Optimization of a synthetic arginine receptor. Systematic tuning of noncovalent interactions

Article abstract of DOI:10.1021/jo0156161

The simple arginine binder 1 could be optimized by strengthening π-cation as well as electrostatic interactions. Electron-donating or -withdrawing substituents in the 5-position provide experimental evidence for π-cation interactions, because binding energies increase by up to 0.6 kcal/mol due to a single benzene-guanidinium interaction. Even more effective is the introduction of a third phosphonate functionality at the correct distance, so that the guanidinium cation is recognized by optimal electrostatic and hydrogen bond interactions. Monte Carlo simulations and NOESY experiments confirm the expected complex geometries. The optimized host molecule 8 binds arginine half an order of magnitude more efficiently than the parent molecule.

Full text of DOI:10.1021/jo0156161

5814
J . Org. Chem. 2001, 66, 5814-5821
Op tim iza tion of a Syn th etic Ar gin in e Recep tor . System a tic Tu n in g
of Non cova len t In ter a ction s
Stephan Rensing,† Markus Arendt,† Andreas Springer,† Thomas Grawe,‡ and
Thomas Schrader*,†
Fachbereich Chemie, Philipps-Universita¨t Marburg, Hans-Meerwein-Strasse,
D-35032 Marburg, Germany, and Institut fu¨r Organische Chemie und Makromolekulare Chemie,
Heinrich-Heine-Universita¨t Du¨sseldorf, Universita¨tsstrasse 1, D-40225 Du¨sseldorf, Germany
schradet@mailer.uni-marburg.de
Received March 6, 2001
The simple arginine binder 1 could be optimized by strengthening π-cation as well as electrostatic
interactions. Electron-donating or -withdrawing substituents in the 5-position provide experimental
evidence for π-cation interactions, because binding energies increase by up to 0.6 kcal/mol due to
a single benzene-guanidinium interaction. Even more effective is the introduction of a third
phosphonate functionality at the correct distance, so that the guanidinium cation is recognized by
optimal electrostatic and hydrogen bond interactions. Monte Carlo simulations and NOESY
experiments confirm the expected complex geometries. The optimized host molecule 8 binds arginine
half an order of magnitude more efficiently than the parent molecule.
In tr od u ction
The molecular recognition of arginine is a key process
in many vital biological control mechanisms. Thrombine
and trypsin cleave peptides directly after an arginine
residue.1 The highly efficient self-organization of the
extremely long DNA double helix in nucleosomes strongly
relies on electrostatic contacts between the negatively
charged phosphate moieties and positively charged basic
amino acids, when it winds around the histone proteins.2
Arginine-rich regions in regulatory proteins often bind
sequence-specifically to the phosphodiester backbone of
RNA and DNA molecules.3
F igu r e 1. Benzylic bisphosphonate host molecules 1 and 2
for strong guanidinium recognition.
Despite their fundamental importance in Nature, these
arginine recognition events have only recently been
imitated by chemists with artificial receptor molecules.
Dougherty and Bell have recently presented two powerful
arginine receptor molecules.4 However, these host mol-
ecules are relatively large and only accessible by a
multistep synthesis. Back in 1998, we introduced the
small, readily available benzylic bisphosphonate 1 as a
new binding site for the guanidinium cation.5 Molecular
tweezers with a sulfonate hinge group 2 are even capable
of rotating their phosphonate anions into the same plane
as the guanidinium cation.6
1). However, the respective binding constants are only
high in dipolar aprotic solvents such as DMSO.5-7 We
asked ourselves, which structural changes could be made
in order to optimize the recognition of arginine deriva-
tives by these small tweezer molecules. To achieve this
goal, we strived to better understand the nature and force
of noncovalent interactions involved in the binding
process. In most cases, it is very difficult to quantify the
contribution of single noncovalent interactions to a host-
guest complex with multiple binding sites.8 Since host
molecule 1 is very small, it represents a good candidate
for the systematic probing of specific noncovalent attrac-
tive forces. To this end, we introduced an additional
substituent in the 5-position of the phosphonate-carrying
benzene ring and systematically varied its electronic and
stereochemical properties to enforce and complement the
existing noncovalent interactions.
This favorable arrangement makes them highly selec-
tive for guanidinium ions over ammonium ions and
especially well suited for the molecular recognition of
arginine derivatives in a peptidic environment (Figure
* To whom correspondence should be adressed. Fax: + 49 6421
2828917. E-mail: schradet@mailer.uni-marburg.de.
† Philipps-Universita¨t Marburg.
‡ Heinrich-Heine-Universita¨t Du¨sseldorf.
(7) For an overview, see: T. Schrader, J . Inclusion Phenom. Macro-
cycl. Chem. 1999, 34, 117.
(1) The Enzymes, Vol. III: Hydrolysis: Peptide Bond; Boyer, Ed.;
Academic Press: New York/London, 1971.
(2) Arents, G.; Moudrianakis, E. N. Proc. Natl. Acad. Sci. U.S.A.
1993, 90, 10489-10493.
(8) For a guanidinium-based carboxylate receptor, this analysis has
recently been carried out: Schmuck, C. Chem. Eur. J . 2000, 4, 709-
718. A concept based on predictable incremental contributions of single
noncovalent interactions was introduced by Schneider and applied to
numerous examples: Schneider, H.-J .; Schiestel, T.; Zimmermann, P.
J . Am. Chem. Soc. 1992, 114, 7698-7703; Schneider, H.-J .; Ru¨diger,
V.; Raevsky, O. A. J . Org. Chem. 1993, 58, 3648-3653. Sartorius, J .;
Schneider, H.-J . Chem. Eur. J . 1996, 2, 1446-1452. Schneider, H.-J .
Angew. Chem., Int. Ed. Engl. 1997, 36, 1072-1073. Eblinger, F.;
Schneider, H.-J . Angew. Chem., Int. Ed. 1998, 37, 826.
(3) Ptashne, M.; Gann, A. Nature 1997, 386, 569.
(4) (a) Bell, T. W.; Khasanov, A. B.; Drew, M. G. B.; Filikov, A.;
J ames, T. L. Angew. Chem., Int. Ed. 1999, 38, 2543-2547. (b) Ngola,
S. M.; Kearney, P. C.; Mecozzi, S.; Russell, K.; Dougherty, D. A. J .
Am. Chem. Soc. 1999, 121, 1192-1201.
(5) Schrader, T. Tetrahedron Lett. 1998, 39, 517-520.
(6) Schrader, T. Chem. Eur. J . 1997, 3, 1537.
10.1021/jo0156161 CCC: $20.00 © 2001 American Chemical Society
Published on Web 07/28/2001

Optimization of a Synthetic Arginine Receptor
J . Org. Chem., Vol. 66, No. 17, 2001 5815
F igu r e 2. Modified m-xylylene bisphosphonate host molecules 3-5 for optimized guanidinium recognition: probing the π-cation
interaction.
F igu r e 3. Modified m-xylylene bisphosphonate host molecules 6-8 for optimized arginine recognition: introducing an additional
negative charge.
F igu r e 4. Dependence of the change in chemical shift of selected NH- and CH-protons of N-tosylarginine methyl ester hydrochloride
(in bold) on addition of the basic bisphosphonate derivative 1 in DMSO at 20 °C.
