An NMR study of cyclodextrin complexes of the steroidal neuromuscular blocker drug Rocuronium Bromide

Article abstract of DOI:10.1002/mrc.1008

The interaction of Rocuronium Bromide, anda model steroid Org 7402, with three cyclodextrins (β-cyclodextrin, γ-cyclodextrin and Org 25969) was studied by solution state NMR experiments. Stoichiometries and binding constants were determined from ¹

Full text of DOI:10.1002/mrc.1008

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2002; 40: 251–260
An NMR study of cyclodextrin complexes of the
steroidal neuromuscular blocker drug Rocuronium
Bromide
∗
Kenneth S. Cameron, Dan Fletcher and Lee Fielding
Akzo-Nobel Pharma Division, Organon Laboratories Ltd, Newhouse, Lanarkshire ML1 5SH, UK
Received 5 October 2001; Accepted 16 November 2001
The interaction of Rocuronium Bromide, and a model steroid Org 7402, with three cyclodextrins
(
b-cyclodextrin, g-cyclodextrin and Org 25969) was studied by solution state NMR experiments.
1
Stoichiometries and binding constants were determined from H chemical shift titrations. All of the
systems formed 1 : 1 complexes. Most of the complexes were in fast exchange with unbound species on the
NMR time scale, but the most tightly bound complex (Rocuronium Bromide–Org 25969) was in the slow
exchange regime. The geometry of the complexes was inferred from H and C NMR shift changes upon
complexation and from intramolecular NOE correlations. Rocuronium Bromide forms a weak complex with
1
13
3
−1
b-cyclodextrin (K
complexes with g-cyclodextrin (K
complexes with the steroid located inside the central void of the cyclodextrin. Copyright  2002 John Wiley
Sons, Ltd.
a
= 3.3 ± 0.5 × 10
M
) and no clear picture of the structure of the complex emerges. The
4
−1
5
−1
a
= 1.8 ± 0.2 × 10
M
) and Org 25969 (K
a
> 10
M
) are true inclusion
&
KEYWORDS: NMR; H NMR; 13C NMR; host–guest chemistry; steroids; cyclodextrins; anaesthetics
1
INTRODUCTION
anaesthetists to have an alternative course of action. One
such line that has been followed at these laboratories was
the development of neuromuscular block reversal agents by
means of sequestering the blocking drug. In the course of
Neuromuscular blocking drugs are used by anaesthetists
during surgical procedures to ease tracheal intubation and to
4
1
improve muscle tone. Many different drugs with different
this research cyclodextrins were considered as prospective
reversal agents because of their known ability to sequester
2
durations of action are available. Rocuronium Bromide (1)
is typical of the highly successful amino steroids class of
5
drugs by forming host–guest complexes.
3
neuromuscular blockers that are in general use worldwide.
Cyclodextrins are cyclic oligosaccharides composed of
a ring of ˛-1,4 linked D-glucosyl residues, and comprising
six (˛-cyclodextrin), seven (ˇ-cyclodextrin (2)) or eight (ꢀ-
cyclodextrin (3)) units. Tertiary structures resemble a hollow,
truncated cone with primary hydroxyl groups crowning the
narrow rim and secondary hydroxyl groups crowning the
wider rim. The low polarity central void is able to encapsulate
smaller molecules and is responsible for the great interest in
OAc
1
2
17
14
1
1
O
16
1
3
+
1
N
9
8
N
2
10
15
5
7
HO
−
6,7
3
Br
cyclodextrins in host–guest chemistry.
4
6
There are many reports of steroid–ˇ-cyclodextrin com-
5
8
1
plex formation, particularly with regard to drug delivery
9
–12
and molecular recognition,
: 1 complexes are typically in the range 10 to 105
and binding constants for
4
ꢀ1 9,11
.
1
M
However, it remains a problem that sometimes patients
are still under the effects of the neuromuscular block (and
are therefore paralysed) after the surgical procedures are
finished. Current practice is to administer an inhibitor
of acetylcholinesterase (e.g. Neostigmine) to reverse the
action of the blocking drug. It would be desirable for
Thus there is little doubt that steroids can at least partially
1
3–16
penetrate the core of ˇ-cyclodextrin.
However, there are
not always clear pictures of the nature of the complexes.
A demonstrated interaction of the steroid with the inter-
nal cyclodextrin protons H-3 or H-5 is usually taken as
proof of internalization. There is a larger literature describ-
ing the binding of bile salts to cyclodextrins, but the evidence
shows that for this class of compounds it is almost always
the alkyl chain (C-20 to C24) that enters the cyclodextrin
Ł
Correspondence to: L. Fielding, Akzo–Nobel Pharma Division,
Organon Laboratories Ltd, Newhouse, Lanarkshire ML1 5SH, UK.
E-mail: l.fielding@organon.nhe.akzonobel.nl
DOI: 10.1002/mrc.1008
Copyright  2002 John Wiley & Sons, Ltd.

