Synthesis of hydroxycinnamic acid glucuronides and investigation of their affinity for human serum albumin

Article abstract of DOI:10.1039/b809965k

Hydroxycinnamic acids (HCAs) are among the most abundant dietary polyphenols. Recent bioavailability studies have shown that HCAs enter the blood circulation mainly as glucuronides, which are thus most likely to express their potential health effects. In

Full text of DOI:10.1039/b809965k

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Synthesis of hydroxycinnamic acid glucuronides and investigation of their
affinity for human serum albumin†
Ste´phanie Galland, Njara Rakotomanomana, Claire Dufour, Nathalie Mora and Olivier Dangles*
Received 12th June 2008, Accepted 16th September 2008
First published as an Advance Article on the web 8th October 2008
DOI: 10.1039/b809965k
Hydroxycinnamic acids (HCAs) are among the most abundant dietary polyphenols. Recent
bioavailability studies have shown that HCAs enter the blood circulation mainly as glucuronides, which
are thus most likely to express their potential health effects. In this work, an efficient synthesis of HCA
O-arylglucuronides is developed. As for many xenobiotics, the resilience of HCA O-arylglucuronides in
plasma and subsequent delivery to tissues could be governed by their binding to human serum albumin
(HSA). Hence, the affinity of HCA O-arylglucuronides for HSA and its possible binding site were
investigated by fluorescence spectroscopy. HCA O-arylglucuronides turn out to be moderate HSA
ligands (K in the range 1–4 ¥ 104 M-1) that bind HSA in sub-domain IIA, competitively or
noncompetitively with other sub-domain IIA ligands such as dansylamide and the flavonol quercetin.
whereas concentrations 4–5 times as low are detected in untreated
Introduction
plasma.8 After consumption of high-bran cereals rich in ferulic
The mean consumption of polyphenols is of the order of 0.5 g per
acid, ferulic acid glucuronide was detected in plasma as the
main metabolite.9 After beer consumption, ferulic and caffeic
acids mainly circulate as sulfates and glucuronides, whereas much
higher levels of free p-coumaric acid are detected.10 After tomato
consumption by humans, ferulic acid is excreted as a mixture of the
free form and glucuronide, and the total recovery amounts to 11–
25% of the ingested dose. At the maximal excretion (approximately
7 h after ingestion), the molar ratio of the glucuronide to that of
the free form is 2–3 : 1.11
Polyphenol metabolites probably circulate in association with
plasma proteins. For instance, flavonol conjugates are bound
to human serum albumin (HSA), the most abundant plasma
protein.12 Binding of xenobiotics (e.g., drugs, dietary components,
toxins) to serum albumin typically modulates the rate of clearance
and delivery to tissues.13 It can also modulate the antioxidant
capacity of polyphenols, in particular their ability to inhibit lipid
peroxidation in HSA–polyunsaturated fatty acid complexes and
in low-density lipoproteins.14,15
person per day, with hydroxycinnamic acids (HCAs) making one
of the highest contributions.1,2 Hence, investigating the bioavail-
ability of HCAs is of great importance for a better understanding
of the potential health effects of polyphenols. In fact, HCAs are
diversely absorbed from the gastro-intestinal tract depending on
their structure. For instance, chlorogenic acid (5-caffeoylquinic
acid), the most abundant caffeic acid derivative in food, can be
directly taken up from the gastric compartment of rats.3 In rats,
p-coumaric and ferulic acids, but not gallic acid, have been shown
to cross the cells of the small intestine via the monocarboxylic acid
transporter.4 As for chlorogenic acid, the absorption mechanism
(determined in vitro with Caco-2 cells) involves removal of the
quinic moiety by a mucosa esterase and subsequent absorption of
caffeic acid via the monocarboxylic acid transporter.5 However,
most of the absorption of chlorogenic acid must take place in the
colon after hydrolysis to caffeic acid by bacterial esterases.
After intestinal absorption, HCAs are extensively metabo-
lized in enterocytes and hepatocytes through sulfation and
glucuronidation.6–10 These conjugates can be returned to the
gastro-intestinal tract via the bile (and partially re-absorbed)
or excreted in urine. Glucuronidation catalyzed by UDP-
glucuronosyltransferases is the main conjugation pathway. For
instance, 2 hours after ingestion of 700 mmol kg-1 of pure caffeic
acid by rats, the plasma concentration of caffeic acid glucuronides
reaches ca. 26 mM, i.e. more than 40% of the total circulating
caffeic acid concentration, whereas the concentrations of free
caffeic and ferulic acids do not exceed 1–2 mM.7 In humans,
1 hour after ingestion of a cup of coffee (ca. 1 mmol of caffeic
acid), the caffeic acid concentration reaches a maximal value in
the range 0.3–1.0 mM after deconjugation by b-glucuronidases
In this work, efficient routes for the chemical synthesis of
O-arylglucuronides of hydroxycinnamic acids are proposed. The
conjugates thus prepared are then tested for their affinity for
HSA. Improved data treatments are devised for the determination
of binding constants, binding sites and possible multiple binding
processes.
Results and discussion
Synthesis of hydroxycinnamic acid O-arylglucuronides
Access to well-characterized chemically pure polyphenol conju-
gates is a critical step toward a better understanding of their
bioavailability and cell effects. Indeed, the low concentration
of circulating polyphenol metabolites requires comparison with
authentic standards for conclusive identification, especially when
it comes to distinguishing between regioisomers. For instance,
glucuronidation of caffeic acid can yield three regioisomers, one
UMR408 Se´curite´ et Qualite´ des Produits d’Origine Ve´ge´tale, INRA, Uni-
versite´ d’Avignon, F-84000, Avignon, France. E-mail: Olivier.Dangles@univ-
avignon.fr; Fax: +33 490 14 44 41; Tel: +33 490 14 44 46
† Electronic supplementary information (ESI) available: 1H NMR spectra
of hydroxycinnamic acid glucuronides. See DOI: 10.1039/b809965k
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4253–4260 | 4253
©

View Article Online
acylglucuronide16 and two arylglucuronides. On the other hand,
the bulk of cell studies published up to now has dealt with
native polyphenols, typically aglycones, whereas conjugated forms,
especially glucuronides, are much more likely to be distributed to
tissues.
