Islet β-Cells Deficient in Bcl-xL Develop but Are Abnormally Sensitive to Apoptotic Stimuli

Animals were housed under specific pathogen-free conditions at the University of Melbourne and at St. Vincent's Institute (Melbourne, Australia). Experiments involving animals were conducted according to our institutional animal ethics committee guidelines. Conditional Bcl-x knockout mice (Bcl-x^fl/fl) (B6;129S6-Bcl2l1^tm1.1Mam), a gift from Dr. Lothar Hennighausen (National Institutes of Health) (16), were backcrossed to C57BL/6 (BL/6) for six generations and then bred to an N10 B6.RIP-Cre transgenic line (B6.Cg-Tg(Ins2-cre)25Mgn/J) (21) to generate mice that lacked Bcl-xL expression in β-cells (RIP2-Cre.Bcl-x^fl/fl) and littermate controls (Bcl-x^fl/fl). B6^bm1.RIP-Bcl-2 transgenic mice (B6-H2^bm1-Tg(RIP2-BclII)407Wehi/J) expressing human Bcl-2 in β-cells under control of the rat insulin promoter have been described previously (7). PCR screening of mice is detailed in the online supplemental data available at http://diabetes.diabetesjournals.org/content/early/2009/06/23/db08-1602/suppl/DC1.

Pancreatic β-cells and non–β-cells were sorted from trypsin-dissociated islets based on auto-fluorescent profile (β-cells have high FL1 auto-fluorescence) (22) using a FACSAria (Becton Dickinson, San Jose, CA). DNA was extracted from sorted islet cells and from homogenized tissues using proteinase K. Bcl-x^fl/fl gene deletion was detected by PCR as detailed in the online supplemental data.

Isolated islets or sorted β- and non–β-cells were solubilized in lysis buffer (0.25M Tris–HCl, pH 6.8, 10% SDS, 20% glycerol, 5% 2-mercaptoethanol, 0.02% bromophenol blue, and 0.5 mg/ml Pefabloc). Wild-type mouse embryonic fibroblasts (MEFs) and MEFs from Bcl-x knockout mice were used as controls. After SDS-PAGE electrophoresis (12% gel) and transfer to immobilon P membrane (Billerica, MA), the blot was probed for Bcl-xL, followed by stripping and reprobing for β-actin. Antibody details are given in the online supplemental data.

Details of the antibodies used and their dilutions are given in the online supplemental data. Intracellular insulin staining and flow cytometric analysis was performed as described (23).

Immunohistology was done on Bouin's solution fixed, paraffin-embedded pancreas. Sections (5 microns) were cut at four levels separated by 100 microns, so each section sampled different islets. For scoring of islet sizes, the numbers of insulin-expressing cells per islet were counted and the islet assigned to a size group ranging from <5 insulin-positive cells per islet to >150 insulin-positive cells per islet. Each size group was then expressed as a percentage of total islets.

Double-label immunofluorescence was done on acetone-fixed, frozen sections of pancreata. Sections were double stained for insulin and Cre recombinase. To score Cre expressing insulin cells, at least 1,000 insulin-positive cells were counted per mouse pancreas. Counting was done using a BIORAD MRC 1024 confocal microscope using BIORAD software.

RIP-Cre.Bcl-x^fl/fl mice and Bcl-x^fl/fl littermate controls were injected intraperitoneally with 35 μg/ml of STZ (Sigma-Aldrich, St. Louis, MO) in citrate buffer daily for 4 days. Sections of 4% (wt/vol) paraformaldehyde fixed, paraffin-embedded pancreas were then subject to TUNEL using biotinylated dUTP that was detected with a horseradish peroxidase–streptavidin conjugate. Staining with anti-insulin antibody was detected with an alkaline phosphatase–conjugated secondary antibody. Antibody details are given in the online supplemental data. At least 45 islets (excluding islets with <20 cells) were scored, per mouse, for TUNEL-positive cells. Data were plotted as the average number of TUNEL-positive cells per islet section.

Islets were isolated from 3 to 4 aged-matched male and female mice and pooled. Following overnight culture, 100 islets of uniform size range were handpicked into 3.5-cm untreated Petri dishes containing 1.1 ml of supplemented CMRL medium, and death agents were added. Islets plus medium containing detached cells were washed in PBS and dissociated at 37°C in Accutase (Chemicon, Millipore, Temecula, CA) for 5 min. After washing in PBS, islet cells were resuspended in 300 μl of a hypertonic buffer containing 50 μg/ml propidium iodide, 0.1% wt/vol tri-sodium citrate, and 0.1% vol/vol Triton-X 100 before being analyzed by flow cytometry for the numbers of cells with a <2C DNA content (apoptotic cells), as previously described (24). Cells were analyzed on a FACSCalibur using Cell Quest software (Becton Dickinson, San Jose, CA) in the FL3 channel.

