A joint structural, kinetic, and thermodynamic investigation of substituent effects on host-guest complexation of bicyclic azoalkanes by β-cyclodextrin

Article abstract of DOI:10.1021/ja011866n

Derivatives of the azoalkane 2,3-diazabicyclo[2,2,2]oct-2-ene (1a) with bridgehead 1,4-dialkyl (1b), 1,4-dichloro (1c), 1-hydroxymethyl (1d), 1-aminomethyl (1e), and 1-ammoniummethyl (1f) substituents form host-guest inclusion complexes with β-cyclodextrin. They were employed as probes to assess substituent effects on the kinetics and thermodynamics of this complexation by using time-resolved and steady-state fluorimetry, UV spectrophotometry, induced circular dichroism (ICD) measurements, and ¹H NMR spectroscopy. The kinetic analysis based on quenching of the long-lived fluorescence of the azoalkanes by addition of host provided excited-state association rate constants between 2.6 × 10⁸ and 7.0 × 10⁸ M^-1 s^-1. The binding constants for 1a (1100 M^-1), 1b (900 M^-1), 1c (1900 M^-1), 1d (180 M^-1), 1e (250 M^-1), and 1f (ca. 20 M^-1) were obtained by UV, NMR, and ICD titrations. A positive ICD signal of the azo absorption around 370 nm was observed for the β-cyclodextrin complexes of 1a, 1d, and 1f with the intensity order 1a ? 1d ≈ 1f, and a negative signal was measured for those of 1b, 1c, and 1e with the intensity order 1c < 1b ≈ 1e. The ICD was employed for the assignment of the solution structures of the complexes, in particular the relative orientation of the guest in the host (co-conformation).

Full text of DOI:10.1021/ja011866n

Published on Web 12/19/2001
A Joint Structural, Kinetic, and Thermodynamic Investigation
of Substituent Effects on Host-Guest Complexation of
Bicyclic Azoalkanes by â-Cyclodextrin
Xiangyang Zhang, Gabriela Gramlich, Xiaojuan Wang, and Werner M. Nau*
Contribution from the Department of Chemistry, UniVersity of Basel, Klingelbergstrasse 80,
CH-4056 Basel, Switzerland
Received July 31, 2001
Abstract: Derivatives of the azoalkane 2,3-diazabicyclo[2,2,2]oct-2-ene (1a) with bridgehead 1,4-dialkyl
(1b), 1,4-dichloro (1c), 1-hydroxymethyl (1d), 1-aminomethyl (1e), and 1-ammoniummethyl (1f) substituents
form host-guest inclusion complexes with â-cyclodextrin. They were employed as probes to assess
substituent effects on the kinetics and thermodynamics of this complexation by using time-resolved and
1
steady-state fluorimetry, UV spectrophotometry, induced circular dichroism (ICD) measurements, and H
NMR spectroscopy. The kinetic analysis based on quenching of the long-lived fluorescence of the azoalkanes
8
8
-1
by addition of host provided excited-state association rate constants between 2.6 × 10 and 7.0 × 10 M
-
1
-1
-1
-1
-1
-1
s
. The binding constants for 1a (1100 M ), 1b (900 M ), 1c (1900 M ), 1d (180 M ), 1e (250 M ),
-1
and 1f (ca. 20 M ) were obtained by UV, NMR, and ICD titrations. A positive ICD signal of the azo absorption
around 370 nm was observed for the â-cyclodextrin complexes of 1a, 1d, and 1f with the intensity order
1
a . 1d ≈ 1f, and a negative signal was measured for those of 1b, 1c, and 1e with the intensity order 1c
1b ≈ 1e. The ICD was employed for the assignment of the solution structures of the complexes, in
particular the relative orientation of the guest in the host (co-conformation).
<
Introduction
which arises from the interaction between an achiral chro-
6
,7,10,11
mophore with the chiral host molecules
has become a
Cyclodextrins (CDs) are naturally occurring water-soluble
container-type host molecules. Their host-guest complexes
powerful tool for obtaining structural information on the relative
orientation of the guest in the host, i.e., the so-called co-
1
with organic guest molecules have application potential in
1
2
conformation.
catalysis, photochemistry,3 drug delivery,4 and analytical
2
Relatively little is also known about aspect (1), i.e., the
association and dissociation kinetics of guest molecules with
5
chemistry. In addition, they are attractive models for molecular
recognition phenomena like enzyme-substrate or drug-target
interactions.
Thermodynamic properties of CD complexes have been
extensively studied. In addition to thermodynamic data,
knowledge of (1) the kinetics of host-guest complexation as
well as (2) structural details is invaluable to assess the
functionality of a particular host system and to develop
structure-reactivity relationships in supramolecular chemistry
in general. With respect to the latter aspect (2), crystallographic
structures of several CD complexes have been obtained from
1
3
CDs. For exceptionally slow kinetics, the rate constants of
CD complex formation can be assessed by stopped-flow
14,15
16,17
methods
or NMR spectroscopy.
However, the inclusion
6,7
of small guest molecules into CDs as well as other host
structures with sufficiently large openings of the host cavity,
18
e.g., calixarenes, is typically a fast process with rate constants
7
8
-1 -1
in the order of 10 -10 M s . While such fast rate constants
of supramolecular association may be desirable from certain
application points of view, e.g., to allow high turnover rates in
catalytic processes, their accurate determination presents an
8
X-ray and neutron diffraction data, but relatively little is known
about the solution structures of these complexes. NMR tech-
niques have proven suitable to assess the depth of inclusion of
the guest and to differentiate between inclusion and association
(10) Harata, K.; Uedaira, H. Bull. Chem. Soc. Jpn. 1975, 48, 375-378.
(
11) Zhdanov, Y. A.; Alekseev, Y. E.; Kompantseva, E. V.; Vergeichik, E. N.
Russ. Chem. ReV. (Engl. Transl.) 1992, 61, 563-575.
(12) Balzani, V.; Credi, A.; Raymo, F. M.; Stoddart, J. F. Angew. Chem., Int.
Ed. 2000, 39, 3348-3391.
9
complexes. In addition, the induced circular dichroism (ICD)
(
13) Petrucci, S.; Eyring, E. M.; Konya, G. In ComprehensiVe Supramolecular
Chemistry; Atwood, J. L., Ed.; Pergamon, New York, 1996; Vol. 8, pp
483-497.
(
(
(
(
(
(
(
(
(
1) Szejtli, J. Chem. ReV. 1998, 98, 1743-1753.
2) Takahashi, K. Chem. ReV. 1998, 98, 2013-2033.
(14) Yoshida, N. J. Chem. Soc., Perkin Trans. 2 1995, 2249-2256.
(15) Abou-Hamdan, A.; Bugnon, P.; Saudan, C.; Lye, P. G.; Merbach, A. E. J.
Am. Chem. Soc. 2000, 122, 592-602.
(16) Yim, C. T.; Zhu, X. X.; Brown, G. R. J. Phys. Chem. B 1999, 103, 597-
602.
3) Bortolus, P.; Monti, S. AdV. Photochem. 1996, 21, 1-133.
4) Uekama, K.; Hirayama, F.; Irie, T. Chem. ReV. 1998, 98, 2045-2076.
5) Hedges, A. R. Chem. ReV. 1998, 98, 2035-2044.
6) Connors, K. A. Chem. ReV. 1997, 97, 1325-1357.
7) Rekharsky, M. V.; Inoue, Y. Chem. ReV. 1998, 98, 1875-1917.
8) Harata, K. Chem. ReV. 1998, 98, 1803-1827.
(17) Ghosh, M.; Zhang, R.; Lawler, R. G.; Seto, C. T. J. Org. Chem. 2000, 65,
735-741.
(18) Franchi, P.; Lucarini, M.; Pedulli, G. F.; Sciotto, D. Angew. Chem., Int.
Ed. 2000, 39, 263-266.
9) Schneider, H.-J.; Hacket, F.; R u¨ diger, V.; Ikeda, H. Chem. ReV. 1998, 98,
1
755-1785.
254 VOL. 124, NO. 2, 2002
9
J. AM. CHEM. SOC.
10.1021/ja011866n CCC: $22.00 © 2002 American Chemical Society

Host−Guest Complexation of Bicyclic Azoalkanes
A R T I C L E S
experimental challenge, since tailor-made probes (as guest
molecules) and fast time-resolved techniques are required for
their measurement. These include ultrasonic relaxation tech-
1
9
20
niques, temperature jump measurements, and more frequently
2
1,22
the absorption, emission, and quenching of triplet probes.