Binding to HSA
As many other xenobiotics, polyphenols could bind to serum
albumin in the blood circulation and form complexes that
modulate their resilience in plasma and their delivery to tissues.12
However, most studies reported up to now19–22 have dealt with
native polyphenols, essentially aglycones, although these forms
(when detectable) remain very minor in plasma after ingestion of
polyphenol-rich foods or supplements. Indeed, dietary polyphe-
nols typically travel in plasma as glucuronides and sulfates of
aglycones as a result of extensive conjugation in the intestinal and
liver cells.1,2 Interestingly, these conjugates themselves could bind
to HSA, as already suggested for flavonol conjugates.12,23 In this
work, the preparation of pure hydroxycinnamic acid glucuronides
offers a good opportunity to investigate their affinity toward HSA
in comparison with the parent aglycones.
Polyphenol–HSA binding can be investigated by a combination
of fluorescence techniques.12,21,24 One method consists of monitor-
ing the fluorescence of the polyphenolic ligand in the presence and
absence of HSA. This approach turned out to be disappointing
with HCAs for several reasons. First, HCAs in their free or
HSA-bound form are poorly fluorescent, so that the relatively
large ligand concentrations required make the data treatment
more complicated because of lack of linearity in the fluorescence
intensity vs. concentration plots and possible multiple bindings.
Second, in the ferulic and caffeic series, substantial photo-induced
Z–E isomerization of the carbon–carbon double bond takes place
that makes the fluorescence intensity unstable. In fact, only 4-O-
b-D-glucuronyl-p-coumaric acid (5a) could be investigated by this
method. Indeed, this ligand is much less prone to isomerization,
The synthetic route summarized on Scheme 1 is close to that
already developed for the synthesis of 5a.17 The key step is the
glucuronidation of methyl hydroxycinnamates (1a–c) by methyl
2,3,4-tri-O-acetyl-1-O-(trichloroacetimidoyl)-a-D-glucuronate (2)
activated by the boron trifluoride–diethyl ether complex in anhy-
drous dichloromethane at room temperature, which afforded the
protected glucuronides (3a–d) in moderate to good yields. In the
case of methyl caffeate, mixtures of 3¢-O-b-D-glucuronide (3d) and
4¢-O-b-D-glucuronide (3c) were obtained. As expected from the
orienting effect of the 2-O-acetyl group, the glucuronidation was
stereoselective and led to the exclusive formation of the b anomers
(J(H1–H2) = 6.9 Hz). As already observed in our previous work
with HCA O-arylglucosides,18 the final cleavage of the acetyl and
methylester groups could not satisfactorily be carried out in one
step in the ferulic and caffeic series. Indeed, 3b and 3c–d had to
be deprotected in two steps involving sugar deacetylation by mild
catalytic methanolysis (compounds 4b–d) followed by methyl ester
saponification using a stoichiometric amount of NaOH (2 equiv.).
All glucuronides were characterized by high-resolution MS and
NMR analyses. Glucuronidation at position 4¢ for 5c and at
position 3¢ for 5d were deduced from analysis of the chemical shifts
of the aromatic protons (glucuronidation lowers the electron-
donating effect of the OH group, thus producing a deshielding
of the aromatic protons ortho and para to the glucuronyl group)
in comparison with the corresponding aglycones and glucosides.18
Scheme 1 Organic synthesis of HCA arylglucuronides. Reagents and conditions: i) Methyl 2,3,4-tri-O-acetyl-1-O-(trichloroacetimidoyl)-a-D-glucuronate
(2, 1–2 equiv.) BF3·Et2O (1–2 equiv.), CH2Cl2; ii) NaOH (8 equiv.) in H2O–MeOH (1 : 1) (R = H) or MeONa (0.1 equiv.) in MeOH, then NaOH (2 equiv.)
in H2O–MeOH (1 : 1) (R = OH, OMe), followed by final acidification by a proton-exchange resin.
4254 | Org. Biomol. Chem., 2008, 6, 4253–4260
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Table 1 Binding of selected ligands to human serum albumin in a pH 7.4 phosphate buffer at 25 ◦C. Fluorescence monitoring following excitation of
the ligand or probe
10-3K/M-1
10-3f P/M-1
10-3f PL/M-1
r2, number of points
lex, lem/nm (eL at lex/M-1cm-1)
5a
21.9 ( 1.4)
107 ( 2)
156 ( 4)
22.2 ( 0.4)
16.2 ( 0.6)
4.0 ( 0.1)
3.4 ( 0.1)
6.2 ( 0.2)
8.2 ( 0.2)
177 ( 1)
348 ( 2)a
0.997, 100
0.9993, 98
0.9994, 58
0.999, 42
0.996, 44
0.997, 56
0.996, 44
0.996, 46
0.997, 58
335, 430
360, 490
360, 490
360, 490 (170)
360, 490 (170)
360, 490 (20)
360, 490 (20)
360, 490 (290)
360, 490 (290)
DNSA
27.5 ( 3.2)
27.7 ( 4.5)
535 ( 2)d
675 ( 4)d
620 ( 2)d
705 ( 2)d
477 ( 2)d
661 ( 3)d
851 ( 3)b
DNSS
1010 ( 4)c
5a (DNSA)
5a (DNSS)
5b (DNSA)
5b (DNSS)
1a (DNSA)
1a (DNSS)
—
—
—
—
—
—
a f L = 12.4 ¥ 103 M-1. b f L = 4625 M-1. c f L = 2420 M-1. d Molar fluorescence of the probe–HSA complex.
