Protein-mimetic, molecularly imprinted nanoparticles for selective binding of bile salt derivatives in water

Article abstract of DOI:10.1021/ja406089c

A tripropargylammonium surfactant with a methacrylate-terminated hydrophobic tail was combined with a bile salt derivative, divinyl benzene (DVB), and a photo-cross-linker above its critical micelle concentration (CMC). Surface-cross-linking with a diazide, surface-functionalization with an azido sugar derivative, and free-radical-core-cross-linking under UV irradiation yielded molecularly imprinted nanoparticles (MINPs) with template-specific binding pockets. The MINPs resemble protein receptors in size, complete water-solubility, and tailored binding sites in their hydrophobic cores. Strong and selective binding of bile salt derivatives was obtained, depending on the cross-linking density of the system.

Full text of DOI:10.1021/ja406089c

Communication
pubs.acs.org/JACS
Protein-Mimetic, Molecularly Imprinted Nanoparticles for Selective
Binding of Bile Salt Derivatives in Water
Joseph K. Awino and Yan Zhao*
Department of Chemistry, Iowa State University, Ames, Iowa 50011-3111, United States
S
* Supporting Information
The key design in our MINPs is the doubly cross-linkable
ABSTRACT: A tripropargylammonium surfactant with a
methacrylate-terminated hydrophobic tail was combined
with a bile salt derivative, divinyl benzene (DVB), and a
photo-cross-linker above its critical micelle concentration
(CMC). Surface-cross-linking with a diazide, surface-
functionalization with an azido sugar derivative, and free-
radical-core-cross-linking under UV irradiation yielded
molecularly imprinted nanoparticles (MINPs) with
template-specific binding pockets. The MINPs resemble
protein receptors in size, complete water-solubility, and
tailored binding sites in their hydrophobic cores. Strong
and selective binding of bile salt derivatives was obtained,
depending on the cross-linking density of the system.
surfactant 1. The tripropargylated surfactant forms micelles in
water with a high density of alkyne on the surface. Covalent
fixation by a diazide cross-linker using Cu(I) catalysts yields
surface-cross-linked micelles (SCMs) that could be easily
functionalized through another round of click reaction.18−21
Unlike previously synthesized tripropargylated surfactants,
however, 1 has a polymerizable methacrylate and thus can
undergo free-radical polymerization orthogonal to the surface-
cross-linking by the click reaction.
Surfactant 1 has three alkyne groups, and cross-linker 3 has
two azides. In the first stage of the reaction, we performed the
surface-cross-linking of the micelles using [3]:[1] = 1.2/1,
allowing good cross-linking while leaving sufficient alkyne groups
on the SCM surface for further functionalization (Scheme
1).18−21 The cross-linked micelles were prepared with 10 mM of
1 in water, above its CMC of 0.55 mM (Figure 1S in Supporting
Information). Our DLS study showed that each SCM contained
ca. 50 surfactants (Figure 2S). Thus, a ratio of [1]:[2] = 1:0.02 in
theory placed one template (i.e., bile salt derivative 2) within
each SCM. The cross-linking chemistry and covalent structure of
olecularly imprinted polymers (MIPs) have binding sites
Mpotentially complementary to guests in size, shape, and
distribution of functional groups.1−3 They are usually prepared
by copolymerization of functional monomers and cross-linkers in
the presence of a template. Tremendous progress has been made
in the last decades in this technology, with imprinted materials
generated for small and large guests,1−3 in macroporous
polymers and on surface4 for molecular recognition and
catalysis,5,6 and even unimolecularly within dendrimers.7,8
A difficult challenge in molecular imprinting remains the
creation of protein-like, water-soluble nanoparticles with high
binding affinity and selectivity for guests.9−11 Part of the
challenge comes from the general difficulty in constructing
synthetic receptors that function in water.12 Hydrogen bonds as
directional intermolecular forces are the most popular tools used
by chemists for molecular recognition but tend to become
ineffective in aqueous solution due to competition from the
solvent. The hydrophobicity of typical MIPs represents another
hurdle, as hydrophobic materials often bind nonpolar molecules
nonspecifically.
