Efficient self-assembly in water of long noncovalent polymers by nucleobase analogues

Article abstract of DOI:10.1021/ja312155v

Molecular self-assembly is widely appreciated to result from a delicate balance between several noncovalent interactions and solvation effects. However, current design approaches for achieving self-assembly in water with small, synthetic molecules do not consider all aspects of the hydrophobic effect, in particular the requirement of surface areas greater than 1 nm² for an appreciable free energy of hydration. With the concept of a minimum hydrophobic surface area in mind, we designed a system that achieves highly cooperative self-assembly in water. Two weakly interacting low-molecular-weight monomers (cyanuric acid and a modified triaminopyrimidine) are shown to form extremely long supramolecular polymer assemblies that retain water solubility. The complete absence of intermediate assemblies means that the observed equilibrium is between free monomers and supramolecular assemblies. These observations are in excellent agreement with literature values for the free energy of nucleic acid base interactions as well as the calculated free energy penalty for the exposure of hydrophobic structures in water. The results of our study have implications for the design of new self-assembling structures and hydrogel-forming molecules and may provide insights into the origin of the first RNA-like polymers.

Full text of DOI:10.1021/ja312155v

Communication
pubs.acs.org/JACS
Efficient Self-Assembly in Water of Long Noncovalent Polymers by
Nucleobase Analogues
Brian J. Cafferty,† Isaac Gallego,† Michael C. Chen,† Katherine I. Farley,† Ramon Eritja,‡
́
and Nicholas V. Hud*,†
†Department of Chemistry and Biochemistry, Parker H. Petit Institute for Bioengineering and Bioscience, Georgia Institute of
Technology, Atlanta, Georgia 30332-0400, United States
‡Institute for Research in Biomedicine, Parc Científic de Barcelona, Barcelona 08028, Spain
S
* Supporting Information
in order for the hydrophobic effect to facilitate molecular self-
assembly.11−14 Inspired by such results, we hypothesized that
ABSTRACT: Molecular self-assembly is widely appreci-
ated to result from a delicate balance between several
noncovalent interactions and solvation effects. However,
current design approaches for achieving self-assembly in
water with small, synthetic molecules do not consider all
aspects of the hydrophobic effect, in particular the
requirement of surface areas greater than 1 nm2 for an
appreciable free energy of hydration. With the concept of a
minimum hydrophobic surface area in mind, we designed a
system that achieves highly cooperative self-assembly in
water. Two weakly interacting low-molecular-weight
monomers (cyanuric acid and a modified triaminopyr-
imidine) are shown to form extremely long supramolecular
polymer assemblies that retain water solubility. The
complete absence of intermediate assemblies means that
the observed equilibrium is between free monomers and
supramolecular assemblies. These observations are in
excellent agreement with literature values for the free
energy of nucleic acid base interactions as well as the
calculated free energy penalty for the exposure of
hydrophobic structures in water. The results of our study
have implications for the design of new self-assembling
structures and hydrogel-forming molecules and may
provide insights into the origin of the first RNA-like
polymers.
small, water-soluble molecules that associate weakly in water
would form highly ordered supramolecular assemblies if the
intermediate structures formed by the association of monomers
were to create hydrophobic surfaces with dimensions greater
than 1 nm2. Here we show the application of this principle to
the generation of some of the longest supramolecular polymers
ever observed in water from well-studied recognition elements
that were previously limited to linear polymer formation in
aprotic solvents.
For over 20 years, the triazines melamine and cyanuric acid
(CA) and the related pyrimidines 2,4,6-triaminopyrimidine
(TAP) and barbituric acid have been used as recognition units
to construct many different three-dimensional noncovalent
assemblies, including the formation of six-membered ro-
settes.15,16 Early studies with these motifs emphasized the
importance of H-bonding in supramolecular assemblies and
were accordingly carried out in “noncompeting” aprotic
solvents. Later attempts to assemble the complementary
monomeric molecules in water resulted in the formation of
irregular precipitates17 or the total loss of molecular
recognition7 unless multiple units were joined by a hydrophilic
covalent linker.6 We were intrigued by the potential of these
historically significant recognition units because (1) the surface
area of a six-membered rosette (e.g., TAP3CA3) is 1.7 nm2 and
(2) the similarity of these monomer units to nucleic acid bases
presents the possibility of understanding the energetics of
assembled structures in terms of well-characterized molecular
interactions. Specifically, decades of research into RNA and
DNA structures have produced some of the most accurate
values available for the standard-state free energies of H-bond
formation and π−π stacking in water.
