Self-assembly of ball-shaped molecular complexes in water

Article abstract of DOI:10.1021/jo025513y

We present a simple and versatile access to spheroidal molecular assemblies with pronounced stability in highly polar solvents. These complexes are composed of doubly and triply charged complementary building blocks based on ammonium or amidinium cations

Full text of DOI:10.1021/jo025513y

J . Org. Chem. 2002, 67, 3755-3763
3755
Self-Assem bly of Ba ll-Sh a p ed Molecu la r Com p lexes in Wa ter
Thomas Grawe,† Thomas Schrader,*,‡ Reza Zadmard,‡ and Arno Kraft*,§
Institut fu¨r Organische Chemie und Makromolekulare Chemie, Heinrich-Heine-Universita¨t Du¨sseldorf,
Universita¨tsstr. 1, D-40225 Du¨sseldorf, Germany, Fachbereich Chemie, Philipps-Universita¨t Marburg,
Hans-Meerwein-Str., D-35032 Marburg, Germany, and Department of Chemistry, Heriot-Watt University,
Riccarton, Edinburgh EH14 4AS, United Kingdom
schradet@mailer.uni-marburg.de
Received J anuary 9, 2002
We present a simple and versatile access to spheroidal molecular assemblies with pronounced
stability in highly polar solvents. These complexes are composed of doubly and triply charged
complementary building blocks based on ammonium or amidinium cations and phosphonate anions.
Their high thermodynamic stability is best explained by the formation of a cyclic array of alternating
positive and negative charges interconnected by a regular network of hydrogen bonds. Association
constants reach 106 M-1 in methanol and often surpass 103 M-1 in water. The broad range of binding
energies correlates well with the varying degree of preorganization of both complex partners. As a
byproduct of these investigations, new recognition motifs for histidine and arginine esters and the
unsubstituted guanidinium ion are proposed. The additional introduction of methyl groups in the
2-, 4-, and 6-positions of the central benzene ring in either cations or anions causes a marked drop
in the corresponding Ka values of 1 order of magnitude; the related rotational barriers were estimated
at 0.3-2.1 kcal/mol. Spontaneous formation of defined 2:1 complexes from three components has
also been observed.
In tr od u ction
coat.5 Other approaches made extensive use of hydro-
phobic interactions as, for example, cyclodextrin dimers.6
Particularly in nonpolar solvents, self-complementary
building blocks capable of forming multiple hydrogen
bonds with each other will self-assemble into closed
architectures.7 These molecular capsules can accom-
modate guest molecules of the right size and polarity. A
Self-organized molecular containers have been exten-
sively used by Nature in various forms. Proteasomes,
which are able to digest almost any protein molecule, are
well-defined barrel-shaped molecular assemblies that
form spontaneously from up to 40 components.1 Viruses
are another case of helical or spherical container super-
structures with a self-assembled protein coat that pro-
tects the viral nucleic acids.2 These natural examples
inspired many supramolecular chemists to design mo-
lecular building blocks that are able to self-assemble into
larger capsule-like ensembles. The simplest access to self-
organized capsules is the well-known preparation of
vesicles from phospholipids. During the past decade,
emulsion and interface polymerization have both been
used to construct various types of concave and porous
polymeric analogues.3 Recent advances include the self-
organization of amphiphilic block copolymers with sub-
sequent cross-linking of their coat and destruction of the
nucleus inside.4 The geometrically well-defined coordina-
tion sphere of transition metal cations has also been
explored for the programmed design of multicomponent
aggregates, in some cases reminiscent of the viroidal
(5) Metal chelates: (a) Lehn, J . M. Angew. Chem., Int. Ed. Engl.
1990, 29, 1304-1319. (b) Lee, J .; Schwabacher, A. W. J . Am. Chem.
Soc. 1994, 116, 8382-8383. (c) Klu¨fers, P.; Schuhmacher, J . Angew.
Chem., Int. Ed. Engl. 1994, 33, 1863-1865. (d) Saalfrank, R. W.;
Burak, R.; Reihs, S.; Lo¨w, N.; Hampel, F.; Stachel, H.-D.; Lentmaier,
J .; Peters, K.; Peters, E. M.; von Schnering, H. G. Angew. Chem., Int.
Ed. Engl. 1995, 34, 993-995. (e) Takeda, N.; Umemoto, K.; Yamaguchi,
K.; Fujita, M. Nature 1999, 398, 794-796. (f) Olenyuk, B.; Whiteford,
J . A.; Fechtenko¨tter, A.; Stang, P. J . Nature 1999, 398, 796-798. (g)
Olenyuk, B.; Levin, M. D.; Whiteford, J . A.; Shield, J . E.; Stang, P. J .
J . Am. Chem. Soc. 1999, 121, 10434-10435. (h) Leininger, S.; Olenyuk,
B.; Stang, P. J . Chem. Rev. 2000, 100, 853-908. (i) Umemoto, K.;
Yamaguchi, K.; Fujita, M. J . Am. Chem. Soc. 2000, 122, 7150-7151.
(j) Fujita, N.; Biradha, K.; Fujita, M.; Sakamoto, S.; Yamaguchi, K.
Angew. Chem., Int. Ed. 2001, 40, 1718-1721.
(6) Hydrophobic capsules: (a) Harada, A.; Takahashi, S. J . Chem.
Soc., Chem. Commun. 1988, 1352-1353. (b) Boulas, P.; Kutner, W.;
J ones, M. T.; Kadish, K. M. J . Phys. Chem. 1994, 98, 1282-1287. (c)
Zhao, S.; Luong, J . H. T. J . Chem. Soc., Chem. Commun. 1995, 663-
664. (d) Venema, F.; Rowan, A. E.; Nolte, R. J . M. J . Am. Chem. Soc.
1996, 118, 257-258.
(7) (a) For an excellent review, see: Conn, M. M.; Rebek, J . Chem.
Rev. 1997, 97, 1647-1668. Some more recent work on self-organized
molecular capsules based on hydrogen bonds: (b) Kikuchi, Y.; Tanaka,
Y.; Sutarto, S.; Kobayashi, K.; Toi, H.; Aoyama, Y. J . Am. Chem. Soc.
1992, 114, 10302-10306. (c) Mammen, M.; Simanek, E. E.; Whitesides,
G. M. J . Am. Chem. Soc. 1996, 118, 12614-12623. (d) Vreekamp, R.
H.; Verboom, W.; Reinhoudt, D. N. J . Org. Chem. 1996, 61, 4282-
4288. (e) Mogck, O.; Bo¨hmer, V.; Vogt, W. Tetrahedron 1996, 52, 8489-
8496. (f) Sessler, J . L.; Wang, R. J . Am. Chem. Soc. 1996, 118, 9808-
9809. (g) Heinz, T.; Rudkevich, D. M.; Rebek, J . Nature 1998, 394, 764-
767. (h) Mart´ın, T.; Obst, U.; Rebek, J . Science 1998, 281, 1842-1845.
(i) Prins, L. J .; Huskens, J .; de J ong, F.; Timmerman, P.; Reinhoudt,
D. N. Nature 1999, 398, 498-502. (j) Hof, F.; Nuckolls, C.; Craig, S.
L.; Mart´ın, T.; Rebek, J . J . Am. Chem. Soc. 2000, 122, 10991-10996.
(k) Rivera, J . M.; Craig, S. L.; Mart´ın, T.; Rebek, J . Angew. Chem.,
Int. Ed. 2000, 39, 2130-2132.
† University of Du¨sseldorf.
‡ University of Marburg. Phone: +49-6421-2825544.
§ Heriot-Watt University.
(1) (a) Voges, D.; Zwickl, P.; Baumeister, W. Annu. Rev. Biochem.
1999, 68, 1015-1068. (b) Kruger, E.; Kloetzel, P. M.; Enenkel, C.
Biochimie 2001, 83, 289-293.
(2) Klug, A. Angew. Chem., Int. Ed. Engl. 1983, 22, 565-582.
(3) (a) Overview: Donath, E.; Sukhorukov, G. B.; Mo¨hwald, H.
