Revisiting cytochrome P450-mediated oxyfunctionalization of linear and cyclic alkanes

Article abstract of DOI:10.1002/adsc.201400410

Cytochrome P450 monooxygenases (CYPs) of the CYP153 family catalyse terminal hydroxylation of n-alkanes. Alkane hydroxylating mutants of self-sufficient CYP102A1 have also been described. We evaluated two CYP153s (a three-component system and a fused self-sufficient CYP), wildtype CYP102A1 and nine CYP102A1 mutants, for the conversion of three cycloalkanes (C₆, C₇ and C₈) and three n-alkanes (C₆, C₈ and C₁₀) using whole cells (WCs) and crude cell-free extracts (CFEs). The aim was to identify substrate-enzyme combinations that give high product titres and space-time yields (STYs). Comparisons were made using total turnover numbers (TTNs) and turnover frequencies (TOFs) to normalize for CYP expression. Reactions were carried out using high enzyme and substrate concentrations compatible with high STYs. Under these conditions CYP102A1 and the double R47L,Y51F mutant, although not regioselective, performed better on all substrates in terms of product titres over 8 h, and thus STYs and TTNs, than heavily mutated variants that have been reported to give very high TOFs. CYP153A6, with its ferredoxin (Fdx) and ferredoxin reductase (FdR), emerged as the superior catalyst for conversion of n-alkanes. In addition to its excellent regioselectivity it also gave the highest final product titres and STYs in WC conversions of hexane and octane. Interaction with FdR and Fdx initially limited performance in CFEs, but with additional FdR and Fdx gave 1-octanol titres of 50 mmol·L_BRM^-1 and TTNs exceeding 12,000 over 18 h, rivalling results reported with self-sufficient CYPs. Selecting biocatalysts for application requires caution, since experimental conditions such as amount of substrate added and solubility as well as cofactor dependence and regeneration can have a profound effect on catalyst performance, while stability and efficiency with regard to cofactor usage (coupling efficiency) are at least as important as TOFs when high product titres and STYs are the target.

Full text of DOI:10.1002/adsc.201400410

FULL PAPERS
DOI: 10.1002/adsc.201400410
Revisiting Cytochrome P450-Mediated Oxyfunctionalization of
Linear and Cyclic Alkanes
Alizꢀ Pennec,a,b Cheri L. Jacobs,a,b Diederik J. Opperman,a,b
and Martha S. Smita,b,*
a
Department of Microbial, Biochemical and Food Biotechnology, University of the Free State, P.O. Box 339, Bloemfontein
9300, South Africa
Fax : (+27)-51-401-9376; phone: (+27)-51-401-2219; e-mail: smitms@ufs.ac.za
South African DST-NRF Centre of Excellence in Catalysis, c*change, University of Cape Town, South Africa
b
Received: May 14, 2014; Revised: August 29, 2014; Published online: November 21, 2014
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201400410.
Abstract: Cytochrome P450 monooxygenases (Fdx) and ferredoxin reductase (FdR), emerged as
(CYPs) of the CYP153 family catalyse terminal hy- the superior catalyst for conversion of n-alkanes. In
droxylation of n-alkanes. Alkane hydroxylating mu- addition to its excellent regioselectivity it also gave
tants of self-sufficient CYP102A1 have also been de- the highest final product titres and STYs in WC con-
scribed. We evaluated two CYP153s (a three-compo- versions of hexane and octane. Interaction with FdR
nent system and a fused self-sufficient CYP), wild- and Fdx initially limited performance in CFEs, but
type CYP102A1 and nine CYP102A1 mutants, for with additional FdR and Fdx gave 1-octanol titres of
À1
the conversion of three cycloalkanes (C6, C7 and C8) 50 mmol·LBRM and TTNs exceeding 12,000 over
and three n-alkanes (C6, C8 and C10) using whole 18 h, rivalling results reported with self-sufficient
cells (WCs) and crude cell-free extracts (CFEs). The CYPs. Selecting biocatalysts for application requires
aim was to identify substrate–enzyme combinations caution, since experimental conditions such as
that give high product titres and space-time yields amount of substrate added and solubility as well as
(STYs). Comparisons were made using total turn- cofactor dependence and regeneration can have
over numbers (TTNs) and turnover frequencies a profound effect on catalyst performance, while sta-
(TOFs) to normalize for CYP expression. Reactions bility and efficiency with regard to cofactor usage
were carried out using high enzyme and substrate (coupling efficiency) are at least as important as
concentrations compatible with high STYs. Under TOFs when high product titres and STYs are the
these conditions CYP102A1 and the double target.
R47L,Y51F mutant, although not regioselective, per-
formed better on all substrates in terms of product
titres over 8 h, and thus STYs and TTNs, than heavi- Keywords: alkane hydroxylation; C H activation;
ly mutated variants that have been reported to give CYP102A1 (P450 BM3); CYP153A6; self-sufficient
very high TOFs. CYP153A6, with its ferredoxin cytochrome P450 monooxygenase
À
Introduction
to water. Two families of CYPs that can even accom-
plish terminal hydroxylation of linear alkanes, a diffi-
Oxyfunctionalization of unactivated carbons in gener- cult and much sought after reaction, are the CYP52
al and in linear and cyclic alkanes specifically is family from alkane utilizing yeasts[6,7] and the CYP153
a problem that is currently receiving much attention family from bacteria.[8,9]
from both chemists and biochemists as is evident
Octane hydroxylation reactions using CYP153A6
from the number of recent reviews.[1–5] One class of have reached space-time yields (STYs) of up to
enzymes that is quite successful at these reactions are 0.8 g·LÀ1 hÀ1,[10] which although not yet feasible for
the cytochrome P450 monooxygenases (CYPs). Cyto- commercial production of a bulk chemical such as 1-
chrome P450s are heme-dependent enzymes that acti- octanol is higher than the minimum requirement of
vate and insert one oxygen atom from molecular 0.1 g·LÀ1 hÀ1 for the application of monooxygenases in
oxygen into C H bonds while the second is reduced the production of fine chemicals.[11] In these studies
À
118
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
Table 1. Published data on alkane conversion by CYP153s, CYP102A1 and some CYP102A1 mutants. Data summarized
here are for reactions that were carried out under analytical conditions using low enzyme and substrate concentrations and
short reaction times.
Octane
Cyclohexane
TOF [minÀ1]
Cyclooctane
TOF [minÀ1]
Enzyme
TOF [minÀ1]
Product ratio 1:2:3:4[a]
Ref.
[8]
CYP153A6 FdRFdx
CYP153A13 FdRFdx[b]
CYP153A13 RhFred
CYP102A1 WT
CYP102A1 1-12G
CYP102 A1 139-3
58
1
57
4
150
480
NA
22
95:5:0:0
100:0:0:0
100:0:0:0
0:17:40:43
5:82:11:3
1:61:20:17
NA
NA
10
3
0.75
NA
NA
14
NA
NA
NA
2
NA
NA
NA
230
[8]
[13]
[14,15]
[16]
[16]
[15]
[14]
CYP102A1 R47L,Y51F
CYP102A1 F87V,A328F
–:92:–:–[c]
NA
[a]
Ratio 1-octanol:2-octanol:3-octanol:4-octanol.
