A New Stereochemical Model from NMR for Benzoylated Cyclodextrins, Promising New Chiral Solvating Agents for the Chiral Analysis of 3,5-Dinitrophenyl Derivatives

Article abstract of DOI:10.1021/jo961562x

Hexakis(2,3-di-O-benzoyl)-α-cyclodextrin and hexakis(2,3,6-tri-O-benzoyl)-α-cyclodextrin have been employed as chiral solvating agents (CSAs) for the NMR determination of the enantiomeric composition of derivatives of chiral amines, amino alcohols, alcohols, carboxyl acids, and amino acids bearing a 3,5-dinitrophenyl moiety. The conformational features of the two cyclodextrins have been carefully analyzed by NMR spectroscopy, and the origin of the symmetry change (C₆ → C₃), detected by NMR for hexakis(2,3-di-O-benzoyl)-α-cyclodextrin in CDC1₃, has been clarified.

Full text of DOI:10.1021/jo961562x

J . Org. Chem. 1997, 62, 827-835
827
A New Ster eoch em ica l Mod el fr om NMR for Ben zoyla ted
Cyclod extr in s, P r om isin g New Ch ir a l Solva tin g Agen ts for th e
Ch ir a l An a lysis of 3,5-Din itr op h en yl Der iva tives
Gloria Uccello-Barretta, Angela Cuzzola, Federica Balzano, Rita Menicagli, Anna Iuliano, and
Piero Salvadori*
Centro di Studio del CNR per le Macromolecole Stereordinate ed Otticamente Attive,
Dipartimento di Chimica e Chimica Industriale, via Risorgimento 35, 56126 Pisa-Italy
Received August 12, 1996X
Hexakis(2,3-di-O-benzoyl)-R-cyclodextrin and hexakis(2,3,6-tri-O-benzoyl)-R-cyclodextrin have been
employed as chiral solvating agents (CSAs) for the NMR determination of the enantiomeric
composition of derivatives of chiral amines, amino alcohols, alcohols, carboxyl acids, and amino
acids bearing a 3,5-dinitrophenyl moiety. The conformational features of the two cyclodextrins
6
have been carefully analyzed by NMR spectroscopy, and the origin of the symmetry change (C f
C
3
), detected by NMR for hexakis(2,3-di-O-benzoyl)-R-cyclodextrin in CDCl
3
, has been clarified.
In tr od u ction
Ch a r t 1
Cyclodextrins1a are cyclic oligosaccharides having an
apolar cavity where apolar moieties can be accom-
modated solely by means of hydrophobic interactions.
Owing to their chirality, the interaction with chiral
substrates can originate enantiodiscrimination phenom-
ena. For this reason such chiral auxiliaries have received
a great deal of attention for a long time both in asym-
1
metric synthesis and in the preparative or analytical
2
separation of chiral compounds by chromatography. In
addition, cyclodextrins have found a very interesting
application as chiral solvating agents (CSAs) for the
determination of the enantiomeric purity by nuclear
magnetic resonance (NMR).3
In particular a conformationally flexible structure of 1,
in CDCl , having a three-fold symmetry axis was evi-
denced; in such a structure adjacent glucose residues
twist each other around their glycosidic linkages. In this
way three units point their benzoyl groups toward the
inside of the cavity and their primary hydroxyl groups
3
Recently we have reported that the commercially
available permethylated â-cyclodextrin is a very efficient
CSA for the NMR determination of the enantiomeric
composition and absolute configuration of chiral trisub-
to the outside, whereas, for the other three units, the C
6
4
stituted allenes and enantiomeric composition of aro-
hydroxyl groups are directed toward the center of the
cavity, thus creating a network of strong intramolecular
hydrogen bonds (Figure 1). IR data were presented to
support this hypothesis.6
matic hydrocarbons.5
These results prompted us to probe the efficiency as
CSAs of other derivatized cyclodextrins; among them, we
focused our attention on the hexakis(2,3-di-O-benzoyl)-
R-cyclodextrin (1) and the hexakis(2,3,6-tri-O-benzoyl)-
R-cyclodextrin (2) (Chart 1), for which previous investi-
,7
In the present paper the use of 1 and 2 as CSAs for
the NMR determination of the enantiomeric composition
of 1-phenyl-1-(3′,5′-dinitrobenzamido)ethane (3), 1-cyano-
6
7
gations by NMR, carried out by Lehn and Stoddart, had
evidenced the remarkable effect of the degree of benzoyl-
ation or solvent nature on their conformational features.
1
-(3′,5′-dinitrobenzamido)-2-methylpropane (4), N-(3,5-
dinitrobenzoyl) derivatives of 2-amino-1-phenyl-1-pro-
panol (5a ), 2-amino-1-phenylethanol (5b), alanine methyl
ester (6),8 t-2-ethoxy-r-1-(hydroxymethyl)-1-methylcy-
X
Abstract published in Advance ACS Abstracts, J anuary 1, 1997.
9
(
1) (a) Szejtli, J . Cyclodextrin Technology; Kluwer: Boston, 1988.
b) Wenz, G. Angew. Chem., Int. Ed. Engl. 1994, 33, 803. (c) Bonchio,
M.; Carofiglio, T.; Di Furia, F.; Fornasier, R. J . Org. Chem. 1995, 60,
clobutane (7), and 3′,5′-dinitrophenyl-2-(p-isobutylphe-
(
nyl)propionamide (8) (Chart 2) has been reported, and
their enantiodiscriminating capability has been dis-
cussed. With the aim to obtain information on the
interaction mechanisms involved in the chiral discrimi-
nation process, the conformational and dynamic features
of 1 and 2 have been reexamined by an accurate NMR
5
1
986. (d) Hattori, K.; Takahashi, K.; Sakai, N. Bull. Chem. Soc. J pn.
992, 65, 2690.
(
2) (a) Francotte, E. J . Chromatogr. A 1994, 666, 565. (b) Li, S.;
Purdy, W. C. Chem. Rev. 1992, 92, 1457. (c) Schurig, V.; Nowotny, H.
