A Selective Receptor for Arginine Derivatives in Aqueous Media. Energetic Consequences of Salt Bridges That Are Highly Exposed to Water

Article abstract of DOI:10.1021/ja982499r

Quantitative measures of salt-bridge-type interactions in a highly exposed aqueous environment have been obtained by modifying the well-studied cyclophane platform 1 to include carboxylates in close proximity to bound, cationic guests, producing hosts 2 and 3. Many guests show significantly enhanced binding to 2 and 3, but cations of the RNMe₃⁺ type show little or no enhancement. We propose that the latter observations result from the fact that RNMe₃⁺ compounds have very diffuse positive charges. Guests that show enhanced binding have focused regions of large, positive electrostatic potential. The highly charged 3 is able to bind very polar, very well-solvated guests, including a series of arginine-based dipeptides. Neutral, water-soluble host 4 was prepared and found to show a decreased affinity for cationic guests. We propose a novel induced dipole mechanism to rationalize these results.

Full text of DOI:10.1021/ja982499r

1192
J. Am. Chem. Soc. 1999, 121, 1192-1201
A Selective Receptor for Arginine Derivatives in Aqueous Media.
Energetic Consequences of Salt Bridges That Are Highly Exposed to
Water
†
†
†
,‡
Sarah M. Ngola, Patrick C. Kearney, Sandro Mecozzi, Keith Russell,* and
Dennis A. Dougherty*,†
Contribution from the DiVision of Chemistry and Chemical Engineering, California Institute of
Technology, Pasadena, California 91125, and Medicinal Chemistry Department, Zeneca Phamaceuticals,
Wilmington, Delaware 19897
ReceiVed July 15, 1998. ReVised Manuscript ReceiVed NoVember 21, 1998
Abstract: Quantitative measures of salt-bridge-type interactions in a highly exposed aqueous environment
have been obtained by modifying the well-studied cyclophane platform 1 to include carboxylates in close
proximity to bound, cationic guests, producing hosts 2 and 3. Many guests show significantly enhanced binding
+
to 2 and 3, but cations of the RNMe3 type show little or no enhancement. We propose that the latter observations
+
result from the fact that RNMe3 compounds have very diffuse positive charges. Guests that show enhanced
binding have focused regions of large, positive electrostatic potential. The highly charged 3 is able to bind
very polar, very well-solvated guests, including a series of arginine-based dipeptides. Neutral, water-soluble
host 4 was prepared and found to show a decreased affinity for cationic guests. We propose a novel induced
dipole mechanism to rationalize these results.
Extensive previous studies of cyclophane host 1 establish that
it presents a well-defined, hydrophobic binding site that can
1
bind a wide range of guests with often very high affinities. In
addition to hydrocarbon-type guests, 1 binds cationic quaternary
ammonium, iminium, guanidinium, and sulfonium compounds
with high affinity and in relatively well-defined binding
orientations. Central to these binding events are cation-π
interactions,2 in which positive charges are stabilized by
interactions with electron-rich faces of aromatic rings. Such
studies have suggested new roles for aromatic residues in a
variety of biological receptors, especially those that bind the
2,3
quaternary ammonium neurotransmitter, acetylcholine (ACh).
Tetraalkylammonium ions are very good guests for 1, but
+
protonated amines, RNH3 , are not well bound. A similar result
is seen with guanidinium ionssextensive alkylation is required
for binding while simple systems such as arginine are not bound.
Theoretical studies, gas-phase measurements, and evidence from
a number of biological structures all indicate that cation-π
+
interactions with RNH3 compounds and simple guanidiniums
are quite strong. We thus concluded that it was the greater
hydration of protonated versus alkylated cations that made them
poorer guests for cyclophane 1. It is always true that binding is
determined by a balance of two issues: the favorable attractions
between host and guest and the solvation properties of the two.
Tetraalkylammoniums are well solvated by water, but protonated
3
amines are better solvated by as much as 30 kcal/mol.
of RNH3+ guests. We naturally wondered whether we could
improve 1 in some way to overcome the additional desolvation
penalty of protonated amines. The obvious route would be to
combine cation-π interactions with some additional binding
interaction such as hydrogen bonding and/or ion pairing.
To that end, we report here the preparation and characteriza-
tion of two new cyclophanes, 2 and 3. In host 2 a “fifth”
carboxylate is appended in close proximity to the binding site
in hopes of providing stabilizing salt bridge interactions with
Apparently, host 1 simply cannot overcome the extra solvation
†
Caltech.
Zeneca.
‡
(
1) Petti, M. A.; Shepodd, T. J.; Barrans, J., R. E.; Dougherty, D. A. J.
Am. Chem. Soc. 1988, 110, 6825-6840. Kearney, P. C.; Mizoue, L. S.;
Kumpf, R. A.; Forman, J. E.; McCurdy, A.; Dougherty, D. A. J. Am. Chem.
Soc. 1993, 115, 9907-9919.
(
2) Dougherty, D. A. Science 1996, 271, 163-168. Ma, J. C.; Dougherty,
D. A. Chem. ReV. 1997, 97, 1303-1324.
(
3) Dougherty, D. A.; Stauffer, D. A. Science 1990, 250, 1558-1560.
1
0.1021/ja982499r CCC: $18.00 © 1999 American Chemical Society
Published on Web 01/22/1999

Receptor for Arg DeriVatiVes in Aqueous Media
J. Am. Chem. Soc., Vol. 121, No. 6, 1999 1193
Figure 1. Synthesis of hosts 2 and 3.
1
specific cationic guests. The highly charged octacarboxylate 3
could develop multiple salt bridges, as well as providing a very
large overall negative charge that could attract cations generally.
In our cyclophane work, a persistent question has been the
role, if any, of the remote carboxylates of 1 in binding cations.
Although they cannot come into close contact with guests, and
they should be well solvated, there remains the possibility that
they could aid cation binding. This issue became more pressing
because of some of the results seen with 2 and 3 (see below),
methodology we developed earlier. The syntheses of 2 and 3
are summarized in Figure 1. Not surprisingly, both hosts are
freely soluble in our aqueous buffer, and binding studies
proceeded quite smoothly.
Neutral Host 4. The key to the synthesis of 4 is the efficient
reaction of an ester with tris(hydroxymethyl)aminomethane (tris)
to make the amide, as developed by Newkome in the context
of preparing novel, water-soluble dendrimers. This reaction,
however, is not compatible with the ester related to 1, perhaps
because of reaction with the olefins or for steric reasons.
8
4
-7
and because a number of biological binding sites for cations
seem to involve close contacts with aromatics but also possible
interactions with more remote carboxylates. To probe potential
long-range ionic interactions in aqueous media, we prepared
the neutral but still water-soluble cyclophane 4. While the amide
groups of 4 are highly polar and well solvated like the remote
carboxylates of 1, they are not charged, and so comparisons
between 1 and 4 should be informative on any possible long-
range Coulombic interactions between cationic guests and the
solvating groups.
Our results establish that even in aqueous media, exposed
salt bridges can contribute significantly to binding, but the
magnitude of the effect depends strongly on the nature of the
charge in the guest. Host 3 is capable of binding highly solvated
arginine dipeptides, making use of generic electrostatic attrac-
tions as well as specific salt bridges. Studies with 4 establish
that the remote carboxylates of 1 can contribute to cation
binding, perhaps via an effect that is transmitted by the aromatic
rings of the cyclophane.
Removing the double bonds to produce ester 5 produces a good
substrate for the tris reaction. Tetraester 5 is readily available
in enantiomerically and diastereomerically pure form using our
1
previously described asymmetric Diels-Alder route. Control
9
studies with the analogue of 1 obtained by direct hydrolysis
of 5 reveal that saturating the etheno bridge does not measurably
alter the binding properties of such hosts.
Synthesis and Characterization of Hosts
The neutral, cyclophane, dodecaalcohol, tetraamide 4 is water-
soluble at sub-millimolar concentrations. However, at concentra-
tions necessary for binding studies, some aggregation does
occur. As a result, most binding studies with 4 were done in a
mixture of 90% (10 mM aqueous NaCl)/10% acetonitrile, in
which no aggregation occurs. We have a large reference base
Pentacarboxylate 2 and Octacarboxylate 3. The syntheses
of 2 and 3 required only minor modifications of the standard
(
4) Sussman, J. L.; Harel, M.; Frolow, F.; Oefner, C.; Goldman, A.;
Toker, L.; Silman, I. Science 1991, 253, 872-879. Sussman, J. L.; Silman,
I. Curr. Opin. Struct. Biol. 1992, 2, 721-729.
(5) Satow, Y.; Cohen, G. H.; Padlan, E. A.; Davies, D. R. J. Mol. Biol.
