Triggered Release of Loads from Microcapsule-in-Microcapsule Hydrogel Microcarriers: En-Route to an "Artificial Pancreas".

A method to assemble stimuli-responsive nucleic acid-based hydrogel-stabilized microcapsule-in-microcapsule systems is introduced. An inner aqueous compartment stabilized by a stimuli-responsive hydrogel-layer (∼150 nm) provides the inner microcapsule (diameter ∼2.5 μm). The inner microcapsule is separated from an outer aqueous compartment stabilized by an outer stimuli-responsive hydrogel layer (thickness of ∼150 nm) that yields the microcapsule-in-microcapsule system. Different loads, e.g., tetramethyl rhodamine-dextran (TMR-D) and CdSe/ZnS quantum dots (QDs), are loaded in the inner and outer aqueous compartments. The hydrogel layers exist in a higher stiffness state that prevents inter-reservoir or leakage of the loads from the respective aqueous compartments. Subjecting the inner hydrogel layer to Zn2+-ions and/or the outer hydrogel layer to acidic pH or crown ether leads to the triggered separation of the bridging units associated with the respective hydrogel layers. This results in the hydrogel layers of lower stiffness allowing either the mixing of the loads occupying the two aqueous compartments, the guided release of the load from the outer aqueous compartment, or the release of the loads from the two aqueous compartments. In addition, a pH-responsive microcapsule-in-microcapsule system is loaded with glucose oxidase (GOx) in the inner aqueous compartment and insulin in the outer aqueous compartment. Glucose permeates across the two hydrogel layers resulting in the GOx catalyzed aerobic oxidation of glucose to gluconic acid. The acidification of the microcapsule-in-microcapsule system leads to the triggered unlocking of the outer, pH-responsive hydrogel layer and to the release of insulin. The pH-stimulated release of insulin is controlled by the concentration of glucose. While at normal glucose levels, the release of insulin is practically prohibited, the dose-controlled release of insulin in the entire diabetic range  is demonstrated. Also, switchable ON/OFF release of insulin is achieved highlighting an autonomous glucose-responsive microdevice operating as an "artificial pancreas" for the release of insulin.

Introduction

The synthesis of microcapsules and their applications have attracted growing interest in recent years.[1−3] Different methods to prepare microcapsules were reported. These include the chemical deposition of polymer or hydrogel coatings on substrate-loaded cores, followed by the etching of the core template.[4−6] For example, CaCO3 core templates were coated by a layer-by-layer deposition process of oppositely charged polyelectrolytes[7−11] or the use of interlayer biorecognition complexes,[12] e.g., lectin/saccharide complexes or the use of covalent bonds,[13,14] such as disulfides,[15] and the etching of the core templates, e.g., by EDTA, resulted in the substrate-loaded microcapsules. Alternatively, microcapsules loaded with substrates were prepared in oil-in-water[16] or water-in-oil[17,18] microemulsions. Different applications of microcapsules were suggested, including their use as drug carriers for slow release,[19−21] sensors,[22−24] microreactors for chemical transformations,[25,26] food and cosmetic additives.[27,28] A subclass of functional microcapsules includes stimuli-responsive capsules that are unlocked to stimulate the selective and programmed release of the loads in the presence of appropriate biomarkers or environmental conditions. For example, pH,[29−31] light,[32] heat,[33,34] gases,[35] salts,[36] chemical reducing agents,[37] carbohydrates,[38] enzymes,[39,40] magnetic field,[41] and ultrasonic or microwave agitation[42−44] were used to release the embedded substrate loads. Specific applications of stimuli-responsive microcapsules include the triggered release of encapsulated drugs by specific biomarkers[45] or appropriate cellular/tissues environmental conditions.[46,47]

The base sequence of nucleic acids dictates their structural reconfiguration functions in the presence of auxiliary triggers. For example, the pH-induced formation of i-motif[48,49] or triplex[50,51] structures, the K+-ions-stabilized G-quadruplexes and their separation by crown ethers,[52,53] and the cooperative stabilization of duplex nucleic acids by metal-ion bridged mismatched bases (e.g., T–Hg2+–T or C–Ag+–C).[54] In addition, sequence-specific recognition of ligands (aptamers)[55−57] or sequence-dictated catalytic properties of nucleic acids (DNAzymes)[58,59] introduce important motifs for controlling the structure and chemical functions of oligonucleotides. Not surprisingly, nucleic acids provide a useful “tool-box” to synthesize stimuli-responsive nucleic acid drug-loaded micro/nanocarriers.[60,61] All-DNA-stabilized drug-loaded microcapsules were reported and their unlocking by light,[32,62] pH,[63] and the formation of aptamer (ATP or VEGF)–ligand complexes were demonstrated.[64] In addition, drug-loaded nucleic-acid-modified hydrogel-stabilized microcapsules were prepared, and the triggered reversible reconfiguration of the oligonucleotide units was used to control the stiffness of the hydrogel coating and to switch the reversible ON/OFF release of the encapsulated drugs. Cell experiments demonstrated the selective cytotoxicity of chemotherapeutic drug-loaded microcapsules toward cancer cells.[65]

Further enhancement of the complexities of stimuli-responsive microcapsules would involve the challenging assembly of microcapsule-in-microcapsule systems, where two aqueous compartments are separated and stabilized by two different stimuli-responsive layers. Such systems could be applied for the programmed release of one of two drugs or the parallel release of two drugs (or prodrug and activator), and could also act as an organized compartmentalized containment for chemical reactions. While microcapsule-in-microcapsule systems were fabricated in microemulsion, the interconnected microcapsules were prepared using microfluidic devices[18,66−68] and the design of stimuli-responsive switchable microcapsule-in-microcapsule systems, and the applications of these structures are basically unexplored.