P. Angew. Chem., Int. Ed. Engl. 1990, 29, 939. (d) Taylor, D. R.; Maher,
K. J . Chromatogr. Sci. 1992, 30, 67. (e) Rogan, M. M.; Altria, K. D.;
Goodall, D. M. Chirality 1994, 6, 25.
investigation in CDCl
3
or DMSO solution involving 2D
(
3) (a) Greatbanks, D.; Pickford, R. Magn. Reson. Chem. 1987, 25,
08. (b) Casy, A. F.; Mercer, A. D. Magn. Reson. Chem. 1988, 26, 765.
4) (a) Uccello-Barretta, G.; Balzano, F.; Caporusso, A. M.; Salvadori,
2
ROESY (Rotating-frame Overhauser Enhancement Spec-
(
10
13
troscopY) and C longitudinal relaxation time mea-
P. J . Org. Chem. 1994, 59, 836. (b) Uccello-Barretta, G.; Balzano, F.;
Caporusso, A. M.; Iodice, A.; Salvadori, P. J . Org. Chem. 1995, 60, 2227.
1
1
1
surements (T ). In addition, some attempts to detect
(
5) Uccello-Barretta, G.; Balzano, F.; Menicagli, R.; Salvadori, P. J .
Org. Chem. 1996, 61, 363.
6) Boger, J .; Corcoran, R. J .; Lehn, J . M. Helv. Chim. Acta 1978,
1, 2190.
7) Ellwood, P.; Spencer, C. M.; Spencer, N.; Stoddart, J . F.; Zarzycki,
R. J . Inclus. Phenom. Molec. Recogn. Chem. 1992, 12, 121.
(
(8) Salvadori, P.; Pini, D.; Rosini, C.; Uccello-Barretta, G.; Bertucci,
C. J . Chromatogr. 1988, 450, 163.
(9) Menicagli, R.; Malanga, C.; Lardicci, L. J . Org. Chem. 1982, 47,
2288.
6
(
S0022-3263(96)01562-9 CCC: $14.00 © 1997 American Chemical Society

8
28 J . Org. Chem., Vol. 62, No. 4, 1997
Uccello-Barretta et al.
Sch em e 1
The 3,5-dinitrobenzoyl derivatives 3 and 4 were pre-
pared by reacting the corresponding available amines
with 3,5-dinitrobenzoyl chloride in the presence of pyri-
dine. Racemic 5a as well as (R)- and (S)-5b and 8 were
1
2
prepared by using EEDQ as condensing agent.
As far as the two enantiomers of 5b are concerned, they
were prepared from the corresponding optically pure
amino alcohols 12, obtained by employing an efficient and
practical route, which avoids the racemate optical resolu-
tion (Scheme 1).13 The asymmetric catalytic dihydroxyl-
1
4
ation of styrene, according to the Sharpless method,
afforded (R)- or (S)-diol 9 in quantitative yield and 97%
ee, when a quinidine or quinine derivative was used in
the reaction, respectively.
Treatment of 9 with 1 equiv of tosyl chloride in pyridine
at -10 °C15 gave only the 1-hydroxy-2-tosyloxy derivative
10. Reaction of 10 with sodium azide in dimethylforma-
mide afforded the hydroxy azide 11, which was succes-
7
F igu r e 1. Proposed schematic representation of the structure
of 1 in CDCl .
3
Ch a r t 2
sively reduced with LiAlH
2.
Ch ir a l Discr im in a tion s by 1 a n d 2. In order to
4
to give the pure amino alcohol
1
probe the efficiency of 1 and 2 as chiral solvating agents
for the NMR determination of the enantiomeric composi-
tion, the 1H NMR spectra in CDCl
of the pure com-
3
pounds 3-8 were compared with those of their corre-
sponding equimolar mixtures with the cyclodextrin 1 or
2
.
In Figure 2 are reported the results obtained for 3 in
the presence of the cyclodextrin 1 (Figure 2b) and of
cyclodextrin 2 (Figure 2c). Both cyclodextrins produce
doubling of the proton absorptions of 3, to indicate that
the originally enantiotopic protons became diastereotopic
in the presence of the chiral auxiliary. Indeed, the 3,5-
dinitrobenzoyl protons of 3 (Figure 2a), which produce
two sharp signals at 9.15 ppm (triplet) and 8.94 ppm
(doublet) in the free compound, exhibit in the presence
of 1 (Figure 2b) two triplets at 9.08 ppm and 9.05 ppm
and two partially superimposed doublets at 8.90 ppm and
8
.89 ppm. As far as the methyl absorptions are con-
cerned, producing the doublet centered at 1.68 ppm in
the free compound, two superimposed doublets at 1.63
and 1.61 ppm (Figure 2b) are observed in the presence
of the chiral auxiliary 1. A minor but appreciable
doubling of the aromatic protons signals is also observed
in the presence of the cyclodextrin 2 (Figure 2c).
Compound 4 (Figure 3a) shows comparable doubling
of the diastereotopic methyl absorptions of the isopropyl
group in the presence of both 1 and 2 (Figures 3b, 3c).
In the case of 5a (Figure 4), each CSA induces non-
equivalence in the methyl absorptions, whereas only 1
intermolecular dipolar interactions between each cyclo-
dextrin and the chiral substrates have been made.
Resu lts a n d Discu ssion
Syn th esis of th e Cyclod extr in s 1, 2 a n d Ch ir a l
Su bstr a tes 3-5 a n d 8. The cyclodextrin derivatives 1
and 2 have been prepared according to the procedure
(12) Belleau, B.; Malek, G. J . Am. Chem. Soc. 1968, 90, 1651.
(13) Green, A. L.; Fielden, R.; Bartlett, D. C.; Cozens, M. J .; Eden,
R. J .; Hills, D.W. J . Med. Chem. 1967, 10, 1006.
reported by J . M. Lehn et al.6
(
14) Sharpless, K. B.; Amberg, W.; Bennani, Y. L.; Crispino, G. A.;
(
10) Neuhaus, D.; Williamson, M. The Nuclear Overhauser Effect;
VCH Publishers, Inc.: New York, 1989.
11) Wehrli, F. W.; Wirthlin, T. Interpretation of Carbon-13 NMR
Spectra; Heyden & Son Ltd.: London, 1976.
Hartung, J .; J eong, K. S.; Kwong, H. L.; Morikawa, K.; Wang, Z. M.;
Xu, D.; Zhang, X. L. J . Org. Chem. 1992, 57, 2768.
(15) Mangold, J . B.; Abdel-Monem, M. M. J . Med. Chem. 1983, 26,
66.
