Precision studies in supramolecular chemistry: A 1H NMR study of hydroxymethoxyacetophenone/β-cyclodextrin complexes

Article abstract of DOI:10.1002/mrc.2762

The association constants for the interactions of 2-hydroxy-4- methoxyacetophenone, 2-hydroxy-5-methoxyacetophenone, 2-hydroxy-6- methoxyacetophenone, 3-hydroxy-4-methoxyacetophenone and 4-hydroxy-3- methoxyacetophenone with β-cyclodextrin in water were measured by ¹H NMR and by isothermal titration calorimetry. Very good agreement was obtained between the different methods. The errors associated with the NMR method for measuring mM binding affinities were estimated to be 10-30%, and by isothermal titration calorimetry, 10-20%. Rotating frame nuclear Overhauser effect spectroscopy studies show that the solution phase host-guest complexes formed by β-cyclodextrin with these hydroxymethoxyacetophenone derivatives are not structurally well defined but that the hydroxymethoxyacetophenone derivatives are mostly associated with the narrow primary hydroxyl rim. Copyright

Full text of DOI:10.1002/mrc.2762

Research Article
Received: 15 January 2011
Revised: 11 March 2011
Accepted: 14 March 2011
Published online in Wiley Online Library: 6 May 2011
(wileyonlinelibrary.com) DOI 10.1002/mrc.2762
Precision studies in supramolecular
chemistry: a 1H NMR study of hydroxy-
methoxyacetophenone/β-cyclodextrin
complexes
Lee Fielding,a∗ Scott C. McKellarb and Alastair J. Florenceb
The association constants for the interactions of 2-hydroxy-4-methoxyacetophenone, 2-hydroxy-5-methoxyacetophenone,
2-hydroxy-6-methoxyacetophenone, 3-hydroxy-4-methoxyacetophenone and 4-hydroxy-3-methoxyacetophenone with β-
cyclodextrin in water were measured by 1H NMR and by isothermal titration calorimetry. Very good agreement was obtained
betweenthedifferentmethods. TheerrorsassociatedwiththeNMRmethodformeasuringmMbindingaffinitieswereestimated
to be 10–30%, and by isothermal titration calorimetry, 10–20%. Rotating frame nuclear Overhauser effect spectroscopy studies
show that the solution phase host-guest complexes formed by β-cyclodextrin with these hydroxymethoxyacetophenone
derivatives are not structurally well defined but that the hydroxymethoxyacetophenone derivatives are mostly associated with
c
the narrow primary hydroxyl rim. Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Keywords: NMR; 1H NMR; 13C NMR; cyclodextrin; hydroxymethoxyacetophenones; stability constants
Introduction
molecules have considerable freedom to reorient, by rotations
inside the cavity[6–8] and by rapid exchange into and out of the
cavity.[9–11] Indeed, such rapid fluxional behaviour is the essence
of the NMR experiment where the systems are almost invariably in
the fast exchange regime. These studies have also established that
for the commonly available α- and β-cyclodextrins and derivatised
cyclodextrins, many diverse kinds of small molecule guests can be
accommodated.Muchoftheliteratureonα-cyclodextrindescribes
interactions with small aromatics,[6,9,10,12] whereas the literature
fromβ-cyclodextrinisconcernedwithawiderrangeofcompounds
including saturated cyclics, fused ring and aliphatic architectures
and even steroids.[13–17] In general terms, anything larger than a
small aromatic ring is generally better accommodated in the larger
cavity of β-cyclodextrin.[18–21]
Much of the interest in cyclodextrin research has been fuelled
by commercial interests and many of the published reports of
NMR studies on small molecule–cyclodextrin interactions have
a focus on pharmaceuticals. This is because formulation with
cyclodextrin can be a way to increase the solubility or stability
or other relevant property of the drug.[22] See, e.g. reports on
the interaction of naproxen,[23,24] tenilsetam,[25] suronacrine,[26]
CI-844,[27] cocaine,[28] salbutamol,[29] nicardipine,[30] doxepin[31]
and midazolam[32] with α- or β-cyclodextrins. These accounts are
also classic illustrations of the NMR approach.
Cyclodextrins are cyclic oligosaccharides consisting of α-1,
4-linked-D-glucopyranose residues. α-Cyclodextrin consists of six
of the glucopyranase residues and β-cyclodextrin 1 consists of
seven of the α-1,4-linked-D-glucopyranose residues. The molecule
has been described as being doughnut shaped or more precisely
as a hollow truncated cone. The internal cavity is relatively
hydrophobic and binds compatibly sized hydrophobic molecules.
Cyclodextrins are a well-characterised class of compounds.[1]
A useful and widely used graphical representation of the
β-cyclodextrin molecule is that of the truncated cone shown
in Scheme 1. The wider rim is composed of the secondary hydroxyl
groups and is ringed on the inside with seven H-3 protons. The
smaller rim is composed of the primary hydroxyl groups and is
ringed on the inside by H-6 and H-6ꢁ. The seven H-5 protons
form a ring around the inner hydrophobic core of the cyclodextrin
molecule. Cyclodextrin hydrogens H-2 and H-4 are arranged as
a band around the outside of the molecule with H-2 near the
wide rim and H-4 near to the narrow rim. The internal space of
β-cyclodextrin is variably described as having has a diameter of
around 6.5–7.5 Å and a depth of around 8 Å. The space inside
the α-cyclodextrin molecule is narrower, around 5.5–6 Å, but of
the same depth. These dimensions are approximately the same as
those of a benzene molecule.[2]
There is a wealth of binding data on cyclodextrins.[3] Much has
been learned about the solution phase chemistry of cyclodextrin
host–guest inclusion complexes from NMR studies. Induced H
NMR shifts of the internal protons, H-3 and H-5, are the hallmark
of internal binding.[4,5] Such reports have demonstrated clearly
that although a guest molecule may be partitioned into the
cyclodextrin host with sometimes remarkable binding affinities,
it is not, even with high binding affinities, held rigidly. The guest
∗
1
Correspondence to: Lee Fielding, Department of Medicinal Chemistry, MSD,
Newhouse ML1 5SH, UK. E-mail: lee.fielding@btinternet.com
a
Department of Medicinal Chemistry, MSD, Newhouse ML1 5SH, UK
b
Strathclyde Institute of Pharmacy and Biomedical Sciences, University of
Strathclyde, Glasgow G4 0NR, UK
c
Magn. Reson. Chem. 2011, 49, 405–412
Copyright ꢀ 2011 John Wiley & Sons, Ltd.

