Production of the Cytotoxic Cardenolide Glucoevatromonoside by Semisynthesis and Biotransformation of Evatromonoside by a Digitalis lanata Cell Culture

Article abstract of DOI:10.1055/s-0043-109557

Recent studies demonstrate that cardiac glycosides, known to inhibit Na ⁺ /K ⁺ -ATPase in humans, have increased susceptibility to cancer cells that can be used in tumor therapy. One of the most promising candidates identified so far is glucoevatromonoside, which can be isolated from the endangered species Digitalis mariana ssp. heywoodii. Due to its complex structure, glucoevatromonoside cannot be obtained economically by total chemical synthesis. Here we describe two methods for glucoevatromonoside production, both using evatromonoside obtained by chemical degradation of digitoxin as the precursor. 1) Catalyst-controlled, regioselective glycosylation of evatromonoside to glucoevatromonoside using 2,3,4,6-tetra- O -acetyl- α -D-glucopyranosyl bromide as the sugar donor and 2-aminoethyldiphenylborinate as the catalyst resulted in an overall 30% yield. 2) Biotransformation of evatromonoside using Digitalis lanata plant cell suspension cultures was less efficient and resulted only in overall 18% pure product. Structural proof of products has been provided by extensive NMR data. Glucoevatromonoside and its non-natural 1-3 linked isomer neo-glucoevatromonoside obtained by semisynthesis were evaluated against renal cell carcinoma and prostate cancer cell lines.

Full text of DOI:10.1055/s-0043-109557

Original Papers
Production of the Cytotoxic Cardenolide Glucoevatromonoside
by Semisynthesis and Biotransformation of Evatromonoside
by a Digitalis lanata Cell Culture*
Authors
Correspondence
Jennifer Munkert1**, Marina Santiago Franco2**,
Elke Nolte3, Izabella Thaís Silva2, Rachel Oliveira Castilho2,
Flaviano Melo Ottoni2, Naira F. Z. Schneider4, Mônica C.
Oliveira2, Helge Taubert3, Walter Bauer5, Saulo F.
Andrade6, Ricardo J. Alves2, Cláudia M. O. Simões4, Fernão
C. Braga2, Wolfgang Kreis1, Rodrigo Maia de Pádua2
Jennifer Munkert
Department of Biology, Division of Pharmaceutical Biology,
Friedrich-Alexander Universität, Erlangen-Nürnberg
Staudtstraße 5, 91058 Erlangen, Germany
Phone: + 4991318528251, Fax: + 4991318528243
jennifer.munkert@fau.de
Affiliations
Supporting information available online at
1
2
3
4
5
6
Department of Biology, Friedrich-Alexander-Universität,
Erlangen-Nürnberg, Germany
http://www.thieme-connect.de/products
Department of Pharmacy, Universidade Federal of Minas
Gerais, Belo Horizonte, Brazil
ABSTRACT
Recent studies demonstrate that cardiac glycosides, known to
inhibit Na+/K+-ATPase in humans, have increased susceptibili-
ty to cancer cells that can be used in tumor therapy. One of
the most promising candidates identified so far is glucoeva-
tromonoside, which can be isolated from the endangered
species Digitalis mariana ssp. heywoodii. Due to its complex
structure, glucoevatromonoside cannot be obtained econom-
ically by total chemical synthesis. Here we describe two meth-
ods for glucoevatromonoside production, both using evatro-
monoside obtained by chemical degradation of digitoxin as
the precursor. 1) Catalyst-controlled, regioselective glyco-
sylation of evatromonoside to glucoevatromonoside using
2,3,4,6-tetra-O-acetyl-α-D-glucopyranosyl bromide as the
sugar donor and 2-aminoethyldiphenylborinate as the cata-
lyst resulted in an overall 30% yield. 2) Biotransformation of
evatromonoside using Digitalis lanata plant cell suspension
cultures was less efficient and resulted only in an overall 18%
pure product. Structural proof of products has been provided
by extensive NMR data. Glucoevatromonoside and its non-
natural 1–3 linked isomer neo-glucoevatromonoside ob-
tained by semisynthesis were evaluated against renal cell car-
cinoma and prostate cancer cell lines.
Department of Urology, Universitätsklinikum Erlangen,
Erlangen, Germany
Department of Pharmaceutical Sciences, Universidade
Federal de Santa Catarina, Florianópolis, Brazil
Department of Chemistry and Pharmacy, Friedrich-
Alexander Universität, Erlangen-Nürnberg, Germany
Department of Pharmaceutical Sciences, Universidade
Federal de Rio Grande do Sul, Porto Alegre, Brazil
Key words
Cardiac glycosides, Digitalis mariana, Digitalis lanata,
Plantaginaceae, regioselective glycosylation, plant cell
suspension culture
received January 31, 2017
revised
March 31, 2017
accepted April 12, 2017
Bibliography
DOI https://doi.org/10.1055/s-0043-109557
Published online | Planta Med © Georg Thieme Verlag KG
Stuttgart · New York | ISSN 0032‑0943
CGs came into focus for their antiviral [2–6] and anticancer activ-
ities as reviewed in [7–11]. CGs inhibit the Na+/K+-ATPase signal-
osome complex at nanomolar concentrations leading to the acti-
vation of various signaling pathways that provoke cell death. In
this respect, digitoxigenin monodigitoxoside [evatromonoside
Introduction
Cardiac glycosides (CGs) are used to treat congestive heart failure
and arrhythmias, since they are known to bind and inhibit Na+/
K+-ATPase, leading to a positive inotropic effect [1]. More recently,
*
Dedicated to Professor Dr. Max Wichtl in recognition of his outstanding
contribution to pharmacognosy research.
** These authors contributed equally to this manuscript.
Munkert J et al. Production of the… Planta Med

Original Papers
▶ Fig. 1 Schematic overview for obtaining GEV (3) by different processes. EV (2) is the common precursor for both approaches. The product
GEV (3) will either be produced by chemical synthesis or biotransformation reaction with plant cell suspension cultures.
(EV) (2)] (▶ Fig. 1) turned out to be a very promising candidate,
which was investigated by Schneider et al. [12] and Wang et al.
