The inclusion complex of rosmarinic acid into beta-cyclodextrin: A thermodynamic and structural analysis by NMR and capillary electrophoresis

Article abstract of DOI:10.1016/j.foodchem.2016.04.008

This work focuses on the characterization of the rosmarinic acid (RA)-β-cyclodextrin (CD) complex in aqueous solution by ¹H NMR (1D- and 2D-ROESY), completed with studies by capillary electrophoresis (CE). From the ¹H NMR data, the s

Full text of DOI:10.1016/j.foodchem.2016.04.008

Accepted Manuscript
The inclusion complex of rosmarinic acid into beta-cyclodextrin: a thermody-
namic and structural analysis by NMR and capillary electrophoresis
Amra Aksamija, Ange Polidori, Raphaël Plasson, Olivier Dangles, Valérie
Tomao
PII:
S0308-8146(16)30516-7
DOI:
http://dx.doi.org/10.1016/j.foodchem.2016.04.008
FOCH 19015
Reference:
Food Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
30 September 2015
2 March 2016
4 April 2016
Please cite this article as: Aksamija, A., Polidori, A., Plasson, R., Dangles, O., Tomao, V., The inclusion complex
of rosmarinic acid into beta-cyclodextrin: a thermodynamic and structural analysis by NMR and capillary
electrophoresis, Food Chemistry (2016), doi: http://dx.doi.org/10.1016/j.foodchem.2016.04.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1
The inclusion complex of rosmarinic acid into beta-cyclodextrin: a
thermodynamic and structural analysis by NMR and capillary
electrophoresis
1
2
3
4
5
6
7
Amra Aksamija,a Ange Polidori,b Raphaël Plasson,a Olivier Dangles,a Valérie Tomaoa*
8
9
10
11
a University of Avignon, INRA, UMR408 SQPOV, 84000 Avignon, France
b University of Avignon, UMR5247 CBSA, 84000 Avignon, France
12
13
14
15
16
17
18
19
* Corresponding author.
E-mail address: valerie.tomao@univ-avignon.fr (V. Tomao).
20

2
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
ABSTRACT
This work focuses on the characterization of the rosmarinic acid (RA) - β-cyclodextrin (CD)
complex in aqueous solution by 1H NMR (1D- and 2D-ROESY), completed with studies by
capillary electrophoresis (CE). From the 1H NMR data, the stoichiometry of the complex was
determined by a Job’s plot and the binding constant was estimated from a linear regression
(Scott’s method). At pH 2.9, the results showed that RA binds CD with a 1:1 stoichiometry
and a binding constant Kb of 445 ( 53) M-1 or 465 ( 81) M-1 depending on the CD protons
(H-5 or H-3) selected for the evaluation. The Kb value was also calculated from the CD-
induced chemical shifts of each RA proton in order to collect information on the structure of
the complex.
The pH dependence of Kb revealed that the RA carboxylic form displays the highest affinity
for CD. An investigation by capillary electrophoresis fully confirmed these results. 2D
ROESY analysis provided detailed structural information on the complex and showed a
strong correlation between H-3 and H-5 of CD and most RA protons. In conclusion, RA, an
efficient phenolic antioxidant from rosemary with a marketing authorization, spontaneously
forms a relatively stable inclusion complex with CD in water.
Keywords
Beta-cyclodextrin, rosmarinic acid, inclusion complex, NMR, ROESY, capillary
electrophoresis

3
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
1. Introduction
Naturally occurring phenolic compounds are currently used in the food industry as additives,
especially in functional foods due to their potential health promotion in terms of antioxidant
and anti-inflammatory protection (Campos-Vega et al., 2010).
Rosmarinic acid (RA), an ester of caffeic acid and 3,4-dihydroxyphenyllactic acid, is a natural
phenolic compound commonly found in many Lamiaceae herbs such as Rosmarinus
officinalis, an aromatic evergreen shrub. Besides its antioxidant and anti-inflammatory
activities, RA has antiallergenic, antiviral, and antibacterial properties and displays a very low
toxicity (Petersen and Simmonds, 2003) and (Furtado et al., 2008). To fulfill the consumer
demand for natural food additives, rosemary extracts have been recently accepted by the EU
food additive legislation as effective and natural alternatives to synthetic antioxidants.
However, the effectiveness of these natural antioxidants depends on the preservation or
improvement of their stability, bioactivity and bioavailability (Fang and Bhandari, 2010).
Nano-encapsulation within food-grade macromolecules represents a remarkable mean to
maintain the structural integrity and potentially enhance the bioavailability of these bioactive
compounds (Munin and Edwards-Lévy, 2011). Cyclodextrins, a group of naturally occurring
macrocyclic oligosaccharides, are convenient encapsulating material widely used in the food
industry as additives for the stabilization of flavors and for the elimination of undesired tastes
(Astray et al., 2009). In particular, β-cyclodextrin (CD) has been listed as a safe food additive
since 1998 (Szente and Szejtli, 2004). Inclusion of natural phenols into CD enables their
protection against enzymatic oxidation and non-enzymatic oxidation by dioxygen and thus
can extend their stability over time (Cravotto et al., 2006).
Numerous analytical methods, such as spectroscopic, electrochemical or separation
techniques, have been used for the characterization of the inclusion complexes and the
estimation of the thermodynamic parameters of binding (Mura, 2014). Among these methods,
1H NMR is the most widely used (Pessine et al., 2012).

