Content uploaded by Mans Broekgaarden
Author content
All content in this area was uploaded by Mans Broekgaarden on Apr 28, 2016
Content may be subject to copyright.
1 Introduction
Raspberry ketone [4-(4-hydroxyphenyl)-butan-2-one] is a
key flavour molecule with typical raspberry flavour char-
acteristics and a low odour threshold. Raspberry ketone is
one of the most expensive flavour components used in the
food industry. Up to $20 000/kg may be paid for the natu-
ral compound, which is found in raspberries and also in
other fruits such as peaches, grapes, apples and various
berries, vegetables (e.g., rhubarb) and in the bark of trees
(e.g., yew, maple and pine). Raspberry ketone can be used
in the aroma formulation of, for instance, strawberry, kiwi,
cherry and other berries. However, none of the fruits list-
ed are currently used to obtain raspberry ketone, as the
low content in these fruits makes the extraction and pu-
rification process unprofitable. For instance, if raspberry
ketone was extracted from raspberry fruit, one would face
the limited availability of this fruit outside the growing
season, and its price. Yields are typically very low: the
content is usually 1–4 mg/kg raspberries [1].
Chemical synthesis of raspberry ketone can be
achieved via the condensation of p-hydroxybenzaldehyde
with acetone [2]. However, according to food laws, this
product of chemical synthesis cannot be marketed as a
“natural” flavour compound [3]. Therefore, we set out to
clone plant genes involved in the biosynthesis of raspber-
ry ketone, and express the pathway in micro-organisms,
to produce natural raspberry flavour compounds.
In the fruit of raspberry (Rubus idaeus), the natural
synthesis of raspberry ketone initiates with the phen-
ylpropanoid pathway [4] (Fig. 1). In the first step,
coumaroyl-CoA (which is present in many plant tissues as
intermediate of the lignin biosynthetic pathway) is con-
densed with one malonyl-CoA to form p-hydroxybenza-
lacetone [4-(4-hydroxyphenyl)-but-3-ene-2-one]. The en-
zyme catalysing this step is called benzalacetone syn-
thase (BAS). In the second step, the double bond in p-hy-
Research Article
Microbial production of natural raspberry ketone
Jules Beekwilder
1
, Ingrid M. van der Meer
1
, Ole Sibbesen
2
, Mans Broekgaarden
1
, Ingmar Qvist
3
,
Joern D. Mikkelsen
2
and Robert D. Hall
1
1
Plant Research International, Wageningen, The Netherlands
2
Danisco Innovation, Copenhagen, Denmark
3
Danisco Flavours, Wellingborough, England
Raspberry ketone is an important compound for the flavour industry. It is frequently used in prod-
ucts such as soft drinks, sweets, puddings and ice creams. The compound can be produced by or-
ganic synthesis. Demand for “natural” raspberry ketone is growing considerably. However, this
product is extremely expensive. Consequently, there is a remaining desire to better understand
how raspberry ketone is synthesized in vivo, and which genes and enzymes are involved. With this
information we will then be in a better position to design alternative production strategies such as
microbial fermentation. This article focuses on the identification and application of genes poten-
tially linked to raspberry ketone synthesis. We have isolated candidate genes from both raspberry
and other plants, and these have been introduced into bacterial and yeast expression systems.
Conditions have been determined that result in significant levels of raspberry ketone, up to 5 mg/L.
These results therefore lay a strong foundation for a potentially renewable source of “natural”
flavour compounds making use of plant genes.
Keywords: Raspberry ketone · Natural flavour · Fermentation
Correspondence: Dr. Jules Beekwilder, PO box 16, 6700 AA Wageningen,
The Netherlands
E-mail: jules.beekwilder@wur.nl
Fax: +31-317-418094
Abbreviations: BAR, benzalacetone reductase; BAS, benzalacetone syn-
thase; CHS, chalcone synthase; 4CL, 4-coumarate-coenzyme A ligase; STS,
stilbene synthase
Received 23 April 2007
Revised 13 June 2007
Accepted 27 June 2007
Biotechnology
Journal
DOI 10.1002/biot.200700076 Biotechnol. J. 2007, 2, 1270–1279
1270 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1271
Biotechnol. J. 2007, 2, 1270–1279 www.biotechnology-journal.com
droxybenzalacetone is reduced, resulting in raspberry ke-
tone. This step is catalysed by benzalacetone reductase
(BAR), and requires the presence of NADPH.
There is evidence that, in raspberry, a relationship ex-
ists between BAS and chalcone synthase (CHS). BAS ac-
tivity during fruit ripening parallels pigment formation [4].
The colour in ripe raspberry fruit is predominantly deter-
mined by the formation of red anthocyanins, which are
derived from the phenyl-propanoid pathway [5]. CHS is a
key enzyme in this pathway. CHS is a member of the type
III polyketide synthase family, and BAS is also likely to be
a member of this family. Whereas BAS condenses one
malonyl-CoA with one coumaroyl-CoA to form p-hydrox-
ybenzalacetone, CHS condenses three malonyl-CoA units
with one p-coumaroyl-CoA to form naringenin chalcone.
This naringenin chalcone may then be further trans-
formed into anthocyanins such as cyanidin sophoroside.
Moreover, a CHS homologue from rhubarb has been
cloned, which, after expression in Escherichia coli and
purification of the recombinant protein, showed in vitro
synthesis of p-hydroxybenzalacetone, but not of narin-
genin [6]. However, such an enzyme has never been iden-
tified in raspberry.
The NADPH-dependent reductase BAR has to date
hardly been investigated. It is known that p-hydroxyben-
zalacetone can be transformed to raspberry ketone by fun-
gi or yeasts such as Pichia, Saccharomyces and Beauve-
ria [7]. However, no gene has yet been identified that is
connected to this enzyme activity. Also, to our knowl-
edge, no such activity has been reported for bacteria.
Attempts to biosynthesise raspberry ketone have pre-
viously been described. Kosjek et al. [8] described a
method using a micro-organism that has a secondary al-
cohol dehydrogenase (ADH), which can convert the pre-
cursor betuloside into raspberry ketone. However, betulo-
side is a compound of limited availability.
We describe a method in which a raspberry CHS is in-
corporated into a raspberry ketone production pathway in
E. coli. Successful production of raspberry ketone was
achieved.
2 Materials and methods
2.1 Materials
Raspberries were obtained from the PPO-field station in
Randwijk (The Netherlands). The p-hydroxybenzalace-
tone was from Pfaltz & Bauer (Waterbury, CT, USA), narin-
genin chalcone, naringenin and raspberry ketone were
from Apin Chemicals Ltd. (Abingdon, UK), while all other
chemicals were from Sigma (St Louis, MO, USA). Restric-
tion enzymes were from Invitrogen. All DNA manipula-
tions were done in E. coli XL-1 blue. Expression was per-
formed in E. coli BL21-CodonPlus-RIL, supplied with plas-
mid pREP4.
