Nitroxide bound β-cyclodextrin: Is there an inclusion complex?

Article abstract of DOI:10.1021/jo061194p

(Figure Presented) 236CDTIPNO, a derivative of the persistent free radical TIPNO (1,1-dimethylethyl 2-methyl-1-phenylpropyl nitroxide) covalently bound to a permethylated-β-cyclodextrin, was prepared. Self-association of 236CDTIPNO in water was proved by

Full text of DOI:10.1021/jo061194p

Nitroxide Bound â-Cyclodextrin: Is There an Inclusion Complex?
David Bardelang,† Antal Rockenbauer,*,‡ Laszlo Jicsinszky,§ Jean-Pierre Finet,*,†
Hakim Karoui,† Sandrine Lambert,† Sylvain R. A. Marque,† and Paul Tordo†
UMR 6517 CNRS et Aix-Marseille UniVersite´, AVenue Escadrille Normandie-Niemen, Marseille 13397,
France, Chemical Research Center, Institute of Structural Chemistry, P.O. Box 17, H-125 Budapest,
Hungary, and Cyclolab Ltd., P.O. Box 435, H-1525 Budapest, Hungary
jean-pierre.finet@up.uniV-mrs.fr; rocky@chemres.hu
ReceiVed June 9, 2006
236CDTIPNO, a derivative of the persistent free radical TIPNO (1,1-dimethylethyl 2-methyl-1-
phenylpropyl nitroxide) covalently bound to a permethylated-â-cyclodextrin, was prepared. Self-association
of 236CDTIPNO in water was proved by solvent- and competition-dependent EPR spectroscopy
experiments with 2,6-di-O-Me-â-cyclodextrin (DIMEB) and permethylated-â-cyclodextrin (TRIMEB)
as external hosts competing for accommodation of the TIPNO moiety. Temperature-dependent EPR spectra
were simulated with a novel two-dimensional (field-temperature) EPR simulation program that afforded
a full determination of the thermodynamic parameters characterizing the rate constants of the self-inclusion
reaction derived from Arrhenius and Eyring models. This method allows separating the line broadening
effects due to relaxation from a chemical exchange, even if only the fast exchange regime is accessible
experimentally. The activation parameters for the forward and backward steps were consistent with an
equilibrium between a nonassociated form and a weakly associated form, with activation free enthalpies
for each reaction of around 34 kJ‚mol-1.
Introduction
corresponding superoxide spin adducts (KNO• > KN+O-).2 The
binding modes of these nitroxides with CDs depend on the
homotopic or heterotopic nature of the spin adducts.2,5,6
Recently, we have reported that CDs play an important role in
stabilizing the superoxide spin adduct, leading to EPR signal
enhancement and in increasing the partial resistance of the
nitroxide against biomimetic reductive conditions.1-3
To enhance the superoxide radical detection, cyclodextrins
(CDs) are interesting supramolecular assistants for EPR spin
trapping experiments involving linear1,2 or cyclic3 nitrones. The
affinity of nitrone spin traps for CDs is substantial in most
cases,2,4 but CDs exhibit higher binding constants for the
However, the need for high CD concentrations can constitute
an important drawback for in vivo EPR spin trapping or for
therapeutic applications. Moreover, once the complex is dis-
* To whom correspondence should be addressed. Tel: + 33 491 288 927.
Fax: + 33 491 288 758.
† Aix-Marseille Universite´.
‡ Institute of Structural Chemistry.
§ Cyclolab Ltd.
(4) Bardelang, D.; Clement, J.-L.; Finet, J.-P.; Karoui, H.; Tordo, P. J.
Phys. Chem. B 2004, 108, 8054-8061.
(5) Lehn, J.-M. Supramolecular Chemistry; VCH: Weinheim, Germany,
1995.
(6) Steed, J. W.; Atwood, J. L. Supramolecular Chemistry; J. Wiley &
Sons: Chichester, UK, 2000.
(1) Karoui, H.; Tordo, P. Tetrahedron Lett. 2004, 45, 1043-1045.
(2) Bardelang, D.; Rockenbauer, A.; Finet, J.-P.; Karoui, H.; Tordo, P.
J. Phys. Chem. B 2005, 109, 10521-10530.
(3) Karoui, H.; Rockenbauer, A.; Pietri, S.; Tordo, P. Chem. Commun.
2002, 24, 3030-3031.
10.1021/jo061194p CCC: $33.50 © 2006 American Chemical Society
Published on Web 09/02/2006
J. Org. Chem. 2006, 71, 7657-7667
7657

Bardelang et al.
FIGURE 1. Chemical structures of the nitroxide-containing â-CDs 1, 2, 3, and 4, permethylated-â-cyclodextrin (TRIMEB) 5a, 2,6-di-O-Me-â-
cyclodextrin (DIMEB) 5b, and nitroxide derivatives 6a, 6b, and 6c.
solved in a biological milieu, the difference of affinity of the
respective partners toward the various compartments, lipophilic
or hydrophilic, can result in separation of the guest from the
host. These limitations encouraged us to investigate the spin
trapping behavior of functionalized nitrone spin traps covalently
bound to a cyclodextrin toward self-recognition. Control of the
spin adduct stabilization could be achieved at lower CD
concentration, by designing a nitrone, in which the balance
between the affinities of the two species for CD (nitrone before
and nitroxide after spin trapping) would be shifted in favor of
the nitroxide.
Before addressing the study of such compounds, the synthesis
of a persistent nitroxide covalently attached to a â-cyclodextrin
was necessary to verify the self-inclusion of this type of spin-
labeled products in water. Cyclic and linear nitroxides have been
used to investigate supramolecular entities7 via simple or
multiple association toward macropolycycles (cyclophane8 and
especially cyclodextrins9-12), but few examples deal with
nitroxides attached to cavity-containing molecules. Rebek and
co-workers13 have reported the molecular recognition of small
molecules by a resorcinarene bearing four TEMPO units. The
first covalent proxyl-type nitroxide bound to a cyclodextrin 1
was described in 1970, but the properties of the host-guest
complexation were not investigated (Figure 1).14 Recently, Ionita
and Chechik reported the preparation of three TEMPO deriva-
tives appended to natural â-cyclodextrins 2, 3, and 4 as hosts
for large supramolecular assemblies (Figure 1).15
We now report the synthesis of 236CDTIPNO 14, a per-
methylated-â-cyclodextrin 5a linked to a derivative of the
persistent free radical TIPNO (1,1-dimethylethyl 2-methyl-1-
phenylpropyl nitroxide),16 6a (Figure 1), which corresponds to
the spin adduct of the 2-propyl radical with PBN (2-phenyl-N-
tert-butylnitrone). The nitroxide-cyclodextrin proximity is of
particular importance since the respective molecular diffusions
for both host and guest are no longer a limitation for the
complexation process, provided the spacer is flexible enough
to avoid interfering with the self-inclusion phenomenon. The
choice of the permethylated-â-cyclodextrin as cyclodextrin
platform was motivated by (i) the limitation of competing
reactions by permethylation of the hydroxy groups, and (ii) the
weak interaction between a nitrone and the host, required in
spin trapping applications. However, permethylation of the
hydroxy groups of each glucopyranose ring is responsible for
the loss of the intramolecular hydrogen bonding network on
the secondary rim, leading to a less organized and more flexible
structure of the host. A carbamate function was selected as the
spacer between the nitroxide and the host walls because of the
(7) Franchi, P.; Lucarini, M.; Mezzina, E.; Pedulli, G. F. Curr. Org.
Chem. 2004, 8, 1831-1849 and references therein.