Apoptosis-inducing reagents used were recombinant murine interferon γ (IFNγ) (Genentech, South San Francisco, CA), recombinant human interleukin-1β (IL-1β) (R&D Systems, Minneapolis, MN), Mega Fas ligand (FasL) (APO-O10; Apoxis/Topotarget, Lausanne, Switzerland), staurosporine (Sts; Sigma-Aldrich) dissolved at 10 mmol/l in DMSO and stored under nitrogen gas at −70°C, thapsigargin (Thap; Calbiochem, EMD Biosciences, San Diego, CA) dissolved at 1.5 mmol/l in 100% ethanol and stored at −20°C and N^G-methyl-l-arginine acetate salt (NMMA) (Sigma-Aldrich) dissolved at 200 mmol/l in medium and stored at −20°C.

Islets from Bcl-x^fl/fl and RIP-Cre.Bcl-x^fl/fl mice were freshly isolated and snap frozen on dry ice. RNA was prepared using the RNeasy Kit (Qiagen, Valencia, CA). First-strand cDNA was prepared from 0.5 to 1.5 μg RNA using the Taqman RT system (Roche, Mannheim, Germany). Real-time PCR was performed using the ABI Prism 7900 (Applied Biosystems) and the QuantiTect SYBR Green PCR Kit (Qiagen) in 15-ml reaction volumes. Bcl-x, Bcl-2, Bcl-w, Mcl-1, and A1 were assayed by quantitative RT-PCR. Data analyses were performed with the ΔC_T method using β-actin as an internal control. Primer sequences are provided in online supplemental data.

Data are represented as means ± SD. Results were analyzed using a two-tailed Student's t test. A P value of <0.05 was considered significant.

Mice carrying a floxed Bcl-x gene (16) were crossed with B6.RIP2-Cre transgenic mice (21,25), and their offspring intercrossed to obtain RIP-Cre.Bcl-x^fl/fl mice and Cre-negative littermates (Bcl-x^fl/fl). The RIP2 promoter has been shown to direct expression of transgenes to β-cells as early as embryonic day 10 (26). Bcl-x^fl/fl and RIP-Cre.Bcl-x^fl/fl offspring were similar in appearance, numbers, weight, and blood glucose levels (see online supplemental data, Fig. S1A–C).

Deletion of the Bcl-x locus was verified by PCR on fluorescence-activated cell sorter (FACS)-sorted β-cells (∼90% pure) using primers that gave rise to a ∼150-bp product when the Bcl-x gene was deleted (Fig. 1,A, lane 2). Bcl-x gene deletion was not seen in non–β-islet cells (Fig. 1, lane 3). Thymus or tail DNA from Bcl-x^fl/fl mice that expressed the Cre recombinase in thymocytes under the lck promoter was used as a positive and negative control, respectively (Fig. 1, lanes 4 and 5). A survey of tissues from RIP-Cre.Bcl-x^fl/fl mice showed deletion of Bcl-x in whole pancreas, as expected (Fig. 1, lane 7), and unexpectedly in salivary gland (Fig. 1, lane 9). The RIP-Cre transgene can show ectopic expression, for example, in the hypothalamus (25) and as our data indicate, apparently also in the salivary gland.

Western blotting of FACS-sorted β-cells (∼90% pure) from RIP-Cre.Bcl-x^fl/fl mice showed a substantial reduction of Bcl-xL protein when compared with β-cells from Bcl-x^fl/fl mice (Fig. 1,B, lanes 1 and 2). Knockdown of Bcl-xL protein was less obvious in whole islets from RIP-Cre.Bcl-x^fl/fl mice when compared with Bcl-x^fl/fl islets (Fig. 1, lanes 3 and 4) because whole islets contain a number of cell populations of which only ∼75% comprise β-cells. Probing of lysates from wild-type and Bcl-x–deficient MEFs was used as a control for the anti–Bcl-xL antibody (Fig. 1, lanes 5 and 6). FACS-sorted non–β-cells showed no reduction in Bcl-xL levels in RIP-Cre.Bcl-x^fl/fl islets (Fig. 1, lanes 7 and 8).

