Role of the Symmetry of Multipoint Hydrogen Bonding on Chelate Cooperativity in Supramolecular Macrocyclization Processes

Article abstract of DOI:10.1002/anie.201508854

Herein, we analyze the intrinsic chelate effect that multipoint H-bonding patterns exert on the overall energy of dinucleoside cyclic systems. Our results indicate that the chelate effect is regulated by the symmetry of the H-bonding pattern, and that the

Full text of DOI:10.1002/anie.201508854

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201508854
German Edition: DOI: 10.1002/ange.201508854
Self-Assembly
Role of the Symmetry of Multipoint Hydrogen Bonding on Chelate
Cooperativity in Supramolecular Macrocyclization Processes
Carlos Montoro-García, Jorge Camacho-García, Ana M. López-PØrez, María J. Mayoral,
Nerea Bilbao, and David Gonzµlez-Rodríguez*
Abstract: Herein, we analyze the intrinsic chelate effect that
multipoint H-bonding patterns exert on the overall energy of
dinucleoside cyclic systems. Our results indicate that the chelate
effect is regulated by the symmetry of the H-bonding pattern,
and that the effective molarity is reduced by about three orders
of magnitude when going from the unsymmetric ADD–DAA
or DDA–AAD patterns to the symmetric DAD–ADA pattern.
ingly used to produce not only discrete cyclic assemblies, but
also supramolecular polymers and functional materials.[7] The
nucleobases are a relevant example,[8] and DNA itself,
composed of combinations of unsymmetric ADD–DAA
guanine–cytosine and symmetric DA–AD adenine–thymine
H-bonded Watson–Crick pairs, can be regarded as the
biological stereotype of a closed assembly. The relative
strength of these multipoint H-bonding interactions is now
well-understood since the investigations described by Jorgen-
sen and Pranata in 1990.[9] Their interpretation takes into
account secondary electrostatic interactions between contig-
uous centers to explain the trend in the association constants
of, for example, triply H- bonded pairs: DDD–AAA > ADD–
DAA > DAD–ADA. However, the intrinsic influence of the
H-bonding pattern on EM, and hence on the chelate
cooperativity of a cyclization process, has never been
addressed, and was the main focus of this study. We have
compared the thermodynamics of the self-assembly of three
related monomers (GC, iGiC, AU) into their respective cyclic
tetramers (cGC4, ciGiC4, cAU4; Figure 1). Our results indi-
cate a huge effect of the symmetry of the binding interaction
on the magnitude of EM, and may thus be highly valuable in
T
he supramolecular synthesis[1] of complex nanostructures
with a precision analogous to that found in the natural world
requires an understanding not only of the noncovalent
interactions involved,[2] but also of cooperative and multi-
valent phenomena that may arise between the individual
constituents, since the control of structure and monodispersity
depends largely on this issue.[3] A molecule with more than
one binding site may assemble into linear (open) or cyclic
(closed) structures. Although the size of linear oligomers can
sometimes be limited within a certain range, the supramolec-
ular product is commonly a statistical distribution of chain
lengths.[4] Therefore, the synthesis of discrete supramolecular
structures has normally been focused on closed (multi)-
macrocyclic systems, in which size and structure are dictated
by the geometric requirements of the monomer and the
binding interaction.[5] The effect that causes the quantitative
formation of a particular ring-closed species is defined as
chelate cooperativity and stems from the fact that an intra-
molecular interaction is favored over an intermolecular
interaction, providing that a series of conditions of enthalpic
and entropic origin are met.[3] The increased stability of
a cyclic oligomer relative to that of the corresponding linear
oligomer is given by the product Kinter·EM, in which Kinter is
the intermolecular binding constant and considers the addi-
tional association to form the macrocyclic ring, and EM, the
key parameter in the quantification of chelate cooperativity,
stands for effective molarity and takes into account that this
last binding event is intramolecular (= Kintra/Kinter).[6]
Table 1: Cyclotetramerization constants (KT), reference intermolecular
association constants (Kref), and effective molarities (EM) obtained for
GC/iGiC/AU from different experiments.