[13]. Other targets of CGs are viral infections, such as those
caused by human Herpes Simplex Virus (HSV). Out of 65 tested car-
denolides, Bertol and coworkers [5] attributed the strongest anti-
herpes effects to EV and to its naturally 4′-glycosylated derivative,
glucoevatromonoside [GEV (3)] (▶ Fig. 1). These compounds in-
hibited HSV-1 and HSV-2 replication at nanomolar concentrations
and the mechanism of action of GEV was elucidated. Additionally,
promising cytotoxic effects, including reduction of proliferation,
migration and invasion in A549 (non-small cell lung cancer) and
U937 cells (human leukemic monocyte lymphoma cell line) were
detected for GEV [14].
Two alternative approaches to obtain GEV and also EV are de-
scribed here. For the production of EV, we propose the rather sim-
ple chemical degradation of digitoxin [DTG, (1)] (▶ Fig. 1), which
can be isolated from any Digitalis extract enriched in A-series car-
denolides. To produce GEV, we established a biotransformation
process using suitable precursors and D. lanata plant cell suspen-
sion cultures, as well as a semisynthetic approach based on a com-
bination of two previously described methods [18,19] (▶ Fig. 1).
Besides the desired GEV (3), the non-natural 3′-glycosylated de-
rivative, neo-glucoevatromonoside [(N‑GEV, (4)], was also ob-
tained and both compounds had their activity evaluated against
tumor cell lines.
Despite their potential anticancer and antiviral effects, CGs
such as EV and GEV are still isolated from plant sources, since total
synthesis is hampered by their structural complexity. GEV is not
even commercially available and the sample employed in our pre-
vious studies [5,14] was isolated from Digitalis lanata leaves [15].
Whereas GEV and EV are only minor CGs in this species, GEV is a
major CG in Digitalis heywoodii (syn. Digitalis mariana ssp. heywoo-
dii, Heywoodʼs foxglove) [16, 17]. Therefore, this species is a
promising source for large-scale production and isolation of GEV.
Unfortunately, Heywoodʼs foxglove is an endemic and endan-
gered species growing only in a few places in the Iberian Penin-
sula. A recent study showed a high content of GEV in D. mariana
ssp. heywoodii plants, which were regenerated from seed-derived
shoot explants through in vitro propagation [17]. This approach is
very labor-intensive and time-consuming, and significant
amounts of GEV have so far only been demonstrated in green
house cultivated plants, but not in field trials.
Results and Discussion
DTG was used as the starting material in the production of EV. It
was isolated from a D. lanata extract enriched in A-series cardeno-
lides. The extract was fractionated by flash chromatography on a
silica gel column and yielded 1 g of DTG per 4 g of extract. Frac-
tions containing DTG were combined and used to prepare EV.
The purity was assessed by UPLC‑ESI‑MS (Fig. 1S, Supporting In-
formation). The chemical degradation of DTG to EV was carried
out according to the method described in US Patent 3,856,944
(1974) [20] with small changes regarding solvents, the molar ratio
of reactants, and reaction conditions. Oxidation of digitoxin using
NaIO4 resulted in dialdehyde I with a 95% yield. This was subse-
quently treated with NaBH4 to produce dimethylol I (95% yield).
This reaction was carried out in MeOH instead of EtOH to improve
NaBH4 solubility. The molar proportion of NaBH4 was increased to
Munkert J et al. Production of the… Planta Med

▶ Fig. 2 Glycosylation reaction. A UPLC chromatogram of glycosylation reaction 1. Compound 2: substrate EV; compound 6: (2,3,4,6)-tetra-O-
acetly-neo-glucoevatromonside, Rt = 6.57 min; compound 5: (2,3,4,6)-tetra-O-acetly-glucoevatromonside, Rt = 6.86 min, detected by a PDA 2996
photo diode array detector. B, C MS spectrum of both products 6 and 5, mass 834 m/z. (7) unidentified product of identical mass.
▶ Table 1 Profile of the intermediated products acetylated GEV (5) and acetylated N‑GEV (6) obtained in different conditions for the glycosylation
of EV (2) calculated w/w.
Reaction
Reaction conditions
Literature conditions
Increased amounts of reagents
Dark
Substrate (2) [%]
Product (5) [%]
Product (6) [%]
Ratio 5/6
80:20
78:22
76:24
77:23
1
2
3
4
32
8
54
72
66
72
14
20
21
21
13
7
Optimized
10 instead of 4, which reduced the reaction time by 4 h. Dimethyl-
ol I was hydrolyzed using 0.05 N hydrochloric acid, resulting in
digitoxigenin-bisdigitoxoside with a 90% yield. This compound
was further oxidized to dialdehyde II (95% yield). The second re-
duction step afforded dimethylol II (88% yield), which was finally
hydrolyzed to EV with an 84% yield. This technique resulted in an
EV production yield of 70% overall on a molar basis. The identity
of EV was confirmed by UPLC‑ESI‑MS and NMR analyses. LC/MS
analysis disclosed the characteristic peak of the molecular ion at
m/z 504 (Fig. 2S, Supporting Information). All 1H and 13C reso-
nance signals in the NMR spectra of EV were assigned by a combi-
nation of 1D and 2D experiments (DEPT, HMQC, and HMBC).
Moreover, the spectroscopic data allowed us to confirm the purity
of the obtained EV (Fig. 5S, Supporting Information). An alterna-
tive approach to synthesize EV has been previously reported using
palladium-catalyzed digitoxosylation of digitoxigenin [19]. This
method also resulted in high yields of the final product. However,
in addition to digitoxigenin, digitoxose had to be obtained and
modified appropriately to carry out the reaction. Therefore, the
chemical degradation of digitoxin described here offers a more ef-
ficient and alternative approach to produce EV that can be directly
used as starting material for biotransformation experiments or
chemical synthesis. The method also produced digitoxigenin-bis-
digitoxoside, another bioactive cardenolide non-commercially
available, and is thus more versatile than the palladium-catalyzed
glycosylation.