(

Stereochemical Model for Benzoylated Cyclodextrins
J . Org. Chem., Vol. 62, No. 4, 1997 829
F igu r e 3. 1H NMR (300 MHz, CDCl
, ppm referenced to TMS
3
as external standard, 25 °C) spectral regions corresponding
to the methyl proton absorptions of (a) free compound 4 (36
mM), (b) equimolar mixture (R)(S)-4/1, and (c) equimolar
mixture (R)(S)-4/2.
F igu r e 2. 1H NMR (300 MHz, CDCl
, ppm referenced to TMS
3
as external standard, 25 °C) spectral regions corresponding
to the 3,5-dinitrobenzoyl and methyl proton absorptions of (a)
free compound 3 (28 mM), (b) equimolar mixture (R)(S)-3/1,
and (c) equimolar mixture (R)(S)-3/2.
doubles significantly its 3,5-dinitrobenzoyl resonances
(Figure 4a). Interestingly, the analogous compound 5b
is discriminated only by 2 (Figure 5), and both the
complexation shifts and the nonequivalences are remark-
ably greater than those measured for 3 and 5a . It is also
to note that the interaction of each enantiomer of 5b with
2
causes a relevant broadening of their proton absorp-
tions, which could arise from a slowing down of their
molecular motion due to the formation of transient
diastereoisomeric species.
The enantiodiscriminating ability of 1 and 2 has been
also probed in the case of 3,5-dinitrophenyl derivatives
of amino acids (compound 6), alcohols (compound 7), and
carboxyl acids (compound 8). In these cases, smaller
nonequivalences were measured ranging between 1.0 and
5
.0 Hz for the 3,5-dinitrophenyl or alkyl protons.
Con for m a tion a n d Dyn a m ics of Ben zoyla ted Cy-
6
,7
clod extr in s. As reported in the literature, in DMSO
solution, both cyclodextrins 1 and 2 show the pattern of
1
6
H NMR absorptions expected for a C symmetry sys-
tem: only one set of resonances is observed for each
glucopyranosyl or aromatic proton.
On the contrary, a dramatic variation in the spectral
parameters of 1 was observed in CDCl : two sets of
3
signals, in an intensity ratio 1:1, for each proton or group
of equivalent protons of the cyclodextrin are detected
Figure 6a). Only negligible chemical shift variations are
F igu r e 4. 1H NMR (300 MHz, CDCl
, ppm referenced to TMS
3
as external standard, 25 °C) spectral regions corresponding
to the 3,5-dinitrobenzoyl and methyl proton absorptions of 5a
(36 mM) of (a) equimolar mixture (R)(S)-5a /1 and (b) equimolar
(
mixture (R)(S)-5a /2.
found for 2 in the same solvent (Figure 6b).
Cyclod extr in 1. The doubling of the absorptions of
The complete assignment of the proton absorptions
(Table 1) required the combined use of 2D DQF-COSY
and 2D ROESY analyses. The protons of each glucopy-
1
, observed in CDCl
kinds of glucopyranoside units named A and B (Scheme
).
3
, reveals the presence of two different
2
ranoside unit, indicated as H
1
-H for A and H1′-H6′ for
6

8
30 J . Org. Chem., Vol. 62, No. 4, 1997
Uccello-Barretta et al.
Sch em e 2
Ta ble 1. 1H NMR Ch em ica l Sh ifts Da ta a (300 MHz,
CDCl3, 25 °C) of 1 a n d 2
1
unit A
5.68
5.59
6.18
unit Bb
5.42
4.82
6.25
4.75
4.02
4.30-4.60
4.30-4.60
8.07
7.20
7.40
6.61
6.24
2
H1
H2
H3
H4
H5
H6
H6
Ho
5.67
5.00
6.29
4.29
4.83
4.89
5.08
7.39
6.86
7.11
7.34
6.81
7.13
8.17
7.49
7.52
4.31
4.70-4.90
4.30-4.60
4.70-4.90
7.30
7.09
7.29
7.95
7.43
7.52
2
2
Hm
2
Hp
Ho
Hm
Hp
Ho
Hm
3
3
3
6.84
6
6
6
Hp
F igu r e 5. 1H NMR (300 MHz, CDCl
, ppm referenced to TMS
3
a ppm referred to TMS as external standard. b Protons in the
as external standard, 25 °C) spectral regions corresponding
to the 3,5-dinitrobenzoyl proton absorptions of (a) free com-
pound 5b (36 mM), (b) equimolar mixture (R)(S)-5b/2, (c)
equimolar mixture (S)-5b/2, and (d) equimolar mixture (R)-
units B are indicated as H , H , etc.
1
′
2′
ortho protons H 2
2
o
of the residue Bz bound to the C
2
-O
at
site of A: they show an intense NOE on the proton H
.68 ppm and a significant NOE on H
Figure 7, trace a). The ortho protons H
were recognized as the ones giving the absorption at 7.95
ppm on the basis of the intense NOE produced on the H
proton (Figure 7, trace b). The proton absorptions of the
two benzoyl groups Bz and Bz of the other glucopyra-
noside residue B have been assigned on the basis of the
exchange effects observed in the 2D ROESY spectrum:
the absorption at 8.07 ppm, showing the exchange peak
1
5
b/2.
5
(
2
at 5.59 ppm
3
3
o
of Bz of A
3
2
′
3′
(
Figure 7, trace c) with the above ortho protons at 7.30
2
ppm (assigned to Bz ), was attributed to the ortho protons
H
originating the exchange peak with H
d), were assigned to H
2
′
2′
o
of Bz . Analogously, the ortho protons at 6.61 ppm,
3
o
(Figure 7, trace
3′. The other aromatic protons
o
were simply assigned on the basis of J correlations.
It is worth noting that the observability of the above
exchange peaks in the 2D ROESY spectrum is also a clear
indication that the two sets of protons, belonging to the
glucopyranoside moieties A and B, are in slow intercon-
version on the NMR time scale.
F igu r e 6. 1H NMR (300 MHz, CDCl
of (a) 1 and (b) 2.