L. Fielding, S. C. McKellar and A. J. Florence
OH
OH
HO
O
HO
OH
O
Hydrophobic cavity
O
O
H
O
3
H
O
OH
OH
O
2
H
OH
O
H
OH
1
H
HO
OH
5
O
4
HO
OH
H
H
OH
HO
6
O
OH
O
(
H
H
HO
OH
nO
O
O
H
O
)
OH
3
5
HO
OH
4
6
HO
1
2
n = 1, α-CD
n = 2, β-CD
n = 3, γ-CD
Hydrophilic exterior
HO
O
OH
Scheme 1. Structure of β-cyclodextrin and diagrams that illustrate the location of hydrogens in the doughnut model.
The ready availability of a series of five different isomers Discussion
of hydroxymethoxyacetophenone offered the opportunity to
The integrity of the hydroxymethoxyacetophenone isomers was
explore the effect of guest shape on binding affinity and binding
geometry, under conditions where all other guest molecule
considerations are identical. Many of the previous reports of
inclusionphenomenaareweightedtowardsstudiesonrotationally
symmetric molecules such as mono-substituted benzenes and 1,
4-disubstituted species. In these cases, the model for host–guest
binding is one where one end or the other of the rod-like molecule
enters the cyclodextrin cavity like an axle threading a wheel.[11,13]
In this study, changes in the positions of the hydroxyl and methoxy
positions produce more diversity to the possible binding modes.
Measurements on a series such as this also offer an opportunity
to consider the sensitivity of the NMR titration technique. Many
published estimates of binding affinity for these systems, which
can cover six to seven orders of magnitude, are often quoted with
no consideration of the errors associated with the measurement.
It was thought that consideration of the variance between the
association constant, Ka, for different isomers coupled with the
variance between replicate measurements would give a handle on
precision.MeasuringtheKasindependentlybythecomplementary
technique, isothermal titration calorimetry (ITC) will provide a
handle to assess accuracy.
Here, we report the results of a study on the binding
of the regioisomers, 2-hydroxy-4-methoxyacetophenone (pae-
onol; 2–4), 2-hydroxy-5-methoxyacetophenone (2–5), 2-hy-
droxy-6-methoxyacetophenone (2–6), 3-hydroxy-4-methoxy-
acetophenone (3–4) and 4-hydroxy-3-methoxyacetophenone,
(acetovanillone; 4–3), Scheme 2, to β-cyclodextrin. The Kas were
measured by the NMR titration method and independently by
ITC. Although a tandem NMR and ITC study is not novel, inde-
pendent evaluation of the association constant with such good
agreement between the two methods is rare. An assessment of
the disposition of the solution state-bound complex was made
from rotating frame nuclear Overhauser effect spectroscopy data.
X-ray crystallography studies of these systems are ongoing and
will be reported elsewhere.
checked by 1H and 13C NMR in DMSO–d6 solutions. All the
materials were confirmed to be >98% pure. The assignments,
based on COSY, HSQC and NOESY data, are reported in Table 1.
These data are in line with expectations for this class of molecules.
The hydroxyl proton shifts in 2–4, 2–5 and 2–6 are approximately
2 ppm downfield compared with 3–4 and 4–3. This is indicative
of some intramolecular hydrogen bonding in the 1-acetoxy-2-
hydroxy moiety. The same phenomena is also reflected in the 13C
data where the carbonyl carbon is at 203–204 ppm in 2–4, 2–5
and 2–6 compared with 196 ppm in 3–4 and 4–3.
To minimise confusion, in the discussion that follows
β-cyclodextrin protons are labelled with a prime (e.g. H-3ꢁ) to
distinguish them from similar labels for hydroxymethoxyace-
tophenone protons. The stoichiometry of binding is assumed
to be 1 : 1 for all the complexes. This has been shown to be the
case for the majority of small molecule/cyclodextrin host–guest
complexes. The 1 : 2 stoichiometries where one guest molecule
is shared between two cyclodextrins are sometimes observed
for appropriately shaped guest molecules and at high cyclodex-
trin/guest ratios. This model can be excluded because the binding
experiments for this study were performed at the opposite condi-
tions (high guest/cyclodextrin ratios). The 2 : 1 stoichiometry can
be discounted on steric grounds – see the discussion of the ROESY
data.
NMR binding data
Binding of small aromatic molecules by cyclodextrins is signalled
by an upfield shift of the internal cyclodextrin protons, H-3ꢁ
and H-5ꢁ. Usually, little significant effect is seen on the other
cyclodextrin protons. This selective effect on only the internal
cyclodextrin protons is generally taken as good evidence for
an interaction where the small guest molecule is internalised
in the hydrophobic core of the cyclodextrin. A graphical plot
of the extent of this shift versus the solution composition at
different cyclodextrin/guest ratios is an isotherm from which Ka
can be derived. The anisotropic shielding region induced by
the ring current of aromatic molecules is effective at perturbing
the cyclodextrin shifts, so the optimum experiment protocol
There is medicinal interest in 2–4 and 4–3, which occur
naturally and are ingredients in traditional Chinese herbal
medicines. They have also been shown to have anti-inflammatory
properties.[33,34]
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2011, 49, 405–412

Hydroxymethoxyacetophenone/β-cyclodextrin complexes
O
O
O
O
O
OH
OH
O
OH
O
OH
O
O
O
OH
4-3
2-4
2-5
2-6
3-4
2-hydroxy-4-methoxyacetophenone, (paeonol), 2-4
2-hydroxy-5-methoxyacetophenone, 2-5
2-hydroxy-6-methoxyacetophenone, 2-6
3-hydroxy-4-methoxyacetophenone, 3-4
4-hydroxy-3-methoxyacetophenone, (acetovanillone), 4-3
Scheme 2. Structures of the hydroxymethoxyacetophenones.
