Glucose-specific poly(allylamine) hydrogels-A reassessment

Article abstract of DOI:10.1016/j.bmcl.2006.09.054

Polymer hydrogels synthesized by crosslinking poly(allylamine hydrochloride) with (±)-epichlorohydrin in the presence of d-glucose-6-phosphate monobarium salt do not show imprinting on the molecular level. A series of hydrogels was prepared using the following five templates: d-glucose-6-phosphate monobarium salt, d-glucose, l-glucose, barium hydrogen phosphate (BaHPO₄), and d-gluconamide; a hydrogel was also prepared in the absence of a template. For all six hydrogels, batch binding studies were conducted with d-glucose, l-glucose, d-fructose, and d-gluconamide. The extent of analyte sugar binding was determined using ¹H NMR. Each hydrogel shows approximately the same relative binding affinity for the different sugar derivatives, and none displays selectivity for either glucose enantiomer. The results of the binding studies correlate with the octanol-water partition coefficients of the sugars, indicative that differential solubilities in the bulk polymer account for the binding affinities observed. Thus, in contrast to templated hydrogels prepared using methacrylate- or acrylamide-based reagents, true imprinting does not occur in this novel, crosslinked-poly(allylamine hydrochloride) system.

Full text of DOI:10.1016/j.bmcl.2006.09.054

Bioorganic & Medicinal Chemistry Letters 17 (2007) 235–238
Glucose-specific poly(allylamine) hydrogels—A reassessment
Furqan M. Fazal and David E. Hansen*
Department of Chemistry, Amherst College, Amherst, MA 01002, USA
Received 15 August 2006; revised 18 September 2006; accepted 19 September 2006
Available online 10 October 2006
Abstract—Polymer hydrogels synthesized by crosslinking poly(allylamine hydrochloride) with ( )-epichlorohydrin in the presence
of D-glucose-6-phosphate monobarium salt do not show imprinting on the molecular level. A series of hydrogels was prepared using
the following five templates: D-glucose-6-phosphate monobarium salt, D-glucose, L-glucose, barium hydrogen phosphate (BaHPO4),
and D-gluconamide; a hydrogel was also prepared in the absence of a template. For all six hydrogels, batch binding studies were
conducted with D-glucose, L-glucose, D-fructose, and D-gluconamide. The extent of analyte sugar binding was determined using
1H NMR. Each hydrogel shows approximately the same relative binding affinity for the different sugar derivatives, and none
displays selectivity for either glucose enantiomer. The results of the binding studies correlate with the octanol–water partition
coefficients of the sugars, indicative that differential solubilities in the bulk polymer account for the binding affinities observed.
Thus, in contrast to templated hydrogels prepared using methacrylate- or acrylamide-based reagents, true imprinting does not
occur in this novel, crosslinked-poly(allylamine hydrochloride) system.
Ó 2006 Elsevier Ltd. All rights reserved.
Over the past decade, many new approaches have been
reported for the generation and characterization of
molecularly imprinted polymers.1,2 For example, im-
prints generated from hydrogels, hydrophilic polymer
networks that have been exploited in a variety of biolog-
ical and pharmacological applications,3 have potential
for use as ‘intelligent, controlled release’ materials.4
Recent reports indicate that imprinted hydrogels can
selectively bind both protein5,6 and small-molecule7
templates, including glucose.8
compared with only slightly more than 100 mg of D-fruc-
tose; control hydrogels formed in the absence of template
were reported to bind just over 100 mg of both D-glucose
and D-fructose per gram of dry polymer. Based upon
these results, the Kofinas group had concluded that the
GPS-Ba templated hydrogels had ‘recognizable cavities
in a water-swollen state with an affinity for the imprint’s
analog, glucose’.9 To better understand the basis of the
binding properties exhibited by this remarkable system,
we undertook the additional studies reported here.
We were particularly interested in the novel, carbohy-
drate-binding system reported by the Kofinas group.9,10
Unlike the majority of imprinted hydrogels reported to
date, this system does not involve the use of methacry-
late- or acrylamide-based reagents—rather the hydrogels
are synthesized by crosslinking poly(allylamine hydro-
chloride) (PAAÆHCl) with ( )-epichlorohydrin (EPI).