, 25 °C, 28 mM) spectrum
3
The analysis of the 2D ROESY traces (Figure 7) also
afforded clear indications on the relative stereochemistry
B, have been simply assigned on the basis of the J
correlations found in the COSY map. The precise as-
signment of each benzoyl group to the corresponding
hydroxyl site required the analysis of the intraring NOEs
measured in the 2D ROESY spectrum. In particular the
doublet centered at 7.30 ppm has been assigned to the
1
of the two kinds of residues: the proton H at 5.68 ppm
of A (Figure 7, trace e) shows an intense NOE on the
proton H4′ (4.75 ppm) of the adjacent residue B, to
indicate that these protons are cisoid. By contrast no
inter-NOE H1′-H
4
is observed. Accordingly, the ortho
2
aromatic protons H
o
(Figure 7, trace a) originate NOEs

Stereochemical Model for Benzoylated Cyclodextrins
J . Org. Chem., Vol. 62, No. 4, 1997 831
F igu r e 9. 13C relaxation times (T
and benzoyl carbons belonging to the two different residues A
, s) of the glucopyranoside
1
2
3
2′
3′
(
Bz , Bz ) and B (Bz , Bz ) (A and B referenced to Scheme 2)
of 1 in CDCl (44 mM) solution.
3
unit are observed (Figure 7, trace c), whereas no inter-
NOEs between the proton H1′ and H
3
or Bz are mea-
sured, as would expected when primary hydroxyl alter-
natively point toward the outside and inside of the ring
F igu r e 7. Traces of the 2D ROESY spectrum (300 MHz,
3
CDCl
ortho protons H
3
, 25 °C, 28 mM, τ
m
) 0.3 s) of 1 corresponding to (a)
2
3
2′
o
o o
, (b) ortho protons H , (c) ortho protons H ,
3
′
o 1
(d) ortho protons H , and (e) H .
(Figure 1).
Therefore, the introduction of benzoyl groups on the
secondary hydroxyls of the cyclodextrin causes a severe
distortion of the molecule, and the NOE data clearly show
that all the primary hydroxyl groups of both units A and
B point at the outside of the ring, as well as the two
2
′
3′
benzoyl groups Bz and Bz of B, whereas the two
2
3
benzoyl groups Bz and Bz of the other residue A point
toward the distorted cavity.
It is worthwhile to note that in such a stereochemical
arrangement (Scheme 2) the Bz2 of B is in proximity of
the primary hydroxyl group of the adjacent residue A,
probably allowing a CdO‚‚‚HsO hydrogen bond.
′
6 3
The C f C reduction of the symmetry with respect
to the unsubstituted cyclodextrin is also accompanied by
a significant distorsion of the residue B, since its H5′
proton does not originate any dipolar interaction with H3′
.
Therefore, the residue B can be schematically repre-
sented as a twisted chair (Scheme 2) having the H5′ and
3
F igu r e 8. Conformational arrangement of 1 in CDCl as
derived by NMR data.
H
3′ protons far from each other and, hence, in a pseu-
3
not only on the ortho aromatic protons H
o
belonging to
doequatorial situation. Probably this distortion is the one
required to have the alternating residues characterized
by H and H4′ cisoid and H1′ and H far from each other.
1 4
the same residue A, but also on the ortho aromatic
3
′
protons H
o
of the adjacent residue B. The aromatic
(Figure 7, trace c) only produce NOE on the
protons of the same residue.
2
′
protons H
o
The comparison between the longitudinal relaxation
times of the methine carbons in the aromatic moieties
with those of the glucosidic methines, to which they were
bound, afforded valuable information on the relative
rigidities of the aromatic nuclei of A and B (Figure 9).
All the aromatic methine carbons of the undistorted
unit A (Figure 9) have relaxation times very similar
(0.16-0.18 s) to that of the glucopyranoside carbons
bearing the substituent. Therefore the aromatic rings
3
′
adjacent H
o
Therefore, by analyzing the triad BAB (Scheme 2), for
the A unit, we find the proton H faced to the proton H4′
of the adjacent B unit and its proton H far from the
1
4
proton H1′ of the unit B on the other side (Scheme 2).
Analogously, for the unit B, the proton H1′ is far from
the proton H
4 1
of A and proton H4′ cisoid to the proton H
of the residue A on the other side (Scheme 2).
2
3
Two different stereochemical pictures can be in keeping
with the above discussed results: in a first structural
arrangement three primary hydroxyl groups point toward
of Bz and Bz follow the global motion of A, and they do
not undergo independent motions about preferential axes.
Rather different results were obtained for the B residue
6
the center of the cavity (as assessed by Lehn and
(Figure 9): the C2′ and C3′ methine T
than those of ortho and meta methines of Bz and Bz
(0.26 s for Co2′, 0.18 s for Cm2′ and 0.31 s for Co3′, 0.17 for
m3′). Moreover these last T s are remarkably longer
1
s (0.14 s) are shorter
7
2′
3′
Stoddart, Figure 1); in a second arrangement all the
primary hydroxyl groups are disposed outside of the ring
(
Figure 8). However, the latter possibility has to be
C
1
2
′
preferred on the basis of NOE data: inter-NOEs H
H
the glucopyranoside protons H
o
-
than those of the corresponding para methines (0.11 s).
This is an indication that these aromatic residues un-
dergo an anisotropic molecular motion around the long
2
′
-H
1
between the ortho protons of Bz2′ and
2
and H
o
1
2
and H of the adjacent

8
32 J . Org. Chem., Vol. 62, No. 4, 1997
Uccello-Barretta et al.
F igu r e 10. Traces of the 2D ROESY spectrum (300 MHz,
DMSO, 25 °C, 28 mM, τ ) 0.3 s) of 1 corresponding to the
ortho aromatic protons of the benzoyl group bound to (a) C -O
m
2
2
3
site (H
o
) and (b) C
3
o
-O site (H ).
molecular axis. The fast rotation around this axis
contributes to the global relaxation of the ortho and meta
carbons, and the effective rotational correlation time of
p-C is longer than those of m-C or o-C.
These results are in keeping with our stereochemical
2
3
model: the two benzoyls Bz and Bz of A are less free to
rotate as they point toward the cavity, whereas the
F igu r e 11. Traces of the 2D ROESY spectrum (300 MHz,
2
′
3′
CDCl , 25 °C, 28 mM, τm ) 0.3 s) of 2 corresponding to the
benzoyls Bz and Bz of B are more free to rotate as they
point toward the outside.