lucidating the physiochemical principles that govern
E
molecular self-assembly in water is essential for under-
standing biological systems at the molecular level, for
developing new materials, and for nanotechnology.1,2 Current
efforts to improve the self-assembly of synthetic molecules in
water include increasing the number and arrangement of H-
bonds,3 adding complementary electrostatic charges for salt-
bridge formation,4 and covalently joining multiple recognition
units with a flexible linker.5,6 Hierarchical assembly methods
involving intermediate structures with large planar hydrophobic
surfaces that stack to form noncovalent polymers have been
demonstrated using both synthetic7−9 and biologically derived
molecules.10 However, a simple description of what is needed
to harness the hydrophobic effect is still lacking, which limits its
use as a design parameter for artificial systems. Both theoretical
studies and recent experimental results indicate that hydro-
phobic surfaces must be larger than a minimum size (ca. 1 nm2)
We constructed a two-component self-assembling system
consisting of CA and TAPAS, a derivative of TAP in which one
exocyclic amine is modified with a succinate group (Figure 1A).
This modification was performed to inhibit sterically the
formation of nonrosette assemblies (e.g., ribbons and sheets),
which otherwise would result in coprecipitation of the
unmodified TAP and CA motifs, and to increase the solubility
of the desired TAPAS−CA rosette assembly. Equimolar
Received: December 12, 2012
Published: February 8, 2013
© 2013 American Chemical Society
2447
dx.doi.org/10.1021/ja312155v | J. Am. Chem. Soc. 2013, 135, 2447−2450

Journal of the American Chemical Society
Communication
Figure 1. (A) Chemical structures of succinate-conjugated 2,4,6-
triaminopyrimidine (TAPAS) and cyanuric acid (CA), a TAPAS−CA
rosette, and the proposed polymer formed by stacking of rosettes. (B)
UV spectra of solutions with only TAPAS (green curve) and 1:1
TAPAS/CA mixtures at 3, 5, and 15 mM. All of the spectra were taken
at 20 °C. CA does not absorb light within the UV range displayed. (C)
Inverted-bottle test of gel formation by solutions with the same
monomer concentrations as in (B). The increasing intensity of the
band at 320 nm in (B) is indicative of ring stacking and correlates with
the gel rigidity in (C). See the SI for more experimental details.
Figure 2. Images and measurements of TAPAS−CA supramolecular
assemblies. (A) AFM topographic image and (B) TEM micrograph of
self-assembled TAPAS−CA fibers. The insets show the profiles
delimited by the red and blue lines in the main panels. The
concentration of each monomer used for the assemblies was (A) 5 and
(B) 10 mM.
mixtures of TAPAS and CA (pH 7, 200 mM sodium
phosphate) remain in solution at concentrations above 50
mM each, whereas mixtures of unmodified TAP and CA
immediately precipitate at concentrations above 10 mM. The
UV absorption spectrum of TAPAS changes substantially in the
presence of CA, with the rise of a longer-wavelength absorption
band at concentrations above 3.5 mM [Figure 1B; also see the
Supporting Information (SI)], a spectral feature associated with
the ordered stacking of chromophores into J-type aggregates.18
Mixtures containing TAPAS and CA at 5 mM or higher form
shear-thinning hydrogels (Figure 1C) for which the rates of gel
formation and gel rigidification increase with monomer
concentration (e.g., solutions containing each monomer at 5
mM gel within 10 min, while 30 mM solutions gel in seconds).