4
In aqueous solution, CDs are able to form inclusion complexes with a variety of phenolic
compounds such as hydroxytyrosol (López-García et al., 2010), baicalin (Li et al., 2009),
isoquercitrin (Wang et al., 2009), quercetin (Koontz et al., 2009), and caffeic acid (Zhang et
al., 2009). This ability is due to their cone-shaped structure with a relatively lipophilic inner
cavity and hydrophilic outer surface. The main driving force in complex formation is the
release of high-enthalpy water molecules from the CD cavity and the development of strong
van der Waals host (CD) – guest (ligand) interactions (Loftsson et al., 2005). Water molecules
are displaced by the more hydrophobic guest molecules with a concomitant decrease of CD
ring strain (Szejtli, 1998).
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
However, only few papers have been published on the inclusion of RA into CDs. Celik et al.
have studied the CD - RA binding and the corresponding changes in antioxidant capacity of
RA in solution using UV-visible and fluorescence spectroscopies (Celik et al. 2011).
Recently, Medronho et al. reported for the first time an NMR study of the β-CD - RA
interactions, giving insight on the complex structure (Medronho et al., 2014) ; however, much
higher values (by one order of magnitude) of the association constant were estimated in this
work.
These conflicting results prompted us to revisit CD – RA binding. As RA could be
encapsulated in CD for application in foods of variable acidity, we also investigated the pH-
dependence of the CD – RA binding to compare the affinity of the carboxylic and carboxylate
forms of RA for the macrocycle. In this work, the CD - RA inclusion complex (figure 1) was
investigated by 1D and 2D (ROESY) 1H NMR in order to refine the structural
characterisation of the complex. Additionally, a complementary study by CE was performed.
The values of the binding constant (Kb) obtained by independent analytical tools (fluorescence
spectroscopy, NMR and CE) in different studies (Celik et al. 2011, Medronho et al. 2014 and
this work) will be discussed.

5
The experimental conditions for the formation of the inclusion complex are consistent with
the EU food additive legislation.
94
95
96
97
2. Materials and methods
98
2.1 Materials
99
Rosmarinic acid, caffeic acid (CA), sodium dihydrogenphosphate dihydrate and disodium
hydrogenphosphate heptahydrate were purchased from Sigma Aldrich (St. Louis, MO, USA).
β-Cyclodextrin was kindly given by Roquette Freres (Lestrem, France). Sodium hydroxide
was purchased from Prolabo (BDH Prolabo, VWR International, Haasrode, Belgium). Phenol
was obtained from Carlo Erba (Carlo Erba Reagents, SdS, Peypin, France). D2O for NMR
analyses was purchased from Euriso-Top (Saint-Aubin, France). Demineralized water was
obtained from VWR (VWR International S.A.S, France). Standard solutions for capillary
conditioning, 0.1 M and 1.0 M NaOH solutions were supplied by Fluka Biochemika (Sigma-
Aldrich Chemie, GmbH, Steinheim, Germany) while milliQ water was produced by an EASY
pure RF compact ultrapure water system (Barnstead, ThermoFicher Scientific, Waltham, MA,
USA). All reagents were of analytical grade quality.
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
2.2. NMR study
All RA and CD solutions were freshly prepared in D2O. For the Job’s plot, different volumes
of 10 mM solutions of RA and CD in D2O were mixed together to a constant volume keeping
the sum of the total RA and CD concentrations equal to 10 mM.
For the determination of the apparent association constant (Scott’s plot), two different series
of samples were prepared.
In the first one, different volumes of a 1 mM CD solution in D2O and of a solution containing
RA (10 mM) and CD (1 mM) in D2O were mixed together to a constant volume with a final
RA/CD ratio ranging from 0.8 to 9.2.