1
986, 190, 593-604.
(8) Newkome, G. R.; Baker, G. R.; Arai, S.; Saunders, M. J.; Russo, P.
S. J. Am. Chem. Soc. 1990, 112, 8458. Newkome, G. R.; Hu, Y. F.;
Saunders, M. J.; Fronczek, F. R. Tetrahedron Lett. 1991, 32, 1133.
(9) Extensive further details on many aspects of this work can be found
in the following: Mecozzi, S. Ph.D. Thesis, California Institute of
Technology, 1997.
(
6) Hibert, M. F.; Trumpp-Kallmeyer; Bruinvels, A.; Hoflack, J. Mol.
Pharmacol. 1991, 40, 8-15. Trumpp-Kallmeyer, S.; Hoflack, J.; Bruinvels,
A.; Hibert, M. J. Med. Chem. 1992, 35, 3448-3462.
(7) Zhong, W.; Gallivan, J. P.; Zhang, Y.; Li, L.; Lester, H. A.;
Dougherty, D. A. Proc. Natl. Acad. Sci. 1998, 95, 12088-12093.

1194 J. Am. Chem. Soc., Vol. 121, No. 6, 1999
Ngola et al.
1
of binding constants for 1 in 10% acetonitrile medium. The
mixed solvent shows a remarkably consistent drop of 2.5 kcal/
mol in binding for cationic guests of the type studied here to
host 1, presumably due to a diminution of the hydrophobic effect
and possible changes in electrostatic/ionic interactions. Impor-
tantly, in no instances are any of the significant trends in binding
data altered by changing from a strictly aqueous to a 10%
acetonitrile medium. Note that in studies of 1, the aqueous
medium is a 10 mM borate buffer, to ensure deprotonation of
all of the carboxylates. We therefore added 10 mM NaCl to the
aqueous medium used in studies of 4, in order to minimize ionic
strength effects.
Binding Studies
The primary tool for evaluating binding constants in these
systems has been NMR spectroscopy. This is now a quite
standard tool in studies of molecular recognition. We simply
note here that all data sets have been through the rigorous
evaluation of statistical significance and precision described in
10
detail elsewhere. As in other studies, we have adopted the
fairly conservative view that -∆G° values are reliable to (0.2
kcal/mol.
Several binding constants have also been determined by
circular dichroism (CD) methods. As discussed in detail
1
1
elsewhere, these cyclophane hosts are especially well suited
to the CD method, and it provides an excellent independent
check on the validity of the NMR method. In addition, CD
allows measurement of some binding constants that are not
accessible by NMR. This is especially true for relatively
insoluble hosts such as 4.
Pentacarboxylate Host 2. Figure 2 summarizes binding data
for host 2, with comparisons to analogous data for 1. The equal
affinities of the neutral guests (6-8) for 1 and 2 indicate that
the additional carboxylate moiety of 2 has not had a major
impact on the hydrophobic binding cavity. One could have
imagined that the carboxylate would decrease the hydrophobicity
of the cyclophane cavity or stabilize an unfavorable conforma-
tion of the host, either of which would have led to a general
decrease in binding for all guests. Since this did not occur, any
observed variations in the binding of cationic guests by 2 can
be considered a direct result of the differing interactions of these
guests with the additional carboxylate of that structure. For the
neutral guests and for all other guests studied, the highly
Figure 2. Binding energies of guest molecules complexed with
cyclophanes 1-3. Value in brackets are in 10% CH CN; all others are
3
in pure borate buffer. a. Determined by CD; all others by NMR. b.
This value is a lower limit.
was simply unable to get into close contact with the additional
carboxylate of 2. However, modeling and chemical shift data
(
which are especially informative for guest 14)13 strongly
1
indicated that guest 14 is ideally positioned to interact with the
carboxylate, yet no effect is seen. Apparently, the strong
solvating ability of water completely damps out any ion-pairing
interactions in the 2/14 pair. If so, then switching to a less
effective solvating medium might allow the interaction to be
seen. Indeed, we find that simply adding 10% acetonitrile to
the aqueous medium leads to a noticeable change (Figure 2).
Now the 2/14 combination displays a significant ion pair
component in its binding. These results clearly demonstrate that
guest 14 can position itself to interact favorably with the
additional carboxylate of 2. In pure water, however, this
interaction is apparently not energetically significant. Note that
the acetonitrile effect is not constant for all of the guests studied
with host 2, unlike the observations for host 1. Although we
have only a few observations for host 2, this is a further example
of the context-sensitivity of ion pair interactions versus the
relatively generic nature of the hydrophobic interaction.
Host 2 is unique compared to 1 and 3 in terms of confor-
mational complexitysthe single carboxylate of 2 completely
desymmetrizes the system. Even considering only the rhomboid
distinctive changes in H NMR chemical shifts that are induced
by binding are identical for hosts 1 and 2.12 Thus, the additional
carboxylate of 2 does not alter the hydrophobic nature of the
binding site, and it does not cause any guests to orient differently
in 2 vs 1.
Studies of cationic guests reveal considerable variability in
the apparent interaction of the guest charge with the additional
carboxylate of 2. Iminium compounds such as 9 and 10 display
an enhanced affinity for 2. Within the class of guanidinium
compounds examined, 11 and 12 exhibit significant increases
in binding affinity, while 13 shows no enhancement in binding.
In contrast, tetraalkylammonium compounds show smaller
enhancements, with the extensively studied 14 showing no
improvement at all.
The failure to see any enhancement in binding of a guest
such as 14 could indicate that the cationic part of the molecule
(
10) Barrans, R. E., Jr.; Dougherty, D. A. Supramolec. Chem. 1994, 4,
21-130.
11) Forman, J. E.; Barrans, R. E., Jr.; Dougherty, D. A. J. Am. Chem.
Soc. 1995, 117, 9213-9228.
12) Extensive further details on many aspects of this work can be found
1
(
(
in the following: Kearney, P. C. Ph.D. Thesis, California Institute of
Technology, 1994.
(13) Shepodd, T. J.; Petti, M. A.; Dougherty, D. A. J. Am. Chem. Soc.
1986, 108, 6085-6087.

Receptor for Arg DeriVatiVes in Aqueous Media
J. Am. Chem. Soc., Vol. 121, No. 6, 1999 1195
2 2
Figure 4. Structures of Arg-NH (19) and Lys-NH (20). Also shown
are the relatiVe upfield shifts induced by binding to 3. The actual
maximum shifts are 3.4 ppm for 19 and 3.0 ppm for 20.
the toroid form is intrinsically less stable in 3 than in 1, or
perhaps in the toroid the carboxylates do not interact well with
the cationic guest. Detailed analysis of NMR shift patterns and
Figure 3. Limiting conformations of host 2. The rhomboid form (left)
1
4
is based on the X-ray structure of the tetraester of 1, with the fifth
carboxylate added in an idealized position. The toroid form (right) is
an idealized structure. Color code: white ) H; gray ) C; black ) O.
15
CD spectra provide support for the notion that conformational
factors contribute to the binding differences we see. An apparent
exception is sulfonium guest 18. Although it binds to the
rhomboid form of the host, it does not benefit from the extra
carboxylates. This could be because a great deal of the positive
charge in such a system is found on the sulfur atom,16 which is
presumably buried in the cyclophane cavity and so not in close
contact with the carboxylates.
Binding of Arginine Derivatives by Host 3. We were quite
encouraged by the enhanced binding of 2 and especially 3 for
cations that are not fully quaternized, such as 11 and 17. Of
course, the challenge with such guests is that relative to the
fully alkylated structures such as 9 and 14, ions such as 11 and
structure, the fifth carboxylate can occupy four distinct positions,
and these conformers interconvert via facile bond rotations.
Detailed analysis of NMR spectra12 provides some evidence that
the guests that benefit from the extra carboxylate select particular
conformations of the host. There would be an entropic cost to
such behavior, which could impact all binding to host 2.
Octacarboxylate Host 3. As the results in Figure 2 indicate,
the more dramatic modifications on going from 1 to 3 have a
broader influence on binding. Neutral guests (7 and 8) have a
decreased affinity for 3 (relative to 1 and 2). Again, one can
imagine two possible reasons for this: a decrease in the
hydrophobic nature of the cyclophane cavity or the development
of an unfavorable conformation of the host.
1
7 are very much better solvated by water. These initial results
suggested that 3 may be able to bind well solvated cations. The
cationic amino acids Arg and Lys seemed like good vehicles
to evaluate this possibility.