2.2 Cloning of raspberry RiCHS
Total RNA was isolated from ripe Tulameen raspberries.
Raspberries (5 g) were homogenized in liquid nitrogen,
and incubated for 1 h at 65°C in 50 mL extraction buffer
(2% cetyltrimethylammonium bromide, 100 mM Tris
pH 8.2, 1.4 M NaCl, 20 mM EDTA, 0.1% 2-mercaptoethanol).
This was extracted twice with chloroform, and RNA was
precipitated with 33 mL 10 M LiCl at 4°C overnight. Pre-
cipitated RNA was dissolved in 1 mL TE buffer and ex-
tracted with phenol/chloroform. Copy DNA was synthe-
sized from 1 µg total RNA using a polyT primer according
to standard procedures. Primers RiCHSa (5’-CAAGT-
GAACCCAGCCATGGTCAAGTTGAAGCTGCCA-
CACTG) and RiCHSs (5’-CTTTCTCCACAGACTCGA-
GATGGTGACCGTCGATGAAGTC) and Pfu polymerase
(Stratagene) were used to amplify the full protein coding
region of a CHS from raspberry (GenBank accession
AF292367; [9]). The fragment, 1136 nucleotides in size,
was digested with NcoI and XhoI, and ligated into
pRSETA (Invitrogen), yielding plasmid pRSETA-RiCHS.
The sequence of the encoded protein was 98% identical
to the aromatic polyketide synthase in AF292367, with 7
residues out of 391 differing. It is even more identical
(99%, 3 out of 391 residues differ) to a protein called CHS6
(AF400567), described in [10].
2.3 Mutagenesis of CHS
The DNA of pRSETA-CHS#1 was mutated by overlap-ex-
tension PCR. Plasmid pRSETA-RiCHS was used as a tem-
plate. Two PCR reactions were performed in parallel, one
using oligonucleotides CHSmutA (5’-CGTCACCGAG-
S-CoA
O
HO
HO
O
S-
CoA
O
HO
HO
OH
O
OH
O
raspberry ketone
anthocyanins
CHS
BAS
OH
O
BAR
OH
O
HO
phenylalanine
lignin
4CL
p-coumaric acid
coumaroyl-CoA
malonyl-CoA
benzalacetone
naringenin chalcone
OH
Figure 1. Biosynthetic pathway of raspberry anthocyanins and raspberry
ketone. The biosynthesis of these compounds starts by the coupling of
p-coumaric acid to CoA by the 4CL enzyme. Subsequently, the coumaroyl-
CoA is converted into either naringenin chalcone, or to p-hydroxybenza-
lacetone, depending on the number of malonyl-CoA units added.
TATGGCTTGGCCCACAAG) and RiCHSs, and the other
using oligonucleotides RiCHSa and CHSmutS (5’-GC-
CAAGCCATACTCGGTGACGGTGCTGC) using Pfu
polymerase. Resulting fragments were purified from gel
and 100 ng of each fragment was used as a template for
fusion PCR using oligonucleotides RiCHSa and RiCHSs.
The resulting fragment was cloned into pRSETA using
XhoI and NcoI restriction enzymes yielding plasmid
pRSET-RiCHS-Mut. Sequence analysis of this plasmid
confirmed the presence of the designed mutations.
2.4 Plasmids pAC-4CL-RiCHS, pAC-4CL-RiCHS-Mut and
pAC-4CL-PhCHS
RiCHS and RiCHS-Mut DNA was amplified from the con-
structs described above using oligonucleotides CHS-
DUETS (5’-TATATAGATCTTGTGACCGTCGATGAAG)
and CHSDUETA (5’-TATATGGTACCTCAAGTTGAAGC
TGCCAC). The amplification products were digested us-
ing restriction enzymes KpnI and BglII, and cloned into
pACYC-DUET-1 (Novagen). The resulting plasmids were
termed pAC-RiCHS and pAC-RiCHS-Mut. The Petunia x
hybrida CHS gene was cleaved from plasmid pFLAP600
([11]; kindly provided by Arnaud Bovy) using restriction
enzymes BglII and XhoI, and ligated into pACYC-DUET-1,
yielding plasmid pAC-PhCHS.
The tobacco 4-coumarate-coenzyme A ligase (4CL)-
2 cDNA was added to all three constructs, and to pACYC-
DUET-1 as described in Beekwilder et al. [12], yielding
plasmids pAC-4CL-RiCHS, pAC-4CL-RiCHS-Mut, pAC-
4CL-PhCHS and pAC-4CL. Plasmid pAC-4CL-VvSTS has
been described elsewhere [12].
RiCHS was amplified from pAC-4CL-CHS by primers
CHSNco (5’-TATAGGTACCTCAGTTGAAGCTGCCACA
CTG) and CHSEco (5’-TATACTCGAGAATTCTCAAGTT
GAAGCTGCCACAC). 4CL was amplified from pAC-4CL-
CHS using oligonucleotides DUETS (5’-GGATCTCGACG
CTCTCCCT) and DUETASEco (5’-TATGAATTCGATTAT
GCGGCCGTGTACAA). Both amplification products were
digested with EcoRI and NcoI and cloned into plasmid
pBAD-HisA (Invitrogen), to yield plasmids pBAD-RiCHS
and pBAD-4CL.
2.5 In vitro BAS assay
Plasmids pRSET-A, pRSET-RiCHS and pRSET-RiCHS-
Mut were used for transformation of E. coli BL21. Individ-
ual transformants were inoculated in LB medium with 1%
glucose and 100 mg/L ampicillin and grown overnight at
37°C. The overnight cultures were diluted 1:100 in 50 mL
2xYT medium with 50 mg/L ampicillin, and further incu-
bated at 37°C. At a density (measured at 600 nm) of 0.4,
IPTG was added to 1 mM and cultures were grown
overnight at 28°C and 250 rpm. The next day, the cultures
were centrifuged and cell pellets were dissolved in 2 mL
buffer A (50 mM Tris pH 8, 300 mM NaCl, 10 mM imida-
zole). After sonication and centrifugation, the super-
natant was transferred to a NiTED Protein Purification
column (ActivMotiv), and treated as prescribed by the
manufacturer. About 10 µg eluted protein was mixed in
200 µL 50 mM Tris pH 7.0, with p-coumaroyl-CoA (10 µM,
kindly provided by Stefan Martens) and malonyl-CoA
(50 µM) and incubated for 1 h at 30°C, after which the
reaction was stopped by addition of 400 µL methanol
with 0.75% formic acid, and samples were analysed on
HPLC (see below).
