Engineered Alkane-Hydroxylating Cytochrome P450BM3Exhibiting
Nativelike Catalytic Properties**
Rudi Fasan, Mike M. Chen, Nathan C. Crook, and Frances H. Arnold*
Cytochrome P450 enzymes (P450s) are exceptional oxygen-
ating catalystswith enormous potential in drug discovery,
chemical synthesis, bioremediation, and biotechnology.[2,3]
Compared to their natural counterparts, however, engineered
P450s often exhibit poor catalytic and cofactor coupling
efficiencies.Obtaining native-like catalytic proficiencies is a
mandatory first step towards utilizing the power of these
versatile oxygenases in chemical synthesis.
Cytochrome P450BM3(119 kDa, B. megaterium) catalyzes
the subterminal hydroxylation of long-chain (C12–C20) fatty
acids.Its high activity and catalytic self-sufficiency (heme
and diflavin reductase domains are fused in a single polypep-
tide chain)[2,4,5]make P450BM3 an excellent platform for
biocatalysis. However, despite numerous reports of the
heme domain being engineered to accept nonnative sub-
strates, including short-chainfattyacids, aromatic compounds,
alkanes, and alkenes,[6–8]reports of preparative-scale applica-
tions of P450BM3remain scarce.
P450BM3function is finely regulated through conforma-
tional rearrangements in the heme and reductase domains
and possibly also through hinged domain motions.[4,10]Hy-
droxylation of fatty acids occurs almost fully coupled to
cofactor (NADPH) utilization (93–96% depending on the
substrate).In the presence of nonnative substrates or when
amino acid substitutions are introduced, the mechanisms
controlling efficient catalysis in P450s are disrupted,
leading to the formation of reactive oxygen species and
rapid enzyme inactivation.High coupling efficiencies on
substrates whose physicochemical properties are substantially
different from the native substrates have not been achieved,
and coupling efficiencies ranging from less than 1% to 30–
40% are typical.[7,8]Strategies for addressing this “coupling
problem” are needed in order to take engineered P450s to
Selective hydroxylation of short alkanes is a long-standing
problem, for which no practical catalysts are available.In
an effort to produce P450BM3-based biocatalysts for selective
hydroxylation of small alkanes, we previously engineered this
enzyme to accept propane and ethane (35E11 variant).
Despite greater than 5000 total turnover (TTN) supported
in vitro, the utility of this catalyst remained limited because of
its poor in vivo performance (see below), which was mostly
due to the low efficiencies for coupling the product formation
to cofactor consumption (17.4% for propane and 0.01% for
Our goal was to engineer a P450BM3variant with native-
like activity and coupling efficiency towards a structurally
challenging, nonnative substrate (propane) and evaluate the
impact of these features on performance in preparative-scale
biotransformations. To this end, we used a domain-based
protein-engineering strategy, in which the heme, flavin
mononucleotide (FMN), and flavin adenine dinucleotide
(FAD) domains of the 35E11 variant were evolved separately
in the context of the holoenzyme, and beneficial mutations
were recombined in a final step (Figure 1). Previous work
suggested that mutations in the reductase and linker regions
can affect catalytic properties.[14,15]However, no systematic
engineering efforts had been undertaken to engineer the
complete 1048 amino acid holoenzyme.
Holoenzyme libraries outlined in Figure 1 were created
using random, saturation, and site-directed mutagenesis and
Figure 1. Outline of the domain engineering strategy used to improve
cytochrome P450BM3heme and reductase domains. HL=heme domain
libraries, RL=reductase domain libraries.
[*] Dr. R. Fasan, M. M. Chen, N. C. Crook, Prof. Dr. F. H. Arnold
Department of Chemistry and Chemical Engineering
California Institute of Technology
1200 California Blvd. MC 210-41, Pasadena, CA 91125 (USA)
[**] This work was supported by a Swiss National Science Foundation
fellowship to R.F., a U.S. NSF Fellowship to M.M.C. and by the U.S.
