Enantioselective Benzylic Hydroxylation Catalysed by P450 Monooxygenases: Characterisation of a P450cam Mutant Library and Molecular Modelling

Article abstract of DOI:10.1002/cbic.201500536

Cytochrome P450 monooxygenases can catalyse the stereoselective C-H activation of a very broad range of substrates. Prediction and control of enantioselectivity of this enzyme class is of great interest for the synthesis of high-value chiral molecules. Here we have used a combination of molecular dynamics simulations and experimental screening to study the enantioselectivity of a library of active-site mutants of chimeric P450cam-RhFRed towards the benzylic hydroxylation of structurally related regioisomers of ethylmethylbenzene. Small variations either in substrate structure or in enzyme active site architecture were shown to lead to dramatic changes in enantioselectivity; this was broadly in agreement with computational predictions. In addition to validating computational approaches, these studies have provided us with a deeper understanding of effects that might control stereoselectivity in these biooxidation reactions.

Full text of DOI:10.1002/cbic.201500536

DOI: 10.1002/cbic.201500536
Full Papers
Enantioselective Benzylic Hydroxylation Catalysed by P450
Monooxygenases: Characterisation of a P450cam Mutant
Library and Molecular Modelling
[
a]
[b]
[a]
[a]
[a]
Anja Eichler, Łukasz Gricman, Susanne Herter, Paul P. Kelly, Nicholas J. Turner,
[b]
[a]
Jürgen Pleiss,* and Sabine L. Flitsch*
Cytochrome P450 monooxygenases can catalyse the stereose-
lective CÀH activation of a very broad range of substrates. Pre-
diction and control of enantioselectivity of this enzyme class is
of great interest for the synthesis of high-value chiral mole-
cules. Here we have used a combination of molecular dynam-
ics simulations and experimental screening to study the enan-
tioselectivity of a library of active-site mutants of chimeric
P450cam-RhFRed towards the benzylic hydroxylation of struc-
turally related regioisomers of ethylmethylbenzene. Small var-
iations either in substrate structure or in enzyme active site ar-
chitecture were shown to lead to dramatic changes in enantio-
selectivity; this was broadly in agreement with computational
predictions. In addition to validating computational ap-
proaches, these studies have provided us with a deeper under-
standing of effects that might control stereoselectivity in these
biooxidation reactions.
Introduction
Catalytic, stereoselective CÀH activation is an important goal in
studies have focussed on stereoselective benzylic hydroxyla-
tions as a way of generating high-value chiral intermediates
organic chemistry and biocatalysis providing contribution to
[1,2]
[7]
such reactions involving oxygen insertion.
One of the major
for the pharmaceutical industry. However, the commonly
classes of enzymes used for CÀH activation is that of the cyto-
chrome P450 monooxygenases (P450s), which can promote
chemo-, regio- and stereoselective oxidations with molecular
used cytochrome P450 monooxygenase P450cam from Pseudo-
monas putida (CYP101) produces benzyl alcohols with only
[
8]
moderate enantioselectivities.
[
3]
oxygen under mild conditions at ambient temperatures. The
limited substrate range of known natural P450s has been ex-
Given the importance of chiral benzylic CÀH activation and
the lack of generic HTS methods for the direct detection of
products, we have started to develop a combination of muta-
[4]
tended by combining active-site mutagenesis with efficient
high-throughput screening (HTS) methodologies. Even more
powerful is the combination of HTS with rational structure-
[
9c]
genesis, surrogate screening and molecular dynamics (MD)
simulation methods as a strategy for faster P450-based biocat-
alyst discovery and optimisation. Our starting point was a self-
sufficient chimeric biocatalyst (P450cam-RhFRed) generated by
combining redox partners from P450RhF from Rhodococcus sp.
(RhFRed, CYP116B2), thus mimicking the natural fusion archi-
[
5a]
guided evolution,
which has recently been shown for
P450BM-3 mutants in regio- and enantioselective hydroxyla-
[5b]
tions of small cyclic Michael acceptors. Based on these initial
successes there is growing interest in applying computational
methods to the directed evolution of P450-mediated biohy-
droxylations, alongside experimentally based HTS methods.
The development of HTS techniques for P450 monooxyge-
nases is not trivial and frequently requires surrogate substrates.
Yang et al. recently used a colorimetric HTS screen for investi-
gations of the subterminal hydroxylation of n-octane and pro-
[
9a]
tecture of P450RhF. The resultant P450cam-RhFRed chimera
was expressed as single protein in the bacterial host, allowing
[
9,10]
fast and surrogate screening of new oxidation activities.
The 96Phe mutation of the P450cam is well established as
a mutation that significantly broadens substrate specificity and
was incorporated into all the mutants studied. Seven
P450cam[Tyr96Phe]-RhFRed double-mutant libraries were gen-
[6]
pylbenzene by P450pyr (CYP153A7). A number of subsequent
[
9c]
erated by targeting the 12 active-site residues specified by
[
8b,11]
[
a] Dipl.-Chem. A. Eichler, Dr. S. Herter, Dr. P. P. Kelly, Prof. N. J. Turner,
Prof. S. L. Flitsch
Loida and Sligar (Figure 1);
these had only recently been
identified as universal selectivity-determining positions
School of Chemistry, Manchester Institute of Biotechnology
University of Manchester, 131 Princess Street
Manchester M1 7DN (UK)
[
12]
through a sequence-based literature algorithm approach.