Nachr. Chem. Technol. Lab. 1999, 47, 400-404 and refs cited therein.
(b) Kong, X. Z.; Kan, C. Y.; Li, H. H.; Yu, D. Q.; Yuan, Q. Polym. Adv.
Technol. 1997, 8, 627-630.
(4) (a) Thurmond, K. B.; Kowalewski, T.; Wooley, K. L. J . Am. Chem.
Soc. 1997, 119, 6656-6665. (b) Donath, E.; Sukhorukov, G. B.; Caruso,
F.; Davies, S. A.; Mo¨hwald, H. Angew. Chem., Int. Ed. 1998, 37, 2202-
2205.
10.1021/jo025513y CCC: $22.00 © 2002 American Chemical Society
Published on Web 05/09/2002

3756 J . Org. Chem., Vol. 67, No. 11, 2002
Grawe et al.
F igu r e 2. Hosseini’s supramolecular networks from dicar-
boxylates and bridging bisamidinium derivatives.11
F igu r e 1. Comparison between a well-known self-organizing
molecular assembly composed of self-complementary building
blocks (e.g., Rebek’s tennisball)8 and our simple system based
on complementary moieties. D/A ) Hydrogen bond donor/
acceptor. Note that the system on the left is held together
merely by hydrogen bonds, whereas the system on the right
makes use of ion-pair-reinforced hydrogen bonds that provide
additional stability in polar organic media such as DMSO and
methanol.
prominent example is the supramolecular tennisball
created by Rebek and co-workers.8 However, most of
these model systems rely on weak, albeit directed,
hydrogen bonds, and they are hence restricted exclusively
to nonpolar solvents.9 We found recently that spheroidal
molecular assemblies in polar organic media can be
achieved with high stability provided that multiple
electrostatic interactions are generated between two
complementary building blocks.10 An alternating array
of positive and negative charges leads to multiple chelate
arrangements, assisted by strong ionic hydrogen bonds.
This design principle was originally recognized with com-
plexes between dications and dianions and later success-
fully transferred to trication-trianion combinations.
F igu r e 3. J ob plot for the complex between bisphosphonate
1 and histidine methyl ester 6. The mole fraction x of 1 is
defined as [1]/([1] + [6]). The total concentration was main-
tained at 2.3 mM, and the change in chemical shift of the
receptor’s methyl ester signal was determined for various
compositions.
parallel the two-dimensional supramolecular networks
that Hosseini et al. obtained with combinations of planar
bisamidinium salts and dicarboxylates in a “crystal
engineering” approach (Figure 2).11
During preliminary investigations, we simply mixed
equimolar solutions of xylylene bisphosphonates12 and
the respective diammonium salts. The first hint of strong
binding came from the observation that the 1:1 complexes
derived from the p-xylylene derivatives precipitated
quantitatively from DMSO in an analytically pure form,
whereas the m-xylylene derivatives instead gave soluble
Resu lts a n d Discu ssion
Com p lexes betw een Bisp h osp h on a tes a n d Or -
ga n ic r,ω-Dica tion s. At the outset of our investigation,
we noted that certain R,ω-bisphosphonates are attracted
by R,ω-diammonium cations. Molecular dynamics calcu-
lations reproducibly lead to highly symmetrical cyclic
arrangements with alternating negative and positive
charges interconnected by a network of hydrogen bonds.
These simple three-dimensional structures bear a certain
resemblance to Rebek’s tennisball (Figure 1). In both
cases, complementary building blocks are attracted to-
ward each other through hydrogen bonds, thereby result-
ing in spheroidal structures of similar overall symmetry.
However, bisphosphonates and diammonium moieties are
ionic species, and they are therefore expected to give
strong complexes even in highly polar organic media. In
this respect, the above-mentioned ionic assemblies also
1
species with sharp signals in the H, 13C, and 31P NMR
spectra. The formation of nonpolymeric complexes with
a 1:1 stoichiometry was evident from a J ob plot (Figure
3).13 The complex geometry postulated in Figure 4 is
further supported by strong upfield shifts (up to 0.8 ppm)
for all bridging diammonium alkyl groups as they experi-
ence the deshielding effect above the host’s benzene ring.
Another indication comes from the strong downfield shift
of the aromatic guest protons H-4 to H-6 in a complex
between 1 and 3. To avoid steric strain, the benzene ring
of the guest is expected to lower down just above one of
the host’s phosphonate esters.
(11) (a) Hosseini, M. W.; Ruppert, R.; Schaeffer, P.; De Cian, A.;
Kyritsakas, N.; Fischer, J . J . Chem. Soc., Chem. Commun. 1994, 2135-
2136. (b) Brand, G.; Hosseini, M. W.; Ruppert, R.; De Cian, A.; Fischer,
J .; Kyritsakas, N. New J . Chem. 1995, 19, 9-13. (c) Hosseini, M. W.;
Brand, G.; Schaeffer, P.; Ruppert, R.; De Cian, A.; Fischer, J .
Tetrahedron Lett. 1996, 37, 1405-1408. (d) Fe´lix, O.; Hosseini, M. W.;
De Cian, A.; Fischer, J . Angew. Chem., Int. Ed. Engl. 1997, 36, 102-
104. (e) Fe´lix, O.; Hosseini, M. W.; De Cian, A.; Fischer, J . Tetrahedron
Lett. 1997, 38, 1933-1936. (f) Russell, V. A.; Ward, M. D. Chem. Mater.
1996, 8, 1654-1666.
(12) Xylylene bisphosphonates were prepared according to published
procedures: Schrader, T. J . Org. Chem. 1998, 63, 264-272.
(13) (a) J ob, P. Ann. Chim. 1928, 9, 113-203. (b) Blanda, M. T.;
Horner, J . H.; Newcomb, M. J . Org. Chem. 1989, 54, 4626-4636.
(8) Molecular tennisballs: (a) Wyler, R.; de Mendoza, J .; Rebek, J .
Angew. Chem., Int. Ed. Engl. 1993, 32, 1699-1701. (b) Branda, N.;
Wyler, R.; Rebek, J . Science 1994, 263, 1267-1268.
(9) Recent advances with hydrogen bonds in polar solvents: (a)
Vysotsky, M. O.; Thondorf, I.; Bo¨hmer, V. Chem. Commun. 2001,
1890-1891. (b) Shivanyuk, A.; Rebek, J . Chem. Commun. 2001, 2374-
2375. (c) Atwood, J . L.; Barbour, L. J .; J erga, A. Chem. Commun. 2001,
2376-2377.
(10) (a) Grawe, T.; Schrader, T.; Gurrath, M.; Kraft, A.; Osterod, F.
Org. Lett. 2000, 2, 29-32. (b) Grawe, T.; Schrader, T.; Gurrath, M.;
Kraft, A.; Osterod, F. J . Phys. Org. Chem. 2000, 13, 670-673.

Self-Assembly of Ball-Shaped Molecular Complexes in Water
J . Org. Chem., Vol. 67, No. 11, 2002 3757
in water by an efficient combination of three or four ion
pairs even in the absence of hydrophobic interactions.
Trisphosphonate 8 is the logical extension of our above-
described concept of alternating charges. The compound
should be able to form a C3v-symmetrical complex with
a matching triammonium ligand. Molecular mechanics
calculations again predict a highly symmetrical spheroi-
dal complex geometry with almost linear hydrogen bonds
(Figure 5).20
This array is calculated to be far more stable than the
respective simple ion pairs.22 A Michaelis-Arbuzow
reaction between trimethyl phosphite and tris(bromo-
methyl)benzenes furnished trisphosphonates 8 and 9 in
high yields (Scheme 1). The phosphonates were subse-
quently monodealkylated with tetrabutylammonium hy-
droxide.
F igu r e 4. Energy-minimized structure of the complex be-
tween m-xylylene bisphosphonate 1 and m-xylylene diammo-
nium salt 3. Left: Lewis structure. Right: cylinder model
(Cerius2, MSI, Dreiding 2.21).
