Contribution of flavonoids and catechol to the reduction of ICAM-1 expression in endothelial cells by a standardised Willow bark extract

Article abstract of DOI:10.1016/j.phymed.2011.08.065

Introduction: A quantified aqueous Willow bark extract (STW 33-I) was tested concerning its inhibitory activity on TNF-α induced ICAM-1 expression in human microvascular endothelial cells (HMEC-1) and further fractionated to isolate the active compounds.

Full text of DOI:10.1016/j.phymed.2011.08.065

Phytomedicine 19 (2012) 245–252
Contents lists available at SciVerse ScienceDirect
Phytomedicine
journal homepage: www.elsevier.de/phymed
Contribution of flavonoids and catechol to the reduction of ICAM-1 expression in
endothelial cells by a standardised Willow bark extract
A. Freischmidta, G. Jürgenliemka, B. Krausa, S.N. Okpanyib, J. Müllerb, O. Kelberb,
D. Weiserb, J. Heilmanna,∗
a Lehrstuhl für Pharmazeutische Biologie, Universität Regensburg, 93040 Regensburg, Germany
b Steigerwald Arzneimittelwerk GmbH, Havelstr. 5, 64295 Darmstadt, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Introduction: A quantified aqueous Willow bark extract (STW 33-I) was tested concerning its inhibitory
activity on TNF-␣ induced ICAM-1 expression in human microvascular endothelial cells (HMEC-1) and
further fractionated to isolate the active compounds.
Cordially dedicated to Prof. Dr. Otto Sticher
(Zurich) on the occasion of his 75th
birthday.
Results: At 50 ␮g/ml the extract, which had been prepared from Salix purpurea L., decreased ICAM-1
expression to 40% compared to control cells without showing cytotoxic effects. Further liquid–liquid
partition revealed an ethyl acetate phase with potent reduction of ICAM-1 expression to 40% at 8 ␮g/ml.
This fraction was comprehensively characterised by the isolation of flavanone aglyca and their corre-
sponding glycosides, chalcone glycosides, salicin derivatives, cyclohexane-1,2-diol glycosides, catechol
and trans-p-coumaric acid. All compounds were investigated for their activity on TNF-␣ induced ICAM-1
expression. The flavonoid and chalcone glycosides were not active up to 50 ␮M, whereas catechol and
eriodictyol at the same concentration showed a significant reduction of ICAM-1 expression to 50% of
control. Interestingly, other isolated flavanone aglyca like taxifolin, dihydrokaempferol and naringenin
showed only weak or moderate inhibitory activity. Eriodictyol was a minor compound in the extract,
whereas the catechol content in the extract (without excipients) reached 2.3%, determined by HPLC. One
of the isolated cyclohexan-1,2-diol glucosides, 6ꢀ-O-4-hydroxybenzoyl-grandidentin, is a new natural
compound.
Keywords:
Salix purpurea
Salicaceae
Willow bark extract
ICAM-1
Catechol
Flavonoids
Conclusion: As catechol is quantitatively important in Willow bark extracts it can be concluded from the
in vitro data that not only flavonoids and salicin derivatives, but also catechol can probably contribute to
the anti-inflammatory activity of Willow bark extracts.
© 2011 Elsevier GmbH. All rights reserved.
Introduction
fractions containing polyphenols also showed inhibitory activity
in vitro on several molecular targets like transcription factors, pro-
Willow bark extracts (Salicis cortex, Salix spec., Salicaceae)
are used since ancient times for the treatment of fever, pain
and inflammation (Hedner and Everts 1998). Salicylic acid, the
in vivo metabolite of salicylic alcohol derivatives, was postu-
lated to be the main active principle of these extracts. Due to
the low achievable salicylic acid plasma concentration, which is
equivalent to only ∼100 mg of acetylsalicylic acid after oral appli-
cation of 240 mg Willow bark extract (Schmid et al. 2001), recent
work concentrated on the additional contribution of polyphenolic
compounds on the anti-inflammatory and analgesic effect. Stan-
dardised extracts rich in polyphenols showed significant clinical
effects (for review Vlachojannis et al. 2009) and extracts as well as
inflammatory cytokines (Bonaterra et al. 2010), cyclooxygenases
and radical production (Khayyal et al. 2005). Consequently, the
overall effect of Willow bark extracts can be most likely explained
by different effects of various phenolic compounds on distinct tar-
gets (Nahrstedt et al. 2007). Wagner and Ulrich-Merzenich (2009)
drafted a multi-component/multi-target concept with regard to
the importance of multi-target synergy effects for combinations
of plant constituents and demonstrated its importance not only
for Willow bark extracts, but also for several other phytophar-
maceuticals. As further compounds and targets may contribute to
the anti-inflammatory effect of Willow bark, a quantified aqueous
extract (STW 33-I; Steigerwald) was investigated in a microvas-
cular endothelial cell line (HMEC-1) on its ability to reduce the
expression of the adhesion molecule ICAM-1. Adhesion molecules
like ICAM-1, VCAM and the selectins are known to be involved in
inflammatory processes at endothelial cells (Gearing and Newman
∗
Corresponding author. Tel.: +49 09419434759; fax: +49 09419434990.
E-mail address: Joerg.Heilmann@chemie.uni-regensburg.de (J. Heilmann).
0944-7113/$ – see front matter © 2011 Elsevier GmbH. All rights reserved.
doi:10.1016/j.phymed.2011.08.065

246
A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
Fig. 1. Isolated compounds 1–20.