In addition, positions Phe98 and Met184 were included in
the mutant libraries because Phe98 is thought to contribute to
E-mail: sabine.flitsch@manchester.ac.uk
[
14]
[
b] Ł. Gricman, Prof. J. Pleiss
substrates’ orientations through hydrophobic interactions
whereas Met184 is part of the P450cam substrate recognition
Institute for Technical Biochemistry, University of Stuttgart
Allmandring 31, 70569 Stuttgart (Germany)
E-mail: juergen.pleiss@itb.uni-stuttgart.de
[
15]
site 2 (SRS 2, Figure 1).
To select for active mutants only, we have previously devel-
oped a colony-based solid-phase HTS relying on the formation
Supporting information and ORCID(s) from the author(s) for this article are
available on the WWW under http://dx.doi.org/10.1002/cbic.201500536.
ChemBioChem 2016, 17, 426 – 432
426
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Results and Discussion
Screening for enantioselectivity in the benzylic oxidation of
ethylmethylbenzenes
On the basis of the structural diversity and improved activity
[
8b]
[8e]
of libraries III (Met184/Thr185) and IV (Leu244/Val247), mu-
tants of library III harbouring nonpolar amino acid residues in
[
8b]
place of Thr185 (Figure 1) were selected and compared to
the 96Phe parent (P ) in an in-depth study of the enantioselec-
0
tive oxidation of the ethylmethylbenzene isomers 1–3
(
Table 1). The oxidation of 1 produced alcohol (R,S)-4 in up to
1% yield (Table 1, entry B, 184Val/186Phe) and exclusively
showed S selectivity for all the library III mutants tested
Table 1 and Table S6 in the Supporting Information). However,
4
Figure 1. Active-site residues in the P450cam domain of P450cam-RhFRed
(
[13]
(
PDB ID: 1PHG). Functional mutants were generated by pairwise random
no improvement in enantioselectivity over the parent P (ee
mutagenesis of the following residues to produce libraries I–VII: Phe87/
Phe96 I (green), Phe98/Thr101 II (blue), Met184/Thr185 III (yellow), Leu244/
Val247 IV (purple), Gly248/Thr252 V (cyan), Val295/Asp297 VI (orange),
Ile395/Val396 VII (red).
0
35%, Table 1, entry A, and Table S6) was observed. In order to
explore the influence of the position of the methyl substituent
on the benzene ring with respect to experimentally achieved
ee values, we extended the substrate set to include ortho- and
meta-configured ethylmethylbenzene derivatives 2 and 3, re-
spectively (Tables 1 and S6). Moderate production of (R,S)-5
with a degree of S selectivity (ee 60%, Table 1, entry A) was
found for P0 with substrate 2. Consistently with this, the
184Val/185Phe mutant oxidised 2 with an ee of 64% [60%
yield of (R,S)-5, Table 1, entry B]. Remarkably, substrate 3 pro-
duced alcohols (R,S)-6 in 62% yield with the 184Val/185Phe
mutant, with a tenfold improved ee (74%, Table 1, entry B)
of indigo, using the mutant P450cam[Tyr96Phe]-RhFRed as
[
9c,10]
a template
and yielding a valuable and diverse selection of
9
3 active P450cam-RhFRed biocatalysts that were subsequently
shown to be active against substrates other than indole.
Among all libraries tested, libraries III (Met184/Thr185) and IV
(
Leu244/Val247) contained mutants with improved yields of al-
[
9c]
cohols 4 (Scheme 1). This suggested that the Met184/Thr185
pair (library III) situated on the top of the active site might
indeed impact the oxidation rate of ethylbenzenes as previous-
relative to P (7%, Table 1, entry A). Overall, library III mutants
0
[
8b]
ly demonstrated. Moreover, 247Leu as one of the mutated
residues in library IV (Leu244/Val247), located more closely to
the haem (Figure 1), has been proposed to dictate P450 enan-
revealed only poor to moderate enantioselectivities ranging
from 7% (S) to 74% (S) ee with substrates 1–3 (Table 1 and
Table S6).
[8e]
tioselectivities in the oxidation of ethylbenzene.
We next turned our attention to library IV (Leu244/Val247),
in which the mutated amino acid residues are located more
closely to the haem and are assumed to control enantioselec-
Here we describe the systematic investigation of these
P450cam mutants from libraries III and IV with regioisomers of
ethylmethylbenzenes (1–3, Scheme 1). The overall aim of the
study is to understand how enantioselectivity is controlled by
subtle structural changes in enzyme·substrate complexes and
to compare experiment with prediction based on MD simula-
[
8e,17,18]
tivity (Table 1).
Hence, six out of 19 mutants of library IV
possessing a Val247Leu mutation along with a Tyr96Phe/
Val247Leu mutant were studied (Table 1 and Table S6). Notably,
opposite enantioselectivities were displayed by 244Phe/247Leu
(ee>99%, R, entry C) and 244Asn/247Leu (ee 99%, S, entry D)
mutants in the oxidation of ethylmethylbenzene isomers 1–3
[
16]
tions.
