Whole-cell one-pot biosynthesis of azelaic acid

Article abstract of DOI:10.1002/cctc.201300787

Polymers benefit from the use of biogenic resources such as fatty acids. They enable easy access to valuable monomeric building blocks, which, in comparison to their exclusively fossil counterparts, lead to products with improved physicochemical properties. Monomers of special interest are medium-chain dicarboxylic acids, which are not easy to obtain by traditional chemical means. Previously, we established an in vitro pathway that combined a 9-lipoxygenase and a 9/13-hydroperoxide lyase, which enabled the conversion of linoleic acid via a hydroperoxy intermediate into 9-oxononanoic acid, the precursor of azelaic acid. Herein, we aimed for the further development of the multi-enzyme cascade, which included the oxidation of 9-oxononanoic acid and the establishment of a suitable whole-cell catalyst. A detailed investigation of the simultaneous in vitro reaction setup revealed that both lipoxygenase activation and the subsequent hydroperoxide lyase reaction depend on the hydroperoxide reaction intermediate. For the activation of lipoxygenase, the hydroperoxide lyase activity, therefore, has to be significantly reduced. In accordance with these observations, we established a suitable dual-expression system and we further demonstrated that endogenous E. coli redox enzymes are feasible to oxidize 9-oxononanoic acid to azelaic acid. The resulting whole-cell catalyst is, therefore, able to perform the direct bioconversion of linoleic acid into azelaic acid. The use of organic solvent as the second phase improved the overall performance of the E. coli host strain. The developed one-pot, single-step process afforded 29 mg L^-1 of azelaic acid within 8 h with a substrate conversion of 34 % and a selectivity of 47 %. Tiny polymer factories: An E. coli whole-cell catalyst is developed for the bioconversion of linoleic acid to azelaic acid, an important building block for biopolymers. The catalytic machinery of the prokaryotic host is equipped with two plant enzymes, a lipoxygenase, and a hydroperoxide lyase. Together with an endogenous oxidoreductase, this three-enzyme cascade reaction catalyzes the oxidative cleavage of the fatty acid substrate.

Full text of DOI:10.1002/cctc.201300787

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300787
Whole-Cell One-Pot Biosynthesis of Azelaic Acid
Konrad B. Otte, Jens Kittelberger, Marko Kirtz, Bettina M. Nestl, and Bernhard Hauer*
[
a]
Polymers benefit from the use of biogenic resources such as
fatty acids. They enable easy access to valuable monomeric
building blocks, which, in comparison to their exclusively fossil
counterparts, lead to products with improved physicochemical
properties. Monomers of special interest are medium-chain di-
carboxylic acids, which are not easy to obtain by traditional
chemical means. Previously, we established an in vitro pathway
that combined a 9-lipoxygenase and a 9/13-hydroperoxide
lyase, which enabled the conversion of linoleic acid via a hydro-
peroxy intermediate into 9-oxononanoic acid, the precursor of
azelaic acid. Herein, we aimed for the further development of
the multi-enzyme cascade, which included the oxidation of
and the subsequent hydroperoxide lyase reaction depend on
the hydroperoxide reaction intermediate. For the activation of
lipoxygenase, the hydroperoxide lyase activity, therefore, has
to be significantly reduced. In accordance with these observa-
tions, we established a suitable dual-expression system and we
further demonstrated that endogenous E. coli redox enzymes
are feasible to oxidize 9-oxononanoic acid to azelaic acid. The
resulting whole-cell catalyst is, therefore, able to perform the
direct bioconversion of linoleic acid into azelaic acid. The use
of organic solvent as the second phase improved the overall
performance of the E. coli host strain. The developed one-pot,
À1
single-step process afforded 29 mgL of azelaic acid within
9
-oxononanoic acid and the establishment of a suitable whole-
8 h with a substrate conversion of 34% and a selectivity of
47%.
cell catalyst. A detailed investigation of the simultaneous in vi-
tro reaction setup revealed that both lipoxygenase activation
Introduction
Biopolymers based on renewable resources have increasingly
desired building blocks for high-performance polyamides and
polyesters. However, their main production processes suffer
[
1]
drawn the attention of researchers and industry. This can be
attributed to the efforts to reduce global CO2 emission, to
evade the rising prices of fossil raw materials, and the unique
several disadvantages. Sebacic acid (C ), for example, is pro-
10
duced from castor oil by energy-consuming alkaline thermal
[2,3]
[6]
properties of polymers composed of biogenic monomers.
pyrolysis, which offers relatively poor yields. The production
[7]
Monomers of special interest in this regard are medium-chain
dicarboxylic acids that can be obtained from fatty acids, which
offer easy access because of their natural functionalization.
Dicarboxylic acids, together with aliphatic diamines, are the
main building blocks of polyamides. The properties of the re-
sulting polymers are greatly reliant upon the aliphatic chain
length of the monomers employed. An increase in the chain
length, for example, leads to reduced water uptake, decreased
of azelaic acid (C ) is based on the ozonolysis of oleic acid. As
9
a result of the instability of the ozonide intermediates, this
venturesome process is the only example of ozonolysis in in-
dustry but is applied because of the lack of adequate alterna-
tives. On that account, safer and more environmentally friendly
[8,9]
processes would be preferable.