1993). Endothelial cells communicate substantially via adhesion
molecules with their counter-receptors on leukocytes to gener-
ate intracellular signals followed by inflammation processes under
active participation of the endothelium (Cook-Mills and Deem
2005). Furthermore, several in vivo and in vitro models revealed
the particular importance of endothelial cells and their expression
of adhesion molecules in atherosclerosis and the inflammatory pro-
cesses initiated therewith (Galkina and Ley 2007). For example,
VCAM-1 and ICAM-1 dependent recruitment of immune cells in
regions of plaque neovascularization is likely also an important
step during the development of atherosclerosis. Therefore, these
molecules are rational targets for in vitro testing of compounds
showing anti-inflammatory effects in vivo. As the ethyl acetate
fraction of STW 33-I was most active all phenolic compounds
(Fig. 1) significantly present in this fraction were isolated and tested
as single compounds. To exclude false positive results caused by
cytotoxicity, a MTT assay under the same assay conditions was
always done in parallel. As the phytochemical characterisation
of the polyphenolic pattern of the Salix species S. purpurea L., S.
daphnoides Vill. and S. fragilis L. is rudimentary (all particularly
mentioned in the Pharmacopeia Europeia (PhEur) for Willow bark)
and the extract studied had been prepared from Salix purpurea L.,
the present study is also a first step towards a systematic phyto-
chemical characterisation of pharmaceutically used Willow bark
species and resulting extracts.
Materials and methods
Extract
Dry extract from Willow bark (PhEur 6.1, Salix purpurea L., Sal-
icaceae; STW 33-I Steigerwald) obtained from water extraction;

A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
247
drug-to-extract ratio (DEV nativ) 16–23:1, overall salicin content
23–26% (m/m), excipients: silica, magnesium stearate.
(ESI-MS: 3 kV, capillary: 250 ◦C). HRESI-MS was determined on
a UHD accurate mass Q-TOF (Agilent Technologies). UV spectra
were measured on a Cary 50 Scan UV–Visible spectrophotometer
(Varian) in MeOH (Uvasol®). Optical rotation was recorded on a
Unipol L1000 (Schmidt + Haensch) in MeOH (Uvasol®), cell length:
5 cm. CD spectra were measured on a J-815 spectropolarimeter
(Jasco), c = 1.5 mg/mL in MeOH at room temperature, quartz
cuvette of 0.1 cm path length. HPLC (semi-preparative): Varian
ProStar, autosampler 410, 2 pumps 210, detector 335, columns:
Varian Pursuit XRs, C18, 5 ␮m, 250 mm × 10 mm, Varian Polaris
C8-A, 5 ␮m, 250 mm × 10 mm, Knauer Eurosphere-100, C18, 7 ␮m,
16 mm × 25 mm. HPLC (analytical): Elite LaChrom VWR Hitachi,
autosampler L-2200, pump L-2130, column oven L-2350, DAD
L-2455, column: Purospher®STAR RP-18e, 250 mm × 4.6 mm.
Flash chromatography system: Armen Spot Liquid Chromatogra-
phy Flash, columns: SPV D40 RP 18, 25–40 ␮m, 90 g (Merck) and
SVF D26 Si60, 15–40 ␮m, 30 g (Merck). Absolute configurations
of the sugars were measured on a Bio-Rad Biofocus 3000 CE
system (70/75 cm, 50 ␮m quartz capillary) according to Noe and
Freissmuth 1995 (modified). Shortly, the glucoside was hydrolyzed
(4 M aqueous TFA, 120 ◦C, 60 min) and afterwards evaporated to
remove the acid until dryness. The hydrolysis product and d-
and l-glucose reference substances (Sigma-Aldrich, Roth) were
derivatised with 0.1 M S-(−)-1-phenylethylamine respectively
and reduced with 0.46 M aqueous sodium cyanoborohydride
solution. The resulting diastereomers were subjected to capillary
electrophoresis (buffer: 50 mM Na2B4O7 in 23% aqueous MeCN,
pH 10.3, injection: 3 psi s, voltage: 30 kV, detection: 200 nm,
temperature: 27 ◦C). Comparison of migration times confirmed the
presence of d-glucose.
Isolation
10 g extract was separated from the excipients by vacuum filtra-
tion and suspended in water (250 ml). The resulting suspension was
partitioned against n-hexane (2 × 100 ml) to extract the lipophilic
compounds. The hexane phase was dried under vacuum to yield
only 50 mg residue which was excluded from further tests due
to the low amount. The water phase was subsequently parti-
tioned against ethyl acetate (EA, 3 × 100 ml) and n-butanol (BUT,
3 × 100 ml) to yield 1.3 and 2.5 g extract after drying over Na2SO4
and evaporation of solvent. The volume of the water phase was
reduced to 100 ml on a rotary evaporator and 500 ml EtOH was
added to yield the EtOH-soluble part and a precipitate only solu-
ble in pure water (H2O-phase) after stirring. After centrifugation
and drying 3.0 g EtOH and 0.5 g H2O phase were obtained (par-
tition process modified according to Nahrstedt et al. 2007). The
procedure was repeated five times. An aliquot of the ethyl acetate
phase (6.3 g) was divided in two portions and fractionated by CC
on Sephadex LH-20 (length 67 cm, ∅ 4.0 cm) eluting with 75%
EtOH (1 ml/min) to yield fractions EA1–EA6. Fractions were com-
bined according to similarity in TLC using Naturstoff A (5% (m/v)
in MeOH) and anisaldehyde reagent (0.5% anisaldehyde, 84.5%
MeOH, 10% glacial acetic acid, 5% sulphuric acid (98%), each v/v) as
spray reagent for detection and resulted in 1.74 g EA1 (∼0–800 ml),
0.15 g EA2 (∼800–850 ml), 2.29 g EA3 (∼850–1000 ml), 0.70 g EA4
(∼1000–1100 ml), 0.47 g EA5 (∼1100–1300 ml) and 0.50 g EA6
(∼1300–1800 ml). Further isolation protocol of compounds 1–20
is described in Chart 1.
The identity of all compounds was confirmed by compari-
son of spectral data (1H NMR, 13C NMR, ESI-MS, melting points,
UV-spectrum, and optical rotation) with literature data of isogran-
didentatin A (1, Si et al. 2009), trichocarposide (3, Fernandes et al.
