CHEMBIOCHEM

FULL PAPERS

DOI: 10.1002/cbic.201402215

A Coenzyme-Independent Decarboxylase/Oxygenase
Cascade for the Efficient Synthesis of Vanillin
Toshiki Furuya,* Misa Miura, and Kuniki Kino*[a]

Vanillin is one of the most widely used flavor compounds in enzyme-independent oxygenase. When Escherichia coli cells
the world as well as a promising versatile building block. The harboring the decarboxylase/oxygenase cascade were incubat-
biotechnological production of vanillin from plant-derived ed with ferulic acid, the cells efficiently synthesized vanillin
ferulic acid has attracted much attention as a new alternative (8.0 mm, 1.2 g L 1) via 4-vinylguaiacol in one pot, without the
to chemical synthesis. One limitation of the known metabolic generation of any detectable aromatic by-products. The effi-
pathway to vanillin is its requirement for expensive coenzymes. cient method described here might be applicable to the syn-
Here, we developed a novel route to vanillin from ferulic acid thesis of other high-value chemicals from plant-derived aro-
that does not require any coenzymes. This artificial pathway matics.
consists of a coenzyme-independent decarboxylase and a co-

Introduction

Vanillin (4-hydroxy-3-methoxybenzaldehyde) is one of the ferulic acid as a starting material provide vanillin as a “biobased
world’s most important flavor and fragrance compounds in building block” for the synthesis of a variety of chemicals.[1, 2]
foods and cosmetics, and is also used as an intermediate in Furthermore, vanillin produced by biotechnological processes
the synthesis of pharmaceuticals.[1] Furthermore, vanillin is a can be labeled as a “natural flavor” under Europe and US legis-
promising building block for the preparation of functionalized lation; this is extremely expensive compared with synthetic
polymers.[2] Most vanillin production (about 12 000 tons per flavor.[1, 5]
year) currently depends on chemical synthesis, including the Many microorganisms that convert ferulic acid to vanillin
glyoxylic method and the nitrose method.[1b, 3] In these, vanillin have been isolated.[5] For example, growing actinomycetous
is synthesized from petrochemical guaiacol through introduc- cells produced 10–20 g L 1 vanillin from ferulic acid.[5] Although
tion of the formyl group into the aromatic ring. These chemical these microorganisms have high potential for the biotechno-
methods, however, do not entirely avoid the use of toxic re- logical production of vanillin, they generally include vanillin
agents and/or the generation of harmful by-products.[3] In degradation pathways, which lead to undesired by-products
recent years, environmentally benign chemical methods have and low product yield.[7, 8] Thus, several researchers have at-
also been extensively studied.[3, 4] Natural vanillin comes from tempted to engineer Escherichia coli to express the genes in-
the beans of the vanilla orchid, but the amount of vanillin volved in vanillin synthesis, because this model microorganism
extracted from plant sources is limited. Less than 1 % of vanillin has no vanillin degradation pathways and because it allows
production is derived from natural sources.[1] high-level expression of heterologous genes. Two genes are re-
Biotechnological and biocatalytic production of vanillin has sponsible for the synthesis of vanillin from ferulic acid
attracted much attention as a green alternative to the conven- (Scheme 1 A).[7] One encodes feruloyl-CoA synthetase (Fcs),
tional methods.[5] Ferulic acid (4-hydroxy-3-methoxycinnamic which catalyzes the conversion of ferulic acid to feruloyl-CoA.
acid) is a practical starting material for the biotechnological The other encodes enoyl-CoA hydratase/aldolase (Ech), which
production of vanillin, because this compound is available catalyzes the conversion of feruloyl-CoA to vanillin; E. coli cells
from renewable resources.[6] For example, a large amount of expressing these two genes were able to synthesize vanillin
ferulic acid can be recovered from agro-industrial waste (e.g., from ferulic acid via feruloyl-CoA.[9] However, the fact that Fcs
from cereal bran).[6] Biotechnological processes that use natural requires ATP and CoA as coenzymes (Scheme 1 A) makes this
biotechnological process complicated, as these expensive
coenzymes must be continuously supplied for Fcs to achieve
[a] Dr. T. Furuya, M. Miura, Prof. Dr. K. Kino high-yield production of vanillin. Recently, Lee et al. attempted
Department of Applied Chemistry to enhance productivity by constructing a CoA regeneration
Faculty of Science and Engineering, Waseda University system in E. coli.[10] If a coenzyme-dependent pathway can be
3-4-1 Ohkubo, Shinjuku-ku, 169-8555 Tokyo (Japan)
replaced with a coenzyme-independent one, construction of
E-mail: tfuruya@fuji.waseda.jp
kkino@waseda.jp a coenzyme regeneration system or addition of coenzymes or
Supporting information for this article is available on the WWW under energy sources (e.g., glucose) are not required for the biocata-
http://dx.doi.org/10.1002/cbic.201402215. lytic processes.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &1&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

Scheme 1. Biocatalytic synthesis of vanillin from ferulic acid. A) Known metabolic pathway: feruloyl-CoA synthetase and enoyl-CoA hydratase/aldolase.
B) Novel metabolic pathway: ferulic acid decarboxylase and 4-vinylguaiacol oxygenase.