(
Table 1). The oxidation of 1 with the 244Phe/247Leu mutant
produced the alcohols (R,S)-4 with poor yield (2%) and moder-
ate R selectivity (ee 43%, Table 1, entry C). However, the
2
44Asn/247Leu mutant produced the alcohol (S)-4 in >99% ee
and 24% yield (Table 1, entry D). Interestingly, substrate 2 pro-
duced opposite enantioselectivities with 244Phe/247Leu (ee>
9
9%, R, Table 1, entry C) and 244Asn/247Leu (ee 96%, S,
Table 1, entry D). However, yields of the alcohol (R)-5 (11%,
Table 1, entry C) were rather low in comparison with those of
the alcohol (S)-5 (60%, Table 1, entry D). In this series of reac-
tions, substrate 3 and the 244Phe/247Leu mutant showed ex-
cellent R selectivity (ee>99%, Table 1, entry C). On the contra-
ry, 244Asn/247Leu displayed only poor R selectivity (ee=16%,
entry D) with substrate 2 (Table 1, entry D). To address the in-
fluence of overoxidation on enantioselectivity, time-course ex-
periments with alcohols (R,S)-4 to -6 (Table S5) and the mu-
Scheme 1. Enantioselective oxidation of ethylmethylbenzenes 1–3 catalysed
by the P450cam[Tyr96Phe]-RhFRed mutants.
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
427
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Comparison of ee values obtained from computational prediction and experimental analysis.
1
2
3
ee [%]
b]
Yield [%]
Exp.
ee [%]
Yield [%]
Exp.
ee [%]
Yield [%]
Exp.
[
[c]
[d]
[b]
[c]
[d]
[b]
[c]
[d]
Mutant
Comp.
Exp.
Comp.
Exp.
Comp.
Exp.
A
B
C
D
96Phe
184Val/185Phe
244Phe/l247Leu
244Asn/247Leu
60 (S)
16 (S)
38 (R)
75 (S)
35 (S)
28 (S)
43 (R)
20
41
2
20 (S)
61 (S)
56 (R)
85 (S)
60 (S)
64 (S)
>99 (R)
96 (S)
55
60
11
60
15 (S)
76 (S)
75 (R)
32 (S)
7 (S)
74 (S)
>99 (R)
16 (R)
34
62
6
[
a]
[
a]
[a]
>99 (S)
24
22
[
a] 184Val/185Phe, 244Phe/247Leu and 244Asn/247Leu mutants have the parental 96Phe mutation. [b] Enantioselectivities were determined by MD simula-
tions. [c] Enantioselectivities were determined by normal-phase HPLC (Tables S3 and S6). [d] Product yields of alcohols 4–6 were determined by GC/FID (Ta-
bles S2, S4 and S5). All product peaks were compared against purchased standards.
[
19]
tants 244Phe/247Leu and 244Asn/247Leu were accomplished
and substantially confirmed the enantioselectivities presented
P450-catalysed oxidation of ethylbenzenes. Moreover, in our
approach the hydrogen abstraction event was modelled,
whereas a further reorientation of the substrate molecule in
the active site might occur before product is formed, depend-
(
Figure S7).
The experimental data obtained from this set of closely re-
[
19]
lated enzyme mutants and substrates provided an ideal plat-
form for a systematic comparison of experimental data with
computational prediction of trends in reaction selectivity.
ing on radical stabilisation.
The simulations of P450cam-RhFRed mutant·substrate com-
plexes were evaluated by extensive conformational sampling.
By means of a heat map analysis (Figure 2) the frequencies of
conformations on the basis of their distance and angle values
were determined (Scheme 2, Figures S8–S29).
Modelling of enantioselectivities of P450cam mutants
toward ethylmethylbenzenes 1–3
Thus, multiple binding modes of substrates 1–3 in the active
site were identified, and the molecular basis of enantioselectiv-
ity was explained, as shown, for example, for substrate 3 and li-
brary III mutant 184Val/185Phe in substrate 3’s pro-R and pro-S
NACs (Figure 2).
To elucidate the molecular basis of enantioselectivity, MD simu-
[13]
lations of four P450cam mutants in complex with ethylme-
thylbenzenes 1–3 were performed (Table 1). The enantioselec-
tivities of P450cam mutant·substrate complexes were modelled
by monitoring the distance dH–O and the angle aC–H–O between
the substrate and the ferryl oxygen of the haem. Each frame of
the simulation was assigned as
In general, the heat maps obtained from MD simulations re-
vealed broad landscapes of conformations, with substrates 1–3
more often found in inactive conformations than in NACs, in
accordance with low conversion rates of substrates 1–3 (Fig-
being in either a near attack con-
[
20]
formation (NAC, either pro-R or
ure S8–S29). Consistently with the “lock-and-key” or the “in-
[
21]
pro-S) or in an inactive conforma-
duced fit” model applied for substrate recognition, the NACs
of substrates 1–3 with the mutants 244Phe/247Leu and
244Asn/247Leu showed a broad range of enantioselectivity,
thus demonstrating that conformational plasticity of P450cam-
RhFRed mutant·substrate complexes explains the experimen-
tally determined enantioselectivities (Table 1 and Figure S8–
S29).
tion on the basis of distance and
[16a]
angle cutoffs.
A NAC was as-
signed if the distance dH–O to one
of the stereocentre hydrogen
atoms was less than 0.35 nm and
the angle aC–H–O was between 135
and 1808 (Scheme 2).
With all mutants, the ortho-configured ethylmethylbenzene
derivative 2 gave the highest ee values of the three substrates
(Table 1). In the MD simulations, the movements of the ortho-
methyl substituent of 2 were restricted, resulting in a low con-
formational flexibility and thus a high enantioselectivity.
MD simulations of the parent mutant Tyr96Phe showed that
the 96Phe mutation increased the active site volume, allowing
for a higher mobility of substrates 1–3 in the binding pocket.