Army Research Office, ARO Contract DAAD19-03-D-0004. We thank
Dr. Christopher Snow for providing the homology model of P450BM3
reductase and Dr. Matthew W. Peters, Dr. Peter Meihnold, and Dr.
Marco Landwehr for helpful discussions regarding the biotransfor-
mations and for providing access to DasGip fermenter.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
? 2007 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimAngew. Chem. Int. Ed. 2007, 46, 8414–8418
screened for activity on apropane surrogate, dimethyl ether.
Positives were confirmed in a re-screen, purified, and
challenged with propane in sealed vials in the presence of a
cofactor regeneration system. As a cumulative measure of
both catalytic and coupling efficiency, improvement in total
turnover (moles of propanol produced per mole of enzyme)
was used as the sole selection criterion.
Measurement of the half-denaturation temperature of
35E11 heme domain demonstrated a considerable reduction
in its stability as a consequence of the 15 accumulated
mutations (T50=43.48 8C vs. 55.08 8C for wild type). We there-
fore subjected 35E11 to an initial thermostabilization step
(HL1), in which stabilizing mutations from a thermostable
P450BM3peroxygenasewere tested singly and in combina-
tionin the 35E11background
Information). Variant ETS8 (DT50=+5.18 8C, DTTNpropane=
?1250) showed the best combination of increased stability
with little decrease in TTN and was selected for further
directed evolution. Using ETS8 as parent, heme-domain
random mutagenesis libraries were generated by error-prone
PCR (HL2). Variant 19A12, with about twofold increase in
TTN (Table 1), was then used to create a pool of active-site
libraries (HL3) in which 17 positions along the substrate
channel (Figure 2a) and near the active site (Figure 2b) were
subjected individually to saturation mutagenesis. Further
improvements in propane-hydroxylating activity were ach-
ieved in multiple variants, including 11-3. Recombination of
the beneficial mutations from the active-site libraries (HL4)
led to variant 1-3. Further fine-tuning of the active site was
pursued with a series of recombination/site-saturation libra-
ries (HL5, see Supporting Information). From these libraries,
7-7 emerged as the most active variant, supporting 20500
turnovers with propane.
Meanwhile, two libraries were constructed in which
random mutations were targeted to the FMN- and FAD-
binding domains of 35E11 (RL1 and RL2, respectively).
Screening of more than 5000 members from each library led
to the identification of eight beneficial mutations (G443D,
V445M, T480M, T515M, P654Q, T664M, D698G, and
E1037G).These positions were further optimized by
saturation mutagenesis in a holoenzyme construct having
the 11-3 heme domain (RL3, RL4). Swapping the heme
domains this way serves to remove mutations whose benefi-
cial effect is solely dependent on the presence of the 35E11
heme domain. Improved 11-3-derived variants were found to
contain G443A, V445R, P654K, T664G, D698G, and E1037G
mutations and showed TTN between 16000 and 20000. In the
final step, a library containing the beneficial reductase
domain mutations was fused to the heme domain of variant
7-7 (L9). The most active variant isolated from this library,
P450PMOR2, supported more than 45000 turnovers and
produced 2- and 1-propanol in a 9:1 ratio. As we expected,
the increase in productivity strongly correlates with the
increase in coupling efficiency, which in the best variant
(P450PMOR2, 98.2%) reaches levels comparable to those
measured for wild-type in the hydroxylation of myristate
(88%), palmitate (93%), or laurate (96%).