Here we wish to report on the synthesis and characterization of hydrogel-based stimuli-responsive microcapsule-in-microcapsule systems that include different loads in separated aqueous compartments comprising the structures. We demonstrate that in the presence of one or two triggers, the selective release of one load, or the two loads, from the carrier proceeds. Also, in the presence of an appropriate trigger, loads embedded in the two aqueous compartments of the carrier can be mixed. In addition, we apply a microcapsule-in-microcapsule system as a functional unit that operates as a closed-loop device that senses glucose and releases insulin, acting as a model of an “artificial pancreas”.

Results and Discussion

The synthesis of dually triggered, stimuli-responsive, nucleic acid-based hydrogel-microcapsule-in-microcapsule microstructures carrying two different loads (e.g., tetramethyl rhodamine-dextran, TMR-D, and CdSe/ZnS quantum dots, QDs) is schematically outlined in Figure . In addition, the mechanism of unlocking the capsules and the dictated release of the loads by two different triggers, Zn2+-ions and/or pH, are schematically exemplified in Figure A. CaCO3 microparticles loaded with TMR-D were coated with a poly(allylamine hydrochloride), PAH, layer followed by the electrostatic adsorption of DNA strand (1) on the coated particles. The strand (1) acts as a promoter strand for inducing the hybridization chain reaction (HCR) in the presence of two carboxymethyl cellulose (CMC) polymers, P1 and P2, functionalized with the anchoring tether (2) that acts as an anchoring site for the stimuli-responsive units (vide infra) and hairpins H1 and H2, respectively (for the detailed sequences see Table S1). Note that hairpin H2 was conjugated to P2 through a tether (x) linked to the polymer. This modification is essential to retain the appropriate directionality of the hairpins for the HCR process. The stimuli-responsive unlocking unit of the first hydrogel layer in this specific example is Zn2+-ions-dependent DNAzyme (4). Accordingly, the strand (3) includes the sequence of the Zn2+-ions-dependent DNAzyme substrate extended at its 5′- and 3′-ends by sequences that are complementary to the anchoring tethers, (2), associated with the polymer chains, P1/P2. The strand (4) includes the sequence of the Zn2+-ions-dependent DNAzyme consisiting of a loop domain for binding Zn2+-ions and two extended arms that bind to the substrate (3) as shown in Figure B. The loading of the nucleic acid units on the CMC chains was evaluated spectroscopically and corresponded to 1:60 (nucleic acid:CMC unit). For experimental details describing the evaluation of the average molecular weights and nucleic acid loadings on the different chains, see the Supporting Information Figures S1–S5 and experimental section.

Figure 1

Synthesis of carboxymethyl cellulose (CMC)-stimuli-responsive nucleic acid-based hydrogel microcapsule-in-microcapsule system loaded with two different fluorophores and unlocking of the system by two different triggers, Zn2+-ions and pH, which results in the release of the loads from the different aqueous compartments. (B) Detailed outline of the substrate (3) and Zn2+-ions-dependent DNAzyme (4) associated with the inner hydrogel layer. (C) Detailed outline of the outer hydrogel layer comprising the pH-responsive reconfiguration of the (5)/(6) duplex nucleic acids into the separated i-motif structures (at pH 5.5, lower stiffness) and the reverse assembly of the (5)/(6) duplex-bridged hydrogel (at pH 7.2, higher stiffness).

