Pyrrolic tripodal receptors effectively recognizing monosaccharides. Affinity assessment through a generalized binding descriptor

Article abstract of DOI:10.1021/ja068754m

Pyrrolic and imino (3) or amino (4) H-bonding ligands were incorporated into a benzene-based tripodal scaffold to develop a new generation of receptors for molecular recognition of carbohydrates. Receptors 3 and 4 effectively bound a set of octylglycoside

Full text of DOI:10.1021/ja068754m

Published on Web 03/16/2007
Pyrrolic Tripodal Receptors Effectively Recognizing
Monosaccharides. Affinity Assessment through a Generalized
Binding Descriptor
Cristina Nativi,‡,£ Martina Cacciarini,‡ Oscar Francesconi,‡ Alberto Vacca,†
Gloriano Moneti,# Andrea Ienco,¶ and Stefano Roelens*,§
Contribution from the Dipartimento di Chimica Organica, Dipartimento di Chimica, Centro
Risonanze Magnetiche (CERM), and Centro Interdipartimentale di Spettrometria di Massa
(CISM), UniVersita` di Firenze, and Istituto di Chimica dei Composti OrganoMetallici (ICCOM)
and Istituto di Metodologie Chimiche (IMC), Consiglio Nazionale delle Ricerche (CNR), Polo
Scientifico e Tecnologico, I-50019 Sesto Fiorentino, Firenze, Italy
Received December 6, 2006; E-mail: stefano.roelens@unifi.it
Abstract: Pyrrolic and imino (3) or amino (4) H-bonding ligands were incorporated into a benzene-based
tripodal scaffold to develop a new generation of receptors for molecular recognition of carbohydrates.
Receptors 3 and 4 effectively bound a set of octylglycosides of biologically relevant monosaccharides,
including glucose (Glc), galactose (Gal), mannose (Man), and N-acetyl-glucosamine (GlcNAc), showing
micromolar affinities in CDCl3 and millimolar affinities in CD3CN by NMR titrations. Both receptors selectively
recognized Glc among the investigated monosaccharides, with 3 generally less effective than 4 but showing
selectivities for the all-equatorial â-glycosides of Glc and GlcNAc among the largest reported for H-bonding
synthetic receptors. Selectivities in CDCl3 spanned a range of nearly 250-fold for 3 and over 30-fold for 4.
Affinities and selectivities were univocally assessed through the BC50 descriptor, for which a generalized
treatment is described that extends the scope of the descriptor to include any two-reagent host-guest
system featuring any number of binding constants. ITC titrations of âGlc in acetonitrile evidenced, for both
receptors, a strong enthalpic contribution to the binding interaction, suggesting multiple H bonding. Selectivity
trends toward RGlc and âGlc analogous to those obtained in solution were also observed in the gas phase
for 3 and 4 by collision-induced dissociation experiments. From comparison with appropriate reference
compounds, a substantial contribution to carbohydrate binding emerged for both the imino/amino and the
pyrrolic H-bonding groups but not for the amidic group. This previously undocumented behavior, supported
by crystallographic evidence, has been discussed in terms of geometric, functional, and coordinative
complementarity between H-bonding groups and glycosidic hydroxyls and opens the way to a new designer
strategy of H-bonding receptors for carbohydrates.
chemistry in the past decade.2 Because of the complexity of
natural processes, carbohydrate recognition has been investi-
gated, focusing mainly on the monosaccharides or short oli-
gosaccharides most frequently encountered as natural epitopes
and largely exploiting synthetic receptors to mimic the interac-
tion occurring with natural receptors and to understand the
molecular basis of recognition events.3
Introduction
Molecular recognition of carbohydrates is crucial in many
biological processes besides carbohydrate metabolism and
transport, including cell-to-cell adhesion, cell infection by
pathogens, immune response, and enzyme activity regulation.1
All such processes rely on selective carbohydrate-to-carbohy-
drate and carbohydrate-to-protein interactions as a fundamental
step. No wonder recognition of carbohydrates has been an
actively investigated subject in bioorganic and supramolecular
Among the many questions still open, a crucial issue is the
comprehension of the structural and functional requirements for
‡ Dipartimento di Chimica Organica, Universita` di Firenze.
† Dipartimento di Chimica, Universita` di Firenze.
£ CERM, Universita` di Firenze.
(2) (a) Lutzen, A. In Highlights in Bioorganic Chemistry; Schmuck, C.,
Wennemers, H., Eds.; Wiley-VCH: Weinheim, Germany, 2004; pp 109-
119. (b) Striegler, S. Curr. Org. Chem. 2003, 7, 81-102. (c) Dam, T. K.;
Brewer, C. F. Chem. ReV. 2002, 102, 387-429. (d) Mellet, C. O.; Defaye,
J.; Garc´ıa Ferna´ndez, J. M. Chem.sEur. J. 2002, 8, 1982-1990. (e) Host-
Guest Chemistry. Mimetic Approaches to Study Carbohydrate Recognition.
In Topics in Current Chemistry; Penade´s, S., Ed.; Springer-Verlag:
Heidelberg, Germany, 2002; Vol. 218. (f) Davis, A. P.; Wareham, R. S.
Angew. Chem., Int. Ed. 1999, 38, 2978-2996. (g) James, T. D.; Sandan-
ayake, K. R. A. S.; Shinkai, S. Angew. Chem., Int. Ed. Engl. 1996, 35,
1910-1922. (h) Lee, Y.C.; Lee, R.T. Acc. Chem. Res. 1995, 321-327.
(3) For a recent comprehensive review on the subject, see: Davis, A. P.; James,
T. D.; In Functional Synthetic Receptors; Schrader, T., Hamilton, A. D.,
Eds.; Wiley-VCH: Weinheim, Germany, 2005; pp 45-109.
# CISM, Universita` di Firenze.
¶ ICCOM, CNR.
§ IMC, CNR.
(1) (a) Ernst, B.; Hart, W.; Sinay¨, P. Carbohydrates in Chemistry and Biology;
Wiley-VCH: Weinheim, Germany, 2000; Part I, Vol. 2 and Part II, Vol.
