Discovery of new biocatalysts for the glycosylation of terpenoid scaffolds

Article abstract of DOI:10.1002/chem.200800548

The synthesis of terpenoid glycosides typically uses a chemical strategy since few biocatalysts have been identified that recognise these scaffolds. In this study, a platform of 107 recombinant glycosyltransferases (GTs), comprising the multigene family of small molecule GTs of Arabidopsis thaliana have been screened against a range of model terpenoid acceptors to identify those enzymes with high activity. Twenty-seven GTs are shown to glycosylate a diversity of mono-, sesquiand diterpenes, such as geraniol, perillyl alcohol, artemisinic acid and retinoic acid. Certain enzymes showing substantial sequence similarity recognise terpenoids containing a primary alcohol, irrespective of the linear or cyclical structure of the scaffold; other GTs glycosylate scaffolds containing secondary and tertiary alcohols; the carboxyl group of other terpenoids also represents a feature that is recognized by GTs previously known to form glucose esters with many different compounds. These data underpin the rapid prediction of potential biocatalysts from GT sequence information. To explore the potential of GTs as biocatalysts, their use for the production of terpenoid glycosides was investigated by using a microbial-based whole-cell biotransformation system capable of regenerating the cofactor, UDP-glucose. A high cell density fermentation system was shown to produce several hundred milligrams of a model terpenoid, geranyl-glucoside. The activities of the GTs are discussed in relation to their substrate recognition and their utility in biotransformations as a complement or alternative to chemical synthesis.

Full text of DOI:10.1002/chem.200800548

DOI: 10.1002/chem.200800548
Discovery of New Biocatalysts for the Glycosylation of Terpenoid Scaffolds
Lorenzo Caputi, Eng-Kiat Lim, and Dianna J. Bowles*[a]
Abstract: The synthesis of terpenoid
glycosides typically uses a chemical
strategy since few biocatalysts have
been identified that recognise these
scaffolds. In this study, a platform of
107 recombinant glycosyltransferases
(GTs), comprising the multigene family
of small molecule GTs of Arabidopsis
thaliana have been screened against a
range of model terpenoid acceptors to
identify those enzymes with high activi-
ty. Twenty-seven GTs are shown to gly-
cosylate a diversity of mono-, sesqui-
and diterpenes, such as geraniol, perill-
yl alcohol, artemisinic acid and retinoic
acid. Certain enzymes showing substan-
tial sequence similarity recognise terpe-
noids containing a primary alcohol, ir-
respective of the linear or cyclical
structure of the scaffold; other GTs
glycosylate scaffolds containing secon-
dary and tertiary alcohols; the carboxyl
group of other terpenoids also repre-
sents a feature that is recognized by
GTs previously known to form glucose
esters with many different compounds.
These data underpin the rapid predic-
tion of potential biocatalysts from GT
sequence information. To explore the
potential of GTs as biocatalysts, their
use for the production of terpenoid gly-
cosides was investigated by using a mi-
crobial-based whole-cell biotransforma-
tion system capable of regenerating the
cofactor, UDP-glucose.
A high cell
density fermentation system was shown
to produce several hundred milligrams
of a model terpenoid, geranyl-gluco-
side. The activities of the GTs are dis-
cussed in relation to their substrate rec-
ognition and their utility in biotransfor-
mations as a complement or alternative
to chemical synthesis.