3
ortho aromatic protons of the benzoyl group bound to (a) C -O
6
6
2
3
site (H
o
), (b) C
2
-O site (H
o
), and (c) C
3
o
-O site (H ).
1
The H NMR spectrum of the same cyclodextrin, in
DMSO as solvent, shows a signal pattern corresponding
to a C
6
symmetry. Namely only one kind of glucopyra-
noside residue is present. The assignment of the ben-
2
3
zoyls Bz and Bz bound to the ring carbons C
2
-O and
C
3
-O, respectively, is immediately obtained by compar-
ing in the 2D ROESY spectrum the traces of the two
kinds of ortho protons resonating at 7.45 and 7.30 ppm
(
Figure 10, traces a, b): the first produce intense and
comparable NOEs on H , H , and H and the latter an
intense NOE on H , but significantly minor NOEs on H
and H . Therefore, they must be assigned to the ortho
protons of Bz and Bz , respectively. Taking into account
3
2
1
3
1
2
2
3
2
3
also that no NOE between H
o o 4
or H and H , directed
toward the outside, is observed, we can conclude that
both benzoyl groups are mainly bent toward the cavity.
Cyclod extr in 2. We have finally compared the con-
formational features in CDCl
those of the cyclodextrin 2, having spectral parameters
in agreement with a C symmetry. As above described,
3
of the cyclodextrin 1 with
6
the benzoate units have been assigned to the correspond-
ing sites on the glucopyranoside ring by an analysis in
which 2D ROESY and 2D DQF-COSY data are compared.
The most interesting traces of the 2D ROESY spectrum
are collected in Figure 11. The protons at 8.17 ppm,
3
F igu r e 12. Conformational arrangement of 2 in CDCl as
derived by NMR data.
6 5
originating NOEs on the protons H and H , have been
6
assigned to the ortho protons of Bz (Figure 11, trace a).
The absorption at 7.39 ppm has been attributed to the
2
ortho protons of the benzoyl Bz as they produce NOEs
not only on the adjacent H
NOE on H (Figure 11, trace b). Finally, the ortho
protons at 7.34 ppm have been assigned to Bz as they
produce a more intense NOE on H and a less intense
NOE on H (Figure 11, trace c).
The observability of inter-NOEs H
2 3
and H , but also an intense
1
3
3
2
6-H
and H 3
-H
o
5
o
3
F igu r e 13. 13C relaxation times (T
CDCl (74 mM) solution.
1
, s) of the benzoate of 2 in
2
or H
o
-H
3
between ortho protons of the benzoate groups
3
and internal protons of the cyclodextrin allow us to
establish that these aromatic moieties are on the average
faced to the two cavities of 2, but their inclusion does
not occur as no inter-NOEs between para aromatic
Once again, the carbon relaxation times (Figure 13)
reflect the dynamic behavior of the cyclodextrin. These
are rather homogeneous for the aromatic methine car-
2
3
protons and internal H
3
or H
5
are measured (Figure 12).
bons of Bz or Bz and the methine directly linked to them

Stereochemical Model for Benzoylated Cyclodextrins
J . Org. Chem., Vol. 62, No. 4, 1997 833
F igu r e 14. Trace of the 2D ROESY spectrum (300 MHz,
CDCl
3
, 25 °C, 80 mM, τ ) 0.3 s) of the mixture (R)-5b/2 (molar
m
ratio 1:1) corresponding to the para aromatic proton of the 3,5-
dinitrobenzoyl moiety of 5b.
(C
2
and C
3
, respectively), whereas remarkably longer
relaxation times are measured for the meta and ortho
6
5
carbons of Bz relatively to the para and C carbons. This
indicates that Bz2 and Bz3 are more rigid than Bz .
6
Accordingly, the trace of the 2D ROESY map correspond-
6
6
ing to the ortho protons H
shows an intense NOE with the external proton H
addition to the above discussed NOE on the internal
proton H
o
of Bz (Figure 11, trace a)
4
, in
5
.
F igu r e 15. Ortho protons H 3
300 MHz, CDCl , 25 °C, 80 mM, τ
equimolar mixture (S)-5b/2, and (c) equimolar mixture (R)-
b/2.
traces of the 2D ROESY spectra
) 0.3 s) of (a) free 2, (b)
o
In ter a ction Mech a n ism 2/(R)-5b a n d 2/(S)-5b. At-
tempts to obtain information on the interaction mecha-
nism of 1 with the chiral substrates 3-8 by detection of
the intermolecular dipolar interactions in the 2D ROESY
maps of their mixtures failed probably as a consequence
of the lability of the diastereoisomeric complexes formed.
However, as evidenced above, the interaction of 2 with
the two enantiomers of 5b caused both a significant
variation of the chemical shifts of (S)-5b or (R)-5b with
respect to the free compounds and a remarkable broad-
ening of their NMR absorptions (Figure 5). This clearly
indicates that the diastereoisomeric complexes of (S)-5b
and (R)-5b with 2 show higher stability constants than
those of the complexes with 1 and also a high degree of
conformational rigidity induced by the complexation. In
fact, the analysis of the 2D ROESY spectra of the
equimolar mixtures 2/(R)-5b or 2/(S)-5b allowed detection
of inter-NOEs (Figure 14) between the 3,5-dinitrobenzoyl
(
3
m
5
two different kinds of glucopyranoside residues are
present: undistorted and distorted. The former main-
tains a chair-type conformation and the latter is a
twisted-chair one. They reside alternatively on almost
perpendicular planes and interchange. The benzoate
2 3
groups at C and C of the undistorted residues point
toward the cavity, whereas the same groups in the
distorted residues point away from it (Figure 8). In this
way all the primary hydroxyl groups are directed to the
outside of the cyclodextrin; in particular, the hydroxyl
function of the undistorted residue is close to the ben-
zoate group at C2′ site of the adjacent distorted residue,
probably allowing the CdO‚‚‚HO hydrogen bond forma-
tion.
protons of (R)-5b or (S)-5b and the proton H
3
lying inside
the cyclodextrin. No further intermolecular NOEs have
been detected; therefore, the interaction between the
Earlier investigations6,7 on the same system led to a
model where three hydroxyl groups point toward the
center of the cavity creating a network of OH‚‚‚OH
hydrogen bonds (Figure 1). This structure is not in
agreement with our NOE measurements. In this respect
3
enantiomers of 5b and 2 takes place at the C site of the
cyclodextrin. This interaction seems to occur at the
external surface of the cyclodextrin and probably no
inclusion occurs taking into account that no inter-NOEs
with the other internal proton H are measured.
