Enantioselective epoxidation of terminal alkenes to (R)- and (S)-epoxides by engineered cytochromes P450 BM-3

Article abstract of DOI:10.1002/chem.200500584

Cytochrome P450 BM-3 from Bacillus megaterium was engineered for enantioselective epoxidation of simple terminal alkenes. Screening saturation mutagenesis libraries, in which mutations were introduced in the active site of an engineered P450, followed by recombination of beneficial mutations generated two P450 BM-3 variants that convert a range of terminal alkenes to either (R)- or (S)epoxidc (up to 83 % ee) with high catalytic turnovers (up to 1370) and high epoxidation selectivities (up to 95%). A biocatalytic system using E. coli lysates containing P450 variants as the epoxidation catalysts and in vitro NADPH regeneration by the alcohol dehydrogenase from Thermoanaerobium brockii generates each of the epoxide enantiomers, without additional cofactor.

Full text of DOI:10.1002/chem.200500584

DOI: 10.1002/chem.200500584
Enantioselective Epoxidation of Terminal Alkenes to (R)- and (S)-Epoxides
by Engineered Cytochromes P450 BM-3
[
a]
Takafumi Kubo, Matthew W. Peters, Peter Meinhold, and Frances H. Arnold*
Abstract: Cytochrome P450 BM-3
from Bacillus megaterium was engi-
neered for enantioselective epoxidation
of simple terminal alkenes. Screening
saturation mutagenesis libraries, in
which mutations were introduced in
the active site of an engineered P450,
followed by recombination of benefi-
cial mutations generated two P450
BM-3variants that convert a range of
terminal alkenes to either (R)- or (S)-
epoxide (up to 83% ee) with high cata-
lytic turnovers (up to 1370) and high
epoxidation selectivities (up to 95%).
A biocatalytic system using E. coli ly-
sates containing P450 variants as the
epoxidation catalysts and in vitro
NADPH regeneration by the alcohol
dehydrogenase from Thermoanaerobi-
um brockii generates each of the epox-
ide enantiomers, without additional co-
factor.
Keywords: alkenes · biocatalysis ·
cofactors
·
cytochrome p450
·
enantioselectivity · epoxidation
[
13]
Introduction
al substrates. We previously demonstrated that a combina-
tion of directed evolution by random mutagenesis over the
whole gene and modification of active-site residues in-
Enantioselective epoxidation of terminal alkenes remains a
synthetic challenge. Catalytic systems such as titanium–tar-
creased alkane hydroxylation activity of BM-3and altered
its regio- and enantioselectivity.
also exhibited respectable activity on several alkenes, but
generated mainly the allyl-hydroxylated product 3 rather
[1]
[2]
[3]
[14]
trate, manganese–salen, and iron–porphyrin complexes
have been developed for asymmetric alkene epoxidation;
however, a chemical catalyst effective on simple linear ter-
Various BM-3mutants
[4]
[15]
minal alkenes has not yet been reported. Several enzymes,
than epoxide 2 from 1-hexene (1, Scheme 1).
Here we
[5]
[6]
including cytochromes P450, methane monooxygenases,
report further engineering of BM-3variants that produce
enantio-enriched epoxides from simple terminal alkenes.
Two engineered cytochromes P450 show good catalytic turn-
overs and high epoxidation selectivities. One variant produ-
ces (R)-epoxides, while the other gives (S)-epoxides with up
to 83% ee.
[7]
[8]
toluene monooxygenases, styrene monooxygenases, and
[9]
chloroperoxidases, as well as microbial whole-cell catalysts
[10]
such as Rhodococcus rhodochrous, catalyze the enantiose-
lective epoxidation of terminal alkenes. High optical purity
is achieved in some cases, but these biocatalytic systems
generally only produce a single enantiomer and only accept
[11]
a limited range of alkene substrates. While chiral terminal
epoxides can be prepared by kinetic resolution of racemic
[12]
epoxides with cobalt–salen catalysts or epoxide hydrolas-
[
11]
es, direct enantioselective epoxidation remains an attrac-
tive goal for its simplicity and higher potential yield.
The fatty acid hydroxylase cytochrome P450 BM-3from
Bacillus megaterium has proven to be a versatile platform
for engineering oxidation catalysts for a variety of nonnatur-
Scheme 1. Oxidation of 1-hexene by P450 BM-3.