In addition, MD simulations identified multiple NACs with the
benzene ring either parallel or perpendicular to the haem
plane, overall producing only low to moderate enantioselectivi-
ties (Figure 3A, Figure S8–S13). In contrast, the 184Val/185Phe
mutant (Table 1, entry B) had a decreased active site volume
due to the bulky Phe side chain at position 185. This residue,
being located above the haem, pushed substrates 1–3 closer
to the haem, thus limiting the number of conformations (Fig-
ures 2 and S14–S17).
The numbers of pro-R and pro-S
conformations per mutant·sub-
strate complex were counted and
used as estimates for the ee values.
A comparison of experimentally
derived ee values with predicted ee
values showed that the latter gen-
erally allowed a semiquantitative
prediction of enantioselectivities,
with deviations between experi-
ment and prediction ranging be-
tween 2 and 48% (Table 1). The
deviations might be a result of ste-
reochemical dynamic events such
as inversion or rotation of the radi-
cal as previously proposed for the
Scheme 2. Representation of
the basis for the prediction of
P450cam-RhFRed mutant en-
antioselectivities in the ben-
zylic oxidation of ethylmethyl-
benzenes 1–3. dH–O represents
the distance between the
ferryl oxygen atom and the
hydrogen atom of the sub-
strate, and aC–H–O the angle
between the ferryl oxygen
atom and the hydrogen and
carbon atoms of the sub-
strate. A conformation was
considered a near attack con-
formation (NAC) if dH–O was
less than 0.35 nm and aC–H–O
was between 135 and 1808.
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
428
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 3. Examples of substrate orientations in the binding pocket of
P450cam mutants explaining trends in enantioselectivity toward the ethyl-
methylbenzene derivatives 1–3. The haem compound I is coloured in red,
and substrates 1–3 in the near attack pro-R conformations are in blue,
whereas the pro-S conformations are green. A) Mutant 96Phe with 3 in pro-
R and pro-S conformations. B) Mutant 184Val/185Phe with 3 in pro-R and
pro-S conformations. C) Mutant 244Phe/247Leu with 2 in pro-R conforma-
tion. D) Mutant 244Phe/247Leu with 1 in pro-R and pro-S conformations.
E) Mutant 244Asn/247Leu with 2 in pro-S conformation. F) Mutant 244Asn/
Figure 2. Heat map representing distances dH–O and angles aC–H–O for A) the
pro-R, and B) the pro-S orientations of substrate 3 in mutant 184Val/185Phe.
NACs are considered to be inside the black frame (0–3.5 nm and 135–1808).
Colours from white to red to black represent low to high numbers of simula-
tion frames for the appropriate substrate orientation.
2
47Leu with 2 in pro-R and pro-S conformations.
In most NACs obtained with the 184Val/185Phe mutant, the
benzene ring of substrates 1–3 was orientated parallel to the
haem plane with the 185Phe residue providing additional sta-
bility. However, the enantioselectivities shown by the 184Val/
binding pocket decreased the mobility of substrates 1–3, ori-
entated the ethyl group close to the haem with its terminal
carbon pointing to the substrate entrance channel (residues
Phe87 and Phe96), and thereby enhanced its R selectivity (Fig-
ure 3C, Figure S18–S23).
1
85Phe mutant toward substrates 1 and 2 were comparable to
those shown by the parent, due to the fact that the ethyl
group was mobile. In contrast, the shape of substrate 3 result-
ed in a sharp increase in the ee from 7% (96Phe) to 74%
(
184Val/185Phe; Figure 2). The pro-S conformation of 3 in the
84Val/185Phe binding pocket indicated that the benzene ring
Oxidation of the of the para-configured ethylmethylbenzene
derivative 1 resulted in moderate R selectivity. This might be
caused by para substitution not allowing constructive NACs,
due to limited space of the binding pocket in comparison with
the situations with substrates 2–3. The orientation of 1 ap-
peared to be more parallel to the haem plane, thus resulting
in an increased space for rotational movements of the ethyl
group, and hence in lower enantioselectivity (Figure 3D). Simi-
larly to the 244Phe/247Leu mutant, the 244Asn/247Leu mutant
revealed a narrow width of the binding pocket, albeit in a less
significant manner. In contrast with the 244Phe/247Leu
mutant, which exclusively showed R selectivity, we found that
the 244Asn/247Leu mutant displayed the highest S selectivity
toward substrates 1 and 2, and low R selectivity toward 3. As
a putative explanation of the opposite enantioselectivity on
a molecular basis, the benzene rings and ethyl groups of the
1
was orientated in a perpendicular position to the haem plane,
with the ethyl group stabilised in a small cavity close to
Thr252 (Figure 3B). Moreover, the high frequency of pro-S con-
formations confirmed the preferred orientation of 3 in the
binding pocket of mutant 184Val/185Phe (Figure 2B).
