α-Phenyl-N-tert-butylnitrone-type derivatives bound to β-cyclodextrins: Syntheses, thermokinetics of self-inclusion and application to superoxide spin-trapping

Article abstract of DOI:10.1002/chem.200700369

α-Phenyl-ZV-Zerr-butylnitrone (PBN) derivatives bound to β-cyclodextrin derivatives have been synthesized. Inclusion of the PBN group into the β-cyclodextrin moiety is host- and temperature-dependent. In the case of the nitrone linked to permethylated cyclodextrin (Me3CD-PBN), the thermokinetic parameters are in favour of a slow chemical exchange between a tight and a loose complex. In contrast. 2,6-di-O-Me-β-cyclodextrin-grafted PBN (Me2CD-PBN) exists either in a fast exchange or as a strongly self-associated complex. The covalent cyclodextrin - PBN compounds have been used to trap carbon and oxygen-centred free radicals. The self-associated forms of the β-CD-spin-traps are compatible with effective spin-trapping, affording spin-adducts with enhanced EPR signal intensities relative to noncovalent CD-nitrone systems or the nitrone alone. This kind of cyclodextrin-bound nitrone is the first type of covalent supramolecular spin-trap and should open new possibilities for the study of biological free radicals in vivo.

Full text of DOI:10.1002/chem.200700369

DOI: 10.1002/chem.200700369
a-Phenyl-N-tert-butylnitrone-Type Derivatives Bound to b-Cyclodextrins:
Syntheses, Thermokinetics of Self-Inclusion and Application to Superoxide
Spin-Trapping
David Bardelang,*[a] Laurence Charles,[b] Jean-Pierre Finet,[a] Laszlo Jicsinszky,[c]
Hakim Karoui,[a] Sylvain R. A. Marque,*[a] ValØrie Monnier,[b] Antal Rockenbauer,[d]
Roseline Rosas,[b] and Paul Tordo[a]
In memory of Dr. Jean-Pierre Finet, who passed away on April 15th, 2007
Abstract: a-Phenyl-N-tert-butylnitrone
(PBN) derivatives bound to b-cyclo-
dextrin derivatives have been synthe-
sized. Inclusion of the PBN group into
the b-cyclodextrin moiety is host- and
temperature-dependent. In the case of
the nitrone linked to permethylated cy-
clodextrin (Me3CD-PBN), the thermo-
kinetic parameters are in favour of a
2,6-di-O-Me-b-cyclodextrin-grafted
radicals. The self-associated forms of
the b-CD-spin-traps are compatible
with effective spin-trapping, affording
spin-adducts with enhanced EPR signal
intensities relative to noncovalent CD–
nitrone systems or the nitrone alone.
This kind of cyclodextrin-bound ni-
trone is the first type of covalent supra-
molecular spin-trap and should open
new possibilities for the study of bio-
logical free radicals in vivo.
PBN (Me2CD-PBN) exists either in a
fast exchange or as a strongly self-asso-
ciated complex. The covalent cyclodex-
trin–PBN compounds have been used
to trap carbon and oxygen-centred free
Keywords: cyclodextrins
·
EPR
spectroscopy · spin trapping · super-
oxide · supramolecular chemistry
slow chemical exchange between
a
tight and a loose complex. In contrast,
Introduction
Detection of oxygen-centred free radicals is of particular in-
terest for better understanding of their importance in the
physiological control of cell functions[1] and in many pathol-
ogies such as neurodegenerative disorders,[2] hypertension,[3]
cancer[4] or diabetes.[5] Over the years, it has emerged that
many of these diseases are related to the occurrence of oxi-
dative stress situations,[6] due to an imbalanced production
[a] Dr. D. Bardelang, Dr. J.-P. Finet, Dr. H. Karoui, Dr. S. R. A. Marque,
Prof. P. Tordo
UMR 6517 CNRS et Aix-Marseille UniversitØ
FacultØ de Saint-JØrôme
Case 521, Avenue Escadrille Normandie Niemen
13397 Marseille Cedex 20 (France)
Fax : (+33)491-288-758
CÀ
of superoxide radical anion (O2 ).[7] One challenge of the
last decade was to develop new techniques and subsequently
E-mail: david.bardelang@nrc-cnrc.gc.ca
sylvain.marque@univ-provence.fr
CÀ
new spin-traps for study of O2 in vivo. On the other hand,
it has been reported that cyclodextrins (CDs) represent
powerful supramolecular assistants[8] of superoxide trapping
in terms of spin-adduct stabilization, EPR signal intensity
enhancement and partial protection of the spin-adducts
against l-ascorbate reduction.[9–12] However, the concentra-
tion of methylated b-cyclodextrins required for observation
of such effects (50 mm) is physiologically unsuitable for in
vivo applications. Another drawback is the presence in bio-
logical fluids of many potential guest molecules that can
compete intensively for guest accommodation with the de-
sired spin-labelled adducts.
[b] Prof. L. Charles, Dr. V. Monnier, R. Rosas
Spectropole, FacultØ de Saint-JØrôme
Case 511, Marseille Cedex 20 (France)
[c] Dr. L. Jicsinszky
Cyclolab Ltd., PO Box 435, 1525 Budapest (Hungary)
[d] Prof. A. Rockenbauer
Chemical Research Center, Institute of Structural Chemistry
PO Box 17, 1525 Budapest (Hungary)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author: Syntheses of 7
and 10, and their precursors, mass spectrometry data, NMR and EPR
procedures.
9344
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354

FULL PAPER
To address these limitations, a covalent model with the ni-
trone spin-trap attached to a monofunctionalized CD is pro-
posed in this paper. Thus, Me2CD-PBN (7) and Me3CD-
PBN (10), derivatives of a-phenyl-N-tert-butylnitrone
(PBN) bound to 2,6-di-O-Me-b-cyclodextrin (DIMEB-de-
rived)[13] and permethylated b-cyclodextrin (TRIMEB-de-
rived), respectively, have been synthesized. Hence, smaller
amounts of CD should be required for complexation of the
spin-adduct, which should furthermore be better stabilized
because of the greater affinity of the CD cavity towards ac-
commodation of the spin-adduct rather than the ni-
trone.[11,12]
radical, with disuccinimidyl carbonate (DSC) encouraged us
to attempt the direct activation of a hydroxy-PBN derivative
by treatment with DSC.[16] Thus, treatment of 1[14] with DSC
in acetonitrile in the presence of triethylamine (TEA) af-
forded the side-chain activated nitrone 3 (Scheme 1). Selec-
tive methylation in the 2- and the 6-positions of cyclodextrin
derivative 4[17] by a modified Szejtli protocol using dimethyl
sulfate in the presence of barium oxide and barium hydrox-
ide, in equivalent volumes of DMF and DMSO, afforded
compound 5. This compound was reduced to the amine de-
rivative 6 by catalytic transfer hydrogenation with hydrazine
monohydrate in the presence of Pd/C in methanol.[18] The
reaction between amine 6 and the carbonate derivative 3 in
the presence of TEA in anhydrous dichloromethane afford-
ed Me2CD-PBN (7). Overmethylation products were almost
completely removed by flash chromatography on silica gel.