In this study, we designed a coenzyme-free route to vanillin of the genome sequences was performed with the amino acid
from ferulic acid. This artificial pathway consists of a coen- sequence of Iso (Table S1 in the Supporting Information). Iso-
zyme-independent decarboxylase that converts ferulic acid to eugenol oxygenase from Pseudomonas nitroreducens Jin1[15] ex-
4-vinylguaiacol (2-methoxy-4-vinylphenol), and then a coen- hibited the highest amino acid similarity to Iso (82 %), followed
zyme-independent oxygenase to convert 4-vinylguaiacol to va- by several putative carotenoid cleavage oxygenases (~ 40 %
nillin (Scheme 1 B). The decarboxylase has been well studied,[11] similarity). Interestingly, the genome sequence of Caulobacter
but an appropriate enzyme for the second reaction has not segnis ATCC 21756 contains three putative carotenoid oxygen-
been reported. We searched for the second-step enzyme by ase genes, Cseg_1716, Cseg_1740, and Cseg_1745, whose
using a genome mining approach and discovered a novel gene products share moderate amino acid identity (42 %) with
coenzyme-independent oxygenase that cleaves the C=C bond Iso (E values 8e-118, 3e-125, and 3e-120, respectively; Fig-
of 4-vinylguaiacol to afford vanillin. We then constructed the ure S1). The gene product of Cseg_1716 shares 46 and 38 %
coenzyme-independent two-step metabolic pathway in Escher- amino acid identity with those of Cseg_1740 and Cseg_1745,
ichia coli and investigated decarboxylase/oxygenase catalysis respectively; the gene product of Cseg_1740 shares 42 % iden-
for the synthesis of vanillin. tity with that of Cseg_1745. These three putative gene prod-
ucts show divergent amino acid sequences and were promis-
ing candidates for 4-vinylguaiacol oxidation. In this study, we
Results and Discussion focused on the complete carotenoid oxygenase complement
of C. segnis. ORFs Cseg_1716, Cseg_1740, and Cseg_1745 were
Genome-wide search for a novel coenzyme-independent
designated cso1, cso2, and cso3, respectively.
oxygenase gene
To create the coenzyme-independent route to vanillin from
Gene cloning and characterization of C. segnis putative
ferulic acid via 4-vinylguaiacol (Scheme 1 B), we first searched
oxygenases
for a second-step enzyme that can convert 4-vinylguaiacol to
vanillin in the absence of any coenzyme by using a genome The three genes cso1, cso2, and cso3 were individually cloned
mining approach. Enzymes of the carotenoid cleavage oxygen- into the pET21a vector to produce proteins with a C-terminal
ase family are candidates for such activity. These enzymes con- His-tag. The resulting plasmids (Table 1) were introduced into
tain iron as a prosthetic group and catalyze oxidative cleavage E. coli BL21 (DE3) cells, and expression was induced with iso-
of the conjugated C=C bond of organic molecules without the propyl-b-d-thiogalactopyranoside (IPTG) under the control of
need for any coenzyme.[12] This enzymatic oxidation also has the T7 promoter. Sodium dodecyl sulfate–polyacrylamide gel
high catalytic potential as an equivalent to chemical ozonoly- electrophoretic (SDS-PAGE) analysis showed major bands corre-
sis.[13] Yamada et al. reported that Iso, an oxygenase of this sponding to Cso1 and Cso3 in soluble fractions, whereas Cso2
family from Pseudomonas putida IE27, converted isoeugenol to was largely insoluble (Figure S2 A). When the three genes were
vanillin in a coenzyme-independent manner.[14] However, its ac- coexpressed with the chaperonin gene groEL and the cocha-
tivity with 4-vinylguaiacol was low (1 % of that with isoeuge- peronin gene groES (plasmid pGro7; Table 1), all genes were ef-
nol).[14] There has been no report of a member of this oxygen- ficiently expressed as soluble products (Figure S2 B). The three
ase family that efficiently converts 4-vinylguaiacol to vanillin. His-tagged proteins were then purified to homogeneity from
Thus, bacterial genome sequences were searched for genes the soluble fractions of the E. coli cells on a nickel column (Fig-
encoding a coenzyme-independent oxygenase. A BLAST search ure S3). The molecular masses of the three proteins were esti-

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &2&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