Whereas library III mutants (Met184/Thr185) such as the
1
84Val/185Phe mutant introduced changes above the haem
plane, library IV mutants (Leu244/Val247) introduced changes
at a-helix I, resulting in a changed width of the P450cam bind-
ing pocket. The library IV mutant 244Phe/247Leu exhibited
moderate to high R selectivity toward substrates 1–3 (Table 1,
entry C). In the majority of the NACs, the benzene ring of 1–3
appeared in an almost perpendicular orientation to the haem
plane. Simultaneously, the narrow width of the 244Phe/247Leu
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
429
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
NACs (Figure S24–S29) of substrate 1–3 were oriented parallel
to the haem plane, thus forcing substrates 1 and 2 into a pro-
S orientation through additional stabilisation by the 247Leu
residue (Figure 3E). The decreased enantioselectivity toward 3
was in agreement with the result from MD simulations, which
demonstrated that substrate 3 was less stabilised in a parallel
position, thus yielding both pro-R and pro-S conformations
Determination of P450 concentrations
Whole cells: Cells (0.5 g) resuspended in buffer (1.3m potassium
phosphate, pH 8) were diluted to a concentration of 90 mg wet
cells per mL. The cell suspension (180 mL) was transferred to a 96-
well plate, and sodium dithionite (0.03m, 20 mL) was added, fol-
[23]
lowed by incubation with a few grains of CORM-3. CO-difference
spectra were monitored in a range between 400 to 500 nm with
use of a plate reader (Tecan Infinite 200 series). All measurements
(
Figure 3F).
[9c]
were accomplished in triplicate.
Cell-free protein: Cells (5 g) were resuspended in buffer [sodium
phosphate (pH 7.2, 50 mm), KCl (100 mm), glycerol (0.4%, v/v)] to
a concentration of 180 mg wet cells/mL. The cells were disrupted
by ultrasonication (10 s burst, 20 s interval), and cell debris was re-
moved by centrifugation (20000 rpm, 15 min, 48C). The P450 con-
centration in cell-free extracts was examined in 1 mL cuvettes.
After addition of a few grains of sodium dithionite, CO gas was
slowly bubbled into the solution for approximately 15 s. CO differ-
ence spectra were recorded in a range between 400 to 500 nm
with use of a Cary 50 UV/Visible spectrophotometer. All measure-
Conclusion
In summary, we have identified enantioselective mutants from
complex P450cam[Tyr96Phe]-RhFRed libraries for the hydroxyl-
ation of the regioisomers of ethylmethylbenzenes 1–3. The in-
troduction of mutations 184Val/185Phe, 244Phe/247Leu and
2
44Asn/247Leu was demonstrated to influence enantioselectiv-
ity toward ethylmethylbenzenes to a high degree. To interpret
these experimental results on a structural basis, MD simula-
tions were used to estimate enantioselectivity. The results from
MD simulations were consistent with experimentally deter-
mined data and reflected the importance of conformational
changes and flexibility of mutant·substrate complexes in en-
forcing excellent enantioselectivity. Moreover, such data might
serve as a basis for enhancing the quality of general P450
mutant libraries to produce high-value chiral molecules, which
in turn might be useful for biotechnological applications.
[9c,23,24]
ments were accomplished in triplicate.
Whole-cell biotransformation reactions: Whole-cell biotransfor-
mation reactions were accomplished in buffer [sodium phosphate
(pH 7.2, 50 mm), KCl (100 mm), glycerol (0.4%, v/v)] with use of
1
80 mg wet cells per mL. Reactions were performed in sealed 7 mL
glass vials in a 2 mL total reaction volume at 208C and 250 rpm for
8 h. Substrates 1–3 were each dissolved in DMSO and added sep-
arately to cell suspensions to a final concentration of 1 mm (DMSO,
.4%, v/v). In addition, biotransformation experiments with racemic
4
0
alcohols (R,S)-4–6 were accomplished in the same fashion to inves-
tigate the degree of P450-catalysed stereospecific overoxidation
Experimental Section
(
Table S7–S9). Negative controls were conducted with whole cells
of E. coli BL21(DE3) harbouring the empty pET-14b vector and
treated in the same ways as cells expressing the P450 enzymes
Chemicals and consumables: All chemicals were of highest avail-
able purity and were obtained from Sigma–Aldrich and Alfa Aesar
unless stated otherwise. GC/FID gases were purchased from BOC
gases (Guildford, UK). Carbon monoxide was sourced from Sigma–
Aldrich. Starting materials, authentic standards and reagents were
purchased from Alfa Aesar and Sigma–Aldrich and used as re-
ceived. Solvents were of analytical or HPLC grade and were pur-
chased dried over molecular sieves where necessary.
(
Figure S4–S6). To address the abiotic removal of substrates 1–3
due to their volatile character, experiments with autoclaved cells
were additionally accomplished under the same conditions as de-
scribed (Table S10). Each experiment was performed in triplicate.
Sample preparation and instrumental analysis: After 48 h,
sample (100 or 500 mL) was extracted into tert-butyl methyl ether
(TBME; 500 mL), vortexed for 1 min and centrifuged (13300 rpm,
Strains, plasmids and media: The gene of the 96Phe mutant
P450cam-RhFRed was cloned into a pET-14b vector (Novagen) as
15 min). Acetic anhydride (Ac O, 2.5 mm) and a catalytic amount of
2
4-(dimethylamino)pyridine (DMAP, 0.02 mm) were added to the
[10,22]
described elsewhere.
Competent E. coli BL21(DE3) cells were
purchased from Merck4Biosciences UK. XL 1 Blue Supercompetent
and XL 10 Gold Ultracompetent cells were obtained from Strata-
gene (Agilent Technologies). Cells were transformed according to
the manufacturer’s instructions. Standard lysogeny broth (LB) and
organic phase, followed by shaking under ambient conditions for
1 h at 250 rpm. Excess of Ac O was removed by addition of dH O
2 2
(250 mL), the sample was mixed for 30 min at room temperature,
and the organic phase was separated (13300 rpm, 15 min) for anal-
ysis.