To address the conformational issue, the host structures were
examined in the presence of various guests. Cyclophanes such
as 1-3 can adopt two general binding conformations, (Figure
Arg-NH2 (19) was chosen as the initial ligand (Figure 4).
Guest 19 has a 5.0 kcal/mol affinity for cyclophane 3, which
we felt was quite promising. Host 1 does not bind simple
arginine derivatives at all. For comparison, we evaluated Lys-
NH2 (20), another amino acid with a cationic side chain.
Although the hydration energies of 19 and 20 are essentially
3) depending on the type of guest bound. For flat aromatic guests
such as 7 or 9, the host adopts a C2-symmetric rhomboid
conformation. This maximizes both the hydrophobic and
π-stacking interactions between the host and the guest. For large
“
spherical” guests such as 14, the host adopts a D2-symmetric
17
the same, there are interesting differences between the two.
toroid conformation. In this form, the host has a much more
open cavity and can encapsulate the larger guest. The existence
One difference is that the cationic side chain of 19 has three
sites for hydrogen-bonding/salt-bridge interactions, while 20
only has one. Also, while both the guanidinium of 19 and the
1
of these two conformers of 1 has been supported by H NMR
1
,11
chemical shift data as well as circular dichroism (CD) data.
ammonium of 20 can experience cation-π interactions, there is
An X-ray structure of the host 1 tetramethyl ester revealed a
rhomboid conformation,14 in agreement with the favored
structure predicted by modeling.
Extensive analysis of the binding conformations of host 3
by both NMR and CD reveal no major changes in the
conformational analysis relative to 1 or 2.15 We thus conclude
that the general diminution of binding for neutral guests
exhibited by 3 results from a disruption of the hydrophobic
binding site caused by the large number of polar groups.
Concerning cationic guests, relatively flat structures such as
evidence that the former interaction is more favorable.2
Consistent with these expectations, 20 has a lesser affinity
for cyclophane 3sthe -∆G° for binding is 4.0 kcal/mol versus
5
2
1
.0 kcal/mol for 19. The NMR shift patterns for bound 19 and
0 (Figure 4) suggest different binding modes for the two. In
9, the most upfield-shifted CH2 (and thus the one most buried
in the cavity) is directly adjacent to the guanidinium, as expected
if a cation-π interaction with this group is important. With 20,
+
no such clear-cut pattern is evident; CH2’s adjacent to the NH3
and adjacent to the R carbon are strongly upfield shifted. This
suggests a coiled conformation that is not controlled by a
cation-π interaction.
Encouraged by the binding of 19 to 3, we decided to try to
design a better guest. Since the guanidinium of 19 is presumably
buried to some extent in the cavity of 3, it may not be optimally
positioned to make a tight salt bridge with a carboxylate of the
host. So, we studied a series of dipeptides in which the side
chain of the second residue could make such a contact. These
results are summarized in Figure 5.
9, 11, 15, and 17 show substantial increases in affinity on going
from 1 or 2 to 3. The more spherical guest 14 shows a decrease
in binding affinity; the other trimethylammonium compound,
1
6, shows only a 0.7 kcal/mol increase (1 f 3) even though it
is a dication designed to position charges to interact with
carboxylates on both faces of the host cavity. A reasonable
explanation for these results is that guests that fit well into the
rhomboid conformation of these hosts can benefit substantially
from the additional carboxylates of 3. Those that prefer the more
open, toroid-like conformation on the host apparently cannot
benefit substantially from the additional carboxylates. Perhaps
The NMR shifts of all of the dipeptides in the presence of
cyclophane 3 support a uniform binding orientation with the
guanidinium group buried in the host cavity and the second
(
14) Forman, J. E.; Marsh, R. E.; Schaefer, W. P.; Dougherty, D. A.
Acta Crystallogr. 1993, B49, 892-896.
15) Extensive further details on many aspects of this work can be found
1
5
residue (X) remaining solvent-exposed. In our initial design,
(
in the following: Ngola, S. M. Ph.D. Thesis, California Institute of
Technology, 1998.
(16) McCurdy, A. Ph.D. Thesis, California Institute of Technology, 1995.
(17) Eisenberg, D.; McLachlan, A. D. Nature 1986, 319, 199-203.

1196 J. Am. Chem. Soc., Vol. 121, No. 6, 1999
Ngola et al.
Figure 6. Graph of the trend in binding energy (-∆G° in kcal/mol)
to 3 with length of alkyl chain (n), where n ) 1 corresponds to Gly-
2
Arg-NH (22). The guests included in this plot are (increasing n) 22,
2
4, 25, 27, and 28. Data are from Figure 5.
Figure 5. Binding energies (-∆G°) of a set of dipeptides complexed
with 3 and the pK values of the amines of the corresponding amino
acids. %P is the percentage of the amine that is protonated at the pH
of the buffer used in the binding studies.
Figure 7. Binding for hosts 4/1 to selected guests. Values are in 10%
CH CN except the data in brackets, which is 100% aqueous (see text).
a
3
1
2
alkyl chain (29) caused a significant reduction in binding relative
to its charged counterpart. Methylation of the terminal amine
had little impact on the peptide’s affinity for the host. To verify
that the position of the amine in the second residue is most
crucial, 23 and 24 were compared. These dipeptides are isomers,
with virtually identical pKa values for the amine. The difference
in their affinity for 3 is 0.6 kcal/mol. This shows that the position
of the amine in the chain is clearly important and can
significantly affect the binding energy.
If a tight electrostatic interaction (i.e., a salt bridge) between
the terminal amine and the carboxylate is important, then
increasing the salt content of the buffer should cause the
peptide’s affinity for 3 to decrease. Indeed, we find that for
guest 24, increasing the salt concentration from 10 mM to 1 M
drops the binding affinity from 5.5 to 4.6 kcal/mol, evidence
of a substantial electrostatic component. Hydrophobic binding
is usually increased by raising the salt concentration.
Asn was chosen as the second residue. A preliminary exercise
with CPK models placed the amide side chain in a position to
interact favorably with the carboxylates. However, the dipeptide
(21) actually had a lower affinity than Arg-NH2 (19) for
cyclophane 3. This is most likely the result of the increased
hydration of the amide side chain.
In our second approach, we started with glycine as the second
residue and varied it by increasing the number of methylene
groups between the terminal ammonium and the amide bond
(see structures in Figure 5). The amine is tethered to the arginine
residue by different lengths which determine its proximity to
the host carboxylates. Although the overall energy changes are
not large, we believe that for a set of such closely related
compounds, the trends across the series are meaningful, even
though any single comparison may present differences that are
only marginally outside our error bars.
The trend in the affinity for 3 of peptides terminating in
primary amines is visible in the graph shown in Figure 6. There
is clearly a peak when the alkyl chain is three units long,
corresponding to γ-Abu-Arg-NH2 (25). This implicates a
specific interaction that depends on the precise position of the
terminal amine relative to the host carboxylates, rather than just
a generic electrostatic effect.
A brief study of the affinity of dipeptide guests for cyclophane
2 was also conducted. In general, we find a roughly 1 kcal/mol
drop in affinity on going from 3 to 2.
Neutral Host 4. Figure 7 summarizes binding data for host
4. Also shown for comparison are corresponding data for
tetraanionic host 1. In all cases, NMR chemical shift changes
seen in guests bound to host 4 are completely analogous to those
9
seen for binding to 1. Thus, the basic binding geometries are
Figure 5 shows that the binding energies do not correlate with
the amount of charge. There is an increase in pKa with increasing
number of methylene groups with the estimated degree of
protonation ranging from 83 to 98%. When n g 2, >90% of
the amine is protonated. If this were purely a general electrostatic
effect, the binding energy should scale with the percentage of
protonated amine. This is certainly not the case, although we
cannot rule out the possibility that binding affects the pKa values
in a way that is not consistent across the series.
the same in the two systems.
An important general observation concerning 4 should be
emphasized. This completely neutral host still shows a sub-
stantial binding preference for cationic guests over neutral
guests. Recall that in 10% CH CN, required for studies of 4,
3
neutral guests analogous to 30-33 do not bind to cyclophanes
such as 1 (or 4) within our detection limits (i.e., -∆G° < 3.5
1
kcal/mol). Thus, the cation-π interaction is still quite obvious
in this system, again establishing that the binding of cations by
1 was not primarily due to the carboxylates. We have previously
studied a neutral analogue of 1 in chloroform, and seen the same
Hydrophobic interactions have a minimal role in binding the
second residue. As seen in Figure 5, the trend in binding energy
does not scale with hydrophobicity. Incorporation of a neutral
1
8
trend.

Receptor for Arg DeriVatiVes in Aqueous Media
J. Am. Chem. Soc., Vol. 121, No. 6, 1999 1197
+
In general, however, 4 binds cations less tightly than 1. The
difference varies somewhat, but averages to about 1 kcal/mol
in -∆G°. Importantly, this difference seems to be independent
of the medium, in that it is essentially the same with 30 whether
in the purely aqueous or the 10% acetonitrile system. It does
appear, then, that the four remote, highly exposed, and thus
highly water-solvated, carboxylates of 4 can contribute to the
binding of a cation. Given a number of studies that would
suggest that highly solvated ion pairs contribute little energeti-
cally, even when the ions are much closer than in the present
system, this could be considered a surprising observation. These
results will be discussed further below.
are focused on the N -H bonds, but, surprisingly, for 9 and
10 they are centered on the C-H hydrogens from C-1, adjacent
to the N . We propose that in order for an electrostatic
+
interaction with a carboxylate to be energetically significant in
an exposed aqueous environment, the cation must have a region
of intense positive electrostatic potential. For a guest such as
14, it is not worth desolvating the carboxylate in order to
establish a tight interaction with the diffuse positive charge of
+
a R-NMe3 group. Recall that studies with the less solvating
10% CH3CN medium establish that guest 14 can bind within
the cavity of host 2 in such a way that the positive charge is
close to the fifth carboxylate. It is just that this interaction is
energetically insignificant in pure water.
Discussion
The large amount of negative charge presented by 3 strongly
influences binding. Now, neutral guests such as 7 and 8 show
diminished binding relative to hosts 1 and 2. Presumably this
is because a primarily hydrophobic binding interaction is
incompatible with the very polar environment around the rim
of the host. However, most cationic guests benefit substantially
from the additional charges, with only 14 and 18 remaining as
stubborn exceptions. The effect can be quite large, approaching
A number of studies of protein and peptide systems have
provided estimates for the importance of a salt bridge in
19
stabilizing protein structure. There is considerable variation,
but such interactions are generally in the 0-3 kcal/mol range.
The higher values are for more buried interactionsssurface
exposed interactions are typically worth e1 kcal/mol. In
contrast, on the basis of a survey of a broad range of synthetic
receptor types, Schneider has suggested a consistent contribution
2
kcal/mol (relative to 1) for guest 9.
2
0
It is clear that host 3 can bind well-solvated guests through
of 1.2 ( 0.2 kcal/mol for an ion pair in aqueous media.
a combination of hydrophobic, cation-π, and ion-pairing interac-
tions. This suggested that even more soluble structures such as
the cationic amino acids Arg and Lys might be bound by 3,
and this is clearly true. Arg is a better guest than Lys (i.e., 19
is bound more tightly than 20 by 1 kcal/mol), although the water
solubilities of the two are similar. We attribute this difference
to the cation-π interaction. A number of arguments suggest that
a guanidinium as in Arg is especially well suited to a cation-π
In the present work we have made several attempts to create
well-defined electrostatic interactions and evaluate their ener-
getic consequences for binding in a relatively exposed, aqueous
environment. While there is some scatter in the data, several
general conclusions emerge.
With host 2 the consequences of a single ion-pair interaction
were evaluated. In some systems, a binding enhancement on
the order of 0.5-0.6 kcal/mol is seen (9-12). Importantly,
sterically similar neutral guests are unaffected by the fifth
carboxylate of 2, greatly simplifying the analysis. With other
guests such as 14-17, the effect is smaller: 0 to 0.3 kcal/mol.
The results for the well-studied guest 14 were surprising. NMR
and computer-modeling studies suggested the quaternary am-
monium group should be able to achieve van der Waals contact
with the fifth carboxylate of 2, yet no binding enhancement is
seen.
2
interaction. The flat delocalized π system stacks well on
2
2
aromatics. In addition, there is some evidence that the face of
a guanidinium is hydrophobic, further encouraging stacking on
aromatics. Finally, surveys of protein structures seem to indicate
that cation-π interactions involving Arg are more common than
those involving Lys, although both occur. The interaction of
Arg stacking on the side chain of Trp seems especially favorable.
To probe for a tight, specific salt-bridge interaction with the
carboxylates surrounding the rim of the host 3, the dipeptide
structures of Figure 5 were studied. The sum of the results from
Figures 5 and 6 strongly suggests a specific and stabilizing salt
In previous studies, we have found that electrostatic potential
surfaces (eps) can provide valuable insights into noncovalent
1
,2,21
interactions that have a significant electrostatic component.
+
Considering the guests of Figure 2,12 the eps of tetraalkyl-
bridge between an RNH3 group of the guest and a carboxylate
(or perhaps more than one) of host 3. The effect peaks at a
ammonium compounds such as 14 are relatively diffusesthere
particular chain length (Figure 6). This indicates that a fairly
precise geometrical arrangement is required, consistent with a
salt bridge, but not with a global, electrostatic interaction. Ionic
strength effects are consistent with a salt bridge. Hydrophobic
effects are not controlling-adding a CH2 can lead to a decrease
in binding. The magnitude of the salt bridge interaction depends
on the reference state. Taking the optimal case, 25, the extra
salt bridge could be worth as much as 1.3 kcal/mol (21 as
reference) or 0.7 kcal/mol (22 as reference).
The binding of 25 by 3 represents a very rare example of
using a highly polar interaction, such as a salt bridge or hydrogen
bond, to enhance molecular recognition in an articifical receptor
in fully aqueous media. The results indicate that in favorable
cases a combination of binding interactionsscation-π, hydro-
phobic, and salt bridgescan lead to effective binding of very
polar guests in water. They also suggest the potential for
developing selective artificial receptors for Arg residues in
peptides and proteins.
are no focal points of intense positive charge. In contrast,
structures with N -H bonds, such as 11 and 12 and quinolinium
compounds such as 9 and 10, do have localized regions of
relatively intense positive charge. For 11 and 12 these regions
+
(
18) Stauffer, D. A.; Dougherty, D. A. Tetrahedron Lett. 1988, 29, 6039-
042.
19) See, for example: Brown, L. R.; De Marco, A.; Richarz, R.; Wagner,
6
(
G.; W u¨ thrich, K. J. Biochem. 1978, 88, 87-95. Dao-pin, S.; Sauer, U.;
Nicholson, H.; Matthews, B. W. Biochemistry 1991, 30, 7142-7153.
Hendsch, Z. S.; Tidor, B. Protein Sci. 1994, 3, 211-226. Serrano, L.;
Horovitz, A.; Avron, B.; Bycroft, M.; Fersht, A. R. Biochemistry 1990, 29,
9
2
7
6
343-9352. Lyu, P. C.; Gans, P. J.; Kallenbach, N. R. J. Mol. Biol. 1992,
23, 343-350. Sali, D.; Bycroft, M.; Fersht, A. R. J. Mol. Biol. 1991, 220,
79. Tissot, A. C.; Vuilleumier, S.; Fersht, A. R. Biochemistry 1996, 35,
786-6794. Scholtz, J. M.; Qian, H.; Robbins, V. H.; Baldwin, R. L.
Biochemistry 1993, 32, 9668-9676. Waldburger, C. D.; Schildbach, J. F.;
Sauer, R. T. Struct. Biol. 1995, 2, 122-128. Blasie, C. A.; Berg, J. M.
Biochemistry 1997, 36, 6218-6222. Schneider, J. P.; Lear, J. D.; DeGrado,
W. F. J. Am. Chem. Soc. 1997, 119, 5742-5743.
(
20) Schneider, H.-J.; Schiestel, T.; Zimmerman, P. J. Am. Chem. Soc.
992, 114, 7698-7703.
21) Mecozzi, S.; West, A. P., Jr.; Dougherty, D. A. Proc. Natl. Acad.
1
(
Sci. U.S.A. 1996, 93, 10566-10571. Mecozzi, S.; West, A. P., Jr.;
Dougherty, D. A. J. Am. Chem. Soc. 1996, 118, 2307-2308.
(22) Boudon, S.; Wipff, G.; Maigret, B. J. Phys. Chem. 1990, 94, 6056-
6061.

1198 J. Am. Chem. Soc., Vol. 121, No. 6, 1999
Ngola et al.
attenuated by the hydration. The important point, though, is that
the intervening medium between the carboxylates of 1 and a
cationic guest is not bulk water, but rather a polarizable benzene
ring. A magnitude of 0.3 kcal/mol for such an interaction seems
quite plausible. The fact that changing the character of the
solvent by adding 10% CH3CN does not change the magnitude
of this effect is consistent with this model. Note that this effect
will not occur if just any aromatic system is positioned randomly
between two charges. The host system locks the aromatic units
into a specific alignment, such that the dipole induced by the
anion is pointed toward the cationic guest. The situation is
clearly different for the additional carboxylates of hosts 2 or 3,
which are closer to the cationic guests, yet in some instances
provide no stabilizing interaction. Now the intervening medium
is best though of as water, and it can substantially, or even
completely, attenuate ionic interactions.
The picture that emerges from studies of 1 and 4 is
reminiscent of a number of biological binding sites for organic
+
cations. Two crystal structures of proteins that bind R-NMe3
4
groups have appeared: the acetylcholine esterase and the
phosphocholine-binding antibody Fab McPC603. In each case
5
Figure 8. Induced-dipole model for the long-range interaction of the
carboxylates of host 1 and a cationic guest. Top: schematic of the
induced dipole. Bottom: the isodesmic reaction used to evaluate the
phenomenon. Results are from ab initio, 6-31G** calculations. Note
that on the “reactant” side, the Na -Cl structure has the ions 7.55 Å
apart; the same separation as on the “product” side.
the quaternary ammonium group is in van der Waals contact
with aromatic rings, while one or more carboxylate groups are
further away, but still within likely interaction distance. A
computation model of G protein-coupled receptors that bind
+
-
6
acetylcholine and other amines suggests a similar arrangement,
and all available data on the nicotinic acetylcholine receptor7
are also consistent with such a model. Thus, inadvertently, host
Given the findings for hosts 2 and 3, the results from
comparing hosts 1 and 4 suggest a surprisingly large long-range
ionic interaction. The average increment by which 1 is a better
host than 4 is about 1 kcal/mol, but since there are four
carboxylates, e0.3 kcal/mol is implied for each interaction.
However, given that in some situations probed here contact ion
pairs appear to have interaction energies <0.3 kcal/mol, it is
difficult to understand the stabilization seen due to the remote
charges of 4. It is not possible for the host to contort in such a
way as to bring these carboxylates close to a charge of a guest.
Also, the carboxylates are highly exposed to water and should
be well solvated.
1
may have captured the essence of biological binding sites for
acetylcholine and related structuressa belt of aromatics for
cation-π interactions augmented by more remote carboxylates
for ion-pair interactions.
Conclusions
A number of interesting conclusions emerge from these
studies. For ion-pair/salt-bridge interactions that are substantially
+
exposed to an aqueous medium, diffuse cations of the RNMe3
type are unlikely to experience significant stabilization. How-
ever, ions with more focused electrostatic potential, such as those
We have considered the possibility that the aromatic rings
of the host provide an especially favorable intervening medium
for transmitting long-range electrostatic effects, and we have
+
containing N -H groups, can substantially benefit from such
interactions. When properly designed, as in the 3/25 pair,
significant binding energy can develop. Through a combination
of interactionssincluding cation-π, hydrophobic, and salt
bridgesa receptor that shows a strong affinity and considerable
selectivity for arginine derivatives can be developed. The results
with host 4 suggest that fairly remote charges can have a small
but nonnegligible effect on binding if the intervening medium
is a properly positioned aromatic.
attempted to model this computationally. On the basis of a
crystal structure of the tetraester of 1,14 we estimate that the
distance from the center of charge of a carboxylate in 1 to the
center of the nearest benzene ring (from the ethenoanthracene)
is 5.1 Å. The question is whether an anion at this distance can
influence the binding of a cation to the opposite face of the
benzene ring. A nearby anion could induce a dipole in the
aromatic in such a way as to enhance binding of a cation to the
opposite face of the aromatic (Figure 8). To evaluate this
Experimental Section
-
possibility we placed (computationally) a Cl 5.1 Å from an
General.9,12,15 NMR spectra were recorded on Bruker AMX500,
aromatic ring and evaluated its influence on the binding of a
Bruker AM500, or General Electric QE-300 spectrometers. Routine
spectra were referenced to the residual proton signals of the solvents
and are reported in ppm downfield of 0.0 as δ values. All coupling
constants, J, are in hertz. Spectra from aqueous binding studies were
referenced to an internal standard of 3,3-dimethyl glutarate (DMG, δ
+
9
+
Na to the opposite face of the ring. With the Na at its optimal
+
-
distance from the benzene (2.45 Å), the Na and the Cl are
.55 Å apart. The induced dipole contribution to Na binding
+
7
can be considered to be ∆E for the isodesmic reaction of Figure
. That is, the “reaction” of Figure 8 evaluates the extent to
1.09). Spectra from mixed solvent (10% acetonitrile in aqueous 10 mM
8
-
+
NaCl) binding studies were referenced to the individual proton signal
of acetonitrile (δ 1.94). Preparative centrifugal chromatography was
performed on a Harrison Research Chromatotron model 7924T using
silica plates. HPLC was performed on a Waters dual 510 pump liquid
chromatograph system equipped with a Waters 996 photodiode array
detector and a Waters 490E wavelength detector. Mass spectral analyses
were performed at the University of California, Riverside, the Nebraska
Center for Mass Spectrometry, and Caltech. Tetrahydrofuran and diethyl
which it is favorable to have Cl and Na bind simultaneously
to opposite faces of a single benzene rather than have each bind
to an individual benzene. The value of -5.1 kcal/mol for ∆E
indicates that the intervening benzene ring does enhance the
long-range electrostatic interaction by a significant amountsit
is better to have an intervening benzene ring than to have a
vacuum. Of course, the 5.1 kcal/mol value will be greatly

Receptor for Arg DeriVatiVes in Aqueous Media
J. Am. Chem. Soc., Vol. 121, No. 6, 1999 1199
ether were distilled from sodium benzophenone ketyl, methylene
chloride and acetonitrile were distilled from calcium hydride, and
anhydrous N,N-dimethylformamide and anhydrous dimethyl sulfoxide
were purchased from Aldrich and used without further treatment. All
guests were either commercially available or were prepared by
exhaustive methylation of the appropriate amines, with the exception
of 11 and 18 which were prepared according to literature procedures.
Commercially available Fmoc amino acids were purchased from
Novabiochem. The peptide solid support resin and the reagents used
in the peptide coupling and deprotection were obtained from Nova-
“Three-Quarters” Host. Into a 100 mL three-necked round-
bottomed flask fitted with a stopper, a septum, and an Ar gas adapter
were placed 1.46 g of cesium carbonate (4.50 mmol, 15 equiv) and
528 mg of (9R,10R)-2,6-dihydroxy-11,12-dicarbomethoxyetheno-
1
anthracene (1.50 mmol, 5 equiv). The flask was then purged with
argon, and 40 mL of anhydrous dimethyformamide was injected into
the vessel. A solution of p-xylene dibromide (79 mg, 0.30 mmol, 1.0
equiv) in 10 mL anhydrous DMF was then delivered via syringe pump
into the ethenoanthracene solution over 4 h. The reaction mixture was
then allowed to stir at room temperature for 2 days. (Note: both the
reaction vessel and the gastight syringe used in the reaction were
covered with foil to shield the reactants and products from light.) The
solution was then acidified using a few drops of concentrated aqueous
HCl until the yellow color of the solution dissipated. The solution was
then filtered to remove residual solids, and the solvent was stripped
off in vacuo. Preparative centrifugal thin-layer chromatography was
employed for further purification. Using 2-mm thick silica gel plates
and a 3:1 (v/v) eluent solution of ether:petroleum ether, the excess
ethenoanthracene (250 mg, 0.71 mmol) was first recovered. Altering
the eluent solution to a 1:1 mixture of diethyl ether:chloroform yielded
the desired compound (147 mg, 0.18 mmol, 61%): 1H NMR (acetone-
2 2
biochem, Fluka, and Aldrich. Arg-NH (19) and Lys-NH (20) were
obtained from Sigma and also used without further purification.
Binding Constant Determinations. Studies with hosts 1 to 3 were
carried out in a standard 10 mM deuterated cesium borate buffer at pD
≈
9 (referred to as borate-d). Studies with host 4 were performed in
1
0% acetonitrile-d in aqueous deuterated 10mM sodium chloride. Stock
3
solutions of hosts 1 to 4 were obtained by dissolving the host in the
appropriate solvent and quantified by integration against a primary
standard solution of DMG in the same solvent. Guest solutions for
NMR binding studies were prepared by dissolution of the compounds
in the appropriate solvent. Guest solution concentrations were deter-
mined gravimetrically, by weight of solute, or through NMR integrations
against DMG. NMR binding studies were performed by sequential
addition of aliquots of guest solutions to an NMR tube containing a
solution of cyclophane at an initial concentration of approximately 200
µM (hosts 1 to 3) or 500 µM (host 4). Binding data were fit to an
appropriate association constant, using the MULTIFIT or EMUL
d ) δ 8.26 (s, 2H), 7.41 (s, 4H), 7.28 (d, J ) 8, 2H), 7.20 (d, J ) 8,
6
2H), 7.13 (d, J ) 2, 2H), 6.96 (d, J ) 2, 2H), 6.60 (dd, J ) 8, 2, 2H),
6.44 (dd, J ) 8, 2, 2H), 5.43 (s, 2H), 5.41 (s, 2H), 5.04 (s, 4H), 3.71
(s, 12H).
Cyclophane 2, Pentamethyl Ester. Cesium carbonate (276 mg, 0.84
mmol, 4 equiv) was placed in a 250 mL three-necked round-bottomed
flask fitted with a stopper, a septum, and an Ar gas adapter. The system
was then purged with argon and charged with 50 mL of anhydrous
DMF. In a separate flask, the above-prepared “three-quarters” host (171
mg, 0.21 mmol, 1 equiv) and methyl 2,5-bis(bromomethyl)benzoate
1
0
programs. Errors bars on the NMR binding constants measurements
were calculated through the analysis packages PORTIA and LU-
1
0,23
CIUS.
All circular dichroism (CD) experiments were carried out
using a JASCO J-600 spectropolarimeter with either 1.0 or 0.5 cm path-
length quartz cells. Solutions for CD studies were prepared in either
borate buffer (pH ) 9) or aqueous 10 mM sodium chloride using water
passed through a Milli-Q purification system. The host concentration
ranged from 1 to 5 µM. In a typical study, CD spectra of six solutions
of equivalent host but varying guest concentrations were used. The
spectra and ∆ꢀ values of the host were fit to an association constant
using the CDFIT program.9
(69 mg, 0.21 mmol, 1 equiv) were dissolved into 20 mL of anhydrous
DMF. The contents of this flask were then drawn into a 25-mL gastight
syringe. The solution was then injected into the cesium carbonate
suspension over 48 h via a syringe pump. After the injection was
complete, the solution was stirred for an additional 24 h. The syringe
and the main reaction vessel were shielded from light throughout the
course of the reaction. The DMF solution was then filtered to remove
cesium salts. The collected solids were washed twice with ap-
proximately 10 mL of DMF. The filtrate and washings were then
combined and the DMF removed in vacuo. The residue was then mixed
Methyl 2,5-Bis(bromomethyl)benzoate. 2,5-Dimethylbenzoic acid
(3.95 g, 26.3 mmol, 1 equiv) was combined with thionyl chloride (4.0
mL, 55 mmol, 2.1 equiv) in a 50 mL round-bottomed flask fitted with
a reflux condenser. A calcium chloride drying tube was attached to
the condenser, and the thionyl chloride was brought to reflux. After 6
h the resulting clear solution was cooled, and the reflux condenser was
replaced with a distillation apparatus. The excess thionyl chloride was
removed under reduced pressure. The residue was then diluted with
with CHCl
was then purified via flash chromatography over silica gel with an eluent
mixture of 5% Et O in CH Cl (v/v). Further purification using
preparative centrifugal thin-layer chromatography (1-mm thick silica
gel plate using an eluent gradient of CH Cl to 5% Et O in CH Cl
v/v)) afforded the pentamethyl ester of the host (53 mg, 0.05 mmol,
3
, filtered, and concentrated in vacuo. The crude material
2
2
2
2
2
2
2
2
2
0 mL of methanol and allowed to stir at room temperature overnight.
(
2
1
The methanol was then removed via rotary evaporation. The resulting
oil was dissolved in methylene chloride (50 mL) and extracted twice
with a saturated solution of sodium bicarbonate (2 × 100 mL), followed
by two further extractions with distilled water (2 × 100 mL). The
organic layer was then dried over magnesium sulfate, filtered, and
6% yield): 1H NMR (CDCl
3
) δ 7.95 (d, J ) 2, 1H), 7.37 (d, J ) 8,
H), 7.37 (dd, J ) 8,2, 1H), 7.20 (s, 4H), 7.09 (d, J ) 8, 1H), 7.07 (d,
J ) 8, 2H), 7.05 (d, J ) 8, 1H), 6.90 (d, J ) 2, 1H), 6.89 (d, J ) 2,
1
6
1
H), 6.88 (d, J ) 2, 1H), 6.84 (d, J ) 2, 1H), 6.40 (dd, J ) 8, 2, 1H),
.39 (dd, J ) 8, 2, 1H), 6.36 (dd, J ) 8, 2, 1H), 6.35 (dd, J ) 8, 2,
H), 5.47 (s, 2H), 5.22 (s, 1H), 5.21 (s, 2H), 5.20 (s, 1H), 5.12-4.99
concentrated to yield the methyl ester (4.05 g, 24.6 mmol, 93%): 1
NMR (CDCl ) δ 7.71 (s, 1H), 7.16 (d, J ) 3, 1H), 7.10 (d, J ) 3,
H),3.86 (s, 3H), 2.53 (s, 3H), 2.32 (s, 3H). This ester (2.46 g, 15.0
mmol, 1 equiv) was dissolved in 50 mL of methylene chloride in a
00 mL round-bottomed flask. N-Bromosuccinimide (5.61 g, 3.15
H
+
3
(m, 6H), 3.91 (s, 3H), 3.75 (s, 12H); FAB-MS m/e 967 (MH ); HRMS
1
+
9
67.2986, calcd for C58
H
47
O
14 (MH ) 967.2965.
Cyclophane 2. The pentaester was dissolved in 2 mL of DMSO,
1
1
2.5 equiv of cesium hydroxide in a 1 M aqueous solution (2.5 equiv
mmol, 2.1 equiv) was added to the flask. The reaction vessel was fitted
with a reflux condenser, and its contents were heated to refluxing. The
flask was irradiated for 7 min using a General-Electric model RSKB
sunlamp with a 275 W, 110-125 V, ac bulb. The solution was allowed
to reflux overnight. It was then cooled and filtered, to remove
precipitated succinimide, and concentrated. The residue was chromato-
for each methyl ester) was added, and the solution allowed to stir
overnight at room temperature. Water (1-2 mL) was then added, and
the solution was allowed to stir for 24 h. The solution was frozen and
lyophilized. The residue was dissolved in 25 mM aqueous ammonium
acetate solution (1-2 mL) with acetic acid being added as necessary
to bring the solution to pH ≈ 7. The pentaacid was then isolated by
preparative HPLC. Using a Whatman Magnum 9 column (50 cm,
Partisil 10, ODS-3), we found the following HPLC conditions suit-
able: flow rate of 4.5 mL/min; observation at 260 nm; eluent gradient
of 25 mM aqueous ammonium acetate (t ) 0 through t ) 10 min),
followed by a gradient eluent of 0 to 30% acetonitrile in 25 mM
ammonium acetate over 50 min. The desired product eluted after
approximately 42 min. The appropriate collected fractions were
combined, frozen, and lyophilized. The material was then purified using
2 2
graphed over silica gel, using CH Cl as an eluent to separate the
product from residual succinimide. The resulting material was twice
recrystallized from cyclohexane to yield the desired dibromide in
1
purified form (980 mg, 3.04 mmol, 20% yield): H NMR (CDCl
3
) δ
7
.97 (d, J ) 2, 1H), 7.50 (dd, J ) 6, 2, 1H), 7.43 (d, J ) 6, 1H), 4.92
(s, 2H), 4.46 (s, 2H), 3.93 (s, 3H).
(
23) Barrans, R. E, Jr. Ph.D. Thesis, California Institute of Technology,
1
992.

1200 J. Am. Chem. Soc., Vol. 121, No. 6, 1999
Ngola et al.
+
a cation exchange column (neutral pH, Dowex 50 × 4, NH
4
form).
Cyclophane 3. Hydrolysis was performed by dissolving the octaester
(11 mg, 10 µmol, 1 equiv) in DMSO (1 mL) and adding CsOH (1 M
aqueous solution, 578 µL, 578 µmol, 60 equiv). After ∼34 h, the
reaction mixture was lyophilized. The resulting solid was redissolved
in water and run down a freshly regenerated (20% aqueous HCl) Diaion
(WK100) weakly acidic ion exchange column eluting with Millipored
water. The UV-active fractions were collected and lyophilized to obtain
a pale yellow solid. The colored impurities were removed by running
the material through a Sephadex G-15 column (1 g) eluting with
Millipored water. (Impurities can also be removed by running the
The fractions containing host were identified by their UV activity and
then combined and lyophilized to give the acid compounds: 1H NMR
(
10% CD
3
CN/90% borate, referenced to internal DMG δ 1.09) δ 7.53
s, 1H), 7.37 (d, J ) 8, 2H), 7.36 (s, 4H), 7.28 (d, J ) 8, 1H), 7.23 (d,
J ) 8, 1H), 7.21 (d, J ) 8, 1H), 7.19 (d, J ) 8, 1H), 7.17 (d, J ) 8,
H), 7.03 (bs, 4H), 6.62 (dd, J ) 8, 2, 1H), 6.60 (dd, J ) 8, 2, 1H),
.59 (dd, J ) 8, 2, 1H), 6.54 (dd, J ) 8, 2, 1H), 5.32 (s, 2H), 5.19 (s,
H), 5.18 (s, 1H), 5.13 (d, J ) 4, 2H), 5.10 (bs, 4H).
,5-Dicyano-p-xylene. 2,5-Dibromo-p-xylene (7.92 g, 30 mmol,
equiv) and CuCN (8.1 g, 90 mmol, 3 equiv) were refluxed in DMF
(
1
6
3
2
4
2
3
1
material down a Sep-Pak C18 cartridge (6 cm , 1 g) and eluting with
(
100 mL) for 2 days under argon. The reaction mixture was poured
into a solution of NH OH (250 mL) yielding a brown precipitate. The
solid was filtered off, washed with NH OH (250 mL) and water (500
3
0-20% CH CN/water). The product was lyophilized to obtain a white
1
4
solid (5.3 mg): H NMR (borate-d) δ 7.40 (s, 4H), 7.22 (d, J ) 8,
4H), 7.03 (d, J ) 2, 4H), 6.53 (dd, J ) 2, 8, 4H), 5.21 (AB, J ) 13,
∆ν ) 19 Hz, 8H), 5.22 (s, 4H).
4
mL), and left open to air-dry. The solid was then hot-extracted with
acetone in a Soxhlet apparatus (250 mL, 3 days) yielding a cloudy
white solution with a white precipitate. The material was concentrated
by removal of the solvent under vacuum to yield a white solid with a
yellow tinge. The residual copper was removed by dissolving the
Host 4. Into a 5 mL round-bottom flask were placed 25 mg (0.027
mmol) of host 5 (prepared according to standard procedures), 68 mg
of tris (0.56 mmol), and 50 mg of potassium carbonate (dried overnight
at 350 °C). The flask was fitted with a septum, and 0.25 mL of
anhydrous DMSO was injected. The slurry was vigorously stirred at
the temperature of 60 °C (the reaction vessel was covered with foil to
shield the reactants and the products from light). After 60 h (TLC,
material in CHCl
crystals (3.78 g, 24 mmol, 81% yield): H NMR (CDCl
H), 2.55 (s, 6H).
,5-Dimethylterephthalic Acid, Dimethyl Ester.25 A stirred mixture
of the dicyano compound (3.48 g, 22 mmol, 1 equiv.), KOH (3.57 g,
9 mmol, 4 equiv), and diethylene glycol (50 mL) was refluxed
overnight under argon. The reaction mixture was then diluted with water
100 mL) and acidified to pH 1 with 10% HCl. A brown solid was
3
and running it down a silica gel plug to obtain white
1
3
) δ 7.56 (s,
2
2
CHCl /MeOH 7:3) the solution was cooled, 0.5 mL of DMSO was
3
added, and the solid (mainly K CO ) was filtered away. The DMSO
2
3
8
solution was then frozen and lyophilized. The remaining residue was
dissolved in 2 mL of water, frozen, and lyophilized again. This
procedure was repeated twice in order to eliminate any trace of DMSO.
The lyophilization products (white powder) were dissolved in 3 mL of
MeOH, and undissolved solid, mainly tris, was filtered away. The
methanolic solution was concentrated and depleted of the excess of
(
filtered off and left to air-dry overnight. The solid was then dissolved
in 10% NaOH, and the resulting solution was decolorized with charcoal.
The solution was then acidified and filtered to yield an off-white solid
which was dried under vacuum. MeOH (100 mL) and MeSO
mL) were placed in a flask with the solid and brought up to reflux for
0 h. The reaction was poured into EtOAc (200 mL), 1 M KH PO
4
buffer (pH ) 7, 400 mL), and brought up to pH 7 with a saturated
aqueous solution of sodium bicarbonate. The organic layer was isolated,
and the aqueous layer was washed with EtOAc. The EtOAc extracts
were combined and dried over sodium sulfate. The solvent was removed
using a rotary evaporator yielding an orange-tinged solid. The solid
was dissolved in ether and run down a silica gel plug. After the solvent
was removed using a rotary evaporator, the resulting material was
3
H (10
tris by a fast column chromatography (CHCl /MeOH 7:3). Pure host 4
3
was recovered by a second column chromatography using a solvent
gradient (95-75% chloroform in methanol) (21.5 mg 0.017 mmol,
1
2
1
62%): H NMR (methanol-d ) δ 7.21 (s, 8H), 7.02 (d, J ) 8, 4H),
4
6.88 (d, J ) 2, 4H), 6.54 (dd, J ) 2,8, 4H), 5.07 (AB, 8H), 4.33 (s,
4H), 3.65 (AB, 24H), 3.19 (s, 4H); 13C NMR (methanol-d ) δ 174.30,
4
156.76, 144.56, 137.06, 132.51, 126.66, 125.06, 111.49, 111.36, 69.12,
+
62.13, 61.27; FAB-MS m/e 1291 (MNa ).
N-(Fmoc), N-Methyl-γ-aminobutyric Acid. N-Me-γ-Abu (1.53 g,
1
0 mmol, 1 equiv) was dissolved in 10% aqueous Na
Dioxane (30 mL) was added yielding a clear solution. Next, Fmoc-Cl
2.6 g, 10 mmol, 1 equiv) was added, and the solution was heated in
2 3
CO (26.5 mL).
washed with hexanes to obtain white crystals (3.91 g, 18 mmol, 80%
1
yield): H NMR (CDCl
3
) δ 7.76 (s, 2H), 3.90 (s, 6H), 2.55 (s, 6H).
(
2
,5-Bis(bromomethyl)terephthalic Acid, Dimethyl Ester. The
above ester (1.00 g, 4.5 mmol, 1 equiv), N-bromosuccinimide (1.68 g,
.5 mmol, 2.1 equiv), and benzoyl peroxide (∼5 pellets) were placed
in CCl (15 mL) under argon. The reaction was initiated by bringing it
a 70 °C oil bath. The solution gradually turned yellow, with bubbling
and formation of a white fluffy precipitate. After 6 h, the bubbling
ceased, and the reaction mixture was poured into water (400 mL) and
extracted twice with ether (100 mL). Concentrated HCl was added to
bring the aqueous layer to pH 3, resulting in a cloudy solution. The
solution was left at 4 °C overnight and filtered to obtain a gummy
9
4
to rapid reflux with a heat gun, and the reflux rate was maintained
using a heating mantle. Additional benzoyl peroxide (∼5 pellets) was
added after 9 h because thin-layer chromatography (1:1 ether/hexanes)
indicated that the reaction had not gone to completion. After a further
yellow solid. This solid was washed with water and dissolved in CH
and the solvent was removed under vacuum to yield a yellow oil,
recrystallized from H O/CH CN to obtain white crystals (1.89 g, 5.6
mmol, 56% yield): 1H NMR (CD
OD) δ 7.79 (d, J ) 7, 2H), 7.60 (d,
J ) 7, 2H), 7.33 (m, 4H), 4.53 (d, J ) 5, 1H), 4.43 (d, J ) 5, 1H),
2 2
Cl ,
1
2 h, all solids were floating on the top of the solution which signaled
the consumption of the N-bromosuccinimide to form succinimide. The
solids were filtered off and recrystallized from CCl . The product was
a white crystalline solid (0.72 g, 1.9 mmol, 42% yield): H NMR
CDCl ) δ 8.05 (s, 2H), 4.93 (s, 4H), 3.98 (s, 6H).
Cyclophane 3, Octamethyl Ester. Macrocyclization was performed
in CH CN according to published procedures. After completion, the
reaction mixture was filtered and the solvent was removed. The residue
was flash-chromatographed over silica gel, eluting first with CH Cl
2
3
3
4
1
4
3
1
.23 (t, J ) 5, 1H), 3.29 (t, J ) 7, 1H), 2.96 (t, J ) 7, 1H), 2.78 (s,
(
3
H), 2.22 (t, J ) 7, 1H), 1.95 (t, J ) 7, 1H), 1.76 (m, 1H), 1.45 (m,
+
H); FAB-MS m/e 340.2 (MH ); HRMS (matrix, NBA) calcd for
+
3
C
H21NO
‚H 340.1549, found 340.1545, ∆ ) 1.3 ppm.
20
4
Solid-Phase Peptide Synthesis. Rink Amide MBHA resin (0.5
2
2
mmol/g) was used to afford carboxyl terminus primary amides. Peptide
synthesis was carried out using Fmoc-protected amino acids. Typical
protocols for coupling a residue involved 50-90 min coupling cycles
with 2 equiv of amino acid. Activated esters were formed in situ using
benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium hexafluorophos-
phate (PyBOP), 1-hydroxybenzotriazole (HOBt), and diisopropylethyl-
amine (DIPEA). Deprotection of Fmoc-protected amine groups was
performed using a 20-min cycle with 20% piperidine/dimethylform-
amide (DMF). The peptides were cleaved from the resin by treatment
with trifluoroacetic acid (TFA)/m-cresol/ethanedithiol (92.5:1.3:6.2) for
and then ether in order to separate the macrocyclic compounds from
baseline impurities. The macrocycle was then isolated from oligomers
using preparative centrifugal thin-layer chromatography (silica gel
plates, 0-10% ether in CH
2 2
Cl ). The product was a white, translucent
1
film (33 mg, 30 µmol, 8% yield): H NMR (CDCl
3
) δ 8.04 (s, 4H),
7
.07 (d, J ) 8, 4H), 6.92 (d, J ) 2, 4H), 6.38 (dd, J ) 2, 8, 4H), 5.39
(
1
AB, J ) 14, ∆ν ) 66 Hz, 8H), 5.23 (s, 4H), 3.78 (s, 12H), 3.75 (s,
+
2H); FAB-MS m/e 1141.3 (MH ); HRMS (matrix, NBA) calcd for
+
C
64
H
52
O
20‚H 1141.3130, found 1141.3188, ∆ ) -5.1 ppm.
9
0 min. The resin was filtered and washed with TFA. The combined
(
24) Rehahn, M.; Schl u¨ ter, A.-D.; Feast, W. J. Synthesis 1988, 386-
88.
25) Anzalone, L.; Hirsch, J. A. J. Org. Chem. 1985, 50, 2128-2133.
filtrates were cooled to 0 °C, thioanisole/trimethylsilylbromide (TMSBr)
(47:53) added and left under argon for 15 min. The peptide was
3
(

Receptor for Arg DeriVatiVes in Aqueous Media
J. Am. Chem. Soc., Vol. 121, No. 6, 1999 1201
+
‚H+ 259.1882,
precipitated with ether and left at 4 °C overnight. The precipitate was
trapped on a fine sintered funnel and washed with a large volume of
22 6 2
(MH ); HRMS (matrix, NBA) calcd for C10H N O
found 259.1873, ∆ ) 3.8 ppm.
N-Me-γ-Abu-Arg-NH
ether. The peptide was then dissolved in H
2
O and lyophilized. The
O/acetonitrile
1
2
(26): H NMR (D
2
O) δ 4.18 (dd, J ) 5, 8,
resulting peptide was triturated with H O/ether and H
2
2
1
7
(
H), 3.15 (t, J ) 7, 2H), 2.99 (t, J ) 8, 2H), 2.65 (s, 3H), 2.39 (t, J )
several times sequentially. The peptides were then lyophilized, and if
impurities were detected by NMR, reverse-phase C18 chromatography
was used for further purification.
, 2H), 1.90 (m, 2H), 1.68 (m, 2H), 1.62 (m, 2H); FAB-MS m/e 273.3
+
+
24 6 2
MH ); HRMS (matrix, NBA) calcd for C11H N O ‚H 273.2039,
found 273.2039, ∆ ) 0.0 ppm.
1
Asn-Arg-NH
2
(21): H NMR (borate-d) δ 4.32 (dd, J ) 5, 8, 1H),
1
5
-Ava-Arg-NH
2
(27): H NMR (borate-d) δ 4.28 (dd, J ) 5, 8,
3
1
.77 (t, J ) 7, 1H), 3.24 (t, J ) 7, 2H), 2.65 (m, 2H), 1.92 (m, 1H),
.79 (m, 1H), 1.69 (m, 2H); FAB-MS m/e 288.2 (MH ); HRMS (matrix,
1
H), 3.23 (t, J ) 7, 2H), 2.98 (t, J ) 7, 2H), 2.38 (t, J ) 7, 2H), 1.86
m, 2H), 1.76 (m, 2H), 1.67 (m, 4H); FAB-MS m/e 273.3 (MH );
HRMS (matrix, NBA) calcd for C11
73.2032, ∆ ) 2.5 ppm.
+
+
(
+
NBA) calcd for C10
ppm.
H
21
N
7
O
3
‚H 288.1784, found 288.1786, ∆ ) -0.8
+
H N O
24 6 2
‚H 273.2039, found
2
1
Gly-Arg-NH
2
(22): H NMR (10% AcOD/D
2
O) δ 4.27 (t, J ) 6,
1
6
2
-Ahx-Arg-NH (28): H NMR (borate-d) δ 4.33 (dd, J ) 5, 8,
1
H), 3.83 (s, 2H), 3.16 (t, J ) 7, 2H), 1.76 (m, 2H), 1.61 (m, 2H);
+
1H), 3.23 (t, J ) 7, 2H), 2.97 (t, J ) 8, 2H), 2.34 (t, J ) 7, 2H), 1.86
(m, 2H), 1.75 (m, 2H), 1.66 (m, 4H), 1.39 (m, 2H); FAB-MS m/e 287.2
FAB-MS m/e 231.2 (MH ); HRMS (matrix, NBA) calcd for C
8 18
H
N
6
O
2
‚
+
H
231.1569, found 231.1579, ∆ ) -4.3 ppm.
+
+
1
(MH ); HRMS (matrix, NBA) calcd for C13
H
26
N
6
O
2
‚H 287.2195,
Sar-Arg-NH (23): H NMR (borate-d) δ 4.35 (dd, J ) 5, 8, 1H),
2
found 287.2203, ∆ ) -2.7 ppm.
Val acid-Arg-NH
3
1
.39 (s, 2H), 3.23 (t, J ) 7, 2H), 2.27 (s, 3H), 1.89 (m, 1H), 1.78 (m,
+
1
H), 1.68 (m, 2H); FAB-MS m/e 245.2 (MH ); HRMS (matrix, NBA)
2
(29): H NMR (D
2
O) δ 4.17 (dd, J ) 5, 8, 1H),
+
calcd for C
9
20
H N
6
O
2
2
‚H 245.1726, found 245.1720, ∆ ) 2.3 ppm.
3.12 (t, J ) 7, 2H), 2.22 (t, J ) 7, 2H), 1.76 (m, 2H), 1.60 (m, 2H),
1.47 (m, 2H), 1.19 (m, 2H), 0.78 (t, J ) 7, 2H); FAB-MS m/e 258.2
1
â-Ala-Arg-NH
(24): H NMR (borate-d) δ 4.33 (dd, J ) 6, 8, 1H),
+
+
3
1
(
.23 (t, J ) 6, 2H), 3.05 (t, J ) 7, 2H), 2.59 (t, J ) 7, 2H), 1.87 (m,
(MH ); HRMS (matrix, NBA) calcd for C11
H
23
N
6
O
2
‚H 258.1930,
+
H), 1.76 (m, 1H), 1.68 (m, 2H); FAB-MS m/e 245.2 (MH ); HRMS
found 258.1929, ∆ ) 0.2 ppm.
+
matrix, NBA) calcd for C
H N
9 20 6
O
2
‚H 245.1726, found 245.1725, ∆
)
0.2 ppm.
Acknowledgment. We thank the Zeneca Pharmaceuticals
Strategic Research Fund for support of this work.
1
γ-Abu-Arg-NH (25): H NMR (borate-d) δ 4.28 (dd, J ) 5, 8,
2
1
(
H), 3.23 (t, J ) 7, 2H), 2.95 (t, J ) 8, 2H), 2.43 (t, J ) 7, 2H), 1.92
m, 2H), 1.86 (m, 1H), 1.75 (m, 1H), 1.68 (m, 2H); FAB-MS m/e 259.2
JA982499R