Not analyzed with its own FdR and Fdx, but with Fdx from Sphingomonas sp. HXN200 and FdR from Mycobacterium
[b]
sp. HXN1500.
[c]
92% 2-octanol reported, other products were not specified.
CYP153A6 was co-expressed from a single operon to- of buffer containing a purified CYP102A1 mutant
gether with its ferredoxin (Fdx) and ferredoxin reduc- (6.6 mM) and cyclohexane in a preparative biphasic
tase (FdR), required to transfer electrons from biotransformation (10-mL scale) to produce ca.
NADH to the heme iron. It is generally assumed that 72 mmol·LBRMÀ1 after 100 h with cofactor regeneration
electron transfer in such three-component systems by a NADPH-dependent mutant formate dehydro-
slows down catalysis compared to CYPs with an elec- genase. In the case of CYP153A6, whole cell (WC)
tron transfer partner fused to the heme domain, the biotransformations are currently deemed the most
so-called self-sufficient CYPs. The best studied exam- practical for preparative reactions, because it compris-
ple of the latter group is CYP102A1 (P450 BM3), es a three-component system and requires cofactor re-
a bacterial subterminal fatty acid hydroxylase.[12] In generation. In reactions reported thus far n-octane
order to overcome these shortcomings some CYP153s was added as a second phase with up to 300 mL
have been fused to the reductase domain from anoth- octane added to 1 mL biotransformation reaction
er natural self-sufficient CYP P450RhF.[9] The most mixtures. In one such a reaction 67 mmol octa-
À1
extensively studied of these self-sufficient CYP153s is nol·LBRM were produced within 24 h by a resting
CYP153A13RhFred.[13] The hydroxylation rate, ex- WC suspension containing 4.5 mM CYP153A6.[17]
pressed as turnover frequencies (TOFs) towards n-
CYP153s and the alkane hydroxylating CYP102A1
octane reported with the CYP153A13RhFred as puri- mutants listed in Table 1, which had been reported to
fied enzyme or in cell-free extracts (CFEs) is much give high TOFs with octane (1-12G and 139-3), cyclo-
higher than that for CYP153A13 in the three-compo- octane (F87V,A328F) and cyclohexane (R47L,Y51F),
nent system, but similar to the rate of CYP153A6 in have never been compared in a single study. In this
the three-component system (Table 1). Although study we set out to compare the two CYP153s (as
these rates are higher than the octane hydroxylation a wild-type three-component system and an artificial-
rates reported with wild-type CYP102A1, octane hy- ly fused self-sufficient CYP) with wild-type
droxylation rates reported for CYP102A1 mutants far CYP102A1 as well as nine CYP102A1 mutants (five
exceed these. Nevertheless, no CYP102A1 mutants previously described and four new combinatorial mu-
have been obtained to date that can perform regiose- tants; Table 2). These twelve CYPs were evaluated
lective terminal hydroxylation as is seen with for the biotransformation of three cycloalkanes (cy-
CYP153s. However, two of the mutants listed in clohexane, cycloheptane and cyclooctane) and three
Table 1 can produce 2-octanol with >80% selectivi- linear alkanes (n-hexane, n-octane and n-decane)
ty.[14,16]
using WCs and crude CFEs. Although regioselectivity
The rates compared in Table 1 were reported for is essential for the practical application of n-alkane
reactions carried out under analytical conditions with hydroxylations, the focus of this study was to compare
reactions mixtures containing relatively low enzyme these different CYPs in terms of their ability to ach-
and substrate concentrations and with excess ieve high product titres and STYs. Comparisons were
NAD(P)H supplied, thus avoiding the need for cofac- done using TTNs and TOFs in order to normalize for
tor regeneration. Such reaction conditions can, how- CYP expression. An important goal was to identify
ever, not be used for preparative scale biotransforma- the substrate–enzyme combinations that could give
tions. Maurer and co-workers[15] used equal volumes the highest STYs in order to use these in future stud-
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
119

FULL PAPERS
Alizꢀ Pennec et al.
Table 2. CYPs investigated in this study.
Enzyme
name
Parent
enzyme
Mutations
Cofactor
dependance
Ref.
Total
number
Amino acid substitutions
in binding pocket
[18]
[9]
153A6FdxFdR
153A13RhFred
102A1
139-3
1-12G
CYP153A6
CYP153A13
CYP102A1
CYP102 A1
CYP102A1
CYP102A1
CYP102A1
CYP102A1
CYP102A1
CYP102A1
CYP102A1
CYP102A1
1
–
–
11
17
2
2
2
–
–
–
NADH
NADPH
NADPH
NADPH
NADPH
NADPH
NADPH
NADH
NADPH
NADH
NADH
[19]
[20]
[16]
[15]
[14]
[21]
V78A
V78A, A82L & A328V
R47L, Y51F
47-51
87-328
F87 V, A328F
102A1NADH
139-3C1
139-3C1NADH
87-328NADH
139C2NADH
13
15
4
R47L, Y51F, V78A
R47L, Y51F, V78A
F87V, A328F
R47L, Y51F, V78A, F87V, A328F
this study
this study
this study
this study
17
NADH
ies to investigate process parameters limiting oxyfunc- Mꢂller and co-workers also reported significantly de-
tionalization by CYPs. Biotransformations were creased expression levels for a CYP102A1 mutant
therefore carried out under conditions of high (19A12) carrying 18 mutations, which included the
enzyme and substrate concentrations compatible with A328F mutation.[21] WC concentrations for the rest of
the conditions required for preparative reactions set- the CYPs typically varied between 4 and 7 mM, but
ting a premium on high STYs. We also sought, by occasionally dropped below 4 mM. CYP recovery in
comparing a large number of enzymes on a range of CFEs was ca. 20% in the case of the CYP153s and on
substrates, to probe issues relating to substrate solu- average ca. 50% for CYP102A1 and its mutants so
bility and uptake, the type and regeneration of cofac- that CYP concentrations in CFE BRM generally
tors and the perceived superiority of one component varied between 1 and 5 mM (Figure 1). The relatively
systems.
low CYP recovery in the CFEs, especially in the case
of the CYP153s, might be improved through the use
Results and Discussion
CYP Expression
The CYPs listed in Table 2 were targeted because
they had previously been reported to give high TOFs
with alkanes while also being relatively regioselective
for hydroxylation at C-2 of n-alkanes in the case of
139-3, 1-12G and 87-328. New combinatorial mutants
were expected to have the same or improved activity
and regioselectivity as the parental mutants. NADH-
dependent versions of wild-type CYP102A1 and some
of the mutants were created to investigate the effect
of cofactor dependence on activity and specificity.
All twelve CYPs were cloned into pET28 and ex-
pressed in E. coli BL21-Gold (DE3) using auto-induc-
tion medium. CYP levels in WCs and CFEs varied
between CYPs and between experiments. Because the
biotransformation reaction mixture (BRM) was
always prepared using a constant wet biomass of
50 g·LÀ1 (ꢀ5 g·LÀ1 dry cell weight) final CYP concen-
trations in the BRM also varied between CYPs
(Figure 1) and between experiments. In the case of 1-
12G, 87-328 and 139-3C2NADH, CYP concentrations
Figure 1. CYP content in WC and CFE containing BRM.