Table 1. Bacterial strains and plasmids used in this study. Table 2. Km and Vmax of the Cso proteins for isoeugenol and 4-vinylguaia-
col.
Strain or Characteristics Ref. or
plasmid source Isoeugenol 4-Vinylguaiacol
Protein Km Vmax Km Vmax
Strains
[mm] [nmol min 1 mg 1] [mm] [nmol min 1 mg 1]
E. coli
DH5a host used for cloning Takara Bio Cso1 7.4 4.9 2.5 0.50
BL21(DE3) host used for expression Invitrogen Cso2 0.85 69 4.1 68
C. segnis Cso3 1.6 6.5 1.6 1.4
NBRC wild type, corresponding to ATCC 21756, from NBRC
15250 which oxygenase genes were cloned
B. pumilus
NBRC wild type, corresponding to ATCC 14884, from NBRC
3813 which a decarboxylase gene was cloned 4-vinylguaiacol as well as for isoeugenol in the absence of
Plasmids coenzyme (Table 2). LC-MS analysis revealed that the reaction
pET21a vector used for cloning, Apr Novagen of these enzymes with 4-vinylguaiacol generated vanillin
pETDuet-1 vector used for cloning, Apr Novagen
(Scheme 1 B, Figure 3 B). The Km values for 4-vinylguaiacol were
pGro7 plasmid containing the chaperonin gene groEL Takara Bio
and the cochaperonin gene groES, Cmr almost the same for the three proteins (1.6–4.1 mm), whereas
pETcso1His pET21a containing cso1 with an in-frame C-ter- this study the Vmax values were different. Only Cso2 had high catalytic ac-
minal His-tag sequence tivity (Vmax 68 nmol min 1 mg 1) for 4-vinylguaiacol (ca. 140 and
pETcso2His pET21a containing cso2 with an in-frame C-ter- this study
49 times higher than Cso1 and Cso3, respectively). Cso1 and
minal His-tag sequence
pETcso3His pET21a containing cso3 with an in-frame C-ter- this study Cso3 exhibited substrate preference for isoeugenol rather than
minal His-tag sequence 4-vinylguaiacol (Table 2), as previously reported for other en-
pETDfdc pETDuet-1 containing fdc in MCS-1 this study zymes in the carotenoid cleavage oxygenase family (e.g., Iso[14]
pETDcso2 pETDuet-1 containing cso2 in MCS-2 this study
and lignostilbene oxygenase from Pseudomonas paucimobi-
pETDfdc- pETDuet-1 containing fdc in MCS-1 and cso2 in this study
cso2 MCS-2 lis).[16] In contrast, interestingly, Cso2 exhibited novel substrate
specificity: Vmax for 4-vinylguaiacol was the same as that for iso-
eugenol (Table 2). Cso2 is the first enzyme in the carotenoid
cleavage oxygenase family that has been shown to efficiently
mated at 56–57 kDa, consistent with masses calculated from cleave the C=C bond of 4-vinylguaiacol. Thus, we selected the
the amino acid sequences of the His-tagged proteins. Cso2 protein for the second-step reaction of the coenzyme-in-
We first examined the catalytic activity of the three purified dependent route to vanillin from ferulic acid.
putative oxygenases towards isoeugenol and found that all
three exhibited catalytic activity in the absence of coenzyme
Construction of expression plasmids for the decarboxylase/
(Table 2). High-performance liquid chromatography (HPLC)
oxygenase cascade
analysis of the reaction mixtures with these enzymes and iso-
eugenol showed a major peak (tR = 7.4 min) in addition to the The coenzyme-independent decarboxylase that converts ferulic
substrate peak (23.0 min). The compound corresponding to acid to 4-vinylguaiacol (first step) has been well studied.[11] This
this peak was identified as vanillin, based on comparison of its type of enzyme contains no prosthetic group and catalyzes
retention time with an authentic
sample of vanillin and its mass
([M H] , m/z 151.1). The Km and
Vmax values were calculated from
Lineweaver–Burk plots (at 20 8C,
pH 7.5). The apparent Km values
of Cso1, Cso2, and Cso3 for iso-
eugenol were 7.4, 0.85, and
1.6 mm, respectively; Vmax values
were 4.9, 69, and 6.5 nmol
min 1 mg 1, respectively. These
results indicate that Cso2 had
the highest affinity and catalytic
activity for isoeugenol among
the three enzymes.
We next explored the catalytic
potential of the three enzymes
Figure 1. SDS-PAGE analysis of the expression of fdc and cso2 in E. coli. A) Samples prepared from E. coli cells car-
for the oxidation of 4-vinylguaia-
rying pETDuet-1 (empty vector) or pETDfdc (fdc). B) Samples prepared from E. coli cells carrying pETDuet-1 (empty
col and found that all three ex- vector), pETDcso2 (cso2), or pETDcso2 and pGro7 (cso2 + gro). S, soluble fraction; W, whole-cell sample. Molecular
hibited catalytic activity for both masses corresponding to Fdc, Cso2, and GroEL are indicated by arrows.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &3&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