Results and Discussion
[
a] T. Kubo, Dr. M. W. Peters, P. Meinhold, Prof. F. H. Arnold
Division of Chemistry and Chemical Engineering
California Institute of Technology
Creation and screening of enzyme libraries: Starting from
[14]
previously isolated BM-3variant 9-10A,
which exhibits
high (hydroxylation) activity on small alkanes and some ep-
oxidation activity towards terminal alkenes, we targeted key
residues in the active site for further modification. Using the
1
200 E. California Blvd. MC210–41, Pasadena, CA 91125 (USA)
Fax : (+1)626-568-8743
E-mail: frances@cheme.caltech.edu
1216
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 1216 – 1220

FULL PAPER
crystal structure of the heme domain of wild-type BM-3
two active-site substitutions, F87V and I263A, in addition to
the 14 that distinguish 9-10A from wild-type BM-3. RH-47
also contains the I263A substitution, but in combination
with A82F and A328V (as well as the 9-10A mutations).
Residues A82, F87, and A328 are known determinants of
[16]
with a bound substrate,
we selected 11 residues within
5
Š of the ten carbon atoms of the (fatty acid) substrate
closest to the heme: A74, L75, V78, F81, A82, F87, T88,
T260, I263, A264, A328. Saturation mutagenesis was used to
generate 11 libraries, each containing all possible amino acid
substitutions at one of these positions. These enzyme libra-
ries were screened for epoxidation activity towards 1-hexene
by using 4-(p-nitrobenzyl)pyridine, which reacts with termi-
nal epoxides to form a blue adduct that can be quantified
[14,19]
regio- and enantioselective oxidation in BM-3.
Catalytic properties of cytochrome P450 BM-3 variants SH-
44 and RH-47: The catalytic properties of wild-type BM-3,
parent 9-10A, and the SH-44 and RH-47 variants reported
here are detailed in Table 2. The total turnover numbers
(TTNs) of SH-44 and RH-47 for 1-hexene epoxidation are
eight- and fivefold higher than parent 9-10A, respectively,
and more than 30-fold higher than wild-type BM-3. Epoxi-
dation selectivities are also greatly improved: SH-44 and
RH-47 produce ~90% epoxide, while wild-type and 9-10A
mainly produce the allylic hydroxylation product. Of partic-
ular note, SH-44 and RH-47 catalyze epoxidation with op-
posite enantioselectivities: SH-44 selectively produces the
(S)-epoxide, while RH-47 makes the (R)-epoxide with mod-
erate enantiomeric excesses (71 and 60% ee, respectively).
Activities towards linear terminal alkenes of length C5 to
C8 were also determined (Table 3). Variant SH-44 supports
more turnovers on shorter alkenes (1370 on 1-pentene vs
200 on 1-octene), and the epoxide accounted for at least
80% of the product for all the alkenes tested. Variant RH-
47 shows similar, moderate TTNs and high regioselectivities
on all the alkenes. The highest enantioselectivity is 83% ee
[17,18]
spectroscopically.
The assay conditions reported previ-
[18]
ously by us were modified to greatly improve the sensitiv-
ity (10 mm for detection of 1,2-epoxyhexane). Five of the li-
braries—at positions L75, A82, F87, I263, A328—yielded
enzymes with improved activity relative to 9-10A. From
these, eight variants of 9-10A containing amino acid substi-
tutions L75S, A82F, A82L, F87I, F87L, F87V, I263A, A328V
were selected and subsequently shown to have not only in-
creased total catalytic turnover number (TTN) for epoxide
production from 1-hexene, but also increased (R)- or (S)-
enantioselectivity (Table 1).
A library containing different combinations of these eight
active-site mutations in 9-10A was created and screened for
epoxidation activity on 1-hexene. Two variants isolated from
this second-generation library, SH-44 and RH-47, are partic-
ularly efficient 1-hexene epoxidation catalysts; they also ex-
hibit opposite enantioselectivities (Table 2). SH-44 contains
(
for (R)-1,2-epoxyoctane) by RH-47. The favored 1,2-epox-
Table 1. Effects of beneficial mutations observed in cytochrome P450
variant 9-10A.
ide remained the S enantiomer with SH-44 and the R enan-
tiomer with RH-47 for all the substrates. No chemical cata-
lysts are known to epoxidize the terminal linear alkenes
[
a]
Mutation
in 9-10A
TTN
Epoxidation
Epoxide optical
[
b]
[c]
selectivity [%]
purity [% ee]
[20]
used here with more than 50% ee. Furthermore, no bio-
catalyst has been engineered to epoxidize these alkenes to
produce both enantiomers prior to this report. The TTNs of
SH-44 and RH-47 towards linear terminal alkenes are com-
parable to those reported for nonasymmetric chemical cata-
lysts. To our knowledge, the best reported TTN is 2200 on
1-octene, by a polyoxometalate with molecular oxygen as
none
140
220
210
220
450
350
670
300
280
25
45
64
49
67
76
7373
68
81
4 (R)
19 (S)
2 (R)
16 (R)
45 (S)
79 (S)
S)
L75S
A82F
A82L
F87I
F87L
F87V
I263A
A328V
(
11 (R)
63 (R)
[21]
the oxidant.