Scheme 1. Preparation of MINP
In this communication, we report a method to prepare
molecularly imprinted nanoparticles (MINPs) for selective
binding of bile salt derivatives in water. Although imprinted
polymeric nanoparticles have been reported in the litera-
ture,13−15 our MINP is characterized by its discrete binding
sites and great resemblance to protein receptors in its
nanodimension, complete water-solubility, functionalizable
exterior, and easily accessible, tailor-made hydrophobic binding
pocket. We chose bile salts as the template/guest molecules
because of their important biological properties and water-
solubility.16,17
Received: June 19, 2013
Published: August 9, 2013
© 2013 American Chemical Society
12552
dx.doi.org/10.1021/ja406089c | J. Am. Chem. Soc. 2013, 135, 12552−12555

Journal of the American Chemical Society
Communication
the SCMs have been previously characterized by mass
spectrometry and TEM.18
After confirming effective binding of the template molecule,
we studied the binding of several other bile salt derivatives (5−9)
to understand the selectivity of the MINPs. After all, molecular
imprinting is meant to create binding sites complementary to the
template. Among these compounds, 5 and 6 have the acyl group
on the amine gradually decrease in size and were designed to test
the size/shape selectivity of the MINPs.
After surface-cross-linking, sugar-derived ligand 4 was added
to the reaction mixture. The surface functionalization, catalyzed
also by Cu(I), made the final MINPs completely hydrophilic and
easy to purify (vide infra). After surface functionalization, the
sample was immediately subjected to UV irradiation to
copolymerize the methacrylate of 1 and DVB solubilized within
the SCMs. The photopolymerization was initiated by DMPA
(i.e., 2,2-dimethoxy-2-phenylacetophenone, a photoinitiator)
added together with DVB at the beginning of the reaction.
1
After 12 h of irradiation, H NMR spectroscopy indicated
complete disappearance of alkenic protons (Figure 3S). DLS
showed a narrow distribution of nanoparticles ca. 4.2 nm in
diameter for alkynyl-SCM, 5.9 nm after surface functionalization,
and 5.0 nm after core-cross-linking (Figure 4S).
Preparation of the MINPs was remarkably simple. The entire
synthesis was a one-pot reaction over 2 d at room temperature in
water. Equally important was the extremely easy purification.
The nanoparticles could be precipitated from acetone after the
core-cross-linking, due to the sugar-derived surface ligand 4.
Repeated washing by methanol/acetic acid and methanol
completely removed the template (as shown by fluorescence
spectroscopy) and afforded the final MINPs in ca. 80% yield. The
materials obtained were fully soluble in water and showed no
change in size compared to the as synthesized MINPs.
An important strategy in the MINP synthesis was the
combination of a cationic cross-linkable surfactant and an
anionic template. Their electrostatic interactions make it easy to
not only incorporate the template inside the micelle and
ultimately inside MINP, but also orient the hydrophobic part of
template within the hydrophobic core of the micelle and the
carboxylate on the surface. The result is easy removal of the
template, which vacates the binding site, and facile rebinding of
the guest.
Table 1 summarizes the binding data of MINPs generated
from 2 and 6 as the templates. For the majority of the bindings,
we employed isothermal titration calorimetry (ITC) because
most of the guests were not fluorescent. An important benefit of
ITC is the simultaneous determination of the number of binding
sites (N) on the MINP. Figure 2 shows three typical ITC titration
curves between MINP(2) and bile salt derivatives. Note that, for
the fluorescent guest (2), the binding constants obtained by the
fluorescence titration and ITC showed excellent agreement
(Table 1, compare entries 1 & 2 and 10 & 11).