4. (b) Lindhorst, T. K. Essentials of Carbohydrate Chemistry and
Biochemistry; Wiley-VCH: Weinheim, Germany, 2000. (c) Essentials of
Glycobiology; Varki, A., Cummings, R., Esko, J., Freeze, H., Hart, G.,
Marth, J., Eds.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor,
NY, 1999.
9
10.1021/ja068754m CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 4377-4385
4377

A R T I C L E S
Nativi et al.
Chart 1
the effective and selective recognition of a specific saccharide.
In nature, molecular recognition of carbohydrates relies on weak
noncovalent interactions, mainly hydrogen bonding, and the
required affinity and selectivity are achieved by multivalency1,4
through a concerted array of H bonds involving multiple
carbohydrate units. In synthetic receptors, such a combination
is difficult to achieve by design, but by capitalizing on
noncovalent interactions,5 encouraging results have been ob-
tained in several cases, mostly in organic solvents, even though
high selectivity remains the most ambitious goal yet to be
achieved.3 Among the numerous artificial receptors reported to
date, benzene-based tripodal structures, both macrocyclic and
acyclic, have been successfully explored for the recognition of
saccharides.6 In this context, we have recently reported a new
prototypical tripodal receptor for the recognition of mono-
saccharides, featuring a triethylbenzene scaffold bearing three
convergent ureidic H-bonding units (1).7 Binding affinities in
the millimolar range and moderate selectivities were measured
for 1 in CDCl3 toward a set of representative R- and â-octyl-
glycosides, selected among the most relevant to biological
recognition processes. In order to improve on binding ability,
we reasoned that more efficient receptors may be obtained by
replacing ureidic groups with H-bonding ligands that could
exhibit better complementarity with the carbohydrate hydroxy
and ether functions but still preserving the flexibility of the
tripodal architecture. The excellent results obtained
for anion binding,10 appear to be, as of yet, essentially
unexplored for the recognition of carbohydrates.11 We thus
thought that amino and pyrrolic binding groups could be
conveniently assembled on the 1,3,5-triethylbenzene scaffold
to afford adaptive receptors of significantly improved recogni-
tion properties toward monosaccharides. Following a systematic
investigation, we now wish to describe a new type of tripodal
receptor, which exhibits some of the best binding affinities
toward biologically relevant monosaccharides reported in the
literature for neutral synthetic receptors, shows a unique
selectivity for the glycosides of â-glucose and N-acetyl-â-
glucosamine, and opens the way to a new generation of acyclic
hosts for recognition of carbohydrates solely based on H-
bonding interactions.
A fundamental issue for the assessment of binding properties
was the evaluation of affinities on a common scale because the
investigated receptor-glycoside systems fit different binding
models depending on the glycoside. To address this issue, we
also wish to report a generalized treatment we developed to
extend the scope of the BC50 descriptor, which we previously
proposed as a comparative index of binding efficacy,7 to include
systems fitting chemical models of any number of equilibria.
In the generalized formulation, the BC50 parameter can, thus,
be used to assess affinities on the same scale for any two-reagent
with a cage receptor featuring amino and pyrrole functions8
showed that these groups can be conveniently employed as
alternative H-bonding ligands. Indeed, amino and hydroxy
groups have been shown to be complementary H-bonding
partners, both geometrically and coordinatively, giving rise to
molecular recognition and self-assembly.9 Likewise, pyrroles,
which are well-established H-bonding donors largely employed
(4) (a) Kiessling, L.; Pontrello, L.; Schuster, M. C. In Carbohydrate-based
Drug DiscoVery; Wong, C. H., Ed.; Wiley-VCH: Weinheim, Germany,
2003; Vol. 2, pp 575-608. (b) Lundquist, J. J.; Toone, E. J. Chem. ReV.
2002, 102, 555-578. (c) Mammen, M.; Choi, S.-K.; Whitesides, G. M.
Angew. Chem., Int. Ed. 1998, 37, 2754-2794.
(5) Although not biomimetic, formation of covalent B-O bonds has also been
largely exploited for the recognition of carbohydrates. See ref 3.
(6) For some representative examples from the recent literature, see: (a) Abe,
H.; Aoyagi, Y.; Inouye, M. Org. Lett. 2005, 7, 59-61. (b) Schmuck, C.;
Schwegmann, M. Org. Lett. 2005, 7, 3517-3520. (c) Mazik, M.; Radunz,
W.; Boese, R. J. Org. Chem. 2004, 69, 7448-7462. (d) Welti, R.; Diederich,
F. HelV. Chim. Acta 2003, 86, 494-503. (e) Kim, Y-H.; Hong, J-I. Angew.
Chem., Int. Ed. 2002, 41, 2947-2950. (f) Zhong, Z.; Anslyn, E. V. J. Am.
Chem. Soc. 2002, 124, 9014-9015.
(7) Vacca, A.; Nativi, C.; Cacciarini, M.; Pergoli, R.; Roelens, S. J. Am. Chem.
Soc. 2004, 126, 16456-16465.
(10) Anion Receptor Chemistry; Sessler, J. L., Gale, P. A., Cho, W.-S., Eds.;
The Royal Society of Chemistry: Cambridge, U.K., 2006.
(11) Apart from a report on anionic carbohydrate recognition by a cationic
receptor through electrostatic interactions in water (ref 6b), we are aware
of only one example of pyrrole-containing receptors for sensing monosac-
charides. See: Fang, J.-M.; Selvi, S.; Liao, J.-H.; Slanina, Z.; Chen, C.-T.;
Chou, P.-T. J. Am. Chem. Soc. 2004, 126, 3559-3566.
(8) Francesconi, O.; Ienco, A.; Moneti, G.; Nativi, C.; Roelens, S. Angew.
Chem., Int. Ed. 2006, 45, 6693-6696.
(9) (a) Hanessian, S.; Simard, M.; Roelens, S. J. Am. Chem. Soc. 1995, 117,
7630-7645. (b) Ermer, O.; Eling, A. J. Chem. Soc., Perkin Trans. 2 1994,
925-944.
