Quantification of cooperativity in the self-assembly of H-bonded rosettes

Article abstract of DOI:10.1039/d0ob00068j

The self-assembly of triaminopyrimidines with barbiturates and with cyanates was investigated in chloroform solution. Equimolar mixtures of two complementary components form stable macrocyclic 3:3 complexes (rosettes). The thermodynamics of self-assembly were quantified by using ¹H NMR titrations to measure the strength of pairwise H-bonding interactions between two rosette components (K), allosteric cooperativity associated with formation of a second H-bonding interaction with each component, and the effective molarity for cyclisation of the rosette motif (EM). Pyrimidine-cyanurate interactions are an order of magnitude more favourable than pyrimidine-barbiturate interactions, so the cyanurate rosettes are significantly more stable than barbiturate rosettes. There is no allosteric cooperativity associated with rosette formation, but the chelate cooperativity quantified by the product K EM is exceptionally high (10²-10⁴), indicating that there are no other species present that compete with rosette assembly. The values of EM for rosette formation are approximately 2 M for all four rosettes studied and are not affected by differences in peripheral substituents or intrinsic H-bond strength.

Full text of DOI:10.1039/d0ob00068j

Organic &
Biomolecular Chemistry
View Article Online
View Journal
PAPER
Quantification of cooperativity in the self-
assembly of H-bonded rosettes†
Cite this: DOI: 10.1039/d0ob00068j
Petr Motloch
and Christopher A. Hunter
*
The self-assembly of triaminopyrimidines with barbiturates and with cyanates was investigated in chloro-
form solution. Equimolar mixtures of two complementary components form stable macrocyclic 3 : 3
complexes (rosettes). The thermodynamics of self-assembly were quantified by using 1H NMR titrations to
measure the strength of pairwise H-bonding interactions between two rosette components (K), allosteric
cooperativity associated with formation of a second H-bonding interaction with each component, and
the effective molarity for cyclisation of the rosette motif (EM). Pyrimidine–cyanurate interactions are an
order of magnitude more favourable than pyrimidine–barbiturate interactions, so the cyanurate rosettes
are significantly more stable than barbiturate rosettes. There is no allosteric cooperativity associated with
rosette formation, but the chelate cooperativity quantified by the product K EM is exceptionally high
(102–104), indicating that there are no other species present that compete with rosette assembly. The
values of EM for rosette formation are approximately 2 M for all four rosettes studied and are not affected
by differences in peripheral substituents or intrinsic H-bond strength.
Received 13th January 2020,
Accepted 11th February 2020
DOI: 10.1039/d0ob00068j
rsc.li/obc
and barbiturates have been described,9–12 the parameters that
govern self-assembly of these exceptionally stable supramole-
Introduction
Self-assembly of multivalent supramolecular systems is an cular architectures have never been measured.
important process in chemistry, biology and materials science.
Although a large number of synthetic systems that form well-
defined discrete self-assembled complexes have been reported,
the factors that govern these processes are still poorly under-
Approach
As shown in Fig. 1, triaminotriazines are commonly used for
stood. For this reason, the design of new supramolecular
motifs with predictable properties remains a challenge. The
key to successful self-assembly is high chelate cooperativity,
which can be quantified by the product K EM, where K is the
association constant for the intermolecular interaction
between two components of the assembly, and EM is the
effective molarity for intramolecular interactions that give rise
to (poly)macrocyclic closed structures.1,2 The values of K are
relatively straightforward to predict based on the chemical
structures of the interacting partners,3 but the relationship
between EM and chemical structure is more problematic.2,4–8
As part of a programme to quantify the values of EM in stable
H-bonded assemblies, we describe here a detailed thermo-
dynamic analysis of the rosette motif first reported by
Whitesides (Fig. 1).9,10 Although many supramolecular assem-
blies based on the 3 : 3 complex formed by triaminotriazines
the self-assembly of H-bonded rosettes,11,12 but replacement of
this component by the corresponding pyrimidine offers some
advantages that we exploit here. Pyrimidines do not suffer
from the complications associated with the presence of mul-
tiple triaminotriazine rotamers,13 and the pyrimidine aromatic
proton provides a useful probe for studying supramolecular
self-assembly using 1H NMR spectroscopy. Fig. 2 illustrates
self-assembly of a H-bonded rosette via a sequence of inter-
molecular interactions to give linear oligomers, followed by
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2
1EW, UK. E-mail: herchelsmith.orgchem@ch.cam.ac.uk
†Electronic supplementary information (ESI) available: Materials and methods,
detailed synthetic procedures, full characterization including 1H and 13C NMR
spectra of all compounds and titration data. See DOI: 10.1039/d0ob00068j
Fig. 1 A H-bonded rosette self-assembled from a triaminotriazine and
a barbiturate. R is a substituent.
