A Novel Approach to Increase Human Islet Cell Mass While Preserving β-Cell Function

Human adult islets were provided through the Juvenile Diabetes Research Foundation Islet Distribution Program and our islet isolation facility at the University of California at San Diego. They were isolated with an automated method as described previously (8) and further purified by handpicking single islets (100- to 150-μm diameter) after dithizone staining (9).

Islets were kept under four tissue culture conditions: free floating in RPMI containing 10% fetal bovine serum and 11 mmol/l glucose with or without a cocktail of growth factors, as well as in fibrin gels with or without the growth factor cocktail. The growth factor cocktail included 20 ng/ml recombinant human HGF (HGF/SF; a generous gift from Genentech), kaposis fibroblast growth factor (hrFGF4, 1 ng/ml; Oncogene Research Products), and 10 mmol/l nicotinamide (Sigma). Fibrin gels were made with human fibrinogen (Sigma) dissolved in PBS (80 mg/ml) and human thrombin (Sigma) dissolved in 40 mmol/l CaCl2 (50 units/ml). After placing the islets in 5-μl droplets of the fibrinogen, an equal volume of the thrombin solution was immediately added to each drop. The liquid polymerized in ∼12 min to a soft gel. At that time, medium was added carefully to the dish. Medium and factors were replenished every 48 h.

Islets in each group were cultured for a total of 6 days. On day 5, after an overnight incubation in RPMI with the glucose concentration reduced to 5 mmol/l, each group was tested for insulin release in response to glucose by incubating at 1.6-mmol/l and 16.7-mmol/l glucose concentrations. On day 6, proliferation rate was assayed by ³H thymidine incorporation as previously described (10). DNA was measured by a fluorometric technique (11). Insulin content and release were measured using a coated antibody radioimmunoassay kit (DPC, Los Angeles) as previously described (10).

Five hundred islets were cultured alone or with fibrin and growth factors for 1 week, and the resulting cell population was then transplanted under the kidney capsule of athymic nude mice (as previously described) (12) under anesthesia (pentobarbital 50 mg/kg i.p.). This method uses a positive pressure pipette to gently implant the islets under the kidney capsule. Because this was technically difficult to accomplish, the gels were dissolved 24 h before transplantation with streptokinase (3 units/unit thrombin; Sigma). One month later, fasted animals were given 3 g/kg glucose i.p., and after 30 min, blood samples were taken for assay of circulating human C-peptide with a radioimmunoassay assay kit (DPC) that is specific for human C-peptide, with no cross-reaction from mouse C-peptide.

In some cultures, 1 mmol/l BrdU was added to the cultures for 24–72 h before harvesting. Islets were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained. Primary antibodies used were sheep anti-human insulin (The Binding Site), rabbit anti-human glucagon (Chemicon), mouse anti-human cytokeratin 4.62 (CK-19; Sigma), and mouse anti-BrdU (Dako). Secondary antibodies used were rhodamine red-conjugated donkey anti-sheep, fluorescein isothiocyanate-conjugated donkey anti-rabbit, or indo-dicarbocyanine-conjugated donkey anti-mouse IgGs (Jackson Immunoresearch). Control slides were incubated with a cocktail of the relevant control antibodies (mouse, sheep, and/or rabbit IgGs). For morphometric analysis, samples were sectioned completely and all sections examined. β-cell mass was quantified by counting the number of insulin-containing cells per intact islet section. Apoptotic nuclei were detected using the DeadEnd Fluorometric TUNEL system (Promega), which measures the fragmented DNA of apoptotic cells by catalytically incorporating fluorescein-12-dUTP(a) at 3′-OH DNA ends using the enzyme terminal deoxynucleotidyl transferase. After secondary staining for insulin, the fluorescein-12-dUTP-labeled DNA was visualized by fluorescence microscopy.

Grafts were removed and fixed in 4% paraformaldehyde and embedded in paraffin. Sections were stained for insulin, glucagon, and CK-19 as above.

Each experimental condition was tested on at least six different preparations of adult islets. Individual in vitro experiments used four replicate cell cultures for each parameter tested. Statistical significance of observed differences was analyzed by ANOVA and Fisher’s protected least significance difference test with 95% level as the limit of significance, or Student’s t test of paired differences using Statview IV (Abacus Concepts).

A few hours after plating in fibrin gels, islets appeared slightly flattened, as observed by phase-contrast microscopy (Fig. 1A). After 6 days, immnuohistochemical staining showed that the majority of the cells in the fibrin gel were insulin positive (Fig. 1B), and CK19-positive cells were seen only rarely (Fig. 1C). BrdU labeling was observed only in the islets that were in fibrin gels and exposed to growth factors (Fig. 1D–F); in this group, BrdU-positive cells were scattered throughout in insulin-containing (arrows) and non-insulin-containing (arrowhead) cells (Fig. 1F), but never in brightly staining insulin-positive cells, as we have previously shown (5). For morphometric analysis, only sections of intact islets were measured; for accuracy, islets that had fused in the fibrin were excluded from analysis. There was a threefold increase in number of β-cells per islet section in fibrin gels treated with growth factors as compared with free-floating islets (100 ± 37 vs. 38 ± 12, P < 0.0001). Single apoptotic nuclei were occasionally seen in intact free-floating islets (Fig. 1G), and many were seen in disintegrating free-floating islets (Fig. 1H), but none were observed in the fibrin-coated islets (Fig. 1I).