This protocol produced indeed much more powerful
host systems in several cases.
postulated.1,2 For a direct proof, small electron-withdraw-
ing as well as -donating substituents were placed in the
5-position of the central benzene ring (compounds 3 and
4, Figure 2). Furthermore, we enlarged the available
π-face (host molecule 5, Figure 2). With these modifica-
tions, only π-cation interactions would furnish signifi-
cantly altered binding constants in a predictable man-
ner.9 A carboxylate group in the 5-position was intended
to increase the negative charge density at this place and
possibly create an additional hydrogen bond if a slight
rotation out of the aromatic plane would be energetically
tolerable (receptor molecule 6, Figure 3). Even better
would be the placement of a phosphonate moiety in this
region, connected to the aromatic core unit by way of a
fairly rigid spacer. Thus, according to molecular modeling
simulations10 the fifth guanidinium NH-proton could be
reached for maximum hydrogen bond contact. Simulta-
We assumed that binding of the guanidinium ion
occurs mainly through three kinds of noncovalent inter-
actions, i.e., electrostatic attraction, hydrogen bonding,
and π-cation interaction. From preliminary experiments,
we knew that in general for the simple xylylene bis-
phosphonates a 100-fold drop was observed in the as-
sociation constants if the solvent was changed from
DMSO to methanol. This is a strong indication for the
dominating influence of electrostatic attractions and the
absence of significant hydrophobic forces. However, in
DMSO strong ionic hydrogen bonds are observed as the
large downfield shifts of all guanidinium NH-protons
demonstrate (Figure 4). Furthermore, a seizable down-
field shift of the amidic NH proton in arginine derivatives
indicates an additional intermolecular hydrogen bond to
one of the phosphonate anions. It correlates well with a
marked increase in Ka of up to 1 order of magnitude.
(9) One could argue, that in methoxy-substituted host 4 an increase
in negative charge density close to the benzene ring would result in
additional electrostatic attraction of the guanidinium cation. This is
indeed the case, as MEP calculations show (vide infra).
Resu lts a n d Discu ssion
(10) Molecular Modeling Program: CERIUS2 from Molecular Simu-
lations Inc., Force field: Dreiding 2.21. Force-field minimizations were
initially carried out in the gas phase.
The contribution of π-cation interactions in complexes
of 1 with guanidinium cations has to date only been

5816 J . Org. Chem., Vol. 66, No. 17, 2001
Rensing et al.
Sch em e 2. Syn th esis of Mod el Host Com p ou n d 7
fr om Ben zen e Tr ica r boxylic Acid
Sch em e 3. Syn th esis of Mod el Host Com p ou n d s 6
a n d 8 In volvin g Selective Clea va ge of th e Meth yl
Ester , F ollow ed by Am id e Cou p lin g w ith BOP -Cl
F igu r e 5. Design of the new arginine receptor molecule 8.
The introduction of the third phosphonate anion at a certain
distance of the m-xylylene bisphosphonate moiety should
facilitate formation of two additional ionic hydrogen bonds with
the fifth guanidinium proton as well as the amidic NH proton
(E ) ester group).
Sch em e 1. Syn th esis of Mod el Host Com p ou n d s 1
a n d 3-5 via Dibr om in a tion a n d Ar bu zov Rea ction
of aqueous tetrabutylammonium hydroxide;15 sensitive
derivatives can also be cleaved with LiBr in dipolar
aprotic solvents (Scheme 1).16
Trisphosphonate 7 can be conveniently prepared from
benzene tricarboxylic acid via reduction of the respective
methyl ester to the benzylic alcohol, bromination, e.g.,
by PBr3, followed by the Arbuzov-step and subsequent
dealkylation, either with aqueous tetrabutylammonium
hydroxide or with lithium bromide in a dipolar aprotic
solvent (Scheme 2).17
A similar sequence starts from 3,5-dimethylbenzoic
acid and furnishes, after bromination, Arbuzov reaction,
and hydrolysis, model compound 6 in high yield (Scheme
3).18 We found, that the first two steps can be performed
in one pot, so that key intermediate 13 is now accessible
in a very economical and fast route (the published route
requires five steps instead of two). After selective hy-
drolysis of the methyl ester in 13, the corresponding
neously the third phosphonate moiety could interact with
amidic NH-protons in the peptide backbone, without the
necessity of an unfavorable folding of the arginine side-
chain. We prepared trisphosphonates 7 and 8 for this
purpose (Figure 3). With them we hoped to achieve a
much more effective and selective molecular recognition
of arginine derivatives in a peptidic environment. In
earlier experiments, a sensitive 1H NMR probe was
discovered which sheds light on the mutual arrangement
of both complex partners: a strong downfield-shift of the
arginine amide NH indicates the additional and often
cooperative interaction of a host phosphonate anion with
the peptide backbone. Usually it is accompanied by a
significant increase in association energy compared with
simple methyl guanidinium chloride (Figure 4).
In all syntheses of the above-mentioned 2nd generation
model compounds (Figure 5) the key-step consists of an
Arbuzov-reaction of m-(bisbromomethyl)arenes, similar
to the preparation of the basic xylylene bisphosphonates.
These are obtained by NBS-bromination from the respec-
tive m-dimethylarenes.11-14 The final hydrolysis is achieved
by refluxing the dialkyl esters with equimolar amounts
(11) Young, R.; Chang, C. K. J . Am. Chem. Soc. 1985, 107, 898-
909.
(12) Karlin, K. D.; Nasir, M. S.; Cohen, B. I.; Cruse, R. W.; Kaderli,
S.; Zuberbu¨hler, A. D. J . Am. Chem. Soc. 1994, 116, 1324-1336.
(13) Dixon, E. A.; Fischer, A.; Robinson, F. P. Can. J . Chem. 1981,
59, 2629-2641.
(14) Taylor, S. D.; Dinaut, A. N.; Thadani, A. N.; Huang, Z.
Tetrahedron Lett. 1996, 37, 8089-8092.
(15) Schrader, T. Angew. Chem. 1996, 108, 2816.
(16) Karaman, R.; Goldblum, A.; Breuer, E. J . Chem. Soc., Perkin
Trans. 1989, 765. H. Krawczyk, Synthetic Commun. 1997, 27, 3151-
3161.
(17) J ohnson, R. A. (Upjohn Co.) PCT Int. Appl. WO 94 07, 501 (Cl.
A61K31/66), 14 Apr 1994, US Appl. 949,738, 23 Sep 1992; J ohnson,
R. A. (Upjohn Co., USA) U.S. US 5,500,417 (Cl. 514-89; A61K31/66),
19 Mar 1996, US Appl. 949,738, 23 Sep 1992; Grawe, T.; Schrader, T.;
Gurrath, M.; Kraft, A.; Osterod, F. Org. Lett. 2000, 2, 29-32.
(18) Nakhle, B. M.; Trammell, S. A.; Sigel, K. M.; Meyer, T. J .;
Erickson, B. W. Tetrahedron 1999, 55, 2835-2846.