This catalyst, however, requires a much
[
a] Total catalytic turnover number for epoxidation of 1-hexene. Epoxide
longer reaction time and higher temperature (over 300 h at
808C) than P450 BM-3, which catalyzes the same reaction
with hundreds of turnovers at room temperature in less than
was quantified by the NBP method. Reaction conditions: 0.1 mm P450 (E.
coli cell lysate), 4 mm substrate (1-hexene), 0.5 mm NADPH, 1 hour at
room temperature. [b] Selectivity for 1,2-epoxyhexane formation. Re-
mainder is mostly 1-hexen-3-ol. These data were obtained by GC analy-
sis. [c] Favored enantiomer is listed in parentheses. Data obtained by GC
analysis.
3
h.
We also determined the activities of SH-44 and RH-47
for hexane hydroxylation (Table 4). The selectivities were
consistent with the relative
product distributions on 1-
[
a]
Table 2. Epoxidation of 1-hexene by wild-type cytochrome P450 BM-3and its variants.
hexene: both SH-44 and RH-47
convert hexane primarily to 1-
and 2-hexanol (cf. 9-10A, which
mostly produces 3-hexanol
from hexane and 1-hexen-3-ol
[
b]
[c]
Enzyme
TTN
Product distribution [%]
Epoxide optical
purity [% ee]
[
c]
epoxide 2
all products
epoxide 2
3-ol 3
other products
Wt
20
130
1090
610
200
530
1240
660
10
25
88
932
85
65
6
5
10
6
17 (R)
4 (R)
71 (S)
R)
9
-10A
SH-44
RH-47
from 1-hexene), indicating
a
5
60 (
shift in the carbon atoms acti-
vated by the iron–oxygen spe-
cies towards terminal positions.
[
a] Reaction conditions: 0.1 mm P450 , 4 mm substrate (1-hexene), 0.5 mm NADPH, 3h at 20 8C. Data were ob-
tained by GC on a chiral column. [b] Total catalytic turnover number. Errors are at most 6%. [c] Errors are
less than 1%.
Chem. Eur. J. 2006, 12, 1216 – 1220
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1217

F. H. Arnold et al.
Table 3. Epoxidation of terminal alkenes by cytochrome P450 BM-3 var-
[
a]
iants SH-44 and RH-47.
[
b]
Variant
Substrate
TTN
Epoxidation
selectivity [%]
Epoxide optical
purity [% ee]
[c]
[c]
SH-44
SH-44
SH-44
SH-44
RH-47
RH-47
RH-47
RH-47
1-pentene
1-hexene
1-heptene
1-octene
1-pentene
1-hexene
1-heptene
1-octene
1370
1090
500
200
570
610
550
560
93
88
85
84
94
9360 (
95
73 (S)
71 (S)
65 (S)
55 (S)
60 (R)
R)
Scheme 2. NADPH is regenerated by coupling to the oxidation of 2-
propanol catalyzed by ADH.
76 (R)
83( R)
92
[
a] Experimental conditions are those for Table 2. [b] Total catalytic turn-
from the lysate. As shown in Table 5, this system converted
over number for epoxidation. Errors are at most 12%. [c] Errors are less
than 1%.
1
-hexene to the 1,2-epoxyhexanes with good yields in 7 h,
by using NADPH regenerated by ADH. The enantioselec-
tivities and epoxidation selectivities of SH-44 and RH-47
observed earlier were preserved in this system. That no ad-
ditional cofactor was necessary makes the synthesis more
economical. We believe that this is the first example of a
monooxygenase-catalyzed epoxidation with in vitro
NADPH regeneration, although epoxidation with regenera-
tion of NADH by NADH-dependent formate dehydrogen-
Table 4. Hydroxylation of hexane by engineered cytochrome P450 BM-3
variants.
[
a]
[
b]
[c]
Variant
TTN
Product distribution [%]
1
-hexanol
2-hexanol
3-hexanol
other
9
-10A
3960
2300
4070
0.6
6
22
22
56
7332
77
34
0.4
4
SH-44
RH-47
[28]
[
1
a] Experimental conditions are those for Table 2. [b] Errors are at most
1%. [c] Errors are less than 1%.
ase has been reported.
Table 5. Epoxidation of 1-hexene by cytochrome P450 BM-3variants
SH-44 and RH-47, using the NADPH regeneration system.