2009), populoside B (4, Zhang et al. 2006a), dihydrokaempferol-7-
O-ˇ-d-glucoside (5, Foo and Karchesy 1989), (2R,2S)-naringenin-7-
O-ˇ-d-glucoside (6, Zapesochnaya et al. 2002), catechol (7, Smith
and Proulx 1976), (2R)-naringenin-5-O-ˇ-d-glucoside (8, Ibrahim
et al. 2007, Tyukavkina et al. 1989, Zapesochnaya et al. 2002), (2S)-
naringenin-5-O-ˇ-d-glucoside (9, Ibrahim et al. 2007; Tyukavkina
et al. 1989; Zapesochnaya et al. 2002), (2R,2S)-eriodictyol-7-
O-ˇ-d-glucoside (10, Pan et al. 2008), trans-p-coumaric acid
(11, Kort et al. 1996), isosalipurposide (12, Zapesochnaya et al.
2002), 6ꢀꢀ-trans-p-coumaroyl-(2R)-naringenin-5-O-ˇ-d-glucoside
(13, Vinokurov and Skrigan 1969), 6ꢀꢀ-trans-p-coumaroyl-(2S)-
naringenin-5-O-ˇ-d-glucoside (14, Vinokurov and Skrigan 1969),
taxifolin (16, Markham and Ternai 1976; Han et al. 2007), dihy-
drokaempferol (17, Foo and Karchesy 1989; Han et al. 2007),
eriodictyol (18, Pan et al. 2008; Zhang et al. 2006b), naringenin
(15, Markham and Ternai 1976; Zhang et al. 2006b), 6ꢀꢀ-trans-
p-coumaroylisosalipurposide (19, Zapesochnaya et al. 2002) and
phelligrin A (20, Mo et al. 2003). Furthermore, 2D NMR mea-
surements (1H,1H COSY, 1H,13C HSQC, 1H,13C HMBC and 1H,1H
ROESY/1H,1H NOESY) for confirmation of the literature signal
assignment were done. In case of R or S assignment also CD spectra
were measured. For determination of the d- or l-configuration of a
sugar a method based on capillary electrophoresis (CE) was applied
(Noe and Freissmuth 1995).
6ꢀ-O-4-Hydroxybenzoylgrandidentin (2)
22
White amorphous powder; UV (MeOH) ꢀmax 257 nm;
˛
=
[ ]D
+41◦ (c = 0.0005); 1H NMR (600 MHz, 298 K, CD3OD): 1.11–1.19
(1H, m, H-5b), 1.22–1.30 (1H, m, H-4b), 1.43–1.51 (2H, m, H-
3b, H-6b), 1.52–1.61 (2H, m, H-4a, H-5a), 1.71–1.78 (1H, m,
H-3a), 1.85–1.92 (1H, m, H-6a), 3.24–3.28 (1H, m, H-2ꢀ), 3.33–3.38
(1H, m, H-4ꢀ), 3.57 (1H, ddd, J = 2.1, 7.0, 9.2, H-5ꢀ), 3.68–3.72
(1H, m, H-1), 3.73–3.78 (1H, m, H-2), 3.42–3.76 (1H, m, H-
3ꢀ), 4.40 (1H, dd, J = 6.9, 11.8, H-6ꢀb), 4.41 (1H, d, J = 7.9, H-1ꢀ),
4.57 (1H, dd, J = 2.1, 11.8, H-6ꢀa), 6.81 (2H, d, J = 8.6, H-3ꢀꢀ, H-
5ꢀꢀ), 7.88 (2H, d, J = 8.6, H-2ꢀꢀ, H-6ꢀꢀ);13C NMR (150 MHz, 298 K,
CD3OD): 22.6 (C-4), 23.1 (C-5), 30.0 (C-6), 31.3 (C-3), 64.9 (C-6ꢀ),
70.3 (C-2), 72.1 (C-4ꢀ), 75.4 (C-2ꢀ, C-5ꢀ), 77.8 (C-3ꢀ), 81.6 (C-1),
104.4 (C-1ꢀ), 116.1 (C-3ꢀꢀ, C-5ꢀꢀ), 122.2 (C-1ꢀꢀ), 132.9 (C-2ꢀꢀ, C-
6ꢀꢀ), 163.6 (C-4ꢀꢀ), 168.0 (C-7ꢀꢀ). LRESI-MS (pos.) m/z 399 [M+H+];
HRESI-MS (pos.) m/z 397.1497 [M−H+], calculated for C19H25O9
397.1504.
6ꢀꢀ-trans-p-Coumaroyl-(2S)-naringenin-5-O-ˇ-d-glucoside (14)
White amorphous powder (MeOH) ꢀmax: 228, 287, 311 nm;
20
˛
= −69◦ (c = 0.001). 1H NMR (600 MHz, 298 K, CD3OD): 2.68
[ ]D
(1H, dd, J = 3.0, 17.3, H-3b), 2.98 (1H, dd, J = 13.1, 17.3, H-3a),
3.48 (1H, dd, J = 8.7, 9.3, H-4ꢀꢀ), 3.51 (1H, dd, J = 8.6, 8.6, H-3ꢀꢀ),
3.59 (1H, dd, J = 7.8, 8.6, H-2ꢀꢀ), 3.74 (1H, dd, J = 2.2, 6.7, H-5ꢀꢀ),
4.32 (1H, dd, J = 6.6, 11.9, H-6ꢀꢀb), 4.59 (1H, dd, J = 2.3, 11.9, H-
6ꢀꢀa), 4.83 (1H, d, J = 7.7, H-1ꢀꢀ), 5.29 (1H, dd, J = 2.9, 13.2, H-2),
6.11 (1H, d, J = 2.2, H-8), 6.37 (1H, d, J = 16.1, H-8ꢀꢀꢀ), 6.44 (1H, d,
J = 2.2, H-6), 6.79 (2H, d, J = 8.6, H-3ꢀ, H-5ꢀ), 6.80 (2H, d, J = 8.6,
H-3ꢀꢀꢀ, H-5ꢀꢀꢀ), 7.28 (2H, d, J = 8.6, H-2ꢀ, H-6ꢀ), 7.46 (2H, d, J = 8.6, H-
2ꢀꢀꢀ, H-6ꢀꢀꢀ), 7.63 (1H, d, J = 16.1, H-7ꢀꢀꢀ); 13C NMR (150 MHz, 298 K,
CD3OD): 46.3 (C-3), 64.5 (C-6ꢀꢀ), 71.7 (C-4ꢀꢀ), 74.7 (C-2ꢀꢀ), 75.9 (C-
5ꢀꢀ), 77.3 (C-3ꢀꢀ), 80.2 (C-2), 99.4 (C-8), 100.5 (C-6), 104.3 (C-1ꢀꢀ),
107.0 (C-10), 114.9 (C-8ꢀꢀꢀ), 116.3 (C-5ꢀ, C-3ꢀ), 116.8 (C-3ꢀꢀꢀ, C-5ꢀꢀꢀ),
Instrumentation
NMR spectra were measured on a Bruker Avance 600 (1H
600 MHz, 13C 150 MHz at 298.0 K) and referenced against
undeuterated solvents. LRMS spectra were recorded on a Finnigan
MAT SSQ 710A (EI-MS: 70 eV), ThermoQuest Finnigan TSQ 7000

248
A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
Chart 1. Detailed isolation protocol of compounds 1–20.