When prepared in the presence of 1.5 mol % of D-glu-
cose-6-phosphate monobarium salt (GPS-Ba), these
hydrogels show preferential binding of D-glucose relative
to D-fructose.9 As measured by batch studies in deion-
ized water, binding capacities of approximately 600 mg
of D-glucose per gram of dry polymer were reported,
Following the general procedure of the Kofinas
group,9,10 we prepared crosslinked hydrogels in the pres-
ence of GPS-Ba, D-glucose, L-glucose, BaHPO4, and D-
gluconamide;11,12 a control hydrogel was also prepared
in the absence of a template. By using D-glucose and
BaHPO4 as templates, we had hoped to dissect the inter-
actions of the GPS-Ba with the PAAÆHCl. We included
D-gluconamide in our studies because it is a potential
substrate for an intramolecular amide cleavage
reaction13 (our ultimate goal is to generate catalytic
molecular imprints). Finally, we were especially interested
in L-glucose—both as a template and an analyte—as a
probe of enantioselective binding.
For all six hydrogels, batch binding studies were con-
ducted in deionized water with the analytes D-glucose,
L-glucose, D-fructose, and D-gluconamide.14 (Although
the Kofinas group had also measured the extent of bind-
Keywords: Hydrogels; Molecular imprinting; Poly(allylamine); Glu-
cose; Fructose.
*
Corresponding author. Tel.: +1 413 542 2731; fax: +1 413 542
2735; e-mail: dehansen@amherst.edu
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.09.054

236
F. M. Fazal, D. E. Hansen / Bioorg. Med. Chem. Lett. 17 (2007) 235–238
The structures of all the templates and analytes are
shown in Figure 1.
While the Kofinas group had determined the extent of
1
binding colorimetrically,9,10 we employed an H NMR
assay that entails integrating signals for the sugar
relative to an internal acetic acid standard; the amount
of sugar remaining in solution is then determined by
reference to a standard curve. The calibration curve
for D-glucose is shown in Figure 2. (This assay does
not require reagents specific to the sugar of interest
and thus is general for any analyte.)
The raw data from our binding studies are shown in
Table 1; the sugar-binding capacities of the hydrogels
(in mg of sugar bound per gram of dry hydrogel) calcu-
lated from these data are shown in Table 2. Our results
indicate that all of the hydrogels, irrespective of tem-
plate, preferentially bind D-glucose over D-fructose.
The values for the separation factors a—defined as
[(binding capacity for glucose)/(binding capacity for
fructose)]— shown in Table 3 indicate that this en-
hanced binding of glucose is essentially independent of
template. Moreover, in contrast to the results reported
by the Kofinas group, we observed that the hydrogel
formed in the absence of template has a higher affinity
both for glucose and for fructose as compared with
the hydrogel formed in the presence of GPS-Ba. The a
value for D-glucose verses D-fructose binding by the
untemplated hydrogel is lower than that for all five of
the templated hydrogels, but the difference is not large.
Figure 1. The structures of the derivatives employed in this study.
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
To further probe the specificity of binding, we measured
the affinity of all six hydrogels for L-glucose—enantio-
selective binding is a hallmark of organic polymers
imprinted with optically pure chiral templates.15 (The
hydrogels prepared with achiral BaHPO4 and in the ab-
sence of template serve as controls.) All of the hydrogels,
however, bound both enantiomers of glucose with essen-
tially equal affinity (Tables 1 and 2).
20
25
30
35
40
45
50
55
Concentration of glucose (in mg/ml)
Figure 2. Calibration curve for D-glucose (n = 6, R2 = 0.981) generated
from the 1H NMR spectra of six solutions of known concentration.
The ratio of the integral of 25 mg/mL D-glucose relative to that of the
acetic acid standard is normalized to 1.0. The error bars indicate the
5% range routinely observed in the measurement of the integrals.
Our results indicate that molecular imprinting is not
responsible for the different binding properties of these
hydrogels. Rather, the binding data correlate to the oct-
anol–water partition coefficients (Poct) of the analyte
sugars. As shown in Figure 3 for the hydrogel generated
in the absence of a template, a plot of log [binding
capacity] vs. logPoct yields a straight line. (Similar plots
are obtained from the data for each of the templated
ing in pH 7 buffer, greater discrimination for glucose
over fructose had been observed in deionized water,9
and we thus did not use buffer in our experiments.)