O-Glycosylation of natural products is a valid strategy to mod-
ify the biopharmaceutical properties of bioactive compounds. In
addition to chemical synthesis, enzyme catalysis, either in vivo
(metabolic engineering, biotransformation) or in vitro (chemo-
enzymatic methods), is a promising approach [18, 21–23]. Cata-
lyst-controlled, regioselective glycosylation of DTG was described
by Beale and Taylor [18] using 2-aminoethyldiphenylborinate and
silver (I) oxide as the catalysts responsible for the regioselectivity
of the reaction [18] (▶ Fig. 1). Stereoselectivity of the reaction
was explained by a selective activation of the equatorial OH group
of cis-1,2-diols motifs of an appropriate pyranoside substrate
emerged by using borinic ester and silver (I) oxide as a catalyst
[18]. We adapted their protocol, described for the synthesis of
purpureaglycoside A and its analogs, to produce GEV using EV ob-
tained by DTG degradation as the educt (see above). The reaction
yielded three products, two of them were identified by UPLC‑
ESI‑MS as 2,3,4,6-tetra-O-acetyl GEV (5) and 2,3,4,6-tetra-O-ace-
tyl-N‑GEV (6) derivatives (Rt = 6.57 min and Rt = 6.86; m/z 834,
respectively) (▶ Fig. 2). The third compound (7) was tentatively
identified as an orthoester regioisomer of GEV, the formation of
which has been previously described [18]. The 2,3,4,6-tetra-O-
acetyl GEV derivative 5 was obtained as the main product. The
ratio of compound 5 to compound 6 was around 80:20, as
indicated by UPLC‑ESI‑MS analyses and NMR data (▶ Fig. 2 and
Table 1).
The method was further optimized to meet our demands. We
found that the molar ratio of acetobromoglucose (ABG) was not
Munkert J et al. Production of the… Planta Med

Original Papers
▶ Fig. 3 Biotransformation reaction of EV (2) by DlK1OHD suspension cells. A GEV (3) production is depicted in relation to the EV concentration
added in mM in cells. B The EV (2) level was followed over time in media. Error bars represent the SD of three biological replicates each consisting
of three technical replicates.
the limiting factor of the reaction. The reaction was conducted in
the dark to stabilize the ABG, and/or in smaller volumes to optimize
the surface/volume ratio. Under optimized conditions, 93% of the
educt was consumed after 20 h, yielding mainly compound 5
(▶ Table 1). The reaction carried out without the catalyst 2-amino-
ethyl-diphenylborinate yielded only small amounts of 6. Different
reactional conditions were tested and are summarized in ▶ Table
1, along with the ratios of the products obtained, according to
UPLC‑ESI‑MS analyses. An 80:20 ratio of compounds 5 and 6 in
the reaction favored glycosylation at the 4′-O position. However,
selectivity was lower compared to that obtained by Beale and Tay-
lor [18], who reported a 95:5 ratio in favor of the 4′′′-O-glycosyla-
tion of digitoxin. Beale and Taylor [18] described that the 4′′′-OH
group of digitoxin experienced efficient organoboron-catalyzed
glycosylation, as it is the most sterically accessible position. Using
EVas the educt, we observed ashift towards glycosylation at the 3′-
O position, though glycosylation at the 4′-O position still is the fa-
vored reaction. This let us assume that EV provides a 3′-hydroxyl
group more easily sterically accessible than the 3′′′-OH of digitoxin
for chemical glycosylation, as it is only bound to the steroid agly-
cone. Using 2-aminoethyldiphenylborinate and silver (I) oxide as
catalysts clearly force regioselective glycosylation on the 4′-O posi-
tion compared to non-catalyst controlled glycosylation reaction.
The non-catalyst controlled reaction resulted in only less conver-
sion of EVand a low amount of product glycosylated in the 3′-O po-
sition.
Extending the reaction time to about 30 h promoted the cleav-
age of the acetyl groups, resulting in a mixture of GEV (3) and
N‑GEV (4) that could not be separated easily due to their similar
polarities. For this reason, the acetylated derivatives of the two
GEV isomers were first isolated and purified. The recovery rates
of compounds 5 and 6 after purification were 45 and 90%, respec-
tively (Fig. 3S, Supporting Information). Compounds 5 and 6 were
then deacetylated [19] (▶ Fig. 1) to yield GEV (3), naturally glyco-
sylated in the 1–4 linkage (77% yield), and N‑GEV (4), non-natu-
rally glycosylated in 1–3 (70% yield), respectively. The identity
and purity of the obtained compounds were accessed by UPLC‑
ESI‑MS and NMR analyses (Fig. 4S, Supporting Information). Gly-
cosylation at the 4′-OH position of GEV (3) was confirmed by the
significant downfield shift of the C-4′ resonance signal as com-
pared to a ring carbon with the free OH-group (▶ Fig. 1 and
Fig. 5S, Supporting Information). Moreover, the downfield 13C sig-
3
nal of C-4′ exhibits a J-coupling with the methyl protons of C-6′
(via HMBC), thus confirming the 4′-OH glycosylation. In summary,
the catalyst-controlled, regioselective glycosylation of EV to GEV
resulted in a 30% overall yield on a molar basis. All individual
chemical reaction starting from the degradation of DTG leading
to EV and the further glycosylation as well as hydrolysis to obtain
GEV yielded conversion rates higher than 70%. The chemical
semisynthesis of GEV therefore offers a potential tool to provide
required amounts to supply further in vitro and in vivo studies.
Digitalis suspension culture cells are able to transform exoge-
nous cardenolides and other steroids [24–26]. Depending on their
chemical structure, exogenous cardenolides may undergo glyco-
sylation/deglycosylation and acetylation/deacetylation reactions
in the side chain, along with hydroxylation of the steroid scaffold
[27–29]. Although several glycosylation reactions such as gluco-
sylation, fucosylation and digitalosylation were reported to take
place in D. lanata tissue cultures [30], digitoxosylation of digitoxi-
genin has not yet been reported to occur in vivo in plant tissue cul-
tures. Therefore, we did not aim at the production of evatromo-
noside using a biotransformation process. Since DlK1OHD cell sus-
pension cultures were described as possessing high glycosyltrans-
ferase activity [25] and to be able to convert EV into GEV [27,29],
we used this particular cell line to establish a biotransformation
process for the production of GEV.