This journal is © The Royal Society of Chemistry 2020
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 2 Self-assembly of a H-bonded rosette from a triaminopyrimidine
and a barbiturate. Intermolecular triple H-bond interactions (K) lead to
the assembly of linear oligomers, and the 3 : 3 complex can cyclise to
form the rosette. EM is the effective molarity for the intramolecular
interaction involved in cyclisation. R is a substituent.
Fig. 4 Complexes used to quantify allosteric cooperativity in self-
assembly of H-bonded rosettes. (a) Allosteric cooperativity between the
two binding sites on a pyrimidine is α = K1/K2. (b) Allosteric cooperativity
between the two binding sites on a barbiturate is α = K1/K2. R is a
substituent.
intramolecular cyclisation of the 3 : 3 complex. The overall
stability of the resulting assembly depends on the association
constant for the intermolecular interactions (K) and the
effective molarity for the intramolecular process (EM). The
value of K can be estimated by measuring the association con-
stant for a suitable reference system that can only form a 1 : 1
complex (Kref ). Here we measure Kref using the complex illus-
trated in Fig. 3, where one of the hydrogen-bonding sites on
each of the two rosette components is blocked by a substitu-
ent. Although this complex could further oligomerise, the
intermolecular interactions would involve doubly H-bonded
contacts which are relatively weak.14 By measuring Kref and the
overall equilibrium constant for self-assembly of the rosette
from the two components (Krosette), the value of EM can be
determined using eqn (1).
ation constants for formation of the 1 : 1 (K1) and the 1 : 2 (K2)
complexes shown in Fig. 4 gives the allosteric cooperativity
factor α (eqn (2)).
K2
K1
α ¼
ð2Þ
Results and discussion
Synthesis
The triaminopyrimidines P and P′ were each synthesised in
one step (Scheme 1). Synthesis of the blocked-pyrimidines bP
and bP′ is shown in Scheme 2. Reacting 4,6-dichloropyrimidin-
2-amine with N-methylpentylamine gave the singly substituted
product bP-Cl, which was converted to bP by heating in neat
n-pentylamine under microwave irradiation. For the synthesis
Krosette
EM ¼
ð1Þ
6
KσKref
where the statistical factor Kσ = 32/3 is determined by the sym- of bP′, bTBA was prepared by a reaction of tert-butyl aniline
metry of the system (see Fig. S20†).15–18
with acetic anhydride followed by reduction using LiAlH4.
However, there is a complication with the rosette structure, bTBA was then used for the synthesis of the singly substituted
because some of the functional groups are involved in bifur- product bP′-Cl, which was converted to bP′ by heating in neat
cated H-bonding interactions, which may differ from the tert-butyl aniline under microwave irradiation.
simple pairwise H-bonding interactions present in the refer-
Commercially available diethyl-barbiturate was not soluble
ence complex shown in Fig. 3. Therefore additional experi- enough in chloroform for NMR titrations, and therefore bar-
ments are required to quantify the extent of any allosteric biturate B with longer alkyl chains was synthesised from urea
cooperativity associated with these interactions. The 1 : 2 com- (Scheme 3). bB was synthesised by using N-methylurea instead
plexes shown in Fig. 4 allow direct measurement of the allo- of urea (Scheme 3). Cyanurate C was prepared from cyanuric
steric cooperativity associated with formation of multiple inter-
actions with either the pyrimidine (Fig. 4a) or the barbiturate
component (Fig. 4b). Comparison of the microscopic associ-
Fig. 3 Complexes used to quantify the intermolecular H-bonding inter-
actions between two rosette components (Kref ). R is a substituent.
Scheme 1 Synthesis of pyrimidines.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Organic & Biomolecular Chemistry
Paper
measured with cyanurate C are approximately one order of
magnitude higher than the corresponding values for barbitu-
rate B.