Mass spectrometry analyses of commercial samples of
DIMEB also revealed the presence of overmethylation
products, which do not alter the recognition properties of
the CD. In fact, only one or two supplementary methyla-
tions in the 3-position are not sufficient to disrupt the hy-
drogen bond network responsible for the pseudo-rigid preor-
ganized b-CD- and DIMEB-type conformations, unlike in
the case of the TRIMEB type.
Results and Discussion
Synthesis: a-Phenyl-N-tert-butylnitrone (PBN), a linear ni-
trone, was selected i) for its simplicity for evaluation of the
feasibility of preparation of a covalent nitrone–CD system,
ii) because of its complementary structure toward the cavity
of b-CD derivatives,[11,14] and iii) to determine the influence
of the grafted hosts on the superoxide spin-trapping proper-
ties in relation to those of noncovalent PBN–CD systems.
By considering DIMEB and TRIMEB derivatives as macro-
cyclic hosts, we also expected to favour spin-adduct com-
plexation over spin-trap complexation. Selective chemical
modification of nitrones without affecting the nitrone
moiety is usually difficult to achieve, due to the pronounced
reactivity of the nitrone function.[15] The easy reaction of a
hydroxy group, present on the tert-butyl part of the 1-
Preparation of the analogous Me3CD-PBN (10) was ach-
ieved by applying the same sequence of reduction and con-
densation (Scheme 1) to 6-monodeoxy-6-monoazido-Me3CD
(8).
The proposed structures of 7 and 10 were supported by
the analytical data such as, for example, the mass spectrum
of Me3CD-PBN (10), displayed in Figure 1 (for MS/MS de-
tails, see Supporting Information).
phenyl-2-methylpropyl-1,1-dimethylethyl-2-nitroxide
free
Abstract in French: a-Phenyl-N-tert-butylnitrones (PBNs)
attachØ à Des b-cyclodextrines fonctionalisØees par lꢀ(a)-
phenyl-N-tert-butylnitrone (PBN) ont ØtØ prØparØes pour la
premi›re fois. Les processus dꢀinclusion du fragment PBN
dans la cavitØ des cyclodextrines dØpend du type de cyclodex-
trines et de la tempØrature. Dans le cas dꢀun fragment PBN
attachØ a une cyclodextrine permØthylØe (Me3CD-PBN), les
param›tres thermodynamiques supportent lꢀØchange lent
entre un complexe ayant un fragment PBN fortement associØ
et un complexe ayant un fragment PBN faiblement associØ.
Par contre, avec la 2,6-di-O-Me-b-cyclodextrine portant un
groupe PBN (Me2CD-PBN), il est difficile de diffØrencier
entre un Øchange rapide entre un fragment PBN inclus dans
la cavitØ et lꢀautre a lꢀextØrieur et un complexe avec un frag-
ment PBN fortement associØ dans la cavitØ. Ces syst›mes co-
valents cyclodextrine–PBN se sont rØvØlØs Þtre des pi›ges effi-
caces à radicaux libres centrØs sur le carbone ou sur lꢀoxyg›-
ne. Les signaux de RPE des adduits sont nettement plus in-
tenses avec des syst›mes covalents quꢀavec des syst›mes binai-
res (nitrones+cyclodextrine) ou quꢀavec la nitrone pure. Ces
molØcules sont la premi›re gØnØration de syst›mes de ce type
et offrent de nouvelles possibilitØs pour lꢀØtude des radicaux
libres in vivo.
Figure 1. a) Positive electrospray mass spectrum of 10 and b) MS/MS
spectrum of [10+H]+ at m/z 1633.7
Chem. Eur. J. 2007, 13, 9344 – 9354
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9345

D. Bardelang, S. R. A. Marque et al.
Scheme 1. Synthesis of the CD-PBN derivatives Me2CD-PBN (7) and Me3CD-PBN (10).