(8) Janzen, E. G.; Kotake, Y.; Diederich, F. N.; Sanford, E. M. J. Org.
Chem. 1989, 54, 5421-5422.
(9) Franchi, P.; Lucarini, M.; Mezzina, E.; Pedulli, G. F. J. Am. Chem.
Soc. 2004, 126, 4343-4354.
(10) Lucarini, M.; Luppi, B.; Pedulli, G. F.; Roberts, B. P. Chem.sEur.
J. 1999, 5, 2048-2054 and references therein.
(14) Paton, R. M.; Kaiser, E. T. J. Am. Chem. Soc. 1970, 92, 4723-
4725.
(11) Kotake, Y.; Janzen, E. G. J. Am. Chem. Soc. 1992, 114, 2872-
2874 and references therein.
(15) Ionita, G.; Chechik, V. Org. Biomol. Chem. 2005, 3, 3096-3098.
(16) (a) Maillard, P.; Massot, J. C.; Giannotti, C. J. Organomet. Chem.
1978, 159, 219-227. (b) Grimaldi, S.; Finet, J.-P.; Zeghdaoui, A.; Tordo,
P.; Benoit, D.; Gnanou, Y.; Fontanille, M.; Nicol, P.; Pierson, J.-F. Polym.
Prepr. 1997, 38, 651-652. (c) Benoit, D.; Chaplinski, V.; Braslau, R.;
Hawker, C. J. J. Am. Chem. Soc. 1999, 121, 3904-3920.
(12) Martinie, J.; Michon, J.; Rassat, A. J. Am. Chem. Soc. 1975, 97,
1818-1823.
(13) Kro¨ck, L.; Shivanyuk, A.; Goodin, D. B.; Rebek, J. J. Chem.
Commun. 2004, 272-273.
7658 J. Org. Chem., Vol. 71, No. 20, 2006

Nitroxide-Bound â-Cyclodextrin
FIGURE 2. Synthesis of 14, a TIPNO nitroxide linked to a permethylated-â-CD.
incompatibility of other systems (for example, thioether that
can interfere with spin trapping)17 in foreseeing extension of
the sequence to the synthesis of nitrone-CD-type spin traps.
Investigation of the binding modes of this paramagnetic species
and temperature-dependent EPR spectroscopy studies gave
thermodynamic and kinetic insights of the nitroxide self-
inclusion reaction for the first time. The activation parameters
of self-inclusion reactions have been extensively investigated
for bimolecular inclusion systems involving cyclodextrins18-21
but scarcely with a nitroxide as guest.22
Results and Discussion
Synthesis of 236CDTIPNO, a Covalent TIPNO-Permeth-
ylated-â-CD Derivative. A cyclodextrin derivative covalently
bound to a persistent TIPNO-derived nitroxide was prepared
(Figure 2). The lack of discrimination between the two rims of
the â-CD cavity in the complexation of PBN4 encouraged us
to graft the TIPNO moiety on the more easily accessible C-6
position of the primary rim. Furthermore, the preferential
recognition of the phenyl moiety in the PBN case4 suggested
grafting the nitroxide by the tert-butyl group. Thus, compound
6b, a hydroxyl derivative of TIPNO, was prepared in five steps
by the method of Klaerner and co-workers.23 Reaction of 6b
with disuccinimidocarbonate in the presence of triethylamine
afforded the crude TIPNO carbonate 11 that was directly treated
with 6-monoamino-6-monodeoxy-O-permethyl-â-cyclodextrin24
13 in dichloromethane in the presence of triethylamine to afford
the CD-nitroxide 14 in 78% yield, based on the cyclodextrin.
(17) Durand, G.; Polidori, A.; Ouari, O.; Tordo, P.; Geromel, V.; Rustin,
P.; Pucci, B. J. Med. Chem. 2003, 46, 5230-5237.
(18) Saudan, S.; Dunand, F. A.; Abou-Hamdan, A.; Bugnon, P.; Lye,
P. G.; Lincoln, S. F.; Merbach, A. E. J. Am. Chem. Soc. 2001, 123, 10290-
10298.
(19) Lock, J. S.; May, B. L.; Clements, P.; Lincoln, S. F.; Easton, C. J.
Org. Biomol. Chem. 2004, 2, 337-344.
(20) Ghosh, M.; Zhang, R.; Lawler, R. G.; Seto, C. T. J. Org. Chem.
2000, 65, 735-741.
(21) Cramer, F.; Saenger, W.; Spatz, H.-C. J. Am. Chem. Soc. 1967, 89,
14-20.
(23) Klaerner, G.; Safir, A.; Nielsen, R.; Jandeleit, B.; Huefner, P.;
Li, Y. PCT Int. Appl. 2000, WO 200053640; Chem. Abstr. 2000, 133,
646046.
(22) Franchi, P.; Lucarini, M.; Pedulli, G. F.; Sciotto, D. Angew. Chem.,
Int. Ed. 2000, 39, 263-266.
(24) Jicsinszky, L.; Iva´nyi, R. Carbohydr. Polym. 2001, 45, 139-145.
J. Org. Chem, Vol. 71, No. 20, 2006 7659

Bardelang et al.
TABLE 1. EPR Parametersa of the 6a-TRIMEB, 6c-TRIMEB, and
6c-DIMEB Systems and Binding Constants for the Formation of 1:1
and 1:2 Complexes in Phosphate Buffer (0.1M, pH 7.4)
to determine the EPR parameters for the species formed in the
course of association. In the simulation, the parameters are
simultaneously optimized for both the mass balance equation
(the two binding constants K1 and K2) and the EPR spectra (g-
factors, hyperfine coupling constants, relaxation parameters).
Since the lines of the nonassociated and associated components
were not resolved, we could not decide a priori whether the
rate of association was slow or fast on the EPR time scale. For
this reason, the spectra were computed either as a superim-
position of the components, corresponding to the case of a
slow association, or by using the weight-averaged parameters
(g-factors, hyperfine coupling constants, relaxation parameters),
corresponding to the case of a fast association. Both assump-
tions gave nearly the same K1 and K2 binding constants (see
Figure 3), and the quality of fit was also similar, thus the data
were not adequate to decide whether the rate of association
is slow or fast. The EPR parameters obtained by the slow
association model are reported in Table 1. Titrations of the
TIPNO carbamate derivative 6c in the presence of 5a or 5b
were also performed under similar experimental conditions
(Table 1).
EPR
parameters
aN
aH
R
â
γ
K
g
(mT) (mT) (mT) (mT) (mT)
(M-1
)
TRIMEB
free 6a
2.00645 1.635 0.249 0.021 0.002 0.007
1:1 complex 2.00654 1.598 0.256 0.030 0.015 0.016 K1 ) 54b
1:2 complex 2.00673 1.528 0.260 0.128 0.053 0.064 K2 ) 78c
free 6c
2.00647 1.595 0.267 0.025 0.003 0.006
1:1 complex 2.00651 1.550 0.269 0.041 0.054 0.047 K1 ) 38d
1:2 complex 2.00648 1.525 0.262 0.140 0.066 0.084 K2 ) 33e
6c-DIMEB
1:1 complex 2.00674 1.528 0.287 0.046 0.030 0.035 K1 ) 6745d
1:2 complex 2.00661 1.544 0.284 0.149 0.089 0.106 K2 ) 18e
a Accuracies of the EPR parameters are on the 5th digit for the Lande´
factor g and 3rd digit for hyperfine coupling constants aN and aH, and the
relaxation parameters. b K1 ) [6a,CD]/[6a] × [CD] with 6a,CD ) 1:1
c
complex. K12 ) [CD,6a,CD]/[6a] × [CD] × [CD] with K12 ) K2 × K1
and CD,6a,CD ) 1:2 complex. d K1 ) [6c,CD]/[6c] × [CD] with 6c,CD )
e
1:1 complex. K12 ) [CD,6c,CD]/[6c] × [CD] × [CD], with K12 ) K2 ×
K1 and CD,6c,CD ) 1:2 complex.