Our first batch of MINPs was prepared with 0.5 equiv of DVB.
ITC shows that MINP(2) synthesized under this condition
bound its template with Ka = 3.5 × 106 M−1 (Table 1, entry 2). As
the acyl group decreased in size, the binding deteriorated, with Ka
going down to 2.5 × 106 for the benzoyl derivative (5) and to
0.05 × 106 when the benzoyl was replaced by acetyl. In other
words, the template itself fitted the binding pocket better than
the other two (smaller) analogues, demonstrating the effective-
ness of the imprinting. Among the naturally occurring bile salts, 7
showed similar binding as 6 but none of the more hydrophilic
compounds (8 and 9) gave any detectable binding.
The dansyl group of 2 allowed us to study its binding by
fluorescence spectroscopy. As shown in Figure 1a, upon the
We then synthesized MINPs using the smaller bile salt
derivative 6 as the template. The result was somewhat
disappointing. On the one hand, the largest bile salt (2) showed
weaker binding to MINP(6) than to MINP(2), as expected from
the smaller size of the binding pocket in the former (Table 1,
compare entries 2 and 6). On the other hand, although 2 bound
to MINP(6) less strongly than 5, both compounds bound much
more strongly than 6 itself (Ka = 0.09 × 106 M−1). It appears that
both hydrophobic effects and size/shape selectivity were playing
roles in this MINP. Essentially, although the binding pocket of
MINP(6) was smaller than that of MINP(2), the stronger
hydrophobicity of 2 and 5 gave them a larger driving force to
occupy the hydrophobic pocket than the somewhat hydrophilic
6. Whereas 6 might fit the binding site of MINP(6) better than
the larger bile salts, its weaker hydrophobicity lowers its tendency
to enter the pocket.
Not satisfied with the above results, we decided to increase the
amount of DVB to 1 equiv to 1 for the core-cross-linking. This
was the highest amount of DVB that could be solubilized by the
surfactant in water. To our delight, binding selectivity increased
dramatically. Using this more highly cross-linked MINP(2), we
were able to distinguish the size of the acyl group easily: the
dansyl, benzoyl, and acetyl derivatives afforded Ka of 3.5, 0.46,
Figure 1. (a) Emission spectra of 2 upon the addition of different
concentrations of MINP(2). [2] = 0.050 μM. The concentration of
MINP was calculated based on a MW of 49800 g/mol determined by
DLS (see Figure 5S for details). (b) Nonlinear least-squares fitting of the
emission intensity of 3 at 475 nm to a 1:1 binding isotherm.
addition of MINP(2), i.e., MINP imprinted against 2, to an
aqueous solution of 2, the dansyl emission at 550 nm
immediately shifted to 475 nm. A large blue shift and enhanced
emission suggest a more hydrophobic environment around
dansyl22 and are frequently observed when dansyl-labeled
compounds are internalized by micelles.23,24 The fluorescence
intensity at 475 nm fitted nicely to a 1:1 binding isotherm to give
an association constant (Ka) of 3.3 × 106 M−1 (Figure 1b). The
binding affinity was among the highest observed between
synthetic hosts and steroid derivatives including bile salt
derivatives.17
12553
dx.doi.org/10.1021/ja406089c | J. Am. Chem. Soc. 2013, 135, 12552−12555

Journal of the American Chemical Society
Communication
a
Table 1. Binding Data for MINPs (Obtained by ITC unless Indicated Otherwise)
entry
MINP
DVB
guest
Ka
(× 106 M−1
−ΔG
N
(equiv)
)
(kcal/mol)