Optimization of a Synthetic Arginine Receptor
J . Org. Chem., Vol. 66, No. 17, 2001 5817
Ta ble 1. Associa tion Con sta n ts Ka [M-1] Mea su r ed in DMSO for th e 1:1 Com p lexa tion of Alk yl Gu a n id in iu m Der iva tives
w ith Recep tor Molecu les 1-7 by Mea n s of NMR Titr a tion Exp er im en ts
guest
1b
2
3
4
5
6
7
methylguanidine HCl 23000 ( 26% 9000 ( 49%c
14000 ( 38% 40000 ( 57%c
45000 ( 30%
3000 ( 19% 5500 ( 25%
N-benzoylarginine
ethyl ester HCl
stoichiometrya
86000 ( 28% 56000 ( 36% 50000 ( 29% 107000 ( 13% 120000 ( 42%c 5000 ( 18% 31000 ( 18%
1:1
1:1
1:1
1:1
1:1
1.6:1
1:1
a
b
Determined with J ob plots. Errors are standard deviations from the nonlinear regressions. c The large standard deviation in this
case can be explained with small chemical shift differences, which give especially large errors with high binding constants.
Ta ble 2. F r ee Bin d in g En th a lp ies ∆G [k ca l/m ol]
benzoic acid derivative can be coupled with aliphatic and
Mea su r ed in DMSO for th e 1:1 Com p lexa tion of Alk yl
aromatic amines. In our case, reaction with aminomethyl-
phosphonic acid dimethyl ester leads to intermediate 14,
which is finally cleaved mildly to the trisphosphonate
lithium salt 8.19 The coupling reaction is not trivial:
Initial attempts to activate the benzoic acid derivative
with thionyl chloride produced the corresponding acid
chloride, but no coupling occurred after addition of
primary amines and triethylamine. DCC gave no product
either, and even the highly efficient peptide coupling
reagent PyClop (trispyrrolidinochlorophosphonium hexa-
fluorophosphate) did not allow us to prepare pure cou-
pling products.20 In this case, the reaction proceeded up
to ∼50% conversion, but the crude product complexed the
byproduct trispyrrolidinophosphinoxide strongly, so that
both substances eluted together from any chromato-
graphic column, irrespective of the coupling partner.
Finally, we were successful with BOPCl, another peptide
coupling reagent, which is normally not recommended
for aromatic acids.21 In this case, however, the amide was
formed smoothly in quantitative yield, and the highly
polar ionic byproduct bis(2-oxo-3-oxazolidinyl) phosphate
could be separated from the desired end products to give
analytically pure compounds in high yields (Scheme 3).
With these new receptor molecules at hand, we carried
out numerous binding experiments in polar organic
solvents. For direct comparison, we titrated methyl
guanidine hydrochloride as well as N-benzoylarginine
ethyl ester hydrochloride with hosts 1-7 as tetrabutyl-
ammonium salts in DMSO. On the other hand, the
association constants between the highly polar lithium
salts of 1, 2, and 8 and both above-mentioned guani-
dinium derivatives were each checked in methanol. The
results of these NMR titrations are shown in Tables 1-4.
From the binding curves, association constants were
calculated by standard nonlinear regression methods
(Table 1).22 J ob plots confirmed a 1:1 stoichiometry in
almost every case.23
Gu a n id in iu m Der iva tives w ith Recep tor Molecu les 1-7
by Mea n s of NMR Titr a tion Exp er im en ts
guest
1
2
3
4
5
6
7
methylguanidine HCl 5.8 5.3 5.6 6.2 6.2 4.7 5.0
N-benzoylarginine
6.6 6.4 6.3 6.8 6.8 5.0 6.0
ethyl ester HCl
interaction operating only in 1. Here the guanidinium
cation and the receptor’s benzene ring are stacked on top
of each other with an ideal distance of ∼3.5 Å.
The additional introduction of electron-withdrawing or
-donating substituents into the core unit of 1 leads to
distinct changes in binding affinity: whereas the electron-
withdrawing nitro group in 3 lowers Ka, the electron-
donating methoxy functionality in 4 rises the binding
constant both by a factor of roughly 2 compared to 1. It
is important to note, that the phosphonate moieties are
electronically separated from the receptor’s benzene ring
by the methylene group, and both the nitro as well as
the methoxy group are geometrically locked in the
aromatic plane of the receptor molecule, so that they
cannot form hydrogen bonds to the guanidinium cation.
Hence the observed changes in binding energy must
reflect the attenuated or enlarged interactions between
the guanidinium cation and the aromatic system. Ac-
cording to the electrostatic model, the variation in ion
binding energies across a series of similar aromatics is
almost completely reflected in the electrostatic term.24
Two good qualitative indicators for the influence of
substituents on the electrostatics are found in the σmeta
Hammett parameter and the electrostatic potential above
the center of the aromatic ring.25 In our case, the strongly
positive σmeta value of 0.71 for the nitro group correlates
well with the marked decrease in binding energy be-
tween receptor molecule 1 and 3 of 0.2-0.3 kcal mol-1
(Table 2).26 Accordingly, the calculated MEP (molecular
electrostatic potential, AM1) shifts from -20 kcal/mol in
benzene to +2 kcal/mol in nitrobenzene. Contrary to
Dougherty’s results from the Na+/benzene system as well
as to its slightly negative σmeta value of -0.1, we find an
activating effect of the methoxy group on the cation-
aromatic interaction, corresponding to an increase in
binding energy of 0.2-0.4 kcal mol-1 from 1 to 4 (Table
2). A careful inspection of anisole’s potential surface,
however, offers a plausible explanation. Although the
electrostatic potential above anisole is essentially the
same as above benzene (-19 kcal/mol), a high negative
charge density resides in the neighboring O atom (MEP:
-56 kcal/mol) at the opposite side of its methyl substitu-
Entries 1 and 2 show the binding affinities of the two
basic receptor molecule structures: although in 2 the
phosphonate moieties can be directed perfectly coplanar
to the chelated guanidinium ion, its recognition by the
m-xylylene bisphosphonate is much more effective. We
attribute this superiority to the additional π-cation
(19) For multiple phosphonate dealkylations, see: Schrader, T.;
Wehner, M.; Finocchiaro, P.; Failla, S.; Consiglio, G. Org. Lett. 2000,
2, 605-608.
(20) Coste, J .; Frerot, E.; J ouin, P. J . Org. Chem. 1994, 59, 2437-
2446.
(21) Colucci, W.; Tung, R. D.; Petri, J . A.; Rich, D. H. J . Org. Chem.
1990, 55, 2895-2903.
(22) (a) Schneider, H. J .; Kramer, R.; Simova, S.; Schneider, U. J .
Am. Chem. Soc. 1988, 110, 6442. (b) Wilcox, C. S. In Frontiers in
Supramolecular Chemistry and Photochemistry; Schneider, H.-J ., Du¨rr,
H., Eds.; VCH: Weinheim, 1991; pp 123-143. (c) Macomber, R. S. J .
Chem. Educ. 1992, 69, 375-378.
(23) (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.
(24) Mecozzi, S.; West, A. P., J r.; Dougherty, D. A. J . Am. Chem.
Soc. 1996, 118, 2307-2308.
(25) Mecozzi, S.; West, A. P., J r.; Dougherty, D. A. Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 10566-10571.
(26) March, J . Advanced Organic Chemistry, 4th ed.; J ohn Wiley &
Sons: New York, 1993; p 280.

5818 J . Org. Chem., Vol. 66, No. 17, 2001
Rensing et al.
acid side chains.30 We would like to point out that our
simple binding motif shown in Figures 1-3 exactly
mimicks the preferred natural arrangement in two ways.