M
Solvent
KT [mÀ3
]
Kref[a] [mÀ1
]
EM [m]
GC
DMF
THF
2.3Æ0.8”105[b] 5.7Æ0.3
218
9.1Æ4.0”1014[c] 1.5Æ0.1”103 180
3.7Æ0.3”1015[d]
730
CHCl3
5.6Æ3.1”1020[e] 2.8Æ0.3”104 910
5.0Æ0.1”1020[f]
813
246
iGiC DMF
3.4Æ1.9”105[b] 6.1Æ0.8
THF
3.7Æ1.2”1015[c] 1.7Æ0.6”103 463
2.2Æ0.5”1015[d]
294
CHCl3
CHCl3
CHCl3/CCl4 (2:3)
3.3Æ0.4”1020[f] 3.2Æ0.5”104 314
2.5Æ0.4”102 0.10[g]
In this context, multipoint H-bonding motifs, constituted
by an array of vicinal H-bonding donor (D) and acceptor (A)
groups, arise as a relevant noncovalent interaction increas-
AU
9.4Æ0.3”1011[c] 2.0Æ0.4”103 0.06
2.8Æ0.2”1011[d]
0.02
CHCl3/acetone (5:1) 7.2Æ1.6”106[h] 0.9Æ0.6”102 0.11
[a] Determined from titration experiments with the mononucleosides:
G+C, iG+iC, A+U.[12] [b] Determined from 1H NMR dilution experi-
ments (see Figure S9). [c] Determined from UV/Vis dilution experiments
(see Figure S13). [d] Determined from temperature-dependent experi-
ments (see Figure S14). [e] Determined from 1H NMR competition
experiments (see Figure S15). [f] Determined from fluorescence com-
petition experiments (see Figure S16). [g] Estimated from the fitting of
the 1H NMR dilution data (see Figure S1 and Figure 2c). [h] Determined
from 1H NMR dilution experiments (see Figure S7B). DMF=N,N-
dimethylformamide, THF=tetrahydrofuran.
[*] C. Montoro-García, J. Camacho-García, Dr. A. M. López-PØrez,
Dr. M. J. Mayoral, N. Bilbao, Dr. D. Gonzµlez-Rodríguez
Nanostructured Molecular Systems and Materials Group
Departamento de Química Orgµnica
Facultad de Ciencias, Universidad Autónoma de Madrid
28049 Madrid (Spain)
E-mail: david.gonzalez.rodriguez@uam.es
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201508854.
Angew. Chem. Int. Ed. 2016, 55, 223 –227
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
223

Angewandte
Communications
Chemie
Figure 1. Structure of lipophilic dinucleoside monomers GC, iGiC, and AU, reference mononucleoside compounds G, C, iG, iC, A, and U, and the
three cyclic tetramers formed in solution: cGC4 (ADD–DAA), ciGiC4 (DDA–AAD) and cAU4 (DAD–ADA).
the future design of suitable monomers that lead quantita-
tively to either discrete closed assemblies or, at the other
extreme, to supramolecular polymers.
ꢀ 3 ” 104 mÀ1) in CHCl3.[12] This can be explained by the
different stabilizing/destabilizing secondary H-bonding inter-
actions between vicinal donor and acceptor groups in the
DAD–ADA (A:U) pair versus the DDA–ADD (G:C or
iG:iC) pair.[9] However, our experimental results (Table 1)
suggest that the decrease in stability of the corresponding
cyclic tetramer is actually much larger. As explained below,
our results indicate that the KT value for AU is not only
reduced by a decrease in Kref, but also by a substantial
decrease in the magnitude of EM.
To evaluate this hypothesis, we performed competition
experiments in which the corresponding complementary
pyrimidine mononucleoside (C/iC/U) was gradually added
to a solution of the associated tetramers (cGC4/ciGiC4/cAU4;
see Figures S15 and S16). The titration process was monitored
in solvents of low polarity by two different techniques:
1H NMR spectroscopy at high concentrations (ca. 10À2 m) and
emission spectroscopy at relatively low concentrations (ca.
