Delivery of Dissociated Islets Cells within Microporous Annealed Particle Scaffold to Treat Type 1 Diabetes

Abstract

Type 1 diabetes (T1D) is caused by the autoimmune loss of insulin-producing beta cells in the pancreas. The only clinical approach to patient management of blood glucose that doesn’t require exogenous insulin is pancreas or islet transplantation. Unfortunately, donor islets are scarce and there is substantial islet loss immediately after transplantation due, in part, to the local inflammatory response. The delivery of stem cell-derived beta cells (e.g., from induced pluripotent stem cells) and dissociated islet cells hold promise as a treatment for T1D; however, these cells typically require re-aggregation in vitro prior to implantation. Microporous scaffolds have shown high potential to serve as a vehicle for organization, survival, and function of insulin-producing cells. In this study, we investigated the use of microporous annealed particle (MAP) scaffold for delivery of enzymatically dissociated islet cells, a model beta cell source, within the scaffold’s interconnected pores. We found that MAP-based cell delivery enables survival and function of dissociated islets cells both in vitro and in an in vivo mouse model of T1D.

Graphical Abstract

In this manuscript, we report the delivery of dissociated islets within the pores of microporous annealed particle (MAP) scaffold to treat Type 1 Diabetes. We found that MAP scaffold supported the viability and function (glucose stimulated insulin secretion) of dissociated islet in vitro, as well as maintained normoglycemia in a streptozotocin-induced mouse model of diabetes.

1. Introduction

Type 1 diabetes (T1D) affects 1.6 million Americans^[1], and is most frequently caused by the autoimmune destruction of the insulin-producing beta cells in the pancreas which can lead to serious or fatal health complications if left untreated. Most patients with T1D manage their diabetes through a combination of blood glucose monitoring (e.g., finger prick testing) and systemic exogenous insulin delivery (e.g., intramuscular injection). While advanced monitoring and delivery techniques do exist (e.g., implantable insulin pumps), the only clinical approach to patient management of blood glucose that doesn’t require exogenous insulin is whole pancreas transplantation or pancreatic islet delivery via portal vein injection. Unfortunately, donor islets are rare; only 1,086 patients were treated with islet transplantation worldwide between 1999 and 2015^[2]. This scarcity of islets has prompted the search for alternative sources of beta cells.

Stem cell-based approaches using induced pluripotent stem cells (iPSCs) and pancreas-derived multipotent precursor cells are currently being explored for beta cell production and represent a promising cell source for treatment of T1D^[3]. Outside of stem-cell derived beta cells, native beta cells derived from dissociated islets are a relatively accessible cell source (i.e., don’t require extensive differentiation protocols) that can serve as a model cell type for stem-cell derived approaches. However, several studies^[4–10] have shown that both dissociated islets and stem-cell derived beta cells typically require cell aggregation in vitro prior to implantation. This facilitates biochemical and biomechanical cell interactions that are necessary for cell differentiation, survival, and function. Additionally, it has been shown that dissociated islets cannot survive in vivo without close cellular communication^[11–14] and extracellular matrix mimicking ligands^[11,15–17]. The formation of “pseudo-islets” typically involves enzymatically dissociating whole islets into single cells or differentiating beta cell precursors, then forming islet-like clusters of a desired size that promote better nutrient diffusion and cell survival. These pseudo-islet clusters can be formed in microwells^[18], Matrigel^[19], or via hanging drop method^[20], and typically require a 3–5 day “pre-conditioning” period in vitro prior to implantation. Another alternative delivery method is via synthetic microporous scaffolds^[17,20–22], which allows for cell self-organization within pores that can mimic native islet structure and facilitate cell-cell signaling.

We chose to investigate the delivery of dissociated islets (as a model cell type for stem-cell derived beta cells) without pre-implantation conditioning using microporous annealed particle (MAP) scaffold, which is an injectable biomaterial composed entirely of hydrogel microspheres that are annealed in situ to provide a material environment with cell-scale porosity^[23]. The MAP scaffold platform has been previously used to accelerate dermal wound healing^[23], but has since been shown to be successful in a multitude of other regenerative medicine applications^[24,25], including serving as a delivery vehicle for stem cells^[26], conveying anti-inflammatory effects to host tissues through recruitment of pro-regenerative macrophages^[27], and promoting a Th2 “tissue repair” type T-cell response^[28]. For this study, we chose to use a recently-designed MAP scaffold formulation comprised of a heterogenous composition that adds heparin-containing microspheres (heparin μislands), which has previously demonstrated the ability to organize endogenous growth factors in a diabetic wound environment and significantly improve re-vascularization^[27,29]. Given our success using MAP scaffold as an angiogenic and immunomodulatory biomaterial capable of cell delivery, we were motivated to use MAP as a delivery platform for dissociated islet cells (containing beta, alpha, and delta cells) to treat T1D in mice. By mixing dissociated islet cells with MAP and assembling the scaffold in situ (Figure 1A), we were able to support cell organization, survival, and function for a prolonged period, which did not require prior in vitro formation of pseudoislets.