Keywords:
biotransformations
·
enzymes · glycosides · glycosylation ·
terpenoids
Introduction
ment[5] and the sesquiterpenoid artemisinin is the active
pharmaceutical ingredient of choice for malaria treatment.[6]
Terpenoids extracted from plants typically occur as glyco-
sides, with the sugars linked to the active groups OH and/or
COOH.[7] Glycosylation influences the chemical properties,
biological activity and utility of terpenoids. For example, in-
creased hydrophilicity impacts on emulsification and deter-
gent properties.[8,9] It is the glycosides rather than the agly-
cones that are extensively used as fragrance ingredients[10,11]
and as food flavours.[12–14] The potential of using terpenoid
glycoconjugates as slow-release aroma compounds in appli-
cations in which the glycosidic bond is cleaved either enzy-
matically or chemically has also been highlighted.[15]
Whilst terpenoid glycosides exist as natural products, their
levels in plant extracts are often limited. This has led to an
increasing interest in synthesis. Several terpenoid glycosides,
such as menthyl-glucoside, linalyl-glucoside, borneoyl-gluco-
side and terpineoyl-glucoside have been synthesized chemi-
cally by using the Koenigs–Knorr reaction involving diverse
catalysts under a range of conditions.[16–19] A limitation in
the use of chemical synthesis for glycosides of terpenoids
with secondary or tertiary alcohols has been observed in
Terpenoids represent one of the major classes of natural
products with more than 40,000 different chemical structures
described in the literature.[1] The structures are oxygen-con-
taining derivatives of terpenes, such as alcohols, aldehydes,
ketones and carboxylic acids. In nature, synthesis occurs
through either the mevalonic acid pathway or the 2-C-
methyl-d-erythritol 4-phosphate/1-deoxy-d-xylulose 5-phos-
phate pathway (MEP/DOXP pathway).[2–4] In terms of utili-
ty, terpenoids are used in many different applications, from
health care and pharmaceutical uses to colour, flavour and
fragrance compounds in food and cosmetics. Terpenoids
such as taxol already have established utility in cancer treat-
[a] L. Caputi, E.-K. Lim, D. J. Bowles
CNAP, Department of Biology, University of York
York, YO10 5DD (UK)
Fax : (+44)1904-328-786
E-mail: djb32@york.ac.uk
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
6656
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 6656 – 6662

FULL PAPER
these works. This arose from steric hindrance of the glycosy-
lation site with consequent low product yields.[16,19]
ticularly for those reactions in which a chemical approach is
neither possible nor appropriate.
Typically, current approaches in chemical synthesis in-
volve organic solvents and heavy-metal catalysts. Whilst the
research on chemical glycosylation continues, nevertheless
there is an increasing demand for natural products, directly
extracted or manufactured by using biotechnologies that are
environmentally benign. For example, recent legislation in
the US and EU have classified “natural” flavours as only
those that are extracted from natural sources or involve bio-
processes with precursors isolated from nature.[20] This clas-
sification has led to a preference for synthetic methods in-
volving biotransformations rather than the use of chemistry.
There are a number of alternative glycosylation tools that
can be applied in biotransformation strategies. These include
the use of glycosidases,[21–24] glycosynthases[25] and glycosyl-
transferases (GTs)[26] and in each case their use can be com-
bined with chemical synthesis to increase the flexibility of
the approach.[27–29] In the context of terpenoid glycosylation,
there have been no reports in the literature on the use of
glycosynthases and only one report on the use of a glycosi-
dase.[24] As yet, relatively few GTs have been identified that
recognise terpenoid acceptors. In this context, by using a
functional genomics strategy, we have identified the entire
multigene family of GTs encoded in the genome of the
model plant Arabidopsis thaliana that recognises small-mol-
ecule scaffolds. These enzymes belong to the Family 1 GTs
in the CAZy classification.[39] GTs in this family catalyse in-
verting glycosylation reactions, and typically use UDP-a-glu-
cose as a sugar donor to form b-glucoside products.[26] Other
nucleotide sugars, such as UDP-a-galactose, UDP-a-glucur-
onic acid, UDP-a-xylose and UDP-b-rhamnose, are also
known to be recognised by this family of GTs. We have
used the plant GT sequences to establish a unique platform
which can be screened for catalytic activities towards natural
and non-natural scaffolds in vitro. The platform provides the
basis for identifying and optimising novel biocatalysts, par-
In this study, we have used our platform of 107 recombi-
nant GTs to explore their activity towards a number of ter-
penoid scaffolds. The data provide insights into sequence-ac-
tivity relationships and a means to identify relevant biocata-
lysts for natural and non-natural terpenoid scaffolds. We
have also demonstrated that the biocatalysts can be used to
prepare research-scale quantities of defined glycosides by
means of a whole-cell biotransformation route.
Results and Discussion
Screen of Arabidopsis GTactivities towards model terpe-
noids: To gain an insight into the activity of Arabidopsis
GTs towards terpenoid scaffolds, members of the entire GT
multigene family were first screened against model com-
pounds in batches of six recombinant enzymes (Table 1).
Table 1. GT grouping in the initial screen.