À1
LB agar plates were supplemented with ampicillin (100 mgL ) as
[22a,b]
selection marker.
Chiral GC analysis was performed with an Agilent system (Agilent,
Santa Clara, CA, USA) equipped with a flame ionisation detector
(FID). For grading of analytes, a CP-Chirasil-Dex CB column (25 m”
0.25 mm”0.25 mm, Agilent) was used, with helium as carrier gas
Recombinant expression of P450 mutants: For expression, E. coli
BL21(DE3) was transformed with pET14-b containing the appropri-
ate P450cam[Tyr96Phe]-RhFRed mutant gene. A single colony was
picked from an overnight LB plate, used to inoculate LB medium
À1
(flow rate 1.5 mLmin with a constant pressure of 15.5 bar). The
(
10 mL) and grown overnight at 308C and 250 rpm. The overnight
injector temperature was set to 2808C and the detector tempera-
ture to 2508C. The following temperature program was used:
1008C for 1 min, 108Cmin to 1208C, held for 10 min, 58Cmin
[22c]
culture (5 mL) was used to inoculate supplemented M9
(
À1
À1
500 mL) in a 2 L Erlenmeyer flask. The cells were grown at 378C
and 250 rpm to an optical density (OD600nm) of 0.8 to 0.9. Subse-
quently, protein expression was induced by addition of IPTG
to 1808C, held for 2 min. Chiral normal-phase HPLC was performed
with an Agilent system equipped with a G1379A degasser, a
G1312A binary pump, a G1329 well plate autosampler unit,
a G1315B diode array detector and a G1316A temperature-con-
trolled column compartment with a CHIRACEL OD-H (0.5 mm parti-
cle size, 4.6 mm diameter”250 mm). For grading of analytes,
(
0.4 mm) and 5-ALA (0.5 mm). The cells were cultivated at 208C
and 250 rpm for a further 16 h. Cells were harvested by centrifuga-
tion at 48C for 20 min and washed once with biotransformation
buffer.
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
430
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
hexane and ethanol were used in a 98:2 ratio with a flow rate of
Acknowledgements
À1
1
mLmin and an injection volume of 10 mL.
The research leading to these results received support from the
FP7-PEOPLE-2011-ITN under grant agreement no. 289217 (P4fifty
to A.E., L.G. and J.P.), the Innovative Medicines Initiative Joint Un-
dertaking under the grant agreement no. 115360 (Chemical man-
ufacturing methods for the 21st century pharmaceutical indus-
tries, CHEM21, to S.H.), the Centre of Excellence for Biocatalysis,
Biotransformations and Biocatalytic Manufacture (CoEBio3, to
P.P.K.) and the Royal Society Wolfson Merit Awards (to N.J.T. and
S.L.F.).
Molecular dynamics simulations: MD simulations of four mutants
[13]
of P450cam (PDB ID: 1PHG) —96Phe, 184Val/185Phe, 244Asn/
47Leu, and 244Phe/247Leu—in complex with substrates 1–3
were performed. The mutants were created by use of the muta-
2
[25]
genesis tool of the PyMOL program, and substrate molecules
were manually placed in the binding pocket by use of the PyMOL
[25]
structure visualisation software
with both hydrogen atoms of
the pro-stereocentre at a distance between 0.25 nm and 0.5 nm
from the compound I ferryl oxygen atom. The benzyl ring was
placed pointing in the direction of residues 87 and 96, which are
part of the main substrate access channel. To ensure extensive con-
formational sampling for each mutant/substrate pair, ten 30 ns mo-
lecular dynamics simulations with different initial orientations of
Keywords: benzylic oxidation
·
CÀH activation
· chiral
alcohols
enzymes
· directed evolution · enantioselectivity · P450
[26]
the substrate were performed, and the AMBER03 force field in
[27]
Gromacs version 5.0.4 was used. All water molecules were delet-
ed from the crystal structure. The haem and its cysteine ligand
were replaced in the crystal structure by a model of haem com-
pound I covalently bound to cysteine, and the corresponding force
field of the cysteine·compound I complex was used as previously
[1] B. Bühler, A. Schmid, J. Biotechnol. 2004, 113, 183–210.
[2] K. Chen, P. Zhang, Y. Wang, H. Li, Green Chem. 2014, 16, 2344–2374.
[3] a) R. Bernhardt, J. Biotechnol. 2006, 124, 128–145; b) S. K. Manna, S. Ma-
zumdar, Dalton Trans. 2010, 39, 3115–3123; c) M. T. Lundemo, J. M.
Woodley, Appl. Microbiol. Biotechnol. 2015, 99, 2465–2483.
[28]
described. The force fields for substrates 1–3 were generated by
use of the substrate structures derived from the PubChem data-
[
29]
[30]
[4] G.-D. Roiban, M. T. Reetz, Chem. Commun. 2015, 51, 2208–2224.
base
and subsequent energy minimisation by YASARA.
The
[5] a) A. Seifert, M. Antonovici, B. Hauer, J. Pleiss, ChemBioChem 2011, 12,
partial charges of the substrates were derived with the aid of the
1346–1351; b) R. Agudo, G.-D. Roiban, M. T. Reetz, ChemBioChem 2012,
[31]
R.E.D. Server with a RESP-A1B charge model. The force field was
1
3, 1465–1473.