10À4 m). Our results show that whereas cGC4/ciGiC4 can resist
up to 60 equivalents of the C/iC mononucleoside, cAU4 was
fully dissociated after the addition of about 3 equivalents of
U, regardless of the solvent system employed, thus indicating
The self-assembly of monomers GC, iGiC, and AU was
analyzed in different solvents by a number of concentration-
and temperature-dependent spectroscopic methods (1D and
1
2D H NMR spectroscopy, as well as absorption, emission,
and CD spectroscopy). These experiments were described in
our previous work[10] and the combined results for the 3
monomers are detailed in the Supporting Information of this
manuscript (see Figures S1–S16).[11] The equilibrium con-
stants of the cyclotetramerization processes (KT) could be
obtained in some cases and are collected in Table 1, together
with the reference association constants for the interaction
between the complementary mononucleosides (Kref). The
whole set of results clearly demonstrate that AU forms
considerably less stable cyclic tetramers than GC or iGiC. On
the other hand, GC or iGiC show comparable qualitative and
quantitative association behavior.
This stability trend was expected, since the individual A:U
binding constant (Kref ꢀ 2.5 ” 102 mÀ1) is typically about two
orders of magnitude lower than that for G:C or iG:iC (Kref
224
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 223 –227

Angewandte
Communications
Chemie
a much weaker chelate effect in cAU4 (Figure 2a). These
curves for each dinucleoside in different solvent systems
experiments, in which the intramolecular and intermolecular
base-pair binding events are made to compete, constitute the
most appropriate way to detach the intrinsic contribution of
the chelate cooperativity from the overall energy of the
system. As a matter of fact, the EM of the system can be
inversely related to the competition equilibrium constant
(KC) as: EM = 1/KC.[11]
The EM values for each cyclotetramerization process
(Table 1) were calculated from the KT values and the
corresponding Kref constants obtained in the different experi-
ments[10,12] by use of the relationship: EM = KT/Kref4. These
Kref and EM values were also used to simulate speciation
(Figure 2b,c). These curves, which relate the concentration of
each supramolecular species to the total concentration,
reproduce quite satisfactorily the dissociation trends observed
for cGC4/ciGiC4/cAU4 in dilution experiments.[11] In all cases,
cGC4 and ciGiC4 fulfill the condition Kref·EM > 185·n (in
which n is the number of monomers in the cycle; n = 4), as
defined by Ercolani for complete cycle assembly.[6] However,
the concentration range in which this condition is met is
clearly wider in solvents that do not compete strongly for
H bonding and thus maintain a high Kref value. On the other
hand, EM values had to be set between 0.05 and 0.1 to
reproduce appropriately the dilution trends of cAU4 in three
solvent systems. They are thus more than
three orders of magnitude lower than
those of the GC/iGiC cyclization process.
As a result, and in line with our exper-
imental observations, the condition
Kref·EM > 185·n is hardly fulfilled by AU
even in the highly apolar CHCl3/CCl4
(2:3) mixture, in which Kref is enhanced.
