Localized drug delivery graphene bioscaffolds for cotransplantation of islets and mesenchymal stem cells.

In the present work, we developed, characterized, and tested an implantable graphene bioscaffold which elutes dexamethasone (Dex) that can accommodate islets and adipose tissue–derived mesenchymal stem cells (AD-MSCs). In vitro studies demonstrated that islets in graphene–0.5 w/v% Dex bioscaffolds had a substantial higher viability and function compared to islets in graphene-alone bioscaffolds or islets cultured alone (P < 0.05). In vivo studies, in which bioscaffolds were transplanted into the epididymal fat pad of diabetic mice, demonstrated that, when islet:AD-MSC units were seeded into graphene–0.5 w/v% Dex bioscaffolds, this resulted in complete restoration of glycemic control immediately after transplantation with these islets also showing a faster response to glucose challenges (P < 0.05). Hence, this combination approach of using a graphene bioscaffold that can be functionalized for local delivery of Dex into the surrounding microenvironment, together with AD-MSC therapy, can significantly improve the survival and function of transplanted islets.

INTRODUCTION

Islet transplantation is a β cell replacement therapy used to treat patients with diabetes who lack the ability to secrete insulin (). Currently, there are two types of islet transplants performed: (i) autologous islet transplantation (a U.S. Food and Drug Administration–approved and reimbursable procedure that is rapidly becoming the “standard of care” for patients who require a total pancreatectomy for the treatment of chronic pancreatitis with intractable abdominal pain) and (ii) allogeneic islet transplantation (currently performed in many countries for patients with type 1 diabetes who have poor glycemic control and suffer from hypoglycemia unawareness). In both autologous and allogenic islet transplantation, up to 60% of islets are immediately lost following transplantation; a main reason for this is due to inflammation from the instant blood-mediated inflammatory reaction (IBMIR) (). One strategy to prevent the IBMIR is to not deliver islets into the liver via the portal vein but instead implant them using a biocompatible three-dimensional (3D) bioscaffold () at extravascular sites like the omentum or subcutaneous space (, ).

Bioscaffolds are rapidly gaining attraction for islet transplantation (–) given that they contain pores that can accommodate islets while simultaneously offering a unique interface that can be modulated to solve several problems currently faced by islets in the immediate posttransplantation period (such as inflammation) (). Bioscaffolds have predominantly been made from synthetic polymers, including polydimethylsiloxane (PDMS) (), poly(lactide-co-glycolide) (), poly(d,l-lactide-co-e-caprolactone) (), poly(ethylene oxide terephtalate)/poly(butylene terephthalate) block copolymer (), and heparin-binding peptide amphiphiles (). Despite promising preclinical data in providing immune isolation of islets (–), the clinical translation of many of these platforms has been limited because of (i) their inability to be reproducibly scaled, (ii) adverse biomaterial–soft tissue interface reactions that adversely affect islets especially for synthetic materials (), and (iii) difficulties in functionalizing the surface of certain bioscaffolds. However, one rapidly emerging material that can address many of these issues is graphene; this one-atom-thick fabric of carbon not only has extremely high mechanical strength with excellent flexibility () but also has excellent electronic and thermal conductivities, making it an attractive material for use in biological applications. Furthermore, its high chemical purity () allows for bioscaffolds to be constructed with low lattice defects () and excellent biocompatibility ().

While graphene will not induce an intense inflammatory response like synthetic materials, there will still be some degree of a host-mediated foreign body response (FBR) following its implantation, which is likely to adversely affect islets (). However, graphene can be easily functionalized () with anti-inflammatory drugs, like dexamethasone (Dex), thereby enabling it to act as a localized anti-inflammatory drug delivery platform. This approach will avoid the systemic side effects from conventional intravenous delivery of Dex while concurrently enabling it to precisely modulate the inflammatory milieu within the transplantation microenvironment. Dex is an immunosuppressive glucocorticoid that significantly reduces the activation of inflammatory pathways and has been shown to polarize monocytes toward the anti-inflammatory (M2) phenotype while still retaining their migratory function (). However, the dose of Dex needs to be carefully tailored as elevated levels can severely impair cell mobility resulting in compromised engraftment and vascularization (), as well as the glucose responsiveness of β cells in transplanted islets ().

In addition to being able to accommodate islets, our graphene bioscaffold will also be able to accommodate other supportive cellular therapies. Hence, we will take advantage of this by cotransplanting islets with mesenchymal stem cells derived from adipose tissue (AD-MSCs). AD-MSCs are able to secrete soluble factors into their surrounding microenvironment that can facilitate angiogenesis, reduce inflammation, as well as modulate both the innate and adaptive immune systems (), thereby helping islets engraft, survive, and function following transplantation. Studies have shown the beneficial effects of cotransplanting islets with MSCs with islets showing better structural organization, revascularization, and reduced peri- and intrainsulitis.

In the present study, we synthesized a 3D graphene bioscaffold using template-directed chemical vapor deposition (CVD) methodology and then functionalized it with a Dex coating via a polydopamine (PDA) nanolayer; the concentration of Dex was optimized to allow it to have an anti-inflammatory effect without affecting β cell function. Our graphene bioscaffold is made with interconnected macropores that are large enough to accommodate both islets and AD-MSCs evenly throughout its lattice; in turn, this will prevent cell clumping and ensure a more even distribution of islets, which will improve their survival and function ().

RESULTS

Bioscaffold synthesis and characterization

Graphene bioscaffolds fabricated by template directed CVD methodology (Fig. 1, A to G) retained a self-standing 3D porous structure (Fig. 1, H and I). By changing the dimensions of nickel foam templates, bioscaffolds with two different dimensions could be fabricated (Fig. 1, H to O): thin graphene bioscaffolds that had a 0.5 ± 0.1–mm thickness by 50 ± 2–mm length by 25 ± 1–mm width corresponding to a volume of 625 ± 185 mm3 with a porosity of 80 ± 5% and a density of 0.02 ± 0.005 mg/mm3 (Fig. 1H); and thick graphene bioscaffolds with 5 ± 0.5–mm thickness by 30 ± 2–mm length by 10 ± 1–mm width corresponding to a volume of 1500 ± 435 mm3 with a porosity of 80 ± 5% and a density of 0.020 ± 0.005 mg/mm3 (Fig. 1L). Scanning electron microscopy (SEM) observation showed that graphene bioscaffolds exhibited a continuous and porous monolith structure, which copied and inherited the interconnected 3D structure of the nickel foam template (Fig. 1, I to K and M to O). Using SEM, the pore size, measured in five different bioscaffolds at five random locations, fall into two groups: thin graphene bioscaffold: big pores = 400 ± 50 μm and small pores = 100 ± 10 μm (Fig. 1, I to K); thick graphene bioscaffold: big pores = 800 ± 50 μm and small pores = 200 ± 10 μm (Fig. 1, M to O). In our work, the thin graphene bioscaffold was selected for further in vitro and in vivo studies because of its pore size distribution (100 to 400 μm) matching the size of islets, which typically fall between 100 and 400 μm (), as well as islet:AD-MSC units. Furthermore, it can easily accommodate islets with the desired number for in vitro (20 islets in a 0.5-mm-thick by 1-mm-long by 1-mm-wide bioscaffold) and in vivo (250 islets in a 0.5-mm-thick by 5-mm-long by 5-mm-wide bioscaffold) experiments. Functionalizing bioscaffolds with a Dex coating slightly changed the bioscaffold porosity and density to 75 ± 5% and 0.03 ± 0.01 mg/mm3, respectively (P > 0.05). Using SEM, Dex microparticles, which were 6 ± 2 μm in size, were clearly detected immobilized onto the surface of graphene bioscaffolds (Fig. 1, K and O). The Raman spectrum demonstrated the successful synthesis of the graphene with corresponding signals estimated from the position of the D band (1415 cm−1), G band (1595 cm−1), and 2D band (2800 cm−1) (Fig. 1P). Using x-ray photoelectron spectroscopy, graphene bioscaffolds showed a peak corresponding to the element of carbon (C) at ~295 eV (Fig. 1Q). Using a two-probe conductivity method, the conductivity of graphene bioscaffolds was also found to be 40 ± 5 S cm−1.