6
In the second series, different volumes of a 1 mM RA solution and of a solution containing
CD (10 mM) and RA (1 mM) in D2O were mixed together to a constant volume with a final
CD/RA ratio ranging from 0.8 to 9.2. Final solutions were equilibrated at room temperature
and protected from light before measurement.
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
1D 1H NMR spectra were recorded on a Bruker AC-400 MHz spectrometer (software Bruker
Top Spin 2.1). 2D-ROESY spectra were recorded in on a Bruker AVL 600 MHz
(Spectropole, Aix-Marseille University).
Chemical shifts are given in ppm (δ) and calculated using the internal reference of the HDO
signal at 4.79 ppm. In all cases, the complexation-induced chemical shift difference is defined
as the difference between the chemical shift of the free molecule to the chemical shift of
bound molecule, ∆δ = δfree – δcomplex.
2.3 Capillary Electrophoresis study
Electrophoresis experiments were performed using an automated capillary electrophoresis
system (Beckman P/ACE MDQ, Fullerton, CA). Fused-silica capillaries (50 µm i.d.× 60 cm,
50 cm to the detector) were used.
Phosphate buffer (pH 7) for CE analysis was prepared with sodium dihydrogenphosphate
dihydrate and disodium hydrogenphosphate heptahydrate in milliQ water (ionic strength = 10
mM) and stored at 4°C. Working buffers were prepared by diluting up to 15 mM of CD in this
phosphate buffer.
Stock solutions of RA and CA, both in final concentration equal to 5 mM, were prepared in
milliQ water and protected from light at 4°C. Sodium thiosulphate (1 mM) was added to
protect the stock solutions against oxidation. Sonication of the stock solutions in an ultrasonic
bath (5 min) was necessary to achieve complete solubility. The stock RA and CA solutions
were renewed every 15 days. Samples for analysis were prepared by dilution of these stock
solutions in milliQ water. After dilution, the concentrations were 0.05 mM for RA, 0.2 mM
for CA and 1 mM for phenol.

7
Prior to first use, new capillaries were conditioned by flushing at 20 psi with milliQ water (2
min), 0.1 M NaOH (10 min), 1.0 M NaOH (5 min), milliQ water (2 min), then with working
buffer for 15 min. In order to remove potentially adsorbed analytes, the capillary was rinsed
with milliQ water (2 min), 0.1 M NaOH (5 min), milliQ water (2 min) and working buffer (5
min) every three consecutive runs. Before each sample run, the capillary was flushed with 0.1
M NaOH (1 min) and working buffer (5 min) and, after analysis, with working buffer (2 min).
The capillary was thermostated at 20 °C. The sample was injected in hydrodynamic mode at
0.5 psi for 5 s and each analysis was performed under an applied voltage of 10 kV. Any
measurement was repeated at least three times in identical conditions.
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
3. Results and Discussion
3.1. Stoichiometry of the inclusion complex
The stoichiometry of inclusion complex between RA and CD was determined by using the
method of continuous variations (Job's plot).
The CD protons that are most sensitive to bound RA are H-3 and H-5, as both points toward
the interior of the cavity. The significant diamagnetic shift observed for those protons is
essentially due to the magnetic anisotropy effects of RA’s π-electrons in agreement with the
formation of an inclusion complex (Schneider et al, 1998; Ali and Upadhyay, 2008). Table S1
shows the chemical shifts differences of H-5 (CD) for a CD mole fraction (rCD) ranging from
0.1 to 0.9. It is also noteworthy that the aromatic protons H-2, H-6, H2’ and H-6’ of RA are
also diamagnetically shifted. By contrast the other CD protons (H-1, H-2, H-4 and H-6),
mostly localized outside the CD cavity, experience insignificant chemical shift variations. The
superposition of the NMR spectra shows a significant shielding of the internal protons H-3
and H-5 of CD (figure 2). The stoichiometry of the inclusion complex was determined by
plotting rCD x ∆δ (H-3, CD) against rCD by using the Job's method (figure S1). The symmetry
of the curve obtained and its maximum at rCD = 0.5 both point to a single complex of 1:1