Excellent simulations in the whole domain of CD concentra-
tions (see Supporting Information) could only be achieved by
assuming a single (1:1 complex, K1) and a double (1:2 complex,
K2) association, typical of a stepwise binding process (Table 1
and Figure 3). The absence of a second type of 1:1 complexes
has already been reported by Lucarini and co-workers9 for chiral
heteroditopic nitroxide-DIMEB complexes and attributed to a
conformational preference of the CD. Interestingly, Lucarini and
co-workers9 noted the absence of formation of a complex
between their chiral nitroxides and TRIMEB, their nitroxides
being a priori less bulky than 6a and 6c. In the case of 6a and
6c, significant changes are only observed for CD concentrations
higher than 15 mM (Figure 3).
Both 6a and 6c exhibit similar values of K1 (54 and 38,
respectively) for formation of the 1:1 complexes with TRIMEB,
likely due to inclusion of the same aromatic fragment in the
cyclodextrin cavity.26 This is in good agreement with the close
values of ∆aN ) aN,free - aN,in (0.037 mT for 6a and 0.045 mT
for 6c). The decrease of aN agrees with the less polar lattice of
the nitroxide moiety expected after inclusion in the CD cavity.
However, as noted by Lucarini and co-workers,9 changes of
FIGURE 3. EPR titration of 6a in the presence of 5a in phosphate
buffer: (left column) (s) experimental spectra, (right column) (‚‚‚)
simulated spectra.
aHâ, which is sensitive to both the spin density on the nitrogen
atom and the dihedral angle <Hâ-C-N-2pz>, would be
expected. In fact, the absence of significant changes (Table 1)
can be due to a compensation effect between the decrease of
spin density on the nitrogen atom (∆aN > 0) and conformational
changes linked to the inclusion (increase of <Hâ-C-N-2pz>
angle).
The interactions between nitroxides 6a and 6c and 5a (Table
1) are weak, and the formation of 1:2 complexes requires very
high CD concentrations (Figure 3). For 6a, the formation of
the 1:2 complex is slightly cooperative (K2 > K1). In the case
of 6c, no cooperativity takes place (K2 > K1) due to an energetic
cost, resulting from either desolvation of the polar amide moiety
or from the lesser hydrophobicity of the CH2OCONHi-Pr group.
In the second step, the better complexation of the t-Bu fragment
Moreover, a second related TIPNO carbamate derivative 6c was
prepared to evaluate the possible influence of the linker group
on the EPR spectrum, which could interfere with the effect of
the CD complexation. Carbamate 6c was obtained by condensa-
tion of the TIPNO carbonate 11 with isopropylamine in the
presence of triethylamine.
EPR Study of the Noncovalent Interactions in Complexes
6a/TRIMEB, 6c/TRIMEB, and 6c/DIMEB. To study the
interactions between nitroxides 6a or 6c and cyclodextrin
derivatives, EPR titrations of the nitroxides 6a or 6c in the
presence of TRIMEB 5a and DIMEB 5b were performed with
increasing concentrations of the cyclodextrin derivative in an
aqueous stock solution of TIPNO (Table 1 and Figure 3). To
decompose the EPR spectra recorded with the different cyclo-
dextrin concentrations (Figure 3), the two-dimensional field
concentration simulation program was utilized.25 This method
allows obtaining a unique decomposition even in the case when
individual spectra do not have adequate amounts of information
in 6a is emphasized by the larger value of ∆′aN ) aN,1:1
-
(25) Rockenbauer, A.; Szabo´-Pla´nka, T.; AÄ rkosi, Sz.; Korecz, L. J. Am.
Chem. Soc. 2001, 123, 7646-7654.
(26) The 1:1 complex with the t-Bu group included in the cyclodextrin
was disregarded because very low (more than 1 order less) binding constant
was obtained.
7660 J. Org. Chem., Vol. 71, No. 20, 2006

Nitroxide-Bound â-Cyclodextrin
TABLE 2. EPR Parametersa of Nitroxides 6a, 6b, 6c, and 14 in
Various Solvents
species
(%)
aN
aH
R
â
nitroxide
solvent
g
(mT) (mT) (mT) (mT)
6a
6b
6c
6c
14
14
14
14
100
100
100
100
100c
100c
100c
waterb
waterb
waterb
2.00645 1.635 0.249 0.021 0.002
2.00647 1.598 0.263 0.024 0.003
2.00647 1.595 0.267 0.025 0.003
t-BuPh 2.00691 1.446 0.286 0.031 0.011
CDCl3 2.00690 1.472 0.264 0.109 0.013
toluene 2.00697 1.441 0.267 0.074 0.001
t-BuPh 2.00699 1.437 0.270 0.080 0.018
38 (A)c waterb
62 (B)d
2.00675 1.543 0.264 0.043 0.035
2.00646 1.571 0.252 0.250 0.021
a Accuracies of the EPR parameters are on the 5th digit for the Lande´
factor g and 3rd digit for hyperfine coupling constants aN and aH, and the
relaxation parameters. b Chelex-treated aqueous phosphate buffer (0.1 M,
pH 7.4). c Weakly included species. d Nonincluded species.
aN,1:2 for formation of the 1:2 complex for 6a (∆′aN ) 0.070
mT) than for 6c (∆′aN ) 0.025 mT). As expected for inclusion
of a nonchiral fragment, the aHâ values do not change
significantly (variations of 0.004 and -0.007 mT for 6a and
6c, respectively).
It is worthwhile noting that K1 for the formation of the 6c-
DIMEB 1:1 complex is more than 150 times greater than K1
for formation of the 6c-TRIMEB 1:1 complex. This is likely
due to an easier access to the DIMEB cavity (with a non-
methylated hydroxyl group in the position 3) for 6c. In their
nitroxide-DIMEB systems, Lucarini and co-workers9 observed
smaller K values probably because their nitroxides are less
hydrophobic than 6c (i-Pr and phenyl groups). More surprising
is the increase of aN (∆′aN ) -0.016 mT) for the formation of
the 1:2 complex of 6c with DIMEB. It has already been
reported27 that the intramolecular hydrogen bond (IHB) matched
the polar effect of electron-withdrawing groups with an increase
of the aN value with increasing IHB strength. The presence of
one free hydroxyl group on each pyranose unit leads the CD
cavity to exhibit different shapes for DIMEB and for TRIMEB.