The RIP-Cre transgene that we used directs expression of Cre to ∼85% of islet β-cells (http://jaxmice.jax.org/strain/003573.html), partially explaining the presence of some Bcl-xL protein in the RIP-Cre.Bcl-x^fl/fl sorted (∼90% pure) β-cell population. It was necessary to determine the efficiency of Bcl-x gene deletion in the β-cell population of our mice. Staining for Bcl-xL in normal islets by immunohistology or intracellular flow cytometry proved intractable. However, because Cre is a highly efficient recombinase requiring only a few molecules to mediate gene deletion (27), the presence of Cre should equate to deletion of Bcl-x. Pancreas sections from RIP-Cre, Bcl-x^fl/fl, and RIP-Cre.Bcl-x^fl/fl mice were examined for Cre expression using immunofluorescence microscopy (Fig. 2,A). Bcl-x^fl/fl β-cells showed no background staining with the anti-Cre antibody. Cre expression in RIP-Cre and RIP-Cre.Bcl-x^fl/fl β-cells was comparable with ∼70% of adult β-cell nuclei expressing Cre at levels detectable by immunofluorescence (Fig. 2 A, inset). There was variability in the intensity of Cre staining among β-cells, and the ∼70% value will be an underestimate given that immunohistology would not detect low numbers of Cre molecules that may still be sufficient to mediate Bcl-x^fl/fl gene deletion. Cre was also detected in neonatal and embryonic β-cells from these animals. Therefore, β-cells can develop and survive in the absence of Bcl-xL expression.

Staining for the islet cell hormones insulin, glucagon, and somatostatin revealed normal islet architecture in RIP-Cre.Bcl-x^fl/fl mice (Fig. 2,B). We analyzed islets from recently weaned (31 days), adult (100 days), and aged mice (300 days) for the range of islet sizes. Both groups showed the same islet size range at these timepoints (Fig. 2,C). The absolute numbers of islets counted in four histology sections were also comparable (Fig. 2 D).

Mice with Bcl-xL–deficient β-cells were treated with multiple low doses of STZ to induce apoptosis in the β-cells (28). TUNEL staining of pancreas sections by immunohistology showed that islets with β-cells deficient in Bcl-xL were more susceptible to this treatment than islets from control littermates (Fig. 2 E). Thus, under conditions of stress, β-cells require Bcl-xL to maintain survival in vivo.

After collagenase isolation, the islet yields from B6.RIP-Cre, Bcl-x^fl/fl, and RIP-Cre.Bcl-x^fl/fl littermates were counted and no major differences found (Fig. 3,A). In addition, islets from all strains appeared microscopically similar postisolation and also after 6 days in culture (online supplemental Fig. S2A). The proportion of β-cells in the islets was quantified by intracellular insulin staining and flow cytometric analysis. All groups contained similar proportions of insulin-producing cells (Fig. 3 B and online supplemental Fig. S2B). These results show that Bcl-xL is not essential for maintaining survival of β-cells during whole-islet extraction and culture.

We next tested the importance of Bcl-xL expression in cultured β-cells under conditions of stress, using whole-islet assays. We compared islets containing Bcl-xL–deficient β-cells with those containing wild-type β-cells or β-cells overexpressing Bcl-2. We expected that Bcl-2 would act as a functional homolog of Bcl-xL (as it does in hemopoietic cells) and allow us to compare β-cells that lacked Bcl-xL with those that overexpressed its functional homolog. We chose stress stimuli, relevant to type 1 diabetes, known to activate different members of the BH3-only, proapoptotic Bcl-2 protein subfamily (29). These included cytotoxic drugs (Sts) and proinflammatory cytokines (IL-1β plus IFNγ) as well as agents that induce endoplasmic reticulum stress (Thap), DNA damage (γ-irradiation [RAD]), or death receptor signaling (FasL) (29).

The broad-spectrum, protein kinase inhibitor, Sts, is a classic apoptosis initiator in most cell types. Islets were cultured for 40 h in Sts, trypsinized, and then assayed for apoptosis by flow cytometric analysis of DNA fragmentation, as described by Nicoletti (24). Bcl-xL deficiency did not enhance islet cell death in response to Sts (Fig. 4,A). We ensured that the vehicle control (DMSO) was not itself toxic to Bcl-xL–deficient β-cells (Fig. 4,B). Overexpression of Bcl-2 protected β-cells from Sts as previously shown (7) (Fig. 4 C). Although supra-physiological levels of Bcl-2 protected against Sts-induced killing, the presence or absence of physiological levels of Bcl-xL made no difference in the response of islet cells to this cytotoxic agent.

Thap inhibits endoplasmic reticulum–specific Ca²⁺ ATPase function causing apoptosis because of endoplasmic reticulum stress. Whole islets with β-cells lacking Bcl-xL were more susceptible to Thap-induced death than control islets (Fig. 4,D), supporting the idea that Bcl-xL is a critical player in controlling endoplasmic reticulum stress–induced apoptosis. The vehicle control (1.3% ethanol) was not toxic to Bcl-xL–deficient β-cells (Fig. 4,E). Overexpression of Bcl-2 resulted in a reduced but not significant protection of β-cells from Thap (Fig. 4,F). This result differs from that of Zhou et al. (6) who found that Bcl-xL overexpression in β-cells gave good protection of mouse islets from Thap-induced apoptosis after 48 h in culture. Islets were also treated with γ-RAD to induce DNA damage. Although only a low amount of islet cell death was found in control samples 6 days after RAD, this increased when Bcl-xL was lacking from β-cells (Fig. 4 G).