First, the guanidinium moiety of arginine is placed
directly on top of the receptor molecule’s benzene ring
for optimal π-cation as well as hydrophobic interactions.
Second, salt bridges and hydrogen bonds are simulta-
neously formed between the guanidinium NH-groups and
anionic binding sites which lead to enhanced lateral
recognition. The effective combination of these two
interactions is beautifully illustrated in the crystal
structure of the human growth hormone receptor, where
interdigitating protein strands are locked relative to each
other by multiple cation-π interactions and salt bridges.31
F igu r e 6. Electrostatic surface potentials (ESP) of benzene,
anisole, and nitrobenzene (AM1 calculations).27 Local MEP’s
have been calculated at the positions indicated by asterisks
and are given below the structures.
Was the introduction of negatively charged substitu-
ents in the 5-position of 1 more effective with respect to
an increased binding effectivity of the guanidinium
receptor? Entries 6 and 7 prove the opposite: In both
cases the obtained values for Ka remain disappointingly
low, much lower than even the parent compound 1. How
can this be explained? In the case of 6, σmeta Hammett
parameters and the electrostatic potential surfaces in-
dicate that no increase in cation-π attraction can be
expected when compared to the benzene case (1). Fur-
thermore, rotation of the carboxylate group out of the
aromatic plane for a potential (kinked) hydrogen bond
with the guanidinium cation is restricted because the
delocalization energy is lost. Another effect adds to this:
A careful analysis of the complex stoichiometry reveals
an average of 1.6:1. Obviously, instead of intramolecular
stabilization, the carboxylate group attracts another
guanidinium molecule, leading in part to a 2:1 complex.
Therefore, the evaluation of the obtained binding curve
according to a simple 1:1 or even 2:1 fit-procedure
produces poor results. For 7, however, we found a perfect
1:1 stoichiometry. The remarkably low binding constants
must hence result from unproductive conformations in
the trisfunctionalized host molecule. Only if each one of
the three phosphonate-carrying arms point upward, can
the host molecule seize the guanidinium cation from all
sides (Figure 7). This is certainly entropically, but also
thermodynamically, costly. Molecular modeling reveals
a strong repulsion between the phosphonate anions in
this conformation.32 We also found only a small increase
in binding energy when in DMSO the simple unsubsti-
tuted guanidinium cation was first titrated with bis-
phosphonate 1 (5.8 kcal mol-1) and then with trisphos-
phonate 7 (6.1 kcal mol-1).33
ent. This (and not the cation-π interaction) is responsible
for an additional electrostatic attraction leading to
enhanced binding of 4 (Figure 6).27
The association constant also rises if the π-face of the
receptor is enlarged as in 5: it is well-known that
π-cation as well as π-stacking interactions depend
linearly on the overall size of the interacting π-faces. In
our case, an additional term may be provided by the
higher polarizability of naphthalene’s larger surface area.
The energy gain compared to benzene is again 0.2-0.4
kcal from 1 to 5 (Table 2). We conclude, that by system-
atically altering the electrostatic potential above the
receptor’s π-face, the existence and important contribu-
tion of π-cation interactions to the overall binding energy
were proven experimentally. The difference in complex-
ation free enthalpies between guanidinium/3 and guani-
dinium/4 amounts to 0.5-0.6 kcal mol-1 (Table 2). If this
can be totally attributed to the cation-π interaction, we
have at least reached the order of magnitude found by
Schneider in systematic experiments with an increasing
number of phenyl rings: He determined that in aqueous
solution the cation-π interaction contributes ∼0.5 kcal
mol-1 per phenyl group.28 The difference in binding
energies between complexes of guanidinium ions with 1
and 2 is also in the range of 0.2-0.5 kcal mol-1 (Table
2). Contrary to 2, only flat 1 can accommodate guani-
dinum ions stacked on top of its benzene ring. Although
the association constants are almost doubled, the relative
change in free binding energy ∆G for the whole complex
is not dramatic, as can be seen from Table 2: a doubled
binding constant refers to an increase in free binding
enthalpy of only ∼7-8%. This reflects the dominating
importance of salt bridges and (in organic solvents) strong
hydrogen bonds in our guanidinium binders.
In recent years the π-cation interaction has been
established as one of the fundamental noncovalent forces
used by Nature in determining protein structures and
protein-ligand interactions.29 Extensive statistical analy-
ses of hundreds of high-resolution protein structures
proved that especially arginine is often involved in
stacking interactions with phenyl rings of aromatic amino
(30) Burley, S. K.; Petsko, G. A. FEBS Lett. 1986, 203, 139-143.
Mitchell, J . B. O.; Nandi, C. L.; McDonald, I. K.; Thornton, J .; Price,
S. L. J . Mol. Biol. 1994, 239, 315-331. Karlin, S.; Zuker, M.; Brocchieri,
L. J . Mol. Biol. 1994, 239, 227-248.
(31) de Vos, A. M.; Ultshc, M.; Kossiakoff, A. A. Science 1992, 255,
306-312.
(32) Modeling Program: MacroModel 7.0 (Schro¨dinger Inc.), 2000.
Energy-minimizations and Monte Carlo-Simulations were carried out
in chloroform with the AMBER*-force-field. The best structure from
the conformational searches was subsequently subjected to a molecular
dynamics calculation at ambient temperature for 10 ps.
(33) Grawe, T.; Schrader, T. Unpublished results. The unsubstituted
guanidinium cation CN3H6+ can form four hydrogen bonds with 1, but
six hydrogen bonds with 7. Although a highly symmetrical energeti-
cally favorable complex structure is calculated for the 1:1 complex
between the guanidinium ion and 7, it is obviously not much stronger
than the related complex with 1. In the first case, the three anionic
phosphonate groups in 7 come close to each other in order to surround
the guanidinium cation. Mutual charge repulsion and unfavorable
entropy losses could also be applicable here and explain the rather
disappointing results of the binding experiments with 7.
(27) Semiempirical calculation with AM1; color coding reaches from
red (-25 kcal mol-1) to blue (+25 kcal mol-1). We thank Prof. Dr. G.
Kla¨rner (Essen) for these calculations. Contrary to the extended
cationic system of the guanidinium cation, the small Na+-cation probe
used by Dougherty et al. does obviously not reach far enough to
experience any additional electrostatic attraction by the phenolic
oxygen.
(28) Schneider, H.-J .; Schiestel, T.; Zimmerman, P. J . Am. Chem.
Soc. 1992, 114, 7698-7703. Schneider, H.-J . Chem. Soc. Rev. 1994,
227-234.
(29) Ma, J . C.; Dougherty, D. A. Chem. Rev. 1997, 97, 1303-1324.

Optimization of a Synthetic Arginine Receptor
J . Org. Chem., Vol. 66, No. 17, 2001 5819
F igu r e 7. Proposed binding mode of hosts 6 and 7 with guanidium guests.
F igu r e 8. Proposed binding mode of host 8 with methyl guanidine hydrochloride. Left: Lewis structure; right: Monte Carlo
simulation in water.