Treatment of the (1)-functionalized microparticles with the polymer chains P1 and P2 initiates the HCR process, where promoter (1) hybridizes with hairpin H1 associated with P1, the “toehold” strand associated with the open H1 hybridizes with H2 and the “toehold” single strand of the opened H2 reopens H1 and vice versa. This HCR process leads to the formation of a CMC hydrogel film, cross-linked by duplexes formed upon the counter-opening of hairpins H1/H2, on the microparticle core. As the HCR process was stimulated in the presence of strands (3) and (4), the free anchoring tether (2) hybridizes with the substrate chain (3) and the single-stranded domain of (3) hybridizes with the arms of (4) that includes in the loop domain the Zn2+-ions-dependent DNAzyme sequence. This yielded a supramolecular structure that in the presence of the Zn2+-ions generates the activation of the Zn2+-ions-dependent DNAzyme. Figure B shows the domains of the duplex nucleic acid cross-linking units generated by the HCR process and the supramolecular structure of the (2)/(3)/(4)/(2) units. The core microparticles coated with the first Zn2+-ions stimuli-responsive hydrogel were then reacted with a mixture of CaCl2, CdSe/ZnS QDs (carboxylic acid functionalized) and Na2CO3. The in situ generated QDs-impregnated CaCO3 particles were deposited on the CMC hydrogel coating layer using Ca2+-ions as “glue” that binds the CaCO3 interlayer to the CMC hydrogel.[69] Subsequently, the CaCO3 interlayer was first modified with PAH and then with the promoter strand (1). The resulting (1)-functionalized particles were interacted with the polymer chains P1 and P2 in the presence of the strands (5) and (6). Strand (5) includes at its 3′-end the (q′)-sequence extended by the sequence (z) that is capable to form the i-motif structure under acidic conditions. The strand (6) includes at its 5′-end the sequence (z′) (complementary to (z)), extended by the sequence (q′) (complementary to (q)), Figure C. Under these conditions, the (1)-stimulated HCR process led to the formation of the second hydrogel layer consisting of the cross-linking duplex nucleic acids generated between H1 and H2 and cooperatively stabilized by the supramolecular structure consisting of the (5)/(6) duplexes anchored through the (q′) toeholds to the (2)-tethers associated with the polymers P1 and P2 (note that the supramolecular Zn2+-ions-dependent DNAzyme (3)/(4), Figure B, and the duplex nucleic acid structure (5)/(6) are cross-linked to the polymer chains by identical (q)/(q′) duplex anchoring site, where q corresponds to strand (2)). The core and interlayer of CaCO3 coated by the two hydrogel layers were, then, etched with EDTA, yielding the microcapsule-in-microcapsule structure consisting of two aqueous compartments separated by two stimuli-responsive hydrogel layers. The inner aqueous reservoir, protected by the Zn2+-ions-dependent stimuli-responsive hydrogel, includes the TMR-D load, and the outer aqueous compartment includes the CdSe/ZnS QDs, protected by the outer pH-responsive hydrogel layer.

Figure A describes the triggered release of the loads associated with the two aqueous compartments of the microcapsule-in-microcapsule system. Treatment of the microcapsule-in-microcapsule system with Zn2+-ions cleaves the substrate (3) and leads to a lower degree of cross-linking (Figure B) and, consequently, to a inner hydrogel layer of lower stiffness, resulting in the mixture of the loads present in the two aqueous compartments. Subjecting the microcapsule-in-microcapsule system to pH 5.5 results in the reconfiguration of the strand (5) into the i-motif structure and in the separation of the (5)/(6) duplex nucleic acids bridges (Figure C). This process leads to an outer-hydrogel layer of lower stiffness that allows the release of the load from the outer aqueous compartment, while the load in the inner reservoir is not released. Treatment of the microcapsule-in-microcapsule system with Zn2+-ions and pH 5.5 leads to the triggering of the inner and outer hydrogel layers into lower stiffness matrixes. This allows the release of the loads from the inner and outer reservoirs into the bulk aqueous solution. The versatility of the method to assemble and trigger the microcapsule-in-microcapsule system should be emphasized: (i) We exemplify the production of CMC–microcapsule-in-microcapsule systems. This concept can be, however, adapted to other stimuli-responsive hydrogel materials. For example, Figure S6 and accompanying discussion describe the synthesis of stimuli-responsive polyacrylamide-based microcapsule-in-microcapsule hydrogel systems using an identical concept. For experimental details describing the evaluation of the average molecular weight and nucleic acid loadings on the different chains, see the Supporting Information Figures S7–S11. (ii) We describe the loading of the aqueous compartments with TMR-D and QDs. The method can be applied, however, to load the different aqueous compartments with any other dyes, particles, proteins, or nucleic acids MW > 5 kDa (vide infra). (iii) We introduce Zn2+-ions-dependent DNAzymes and the pH-stimulated formation of the i-motif structures as unlocking motifs for releasing the loads. Nonetheless, the unlocking mechanisms are not limited to these hydrogel uncaging principles. Other methods to reconfigure the bilayer microcapsule-in-microcapsule hydrogel system may be envisaged, including the application of other DNAzymes or the use of other stimuli-responsive DNA-reconfiguration principles, e.g., K+-ions-stabilized G-quadruplexes (vide infra). (iv) We note that the bilayered microcapsule-in-microcapsule microstructures reveal in/out permeation of low-molecular-weight substrates or products. Nonetheless, loads exhibiting MW >5 kDa are confined to the respective aqueous compartments, with no leakage through the hydrogel layers of the respective compartment, provided that the hydrogels exist in their locked higher stiffness configurations.

In the first stage, the bilayer microcapsule-in-microcapsule structures were characterized. Figure A, panel I, shows the SEM image of bare inner CaCO3 core microparticles. Figure A, panel II shows the SEM image of a CaCO3 microparticle-in-microparticle after the stepwise deposition of the Zn2+-ions-dependent DNAzyme and pH responsive hydrogel layers. A rough, porous coating is observed, consistent with the formation of hydrogel-coated particles. The formation of a bilayer hydrogel coating on the CaCO3 microparticles is supported by identifying “defective”, broken bilayer-coated particles, Figure A, panel III. The yield of intact bilayer hydrogel coated microparticles is very high, ∼99% yet ∼1% consists of defective structures, such as displayed in panel III. The SEM images clearly demonstrate the bilayer coating of the particles and eventually show “broken-off” pieces of microparticles coated by hydrogel. The formation of the bilayer hydrogel on the CaCO3 microparticles is further supported by focused ion-beam (FIB) imaging. Figure B, panel I shows a FIB image of a cut particle coated only with the first hydrogel layer. The thickness of the hydrogel layer is ∼150 nm. Figure B, panel II shows a FIB image of the two-hydrogel coating layers, separated by the outer CaCO3 template (thickness 150–200 nm). The two hydrogel layers reveal a similar thickness (∼150 nm).