The sequence of mutational events leading to P450PMO
generation reveals a continuous rearrangement of substrate-
channel and active-site residues (Table 1), presumably in
search of an optimal configuration for accommodating
propane. Additional beneficial mutations in the hydroxylase
domain include L188P and G443A. Leucine 188 is located
along helix F, which together with helix G forms a lid covering
Table 1: In vitro propane oxidation activities of most representative P450BM3variants.[a]
VariantLibrary Mutations versus 35E11[b]
Rate[c][equivmin?1] Coupling[d][%] Total turnovers
heme domain reductase domain
L52I, L188P, I366V
L52I, A74S, L188P, I366V
L52I, A74S, L188P, I366V
L52I, A74S, V184A, L188P, I366V
L52I, A74E, S82G, A184V, L188P, I366V
L52I, A74E, S82G, A184V, L188P, I366V, G443A
L52I, A74E, S82G, A184V, L188P, I366V, G443A
[a] Mean values from at least three replicates ?10% error. [b] Mutations in 35E11 are R47C, V78F, A82S, K94I, P142S, T175I, A184V, F205C, S226R,
H236Q, E252G, R255S, A290V, A328F, L353V, E464G, I710T. [c] Over the first 20 s. [d] Ratio between propanol formation rate and NADPH oxidation
rate in propane-saturated buffer.
Figure 2. a) Substrate channel and b) active-site residues targeted for
saturation mutagenesis mapped on the palmitate-bound structure of
P450BM3heme domain (PDB 1FAG). Heme (white) and fatty acid
(orange) are shown in space-filling mode.
Angew. Chem. Int. Ed. 2007, 46, 8414–8418? 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
the active site.Glycine 443 lies on a loop at the C-terminal
end of hairpin b4, which inserts into the active site.
Interestingly, the activity-enhancing substitutions in the
reductase domain are clustered in the same region in the FAD
domain (T664G, D698G, E1037G) and nearby linker to the
Information). Perturbation of electrostatic charge distribu-
tion appears to be a prevailing trend, suggesting a more
important role of these forces in P450BM3 function than
previously proposed.A smaller contribution was obtained
by mutating the FMN domain. This effect may reflect its
higher sensitivity to mutagenesis, as judged by the signifi-
cantly lower fraction of functional variants in the FMN
libraries compared to the FAD libraries (data not shown).
Chemical and thermal denaturation studies have shown that,
among the three cofactors, FMN is most weakly bound to the
A common strategy to reduce the prohibitive costs of
NADPH-driven biotransformations is the use of cofactor
regeneration systems.[9c,21]For bulk chemical transformations
such as alkane hydroxylation, these in vitro approaches are
not viable.The propane-hydroxylating P450 variants were
therefore evaluated in whole-cell biotransformations using
resting E. coli cells (Figure 3).
The expression levels ofthese variants in minimal medium
(initially less than 0.5% of total cell mass) were first
optimized to achieve 6–11% of total cell mass as soluble
P450 enzyme. Experiments were carried out in a 100-mL
fermenter using cell suspensions in nitrogen-free minimal
medium (supplemented with glucose) and a propane/air
mixture as substrate and oxidant feed. Under these condi-
tions, cell densities less than 1 gcdwL?1(typically 0.5–
mapin the Supporting
0.9 gcdwL?1; cdw=cell dry weight) were used to avoid
oxygen-transfer limitations. Activities of 80–120 Ug?1cdw
(where 1 U=1 mmolpropanolmin?1) were measured for
P450PMOR1 and P450PMOR2 in various E. coli strains (Fig-
ure 3a, Table 2). The experiment was repeated in a larger
fermenter (0.3 L, pH and dissolved oxygen control). A
suspension of P450-expressing DH5a cells was fed with a
1:1 mixture of pure oxygen and propane, and propanol
formation was monitored for up to 9 h (Figure 3b). Under
these conditions, very high activities (up to 180 Ug?1cdw)
were obtained.In comparison,
30 Ug?1cdw on n-nonene were reported for the natural
AlkB alkane hydroxylase system in both homologous (P.
oleovorans) and heterologous strains (E. coli).[23a]
At moderately higher cell densities (ca. 4 gcdwL?1),
propanol accumulated to a concentration of more than
15 mm over 4 h (Figure 4, upper panel). The improved
coupling efficiencies result in considerably extended periods
of whole-cell activity (6 vs. 0.5 h, Figures 3b and 4). To
investigate the possible causes of the decrease in productivity
over time, we monitored the biocatalyst concentration over
the course of the biotransformation (Figure 4, lower panel).
