Partially oxidized alginate as a biodegradable carrier for glucose-responsive insulin delivery and islet cell replacement therapy

Abstract

Self-regulated insulin delivery that mimics native pancreas function has been a long-term goal for diabetes therapies. Two approaches towards this goal are glucose-responsive insulin delivery and islet cell transplantation therapy. Here, we develop biodegradable, partially oxidized alginate carriers for glucose-responsive nanoparticles or islet cells. We tune material composition and formulation in each of these contexts to enable glycemic control in diabetic mice. For injectable, glucose-responsive insulin delivery, 0.5 mm 2.5% oxidized alginate microgels facilitate repeat dosing and consistently provide 10 days of glycemic control. For islet cell transplantation, 1.5 mm capsules comprised of a blend of unoxidized and 2.5% oxidized alginate maintain cell viability and glycemic control over a period of more than 2 months while reducing the volume of non-degradable material implanted. These data show the potential of these biodegradable carriers for controlled drug and cell delivery for the treatment of diabetes with limited material accumulation in the event of multiple doses.

Graphical Abstract

Partially oxidized alginate is used to provide biodegradable carriers for self-regulated diabetes therapies. Material composition and formulation is tuned to encapsulate and deliver glucose-responsive, insulin-loaded nanoparticles or insulin-secreting pancreatic islet cells. Diabetic animals receiving these therapies experience glycemic control with minimal accumulation of material over time.

1. INTRODUCTION

Alginate is a naturally occurring polysaccharide and popular biomaterial for drug delivery and cell therapy due to its biocompatibility, mild gelation conditions, and natural abundance^1,2. Commercial alginates extracted from brown algae are non-degradable in mammals, which lack the enzyme to digest the polysaccharide backbone³. Although hydrogels may dissociate over time due to exchange of divalent cation crosslinking agents with physiologically abundant monovalent cations, the polymer chains remain intact. Since these chains are often larger than the molecular weight cutoff for renal clearance, alginate may not be completely removed from the host and accumulate over multiple treatments⁴. To aid in renal clearance, the partial oxidation of alginate chains renders them susceptible to hydrolysis and degradation into lower molecular weight species in mammalian systems^5–8. Oxidized alginate microgels have shown promise in applications of tissue engineering and regenerative medicine^9–13. Here, we explore the use of oxidized alginate microgels as degradable drug and cell carriers for applications in self-regulated diabetes therapies (Fig. 1a).

Figure 1.

Oxidation of alginate can be tuned for applications in controlled drug delivery and cell therapy. a, Schematic of oxidized alginate capsules used for drug delivery and cell therapy. Microgels act as a depot for sustained, stimuli-responsive release of drug when injected s.c. Larger alginate capsules provide an immunoprotective barrier for cells when transplanted intraperitoneally. b, Concentration of aldehyde groups on alginates oxidized to varying extents. Percentages correspond to the ratio of sodium periodate to number of alginate repeat units. Data represent mean ± SD. c, Example structure of oxidized alginate and its degradation products. d, Bright field images of Ca²⁺-crosslinked 7.5% oxidized alginate microgels before (top) and after (bottom) 6 solution exchanges causing dissolution in vitro. Scale bar = 1 mm. e, Molecular weight by gel permeation chromatography of alginates before and after incubation in 10 mM NaOH for 1 week.

Therapies that self-regulate insulin delivery have the potential to improve glycemic control and quality of life for diabetic individuals. Two avenues that have been investigated to achieve this goal are glucose-responsive insulin delivery and islet cell transplantation^14–16. In glucose-responsive insulin delivery, stimuli-responsive biomaterials may be coupled to glucose sensors to modulate insulin release, limit hypoglycemia, and improve glycemic control compared to conventional patient administration¹⁷. Common glucose sensors are sugar-binding lectins^18–20, phenylboronic acids^21–23, and glucose oxidase (GOx)^24–26. GOx converts glucose to gluconic acid, thus translating a change in glucose concentration to a change in local environmental pH²⁷. Catalase is often delivered with GOx to convert byproduct hydrogen peroxide into water and oxygen²⁷. We have previously developed pH-responsive acetalated-dextran nanoparticles loaded with insulin, GOx, and catalase to enable rapid glycemic control that lasts for 16 h in a mouse model of diabetes²⁸. We subsequently encapsulated these nanoparticles into alginate microgels to extend glycemic control on the order of weeks²⁹. To facilitate repeat dosing, here we investigate the use of oxidized alginate to form degradable, nanoparticle-loaded microgels.