The latter has been recently reviewed.23 Besides these direct
spectroscopic methods, EPR line broadening was also exploited
to examine the association kinetics for complexation of persistent
nitroxide radicals with CDs.2
4,25
We have recently introduced a fluorescence-based method
2
6
for assessing the kinetics of complexation by CDs. In this
photophysical method, 2,3-diazabicyclo[2.2.2]oct-2-ene (1a) is
employed as a guest molecule, which serves also as a “dynamic”
probe to monitor the kinetics of host-guest complexation by
the fluorescence quenching inside the CD cavity. The exceed-
ingly long-lived fluorescence of azoalkane 1a, e.g., 730 ns in
performed on silica gel (60-200 µm). The azoalkane 2,3-diazabicyclo-
[2.2.2]oct-2-ene (1a)35 and its derivatives 1b, 1c, 1d, and 1e were
synthesized according to literature procedures. The detailed procedure
and spectroscopic data for 1d, which have not been previously reported,
are given below. The azoalkanes were purified by sublimation at
reduced pressure (not for 1d) and subsequent 2-fold recrystallization
29
36
37
38
27,28
deaerated D2O,
is the prerequisite for the direct measurement
of absolute kinetic rate data by fluorescence, since the associa-
26
tion process must occur within the lifetime of the excited state.
In addition to the kinetic measurements, azoalkane 1a offered
the possibility of thermodynamic measurements through several
alternative methods, and it was also possible to assess the
from n-hexane (1a, 1b, 1e), methanol (1c), or diethyl ether (1d). D
> 99.9%, Glaser AG, Basel, Switzerland) was used as solvent for all
measurements.
2
O
(
29,30
solution structure of the resulting CD complex by ICD.
The
characteristic ICD of the n,π* chromophore band around 370
nm depends sensitively on the alignment of this small chro-
mophore within the CD complex, thus providing a unique tool
1-(Hydroxymethyl)-2,3-diazabicyclo[2.2.2]oct-2-ene (1d). The start-
ing material was the urazole 1-(hydroxymethyl)-4-methyl-2,4,6-
triazatricyclo[5.2.2.02 ]undecane-3,5-dione, which was synthesized
according to literature37 from 2,3-dihydrobenzyl alcohol and 4-methyl-
urazole and subsequent hydrogenation.37 A 2.5 g (11.1 mmol) amount
of the resulting urazole was dissolved in 50 mL of 2-propanol, and
KOH pellets (4.8 g, 85.5 mmol) were added in small portions while
stirring. After the solution was refluxed under argon for 15 h and cooled
to room temperature, the solids were filtered off and washed with
.6
for structure determination2
9-32
in aqueous solution.
In this work, we employ bridgehead-substituted derivatives
of the parent azoalkane 1a as new probes to examine the effect
of molecular structure of the guest molecule on the association
rate constant and to compare it with the thermodynamics of
association and the co-conformations of the host-guest com-
plexes: 4-methyl-1-isopropyl-2,3-diazabicyclo[2.2.2]oct-2-ene
2-propanol. Rotary evaporation of the filtrates gave a slurry which was
suspended in CH Cl and filtrated. After removal of solvent by rotary
2
2
(1b), 1,4-dichloro-2,3-diazabicyclo[2.2.2]oct-2-ene (1c), 1-hy-
evaporation, the product was purified by silica gel chromatography (1.20
g, 8.55 mmol, 77%). Recrystallization from diethyl ether afforded
colorless crystals of 1d (mp 75-76 °C). UV (benzene): λmax 380 nm,
droxymethyl-2,3-diazabicyclo[2.2.2]oct-2-ene (1d), and 1-amino-
methyl-2,3-diazabicyclo[2.2.2]oct-2-ene in its neutral (1e) and
protonated forms (1f) at pH 11 and pH 5, respectively. â-CD,
which is composed of seven R-D-glucose units, was preferred
over the smaller R-CD and the larger γ-CD forms, since it shows
-
1
-1 1
ꢀ 220 cm
CH ), 3.30 (1 H, s br, OH), 4.06 (2 H, s, CH
CH). 13C NMR (101 MHz, CDCl
): δ 22.0 (2 C, CH
CH ), 62.3 (CH), 67.3 (CH OH), 68.0 (C ). Anal. Calcd for
O: C, 59.97; H, 8.63; N, 19.98; O, 11.41. Found: C, 60.14;
M
. H NMR (400 MHz, CDCl
3
): δ 1.33-1.79 (8 H, m,
OH), 5.25 (1 H, s br,
), 23.3 (2 C,
2
2
3
2
26,33,34
2
2
q
a 1-2 orders of magnitude stronger binding (e.g., for 1a).
7 12 2
C H N
Experimental Section
H, 8.60; N, 20.04; O, 11.47.
Spectroscopic Measurements. All experiments were performed at
All commercial materials, including â-CD, were obtained from Fluka
or Aldrich and were used as received. Column chromatography was
ambient temperature in D
and the ammonium form 1f, the pH values were adjusted to ca. 11 and
5 (by addition of NaOD or D SO ) to bypass complications from the
protonation equilibria, i.e., to work well below or above the pK value
of the guest (pK ) 9.2, see below), while avoiding deprotonation of
the host, cf. pK
32 pH meter with a combined pH glass electrode (METROHM,
2
O. For experiments with the amine form 1e
(
(
(
(
19) Nishikawa, S.; Yokoo, N.; Kuramoto, N. J. Phys. Chem. B 1998, 102,
830-4834.
20) Cramer, F.; Saenger, W.; Spatz, H.-C. J. Am. Chem. Soc. 1967, 89, 14-
0.
21) Monti, S.; Flamigni, L.; Martelli, A.; Bortolus, P. J. Phys. Chem. 1988,
2
4
4
a
2
a
6
a
(â-CD) ) 12.34. pH readings were taken from a
9
2, 4447-4451.
6
22) Okano, L. T.; Barros, T. C.; Chou, D. T. H.; Bennet, A. J.; Bohne, C. J.
Phys. Chem. B 2001, 105, 2122-2128.
Switzerland). Most spectroscopic experiments were performed in 4 mL
cuvettes by using 3 mL of 4 mM stock solutions, except for 1c (1.0
mM). Deaerated solutions, where required, were obtained by two
freeze-pump-thaw degassing cycles using homemade quartz cells (4
(
(
(
23) Bohne, C. Spectrum 2000, 13 (3), 14-19.
24) Kotake, Y.; Janzen, E. G. J. Am. Chem. Soc. 1992, 114, 2872-2874.
25) Lucarini, M.; Luppi, B.; Pedulli, G. F.; Roberts, B. P. Chem. Eur. J. 1999,
5
, 2048-2054.
(
(
26) Nau, W. M.; Zhang, X. J. Am. Chem. Soc. 1999, 121, 8022-8032.
27) Nau, W. M.; Greiner, G.; Rau, H.; Wall, J.; Olivucci, M.; Scaiano, J. C. J.
Phys. Chem. A 1999, 103, 1579-1584.
×
1 × 1 cm) with high-vacuum Teflon stopcocks.
A XeF excimer laser pulse from a Lambda Physics COMPex 205
(
(
(
28) Nau, W. M. J. Am. Chem. Soc. 1998, 120, 12614-12618.
29) Zhang, X.; Nau, W. M. Angew. Chem., Int. Ed. 2000, 39, 544-547.
30) Mayer, B.; Zhang, X.; Nau, W. M.; Marconi, G. J. Am. Chem. Soc. 2001,
laser (351 nm, pulse width ca. 20 ns, pulse energy 40-175 mJ) or a
1
23, 5240-5248.
(35) Askani, R. Chem. Ber. 1965, 98, 2551-2555.
(36) L u¨ ttke, W.; Schabacker, V. Justus Liebigs Ann. Chem. 1965, 687, 236-
240.
(37) Engel, P. S.; Horsey, D. W.; Scholz, J. N.; Karatsu, T.; Kitamura, A. J.
Phys. Chem. 1992, 96, 7524-7535.
(38) Hudgins, R. R.; Huang, F.; Gramlich, G.; Nau, W. M. J. Am. Chem. Soc.,
in press.
(
(
(
31) Krois, D.; Brinker, U. H. J. Am. Chem. Soc. 1998, 120, 11627-11632.
32) Bobek, M. M.; Krois, D.; Brinker, U. H. Org. Lett. 2000, 2, 1999-2002.