NMR study of PBN/TRIMEB: Before the NMR study of
covalent compounds 7 and 10 was addressed, comparison of
the influence of the CD substitution on the recognition
properties in the two corresponding noncovalent systems
was necessary. Interactions in noncovalent bimolecular sys-
tems involving PBN and substituted CDs have only been re-
ported for RAMEB (randomly methylated b-cyclodextrin)
and DIMEB as hosts.[11] In this latter case, a 1:1 complex
was observed and a binding constant of around 80mÀ1 was
found by NMR titration. 1H NMR titration of the PBN/
TRIMEB system was performed with the assumption of a
1:1 complex formation. The chemical shift of the most sensi-
tive (H9) proton was monitored with increasing CD concen-
trations. A fast exchange regime was observed and a nonlin-
ear curve fit according to the Macomber model[19–21] gave a
very small binding constant of 4mÀ1 (Figure S9). In
9346
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354

Nitrones Bound to b-Cyclodextrins
FULL PAPER
TRIMEB, permethylation of the b-cyclodextrin hydroxy
groups results in the loss of rigidity and absence of preorga-
nization of the CD cavity, so the conformations of TRIMEB
are less favourable than those of DIMEB for efficient bind-
ing of PBN. Indeed, the absence of any hydrogen bonding
network in TRIMEB is correlated with the very small value
of KH9 (KH9 =4mÀ1) in relation to the previously reported
values for DIMEB (K1 =80mÀ1)[11] and natural b-CD (K1 =
117mÀ1).[14]
(d=8.28 ppm) and at d=7.85ppm. The lack of signal (see
below; Figure 3) in the H9 region (d=8.05ppm) excluded
the formation of head-to-tail complexes featuring a non-
self-included nitrone fragment. A tentative description of
the new species would be a head-to-tail dimer with one ni-
trone partly included in intermolecular fashion and a second
nitrone partly excluded in intramolecular inclusion (Fig-
ure 2c). The absence of an H9 signal around 8 ppm—which
would otherwise correspond to the non-included nitrone
fragment—at low concentrations led us to consider either
the occurrence of a fast exchange between in- and out-com-
plexes or the existence of a strongly bound complex. To
obtain deeper insight into these alternatives, we performed
two sets of competition experiments with external competi-
tors. In the first series of experiments, DIMEB was used to
observe the competition for PBN accommodation between
the cavity of Me2CD and the cavity of DIMEB. Addition of
DIMEB as an external host led to the appearance of a new
1
NMR study of Me2CD-PBN (7): In the H NMR spectrum
of 7 in D2O (Figure 2b), the signals of the aromatic protons
H7 and H8 and the nitronyl proton H9 are broader than the
corresponding signals observed for free PBN in D2O (Fig-
1
ure 2a). The H NMR chemical shift of H9 in compound 7 is
significantly shielded, by up to 0.5ppm, in comparison with
the maximum complexation-induced shift (CIS) reported in
the noncovalent PBN/DIMEB case (Ddꢀ0.41 ppm).[11] Such
[22]
^
a strong upfield shift of H9 is indicative of a deep inclu-
signal in the H8 region ( , Figure 2d), probably due to inter-
sion of the PBN part inside the CD cavity. The absence of
significant changes in the spectrum of 7 (Figure 2b) in the
1–9 mm range of concentrations excluded the formation of
intermolecular complexes. At higher concentrations (Fig-
ure 2c), however, new signals appeared in the region of H8
molecular nitrone–DIMEB complexes (this is further sup-
ported by the absence of H8 and H9 signals corresponding to
free PBN). 1H NMR integration of H8 indicated that the
presence of 10% of intermolecular DIMEB:7 complex pro-
moted strong self-inclusion of the PBN moiety in the
Me2CD cavity, as the included guest can hardly be com-
plexed intermolecularly in spite of the presence of a sixfold
excess of competitor. It is worth mentioning that a clear dif-
ference is observed for the shift of H8, although the cavity
(DIMEB) used for the intermolecular complex formation is
very similar to that in the self-included Me2CD complex.
This denotes the high sensitivity of H8 to its chemical envi-
ronment.
In the second set of competition experiments, adamantan-
1-ol (Kꢀ104–105 mÀ1)[23] was used to expel the nitrone
moiety outside of the Me2CD cavity. The 1H NMR spectrum
of 7 was strikingly modified upon addition of adamantan-1-
ol (Figure 2e), with the appearance of a peak at d=
7.99 ppm, corresponding to the non-included nitronyl proton
H9 (d=8.01 ppm for PBN, and d=7.95ppm for H 9 of 10;
see below). However, the only, scarce, modifications ob-
served in the H8 proton region were not compatible with the
induction of complete expulsion of the nitrone moiety by
the inclusion of adamantan-1-ol. Indeed, in such a case, an
unambiguous signal for H8 should be observed around
8.25ppm. Hence, the absence of a dH8 shift led us to consid-
er the possibility that the cavity may accommodate both the
nitrone moiety and the adamantan-1-ol competitor, by push-
ing only a part of the nitrone moiety outside of the cavity:
that is, the nitronyl proton H9 (Figure 2e). Consequently,
this experiment further confirms strong binding of the ni-
trone moiety in the appended Me2CD cavity. Moreover, an
1
absence of changes in the H NMR spectrum was observed
when the temperature was increased from 13 to 578C (9 mm
in water; see below).
Figure 2. Aromatic parts of 1H NMR spectra in D2O of a) PBN, b) 7
(2 mm), c) 7 (50 mm), d) 7 (8 mm) with DIMEB (50 mm), and e) 7 (2 mm)
with adamantan-1-ol (4.8 mm). Truncated cones are used to symbolize
^
CD derivatives, egg shapes for PBN and spheres for adamantan-1-ol;
NMR study of Me3CD-PBN (10): In contrast with that of
molecule 7, the 1H NMR spectrum of Me3CD-PBN (10;
^
and
symbols represent possible species accounting for additional sig-
nals; see text.
Chem. Eur. J. 2007, 13, 9344 – 9354
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9347

D. Bardelang, S. R. A. Marque et al.
with that of the free PBN mol-
ecule
ROESY spectrum showed cor-
relation peaks and
(Figure S10).
The
A
B
(Figure 4) between the aromat-
ic protons and methyl groups
of the PBN moiety and the
inside cavity protons of the
CD. These correlation peaks
are indicative of a deep inclu-
sion of a nitrone moiety self-
included inside the CD cavity.
The absence of significant
changes in the 1H NMR spec-
trum (Figure 3b) upon varia-
tion of the concentration of 10
from 0.1 to 10 mm excluded
the formation of intermolecu-
lar complexes. However, at
higher concentrations (50 mm,
Figure 3c), new broad singlets
^
( ) appeared at d=7.48 ppm
and d=7.73 ppm, correspond-
ing to H7 and H9, respectively.
At the same time, the self-in-
cluded species disappeared and
the amount of non-included
(or weakly associated; see
Figure 3. Aromatic part of the 1H NMR spectra in D2O of a) PBN, b) 10 (2 mm), c) 10 (50 mm), d) 10 (10 mm)
with DIMEB (50 mm), and e) 10 (2 mm) with adamantan-1-ol (4.8 mm). Truncated cones are used to symbolize
below)
species
decreased
^
^
CD derivatives, egg shapes for PBN and spheres for adamantan-1-ol; and symbols represent possible spe-
slightly. The absence of signifi-
cant changes for H8 (no shield-
ing such as observed in the
case of 7), the slight shielding
cies accounting for additional signals; see text.
Figure 3)
exhibits
several
peaks in the region of protons
H8 and H9 ascribed both to the
non-included (or weakly asso-
ciated; see below) nitrone
moiety (broad doublet at d=
8.25–8.33 ppm for H8, singlet at
d=7.95ppm for H 9, and mul-
tiplet at d=7.50–7.64 ppm for
*
H7 in Figure 3b ( ) and to the
self-included nitrone moiety
(broad singlet at d=8.51 ppm
for H8, singlet at d=7.77 ppm
for H9, and multiplet at d=
7.50–7.64 ppm for H7 in Fig-
*
ure 3b ( ). The occurrence of
such kinetically stable host–
guest systems has seldom been
reported, especially in cases in-
volving host–guest appended
compounds.[24]
The HMQC spectrum exhib-
ited additional signals in the
aromatic region in comparison
Figure 4. 500 MHz ROESY spectrum of 10 in D2O (9 mm) illustrating intramolecular ROE cross-correlations
due to interactions between the methyl (A) and aromatic (B) protons of the PBN moiety and the protons lo-
cated inside the Me3CD cavity.