Group
GTs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
82A1, 92A1, 84A3, 84A4, 84A2, 84A1
84B2, 84B1, 75B2, 75B1, 75D1, 75C1
74E1, 74E2, 74C1, 74D1, 74F1, 74F2
74B1, 86A1, 86A2, 87A2, 87A1, 83A1
76B1, 76E5, 76E6, 76E3, 76E4, 76E1
76E2, 76E7, 76E11, 76E12, 76D1, 76C4
76C2, 76C3, 76C5, 76C1, 76F2, 76F1
85A3, 85A1, 85A2, 85A7, 85A5, 85A4
78D1, 78D3, 78D2, 71B8, 71B7, 71B6
71B5, 71B2, 71B1, 71D1, 71D2, 71C1
71C2, 88A1, 72E2, 72E3, 73E1, 72D1
72C1, 72B2, 72B3, 72B1, 71C5, 71C4
71C3, 73C3, 73C4, 73C2, 73C5, 73C6
73C1, 73B2, 73B3, 73B1, 73B4, 76E9
73B5, 73D1, 73C7, 90A1, 90A2, 90A4
89B1, 89C1, 89A2, 79B11, 79B10, 79B9
79B8, 79B7, 79B4, 79B5, 79B6, 79B2
79B3, 79B1, 91A1, 91C1, 91B1, GST
The screen used UDP-[14C]-glucose as the sugar donor,
which was incubated with each batch of enzymes and terpe-
noid acceptor molecule, followed by analysis of the reaction
mixture by using a TLC system known to separate the agly-
cones from glycosides. The presence of radioactively la-
belled compounds was detected by using phosphor-imaging
screens and provided an indication of putative product for-
mation. The data for two model terpenoids, geraniol and far-
nesol, are illustrated in Figure 1A and B, respectively. For
example, in the geraniol screen, incorporation of [14C]-glu-
cose into putative products could be detected in reaction
mixtures from batch groups 4, 6, 8, 11, 13 and 14. In the far-
nesol screen, only four groups displayed potential activity
(groups 8, 11, 13 and 14). The comprehensive screening re-
sults of GT batches against other terpenoid scaffolds are
provided in Figure S1 in the Supporting Information.
Chem. Eur. J. 2008, 14, 6656 – 6662
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6657

D. J. Bowles et al.
Figure 1. TLC analysis of the reaction mixtures from the initial activity screen. Two examples of the activity screen are shown, A) geraniol screen and
B) farnesol screen. The groups that displayed putative activity were further analysed with the individual enzymes assayed towards the substrates.
C) and D) show the detailed analysis of group 8 towards geraniol and farnesol, respectively.
When putative products were detected in the batch
screen, individual GTs from those batches were assayed sep-
arately against the acceptors and the reaction mixtures ana-
lysed as before. In Figure 1C, the activities of the six GTs
comprising group 8 are illustrated; this shows that only four
of the six enzymes were potentially active towards geraniol.
Similarly, in Figure 1D, when batch group 8, displaying ac-
tivity towards farnesol was further analysed, incorporation
of [14C]-glucose into putative products was found in only
four reaction mixtures.
In total, this screening strategy revealed that twenty-seven
Arabidopsis GTs, from a total number of 107 enzymes, dis-
played potential activity in vitro towards the nine model ter-
penoids. The activities were restricted to groups D, E, G, H
and L of the phylogenetic tree of the GT1 multigene family
of Arabidopsis.[30] These data are summarized in Table 2,
which provides the GT, the phylogenetic group of the tree
within which it is located and its activity towards the accept-
or. The activities from this screen are defined as the per-
centage incorporation of [14C] glucose into putative product
and are colour-coded as low (1–10%), medium (10–40%) or
Table 2. Arabidopsis GT activities towards model terpenoids.[a]
[a]
: 1–10%;
: 10–40%;
: >40%.
6658
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 6656 – 6662

Glycosylation of Terpenoid Scaffolds
FULL PAPER
high (>40%). The products from these reactions were con-
of each terpenoid acceptor. The fragmentation data indicat-
firmed by HPLC-MS, as described in the next section.
ed that O-glucosides were formed.
Characterization of terpenoid glycosides: To investigate the
nature of the putative products in the reaction mixtures
from the initial screen of Arabidopsis GTs, assays were re-
peated with unlabelled UDP-glucose and products were an-
alysed by using HPLC-MS. Two representative examples are
illustrated in Figure 2 in which the reaction mixtures from
Whole-cell biotransformations of terpenoids into glycosides:
GTs are known to carry out whole-cell biotransformations
in microbial cells to form glycosides of the substrates added
to the culture medium.[26] The chemicals applied to this ap-
proach in earlier studies were all phenolic compounds.[26]
This study is the first to investigate the utility of GTs as bio-
catalysts to synthesize research-
scale quantities of terpenoid
glycosides. Each of the model
terpenoid scaffolds was provid-
ed as a substrate in a whole-cell
biocatalysis system, which con-
sisted of E. coli expressing a re-
combinant GT with known ac-
tivity towards the specific scaf-
fold. Control batches, in which
E. coli expressed only the
empty vector, were used to con-
firm that the glycoside products
were formed by the recombi-
nant GTs. After 18 h of incuba-
tion, the glycosides produced
through these processes were
purified from the culture
medium by using preparative
HPLC, and were analyzed by
HPLC-MS for purity assess-
ment and NMR spectroscopy
for identity confirmation. The
Figure 2. HPLC-MS analysis. A) and B) geranyl-glucoside. C) and D) farnesyl-glucoside.
analyses showed that the purity
of the products was >95%.