Figure 1.

A) Overview of whole islet dissociation, filtration, and implantation with MAP scaffold under the renal subcapsular space. The MAP scaffold contains heparin-μislands, which has previously demonstrated to significantly improve vascularization within the scaffold^[27]. B) 3D rendering of MAP scaffold pores (green dextran) and heparin-μisland particles (red) in Imaris software (Oxford Instruments). Imaris software was also used to quantify the surface area of the pores, which was then converted to circular pore diameters.

2. Results and Discussion

2.1. Microgel Synthesis and Scaffold Pore Sizing

Our investigation focused on a MAP microgel formulation that consisted of a 4-arm polyethylene glycol (PEG)-maleimide backbone, an enzymatically degradable peptide crosslinker, a cell adhesive peptide pendant group, and a custom MethMal annealing macromer (previously described for accelerated MAP assembly in situ^[30]). For animal studies, a minority of microgels containing immobilized thiolated heparin were included heterogeneously throughout the MAP scaffold to promote accelerated vascularization^[27]. All microgels were synthesized using a high-throughput microfluidics technique as previously described^[31]. After separate microgel synthesis and purification, heparin-containing microgels were mixed in with the no-heparin microgels at a 1:10 ratio (10% heparin-μislands). This ratio was chosen based on previously published data, where 10% heparin-μislands facilitated significantly increased endothelial cell behavior in vitro and in vivo^[27]. To visualize scaffold porosity, 200 μm thick MAP scaffolds (n=3) containing 10% heparin islands were incubated with fluorescent dextran and imaged with a Zeiss 710 laser scanning confocal microscope (Figure 1B). The average median pore diameter across three scaffolds was 36.7 ± 9.62 μm.

2.2. Dissociated islet viability is improved in MAP scaffold at high cell densities in vitro

Initially, we assessed the impact of MAP scaffold on dissociated islet cell viability in vitro. Islets harvested from healthy donor C57BL/6 mice were isolated by collagenase digestion and density gradient centrifugation using our established protocol. Isolated islets were resuspended in 300 μL of 0.05% Trypsin-EDTA and incubated in 37°C water bath for 10 minutes with tapping to mechanically dissociate the islet clusters. The resulting digestate was filtered with a 35 μm cell strainer to isolate a single cell suspension. The MAP scaffold was sterilized and mixed 1:1 with a 0.2 mM lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) photoinitiator solution. For each condition, 50 islets were dissociated, filtered, and the resulting cell suspension was mixed with 80 μL of MAP gel or 80 μL of media and transferred to a 24 transwell plate insert. The MAP scaffold was annealed with a 365 nm LED light (ThorLabs, 20 mW/cm² intensity) for 30 seconds. At each timepoint (1, 24, 48, and 72 hrs), the transwell inserts (n=3 per timepoint) were removed from media and placed into 1 mL of PBS containing 50 ng/mL of fluoresceine diacetate (live cell stain) and 14.5 μg/mL of propidium iodide (dead cell stain). Wells were imaged with 4X objective on an EVOS FL Auto microscope and images were analyzed in ImageJ. After 1 hour, there was a slight decline in cell viability in both conditions (~10% loss in viability), likely due to cell death that occurred during the dissociation process (Figure 2A). After 24 hours, the cell viability in the cells only condition had decreased significantly (60% live) compared to MAP + cells condition (89% live). Cell viability in cells only condition continued to decline after 48 hr (38% live) and 72 hr (21% live) incubation. The dissociated islet cells mixed with MAP scaffold maintained significantly higher viability after 48 hr (87% live) and 72 hr (79% live) incubation time periods. It is notable that MAP scaffold facilitates dissociated islet cell survival up to 72 hours in vitro (Figure 2E).

Figure 2.