Fig. 1.

Bioscaffold fabrication using template-directed CVD method and characterizations.

(A) Photograph and (B and C) scanning electron microscopy (SEM) image of a nickel foam in low (B) and high (C) magnification. (D to F) Photos of our CVD chamber before (D), during (E), and after (F) the graphene growth on the nickel foam (the arrows indicate the nickel foam placed inside the chamber). The nickel foam was first annealed at 1000°C for 5 min under Ar and H2 atmosphere with the flow rates of 500 and 200 standard cubic centimeters per minute (sccm), respectively, to clean its surfaces and eliminate the surface oxide layer. The CH4 gas with the flow rate of 50 sccm was then introduced into the reaction chamber. After 5 min of reaction-gas mixture flow, the foam was rapidly cooled down to room temperature under Ar and H2 atmosphere with the flow rates of 500 and 200 sccm, respectively. (G) Schematic representation showing graphene bioscaffolds before and after the transfer process of graphene bioscaffold from nickel-graphene foam. The nickel-graphene foam was first coated with a thin polymethyl methacrylate (PMMA) layer to support the graphene structure and prevent its structural failure when the nickel foams is etched away in the next step. Then, the nickel-graphene-PMMA foam was immersed into a mixture of FeCl3 and HCl at 80°C for 72 hours to dissolve the nickel. Last, graphene bioscaffolds were obtained by dissolving the PMMA with acetone. (H to O) Photos [(H) and (L); showing a self-standing 3D porous structure] and SEM images [(I) to (K) and (M) to (O); showing a monolith of continuous and porous structure, which copied and inherited the interconnected 3D structure of the nickel foam template] of thin (I to K) and thick (M to O) graphene bioscaffolds. SEM images also indicate the Dex microparticles that has been immobilized on the surface of graphene bioscaffolds (K and O). (P and Q) Bioscaffold physicochemical characterization include Raman spectroscopy (P) and x-ray photoelectron spectroscopy (Q). Photo credit: Mehdi Razavi, Stanford University.

Bioscaffold coating with Dex

Using SEM, Dex particles were found attached onto the surface of graphene bioscaffolds and uniformly distributed, with increases in the concentration of Dex from 0.25 to 1 w/v% resulting in more Dex particles seen (Fig. 2A). Graphene-Dex bioscaffolds can release Dex for at least 14 days with the release rate significantly increasing as the concentration of Dex increases from 0.25 to 1 w/v% (P < 0.05). In the first day, the Dex release rate reached 3.96 ± 0.21 ng/ml per hour for graphene–0.25 w/v% Dex bioscaffolds; this significantly increased to 4.53 ± 0.31 and 10.20 ± 0.14 ng/ml per hour for graphene–0.5 w/v% Dex and graphene–1 w/v% Dex bioscaffolds, respectively (P < 0.05). From days 1 to 14, the Dex release rate significantly decreased to 1.04 ± 0.85 for graphene–0.25 w/v% Dex bioscaffolds. This amount was insignificant compared to when 0.5 w/v% Dex and 1 w/v% Dex was used in bioscaffolds (day 14: 1.15 ± 0.05 and 1.84 ± 0.62 ng/ml per hour, respectively; Fig. 2B, P < 0.05).

Fig. 2.

Bioscaffold coating with Dex.

(A) SEM images from graphene-Dex bioscaffolds with increasing Dex concentrations from 0 to 1 w/v% showing Dex particles that have been attached onto the surface of graphene bioscaffolds and by increasing the concentration of Dex from 0.25 to 1 w/v%, more Dex particles are seen. (B) Dex release profile showing the ability of graphene-Dex bioscaffolds to release Dex for at least 14 days with the release rate significantly increasing as the concentration of Dex increases from 0.25 to 1 w/v% (P < 0.05). Significant differences: (B) aP < 0.05: graphene–0.25 w/v% Dex bioscaffolds versus graphene–0.5 w/v% Dex bioscaffolds or graphene–1 w/v% Dex bioscaffolds, bP < 0.05: graphene–0.5 w/v% Dex bioscaffolds versus graphene–1 w/v% Dex bioscaffolds; *Day 1 versus days 3, 7, and 14.

In vitro interaction of our bioscaffolds with islets ± AD-MSCs

Islets seeded in graphene–0.5 w/v% Dex bioscaffolds were uniformly dispersed as shown by their presence on the top and the center of the bioscaffold. In contrast, after 7 days, islets cultured alone in the control cell culture plates showed clumping (Fig. 3, A to C). Compared to the control, islets in graphene-Dex bioscaffolds had a significantly greater percentage of live cells (determined using the Live/Dead assay, P < 0.05; Fig. 3D), viability [determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, P < 0.05; Fig. 3E], and functionality [determined using a glucose-stimulated insulin secretion (GSIS) assay, P < 0.05; Fig. 3F]. Comparing graphene-Dex bioscaffolds, cell viability was greatest for islets in graphene–1 w/v% Dex bioscaffolds, where there was an increase in live cells (90 ± 4% versus 35 ± 10%, P < 0.05; Fig. 3, B and C), and cell viability (islet viability ratio: 2.0 ± 0.2 versus 1.0 ± 0.3, P < 0.05; Fig. 3D) compared to the control group. However, the greatest islet functionality was seen when islets were seeded into graphene–0.5 w/v% Dex bioscaffolds; here, islets secreted the highest level of insulin—low-glucose stimulation: 1.52 ± 0.38 ng/ml versus 0.89 ± 0.21 ng/ml and high-glucose stimulation: 4.28 ± 0.25 ng/ml versus 1.05 ± 0.16 ng/ml (P < 0.05; Fig. 3F)—compared to the control group. Calculation of the insulin stimulation index (SI; ratio of insulin secretion from high-glucose stimulation relative to basal conditions) also showed a significant increase when islets were seeded in graphene–0.5 w/v% Dex bioscaffolds compared to the control group (2.82 ± 0.04 versus 2.29 ± 0.06; P < 0.05; Fig. 3G). When islet:AD-MSC units were seeded into graphene–0.5 w/v% Dex bioscaffolds, there was a significant increase in islet viability (Live/Dead assay: 94 ± 3% versus 87 ± 3 and 35 ± 10%, P < 0.05; and MTT assay: 1.78 ± 0.03 fold change versus 1.72 ± 0.03 and 1 ± 0.01 fold change versus control, P < 0.05) and function (GSIS assay: 5.76 ± 0.47 ng/ml versus 4.28 ± 0.25 and 2.98 ± 0.23 ng/ml insulin secretion, P < 0.05; and insulin index: 5.51 ± 0.44 versus 2.82 ± 0.04 and 3.23 ± 0.25, P < 0.05; fig. S1) compared to islets alone in graphene–0.5 w/v% Dex bioscaffolds and islets alone (i.e., control islets).

Fig. 3.

Bioscaffold interactions with pancreatic islets in vitro.