Ta ble 3. Associa tion Con sta n ts Ka [M-1] Mea su r ed in
MeOD for th e 1:1 Com p lexa tion of Alk yl Gu a n id in iu m
Der iva tives w ith Recep tor Molecu les 1, 2, a n d 8 by
of the amide spacer when going from 7 to 8. During the
Mea n s of NMR Titr a tion Exp er im en ts
The counterproductive mutual electrostatic repulsion of
the anionic groups (as in 7!) is minimized by introduction
Monte Carlo simulations, several energetically favorable
guest
1
2
8
structures were found which mainly differ in the number
of oxygen atoms per phosphonate unit, determining the
hydrogen bond pattern in the complex. These structures
differ distinctly from those obtained in chloroform in one
respect: Only one of the negatively charged oxygen atoms
on the phosphonate anions points to the guanidinium
moiety; the other one rotates outward to gain solvation
energy with water (Figure 8). Nevertheless, five strong
hydrogen bonds are found in the minium energy struc-
tures: In addition to the oxygen anion, the ester oxygen
also participates in guanidinium binding. Here the
atomic charge is still calculated at -0.35 as opposed to
-0.65 in the anionic oxygen. This unusual binding motif
has been repeatedly found in Monte Carlo simulations
in water if multiple phosphonate anions were used to
recognize one positively charged cation (amidinium and
ammonium binding).
To confirm the contribution of all three phosphonate
arms in guanidinium binding, we performed a NOESY
experiment of the complex between methylguanidinium
chloride and 8 in methanol. In addition to relatively weak
intermolecular NOE’s between guanidium’s methyl group
and the two methyl phosphonates, both the methyl and
methylene signal of the ethyl phosphonate produced
relatively strong NOE contacts. This is in accord with
the above postulated complex structure, depicted in
Figure 9. Since both methyl phosphonates are involved
in two hydrogen bonds, the alkyl substituent of the
guanidine must occupy a position close to the third
receptor arm with its ethyl phosphonate functionality.
For a closer inspection of potential contacts between
the receptor arms and the guanidinium NH protons, we
also performed a NOESY experiment in a mixture of
deuterated and nondeuterated methanol (4:1) with sup-
pression of the solvent signal. At room temperature, only
one significant NOE contact could be identified, i.e.,
between the guanidinium NH2 signals and the two
aromatic receptor protons close to the benzamide group.
This shows that the guanidinium moiety lies on top of
the receptor aromatic, which is necessary for π-cation
methylguanidine HCl 800 ( 6%
490 ( 15% 3500 ( 8%
800 ( 19% 570 ( 17% 3500 ( 16%
N-benzoylarginine
ethyl ester HCl
stoichiometry
1:1
1:1
1:1
Ta ble 4. F r ee Bin d in g En th a lp ies ∆G [k ca l/m ol]
Mea su r ed in MeOD for th e 1:1 Com p lexa tion of Alk yl
Gu a n id in iu m Der iva tives w ith Recep tor Molecu les 1, 2,
a n d 8 by Mea n s of NMR Titr a tion Exp er im en ts
guest
1
2
8
methylguanidine hydrochloride
N-benzoylarginine ethyl ester
hydrochloride
4.0
4.0
3.6
3.7
4.8
4.8
We would like to emphasize that in DMSO the arginine
derivative was always bound three to six times more
strongly by hosts 1-7 than the simple methylguanidine,
which was always indicated by a seizable downfield shift
of the amidic arginine-NH proton of 1-2 ppm. Thus,
receptor molecules 1-7 are all arginine-selective.
The less symmetrical trisphosphonate 8 could only be
synthesized as the trilithium salt and was completely
insoluble in DMSO. Therefore, we prepared the respec-
tive lithium salts of 1 and 2 and compared their binding
properties with those of 8 in pure methanol. As Tables 3
and 4 show, 1 is again more effective than 2, and we
explain this difference once more with the additional
π-cation interaction. However, to our great pleasure,
host molecule 8 is half an order of magnitude more
efficient in guanidine binding than 1 and 2. This agrees
with comparable results from conformational searches by
Monte Carlo simulations:34 The distance between the
three converging phosphonate anions of 8 in the complex
is tailored for optimal electrostatic attraction as well as
hydrogen bonding with all five guanidinium NH-protons.
(34) Modeling Program: MacroModel 7.0 (Schro¨dinger Inc.), 2000.
Energy-minimizations and Monte Carlo Simulations were carried out
in water with the AMBER* force-field. The best structure from the
conformational searches was subsequently subjected to a molecular
dynamics calculation at ambient temperature for 10 ps.

5820 J . Org. Chem., Vol. 66, No. 17, 2001
Rensing et al.
cool to ambient temperature. The solvent is distilled off, and
products 10a and 10b are purified over silica gel (KG60,
hexanes/dichloromethane 3:2), whereas methyl benzoate 10c
and naphthalene derivative 10d are used without further
purification. 3,5-Bis(br om om eth yl)n itr oben zen e 10a :
1
Yield: 18%; mp: 103-104 °C; H NMR (200 MHz, CDCl3) δ
4.46 (s, 4H), 7.68 (s, 1H), 8.13 (s, 2H); 13C NMR(50 MHz,
CDCl3) δ 30.3 (s), 123.3 (s), 134.8 (s), 139.9 (s), 148.1 (s). Anal.
Calcd for C8H7Br2NO2: C, 31.10; H, 2.28; N, 4.53. Found: C,
32.01; H, 2.96; N, 5.00. 3,5-Bis(br om om eth yl)m eth oxyben -
1
zen e 10b: Yield: 27%; H NMR (200 MHz, CDCl3) δ 3.72 (s,
F igu r e 9. Summarized NOE contacts between 8 and the
methylguanidinium cation in methanol.
3H), 4.37 (s), 6.82 (s, 2H), 7.06 (s, 2H); 13C NMR(50 MHz,
CDCl3) δ 32.8 (s), 55.4 (s), 114.7 (s), 121.8 (s), 136.6 (s), 160.0
(s). Anal. Calcd for C9H10Br2O: C, 36.77; H, 3.43. Found: C,
36.32; H, 3.26.
interactions. Since all the guanidinium NH signals are
broad and not separated from each other, only very strong
intermolecular NOE’s could be detected. In addition, a
dynamic equilibrium must be assumed at this tempera-
ture with fast interconversion of different complex ge-
ometries and hydrogen bond patterns. Therefore, the
above-mentioned experiment can only give indications for
the complex geometry in averaged structures. In the
future we intend to use well soluble complexation part-
ners to carry out these NOESY experiments at low
temperatures (<-50 °C). In a related example Gschwind
and co-workers were able to generate five different NH
signals for the guanidinium cation in arginine involved
in complex formation with a bisphosphonate tweezer by
cooling the NMR tube down to -73 °C.35
Only one drawback remains: In methanol, the receptor
molecules cannot distinguish between methylguanidine
and amide-protected arginine ester. The additional co-
operative hydrogen bond formed between the receptor’s
phosphonate and one of the peptidic NH-groups in DMSO
is absent in methanol. This may be different in the
hydrophobic active sites of enzymes.
