Characterization of Polyethylene Glycol–Reinforced Alginate Microcapsules for Mechanically Stable Cell Immunoisolation

Abstract

Islet transplantation within mechanically stable microcapsules offers the promise of long-term diabetes reversal without chronic immunosuppression. Reinforcing the ionically gelled network of alginate (ALG) hydrogels with covalently linked polyethylene glycol (PEG) may create hybrid structures with desirable mechanical properties. This report describes the fabrication of hybrid PEG-ALG interpenetrating polymer networks and the investigation of microcapsule swelling, surface modulus, rheology, compression, and permeability. It is demonstrated that hybrid networks are more resistant to bulk swelling and compressive deformation and display improved shape recovery and long-term resilience. Interestingly, it is shown that PEG-ALG networks behave like ALG during microscale surface deformation and small amplitude shear while exhibiting similar permeability properties. The results from this report’s in vitro characterization are interpreted according to viscoelastic polymer theory and provide new insight into hybrid hydrogel mechanical behavior. This new understanding of PEG-ALG mechanical performance is then linked to previous work that demonstrated the success of hybrid polymer immunoisolation devices in vivo.

1. Introduction

Type-1 diabetes (T1D), is a debilitating disease whereby the body’s immune system targets and eliminates pancreatic β cells, which produce and secrete insulin to regulate blood glucose.^[1,2] The transplantation of cadaveric islets presents a promising biological therapy for T1D, but islet grafts require chronic immunosuppression to prevent rejection.^[3,4] Unfortunately this systemic immunosuppression is toxic to islets and greatly increases patient susceptibility to infection and cancer.^[5,6] Islet encapsulation may allow for trans-plantation without chronic immunosuppression.^[7–10] The capsule would act as a physical barrier to the immune system, preventing islet destruction. Conversely, the capsule would be permeable to crucial nutrients and metabolic products like insulin, thereby enabling normal islet functionality and T1D reversal.^[11] Historically, encapsulated islet transplantation demonstrated a lack of long-term success in large animals and humans, which may have been partially due to capsule mechanical failure.^[12–16] Many capsules previously evaluated in clinical settings were composed of the natural hydrogel alginate (ALG).^[17–19] Though ALG was chosen for its high biocompatibility, several reports indicated that mechanical instability could constrain its usefulness in transplantation.^[20–24] Theoretically, mechanical instability of an immunoisolation device could lead to islet immune expo-sure and ultimately to destruction of the islet graft.^[25] Thus a variety of reports have focused on modifying and improving ALG-based immunoisolation systems, leading to long-term immunoprotection and stability in vivo.^[26–33] One potential strategy involves the reinforcement of ionically gelled ALG with polyethylene glycol (PEG), yielding an interpenetrating polymer network.^[15,34–36] Unlike ALG, PEG is covalently crosslinked and resistant to osmotic pressure.^[37–41] We previously demonstrated that islet encapsulation in PEG-reinforced ALG hydrogels display superior function in reversing diabetes long-term after implantation in the peritoneal cavity of fully MHC-mismatched mice.^[36] To explain this improvement we hypothesized that PEG-reinforced ALG-based capsules would display superior mechanical properties and overall stability compared to ALG-only capsules. Here, three different capsule formulations were studied: ALG-only, PEG-only, and PEG-ALG hybrids. For PEG-containing gels, two different crosslinkers with high and low molecular weight were assessed to modulate PEG gel crosslinking density. After fabrication, osmotic stress testing, atomic force microscopy (AFM), and diffusion assays were used to determine hybrid capsule swelling properties, surface elastic moduli, and permeability. Parallel-plate rheology and uniaxial compression were used for evaluation of hybrid gel responses to shear strain and compressive strain. The design, fabrication, and testing of the materials is summarized in Figure 1.

Figure 1.

Design, fabrication, and testing of PEG-ALG hybrid materials. A) Schematic representation of ionically-gelled Alginate capsules, chemically-gelled PEG capsules, and hybrid MicroMix capsules composed of PEG-ALG interpenetrating polymer networks. B) Schematic of electrostatic droplet generation for ionic Alginate capsules, hand-pipetting with pre-mixed crosslinker for PEG capsules, and two-step electrostatic droplet generation and covalent crosslinker addition for hybrid MicroMix capsules. C) Overview of the study’s testing scheme featuring bulk osmotic pressure swelling, local AFM indentation, rheological shear performance, uniaxial compression performance, and capsule diffusion testing.