nonoxidative decarboxylation without a coenzyme.[17] The Vanillin synthesis using E. coli cells harboring the
ferulic acid decarboxylase Fdc from Bacillus pumilus ATCC decarboxylase/oxygenase cascade
14884 was reported to exhibit high catalytic activity.[11a] Thus,
B. pumilus fdc was cloned into multicloning site 1 (MCS-1) of To optimize this two-step reaction, we examined the pH de-
the pETDuet-1 vector, and the resulting plasmid, pETDfdc pendency of the first- and second-step reactions in E. coli cells
(Table 1), was introduced into E. coli BL21 (DE3). Expression carrying pETDfdc, and in cells carrying pETDcso2 and pGro7.
analysis by SDS-PAGE revealed that Fdc was efficiently pro- Whole-cell reactions were performed on cells that were har-
duced in the soluble fraction of the E. coli cells (Figure 1 A). vested and suspended in potassium phosphate buffer (pH 6.0–
7.5) or glycine/NaOH buffer (pH 9.0–10.5). The reactions were
performed at 20 8C, as Cso2 (second step) was fairly unstable
at 30 8C (relative to 20 8C): preliminary experiments indicated
that the initial rate of 4-vinylguaiacol oxidation by Cso2 at
30 8C was higher than that at 20 8C, whereas the production of
vanillin after 24 h at 30 8C was lower than that at 20 8C. Cells
expressing fdc almost completely converted 10 mm ferulic acid
to 4-vinylguaiacol in 1 h over pH 6.0–9.0 (Figure 2 A), whereas
under alkaline conditions (pH 10.5), the decarboxylation rate
decreased: 8.2 mm 4-vinylguaiacol was synthesized from
10 mm ferulic acid in 8 h (Figure 2 A). The pH dependency of
the second-step reaction tended to be opposite to that of the
first-step reaction (Figure 2 B). As the pH of the reaction mix-
ture was increased, E. coli cells expressing the cso2 gene more
rapidly converted 4-vinylguaiacol to vanillin: at pH 10.5,
7.5 mm vanillin was synthesized from 10 mm 4-vinylguaiacol in
24 h (Figure 2 B). These results suggest that the optimal pH of
the two-step reaction for vanillin synthesis is pH 9.0–10.5.
We attempted to synthesize vanillin from 10 mm ferulic acid
by using E. coli cells carrying pETDfdc-cso2 and pGro7. When
tested over the range pH 9.0–10.5, vanillin was most efficiently
synthesized at pH 9.5 (Figure S4). E. coli cells harboring the de-
carboxylase/oxygenase cascade converted ferulic acid to vanil-
lin via 4-vinylguaiacol (Figure 3 A). After 1 h of reaction, 4-vinyl-
guaiacol temporarily accumulated in the reaction mixture. This
intermediate was subsequently converted to vanillin (6.8 mm
Figure 2. pH dependency of the first- and second-step reactions for vanillin vanillin after 8 h; 8.0 mm (1.2 g L 1) after 32 h). HPLC analysis
synthesis by transformed E. coli cells. A) Concentration of 4-vinylguaiacol revealed that the engineered E. coli cells did not generate any
synthesized from ferulic acid by Fdc. B) Concentration of vanillin synthesized detectable aromatic by-products (Figure 3 B). Formaldehyde,
from 4-vinylguaiacol by Cso2. *: pH 6.0, ~: pH 7.5, &: pH 9.0, *: pH 10.5.
the other product generated by the oxidative cleavage of the
Data are mean  SD from two independent experiments.
4-vinylguaiacol C=C bond (Scheme 1 A), was partially con-
sumed by the cells (Figure 3 A). It has been reported that form-
The gene encoding the coenzyme-independent oxygenase aldehyde is degraded by several enzymes in E. coli, including
Cso2 (second step) was cloned into MCS-2 of pETDuet-1 to formaldehyde dehydrogenase and S-formylglutathione hydro-
yield the gene product without a His-tag. The resulting plas- lase.[18] Residual formaldehyde can be removed from vanillin by
mid, pETDcso2 (Table 1), was introduced into E. coli BL21 (DE3). standard separation methods, such as evaporation and chro-
We confirmed by SDS-PAGE analysis that Cso2 without a His- matography.
tag was more highly expressed in the soluble fraction (Fig- The approach described here provides not only a new,
ure 1 B) than the His-tagged Cso2 protein described above simple, and efficient method for vanillin synthesis, but also
(Figure S2; chaperonin GroEL and the cochaperonin GroES novel insights into the molecular mechanisms that underlie
coexpressed from the pGro7 plasmid). Based on these expres- the microbial metabolism of ferulic acid and 4-vinylguaiacol in
sion studies, from pETDfdc and pETDcso2 we constructed a co- the environment. Recent reports suggest that several microor-
expression plasmid carrying fdc in MCS-1 and cso2 in MCS-2. ganisms metabolize ferulic acid via 4-vinylguaiacol and vanil-
The resulting plasmid, pETDfdc-cso2 (Table 1), was introduced lin.[19] Although it was confirmed that ferulic acid decarboxy-
with pGro7 into E. coli BL21 (DE3) to assemble the novel meta- lase was responsible for the conversion of ferulic acid into 4-vi-
bolic pathway to vanillin from ferulic acid in E. coli. nylguaiacol in Enterobacter sp. Px6-4,[20] the enzymes involved
in the subsequent metabolic steps from 4-vinylguaiacol have
not yet been elucidated. The pathway engineering described
here suggests that enzymes homologous to the 4-vinylguaia-