Hence, the DIMEB cavity can accommodate 6c in a way
favoring either an intermolecular hydrogen bond between the
nitroxide moiety of 6c and one or several of the free hydroxyl
groups carried by the CD cavity, or an intramolecular hydrogen
bond with the H atom of the amide function. The comments on
FIGURE 4. EPR spectra of 236CDTIPNO 14 (5 × 10-4 M) in organic
and aqueous media: (upper curve) (s) experimental spectra, (lower
curve) (‚‚‚) simulated spectra.
of the stabilization of the charged canonic form of the nitroxide
moiety in water.29 The slight changes of aHâ denote minor
changes in the conformations. The EPR spectra of 14 recorded
in CDCl3, toluene, and tert-butylbenzene were satisfactorily
simulated assuming the existence of only one species (Figure
4). The ESR parameters of 14 in tert-butylbenzene (aN ) 1.437
mT and aH ) 0.270 mT) are very close to those reported for 6c
(aN ) 1.446 mT and aH ) 0.286 mT), a good model for the
CD-bound nitroxide 14 with a nonincluded TIPNO moiety (see
Table 2). The CD-included forms could not be detected in the
three organic solvents. When going from tert-butylbenzene,
aHâ done for the formation of the complexes involving TRIMEB
and 6a or 6c hold for the DIMEB-6c system.
toluene to CDCl3, the aN value increases with the solvent
29,30
The significant increase of the relaxation parameters R, â,
and γ in each system (Table 1) denotes that the association with
CD slows down the rotational motion of the nitroxide. It is worth
noting that the relaxation parameters R, â, and γ for the 1:1
and 1:2 6c-TRIMEB complexes are slightly greater than those
for the homologous 6a-TRIMEB complexes. This can be
related to the difference of steric bulkiness between CMe2CH2-
OCONHi-Pr and the t-Bu group. On the other hand, 1:1 and
1:2 complexes of 6c with DIMEB and TRIMEB exhibit very
similar relaxation parameters. Thus, despite a weak molecular
recognition, these data show that TRIMEB can bind TIPNO
derivatives.
EPR Study of 236CDTIPNO, 14. When the EPR spectra
of the three different free nitroxides 6a, 6b, and 6c in aqueous
solutions are compared, a clear decrease27 of aN (Table 2) is
observed, and it can be correlated with the electron-withdrawing
capacity of the substituent:28 Me (σI ) -0.01) to CH2OH (σI )
0.11)27 and CH2OCONHi-Pr (σI ) 0.17). The aN values of 6a,
6b, and 6c are greater in water than in organic solvents because
polarity.
It is noteworthy that the relaxation parameters R
and â are nearly insensitive to the polarity of the organic
solvents. The observed reduction of the tumbling rate is charac-
teristic of nitroxides attached to bulky molecules.31 Moreover,
whatever the solvent, comparison of the EPR spectra of 6c and
14 reveals an increase of the relaxation parameters R and â
with the increasing size of the molecule (Table 2). The relaxation
parameter γ followed an identical trend as parameter â.
In aqueous solution, the asymmetry of the EPR spectrum of
14 is more pronounced with stronger line broadening, particu-
(27) Marque, S.; Fischer, H.; Baier, E.; Studer, A. J. Org. Chem. 2001,
66, 1146-1156.
(28) Charton, M. Prog. Phys. Org. Chem. 1981, 13, 119-251.
(29) Beckwith, A. L. J.; Bowry, V. W.; Ingold, K. U. J. Am. Chem. Soc.
1992, 114, 4983-4992.
(30) Herrenschmidt, Y.-L.; Guette´, J.-P. Constantes Physico-chimiques.
In Techniques de l′Inge´nieur; Editions T. I.: Paris, France, 1988; pp K310-
311.
(31) Cle´ment, J.-L.; Gilbert, B. C.; Rockenbauer, A.; Tordo, P. J. Chem.
Soc., Perkin Trans. 2 2001, 1463-1470.
J. Org. Chem, Vol. 71, No. 20, 2006 7661

Bardelang et al.
an intermolecular one41,42 (Figure 5b). The formation of
micellar-type structures, such as that depicted in Figure 5c, with
the hydrophobic TIPNO part self-organized keeping the hydro-
philic cyclodextrin moiety on the outer sphere of the micelle
could be considered.43-45 However, the spin-spin interactions
between the nitroxide functions of the 236CDTIPNO molecules
that are expected in a micellar model even at low nitroxide
concentrations were not observed. Only at concentrations higher
than 5 × 10-3 M further line broadening was observed that
can be attributed to the spin-spin interactions classically
occurring at high concentrations of nitroxides.46,47 Addition of
dioxane, known to disorganize micelles, to the nitroxide solution
(up to 40% v/v) failed to modify the EPR spectrum.48
Furthermore, a broader line width is expected for the higher
level of organization corresponding to a micelle architecture,
but the observed values of the line width of the two species,
A
B
∆Hpp ) 0.046 mT (weakly associated) and ∆Hpp ) 0.250
mT (nonassociated), follow an opposite trend, thus excluding
the micellar model (Figure 5c). The alternative possibility
involving the presence of two diastereoisomers with different
association constants, resulting from enantiomeric discrimination
between the two enantiomers of the TIPNO moiety by the chiral
cyclodextrin (Figure 5d)sknown for amino acid functionalized
â-CDswas disregarded (see Experimental Section).49-52 Equi-
libria between nonassociated and strongly associated species
(Figure 5e) or between weakly associated and strongly associ-
ated species (Figure 5f) are also possible. These two possibilities
are discarded due to the aN values observed (vide supra) which
are noncompatible.
The R relaxation parameter, smaller for species A (0.043 mT)
than for species B (0.250 mT), means that the weakly self-
associated species exhibits a more compact geometry, allowing
a faster rotation. Indeed, if the molecule is not self-associated,
FIGURE 5. Possible modes of binding of 236CDTIPNO 14 in water.
larly, for the two lines at high magnetic field (Figure 4, spectrum
d). In this case, a good simulation could be achieved only by
assuming the existence of two nearly equally populated species
that are undergoing a fast exchange (Table 2).
(36) Dunbar, R. A.; Bright, F. V. Supramol. Chem. 1993, 3, 93-99.
(37) Eddaoudi, M.; Parrot-Lopez, H.; Frizon de Lamotte, S.; Ficheux,
D.; Prognon, P.; Coleman, A. W. J. Chem. Soc., Perkin Trans. 2 1996,
1711-1715.
(38) Bellanger, N.; Perly, B. J. Mol. Struct. 1992, 273, 215-226.
(39) Ikeda, T.; Yoshida, K.; Schneider, H.-J. J. Am. Chem. Soc. 1995,
117, 1453-1454.
Cyclodextrins are considered as polar molecules with a
nonpolar cavity. Hence, when a nitroxide is attached to CD, a
slight decrease (0.01-0.02) of aN due to the presence of a fairly
electron-withdrawing group (cyclodextrin contains 35 oxygen
atoms out of 210 atoms) can be expected. Thus, the small ∆aN
(0.024 mT) between 6c and 14B led us to consider species B
as exhibiting the nitroxide moiety outside of the inner cavity.
The small ∆aN (0.028 mT) between species A and B (Table 2)
does not agree with a strongly included nitroxide moietys1:2
complex (∆aN ) 0.070 mT and aN expected smaller than 1.525
mT) or 1:1 complex (∆aN ) 0.045 mT)sfor species A. Thus,
species A is likely to be a weakly associated nitroxide, with
the nitroxide moiety located amidst the methoxy crown of the
narrow rim of the CD cavity, as depicted in Figure 5a. However,
several alternative possibilities, displayed in Figure 5b-f, cannot
be excluded. The concentration independence of the EPR
spectrum of 14 in the 10-5 to 10-3 M range is in favor of a
reversible intramolecular model32-40 (Figure 5a,d-f) rather than
(40) McAlpine, S. R.; Garcia-Garibay, M. A. J. Org. Chem. 1996, 61,
8307-8309.
(41) McAlpine, S. R.; Garcia-Garibay, M. A. J. Am. Chem. Soc. 1998,
120, 4269-4275.
(42) Liu, Y.; Fan, Z.; Zhang, H.-Y.; Yang, Y.-W.; Ding, F.; Liu, S.-X.;
Wu, X.; Wada, T.; Inoue, Y. J. Org. Chem. 2003, 68, 8345-8352.