Table 1. 1H and 13C assignments of hydroxymethoxyacetophenones
a
recorded from 50 mM solutions in DMSO–d6
4–3
3–4
2–4
2–5
2–6
1H
H-2
7.43
7.34
H-3
6.47
6.91
6.50
H-4
7.17
7.33
6.56
H-5
6.86
7.50
2.49
3.82
7.01
7.47
2.47
3.84
9.4
6.53
7.84
2.57
3.82
H-6
7.32
2.65
3.76
CH3CO
CH3O
OH
2.52
3.82
10.0
12.6
11.5
11.8
3J (Hz)
4J (Hz)
8.3
b
8.4
8.9
2.5
9.0
3.1
8.3
0.7
2.1
13C
Figure 1. Stack plot of 1H NMR data showing the influence of3–4 on the
chemical shifts of β-cyclodextrin protons. (A) 1.23 mM 3–4, (B) 2.45 mM
3–4, (C) 6.13 mM 3–4. β-cyclodextrin = 0.0126 mM throughout.
C-1
C-2
C-3
C-4
C-5
C-6
128.8
111.0
147.4
151.6
114.8
123.4
196.0
55.5
130.0
113.7
164.0
100.7
165.7
107.2
133.3
203.1
55.7
120.2
154.8
118.5
123.7
151.5
113.6
203.8
55.6
114.0
159.4
109.3
134.1
102.1
159.8
203.7
55.9
114.3
146.2
152.0
111.1
121.3
196.4
55.6
A result from a typical binding titration is shown in Fig. 1.
This figure shows how the cyclodextrin protons, H-3ꢁ and H-5ꢁ,
shift upfield in the presence of the acetophenone derivative.
The movement of H-3ꢁ is obvious, and also H-6/6ꢁ, although this
smaller shift was not considered for the Ka data treatment. The
large effect on H-5ꢁ is most apparent. The double doublet and
triplet from H-2ꢁ and H-4ꢁ are almost unperturbed. The large signal
at 3.8 ppm is water. The singlet at 3.62 ppm is an impurity present
in 3–4.
C
O
CH3O
CH3
26.2
26.3
26.6
28.0
32.9
a 1H chemical shifts are referenced to TMS (0 ppm). 13C shifts are
referenced to the residual DMSO-d5 signal (39.45 ppm).
b Not resolved.
The graphical representation of this data is shown in Fig. 2. The
attenuating response is a reflection of how the cyclodextrin bind-
ingcavitybecomessaturatedwithhydroxymethoxyacetophenone
at high(3–4), although this is a weak binding event and high mole
ratios of hydroxymethoxyacetophenone are required to approach
saturation. Note that the two solid lines drawn on this figure
correspond to the optimised binding parameters and are a three
parameter fit to 21 data points.
A titration that illustrates weaker binding is shown in Fig. 3.
The curve is less inclined to saturate and the endpoint which
corresponds to the chemical shift of 100% bound cyclodextrin is
less obvious, although this parameter is reliably located from the
computer fitting.
is to observe the cyclodextrin protons rather than the guest
protons (although the inverse protocol is sometimes used).
Sometimes, 13C NMR may be used to report on the binding
events, but in the case reported here, the limited solubility of the
hydroxymethoxyacetophenones restricts the study to solution
compositions where only 1H is a realistic observation. Some
preliminary experiments in the solvent DMSO–d6 showed that
the advantage offered by better solubility of the chemical system
was outweighed by some problems of poor dispersion of the
cyclodextrin signals and overlap of key signals with the residual
DMSO solvent peak. So this study was performed in nonbuffered
D2O throughout.
c
Magn. Reson. Chem. 2011, 49, 405–412
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

L. Fielding, S. C. McKellar and A. J. Florence
0.2
0.15
0.1
Table 2. Results of NMR titrations for the interaction of five
hydroxymethoxyacetophenones with β-cyclodextrin in D2O at 25 ◦Ca
H-3ꢁ ꢀδmax
(ppm)
H-5ꢁ ꢀδmax
(ppm)
Ka(M−1
)
H-3’
3–4
2–4
2–5
4–3
2–6
370 20
310 80
190 50
110 10
100 30
0.12 0.01
0.15 0.03
0.11 0.02
0.13 0.01
0.09 0.03
0.26 0.01
0.21 0.02
0.18 0.04
0.21 0.01
0.11 0.05
H-5’
H-3’ calc
H-5’ calc
0.05
0
a The results are the average and standard deviation of three (or four)
independent titrations.
0
200
400
600
[G]/[H]
they are all in the range 0.10–0.15 ppm upfield of uncomplexed
β-cyclodextrin, and that is consistent with an aromatic molecule
located inside the cavity. The first four isomers in the list can be
bracketed in the range as ꢀδmax = 0.13 0.02 ppm, i.e. essentially
the same for all.
Figure 2. Binding isotherm for the titration of 3–4 into β-cyclodextrin.
The chemical shift displacement ꢀδ is the difference between δ for the
relevant cyclodextrin proton in the presence of and in the absence of the
hydroxymethoxyacetophenone. The solution composition is expressed as
the ratio of [3–4] to [β-cyclodextrin] and the experiment is performed
at a fixed [β-cyclodextrin] = 0.0126 mM. The solid lines are the
computed curves for Ka = 373 M−1, H-3ꢁ ꢀδmax = 0.126 ppm and H-
5ꢁ ꢀδmax = 0.258 ppm.