9
4378 J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007

Pyrrolic Tripodal Receptors Recognizing Monosaccharides
A R T I C L E S
Chart 2
host-guest systems, irrespective of the nature of the binding
partners. Making use of the generalized BC50 parameter, the
affinities of the receptors for the investigated set of glycosides
have been univocally assessed and may be compared with those
calculated for any other receptor.
for the amino-pyrrolic receptor 4. Eventually, receptors 8 and
9 were prepared from the homologue triamine 7 by condensation
with pyrrole-2-carboxaldehyde to elucidate the effect of the
spacer length on the binding capabilities of the imino- and
amino-pyrrolic receptors and to validate predictions from
molecular modeling calculations.
Results and Discussion
Binding Studies. The binding affinities toward the set of
octylglycosides of biologically relevant monosaccharides de-
picted in Chart 2 were measured by 1H NMR titrations in CDCl3
at T ) 298 K. Glc, Gal, Man, and GlcNAc were selected among
the most frequently encountered monosaccharidic epitopes,
present as terminal R- or â-glycosides in more complex oligo-
or polysaccharides on cell surfaces and in glycoconjugates.
An in-depth investigation of all of the aspects involved in
the correct determination of binding affinities toward glycosides
by NMR titrations (reactant self-association, chemical model,
choice of signals, etc.) has been described in a previous paper.7
In the present study, we closely followed the established titration
protocol and the optimized analysis of data, which consisted of
diluting the receptor with a stock solution of the glycoside and
simultaneously fitting of all the available signals of both reagents
to the appropriate model of chemical equilibria, respectively.
Unfortunately, the hydroxylic proton signals, usually experienc-
ing large shifts due to H bonding, could not be followed in the
titration experiments because they all collapsed into a broad
average signal, together with the water signal and with the
aminic protons for receptor 4. For the iminic receptor 3,
interestingly, with a lack of aminic protons, the signals involved
in fast exchange averaging with the hydroxylic protons were
the pyrrolic NH protons. In all cases, however, complexation-
induced shifts could be accurately monitored for several CH
signals of both binding partners, some of which exhibited
remarkably large values (see Supporting Information). The
titration of âGlcNAc with 3 is reported in Figure 1 as a typical
example, showing the excellent agreement obtained between
the experimental and the calculated shifts for all of the detected
signals. In general, binding of the receptors to the investigated
glycosides fit a model including 1:1 and 2:1 host-to-guest
association equilibria, in addition to the dimerization of the
receptor. For the latter self-association equilibrium, the corre-
sponding constants were measured independently and used as
invariant parameters in the nonlinear least-squares regression
analysis.14 The results obtained for receptors 3 and 4 with the
set of glycosides of Chart 2 are reported in Table 1 as cumulative
log â values for the formation of the 1:1 and 2:1 receptor-to-
Design and Synthesis. The design of the new pyrrolic
tripodal receptors was based on the same 1,3,5-substituted 2,4,6-
triethylbenzene scaffold12 of the parent ureidic host 1 on account
of its marked preference for the alternate substituents pattern,
directing the three binding arms toward the same side of the
aromatic ring, which was shown to provide the correct geometry
for binding monosaccharides.7 Molecular mechanics calculations
indicated that, for replacing the ureidic group with an amino-
pyrrole motif, an appropriate assemblage to achieve the correct
geometry for binding would require the amine and the pyrrole
groups to be spaced by one methylene unit, with the spacer
connected to the 2 position of the pyrrole ring. Such an
arrangement could be conveniently realized by condensation
of the parent triamine 2 with pyrrole-2-carboxaldehyde through
the formation of the corresponding Schiff base 3 (Chart 1).
Reduction to the amine 4 would provide an amino-pyrrole
receptor endowed with the functional, structural, and flexibility
features appropriate for H bonding to monosaccharides. Al-
though structurally similar to 4, the iminic receptor 3 may exhibit
significantly different binding properties because the confor-
mational constraint imposed by the iminic double bond, coplanar
with the pyrrolic ring due to conjugation, may lock the
conformation into a chelating arrangement of the two nitrogen
atoms. Since amidic groups are largely employed for H bonding
in carbohydrate recognition,3 the structurally related amido-
pyrrole 5 was prepared by standard peptide chemistry from
pyrrole-2,5-dicarboxylic acid monobenzylester,13 with the aim
of assessing whether the aminic group would be more effective
than the amidic group in binding carbohydrates. The benzylester
moiety in the 5 position of pyrrole was necessary to overcome
the poor solubility we observed with the corresponding 2,5-
diamidic (5a) and 5-unsubstituted (5b) amidic receptors, which
prevented an evaluation of their binding properties. To ascertain
the contribution from the pyrrolic groups to the recognition of
monosaccharides, the plain triacetamide 6, to be compared to
the amidic receptor 5, was prepared by acetylation of the
triamine 2; likewise, the plain triamine 2 was used as a reference
(12) Tripodal receptors based on hexasubstituted benzene rings have become
quite popular in the last decade. For a recent review, see: Hennrich, G.;
Anslyn, E. V. Chem.sEur. J. 2002, 8, 2218-2224.
(14) Measured self-association constants, log âdim, in CDCl3 at T ) 298 K for
1 and the receptors of Chart 1: 1.732 ( 0.009 (1); 1.83 ( 0.02 (2); 0.92
( 0.02 (3); 1.07 ( 0.01 (4); n.d. (5); n.d. (6).
(13) Schmuck, C.; Schwegmann, M. J. Am. Chem. Soc. 2005, 127, 3373-3379.
9
J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007 4379

A R T I C L E S
Nativi et al.