and its excitation spectrum (Fig. 1) makes it possible to set the
irradiation at a wavelength (335 nm) where absorption is weak (a
condition for proportionality between fluorescence intensity and
concentration25). Consequently, a saturatable binding isotherm
could be constructed from which the binding constant (pure 1 :
1 binding assumed) was extracted (Fig. 2, Table 1): K1 ª 22 ¥
103 M-1.
so that a correction of the fluorescence intensity at 340 nm for
this inner filter effect has to be applied in the data treatment
(see experimental part and ref. 24). Although acceptable curve-
fittings were achieved within the hypothesis of pure 1 : 1 binding,
significant improvements were noted in the fitting of the final part
of the curves (high ligand concentration) by assuming additional
1 : 2 protein–ligand binding (Fig. 3, Table 2). In the initial
calculations, both the 1 : 1 and 1 : 2 complexes are assumed
to be nonfluorescent. With 5a, the K1 values thus obtained can
be compared with the one deduced from the previous method
(enhancement of ligand fluorescence: K1 ª 22 ¥ 103 M-1): K1
ª
14 ¥ 103 M-1 (quenching of Trp fluorescence by pure 1 : 1
binding), K1 ª 10 ¥ 103 M-1 (quenching of Trp fluorescence by
1 : 1 and 1 : 2 bindings). Finally, if we consider that the K1
value deduced from the enhancement of ligand fluorescence is
most reliable because no inner filter correction is required, the
best compromise seems to refine the fluorescence quenching data
with a K1 value set constant at 22 ¥ 103 M-1 and to assume both
1 : 1 and 1 : 2 bindings and an intermediate fluorescent 1 : 1
complex (molar fluorescence intensity f PL π 0). A more reliable
value for the stepwise 1 : 2 binding constant is thus obtained:
K2 ª 13 ¥ 103 M-1. Hence, HSA seems to display two sites of close
affinity for accommodating 5a. For the other HCAs, such cross-
control cannot be carried out since K1 cannot be independently
determined from the ligand fluorescence method. However, for
the other glucuronides, a reasonable strategy could be to use the
f PL value deduced from L = 4-O-b-D-glucuronyl-p-coumaric acid
and fit the curves with f P (molar fluorescence intensity of free
HSA), K1 and K2 as the optimizable parameters (Table 2). As
an example, with 4-O-b-D-glucuronylferulic acid (5b), we obtain:
K1 ª 21 ¥ 103 (1 : 1 binding), 16 ¥ 103 (1 : 1 + 1 : 2 bindings,
f PL = 0) and 36 ¥ 103 (1 : 1 + 1 : 2 bindings, f PL π 0) M-1. These
small discrepancies reflect the difficulties inherent to quantitatively
investigating the binding to serum albumin of relatively small and
moderate ligands that can possibly partition over several binding
sites, even within the same sub-domain. Anyway, it is comforting
to note that the quercetin–HSA binding constant is consistent
with previous estimations using a variety of techniques.19–21 The
order of magnitude of the K1 values for the HCAs (aglycones) is
also in general agreement with previous reports.27,28 Taking the K1
values deduced from simple 1 : 1 binding as an acceptable scale of
relative affinity for HSA, it can be noted that the bulky glucuronyl
moiety only moderately lowers the affinity of HCAs for HSA (at
most, by a factor of 2).
Fig. 1 Absorption (1), excitation (2) and emission (3) spectra of
4-O-b-D-glucuronyl-p-coumaric acid (5a) in the absence (A) or presence
(B) of HSA (pH 7.4 phosphate buffer, 25 ◦C). Excitation and emission
wavelengths are set at 335 and 430 nm, respectively.
For the other ligands, we had to consider the HSA intrinsic fluo-
rescence due to its single Trp residue (Trp-214), which is located in
sub-domain IIA where small negatively charged aromatic ligands
are most likely to bind.26 The signal intensity and its sensitivity
to quenching by sub-domain IIA binders make it possible to
use small protein and ligand concentrations. However, all the
selected ligands absorb at the excitation wavelength (295 nm)
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4253–4260 | 4255
©

View Article Online
Fig. 2 Changes in the fluorescence intensity of 4-O-b-D-glucuronyl-p-coumaric acid (5a) at 430 nm (lex = 335 nm) as a function of its concentration.
Initial HSA concentration = 75 mM (pH 7.4 phosphate buffer, 25 ◦C). Fluorescence of HSA-bound ligand (●) and free ligand (ꢀ), absorbance of free
ligand at 335 nm (▲).
Table 2 Binding of selected ligands to human serum albumin (3.75 mM) in a pH 7.4 phosphate buffer at 25 ◦C. Fluorescence monitoring at 340 nm
following excitation of HSA (single Trp residue) at 295 nm
a
Ligand
10-3K/M-1
10-5f P/M-1
10-5f PL/M-1
r2, number of points
eL(295 nm)/M-1 cm-1
1a
39.5 ( 1.3)
28.7 ( 1.7), 14.6 ( 3.0)
33.6 ( 1.1)
23.4 ( 1.3), 15.6 ( 2.7)
41.8 ( 0.9)
32.1 ( 0.8), 11.9 ( 1.1)
14.1 ( 0.3)
10.1 ( 0.4), 8.3 ( 1.0)
21.9d, 13.2 ( 0.9)
41.8 ( 0.9)
32.1 ( 0.8), 11.9 ( 1.1)
35.6 ( 1.2), 17.5 ( 0.6)
19.8 ( 0.7)
12.4 ( 0.8), 15.9 ( 2.5)
26.2 ( 2.5), 22.3 ( 1.9)
28.7 ( 1.1)
95.1 ( 0.7)
93.1 ( 0.5)
95.9 ( 0.6)
93.6 ( 0.5)
106.3 ( 0.5)
104.1 ( 0.3)
116.8 ( 0.4)
114.9 ( 0.3)
114.6 ( 0.3)
106.3 ( 0.5)
104.1 ( 0.3)
114.9 ( 0.2)
111.4 ( 0.7)
108.6 ( 0.5)
108.0 ( 0.5)
118.5 ( 0.9)
114.7 ( 0.4)
113.9 ( 0.4)
114.2 ( 0.7)
112.4 ( 0.5)
125.8 ( 0.6)
—
—
—
—
—
—
—
—
0.997, 51
0.998, 51
0.997, 50
0.999, 50
0.998, 54
0.9997, 54
0.998, 63
0.9995, 63
0.9995, 63
0.998, 54
0.9997, 54
0.9997, 58
0.997, 59
0.999, 59
0.999, 59
0.997, 59
0.9994, 59
0.9993, 59
0.999, 45
0.9993, 63
0.999, 56
14420
14100
13110
10840
1ab
1b
1bb
1c
1cb
5a
5ab
5ac
65.9 ( 2.7)
—
5b
13110
12850
14310
9410
5bb
—
5bc
66d
5c
—
5cb
—
5cc
66d
5d
—
5db
16.3 ( 0.8), 23.3 ( 2.6)
30.2 ( 2.2), 34.8 ( 2.2)
222.0 ( 4.7)
—
5dc
66d
Quercetin
DNSA
DNSS
—
107d, 12.9 ( 0.6)
156d, 10.1 ( 0.7)
52.9 ( 1.1)
55.7 ( 1.3)
a Fluorescence intensity corrected by a factor exp(-eLlLt) with l = 0.65 cm (Lt: total ligand concentration). b 1 : 1 and 1 : 2 bindings (nonfluorescent 1 : 1
and 1 : 2 complexes). c 1 : 1 and 1 : 2 bindings (fluorescent 1 : 1 complex, nonfluorescent 1 : 2 complex). d Set constant.
