Rapid identification of cytochrome P450cam variants by in vivo screening of active site libraries

Article abstract of DOI:10.1016/j.tetasy.2004.06.053

The selection of cytochrome P450 enzymes from large variant libraries, and the subsequent use of these enzymes in preparative scale biotransformations, remains a formidable challenge due to the complexities of the associated electron transport systems. Here, a powerful approach for the generation and screening of P450_cam libraries for new function is presented that is both flexible and robust. A targeted library was generated wherein only the P450_cam active-site amino acids Y96 and F98 were fully randomized and biotransformations, using a novel P450_cam whole-cell system, were screened by GC-MS for the hydroxylation of diphenylmethane. One in 50 of the reactions screened, including 16 different variants, produced 4-hydroxydiphenylmethane with up to 92% conversion observed in the case of the Y96A variant. These results demonstrate a primary example of the screening of P450_cam libraries in a format that is compatible with extension to preparative scale reactions.

Full text of DOI:10.1016/j.tetasy.2004.06.053

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 15 (2004) 2829–2831
Rapid identification of cytochrome P450cam variants by
in vivo screening of active site libraries
Robert E. Speight,a Fred E. Hancock,b Chris Winkel,c Hanamanthsa S. Bevinakatti,d
Manish Sarkar,e Sabine L. Flitscha and Nicholas J. Turnera,*
aSchool of Chemistry, The University of Edinburgh, KingÕs Buildings, West Mains Road, Edinburgh, EH93JJ, UK
bSynetix, PO Box 1, Billingham, TS23 1LB, UK
cQuest International, Huizerstraatweg 28, 1411 GP Naarden, The Netherlands
dUniqema, Bebington, Wirral, Merseyside, L62 4UF, UK
eICI Paints, Wexham Road, Slough, Berkshire, SL2 5DS, UK
Received 2 June 2004; accepted 18 June 2004
Available online 11 September 2004
Abstract—The selection of cytochrome P450 enzymes from large variant libraries, and the subsequent use of these enzymes in pre-
parative scale biotransformations, remains a formidable challenge due to the complexities of the associated electron transport sys-
tems. Here, a powerful approach for the generation and screening of P450cam libraries for new function is presented that is both
flexible and robust. Atargeted library was generated wherein only the P450 cam active-site amino acids Y96 and F98 were fully ran-
domized and biotransformations, using a novel P450cam whole-cell system, were screened by GC–MS for the hydroxylation of diphen-
ylmethane. One in 50 of the reactions screened, including 16 different variants, produced 4-hydroxydiphenylmethane with up to 92%
conversion observed in the case of the Y96Avariant. These results demonstrate a primary example of the screening of P450
libraries in a format that is compatible with extension to preparative scale reactions.
cam
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
thermore, since enzymatic hydroxylation reactions are
in practice carried out using whole cell systems, it would
be desirable to screen for activity using in vivo rather
than in vitro systems.
The cytochrome P450 (CYP) enzyme family catalyzes
the regio and stereospecific hydroxylation of a wide
range of molecules, typically at nonactivated carbon
atoms.1 Such hydroxylations are particularly difficult
using traditional chemical methods, suggesting that
these enzymes could be highly useful biocatalysts. How-
ever, the application of CYP enzymes in biotransforma-
tions has been limited by (i) the requirement for
heterologous electron transport proteins that utilize
NAD(P)H in stoichiometric quantities and (ii) the sub-
strate specificity, which is not usually optimized for rel-
evant reactions. One solution for tailoring substrate
specificity involves random genetic mutation followed
by identification of proteins with the desired activity.
Screening large CYP variant libraries is challenging
due to the complexities of the required electron trans-
port systems and the lack of high-throughput screens
that are generally applicable, robust and powerful. Fur-
Successful reports of directed evolution have been lim-
2
ited to the self-sufficient protein CYP102 (P450BMÀ3
)
and a variant of CYP101 (P450cam) that circumvented
the need for NADH and associated electron transport
proteins by utilizing the peroxide shunt pathway.3 How-
ever, P450cam, which catalyzes the hydroxylation of
camphor 1 to 5-exo-hydroxycamphor 2 (Scheme 1), is
an attractive target for protein engineering due to the
O
O
OH
1
2
*
Scheme 1. P450cam native activity, the hydroxylation of camphor 1 to
5-exo-hydroxycamphor 2.
Corresponding author. Tel.: +44 1316504717; fax: +44
1316504717; e-mail: n.j.turner@ed.ac.uk
0957-4166/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2004.06.053