Figure 2

(A) SEM images corresponding to: Panel I, the CaCO3 microparticles before modification. Panel II, the microparticles coated with the hydrogel layers (prior to etching off the CaCO3 component). Panel III, an example of a broken microparticle consisting of two-deposited hydrogel layers. (B) Focus-ion beam (FIB) images of: Panel I, the microparticle core coated with the first hydrogel layer. Panel II, the microparticle consisting of CaCO3 core-hydrogel layer/CaCO3 interlayer-hydrogel layer.

The formation of the bilayer microcapsules after etching the CaCO3 template units was supported by confocal fluorescence microscopy imaging experiments. Figure A shows the confocal fluorescence microscopy images of the TMR-D (red) and QDs (green) loaded bilayer microstructures, and the respective bright-field images before etching the CaCO3 templates, panel I, and after the EDTA-stimulated removal of the templates, panel II. The overlaid fluorescence images indicate the formation of two separated fluorophore-containing compartments before and after etching (outside compartment green, inside compartment red). The microstructures generated after etching-off the CaCO3 template show a void internal volume that implies the formation of the two-aqueous compartment microcapsule-in-microcapsule system Figure A, panel II, a,b. The overlay of a and b shows the location of the two different fluorophore-containing compartments, Figure A, panel II, c. In addition, Figure B shows the orthogonal projections of the bilayer microparticles. Distinct and separated compartments are visible at the ZX projection and ZY projection.

Figure 3

(A) Confocal microscopy images corresponding to the bilayer hydrogel coated CaCO3 microparticles that includes: (a) CdSe/ZnS QDs in the CaCO3 interlayer (λex = 464 nm; λem = 482 nm); (b) TMR-D in the inner CaCO3 core (λex = 546 nm; λem = 580 nm); (c) overlay of the channel-separate fluorescence images shown in parts a and b; (d) bright-field image of the bilayer hydrogel functionalized microparticles. Panel I-before etching and Panel II after etching. (B) Orthogonal projections of the overlaid confocal microscopy images of the bilayer hydrogel-functionalized microparticles that include the CdSe/ZnS QDs in the outer CaCO3 core and TMR-D in the inner CaCO3 core.

In the next step, the triggered release of the loads from the bilayer microcapsules was examined. Treatment of the microcapsules at pH 5.5 resulted in the release of the QDs, Figure A, curve a. Under these conditions, the TMR-D load, entrapped in the inner compartment, was not released, Figure A, curve b. In addition, at pH 7.2 no release of the QDs was observed, Figure S12, curve a. This allowed the switchable pH-stimulated release of the QDs from the outer compartment of the microcapsules, Figure A, inset. Figure B depicts the time-dependent release of the QDs (from the outer compartment), curve a, and of the TMR-D (from the inner compartment), curve b, upon subjecting the microcapsules to pH 5.5 and Zn2+-ions (20 mM). Under these conditions, the two loads were released from the bilayer compartments. Control experiments revealed that at pH 7.2, and in the absence of Zn2+-ions, no release of the QDs (curve a) or TMR-D (curve b) was detected, Figure S12. (The triggered release of the loads from the polyacrylamide bilayer microcapsules was also examined, Figures S13–S17).

Figure 4

Triggered time-dependent release of the loads from the bicompartmentalized aqueous compartments of the microcapsule-in-microcapsule system loaded with TMR-D in the inner compartment and CdSe/ZnS QDs in the outer compartment. The compartments are separated by a Zn2+-ions-dependent DNAzyme hydrogel (inner compartment). The outer compartment and the bulk solution are separated by a pH-responsive hydrogel. (A) The release of (a) CdSe/ZnS QDs and (b) TMR-D, upon subjecting the microcapsule-in-microcapsule system to pH 5.5. Inset: switchable “ON” and “OFF” release of the CdSe/ZnS QDs from the microcapsule-in-microcapsule system upon the reversible treatment at pH 5.5 and pH 7.2. (B) The release of (a) CdSe/ZnS QDs and (b) TMR-D upon treatment of the microcapsule-in-microcapsule system with Zn2+-ions, 20 mM and pH 5.5.