At the end of the experiment, approximately 52% of the
Table 2: In vivo propane oxidation activities of P450BM3variants.[a]
0.5 h 3 h
P450PMOR1 air (1:1)
P450PMOR2 air (1:1)
[a] Mean values from two biological replicates ?15% error. n.d.=not
determined. [b] At cell density=0.5–0.9 gcdwL?1. [c] Calculated from
the first hour of biotransformation.
Figure 4. Concentration of propanol during biotransformation of pro-
pane with DH5a cells expressing P450PMOR1 (&) and P450PMOR2 (~) at
medium cell density (4 gcdwL?1). In the lower panel, relative P450
concentration as determined from CO-binding difference spectra on
cell lysate; OD=optical density.
Figure 3. Whole-cell biotransformation of propane. a) Initial activities
of selected P450BM3variants in different E. coli strains using air/
propane (1:1) feed (pH 7.2, 258 8C). b) Time course of propane
biotransformation using recombinant DH5a cells using oxygen/pro-
pane (1:1) feed (pH 7.2, 258 8C). Product amount is given per gram cell
dry weight to facilitate comparison among variants. Control: no
propane in the gas feed.
? 2007 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimAngew. Chem. Int. Ed. 2007, 46, 8414–8418
initial P450PMOR2 was still correctly folded in the cells.
Control experiments using P450PMOR2-expressing cells and
propanol concentrations up to 30 mm showed no product
inhibition nor overoxidation to acetone, suggesting that host-
related rather than biocatalyst-dependent factors are limiting.
Indeed, 40–60% of the initially measured activity could be
restored by resuspending cells from the plateau phase (i.e.
after 4–6 h reaction) in fresh medium. In addition, the rate of
biocatalyst inactivation could be reduced by varying the
relative concentration of oxygen in the gas feed, with more
extended whole-cell activity periods obtained at a propane/
oxygen ratio of 4:1 compared to 1:1 (Table 2). Optimization
of this parameter as well as the availability of more robust
host strainsis expected to further enhance the whole-cell
productivity of this engineered P450BM3.
Overall, a domain-based directed evolution strategy has
enabled us to engineer a finely-tuned, multicofactor, multi-
domain enzyme to exhibit nativelike catalytic properties on a
substrate significantly different from the native substrate.
With this approach, we could use relatively small and targeted
libraries to identify beneficial mutations throughout the
enzyme, which were recombined to yield the most efficient
engineered P450 reported to date. This strategy should prove
useful for engineering other enzymes with multiple, interact-
ing functional domains. With high activity and coupling
efficiency for propane oxidation, P450PMOs could be used in
whole-cell biohydroxylation of propane at room temperature
and pressure with air as oxidant. Total activities and product
formation rates exceeding those obtained with naturally
occurring alkane monooxygenases on their native sub-
strates[23,24]were achieved in this first report of whole-cell
bioconversion of propane to propanol in E. coli.These
results open the door to considering P450-based oxidations of
short-chain alkanes, with promise for green conversion of
gaseous hydrocarbons into liquid fuels and chemicals.
Received: June 15, 2007
Published online: September 20, 2007
cytochrome P450 · directed evolution · protein engineering
Keywords: alkane oxidation · biotransformations ·
 a) M. Sono, M. P. Roach, E. D. Coulter, J. H. Dawson, Chem.
Rev. 1996, 96, 2841–2888; b) I. G. Denisov, T. M. Makris, S. G.
Sligar, I. Schlichting, Chem. Rev. 2005, 105, 2253–2277.