2830
R. E. Speight et al. / Tetrahedron: Asymmetry 15 (2004) 2829–2831
availability of an X-ray crystal structure4 and the diverse
set of reactions that P450cam variants, generated by site-
directed mutagenesis, can perform.5 Herein, we report
an efficient strategy for identifying P450cam variants
based upon generating fully randomized active site
libraries coupled with screening for product formation
through in vivo biotransformations.
camphor 2 in 61% yield after chromatographic purifica-
tion.10 These experiments suggested that any selected
P450cam variants could be immediately transferred to
preparative scale biotransformations without further
development, a significant advantage of the current
approach.
With a functional whole-cell catalyst in hand, we pro-
ceeded to devise a strategy for library generation and
screening. In order to be able to screen generically for
substrate to product conversions we elected to use
2. Results
To enable screening of variant P450cam activities present
in whole cells we devised an in vivo system containing
two different plasmids in each Escherichia coli cell
(Fig. 1). The auxiliary proteins putidaredoxin (Pd) and
putidaredoxin reductase (PdR) were expressed from
one plasmid and P450cam, or variants thereof, were ex-
pressed from the other with NADH supplied in vivo.
Previously, CYP in vivo catalytic systems have been
produced with CYP102,6 engineered fusion proteins
based on CYP1027 and co-expression of the separate
proteins,8 including a tricistronic plasmid for P450cam
expression and activity in E. coli.9 However, we envis-
aged that our approach would allow the facile and
high-efficiency introduction of P450cam plasmid libraries
into a host strain that already contained plasmids
encoding Pd and PdR to establish an efficient whole-cell
biocatalyst.
LC–MS/GC–MS. Such methods have
a
modest
throughput (ca. 150–300 samples/day) and hence are
best used for screening smaller libraries. Although fully
random approaches can lead to the generation of substi-
tutions at unpredictable positions, away from the active
site, there is a strong precedent for targeting only the
P450cam substrate binding site (Fig. 2). In previous re-
ports, the substrate specificity of P450cam has been
extended to molecules as diverse as large polycyclic aro-
matics and propane by site directed mutagenesis of F87,
Y96, F98, T101 and V247.4 Herein, a library of 400 pos-
sible variants of P450cam was produced in which the two
amino acids Y96 and F98 were simultaneously mutated
to each of the remaining 19 amino acids. Y96 and F98
are both proximal to the substrate binding site of
P450cam and a Y96A variant has previously been re-
ported to catalyze the hydroxylation of diphenylmeth-
ane 3 to 4-hydroxydiphenylmethane 4 (Scheme 2).11 In
order to validate the biotransformation system, library
generation and screening methodologies, the library
was screened against diphenylmethane 3 for the selec-
tion of the Y96A variant along with the possibility of
new or improved P450cam proteins.
XmnI
PmeI
NdeI
SacI
XmnI
rrnB terminator
p15A origin
AmpR
XmnI
Pd
P450cam
pRES27
pRES18
(4521 bp)
(6077 bp)
PdR
KanR
NcoI
pBAD promoter
BamHI
ColE1 origin
AraC
Lac promoter
SapI
BamHI
HindIII
BamHI SacI
Figure 1. The two plasmids involved in the E. coli P450cam whole-cell
catalyst. Pd and PdR were expressed from a Pseudomonas putida
genomic fragment cloned into pUC18 (New England Biolabs) to
produce pRES27. P450cam expression was induced with arabinose, in a
concentration dependent manner, under the control of the pBAD
promoter in a plasmid that is a hybrid of pACYC177 (New England
Biolabs) and pBAD-Thio (Invitrogen).
To ensure that the engineered whole-cell biocatalyst
could be used to screen variant libraries, the conversion
of camphor 1 to 5-exo-hydroxycamphor 2 by native
P450cam was carried out using growing cells of E. coli
DH5a [pRES18][pRES27] with GC–MS analysis.
Experiments were performed with substrate concentra-
tions up to 10mM and the effect of increasing P450cam
expression was evaluated at increasing arabinose con-
centrations. Up to 80% substrate conversion was
achieved in only 5h, with most reactions proceeding to
over 99% conversion after 20h. Performing the reaction
at 10mM (1.52gLÀ1 camphor) 1 yielded 5-exo-hydroxy-
Figure 2. Structure of selected P450cam substrate binding pocket
amino acids with camphor (red) and the randomized amino acids
(blue).
Atotal of 1056 biotransformations in 96-well microtitre
plates, each containing pRES27 and a distinct pRES18
library member, were analyzed by GC–MS. Atotal of
21 reactions were found to produce 4-hydroxydiphenyl-
methane 4, a hit rate of ca. 1 in 50. Those variants that
showed activity were subsequently grown on a 10mL
scale, to quantify the conversion of each reaction, fol-