b
b
1
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(6)
MINP(6)
MINP(6)
MINP(6)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(2)
MINP(6)
MINP(6)
MINP(6)
MINP(6)
MINP(2)2
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
0.5
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
2
2
3.3 0.5
3.5 0.2
2.5 0.1
8.9
8.9
8.7
6.4
6.0
8.4
8.6
6.8
5.9
9.0
8.9
7.7
7.4
-
2
1.0
0.9
0.7
3
5
4
6
0.05 0.01
0.03 0.01
1.39 0.02
1.99 0.03
0.09 0.01
0.02 0.01
c
5
7
0.3
6
2
0.5
0.7
0.8
7
5
8
6
c
9
7
0.2
b
b
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
2
3.7 0.8
-
2
3.5 0.2
0.46 0.07
0.28 0.04
1.0
0.7
0.6
5
6
d
d
d
7
-
-
-
c
10
11
12
0.07 0.01
0.0045 0.0002
0.28 0.03
3.21 0.02
3.28 0.02
0.27 0.12
0.58 0.02
1.1 0.2
6.6
5.0
7.4
8.9
8.9
7.4
7.9
8.2
0.3
c
0.2
0.6
1.0
1.0
0.7
0.6
0.8
e
2
f
2
2
5
6
7
2
d
d
d
-
-
-
g
3.56 0.01
3.07 0.04
8.9
8.8
1.2
1.2
a
The titrations were generally performed in duplicates and the errors between the runs were <10%. Binding was measured in 50 mM Tris buffer (pH
b
= 7.4) with 150 mM NaCl unless otherwise noted. Compounds 8 and 9 showed no detectable binding by ITC with any of the MINPs. Binding data
c
d
were obtained from fluorescence titration using a 1:1 binding model. The weak binding made the curve fitting not as accurate. Binding (Ka <1000
e
f
M−1) was not detectable by ITC. Binding was measured in 50 mM Tris buffer (pH = 7.4) with 200 mM NaCl. Binding was measured in 50 mM
g
Tris buffer (pH = 7.4) without NaCl. The MINPs were prepared with [1]:[2] = 1:0.04, i.e., twice as much of the template was employed as
compared to other examples. The two binding constants were for the two independent binding sites, respectively.
For the more highly cross-linked MINP(2), we also studied
several non-steroid aromatic guests (10−12) to test the binding
selectivity. Dansyl sulfonate 10 in a sense was a “half-match” for
the binding site generated from dansylated 2. Its Ka (= 0.07 × 106
M−1) was reduced by ca. 50 times from that of 2. A further
decrease of the hydrophobic size made naphthalene-1-
carboxylate (11) an even poorer guest, whose Ka was only
0.0045 × 106 M−1 or about 800 times weaker than that of 2. As
soon as the guest size increased, binding resumed, as
pyrenebutyrate 12 displayed identical binding constant to that
Figure 2. ITC titration curves obtained at 298 K for the binding of 2 (a),
5 (b), and 6 (c) by MINP(2) prepared with 1 equiv of DVB. The data
correspond to entries 11−13 in Table 1. Additional ITC titration curves
can be found in the Supporting Information (Figures 6S−10S). In
general, an aqueous solution of an appropriate bile salt in Tris buffer (50
mM Tris, 150 mM NaCl, pH = 7.4) was injected in equal steps into
1.428 mL of the corresponding MINP solution (4.0 mg/mL) in the
same buffer. The top panel shows the raw calorimetric data. The area
under each peak represents the amount of heat generated at each
ejection and is plotted against the molar ratio of the bile salt to the
MINP. The smooth solid line is the best fit of the experimental data to
the sequential binding of N equal and independent binding sites on the
MINP. The heat of dilution for the bile salt, obtained by adding the bile
salt to the buffer, was subtracted from the heat released during the
binding. Binding parameters were autogenerated after curve fitting using
Microcal Origin 7.
of 6 (Ka = 0.28 × 106 M−1) or 1/13 of that of 2.
