Angewandte
Chemie
formation of a skin layer. These observations are schemati-
cally presented in Figure 2c.
to the previous experiments
(1.5m). As expected, the leakage
of insulin became barely visible
owing to the increased network
density or suppressed diffusion
of insulin, thus reducing the
induction times. This observa-
tion is of critical importance to
prevent overdosing of insulin
once a normoglycemic condition
is reached. Remarkably enough,
this gel did not allow leakage of
insulin for a few days while
normoglycemia was maintained,
and upon re-increase of the
glucose concentration, the gel
did allow a burst release of
insulin with no visible lag time
(see the Supporting Informa-
tion, Figure S5). Neither deteri-
oration in the shape of the gel
nor cracks were observed after
repeated formation of the skin
layer over two weeks. Also
worthy of mention is that the
FITC-labeled (bovine) insulin was loaded onto the gel by
first allowing the gel to be hydrated (swollen) in a solution
containing insulin and then introducing the skin layer onto its
surface. Figure 3a–c show time-course profiles of the released
insulin under different addition patterns of glucose under
physiological conditions. It is clearly demonstrated that the
release of insulin from the gel can be continuously controlled
Figure 4. Improved insulin
release using a gel with
a dense polymer network
with twice the monomer
concentration compared to
Figure 3. (Top) Time-course
change in fluorescence inten-
sity of FITC-labeled bovine
insulin released from the gel
under physiological aqueous
conditions. (Bottom) Tempo-
ral pattern of the glucose
concentration investigated in
the experiment.
Figure 3. (Top) Time-course changes in the fluorescence intensity of
FITC-labeled bovine insulin released from the gel under physiological
conditions (pH 7.4, I=0.15, 378C). (Bottom) Temporal patterns of the
fluctuation in glucose concentration, investigated in each experiment.
normal pattern of physiological response to a stepwise
increase in blood glucose concentration is initially a rapid
and pulsatile release of insulin followed by a ramp in
secretion.[9] Indeed, this type of pattern could also be imitated
by lowering the amount of insulin remaining in the gel (see
the Supporting Information Figure S4).
by the skin layer (see also the Supporting Information,
Figure S4) with close correspondence to each addition pattern
of glucose. In Figure 3a, the glucose concentration was varied
between 1 glÀ1 and 2 glÀ1, which correspond to normogly-
cemia and the typical cut-off value for the diagnosis of
diabetes, respectively. Note that the profile in Figure 3b, for
which the glucose concentration is increased to up to 3 glÀ1, is
displayed at 10 times the scale as for Figure 3a and Figure 3c.
By comparing these spectra, it can be determined that under
such more severe hyperglycemic conditions the rate of release
is 10 times greater (Figure 3b; 3 glÀ1) as compared to the
other glucose concentrations (Figure 3a; 2 glÀ1). On the
other hand, under normoglycemic conditions (shown in
Figure 3c) the release is effectively halted over the timescale
of a day (at least 1000 min). Overall, sharp onsets of the
release in response to increases in glucose concentration are
readily achievable, whereas during the offset processes, that is
upon decreased glucose concentration, somewhat substantial
induction times are seen depending on the addition pattern of
glucose. This delayed shut-off response can be attributed to
a leakage of insulin through the skin layer at the premature
stage of the polymer-network contraction. Based on the fact
that the diffusion coefficient of a solute molecule in a gel is
dependent on the density of the polymer network, that is,
effective mesh size relative to the size of solute molecule, in
the present case insulin,[8] we anticipated that the leakage
could be reduced by increasing the density of the cross-linker
(N,N’-methylene-bis(acrylamide)) or the total concentration
of monomers at preparation (with consideration to the
molecular weight of insulin). Figure 4 verifies the validity of
such a remedy. For this particular experiment, the gel was
prepared with twice as high a polymer-network density
(monomer concentration at preparation: 3m) as compared
Our results demonstrate that a nonprotein-based, totally
synthetic smart gel can achieve “lasting” control over the
provision of insulin under conditions closely associated with
human glucose homeostasis. Owing to the significant rate and
magnitude of the glucose-induced phase transition, the gel
can promptly form a skin layer, thus providing a rationale for
the surface-controlled provision of insulin. This non-equilib-
rium mechanism leads to a remarkably shortened response
time (owing to the short diffusion distance across the skin
layer: typically on the order of 100 mm); this shortened
response time is an important criterion for the tight control of
the insulin dosage. Such a surface-limited mechanism (as
opposed to more bulk-dependent systems) in theory allows
the prediction of the release profile, independent of the size
and topological structure of the gel. This feature may
facilitate ease and accuracy in the management of the
administration of insulin. Furthermore, the size independency
may lead to compatibility with other well-developed tech-
nologies such as dwelling needles and semi-embedded
devices. Aside from the skin-layer-based strategy described
herein, the material can also be exploited as other mechan-
ically functional elements such as chemical valves or pumps
for related applications.[10] In addition, with its ability to cause
dramatic changes to other parameters including size, charge
density, conductivity, and permittivity (owing to the change in
hydration), the material provides a unique platform for
electrochemical sensors and actuators. There is also room for
size control (e.g., preparation in micro- or nanoporous
Angew. Chem. Int. Ed. 2012, 51, 2124 –2128
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2127