Measurement of allosteric cooperativity (α) in 1 : 2 complexes
To determine the values of α for different combinations of
rosette components, 1H NMR titrations were carried out in
CDCl3 at 298 K using the ditopic pyrimidines (P or P′) as hosts
and blocked-barbiturate/cyanurate (bB or bC) as guests, or the
ditopic barbiturate/cyanurate (B or C) as hosts and blocked-
pyrimidines (bP or bP′) as guests. In order to assess whether
there was any deviation of the titration data from a simple
non-cooperative isotherm, the data were analysed in three
different ways:
Scheme 2 Synthesis of blocked pyrimidines.
1. A non-cooperative binding isotherm, which assumes that
only two variables can be used to fit the data. The first variable
is defined as Knc, where Knc = K1 = K2. The second variable was
the change in chemical shift for formation of the 1 : 1 complex
that was set to be identical to the subsequent change in chemi-
cal shift for formation of the 1 : 2 complex.
2. A non-cooperative binding isotherm assuming that K1 =
K2 = Kref, which was fixed at the value for the corresponding
1 : 1 complex in Table 1. The change in chemical shifts for-
mation of the 1 : 1 complex and the subsequent change in
chemical shift for formation of the 1 : 2 complex were fitted to
the data as two independent variables.
3. A binding isotherm that allows for cooperativity with K1 =
Kref, which was fixed at the value for the corresponding 1 : 1
complex in Table 1, and K2 a variable that was fitted to the
data. The change in chemical shifts formation of the 1 : 1
complex and the subsequent change in chemical shift for for-
mation of the 1 : 2 complex were fitted to the data as two
additional variables.
Scheme 3 Synthesis of cyanurates and barbiturates.
acid and iodopentane, and bC was also obtained as a side
product in this reaction (Scheme 3).
Measurement of Kref in 1 : 1 complexes
Table 2 summarises the results along with the global error
E of the fit. Isotherm 3 uses an additional variable in the fit
compared with isotherms 1 and 2, and so the value of E is
always lower for the cooperative isotherm. When the ditopic
pyrimidines were used as hosts, the titration data fit well to all
three isotherms, and the values of Knc and K2 are similar to
the corresponding values of Kref measured for the simple 1 : 1
complexes. These results suggested that the value of α is
approximately one in all cases, i.e. the allosteric cooperativity
To determine the values of Kref for different combinations of
rosette components, 1H NMR titrations were carried out in
CDCl3 at 298 K using blocked-pyrimidines (bP or bP′) as hosts
and blocked-barbiturate/cyanurate (bB or bC) as guests. 1H
NMR dilution experiments show that the pyrimidines do not
self-associate to a significant extent at the concentrations used
in these experiments (K ≈ 1 M−1, see Fig. S5 and S6†).
Similarly, the self-association constants for barbiturates and
cyanurates in chloroform are too low to affect the titration
experiments (K < 10 M−1).11 The titration data fit well to a 1 : 1
binding isotherm in all cases (see Fig. S7–S10†), and the
results are summarised in Table 1. The association constants
Table 2 Association constants (Knc and K2) and global errors (E)
obtained by fitting 1H NMR titration data measured in CDCl3 at 298 K to
different isothermsa
Table 1 Association constants measured in CDCl3 at 298 K by 1H NMR
Isotherm 1
Complex Knc [M−1
Isotherm 2
E [%] Kref [M−1
Isotherm 3
E [%] K2 [M−1
titrationsa
]
]
]
E [%]
Complex
Kref [M−1
]
P·bB2
P′·bB2
P·bC2
P′·bC2
200 10
220 20
3700 200
690 100
1400 200
0.7
0.9
2
242
200 10
3200 40
1200 200 0.1
200 10
5
0.7
0.9
1
334
320 40
1600 10 0.5
1200 30 0.1
5
0.1
0.3
bP·bB
bP′·bB
bP·bC
bP′·bC
240
200 10
3200 40
5
0.5
4
1200 200 B·bP′2
1
200 50
1
a Errors are quoted at the 95% confidence limit.
a Errors are quoted at the 95% confidence limit.
This journal is © The Royal Society of Chemistry 2020
Org. Biomol. Chem.

