The Effects of Buffers on the Thermodynamics and Kinetics of Binding between Positively-Charged Cyclodextrins and Phosphate Ester Guests

Article abstract of DOI:10.1021/jo991616d

Cyclodextrin derivatives in which the primary hydroxyl groups at the 6-positions of the sugar units have been replaced with 2-hydroxethylamino groups are known to bind phosphate esters with significant affinity. Association between the host and guest is mediated by a combination of hydrophobic and electrostatic interactions. Association constants for two cyclodextrin hosts with three phosphate ester guests are measured in a variety of buffer solutions, and it is found that the buffer has a large impact on the electrostatic component of the binding interaction. Negatively charged buffers such as phosphate and ADA (N-(2-acetamido)iminodiacetic acid) compete with the phosphate ester guests for binding to the positively charged cyclodextrins. This competition lowers the effective association constant between host and guest. Positively charged buffers such as Tris-HCl do not strongly compete and are thus more conducive to binding. The kinetics of dissociation of the host-guest complexes were measured in several buffers using ³¹P EXSY experiments. These measurements demonstrate that negatively charged buffers decrease the equilibrium constant by lowering the rate of association of these complexes but in most cases do not effect the dissociation rates.

Full text of DOI:10.1021/jo991616d

J . Org. Chem. 2000, 65, 735-741
735
Th e Effects of Bu ffer s on th e Th er m od yn a m ics a n d Kin etics of
Bin d in g betw een P ositively-Ch a r ged Cyclod extr in s a n d P h osp h a te
Ester Gu ests
Mousumi Ghosh, Rui Zhang, Ronald G. Lawler, and Christopher T. Seto*
Department of Chemistry, Brown University, 324 Brook Street, Box H, Providence, Rhode Island 02912
Received October 15, 1999
Cyclodextrin derivatives in which the primary hydroxyl groups at the 6-positions of the sugar units
have been replaced with 2-hydroxethylamino groups are known to bind phosphate esters with
significant affinity. Association between the host and guest is mediated by a combination of
hydrophobic and electrostatic interactions. Association constants for two cyclodextrin hosts with
three phosphate ester guests are measured in a variety of buffer solutions, and it is found that the
buffer has a large impact on the electrostatic component of the binding interaction. Negatively
charged buffers such as phosphate and ADA (N-(2-acetamido)iminodiacetic acid) compete with the
phosphate ester guests for binding to the positively charged cyclodextrins. This competition lowers
the effective association constant between host and guest. Positively charged buffers such as Tris-
HCl do not strongly compete and are thus more conducive to binding. The kinetics of dissociation
of the host-guest complexes were measured in several buffers using 31P EXSY experiments. These
measurements demonstrate that negatively charged buffers decrease the equilibrium constant by
lowering the rate of association of these complexes but in most cases do not effect the dissociation
rates.
In tr od u ction
demonstrated that the binding interactions between
phosphotyrosine and aminoCDs can be used to inhibit
hydrolysis of the phosphate ester linkage in phospho-
tyrosine catalyzed by phosphatase enzymes.9 Addition-
ally, the investigations of Thatcher and co-workers have
shown that aminoCDs bind to glycosaminoglycan sulfates
and inhibit growth of neurites.10 In vitro biological studies
of this type typically require the use of solutions that
contain buffers, and buffers are likely to mediate the
strength of the electrostatic binding interactions between
aminoCDs and negatively charged guest molecules.
Therefore, the goal of the present study is to clarify the
role that buffer structure and concentration plays in
determining both the thermodynamics and kinetics of
association between aminoCD hosts and phosphate ester
guests. From a broader perspective, these studies provide
a model system to study the effects of buffers on binding
interactions, including small molecule-protein and pro-
tein-protein interactions, that occur in aqueous solution
and that are mediated by electrostatic forces.
In aqueous solution, cyclodextrins bind hydrophobic
molecules in the interior of their cavities. The synthesis
of modified cyclodextrins has expanded their recognition
properties to include noncovalent binding motifs beyond
simple hydrophobic interactions.1 For example, the bind-
ing properties of a large number of modified cyclodextrins
have been investigated that incorporate hydrogen bond-
ing, electrostatic, and metal binding interactions.2 Ami-
nocyclodextrins (AminoCDs) are cyclodextrin derivatives
that incorporate amino-containing functionality in place
of the primary hydroxyl groups at the 6-positions of the
sugar units. Such aminoCDs were first demonstrated to
bind to phosphate esters such as benzyl phosphate
through a combination of hydrophobic and electrostatic
interactions by Knowles and Boger in 1979.3 Since that
time, a variety of aminoCDs have been developed that
recognize phosphate esters,4,5 including several with
biological significance such as nucleotide mono- and
triphosphates6,7 and phosphotyrosine.8 We have recently
Resu lts a n d Discu ssion
(1) For a recent series of comprehensive reviews on all aspects of
cyclodextrin chemistry, see: Chem. Rev. 1998, 98(5).
Cyclod extr in Hosts a n d P h osp h a te Ester Gu ests.