Table 1. Cumulative Association Constants (log ân) with Standard
Deviations (σ) for 1:1 (Upper Value) and 2:1 (Lower Value)
Complexes of Receptors 1, 3, and 4 with Octylglycosidesa
glycoside
1b
3
4
RGlc
2.77 ( 0.01
4.80 ( 0.02
2.67 ( 0.04
4.88 ( 0.06
2.82 ( 0.01
4.82 ( 0.02
2.92 ( 0.01
5.01 ( 0.03
2.82 ( 0.05
4.60 ( 0.11
2.31 ( 0.18
3.98 ( 0.29
2.75 ( 0.04
4.31 ( 0.13
2.62 ( 0.03
4.50 ( 0.06
3.573 ( 0.004
c
5.30 ( 0.05
9.04 ( 0.09
3.437 ( 0.002
c
3.23 ( 0.01
5.22 ( 0.04
4.61 ( 0.03
7.79 ( 0.06
3.10 ( 0.01
4.75 ( 0.10
4.153 ( 0.008
5.59 ( 0.12
4.37 ( 0.01
5.75 ( 0.08
4.43 ( 0.02
6.45 ( 0.09
4.14 ( 0.04
6.05 ( 0.47
4.72 ( 0.04
8.25 ( 0.06
âGlcd
RGal
âGal
3.921 ( 0.004
c
RMan
3.583 ( 0.006
c
âMan
3.185 ( 0.009
c
RGlcNAc
âGlcNAcf
2.937 ( 0.001
n.d.e
4.49 ( 0.04
7.95 ( 0.07
1
a Measured by H NMR (400 MHz) from titration experiments at T )
298 K in CDCl3 on 0.8-1.2 mM stock solutions of glycoside using receptor
concentrations up to 20 mM. Binding constants â11 (M-1) and â21 (M-2
)
were calculated by simultaneous fit of the shifts of all of the available
signals. The receptor’s dimerization constant was measured independently
under the same conditions and set invariant in the nonlinear regression
c
analysis. For log âdim values, see ref 14. b Data from ref 7. Too low to be
correctly determined. d Determined at 900 MHz. e Nondetectable. f Stock
solution, 0.27 mM.
Table 2. Association Constants Ka (M-1) and Thermodynamic
Parameters (kcal mol-1) for 1:1 Association of Receptors 3 and 4
with âGlca
receptor
Ka
−∆G
°
−∆H
°
−T∆S°
3
4
129.0 ( 1.6
87.4 ( 1.7
2.87 ( 0.01
2.65 ( 0.01
11.3 ( 0.8
6.0 ( 0.8
8.4
3.4
a Measured by ITC from titration experiments at T ) 298 K in CH3CN
in 1.0 mM solutions of receptor-injecting 25-50 mM solutions of glycoside.
calorimetry (ITC), whose results are reported in Table 2. Besides
the very good agreement between the association constants
obtained by the ITC and NMR techniques, thermodynamic
parameters evidenced a strong enthalpic contribution to the
association, which resulted in much smaller binding free-energy
values because of the large and adverse entropic contribution.
It should be mentioned that receptor 4 was also tested with fully
acetylated âGlc, giving no evidence of binding in CDCl3 and
clearly showing that the strong binding observed with âGlc is
fully depleted in the absence of free hydroxyl groups. Recogni-
tion of glycosides by the tripodal pyrrolic receptors can, thus,
be ascribed to a strong enthalpic interaction, likely resulting
from multiple H bonding.
As the largest constants were observed for âGlc, it was used
as a test ligand to evaluate the affinities of the reference
compounds 2 and 5-9. The parent triamine 2 showed a 1:1
association equilibrium, with a log â value of 2.616 ( 0.004,
together with a significant self-association constant.14 The
trisamidopyrrole 5 showed 1:1 and 2:1 association equilibria
with a negligible self-association constant; the corresponding
log â values were 2.22 ( 0.01 and 3.59 ( 0.03, respectively.
To ascertain any R/â selectivity, binding constants were also
measured toward RGlc, giving log â values for the 1:1 and 2:1
adducts of 2.02 ( 0.02 and 3.24 ( 0.06, respectively. In the
case of the triacetamido derivative 6, only a 1:1 association could
be revealed for âGlc, with a log â value of 1.21 ( 0.01 and
undetectable self-association. Finally, attempts to evaluate the
Figure 1. Plot of complexation induced shifts versus the receptor’s total
concentration in the titration of âGlcNAc (0.27 mM) with 3 in CDCl3 at
400 MHz and T ) 298 K. Bottom: glycoside CH signals (b CH-1; ×
CH-2; 1 CH-3; left arrow CH-4; + CH-5; 9 CH-6; ( CH-6′; 2 CH-7;
right arrow CH-7′). Top: receptor CH signals (b CHN-im; 9 CH-pyrr; 2
CH-pyrr; ( CH2N). For glycoside proton numbering, see Chart 2; for proton
assignment, see 2D-NMR spectra in Supporting Information. Symbols are
experimental data points; solid lines are best-fit curves obtained through
nonlinear regression by simultaneous fit of all data.
glycoside adducts. Corresponding data for 17 are also included
for direct comparison. While for RGlcNAc no evidence of the
2:1 complex species was found, in several cases, the concentra-
tion of the latter was too low to allow a correct determination
of the corresponding constant. In both instances, the system was
correctly described with good agreement by the 1:1 association
constant. From the inspection of Table 1, particularly large
values can be noted for âGlc with receptors 3 and 4. Since these
values were obtained in a noncompetitive solvent, binding
constants were also measured in CD3CN to ascertain whether
recognition would still occur in a polar medium. A 1:1
association with âGlc could, indeed, be detected, with log â
values of 2.169 ( 0.004 (Ka ) 147 ( 1.2 M-1) and 1.721 (
0.002 (Ka ) 52.5 ( 0.3 M-1) for 3 and 4, respectively. An
independent support was obtained by isothermal titration
9
4380 J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007

Pyrrolic Tripodal Receptors Recognizing Monosaccharides
A R T I C L E S
affinities of 8 and 9 gave no evidence of binding toward âGlc
and a clear-cut answer to the question about the choice of the
spacer length, in agreement with predictions based on molecular
modeling calculations.
this way, the affinity and the selectivity of 1 toward the
glycosides of Chart 1, all sharing a common three-constant
model.7
In order to be a general descriptor of binding affinity, BC50
must fulfill two basic requirements, namely, (1) it must be a
function of variables directly related to binding affinity, and
(2) to allow comparison of different systems, it must be
calculated under the same binding conditions. For systems fitting
the same chemical model, parity of conditions, besides solvent
and temperature, is ensured by calculating BC50 at the same TB
value. Unfortunately, this is no longer true for systems fitting
different models, as TB appears in the BC50 expression with a
different coefficient (thus having a different weight) for
complexes of different stoichiometry.16 For a generalized validity
of the descriptor, a more convenient quantity to refer to is the
fraction of bound A (eq 4), which identifies the extent of
saturation of the binding reagent
Although a whole set of binding constants was obtained for
the recognition of glycosides, the results reported in Table 1
cannot be directly used for assessing the binding ability of the
tripodal pyrrolic receptors for several reasons: (a) each constant
of the two complexation steps alone does not describe the overall
binding ability; (b) comparison of binding constants of different
orders is unfeasible; and (c) the investigated systems are not
homogeneously characterized by the same model. To assign an
univocal value to the affinity of each receptor for each glycoside
on a common scale, a general descriptor of binding affinity is
necessarily required.