127.2 (C-1ꢀꢀꢀ), 129.0 (C-2ꢀ, C-6ꢀ), 131.0 (C-1ꢀ), 131.3 (C-2ꢀꢀꢀ, C-6ꢀꢀꢀ),
146.9 (C-7ꢀꢀꢀ), 159.0 (C-4ꢀ), 161.3 (C-4ꢀꢀꢀ), 162.1 (C-5), 166.5 (C-9),
167.3 (C-7), 169.1 (C-9ꢀꢀꢀ), 193.0 (C-4). LRESI-MS (pos.) m/z 581
[M+H+].
Determination of catechol
The content of catechol (7) was determined by HPLC. STW
33-I (recipients removed) was dissolved in HPLC grade water and

A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
249
sonicated for 15 min. 75 ␮l were injected and analyzed by analyt-
ical HPLC, n = 3. Conditions: A: H2O + 0.1% formic acid, B: MeCN:
H2O 95/5; 0–10 min 90/10 → 87/13, 20–30 min 87/13 isocratic;
0.75 ml/min flow, oven temperature 30 ◦C. A calibration curve of
catechol was established (12 point, n = 3, range 0.05–125 ␮g/ml)
by height of the catechol peaks (y = 26,591x + 9823.2, R2 = 0.9982).
The content of catechol in the extract was determined
to be 2.3 0.1%.
Cell culture
Confluent grown human microvascular endothelial cells
(HMEC-1, Ades et al. 1992) were pretreated either with test com-
pounds, parthenolide (Calbiochem, purity ≥97%, 5 or 10 ␮M, pos-
itive control) or medium (ECGM, endothelial cell growth medium
(Provitro) + 10% FCS + antibiotics + supplements (both Provitro)) as
negative control in 24-well plates. 30 min later, 10 ng/ml TNF-␣
(Sigma–Aldrich) were added to stimulate the ICAM-1 expres-
sion. After 24 h of incubation (New Brunswick Scientific, 37 ◦C,
5% CO2), cells were washed with PBS, removed from the plate
with trypsin/EDTA and fixed with formalin. After incubating with
FITC-labelled mouse antibody against ICAM-1 (Biozol) for 20 min,
the fluorescence intensity was measured by FACS analysis (Becton
Dickinson FacscaliburTM). ICAM-1 expression of cells treated with
TNF-␣ only was set as 100%.
The MTT viability assay was performed according to Mosman
1983 (modified). Briefly, confluent grown HMEC-1 cells were incu-
bated with fractions and test compounds (concentration according
to the ICAM-1 assay) in 96-well plates in medium (pure medium
for negative control; n = 6 for each value). After 24 h the frac-
tions/substances and the supernatants were removed and 10 ␮l
MTT (5 mg/ml) in 90 ␮l medium was added for 3 h. This solution
was exchanged for 10% sodium dodecyl sulphate and 24 h later the
absorbance was measured with a SpectraFluor plus plate reader
(Tecan, Crailsheim, Germany) at 560 nm.
Fig. 2. (A) Inhibition of TNF-␣ induced ICAM-1 expression by extract STW 33-I (10
and 50 ␮g/ml) and ethyl acetate (EA), n-butanol (BUT), ethanol (EtOH) and H2O frac-
tions obtained via liquid–liquid partition. (B) Inhibition of TNF-␣ induced ICAM-1
expression by fractions EA1–EA6 obtained by separation of the ethyl acetate frac-
tion on Sephadex LH-20. Cells were pretreated with the obtained fractions (10 ␮M
parthenolide (P) as positive control) for 30 min followed by treatment with 10 ng/ml
TNF-␣ for 24 h; HMEC #4–11; n = 3 in duplicates; mean SD; *p < 0.05 vs. neg. con-
trol with TNF-␣, ICAM expression of cells without any treatment ≈2.5%.
Statistical analysis
Each experiment was performed 3 times in duplicates for ICAM-
1 assay and 3 times in sextuplicates for viability assay. Results are
expressed as mean SD. Statistical significance between control
and test compounds was determined by one-way ANOVA (p < 0.05)
followed by Dunetts Post hoc test.