Recently, several biotechnological alternatives were pub-
lished that demonstrate an enzymatic access to medium-chain
dicarboxylic acids. Song et al. published a multi-enzyme ap-
proach from which sebacic acid is produced from oleic acid.
The enzymatic cascade was based on the formation of a car-
bonyl moiety, its Baeyer–Villiger oxidation, and the subsequent
[
4]
specific gravity, and increased chemical resistance. Compared
to classic polyamides, such as PA 6.6, these improved proper-
ties render the resulting polymers high-performance materials
because new applications can be realized in which classic poly-
amides fall by the wayside. The use of renewables in the field
of biopolymers is, therefore, not only reasonable in terms of
[10]
cleavage of the arising ester by a lipase. Our recent attempt
aimed to produce 9-oxononanoic acid, a precursor of azelaic
[
5]
[11]
eco-efficiency but also functionality. Medium-chain dicarbox-
ylic acids (C –C ) are of special interest because they are the
acid. We demonstrated the successful hydroperoxidation of
linoleic acid by the lipoxygenase St-LOX1 from Solanum tuber-
osum followed by the successive cleavage of the intermediate
by the hydroperoxide lyase Cm-9/13HPL from Cucumis
8
12
+
+
[
a] K. B. Otte, J. Kittelberger, M. Kirtz, Dr. B. M. Nestl, Prof. Dr. B. Hauer
Department of Chemistry
[12–14]
melo.
The combination of lipoxygenase and hydroperoxide
Institute of Technical Biochemistry
Universitaet Stuttgart
Allmandring 31, 70569 Stuttgart (Germany)
E-mail: Bernhard.Hauer@itb.uni-stuttgart.de
lyase is an established enzyme system in the flavor and fra-
grance industry to produce C odor compounds, the so-called
6
[15]
green-leaf volatiles.
In contrast to the enzymes currently
+
used in industry, the enzymes deployed in our previous work
[
] These authors contributed equally to this work.
showed nearly exclusive C selectivity, which enabled the for-
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201300787.
9
mation of C compounds such as 9-oxononanoic acid.
9
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1003

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Lipoxygenase (LOX) is a non-heme iron dioxygenase that
catalyzes oxygen insertion into polyunsaturated fatty acids
such as linoleic acid, which contain a (1Z,4Z)-pentadiene motif.
Mechanistically, the reaction occurs by hydrogen abstraction at
the double allylic position C-11 followed by radical rearrange-
ment to either position C-9 or C-13. Plant lipoxygenases are
therefore practically classified into 9- and 13-lipoxygenases ac-
Results
Lag-phase investigation indicates the pivotal role of HPODE
in St-LOX1 activation
Lipoxygenases in combination with hydroperoxide lyases are
already used for the production of green-leaf volatiles. In previ-
ous work, we investigated whether this reaction setup is also
suitable for the production of 9-oxononanoic acid, a precursor
[16,17]
cording to the regioselectivity of the oxygen insertion.
In contrast to lipoxygenase, hydroperoxide lyase (HPL) is
a heme iron enzyme, which belongs to the CYP74 family of
the P450 macrofamily. The classification of HPLs is also based
on their regioselectivity: 13-HPLs, which are only active on
both (13S)-hydroperoxy-(9Z,11E)-octadecadienoic acid ((13S)-
HPODE), and 9/13-HPLs, which are active on (9S)- and (13S)-
[11]
of azelaic acid. We demonstrated that the established in vitro
St-LOX1 and Cm- or Cs-9/13HPL system is indeed able to pro-
duce 9-oxononanoic acid in 73% yield.
If we performed St-LOX1 in vitro experiments with low pro-
À1
tein concentrations (up to 26.6 mgmL ), we observed a signifi-
[
18,19]
HPODE.
cant lag time in the 235 nm diene assay, which has already
[20]
Herein, we investigate the simultaneous lipoxygenase and
hydroperoxide lyase reaction and transfer the previously re-
ported in vitro cascade into a whole E. coli cell (Scheme 1). The
been reported for other lipoxygenases (Figure 1A).
The
assay monitors the emerging conjugated diene system as it is
formed in HPODE by the action of a lipoxygenase or cleaved
by a hydroperoxide lyase. Other
studies report that the lag time
can be reduced or even averted
by applying high lipoxygenase
concentrations or by adding
HPODE, which in trace amounts
is believed to be responsible for
the initial lipoxygenase activa-
[20–22]
tion.
This behavior was veri-
fied for St-LOX1 by the addition
of HPODE prior to the start of
the reaction. As shown in
Figure 1, 50 pm of HPODE was
already sufficient to reduce the
lag time from 4 to 2 min. Like-
wise, the addition of HPODE led
to a 13-fold increase in the initial
velocity of the lipoxygenase re-
action, which indicates the pivo-
tal role of HPODE for the
St-LOX1 reaction.
