Probing polyvalency in artificial systems exhibiting molecular recognition

Article abstract of DOI:10.1021/jo0110799

An approach to the study of polyvalency - the interaction of polyvalent receptors with polyvalent ligands - in unnatural systems is outlined. In this study, the complexation of dibenzylammonium cations by dibenzo [24] crown-8 or benzometaphenylene[25]crown-8 is utilized as the component receptor-ligand interaction. Two analogous multivalent receptors - each containing either seven dibenzo[24]crown-8 (DB24C8 CLUSTER) or seven benzometaphenylene[25] crown-8 (BMP25C8 CLUSTER) moieties appended to a modified β-cyclodextrin core - were prepared in moderate yields. For each of these multivalent receptors, complementary mono- and divalent ligands containing one or two dialkylammonium centers, respectively, were prepared in good yields. These ligands contained fluorine atom substituents to allow their interactions with crown ether compounds to be probed by ¹⁹F NMR spectroscopy. The complexation of these monovalent ligands with the DB24C8 CLUSTER and the BMP25C8 CLUSTER was studied by determining the average binding constant (K_AVE) between the receptors and ligands. The abilities of the crown ether clusters to complex with these monovalent ligands was compared with those of the monovalent crown ethers dibenzo[24]crown-8 and benzometaphenylene[25]crown-8. In both instances, it was found that clustering seven crown ethers together into one molecule is detrimental to the abilities of the crown ether moieties to complex with monovalent dialkylammonium ligands. The complexation of the divalent ligands by the DB24C8 CLUSTER and the BMP25C8 CLUSTER was then studied - again by determining K_AVE - and their abilities to complex with these ligands was compared with those of their respective component interactions. By determining K_AVE for the polyvalent interaction, it was possible to calculate an association constant, K_POLY, for the binding of the divalent ligands by the DB24C8 CLUSTER and the BMP25C8 CLUSTER compounds. In both instances K_POLY for the polyvalent interaction was found to be approximately 2 orders of magnitude higher than the association constants, K_A, for the component interaction.

Full text of DOI:10.1021/jo0110799

P r obin g P olyva len cy in Ar tificia l System s Exh ibitin g Molecu la r
Recogn ition
David A. Fulton, Stuart J . Cantrill,‡ and J . Fraser Stoddart*
Department of Chemistry and Biochemistry, University of CaliforniasLos Angeles, 405 Hilgard Avenue,
Los Angeles, California 90095-1569
stoddart@chem.ucla.edu
Received November 14, 2001
An approach to the study of polyvalencysthe interaction of polyvalent receptors with polyvalent
ligandssin unnatural systems is outlined. In this study, the complexation of dibenzylammonium
cations by dibenzo[24]crown-8 or benzometaphenylene[25]crown-8 is utilized as the component
receptor-ligand interaction. Two analogous multivalent receptorsseach containing either seven
dibenzo[24]crown-8 (DB24C8 CLUSTER) or seven benzometaphenylene[25]crown-8 (BMP 25C8
CLUSTER) moieties appended to a modified â-cyclodextrin coreswere prepared in moderate yields.
For each of these multivalent receptors, complementary mono- and divalent ligands containing
one or two dialkylammonium centers, respectively, were prepared in good yields. These ligands
contained fluorine atom substituents to allow their interactions with crown ether compounds to be
probed by 19F NMR spectroscopy. The complexation of these monovalent ligands with the DB24C8
CLUSTER and the BMP 25C8 CLUSTER was studied by determining the average binding
constant (KAVE) between the receptors and ligands. The abilities of the crown ether clusters to
complex with these monovalent ligands was compared with those of the monovalent crown ethers
dibenzo[24]crown-8 and benzometaphenylene[25]crown-8. In both instances, it was found that
clustering seven crown ethers together into one molecule is detrimental to the abilities of the crown
ether moieties to complex with monovalent dialkylammonium ligands. The complexation of the
divalent ligands by the DB24C8 CLUSTER and the BMP 25C8 CLUSTER was then studieds
again by determining KAVEsand their abilities to complex with these ligands was compared with
those of their respective component interactions. By determining KAVE for the polyvalent interaction,
it was possible to calculate an association constant, KPOLY, for the binding of the divalent ligands
by the DB24C8 CLUSTER and the BMP 25C8 CLUSTER compounds. In both instances KPOLY
for the polyvalent interaction was found to be approximately 2 orders of magnitude higher than
the association constants, KA, for the component interaction.
In tr od u ction
interactions in nature. Indeed, polyvalent interactions
involving multiple ligands and multiple receptors are
almost certainly common in biology,2 possibly as an
evolutionary consequence of the relative ease of increas-
ing binding affinities by multiplying the number of
existing interactions, rather than addressing the more
complicated task of evolving a new, stronger interaction.
There are several recent examples3-5 that highlight the
power of polyvalent ligands to act as extremely effective
inhibitors of biological processes. One of these examples,
the so-called STARFISH ligand, which is an oligovalent
carbohydrate ligand that is able to inhibit Shiga-like
toxins with subnanomolar activity, is a triumph of
rational design.3 Consisting of five “arms”seach termi-
Natural phenomena provide a never-ending and fas-
cinating source of inspiration for chemists seeking to
transfer concepts from the biological world into the
materials world. The so-called “glycoside cluster” effect1
is a particularly interesting phenomenon in glycobiology.
By exploiting the glycoside cluster effect, nature com-
pensates for the inherently low binding affinities of
carbohydrate ligands for lectins by clustering together
carbohydrates and lectins on cell surfaces, leading to
stronger, cooperative polyvalent interactions. The result-
ing gain in binding affinity can often be several orders
of magnitude larger than that of a single component
carbohydrate-lectin binding interaction. The interactions
between carbohydrates and lectins involved in glycoside
cluster effects are not the only examples of polyvalent
(2) Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int.
Ed. 1998, 37, 2754-2794.
(3) Kitov, P. I.; Sadawska, J . M.; Mulvey, G.; Armstrong, G. D.; Ling,
H.; Pannu, N. S.; Read, R. J .; Bundle, D. R. Nature 2000, 403, 669-
672.
(4) Mourez, M.; Kane, R. S.; Mogridge, J .; Metallo, S.; Deschatelets,
P.; Sellman, B. R.; Whitesides, G. M.; Collier, R. J . Nature Biotechnol.
2001, 19, 958-961.
(5) For a number of examples of multivalency in simple biological
systems, such as antibodies, see ref 2.
* Corresponding author. Tel: (310) 206 7078. Fax: (310) 206 1843.
‡ Current Address: Division of Chemistry and Chemical Engineer-
ing, California Institute of Technology, Pasadena, CA 91125.
(1) Lee, R. T.; Lee, Y. C. in Neoglycoconjugates: Preparation and
Applications; Lee, R. T., Lee, Y. C., Eds.; Academic Press: San Diego,
1994, pp 23-50.
10.1021/jo0110799 CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/24/2002
7968
J . Org. Chem. 2002, 67, 7968-7981

Probing Polyvalency in Artificial Systems
nated with a pair of Gb3 trisaccharide analoguess
emanating from a central core, the STARFISH ligand has
been tailored to bind to the B-subunit pentamer of a
Shiga-like toxin. Using crystallographic information to
determine the spatial relationships between the binding
sites of the Shiga-like toxin, the dimensions of the
STARFISH ligand were designed so that it engages
simultaneously, in an extremely effective fashion, to all
10 units of a pair of toxin B subunit pentamers.
Such rational design, however, is not necessarily
required for the construction of an effective polyvalent
inhibitor. It has been demonstrated recently that a
polyacrylamide derivative bearing dodecapeptide ligands
(approximately one ligand for every forty acrylamide
units) is an efficacious inhibitor (IC50 ) 20 nM in vitro)
of the action of the anthrax toxin in an animal model.4
The effectiveness of such “random” polyvalency, although
poorly understood, does raise the issue of quantity versus
quality in the design of polyvalent ligands. Given the
amount of synthetic effort required to prepare effective,
rationally designed, polyvalent ligandssas well as the
realization that suitable crystallographic information is
often not availablesit may be that the most effective
design of polyvalent receptors and ligands often will be
one that emphasizes the quantity of interactions over the
quality of design.
F IGURE 1. A graphical representation of a multivalent
glycoconjugate binding with a multivalent lectin. In this
instance, the complex is stabilized by multiple noncovalent
interactions.
assemble together without freezing too many conforma-
tionally flexible bonds. Even when this feature is absent,
this study demonstrates that extremely high binding
constants can still be achieved in multivalent systems.
There are various contributing mechanisms10 respon-
sible for these gains in binding affinity that occur as a
consequence of glycoside cluster effects, with chelate
effects and statistical effects being two of the most
important. In the chelate effect, a polyvalent carbohy-
drate ligand occupies several binding sites within a single
lectin (Figure 1), leading to a stronger, cooperative
binding interaction. In the case of the statistical effect,
a polyvalent carbohydrate ligand presents a high local
concentration of saccharide epitopes to a single binding
site on the lectin (Figure 2), leading to slower dissociation
Systems that exhibit extremely strong association
between “unnatural” ligands and receptors bearing mul-
tiple copies of their host and guest subunits also have
been reported.6,7 A structurally simple trivalent recep-
tor-ligand pair, based upon the threading of dialkylam-
monium ions through the cavities of large-ring crown
ethers, has been developed and displays very tight
binding in relatively nonpolar organic solvents.6,8 The
association constant for this assembly in CDCl3/CD3CN
1
was determined by H NMR spectroscopy to be greater
than 106 M-1, significantly larger than that9 of the
monomeric receptor and ligand pair (ca. 2 × 103 M-1).
A receptor presenting three subunits of the antibiotic
vancomycin about a trigonal hub binds extremely strongly
(Kd ∼10-17 M) in water to a complementary ligand
presenting three L-Lys-D-Ala-D-Ala tripeptide units.7 Very
careful analysis of the thermodynamics of the multivalent
system led to the conclusion that although the enthalpy
of binding was approximately 3 times that of the basic
receptor-ligand pair (as is to be expected in an ideal
trivalent host-guest system), the gains in the transla-
tional and rotational entropy of bindingsbrought about
by linking the units togetherswere offset by a large loss
of conformational entropy. This elegant study brings
home the point that ideal multivalent receptors and
ligands are those designed such that their subunits
F IGURE 2. A graphical representation of the binding of a
multivalent glycoconjugate with a monovalent lectin.
rates and thus a larger binding affinity. Given the
importance of the glycoside cluster effect in biological
systems, it was decided to design and investigate an
“unnatural” system that would allow an assessment of
the contributions chelate and statistical effects make to
polyvalent interactions. The ability to enhance noncova-
lent binding affinities within a given system, simply by
expressing its component molecular recognition elements
in a polyvalent fashion, remains a relatively unexplored11
concept in supramolecular chemistry that could have
important applications in materials science, such as in
the preparation8 of novel supramolecular polymers.12
(6) Fyfe, M. C. T.; Lowe, J . N.; Stoddart, J . F.; Williams, D. J . Org.