33) Eftink, M. R.; Andy, M. L.; Bystrom, K.; Perlmutter, H. D.; Kristol, D. S.
J. J. Am. Chem. Soc. 1989, 111, 6765-6772.
(
34) Stoedeman, M.; Berg, U.; Svensson, A. J. Chem. Soc., Faraday Trans.
1
998, 94, 1737-1741.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002 255

A R T I C L E S
Zhang et al.
Table 2. Binding Constants, ICD Ellipticities, and Association
Rate Constants of the â-CD Complexes of Azoalkanes 1
∆
ꢀ/
k
ass
/
diss
k /
105 s-1
K/M-
1a
θ/mdegb
(M-1 cm
)
-1 c
10 M s-1 d
8
-1
guest
1a
1b
1c
1d
1e
1f
1100 ( 200
900 ( 150
1900 ( 300
180 ( 20
19.3
-4.0
-5.6
5.6
-1.9
1.1
+0.162
-0.034
-0.119
+0.069
-0.025
+0.045
4.8 ( 0.5
f
7.0 ( 0.7
3.3 ( 0.4
2.6 ( 0.6
3 ( 1h
(4.4)e
f
e
(3.7)
(18)
(10)
e
e
250 ( 20
20 ( 10g
120 ( 50
h
a
Ground-state binding constants from UV and H NMR. b ICD ellipticity
1
at band maximum for 4 mM solutions of 1 (1.0 mM for 1c) in the presence
of â-CD (12.0 mM) in D2O. c Molar ellipticity calculated as ∆ꢀ )
θ/(32982 l c), cf. ref 76, with l ) path length and c ) actual concentration
of complex in solution calculated by using the binding constants in this
table. d Association rate constants in the excited state from the time-resolved
and steady-state fluorescence quenching data by employing simultaneous
Figure 1. UV absorption spectra of azoalkane 1a (s), 1b (‚‚‚), 1c (- - -),
1
d (s‚‚‚s), and 1e (s s‚s s), all in D2O. The spectrum of 1f is virtually
e
fitting according to eqs 6 and 9. Dissociation rate constant estimated as
superimposable with that of 1e and is not shown for clarity. The inset shows
the expanded spectra from 300 to 400 nm.
kass/K. f Quenching effect too small to analyze kinetics, cf. text. g Determined
h
from ICD titration. Obtained from global fitting to eqs 6 and 7, cf. text.
Table 1. Photophysical Properties of Azoalkanes 1 in D2O,
n-Hexane, and Their â-CD Complexes
All azoalkanes 1a-f display exceedingly long fluorescence
lifetimes in deaerated D2O between 730 and 920 ns (Table 1).
Table 1 contains also the fluorescence lifetime in the corre-
sponding â-CD‚1 complex (τCD values), which can be obtained
from kinetic measurements in the presence of â-CD (see below).
These reflect the reactivity of the excited azoalkane toward
abstractable C-H hydrogens in the CD cavity. The method to
determine the association rate constants (see below) relies on a
fast quenching inside the CD cavity, i.e., small τCD values.
Otherwise, exit from the cavity may compete with deactivation,
which complicates the data analysis. The observed trend of the
τCD values, i.e., 1c , 1a, 1f < 1d < 1e , 1b, corresponds to
that observed for the solvent-induced quenching by n-hexane,
cf. lifetimes in this solvent in Table 1: 1c (30 ns) , 1a (340
ns) ≈ 1d (335 ns) < 1e (420 ns) , 1b (770 ns). This correlation
is reasonable since the same quenching mechanism applies for
both n-hexane and the interior of â-CD, namely an “aborted”
1
τ/ns [deaerated]b
λ
max/nm [ꢀ/(M- cm-1)]
λ
iso
nma
/
D O n-hexane â-CD‚1
2
c
2
D O
n-hexane
â-CD‚1
(τ
0
)
(τCD)
1a
1b
1c
1d
1e
1f
365 [48] 376 [200] 371
373 [54] 383 [200] 379
358 [40] 367 [190] 364
368 [54] 380 [140] 370
369 [43] 381 [120] 373
358
376
356
371
371
505 [730] [340]
590 [810] [770]
80
430
40
95
110
75
705 [750]
[30]
565 [840] [335]
660 [880] [420]
e
e
370 [43]
d
370
(370) 670 [920]
d
a
b
Isosbestic point in the UV titration by â-CD in D2O. Fluorescence
c
lifetime in aerated [deaerated] solution; 5% error. Fluorescence lifetime
in the â-CD complex in aerated D2O obtained by simultaneous fitting of
the time-resolved and steady-state fluorescence data according to eqs 6 and
d
e
9
; 10% error. Insoluble in n-hexane. No significant UV shift was
observed upon addition of â-CD.
Continuum Minilite Nd:YAG laser (355 nm, pulse width 3-7 ns, 7
mJ) was used for excitation to obtain the time-resolved fluorescence
decays. The decays were monitored with a monochromator-photo-
multiplier setup at 420-500 nm, depending on signal intensity. The
kinetic traces were registered by means of a transient digitizer and
analyzed by nonlinear least-squares fitting of monoexponential or
biexponential decay functions, where appropriate. Steady-state fluo-
rescence spectra were measured with a Spex Fluorolog fluorimeter or
with an FLS900 setup from Edinburgh Instruments. Steady-state
fluorescence quenching experiments with â-CD were performed with
excitation at the isosbestic points (λiso) which were obtained from UV
titrations. UV spectra were obtained with a Hewlett-Packard 8452 diode
array spectrophotometer (2 nm resolution) or with a Perkin-Elmer
Lambda 19 spectrophotometer (0.1 nm resolution). The NMR spectra
were obtained with a Bruker DPX 400 MHz Avance NMR spectrom-
eter. Induced circular dichroism spectra were recorded with a Jasco
J-720 spectropolarimeter (0.2 nm resolution, 25 accumulations, 1 cm
cell) or an AVIV circular dichroism spectrometer (model 62A DS,
Lakewood, NJ).
27,39
hydrogen abstraction from C-H bonds.
Shorter fluorescence lifetimes correspond to higher reactivity
for hydrogen abstraction. Accordingly, the chlorinated derivative
1
1
c is the most reactive one, while the bis-alkylated derivative
b displays the lowest reactivity. The hydroxymethyl- and
4
0
aminomethyl-substituted azoalkanes fall in between. Hence, it
appears that electron-withdrawing groups enhance the reactivity
of these azoalkanes toward hydrogen donors, while electron-
4
0
donating groups lower it. This variation can be related to the
changes in excitation energy, since the chlorinated derivative
exhibits a significantly hypsochromically shifted absorption
maximum (Figure 1), corresponding to a higher excitation
energy. In addition to variations of the excitation energy,
different contributions of charge transfer to the quenching
4
1,42
mechanism may apply.
Acid-Base Equilibrium of Azoalkanes 1e and 1f. The pKa
value of azoalkane 1e can be readily determined through
1
Results
measurement of the H NMR chemical shift of the R-CH2
protons in D2O (δ1e ) 3.18 ppm and δ1f ) 3.62 ppm) as a
function of pH. A pKa value of 9.2 in D2O was obtained by
Photophysical Properties. The UV spectra of 1a-f in D2O
are quite similar (Figure 1). The substituents have only a minor
effect on the n,π* transition near 370 nm (Table 1), causing a
slight bathochromic shift for electron-donating groups (1b) and
a hypsochromic one for electron-withdrawing groups (1c). A
substituent-dependent shift was also observed for the π,π*
(39) Nau, W. M.; Greiner, G.; Wall, J.; Rau, H.; Olivucci, M.; Robb, M. A.
Angew. Chem., Int. Ed. 1998, 37, 98-101.
(
40) Zhang, X.; Nau, W. M. J. Inf. Recording 2000, 25, 323-330.
(41) Pischel, U.; Zhang, X.; Hellrung, B.; Haselbach, E.; Muller, P.-A.; Nau,
W. M. J. Am. Chem. Soc. 2000, 122, 2027-2034.
(
42) Sinicropi, A.; Pischel, U.; Basosi, R.; Nau, W. M.; Olivucci, M. Angew.
3
0
transition below 220 nm (Figure 1).
Chem., Int. Ed. 2000, 39, 4582-4586.
256 J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002

Host−Guest Complexation of Bicyclic Azoalkanes
A R T I C L E S
Scheme 1
Figure 2. ICD spectra of 4.0 mM solutions (1.0 mM for 1c) of azoalkane
nonlinear least-squares fitting of the chemical shifts according
to eq 1.43 This value falls in the range of those for other amines
1
a (s), 1b (‚‚‚), 1c (- - -), 1d (s‚‚‚s), 1e (s s‚s s), and 1f (- - -) in
the presence of â-CD (12.0 mM) in D O.