252
K. S. Cameron, D. Fletcher and L. Fielding
OH
NaO C
2
8
O
7
OH
O
O
HO
CO Na
2
S
6
5
O
OH
NaO C
O
1
O
2
HO
4
O
S
O
HO
HO
O
O
OH
2
S
HO 3
OH
O
HO
OH
OH
HO
HO
O
OH
OH
HO
O
O
O
OH
S
CO Na
2
O
OH
HO
O
HO
O
S
OH
OH O
NaO C
OH
O OH
O
2
OH
HO
HO
HO
OH
OH
OH
HO
O
OH
O
O
S
O
S
O
O
HO
O
O
S
O
CO Na
2
NaO C
HO
2
2
CO Na
2
4
OH
6
CH OH
2
+
−
5
N
N
O
4
1
Br
HO
HO
O
3
2
5
OH
(n = 8)
n
3
EXPERIMENTAL
ˇ-Cyclodextrin [85,608-8] was purchased from Aldrich;
-cyclodextrin [17465-86-0] was purchased from Fluka;
Rocuronium Bromide, Org 7402 and Org 25969 were from
Organon Laboratories Ltd. All materials were dried at 70 °C
for 48 h before use. Solutions for binding studies were
prepared by mixing and dilution of stock solutions in 0.2 M
ꢀ
cavity.17–21 Recent ROESY studies of ursodeoxycholic acid
with ˇ-cyclodextrin2 and sodium cholate and sodium deoxy-
2
2
3
cholate with ˇ-cyclodextrin are producing more detailed
pictures. Again, the aliphatic side chain plays the major role
in binding. Some reports implicate that ring A of (non-bile)
phosphate-buffered D
2
O. For instance, a series of solutions
1
3,16,23,24
steroids is involved in formation of the complex.
Less
of cyclodextrin was made to cover the concentration range
work has been published on the interactions of steroids with
the larger ꢀ-cyclodextrin.
0
4
.2 to 6 mM (each 400 µl). To each solution was then added
00 µl of 0.820 mM Rocuronium Bromide solution. Solutions
A modified ꢀ-cyclodextrin, Org 25969 (4), was developed
as a potent and rapid reversal agent for the neuromuscular
for Job plots were prepared by mixing appropriate quantities
of 1 mM Rocuronium Bromide with 1 mM Org 25969 (all in
4
blocking drug Rocuronium Bromide. A natural question
2
0.2 M phosphate-buffered D O at pH 7.5, pH 4.3 or pH 9.1)
arises as to what the structure of the cyclodextrin–steroid
complex is.
The principal aim of the study was to elucidate the
structure of the Rocuronium Bromide–Org 25969 complex
so that a range of different solution compositions was
sampled, but the total molarity remained 1 mM. Equimolar
solutions were prepared for the structural studies and for
the purposes of tabulating bound shifts. Concentrations are
given in the text. All NMR experiments were performed
on a Bruker DRX 400 spectrometer fitted with an inverse
(
1 : 4). This was a challenging problem, and we found
it convenient to study also some simpler systems, and
so we also report on the properties of the complexes
of Rocuronium Bromide with the naturally occurring ˇ-
cyclodextrin and ꢀ-cyclodextrin, and on a model system, the
Org 7402–Org 25969 complex (5 : 4). This communication
reports on the stoichiometry and formation constants of the
steroid–cyclodextrin complexes 1 : 2, 1 : 3 and 1 : 4. Evidence
is presented for internalization of the steroid in complexes
with the larger ꢀ-cyclodextrin 3 and the ꢀ-cyclodextrin
derivative 4.
1
geometry 5 mm probe and under VT control. 1D H NMR
spectra for binding studies and Job plots were acquired
with a digital resolution of 0.0002 ppm/point and using
a small amount of MeOH as an internal chemical shift
reference. ROESY experiments were performed in the phase-
sensitive mode and with presaturation of the water signal, 16
1
scans, sw D 6.4 ppm, 256 t increments, mixing time 400 ms.
1
3
C chemical shifts were recorded from the HSQC spectra.
HSQC data were obtained with eight scans, sw
H
D 6.3 ppm,
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

NMR of cyclodextrin complexes 253
sw
C
D 180 ppm, and 256 t
1
increments. Analysis of the
Rocuronium Bromide complexes with ˇ-cyclodextrin and
ꢀ-cyclodextrin (1 : 2 and 1 : 3)
NMR binding data was performed within the environment
of an Excel spreadsheet. The spreadsheet was configured
to solve the quadratic equation that relates specific species
concentrations to the known total concentrations present and
For the determination of stoichiometry and binding constants
we simply need to identify protons that are sensitive to
complex formation. In the present case this was generally
the cyclodextrin H-3 proton and the steroid axial methyl
groups (18-CH and 19-CH ) or the steroid H-9˛ proton. In
2
5
the association constant for formation of a 1 : 1 complex.
The embedded solver tool was used to minimize the
difference between the calculated curve and the experimental
data. This spreadsheet is available from the authors. The
following discussion contains frequent references to steroid
and cyclodextrin protons. To clarify any confusion about
which protons are being discussed, cyclodextrin protons are
always referred to as H-3, H-2, etc., with no subscripts,
whereas notations for steroid protons bear an additional ˛
or ˇ identification.
3
3
this section we are only concerned with markers of complex
formation, and not with structure. A good marker will be
a peak that is easy to measure, for instance a sharp singlet,
and/or is a peak that shifts significantly upon complexation.
1
13
The H and C NMR spectra of 1 have been fully
2
6
assigned. The stoichiometry of the complexes 1 : 2 and
1 : 3 were determined to be 1 : 1 by Job’s method (data
2
7
not shown). Results of binding assays are summarized
in Table 1.
Binding curves observed from the system Rocuronium
Bromide and ˇ-cyclodextrin at room temperature are shown
RESULTS AND DISCUSSION
in Fig. 1. The change in chemical shift of the 19-CH
3
peak of
Stoichiometry and binding constants
1
as a function of the concentration of 2 is shown in Fig. 1(A).
Quantitative descriptors of binding are stoichiometry and
binding constants. NMR is routinely used to measure these
parameters for host–guest complexes. The terms of reference
for such studies are either fast exchange or slow exchange. In
the slow exchange regime, all components of a mixture
3
The 19-CH group is a good marker because it is a sharp
singlet that can be seen in a simple 1D spectrum and it shifts
significantly (>0.2 ppm downfield) in the presence of 2. In
the form of binding curve shown here, the concentration
of 1 is kept constant and the concentration of 2 is varied.