TTNs on hexane were significantly higher than on 1-hexene,
which suggests that epoxidation of linear terminal alkenes
leads to more rapid enzyme inactivation. Possible inacti-
vation mechanisms include heme alkylation and peptide
[a]
Variant
SH-44
RH-47
[22]
added substrate [mm]
10
10
remaining substrate [mm]
0.1
7.6
8.0
81
95
71 (S)
1.2
7.2
7.4
86
98
57 (R)
[
23,24]
modification, as reported for other P450 s.
terminal alkene that is known not to cause P450 inactiva-
Styrene is a
1
,2-epoxyhexane [mm]
total products [mm]
[
b]
[23]
mass balance [%]
epoxidation selectivity [%]
epoxide optical purity [% ee]
tion. Variants SH-44, RH-47, and even 9-10A exhibit very
high TTNs for styrene epoxidation (14000 for SH-44, 10000
for RH-47, 4000 for 9-10A).
[
c]
[
a] Reactions were performed under the following conditions: 10 mm
À1
P450 (E. coli cell lysate), 1 UnitmL ADH, 1 vol% 2-propanol, and 7 h
reaction at room temperature. Oxygen was initially pressurized to
0.15 MPa (gauge) and added at 2 h and 4 h (1–2 mL at ambient pressure).
Epoxidation with NADPH regeneration: Cytochrome P450
BM-3-catalyzed epoxidation requires an equivalent of
NADPH to convert an equivalent of alkene substrate. Thus
efficient use of the enzyme requires cofactor recycling,
1
0 mm substrate (1-hexene) was added over 3h. [b] Mass balance ={c(re-
maining substrate + c (total products)}/c(added substrate)”100. 19%
and 14% of substrate could not be recovered; this seems to be caused by
the high volatility of the substrate and the product epoxide. No signifi-
cant by-products were detected. [c] Epoxidation selectivity=c(1,2-epoxy-
hexane)/c(total products)”100.
[25]
which can be accomplished in vitro or in vivo, in whole
cells. The whole-cell reaction is simple, but sometimes diffi-
cult when the substrate is hydrophobic, as are the alkenes
investigated here. Therefore, we tested in vitro NADPH re-
generation to perform enantioselective epoxidation with the
two BM-3variants. Dehydrogenases such as alcohol dehy-
drogenase, glucose dehydrogenase, isocitric dehydrogenase
and formate dehydrogenase can be used as the cofactor-re-
Conclusion
We have engineered P450 BM-3for enantioselective epoxi-
dation of terminal alkenes by using saturation mutagenesis
and recombination coupled with a colorimetric high-
[
26]
generating enzyme.
ADH), because the alcohol can serve not only as the regen-
We chose alcohol dehydrogenase
(
eration driving force, but also as the cosolvent for hydropho-
bic substrates. We employed the NADPH-dependent alco-
hol dehydrogenase from Thermoanaerobium brockii, which
is highly thermo- and solvent-stable, and specific to shorter
throughput screen for epoxide formation to identify efficient
epoxidation catalysts. P450 variants isolated in this way con-
vert a range of terminal alkenes to the (R)- or (S)-epoxides
with high catalytic turnovers and high epoxidation selectivi-
ties. We also showed that P450-catalyzed epoxidation could
be performed by using cell lysate containing the P450 and
alcohol dehydrogenase for regeneration of the expensive
NADPH cofactor, without additional NADPH. We antici-
pate that improvements in catalytic properties including
[27]
and secondary alcohols.
NADPH was recycled by the
ADH-catalyzed oxidation of 2-propanol to acetone coupled
+
to the reduction of NADP (Scheme 2).
An E. coli cell lysate containing the variant P450 s was
used for epoxidation so that the initial cofactor is supplied
1218
www.chemeurj.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 1216 – 1220

Biocatalysis
FULL PAPER
enantioselectivity on desired target substrates can be ach-
ieved using the methods of mutagenesis and screening out-
lined here. This and our recent report describing enantiose-
lective hydroxylation of protected carboxylic acids by other
for each saturation library contained all possible combinations of bases,
NNN (N=A, T, G, or C), at the codon for a particular residue. The pri-
mers in the forward direction for each library were: 74NNNfor (5’-
GTCAANNNCTTAAATTTGCACG-3’),
GTCAAGCGNNNAAATTTGCACG-3’),
GCTTAAATTTNNNCGTGATTTTGCAGG-3’),
75NNNfor
78NNNfor
(5’-
(5’-
(5’-
[29]
BM-3variants
demonstrate that the BM-3active site can
81NNNfor
CGTGATNNNGCAGGAGAC-3’), 82NNNfor (5’-CGTGATTTTNNNG-
GAGAC-3’), 87NNNfor (5’-GAGACGGGTTANNNACAAGCTGGAC-
be readily molded for enantio- and regioselective oxidation
of a range of substrates.