8
stoichiometry. The same conclusion was reached with H-5. These results are in agreement
with previous investigations by fluorescence spectroscopy (Çelik et al., 2011) and NMR
(Medronho et al., 2014).
171
172
173
174
175
176
3.2. Stability of the inclusion complex
The apparent binding constant Kb was calculated according to the conventional Scott's
equation assuming 1:1 binding.
ꢀꢁꢂꢃ
ꢀꢁꢂꢃ
ꢋ
177
=
+
(1)
∆ꢄ
∆ꢄ
ꢈꢉꢊ
ꢌ ∆ꢄ
ꢆ
ꢈꢉꢊ
ꢅꢆꢇ
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
∆δobs represents the observed chemical shift difference of CD proton H-3 or H-5 between free
CD and the CD + RA mixtures. ∆δmax is the chemical shift difference at saturation. The
[RA]/∆δobs ratio for H-5 was plotted as a function of [RA], thus resulting in an excellent linear
fit (figure 3). This confirms the 1:1 stoichiometry of the inclusion complex in agreement with
the Job's plot. From the slope of the plot, one obtains: ∆δmax = 0.22 0.01 ppm. The same plot
with H-3 yields: ∆δmax = 0.10 0.01 pm. The slope-to-intercept ratio equals the binding
constant. From the H-5 and H-3 plots respectively, one obtains: Kb = 445 53 M-1 and 465
81 M-1. As expected, both values are identical within experimental error.
RA displays two phenolic rings and each of them may be involved in the binding. A second
series of measurements, in which RA is in low and constant concentration (1 mM) and CD in
variable concentration, was performed to collect information on the structure of the complex.
Although the CD-induced chemical shift displacements of the RA protons are weak, Scott’s
plots could be constructed to estimate ∆δc and Kb. The Kb values thus obtained are slight
lower than the one obtained by monitoring the CD protons (Table S2). The latter are
considered more reliable based on the higher sensitivity of the CD protons (especially H-5) to
RA-CD binding, which is obvious from the larger ∆δmax values.
3.3. Structure of the inclusion complex
The assignments of RA protons were made on the basis of their specific coupling constants
and on the COSY spectrum. The complete 1H-NMR data for each of RA protons is presented

9
in the Table S3 and is consistent with those previously obtained in the literature (Lecomte et
al, 2010). Analysis of the NMR spectrum of RA (1 mM) shows signals of six aromatic
protons (H-2, H-5, H-6 and H-2', H-5', H-6'), two vinylic protons (H-7, H-8) and three
aliphatic protons (H7’a, H7’b and H-8').
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
1
In figure S2, the H NMR data show the large variations induced by RA on the chemical
shifts of the CD protons located inside the cavity (H-3, H-5), compared with the weak
variations of the CD protons located outside (H-1, H-2, H-4). Upon complex formation, the
large shielding of H-3 and H-5 reflects the presence of one of the RA aromatic rings in the
CD cavity.
1
In figure S3, the H NMR data of free and bound RA are compared. Maximal CD-induced
deshielding occurs for aromatic protons.
The geometry of the RA-CD complex was further investigated via a ROESY experiment.
Two different mixing times were used to obtain the NOE correlations between RA (350 ms)
and CD (150 ms) (figure 4). As expected from the 1D NMR spectra, strong correlations were
observed between the CD H-3 and H-5 protons on the one hand and most of the RA protons
on the other hand, especially H-2' and H-5' but also H-2. No correlation was observed
between the RA protons and the protons of the CD outer surface (H-2, H-4). These results
confirm without ambiguity the encapsulation of RA inside the CD cavity. The correlations
between the CD protons and all RA aromatic protons suggest that both phenolic moieties can
interact with the CD cavity. It can thus be assumed that two inclusion complexes are formed,
one involving the caffeoyl moiety (complex 1, binding constant K1) and the other one
involving the 3,4-dihydroxyphenyllactic moiety (complex 2, binding constant K2), both being
in fast equilibrium via free RA. Overall, only a single averaged NMR spectrum is observed
for the three species.