View Article Online
Paper
Organic & Biomolecular Chemistry
associated with self-assembly of H-bonded rosettes is negli- rate NH group and two signals at 8–11 ppm due to the
gible. When the ditopic barbiturate/cyanurate was used as H-bonded pyrimidine NH and NH2 groups.9 At 298 K, the
hosts, the titration data generally did not fit well to any of the spectra were much broader. However, the H-bonded rosette
isotherms, which indicates that additional equilibria are signals were still observed for the cyanurate C mixtures, which
present, so these systems cannot be used to measure α. indicates that these rosettes are kinetically more stable than
However, the titration data for B·bP′2 did fit well to the non- the corresponding barbiturate B rosettes. This result is consist-
cooperative isotherm 2 and to isotherm 3 giving a value of K2 ent with the higher value of Kref measured for the cyanurate
identical to K1 and Kref. The fit to isotherm 1 for this 1 : 1 complexes compared with the barbiturate 1 : 1 complexes
complex is poor, as judged by the large value of E, which (Table 1).
implies that chemical shift changes differ significantly for the
Equilibrium constants for self-assembly of the H-bonded
first and second binding event. Since no allosteric rosettes were determined by 1H NMR titrations in CDCl3 at
cooperativity was observed in the formation of any of the 1 : 2 298 K using pyrimidines (P or P′) as hosts and barbiturate/cya-
complexes, the values of Kref from Table 1 were used to nurate (B or C) as guests. The signal due to the aromatic
assess the chelate cooperativity associated with rosette self- proton of the pyrimidine stayed sharp for the whole titration,
assembly.
and the changes in chemical shift fit well to an all-or-nothing
3 : 3 isotherm in all cases (see Fig. S16–S19†). Fitting to an
alternative isotherm that also allowed for opening of the
rosette to give 1 : 2 complexes in the presence of excess guest
did not change the result, because the 1 : 2 complexes are not
significantly populated at the concentrations used. The values
of Krosette are summarised in Table 3. Rosettes based on cyanu-
rates are significantly more stable than rosettes based on bar-
biturates, because of the order of magnitude difference in the
intrinsic H-bond strength, i.e. Kref. However, the values of EM
calculated using eqn (1) are practically identical (1–3 M) for all
four rosettes. Whitesides used the peripheral crowding associ-
ated with t-butylphenyl substituents in P′ to favour rosette for-
mation in the crystalline state,9 but the results in Table 3
show that in solution these bulky groups do not increase the
EM for rosette formation compared with simple alkyl
substituents.
Measurement of chelate cooperativity (EM) in 3 : 3 rosettes
To establish rosette formation, millimolar solutions of 1 :
1 mixtures of two complementary components (namely P + B,
P′ + B, P + C or P′ + C) were prepared in CDCl3 and the 1H
NMR spectra were recorded at 298 K and 233 K (Fig. 5). At
233 K, all of the mixtures showed the characteristic rosette
signal at 14–15 ppm due to the H-bonded barbiturate/cyanu-
The value of EM for the H-bonded rosettes is at the high
end of the values reported in the literature for supramolecular
systems,2 but it is substantially lower than the highest values
of EM (102–103 M) found for H-bonded squares based on gua-
nosine-cytosine base-pairing. The statistical correction Kσ
lowers the value of EM measured for the rosettes by a factor of
10, but even without this correction, the EM would be orders
of magnitude lower than the values reported for the González-
Rodríguez systems.19 The chelate cooperativity associated with
rosette formation is quantified by the product Kref EM, which
is the equilibrium constant between the open and closed 3 : 3
complexes shown in Fig. 2. The values of Kref EM in Table 3
are 102–103 for the barbiturate rosettes and 103–104 for the cya-
Table 3 Equilibrium constants and effective molarities for self-assem-
bly of H-bonded rosettes in CDCl3 at 298 K measured by 1H NMR
titrationsa
Complex
Krosette [M−5
]
Kref [M−1
]
EM [M]
Kref EM
P3·B3
P′3·B3
P3·C3
P′3·C3
(6.3 0.1) × 1015
(8.4 0.4) × 1014
(2.7 0.2) × 1022
(3.0 0.1) × 1019
240
200 10
3200 40
1200 200
5
3.1 0.4
1.2 0.4
2.4 0.2
1.0 0.9
740 80
250 60
7500 700
1100 900
Fig. 5 Partial 1H NMR spectra (500 MHz, CDCl3) of rosette assemblies
at (a) 298 K and (b) 233 K. The concentration of each component was
3 mM. Barbiturate/cyanurate NH signal highlighted in red, and pyrimi-
dine NH/NH2 signals highlighted in blue.
a Errors are quoted at the 95% confidence limit.
Org. Biomol. Chem.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Organic & Biomolecular Chemistry
Paper
nurate rosettes, which means that less than 1% of the species
present are not in the form of fully assembled rosettes.
Notes and references
1 C. A. Hunter and H. L. Anderson, Angew. Chem., Int. Ed.,
2009, 48, 7488.