Figure 1 shows the structures of compounds 1â and 2R,
the two aminoCD hosts that are used in this study. The
syntheses of these compounds has been reported by
Darcy7 and Thatcher,11 respectively. Three aryl phos-
phates were used as guests; p-nitrophenyl phosphate (p-
NPP), phosphotyrosine (Tyr(PO32-)), and N-acetyl-phos-
photyrosine methyl ester (AcTyr(PO32-)OMe). The latter
(2) For several recent examples of modified cyclodextrins binding
to amino acids, see: (a) Bonomo, R. P.; Pedotti, S.; Vecchio, G.;
Rizzarelli, E. Inorg. Chem. 1996, 35, 6873. (b) Liu, Y.; Zhang, Y.-M.;
Qi, A.-D.; Chen, R.-T.; Yamamoto, K.; Wada, T.; Inoue, Y. J . Org. Chem.
1997, 62, 1826. (c) Liu, Y.; Zhang, Y.-M.; Qi, A.-D.; Chen, R.-T.;
Yamamoto, K.; Wada, T.; Inoue, Y. J . Org. Chem. 1998, 63, 10085.
(3) (a) Boger, J .; Knowles, J . R. J . Am. Chem. Soc. 1979, 101, 7631.
(b) Boger, J .; Brenner, D. G.; Knowles, J . R. J . Am. Chem. Soc. 1979,
101, 7630.
(4) Vizitiu, D.; Thatcher, G. R. J . J . Org. Chem. 1999, 64, 6235.
(5) Breslow, R.; Schmuck, C. J . Am. Chem. Soc. 1996, 118, 6601.
(6) (a) Eliseev, A. V.; Schneider, H.-J . Angew. Chem., Int. Ed. Engl.
1993, 32, 1331. (b) Eliseev, A. V.; Schneider, H.-J . J . Am. Chem. Soc.
1994, 116, 6081.
(9) Ghosh, M.; Sanders, T. C.; Zhang, R.; Seto, C. T. Org. Lett. 1999,
1, 1945.
(7) (a) Schwinte, P.; Darcy, R.; O′Keeffe, F. J . Chem. Soc., Perkin
Trans. 2 1998, 805. (b) Ahern, C.; Darcy, R.; O′Keeffe, F.; Schwinte,
P. J . Incl. Phenom. Mol Recog. Chem. 1996, 25, 43.
(10) Borrajo, A. M. P.; Gorin, B. I.; Dostaler, S. M.; Riopelle, R. J .;
Thatcher, G. R. J . Bioorg. Med. Chem. Lett. 1997, 7, 1185.
(11) Vizitiu, D.; Walkinshaw, C. S.; Borin, B. I.; Thatcher, G. R. J .
J . Org. Chem. 1997, 62, 8760.
(8) Cotner, E. S.; Smith, P. J . J . Org. Chem. 1998, 63, 1737.
10.1021/jo991616d CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/14/2000

736 J . Org. Chem., Vol. 65, No. 3, 2000
Ghosh et al.
F igu r e 2. Structures of buffers.
finity (Ka ∼ 104-105 M-1). These values were measured
using potentiometric titration in nonbuffered aqueous
NaCl solution.6,7 Knowles has reported that benzyl
phosphate binds to a positively charged cyclodextrin in
triethanolamine buffer with similar affinity (Ka ∼ 3.2 ×
104 M-1).3 In contrast, Breslow and Smith have shown
that binding of aryl phosphates to positively charged
cyclodextrins in phosphate buffer is somewhat weaker (Ka
∼ 102-103 M-1).5,8 The 2- to 3-order of magnitude
difference between these reported binding constants may
simply reflect the differences in structures between the
various cyclodextrin hosts and phosphate ester guests
used in the studies. However, since electrostatic interac-
tions are important in determining the strength of
binding in all of these systems, it is also possible that
the aqueous environment, specifically the buffer, plays
a significant role in determining the association con-
stants. Negatively charged buffers may have reasonable
affinities for the positively charged cyclodextrins and
effectively compete with a phosphate ester guest for
binding with the cyclodextrin. This competition would
have the effect of lowering the observed binding constant
between host and guest.
F igu r e 1. Structures of positively charged cyclodextrin hosts
and aryl phosphate guests.
Sch em e 1a
a
Reagents and conditions: (a) (BnO)2PNEt2, tetrazole; (b)
m-CPBA, 0 °C; (c) 5% Pd/C, H2, MeOH.
compound was included in the study to determine if
blocking the charged ammonium carboxylate portion of
phosphotyrosine causes enhanced affinity to the cyclo-
dextrin hosts. The synthesis of AcTyr(PO32-)OMe (6) is
outlined in Scheme 1. N-Acetyltyrosine methyl ester was
phosphorylated with dibenzyl-N,N-diethylphosphoramid-
ite to give phosphite 4. This compound was oxidized with
m-CPBA to give the corresponding phosphate triester 5.
Finally, removal of the benzyl protecting groups with
hydrogen and Pd/C gave the desired phosphate mono-
ester 6.