As a longer-term alternative to glucose-responsive insulin delivery, the transplantation of insulin-secreting islet cells by the Edmonton protocol has resulted in improved clinical outcomes for certain diabetic patients^30,31. However, islet graft function diminishes over time³². In a summary of 255 islet transplant recipients, median graft survival was 5.9 years³³. Furthermore, 88% of patients required two or more islet infusions and 37% required three or more³³. To enable islet transplantation without systemic immunosuppression, alginate microgels have been used to encapsulate and deliver islet clusters, functioning as semipermeable barriers that exclude host immune cells while permitting insulin release and nutrient exchange³⁴. We postulated that oxidized alginate capsules could be tuned to provide an immune barrier during islet function and subsequently degrade after the graft no longer provides glycemic control. Since several islet doses are often needed to achieve glycemic control, this approach could limit the amount of material accumulation over time.

In this report, we develop injectable oxidized alginate microgels that exhibit favorable degradation kinetics for the delivery of glucose-responsive insulin-loaded nanoparticles. Additionally, we form larger capsules comprised of a blend of non-oxidized and partially oxidized alginate that isolate xenogeneic islets from the host immune system and degrade around the time of graft failure.

2. RESULTS

To determine the optimal microgel composition for glucose-responsive insulin delivery and islet cell therapy, we first made various oxidized alginate polymers with different extents of oxidation, ranging from 0 to 7.5% (see methods). Oxidation of ultrapure high molecular weight alginate (SLG100) with sodium periodate leads to the formation of aldehyde functional groups⁶, and we show that the concentration of aldehydes linearly increases with percent oxidation (Fig. 1b). These aldehydes react with neighboring hydroxyl groups to form cyclic hemiacetals which are susceptible to hydrolysis, leading to degradation of the alginate backbone⁶ (Fig. 1c). Microgels comprised of oxidized alginate thus swell and dissolve over several solution exchanges (Fig. 1d). After synthesizing these polymers, we measured their molecular weights (Mw) by gel permeation chromatography (GPC, Fig. S1). Since the oxidation process reduces the initial Mw of alginate, we use a lower molecular weight ultrapure alginate (SLG20) as the 0% oxidized control. To ensure maximal degradation and determine the Mw of end-stage degradation products, we incubated the alginates under basic conditions for one week. SLG20 has a Mw around 175 kDa which remains constant upon incubation (Fig. 1e). The oxidized alginates have initial Mw between 125 and 200 kDa and final Mw less than 60 kDa (Fig. 1e, Table S1), which is often considered the cutoff for renal clearance⁴.

2.1. 0.5 mm microgels comprised of 2.5% oxidized alginate provide an injectable, degradable insulin delivery system

In the context of drug delivery, injectability can simplify administration, and particles of less than 0.5 mm can be injected subcutaneously (s.c.) through standard needle sizes. Therefore, we formed 0.5 mm diameter Ca²⁺-crosslinked microgels comprised of the different polymers for applications in glucose-responsive insulin delivery. We first tested in vitro stability by imaging microgels immediately after formation and then after multiple solution exchanges or washes (Fig. 1d, Fig. S2a). After 4 washes, the 7.5% oxidized alginate microgels swelled to almost 3 times their initial size before completely dissolving (Fig. S2b). The 5% oxidized alginate microgels experienced initial swelling which remained relatively constant over 28 washes while neither the 2.5% nor 0% oxidized alginate microgels swelled substantially. Based on these results, we anticipated that more highly oxidized microgels would degrade more rapidly in vivo.

To evaluate microgel degradation in a physiological setting, we then injected mice s.c. with 100 μL of microgels and measured the volume of the resulting implant for the following 31 d. The unoxidized implant swelled initially in vivo, and then remained constant over the subsequent month (Fig. 2a). Each of the oxidized alginate implants decreased in size over time, and degradation rate correlated with percent oxidation. To quantify the amount of degradation, we calculated the area under the implant volume vs. time curve. Using this metric, all oxidized alginate implants had significantly reduced volumes over time compared to the unoxidized implant, and the 7.5% oxidized implant was significantly reduced compared to the 2.5% (Fig. 2b).