2 P. Motloch and C. A. Hunter, in Adv. Phys. Org. Chem, ed.
I. H. Williams and N. H. Williams, Academic Press, 2016,
vol. 50, pp. 77–118.
3 C. A. Hunter, Angew. Chem., Int. Ed., 2004, 43, 5310.
4 C. A. Hunter, M. C. Misuraca and S. M. Turega, J. Am.
Chem. Soc., 2011, 133, 20416.
5 C. A. Hunter, M. C. Misuraca and S. M. Turega, Chem. Sci.,
2012, 3, 2462.
6 H. Adams, E. Chekmeneva, C. A. Hunter, M. C. Misuraca,
C. Navarro and S. M. Turega, J. Am. Chem. Soc., 2013, 135,
1853.
7 H. Sun, C. A. Hunter and E. M. Llamas, Chem. Sci., 2015, 6,
1444.
8 M. A. Jinks, H. Sun and C. A. Hunter, Org. Biomol. Chem.,
2014, 12, 1440.
9 G. M. Whitesides, E. E. Simanek, J. P. Mathias, C. T. Seto,
D. Chin, M. Mammen and D. M. Gordon, Acc. Chem. Res.,
1995, 28, 37.
10 J. P. Mathias, E. E. Simanek, J. A. Zerkowski, C. T. Seto and
G. M. Whitesides, J. Am. Chem. Soc., 1994, 116, 4316.
11 A. G. Bielejewska, C. E. Marjo, L. J. Prins, P. Timmerman,
F. de Jong and D. N. Reinhoudt, J. Am. Chem. Soc., 2001,
123, 7518.
Conclusions
Mixing stoichiometric quantities of a pyrimidine with a bar-
biturate or a cyanurate in non-polar solvents leads to quantitat-
ive formation of 3 : 3 rosette complexes, which are held
together by multiple H-bonding interactions. By using rosette
components in which one of the H-bonding edges is blocked
by a substituent, it is possible to dissect the free energy contri-
butions that govern the self-assembly of H-bonded rosettes. 1H
NMR titrations were used to measure the equilibrium con-
stants for formation of 1 : 1 complexes between two blocked
components, 1 : 2 complexes between one rosette component
and a blocked component, and 3 : 3 rosette complexes. The
H-bonding interactions involved in pairwise interaction of two
components are an order of magnitude stronger for cyanurates
than for barbiturates and the rosette complexes are corre-
spondingly more stable. The 1 : 2 complexes show that there is
no allosteric cooperativity in H-bonded rosettes. The chelate
cooperativity associated with the intramolecular interactions
required to cyclise the rosette was quantified by the
effective molarity, which is approximately 2 M for all four
rosettes studied, and the product Kref EM is 102–104 indicating
that rosette self-assembly is highly cooperative with no other
species populated to a significant extent in these systems.
12 P. Timmerman and L. J. Prins, Eur. J. Org. Chem., 2001,
3191.
13 H. E. Birkett, R. K. Harris, P. Hodgkinson, K. Carr,
M. H. Charlton, J. C. Cherryman, A. M. Chippendale and
R. P. Glover, Magn. Reson. Chem., 2000, 38, 504.
14 L. J. Prins, D. N. Reinhoudt and P. Timmerman, Angew.
Chem., Int. Ed., 2001, 40, 2382.
15 S. W. Benson, J. Am. Chem. Soc., 1958, 80, 5151.
16 D. M. Bishop and K. J. Laidler, J. Chem. Phys., 1965, 42,
1688.
Conflicts of interest
There are no conflicts to declare.
17 W. F. Bailey and A. S. Monahan, J. Chem. Educ., 1978, 55,
489.
Acknowledgements
This project has received funding from the European Union’s 18 G. Ercolani, C. Piguet, M. Borkovec and J. Hamacek, J. Phys.
Horizon 2020 research and innovation programme under the Chem. B, 2007, 111, 12195.
Marie Sklodowska-Curie grant agreement no. 642793. We are 19 C. Montoro-García, J. Camacho-García, A. M. López-Pérez,
grateful to Dr Filip Szczypiński and Professor Harry Anderson
N. Bilbao, S. Romero-Pérez, M. J. Mayoral and D. González-
for valuable discussion.
Rodríguez, Angew. Chem., Int. Ed., 2015, 54, 6780.
This journal is © The Royal Society of Chemistry 2020
Org. Biomol. Chem.