2. Experimental Section

2.1. Material Fabrication

2.1.1. Fabrication of Alginate Microcapsules

Fabrication of 800 μm-diameter 1.2% MVG (Pronova) alginate (ALG) capsules was performed with an electrostatic droplet generator (Nisco) and 50 mM CaCl₂ incubation as previously described.^[36] The alginate was dissolved in Hank’s balanced salt solution without Ca²⁺ or Mg²⁺ (HBSS⁻⁻) (Invitrogen). The ALG solution was then extruded from a 0.4 mm-diameter needle and an 8.8 kV voltage potential was applied. A 50 mM CaCl₂ solution osmotically balanced to 300 mOsm L⁻¹ was placed downstream of the droplet generator to provide the divalent ions needed for ALG gelation.

2.1.2. Fabrication of Polyethylene Glycol Microcapsules

Fabrication of 800 μm-diameter 5% polyethylene glycol (PEG-maleimide 10 kDa 8 arms, Jenkem) capsules was performed by adapting a previously described method (Figure 1).^[42] Briefly, 0.4 μL of PEG (pre-mixed with dithiothreitol (DTT, Sigma) or PEG-dithiol 2 kDa (PEGSH, Jenkem) at pH 4–5) was hand-pipetted into polypropylene glycol (Sigma) containing 0.02% triethanolamine (Sigma) and 10% Span-80 (Sigma). The process was designed to promote PEG-MAL crosslinking and to shape the capsules into spheres of comparable size and geometry.

2.1.3. Fabrication of PEG-ALG Interpenetrating Polymer Network (MicroMix) Microcapsules

Fabrication of 800 μm-diameter 5% PEG (PEG-maleimide 10 kDa 8 arms, Jenkem)—1.2% MVG (Pronova) ALG hybrid MicroMix capsules was also performed with an electrostatic droplet generator and 50 mM CaCl₂ incubation as previously described.^[36] PEG crosslinking was achieved by further incubating capsules with 4:1 molar ratio of either DTT or PEGSH solutions for 1 min. MicroMix capsules were then washed three times in Hank’s balanced salt solution containing Ca²⁺ and Mg²⁺ (HBSS⁺⁺) (Invitrogen).

2.2. Material Characterization

2.2.1. Anti-PEG Staining

Capsules were blocked and permeabilized in HBSS containing Ca²⁺ and Mg²⁺ (HBSS⁺⁺) and supplemented with 2% bovine serum albumin (BSA, Sigma), 10% chicken serum (Sigma) and 0.2% Triton X-100 (Sigma). Then, the sample was incubated in anti-PEG biotinylated antibody solution (Abcam, 1:500) in 2% BSA supplemented HBSS⁺⁺, washed with HBSS⁺⁺, and then incubated in Alexa Fluor 488-conjugated streptavidin (Life Technologies, 1:200) solution in 2% BSA in HBSS. Samples were imaged with a SP5 Leica inverted confocal microscope and processed using Leica Application Suite or ImageJ.

2.2.2. Bulk Mechanical Analysis of Microcapsules by Osmotic Pressure Test

Mechanical evaluation of capsules was performed by osmotic pressure testing. In the osmotic test, capsules (n > 30) were subjected to sequential exposure to ddH₂O (2 h, 0 mOsm kg⁻¹) and HBSS++ (0.2–120 h, 270–305 mOsm kg⁻¹). Capsule diameter was measured before testing (HBSS), after 2 h of osmotic stress (ddH₂O), and at 0.2, 24, 72, and 120 h after stress removal (HBSS). Graphs were produced from data that had been normalized to baseline (pre-stress) capsule diameters. Capsule resistance to swelling was defined by the maximum deformation attained at the end of the osmotic stress period. Capsule elasticity or plasticity was defined by the diameter recovery achieved after the stress was removed. Osmotic pressure testing was performed on capsules at two specific time points: 2 and 7 days after fabrication. The 2 day test provides information on capsule stability and performance after gelation and when PEG crosslinking was assumed to be completed. The 7-day test provided information on long-term stability and performance of the capsules.

2.2.3. Analysis of Single Microcapsules by Atomic Force Microscopy

Capsule immobilization on 35 mm petri dishes for AFM measurements was performed by spreading 50 μL of 0.5% (wt) agarose solution in HBSS++ on preheated 35 × 10 mm tissue culture plates (VWR), producing an approximately 50 micron thick liquid agarose layer to trap capsules. Then, approx. 3–5 beads were withdrawn and allowed to settle at the tip of a 200 μL micropipette, before being expelled into the center of the agarose-coated dish. The dish was cooled for 1 h at room temperature, followed by 1 h at 4°C to complete agarose gelation, leading to capsule immobilization on the dish before AFM characterization.