Ta ble 1. Associa tion Con sta n ts Ka for th e 1:1 Com p lexes
betw een r,ω-Dia m m on iu m Der iva tives a n d 1 fr om NMR
Titr a tion s in DMSO-d 6 a t 20 °Ca
The tricationic counterparts are displayed in Scheme
4. They were synthesized either as trisamidinium or as
triammonium derivatives. Heterocyclic trisamidines 10
and 11 were prepared by a one-step route from benzene-
1,3,5-tricarboxylic acid.23 Trisamidine 12 was obtained
in two steps from the corresponding ethyl amide.24
Cyclohexanetriamine 13 was prepared from cis,cis-1,3,5-
cyclohexanetricarboxylic acid by a Curtius rearrange-
ment.25 The mesitylene triamines 14 and 15 were acces-
sible via the corresponding trisphthalimides and sub-
sequent hydrazinolysis (Scheme 2).
diamine hydrochloride
Ka [M-1
]
1,2-diaminocyclohexane (2)
m-xylylenediamine (3)
1,3-diaminopropane (4)
N-methyl-1,3-diaminopropane
lysine methyl ester (5)
1.4 × 103
1.6 × 103
2.8 × 103
3.2 × 103
5.0 × 103
8.0 × 103
1.6 × 103
histidine methyl ester (6)
arginine methyl ester (7)
a
As a result of the strongly hygroscopic character of both
titration partners, the DMSO-d6 solutions contained ∼0.1% of
water. Errors in Ka are standard deviations from the nonlinear
regression; they were calculated at 10% e Ka e 50%.
Finally, tris(pyrazolyl)-substituted mesitylenes 16 and
17 were obtained from the same two 1,3,5-tris(bromo-
methyl)benzenes after triple nucleophilic substitution
with pyrazole (Scheme 3).26 C3-Symmetrical trispyrazole
compounds such as 16 and 17 in the form of the free base
have recently been used as metal-encapsulating ligands27
and as highly efficient ammonium receptors.28 Crystal
structures show that all pyrazole groups are oriented
toward one side of the central benzene ring with the basic
We carried out NMR titrations of bisphosphonate 1
with a broad range of diammonium salts in DMSO-d6 and
calculated the association constants by nonlinear regres-
sion.14 Table 1 shows that all association constants Ka
lie between 103 and 104 M-1, which incidentally is not
much higher than the comparable association constants
of 1 with simple monoamines.15 Obviously, bond distances
and angles in the bisphosphonate-bisammonium hydro-
gen bond network are far from ideal. This may stem from
the fact that the second ammonium group has to push
the first one out of its chelated arrangement between
both phosphonates.15 It is interesting to note that within
this series, binding constants increase with increasing
length of the alkyl chain between the cationic head and
tail group (2-5). Although the bisphosphonate distance
remains constant throughout the series, it may constitute
another example of the beneficial effect of guest flexibility
as opposed to loss of preorganization.16
The highest association constants are measured for two
basic amino acids. Histidine ester 6 is particularly
interesting because, during its NMR titration with 1, both
aromatic imidazolium-CH protons are shifted upfield by
1.1 and 1.2 ppm, respectively. The imidazolium ring is
apparently located just above the host’s aromatic ring
and, thus, involved in efficient π-cation as well as π-π
interactions.17
(17) A similar observation was made in macrocyclic tetraphos-
phonate hosts that bind histidine esters strongly in water: Grawe,
T.; Schrader, T.; Finocchiaro, P.; Consiglio, G.; Failla, S. Org. Lett.
2001, 3, 1597-1600.
(18) Metzger, A.; Lynch, V. M.; Anslyn, E. V. Angew. Chem., Int.
Ed. Engl. 1997, 109, 862-865.
(19) (a) Hosseini, M. W.; Lehn, J .-M.; Mertes, M. P. Helv. Chim. Acta
1983, 66, 2454-2466. (b) Hosseini, M. W.; Lehn, J .-M.; Maggiora, L.;
Mertes, K. B.; Mertes, M. P. J . Am. Chem. Soc. 1987, 109, 537-544.
(20) (a) A relatively vague self-organization of a molecular capsule
based on flexible ammonium carboxylate ion pairs has been de-
scribed: Lee, S. B.; Hong, J .-I. Tetrahedron Lett. 1996, 37, 8501-8504.
(b) Recently, a highly stable cage-like complex has been assembled from
tetracationic Zn-porphyrinates and tetraanionic calixarenes: Fiam-
mengo, R.; Timmerman, P.; de J ong, F.; Reinhoudt, D. N. Chem.
Commun. 2000, 2313-2314.
(21) (a) Connolly, M. L. Science 1983, 221, 709-713. (b) Connolly,
M. L. J . Appl. Crystallogr. 1983, 16, 548.
(22) For flat structures, which form simple ion pairs, see: Schneider,
H.-J .; Schiestel, T.; Zimmermann, P. J . Am. Chem. Soc. 1992, 114,
7698-7703.
(23) (a) Kraft, A.; Fro¨hlich, R. Chem. Commun. 1998, 1085-1086.
(b) Kraft, A.; Osterod, F. J . Chem. Soc., Perkin Trans. 1 1998, 1019-
1025. (c) Kraft, A.; Reichert, A. Tetrahedron 1999, 55, 3923-3930.
(24) (a) Kraft, A. J . Chem. Soc., Perkin Trans. 1 1999, 705-714.
Three crystal structures of an N,N′-diethyl-substituted benzamidine
and its complexes confirm the almost perpendicular torsion angle
between the amidine moiety and an adjacent benzene ring: (b) Peters,
L.; Fro¨hlich, R.; Boyd, A. S. F.; Kraft, A. J . Org. Chem. 2001, 66, 3291-
3298.
Com p lexes betw een Tr isp h osp h on a tes a n d Or -
ga n ic Tr ica tion s. Anslyn18 und Lehn19 have previously
demonstrated that it is possible to achieve strong binding
(14) (a) Schneider, H.-J .; Kramer, R.; Simova, S.; Schneider, U. J .
Am. Chem. Soc. 1988, 110, 6442-6448. (b) Wilcox, C. S. In Frontiers
in Supramolecular Chemistry and Photochemistry; Schneider, H.-J .,
Du¨rr, H., Eds.; VCH: Weinheim, Germany, 1991; pp 123-143. (c)
Macomber, R. S. J . Chem. Educ. 1992, 69, 375-378.
(15) (a) Schrader, T. Angew. Chem., Int. Ed. Engl. 1996, 35, 2649-
2651. (b) Schrader, T. J . Org. Chem. 1998, 63, 264-272.
(16) Eblinger, F.; Schneider, H.-J . Angew. Chem., Int. Ed. 1998, 37,
826-829.
(25) Bowen, T.; Planalp, R. P.; Brechbiel, M. W. Bioorg. Med. Chem.
Lett. 1996, 6, 807-810.
(26) Hartshorn, C. M.; Steel, P. J . Aust. J . Chem. 1995, 48, 1587-
1599.
(27) Hartshorn, C. M.; Steel, P. J . Angew. Chem., Int. Ed. Engl.
1996, 35, 2655-2657.
(28) Chin, J .; Walsdorff, C.; Stranix, B.; Oh, J .; Chung, H. J .; Park,
S.-M.; Kim, K. Angew. Chem., Int. Ed. 1999, 38, 2756-2759.

3758 J . Org. Chem., Vol. 67, No. 11, 2002
Grawe et al.
F igu r e 5. Energy-minimized structure of the complex between 1,3,5-trisphosphonate 8 and its analogous triammonium salt: (a)
Lewis structure; (b) front view of the corresponding CPK model; (c) calculated van der Waals surface shown as Connolly surface21
(dotted solvent-accessible area around the complex) with an internal cavity.
