Combining Magnetic Resonance Spectroscopies, Mass Spectrometry, and Molecular Dynamics: Investigation of Chiral Recognition by 2,6-di-O-Methyl-β-cyclodextrin

Article abstract of DOI:10.1021/ja049713y

EPR spectroscopy has been employed to investigate the formation of complexes between heptakis-(2,6-O-dimethyl)-β-cyclodextrin (DM-β-CD) and different enantiomeric pairs of chiral nitroxides of general structure PhCH₂N(O·)CH(R)R′. Accurate equilibrium measurements of the concentrations of free and included radicals afforded the binding constant values for these nitroxides. The relationship between the stereochemistry of the DM-β-CD complexes and the thermodynamics of complexation was elucidated by correlating EPR data with ¹H-¹H NOE measurements carried out on the complexes between DM-β-CD and the structurally related amine precursors of nitroxides. NOE data suggested that inclusion of the stereogenic center in the DM-β-CD cavity occurs only when the R substituent linked to the chiral carbon contains an aromatic ring. For these types of complexes, molecular dynamics simulation indicated that the depth of penetration of the stereogenic center into the cyclodextrin cavity is determined by the nature of the second substituent (R′) at the asymmetric carbon and is responsible for the observed chiral selectivity. Analysis of mass spectra showed that, for the presently investigated amines, electrostatic external adducts of CDs with protonated amines are detected by ESI-MS.

Full text of DOI:10.1021/ja049713y

Published on Web 03/12/2004
Combining Magnetic Resonance Spectroscopies, Mass
Spectrometry, and Molecular Dynamics: Investigation of
Chiral Recognition by 2,6-di-O-Methyl-â-cyclodextrin
Paola Franchi, Marco Lucarini,* Elisabetta Mezzina,* and Gian Franco Pedulli
Contribution from the Department of Organic Chemistry “A. Mangini”,
UniVersity of Bologna, Via San Donato 15, 40127 Bologna, Italy
Received January 16, 2004; E-mail: lucarini@alma.unibo.it; elisabetta.mezzina@unibo.it
Abstract: EPR spectroscopy has been employed to investigate the formation of complexes between
heptakis-(2,6-O-dimethyl)-â-cyclodextrin (DM-â-CD) and different enantiomeric pairs of chiral nitroxides of
general structure PhCH2N(O‚)CH(R)R′. Accurate equilibrium measurements of the concentrations of free
and included radicals afforded the binding constant values for these nitroxides. The relationship between
the stereochemistry of the DM-â-CD complexes and the thermodynamics of complexation was elucidated
by correlating EPR data with 1H-1H NOE measurements carried out on the complexes between DM-â-CD
and the structurally related amine precursors of nitroxides. NOE data suggested that inclusion of the
stereogenic center in the DM-â-CD cavity occurs only when the R substituent linked to the chiral carbon
contains an aromatic ring. For these types of complexes, molecular dynamics simulation indicated that the
depth of penetration of the stereogenic center into the cyclodextrin cavity is determined by the nature of
the second substituent (R′) at the asymmetric carbon and is responsible for the observed chiral selectivity.
Analysis of mass spectra showed that, for the presently investigated amines, electrostatic external adducts
of CDs with protonated amines are detected by ESI-MS.
Introduction
Several investigations have also explored the selective precipita-
tion of optically active compounds with cyclodextrins, and it
Cyclodextrins1 (CDs) are cyclic oligomers of 1,4-linked, R-D-
glucose monomers that have become the focus of intense study
by technologists interested in application of guest-host com-
plexation2 as well as by scientists interested in fundamental
issues of molecular recognition.3 Interest in these compounds
derives from the fact that they act as host molecules to form
inclusion complexes with a wide variety of guests. The
cyclodextrins exist as single enantiomers, with the consequence
that when they act as host molecules, interaction with a racemic
guest will lead to the formation of diastereomeric complexes
which may have different thermodynamic stability. The stereo-
selective complex formation of cyclodextrins has been applied
to separate enantiomers in planar chromatography (TLC),4 high
performance liquid chromatography (HPLC),5 gas-liquid-phase
chromatography (GLC)6 and capillary electrophoresis (CE).7
has been found that precipitation with cyclodextrins may
effectively lead to separation of some racemic compounds.8
Although chiral discrimination by cyclodextrins have been
investigated in great detail, only relatively limited efforts have
been devoted to understand the relationship between the
stereochemistry of the complex and the complexation thermo-
dynamic. By using computational techniques Lipkowitz and co-
workers showed that in permethylated â-cyclodextrin enantiod-
ifferentiating regions are at the interior rather than the exterior
of the macrocycle and that the chiral discriminating forces are
short-range dispersion forces.9
More recently, Rekharsky and Inoue investigated a wide
variety of chiral compounds by employing isothermal titration
calorimetry.10 These papers represent the most important attempt
to determine a direct relationship between thermodynamic
parameters and chiral recognition by CDs.11 Although a general-
ized pictured was not obtained, an important conclusion was
that a direct correlation between the mode of penetration and
chiral recognition is a fairly general rule applicable to a variety
(1) For an introduction and a general overview of cyclodextrin chemistry,
see: Szejtli, J. Chem. ReV. 1998, 98, 1743-1753.
(2) Hedges, A. R. Chem. ReV. 1998, 98, 2035-2044.
(3) ComprehensiVe Supramolecular Chemistry, Vol. 3: Cyclodextrins; Szejtli,
J., Osa, T., Eds.; Pergamon: Oxford, UK, 1996.
(4) For reviews see: (a) Ward, T. J.; Armstrong, D. W. J. Liq. Chromatogr.
1986, 9, 407-423. (b) Martens, J.; Bhushan, R. Int. J. Pept. Protein Res.
1989, 34, 433-444. (c) Martens, J.; Bhushan, R. J. Pharm. Biomed. Anal.
1990, 8, 259-269. (d) Bhushan, R.; Joshi, S. Biomed. Chromatogr. 1993,
7, 235-250.
(8) Harata, K. Chem. ReV. 1998, 98, 1803-1827.
(9) (a) Lipkowitz, K. B.; Pearl, G.; Coner, B.; Peterson, M. A. J. Am. Chem.
Soc. 1997, 119, 600-610. (b) Lipkowitz, K. B.; Coner, B.; Peterson, M.
A. J. Am. Chem. Soc. 1997, 119, 11 269-11 276.
(5) For review, see: Li, S.; Purdy, W. C. Chem. ReV. 1992, 92, 1457-1470.
(6) For reviews see: (a) Armstrong, D. W.; Han, S. CRC Crit. ReV. Anal.
Chem. 1988, 19, 175. (b) Schurig, V.; Nowotny, H.-P. Angew. Chem., Int.
Ed. Engl. 1990, 29, 939-957.
(7) For reviews see: (a) Vespalec, R.; Bocek, P. Electrophoresis 1994, 15,
755-762. (b) Nishi, H.; Terabe, S. J. Chromatogr. A 1995, 694, 245-
276.
(10) (a) Rekharsky, M.; Inoue, Y. J. Am. Chem. Soc. 2000, 122, 4418-4435.
(b) Rekharsky, M. V.; Inoue, Y. J. Am. Chem. Soc. 2000, 122, 10 949-
10 955. (c) Hembury, G.; Rekharsky, M.; Nakamura, A.; Inoue, Y. Org.
Lett. 2000, 2, 3257-3260. (d) Rekharsky, M. V.; Inoue, Y. J. Phys. Org.
Chem. 2001, 14, 416-424. (e) Rekharsky, M. V.; Inoue, Y. J. Am. Chem.
Soc. 2002, 124, 813-826.
9
10.1021/ja049713y CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 4343-4354
4343