These purified glycosides were
an incubation of GT 73C5 with geraniol and GT 85A1 with
farnesol were analysed. The products were confirmed to be
geranyl-glucoside (Figure 2A and B) and farnesyl-glucoside
(Figure 2C and D). The positive ion-mode MS spectrum of
geranyl-glucoside gave two ions at m/z: 334.23 and 339.19,
used to generate the standard curves for glycoside quantifi-
cation. An example of these analyses is provided in Figure
S3 and Table S1 in the Supporting Information.
Samples from the whole-cell biotransformations were re-
analyzed and quantified by using the standard curves, and
the process yields and productivity are summarized in
Table 3. Thus, after 18 h of incubation, the molar yields of
substrates converted into products ranged from 5% for far-
nesyl-glucoside to 52.8% for menthyl-glucoside. Essentially,
+
which corresponded to the NH4 and the Na+ adducts, re-
spectively (Figure 2A). Further in-source fragmentation re-
sulted in the loss of the glucose moiety and gave rise to the
product ion at m/z: 137.15 (Figure 2B), which indicated that
a fragmentation had occurred at the glucosidic linkage. The
positive ion-mode MS spectrum of the farnesyl-glucoside
was very similar to that of geranyl-glucoside, showing the
Table 3. Small-scale (50 mL) whole-cell biotransformations after 18 h in-
cubation.
+
NH4 adduct (m/z: 402.16) and Na+ adduct (m/z: 407.11)
Product
GT
Total yield
Productivity
(Figure 2C) and the in-source fragmentation leading to the
breakage of the glucosidic bond and formation of the agly-
cone product ion at m/z: 205.22 (Figure 2D). Examples of
HPLC-MS confirmation of other terpenoid products pro-
duced by GTs from the in vitro reactions are shown in
Figure S2 in the Supporting Information.
In summary, the presence of a radioactively labelled prod-
uct in the reaction mixtures from the initial TLC screen
could be confirmed to represent the corresponding glycoside
G
]
geranyl-glucoside
citronellyl-glucoside
farnesyl-glucoside
terpineoyl-glucoside
perillyl-glucoside
linalyl-glucoside
73C5 10.7
73C5
73C5
73C5 15.8
73C5 10.3
73C5 10.7
73C5 26.4
3.4
1.1
0.9
5.0
3.2
3.4
8.4
3.7
21.4 3.8
7.2 1.3
4.6 1.0
31.6 5.6
20.7 3.6
21.4 3.8
52.8 9.7
18.6 4.1
3.6
2.3
menthyl-glucoside
artemisinic acid glucose ester 75D1 9.3
Chem. Eur. J. 2008, 14, 6656 – 6662
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6659

D. J. Bowles et al.
whole-cell biocatalysis using Arabidopsis GTs was successful
in producing milligram quantities of glycosides of each of
the monoterpenoid scaffolds and two model sesquiterpe-
noids. This whole-cell system was not able to produce the
glucose ester of retinoic acid; this may due to the instability
of the compound in the culture medium.[31]
To test the possibility of a larger-scale production of ter-
penoid glycosides in a more controlled process, glycosylation
of geraniol by E. coli expressing GT 73C5 was studied in a
3L bioreactor. During the transformation process 247 mg of
geranyl-glucoside were formed from 764 mg of geraniol
added to the bioreactor. The highest conversion was 41%
(0.62 mmol, 196 mg of product), obtained at 6 h of the pro-
duction phase (Figure 3).
of terpenoid scaffolds containing secondary and tertiary al-
cohols, the major activities were detected in groups D and
H. GTs from these groups are also known to glycosylate sec-
ondary alcohols present in a diverse range of compounds,
such as benzoates,[33] phenylpropanoids[34] and flavonoids.[35]
The broad substrate range of these GTs enhances their po-
tential utility in biocatalysis involving natural products or
non-natural scaffolds carrying secondary or tertiary alcohols.
Third, enzymes in one of the phylogenetic groups, group L,
all shared the ability to form ester linkages. This ability of
group L enzymes to recognize carboxyl groups has been pre-
viously reported for other scaffolds, such as benzoates and
phenylpropanoids,[34] and has again been demonstrated in
this study towards terpenoids (i.e. artemisinic acid and reti-
noic acid).