5
It is worthwhile to consider an interesting point arising
7
it is interesting to note that the same authors report a
C
6
symmetry for the analogous benzylated cyclodextrin
(
PhCH vs PhCO) in which the benzyl groups could not
2
hinder the formation of OH‚‚‚OH hydrogen bonds. There-
fore, the existence of interresidue CdO‚‚‚HsO hydrogen
from the comparison between the traces of the protons
3
H
o
of the 2D ROESY maps of free 2 and its mixture with
bond seems better account for the C
As far as the perbenzoylated cyclodextrin 2 is con-
cerned, the C symmetry is retained and the exhaustive
6 3
f C transition.
one or another enantiomer of 5b (Figure 15), i.e. for free
3
2
the ortho protons of the benzoate (Bz ) bound to C
3
does
not produce NOE with H (Figure 15, trace a); therefore,
4
6
this benzoate must be faced to the wider cavity, as
already discussed. By contrast, in the mixtures, the same
proton (Figure 15, traces b, c) produces a relevant NOE
on H and, hence, it must be rotated toward the external
4
part of the intermolecular complex.
derivatization creates a closed structure where the ben-
zoyl groups occlude both the wide and narrow rims and
probably prevent the insertion of substrates. In spite of
both the conformational distortion found for 1 and cavity
crowding in 2, the two cyclodextrins show enantiodis-
criminating ability toward chiral substrates, and their
use as chiral solvating agents for NMR spectroscopy can
be proposed. However, since it seems unlikely that the
enantiodiscrimination could originate from classical in-
clusion phenomena, superficial interactions of the chiral
substrates with the polar functionalities of the deriva-
tized cyclodextrin can be proposed. In the case of 2, the
noninclusive interaction with the enantiomers of 5b has
Con clu sion s
The above described results allow the nature of the
distortion introduced in R-cyclodextrin by benzoylation
of the hydroxyl groups to be established. Compound 1
shows, in DMSO, a C
are faced to the wider cavity without undergoing inclu-
sion. In CDCl , the symmetry of 1 reduces to C , and
6
symmetry, and the benzoyl groups
3
3
3
been proved to involve the C site of the cyclodextrin and

8
34 J . Org. Chem., Vol. 62, No. 4, 1997
Uccello-Barretta et al.
to cause the rotation of the benzoyl groups toward the
outside of the cavity.
(R)- a n d (S)-2-Hyd r oxy-1-p h en yleth a n ol [(R)- a n d (S)-
]. According to the Sharpless recipe,14 (R)-9 and (S)-9 were
obtained starting from styrene (16.0 mmol) in quantitative
9
yield after flash cromatography (ethyl acetate): mp 66 °C;
[R]21 -38.1 (c 1.25, EtOH) for (R)-9 and [R]
21
+38.1 (c 1.25,
Exp er im en ta l Section
D
D
1
7 1
EtOH) for (S)-9 according to literature; H NMR (CDCl
3
, 25
Gen er a l Meth od s. NMR measurements were performed
on a Varian VXR-300 spectrometer equipped with a temper-
ature control unity ((0.1 °C). The 2D NMR spectra were
obtained by using standard sequences. The double-quantum-
filtered (DQF) COSY experiments were recorded with a
spectral width of 1800 Hz; 512 increments of 8 scans and 2K
data points were acquired. The relaxation delay was 5 s. The
data were zero-filled to 2K × 1K, and a Gaussian function was
applied for processing in both dimension. The HETCOR
°
2
C, ppm referenced to TMS as external standard): 2.30 (1H),
.75 (1H), 3.60-3.80 (2H), 4.85 (1H), 7.20-7.50 (5H). Anal.
: C, 69.54%; H, 7.30%. Found for (R)-9: C,
9.57%; H, 7.24%. Found for (S)-9: C, 69.52%; H, 7.23%.
R)- a n d (S)-1-P h en yl-2-(tosyloxy)eth a n ol [(R)- a n d (S)-
0]. A stirred solution of (R)-9 (14.5 mmol) in 20 mL of
8 10 2
Calcd for C H O
6
(
1
anhydrous pyridine was treated, at -10 °C, with a stoichio-
metric amount of TsCl dissolved in the same solvent. The
reaction mixture was stirred at -10 °C for 4 h and then
quenched with ice-water. The organic materials were ex-
spectra were acquired with a spectral width of 11000 Hz in F
and 2000 Hz in F in 2K data points using 32 scans of the 256
2
1
tracted with Et
O, and dried with Na
pressure to give 11.0 mmol (76% yield) of (R)-10: mp 65-66
O, washed with HCl (10%), H
2 3
O, NaHCO , and
2
increments. The relaxation delay was 5 s. The data were zero-
filled to 2K × 1K and a Gaussian function was applied for
processing in both dimension. The phase-sensitive ROESY10
spectra were acquired with a spectral width of 2000-3000 Hz
H
2
2
SO . Ether was removed at reduced
4
21
1
°
C, [R]
referenced to TMS as external standard): 2.10 (1H), 2.45 (3H),
.00-4.20 (2H), 5.00 (1H), 7.30-7.40 (7H), 7.76-7.82 (2H).
Anal. Calcd for C15 S: C, 61.63%; H, 5.52%; S, 10.95%.
D
-51.1 (c 1.00, CHCl
3 3
). H NMR (CDCl , 25 °C, ppm
in 2K data points using 8 scans for each of the 512 t
1
4
increments. The spin-lock time was set to 300 ms. The data
were zero-filled to 2K × 1K, and a Gaussian function was
applied for processing in both dimensions. The 13C spin-lattice
relaxation times were measured by the inversion-recovery
method.
16 4
H O
Found: C, 61.60%; H, 5.56%; S, 11.00%.
According to the above procedure, from (S)-9 was obtained
21
(S)-10: 78% yield, [R]
D
3
+51.1 (c 1.00, CHCl ). Anal. Calcd
for C15H O S: C, 61.63%; H, 5.52%; S, 10.95%. Found: C,
16 4
Thin layer chromatography (TLC) was carried out on silica
gel plates (Merck, Silica G-60 0.2 mm), and compounds were
visualized with iodine or by examination under UV light.