All monomers share a common, rigid
structure that was designed to produce
cyclic square-shaped assemblies devoid of
strain and with minimal conformational
entropy loss. These properties were made
possible by the 908 angle that the 8-purine
and 5-pyrimidine positions adopt upon
Watson–Crick complementary base pair-
ing and by the use a rigid central block to
connect the bases with only four rotatable
linear p-conjugated bonds. Rotation
about these bonds can produce different
conformations in which the Watson–Crick
edges alternate between syn and anti
relative arrangements (Figure 3a). How-
ever, cycle formation demands that the
Watson–Crick edges are in a syn relative
conformation. This is a degree of freedom
that is lost when comparing cyclic and
open GC, iGiC, and AU oligomers, and
must contribute to a reduction, of entropic
origin, in the maximum attainable EM of
the cyclic system. Now, the AU monomer,
which contain complementary nucleosides
that pair with a symmetric DAD–ADA H-
bonding pattern, have the additional pos-
sibility to self-assemble through either
Watson–Crick or reverse Watson–Crick
interactions (Figure 3a). Each binding
mode provides a different association
angle (908 and 2108), and their relative
energy is assumed to be comparable, as
Figure 2. a) Competition experiments. Plots of the degree of cGC4, ciGiC4, and cAU4
association, as measured by either 1H NMR or fluorescence spectroscopy, as a function of the
equivalents of C/iC/U added. b,c) Simulated speciation curves (lines) and experimental
dilution data (circles/triangles/squares) indicating the degree of cM4 association of b) GC/
iGiC in CHCl3 (see Figure S1), THF (see Figure S13A,B), and DMF (see Figure S9), and c) AU
in CHCl3/CCl4 (2:3; see Figure S13C), CHCl3 (see Figure S1), and CHCl3/acetone (5:1; see
Figure S7B).
previous studies with the adenine–thy-
mine pair have demonstrated.[13,14] This
property introduces additional degrees of
freedom, not available in the unsymmetric
ADD–DAA or DDA–AAD patterns, that
allow the linear AUn oligomers to access
Angew. Chem. Int. Ed. 2016, 55, 223 –227
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
225

Angewandte
Communications
Chemie
a higher number of binding and conforma-
tional possibilities. However, such freedom
must be lost upon cycle formation because
the cyclotetramerization process exclu-
sively demands a 908 Watson–Crick inter-
action. Hence, we assume that the entropy
loss associated with cyclization becomes
much larger for cAU4 than for cGC4 or
ciGiC4, thus resulting in a supplementary
and notable reduction of the EM values.[15]
It would have been highly interesting to
compare the cyclization process of our AU,
GC, and iGiC molecules with that of
a related dinucleoside with complementary
DDD–AAA
base
pairs.
However,
a purine–pyrimidine couple with such a H-
bonding pattern that would maintain the
same geometric requirements is simply not
available. We propose, given the conclu-
sions drawn from this study, that such
a cyclic tetramer bound by symmetric
DDD–AAA pairs would have both a high
Kref value (ca. 105–106 m in CHCl3, as
reported previously)[7a] and a low EM
(0.01–0.1, comparable to that of cAU4),
owing to the possibility of binding with two
different angles. From these values, we
simulated in Figure 3b the speciation
curves of this hypothetical symmetric
DDD–AAA system in CHCl3 and com-
pared them with those obtained for unsym-
metric ADD–DAA/DDA–AAD and sym-
metric DAD–ADA H-bonded systems.
It is clear that the use of an ADD–
DAA/DDA–AAD H-bonding pattern, pro-
viding moderate Kref and high EM values,
leads to cyclic tetramer assemblies that
persist as the main species in solution over
a much broader concentration range. Only
at very low concentrations is cM4 dissoci-
ated as a monomer, but no other associated
species is seen to compete within the 10À8–
105 m concentration range, thus underlining
the strong all-or-nothing behavior of this H-
bonded system. At very high concentra-
tions, intra- and intermolecular processes
begin to compete, and above 105 m linear
polymers of high molecular weight grow at
the expense of the cM4 species. On the
other hand, the DDD–AAA pattern leads
to more strongly bound assemblies along
the whole concentration scale as a result of
a high Kref value, and the monomer is only
present at concentrations below 10À5 m. The
cM4 species is dominant in the 10À5–1m
range; however, in sharp contrast to the
unsymmetric pattern, higher-order linear
oligomers begin to compete strongly at
moderate concentrations. Finally, in the
Figure 3. a) Unsymmetric versus symmetric H-bonding patterns. b) Simulated speciation
curves for the H-bonded self-assembly of hypothetical DDD–AAA (Kref =106 mÀ1
;
EM=0.05m), ADD–DAA/DDA–AAD (Kref =104 mÀ1; EM=500m), and DAD–ADA
(Kref =102 mÀ1; EM=0.05m) ditopic monomers with identical geometrical features to those
examined in this study. For the sake of clarity, only a few supramolecular species are
represented within the 10À8–108 m concentration range: M, M2, M3, M4, cM4, M5, and, as
examples of higher-order H-bonded linear oligomers: M10, M15, and M20.