The solution composition is expressed on the x-axis as
a mole fraction of 2, i.e. [2]/([1] C [2]). Throughout this
range the spectra appeared as a two-component mixture of
Rocuronium Bromide and cyclodextrin. This confirms that
the complexed and non-complexed components are in fast
exchange on the NMR time scale.
(steroid, cyclodextrin and steroid–cyclodextrin complex)
give rise to their own discrete signals in the NMR spectrum.
If spectral dispersion is sufficient for these discrete signals to
be resolved, analysis of the data is simple. The spectrum is
assigned and information about stoichiometry and binding
comes directly from integrating the spectrum.
In the fast exchange regime, the observed nuclear spin
appears as though it were in two sites at the same time, and
the observed NMR frequency is the mole fraction weighted
average between the frequencies in the native form and the
complexed form of the observed molecule. For example, if
the observed proton is located on the host (cyclodextrin)
molecule.
The solid line is the least squares fitted curve for a
ꢀ
1
1 : 1 complex with Ka D 3500 M . Figure 1(B) shows the
experimental data and the corresponding calculated curve
from the proton (H-5) on the cyclodextrin. The chemical shift
range experienced by this proton is considerably less than
that seen on the steroid axial methyl group, but it is enough of
ꢀ
1
a shift to calculate K reliably. The value obtained (2600 M
a
)
is in satisfactory agreement with the value obtained from
observations on the steroid, and averaging all of the available
υobs D X
H
υ
H
C XHG
υ
HG
ꢁ1ꢂ
where υobs, υ
observed nucleus in the experiment, in the host molecule
and in the complex respectively, and X and XHG are the
H
and υHG are the chemical shifts of the
Table 1. Binding constants of Rocuronium Bromide with
cyclodextrinsa
H
mole fractions of host molecule distributed between the
two sites. In this case the analysis is more complicated
than the slow exchange case because the chemical shift
in the complex cannot be observed. This problem may be
overcome by obtaining information on chemical shift over a
range of different solution compositions—an NMR titration.
Once the stoichiometry of the complex is established, a
binding equation can be written, which can be combined
with Eqn (1) to produce non-linear expressions relating
Observed
protonb
ˇ-Cyclodextrin
ꢀ-Cyclodextrin
Org 25969
2
3
4
H-9˛
20 400
>60 000
1
9-CH3
3 500
3 900
2 600
H-3
H-5
17 000
17 800
a
From curve fitting the NMR titration data. The υ , υ or
H
G
υHG parameters that accompany this fitting are as follows.
the observed chemical shift to K
a
and υHG. The values
1 : 2, 19-CH3 υG D 0.87, υ
D 1.10 ppm; H-3 υ D 3.94,
HG
H
K
a
and υHG are determined by minimizing the difference
υHG D 3.89 ppm. 1 : 3, H-3 υ D 3.91, υ
D 3.81 ppm; H-5
H
HG
between the experimentally observed data and the calculated
curve.
υH D 3.83, υ
D 3.64 ppm; H-9˛ υ D 0.78, υHG D 0.63 ppm.
HG
G
1 : 4, H-9˛ υ D 0.81, υ D 0.68 ppm.
G
HG
b
In this study, fast exchange was observed in complexes
H-9˛ and 19-CH3 are located on the steroid, and H-3 and H-5
are the ‘internal’ protons in cyclodextrins.
1
: 2 and 1 : 3, and slow exchange was observed in 1 : 4.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

254
K. S. Cameron, D. Fletcher and L. Fielding
A 1.15
A
0.8
.75
1
1
1
0
0
0
.10
.05
.00
.95
.90
.85
0
0
0.7
.65
0.6
0
.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
mol fraction β-cyclodextrin
mol fraction γ-cyclodextrin
B 3.85
B 3.92
3
3
3
3
.84
.83
.82
.81
3
3
3
3
3
3
.90
.88
.86
.84
.82
.80
3.8
3
.79
0
.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
mol fraction β-cyclodextrin
mol fraction γ-cyclodextrin
Figure 1. 1H NMR chemical shift changes observed during the
titration of Rocuronium Bromide with ˇ-cyclodextrin. (A) The
behaviour of the steroid 19-CH3 proton signal. (B) The
behaviour of the cyclodextrin H-5 signal. The points are the
experimental data, [1] D 0.41 mM, [2] ranging from 0.013 to
Figure 2. 1H NMR titration curves for Rocuronium Bromide
with ꢀ-cyclodextrin. (A) The behaviour of the steroid H-9˛
proton signal. (B) The behaviour of the cyclodextrin H-3 signal.
The points are the experimental data, [1] D 0.41 mM and [3]
ranging from 0.013 to 8.2 mM at 30 °C, and the solid line is the
6
.6 mM at 30 °C, and the solid line is the calculated curve using
least squares fitted calculated curve using parameters from
Table 1.
parameters from Table 1.
3
ꢀ1
0
assumed to be equal to the starting concentration [H] and
data gives K
a
D 3.3 š 0.5 ð 10
M
for formation of a 1 : 1
2
8
linear graphical solutions ensue. One such linearization
that has been widely used in studies of host–guest com-
complex. Similar curves were obtained from the titration of
Rocuronium Bromide with ꢀ-cyclodextrin (Fig. 2). Again, we
see (Table 1) a good correlation between the K
a steroid proton shift and that derived from a cyclodextrin
2
9
plexation is the Benesi–Hildebrand method. The data from
both of the binding experiments were also suitable for treat-
a
derived from
ment by the Benesi–Hildebrand method ([H]
0
× [G]
0
). This
4
ꢀ1
proton, and K
a
D 1.8 š 0.2 ð 10
M
. With ꢀ-cyclodextrin,
data treatment was applied and very similar results were
H-9˛ of the steroid also proved to be very sensitive to the
presence of the cyclodextrin, and so this was used as the
marker. In plots such as Figs 1 and 2 the difference between
the chemical shift of the observed proton in the complex
ꢀ
1
obtained: K
a
D 3100 M for 1 : 2 from the data of 19-CH
3
;
K
a
D 11 000 Mꢀ for 1 : 3 from the data of H-9˛ .