Sch em e 1. Syn th esis of Tr isp h osp h on a te Bu ild in g Block s
Sch em e 2. Syn th esis of Mesitylen e Tr ia m m on iu m Bu ild in g Block s
12. An ESI mass spectrum could be obtained from a 10-7
M solution in methanol. The molecular ion peak M + H+
(7% at m/z ) 775) was found in the positive mass range,
whereas in the negative mass range an M- ion peak (5%
at m/z ) 774) was observed. Apparently, the capsule can
pick up a proton with its anionic building block or lose a
proton from its cationic moiety without falling apart. We
note that ESI-MS could be a promising analytical tool
for identifying changes in either half-sphere of the ionic
capsules. This may, at a later stage, be of great value for
detecting capsules with included neutral or charged guest
molecules.
Sch em e 3. Syn th esis of Tr isp yr a zolylben zen e
Bu ild in g Block s
Chemically induced upfield shifts for CH-protons in
some cases surpassed 0.5 ppm in DMSO-d6, pointing to
strong interactions. The results of selected NMR titra-
tions with association constants Ka in DMSO-d6 are
shown in Table 2. All these entries indicate a slightly
increased affinity of the complementary counterparts
compared to our previously discussed dication-dianion
nitrogen atoms pointing into the interior of the mol-
ecule.26,27
The high number of possible half-spheres resulting in
assemblies of varying size and geometry demonstrates a
substantial gain in flexibility as opposed to the self-
complementary “one subunit-one complex” strategy. All
the complexes prepared from 8 and various tricationic
components produced sharp NMR signals as well as
perfect 1:1 stoichiometries.29 Chemical ionization mass
spectrometry produced a weak molecular ion peak for the
1:1 complex of 8 and trisamidine 12 (1% for the M +
(29) In principle, 2:2 or 3:3 complexes could also be formed. However,
at high contrentations, we observed another interesting effect, namely,
extreme line-broadening of both NMR signals of the phosphonate esters
in the trisphosphonate half-sphere 8 in its complex with trisamidine
12. This must be a concentration-dependent dynamic effect: contrary
to the preorganized amidinium groups, the phosphonate esters have
to be frozen in their rotation around their benzylic and ester bonds, if
a the proposed 1:1-complex geometry is reached, with an alternating
array of arms coming from the top and bottom parts. Thus, the selective
line-broadening of these two phosphonate signals constitutes a good
indicator for formation of the 1:1-complex.
+
NH4 ion at m/z ) 792), and no additional peaks were
observed for higher aggregates. Electrospray ionization
(ESI) mass spectrometry is a much milder technique,
which proved to be advantageous for detecting the
molecular ion peak of the same complex between 8 and

Self-Assembly of Ball-Shaped Molecular Complexes in Water
J . Org. Chem., Vol. 67, No. 11, 2002 3759
Sch em e 4 C3v-Sym m etr ica l Tr ia m m on iu m /Tr isa m id in iu m Ca tion s a n d Hexa za -18-cr ow n -6 (19) Used for
NMR Titr a tion s w ith Tr isp h osp h on a te 1a
a
All structures are depicted in the conformation necessary for complexation.
Ta ble 2. Associa tion Con sta n ts Ka for th e 1:1 Com p lexes betw een Selected Tr ica tion s a n d Tr isp h osp h on a te 8 fr om NMR
Titr a tion s in DMSO-d 6 a t 20 °Ca
trisamidine hydrochloride
Ka [M-1
]
other C3-symmetrical cations
Ka [M-1
]
trisimidazoline 10
tris(tetrahydropyrimidine) 11
9.5 × 103
guanidinium
1,4,7-triazacyclononane
3.6 × 104
2.1 × 104
(2:1)b
a
As a result of the strongly hygroscopic character of both titration partners, the DMSO-d6 solution contained ∼0.1% water. Errors in
b
Ka are standard deviations from the nonlinear regression; they were calculated at 10% e Ka e 50%. Strongly sigmoidal binding curve
could not be fitted with conventional 2:1 equations; a highly cooperative binding is assumed.
The guanidinium cation is C3-symmetrical, too. Al-
though it carries only a single positive charge, its binding
constant clearly surpasses that of trications 10 and 11.
We attribute this to additional π-cationic interactions
that become effective once the guanidinium moiety is
located directly on top of the central phenyl ring at a
calculated, almost ideal, distance of 3.2 Å. With the
smallest possible azacrown (1,4,7-triazacyclononane), a
strongly sigmoidal binding curve is obtained, which just
about reaches its saturation point after addition of a
second equiv of trisphosphonate 8. A 2:1 complex is
formed, as confirmed by a J ob plot, with a high degree
of cooperativity. We assume that the initial binding of a
trisphosphonate preorients the azacrown, which then
facilitates the docking of a second molecule of 8.
F igu r e 6. Proposed complex geometry between trisimidazo-
line 10 and trisphosphonate 8 according to molecular mechan-
ics calculations.
complexes, albeit the difference is not very pronounced.
Modeling experiments reveal that the two cyclic amidines
10 and 11 have to rotate the amidinium moieties out of
the plane of the central benzene ring prior to binding to
the trisphosphonate (Figure 6). The result is a substantial
loss of delocalization energy reflected in the moderate
binding constants. In addition, a notable upfield shift of
the aromatic sensor proton30 was observed as, obviously,
the twisted amidinium substituents in the complex no
longer withdraw electron density from the adjacent
benzene ring.31
With more flexible or better preorganized guest mol-
ecules, however, binding energies rise sharply and be-
come too high in DMSO to be determined accurately by
NMR techniques. We turned therefore to the more polar
solvent methanol, which normally reduces association
constants for purely electrostatic interactions by a factor
of about 100. As Table 3 demonstrates, Ka values in
methanol reach from 103 to 106 M-1 in the most favorable
cases. Hence, they are up to 5 orders of magnitude larger
than those obtained with their C2v-symmetrical analogues
(Scheme 4, Table 3).
(30) Complexes of 10 and 11 with monocarboxylates show the
contrary, namely, a distinct downfield shift of the aromatic sensor
protons. This is explained by the overall flattened complex geometry
and has been confirmed by crystal structures: Osterod, F.; Peters, L.;
Kraft, A.; Sano, T.; Morrison, J . J .; Feeder, N.; Holmes, A. B. J . Mater.
Chem. 2001, 11, 1625-1633. In these complexes, the bridging car-
boxylates come into close contact with the trisamidine’s benzene
protons, thus causing the above-mentioned downfield shift.
(31) The torsion angle from force-field calculations amounts to 27°
(11) and 31° (10), bringing both central benzene rings in complex 8‚
10 closer to each other: the distance between both central benzene
rings is ∼4.33 Å in the complex with 11, but only ∼4.21 Å in a complex
with 10. This correlates well with the higher Ka value for the
tetrahydropyrimidine. In both complexes, the hydrogen bonds are all
1.98 Å long.
We did not observe any chemically induced shifts for
1
the H NMR signals of the tetrabutylammonium coun-
terions. Nevertheless, counterions could still play an
important role in the early phase of the binding process.
It is conceivable, for example, that even before complex-
ation the chloride counterions of the trisammonium
moiety occupy those places that are later taken over by
the phosphonate groups. Hence, they would build a
network of alternating positive and negative charges with
a certain degree of productive preorientation.

3760 J . Org. Chem., Vol. 67, No. 11, 2002
Grawe et al.
Ta ble 3. Associa tion Con sta n ts Ka a n d F r ee Bin d in g En th a lp ies ∆G ) -RT ln Ka for 1:1 Com p lexes betw een
Sym m etr ica l Tr ica tion s a n d 8 Deter m in ed by NMR Titr a tion s in Meth a n ol-d 4 or D2O a t 20 °C
tricationa
Ka [M-1] in CD3OD
-∆G [kcal/mol]
Ka [M-1] in D2Oc
-∆G [kcal/mol]
tris(tetrahydropyrimidine) 10
trisimidazoline 11
tren 18
mesitylene triamine 14 (R ) Me)
mesitylene triamine 15 (R ) H)
cyclohexanetriamine 13
trispyrazole 16 (R ) Me)
trispyrazole 17 (R ) H)
trisamidine 12
1.8 × 103 ( 18%b
3.0 × 103 ( 16%b
4.8 × 104 ( 48%
8.0 × 104 ( 34%
1.4 × 105 ( 11%
9.9 × 105 ( 32%
9.8 × 105 ( 10%
1.0 × 106 ( 23%
1.1 × 106 ( 8%
4.4
4.7
6.4
6.7
7.0
8.1
8.0
8.1
8.2
3.8 × 103 ( 14%
2.6 × 103 ( 13%
4.0 × 103 ( 6%
1.1 × 103 ( 15%
4.9
4.6
4.9
4.1
1.0 × 103 ( 9%
4.1
a
b
Ammonium and amidinium chlorides (except for 13, which was used as a bromide). Errors are standard deviations. c pD = 7.0.