 V. B. Urlacher, S. Eiben, Trends Biotechnol. 2006, 24, 324–330.
 R. Bernhardt, J. Biotechnol. 2006, 124, 128–145.
 A. W. Munro, D. G. Leys, K. J. McLean, K. R. Marshall, T. W.
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,
 A. J. Warman, O. Roitel, R. Neeli, H. M. Girvan, H. E. Seward,
S. A. Murray, K. J. McLean, M. G. Joyce, H. Toogood, R. A.
Holt, D. Leys, N. S. Scrutton, A. W. Munro, Biochem. Soc. Trans.
2005, 33, 747–753.
 a) T. W. Ost, C. S. Miles, J. Murdoch, Y. Cheung, G. A. Reid,
S. K. Chapman, A. W. Munro, FEBS Lett. 2000, 486, 173–177;
b) W. T. Sulistyaningdyah,J. Ogawa,Q. S. Li,C.Maeda,Y.Yano,
R. D. Schmid, S. Shimizu, Appl. Microbiol. Biotechnol. 2005, 67,
556–562; c) D. Appel, S. Lutz-Wahl, P. Fischer, U. Schwaneberg,
R. D. Schmid, J. Biotechnol. 2001, 88, 167–171; d) A. Glieder,
E. T. Farinas, F. H. Arnold, Nat. Biotechnol. 2002, 20, 1135–
1139; e) Q. S. Li, J. Ogawa, R. D. Schmid, S. Shimizu, FEBS
Lett. 2001, 508, 249–252; f) T. Kubo, M. W. Peters, P. Meinhold,
F. H. Arnold, Chem. Eur. J. 2006, 12, 1216–1220.
 a) Q. Li, J. Ogawa, R. D. Schmid, S. Shimizu, Appl. Environ.
Microbiol. 2001, 67, 5735–5739; b) A. B. Carmichael, L. L.
Wong, Eur. J. Biochem. 2001, 268, 3117–3125.
 M. W. Peters, P. Meinhold, A. Glieder, F. H. Arnold, J. Am.
Chem. Soc. 2003, 125, 13442–13450.
 a) S. Schneider, M. G. Wubbolts, G. Oesterhelt, D. Sanglard, B.
Witholt, Biotechnol. Bioeng. 1999, 64, 333–341; b) R. J. Sowden,
S. Yasmin, N. H. Rees, S. G. Bell, L. L. Wong, Org. Biomol.
Chem. 2005, 3, 57–64; c) J. R. Falck, K. Y. Reddy, D. C. Haines,
K. M. Reddy, U. M. Krishna, S. Graham, B. Murry, J. A.
Peterson, Tetrahedron Lett. 2001, 42, 4131–4133; d) S. C.
Maurer, K. Kuhnel, L. A. Kaysser, S. Eiben, R. D. Schmid,
V. B. Urlacher, Adv. Synth. Catal. 2005, 347, 1090–1098. Among
these, only two examples involve a nonnative substrate, with the
highest TTN (12850) achieved by Maurer et al. on cyclohexane
over a 100-h period using an isolated P450BM3variant, a cofactor
regeneration system, and a cyclohexane/buffer biphasic system.
 a) D. C. Haines, D. R. Tomchick, M. Machius, J. A. Peterson,
Biochemistry 2001, 40, 13456–13465; b) M. B. Murataliev, R.
Feyereisen, Biochemistry 1996, 35, 15029–15037.
 a) M. A. Noble, C. S. Miles, S. K. Chapman, D. A. Lysek, A. C.
Mackay, G. A. Reid, R. P. Hanzlik, A. W. Munro, Biochem. J.
1999, 339, 371–379; b) M. J. Cryle, R. D. Espinoza, S. J. Smith,
N. J. Matovic, J. J. De Voss, Chem. Commun. 2006, 2353–2355.
 a) P. J. Loida, S. G. Sligar, Biochemistry 1993, 32, 11530–11538;
b) S. Kadkhodayan, E. D. Coulter, D. M. Maryniak, T. A.