1
The complex of Rocuronium Bromide with Org 25969
1 : 4)
(
υ
HG) and the non-complexed molecule (υ
the total chemical shift range of the y-axis (often referred
to as υmax) and the value of K determines the degree of
curvature. Hence, compared with the flatter curves of Fig. 1,
the S-shaped curves observed in Fig. 2 graphically illustrate
the stronger affinity of ꢀ-cyclodextrin over ˇ-cyclodextrin for
Rocuronium Bromide.
H G
or υ ) defines
(
Org 25969 has eight pendant mercapto-propionate groups
ringing the primary face of the cyclodextrin torus. The studies
with the 1 : 4 system were performed at several different
conditions of pH to probe the importance of ion interactions.
The studies were also performed at a range of temperatures
in order to explore the fast exchange and slow exchange
NMR time scales.
a
a
The non-linear relationship between υobs and K neces-
1
sitates the curve fitting approach discussed above. An
alternative treatment, dating back to early spectroscopic
studies of complexation, is to make a numerical approxi-
mation that results in a linear relationship between υobs and
K . If the not observed species (e.g. the host) is present
a
in large excess, then the concentration of free host can be
The room temperature H NMR spectra of Rocuronium
Bromide–Org 25969 mixtures were more complicated than
mixtures of 1 with ˇ-cyclodextrin or ꢀ-cyclodextrin. This
system was in slow exchange on the NMR time scale, and so
both free 1, free 4, and the 1 : 4 complex contributed to the
1
spectrum. A portion of the 400 MHz H spectrum of a mixture
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

NMR of cyclodextrin complexes 255
2
R for the least squares fitted straight line is 0.982, and the
line intersects the 100% bound axis at a stoichiometry of
0.47. The binding constant for the 1 : 4 complex is too high to
bound 18-CH3
bound 19-CH3
free 19-CH3
free 18-CH3
measure from this data. In order to measure K , there must
a
be a measurable quantity of uncomplexed 1 or 4 present at
sub-stoichiometric ratios. If we assume that we should be
able to detect at least 0.1 mM of the uncomplexed species, the
4
ꢀ1
lower limit of K
a
is 10
M
.
As the temperature was increased, the signals due
to non-complexed and complexed species coalesced and
then sharpened. At elevated temperatures the system was
completely in the fast exchange regime, and data could be
treated by the methods detailed above for ˇ-cyclodextrin
bound H-9α
1
.10 1.05 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60
and ꢀ-cyclodextrin complexes. Figure 5 shows the result of
a Job plot analysis for the 1 : 4 system at pH 9.1 and 70 °C. In
ppm
1
1
Figure 3. Part of the 400 MHz H NMR spectrum of a mixture
the Job plot a physical parameter (the H chemical shift) is
of Rocuronium Bromide (0.5 mM) and Org 25969 (0.3 mM) in
D2O at pH 7.5, 30 °C, showing peaks for both free and bound
Rocuronium Bromide.
measured at different ratios of [1]/[4], while keeping the total
concentration [1] C [4] constant. This data is a confirmation of
the 1 : 1 stoichiometry observed in slow exchange conditions
at 30 °C (Fig. 4). An almost identical plot was obtained at
pH 4.3 and 60 °C , indicating that the stoichiometry of the
complex is independent of pH.
of 1 and 4 (5 : 3 mole ratio) is shown in Fig. 3. This spectrum
displays the doubling of signals that results from observing
the steroid axial methyl signals (18-CH and 19-CH ) in
3 3
Attempts to extract a value for K
temperature fast-exchange data failed for the same reason
sensitivity) as outlined above, but consideration of the
fast exchange data allowed us to revise the lower limit
a
from the higher-
two different environments. In this solution the steroid is
present in higher concentration than the cyclodextrin, and
the integrated peak ratios (bound versus free) are consistent
with very strong binding and formation of a 1 : 1 complex (i.e.
the amount of complex formed is limited by the amount of
cyclodextrin present, [1 : 4] D [4]). Also visible in this portion
of the spectrum is the signal from H-9˛ in the complex. This
peak is shifted from 0.81 ppm in the uncomplexed steroid.
For a system in slow exchange on the NMR time scale,
the concentrations of constituents can be measured directly
from signal intensities. Hence, the stoichiometry of the
complex is determined in a straightforward fashion from
the binding isotherm, which is expected to be in the form
of two straight lines intersecting at the stoichiometry of the
complex. Figure 4 shows the pH 7.5 data plotted to show
the proportion of steroid bound in the complex (determined
from the peak height ratios of spectra such as Fig. 3) versus
the mole fraction of Org 25969. The correlation coefficient
(
4
ꢀ1
to K
a
> 6 ð 10
M
. At 30 °C (slow exchange regime) there
was a clear trend for the spectra to become sharper as pH
increased. This indicates that the exchange (on/off rate) for
complex formation is faster at higher pH, which in turn
suggests that the formation constant K
a
decreases as pH
increases. Microcalorimetry experiments indicate that the
7
ꢀ1 30
complex 1 : 4 has an association constant of ca 1 ð 10
M
.