Results from viability and GSIS assays on dissociated islet cells incubated with and without MAP gel in a transwell plate. A) Mean percentage of live cells (error bars = standard deviation) imaged at 1, 24, 48, and 72 hr. n=3 wells per timepoint, two-way ANOVA with Sidak’s multiple comparison test (GraphPad Prism), alpha = 0.05, ^nsp > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. B) Mean insulin concentration (error bars = standard deviation) detected by ELISA in the high and low glucose buffers that dissociated islets cells were exposed to for 1 hour each after 1, 24, 48, and 72 hr incubation with or without MAP scaffold (GSIS assay). n=3 wells per timepoint, two-way ANOVA with Sidak’s multiple comparison test (GraphPad Prism), alpha = 0.05, ^nsp > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. C) Mean percent viability (error bars = standard deviation) of dissociated cells from 50, 12.5, or 2.5 islets incubated in MAP gel for 72 hours. n=3 wells per condition, one-way ANOVA with Tukey’s multiple comparison test (GraphPad Prism), alpha = 0.05, ^nsp > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. D) Direct comparison of dissociated islet cell viability when exposed to conditions with LAP or Eosin Y photoinitiator. Graph represents mean percent viability (error bars = standard deviation). n=3 wells per condition, two-way ANOVA with Sidak’s multiple comparison test (GraphPad Prism), alpha = 0.05, ^nsp > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. E) Representative live/dead images of dissociated islets cells incubated with or without MAP scaffold at 1-hour and 72-hour timepoints. Live cells were stained with fluorescein diacetate (FDA) and dead cells stained with propidium iodine (PI) and imaged at 4× (scale bar = 500 μm).

Based on previously published data that dissociated islet viability is dependent on cell-packing density^[17], we wanted to investigate the effects of MAP scaffold on three different concentrations of dissociated islets and assessed cell viability at 72 hours. Dissociated cells from 50, 12.5, or 2.5 islets (approximately 2,000 cells/islet) were mixed with MAP as described above, and incubated in a transwell plate for 72 hours (n=3 per condition). Cell viability in the 50 and 12.5 dissociated islet conditions was on average >78%, and there was no significant difference between groups (Figure 2C). However, the dissociated cells from 2.5 islets were only 56% viable after 72 hours, which was a significant decrease compared to the other two conditions. This indicates that even in MAP scaffold, dissociated islet cell viability is still density dependent.

The annealing step of MAP scaffold assembly relies on the use of a light-activated photoinitiator of radical polymerization and our selection of appropriate photoinitiator is application dependent. For example, lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) photoinitiator does not fluoresce, and does not interfere with the fluorescent dyes used for the in vitro viability assays. Therefore, our group chooses to use LAP for MAP scaffold annealing during in vitro experiments. However, LAP uses long-wave UV ($\lambda$_max~365nm) that is highly adsorbed by tissue^[32]. By contrast, Eosin Y is activated by visible wavelengths (505 nm) that have deeper penetration into tissue and enhances the capability of MAP scaffold to be annealed beneath tissue structures (e.g., skin, kidney capsule, etc.). As a result, our in vivo experiments use Eosin Y for radical generation and scaffold annealing. To demonstrate that both methods provide equal cytocompatibility, we directly compared the viability of dissociated islets cells in vitro after annealing the MAP scaffold under conditions with the LAP and Eosin Y photointiators, and found no difference in viability at 24 hours (Figure 2D).

2.3. Dissociated islets within MAP scaffold have improved glucose-sensing function in vitro

We further evaluated the effects of MAP scaffold on dissociated islet cell function using a glucose stimulated insulin secretion (GSIS) assay. For each condition, 30 islets were dissociated, filtered, and the resulting cell suspension was mixed with 80 μL of sterile MAP scaffold (no-heparin) or 80 μL of media and transferred to a 24 transwell plate insert. The MAP scaffold was annealed with a 365 nm LED light (ThorLabs, 20 mW/cm² intensity) for 30 seconds. At each timepoint (1, 24, 48, and 72 hrs), the transwell inserts (n=3 per timepoint) were removed from media and placed into Krebs-Ringer bicarbonate (KRB) buffer supplemented with low (2.8 mM) glucose for 1 hour. Then, the transwell inserts were transferred to high (28 mM) glucose buffer for 1 hour. The buffer from low and high glucose conditions was collected and assessed by insulin ELISA (mouse insulin ELISA kit, Mercodia Inc). It was observed that the cells within the MAP scaffold retained their glucose-sensing abilities, and secreted significantly higher amounts of insulin in response to the high-glucose environment compared to the low glucose environment (Figure 2B). The cells seeded without MAP scaffold exhibited a significant decrease in function, even after 1 hour incubation. There was no significant difference in insulin secretion between high glucose and low glucose conditions at any timepoint for the cells only condition. This indicates that dissociated islet cells lose their glucose-sensing capabilities and are unable to secrete insulin in response to glucose challenge even after 1 hour incubation on a 2D plastic substrate. Thus, indicating that our MAP scaffold offers the environment needed to maintain beta cell function.