Flash chromatography was carried out using Silica Gel 60
61.68%; H, 5.50%; S, 10.99%.
(R)- a n d (S)-2-Azid o-1-p h en yleth a n ol [(R)- a n d (S)-11].
A solution of (R)-10 (8.6 mmol) and NaN (12.0 mmol) in DMF
3
(70 mL) was stirred at 90-95 °C for 2 h. The reaction mixture
(Fluka, 0.035-0.070 mm particle size, 220-440 mesh ASTM).
was treated with H O, and the organic materials were
2
High performance liquid chromatography (HPLC) was
extracted with Et
elution (SiO , CHCl
byproducts, and CHCl
(93%) pure (R)-11 as a colorless liquid: [R]
2
O. After the usual workup, chromatographic
/petroleum ether 1:1 for the elution of
/Et O 1:1 for the azido elution) yielded
performed on a Perkin Elmer series 410 HPLC with UV
detection at 254 nm, using a LiCrospher 100 RP-18 (5 µm)
column. Solvent systems for chromatography are v/v.
Melting points were determined using a Koffler hot-stage
apparatus. Optical rotations were measured using a Perkin
Elmer 142 polarimeter.
2
3
3
2
1
9
D
-92.8 (c 1.08,
, 25 °C, ppm referenced to TMS as
external standard): 2.45 (1H), 3.40-3.60 (2H), 4.85-4.95 (1H),
7.30-7.45 (5H). Anal. Calcd for C O: C, 58.87%; H,
.56%; N, 25.76%. Found: C, 58.83%; H, 5.58%; N, 25.80%.
According to the above procedure, from (S)-10 was obtained
1
3 3
CHCl ). H NMR (CDCl
8
9 3
H N
5
Ma ter ia ls. Compound 1-cyano-1-amino-2-methylpropane
was kindly provided by DMS Research (Geleen, The Nether-
lands). All chemicals were purified prior to use by standard
1
9
(
S)-11: 90% yield, [R]
D
+92.8 (c 1.08, CHCl
3
). Anal. Calcd
1
6
for C
8
H
9
N
3
O: C, 58.87%; H, 5.56%; N, 25.76%. Found: C,
methods.
mmHg, in the presence of P
R-Cyclodextrin was dried (12 h) at 110 °C/0.1
5
8.90%; H, 5.53%; N, 25.72%.
R)- a n d (S)-2-Am in o-1-p h en yleth a n ol [(R)- a n d (S)-12].
To a stirred suspension of LiAlH (32.7 mmol) in dry Et O (100
2
5
O .
(
R-Cyclodextrin and 2-amino-1-phenyl-1-propanol hydrochlo-
ride were obtained from Fluka.
4
2
mL) a solution of (R)-11 (7.4 mmol) in the same solvent (50
mL) was added. The reaction mixture was refluxed for 3 h,
stirred at room temperature for 12 h, hydrolyzed with ice-
water, and filtered. After the usual workup, (R)-12 was
Hexa k is(2,3,6-tr i-O-ben zoyl)-r-cyclod extr in (2). Fol-
6
lowing the experimental procedure reported, chemically pure
2
8
+
(TLC benzene/ethanol ) 4/1: R
f
) 0.8; HPLC MeCN/H O )
2
22
5/15,1 mL/min: t
R
) 30 min) having mp 157-160 °C, [R]
D
2
3
recovered (90% yield): mp 52-55 °C, [R]
D
-45.0 (c 1.32, C
according to the literature.18 H NMR (CDCl
, 25 °C, ppm
referenced to TMS as external standard): 1.90 (3H), 2.70-
.10 (2H), 4.60 (1H), 7.20-7.40 (5H). Anal. Calcd for C
NO: C, 70.04%; H, 8.08%; N, 10.21%. Found: C, 70.00%; H,
6 6
H )
6
22
50.8 (c 1.04, CHCl
recovered in 94% yield. H NMR (DMSO-d
3
) [lit. [R]
D
+54 (c 1, CHCl
, 25 °C, ppm
, 6H), 4.89
, 6H), 5.65 (H , 6H),
3
)], was
1
1
3
6
referenced to TMS as external standard): 4.57 (H
and H , 12H), 5.10 (H , 6H), 5.14 (H
4
3
8 11
H -
(H
5
6
2
6
1
3
3
2
6
7
7
.25 (H
.27 (H
3
, 6H), 6.87 (H
, 6H), 7.28 (H
m
, 12H), 6.98 (H
, 12H), 7.49 (H
m
2
p
, 12H), 7.24 (H , 6H),
8
.10%; N, 10.23%.
2
3
6
p
o
o
, 12H), 7.52 (H
m
, 12H),
According to the above procedure, from (S)-11 was obtained
6
6
p o 132 48
.61 (H , 6H), 8.13 (H , 12H). Anal. Calcd for C162H O :
2
3
(
S)-12: 80% yield; [R]
D
+45.0 (c 1.32, C
literature. Anal. Calcd for C 11NO: C, 70.04%; H, 8.08%;
N, 10.21%. Found: C, 70.03%; H, 8.03%; N, 10.22%.
6 6
H ) according to the
C, 64.85%; H, 4.90%. Found: C, 64.81%; H, 4.93%.
1
8
8
H
H exa k is(2,3-d i-O-b en zoyl)-r-cyclod ext r in (1). Com-
6
pound 2 was converted into 1 according to Lehn. Accurate
12
3
,5-Din itr oben zoyl Der iva tives of 5. A tetrahydrofu-
purification of crude 1 by flash chromatography (benzene/
ethanol ) 6/1) gave (41% yield) a chemically pure sample (TLC
ran solution (300 mL) of 3,5-dinitrobenzoic acid (15.0 mmol)
and 2-ethoxy-1-(ethoxycarbonyl)-1,2-dihydroquinoline (EEDQ)
benzene/ethanol ) 6/1 R
f
) 0.26; HPLC MeCN/H O ) 85/15,
2
(15.0 mmol) was stirred for 3 h, and then the suitable
7
1
°
mL/min: t
R
) 17.25 min) showing mp 207 °C (lit. 209-210
â-hydroxyphenethylamine (15.0 mmol) was added. The reac-
tion mixture was stirred at room temperature overnight, and
then the solvent was removed at reduced pressure. The crude
2
2
6,7
22
C), [R]
D
+23.4 (c 1.14, CHCl
and +11 (c 1, CHCl
3
) [lit. [R]
D
+9 (c 1.1, CHCl
3
)
, 25
1
3
) respectively]. H NMR (DMSO-d
6
°
C, ppm referenced to TMS as external standard): 3.88 (H
H), 4.17 (H , 6H), 4.26 (H , 6H), 4.40 (H , 6H), 4.89 (H , 6H),
.93 (OH, 6H), 5.42 (H , 6H), 6.07 (H , 6H), 6.90 (H
6
,
residue, crystallized from CHCl
3
/pentane (1:3), gave pure 5a ,b.