The architecture of the TAPAS−CA assemblies was
visualized by atomic force microscopy (AFM) and transmission
electron microscopy (TEM). Both techniques revealed large
(>1 μm) linear and branched fibrillar structures with extremely
high aspect ratios (Figure 2). AFM showed that the diameter of
a single fiber perpendicular to the image plane (i.e., the height)
was between 1.5 and 2.2 nm, corresponding to the predicted
width of a rosette. Large aggregated, fibrillar structures created
higher-order network assemblies that were also observed in
some of the AFM images (Figure S2 in the SI), consistent with
the formation of a gel matrix. TEM micrographs also showed
the formation of long individual fibers that were laterally
associated into bifurcating bundles. An electron density profile
of one such bundle confirmed the width of an individual fiber
to be ca. 2 nm (Figure 2B).
and the formation of polymers longer than 100 nm required
covalent tethering of two monomers. Thus, to the best of our
knowledge, the TAPAS−CA assemblies are the longest
supramolecular polymers generated to date by untethered,
monocyclic monomers in water.
NMR spectroscopy was used to follow the association of
TAPAS and CA as a function of concentration to determine
their association constant. The 1H spectrum of equimolar
mixtures of TAPAS and CA from 1 to 3 mM exhibited TAPAS
chemical shifts identical to those in the spectrum of TAPAS in
the absence of CA, with resonance intensities directly
proportional to the TAPAS concentration (Figure 3A). At
concentrations above 3.5 mM, the integrated intensity of the
TAPAS resonances remained identical to that at 3.5 mM. The
lack of an observed change in the TAPAS chemical shifts or
signal intensity or the appearance of new TAPAS resonances
indicates that TAPAS molecules were in slow exchange
between their unbound and assembled states (i.e., exchange
time constant >100 ms). In these experiments, the TAPAS−
CA assemblies were too large to be observed by solution NMR
spectroscopy because of extreme resonance line broadening,
consistent with the structures having very large molecular
weight, such as those observed by AFM and TEM. The
formation of a TAPAS−CA assembly is therefore a highly
cooperative transition in which only free monomers and large
assemblies coexist (i.e., there are no intermediate structures).19
Additional evidence of a phase transition was also provided by
In a previous study, Fenniri and co-workers demonstrated
the formation of micrometer-length polymer bundles from
monomers with similar H-bonding edges.8 However, the
stacking surface of their monomers was larger (i.e., bicyclic),
2448
dx.doi.org/10.1021/ja312155v | J. Am. Chem. Soc. 2013, 135, 2447−2450

Journal of the American Chemical Society
Communication
between nucleobases within folded RNA structures. Specifically,
each H-bond contributes ca. −0.5 kcal·mol−1 to the stability of
an RNA secondary structure,23 and the stacking of two six-
membered rings (derived from the overlap of purine bases on
opposite strands of a duplex) contributes −1.8 kcal·mol−1 to
duplex stability.24,25 On the basis of these values, the free
energy for the three H-bonds and π−π stacking of each
monomer in stacked TAPAS−CA rosettes is predicted to be
−3.3 kcal·mol−1.
At pH 7, free CA (pKa = 6.9) and the pyrimidine ring of free
TAPAS (pKa = 6.5; see the SI) both exist in their neutral and
ionized states in approximately in equal proportions. The
assembly of a supramolecular structure by TAPAS and CA
would be expected to maintain a net zero charge within the
stacked rosettes, as otherwise the electrostatic repulsion in the
low-dielectric interior of this structure would prevent its
formation. In a solution with pH above the pKa of CA (i.e.,
>6.9), three protons per rosette must be abstracted from the
solvent upon formation of the supramolecular assembly. As a
test of this prediction, the concentration dependence of the
1
TAPAS−CA assembly was investigated by H NMR spectros-
copy at pH 8 (Figure 3B). In this case, the plateau in the
integrated resonance intensity was observed at ca. 10 mM
TAPAS and CA. This 3-fold increase in the concentration
required for TAPAS−CA assembly is in excellent agreement
with the free energy difference for abstraction of a proton at pH
7 versus pH 8. We estimate the free energy of monomer
incorporation into the polymeric assemblies at pH 8 to be −2.7
kcal·mol−1 on the basis of the free TAPAS and CA
concentrations of 10 mM. As three protons per rosette must
be captured from solution at a free energy that is 1.34
kcal·mol−1 greater at pH 8 than at pH 7, this average change in
free energy of 0.67 kcal·mol−1 per monomer is in excellent
agreement with the apparent change in the TAPAS/CA
association free energy from −3.3 kcal·mol−1 at pH 7 to ca.