Scheme 1. Whole-cell biotransformation of linoleic acid. The transformation consists of a three-step multiple-
enzyme cascade. In the first step, the lipoxygenase St-LOX1 catalyzes the hydroperoxidation of linoleic acid, which
is followed by the subsequent cleavage of the nascent hydroperoxy moiety by the action of a hydroperoxide
lyase Cs-9/13HPL to yield 9-oxononanoic acid. In the last step, 9-oxononanoic acid is oxidized by an endogenous
E. coli oxidoreductase to azelaic acid.
Lipoxygenase and hydroper-
oxide lyase compete for
HPODE
In previous work, we further
in vivo system provides the advantages that expensive cofac-
tors, necessary for the last step in the cascade, can be regener-
ated, that higher substrate concentrations are tolerated, and
that all reaction steps can be performed in one single step. We
show that lipoxygenase and hydroperoxide lyase compete for
the hydroperoxy intermediate and that a careful balance be-
tween both enzyme activities is mandatory. Furthermore, we
indicate that endogenous E. coli oxidoreductases are sufficient
to catalyze the oxidation of 9-oxononanoic acid to azelaic acid.
The resulting whole-cell catalyst is, therefore, able to perform
the direct bioconversion of linoleic acid into azelaic acid.
demonstrated that compared to the successive reaction setup,
the simultaneous approach gave only very poor yields of
9-oxononanoic acid with concurrent linoleic acid conversion if
St-LOX1 and either one of the hydroperoxide lyases Cs- or Cm-
[11]
9/13HPL were used. However, higher conversions and 9-oxo-
nonanoic acid yields were expected because of the direct con-
version of the intermediate (9S)-HPODE and the suppression of
side reactions because of the shorter dwell time.
The photometric assay has shown that the addition of
50 pm HPODE was sufficient to reduce the lag time of St-LOX1
and is, therefore, important for the activation of St-LOX1. Un-
fortunately, HPODE is also the substrate for the hydroperoxide
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1004

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Scheme 2. Inactive St-LOX1 and Cs-9/13HPL compete for 9-hydroperoxy
octadecadienoic acid. St-LOX1 is a) activated by trace amounts of HPODE.
However, in a simultaneous multiple-enzyme reaction setup with Cs-9/
13HPL, HPODE is b) removed from the reaction mixture by the action of the
hydroperoxide lyase. AS=active site of enzyme.
Adjustment of Cs-9/13HPL expression levels
The reduction of the hydroperoxide lyase activity in the simul-
taneous one-pot reaction setup improved the lipoxygenase ac-
tivity significantly and thus, enabled the coincident catalysis of
lipoxygenase and hydroperoxide lyase. The adjustment of Cs-
9/13HPL with St-LOX1 was further transferred to a whole-cell
system, in which the HPL expression levels had been reduced.
In the system used, St-LOX1 is expressed in the commercially
available pQE-expression system (Qiagen, Hilden, Germany).
The host strain E. coli SG13009 harbors an additional pREP4
plasmid for the constitutive trans-expression of the lacI re-
pressor gene, which prevents the leaky expression of target
genes driven by the strong T5 promoter on the actual expres-
[
23]
Figure 1. Photometric 235 nm diene assay using cell lysates. A) Lag time in-
sion vector pQE-30. A comparison of the origins of replica-
tion of both plasmids, ColE1 (high copy) for pQE-30 and p15A
vestigation of St-LOX1. Lag time is observed by using low concentrations of
À1
2
6.6 mgmL St-LOX1 (c). The addition of 50 pm of HPODE is sufficient to
(
low copy) for pREP4, reveals that the intracellular concentra-
reduce the lag time and to increase the initial velocity (c). E. coli lysate
without St-LOX1 was used as a negative control (c). B) Simultaneous
St-LOX1 and Cs-9/13HPL reactions. The reaction of single St-LOX1 was used
as a reference (c). The activity ratio of 20:1 St-LOX1 to Cs-9/13HPL nearly
halves the LOX activity (c). A ratio of 2:1 St-LOX1 to Cs-9/13HPL resulted
in an approximately 10-fold decrease in LOX activity (c).
2
tions of pQE-30 are higher by a factor of ꢀ10 . The prepara-
tion of a dual-expression vector based on pQE-30 should result
in similar expression levels of St-LOX1 and Cs-9/13HPL.
The integration of the whole Cs-9/13HPL expression cassette
into the low copy vector pREP4 should reduce the intracellular
concentrations of hydroperoxide lyase significantly and, there-
fore, enable the simultaneous intracellular lipoxygenase and
hydroperoxide lyase reaction (Figure S1). Interestingly, the spe-
lyase, the second enzyme in the cascade. Hence, the lipoxyge-
nase cannot be fully activated if HPODE reacts too quickly in
the subsequent hydroperoxide lyase reaction (Scheme 2).