Furthermore, at pH 5.5 and in the presence of Zn2+-ions the release of the TMR-D from the microcapsules was controlled by the concentration of Zn2+-ions, Figure S18. As the concentration of Zn2+-ions increases, the release of TMR-D is enhanced, consistent with the improved unlocking of the inner compartment by the Zn2+-ions-dependent cross-linking DNAzyme. Treatment of the microcapsules with Zn2+-ions at pH 7.2 did not lead to the release of TMR-D or QDs to the bulk solution. Nonetheless, under these conditions, the inner compartment was unlocked by the Zn2+-ions-dependent DNAzyme and this allowed the mixing of the fluorophores between the two compartments. The selective unlocking of the inner-compartment of the microcapsules by the Zn2+-ions dependent DNAzyme cross-linkers and the mixture of the loads between the two compartments was confirmed by confocal fluorescence microscopy imaging. Figure shows the confocal fluorescence microscopy images and bright-field images of the microcapsule-in-microcapsule fluorophore-labeled microstructures before the addition of Zn2+-ions to the system, panel I, and after treatment of the microcapsules with Zn2+-ions, 20 mM, for different time intervals, panels II–IV. The nontreated microcapsule-in-microcapsule microstructures, panel I, show the specific green fluorescence (QDs) and the red fluorescence (TMR-D) upon the single-channel excitation of the fluorophores, images a and b, respectively. Image c depicts the overlay of the two-channel fluorescence images of the microcapsule-in-microcapsule microstructures. An inner red fluorescence separated from an outer green fluorescence is observed, indicating that the fluorophores are confined in the two separated compartments. Treatment of the microcapsule-in-microcapsule microstructures with Zn2+-ions results in a yellow boundary between the inner red compartment and the outer green compartment that expands with time to a fully overlaid yellow fluorescence image, indicating the complete mixing of the two fluorophores as a result of the DNAzyme-catalyzed unlocking of the inner compartment. For additional confocal fluorescence microscopy images corresponding to control experiments probing the release of QDs and of TMR-D upon triggering the bicompartment-loaded microcapsule-in-microcapsule system, see Figure S19.

Figure 5

Confocal microscopy images and bright-field images of the microcapsule-in-microcapsule system loaded with CdSe/ZnS QDs and TMR-D upon triggering the system with Zn2+-ions, 20 mM, at different time intervals. (a) Single channel QDs fluorescence (green) (b) Single channel TMR-D fluorescence (red). (c) Overlaid fluorescence of parts a and b. (d) Bright-field image. Panel I, t = 0 min; panel II, t = 5 min; panel III, t = 10 min; panel IV, t = 25 min. Note that after 25 min an overlaid yellow image is observed confirming the mixture of the fluorophores in the two aqueous compartments.

The concept of synthesizing stimuli-responsive microcapsule-in-microcapsule microstructures and the triggered release of the loads from the bilayer assemblies were expanded to include another switchable trigger. Figure S20 shows the synthesis of microcapsule-in-microcapsule microstructures consisting of an outer hydrogel layer cross-linked by K+-ions-stabilized G-quadruplexes and an inner hydrogel layer composed of the Zn2+-ions-dependent DNAzyme as cross-linkers. As before, the outer aqueous compartment was loaded with QDs, and the inner aqueous compartment was loaded with TMR-D. Subjecting the microcapsule-in-microcapsule system to 18-crown-6-ether (CE) separated the G-quadruplex cross-linking bridges, resulting in a hydrogel layer of lower stiffness that led to the release of the QDs, Figure S21A, curve a. Under these conditions the TMR-D, confined to the inner compartment, was not released from the microstructures, Figure S21A, curve b. The unlocking of the outer hydrogel layer could be switched between “ON” and “OFF” states by the cyclic treatment of the microstructures with CE and K+-ions, Figure S21A, inset. Treatment of the microcapsule-in-microcapsule system with CE and Zn2+-ions resulted in the triggered unlocking of the two hydrogel layers leading to the release of the loads from the two aqueous compartments, Figure S21B. In addition, subjecting the microcapsule-in-microcapsule system to Zn2+-ions resulted only in the mixing of the fluorophores in the two aqueous compartments, with no release of the loads to the bulk solution.

The development of a versatile method to prepare two-reservoir stimuli-responsive microcapsule-in-microcapsule systems turns these ensembles into ideal drug carriers for controlled switchable release. Indeed, as a proof-of-concept, we applied these systems to tailor a glucose-regulated insulin release microcapsule-in-microcapsule system. Diabetes mellitus is a major public health problem across the world, accompanied by the constant increase of diabetes patients.[70−73] Besides oral administration of drugs, the injection of insulin is a frequent practice to control glucose levels in blood. Nevertheless, poor control over glucose levels is often experienced,[74,75] and complications such as hypoglycemia[76] are often encountered. Indeed, major efforts were directed in the past decades to develop autonomous glucose-responsive materials for the controlled release of insulin.[75,77−80] Insulin-loaded polymer nanoparticles responding to pH changes generated by the glucose oxidase (GOx)-catalyzed oxidation of glucose and accompanied by the generation of gluconic acid[81] or glucose-induced swelling of boronic acid-functionalized polymers acted as useful glucose-triggered insulin release matrices.[82−87] In addition, pH-responsive polysaccharide particles loaded with insulin, glucose oxidase, and catalase were used as functional carriers for the pH-stimulated release of insulin and the concomitant catalase-induced degradation of accompanying GOx generated H2O2 that could lead to harmful reactive oxygen species (ROS).[88] As well, a closed-loop insulin delivery system consisting of an insulin-loaded silicon reservoir gated by enzyme-loaded pH-responsive hydrogel nanoparticles was used for the release of insulin.[89] Although substantial progress in designing closed-loop insulin release systems was demonstrated, the development of other carriers revealing increased insulin loading, fast, selective, and reversible insulin release functions under high/normal levels of glucose, injectability of the carrier (e.g., subcutaneous injectability with microneedle arrays), and elimination of immunogenic effects of undesired leakage of proteins, such as GOx, are desired. In contrast to the reported carriers that include the insulin and stimuli-release element in one compartment, the two-reservoir, hydrogel-based microcapsule-in-microcapsule system might introduce several advantages: (i) The thin hydrogel layers separating the two compartments are anticipated to allow effective switchable and selective release of insulin while protecting GOx in a confined inner compartment against leakage. (ii) The availability of an aqueous compartment for the solubilization of insulin might allow enhanced loading of the drug into the carrier. (iii) The hydrogel-based microcapsule-in-microcapsule carriers are suspendable in water/buffer solutions and, thus, their injectability could be of useful practice. Accordingly, we applied the bilayer hydrogel layers microcapsule-in-microcapsule to tailor a model system acting as an artificial pancreas. Figure depicts the assembly of the bilayer microcapsules carrying GOx and insulin in the distinct compartments comprising the microcapsules. The synthesis of GOx/insulin-loaded bilayer microcapsule-in-microcapsule system followed the same protocol used in the previous systems, where GOx is confined to the inner aqueous compartment and insulin is confined to the outer aqueous compartment. It should be noted that in this microcapsule-in-microcapsule system the inner layer consists of the (2)/(3)/(2) bridged cross-linking units in the absence of the added DNAzyme sequence (4), Figure B. Thus, these bridges are stable under all external conditions.