activity in fraction EA4 (Fig. 2B). All fractions were worked up to
yield the flavanone glycosides (2R)-naringenin-5-O-ˇ-d-glucoside
(8), (2S)-naringenin-5-O-ˇ-d-glucoside (9), (2R,2S)-naringenin-
7-O-ˇ-d-glucoside (6), 6ꢀꢀ-trans-p-coumaroyl-(2R)-naringenin-5-
O-ˇ-d-glucoside (13), 6ꢀꢀ-trans-p-coumaroyl-(2S)-naringenin-5-O-
ˇ-d-glucoside (14), (2R,2S)-eriodictyol-7-O-ˇ-d-glucoside (10),
the flavanonol glycoside dihydrokaempferol-7-O-ˇ-d-glucoside
(5), the chalcone glycosides isosalipurposide (12), 6ꢀꢀ-trans-
p-coumaroylisosalipurposide (19), the flavanone or flavanonol
aglycones naringenin (15), taxifolin (16), dihydrokaempferol (17),
eriodictyol (18) and phelligrin A (20), the salicyl alcohol deriva-
tives trichocarposide (3) and populoside B (4) as well as catechol (7)
and trans-p-coumaric acid (11, Fig. 1) by subsequent open column
chromatography, flash chromatography and HPLC. Additionally,
compounds 1 and 2, representing rare coumaroyl- and benzoyl-
glycosides with cyclohexanol substitution were isolated of which
compound 2 is a new one. Isolated compounds were tested at 20
and 50 ␮M for their ability to reduce ICAM-1 expression in HMEC-
1. While none of the glucosides were active (data not shown),
probably due to their high hydrophilicity, only the aglyca cate-
chol (7) and the flavanone eriodictyol (18, Figs. 3 and 4) exhibited
significant activity. Catechol was further tested in different concen-
trations showing a dose dependent decrease of ICAM-1 expression
(Fig. 4) without any signs of cytotoxicity in the MTT assay (Fig. 5B).
Interestingly, neither the isolated flavanonol aglyca (introduction
of an additional hydroxyl group at C-3) nor naringenin (15) showed
comparable activity to 18. Whereas 18 is only a minor compound,
the amount of 7 in the extract is noteworthy and reached 2.3%.
Results
An aqueous extract of Willow bark was separated from its
excipients (magnesium stearate and silicium dioxide) and in vitro
tested at 10 and 50 ␮g/ml in HMEC-1 for reduction of TNF-␣
induced ICAM-1 expression. The extract showed a potent and
dose-dependent reduction of ICAM-1 expression to 70% and 40%
of control (Fig. 2A) without exhibiting cytotoxic effects (Fig. 5A)
Following a liquid–liquid separation strategy (“Material and meth-
ods”) of the extract four fractions were obtained for further
pharmacological testing. The ethyl acetate (EA), n-butanol (BUT),
ethanol (EtOH) and H2O soluble phase were tested in different con-
centrations resembling their yield from the extract. The EA phase
showed the strongest activity and decreased the ICAM-1 expression
to ∼40% of control at 8 ␮g/ml and was chosen for further frac-
tionation. The BUT (15 ␮g/ml) and EtOH (19 ␮g/ml) phase showed
moderate effects with a decrease of ICAM-1 expression to 75 and
82% of control, respectively, whereas the H2O-soluble phase was
not active at 3 ␮g/ml (Fig. 2A). None of the tested phases showed
cytotoxic effects at the used concentration (Fig. 5A).
The EA phase was worked up with column chromatography
on Sephadex LH-20 to obtain 6 fractions showing a significant

250
A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
Fig. 3. Inhibition of TNF-␣ induced ICAM-1 expression by compounds 15–18 and
20 isolated from the ethyl acetate fraction. Cells were pretreated with 20 and
50 ␮M of the compound (5 ␮M parthenolide (P) as positive control) for 30 min fol-
lowed by treatment with 10 ng/ml TNF-␣ for 24 h; HMEC #3–16; n = 3 in duplicates;
mean SD; *p < 0.05 vs. neg. control with TNF-␣, ICAM expression of cells with-
out any treatment ≈2.5%, EtOH (E) and DMSO (D) were used as solvents for the
compounds (0.1% in medium).
Despite of several phytochemical investigations regarding the gen-
era Salix and Populus (both Salicaceae) also a hitherto unknown
esterified cyclohexandiol-ˇ-O-glucoside (2) was isolated. The 1H
NMR spectrum substantiated the presence of a ˇ-configurated glu-
cose residue (Domisse et al. 1986; Si et al. 2009). Furthermore,
the downfield region showed two doublets at ı = 6.81 and 7.88
(J = 8.6, 2H, each) indicating together with a quaternary carbon sig-
nal at ı = 168.0, which was coupled to the phenyl protons in HMBC,
the presence of a 4-hydroxybenzoyl moiety. The resonance sig-
nals of the protons at C-6 of glucose (ı = 4.40, 4.57) were shifted
significantly downfield in comparison with unsubstituted glucose
pointing to C-6ꢀ as position for esterification. This was confirmed by
a HMBC experiment showing 3J correlations between the ester car-
bonyl and H-6ꢀ. The 1,2-dihydroxycyclohexane unit was deduced
from 4 pairs of broad geminal proton signals resonating between
ı = 1.11–1.92 as well as from two additional downfield shifted and
coupling proton signals at ı = 3.68–3.78. This unit is linked at posi-
tion 1 to the glucose at C-1ꢀ what was substantiated in the NOESY
spectrum as well as by the corresponding 3J C,H correlations in the
HMBC. Extensive 13C and 2D NMR measurements corroborated the
presence of a 1,2-dihydroxycyclohexyl-6ꢀ-O-4-hydroxy-benzyl-␤-
d-glucopyranoside.
Fig. 5. Lacking cytotoxicity in the MTT assay: (A) extract and fractions (B). Single
compounds 7, 15–18 (in highest concentrations tested).
were co-chromatographed as reference substances. After
derivatisation with anisaldehyde reagent and heating at
120 ◦C, cis-1,2-hydroxycyclohexane was identified by its purple
colour and Rf value. Thus, structure of 2 was unambiguously
elucidated
as
the
new
cis-1,2-dihydroxycyclohexyl-6ꢀ-
O-4-hydroxybenzyl-␤-d-glucopyranoside
and
named
6ꢀ-O-4-hydroxy-benzoylgrandidentin.