The ꢀδmax values for H-5ꢁ are more diverse. They are in the
range 0.11–0.26 ppm, and they appear to point towards a rough
correlation with Ka (tighter binding equates to larger downfield
shifts). But there is no a priori reason to expect any correlation
between ꢀδmax and Ka. Perhaps these differences are genuinely
due to structural differences between the isomeric acetophenone
derivatives, perhaps they reflect different binding modes? What
is consistent and apparent is that ꢀδmax values for H-5ꢁ are al-
ways higher than for H-3ꢁ. This observation is less consistent
with a binding model in which hydroxymethoxyacetophenone
molecules are fully embedded inside the cyclodextrin molecule
and more consistent with a model where the hydroxymethoxy-
acetophenone is associated with the narrower opening to the
cavity.
0.1
H-3’
H-5’
0.05
H-3’ calc
H-5’ calc
Note that although the observed upfield (increased shielding)
shift of H-3ꢁ and H-5ꢁ is a general rule with β-cyclodextrin, a
reversal of the shift effect on H-5ꢁ is sometimes observed (i.e.
H-5ꢁ is shifted downfield in the presence of small aromatics)
when the smaller α-cyclodextrin is the host.[9] This has not been
satisfactorily explained. The observed small shifts of the external
protons (H-2ꢁ and H-4ꢁ) are at least an order of magnitude less
than those observed for the inner protons. Such shifts are not
without precedent and probably result from weak and nonspecific
interactions.
0
0
200
400
600
[G]/[H]
Figure 3. Binding isotherm for the titration of 4–3 into β-cyclodextrin
(0.0126 mM). The solid lines are the computed curves for Ka = 105 M−1
,
H-3ꢁ ꢀδmax = 0.131 ppm and H-5ꢁ ꢀδmax = 0.212 ppm.
Table 2 summarises the results of the NMR study on the
interaction of five isomeric acetophenone derivatives with
β-cyclodextrin. The range of measured Kas is well outside the
measurement errors that range from 10 to 30% and so the
NMR titration method has good precision. There are clear and
significant differences in Ka between most of the isomers. So, even
though the interactions are all weak, and cover a limited range of
association constants (100–400 M−1), an assessment of the impact
of regiochemistry on binding affinity is possible. 3–4 has clearly
the tightest interaction with β-cyclodextrin, 2–4 less so and 2–5
less so again. 4–3 and 2–6 have the weakest interaction and are
virtually indistinguishable from each other.
The limiting chemical shift (ꢀδmax) is a parameter that provides
a view of the bound state without the obfuscation of dynamic
averaging effects. It is essentially the chemical shift of the fully
complexed cyclodextrin and is therefore a source of structural
information on the 1 : 1 complex. The ꢀδmax values for H-3ꢁ are
slightlydifferentforthedifferentisomers, buttherearenogrounds
for interpreting at this level of detail. In rather general terms,
ROESY data
It is often found that the correlation times of cyclodextrin
complexes are such that near-zero correlation intensities are seen
from NOESY experiments, but the ROESY experiments work well.
However, for this series of systems, the observed ROESY cross peak
intensities are very low (<0.1%). This is partly due to the poor
water solubility of the hydroxymethoxyacetophenones, which
limits the solution concentrations. It is also due to the modest
binding affinities of the complexes. The solution compositions
were optimised to maximise the correlations, but nevertheless,
only about 10% of the hydroxymethoxyacetophenone that is
present in solution is in the form of a 1 : 1 host/guest complex.
Althoughthecrosspeakswereofgreaterintensitythanthethermal
noise, some of the cross peaks were of a similar magnitude to
those due to artefacts in the 2D data. Instead of increasing the
number of acquisitions, which produces diminishing returns in
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2011, 49, 405–412

Hydroxymethoxyacetophenone/β-cyclodextrin complexes
unspecified binding geometries. The preference for the narrow
rather than the wide rim may be due to a more complementary fit
that leaves less unoccupied space between the molecules.
This observation is in contrast to the many studies on mono-
substituted benzenes and 1,4-disubstituted benzenes, which are
forced to align along the cyclodextrin axis and thus have only one
or two possible binding modes. Interestingly, both specific and
nonspecific binding modes can be found in the rotationally sym-
metric systems. Some studies reveal a unique one-way threading
of the guest (usually readily explained on steric or electrostatic
grounds).[9,10] Some studies reveal a dynamic situation with both
up and down modes present.[11] More heavily substituted systems
(as here) are expected to require more elaborate models. Only
one of the isomeric acetophenone derivatives has a C2 rotational
axis (2–6), and therefore, for the other four examples, there is
no long axis and no obvious preferred direction of approach
to the β-cyclodextrin cavity. One would imagine that for all the
hydroxymethoxyacetophenones, the preferred approach would
be towards the wide rim. The observed ROESY data certainly
contradict this naive model. For cyclodextrin complexes, it is not
too unusual to have difficulty in rationalising the observed NOEs
in terms of a single orientation of the guest molecule. Previous
workers have invoked contributions from opposite arrangements
to explain such data.[13,15,16]
Figure 4. Expansion from the 400 MHz ROESY spectrum of a solution
containing 7.22 mM 3–4 and 1.02 mM β-cyclodextrin in D2O at 25 ◦C. The
1H spectrum of 3–4 is projected onto the y axis and the 1H spectrum of
β-cyclodextrin is shown projected onto the x axis.
Steric forces must be a strong influence on binding behaviour
but are exhibited in a complex way. Bergeron[10] has studied the
space requirements of a series of nitrophenols and concluded that
the steric demands do lead to preferred binding modes.