2.6 Bioconversion of p-coumaric acid by E. coli
Plasmids were transferred into E. coli BL21 cells and se-
lected on the appropriate antibiotics (30 mg/L chloram-
phenicol in case of pAC-plasmids; 100 mg/L ampicillin in
case of pRSET plasmids or pBAD plasmids) in the pres-
ence of 1% glucose. Overnight cultures of individual
colonies were made in LB medium with antibiotics and
1% glucose. For expression, the overnight cultures were
diluted 1:100 in 50 mL 2xYT medium, and grown at 37°C
until a density (measured at 600 nm) of 0.4 was reached.
At this point, p-coumaric acid was added to 3 mM and
IPTG to 1 mM, and cultures were grown overnight at 28°C
and 250 rpm. The next day (after 20 h), the cultures were
transferred to separation funnels, and 20 mL ethyl acetate
was added under vigorous mixing. The upper (ethyl ac-
etate) phases were collected in centrifuge tubes, sonicat-
ed for 10 min and centrifuged for 5 min at 1200 × g. From
the clear ethyl acetate phases, solvent was aspired under
a nitrogen flow. Dried product was dissolved in 1 mL eth-
yl acetate, and analysed on HPLC and GC-MS.
2.7 Expression in yeast
For expression in yeast, the RiCHS gene was amplified
from pRSET-RiCHS and pRSET-RiCHS-Mut using oligo-
nucleotides ESCCHSfw (5’-ATATGGATCCACCATGGTG
ACCGTCGATGAAG) and ESCCHSre (5’-TATATACTCG
AGTCAAGTTGAAGCTGCCAC), and cloned into vector
pESC-trp (Stratagene) using BamHI and XhoI restriction
enzymes. This yielded plasmids pESC-RiCHS and pESC-
RiCHS-Mut. Subsequently, the 4CL fragment was re-
cloned from pAC-4CL-CHS into pESC-RiCHS and pESC-
RiCHS-Mut using EcoRI and NotI restriction enzymes.
The resulting plasmids (pESC-4CL-CHS and pESC-4CL-
CHS-Mut) and control vector pESC-Trp were introduced
into the yeast S. cerevisiae YPH499, where they were
maintained by tryptophan auxotrophy complementation.
Cultures (50 mL, YPGal-medium) were grown in the pres-
ence of galactose, to induce recombinant gene expres-
sion. At a culture density of 0.4 (at 600 nm), p-coumaric
acid (3 mM) was added, after which the cultures are
grown overnight at 30°C and 250 rpm. After 20 h, the cul-
tures were extracted with ethyl acetate as described
above, and extracts were analysed on HPLC.
Biotechnology
Journal
Biotechnol. J. 2007, 2, 1270–1279
1272 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1273
Biotechnol. J. 2007, 2, 1270–1279 www.biotechnology-journal.com
2.8 Fermentation
E. coli BL21 containing the pAC-4CL-RiCHS vector was
grown in an 800-mL fermenter (Braun Biotech, Biostat Q),
using 2xYT as growth medium. Temperature was kept at
25°C, and pH at 7.5. Samples were taken at various time
points. Samples were extracted with ethyl acetate and
analysed on a Perkin Elmer Autosystem XL GC fitted with
an FID detector and a Varian FactorFour VF-5ms 30 m ×
0.25 mm column with helium as carrier. Injector tempera-
ture was 270°C, detector temperature 300°C. Oven tem-
perature was at 50°C for the first 2 min after injection, and
thereafter raised by 15°C/min to 230°C. Ethylvanillin
(100 mg/L) was used as an internal standard in the
samples. The chromatograms were normalized according
to the standard, and raspberry ketone was measured
using a standard curve generated by using authentic
compounds.
2.9 HPLC and GC-MS analysis
From each extract, 10 µL was injected into an HPLC sys-
tem using a Luna 3u C18(2) 150 × 4.6 mm column (Phe-
nomenex). The HPLC set-up was composed of a Waters
600 controller and a Waters 996 Photo Diode Array detec-
tor. The column was used at 40°C, flow rate 1 mL/min. The
products were eluted with a linear gradient from 95%
buffer A (0.1% formic acid in water) and 5% buffer B (100%
acetonitrile) to 50% buffers A and B in 37 min.
Of each sample, 2 µL was analysed by GC-MS using a
gas chromatograph (5890 series II, Hewlett-Packard)
equipped with a 30 m × 0.25 mm 5MS column (Hewlett-
Packard) and a mass-selective detector. The GC was pro-
grammed at an initial temperature of 45°C, with a ramp of
10°C/min to 220°C and final time of 5 min and a constant
column flow of 1.0 mL/min. The injection port, interface,
and MS source temperatures were 250°C, 290°C, and
180°C, respectively. Compounds were detected in the se-
lected-ion-monitoring mode (70 eV): m/z 107, 121 and 164.
Masses at m/z 107 and 164 are typical for raspberry ke-
tone. The peak surface at m/z 107, eluting around 15 min,
was used as a measure for the amount of raspberry ketone
formed. Values were normalized by use of an internal stan-
dard [1 µg/mL 4-(4-methoxyphenyl)-2-butanone (Aldrich)]
in the ethyl acetate used for extraction, which was de-
tected at m/z 121.
3 Results
3.1 CHS and BAS activity in vitro
The first aim of this study was to obtain a CHS-like gene
from raspberry that could drive synthesis of p-hydroxy-
benzalacetone. A known CHS gene was amplified from
ripe raspberry (Rubus idaeus) fruit cDNA. The cDNA was
cloned into the E. coli expression vector pRSETA (yielding
pRSET-RiCHS), in such a way that the CHS protein is
fused to an N-terminal His-tag, which allows convenient
purification for enzymatic tests (Fig. 2). Subsequently,
mutations were introduced into this plasmid (Fig. 2B),
which were aimed at altering the properties of the
coumaroyl-CoA binding site at residues 214 (Leu) and 215
(Phe) in RiCHS. The introduced mutations alter these
residues to conform to those present in rhubarb BAS (Ile
and Leu, respectively; Fig. 2B, [13]), and the resulting
plasmid was referred to as pRSET-RiCHS-Mut (Fig. 2).
Both wild-type and mutant CHS fusion proteins were pro-
duced in E. coli, and subsequently purified using their N-
terminal His-tag. They were then incubated with
coumaroyl-CoA and malonyl-CoA and products were
analysed using HPLC. As expected from literature, the
RiCHS protein converted the substrates mainly into
naringenin chalcone (not shown; [10]).