All the bindings were measured in 50 mM Tris buffer with 150
mM NaCl. When the salt concentration was raised to 300 mM,
the MINPs were found to precipitate out of the buffer. In 200 and
0 mM NaCl (Table 1, entries 18 and 19), similar binding
constants were obtained for the MINP(2)−2 complex and were
essentially within the experimental error from that in 150 mM
NaCl (entry 11). We attributed the insensitivity of binding to salt
(at least over 0−200 mM NaCl) to the two opposing binding
forces present in the system: whereas salt tends to strengthen
hydrophobic interactions, it weakens the electrostatic inter-
actions between the positively charged MINP and the negatively
charged guest.
Importantly, when 1 equiv of DVB was used in the core-cross-
linking, selective binding pockets could be created for the smaller
bile salt 6 as well. As the acyl group became smaller, the bile salts
exhibited a consistent increase in their binding affinity toward the
and 0.28 × 106 M−1, respectively (Table 1, entries 11−13).
Lithocholate 7 displayed no binding at all.
12554
dx.doi.org/10.1021/ja406089c | J. Am. Chem. Soc. 2013, 135, 12552−12555

Journal of the American Chemical Society
Communication
highly cross-linked MINP(6), with Ka increasing from 0.27 × 106
to 0.58 × 106 and further to 1.1 × 106 M−1 (entries 20−22). This
trend was opposite to the hydrophobicity of the guest and
different from what was observed for MINP(6) prepared with 0.5
equiv of DVB. Clearly, the higher cross-linking density of the
material significantly enhanced the rigidity of the binding
pockets. Under this condition, even though the more hydro-
phobic guests (2 and 5) possess stronger thermodynamic
“desires” to enter the hydrophobic pocket, they were excluded
most likely because of their misfit to the less “forgiving” binding
sites.
It should be pointed out that, unlike conventional MIPS and
the reported molecularly imprinted nanoparticles,13−15 our
MINPs on average possessed approximately one guest-binding
site per particle. Except when weak binding made the curve fitting
less accurate, the number of independent binding sites (N)
obtained by ITC was close to 1 in most cases for the MINPs
(Table 1). This feature comes directly from the stoichiometry of
template used in the synthesis relative to the micelle aggregation
number of 1.
ASSOCIATED CONTENT
* Supporting Information
Experimental details for the syntheses, DLS data, ITC titration
curves, and NMR spectra for key compounds. This material is
available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
zhaoy@iastate.edu
Notes
■
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We thank NSF (CHE-1303764) for partial support of the
research.
■
REFERENCES
(1) Wulff, G. Angew. Chem., Int. Ed. Engl. 1995, 34, 1812.
■
(2) Mosbach, K.; Ramstrom, O. Nat. Biotechnol. 1996, 14, 163.
̈
(3) Haupt, K.; Mosbach, K. Chem. Rev. 2000, 100, 2495.
(4) Sellergren, B. Molecularly Imprinted Polymers: Man-Made Mimics of
Antibodies and Their Applications in Analytical Chemistry; Elsevier:
Amsterdam, 2001.
To demonstrate the tunability in binding stoichiometry, we
prepared MINP(2)2 with [1]:[2] = 1:0.04, i.e., doubling the
amount of template employed during the imprinting. Figure 3
(5) Wulff, G. Chem. Rev. 2001, 102, 1.
(6) Komiyama, M. Molecular Imprinting: From Fundamentals to
Applications; Wiley-VCH: Weinheim, 2003.
(7) Zimmerman, S. C.; Wendland, M. S.; Rakow, N. A.; Zharov, I.;
Suslick, K. S. Nature 2002, 418, 399.
(8) Zimmerman, S. C.; Zharov, I.; Wendland, M. S.; Rakow, N. A.;
Suslick, K. S. J. Am. Chem. Soc. 2003, 125, 13504.
(9) Dirion, B.; Cobb, Z.; Schillinger, E.; Andersson, L. I.; Sellergren, B.
J. Am. Chem. Soc. 2003, 125, 15101.
(10) Cutivet, A.; Schembri, C.; Kovensky, J.; Haupt, K. J. Am. Chem.
Soc. 2009, 131, 14699.