A R T I C L E S
Franchi et al.
Chart 1
of cyclodextrin complexations. In particular, it has been shown
that the spatial interaction between chiral guest and CD depends
on the depth of guest penetration and therefore on the position
of the most hydrophobic group around the chiral center of the
guest.
However, these measurements are severely hampered by small
enthalpic differences, which additionally are compensated by
counteracting entropic contributions, leading to rather moderate
differences in affinity. Moreover, in some cases, the cyclodex-
trins form complexes with higher stoichiometries that force one
to explore a limited concentration range in order to obtain only
1:1 complexes.
Even if the complexation of a chiral guest with CDs gives
rise to diastereomeric pairs,12 EPR spectroscopy has only been
rarely used to investigate the complexation of chiral radical
species by CDs.13 Recently, we have reported the EPR observa-
tion of the complexes between CDs and dialkyl nitroxide
radicals in water.14,15 Due to the shorter time scale of EPR
spectroscopy compared to that of complexation processes, EPR
spectra resulting from the superposition of separate signals from
free and included species are usually observed when using free
radicals as guest molecules. In particular, with the tert-butyl
benzyl nitroxide radical the formation of an inclusion complex
with CDs produces in the EPR spectrum both a decrease of the
overall splitting and significant differences in the resonance
fields of the MI(2Hâ) ) (1 lines, compared to the free species
in water. These large spectral changes arise both from the
decrease of the nitrogen hyperfine splitting, a(N), induced by
the less polar environment of the CD host cavity, and from the
strong reduction of the benzylic protons coupling, a(2Hâ), due
to conformational changes occurring upon complexation.16
Because of this favorable feature, the determination of the
association constants is easily achieved by double integration
of the EPR lines.
In contrast to EPR spectra, NMR spectra obtained from most
CD complexes represent concentration-weighted averages since
exchange between the free and complexed guest molecule is
usually fast in the NMR time scale. Therefore, to determine
association constants with CDs, non linear fitting of chemical
shift changes as a function of concentration (the so-called NMR
titration) is usually performed.17
On the basis of this, we decided to employ EPR spectroscopy
to obtain accurate equilibrium constants for the formation of
complexes between heptakis-(2,6-O-dimethyl)-â-CD (DM-â-
CD) and different enantiomeric pairs of chiral nitroxides (2a-
2i). The selected chiral nitroxides were chosen because they
maintain both the favorable EPR features found in the related
benzyl tert-butyl nitroxide and can be obtained easily as pure
optical isomers by oxidation of the corresponding chiral amine
precursors 1a-1i. It will be shown that the reported chiral
nitroxides are particularly suitable probes for the study of
inclusion by cyclodextrins (and probably by other chiral
complexing agents) by EPR spectroscopy, because their dia-
stereomeric complexes are characterized by spectroscopic
parameters differing significantly more than for any other type
of probe previously used for this purpose. In addition, small
differences in the affinity of the guest for the CDs cavity are
reflected by a significant change of the EPR line intensity and
formation of complexes with different stoichiometry give rise
to additional EPR lines easily distinguishable from those of 1:1
complexes.15
To fully clarify the relationship between the stereochemistry
of the CD complexes and the complexation thermodynamics,
(11) Other papers covering this aspect have appeared in the literature: (a) Cooper,
A.; MacNicol, D. D. J. Chem. Soc., Perkin Trans. 2 1978, 760-763. (b)
Harada, A.; Saeki, K.; Takahashi, S. Carbohydr. Res. 1989, 192, 1-7. (c)
Barone, G.; Castronuovo, G.; Di Ruocco, V.; Elia, V.; Giancola, C.
Carbohydr. Res. 1989, 192, 331-341. (d) Schmidtchen, F. P. Chem. Eur.
J. 2002, 8, 3522-3529.
1
EPR data were correlated with H-1H NOE measurements17
carried out on the complexes between CDs and the structurally
related amine precursors 1a-1i.18 Geometries inferred from
NMR data were clarified by molecular dynamics simulations.
Finally, the data obtained from magnetic resonance experiments
were compared to those extracted from ESI-MS spectra.
(12) Schuler, P.; Schaber, F.-M.; Stegmann, H. B.; Janzen, E. Magn. Reson.
Chem. 1999, 37, 805-813.
(13) (a) Michon, J.; Rassat, A. J. Am. Chem. Soc. 1979, 101, 995-996. (b)
Kotake, Y.; Janzen, E. G. J. Am. Chem. Soc. 1992, 114, 2872-2874.
(14) (a) Lucarini, M.; Roberts, B. P. J. Chem. Soc., Chem. Commun. 1996,
1577-1578. (b) M. Lucarini, M.; Luppi, B.; Pedulli, G. F.; Roberts, B.
Chem. Eur. J. 1999, 5, 2048-2054.
(15) Franchi, P.; Lucarini, M.; Pedulli, G. F. Angew. Chem., Int. Ed. Engl. 2003,
42, 1842-1845.
(16) For other examples of application of benzyl tert-butyl nitroxide in
supramolecular environments, see: (a) Franchi, P.; Lucarini, M.; Pedulli,
G. F.; Sciotto, D. Angew. Chem., Int. Ed. Engl. 2000, 39, 263-266. (b)
Brigati, G.; Franchi, P.; Lucarini, M.; Pedulli, G. F.; Valgimigli, L. Res.
Chem. Intermed. 2002, 28, 131-141.
(17) For reviews, see: (a) Schneider, H. J.; Hacket, F.; Rudiger, V.; Ikeda, H.
Chem. ReV. 1998, 98, 1755-1785 and references therein. (b) Schneider,
H. J.; Yatsimirsky, A. Principles and Methods in Supramolecular Chemistry;
Wiley and Sons: Chichester, 2000.
Results
Preparation of the Optical Active Amines and Nitroxides.
Chart 2 summarizes the synthetic methods used to prepare the
optical pure amines. Briefly, the commercially available amino
acids, amino alcohols and methylbenzylamine were reacted with
(18) Lucarini, M.; Mezzina, E.; Pedulli, G. F. Eur. J. Org. Chem. 2000, 3927-
3930.
9
4344 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004

Recognition by 2,6-di-O-Methyl-â-cyclodextrins
A R T I C L E S
Chart 2
benzoyl chloride to afford the corresponding amides which were
then reacted with LiAlH4 or LiAlD4 to give the desired amines
in yield ranging from 60% to 70%. To obtain the O-methylated
derivatives (1h-1i) the corresponding amide alcohols were
reacted in two steps with NaH and (CH3)2SO4 before reduction.
Products deuterated at the aromatic benzyl ring were obtained
in a similar way by using d5-PhCOCl. In all cases, the
configuration at the chiral carbon in the starting amine was
completely maintained in the final products.
The optically pure nitroxides were generated directly inside
an EPR tube by oxidation of the corresponding chiral amines
with the magnesium salt of monoperoxyphthalic acid (MMPP)
in water (eq 1).
Figure 1. EPR spectra at 300 K of (a) racemic 2e in water. (b) (R)-2e in
the presence of DM-â-CD 22.5 mM. (c) (S)-2e in the presence of DM-â-
CD 22.5 mM. In red are shown the corresponding simulations obtained by
using the hfsc reported in Table 1.
EPR Spectra. Good EPR spectra of nitroxides 2a-2i, were
obtained by oxidation of amines 1a-1i (ca. 1.0 mM) with
MMPP (ca. 1.0 mM) in water at 300 K. As an example in Figure
1a is reported the spectrum obtained by oxidizing the racemic
mixture of amine 1e in water at 300 K. All spectra were easily
interpreted on the basis of the coupling of the unpaired electron
with nitrogen, with the single proton bound to the chiral carbon
and with the two diastereomeric benzylic protons (see Table
1). By recording the EPR spectra of the deuterated nitroxides
labeled at the benzylic protons it was possible, in all cases, to
assign the hydrogen hyperfine splitting constant (hfsc) to the
single â-proton.
It should be remarked that for a given couple of enantiomeric
amines, the oxidation of the two antipodes in an achiral medium,
such as water, gave rise to the same EPR spectrum.12
When the EPR spectra of radicals 2a-2i were recorded in
the presence of DM-â-CD, additional signals were observed
assigned to the radical included in the cavity of cyclodextrin,
in equilibrium with the free nitroxide (eq 2). As an example
Figure 1 reports the spectra of (R)-2e and (S)-2e recorded in
the presence of 22.5 mM DM-â-CD.
By increasing the absolute concentration of DM-â-CD, the
ratio between the included and the free species increased and
eventually the spectrum of the 1:1 inclusion complex became
dominant allowing for easy measurement of the spectroscopic
parameters for the included nitroxides (see Table 1).
With heptakis-(2,3,6-O-trimethyl)-â-CD (TM-â-CD) no ad-
ditional spectral lines were observed beside those due to the
nitroxide radical in water solution, thus indicating that this
cyclodextrin is not able to complex nitroxides 2a-2i.
Inspection of Table 1 clearly indicates that the inclusion by
the DM-â-CD chiral cavity of the (R) and (S) antipodes gave
rise to diastereomeric inclusion complexes characterized by
different EPR spectroscopic parameters. Actually, both a(N) and
a(2Hbenzylic) change significantly upon inclusion into the DM-
â-CD cavity, thus giving rise to significant differences in the
resonance fields for the MI(2Hâ) ) (1 lines of the complexed
and free nitroxides. Due to this favorable feature, the ratio
between the concentrations of the two species could be easily
obtained by measuring the relative intensity of their low field
EPR lines (see Figure 2). By plotting the ratio between the
concentrations of the complexed and free species as function
of the DM-â-CD concentration in water (see Figure 3) we could
9
J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004 4345

A R T I C L E S
Franchi et al.
Table 1. Hyperfine Splitting Constants for Free and Included
Nitroxides (1:1 complex) in Water at 300 K
Hpf splitting (G ) 0.1 mT)a
radical
enantiomer
host
H−C*
N−CH −Ph
N
2
2a
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
(S)
(R)
racemic
(S)
(R)
3.71
3.55
3.63
3.28
3.34
2.76
2.69
2.68
2.69
2.70
2.29
2.21
2.01
6.13
7.00
6.80
7.04
7.18
2.24
1.80
1.72
1.94
1.85
2.16
1.29
1.46
7.79
8.66
8.59
8.73
8.63
2.76
2.45
2.44
2.56
2.57
5.50
5.95
5.66
10.82, 11.30
7.34, 10.53
7.33, 10.48
6.74, 10.74
6.64, 11.01
11.45, 11.56
7.69, 11.29
7.63, 11.53
7.43, 11.27
6.97, 12.03
10.98, 11.96
7.24, 12.63
7.06, 13.60
10.39, 11.42
9.24, 9.67
16.21
15.54
15.54
15.70
15.70
15.95
15.34
15.36
15.44
15.49
15.99
15.38
15.55
15.91
15.45
15.53
15.50
15.55
15.80
15.44
15.42
15.59
15.54
15.97
15.79
15.73
16.49
15.66
15.62
15.80
15.77
16.05
15.39
15.40
15.52
15.56
16.12
15.65
15.72
DM-â-CD
DM-â-CD
â-CD
â-CD
2b
DM-â-CD
DM-â-CD
â-CD
â-CD
2c
DM-â-CD
DM-â-CD
2d
DM-â-CD
DM-â-CD
â-CDb
9.65, 10.22
9.63, 9.85
â-CDb
9.88, 10.29
11.63, 12.75
10.97, 11.01
9.98, 12.52
10.82, 10.85
10.43, 12.00
11.69, 12.21
10.22, 12.04
9.55, 12.33
10.62, 11.34
8.93, 9.69
2e
DM-â-CD
DM-â-CD
â-CD
Figure 2. 300 K EPR spectra of (R)-2b in water in the presence of variable
amount of DM-â-CD. The dotted lines correspond to the resonance field
of the free and complexed species.
â-CD
2f
DM-â-CD
DM-â-CD
2g
DM-â-CD
DM-â-CD
â-CD
8.62, 9.70
8.75, 10.09
8.68, 9.90
â-CD
2h
2i
11.55, 11.74
8.10, 10.96
7.67, 11.47
7.80, 10.86
7.32, 11.46
10.28, 11.75
9.00, 10.18
9.58, 10.48
DM-â-CD
DM-â-CD
â-CD
â-CD
DM-â-CD
DM-â-CD
a The reported Hpf splitting constants refers to the nucleus indicated in
bold. b T ) 292 K.
determine the corresponding complex stability constants, K2 (see
Table 2). The reported uncertainties in the equilibrium constants
were calculated as twice the standard error of the regression
coefficient attained by analyzing up to 10 different EPR spectra
recorded in the presence of variable concentrations of DM-â-
CD. The uncertainties in the observed equilibrium constants are
less than 4% except for the derivative 2f for which the similarity
of the spectroscopic parameters of the free and included species
did not allow us to obtain well resolved spectra.
Figure 3. Plot of the concentration ratio between the 1:1 complex and the
free nitroxide versus the concentration of DM-â-CD in water at 300 K for
(S)-2g and (R)-2g.
representative nitroxides in the presence of the unmodified â-CD
(see Table 1). Nitroxides having a second aromatic ring at R
substituent (2d, 2e, and 2g), in the presence of â-CD at
concentrations close to the solubility limit (13.0 mM), showed
EPR spectra containing additional lines19 beside those of the
free and 1:1 complexed species (see the Supporting Information).
These lines were attributed to a third radical species identified
as an inclusion complex between one guest radical and two
â-cyclodextrin units on the basis of experimental evidence
similar to those reported previously for the dibenzyl nitroxide.15
Because the driving force responsible for the â-CD dimer
stabilization is the formation of intermolecular hydrogen bonds
between the secondary hydroxyl groups of the two larger CD
rings, the different behavior observed when using 2,6-O-
dimethyl â-cyclodextrin as host is presumably due to the absence
of some of the hydroxylic hydrogens. The results obtained with
â-CD illustrates the ease by which complexes with stoichiom-
The enantioselectivity (ES) of DM-â-CD toward a given
nitroxide enantiomeric couple was estimated quantitatively from
the relative complex stability constant by means of eq 3
|K2(S) - K2(R)|
K2(S) + K2(R)
Enantioselectivity (ES) )
× 100 (3)
Further increase of DM-â-CD concentration (up to 0.1 M)
did not lead to the appearance of any new EPR signals. This
experimental observation was taken as evidence that DM-â-
CD forms inclusion complexes with the investigated nitroxides
with a unique geometry from the interaction between a single
molecule of the guest and one host (1:1 complex). For
comparison purposes, we also recorded the EPR spectra of some
(19) Data to be published.
9
4346 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004

Recognition by 2,6-di-O-Methyl-â-cyclodextrins
A R T I C L E S
Table 2. Complex Stability Constants K2 at 300 K for the
Inclusion by DM-â-CD
0.17 ppm, respectively, in the presence of equimolar amounts
of the guest molecules. This behavior has been taken as an
indication of the occurrence of a “host-guest” interaction.17
The formation of the DM-â-CD-guest complex is also confirmed
by the observation of significant signal shifts of the amino guest
protons. In no cases, however, did complexation by the
cyclodextrin cavity produce the splittings of the aromatic protons
of the N-benzylic substituent of the guest.
Heptakis-(2,3,6-O-trimethyl)-â-CD (TM-â-CD) was treated
with amines 1a-1i under the same experimental conditions for
comparison. No significant spectral shifts of the cyclodextrin
inner protons were observed, thus indicating that no complex-
ation of amines 1a-1i takes place at the cyclodextrin concentra-
tion employed.
Being interested in comparing the enantioselectivity of DM-
â-CD toward nitroxides with respect to that of the parent amine
measured by EPR, we tried to determine the affinity constants
of some representative amines by NMR titrations.17 Contrary
to what was observed by EPR spectroscopy, the experimental
NMR spectra represent the concentration-weighted average of
the spectra of the amine in water and of the amine included in
the CD cavity. Therefore, the association constant could be
calculated by non linear fitting of the concentration dependence
of the δ shift values of the guest or the host protons. To obtain
accurate measurements of the affinity constants the measure of
the chemical shift value corresponding to 100% complexation
(complexation induced shift, CIS) is generally required but this
is not always possible due to signal overlapping and formation
of complexes with different stoichiometry. Moreover, recording
of many 1H NMR spectra is required in order to have relatively
small uncertainties, which in turn would lead to a very time-
consuming determination of the equilibrium constant.
For these reasons, we decided to investigate only one
enantiomeric pair of amines (R)-1g and (S)-1g, which by
oxidation give aminoxyls displaying the highest enantiomeric
discrimination (13.42%) toward DM-â-CD. These amines were
also chosen because they undergo the most substantial signal
shifts during complexation.
guest
K /M-1
ES/%
2
(S)-2a
(R)-2a
(S)-2b
(R)-2b
(S)-2c
(R)-2c
(S)-2d
(R)-2d
(S)-2e
(R)-2e
(S)-2f
(R)-2f
(S)-1g
(R)-1g
(S)-2g
(R)-2g
(S)-2h
(R)-2h
(S)-2i
249 ( 9a
245 ( 10
277 ( 12
262 ( 11
355 ( 15
341 ( 13
244 ( 10
273 ( 11
751 ( 37
678 ( 34
560 ( 61
588 ( 59
810 ( 156b
662 ( 106b
1500 ( 64
1145 ( 48
440 ( 17
409 ( 15
459 ( 18
382 ( 15
0.81
2.78
2.01
5.61
5.11
2.43
10.05
13.42
3.65
9.16
(R)-2i
a Errors are twice standard errors of equilibrium constants. b Determined
1
by H NMR titrations.
etries higher than 1:1 can be distinguished by their EPR spectra
and lead us to the conclusion that these are not formed in the
presence of DM-â-CD.
With nitroxides containing a phenyl and a CH2OH (or CH2-
OCH3) substituent at the chiral carbon (2d and 2i) a marked
selective broadening of the Mi(2Hâ) ) +1, Mi(H*â) ) -1/2 or
Mi(2Hâ) ) -1, Mi(H*â) ) +1/2 lines was visible when their
EPR spectra were recorded in the presence of DM-â-CD at room
temperature (see Supporting Information). These line width
effects20 indicate that the rate of rotation about one of the two
N-CR bonds is significantly reduced with respect to the free
species when nitroxides 2d and 2i are included inside the DM-
â-CD cavity.21 Although a decrease of the intramolecular
flexibility upon inclusion by a model receptor has been already
observed in paramagnetic guests,22 to the best of our knowledge
this is the first example of an internal motion restriction in an
organic free radical included in a cyclodextrin cavity.
Titrations were carried out in D2O solutions containing CD3-
OD (10% v/v) at 25 °C by keeping the 1g concentration constant
(2.15 mM) and increasing gradually the amount of DM-â-CD.
All solutions were prepared in a 0.1 M phosphate buffer adjusted
at pH 7.4.
1
NMR Measurements. Room temperature H NMR spectra
of DM-â-CD were recorded in D2O solutions in the presence
of equimolar amounts of amines 1a-1i (10-20 mM). In some
cases, CD3OD (10% v/v) was added to the sample to improve
the amine solubility. The signals of the DM-â-CD recorded in
the presence of the guest showed some differences with respect
to those of the free CD. In particular, the internal H3 and H5
signals of the host underwent upfield shifts of about 0.11 and
Experiments were performed by following the chemical shift
change of the proton attached to the chiral carbon (this in the
free amines falls at 4.22 ppm and after 100% of complexation
is 0.30 and 0.26 ppm shifted toward lower frequencies for the
(S) and (R) enantiomers, respectively). The calculated binding
constants are K2(S) ) 810 ( 156 M-1, K2(R) ) 662 ( 106
M-1 leading to an ES ratio of 10.05% (see Figure 4). In the
present case application of the fitting procedure yields equilib-
rium constants with uncertainties (calculated as twice the
standard error) of about 16-18%. These large errors, which
are quite common when the CIS value is introduced as
adjustable parameters,17b reduce significantly the utility of NMR
titrations in evaluating small differences of affinity constants.
The solution structures of the complexes between DM-â-CD
(20) (a) Freed, J. H.; Fraenkel, G. K. J. Chem. Phys. 1963, 39, 326-348. (b)
Hudson, A.; Luckhurst, G. R. Chem. ReV. 1969, 69, 191-225.
(21) It is possible to reproduce the observed line width alternating effect by
considering the modulations of â hydrogen hyperfine splittings arising from
exchange between two conformations which differ in the average value of
cos2θ, where θ is the dihedral angle between a â C-H bond and the axis
of the 2pz orbital on N. As an example, the experimental spectrum of the
radical (R)-2d is well reproduced (see Supporting Information) by using
for each conformer two sets of three â-hydrogen couplings whose sum is
the same (26.7 G) and the averaged values are those reported in Table 1.
To correctly simulate the alternating line width effects for one conformer
the two CH2-Ph couplings must be larger and the C*-H coupling smaller
with respect to the other. However, determining the absolute values of these
couplings would require us to record spectra at temperatures lower than 0
°C to freeze the internal rotation around the N-C bond in the EPR time
scale, this being not possible in water solution.
1
and amines were investigated by H-1H NOE experiments in
(22) (a) Ja¨ger, M.; Stegmann, H. B J. Am. Chem. Soc. 1996, 35, 1815-1818.
(b) Ja¨ger, M.; Schuler, P.; Stegmann, H. B.; Rockenbauer, A. Magn. Reson.
Chem. 1998, 36, 205-210.
order to evaluate the geometry of the aggregates and to estimate
the position of the amine chiral center within the cyclodextrin
9
J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004 4347

A R T I C L E S
Franchi et al.
Table 3. Cross Peaks Observed in the ROESY Spectra of Some
Representative Amine/DM-â-CD Complexes in D2O at 298 K
(mixing times ) 200 ms)
guest protons
host protons
H5
guest
orientation of
H6 the complex
R′
N-benzyl
other
C*H
H3
1b-d2
Ho, Hm, Hp
++a +++ + N-benzyl in
b
Et
+
1d-d9 Ho, Hm, Hp
++ ++
+
-
C*-R in
C*-R in
C*Hc
1e-d9 Ho
++ ++
Hm
Hp
+
+++ +
-
+
+
-
-
-
-
-
-
-
C*H
+
-
CH2
1f-d4 d H4
H5
+
-
+
+
C*-R in
-
++
++
+
H6
H7
-
Figure 4. Plot of the complexation-induced chemical shift (CIS) of the
hydrogen bound to the chiral carbon for (S)-1g and (R)-1g versus the
concentration of DM-â-CD. The lines represent the theoretical dependence
of CIS on the DM-â-CD concentration calculated by taking into account
the formation of the inclusion complexes and were obtained by numerically
fitting the experimental data by introducing as adjustable parameters the
values of K2 and CIS at 100% of complexation.
+
Ho, Hm, Hp
++
-
C*Hc
CH2
c
1g-d7 Ho, Hm, Hp
1h-d2
++ +++ + C*-R in
C*Hc
CH3
++
Chart 3
Ho, Hm, Hp
+
+++ ++ N-benzyl in
a
C*H
Et
+
1i-d7 Ho, Hm, Hp
++ +++ + C*-R in
C*Hc
CH2OMec
c
OCH3
a +++, strong; ++, medium; +, weak; -, no effect. b A cross-peak
with H3 is recorded at higher mixing times. c The signal partially overlaps
with some signals of DM-â-CD. d Measurement were carried out by using
an amine and DM-â-CD concentration of 2.2 mM.
cavity. It should be pointed out that, for a given amine, each
couple of enantiomers gives rise to similar 1H-1H NOE
interactions.
Initially we studied the interactions of DM-â-CD with amines
1a-1c, which differ only in the length of the alkyl chain (R,
see Chart 1) bound to the chiral carbon. However the interpreta-
tion of the NMR spectra of the inclusion complexes was difficult
to achieve, due to the overlapping of the cyclodextrin H2-H6
proton signals (3.3-4.0 ppm) with the benzylic proton signals
and, to some extent, the CH2OH proton signals of the amines.
1
To overcome this problem, H-1H NOE interactions of the
corresponding deuterated amines (1a,c)-d4 and 1b-d2 (see Chart
1) were investigated in the presence of DM-â-CD. ROESY
spectra of water solutions of the above-mentioned deuterated
amines (15 mM) containing equimolar amounts of DM-â-CD
showed significant interactions (see Table 3) of the aromatic
protons of the guest molecule with the inner H3 (medium) and
H5 (strong) cyclodextrin protons as well as with the H6
(medium) protons located at the smaller edge of the macrocyclic
host (as an example see Figure 5). On the contrary only a weak
interaction of the R alkyl group with the cyclodextrin H3 protons
was observed. These results indicate that the inclusion com-
plexes experience the same type of NOE interactions despite
the different chain length of the R substituent bound to the chiral
center. Moreover, from the strong interactions between the
aromatic ring and the inner cyclodextrin protons, the geometry
Figure 5. Portion of the ROESY spectrum of an equimolar solution (15
mM) of DM-â-CD/1b-d2 in D2O at room temperature.
of the inclusion complexes can be inferred to be very similar
in all cases with the N-benzyl part of the molecule deeply
included inside the CD cavity. We call this orientation of the
complex as N-benzyl in (see Chart 3).
The amine 1h-d2, where the hydrophilic CH2OH group has
been converted to the corresponding CH2OCH3 group, showed
NOE interactions similar to those of the structurally related
aminoalchol 1b-d2. This result suggests that, also in this case,
inclusion of the guest species into the CD cavity occurs from
9
4348 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004

Recognition by 2,6-di-O-Methyl-â-cyclodextrins
A R T I C L E S
hydrogens with H3 protons of the host were found to have
different intensities; in particular the larger cross-peak volume
detected for Ho with respect to Hm and the absence of correlation
between Hp and H3 indicates that Hp is directed toward the
smaller primary rim of the cavity far from H3. Similar
considerations hold in the analysis of the interactions of the
aromatic protons with H5 and H6 of the host. Actually, the
intensities of the NOE interactions between Ho, Hm, Hp, and
H5, H6 (Hm > Ho > Hp) again indicate that the phenyl of the
benzylic group bound to the chiral carbon is included inside
the CD cavity. This conclusion is also confirmed by the presence
of NOE interactions of benzylic and H-C* protons with H3
only (data not shown).
An analysis of the geometry of the complex involving amine
1f-d4 was carried out on a 2.1 mM solution containing an
equimolar amount of DM-â-CD. In the spectrum of the complex
recorded at this concentration each aromatic proton of the indole
moiety shows distinct signals and the N-benzylic protons are
easily distinguishable. The ROESY data can be summarized as
follows: (a) protons H4 and H7 of the indolic group interact
with both H3 and H5 of the host; (b) the indolic H5 and H6
show cross-peaks only with H5 of the host; (c) the aromatic
protons of the N-benzylic moiety display correlations only with
the internal H3 of the host (see Table 3). These results are
consistent with a geometry in which the indolic fragment of
the amine is complexed by the cavity of the cyclodextrin and
the phenyl group is outside the wider rim.
Molecular Dynamics. To obtain a more detailed picture of
the geometry of the complexes investigated by NOE experi-
ments, stochastic dynamics (SD) simulations were performed
by using the AMBER* force field of Macromodel 7.0 program.
For isolated guests 1a-1i, a Monte Carlo conformational
search was carried out by rotating all C(sp3)-C(sp3), Ar-C(sp3),
O-C(sp3), and N-C(sp3) bonds in the molecule. The most
stable conformation found by this procedure was then used in
the dynamic simulation. All computations were performed on
the S enantiomer with the exception of amine (R)-1g.
The computational approach taken in this study is to dock
the guest molecule inside the host cavity, energy minimize the
complex, and carry out standard equilibrations and production
runs to derive averaged distances. To accomplish this, we began
with a 7-fold symmetric cyclodextrin with the guest molecule
docked in its interior. We selected a symmetric CD only as a
guide for the placement of the guest and with the understanding
that these molecules will deform upon geometry optimization.23
The origin of a Cartesian reference frame was placed at the
center of mass of the CD with the z axis aligned with the C7
symmetry axis of CD. A second reference frame was placed
on the guest molecule with the z axis passing through the C*-N
or PhCH2-N bond. The guest was translated along the z axis
in order to have the nitrogen atom located in the plane passing
through the seven acetal oxygens of the CD. For each complex
only the experimental orientations previously found by NOE
experiments were investigated (see Table 4). At the end of this
procedure, minimization of the ensemble CD-amine lead to the
starting geometry for stochastic dynamics (SD) simulations.
The simulations were run at 300 K with time steps of 1 fs
and an equilibrium time of 500 ps before each dynamic run.
Figure 6. Portion of the ROESY spectrum of an equimolar solution (1
mM) of DM-â-CD/1e-d9 in D2O/CD3OD (10:1 v/v) at room temperature.
the N-benzyl portion of the amine. However, the presence of
an additional cross-peak due to a NOE interaction between the
aromatic protons of (1h)-d2 and the 6-OCH3 protons of CD is
indicative of a more profound interaction of the aminoether with
the primary edge of the cavity when compared to the corre-
sponding amino alcohol 1b.
We then examined the amines containing two aromatic rings,
the N-benzylic one and that one bound to the asymmetric carbon
(1d, 1g, and 1i). To distinguish the aromatic protons of these
two groups, amines completely deuterated at the N-benzylic
substituent (1d-d9, 1g-d7, and 1i-d7) were employed. ROESY
experiments showed significant NOE interactions between the
phenyl ring protons of the guest and the H3 (strong), H5
(strong), and H6 (medium) protons of DM-â-CD, thus indicating
that complex formation occurs by the inclusion in the CD cavity
of the phenyl ring directly linked to the stereogenic center. We
call this orientation of the complex as C*-R in (see Chart 3).
This behavior is opposite to that found with amines 1a-1c and
1h where NOE data suggested that the N-benzyl part of the
molecule was complexed by the CD cavity.
Unfortunately, NMR data of the complex geometry concern-
ing compounds 1d, 1g and 1i did not provide information on
the extent of inclusion of each amine into the cavity. This is
due to the overlap of the signals for the aromatic ortho, meta,
and para protons, that did not allow us to compare the single
contributions of each kind of protons with the H3, H5, and H6
protons of DM-â-CD.
We also investigated the amino alcohol containing two
benzylic groups (1e). Again the N-benzylic deuterated product
(1e-d9) was used to identify unambiguously the geometry of
the complex. 1H NMR spectra of the solution containing
equimolar amounts of DM-â-CD and 1e-d9 showed separate
signals for each proton of the guest. In particular, the benzyl
and H-C* proton signals found under 3 ppm did not overlap
with the CD protons, whereas the phenyl ring showed complete
separation of the three distinct spin systems corresponding to
the ortho, meta, and para protons. This favorable feature allowed
us to assign unambiguously the most important NOE interac-
tions. In the portion of the ROESY spectrum shown in Figure
6, the cross-peaks correlating the aromatic ortho and meta
(23) (a) Lipkowitz, K. B. Chem. ReV. 1998, 98, 1829-1873. (b) Lipkowitz, K.
B. J. Org. Chem. 1991, 56, 6357-6367.
9
J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004 4349

A R T I C L E S
Franchi et al.
Table 4. Calculated Averaged Energies (kJ mol-1) and Distances
(Å) at 300 K for the Inclusion Complexes between DM-â-CD and
Amines 1a-1i
orientation
of complex
energy of
complexa
b
c
b
c
guest
dC*,H3
dC*,H5
dHp,H3
dHp,H5
(S)-1a
(S)-1b
(S)-1c
(S)-1d
(S)-1e
(S)-1f
(R)-1g
(S)-1h
(S)-1i
N-benzyl in
N-benzyl in
N-benzyl in
C*-R in
1319.7
1318.1
1338.2
1340.2
1335.0
1331.3
1354.1
1357.2
1368.0
3.63
1.70
2.54
1.78
1.44
2.54
0.01
2.43
0.93
6.13
4.20
5.04
4.28
3.94
4.94
2.51
4.93
3.43
4.60
4.63
2.10
2.13
C*-R in
C*-R in
C*-R in
N-benzyl in
C*-R in
5.35
4.69
2.85
2.19
a Potential energy of the entire complex averaged over the simulation
time. b Distances to the plane defined by all H3 of DM-â-CD. c Distances
to the plane defined by all H5 of DM-â-CD.
The total simulation time was set to 10 000 ps in order to achieve
full convergence.9a During the simulation the distance between
the chiral center of the guest and the H3 and H5 protons of CD
were monitored. In the case of amines having a phenyl or benzyl
group linked to the chiral center, the distances between the para-
hydrogen of the guest and the H3 and H5 CD protons were
also monitored.
As a measure of the degree of inclusion of the guests we
chose the distances from the chiral center or from the para-
hydrogen of the R group to the planes defined by all H3 and
H5 protons of DM-â-CD within a particular complex. The
results are summarized in Table 4.
Figure 7. Clustered molecular display. Dynamics of the complex of DM-
â-CD with 1c (a) and 1g (b). The drawings include only 5 structures that
refer to the time interval of simulation between 5000th to the 6000th ps.
Hydrogen atoms have been omitted for clarity.
Mass Spectra. The chiral discrimination of amines 1a-1i
by DM-â-CD has been also investigated by using the mass
spectrometry (MS)-enantiomer labeled guest method.24 This
method consists of using an isotopically labeled enantiomer of
a selected guest to distinguish the diasteromeric host-guest
complex ions in a mass spectrum. A 1/1 mixture of a labeled
and an unlabeled guest enantiomer (i.e., R-dn + S or R + S-dn)
is complexed with a target chiral host. The enantioselectivity
of the host DM-â-CD toward a given racemic mixture is
estimated quantitatively from the relative peak intensity value
(IR/IS-dn or IS/IR-dn) of the two host-guest diastereomeric
complex ion peaks in the mass spectrum. In our case, electro-
spray ionization mass spectrometry (ESI-MS) was employed.
A 1:1 racemic solution of enantiomer guests was prepared
by mixing an equal amount of a water/ACN solution (2/1 v/v)
of each enantiomer (10 µM) in the presence of DM-â-CD 40
µM. A typical ESI-MS spectrum recorded in the region 1542-
1556 m/z units for the mixture (R)-1g and (S)-1g-d7 is shown
in Figure 8.
Figure 8. Positive ESI spectra of water/ACN solution (2/1) containing an
equimolar amount of (R)-1g and (S)-1g-d7 (15 µM) in the presence of DM-
â-CD 40 µM.
Analyses of the ESI-MS peak intensities led to an ES ratio
smaller than 1% for all investigated amines (1a-1i). The
apparent absence of chiral discrimination resulting from ESI-
MS experiments contrasts with the enantioselectivity ratios
found when using magnetic resonance spectroscopies (see
discussion).
â-CD and TM-â-CD for comparison. The ESI-MS spectra
showed besides the signal of the DM-â-CD-guest complex also
a strong signal due to the TM-â-CD-guest complex (see
Supporting Information).
We also recorded ESI-MS spectra of amine 1g in a water/
ACN solution (2/1 v/v) containing an equimolar mixture of DM-
Discussion
(24) (a) Fales, H. M.; Wright, G. J. J. Am. Chem. Soc. 1977, 99, 2339-2340.
(b) Sawada, M.; Takai, Y.; Yamada, H.; Hirayama, S.; Kaneda, T.; Tanaka,
T.; Kamada, K.; Mizooku, T.; Takeuchi, S.; Ueno, K.; Hirose, K.; Tobe,
Y.; Naemura, K. J. Am. Chem. Soc. 1995, 117, 7726-7736. (c) Sawada,
M.; Shizuma, M.; Takai, Y.; Adachi, H.; Takeda, T.; Uchiyama, T. Chem.
Commun. 1998, 1453-1454. (d) Sawada, M. Mass Spectrom. ReV. 1997,
16, 73-90. (e) So, P. M.; Wan, T. S. M.; Dominic Chan, T.-W. Rapid
Commun. Mass Spectrom. 2000, 14, 692-695.
EPR Parameters. Inspection of Table 1 shows that the value
of a(N) decreases significantly when the aminoxyl is located
in the less polar environment of the host cavity. This reduction
is in the range from -0.21 G for the indolyl derivative (2f) to
-0.85 G for 2g. The reduction of a(N), reported in previous
9
4350 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004

Recognition by 2,6-di-O-Methyl-â-cyclodextrins
A R T I C L E S
Chart 4
rotamers. Consequently, the methylene protons will be magneti-
cally nonequivalent and the energies associated with the two
rotamers will likewise be different. When the interconversion
between the two nonequivalent conformers is fast on the EPR
time scale the observed â-hydrogen splittings will be time
averaged over the residence time in each conformation. Because
the two conformations have different energies the two splittings
will be different, their values being closer to those of the favored
conformer.27
On the basis of these considerations the increased magnetic
non equivalence of the benzylic protons observed in the N-benzyl
in complexes should be attributed to an increase of the energy
difference for the two rotamers when the nitroxides are included
by DM-â-CD (Chart 4, equation b). This means that the
differences in the two rotamers concerning the spatial orientation
of the two â-hydrogens with respect to both the phenyl ring
and the C* substituents are enhanced when the benzylic aromatic
ring is included in the DM-â-CD cavity.
It should be pointed out that the different behavior of the
N-benzyl in and C*-R in complexes originates from confor-
mational effects rather than from a different affinity for the
cyclodextrin cavity of the related nitroxides. Thus, a large
variation of the spectroscopic parameters when passing from
the free to the included species does not depend on the affinity
of the guest for the CD cavity.29
Also, the variation of the hyperfine splitting of the hydrogen
bound to the chiral center a(C*-Hâ) observed when passing
from the free to the complexed species, depends to some extent
on the nature of the inclusion complexes. With nitroxides
forming N-benzyl in complexes an averaged decrease of 0.16
G is observed by passing from the free to the complexed radical
while with nitroxides giving rise to C*-R in complexes a larger
average difference (0.55 G) is found. Since the value of the
coupling for the single â-hydrogen depends on the conformation
adopted around the N-C* bond, inclusion of the chiral part of
the molecule has a larger influence on its value.
A final consideration concerns the value of hyperfine splitting
of the single proton bound to the chiral carbon which is much
larger in nitroxides where the R substituent is a phenyl group
(2d, 2g, and 2i). For a methyne group the lowest energy
conformation, i.e., that one in which the carbon-hydrogen bond
is eclipsed by the benzylic group, is characterized by a small
value of the Hâ splitting (Chart 5, structure a). With nitroxides
where RdPh, the conformation having the phenyl group eclipsed
by the pz orbital on the nitrogen atom (Chart 5, structure b)
seems to be important, thus leading to the higher value of the
Hâ splitting observed experimentally.
studies carried out with different nitroxide radicals14,15,25 has
been attributed to the larger weight of the nitroxide mesomeric
forms in media of low polarity in which the unpaired electron
is localized on the oxygen rather than on the nitrogen atom.
If we consider the nitroxides forming an inclusion complex
with the C*-R in orientation and take the calculated distances
from the chiral center to the planes defined by all H3 (see Table
4) as a measure of the degree of inclusion (a larger inclusion
corresponds to a more hydrophobic environment), an acceptable
correlation (r ) 0.94) is obtained between this distance and the
reduction of the nitrogen splitting (a(N)water - a(N)CD) measured
by EPR (see Supporting Information).
The value of the hyperfine splitting constant at the â-protons
in alkyl nitroxides is sensitive to conformational changes and
depends on the spin population in the 2pz orbital of nitrogen
(FN) and on the dihedral angle (ϑ) between the symmetry axis
of this orbital and the N-C-Hâ plane according to the Heller-
McConnel equation26
a(Hâ) ) FN(AN + BN cos2ϑ )
(4)
The nitroxides investigated in this work have two types of
â-protons: the two benzylic protons and the single proton bound
to the chiral carbon. As far as the benzylic protons are
concerned, analysis of the corresponding hyperfine splitting
constants allowed us to differentiate the behavior of the nitroxide
forming N-benzyl in complexes from those giving rise to C*-R
in complexes. In the former complexes (2a-2c, 2h), the
difference in benzylic couplings (∆a2Hâ) is much larger in the
radical included in the DM-â-CD cavity than in the free
nitroxide (in water the average ∆a2Hâ is 0.44 G, whereas in the
complexes with DM-â-CD is 4.05 G; see Table 1). On the other
hand, nitroxides (2d-2g, 2i) affording C*-R in complexes, are
characterized by differences in the value of the two benzylic
couplings similar to those observed in the free paramagnetic
species (in water the average ∆a2Hâ is 0.97 G, whereas in the
complexes with DM-â-CD it is 1.16 G; see Table 1).
Comparison between Amine and Nitroxides Guests. To
correlate the geometries of the amine/DM-â-CD complexes
In the investigated nitroxide radicals, the lower energy
conformations around the PhCH2-N bond are those where the
chiral center avoids the phenyl group (see Chart 4, equation a).
Due to the presence of the carbon chiral center, the spatial
orientation of the two â-hydrogens with respect to both the
phenyl ring and the C* substituents will differ for the two
(27) It should be pointed out that even in the case that all conformations were
equally populated, magnetic nonequivalence would still be observed since
in each conformer the two methylene splittings can be coincident only by
accident. The residual nonequivalence still present when all conformations
are energetically degenerate, is called intrinsic nonequivalence. We have
recently demonstrated that the magnetic nonequivalence of diastereomeric
â-protons in the EPR spectra mainly arise from the population difference
of the various conformations.28
(28) Franchi, P.; Lucarini, M.; Pedulli, G. F.; Bandini, E. J. Chem. Soc., Chem.
Commun. 2002, 560-561.
(25) (a) Kotake, Y.; Jansen, E. J. Am. Chem. Soc. 1988, 110, 3699-3701. (b)
Kotake, Y.; Jansen, E. J. Am. Chem. Soc. 1989, 111, 5138-5140. (c)
Fathallah, M.; Fotiadu, F.; Jaime, C. J. Org. Chem. 1994, 59, 1288-1293.
(d) Pe´rez, F.; Jaime, C.; Sa´nchez-Ruiz, X. J. Org. Chem. 1995, 60, 3840-
3845.
(29) This behavior has already been observed by NMR spectroscopy. See for
examples: (a) Dodziuk, H.; Ejchart, A.; Lukin, O.; Vysotsky, M. O. J.
Org. Chem. 1999, 64, 1503-1507. (b) Barretta-Uccello, G.; Balzano, F.;
Caporusso, A. M.; Iodice, A.; Salvadori, P. J. Org. Chem. 1995, 60, 2227-
2231. (b) Barretta-Uccello, G.; Balzano, F.; Caporusso, A. M.; Salvadori,
P. J. Org. Chem. 1994, 59, 836-839.
(26) Heller, C.; McConnell, H. M. J. Chem. Phys. 1960, 32, 1535-1539.
9
J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004 4351

A R T I C L E S
Franchi et al.
Chart 5
the loss of hydrogen bonding) as indicated by the larger affinity
constants found by EPR for the corresponding nitroxide.
However, poor enantiodiscrimination is again observed due to
the small interaction between the stereogenic center of the guest
and the chiral group of the CD-cavity as suggested by the
calculated distance between the chiral center and the plane
containing the cyclodextrin H3.
We can conclude that the poor enantiodiscrimination of
amines and nitroxides forming N-benzyl in complex is due to a
favorable interaction of the host with the benzylic phenyl ring
which is far from the stereogenic center. Thus, any change on
the alkyl substituents bound to the stereogenic carbon is not
able to affect substantially the chiral recognition process.
determined by NMR or computed by MD experiments, with
the enantiomeric discrimination of DM-â-CD toward the
corresponding nitroxides obtained by EPR spectroscopy, we
point out that the structure of the inclusion complexes is similar
for the nitroxide radicals and their amine precursors. This is
true for the DM-â-CD complexes of some achiral secondary
amines and of the corresponding nitroxides, as it has been
demonstrated previously by us when comparing NMR18 and
EPR14b data.
In the series of products investigated here, the complexes of
compounds 1 and 2 have closely related structures. Support for
this assumption comes from the similar enantioselectivity ratio
measured by NMR for the amine 1g (ES ) 10.05%) and by
EPR for the corresponding nitroxide 2g (ES ) 13.42%). This
result, besides suggesting a similar geometry of inclusion for
the two guests, indicates that the change of hybridation on the
nitrogen atom on passing from 1 (sp3) to 2 (sp2) is not an
important factor for chiral discrimination.
Additional evidence in this direction comes from the analysis
of the variation of hfsc of the benzylic protons when the
nitroxides are included inside the CD cavity. With amines
forming N-benzyl in complexes the corresponding nitroxides
show a large increase in the difference of the two â-splittings.
On the other hand, amines forming C*-R in complexes give
rise by oxidation to nitroxides that show small variations in the
benzylic hfsc upon inclusion. These conclusions strongly support
the assumption that the structure of the amine/DM-â-CD
complex is likely to also represent the association mode of the
nitroxide radicals.
Geometry of the Inclusion Complexes. We have previously
emphasized that the same guest can give rise to two type of
complexes: those having the N-benzyl side of the guest inside
the cyclodextrin cavity (N-benzyl in) and those in which the
portion of the guest included inside the cavity bears the chiral
center (C*-R in).
Calculated molecular models for complexes belonging to the
first group (1a, 1b, 1c, and 1h) show that the guest is inserted
with the N-benzyl group in the smaller primary rim of the CD
and the alkyl group R placed near the larger secondary rim of
the host (see Figure 7a). In this arrangement, the stereogenic
center interacts weakly with the cyclodextrin interior, being
located in the middle of the larger rim. In addition, the other R′
substituent (CH2OH) at the chiral carbon presumably behaves
as a hydrogen bond donor toward one of the oxygen atoms
present in the larger rim of DM-â-CD, with a consequent poor
penetration of the guest into the host cavity. These results are
probably the main reason for the lack of chiral discrimination
of the corresponding chiral nitroxide radicals (Table 2). Due to
the geometry of these complexes an increase in the substituent
R lipophilicity (R ) Me, Et, i-Bu) does not affect the
enantiomeric discrimination.
The situation is reversed in the case of guests where R is or
contains an aromatic ring (1d-1g, 1i) giving C*-R in
complexes, independently of the nature of the aliphatic group
R′ (CH2OH, CH2OMe, CH3). In this case, the host cavity shows
a greater affinity for the aromatic system bonded to the chiral
center than for the N-benzyl group, forcing the chiral carbon to
be included within the cyclodextrin cavity. With this geometry,
the nature of the other substituent at the asymmetric carbon
determines the depth of penetration of the stereogenic center
into the cyclodextrin cavity and thus is responsible for the
observed chiral selectivity. In particular with nitroxides, where
the R substituent is a phenyl group, as in 2d, 2i, and 2g, the
degree of enantioselection increases proportionally to the
hydrophobicity of the R′ group (R′ ) CH2OH, CH2OMe, CH3,
respectively), the best value of chiral discrimination being that
found with nitroxide 2g (ES ) 13.42%).
Molecular dynamics calculations also indicate that increasing
of the R′ substituent hydrophobicity gives rise to a deeper
penetration of the chiral carbon into the receptor cavity.
Actually, the average distance between the guest chiral center
and the plane defined by all H3 protons of DM-â-CD increases
from 0.01 Å for R′ ) CH3 to 0.93 and 1.78 Å for R′ ) CH2-
OCH3 and R′ ) CH2OH, respectively. The more pronounced
penetration of the guest with R′ ) CH3 (1g) with respect to 1d
and 1i is also confirmed by the larger distance, computed in
the inclusion complex of 1g, between the para-hydrogen of the
Ph-C* group and the plane defined by all H3 protons (see Table
4).
Molecular dynamics simulations show that amine 1e, contain-
ing a methylene unit between the phenyl group and the chiral
center, behaves similarly to 1d. Actually, the stereogenic center
of the guest interacts weakly with the cyclodextrin interior
leading to the low enantiodiscrimination observed in the
corresponding nitroxides. However, the distance of the para-H
from the plane passing through the H3 protons of cyclodextrin
is similar to that found for 1d, implying that the aromatic ring
is more tilted in 1e than in 1d.
Contrary to the other enantiomers, (R)-2f and (S)-2f (R′ )
3-indolilmethyl) which contain aromatic substituent at the chiral
carbon, are not selectively recognized by DM-â-CD (ES )
2.43%). Both the small change of the EPR nitrogen coupling
when passing from the free to the complexed form and the MD
simulations indicate that only the indolic ring is included inside
the CD cavity while the N-benzylic group is exposed to solvent.
This geometry of the complex rationalizes the large distance
between the chiral center and the interior of the CD-cavity and
thus the observed lack of enantiodiscrimination.
Conversion of CH2OH to a CH2OMe group (compound 1h)
allows for a better insertion of the guest (presumably due to
9
4352 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004

Recognition by 2,6-di-O-Methyl-â-cyclodextrins
A R T I C L E S
On the basis of the above discussion, it can be concluded
that better chiral recognition is obtained when the chiral center
of the guest interacts strongly with the internal chiral field of
the cyclodextrin. If we assume that a greater intermolecular
interaction between the chiral groups of guest and host is
operating at the center of the cyclodextrin cavity, the dependence
of the chiral discrimination on the depth of guest penetration
could be modeled by means of a Gaussian distribution centered
at the acetalic-O plane. Figure 9 shows that a good fit of the
experimental data is obtained by means of this model by using
a half width σ equal to 1.7 Å.
This simplified model implies that the mechanism of enan-
tiomeric discrimination is determined by the mode of guest
penetration, with the magnitude of enantioselection dependent
upon the degree of intermolecular contact inside the cyclodextrin
cavity. Application of this hypothetical model implies that, after
reaching a maximum in the enantiodiscrimination process,
accommodation of the chiral center at the smaller rim of the
cavity should lead to a decrease in enantiodiscrimination. Design
of additional chiral probes is required to confirm this hypoth-
esis.30
Figure 9. Plot of chiral discrimination versus the distance between the
guest chiral carbon and the H3 plane in DM-â-CD/amine complexes. The
line represents a hypothetical Gaussian behavior with the maximum centered
at the acetalic-O plane and was obtained by numerical fitting of the
experimental data by introducing the half width σ as an adjustable parameter.
ESI-MS Results. Analysis of the ESI-MS peak intensities
of the signals due to labeled and unlabeled guest enantiomers
(i.e., R-dn + S or R + S-dn) complexed with DM-â-CD, provided
an ES ratio smaller than 1% for all of the investigated amines.
The poor chiral discrimination resulting from ESI-MS experi-
ments could be attributed to the fact that expected inclusion
complexes are not formed in the gas phase but instead only
external electrostatic adducts of DM-â-CD with protonated
amines are formed. This result is, perhaps, not surprising since
binding contributions from solvophobic interactions are absent
in the gas phase and purely hydrophobic adducts can hardly
survive in ESI conditions.31
The statement that hydrophobic interactions play an important
role in controlling the stabilities of the complexes between our
probes and DM-â-CD comes from the analysis of the affinity
constants that depend strongly on the relative hydrophobicities
of the substituents at the chiral carbon. For instance, the affinity
constant increases in the series 2a < 2b < 2c due to the
increasing lipophilicity of the R substituent which changes from
CH3, to CH2CH3 and to CH2CH(CH3)2 in this series. Similarly,
K2 is larger in 2h than in 2b due to the replacement of the CH2-
OH substituent by CH2OCH3; in the series 2d < 2i < 2g the
value of the affinity constant parallels the lipophilicity of the
R′ substituent, i.e., CH2OH < CH2OCH3 < CH3.
Additional support for this view comes from the spectra of
1g recorded in the presence of two different cyclodextrins: DM-
â-CD, which in water solution is able to include the phenyl
ring bound to the chiral carbon, and TM-â-CD for which no
complexation was observed by EPR and NMR spectroscopies.
Nevertheless, the ESI-MS peak intensity of the [TM-â-CD +
1g + H+] is comparable to that of [DM-â-CD + 1g + H+].
On the basis of these observations we may conclude that in
the case of amines 1a-1i, ESI-MS detects electrostatic external
adducts of CDs with protonated amines which are more stable
in the gas-phase32 instead of the expected inclusion complexes
with CDs preformed in solution.
Conclusions
By using specifically designed EPR spin probes and their
corresponding diamagnetic precursors, we could clearly distin-
guish different geometries of inclusion complexes and succeed
in identifying the factors responsible for chiral recognition by
DM-â-CD. The use of EPR and NMR spectroscopies together
with molecular dynamics provides a detailed description of
supramolecular complexes in solution. Work to extend this
approach to other supramolecular systems is in progress.
Experimental Section
EPR Measurements. EPR spectra were obtained using a Bruker
ESP300 spectrometer equipped with an NMR gaussmeter for field
calibration and a Hewlett-Packard 5350B microwave frequency counter
for the determination of the g-factors, which were referred to that of
the perylene radical cation in concentrated H2SO4 (g ) 2.002 58). The
sample temperature was controlled with a standard variable temperature
accessory and was monitored before and after each run using a copper-
constantan thermocouple. The instrument settings were as follows:
microwave power 5.0 mW, modulation amplitude 0.05 mT, modulation
frequency 100 kHz, scan time 180 s. Digitized EPR spectra were
transferred to a personal computer and were analyzed using digital
simulations carried out with a program developed in our laboratory
and based on a Monte Carlo procedure.14b Nitroxide radicals were
generated by mixing a solution of the corresponding amine (ca. 1 mM)
with a solution of the magnesium salt of monoperoxyphthalic acid
(Aldrich, technical grade) (ca 1 mM). To achieve a sufficiently large
radical concentration, the mixed solution was sometimes heated at 60
°C for 1-2 min. Aliquots from a concentrated CD solution were added
to the solution of nitroxide to yield the required concentrations. Samples
were then transferred in capillary tubes (1 mm i.d.) and the EPR spectra
were recorded.
(32) It has been reported33 that protonated amino acids form inclusion complexes
with TM-â-CD in the gas-phase. The existence of the included structure
even in the gas-phase has been attributed to the “three-point attachment”
interaction in which -NH3+ and -COOH groups are inside the CD cavity.
(33) (a) Lebrilla, C. B. Acc. Chem. Res. 2001, 34, 653-661. (b) Ramirez, J.;
Ahn, S.; Grigorean, G.; Lebrilla, C. B. J. Am. Chem. Soc. 2000, 122, 6884-
6890.
(30) We selected a Gaussian function because the variation in the magnitude of
enantioselection can be correctly reproduced by a bell shape function, with
the understanding that the real function can be determined only by studying
additional chiral probes which are able to accommodate the chiral center
in the smaller rim of the cavity.
(31) Cunniff, J. B.; Vouros, P. J. Am. Soc. Mass Spectrom. 1995, 6, 437-447.
9
J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004 4353

A R T I C L E S
Franchi et al.
NMR Measurements. All 1D and 2D NMR spectra were recorded
at 298 K on Varian Inova spectrometers operating at 300 and 600 MHz,
respectively, and on a Varian Mercury instrument operating at 400 MHz.
NMR experiments were performed in D2O solutions using residual
HOD as an internal standard (4.76 ppm). In some cases CD3OD (10%
v/v) was added in order to improve the amine solubility. Chemical
shifts are reported in parts per million (δ scale).
Amine titrations with DM-â-CD were carried out on a Varian Inova
300 at 298 K. NMR determination of the binding constants of
diastereomeric complexes of (R) and (S)-1g were performed on 2.15
mM solutions of the guest containing increasing amounts of DM-â-
CD (0-20 mM), in D2O/CD3OD (10:1 v/v). All solutions were prepared
in a 0.1 M phosphate buffer adjusted at pH 7.4.
ROESY data were collected on a Varian Mercury 400 and on a
Varian Inova 600, using a 90° pulse width of 13.5 µs and 6.1 µs and
a spectral width of 4000 and 6000 Hz in each dimension, respectively.
The data were recorded in the phase sensitive mode using a CW spin-
lock field of 2 kHz, without spinning the sample. Acquisitions were
recorded at mixing times 50-400 ms. Other instrumental settings are:
256 increments of 2 K data points, 8 scans per t1, 2.5 s delay time for
each scan.
Dynamic Simulations. SD simulations were carried out using the
MacroModel 7.0 program. Extended nonbonded cutoff distances were
set to 8 Å and 20 Å for the van der Waals and electrostatic interactions.
All C-H and O-H bond lengths were held fixed using the SHAKE
algorithm. Translational and rotational momentum were removed every
0.1 ps.
ESI-MS Measurements. Mass spectra were recorded with Micro-
mass ZMD ESI-MS spectrometer by using the following instrumental
settings: positive ions; desolvation gas (N2) 220 L/h; cone gas
(skimmer): 32 L/h; desolvation temp. 120 °C; capillary voltage: 2.8
kV; cone voltage: 45 V; hexapole extractor: 3 V.
Materials. â-CD, DM-â-CD, TM-â-CD, D,L-amino acids, (R) and
(S)-2-amino-1-butanol, (R) and (S)-R-methylbenzylamine, and (R) and
(S)-N-benzyl-R-methylbenzylamine are commercially available (Sigma-
Aldrich) and were used as received.
Acknowledgment. Financial support from MURST (Research
project “Free Radical Processes in Chemistry and Biology:
Synthesis, Mechanisms, Applications”) and University of
Bologna is gratefully acknowledged. We wish also to thank
Fiammetta Ferroni for technical assistance.
Supporting Information Available: Preparation of amines
1a-1i and of their deuterated analogues, some representative
EPR spectra of nitroxides 2a, 2c-2g, ESI-MS spectrum of 1g
in the presence of DM-â-CD and TM-â-CD, and a plot of the
decrease of a(N)water-a(N)CD versus the distance between the
guest chiral carbon and the H3 plane in DM-â-CD/amine
complexes (PDF). This material is available free of charge via
the Internet at http://pubs.acs.org.
JA049713Y
9
4354 J. AM. CHEM. SOC. VOL. 126, NO. 13, 2004