Averages and standard deviations are for six and five differ-
in the WC BRM were always less than 2.5 mM. ent experiments, respectively.
120
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
Figure 2. Time course of whole cell (WC) biotransformations of (A) cyclohexane and (B) cyclooctane using 139-3C1NADH
to investigate the effect of different amounts of substrate (4 or 250 mL) added to 1 mL BRM. (C) Time course of cyclohex-
ane (4 or 250 mL) conversion by 139-3C1NADH using cell free extracts (CFE) containing BmGDH for cofactor regenera-
tion. Error bars are for duplicate reactions using WCs or CFEs from the same culture.
of a different buffer and requires further investiga- This enzyme was selected because it accepts both
tion.
NAD+ and NADP+ with similar turnover numbers.[22]
Amount of substrate added had opposite effects in
the case of cyclohexane and octane. In the case of
octane it was evident from the statistically designed
experiments that low substrate concentrations not
Reaction Conditions for Comparison
Preliminary WC experiments had indicated that 139- only resulted in lower hydroxylase activity but also in
3C1 and 139-3C1NADH were promising catalysts for over-oxidation to give significant amounts of the cor-
the conversion of cycloalkanes. Cyclohexane conver- responding diol and carboxylic acid. In the case of cy-
sion by 139-3C1 and octane conversion by clohexane time-course experiments with all three cy-
153A6FdRFdx were therefore used to investigate WC cloalkanes indicated that only cyclohexane was inhibi-
biotransformation conditions by statistically designed tory to WCs when 250 mL were added, while CFEs
experiments (see the Supporting Information, Section yielded higher product concentrations when 250 mL
A). Conditions which previously had given promising cyclohexane were used (Figure 2). It was thus decided
results for WC conversion of octane by to use 4 mL per 1 mL BRM in WC biotransformations
153A6FdRFdx[10,17] were used as starting point for of cyclohexane but 250 mL per 1 mL BRM in the case
these experiments. Amount of substrate added, the of WC reactions with the other five substrates as well
type of buffer and the buffer concentration emerged as for all CFE reactions. From the time-course experi-
as significant factors affecting activity in 2 h and 24 h ments reaction times of 8 h were selected to compare
reactions. Buffer (Tris-HCl, pH8, 200 mM) as well as the different enzymes in WCs, while 1 h and 8 h reac-
other parameters were chosen to select a set of condi- tions were selected for CFE comparisons. Even with
tions that would favour both enzyme–substrate com- very active enzymes such as 139-3C1NADH on cyclo-
binations. Although the difference in results obtained hexane, 1 h was still at the end of the linear range for
with 24-mL and 40-mL vials were not statistically sig- the calculation of TOFs, while 8 h measurements
nificant, reactions were performed in 40-mL vials to were used for comparison of WC and CFE conver-
ensure that no oxygen limitation occurred.
sions using TTNs. Particularly with whole cells there
The same conditions used for WC reactions were were most likely in many cases, i.e., cyclooctane con-
used for CFE reactions. Cellular metabolism of glu- version by 139-2C1NADH (Figure 2B), still residual
cose and glycerol was relied on to take care of cofac- activity left after 8 h, so that the TTNs calculated
tor regeneration in WC reactions. In the case of CFE after 8 h represent a lower boundary for comparison.
reactions, CFEs from E. coli BL21-Gold (DE3) ex-
All biotransformations were performed at 208C be-
pressing the Bacillus megaterium glucose dehydrogen- cause we had previously observed with WC biotrans-
ase (BmGDH) was added for cofactor regeneration. formations of octane by 153A6FdRFdx that this
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
121

FULL PAPERS
Alizꢀ Pennec et al.
lower temperature ensured that activity could be
maintained for longer so that higher product titres
could be achieved over 24 h.[10] CFE conversions of
cyclohexane and octane by 139-3C1NADH and
153A6FdRFdx, respectively, confirmed that 208C fav-
oured higher product titres and thus higher TTNs
over 8 h.
Cycloalkane Conversion
The cycloalkanes were very poor substrates for the
terminal
hydroxylases
153A6FdRFdx
and
153A13RhFred (Figure 3 and Supporting Information,
Figure S3) and thus this section focuses on the results
obtained with CYP102A1 and its mutants. Compari-
son of the TTNs for the different CYPs obtained with
WCs and CFEs, indicated that CFEs generally per-
formed better than WCs, despite the fact that activi-
ties in CFEs generally levelled off after 1 h (Support-
ing Information, Figure S3). TTNs calculated over 8 h
were thus in some cases up to five times higher with
CFEs than with WCs (Figure 3) with TTNs of more
than 8000 achieved with 1-12G and 47-51 on cyclo-
hexane and cycloheptane. The higher TTNs observed
with CFEs are most likely due to substrate uptake
being limited in WCs.
Activities towards cyclohexane and cycloheptane
were generally higher than towards cyclooctane with
TOFs of ca. 100 minÀ1 achieved with cycloheptane
during conversions with CFEs using 1-12G, 139-3C1
and 139-3C1NADH. The highest TOFs with cyclohex-
ane were ca. 80 minÀ1 obtained with CFEs containing
1-12G, 47-51 and 139-3C1, while the highest TOFs
with cyclooctane were only ca. 60 minÀ1 obtained with
1-12G, 139-3C1 and 139-3C1NADH. The TOF ob-
tained with 47-51 with cyclohexane was higher than
the
activity
previously
described
(14 minÀ1;
Table 1[15]). These authors had used 0.4 mM cyclohex-
ane for these assays, but had already pointed out the
increase in activity with increase in amount of cyclo-
hexane added. In a biphasic reaction using 6.6 mM
purified 47-51, they had obtained over 10 h ca.
20 mmolalcohol·LBRMÀ1, while we obtained with 3 mM
47-51 in CFEs over 8 h 26 mmolalcohol·LBRMÀ1. The 1-
12G and 139-3 mutants, which had not previously
been tested with cyclic substrates, both displayed
higher TOFs with cycloheptane and cyclooctane than
102A1. Wild-type 102A1 on the other hand displayed
higher TOFs with cyclohexane and cyclooctane than
Figure 3. Comparison of activities for the hydroxylation of
cyclohexane, cycloheptane and cyclooctane by twelve differ-
ent CYPs. (A) Total turnover numbers (TTNs) achieved
over 8 h in whole-cell (WC) biotransformtions. (B) TTNs
achieved over 8 h in cell-free extracts (CFE) biotransforma-
activities previously reported[14,15] (Table 1). This tions. (C) Turnover frequencies (TOFs) for CFE reactions
calculated for 1 h reactions. Initial CYP concentrations in
the BRM were used to calculate TOFs. Results are the aver-
ages for triplicate reactions using WCs or CFEs from the
same culture. The same trends were observed in two inde-
pendent experiments.