In order to investigate the stereochemistry of the 1,2-
hydroxycyclohexane-ring, the acetal was cleaved by acid
hydrolysis (TFA 1 M, 50% MeOH in a Wheaton V-Vial, 120 ◦C, 15 min)
and subsequent TLC of the reaction products on silica gel (mobile
phase: hexane/ethyl acetate/tetrahydrofuran/methanol/formic
acid 12/10/3/3). Cis and trans 1,2-hydroxycyclohexane (Sigma)
For compounds 13 and 14, isolated the first time by Vinokurov
and Skrigan (1969), the NMR data were not given in the litera-
ture and thus are presented for compound 14 in the experimental
part. Compound 13 showed identical mass and UV absorption, but
20
optical rotation was different
˛
= −34. 1H NMR signals of 13
[ ]D
were identical to those of 14 ( 0.05 ppm) except for H-6: ı = 6.36,
d, J = 2.2 Hz, H-1ꢀꢀ: ı = 4.90, d, J = 7.5 Hz, H-3a: ı = 3.04, dd, J = 12.9,
16.4 Hz and H-3b: ı = 2.61, dd, J = 3.0 and 16.4 Hz. Signals in the
13C NMR spectrum were also identical (ı 0.3 ppm), except for
C-5 (ı = 161.4 ppm), C-10 (ı = 106.4 ppm) and C-1ꢀꢀ (ı = 103.3 ppm).
Stereochemical assignment (8/9 and 13/14) was carried out by CD
spectroscopy and comparison with the literature (Tyukavkina et al.
1989).
ICAM-1: Catechol
120
100
80
*
60
*
*
*
Discussion
40
20
0
As the tested Willow bark extract showed significant activity
on the TNF-␣ induced inhibition of the adhesion molecule ICAM-1
in HMEC-1 cells, a liquid/liquid partition protocol was applied to
identify the compounds responsible for the observed activity. The
EA-phase, containing a variety of phenols from different structural
classes, exhibited the most pronounced activity which resulted
at least mostly from the presence of the compound catechol (7).
Among the other isolated phenols only the flavanone aglycone eri-
odictyol (18) showed significant in vitro activity. The activity of 18
1
10
20
50
100
Fig. 4. Inhibition of TNF-␣ induced ICAM-1 expression by different concentrations
of catechol (7; 1, 10, 20, 50, 100 ␮M). Cells were pretreated for 30 min followed
by treatment with 10 ng/mL TNF-␣ for 24 h; HMEC #3–16; n = 3 in duplicates;
mean SD; *p < 0.05 vs. neg. control with TNF-␣, ICAM expression of cells without
any treatment ≈2.5%. 5 ␮M parthenolide was used as positive control.

A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
251
is interesting from the structural point of view as neither 15 nor the
corresponding flavanonols 16 and 17 were active. These points to a
relatively specific mechanism besides the always postulated effect
caused by the 3,4-dihydroxyphenyl moiety of flavonoids. Unfortu-
nately, this part of investigation was not expendable to flavones
and flavonols, as luteolin, apigenin and quercetin are active in the
ICAM-1 assay, but showed partially unspecific cytotoxic effects in
concentrations above 20 ␮M (data not shown). In literature, spe-
cific effects of 18 in comparison to luteolin or 15 are relatively
scarce and reported for example in the field of TNF-␣ mediated
cytotoxicity (Haptemariam 1997), whereas a stronger activity of
luteolin in comparison to 18 is more common (e.g. van Zanden
et al. 2004; Takano-Ishikawa et al. 2003; Sasaki et al. 2003) and
often attributed to the additional 2,3-double bond. The importance
of the latter structural feature is also addressed in the field of atten-
uation of VCAM-1, ICAM-1 and E-selectin, in other endothelial cells,
but interestingly without testing eriodictyol (Lotito and Frei 2006).
Nevertheless, Takano-Ishikawa et al. (2003) reported in accordance
with our results on a significant inhibition of E-selectin (evoked by
TNF-␣ addition) in HUVECs by eriodictyol (18). As luteolin and 18
were both reported to be able to reduce Toll like receptor medi-
ated inflammatory processes in vitro (Lee et al. 2009), an in vitro
reduction of adhesion factor expression by 18 is likely. As it is only a
trace compound in the extract a substantial contribution to the anti-
inflammatory in vivo effects seems to be questionable. In contrast, 7
is present in higher amounts (>2%) likely resulting from the degra-
dation of the 1-hydroxy-6-oxo-2-cyclohexene-1-carboxylic acid
moiety of salicin and catechin derivatives. Catechol (7) is not only
able to reduce the ICAM-1 expression in HMEC-1 cells, but is also
reported as a potent inhibitor of the transcription factor NF-kB (Ma
et al. 2003) resulting in the inhibition of LPS induced NO-production
(Zheng et al. 2008). Ruuhola et al. 2003 observed the facile degrada-
tion of 1-hydroxy-6-oxo-2-cyclohexene-1-carboxylic acid moiety
of salicortin to 7 in plant material under different pH conditions,
what is confirmed here by the presence of free 7 in the aqueous
extract. Furthermore, Knuth et al. (2011) were able to show that
salicortin can produce 7 also under cell culture conditions. As this
process is also mentionable in vivo for all compounds exhibiting a
cyclohexenon carboxylic acid moiety, like the salicyl alcohol deriva-
tives salicortin, 2ꢀ-acetylsalicortin and tremulacin (Poblecka-Olech
et al. 2007) or the catechin derivative catechin-3-O-(1-hydroxy-
6-oxo-2-cyclohexene-1-carboxylic acid)-ester (Jürgenliemk et al.
2007) the amount of 7 would further increase in vivo. Thus,
the release, presence and metabolism of 7 in vivo and its possi-
ble contribution on the anti-inflammatory effects of Willow bark
preparations is an interesting topic of further research currently
in progress. As catechol is a simple phenol also occurring in food
like coffee, malt, bread crust or cocoa powder (Lang et al. 2008)
the data are of general interest and complement the results pub-
lished by Zheng et al. 2008 on possible beneficial effects of this
compound.
spectrum of its respective host for its own secondary metabolite
profile.
