The Edmonton Protocol for treatment of type 1 diabetes requires islets from two or more donors to achieve euglycemia in a single recipient, primarily because soon after portal infusion, the majority of the transplanted cells undergo apoptosis due to hypoxia and hypoxia reperfusion injury. X-linked inhibitor of apoptosis protein (XIAP) is a potent endogenous inhibitor of apoptosis that is capable of blocking the activation of multiple downstream caspases, and XIAP overexpression has previously been shown to enhance engraftment of a murine β-cell line. In this study, human islets transduced with a XIAP-expressing recombinant adenovirus were resistant to apoptosis and functionally recovered following in vitro stresses of hypoxia and hypoxia with reoxygenation (models reperfusion injury). Furthermore Ad-XIAP transduction dramatically reduced the number of human islets required to reverse hyperglycemia in chemically diabetic immunodeficient mice. These results suggest that by transiently overexpressing XIAP in the immediate posttransplant period, human islets from a single donor might be used to effectively treat two diabetic recipients.
The recent introduction of the Edmonton Protocol has demonstrated that islet transplantation is a viable route to achieve insulin independence in a population of patients with type 1 diabetes (1). Despite its promise, islet transplantation remains restricted to patients with severe hypoglycemia or glycemic lability and is currently unsuitable for the majority of patients with type 1 diabetes for several reasons. Most recipients require two or more islet transplant procedures (combined mass of >10,000 islet equivalents [IEQs]/kg body wt) in order to become insulin independent, which is a serious drawback given the prevalence of diabetes and the limited cadaveric organ donor pool (2, 3). Also, the risks associated with islet transplantation appear to increase with the number of infusions and with the total packed cell volume of cumulative grafts (4).
Expansion of clinical islet transplantation has been limited by the large requirement for donor tissue. The fact that most patients must receive >10,000 IEQs/kg to become insulin independent suggests that a large portion of the infused islets fail to engraft sufficiently. In fact, in murine models of islet transplantation, it has been determined that even under ideal circumstances, >60% of syngeneic islet graft mass is lost due to apoptosis (5). In clinical islet transplantation, it has been estimated that more than two-thirds of the implanted islets never become functional (2).
This early profound loss in islet mass can be attributed to several factors. Within a healthy pancreas, islet function is maximized by the intimate proximity of the β-cells and circulating blood, and, as a result, β-cells require a microenvironment with highly oxygenated blood (pO2 of 40 mmHg) and abundant nutrients (6). The current method of human islet isolation and purification destroys the capillary network in islets, causing the rapid onset of hypoxia (7). Islet hypoxia immediately after transplantation into the portal circulation further extends the postisolation hypoxic period (pO2 of 5–10 mmHg or <1% O2), and the revascularization process leads to reperfusion injury and death in islets (6). Thus, the majority of the islet graft rapidly fails to engraft after injection and undergoes apoptosis, which begins within hours posttransplant and continues for up to 2 weeks (8, 9). The immediate physiological burden faced by transplanted islets is also exacerbated by high levels of tissue factor expression in islets (10, 11). The Uppsala Group has demonstrated that this causes an instant blood-mediated inflammatory response to transplanted islets, with platelet deposition and subsequent macrophage-mediated islet destruction (10, 11).
Since the majority of the islet tissue is lost to apoptosis for the reasons listed above, intervention with antiapoptotic agents may substantially enhance preservation of islet mass following transplantation (5). Both extracellular and intracellular pathways of apoptosis have been implicated in β-cell death, and many studies have described inhibition of a variety of apoptosis-associated proteins, including cFLIP (prevents caspase-8 activation), A20 (inhibits NF-κB activation), Bcl-2, and Bcl-XL (mitochondria-associated antiapoptotic proteins) (12–18). In terms of preventing islet apoptosis and graft loss in the early posttransplant period, results thus far have been largely disappointing. Only A20 has been shown to moderately reduce the islet mass required for syngeneic islet transplantation (16, 17). Despite extensive investigation of these apoptosis prevention strategies in rodent models of islet transplantation, none of these antiapoptotic gene transfer strategies have been reported to be successful in enhancing human islet engraftment.