(A) SEM images of the top surface, and center, of our graphene-Dex bioscaffold seeded with islets. (B) Bright-field images of islets cultured in conventional culture plates. (C) Confocal images of islets cultured in culture plates (islets only) or in graphene-alone bioscaffolds or graphene-Dex bioscaffolds with 0.25, 0.5, and 1 w/v% Dex at day 7. Green represents live cells and red represents dead cells. Results of (D) Live/Dead, (E) 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), and (F and G) GSIS assays for islets alone and islets in graphene bioscaffolds without, and with, Dex at day 7. Red, dead cells stained with propidium iodide. Green, live cells stained with fluorescein diacetate. Significant differences: (D to G) aP < 0.05: islets alone versus graphene and graphene–0.25, 0.5, and 1 w/v% Dex bioscaffolds; bP < 0.05: graphene bioscaffolds versus graphene–0.25, 0.5, and 1 w/v% Dex bioscaffolds; cP < 0.05: graphene–0.25 w/v% bioscaffolds versus graphene–0.5 and 1 w/v% Dex bioscaffolds; dP < 0.05: graphene–0.5 w/v% bioscaffolds versus graphene–1 w/v% Dex bioscaffolds; *Low glucose (LG) versus high glucose (HG).

In vivo interaction of our bioscaffolds with islets ± AD-MSCs

Experimental details of our in vivo experiments and transplantation procedure are outlined in Fig. 4A. Following streptozotocin (STZ) treatment, all animals became hyperglycemic [nonfasting blood glucose values increasing from 133 ± 8 mg/dl (day −2) to 502 ± 35 mg/dl (day 0)]. Following islet transplantation, immediate reestablishment of glycemic control, with nonfasting blood glucose values returning to baseline values, was only seen in mice that received islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds (i.e., 181 ± 32 mg/dl versus 380 ± 50 mg/dl compared to control animals receiving islets alone; P < 0.05; Fig. 4B), and this effect was maintained for 30 days. In contrast, diabetic mice, which received islets alone or islets in graphene bioscaffolds without Dex, were not able to restore glycemic control throughout the course of this study. Diabetic mice that received islets alone in 0.5 w/v% Dex bioscaffolds were not able to initially restore glycemic control; however, by day 14, their nonfasting blood glucose did return to baseline values, and this was maintained to the end of the study at day 30. This effect was also seen when comparing the percentage of animals that exhibited normoglycemia at each week during this study (Fig. 4C). At day 0, all mice have a similar weight of 18 ± 1 g. Following transplantation, the body weight of all mice increased; however, this increment was significantly higher for mice transplanted with islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds (P < 0.05; Fig. 4D).

Fig. 4.

Bioscaffold interactions with pancreatic islets in vivo.

(A) Experimental details of our in vivo experiment and schematic representation of our bioscaffold transplantation in the epididymal fat pad (EFP). Results of (B) nonfasting blood glucose measurements, (C) percentage of normoglycemia, (D) body weight, (E) IPGTT [i.e., changes of fasting blood glucose versus baseline (0-min time point)], and (F) results of calculation of area under the curve (AUC0–120min) of IPGTT curves (i.e., glucose clearance rate). (G) Photographs of the transplantation procedure of islets alone and islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds. Photo credit: Mehdi Razavi, Stanford University. Significant differences: (B to G) aP < 0.05: islets alone versus islets in graphene-alone bioscaffolds or islets in graphene–0.5 w/v% Dex bioscaffolds or islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; bP < 0.05: islets in graphene-alone bioscaffolds versus islets in graphene–0.5 w/v% Dex bioscaffolds or islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; cP < 0.05: islets in graphene–0.5 w/v% Dex bioscaffolds versus islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds. (B) *P < 0.05: baseline (day −2) versus all other time points [two-way ANOVA (analysis of variance), Tukey post hoc test]. (C and D) *P < 0.05: post-transplant week 0 versus post-transplant weeks 1, 2, 3, and 4 (two-way ANOVA, Tukey post hoc test). (E) *P < 0.05: 0 min versus 30, 60, 90, and 120 min (two-way ANOVA, Tukey post hoc test). (F) One-way ANOVA, Tukey post hoc test.

When assessing the dynamic responses to glucose challenges, mice transplanted with islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds demonstrated the smallest peak change in fasting blood glucose values at 30 min when compared to all other groups (P < 0.05; Fig. 4, E and F). At 120-min post-glucose load, the blood glucose value returned to normal range similar to baseline (0 min), indicating normal glucose tolerance. At sacrifice, graphene–0.5 w/v% Dex bioscaffolds were tightly wrapped within the epididymal fat pad (EFP), and, following their extraction, they were noted to be well engrafted within the fat tissue with no evidence of adhesions/fibrous bands (Fig. 4G).

When islets were transplanted into bioscaffolds, they could be easily identified within the bioscaffold pores (black arrows; Fig. 5A). Islets cotransplanted with AD-MSCs (i.e., islet:AD-MSC units) in graphene–0.5 w/v% Dex bioscaffolds were found on histological analysis to be significantly greater in number compared to animals that received islets alone in graphene–0.5 weight % (wt %) Dex bioscaffolds, graphene bioscaffolds alone, or islets only (total islet area: 0.86 ± 0.17 mm2 versus 0.55 ± 0.15 or 0.34 ± 0.09 or 0.09 ± 0.03 mm2, respectively, P < 0.05; Fig. 5A). In animals that were transplanted with islets alone into the EFP, there were few intact islets visualized; furthermore, these remaining islets had lost their spherical morphology and intrinsic architecture with insulin-staining cells now noted to be dispersed around the islet rather than being localized in discrete islet structures. In contrast, islets transplanted alone, or with AD-MSCs, in graphene–0.5 wt % Dex bioscaffolds retained their native size and spherical morphology and maintained their intrinsic architecture with β cells (positive insulin staining) located in the center of the islets (Fig. 5A). When transplanted islets are healthier, they are normally intact and have a spherical structure (). However, when islets start to die, they lose their shape because of a loss in plasma membrane integrity and cell death (). Here, our results support the cytoprotective effect provided by AD-MSCs and graphene-Dex bioscaffolds given that islet survival and function was improved; accordingly, islets had a more spherical and organized structure when compared to transplanted islets alone.

Fig. 5.

Histological and molecular analyses.

(A) Representative histological [hematoxylin and eosin (H&E) staining] and immunohistochemical images [insulin, von Willebrand factor (vWF), and tumor necrosis factor–α (TNF-α) staining] of the EFP containing islets alone or islets into graphene-alone and graphene–0.5 w/v% Dex bioscaffolds or islet:AD-MSCs units into graphene–0.5 w/v% Dex bioscaffolds. Red stars, islets; black arrows, bioscaffolds; blue arrows, positive (brown) staining. (B) Cytokine expression profile within the EFP tissue. The level of insulin within the (C) EFP and (D) blood serum. Quantification of positive (E) insulin, (F) vWF, and (G) TNF-α staining. Results were analyzed with at least 15 to 20 islets from five different sections through the EFP of each animal. Significant differences: (B) aP < 0.05: islets in graphene-alone bioscaffolds versus islets in graphene–0.5 w/v% Dex bioscaffolds or islets:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; bP < 0.05: islets in graphene–0.5 w/v% Dex bioscaffolds versus islets:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; *P < 0.05: islets alone (control) versus islets in graphene-alone bioscaffolds or islets in graphene–0.5 w/v% Dex bioscaffolds or islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds (one-way ANOVA, Tukey post hoc test). (C to G) aP < 0.05: islets alone versus islets in graphene-alone bioscaffolds or islets in graphene–0.5 w/v% Dex bioscaffolds or islet:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; bP < 0.05: islets in graphene-alone bioscaffolds versus islets in graphene–0.5 w/v% Dex bioscaffolds or islets:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds; cP < 0.05: islets in graphene–0.5 w/v% Dex bioscaffolds versus islets:AD-MSCs units in graphene–0.5 w/v% Dex bioscaffolds.