To validate this theory, we performed the simultaneous
St-LOX1 and Cs-9/13HPL one-pot reaction with different con-
centrations of Cs-9/13HPL. The results of these experiments are
shown in Figure 1B. The activity of single St-LOX1 was used as
a reference. If the reaction was performed with a St-LOX1 to
Cs-9/13HPL ratio of 2:1, an 89% decrease in diene formation
was observed. A ratio of 20:1 only led to a 38% decrease in
diene formation (Table S1).
À1
cific Cs-9/13HPL activity expressed from pREP4 was 1.3 Umg ,
which was 3.2-fold higher than that expressed from the origi-
nal pQE-expression vector. However, the co-expression of
St-LOX1 and Cs-9/13HPL resulted in a 25-fold reduction of hy-
droperoxide lyase activity in the corresponding crude extracts
or a 6.3-fold reduction compared to the original pQE-expres-
sion construct (Figure 2A). The activity of both enzymes in the
crude extract of a dual-expression approach of St-LOX1 and
Cs-9/13HPL was confirmed by measuring the 9-oxononanoic
acid formation by GC–FID.
The formation of 9-oxononanoic acid and, therefore, activi-
ties of both St-LOX1 and Cs-9/13HPL in the simultaneous reac-
tion mixture were confirmed by GC with flame ionization de-
tection (FID).
Parameter evaluation of whole-cell biotransformations
The decrease of the hydroperoxide lyase activity by an adjust-
ment of the hydroperoxide lyase expression level enabled the
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1005

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
phase increased the overall productivity of the host strain. The
linoleic acid conversion and the product formations of azelaic
acid and 9-oxononanoic acid were higher for the two-phase
system. At 250 mm, a linoleic acid conversion of 56% was ob-
tained in the two-phase system, whereas the one-phase
system only led to 42%. The same effect was observed for the
product yield of azelaic acid. With the two-phase system, an
overall yield of 26% was observed in comparison to the one-
phase system, which yielded only 7%.
Dependence on the substrate concentration
In our in vitro system reported previously, the productivity was
limited to low substrate concentrations because of denatura-
tion of the employed enzymes. At a concentration of 5 mm in
the in vitro system, nearly no substrate conversion and only
very poor yields of 9-oxononanoic acid were detected.
To compare the in vivo system with the in vitro system re-
ported previously, we used a similar experimental setup to in-
vestigate the dependence of the substrate concentration. The
following experiments were performed in the two-phase setup
because higher conversions and yields were observed com-
pared those in the one-phase reaction. The results are shown
in Figure 3.
In the first 2 h of the whole-cell biotransformation of linoleic
acid, the reaction proceeded quite moderately. If a 250 mm
substrate concentration was used, after 2 h 32.6Æ2.1% of the
substrate was already converted and 62.2Æ2.9 mm of azelaic
acid was formed (24.9Æ2.1% yield). Longer reaction times did
not significantly improve the product formation. However, if
we used higher linoleic acid concentrations, a steady increase
in product formation was detectable throughout the whole re-
action time of 32 h. Interestingly, reactions with either 1 or
Figure 2. A) Photometric 235 nm diene Cs-9/13HPL activity assay. Single ex-
pression of Cs-9/13HPL from pREP4 in SG13009 (c). Dual expression of
Cs-9/13HPL from pREP4 and St-LOX1 from pQE-30 in SG13009 (c).
B) Comparison of the whole-cell biotransformations (dual expression of St-
LOX1 and Cs-9/13HPL) under single-phase (left) and two-phase conditions
5
mm substrate gave very similar yields of azelaic acid and
a steady increase in product formation. After 2 h, the 1 mm
process gave 91.7Æ3.4 mm and 200.6Æ8.0 mm after 32 h. The
5 mm approach yielded 85.0Æ3.6 mm azelaic acid after 2 h and
(right) using cyclohexane.
210.7Æ5.4 mm after 32 h. In the case of the 1 mm reaction
after 32 h, 48.7Æ2.2% linoleic acid was converted and in the
case of the 5 mm reaction, the conversion was 31.7Æ0.0%.
Consequently, this resulted in a better reaction performance
for lower substrate concentrations in terms of selectivity. With
the established whole-cell system, the formation of approxi-
mately 200 mm azelaic acid represented an upper threshold,
above which no additional azelaic acid was formed. For con-
centrations equal to or greater than 5 mm, the formation of
the precursor 9-oxononanoic acid was detectable (Figure 3).
This applies to all concentrations higher than 5 mm and in-
creased with higher substrate concentrations (Figures S3 and
S4). After 32 h with 5 mm linoleic acid, an additional 14.7Æ
1.9 mm of 9-oxononanoic acid was formed. The best reaction
performance in terms of productivity and selectivity was found
in the 1 mm approach, in which 33.8Æ1.1% linoleic acid was
converted to 157.6Æ7.3 mm azelaic acid (15.8Æ0.72% yield)
conversion of linoleic acid into 9-oxononanoic acid in the si-
multaneous one-pot reaction setup. Preliminary results showed
that the developed dual-expression system is indeed able to
produce both enzymes and thus, enables the whole-cell bio-
conversion of linoleic acid with resting cells. Investigations of
the product distribution revealed that azelaic acid was the
main product. We evaluated several parameters to determine
the optimal reaction conditions.