BrdU labeling data were supported by the quantitative assays for thymidine incorporation. After 6 days’ culture, proliferation was only observed in the group in which islets were cultured in fibrin gels in the presence of the growth factor cocktail (n = 4, P < 0.0001) (Fig. 2A); in this group, the total DNA also increased threefold (n = 4, P < 0.0001) (Fig. 2B). The islet cells were unable to proliferate in the presence of the growth factor cocktail alone. Although no proliferation was observed in the islets in fibrin without growth factors, as determined both by BrdU (Fig. 1E) and ³H thymidine incorporation (Fig. 2A), there was a significantly higher total DNA as compared with free-floating islets (n = 4, P < 0.005) (Fig. 2B).

Both insulin/DNA and total insulin were increased significantly after 6 days in fibrin gels (n = 4, P < 0.005; Fig. 2C and n = 4, P < 0.05; Fig. 2D, respectively) as compared with free-floating islets. Thus, the addition of fibrin alone is sufficient to increase insulin content. When the growth factors were added, the total insulin increased threefold over free-floating controls (n = 4, P < 0.0001) (Fig. 2D), which is in agreement with the morphometric data. These findings differ significantly from the situation when cells are grown in monolayer in the presence of growth factors with a concomitant loss of insulin expression (5). The rationale for the growth factors used was based on previous findings: 1) HGF/SF is mitotic to β-cells (2,6); 2) fibroblast growth factor-4 has been shown to induce β2/NeuroD in isolated mouse endoderm (13), and, in preliminary experiments, it preserved insulin content in HGF/SF-expanded islet cells (data not shown); and 3) nicotinamide increases insulin content in β-cells (6,10).

After 5 days in culture, all groups, except for the islets that were free floating in the presence of the growth factor cocktail, were stimulated more than twofold over baseline levels after a glucose challenge (Fig. 2E).

Macroscopically, grafts were larger in mice that had been transplanted with islets grown in fibrin gels than in the same number of islets cultured free floating. This observation was validated by immunostaining (Fig. 1J), which showed large compact grafts of endocrine cells (most stained for insulin or glucagon), and CK19 cells were rarely seen. For comparison, grafts from free-floating islets were smaller and showed scattered endocrine cells and also some ductal structures (Fig. 1K). After glucose challenge of fasted mice, circulating human C-peptide was significantly higher in mice receiving the graft from fibrin-treated islets than in the mice receiving a comparable number of free-floating cultured islets from the same islet preparation (n = 6, P < 0.005) (Table 1). In two experiments, we also transplanted islets within 24 h of isolation. Higher C-peptide levels were also observed when grafts from fibrin-treated islets were compared with grafts of freshly isolated islets from the same preparation transplanted without the 7-day culture period (data not shown).

3-D matrices such as collagen gel have previously been shown to promote survival and differentiation of islets in culture (14–19), but neither their effect on proliferation nor the mechanisms responsible for their beneficial effect have been studied. In this report, we show that a 3-D fibrin matrix support improves survival, reduces apoptosis, and allows for HGF-mediated proliferation of islet cells, leading to increased β-cell mass without loss of function.

Previous data have shown that islet proliferation needs the interaction in monolayer with matrices such as 804G or HTB-9 (2,3). Our findings in this report show that although growth factors are ineffective in islets that are free floating, cellular interaction in a 3-D configuration with a fibrin matrix allows proliferation to occur. Thus, avoiding cell growth in monolayer eliminates the need to disrupt cell matrix interactions and subsequent anoikis (7). Moreover, we show that glucose-responsive β-cell function is not impaired in proliferating islet cells in fibrin, which is in contrast to the loss of function seen previously in islets expanded in monolayer on matrices such as HTB-9 (5).

We and others (20,21) have shown that loss of islet cell mass through apoptosis occurs rapidly after islet isolation. The higher total insulin and DNA contents observed in the islets in fibrin without growth factors, as compared with free-floating islets, was concomitant with the lack of cell proliferation, as determined both by BrdU and ³H thymidine incorporation and the absence of apoptotic nuclei, and could be indicative of increased cell survival and/or suppression of cell death mediated by β-cell fibrin-serum interactions.

Fibrin gels have been shown to support cell survival, growth, and differentiation through the intermediary function of specific integrins, including α5β1, αvβ3, and αvβ1 (22–24) Based on preliminary studies using function-blocking antibodies, we have identified αvβ1 as an important candidate in the adhesion of adult β-cells to purified fibrin (data not shown). However, further studies are required to identify whether this integrin or others are ultimately responsible for supporting long-term β-cell survival or proliferation in fibrin gels supplemented with growth factors and serum.

The results of the islet transplants show that previous culture in fibrin gels ensures a robust graft that is capable of withstanding the deleterious conditions of engraftment. After glucose challenge, mice with grafts from fibrin-treated islets responded with higher C-peptide levels than mice with islets transplanted either within 24 h after isolation (as in the Edmonton protocol) or after 1 week in culture. Thus, the improvement in islet survival and function was due to the previous culture period in fibrin, and graft survival and function could perhaps be further improved by transplantation of the islets in fibrin, given that fibrin is also known to promote angiogenesis (25,26). However, dissolution of the gel by substances such as streptokinase, as described here, would facilitate the clinical use of islets for intraportal delivery.

In summary, fibrin improves survival of β-cells in vitro and reduces apoptosis, and addition of growth factors promotes proliferation and increased β-cell mass without loss of function. Thus, limited ex vivo islet expansion may improve recipient/donor tissue ratios to equal the functional results of whole-organ pancreas transplants.

This work was supported by Juvenile Diabetes Research Foundation grant nos. 1-2001-365 (to A.H.) and 1-2001-506 (to A.M.P.M.) and by grants from the Larry L. Hillblom Foundation and the Iacocca Foundation.

We thank the Surgical Medical Research Institute, University of Alberta, the Diabetes Research Institute (Miami, FL), the Islet Isolation Core Facility (St. Louis, MO), and the Pacific Northwest Research Center (Seattle, WA) for islet preparations.