2
+
Scheme 2
δ K + δ [H ]
1e
a
1f
δobs
)
(1)
+
K + [H ]
a
with electron-withdrawing â substituents.44 Note that the azo
group in 1e can be considered as such an electron-withdrawing
4
5
â substituent (cf. Hammett σ value of 0.39 for p-NdNPh).
Binding Constants. According to Scheme 1, a ground-state
equilibrium exists between the free, uncomplexed form (1) and
the complex (CD‚1), which is characterized by the binding
constant K (Table 2). The six differently substituted guest
molecules show significant variations of the binding constants,
which can be obtained by UV, NMR, ICD, and fluorescence
titrations.26 At least two independent techniques, generally UV
and NMR, were employed for each guest, and the results were
the same within error. The obtained data confirmed the
formation of 1:1 complexes in all cases. For the protonated
azoalkane 1f, the UV and NMR spectral changes were too small
to allow a reliable determination of the binding constant by these
methods. Here, the more sensitive ICD measurement was
are ICD silent, this titration method may in some cases be more
sensitive to determine the binding constants. For 1f, ICD proved
to be the method of choice to determine the binding constant
since shifts in the UV and NMR spectra were too small to allow
accurate titrations.
Addition of â-CD to a solution of 1e results in a multiple
3
3,47
acid-base equilibrium (Scheme 2),
and the variation of the
ICD spectra with increasing pH reveals an interesting inversion
in the sign of the ICD (Figure 3). This can be related to a
structural change upon deprotonation (cf. Discussion).
From the thermodynamic cycle in Scheme 2, it is possible to
determine the pK ′ value of the â-CD‚1e complex from eq 2.
a
-
1
employed to derive a much smaller binding constant (20 M ,
One obtains a pK ′ value of 8.1, which reveals an apparently
a
-
1
see below) than for the amine 1e (250 M ).
K ′ ) K K /K
a 1e 1f
(2)
Induced Circular Dichroism. The analysis of the sign and
intensity of the ICD effect arising from the n,π* transition of
azoalkanes has recently been recognized as a powerful tool for
a
lower basicity for the â-CD‚1e complex compared to the free
amine 1e (pKa ) 9.2). This is synonymous with a higher
preference of the ammonium compared to the amine form to
remain in the aqueous phase. The known values for the two
binding constants (Table 2) and the two pKa values allow one
also to fit the pH profile of the variation of the observed
2
9-32
structure elucidation of their CD complexes in solution.
The ICD spectra of all â-CD complexes of azoalkanes 1a-f
were measured (Figure 2).46 The molar ellipticities (∆ꢀ) were
quantified through the respective concentrations and were
corrected for the percentage of complexed guest by using the
known binding constants (Table 2). For the determination of
ICD effects of the amine form 1e and the ammonium form 1f,
the pH values were again adjusted to 11 and 5. The parent
compound 1a gave a strong, positive ICD signal, and its
monosubstituted derivatives 1d and 1f showed also a positive,
but weaker signal, whereas the disubstituted derivatives 1b and
(
46) ICD spectra were also obtained for azoalkanes 1 in the presence of R-CD
and γ-CD as inducers. The experimental ICD intensities (θ/mdeg) at the
band maximum of 4 mM solutions of azoalkanes 1 (1.0 mM for 1c) in the
2
presence of R-CD (12.0 mM) in D O were as follows: +14 for 1a, +0.5
for 1b, 0 for 1c, and +0.5 for 1d. In the presence of γ-CD (12.0 mM) the
intensities were +1.5 for 1a, +8.0 for 1b, +2.1 for 1c, and +2.5 for 1d.
As can be seen, the ICD signals were consistently positive, but for 1b-d
in the presence of R-CD, hardly any effect was noticed, which paralleled
the absence of a significant effect on the UV spectra in these cases. The
binding constants for R-CD and γ-CD are generally much lower than those
1
c, as well the amino derivative 1e, gave rise to negative ICD
-
1
-1
effects. Since the ICD signals report selectively on the popula-
tion of the â-CD complexes, while the uncomplexed components
for â-CD; e.g., for R-CD‚1a ca. 50 M , for â-CD‚1a 1100 M , for γ-CD‚
1 -1 -1
1
a ca. 6 M- (ref 26); for R-CD‚1b <3 M , for â-CD‚1b 900 M , and
-1
for γ-CD‚1b 150 M (this work). More detailed investigations with the
other cyclodextrins were not undertaken, since the weaker binding imposes
experimental complications. Moreover, R-CD appears to be too small to
form deep inclusion complexes with the guests, cf. absence of binding for
1b-d, while γ-CD may be too large to cause a clear preference for a
particular co-conformation.
(
43) Arrowsmith, C. H.; Guo, H.; Kresge, A. J. J. Am. Chem. Soc. 1994, 116,
8
890-8894.
(
44) Jencks, W. P.; Regenstein, J. In CRC Handbook of Biochemistry and
Molecular Biology, 2nd ed.; Sober, H. A., Ed.; CRC Press: Cleveland,
OH, 1970; pp J187-J226.
(47) Yoshida, N.; Seiyama, A.; Fujimoto, M. J. Phys. Chem. 1990, 94, 4254-
(45) Hansch, C.; Leo, A.; Taft, R. W. Chem. ReV. 1991, 91, 165-195.
4259.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002 257

A R T I C L E S
Zhang et al.
Figure 4. Steady-state and kinetic (inset) fluorescence quenching plots
for 1c (1.0 mM) by â-CD in D2O and fits according to eqs 6 and 9 and the
parameters in Table 2 (solid lines).
Figure 3. ICD spectra for azoalkane 1e (4.0 mM) in the presence of â-CD
(
12.0 mM) in D2O at varying pH values. The inset shows a plot of the
ellipticity at 381 nm versus pH fitted according to eq 3.
and in the presence of CD (I) is plotted against the total
ellipticity (θobs), e.g., in the region of maximum variation at
concentration of CD ([CD]0) according to the previously
48
3
81 nm. We have derived eq 3 to describe the pH dependence
2
6
introduced eq 4, in which the concentration of uncomplexed
azoalkane, [1], is defined by eq 5. Because the lifetime of the
uncomplexed form (τ0 in eq 4) and the binding constants (K in
eq 5, Table 2) are known otherwise (Table 1), such plots (see
Figure 4) are suitable to provide values for the association rate
constant (kass), the dissociation rate constant (kdiss), and the
fluorescence lifetime of the complex (τCD). τQ in eq 4 is the
lifetime of the free excited guest, i.e., before it either deactivates
or complexes with â-CD.
49
of the ellipticity (l ) path length). The resulting fit (inset in
Figure 3) corroborates the applicability of Scheme 2. It should
be noted that the analytical form of eq 3 should be generally
applicable to describe related pH effects on ICD spectra.
θobs
)
)
∆ꢀ [CD‚1e] + ∆ꢀ [CD‚1f]
1e 1f
3
2982 l
+
∆ꢀ PK ′ + ∆ꢀ P[H ]
(3)
1e
a
1f
+
τQ
τCD
τCD
I
[
1]
[1]
[1]
[
1e] + [CD]0
K ′([H ] + K )
0
a
a
)
+
QR
+ 1 -
R
(4)
with P )
+
-
(
)
+
+
2
I
0
[1] τ
0
[1]0
τ
0
[1]0
τ
0
0
2
(K ′ + [H ]) 2K K (K ′ + [H ])
a
1e
a
a
k τ [CD]
0
+
+
2
ass
0
{
(K K ([1e] + [CD] )(K ′ + [H ]) + K ′([H ] + K )) -
with Q )
,
1
e
a
0
0
a
a
a
1
+ k τ [CD]
ass 0 0
+ k τ [CD] + kdissτ
ass 0 0
2
2
+
2
1/2
+
2
4
[1e] [CD] K K (K ′ + H ]) } /2K K (K ′ + [H ])
0
0
1e
a
a
1e
a
a
1
0
1
R )
, and τ )
Q
1
+ k τ [CD] + kdissτ
k + k [CD]
0 ass 0
The maxima of the ICD bands matched the UV absorption
ass
0
0
CD
maxima for all â-CD‚1 complexes except for 1b and 1e. A
hypsochromic shift from 379 nm (UV) to 370 nm (ICD) was
registered for 1b and a bathochromic shift from 373 nm (UV)
to 382 nm (ICD) for 1e. The ICD spectrum for 1e exhibited
also a distorted band shape, cf. pH 11 spectrum in Figure 2.