3
2
2
’), 88NNNfor (5’-GGAGACGGGTTATTTNNNAGCTGGACG-3’),
60NNNfor
63NNNfor
(5’-CAAATTATTNNNTTCTTAATTGCGGGAC-3’),
(5’-ACATTCTTANNNGCGGGACACGAAAC-3’),
Experimental Section
264NNNfor (5’-ACATTCTTAATTNNNGGACACGAAAC-3’), and
28NNNfor (5’-CCAACTNNNCCTGCGTTTTCC-3’). The reverse pri-
3
mers for each of these libraries complement their corresponding forward
primers. For each mutation, two separate PCRs were performed, each
using a perfectly complementary primer, BamHIfor (5’-GGAAACAG-
GATCCATCGATGC-3’) and SacIrev (5’-GTGAAGGAATACCGC-
CAAGC-3’), at the end of the sequence and a mutagenic primer. The re-
sulting two overlapping fragments that contain the mutations were then
annealed in a second PCR to amplify the complete mutated gene. The
full gene was then cut with BamHI and SacI restriction enzymes and li-
gated with T4 ligase (Invitrogen) into pBM3_WT18-6, previously cut
with BamHI and SacI to remove the wild-type gene. The ligation mix-
tures were then transformed into E. coli DH5a electrocompetent cells
and plated onto LB agar plates to form single colonies for picking. 176
clones were screened for each library in the first screening. Positive
clones that showed higher epoxide formation than parent 9-10A were re-
assayed in the second screening, in which each clone was inoculated in
one column (8 wells). P450 concentration was also measured in the
second screening by CO difference spectroscopy to correct for the
enzyme expression level. Positive clones were grown on TB medium
Materials: 1-Pentene (99%), 1-hexene (99%), 1-octene (98%), styrene
(
99%), product standards such as racemic epoxides (except 1,2-epoxy-
heptane), (R)-1,2-epoxyhexane, (R)-1,2-epoxyoctane, 1-hexen-3-ol, hexa-
nols and solvents were purchased from Sigma-Aldrich. 1-Heptene (98%)
and 1,2-epoxyheptane were purchased from TCI America. n-Hexane
(
99%) was purchased from Fluka. NADPH was obtained from Biocata-
lytics, Inc.
Plasmids and expression of P450 BM-3: P450 BM-3was expressed as de-
[
30]
scribed.
The P450 BM-3gene or mutants thereof, which include a
silent mutation to introduce a SacI site 130bp upstream of the end of the
heme domain, were cloned behind the double tac promoter of the expres-
[
31]
sion vector pCWori (pBM3WT18–6). E. coli DH5a, transformed with
these plasmids, was used for expression of P450 BM-3on a 500 mL scale
as well as for expression in 96-well plates.
Purification of P450 BM-3: For protein production, supplemented terrific
À1
broth (TB) medium (500 mL, 100 mgmL ampicillin) in a 2.8 L flask was
inoculated with an overnight culture (3mL) and incubated at 30 8C and
1
80–200 rpm shaking. After 24 h of incubation, d-aminolevulinic acid hy-
(
25 mL) to prepare cell lysates under the same conditions that are de-
drochloride (ALA; 0.5 mm) was added, and expression was induced by
addition of isopropyl-b-d-thiogalactoside (IPTG; 1 mm). Cells were har-
vested by centrifugation 24 h after induction. The enzymes were purified
scribed in the “Purification of P450 BM-3” section. Cell lysates were
used for characterization of mutants.
Recombination of P450 BM-3 mutants: Recombination of BM-3muta-
tions identified in the saturation mutagenesis libraries was performed by
repeating PCR overlap extension mutagenesis with mutagenic primers
and mixed DNA templates as described below. Step 1 (recombination of
L and S at residue 75 and of A, F, and L at residue 82): a PCR was per-
formed by using a mutagenic primer 75LSfor (5’-CTTAAGTCAAGCG-
TYRAAATTTGCACG-3’), SacIrev, and a DNA plasmid mixture of
mutant A82F, mutant A82L, and 9-10A as PCR template. These plas-
mids were mixed at a same concentration (estimated from ABS260). An-
other PCR was also performed using the complementary reverse primer
of 75LSfor, BamHIfor, and the plasmid of 9-10A as a template. The re-
sulting two overlapping fragments were then assembled in a second PCR
to give the whole gene recombined at residue 75 and 82. Step 2 (recombi-
nation of F, V, I, and L at residue 87): a PCR with a mutagenic primer
[
31]
following published procedures. Enzyme concentration was measured
[
32]
in triplicate from the CO difference spectra.