Lett. 2000, 2, 1221-1224.
(7) Rao, J .; Lahiri, J .; Weis, R. M.; Whitesides, G. M. J . Am. Chem.
Soc. 2000, 122, 2698-2710.
(8) It may be possible to extend this concept to the formation of
extremely stable supramolecular polymers, wherein each “dipolytopic”
monomer is bound to each of its “dipolytopic” neighbors through
extremely strong polyvalent noncovalent interactions.
(9) (a) Ashton, P. R.; Campbell, P. J .; Chrystal, E. J . T.; Glink, P.
T.; Menzer, S.; Philp, D.; Spencer, N.; Stoddart, J . F.; Tasker, P. A.;
Williams, D. J . Angew. Chem., Int. Ed. Engl. 1995, 34, 1865-1869.
(b) Ashton, P. R.; Chrystal, E. J . T.; Glink, P. T.; Menzer, S.; Schiavo,
C.; Spencer, N.; Stoddart, J . F.; Tasker, P. A.; Williams, D. J . Chem.
Eur. J . 1996, 2, 709-728.
(10) (a) Pohl, N. L.; Kiessling, L. L. Synthesis 1999, 1515-1519. (b)
Mann, D. A.; Kiessling, L. L. In Glycochemistry; Wang, P. G., Bertozzi,
C. R., Eds.; Marcel Dekker: New York, 2001, pp 221-277.
J . Org. Chem, Vol. 67, No. 23, 2002 7969

Fulton et al.
the CD to complex with guest moleculessit can be safely
assumed that the CD moieties of the cluster molecules
will not have any significant effect on the outcome of
these binding experiments.
This paper describes the synthesis of two CD-based
crown ether cluster molecules, both of which can act as
polyvalent receptors for mono- and divalent dialkylam-
monium ligands. The interaction of these polyvalent
clusters with the mono- and divalent dialkylammonium
ligands was then studied by NMR spectroscopic tech-
niques in an effort to determine the contributions of
chelate and statistical effects to the binding affinities of
these systems.
FIGURE 3. The complexation of dibenzylammonium hexafluo-
rophosphate (DBA‚PF6) with the crown ether DB24C8 to form
a 1:1 complex with a pseudorotaxane geometry.
Resu lts a n d Discu ssion
Syn th etic Str a tegy. The grafting of appendages onto
CD cores is the key step in the synthesis of any CD-based
cluster compound. The perfunctionalization of a hepta-
kisamino-â-CD with N-protected amino acids via amide
bond formation to afford, after deprotection, water-soluble
derivatives has already been successfully accomplished
within our research group.16 This amide bond-forming
methodology has been employed also for the attachment
of carbohydrate-containing appendages to heptakisamino-
â-CD cores in the preparation of cyclodextrin-based
carbohydrate cluster compounds.17 Thus, extension of this
methodology should permit the synthesis of CDs per-
functionalized with crown ether appendages. The hep-
takisamino-â-CD 1, prepared17 as its heptakishydrochlo-
The complexation (Figure 3) of dibenzylammonium
cations by dibenzo[24]crown-8 (DB24C8) is a well-studied
14
binding motif in our laboratory13 and others’ and
therefore makes a good candidate for incorporation into
multivalent conjugates. Additionally, cyclodextrins15 (CDs)
constitute interesting candidates as scaffolds for the
construction of multivalent conjugates. It is possible to
functionalize selectively the primary and/or secondary
faces of CDs, allowing recognition elements to be attached
to the CD scaffold spatially in a very precise manner, and
with a large degree of directionality, i.e., all recognition
elements are pointing in more or less one particular
direction. The attachment of seven recognition elements
to a â-CD scaffold also allows for the C7 symmetry of the
CD to be retained in the resulting conjugate, which could
help to simplify the study of these compounds by NMR
spectroscopic techniques. As all the experiments de-
scribed in this paper have been performed in organic
solventsswhich essentially “switches off” the ability of
(11) (a) Breslow, R.; Greenspoon, N.; Guo, T.; Zarzycki, R. J . Am.
Chem. Soc. 1989, 111, 8296-8297. (b) Zhang, B.; Breslow, R. J . Am.
Chem. Soc. 1993, 113, 9353-9354. (c) Breslow, R.; Duggan, P. J .;
Wiedenfeld, D.; Waddell, S. T. Tetrahedron Lett. 1995, 36, 2707-2710.
(d) Breslow, R.; Zhang, B. J . Am. Chem. Soc. 1996, 118, 8495-8496.
(e) Venema, F.; Nelissen, H. F. M.; Berthault, P.; Birlirakis, N.; Rowan,
A. E.; Feiters, M. C.; Nolte, R. J . M. Chem. Eur. J . 1998, 4, 2237-
2250. (f) Breslow, R.; Belvedere, S.; Gershell, L.; Leung, D. Pure Appl.
Chem. 2000, 72, 333-342.
(12) (a) Archer, R. D. Coord. Chem. Rev. 1993, 128, 49-68. (b)
Rehahn, M. Acta Polym. 1998, 49, 201-224. (c) Moore, J . S. Curr. Opin.
Colloid Interface Sci. 1999, 4, 108-116. (d) Supramolecular Polymers;
Ciferri, A., Ed.; Marcel Dekker: New York, 2000.
ride salt from â-CD in four steps by a highly efficient
methodology (89% overall yield), was chosen as a suitable
CD core on account of the relative ease of its preparation
and the potential for the perfunctionalization of its
primary face via amide bond formation.
Syn th esis of a Diben zo[24]cr ow n -8 Clu ster . The
DB24C8 derivative 5 (Scheme 1), which contains an
active carboxylic acid function at the end of a short spacer
(13) (a) Glink, P. T.; Schiavo, C.; Stoddart, J . F.; Williams, D. J .
Chem. Commun. 1996, 1483-1490. (b) Ashton, P. R.; Baxter, I.; Fyfe,
M. C. T.; Raymo, F. M.; Spencer, N.; Stoddart, J . F.; White, A. J . P.;
Williams, D. J . J . Am. Chem. Soc. 1998, 120, 2297-2307. (c) Cantrill,
S. J .; Fulton, D. A.; Heiss, A. M.; Pease, A. R.; Stoddart, J . F.; White,
A. J . P.; Williams, D. J . Chem. Eur. J . 2000, 6, 2274-2287. (d) Cantrill,
S. J .; Pease, A. R.; Stoddart, J . F. J . Chem. Soc., Dalton Trans. 2000,
3715-3734.
SCHEME 1. Syn th esis of a DB24C8 Der iva tive 5,
Wh ich Con ta in s a n Active Ca r boxyl F u n ction a t
th e En d of a Sh or t Sp a cer Ar m
(14) (a) Kolchinski, A. G.; Busch, D. H.; Alcock, N. W. J . Chem. Soc.,
Chem. Commun. 1995, 1289-1291. (b) Yamaguchi, N.; Gibson, H. W.
Angew. Chem., Int. Ed. Engl. 1999, 38, 143-147. (c) Yuya, T.; Kihara,
N.; Ohga, Y.; Takata, T. Chem. Lett. 2000, 806-807. (d) J ohnson, B.
F. G.; J udkins, C. M. G.; Matters, J . M.; Shephard, D. S.; Parsons, S.
Chem. Commun. 2000, 1549-1550. (e) Zehnder, D. W.; Smithrud, D.
B. Org. Lett. 2001, 3, 2485-2487. (f) Duggan, S. A.; Fallon, G.;
Langford, S. J .; Lau, V. L. Satchell, J . F.; Paddon-Row: M. N. J . Org.
Chem. 2001, 66, 4419-4426.
(15) (a) Ducheˆne, D. Cyclodextrins and Their Industrial Use; Edi-
tions de Sante´: Paris, 1987. (b) Wenz, G. Angew. Chem., Int. Ed. Engl.
1994, 33, 803-822. (c) Szejtli, J . Chem. Rev. 1998, 98, 1743-1755. (d)
Khan, A. R.; Forgo, P.; Stine, K. J .; D’Souza, V. T. Chem. Rev. 1998,
98, 1977-1996.
7970 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
SCHEME 2. Rea ction of th e DB24C8 Der iva tive 5
w ith th e Hep ta k isa m in o-â-CD Der iva tive 1 To
Affor d th e DB24C8 Clu ster
equiv of the acid 5 with the heptakisamino-â-CD 1 in the
presence of 14 equiv of HBTU‚BF4 and 16 equiv of i-Pr2-
NEt in DMF afforded the persubstituted DB24C8 CLUS-
TER in 37% yield after purification by gel filtration
chromatography.
Syn th esis of a Ben zom eta p h en ylen e[25]cr ow n -8
Clu ster . A benzometaphenylene[25]crown-8 (BMP 25C8)
analogue of 5, which also contains an active carboxylic
acid function at the end of a short spacer arm, was
prepared in three synthetic steps starting from the
known22 BMP 25C8 benzyl alcohol derivative 6 (Scheme
3). Treatment of 5 with methanesulfonyl chloride and
Et3N in CHCl3 afforded quantitatively the mesylate 7,
which, upon reaction with methyl 3-mercaptopropionate,
was transformed cleanly to the methyl ester derivative
8. As observed previously, direct reaction of the mesylate
7 with 3-mercaptopropionic acid failed to produce the
desired product either cleanly or in a reproducible
fashion. Saponification of 8 with NaOH in THF/H2O then
afforded the desired carboxylic acid 9 in an overall yield
of 76% from alcohol 6. The crown ether derivative 9 was
attached to the heptakisamino-â-CD 1 in the same
manner as used for the formation of the DB24C8
CLUSTER. Reaction (Scheme 4) of 14 equiv of the acid
9 with the heptakisamino-â-CD 1 in the presence of 14
equiv of HBTU‚BF4 and 16 equiv of i-Pr2NEt in DMF
afforded the persubstituted BMP 25C8 CLUSTER in
36% yield after purification by gel filtration chromatog-
raphy.
SCHEME 3. Th e Syn th esis of a BMP 25C8
Der iva tive 9, Wh ich Con ta in s a n Active Ca r boxyl
F u n ction a t th e En d of a Sh or t Sp a cer Ar m
arm, was prepared in four steps from the known18 formyl-
substituted DB24C8 macrocycle 2. Reduction of 2 with
LiAlH4 in THF afforded smoothly the corresponding
alcohol19 3 in 96% yield, which, following treatment with
methanesulfonyl chloride and Et3N in CHCl3, yielded the
mesylate 4. Nucleophilic displacement with methyl 3-mer-
captopropionate20 afforded cleanly the methyl ester de-
rivative that was then transformed readily to the car-
boxylic acid 5 upon saponification with NaOH in MeOH.