R. E. Speight et al. / Tetrahedron: Asymmetry 15 (2004) 2829–2831
2831
The high hit rate and variety of mutations that enabled
the hydroxylation of diphenylmethane demonstrates
that combinatorial mutagenesis of active site amino
acids, coupled with an in vivo biotransformation system
and GC–MS/LC–MS screen, is a powerful method for
discovering useful P450cam variants. As this approach
is much faster than rational site-directed mutagenesis,
it should be possible to extend these experiments to
the identification of a variety of new activities in a much
shorter time. Targeting three or more amino acids cou-
pled with increased screening throughput, for example
using capillary array electrophoresis,12 should further
accelerate this discovery process.
OH
3
4
Scheme 2.
lowed by DNAsequencing to identify the specific
mutations.
The results shown in Figure 3 reveal that a range of var-
iants possessed the ability to hydroxylate diphenylmeth-
ane 3 including the expected Y96Amutant, which had
the highest activity. Phe at position 98 was frequently
maintained, especially in the high activity mutants,
and this was typically coupled with a small amino acid
(e.g., Gly) at position 96, possibly to help accommodate
one of the phenyl rings of the substrate. In addition, leu-
cine at position 98 was a common substitution, espe-
cially with the lower activity mutants. Aparticular
variant of interest had the substitutions Y96F and
F98G wherein the positions for space generation and
p–p stacking with the substrate are reversed as com-
pared to the Y96Aprotein. In addition, some variants
were selected more than once, including Y96C (3 times),
Y96S (3) and Y96M F98L (2).
Acknowledgements
We are grateful to the ICI group for a postdoctoral fel-
lowship (R.E.S.). We also thank the Wellcome Trust for
financial support.
References
1. For a recent review see: Urlacher, V.; Schmid, R. D. Curr.
Opin. Biotechnol. 2002, 13, 557.
2. (a) Lentz, O.; Li, Q.-S.; Schwaneberg, U.; Lutz-Wahl, S.;
Fischer, P.; Schmid, R. D. J. Mol. Catal. B: Enzym. 2001,
15, 123; (b) Glieder, A.; Farinas, E. T.; Arnold, F. H. Nat.
Biotechnol. 2002, 20, 1135; (c) Li, Q.-S.; Schwaneberg, U.;
Fischer, P.; Schmid, R. D. Chem. Eur. J. 2000, 6,
1531.
3. Joo, H.; Lin, Z.; Arnold, F. H. Nature 1999, 399, 670.
4. Poulos, T. L.; Finzel, B. C.; Howard, A. J. J. Mol. Biol.
1987, 195, 687.
100
5. (a) Jones, J. P.; OÕHare, E. J.; Wong, L.-L. Eur. J.
Biochem. 2001, 268, 1460; (b) Bell, S. G.; Stevenson, J.-A.;
Boyd, H. D.; Campbell, S.; Riddle, A. D.; Orton, E. L.;
Wong, L.-L. Chem. Commun. 2002, 490; (c) French, K. J.;
Rock, D. A.; Rock, D. A.; Manchester, J. I.; Goldstein, B.
M.; Jones, J. P. Arch. Biochem. Biophys. 2002, 398, 188.
6. (a) Schneider, S.; Wubbolts, M. G.; Sanglard, D.; Witholt,
B. Appl. Environ. Microb. 1998, 64, 3784; (b) Schneider, S.;
Wubbolts, M. G.; Sanglard, D.; Witholt, B. Tetrahedron:
Asymmetry 1998, 9, 2833; (c) Schwaneberg, U.; Otey, C.;
Cirino, P. C.; Farinas, E.; Arnold, F. H. J. Biomol. Screen
2001, 6, 111.
80
60
40
Ala
20
Cys
Asp
0
Asn
Po
Glu
s
i
t
His
i
o
n
Ile
7. Lacour, T.; Ohkawa, H. Biochim. Biophys. Acta 1999,
1433, 87.
8. Parikh, A.; Gillam, E. M. J.; Guengerich, F. P. Nat.
Biotechnol. 1997, 15, 784.
9
6
8
9
Lys
n
o
i
t
i
Phe
s
Po
Trp
9. Bell, S. G.; Harford-Cross, C. F.; Wong, L.-L. Protein
Eng. 2001, 14, 797.
Figure 3. Hydroxylation of diphenylmethane (DPM) 3 by P450cam
variants. The different colours of the bars are used for clarity to group
variants that have a common residue at position 98. Only those clones
that showed >5% conversion of 3 were sequenced and hence included
in this analysis. Atotal of 1056 clones was analyzed in order to ensure
that the library had a >99% chance of containing all 400 possible
variants.
10. The yield of 5-exo-hydroxycamphor was lowered by the
formation of 5-oxo-camphor in ca. 20% yield after 40h
incubation. However, after 19h, 97% of the substrate had
been consumed but less than 4% had been converted to 5-
oxo-camphor, as previously observed.9.
11. Fowler, S. M.; England, P. A.; Westlake, A. C. G.; Rouch,
D. A.; Nickerson, D. P.; Blunt, C.; Braybrook, D.; West,
S.; Wong, L.-L.; Flitsch, S. L. Chem. Commun. 1994, 22,
2761.
These results demonstrate a novel method for the gener-
ation and rapid screening of libraries of P450cam vari-
ants for new function that is both flexible and robust.
12. Reetz, M. T.; Kuhling, K. M.; Deege, A.; Hinrichs, H.;
Belder, D. Angew. Chem., Int. Ed. 2000, 39, 3891.