The greater degree of vascularity surrounding, and within, islets was further confirmed by hematoxylin and eosin (H&E) and von Willebrand factor (vWF) staining. Results of H&E staining showed significantly increased microvessel density for transplanted islet:AD-MSCs units in graphene–0.5 wt % Dex bioscaffolds compared to islets transplanted alone or in graphene-alone or graphene–0.5 wt % Dex bioscaffolds (95 ± 3 vessels/mm2 versus 11 ± 2, 35 ± 5, and 54 ± 6 vessels/mm2, respectively; Fig. 5A). In keeping with this, there was a significantly higher insulin staining in graphene–0.5 w/v% Dex bioscaffolds that contained islet:AD-MSC units compared to islets in graphene–0.5 w/v% Dex bioscaffolds, graphene-alone bioscaffolds, or islets transplanted alone without a bioscaffold (percentage of insulin expression per islet: 86.5 ± 10.2% versus 55.3 ± 6.2, 43.2 ± 4.5, or 33.5 ± 7.3%, respectively; P < 0.05; Fig. 5B). Islets transplanted with AD-MSCs, in graphene–0.5 w/v% Dex bioscaffolds, demonstrated increased vascularization as evidenced by an increase in vWF staining when compared to transplanted islets alone or islets seeded alone in graphene or graphene–0.5 w/v% Dex bioscaffolds (percentage of vWF expression per islet: 58.5 ± 6.2% versus 35.7 ± 7.2, 10 ± 2.5, or 8.5 ± 3.5% P < 0.05; Fig. 5C). Islets transplanted alone, or with AD-MSCs, in graphene–0.5 w/v% Dex bioscaffolds demonstrated reduced inflammation as evidenced by a reduction in tumor necrosis factor–α (TNF-α) staining when compared to transplanted islets alone (percentage of TNF-α expression per islet: 3.5 ± 1.2 or 4.3 ± 1.1% versus 18.2 ± 4.1%; P < 0.05; Fig. 5D). Correspondingly, the levels of insulin in the blood was significantly higher in mice that received islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds compared to mice transplanted with islets alone without a bioscaffold (1.53 ± 0.36 ng/ml versus 0.34 ± 0.11 ng/ml; P < 0.05). No significant difference was seen between islets transplanted in either graphene-alone bioscaffolds or graphene–0.5 w/v% Dex bioscaffolds (0.89 ± 0.17 ng/ml versus 1.02 ± 0.36 ng/ml, P > 0.05; Fig. 5E). Similar results were also seen for the levels of insulin in the EFP tissue (2.13 ± 0.45 μg/ml versus 0.50 ± 0.01 μg/ml; P < 0.05; Fig. 5F).

In addition, the EFP tissue from animals that had received islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds demonstrated up-regulation of macrophage inflammatory protein-1α (MIP-1α: +5.29 ± 0.58 fold change), EOTAXIN (+1.54 ± 0.39 fold change), interferon-α (IFN-α: +0.89 ± 0.01–fold change), and down-regulation of IFN-γ (−0.42 ± 0.09 fold change), LEPTIN (−0.34 ± 0.06 fold change), interleukin-27 (IL-27; −0.43 ± 0.09 fold change), IL-3 (−0.56 ± 0.18 fold change), IL-5 (−0.43 ± 0.09 fold change), TNF-α (−0.54 ± 0.14 fold change), IL-13 (−0.75 ± 0.15 fold change), IL-10 (−0.80 ± 0.23 fold change), granulocyte-macrophage colony-stimulating factor (−0.82 ± 0.29 fold change), and IL-15 (−0.93 ± 0.57 fold change) when compared to control animals that received islets alone (P < 0.05; Fig. 5F). Up-regulated expression of MIP-1α, EOTAXIN, macrophage colony-stimulating factor, vascular endothelial growth factor, and monocyte chemoattractant protein 1 (MCP-1), as well as down-regulation expression of IL-1β, IL-5, TNF-α, and IL-10 were noted within the EFP tissue of animals that had received islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds compared to islets seeded into graphene–0.5 w/v% Dex, or graphene-alone, bioscaffolds (P < 0.05; Fig. 5F).

Bioscaffold biodegradability and biocompatibility

After 12 weeks of incubation in phosphate-buffered saline (PBS), the biodegradation degree of graphene–0.5 w/v% Dex bioscaffolds was 5.40 ± 0.23% (Fig. 6A). At 3 months after implantation, graphene–0.5 w/v% Dex bioscaffolds were well integrated into the EFP or subcutaneous tissue (Fig. 6B). Histological images show minimal foreign body reaction as demonstrated by the formation of fibrous tissue (Fig. 6, C and D) and a minimal inflammatory response demonstrated by TNF-α staining, inside the bioscaffold pores and around the bioscaffolds (Fig. 6, D to F). However, blood analysis at 3 months demonstrated no significant elevation in any parameter with average values from all animals remaining within their respective normal ranges (Table 1) (, ).

Fig. 6.

Bioscaffold biodegradability and biocompatibility.

(A) Biodegradation profile and photographic of graphene–0.5 w/v% Dex bioscaffolds before and after incubation in PBS for 3 months. (B) Photographic and representative histological (H&E staining) of the (C) EFP and (D) subcutaneous tissue implanted with graphene–0.5 w/v% Dex bioscaffolds. Red arrow, bioscaffold; blue arrows, blood vessels (photographs and H&E staining images). Representative TNF-α staining of the (E) EFP and (F) subcutaneous tissue implanted with graphene–0.5 w/v% Dex bioscaffolds. Black arrow, bioscaffold; red arrows, positive (brown) staining. Photo credit: Mehdi Razavi, Stanford University.

Table 1.

Blood metabolic, chemistry, and liver panels from mice that had been implanted with graphene–0.5 w/v% Dex bioscaffolds in EFPs for 3 months.

The normal range for each parameter is also listed in the table.

Test	Mice implanted with graphene–0.5 w/v% Dex bioscaffolds	Normal range
	Metabolic panel
Sodium	151 ± 1	146–151 mM
Chloride	109 ± 2	107–111 mM
Carbon dioxide	18 ± 2	20–29 mM
Potassium	4.1 ± 0.2	3.0–9.6 mM
Blood urea nitrogen	24 ± 1	20.3–24.7 mg/dl
Creatinine	0.18 ± 0.02	0.1–1.1 mg/dl
	Chemistry panel
Calcium	9.7 ± 0.1	8.9–9.7 mg/dl
Phosphorous	8.3 ± 0.4	4.2–8.5 mg/dl
T.Protein	5.2 ± 0.1	4.5–6.5 g/dl
Albumin	3 ± 0.1	2.5–2.8 g/dl
	Liver panel
Alkaline phosphatase	65 ± 1	44–147 IU/liter
Aspartateaminotransferase (AST)	34 ± 5	10–45 U/liter
Alanineaminotransferase (ALT)	26 ± 2	10–35 U/liter
Total bilirubin	0.3 ± 0.1	0–1.0 mg/dl

DISCUSSION

Recently, graphene bioscaffolds have gained interest due to their larger interconnected macropores (i.e., super-macropores) and enhanced mechanical stability compared to traditional hydrogel and polymeric constructs (). We therefore fabricated a graphene bioscaffold via a template-directed CVD method that involves deposition of methane gas on a sacrificial nickel template where the porous structure of this template dictates the final architecture of the bioscaffold. When islets were seeded into these graphene bioscaffolds, the large pores (400 ± 50 μm) enabled islets to be evenly distributed throughout its 3D matrix. In turn, this prevented islets from clumping, resulting in their improved function and survival, as similarly observed in previous studies (). In addition, islets seeded within our bioscaffolds were able to maintain their native rounded morphology, size, and architecture, all of which have been shown to play a crucial role in their function and outcome following transplantation (). Studies have also shown that the small pores (100 ± 10 μm) of a bioscaffold, similar to the ones in our graphene bioscaffolds, allow mass transfer of metabolites and ingrowth of blood vessels resulting in an intraislet vascular density that is comparable to native islets (). This is supported by our in vitro (MTT, Live/Dead, and GSIS assays) and in vivo (metabolic analysis) data showing that bioscaffolds made from graphene improved islet survival and function when compared to islets alone.