The optimal pH for the whole-cell bioconversion was found
to be pH 6 (Figure S2). We also investigated whether it is ad-
vantageous to perform the reaction in a two-phase system be-
cause of the low solubility of fatty acids in water and the easy
substrate and product recovery. The data obtained are shown
in Figure 2B. In the two-phase setup cyclohexane was used as
it is a commonly available organic solvent. We tested the sol-
vent dependence at two different concentrations, 250 mm and
À1
after 8 h. This results in a productivity of 29 mgL in 8 h with
1
mm. At both tested concentrations, the use of a second
a selectivity of 47%.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1006

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
tion of hydroperoxide lyase products, which are also
[30,31]
known to cause lipoxygenase inhibition.
We were able to demonstrate that a 20-fold reduc-
tion of hydroperoxide lyase activity is sufficient to
enable the coupled in vitro lipoxygenase and hydro-
peroxide lyase cascade reaction (Figure 1B). To estab-
lish a whole-cell system with reduced hydroperoxide
lyase activity, we cloned the Cs-9/13HPL expression
cassette into the low copy pREP4 vector. Surprisingly,
the single expression of Cs-9/13HPL from pREP4 re-
sulted in a higher HPL activity than that expressed
from the original expression construct in pQE-30.
However, the co-expression of St-LOX1 and
Cs-9/13HPL led to a 25-fold reduction of hydroperox-
ide lyase activity.
Especially in the case of multistep reactions, in
which cofactors have to be regenerated, whole-cell
[32–34]
catalysis is favored over free-enzyme systems.
For the best reaction performance we tested several
parameters in the whole-cell system. Changes of ex-
tracellular pH do not only influence the intracellular
pH but also the substrate and product solubility. The
optimum pH was determined to be 6, therefore, the
intracellular pH can be predicted to be 7.46 by the
Figure 3. Whole-cell biotransformations of linoleic acid with cells that express St-LOX1
and Cs-9/13HPL dependent of the substrate concentration (250 mm, 1 and 5 mm) over
the 32 h. *: Data point extrapolated.
[35]
Discussion
formula of Slonczewski et al.
The pH optima for
St-LOX1 and Cs-9/13HPL are reported to be 6.5 and 5.5, respec-
[
13,19]
In this study, we focused on the establishment of an E. coli
whole-cell system for the biotransformation of biogenic linoleic
acid into azelaic acid, an important polymer building block.
Our previous work indicated that the coupled lipoxygenase
and hydroperoxide lyase reaction can only be performed in
a successive rather than in a simultaneous manner, which is
a crucial requirement for whole-cell catalysis. To enable the si-
multaneous in vivo two-enzyme cascade, additional in vitro ex-
periments have been performed, which support the conclusion
that the hydroperoxide lyase competes with the lipoxygenase
for the HPODE intermediates, compounds mandatory for
lipoxygenase activation.
tively.
Hence, the optimal external pH was expected to be
lower, but the linoleic acid solubility rapidly decreases at lower
[36]
pH. To overcome solubility issues and to reduce the in vivo
amounts of substrate and product, we investigated the applic-
ability of a two-phase system. We chose cyclohexane as a non-
toxic, commonly used laboratory solvent without taking the
logPoct value into account. There is no strict correlation be-
tween maleficent solvent effects and specific enzyme activities,
[37,38]
which presupposes extensive solvent pre-screening.
The
use of a second phase resulted in higher conversions and
product yields.
The initial substrate concentration of linoleic acid had an im-
portant effect on the final product formation. If we used
250 mm of substrate, a maximum of 70 mm of product was
formed. This upper threshold was reached after 2 h and did
not increase significantly over the remaining reaction time.
Higher concentrations of 1 and 5 mm yielded approximately
90 mm of product after 2 h. At both concentrations, the final
product concentrations were further enhanced during the
course of the reaction to approximately 200 mm. At concentra-
tions equal to or higher than 5 mm, the accumulation of 9-oxo-
nonanoic acid was observed. After reaching a maximum of
200 mm azelaic acid, only the formation of additional
9-oxononanoic acid is observed (Figures S4 and S5).
The appearance of lag times in lipoxygenase catalysis has al-
[
21,22]
ready been described.
It was also shown that the lag time
can be reduced by adding the product of the lipoxygenase re-
action HPODE or by increasing the initial lipoxygenase concen-
[
20]
tration. The reduction of the lag phase explained that in lip-
oxygenase catalysis two different iron species are involved, of
2
+
which the inactive Fe -species has to be oxidized by HPODE
3
+
[24–26]
to a Fe ÀOH moiety for activation (Scheme 2).