Figure 2.

Upon s.c. injection, oxidized alginate microgels degrade over time. a, Total implant volume of s.c. injected oxidized alginate microgels (0.5 mm diameter) decreases over the course of a month corresponding to the percent of oxidation. Unoxidized alginate microgels initially swell, and the total implant volume remains relatively constant. b, Area under the curve for implant volumes in (a). Statistical significance was determined by one-way ANOVA with Bonferroni multiple comparison correction. **** p < 0.0001 in comparison to 0% oxidation, ^## p = 0.006. c, H&E staining of representative histological sections of the implants after 31 days. Microgels stain pink. Implant thickness decreases with increasing percent oxidation. Scale bar represents length indicated for each group. d, Masson’s trichrome staining of representative histological sections of the implants after 31 days. Arrows represent thin fibrous capsule. Microgels stain blue. Scale bar = 0.2 mm. n=5 mice/group; data represent mean ± SEM.

At the end of 31 d, the unoxidized implants were on average ~150% of their initial volume, while the 7.5% implants were no longer visible or palpable (Fig. 2a, S3a). At this time, individual microgels were retrieved from the implants and imaged, or the implants and surrounding tissues were excised and processed for histology. While a large volume of microgels remained from the unoxidized implants, minimal material was retrieved from the 7.5% oxidized implants (Fig. S3b). Consistent with macroscopic implant measurements and retrieved volumes, H&E staining showed that the final thickness of the implant decreases with increased percent oxidation (Fig. 2c). Higher resolution H&E images show there are more cellular infiltrates in the oxidized implants (Fig. S4).

We also used Masson’s trichrome to identify connective tissue and collagen deposition. The unoxidized microgels have a thin fibrous capsule (arrow, Fig. 2d) surrounding the entire implant, with no visible connective tissue between individual microgels. Degraded oxidized alginate implants have a similar fibrous capsule (arrows, Fig. 2d) with increasing amounts of collagen between individual microgels, corresponding to areas of increased cellular infiltration (Fig. S4).

Although histological processing causes the microgels to shrink and alters their morphology, the unoxidized microgels remain relatively spherical compared to the flattened structures of the oxidized microgels (stained pink in Fig. 2c, S4 and blue in Fig. 2d). In sum, these results show that oxidized alginate microgels, but not unoxidized controls, degrade over time in vivo.

To create a glucose-responsive insulin delivery system, we next encapsulated acetalated-dextran nanoparticles containing insulin, glucose oxidase, and catalase (NPs) into 0.5 mm oxidized alginate microgels. We investigated the release kinetics of insulin from the microgels upon acid-mediated NP degradation and found that 35–45% of insulin is released within the first 30 min, followed by controlled release over 6 h (Fig. S5). Alginates with higher degrees of oxidation resulted in more rapid insulin release.

To determine their efficacy as a self-regulated insulin therapy, we administered a single dose of NP-encapsulated microgels (60 IU/kg insulin), free insulin (3 IU/kg), or empty microgels (0 IU/kg insulin) s.c. to STZ-induced type 1 diabetic mice. Over the first 1.5 h, all NP-encapsulated microgels (2.5%, 5%, and 7.5% oxidized) reduced the blood glucose levels (BGs) of diabetic mice into the normoglycemic range at the same rate as free insulin (Fig. 3a, S6a,b). After 3 h, mice were given an intraperitoneal (i.p.) glucose tolerance test (GTT, 1.5 g/kg) to evaluate their response to an increase in BG. Following this test, BGs of mice receiving free insulin returned to their initial glycemic state (~ 500 mg/dL). Conversely, BGs of mice receiving NP-encapsulated microgels remained at normoglycemic levels (< 200 mg/dL). These trends continued for 3 additional GTTs over the subsequent 7 h, and glycemic control was similar to that of healthy mice (Fig. S6b,c).

Figure 3.