Figure 6

(A) Schematic synthesis of the microcapsule-in-microcapsule system composed of glucose oxidase (GOx) entrapped in the inner aqueous compartment, stabilized by a supramolecular duplex nucleic acid-bridged hydrogel, and the fluorophore-labeled insulin loaded in the outer aqueous compartment, stabilized by a pH-responsive hydrogel layer. Scheme depicts the switchable pH-stimulated release of the insulin by the reconfiguration of the duplex units, bridging the outer hydrogel layer into the i-motif structures. Inset: schematic biocatalytic reaction of GOx entrapped in the inner aqueous reservoir. Note that the release of insulin is reversibly controlled by the concentration of glucose and the accompanying GOx-stimulated pH changes. (B) Detailed outline of the nucleic acid bridging elements associated with the inner hydrogel layer. (C) Detailed outline of the nucleic acid bridging units comprising the outer pH-responsive reconfiguration of the (5)/(6) supramolecular duplexes into the separated i-motif structures (at pH 5.5, lower stiffness) and the reverse assembly of the (5)/(6) duplex-bridged hydrogel (at pH 7.0, higher stiffness) associated with the outer hydrogel layer.

GOx was loaded in the core CaCO3 microparticles, and these were coated with PAH and further functionalized with the promoter strand (1). The (1)-modified particles were subjected to the polymer chains P1 and P2, where P1 was modified with hairpin H1 and P2 was modified with hairpin H2. The two polymer chains included identical nucleic acid anchoring tether (2). The (1)-stimulated HCR process, in the presence of strand (3), resulted in the first hydrogel layer, Figure B. Subsequently, CaCO3 loaded with insulin was deposited on the first layer hydrogel. The deposited CaCO3 was then modified with PAH and the promoter strand (1). The (1)-functionalized surface was used to activate the HCR process in the presence of P1, P2, and the strands (5) and (6) to yield the second (outer) hydrogel layer that is cooperatively stabilized by the duplex nucleic acids H1open/H2open and the superstructure bridge (2)/(5)/(6)/(2), Figure C. After etching the CaCO3 template with EDTA, the microcapsule-in-microcapsule system is formed, where GOx and insuline are entrapped in the inner and outer aqueous compartments, respectively.. The inner hydrogel layer is stabilized by the duplex nucleic acids H1open/H2open and the (2)/(3)/(2) bridges. The outer hydrogel layer is composed of the duplex nucleic acids H1open/H2open and the pH-responsive complex (2)/(5)/(6)/(2) (cf. Figure B and Figure C).

The strand (5) associated with the bridging unit is cytosine rich, and at pH 5.5 it reconfigures into an i-motif structure. This results in the separation of the bridging units and the unlocking of the hydrogel toward the release of insulin. The further neutralization of the pH locked the structure by dissociation of the i-motif units and regeneration of the gated higher stiffness hydrogel that prohibits the release of insulin. Thus, by the cyclic control of the pH at the hydrogel boundary, the switchable ON/OFF release of insulin proceeds.