Binding Descriptors. In a previous work,7 the affinity of a
reagent A for a reagent B has been assessed through the BC50
parameter, which we proposed as a descriptor of binding affinity.
In analogy to the IC50 parameter widely employed in biochem-
istry, the median binding concentration, BC50, was defined as
the total concentration of A (TA, in mol L-1) necessary for
binding 50% of B, that is, for which
xA ) TAbound /TA
(4)
BC50 is most usefully expressed as a function of this quantity
because xA has two main advantages over TB or the extent of
binding TAbound/TB. First, in contrast to TB, it does not appear
in the BC50 expression with coefficients dependent on the
complex’s stoichiometry, as long as complex species multi-
nuclear in B are absent (see Appendix in the Supporting
Information). Second, in contrast to TAbound/TB, it ranges from
0 to 1 for all systems, irrespective of the model and the
stoichiometry of the complexes formed. These properties ensure
the univocal definition of parity of conditions, that is, the same
extent of saturation of reagent A for systems fitting different
models and stoichiometries. Under these conditions, BC50
describes, on a common scale, the binding ability of reagents
when they have the same tendency to form complexes. Indeed,
from the expressions obtained for different models (see Ap-
pendix in the Supporting Information), it can be seen that BC50
depends solely on xA and on all binding constants and, therefore,
TBbound ) TB/2
(1)
or, equivalently
xB ) TBbound /TB ) 0.5
(2)
where xB is the fraction of bound B. Thus, like for IC50, the
higher the affinity, the lower the BC50 value. BC50 becomes
essential to assess affinities for systems featuring more than
one association equilibrium, in which the binding ability of a
species is not directly defined by a single binding constant. In
such cases, BC50, which can be calculated from the knowledge
of the set of binding constants, has been shown to depend on
all of the constants involved and thus takes into account all of
the complex species.7
BC50 is a conditional parameter that depends on the total
concentration of the analyte, TB, at which it is calculated (eq
1), which must be, therefore, specified together with the
temperature.15 A very useful property of BC50 is that, when the
total concentration of the analyte TB becomes negligible, the
value of BC50 becomes constant, that is,
for a specific xA value, exclusively on affinity, making BC50
a
binding descriptor of general scope. Thus, comparing the binding
abilities of reagents for two host-guest systems fitting different
chemical models translates into comparing BC50 values calcu-
lated at the same fraction of bound reagent xA, that is, calculating
the total concentration of A that binds 50% of B when A is
saturated to the same extent for both systems.
Analogous to BC50 as a function of TB, when xA becomes
negligible, the value of BC50 becomes constant, that is,
lim BC50 ) const ) BC500
(3)
lim BC50 ) const ) BC500
(5)
T
B f 0
x
A f 0
In contrast to BC50, BC0 , which we called intrinsic median
binding concentration, depends on all of the equilibrium
constants involved in the system but not on specific (concentra-
tion) conditions.
The BC50 descriptor is very useful when comparing binding
abilities for different host-guest systems. For example, the
affinity of two or more receptors for a ligand or the selectivity
of a receptor toward two or more ligands can be directly assessed
by comparing the corresponding BC50 values, whatever the
number of equilibria involved. We have successfully assessed,
50
For a two-reagent system, BC050 as a function of TB and xA
coincide because, when the concentration of B becomes
negligible, the fraction of bound A becomes negligible as
well. However, BC050 as a function of xA assumes the mean-
ing of the affinity of reagent A for reagent B in the absence
of formed complex species, that is, when forming the first
complex molecule. Thus, BC050 identifies the maximum bind-
ing ability of A toward B since, with increasing xA, A becomes
(16) This feature can be appreciated, for example, from the BC50 expressions
relative to complexes of the type AB, A2B, and A3B, respectively. AB:
-1/3
(15) We name the species (B) toward which the binding affinity is being
evaluated the “analyte”.
BC50 ) â11-1 + 0.5TB. A2B: BC50 ) â21-1/2 + 1TB. A3B: BC50 ) â31
+ 1.5TB.
9
J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007 4381

A R T I C L E S
Nativi et al.
progressively saturated. When comparing, for example, the
affinities of two reagents A and A′ (e.g., two hosts H and H′)
for the same ligand (G), while the selectivity ratio BC50(H)/
BC50(H′) will remain constant in the entire xH range and equal
to the BC0 (H)/BC0 (H′) ratio, the actual BC50 values will
Table 3. Intrinsic Median Binding Concentration BC050 (µM) with
Standard Deviation (σ) for Complexes of Receptors 1, 3, and 4
with Octylglycosides in CDCl3 at T ) 298 Ka
BC050
(σ)
glycoside
1b
3
4
50
50
increase with increasing xH from BC0 to saturation, reflecting
RGlc
âGlc
1720(40)
1970(90)
268(2)
4.8(5)
570(20)
24(2)
50
the decreasing binding ability of the receptors with the increasing
extent of saturation (see Supporting Information). We believe
that the BC50 and BC050 descriptors represent the most useful
and most general indicators of binding affinity.17
6780(50)c
7750(90)d
368(1)
19000(1000)c
11500 (300)d
790(20)
70(1)
RGal
1520(30)
1190(30)
1600(200)
6000(200)
2000(200)
2600(200)
âGal
120(1)
262(4)
660(10)
1179(3)
30(2)
RMan
43(1)
37(1)
72(7)
18(1)
It should be emphasized that, although not necessary, BC50
and BC050 can be appropriately employed in cases where a 1:1
association is the only equilibrium involved, allowing for a direct
comparison of results with those from more complex systems
and, thus, heterogeneous in nature. In this case, it has been
demonstrated7 that BC500 coincides with the dissociation con-
stant Kd ) 1/Ka, also known as the affinity constant, where Ka
is the association constant of the 1:1 binding equilibrium,
making apparent the chemical meaning of the intrinsic median
binding concentration, which can be viewed as a “global”
affinity constant.18
âMan
RGlcNAc
âGlcNAc
a Calculated from the log â values reported in Table 1. b Calculated from
log â values from ref 7. c Calculated from the log â values measured by
NMR in CD3CN. d Calculated from the log â values measured by ITC in
CH3CN.