Oxidized alginate microgels loaded with glucose-responsive nanoparticles provide extended glycemic control in diabetic mice with reduced hypoglycemia in healthy mice. a, Blood glucose levels of diabetic mice receiving empty microgels, free insulin (3 IU/kg), or nanoparticle-encapsulated microgels (2.5%, 5%, and 7.5% oxidized alginate; 60 IU/kg insulin) over the first 10 h with four i.p. glucose tolerance tests (GTT, black arrows, 1.5 g/kg). b, Blood glucose levels of treated mice following a GTT (black arrow, 3 g/kg) three days after injection, compared to healthy mice. c, Area under the curve quantifying results in (b). d, Blood glucose levels of mice receiving one (5% and 7.5%; d0) or two (2.5%; d0 and d12) doses of nanoparticle-encapsulated microgels (blue arrows; 60 IU/kg insulin). e, Blood glucose levels of healthy mice upon administration of free insulin (3 IU/kg) or nanoparticle-encapsulated microgels (2.5%, 5%, and 7.5% oxidized alginate; 60 IU/kg insulin). f, Area above the curve and below the initial average starting blood glucose level for each group to quantify results in (e). Horizontal lines in (a), (b), and (e) indicate normoglycemia (<200 mg/dL). n=4–5 mice/group; data represent mean ± SEM. Statistical significance was determined by one-way ANOVA with Bonferroni multiple comparison correction. NS p > 0.9999, **** p < 0.0001 in comparison to free insulin.

Although all NP-encapsulated oxidized alginate microgels provided similar glycemic control on the day of injection, only mice receiving 2.5% microgels responded to a GTT (3 g/kg) 3 days post-injection similarly to healthy mice (Fig. 3b,c). While 5% and 7.5% microgels failed to correct hyperglycemia beyond d0, 2.5% microgels provided ~10 d of glycemic control (Fig. 3d). Mice receiving 2.5% microgels were administered a second dose on d12, which provided an additional 10 d of glycemic control. In healthy, normoglycemic mice, all microgels mitigated BG reduction compared to free insulin (Fig. 3e,f, S7), indicative of their glucose-responsive behavior and limited hypoglycemic effect. Therefore, a single dose of 2.5% microgels consistently provides over a week of glycemic control with reduced risk of hypoglycemia compared to free insulin.

2.2. 1.5 mm capsules comprised of a blend of 0% and 2.5% oxidized alginate provide immunoprotection for cell therapy with reduced material accumulation

To further extend glycemic control from the order of weeks to the order of months, we explored the potential of using oxidized alginate microgels for islet cell replacement therapy. In this context, we have previously shown that 1.5 mm alginate capsules prolong cell viability and glycemic control in rodents and nonhuman primates compared to those with 0.5 mm diameter ³⁵. Therefore, here we formed 1.5 mm Ca²⁺-crosslinked capsules comprised of oxidized alginates and evaluated their in vitro stability. Both 7.5% and 5% capsules swelled and dissolved over several washes, and 2.5% capsules experienced initial swelling (Fig. S8). Due to the lower in vitro stability of the larger oxidized capsules compared to 0.5 mm microgels, we also formed capsules made from a 50:50 blend of 2.5% and 0% oxidized alginate, which did not swell substantially. We then implanted capsules made from 2.5%, the 2.5% / 0% blend, or 0% oxidized alginate in the i.p. space of mice to evaluate their degradation profile. At indicated time points, we retrieved and imaged the capsules. On d7, none of the 2.5% capsules could be retrieved, suggesting that they degrade rapidly in the i.p. space (Fig. S9a). About 60% of volume of the blend and 100% of the unoxidized capsules were retrieved. On d28 and d115, ~60% and ~40% of blend capsules were retrieved, respectively. Images of the retrieved material on d28 show spherical capsules in the unoxidized control and degraded capsules in the blend group (arrows, Fig. S9b). Therefore, we hypothesized that the blend may be advantageous for supporting islet function with reduced material accumulation in vivo.