Table 1. The results of the 1H NMR binding assaya
Template
Analyte sugar (mg/mL remaining in solution after incubation of a 50 mg/mL solution with hydrogel)
D-Fructose
D-Glucose
L-Glucose
D-Gluconamide
GPS-Ba
BaHPO4
45.7 2.3
43.5 1.0
43.6 0.7
45.2 0.4
46.3 0.4
41.8 0.5
38.6 2.2
34.3 0.1
35.3 1.1
35.7 0.8
36.2 0.0
32.7 1.0
37.7 2.4
34.2 2.3
36.1 1.3
36.2 2.6
38.2 0.5
35.4 1.2
31.2 1.7
32.6 1.8
30.3 1.4
31.0 1.6
31.4 1.1
28.8 0.7
D-Glucose
L-Glucose
D-Gluconamide
None
a The final concentration of analyte sugar (in mg/mL) remaining in solution after incubation of a 50 mg/mL solution of sugar in deionized water with
the hydrogels. A 0.40:15 w/v ratio of the dry polymer hydrogels to the analyte sugar solution was used. Each value shown is the average of two
independent measurements with the range indicated as ‘ ’.

F. M. Fazal, D. E. Hansen / Bioorg. Med. Chem. Lett. 17 (2007) 235–238
Table 2. Sugar-binding capacity (mg of sugar bound/g of dry polymer hydrogel)a
237
Template
Analyte sugar
D-Fructose
D-Glucose
L-Glucose
D-Gluconamide
GPS-Ba
BaHPO4
161 86
244 38
240 26
180 15
139 15
308 19
428 83
589 04
551 41
536 30
518 00
649 38
461 90
593 86
521 49
518 98
443 19
548 45
705 64
653 68
739 53
713 60
698 41
795 26
D-Glucose
D-Glucose
D-Gluconamide
None
a The values shown are calculated from the raw data in Table 1: Sugar-binding capacity = [(50 mg/mL initial analyte sugar concentration À analyte
sugar concentration remaining in solution)/0.02667 g polymer per mL analyte sugar solution]. The larger errors than in Table 1 are due to the small
amount of sugar bound relative to that remaining in solution.
Table 3. Separation factors a for D-glucose vs. D-fructose binding for
ing of a series of drugs into the hydrogel poly-2-
hydroxyethyl methacrylate also revealed a linear depen-
dence on analyte polarity, but with more hydrophobic
analytes preferentially binding in the hydrogel.)
each hydrogela
Template
Range
Average
GPS-Ba
BaHPO4
1.39–6.80
2.08–2.87
1.92–2.77
2.60–3.43
3.37–4.18
1.87–2.38
2.65
2.42
2.30
2.98
3.73
2.11
Finally, we would note that two factors likely account
for the discrepancies in the values reported here, as com-
pared with those reported earlier by the Kofinas group,9
for the glucose and fructose-binding affinities of the
hydrogels prepared with GPS-Ba and prepared in the
absence of template: one, the binding assays are run in
unbuffered, deionized water (for the reasons discussed
earlier), and thus small differences in pH are inevitable;
and two, binding equilibrium is not reached in the ‘stan-
dard testing time’ of 4 h,10 and thus kinetic factors will
influence the data obtained.
D-Glucose
L-Glucose
D-Gluconamide
None
a The a value = [(binding capacity for D-glucose)/(binding capacity for
D-fructose)]. The range in each a value reflects the experimental error
in Tables 1 and 2; the average a values are calculated from the
average binding values in Table 2.
3.0
2.9
2.8
2.7
2.6
2.5
2.4
In summary, polymer hydrogels prepared by crosslink-
ing poly(allylamine hydrochloride) with epichlorohydrin
in the presence of sugar templates do not exhibit
imprinting on the molecular level. Instead, differences
in the solubilities of the analyte sugars in the polymer
hydrogel bulk account for the observed data, a conclu-
sion consistent with the fact that far more sugar is
bound by the hydrogels than the 1.5 mol % of template
used in their generation.