F igu r e 7. (a) NMR titration of 12 with 8 (from top to bottom: 0, 0.5, 1.0, and 2.0 equiv of 8 added). The two sets of signals for
the N-ethyl groups indicate that the amidine has a preference for an (E,Z)-configuration around the C-N partial double bond.24b
Tetrabutylammonium signals are marked by an asterisk. (b) NMR titration curve showing the change in the chemical shift ∆δ
) δobserved - δ0 for two 1H NMR signals of trisamidine 12 (concn ) 0.1 mM) on addition of trisphosphonate 8 in CD3OD. (c) Typical
NMR titration curve for mesitylene triamine 14 with 8 in D2O.
1
Obviously, the binding affinity of C3-symmetrical
complementary components for each other is highly
dependent on stereoelectronic effects. Whereas the fully
conjugated, planar trisimidazoline 10 and the related
tris(tetrahydropyrimidine) 11 have to be strongly twisted
for complexation by 8 (see above), tris(aminoethyl)amine
18 is much too flexible, and complexation rotations
around nine bonds have to be restricted. By contrast,
trisamidine 12 and cyclohexanetriamine 13 are more
favorably preoriented. The N,N′-diethyl-substituted ami-
dine 12 with its sterically demanding ethyl groups was
found to prefer the (E,Z)-configuration during complex-
ation with 8 according to 1H NMR spectra of the complex
(Figure 7a). In addition, all three amidinium groups of
12 are likely to be strongly twisted relative to the central
benzene ring judging from the H NMR chemical shift of
the aromatic proton (δH ∼ 8.0) being upfield compared
to that of 10 (δH ∼ 8.4) and 11 (δH ∼ 8.9). This is in
accordance with a previously reported crystal structure
in which the torsion angle between a single N,N′-diethyl-
substituted amidinium group and an adjacent phenyl
substituent is as high as 66-72°.24b Preorganization thus
increases the binding constant (Figure 7b,c).
The broad range of association constants that is
observed for complexes of 8 with C3-symmetrical tri-
cations therefore reflects the varying degree of pre-
organization of both complex partners, but there seems
to be no obvious correlation with the pKa values of the
protonated triamines/trisamidines.32 Both components of
the complex are flexible molecules possessing various

Self-Assembly of Ball-Shaped Molecular Complexes in Water
J . Org. Chem., Vol. 67, No. 11, 2002 3761
rotatable bonds that need to be conformationally fixed
in the final complex. In this respect, the required entropy
loss for the self-organization process lies between that
of Rebek’s tennisball (composed of rigid half-spheres) and
Whitesides’ rosette (made up of highly flexible compo-
nents).8,33 Several other groups have exploited the multi-
valency of polytopic ligands, often in a pseudo C3-
symmetrical arrangement.34 In water, however, solvation
effects may become more prominent than the relatively
low torsional entropy differences that are in the range
of 0.5-3 kcal/mol at 293 K for our aggregates.
We found that most of our complementary trication-
trianion complexes are surprisingly stable even in water.
Binding constants range from 103 to 104 M-1 for both
ammonium and amidinium ligands (Table 3). In aqueous
solution, the dominating influence of the electrostatic
interaction renders the hard ammonium cation superior
to the softer amidinium cation with its delocalized posi-
tive charge. Hence, the distribution of association con-
stants becomes much more uniform. Such high binding
energies constitute a strong argument for the spheroidal
structure of the complexes. In the majority of our C3-
symmetrical arrangements, each ammonium cation is
located in the center of two surrounding phosphonate
anions, and vice versa. This geometry is favorable
considering that the trication and the trianion act as
chelates for each other. An alternating order of molecular
arms coming from the top and bottom unit generates a
compact ball-like structure (Figure 5). Similar combina-
tions of weak multipoint interactions are, of course, also
used by Nature for selective and strong self-organization
processes.35
+3.5 kcal/mol and ∆S ) +12.6 cal K-1 mol-1. This means
that, contrary to self-organization processes in unpolar
solvents, the enthalpic gain and entropic cost of the
interaction between both components are completely
overridden by solvation effects. Similar to many other
ion-pairing processes in water, the small capsulelike
complexes suffer from decreased solvation compared to
their highly polar components.36 In fact, the aggregates
are often only poorly soluble in a polar solution. On the
other hand, the highly ordered solvation shell around
each building block is broken, and free water molecules
are released into the solution, leading to entropic gain.
In the future, care should be taken to ensure a good
solvation of the whole capsule in order to keep the
unfavorable positive enthalpy term as low as possible.
We wondered whether the introduction of three ethyl
(or methyl) substituents in the 2-, 4-, and 6-positions of
trisphosphonate 8 would perhaps increase binding due
to preorganization, following an idea that Anslyn and
others have previously demonstrated with cationic and
neutral receptor molecules.37 The introduction of three
methyl groups in the 2-, 4-, and 6-positions was thought
to have the same beneficial effect on triammonium
complexation. For practical reasons, we decided to use
the more easily accessible trimethyl instead of the
triethyl derivatives favored by Anslyn.18,37b,c,38 This was
not without precedence as several groups have used core
units of this type and reported crystal structures with
an alternating order of substituents in which all the
binding sites lie on the same side of the central benzene
ring.39 NMR titrations with various trications and the
readily available new “template”, 2,4,6-trimethyl-substi-
tuted trisphosphonate 9, proved again that binding in
methanol is quite strong and results in the formation of
specific 1:1 complexes (Table 4). However, as a direct
comparison shows, binding constants are consistently
lower than those obtained with the unsubstituted 8 by
almost an order of magnitude. We assume that in the
lowest energy state, not all phosphonate groups are
located on the same side of the central benzene ring of
9. If this is the case, any phosphonate arm, which has to
be rotated to the opposite side of the central benzene ring
for three-point binding, experiences an additional steric
strain when it passes a neighboring benzylic methyl
substituent, thus leading to a reduction in the overall
association energy. The same argument applies to the
related mesitylene triamines 14 and 15 (Table 3). From
the difference in binding energies, a rotational barrier
To gain more insight into the forces governing the self-
organization process we carried out microcalorimetric
measurements with the aggregate between trisphos-
phonate 8 and trisamidium salt 12 in water (Microcal
MCS-ITC, 20 °C). Aliquots of aqueous 8 were added via
microsyringe to a 3 mM aqueous solution of 12. Strong
complexation heats of more than 20 µcal/s were produced
even with our small ligands. The titration curve resulted
in a Ka value of 400 M-1, a little lower than that of the
NMR titration (103 M-1). The stoichiometry factor was
0.9, confirming again the 1:1 complex formation. Positive
enthalpy and entropy terms were calculated as ∆H )
(32) Potentiometric studies were conducted with an autotitrator and
a pH glass electrode with a 3 M KCl internal filling. Titrations were
carried out with 0.4905 M NaOH, and about 100 data points were
collected for each titration. Protonation constants (pKa) were obtained
by a nonlinear curve fitting of the titration curve or by calculating the
pH dependence of the degree of protonation: (a) Billo, E. J . Excel for
Chemists: A Comprehensive Guide, 2nd ed.; Wiley-VCH: New York,
2001; Chapter 22. (b) Kraft, A. J . Chem. Educ., submitted for
publication. In methanol, the different basicities of amines, imidazo-
lines, and amidines are obviously of minor importance. They are much
more pronounced in water, and pKa values could be determined in 0.1
M aqueous KCl at 25 °C for protonated 10 (12.0 ( 0.3, 11.0 ( 0.3, 9.9
( 0.2), 11 (9.9, 8.7, 7.4), 12 (>11), 13 (10.4, 9.0, 7.2), and 18 (10.3, 9.5,
8.4), in most cases with an error less than (0.1. The fact that some
triamines are not as basic as others and even exist predominantly as
diprotonated species at pH 7 (11, 13, 18) does not seem to affect their
association constant with 8 or 9. We deduce, therefore, that preorien-
tation plays a more important role than the pKa of the tricationic
component.