(43) Auzely-velty, R.; Pean, C.; Djeda¨ıni-Pilard, F.; Zemb, T.; Perly, B.
Langmuir 2001, 17, 504-510.
(44) Auzely-velty, R.; Djeda¨ıni-Pilard, F.; Desert, S.; Perly, B.; Zemb,
T. Langmuir 2000, 16, 3727-3734.
(45) Zhang, P.; Parrot-Lopez, H.; Tchoreloff, P.; Baszkin, A.; Ling, C.
C.; De Rango, C.; Coleman, A. W. J. Phys. Org. Chem. 1992, 5, 518-
528.
(46) Bratt, P. J.; Kevan, L. J. Phys. Chem. 1993, 97, 7371-7374.
(47) Gerson, F.; Huber, W. Electron Spin Resonance Spectroscopy of
Organic Radicals; Wiley-VCH: Heppenheim, Germany, 2003.
(48) Chalier, F.; Ouari, O.; Tordo, P. Org. Biomol. Chem. 2004, 2, 927-
934.
(49) Parrot-Lopez, H.; Galons, H.; Coleman, A. W.; Djeda¨ıni, F.; Keller,
N.; Perly, B. Tetrahedron: Asymmetry 1990, 1, 367-370.
(50) Wang, H.; Cao, R.; Ke, C.-F.; Liu, Y.; Wada, T.; Inoue, Y. J. Org.
Chem. 2005, 70, 8703-8711.
(32) Liu, Y.; Zhao, Y.-L.; Zhang, H.-Y.; Fan, Z.; Wen, G.-D.; Ding, F.
J. Phys. Chem. B 2004, 108, 8836-8843.
(51) Ikeda, H.; Nakamura, M.; Ise, N.; Oguma, N.; Nakamura, A.; Ikeda,
T.; Toda, F.; Ueno, A. J. Am. Chem. Soc. 1996, 118, 10980-10988.
(52) One referee suggested that the spectra recorded might be described
by a fast exchange between weakly included speciessone diastereoisomer
and a mixture of both diastereoisomers not included in the cavity, i.e., a
combination of enantioselection and fast exchange; Figure 5a and d. This
possibility cannot be ruled out with the available data. However, this
complex scheme was not taken into account.
(33) Liu, Y.; You, C.-C.; Zhang, H.-Y.; Zhao, Y. L. Eur. J. Org. Chem.
2003, 1415-1422.
(34) Hamasaki, K.; Usui, S.; Ikeda, H.; Ikeda, T.; Ueno, A. Supramol.
Chem. 1997, 8, 125-135.
(35) McAlpine, S. R.; Garcia-Garibay, M. A. J. Am. Chem. Soc. 1996,
118, 2750-2751.
7662 J. Org. Chem., Vol. 71, No. 20, 2006

Nitroxide-Bound â-Cyclodextrin
FIGURE 6. Competition experiments between 236CDTIPNO 14 and cyclodextrins as external competing hosts: (a) 14 (5 × 10-4 M) alone and
with a CD (b) 1 equiv, (c) 10 equiv, (d) 100 equiv, (e) 300 equiv. Left column: TRIMEB (light gray annulus). Right column: DIMEB (dark gray
annulus): (upper curve) (s) experiments, (lower curve) (‚‚‚) simulations.
TABLE 3. EPR Parametersa of A and B Species in Fast Exchange, Representative of the Two Conformers of 14b in the Absence and Presence
of â-CD Derivatives as Competing Hosts
aN (mT)
Populations (%)
R (mT)
â (mT)
entry
CD
equiv
A
B
A
B
A
B
A
B
1
2
3
4
5
6
7
no
DIMEB
0
1
10
100
1
1.543
1.542
1.522
1.446
1.543
1.542
1.526
1.571
1.540
1.487
1.494
1.569
1.571
1.551
52
45
22
8
57
53
28
48
55
78
92
43
47
72
0.043
0.049
0.068
0.078
0.049
0.049
0.058
0.250
0.201
0.206
0.223
0.196
0.204
0.211
0.035
0.036
0.052
0.053
0.034
0.036
0.050
0.021
0.064
0.099
0.124
0.027
0.036
0.089
TRIMEB
10
100
a Accuracies of the EPR parameters are on the 5th digit for the Lande´ factor g and 3rd digit for hyperfine coupling constants aN and aH, and the relaxation
parameters. b 5 × 10-4 M.
the moment of inertia will be rather large, due to the long linear
size of the molecule. If the effective radius is reduced by self-
association, the rotation will be accelerated, resulting in a
narrower line width.53
To get better insight in the species involved in the host-
guest equilibrium, addition of TRIMEB or DIMEB as external
FIGURE 7. Proposed binding modes for 14 in water, alone (γ and
competing complexation hosts was performed and showed
important changes in the pattern of the EPR spectra of 14
(Figure 6).
â), or in the presence (δ, γ, â, and R) of DIMEB (dark gray annulus)
as an external competing host.
spectrum of the weakly included species â and the weakly 1:2
complex R (Figure 7).
An increasing anisotropy of the EPR spectrum of 14 is
observed with increasing concentrations of the competing
cyclodextrins, and that is likely due to a progressive nitroxide
immobilization by bimolecular encapsulation of the TIPNO
moiety. The difference in binding behavior of the two competing
methylated CDs is revealed by the similar variations of the EPR
spectrum, detected on the B form of 14 (Table 3, entries 3 and
7). This effect is underlined by the same aN values for species
A with 10 equiv of DIMEB and 100 equiv of TRIMEB. EPR
spectra were correctly simulated assuming two species A and
B (Table 3, Figure 7), even though more species are present.
Thus, the spectrum of species B is an average of the spectra of
the nonincluded species γ and the intermolecular included
species δ, whereas the spectrum of the species A is an average
The higher solubility of DIMEB in water allowed the EPR
spectrum of 14 to be recorded in the presence of up to 300
equiv of DIMEB. The strong asymmetry of the resulting
spectrum (Figure 6e) is probably related to a marked im-
mobilization of the nitroxide in the cavity of CD aggregates.54
However, the pronounced anisotropy in this limiting case is close
to the situation encountered in the solid state or in the folding
of spin-labeled proteins.55 Whatever the nature of the competing
CD, the aN value of the TIPNO derivative decreases with
(54) Loftsson, T.; Ma´sson, M. Proceedings of 12th International Cyclo-
dextrin Symposium; Montpellier, May 16-19, 2004, APGI: Chatenay-
Malabry, France, pp 409-416.
(55) Holtzman, J. L. Spin Labeling in Pharmacology; Academic Press:
Orlando, 1984.
(53) Suzuki, M.; Szejtli, J.; Szente, L. Carbohydr. Res. 1989, 192, 61-68.
J. Org. Chem, Vol. 71, No. 20, 2006 7663

Bardelang et al.