Elasticity characterization was performed using a custom-built AFM system.^[43,44] A spherical, gold-coated, silicon nitride AFM cantilever with a borosilicate glass particle tip (2 μm diameter, 0.1 N m⁻¹ spring constant Novascan) was used to probe the capsules. The contact interaction between the cantilever and the capsules was recorded using a custom code based in Igor Pro (Wavemetrics). Recorded cantilever deflection-indentation curves were used to derive force-indentation curves, after factoring out the cantilever deflection on a hard surface and incorporating the spring constant of the cantilever. According to the Hertz contact mechanical model, the force-indentation relationship is a function of Young’s modulus of elasticity:

$$F=\frac{4E\sqrt{R}}{3\left(1-v^2\right)}{D}^{3/2}$$ (1)

where F is the measured force, E is Young’s modulus, v is Poisson’s ratio (assumed to be 0.49), R is the radius of the spherical indenter, and D is the measured indentation.^[45] Ten measurements were taken per capsule in the same location. Measurements were taken at two different time points for each type of capsule: 1 day and 7 days after capsule fabrication, both at room temperature. Measurements were taken in HBSS⁺⁺ to prevent capsule dehydration. A custom curve fitting MATLAB program was used to analyze the data. Data obtained was analyzed for statistical significance with a t-test.

2.2.4. Rheology of Macroscale Hydrogel Disks

Alginate disks (h = 1.0 mm, d = 35.0 mm) were formed by adding 1.0 mL of 1.2% MVG into transwells (Corning) and incubating in 50 mM CaCl₂ for 1 hr. Disks were removed manually with a spatula and washed with HBSS⁺⁺ before storing in a well with fresh HBSS⁺⁺. MicroMix disks were formed according to the same procedure with an additional step for PEG-crosslinking (addition of DTT or PEGSH crosslinker, incubation for 5 min prior to HBSS++ washing). PEG disks were formed by pre-mixing PEG with crosslinker (DDT, PEGSH) and then pipetting into transwells for gelation. Disks were incubated in HBSS⁺⁺ for 2 days or for 7 days before removal and mechanical testing via rheometry. A Haake Rheostress6000 Rheometer or a TA Discovery HT-1 with parallel-plate geometry was used. Frequency sweeps from 0.1 to 100 Hz with an applied strain of 1% were performed on hydrogel disks. G′ and G″ were calculated from experimental curves.

2.2.5. Compression of Macroscale Hydrogel Disks

Alginate, PEG, and MicroMix disks were formed according to the procedure outlined above for rheology. For consistency, all disks were composed of equal volumes of precursor gel. Disks were placed between the parallel plates of a Haake Rheostress6000 Rheometer and the upper plate was dropped to a pre-defined height of 1.1 mm for all disk compositions analyzed. The instantaneously resulting normal force (Fn) was recorded and the Fn was subsequently measured every minute for 5 min in order to measure hydrogel relaxation.

2.2.6. Microcapsule Permeability

During capsule fabrication, capsules were loaded with 4 or 10 kDa FITC-Dextran (Sigma) and subsequently incubated overnight in an HBSS⁺⁺ solution with the same FITC-Dextran concentration. Capsules were then removed from the FITC-HBSS⁺⁺ solutions and placed into wells with 1 ml fresh HBSS⁺⁺. Three wells with eight capsules per well were tested for each of the different capsule compositions. Samples were taken at 17 time points (1, 3, 5, 8, 10, 15, 20, 25, 30, 45, 60, 90, 120, 150, 180, 240 min and 24 h). Diffusion coefficients were obtained using a best fitting between the experimental solute concentration profiles and the analytical solution of the equation describing solute diffusion out of spheres or a plane sheet into initially solute-free liquid of a constant volume.^[46]

3. Results and Discussion

3.1. Microcapsule Fabrication and Experimental Strategy

We evaluated the properties of physically-gelled alginate (ALG), covalently-gelled polyethylene glycol (PEG), and two-step interpenetrating polymer networks composed of both ALG and PEG (MicroMix) (Figure 1A). ALG and MicroMix capsules were fabricated via electrostatic droplet generation, which involved pump driven extrusion of precursor solution into a CaCl₂ gelation bath, with the additional step of PEG crosslinking for the MicroMix capsules. (Figure 1B).^[36] PEG capsules were fabricated by briefly pre-mixing solution and crosslinker and then immediately micropipetting into a polypropylene glycol bath (Figure 1B).^[36] The PEG networks were crosslinked with either dithiothreitol (DTT) or PEG-dithiol 2 kDa (PEGSH), yielding MicroMix-DTT (MM-DTT), MicroMix-PEGSH (MM-PEGSH), PEG-DTT, and PEG-PEGSH. In this in vitro study we analyzed the bulk swelling properties (osmotic stress test) and local surface modulus (atomic force microscopy) of capsules (Figure 1C). The shear behavior (rheology) and compressive behavior (stress relaxation after uniaxial compression) of the materials was also documented (Figure 1C). Lastly the capsule permeability was assessed (diffusion assay). (Figure 1C). The in vivo performance of these capsule compositions in islet transplantation models in mice was previously reported.^[36] Prior to testing, the presence of PEG in hybrid MicroMix networks was confirmed through anti-PEG staining and confocal microscopy (Figure 1A,B, Supporting Information).