226
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 223 –227

Angewandte
Communications
Chemie
weaker DAD–ADA H-bonded system, no associated species
is seen below 10À4 m. At intermediate concentrations, a dis-
tribution of small oligomers, among which cM4 is one of the
main species, is observed. In analogy to the DDD–AAA
system, when the concentration is increased above 1m, the
higher-order linear oligomers dominate. However, high-
molecular-weight distributions are attained much faster with
the DDD–AAA system than with the DAD–ADA H-bond-
ing pattern, as a result of a Kref value that is about four orders
of magnitude higher. It should be noted that the quantitative
results derived from Figure 3b must be strictly applied to the
specific monomer geometries investigated in this study.
In summary, in this study we have been able to dissect and
analyze independently the contributions of the H-bonding
strength between complementary nucleobases, as explained
by the Jorgensen model, and the intrinsic chelate effect that
they exert in cyclic systems. The results presented clearly
demonstrate that cyclic assemblies constructed from sym-
metric DAD–ADA H-bonding pairs are much less stable than
the homologues from unsymmetric ADD–DAA or DDA–
AAD pairs. On one hand, the DAD–ADA bonding pattern
reduces considerably the enthalpy of intermolecular associ-
ation owing to the absence of attractive secondary interac-
tions between vicinal H-bonding groups. On the other, the
symmetry of this pattern introduces the possibility of multiple
binding modes and hence a higher number of degrees of
freedom in linear oligomers, which are then lost upon
macrocyclization. This effect, of entropic origin, has a large
impact on the EM of the system; in our case, the EM was
reduced by about three orders of magnitude. Our conclusions
could in principle be extended to many linear or cyclic
supramolecular systems assembled through multipoint bind-
ing interactions. If a discrete, well-defined closed architecture
is to be designed, rigid monomers with a suitable geometry in
combination with an unsymmetric binding motif should be
used to enhance the EMs of the cyclization process. In other
words, the binding interaction should also contribute to the
preorganization of the system towards a specific structure,
thus reducing the degrees of freedom of any other compet-
itive supramolecular species. If, on the other hand, linear
supramolecular polymers are pursued, a strong symmetric
binding interaction would be the best choice to minimize
chelate cooperativity and hence the tendency of the supra-
molecular system to form undesired cycles.
[1] a) L. J. Prins, D. N. Reinhoudt, P. Timmerman, Angew. Chem.
Int. Ed. 2001, 40, 2382 – 2426; Angew. Chem. 2001, 113, 2446 –
2492; b) D. N. Reinhoudt, M. Crego-Calama, Science 2002, 295,
2403 – 2407.
[2] D. Gonzµlez-Rodríguez, A. P. H. J. Schenning, Chem. Mater.
2011, 23, 310 – 325.
[3] a) J. D. Badjicä, A. Nelson, S. J. Cantrill, W. B. Turnbull, J. F.
Stoddart, Acc. Chem. Res. 2005, 38, 723 – 732; b) Focus Issue on
Cooperativity, Nat. Chem. Biol. 2008, 4, 433 – 507; c) C. A.
Hunter, H. L. Anderson, Angew. Chem. Int. Ed. 2009, 48, 7488 –
7499; Angew. Chem. 2009, 121, 7624 – 7636; d) G. Ercolani, L.
Schiaffino, Angew. Chem. Int. Ed. 2011, 50, 1762 – 1768; Angew.
Chem. 2011, 123, 1800 – 1807.
[4] T. F. A. de Greef, M. M. J. Smulders, M. Wolffs, A. P. H. J.
Schenning, R. P. Sijbesma, E. W. Meijer, Chem. Rev. 2009, 109,
5687 – 5754.
[5] a) M. Fujita, M. Tominaga, A. Hori, B. Therrien, Acc. Chem. Res.