(36) Other groups have found similar ITC results for molecular
complexes based on salt bridges in polar solvents: (a) Haj-Zaroubi,
M.; Mitzel, N. W.; Schmidtchen, F. P. Angew. Chem., Int. Ed. 2002,
41, 104-107. (b) Schmidtchen, F. P. Org. Lett. 2002, 4, 431-434. (c)
Linton, B.; Hamilton, A. D. Tetrahedron 1999, 56, 6027-6038. (d) Sebo,
L.; Schweizer, B.; Diederich, F. Helv. Chim. Acta 2000, 83, 80-92. (e)
Enthalpy-entropy compensation often takes place in polar solvents:
Dunitz, J . D. Chem. Biol. 1995, 2, 709-712.
(37) Anion receptors: (a) Ref 18. (b) Metzger, A.; Anslyn, E. V.
Angew. Chem., Int. Ed. 1998, 37, 649-652. (c) Lavigne, J . J .; Anslyn,
E. V. Angew. Chem., Int. Ed. 1999, 38, 3666-3669. Ammonium
receptors: (d) Ref 28. (e) Ahn, K. H.; Kim, S.-G.; J ung, J .; Kim, K.-H.;
Kim, J .; Chin, J .; Kim, K. Chem. Lett. 2000, 170-171.
(38) As all attempts to chloromethylate 1,3,5-triethylbenzene failed,
we chose the commercially available 1,3,5-tris(chloromethyl)-2,4,6-
trimethylbenzene instead.
(33) (a) Seto, C. T.; Whitesides, G. M. J . Am. Chem. Soc. 1993, 115,
905-916. (b) Seto, C. T.; Whitesides, G. M. J . Am. Chem. Soc. 1993,
115, 1330-1340.
(34) (a) Prohens, R.; Toma`s, S.; Morey, J .; Deya`, P. M.; Ballester,
P.; Costa, A. Tetrahedron Lett. 1998, 39, 1063-1066. (b) Rao, J .; Lahiri,
J .; Weis, R. M.; Whitesides, G. M. J . Am. Chem. Soc. 2000, 122, 2698-
2710.
(35) Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem.,
Int. Ed. 1998, 37, 2754-2794.
(39) (a) Garratt, P. J .; Ibbett, A. J .; Ladbury, J . E.; O’Brien, R.;
Hursthouse, M. B.; Malik, K. M. A. Tetrahedron 1998, 54, 949-968.
(b) Sato, K.; Arai, S.; Yamagishi, T. Tetrahedron Lett. 1999, 40, 5219-
5222. (c) Oh, K. S.; Lee, C.-W.; Choi, H. S.; Lee, S. J .; Kim, K. S. Org.
Lett. 2000, 2, 2679-2681. (d) Yuan, Y.; Zhou, H.; J iang, Z.; Yan, J .;
Xie, R. Acta Crystallogr., Sect. C 2000, 56, e34-e35.

3762 J . Org. Chem., Vol. 67, No. 11, 2002
Grawe et al.
Ta ble 4. Associa tion Con sta n ts Ka a n d F r ee En th a lp ies
∆G ) -RT ln Ka for 1:1 Com p lexes betw een Sym m etr ica l
Tr ica tion s a n d 1,3,5-Tr im eth yl-Su bstitu ted
Tr isp h osp h on a te 9 Deter m in ed by NMR Titr a tion s in
Meth a n ol-d 4 a t 20 °C
sensors, markers, or reagents by such artificial container
molecules would, especially in a physiological solution,
open the door for a wide range of possible applications.7
Exp er im en ta l Section
Ka [M-1
]
-∆G
[kcal/
mol]
∆∆Gb
[kcal/
mol]
in
Gen er a l. DMSO-d6, deuterium oxide, and methanol-d4 were
all purchased in g99.8% purity. Thin-layer chromatography
(TLC) analyses were performed on silica gel 60 F254 with a 0.2
mm layer thickness. Preparative chromatography columns
were packed with silica gel 60. All solvents were dried and
freshly distilled before use. Compounds 1,15 10,23b 11,23c 12,24a
and 1325 were prepared according to literature procedures.
1,3,5-Tr is(d im eth oxyp h osp h or ylm eth yl)ben zen e. 1,3,5-
Tris(bromomethyl)benzene42 (1.00 g, 2.80 mmol) was dissolved
in trimethyl phosphite (5 mL) and refluxed for 3 h. Evaporation
of the solution gave a colorless oil, which was further purified
by column chromatography (9:1 CHCl3/MeOH, Rf ) 0.37) to
tricationa
CD3OD
tris(tetrahydropyrimidine) 11
tren 18
trisimidazoline 10
mesitylene triamine 14 (R ) Me)
mesitylene triamine 15 (R ) H)
cyclohexanetriamine 13
trisamidine 12
1.2 × 102
1.4 × 103
1.7 × 103
3.4 × 104
5.8 × 104
1.9 × 105
2.1 × 105
2.8
4.3
4.3
6.2
6.5
7.1
7.2
1.6
2.1
0.3
0.5
0.5
1.0
1.0
a
Ammonium and amidinium chlorides (except for 13, which was
used as a bromide). Difference between ∆G values for trisphos-
phonates 8 and 9. These energy differences are attributed mainly
to the rotational barrier for one phosphonate moiety as it passes
by the neighboring benzylic methyl substituent in 9.
b
1
furnish a colorless solid. Yield: 1.05 g (84%). Mp: 98 °C. H
NMR (200 MHz, CDCl3): δ 3.15 (d, J ) 22.1 Hz, 6 H), 3.68 (d,
J ) 10.7 Hz, 18 H), 7.14 (br q, J ) 2.4 Hz, 3 H). 13C NMR (75
MHz, CDCl3): δ 32.7 (d, 1J CP ) 138 Hz), 52.9 (m), 129.9, 132.9.
31P NMR (202 MHz, CDCl3): δ 29.6. MS (CI, NH3): m/z 462
(80%, M + NH4+), 445 (25, M + H+). Anal. Calcd for
15H27O9P3: C, 40.55; H, 6.13. Found: C, 40.30; H, 6.29.
1,3,5-Tr is(d im eth oxyp h osp h or ylm eth yl)-2,4,6-tr im eth -
ylben zen e. 1,3,5-Tris(bromomethyl)-2,4,6-trimethylbenzene
(1.00 g, 2.51 mmol) was dissolved in trimethyl phosphite (5
mL) and refluxed for 12 h. Evaporation of the solution gave a
colorless oil, which was further purified by column chroma-
tography (9:1 CHCl3/MeOH, Rf ) 0.33). Yield: 0.91 g (74%).
1H NMR (200 MHz, CDCl3): δ 2.44 (s, 9 H), 3.38 (d, J ) 23.5
Hz, 6 H), 3.64 (d, J ) 10.7 Hz, 18 H). 13C NMR (75 MHz,
CDCl3): δ 18.10, 25.61 (d, J CP ) 130.2 Hz), 52.50 (d, J CP ) 40
Hz), 130.58, 138.80. 31P NMR (202 MHz, CDCl3): δ 30.74. MS
(CI, NH3): m/z 504 (90%, M + NH4+), 487 (35, M + H+). Anal.