Syn th esis of m -Xylylen e Bisp h osp h on a tes. Gen er a l
P r oced u r e. The dibromoderivative 10 is dissolved in 3 equiv
of trimethyl phosphite and refluxed for 3-4 h. Remaining
trimethyl phosphite and methylphosphonic acid dimethyl ester
are removed by vacuum distillation at 80 °C/0.1 mbar. The
product is purified over silica gel (KG 60, ethyl acetate/
methanol 10:1). Recrystallization from ethyl actetate/hexanes
affords colorless crystals. 5-Nit r o-m -xylylen e b isp h os-
p h on ic a cid tetr a m eth yl ester 11a : Yield: 98%; mp: 103-
2
104 °C; 1H NMR (300 MHz, CDCl3) δ 3.18 (d, 4H, J HP ) 21.9
3
Hz), 3.66 (d, 12H, J HP ) 11.0 Hz); 7.53 (s, 1H), 7.98 (q, 2H,
1
4J HP ) 1.7 Hz); 13C NMR(50 MHz, CDCl3) δ 32.9 (d, J CP
)
2
139.0 Hz), 53.5 (d, J CP ) 6.8 Hz), 123.7 (m), 134.5 (s), 137.4
(m).31P NMR(80 MHz, CDCl3) δ 27.4. Anal. Calcd for C12H19
-
NO8P2: C, 44.33; H, 6.30. Found: C, 43.74; H, 6.33. 5-Meth -
oxy-m -xylylen e b isp h osp h on ic a cid t et r a m et h yl est er
11b: Yield: 98%; 1H NMR (200 MHz, D2O) δ 3.11 (d, 4H, 2J HP
3
) 22.0 Hz), 3.47 (d, 12H, J HP ) 13.0 Hz), 3.61 (s), 6.55-6.75
1
(m, 3H); 13C NMR (50 MHz, D2O) δ 31.1 (d, J CP ) 135.0 Hz),
53.7 (d, 2J CP ) 8 Hz), 55.6 (s), 114.6 (m), 124.0 (s), 133.2 (m).31P
NMR(80 MHz, D2O) δ 34.0. Anal. Calcd for C13H22O7P2: C,
44.33; H, 6.30. Found: C, 43.74; H, 6.33. 5-Meth oxyca r bon yl-
m -xylylen e bisp h osp h on ic a cid tetr a m eth yl ester 11c:
Yield: 35% (both steps); mp: 110-111 °C; 1H NMR (300 MHz,
2
3
CDCl3) δ 3.14 (d, 4H, J HP ) 21.9 Hz), 3.63 (d, 12H, J HP
)
Con clu sion s
11,0 Hz), 3.83 (s, 3H), 7.39 (s, 1H), 7.79 (s, 2H); 13C NMR (50
In conclusion, the binding strength of a simple arginine
receptor molecule could be considerably increased by
strengthening π-cation as well as electrostatic interac-
tions. The selective introduction of electron-donating or
-withdrawing substituents in the 5-position of the parent
bisphosphonate revealed the presence of π-cation inter-
actions, which contribute at least 0.5 kcal/mol for a single
benzene-guanidinium stack. Much more effective, how-
ever, was the introduction of a third phosphonate func-
tionality at the correct distance, so that five hydrogen
bonds can be formed to all five guanidinium NH protons,
surrounded by three phosphonate arms. Monte Carlo
simulations and NOESY experiments confirm these
complex geometries. The optimized host molecule binds
arginine half an order of magnitude more efficiently than
the parent molecule. With the new guanidinium host 8,
we are currently developing an artificial receptor mol-
ecule for the RGD sequence, which is essential for many
cell surface recognition processes.36
MHz, CDCl3) δ 32.9 (d, 1J CP ) 138.5 Hz), 52.6 (s), 53.3 (d, 2J CP
4
3
) 6.8 Hz), 129.9 (m), 131.3 (t, J CP ) 2.8 Hz), 132.7 (t, J CP
)
6.2 Hz), 135.8 (t, J CP ) 6.2 Hz); 166.8 (s); 31P NMR(80 MHz,
CDCl3) δ 28.5. Anal. Calcd for C14H22O8P2: C, 44.22; H, 5.83.
Found: C, 44.64; H, 5.57. 1,3-Bis(d im eth oxyp h osp h or yl-
m eth yl)n a p h th a len e 11d : Yield: 32% (both steps); mp.: 118
°C; 1H NMR (200 MHz, CDCl3) δ 3.56 (d, 2H, 2J HP ) 20.2 Hz),
3.24 (d, 2H, 2J HP ) 20.2 Hz), 3.54 (d, 6H, 3J HP ) 10.1 Hz), 3.61
(d, 6H, 3J HP ) 10.1 Hz), 7.94-7.21 (m, 6H); 13C NMR (50 MHz,
3
1
1
CDCl3) δ 30.3 (d, J CP ) 139.0 Hz), 33.2 (d, J CP ) 139.0 Hz),
53.3 (d, 2J CP ) 6.8 Hz), 53.4 (d, 2J CP ) 6.8 Hz), 124.4 (s), 126.6
(s), 128.5-130.0 (m), 130.6 (m), 131.3 (m), 134.4 (s). 31P NMR
(80 MHz, CDCl3) δ 29.8; 29.2. Anal. Calcd for C16H22 O6P2: C,
51.62; H, 5.96. Found: C, 49.91; H, 6.16.
Syn th esis of Tetr a bu tyla m m on iu m Sa lts of m -Xylyl-
en e Bisp h osp h on a tes. Gen er a l P r oced u r e. A 1 g amount
of the respective m-xylylene bisphosphonate is dissolved in 50
mL of water and treated with 2 equiv of 1 N aqueous
tetrabutylammonium hydroxide. The solution is then refluxed
for 7 days. Evaporation of the solvent affords a colorless oil.
Bis(tetr a bu tyla m m on iu m )-5-n itr o-m -xylylen e d im eth yl
bisp h osp h on a te 3a : Yield: quantitative; 1H NMR (300 MHz,
3
CDCl3) δ 0.89 (t, 36H, J HH ) 7.31 Hz), 1.31 (m, 24H), 1.51
Exp er im en ta l Section
2
(m, 24H), 2.87 (d, 4H, J HP ) 20.2 Hz), 3.13 (m, 24H), 3.43 (d,
3
6H, J HP ) 11.0 Hz), 7.23 (s, 1H), 7.67 (s, 2H); 13C NMR (50
Syn th esis of R,R′-Dibr om oxylylen e Der iva tes. Gen er a l
Dibr om in a tion P r oced u r e. A 5 g amount of the xylylene
derivative, 2 equiv of N-bromosuccinimide, and a catalytic
amount of AIBN are dissolved in 200 mL of tetrachloro-
methane. After refluxing the reaction mixture for 3 h, complete
conversion is reached, and the reaction mixture is allowed to
1
MHz, CDCl3) δ 13.6 (s), 19.6 (s), 23.9 (s), 34.8 (d, J CP ) 125.2
Hz), 51.6 (d, 2J CP ) 6.2 Hz), 58.5 (s), 121.3 (m), 130.5 (m), 137.8
(m), 138.8 (m,); 147.7 (s); 31P NMR(80 MHz, CDCl3): 18.0 (s).
MS (ESIneg): m/z 337.16, found 168 (M2-), 337 ((M + 1)-),
579 ((M + NBu4+)-). Anal. Calcd for C58H121N3O8P2‚4H2O: C
56.42, H 10.48, N 4.70. Found: C 56.42, H 10.83, N 4.53. Bis-
(tetr a bu tyla m m on iu m )-5-m eth oxy-m -xylylen e d im eth yl
bisp h osp h on a te 4a : Yield: quantitative; 1H NMR (200 MHz,
(35) Gschwind, R.; Armbru¨ster, M. diploma thesis, Marburg, 2001.