It is well established that inflammation plays a key role in decreasing islet viability and graft efficacy following their transplantation (). Activation of the inflammatory cascade starts early, whereby proinflammatory mediators, such as tissue factor and high-mobility group box-1 protein, are up-regulated by stressed islets following their isolation from the pancreas. Conventionally, islets are transplanted by delivery into the portal vein, which results in an innate immune attack including activation of both the coagulation and complement systems; this causes a rapid binding of activated platelets to the surface of islets and infiltration of polymorphonuclear leukocytes in what is commonly referred to as the IBMIR. As a result, there is a significant loss of islets (∼60% of the islet mass) immediately following their transplantation. While implanting islets in nonendovascular locations can mitigate the IBMIR, islets ideally need a 3D matrix or scaffold to be housed in to ensure their survival; unfortunately, these can trigger an FBR (), which again causes inflammation that can adversely affect islets. One advantage of graphene bioscaffolds is their excellent biocompatibility given that the surface of graphene sp2 carbon allows for strong, nondestructive, interfacial interactions at a cellular level, which can be further improved by chemical functionalization (). While graphene bioscaffolds did not induce a significant FBR reaction following implantation in animals (as commonly seen with synthetic bioscaffolds like PDMS), our results did show that there was a minimal inflammatory response around bioscaffolds as demonstrated on histology by cellular infiltrates and some positive TNF-α staining. Unfortunately, this response was enough to likely damage the transplanted islets given that these animals remained hyperglycemic and were unable to reestablish glucose homeostasis. Hence, to address this issue of the host inflammatory response to our graphene bioscaffold, we decided to functionalize our bioscaffold with Dex because it is an immunosuppressive glucocorticoid capable suppressing inflammatory pathways (), as well as inhibiting the expression of many inflammatory mediators (). This approach will enable us to precisely modulate the local microenvironment following bioscaffold implantation, thereby mitigating the need for intravenous administration of Dex which is often associated with significant systemic side effects, which include fluid retention (swelling in hands or ankles), headache, dizziness, increased blood pressure, and slow wound healing (, ).

To allow our graphene bioscaffold to be functionalized with Dex, we first applied a PDA coating to our bioscaffold; this served as the interface to enable our bioscaffold to then be coated with Dex given its strong adsorption properties through covalent bonding and intermolecular interactions (, ). Although Dex is a potent anti-inflammatory and immunosuppressive drug, at high doses, it reduces insulin secretion by islets by making them insensitive to glucose (–). Studies have attributed this to Dex reducing the protein expression of the glucose transporter type 2 (GLUT2) transporter in β cells (), which then attenuates the Km glucose transport into β cells resulting in an absent insulin response to glucose (, ). Hence, characterization of the release kinetics of Dex from graphene bioscaffolds is of paramount importance, and in our study, we found that the cumulative release profile of Dex (100 to 400 ng/ml over 14 days) was related to both the loading dose of the drug and time, which is typical of Dex delivery platforms (–). Although increasing the concentration of Dex on our graphene bioscaffolds was able to increase the amount of Dex released, this did not correlate with improved islet survival and function. We found that incorporation of 0.5 w/v% Dex onto the surface of our graphene bioscaffolds resulted in the highest degree of islet survival (measured by an MTT and Live/Dead assay) and function (measured by GSIS assay) when compared to graphene bioscaffolds incorporated with higher concentrations of Dex (i.e., 1 w/v% Dex). This was then validated in vivo in diabetic animals with graphene–0.5 w/v% Dex bioscaffolds loaded with islets reversing hyperglycemia and restoring glycemic control during basal conditions after 14 days, while also improving dynamic responses to glucose challenges. In addition, the EFP at day 30 from these mice contained a significantly higher amount of insulin compared to mice transplanted with islets alone or islets in graphene-alone bioscaffolds, thereby indicating that there was a higher amount of viable and functional islets within these animals. The functionalization of Dex on our graphene bioscaffolds also resulted in a greater reduction in proinflammatory cytokines, such as IL-1β [which is a well-established inducer of both insulin resistance and impaired pancreatic islet function ()] and IL-18 [which induces IFN-γ () that then drives programmed death-ligand 1] expression on islet β cells ()) within the EFP. The lower efficiency of islet transplantation in islets alone transplanted into the EFP compared to previous studies (–) can be due to (i) the use of 500 islets with islet diameters in the range of 50 to 150 μm—this is less than 500 islet equivalents or IEQ (~ 150 μm) used in previous studies (–)—and (ii) higher initial blood glucose of our animals at the transplantation day (i.e., 502 ± 35 mg/dl) compared to previous studies (i.e., 250 to 500 mg/dl) (–)—this would mean that our animals had a higher degree of dysglycemia and were more dependent on the success of their islet transplant. While graphene–0.5 w/v% Dex bioscaffolds were eventually able to restore hyperglycemia, they did so only after 14 days, which resulted in diabetic animals for the first 2 weeks following transplantation not being able to regain glucose homeostasis. Hence, to improve islet engraftment during this time, we added a supportive cellular therapy in the form of AD-MSCs; these cells have been shown to enhance the survival and function of transplanted islets by promoting their revascularization as well as suppressing any localized inflammatory responses. For AD-MSCs to be effective, they need to be in close proximity to islets to first sample the surrounding microenvironment and then modulate it to facilitate islet engraftment by releasing the appropriate paracrine factors into the same local microenvironment. Hence, to ensure that AD-MSCs were “spatially located” next to islets at the time of transplantation, we first coculture them together to create islet:AD-MSC units. These islet:AD-MSC units were then seeded in graphene–0.5 w/v% Dex bioscaffolds that resulted in animals having an almost immediate response with glycemic control being reestablished for 24 hours and then sustained for 30 days. Together, these results would suggest that graphene–0.5 w/v% Dex bioscaffolds are essential for islet survival and that the initial engraftment and function of islets can be improved when they are seeded into these bioscaffolds as islet:AD-MSC units compared to when they seeded as islets alone. Significant differences between islets and islet:AD-MSC units were noted in animals when assessing glycemic control and the percentage of animals that became normoglycemic over the first 14 days as well as in the long term as shown by animal body weight over 30 days. While there was also the suggestion that islets within islet:AD-MSC units demonstrated better dynamic responses to glucose challenges, healthier morphology, and higher levels of serum insulin, these changes were not significant. One explanation for this was that the intraperitoneal glucose tolerance test (IPGTT) was done at 14 days and the molecular assessment of the transplants were done at 30 days (i.e., the time of animal sacrifice and tissue harvesting), and by this time, the islets in both groups were likely fully established and engrafted (as demonstrated in their basal glycemic control) and that any differences may have been more significant at earlier time points. Transplanted islets also demonstrated reduced inflammation shown by a reduced TNF-α expression on histology as well as down-regulation of proinflammatory cytokines IL-1β (), IL-10 (), IL-5 (), and TNF-α () in the tissue lysate of the islet transplant. Together, these results show the ability of this approach to help islet engraftment at the site of transplantation by lowering the islet inflammation.

The current clinical approach for islet transplantation is to infuse islets into the liver; however, this renders them irretrievable () as well as exposing them to adverse conditions such as hypoxia and inflammation. Given these challenges associated with the liver, other extravascular sites are now being evaluated such as the omentum or subcutaneous space; however, for islets to be transplanted and survive in these areas, they need a supportive structure such as a bioscaffold. While the subcutaneous space can be easily accessed for bioscaffold implantation, the omentum also has several promising attributes, which include its large well-vascularized surface area (). Both sites also have the ability to accommodate 3D bioscaffolds (, , ), as well as the ability to retrieve these platforms should something adverse happen. Furthermore, human clinical trials are already under way examining the feasibility of the omentum as a site for PDMS-based bioscaffolds for islet transplantation (). In the present study, we therefore examined the biocompatibility of our graphene-Dex in both the EFP [which, in mice, is used as a surrogate site for the omentum (, , )] and the subcutaneous space. At both sites, our graphene bioscaffold showed that they were well integrated into the surrounding tissues with minimal, if any, evidence of ongoing/active inflammation or fibrosis; this is essential to ensure the long-term safety and integrity of our implants (). In addition, all blood biochemistry and electrolyte markers were within their normal range with no perturbations noted.