TABLE 4. Temperature Dependence of the Normalized
Populations of the Weakly Associated Species A (PA) and the
Nonassociated Species B (PB) of 14 and Kinetic Constants of the
Exchange Process Given by the Arrhenius Model
T/°C
T/K
PA
PB
K
k-1/s-1
k1/s-1
20
30
40
50
60
70
80
82
85
293
303
313
323
333
343
353
355
358
0.380
0.388
0.396
0.403
0.410
0.416
0.422
0.423
0.425
0.620
0.612
0.604
0.597
0.590
0.584
0.578
0.577
0.575
0.613
0.634
0.656
0.675
0.695
0.712
0.730
0.733
0.739
8.3 × 106
5.1 × 106
5.7 × 106
6.3 × 106
6.8 × 106
7.5 × 106
8.1 × 106
8.7 × 106
8.8 × 106
9.0 × 106
8.9 × 106
9.5 × 106
10.1 × 106
10.7 × 106
11.3 × 106
11.9 × 106
12.0 × 106
12.2 × 106
anymore. Indeed, only a six-line pattern can be seen, where the
position of the lines can be determined as population weighted
averages (see Experimental Section). In the case of an inter-
mediate exchange frequency, the exchange process could
produce an extra line broadening proportional to PAPBδω2τexc
,
where PA and PB are the respective populations, δω is the
difference of the frequency of the lines for the two conformers,
and τexc is the average exchange time (1/τexc ) PA/τA + PB/τB).
The lines of the nitrogen triplet will have a further broadening
due to the orientation averaging of g and hyperfine anisotropy,
which can be described by the usual relaxation formula, where
MN is the magnetic quantum number of 14N nuclei
FIGURE 8. Temperature dependence of the EPR spectrum of 14
(5 × 10-4 M) in phosphate buffer solution: (upper curve) (s)
experimental spectra, (lower curve) (‚‚‚) simulated spectra.
increasing concentrations of the added cyclodextrins as observed
for the noncovalent system. The trend is more pronounced for
DIMEB when 100 equiv is added with a decrease of 0.077 mT
for species B. Moreover, the proportion of species B increases
from 48 to 92%. These observations can be tentatively explained
by an increasing intermolecular type complexation of 14 with
species B going from a nonincluded form to a strongly
associated 1:2 complex δ (Figure 7).
A similar trend is observed for species A. The increase of
the R and â parameters indicates once again that the molecular
rotation of the nitroxide is slowed upon addition of DIMEB. In
the case of the weakly self-included species â, the presence of
DIMEB leads to complexation of the alkyl parts out of the
methoxy crown of the narrow rim of the CD cavity to form 1:2
heterocomplexed covalent analogue R, although in a modest
percentage (8%). Thus, addition of increasing amounts of
methylated â-cyclodextrins favors the intermolecular binding
mode, with formation of the intermolecular dimer δ (92%).
Moreover, the higher value of RB compared to RA related to
the greater flexibility of the species δ (noncovalent intermo-
lecular type association) compared to species R (covalent
intramolecular type association) is worth noting. These data are
probably due to the higher rotation possibilities offered by the
more compact intramolecular dimer R compared to the more
extended intermolecular dimer δ.
2
W(MN) ) R + âMN + γMN
However, the nonassociated and weakly associated forms of
14 experience different relaxations since the more compact
molecule in the compact form can rotate more easily. Thus, six
different relaxation parameters and the exchange time can
describe the line broadening. Since there are only three
independent line width data in the spectra, the unique parameter
determination is impossible, when the spectra are simulated one
by one. To overcome this shortcoming, we developed recently
a new two-dimensional simulation approach, where the second
coordinate is the temperature.56 Then the aggregate information
of n spectra offers altogether 3n data for the line width. In the
simultaneous spectrum fitting, all spectroscopic parameters were
not adjusted independently, except for the first few coefficients
in their power expansion (eq 1):
f(T - T0) ) f0 + f1(T - T0) + f2(T - T0) 2 + f3(T - T0)3
(1)
where T0 ) 273 K was used in the program.
Typically for the g-factors and the small aH values, two
coefficients were adjusted, while three and sometimes four
coefficients were adjusted for aN and the R, â, and γ relaxation
parameters. Concerning the exchange time, it can be described
either by the Arrhenius equation (eq 2)
Kinetics and Thermodynamics of the Autocomplexation
of 14. The temperature dependence of the EPR spectrum of 14
(5 × 10-4 M) in aqueous phosphate buffer shows a progressive
and reversible symmetrization of the signal related to the
acceleration of the inclusion and exclusion rates of the TIPNO
moiety in the CD cavity. Above 85 °C, decomposition of the
product started to be significant (Figure 8).
k ) 1/τexc ) A exp(-Ea/RT)
(2)
or by the Eyring equation (eq 3)
k ) 1/τexc ) (kBT/h)exp(∆Sq/R)exp(-∆Hq/RT)
The positions of the six-line spectra of 14 are determined by
the isotropic g, aN, and aH constants. If we assume that two
species are present in a slow exchange, two different sets of
(3)
parameters (gA, aNA, aH and gB, aNB, aHB) will determine the
A
positions of the lines in the superimposed spectra. If the
exchange is fast, superimposed lines should not be seen
(56) Rockenbauer, A.; Nagy, N. V.; Le Moigne, F.; Gigmes, D.; Tordo,
P. J. Phys. Chem. A 2006, 110, 9542-9548.
7664 J. Org. Chem., Vol. 71, No. 20, 2006

Nitroxide-Bound â-Cyclodextrin
TABLE 5. Thermodynamic and Kinetic Parameters for the Exchange between the Weakly and Nonassociated Species A and B of 14
a
∆GR
∆HR
∆SR
Ea
A
∆Gqa
∆Hq
∆Sq
(kJ‚mol-1
)
(kJ‚mol-1
)
(J‚K-1‚mol-1
)
(kJ‚mol-1
)
(s-1
)
(kJ‚mol-1
)
(kJ‚mol-1
)
(J‚K-1‚mol-1
)
a
Arrheniusb
Eyringb
association
dissociation
1.2
1.2
2.5
2.5
4.5
4.5
6.6
9.6 × 107
33.9
34.5
33.3
3.5
4.8
2.3
-102.2
-104.5
-100.0
7.9
5.4
12.6 × 107
1.2
2.5
4.5
7.3 × 107
a At 298 K. b Exchange process. The errors for ∆GR and ∆HR are 0.2, for ∆SR and Ea 1.0, for A 2.0 × 107, for ∆Gq and ∆Hq 2.0 and for ∆Sq 10.0.
arbitrary decomposition of a single spectrum (around 50/50).
However, the confidence of the novel proportions is better as
the calculations are based on a set of temperature-dependent
spectra and the populations are derived from the thermodynamic
relations. The data in Table 5 are only slightly affected whether
we use the Arrhenius (eq 2) or the Eyring relation (eq 3) in the
2D simulations.
Statistical analyses were carried out to characterize the quality
of the different models. The Eyring and Arrhenius models were
compared to the cases supposing slow or fast exchanges. The
respective regression coefficients R, defined in ref 25, were
0.998113, 0.998069, 0.997559, and 0.995372 in the four cases.
The noise equivalent ∆R factor25 offering the criterion for signif-
icance was 0.000024. Consequently, according to the R values,
the best kinetic model was given by the Eyring equation, but
the regression in the Arrhenius model was smaller only by 2∆R,
FIGURE 9. Suggested reaction pathway between the non- and the
thus the difference cannot be considered as significant. On the
other hand, the computations, assuming slow or fast exchanges,
gave a significant deterioration of the fit since the drops of R
were 23 and 114 times larger, respectively, than the noise equiv-
alent ∆R change of regression. Thus, the assumption of a chem-
ical exchange between the open and closed species is necessary
to obtain a correct description for the temperature dependence
of the line width in the EPR spectra of 236CDTIPNO.
weakly associated forms of 236CDTIPNO 14 in phosphate buffer.