Calcd for C18H33O9P3: C, 44.45; H, 6.84. Found: C, 44.30; H,
6.99.
between 0.3 and 2.1 kcal/mol can be estimated in the
above-mentioned series of complexes.40
Molecular mechanics calculations of all examined
complexes in various environments (gas phase, chloro-
form, and water) reproducibly led to the same sym-
metrical complex structures. Detailed conformational
searches (Monte Carlo simulations, followed by molecular
dynamics calculations) invariably produced global minima
with all the flexible arms of each molecule pointing in
the same direction, very much like fingers of a claw or
like an open mouth. Both complex partners are locked
together similar to gear wheels and thus form a molec-
ular capsule. In some of the lower-energy structures, one
arm of a half-sphere was opened to the solvent. We
assume that opening and closing of arms occurs rapidly
and constantly. This, however, distorts only slightly the
compact ball-shaped overall structure of the assembly.
Calculations of the respective Connolly surfaces21 re-
vealed a tiny internal cavity. Any attempted inclusion
of even small diatomic guests predictably leads to a
widening and, thus, destabilization of the capsule; these
spheroidal assemblies are considered to be too small for
guest encapsulation.41
C
1,3,5-Tr is(h yd r oxym e t h oxyp h osp h or ylm e t h yl)b e n -
zen e, Tr is(t et r a b u t yla m m on iu m ) Sa lt (8). 1,3,5-Tris-
(dimethoxyphosphorylmethyl)benzene was treated with 3.0
equiv of aqueous [NBu4]OH and refluxed for 2 weeks. After
evaporation of the solvent, the crude product was extracted
with chloroform. The solution was dried over MgSO4, filtered,
and evaporated to dryness. The colorless amorphous solid was
1
then dried at 50 °C/1 mbar. Yield: 87%. H NMR (500 MHz,
CDCl3): δ 0.98 (t, J ) 7.6 Hz, 36 H), 1.41 (sextet, J ) 7.6 Hz,
24 H), 1.61 (m, 24 H), 2.90 (d, J ) 20.1 Hz, 6 H), 3.25 (m, 24
H), 3.54 (d, J ) 10.1 Hz, 9 H), 7.14 (br q, J ) 1.9 Hz, 3 H). 13
C
Con clu sion
1
NMR (125 MHz, CDCl3): δ 13.76, 19.72, 24.09, 34.87 (d, J CP
2
) 124.2 Hz), 51.68 (d, J CP ) 5.0 Hz), 58.30, 128.18, 136.49.
With symmetrical hexacations, it is possible to achieve
the spontaneous formation of higher aggregates. Even
in water, a 2:1 complex between two half-spheres of 8
and a belt of hexaza-18-crown-6 19 was formed as we
were able to prove by means of a J ob plot. The concept
for a new design of molecular capsules based on pre-
orientation and alternating charges can thus be trans-
ferred to systems with more than two components.
Experiments for the directed formation of multicompo-
nent assemblies along these lines are underway in our
laboratory. We are also devising larger, more rigid
capsules by the same design principles involving cavities
large enough to accommodate polar aliphatic and aro-
matic guest molecules in water. The transport of drugs,
31P NMR (505 MHz, CDCl3): δ 18.58. MS (FAB, glycerol
matrix, Xe): m/z 1126 (60%, M + H+), 242 (100, NBu4+). Anal.
Calcd for C60H126N3O9P3‚6H2O: C, 58.37; H, 11.27; N, 3.40.
Found: C, 58.45; H, 11.51; N, 3.67.
1,3,5-Tr is(h yd r oxym et h oxyp h osp h or ylm et h yl)-2,4,6-
tr im eth ylben zen e, Tr is(tetr a bu tyla m m on iu m ) Sa lt (9).
1,3,5-Tris(dimethoxyphosphorylmethyl)-2,4,6-trimethylben-
zene was treated with 3.0 equiv of 40% aqueous [NBu4]OH
and refluxed for 3 weeks while the reaction was followed by
occasional 31P NMR control. After evaporation of the solvent,
the crude product was extracted with chloroform. The solution
was dried over MgSO4, filtered, and evaporated to dryness.
The colorless amorphous solid was then dried at 50 °C/1 mbar.
Yield: 65%. 1H NMR (500 MHz, CDCl3): δ 0.99 (t, J ) 7.3
Hz, 36 H), 1.43 (sextet, J ) 7.6 Hz, 24 H), 1.63 (m, 24 H), 2.56
(s, 9 H), 3.13 (d, J ) 20.5 Hz, 6 H), 3.29 (m, 24 H), 3.57 (d, J
) 10.4 Hz, 9 H). 13C NMR (125 MHz, CDCl3): δ 13.76, 18.01,
(40) In comparing trisphosphonates 8 and 9, we assume that the
changes in solvation energies on complex formation with their guests
are the same for both hosts, since they differ only in three methyl
groups on the central phenyl ring.
(41) According to modeling geometries, the egg-shaped inner volume
of our capsules was calculated from Connolly surfaces at 0-8 Å3. Thus,
these containers are too tiny even for the smallest conceivable guest,
a noble gas atom.
1
2
19.72, 24.09, 34.87 (d, J CP ) 124.2 Hz), 51.68 (d, J CP ) 5.0
Hz), 58.30, 128.18, 136.49. 31P NMR (505 MHz, CDCl3): δ
19.88. MS (FAB, glycerol matrix, Xe): m/z 1168 (56%, M +
(42) Plater, M. J .; Praveen, M.; Stein, B. K.; Ballantine, J . A.
Tetrahedron Lett. 1996, 37, 7855-7856.

Self-Assembly of Ball-Shaped Molecular Complexes in Water
J . Org. Chem., Vol. 67, No. 11, 2002 3763
H+), 242 (100, NBu4+). Anal. Calcd for C63H132N3O9P3: C,
64.75; H, 11.37; N, 3.60. Found: C, 65.01; H, 11.47; N, 3.78.
Gen er a l P r oced u r e for th e P r ep a r a tion of 1,3,5-Tr is-
(p h th a lim id om eth yl)ben zen e Com p ou n d s. The 1,3,5-tris-
(bromomethyl)benzene compound, 3.6 equiv of potassium
phthalimide, and 0.3 equiv of 18-crown-6 were dissolved in
toluene under Ar. After the mixture was heated to 100 °C for
24 h, water was added to the reaction mixture. The aqueous
layer was extracted four times with CH2Cl2. The combined
organic layers were dried over Na2SO4 and evaporated, and
the residue was purified by column chromatography (20:1
CH2Cl2/acetone).
1,3,5-Tr is(p h th a lim id om eth yl)ben zen e. Yield: 71%. Rf
) 0.55 (20:1 CH2Cl2/acetone). 1H NMR (500 MHz, CD2Cl2/
CDCl3): δ 4.78 (s, 6 H), 7.35 (s, 3 H), 7.70-7.82 (AA′BB′, 12
H). 13C NMR (125 MHz, CD2Cl2/CDCl3): δ 40.23, 122.40,
126.68, 132.16, 133.07, 136.50, 166.95. MS (CI, NH3): m/z 573
(M + NH4+). Anal. Calcd for C33H29N3O6: C, 71.34; H, 3.81;
N, 7.56. Found: C, 71.46; H, 3.70; N, 7.58.
Micr oca lor im et r ic Mea su r em en t s. Microcalorimetric
measurements were carried out with the aggregate between
trisphosphonate 8 and trisamidium salt 12 in water (20 °C).
Aliquots of aqueous 8 were added via microsyringe to a 3 mM
aqueous solution of 12. Complexation heats were more than
20 µcal/s (Ka ) 400 M-1; the stoichiometry factor was 0.9; ∆H
) +15 kJ /mol, and ∆S ) +53 J K-1 mol-1). The evaluation
software Origin 5.0 was used. The following were the experi-
mental parameters: measuring temperature, 20 °C; room
temperature, 17-18 °C; cell volume, 1.414 mL; syringe volume,
250 µL; solvent, triply distilled water; first injection, 5 µL; all
other injections, 20 µL; injection period, 60 s; delay time
between two injections, 180 s; total number of injections, 11;
cell concentration of 12, 7.34 mg per 2 mL (concn ) 3.26 mM);
syringe concentration of 8, 32.8 mg per 1.5 mL (concn ) 45.4
mM).