However, the mutant enzyme displayed poorly repro-
ducible behaviour when it was tested a few times under
the same conditions. It never produced any naringenin
chalcone, but in some tests, the formation of p-hydroxy-
benzalacetone was observed. In others the mutant en-
zyme appeared to be inactive, as was also reported for a
similar Scutellaria CHS mutant [13]. It was therefore con-
ceived that the mutant RiCHS enzyme would perform bet-
4CL
RiCHS-Mut
His
RiCHS
T7
T7 T7
4CL RiCHS-Mut
T7 T7
4CL PhCHS
T7 T7
4CL VvSTS
T7 T7
4CL
T7
RiCHS
T7
RiCHS
His
T7
4CL
BAD
RiCHS
BAD
RiCHS
Gal1
4CL
Gal10
RiCHS-Mut
Gal1
4CL
Gal10
pESC-TrpTrppESC-4CL-CHS-Mut
pESC-TrpTrppESC-4CL-CHS
pBAD-HisAAmppBAD-RiCHS
pBAD-HisAAmppBAD-4CL
pACYC-DUET-1CampAC-RiCHS
pACYC-DUET-1CampAC-4CL
pACYC-DUET-1CampAC-4CL-VvSTS
pACYC-DUET-1CampAC-4CL-PhCHS
pACYC-DUET-1CampAC-4CL-RiCHS-Mut
pACYC-DUET-1CampAC-4CL-RiCHS
pRSET-AAmppRSET-RiCHS-Mut
pRSET-AAmppRSET-RiCHS
vectorselectionname
747 783
Leu Val Gly Gln Ala Leu Phe
Gly Asp Gly Ala Ala
RiCHS CTT GTG GGC CAA GCC TTG TTC GGT GAC GGT GCT GCA
RiCHS-Mut CTT GTG GGC CAA GCC ATACTC GGT GAC GGT GCT GCA
Leu Val Gly Gln Ala Ile Leu
Gly Asp Gly Ala Ala
Figure 2. Constructs used in this study. (A) Indicated are the promoters
(arrows), terminators (crosses), construct name, antibiotic resistance and
vector. (B) The region from nucleotide 747 to 783 in RiCHS and RiCHS-
Mut, and the encoded amino acid sequences. Altered residues are indicat-
ed in bold.
AB
ter in an in vivo system, where circumstances (the cyto-
plasmic environment) would be more constant and fresh
enzyme is continuously being synthesized.
3.2 E. coli and S. cerevisiae are hardly affected by raspberry
ketone
The suitability of two microbial hosts for producing rasp-
berry ketone was investigated. Firstly, the maximum con-
centrations of raspberry ketone and the intermediate
product p-hydroxybenzalacetone that could be tolerated
by cultures without severely affecting microbial growth
were determined. In Table 1, the concentrations of each
compound are given leading to a reduction of 50% in the
density of the culture (IC
50
). Inhibitory concentrations are
between 0.1 and 1 g/L, indicating that microbial growth
is unlikely to be significantly affected when these mi-
crobes are used in a production system. Thus, it appears
that both E. coli and yeast are not very sensitive for either
intermediate or product of raspberry ketone synthesis.
3.3 Bioconversion of benzalacetone by E. coli
The activity of BAR can easily be recruited from fungi like
Saccharomyces [7]. However, no such reductase activity
has been documented for bacteria. To assess the reduc-
tase activity of E. coli cultures, a fresh overnight culture of
E. coli BL21 was diluted 200-fold in 2xYT medium, sup-
plemented with 10 µM p-hydroxybenzalacetone, and in-
cubated overnight. The next day, raspberry ketone was
analysed using GC-MS, and a concentration of 4 µM was
detected. Consequently, the E. coli culture demonstrated
a capacity to turn over 40% of p-hydroxybenzalacetone
into raspberry ketone.
To further characterize the BAR activity of E. coli, a
fresh overnight culture of E. coli BL21 was divided into an
extracellular fraction (centrifuged medium) and intracel-
lular fraction (pelleted cells, resuspended in buffer, soni-
cated and centrifuged again). In both fractions, hardly any
reductase activity was observed when incubated with
5 mM p-hydroxybenzalacetone. However, when 5 mM
NADPH was added, the intracellular fraction was capable
of producing 2.4 mM raspberry ketone in 30 min. The ex-
tracellular fraction remained inactive. Thus, it was con-
cluded that E. coli BL21 contains intracellular, NADPH-
dependent BAR activity.
3.4 Raspberry CHS produces benzalacetone in E. coli in vivo
A microbial production system for raspberry ketone was
designed. For the synthesis of raspberry ketone from a
simple substrate, the micro-organism would need to (i)
convert p-coumaric acid into coumaroyl-CoA, (ii) con-
dense coumaroyl-CoA into p-hydroxybenzalacetone, and
(iii) convert p-hydroxybenzalacetone into raspberry ke-
tone (Fig. 1). From previous work it is known that
coumaroyl-CoA can be provided in E. coli cells by expres-
sion of the 4CL-2 gene from tobacco, and feeding with p-
coumaric acid [12]. Moreover, it was also observed that E.
coli contains an endogenous BAR activity. We thus aimed
to establish the raspberry ketone synthesis pathway by
introducing the RiCHS-Mut protein, which seemed to be
able to convert coumaroyl-CoA into p-hydroxybenzalace-
tone. Wild-type RiCHS protein would serve as a control to
test the performance of the system by monitoring narin-
genin production.
To establish the putative production system, both
CHS variants (RiCHS and RiCHS-mut) were recloned into
E. coli vector pACYC-DUET-1, together with the 4CL-2
gene. In this system, both genes (4CL-2 and RiCHS) are
under control of the T7 promoter. In a previous publica-
tion, we described this system for stilbene synthase (STS)
[12]. Both plasmids were expressed in E. coli BL21 and
compared to the empty vector.