Taken collectively, these results suggest that the best strategy to enhance islet survival in the harsh posttransplant engraftment period would be to target the executioner caspases that function late in apoptosis, beyond the convergence point of both intracellular and extracellular signaling pathways. The inhibitor of apoptosis protein (IAP) family of apoptosis inhibitors includes a number of potent endogenous antiapoptotic genes, including X-linked IAP (XIAP), which binds to the active site of the main effector caspases 3, 7, and 9. Recent studies (19, 20) examining the effect of XIAP overexpression in islet transplantation have been very promising. When XIAP was overexpressed in human islets, islet death and loss of function was prevented following treatment with the immunosuppressive drugs tacrolimus, sirolimus, and mycophenolic acid in vitro (20). The protective effects of XIAP were further magnified in the Tet-regulatable murine β-cell line, βTC-Tet, following Ad-XIAP transduction (19). XIAP overexpression markedly enhanced β-cell survival and functional recovery during periods of stress due to hypoxia, reoxygenation, and cytokine insult in vitro (19). These results were confirmed in vivo by the finding that the mean time to achieve normoglycemia in recipient mice was decreased nearly sevenfold in XIAP-overexpressing grafts (19). Based upon these positive results, the impact of adenoviral-mediated XIAP overexpression in human islets during periods of hypoxia, reperfusion, and during posttransplant engraftment was investigated in the present study.
RESEARCH DESIGN AND METHODS
Human islet isolation.
Cadaveric donor pancreata were removed with prior informed written consent and stored in chilled University of Wisconsin solution or two-layer system before islet isolation. Islet isolation was performed as previously described for human islets (1, 21, 22). Immediately postisolation, islets were quantified with dithizone and assessed for viability with ethidium bromide/Cyto Green staining using established methods (Invitrogen, Burlington, ON, Canada) (23, 24). The islet preparations used were collected from the top fraction, ranged from 70 to 85% purity as measured by dithizone staining using a standard diameter of 150 μm, and were transduced in our lab <1 h following the completion of the isolation (23, 24). CMRL 1066 culture medium supplemented with calcium chloride (2.13 mmol/l), glutamine (2 mmol/l), HEPES (20 μmol/l), human serum albumin (0.625%), insulin (1.7 μmol/l), nicotinamide (10 mmol/l), selenium (8 nmol/l), sodium pyruvate (5 mmol/l), transferrin (5.5 mg/l), trasylol (20,000 IU), and zinc sulfate (16.7 μmol/l) (all from Sigma-Aldrich, Oakville, ON, Canada) was used for all experiments, unless otherwise noted.
Adenoviral transduction of human islets.
The first-generation recombinant adenoviral constructs used in these experiments have been previously described (25). Islets were transduced at 10 virus particles/cell (based on the average cellular content in a human islet being roughly 103 cells/IEQ), with Ad-XIAP or Ad-LACZ for 1–2 h in a volume of 1 ml CMRL culture medium/2,000 IEQs. The islet-virus mixtures were incubated for 1.5 h at 37°C, 5% CO2, after which additional culture medium (10-fold increase in volume) was added and incubation continued for another 24 h. Following transduction, the islets were washed twice with culture medium and incubated for at least 48 h before all in vitro assays or 7–10 days for transplant studies.
Analysis of adenoviral transgene expression: Ad-XIAP expression.
The presence of human XIAP protein (Hu-XIAP) was detected by immunoblot using anti-Hu-XIAP monoclonal antibody as previously described (Clone 48; BD Pharmingen, Mississauga, ON, Canada) (19). Ad-LACZ expression-β galactosidase (GAL) activity was assessed following 4 h of staining using the method outlined by Sanes et al. (26).
Adenovirus quantification in islet supernatants.
Aliquots of culture supernatant were harvested before adenovirus transduction and at days 2, 4, 6, 8, and 10 posttransduction. After each sample was collected, the islets were washed twice with PBS and resuspended in fresh culture medium. Samples of culture supernantants collected following Ad-LACZ transduction of human islets were serially diluted, transferred to murine A9 fibroblast cells, and allowed to incubate for 10 h. After an additional 24 h, the cells were stained for βGAL activity and quantified to determine the amount of infectious virus present in each supernatant sample.
In vitro hypoxia and reoxygenation studies.