9348
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354

Nitrones Bound to b-Cyclodextrins
FULL PAPER
of H9 and the presence of a
new shielded singlet peak for
H7 make the presence of a
head-to-tail complex unlikely.
On the contrary, a head-to-
head complex should increase
the shielding of H7 with minor
changes in the region of H8.
Furthermore, such a complex
may force the inclusion of the
PBN fragment deeper into the
cavity, entailing a slightly more
shielded H9 signal. To obtain
deeper insight into these inclu-
sion-exclusion processes, we
performed competitive experi-
ments using either DIMEB
(K=80mÀ1) to pull the nitrone
fragment out of the Me3CD
cavity to form an intermolecu-
lar
complex
inside
the
DIMEB, or adamantan-1-ol (K
ꢀ104–105 mÀ1[23]) to push the ni-
Figure 5. 1H NMR temperature dependence of the H8 and H9 signals of 10 (9 mm) in D2O.
trone fragment out of the
Me3CD cavity. In the presence
of an excess of DIMEB, the signals of H8 and H9 previously
attributed to the non-included (or weakly associated; see
for non-included (or weakly associated; see below) and self-
included species coalesce (TC ꢀ458C; Figure 5).
*
below) nitrone conformer ( in Figure 3b) were shifted sig-
^
nificantly ( in Figure 3d). However, the signal of the self-
Fast versus slow exchange processes: As mentioned above,
in the case of Me2CD-PBN (7) there was no temperature
dependence of the H8 and H9 lines, which in general repre-
sents a fast exchange process on the NMR timescale. Such a
fast exchange is the general case for host–guest interactions
involving cyclodextrins as hosts.[25] On the other hand, the
temperature dependencies of H8 and H9 observed for 10 are
characteristic of slow exchange processes, which are less
common for host–guest interactions involving cyclodextrins
as hosts. Simulations of the NMR line-shapes of the signals
(superimposed lines, Figure 5) were carried out with a 2D
simulation program,[26] affording the rate constants k1 for
the self-inclusion (forward reaction) of the nitrone moiety
inside the Me3CD cavity and kÀ1 for the exclusion (back-
ward reaction), together with the corresponding populations
(P1, PÀ1; see Supporting Information). The averaged values
of k1 and kÀ1 obtained for H8 and H9 were used to deter-
mine the Arrhenius (kinetic, frequency factors A and activa-
tion energies Ea, Figure 6 and Table 1) and Vanꢁt Hoff (ther-
modynamic, Gibbs energy DGr, reaction enthalpy DHr, and
reaction entropy DSr, Figure 6 and Table 1) reaction param-
eters. It is worth mentioning the enrichment of the non-asso-
ciated species with increasing temperature (decrease in K).
The thermodynamic parameters show that the self-inclusion
reaction is enthalpically favoured (DHr ꢀÀ11 kJmolÀ1) but
entropically (DSr =À35.0 JKÀ1 molÀ1) compensated (DGr
ꢀ0 kJmolÀ1).
included nitrone fragment was only slightly affected (that is,
a slight decrease in the amount of the self-included species).
The shielding only of the H9 proton in the presence of
DIMEB is in agreement with a fast exchange between the
non-included (or weakly associated; see below) nitrone frag-
ment and DIMEB. This indicates that the self-included com-
plex is strong enough to compete with the DIMEB complex.
On the other hand, when adamantan-1-ol was added as a
competitor, the spectrum changed strikingly (Figure 3e).
The H8 and H9 peaks corresponding to the self-included spe-
*
cies conformer ( in Figure 3b) disappeared and new peaks
appeared (that is, d=8.45ppm for H 8 and d=7.90 ppm for
H9). Integration of H9 for the non-included (or weakly asso-
ciated; see below) species increased from 0.50H to 0.79H,
while the new conformer accounted for only 0.21H. One
might expect that inclusion of adamantan-1-ol in the empty
CD cavity should not modify the chemical shift of PBN, so
the increase in intensity for H9 can be attributed to competi-
tion between adamantan-1-ol and the self-included nitrone
fragment, the latter being expelled out of the cavity. The
presence of additional peaks can be ascribed to a species ac-
commodating both the nitrone fragment and the adaman-
tan-1-ol in the cavity. Unlike in the case of 7, however, this
species is minor, probably because the Me3CD cavity is less
suited than the Me2CD cavity to accommodate both the ni-
trone and the adamantan-1-ol. The temperature dependence
of H8 and H9 showed dramatic changes when the tempera-
ture was increased from 258C to 458C, at which the signals
This is typical of nonclassical hydrophobic interactions
(DHr ! 0, TDSr ! 0).[27] The entropic cost can be ascribed
Chem. Eur. J. 2007, 13, 9344 – 9354
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9349

D. Bardelang, S. R. A. Marque et al.
attributed to the energetic cost for the desolvation of the
polar nitrone function for the forward reaction and to the
loss of stabilizing van der Waals host–guest interactions for
the backward reaction. In the latter case, the large positive
value of DSꢀÀ1 highlights a transition state less sterically hin-
dered (loose TS) than the self-included species. Notably, an
unexpected positive DSꢀ1 value for the forward reaction is
observed, whereas one would expect a sterically hindered
approach of the nitrone fragment (DSꢀ1 < 0) due to shield-
ing of the cavity entrance because of the methylation of the
hydroxy groups. Indeed, the spacer holds the PBN fragment
close to the entrance of the cavity (spacer effect), forcing in-
clusion of the PBN fragment deeper in the cavity, giving rise
to a positive DS1ꢀ value.[29,30] Consequently, the expected en-
tropic cost (DSꢀ < 0) for the host–guest adaptation is over-
balanced by the spacer effect. All these energetic observa-
tions are accounted for well by an out–in complex equilibri-
um: that is, an equilibrium between a deeply included ni-
trone fragment and a nitrone fragment close to the entrance
of the cavity as displayed in Figure 7.