This displacement/saponification approach to the syn-
thesis of 5 was found to be far superior to the nucleophilic
displacement reaction between the mesylate 4 and
3-mercaptopropionic acid. The amide coupling reagent
O-benzotriazol-1-yl-N,N,N′N′-tetramethyluronium tet-
rafluoroborate (HBTU‚BF4) has been used effectively for
the attachment of carbohydrate-containing ligands to
cyclodextrin cores.17,21 Thus, reaction (Scheme 2) of 14
Ch a r a cter iza tion of Cr ow n Eth er Clu ster s. A
1
detailed analysis of the H NMR spectra of the DB24C8
CLUSTER and the BMP 25C8 CLUSTER (the latter is
depicted in Figure 4) was performed in order to ensure
that complete perfunctionalization of the CD cores with
(16) Alker, D.; Ashton, P. R.; Harding, V. D.; Ko¨niger, R.; Stoddart,
J . F. J . Org. Chem. 1996, 61, 903-908.
1
crown ethers had occurred. The H NMR spectra of the
(17) Fulton, D. A.; Pease, A. R.; Stoddart, J . F. Israel J . Chem. 2000,
40, 325-333.
DB24C8 CLUSTER and the BMP 25C8 CLUSTER
were reasonably sharp and well-defined, features that
are indicative of complete substitution of the CD core
with crown ether appendages. A small degree of line
(18) Ashton, P. R.; Baxter, I.; Cantrill, S. J .; Fyfe, M. C. T.; Glink,
P. T.; Stoddart, J . F.; White, A. J . P.; Williams, D. J . Angew. Chem.,
Int. Ed. 1998, 37, 1294-1297.
(19) (a) Yamaguchi, N.; Gibson, H. W. Angew. Chem., Int. Ed. 1999,
38, 143-147. (b) Diederich, F.; Echegoyen, L.; Go´mez-Lo´pez, M.;
Kessinger, R.; Stoddart, J . F. J . Chem. Soc., Perkin Trans. 2 1999,
1577-1586.
1
broadening is evident in both of these H NMR spectra,
a phenomenon that has been observed23 in several other
instances of CDs persubstituted with carbohydrate ap-
pendages, which is thought to be a consequence of
relatively slow dynamic processes occurring in these
molecules. Nevertheless, it is observed (Figure 4) that
(20) Baker, B. R.; Merle, V. Q.; Bernstein, S.; Safir, S. R.; Subbarow,
Y. J . Org. Chem. 1947, 12, 167-173.
(21) Attempts to utilize dicyclohexylcarbodiimide (DCC) and hy-
droxybenzotriazole as activating agents were unsuccessful, largely on
account of difficulties encountered in removing the byproducts of the
reactions. Although benzotriazol-1-yloxytripyrrolidinophosphonium
hexafluorophosphate (Py-BOP) was found to act as a suitable activator,
the removal of the tripyrrolidinophosphonium oxide byproduct was far
from a straightforward exercise.
(22) Cantrill, S. J .; Youn, G. J .; Stoddart, J . F.; Williams, D. J . J .
Org. Chem. 2001, 66, 6857-6872.
J . Org. Chem, Vol. 67, No. 23, 2002 7971

Fulton et al.
in the 13C NMR spectra of these compounds. The partially
assigned 13C NMR spectrum of the DB24C8 CLUSTER
is shown in Figure 5. The signals are sharp and well-
defined, a feature that is also evident in the spectrum of
the BMP 25C8 CLUSTER. The signals corresponding to
carbon atoms in the CD torus have weaker intensities
than those signals associated with carbon atoms in the
crown ether appendages. Again, this phenomenon has
been observed23 previously in the 13C NMR spectra of CD-
based carbohydrate clustersspresumably as a result of
some dynamic phenomenasand is thus not unexpected.
Finally, further evidence to support the homogeneity of
the DB24C8 CLUSTER and the BMP 25C8 CLUSTER
is found in the MALDI-TOF mass spectra of both of these
compounds. In each instance, only a single peak corre-
sponding to the molecular ion can be detected; there are
no signals corresponding to ions derived from CD cores
having less than seven crown ether appendages.
SCHEME 4. Rea ction of th e BMP 25C8 Der iva tive
9 w ith th e Hep ta k isa m in o-â-CD Der iva tive 1 To
Affor d th e BMP 25C8 Clu ster
Syn th esis of Mon o- a n d Diva len t Dia lk yla m m o-
n iu m Sa lts. Two pairs of mono- and bis(dialkylammo-
nium) salts were prepared and were used as mono- and
divalent ligands in the study of the binding properties of
the DB24C8 CLUSTER and the BMP 25C8 CLUSTER.
Each ligand contains fluorine atom substituents, which
allowed the binding with the crown ether clusters to be
studied using 19F NMR spectroscopy, a technique that
has proven extremely useful22 for studying the binding
of complex dialkylammonium ion/crown ether systems.
+
The mono- and dications DBA-F 2 and bis-DBA-F 22+s
each of which contains two para-substituted fluorine
atomsswere used as ligands in investigations with the
DB24C8 CLUSTER. The corresponding pair of mono-
the signals corresponding to the CD protons H-1 and H-2
in the H NMR spectrum of the BMP 25C8 CLUSTER
1
+
2+
and dications DBA-F 4 and bis-DBA-F 4 (Scheme 5)
were used as ligands in binding studies with the
BMP 25C8 CLUSTER. It has been shown22 that bis(3,5-
difluorobenzyl)ammonium cations thread through the
cavity of BMP 25C8 at a rate such that slow exchange of
complexed and uncomplexed species is observed on the
19F NMR time scale at 376 MHz. Therefore, binding
constants can be determined24 simply by integrating the
signals of the bound and unbound species present in the
19F NMR spectrum of a solution of the crown ether and
the dialkylammonium cation, as is the case9 with the
binding of dibenzylammonium salts with DB24C8 de-
rivatives.
appear as sharp multiplets of correct integration, spec-
troscopic features that support very strongly the hypoth-
esis that the CD core is persubstituted with seven crown
ether appendages. The 1H NMR spectrum of the DB24C8
CLUSTER is slightly less well defined than that of the
BMP 25C8 CLUSTER; in this instance the signals
corresponding to the H-1 and H-2 protons in the CD torus
appear as an almost-resolved doublet and a doublet of
doublets, respectively. The quality of both these 1H NMR
spectra strongly supports the hypothesis that both the
DB24C8 CLUSTER and the BMP 25C8 CLUSTER are
fully substituted with seven crown ether-containing
appendages.
Thus, as reported previously,22 condensation of the
aldehyde 10 with the benzylamine 11 afforded (Scheme
5)safter reduction of the resulting imine, protonation
with HCl and counterion exchange with NH4PF6sthe
+
p-fluoro-substituted monovalent cation DBA-F 2 as its
PF6 salt. Using an identical procedure,22 the aldehyde
-
12 was reacted with the amine 13 to afford the PF6- salt
of DBA-F 4+. Condensation of 11 with the dialdehyde 14
affordedsafter reduction of the resultant diimine with
NaBH4 followed by direct protonation with NH4PF6sthe
(23) (a) Garc´ıa-Lo´pez, J . J .; Santoyo-Gonza´lez, F.; Vargas-Berenguel,
A.; Gime´nez-Mart´ınez, J . J . Chem. Eur. J . 1999, 5, 1771-1784. (b)
Garc´ıa-Lo´pez, J . J .; Herna´ndez-Mateo, F.; Isac-Garc´ıa, J .; Kim, J . M.;
Roy, R.; Santoyo-Gonza´lez, F.; Vargas-Berenguel, A. J . Org. Chem.
1998, 64, 522-531. (c) Furuike, T.; Aiba, S.; Nishimura, S.-I. Tetra-
hedron 2000, 56, 9909-9915. (d) Fulton, D. A.; Stoddart, J . F. Org.
Lett. 2000, 2, 1113-1116.
F IGURE 4. 1H NMR spectrum (500 MHz, CDCl3) of the
BMP 25C8 CLUSTER.
(24) Hill, Z. D.; MacCarthy, P. J . Chem. Educ. 1986, 63, 162-167.
(b) Connors, K. A. Binding Constants; Wiley-Interscience: New York,
1987, pp 191-192. (c) Adrian, J . C.; Wilcox, C. S. J . Am. Chem. Soc.
1991, 113, 678-680. (d) Fielding, L. Tetrahedron 2000, 56, 6151-6170.
Further evidence that the CD cores in the DB24C8
CLUSTER and the BMP 25C8 CLUSTER are each fully
substituted with seven crown ether appendages is present
7972 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
F IGURE 5. 13C NMR Spectrum (125 MHz, CDCl3) of the DB24C8 CLUSTER.
+
+
SCHEME 5. Syn th esis of th e Mon ova len t Am m on iu m Ca tion s DBA-F 2 a n d DBA-F 4 a n d th e Diva len t
Bisa m m on iu m Dica tion s Bis-DBA-F 2 a n d Bis-DBA-F 4
2+
2+
dication bis-DBA-F 22+. Condensation of 13 with 14
furnished, after similar treatment, the dication bis-DBA-
measure of the strength of the noncovalent binding
interaction between a single crown ether moiety and a
single dialkylammonium center,” is defined mathemati-
cally as
F 4
.
19F NMR spectroscopy was used to confirm that the
2+
compounds DBA-F 2+, DBA-F 4+, bis-DBA-F 22+, and bis-
2+
-
DBA-F 4 had been prepared as their PF6 salts. Inte-
grations of the signals corresponding to the fluorine atom
substituents and the PF6- anion gave the expected ratios
in the case of all four compounds.
KAVE
)
[complexed NH2+ sites]
[uncomplexed NH2+ sites][uncomplexed crown ethers]
(1)
Sta tistica l Effects. To probe statistical effects in
polyvalency, a simple question was posed, namely, will
a crown ether moiety bind a single dialkylammonium
cation more strongly when the crown ethers are clustered
together in a single molecule than when a single crown
ether binds a single dialkylammonium cation? If the
answer to this question is “yes”, it can be concluded that
clustering crown ethers together in a single molecule
should lead to an increase in binding as a consequence
of statistical effects.25 Quantitatively, this question can
be answered by determining the average binding con-
stant (KAVE) for the complexation of a dialkylam-
monium cation by a single crown ether moiety. The
average binding constant, which is defined here as “a
+
where an “NH2 site” is a single dialkylammonium
moiety within a compound containing one or more
dialkylammonium sites.
In the first experiment performed, the binding of
+
DB24C8 with the dialkylammonium thread DBA-F 2
was investigated by NMR spectroscopy. This experiment
served as the model upon which the following NMR
experiments were based. From the 19F NMR spectrum
(Figure 6) of a solution of DB24C8 (7 mM) and the
dialkylammonium thread DBA-F 2+ (14 mM) in CD3CN,
it is straightforward to calculate the concentrations of
complexed and uncomplexed DBA-F 2+. The concentra-
tions of complexed and uncomplexed DB24C8 can there-
fore be calculated, as it is known22 that DB24C8 binds
DBA-F 2+ in a 1:1 complex with a pseudorotaxane geom-
(25) It is assumed that the electronic properties of the crown ethers
are not significantly altered as a consequence of the presence of the
tether that attaches the crown ether to the CD core.