The aim of this study has been to establish a foundation
for the rapid prediction and selection of biocatalyst candi-
dates towards this class of substrates. This advanced knowl-
edge will aid future research in the identification of GT bio-
catalysts for terpenoid glycoside production.
Conclusion
This study has demonstrated that plant GTs can be identi-
fied that recognize terpenoid scaffolds and can be used to
produce a diverse range of natural product glycosides. The
biotransformation route of the synthesis is simple, involves a
single-step and is a useful complement or alternative to the
chemical route of production.
In comparison to the chemical route, enzymatic synthesis
does not require organic solvents and heavy-metal catalysts.
Biotransformation thereby allows terpenoid glycosides to be
made by using a “green” process that can meet regulations
for products to be classified as “natural”.[20] Also, it is signif-
icant that GTs have been identified in this study that glyco-
sylate terpenoids with secondary and tertiary alcohols. Use
of these biocatalysts overcomes the recognized difficulties in
chemical glycosylation of terpenoids, such as menthol and li-
nalool.[16,18,19]
The bioprocess can be readily scaled-up by using microbi-
al whole-cell biocatalysis systems that regenerate sugar
donors and enable good yields of glycosides without the re-
quirement for cofactor addition.[35–37] Our data underpin the
future design of novel routes of terpenoid glycoside synthe-
sis for those applications that increasingly require green
chemistry, such as those in the fine chemical, flavour, fra-
grance and food additive sectors.
Figure 3. Production of geranyl-glucoside through fed-batch fermenta-
tion. The fermentation process was divided into a growth phase and a
production phase. Bacterial cells were grown to a high density in the cul-
ture phase, and were fed with geraniol substrate during the production
phase. The quantities of geranyl-glucoside produced in the fermenter and
the amount of geraniol added during the production phase are plotted in
the insert figure.
The relation of substrate recognition and protein primary
structures: In this study, we explored the terpenoid features
recognized by the Arabidopsis GTs by using scaffolds with
different structures (linear and cyclic), sizes (C10-C20) and
functional groups (primary, secondary and tertiary alcohol,
and carboxyl group). Through the evaluation of their in
vitro activity, twenty-seven enzymes from five different phy-
logenetic groups were identified as potential terpenoid bio-
catalysts. The activity screen highlighted several distinct fea-
tures. First, and most significantly, the entire group G recog-
nized terpenoids with primary alcohols as substrates irre-
spective of linear or cyclic structure. This activity can be
found in other groups but is restricted to only a few mem-
bers, for example 4 out of 13 GTs in group D and 4 out of
19 GTs in group H. The preference of group G for primary
alcohols is also reflected in our earlier study reporting the
glycosylation of the primary alcohol on the N6 side chain of
trans-zeatin and dihydrozeatin (adenine scaffolds) by GT
85A1.[32] Thus, the group G enzymes can be prioritized can-
didates for future activity screening towards compounds car-
rying primary alcohol groups. Second, for the glycosylation
Experimental Section
Model substrates: (Æ)-Linalool, menthol, farnesol (mixed isomers), (Æ)-
b-citronellol, (S)-(À)-perillyl alcohol, terpineol (mixed isomers), geraniol
and all-trans-retinoic acid were purchased from Sigma-Aldrich. Artemi-
sinic acid (arteannuic acid) was purchased from Apin Chemicals.
6660
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 6656 – 6662

Glycosylation of Terpenoid Scaffolds
FULL PAPER
Preparation of recombinant GTs: Recombinant GTs were expressed as
fusion proteins with glutathione-S-transferase (GST) attached to the N-
terminus of the GTs, by using the GST gene fusion vector pGEX-2T
(Amersham Biotech). The recombinant proteins were prepared as previ-
ously described[30] and quantified by using the Bradford method with
bovine serum albumin as the reference.
umes of water, followed by 3 column volumes of methanol (25%). To
elute the glycosides, the column was titrated with methanol (70%). The
eluents were evaporated to dryness, suspended in methanol (50%, 2 mL)
and subjected to preparative HPLC (¾kta purifier 10, Pharmacia)
equipped with a fraction collector. Preparative HPLC was performed by
using a Luna 5 m C18 column (250”10 mm, Phenomenex) with a linear
gradient of solvent A (methanol, 10–100%) against solvent B (10 mm
ammonium acetate) over 25 min at a flow rate of 3 mLminÀ1. The
column was then washed with A (100%) for 5 min. Chromatography was
monitored at 210 nm. Fractions containing the glycosides were pooled,
evaporated to dryness and stored at 48C.