3.2. Bulk Osmotic Stress Resistance and Microcapsule Swelling Properties

3.2.1. Experimental Design

We first assessed the capsule swelling properties via osmotic pressure test, where the manipulation of solution concentrations produced external pressures that drove sample volume change. Such a test probes bulk capsule resistance to deformation and allows for distinction between elastic recovery and plastic deformation. Volumetric capsule stability is highly relevant to encapsulated islet graft success because uncontrolled swelling and irreversible deformation lead to a loss of islet immunoprotection. We performed analysis 2 days after capsule fabrication because that is the approximate time-point when they would be implanted in clinical settings (Figure 2A–D). Tests were also performed at 7 days post-fabrication to ensure there was not a loss of capsule mechanical integrity (Figure 2E–H).

Figure 2.

Bulk volumetric deformation and recovery of microcapsules. A,E) Full capsule deformation profile (normalized to baseline diameter) over time (t = 0 h to t = 120 h) for samples tested 2 and 7 days post-fabrication. The red box represents the period of osmotic stress application. B,F) Maximum capsule deformation (deformed diameter normalized to baseline diameter) measured at the end of the constant applied osmotic stress phase (2 h) for samples tested 2 and 7 days post-fabrication. C,G) Short-term capsule recovery (immediately recovered diameter normalized to baseline diameter) measured approximately 10–12 min after the removal of the constant stress, for samples tested 2 and 7 days post-fabrication. D,H) Long-term capsule recovery (final recovered diameter normalized to baseline diameter) measured approximately 120 h after the removal of the constant stress, for samples tested 2 and 7 days post-fabrication. I) Long-term osmotic stability test showing capsule diameter trends (normalized to baseline) over time (t = 0 to t = 50 days) for all compositions analyzed. J) Final capsule diameter (normalized to baseline) after long-term stability incubation period (50 days). K) Burgers viscoelastic model combining a Kelvin-Voigt solid and a Maxwell fluid in series. The dashpots represent viscous Newtonian fluid elements and the springs represent Hookean elastic solid elements. Here this model allows for a qualitative interpretation of the viscoelastic behavior we observed in volumetric deformation and recovery. MM: MicroMix.

3.2.2. Microcapsule Resistance to Volumetric Deformation

Capsule diameters were normalized to the pre-stress dimensions in order to accurately compare sample deformations across the various capsules tested. The full creep and recovery curves illustrated a large capsule deformation in the osmotic stress phase (red box) followed by both short-term and long-term recovery of capsule diameter following osmotic stress removal (Figure 2A,E). We measured the maximum capsule diameter at the end of the applied constant stress phase to understand total capsule resistance to volumetric deformation (Figure 2B,F). At 2 days post-fabrication the ALG-only capsules failed to resist osmotic pressure, as demonstrated by 136% strain after 2 h of continuous osmotic stress. As expected, both of the PEG-only capsule formulations resisted osmotic pressure and exhibited significantly less swelling after continuous stress (20–30% strain). The PEG/ALG networks (MicroMix: MM) displayed hybrid properties between these extrema, with both MM-DTT (109% strain) and MM-PEGSH (100% strain) showing significantly less capsule deformation than the unreinforced ALG-only networks (p < 0.0001; p < 0.0001) (Figure 2B). The same trend was observed at 7 days post-fabrication, with the MM-DTT and MM-PEGSH capsules swelling significantly less than the ALG capsules under the same applied osmotic stress (p = 0.0129; p < 0.0001) (Figure 2F). Thus, PEG-reinforcement of ALG leads to significantly improved bulk resistance to volumetric deformation.