J . Org. Chem, Vol. 67, No. 23, 2002 7973

Fulton et al.
F IGURE 6. Partial 19F NMR spectrum (376 MHz, CD3CN)
+
of a solution of DB24C8 (7 mM) and DBA-F 2 (14 mM),
showing signals for both complexed and uncomplexed DBA-
+
F 2
.
26
etry. Thus, the average association constant (KAVE
)
for
+
the binding interaction between DB24C8 and DBA-F 2
can be calculated using eq 1. To obtain a better estimate
for the value of KAVE for this binding interaction, an NMR
titration experiment was performed, in which the con-
centration of DB24C8 was held constant at 7 mM
throughout, while incrementally increasing the concen-
tration of DBA-F 2+. By performing this titration experi-
ment, it is possible to plot a binding isotherm27,28 (Figure
F IGURE 7. “Best fit” binding isotherms (solid lines) for (i)
+
complexation of DBA-F 2 by DB24C8 and (ii) complexation
+
of DBA-F 2 by the DB24C8 CLUSTER. (a) The values
represented on this axis represent the measured concentration
of complexed dialkylammonium centers divided by the total
concentration of crown ether in solution. Error bars on this
axis are set at (10% to reflect errors in the preparation of the
sample and in determination of values of the integrals. (b) The
values represented on this axis represent the total concentra-
tion of dialkylammonium cations in solution. Error bars on
this axis are set at (4% to reflect errors in the preparation of
the sample.
+
7) for the complexation of DBA-F 2 with DB24C8. A
“best-fit” isotherm has been drawn and serves as a useful
comparison with other isotherms.29 An accurate value of
KAVE was therefore determined by simply taking the
mean of the KAVE values, each of which corresponds to a
data point on the binding isotherm. The standard devia-
tion from this pool of data serves as an indication of the
error in the calculated value of KAVE. For the binding of
DB24C8 with the thread DBA-F 2+, the mean value of
KAVE was calculated to be 790 ( 240 M-1, a value that
compares well30 to the complexation of other dialkylam-
monium ions by DB24C8.
A similar NMR titration experiment was performed to
determine the mean value of KAVE for the binding of
+
thread DBA-F 2 with the DB24C8 CLUSTER. In this
NMR titration experiment, the concentration of the
DB24C8 CLUSTER was held constant throughout at 1
mM, i.e., the concentration of crown ethers in solution
remains constant at 7 mM, a concentration that is
identical to that of DB24C8 used in the previous experi-
ment. A sample 19F NMR spectrum from this titration
experiment is shown in Figure 8. As expected, signals
(26) In this instance, KAVE ) KA, where KA is the binding constant
+
between DB24C8 and DBA-F 2
.
+
corresponding to complexed and uncomplexed DBA-F 2
(27) We have chosen to use binding isotherms, rather than Scat-
chard plots, to display our data. The main advantage of displaying
data in Scatchard plots is that curvature of the data points suggests
positive or negative cooperativity. (For a recent example of the use of
Scatchard plots in determining the degree of cooperativity in a host-
guest system similar to ours, see: Gibson, H. W.; Yamaguchi, N.;
Hamilton, L.; J ones, J . W. J . Am. Chem. Soc. 2002, 124, 4653-4665.)
Nonideal polyvalent systems are invariably uncooperative. It should
be noted that Scatchard plots are mathematical approximations of the
actual binding isotherm and thus are subject to the limitations of such
an approximation. (There is some controversy about the advantages
and limitations of Scatchard analysis; see ref 24b, as well as Klotz, I.
M. Acc. Chem. Res. 1974, 7, 162-168). We decided, therefore, to model
our systems on a 1:1 binding isothermsarguably the purest way in
which to represent receptor-ligand interactionssas polyvalent inter-
actions can be simplified in their analysis to the interaction between
one ligand and one receptor. Our experimental evidence reported in
this paper suggests that this model is a valid one.
(28) To aid clarity, all binding isotherms presented in this work have
been plotted using the approximation that [free ligand] ) [total ligand].
(29) As the binding isotherms described here are only qualitative,
only four to six data points were determined for every system discussed.
Even this small number of data points allows a binding isotherm to
be determined that is sufficiently accurate for the purposes of qualita-
tive comparison. Bear in mind also that the association constant is
calculated at each point in the isotherm using the single-point
technique, and thus, we do not require multiple data points in order
can be observed in the NMR spectrum. The signal
+
corresponding to complexed DBA-F 2 is broadened,
presumably because (1) the fluorine atoms of complexed
+
DBA-F 2 are diastereotopic because of the chirality of
the CD and (2) there are many different isomers of the
complex formed when the cluster complexes with two or
more dialkylammonium ion threads. Taken together, it
+
is clear that the fluorine atoms of bound DBA-F 2 can
find themselves in many slightly differing magnetic
environments upon complexation with the DB24C8
CLUSTER, leading to a complex, broadened signal.31
Nevertheless, from the NMR spectrum, it is straightfor-
ward to determine the concentrations of uncomplexed
+
and complexed DBA-F 2 in solutionsand therefore the
concentrations of uncomplexed and complexed crown
ether moietiessallowing a value of KAVE to be obtained.
This NMR experiment was repeated at different concen-
+
trations of DBA-F 2 and the binding isotherm (Figure
7) plotted from the data points. From these data points,
the mean KAVE was determined to be 110 ( 33 M-1
.
to utilize a curve-fitting approach to solve for KAVE
.
(30) (a) Ashton, P. R.; Fyfe, M. C. T.; Hickingbottom, S. K.; Stoddart,
J . F.; White, A. J . P.; Williams, D. J . J . Chem. Soc., Perkin Trans. 2
1998, 2117-2128. (b) Ashton, P. R.; Bartsch, R. A.; Cantrill, S. J .;
Hanes, R. E., J r.; Hickingbottom, S. K.; Lowe, J . N.; Preece, J . A.;
Stoddart, J . F.; Talanov, V. S.; Wang, Z.-H. Tetrahedron Lett. 1999,
40, 3661-3664.
The abilities of BMP 25C8 and the BMP 25C8 CLUS-
+
TER to bind the dialkylammonium ion DBA-F 4 were
(31) Although this signal is broad, we believe that it is still possible
to obtain an accurate value for its area by integration.
7974 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
F IGURE 8. Partial 19F NMR spectrum (376 MHz, CD3CN) of a solution of the DB24C8 CLUSTER (1 mM) and DBA-F 2 (14
+
+
mM), which shows signals for both complexed and uncomplexed DBA-F 2
.
evaluated in a manner similar to that described for the
ether moiety binds a single ammonium cation more
strongly when the crown ethers are clustered together
in a single molecule, compared to a single crown ether
+
binding of DBA-F 2 by DB24C8 and the DB24C8
CLUSTER. The binding isotherms32 for the complexation
+
of DBA-F 4 by BMP 25C8 and the BMP 25C8 CLUS-
TABLE 1. Va lu es of Mea n KAVE a n d ∆G° for th e
AVE
TER are shown in Figure 9. The mean KAVE for the
binding22 of DBA-F 4+ by BMP 25C8 was calculated to be
20 ( 6 M-1, and the mean KAVE for the binding of this
ion by the BMP 25C8 CLUSTER was calculated to be
In ter a ction s in CD3CN a t 300 K of DB24C8 a n d th e
+
DB24C8 CLUSTER w ith DBA-F 2 a n d of BMP 25C8 a n d
+
th e BMP 25C8 CLUSTER w ith DBA-F 4
mean
∆G°
AVE
11 ( 3 M-1
.
host
guest
KAVE (M-1
)
(kcal mol-1
)
+
DB24C8
DB24C8 CLUSTER
BMP 25C8
DBA-F 2+
DBA-F 2+
DBA-F 4+
790 ( 240
110 ( 33
20 ( 6
-3.9 ( 0.2
-2.8 ( 0.2
-1.8 ( 0.2
-1.4 ( 0.2
BMP 25C8 CLUSTER DBA-F 4
11 ( 3
binding a single ammonium cation, can be answered with
a very definite “no”. This negative answer is probably a
(32) The Weber rule of thumb (see, for instance: Tsukube, H.;
Furuta, H.; Odani, A.; Takeda, Y.; Kudo, Y.; Inoue, Y.; Liu, Y.;
Sakamoto, H.; Kimura, K. In Comprehensive Supramolecular Chem-
istry, Davies, J . E., Ed.; Pergamon Press: Oxford, 1996; Vol. 8, pp 425-
482) states that accurate association constants can be obtained only
at total concentrations of host and guest at which the minor component
in the system is between 20 and 80% complexed. In the upper isotherm
of Figure 9, all data are within this range, even the point labeled 28
M-1, which corresponds to ca. 25% of the ammonium ions being
complexed. Every point in the lower isotherm, however, corresponds
to a situation in which there is less than 20% complexation of all of
the species in solution. This situation arises because of the very weak
association constant in this system (see discussion in the text), which
results in only ca. 18% of the crown ether units being bound at 28 mM
of DBA-F 4+. Presumably, adding more of this salt would eventually
result in greater than 20% of the crown ether units being complexed,
but we did not wish to overload our solutions with salt in order to avoid
affecting the dielectric constant of the medium. As such, the data points
in this binding isotherm must be treated with caution, but, even so,
the value of KAVE in this system obviously is very low.
F IGURE 9. “Best fit” binding isotherms for (i) complexation
of DBA-F 4 by BMP 25C8 and (ii) complexation of DBA-F 4
+
+
by the BMP 25C8 CLUSTER. (a) The values represented on
this axis represent the measured concentration of complexed
dialkylammonium centers divided by the total concentration
of crown ether in solution. (b) The values represented on this
axis represent the total concentration of dialkylammonium
cations in solution. Error bars are set as defined in Figure 7.
(33) It appears from the data presented in Table
1 that the
magnitude of this detrimental binding effect, which occurs when crown
ethers are clustered, is somewhat more pronounced in the DB24C8
systems than in the BMP 25C8 systems. There are two principal
reasons, however, why it is unwise to draw comparisons between the
DB24C8 and BMP 25C8 systems. First, these systems were studied
at different concentrations of crown ether moieties; in the DB24C8
system, the concentration of crown ethers was kept constant at 7 mM,
while in the BMP 25C8 system, the concentration of crown ethers was
kept constant at 14 mM. Recent research (Chang, T.; Heiss, A. M.;
Cantrill, S. J .; Fyfe, M. C. T.; Pease, A. R.; Rowan, S. J .; Stoddart, J .
F.; White, A. J . P.; Williams, D. J . Org. Lett. 2000, 2, 2947-2950) has
indicated that the binding of dibenzylammonium cations by crown
ethers is concentration dependent, that is, association constants can
differ depending on the concentrations at which they are measured.