The highly sensitive fluorescent probes dansylamide (DNSA)
and dansylsarcosine (DNSS) were used for competitive exper-
iments with the HCAs to gain information about substantial
differences in binding locations. DNSA is a sub-domain IIA binder
whereas DNSS could also partition in sub-domain IIIA.29 The
strategy developed above for 4-O-b-D-glucuronyl-p-coumaric acid
(5a) was also applied to the dansyl probes to estimate the values
of their 1 : 1 and 1 : 2 binding constants to HSA. Thus, K1
was accurately determined from the ligand fluorescence method
(excitation at 360 nm, emission at 490 nm) and this value was
used in the quenching of the Trp fluorescence to estimate K2
(Tables 1 and 2). Then, the fluorescence of the dansyl probe–HSA
complex was gradually quenched by increasing concentrations of
5a. The quenching curves were analyzed within the hypothesis
of pure competitive 1 : 1 binding for both the probe and ligand
and by taking into account a very weak inner filter effect due
to absorption of the ligand at the excitation wavelength (Fig. 4).
These assumptions were nicely validated by the K1 values thus
extracted (K1 ª 22 ¥ 103 M-1 with DNSA, 16 ¥ 103 M-1 with
DNSS) which are in agreement with our previous estimate (K1 ª
22 ¥ 103 M-1) (Table 1). Although the K1 values of the other
HCAs are less accurately determined, it can be safely stated that
this competitive binding is not a general rule. For instance, the
quenching of both DNSS–HSA and DNSA–HSA fluorescence
4256 | Org. Biomol. Chem., 2008, 6, 4253–4260
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Fig.
4
Changes in the fluorescence intensity of dansylamide at
490 nm (lex
=
360 nm) as a function of the concentration of
4-O-b-D-glucuronyl-p-coumaric acid (5a). Initial HSA and DNSA con-
centrations = 75 mM (pH 7.4 phosphate buffer, 25 ◦C). The solid line is
the result of the curve-fitting.
Fig. 3 Changes in the fluorescence intensity of HSA at 340 nm (lex
=
295 nm) as a function of the 4-O-b-D-glucuronylferulic acid (5b) concen-
tration (pH 7.4 phosphate buffer, 25 ◦C). Initial HSA concentration =
3.75 mM. A: curve-fitting assuming pure 1 : 1 binding, B: curve-fitting 1 :
1 and 1 : 2 bindings.
by p-coumaric acid (1a) and 4-O-b-D-glucuronylferulic acid
(5b) gave K1 values significantly weaker than the lower limit
(15 ¥ 103 M-1) determined from the quenching of HSA fluo-
rescence, which is indicative of possible noncompetitive binding.
Hence, glucuronidation, or even a more minor structural change
such as an additional methoxy group, is enough to induce a
substantial move of the ligand within sub-domain IIA.
Fig.
530 nm (lex
5
Changes in the fluorescence intensity of quercetin at
450 nm) as function of the concentration of
=
a
4-O-b-D-glucuronyl-p-coumaric acid (5a). Initial HSA and quercetin
concentrations = 75 mM (pH 7.4 phosphate buffer, 25 ◦C). The solid
line is the result of a simulation assuming competitive binding.
Finally, 5a acid barely lowers the fluorescence of the quercetin–
HSA complex at 530 nm. The corresponding IF vs. Lt curve
remains much higher than the theoretical curve assuming com-
petitive binding and construction from K1 ª 22 ¥ 103 M-1 for
the ligand–HSA complex and 222 ¥ 103 M-1 for the quercetin–
HSA complex, and from the values for the molar fluorescence
(data not shown) is also consistent with their binding to HSA
noncompetitively with quercetin.22
In conclusion, hydroxycinnamic acids and their O-arylglucu-
ronides moderately bind HSA in the IIA-domain (K1 = 1–4 ¥
104 M-1). Their interaction with HSA is clearly noncompetitive
with quercetin and either competitive or noncompetitive with
DNSS and DNSA. The moderate affinity of HCA glucuronides for
HSA should be large enough to ensure substantial in vivo binding
given the high concentration of HSA in plasma (ca. 0.6 mM)
and the low concentrations of circulating polyphenol metabolites
(<1 mM).
intensity of free and HSA-bound quercetin (f Q ª 4290 M-1, f QP
ª
3 ¥ 105 M-1 at 530 nm) (Fig. 5). It can thus be concluded that the
bindings of 4-O-b-D-glucuronyl-p-coumaric acid and quercetin are
noncompetitive. Unlike competition with the dansyl probes, this
behaviour appears fairly general in the series of HCAs investigated.