To test this hypothesis, we encapsulated rat islets inside 1.5 mm capsules made from 0% oxidized alginate, 2.5% alginate, or the 2.5% / 0% blend. Following encapsulation, the morphology of the islets and the capsules were visually similar (Fig. S10). We then transplanted the encapsulated rat islets (500 IE) into diabetic mice and measured BG levels over the following 3 months. None of the mice receiving 2.5% capsules reached normoglycemia. Conversely, glycemic control of mice receiving the blend was similar to that from mice receiving unoxidized control capsules for the first two months (Fig. 4a). The median periods of normoglycemia for the blend and unoxidized capsules were 67 and 81 d, respectively (Fig. 4b). Mice receiving the 0% / 2.5% blend all became hyperglycemic around the same time while mice receiving control 0% capsules had a more heterogeneous response. Until the point where at least one mouse from each group was hyperglycemic (d63), the areas under the BG curves were similar for the 0% and 0% / 2.5% groups (Fig. 4c). After 92 d, the capsules were retrieved and imaged. While almost all of the initial volume of 0% capsules was retrieved, only ~60% of the volume of the blend remained (Fig. 4d). The unoxidized capsules retained their spherical shape, and the blend capsules were visually degraded (Fig. 4e). Both groups showed a small amount of collagen deposition by phase contrast imaging. Since oxidized alginate capsules produce a more consistent response and result in less material accumulation, they may be advantageous for the transplantation of therapeutic cells.

Figure 4.

Oxidized alginate capsules encapsulating rat pancreatic islet cells (500 IE) provide long-term glycemic control in diabetic mice with a significant reduction in retrieved material. a, Blood glucose levels of diabetic mice for three months following i.p. transplantation of islet cells encapsulated in 1.5 mm oxidized alginate capsules (0% oxidized, 0% / 2.5% indicating a 50:50 blend by volume of 0% and 2.5% oxidized, or 2.5% oxidized). b, Kaplan-Meier plot showing the percent of normoglycemic mice in (a). Statistical significance was determined by two-sided log-rank (Mantel-Cox) test. c, Area under the curve in (a) according to the interval before (d0 - d63) or after (d63 - d89) the time when each group had at least one mouse reach hyperglycemic criteria. d, Total volume of capsules retrieved from implanted volume of 500 μL after d92. e, Representative phase contrast images of retrieved capsules showing minimal fibrosis in each group. Scale bar = 1 mm. n=4–5 mice/group; data represent mean ± SEM. Statistical significance was determined by one-way ANOVA with Bonferroni multiple comparison correction unless otherwise indicated. NS p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 in comparison to 0% oxidized capsules.

3. DISCUSSION

The objective of this study was to develop degradable alginate carriers for drug and cell delivery in the context of diabetes therapies. To accomplish this goal, we first oxidized alginate to varying degrees and showed that the molecular weights of maximally degraded species were all below the threshold of renal clearance⁴. This key design feature enables eventual degradation products to be filtered through the kidney and excreted, thereby preventing accumulation of material over time.

For applications in glucose-responsive insulin delivery, we evaluated in vivo degradation through s.c. injection of 100 μL of 0.5 mm oxidized alginate microgels. This implant volume enabled a macroscopic readout such that we could reliably track degradation and obtain longitudinal data. While this study provides important evidence of microgel degradation in vivo, there are several limitations. The degradation rates may be altered in the context of glucose-responsive insulin delivery due to encapsulation of nanoparticles, an acidic microenvironment from GOx, altered metabolic processes of diabetic animals, and much lower volumes of microgels (~20 μL) needed for glycemic control. We expect that lower microgel volumes dispersed throughout the s.c. space will have higher surface area exposure and faster degradation rates.

Following in vivo degradation, we excised the remaining microgel implants for histological analysis. While the implants comprised of 5% and 7.5% oxidized alginate microgels were no longer macroscopically visible, some of the material remained. Cellular infiltration and collagen deposition surrounding these highly oxidized microgels could be due to degradation and natural host remodeling or to an elevated host response to the modified material^36,37. Further investigation into the kinetics of cellular infiltration as well as analysis of cell types and phenotypes are needed to distinguish between these two processes.

In all short term quantitative metrics of glycemic control in vivo (i.e., in Fig. 3c,f, S6b,d), 2.5% oxidized alginate microgels behave similarly to unoxidized microgels (p > 0.05)²⁹. The advantages of the 2.5% oxidized microgels lie in their enhanced degradability compared to unoxidized microgels (Fig. 1e, 2a) and their more reproducible glycemic profile upon multiple doses (Fig. 3d) compared to previously published results²⁹. Therefore, these degradable microgels show promise in facilitating repeat dosing with reduced material accumulation, and further investigation into long-term models (e.g., 3 months) with several doses is warranted.