−2.7 kcal·mol−1 at pH 8.
Figure 3. Plots of the apparent solution-phase concentration of
TAPAS (as determined by methylene 1H NMR resonance integration)
vs the actual TAPAS concentration in 1:1 solutions with CA at (A) pH
7 and (B) pH 8. Horizontal lines indicate the concentration observed
for supramolecular assembly (ca. 3.5 mM at pH 7 and ca. 10 mM at
pH 8). The insets show representative spectra of the TAPAS
methylene protons. Line broadening in the more concentrated samples
likely occurred because the solvent became viscous with gel formation.
The peculiar methylene resonance line structure observed for the pH 7
spectra at higher concentrations appears to be the result of residual
dipolar coupling, which can be considerable in samples containing
linear supramolecular assemblies that are partially aligned by the
magnetic field.20
dynamic light scattering (DLS), which revealed a dramatic
increase in scattering intensity when the concentration of each
monomer was above 3.5 mM (Figure S3A). Cooling to 5 °C
decreased the minimal assembly concentration 5-fold, to below
1 mM (Figure S4).
While the assembly of the monomers is an enthalpically
driven process, as indicated by thermal denaturation experi-
ments (Figure S4), the very long assemblies observed by AFM
and TEM as well as the absence of evidence for intermediate
structures (e.g., individual rosettes, rosette dimers, trimers, etc.)
Our NMR results show similarities and differences in
comparison with those for previously reported self-assembling
systems. For example, previous studies of small-molecule
systems that self-assemble in water through the stacking of
rosettes7 or G-tetrads (see below)21 were observed directly by
solution NMR spectroscopy, indicating the formation of
supramolecular polymers much smaller than those reported
here. In a different study, Whitesides and co-workers also
1
in our H NMR spectra can be explained by the hydrophobic
effect. Theoretical studies estimate the free energy for cavity
formation in water around a hydrophobic molecule or assembly
to be ca. 8 kcal·nm−2·mol−1 (calculated for a sphere of radius
0.8 nm).13 On the basis of this value, we estimate the free
energy cost of exposing the two hydrophobic faces of a single
rosette (each face with surface area of 1.7 nm2) to be on the
order of 27 kcal·mol−1. The same positive free energy would
apply to the two solvent-exposed ends of a rosette stack and
thus may be the origin of the highly cooperative nature of
TAPAS−CA self-assembly. That is, the free energy associated
with doubling the average stack length (i.e., combining pairs of
existing stacks to eliminate half of the total number of exposed
rosette faces) is on the order of −27 kcal·mol−1. In contrast, the
free energy of mixing associated with decreasing the
concentration of the stacks by a factor of 2 is only +0.3
kcal·mol−1. In view of the large net free energy for stack
consolidation, the hydrophobic effect is expected to drive the
assembly of rosettes into polymers with lengths that would be
limited only by cyclization and the steady-state breakage rates
that result from kinetic fluctuations, a prediction that is fully
1
observed a plateau in the H NMR signal intensity associated
with the self-assembly of bis-modified melamine and bis-
modified CA monomers in an organic solvent, but in their
system, signal loss occurred over several hours, indicating much
slower growth of large structures compared with that reported
here for monomeric recognition units.22
The free energy associated with TAPAS and CA monomer
incorporation into supramolecular assemblies can be estimated
from the free energy of dilution corresponding to the free
TAPAS and CA concentrations that coexist with these
assemblies. At 20 °C and pH 7, the 3.5 mM free monomer
concentration corresponds to a standard-state free energy of
dilution of −3.3 kcal·mol−1. This value is in excellent agreement
with free energy values previously determined for interactions
2449
dx.doi.org/10.1021/ja312155v | J. Am. Chem. Soc. 2013, 135, 2447−2450

Journal of the American Chemical Society
Communication
consistent with the observed phase transition between free
monomers and large supramolecular assemblies.