2.4. Delivery of dissociated islets within MAP scaffold induces normoglycemia in vivo

Following the positive in vitro results described above, we investigated the efficacy of MAP delivery of dissociated islet cells in a syngeneic mouse model of diabetes. Recipient C56BL/6 mice were rendered diabetic by a single intraperitoneal injection of streptozotocin (STZ, 250 mg/kg). For transplantation at the renal subcapsular site, diabetic recipient mice were anesthetized, and the left kidney exposed through the left flank. 100 islets harvested from C56BL/6 donors were dissociated and mixed with 15 μL of MAP scaffold (containing 10% heparin-μislands) and injected through a small incision in the renal capsule using a positive displacement pipette. Control animals received 15 μL of MAP scaffold alone. The scaffold was annealed in situ by exposing the injection site to a 505 nm LED light (ThorLabs, 104 mW/cm² intensity) for 2 minutes. For control groups, dissociated cells from 100 islets were injected directly under renal capsule (no MAP) or the animals received no treatment. The kidney was returned to anatomical position, and the skin and muscle closed separately. Recipient blood glucose levels (non-fasting) were monitored daily (AccuChek blood glucose meter, Roche), and diabetes was defined as blood glucose (BG) >300 mg/dl on 2 consecutive days, with cure defined as a return to normoglycemia (BG <200 mg/dl for 2 consecutive days). Blood glucose levels of recipient mice were monitored every day for up to 44 days. By day 16, the mice that received MAP only, dissociated islets only, or no treatment had achieved hyperglycemic blood glucose levels (>600 mg/dL) and were sacrificed in accordance with the University of Virginia Animal Care and Use Committee Protocol No. 3347. By day 17, the mice that received MAP mixed with dissociated islets achieved normoglycemia (<200 mg/dL) and maintained normal glucose levels for up to 40 days post-transplant (Figure 3A). Left nephrectomy at day 40 caused a spike in blood glucose, which indicated that the implant itself was regulating the blood glucose. The dissociated islet cells implanted with MAP scaffold retained their ability to secrete insulin and maintain normoglycemia in a syngeneic implant model of STZ-induced diabetes, whereas the cells alone did not. Kidneys removed at the study endpoint were fixed, sectioned, and stained with anti-Insulin antibody. Immunohistochemistry revealed positive insulin staining of beta cells in the MAP implant (day 40 timepoint, Figure 3C), whereas no positive staining is seen in the implant site of the cells only transplant (Figure 3B).

Figure 3.

Blood glucose data and representative immunohistochemistry (IHC) images from syngeneic implant. A) Daily blood glucose levels (mg/dL) of recipient diabetic mice. Graphed points represent mean blood glucose measurements from n=6 mice (error bars = standard deviation). Left nephrectomy at day 40 was done to confirm return of hyperglycemia. B) Representative IHC image of implant site that received dissociated islets only. C) Representative IHC image of MAP implant site that contained dissociated islet cells. Dark-brown DAB chromogen stain indicates positive staining for insulin (indicated by black arrows). Scale bar = 150 μm.

3. Conclusion

In summary, we present the novel and successful use of our injectable biomaterial, microporous annealed particle (MAP) scaffold, as a platform for delivery of dissociated islet cells. Specifically, we demonstrated that MAP scaffold maintains dissociated islet cell survival and function (i.e., glucose-sensing and insulin secretion) in vitro for at least 3 days and in vivo for up to 40 days post-transplant. The specific mechanisms by which MAP scaffold may support function of single cell suspensions of beta cells (e.g., immunomodulation, enhanced angiogenesis, cellular pathways, etc.) will be investigated in future studies. Importantly, this is the first study that demonstrates the feasibility of delivering dissociated islets within MAP scaffold to re-establish normoglycemia in a model of T1D. These findings may lead to translational approaches that will address the scarcity of donor pancreatic tissue needed to treat patients suffering from T1D. In future studies, we intend to use this work as the first step to MAP scaffold-assisted delivery of clinically scalable sources of insulin-secreting cells (e.g., iPSC-derived beta cells).