To test the performance of the system, fresh cultures
of pAC-4CL-RiCHS, pAC-4CL-RiCHS-Mut and pACYC-
DUET were incubated overnight with 3 mM p-coumaric
acid. Cultures were extracted with ethyl acetate, and
HPLC-PDA analyses and GC-MS analyses were per-
formed. Using HPLC analysis, a peak co-eluting with
naringenin was detected in the chromatogram at 280 nm
of the pAC-4CL-RiCHS culture extract, but not in the oth-
er extracts (Fig. 3A). This indicated that p-coumaric acid
was modified by the 4CL enzyme into coumaroyl-CoA,
which was in turn accepted by the CHS enzyme as a sub-
strate. The expected product from CHS, naringenin chal-
cone, spontaneously converts into naringenin at neutral
pH [14]. When the chromatogram at 312 nm was inspect-
ed, a significant peak was observed at the position of p-
hydroxybenzalacetone, with an absorption maximum at
323 nm. This peak was not predominantly present in the
candidate BAS strain pAC-4CL-RiCHS-Mut extract, but,
surprisingly, in the pAC-4CL-RiCHS extract, where it was
clearly distinguishable from the background (Fig. 3B). The
identity of the product was confirmed as p-hydroxyben-
zalacetone using MS (see below). Thus, we concluded
that the wild-type RiCHS was unexpectedly able to con-
vert p-coumaric acid into p-hydroxybenzalacetone in
vivo.
Biotechnology
Journal
Biotechnol. J. 2007, 2, 1270–1279
1274 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Table 1. IC
50
values of raspberry ketone and p-hydroxybenzalacetone in
E. coli and yeast
a)
p-Hydroxy-
Raspberry ketone
benzalacetone
IC
50
(g/L)
IC
50
(g/L)
E. coli BL21 0.3 ± 0.1 0.9 ± 0.2
S. cerevisiae YPH499 0.10 ± 0.03 0.5 ± 0.2
a) Fresh overnight cultures of E. coli BL21 and S. cerevisiae YPH499 were dilut-
ed 200-fold in 30 mL of 2xYT medium to which 40 μL ethanol was added
containing increasing concentrations of either p-hydroxybenzalacetone or
raspberry ketone. After overnight incubation, the density of the cultures was
measured at 600 nm.
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1275
Biotechnol. J. 2007, 2, 1270–1279 www.biotechnology-journal.com
GC-MS analysis further confirmed the establishment
of a raspberry ketone production pathway in E. coli. A
very clear peak was observed in the GC-MS chro-
matogram, recorded at m/z 164 (Fig. 3C), with the same
retention time and mass spectrum as raspberry ketone.
This peak was also present in the pAC-4CL-RiCHS-Mut
extract, but was much lower. Thus, it appeared that the
reductase activity of E. coli had converted p-hydroxyben-
zalacetone into raspberry ketone.
In a second experiment, the localization of raspberry
ketone was tested. A culture of pAC-4CL-RiCHS after it
had been fed with p-coumaric acid for 2 days was sepa-
rated into cell pellet and culture medium by centrifuga-
tion, and both fractions were extracted. It was observed
that 0.7 mg/L raspberry ketone could be isolated from the
culture medium, while only 0.1 mg/L was detected in the
bacterial pellet. This indicates that most synthesized
raspberry ketone had been secreted by the bacteria into
the medium.
3.5 Raspberry ketone production in yeast in vivo
The p-hydroxybenzalacetone synthase activity observed
for RiCHS was not restricted to the environment of a liv-
ing E. coli cell, but appeared also be achievable using oth-
er microbial systems. This was demonstrated when the
RiCHS gene was introduced in the yeast S. cerevisiae
YPH499. For expression in yeast, the 4CL gene and the
RiCHS gene or the mutant RiCHS-Mut gene were intro-
duced into the yeast expression vector pESC-Trp (Fig. 2),
which allows expression of both genes when induced by
galactose. The pESC-Trp, pESC-4CL-RiCHS and pESC-
4CL-RiCHS plasmids were introduced in the yeast strain
YPH499. Cultures (10 mL) were grown in the presence of
galactose to induce recombinant gene expression, and p-
coumaric acid (3 mM) was added, after which the cultures
were grown overnight at 30°C. The cultures were extract-
ed with ethyl acetate, and analysed using HPLC. Again,
as was the case with E. coli, the pESC-4CL–RiCHS strain
showed production of p-hydroxybenzalacetone, in addi-
tion to naringenin chalcone and naringenin (Table 2). The
extract from the yeast cultures was also analysed on an
HPLC coupled to a Q-TOF-MS (Fig. 4). Figure 3C shows
that the peak eluting at Rf =27.8 min in the HPLC-MS is
dominated by a compound of m/z 163.0761 [M+H], which
is only 1.2 ppm different from the predicted mass of p-hy-
droxybenzalacetone (C
10
H
10
O
2
; calculated mass [M+H] =
163.0759). Moreover, this compound has the same re-
tention time and absorption spectrum (λ
max
=323 nm) as a
p-hydroxybenzalacetone standard, which confirms the
presence of p-hydroxybenzalacetone in the yeast culture
medium.
In the pESC-4CL–RiCHS-Mut culture, no naringenin
or naringenin chalcone was observed, and only a low
13.60 14.00 14.40 14.80 15.20 15.60
C
t
Minutes
23.00 24.00 25.00 26.00 27.00 28.00 29.00
t
Minutes
23.00 24.00 25.00 26.00 27.00 28.00 29.00
312nm
B
m/z 164
A
280nm
MinutesMinutes
38.5 39.0 39.5 40.0 40.5 41.0 41.5 42.0 42.5 43.0
A
280nm
MinutesMinutes
38.5 39.0 39.5 40.0 40.5 41.0 41.5 42.0 42.5 43.0
Figure 3. Comparison of culture extracts from BL21-pAC-DUET1 (lower),
pAC-4CL-RiCHSmut (middle) and pAC-4CL-RiCHS (upper). (A) HPLC
analysis: area from 38 to 43 min, recorded at 280 nm. The position of
naringenin is indicated (arrow). (B) HPLC analysis: area from 23 to
30 min, recorded at 312 nm. The position of p-hydroxybenzalacetone is
indicated (arrow). (C) GC-MS chromatogram, recorded at m/z 164. The
position of raspberry ketone is indicated (arrow).
Table 2. Production of p-hydroxybenzalacetone, naringenin and naringenin
chalcone in yeast
a)
p-Hydroxy- Naringenin
Naringenin
construct benzalacetone chalcone
(mg/L)
(mg/L) (mg/L)
pESC-Trp 0.000 0.000 0.000
pESC-4CL-RiCHS 0.060 0.086 0.296
pESC-4CL-RiCHS-Mut 0.004 0.000 0.000
a) Concentrations are the average of two independent measurements.
amount of p-hydroxybenzalacetone (Table 2). GC-MS
analysis was used to detect raspberry ketone in the yeast
culture extracts. However, as observed for the production
of resveratrol in yeast [12], most p-coumaric acid was
found to be converted into hydroxyphenyl-propionic acid,
which hampered proper detection of raspberry ketone in
the GC-MS. Nevertheless, it was concluded that the
RiCHS was capable of BAS activity, also in a yeast envi-
ronment.