GTactivity assay : In the initial screen, each reaction mixture (20 mL)
contained Tris-HCl (100 mm, pH 7.0), UDP-[14C] glucose (3.7 mm,
11.6 GBq/mmol, Amersham), substrate (1 mm) and a pool of six different
recombinant GTs (300 ng each). The reaction mixture was incubated at
308C for 2 h and stored at À208C before TLC analysis. The reaction mix-
tures that showed positive signals in the TLC analysis were further ana-
lyzed by using the same conditions but were incubated with one enzyme
(300 ng) in each assay to identify the GTs active in the initial screen. The
active enzymes were also assessed by HPLC methods. The reaction mix-
ture (200 mL) for HPLC analysis contained Tris-HCl (100 mm, pH 7.0),
UDP-glucose (2.5 mm), substrate (1 mm) and enzyme (1 mg). The reaction
was incubated at 308C for 2 h and stored at À208C prior to HPLC analy-
sis.
Quantification of terpenoid glycosides by HPLC-MS: The quantification
of glycosides was carried out by interpolation of the peak areas obtained
by HPLC-MS with the standard curves. The standard curves were pre-
pared by HPLC-MS analysis of glycosides (0.1–1000 ng/50 mL) produced
by whole-cell transformations and purified by preparative HPLC (>95%
purity by HPLC). The HPLC-MS analysis for glucoside quantification
was performed with a SpectraSYSTEM HPLC system (ThermoQuest)
coupled with a LCQ ion trap mass spectrometer equipped with a APCI
source (Finnigan MAT). The chromatographic separation was performed
by using a Columbus 5 m C18 (150”4.6 mm, Phenomenex) column, at a
flow rate of 0.5 mLminÀ1 with a linear gradient of solvent A (methanol,
10–50%) against solvent B (10 mm ammonium acetate) over 10 min, fol-
lowed by a linear gradient A (50–100%) against B over 20 min. The
column was then washed with A (100%) for 5 min and re-equilibrated
for 5 min. The mass spectrometry analysis was performed in positive
mode (source voltage, 5.09 kV; source temperature, 4508C; nebulizing
sheath gas flow rate 63.48; auxiliary gas flow rate 29.27; capillary voltage
10.24 V; capillary temperature 514.68C). The instrument was operated at
unit resolution in full-scan MS-MS mode, scanning the product ion spec-
trum from m/z: 50–800. The LCQ was interfaced to a computer worksta-
tion running Xcalibur 2.0 software.
TLC analysis of the GT reaction mixture: TLC analysis was performed
on Silica gel 60 TLC plates in a solvent system consisting of ethylacetate/
acetone/dichloromethane/methanol/water (20:15:6:5:4, v/v/v/v/v). TLC
plates were dried and exposed to phosphor-imaging screens (Molecular
Dynamics) for 24 h. The screens were read by using a Molecular Imager
FX scanner (BioRad) supplied with Quantity One software (BioRad).
The amount of UDP-[14C]-glucose transferred by the enzymes to the sub-
strates was calculated by using a regression equation obtained by analy-
sing UDP-[14C]-glucose standards (0.008–0.555 kBq) with the TLC
method described above.
HPLC analysis of the GTreaction mixture : Reverse-phase HPLC (Spec-
traSYSTEM HPLC system and UV6000LP photodiode array detector,
ThermoQuest) was carried out by using a Columbus 5m C18 column
(250”4.6 mm, Phenomenex) at a flow rate of 1 mLminÀ1 with a linear
gradient of solvent A (methanol, 10–50%) against solvent B (10 mm am-
monium acetate) over 10 min, followed by a linear gradient A (50–
100%) over 20 min against B. The column was then washed with A
(100%) for 5 min. Chromatography was monitored at 210 nm.