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &4&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

ods, although many researchers have been extensively study-

ing environmentally benign chemical and biocatalytic meth-
ods. Here, we designed a coenzyme-independent decarboxy-
lase/oxygenase cascade for the efficient biocatalytic synthesis
of vanillin from plant-derived ferulic acid via 4-vinylguaiacol.
The discovery of an oxygenase that converts 4-vinylguaiacol to
vanillin allowed the construction of the novel sustainable
route. We demonstrated that the decarboxylase/oxygenase
cascade functions well in E. coli: cells harboring this cascade
achieved gram-per-liter-scale production of vanillin in one pot.
The cells were able to provide vanillin without undesired by-
products, although the production was low compared with
that by actinomycetous strains.[5] In the two-step route, the
second-step reaction was the rate-determining step. The cata-
lytic activity of Cso2 (68 nmol min 1 mg 1) was lower than that
of Fdc (10.2 mmol min 1 mg 1)[11a] and of Fcs from a Streptomyces
strain (2.77 mmol min 1 mg 1).[21] Further investigations will
therefore focus on screening for and/or genetic engineering of
a higher-activity second-step oxygenase, to obtain higher-yield
production of vanillin. More generally, the developed method
might serve as a model for the design of other biocatalytic
and chemical reactions for the environmentally benign synthe-
sis of high-value chemicals from plant-derived aromatics, for
example, the synthesis of benzaldehydes from cinnamic acids.

Experimental Section
Chemicals: Ferulic acid, vanillin, and isoeugenol were purchased
from Wako Pure Chemicals (Osaka, Japan). 4-Vinylguaiacol was pur-
chased from Sigma–Aldrich. All other chemicals were of analytical
grade.
Bacterial strains, plasmids, and culture medium: Bacterial strains
and plasmids are listed in Table 1. Bacteria were grown in lysogeny
broth (LB): Bacto tryptone (10 g), Bacto yeast extract (5 g), and
NaCl (10 g) in water (1 L, pH 7.0).
Construction of expression plasmids for oxygenases: Three puta-
tive oxygenase genes, cso1, cso2, and cso3 (C. segnis ATCC 21756
ORFs Cseg_1716 (GenBank accession number, YP_003592815.1),
Figure 3. Vanillin synthesis by E. coli cells coexpressing Fdc and Cso2. Cseg_1740 (YP_003592837.1), and Cseg_1745 (YP_003592842.1),
A) Time courses of ferulic acid (*), 4-vinylguaiacol (~), formaldehyde (&), and respectively) were cloned into pET21a (Table 1) to obtain each
vanillin (*) concentration. Data are mean  SD from two independent ex- gene product with a C-terminal His-tag. Two oligonucleotide pri-
periments. B) HPLC chromatograms of the reactions after 0, 1, 8, and 32 h. mers, cso1-F and cso1-R (Table S2), were designed to amplify the
Peaks 1, 2, and 3 were found to correspond to ferulic acid, 4-vinylguaiacol, cso1 gene, based on the genome sequence of C. segnis ATCC
and vanillin, respectively. Peaks shown by an asterisk are detected also in 21756 (GenBank accession number, NC 014100). The region be-
the reaction mixture without ferulic acid, thus indicating that these peaks
tween the two oligonucleotide primers was amplified by PCR from
were not derived from the substrate.
genomic DNA of C. segnis NBRC 15250 (ATCC 21756). This amplified
DNA fragment was digested with NdeI and HindIII, and inserted
into pET21a digested with the same enzymes. The resulting plas-
col oxygenase identified in this study might play important
mid, pETcso1His (Table 1), was amplified in E. coli DH5a (Table 1).
roles in 4-vinylguaiacol metabolism in various microorganisms. Plasmids pETcso2His and pETcso3His (Table 1) were similarly con-
structed (primers in Table S2).
Conclusion After the correct generation of the recombinant plasmids had
been confirmed by sequencing, these plasmids were introduced
Vanillin is an important and versatile industrial compound that into E. coli BL21 (DE3) cells (Table 1) by electroporation. When
finds use in numerous applications, including the production required, the pGro7 plasmid (Table 1; carrying chaperonin gene
of foods, cosmetics, pharmaceuticals, and polymers. Industrial groEL and the cochaperonin gene groES) was simultaneously intro-
vanillin production is mainly by the glyoxylic and nitrose meth- duced.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &5&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