EV substrate concentrations between 0.3–1.2 mM were added
to cell suspension cultures 3 days after transfer into fresh culture
medium followed with GEV accumulation in the cells. The highest
product levels were obtained after administration of 1.2 mM EV
and the highest substrate:product ratio (1:1) was obtained when
0.3 mM EV was given (▶ Fig. 3). The highest glycosylation rates
were recorded at 3 to 7 days after substrate administration. This
was also reported for digitoxin glycosylation using the same cell
culture [25]. GEV concentration decreased after 7 days, probably
as a result of degradation [27, 30]. A series of attempts were car-
ried out to increase the biotransformation efficacy. For example,
10 mM ascorbate or 50 µM MeJA were added to the suspension
Munkert J et al. Production of the… Planta Med

gated cell lines. EV and GEV showed higher cytotoxicity compared
to DTG towards all four tested cancer cell lines. The DU145 prostate
cancer cell line and 786-O renal cell carcinoma cell line expressed
similar sensitivity for EV and GEV. However, especially for the
Caki-1 renal cell carcinoma cell line and the PC3 prostate cancer cell
line, GEV (4) was the most bioactive and cytotoxic compound. It
was already shown that cardiac glycosides might have a different
impact on various cancer cell lines [36]. In case of, non-small cell
lung cancer A549 cells, EV seems a very promising candidate [12],
and a well-established synthesis for this compound is an additional
advantage of this study. However, in the cases of renal and prostate
cancer cell lines mentioned above as well as in antiviral activity, GEV
is far more active [5]. The IC50 values obtained for the synthetic
non-natural isomer N‑GEV were significantly higher than those of
GEV (▶ Table 2). A similar pattern of cytotoxicity was shown for
the acetylated compounds 5 and 6 against PC3 cell lines. The IC50
value of 5 was determined to be 37.23 nM and 6 was 390 nM. Along
with the IC50 values of GEV and N‑GEV, this indicates the impor-
tance of a free 3′-OH group within the sugar moiety of the CGs. This
coincides with the values for the inhibition of the α1, 2, 3 subunit of
the porcine Na+/K+-ATPase (▶ Table 2). The inhibition rate in that
case was not measured for a specific Na+/K+-ATPase isoform, but of-
fers a clue on the overall inhibition activity. A free hydroxyl group at
position 3′ seems to be important for the bioactivity. The reduced
activity of N‑GEV might be associated with the lower binding affin-
ity towards Na+/K+-ATPase rather than enzyme inhibition. This can
be explained as the sugar moiety is not necessary for the inhibition
of the Na+/K+-ATPase [37]. However, the glycosylation pattern and
hydrophilic surface have an influence on the binding affinity. Steric
hindrance also has a great impact on the proper orientation of the
molecule in the enzyme complex [38]. In addition, the sugar moi-
ety is described as the key feature towards specificity of CG on
Na+/K+-ATPase binding [38, 39]. Amino acids within the core re-
gions of the protein are highly conserved, however, amino acids
around the entrance of the CG binding cavity show isoform specific
differences [38,40]. The non-natural 1–3 linkage found in N‑GEV is
influencing the binding affinity. The sugar residues have been dem-
onstrated to stabilize the enzyme-CG complex, especially by the in-
teractions of the 3′-α-hydroxyls with both proton-donating and
proton-accepting groups on the enzyme [41]. Possible amino acids
for the interaction with the 3′-α-hydroxyl group are Glu-319 and
Arg-887 (aa numbering for the shark Na+/K+-ATPase) [42]. For
N‑GEV, the position of 3′-α-hydroxyls is blocked, and therefore
the stability of the CG-enzyme complex is decreased, resulting in
a lower bioactivity.
▶ Fig. 4 Glycosylation of EV (2) by biotransformation reaction under
different conditions. Influence on GEV (3) formation, extracted
from the suspension cells, in heat, and darkness versus control reac-
tions performed in light. Dashed line marks 50% conversion rate.
Error bars represent the SD of at least three biological replicates each
consisting of three technical replicates.
cultures. MeJA is known to induce the biosynthesis of various nat-
ural products in planta and can also affect cardenolide biosynthe-
sis [31]. However, it had no influence on the desired biotransfor-
mation reaction. Biotransformation experiments were also per-
formed in dark and light conditions, but the production rates
were similar (▶ Fig. 4). Cultivation in the dark is the preferred op-
tion since no chlorophyll is formed, which may impair purification,
and energy costs are reduced (▶ Fig. 5).
In order to obtain GEV and to investigate and compare the effi-
cacy of the biotransformation reaction, EV was added to batch
cultivation of DlK1OHD suspension cells at a concentration of
0.3 mM, a similar concentration used in the chemical reaction.
The two-step purification afforded 98% pure GEV. Using the
added substrate as an indicator, GEV resulted in an 18% overall
yield on a molar basis.
At this stage, chemical synthesis starting from EV seems to be
the more promising and straightforward approach to produce
GEV in bulk amounts than glycosylation by plant cell cultures.
The biotransformation process can be further optimized and
scaled up to larger vessels and higher cell density to improve the
production rates. In addition, cell cultures with a higher glycosyla-
tion capacity might well be selected by cell aggregate cloning as
described earlier [25]. Another approach would be the identifica-
tion of glucosyltransferase genes encoding enzymes capable of
converting EV or similar cardenolides. These enzymes can be re-
combinantly expressed in Escherichia coli or yeast, and these
strains can be used for biotransformation in vivo or in vitro as pre-
viously described for flavonoids [32–34]. Work in this area is in
progress. Although it is feasible to carry out enzymatic conver-
sions in vitro, the use of glucosyltransferases is a less promising al-
ternative since they require activated sugars as co-substrates [34].
The effect of the synthesized EV (2), GEV (3), and N‑GEV (4) on
the viability of renal cell carcinoma (Caki-1, 786-O) and prostate
cancer (DU-145 and PC3) cell lines was investigated and compared
to DTG (1) determined by [35,36]. All cardiac glycosides exhibited
cytotoxic effects against renal and prostate cancer cell lines
(▶ Table 2), however, they differed in their impact on the investi-
In summary, we described two different approaches to obtain
the antiviral and cytotoxic cardiac glycoside GEV. The synthesis of
EV from DTG was optimized, so EV could be used as a precursor
for both approaches and also for further pharmaceutical studies,
as EV also shows promising anticancer activity [12]. The molar
product yield of GEV obtained by the biotransformation reaction
with plant suspension cultures was only half of that achieved by
chemical semisynthesis. The straightforward chemical synthesis
approach produces bulk amounts of the non-commercially avail-
able GEV, which are required for further pharmacological and
pharmaceutical studies, including in vivo studies. In addition, the
reaction side product N‑GEV showed lower cytotoxic activity
Munkert J et al. Production of the… Planta Med

Original Papers
▶ Fig. 5 DlK1OHD suspension cell extract of biotransformation reaction with EV (2) as the substrate. A Chromatogram of extracted cells, 5 h after
feeding in light conditions. B Chromatogram of extracted cells, previously adapted to the dark, 5 h after feeding in dark conditions. GEV (3):
Rt = 14.8 min; EV (2): Rt = 22.0 min; internal standard digitoxigenin (IS): Rt = 23.8 min. For chromatographic conditions, see Material and Methods.