might again be explained by the fact that these au-
thors added less than 0.5 mM substrate instead of
a second phase as we used. Surprising though is that
the cyclooctane activity we observed with 87-328 was
the same as the activity with 102A1 while Weber and
co-workers[14] had reported that activity of the
122
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
F87V,A328F mutant towards cyclooctane was 115
times improved (Table 1). From our results it appears
as if the F87V,A328F mutations destroyed activity to-
wards cyclic substrates, since TOFs of 139-3C2NADH
towards all three cycloalkanes are less than 40% of
those of 139-3C1NADH
Although some enzymes such as 102A1 and 47-51
had lower TOFs, they were able to produce over 8 h
more cyclohÀe1xanol and cycloheptanol (up to
25 mmol·LBRM
) than 1-12G, 139-3C1 and 139-
3C1NADH (Supporting Informatrion, Figure S3), be-
cause their concentrations in CFEs were higher and
they also maintained activity over longer reaction
times. It is not clear why activity of the highly active
enzymes almost ceased after 1 h. This levelling off in
activity after 1 h is, however, not strictly activity relat-
ed, since alcohol production from cyclooctane also
levelled off even though cyclooctanol concentrations
À1
after 1 and 8 h were between 5 and 10 mmol·LBRM
less than from cyclohexanol and cycloheptanol.
Mꢂller and co-workers also observed that heptane hy-
droxylation activity of two highly active CYP102A1
mutants levelled off after 1 h.[21] They had used puri-
fied enzymes and lower enzyme and substrate concen-
trations so that activity levelled off when only
À1
0.6 mmol·LBRM heptanols (mixture of regioisomers)
had been produced.
Linear Alkane Conversion
All the enzymes were also tested for the conversion
of n-hexane, n-octane and n-decane. Enzymes
153A6FdRFdx, 153A13RhFred, 1-12G, 139-3 and 87-
328 were included in this study because of their previ-
ously reported activities towards linear alkanes
(Table 1), but activity data have not been reported for
all of them on all these substrates. For linear alkanes
regiospecific hydroxylation is an important require-
ment. Regiospecificities of the previously described
enzymes were as expected and changing the cofactor
specificity of 102A1 and its mutants did not affect the
Figure 4. Regioselectivity of (A) hexane, (B) octane and (C)
decane hydroxylation by twelve different CYPs. Data from
8 h reactions with cell free extracts were used to calculate
regioselectivities. Results are the averages for triplicate reac-
tions using WCs or CFEs from the same culture. The same
trends were observed in two independent experiments.
regioselectivity (Figure 4). The new combinatorial octanol, respectively, with TTNs of up to 4000 (Sup-
constructs 139-3C1 and 139C1NADH displayed for all porting Information, Figure S4 and Figure 5). The
three n-alkanes the same regioselectivity as 139-3, 102A1s in general, and specifically mutants 47-51 and
while 139-3C2NADH displayed the same strong pref- 139-3C1, also performed very well in WC biotransfor-
erence for hydroxylation at the 2-position as the pre- mations of n-hexane achieving TTNs of more than
viously described 87-328 and 1-12G. These results 4000. However, 102A1 and its mutants gave com-
again confirm that mutation of Ala at position 328 to pared to 153A6FdRFdx relatively poor conversions in
a larger residue forces hydroxylation to the 2-posi- WC biotransformations of n-octane, while WC bio-
tion.[14] It is unfortunate that all these 2-specific en- transformation of n-decane yielded very poor results
zymes, except 87-328NADH were expressed at rela- with all the enzymes. This probably points to sub-
tively low levels compared with 102A1 and its other strate uptake limitation that is more pronounced at
mutants.
In WC reactions 153A6FdRFdx performed very
lower water solubilities.
In reactions with CFEs, TOFs were generally lower
well over 8 h with both n-hexaneÀa1nd n-octane pro- than observed for cycloalkanes, with the exception of
ducing up to 36 and 23 mmol·LBRM 1-hexanol and 1- 47-51 which displayed very high activity with n-
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
123

FULL PAPERS
Alizꢀ Pennec et al.
hexane. In the case of n-hexane and n-octane, 139-3
and its two derivatives 139-3C1 and 139-3C1NADH
produced little additional products between 1 and 8 h
(Supporting Information, Figure S4). As in the case of
the cycloalkanes this is probably an indication that
these enzymes have very high activities, but activities
could not be maintained. Interestingly, although
TOFs with n-decane for 47-51, 139-3C1 and 139-
3C1NADH were approximately 30% of that for n-
octane, activity lasted longer, resulting in decane
TTNs over 8 h being 60 to 70% of octane TTNs. This
is an indication that, different from what we argued in
the case of the cyclooctane, deactivation of these
102A1 mutants is, after all, activity related.
This discrepancy between the cyclooctane and n-
decane results can be explained if these enzymes are
more uncoupled with cyclooctane as substrate than
with n-decane. In the case of 153A6FdRFdx,
153A13RhFred as well as 102A1 and its NADH
mutant there was still significant product formation
by CFEs between 1 h and 8 h on all three linear alk-
AHCTUNGTREGaNNUN nes (Supporting Information, Figure S4). Activity of
these enzymes could most likely be maintained for
longer, because their initial activities were slower and
they are less uncoupled than some of the 102A1 mu-
tants (uncoupling of these mutants is discussed later).
It is common practice to add catalase to the BRM
when purified CYPs are used, in order to scavenge
hydrogen peroxide produced by uncoupling.[15,21,23]
However, addition of catalase to CFE reactions car-
ried out with 153A6FdRFdx and 139-3C1NADH did
not have a significant effect. Hydrogen peroxide scav-
enging activity in CFEs of E. coli is probably suffi-
cient to eliminate hydrogen peroxide produced by the
CYPs.[24] However, it is still possible that hydrogen
peroxide or other reactive oxygen species cause im-
mediate damage to the enzyme thereby contributing
to the instability of the 102A1 mutants.
From the TTNs for CFE reactions it can be con-
cluded that the enzymes with the highest sustained ac-
tivity towards n-hexane are 47-51 and 102A1 followed
by 153A6FdRFdx, 1-12G and 87-328. When regiose-
lectivity is also considered 153A6FdRFdx is of inter-
est for 1-hexanol production, while 1-12G and 87-328
might be of interest for 2-hexanol production.
In the case of n-octane the TOFs for CFE reactions
indicate that 47-51 as well as the combinatorial mu-
tants 139-3C1 and 139-3C1NADH have the highest
activities with n-octane (33–36 minÀ1). However, their
regiospecificities are poor, making them uninteresting
for practical application. On the other hand
Figure 5. Comparison of activities for the hydroxylation of
hexane, heptane and decane by twelve different CYPs. (A)
TTNs achieved over 8 h in WC biotransformtions. (B) TTNs
achieved over 8 h in CFE biotransformations. (C) TOFs for
CFE reactions calculated for 1 h reactions. Initial CYP con-
centrations in the BRM were used to calculate TTNs and
TOFs. Results are the averages for triplicate reactions using 153A6FdRFdx and the newly created combinatorial
WCs or CFEs from the same culture. The same trends were
observed in two independent experiments.
mutant 139-3C2NADH gave somewhat lower TOFs
(19–22 minÀ1), but both displayed comparable or
better TTNs (ca. 4000) and excellent regioselectivity
to produce 1- and 2-octanol, respectively.