Construction of expression plasmids for the decarboxylase/oxy- Reactions using whole cells: The reaction mixture (1 mL) con-
genase cascade: The B. pumilus ATCC 14884 gene encoding ferulic tained transformed E. coli cells (50 g wet cell weight per liter), sub-
acid decarboxylase (Fdc; GenBank accession number, AJ278683)[11a] strate (10 mm), dimethylsulfoxide (10 % v/v) in potassium phos-
was cloned into MCS-1 of the pETDuet-1 vector (Table 1). Primers phate (100 mm, pH 6.0–7.5) or glycine-NaOH (100 mm, pH 9.0–10.5)
fdc-F and fdc-R (Table S2) were designed to amplify fdc without a containing glycerol (10 % v/v). The reactions were performed at
His-tag. The region between the two oligonucleotide primers was 20 8C with vigorous shaking.
amplified by PCR from genomic DNA of B. pumilus NBRC 3813
Product analysis: HPLC analysis was performed in a 10AVP system
(ATCC 14884). This amplified DNA fragment was digested with
(Shimadzu, Kyoto, Japan) with a Wakosil-II 5C18 HG column (4.6 
NcoI and BamHI and inserted into MCS-1 of pETDuet-1 to obtain
150 mm; particle size, 4.2–4.7 mm; Wako Pure Chemicals). The reac-
pETDfdc (Table 1).
tion mixture (purified-enzyme reaction mixture, 250 mL + whole-cell
Primers cso2-F and cso2-R’ (Table S2) were used to PCR amplify reaction mixture, 1 mL) was acidified by addition of HCl (pH 2–3).
cso2 (4-vinylguaiacol oxygenase gene) from C. segnis NBRC 15250 Methanol (250 mL) was added to the purified-enzyme reaction mix-
genomic DNA (without a His-tag). This amplified DNA fragment ture; ethanol (1 mL) was added to the whole-cell reaction mixture.
was digested with NdeI and MunI and inserted into MCS-2 of pET- After vigorous shaking and centrifugation, the supernatant (10 mL)
Duet-1 to obtain pETDcso2 (Table 1). was injected into the HPLC system (mobile phases A: formic acid
(0.1 % in water), B: methanol; flow rate 0.5 mL min 1; 40 % B
The coexpression plasmid (fdc in MCS-1, cso2 in MCS-2) was con- (8.5 min), increasing to 85 % B (over 1 min), 85 % B (15.5 min)).
structed from pETDfdc and pETDcso2. cso2 was excised from Compounds were detected spectrophotometrically (280 nm). Mass
pETDcso2 by digestion with SacI and KpnI and inserted into MCS-2 analysis was performed in an LCQ spectrometer (Thermo Finnigan)
of pETDfdc digested with the same restriction enzymes. The result- with electrospray ionization, as described previously.[23]
ing plasmid, pETDfdc-cso2 (Table 1), was simultaneously introduced
Formaldehyde concentration was measured with a formaldehyde
with pGro7 into E. coli BL21 (DE3) cells to construct the novel met-
test kit (Wako) according to the instruction manual. We confirmed
abolic pathway to vanillin from ferulic acid.
that this kit was able to quantify formaldehyde in a reaction mix-
Preparation of whole cells: The transformed E. coli BL21 (DE3) ture containing 10 mm ferulic acid, 4-vinylguaiacol, or vanillin.
cells were cultured at 37 8C in LB medium supplemented with am- Sequence analysis: Sequence similarity was analyzed with the
picillin (50 mg mL 1). For E. coli carrying pGro7, the medium was fur- BLAST program (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Multiple
ther supplemented with chloramphenicol (30 mg mL 1) and arabi- sequences were aligned in GENETYX (Win version 5.0.2; Genetyx
nose (4 mg mL 1). After culturing for 3 h (OD600 = 0.8–1.0), IPTG Corporation, Tokyo, Japan).
(0.1 mm) was added, and culturing was continued for an additional
16 h at 25 8C. Cells were harvested by centrifugation (5000 g,
10 min, 4 8C) and washed with potassium phosphate buffer Acknowledgements
(50 mm, pH 7.5) containing glycerol (10 % v/v). These cells were