▶ Table 2 Cytotoxic effects of GEV and N‑GEV on renal cell carcinoma (Caki-1,786-O) and prostate cancer (DU-145 and PC3) cell lines and inhibi-
tion of Na+/K+-ATPase. IC50 values for cytotoxicity are in nM (48 h) and IC50 values for the α1, 2, 3 subunit of the porcine Na+/K+-ATPase inhibition
are in µM. The IC50 values represent the mean from three experiments each consisting of three technical replicates; values for DTG were evalutated
by * [35] and † [36].
Compound
GEV (3)
Caki-1
786-O
25.54 nM
1513 nM
21.27 nM
57.13 nM†
DU-145
27.82 nM
1998 nM
27.30 nM
43.80 nM*
PC3
38.01 nM
Inhibition Na+/K+-ATPase
5.50 µM
25.66 nM
n.d.
N‑GEV (4)
EV (2)
911.9 nM
69.53 nM
121.8 nM*
14.0 µM
54.69 nM
33.31 nM†
2.57 µM
DTG (1)
12.1 µM†
against tumor cell lines than GEV. This may be explained by a low-
er affinity of N‑GEV to the targeted Na+/K+-ATPase or a reduced
stability of the inhibitor-enzyme complex.
Detailed isolation and purification protocol is provided in Support-
ing Information.
General procedure for preparation of evatromonoside
The synthesis of EV was based on the US Patent 3,856,944 (1974)
[20], according to the examples 1, 18, and 7 thereof. Some mod-
ifications were introduced regarding solvents, the molar ratio of
reactants, and the hydrolysis conditions to optimize the prepara-
tion procedure. The reaction was conducted in two main steps.
The first one includes the obtaining of digitxogenin-bisdigitoxo-
Material And Methods
Isolation of digitoxin from Digitalis lanata extract
Digitoxin was isolated and purified from a D. lanata methanolic ex-
tract enriched in A-series cardenolide by flash chromatography.
Munkert J et al. Production of the… Planta Med

side and the second the preparation of evatromonoside. Detailed
synthesis protocol is available in Supporting Information.
For each reactional condition assayed, the resulting materials
were purified by flash column chromatography (0 = > 100% ethyl
acetate in dichloromethane, 10 mL fractions, 0, 20, 40, 50, 60,
80, 100%).
Chemical synthesis of glucoevatromonoside
The chemical synthesis of GEV consisted of two individual reac-
tion steps containing an intermediate purification step of the in-
terim reaction compounds. First, a glycosylation reaction was per-
formed after Beale and Taylor [18], whereas removal of the acetyl
groups was based on Zhou and OʼDoherty [19].
Literature reaction (reaction 1). The reaction was performed
according to Beale and Taylor [18], with EV as the substrate. To
EV (50 mg, 0.1 mmol) and Ag2O (47 mg, 0.2 mmol), 2-amino-
ethyldipheynlborinate (5.6 mg, 0.025 mmol) in 2 mL dried CH2Cl2
and ABG (82.5 mg, 0.2 mmol) in 3 mL CH2Cl2 were added. The re-
action was stirred for 20 h at room temperature. The reactional
medium was diluted with dichloromethane, filtrated, and the or-
ganic solvent was evaporated under reduced pressure for purifica-
tion.
Hydrolysis of acetyl groups. LiOH*H2O (30 mg, 715 µmol, 1:3)
was added to a solution of the acetylated product (200 mg,
240 µmol) in a mixture of MeOH/H2O (4:1, 50 mL), and the reac-
tion was stirred for a maximum of 1 h at room temperature. The
reaction was then neutralized with 0.05 M HCl (17 mL). The organ-
ic solvent was reduced and the remaining residue was diluted with
water (17 mL). The mixture was extracted with 3 × 50 mL EtOAc
and washed with 3 × 50 mL water. The organic layer was dried
over anhydrous sodium sulfate and evaporated to dryness. The re-
action was analyzed by UPLC‑ESI‑MS and TLC. For the latter, a mo-
bile phase of methanol:acetone (5:95) was used.
Biotransformation experiments and purification
of glucoevatromonoside
Reaction with double proportion of reagents (reaction 2). The
reaction was performed as described for reaction 1, but with
changes in the amounts of reagents: EV (100 mg, 0.2 mmol) and
Ag2O (95 mg, 0.4 mmol) were added to 2-aminoethyldipheynl-
borinate (11.25 mg, 0.05 mmol) in 2 mL dried CH2Cl2 and ABG
(165.2 mg, 0.4 mmol) in 3 mL CH2Cl2. The reactional medium
was diluted with dichloromethane, filtrated, and the organic sol-
vent was evaporated under reduced pressure for purification.
Reaction in the dark (reaction 3). The reaction was performed
as described for reaction 1. However, the suspension was stirred
for 20 h at room temperature in the dark. Afterwards, the reaction
medium was diluted with dichloromethane, filtrated, dried, and
the reaction mixture was analyzed on TLC and UPLC‑ESI‑MS.
Optimized reaction conditions (reaction 4). EV (100 mg,
0.2 mmol) and Ag2O (95 mg, 0.4 mmol) were added to a glass
flask containing a stir bar. The substrate was dissolved by adding
the solutions of 2-aminoethyldiphenylborinate (11.25 mg,
22.5 mM, 0.05 mmol) in dried CH2Cl2 (2 mL) and acetyl protected
glycosyl bromide (2,3,4,6-tetra-O-acetyl-α-D-glucopyranosyl bro-
mide, ABG, 165.2 mg, 133 mM, 0.4 mmol) in dried CH2Cl2 (3 mL).
The resulting suspension was stirred at room temperature in total
for 20 h in the dark. After 15 h, the reaction was monitored by
TLC. A mobile phase of ethyl acetate:dichloromethane (1:1) was
used for elution, and detection was made by spraying anisalde-
hyde/sulphuric acid and Kedde reagent. In case there was still
more than around 30% of substrate remaining, quantified by Im-
ageJ software (https://imagej.nih.gov/ij/, 1.48 v), the volume of
the reaction mixture was reduced by evaporation and the reaction
was stopped, at the latest, after 20 h by diluting the suspension
with CH2Cl2 and filtrating the reaction mixture twice. The organic
solvent was evaporated and the reaction mixture was purified by
silica gel column chromatography.