AFM, M. Zhou for mass spectrometry, and the CNC EM
center at Georgia Tech. This research was supported by NASA
Exobiology [NNX08A014G] (B.J.C., N.V.H.) and jointly
supported by the NSF and the NASA Astrobiology Program
under the NSF Center for Chemical Evolution [CHE-1004570]
(I.G., M.C.C., K.I.F., N.V.H.), the Parker H. Petit Endowment
The results presented here illustrate how relatively simple
small molecules can efficiently form supramolecular polymers
in water if they assemble into intermediate structures with
hydrophobic surfaces having areas greater than ca. 1 nm2. The
H-bonding and π−π stacking of TAPAS and CA monomers
within our system is similar to the association of nucleobases
within RNA and DNA duplexes. However, these associations
clearly differ in that the canonical nucleobases (i.e., G, A, C, U,
and T) and their corresponding free mononucleosides do not
form H-bonded Watson−Crick base pairs in water.26 A notable
exception is the G-tetrad formed by guanosine and its
derivatives with the aid of cation coordination, which does
form polymeric structures in water and can result in
gelation.10,27 Our explanation for TAPAS−CA assembly also
applies to G-tetrad formation, as the area of a G-tetrad is on the
order of 1 nm2. However, in the case of G-tetrads, two stacked
tetrads are observed as an intermediate assembly,10 and
polymer formation is less cooperative than in the TAPAS−
CA system. This tolerance of intermediate assemblies could be
due to the smaller stacking area of the G-tetrad and electrostatic
repulsions due to coordinated cations in the center of the G-
tetrads and phosphate groups on each monomer, as
incremental changes in monomer charge can greatly affect
the degree of self-assembly in water.10,28
The TAPAS−CA assembly system presented here demon-
strates that the hexameric rosette can be used as a functional
architecture to generate hydrogels, which may be favorable for
soft-material design and applications because of their chemical
simplicity. For example, rosette nanotubes have recently been
used in the formulation of hydrogel materials that enhance
tissue growth.29
Finally, the inability of the RNA and DNA nucleobases to
base-pair and assemble further in water has led to the proposal
that the canonical nucleobases are products of evolution.30 The
TAPAS−CA system presented here forms polymer assemblies
greater than 1 μm in length even at concentrations below 10
mM. Such assemblies contain over 18 000 highly organized
monomers. This observation, along with the structural
similarity of TAPAS−CA pairing with Watson−Crick base-
pairing, suggests a possible prebiotic mechanism for proto-
nucleobase selection from a complex mixture and organization
into gene-length polymers even before linkage by a common
backbone.
́
(B.J.C.), and Consejo Superior de Investigaciones Cientificas
(CSIC) [MEC, SAB2010-0163] (R.E., N.V.H.).
REFERENCES
■
(1) Aida, T.; Meijer, E. W.; Stupp, S. I. Science 2012, 335, 813.
(2) De Greef, T. F. A.; Smulders, M. M. J.; Wolffs, M.; Schenning, A.
P. H. J.; Sijbesma, R. P.; Meijer, E. W. Chem. Rev. 2009, 109, 5687.
(3) Prins, L. J.; Reinhoudt, D. N.; Timmerman, P. Angew. Chem., Int.
Ed. 2001, 40, 2382.
(4) Rehm, T. H.; Schmuck, C. Chem. Soc. Rev. 2010, 39, 3597.
(5) Hirschberg, J. H. K. K.; Brunsveld, L.; Ramzi, A.; Vekemans, J. A.
J. M.; Sijbesma, R. P.; Meijer, E. W. Nature 2000, 407, 167.
(6) Ma, M. M.; Bong, D. Langmuir 2011, 27, 8841.
(7) Fenniri, H.; Packiarajan, M.; Vidale, K. L.; Sherman, D. M.;
Hallenga, K.; Wood, K. V.; Stowell, J. G. J. Am. Chem. Soc. 2001, 123,
3854.
(8) Moralez, J. G.; Raez, J.; Yamazaki, T.; Motkur, K.; Kovalenko, A.;
Fenniri, H. J. Am. Chem. Soc. 2005, 127, 8307.
(9) Bozsonyi, G.; Beingessner, R. L.; Yamazaki, T.; Cho, J.; Myles, A.