3.6 Raspberry ketone is an in vivo product of other enzymes
in the CHS family
In the previous paragraphs, the capability of the RiCHS
protein to drive synthesis of p-hydroxybenzalacetone in
vivo was observed. As a next step, we quantitatively com-
pared its production to that of other members from the
CHS family. The Petunia CHS gene, involved in produc-
tion of flower colour, and the grape STS, involved in the
production of resveratrol, were introduced into the same
E. coli expression system in combination with the 4CL
gene (Fig. 2). Both constructs were compared to pAC-
4CL-RiCHS and the empty vector. RiCHS was the most ef-
ficient enzyme, but all enzymes tested were capable of
producing some raspberry ketone (Table 3). The amount
of naringenin produced by raspberry RiCHS (0.44 mg/L)
only slightly exceeded the sum of the raspberry ketone
and p-hydroxybenzalacetone produced (0.39 mg/L). The
petunia CHS produced 0.24 mg/L naringenin, and
0.10 mg/L p-hydroxybenzalacetone and raspberry ketone.
The grape VvSTS was clearly more focussed on stilbene
synthesis (5 mg/L resveratrol) than on raspberry ketone
synthesis (0.07 mg/L). Thus, we conclude that the RiCHS
may be somewhat more active as a BAS than the other
two tested polyketide synthases, but it is certainly not
unique for this activity.
3.7 RiCHS is the limiting enzyme in the E. coli raspberry
ketone biosynthesis pathway
In the system described above, both RiCHS and 4CL are
under control of the same T7 promoter. We wanted to test
the role of the individual enzymes for synthesis of rasp-
berry ketone. For this, the 4CL and RiCHS genes were
placed on separate, compatible plasmids, each under
control of either the T7 promoter or the arabinose-in-
ducible P(BAD) promoter, forming plasmids pBAD-RiCHS
and pBAD-4CL (arabinose-controlled promoter) and pAC-
4CL and pAC-RiCHS (IPTG inducible) (Fig. 2). Plasmids
were combined in E. coli BL21, and production of narin-
genin and raspberry ketone was monitored using HPLC
and GC-MS.
In one experiment, E. coli with pBAD-RiCHS and pAC-
4CL was tested. The RiCHS expression was varied
(0–0.1% arabinose), while 4CL expression was kept con-
stant (1 mM IPTG) (Fig. 5A). Conspicuously, the raspber-
ry ketone formation had an optimum at 0.03–0.05% arabi-
nose, and was less efficient at higher concentration of ara-
binose, while the naringenin production was always high-
er at higher concentrations of arabinose. Apparently, BAS
activity and CHS activity behaved differently upon in-
crease of RiCHS expression.
In a second experiment, E. coli BL21 with pAC-RiCHS
and pBAD-4CL was tested. In this the 4CL expression
Biotechnology
Journal
Biotechnol. J. 2007, 2, 1270–1279
1276 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 4. p-Hydroxybenzalacetone production by yeast. Culture extracts
from YPH499-pESC-4CL-RiCHS (A) and YPH499-pESC-Trp (B) were com-
pared on HPLC-MS. Shown are total mass counts from the chromatogram
from Rf = 27 to 30 min. In (C), the mass spectrum recorded at 27.8 min is
shown. Also indicated is the structure of p-hydroxybenzalacetone, and its
calculated mass ([M+H]).
Table 3. Production of p-hydroxybenzalacetone, raspberry ketone and
naringenin in E. coli
p-Hydroxy- Raspberry
Naringenin
Construct benzalacetone ketone
(mg/L)
(mg/L) (mg/L)
pAC-DUET-1 0.000 0.000 0.000
pAC-4CL-RiCHS-mut 0.005 0.006 0.000
pAC-4CL-RiCHS 0.108 0.284 0.440
pAC-4CL-PhCHS 0.022 0.074 0.244
pAC-4CL-VvSTS 0.015 0.052 5.01
a)
a) Here the concentration of resveratrol instead of naringenin is given.
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1277
Biotechnol. J. 2007, 2, 1270–1279 www.biotechnology-journal.com
was regulated (0–0.1% arabinose), while RiCHS expres-
sion was kept constant (1 mM IPTG) (Fig. 5B). The maxi-
mal amounts of raspberry ketone and naringenin pro-
duced in this experiment were both relatively low
(0.016 mg/L and 0.07 mg/L; 7.5 and 2.5 times less than
that seen in Fig. 5A, respectively). The quantity of each
compound seemed to be related, and reached a plateau at
0.05% arabinose. These data indicate that the RiCHS is
the limiting factor in the production system.
3.8 Fermentation yields 5 mg/L raspberry ketone
E. coli BL21 containing the pAC-4CL-RiCHS vector was
grown in a fermenter, and enzyme production was in-
duced after 28 h by addition of IPTG and p-coumaric acid.
A distinct raspberry ketone peak was present in the 42-h
and 46-h samples; after 46 h this corresponded to a con-
centration of 5 mg raspberry ketone/L (Fig. 6). p-Hydrox-
ybenzalacetone was not observed, indicating that gener-
ation of p-hydroxybenzalacetone is rate limiting under
these circumstances. Induction of the cells also resulted
in the accumulation of a large amount of another com-
pound (Fig. 6: peak at 9.6 min). This compound was iden-
tified as indole by GC-MS analysis.
4 Discussion
The enzymes responsible for raspberry ketone formation
are interesting from a biotechnological point of view. Such
enzymes could be recruited for production of raspberry
ketone by micro-organisms or plants, to achieve a sus-
tainable production system for producing this flavouring
compound, to include it in artificial, but “natural” aromas.
For these reasons, identification of the enzymes has been
pursued by several groups. The focus has been on the p-
hydroxybenzalacetone synthesis, since the reductase ac-
tivity can be performed by yeasts, such as Beauveria [7].
So far, it has been assumed that raspberry contains a BAS,
which is distinguished from CHS. In rhubarb, such a BAS
has indeed been found and this enzyme produces exclu-
sively p-hydroxybenzalacetone, and not naringenin chal-
cone [6].
A biochemical strategy to isolate BAS from raspberry
was unsuccessful. When purifying BAS activity from
raspberry, Borejsza-Wysocki and Hrazdina [15] showed
that CHS and BAS co-purify, and reacted with the same
antisera. However, CHS and BAS activity in the same en-
zyme preparation showed a different response to treat-
ments with, for instance, 2-mercaptoethanol and ethylene
glycol. Also, CHS activity and BAS activity showed dif-
ferent induction patterns upon treatment of raspberry cell
cultures with yeast extract, suggesting that these en-
zymes are not one and the same molecule. However, CHS
and BAS could not be separated by biochemical tech-
niques.