4. Supplemental Experimental Section/Methods

Microgel production and purification:

The no-heparin 3.2 wt% (w/v) microgel precursor solution consisted of PEG-Maleimide (10 kDa, Nippon Oil Foundry, Japan), MMP-2 degradable crosslinker (Ac-GCGPQGIAGQDGCG-NH2, Watson Bio), RGD (Ac-RGDSPGGC-NH2, Watson Bio) and MethMal^[30] annealing macromer. The 2.2 wt% 6 mg/mL heparin microgel precursor solution contained thiolated heparin which was prepared as previously described^[27]. Microgels were synthesized using a high-throughput microfluidics technique as previously described^[31]. The aqueous phase was run at 3 mL/hr and the surfactant solution (2% Pico-Surf in NOVEC 7500) was run at 6 mL/hr through the microfluidic device, and microgels were collected in a 50 mL conical tube. An oil solution with 3% vol/vol triethylamine was added to the microgel suspension to increase the pH and initiate microgel crosslinking. After complete gelation, microgel purification and sterilization was performed as previously described. For animal studies, heparin-containing microgels were mixed in with the no-heparin microgels at a 1:10 ratio (10% heparin-islands). Prior to annealing, microgels were mixed 1:1 with a 0.2 mM lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) solution for cell studies or mixed 1:1 with a 40μM Eosin Y solution for animal studies.

Scaffold pore sizing:

Heparin-containing microgels (labeled with TexasRed fluorophore) were mixed in with the no-heparin microgels at a 1:10 ratio (10% heparin-μislands) and incubated with 0.2 mM LAP for 15 minutes. Next, three 5 μL pucks of gel were added to a glass slide with a 1 mm-thick spacer and annealed with 365 nm LED light for 30 seconds. A 300 μg/mL dextran solution (Oregon Green 488, 70 kDa) was added on top of each gel puck and allowed to diffuse into the pores for at least 30 minutes prior to imaging. A Zeiss 710 Laser Scanning Microscope was used in 2-Photon mode to image at least a 200 μm z-stack (1 μm or 10 μm step-size) through each scaffold (n=3). The z-stacks from three different scaffolds were then exported to Imaris Microscopy Image Analysis software (Oxford Instruments) to produce 3D rendering of the pores (green dextran) and heparin-μislands particles (red) (Figure 1B). Imaris software was also used to threshold the area of the green dextran (pores) in each slice and the median pore area was calculated.

Isolation and dissociation of pancreatic islets:

Islets were isolated by collagenase digestion and density gradient centrifugation. Under anesthesia, the pancreas was exposed by midline laparotomy, the common bile duct identified and cannulated, and the pancreas distended [collagenase P, 2.5 ml, 2 mg/ml in Hank’s balanced salt solution (HBSS); Roche Molecular Biochemicals, Indianapolis, IN] in accordance with the University of Virginia Animal Care and Use Committee Protocol No. 3347. The distended pancreas was excised, digested (13 min, 37°C), and shaken vigorously to mechanically disrupt the tissue (30 seconds). Digestion was quenched [excess cold HBSS with 10% fetal calf serum (FCS); Invitrogen, Grand Island, NY], and the digested tissue filtered through a nylon mesh (1,000 μm, Nitex; Fisher Scientific) and washed (3x, HBSS, 0°C). Islets were purified by density gradient centrifugation using Histopaque 1077 (Sigma-Aldrich, St. Louis, MO). Islets were washed and cultured (~18 h, 37°C, 5% CO₂) in DMEM (Invitrogen) supplemented with 10% FCS, 2 mM l-glutamine, 1% penicillin/streptomycin, 0.1 mM MEM nonessential amino acids, and 25 mM HEPES buffer (all supplements from Invitrogen; 20 mL per plate). Islets were then hand-picked and transferred to sterile Eppendorf tubes and washed with PBS twice following which islets were resuspended in 300 μL of 0.05% Trypsin-EDTA (Gibco) and incubated in a 37°C waterbath for 10 minutes. The tube was gently tapped every 2 min to mechanically dissociate cells. 300 μL of DMEM media with 10% FCS was added to the cells to neutralize the trypsin-EDTA. The cell suspension was filtered with a 35 μm cell strainer to isolate single cells and remove residual extracellular matrix. Dissociated cells were centrifuged at 0.2(×1000)g for 5 min and supernatant carefully removed. The cell pellet was resuspended with fresh DMEM media.