Furthermore, the relative populations can be given by the van’t
Hoff relation (eq 4):
K ) PA/PB ) exp(∆SR/R - ∆HR/RT)
(4)
Consequently, the number of adjusted parameters determining
the line width could be 6 × 3 ) 18 relaxation and 2 thermo-
dynamic quantities in a typical two-dimensional simulation,
which needs at least 7 spectra having 21 independent line width
data.57 The 2D EPR simulations give the thermodynamic
quantities defined in eqs 2-4, which allow the computation of
association constants K as well as the average exchange rates
k. The populations of A and B can be given as PA ) K/(1 + K)
and PB ) 1/(1 + K), respectively, while the association rate is
To the best of our knowledge, studies of thermodynamic
(∆HR, ∆SR) and kinetic (∆Hq , ∆Sq , ∆Hq-1, ∆Sq-1) parameters,
1
1
in the field of nitroxide recognition, have only been reported
by Lucarini and co-workers22 for the bimolecular inclusion of
a nitroxide into a calixarene cavity. Although their values
strongly differ from ours, the trends found in the noncovalent
mode for nitroxide inclusion into calixarene may also hold for
the TIPNO in permethylated CD carrier system. Hence, on
thermodynamic grounds, the small observed ∆HR means that
the nonassociated B form is as stabilized as the weakly
associated A form. This is likely due to the steric strain of the
isopropyl group of the TIPNO moiety overbalancing the
stabilizing hydrophobic effect of the methoxy crown.
k1 ) kK(1 + K)/(1 + K2) and the dissociation rate is k-1
)
k(1 + K)/(1 + K2). All these data are reported in Table 4. The
individual thermodynamic parameters for the association and
dissociation processes were derived by eqs 5-9.
A1/A-1 ) exp(∆SR/R)
Ea - Ea ) ∆HR
(5)
(6)
The non-negligible value of ∆Hq1 is likely due to steric strain
in the transition state because the crown or the cavity must
accommodate the nitroxide. In our system, ∆Hq and ∆Hq
1
-1
-1
1
are very small (the inclusion and exclusion reactions are almost
barrierless), meaning that the strain and stabilization in the
transition state and in forms A and B are either very similar or
match each other. In agreement with the observations of Lucarini
and co-workers,7,22 negative ∆Sq values were observed, meaning
that the transition state is more ordered or “tight” than either
the self-included or the open form. Thus, to reach the transition
state, the TIPNO moiety must adopt a peculiar conformation
∆Gq - ∆Gq-1 ) ∆GR
(7)
(8)
(9)
1
∆Hq - ∆Hq-1 ) ∆HR
1
∆Sq - ∆Sq-1 ) ∆SR
1
In this set of experiments, populations PA and PB are different
at 293 K in comparison with the values estimated from the rather
(spacer effect). The negative ∆Sq can be due either (i) to the
1
desolvation entropy, (ii) to the approach of the cavity hampered
by the crown of methoxy groups, or (iii) to the rigidity of the
cyclodextrin backbone (which is unable to modify its conforma-
tion for fitting to the nitroxide fragment), although the TIPNO
(57) However, the confidence of this procedure is greatly enhanced if
the number of spectra recorded at different temperatures is equal to or greater
than 10. See ref 56.
J. Org. Chem, Vol. 71, No. 20, 2006 7665

Bardelang et al.
acetonitrile (6 mL) under a nitrogen atmosphere. After stirring the
mixture for 16 h at room temperature, the solvent was distilled
under reduced pressure. The residue was dissolved in methylene
chloride (20 mL) and washed with distilled water (8 mL) and with
a saturated NaHCO3 solution (8 mL). The organic phase was then
washed with brine (8 mL) and dried over Na2SO4. Distillation of
the solvent under reduced pressure gave the crude product (170
mg) which was used directly in the next step without further
purification. C19H25N2O6, M ) 377.42; MS (3 mM ammonium
acetate in methanol, ESI+) m/z 378 [M + H]+, 395 [M + NH4]+,
400 [M + Na]+, 416 [M + K]+.
2-N-Isopropylcarbamoyl-1,1-dimethylethyl 2-methyl-1-phen-
ylpropyl nitroxide (6c). A solution of TIPNO-carbonate 11 (65
mg, 0.17 mmol) and triethylamine (17.5 mg, 0.15 mmol) in
dichloromethane (2 mL) was added dropwise to a solution of iso-
propylamine (31 mg, 0.52 mmol) and triethylamine (17.5 mg, 0.15
mmol) in anhydrous dichloromethane (3 mL) under nitrogen. The
solution was then stirred for 50 min and washed with distilled water
(10 mL). The organic phase was washed with a saturated aqueous
NaHCO3 solution (2 × 10 mL) followed by brine (10 mL) and
dried with anhydrous MgSO4. Distillation under reduced pressure
yielded an orange oil (41 mg, 85%) which was purified by pre-
parative thin-layer chromatography [eluent CH2Cl2/EtOH (95/5)]
to afford 6c. C18H29N2O3, M ) 321.44; EPR (water, 6 lines, aN )
1.595 mT, aH ) 0.267 mT); MS (3 mM ammonium acetate in
methanol, ESI+) m/z 322 [M + H]+, 339 [M + NH4]+, 344
[M + Na]+, 360 [M + K]+.
6-Monoamino-6-monodeoxy-(2,3-di-O-methyl)hexakis(2,3,6-
tri-O-methyl)-â-cyclodextrin hydrochloride (13). 6-Monoamino-
6-monodeoxy-TRIMEB-â-CD was prepared by a slight modifica-
tion of the method of Jicsinszky and co-workers24 [20 min instead
of 10 min heating time and a larger excess of hydrazine hydrate (8
equiv instead of 5 equiv)]: Rf ) 0.20-0.25 (acetone), 70%, white
powder, mp 140-143 °C (lit.24 158-161 °C); δH 3.17-3.97 (m,
42H), 3.39 (s, 18H), 3.51 (s, 21H), 3.61 (s, 21H), 5.02-5.28 (m,
7H), 8.27 (s, 3H).
2-N-6-[-6-Deoxy-2,3-di-O-methyl)hexakis(2,3,6-tri-O-methyl)-
â-cyclodextrinyl]carbamoyl-1,1-dimethylethyl 2-methyl-1-phen-
ylpropyl nitroxide, 236CDTIPNO (14). Under nitrogen, triethyl-
amine (19 µL, 1 equiv) was added to a vigorously stirred solution
of monoamino-6-monodeoxy-permethyl-â-CD hydrochloride 13
(200 mg, 1 equiv) in anhydrous dichloromethane (14 mL). A
solution of the above-described crude 11 (170 mg) in dichloro-
methane (8 mL) was then added, followed by a second equivalent
of triethylamine. The reaction evolution was followed by TLC. After
1 h, the organic phase was washed with water (10 mL) and with a
saturated NaHCO3 solution (10 mL). The organic phase was then
washed with brine (10 mL) and dried over Na2SO4. After filtration,
the solvent was distilled under reduced pressure, and the residue
was purified by flash chromatography on silica gel (gradient CH2-
Cl2, 100%; CH2Cl2/EtOH 95, 19/1) to afford 14 as a viscous orange
oil (180 mg, 78%). C77H131N2O37, M ) 1675.9; EPR (CDCl3, 6
lines, aN ) 1.472 mT, aH ) 0.264 mT); MS (3 mM ammonium
acetate in methanol, ESI+) m/z 1694.0 [M + NH4]+, 1699.1 [M +
Na]+, 1715.0 [M + K]+, 856.2 [M + 2NH4]2+. C77H131N2O37, 2
EtOH, 1.1 CH2Cl2 (1676.87 g mol-1) requires C, 54.35; H, 8.07;
N, 1.54. Found: C, 54.05; H, 8.37; N, 1.64. Exact mass: calculated
for [C77H131N2O37]+Na+ 1698.8328. Found 1698.8311.
moiety is prepositioned in the space close to the narrow rim of
the cavity. On the other hand, the negative ∆Sq-1 is likely due
to the spacer effect requiring a peculiar conformation in the
transition state of the expulsion reaction.