These shifts may be indicative of a large co-conformational
variability or nonuniformity (cf. Discussion).
2
[
1] ) (K[1] - K[CD] - 1 + {(K[1] + K[CD] + 1) -
0
0
0
2
0
1/2
4K [1] [CD] } )/2K (5)
0
0
The kdiss values recovered from fits according to eq 4 were
very small for azoalkanes 1a-e, and the errors were as large
as the values themselves. This result confirmed that exit from
the excited complex must be indeed insignificant for azoalkanes
Association Rate Constants. It should be noted in advance
that our kinetic interpretations of the kinetic parameters are
2
6
1
a-e, as previously demonstrated for 1a. This was not the
case for the ammonium derivative 1f, for which the value of
kdiss, within a large error, was only 1 order of magnitude below
that of kass. This indicated the significance of exit in this case.
If one assumes that the binding constants are similar in the
ground and excited states (see Discussion), one may obtain
estimates of the dissociation rate constants (kdiss, values in
parentheses in Table 2). These values are much smaller than
kCD ) 1/τCD, the quenching rate constant inside the complex,
except for 1f. This supports the notion that kdiss cannot be
neglected for 1f, but it can be for the other derivatives 1a-e,
for which the simplified eq 6 can then be applied for fitting
and extracting the values of kass.
based on Scheme 1, which has been demonstrated to hold for
the parent compound 1a.26 Excitation by UV light around 370
nm populates the fluorescent singlet-excited state of the free
azoalkane (1*) and the complex (CD‚1*). The environmental
dependence of the fluorescence of 1 (quenching by abstractable
C-H bonds) results in a shorter fluorescence lifetime within the
complex (τCD) compared to the uncomplexed form (τ0), see
above. The kinetics can be analyzed through steady-state
fluorescence measurements, which rely on the concomitant
decrease of integrated fluorescence intensity upon addition of
CD. These measurements require excitation at the isosbestic
point (λiso, Table 1).26 The intensity ratio in the absence (I0)
τCD
) τ0
[1]
[1]0(
τCD
1
(
48) Yi, Z.; Chen, H.; Huang, Z.; Huang, Q.; Yu, J. J. Chem. Soc., Perkin Trans.
I
I0
2
2000, 121-127.
+
1 -
, for kdiss , kCD
)(
)
τ0 1 + k τ [CD]
(49) The computer program Pro Fit 5.5.0 (QuantumSoft, Zurich) was employed
for individual fitting and global data analysis.
ass
0
0
(6)
2
58 J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002

Host−Guest Complexation of Bicyclic Azoalkanes
A R T I C L E S
The time-resolved fluorescence decays were obtained by
laser-flash photolysis. The resulting kinetics can be analyzed
according to the formal kinetic scheme derived by Andriessen
et al. In our fluorescence measurement, both the free fluoro-
phores and the complexes are simultaneously excited due to
be readily rationalized. The first time constant corresponds to
a static component (τCD), which is due to quenching inside the
complex (Scheme 1) and therefore independent of CD concen-
tration. The second, dynamic component (τQ) corresponds to
the lifetime of the free form before complexation, which depends
on the bimolecular association rate constant (kass) and the CD
concentration. kass can then be extracted from the dependence
of the τQ values on the total CD concentration (eq 9). Here, the
analysis according to eq 8 rather than eq 7 has the advantage
of a simple linearized data representation (eq 9) through a
regression analysis of a kinetic quenching plot (inset in Figure
4).
5
0
the very similar UV absorption spectra. Moreover, since the
fluorescence spectra are virtually superimposable,2
6,27
the sum
of the fluorescence from both species is observed. If one further
ensures equal absorption probabilities by excitation at the
isosbestic point (λiso, Table 1) or corrects for differential
absorption in cases where the excitation wavelength cannot be
adjusted (e.g., for laser excitation), and in addition relies on
the observation that the natural radiative lifetimes display no
pronounced solvent dependence,27 the emission intensities of
each species are linearly related to its ground-state concentration.
The time dependence of the observed fluorescence intensity,
I(t), normalized to unity at t ) 0, is then given by eq 7, where
τ0
)
1 + k τ [CD]
0
(9)
ass
0
τQ
In a final procedure, to obtain the best estimates of the
association rate constants for azoalkanes 1a-e, where exit is
insignificant, the time-resolved and steady-state data were
subjected to a simultaneous nonlinear fitting analysis according
to eqs 6 and 9, putting arbitrarily an equal weight to both data
[1] is defined by eq 5.
I(t) )
-
[1](k + γ ) - [CD‚1](k + γ2)
0 2 CD
1
γ t
e
+
4
9
(
(
2
)
)
sets. The resulting values are entered in Table 2. For the
ammonium derivative 1f, the biexponential fitting of the excited-
state lifetimes did not yield a static, i.e., constant, component
at different CD concentrations. This, along with the indications
from the steady-state experiments (see above), suggested that
exit cannot be neglected and that eqs 6 and 9 no longer apply.
This prevented also an analysis of the time-resolved data
[
1]0 (k + k [CD] - k - kdiss) + 4k kdiss[CD]0
x
0
ass
0
CD
ass
[
1](k + γ ) + [CD‚1](k + γ1)
0
1
CD
γ t
2
e
2
[
1]0 (k + k [CD] - k - kdiss) + 4k kdiss[CD]0
x
0
ass
0
CD
ass
(7)
The time constants γ1 and γ2 in eq 7 are defined as follows:
4
9
according to eq 9. A global analysis, i.e., simultaneous fitting
of the individual decay traces according to eq 7, is indicated in
1
2
γ ) - {k + k [CD] + k + kdiss
-
1
0
ass
0
CD
50
such cases, which includes kdiss as a fitting parameter.
Moreover, we have also included the steady-state treatment
according to eq 4, which considers also kdiss, in the final analysis,
putting again an equal weight to the steady-state and the
combined time-resolved data. The resulting constants for 1f are
entered in Table 2. It should be mentioned that the binding
constant obtained from the ratio of the recovered rate constants
2
(
k + k [CD] - k - kdiss) + 4k k [CD]0}
x
0 ass 0 CD ass diss
1
2
γ ) - {k + k [CD] + k + kdiss
+
2
0
ass
0
CD
2
(
x
k + k [CD] - k - kdiss) + 4k k [CD]0}
0 ass 0 CD ass diss
-
1
(
kass/ kdiss ) 25 M ) supports also the suggestion that the
When dissociation of the complex during its excited-state
lifetime is negligible (kdiss , kCD), eq 7 simplifies to the equation
employed in our previous work (here rearranged):
-1
excited-state binding resembles that in the ground state (20 M ),
which was obtained by ICD (see above).
For azoalkane 1b, for which the fluorescence lifetime of the
complex (τCD) was very long (430 ns) and too similar to that
of the uncomplexed guest (590 ns, under air), a reliable
differentiation of the static and dynamic components through
biexponential fitting of the time-resolved decay traces could not
be performed. In general, the differentiation of two exponential
decay components becomes a formidable task when the lifetimes
-
(τCD)-1t
-(τQ)-1
t
I(t) ) (1 - S) e
+ S e
[CD]
τ0/τCD - k τ [CD] - 1
[
1]
k
ass
τ
0
0
with S )
1 +
(8)
(
)
[1]0
ass
0
0
1
and τ )
, for kdiss , kCD
51
Q
differ by less than a factor of 2. The effects on the steady-
k + k [CD]
0
ass
0
state fluorescence intensity were also too small in relation to
the error to allow a reliable fitting according to eq 6. As a
consequence, no kinetic data for 1b can be reported. The failure
for 1b reveals the limitations of our fluorescence-based method
for determining association rate constants.
An analysis of the preexponential factors in eqs 7 and 8,
26
which are subject to a larger error, was generally not attempted,
but the two time constants (τCD and τQ) in the experimental
decay traces were extracted by biexponential fitting. For
azoalkanes 1a-e, one component was found to be constant
within error, i.e., independent of CD concentration. This
suggested that eq 8 applied, in which exit is considered
insignificant for 1a-e, as already borne out by the steady-state
analysis above. The meaning of the time constants in eq 8 can
Discussion
The changes of the UV, ICD, and NMR spectra of azoalkanes
1
upon addition of â-cyclodextrin (â-CD) can be understood in
terms of the formation of inclusion complexes, where the guest
is positioned inside a hydrophobic cavity. This is confirmed,
(
50) Andriessen, R.; Boens, N.; Ameloot, M.; De Schryver, F. C. J. Phys. Chem.