used for protein analysis, protein purification, and whole-cell re-
actions.
This work was supported by Japan Science and Technology
Agency (JST), Adaptable & Seamless Technology Transfer Program
Protein analysis: Cells were suspended in potassium phosphate through Target-driven R&D (A-STEP) AS251Z01150N to T.F.
buffer and were disrupted by sonication. After centrifugation
(4000 g, 30 min, 4 8C), the supernatant was used as the soluble frac-
tion. Protein concentration was measured with a Coomassie (Brad- Keywords: biocatalysis · cascade reaction · decarboxylation ·
ford) protein assay kit (Thermo Scientific) with bovine serum albu- oxidation · sustainable chemistry
min as the standard.[22] The expression levels of proteins in the
E. coli cells were examined by SDS-PAGE analysis. [1] a) S. Wenda, S. Illner, A. Mell, U. Kragl, Green Chem. 2011, 13, 3007 –
3047; b) S. R. Rao, G. A. Ravishankar, J. Sci. Food Agric. 2000, 80, 289 –
Purification and characterization of oxygenases: The three His- 304; c) N. J. Walton, M. J. Mayer, A. Narbad, Phytochemistry 2003, 63,
tagged proteins, Cso1, Cso2, and Cso3, were purified from the 505 – 515.
soluble fractions of E. coli by using a HisTrap kit (GE Healthcare) ac- [2] a) A. S. Amarasekara, B. Wiredu, A. Razzaq, Green Chem. 2012, 14, 2395 –
cording to the instruction manual. The soluble fractions were ap- 2397; b) M. Fache, E. Darroman, V. Besse, R. Auvergne, S. Caillol, B. Bou-
plied to a HisTrap HP 1 mL column equilibrated with potassium tevin, Green Chem. 2014, 16, 1987 – 1998; c) M. Firdaus, M. A. R. Meier,
Eur. Polym. J. 2013, 49, 156 – 166; d) J. Kwon, J. Kim, S. Park, G. Khang,
phosphate (50 mm, pH 7.4) containing NaCl (500 mm) and imida-
P. M. Kang, D. Lee, Biomacromolecules 2013, 14, 1618 – 1626; e) J. F.
zole (20 mm). The column was then washed with 10 column vol-
Stanzione III, J. M. Sadler, J. J. La Scala, K. H. Reno, R. P. Wool, Green
umes of the same buffer. Bound His-tagged protein was eluted Chem. 2012, 14, 2346 – 2352.
with potassium phosphate (50 mm, pH 7.4) containing NaCl [3] J. Hu, Y. Hu, J. Mao, J. Yao, Z. Chen, H. Li, Green Chem. 2012, 14, 2894 –
(500 mm) and imidazole (500 mm). After individual fractions were 2898.
analyzed by SDS-PAGE, the fractions containing each Cso protein [4] a) V. Augugliaro, G. Camera-Roda, V. Loddo, G. i Palmisano, L. Palmisano,
were combined and desalted with a PD-10 column (GE Healthcare) F. Parrino, M. A. Puma, Appl. Catal. B 2012, 111 – 112, 555 – 561; b) J.-A.
equilibrated with potassium phosphate (50 mm, pH 7.5) containing Jiang, C. Chen, Y. Guo, D.-H. Liao, X.-D. Pan, Y.-F. Ji, Green Chem. 2014,
glycerol (10 % v/v). The catalytic activities of these three oxygenas- 16, 2807 – 2814.
[5] a) B. Kaur, D. Chakraborty, Appl. Biochem. Biotechnol. 2013, 169, 1353 –
es for isoeugenol and 4-vinylguaiacol were determined by using
1372; b) H. Priefert, J. Rabenhorst, A. Steinbchel, Appl. Microbiol. Bio-
these purified enzymes. The reaction mixture (250 mL) contained
technol. 2001, 56, 296 – 314; c) N. J. Walton, A. Narbad, C. B. Faulds, G.
purified enzyme (0.2–1 mg mL 1), substrate (0.5–2 mm), and di- Williamson, Curr. Opin. Biotechnol. 2000, 11, 490 – 496.
methylsulfoxide (1 % v/v) in potassium phosphate (50 mm, pH 7.5) [6] a) D. Di Gioia, L. Sciubba, L. Setti, F. Luziatelli, M. Ruzzi, D. Zanichelli, F.
with glycerol (10 % v/v). The reactions were performed at 20 8C Fava, Enzyme Microb. Technol. 2007, 41, 498 – 505; b) S. Mathew, T. E.
with vigorous shaking. Abraham, Crit. Rev. Biotechnol. 2004, 24, 59 – 83; c) J. P. N. Rosazza, Z.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &6&
These are not the final page numbers! ÞÞ
CHEMBIOCHEM
FULL PAPERS www.chembiochem.org