124
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
The superior performance of 153A6FdRFdx in
WCs towards n-hexane and n-octane, when compared
to all the other enzymes and when compared to its
relative performance in CFE, should be noted. Per-
formance of 153A6FdRFdx is probably relatively
better in WCs, because it functions as a three-compo-
nent system with separate FdR and Fdx and interac-
tion with the redox partners is improved in the con-
fined space within the cell.
Factors Limiting Activity: Substrate Solubility
The fact that TTNs over 8 h for WC reactions were
generally much lower than for CFE reactions, al-
though many CFE reactions levelled off after 1 h,
gives a very strong indication that substrate uptake
was limiting. This has previously been reported by
several authors using such hydrophobic sub-
strates.[25–27] When polymyxin B, which is known to
permeabilize E. coli cells through disruption of the
cell wall and membranes,[28] was added to the BRM
prior to biotransformations with linear alkanes using
153A6FdRFdx and 153A13RhFred, activity was en-
hanced in the case of n-octane and n-decane, but not
for n-hexane (Supporting Information, Figure S5).
Also the improvement in activity was more with n-
decane than with n-octane, again supporting the
notion that substrate uptake becomes more limiting
as solubility decreases. In the case of n-hexane the ad-
dition of polymyxin B caused reduced product forma-
tion, probably because the stress caused by cell per-
meabilization outweighed the advantage of improved
substrate uptake in this case where the substrate is
more water soluble.
Even with CFEs substrate solubility apparently
limits activity. When maximum activities of the differ-
ent 102A1 mutants towards the different cyclic and
linear alkanes were plotted against solubility data (ex-
perimentally observed and predicted by two different
algorithms[29,30]),
a
clear correlation emerged
(Figure 6). The correlation with the predicted solubili-
ty data is better, although the predicted solubility
values from the two models are quite different. Two
opposing solubility related factors probably influence
activity, so that activity initially increases as solubility
increases, but a maximum is reached at around the
solubility of cycloheptane and n-hexane. Thus when
solubility further increases the substrate becomes in-
hibitory (toxic) to the enzyme. This observation is
particularly intriguing when it is further considered
that activity increases as the ratio of substrate to
aqueous phase increases even when the smallest
Figure 6. Relationship between water solubility and maxi-
mum TOFs observed with 102A1 mutants towards the dif-
ferent substrates fitted to polynomic functions. Predicted
solubility data (A),[30] (B)[29] vcclab.org/web/alogps/and
(C)[30] experimental data. Large discrepancies exist between
experimentally determined and predicted solubility of
hexane.
amount of substrate added already exceeds the solu-
À1
bility limit, i.e., the 37 mmol·LBRM
cyclohexane
added to CFE of 139-3C1NADH was more than the still improved when more cyclohexane was added
solubility of cyclohexane in water, but activity was (Figure 2C).
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
125

FULL PAPERS
Alizꢀ Pennec et al.
Factors Limiting Activity: Cofactor Dependence and
Regeneration
therefore wasting reducing equivalents; 139-3 and 1-
12G reportedly display respectively only 22 and 37%
coupling to NADPH.[16] It has also been reported that
153A6FdRFdx
is
NADH-dependent,
while the gluconolactone produced by the GDH causes
153A13RhFred as well as 102A1 and all its original acidification.[34] The pH in the reactions of the 102A1
mutants are NADPH-dependent. Since NAD(H) mutants thus probably drops much faster causing de-
levels in E. coli WCs are higher than NADP(H) naturation of the CYPs. This problem was not antici-
levels[31] and attractive cofactor regenerating enzymes pated, because the BRM contains 200 mM Tris buffer
such as formate dehydrogenase and phosphite dehy- and product formation in CFE never reached concen-
drogenase are as WT enzymes NAD+-dependent,[32,33] trations of more than 27 mmol·LBRMÀ1. Additionally
it would have been convenient if all the enzymes used 153A6FdRFdx reactions carried out with CFEs from
in this study could have been NADH-dependent. In more concentrated cell suspensions demonstrated that
order to determine whether this would have been fea- cofactor regeneration by GDH caused inhibition of
sible and whether changing cofactor dependence will the CYP reaction after 1 h (Supporting Information,
change other properties of the enzyme, cofactor de- Figure S7). It thus appears that wild-type enzymes in
pendence of 102A1 and two of its mutants 139-3C1 E. coli are also wasting cofactors regenerated by the
and 87-328 was changed to NADH. Changing cofac- GDH, thus contributing to the acidification.
tor dependence did not significantly change substrate
specificity or regioselectivity. However, despite the
fact that an excess of the relevant added cofactors as Factors Limiting Activity: Redox Partners – Three
well as GDH were present in CFE reactions, Component System vs. a Single Protein
102A1NADH as well as 87-328NADH in most instan-
ces displayed lower TOFs than their original TTNs over 8 h for 153A6FdRFdx were similar with
NADPH-dependent counterparts, while performance WC and CFE reactions of n-hexane and n-octane,
of 139-3C1NADH was mostly the same as its parent. while 102A1 and its mutants generally performed sig-
Strangely, expression of 87-328NADH was consistent- nificantly better in CFE reactions (Figure 5). One
ly much higher than expression of 87-328 (Figure 1). possible explanation for this might be that the sepa-
Several E. coli transformants carrying 87-328 were rate FdR and Fdx might become too dilute when they
tested and all displayed low expression levels. It is not are no longer confined to the cell.[18] It was thus de-
possible to explain without further experiments and cided to test reactions with more concentrated cell
creating more mutants why the 87-328 mutants had suspensions. Initially in the presence of GDH the re-
lower expression levels and how the mutations in the actions stopped after 1 h as described above, and
CPR domain could alleviate the problem. One might there was no improvement in the TOF over 1 h (Sup-
speculate that it is due to altered codons that disrupt porting Information, Figure S8). When the GDH was
(in the case of the 87-328 mutations) and restore (in omitted the TOF remained the same but the reaction
the case of the CPR mutations) synchronization be- continued for 8 h. When in a new experiment addi-
tween transcription and translation.