1
991, 95, 2047-2058.
(51) Grinvald, A.; Steinberg, I. Z. Anal. Biochem. 1974, 59, 583-598.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002 259

A R T I C L E S
Zhang et al.
for example, by the bathochromic shift of the near-UV azo
absorption band (Table 1) and the chemically induced H NMR
Scheme 3
1
shifts (CIS), which were observed for the inner H-3 and H-5,
yet not for the outer H-2 and H-4 protons of â-CD (the
approximate locations of these hydrogens are entered in Scheme
1
, bottom left). The CIS values are significant not only for the
upper H-3 (ca. 0.10 ppm) but also for the lower H-5 protons
ca. 0.08 ppm). This applies for all azoalkanes 1a-f when the
(
effects are extrapolated to quantitative binding. A small effect
was also observed for the H-6 protons. The observed NMR shifts
can be conclusively interpreted in terms of a deep inclusion for
all complexes, as previously corroborated for azoalkanes 1a and
2
6,29
1
b.
stoichiometry; e.g., the UV titrations show an isobestic point
Table 1) and the data from UV, NMR, fluorescence, and ICD
All data support the formation of complexes with 1:1
(
titrations can be fitted with a 1:1 complexation model.
through force-field computed structures and energies as well
as quantum-chemical calculations of ICD effects.
The Discussion is arranged such that we will first analyze
the structure of the host-guest complexes in more detail, in
particular with respect to the relative alignment of the guest to
the host, i.e., the co-conformation of the complex.12 In a second
part, substituent effects on the thermodynamics of binding will
be examined, keeping in mind a possible influence of structural
variations. Finally, we will question to which degree the kinetics
of host-guest complex formation can be understood in terms
of the observed structural and thermodynamic trends.
3
0
The co-conformational assignments of the presently examined
derivatives 1c-e are made accordingly (Scheme 3), keeping in
mind the NMR results (see above) which suggest the formation
of deep inclusion complexes in all cases. Further, we adhere to
the general notion that polar substituents reside near the
secondary hydroxyl rim while more hydrophobic groups tend
6
to protrude into the cavity. For 1b, in particular, the penetration
of the isopropyl rather than the methyl group has been
Complex Structures by Induced Circular Dichroism and
NMR. The induced circular dichroism (ICD) provides a unique
analytical tool to investigate the structures of chiral host-guest
30
substantiated by force-field calculations. The dichloro deriva-
tive 1c resembles the bis-alkylated derivative 1b in that it bears
two bridgehead substituents. The van der Waals diameters along
the bridgehead axis of 1c (ca. 8 Å) and 1b (ca. 9 Å) both exceed
6,7,11
(host-chromophore) inclusion complexes in solution.
Sev-
eral rules for the assignment of the co-conformations and the
1
the diameter of the â-CD cavity (6.0-6.5 Å) significantly,
depth of inclusion have been developed for cyclodextrins (CDs)
which promotes a frontal type of inclusion due to steric
1
0,52,53
as host systems.
In most studies, aromatic chromophores
2
9
restraints. The strong negative ICD for 1c supports this
structural assignment (Scheme 3).
have been examined, whereas aliphatic ones have received
comparably little attention. Aliphatic azoalkanes and diazirines
The van der Waals diameter of the hydroxymethyl, amino-
methyl, and ammoniummethyl derivatives 1d-f (ca. 7 Å along
bridgehead axis) lies between those for the parent 1a (ca. 6 Å)
and the disubstituted cases 1b and 1c. This reduces the
preference for frontal inclusion, but a lateral or tilted lateral
inclusion may still be viable. Indeed, the hydroxy and am-
monium derivatives 1d and 1f display positive ICD effects,
which suggest a more lateral co-conformation, as has been
previously observed and analyzed for the parent compound 1a.
Note that among the four principal co-conformations (lateral,
29-32
have recently been employed in ICD measurements,
which
has allowed several new insights into the inclusion geometries
of aliphatic guests.
The ICD method was also applied in this work to the
complexes between â-CD as chiral host and azoalkanes 1 as
achiral chromophores.46 In essence, the n,π* transition of the
azo chromophore around 370 nm gives rise to a characteristic
ICD signal when included in CD.2
9-32
The magnitude and sign
of the ICD effect are directly related to the orientation of the
electric dipole transition moment relative to the axis of the CD.
frontal, apical, basal), only the lateral one should give rise to a
According to the rule of Harata,1
0,29,30,53
which applies to
29,30
positive ICD.
Due to the angular dependence of the ICD
chromophores immersed in the CD cavity, co-conformations
with a parallel alignment produce a positive ICD signal, while
those with an orthogonal alignment produce only a half as
strong, negative ICD signal. In the case of azoalkanes, the
electric dipole transition moment points along the azo π orbital
and lies orthogonal to the plane defined by the CsNdNsC
effect in CD inclusion complexes, which predicts positive
contributions to be twice as strong as negative ones for the same
10,29,30,53
electronic transition,
a positive ICD signal is expected
even for significant deviations from the ideal angle (Harata’s
30
rule, see above).
Strikingly, the aminomethyl derivative produced a negative
ICD effect, which resembles that observed for the bis-substituted
derivatives 1b and 1c. This suggests that a frontal co-
conformation is populated for 1e. But why does the substitution
of the hydroxy for the amino group (1d versus 1e) or even the
deprotonation of the ammonium group (1f versus 1e) result in
an inversion of the ICD sign from positive to negative, i.e., a
pronounced change in the co-conformation of the host-guest
complex? We propose that differences in host-guest hydrogen
bonding are important. The intramolecular network of secondary
29
linkage. Structural assignments of the â-CD complexes of the
parent compound 1a and the bis-alkylated derivative 1b based
on ICD effects have been communicated (Scheme 3, arrows
29
indicate the direction of the electric dipole transition moment).
We refer to the respective co-conformations as lateral (1a) and
frontal (1b).29 These assignments have been recently confirmed
(
52) Kajt a´ r, M.; Horv a´ th-Tor o´ , C.; Kuthi, EÄ .; Szejtli, J. Acta Chim. Acad. Sci.
Hung. 1982, 110, 327-355.
(
53) Kodaka, M. J. Phys. Chem. A 1998, 102, 8101-8103.
260 J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002

Host−Guest Complexation of Bicyclic Azoalkanes
A R T I C L E S
Scheme 4
Scheme 5
to that depicted in Scheme 3 for 1e and does not engage in
6
2
hydrogen bonding with the enclosing host, either.
1
Based on the relative magnitude of the H NMR shifts, H-3
>
H-5 . H-6, we propose for all azoalkanes 1a-f a location
with the center of mass somewhat displaced toward the upper
rim, as previously found for 1b by force-field calculations.30
This location renders also hydrogen bonding of the functional
groups in 1d and 1f with the lower rim O-6 protons unlikely,
i.e., a 180° rotation of these guest molecules in Scheme 3. Such
alternative hydrogen bonding with the primary hydroxyl rim
has been occasionally invoked for very small guest molecules.62
However, bonding with the upper secondary hydroxyl rim as
depicted for 1d and 1f in Scheme 3 is more common.4
O2(n)‚‚‚H‚‚‚O3(n - 1) interglucose hydrogen bonds9,54,55 sta-
bilizes the macrocyclic host structures. These intramolecular
hydrogen bonds at the upper rim are particularly strong for
â-CD56 compared to R-CD and γ-CD, which is also held
8,60,61,63-66
57
responsible for the reduced solubility of â-CD. It is conceiv-
able that the intramolecular network prefers intermolecular
interactions with hydrogen bond donors over those with
hydrogen bond acceptors (Scheme 4). In fact, in most docu-
mented cases for hydrogen bonding, the guests, in particular
While we assume that the structure of the â-CD‚1e complex
in Scheme 3 gives rise to the negative ICD signal, scrutiny of
the ICD spectra reveals that the complex of azoalkane 1e differs
from the other derivatives in that it gives rise to an ICD band,
which is significantly distorted, cf. Figure 2, and shows also a
bathochromic shift relative to the UV spectrum (cf. Results). A
small shift of the band maximum of the UV and ICD spectrum
was also observed for azoalkane 1b, but not for the other four
derivatives, for which the band shape and maximum were
virtually the same. Moreover, the ICD of 1e shows a small
positive component at shorter wavelengths in addition to the
stronger negative one. These irregularities suggest a large co-
conformational variability, which has been previously cor-
7
,58-60
phenols,
function as hydrogen bond donors.