Fermentation: Fed-batch fermentation was performed in a stirrer tank bi-
oreactor (3 L, Applikon). The fermentation broth (1 L) contains KH2PO4
(2 g), K2HPO4 (5.79 g), (NH4)2SO4 (0.5 g), glucose (5 g), MgSO4·7H2O
(0.31 g), trace element solution (1.25 mL), vitamin solution (1.25 mL)
ampicillin (50 mgmLÀ1).[38] Prior to inoculation, the bioreactor was condi-
tioned (378C, 150 rpm, 50% dO2 relative to air, pH 7.4) and the pH was
maintained by using NH4OH (20%) and H2SO4 (2n). An overnight cul-
ture was prepared by inoculating 2”YT (30 mL) containing ampicillin
(50 mgmLÀ1) with a freshly streaked colony and incubating at 378C with
agitation (180 rpm). The overnight culture (10 mL) was added to the con-
ditioned fermenter at the beginning of the fermentation process. After
approximately 8 h, when the glucose in the bioreactor was completely
consumed, the culture was fed started Feed 1 (60% glucose, 1.2%
MgSO4, 0.6% (NH4)2SO4, 15 mLLÀ1 trace element solution and
15 mLLÀ1 vitamin solution) at a rate of 200 mLdayÀ1. After 12 h of cul-
turing time, the dO2 was set to 25% and the temperature at 258C. IPTG
(1 mm) was then added to the culture. After 25 h post-inoculation, the
feed was changed to Feed 2 (4% glycerol, 1.2% MgSO4, 15 mLLÀ1 trace
HPLC-MS analysis of glycosides: Glycosides formed in the enzymatic re-
actions were confirmed by using an Agilent 1100 Series HPLC system
(Agilent Technologies) coupled with a QSTAR hybrid quadrupole-TOF
mass spectrometer (Applied Biosystems). HPLC was performed with a
Columbus 5 m C18 column (150”3.2 mm, Phenomenex) at a flow rate of
0.5 mLminÀ1 by following the gradient described in the previous section.
MS analysis was carried out in a positive-ion mode. The mass spectrome-
ter was operated with a capillary voltage of 4.5 kV, by using nitrogen as
the drying gas at 2008C. Ion-source fragmentation was achieved by using
declustering potentials of 10, 20 and 30 V. Full-scan spectra were record-
ed by scanning from m/z: 150 to 600 for MS analysis and from m/z: 50–
600 for MS-MS studies. Total ion current and ion traces for specific [M+
+H], [M++NH4] and [M++Na] adduct ions were used to detect the com-
pounds. MS and MS-MS analysis were performed simultaneously by
using the Information Dependent Acquisition (IDA) software tool. Data
were collected and processed by using ANALYST QS (Applied Biosys-
tems) software.
element solution and 15 mLLÀ1 vitamin solution) at
a rate of
200 mLdayÀ1 and 1 h later geraniol was added into the bioreactor at a
rate of 26.3 mghÀ1. Samples were harvested at intervals for HPLC analy-
sis.
Production of terpenoid glycosides by whole-cell biotransformations: E.
coli BL21 cultures for whole-cell biotransformations were grown over-
night at 378C in 2”YT medium (50 mL) containing ampicillin (50
mgmLÀ1). Cells were harvested by centrifugation (5,000”g, 5 min), and
suspended in M9 minimal medium (50 mL, pH 7.0) containing glucose
(1%) to an OD600 nm reading of 1.0. IPTG (1 mm) was added to the bacte-
rial cultures and terpenoid substrate (50 mmol) was added 6 h later. The
biotransformation processes were carried out for 3 days at 258C in a
shaker set at 150 rpm. Samples were harvested at intervals and analyzed
by HPLC-MS for the presence of glycosides in the medium.
Acknowledgements
The authors thank Professor J. Thomas-Oates (Department of Chemistry,
University of York, York, UK) for the interpretation of the mass spectra
and the European Union for funding (CHEMCELL project, Chemical
Biology in Reactors and Cells, contract no. MEST-CT-2004–504345).
Purification of glycosides from the culture medium: The culture broths
containing the glucoside products were harvested by centrifugation
(5000”g, 5 min). The supernatants were applied to an Amberlite XAD-2
column (100”15 mm). The column was then washed with 3 column vol-
[1] S. C. Roberts, Nat. Chem. Biol. 2007, 3, 387–395.
[2] D. J. Mc Garvey, R. Croteau, Plant Cell 1995, 7, 1015–1026.
[3] M. Rohmer, Nat. Prod. Rep. 1999, 16, 565–579.
Chem. Eur. J. 2008, 14, 6656 – 6662
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6661

D. J. Bowles et al.
[4] D. Arigoni, S. Sagner, C. Latzel, W. Eisenreich, A. Bacher, M. H.
Zenk, Proc. Natl. Acad. Sci. USA 1997, 94, 10600–10605.
[5] V. Srivastava, A. S. Negi, J. K. Kumar, M. M. Gupta, S. P. Khanuja,
Bioorg. Med. Chem. 2005, 13, 5892–5908.
[23] B. M. de Roode, M. C. Franssen, P. A. van der, R. M. Boom, Bio-
technol. Prog. 2003, 19, 1391–1402.
[24] B. M. de Roode, L. Oliehoek, P. A. van der, M. C. Franssen, R. M.
Boom, Biotechnol. Prog. 2001, 17, 881–886.
[6] C. Liu, Y. Zhao, Y. Wang, Appl. Microbiol. Biotechnol. 2006, 72,
11–20.