Mea su r em en t of Associa tion Con sta n ts by NMR
Sp ectr oscop y. A variety of experimental methods have
been used to measure association constants of cyclodex-
trins with guest molecules. These include UV and CD
spectroscopies, calorimetry, potentiometric titration, and
NMR spectroscopy. Potentiometric titration is one of the
most common methods for investigating the binding
interactions between positively charged cyclodextrins and
phosphate esters.6,7 This technique is useful because it
provides information about both association constants
and pKa’s of ionizable groups. However, potentiometric
titration measurements must be performed in the ab-
sence of buffer, and buffers are typically required when
investigating in vitro biological activity. Thus, association
constants measured using this method may not ac-
curately predict what will occur in biological systems.
To investigate this issue, we have measured the
association constants of cyclodextrins 1â and 2R with
three aryl phosphates in five different buffer systems in
order to clarify the role that buffer structure and con-
centration plays in determining the magnitude of the
binding interactions. Association constants were mea-
sured using NMR spectroscopy because this method is
compatible with a variety of aqueous solutions. The
structures of the buffers used in this study are shown in
Figure 2. These buffers encompass a range of functional
groups, sizes, and charge states that vary from -2 for
phosphate at pH 7 to +1 for imidazole and Tris.
Three NMR methods were used to measure the as-
1
sociation constants: 1H NMR titrations and H and 31P
NMR dilution experiments. We chose a particular experi-
ment for a given complex based upon two considerations;
signal dispersion of the aromatic protons of the guest and
the rate of exchange between the free and bound guest.
In general, complexes with p-NPP as the guest were in
fast exchange on the NMR time scale, while complexes
with Tyr(PO32-) and AcTyr(PO32-)OMe were in the slow
exchange regime. For complexes that displayed good
signal dispersion and that were in fast exchange on the
NMR time scale, the 1H NMR titration method was used.
In these experiments, the chemical shift of the aryl
phosphate protons were monitored as a function of added
cyclodextrin. Data were analyzed using a nonlinear curve
fitting procedure to an equation, reported by Wilcox, that
is based upon a simple 1:1 binding isotherm.12 An
Several reports show that nucleotide phosphates bind
to positively charged cyclodextrins with significant af-

Effects of Buffers on the Thermodynamics and Kinetics of Binding
J . Org. Chem., Vol. 65, No. 3, 2000 737
Ta ble 1. Associa tion Con sta n ts for Cyclod extr in s 1â a n d 2r w ith Ar yl P h osp h a tes in a Va r iety of Bu ffer sa
Kassoc (M-1
)
-2
-2
CD
guest
Tris
imidazole
D2O, NaCl
BES
ADA
10 mM HPO4
100 mM HPO4
1â
p-NPP
Tyr(PO3
60 000b
( 15 000
130 000c
( 25 000
610 000c
( 120 000
30 000b
210 000b
( 60 000
65 000b
( 13 000
34 000b
( 8000
16 000c
(3000
2700b
( 600
1500d
( 300
2200b
(400
410b
( 80
2-
1â
1â
2R
2R
2R
)
710d
( 140
AcTyr(PO32-)OMe
430 000c
( 90 000
70 000b
p-NPP
370d
( 80
( 6000
( 10 000
2-
Tyr(PO3
)
42 000c
( 8000
AcTyr(PO32-)OMe
150 000c
( 30 000
33 000c
( 6000
a
Buffer concentrations are 100 mM unless noted otherwise. The pH of all solutions was 7.0. b Measured by 1H NMR titration experiments.
c Measured 1H NMR dilution experiments. Measured by 31P NMR dilution experiments.
d
Association constants for the two cyclodextrin hosts
with the three aryl phosphate guests in a variety of
buffers are shown in Table 1. From these data, it is
apparent that buffers have a dramatic impact on the
value of the measured association constants. Differences
of 3-4 orders of magnitude are observed on the basis of
the structure and concentration of the buffer. The weak-
est binding constants were measured in 100 mM phos-
phate. Phosphate can associate with the ammonium
groups of the cyclodextrin13 and must be displaced before
the aryl phosphate guest can bind. Part of the free energy
that is gained from binding the aryl phosphate to the
cyclodextrin is offset by the loss of binding interactions
between the buffer and the cyclodextrin. This has the
effect of lowering the observed binding constant. Lower-
F igu r e 3. Titration of p-NPP (1.5 mM) with cyclodextrin 1â
in 100 mM ADA buffer as monitored by 1H NMR spectroscopy.
The curve corresponds to the best fit of the data to a simple
ing the phosphate concentration from 100 to 10 mM
increases the association constant of p-NPP with 1â by
a factor of 5 because there is less competition from the
buffer. Measurements in 100 mM ADA buffer, which has
one negative charge compared to phosphate with two
negative charges, gives higher association constants. This
result indicates that ADA is bound less tightly by the
cyclodextrin than is phosphate.
1:1 binding isotherm.
example of one data set, along with the curve fit, is shown
in Figure 3.
Complexes that were in slow exchange on the NMR
time scale showed well-resolved signals for the free and
bound guest. Integration of these resonances gave a
measure of the relative concentrations of free and bound
aryl phosphate. In these cases, an NMR sample contain-
ing aryl phosphate and a slight excess of cyclodextrin was
prepared. The solution was diluted with buffer in several
A zwitterionic, but overall neutral, buffer such as BES
does not compete as strongly for the cyclodextrin host.