(11) Ma, Y.; Pan, G. Q.; Zhang, Y.; Guo, X. Z.; Zhang, H. Q. Angew.
Chem., Int. Ed. 2013, 52, 1511.
(12) Oshovsky, G. V.; Reinhoudt, D. N.; Verboom, W. Angew. Chem.,
Int. Ed. 2007, 46, 2366.
(13) Hoshino, Y.; Kodama, T.; Okahata, Y.; Shea, K. J. J. Am. Chem. Soc.
2008, 130, 15242.
(14) Yang, K. G.; Berg, M. M.; Zhao, C. S.; Ye, L. Macromolecules 2009,
42, 8739.
Figure 3. ITC titration curves obtained at 298 K for the binding of 2 by
MINP(2)2 prepared with 1 equiv of DVB. The data correspond to entry
24 in Table 1. The smooth solid line is the best fit of the experimental
data to the sequential binding of 2 equal and independent binding sites
on the MINP.
(15) Zeng, Z. Y.; Patel, J.; Lee, S. H.; McCallum, M.; Tyagi, A.; Yan, M.
D.; Shea, K. J. J. Am. Chem. Soc. 2012, 134, 2681.
(16) Danielsson, H.; Sjovall, J. Sterols and Bile Acids; Elsevier:
Amsterdam, 1985.
(17) Wallimann, P.; Marti, T.; Furer, A.; Diederich, F. Chem. Rev. 1997,
97, 1567.
(18) Zhang, S.; Zhao, Y. Macromolecules 2010, 43, 4020.
(19) Zhang, S.; Zhao, Y. J. Am. Chem. Soc. 2010, 132, 10642.
(20) Li, X.; Zhao, Y. Bioconjugate Chem. 2012, 23, 1721.
(21) Peng, H.-Q.; Chen, Y.-Z.; Zhao, Y.; Yang, Q.-Z.; Wu, L.-Z.; Tung,
C.-H.; Zhang, L.-P.; Tong, Q.-X. Angew. Chem., Int. Ed. 2012, 51, 2088.
(22) Li, Y. H.; Chan, L. M.; Tyer, L.; Moody, R. T.; Himel, C. M.;
Hercules, D. M. J. Am. Chem. Soc. 1975, 97, 3118.
(23) Zhong, Z.; Zhao, Y. J. Org. Chem. 2008, 73, 5498.
(24) Zhao, Y. J. Org. Chem. 2009, 74, 834.
shows the ITC titration curve for the rebinding of 2. The
distinctively different curve as compared to those in Figure 2
fitted best to a binding model with two independent binding sites
per nanoparticle. As shown by entry 24 in Table 1, the two
binding sites had very similar binding constants (Ka = 3.56 and
3.07 × 106 M−1), which were essentially the same as that of the
MINP(2)−2 complex (Ka = 3.5 × 106 M−1). Thus, the same
hydrophobic and electrostatic interactions were behind all these
binding events, whether the MINP contained one or two binding
sites.
To summarize, we have developed a facile method to create
protein-like, water-soluble receptors for selective binding of bile
salt derivatives in water. Unlike proteins, however, these
nanoparticles are extremely robust and have outstanding
tolerance for organic solvents and adverse temperature/pH
conditions. The robustness comes directly from their highly
cross-linked nature and was demonstrated by our washing
conditions in the purification. Their nanodimension, readily
modified structure,18−21,25−27 and excellent properties of
molecular recognition should make them highly useful materials
for chemistry and biology.
̈
(25) Li, X.; Zhao, Y. Langmuir 2012, 28, 4152.
(26) Zhang, S.; Zhao, Y. Chem. Commun. 2012, 48, 9998.
(27) Chadha, G.; Zhao, Y. J. Colloid Interface Sci. 2013, 390, 151.
12555
dx.doi.org/10.1021/ja406089c | J. Am. Chem. Soc. 2013, 135, 12552−12555