Given that our data show that our graphene-Dex bioscaffold can augment islet survival and function with excellent biocompatibility, future research will aim to examine using this platform at both these sites with human islets in appropriate animal models over the short and long term. In humans, islets are typically given at 5000 islets/kg (i.e., a weight-based approach) which, on average, translates to approximately 350,000 to 450,000 islets (). The thin graphene bioscaffolds (0.5-mm thickness by 5-mm length by 5-mm width) that were used in our experiments can accommodate up to 500 islets. However, using our established CVD protocol, we can easily manufacture bigger graphene bioscaffolds, thereby easily ensuring the clinical use of graphene bioscaffolds with a volume > 100 cm3, which we have calculated can accommodate the required number of islets for human transplantation. Given that islets will also require protection against inflammatory reactions triggered by the FBR to the bioscaffold itself, combining our graphene bioscaffolds with Dex can address this issue to ensure that islets have the optimal microenvironment following their transplantation. While graphene has recently been used for drug delivery systems, and ultrasensitive electrochemical biosensors for detection of biomolecules, this is the first study to show its potential use as a bioscaffold for a cellular therapy (i.e., pancreatic islet transplantation). Hence, graphene is well placed to be translated into human clinical trials following further optimization and validation.

The in vitro biodegradability rate was low such that bioscaffolds would be expected to provide a stable space to accommodate islets until engraftment. However, if bioscaffolds induce a strong inflammatory response, then this will result in rapid biodegradation and loss of mechanical and structural integrity. Hence, to alleviate this effect, we can also enhance the biodegradability of our bioscaffolds by reducing the methane gas concentration in the CVD reactor during graphene synthesis or using a shorter growth time. Nevertheless, the in vitro biodegradability results need to be interpreted with caution as they do not completely predict the in vivo biodegradability (, ). The in vivo biodegradation rate will be strongly dependent on the surrounding blood flow, bioscaffold implantation site, oxygen supply, pH values, and the amount of water and ion content in the surrounding microenvironment (). Hence, although we expect >5-year stability of our bioscaffolds, a long-term in vivo biodegradability study will need to be performed to fully assess this.

In this study, syngeneic islet transplantation has been studied in which mice from the same strain have been used as both the donor and the recipient. This model allows us to evaluate the innate immune response toward the bioscaffold and, more specifically, the FBR toward the biomaterial from which our bioscaffold is constructed. This approach is also very pertinent to the autologous islet transplant setting. In contrast, for allogenic islet transplants, we will need to use a different model in which we have an major histocompatibility complex (MHC) mismatch, thereby allowing us to evaluate and focus more on the adaptive immune response. Future studies will therefore evaluate this setting as well as the ability of our bioscaffold to support human islet cargo in nude or immunocompromised mice models. Future work will also examine further optimization of our graphene–0.5 w/v% Dex bioscaffold with the incorporation of extracellular matrix molecules or growth factors in animal models that can accommodate human islets (i.e., NSG mice). We will also focus on the long-term outcome (e.g., >6 months) from both the biological (i.e., maintenance of basal and dynamic glycemic control) and biocompatibility perspectives using our graphene–0.5 w/v% Dex bioscaffold for islet transplantation. This bioscaffold can be produced in accordance with good manufacturing practice (GMP) guidelines that will also facilitate its clinical translatability. Although not performed in the current study, mechanical tests can be performed in the future to determine the tensile strength and percentage of elongation of both thin and thick bioscaffolds. However, our graphene bioscaffolds (both thin and thick) demonstrated adequate flexibility for in vitro culture experiments and in vivo implantation studies. If any wear or fracture of our bioscaffolds is encountered in vivo, then graphene-polymer composite bioscaffolds can also be developed in the future; however, the changes in their biological characteristics will need to be studied. In summary, this combination approach of using a graphene bioscaffold that can be functionalized for localized Dex delivery for 3 weeks, together with AD-MSC therapy, can significantly improve the survival and function of transplanted islets resulting in a rapid and stable restoration glycemic control in diabetic animals following their transplantation.

MATERIALS AND METHODS

Bioscaffold fabrication and characterization

Graphene bioscaffolds were fabricated using template-directed CVD as previously described (). Briefly, a nickel foam (Goodfellow Inc., USA) with a density of 0.45 g cm−3, 95% porosity, and 40-mm length, 30-mm width, and 1.6-mm thickness (thin) or 6-mm thickness (thick) was used as the template for the CVD growth of graphene. The thin or thick nickel foam was placed in a CVD furnace (Aixtron Black Magic, USA) dedicated to graphene synthesis with methane (CH4) and hydrogen (H2) available as the process gases. The foam was first annealed at 1000°C for 5 min under argon (Ar) and H2 atmosphere with the flow rates of 500 and 200 standard cubic centimeters per minute (sccm), respectively to clean its surfaces and eliminate the surface oxide layer. The CH4 gas, with the flow rate of 50 sccm, was then introduced into the reaction chamber. After 5 min of reaction-gas mixture flow, the foam was rapidly cooled down to room temperature under Ar and H2 atmosphere with flow rates of 500 and 200 sccm, respectively. Following graphene growth, the nickel-graphene foams were coated with a thin polymethyl methacrylate (PMMA) layer to support the graphene structure and prevent its structural failure when the nickel foams are etched away in the next step. The nickel-graphene foams were therefore dip-coated with PMMA (average molecular weight of 996,000; Thomas Scientific, USA) dissolved in ethyl l-lactate (4 wt %; Alfa Aesar, USA), and then baked at 180°C for 2 hours to obtain a nickel-graphene-PMMA foam. This process solidifies PMMA to form a thin film on the graphene surface. Next, the nickel-graphene-PMMA foam was immersed into a mixture of iron(III) chloride (FeCl3; Fisher Scientific, USA) and hydrochloric acid (HCl; Fisher Scientific, USA) with the ratio of FeCl3/HCl:1 M/1 M solution at 80°C for 72 hours to dissolve the nickel. Last, graphene bioscaffolds were obtained by dissolving the PMMA with acetone (Fig. 1, A to G). The graphene bioscaffolds were then washed three times with distilled water to ensure removal of all before being sectioned to obtain cubes measuring 0.5 mm in thickness by 5 mm in length by 5 mm in width (thin) or 3 mm in thickness by 5 mm in length by 5 mm in width (thick). The thin graphene bioscaffold was used in our in vitro and in vivo experiments. Next, microstructural, chemical, and physical characteristics of bioscaffolds were analyzed (see the Supplementary Materials).