Table 4. Intrinsic Median Binding Concentration BC050 (µM) with
Standard Deviation (σ) for Complexes of Receptors 2, 5, and 6
with Octylglycosides in CDCl3 at T ) 298 Ka
receptor
glycoside
BC050
(σ)
Glycoside Binding Affinity and Selectivity. Using the
affinity descriptors, we are now able to quantitatively evaluate
the binding ability of the pyrrolic tripodal receptors, which can
be most appropriately discussed in terms of intrinsic median
binding concentrations. The BC050 values calculated for 3 and 4
from â11, â21, and âdim of Table 1 are reported in Table 3
together with the corresponding values calculated for 1 from
previously reported data7 for direct comparison. In contrast to
1, whose affinities are in the millimolar range, BC050 values for
3 and 4 lie in the micromolar range, demonstrating a dramati-
cally improved binding ability with respect to the ureidic
progenitor. Besides the remarkable figure of 4.8 µM shown by
3 for âGlc, which improves on the affinity of 1 by over 400-
fold, in general, the observed affinities compete favorably with
data reported in the chemical literature, establishing the
described tripodal hosts as a new generation of highly effective
receptors for monosaccharides.
2
5
5
6
âGlc
RGlc
âGlc
âGlc
3690(50)
8300(300)
5400(100)
62000(1000)
a Calculated from the log â values reported in text.
are strongly bound to 4; on the contrary, âGlc is preferred by
3 by orders of magnitude with respect to the other mono-
saccharides.
To our knowledge, the selectivities shown by 3 for âGlc,
not only versus the R-anomer but also versus the other
monosaccharides, appear to be among the largest reported for
neutral synthetic receptors. The fact that âGlcNAc, which, like
âGlc, possesses all equatorial substituents, is bound only 6-fold
less effectively than âGlc indicates that the correct comple-
mentarity is achieved for equatorial H-bonding groups. An
analogous conclusion can be drawn for 4, for which âGlcNAc
is bound even more strongly than âGlc. In contrast, axial
hydroxyl groups seem to feature a mismatched binding geom-
etry, affecting 3 distinctly more than 4 and showing that
geometric and coordinative requirements are significantly more
strict for the former than for the latter. It can be concluded that
the pyrrolic tripodal architecture is well suited to preferentially
bind to the all-equatorial conformation of glucose and glu-
cosamine, while conformational restrictions imposed by the
imine double bonds of 3 significantly improve selectivity with
respect to the aminic receptor 4, as a result of a reduced
flexibility.
A peculiar feature emerging from Table 3 is that the aminic
receptor 4 is generally more effective than the iminic receptor
3, whereas the latter is distinctly more selective than the former.
Indeed, both are selective for âGlc, but selectivity spans a range
of over 30-fold for 4 and nearly 250-fold for 3. Except for RGlc
and RGal, for which lower affinities are observed, all glycosides
(17) Care must be paid to the presence of complex species multinuclear in the
analyte B. From BC50 equations (see Appendix in the Supporting Informa-
tion), it is apparent that, in contrast to all other species, they contribute to
the descriptor with a coefficient that depends on the stoichiometry of the
species. When comparing systems fitting different models and featuring
complex species multinuclear in B of different stoichiometry, the contribu-
tion of the latter species to BC50 will not be equally scaled unless weighted
for the corresponding coefficients. It must be stressed that this represents
a matter of concern only when complex species multinuclear in the analyte
B of different stoichiometry are present, for example, when comparing
binding abilities of two reagents A and A′ toward a common reagent B
and species AB2 and A′B3 are formed. In these cases, it is easily seen that
Comparison with the reference compounds highlights the
contributions to binding of the pyrrolic and the imino/amino
groups. The BC050 values calculated for 2, 5, and 6 from the
measured constants are reported in Table 4. Although the
triaminic receptor 2 is indeed capable of binding âGlc with
millimolar affinity, the improvement brought by the pyrrolic
binding sites is 770-fold for 3 and 150-fold for 4. It is also
worth noting that the affinity of 2 for âGlc is only 2-fold lower
than that of the triureidic receptor 1. It is evident that the
contribution of the pyrrolic substituents is substantial, whereas
converting the triaminic receptor to the corresponding triureidic
the contributions of these species to BC050 become indefinite, while those
to BC50 will not be equally weighted (see Appendix in the Supporting
Information). Since, most frequently, multinuclear complex species, if any,
appear for only one of the two reagents, typically the titrating agent, an
alternative approach in such cases would consist of calculating the BC50
value toward the reagent lacking multinuclear complex species.
(18) To expedite the calculation of the descriptors from the values of the binding
constants, the computer program “BC50 Calculator” has been developed
and made available for free upon request. The equations used by the
program to calculate BC50 values are described in the Appendix in the
Supporting Information.
9
4382 J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007

Pyrrolic Tripodal Receptors Recognizing Monosaccharides
A R T I C L E S
host 1 brings little advantage. Somewhat surprisingly, affinities
even lower than that for 1 were detected for the amido-pyrrolic
receptor 5, which also shows essentially no â/R selectivity
toward Glc, despite the presence of pyrrolic binding sites. A
clue to this unexpected result is provided by the triamidic
receptor 6, for which a drop of an order of magnitude in the
affinity for âGlc is observed with respect to 5 and is even more
pronounced with respect to 2, showing that the decreased affinity
can be ascribed to the amidic function, whereas pyrrole and
amino groups are both effective in binding the glycoside. It thus
emerges that aminic and pyrrolic binding groups are far better
H-bonding ligands than amidic groups, which are largely
employed in synthetic receptors for the recognition of carbo-
hydrates. We believe that the reason for this ability resides in
the good donor/acceptor H-bonding complementarity of amine
and pyrrole with glycosidic hydroxyls, whereas that of amide
groups is poorer than commonly thought.19
One comment is due to the results obtained in acetonitrile
(Table 3). Although a drop of 3 orders of magnitude in the
binding affinity for âGlc was observed with respect to chloro-
form, it is quite remarkable that an affinity at the millimolar
level could still be detected, testifying that the association is
strong enough to survive in pure acetonitrile. Calorimetric data
confirmed that the interaction of the pyrrolic receptors with âGlc
results in a distinct association event, enthalpic in origin, which
survives even in a polar solvent. Eventually, it should be
underlined that contribution of the spacer length to the binding
geometry appears unexpectedly critical; binding was completely
depleted by chain elongation of a single methylene group, even
though the receptor is nonpreorganized and binding relies on
adaptivity to the glycosidic guest.