The fact that the dansyl probes, despite their relatively high affinity
for HSA, only weakly quench the fluorescence of bound quercetin
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4253–4260 | 4257
©

View Article Online
Compounds 3c–3d. Same procedure as for 3a except for the
concentrations of BF3·Et2O and 2 (1 equiv.). The two regioisomers
were isolated together. Rf (AcOEt–C6H12, 6 : 4) = 0.62, yield 35%.
1H-NMR (CDCl3): 7.59 (1H, d, J = 15.9, Hb), 7.27–6.96 (3H,
m, H2, H5, H6), 6.28 (1H, d, J = 15.9, Ha), 5.60–4.80 (3.4 H,
m, H1¢, H2¢, H3¢, H4¢), 4.61 (0.6 H, d, J = 9.9, H1¢), 4.24 (1H,
d, J = 9.0, H5¢), 3.81–3.76 (6H, m, 2 CO2CH3), 2.14–2.05 (9H,
m, 3 OCOCH3). 13C-NMR (CDCl3): 170.8, 170.2, 168.3, 167.5
(2 CO2CH3, 3 OCOCH3), 150.3–148.3–145.5 (C4, C3, Cb), 132.4–
128.0–126.7 (C1, C2, C6), 118.3 (C5), 116.7 (Ca), 91.0 (C1¢), 73.4,
72.0, 71.9, 69.7 (C2¢, C3¢, C4¢, C5¢), 54.0, 52.5 (2 CO2CH3), 21.5 (3
OCOCH3).
Experimental
Chemicals
p-Coumaric acid, ferulic acid, caffeic acid, quercetin, dansylamide
(DNSA), dansylsarcosine (DNSS), glucurono-3,6-lactone and
HSA (fraction V, MM = 66500 g mol-1) were all purchased
from Sigma-Aldrich. Glucuronyl donor 2 was synthesized from
glucurono-3,6-lactone according to a 3-step procedure already
described in the literature.17,30 Methyl p-coumarate, ferulate and
caffeate (1a–1c) were prepared as already reported.18
Analyses
4-O-b-D-Glucuronyl-p-coumaric acid (5a). 3a (0.5 mmol) was
stirred in 1 M NaOH (8 equiv.)–MeOH–H2O (1 : 3 : 2) (24 ml)
at room temperature for 15 h. After acidification with a proton-
exchange resin and concentration, 5a was isolated. Rf (CH2Cl2–
MeOH, 1 : 1) = 0.32, yield 95%. 1H-NMR (CD3OD): 7.65 (1H, d,
J = 15.9, Hb), 7.58 (2H, d, J = 8.7, H2, H6), 7.13 (2H, d, J = 8.7,
H3, H5), 6.39 (1H, d, J = 15.9, Ha), 5.07 (1H, d, J = 7.8, H1¢), 4.04
(1H, d, J = 9.6, H5¢), 3.64–3.52 (3H, m, H2¢, H3¢, H4¢). 13C-NMR
(CD3OD): 171.5 (C6¢), 169.9 (CO2H), 145.2 (C4, Cb), 130.2 (C2,
C6), 130.1 (C1), 117.3, 117.0 (C3, C5, Ca), 101.2 (C1¢), 76.4, 76.3,
73.9, 72.3 (C2¢, C3¢, C4¢, C5¢). HRMS: m/z 339.0718 ([M - H]-)
(339.0722, calcd for C15H16O9). [a]D-72.1 (c 1, MeOH).
1H- and 13C-NMR spectra were recorded on a 300 MHz Bruker
apparatus at 27 ◦C. Chemical shifts (d) are given in ppm
(solvent as the internal reference), coupling constants (J) are
given in Hz. High-resolution mass analyses were carried out
on Qstar Elite instrument (Applied Biosystems SCIEX, Foster
City, USA) equipped with API. Mass detection was performed
in the negative or positive electrospray ionization mode. Steady-
state fluorescence spectra were recorded on a thermostatted Safas
Xenius spectrofluorometer. The excitation and emission slit widths
were set at 10 nm. All studies were performed at 25 ( 1) ◦C. Two
types of excitation–emission conditions were used: a) excitation at
335 or 360 nm (4-O-b-D-glucuronyl-p-coumaric acid and dansyl
probes, respectively), emission light collected between 350 and
530 nm; or b) excitation at 295 nm (HSA Trp residue), emission
light collected between 260 and 400 nm.
Compound 4b. 3b (0.6 mmol) was stirred in MeOH (30 ml)
containing ca. 0.1 equiv. MeONa at room temperature for 24 h.
After acidification with a proton-exchange resin, concentration
and chromatography on silica gel (eluent AcOEt), 4b◦was isolated.
Rf (AcOEt) = 0.15, yield 64%, melting point 180–181 C. 1H-NMR
(CD3OD): 7.66 (1H, d, J = 15.9, Hb), 7.29–7.20 (3H, m, H2, H5,
H6), 6.50 (1H, d, J = 15.9, Ha), 4.96 (1H, H1¢, masked by water
signal), 3.91–3.37 (13H, m + 3 s, H2¢, H3¢, H4¢, H5¢ + 3 OCH3). 13C-
NMR (CD3OD): 172.6 (C6¢), 167.3, 155.3 (CO2CH3), 149.2 (Cb),
135.1, 127.4, 126.4, 122.6 (C1, C2, C4, C6), 121.0, 117.1, 115.8 (C3,
C5, Ca), 103.0 (C1¢), 74.2, 70.4 (C2¢, C3¢, C4¢, C5¢), 59.9, 56.2, 55.4
(3 OCH3).