The sulfonate anion of BES is significantly stabilized by
the intramolecular ammonium ion and gains only a small
amount of additional stabilization by complexing with the
cyclodextrin. Therefore, association constants measured
in this buffer are significantly higher than those mea-
sured in ADA and phosphate buffers. In D2O/NaCl
solution that has been pH adjusted to 7.0, the association
constants for p-NPP and AcTyr(PO32-)OMe with 1â are
in the range of 104-105 M-1. These values are similar to
those reported for binding of nucleotide phosphates to
positively charged cyclodextrins as measured by poten-
tiometric titration under similar conditions.6,7 Association
constants measured in buffers that bear a positive charge
such as Tris and imidazole are of similar orders of
magnitude. These solutions allow optimal interactions
between the aryl phosphates and positively charged
cyclodextrins, with the chloride counterion of the buffers
providing little competition to the binding event.
1
steps, and the H NMR spectrum was monitored until it
showed approximately a 1:1 mixture of free and bound
aryl phosphate. At this point, the signals were integrated
and the binding constant calculated on the basis of the
known concentrations of aryl phosphate and cyclodextrin.
1
For several of the complexes in this study, the H NMR
spectra were not useful for measuring association con-
stants because either exchange was intermediate on the
NMR time scale and resonances were broad or the
protons of the aryl phosphate were overlapping. In these
cases, the 31P NMR spectra showed well-resolved reso-
nances for free and bound aryl phosphate. Thus, 31P NMR
dilution experiments were used to measure association
constants. In all cases, there was a significant chemical
shift difference between free and bound aryl phosphate
in the 31P spectrum, reflecting the large difference in
environment experienced by the phosphate group in the
free and bound forms. Relaxation rates (T1) were mea-
sured for several of the free and complexed aryl phos-
phates to ensure that the relaxation delays used in the
1H and 31P NMR spectra were long enough to allow for
accurate integration of resonances.
Figure 4 shows a plot of the free energy of association
for the various cyclodextrin-aryl phosphate combinations
as a function of the buffer. From this pictorial represen-
(13) Schneider has reported that the binding constant between an
(12) Wilcox, C. S.; Cowart, M. D. Tetrahedron Lett. 1986, 27, 5563.
aminoCD and inorganic phosphate is 3.7 × 103 M-1. See ref 6.

738 J . Org. Chem., Vol. 65, No. 3, 2000
Ghosh et al.
F igu r e 4. Free energy for the binding interaction between
the cyclodextrin hosts and the aryl phosphate guests as a
function of buffer structure.
tation of the binding data it becomes apparent that the
two cyclodextrin hosts bind the three aryl phosphate
guests with similar affinity in any particular buffer
system. The only exception is the complex between 1â
and AcTyr(PO32-)OMe, which has an association constant
that is significantly higher than the other complexes. The
difference in affinity between the complexes of 1â with
Tyr(PO32-) and AcTyr(PO32-)OMe suggests that the later
compound makes better hydrophobic contacts with the
cyclodextrin host. In addition, molecular models suggest
that there is a good steric compatibility between AcTyr-
(PO32-)OMe and cyclodextrin 1â, while the cavity of 2R
is too narrow for an optimal fit.
F igu r e 5. 31P EXSY spectrum of the complex between
cyclodextrin 1â and Tyr(PO3-2) in 100 mM ADA buffer at pH
7.0. The mixing time in this experiment was 50 ms.
Ta ble 2. Ra te Con sta n ts, koff, for Dissocia tion of
Com p lexes betw een P ositively-Ch a r ged Cyclod extr in s
a n d Tyr (P O32-) Deter m in ed by 31P EXSY Exp er im en ts (T
) 25 °C)
The binding data presented here demonstrate that
negatively charged buffers weaken the association be-
tween aryl phosphate guests and positively charged
cyclodextrin hosts by competing with the guest for the
available binding sites. If these cyclodextrin derivatives
are used to influence the activity of biological macromol-
ecules in vitro, positively charged buffers should be used
in order to maximize binding interactions between nega-
tively charged guests and the aminoCD hosts.
Mea su r em en t of Dissocia tion Ra tes by Exch a n ge
Sp ectr oscop y. The equilibrium constants discussed in
the previous section provide insights into how buffers
affect the thermodynamics of aryl phosphates binding to
cyclodextrins. We are also interested in how buffers
influence the kinetics of such binding events. In previous
work we have used ROESY experiments to make rough
estimates of the dissociation rate constants of several of
the complexes.9 To determine these rate constants more
precisely, we have measured them using 31P 2D exchange
spectroscopy (EXSY). These experiments were performed
on the complex between Tyr(PO3-2) and cyclodextrin 1â
in a variety of buffers in order to examine the relationship
between dissociation rate and buffer structure.