Figure 2. ORTEP projections of the X-ray structure of 3‚EtOH. Left: side
view; right: top view. Ellipsoids are at 50% probability. Nitrogen and
oxygen atoms are represented as shaded ellipsoids. Hydrogen atoms
are omitted for clarity, except for those involved in H bonding. Selected
distances and angles: N-H‚‚‚O, 2.10 Å (161.6°); N(H)‚‚‚O, 2.93 Å;
O-H‚‚‚N, 1.76 Å (153.6°); O(H)‚‚‚N, 2.80 Å.
X-ray Studies. The X-ray structures of receptors 3 and 5b
supported the above conclusions, providing an insight into the
origin of the observed binding features. The ORTEP projections
of the structure of 3 crystallized from CHCl3/EtOH, depicted
in Figure 2, show the expected alternate arrangement of
substituents, with the three pyrrolic arms on the same side of
the aromatic ring forming a cleft, in the center of which is a
captured ethanol molecule. As anticipated, the iminic and the
pyrrolic nitrogen atoms lie coplanar in all of the three side chains
because of the conjugation of the iminic double bond with the
pyrrole ring; rotation about the CH2-NH single bond brings
one of the three arms to converge toward the inside of the cleft
and to chelate the alcoholic hydroxyl with the two nitrogen
atoms. The H-bonding chelate arrangement is noteworthy not
only for the nearly perfect planar geometry of the assembly but
also for the matched complementarity of the involved functional
groups, with the hydroxyl accepting one H bond from the pyrrole
NH and donating one H bond to the imine nitrogen; this way,
the chelating donor/acceptor diad of the receptor perfectly
matches the dual donor/acceptor nature of the hydroxyl group.
Quite remarkably, a nearly identical structure was observed
from crystals obtained from CHCl3/MeOH, indicating that the
H-bonded chelate with a hydroxylic species included in the cleft
represents a structural preference for the pyrrolic tripodal
receptor. Unfortunately, crystals suitable for X-ray structure
analysis could not be obtained for any of the adducts with the
Figure 3. ORTEP projections of the X-ray structure of 5b. Left: side view;
right: top view. Ellipsoids are at 50% probability. Nitrogen and oxygen
atoms are represented as shaded ellipsoids. Hydrogen atoms and a water
molecule of crystallization are omitted for clarity.
investigated glycosides; however, it is plausible that, in the
presence of glycosidic, guests of the appropriate size and
possessing appropriately located hydroxyl groups, all three of
the pyrrolic side chains may converge to cooperatively engage
more than one H bond, giving rise to a reinforced enthalpic
interaction and enhanced selectivities. In contrast to 3, a refined
X-ray structure could not be obtained for the aminic receptor
4. Preliminary diffraction data indicated that disorder in the
crystal structure prevented the refinement of data, most likely
caused by a much floppier configuration of the receptor and
the absence of solvent molecules bound in the cleft, although
the alternate substituent pattern appeared to be conserved. A
refined structure could instead be obtained for the amidic
receptor 5b, which resulted to be quite informative about the
binding properties of the amidic receptor family (Figure 3). The
main feature is the unexpected lack of an alternate arrangement
of substituents and thus of a binding cleft; only two pyrrolic
arms lie on one side of the aromatic ring, the third being located
(19) These results suggest that the binding of anionic carbohydrates reported
for the cationic tripodal receptor featuring amido-pyrrolic binding sites
(ref 6b) may be substantially ascribed to electrostatic interactions rather
than to H bonding.
9
J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007 4383

A R T I C L E S
Nativi et al.
on the other side together with the three ethyl groups. This
evidence demonstrates that the alternate pattern of substituents
is not a general feature of hexasubstituted benzenes. Of course,
in solution, all conformations are expected to be present, but
crystallographic evidence indicates that the alternate distribution
may not be the preferred conformation.20 A further feature is
relevant to binding properties; as expected, the amidic group is
coplanar with the pyrrole ring due to conjugation, but in contrast
to the iminic receptor, the pyrrolic NH lies on the same side of
the carbonyl and opposite the amidic NH in a preferred anti
configuration. Thus, if the latter would point toward the inside
of the virtual binding cleft, the former would point toward the
outside of it and would be, therefore, unable to bind. The
alternative binding mode, with the pyrrole NH and the carbonyl
as chelating elements, does not meet the geometrical require-
ments for binding, being located one bond further from the
appropriate geometry; moreover, the arrangement of the chelat-
ing groups would be energetically unfavorable when converg-
ing toward the inside of the cleft for binding, and because of
the repulsive interactions, it would induce between pairs of
carbonyls pointing toward each other. The latter factor may also
be responsible for the lack of an alternate conformation in the
solid state in order to minimize carbonyl group repulsion.
Finally, in the crystal packing, two contiguous molecules are
H bonded through a pair of short contacts between the pyrrolic
NH of one arm and the amidic CO of the corresponding other
molecule’s arm in a head-to-tail fashion. Since 5b was crystal-
lized from the same alcoholic solvent used for 3, this evidence
shows that amidic groups preferentially bind to pyrrolic NH
rather than to solvent hydroxyl groups and, most likely, to
carbohydrate hydroxyls. Altogether, X-ray studies showed that,
at least in the tripodal scaffold, the amidic group performs poorly
as a H-bonding ligand for carbohydrates for structural, geo-
metrical, and functional complementarity reasons.
Figure 4. (+)ESI-MS spectra of (bottom) 3 + âGlc, 0.2 mM each; m/z
481.3 [3+H]+, 773.5 [3‚âGlc+H]+, 961.6 [3‚3+H]+. (Top) 4 + âGlc, 0.2
mM each; m/z 487.3 [4+H]+, 779.5, [4‚âGlc+H]+, 973.7 [4‚4+H]+.