Aliquots of 500 human islets that had been previously transduced with Ad-XIAP or Ad-LACZ were cultured in a modular incubator chamber (Billups-Rothenberg, DeMar, CA) at <1% oxygen conditions to mimic the hypoxic environment in the early posttransplant period (19, 27). After a period of 24 h within the chamber, the plates were removed and either analyzed immediately or cultured for an additional 24 h under normal nonhypoxic conditions (20% O2, 5% CO2, 37°C) to simulate reperfusion injury. Islets cultured in either hypoxic or hypoxic/reoxygenation environments were compared with matched transduced islets that had been cultured for the same time period under normal conditions.
Apoptosis in control Ad-LACZ- or Ad-XIAP-transduced human islets following exposure to hypoxia or hypoxia with reoxygenation was quantified using TdT-mediated dUTP nick-end labeling (TUNEL) staining. Control, hypoxic, or reoxygenated Ad-XIAP- or Ad-LACZ-transduced human islets were fixed in 4% paraformaldehyde, embedded in agar, processed, and embedded in paraffin (28). Ten-micrometer sections were stained for human insulin to identify β-cells using guinea pig anti-insulin (1:1,000; Dako, Carpinteria, CA) and labeled using phycoerythrin-conjugated anti-guinea pig IgG (Jackson Immunoresearch, West Grove, PA). Afterward, the apoptotic nuclei were labeled with fluorescein isothiocyanate-dUTP (Roche) using the TdT enzyme (Deadend Fluorometric TUNEL System; Promega, Madison, WI). The stained slides were analyzed using fluorescent microscopy. Each sample was stained in triplicate, and the number of TUNEL plus cells per islet cross-section was determined. At least 50 islets from each slide were examined.
Glucose-stimulated insulin release assays.
For each experimental condition, duplicate aliquots containing ∼500 islets were washed three times in low-glucose medium (Dulbecco’s modified Eagle’s medium containing 100 mg/dl d-glucose supplemented with 10% FCS) followed by gravity sedimentation. The media was then aspirated and replaced with either 1 ml low- or high-glucose medium (Dulbecco’s modified Eagle’s medium containing 450 mg/dl d-glucose supplemented with 10% FCS) and incubated for 2 h at 37°C, 5% CO2. Aliquots from the supernatant (250 μl) were analyzed in triplicate for human insulin content using a radioimmunoassay (Linco Research, St. Charles, MO). In each experimental condition, the fold stimulation was calculated by dividing the mean insulin released from islets cultured in high-glucose medium by the mean insulin released from islets cultured in low-glucose medium in parallel.
Immunodeficient nonobese diabetic (NOD)-RAG−/− mice (NOD.129S7[B6]-Rag1tm1Mom/J) were obtained from The Jackson Laboratory (Bar Harbor, ME) and housed under specific pathogen-free conditions. Ethical approval was obtained from the animal welfare committee at the University of Alberta, and all mice were cared for according to the guidelines of the Canadian Council on Animal Care.
Islet transplantation studies.
Diabetes was induced in 8- to 10-week-old male NOD-RAG−/− mice using a single intraperitoneal injection of streptozotocin (200 mg/kg; Sigma). Animals were considered to be diabetic after two consecutive blood glucose measurements ≥325 mg/dl. Grafts containing a mass of 1,000 or 600 islets were transplanted under the left kidney capsule in confirmed diabetic mice. In all transplantation studies, the nonfasting blood glucose levels were measured in each animal everyday for the 1st week to establish time to engraftment and on alternating days thereafter using a One-Touch Ultra Glucometer (Johnson & Johnson, New Brunswick, NJ). Islet grafts were considered to be functional when the nonfasting blood glucose returned to normoglycemic levels (<220 mg/dl). Euglycemic animals from each cohort were selected randomly, and the graft-bearing kidney was removed to establish that the human islet graft was functional, as determined by a return to hyperglycemia postnephrectomy.
Human C-peptide assays.
To determine the change in nonfasting serum human C-peptide levels following islet transplantation, serum was collected from posttransplant animals via tail vein bleeds each week. The serum human C-peptide levels for individual mice over time were quantified using an ultrasensitive human C-peptide enzyme-linked immunosorbent assay kit (Alpco Diagnostics, Windham, NH), and each serum sample was assayed in triplicate.
SigmaPlot 2000 (SPSS, Chicago, IL) was used for all statistical analyses in this study. Results are expressed as means ± SE. To compare results from in vitro hypoxia and hypoxia with reoxygenation experiments, Student’s t tests were used.