The EXSY experiment uses the same pulse sequence
as a NOESY experiment, and cross-peaks of opposite sign
are observed for NOEs and exchange. For the complex
between Tyr(PO32-) and 1â, we observe separate reso-
nances for free and bound Tyr(PO32-) in the 31P NMR
spectrum, and no cross-peaks can occur for NOE interac-
tions. If the mixing time used in the experiment is short
compared to the rate of exchange between free and bound
Tyr(PO32-), then no cross-peak is observed between these
species. However, if the mixing time is of the same order
of magnitude or longer than the exchange rate, then a
cross-peak will be observed. The volume of the cross-peak
is proportional to the amount of exchange that has
a
CD
buffer
koff (s-1
)
kon (M-1s-1
)
1â 100 mM phosphate
1â 100 mM ADA
1â 100 mM BES
1â 100 mM Tris
2R 100 mM Tris
45 ( 5
7 ( 1
7 ( 1
7 ( 1
32 000
10 500
110 000
910 000
84
2 × 10-3 ( 0.4 × 10-3 b
a
kon values are inferred based upon the measured koff and Kassoc
values shown this table and Table 1. This value is extrapolated
from measurements that were performed at higher temperatures
(see text).
b
occurred during the mixing time and, thus, can be used
to calculate the exchange rate. One EXSY spectrum is
shown in Figure 5 as an example, and a summary of the
exchange data is given in Table 2.
When ADA, BES, and Tris are used as buffers, the
dissociation rate constants are 7 s-1, despite the fact that
the binding constants in these buffers range from 1500
M-1 for ADA to 130 000 M-1 for Tris. The fact that these
off-rates are idential shows that the buffer does not
participate during the rate-limiting step for dissociation.
Thus, the dissociation rate constant of 7 s-1 simply
represents the rate at which phosphotyrosine dissociates
from the cyclodextrin in aqueous solution. This slow step
is then followed by a second fast step that involves
binding of the buffer to the empty cyclodextrin. In
contrast, the buffer is critical for determining the rate
that Tyr(PO32-) associates with the cyclodextrin. Associa-
tion rate constants (kon) are provided in Table 2 and have
been calculated on the basis of the measured Kassoc and
koff values. These data are consistent with an interpreta-
tion in which the cyclodextrin, in the absence of Tyr-
(PO32-), binds to buffers that incorporate a negatively
charged functional group. In order for Tyr(PO32-) to bind,
the buffer must first dissociate, and it is this dissociation
reaction that controls the overall rate at which Tyr(PO3
2-
)

Effects of Buffers on the Thermodynamics and Kinetics of Binding
J . Org. Chem., Vol. 65, No. 3, 2000 739
state of the dissociation reaction. The calculated values
are ∆Gq ) 21 kcal/mol, ∆Hq ) 33 kcal/mol, and ∆Sq ) 41
eu (T∆Sq ) 12 kcal/mol at 25 °C). The reaction is
entropically favored as is expected for a dissociation
process, and the enthalpy change on going to the transi-
tion-state is surprisingly high for a reaction that involves
only noncovalent interactions. There is likely to be
significant error in the entropy calculation because the
temperature range covered by these experiments is fairly
small.
The dissociation rate constant for the Tyr(PO32-)-2R
complex is 3 orders of magnitude smaller than the
complex with 1â, and the calculated rate constant for
association is also small. The interior diameter of
the cavity in R-cyclodextrin is significantly narrower
than that of â-cyclodextrin, and presents a sterically
constrained hydrophobic pocket through which the phos-
phate group of Tyr(PO32-) must pass in order for associa-
tion or dissociation to occur. In the case of 1â, the cavity
is large enough to allow the phosphate group, perhaps
in the solvated form, to pass through it during these
processes. This lowers the energy of the transition state
for exchange. However, in the case of 2R the tight
confines of the cavity do not easily allow Tyr(PO32-), in
the free or solvated form, to pass through it during
exchange. Therefore, there is a large energy barrier for
both association and dissociation. This interpretation is
supported by molecular models of the two complexes.14
F igu r e 6. Arrenhius plot for dissociation of the complex
between cyclodextrin 2R and Tyr(PO3-2) in 100 mM Tris buffer
at pH 7.0. The y-intercept gives the preexponential factor (A
) 1.4 × 1022 s-1) and the slope gives Ea/R. The activation
energy for this dissociation reaction is 34 kcal/mol. The rate
constants for dissociation at the four temperatures are k )
0.064 s-1 (45 °C), 0.22 s-1 (51 °C), 0.42 s-1 (55 °C), and 0.98
s-1 (62 °C).
binds to the cyclodextrin. After the buffer dissociates from
the cyclodextrin, the Tyr(PO32-) can bind to the empty
cavity in a faster second step. Tris, which is positively
charged and has chloride as the counterion, is not likely
to complex strongly with the cyclodextrin and provides
the fastest association rate between Tyr(PO32-) and the
cyclodextrin. BES, which is neutral overall but possesses
a negatively charged sulfonate group, slows the associa-
tion rate by almost 1 order of magnitude. The association
rate is slowest in ADA buffer, which has an overall
negative charge and is likely to bind well to the cyclo-
dextrin.