Geometry of the complexes
The previous section dealt with the thermodynamic descrip-
tors of binding. We turn now to structural issues and ask
questions about the geometry of the steroid–cyclodextrin
complex. The cyclodextrin molecule is analogous to a short
pipe with one opening (the side of the secondary hydroxy
groups) larger than the other. The depth of the cavity is 7.9
˚
˚
to 8.0 A, and the diameter varies between 6.0 and 6.5 A for
100
0.03
0
0
0
.025
.02
8
6
4
2
0
0
0
0
0
0
.015
0.01
.005
0
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
mole fraction 1
mole fraction cyclodextrin
Figure 5. Job plot for the fast exchange system, Rocuronium
Bromide and Org 25969 at pH 9.1, 70 °C. The total
concentration [1] C [3] D 1.0 mM.
Figure 4. Binding isotherm for the slow exchange system,
Rocuronium Bromide with Org 25969 at pH 7.5, 30 °C.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

256
K. S. Cameron, D. Fletcher and L. Fielding
Table 2. Compiled H chemical shift perturbationsa and NOE
datab observed in Rocuronium Bromide upon binding to
cyclodextrins
1
˚
ˇ-cyclodextrin and between 7.5 and 8.3 A for ꢀ-cyclodextrin.
It is well established that steroids are able to enter the cavity
in ˇ-cyclodextrin.
One source of structural information is the chemical shift
changes relative to the unbound forms observed upon for-
mation of the complex. This parameter is referred to as the
complexation-induced shift (CIS, or υ). Such information
from the steroid is likely to be useful to distinguish between
postulated binding models. A consideration of cyclodextrin
chemical shift changes is expected to be even more profitable,
and indeed it is well established that changes in the chemical
shifts of cyclodextrin H-3 and H-5 protons are indicative of
internalization in guest–cyclodextrin complexes. Observa-
tion of υH-3 > υH-5 is taken to indicate guest binding
from the secondary side.
A second NMR probe of the structure of the complex is the
detection of through-space dipole–dipole interactions (NOE
and/or ROESY) between the steroid and the cyclodextrin.
Again, NOESY contacts between the guest and the cyclodex-
trin protons H-3 and H-5 are hallmarks of internalization.
As with the CIS method, stronger NOEs from H-3 compared
with H-5 are taken to indicate entry of the guest molecule
from the secondary side. With only a few exceptions, guests
generally produce stronger NOEs and larger υ effects at
H-3 than H-5. ROESY experiments are generally preferred to
NOESY because of the short correlation times of these small
molecules.
ˇ-Cyclo-
dextrin
ꢀ-Cyclo-
dextrin
(3)
Org
25969
(4)
(
2)
1˛
ꢀ0.10
1
2
3
ˇ
˛
ˇ
ꢀ0.12
4
4
˛
ˇ
ꢀ0.12
ꢀ0.13
ꢀ
0.07
0.09
ꢀ0.16
5˛
ꢀ
6
6
˛
ˇ
ꢀ
ꢀ
0.14
0.10
7˛
C0.07
C0.10
ꢀ0.10
ꢀ0.15
7
ˇ
8
9
ˇ
˛
C0.14
C0.07
C0.14
1
1˛
ꢀ0.12
ꢀ0.15
11ˇ
C0.08
C0.10
1
1
2˛
2ˇ
C0.14
ꢀ0.16
1
1
4˛
5˛
Mapping the complexation-induced chemical shift
changes
ꢀ0.10
Table 2 summarizes the observed CIS values for Rocuronium
Bromide in the three cyclodextrin complexes. For clarity, only
the largest shifts are displayed. There is a wealth of structure
information in this data. The objective of this part of the
analysis is to look for patterns in the magnitude of these
shifts that might indicate possible complex geometries. For
instance, a preference for binding to the face of the steroid
instead of the edge of the steroid would be signalled by
larger shifts of axial protons over equatorial protons. Larger
shifts of ˛ protons compared with ˇ protons would indicate
stronger binding to one face of the steroid compared with
the other.
15ˇ
16˛
17˛
ꢀ0.31
ꢀ0.10
1
8-CH3
ꢀ
C0.13
0.19
1
2
9-CH3
ˇ-Morpholine
c
ꢀ0.12
C0.07
ꢀ0.09
d
1
6ˇ-Pyrolli-
ꢀ0.15
ꢀ0.10
dine/allyl
1
7ˇ-CH3CO
a
Chemical shift change, (υ D υfree ꢀ υbound). The threshold
value was set to š0.10 ppm for ˇ-cyclodextrin and Org
25969 and š0.07 ppm for ꢀ-cyclodextrin. Data from equimolar
Rocuronium Bromide–ˇ-cyclodextrin (1 : 2)
There are 11 entries in Table 2 for the complex with ˇ-
cyclodextrin (1 : 2). We have noted already that the largest
mixtures (6.1 mM of each component) at 30
°
C.
b
The threshold value for NOEs was arbitarily set so that only
the strongest eight to ten NOEs appear in the table (grey 15%
applied to cell). Data from equimolar mixtures (33 mM of each
shift occurs for the axial 19-CH
3
signal (C0.19 ppm). Nine
other protons have shifted by >0.1 ppm, and with only one
exception (H-15˛) they are located on rings A, B and C of
the steroid. Seven of these strongly shifted protons occupy
axial positions, and only three are equatorial. This may be
suggestive of a face-on interaction of the steroid, rather than
an edge-on binding. The protons with large CIS are equally
distributed over both sides of the steroid (five on the ˛
component) at 30
°
C.
c
OCH2.
NCH2.
d
steroid are H-3 and H-5, which are located inside the torus-
shaped cyclodextrin (Fig. 6). The shifts that are observed on
other protons are insignificant. This suggests that significant
internalization of the steroid occurs.
These results are inconclusive and even contradictory.
They are suggestive that, in the complex 1 : 2, Rocuronium
Bromide is incorporated into the cyclodextrin cavity with
1
3
face, and five on the ˇ face). CIS effects in the C spectrum
were significant, but small (typically ca š0.3 ppm, and up
to C1.7 ppm) and without an apparent pattern. The CIS in
1
the H spectrum of the ˇ-cyclodextrin are listed in Table 3.