&
*
Figure 6. Arrhenius (top; : k1, : kÀ1) and Vanꢁt Hoff (bottom) plots for
the outQin equilibrium of 10 in D2O.
Nitrone versus nitroxide inclu-
sion-exclusion processes: In
Table 1. Kinetic and thermodynamic parameters for the exchange process between the free and self-included
conformers of Me3CDPBN (10) and Me3CDTIPNO (11) in water.
previous attempts to model
CÀ
the formyl (CO2 ) and super-
[d]
DGr[a,b]
DHr[a]
DSr[c]
Ea
A[e]
DG#[f,b]
DH#[f]
DS#[g]
ÀC
oxide (O2 ) spin-adducts, ni-
10
À0.26
À10.7
À35.0
73.2[h]
1.6”1014[h]
65.9[h]
66.3[i]
34.5[h]
33.3[i]
84.4[h]
96.5[i]
4.8[h]
2.3[i]
62.1[h]
troxide 11, Me3CD-TIPNO
(Figure 8), was prepared and
its inclusion-exclusion process-
es were studied by EPR.[16]
Those data are now discussed
85.4[i]
1.9”1016[i]
12.6”107[h]
7.3”107[i]
101.4[i]
À104.5[h]
À100.0[i]
11[j]
1.2
2.547.5.9
[h]
5.4[i]
[a] kJmolÀ1, error ca. 2 kJmolÀ1. [b] At 298 K. [c] JKÀ1 molÀ1, error ca. 5JK À1 molÀ1. [d] kJmolÀ1, error ca.
8 kJmolÀ1. [e] In sÀ1, error of a factor of 3. [f] kJmolÀ1, error ca. 7 kJmolÀ1. [g] JKÀ1 molÀ1, error ca.
20 JKÀ1 molÀ1. [h] Association process. [i] Dissociation process. [j] For molecule 11, see ref. [16] and Figure 8.
both to the amide-type spacer and to the rigidity of the
planar aromatic nitrone moiety, which reduces the mobility
of the guest included in the cavity. Thus, the enthalpic gain
due to the van der Waals interactions inside the cavity is
merely sufficient to compensate the entropic cost due to the
adoption of a strained supramolecular conformation. The
Arrhenius parameters show large activation energies Ea,1
and Ea,À1 and unexpectedly high frequency factors A1 and
AÀ1 that are typical of a loose transition state. These high
values might be partly due to the EaÀA compensation error
effect,[28] and allowed only a qualitative discussion of the
Figure 7. Reversible pathway from weakly to strongly associated PBN
part for 10 in D2O.
Eyring parameters DHꢀ and DSꢀ. The large DHꢀ values are
Figure 8. Nitroxide models 11–13 used for EPR study of the spin-adducts.
9350
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354

Nitrones Bound to b-Cyclodextrins
FULL PAPER
in comparison with those of
10.
Dramatically different Ar-
rhenius and Vanꢁt Hoff param-
eters were observed. In con-
trast to Me3CD-PBN (10), the
non-included nitroxide frag-
ment species is preferred over
the self-included species (DGr
> 0), and the DHr and DSr pa-
rameters
exhibit
positive
values. However, the magni-
tudes for the parameters re-
ported for 11 are clearly small-
er than those observed for 10,
meaning small differences be-
tween the species involved in
the exchange processes for 11.
In this case, the exchange pro-
cess involves two conformers
in equilibrium with the nitro-
Scheme 2. Proposed general situation for nitrone and nitroxide self-inclusion equilibria prior to (top) and after
(bottom) superoxide spin-trapping by CD-functionalized spin-traps 7 and 10.
xide fragment in non-included and in partly included
(mainly lying in the middle of the methoxy crown of the
Me3CD cavity) positions, in contrast with the equilibrium
observed in the case of nitrone Me3CD-PBN (10). The neg-
ative DSꢀ1 observed for the exchange in 11 in contrast to 10
denotes a highly hindered TS, probably due both to the
change in hybridization on the Ca carbon, from sp2 in the
nitrone moiety to sp3 in the nitroxide moiety, and to the
presence of the bulky isopropyl group in the b position.
These two modifications make the forward reaction (inclu-
sion process) for 11 more sensitive to steric hindrance, due
to the methoxy crown shielding the entrance of the cavity.
Furthermore, DHꢀ1 is clearly smaller for 11 than for 10, as
would be expected for an easier desolvation of the less polar
nitroxide function. However,
with covalent nitrones 7 and 10 to exhibit greater persis-
tence, due both to conformational changes decreasing frag-
mentation reactions[31] and to the self-inclusion of the nitro-
xide fragment, making the bioreduction less efficient
(Scheme 2).
CÀ
CO2 spin-trapping EPR experiments: Spin-trapping of a
CÀ
carbon-centred radical such as CO2 , which commonly gives
persistent spin-adducts with intense EPR signals, was first
investigated with both the bimolecular PBN/TRIMEB and
the unimolecular Me3CD-PBN (10) systems to check the
spin-trapping efficiency of molecule 10. The bimolecular
PBN/TRIMEB was studied first (entries 1 and 2, Table 2)
with equimolar amounts of spin-trap and host. The EPR pa-
for the included nitroxide frag-
CÀ
Table 2. EPR parameters for the exchange between the free and the partly included forms of the CO2 ad-
ment of 11, one would expect
a larger DHꢀÀ1 value than for
10, because of stronger stabi-
lizing van der Waals interac-
tions between the apolar
cavity and the nitroxide frag-
ment, which is less polar than
the nitrone moiety. Hence, the
smaller DHꢀÀ1 for the exclusion
of nitroxide than for the ni-
trone fragment is consistent
ducts of PBN/TRIMEB and 10 systems in phosphate buffer (pH 7.4).