2005, 38, 371 – 380; b) M. J. Mayoral, N. Bilbao, D. Gonzµlez-
Rodríguez,
ChemistryOpen
2015,
DOI:
10.1002/
open.201500171.
[6] a) G. Ercolani, J. Phys. Chem. B 1998, 102, 5699 – 5703; b) G.
Ercolani, J. Phys. Chem. B 2003, 107, 5052 – 5057; c) G. Ercolani,
Struct. Bonding (Berlin) 2006, 121, 167 – 215; d) H. Sun, C. A.
Hunter, C. Navarro, S. Turega, J. Am. Chem. Soc. 2013, 135,
13129 – 13141.
[7] a) T. F. A. de Greef, E. W. Meijer, Nature 2008, 453, 171 – 173;
b) B. A. Blight, C. A. Hunter, D. A. Leigh, H. McNab, P. I. T.
Thomson, Nat. Chem. 2011, 3, 244 – 248; c) S. K. Yang, S. C.
Zimmerman, Isr. J. Chem. 2013, 53, 511 – 520.
[8] a) S. Sivakova, S. J. Rowan, Chem. Soc. Rev. 2005, 34, 9 – 21;
b) J. L. Sessler, C. M. Lawrence, J. Jayawickramarajah, Chem.
Soc. Rev. 2007, 36, 314 – 325; c) M. Fathalla, C. M. Lawrence, N.
Zhang, J. L. Sessler, J. Jayawickramarajah, Chem. Soc. Rev. 2009,
38, 1608 – 1620.
[9] W. L. Jorgensen, J. Pranata, J. Am. Chem. Soc. 1990, 112, 2008 –
2010.
[10] a) C. Montoro-García, J. Camacho-García, A. M. López-PØrez,
N. Bilbao, S. Romero-PØrez, M. J. Mayoral, D. Gonzµlez-
Rodríguez, Angew. Chem. Int. Ed. 2015, 54, 6780 – 6784;
Angew. Chem. 2015, 127, 6884 – 6888; b) S. Romero-PØrez, J.
Camacho-García, C. Montoro-García, A. M. López-PØrez, A.
Sanz, M. J. Mayoral, D. Gonzµlez-Rodríguez, Org. Lett. 2015, 17,
2664 – 2667.
[11] See the Supporting information for further details.
[12] J. Camacho-García, C. Montoro-García, A. M. López-PØrez, N.
Bilbao, S. Romero-PØrez, D. Gonzµlez-Rodríguez, Org. Biomol.
Chem. 2015, 13, 4506 – 4513.
[13] a) R. L. Ornstein, J. R. Fresco, Proc. Natl. Acad. Sci. USA 1983,
80, 5171 – 5175; b) S. N. Rao, P. A. Kollman, Biopolymers 1986,
25, 267 – 280.
[14] Additional A–U binding modes could be considered that do not
differ much in association strength from the Watson–Crick
mode, such as the double H-bonding interaction of the U base
with the 2-aminoadenine Hoogsten edge.
Acknowledgements
[15] It is interesting to note that related cyclic tetramers based on
pyridine–metal coordination (a strong single-point interaction
that allows for some degree of torsional and rotational flexi-
bility) afford EM values that are intermediate between our
symmetric and unsymmetric patterns (EM = 0.1–20m). As
previous studies on supramolecular systems have also shown
(see Ref. [6d]), these EM values do not seem to follow any clear
relationship with respect to solvent polarity or H-bond strength.
Funding from the European Union (ERC-Starting Grant
279548) and MINECO (CTQ2011-23659 and CTQ2014-
57729-P) is gratefully acknowledged.
Keywords: chelate effect · effective molarity ·
noncovalent synthesis · nucleoside self-assembly ·
supramolecular chemistry
How to cite: Angew. Chem. Int. Ed. 2016, 55, 223–227
Angew. Chem. 2016, 128, 231–235
Received: September 21, 2015
Published online: November 20, 2015
Angew. Chem. Int. Ed. 2016, 55, 223 –227
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
227