Second, the DB24C8 and BMP 25C8 systems are, almost certainly,
not identical in terms of the enthalpy and entropy changes that occur
when dibenzylammonium cations are bound. Again, it is concluded that
it is unwise to draw comparisons between them.
The results from all four experiments are summarized
in Table 1. It appears that the clustering of seven
DB24C8 moieties into one cluster molecule is detrimen-
tal to the abilities of its crown ether moieties to bind the
dialkylammonium cation DBA-F 2+, relative to the ability
of DB24C8 to bind the same dialkylammonium cation.
In terms of free energy, the costs of clustering DB24C8
moieties together is ca. 1.1 kcal mol-1 in this instance.
The same trend33 is observed with the BMP 25C8 CLUS-
TER, whose crown ether moieties bind the ammonium
+
thread DBA-F 4 less effectively than BMP 25C8 by ca.
0.4 kcal mol-1. Therefore, the question of whether a crown
J . Org. Chem, Vol. 67, No. 23, 2002 7975

Fulton et al.
SCHEME 6. A Gr a p h ica l Rep r esen ta tion of th e Com p lexa tion of DB24C8 w ith th e Diva len t Bisa m m on iu m
2+
Dica tion Bis-DBA-F 2
consequence of unfavorable electrostatic and steric ef-
fects, which occur when the crown ether clusters bind
dialkylammonium ions. When a cluster binds a single
dialkylammonium ion, the resulting complex carries a
unit of positive charge. Even though solvent and anions
will help to shield this bound positive charge from other
unbound dialkylammonium ions, there will still be a
certain degree of electrostatic repulsion between the
positively charged complex and the positively charged
unbound ions, thus reducing the ability of the cluster to
bind any further dialkylammonium cations. Indeed, the
more dialkylammonium cations complexed per cluster,
the greater this electrostatic repulsion will become. Steric
hindrance from bound threads and anionsswhich prob-
ably form loose ion pairs with bound ionsscould also
make it even more difficult for the crown ether clusters
to bind any further dialkylammonium ions. Yamaguchi
and Gibson reported34 similar findings for the binding of
dibenzylammonium cations by polymers containing ap-
pended DB24C8 moieties. In that study, the authors
suggested that electrostatic repulsion and steric hin-
drance were responsible for reducing the abilities of the
appended crown ethers to complex with dibenzylammo-
nium cations.
one signal (the most downfield) corresponds to the
fluorine atom adjacent to the uncomplexed dialkylam-
monium centers and the other signal (the most upfield)
corresponds to the fluorine atoms adjacent to the com-
plexed dialkylammonium center. From integrations of
these signals, it is straightforward to calculate the
concentrations of all species in solution, and thus, KAVE
can be calculated using eq 1. The microscopic binding
constants K1 (1500 ( 440 M-1) and K2 (730 ( 220 M-1
)
were also determined.35 As before, KAVE was determined
at various concentrations of bis-DBA-F 22+ and a binding
isotherm plotted (Figure 11).36 A mean value of KAVE was
calculated from the data points, with the standard
deviation taken to be a measure of the error in KAVE. The
mean value of KAVE for the binding of bis-DBA-F 22+ with
DB24C8 was calculated to be 1100 ( 340 M-1
.
P r obin g Ch ela te Effects. The study of the contribu-
tion of chelate effects to polyvalent interactions is not
straightforward. In this discussion, the system of nomen-
clature describing polyvalent interactions proposed by
Whitesides and co-workers2 is used. A polyvalent ligand
can interact with a polyvalent receptor in numerous
different ways to afford potentially a large number of
species in solution. The value of KAVE for a polyvalent
system, relative to KA for its component interaction, is
by no means indicative of the strength of a polyvalent
interaction. In other words, KAVE for a polyvalent interac-
tion can be of smaller value than KA for the component
interaction, but the binding constant for the polyvalent
interaction (KPOLY) can still be extremely high. However,
F IGURE 10. Partial 19F NMR spectra (376 MHz, CD3CN) of
(a) a solution of DB24C8 (7 mM) and Bis-DBA-F 2 (7 mM)
2+
and (b) a solution of the DB24C8 CLUSTER (1 mM) and Bis-
2+
DBA-F 2 (7 mM).
K
AVE for a polyvalent interaction is still a crucial quantity,
The binding of bis-DBA-F22+ with the DB24C8 CLUS-
TER was then investigated. The 19F NMR spectrum of a
solution containing the cluster (1 mM cluster, concentra-
since, with its careful interpretation, KAVE can be used
to determine KPOLY and, therefore, the free energy change
for the polyvalent interaction, ∆G°
.
POLY
In the first NMR experiment, the binding of bis-DBA-
F 22+ by DB24C8 in CD3CN, which follows the equilibria
outlined in Scheme 6, was examined. The 19F NMR
(35) The ratio of the microscopic binding constants (K1/K2) is
approximately 2. For a ditopic system in which the recognition sites
act independently, this ratio should be 4 (see ref 24b). The smaller
observed ratio seems to suggest that this system behaves in a positively
cooperative manner. This conclusion, however, might be erroneous. The
assumption that the recognition sites act independently does not hold
in such a system, because the crown ether unit in a 1:1 complex can
2+
spectrum of a solution of bis-DBA-F 2 (7 mM) and
DB24C8 (7 mM) is shown in Figure 10a. Four signals
are observed in this spectrum: one signal corresponding
to uncomplexed bis-DBA-F 22+, one signal for the 2:1
complex, and two signals corresponding to the 1:1 com-
plex. Of the two signals arising from the 1:1 complex,
+
shuttle between the two NH2 centers of the dication. It is not
intuitively obvious what effect such shuttling will have on the ratio
K1/K2, but it seems possible that the rate at which a crown ether unit
slips on to a dicationic thread (kon) of a shuttling system will be double
that exhibited in a system that cannot shuttle. If true, the effect would
be to double K2, and so K1/K2 would be expected statistically to be 2,
rather than 4.
(34) Yamaguchi, N.; Gibson, H. W. Macromol. Chem. Phys. 2000,
201, 815-824.
7976 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
2+
tion of crown ethers ) 7 mM) and the bis-DBA-F 2 (7
mM) is shown in Figure 10b. Two distinct signals are
observed in this spectrum: a sharp signal (-113.5 ppm)
corresponding to uncomplexed bis-DBA-F 22+ and a very
broad signal (-114.1 ppm) corresponding to complexed
NMR spectrum of a solution of BMP 25C8 (14 mM) and
bis-DBA-F 42+ (9 mM) in CD3CN is shown in Figure 12a.
Three signals can be observed in this NMR spectrum: a
2+
sharp singlet corresponding to unbound bis-DBA-F 4
,
a sharp singlet corresponding to the 2:1 complex, and a
broadened signal corresponding to the 1:1 complex. This
signal for the 1:1 complex is broadened, since presumably
2+
bis-DBA-F 22+, where bis-DBA-F 2 is complexed pre-
sumably by two crown ether moieties of the cluster. This
assumption is made after comparison with the NMR
spectrum in Figure 10a. In Figure 10b, there do not
appear to be any signals at ca. -114.1 ppmsat least as
can be detected by 19F NMR spectroscopyscorresponding
+
the bound crown is shuttling between the two NH2
centers of the threadlike dication at a rate approaching
that of the 19F NMR time scale at 376 MHz. Despite the
fact that there is some partial overlap of the broadened
signal (corresponding to the 1:1 complex) and the sharp
singlet corresponding to uncomplexed bis-DBA-F 42+, it
is still possible to integrate each of these signals and
therefore calculate the concentrations of each of the
species in solution and, hence, the values of the micro-
scopic binding constants38 K1 and K2, as well as KAVE. By
performing this NMR experiment at various concentra-
tions of bis-DBA-F 42+, again it is possible to construct a
binding isotherm (Figure 13). From the data points, the
2+
to the case in which a bis-DBA-F 2 unit is complexed
by only a single crown ether moiety of the cluster; i.e., in
2+
solution, bis-DBA-F 2 appears to be either completely
+
uncomplexed or both of its NH2 centers are complexed
by crown ether moieties. The signal corresponding to
complexed bis-DBA-F 22+ has now moved downfield rela-
2+
tive to the signal observed when bis-DBA-F 2 is com-
plexed by DB24C8, presumably as a consequence of the
unique microenvironment that exists within the crown
ether cluster.37 By integration of the signals in the NMR
mean value of KAVE was found to be 30 ( 9 M-1
.
spectrum, again it is straightforward to calculate KAVE
.
By performing this NMR experiment at various concen-
trations of bis-DBA-F 22+, it is possible to construct a
binding isotherm (Figure 11) and, by taking the average
of these data points, calculate the mean KAVE to be 160
( 48 M-1
.
F IGURE 11. “Best fit” binding isotherms for (i) complexation
of Bis-DBA-F 2+ by DB24C8 and (ii) complexation of Bis-DBA-
2+
F 2 by the DB24C8 CLUSTER. (a) The values represented
on this axis represent the measured concentration of com-
plexed dialkylammonium centers divided by the total concen-
tration of crown ether in solution. (b) The values represented
on this axis represent the total concentration of dialkylam-
monium cations in solution. Error bars are set as defined in
Figure 7.
F IGURE 12. Partial 19F NMR spectra (376 MHz, CD3CN) of
(a) a solution of BMP 25C8 (14 mM) and Bis-DBA-F 42+ (9 mM)
and (b) a solution of the BMP 25C8 CLUSTER (2 mM) and
2+
Bis-DBA-F 4 (4 mM).
The 19F NMR spectrum of a solution of the BMP 25C8
CLUSTER (2 mM, concentration of crown ethers ) 14
The binding of bis-DBA-F 42+ with BMP 25C8 and the
BMP 25C8 CLUSTER was then investigated. The 19F
2+
mM) and bis-DBA-F 4 (4 mM) in CD3CN is shown in
Figure 12b. Two signals are observed in this spectrum:
2+
one signal corresponding to uncomplexed bis-DBA-F 4
(36) A couple of data points in the upper isotherm in Figure 11
exhibit greater than 80% complexation of the minor (crown ether)
component in the equilibrium, and one point exhibits less than 20%
complexation. As discussed in ref 32, such data may lead to unreliable
values of the association constant. We believe, however, that the trend
is clear in this series of experiments and that the accuracy of
integration of 19F NMR spectra makes these values significant enough
and another broad signal corresponding to aggregates in
which bis-DBA-F 42+ units are each complexed by either
one or two BMP 25C8 moieties. In the previous example
2+
of the binding of the b is-DBA-F 2 by the DB24C8
CLUSTER , the NMR spectrum (Figure 10b) provided
to include them in the calculation of KAVE
.
(37) The possibility exists that dibenzylammonium cations can be
bound by two crown ether moietiesscontained within a single clusters
in a sandwich-like complex, as opposed to in a threaded pseudorotax-
ane-like geometry.