Expression of Ad-XIAP and Ad-LACZ in human islets.
To study the effect of XIAP overexpression on β-cell survival, human islets were transduced with an adenoviral vector containing the Hu-XIAP cDNA (Ad-XIAP) or a control vector containing the βGAL cDNA (Ad-LACZ). In both cases, the transgene was driven by the chicken β-actin promoter, which gives stable high levels of gene expression (25). Immunoblot analysis of Ad-XIAP-transduced human islets revealed that following transduction, the islets expressed high levels of the recombinant human XIAP protein, which varied slightly among different preparations of islets (Fig. 1A). Since the Hu-XIAP gene was overexpressed in human islets, we were unable to quantify the number of Hu-XIAP-expressing islets by immunohistochemistry without generating a low background staining of endogenous XIAP protein. However, when a similarly grown and titered Ad-LACZ virus was used to transduce human islets in parallel at the same multiplicity of infection as Ad-XIAP, βGAL staining showed that >95% of the islets present in the sample were transduced, at least on the surface (Fig. 1C), while untransduced human islets exhibited no evidence of βGAL staining (Fig. 1B). To determine the penetrance of the adenoviral transduction in human islets, X-GAL-stained islets were embedded in agar and sectioned. As shown in Fig. 1E, most of the cell clusters were completely transduced by the Ad-LACZ vector. Immunohistochemical analysis for the ductal cell marker CK19 in X-GAL-stained islets illustrated that the larger cell clusters with no visible staining in the middle were in fact ductal in origin, suggesting that all of the smaller completely transduced clusters were islets (Fig. 1F). Immunohistochemical staining for insulin in these sections was not possible due to the destruction of insulin epitopes following the X-GAL staining procedure, but dithizone staining in Ad-LACZ- or Ad-XIAP-transduced islets consistently demonstrated >90% islet purity (data not shown). Ad-XIAP- or Ad-LACZ-transduced human islets showed no loss in their ability to secrete insulin in response to glucose stimulation.
Hypoxia and hypoxia reoxygenation-induced apoptosis is inhibited by XIAP overexpression.
To examine the protective effects of XIAP during periods of hypoxia and following hypoxia reoxygenation injury, a well-characterized in vitro hypoxia reperfusion model system that has proven to induce apoptosis in β-cells was utilized (19, 27). After being exposed to hypoxia for 24 h (<1% O2, 5% CO2), 53.75 ± 4.84 cells/islet cross-section in the LACZ-transduced human islets stained TUNEL positive, while the levels of spontaneous apoptosis in the parallel untreated control islets remained at 5.71 ± 1.46 cells/islet cross-section (Fig. 2). However, when XIAP was overexpressed in the islets during hypoxia, very few TUNEL-positive apoptotic cells were observed (9.60 ± 1.36 cells/islet cross-section; Fig. 2). When islets were subjected to “reperfusion injury,” the protective effect of XIAP was even more apparent, as the number of apoptotic cells in XIAP-transduced islets (32.83 ± 4.78 cells/islet cross-section) was significantly reduced compared with Ad-LACZ-transduced islets (171.825 ± 9.71 cells/islet cross-section) (Fig. 2B). Although XIAP overexpression protected human islets from apoptotic death during hypoxia, it did not prevent the transient loss of β-cell function (Fig. 3). This result was confirmed by the low levels of insulin immunohistochemical staining in both control and XIAP-transduced islets, suggesting that hypoxia results in massive β-cell degranulation (Fig. 2A). Still, XIAP overexpression enhanced islet functional recovery following reoxygenation, since XIAP-transduced cells showed improved glucose-stimulated insulin secretion, while control LACZ-transduced islets were unable to respond to glucose stimulation after the “reperfusion” (i.e., reoxygenation) period (Fig. 3).
Presence of adenovirus in supernatants of cultured human islets following transduction.