Suspension cell cultures K1OHD of D. lanata that were described
having a high glycosyltransferase activity [25] were subcultured
weekly to fresh MS media. Fifteen g of cells were transferred to
an Erlenmeyer flask containing 100 mL of growth medium. Erlen-
meyer flasks were kept on rotary shakers (110 rpm) at 21 2°C
under 16 h of light and 8 h of dark.
Erlenmeyer flasks containing 100 mL of growth medium and
15 g of cell suspensions culture of D. lanata, 3 days old, were feed
with 1 mL of the substrate solution in DMSO (fc: 0.3–1.2 mM)
under aseptic conditions. Each experiment included a control with
1 mL of DMSO and a media control with the respective substrate.
After feeding, the Erlenmeyer flasks were kept on a rotary shaker
(110 rpm) at 21 2°C under 16 h of light and 8 h of dark or com-
pletely in the dark. To enhance the glycosylation reaction, 50 µM
methyl jasmonate and 10 mM ascorbate were added to the sus-
pension cell culture. In addition, the biotransformation experi-
ment was conducted at 35°C. Cell material (1 g) and 1 mL of
growth medium aliquots were removed for analysis after intervals
of 1, 24, 48, 96, 168 and 240 h of feeding.
The frozen cells (300 mg) were extracted with 600 µL of HPLC
grade methanol in a 2-mL Eppendorf flask by vortexing and keep-
ing the mixture in an ultrasonic bath for 10 min at 40°C. Ten µL of
methanol solution of digitoxigenin (stock solution: 10 mg/mL)
were added as the internal standard. Afterwards, the extract was
centrifuged for 10 min at 10000 g and the supernatant was col-
lected and analyzed by HPLC‑ESI‑MS or TLC.
The growth medium (0.5 mL) was extracted with 250 µL of
chloroform:isopropanol (3:2, v/v) in a 2-mL Eppendorf flask.
Ten µL a methanol solution of digitoxigenin (stock solution
10 mg/mL) were added as the internal standard. The mixture was
kept in an ultrasounic bath at room temperature for 5 min. The
extract was centrifuged for 5 min at 10000 g and the organic
phase was separated and evaporated in a heating block (34°C) to
dryness. The samples were kept at − 18°C until analysis. The resi-
dues were dissolved in 100 µL HPLC grade methanol and analyzed
by HPLC or TLC.
Non-catalyzed reaction. EV (50 mg, 0.1 mmol) and silver oxide
(47 mg, 0.2 mmol) were dissolved in a solution of 82.5 mg ABG in
5 mL of dichloromethane. The reaction was stirred for 20 h at
room temperature. Afterwards, the reactional medium was di-
luted with dichloromethane, filtrated, and the reaction mixture
was analyzed on TLC.
Suspension cells were harvested by vacuum filtration and
grounded to a fine powder by liquid nitrogen. The cells (15 g)
were extracted by MeOH (30 mL) under sonication (30 min). After
Munkert J et al. Production of the… Planta Med

Original Papers
Statistical analysis
filtration of the grounded cell material, the methanol was re-
moved by evaporation under reduced pressure at 50°C and the
resulting residue was purified by flash chromatography over a sili-
ca gel column (0 = > 100% acetone in dichloromethane; 0, 60, 70,
100%; 20 > 100% MeOH in acetone, 20%, 100%). Fractions eluted
with methanol:acetone (20:80) were shown to concentrate GEV,
according to TLC analysis. They were combined, the solvent re-
moved, and the obtained residue (55 mg) submitted to purifica-
tion over a Sephadex LH20 column. Isocratic elution with metha-
nol afforded 130 fractions, 1 mL each, which were analyzed by TLC
or HPLC after solvent removal.
All data are expressed as the mean standard deviation of the
mean. Means between the various groups were compared by
one-way analysis of variance (ANOVA followed by Tukeyʼs post
hoc test). In case of multiple comparisons, a post hoc Bonferroni
correction was applied. P values < 0.001 were considered statisti-
cally significant. Data were analyzed using GraphPad Prism 5 Soft-
ware (GraphPad).
Supporting information
UPLC/MS data and HPLC-DAD chromatograms obtained for com-
pounds 1–6, isolation and purification protocol for compound 1, a
plot of the HMQC‑NMR spectrum of compound 2 as well as the
dose-response curves of MTT assays for compounds 2–4 and the
conditions of the Na+/K+-ATPase assay along with the bar graphs
for compounds 1–4 are available as Supporting Information. De-
tailed descriptions of EV (2) synthesis and experimental condi-
tions of UPLC‑ESI‑MS, HPLC, TLC, and NMR analysis are also pro-
vided as Supporting Information.
Chromatography analysis
Detailed information for UPLC‑ESI‑MS, HPLC, and TLC analysis are
provided in Supporting Information.
NMR analysis
NMR spectra were recorded on a JEOL Alpha500 spectrometer
(500 MHz, 11.7 T) by using a tunable 5 mm probe head and an in-
verse field gradient 5 mm probe head, respectively. Detailed infor-
mation on NMR analysis is available in Supporting Information.
Acknowledgements
Cell lines
The renal cell carcinoma cell lines (Caki-1, 786-O) and the pros-
tate cancer cell lines (DU145, PC-3) were obtained from the Ger-
man Collection of Microorganisms and Cell Cultures (DSMZ,
Braunschweig, Germany). Cells were cultured in RPMI-1640 me-
dium or DMEM (Caki-1, 786-O, and PC-3; Sigma) supplemented
with 10% heat-inactivated fetal bovine serum, 1 nmol/L L-gluta-
mine, 100 U/mL penicillin, and 100 µg/mL streptomycin at 37°C
in a humidified atmosphere of 5% CO2.
We are very thankful for the financial support of this study by a
grant from FFL Stipendiums, FAU (Programm zur Förderung der
Chancengleichheit für Frauen in Forschung und Lehre, JM) and
FAU Unibund Anschubfinanzierung (J. M.), a grant from EU FP7
IRSES (grant 295251, W.K.), and a grant from CNPq (MCTI, Brazil,
grant 482244/2013-5, RP). F.C. B., R.J.A., M.C. O., and C. M. O.S.
thank CNPq for their research fellowships. Authors Contributions:
J. M., R.P., W. K., and F.B. conceived and designed the study. J.M.,
M. F., and R. P. performed all experiments with help from E. N.,
I.T. S., S.A., and R. A.W. B. performed the NMR studies. All au-
thors interpreted and discussed the data and contributed to the
preparation of the final manuscript.