In phosphate buffer, the dissociation rate constant is
six to seven times faster than what is observed in ADA,
BES, and Tris buffers. The higher off-rate suggests that
phosphate, unlike the other buffers, is playing an active
role in facilitating dissociation of Tyr(PO32-) from 1â. We
suggest two possible interpretations of this result. First,
phosphate may actively catalyze displacement of the
Tyr(PO32-) from the cyclodextrin by stabilizing the
transition state for dissociation through formation of
electrostatic interactions with the ammonium groups of
the cyclodextrin. A second possibility is that the ground
state of the complex actually involves a ternary aggregate
between Tyr(PO32-), 1â, and inorganic phosphate. Elec-
trostatic interactions between inorganic phosphate and
the cyclodextrin would weaken the interactions with
Tyr(PO32-) and thus decrease the amount of energy
necessary for dissociation.
The dissociation rate constant for the complex between
Tyr(PO32-) and the R-cyclodextrin derivative 2R is too
small to measure at room temperature using the 31P
EXSY experiment. The phosphorus nuclei relax com-
pletely to the ground state before any exchange takes
place, and therefore, no cross-peaks are observed in the
EXSY spectrum no matter how long the mixing time that
is used. To circumvent this problem, we have measured
the dissociation rate constant at several different tem-
peratures above 25 °C in a range in which the rate can
be measured conveniently by EXSY spectroscopy. Extra-
polation of the rate constant back to room temperature
gives the value for this complex shown in Table 2. Figure
6 shows an Arrenhius plot of the exchange data at four
different temperatures, and these data can be used to
calculate thermodynamic parameters for the transition-
Con clu sion s
The structure and concentration of buffers play a major
role in controlling both the thermodynamics and kinetics
of binding interactions of aminoCDs 1â and 2R with aryl
phosphate guest molecules. Negatively charged buffers
compete with the aryl phosphate guest for the positively
charged binding site of the cyclodextrin, and this com-
petition lowers the effective binding constant. Based upon
these results, we conclude that positively charged buffers
such as Tris and imidazole should be used when studying
the influence of these cyclodextrins on biological systems.
The 31P EXSY experiments show that, with the exception
of phosphate, buffers exert a major influence on the
association rates of aryl phosphates with cyclodextrins,
but have no effect on the dissociation rates.
One of the disadvantages of the aminoCDs in their
present form is that they show low selectivity and will
thus bind to many sterically compatible phosphate esters.
To make the aminoCDs more useful for biological studies,
we are modifying their structure to make them more
specific for particular phosphotyrosine-containing se-
quences and structures within peptides and proteins.
Such modifications include addition of a second CD unit
that is designed to bind to the side chain of a nearby
hydrophobic residue, and functionalization of the amino-
CDs with Lewis acidic groups such as activated carbonyl
compounds or boronic acids that are designed to interact
with the side chains of proximal nucleophilic amino acids.
Exp er im en ta l Section
Gen er a l Meth od s. Reactions were conducted under an
atmosphere of dry nitrogen in oven-dried glassware. Anhy-
(14) Thatcher and co-workers have recently reported that the
coalescence temperature for free 4-isopropylphenyl phosphate and
4-isopropylphenyl phosphate that is bound to an aminoCD is ap-
proximately 100 °C. These authors also attribute the large barrier to
exchange to the passage of the phosphate group through the hydro-
phobic cavity of the CD. See ref 4.

740 J . Org. Chem., Vol. 65, No. 3, 2000
Ghosh et al.
drous procedures were conducted using standard syringe and
cannula transfer techniques. THF was distilled from sodium
and benzophenone. Other solvents were of reagent grade and
were stored over 4 Å molecular sieves. All other reagents were
used as received. Organic solutions were dried over MgSO4
unless otherwise noted. Solvent removal was performed by
rotary evaporation at water aspirator pressure.
0.018 mM, aromatic protons of the aryl phosphate. 1â-Tyr-
(PO3-2): 100 mM BES, 0.14 mM, 0.087 mM, aromatic protons
of the aryl phosphate. 1â-AcTyr(PO3-2)OMe: D2O/NaCl, 0.026
mM, 0.022 mM, N-Ac protons of the aryl phosphate. 2R-Tyr-
(PO3-2): 100 mM Tris, 0.093 mM, 0.067 mM, aromatic protons
of the aryl phosphate. 2R-AcTyr(PO3-2)OMe: 100 mM Tris,
0.063 mM, 0.050 mM, N-Ac protons of the aryl phosphate. 2R-
AcTyr(PO3-2)OMe: D2O/NaCl, 0.055 mM, 0.050 mM, N-Ac
protons of the aryl phosphate.