Chemical syntheses
Compound 3a. BF3·Et2O (2 equiv.‡ ) was added to a solution
of 1a (1 mmol) and 2 (2 equiv.) in dry CH2Cl2 (4 ml) under N2 in the
˚
presence of 4 A molecular sieves. The mixture was stirred at room
temperature for 1 h, then concentrated and directly subjected to
chromatography on silica gel (eluent AcOEt–C6H12, 3 : 7) to afford
3a. Rf (AcOEt–C6H12, 6 : 4) = 0.53, yield 43%. 1H-NMR (CDCl3):
7.66 (1H, d, J = 15.9, Hb), 7.50 (2H, d, J = 8.7, H2, H6), 7.02 (2H,
d, J = 8.7, H3, H5), 6.36 (1H, d, J = 15.9, Ha), 5.39–5.31 (3H, m,
H2¢, H3¢, H4¢), 5.21 (1H, d, J = 6.9, H1¢), 4.23 (1H, d, J = 9.3, H5¢),
3.82 (3H, s, CO2CH3), 3.75 (3H, s, CO2CH3 (GlcU)), 2.08–2.06
(9H, 3 s, 3 OCOCH3). 13C-NMR (CDCl3): 170.5, 169.7, 169.6,
167.2 (2 CO2CH3, 3 OCOCH3), 158.4 (C4), 144.3 (Cb), 130.0 (C2,
C6), 128.3 (C1), 117.6, 117.2 (C3, C5, Ca), 99.0 (C1¢), 73.1, 72.1, 71.4,
69.4 (C2¢, C3¢, C4¢, C5¢), 53.4–52.1 (2 COOCH3), 21.0 (3 OCOCH3).
The NMR data of 3a are consistent with the literature.17
Compounds 4c–4d. Same procedure as for 4b. Rf (AcOEt) =
0.2, yield 50%. 1H-NMR (CD3OD): 7.60 (0.6H, d, J = 15.9, Hb(4d)),
7.58 (0.4H, d, J = 15.9, Hb(4c)), 7.44 (0.6H, d, J = 2.1, H2(4d)), 7.21
(0.6H, dd, J = 2.1 and J = 8.4, H6(4d)), 7.13 (0.4H, d, J = 2.1,
H2(4c)), 7.12 (0.4H, d, J = 8.4, H5(4c)), 7.05 (0.4H, dd, J = 2.1,
J = 8.4, H6(4c)), 6.89 (0.6H, d, J = 8.4, H5(4d)), 6.38 (0.4H, d, J =
15.9, Ha(4c)), 6.34 (0,6H, d, J = 15.9, Ha(4d)), 4.96 (0.4H, d, J =
7.5, H1¢(4c)), 4.95 (0.6H, d, J = 7.5, H1¢(4d)), 4.10 (0.6H, d, J = 9.6,
H5¢(4d)), 4.07 (0.4H, d, J = 9.6, H1¢(4c)), 3.83 (1.8H, s, OCH3(4d)), 3.80
(1.2H, s, OCH3(4c)), 3.78 (3H, s, OCH3(GlcU)), 3.73–3.51 (3H, m, H2¢,
H3¢, H4¢).
Compound 3b. Same procedure as for 3a. Rf (AcOEt–C6H12,
6 : 4) = 0.56, yield 84%. 1H-NMR (CDCl3): 7.63 (1H, d, J = 15.9,
Hb), 7.11 (3H, m, H2, H5, H6), 6.36 (1H, d, J = 15.9, Ha), 5.36–
5.30 (3H, m, H2¢, H3¢, H4¢), 5.10 (1H, d, J = 6.9, H1¢), 4.13 (1H, d,
J = 9.3, H5¢), 3.86, 3.82, 3.76 (9H, 3 s, 3 OCH3), 2.10–2.05 (9H,
3 s, 3 OCOCH3). 13C-NMR (CDCl3): 170.5, 169.7, 167.8, 159.6 (2
CO2CH3, 3 OCOCH3), 151.2, 147.9, 144.6 (C3, C4, Cb), 131.5 (C1),
120.4 (C6), 117.6, 117.4 (Ca, C5), 111.9 (C2), 100.7 (C1¢), 73.1, 72.1,
71.4, 69.6 (C2¢, C3¢, C4¢, C5¢), 56.5 (OCH3), 53.4, 52.1 (2 CO2CH3),
21.1, 20.9 (3 OCOCH3).
4-O-b-D-Glucuronylferulic acid (5b). 4b (0.5 mmol) was stirred
in 1 M NaOH (2 equiv.)–MeOH–H2O (1 : 3 : 2) (6 ml) at room
temperature for 15 h. After acidification with a proton-exchange
resin and concentration, 5b was isolated. Rf (C18-silica gel, 0.05%
aq. HCO2H–MeCN, 1 : 1) = 0.89, yield 80%. 1H-NMR (CD3OD):
7.64 (1H, d, J = 15.9, Hb), 7.28–7.18 (3H, m, H2, H5, H6), 6.42
(1H, d, J = 15.9, Ha), 5.06 (1H, d, J = 7.5, H1¢), 3.99 (1H, d, J =
9.6, H5¢), 3.92 (3H, s, OCH3), 3.67–3.52 (3H, m, H2¢, H3¢, H4¢). 13C-
NMR (CD3OD): 171.5 (C6¢), 169.9 (CO2H), 150.6, 149.0, 145.4
(Cb, C3, C4), 130.4, 122.6, 117.4, 117.2, 112.0 (C1, C2, C5, C6, Ca),
‡ Concentration not optimized: a catalytic amount could suffice.17,31
4258 | Org. Biomol. Chem., 2008, 6, 4253–4260
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
101.7 (C1¢), 76.6, 76.0, 73.9, 72.3 (C2¢, C3¢, C4¢, C5¢), 55.3 (OCH3).
HRMS: m/z 369.0825 ([M - H]-) (369.0828, calcd for C16H18O10).
[a]D -19.9 (c 1, MeOH).
and C the total protein concentration. The molar fluorescence
intensity of the free ligand (f L) was estimated from a linear plot of
the fluorescence intensity vs. ligand concentration in the absence
of HSA. The weak fluorescence intensity of free HSA (molar
fluorescence intensity f P) detected in the absence of ligand was
taken into account.