The protons that are most sensitive to the presence of the
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

NMR of cyclodextrin complexes 257
Table 3. 1H chemical shift changes observed in cyclodextrins
upon forming a 1:1 complex with Rocuronium Bromide
equal distribution of strongly shifted protons around rings
B, C and D, but only one entry from ring A. There are
almost equal representations from protons on the ˛ face of
the steroid (six) and from the ˇ face (five), and there is a
nearly equal distribution of entries from axial (three) and
equatorial steroid protons (four). As noted previously for
ˇ-Cyclodextrin
2)
ꢀ-Cyclodextrin
(3)
Org 25969
(
(4)
H-1
H-2
H-3
H-4
H-5
H-6a
H-6b
H-7
H-8
ꢀ0.01
ꢀ0.02
C0.05
ꢀ0.02
C0.08
C0.01
C0.01
ꢀ0.03
C0.10
ꢀ0.02
C0.18
C0.02
C0.04
ꢀ0.07
C0.19
C0.07
C0.19
C0.07
ꢀ0.25
ꢀ0.06
ꢀ0.01
1
3
the other two systems, the C CIS values for the 1 : 4 system
were small and there was no discernable pattern when they
1
were mapped onto the steroid skeleton. Hence the H and
1
3
the C steroid CIS values indicate no preference for edge
binding, face binding, or for any particular face of the steroid.
The CIS values from the ꢀ-cyclodextrin (Table 3) suggest
incorporation of the steroid into the cavity. The observed
shifts for the internal protons H-3 and H-5 are two times
greater than those observed for analogous ˇ-cyclodextrin
protons and are an order of magnitude greater than those
observed for any other ꢀ-cyclodextrin protons.
Taken together, the CIS from both 1 and 4 suggest that
the steroid is incorporated in the cyclodextrin and that it
rotates freely inside the cavity. This model is also consistent
a
with the large K measured for the 1 : 4 complex.
Mapping the NOE contacts
More direct information on the complex geometry came from
nuclear Overhauser experiments. The rotational correlation
times of the complexes studied here are such that NOEs are
very weak, and so the 2D ROESY experiment was used to
detect steroid–cyclodextrin interactions. The results of the
ROESY experiments are summarized in Table 2, where the
shading indicates the presence of a correlation between a
steroid proton and H-3 or H-5 of cyclodextrin.
Figure 6. Sketch of the internal proton positions in
cyclodextrins.
no single preferred conformation. However, the ‘face-on’
model suggested by the steroid CIS values is not compatible
with the inclusion model suggested by the cyclodextrin CIS
values.
Rocuronium Bromide–ˇ-cyclodextrin (1 : 2)
For the complex 1 : 2, the interpretation of ROESY data
was straightforward. Only correlations between steroid and
H-3 and H-5 on cyclodextrin were observed (there were
no correlations to any other cyclodextrin protons). The
distribution of NOE contacts follows roughly the general
pattern of large CISs, so that correlations are noted to rings
A, B and C, and there are none to ring D. However, for
the rings A, B and C, there is no correlation between
the observed NOEs and the larger CISs. Only half of the
previously discussed CISs have NOE boxes shaded, and half
of the noted NOE contacts do not correlate with a large CIS.
Like the distribution of CISs, the NOE contacts are equally
distributed over the ˛ face and the ˇ face of the steroid,
and are equally distributed between axial and equatorial
protons. The general conclusion from the consideration of
both the CIS values and the NOE data is that rings A, B and
C of Rocuronium Bromide are buried within the cavity of
ˇ-cyclodextrin. A likely model would be for the cyclodextrin
to form an annular belt around ring B of the steroid, leaving
the ring A substituents projecting into the solvent space and
ring D with its bulky substituents plugging the other side. It
is implicit in such a model that ring A of the steroid passes
through the torus first. This is possible because ring A of
Rocuronium Bromide is an equilibrium mixture of chair and
twist-boat conformations (see below).
Rocuronium Bromide–ꢀ-cyclodextrin (1 : 3)
The CIS values observed in the 1 : 3 system were weaker, and
so the threshold value for inclusion in Table 2 was lowered.
Smaller CIS effects in the ꢀ-cyclodextrin system compared
with the ˇ-cyclodextrin system may indicate that the υmax
values are less in this system, in turn suggesting a looser fit
in the larger ꢀ-cyclodextrin ring. All of the significant shifts
were again seen from protons on steroid rings A, B and
C. Of the seven entries for framework protons, there is no
preference for axial (four) to equatorial sites (three). Neither
is there any argument to support preferential binding to the ˛
or ˇ face of the steroid. The data from observations on 3 again
show much stronger CISs for H-3 and H-5 than for any other
cyclodextrin protons. These results strongly suggest that the
steroid is included into the centre void of ꢀ-cyclodextrin.
A model where ring B of the steroid was centred in the
cyclodextrin torus, but with no preferred orientation, would
explain most of these observations.
Rocuronium Bromide–Org 25969 (1 : 4)
There are 11 entries in Table 2 for CISs > 0.1 ppm, and
the largest shift is observed for H-17˛. There is a nearly
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

258
K. S. Cameron, D. Fletcher and L. Fielding
Rocuronium Bromide–ꢀ-cyclodextrin (1 : 3)
Analysis of the ROESY data for the complex 1 : 3 was
hindered by the poor dispersion of the ꢀ-cyclodextrin signals.