Com p u ta tion a l Meth od s. Molecular mechanics calcula-
tions were performed using Cerius2 software, Molecular Simu-
lations, Inc. The Dreiding 2.21 force field was used. For Monte
Carlo simulations and molecular dynamics, the program
MacroModel 7.0 was used for model-building procedures and
as a graphical interface. Force-field parameters were taken
from the built-in force fields, which were in some cases
modified versions of the classical published versions. Amber*
and OPLAA were chosen for all minimizations and Monte
Carlo simulations as well as MD calculations. Both force fields
produce essentially the same results. Minimizations were
initially carried out in the gas phase and then in aqueous
solution. Most complex structures were virtually identical
under both conditions, indicating the strong enthalpic prefer-
ence and hence the stability of these arrangements. Energy
minimizations were conducted over 2000 iterations on a Silicon
Graphics O2 workstation. The best structures were subjected
to conformational searches with 2000-step Monte Carlo simu-
lations. For the optimized conformations, molecular dynamics
calculations were subsequently carried out at room tempera-
ture for 100 ps and without any external restraints (such as
hydrogen bonds, etc.).
1H NMR Titr a tion s. Ten NMR tubes were filled each with
0.80 mL of a solution of the guest compound (concnguest ) 0.5-4
mM) in a deuterated solvent (CD3OD or D2O). The host
compound (1.525 equiv corresponding to the guest) was
dissolved in 0.61 mL of the same solvent, and the resulting
solution was added, in amounts increasing from 0 to 5.0 equiv,
to the 10 guest solutions. Due to their strong hygroscopicity,
the tetrabutylammonium phosphonate solutions contained
approximately 0.3-0.6% water. Volume and concentration
changes were taken into account during analysis. The associa-
tion constants were calculated by nonlinear regression meth-
ods.14
1,3,5-Tr is(p h t h a lim id om e t h yl)-2,4,6-t r im e t h ylb e n -
1
zen e. Yield: 82%. Rf ) 0.60 (20:1 CH2Cl2/acetone). H NMR
(500 MHz, CD2Cl2/CDCl3): δ 2.42 (s, 9 H), 3.57 (s, 6 H), 7.60-
7.69 (AA′BB′). 13C NMR (125 MHz, CD2Cl2/CDCl3): δ 16.27,
37.54, 122.23, 128.39, 131.01, 132.92, 137.34, 171.03. MS (CI,
NH3): m/z 615 (M + NH4+). Anal. Calcd for C36H27N3O6: C,
72.41; H, 4.56; N, 7.04. Found: C, 72.26; H, 4.45; N, 6.88.
Gen er a l P r oced u r e for th e P r ep a r a tion of 1,3,5-Tr is-
(a m in om eth yl)ben zen e Com p ou n d s. The phthalimide com-
pound was dissolved in a hot mixture of dry EtOH/toluene (2:
1) and refluxed with 6 equiv of hydrazine hydrate for 72 h.
The reaction mixture was evaporated, and the residue was first
suspended in ether and then shaken with a cold 40% aqueous
solution of KOH. The extraction was repeated four times, and
the combined organic extracts were dried over Na2SO4 and
filtered. Dry hydrogen chloride was then bubbled through the
filtrate. The yellow precipitate was collected by suction filtra-
tion and dried.
1,3,5-Tr is(a m in om eth yl)-2,4,6-tr im eth ylben zen e Tr i-
1
h yd r och lor id e (14). Yield: 62%. Mp: 108 °C. H NMR (500
MHz, CD3OD): δ 2.56 (s, 9 H), 4.36 (s, 6 H). 13C NMR (125
MHz, CD3OD): δ 17.14, 38.90, 131.34, 141.29. MS of the free
base (CI, NH3): m/z 225 (M + NH4+). Anal. Calcd for C12H24
-
Cl3N3: C, 45.74; H, 7.68; N, 13.34. Found: C, 45.46; H, 7.73;
N, 13.45.
1,3,5-Tr is(am in om eth yl)ben zen e Tr ih ydr och lor ide (15).
Yield: 53%. Mp: 102 °C. 1H NMR (500 MHz, CD3OD): δ 4.22
(s, 6 H), 7.67 (s, 3 H). 13C NMR (125 MHz, CD3OD): δ 43.81,
131.73, 136.37. MS of the free base (CI, NH3): m/z 183 (M +
NH4+). Anal. Calcd for C9H18Cl3N3: C, 39.58; H, 6.64; N, 15.39.
Found: C, 39.22; H, 6.45; N, 15.56.
Gen er a l P r oced u r e for th e P r ep a r a tion of 1,3,5-Tr is-
(1N-p yr a zolylm eth yl)ben zen e Com p ou n d s. A mixture of
the tris(bromomethyl)benzene (1 mmol), 13.2 equiv of pyrazole,
four drops of 40% aqueous [NBu4]OH, and 10 mL of 40%
aqueous NaOH in 40 mL of benzene was refluxed for 48 h.
The organic layer was separated, dried over Na2SO4, and
concentrated in vacuo. The residue was dissolved in methanol
and treated with 3 equiv of aqueous HCl. After evaporation
to dryness, the hydrochloride salt was dried in vacuo.
1,3,5-Tr is(1N -p yr a zolylm e t h yl)-2,4,6-t r im e t h ylb e n -
J ob P lots. Equimolar solutions (10 mmol/10 mL, ap-
proximately 10 µM) of trication and trisphosphonate were
1
prepared and mixed in various ratios. H NMR spectra of the
mixtures were recorded, and the chemical shifts were analyzed
by J ob’s method13a modified for NMR results.13b J ob plots were
carried out for complexes of bisphosphonate 1 with dications
2-7 as well as for the complexes of trisphosphonate 8 with
trications 1,4,7-triazacyclononane trihydrochloride, 11, 12, 15,
and 19 and hexacation 19.
1
Ma ss Sp ectr om etr ic Mea su r em en ts. Samples (20 µL) for
ESI mass spectra were introduced as 10-7 M solutions in
HPLC-grade methanol at flow rates of 20 µL/min. The heated
capillary temperature was 150 °C. The ion spray potential was
3.5 kV (positive ESI), 3.0 kV (negative ESI). About 20-30
scans were averaged to improve the signal-to-noise ratio.
zen e (16). Yield: 86%. H NMR (500 MHz, CD3OD): δ 2.32
(s, 9 H), 5.69 (s, 6 H), 6.61 (m, 3 H), 7.91 (d, J ) 2.2 Hz, 3 H),
7.99 (d, J ) 2.2 Hz, 3 H). 13C NMR (125 MHz, CD3OD): δ
17.42, 48.10, 105.90, 129.10, 138.43, 139.40, 140.92. MS of the
free base (FAB, glycerol matrix, Xe): m/z 361 (55%, M + H+).
Anal. Calcd for C21H24N6: C, 70.02; H, 6.72; N, 23.33. Found:
C, 70.12; H, 6.73; N, 23.45.
1,3,5-Tr is(1N-pyr azolylm eth yl)ben zen e (17). Yield: 90%.
1H NMR (500 MHz, CD3OD): δ 5.34 (s, 6 H), 6.36 (t, J ) 2.2
Hz, 3 H), 7.58 (m, 3 H), 7.72 (d, J ) 2.2 Hz, 3 H), 8.21 (s, 3 H).
13C NMR (125 MHz, CD3OD): δ 56.16, 105.90, 129.10, 138.43,
140.10, 145.87. MS of the free base (FAB, glycerol matrix,
Xe): m/z 319 (62%, M + H+). Anal. Calcd for C18H21Cl3N6: C,
50.54; H, 4.95; N, 19.65. Found: C, 50.42; H, 4.73; N, 19.85.
Ack n ow led gm en t. We thank Prof. Dr. G. Wulff
(Universita¨t Du¨sseldorf) for generous support during the
past three years of T. Grawe’s Ph.D. work.
J O025513Y