4-O-b-D-Glucuronylcaffeic acid (5c) and 3-O-b-D-glucuronyl-
caffeic acid (5d). Same procedure as for 5b. Rf (C18 silica gel,
0.05% aq. HCO2H–MeCN, 1 : 1) = 0.77, yield 80%. 5c and 5d
were separated by semi-preparative chromatography on C18 silica
gel (eluent 0.05% aq. HCO2H–MeOH 9 : 1 to pure MeOH).
IF = f L[L] + f P[P] + f PLK1[L][P]
Lt = [L](1 + K1[P])
(1)
(2)
(3)
1
Data for 5c: H-NMR (CD3OD): 7.58 (1H, d, J = 15.9, Hb),
7.17 (1H, d, J = 8.4, H5), 7.13 (1H, d, J = 2.1, H2), 7.06 (1H, dd,
J = 8.4 and J = 2.1, H6), 6.34 (1H, d, J = 15.9, Ha), 4.94 (1H,
d, J = 7.2, H1¢), 4,00 (1H, d, J = 9.6, H5¢), 3.64–3.53 (3H, m, H2¢,
H3¢, H4¢). 13C-NMR (CD3OD): 169.9 (2 CO2H), 148.0, 147.9, 145.4
(Cb, C3, C4), 130.9, 121.4, 117.7, 117.3, 115.5 (C1, C2, C5, C6, Ca),
103.0 (C1¢), 76.3, 76.0, 73.8, 72.4 (C2¢, C3¢, C4¢, C5¢). HRMS: m/z
357.0815 ([M + H]+), 374.1077 ([M + NH4]+) (357.0816, 374.1081,
calcd for C16H18O10). [a]D -84.4 (c 1, MeOH).
C = [P](1 + K1[L])
Quenching of the intrinsic fluorescence of HSA by the ligands.
The excitation wavelength was selected so as to maximize the
fluorescence of the single Trp residue of HSA. However, most
ligands substantially absorb light at the excitation wavelength
(295 nm) so that an inner filter correction is necessary. Hence,
the protein fluorescence intensity is expressed as eqn (4), which is
used with eqn (2) and eqn (3) in the curve-fitting procedure.
1
Data for 5d: H-NMR (CD3OD): 7.59 (1H, d, J = 15.9, Hb),
7.47 (1H, d, J = 2.1, H2), 7.22 (1H, dd, J = 8.1 and J = 2.1,
H6), 6.90 (1H, d, J = 8.1, H5), 6.31 (1H, d, J = 15.9, Ha), 4.92
(1H, d, J = 7.2, H1¢), 4.03 (1H, d, J = 9.6, H5¢), 3.65–3.54 (3H,
m, H2¢, H3¢, H4¢). 13C-NMR (CD3OD): 171.7 (C6¢), 170.2 (CO2H),
150.6, 146.1, 145.7 (Cb, C3, C4), 127.4, 125.6, 117.9, 117.0 (C1, C2,
C5, C6), 115.9 (Ca), 103.7 (C1¢), 76.4, 75.9, 73.9, 72.3 (C2¢, C3¢, C4¢,
C5¢). HRMS: m/z 357.0814 ([M + H]+), 374.1080 ([M + NH4]+)
(357.0816, 374.1081, calcd for C16H18O10). [a]D -73.4 (c 1, MeOH).
IF = f P[P]exp(-eLlLt)
(4)
In eqn (4), eL stands for the molar absorption coefficient of the
ligand at the excitation wavelength, and has been checked to be
identical for the bound or free ligand. Its value is determined
independently by UV-visible spectroscopy from a Beer’s plot.
Finally, l is the mean distance travelled by the excitation light
at the site of emission light detection. For the spectrometer used
in this work, l is estimated to be 0.65 cm.
Fluorescence measurements
When necessary, additional 1 : 2 protein-ligand binding (step-
wise binding constant K2) was taken into account, leading to eqns
(5)–(7):
Solutions were prepared daily by dissolving HSA in a pH 7.4
buffer (50 mM phosphate–100 mM NaCl).
Interaction of albumin with a single ligand. Small aliquots (1–
80 mL) of a concentrated ligand solution in MeOH or in MeOH–
DMSO (9 : 1) were added via syringe to 2 mL of a 3.75 mM
HSA solution placed in a quartz cell (path length: 1 cm). In all
experiments, the maximal cosolvent concentration was 5%.
Lt = [L] + K1[P][L] + K1K2[P][L]2
C = [P](1 + K1[L] + K1K2[L]2)
(5)
(6)
(7)
IF = [P](f P + f PLK1[L])exp(-eLlLt)
In eqn (7), the fluorescence of complex PL2 is neglected.
Interaction of albumin with a dansyl probe and a second ligand.
To 2 mL of a 75 mM HSA solution were successively added 20 mL of
a 7.5 mM solution of the dansyl probe in MeOH (1 : 1 HSA–probe
molar ratio) and aliquots (1–300 mL) of a concentrated solution
of the second ligand in MeOH or MeOH–DMSO (9 : 1).
Quenching of the fluorescence of dansyl probe–HSA complexes
by the ligands. The excitation wavelength was selected so as
to maximize the fluorescence of the bound probe. Assuming
competition between the dansyl probe D and ligand L and pure 1 :
1 binding, eqns (8) and (9) can be derived and used in the curve-
fitting of the IF vs. Lt curve for the determination of optimized
values for parameters K1 and f DP (after preliminary determination
of KD in the absence of L and of f D in the absence of L and P).