10
The observed chemical shift of any RA proton can be expressed as: δobs = x0δ0 + x1δ1 + x2δ2,
δ0, δ1 and δ2 being the chemical shifts of the proton in free RA, complex 1 and complex 2,
respectively, and x0, x1 and x2 the corresponding mole fractions.
221
222
223
224
Simple solution chemistry gives:
δ0 +
(
δ1K1 +δ2K2
)
[
CD
]
δ0 +δmax Kb
[
CD
]
225
226
(2)
δobs
=
=
[
]
[
]
1+ (K1 + K2 ) CD
1+ Kb CD
δ1K1 +δ2K2
K1 + K2
With K = K + K and
δmax
=
b
1
2
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
Therefore, whatever the proton signal detected, the apparent binding constant derived from
the chemical shift variations is the same and equals the sum of the individual binding
constants. It can thus be assumed that differences in Kb values depending on the RA proton
detected merely reflect differences in sensivity.
However, it can be noted that the NOE correlations are especially intense between with the
CD H-3 and H-5 protons and the vinylic protons of the caffeoyl part, which suggests a deep
inclusion of this moiety into the cavity.
3.4. pH dependence of binding
The pH-dependence of the CD-RA complexation was investigated. At pH 2.9, RA is a
-
mixture of neutral carboxylic (RA-CO2H) and anionic carboxylate (RA-CO2 ) forms in nearly
equal proportions. When the pH was decreased to 1 (pure RA-CO2H), the Kb values deduced
from Scott’s plots for the RA protons (variable CD concentration) were significantly higher.
-
By contrast, when the pH was increased to 6 (pure RA-CO2 ), the Kb values were lower.
Clearly, the less hydrophilic carboxyl form display a higher affinity for the CD cavity than the
corresponding carboxylate. On the other hand, at pH 1 and 6, the Job’s plots confirmed the
1:1 stoichiometry of the inclusion complex (data not shown). As shown in table 1, the Kb
values obtained at pH 6 (ca. 200 M-1) are of the same order of magnitude as those obtained by
Celik et al. (164 M-1) by fluorescence but much lower than those obtained by Medronho et al.
by NMR. Indeed, at pH 6, Medronho et al. have estimated Kb at ca. 1180 M-1 and 2030 M-1

11
depending on which RA aromatic ring is monitored and interpreted this difference by
assuming two complexes in solution, the first one involving the caffeoyl moiety and the
second the 3,4-dihydroxyphenyllactic moiety. As stated above, this interpretation is incorrect
as both complexes are indistinguishable by NMR due to fast chemical exchange. Hence,
differences in Kb values must be ascribed to differences in sensitivity depending on the
protons monitored and techniques adopted (detection on RA at variable CD concentration vs.
detection on CD at variable RA concentration).
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
3.5 Analysis of RA-CD inclusion complexes by capillary electrophoresis
To confirm the NMR data, the RA-CD inclusion complex was also studied by CE. The
electrophoretic mobility of RA was measured in buffered CD solutions at different
concentrations (Li and Waldran, 1999). The binding kinetics being fast with respect to the
separation time, the measured mobility ꢍ is the average mobility of the free (ꢍꢁꢂ) and bound
(ꢍꢁꢂꢎꢏꢐ) RA forms:
ꢑ
ꢓ
ꢍ = 1 − ꢒ ∙ ꢍꢁꢂ + ꢒ ∙ ꢍꢁꢂꢎꢏꢐ
(3)
ꢒ being the bound fraction of RA. For a large CD-to-RA molar ratio, the free CD
concentration can be considered constant and equal to the total CD concentrationꢔꢕ, so that
the binding constant can be expressed as:
ꢃ
ꢃ
ꢖꢗ = ꢀꢀ
=
(4)
ꢁꢂꢎꢏꢐ
ꢘ
263
ꢃꢀ
ꢁꢂ ꢏꢐ
ꢑ
ꢓ
ꢋꢎꢘ ∙ꢏ
264
265
Hence, the measured electrophoretic mobility can be expressed as a function of ꢕ:
ꢍ = ꢙ
(5)
ꢜꢌ ꢏ∙ꢙ
ꢚꢛ
ꢚꢛꢝꢞꢟ
ꢆ
ꢋꢜꢌ ꢏ
ꢆ
266
267
268
269
On top of binding effects, an increase in CD concentration leads to an increase in the buffer
viscosity (Paduano et al., 1990), and this in turn influences all the electrophoretic mobilities,
which are inversely proportional to the buffer viscosity (Plasson and Cottet 2005). The
electroosmotic flow ꢍꢠꢡꢢ being inversely proportional to the buffer viscosity too (Corradini
270
271
and Sprecacenere, 2003), it was used for evaluating the viscosity correction factor to be
applied to the measured electrophoretic mobilities (Li and Waldran, 1999). A linear