(36) Haubner, R.; Finsinger, D.; Kessler, H. Angew. Chem. 1997,
109, 1440-1456.
3
CDCl3) δ 0.90 (t, 36H, J HH ) 7.21 Hz), 1.30 (m, 24H), 1.52

Optimization of a Synthetic Arginine Receptor
J . Org. Chem., Vol. 66, No. 17, 2001 5821
2
(m, 24H), 2.95 (d, 4H, J HP ) 20.1 Hz), 3.18 (m, 24H), 3.45 (d,
subsequently overnight at room temperature. The solvent is
removed to afford the product as a colorless powder. Lith iu m
bis(d im eth oxyp h osp h or ylm eth yl)ben zoa te: Yield: 89%;
12H, J HP ) 10.2 Hz), 7.58 (s, 1H), 7.97 (s, 1H); 13C NMR (50
3
1
MHz, CDCl3) δ 14.1 (s), 20.0 (s), 24.3 (s), 31.1 (d, J CP ) 135.0
2
1
Hz), 53.7 (d, J CP ) 8 Hz), 55.6 (s), 59.0 (s), 114.6 (m), 124.0
1H NMR (200 MHz, D2O) δ 3.24 (d, 4H, J HP ) 21.5 Hz); 3.60
(s), 133.2 (m). 31P NMR (80 MHz, CDCl3) δ 20.1(s). Tr is-
(d, 12H, J HP ) 10.8 Hz); 7.15 (s, 1H);7.70 (m, 2H); 13C NMR
2
1
(t et r a b u t yla m m on iu m )-5-ca r b oxyla t o-m -xylylen e
d i-
(80 MHz, D2O) δ 32.9 (d, J CP ) 138.5 Hz), 52.6 (s), 53.3 (d,
1
4
m eth yl bisp h osp h on a te 6a : Yield: quantitative; H NMR
2J CP ) 6.8 Hz), 129.9 (m), 131.3 (t, J CP ) 2.8 Hz), 132.7 (t,
3
(300 MHz, CDCl3) δ 0.90 (t, 36H, J HH ) 7.21 Hz), 1.35 (m,
3J CP ) 6.2 Hz), 135.8 (t, 3J CP ) 6.2 Hz); 169.1 (s); 31P NMR (80
MHz, D2O) δ 34.5. A solution of 500 mg of lithium bis-
(dimethoxyphosphorylmethyl)benzoate in ethanol/dichloro-
methane (1:2) is treated with 1 equiv of aminomethylphos-
phonic acid diethyl ester, 1 equiv of diisopropylethylamine, and
1 equiv of bis(2-oxo-3-oxazolidinyl)phosphinic acid chloride and
stirred for 3 days at room temperature. After filtration, the
solvent is removed in vacuo, and the product (14) is purified
over silica gel (KG 60, ethanol). 3,5-Bis-(d im eth oxyp h os-
p h or ylm eth yl)ben zoic a cid d ieth oxyp h osp h or ylm eth yl
a m id e 14: Yield: 75%; 1H NMR (200 MHz, CDCl3) δ 1.15 (dt,
2
24H), 1.55 (m, 24H), 2.97 (d, 4H, J HP ) 20.1 Hz), 3.20 (m,
3
24H), 3.46 (d, 6H, J HP ) 10.2 Hz), 7.45 (s, 1H), 7.70 (s, 2H);
13C NMR (50 MHz, CDCl3) δ 13.5 (s), 19.5 (s), 23.8 (s), 34.4 (d,
1J CP ) 128.9 Hz), 51.7 (d, J CP ) 5.7 Hz), 58.6 (s), 128.3 (m),
2
130.4 (m), 131.6 (m), 135.2 (m); 169.3 (s); 31P NMR (80 MHz,
CDCl3): 20.0 (s). MS (ESIneg): m/z 335.16, found 335 ((M +
1)-), 577 ((M + NBu4+)-). Anal. Calcd for C59H121N3O8P2‚
11H2O: C 56.21, H 11.43, N 3.33. Found: C 56.36, H 11.08, N
3.62.
Syn t h esis of Lit h iu m Sa lt s of m -Xylylen e Bisp h os-
p h on a tes. Gen er a l P r oced u r e. A 1 g amount of the respec-
tive m-xylylene bisphosphonate tetraester is dissolved in a
small amount of dry acetonitrile and treated with 2.5 equiv of
lithium bromide. The solution is refluxed overnight. After
cooling to room temperature, the precipitate is filtered off,
washed two times with a small amount of hot acetonitrile, and
dried in vacuo. Dilith iu m -m -xylylen e d im eth yl bisp h os-
p h on a te 1b: Yield: 79%; mp >350 °C (dec); 1H NMR (200
3
4
2
6H, J HH ) 7.0 Hz, J HP ) 1.0 Hz), 3.12 (d, 4H, J HP ) 21.8
Hz), 3.62 (d, 12H, J HP ) 10.1 Hz), 3.84 (dd, 2H,3J HH ) 6.4
3
2
Hz, J HP ) 10.8 Hz), 4.11 (m, 4H), 7.30 (s, 1H), 7.59 (m, 2H);
13C NMR (80 MHz, CDCl3) δ 16.3 (d, J CP ) 5.9 Hz), 32.3 (d,
2
1J CP ) 138.5 Hz), 35.2 (d, 1J CP ) 156.9 Hz), 52.4 (d, J CP ) 6.8
2
2
Hz), 62.6 (d, J CP ) 6.5 Hz), 126.9 (m), 127.3 (m), 131.5 (m),
132.1 (m),134.0 (m), 134.9 (m), 167.2 (dd, 2J CP ) 34.0 Hz, 5J CP
) 4.4 Hz). 31P NMR (80 MHz, CDCl3) δ 28.7 (s); 23.5 (s). Anal.