A molecular strategy to identify a raspberry BAS has
so far also been unsuccessful. In total, six different rasp-
berry CHS-like proteins have been isolated by amplifying
raspberry CHS-like sequences [9, 10]. When tested in an
in vitro assay, using purified recombinant enzyme and p-
coumaroyl-CoA and malonyl-CoA as substrates, “negligi-
ble benzalacetone synthesis was observed” [10], which
we confirmed in this work. Tested in vivo, when CHS is
co-expressed with 4CL, highly significant BAS is exhibit-
ed by RiCHS. In E. coli, the production of raspberry ketone
(0.28 mg/L), and its precursor p-hydroxybenzalacetone
(0.11 mg/L) was only slightly lower than production of
naringenin (0.44 mg/L). We conclude from this finding
that the natural in vivo environment in these living cells is
conducive to p-hydroxybenzalacetone synthesis by
RiCHS. Apparently, this in vivo environment is difficult to
mimic in vitro. Nevertheless, combining 4CL activity,
CHS/BAS activity and BAR activity, provided by an en-
dogenous E. coli enzyme, leads to a system where signif-
icant amounts of the valuable flavour compound can be
produced.
Arabinose-induced RiCHS expression
constant 4CL expression
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
concentration arabinose (%)
raspberry ketone (mg/L)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
naringenin (mg/L)
raspberry ketone
naringenin
Arabinose-induced 4CL expression
constant RiCHS expression
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0.02
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
concentration arabinose (%)
raspberry ketone (mg/L)
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
naringenin (mg/L)
raspberry ketone
naringenin
A
B
Figure 5. Influence of variation in expression of RiCHS (upper) or 4CL
(lower). Cultures of BL21with pBAD-RiCHS and pAC-4CL (top) or BL21
with pBAD-4CL and pAC-RiCHS were grown overnight in the presence of
3 mM p-coumaric acid, 0.1 mM IPTG and increasing concentrations of
arabinose. HPLC analysis was used to monitor naringenin concentration
(open squares), and GC-MS to monitor raspberry ketone concentration
(solid diamonds).
The differences between the in vitro and the in vivo
properties of RiCHS are difficult to interpret. Some clue
may be provided by the experiment shown in Fig. 5. In-
creasing the amount of RiCHS in a cell (by inducing its
promoter more strongly) led to an increased CHS activity,
while BAS activity was decreased (Fig. 5A). This result
was further confirmed when an E. coli was tested where
both pAC-4CL-RiCHS (low copy) and pRSET-RiCHS (high
copy) were used for expression (not shown). In that case,
the presence of pRSET-RiCHS (producing much more
RiCHS) led to five times more naringenin, while the
amount of raspberry ketone was unaffected. Thus, also in
this case, CHS activity and BAS activity, deriving from
the same gene, respond differently to circumstances such
as enzyme concentration. The effect of enzyme concen-
tration could relate to the fact that CHS enzymes usually
occur as a homodimer [16, 17]. The increased concentra-
tion of RiCHS in Fig. 5 may have led to an increase in the
number of dimers, and a slight decrease in the number of
monomers. One could envisage that the active site pock-
et of a CHS in a dimer may have properties different from
a CHS in a monomer. In the monomer, this could possibly
lead to an increased premature release of a p-hydroxy-
benzalacetone polyketide. However, many more experi-
ments are needed to support this hypothesis.
Raspberry ketone synthesis in raspberry fruit may also
be mediated by CHS. The in vivo activity in E. coli and
yeast of a 4CL-CHS pathway may reflect the situation in
raspberry. The concentration of raspberry ketone in rasp-
berry fruit is much lower than that of anthocyanins, which
are the major end-product of CHS activity in raspberry.
Where raspberry ketone occurs in concentrations be-
tween 1 and 4 mg/kg fruit [1], anthocyanins occur at
400–500 mg/kg fruit [18]. It is easily conceivable that this
relatively small amount of raspberry ketone in the fruit is
a side product from RiCHS, and that this enzyme medi-
ates both CHS and BAS activity.
The produced quantities of both naringenin and rasp-
berry ketone do not dramatically differ from other plant-
specific phenyl-propanoids produced in microbes by re-
combinant expression. In our system, feeding p-coumar-
ic acid and expressing tobacco 4CL2 and raspberry CHS,
we observed production of 0.3 mg/L (flasks) to 5 mg/L (fer-
menter). In another study, three genes were co-expressed
in E. coli, including a phenylalanine-ammonia lyase (PAL)
from Rhodotorula rubra, 4CL from Streptomyces coelicol-
or and CHS from Glycyrrhiza echinata [19]. Using tyrosine
as a precursor, formation of naringenin was observed to a
maximum concentration of 0.45 mg/L. Similarly, the 4CL
and CHS from Arabidopsis thaliana have been expressed
in E. coli, together with the tyrosine-ammonia lyase (TAL)
from Rhodobacter sphaeroides, which led to a production
of about 20 mg/L naringenin [20].
We observed that, in our system, expression of the
RiCHS enzyme was the limiting factor (Fig. 5B). This is in
keeping with turnover numbers reported for CHS
(k
cat
=2.2/min; [21]) and 4CL (k
cat
is around 1200/min, cal-
culated from [22]), which indicate that 4CL is a much
more efficient enzyme. It is known that the CHS enzyme
is part of an enzyme complex, in which the substrate is
channelled from one enzyme to the other [23]. Deriving
both genes from the same organism may enhance this
complex formation, mediated through specific surface in-
teractions between 4CL and CHS proteins, and thereby
improve channelling of the substrate. This may lead to a
higher metabolic flux and to a higher production of valu-
able compounds.
None of the previously published studies on produc-
tion of flavonoids in E. coli reported any formation of rasp-
berry ketone or p-hydroxybenzalacetone [19, 20]. Our data
show that not only RiCHS but also PhCHS and VvSTS are
capable of BAS activity, which would suggest that also
the CHS enzymes from Arabidopsis and Glycyrrhiza pro-
duce these compounds in E. coli. However, the authors
Biotechnology
Journal
Biotechnol. J. 2007, 2, 1270–1279
1278 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 6. Fermentation of E. coli poAC-4CL-RiCHS. Cells were induced by
addition of 1 mM IPTG after 28 h of growth, at the same time 3 mM
p-coumaric acid and the culture was further grown until 46 h after inocula-
tion. An OD
600
of 20 was reached after 28 h and this remained constant af-
ter induction. Shown are compounds extracted from the culture at differ-
ent time points, and analysed on GC using flame-ionisation detection.
Compounds raspberry ketone, ethylvanillin and indole are indicated.