In the 1 : 1 mixtures the cyclodextrin H-3 signal overlaps with
H-6, and H-5 overlaps with H-2. Hence it was impossible
to distinguish a binding mode from this data. In addition,
weak correlations were observed to almost every proton in
the steroid (Table 2). But, taken together with the CIS data,
and especially with the noted strong CIS of H-3 and H-5
of the cyclodextrin, these results are strongly suggestive of
internalization of the steroid into the cyclodextrin.
Rocuronium Bromide–Org 25969 (1 : 4)
Attempts to observe NOE contacts between Rocuronium
Bromide and Org 25969 were greatly hindered by the inad-
equate dispersion at 400 MHz, and the broad lines that
characterize this system. The situation was particularly dif-
ficult because signals from the morpholine and pyrrolidine
signals of 1 overlap with the signals from 4. This made it vir-
tually impossible to distinguish intramolecular NOEs from
intermolecular NOEs. Accordingly, the remainder of this
communication discusses the simpler model systems 5 : 2,
Figure 7. A section of the ROESY spectrum of Org 7042
6 mM) and ˇ-cyclodextrin (6 mM) in pH 7.6 buffered D2O
showing strong correlations between the steroid and the H-3
(
5
: 3 and 5 : 4.
ˇ-cyclodextrin proton, and weaker correlations to H-5.
NOESY studies of the model steroid Org 7402 (5)
Several benefits were obtained by choosing Org 7402 (5)
as a model system for Rocuronium Bromide. Changing the
morpholine and pyrrolidine rings to piperidine completely
only spurious peaks to the H-1, H-2 and H-4 signals of the
cyclodextrin to the steroid, showing there was no binding
to the outside of the cyclodextrin. There was no evidence of
any binding to the primary face.
1
solved the problem of overlaps between steroid H signals
and essential cyclodextrin signals. Replacement of the 17ˇ-
OAc by 17ˇ-OH made a more stable system (slow hydrolysis
of the 17ˇ-OAc group was noted in the aqueous solutions of
Org7402–ꢀ-cyclodextrin (5 : 3)
As with the previous sample, both species showed changes
in their chemical shifts when mixed together. This was very
apparent for 3-H and 5-H of the cyclodextrin. A section of the
ROESY spectrum is shown in Fig. 8. It is evident that there
1
). The structural changes between 5 and 1 are considered
to be minor with respect to the size and shape of the
steroid, and so we assume that they have no impact on the
gross morphology of the complex. Org 7402 is a metabolite
of a similar steroidal neuromuscular blocker—Vecuronium
3
1
Bromide.
Org 7402–ˇ-cyclodextrin (5 : 2)
Both species showed changes in their chemical shifts
when mixed together, showing that some interaction was
occurring, although no clear pattern emerged from these
changes. A section from the ROESY spectrum of an equimolar
2
mixture of 5 and 2 in D O is shown in Fig. 7. As can be
seen, several intermolecular ROE cross-peaks appear. In the
following discussion, stronger NOE contacts are indicated
by bold type. Protons H-3ˇ (not visible in Fig. 7), H-4˛, H-
4
ˇ, H-5˛, H-6˛, H-7˛ and/or H-14˛, H-11˛, H-11ˇ, H-12ˇ,
H-15˛, H-17˛, 18-CH and 19-CH are close to H-3 of the
cyclodextrin. There are weak interactions from H-3ˇ, H-6˛,
H-7˛ and/or H-14˛, H-12ˇ, H-12˛, H-11˛, H-15˛ and 19-CH
to H-5 of ˇ-cyclodextrin, whereas H-7ˇ, and 18-CH show
medium interactions. The 18-CH also shows a correlation to
3
3
3
3
3
H-6. This result would appear to indicate that the steroid can
sit on the secondary face of the cyclodextrin with only small
sections at both ends of the skeleton entering the cavity. No
correlations were observed between the piperidine rings in
the 2ˇ and 16ˇ positions and the cyclodextrin, and there were
Figure 8. A section of the ROESY spectrum of Org 7042
(6 mM) and ꢀ-cyclodextrin (6 mM) showing clear correlations
between the steroid protons and the cyclodextrin protons H-3
and H-5.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

NMR of cyclodextrin complexes 259
A
B
3
3
5
5
6
6
Figure 9. Model of the 1 : 1 complex formed between Org 7042 and ꢀ-cyclodextrin.
are many more correlations from H-5 of the cyclodextrin to
steroid signals, compared with the ˇ-cyclodextrin mixture.
Unfortunately, the signals of H-3 and H-6 overlap, making
any assignments difficult, although most steroid signals
appear to correlate to both H-3 and H-6. The steroid protons
showing correlations to H-3 and/or H-6 are H-6˛ and/or
H-11˛, H-7ˇ, H-8ˇ, H-9˛(?), H-12ˇ, H-14˛ and/or H-7˛, 18-
the plane of the molecule. Ring A of Rocuronium Bromide is
known to exist in a dynamic equilibrium between chair and
twist-boat conformations, with both conformers significantly
populated and in fast exchange at room temperature.3
2,33
In
the ring A chair conformation the 2ˇ-morpholine and the
3˛-ol are trans axial, and would present a substantial steric
constraint to passage of the steroid into the cyclodextrin. In
the ring A twist-boat conformation the 2ˇ-morpholine and
the 3˛-ol are equatorial and project along the long axis of the
steroid. This arrangement would certainly allow the steroid
to thread through the centre of the cyclodextrin.
CH
to H-5 are H-6ˇ, H-6˛ and/or H-11˛, H-7ˇ, H-8ˇ, H-9˛,
H-11ˇ, H-12˛, H-12ˇ, 18-CH and 19-CH
3 3
and 19-CH . The steroid protons showing correlations
3
3
.
There are strong interactions between the protons in the
cyclodextrin cavity and the protons in the central rings (B
and C) of the steroid. This result would seem to indicate that
the main skeleton of the steroid lies within the cavity of the
cyclodextrin in two ways, as sketched in Fig. 9.