For applications in islet transplantation, we had similar goals. We aimed to develop carriers that protect encapsulated cargo (nanoparticles or islets) during their functional lifetime and then degrade once the cargo has been exhausted (due to insulin release or islet graft failure, respectively) to facilitate repeat dosing. Alginate capsules offer an immunoprotective barrier to avoid the need for systemic immunosuppression³⁸. Without this barrier, transplant rejection of even allogeneic tissues occurs³⁹. Here, we use a more rigorous model of xenogeneic islet transplantation and show by BG measurements that 0% / 2.5% blend capsules provide immunoprotection and support prolonged islet function in vivo (Fig 4a).

The blend of 0% and 2.5% oxidized alginate was also found to be necessary to achieve in vitro capsule stability upon several solution exchanges (Fig. S8). The lower stability of the 1.5 mm 2.5% capsules compared to the smaller 0.5 mm microgels may be due to the limited penetration depth of the calcium ions, resulting in a larger volume of un-crosslinked alginate in the center of the spheres^40,41. In vivo, we found that the blend supports islet function similar to unoxidized alginate. In fact, up until the time when at least once mouse from each group experiences graft failure, the areas under the curve for these two groups are statistically similar (Fig. 4c). At this point, the group receiving blend capsules has a homogenous response which may be beneficial for repeat dosing, whereas the group receiving unoxidized capsules is much more heterogeneous.

Even when islets are successfully engrafted into immunosuppressed hosts, their functional lifetime is finite³³. Encapsulated islets may also decline in function over time for a number of reasons including fibrosis and hypoxia^34,42. As a result, there are many promising approaches aimed at preventing graft failure, for example by developing super-biocompatible materials⁴³ or administering external oxygen sources⁴⁴. Our goal here is not to prevent graft failure but to design capsules that degrade upon inevitable graft failure to limit material accumulation. We accomplished this goal by reducing retrieved capsule volume by ~35% (Fig. 4d). While the 50:50 blend worked well in this context, future studies should be performed to determine the optimal degradation kinetics needed to support the graft while it is functioning and subsequently degrade to limit material accumulation upon graft failure. The extent of oxidation and composition of the blend can be tuned to match these optimal rates in future work.

We hypothesize that the 2.5% alginate capsules exhibited reduced stability and failed to provide an immunoprotective barrier for the islets thus failing to restore normoglycemia in recipient mice (Fig. 4a,b). Our degradation studies using empty capsules support this hypothesis; we were unable to retrieve any empty 2.5% capsules 7 days after transplantation (Fig. S9a), suggesting they degrade rapidly in this context. It cannot be ruled out, however, that the 2.5% oxidized alginate itself affects islet health, and a blend of unoxidized alginate is necessary to support islet function. Although previous literature indicates that oxidized alginate exhibits high cytocompatibility^45–48, future studies on islet viability and more granular degradation kinetics are needed to better understand why the 2.5% capsules failed.

4. CONCLUSION

In summary, we report the development and formulation of degradable hydrogel capsules for glucose-responsive insulin delivery and islet cell replacement therapy. These proof-of-concept studies in two different models provide encouraging data supporting the use of oxidized alginate in diabetes therapies. We anticipate this biodegradable material will facilitate repeat doses of insulin and support islet cell transplantation without excess accumulation of material.

MATERIALS AND METHODS

All chemicals were obtained from Millipore Sigma (St. Louis, MO) and cell culture reagents from Life Technologies (Carlsbad, CA) unless otherwise noted. PRONOVA SLG100 and SLG20 were purchased from NovaMatrix (Sandvika, Norway). Recombinant human insulin (Gibco) was purchased from ThermoFisher Scientific (Waltham, MA). All animal protocols were approved by the MIT Committee on Animal Care, and animals were cared for under supervision of MIT’s Division of Comparative Medicine.