6
4
7
7
4
5
4
6
2
2
-(3′,5′-Dinitrobenzamido)-1-phenyl-1-propanol (5a ): mp
3
1
3
m
, 12H),
1
1
29-131 °C. H NMR (CDCl
as external standard): 1.14 (3H), 2.55 (1H), 4.53 (1H), 5.02
1H), 6.57 (1H), 7.26-7.50 (5H); 8.91 (2H), 9.14 (1H). Anal.
3
, 25 °C, ppm referenced to TMS
2
3
3
2
.05 (H
.45 (H
m
, 12H), 7.27 (H
p
, 6H), 7.30 (H
o
, 12H), 7.31 (H
p
, 6H),
2
o
, 12H). Anal. Calcd for C120
H
108
O42: C, 68.34%; H,
(
.68%. Found: C, 68.36%; H, 4.66%.
(
17) King, R. B.; Bakos, J .; Hoff, C. D.; Mark o´ , L. J . Org. Chem.
(16) Perrin, D. D.; Armarego, W. L. F.; Perrin, D. R. Purification of
1979, 44, 1729.
laboratory chemicals; Pergamon Press: Toronto, 1983.
(18) Ziegler, T.; H o¨ rsch, B.; Effenberger, F. Synthesis 1990, 575.

Stereochemical Model for Benzoylated Cyclodextrins
J . Org. Chem., Vol. 62, No. 4, 1997 835
Calcd for C16
Found: C, 55.61%; H, 4.40%; N, 12.20%.
R)- and (S)-2-(3′,5′-Dinitrobenzamido)-1-phenylethanol [(R)-
H
15
N
3
O
6
:
C, 55.65%; H, 4.38%; N, 12.17%.
t-2-Ethoxy-r-1-(hydroxymethyl)-1-methylcyclobutane 3,5-di-
nitrobenzoate (7):9 1H NMR (CDCl
, 25 °C, ppm referenced to
3
(
TMS as external standard): 1.21 (3H), 1.26 (3H), 1.57 (2H),
2.02 (1H), 2.22 (1H), 3.43 (1H), 3.48 (1H), 3.84 (1H), 4.29 (1H),
4.36 (1H), 9.15 (2H), 9.23 (1H).
1
and (S)-5b]: mp 95 °C. H NMR (CDCl
3
, 25 °C, ppm
referenced to TMS as external standard): 2.92 (1H), 3.56 (1H),
4
(
1
.01 (1H), 5.00 (1H), 6.88 (1H), 7.30-7.50 (5H), 8.94 (2H), 9.16
1H). Anal. Calcd for C15 : C, 54.38%; H, 3.96%; N,
2.69%. Found for (R)-5b: C, 54.41%; H, 3.94%; N, 12.66%.
N-(3′,5′-Dinitrophenyl)-2-(p-isobutylphenyl)propionamide
H
13
N
3
O
6
(8): mp 195-200 °C. 1H NMR (CDCl
3
, 25 °C, ppm referenced
to TMS as external standard): 0.95 (6H), 1.63 (3H), 1.85 (1H),
2.49 (2H), 3.80 (1H), 7.23 (4H), 7.71 (1H), 8.72 (3H). Anal.
Found for (S)-5b: C, 54.36%; H, 3.98%; N, 12.71%.
Ch a r a cter iza tion Da ta of 3, 4, 6, 7, a n d 8. 1-Phenyl-1-
3′,5′-dinitrobenzamido)ethane (3): mp 145 °C. H NMR
21 3 5
Calcd for C19H N O : C, 61.45%; H, 5.70%; N, 11.31%.
Found: C, 61.41%; H, 5.66%; N, 11.35%.
1
(
(
1
(
1
1
CDCl
.68 (3H), 5.36 (1H), 6.81 (1H), 7.27-7.50 (5H), 8.94 (2H), 9.15
1H). Anal. Calcd for C15 : C, 57.14%; H, 4.16%; N,
3.33%. Found: C, 57.10%; H, 4.18%; N, 13.30%.
-Cyano-1-(3′,5′-dinitrobenzamido)-2-methylpropane (4): mp
3
, 25 °C, ppm referenced to TMS as external standard):
Ack n ow led gm en t. This work has been partially
supported by CNR, Roma, Italy.
13 3 5
H N O
1
Su p p or tin g In for m a tion Ava ila ble: 1H NMR spectra of
1
3
27-129 °C. H NMR (CDCl , 25 °C, ppm referenced to TMS
1
and 2 in DMSO-d
6
at 25 °C; 2D DQF-COSY and 2D ROESY
as external standard): 1.18 (3H), 1.22 (3H), 2.25 (1H), 5.04
1H), 6.84 (1H), 8.99 (2H), 9.23 (1H). Anal. Calcd for
: C, 49.31%; H, 4.14%; N, 19.17%. Found: C,
9.34%; H, 4.11%; N, 19.20%.
N-(3,5-dinitrobenzoyl)alanine methyl ester (6): 1H NMR
CDCl , 25 °C, ppm referenced to TMS as external standard):
.60 (3H), 3.85 (3H), 4.85 (1H), 7.55 (1H), 8.90 (2H), 9.15 (1H).
maps of 1 in CDCl ; scheme summarizing the characterization
3
(
strategy for 1 (4 pages). This material is contained in libraries
on microfiche, immediately follows this article in the microfilm
version of the journal, and can be ordered from the ACS; see
any current masthead page for ordering information.
12 12 4 5
C H N O
4
8
(
1
3
J O961562X