(38) The two values (K1 ) 55 ( 17 M-1, K2 ) 25 ( 8 M-1; K1/K2
)
ca. 2.2) once again seem to suggest that positive cooperativity is in
effect.
J . Org. Chem, Vol. 67, No. 23, 2002 7977

Fulton et al.
that of the mean KAVE for the interaction of the dicationic
thread with BMP 25C8 and approximately 3 times larger
than that of the component interaction. Again, given the
differences in concentrations at which the DB24C8 and
BMP 25C8 systems were investigated, and the intrinsic
differences between both systems, it is probably unwise
to try to draw comparisons.
With careful interpretation of the mean KAVE, it is
possible to calculate the binding constant (KPOLY) for the
binding of 1 equiv of the bis-DBA2+ threads by 1 equiv
of the CLUSTER compounds to afford the 1:1 complex,
i.e., where one bis-DBA2+ molecule is complexed by one
CLUSTER molecule, irrespective of the geometry of this
complex. KAVE has been defined in eq 1. Therefore,
∆G° can be defined as in eq 2.
AVE
F IGURE 13. Binding isotherms for (i) complexation of Bis-
DBA-F 42+ by the BMP 25C8 CLUSTER and (ii) complexation
of Bis-DBA-F 4 by BMP 25C8. (a) The values represented
∆G° ) -RT lnKAVE
(2)
AVE
2+
on this axis represent the measured concentration of com-
plexed dialkylammonium centers divided by the total concen-
tration of crown ether in solution. (b) The values represented
on this axis represent the total concentration of dialkylam-
monium cations in solution. Error bars are set as defined in
Figure 7.
For a polyvalent interaction, the total free energy change
can be defined2 by eq 3
∆GP°OLY ) N∆G°
(3)
AVE
where N can be considered as the number of individual
component interactions that constitute the multivalent
interaction. In the case of the interaction where one bis-
DBA2+ dication is complexed by one CLUSTER, two
NH2+ centers are simultaneously complexed by two crown
+
good evidence that both NH2 sites on the dication are
complexed by crown ether moieties. Therefore, it follows
logically that, in the present case of the binding of bis-
2+
DBA-F 4 by the BMP 25C8 CLUSTER, either both
ether moieties (N ) 2). Thus, ∆G°
can be calculated
from eq 3, an equation that can also be expressed in
POLY
+
NH2 centers of this dication are complexed simulta-
neously by two crown ether moieties or none are com-
plexed at all. Making this assumption, it is possible to
calculate KAVE and, by repeating the experiment at
various concentrations of bis-DBA-F 42+, a binding iso-
therm was constructed (Figure 13). Taking the mean of
terms of KPOLY and KAVE (eq 4).
N
KPOLY ) (KAVE
)
(4)
2+
Thus, for the binding of the b is-DBA-F 2 and b is-
DBA-F42+ by the DB24C8 CLUSTER and the BMP 25C8
CLUSTER, respectively, the quantities ∆G°
the data points yields a value of KAVE ) 59 ( 18 M-1
.
and
The results from all four experiments conducted with
the dicationic threads are summarized in Table 2. In the
case of the DB24C8-based series of experiments, it
appears that the mean KAVE for the interaction of the
POLY
K
POLY can be calculated, and they are shown in Table 3.
2+
For the complexation of bis-DBA-F 2 by the DB24C8
CLUSTER, KPOLY was calculated to be 2.6 × 104 M-1
(∆G°
) -6.0 kcal mol-1), which is over 2 orders of
2+
DB24C8 CLUSTER with bis-DBA-F 2 is lower than
POLY
2+
magnitude larger than the component interaction, for
that for the interaction of DB24C8 with DBA-F 2 by
which KAVE ) 790 M-1 (∆G°
) -3.9 kcal mol-1). A
almost an order of magnitude. Additionally, the value of
the mean KAVE for the interaction of the DB24C8
CLUSTER with bis-DBA-F 22+ is approximately 5 times
smaller than the mean KAVE for the component interac-
tion of DB24C8 with DBA-F 2+. The reason for this
reduction in the mean KAVE is probably, at least in part,
a consequence of the electrostatic and steric effects
discussed previously for the complexation of monovalent
AVE
similar trend is observed for the complexation of bis-
2+
DBA-F 4 by the BMP 25C8 CLUSTE R, where KPOLY
was calculated to be 3.5 × 103 M-1 (∆G°POLY ) -4.8 kcal
mol-1), which is over 2 orders of magnitude larger than
that of the component interaction, for which KAVE ) 20
M-1 (∆G° ) -1.8 kcal mol-1). These results indicate
AVE
that significant gains in binding affinity can be achieved
within a supramolecular system by simply expressing its
component interaction in a polyvalent fashion, even when
each of the individual interactions within the polyvalent
array are weaker than the component interaction. In both
the DB24C8 and BMP 25C8 systems, these gains in
binding affinities are quite significant, especially when
one considers that the multivalent receptor molecules are
not particularly rigid, with their crown ether appendages
having a large degree of conformational freedom, which
in turn incurs a greater entropic penalty upon complex-
ation.
TABLE 2. Va lu es of Mea n KAVE a n d ∆G° for th e
AVE
In ter a ction s in CD3CN a t 300 K of DB24C8 a n d th e
2+
DB24C8 CLUSTER w ith Bis-DBA-F 2 a n d of BMP 25C8
2+
a n d th e BMP 25C8 CLUSTER w ith Bis-DBA-F 4
mean
∆G°
AVE
host
guest
KAVE (M-1 (kcal mol-1
) )
2+
2+
2+
2+
DB24C8
DB24C8 CLUSTER
BMP 25C8
Bis-DBA-F 2
Bis-DBA-F 2
Bis-DBA-F 4
1100 ( 340 -4.1 ( 0.2
160 ( 48
30 ( 9
-3.0 ( 0.2
-1.8 ( 0.2
-2.4 ( 0.2
BMP 25C8 CLUSTER Bis-DBA-F 4
59 ( 18
+
+
threads DBA-F 2 and DBA-F 4 by the clusters. In the
case of the BMP 25C8-based series of experiments, the
value of the mean KAVE for the interaction of the
BMP 25C8 CLUSTER with bis-DBA-F 42+ is about twice
Con clu sion s
The syntheses of two analogous multivalent receptors,
the DB24C8 CLUSTER and the BMP 25C8 CLUSTERs
7978 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
TABLE 3. Ca lcu la ted Va lu es of KP OLY a n d ∆G°
for th e In ter a ction s in CD3CN a t 300 K of DB24C8 a n d th e DB24C8
CLUSTER w ith Bis-DBA-F 2 a n d of BMP 25C8 a n d th e BMP 25C8 CLUSTER w ith Bis-DBA-F 4
P OLY
2+
2+
mean
∆G°
KPOLY
(M-1
∆G°
POLY
AVE
host
guest
KAVE (M-1
)
(kcal mol-1
)
)
(kcal mol-1
)
+
DB24C8
DB24C8 CLUSTER
BMP 25C8
DBA-F 4
790 ( 240
160 ( 48
20 ( 6
-3.9 ( 0.2
-3.0 ( 0.2
-1.8 ( 0.2
-2.4 ( 0.2
2.6((0.8) × 104
3.5((1.0) × 103
-6.0 ( 0.2
-4.8 ( 0.2
2+
2+
Bis-DBA-F 2
+
DBA-F 4
BMP 25C8 CLUSTER
Bis-DBA-F 4
59 ( 18
each containing seven crown ether moieties attached to
a â-CD coreshave been accomplished. A series of comple-
mentary mono- and divalent dialkylammonium ligandss
containing judiciously sited fluorine atom substituents39s
was also prepared. These multivalent receptors and
mono- and divalent ligands were used to study certain
aspects of polyvalency, namely, statistical and chelate
effects.
Studies of the interaction of both the DB24C8 CLUS-
TER and the BMP 25C8 CLUSTER with their corre-
sponding monovalent dialkylammonium ligands led to
the finding that the crown ethers within the multivalent
receptors were poorer hosts for dialkylammonium cat-
ions, when compared to the parent systems. The reason
for this reduced binding by the multivalent receptors is
probably a consequence of electrostatic repulsion and
steric effects, both of which result from the build-up of
ligands and their associated PF6- counterions within the
complex.40
The abilities of the multivalent receptors, the DB24C8
CLUSTER and the BMP 25C8 CLUSTER, to bind their
corresponding divalent dialkylammonium ligands were
investigated in an attempt to study the contributions of
chelate effects to multivalency. In the case of the DB24C8
CLUSTER, we found that the KAVE was lower than that
of the component interaction, i.e., the crown ether
moieties within this receptor bind a single dialkylam-
monium centerscontained within the divalent ligands
less effectively than that of a single crown ether moiety
binding a monovalent dialkylammonium cation. The
opposite effect was found in the case of the BMP 25C8
CLUSTER binding its corresponding divalent ligand. For
both multivalent receptors, however, we found that KPOLY
was approximately 2 orders of magnitude larger than
their respective component interactions.
appears that expressing a recognition motif in a polyva-
lent fashion leads to high binding affinities, even when
each interaction within the polyvalent assembly is not
as strong as the component interaction. In other words,
the quantity of binding interactions, as opposed to the
quality, can be an effective strategy to adopt when
seeking ways to construct supramolecular assemblies
where each subunit is noncovalently bonded to another
subunit with a strong affinity. The results of this initial
study into polyvalency in a quite complicated unnatural
system suggest that a more systematic study of polyva-
lency, whereby structurally simple di- and trivalent
receptors and ligands are made by attaching crown ether
and dialkylammonium moieties to a simple scaffold,
should be undertaken before any detailed analysis of the
factors at work in this system is done. Modifying the
structures of the receptors and ligands by, for example,
changing the rigidity of the scaffold onto which the
receptor and ligand moieties are attached could yield
valuable information about the subtle interplay between
entropy and binding affinities, which is probably the most
important and least understood aspect in the study of
polyvalency.
Exp er im en ta l Section
Gen er a l. Compounds 1,17 2,18 6,22 DBA-F 2‚P F 6,22 and DBA-
F 4‚P F 622 were prepared according to published procedures. All
solvents were used as purchased except for CH2Cl2, which was
distilled from CaH2, and THF, which was distilled from Na/
benzophenone. Thin-layer chromatography (TLC) was carried
out on aluminum sheets precoated with silica gel 60 F. The
plates were examined under UV light or developed by charring
with 5% H2SO4 in EtOH. Column chromatography was carried
out using silica gel 60 F (230-400 mesh). Gel filtration
chromatography was performed using a column packed with
Sephadex LH-20 (3 cm × 90 cm) eluting with either MeOH or
CHCl3 at 1 mL min-1. In both cases, the eluant was monitored
with a refractive index detector. Fast atom bombardment mass
spectra (FAB-MS) were obtained from a mass spectrometer.