Table 3. ROESY data for hydroxymethoxyacetophenone/β-
cyclodextrin systemsa
H-3ꢁ
H-5ꢁ
H-6ꢁ
These results broadly support the findings of an earlier study
which incorporated some of the hydroxymethoxyacetophenones
that are reported here.[35] This previous study on the 2–4, 2–5
and 4–3 interacting with β-cyclodextrin, focussed on calorimetry
3–4
2–4
H-2 (0.12)
H-5 (0.06)
H-6 (0.13)
H-3 (0.02)
H-5 (0.04)
H-6 (0.02)
H-6 (0.03)
H-6 (0.05)
H-6 (0.03)
1
and included some limited H NMR data. This report concluded
(from a consideration of binding-induced chemical shifts) that
all three compounds were preferentially associated with the
narrow, primary hydroxyl side of the cavity. A more recent
study on the behaviour of 2–4, which includes ROESY data, also
concludesthatthisacetophenoneisinsertedintothenarrowrimof
β-cyclodextrin.[36]
2–5
4–3
2–6
H-6 (0.03)
Ac (0.04)
H-2 (0.05)
Ac (0.02)
–
Ac (0.02)
–
–
a The intensity of the cross peak (brackets) is the integral of a row from
the 2D spectrum with the diagonal peak set at −100%.
ITC data
Typical ITC titration data are shown in Fig. 5, and the thermody-
namic parameters resultant from the ITC study are listed in Table 4.
The measured Kas are in good agreement with those measured
by the NMR method. The ITC measurement errors range around
10–20%. So, within the error bars established by replicate deter-
minations, both methods are almost completely concurrent. This
is especially satisfying given the very different range of solution
compositions used by the two methods.
signal-to–noise ratio, we ran each ROESY experiment three or
four times. An example of one of the clearer ROESY results is
shown in Fig. 4. Note the intermolecular correlations from all three
aromatic protons of 3–4 to H-5ꢁ, and the correlation from only one
aromatic proton (H-6) to H-3ꢁ. A strong intramolecular correlation
is visible from H-5 to OCH3. The other strong correlations in the
3.8 ppm region, but displaced slightly ( 0.02 ppm) upfield or
downfield of the OCH3 frequency are artefacts. Only cross peaks
that were reproducible in several of the replicate ROESY data sets
are recognised as real and are listed in Table 3.
The observed ROESY cross peaks cannot be rationalised by any
specific binding model other than that of the general picture of
an embedded molecule that spends more time at the narrow
primary hydroxyl rim than at the wider secondary hydroxyl rim.
That is, there is apparently enough space to accommodate more
than one binding mode, and weak dynamic coupling within
the cyclodextrin cavity allows rapid re-orientation among several
The values obtained for standard Gibbs free energy (ꢀG◦),
enthalpy (ꢀH◦) and entropy (ꢀS◦) compliment those for Ka
and mirror the data trend. Negative enthalpies and negative
Gibbs energies confirm that the complexation reactions are
exothermic and spontaneous. These parameters can also be used
to gauge the strength of the hydroxymethoxyacetophenone/
β-cyclodextrin interaction, irrespective of the binding constant. A
stronger interaction is typified by a more negative ꢀH◦, wh◦ile a
‘more spontaneous’ interaction will have a more negative ꢀG .[37]
Complexation of organic molecules by cyclodextrins is believed
to be driven by both hydrophobic interactions, arising from the
penetration of the hydrophobic region of the guest molecule
c
Magn. Reson. Chem. 2011, 49, 405–412
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

L. Fielding, S. C. McKellar and A. J. Florence
Within the series presented here, ꢀH◦ and ꢀG◦ decrease
(become less negative) as Ka decreases. This indicates that van der
Waals forces become less pronounced in the complexes from 3–4
>2–4>2–5>4–3>2–6.Withthehigheraffinitycomplexes3–4
and 2–4, the hydroxymethoxyacetophenone will be more tightly
bound in the β-cyclodextrin cavity. Like interactions that give rise
to cross peaks in ROESY data, van der Waals forces are dependent
on intermolecular distance; thus, the ꢀH◦ values loosely reflect
the ROESY results, wherein the most well-defined cross peaks are
in the systems with highest (most negative) enthalpy, while cross
peaks are absent in the lowest enthalpy system (2–6).
Time (min)
0
20 40 60 80 100 120 140
0.00
-0.75
-1.50
Conversely, ꢀS◦ increases (becomes more positive) over the
seriesintheorder2–4>3–4>2–5>2–6>4–3.Thoughslightly
different from the order described previously, the general entropic
trend remains consistent with the previous thermodynamic
observations. As the enthalpic contribution to ꢀG◦ diminishes,
the entropic contribution increases. It would be presumptive
to designate any one system enthalpy-driven or entropy-driven,
not least because enthalpy is more substantial and outweighs
entropy in the calculation of ꢀG◦.[40] It is clear, however, that
the complexation reaction over the series 3–4, 2–4, 2–5, 4–3
and 2–6 is influenced less by van der Waals forces and more by
hydrophobic interactions.
-2.25
0.0
-0.1
-0.2
-0.3
-0.4
Data: SMCD34Phe_NDH
Model: One Sites
Chi^2/DoF = 6.794
N
K
1.00
389
0 Sites
4.73 M
-1
∆H -2864
21.11 cal/mol
∆S
2.24 cal/mol/deg
It is worth noting at this point the potential sources of error in
the ITC data. Though there is a ‘binding hierarchy’ within the set
of isomers, even 3–4, the highest Ka complex, is still a very low-
affinitysystem,asshowngraphicallyinFig. 5.Thebindingisotherm
is logarithmic in shape, rather than the classical sigmoidal curve
resultant of high-affinity systems. It is possible to accurately study
low-affinity complexes provided that an ample section of the
binding curve is used for analysis, that the concentrations of host
and guest are known accurately and that the signal-to-noise ratio
is satisfactory.[41] These requirements have been met in this study.