The tendency for formation of hydrogen bonds in which the
functional groups act as donors should approximately decrease
with the group acidity, which follows the well-known order
+
R-NH3 > R-OH . R- NH2. The hydroxyl and ammonium
groups in 1d and 1f may thus form hydrogen bonds with the
48,61
secondary hydroxyl rim (Scheme 4)
and, thus, adapt a tilted
lateral co-conformation (Scheme 3). In contrast, the amino group
as the poorest hydrogen bond donor may not form such
hydrogen bonds. Steric effects (as those for 1b and 1c) may
now prevail and induce a frontal co-conformation with the
expected negative ICD. Hence, the variations in the signs of
the experimental ICD effects may be due to a change of the
co-conformations of azoalkanes 1d-f as induced by hydrogen
bonding with the secondary hydroxyl rim for the hydroxyl and
ammonium but not for the amino group.
3
0
roborated for 1b. This means that complexes with different
tilt angles (R in Scheme 5) may be present in solution. Note, in
particular, that for 1e, in contrast to 1d and 1f, the co-
conformational space may not be restricted by hydrogen bonding
with the upper rim. The co-conformers with different tilt angles
may have quite different UV spectra (which reflect the depth
of inclusion into the nonpolar environment) and ICD effects
(which reflect the geometrical alignment with respect to the azo
These tentative structural assignments are in line with
literature findings. First, a lower tendency for hydrogen bonding
of the amino compared to the hydroxyl group has been
previously held responsible for a similar structural effect, namely
chromophore) and, thus, account for the observed shifts for 1b
and 1e. For example, a tilted, less deeply immersed complex
could be responsible for the small positive ICD contributions
for 1e at short wavelengths.
The possible involvement of more than one co-conformer and
a large co-conformational variability constitute standing prob-
lems in the assignment of the solution structures of host-guest
complexes. This is in contrast to crystal structures where one
co-conformer may crystallize out preferentially, which does not
58
a more axial co-conformation of anilines compared to phenols.
Second, according to crystal structures, the amino group in
p-iodoaniline in its R-CD complex adapts a position very similar
(
54) Jacob, J.; Gessler, K.; Hoffmann, D.; Sanbe, H.; Koizumi, K.; Smith, S.
M.; Takaha, T.; Saenger, W. Angew. Chem., Int. Ed. 1998, 37, 605-609.
55) Harata, K. Bull. Chem. Soc. Jpn. 1982, 55, 2315-2320.
(
(
(
(
56) Steiner, T.; Saenger, W. Angew. Chem., Int. Ed. 1998, 37, 3404-3407.
57) Szejtli, J. Cyclodextrin Technology; Kluwer: Dordrecht, 1988.
58) Kamiya, M.; Mitsuhashi, S.; Makino, M.; Yoshioka, H. J. Phys. Chem.
(62) Saenger, W.; Beyer, K.; Manor, P. C. Acta Crystallogr. 1976, B32, 120-
128.
1
992, 96, 95-99.
(63) Wenz, G.; Hoefler, T.; Wehrle, S.; Schneider, M. Polym. Prepr. (Am. Chem.
Soc., DiV. Polym. Chem.) 1998, 39, 202-203.
(64) Kano, K.; Tatsumi, M.; Hashimoto, S. J. Org. Chem. 1991, 56, 6579-
6585.
(
59) Rekharsky, M. V.; Goldberg, R. N.; Schwarz, F. P.; Tewari, Y. B.; Ross,
P. D.; Yamashoji, Y.; Inoue, Y. J. Am. Chem. Soc. 1995, 117, 8830-
8
840.
(
60) Marconi, G.; Monti, S.; Mayer, B.; K o¨ hler, G. J. Phys. Chem. 1995, 99,
943-3950.
61) Roberts, E. L.; Dey, J.; Warner, I. M. J. Phys. Chem. A 1997, 101, 5296-
301.
(65) Amato, M. E.; Djedaini-Pilard, F.; Perly, B.; Scarlata, G. J. Chem. Soc.,
Perkin Trans. 2 1992, 2065-2069.
(66) Krishnamoorthy, G.; Dogra, S. K. J. Phys. Chem. A 2000, 104, 2542-
2551.
3
(
5
J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002 261

A R T I C L E S
Zhang et al.
necessarily provide evidence for the prevalence of this particular
co-conformer in solution. The possibility of several populated
co-conformers exposes also the limitations of ICD spectra for
structural assignments, since ICD spectra reflect a Boltzmann-
weighted average of all thermally accessible structures. In
general, one must note that the proposed structural assignments
based on NMR and ICD effects (Scheme 3) are tentative ones,
since they rely, for example, on empirical rules for the
interpretation of ICD effect. However, one must also recall that
the determination of the solution structures of cyclodextrin
complexes by NMR and ICD presents a great challenge since
alternative experimental methods have not been established.
frontal mode of inclusion must be adapted (Scheme 3). This
co-conformation is energetically less favorable, which may offset
the increased driving force due to its higher hydrophobicity and
results in essentially the same binding constant as for 1a.
Generally, the more hydrophilic guests 1d-f show lower
binding constants than 1a-c. The ionic ammonium guest 1f
displays the weakest binding. The difference in binding constants
-1
-1
between 1e (250 M ) and 1f (20 M ) is in line with the general
observation that the ionization of the guest causes a destabiliza-
3
3,68,69
tion of CD complexes in water.
We have shown above
that hydrogen bonding is presumably an important structure-
determining factor. However, the postulated hydrogen bonding
of the hydroxyl and ammonium groups with the hydroxyl rim
of the CD is not sufficiently large to result in an increased
binding constant compared to that of 1a. Moreover, the hydroxyl
and ammonium derivatives 1d and 1f, for which hydrogen
bonding is presumed, show actually a weaker binding than the
amino derivative, for which hydrogen bonding to the CD may
not play a role at all. The absence of a significant thermody-
namic stabilization due to hydrogen bonding is in agreement
The absolute magnitude of the ICD effects can be compared
on the basis of the molar ellipticity values (∆ꢀ, Table 2), which
extrapolate the experimental θ values to quantitative complex-
ation and, thus, present a comparable quantity for all complexes.
The order observed for the positive ICD effects, i.e., 1a > 1d
≈
1f, is in agreement with the suggestion that 1d and 1f adapt
a similar co-conformation, which, however, is somewhat less
ideal than for azoalkane 1a, i.e., the lateral co-conformation
which gives rise to the positive ICD is somewhat tilted. The
order observed for the negative ICD effects, i.e., 1c < 1e ≈
6
,7
with the results of other authors. Consequently, the thermo-
dynamics for binding of azoalkanes 1 is dominated by other
factors. Namely, it can be adequately accounted for in terms of
guest hydrophobicity and a change in the co-conformation due
to the introduction of bridgehead substituents.
1
b, suggests that the dichloro derivative 1c adapts the “most
perfect” frontal co-conformation, which may be caused by a
particularly strong binding and deep penetration of this deriva-
tive into the cavity (see below). For comparison, 1b may not
protrude as deeply due to steric repulsion of the isopropyl group
caused by the lower, tighter rim (note the conical shape of CD).
This leaves the center of mass of 1b more displaced toward the
upper rim and may also be the reason for the large co-
conformational variability, which is indicated for 1b, yet not
Complexation Kinetics. Knowledge of the association and
dissociation rate constants of guest molecules with CDs is
1
9-23
restricted to case studies,
and structural effects have only
70
recently received attention. Substituent effects have apparently
not been examined for a common guest structure. This chal-
lenging aspect can now be studied by using the substituted azo-
alkanes 1 as dynamic fluorescent probes. As noted in the outset,
the fluorescence quenching in the â-CD cavity allows one to
obtain the association rate constants, while other techniques have
proven particularly suitable for determining dissociation rate
3
0
for 1c (see above).
Binding Constants. Relatively few thermodynamic studies
of substituent effects have been devoted to aliphatic guest
molecules3
3,34
as opposed to aromatic guests.
6,7,67
The binding
2
2,23
constants.