J.; Malac, M.; Egerton, R.; Kawasaki, M.; Ishizuka, K.; Kovalenko, A.;
Fenniri, H. J. Am. Chem. Soc. 2010, 132, 15135.
(10) Davis, J. T. Angew. Chem., Int. Ed. 2004, 43, 668.
(11) Davis, J. G.; Gierszal, K. P.; Wang, P.; Ben-Amotz, D. Nature
2012, 491, 582.
(12) Li, I. T. S.; Walker, G. C. Proc. Natl. Acad. Sci. U.S.A. 2011, 108,
16527.
(13) Chandler, D. Nature 2005, 437, 640.
(14) Raschke, T. M.; Tsai, J.; Levitt, M. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 5965.
(15) Whitesides, G. M.; Simanek, E. E.; Mathias, J. P.; Seto, C. T.;
Chin, D. N.; Mammen, M.; Gordon, D. M. Acc. Chem. Res. 1995, 28,
37.
(16) Lehn, J.-M.; Mascal, M.; Decian, A.; Fischer, J. J. Chem. Soc.,
Perkin Trans. 2 1992, 461.
(17) Bohanon, T. M.; Denzinger, S.; Fink, R.; Paulus, W.; Ringsdorf,
H.; Weck, M. Angew. Chem., Int. Ed. Engl. 1995, 34, 58.
(18) Jelley, E. E. Nature 1936, 138, 1009.
(19) Hirst, A. R.; Coates, I. A.; Boucheteau, T. R.; Miravet, J. F.;
Escuder, B.; Castelletto, V.; Hamley, I. W.; Smith, D. K. J. Am. Chem.
Soc. 2008, 130, 9113.
(20) Prestegard, J. H.; Al-Hashimi, H. M.; Tolman, J. R. Q. Rev.
Biophys. 2000, 33, 371.
(21) Wong, A.; Ida, R.; Spindler, L.; Wu, G. J. Am. Chem. Soc. 2005,
127, 6990.
(22) Choi, I. S.; Li, X. H.; Simanek, E. E.; Akaba, R.; Whitesides, G.
M. Chem. Mater. 1999, 11, 684.
ASSOCIATED CONTENT
* Supporting Information
■
S
(23) Klostermeier, D.; Millar, D. P. Biochemistry 2002, 41, 14095.
(24) Mohan, S.; Hsiao, C.; Bowman, J. C.; Wartell, R.; Williams, L.
D. J. Am. Chem. Soc. 2010, 132, 12679.
(25) Frieir, S. M.; Alkema, D.; Sinclair, A.; Neilson, T.; Turner, D. H.
Biochemistry 1985, 24, 4533.
(26) Ts’o, P.; Melvin, I.; Olson, A. J. Am. Chem. Soc. 1963, 85, 1289.
(27) Yu, Y.; Nakamura, D.; DeBoyace, K.; Neisius, A. W.; McGown,
L. B. J. Phys. Chem. B 2008, 112, 1130.
(28) Besenius, P.; Portale, G.; Bomans, P. H. H.; Janssen, H. M.;
Palmans, A. R. A.; Meijer, E. W. Proc. Natl. Acad. Sci. U.S.A. 2010, 107,
17888.
Detailed experimental procedures and additional figures. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
hud@chemistry.gatech.edu
■
Notes
The authors declare no competing financial interest.
(29) Zhang, L.; Rakotondradany, F.; Myles, A. J.; Fenniri, H.;
Webster, T. J. Biomaterials 2009, 30, 1309.
(30) Engelhart, A. E.; Hud, N. V. Cold Spring Harbor Perspect. Biol.
2010, DOI: 10.1101/cshperspect.a002196.
ACKNOWLEDGMENTS
■
We thank A. E. Engelhart, F. A. L. Anet, G. B. Schuster, L. A.
Lyon, L. D. Williams, and R. Krishnamurthy for discussions; S.
Grijalvo and M. Terrazas for synthesis advice, L. Bottemely for
2450
dx.doi.org/10.1021/ja312155v | J. Am. Chem. Soc. 2013, 135, 2447−2450