However, caution should still be exercised before putting too
much credence in the values obtained for ꢀH◦ and ꢀS◦. These
values serve as good indication of trends within a data set, but
slight inaccuracies in concentration, e.g. can affect these values
more so than the Ka value. This may account for the error on
some of the enthalpy and entropy values and also for the slight
difference in the ꢀS◦ series when compared with Ka and ꢀH◦. It
is also a requirement that the binding stoichiometry be known.[41]
The stoichiometry was assumed to be 1 : 1 and fixed during least-
squares curve fitting. However, therein lays one source of disparity
between this work and a previous study of 2–4, 4–3 and 2–5,[35]
which used a different ITC experiment and apparatus and found
the stoichiometry of the complex of 2–5 with β-cyclodextrin to
be a co-existing mixture of 1 : 1 and 2 : 1. Also, with the exception
of the enthalpy of the 1 : 1 2–5 β-cyclodextrin complex, there is
little agreement between any of the numerical thermodynamic
values. However, the complex stabilities based on ꢀG◦ values
were ordered 2–4 > 4–3 > 2–5, which is similar to the order
determined in this study (2–4 > 2–5 ≈ 4–3).
0
1
2
3
4
5
6
7
8
9
Molar Ratio
Figure 5. Typical ITC data for a titration of β-cyclodextrin into 3–4. The
upper panel shows the exothermic heat released by injecting a series of
10-µl aliquots of β-cyclodextrin solution (19 mM) into the solution of 3–4
(0.43 mM). The lower panel shows integrated heat data with a differential
binding curve fitted to a one-site binding model used to calculate the
values for the association constant (Ka), enthalpy (ꢀH◦) and entropy (ꢀS◦).
The stoichiometry (N) was fixed at 1 : 1.
Table 4. Thermodynamic parameters of hydroxymethoxyaceto-
phenone/β-cyclodextrin systems from ITC data, in water at 25 ◦Ca
Ka
ꢀH◦
TꢀS◦
ꢀG◦
(M−1
)
(kJ mol−1
)
(kJ mol−1
)
(kJ mol−1
)
3–4
2–4
2–5
4–3
2–6
400 30
230 30
130 20
120 20
90 10
−12.1 0.2
−11.1 0.8
−9.1 0.6
−7.3 0.7
−6.7 0.8
2.8 0.1
2.4 1.0
3.0 0.4
4.6 1.1
4.5 0.9
−14.8 0.2
−13.5 0.3
−12.1 0.3
−11.8 0.4
−11.2 0.2
a The results are the average and standard deviation of three (or four,
in the case of 2–4) independent titrations.
into the cyclodextrin cavity, de-solvation of the guest and the
release of water from the cyclodextrin cavity to the bulk water
and/or van der Waals forces.[38] The inclusion process is usually
a weighted combination of these two factors. Indeed, this is
reflected thermodynamically in the value of ꢀG◦ by a negative
enthalpy contribution from van der Waals and a positive entropy
contribution from hydrophobic interactions. Rationalised simply,
attractive van der Waals forces in a high-affinity complex would
yield a tightly bound guest molecule (large negative ꢀH◦) with
restricted degrees of freedom (low ꢀS◦), while the reverse would
be true of a complex characterised predominantly by hydrophobic
interactions.[39,40]
Experimental
The water content of β-cyclodextrin hydrate (C42H70O35 · xH2O,
Arcos Organics) was determined by difference from a quan-
titative NMR assay of the cyclodextrin content.[42,43] This
gave a composition of 86 1% β-cyclodextrin and 14% wa-
ter, which translates to C42H70O35· 10H2O and a molecu-
lar weight of 1320. 2-Hydroxy-4-methoxyacetophenone (2–4),
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2011, 49, 405–412

Hydroxymethoxyacetophenone/β-cyclodextrin complexes
2-hydroxy-5-methoxyacetophenone (2–5), 4-hydroxy-3-me-
thoxyacetophenone (4–3), 2-hydroxy-6-methoxyacetophenone
(2–6) and 3-hydroxy-4-methoxyacetophenone (3–4) were pur-
chased from Arcos Organics or Sigma-Aldrich.
(stirring at 307 rpm) at 5-min intervals into a 1.4334-ml sample cell
filled with the desired hydroxymethoxyacetophenone solution.
Titrations for each system were performed in triplicate, except for
2–4, which was performed in quadruplicate. The concentration of
β-cyclodextrin in the injection well was varied accurately between
experiments over the range 17–33 mM and the concentration of
2–4, 2–5, 4–3, 2–6 or 3–4 in the sample well was 0.4–1.3 mM.
Titrations of the same compounds at different concentrations
were carried out to test the robustness of the method and ensure
comparable results were obtained when a different molar ratio
of the same compounds was used. Control experiments were
performed wherein β-cyclodextrin solution was titrated into pure
distilled water using the same procedure and concentrations as
described above. Distilled water was also titrated into solutions of
2–4, 2–5, 4–3, 2–6 or 3–4 as described.
The poor solubility in water of hydroxymethoxyacetophenones
is the limiting factor that constrains the solution NMR experi-
mental conditions. All protocols were designed to optimise the
experimental hydroxymethoxyacetophenone concentrations.