Alginate oxidation and characterization

Oxidized alginate preparation was adapted from a previously described protocol.⁵ SLG100 (250 mg) was solubilized in Milli-Q ultra-pure water (12.5 mL) by shaking at room temperature (RT) overnight. An aqueous solution of sodium periodate (6.25 mg, 12.5 mg, or 18.75 mg in 12.5 mL corresponding to 2.5%, 5%, or 7.5% oxidized) was added, and the reaction proceeded under shaking at RT in the dark. After 24 h, ethylene glycol (0.125 mL) was added to quench unreacted periodate, and the solution continued shaking for an additional 1 h. The solution was then dialyzed against ultra-pure water for 3 d (Spectra/Por regenerated cellulose membrane, MWCO 3.5 kDa) and lyophilized to form a white fluffy product. The final concentration of aldehydes was determined using a commercial colorimetric aldehyde assay kit (Millipore Sigma). Alginate molecular weight was determined by gel permeation chromatography (GPC) using a Waters system with a PL aquagel-OH column, aqueous mobile phase (20% MeOH), and linear PEO molecular weight standards. GPC was performed immediately after solubilization of the final product (pre-degradation) and after 7 d at 37 °C in PBS + 10 mM NaOH to represent completely degraded polymer (post-degradation).

Microgel fabrication

Alginate microgels were fabricated with a custom-designed, electro-spray system comprised of a vertically mounted syringe pump, a voltage generator, and a grounded metal collecting dish containing a CaCl₂ (50 mM) gelling solution. Due to differences in viscosity, the weight percent of alginate was optimized independently for each of the oxidized polymers. For 0% oxidized alginate (SLG20), 1.4% alginate was used as previously described.²⁹ 1.6%, 2%, and 2.5% alginate were used for 2.5%, 5%, and 7.5% oxidized alginate, respectively. 1.6% was also used for the blend of 0% and 2.5% oxidized alginate.

To form empty or nanoparticle-encapsulated 0.5 mm microgels, the aforementioned weight percent of alginate was solubilized in 0.9% saline or 0.9% saline containing 20 mg/mL NPs (synthesized as previously described^28,29; see supplemental methods). A 25 G 1.5 in blunt needle and a flow rate of 0.18 mL/min were used to generate microgels. The voltage was optimized for each solution between 6–8 kV to ensure particle diameter around 0.5 mm. After crosslinking, the microgels were washed with saline containing 2 mM CaCl₂.

To form empty or islet-encapsulated 1.5 mm microgels, the aforementioned weight percent of alginate was solubilized in 0.9% saline or 0.9% saline containing 1,000 Sprague-Dawley rat islets per 0.75 mL alginate solution. An 18 G 1 in blunt needle and a flow rate of 0.18 mL/min were used to generate microgels. Islets were prepared as described below and divided into thirds. One third was encapsulated in 0% oxidized alginate, one third was encapsulated in a 50:50 blend of 0% and 2.5% oxidized alginate, and the remaining third was encapsulated in 2.5% oxidized alginate. The voltage was optimized for each solution between 5–7 kV to ensure particle diameter around 1.5 mm. After crosslinking, the microgels were washed with HEPES supplemented with CaCl₂ and transferred into RPMI Medium 1640 with 10% heat-inactivated FBS and cultured overnight at 37 °C prior to transplantation.

Subcutaneous microgel implantation, retrieval, and histology

100 μL of microgels in saline was s.c. injected into the backs of C57BL/6J mice through an 18 G needle. The implant volumes were measured with calipers 2 h post-injection once the injected saline had dissipated. The implants were measured the following day and then every 2–4 d for a total of 31 d. On d31, mice were euthanized by CO₂ administration followed by cervical dislocation. An incision was made near the implant along the skin and peritoneal wall. The whole implant was excised from the epidermis to the peritoneal wall so that it could be examined in situ. Retrieved implants were fixed in Bouin’s solution for 48 h followed by several days of washes in water and 70% EtOH. Implants were processed for histological analysis, and 5 μm sections were taken near the beginning, middle, and end of the implant. Sections were stained with H&E and Masson’s trichrome. A pathologist was consulted in analysis of the samples.