[32]
built with the ANTECHAMBER module of AMBER 10 and convert-
ed into the GROMACS format. An octahedral water box of SPC/E
water with periodic boundaries of at least 1.2 nm from the protein
was used. Simulations were run at 300 K and 1 bar in a constant-
temperature, constant-pressure (NPT) ensemble. Pressure coupling
[
6] Y. Yang, J. Liu, Z. Li, Angew. Chem. Int. Ed. 2014, 53, 3120–3124; Angew.
Chem. 2014, 126, 3184–3188.
[7] a) J. T. Groves, P. Viski, J. Am. Chem. Soc. 1989, 111, 8537–8538; b) R.
Zhang, W. Y. Yu, T. S. Lai, C. Ming, Chem. Commun. 1999, 1791–1792; M.
Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßeler, R. Stürmer, T. Zelin-
ski, Angew. Chem. Int. Ed. 2004, 43, 788–824; Angew. Chem. 2004, 116,
[33]
was performed with a Parrinello–Rahman barostat,
and the
806–843; c) M. Landwehr, L. Hochrein, C. R. Otey, A. Kasrayan, J.-E. Bäck-
NosØ–Hoover coupling scheme was used for temperature coupling
[34]
vall, F. H. Arnold, J. Am. Chem. Soc. 2006, 128, 6058–6059; d) Y. Z. Chen,
F. Lie, Z. Li, Adv. Synth. Catal. 2009, 351, 2107–2112; e) A. Li, S. Wu, J. P.
Adams, R. Snajdrova, T. Li, Chem. Commun. 2014, 50, 8771–8774.
8] a) D. Filipovic, M. D. Paulsen, P. J. Loida, S. G. Sligar, R. L. Ornstein, Bio-
chem. Biophys. Res. Commun. 1992, 189, 488–495; b) P. J. Loida, S. G.
Sligar, Biochemistry 1993, 32, 11530–11538; c) P. J. Loida, S. G. Sligar,
Protein Eng. 1993, 6, 207–212; d) O. Sibbesen, Z. Zhang, P. R. Ortiz de
Montellano, Arch. Biochem. Biophys. 1998, 353, 285–296; e) S. G. Bell,
C. F. Harford-Cross, L.-L. Wong, Protein Eng. 2001, 14, 797–802.
9] a) A. Robin, G. A. Roberts, J. Kisch, F. Sabbadin, G. Grogan, N. Bruce, N. J.
Turner, S. L. Flitsch, Chem. Commun. 2009, 2478–2480; b) A. Robin, V.
Kçhler, A. Jones, A. Ali, P. P. Kelly, E. O’Reilly, N. J. Turner, S. L. Flitsch, Beil-
stein J. Org. Chem. 2011, 7, 1494–1498; c) P. P. Kelly, A. Eichler, S. Herter,
D. C. Kranz, N. J. Turner, S. L. Flitsch, Beilstein J. Org. Chem. 2015, 11,
1713–1720.
with a coupling constant of 0.5 ps.
Long-range electrostatic
interactions were treated by use of the particle-mesh Ewald
[35]
method. The initial velocities were randomly assigned. The LINCS
algorithm was applied to constrain all bonds containing hydrogen
[
[
[36]
+
atoms. Eighteen Na counter ions were added to maintain the
neutral charge of each system. Energy minimisation was performed
by the steepest descent method with positional restraints on pro-
À1
tein heavy atoms and a maximum force of 1000 kJmol nm. A 2 fs
time step was used, and coordinates were saved every 1000 steps
(
2 ps). All molecular systems were energy-minimised, gradually
heated to 300 K for 1 ns, and simulated for 30 ns. The final 25 ns of
each simulation were used for analyses. The selectivity prediction
[16]
method was based on previously published work that predicted
the selectivity of P450cam through MD simulations by measuring
the distance dH–O between the ferryl oxygen atom and the hydro-
gen atom of the substrate and the angle aC–H–O between the ferryl
oxygen atom and the hydrogen and carbon atoms of the substrate
[10] a) E. M. Gillam, A. M. Aguinaldo, L. M. Notley, D. Kim, R. G. Mundkowski,
A. A. Volkox, F. H. Arnold, P. Soucek, J. J. De Voss, F. P. Guengerich, Bio-
chem. Biophys. Res. Commun. 1999, 265, 469–472; b) E. M. Gillam, L. M.
Notley, H. Chai, J. J. De Voss, F. P. Guengerich, Biochemistry 2000, 39,
13817–13824; c) Q. S. Li, U. Schwaneberg, P. Fischer, R. D. Schmid,
(
Scheme 2). Pro-R orientations were assigned when aC–H–O for the R
Chemistry 2000, 6, 1531–1536; d) A. C¸ elik, R. E. Speight, N. J. Turner,
Chem. Commun. 2005, 3652–3654.
hydrogen atom was 135–1808 and dH–O was shorter than 0.35 nm
and shorter than the distance for the S hydrogen atom. All snap-
shots of the MD trajectories were classified on the basis of these
criteria and assigned variously to pro-R, pro-S, or nonproductive
orientations. For each mutant/substrate pair, the numbers of pro-R
and pro-S orientations from ten simulation runs were counted.
The percentages of pro-R and pro-S orientations were used as an
estimate for the experimentally determined product formation
[
11] M. T. Reetz, M. Bocola, J. D. Bocola, J. D. Carballeira, D. Zha, A. Vogel,
Angew. Chem. Int. Ed. 2005, 44, 4192–4196; Angew. Chem. 2005, 117,
4264–4268.