All NMR data were acquired at 25 ◦C on a Bruker DRX400
equipped with a 5-mm DCH Cryoprobe operating at a 1H fre-
quency of 400.23 MHz. The 1D 1H data were obtained with a 30◦
flip angle (90◦ pulse = 9.7 µs), an acquisition time of 1.99 s and a
relaxation delay of 1 s, into a 32k data array, zero filled to 64 k and
process with a 0.3 Hz line broadening. Solutions for NMR titration
data were prepared by volumetric dilution and mixing of gravi-
metrically prepared solutions in D2O. The chemical shift reference
for the titrations was a trace amount of acetone (0.01%), which
generated a small peak (ca 2% of the height of analyte peaks)
at 2.10 ppm. The protocol was to obtain 1H NMR data (observed
species – β-cyclodextrin) at a fixed β-cyclodextrin concentration
and over a range of hydroxymethoxyacetophenone concentra-
tions extending to 400× or 1000× molar excess. The strategy
was to prepare solutions at the maximum achievable hydrox-
ymethoxyacetophenone concentration that would give a stable,
noncrystallising solution. In a typical experiment, stock solutions
of 0.1 mM β-cyclodextrin and 6 or 7 mM hydroxymethoxyace-
tophenone were mixed and diluted with D2O to give a series of
solutions where the β-cyclodextrin was fixed at a <0.02 mM and
the hydroxymethoxyacetophenone ranged from 0 to 6 or 7 mM
over 10 steps. These titrations were repeated three or four times
with newly prepared stock solutions so that at least three different
binding curves were produced for each isomeric acetophenone
derivative. A typical 3× replicate set contained 1H NMR data from
titrations at 0.005, 0.01 and 0.02 mM β-cyclodextrin. The chemical
shifts of cyclodextrin protons H-3ꢁ and H-5ꢁ were measured to
0.001 ppm.
The optimum ratio of β-cyclodextrin to hydroxymethoxyace-
tophenone for the ROESY experiments was determined to be
1 : 7. At this ratio, the intensities of the peaks in the 1D spectrum
due to β-cyclodextrin approximately equal those due to hydrox-
ymethoxyacetophenone, and in the 2D spectrum, intermolecular
cross peaks were optimised. Differentratios produced significantly
smaller responses. The solutions were made up at the highest pos-
sible stable hydroxymethoxyacetophenone concentrations. The
phase sensitive TPPI ROESY[44] data were acquired with a spin lock
power of 17 dB, a mixing time of 200 ms and a relaxation delay
of 2 s. Data were accumulated by co-adding 16 or 32 FIDs per F1
increment (256) and the total experiment time was around 5–6 h.
The F1 dimension was zero filled to 1024 and a QSINE window
function was applied to both dimensions. The cross peak signal
intensities were recorded by selecting a row from the 2D data and
setting the integral of the diagonal peak to −100.
Data analysis
The equilibrium between the isomeric acetophenone deriva-
tives and β-cyclodextrin is in the fast exchange regime on
the 1H NMR time scale. So, although the bound hydroxy-
methoxyacetophenone/β-cyclodextrin complex has different
NMR properties to either of the free hydroxymethoxyacetophe-
none or free β-cyclodextrin species, the observed 1H NMR
spectrum for any given solution is an averaged picture. Hence,
there can be no direct measurement of the solution composition.
The deconvolution of the data by means of assembling a titration
curve and computer fitting to a model of the binding condition is
well established.[45,46]
A 1 : 1 binding model is assumed and an Excel spreadsheet
was used to fit the data to the quadratic binding equation. Each
curve in Fig. 2 shares the same value of Ka but has a different
ꢀδmax. The computer fitting was configured to produce a three
parameter fit to the data (Ka and two independent values of
ꢀδmax). The fitting routine can accommodate data from as many
distinct protons as are available. All cyclodextrin 1H signals were
monitored during the titrations. Large effects were only seen
at H-3ꢁ and H-5ꢁ. The magnetically inequivalent geminal pair of
protons at H-6/6ꢁ produces a multiplet signal that was dependent
on hydroxymethoxyacetophenone concentration but was difficult
to monitor reliably so it was excluded from the data analysis. Small
shifts (<10% of that seen for H-5ꢁ) were observed at H-1ꢁ and H-4ꢁ
and no significant shifts were observed at H-2ꢁ.
For the determination of thermodynamic parameters from
ITC titrations, nonlinear least-squares curve fitting was carried
out in Origin version 7.0 using the noninteracting ‘one site of
binding sites’ model provided by MicroCal. Heats of dilution were
determined from the control experiments and were subtracted
from the titration data prior to curve fitting. The initial 1-µl
injection was removed from each titration curve to prevent any
artefacts from diffusion at the injection tip during experiment
equilibration. The stoichiometry parameter (N) was fixed at 1. The
resultant binding constant was used to calculate Gibbs free energy
according to the equation:
For ITC titrations, solutions of hydroxymethoxyacetophenone
and β-cyclodextrin were prepared by accurately weighing the
desired quantity of sample into a glass vial and dissolving in 5 ml
of distilled water (measured with a calibrated micropipette). For
particularlylowconcentrationsofhydroxymethoxyacetophenone,
dilutions were carried out as necessary to prevent error that may
occur while weighing very low sample masses. ITC was performed
on a previously calibrated MicroCal VP-ITC unit at 25 ◦C. Each
titration consisted of 26 successive 10-µl injections (with an initial
1-µl injection) of β-cyclodextrin solution from a 0.28-ml syringe
ꢀG◦ = −RT ln K
These values are in good agreement when ꢀG◦ is calcu-
lated from
ꢀG◦ = ꢀH◦ − TꢀS◦
using experimentally determined ꢀH◦ and ꢀS◦ values.
c
Magn. Reson. Chem. 2011, 49, 405–412
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

L. Fielding, S. C. McKellar and A. J. Florence
Conclusion
[19] M. V. Rekharsky, R. N. Goldberg, F. P. Schwarz, Y. B. Tewari, P. D.
Ross, Y. Yamashoji, Y. Inoue, J. Am. Chem. Soc. 1995, 117, 8830.
[20] P. M. Ivanov, D. Salvatierra, C. Jaime, J. Org. Chem. 1996, 61, 7012.
[21] A. Botsi, B. Perly, E. Hadjoudis, J. Chem. Soc. Perkin Trans. 1997, 2, 89.
[22] M. E. Davis, M. E. Brewster, NatureRev. DrugDiscovery 2004, 3, 1023.
[23] G. Bettinetti, F. Melani, P. Mura, R. Monnanni, F. Giordano, J. Pharm.
Sci. 1991, 80, 1162.