Acknowledgements
Special thanks are due to Steigerwald Arzneimittelwerk GmbH
for financial support, to Dr. U. Kroll and Dr. E.-U. Heinrich from this
company for providing the extract, to Dr. E. Ades, F.J. Candal of CDC
(USA) and to Dr. T. Lawley of Emory University (USA) for providing
the HMEC-1. We thank J. Kiermaier and W. Söllner (Central Ana-
lytic of NWF IV, University of Regensburg) for measuring the mass
spectra. We are indebted to Dr. T. Burgemeister and F. Kastner (Cen-
tral Analytic of NWF IV, University of Regensburg) for recording the
NMR spectra. S. Knuth (Department of Pharmaceutical Biology, Uni-
versity of Regensburg) is acknowledged for measuring the ICAM-1
assays of phelligrin A.
References
Ades, E.W., Candal, F.J., Swerlick, R.A., George, V.G., Summers, S., Bosse, D.C., Lawle,
T.J., 1992. HMEC-1: Establishment of an immortalized human microvascular
endothelial cell line. J. Invest. Dermatol. 99, 683–690.
Bonaterra, G.A., Heinrich, E.U., Kelber, O., Weiser, D., Metz, J., Kinscherf, R., 2010. Anti-
inflammatory effects of the Willow bark extract STW 33-I (Proaktiv®) in LPS-
activated human monocytes and differentiated macrophages. Phytomedicine
17, 1106–1113.
Cook-Mills, J.M., Deem, T.L., 2005. Active participation of endothelial cells in inflam-
mation. J. Leukoc. Biol. 77, 487–495.
Fernandes, C.C., de Carvalho Cursino, L.M., de Abreu Pinheiro Novaes, J., Avelino
Demetrio, C., Liborio Pereira Jr., O., Nunez, C.V., do Amaral, I.L., 2009. Salicylates
isolated from leaves and stems of Salix martiana Leyb. (Salicaceae). Quim. Nova
4, 983–986.
Domisse, R.A., van Hoof, L., Vlietinck, A.J., 1986. Structural analysis of phenolic glu-
cosides from Salicaceae by NMR spectroscopy. Phytochemistry 5, 1201–1204.
Foo, L.Y., Karchesy, J.J., 1989. Polyphenolic glycosides from Douglas fir inner bark.
Phytochemistry 4, 1237–1240.
Galkina, E., Ley, K., 2007. Vascular adhesion molecules in atherosclerosis. Arte-
rioscler. Thromb. Vasc. Biol. 27, 2292–2301.
Gearing, A.J.H., Newman, W., 1993. Circulating adhesion molecules in disease.
Immunol. Today 14, 506–512.
Han, X.H., Hong, S.S., Hwang, J.S., Lee, M.K., Hwang, B.Y., Ro, J.S., 2007. Monoamine
oxidase inhibitory components from Cayratia japonica. Arch. Pharm. Res. 1,
13–17.
Haptemariam, S., 1997. Flavonoids as inhibitors or enhancers of the cytotoxicity of
tumor necrosis factor-alpha in L-929 tumor cells. J. Nat. Prod. 60, 775–778.
Hedner, T., Everts, B., 1998. The early clinical history of salicylates in rheumatology
and pain. Clin. Rheumatol. 17, 17–25.
Ibrahim, L.F., El-Senousy, W.M., Hawas, U.W., 2007. NMR spectral analysis of
flavonoids from Chrysanthemum coronarium. Chem. Nat. Compd. 6, 659–662.
Jürgenliemk, G., Petereit, F., Nahrstedt, A., 2007. Flavan-3-ols and procyanidins from
the bark of Salix purpurea L. Pharmazie 62, 231–234.
Khayyal, M.T., El-Ghazaly, M.A., Abdallah, D.M., Okpanyi, S.N., Kelber, O., Weiser, D.,
2005. Mechanisms involved in the anti-inflammatory effect of a standardized
willow bark extract. Arzneim. Forsch./Drug Res. 55, 677–687.
Knuth, S., Schübel, H., Hellemann, M., Jürgenliemk, G., 2011. Catechol, a bioactive
degradation product of salicortin, reduces TNF-␣ induced ICAM-1 expression in
human endothelial cells. Planta Med. 77, 1024–1026.
Kort, R., Vonk, H., Xu, X., Wolff, W.D., Crielaard, W., Hellingwerf, K.J., 1996. Evidence
for trans–cis isomerisation of the p-coumaric acid chromophore as the photo-
chemical basis of the photocycle of photoactive yellow protein. FEBS Lett. 382,
73–78.
Lang, R., Müller, C., Hofmann, T., 2008. Development of a stable isotope dilution anal-
ysis with liquid chromatography–tandem mass spectrometry detection for the
quantitative analysis of di- and trihydroxybenzenes in foods and model systems.
J. Agric. Food Chem. 54, 5755–5762.
Lee, J.K., Kim, S.Y., Kim, Y.S., Lee, W.H., Hwang, D.H., Lee, J.Y., 2009. Suppression of the
TRIF-dependent signaling pathway of Toll-like receptors by luteolin. Biochem.
Pharmacol. 77, 1391–1400.
Lotito, S.B., Frei, B., 2006. Dietary flavonoids attenuate tumor necrosis factor
alpha-induced adhesion molecule expression in human aortic endothelial cells.
Structure–function relationships and activity after first pass metabolism. J. Biol.
Chem. 281, 37102–37110.
Noticeable from the phytochemical point of view
is, besides the isolation of the new compound 6ꢀ-O-4-
hydroxybenzoylgrandidentin (2), the presence of flavanone (13,
14) and chalcone glycosides (19) substituted at 6-OH of glucose
with a coumaroyl moiety. It has to be evaluated by comprehensive
phytochemical analyses whether this is characteristic for S. pur-
purea in comparison to other Salix species. Moreover, the isolation
of 20 is noticeable as this compound has only been isolated from
Phellinus igniarius (L. ex Fr.) Quel. an edible and pharmaceutically
used mushroom of the Hymenochaetaceae (Agaricomyceti-
dae) family (Mo et al. 2003). As there is no further report on
the isolation and production of flavonoids associated to ortho-
hydroxybenzyl moiety (typical for salicyl alcohol derivatives), and
P. igniarius is often growing on Salix spec., the question arises
whether P. igniarius uses the services of the secondary metabolite
Ma, Q., Kineer, K., Ye, J., Chen, B.J., 2003. Inhibition of nuclear factor kappaB by phe-
nolic antioxidants: interplay between antioxidant signaling and inflammatory
cytokine expression. Mol. Pharmacol. 64, 211–219.