In the clinical context, transduction of islets ex vivo is attractive since the majority of the adenoviral vector could be washed away, in theory, before portal vein infusion, reducing the risk of nontargeted gene transfer to the recipient. This hypothesis has not been verified, so culture supernatants from Ad-LACZ-transduced islets were collected and transferred to murine A9 fibroblast cells (which are very susceptible to adenovirus transduction) to determine whether infectious virions were present for up to 10 days posttransduction. Samples were collected every 2nd day after virus exposure, beginning on day 2 (at which point no washes had yet been performed after the addition of the vector on day 0). Following collection of the sample, the islets were washed twice with PBS and resuspended in fresh culture medium. Pretransduction culture supernatant was used as a negative control and had no infectious Ad-LACZ present, as indicated by the complete absence of βGAL staining in treated A9 cells. As expected, supernatants collected on day 2 had a high concentration of infectious Ad-LACZ, since none of the free virus had been washed away (Fig. 4). The amount of infectious Ad-LACZ in human islet culture supernatants decreased dramatically until no detectable βGAL staining was observed in A9 cells treated with day 10 supernatants (Fig. 4).
Reversal of diabetes required transplantation of a significantly reduced number of Ad-XIAP-transduced human islets.
Based upon the observation that XIAP-transduced human islets were protected from hypoxia and hypoxia reoxygenation death in vitro, we set out to evaluate the role of apoptosis in the acute posttransplant period in human islet transplantation. In our experience, a graft mass of 2,000 IEQs/mouse is necessary to reverse hyperglycemia in diabetic mice, which is in keeping with other reports (29–31). Marginal mass grafts containing 1,000 or 600 XIAP- or LACZ-transduced human islets were transplanted in streptozotocin-induced diabetic NOD-RAG−/− recipients (Fig. 5). Control grafts containing 1,000 Ad-LACZ-transduced human islets reversed hyperglycemia only 10% of the time (n = 10, Fig. 5A), while grafts containing only 600 islets never became functional (n = 7, Fig. 5A). In the presence of XIAP overexpression, islet grafts containing 600 islets were able to engraft rapidly and restore euglycemia in 89% of the diabetic mice (n = 9, Fig. 5A). The superior survival characteristics of the XIAP-transduced grafts were confirmed by measurement of serum human C-peptide levels in transplanted animals. Compared with serum from animals that had received 600 or even 1,000 Ad-LACZ-transduced human islets, the serum from animals receiving 600 Ad-XIAP-transduced islets showed significantly higher levels of human C-peptide, and these levels remained constant for several weeks posttransplant (Fig. 5B; animals in the control LACZ cohorts were hyperglycemic and had to be killed before the 21-day time point).
The present study demonstrates that XIAP overexpression enhances human islet survival during the posttransplant engraftment period by preventing hypoxia and hypoxia reperfusion-induced apoptosis. Inhibition of effector caspases via XIAP overexpression significantly enhanced human islet survival posttransplant, allowing a 70% reduction in human islet mass needed to reverse diabetes—the most profound reduction in graft mass capable of maintaining euglycemia in mice to be reported to date (Fig. 5). Although it was not measured directly, the success of these XIAP-transduced marginal mass islet grafts supports the fact that apoptosis is a critical mediator of β-cell survival posttransplant. This is consistent with the data obtained using an in vitro system that mimics hypoxia and reperfusion injury. XIAP-transduced human islets were significantly less apoptotic and recovered functional response to glucose stimulation following these insults, suggesting that XIAP mediates islet survival in vivo by inhibiting apoptosis during hypoxia and reperfusion stress (Figs. 2 and 3). These data confirm that the protective effect of XIAP overexpression previously observed in murine β-cells is reproducible and relevant in human islets (19).
The in vivo benefit of XIAP overexpression compared with other antiapoptotic molecules is most likely related to its ability to potently block activation of late, effector apoptotic caspases, which prevents β-cell death in the multifactorial apoptotic environment present following portal vein infusion. This suggests that using XIAP in the clinical setting could immediately enhance the availability of islet transplantation by dramatically reducing the amount of islet tissue necessary to obtain insulin independence, effectively removing the need for multiple donor infusions. The recent announcement (32) of a successful case of living-donor islet transplantation in Japan has provided renewed hope for many patients with type 1 diabetes. To bring living-donor islet transplantation to the mainstream and enhance recipient benefit for potential donor risk, success rates with marginal mass infusions must be improved, since only <50% of a donor’s pancreas could be excised safely, and our data implies that potent apoptosis prevention with XIAP could have an immediate impact in these studies.