MTT assays
The colorimetric MTT assay was performed using 3-(4,5-dimethyl-
1,3-thiazole-2-yl)-2,5 diphenyltetrazolium bromide (MTT). Briefly,
prostate (DU-145 and PC3) and renal (Caki-1, 786-O) cells were
seeded in 96-well culture plates (6 × 103 and 4 × 103 cells/well, re-
spectively) and, after 24 h, were treated with different concentra-
tions of the compounds. Then, 24, 48, and 72 h later, the medium
was replaced by 50 µL of MTT reagent (1 mg/mL) and the cells
were incubated for a further 4 h. DMSO was added in order to dis-
solve the formazan crystals, and the absorbance was measured at
570 nm using a plate spectrophotometer (VersaMax ELISA Micro-
plate Reader, Molecular Devices). Cells treated with 0.5% DMSO
served as a negative control to define 100% of viability. The per-
centage of viable cells was plotted against the compound concen-
tration and the IC50 values were determined by nonlinear regres-
sion and calculated using log (compound) compared with a nor-
malized response (variable slope) using GraphPad Prism 5 Soft-
ware (GraphPad).
Conflict of Interest
All authors declare to have no conflicts of interest.
References
[1] Ehle M, Patel C, Giugliano RP. Digoxin: clinical highlights: a review of
digoxin and its use in contemporary medicine. Crit Pathw Cardiol 2011;
10: 93–98
[2] Ashbrook AW, Lentscher AJ, Zamora PF, Silva LA, May NA, Bauer JA,
Morrison TE, Dermody TS. Antagonism of the sodium-potassium ATPase
impairs chikungunya virus infection. MBio 2016; 7: e00693-16
[3] Cheung YY, Chen KC, Chen H, Seng EK, Chu JJH. Antiviral activity of lana-
toside C against dengue virus infection. Antiviral Res 2014; 111: 93–99
[4] Wong RW, Balachandran A, Ostrowski MA, Cochrane A. Digoxin sup-
presses HIV‑1 replication by altering viral RNA processing. PLoS Pathog
2013; 9: e1003241
Na+/K+-ATPase assay
Enzymatic activities of the Na+/K+-ATPase α1, 2, 3 subunit of por-
cine cortex (Sigma) were assayed as described by [36,43]. De-
tailed information of the assay conditions is provided within the
Supporting Information.
[5] Bertol JW, Rigotto C, de Padua RM, Kreis W, Barardi CRM, Braga FC,
Simoes CMO. Antiherpes activity of glucoevatromonoside, a cardenolide
isolated from a Brazilian cultivar of Digitalis lanata. Antiviral Res 2011;
92: 73–80
Munkert J et al. Production of the… Planta Med

[6] Su CT, Hsu JT, Hsieh HP. Anti-HSV activity of digitoxin and its possible
[26] Kreis W, Reinhard E. 12 beta-Hydroxylation of digitoxin by suspension-
cultured Digitalis lanata cells: production of digoxin in 20-litre and 300-
litre air-lift bioreactors. J Biotechnol 1992; 26: 257–273
mechanisms. Antiviral Res 2008; 79: 62–70
[7] Diederich M, Muller F, Cerella C. Cardiac glycosides: From molecular tar-
gets to immunogenic cell death. Biochem Pharmacol 2017; 125: 1–11
[27] Theurer C, Treumann HJ, Faust T, May U, Kreis W. Glycosylation in carde-
nolide biosynthesis. PCTOC 1994; 38: 327–335
[8] Calderon-Montano JM, Burgos-Moron E, Orta ML, Maldonado-Navas D,
Garcia-Dominguez I, Lopez-Lazaro M. Evaluating the cancer therapeutic
potential of cardiac glycosides. Biomed Res Int 2014; 2014: 794930
[28] Kreis W, Reinhard E. 12beta-Hydroxylation of digitoxin by suspension-
cultured Digitalis lanata cells. Production of deacetyllanatoside C using
a two-stage culture method. Planta Med 1988; 54: 143–148
[9] Cerella C, Dicato M, Diederich M. Assembling the puzzle of anti-cancer
mechanisms triggered by cardiac glycosides. Mitochondrion 2013; 13:
225–234
[29] Kreis W, May U, Reinhard E. UDP-glucose: digitoxin 16′-O-glucosyltrans-
ferase from suspension-cultured Digitalis lanata cells. Plant Cell Rep
1986; 5: 442–445
[10] Mijatovic T, Dufrasne F, Kiss R. Cardiotonic steroids-mediated targeting
of the Na(+)/K(+)-ATPase to combat chemoresistant cancers. Curr Med
Chem 2012; 19: 627–646
[30] Theurer C, Kreis W, Reinhard E. Effects of digitoxigenin, digoxigenin, and
various cardiac glycosides on cardenolide accumulation in shoot cultures
of Digitalis lanata. Planta Med 1998; 64: 705–710
[11] Elbaz HA, Stueckle TA, Tse W, Rojanasakul Y, Dinu CZ. Digitoxin and its
analogs as novel cancer therapeutics. Exp Hematol Oncol 2012; 1: 4
[31] Munkert J, Ernst M, Muller-Uri F, Kreis W. Identification and stress-in-
duced expression of three 3beta-hydroxysteroid dehydrogenases from
Erysimum crepidifolium Rchb. and their putative role in cardenolide bio-
synthesis. Phytochemistry 2014; 100: 26–33
[12] Schneider NFZ, Geller FC, Persich L, Marostica LL, Padua RM, Kreis W,
Braga FC, Simoes CMO. Inhibition of cell proliferation, invasion and migra-
tion by the cardenolides digitoxigenin monodigitoxoside and convalla-
toxin in human lung cancer cell line. Nat Prod Res 2016; 30: 1327–1331
[32] Singh S, Vishwakarma RK, Kumar RJS, Sonawane PD, Ruby, Khan BM.
Functional characterization of a flavonoid glycosyltransferase gene from
Withania somnifera (Ashwagandha). Appl Biochem Biotechnol 2013;
170: 729–741
[13] Wang HY, Xin W, Zhou M, Stueckle TA, Rojanasakul Y, OʼDoherty GA.