Markham, K.R., Ternai, B., 1976. 13C-NMR of flavonoids-II, flavonoids other than
flavone and flavonol aglycones. Tetrahedron 32, 2607–2612.
Mo, S.Y., He, W.Y., Yang, Y.C., Shi, J.G., 2003. Two benzyl dihydroflavones from Phelli-
nus igniarius. Chin. Chem. Lett. 14, 810–813.

252
A. Freischmidt et al. / Phytomedicine 19 (2012) 245–252
Mosman, T., 1983. Rapid colorimetric assay for cellular growth and survival: Appli-
cation to proliferation and cytotoxicity assays. J. Immunol. Methods 65, 55–63.
Nahrstedt, A., Schmidt, M., Jäggi, R., Metz, J., Khayyal, M.T., 2007. Willow bark extract:
the contribution of polyphenols to the overall effect. Wien Med. Wochenschr.
157, 348–351.
Noe, C.R., Freissmuth, J., 1995. Capillary zone electrophoresis of aldose enantiomers:
separation after derivatization with S-(−)-1-phenylethylamine. J. Chromatogr.
A 704, 503–512.
Pan, J., Zhang, S., Yan, L., Tai, J., Xiao, Q., Zou, K., Zhou, Y., Wu, J., 2008. Separation
of flavanone enantiomers and flavanone glucoside diastereomers from Bal-
anophora involucrata Hook. f. by capillary electrophoresis and reversed-phase
high-performance liquid chromatography on a C18 column. J. Chromatogr. A
1185, 117–129.
Tyukavkina, M.A., Dem’yanovich, V.M., Kolesnik, Yu.A., Ruchkin, V.E., Rulenko, I.A.,
Litvinenko, V.I., 1989. Configuration of stereoisomers of naringenin 5-glucoside
present in flamin. Chem. Nat. Compd. 2, 157–160.
Takano-Ishikawa, Y., Goto, M., Yamaki, K., 2003. Inhibitory effects of several
flavonoids on E-selectin expression on human umbilical vein endothelial
cells stimulated by tumor necrosis factor-alpha. Phytother. Res. 17, 1224–
1227.
Wagner, H., Ulrich-Merzenich, G., 2009. Synergy research: approaching a new gen-
eration of phytopharmaceuticals. Phytomedicine 16, 97–110.
van Zanden, J.J., Geraets, L., Wortelboer, H.M., van Bladeren, P.J., Rietjens, I.M.,
Cnubben, N.H., 2004. Structural requirements for the flavonoid-mediated mod-
ulation of glutathione S-transferase P 1-1 and GS-X pump activity in MCF7 breast
cancer cells. Biochem. Pharmacol. 67, 1607–1617.
Poblecka-Olech, L., van Nederkassel, A.-M., vander Heyden, Y., Krauze-Baranowska,
M., Glod, D., Baczek, T., 2007. Chromatographic analysis of salicylic compounds
in different species of the genus Salix. J. Sep. Sci. 30, 2958–2966.
Ruuhola, T., Julkunen-Tiitto, R., Vainiotalo, P., 2003. In vitro degradation of willow
salicylates. J. Chem. Ecol. 29, 1083–1097.
Vinokurov, I.I., Skrigan, A.I., 1969. Properties of a flavanone glucoside from the bark
of the sharp-leaved willow. Seryya Khimichnykh Navuk 6, 116–118.
Vlachojannis, J.E., Cameron, M., Chrubasik, S., 2009. A systematic review on the effec-
tiveness of willow bark for musculoskeletal pain. Phytother. Res. 23, 897–900.
Zapesochnaya, G.G., Kurkin, V.A., Braslavskii, V.B., Filatova, N.V., 2002. Phenolic com-
pounds of Salix acutifolia bark. Chem. Nat. Compd. 4, 314–318.
Zhang, X., Thuong, P.T., Min, B.S., Ngoc, T.M., Hung, T.M., Lee, I.S., Na, M., Seong, Y.H.,
Song, K.S., Bae, K., 2006a. Phenolic glucosides with antioxidant activity from the
stem bark of Populus davidiana. J. Nat. Prod. 69, 1370–1373.
Zhang, X., Hung, T.M., Phuong, P.T., Ngoc, T.M., Min, B.S., Song, K.S., Seong, Y.H., Bae,
K., 2006b. Anti-inflammatory activity of flavonoids from Populus davidiana. Arch.
Pharm. Res. 12, 1102–1108.
Zheng, L.T., Ryu, G.-M., Kwon, B.-M., Lee, W.-H., Suk, K., 2008. Anti-inflammatory
effects of catechols in lipopolysaccharide-stimulated microglia cells: inhibition
of microglial neurotoxicity. Eur. J. Pharmacol. 588, 106–113.
Sasaki, N., Toda, T., Kaneko, T., Baba, N., Matsuo, M., 2003. Protective effects
of flavonoids on the cytotoxicity of linoleic acid hydroperoxide toward rat
pheochromocytoma PC12 cells. Chem. Biol. Interact. 145, 101–116.
Schmid, B., Kötter, I., Heide, L., 2001. Pharmacokinetics of salicin after oral admin-
istration of a standardised willow bark extract. Eur. J. Clin. Pharmacol. 57,
387–391.
Smith, W.B., Proulx, T.W., 1976. The 13C NMR Spectra of Nine ortho-substituted
phenols. Org. Magn. Reson. 8, 205–207.
Si, C.-L., Kim, J.-K., Bae, Y.-S., Li, S.-M, 2009. Phenolic compounds in the leaves of
Populus ussuriensis and their antioxidant activities. Planta Med. 75, 1165–1167.