12
272
relationship between the elution time of a neutral marker (phenol) ꢣꢠꢡꢢ and the CD
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
concentration was observed (correlation coefficient R = 0.92). It corresponds to a linear
variation of the viscosity with a maximal variation of 8% for a CD concentration of 15 mM.
The ꢖꢗ values can be obtained from the non-linear curve-fitting of the plot expressing the
viscosity-corrected electrophoretic mobility ꢍ as a function of ꢕ according to eq. (5), thus
enabling the determination of the numerical values of ꢍꢁꢂ, ꢍꢁꢂꢎꢏꢐ and ꢖꢗ. The same method
was repeated with caffeic acid (CA) for comparison purposes, as CA is structurally related to
RA (see figure 1). The measurements were performed at 20 °C. The pH was fixed at 7 so as to
ensure that the carboxyl groups of RA and CA are fully deprotonated. The ionic strength was
fixed to a low value of 10 mM in order to avoid any heating of the capillary. Moreover, in
those conditions, the ionic and actual electrophoretic mobilities can be taken equal (Plasson
and Cottet, 2005).
The non-linear curve fitting yields: ꢖꢗ= 197 ( 14) M-1, ꢍꢁꢂ = 1.520 ( 0.006) x 10-8 m2V-1s-1
and ꢍꢁꢂꢎꢏꢐ = 0.934 ( 0.014) x 10-8 m2V-1s-1 for RA, ꢖꢗ= 176 ( 4) M-1, ꢍꢏꢂ = 2.101 (
0.003) x 10-8 m2V-1s-1 and ꢍꢏꢂꢎꢏꢐ = 1.047 ( 0.014)x10-8 m2V-1s-1 for CA (see figure 5).
The similar Kb values for RA and CA suggest that the caffeoyl moiety of RA is mostly
responsible for the affinity of RA for CD and that the less polarizable dihydroxyphenyllactic
moiety experiences a weaker binding. Taking hydroxytyrosol as a structural analog of the
dihydroxyphenyllactic moiety, it can actually be noted that this olive phenol only weakly
binds CD with a Kb value of ca. 90 M-1 (López-García et al., 2010).
The electrophoretic mobilities can also be used to evaluate the hydrodynamic radius of each
species (Plasson and Cottet, 2005):
ꢦ
ꢤ =
ꢥ
(6)
ꢧꢨꢩꢙ
ꢪ
295
296
297
with q the electric charge of the compounds (1.6x10-19 C) and η the water viscosity at 20 °C
(10-3 Pa.s). This gives rRA = 5.6 Å, rCA = 4.0 Å, rRA-CD = 9.1 Å, rCA-CD = 8.1 Å. These values can
be compared with the hydrodynamic value of β-CD: rCD = 7.7 Å (Pavlov et al. 2010). They

13
are consistent with the inclusion of either RA or CA inside the β-CD cavity, as evidenced by a
slight radius increase from free β-CD to the CA- β-CD and RA- β-CD complexes.
Close Kb values are thus obtained for RA- β-CD with this method and with the NMR method
in neutral conditions (about 200 M-1, see Table 1) with a similar precision (about 10% error).
This value is also in good agreement with the one determined by Celik et al. by fluorescence
(164 65 M-1) rather than with the value obtained by NMR by Medrohno et al. (ca. 2000 M-
1). The consistency of EC methods with fluorescence methods was furthermore checked by
performing a similar study for the association of RA with methyl-β-CD. A Kb value of 372 (
20) M-1 was determined; once again, this value is in perfect agreement with the value
measured by Celik et al. (328 39 M-1).
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
4. Conclusion
The complementary use of 1D and 2D ROESY NMR and EC methods have led to consistent
results concerning the complex formation between RA and β-CD. In all conditions, a mixture
of 1:1 complexes in fast equilibrium was obtained, both catechol subunits of RA being
potentially inserted inside the CD hydrophobic cavity, while no interactions of RA with the
outside of the CD unit could be reliably detected. Typically, the apparent binding constant is
ca. 450 M-1 at pH 2.9 (equimolar mixture of the carboxyl and carboxylate forms),
corresponding to an encapsulation efficiency of ca. 50% when the two species are mixed in an
equimolar concentration of 5 mM each. The pH dependence of Kb shows that the carboxyl
form of RA displays a higher affinity for the macrocycle than the carboxylate form.
In conclusion, RA forms a relatively stable complex with β-CD, especially in acidic
conditions. By accommodating the catechol nuclei inside the CD cavity, the binding could
inhibit their interactions with redox-active metal traces, thereby providing higher stability for
food applications. The binding could also modulate the release of RA in the digestive tract as
a function of pH.