Solvent, CHCl3/CH3CN 1:1; ESI voltage, 6 kV; sampling cone potential,
56 V.
Mass Studies. Independent evidence of the recognition
properties of the pyrrolic tripodal receptors, in agreement with
binding studies in solution, was obtained in the gas phase from
mass experiments. In the positive ion mode ESI-MS spectrum
of an equimolar mixture of 3 and âGlc, the [3‚âGlc+H]+
complex was present as the major peak, after the base peak of
the free receptor, together with a peak of smaller intensity for
dimeric 3 (Figure 4, bottom). An analogous spectrum was
obtained by injecting an equimolar mixture of 3 and RGlc,
showing the same set of peaks in comparable intensities, with
the [3‚RGlc+H]+ complex present as the minor of the three
peaks. The observed relative intensities suggested the formation
of a more stable complex for âGlc than for RGlc. Experiments
performed under the same conditions on receptor 4 with RGlc
and âGlc gave very similar results, showing the corresponding
peaks of the receptor, of its dimer, and of the complex present
in comparable intensities for the two anomeric glycosides
(Figure 4, top). The common features exhibited by this set of
ESI-MS spectra, although just providing a qualitative picture,
demonstrated the presence of the complex species as a major
component for all of the investigated mixtures. A more
quantitative description of the relative affinities of 3 and 4 for
RGlc and âGlc could be obtained through collision-induced
dissociation (CID) experiments performed on a triple-quadrupole
Figure 5. CID MS/MS analysis of the complex detected as the [M+H]+
ion at m/z 779.5. Products, 779.5 ([M+H]+, dotted line), 487.3 ([4+H]+,
solid line). Solvent, CHCl3/CH3CN 1:1; ESI voltage, 6 kV; sampling cone
potential, 56 V; signal acquisition, 0.36 min; 53 scans over collision energies
from - 8.0 to - 34.0 eV in 0.5 eV steps (Q0 ) -12 eV); collision-gas
pressure, P ) 2.64 10-5 Torr. Top: 4 + âGlc, 0.2 mM each. Bottom: 4 +
RGlc, 0.2 mM each.
mass spectrometer. A scan of the intensity of the [4‚âGlc+H]+
and [4+H]+ ions originating from the ion of the complex
selected at m/z 779 with increasing potential gave the profiles
shown in Figure 5 (top), which crossed for a collision energy
value of 7.4 eV, corresponding to the energy required to
dissociate 50% of the complex under the specific experimental
conditions. The corresponding CID profiles originating from
the [4‚RGlc+H]+ ion under identical conditions (Figure 5,
bottom) exhibited a crossing point for a collision energy value
of 6.4 eV, showing that dissociation of the RGlc complex
required a collision energy smaller by 1 eV than that required
by the âGlc complex. CID profiles were analogously obtained
(20) This conformation in the solid state does not appear to be exceptional.
See, for example, ref 6c.
9
4384 J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007

Pyrrolic Tripodal Receptors Recognizing Monosaccharides
A R T I C L E S
for complexes of 3 with RGlc and âGlc run under identical
experimental conditions in order to obtain comparable results.
The corresponding profiles, similar in all respects to those
depicted in Figure 5, gave crossing points for collision energies
of 8.1 and 7.7 eV for âGlc and RGlc, respectively, demonstrat-
ing a stronger binding to the former glycoside also for 3 (see
Supporting Information). Quite gratifyingly, results in the gas
phase showed the same trend observed in solution; besides the
same â/R selectivity order, both glycoside anomers were more
strongly bound to 3 than to 4 in solution and in the gas phase.
and quantitatively assessed on a common scale by generalized
binding descriptors, that is, the median binding concentration
BC50 and the intrinsic median binding concentration BC500
parameters, which we developed for host-guest systems of any
nature and involving any number of binding constants, and
which we propose as the most useful and most general indicators
of binding affinity. These findings may open the way to a new
generation of carbohydrate receptors, featuring hitherto unex-
plored pyrrolic H-bonding donors, and to a new and general
way of assessing binding affinities.
Conclusion
Acknowledgment. The contribution of Dr. P. Gans, Univer-
sity of Leeds, and of Prof. A. Sabatini, Universita` di Firenze,
to the upgrade of the HypNMR program is gratefully acknow-
ledged. High-field NMR spectra were acquired at the Magnetic
Resonance Center (CERM) facilities of the Universita` di Firenze.
We thank Mr. M. Lucci for technical assistance. Ente Cassa di
Risparmio di Firenze (Italy) is acknowledged for granting a 400
MHz NMR spectrometer.
In the present work, we have shown that very effective
receptors for monosaccharides have been obtained by
assembling the correctly matching H-bonding groups on a
tripodal scaffold appropriately designed to form an adaptive cleft
around the carbohydrate moiety. Following previous findings,
we demonstrated that pyrrolic donors and iminic/aminic accep-
tors are far better H-bonding ligands for glycosidic hydroxyls
than the most largely employed amidic and ureidic groups,
showing geometric, coordinative, and functional complemen-
tarity to the donor/acceptor properties of aliphatic hydroxyls.
When correctly located and precisely spaced in the side arms
of the tripodal scaffold, since spacer length has been shown to
be crucial, these H-bonding groups gave rise to a new generation
of receptors showing micromolar affinities for monosaccharidic
glycosides in chloroform and millimolar affinities in a polar
solvent such as acetonitrile due to a strongly enthalpic interac-
tion. Remarkable selectivities were also observed, reaching
unprecedented selectivity factors for âGlc with the imino-
pyrrolic receptor. Affinities and selectivities were univocally
Supporting Information Available: General methods,
materials, and synthetic procedures for compounds 1-9. Data
tables, result pages, and plots of experimental and calculated
shifts from titrations of glycosides of Chart 2. Plots of ITC
titrations and CID profiles from ESI-MS experiments. Crystal-
lographic tables and details. Appendix: generalized treatment
of BC50 and BC050 for host-guests systems. Two-dimensional
NMR spectra. X-ray crystallographic data in the electronic
Crystallographic Information File (CIF) format. This material
is available free of charge via the Internet at http://pubs.acs.org.
JA068754M
9
J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007 4385