Finally, kinetic data strongly suggest that the inclusion-
exclusion reactions are entropy controlled for the TIPNO-
permethyl-â-CD covalent derivative. The thermodynamic values
agree with this conclusion, even though interpretation of the
results related to ∆SR must be made with caution. The enthalpic
and entropic effects seem to be balanced, involving only a small
stability difference between the nonincluded form B and the
weakly self-included form A (i.e., ∆GR > 0 and ∆Gq-1 > ∆Gq ).
1
The dynamics of the self-association process of 14 is better
understood as an exchange between the nonincluded species B
and the weakly self-included form A as depicted in Figure 9.
Conclusion
A TIPNO-type persistent nitroxide has been grafted on a
permethylated-â-cyclodextrin, and self-association in water was
observed. EPR spectra were simulated assuming a fast exchange
between a weakly (loose complex) and a nonassociated species.
Competition experiments with methylated â-CDs as external
complexing agents revealed that the initial equilibrium was
extended by intermolecular complexation of the TIPNO moiety.
Simulation of the temperature-dependent EPR spectra with a
novel two-dimensional (field-temperature) simulation program
afforded thermodynamic and kinetic insights of the self-inclusion
process. Several factors influence the binding mode of TIPNO
grafted on permethylated-â-CD, among them, the length and
flexibility of the spacer arm that appear to be appropriate for
nitroxide self-inclusion as well as the shape and polarity of the
cavity that is complementary to that of TIPNO. So the binding
mode of nitroxide 14 in water can be expressed as a fast
exchange between a weakly self-included species A and a non-
self-included species B. In view of these data, the observed
changes in the binding modes of 14 with the addition of other
CDs are in line with applications of this class of spin-labeled
material for the recognition of large supramolecular architectures
through nitroxide binding.13 Finally, this nitroxide represents a
good model for further investigations of nitrone-appended
cyclodextrins for EPR spin trapping, in terms of the kinds of
cyclodextrin, guest, and spacer.
Experimental Section
General. All reactants were used as received without further
purification. Crude materials were purified by flash chromatography
on silica gel 60 (0.040-0.063 mm). 1H NMR and 13C NMR spectra
were recorded at 300.13 and 75.54 MHz, respectively. Melting
points were measured on a Bu¨chi B-540 apparatus and are un-
corrected. Mono-6-azido-6-deoxy-2,3-di-O-methyl hexakis(2,3,6-
tri-O-methyl)-â-cyclodextrin was kindly provided by Cyclolab R&D
(Budapest, Hungary).
2-Hydroxy-1,1-dimethylethyl 2-methyl-1-phenylpropyl ni-
troxide, TIPNOOH (6b). TIPNOOH was prepared by the proce-
dure of Klaerner et al.23 with a reaction time of 3.5 h. Flash
chromatography: Rf ) 0.3; pentane/AcOEt, 16/4, 66%, red orange
oil; C14H22NO2, M ) 236.33; EPR (water, 6 lines, aN ) 1.598 mT,
aH ) 0.261 mT); MS (3 mM ammonium acetate in methanol, ESI+)
m/z 237 [M + H]+, 254 [M + NH4]+, 259 [M + Na]+, 275 [M + K]+.
2-Succinimidyloxycarbonyloxy-1,1-dimethylethyl 2-methyl-
1-phenylpropyl nitroxide, TIPNO-C (11). Triethylamine (1.2
equiv) was added dropwise to a solution of TIPNOOH 6b (200
mg, 1 equiv) and disuccinimidyl carbonate (304 mg, 1.4 equiv) in
EPR Measurements. EPR spectra were recorded at room
temperature, unless otherwise specified, on a Bruker ESP 300 EPR
spectrometer at 9.5 GHz (X-band) employing 100 kHz field
modulation and equipped with a Variable Temperature Unit Bruker
ER 4111VT. Reaction mixtures were prepared in a chelex-treated
phosphate buffer (0.1 M, pH 7.4) with a fixed concentration of
nitroxide (5 × 10-4 M) unless otherwise specified. Typical spec-
trometer settings: microwave power, 10 mW; amplitude modula-
tion, 0.024 mT; time constant, 0.640 ms; receiver gain, 5 × 104;
sweep width, 6 mT; sweep time, 84 s. EPR spectra were simulated
with two kinds of 2D EPR software: a 2D simulation of the
7666 J. Org. Chem., Vol. 71, No. 20, 2006

Nitroxide-Bound â-Cyclodextrin
magnetic field versus concentration25 and a 2D simulation of the
magnetic field versus temperature.56
from the program depending on the preliminary fixed EPR
parameters. However, the trend for the calculated k1, k-1, PA, PB,
and K values was not affected by the choice of the initial values of
aN, aH, and ∆Hpp, only the scale was shifted, which allows a reliable
determination of the kinetics and thermodynamics of the self-
inclusion reaction. The better fit obtained by the exchange model
compared to the case when only superimposition was assumed
strongly suggests a chemical exchange between the A and B species
instead of the diastereoisomeric equilibrium.
Binding Constant Calculations. EPR titrations were performed
at a constant concentration (5 × 10-4 M) of 6a and 6c with pro-
gressively increasing the amount of cyclodextrin (from 0.05 × 10-3
to 105 × 10-3 M). The samples were prepared by mixing pre-
determined volumes of freshly made stock solutions of 6a and 6c
and of TRIMEB, changing progressively the cyclodextrin concen-
trations and completing, when necessary, with phosphate buffer to
reach the desired final concentrations.
Acknowledgment. The authors thank the Conseil Re´gional
Provence Alpes Coˆte d′Azur, the CNRS, the Universite´ de
Provence and TROPHOS Company for financial support.
Laurence Charles and Vale´rie Monnier (Spectropole, Universite´s
d′Aix-Marseille I et III), are thanked for the mass spectrometry
study of 6c, 11, 12, and 14. A.R. expresses his thanks for partial
financial support of the Hungarian Scientific Research Fund
(OTKA T-046953) and the grant NKFP 1/A/005/2004 “Medi-
Chem2”.
Temperature-Dependent EPR Spectra. Two models have been
considered for the simulations of the recorded EPR spectra. The
first supposes a superimposition of two species in agreement with
the diastereoisomeric model (d) of Figure 5. The simulations
allowed the determination of the EPR parameters of both species
at each temperature. The second way to simulate the EPR spectra
supposes a chemical exchange between two species in line with
the model (a) of Figure 5, with the rate constants and the populations
that can be deduced from the mathematical treatment. In this case,
the only limitation is that the program simulating the spectra on a
one-by-one basis cannot derive both EPR parameters and the kinetic
constants at each temperature. Therefore, the EPR parameters have
been fixed for A (aN ) 1.543 mT, aH ) 0.264 mT, ∆Hpp ) 0.057
mT) and B (aN ) 1.571 mT, aH ) 0.252 mT, ∆Hpp ) 0.172 mT),
and the populations and exchange time were adjusted to obtain the
best fit. We have to stress that quite different values were derived
Supporting Information Available: Experimental and calcu-
lated spectra for inclusion of 6a in CD are displayed in Figure 1 of
the Supporting Information. This material is available free of charge
via the Internet at http://pubs.acs.org.
JO061194P
J. Org. Chem, Vol. 71, No. 20, 2006 7667