As HPODE is crucial for lipoxygenase activation, its abolition
prior to lipoxygenase activation prevents the activation of the
enzyme. Similar observations were made by Lands et al., who
reported that HPODE was reduced by a glutathione peroxidase
in the presence of glutathione, which inhibited LOX activa-
The best reaction performance was found if we used
a 1 mm substrate concentration. After 8 h, 34% of linoleic acid
was converted with an azelaic acid yield of 16%, which corre-
[
27]
tion. However, the consequent need to lessen the HPL ex-
pression levels cannot be overstated, as excessive amounts of
HPODE are known to irreversibly inhibit hydroperoxide
À1
sponds to a product titer of 29 mgL . Notably, no metabolic
[
28,29]
lyases.
Inhibition-wise, the established in vivo system pro-
engineering of the production strain was undertaken at this
time. The knock-out of b-oxidation genes and genomic inte-
gration of the target genes still harbors great potential for the
fits from the fact that endogenous E. coli enzymes are capable
of oxidizing 9-oxononanoic acid. This reduces the concentra-
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1007

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
overall productivity of the host strain. Previously, we were able
Expression of St-LOX1 and Cs-9/13-HPL in E. coli
À1
to produce 94 mgL of 9-oxononanoic acid, the precursor of
The dual expression of St-LOX1 (pQE-30) and Cs-9/13HPL (pREP4)
was performed in E. coli SG13009. An overnight LB agar
azelaic acid, in a successive two-step reaction that employed
[
11]
cell lysates with 54% yield.
Despite the lower yield, the
À1
À1
(
30 mgmL kanamycin and 100 mgmL ampicillin) culture was
À1
whole-cell system is favored over the in vitro system. All three
reactions can be performed simultaneously rather than only
the first two in a successive manner. In addition, Buchhaupt
et al. reported a similar whole-cell yeast for the production of
used to inoculate an overnight culture LB (5 mL; 30 mgmL kana-
À1
mycin and 100 mgmL ampicillin). The culture was transferred to
À1
À1
LB media (400 mL; 30 mgmL kanamycin and 100 mgmL ampicil-
lin) and incubated at 378C (180 rpm) to an OD600 of 0.6. The flask
was cooled to 168C (180 rpm), induced with isopropyl b-d-1-thio-
galactopyranoside (IPTG; 0.1 mm) and expressed for 48 h. Single
St-LOX1 and Cs-9/13HPL expression was performed as previously
À1
C6 volatiles. The yeast system afforded 63.1 mgL of products
(
58% yield) by utilizing the coupled LOX and HPL reaction
[
39]
with C13 regioselectivity. In contrast to S. cerevisiae, E. coli is
[44]
reported.
favorable because the substrate linoleic acid is toxic towards
Cells were harvested by centrifugation (10820ꢁg, 48C, 30 min),
washed and frozen at À208C until further use. Disruption of the
cells and protein and DNA analysis were performed as described
yeast cells, whereas E. coli is known to tolerate at least
À1 [39–42]
5
gL .
[11]
previously.
Conclusions
Based on previous work on the establishment of an in vitro
multiple-enzyme cascade, we have developed a whole-cell bio-
catalytic process for the direct synthesis of azelaic acid. By
using E. coli as the host, a laboratory-scale process able to syn-
thesize azelaic acid from linoleic acid has been implemented
for the first time. Additionally, a straightforward use of two ex-
pression vectors and their balancing has been devised for the
facile expression of lipoxygenase and hydroperoxide lyase,
mandatory for the performance of the cascade in a one-pot re-
action. We are currently working on the genomic integration
of the pathway genes for robust expression and improvement
of the cofactor recycling system. With the optimization of the
overall cascade as well as metabolic engineering of the host to
improve the overall productivity, the enzymatic route to afford
azelaic acid as a polymer building block can be a competitive
and attractive biotechnological alternative.
Enzyme assay
The conjugated diene system of 9-hydroperoxy-(10E,12Z)-octadeca-
dienoic acid (9-HPODE) has an absorption maximum at 235 nm
[45,46]
(e=23000).
This feature was exploited to measure the forma-
tion and further reaction of 9-HPODE. Measurements were per-
formed by using an Ultrospec 3000 photometer (Pharmacia bio-
tech, Uppsala). A typical reaction was performed on a 1 mL scale
with linoleic acid (50 mm) in potassium phosphate buffer (100 mm,
pH 6). The reactions were normalized to the protein concentration:
À1
À1
1
7.43 mgmL for St-LOX1 and 11 mgmL for Cs-9/13HPL.