[12] Ł. Gricman, C. Vogel, J. Pleiss, Proteins 2015, 83, 1593–1603.
[13] T. L. Poulos, A. J. Howard, Biochemistry 1987, 26, 8165–8174.
14] T. L. Poulos, B. C. Finzel, A. J. Howard, J. Mol. Biol. 1987, 195, 687–700.
15] O. Gotoh, J. Biol. Chem. 1992, 267, 83–90.
16] a) M. D. Paulsen, R. L. Ornstein, J. Comp. Aided Mol. Des. 1992, 6, 449–
[
[
[
(
Table 1).
460; b) R. J. F. Branco, A. Seifert, M. Bude, V. B. Urlacher, M. J. Ramos, J.
Pleiss, Proteins 2008, 73, 597–607.
[17] T. C. Pochapsky, S. Kazanis, M. Dang, Antioxid. Redox Signaling 2010, 13,
1273–1296.
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
431
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
[
18] a) J. J. De Voss, O. Sibbesen, Z. Zhang, P. R. Ortiz de Montellano, J. Am.
Chem. Soc. 1997, 119, 5489–5498; b) K. J. French, D. A. Rock, D. A. Rock,
J. I. Manchester, B. M. Goldstein, J. P. Jones, Arch. Biochem. Biophys.
Bjelkmar, R. Apostolov, M. R. Shirts, J. C. Smith, P. M. Kasson, D. van der
Spoel, B. Hess, E. Lindahl, Bioinformatics 2013, 29, 845–854.
[28] A. Seifert, S. Tatzel, R. D. Schmid, J. Pleiss, Proteins 2006, 64, 147–155.
[29] Y. Wang, J. Xio, T. O. Suzek, J. Zhang, J. Wang, S. H. Bryant, Nucleic Acids
Res. 2009, 37, W623–W633.
[30] E. Krieger, G. Vriend, Bioinformatics 2014, 30, 2981–2982.
[31] a) C. I. Bayly, P. Cieplak, W. D. Cornell, P. A. Kollman, J. Phys. Chem. 1993,
97, 10269–10280; b) F.-Y. Dupradeau, A. Pigache, T. Zaffran, C. Savineau,
R. Lelong, N. Grivel, D. Lelong, W. Rosanski, P. Cieplak, Phys. Chem.
Chem. Phys. 2010, 12, 7821–7839; c) E. Vanquelef, S. Simon, G. Mar-
quant, E. Garcia, G. Klimerak, J. C. Delepine, P. Cieplak, F.-Y. Dupradeau,
Nucleic Acids Res. 2011, 39, W511–W517.
[32] D. A. Case, T. A. Darden, T. E. Cheatham, C. L. Simmerling, J. Wang, R. E.
Duke, R. Luo, M. Crowley, R. C. Walker, W. Zhang, AMBER 10, University
of California, San Francisco, CA, 2008.
[33] M. Parrinello, A. Rahman, J. Appl. Phys. 1981, 52, 7182–7190.
[34] S. NosØ, Mol. Phys. 1984, 52, 255–268.
2
002, 398, 188–197.
19] R. E. White, J. P. Miller, L. V. Favreau, A. Bhattachanyya, J. Am. Soc. 1986,
08, 6024–6031.
1
[
[
[
20] E. Fischer, Ber. Dtsch. Chem. Ges. 1894, 27, 2984–2993.
21] D. E. Koshland, Proc. Natl. Acad. Sci. USA 1958, 44, 98–104.
22] a) G. Bertani, J. Bacteriol. 1951, 62, 293–300; b) J. Sambrook, D. W. Rus-
sell, Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Labo-
ratory Press, Cold Spring Harbor, NY, 2001; c) H. Maike in Basic Methods
in Molecular Biology, Vol. 29 (Eds.: L. G. Davis, M. D. Dibner, J. F. Battey),
Elsevier, London, 1986, p. 336.
[
[
23] S. García-Gallego, G. J. L. Bernardes, Angew. Chem. Int. Ed. 2014, 53,
9
712–9721; Angew. Chem. 2014, 126, 9868–9877.
24] a) T. Omura, R. Sato, J. Biol. Chem. 1964, 239, 2370–2378; b) M. Geier, A.
Braun, A. Emmerstorfer, H. Pichler, A. Glieder, Biotechnol. J. 2012, 7,
1
346–1358; c) R. K. Gudiminchi, M. Geier, A. Glieder, A. Camattari, Bio-
[35] T. Darden, D. York, L. Pedersen, J. Chem. Phys. 1993, 98, 10089–10092.
[36] B. Hess, J. Bekker, H. J. C. Berendsen, J. G. E. M. Fraaije, J. Comp. Chem.
1997, 18, 1463–1472.
technol. J. 2013, 8, 146–152.
[
[
25] W. L. DeLano, available at: http://www.pymol.org, 2002.
26] Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P.
Cieplak, R. Luo, T. Lee, J. Caldwell, J. Wang, P. A. Kollmann, J. Comp.
Chem. 2003, 24, 1999–2012.
27] a) B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, J. Chem. Theory
Comput. 2008, 4, 435–437; b) S. Prouk, S. Pµll, R. Schulz, P. Larsson, P.
Manuscript received: October 14, 2015
Accepted article published: December 23, 2015
Final article published: February 16, 2016
[
ChemBioChem 2016, 17, 426 – 432
www.chembiochem.org
432
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim