Pyromellitamide aggregates and their response to anion stimuli

Article abstract of DOI:10.1021/ja0713781

The N,N,N′,N″-1,2,4,5-tetra(ethylhexanoate) pyromellitamide is found to be capable of both intermolecular aggregation and binding to small anions. It is synthesized by aminolysis of pyromellitic anhydride with ethanolamine, followed by a reaction with hex

Full text of DOI:10.1021/ja0713781

Published on Web 05/11/2007
Pyromellitamide Aggregates and Their Response to Anion
Stimuli
James E. A. Webb, Maxwell J. Crossley, Peter Turner, and Pall Thordarson*
Contribution from the School of Chemistry, The UniVersity of Sydney NSW 2006, Australia
Received February 27, 2007; E-mail: p.thordarson@chem.usyd.edu.au
Abstract: The N,N′,N′′,N′′′-1,2,4,5-tetra(ethylhexanoate) pyromellitamide is found to be capable of both
intermolecular aggregation and binding to small anions. It is synthesized by aminolysis of pyromellitic
anhydride with ethanolamine, followed by a reaction with hexanoyl chloride. The single-crystal X-ray structure
of the pyromellitamide shows that it forms one-dimensional columnar stacks through an intermolecular
hydrogen-bonding network. It also forms self-assembled gels in nonpolar solvents, presumably by a
hydrogen-bonding network similar to the solid-state structure as shown by IR and XRD studies. Aggregation
by intermolecular hydrogen bonding of the pyromellitamide is also observed by NMR and IR in solution.
Fitting of NMR dilution data for pyromellitamide in d6-acetone to a cooperative aggregation model gave KE
) 232 M-1 and positive cooperativity of aggregation (F ) 0.22). The pyromellitamide binds to a range of
small anions with the binding strength decreasing in the order chloride > acetate > bromide > nitrate ≈
iodide. The data indicate that the pyromellitamide binds two anions and that it displays negative cooperativity.
The intermolecular aggregation of the pyromellitamide can also be altered using small anion stimuli; anion
addition to preformed self-assembled pyromellitamide gels causes their collapse. The kinetics of anion-
induced gel collapse are qualitatively correlated to the binding affinities of the same anions in solution. The
cooperative anion binding properties and the sensitivity of the self-assembled gels formed by pyromellitamide
toward anions could be useful in the development of sensors and switching/releasing devices.
Introduction
The vast majority of these concern the use of cationic or neutral
guests (effectors), while relatively few discrete multivalent hosts
Stimuli responsive biological systems provide some of the
most elegant demonstrations of the importance of supramo-
lecular interactions in nature. A well-known example is the
cooperative binding of small ligands to multivalent hosts, as in
the case of allosteric1 oxygen binding to hemoglobin,2 where
successive binding of the ligand(s) changes the structure and/
or function of the host in a nonlinear fashion. External stimuli
can also be used to influence intermolecular interactions as in
the case of self-assembled protein filaments, where the structure
can be modulated in vivo and in vitro with various small-
molecule effectors. For instance, iodide is known to cause the
depolymerization (collapse) of actin filaments (F-actin)3 and the
corresponding transition from a gel-like to solution state (gel
f sol transition), possibly by influencing the unique hydration
shell that is thought to surround and stabilize the F-actin
polymer.4
displaying cooperative binding toward anions have been
reported.6c,7 The intermolecular interactions of synthetic self-
assembled systems, including macroscopic self-assembled mo-
lecular gels,8 can also be tuned using small ionic or molecular
effectors. Self-assembled molecular gels are showing increased
potential for applications in biomedicine for drug delivery9 and
in tissue engineering10 (e.g., by stimulating nerve regrowth in
spinal injuries).11 To date, there have been reports on molecular
(5) For reviews, see: (a) Rebek, J., Jr. Acc. Chem. Res. 1984, 17, 258. (b)
Shinkai, S.; Ikeda, M.; Sugasaki, A.; Takeuchi, M. Acc. Chem. Res. 2001,
34, 494. (c) Takeuchi, M.; Ikeda, M.; Sugasaki, A.; Shinkai, S. Acc. Chem.
Res. 2001, 34, 865. (d) Kovbasyuk, L.; Kra¨mer, R. Chem. ReV. 2004, 104,
3161.
(6) For some recent examples, see: (a) Thordarson, P.; Bijsterveld, E. J. A.;
Elemans, J. A. A. W.; Kasa´k, P.; Nolte, R. J. M.; Rowan, A. E. J. Am.
Chem. Soc. 2003, 125, 1186. (b) Thordarson, P.; Coumans, R. G. E.;
Elemans, J. A. A. W.; Thomassen, P. J.; Visser, J.; Rowan, A. E.; Nolte,
R. J. M. Angew. Chem., Int. Ed. 2004, 43, 4755. (c) Hirata, O.; Takeuchi,
M.; Shinkai, S. Chem. Commun. 2005, 3805. (d) Wakabayashi, R.; Kubo,
Y.; Hirata, O.; Takeuchi, M.; Shinkai, S. Chem. Commun. 2005, 5742. (e)
Thomassen, P. J.; Foekema, J.; Jordana, i Lluch, R.; Thordarson, P.;
Elemans, J. A. A. W.; Nolte, R. J. M.; Rowan, A. E. New. J. Chem. 2006,
30, 148. (f) Sessler, J. L.; Tomat, E.; Lynch, V. M. J. Am. Chem. Soc.
2006, 128, 4184. (g) Ikeda, T.; Hirata, O.; Takeuchi, M.; Shinkai, S. J.
Am. Chem. Soc. 2006, 128, 16008. (h) Heo, J.; Mirkin, C. A. Angew. Chem.,
Int. Ed. 2006, 45, 941.
(7) Apart from ref 6c, the only neutral non-metallic hosts that show cooperative
binding toward anions are fluoride receptors such as: (a) Kavallieratos,
K.; Bertao, C. M.; Crabtree, R. H. J. Org. Chem. 1999, 64, 1675. Examples
of organometallic hosts showing cooperative binding to anions other than
fluoride include: (b) Uppadine, L. H.; Drew, M. G. B.; Beer, P. D. Chem.
Commun. 2001, 291. (c) Lazarides, T.; Miller, T. A.; Jeffery, J. C.; Ronson,
T. K.; Adams, H.; Ward, M. D. Dalton Trans. 2005, 528. (d) Kim, D.-S.;
Miyaji, H.; Chang, B.-Y.; Park, S.-M.; Ahn, K. H. Chem. Commun. 2006,
3314.
Much effort has been directed toward mimicking biological
systems that respond to external stimuli. There are numerous
examples of simple synthetic allosteric host-guest systems.5,6
(1) (a) Monod, J.; Changeux, J.-P.; Jacob, F. J. Mol. Biol. 1963, 6, 306. (b)
Monod, J.; Wyman, J.; Changeux, J.-P. J. Mol. Biol. 1965, 12, 88.
(2) (a) Ackers, G. K.; Doyle, M. L.; Myers, D.; Daugherty, M. A. Science
1992, 255, 54. (b) Huang, Y.; Doyle, M. L.; Ackers, G. K. Biophys. J.
1996, 71, 2094. (c) Johnson, M. L. Methods Enzymol. 2000, 323, 124.
(3) (a) Straub, F. B. Stud. Inst. Med. Chem. UniV. Szeged 1942, 2, 4. (b) Straub,
F. B. Stud. Inst. Med. Chem. UniV. Szeged 1942, 2, 7. (c) Guba, F. Nature
1950, 165, 439.
(4) Kabir, S. R.; Yokoyama, K.; Mihashi, K.; Kodama, T. Biophys. J. 2003,
85, 3154.
9
10.1021/ja0713781 CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 7155-7162
7155

A R T I C L E S
Chart 1
Webb et al.
Scheme 1. Synthesis of 1
gels that are responsive to pH changes,12a light,12b,c catalysis
(both chemical12d and enzymatic12e-g), and cation recognition.12h,i
The use of metal salts (e.g., AgBF4) has also been found to
template the formation of a gel from a simple bis(urea)
compound and the anion shown to influence the gel structure
by hydrogen-bonding interactions.13 Prior to this work, however,
there has only been one communication of using anions to
inhibit the formation of self-assembled molecular gels.14 This
is somewhat surprising, as many low-molecular mass organic
gelators (LMOG),8 that is, compounds that form self-assembled
gels, aggregate by the virtue of bis-urea groups or amide groups
often found in anion receptors. Among amide-based anion
receptors, the isophthalamides15 were the first acyclic anion
receptor to gain popularity in supramolecular chemistry and are
now one of the most heavily used in this area of research.7a,16
To date, however, there have been no reports of a ditopic version
of this receptor, or to the best of our knowledge, no reports on
self-assembled gels based on this class of molecules.
We report herein the synthesis of N,N′,N′′,N′′′-1,2,4,5-tetra-
(ethylhexanoate) pyromellitamide 1 (Chart 1). Remarkably,
although the parent pyromellitamide 2 has been known since
(8) For reviews, see: (a) Terech, P.; Weiss, R. G. Chem. ReV. 1997, 97, 3133.
(b) Abdallah, D. J.; Weiss, R. G. AdV. Mater. 2000, 12, 1237. (c) van Esch,
J. H.; Feringa, B. L. Angew. Chem., Int. Ed. 2000, 39, 2263. (d) Estroff, L.
A.; Hamilton, A. D. Chem. ReV. 2004, 104, 1201. (e) Hirst, A. R.; Smith,
D. K. Chem.-Eur. J. 2005, 11, 5496. (f) de Loos, M.; Feringa, B. L.; van
Esch, J. H. Eur. J. Org. Chem. 2005, 3615. (g) George, M.; Weiss, R. G.
Acc. Chem. Res. 2006, 39, 489.
1914,17 there are not many reports of pyromellitamides in the
literature18 and no reports regarding their supramolecular
properties. The pyromellitamides can be viewed structurally as
the ditopic version of the isophthalamides. The pyromellitamide
1 reported here displays aggregation behavior, including gel
formation, and binds to anions with a 1:2 host/anion stoichi-
ometry and negative cooperativity. The ability of pyromellita-
mide 1 to form self-assembled gels is not unexpected as it fits
well within the “rigid core/flexible tail” motif,8,19 commonly
found among other LMOG, being composed of a rigid aromatic
core with four amides linked to four flexible ethyl hexanoate
tails. The gels formed by 1 respond to the addition of anions,
by collapsing back to the solution state (anion-induced gel f
sol transition).
(9) (a) Xing, B. G.; Yu, C. W.; Chow, K. H.; Ho, P. L.; Fu, D. G.; Xu, B. J.
Am. Chem. Soc. 2002, 124, 14846. (b) Motulsky, A.; Lafleur, M.; Couffin-
Hoarau, A.-C.; Hoarau, D.; Boury, F.; Beonit, J.-P.; Leroux, J.-C.
Biomaterials 2005, 26, 6242-6253.
(10) Rajangam, K.; Behanna, H. A.; Hui, M. J.; Han, X.; Hulvat, J. F.; Lomasney,
J. W.; Stupp, S. I. Nano Lett. 2006, 6, 2086.
(11) Silva, G. A.; Czeisler, C.; Niece, K. L.; Beniash, E.; Harrington, D. A.;
Kessler, J. A.; Stupp, S. I. Science 2004, 303, 1352.
(12) For some representative examples, see: (a) Pozzo, J.-L.; Clavier, G. M.;
Desvergne, J.-P. J. Mater. Chem. 1998, 8, 2575. (b) Murata, K.; Aoki, M.;
Nishi, T.; Ikeda, A.; Shinkai, S. J. Chem. Soc., Chem. Commun. 1991,
1715. (c) Murata, K.; Aoki, M.; Suzuki, T.; Harada, T.; Kawabata, H.;
Komori, T.; Ohseto, F.; Ueda, K.; Shinkai, S. J. Am. Chem. Soc. 1994,
116, 6664. (d) Love, C. S.; Chechik, V.; Smith, D. K.; Ashworth, I.;
Brennan, C. Chem. Commun. 2005, 5647. (e) van Bommel, K. J. C.; Stuart,
M. C. A.; Feringa, B. L.; van Esch, J. Org. Biomol. Chem. 2005, 3, 2917.
(f) Toledano, S.; Williams, R. J.; Jayawarna, V.; Ulijn, R. V. J. Am. Chem.
Soc. 2006, 128, 1070. (g) Yang, Z.; Ho, P.-L.; Liang, G.; Chow, K. H.;
Wang, Q.; Cao, Y.; Guo, Z.; Xu, B. J. Am. Chem. Soc. 2007, 129, 266. (h)
Engelkamp, H.; Middelbeek, S.; Nolte, R. J. M. Science 1999, 284, 785.
(i) Hanabusa, K.; Maesaka, Y.; Suzuki, M.; Kimura, M.; Shirai, H. Chem.
Lett. 2000, 1168.
Results and Discussion
(13) Applegarth, L.; Clark, N.; Richardson, A. C.; Parker, A. D M.; Radosav-
ljevic-Evans, I.; Goeta, A. E.; Howard, J. A. K.; Steed, J. W. Chem.
Commun. 2005, 5423.
Synthesis and Solid-State Structure of Pyromellitamide
1. The pyromellitamide 1 was synthesized in two steps by the
(14) Stanley, C. E.; Clarke, N.; Anderson, K. M.; Elder, J. A.; Lenthall, J. T.;
Steed, J. W. Chem. Commun. 2006, 3199.
(15) Gale, P. A. Acc. Chem. Res. 2006, 39, 465.
(17) Meyer, H.; Steiner, K. Monatsh. Chem. 1914, 35, 391.
(16) See, for example: (a) Kavallieratos, K.; de Gala, S. R.; Austin, D. J.;
Crabtree, R. H. J. Am. Chem. Soc. 1997, 119, 2325. (b) Wisner, J. A.;
Beer, P. D.; Drew, M. G. B. Angew. Chem., Int. Ed. 2001, 40, 3606. (c)
Wisner, J. A.; Beer, P. D.; Berry, N. G.; Tomapatanaget, B. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 4983. (d) Wisner, J. A.; Beer, P. D.; Drew, M.
G. B.; Sambrook, M. R. J. Am. Chem. Soc. 2002, 124, 12469. (e) Curiel,
D.; Beer, P. D.; Paul, R. L.; Cowley, A.; Sambrook, M. R.; Szemes, F.
Chem. Commun. 2004, 1162. (f) Sambrook, M. R.; Beer, P. D.; Wisner, J.
A.; Paul, R. L.; Cowley, A. R.; Szemes, F.; Drew, M. G. B. J. Am. Chem.
Soc. 2005, 127, 2292. (g) Sambrook, M. R.; Beer, P. D.; Lankshear, M.
D.; Ludlow, R. F.; Wisner, J. A. Org. Biomol. Chem. 2006, 4, 1529. (h)
Chmielewski, M. J.; Jurczak, J. Chem.-Eur. J. 2006, 12, 7652.
(18) (a) Lawton, E. A.; McRitchie, D. D. J. Org. Chem. 1959, 24, 26. (b) Imai,
Y. Polym. Lett. 1970, 8, 555. (c) Saini, A. K.; Carlin, C. M.; Patterson, H.
H. J. Polym. Sci., Part A: Polym. Chem. 1992, 30, 419. (d) Kaupp, G.;
Schmeyers, J.; Boy, J. Tetrahedron 2000, 56, 6899. (e) Watson, K. G.;
Cameron, R.; Fenton, R. J.; Gower, D.; Hamilton, S.; Jin, B.; Krippner, G.
Y.; Luttick, A.; McConnell, D.; MacDonald, S. J. F.; Mason, A. M.;
Nguyen, V.; Tucker, S. P.; Wu, W.-Y. Bioorg. Med. Chem. Lett. 2004, 14,
1589.
(19) (a) van Bommel, K. J. C.; van der Pol, C.; Muizebelt, I.; Friggeri, A.;
Heeres, A.; Meetsma, A.; Feringa, B. L.; van Esch, J. H. Angew. Chem.,
Int. Ed. 2004, 43, 1663. (b) de Loos, M.; Friggeri, A.; van Esch, J. H.;
Kellogg, R. M.; Feringa, B. L. Org. Biomol. Chem. 2005, 3, 1631.
9
7156 J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007

Pyromellitamide Aggregate Response to Anion Stimuli
A R T I C L E S
Figure 1. (a) X-ray crystal structure of 1 showing in blue (direct) and red (indirect, by water molecule) the intermolecular hydrogen-bonding network that
results in one-dimensional offset stacks (hydrogen atoms have been removed for clarity). (b) Side and (c) top views of the two possible anion-binding sites,
each occupied by a water molecule within the crystal structure.
Table 1. Solvents Forming a Gel with Pyromellitamide 1
solventa
min % (w/w)
Tg (°C)
hexane
1.3
1.0
9.7
45
42
N/A
46
cyclohexane
diethyl ether
toluene
10.0
a Solvents tested that did not form a gel included water, acetonitrile,
dimethyl sulfoxide, chloroform, dichloromethane, methanol, ethanol, and
1-butanol.
aminolysis of N,N′-hydroxyethyl pyromellitimide 3,20 giving the
pyromellitamide tetraethanol 4 that was then coupled to hex-
anoyl chloride to give 1 in an overall 55% yield in two steps
(Scheme 1).
An X-ray crystal structure of pyromellitamide 1 was obtained
from a crystal grown by slow evaporation of a 1-butanol solution
of 1 (see Supporting Information for details). The crystal
structure (Figure 1) shows that each molecule is in contact with
six water molecules. The pyromellitamide molecules are aligned
in one-dimensional columnar stacks. Two water molecules
occupy sites similar to the anion binding site in Crabtree’s
isophthalamide-based anion receptor.7a,16a The other four water
molecules form hydrogen-bonding bridges between adjacent
columnar stacks in the crystal (Figure 1a).
Study of Gelation and Aggregation in Solution. The
formation of gels from LMOG is thought to proceed from the
self-assembly of columnar stacks, helical ribbons, and similar
one-dimensional aggregates which then form a dense three-
dimensional network encapsulating the solvent and forming a
gel.8 The crystal structure of pyromellitamide 1 shows one-
dimensional columnar stacks, indicating it is a good candidate
for the formation of gels in the appropriate solvent. Gel
formation was studied by dissolving pyromellitamide 1 in
various solvents by heating and sonication, followed by slow
cooling (Table 1).
Figure 2. (a) XRD pattern obtained of a gel from pyromellitamide 1 in
hexane (black solid line) compared to the calculated21 XRD pattern (red
broken line) from the single-crystal X-ray structure in Figure 1. (b)
Enlargement (× 50) of the gel pattern in the 2θ range of 10-30°. The gel
was loaded directly onto the sample holder, and scans were carried out
immediately.
2θ ≈ 6° corresponds to a d spacing of ca. 14-16 Å; however,
this does not match with any of the reflections calculated from
the solid-state structure. A very small peak at 2θ ) 10.54° aligns
well with the (110) reflection and the smaller (010) reflections
calculated in the solid-state structure at 2θ ) 10.62° and 10.12°,
respectively. The d spacing values for the hexane gel from 1
obtained by XRD are consistent with one-dimensional stacks
with similar dimensions as in the solid-state structure (Figure
1) but where the intercolumnar distance has been increased from
ca. 9 to 15 Å.
The microstructure of the gels formed was also studied using
field emission gun scanning electron (FEG-SEM), transmission
electron (TEM), and atomic force (AFM) microscopy. The TEM
micrograph of a dried diluted solution of pyromellitamide 1 in
hexane deposited on amorphous carbon shows twisted fibers
(Figure 3a and Figures S1-S3 in Supporting Information).
These twisted fibers may be related to helical ribbons that
have previously been reported as possible precursor structures
to the formation of a gel from LMOG.8d,f,22 The AFM (see
Freshly prepared samples of a hexane gel from 1 were
analyzed using XRD. The XRD pattern (Figure 2) shows two
closely spaced intense peaks at 2θ ) 4.30° and 4.72°, corre-
sponding to a d spacing of 20.5 and 18.7 Å, respectively. These
peaks match well with the calculated21 (001) reflection from
the solid-state structure of 1. A second set of three peaks around
(21) Calculated using Mercury, Crystal Structure Visualisation and Exploration
software (http://www.ccdc.cam.ac.uk/free_services/mercury/), see: Macrae,
C. F.; Edgington, P. R.; McCabe, P.; Pidcock, E.; Shields, G. P.; Taylor,
R.; Towler, M.; van de Streek, J. J. Appl. Crystallogr. 2006, 39, 453.
(22) D´ıaz, N.; Simon, F.-X.; Schmutz, M.; Rawiso, M.; Decher, G.; Jestin, J.;
Me´sini, P. J. Angew. Chem., Int. Ed. 2005, 44, 3260.
(20) This compound has been previously reported: Cowan, J. A.; Sanders, J.
K. M. J. Chem. Soc., Perkin Trans. 1 1985, 2435. See Supporting
Information for experimental details for the synthesis of 3.
9
J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007 7157

A R T I C L E S
Webb et al.
in intermolecular hydrogen bond driven aggregation23 of 1 in
the order solid state > cyclohexane > chloroform > (CH3)2-
SO.
The hydrogen bond driven intermolecular aggregation of
pyromellitamide 1 was also evident from a comparison of the
1H NMR spectra of 1 in various solvents (Figure 4a-c). Here,
a similar trend appears as from the FT-IR solution data; in
CDCl3 the “core” amide (δN-H) and aromatic (δAr-H)
proton resonances are very broad compared to the same
peaks in d6-acetone, which in turn are broader than those
observed in d6-(CH3)2SO. It is also noteworthy that, in both
CDCl3 and d6-acetone, the proton resonances corresponding
to the hexanoyl side chain are significantly sharper than the
δN-H and δAr-H, in line with the rigid core/flexible tail
model mentioned above. In d12-cyclohexane, even at elevated
temperatures, the spectrum (not shown) was too broad to resolve
any peaks, indicating that the degree of aggregation for 1 is
greater than in any of the other solvents shown in Figure 4.
Given this and the broadness of the CDCl3 spectrum, we decided
to focus on d6-acetone to study the aggregation of 1 in more
details. When plotted against increasing concentration of 1, the
1H NMR data in d6-acetone (Figure 4e) show significant
downfield and upfield shifts, for the δN-H and δAr-H,
respectively.
The δN-H downfield shift in Figure 4e is consistent with
an increase in intermolecular hydrogen bonding as the concen-
tration of 1 increases. The upfield shift for δAr-H suggests
co-facial stacking of 1 in the aggregates formed, possibly similar
to that observed in the single-crystal X-ray structure (Figure
1). We fitted the data to three different aggregation models
(Figure 5): a simple 1:1 dimerization (Dim)24,25 and both the
simple noncooperative Equal K model (EK) and its cooperative
version (coEK) described by Martin.25
Figure 3. Micrographs of pyromellitamide 1 gels. (a) TEM image of a
dried dilute hexane solution of 1 deposited on a carbon-coated Cu
grid. (b) FEG-SEM of 1 cyclohexane gel, dried on a carbon tape, platinum-
coated.
Figure S4 in Supporting Information) and FEG-SEM (Figure
3b) micrographs from the dried gel show various twisted fibers
and bundles of fibers as is commonly reported for LMOG
gels.8,22
The best fit for the aggregation of 1 in d6-acetone was
obtained with the coEK model, giving KE ) 232 M-1 and F )
0.22 (corresponding to K2 ) 51 M-1). The goodness of fit as
judged by comparing the chi squared (ø2) and the covariance
of the fit (coVfit)26 for the coEK model was an order of
magnitude better than either the noncooperative EK model or
the simple dimerization. According to the coEK model,25 when
F < 1, as observed here, it indicates that the formation of the
dimer 1/1 is more difficult than subsequent addition of 1 to a
growing stack (i.e., the aggregation of 1 is positively coopera-
tive). This could be simply due to favorable entropy for
aggregation as a dimer formation will lead to some loss of
orientational freedom while addition of another molecule of 1
to the growing aggregate involves only the loss of entropy of
one monomer.25,27 In our case, the well-known cooperative
hydrogen-bonding effect,28 where neighboring protic groups
increase the strength of hydrogen bonds, undoubtedly plays a
role as well.
The pyromellitamide 1 gels only nonpolar, aprotic solvents,
forming organogels and not hydrogels.8 This observation and
XRD data above support the hypothesis that it is the same or
similar hydrogen-bonding network observed in the crystal
structure for 1 (Figure 1) that generates the supramolecular
stacks responsible for the gel structure.
Further evidence for the key role that intermolecular hydrogen-
bonding plays in the formation of gels from pyromellitamide 1
comes from FT-IR spectroscopy studies (Table 2).
The FT-IR spectral data (entry 2) for the amide groups for
the cyclohexane solution of 1 (at a concentration well below
the gelation point) is almost identical to the solid-state spectrum
(entry 1), indicating similar hydrogen-bonding interactions in
cyclohexane. In contrast, the IR spectrum of 1 in chloroform
(entry 3) shows two peaks for the N-H stretch, one similar to
the solid state and cyclohexane spectra at 3261 cm-1, but the
other peak is blue-shifted to 3431 cm-1, characteristic of non-
hydrogen-bonded amide groups. In (CH3)2SO, only two peaks
above 3400 cm-1 (entry 4) are observed, indicating that there
are no intermolecular hydrogen-bonding interactions in (CH3)2-
SO. The data for the amide I (CdO stretching) and amide II
(N-H in-plane bending coupled to a C-N stretch) peaks show
that the former is blue-shifted in (CH3)2SO and slightly red-
shifted in chloroform, while the latter is red-shifted in both
(CH3)2SO and chloroform compared to the solid state and
cyclohexane solution spectra. This is consistent with a decrease
(23) Miyazawa, T.; Shimanouchi, T.; Mizushima, S.-I. J. Chem. Phys. 1956,
24, 408.
(24) Connors, K. A. Binding Constants; Wiley & Sons: New York, 1987.
(25) Martin, R. B. Chem. ReV. 1996, 96, 3043.
(26) The covariance of the fit (coVfit) is calculated by dividing the covariance
of the residual (raw - calculated data) with the covariance of the raw data.
This value is independent of the number of variables but reflects the
distributions of the residuals; the lower the coVfit value, the better the fit.
(27) Sarolea-Mathot, L. Trans. Faraday Soc. 1953, 49, 8.
(28) (a) Allen, F. H.; Kennard, O.; Taylor, R. Acc. Chem. Res. 1983, 16, 146.
(b) Guo, H.; Karplus, M. J. Phys. Chem. 1994, 98, 7104. (c) Schneider,
H.-J.; Yatsimirsky, A. Principles and Methods in Supramolecular Chem-
istry; Wiley & Sons: Chichester, 2000.
9
7158 J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007

Pyromellitamide Aggregate Response to Anion Stimuli
A R T I C L E S
Table 2. Selected Data for the Amide Peaks from FT-IR Spectra of Pyromellitamide 1 (Entries 1-4) and Its Corresponding TBA-Br
Complexes (Entries 5-7)
1
1
1
entry
sample
νN
−
H (cm-
)
amide I (cm-
)
amide II (cm-
)
1
2
3
4
5
6
1 (thin film)
1 (1 mM in cyclohexane)
1 (1 mM in CHCl3)
1 (1 mM in (CH3)2SO)
1 + 10 equiv of TBA-Br (thin film)
1 + 10 equiv of TBA-Br
(1 mM in CHCl3)
3242
3253
1635
1635
1622
1667
1662
1658
1566
1558
3431, 3261a
3440, 3496a
3211
1533
∼1540b
1537
3392, 3211a
1544
7
1 + 1 equiv of TBA-Br
(1 mM in cyclohexane)
3443, 3315, 3156c
1684, 1635a
1533
a These peaks have similar intensities. b This peak is extremely weak, broad (1510-1570 cm-1), and partially obscured by a large solvent peak. c These
peaks are very broad and overlap each other.
Figure 5. (a) Successive equilibria for self-association of molecule A and
the three models (b-d) used here fit the data for the aggregation of 1. The
factor F in (d) describes the cooperativity of aggregation with F < 1 positive
cooperativity for aggregation vs dimerization, F ) 1 no cooperativity (same
as EK-model in c), and F > 1 negative cooperativity for aggregation vs
dimerization. See ref 25 and text for details.
Figure 4. 1H NMR spectra at 300 K of 1 in (a) CDCl3 (200 MHz), (b)
Figure 6. 1H NMR spectra (400 MHz, d6-acetone) from titration experi-
d6-acetone (400 MHz), (c) d6-(CH3)2SO (400 MHz), and (d) CDCl3 with
ments with (a) TBA-Cl and (b) TBA-Br added to pyromellitamide 1.
2.0 equiv of TBA-Cl added (200 MHz). In all these spectra, the residual
The resonances shown are the δN-H (amide, right) and δAr-H (aromatic
water, acetone, (CH3)2SO, and CHCl3 resonances are indicated with an
proton, left). (c) A more detailed view of the δAr-H resonances upon the
1
asterisk (*). (e) H NMR (400 MHz, d6-acetone) at 300 K of the δN-H
addition of TBA-Cl to 1. The numbers of equivalents added are indicated
(blue open squares) and δAr-H (red solid circles) of pyromellitamide 1 at
different concentrations of 1. Also shown are the calculated aggregation
on the left (a, b) or right (c) of the spectra.
isotherms for the cooperative (coEK, solid line) and noncooperative (EK,
To explore the anion binding and the interaction between these
sites in 1, we titrated it with tetrabutylammonium (TBA+) salts
of chloride (Cl-), bromide (Br-), iodide (I-), nitrate (NO3-),
acetate (OAc-), and hexafluorophosphate (PF6-) and monitored
the binding by 1H NMR in d6-acetone. Titration of 1 with TBA-
Cl causes the aromatic and amide host resonances to shift
downfield (Figure 6a). Titration with TBA-Br causes similar
but less pronounced shifts, indicating weaker binding (Figure
6b). For TBA-I and TBA-NO3, these shifts are smaller still,
while TBA-OAc causes even larger shifts than TBA-Cl. No
broken lines) Equal K aggregation models, the latter also equal to the simple
dimerization (Dim) model. The insert in the top left corner of (e) shows an
enlargement of the region between 0 and 2 mM for the δN-H. See text
for details.
Anion Binding. According to earlier work, isophthalamide
receptors bind anions between the two amide N-H’s.7a,16a In
the X-ray structure of pyromellitamide 1, these potential binding
sites are co-occupied by a carbonyl of an adjacent 1 and a water
molecule (Figure 1b,c). The two isophthalamide sites in 1 are
in a chair conformation with respect to each other (Figure 1b).
9
J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007 7159

A R T I C L E S
Webb et al.
The symmetry breaking for Cl- and the calculated (apparent)
1:2 association constants for the other anions in Table 3 suggest
that 1 displays negative cooperativity toward anions as the
interaction cooperativity parameter26,30 (R ) 4K2a/K1a) is always
less than 1. This negative cooperativity appears to be allosteric
in nature (i.e., due to reversible discrete conformational changes
in the host that take place in the first binding event).1,6 Using
the nomenclature of Crabtree,7a we find that the isophthalamide
binding site must adopt a (syn-syn) orientation to bind the first
anion (Figure 7). To bind a second anion, the amide groups of
the opposite binding site must adopt the same conformation,
resulting in an overall (syn-syn)-(syn-syn) conformation for
1 in which there would be some electronic repulsion between
amide-carbonyl oxygen lone pairs. If the empty binding site
was to adopt an (anti-anti) or an (anti-syn) conformation,
however, it cannot bind to the second anion but instead hydrogen
bonds are formed to the carbonyl of the (first) occupied anion-
binding site. These conformers would be more stable than the
(syn-syn) conformer required for anion binding, resulting in
the observed negative cooperativity for the second binding event.
Simple molecular modeling studies (Hyperchem31 - MM+
force field) on the three uncomplexed conformers in Figure 7
support this explanation: the di-destabilized (syn-syn)-(anti-
anti) conformer is 29 kJ mol-1 and the monostabilized (syn-
syn)-(syn-anti) is 9 kJ mol-1 higher in energy than the
distabilized (syn-syn)-(anti-anti) conformer. In comparison,
the energy penalty6a,b for the second anion binding to 1 is 2-12
kJ mol-1 on the basis of the data in Table 3.
Table 3. Apparent Association Constants Ka for the Complexation
of 1 toward Various Anions (as TBA Salts) Obtained from NMR
Titrations in d6-Acetone at 300 Ke
1:1 binding model
1:2 binding model
1
a
1
a
1 a
anion
K
a (M-
)
covfit
(×
103)b
K
1a (M-
)
K
2a (M-
)
Rc
covfit (×
103)b
Cl-
Br-
I-
71000
1100
92
130
1300
4.0
2.6
2.3
1.5
2.9
data could not be fittedd
14000
420
460
430
40
0.12
0.38
0.61
0.007
3.6
1.7
0.1
0.6
-
NO3
70
OAc-
120000
200
a Estimated errors: 50%. b See ref 26. c Interaction cooperativity pa-
rameter: see refs 24 and 30. d The data could not be fitted to a realistic 1:2
binding model (K1a and K2a . 107, coVfit ) 20 × 10-3). e Results for both
a 1:1 and 1:2 binding models are shown. The concentration of 1 is in each
case fixed at 2-3 mM. See text for details.
1
significant chemical shifts are observed in the H NMR of
pyromellitamide 1 after adding excess TBA-PF6, demonstrating
that only relatively small anions can bind to 1 and that the TBA
cation has no effect on 1. These results are consistent with earlier
anion binding studies on other isophthalamide receptors. The
binding of anions to 1 is not limited to d6-acetone; in deuterated
chloroform the broad resonances for 1 (due to aggregation) get
significantly sharper upon the addition of TBA-Cl (compare
parts a and d of Figure 4).
Detailed analysis of the binding data and the stoichiometry
of binding was complicated by the fact that 1 aggregates in
d6-acetone. Attempts to determine stoichiometry by the Job plot
method failed (see Supporting Information), as the changes in
δN-H and δAr-H at different concentrations of 1 (Figure 4e)
either add to or counteract the chemically induced shifts of
δN-H and δAr-H caused by additions of anions. This causes
the Job plot to show different outcomes for the δN-H and δAr-
H, making it impossible to use to determine stoichiometry. To
overcome the problem with the aggregation of 1, we performed
all the anion titrations at a constant concentration of 1 (as
described in Supporting Information). At the concentrations used
in our experiments (2-3 mM), we calculated that 65-75% of
1 is in the monomer state, on the basis of the data we obtained
from the coEK model (Figure 5). We tried to fit the data to
binding models that take aggregation of the host into account,29
but due to large numbers of independent parameters in the
binding equation, no sensible results could be obtained. We then
fitted the data to classical 1:1 and 1:2 binding models,24 ignoring
the effect of aggregation of 1 (Table 3).
Effect of Anions on Gel Formation and Stability. Addition
of as little as 0.25 equiv of TBA-Cl or TBA-Br to pyromel-
litamide 1 in cyclohexane, under the same conditions as in Table
1, was enough to inhibit the formation of a gel. Addition of
TBA-PF6 (1 equiv) did not affect the ability of 1 to form a
gel. Adding these salts to preformed 1 cyclohexane gels had
the same effect on their stability (i.e., TBA-Cl and TBA-Br
caused the gels to collapse whereas TBA-PF6 did not). The
complete gel disruption by TBA-Cl and TBA-Br depended
on the mixing efficiency; if the gel was sonicated briefly after
adding the salts it collapsed almost instantaneously, while it
took from a few tens of seconds to several minutes for the gel
to collapse when the salts were simply placed on top of the
gel. Other TBA-salts, including TBA-I, TBA-OAc, and
TBA-NO3, had a similar effect. It is noteworthy that the time
it takes for anions to collapse the gel correlates qualitatively
with the anion strength reported above (Table 3) in the order
Our results suggest that the binding data fit better to a 1:2
binding model. The only apparent exception to this in Table 3
is Cl-, which could not be fitted to a 1:2 model but that is
probably because the association constants are too high to be
determined from NMR data (aside from the complications
arising from aggregation of 1). The strongest evidence for 1:2
binding, however, comes from breaking of the symmetry of the
δAr-H (Figure 6c) upon addition of 1 equiv of Cl-. The
symmetry of the δAr-H is then gradually restored as more Cl-
is added, which is expected as 1 binds to the second anion.
This data also suggest that the relative strength of anion binding
-
Cl- > AcO- > Br- > NO3 > I-, with the TBA-Cl salt
causing the quickest collapse (within few tens of seconds) and
TBA-I the slowest (up to 1 h). The solubility of the anion in
the gelating solvent also plays a role. Adding an aqueous
solution of NaCl did not disrupt the gels, instead a phase
separation was observed just as in the case of pure water and 1
cyclohexane gels.
The ability of anions to inhibit and destabilize the gelation
of 1 in nonpolar solvent is in line with NMR studies in
deuterated chloroform and acetone (Figures 4a,d and 6). The
FT-IR data in Table 2 (entries 5-7) also show that anions
destabilize the hydrogen-bonding network that drives the
aggregation of 1. The data for the solid-state complex between
-
to 1 decreases in the order Cl- > OAc- > Br- > NO3 ≈ I-,
which is in good agreement with earlier work on the related
isophthalamides.7a,15,16
(30) Connors, K. A.; Paulson, A.; Toledo-Velasquez, D. T. J. Org. Chem. 1988,
53, 2023.
(31) Hyperchem 6.03; Hypercube, Inc.: Gainesville, FL, 2000.
(29) (a) Steiner, R. F. J. Theor. Biol. 1974, 45, 93. (b) Wolfer, G. K., Jr.; Neil,
J. L.; Rippon, W. B. J. Protein Chem. 1987, 6, 441.
9
7160 J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007

Pyromellitamide Aggregate Response to Anion Stimuli
A R T I C L E S
Figure 7. Three different conformers that pyromellitamide 1 can adopt after binding to a single anion. Only the least stable conformer on the right is
capable of binding another anion. See text for details.
4.65 (t, J ) 5.7 Hz, 4H), 7.57 (s, 2H), 8.29 (t, J ) 5.6 Hz, 4H);
13C NMR (d6-(CH3)2SO, 75.4 MHz) 42.0, 59.5, 127.0, 136.5, 167.1;
FTIR (KBr, cm-1) 3404 (s), 3242 (s), 3080 (m), 2927 (m), 2869 (m),
1627 (s), 1556 (s), 1315 (s), 1068 (s); m/z HRMS (ESI) calcd for
C18H26N4O8 ([M + H]+) 427.1829, found 427.1822. Anal. Calcd for
C18H26N4O8 (426.421): C 50.70, H 6.15, N 13.14. Found C 50.82, H
6.15, N 13.16.
1 and TBA-Br (compare entries 1 and 5 in Table 2) show a
red shift of both the N-H stretch and the amide II peaks and
blue shift for the amide I peaks. Similar trends are also observed
in chloroform and cyclohexane solutions, suggesting the desta-
bilizing effects of TBA-Br are the same in these solvents and
the solid state.
N,N′,N′′,N′′′-1,2,4,5-Tetra(ethylhexanoate) Pyromellitamide (1).
Compound 4 (0.35 g, 0.82 mmol) was added to acetonitrile (25 mL)
with pyridine (0.47 g, 5.9 mmol) and 4-dimethylaminopyridine
(0.080 g, 0.65 mmol). To this was added hexanoyl chloride (0.82 g,
6.1 mmol). After 1 h, all solid material had dissolved, and the solution
was then stirred overnight at room temperature. The reaction mixture
was then poured onto an ice-water slurry (30 g) and stirred for 1 h.
The product was then extracted into dichloromethane (50 mL), washed
with saturated sodium bicarbonate (2 × 50 mL) and water (50 mL),
dried over anhydrous magnesium sulfate, filtered, and evaporated to
dryness to yield 1 (0.56 g, 88%) as a pastelike material that hardened
Conclusions
Pyromellitamide 1 forms aggregates by a one-dimensional
intermolecular hydrogen-bonding network similar to that ob-
served in the single-crystal X-ray structure of 1. In nonpolar
solvents such as hexane, the aggregation is strong enough for 1
to form gels. In deuterated acetone, where 1 does not aggregate
as strongly, the NMR dilution data could be fitted to the
cooperative equal K (coEK) aggregation model with positive
cooperativity.
Pyromellitamide 1 also binds to a range of small anions with
similar selectivity as observed for other isophthalamides. The
host 1 has two potential anion-binding sites, and although the
data analysis is complicated by the aggregation of 1, we
conclude that 1 can indeed bind two anions, albeit with negative
cooperativity. The NMR and IR study of anion complexation
with 1 clearly indicates that these anions interfere and disturb
the hydrogen-bonding network that is responsible for the
aggregation of 1. This is not surprising as these anions bind to
the same amide groups in 1 that are responsible for its
aggregation. The ability of anions to act as a stimuli for the
dissociation of 1 was then demonstrated on gels from 1. The
stimuli-response time for the dissociation of cyclohexane gels
from 1 was qualitatively correlated to the measured binding
strength of 1 in deuterated acetone. This suggests that the anion
sensitivity of gels formed from 1 might find applications in anion
sensors and release systems for biomedical applications. To
realize these applications and explore the gelating and anion
properties of pyromellitamides, we are currently synthesizing
a range of these compounds, including compounds that could
form gels in water.
1
slightly after drying: mp 115 °C; H NMR (d6-(CH3)2SO, 400 MHz)
δ 0.85 (t, J ) 6.8 Hz, 12H) 1.25 (m, 16H), 1.53 (m, 8H), 2.30 (t, J )
7.5 Hz, 8H), 3.42 (m, 8H), 4.10 (t, J ) 5.9 Hz, 8H), 7.52 (s, 2H), 8.48
(t, J ) 5.9 Hz, 4H); 13C NMR (d6-(CH3)2SO, 50.3 MHz) 13.8, 21.8,
24.0, 30.7, 33.4, 40.8, 62.1, 127.1, 136.7, 167.2, 173.0; FTIR (thin
film, cm-1) 3242 (s), 3072 (s), 2956 (d), 2860 (d), 1741 (s), 1635 (s),
1566 (s), 1168 (s); m/z HRMS (ESI) calcd for C42H66N4O12
([M
+
Na]+) 841.4570, found 841.4559. Anal. Calcd for
C42H66N4O12 (818.468): C 61.59, H 8.12, N 6.84. Found C 61.57, H
8.25, N 6.94.
Gel Formation: The Inversion Test. In a typical experiment, a
weighed amount of 1 and the solvent (1.0 mL) were placed in a vial.
When applicable, 0.25-1.0 equiv of the TBA-anion salt was also
added to this vial. The vial was heated around the boiling point of the
solvent being tested until the mixture turned into a clear solution. The
solution was allowed to cool slowly to room temperature, and the vial
with a closed lid was inverted to determine whether the solution was
completely immobilized to achieve gelation.
Determination of Gelation Temperature (Tg). In a typical
experiment, the vial containing a gel was immersed inversely in a
water bath while the temperature was raised slowly. Tg was
determined as the temperature at which the gel collapses such that the
gel was no longer immobilized according to the inversion test described
above.
Experimental Section
Anion-Induced Gel Collapse. To a preformed gel of typically
1.5% (w/w) of 1 in cyclohexane, 1 equiv of the TBA-anion salt was
added to the surface and allowed to diffuse through the gel. The time
taken for the gel f sol transition was then measured by periodically
testing the gel by the inversion test. Due to the hygroscopic nature of
TBA-OAc, the entire system was contained under an atmosphere of
nitrogen.
General Remarks. Details of equipment and techniques used,
including binding and microscopy studies, and the synthesis of the
previously reported20 compound 3 are given in the Supporting Informa-
tion.
N,N′,N′′,N′′′-1,2,4,5-Tetra(hydroxyethyl) Pyromellitamide (4). Di-
imide 3 (0.10 g, 0.3 mmol) was suspended in ethanol (6 mL) at 60 °C,
and 2-aminoethan-1-ol (0.17 g, 2.8 mmol) added. The diimide 3
gradually dissolved, after which a fine white precipitate formed.
After 30 min, the reaction was filtered and washed with ethanol,
yielding 3 (0.08 g, 62%) as a fine white powder: mp 197-198 °C;
1H NMR (d6-(CH3)2SO, 200 MHz) δ 3.26 (m, 8H), 3.50 (m, 8H),
Acknowledgment. We acknowledge the financial support of
the Australian Research Council (ARC), including an APAI
scholarshiptoJ.E.A.W.fromanARCLinkageGrant(LP0455238)
to M.J.C. and P.T. and a Discovery Project Grant and an
9
J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007 7161

A R T I C L E S
Webb et al.
Australian Research Fellowship (DP0666325) to P.T. The
University of Sydney is thanked for providing a Sesqui
Postdoctoral Research Fellowship to P.T. We thank the staff
in the NANO Major National Research Facility at the
Electron Microscope Unit, The University of Sydney, particu-
larly Tony Romeo and Emine Korkmaz, for help with FEG-
SEM studies.
Supporting Information Available: Details of equipment and
techniques used, synthesis of 3, methodology for binding studies
and binding models used, Job plot, AFM images, NMR spectra
of 1, and details for single-crystal X-ray diffraction analysis of
1. This material is available free of charge via the Internet at
http://pubs.acs.org.
JA0713781
9
7162 J. AM. CHEM. SOC. VOL. 129, NO. 22, 2007