Viability assay with LAP photoinitiator:

Cell survival was determined using fluorescein diacetate (FDA, Sigma) and propidium iodide (PI, Sigma) live/dead stains. For each condition, 50 islets were dissociated with 0.05% trypsin-EDTA, filtered, and the resulting cell suspension was mixed with 80 μL of MAP gel or 80 μL of media and transferred to a 24 transwell plate insert. MAP scaffolds were annealed with 365 nm LED light (ThorLabs, 20 mW/cm² intensity) for 30 seconds. Wells received 1 mL of DMEM media and the plate was incubated at 37°C in between timepoints. At each timepoint (1, 24, 48, and 72 hrs), the transwell inserts (n=3 per timepoint) were removed from media and placed into 1 mL of PBS containing 50 ng/mL of FDA and 14.5 μg/mL of PI. After incubation for 30 seconds with gentle shaking at room temperature, the transwell was removed and excess moisture was removed by gently blotting bottom of membrane with a Kimwipe. Wells were imaged with 4X objective on an EVOS FL Auto microscope. Images were imported into ImageJ, and the red (dead) and green (live) channels were thresholded and cells were counted using the ImageJ Particle Analysis plugin. Cell viability is reported as [(sum of live cells) / (sum of live + dead cells)]×100%.

Viability assay with Eosin Y photoinitiator:

Due the Eosin Y autofluorescence in the green channel, cell survival was determined using ReadyProbes cell viability imaging kit, blue/red (Invitrogen). For each condition, 50 islets were dissociated with 0.05% trypsin-EDTA, filtered, and the resulting cell suspension was mixed with 80 μL of MAP gel or 80 μL of media and transferred to a 24 transwell plate insert. MAP scaffolds were annealed with 505 nm LED light (ThorLabs, 104 mW/cm² intensity) for 2 minutes. Wells received 1 mL of DMEM media and the plate was incubated at 37°C in between timepoints. After a 24-hour incubation, 2 drops of NucBlue Live reagent (Hoechst 33342) and 2 drops of propidium iodide were added directly to the well. After incubation for 10 minutes at room temperature, the transwells were imaged with 20X objective (1.4 mm correction collar) on a Molecular Devices ImageXpress Micro confocal microscope. Images were imported into ImageJ, and the red (dead) and blue (live) channels were thresholded and cells were counted using the ImageJ Particle Analysis plugin. Cell viability is reported as [(sum of live cells) / (sum of live + dead cells)]×100%.

Immunohistochemistry:

Kidneys containing the graft were excised at the study endpoint and were immediately fixed in formalin. Prior to paraffin embedding, kidneys were cut in half and the half containing the graft was submitted to the University of Virginia Histology Core. Samples were embedded in paraffin and sectioned transversely with a microtome into sections with 5 μm thickness. Immunohistochemistry was performed by the University of Virginia Biorepository and Tissue Research Facility (BTRF). Recombinant Anti-Insulin antibody (rabbit monoclonal [EPR17359] to insulin) was purchased from Abcam (Cat# ab181547) and optimized by the BTRF. Positive-staining was confirmed by dark-brown DAB chromogen stain.

Statistical Analysis:

All statistical analysis was performed in GraphPad Prism software. For the pore size analysis, the scaffold pore surface area was acquired in Imaris software and converted to a circular pore diameter. A ROUT outlier analysis (Q = 1%) was performed on the data and outliers were removed. Then, the median pore diameter from each scaffold (n=3) was calculated in GraphPad Prism. The final median pore diameter value that is reported in this manuscript (36.7 ± 9.62 μm) is the mean of the median diameter values from each scaffold (n=3) and the standard deviation. There was no need for any pre-processing (i.e., outlier analysis, normalization, etc.) of the in vitro cell data. An ordinary two-way ANOVA with Sidak’s multiple comparison test (single pooled variance, alpha = 0.05) was used to compute differences between groups in the cell viability data (Fig. 2A, n=3 wells per timepoint), GSIS data (Fig. 2B, n=3 wells per timepoint), and photoinitiator comparison data (Fig. 2D, n=3 wells per group). An ordinary one-way ANOVA with Tukey’s multiple comparison test (alpha = 0.05) was used to compute differences between groups in the dissociated islet cell density viability experiment (Fig. 2C, n=3 wells per group).