Calcd for C18H32NO10P3: C, 41.95; H, 6.26, N, 2,72. Found: C,
41.62; H, 6.16; N 2,70. A 1 g amount of intermediate 14 is
dissolved in a small amount of acetonitrile and treated with
2.5 equiv of lithium bromide. The solution is refluxed over-
night. After cooling to room temperature, the precipitate (8)
is filtered off, washed two times with a small amount of hot
acetonitrile, and dried in vacuo. Tr ilith iu m 3,5-bis(m eth yl-
p h osp h on a t om et h yl)b en zoic a cid et h ylp h osp h on a t o
m eth yl a m id e 8: Yield: quantitative; mp > 350 °C (dec); 1H
2
3
MHz, D2O) δ 2.86 (d, 4H, J HP ) 20.6 Hz), 3.35 (d, 6H, J HP
)
10.2 Hz), 7.01 (m, 3H).13C NMR (50 MHz, D2O) δ 33.6 (d, 1J CP
) 129.5 Hz), 52.0 (d, 2J CP ) 6.4 Hz), 127.5 (m), 128.8 (m), 130.8
(m), 135.0 (m).31P NMR (80 MHz, D2O) δ 27.3 (s). Anal. Calcd
for C10H13O6 P2 Li2‚3H2O: C, 35.11; H, 5.30. Found: C, 34,86;
H, 5,32. Dilith iu m 1,3-bis(m eth ylp h osp h on a tom eth yl)-
1
n a p h th a len e 5b: Yield: 89%; mp > 350 °C (dec); H NMR
2
(200 MHz, D2O) δ 3.13 (d, 2H, J HP ) 20.5 Hz), 3.36 (d, 3H,
3J HP ) 10.3 Hz), 3.45 (d, 2H, 2J HP ) 20.5 Hz), 3.48 (d, 3H, 3J HP
2
) 10.3 Hz), 7.32 (m, 1H), 7.49 (m, 2H), 7.61 (m, 1H), 7.81 (m,
NMR (200 MHz, D2O) δ 1.33 (t, 3H, J HP ) 7.2 Hz), 3.20 (d,
1
2
2
1H), 8.06 (m, 1H). 13C NMR (50 MHz, D2O) δ 31.1 (d, J CP
)
4H, J HP ) 20.2 Hz), 3.63 (d, 6H, J HP ) 12.0 Hz), 3.74(d, 2H,
20.0 Hz), 4.07 (m, 2H), 7.48 (s, 1H), 7.62 (s, 1H); 13C NMR (80
MHz, D2O) δ 17.4 (d, 2J CP ) 5.8 Hz), 30.3 (d, 1J CP ) 139.0 Hz),
1
2
143.1 Hz), 33.7 (d, J CP ) 141.0 Hz), 52.2 (d, J CP ) 6.2 Hz),
124.8 (m), 125.9 (m), 126.4 (m), 126.9 (m), 128.4 (m), 130.2
(m), 130.5 (m), 131.5 (m), 132.2 (m), 134.0 (m). 31P NMR (80
MHz, D2O) δ 29.7 (s), 29.5 (s). Anal. Calcd for C14H16O6P2Li2‚
3H2O: C 41.00; H, 5.41. Found: C 40.36; H 5.44.
1
2
2
33.2 (d, J CP ) 139.0 Hz), 53.3 (d, J CP ) 6.8 Hz), 53.4 (d, J CP
) 6.8 Hz), 124.4 (s), 126.6 (s), 128.5-130.0 (m), 130.6 (m), 131.3
(m), 134.4 (s) 169.1 (d, 2J CP ) 34.0 Hz). 31P NMR(80 MHz, D2O)
δ 26.6 (s); 20.9 (s). MS (ESIneg): m/z 456.24, found 456 (M-),
Syn th esis of Tetr a bu tyla m m on iu m Sa lt 5a . A 1 g
amount of dilithium salt 5b is dissolved in a small amount of
water and acidified with 3 equiv of 1 N hydrochlorid acid. The
precipitate is filtered off, washed two times with cold water,
564 ((M + Li+)-), 470 ((M + 2Li+)-). Anal. Calcd for C14H21
-
Li3NO10P3‚3H2O: C, 31.66; H, 5.12, N, 2.64. Found: C, 31.65;
H, 4.95, N, 2.60.
1
NMR Titr a tion s. Ten NMR tubes were filled each with 0.8
mL of a solution of the guest compound (cguest ) 0.5-4 mM) in
a deuterated solvent (DMSO-d6, or methanol-d4). 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 with increasing volumes from 0 to 5 equiv
to the guest solution in 10 NMR tubes. In some cases, guest
solutions were added with increasing volumes from 0 to 5 equiv
to the host solution in 10 NMR tubes.
Volume and concentration changes were taken into account
during analysis. The association constants were calculated by
nonlinear regression methods.
J ob P lots. Equimolar solutions (10 mmol/10 mL, ap-
proximately 1 mM) of guanidinium compound and bis- or
trisphosphonate host were prepared and mixed in various
amounts. 1H NMR spectra of the mixtures were recorded, and
the chemical shifts were analyzed by J ob’s method modified
for NMR results.
and dried in vacuo. Yield: 99%; H NMR (200 MHz, DMSO-
2
2
d6) δ 3.27 (d, 2H, J HP ) 21.7 Hz), 3.57 (d, 2H, J HP ) 21.7
Hz), 3.56 (d, 3H, 3J HP ) 10.8 Hz), 3.59 (d, 3H, 3J HP ) 10.8 Hz),
7.35 (m, 2H), 7.62 (m, 2H), 7.70 (m, 1H), 8.35 (m, 1H). 13C NMR
1
1
(50 MHz, DMSO-d6) δ 30.7(d, J CP ) 144.3 Hz), 33.3 (d, J CP
2
2
) 143.1 Hz), 52.0 (d, J CP ) 6.2 Hz), 51.9 (d, J CP ) 6.2 Hz),
125.3 (m), 125.6 (m), 126.1 (m), 128.1 (m), 130.7 (m), 133.8
(m), 140.1 (m), 140.5 (m); 31P NMR(80 MHz, DMSO-d6) δ 29.7
(s), 29.5 (s). The colorless product is triturated with a small
amount of water and then treated with 2 equiv of tetrabutyl-
ammonium hydroxide. The mixture is stirred until the acid is
completely dissolved; then the solution is stirred for another
5 min. The solvent is removed to afford a colorless oil.
Bis(tetr a bu tyla m m on iu m )-1,3-bis-(m eth ylp h osp h on a to-
m eth yl)n a p h th a len e 5a : Yield: quantitative; 1H NMR (200
MHz, CDCl3) δ 0.94 (t, 36H, J ) 7.34 Hz), 1.31 (m, 24H), 1.45
2
(m, 24H), 3.02 (m,24H), 3.16 (d, 2H, J HP ) 20.2 Hz), 3.44 (d,
2H, 2J HP ) 20.2 Hz), 3.51 (d, 3H, 3J HP ) 10.1 Hz), 3.58 (d, 3H,
3J HP ) 10.1 Hz), 7.32-7.35 (m), 7.62 (m), 7.70 (m), 8.35 (m);
Ack n ow led gm en t. We thank Prof. Dr. G. Wulff
(Du¨sseldorf University) for generous support during
recent years.
13C NMR (80 MHz, CDCl3) δ 30.3 (d, J CP ) 139.0 Hz), 33.2
1
1
2
2
(d, J CP ) 139.0 Hz), 53.3 (d, J CP ) 6.8 Hz), 53.4 (d, J CP
)
6.8 Hz), 124.4 (s), 126.6 (s), 128.5-130.0 (m), 130.6 (m), 131.3
(m), 134.4 (s). 31P NMR (80 MHz, CDCl3) δ 18.8 (s); 18.5 (s).
MS (ESIneg): m/z 342.16, found 171(M2-), 342 ((M + 1)-), 584
((M + NBu4+)-).
Su p p or tin g In for m a tion Ava ila ble: NMR titration data,
titration curves, and J ob plots for all new hosts with methyl-
guanidinium hydrochloride and N-benzoylarginine ethyl ester
hydrochloride. This material is available free of charge via the
Internet at http://pubs.acs.org.
Syn th esis of Tr ilith iu m Tr isp h osp h on a te 8. The methyl
benzoate 13 is dissolved in 30 mL of methanol/water (2:1),
cooled to -15 °C, and treated with 1.5 equiv of lithium
hydroxide. The solution is stirred at -10 °C for 6 h and
J O0156161