By reducing the total number of apoptotic islets within a graft, one can speculate that XIAP may reduce immune stimulation caused by host antigen-presenting cells presenting graft antigens. This is an important distinction from previously reported genetic manipulations to human islets that have moderately enhanced marginal mass graft survival by targeting overexpression of β-cell growth factors like hepatocyte growth factor or erythropoietin or by overexpression of the insulin gene to enhance insulin output per islet (29, 33, 34). These strategies function by enhancing the survival, growth, and/or function of the islets that manage to engraft but still exert little effect on apoptotic islets posttransplant. Thus, a large amount of donor antigen would theoretically still be presented to the recipient’s immune system, potentially limiting the possibility of immunosuppressive therapy reduction or tolerance induction. Minimizing islet apoptosis posttransplant should prolong graft longevity, resulting in a more quiescent immunological state, and thereby enhance long-term rates of insulin independence. If this could be achieved, islet transplantation would be potentially safer and therefore more available to a broader spectrum of patients with type 1 diabetes, including children.
The primary limitation in the clinical application of genetic modification to islets lies in the efficiency and reproducibility of gene transfer. Over the years, it has become apparent that adenoviral vectors represent the most reliable and efficient method to deliver genes to intact islets without the worrisome long-term effect of nontargeted genomic integration associated with lentiviral vectors. That being said, many groups have struggled to obtain transduction efficiencies >50%, which limits the opportunity to observe any protective effect, especially with a transgene such as XIAP, which functions intracellularly (29, 34). One approach to improve transduction efficiency involves the use of more virus (i.e., increase the multiplicity of infection per cell), and while this does result in enhanced levels of transgene expression, the consequence is that the viral load itself becomes toxic to the islets (35). We hypothesized that transduction of islets very soon after isolation would enhance transduction efficiencies, since the microcapillary network in each islet would still be intact, allowing for improved adenoviral penetrance. Though collaboration with the University of Alberta Clinical Islet Isolation Laboratory, we were able to obtain and transduce islets <1 h following the completion of the isolation procedure. This resulted in highly efficient gene transfer to the islets, with >95% of Ad-LACZ-stained islets exhibiting at least 75% X-GAL staining through the islet (Fig. 1E). This result is especially remarkable considering that only 10 virus particles/cell were used, while other groups have obtained at most 50% transduction efficiency in human islets using 500–1,000 virus particles/cell (29, 34, 35). Even with the 50- to 100-fold decrease in adenovirus used to transduce the islets in our experiments, infectious virus was still detectable up to 8 days posttransduction, after >20 wash steps, suggesting that adenovirally transduced islets must be cultured for at least a week before transplantation to ensure that no vector is inadvertently transferred to the recipient (Fig. 4). Islet mass generally decreases in the culture setting due to hypoxic and other stress; however, the use of an antiapoptotic agent like XIAP would tend to enhance islet survival during this period.
Our studies suggest that effector caspase inhibition may be only transiently required to improve islet engraftment in the first few days or weeks following transplantation. Since first-generation adenoviral vectors were used in this study, and it is known that expression levels from these vectors declines 60–90 days posttransduction, we believe that XIAP overexpression would naturally taper off posttransplant, removing the need for the design of complex vectors with drug-regulatable promoters (36). Our studies also suggest that XIAP gene transfer and overexpression could be circumvented altogether, should pharmaceutical agents targeted at preventing caspase activation become available, allowing the positive effects observed following XIAP overexpression to be reproduced using a transient drug therapy. Taken together, these data confirm that XIAP overexpression in human islets dramatically enhances engraftment and in so doing reduces the islet mass necessary to achieve euglycemia, suggesting that clinical application of this protocol could immediately and greatly enhance the availability and long-term outcome of islet transplantation for type 1 diabetes.
This work was funded by the Juvenile Diabetes Research Foundation (JDRF) and the Canadian Institutes of Health Research (CIHR). The clinical islet laboratory is funded by the JDRF, Roberts Trust, and Alberta Building Trades. J.A.E. is supported by a studentship award, and J.R.T.L., A.M.J.S, and J.F.E are supported by research scientist awards, all from the Alberta Heritage Foundation for Medical Research. A.M.J.S. is also a recipient of a CIHR/Wyeth Research Chair in Transplantation.
The authors thank Aegera Therapeutics (Ottawa, ON) for providing the Ad-XIAP and Ad-LACZ stocks and the clinical islet laboratory of the Capital Health Authority for the isolation of human islets used in this study.