N-Acetyl-L-tyr osin e Diben zyl P h osp h a te Meth yl Ester
5. 1H-Tetrazole (1.68 g, 23.5 mmol), N-acetyl-L-tyrosine methyl
ester (2.0 g, 7.8 mmol), and dibenzyl N,N-diethylphosphor-
amidite (4.7 mL, 15.6 mmol) were combined in 8 mL of dry
THF, and the solution was stirred at 25 °C for 1 h. The mixture
was then cooled to -40 °C in a dry ice-acetone bath, and a
solution of 77% m-CPBA (2.6 g, 11.6 mmol) dissolved in 20
mL of methylene chloride was added via syringe. After the
addition was complete, the reaction was warmed to 25 °C and
stirred for 30 min, then 10 mL of a 10% aqueous solution of
NaHSO3 was added and the reaction was stirred for an
additional 10 min. The mixture was extracted with ether, and
the organic layer was washed twice with 10% aqueous
NaHSO3, twice with 5% aqueous NaHCO3, and once with brine
and dried over MgSO4. The solvent was removed by rotary
evaporation, and the product was purified by flash chroma-
tography (1:9 hexanes/ethyl acetate) to give 1.9 g (3.9 mmol,
50%) of N-acetyl-L-tyrosine dibenzyl phosphate methyl ester:
1H NMR (300 MHz, CDCl3) δ 1.94 (s, 3H), 3.02-3.08 (m, 2H),
3.67 (s, 3H), 4.82 (dd, J ) 14.6, 6.3 Hz, 1H), 5.10 (d, J ) 8.3
Hz, 4H), 6.43 (d, J ) 7.7 Hz, 1H), 7.03 (d, J ) 8.6 Hz, 2H),
7.07 (d, J ) 8.6 Hz, 2H), 7.31 (s, 10H); 13C NMR (75 MHz,
CDCl3) δ 22.7, 36.8, 52.1, 69.5 (d, J ) 5.6 Hz), 119.8 (d, J )
4.7 Hz), 127.8, 128.3, 128.4, 130.3, 132.8, 135.1 (d, J ) 7.1
Hz), 149.4 (d, J ) 7.1 Hz), 169.7, 171.8; HRMS-FAB (M + H+)
calcd for C26H29NO7P 498.1680, found 498.1692.
31P NMR Dilu tion Exp er im en ts. For the 31P NMR dilu-
tion experiments, the conditions that were used during each
experiment are provided in the following format: Complex:
buffer, final cyclodextrin concentration, final aryl phosphate
concentration. In all of these experiments, the resonance of
the phosphate group of the aryl phosphate was monitored. 1â-
Tyr(PO3-2): 100 mM ADA, 1.40 mM, 1.12 mM. 1â-Tyr(PO3-2):
100 mM phosphate, 2.9 mM, 2.6 mM. 2R-p-NPP: 100 mM
phosphate, 8.0 mM, 4.5 mM.
31P 2D Exch an ge Spectr oscopy Exper im en ts. The phase-
sensitive 2D 31P-31P EXSY spectra were recorded at 27 °C at
161.9 MHz on a Bruker AMX 400 spectrometer equipped with
an SGI computer, using a 5 mm QNP probe. The 2D EXSY
maps were obtained from the basic NOESY pulse sequence
that involves three 90° pulses with time-proportional phase
increments and 16-order phase cycling. Sixteen scans were
accumulated for each of the 128 t1 increments, zero filled to
512 W in the F1 dimension and using 512 W in the F2
dimension with no zero filling, a 2 s relaxation delay and a
spectral window of 3221 Hz. Preliminary experiments with
longer relaxation delays did not reveal any significant change
in the relative integrals of the resonances, indicating es-
sentially full relaxation within the recycling delay. The free
induction delay was multiplied by a sine bell function with φ
) π/2. The 2D spectra were phase and baseline corrected in
both dimensions and diagonal and cross-peak volumes were
determined using the standard Bruker Avance software. The
rate constants were evaluated using the method of Harzell and
co-workers.15
For cases in which the T1 values of the free and bound aryl
phosphate were much less than the exchange rate constant
(kex ≡ koff), the effects of longitudinal cross relaxation rates
were ignored. For cases in which the T1 values were of the
same order of magnitude as the exchange rate constant, the
kex value was corrected for the effects of longitudinal cross
relaxation using the method described by Macura and Ernst.16
The equation relating the values of kex to the ratios of the areas
of bound diagonal to cross-peaks, Iii/Iij, for the case of equal
concentration of free and bound species, is shown below and
is taken from ref 16.
N-Acetyl-L-tyr osin ep h osp h a te Meth yl Ester 6. To a
solution of 5 (0.77 g, 1.4 mmol) in 2 mL of MeOH was added
a catalytic amount of 5% Pd/C, and hydrogen gas was bubbled
into the reaction mixture. The reaction was stirred for 3 h at
25 °C under an atmosphere of hydrogen. The catalyst was
removed by filtration, and the solvent was removed by rotary
evaporation to give 421 mg (1.27 mmol, 91%) of compound 6
as a clear oil: 1H NMR (300 MHz, MeOH-d4) δ 1.93 (s, 3H),
2.95 (dd, J ) 13.6, 9.3 Hz, 2H), 3.14 (dd, J ) 13.6, 5.7 Hz,
1H), 3.69 (s, 3H), 4.65 (dd, J ) 9.3, 5.7 Hz, 1H), 7.15 (d, J )
8.6 Hz, 2H), 7.21 (d, J ) 8.6 Hz, 2H); 13C NMR (75 MHz,
MeOH-d4) δ 21.2, 36.6, 51.6, 54.3, 120.3 (d, J ) 4.7 Hz), 130.3,
133.6, 150.7 (d, J ) 6.2 Hz), 172.2, 172.4; HRMS-FAB (M +
Na+) calcd for C12H16NNaO7P 340.0561, found 340.0562.