Bioscaffold coating

Dex (Enzo Life Sciences, USA) was incorporated onto the surface of graphene bioscaffolds using a PDA coating as we previously reported (, ). PDA was used as an adhesive () and biocompatible () coating nanolayer to hold Dex onto bioscaffolds. Briefly, bioscaffolds were immersed in a dopamine solution (2 mg/ml in 10 mM tris, pH 8.5 in the dark). Dex was subsequently added to the dopamine solution at the desired percentage [0.25, 0.5, and 1 w/v%, which results in a total mass of 45, 90, and 180 mg of Dex per bioscaffold (0.5-mm thickness by 5-mm length by 5-mm width), respectively]. The mixture containing the graphene bioscaffolds, dopamine solution, and Dex was then placed on a tube rotisserie at 18 rpm for 30 min at 25°C to self-polymerize the dopamine. The graphene-Dex bioscaffolds were washed three times in PBS (Gibco, USA) to ensure removal of all chemicals. A Kimwipe was then used to wick away any residual water before leaving bioscaffolds to dry at room temperature for further analysis. To avoid any reaction of PDA with PBS, and Dex release, during the washing process, we minimized the process of washing graphene-Dex bioscaffolds with PBS (<15 s). Here, the graphene bioscaffolds coated with 0.25, 0.5, and 1 w/v% Dex were called graphene–0.25 w/v% Dex, graphene–0.5 w/v% Dex, and graphene–1 w/v% Dex bioscaffolds, respectively. To evaluate the release kinetic of Dex, bioscaffolds were incubated in benzalkonium chloride (BKC; 5 ml, 1 w/v%, Sigma-Aldrich, USA) for 14 days with samples collected at days 1, 2, 3, 7, 10, and 14, and the Dex concentration at each of these time points was measured using a Dex enzyme-linked immunosorbent assay (ELISA) kit (Neogen, USA). BKC was used as a solubilizing agent to promote a high sink condition and mimic the clearance of the steroid in vivo as previously reported for the measurement of Dex release from different materials (). Dex-free graphene bioscaffolds were used as the control group.

All assays were carried out on islets alone or islet:AD-MSC units, i.e., islets cocultured with AD-MSCs for 24 hours in a 1:100 ratio according to our previously published protocol (), manually picked under a microscope, and seeded directly into bioscaffolds (20 islets or 20 islet:AD-MSC units) or into a 96-well plate. For all in vitro experiments, isolated islets were individually counted and manually picked up under a microscope to achieve a density of 20 islets per well [20 islets in 200 μl of complete medium: RPMI 1640 medium (Gibco, USA)] supplemented with 10% fetal bovine serum (FBS; Invitrogen, USA) and penicillin (50 U/ml)/streptomycin (50 μg/ml) (P/S; Invitrogen, USA) were added into 96-well nonadherent tissue culture plates. All experiments were performed in n = 5 on the following experimental groups: (i) islets alone or islets in (ii) graphene-alone bioscaffolds, (iii) graphene–0.25 w/v% Dex bioscaffolds, (iv) graphene–0.5 w/v% Dex bioscaffolds, (v) graphene–1 w/v% Dex bioscaffolds, and (vi) islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds (i.e., our optimized bioscaffold configuration).

Islet isolation

Pancreatic islets were isolated from the pancreas of C57/B6 mice (female, 6 to 8 weeks old, Charles River Laboratories, USA), as we previously described (). Briefly, mice were anesthetized and then euthanized by cervical dislocation. The abdomen was opened, the bowel was moved to the left side, and the pancreas and common duct were then exposed. A hemostat clamp was placed on either side of the small intestine, and the pancreas was inflated through the bile duct with a 30-gauge needle and a 5-ml syringe containing 3 ml of cold collagenase solution. The pancreas was then removed from the body and placed in a vial containing 2 ml of collagenase. Digestion lasted for 10 min, and then the pellet was resuspended in Ficoll of different densities. The islet layer was identified, picked, and washed with Hanks’ balanced salt solution (Gibco, USA) supplemented with 0.1% bovine serum albumin (Gibco, USA). Islets were then individually counted and picked manually under a microscope.

AD-MSC isolation

AD-MSCs were isolated from the adipose tissue of C57BL6 mice (female, 6 to 8 weeks old, Charles River Laboratories, USA), as described in our previous works (, ). Briefly, mouse adipose tissue was obtained from the lower abdomen of mouse. Harvested adipose tissue was then washed with sterile PBS, minced with scissors, and then digested with type I collagenase (1 mg/ml) (Sigma-Aldrich, USA) in serum-free medium at 37°C for 3 hours. The digestion was then inactivated with an equal volume of complete medium. All samples were then filtered through a 100-μm mesh filter to remove any debris. The cellular pellets were collected and then resuspended in complete medium in a humidified incubator at 37°C with 5% carbon dioxide. AD-MSCs from passage numbers 3 to 5 were used in our studies.

Creating islet:AD-MSC units and bioscaffold seeding of cellular therapies

Islets were cultured in a complete medium containing the following: low-glucose Dulbecco’s modified Eagle’s medium (glucose concentration: 1 g/liter, Gibco, USA) supplemented with 10% FBS (Invitrogen, USA) and penicillin (50 U/ml)/streptomycin (50 μg/ml) in a humidified incubator under normal conditions [0.2 mM (20%) O2 and 5% CO2] at 37°C. The cultured AD-MSCs were enzymatically detached from culture plates upon reaching 70 to 80% confluence using 0.25% trypsin solution to obtain a single-cell suspension and resuspended in complete medium. AD-MSCs were then counted on a hemocytometer under a bright-field microscope and cocultured with AD-MSCs for 24 hours in a 1:100 ratio according to our previously published protocol (). Bioscaffolds were sterilized by soaking them in 70% ethanol for 0.5 hours after which time they were rinsed three times in sterilized PBS. Islets alone or islet:AD-MSC units (i.e., islets cocultured with AD-MSCs for 24 hours in a 1:100 ratio) were then hand-picked and seeded into sterilized bioscaffolds, achieving a density of 20 islets in 200 μl of complete medium per bioscaffold (cubes measuring 0.5-mm in thickness by 1 mm in length by 1 mm in width); these were then placed within each well of a 96-well plate.

Islet viability

Following procedures previously described in (), we determined the viability of islets using a MTT assay. Here, 50 μl of MTT solution (0.5 mg/ml) was added to the complete medium in each well and left to incubate at 37°C for 4 hours. Water-soluble MTT is taken up by viable cells and converted to an insoluble formazan. Next, 200 μl of dimethyl sulfoxide (to dissolve the formazan) was added to each well and left at 37°C for a further 10 min before the absorbance was measured at 570 nm using a microplate spectrophotometer system—the absorbance directly relates to the number of viable cells present (). Cell viability was determined using Eq. 1

ODsample is the optical density (absorbance) of islets, and ODcontrol is the optical density (absorbance) of islets that were not exposed to any bioscaffolds. To mitigate potential sources of errors caused by the effect of graphene or Dex on MTT readings, graphene-alone and graphene-Dex bioscaffolds without any MTT solution were used as background controls.

Islet function

Islets seeded in graphene-Dex bioscaffolds were cultured in a humidified incubator under normal conditions (37°C/5% CO2/20% O2). After 7 days of incubation, the ability of islets to secrete insulin was assessed by exposing them to low-glucose (i.e., basal conditions) and high-glucose (i.e., stimulated conditions) media (200 μl per well). Islets were incubated in Krebs-Ringer buffer (Sigma-Aldrich, USA) spiked with 2.8 mM glucose (low) for 2 hours followed by 16.7 mM glucose (high) for 2 hours at 37°C/5% CO2/20% O2 (). The supernatants were collected at the end of incubation for both basal and stimulated conditions, and insulin levels were quantified using a mouse insulin ELISA kit (Mercodia Developing Diagnostics, USA). The total insulin content of islets (i.e., 20 islets) was then normalized to present the amount of insulin secreted per islet.