3.2.3. Microcapsule Volume Recovery after Swelling Event

After the continuous stress phase, the osmotic pressure was removed. Measurement of the capsule diameter 10 min post-stress allowed us to identify the immediate elastic contribution to capsule recovery (Figure 2C,G). The 2-day ALG samples only recovered to 80% strain after stress removal, indicating significant viscous retardation of sample recovery. On the other hand, the PEG samples recovered to 2–3% strain immediately after stress removal, demonstrating an almost completely elastic recovery. Again, both MicroMix (MM) samples displayed hybrid properties (42–46% strain), with a significantly greater capsule recovery response than the ALG-only capsules (p < 0.0001; p < 0.0001) (Figure 2C). A similar pattern was observed for the 7-day samples, with the MM-PEGSH showing significantly greater instant capsule recovery than the ALG samples (p < 0.0001) (Figure 2G). Therefore, the addition of a covalent PEG network to an ionic ALG capsule results in significantly increased elastic recovery following bulk deformation.

We continued to monitor the capsule recovery over time to describe the complete viscoelastic response. The last time point, 5-days (120 h) after stress removal, was chosen to represent the “long-term” sample recovery (Figure 2D,H). The 2-day ALG samples continued to slowly recover to 28% strain, demonstrating a viscously dampened elastic response compared to the instantaneous elastic response of the previous figure. The non-zero final deformation value indicates permanent viscous flow. The PEG capsules remained steady at 2–3% strain, thus illustrating a highly elastic response with a negligible degree of permanent deformation. Like ALG, the MicroMix (MM) samples recovered additional volume through a dampened long-term elastic response. Though the MicroMix displayed some degree of plastic deformation, both MicroMix capsule formulations recovered to a greater degree (MM-DTT: 19% strain; MM-PEGSH: 23% strain) than the ALG-only capsules (p = 0.5629; p = 0.0076). As before, the same trend held for the samples tested at 7 days post-fabrication. Therefore, PEG-reinforcement allows for a more complete elastic recovery of initial sample volume, compared to ALG-only capsules.

3.2.4. Theoretical Interpretation of Experimental Bulk Microcapsule Properties

We interpreted these results using the Burgers model (Figure 2K). This viscoelastic model contains a Kelvin-Voigt solid (Hookean spring and Newtonian dashpot in parallel) with a Maxwell fluid (Hookean spring and Newtonian dashpot in series).^[47–50] When a stress is applied, the spring elements elastically resist linear strain, while the dashpot elements viscously retard the strain rate. During the continuous stress phase, the elastic portion of the network reversibly deforms and stores energy, while the viscous portion irreversibly dissipates energy through rearrangement. During the recovery phase, the dashpot components remain deformed or dampen the elastic recovery as the spring components return to their initial position. From our data, the PEG-only samples behaved almost completely elastically. There was no significant retardation of the recovery and no significant permanent network deformation, indicating little to no contribution from the dissipative dashpots. Thus, the covalently linked PEG chains deformed like elastic springs and were able to rapidly return to their original shape following stress removal. On the other hand, the ALG samples showed a highly dampened recovery response and substantial permanent deformation, indicating contribution from both the Kelvin dashpot (delayed recovery) and Maxwell dashpot (irreversible deformation). Therefore, the ionically gelled ALG network behaves like a viscoelastic material with a significant contribution from the viscous chain rearrangement. However, the hybrid MicroMix samples displayed increased resistance to stress, greater instantaneous recovery, and a lesser degree of permanent deformation compared to the ALG-only samples. Thus, the incorporation of a covalent PEG network in an ionic ALG network produced more elastic behavior, thereby allowing for superior resistance and recovery from destructive forces in vivo.^[36]

Interestingly, all of the ALG-based capsules displayed improved mechanical properties over time. ALG-only, MM-DTT, and MM-PEGSH showed significantly improved total resistance to stress, instantaneous recovery, and long-term recovery when the 7-day capsules were compared to the 2-day capsules (Figure 2A–C, Supporting Information). This increased elasticity has been reported before and can be due to spontaneous thermodynamically favored ALG chain rearrangements or the sequestration of additional calcium.^[51–54] Alginate tends to not only increase junctional density, but also develop ordered fibrillar structures over time.^[55,56] These spontaneous semicrystalline structures could also explain the increase in opacity observed during long-term quiescent incubations, as the boundaries between ordered microstructures and amorphous polymer can increase light scattering.^[57,58] Additionally, the PEGSH crosslinked capsules generally showed improved mechanical properties compared to the DTT-crosslinked capsules (Figure 3A–C, Supporting Information). This could be explained by the crosslinker size difference, since the PEGSH (2 kDa) is almost 13 times larger than the DTT (153 Da). There is evidence that an increase in crosslinker size can produce stronger mechanical properties in other crosslinked polymer systems.^[59]