[25] M. Yang, G. J. Davies, B. G. Davis, Angew. Chem. 2007, 119, 1;
Angew. Chem. Int. Ed. 2007, 46, 3885–3888.
[7] D. Bowles, E. K. Lim, B. Poppenberger, F. E. Vaistij, Annu. Rev.
Plant Biol. 2006, 57, 567–597.
[8] W. Scholz, G. Shosser, W. Schneider, H. Schelges, M. Waldmann-
Lave (Henkel), US5286406, 1994.
[9] I. Johansson (Berol Nobel AB), US5644041, 1997.
[10] O. R. Jensen, T. F. Okkels (Poalis A/S), WO2006087370, 2006.
[11] H. Luther, A. Stehlin, M. Minkley (Ciba Speciality Chemicals Hold-
ing Inc.), US5980872, 1999.
[12] A. A. Murakami, A. L. Navarro, D. S. Ryder, H. Goldstein (Miller
Brewing), WO03078562, 2003.
[26] E. K. Lim, Chem. Eur. J. 2005, 11, 5486–5494.
[27] K. M. Koeller, C. H. Wong, Chem. Rev. 2000, 100, 4465–4494.
[28] S. Riva, B. Danieli, M. Luisetti, J. Nat. Prod. 1996, 59, 618–621.
[29] M. Zarevucka, M. Vacek, Z. Wimmer, K. Demnerova, M. Mackova,
Chirality 1998, 10, 676–681.
[30] J. Ross, Y. Li, E. Lim, D. J. Bowles, Genome Biol. 2001, 2, 3004.1–
3004.6.
[31] D. K. Bempong, I. L. Honigberg, N. M. Meltzer, J. Pharm. Biomed.
Anal. 1995, 13, 285–291.
[32] B. Hou, E. K. Lim, G. S. Higgins, D. J. Bowles, J. Biol. Chem. 2004,
279, 47822–47832.
[33] E. K. Lim, C. J. Doucet, Y. Li, L. Elias, D. Worrall, S. P. Spencer, J.
Ross, D. J. Bowles, J. Biol. Chem. 2002, 277, 586–592.
[34] E. K. Lim, G. S. Higgins, Y. Li, D. J. Bowles, Biochem. J. 2003, 373,
987–992.
[13] K. Imaoka, S. Kameda, T. Gomi (Delphi), JP11009194, 1999.
[14] B. Fabry (Cognis Corp.), US20060093559, 2006.
[15] A. Herrmann, Angew. Chem. 2007, 119, 1; Angew. Chem. Int. Ed.
2007, 46, 5836–5863.
´
´
´
ˇ
´
´
´
[16] J. Mastelic, I. Jerkovic, M. Vinkovic, Z. Dzolic, D. Vikic-Topic,
Croat. Chem. Acta 2004, 77, 491–500.
[17] T. Ohta, T. Omori, H. Shimojo, K. Hashimoto, T. Samuta, T. Ohba,
Agric. Biol. Chem. 1991, 55, 1811–1816.
[18] H. Paulsen, B. LÞ-NguyÞn, V. Sinnwell, V. Heemann, F. Seehofer,
Liebigs Ann. Chem. 1985, 8, 1513–1536.
[19] G. Desmares, D. Lefebvre, G. Renevret, C. Le Drian, Helv. Chim.
Acta 2001, 84, 880–889.
[35] E. K. Lim, D. A. Ashford, B. Hou, R. G. Jackson, D. J. Bowles, Bio-
technol. Bioeng. 2004, 87, 623–631.
[36] M. G. Willits, M. Giovanni, R. T. Prata, C. M. Kramer, L. De, J. C.
Steffens, G. Graser, Phytochemistry 2004, 65, 31–41.
[37] J. Arend, H. Warzecha, T. Hefner, J. Stockigt, Biotechnol. Bioeng.
2001, 76, 126–131.
[38] J. Lau, C. Tran, P. Licari, J. Galazzo, J. Biotechnol. 2004, 110, 95–
103.
[39] http://www.cazy.org/.
[20] S. Serra, C. Fuganti, E. Brenna, Trends Biotechnol. 2005, 23, 193–
198.
[21] K. Kurashima, M. Fujii, Y. Ida, H. Akita, Chem. Pharm. Bull. 2004,
52, 270–275.
[22] C. Gao, P. Mayon, D. A. MacManus, E. N. Vulfson, Biotechnol.
Bioeng. 2000, 71, 235–243.
Received: March 25, 2008
Published online: June 18, 2008
6662
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 6656 – 6662