Stereochemical survey of digitoxin monosaccharides: new anticancer
analogues with enhanced apoptotic activity and growth inhibitory effect
on human non-small cell lung cancer cell. ACS Med Chem Lett 2011; 2:
73–78
[33] Sharma LK, Madina BR, Chaturvedi P, Sangwan RS, Tuli R. Molecular clon-
ing and characterization of one member of 3beta-hydroxy sterol gluco-
syltransferase gene family in Withania somnifera. Arch Biochem Biophys
2007; 460: 48–55
[14] Schneider NFZ. Avaliação da ação citotóxica de cardenolídeos em células
tumorais [thesis]. Florianópolis: Universidade Federal de Santa Catarina;
2015
[34] Lim EK, Ashford DA, Hou B, Jackson RG, Bowles DJ. Arabidopsis glycosyl-
transferases as biocatalysts in fermentation for regioselective synthesis
of diverse quercetin glucosides. Biotechnol Bioeng 2004; 87: 623–631
[15] Braga FC, Kreis W, Braga de Oliveira A. Isolation of cardenolides from a
Brazilian cultivar of Digitalis lanata by rotation locular counter-current
chromatography. J Chromatogr A 1996; 756: 287–291
[35] Nolte E, Sobel A, Wach S, Hertlein H, Ebert N, Muller-Uri F, Slany R, Taubert
H, Wullich B, Kreis W. The new semisynthetic cardenolide analog 3beta-2-
(1-Amantadine)-1-on-ethylamine-digitoxigenin (AMANTADIG) efficient-
ly suppresses cell growth in human leukemia and urological tumor cell
lines. Anticancer Res 2015; 35: 5271–5275
[16] Luckner M, Wichtel M. Digitalis. Stuttgart: Wissenschaftliche Verlagsge-
sellschaft mbH; 2003
[17] Kreis W, Haug B, Yücesan B. Somaclonal variation of cardenolide content
in Heywoodʼs foxglove, a source for the antiviral cardenolide glucoeva-
tromonoside, regenerated from permanent shoot culture and callus. In
Vitro Cell Dev Biol Plant 2014; 51: 35–41
[36] Nolte E, Wach S, Thais Silva I, Lukat S, Ekici AB, Munkert J, Muller-Uri F,
Kreis W, Simoes CMO, Vera J, Wullich B, Taubert H, Lai X. A new semi-
synthetic cardenolide analog 3beta-2-(1-amantadine)-1-on-ethylamine-
digitoxigenin (AMANTADIG) affects G2/M cell cycle arrest and miRNA
expression profiles and enhances proapoptotic survivin-2B expression
in renal cell carcinoma cell lines. Oncotarget 2017; 8: 11676–11691
[18] Beale TM, Taylor MS. Synthesis of cardiac glycoside analogs by catalyst-
controlled, regioselective glycosylation of digitoxin. Org Lett 2013; 15:
1358–1361
[19] Zhou M, OʼDoherty GA. A stereoselective synthesis of digitoxin and digi-
toxigen mono- and bisdigitoxoside from digitoxigenin via a palladium-
catalyzed glycosylation. Org Lett 2006; 8: 4339–4342
[37] Katz A, Lifshitz Y, Bab-Dinitz E, Kapri-Pardes E, Goldshleger R, Tal DM,
Karlish SJD. Selectivity of digitalis glycosides for isoforms of human Na,
K-ATPase. J Biol Chem 2010; 285: 19582–19592
[20] Daisuke Satho. United States Patent US 3,856,944. Pharmaceutical
Compositions; 1974
[38] Laursen M, Yatime L, Nissen P, Fedosova NU. Crystal structure of the
high-affinity Na+K+-ATPase-ouabain complex with Mg2+ bound in the
cation binding site. Proc Natl Acad Sci U S A 2013; 110: 10958–10963
[21] Gantt RW, Peltier-Pain P, Cournoyer WJ, Thorson JS. Using simple donors
to drive the equilibria of glycosyltransferase-catalyzed reactions. Nat
Chem Biol 2011; 7: 685–691
[39] Yatime L, Laursen M, Morth JP, Esmann M, Nissen P, Fedosova NU. Struc-
tural insights into the high affinity binding of cardiotonic steroids to the
Na+,K+-ATPase. J Struct Biol 2011; 174: 296–306
[22] Gantt RW, Peltier-Pain P, Thorson JS. Enzymatic methods for glyco(diver-
sification/randomization) of drugs and small molecules. Nat Prod Rep
2011; 28: 1811–1853
[40] Toyoshima C, Cornelius F. New crystal structures of PII-type ATPases: ex-
citement continues. Curr Opin Struct Biol 2013; 23: 507–514
[23] Shilpa K, Varun K, Lakshmi BS. An alternate method of natural drug pro-
duction: Elciting secondary metabolite production using plant cell cul-
ture. J Plant Sci 2010; 5: 222–247
[41] Yoda A. Structure-activity relationships of cardiotonic steroids for the in-
hibition of sodium- and potassium-dependent adenosine triphospha-
tase. I. Dissociation rate constants of various enzyme-cardiac glycoside
complexes formed in the presence of magnesium and phosphate. Mol
Pharmacol 1973; 9: 51–60
[24] Padua RM, Meitinger N, Dias de Souza JF, Waibel R, Gmeiner P, Braga FC,
Kreis W. Biotransformation of 21-O-acetyl-deoxycorticosterone by cell
suspension cultures of Digitalis lanata (strain W.1.4). Steroids 2012; 77:
1373–1380
[42] Cornelius F, Kanai R, Toyoshima C. A structural view on the functional im-
portance of the sugar moiety and steroid hydroxyls of cardiotonic ste-
roids in binding to Na,K-ATPase. J Biol Chem 2013; 288: 6602–6616
[25] Kreis W, Fulzele D, Hoelz H, Val J, Reinhard E. Production of Cardenolides
by Digitalis Cell Cultures – Models and Process Options. In: Oono K,
Hirabayashi T, Kikuchi S, Handa H, Kajiwara K, eds. Plant Tissue Culture
and Gene Manipulation for Breeding and Formation of Phytochemicals.
Tsukuba: National Institute of Agrobiological Resources (NIAR); 1992:
335–354
[43] Baykov AA, Evtushenko OA, Avaeva SM. A malachite green procedure for
orthophosphate determination and its use in alkaline phosphatase-
based enzyme immunoassay. Anal Biochem 1988; 171: 266–270
Munkert J et al. Production of the… Planta Med