3.2.5. Characterization of Long-Term Stability of Microcapsules

Quantification of capsule diameter trend (Figure 2I) and final capsule deformation (Figure 2J) during long-term (50 days) incubation in osmotically balanced solutions showed that ALG capsules had a tendency of shrinking their diameter (average ratio of final to initial diameter: 0.875), while PEG capsules displayed mild swelling (average ratio of final to initial diameter: DTT, 1.061; PEGSH, 1.045). The hybrid MicroMix (MM) displayed very slight shrinking compared to ALG capsules (average ratio of final to initial diameter: MM-DTT, 0.981, p = 0.0011; MM-PEGSH, 0.958, p = 0.0240). The spontaneous syneresis of ALG under quiescent conditions has been described before, and it appears the incorporation of a PEG-network provides a counter-balance to abrogate the ALG shape change.^[51,60–62]

3.3. Elastic Moduli of Single Microcapsules via Atomic Force Microscopy

We analyzed the local mechanical properties of individual capsules by AFM, which describes the gel response to micro-indentation and allows calculation of the surface compressive elastic modulus (E′) of each single capsule. Capsules were successfully immobilized and the exposed capsule surface was microscopically probed with the AFM cantilever to determine the local E′. (Figure 3A). AFM data showed that the E′ of PEG capsules was higher (DTT 66.73 kPa; PEGSH 64.58 kPa) than both the MicroMix (MM) and ALG-only capsules at 1 day post-fabrication (Figure 3B). However, the local moduli of the MicroMix capsules (MM-DTT 18.44 kPa; MM-PEGSH 15.44 kPa) and the ALG capsules (17.17 kPa) were nearly identical (Figure 3B). When measurements were performed 7 days after capsule fabrication and compared to day 1 measurements, there was a significant decrease in the E′ of PEG capsules (DTT 40.25 kPa; PEGSH 41.06 kPa) (Figure 2I, Supporting Information), a slight increase in the ALG E′ (24.12 kPa) (Figure 2G, Supporting Information), and negligible change in the MicroMix E′ (MM-DTT 16.54 kPa; MM-PEGSH 17.29 kPa) (Figure 2H, Supporting Information). Again, there was not a significant difference between the ALG-only and MicroMix capsules (Figure 3C). Taken together, these data suggest that although PEG-only networks display a much greater surface elastic modulus than ALG-only, PEG-reinforcement of an ALG network does not produce an increase in the hybrid capsule’s local compressive elastic modulus. Compared to the substantial improvements in bulk mechanical properties observed in the whole-capsule swelling test (Figure 2), PEG incorporation in the interpenetrating polymer network does not appear to affect microscale capsule surface compressive elastic properties (Figure 3).

Figure 3.

Surface Elasticity of Microcapsules. A) Image of the AFM probe making a microscale indentation on the capsule surface. B) Local elastic moduli of ALG, MicroMix and PEG samples, 1 day after fabrication, measured via AFM. C) Local elastic moduli of ALG, MicroMix and PEG samples, 7 days after fabrication, measured via AFM.

3.4. Rheological Shear Response of Macroscale Hydrogel Disks

The hydrogel responses to shear stress and compression were analyzed with a parallel plate rheometer. These hydrogel mechanical properties are important because capsules in vivo can be exposed to tangential shear forces and destructive compressive forces due to the implantation procedure and/or the transplant site environment. Our microscale capsule fabrication protocols were adapted to create macroscale hydrogel disks of constant pre-determined diameter and volume. After performance of an amplitude sweep to determine the linear viscoelastic region, we performed frequency sweeps on the gel disks to analyze the response to increasing rates of small amplitude oscillatory shear strain. Dynamic rheology allows for measurement of the gel storage modulus (G′) and loss modulus (G″), which are measures of solid-like elastic energy storage and liquid-like viscous energy loss.^[63,64] The ALG and MicroMix (MM) hydrogel samples displayed nearly identical responses to shear stress at both 2 and 7 days post-fabrication (Figure 4A,D). Specifically, at lower frequencies there was a frequency-independent linear plateau region where the storage modulus was much greater than the loss modulus, indicating the presence of a stable crosslinked network where elastic solid behavior dominated over viscous flow behavior.^[65,66] Then, after the rubbery plateau there was a glass-like increase in storage modulus as the frequency of applied shear strain surpassed the frequency of chain rearrangement.^{[15,64,67,68]} Compared to the ALG-based gels, the PEG-only gels displayed a frequency-independent plateau the entire time (Figure 4A,D). Similar to the AFM results on single capsules (Figure 3), there was no significant difference between the elastic storage modulus of the ALG and MicroMix samples, both at 2 days (p = 0.1447; p = 0.3615) and 7 days (p = 0.9991; p = 0.8524) post-fabrication (Figure 4B,E). Interestingly, the PEG-PEGSH sample exhibited a significantly greater storage modulus than the PEG-DTT (Figure 3F, Supporting Information). As seen in the initial rheological curves, there was a significant difference between the low-frequency shear behavior and the high-frequency shear behavior for both ALG and MicroMix samples both at 2 and 7 days post-fabrication (p < 0.0001 for all samples) (Figure 4C,F). Conversely, there was no significant change between the low and high shear behavior for the PEG-only samples. Such a frequency-dependent response for the ALG-based samples could be explained by the viscoelastic model proposed earlier (Figure 2K), where the viscous dashpot elements highly prevalent in the ionic ALG networks were not capable of performing chain rearrangement under the rapidly applied deformations.^[64,68] This response contrasts with that of the spring-like covalent PEG samples, which exhibit linear elastic behavior over three decades of applied shear frequency. Because of the nearly identical rheological behavior exhibited by both the ALG and MicroMix samples, the ALG network provides the dominant contribution to the overall capsule response to tangential shear forces. Thus PEG-reinforcement does not lead to a noticeable change in capsule shear behavior.