Biotransformations and analytics
The reactions were performed on a 400 mL scale in potassium
phosphate buffer (50 mm, pH 6). A typical reaction setup consisted
of linoleic acid (250 mm) and cells (50 mgmL , wet weight). The
Experimental Section
À1
The chemicals used were standard products from the following
suppliers: linoleic acid was purchased from Fluka (Buchs, CH); N,O-
bis(trimethylsilyl)trifluoroacetamide containing 1% v/v trimethyl-
chlorosilane and azelaic acid were purchased from Alfa Aesar
two-phase reactions were performed in the presence of additional
cyclohexane (80 mL). The reaction was started by the addition of
substrate and quenched with HCl (50 mL, 1m). The products were
extracted twice with methyl t-butyl ether (400 mL). The organic
phases were combined, evaporated, and derivatized with N,O-
bis(trimethylsilyl)trifluoroacetamide containing 1% v/v trimethyl-
chlorosilane (40 mL). The reactions were analyzed by GC–FID (Shi-
madzu GC-2010 Plus, AOC-20i/s) with a DB-5 column (Agilent,
(
Ward Hill, US). Soybean lipoxidase (Glycine max) was purchased
from Sigma–Aldrich (St. Louis, MO). Oligonucleotides were synthe-
sized by Metabion (Martinsried, Germany); pQE-St-LOX1 (Solanum
tuberosum) was kindly donated by Ellen Hornung;
[13,43]
pQE-Cs-9/
[
19]
1
3HPL (Cucumis sativus) was kindly donated by Kenji Matsui;
3
0 mꢁ0.25 mmꢁ0.25 mm). Hydrogen was used as the carrier gas
9
-oxo-nonanoic acid was synthesized by the ozonolysis of oleic
À1
[11]
with a linear velocity of 30 cms . The injector and detector tem-
peratures were set to 250 and 3208C, respectively. The initial tem-
perature was set to 508C hold for 2 min, increased to 1008C
acid as described elsewhere.
À1
À1
(10 Kmin ), then increased to 2008C (30 Kmin ), further increased
Cloning of pREP4-Cs-9/13HPL
À1
À1
to 2408C (10 Kmin ), finally increased to 3008C (40 Kmin ), and
The whole Cs-9/13HPL expression cassette, which included pro-
moter, Cs-9/13HPL gene, and ter-
held for an additional 3 min.
minator, was amplified from pQE-
Cs-9/13HPL by the polymerase
chain reaction (PCR) by using oli-
Table 1. Oligonucleotides.
gonucleotides 1 and 2 (Table 1).
The insert was ligated into the
pREP4 vector using the Bsu36I and
NruI restriction sites.
Sequence
Site
1
2
5’-AAC CAA TGC ATT CCC TCA GGC GGC TCG AGA AAT CAT AAA AAA T-3’
5’-CGT ATG CAT CAG CTT CGC GAC CAT GGT TGG ATT CTC ACC AAT AAA AAA-3’
Bsu36I
NruI
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1008

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
22] A. P. Kulkarni, A. Mitra, J. Chaudhuri, J. Z. Byczkowski, I. Richards, Bio-
chem. Biophys. Res. Commun. 1990, 166, 417–423.
Acknowledgements
[
[
23] P. J. Farabaugh, Nature 1978, 274, 765–769.
24] M. J. Schilstra, G. A. Veldink, J. Verhagen, J. F. G. Vliegenthart, Biochem-
istry 1992, 31, 7692–7699.
The research leading to these results has received funding from
the European Union’s Seventh Framework Programme FP7/2007–
[
[
25] R. C. Scarrow, M. G. Trimitsis, C. P. Buck, G. N. Grove, R. A. Cowling, M. J.
Nelson, Biochemistry 1994, 33, 15023–15035.
26] M. C. Feiters, R. Aasa, B. G. Malmstrçm, G. A. Veldink, J. F. G. Vliegenthart,
Biochim. Biophys. Acta Protein Struct. Mol. Enzymol. 1986, 873, 182–189.
2
013 under grant agreement no. 266025 and the Landesgraduier-
tenfçrderung (LGF) Baden-Wꢀrttemberg. Furthermore, we want
to thank Sabine Rosahl, Ellen Hornung, Alan Brash, and Kenji
Matsui for the kind plasmid donations.
[27] W. Lands, R. Lee, W. Smith, Ann. N. Y. Acad. Sci. 1971, 180, 107–122.
[
28] M. P. Santiago-Gꢄmez, C. Vergely, C. Policar, J.-M. Nicaud, J.-M. Belin, L.
Rochette, F. Husson, Enzyme Microb. Technol. 2007, 41, 13–18.
Keywords: biocatalysis
oxidoreductases · oxidation · polymers
·
biotransformations
·
[29] C. N. S. P. Suurmeijer, M. Pꢅrez-Gilabert, D.-J. van Unen, H. T. W. M. van
der Hijden, G. A. Veldink, J. F. G. Vliegenthart, Phytochemistry 2000, 53,
177–185.
[
[
[
30] F.-C. Huang, W. Schwab, Plant Biotechnol. J. 2012, 10, 1099–1109.
31] H. Gardner, N. Deighton, Lipids 2001, 36, 623–628.
32] R. Leꢄn, P. Fernandes, H. M. Pinheiro, J. M. S. Cabral, Enzyme Microb.
Technol. 1998, 23, 483–500.