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1279
Biotechnol. J. 2007, 2, 1270–1279 www.biotechnology-journal.com
may have overlooked this pathway, since most p-hydrox-
ybenzalacetone is converted into raspberry ketone, and
this compound is only efficiently detected by GC-MS.
In conclusion, we report on the establishment of a mi-
crobial production system for the flavour compound rasp-
berry ketone, using enzymes from plant sources. Future
efforts will be focused on expression of this plant-specific
pathway in food-grade micro-organisms, such as Lacto-
coccus lactis [24].
The authors acknowledge Francel Verstappen and Harry
Jonker for technical advise. J.B., I.v.d.M. and R.H. ac-
knowledge Danisco Flavours for financial support. Part of
this research was carried out in the frame of the EU 6th
Framework project FOOD-CT-2005-007130 “FLORA”.
5 References
[1] Larsen, M., Poll, L., Callesen, O., Lewis, M., Relations between the
content of aroma compounds and the sensory evaluation of 10 rasp-
berry varieties (Rubus Idaeus L). Acta Agric. Scand. 1991, 41,
447–454.
[2] Smith, L. R., Rheosmin (“raspberry ketone”) and zingerone, and their
preparation by crossed aldol-catalytic hydrogenation sequences.
Chemical Educator 1996, 1, 1–18.
[3] Vandamme, E. J., Soetaert, W., Bioflavours and fragrances via fer-
mentation and biocatalysis. J. Chem. Technol. Biotechnol. 2002, 77,
1323–1332.
[4] Borejsza-Wysocki, W., Hrazdina, G., Biosynthesis of p-hydrox-
yphenylbutan-2-one in raspberry fruits and tissue cultures. Phyto-
chemistry 1994, 35, 623–628.
[5] Beekwilder, J., Jonker, H., Meesters, P., Hall, R. D., et al., Antioxi-
dants in raspberry: On-line analysis links antioxidant activity to a di-
versity of individual metabolites. J. Agric. Food Chem. 2005, 53,
3313–3320.
[6] Abe, I., Takahashi, Y., Morita, H., Noguchi, H., Benzalacetone syn-
thase – A novel polyketide synthase that plays a crucial role in the
biosynthesis of phenylbutanones in Rheum palmatum. Eur. J.
Biochem. 2001, 268, 3354–3359.
[7] Fuganti, C., Zucchi, G., Product distribution in the microbial bio-
generation of raspberry ketone from 4-hydroxybenzalacetone. J.
Mol. Catal. B Enzym. 1998, 4, 289–293.
[8] Kosjek, B., Stampfer, W., van Deursen, R., Faber, K. et al., Efficient
production of raspberry ketone via ‘green’ biocatalytic oxidation.
Tetrahedron 2003, 59, 9517–9521.
[9] Zheng, D. S., Schröder, G., Schröder, J., Hrazdina, G., Molecular and
biochemical characterization of three aromatic polyketide synthase
genes from Rubus idaeus. Plant Mol. Biol. 2001, 46, 1–15.
[10] Kumar, A., Ellis, B. E., A family of polyketide synthase genes ex-
pressed in ripening Rubus fruits. Phytochemistry 2003, 62, 513–526.
[11] Colliver, S., Bovy, A., Collins, G., Muir, S. et al., Improving the nutri-
tional content of tomatoes through reprogramming their flavonoid
biosynthetic pathway. Phytochem. Rev. 2002, 1, 113–123.
[12] Beekwilder, J., Wolswinkel, R., Jonker, H., Hall, R. et al., Production
of resveratrol in recombinant microorganisms. Appl. Environ. Mi-
crobiol. 2006, 72, 5670–5672.
[13] Abe, I., Sano, Y., Takahashi, Y., Noguchi, H., Site-directed mutagen-
esis of benzalacetone synthase – The role of PHE215 in plant type III
polyketide synthases. J. Biol. Chem.
2003, 278, 25218–25226.
[14] Jez, J. M., Noel, J. P., Reaction mechanism of chalcone isomerase –
pH dependence, diffusion control, and product binding differences.
J. Biol. Chem. 2002, 277, 1361–1369.
[15] Borejsza-Wysocki, W., Hrazdina, G., Aromatic polyketide synthases
– Purification, characterization, and antibody development to ben-
zalacetone synthase from raspberry fruits. Plant Physiol. 1996, 110,
791–799.
[16] Tropf, S., Karcher, B., Schroder, G., Schroder, J., Reaction-mecha-
nisms of homodimeric plant polyketide synthases (stilbene and chal-
cone synthase) – A single active-site for the condensing reaction is
sufficient for synthesis of stilbenes, chalcones, and 6’-deoxychal-
cones. J. Biol. Chem. 1995, 270, 7922–7928.
[17] Ferrer, J. L., Jez, J. M., Bowman, M. E., Dixon, R. A. et al., Structure
of chalcone synthase and the molecular basis of plant polyketide
biosynthesis. Nat. Struct. Biol. 1999, 6, 775–784.
[18] Koponen, J. M., Happonen, A. M., Mattila, P. H., Torronen, A. R., Con-
tents of anthocyanins and ellagitannins in selected foods consumed
in Finland. J. Agric. Food Chem. 2007, 55, 1612–1619.
[19] Hwang, E. I., Kaneko, M., Ohnishi, Y., Horinouchi, S., Production of
plant-specific flavanones by Escherichia coli containing an artificial
gene cluster. Appl. Environ. Microbiol. 2003, 69, 2699–2706.
[20] Watts, K. T., Lee, P. C., Schmidt-Dannert, C., Exploring recombinant
flavonoid biosynthesis in metabolically engineered Escherichia coli.
Chembiochem 2004, 5, 500–507.
[21] Jez, J. M., Ferrer, J. L., Bowman, M. E., Dixon, R. A. et al., Dissection
of malonyl-coenzyme A decarboxylation from polyketide formation
in the reaction mechanism of a plant polyketide synthase. Biochem-
istry 2000, 39, 890–902.
[22] Schneider, K., Hovel, K., Witzel, K., Hamberger, B. et al., The sub-
strate specificity-determining amino acid code of 4-coumarate:CoA
ligase. Proc. Natl. Acad. Sci. USA 2003, 100, 8601–8606.
[23] Winkel, B. S. J., Metabolic channeling in plants. Annu. Rev. Plant
Biol. 2004, 55, 85–107.
[24] Hernandez, I., Molenaar, D., Beekwilder, J., Bouwmeester, H. et al.,
Expression of plant flavor genes in Lactococcus lactis. Appl. Envi-
ron. Microbiol. 2007, 73, 1544–1552.