Figure 4.

Rheological Analysis of Macroscale Hydrogels. A,D) Complete small amplitude oscillatory shear (SAOS, strain = 1%) rheological output over three decades of applied shear frequency (f = 0.1 to f = 100) for all compositions analyzed, at 2 and 7 days post-fabrication, respectively. Testing was performed on macroscale hydrogel disks due to the geometric constraints imposed by the measuring device that precluded the use of microspheres. Frequency sweep outputs (G′ = storage modulus, G″ = loss modulus). B,E) Elastic storage moduli (G′) at 0.1 Hz of applied frequency for all compositions analyzed at 2 and 7 days post-fabrication, respectively. C,F) Comparison of low frequency (0.1 Hz) and high frequency (100 Hz) shear response for all compositions analyzed at 2 and 7 days post-fabrication, respectively.

3.5. Uniaxial Compression Response of Macroscale Hydrogel Disks

We also investigated the hydrogel behavior during vertical compression for gels analyzed 2 and 7 days post-fabrication. Disks of equal diameter and pre-gelation volume were placed under a constant compressive strain and the gel response (normal force decay) was measured over time (Figure 5A,D). The instantaneous stress response was significantly greater for the 2-day MicroMix samples (Fn (t = 0): MM-DTT, 6.61 N; MM-PEGSH 7.23 N) than the 2-day ALG samples (Fn (t = 0): 3.05 N), indicating improved gel resistance to compression (p < 0.0001; p < 0.0001) (Figure 5B). Following the instantaneous response, each of the samples exhibited a viscoelastic stress relaxation. MicroMix samples (Fn (t = 5 min): MM-DTT, 0.85 N; MM-PEGSH 1.34 N) displayed equilibrium stress plateaus that were much higher than ALG samples (Fn (t = 5 min): 0.36 N), demonstrating increased elastic resistance to permanent compressive deformation (Figure 5C). The same trend was observed for the 7-day samples, with MicroMix displaying both significantly enhanced immediate resistance to compression (p < 0.0001; p < 0.0001) and significantly greater equilibrium resistance to compression (p = 0.262; p = 0.0001) (Figure 5E,F). A comparison of hydrogels fabricated 2 days versus 7 days before characterization revealed a non-statistically significant increase in the equilibrium gel stress under strain (Figure 2E, Supporting Information). A comparison of crosslinkers revealed that the PEGSH produced networks with superior immediate resistance and equilibrium resistance compared to DTT for the 7-day MicroMix samples (Figure 3C,D, Supporting Information). For a more complete viscoelastic interpretation, we present the Standard Linear Solid model with a Maxwell arm (Figure 5G).^[48–50] At t = 0 there is no time for viscous chain rearrangement, and therefore the dashpot element does not contribute. Thus, the immediate measured output is the combined elastic spring and Maxwell spring resistance to a sudden network strain. However, as time goes on the viscous chains can rearrange and undergo plastic deformation, thereby relaxing the overall stress of the strained polymer network.^[69–72] In the case of the MicroMix samples, the covalent PEG-reinforcement confers increased spring-like elasticity to the network, both in the short and long term, despite the viscous dashpot deformation of the ionic ALG chains.

Figure 5.