1
1H NMR Titr a tion Exp er im en ts. For the H NMR titra-
tion experiments, the conditions that were used during each
titration are provided in the following format: Complex:
buffer, cyclodextrin concentration or range of concentrations
used, aryl phosphate concentration or range of concentrations
used, identity of the proton or protons that were monitored
during the titration experiment. 1â-p-NPP: 100 mM Tris,
0-0.25 mM, 0.05 mM, aromatic protons of the aryl phosphate.
1â-p-NPP: 100 mM imidazole, 0.05 mM, 0-0.20 mM, H4 of
the cyclodextrin. 1â-p-NPP: D2O/NaCl, 0-0.38 mM, 0.1 mM,
aromatic protons of the aryl phosphate. 1â-p-NPP: 100 mM
BES, 0-1.0 mM, 0.21 mM, aromatic protons of the aryl
phosphate. 1â-p-NPP: 100 mM ADA, 0-6.1 mM, 1.5 mM,
aromatic protons of the aryl phosphate. 1â-p-NPP: 10 mM
phosphate, 0-4.4 mM, 0.74 mM, aromatic protons of the aryl
phosphate. 1â-p-NPP: 100 mM phosphate, 0-10 mM, 1.0
mM, aromatic protons of the aryl phosphate. 2R-p-NPP: 100
mM Tris, 0.05 mM, 0-0.25 mM, H4 of the cyclodextrin. 2R-
p-NPP: D2O/NaCl, 0.1 mM, 0-0.63 mM, H4 of the cyclodex-
trin.
Iii/Iij ) 1/2(Rc/kex)[(1 + e-Rcτm)/(1 - e-Rcτm)] - 1/2(∆R/kex)
In this equation, ∆R ) R(bound) - R(free) ) 1/T1(bound) - 1/T1(free)
,
and Rc ) [∆R2 + 4kex 1/2. The value of kex was obtained using
]
2
a computer program to solve in an iterative fashion for Iii/Iij,
starting from a given value of kex and the observed values of
T1 for each temperature. Iterations were continued until the
calculated Iii/Iij ratio agreed to better than 1% with the
experimentally observed value. Data for the complex between
cyclodextrin 2R and Tyr(PO32-) in 100 mM Tris buffer at four
different temperatures are as follows, where kex,uncorr corres-
ponds to the experimentally observed values and kex,corr are
the values that have been corrected for the effects of longitu-
dinal cross relaxation. 45 °C: kex,uncorr ) 0.078 s-1, kex,corr
0.064 s-1; 51 °C: kex,uncorr ) 0.239 s-1, kex,corr ) 0.217 s-1; 55
°C: kex,uncorr ) 0.457 s-1, kex,corr ) 0.415 s-1; 62 °C: kex,uncorr
1.09 s-1, kex,corr ) 0.980 s-1
)
)
1H NMR Dilu tion Exp er im en ts. For the 1H NMR dilution
experiments, the conditions that were used during each
experiment are provided in the following format: Complex:
buffer, final cyclodextrin concentration, final aryl phosphate
concentration, identity of the proton or protons that were
monitored during the dilution experiment. 1â-AcTyr(PO3-2)-
OMe: 100 mM Tris, 0.027 mM, 0.025 mM, N-Ac protons of
the aryl phosphate. 1â-Tyr(PO3-2): 100 mM Tris, 0.019 mM,
.
T1 Mea su r em en ts. Phosphorus T1 relaxation times were
measured using a 31P inversion recovery sequence with
increasing delays between the 180° inversion pulse and the
(15) Hartzell, C. J .; Mente, S. R.; Eastman, N. I.; Beckett, J . L. J .
Phys. Chem. 1993, 97, 4887.
(16) Macura, S.; Ernst, R. R. Mol. Phys. 1980, 41, 95, eq 23.

Effects of Buffers on the Thermodynamics and Kinetics of Binding
J . Org. Chem., Vol. 65, No. 3, 2000 741
90° observe pulse. The intensities of the peaks were plotted
against the delay times, and the T1 values were calculated
Ack n ow led gm en t. This research was supported by
the U.S. Army Medical Research and Material Com-
mand-Breast Cancer Research Initiative through a
Career Development Award to C.T.S. (Grant No.
DAMD17-96-1-6161).
using standard Bruker software. To measure the T1 value of
the phosphate group in a cyclodextrin-aryl phosphate com-
plex, a solution of the complex was prepared in which greater
than 99% of the aryl phosphate was present in the complexed
form. 1â-Tyr(PO32-): T1 ) 1.01 s (100 mM phosphate, 25 °C);
Tyr(PO32-): T1 ) 6.29 s (100 mM phosphate, 25 °C); 2R-Tyr-
(PO32-): T1 ) 1.89 s (100 mM Tris, 62 °C); 2R-Tyr(PO32-):
T1 ) 1.33 s (100 mM Tris, 45 °C); Tyr(PO32-): T1 ) 6.88 s
(100 mM Tris, 62 °C); Tyr(PO32-): T1 ) 5.23 s (100 mM Tris,
45 °C).
Su p p or tin g In for m a tion Ava ila ble: Copies of 1H and 13C
NMR spectra for compounds 5 and 6. This material is available
free of charge via the Internet at http://pubs.acs.org.
J O991616D