[a]
Entry
Spin-adduct
aN[a] [mT]
ab,H [mT]
a
[a] [mT]
b[a] [mT]
g
[a] [mT]
CÀ[b]
1
2
3
4
TRIMEB/PBN-CO2
TRIMEB/PBN-CO2
1.581
1.566
1.512
1.494
0.322
0.338
0.300
0.310
0.137
0.048
0.217
0.092
0.040
0.006
0.062
0.036
0.047
0.005
0.077
0.035
CÀ[c]
CÀ[d,e]
10-CO2
CÀ[d,f]
10-CO2
[a] Nitrogen aN and b-hydrogen ab,H hyperfine coupling constants, and relaxation parameters a, b, and g are de-
termined as in ref. [16]. [b] Free spin-adduct. [c] Included spin-adduct. [d] Concentration of 10 in the 3.5to
25m m range. [e] Non-included spin-adduct fragment. [f] Partly self-included spin-adduct fragment.
with weak van der Waals interactions, as would be expected
for the nitroxide part lying in the methoxy crown.
rameters were obtained with the assumption of an exchange
between a non-included (entry 1, Table 2) and an included
(entry 2, Table 2) species. In contrast with the EPR model
experiments performed with nitroxides 12 and 13 (Figure 8,
DaN =0.04 mT)[16] in the presence of TRIMEB, the observed
DaN =0.015mT is rather small, probably denoting a partly
included nitroxide fragment. Surprisingly, the relaxation pa-
rameters a, b and g are larger for the free spin-adduct PBN-
Spin-trapping experiments with Me2CD-PBN (7) and
Me3CD-PBN (10): In previous studies involving bimolecu-
lar complexes, we have shown that the PBN–superoxide ni-
troxide spin-adducts associate more strongly than the parent
PBN nitrones with methylated b-cyclodextrins.[11] Hence,
one might expect the spin-adducts obtained by spin-trapping
CÀ
CÀ
/
CO2 than for the supramolecular spin-adduct PBN-CO2
Chem. Eur. J. 2007, 13, 9344 – 9354
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9351

D. Bardelang, S. R. A. Marque et al.
TRIMEB, whereas one would expect a slower rotational
motion for the supramolecular system than for the free spin-
adduct. This might be explained by the strong solvation of
the carboxylate group making small molecules resemble
large ones, whereas inclusion into the cavity of TRIMEB in-
volves a partial desolvation of the spin-adduct. Hence, the
solvation being smaller for the supramolecular system, the
rotation motion is easier and the relaxation parameters are
smaller (entry 2, Table 2).
With concentrations of 10 in the 3.5to 25m m range, in-
tense EPR signals exhibiting strong distortion were record-
ed. The absence of changes in the EPR features with dilu-
tion was consistent with an intramolecular association pro-
cess. The strongly distorted signals are due to species experi-
encing strongly impeded rotational motion, specific of large
molecules (Figure 9).
cluded species exhibit the same trends as already observed
for the nitroxide model 11 and so need no further com-
ments. On the other hand, after 450 s, the global intensities
of the signals had decreased by ca. 40% in the case of the
CÀ
spin-adduct 10-CO2 and by ca. 60% in that of the bimolec-
CÀ
ular spin-adduct system PBN-CO2 /TRIMEB, but were
CÀ
almost unchanged in that of the free PBN-CO2 spin-
adduct. It is noteworthy that the separate evolution of the
two spin-adducts of 10 shows a faster decay for the non-in-
cluded species than for the partly included one, leading to
an artificial enrichment in the partly self-included species. In
this case, the presence of cyclodextrin—attached or not at-
tached—to the nitrone is detrimental to the persistence of
the formate spin-adduct. As the occurrence of fast side-reac-
tions between the nitroxide moiety and the CD moiety are
unlikely at room temperature, the faster decay may thus be
due to conformational changes facilitating, for example, a
CD-catalysed b-fragmentation (i.e., back reaction) of the
formyl group.[31] In the case of a bimolecular system, the
slight increase in aH,b (although a decrease would be expect-
ed, due to the decrease in spin density on the nitrogen
atom: DaN > 0), denotes conformational changes around
the nitroxide moiety. That is, the increase in aH,b may involve
a decrease in the <Hb-C-N-2p> dihedral angle and conse-
C
quently a decrease in the <2p-N-C-CO2 > dihedral angle,
À
À
which may favour overlapping between s*C CO2 and the
2p orbital containing the odd spin density, hence accelerat-
ing the b-fragmentation reaction. On the other hand, the
presence of the bulky CD and the positioning of the nitrone
fragment close to the cavity probably induced diastereose-
lective scavenging. Thus, because of the steric hindrance of
the CD, the nitroxide moiety probably adopted different
conformations for partly included and non-included forms—
as highlighted by the small changes in aH,b—leading to less
stable non-included species for the same reasons as de-
scribed for the bimolecular system. Nevertheless, Me3CD-
PBN (10) appears to be an efficient trap for the formyl radi-
cal, in spite of the significant nitrone self-association.
CÀ
Figure 9. Evolution of the Me3CD-PBN-CO2 spin-adduct EPR spec-
trum and calculated proportions between included and non-included ni-
troxides within the timescales: a) 90 s, b) 180 s, c) 300 s, d) 420 s, and
e) 540 s.
CÀ
O2 spin-trapping EPR experiments: In absence of CD, the
intensity of the EPR signal of the PBN superoxide (PBN-
OOH) spin-adduct is very weak and disappears in less than
one minute (Figure 10a).
On the other hand, a longer persistence and a higher
EPR signal intensity of the PBN-OOH spin-adduct in the
EPR data were therefore obtained by spectrum simula-
tions with the assumption of the existence of two species,
one with non-included and one with (partly) partly self-in-
cluded spin-adduct fragments. As in the cases of the bimo-
lecular system and the nitroxide model 11 (DaN =0.03 mT),
the small DaN =0.018 mT between these two species (en-
tries 3 and 4, Table 2) led us to assume an equilibrium be-
tween a non-included and a partly self-included species. The
spin-adduct fragment is probably lying in the methoxy
crown of the narrow ring. Such a form of partial inclusion is
due both to the bulkiness of the solvated COOÀ group and
to the disfavoured desolvation of the polar COOÀ moiety
impeding the inclusion of the spin-adduct fragment. An in-
crease in the polarity of the spin-adduct fragment is also
likely to weaken the stabilizing van der Waals interactions.
The relaxation parameters for the non-included and self-in-
presence of
a cyclodextrin derivative—TRIMEB (Fig-
ure 10b) or DIMEB (Figure 10d)—were observed for the bi-
molecular PBN/CD systems, with DIMEB affording the
more intense signal and the more persistent supramolecular
spin-adduct. The same trends are observed for the superox-
ide spin-adducts with the covalent molecules 10 (Figure 10c)
and 7 (Figure 10e). Moreover, the signals are more intense
for the monomolecular systems than for the bimolecular
ones, owing to the forced host–guest proximity, which fa-
vours the self-inclusion processes. However, whatever the
system—bimolecular or monomolecular–-and the type of
CD derivative—DIMEB or TRIMEB—the decay of the
9352
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354

Nitrones Bound to b-Cyclodextrins
FULL PAPER
species. This sharp contrast between the two spin-adducts 7-
CÀ
OOH and 10-CO2 is probably due both to the better com-
plexing properties of the Me2CD over Me3CD and to the
trapped radical. However, the self-included/non-included
spin-adduct fragment ratio is in favour of the non-included
species (75%), probably because of the disfavoured desolva-
tion of the polar and protic OOH group. Nonetheless, super-
oxide spin-adducts of 7 and 10 led to more persistent species
that were detectable up to 10 minutes after their generation
Spin-adduct reduction by sodium l-ascorbate: The persis-
tence of the superoxide spin-adducts obtained with 7 and 10
was measured in the presence of l-ascorbate (0.1 mm) to
mimic the physiological reductive conditions that destroy
the nitroxide. Despite the observed higher signal intensities
and longer persistence of the superoxide spin-adducts, no
significant lifetime enhancements were observed, meaning
that the resistance to l-ascorbate reductive conditions was
not higher for monomolecular systems than it was for bimo-
lecular systems. Improvements were attempted by the addi-
tion of extra DIMEB (50 mm) to the superoxide spin-ad-
ducts of 7 and 10 (see Supporting Information), leading to
an increase in the EPR spectrum intensity for the 7-OOH
spin-adduct and also to a slight one for the 10-OOH spin-
adduct. However, the half-life times of these species were
not longer than one minute, as observed without addition of
DIMEB.
As stated above, the non-included species was favoured
over the self-included species, indicating that the complexa-
tion rate was low and could probably not compete with the
high reduction rate. Consequently, the spin-adduct was
strongly exposed to fast reductants such as sodium l-ascor-
bate and no effect due to the inclusion process was ob-
served.
CÀ
Figure 10. Differences in EPR signal shapes and intensities of O2 spin-
adducts as a function of the spin-trap system: a) PBN alone (25m m),
b) PBN/TRIMEB (25/25 mm), c) 10 (25m m), d) PBN/DIMEB (25/
^
25m m), and e) 7 (25m m). ( )=three-lines EPR signal attributed to the
acyl-type decomposition product,[11] c=experimentally measured spec-
tra, g=simulated spectrum. Truncated cones are used to symbolize
CD derivatives, and egg shapes for PBN or spin-adduct.
spin-adduct generated a three-line signal ascribed to an
acyl-type nitroxide.[11] Such signals arose from the sequential
Hb abstraction and b-fragmentation of the hydroperoxide
group. Determination of reliable and accurate EPR parame-
ters by simulation of 10-OOH spin-adduct EPR spectra was
not possible because of the broad linewidths (DHpp ꢀ0.14–
0.18 mT). However, the large anisotropy of the signal at
high field underlines the presence of the self-included spin-
adduct fragment as already observed in the case of the ni-
troxide model 11. On the other hand, in the case of the
spin-adduct 7-OOH, simulations afforded clear evidence for
an exchange between two species. Furthermore, the value of
DaN =0.075mT is unambiguously larger than the value
DaN =0.054 mT for the PBN-OOH (aN =1.480 mT) and the
1:2 complex DIMEB/PBN-OOH/DIMEB (aN =1.426 mT),
and also strikingly larger than the DaN =0.018 mT for the
10-CO2H spin-adduct species. Consequently, such a large
DaN value suggests an equilibrium between non-included
(entry 1, Table 3) and deeply self-included (entry 2, Table 3)
Conclusion
The first grafting of a nitrone onto b-cyclodextrins has been
achieved under mild conditions that preserved the nitrone
functionality, affording the PBN-derived DIMEB and
TRIMEB derivatives as spin-traps for superoxide and
formyl radical spin-trapping. NMR studies of the CD-PBN
derivatives 7 and 10 showed that the nitrone moiety is in
each case strongly self-included in the corresponding cavity.
However, this inclusion process does not preclude efficient
spin-trapping of the superoxide and formyl radicals whatev-
er the spin-trap considered (7 or 10), as demonstrated by
EPR spectroscopy. The resulting spin-adducts even exhibit-
ed enhanced EPR signals in relation to the noncovalent
CD–nitrone systems or the nitrone alone used as spin-traps.
Moreover, the half-life times for both supramolecular super-
oxide spin-adducts were significantly increased, but no
effect on their bioreduction in the presence of l-ascorbate
was detected. These results demonstrate that cyclodextrin
grafting does not affect the spin-trapping properties of well
known spin-traps and can afford additional benefits due to
Table 3. EPR parameters for the exchange between the non-included
and the self-included conformers of the superoxide spin-adduct of 7 in
phosphate buffer (pH 7.4).
[a]
[a]
Entry
Species
aN [mT]
aH,b [mT]
n[a] [mT]
%
1
2
3
7-OOH[b,c]
7-OOH[b,d]
1.4150.32
1.340
0.815–
0.20
0.55
73
0.13
0.10
20
7
[f]
acyl nitroxide[e]
[a] Nitrogen aN and b-hydrogen aH,b hyperfine coupling constants, and re-
laxation parameter n given by the simulation program as in ref. [16].
[b] [7]=25m m. [c] Non-included spin-adduct fragment. [d] Self-included
spin-adduct fragment. [e] Degradation product. [f] No hydrogen in posi-
tion b; see text.
Chem. Eur. J. 2007, 13, 9344 – 9354
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
9353

D. Bardelang, S. R. A. Marque et al.
TRIMEB derivatives in which one 6-MeO has been replaced by a
PBN-derived moiety.
[14] D. Bardelang, J.-L. Clement, J.-P. Finet, H. Karoui, P. Tordo, J. Phys.
Chem. B 2004, 108, 8054–8061.
the proximal positions of the cyclodextrin hosts as supra-
molecular protectors.
[15] H. G. Aurich, Nitrones, Nitronates and Nitroxides, Wiley, New York,
1989.
Acknowledgement
[16] The resulting grafted nitroxide was demonstrated to self-associate in
the TRIMEB cavity as a model of a trapped isopropyl radical by
Me3CD-PBN (10): see D. Bardelang, A. Rockenbauer, L. Jicsinszky,
J.-P. Finet, H. Karoui, S. Lambert, S. R. A. Marque, P. Tordo, J. Org.
Chem. 2006, 71, 7657–7667. For other work on nitroxide spin-la-
belled cyclodextrins see a) R. M. Paton, E. T. Kaiser, J. Am. Chem.
Soc. 1970, 92, 4723–4725; b) G. Ionita, V. Chechik, Org. Biomol.
Chem. 2005, 3, 3096–3098; c) V. Chechik, G. Ionita, Org. Biomol.
Chem. 2006, 4, 3505–3510.
[17] H. Parrot-Lopez, H. Galons, S. Dupas, M. Miocque, G. Tsoucaris,
Bull. Soc. Chim. Fr. 1990, 127, 568–571.
[18] L. Jicsinszky, R. Ivꢂnyi, Carbohydr. Polym. 2001, 45, 139–145.
[19] R. S. Macomber, J. Chem. Educ. 1992, 69, 375–378.
[20] K. A. Connors, Binding Constants: the measurement of molecular
complex stability, Wiley, New York, 1987.
The financial support provided by the Conseil RØgional Provence Alpes
Côte dꢁAzur, the CNRS, the UniversitØ de Provence and TROPHOS
Company (Marseille) is gratefully acknowledged.
[1] a) W. Drçge, Physiol. Rev. 2002, 82, 47–95; b) J. Foreman, V. Dem-
idchik, J. H. F. Bothwell, P. Mylona, H. Miedema, M. A. Torres, P.
Linstead, S. Costa, C. Brownlee, J. D. G. Jones, J. M. Davies, L.
Dolan, Nature 2003, 422, 442–446; c) Y. Shibata, R. Branicky, I. O.
Landaverde, S. Hekimi, Science 2003, 302, 1779–1782.
[2] a) M. P. Mattson, Nature 2004, 430, 631–639; b) E. Gaggelli, H. Ko-
zlowski, D. Valensin, G. Valensin, Chem. Rev. 2006, 106, 1995–2044;
c) M. T. Lin, M. F. Beal, Nature 2006, 443, 787–795.
[3] S. Cuzzocrea, E. Mazzon, L. Dugo, R. Di Paola, A. P. Caputi, D. Sal-
vemini, FASEB J. 2004, 18, 94–101.
[4] J. H. J. Hoeijmakers, Nature 2001, 411, 366–374.
[5] a) M. Brownlee, Diabetes 2005, 54, 1615–1625; b) N. Houstis, E. D.
Rosen, E. S. Lander, Nature 2006, 440, 944–948.
[6] a) H. Sies, Angew. Chem. 1986, 98, 1061–1076; Angew. Chem. Int.
Ed. Engl. 1986, 25, 1058–1071; b) D. L. Korthuis, D. L. Carden,
D. N. Granger, Biological consequences of oxidative stress, Oxford
University Press, London 1992, p. 50; c) H. Wiseman, B. Halliwell,
Biochem. J. 1996, 313, 17–29; d) B. Kalyanaraman, Oxygen Radicals,
Disease and Disease Process (Ed.: C. E. Thomas), Harwood, Am-
sterdam, 1997.
[21] K. A. Connors, Chem. Rev. 1997, 97, 1325–1357.
[22] The H9 chemical shift was determined unambiguously at 7.50 ppm
by a HMQC sequence and by comparing the observed cross correla-
tions with those of the PBN in water. However a satisfactory
13C NMR spectrum of 7 could not be obtained in spite of a 50 mm
concentration and 3072 scans.
[23] M. V. Rekharsky, Y. Inoue, Chem. Rev. 1998, 98, 1875–1917.
[24] a) T. Haino, D. M. Rudkevich, J. Rebek, Jr., J. Am. Chem. Soc.
1999, 121, 11253–11254; b) S. M. Biros, J. Rebek, Jr., Chem. Soc.
Rev. 2007, 36, 93–104.
[25] H. J. Schneider, F. Hacket, V. Rꢃdiger, H. Ikeda, Chem. Rev. 1998,
98, 1755–1785.
[7] a) D. Salvemini, Z.-Q. Wang, J. L. Zweier, A. Samouilov, H. Macar-
thur, T. P. Misko, M. G. Currie, S. Cuzzocrea, J. A. Sikorski, D. P.
Riley, Science 1999, 286, 304–306; b) M. J. Davies, G. S. Timmins,
Electron Paramagn. Reson. 2000, 17, 1–42.
[8] a) J.-M. Lehn, Supramolecular Chemistry, VCH, Weinheim, 1995;
b) J. W. Steed, J. L. Atwood, Supramolecular Chemistry, Wiley-
VCH, London, 2000.
[9] H. Karoui, A. Rockenbauer, S. Pietri, P. Tordo, Chem. Commun.
2002, 3030–3031.
[10] H. Karoui, P. Tordo, Tetrahedron Lett. 2004, 45, 1043–1045.
[11] D. Bardelang, A. Rockenbauer, J.-P. Finet, H. Karoui, P. Tordo, J.
Phys. Chem. B 2005, 109, 10521–10530.
[12] D. Bardelang, A. Rockenbauer, H. Karoui, J.-P. Finet, I. Biskupska,
K. Banaszak, P. Tordo, Org. Biomol. Chem. 2006, 4, 2874–2882.
[13] DIMEB is an acronym used to describe 2,6-di-O-Me-b-cyclodextrin
and TRIMEB for the analogous 2,3,6-permethylated b-cyclodextrin.
In this paper we use Me2CD and Me3CD to describe DIMEB or
[26] A. Rockenbauer, L. Korecz, Appl. Magn. Reson. 1996, 10, 29–43.
[27] E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem. 2003,
115, 1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250.
[28] K. HØberger, S. KemØny, T. Vidoczy, Int. J. Chem. Kinet. 1987, 19,
171–181.
[29] P. Franchi, M. Lucarini, G. F. Pedulli, D. Sciotto, Angew. Chem.
2000, 112, 269–272; Angew. Chem. Int. Ed. 2000, 39, 263–266.
[30] S. Saudan, F. A. Dunand, A. Abou-Hamdan, P. Bugnon, P. G. Lye,
S. F. Lincoln, A. E. Merbach, J. Am. Chem. Soc. 2001, 123, 10290–
10298.
[31] The main decay process of the glutathiyl spin-adduct of PBN is the
fragmentation reaction regenerating the thiyl radical and the PBN.
This reaction is slowed down several times for the corresponding
spin-adduct of 7. Personal communication from Prof. E. G. Bagryan-
skaya (Novosibirsk).
Received: March 5, 2007
Published online: August 29, 2007
9354
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2007, 13, 9344 – 9354