Under acidic conditions, the separation of the duplex nucleic acids (5)/(6) leads to a hydrogel of lower stiffness that provides the key properties for the operation of the “artificial pancreas”. Note that the bilayer hydrogel microcapsules are freely permeable to low-molecular weight substrates, and while the lower stiffness hydrogel is permeable to proteins, <5 kDa, the higher stiffness hydrogel is nonpermeable to proteins. These features of the bilayer microcapsule boundaries enable the operation of the “artificial pancreas”. The permeation of glucose across the two layers of the microcapsule-in-microcapsule boundaries leads to the GOx-catalyzed aerobic oxidation of glucose to gluconic acid and H2O2. The formation of gluconic acid acidifies the inner aqueous compartment and protons are permeating to the outer compartment and acidifying it. The acidic pH induces the reconfiguration of the duplex nucleic acids (5)/(6) associated with the outer hydrogel layer into the i-motif structure, a process which leads to the formation of a lower stiffness hydrogel that allows the release of insulin to the bulk solution. Note that the pH-changes stimulated by GOx and, thus, the stiffness changes and release efficiency by the outer hydrogel layer are controlled by glucose concentration. Furthermore, the control over the release of insulin by the concentrations of glucose is anticipated to yield a switchable dose release mechanism as required for an artificial pancreas-mimicking device. Figure A shows the time-dependent release of fluorescein-labeled insulin from the microcapsule system in the presence of different concentrations of glucose. In the absence of glucose, no release of insulin is observed, curve a. At a glucose concentration of 5 mM (normal levels of glucose in human blood), inefficient release of insulin is observed, curve b. At higher concentrations of glucose, ≥10 mM, effective release of insulin is observed, and the release of insulin is enhanced as the concentration of glucose increases, curves c–f. At the high glucose concentration, 30 mM, the release of insulin reaches a saturation value after ∼35 min, curve f. Further increase in the concentration of glucose does not affect the insulin release profile, and the release curve overlap curve f, Figure A. That is, the saturated release of insulin corresponds to the release of the entire insulin loaded in the microcapsule-in-microcapsule system. Using an appropriate calibration curve for the fluorophore-labeled insulin and knowing the concentration of the carrier, we evaluated the loading of insulin in the bilayer microcapsule to be 7.4 × 10–13 mol/capsule, see Supporting Information.

Figure 7

(A) Time-dependent fluorescence changes corresponding to the release of the fluorescein-labeled insulin from the microcapsule-in-microcapsule systems subjected to different concentrations of glucose: (a) 0 mM, (b) 5 mM, (c) 10 mM, (d) 15 mM, (e) 20 mM, and (f) 30 mM. (B) Switchable “ON” and “OFF” release of the insulin upon subjecting the microcapsule-in-microcapsule system to “elevated”/“normal” glucose levels. The capsules are initially introduced into a glucose solution, 10 mM, and this results in the release of insulin for 50 min and the leveling off of the insulin release (where glucometer evaluated values of the glucose in the bulk solution correspond ca. to 4 mM). At point a, the capsules are subjected to an additional added concentration of glucose, 10 mM. This results in the reactivation of the release of insulin and the leveling off of the release process after ∼60 min.

Figure B shows the switchable glucose-triggered release of insulin from the microcapsules system. In this experiment, the insulin-loaded microcapsules were subjected to glucose, 10 mM, and after a time-interval of ∼50 min the release of insulin reaches a saturation value. By applying the appropriate calibration curve, we estimated that ∼20% of the insulin loaded in the carriers was released. Parallel measurement of the glucose concentration in the solution, using a glucometer, resulted in a decrease of the glucose concentration from ∼200 mg/dL to 75 mg/dL (10 mM to 4 mM). The saturation value observed after 50 min is consistent with the inefficient release of insulin at glucose concentrations lower than 5 mM. At point a, marked with an arrow, the microcapsules were subjected to an additional increase in the glucose concentration (10 mM or 200 mg/dL). This switches on the release of the fluorophore-labeled insulin, resulting in a saturation value after ∼60 min. In this second step, an additional 40% of the loaded insulin was released. Using the respective calibration curve, we estimate that ∼25 μM of insulin was released within this step. Note that even though the additional concentration of glucose in the second step was 10 mM, a significantly higher insulin release is observed as compared to the release of insulin in the first step. The glucometer-measured values of glucose outside the microcapsules indicated that even after a time-interval of 100 min, where the immediate release reaches saturation, the outside concentration of glucose is ∼4 mM. Presumably, the flux of glucose penetration into the microcapsules decrease upon completion of the first insulin release cycle. The residual concentration of “inactive” glucose in the bulk solution increases the glucose concentration in the solution to ∼15 mM (∼4 mM from the first cycle that were not consumed in the addition to 10 mM of the second cycle). This results in enhanced pH changes, cf. Figure A, and increased concentrations of the released insulin. Attempts were directed to evaluate the number of switchable insulin release cycles and to establish principles to optimize the functions of the “artificial pancreas”. Toward this goal we prepared microcapsules of higher loading of insulin (1.3 × 10–12 mol/capsule compared to 7.4 × 10–13 mol/capsule). Using microcapsules at a concentration corresponding to 1700 capsules/μL and repeated added concentrations of glucose, each cycle 7.0 mM, we were able to switch the ON/OFF release of insulin for three cycles, Figure S22. The cyclic release of insulin demonstrated in Figure S22 indicates a 50–60% release of insulin from the total insulin (Figure S23 and accompanying discussion). Thus, the insulin release can be, in principle, further cycled.

In addition, the cyclic and reversible performance of the bilayer, insulin-loaded carriers was examined, Figure S24. In this experiment, the insulin loaded carriers were washed after the insulin-release cycle to remove exterior glucose and insulin. Glucose, 7 mM, in a new HEPES buffer solution was then added to reactivate the release of insulin. Figure S24 demonstrates that, at least, five ON/OFF release cycles could be stimulated by the microcapsule carriers. Thus, in principle, by increasing the number of capsules and the loaded insulin, the further optimization of the release system could be envisaged.