Binding data treatment
All calculations were carried out with the least-square regression
program Scientist (MicroMath, Salt Lake City, USA). Standard
deviations and correlation coefficients are reported. Beside the
expression of the fluorescence intensity IF, the typical relationships
used in the curve-fitting procedures were combinations of the mass
law for the complexes and mass conservation for the ligand L,
protein P and dansyl probes D (see below).
fD + fDPKD[P]
1+ KD[P]
IF = Dt
exp(-eLlL )
(8)
(9)
t
Ê
ˆ
KDDt
K1Lt
C = [P] 1+
+
Fluorescence of the ligands in the presence of HSA. The excita-
tion wavelength was selected so as to maximize the fluorescence of
the bound ligand or probe (4-O-b-D-glucuronyl-p-coumaric acid,
DNSA, DNSS). Assuming 1 : 1 binding, optimized values for
the binding constant (K1) and the molar fluorescence intensity of
the complex (f PL) were estimated by fitting the IF vs. Lt curves
against eqns (1)–(3) where Lt is the total ligand concentration
Á
Ë
˜
1+ KD[P] 1+ K [P]
¯
1
where f D and f DP are the molar fluorescence intensities of the
dansyl probe and dansyl probe–albumin complex, respectively, Dt
is the total concentration of the dansyl probe, KD the probe–HSA
binding constant and eL the molar absorption coefficient of the
ligand at the excitation wavelength.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 4253–4260 | 4259
©

View Article Online
15 K. M. Janisch, G. Williamson, P. Needs and G. W. Plumb, Free Radical
Res., 2004, 38, 877–884.
16 A. V. Stachulski, J. R. Harding, J. C. Lindon, J. L. Maggs, B. K. Park
and I. D. Wilson, J. Med. Chem., 2006, 49, 6931–6945.
17 A. G. Pearson, M. J. Kiefel, V. Ferro and M. V. Itzstein, Carbohydr.
Res., 2005, 340, 2077–2085.
References
1 M. Clifford and J. E. Brown, in: Flavonoids: Chemistry, Biochemistry
& Applications, ed. O. Andersen and K. Markham, CRC Press, Boca
Raton, 2006, pp. 319–370.
2 C. Manach, G. Williamson, C. Morand, A. Scalbert and C. Remesy,
Am. J. Clin. Nutr., 2005, 81, 230S–242S.
3 S. Lafay, A. Gil-Izquierdo, C. Manach, C. Morand, C. Besson and A.
Scalbert, J. Nutr., 2006, 136, 1192–1197.
4 Y. Konishi, Y. Hitomi and E. Yoshioka, J. Agric. Food Chem., 2004, 52,
2527–2532.
5 Y. Konishi and S. Kobayashi, J. Agric. Food Chem., 2004, 52, 2518–
2526.
6 S. M. Kern, R. N. Bennett, P. W. Needs, F. A. Mellon, P. A. Kroon
and M. T. Garcia-Conesa, J. Agric. Food Chem., 2003, 51, 7884–
7891.
18 S. Galland, N. Mora, M. Abert-Vian, N. Rakotomanomana and O.
Dangles, J. Agric. Food Chem., 2007, 55, 7573–7579.
19 D. W. Boulton, U. K. Walle and T. Walle, J. Pharm. Pharmacol., 1998,
50, 243–249.
20 F. Zsila, Z. Bikadi and M. Simonyi, Biochem. Pharmacol., 2003, 65,
447–456.
21 B. Sengupta and P. K. Sengupta, Biochem. Biophys. Res. Commun.,
2002, 299, 400–403.
22 C. Dufour and O. Dangles, Biochim. Biophys. Acta, 2005, 1721, 164–
173.
7 K. Azuma, K. Ippoushi, M. Nakayama, H. Ito, H. Higashio and J.
Terao, J. Agric. Food Chem., 2000, 48, 5496–5500.
8 M. Nardini, E. Cirillo, F. Natella and C. Scaccini, J. Agric. Food Chem.,
2002, 50, 5735–5741.
9 S. M. Kern, R. N. Bennett, F. A. Mellon, P. A. Kroon and
M. T. Garcia-Conesa, J. Agric. Food Chem., 2003, 51, 6050–
6055.
23 C. Manach, C. Morand, V. Crespy, C. Demigne, O. Texier, F. Regerat
and C. Remesy, FEBS Lett., 1998, 426, 331–336.
24 D. E. Epps, T. J. Raub, V. Caiolfa, A. Chiari and M. Zamai, J. Pharm.
Pharmacol., 1999, 51, 41–48.
25 B. Valeur, Molecular fluorescence, Wiley-VCH, Weinheim, 2002.
26 J. Ghuman, P. A. Zunszain, I. Petitpas, A. A. Bhattacharya, M. Otagiri
and S. Curry, J. Mol. Biol., 2005, 353, 38–52.
10 M. Nardini, F. Natella, C. Scaccini and A. Ghiselli, J. Nutr. Biochem.,
27 J. Min, X. Meng-Xia, Z. Dong, L. Yuan, L. Xiao-Yu and C. Xing,
J. Mol. Struct., 2004, 692, 71–80.
2006, 17, 14–22.
11 L. C. Bourne and C. Rice-Evans, Biochem. Biophys. Res. Commun.,
1998, 253, 222–227.
12 O. Dangles, C. Dufour, C. Manach, C. Morand and C. Remesy,
Methods Enzymol., 2001, 335, 319–333.
28 J. Kang, Y. Liu, M.-X. Xie, S. Li, M. Jiang and Y.-D. Wang, Biochim.
Biophys. Acta, 2004, 1674, 205–214.
29 D. E. Epps, T. J. Raub and F. J. Kezdy, Anal. Biochem., 1995, 227,
342–350.
13 T. Peters, All about albumin: biochemistry, genetics and biomedical
applications, Academic Press, San Diego, 1996.
14 C. Dufour, M. Loonis and O. Dangles, Free Radical Biol. Med., 2007,
43, 238–250.
30 B. Fischer, A. Nudelman, M. Ruse, J. Herzig, H. E. Gottlieb and E.
Keinan, J. Org. Chem., 1984, 49, 4988–4993.
31 R. T. Brown, F. Scheinmann and A. V. Stachulski, J. Chem. Res. (S),
1997, 370–371.
4260 | Org. Biomol. Chem., 2008, 6, 4253–4260
This journal is
The Royal Society of Chemistry 2008
©