tional CFE from a strain expressing only FdR and
In WCs the glucose and glycerol metabolisms of E. Fdx was added, the TOF increased significantly when
coli should be adequate for cofactor regeneration.[5] additional FdR and Fdx were added (Figure 7). The
Efforts to co-express various dehydrogenases using TOF of the reaction without additional FdR and Fdx
Duet vectors repeatedly resulted in significantly lower was lower than in the previous experiment (Support-
expression levels of the CYPs and this route was not ing Information, Figure S7), because reactions in this
pursued. In order to ensure that cofactor regeneration case were carried out without the addition of exoge-
in CFEs will be sufficient, CYP containing CFEs were nous NADH, thus relying solely on the NADH avail-
mixed with an equal volume of CFE from a strain ex- able from the cytosol. By using concentrated cell sus-
pressing the BmGDH. Comparing results for pensions, excluding GDH and adding additional FdR
153A6FdRFdx, 139-3C1 and 139-3C1NADH from re- and Fdx it was thus possible to nearly double the
actions carried out with and without GDH showed TTN of 153A6FdRFdx over 8 h. The TTN over 18 h
that in the case of the 102A1 mutants the added exceeded 12,000 (volumetric yield of 50 mmol
GDH significantly improved activity, but with LBRMÀ1·1-octanol and STY 0.4 g·LBRMÀ1 hÀ1) while the
153A6FdRFdx this was not the case (Supporting In- reaction was still not levelling off. Similar TTN have
formation, Figure S6). It thus appears that the 102A1 been obtained over 100 h (STY 0.02 g·LBRMÀ1 hÀ1) in
mutants with higher initial reaction rates become lim- a bioreactor conversion of cyclohexane with 47-51
ited by cofactor regeneration if the GDH is not pres- using purified CYP and a commercially available for-
ent, while the wild-type E. coli enzymes are sufficient mate dehydrogenase.[15] Bordeaux and co-workers[23]
for the slower 153A6FdRFdx. Additionally the 102A1 have recently reported
a TTN of 3247 (STY
mutants are probably much more uncoupled and 0.003 g·LBRMÀ1 hÀ1) for a 78 h reaction with purified
126
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
substrate specificity apparently changes so that 102A1
and the 47-51 mutant generally performed better in
terms of product titres, and thus STYs over 8 h, than
1-12G, 139-3 and combinatorial mutants created from
139-3. This is remarkable since 1-12G and 139-3 were
reported as promising alkane hydroxylases obtained
from several rounds of directed evolution, while WT
102A1 is often proclaimed as having negligible activi-
ty towards alkanes. Although substrate uptake gener-
ally limited WC reactions, regioselectivity, final prod-
uct titres and STYs obtained in WC conversions of n-
hexane and n-octane using 153A6FdRFdx revealed
this natural alkane hydroxylase as the superior cata-
lyst for conversion of linear alkanes, when compared
with 102A1 and its mutants. Although improved ex-
pression might still make 1-12G and 139-3C2NADH
attractive as 2-hydroxylases, poor coupling efficiencies
will always hamper the heavily mutated 102A1s, even
if they are to be used with an alternative cofactor re-
generation system that will not cause acidification.
However, even for 2-hydroxylation a CYP153 might
become the preferred catalyst, since Yang and co-
workers have recently created mutants of another
CYP153 (CYP153A7; P450pyr), which are >99% re-
gioselective and >95% enantioselective for C-2 hy-
droxylation of octane while maintaining >90% activi-
ty when compared with the wild-type CYP.[35]
Figure 7. Effect of additional FdR and Fdx on octane con-
version by CYP153A6FdRFdx when using CFEs from con-
centrated cell suspensions (500 gWCW·LÀ1). CYP containing
CFE (250 mL) and CFEs from FdRFdx expressing cells or
cells carrying empty vector (125 or 250 mL) were used to
prepare BRM (final volume 1.2 mL). No NADH was added.
TTNs were calculated using the initial CYP concentration
(4.1 mM).
The STY of 0.37 g·LBRMÀ1 hÀ1 and final product titre
À1
of 23 mmol·LBRM obtained with WC octane bio-
transformations using 153A6FdRFdx were higher
than those obtained in bioreactor studies with WCs of
E. coli expressing the diiron monooxygenase from
Pseudomonas putida GPo1 (STY 0.23 g·LBRMÀ1 hÀ1
and final total product titre 11 mmol·LBRMÀ1),[25] but
not as high as the STY of 1 g·LBRMÀ1 hÀ1 obtained in
deep well plates with an E. coli strain co-expressing
153A13RhFred, using a commercial isocitrate dehy- the outer membrane protein AlkL, which facilitates
drogenase for cofactor regeneration. These results the uptake of hydrophobic substrates by E. coli.[36]
demonstrate that it is possible to achieve with Co-expression of AlkL also allowed WC conversion
a three-component system TTNs and STYs rivalling of dodecane by the diiron monooxygenase which ac-
those obtained with self-sufficient CYPs, even in CFE cepts longer chain alkanes than CYP153A6. AlkL
reactions.
however did not affect WC activity towards hexane,
confirming our own results that permeabilization did
not improve CYP activity towards hexane. A draw-
back of WC biotransformations with the diiron mono-
oxygenase is that alcohols are over oxidised to the
Conclusions
CYPs have been extensively studied for hydroxylation corresponding acids.
of hydrocarbons such as linear and cyclic alkanes. Al-
Co-expression of AlkL with the diiron monooxyge-
though very high TOFs have been reported for some nase has demonstrated the potential of E. coli whole
CYP102A1 mutants, the STYs achieved with these re- cells to hydroxylate hydrophobic substrates. By com-
actions were too low to be of interest for synthetic ap- bining reaction engineering with catalyst engineering
plication, because relatively low CYP and substrate Schrewe et al. has managed to obtain within 17 h total
concentrations were used. We confirmed that activi- product titres of more than 200 mM in WC conver-
ties of both CYP153s and CYP102A1s towards cyclic sions of dodecanoic acid methyl ester.[37] It is there-
and linear alkanes benefit when relatively large vol- fore quite likely that such productivities can also be
umes of substrate, equivalent to what would be used achieved with the three-component 153A6FdRFdx
for a co-solvent, are added. Under these conditions through further catalyst and reaction engineering.
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
127

FULL PAPERS
Alizꢀ Pennec et al.
TACCCAGCCGACACGTCTTTTGCGTATC-3’. PCR reactions
consisted of 1X KOD Hot Start Polymerase buffer, 1.5 mM
MgSO4, 0.2 mM (each) deoxynucleoside triphosphates
(dNTPs), 0.02 U·mLÀ1 KOD Hot Start polymerase (Nova-
gen), 0.4 ng·mLÀ1 template DNA and 0.1 mM of both forward
and reverse primers. PCR conditions consisted of an initial
denaturation step (958C, 2 min), followed by 7 cycles of de-
naturation at 958C (20 s), annealing at 628C (10 s) and elon-
gation at 708C (15 s) and 22 cycles of denaturation at 958C
(20 s) and annealing and extension of the megaprimer at
708C (4 min) with a final extension of 10 min at 708C. Tem-
plate DNA was removed with DpnI digestion (10 U) at
378C overnight, whereafter the PCR products were purified
from a DNA agarose gel (Biospin gel extraction kit, Bio-
Flux) and transformed into E. coli TOP10 (Invitrogen).
Single colonies were selected and grown overnight in 5 mL
LB medium containing 30 mg·mLÀ1 kanamycin. Plasmid
DNA was isolated (Biospin plasmid DNA extraction kit,
BioFlux) and verified by DNA sequencing.