Islet imaging

Following procedures previously described in (), we labeled islets using Hoechst 33342 (for live cells, Thermo Fisher Scientific, USA), propidium iodide (PI; for dead cells, Thermo Fisher Scientific, USA), and fluorescein diacetate (FDA; for AD-MSCs, Thermo Fisher Scientific, USA). The culture medium was removed under a bright-field microscope, and the staining solution [Hoechst 33342 (50 μl per well), PI (50 μl per well), and FDA (50 μl per well)] was added and then left to incubate in the dark at 5% CO2 and 37°C for 20 min. At the end of the incubation time, the staining solution was removed and cells were washed three times with PBS. The live cell imaging solution (Thermo Fisher Scientific, USA) was then added to each well before imaging. Images were acquired with a Zeiss LSM710 confocal microscope at a magnification of ×20, and figures were created with the FIJI software (ImageJ, GNU General Public License). Islets were further visualized with SEM by acquiring images from three to five random locations within each bioscaffold. For SEM imaging, sectioned bioscaffolds were washed three times with PBS, fixed using 4% paraformaldehyde for 0.5 hours at room temperature, and then dehydrated in graded ethanol solutions (50, 70, 90 and, lastly, 100% absolute ethanol). All bioscaffolds were then dried at room temperature, sectioned, sputter-coated with Au-Pd, and then analyzed with SEM.

Diabetes induction and islet transplantation

All procedures were performed in accordance with the regulations approved by the Institutional Animal Care and Use Committee of Stanford University. The study was conducted on a diabetic mouse model (C57BL/6, male, 6 to 8 weeks old, Charles River Laboratories, USA). To induce diabetes, each mouse received an intraperitoneal injection of STZ at the dose of 180 mg/kg; this technique is a well-established model for inducing diabetes in rodents and hence for studying islet transplantation (, , –) given that STZ selectively causes destruction of insulin-producing β cells within pancreatic islets (). Tail vein blood was collected to measure blood glucose concentrations with a portable glucose meter (Bayer Contour Glucose Meter, USA) for the next 5 days. Animals were determined as diabetic and ready for islet transplantation when they became hyperglycemic [nonfasting blood glucose >350 mg/dl on two consecutive days as previously documented ()]. These diabetic recipient mice were then prepared for surgery and anesthetized with 2% isoflurane in oxygen. The abdomen was shaved and swabbed with betadine and isopropanol to sterilize the skin. Next, a 7-mm incision was made through the peritoneal wall in the midline close to the genital area. The EFP on one side was gently grabbed and pulled out from the abdominal cavity using forceps. A bioscaffold piece (cubes measuring 0.5 mm in thickness by 2.5 mm in length by 2.5 mm in width) was placed on the unfolded EFP (all bioscaffolds were sterilized by incubating them in 70% ethanol for 30 min and washed three times under sterile conditions in PBS before they were transplanted). A total of 250 islets alone or islet:AD-MSC units mixed with 5 μl of Matrigel (Thermo Fisher Scientific, USA) were then seeded into each bioscaffold. The EFP was folded and sutured around the bioscaffold before being placed back into the abdominal cavity of the animal. This procedure was repeated for the contralateral EFP thereby achieving 500 islets alone or islet:AD-MSC units seeded into two bioscaffolds for each recipient animal. Control animals (i.e., islets only) also received 500 islets using a similar technique with 250 islets alone seeded into the EFP on each side of the animal; this amount has been previously shown to be subtherapeutic. At the end of the surgery, the abdomen was closed and animals were left to recover for 24 hours. A total of four experimental groups were used (n = 8 animals per group): group 1: mice transplanted with islets alone; group 2: mice transplanted with islets in graphene-alone bioscaffolds; group 3: mice transplanted with islets alone in graphene–0.5 w/v% Dex bioscaffolds; group 4: mice transplanted with islet:AD-MSC units in graphene–0.5 w/v% Dex bioscaffolds.

Blood glucose measurement

Blood glucose was monitored daily from the tail vein blood for 30 days after transplantation. Mice were considered normoglycemic when blood glucose levels were less than 200 mg/dl (). Time to reestablish normoglycemia was defined as the number of days required to reestablish blood glucose levels consistently lower than 200 mg/dl.

Intraperitoneal glucose tolerance tests

The function of the islets was examined further with IPGTT performed at day 14 after transplantation. Mice were fasted overnight before receiving an intraperitoneal glucose bolus (2 g/kg). Blood glucose levels were monitored at 0, 30, 60, 90, and 120 min after injection. The changes in blood glucose levels at 30-, 60-, 90-, and 120-min time points versus baseline (0-min time point) were presented. The slope of blood glucose changes versus time after injection and the area under the curve were calculated between transplant groups.

Histological analysis

Sections were prepared for histological and immunohistochemical analyses to determine islet structure and viability (H&E and insulin staining), evidence of vascularization (vWF antibody staining), and evidence of inflammation (TNF-α staining) via standard procedures. The stained sections were then imaged using a NanoZoomer slide scanner 2.0-RS (Hamamatsu, Japan). Results were analyzed using FIJI ImageJ software from five different sections through the EFP of each animal.

Molecular analysis

At day 30 after transplantation, mice were euthanized and serum and tissue (i.e., the EFP with or without graphene-Dex) samples collected to determine insulin levels (insulin ELISA kit; Mercodia). The frozen EFP tissue was then homogenized as follows: Tissue samples were placed in a homogenization buffer at a ratio of 1 kidney/1-ml buffer; the buffer contained a protease inhibitor combination (Sigma-Aldrich, USA) including 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (2 mM), aprotinin (0.3 μM), bestatin (116 μM), trans-epoxysuccinyl-l-leucylamido(4-guanidino)butane (E-64, 14 μM), leupeptin (1 μM), and EDTA (1 mM) in tissue protein extraction reagent (Thermo Fisher Scientific, USA) containing phenylmethylsulfonyl fluoride. All homogenized EFP samples were sonicated three times for a total of 8 s (Branson SLPe) mixed for 45 min at 4°C, before then being centrifuged at 15000 rpm (for 15 min at 4°C). The tissue supernatant was then collected and the insulin content measured (mouse insulin ELISA kit; Mercodia); results were normalized per EFP for each mouse. The level of tissue cytokines was also measured using a mouse multiplex ELISA (eBiosciences/Affymetrix/Fisher). Briefly, beads were first added to a 96-well plate and washed (Biotek ELx405). Samples were then added to the plate containing the mixed antibody-linked beads and incubated at room temperature for 1 hour followed by overnight incubation at 4°C on a plate shaker (500 rpm). Biotinylated detection antibody was then added, after which the plates were incubated at room temperature for 75 min on the plate shaker (500 rpm). Next, the samples were washed and streptavidin-phycoerythrin added followed by incubation of the plate 30 min at room temperature on the plate shaker (500 rpm). The plate was then washed and a reading buffer added to all the wells. Last, a Luminex Flex 3D instrument was used to read the plates with a lower bound of 50 beads per sample per cytokine. Control assay beads (Radix Biosolutions) were added to all wells.

Biodegradability

Graphene–0.5 w/v% Dex bioscaffolds (our optimized bioscaffold) were weighed [dry weight (Wd1)] and then incubated in PBS at 37°C for 3 months. Every week, bioscaffolds were removed from PBS, dried overnight, and reweighed [dry weight (Wd2)]. The degree of bioscaffold biodegradation was calculated as follows: [(Wd1 − Wd2)/Wd1 × 100] ().

Biocompatibility

Graphene–0.5 w/v% Dex bioscaffolds were implanted into the EFP and subcutaneous tissue of C57/B6 mice. After 3 months, mice were euthanized, and the EFP and subcutaneous tissue containing the implanted bioscaffolds were harvested for macroscopic and microscopic (i.e., histology with H&E and immunohistochemistry with TNF-α) examination, specifically looking at the bioscaffold and surrounding tissue. Blood samples were also collected for routine analysis (i.e., electrolyte, metabolic, chemistry, and liver panels) ().

Statistical analysis

All experiments were performed in n = 5 for in vitro or n = 8 for in vivo experiments, and all results were expressed as means ± sem. Statistical analysis of all quantitative data was performed using a one- or two-way ANOVA (analysis of variance) with Tukey post hoc test (Astatsa.com; Online Web Statistical Calculators, USA) with any differences considered statistically significant when P < 0.05.