[24] A. Ganza-Gonzalez, J. L. Vila-Jato, S. Anguiano-Igea, F. J. Otero-
Espinar, J. Blanco-Mendez, Int. J. Pharmaceut. 1994, 106, 179.
[25] M. E. Amato, F. Djedaini-Pilard, B. Perly, G. Scarlata, J. Chem. Soc.
Perkin Trans. 1992, 2, 2065.
β-Cyclodextrin binds all five isomeric acetophenone derivatives at
its narrow, primary hydroxyl rim with association constants in the
range 100–400 M−1. The precision of the NMR and ITC titration
methods are around 10–30 and 10–20%, respectively, for
systems with mM binding affinities. Such precision is more than
adequate to detect the different binding interactions between
these compounds and β-cyclodextrin that arise from differences
in guest regiochemistry.
[26] M. E. Amato, G. C. Pappalardo, B. Perly, Magn. Reson. Chem. 1993,
31, 455.
[27] M. E. Amato, K. B. Lipkowitz, G. M. Lombardo, G. C. Pappalardo,
Magn. Reson. Chem. 1998, 36, 693.
[28] N. Nesnas, J. Lou, R. Breslow, Biorg. Med. Chem Lett. 2000, 10, 1931.
[29] E. Estrada, I. Perdomo-Lopez, J. J. Torres-Labandeira, J. Org. Chem.
2000, 65, 8510.
[30] C. M. Fernades, R. A. Carvalho, S. PereiradaCosta, F. J. B. Veiga,Euro.
J. Pharma. Sci. 2003, 18, 285.
[31] J. R. Cruz, B. A. Becker, K. F. Morris, C. K. Larive, Magn. Reson. Chem.
2008, 46, 838.
References
[1] J. Szejtli, Chem. Rev. 1998, 98, 1743.
[2] I. Gomez-Orellana, D. Hallen, Thermochim. Acta. 1993, 221, 183.
[3] K. A. Connors, Chem. Rev. 1997, 97, 1325.
[4] P. V. Demarco, A. L. Thakkar, Chem. Comms. 1970, 2.
[5] H.-J. Schneider, F. Hacket, V. Rudiger, H. Ikeda, Chem. Rev. 1998, 98,
1755.
[32] S. M. Ali, S. K. Upadhyay, Magn. Reson. Chem. 2008, 46, 676.
[33] T.-C. Chou, Brit. J. Pharmacol. 2003, 139, 1146.
[34] E. A. Peters, J. T. N. Hiltermann, J. Stolk, Free Radical Biol. Med. 2001,
31, 1442.
[35] D. Z. Sun, L. Li, X. M. Qiu, F. Liu, B. L. Yin, Int. J. Pharmaceut, 2006,
316, 7.
[36] Y. Tsai, H. H. Tsai, C. P. Wu, F. J. Tsai, Food Chem. 2010, 120, 837.
[37] K. Bouchemal, Drug Discovery Today 2008, 13, 960.
[38] M. V. Rekharsky, Y. Inoue, Chem. Rev. 1998, 98, 1875.
[39] M. V. Rekharsky, M. P. Mayhew, R. N. Goldberg, P. D. Ross,
Y. Yamashoji, Y. Inoue, J. Phys. Chem. B 1997, 101, 87.
[40] G. Castronuovo, M. Niccoli, Bioorg. Med. Chem. 2006, 14, 3883.
[41] W. B. Turnbull, A. H. Daranas, J. Am. Chem. Soc. 2003, 125, 14859.
[42] G. K. Pierens, A. R. Carroll, R. A. Davis, M. E. Palframan, R. J. Quinn,
J. Nat. Prod. 2008, 71, 810.
[6] J. P. Behr, J. M. Lehn, J. Am. Chem. Soc. 1976, 98, 1743.
[7] Y. Inoue, F.-H. Kuan, R. Chujo, Bull. Chem. Soc. Jpn. 1987, 60, 2539.
[8] S. Cahill, S. Bulusu, Magn. Reson. Chem. 1993, 31, 731.
[9] D. J. Wood, F. E. Hruska, W. Saenger, J. Am. Chem. Soc. 1977, 99,
1735.
[10] R. J. Bergeron, M. A. Channing, G. J. Gibeily, D. M. Pillor, J. Am. Chem.
Soc. 1977, 99, 5146.
[11] R. J. Bergeron, M. A. Channing, K. A. McGovern, J. Am. Chem. Soc.
1978, 100, 2878.
[12] S. Simova, H.-J. Schneider, J. Chem. Soc. Perkin Trans. 2000, 2, 1717.
[13] D. Salvatierra, D. Jaime, A. Virgili, F. Sanchez-Ferrando, J. Org. Chem.
1996, 61, 9578.
[14] P. Wallimann, T. Marti, A. Furer, F. Diederich, Chem. Rev. 1997, 97,
1567.
[15] D. Salvatierra, X. Sanchez-Ruiz, R. Garduno, E. Cervello, C. Jaime,
A. Virgili, F. Sanchez-Ferrando, Tetrahedron 2000, 56, 3035.
[16] M. Zubiaur, C. Jaime, J. Org. Chem. 2000, 65, 8139.
[17] E. Bednarek, W. Bocian, J. Poznanski, J. Sitkowski, N. Sadlej-
Sosnowska, L. Kozerski, J. Chem. Soc. Perkin Trans. 2002, 2, 999.
[18] A. Botsi, K. Yannakopoulou, B. Perly, E. Hadjoudis, J. Org. Chem.
1995, 60, 4017.
[43] U. Holzgrabe, Prog NMR Spectros. 2010, 57, 229.
[44] A. Bax, D. G. Davis, J. Magn. Reson. 1985, 63, 207.
[45] R. S. Macomber, J. Chem. Ed. 1992, 69, 375.
[46] L. Fielding, Tetrahedron 2000, 56, 6151.
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2011 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2011, 49, 405–412