153A6FdRFdx did not initially perform as well in
CFE reactions as it did in WC reactions, while the
self-sufficient CYPs benefitted significantly when the
substrate uptake barrier was removed. This was prob-
ably because interactions with FdR and Fdx were not
optimal in CFEs. This problem was alleviated by the
addition of CFEs containing only FdR and Fdx to the
BRM, giving a STY of 0.4 g·LBRMÀ1 hÀ1 over 18 h and
a final 1-octanol titre of 50 mmol·LBRMÀ1. These re-
sults are in terms of final product titre and STY
better than any reported thus far for n-alkane conver-
sion by a self-sufficient CYP and point the way for
further improvement of the 153A6FdRFdx system in
both CFEs and WCs.
This study demonstrated that given the importance
of high product titres and STYs, efficient heterolo-
gous expression, stability (TTN) and efficiency with
regard to cofactor usage (coupling efficiency) out-
weigh TOFs derived from short reaction times, espe-
cially from kinetic assays, when comparing recombi-
nant CYPs for preparative biocatalysis. It also indi-
cates that one should tread with caution when com-
paring biocatalysts either based on published data or
in your own experiments, since experimental condi-
tions such as amount of substrate added and solubility
as well as cofactor dependence and regeneration can
have a profound effect on catalyst performance.
The gene encoding GDH from Bacillus megaterium
(BmGDH) was kindly provided by Dr. Dirk Holtmann
(Dechema, Germany) in pETDuet (Novagen) multiple clon-
ing site 2 (MC2): cloned using the NdeI and XhoI restriction
sites.
Heterologous Expression of Enzymes
For expression of the CYPs and BmGDH, E. coli BL21-
Gold(DE3) (Stratagene) was transformed with the relevant
plasmids and transformants selected on LB-plates contain-
ing 30 mg·mLÀ1 kanamycin or 100 mg·mLÀ1 ampicillin. Ex-
pression was performed by using ZYP-5052 auto-induction
medium[39] containing 1 mM 5-aminolevulinic acid and
0.05 mM FeCl3. Cells were cultured for 48 h at 208C, after
which they were harvested through centrifugation (6000ꢃg,
10 min) and 1 g (wet weight) resuspended in 10 mL or 2 mL
Tris-HCl buffer (pH 8, 200 mM). CFEs were obtained by
a single passage of the suspended cells through a One Shot
Cell Disrupter (Constant Systems) at 207 MPa. The soluble
fraction was separated from unbroken cells by centrifuga-
tion (20,000ꢃg, 20 min).
CYP450 concentrations were determined using CO-differ-
ence spectra[40] recorded in F8 Maxisorp Nunc-ImmunoTM
Modules. Samples (2ꢃ200 mL) of a given preparation were
added into wells in two different microtiter strips and re-
duced by adding a few grains of sodium dithionite. One well
of each sample was saturated with carbon monoxide. Spec-
tra were recorded between 400 and 500 nm using a Spectra-
max M2 Microtiter Plate Reader (Molecular Devices Cor-
poration). The A450–A490 difference was corrected for
a path length of 1 cm, and the CYP concentration was calcu-
lated using an extinction coefficient of 91 mMÀ1 cmÀ1.
Experimental Section
Cloning of Enzymes
Cloning of the CYP153A6 operon into pET28b(+) was as
previously described.[17] The complete open reading frame
of CYP153A13 fusion was synthesized (GenScript) with
added 5’ NdeI and 3’ HindIII restriction sites (supplied in
pMK). CYP153A13 fusion was sub-cloned into pET28b(+)
(Novagen) using these restriction sites.
CYP102A1 (P450 BM3) in pET28a(+) was kindly pro-
vided by Prof. Vlada Urlacher (then at Institute for Techni-
cal Biochemistry, University of Stuttgart, Stuttgart, Germa-
ny). The N-terminal heme-domains (first 1182 bp of ORF)
of 1-12G and 139-3C1 were synthesized by GenScript (sup-
plied in pUC57) and were used to construct 1-12G and 139-
3C1 by replacing these fragments in the complete open
reading frames of CYP102A1 in pET28a using restriction
sites BamHI-PmeI. Similarly, the 139-3 and 47-51 mutants
were created from CYP102A1 using the AflII-PmeI and
BamHI-AflII restriction sites, respectively. The 87-328 and
139-3C2 mutants were created using the “megaprimer” PCR
method[38] with CYP102A1 and 139-3C1 as templates and
Biotransformations
All chemicals were from Sigma–Aldrich and were used
without further purification. Biotransfromations were per-
formed in 40-mL amber glass vials. The cell suspensions de-
fined as the biotransformation reaction mixture (BRM) con-
tained, unless otherwise stated, a final concentration of
0.05 g wet weight cells·mLÀ1 (or CFE from a corresponding
cell suspension) and 100 mM of glucose and glycerol in
200 mM Tris-HCl buffer (pH 8). In the case of CFE reac-
primers:
102A1-F87V-F:
5’-CAGGAGACGGGTTAGTTA
CAAGCTGGAC-3’ and 102A1-A328F-R: 5’-GGAAAACGCAG
GAAAAGTTGGCCATAAG-3’. NADH-dependent variants were
similarly created by introducing the mutations R966D and
W1046S into the respective parent plasmids by using pri-
mers:
102A1-R966D-F:
5’-CTTCATACCGCTTTTTCTGA
CATGCCAAATCAGC-3’ and 102A1-W1046S-R: 5’-GAATTCT
128
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130

FULL PAPERS
Revisiting Cytochrome P450-Mediated Oxyfunctionalization
tions the BRM also contained, unless otherwise stated, CFE
from an equally concentrated cell suspension of a culture
expressing BmGDH. In cases where the GDH was omitted
GDH containing CFE was replaced with CFE from an
empty vector control strain. NAD(P)H (1 mM) was, unless
otherwise stated, added to CFE reactions. Four or 250 mL of
alkane were added to each vial containing 1 mL of BRM
before it was placed on an orbital shaker at 208C, 200 rpm,
oscillation amplitude 26 mm. Vials were removed at specific
time intervals for extraction. Reactions were stopped and
extracted using an equal volume of ethyl acetate containing
2 mM 2-decanol or 1-undecanol as internal standard. GC-
FID and GC-MS analyses were carried out on a Shimadzu
GC and Thermo Trace GC ultra chromatograph with DSQ
mass spectrometer, both equipped with a FactorFour VF-
5 ms column (60 mꢃ0.32 mmꢃ0.25 mm, Varian). Alcohol
concentrations were calculated using conversion factors de-
rived from standard curves of authentic samples (see the
Supporting Information, Section B).
[10] O. A. Olaofe, C. J. Fenner, R. K. Gudiminchi, M. S.
Smit, S. T. L. Harrison, Microb. Cell Fact. 2013, 12, 8.
[11] M. K. Julsing, S. Cornelissen, B. Bꢂhler, A. Schmid,
Curr. Opin. Chem. Biol. 2008, 12, 177–186.
[12] C. J. C. Whitehouse, S. G. Bell, L.-L. Wong, Chem. Soc.
Rev. 2012, 41, 1218–1260.
[13] M. Bordeaux, A. Galarneau, F. Fajula, J. Drone,
Angew. Chem. 2011, 123, 2123–2127.
[14] E. Weber, A. Seifert, M. Antonovici, C. Geinitz, J.
Pleiss, V. B. Urlacher, Chem. Commun. 2011, 47, 944–
946.