Acknowledgements
This work represents only one aspect of a large team effort to develop
neuromuscular block reversal agents. We are grateful to the entire
reversal agents team for both supplying the materials and creating
the opportunity for us to perform and complete this work. We are
particularly indebted to Dr D. C. Rees, Dr M. Zhang and Dr L.
Orbons for their constant interest in this work.
Org 7402–Org 25969 (5 : 4)
This mixture was much more difficult to study. The signals
at room temperature were extremely broad, presumably
because of the smaller K
to be heated to 343 K to make the signals reasonably sharp.
The piperidine CH signals overlapped with the cyclodextrin
signals, and the signals of H-3 and H-5 of the cyclodextrin
overlapped. This made the separation of intramolecular and
intermolecular correlations impossible, so no conclusions
on the structure of the steroid–cyclodextrin complex could
be made.
a
compared with 1 : 4. The sample had
REFERENCES
1. Katz RL (ed.). Muscle Relaxants, Monographs in Anaesthesiology,
2
Vol. 3. Elsevier: New York, 1975.
2
3
. Rees DC, Hill DR. Annu. Rep. Med. Chem. 1996; 31: 41.
. Szpilfogel SA, Zeelen FJ. Steroids 1996; 61: 483.
4. Zhang M, Palin R, Bennett DJ. Akzo Nobel. Patent WO2001-
0316-A1.
. Wallimann P, Marti T, F u¨ rer A, Diederich F. Chem. Rev. 1997; 97:
567.
4
5
1
6. Connors KA. Chem. Rev. 1997; 97: 1325.
7
. Schneider H-J, Hacket F, R u¨ diger V, Ikeda H. Chem. Rev. 1998;
9
CONCLUSIONS
8: 1755.
. Jackson K, Young D, Pant S. Pharmaceut. Sci. Technol. Today 2000;
: 336.
8
9
It is difficult to imagine that non-core binding would be
responsible for the impressive association constant between
Rocuronium Bromide and Org 25969. Although individual
experiments seem sometimes to give conflicting results,
when taken as a whole the balance of the evidence
suggests strongly that Rocuronium Bromide is completely
encapsulated in the central void of Org 25969. The hallmarks
of inclusion into a cyclodextrin (CIS to the H-3 and H-
3
. Hamasaki K, Ikeda H, Nakamura A, Ueno A, Toda F, Suzuki I,
Osa T. J. Am. Chem. Soc. 1993; 115: 5035.
10. B u¨ gler J, Engbergsen JFJ, Reinhoudt DN. J. Org. Chem. 1998; 63:
339.
1. Sato M, Narita M, Ogawa N, Hamada F. Anal. Sci. 1999; 15: 1199.
2. De Jong MR, Engbergsen JFJ, Huskens J, Reinhoudt DN. Chem.
Eur. J. 2000; 6: 4034.
5
1
1
1
3. Uekama K, Fujinaga T, Hirayama F, Otagari M, Yamasaki M.
5
protons of the cyclodextrin, and intermolecular ROESY
Int. J. Pharm. 1982; 10: 1.
1
1
1
4. Liu F-Y, Kildsig DO, Mitra AK. Pharmaceut. Res. 1990; 7: 869.
5. Djedaini F, Perly B. J. Pharmaceut. Sci. 1991; 80: 1157.
6. Badawy SIF, Marshall AL, Ghorab MM, Adeyeye CM. Drug Dev.
Ind. Pharm. 1996; 22: 959.
7. Tan X, Lindenbaum S. Int. J. Pharm. 1991; 74: 127.
18. Tan ZJ, Zhu XX, Brown GR. Langmuir 1994; 10: 1034.
correlations between guest protons and H-3 and H-5) are
clearly demonstrated.
Does physical reality (the size and shapes of these
molecules) permit this conclusion? Ring D of Rocuronium
Bromide is bulky, with large substituents projecting above
1
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260

260
K. S. Cameron, D. Fletcher and L. Fielding
1
9. Vandelli MA, Salvioli G, Mucci A, Panini R, Malmusi L, Forni F.
Int. J. Pharm. 1995; 118: 77
25. Fielding L. Tetrahedron 2000; 56: 6151.
26. Fielding L. Magn. Reson. Chem. 1998; 36: 387.
27. Gil VMS, Oliverira NC. J. Chem Educ. 1990; 67: 473.
28. Foster R, Fyfe CA. Prog. Nucl. Magn. Reson. Spectrosc. 1969; 4: 1.
29. Hanna MW, Ashbaugh AL. J. Phys. Chem. 1964; 68: 811.
30. Cameron KS, Fletcher D, Fielding L. Submitted to Magn. Reson.
Chem.
31. Ohta Y. Acta Med. Okayama 1985; 39: 471.
32. Fielding L, Grant G. J. Am. Chem. Soc. 1991; 113: 9785.
33. Fielding L, Grant G. J. Am. Chem. Soc. 1993; 115: 1902.
20. Mucci A, Vandelli MA, Salvioli G, Malmusi L, Forni F,
Schenetti L. Supramol. Chem. 1996; 7: 125.
2
2
1. Yim CT, Zhu XX, Brown GR. J. Phys. Chem. B 1999; 103: 597.
2. Mucci A, Schenetti L, Vandelli MA, Forni F, Ventura P,
Salvioli G. J. Chem. Soc. Perkin Trans. 2 1996; 2347.
3. Cabrer PR, Alvarez-Parrilla E, Meijide F, Seijas JA, N u´ nez ER,
Tato JV. Langmuir 1999; 15: 5489.
2
2
4. Chun IK, Yun DS. Int. J. Pharm. 1993; 96: 91.
Copyright  2002 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2002; 40: 251–260