In vivo blood glucose studies for glucose-responsive insulin delivery

The efficacy of NP-encapsulated oxidized alginate microgels was evaluated in diabetic C57BL/6J mice (Jackson Labs). To induce diabetes, 8-week-old male mice were fasted overnight and injected with a single dose of 150 mg/kg streptozotocin (STZ). Blood glucose levels were monitored for the following week, and only mice with blood glucose levels consistently over 400 mg/dL were considered diabetic. Groups of at least 4 mice were fasted for 10 h and s.c injected with microgels (60 IU/kg insulin) through an 18 G needle or naked insulin (3 IU/kg insulin) through a 30.5 G needle. A ~1 μL drop of blood from the tail vain was used to monitor blood glucose levels every 30 min with Clarity BG1000 Blood Glucose Meters. For long term studies, mice were fasted 10 h prior to measuring their blood glucose. Glucose tolerance tests were performed by i.p. administration of 1.5–3 g/kg glucose in saline. Mice that were hyperglycemic for more than 7 days in this model were euthanized due to systemic toxicity of high-dose STZ. To evaluate the hypoglycemic effect, healthy 8-week-old male C57BL/6J mice were similarly injected with microgels (60 IU/kg) or naked insulin (3 IU/kg), and their blood glucose was monitored every 0.5 h for the initial 2.5 h following administration.

Islet isolation

Islets were isolated and cultured as previously described⁴⁹. Briefly, male Sprague–Dawley rats (Taconic) were anaesthetized with xylazine and ketamine, the bile duct was cannulated, and the pancreas was distended with an injection of Liberase (Roche). Rats were sacrificed by cutting the descending aorta, and the distended pancreatic organs were removed. Pancreases were digested for 30 min at 37 °C, washed twice with RPMI 1640 media with 10% heat-inactivated FBS, and filtered through a 450-μm cell strainer. Islets were suspended in a Histopaque-1077, centrifuged at 1,700 RCF at 4 °C, collected from the gradient, and further purified by gravity sedimentation. Purified islets were counted by hand, washed three times in sterile PBS, and cultured in RPMI 1640 media with 10% heat-inactivated FBS and 1% penicillin-streptomycin overnight prior to encapsulation.

In vivo blood glucose studies for islet cell transplantation

Diabetic C57BL/6J were ordered from Jackson Laboratory following a low-dose, multiple injection protocol (50 mg STZ/kg for 5 consecutive days). Mice were STZ-induced d The blood glucose levels of all mice were tested upon arrival after a 2 h fasting period, and only mice with blood glucose levels over 400 mg/dL were considered diabetic and underwent transplantation. Mice were anesthetized with 3% isoflurane in oxygen, and their abdomens were shaved and sterilized with betadine and isopropanol. Preoperatively, mice were s.c. administered 0.05 mg/kg extended-release buprenorphine and 0.3 mL of 0.9% saline. A small incision was made along the midline of the abdomen, and a second incision was made through the peritoneal wall. Empty microgels or microgels encapsulating islets (500 islet equivalents) were implanted into the peritoneal cavity through a transfer pipette. The peritoneal incision was then closed using 5–0 taper-tipped polydioxanone (PDS II) absorbable sutures. The skin was closed over the incision using a wound clip and tissue glue. Mice were monitored daily for 3 d postoperatively. Blood glucose levels were monitored ~every 3 d for a period of ~3 months. A ~1 μL drop of blood from the tail vain was used to monitor blood glucose levels with Clarity BG1000 Blood Glucose Meters. For the Kaplan-Meier plot, a curative normoglycemia threshold of 200 mg/dL for two consecutive blood glucose measurements was applied. Once normoglycemia was reached, hyperglycemia was defined as two consecutive measurements above 250 mg/dL.

Intraperitoneal microgel retrieval and imaging

At indicated time points post-implantation, mice were euthanized by CO₂ administration followed by cervical dislocation. An incision was made along the abdomen skin and peritoneal wall, and Krebs buffer was used to wash out the microgels from the i.p. space. Microgels and lavage were collected into 50 mL conical tubes. Several washes were performed. All tissues were extremely thoroughly inspected for remaining microgels. Microgels which had adhered to tissues were manually retrieved with forceps. After ensuring all microgels were retrieved, they were transferred to 0.6 mL centrifuge tubes, and the retrieved volume was estimated. Microgels were gently washed twice with Krebs buffer and transferred into 35 mm petri dishes for phase contrast microscopy using an EVOS Xl microscope (Advanced Microscopy Group).

Statistical analysis

Statistical analysis was performed with GraphPad Prism 9.1.1. Data are expressed as mean ± SEM unless otherwise indicated, and n = 4 – 6 randomly assigned mice per time point and per group. Data were analyzed for statistical significance by one-way ANOVA with Bonferroni multiple comparison correction unless otherwise indicated.