Acknowledgments
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
We are grateful to the National Institutes of Health
(R15 GM63776) and to the Amherst College Faculty
Research Award Program, as funded by the H. Axel
Schupf ’57 Fund for Intellectual Life, for financial sup-
port. We are also pleased to thank the Howard Hughes
Medical Institute for support of F.M.F. (through an
Undergraduate Biological Sciences Education Program
award to Amherst College).
log Poct
Figure 3. Log [binding capacity untemplated hydrogel] vs. logPoct for
the analyte sugars. The binding capacity for glucose was taken to be
the weighted average of the binding capacity for D-glucose and L-
glucose. The logPoct values at 25 °C for glucose (both enantiomers will
have the identical value), D-fructose, and D-gluconamide have been
calculated to be À3.169 0.858, À1.629 0.870, and À3.700 0.870,
respectively.17 The logPoct value for glucose has also been experimen-
tally determined as À3.24 by Sangster18 and À2.82 0.04 by Mazzobre
et al.,19 numbers that agree, within experimental error, with the
calculated value.
References and notes
1. Molecularly Imprinted Materials: Science and Technology;
Yan, M., Ramstro¨m, O., Eds.; Marcel Dekker: New York,
2005.
2. van Nostrum, C. F. Drug Discov. Today: Technol. 2005, 2,
119.
hydrogels.) As logPoct becomes more negative, indica-
tive of an increased sugar polarity, adsorption into the
hydrogel also increases. Thus, the differential solubilities
of each sugar in the bulk polymer account for the bind-
ing data obtained. (An earlier study16 on the partition-
3. Kashyap, N.; Kumar, N.; Ravi Kumar, M. N. V. Crit.
Rev. Ther. Drug Carrier Sys. 2005, 22, 107.

238
F. M. Fazal, D. E. Hansen / Bioorg. Med. Chem. Lett. 17 (2007) 235–238
4. Byrne, M. E.; Kinam, P.; Peppas, N. A. Adv. Drug
Delivery Rev. 2002, 54, 149.
5. Hawkins, D. M.; Stevenson, D.; Reddy, S. M. Anal. Chim.
Acta 2005, 542, 61.
6. Xia, Y-q.; Guo, T-y.; Song, M-d.; Zhang, B-h.; Zhang,
B-l. Biomacromolecules 2005, 6, 2601.
allowed to sit for 24 h at room temperature. The solvent
was removed by rotary evaporation, and the resultant
solid was dried under vacuum for 2–3 days. The product
obtained was used without further purification [mp 147.2–
155.1 °C; 13C NMR (DMSO-d6): d 175.7, 73.0, 72.6, 71.9,
71.0 and 63.9 ppm].
7. Hiratani, H.; Alvarez-Lorenzo, C. Biomaterials 2004, 25,
1105.
13. Wolfrom, M. L.; Bennett, R. B.; Crum, J. D. J. Am. Chem.
Soc. 1958, 80, 944.
8. Oral, E.; Peppas, N. A. J. Biomed. Mat. Res., Part A 2006,
78A, 205.
14. For the batch binding studies, a 0.40:15 w/v ratio of the
dry polymer hydrogels to a 50 mg/ml solution of the
sugar analyte (i.e., D-glucose, L-glucose, D-gluconamide
or D-fructose) in deionized water (prepared from house
distilled water using a Corning MP-3A still) was used.
Typically the binding studies were performed on a scale of
50 mg of dried hydrogel in 1.875 mL of sugar solution.
The mixture was gently agitated for 4 h,9,10 and the
amount of sugar still in solution (and hence not bound to
hydrogel) was then immediately measured using 1H NMR
(recorded at 400 MHz on a JEOL GSX spectrometer). An
aliquot of the solution was removed (typically 0.5 mL)
and added to an equal volume of a standard acetic acid
solution in D2O [5:1000 (v/v) glacial acetic acid in D2O].