Compressive Stress Relaxation of Macroscale Hydrogels. A,D) Complete stress relaxation response (measured in normal force Fn, from t = 0 min to t = 5 min) under constant applied axial strain for all samples analyzed at both 2 and 7 days post-fabrication, respectively. B,E) Instantaneous gel stress response (Fn at t = 0) after application of a constant uniaxial compressive strain for all samples analyzed at both 2 and 7 days post-fabrication, respectively. C,F) Equilibrium gel stress response (Fn at t = 5) after application of a constant uniaxial compressive strain for all samples analyzed at both 2 and 7 days post-fabrication, respectively. G) Standard linear solid viscoelastic model combining a Maxwell fluid and a Hookean spring in parallel. The dashpots represent viscous Newtonian fluid elements and the springs represent Hookean elastic solid elements. Here this model allows for a qualitative interpretation of the viscoelastic behavior we observed in uniaxial stress relaxation.

3.6. Characterization of Microcapsule Permeability Properties

Following mechanical analysis, we investigated the effects of gel composition on capsule transport properties, which are critical for immunoisolated cell survival and functionality. For use in islet transplantation, the PEG-reinforced capsules must allow unhindered passage of crucial cellular products like insulin. To study permeability, the capsules were loaded with 4 or 10 kDa FITC-dextran and transferred to a new bath. The supernatant was measured over time to quantify relative diffusion and diffusion coefficients for FITC-dextran through the capsules. All samples demonstrated a rapid release of their cargo, with a faster and greater release of 4 kDa FITC than 10 kDa FITC, as expected (Figure 6). The PEG release curves (Figure 6D,E) showed the lowest diffusion equilibrium plateaus, while the MicroMix samples (Figure 6B,C) displayed 4 kDa diffusion curves very similar to ALG (Figure 6A), and slightly lower equilibrium plateaus than ALG for the 10 kDa FITC. Experimental data was fitted into our model to extrapolate diffusion coefficients, which revealed similar diffusivity values for all samples analyzed (Figure 6F). Thus, PEG-reinforcement of an ALG network did not significantly alter the capsule permeability properties, which explains the good in vivo and in vitro islet viability and function we previously reported.^[36]

Figure 6.

Microcapsule Permeability Analysis and Extrapolated Diffusion Coefficients. A–E) 4 and 10 kDa FITC-Dextran release curves for the ALG, MM-DTT, MM-PEGSH, PEG-DTT, and PEG-PEGSH capsules, respectively. FITC diffusion into solution was sampled periodically over time from t = 0 h to t = 24 h. F) Theoretical diffusion coefficients for the materials were extrapolated from the experimental data.

4. Conclusions

We here presented an analysis of the bulk volumetric swelling, microscale surface elasticity, tangential shear behavior, axial compressive response, and solute transport properties of hydrogels used for immunoisolation of insulin-secreting cells: PEG, ALG, and PEG-reinforced ALG-based “MicroMix.” Compared to ALG-only capsules, the hybrid MicroMix displayed improved resistance to bulk osmotic stress, spontaneous shape deformation, and axial compressive stress. Additionally, the MicroMix displayed more rapid and complete shape recovery and improved equilibrium resistance to compressive deformation. This clear improvement in bulk and axial mechanical properties could explain our previously reported in vivo results, where the stable MicroMix capsules improved islet engraftment in mechanically stressful implant sites compared to ALG-only.^[36] Conversely, the MicroMix and the ALG capsules and hydrogels exhibited very similar microscale surface elasticity and bulk tangential shear response. From our previous in vivo work, we observed fibrotic capsule formation on PEG-coated ALG capsules, whereas the MicroMix and ALG-only capsules showed a high degree of biocompatibility.^[36] Cellular response to a foreign implant is highly dependent on implant mechanics, and thus the similar surface and shear properties of the interpenetrating MicroMix capsules may confer alginate-like biocompatibility to the hybrid networks.^[73–75] Lastly PEG-reinforcement doesn’t significantly affect the capsule transport properties, which aligns with our previous studies that demonstrated a high degree of encapsulated islet viability and function.^[36]

Mechanical instability of previous capsule systems was well-established and presented a sizable hurdle for the field of encapsulated cell transplantation.^{[20,22,76,77]} Our study results substantiated those observed in recent gel reinforcement studies, particularly the noticeable change in mechanics following the incorporation of an interpenetrating polymer network.^[15,34,37] With this work we offer a more detailed interpretation of the various viscoelastic consequences of covalent network addition to an ionically gelled polysaccharide. Relevant material parameters (bulk, surface, shear, compressive, diffusive) have been linked to previously observed in vivo behaviors (capsule stability, biocompatibility, islet functionality). We believe that this comprehensive mechanical analysis of PEG-reinforced ALG networks will add to the current body of literature and provide a more complete framework for future manipulation and improvements.^[75] Ultimately, mechanically stable micro-capsules should allow for the long-term immunoisolation and transplantation of allogeneic cells in human patients.