Strictly speaking, the association rate constants
constants in Table 2 for azoalkanes 1 reflect the differential
stabilization of the various guest molecules in the bulk water
and the CD cavity. This results from a complex interplay
between van der Waals and hydrophobic interactions, hydrogen
refer to the excited state, but it is presumed that these are
comparable with those in the ground state, since the structure
of these rigid bicyclic molecules and their dipole moments
remain virtually unaffected by the electronic excitation (e.g.,
bonding, release of high-energy water from the CD cavity, and
3
.5 D in the ground state vs 3.2 D in the singlet-excited state
relief of CD strain energy.6
,33,67
A quantitative analysis of the
7
1
of 1a). This is in contrast to other guest molecules such as
trends of the binding constants of azoalkanes 1 (Table 2) is not
attempted, but some peculiarities and clear-cut rationalizations
are addressed.
ketones, which undergo large dipole moment changes upon
2
2,23
excitation.
Moreover, the very low pKa values of the azo
group in the ground state (0.5) and excited state (ca. -8) exclude
The parent compound 1a serves as a reference with a binding
72
the possibility of differential stabilization due to protonation.
-
1
constant of about 1000 M . Based on the lower solubility of
the dichloro derivative 1c in water (1.0 mM) compared to the
other azoalkanes (>10 mM) and well-known solubility trends
In any case, the substituents hardly interact with the azo
chromophore (cf. very similar UV spectra), such that the relatiVe
effects on the association rate constants, i.e., the substituent
effects, should be meaningful even if small changes in the
absolute thermodynamics due to excitation of the azo group
apply.
(e.g., higher water solubility of alcohols compared to homolo-
gous amines), the hydrophobicity order of the examined guest
molecules can be established as 1c > 1b > 1a > 1e > 1d >
1
f. A higher hydrophobicity is expected to result in a larger
As can be seen from Table 2, the overall variation in the
association rate constants covers only a factor of 2-3, much
binding constant, which is in an overall agreement with the
-
1
-1
experimental findings (1900 M for 1c > 1000 M for 1a,1b
>
-
1
-1
-1
(67) Liu, L.; Guo, Q. J. Phys. Chem. B 1999, 103, 3461-3467.
250 M for 1e > 180 M for 1d > 20 M for 1f).
(
68) Matsuura, N.; Takenaka, S.; Tokura, N. J. Chem. Soc., Perkin Trans. 2
1977, 1419-1421.
Interestingly, azoalkane 1b does not show a higher binding
affinity than 1a despite its higher hydrophobicity. This can be
rationalized in terms of the solution structures, which indicate
that the lateral co-conformation, which is the energetically most
(69) Bergeron, R. J.; Channing, M. A.; McGovern, K. A. J. Am. Chem. Soc.
1978, 100, 2878-2883.
(70) Christoff, M.; Okano, L. T.; Bohne, C. J. Photochem. Photobiol. A: Chem.
000, 134, 169-176.
71) Nau, W. M.; Pischel, U. Angew. Chem., Int. Ed. 1999, 38, 2885-2888.
(72) Nau, W. M. EPA Newsl. 2000, 70, 6-29.
2
(
30
favorable one for 1a, cannot be attained for 1b. Instead, a
262 J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002

Host−Guest Complexation of Bicyclic Azoalkanes
A R T I C L E S
less than that for the binding constants. Efficient quenching upon
complexation is a prerequisite for determining association rate
constants by the fluorescence-based method. This is not met
for azoalkane 1b, such that this derivative could not be included
in the kinetic analysis. Until recently, most reported association
rate constants of small organic guest molecules with â-CD fell
(cf. binding constants). This factor may be a common denomi-
nator of the dissociation rate constants.
Conclusions
Thermodynamic, kinetic, and structural aspects of complex-
ation by cyclodextrins (CDs) play an essential role in the
understanding of their functions and for the development of
prospective applications. A fast kinetics, for example, is
desirable for catalytic activity, and a regulated release of the
guest is essential in drug delivery. On the other hand, a strong
and selective binding may be of interest for analytical or
environmental purposes. Finally, the solution structures are
intimately related to molecular recognition, and the specific
location of functional groups in the CD cavity, in particular in
solution, may be the prerequisite for enzymatic specificity. A
pertinent example is that of ester hydrolysis, where the second-
ary hydroxyl rim catalyzes the reaction at the acyl site.7
8
-1 -1
26
in the range of (4-5) × 10 M
s
or below. Recent
measurements suggest that this does not constitute an upper limit
22,70
and that rate constants may exceed this value.
The value
8
-1 -1
measured for azoalkane 1c (7.0 × 10 M s ) is in line with
this observation. Since the value for the dichloro derivative 1c
is the largest, a steric effect, which would predict a slower
association rate, is clearly not observed. Moreover, the presence
of hydrophilic substituents such as hydroxy, amino, and
ammonium has no large effect on the kinetics. Consequently,
guest desolvation does not dominate the complexation kinetics
as in other cases, e.g., in the complexation of cations by crown
4,75
1
3
ethers. Rather, the observed order of the available association
rate constants, namely, 1c > 1a > 1d-f, follows roughly the
order of the binding constants but does not allow definitive
conclusions. However, it should be noted that a correlation
between the kinetics of complexation and the binding constants
is not a priori expected, and contrasting trends were observed
in some cases.7 This circumstance emphasizes the need for
measurement of rate constants in addition to binding constants.
The dissociation rate constants (kdiss) define the lifetimes of
the guests in the CD cavity (τdiss ) 1/kdiss) and may be directly
related to the efficiency of drug delivery and chemical protection
We have provided herein a comprehensive data set for
substituent effects on the binding constants and association rate
constants for complexation of azoalkanes 1 by â-CD and the
solution structures of the resulting complexes (co-conforma-
tions). This joint study of structural, kinetic, and thermodynamic
effects was made possible by the application of several
independent spectroscopic techniques in aqueous solution.
Azoalkanes 1 serve as molecular probes, which can be examined
by UV, time-resolved and steady-state fluorescence, ICD, and
NMR. We have shown that the structures of the CD complexes
of azoalkanes 1, which can be examined by ICD and NMR,
are mainly governed by steric effects and hydrogen bonding.
An unexpected structural variation, signaled by the inversion
of the ICD sign, was observed for the hydroxyl, amino, and
ammonium substituents. The complex stability appears to be
dominated by an interplay between hydrophobic and specific
structural effects, while the association rate constants, which
can be determined by fluorescence decay, are relatively insensi-
tive to substitution. The estimated dissociation rate constants
show a larger variation and appear to be correlated with the
hydrophobicity of the guest.
3
1
7,19
of guests.
These values cannot be directly measured with
our method, but they can be calculated from the experimental
binding constants (K) and excited-state association rate constants
(kass) subject to the above assumption that photochemical
excitation does not significantly modify the affinity for binding.
5
7
-1
The calculated values for kdiss range from 10 to 10 s and
are included in Table 2. For 1a-e, these values lie far below
the rate constants for fluorescence in the CD cavity (kCD ) 1/τCD
7
-1
≈
1 × 10 s ), but for 1f exit occurs with a similar rate
constant. This quantifies the previous notion (expressed for 1a)
that exit from the cavity does not compete significantly with
deactivation for 1a-e. Only for 1f does exit become significant,
such that the kinetics required a more evolved analysis in this
case (cf. Results).
Acknowledgment. This work was supported by the Swiss
National Science Foundation (projects 620-58000 and 4047-
057552) within the Swiss National Research Program “Su-
pramolecular Functional Materials” and MHV grant 2134-
62567.00 for G.G.
As previously emphasized, structural effects manifest them-
selves much more strongly in the exit than in the entry rate
constants.70 This notion is supported for the substituent effects
on the dissociation rate constants for azoalkanes 1, which display
a much larger variation than the association rate constants:
JA011866N
7
-1
6
-1
6
-1
1
>
.2 × 10 s (1f) > 1.8 × 10 s (1d) > 1.0 × 10 s (1e)
(
(
(
74) VanEtten, R. L.; Sebastian, J. F.; Clowes, G. A.; Bender, M. L. J. Am.
5
-1
5
-1
4.4 × 10 s (1a) > 3.7 × 10 s (1c). Interestingly, this
Chem. Soc. 1967, 89, 3242-3253.
75) Tee, O. S.; Bozzi, M.; Hoeven, J. J.; Gadosy, T. A. J. Am. Chem. Soc.
order is the reverse of that established for the hydrophobicity
1
993, 115, 8990-8998.
76) Rodger, A.; Nord e´ n, B. Circular Dichroism and Linear Dichroism; Oxford
(73) Liao, Y.; Bohne, C. J. Phys. Chem. 1996, 100, 734-743.
University Press Inc.: New York, 1997.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 2, 2002 263