[15] S. C. Maurer, K. Kꢂhnel, L. A. Kaysser, S. Eiben, R. D.
Schmid, V. B. Urlacher, Adv. Synth. Catal. 2005, 347,
1090–1098.
[16] M. W. Peters, P. Meinhold, A. Glieder, F. H. Arnold, J.
Am. Chem. Soc. 2003, 125, 13442–13450.
[17] R. K. Gudiminchi, C. Randall, D. J. Opperman, O. A.
Olaofe, S. T. L. Harrison, J. Albertyn, M. S. Smit, Appl.
Microbiol. Biotechnol. 2012, 96, 1507–1516.
[18] E. G. Funhoff, U. Bauer, I. Garcꢅa-Rubio, B. Witholt,
J. B. van Beilen, J. Bacteriol. 2006, 188, 5220–5227.
[19] L. O. Narhi, A. J. Fulco, J. Biol. Chem. 1986, 261, 7160–
7169.
Acknowledgements
[20] A. Glieder, E. T. Farinas, F. H. Arnold, Nat. Biotechnol.
2002, 20, 1135–1139.
[21] C. A. Mꢂller, B. Akkapurathu, T. Winkler, S. Staudt,
W. Hummel, H. Grçger, U. Schwaneberg, Adv. Synth.
Catal. 2013, 355, 1787–1798.
Financial support by the University of the Free State, SASOL
Ltd and the South African Department of Science and Tech-
nology/National Research Foundation Centre of Excellence
in Catalysis is gratefully acknowledged. We also thank Sarel
Marais for technical assistance with GC analyses. The
CYP102A1 (BM3) gene was kindly provided by Prof. Vlada
Urlacher (currently at the University of Dꢀsseldorf, Germa-
ny) and the BmGDH gene by Dr. Dirk Holtmann (Dechema,
Germany).
[22] T. Nagao, Y. Makino, K. Yamamoto, I. Urabe, H.
Okada, FEBS Lett. 1989, 253, 113–116.
[23] M. Bordeaux, D. Girval, R. Rullaud, M. Subileau, E.
Dubreucq, J. Drone, Appl. Microbiol. Biotechnol. 2014,
DOI 10.1007s00253-014-5671-1.
[24] L. C. Seaver, J. A. Imlay, J. Bacteriol. 2001, 183, 7182–
7189.
[25] C. Grant, J. M. Woodley, F. Baganz, Enzyme Microb.
Technol. 2011, 48, 480–486.
References
[26] M. K. Julsing, M. Schrewe, S. Cornelissen, I. Hermann,
A. Schmid, B. Bꢂhler, Appl. Environ. Microbiol. 2012,
78, 5724–5733.
[27] D. Scheps, S. Honda Malca, S. M. Richter, K. Marisch,
B. M. Nestl, B. Hauer, Microb. Biotechnol. 2013, 6,
694–707.
[1] J. C. Lewis, P. S. Coelho, F. H. Arnold, Chem. Soc. Rev.
2011, 40, 2003–2021.
[2] M. Bordeaux, A. Galarneau, J. Drone, Angew. Chem.
2012, 124, 10870–10881; Angew. Chem. Int. Ed. 2012,
51, 10712–10723.
[3] M. Morikawa, Appl. Microbiol. Biotechnol. 2010, 87,
1595–1603.
[28] K. Schmçlzer, K. Mꢄdje, B. Nidetzky, R. Kratzer, Bio-
resour. Technol. 2012, 108, 216–223.
[29] I. V. Tetko, V. Y. Tanchuk, J. Chem. Inf. Comput. Sci.
2002, 42, 1136–1145.
[4] E. Roduner, W. Kaim, B. Sarkar, V. B. Urlacher, J.
Pleiss, R. Glꢄser, W.-D. Einicke, G. A. Sprenger, U.
Beifuß, E. Klemm, C. Leibner, H. Hieronymus, S.-F.
Hsu, B. Plietker, S. Lachat, ChemCatChem 2013, 5, 82–
112.
[5] M. Schrewe, M. K. Julsing, B. Bꢂhler, A. Schmid,
Chem. Soc. Rev. 2013, 42, 6346–6377.
[6] D. L. Craft, K. M. Madduri, M. Eshoo, C. R. Wilson,
Appl. Environ. Microbiol. 2003, 69, 5983–5991.
[7] T. Iida, T. Sumita, A. Ohta, M. Takagi, Yeast 2000, 16,
1077–1087.
[8] E. G. Funhoff, J. Salzmann, U. Bauer, B. Witholt, J. B.
van Beilen, Enzyme Microb. Technol. 2007, 40, 806–
812.
[30] T. J. Hou, K. Xia, W. Zhang, X. J. Xu, J. Chem. Inf.
Comput. Sci. 2004, 44, 266–275.
[31] B. D. Bennett, E. H. Kimball, M. Gao, R. Osterhout,
S. J. Van Dien, J. D. Rabinowitz, Nat. Chem. Biol. 2009,
5, 593–599.
[32] T. W. Johannes, R. D. Woodyer, H. Zhao, Biotech-
nology 2007, 96, 18–26.
[33] V. I. Tishkov, V. O. Popov, Biomol. Eng. 2006, 23, 89–
110.
[34] S. Q. Pham, P. Gao, Z. Li, Biotechnol. Bioeng. 2013,
110, 363–373.
[35] Y. Yang, J. Liu, Z. Li, Angew. Chem. Int. Ed. 2014, 53,
3120–3124.
[9] M. Kubota, M. Nodate, M. Yasumoto-Hirose, T.
Uchiyama, O. Kagami, Y. Shizuri, N. Misawa, Biosci.
Biotechnol. Biochem. 2005, 69, 2421–2430.
Adv. Synth. Catal. 2015, 357, 118 – 130
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
129

FULL PAPERS
Alizꢀ Pennec et al.
[36] C. Grant, D. Deszcz, Y.-C. Wei, R. J. Martꢅnez-Torres,
P. Morris, T. Folliard, R. Sreenivasan, J. Ward, P. Dalby,
J. M. Woodley, F. Baganz, Sci. Rep. 2014, 4, 5844.
[37] M. Schrewe, M. K. Julsing, K. Lange, E. Czarnotta, A.
Schmid, B. Bꢂhler, Biotechnol. Bioeng. 2014, 111,
1820–1830.
Lohmer, J. J.-P. Peyralans, J. Podtetenieff, S. Prasad, P.
Soni, A, Taglieber, S. Wu, F. E. Zilly, M. T. Reetz,
Appl. Microbiol. Biotechnol. 2008, 81, 387–397.
[39] F. W. Studier, Protein Expr. Purif. 2005, 41, 207–234.
[40] F. P. Guengerich, M. V. Martin, C. D. Sohl, Q. Cheng,
Nat. Protoc. 2009, 4, 1245–1251.
[38] J. Sanchis, L. Fernꢆndez, J. D. Carballeira, J. Drone, Y.
Gumulya, H. Hçbenreich, D. Kahakeaw, S. Kille, R.
130
asc.wiley-vch.de
ꢁ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 118 – 130