Huang, L. Dostal, T. Volm, B. Rousseau, J. Ind. Microbiol. 1995, 15, 457 – [17] a) W. Gu, J. Yang, Z. Lou, L. Liang, Y. Sun, J. Huang, X. Li, Y. Cao, Z.
471. Meng, K.-Q. Zhang, PLoS One 2011, 6, e16262; b) H. Rodrguez, I.
[7] a) S. Achterholt, H. Priefert, A. Steinbchel, Appl. Microbiol. Biotechnol. Angulo, B. de las Rivas, N. Campillo, J. A. Pez, R. MuÇoz, J. M. Manche-
2000, 54, 799 – 807; b) E. Masai, K. Harada, X. Peng, H. Kitayama, Y. Ka- Ço, Proteins Struct. Funct. Bioinf. 2010, 78, 1662 – 1676.
tayama, M. Fukuda, Appl. Environ. Microbiol. 2002, 68, 4416 – 4424; c) A. [18] a) C. F. Gonzalez, M. Proudfoot, G. Brown, Y. Korniyenko, H. Mori, A. V.
Narbad, M. J. Gasson, Microbiology 1998, 144, 1397 – 1405. Savchenko, A. F. Yakunin, J. Biol. Chem. 2006, 281, 14514 – 14522;
[8] X.-k. Ma, A. J. Daugulis, Biotechnol. Prog. 2014, 30, 207 – 214. b) W. G. Gutheil, B. Holmquist, B. L. Vallee, Biochemistry 1992, 31, 475 –
[9] a) P. Barghini, D. Di Gioia, F. Fava, M. Ruzzi, Microb. Cell Fact. 2007, 6, 13; 481; c) C. D. Herring, F. R. Blattner, J. Bacteriol. 2004, 186, 6714 – 6720;
b) S.-H. Yoon, C. Li, J.-E. Kim, S.-H. Lee, J.-Y. Yoon, M.-S. Choi, W.-T. Seo, d) B. K. Hunter, K. M. Nicholls, J. K. M. Sanders, Biochemistry 1984, 23,
J.-K. Yang, J.-Y. Kim, S.-W. Kim, Biotechnol. Lett. 2005, 27, 1829 – 1832. 508 – 514.
[10] E.-G. Lee, S.-H. Yoon, A. Das, S.-H. Lee, C. Li, J.-Y. Kim, M.-S. Choi, D.-K. [19] a) W. J. Hunter, D. K. Manter, D. van der Lelie, Curr. Microbiol. 2012, 65,
Oh, S.-W. Kim, Biotechnol. Bioeng. 2009, 102, 200 – 208. 752 – 757; b) B. Karmakar, R. M. Vohra, H. Nandanwar, P. Sharma, K. G.
[11] a) L. Barthelmebs, C. Divis, J.-F. Cavin, Appl. Environ. Microbiol. 2001, 67, Gupta, R. C. Sobti, J. Biotechnol. 2000, 80, 195 – 202; c) X. Li, J. Yang, X.
1063 – 1069; b) J. Yang, S. Wang, M.-J. Lorrain, D. Rho, K. Abokitse, Li, W. Gu, J. Huang, K.-Q. Zhang, Process Biochem. 2008, 43, 1132 – 1137.
P. C. K. Lau, Appl. Microbiol. Biotechnol. 2009, 84, 867 – 876. [20] W. Gu, X. Li, J. Huang, Y. Duan, Z. Meng, K.-Q. Zhang, J. Yang, Appl. Mi-
[12] a) D. P. Kloer, G. E. Schulz, Cell. Mol. Life Sci. 2006, 63, 2291 – 2303; crobiol. Biotechnol. 2011, 89, 1797 – 1805.
b) E. K. Marasco, C. Schmidt-Dannert, ChemBioChem 2008, 9, 1450 – [21] W. Yang, H. Tang, J. Ni, Q. Wu, D. Hua, F. Tao, P. Xu, PLoS One 2013, 8,
1461. e67339.
[13] R. Criegee, Angew. Chem. Int. Ed. Engl. 1975, 14, 745 – 752; Angew. [22] M. M. Bradford, Anal. Biochem. 1976, 72, 248 – 254.
Chem. 1975, 87, 765 – 771. [23] a) T. Furuya, K. Kino, ChemSusChem 2009, 2, 645 – 649; b) T. Furuya, Y.
[14] M. Yamada, Y. Okada, T. Yoshida, T. Nagasawa, Arch. Microbiol. 2007, Arai, K. Kino, Appl. Environ. Microbiol. 2012, 78, 6087 – 6094.
187, 511 – 517.
[15] J.-Y. Ryu, J. Seo, S. Park, J.-H. Ahn, Y. Chong, M. J. Sadowsky, H.-G. Hur,
Biosci. Biotechnol. Biochem. 2013, 77, 289 – 294. Received: May 7, 2014
[16] S. Kamoda, Y. Saburi, Biosci. Biotechnol. Biochem. 1993, 57, 931 – 934. Published online on && &&, 0000

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &7&
These are not the final page numbers! ÞÞ
FULL PAPERS
T. Furuya,* M. Miura, K. Kino*
&& – &&

A Coenzyme-Independent
Decarboxylase/Oxygenase Cascade for
the Efficient Synthesis of Vanillin
Vanillin, a valuable building block and that is, a “coenzyme-independent de-
flavor, was efficiently synthesized from carboxylase/oxygenase cascade”, allows
ferulic acid, which is available from re- high-yield production of vanillin in one
newable resources. A new biocatalysis, pot.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemBioChem 0000, 00, 1 – 8 &8&
These are not the final page numbers! ÞÞ