[
[
1] A. H. Tullo, Chem. Eng. News 2008, 86, 21–25.
2] D. Peters, N. Holst, B. Hermann, S. Lulies, H. Stolte, Nachwachsende Roh-
stoffe in Der Industrie, Fachagentur Nachwachsender Rohstoffe E. V.,
Gꢂlzow, 2010.
[
[
[
33] A. Schmid, J. S. Dordick, B. Hauer, A. Kiener, M. Wubbolts, B. Witholt,
[
[
[
[
[
3] H. Hꢃger, J. Limper, Elem. 28—Evonik Sci. Newsl. 2009, 24–30.
4] A. H. Tullo, Chem. Eng. News 2013, 19, 28–30.
5] V. Reimer, A. Kꢂnkel, S. Philipp, Kunststoffe 2008, 32–36.
6] D. S. Ogunniyi, Bioresour. Technol. 2006, 97, 1086–1091.
7] B. Cornils, P. Lappe, Dicarboxylic Acids, Aliphatic, Wiley-VCH, Weinheim,
Nature 2001, 409, 258–268.
34] A. Schmid, F. Hollmann, J. B. Park, B. Bꢂhler, Curr. Opin. Biotechnol. 2002,
1
3, 359–366.
35] J. L. Slonczewski, B. P. Rosen, J. R. Alger, R. M. Macnab, Proc. Natl. Acad.
Sci. USA 1981, 78, 6271–6275.
36] A. Mabrouk, L. R. Dugan Jr., J. Am. Oil Chem. Soc. 1961, 38, 9–13.
37] G. Carrea, Trends Biotechnol. 1984, 2, 102–106.
2000.
[
[
[
[
[
8] A. Kçckritz, A. Martin, Eur. J. Lipid Sci. Technol. 2008, 110, 812–824.
9] A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411–2502.
38] M. G. Wubbolts, O. Favre-Bulle, B. Witholt, Biotechnol. Bioeng. 1996, 52,
[
[
10] J.-W. Song, E.-Y. Jeon, D.-H. Song, H.-Y. Jang, U. T. Bornscheuer, D.-K. Oh,
301–308.
J.-B. Park, Angew. Chem. 2013, 125, 2594–2597; Angew. Chem. Int. Ed.
2
[
[
[
39] M. Buchhaupt, J. Guder, M. Etschmann, J. Schrader, Appl. Microbiol. Bio-
technol. 2012, 93, 159–168.
40] T. Q. Do, J. R. Schultz, C. F. Clarke, Proc. Natl. Acad. Sci. USA 1996, 93,
013, 52, 2534–2537.
11] K. B. Otte, M. Kirtz, B. M. Nestl, B. Hauer, ChemSusChem. 2013, DOI:
0.1002/cssc.201300183.
12] A. Andreou, I. Feussner, Phytochemistry 2009, 70, 1504–1510.
13] A. Geerts, D. Feltkamp, S. Rosahl, Plant Physiol. 1994, 105, 269–277.
14] N. Tijet, C. Schneider, B. L. Muller, A. R. Brash, Arch. Biochem. Biophys.
1
7
534–7539.
41] E. Skrivanova, M. Marounek, V. Benda, P. Brezina, Vet. Med.-Czech 2006,
1, 81.
[
[
[
5
[
[
42] M. Marounek, E. Sk rˇ ivanovꢆ, V. Rada, Folia Microbiol. 2003, 48, 731–735.
43] A. Z. Andreou, E. Hornung, S. Kunze, S. Rosahl, I. Feussner, Lipids 2009,
2001, 386, 281–289.
[
15] B. Muller, A. Gautier, C. Dean, J.-C. Kuhn (Firmenich, S. A.), U.S. Patent
WO/1993/024644, 1993.
16] A. Liavonchanka, I. Feussner, J. Plant Physiol. 2006, 163, 348–357.
4
4, 207–215.
44] K. Matsui, J. Wilkinson, B. Hiatt, V. Knauf, T. Kajiwara, Plant Cell Physiol.
999, 40, 477–481.
[
[
[
[
[
1
17] A. R. Brash, J. Biol. Chem. 1999, 274, 23679–23682.
[
[
45] C. Ingram, A. Brash, Lipids 1988, 23, 340–344.
46] M. J. Gibian, P. Vandenberg, Anal. Biochem. 1987, 163, 343–349.
18] I. Feussner, C. Wasternack, Annu. Rev. Plant Biol. 2002, 53, 275–297.
19] K. Matsui, C. Ujita, S. Fujimoto, J. Wilkinson, B. Hiatt, V. Knauf, T. Kaji-
wara, I. Feussner, FEBS Lett. 2000, 481, 183–188.
[
[
20] W. L. Smith, W. E. M. Lands, J. Biol. Chem. 1972, 247, 1038–1047.
21] J. L. Haining, B. Axelrod, J. Biol. Chem. 1958, 232, 193–202.
Received: September 17, 2013
Published online on December 20, 2013
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1003 – 1009 1009