Preparation of cell lysates for high-throughput screening: Single colonies
were picked and inoculated into 1 mL deep-well plates containing Luria–
À1
Bertani (LB) medium (350 mL, supplemented with 100 mgmL ampicil-
lin). The plates were incubated at 308C, 250 rpm, and 80% relative hu-
midity. After 24 h, clones from this preculture were inoculated using a
9
6-pin replicator into 2 mL deep-well plates containing TB medium
À1
(
500 mL, supplemented with 100 mgmL ampicillin, 10 mm IPTG and
0
.5 mm ALA). The cultures were grown at 308C, 250 rpm, and 80% rela-
tive humidity for another 24 h. Cells were then pelleted and stored
frozen at À208C until they were resuspended in 500 mL 100 mm phos-
À1
À1
phate buffer (pH 8, containing 0.5 mgmL lysozyme, 2 UnitsmL
2
DNase I, and 10 mm MgCl ). After 30 min at 378C, the lysates were cen-
trifuged and the supernatant was used for activity measurements in 96-
well microtiter plates.
8
7FVILfor (5’-GAGACGGGTTANTYACAAGCTGGAC-3’), SacIrev,
and the plasmid DNA of 9-10A as a template was performed. Another
PCR using the complementary reverse primer of 87FVILfor, BamHIfor,
and the assembled DNA fragment from step 1 as a template was also per-
formed. The resulting two fragments were then assembled in a second
PCR to give the whole gene recombined at 75, 82, and 87. Step 3(recom-
bination of I and A at residue 263and of A and V at residue 32 8): a
PCR with a mutagenic primer 263IAfor (5’-CAAATTATTACATTCT-
TARYHGCGGGACACG-3’), SacIrev, and a DNA plasmid mixture of
mutant A328 V and 9-10A as a template was performed. Another PCR
with the complementary reverse primer of 263IAfor and BamHIfor with
the assembled DNA fragment from step 2 as a template was also per-
formed. The resulting two fragments were then assembled in a second
PCR to give the whole gene recombined at residues 75, 82, 87, 263, and
High-throughput epoxide formation assay: For high-throughput screening
[
18]
of mutants based on epoxide formation, the published method using 4-
p-nitrobenzyl)pyridine (NBP) was modified as follows: to E. coli super-
(
natant (30–40 mL), phosphate buffer (100 mm, pH 8) was added to 100 mL
of total volume on the 96-well plate. After addition of a solution 0.4m 1-
hexene in ethanol (1.5 mL) as substrate, oxidation was started by addition
of 50 mL of 1.5 mm NADPH solution (in buffer). The plate was sealed
with a plastic film (GeneMate) immediately so that epoxide does not
evaporate. After 30 minutes incubation at room temperature, a solution
of NBP (100 mL; 6 wt/vol% NBP, 19.5 vol% 1-butanol, 80 vol% propyl-
ene glycol, 0.5 vol% acetic acid) was added, and the plate was sealed
again with a film. The plate was placed in an oven at 808C for 2.5 h and
chilled on ice for 10 minutes. Subsequently, solution of triethylamine in
acetone (100 mL, 50 vol%) was added to develop color, followed by
measurement of absorbance at 580 nm on a plate-reader.
3
28. The recombined gene was then cut with BamHI and SacI restriction
enzymes, ligated into vector DNA, and transformed into E. coli DH5a.
The library size was 192, and about 500 clones were screened for epoxide
formation in the first screening. 22 clones, which showed same or higher
epoxide formation than the mutant F87V, were re-assayed in the second
screening. Positives from this were then grown on TB medium (25 mL)
for preparation of cell lysates and characterization. Two mutants (SH-44
Construction and screening of saturation mutagenesis libraries: Muta-
tions were introduced into mutant 9-10A by PCR overlap extension mu-
[
33]
tagenesis using Pfu turbo DNA polymerase (Stratagene). The primers
Chem. Eur. J. 2006, 12, 1216 – 1220
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1219

F. H. Arnold et al.
and RH-47) that showed the highest turnovers with different enantiose-
lectivity were grown on a 500 mL scale and purified as described above.
[4] V. B. Valodkar, G. L. Tembe, R. N. Ram, H. S. Rama, Catal. Lett.
2003, 90, 91–94.
[
[
5] C. A. Martinez, J. D. Stewart, Curr. Org. Chem. 2000, 4, 263–282.
6] S. J. Elliott, M. Zhu, L. Tso, H. H. T. Nguyen, J. H. K. Yip, S. I.
Chan, J. Am. Chem. Soc. 1997, 119, 9949–9955.
Quantification of epoxides by the NBP method: Epoxides were quanti-
fied using NBP under conditions described in the “High-throughput ep-
oxide formation assay” section. Enzymatic reactions were performed on
a 0.6 mL scale in 1.7 mL tubes. Calibration of the epoxide was conducted
using a standard reagent of 1,2-epoxyhexane.