14
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
Acknowledgements
The authors thank Roquette Freres (Lestrem, France) for providing β-cyclodextrin.
Appendix. Supplementary data
Supplementary data associated with this article can be found in Appendix.
References
Ali, S. M., Upadhyay, S. K. (2008). Complexation studies of pioglitazone hydrochloride and
β-cyclodextrin: NMR (1H, ROESY) spectroscopic study in solution. Journal of Inclusion
Phenomena and Macrocyclic Chemistry, 62, 161-165.
Astray, G., Gonzalez-Barreiro, C., Mejuto, J.C., Rial-Otero, R., Simal-Gándara, J. (2009). A
review on the use of cyclodextrins in foods. Food Hydrocoll., 23, 1631-1640.
Çelik, S.E., Özyürek, M., Tufan, A.N., Güçlü, K., Apak, R. (2011). Spectroscopic study and
antioxidant properties of the inclusion complexes of rosmarinic acid with natural and
derivative cyclodextrins. Spectrochim. Acta. A. Mol. Biomol. Spectrosc., 78, 1615-1624.
Corradini, D., and Sprecacenere, L. (2003). Dependence of the Electroosmotic Flow in Bare
Fused-Silica Capillaries from pH, Ionic Strength and Composition of Electrolyte Solutions
Tailored for Protein Capillary Zone Electrophoresis. Chromatographia, 58, 587-596.
Cravotto, G., Binello, A., Baranelli, E., Carraro, P., Trotta, F. (2006). Cyclodextrins as Food
Additives and in Food Processing, Current Nutrition and Food Science, 2, 343-350.
Fang, Z., Bhandari, B. (2010). Encapsulation of polyphenols. Trends Food Sci. Technol., 21,
510-523.
Furtado, M.A., de Almeida, L.C.F., Furtado, R.A., Cunha, W.R., Tavares, D.C. (2008).
Antimutagenicity of rosmarinic acid in Swiss mice evaluated by the micronucleus assay.

15
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
Mutat. Res., 657, 150-154.
Koontz, J. L., Marcy, J. E., O’Keefe, S. F., Duncan, S. E. (2009). Cyclodextrin Inclusion
Complex Formation and Solid-State Characterization of the Natural Antioxidants α-
Tocopherol and Quercetin. J. Agric. Food Chem., 57, 1162-1171.
Lecomte, J., Giraldo López, L.J., Laguerre, M., Baréa, B., Villeneuve P. (2010). Synthesis,
Characterisation and Free Radical Scavening Properties of Rosmarinic Acid Fatty Esters.
Journal of the American Oil Chemists Society, 87, 615-620.
Li, J., and Waldron, K.C. (1999). Estimation of the pH-independent binding constants
of alanylphenylalanine and leucylphenylalanine stereoisomers with β-cyclodextrin in the
presence of urea. Electrophoresis, 20, 171-179.
Li, J., Zhang M., Chao, J., Shuang, S. (2009). Preparation and characterization of the
inclusion complex of Baicalin (BG) with β-CD and HP-β-CD in solution: An antioxidant
ability study. Spectrochimica Acta Part A, 73, 789-793.
Loftsson, T., Jarho, P., Másson, M., Järvinen, T. (2005). Cyclodextrins in drug delivery.
Expert Opin. Drug Deliv., 2, 335-351.
López-García, M.Á., López, Ó. Maya, I., Fernández-Bolaños, J.G. (2010). Complexation of
hydroxytyrosol with β-cyclodextrins. An efficient photoprotection. Tetrahedron, 66, 8006–
8011.
Munin, A., Edwards-Lévy, F. (2011). Encapsulation of Natural Polyphenolic Compounds.
Pharmaceutics 3, 793-829.
Mura, P., n.d. Analytical techniques for characterization of cyclodextrin complexes in
aqueous solution. J. Pharm. Biomed. Anal. doi:10.1016/j.jpba.2014.02.022.
Paduano, L., Sartorio, R., Vitagliano, V. and Costantino, L. (1990). Diffusion properties of
cyclodextrins in aqueous solution at 25 °C. J. Sol. Chem., 19, 31-39.
Pavlov, G.M., Korneeva, E.V., Smolina, N.A., and Schubert, U.S. (2010). Hydrodynamic
properties of cyclodextrin molecules in dilute solutions. Eur. Biophys. J., 39, 371-379.