An 1H NMR spectrum was recorded and integrated
(though the final solution contained 50% H2O, the signals
for sugar and the acetic acid methyl group were well
resolved from the large water peak). The amount of sugar
still present in solution was determined by reference to a
calibration curve (see below), and the binding capacity of
the hydrogels (mg of sugar bound per gram of dry
hydrogel) then calculated. All binding studies were
performed in duplicate, and the average of the two
measurements was used to calculate the binding capaci-
ties. Calibration curves were constructed for D-glucose, L-
glucose, D-gluconamide, and D-fructose by employing
solutions of known concentration (25, 30, 35, 40, 45, and
50 mg/mL). As above, an aliquot of each of these
solutions (usually 0.5 mL) was added to the same volume
of the 5:1000 glacial acetic acid solution in D2O, and an
1H NMR spectrum recorded. The integration for signals
of the analyte compound and the acetic acid methyl
group was measured, and the ratio for the 25 mg/mL
solution of analyte normalized to 1.0. A calibration curve
was then constructed for each of the four analytes (n = 6;
R2 = 0.991 for D-fructose, 0.996 for D-gluconamide, 0.981
for D-glucose, and 0.994 for L-glucose).
9. Parmpi, P.; Kofinas, P. Biomaterials 2004, 25, 1969.
10. Wizeman, W. J.; Kofinas, P. Biomaterials 2001, 22, 1485.
11. In a typical procedure, a 25% w/v aqueous solution of
poly(allylamine hydrochloride) (average molecular weight
15,000) [1.0 g PAAÆHCl (10.7 mmol allylamineÆHCl mono-
mer) in 4.0 deionized water] was allowed to stir with
0.16 mol of the template (GPS-Ba, D-glucose, L-glucose,
BaHPO4, or D-gluconamide) for 2 h. 0.534 mL of a 10 M
NaOH solution (5.34 mmol) was then added and the
resultant solution was allowed to stir for 20 minutes,
followed by the addition of 0.109 mL of ( )-epichlorohy-
drin (1.39 mmol). As indicated by the numbers above, the
molar ratio of monomer/template/NaOH/crosslinker was
200:3:100:26, the values employed in the earlier reports.9,10
Upon addition of the EPI, gelation occurred in 10–15 min,
and the hydrogel was then allowed to sit undisturbed
overnight. The next day, the hydrogel was cut into
approximately 4 mm cubes with a razor blade and was
then washed with gentle shaking in 4 M NaOH solution
for 24 h to remove the template and any unreacted
reagents. The NaOH solution was decanted off the
hydrogel cubes, which were then repeatedly washed by
gently shaking in deionized water over a period of 5 days.
Each day, the hydrogels were washed 3–4 times for 1–2
hours, and after every wash the solution was decanted,
and the pH measured with pH paper. After the 5 day
period, overnight incubation of the hydrogels prepared
with D-glucose, L-glucose, and D-fructose, as well as the
one prepared in the absence of a template molecule,
yielded
a wash solution that was no longer basic
(pH ꢀ 6.5). However, even after the 5 days of washing,
the pH of the wash solution for the GPS-Ba and BaHPO4
hydrogels was still slightly basic (pH ꢀ 8). These hydrogels
were thus washed repeatedly for an additional 3–4 days,
but the pH did not drop further. All the polymers were
then dried, open to the air, for 18–24 h in an oven at 50 °C.
Between 0.6 and 0.7 g of washed and dried hydrogel was
routinely obtained. Assuming that all of epichlorohydrin
had fully reacted (with loss of HCl) with the polyallyl-
amine and that all salts and template had been fully
washed away, 0.69 g of dried hydrogel would correspond
to a 100% yield.
15. Ramstro¨m, O.; Yan, M. In Molecularly Imprinted Mate-
rials: Science and Technology; Yan, M., Ramstro¨m, O.,
Eds.; Marcel Dekker: New York, 2005; p 1.
16. Pitt, C. G.; Bao, Y. T.; Andrady, A. L.; Samuel, P. N. K.
Int. J. Pharm. 1988, 45, 1.
17. Data obtained from SciFinderÒ Scholar [calculated using
Advanced Chemistry Development (ACD/Labs) Software
V8.14 for Solaris, Ó 1994–2006 ACD/Labs].
12. D-Gluconamide was synthesized using the method of
Wolfrom et al.13 5.0 g of d-gluconolactone (28.0 mmol)
was dissolved in 18.3 mL of concentrated ammonium
hydroxide (28–30%, ꢁ0.3 mol), and the solution was
18. Sangster, J. J. Phys. Chem. Ref. Data 1989, 18, 1111.
´
19. Mazzobre, M. F.; Roman, M. V.; Mourelle, A. F.; Corti,
H. R. Carbohydr. Res. 2005, 340, 1207.