[
[
[
7] K. McClay, B. G. Fox, R. J. Steffan, Appl. Environ. Microbiol. 2000,
6
6, 1877–1882.
8] A. Schmid, K. Hofstetter, H. J. Feiten, F. Hollmann, B. Witholt,
Adv. Synth. Catal. 2001, 343, 732–737.
9] A. F. Dexter, F. J. Lakner, R. A. Campbell, L. P. Hager, J. Am.
Chem. Soc. 1995, 117, 6412–6413.
Chiral gas chromatography: A CycloSil-B chiral capillary column (Agi-
lent Technologies, 30 m length, 0.32 mm ID, 0.25 mm film thickness) and
a Flame Ionization Detector (FID) were used. The reaction products
were determined and quantified by using standard racemic epoxides and
alcohols. The absolute configurations of epoxides were determined by
comparing the reaction product with a standard (R)-enantiomer-enriched
epoxide in the case of 1,2-epoxyhexane and 1,2-epoxyoctane. For 1,2-ep-
oxypentane and 1,2-epoxyheptane, the standard (S)-enantiomer-enriched
epoxides were prepared using the yeast strain Rhodosporidium toruloides
[
10] K. Furuhashi, Biological Routes to Optically Active Epoxides, Wiley,
New York, 1992, pp. 167–186.
11] A. Archelas, R. Furstoss, Annu. Rev. Microbiol. 1997, 51, 491–525.
12] M. Tokunaga, J. F. Larrow, F. Kakiuchi, E. N. Jacobsen, Science
[
[
1
997, 277, 936–938.
[
[
13] A. W. Munro, D. G. Leys, K. J. McLean, K. R. Marshall, T. W. B.
Ost, S. Daff, C. S. Miles, S. K. Chapman, D. A. Lysek, C. C. Moser,
C. C. Page, P. L. Dutton, Trends Biochem. Sci. 2002, 27, 250–257;
V. B. Urlacher, S. Lutz-Wahl, R. D. Schmid, Appl. Microbiol. Bio-
technol. 2004, 64, 317–325.
14] M. W. Peters, P. Meinhold, A. Glieder, F. H. Arnold, J. Am. Chem.
Soc. 2003, 125, 13442–13450.
15] E. T. Farinas, M. Alcalde, F. Arnold, Tetrahedron 2004, 60, 525–528.
16] D. C. Haines, D. R. Tomchick, M. Machius, J. A. Peterson, Biochem-
istry 2001, 40, 13456–13465.
(
ATCC 10657) expressing an epoxide hydrolase that is known to hydro-
[
34]
lyze (R)-epoxides preferentially. Conditions for GC analysis for 1,2-ep-
oxyhexane were: isothermal at 708C for 2.5 min, linear gradient to 2008C
in 13min and isothermal at 200 8C for 5 min. The conditions were slightly
modified for different epoxides. The separation of the R and S enantio-
mers of the epoxides gave the following retention times: 5.04 and
5
.14 min (1,2-epoxypentane), 5.21 and 5.29 min (1,2-epoxyhexane), 14.17
[
[
and 14.35 min (1,2-epoxyheptane) and 19.39 and 19.46 min (1,2-epoxyoc-
tane).
Analysis of alkene epoxidation and alkane hydroxylation: Oxidations of
[
[
[
17] F. Zocher, M. M. Enzelberger, U. T. Bornscheuer, B. Hauer, R. D.
1
-pentene, 1-hexene, 1-heptene, 1-octene, styrene, and n-hexane were
Schmid, Anal. Chim. Acta 1999, 391, 345–351.
performed on 1 mL scale in phosphate buffer (pH 8, 100 mm) containing
P450, substrate, and 1% ethanol in 1.7 mL tubes. Reactions were started
by adding NADPH (dissolved in buffer). After reaction, a solution of 1-
octanol in ethanol (10 mL; 10 mm) was added as the internal standard
and chloroform (300 mL) was also added to extract products. The tubes
were vortexed vigorously for 1 min, and subsequently centrifuged for
18] M. Alcalde, E. T. Farinas, F. H. Arnold, J. Biomol. Screening 2004, 9,
1
41–146.
19] S. Graham-Lorence, G. Truan, J. A. Peterson, J. R. Falck, S. Z. Wei,
C. Helvig, J. H. Capdevila, J. Biol. Chem. 1997, 272, 1127–1135;
Q. S. Li, J. Ogawa, R. D. Schmid, S. Shimizu, FEBS Lett. 2001, 508,
49–252.
20] R. Sinigalia, R. A. Michelin, F. Pinna, G. Strukul, Organometallics
987, 6, 728–734; R. I. Kureshy, N. H. Khan, S. H. R. Abdi, S. T.