The GOx biocatalyzed oxidation of glucose in the core of the bicompartment microcapsule system is, however, accompanied by the generation of H2O2. This product might be harmful as it acts as precursor for ROS. To eliminate this disturbing effect, the enzyme catalase, which disproportionates H2O2 into H2O and O2, was added to the GOx-loaded insulin-modified microcapsule-in-microcapsule system, Figure S25. (For confocal microscopy images confirming the confinement of GOx and catalase/insulin to the respective aqueous containments, see Figure S26.) Using the respective calibration curves, Figure S27, we estimate that the loading of the catalase in the inner core corresponded to ∼7.8 × 10–14 mol/capsule and the loading of GOx in the same inner compartment corresponded to ∼5.3 × 10–14 mol/capsule. Figure S28, curve a depicts the fluorescence change upon subjecting the GOx and catalase/insulin bicompartment microcapsule system to the Amplex-Red/Resorufin assay. No fluorescence change is observed implying that the coadded catalase degraded any harmful peroxide. Figure S28, curve b shows the time-dependent fluorescent change upon analyzing the glucose-generated H2O2, using the Amplex-Red oxidation to the fluorescent Resorufin as assay, and applying the GOx-insulin bireservoir microcapsule-in-microcapsule system (in the absence of catalase).

We realize that previous efforts were directed to develop an “artificial pancreas” system[88,90,91] and, thus, the evaluation of the present system in comparison to previous art is important. Accordingly, we selected to compare our system to a couple of relevant reports on the design of “artificial pancreas” systems, see Table S2 and accompanying discussion.

Conclusions

The study has introduced a versatile method to assemble stimuli-responsive hydrogel-based microcapsule-in-microcapsule systems. The resulting microcapsule systems included two aqueous reservoirs, loaded with different loads, separated by stimuli-responsive hydrogel layers. We have demonstrated that the microcapsule-in-microcapsule systems can be based on nucleic acid-functionalized CMC or nucleic acid-modified polyacrylamide scaffolds as hydrogel-building scaffolds. The nucleic acids embedded in the hydrogel layer provide cross-linking duplex units and, most important, stimuli-responsive nucleic acids that upon reconfiguration control the stiffness of the hydrogel layers and their permeability toward loads. Specifically, Zn2+-ions were used as irreversible triggers to unlock one of the hydrogel layers, and pH or K+-ions/crown ether triggers were successfully applied to reversibly control the stiffness/permeability of the hydrogel through the switchable reconfiguration of i-motif or G-quadruplex structures, respectively.

These results suggest that other cofactor-dependent DNAzymes or irreversible light-induced cleavage of DNA duplex nucleic acids could lead to irreversible unlocking of the hydrogel layers. In addition, the reversible control over the stiffness of the hydrogel by means of metal-ions/ligands (e.g., T-Hg2+-T bridges/cysteine) or light (cis/trans isomerization of azobenzene intercalators) could be used to control the reversible permeability of the hydrogel layers.

We found that the specific hydrogel layers used in the present study revealed a substrate permeability cutoff ≤ 5 kDa, lower molecular weight substrates will permeate through the hydrogel layers. Substrates of higher molecular weight were nonpermeable at in the higher stiffness state of the hydrogel (no leakage phenomena were detected on a time-scale of 10 days). For a detailed evaluation of the molecular-weight cutoff properties of the bilayer microcapsule system, see Figure S29, Table S3, and accompanying discussion. These features of the systems are dictated by the degree of cross-linking of the hydrogel by the respective nucleic acid bridges. Tuning of the loading and the recovery of nucleic acid bridges could further control the permeability of the hydrogels. Finally, to adapt the controlled release of the loads to the triggered stiffness properties of the hydrogels, we applied luminescent quantum dots and drug-functionalized polymers as loads. One may, however, apply other nanoparticles or macromolecular nanostructures as loads. Besides the controlled release of the loads associated with the two aqueous reservoirs in the presence of the appropriate triggers, an interesting enzyme-stimulated release of a drug (insulin) from the bicompartmentalized microcapsule was demonstrated, thereby providing a microencapsulated system acting as an “artificial pancreas”. That is, the enzyme glucose oxidase was encapsulated in the inner aqueous compartment of the microcapsules-in-microcapsules, and insulin was loaded in the outer aqueous compartment. The freely permeating glucose led to the GOx-catalyzed aerobic oxidation of glucose to gluconic acid. The acidification of the microcapsule-in-microcapsule microenvironment and the accompanying release of insulin were then guided by the concentration of glucose. That is, an autonomous microcarrier for the release of insulin in response to up-regulated contents of glucose is demonstrated under conditions expected to prevent hypoglycemia. In principle, other enzymes altering the pH of the microcarriers may be loaded in the microcapsules for releasing other therapeutics. For example, acetylcholine esterase could be loaded in the microcapsules. Changes driven upon the hydrolysis of acetylcholine, upon overactivation or inhibition of the enzyme may, then, release drugs that perturb the neural system. Furthermore, the triggered intercommunication (mixing) of the two aqueous compartments might lead to signal-dictated reactions in microreactors. The structural diversity stimuli-responsive microcapsule-in-microcapsule systems and the variability of loads in the bicompartmentalized microcapsule systems provide different versatile applications of such carriers.