16
Pessine, F.B., Calderini, A., Alexandrino G.L. (2012). a review: Cyclodextrin inclusion
complexes probed by NMR techniques, D.H. Kim (Ed.), Magnetic resonance spectroscopy,
InTech Publishing.
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
Petersen, M., Simmonds, M.S. (2003). Rosmarinic acid. Phytochemistry, 62, 121-125.
Plasson, R., and Cottet, H. (2005). Determination of homopolypeptide conformational
changes by the modeling of electrophoretic mobilities. Anal. Chem., 77, 6047-6054.
Rocio Campos-Vega, Guadalupe Loarca-Piña, B. Dave Oomah, (2010). Minor components of
pulses and their potential impact on human health. Food Research International, 43, 461–482.
Schneider, H., Hacket, F., and Rudiger, V. (1998). NMR Studies of Cyclodextrins and
Cyclodextrin Complexes. Chem. Rev., 98, 1755-1785.
Szejtli, J. (1998). Introduction and general overview of cyclodextrin chemistry. Chem Rev, 98,
1743-1753.
Szente, L., Szejtli, J., 2004. Cyclodextrins as food ingredients. Trends Food Sci. Technol., 15,
137-142.
Wang, Y., Qiao, X., Li, W., Zhou, Y., Jiao, Y., Yang, C., Dong, C., Inoue, Y., Shuang, S.
(2009). Study on the complexation of isoquercitrin with β-cyclodextrin and its derivatives by
spectroscopy. Analytica Chimica Acta, 650, 118–123.
Zhang, M., Li, J., Zhang, L., Chao, J. (2009). Preparation and spectral investigation of
inclusion complex of caffeic acid with hydroxypropyl-β-cyclodextrin. Spectrochimica Acta
Part A, 71, 1891–1895.

17
1
397
398
399
400
Table 1. Maximal CD-induced H-NMR chemical shift displacements (∆δmax = δmax – δ0 (no
CD)) of RA protons and binding constant (Kb) at variable pH values in D2O (RA
concentration = 1 mM).
pH 1
pH 2.9
pH 6
RA
∆δmax
Kb
∆δmax
Kb
∆δmax
Kb
proton
(ppm)
(M-1)
(ppm)
(M-1)
(ppm)
(M-1)
H-2
H-2'
H-5
H-5'
H-6
H-6'
H-7
H-7’
H-8
H-8’
0.08
0.08
0.09
0.03
0.07
0.09
0.03
0.05
0.11
0.11
320
352
314
328
300
330
342
468
466
466
0.01
0.06
0.02
0.03
0.09
0.06
0.02
0.03
0.09
0.08
265
316
260
325
260
328
313
390
393
299
0.14
0.012
0.4
209
319
256
284
202
238
222
230
227
227
0.03
0.13
0.02
0.03
0.04
0.07
0.07
401
402

Figure 1. Chemical structures of (a) β-cyclodextrin (CD) and (b) rosmarinic acid (RA).
H-1
(a)
H-3
H-2
H-4
H-5
H-6
5
6
(b)
8
8’
2
7
2’
5’
7’
6’
(c)

Figure 2. Expansion of the 1H NMR spectrum showing the displacement of H-5 and H-3 (CD)
for a CD mole fraction (rCD) ranging from 0.1 to 0.9 in D2O at pH 2.9. Different volumes of 10
mM solutions of RA and CD were mixed together to a constant volume keeping the sum of the
total RA and CD concentrations equal to 10 mM in D2O at pH 2.9.
H-3
H-5

Figure 3. Scott's plot for CD proton H-5 with CD concentration set at 1 mM and variable RA
concentration in D2O at pH 2.9.

Figure 4. ROESY spectrum of the 1:1 RA-CD complex (5 mM of each partner) in D2O at pH
2.9 (mixing time = 150 ms).
H-2’ H-5’
H-2
H-6
H-7
H-6’
H-8
H-5
RA

Figure 5. Variation of the electrophoretic mobility of RA (solid line and squares) and CA
(dotted line and circles) as a function of the total CD concentration, in phosphate buffer at pH
7, 20 °C, I = 10 mM, [RA] = 0.05 mM, [CA] =0.2 mM. Circles and squares are experimental
measurements for respectively CA-CD and RA-CD, each of them being repeated three times.
The lines were obtained by non-linear curve-fitting of experimental data using Eq. 5.

18
403
404
405
Highlights
•
•
•
•
•
Investigation of the RA-β-CD complex by 1H NMR and CE
Estimation of the stoichiometry and binding constant by 1H NMR
Structural analysis of the RA-β-CD complex by ROESY
pH dependence of the binding constant
406
407
408
409
Estimation of the binding constants of the RA-β-CD, RA-Me-β-CD and CA-β-CD
complexes by CE
410