Patel, P. Iyer, E. Suresh, P. Dastidar, J. Mol. Catal. A 2000, 160,
17–227.
2
1
min at 14000 rpm. The lower organic layer was collected and analyzed
[
by gas chromatography to determine total catalytic turnover numbers,
product distributions, and enantiomeric excess. Reactions were per-
formed in triplicate.
1
2
Epoxidation with NADPH regeneration: E coli cells were grown in a TB
culture (500 mL), and P450 expression was induced as described above.
Cells were collected, suspended in phosphate buffer (20 mL, 100 mm,
pH 8), and disrupted by sonication. The cell lysate was recovered by cen-
trifugation, and P450 concentration was measured by CO binding (SH-
[21] Y. Nishiyama, Y. Nakagawa, N. Mizuno, Angew. Chem. 2001, 113,
3751–3753; Angew. Chem. Int. Ed. 2001, 40, 3639–3641.
[22] Enzymes are rapidly inactivated. Initial rates on 1-hexene epoxida-
À1
tion are very high (80 turnovers min for S-44 and 160 turnovers
À1
min for R-47).
4
4, 63 mm, RH-47, 30 mm). Epoxidation reactions were performed in a vial
[23] P. R. Ortiz de Montellano, M. A. Correia, Annu. Rev. Pharmacol.
Toxicol. 1983, 23, 481–503.
(
12 mL), with phosphate buffer (100 mm, pH 8), E. coli cell lysate con-
taining P450 and the alcohol dehydrogenase from Thermoanaerobium
brockii (purchased from Sigma). The contents were mixed and the vial
sealed with a crimp cap (Agilent Technologies). Oxygen was initially
pressurized from the top and added every few hours using a syringe. 1-
Hexene was dissolved in 2-propanol (1m), and the solution was slowly
added to the vial with a syringe over a few hours with stirring. A portion
of the reaction mixture was sampled from the vial and extracted with
chloroform, and the extract was analyzed by gas chromatography. Total
reaction volume was 9 mL.
[24] K. He, Y. A. He, G. D. Szklarz, J. R. Halpert, M. A. Correia, J. Biol.
Chem. 1996, 271, 25864–25872.
[25] W. A. van der Donk, H. M. Zhao, Curr. Opin. Biotechnol. 2003, 14,
421–426.
[26] R. Wichmann, D. Vasic-Racki, Adv. Biochem. Eng./Biotechnol. 2005,
92, 225–260.
[27] L. S. al-Kassim, C. S. Tsai, Biochem. Cell Biol. 1990, 68, 907–913.
[28] K. Hofstetter, J. Lutz, I. Lang, B. Witholt, A. Schmid, Angew. Chem.
2004, 116, 2215–2218; Angew. Chem. Int. Ed. 2004, 43, 2163–2166.
[
29] D. Munzer, P. Meinhold, M. W. Peters, S. Fiechtenhofer, H. Griengl,
A. Glieder, F. H. Arnold, A. de Raadt, Chem. Commun. 2005, 20,
2
597–2599.
30] A. Glieder, E. T. Farinas, F. H. Arnold, Nat. Biotechnol. 2002, 20,
135–1139.
Acknowledgements
[
[
1
The authors thank Dr. Nathan Dalleska for assistance with gas chroma-
tography. This work is supported by the National Science Foundation
BES-0313567), the Donors of the American Chemical Society Petrole-
um Research Fund (ACS PRF Alternative Energy Postdoctoral Fellow-
ship for MWP) and the Nippon Shokubai Co., Ltd.
31] E. T. Farinas, U. Schwaneberg, A. Glieder, F. H. Arnold, Adv. Synth.
Catal. 2001, 343, 601–606.
32] T. Omura, R. Sato, J. Biol. Chem. 1964, 239, 2370–2378.
33] R. Higuchi, B. Krummel, R. K. Saiki, Nucleic Acids Res. 1988, 16,
(
[
[
7
351–7367.
[
34] A. L. Botes, C. A. G. M. Weijers, P. J. Botes, M. S. van Dyk, Tetrahe-
dron: Asymmetry 1999, 10, 3327–3336.
[
[
1] T. Katsuki, K. B. Sharpless, J. Am. Chem. Soc. 1980, 102, 5974–5976.
2] W. Zhang, J. L. Loebach, S. R. Wilson, E. N. Jacobsen, J. Am. Chem.
Soc. 1990, 112, 2801–2803.
Received: May 26, 2005
[
3] E. Rose, Q. Z. Ren, B. Andrioletti, Chem. Eur. J. 2004, 10, 224–230.
Published online: October 20, 2005
1220
www.chemeurj.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 1216 – 1220