Matrix-assisted, laser-desorption-ionization, time-of-flight
mass spectra (MALDI-TOF-MS) were recorded using a trans-
indole acrylic acid matrix and an average of 50 laser shots per
sample. 1H NMR spectra were recorded at 200, 300, 400 or
500 MHz with either the residual solvent or external TMS (for
1H NMR) as calibrants. 13C NMR spectra were recorded at 75,
100 or 125 MHz. 19F NMR spectra were recorded at 376 MHz
using C6F6 as internal standard. The chemical shifts are
expressed in ppm and the coupling constants from the 1H NMR
spectra in hertz (Hz) and are within a ca. (0.5 Hz error range.
The following abbreviations are used to explain the multiplici-
ties: s, singlet; br s, broad singlet; br m, broad multiplet; d,
doublet; t, triplet; m, multiplet; dd, doublet of doublets.
Microanalyses were performed by Quantitative Technologies
Inc., Whitehouse, NJ .
These results indicate that relatively high values of
KPOLY can be obtained between two moieties simply by
constructing receptors and ligands that express compo-
nent recognition motifs in a polyvalent fashion.41 It
(39) 19F NMR spectroscopy was found to be an invaluable tool in
the study of polyvalency in this unnatural system, allowing quantita-
tive data to be obtained from binding experiments, which would not
be possible by utilizing 1H NMR spectroscopic techniques.
(40) It should be noted, however, that great care must be taken when
making generalizations on the importance of statistical contributions
to polyvalency based on the results reported here. The component
interaction used in this study involves positively charged ligands and
neutral receptors, which leads to complexes with resulting overall
positive charge, a factor that is detrimental to the ability of the complex
to further bind ligands. This phenomenon would not occur with a
polyvalent system in which both the receptor and ligand, and their
resulting complexes, have no overall charge.
(41) We do not wish to imply that it is easy to achieve significant
enhancements in binding in multivalent systems over monomeric ones
merely by force of numbers alone. These systems must still be designed
such that recognition sites are ligated optimally, which means that,
among other considerations, spacer units, linkers, and hubs must be
complementary, relatively conformationally rigid, and not too sterically
bulky so as to disturb the molecular recognition event.
(2-Hyd r oxym eth yl)d iben zo[24]cr ow n -8 (3). To a solu-
tion of the aldehyde 2 (2.5 g, 5.25 mmol) in dry THF (100 mL)
was added carefully LiAlH4 (300 mg, 7.9 mmol) portionwise.
The resulting suspension was left to stir under an atmosphere
J . Org. Chem, Vol. 67, No. 23, 2002 7979

Fulton et al.
of Ar at room temperature for 1 h. The reaction mixture was
quenched by the dropwise addition of H2O (ca. 1 mL) to the
reaction, and MgSO4 (ca. 1 g) then added to remove unreacted
H2O from the reaction mixture. The reaction mixture was
filtered and evaporated to dryness to afford the alcohol 3 as a
white solid (2.35 g, 96%). 1H NMR (300 MHz, CDCl3): δ )
3.82 (8H, s), 3.89-3.92 (8H, m), 4.12-4.15 (8H, m), 4.75 (2H,
s), 6.83-6.90 (7H, m). 13C NMR (75 MHz, CDCl3): δ ) 65.2,
69.4, 69.9, 71.3, 112.9, 113.8, 114.0, 119.9, 121.4, 134.2, 148.9.
FAB-MS: m/z 478 [M]+. Anal. Calcd for C25H34O9: C, 62.75;
H, 7.16. Found: C, 62.93; H, 7.02. Spectral data are consistent
with those reported previously.19
dissolved in EtOAc (30 mL) and washed with 1 M HCl (30
mL), saturated NaHCO3 (30 mL), and brine (30 mL). The
organic layer was then dried (MgSO4), filtered, and evaporated
to dryness to afford a clear oil. Purification by gel filtration
chromatography (LH-20, 1:1 MeOH/CHCl3) afforded the title
compound (26 mg, 37%). 1H NMR (500 MHz, CDCl3): δ )
2.43-2.58 (14H, br m), 2.61-2.76 (14H, br m), 3.15 (7H, d, J
) 7.0 Hz), 3.27 (7H, br m), 3.78-3.84 (56H, m), 3.85-3.93
(70H, m), 4.07-4.17 (56H, m), 5.07 (7H, br s), 6.71-6.81 (21H,
m), 6.83-6.88 (28H, m). 13C NMR (125 MHz, CDCl3): δ ) 26.3,
29.6, 35.9, 46.2, 58.5, 61.1, 69.3, 69.7, 70.5, 71.0, 80.9, 81.9,
82.7, 99.2, 113.7, 114.0, 114.2, 121.4, 121.6, 131.2, 147.7, 148.8,
172.0. MALDI-TOF: m/z 5188 [M + Na]+. Anal. Calcd for
2-(Meth a n esu lfon yloxym eth yl)d iben zo[24]cr ow n -8 (4).
A stirred solution of the alcohol 3 (315 mg, 0.66 mmol) and
Et3N (552 µL, 3.96 mmol) in dry CH2Cl2 (30 mL) under an
atmosphere of Ar was cooled to 0 °C on an ice bath. MsCl (204
µL, 2.64 mmol) was added dropwise and the solution left to
stir at room temperature for 4 h. H2O (30 mL) was added and
the mixture stirred for 10 min, before being transferred to a
separating funnel. The organic layer was dried (MgSO4),
filtered, and evaporated to dryness to afford a clear oil.
Purification by column chromatography (SiO2, EtOAc) afforded
the pure mesylate 4 as an oil, which slowly solidified on
standing (260 mg, 71%). 1H NMR (500 MHz, CDCl3): δ ) 3.05
(3H, s), 3.77 (8H, s), 3.84-3.87 (8H, m), 4.08-4.11 (8H, m),
4.48 (2H, s), 6.75-6.85 (7H, m). 13C NMR (125 MHz, CDCl3):
δ ) 31.6, 46.5, 69.27, 69.35, 69.36, 69.74, 69.75, 69.85, 71.17,
71.20, 113.5, 114.0, 114.3, 121.4, 121.7, 139.0, 148.9, 149.0.
FAB-MS: m/z 556 [M]+. Anal. Calcd for C26H36O11S: C, 56.11;
H, 6.52. Found: C, 55.89; H, 6.62.
C
252H357O91N7S7 ‚2H2O: C, 58.19; H, 6.99. Found: C, 58.52;
H, 7.13.
Ben zo(5-m eth a n esu lfon yloxym eth ylm eta p h en ylen e)-
[25]cr ow n -8 (7). A stirred solution of the alcohol 6 (1.013 g,
2.13 mmol) and Et3N (880 µL, 6.32 mmol) in CHCl3 (40 mL)
was cooled to 0 °C. MsCl was then added dropwise (333 µL,
4.32 mmol) and the reaction was stirred for 10 min and then
quenched with H2O (40 mL). The mixture was transferred to
a separating funnel, and the organic layer washed with H2O
(2 × 40 mL), dried (MgSO4), and evaporated to dryness to
afford the mesylate 7 as a clear oil which solidified upon
standing (1.166 g, 99%). 1H NMR (200 MHz, CDCl3): δ ) 2.84
(3H, s), 3.65 (8H, s), 3.73-3.80 (8H, m), 4.05-4.10 (8H, m),
5.05 (2H, s), 6.47 (2H, d, J ) 2.2 Hz), 6.66 (1H, t, J ) 2.2 Hz),
6.83 (4H, s). 13C NMR (125 MHz, CDCl3): δ ) 38.3, 68.1, 68.8,
69.8, 70.88, 70.93, 71.3, 103.1, 108.1, 115.1, 121.6, 135.1, 148.9,
160.3. FAB-MS: m/z ) 556 [M]+, 461 [M - SO2CH3]+. Anal.
Calcd for C26H36O11S: C, 56.11; H, 6.52. Found: C, 55.92; H,
6.36.
5-(2-(Met h oxyca r b on yl)et h ylt h iom et h yl)-1,3-p h en yl-
en eben zo[25]cr ow n -8 (8). To a stirred suspension of the
mesylate 7 (960 mg, 1.73 mmol), K2CO3 (2.38 g, 17.2 mmol),
and KI (10 mg) in DMF (30 mL) was added methyl 3-mercap-
topropionate (1.02 g, 8.5 mmol). The suspension was stirred
under an atmosphere of Ar at room temperature for 3 d. The
reaction mixture was evaporated to near dryness and the
residue partitioned between 1 M HCl (50 mL) and EtOAc (50
mL). The organic layer was washed with brine (50 mL), dried
(MgSO4), and evaporated to dryness. Purification by column
chromatography (SiO2; hexanes/EtOAc, 3:7) afforded the ester
8 as a clear oil which solidified on standing (839 mg, 83%). 1H
NMR (500 MHz, CDCl3): δ ) 2.50 (2H, t, J ) 7.2 Hz), 2.63
(2H, t, J ) 7.2 Hz), 3.56 (2H, s), 3.61 (3H, s), 3.65-3.69 (8H,
m), 3.75-3.81 (8H, m), 4.08-4.10 (8H, m), 6.43 (2H, d, J )
2.0 Hz), 6.54 (1H, t, J ) 2.0 Hz), 6.82-6.87 (4H, m). 13C NMR
(125 MHz, CDCl3): δ ) 25.9, 33.9, 36.1, 51.4, 67.8, 68.7, 69.6,
70.66, 70.70, 101.0, 108.14, 114.9, 121.4, 139.8, 148.8, 159.8,
172.0. FAB-MS: m/z 580 [M]+. Anal. Calcd for C29H40O10S: C,
59.98; H, 6.94. Found: C, 60.21; H, 7.06.
5-(2-(H yd r oxyca r b on yl)et h ylt h iom et h yl)-1,3-p h en yl-
en eben zo[25]cr ow n -8 (9). A mixture of the ester 8 (830 mg,
1.43 mmol), NaOH (251 mg, 6.3 mmol), H2O (2.8 mL), and THF
(25 mL) was stirred vigorously for 22 h. EtOH (5 mL) was
added to the mixture, and Amberlite resin IR-120 (H+ form)
was added until the pH reached ∼4. The mixture was filtered,
and the filtrate evaporated to dryness. The resulting residue
was redissolved in CHCl3 and dried (MgSO4). The organic layer
was collected and evaporated to dryness to afford the acid 9
as an off-white solid (975 mg, 93%). 1H NMR (500 MHz,
CDCl3): δ ) 2.52 (2H, t, J ) 7.0 Hz), 2.64 (2H, t, J ) 7.0 Hz),
3.69-3.73 (8H, m), 3.79-3.81 (4H, m), 3.83-3.84 (4H, m),
4.10-4.14 (8H, m), 6.46 (2H, d, J ) 2.2 Hz), 6.57 (1H, t, J )
2.2 Hz), 6.85-6.91 (4H, m). 13C NMR (125 MHz, CDCl3): δ )
25.7, 34.0, 36.2, 67.8, 68.8, 69.6, 70.72, 70.74, 101.2, 108.2,
115.0, 121.5, 139.9, 148.8, 159.7, 159.8, 176.3. FAB-MS: m/z
566 [M]+. Anal. Calcd for C28H38O10S: C, 59.35; H, 6.76.