Bryson, J. H. Dawson, J. Biol. Chem. 1995, 270, 28042–28048.
 a) J. A. Labinger, J. E. Bercaw, Nature 2002, 417, 507–514;
b) G. B. Shul?pin, G. Suss-Fink, L. S. Shul?pina, J. Mol. Catal. A
2001, 170, 17–34; c) I. Yamanaka, S. Hasegawa, K. Otsuka,
Appl. Catal. A 2002, 226, 305–315.
 P. Meinhold, M. W. Peters, M. M. Chen, K. Takahashi, F. H.
Arnold, ChemBiochem 2005, 6, 1765–1768; Wong and co-
workers have recently reported an engineered P450cam for
in vitro hydroxylation of propaneand ethane (F. Xu, S. G. Bell, J.
Lednik, A. Insley, Z. Rao, L.-L. Wong , Angew. Chem. 2005, 117,
4097–4100; Angew. Chem. Int. Ed. 2005, 44, 4029–4032) using a
reconstituted redox system containing P450cam/putidaredoxin/
putidaredoxin reductase in a 1:16:1 ratio.
 a) S. Govindaraj, T. L. Poulos, Biochemistry 1995, 34, 11221–
11226; b) O. Roitel, N. S. Scrutton, A. W. Munro, Biochemistry
2003, 42, 10809–10821.
 O. Salazar, P. C. Cirino, F. H. Arnold, ChemBiochem 2003, 4,
 Residues G443 and V445 are in the heme domain. The region
targeted for construction of FMN libraries comprises 32 residues
located at the C-terminus of P450BM3heme domain, which are
not included in the heme domain libraries (see the Supporting
 O. Pylypenko, I. Schlichting, Annu. Rev. Biochem. 2004, 73, 991–
 D. R. Davydov, A. A. Kariakin, N. A. Petushkova, J. A. Peter-
son, Biochemistry 2000, 39, 6489–6497.
 A. W. Munro, J. G. Lindsay, J. R. Coggins, S. M. Kelly, N. C.
Price, Biochim. Biophys. Acta Protein Struct. Mol. Enzymol.
1996, 1296, 127–137.
 a) U. Schwaneberg, C. Otey, P. C. Cirino, E. Farinas, F. H.
Arnold, J. Biomol. Screening 2001, 6, 111–117; b) S. C. Maurer,
H. Schulze, R. D. Schmid, V. B. Urlacher, Adv. Synth. Catal.
2003, 345, 802–810.
 W. A.Duetz,J. B.van Beilen,B.Witholt,Curr.Opin.Biotechnol.
2001, 12, 419–425.
Angew. Chem. Int. Ed. 2007, 46, 8414–8418 ? 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
 a) I. E. Staijen, J. B. Van Beilen, B. Witholt, Eur. J. Biochem. Download full-text
2000, 267, 1957–1965; b) T. Fujii, T. Narikawa, K. Takeda, J.
Kato, Biosci. Biotechnol. Biochem. 2004, 68, 2171–2177; c) M.
Kubota, M. Nodate, M. Yasumoto-Hirose, T. Uchiyama, O.
Kagami, Y. Shizuri, N. Misawa, Biosci. Biotechnol. Biochem.
2005, 69, 2421–2430.
 a) T. Furuto, M. Takeguchi, I. Okura, J. Mol. Catal. A 1999, 144,
257–261; b) S. G. Lee, J. H. Goo, H. G. Kim, J.-I. Oh, Y. M. Kim,
S. W. Kim, Biotechnol. Lett. 2004, 26, 947–950.
 Previous whole-cell bioconversions of gaseous alkanes have
relied on partial inhibition (i.e. blockage of methanol dehydro-
genase activity) of methanotrophic bacteria carrying multi-
component nonheme methane monooxygenases.Accumula-
tions of 5–8 mm methanol over periods of 40 h were reported.
 H. Li, T. L. Poulos, Nat. Struct. Biol. 1997, 4, 140–146.
? 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2007, 46, 8414–8418