Found: C, 59.53; H, 6.62.
2-(2-(Hyd r oxyca r bon yl)eth ylth iom eth yl)d iben zo[24]-
cr ow n -8 (5). To a stirred suspension of the mesylate 4 (698
mg, 1.25 mmol), K2CO3 (1.77 g, 12.8 mmol), and KI (10 mg) in
DMF (20 mL) was added methyl 3-mercaptopropionate (0.8 g,
6 mmol). The suspension was stirred under an atmosphere of
Ar at room temperature for 3 d. The reaction mixture was then
evaporated to near dryness and the residue partitioned
between H2O (30 mL) and EtOAc (30 mL). HCl (2 M) was
added dropwise and the mixture briefly shaken. This procedure
was repeated until the pH of the water layer was ∼4. The
organic layer was then dried (MgSO4), filtered, and evaporated
to dryness to afford the crude ester (990 mg), which was used
without further purification in the next reaction. 1H NMR (400
MHz, CDCl3): δ ) 2.72 (2H, t, J ) 7.0 Hz), 2.90 (2H, t, J )
7.0 Hz), 3.65 (2H, s), 3.67 (3H, s), 3.80 (8H, s), 3.86-3.90 (8H,
m), 4.11-4.13 (8H, m), 6.74-6.87 (7H, m). 13C NMR (100 MHz,
CDCl3): δ ) 26.2, 34.5, 36.1, 69.5, 69.6, 70.00, 70.02, 71.4,
113.8, 114.2, 114.5, 121.6, 131.1, 148.1, 149.0, 172.2. FAB-
MS: m/z 603 [M + Na]+. Anal. Calcd for C29H40O10S: C, 59.98;
H, 6.94. Found: C, 60.14; H, 7.01. A mixture of crude ester
(990 mg), 1 M NaOH (13 mL), and tetrabutylammonium iodide
(10 mg) in THF (20 mL) was vigorously stirred for 24 h. HCl
(2 M) was then added dropwise until the pH of the mixture
was ∼4, and the mixture evaporated to dryness. The resulting
residue was then partitioned between H2O (30 mL) and EtOAc
(30 mL). The organic layer was dried (MgSO4), filtered, and
evaporated to dryness to afford the acid 5 as an oil that
1
solidified upon standing (827 mg, 86% from 4). H NMR (500
MHz, CDCl3): δ ) 2.52 (2H, t, J ) 7.0 Hz), 2.64 (2H, t, J )
7.0 Hz), 3.65 (2H, s), 3.83 (8H, s), 3.89-3.91 (8H, m), 4.11-
4.16 (8H, m), 6.76-6.89 (7H, m). 13C NMR (125 MHz, CDCl3):
δ ) 25.8, 34.2, 36.0, 69.3, 69.4, 69.80, 69.83, 71.1, 113.7, 114.0,
114.3, 121.4, 121.5, 130.9, 147.9, 148.81, 148.89, 176.3. FAB-
MS: m/z 589 [M + Na]+, 566 [M]+. Anal. Calcd for C28H38
-
O
10S: C, 59.35; H, 6.76. Found: C, 59.03; H, 6.62.
DB24C8 CLUSTER. To a stirred solution of the acid 5 (116
mg, 0.20 mmol) in DMF (10 mL) was added HBTU‚BF4 (86
mg, 0.23 mmol) and the reaction stirred at room temperature
under an atmosphere of Ar for 2 h. i-Pr2NEt (40 µL, 0.23 mmol)
and the heptakisamino-â-CD derivative 1 (22 mg, 14 µmol)
were added, and the reaction stirred for a further 3 d. The
mixture was evaporated to dryness, and the residue was
BMP 25C8 CLUSTER. To a stirred solution of the acid 9
(159 mg, 0.28 mmol) in DMF (10 mL) was added HBTU‚BF4
7980 J . Org. Chem., Vol. 67, No. 23, 2002

Probing Polyvalency in Artificial Systems
(130 mg, 0.34 mmol) and the reaction stirred at room temper-
ature under an atmosphere of Ar for 2 h. i-Pr2NEt (65 µL, 0.38
mmol) and the heptakisamino-â-CD derivative 1 (36 mg, 23
µmol) were then added, and the reaction stirred for a further
3 d. The reaction was evaporated to dryness, and the residue
was dissolved in EtOAc (50 mL) and washed with 1 M HCl
(50 mL), saturated NaHCO3 (50 mL), and brine (50 mL). The
organic layer was then dried (MgSO4) and evaporated to
dryness to afford a clear oil. Purification by gel filtration
chromatography (LH-20, CHCl3) afforded the title compound
(43 mg, 36%). 1H NMR (500 MHz, CDCl3): δ ) 2.44-2.53 (14H,
m), 2.64-2.77 (14H, m), 3.14 (7H, dd, J ) 2.9 Hz, 9.4 Hz),
3.31 (7H, t, J ) 8.7 Hz), 3.47 (21H, br s), 3.46-3.51 (7H, m),
3.59 (35H, br s), 3.66-3.74 (63H, m), 3.73-3.77 (28H, m),
3.82-3.86 (42H, m), 4.08-4.15 (56H, m), 5.07 (7H, d, J ) 2.9
Hz), 6.43-6.48 (14H, br m), 6.53-6.57 (7H, br m), 6.86-6.91
(28H, m). 13C NMR (125 MHz, CDCl3): δ ) 25.1, 34.5, 36.2,
40.1, 58.5, 61.0, 67.9, 68.9, 69.7, 70.4, 70.8, 80.8, 81.9, 82.6,
99.0, 101.1, 108.4, 115.0, 121.5, 140.3, 148.9, 159.8, 171.9.
MALDI-TOF: m/z 5185 [M + Na]+. Anal. Calcd for C252H357O91-
N7S7‚2H2O: C, 58.19; H, 6.99. Found: C, 58.52; H, 7.30.
r,r′-Bis(4-flu or oben zyl)a m m on iu m -p-xylen e Bis(h exa -
flu or op h osp h a te) (Bis-DBA-F 22+). A solution of 4-fluoroben-
zylamine (11) (821 mg, 6.57 mmol) and terephthaldicarbox-
aldehyde (14) (436 mg, 3.25 mmol) in PhMe (50 mL) was
heated under reflux for 20 h using a Dean-Stark apparatus.
The resulting solution was evaporated to dryness, the residue
was dissolved in dry MeOH (50 mL), NaBH4 (1.24 g, 32.63
mmol) was added portionwise over a period of 10 min, and
the reaction was left to stir at room temperature overnight.
The reaction mixture was evaporated to dryness, and 2 M HCl
(50 mL) was added to the residue. The resulting suspension
was washed with CHCl3 (50 mL), and 2 M NaOH solution was
added to the aqueous layer until the pH reached ∼14. The
aqueous layer was extracted with CHCl3 (5 × 50 mL), and the
organic extracts were combined, dried (MgSO4), and evapo-
rated to dryness. The resulting oil was dissolved in MeOH (25
mL), and saturated methanolic NH4PF6 solution was added
dropwise until the precipitation of white solids ceased. The
solids were filtered and dried to afford the title compound as
r,r′-Bis(3,5-d iflu or oben zyl)a m m on iu m -p-xylen e Bis-
(h exa flu or op h osp h a te) (Bis-DBA-F 42+). A solution of 3,5-
difluorobenzylamine (13) (959 mg, 6.70 mmol) and terephthal-
dicarboxaldehyde (14) (447 mg, 3.33 mmol) in PhMe (50 mL)
was heated under reflux for 20 h using a Dean-Stark
apparatus. The resulting solution was evaporated to dryness,
the residue was dissolved in dry MeOH (50 mL), NaBH4 (2.53
g, 66.6 mmol) was added portionwise over a period of 10 min,
and the reaction was stirred at room temperature overnight.
The reaction mixture was worked up using a procedure similar
2+
to that described above for bis-DBA-F 2 to afford the title
compound as a white powder (1.34 g, 61%). 1H NMR (400 MHz,
CD3CN): δ ) 4.24 (4H, s), 4.25 (4H, s), 7.04-7.12 (10H, m),
7.56 (4H, s). 19F NMR (376 MHz, CD3CN): δ ) -72.3 (6F, d,
J ) 707 Hz), -110.4 (4F, s). 13C NMR (100 MHz, CD3CN): δ
) 51.4, 52.0, 114.3, 114.5, 118.8, 131.9, 132.8, 134.9. FAB-
MS: m/z ) 535 [M - PF6]+. Anal. Calcd for C22H22F16N2P2:
C, 38.84; H, 3.26. Found: C, 38.73; H, 3.30.
P r ep a r a tion of Solu tion s for NMR Stu d ies. A stock
+
solution of DBA-F 2 was prepared by dissolving this salt in
an appropriate volume of CD3CN, as measured by a Gilson
pipet, such that the concentration of the solution was 35 mM.
A 14 mM stock solution of DB24C8 in CD3CN was prepared
in the same manner. Solutions for NMR spectroscopic analysis
were then prepared as follows. To a small vial were added,
with an appropriate Gilson pipet, known volumes of the stock
+
solutions of DBA-F 2 and DB24C8, and the volume of CD3-
CN was adjusted such that the final concentrations of the two
components were as desired. After careful mixing, the resulting
solution was transferred to an NMR tube and analyzed
immediately. Solutions were discarded after use. This proce-
dure was repeated to prepare samples corresponding to all
other points on the binding isotherm, using the same stock
solutions. For further experiments involving other dialkylam-
monium salts or crown ethers/crown ether clusters, fresh stock
solutions of suitable concentrations were prepared, and the
solutions were mixed as described using appropriate volumes
of solutions and then diluted with CD3CN so that the final
solution contained both host and guest at the desired concen-
trations.
1
a white powder (1.19 g, 62%). H NMR (400 MHz, CD3CN): δ
Ack n ow led gm en t. This research was supported by
the National Science Foundation (CHE-9910199, and
equipment grant CHE-9974928). We thank Dr. Peter
T. Glink for all his considerable help in producing the
final version of this manuscript.
) 4.34 (4H, s), 4.35 (4H, s), 7.10 (4H, br s), 7.28-7.33 (4H,
m), 7.58-7.62 (4H, m), 7.61 (4H, s). 19F NMR (376 MHz, CD3-
CN): δ ) -75.3 (6F, d, J ) 707 Hz), -113.5 (2F, s). 13C NMR
(100 MHz, CD3CN): δ ) 51.77, 51.85, 116.8, 117.0, 131.8,
132.9, 133.7, 133.8. FAB-MS: m/z ) 500 [M - PF6]+. Anal.
Calcd for C22H24F14N2P2: C, 41.01; H, 3.75. Found: C, 41.19;
H, 3.98.
J O0110799
J . Org. Chem, Vol. 67, No. 23, 2002 7981