In summary, Bcl-xL is needed to help protect β-cells from agents that induce endoplasmic reticulum stress or DNA damage. In addition, it appeared that in β-cells Bcl-2 and Bcl-xL may not be functionally equivalent.

Whole rodent primary islets undergo cell death from upregulation of inducible nitric oxide synthase (iNOS) when exposed to high levels of inflammatory cytokines, such as IL-1 plus IFNγ. NO may inhibit mitochondrial electron transfer by inactivating aconitase (see 30) resulting in cell death. NO may also induce endoplasmic reticulum stress in mouse or rat β-cells (31,32) and deficiency in the endoplasmic reticulum stress–induced protein, C/EBP homologous protein, can partially protect (31). Antiapoptotic Bcl-2 family members, such as Bcl-xL or Bcl-2, are likely to be critical for protecting islets against both of these apoptotic pathways. We found that islets with β-cells lacking Bcl-xL were more susceptible to IL-1β plus IFNγ (Fig. 5,A). This death could be blocked in both control and Bcl-xL–deficient islets by coaddition of the iNOS inhibitor NMMA (Fig. 5,B), indicating that the increased β-cell death in Bcl-xL–deficient islets was mediated primarily by NO. Overexpression of Bcl-2 in β-cells did not protect the cytokine-treated islets from death as previously shown (see 33 and Fig. 5 C).

Islet cells do not normally express Fas, but low concentrations of IL-1 plus IFNγ (that do not kill) can induce Fas on β-cells (34). After culture with low levels of IL-1β plus IFNγ to induce Fas and with FasL to induce cell killing, death was substantially increased in islets with β-cells that lacked Bcl-xL (Fig. 5 D). Our lab has previously shown that β-cell death through the death receptor pathway is mediated by the BH3-only protein Bid and can be inhibited by Bcl-2 overexpression (35). In summary, Bcl-xL is critical for protection of β-cells from death through the death receptor pathway.

Given there was so little impact of Bcl-xL deficiency on the development of islets and their survival in the absence of cytotoxic stressors, we tested whether other antiapoptotic members of the Bcl-2 family were transcriptionally upregulated in the absence of Bcl-xL. Whole islets from Bcl-x^fl/fl and RIP-Cre.Bcl-x^fl/fl mice were assayed for levels of mRNA for antiapoptotic Bcl-2 family members (Bcl-xL, Bcl-2, Bcl-w, Mcl-1, and A1). As expected, there was a notable reduction in Bcl-xL mRNA in islets with Bcl-xL–deficient β-cells (Fig. 6) consistent with the Western analysis of whole islets (Fig. 1 B). Unexpectedly, a twofold reduction in Bcl-2 mRNA was also noted, but the biological implications of this reduction are at present unclear. There was no significant upregulation of the other gene transcripts in islets with Bcl-xL–deficient β-cells.

To address the role of Bcl-xL in pancreatic islets we used the Cre-loxP system to delete the Bcl-x gene specifically in β-cells. Loss of Bcl-xL in β-cells was not lethal. The numbers of islets were normal and the proportion of β-cells that made up the islets was similar to controls. Although a proportion of β-cells (∼20%) do not express Cre (25) and would remain Bcl-x sufficient, these cells did not have a survival advantage over their Bcl-x–deficient neighbors, again indicating that Bcl-xL is not essential for β-cell growth and survival. Although other prosurvival members were expressed but not transcriptionally upregulated in response to Bcl-xL loss in β-cells, it appears likely that normal levels could substitute for Bcl-xL deficiency, with Mcl-1 being a plausible candidate given its ready detection in islets at the protein level (2,5). Interestingly, Hager et al. (36) recently showed that Bcl-x gene deficiency did not affect growth of β-cell tumors transformed by the SV40 T antigen and that other antiapoptotic members were not upregulated at the transcriptional level in these tumors.

Our findings in β-cells differ from those for hematopoietic cells (13,15), neurons (13,37,38), and germ cells (16) in which Bcl-xL was found to play an indispensable role in cell survival during development. Our results are more like those observed for another epithelial tissue, the mammary gland (17), where Bcl-xL depletion did not affect normal organ development or function during mammopoiesis (17), although it appears to be the most highly expressed antiapoptotic Bcl-2 family member in this tissue (17,39). Bcl-xL was important, however, during the first phase of mammary gland involution after weaning, a stage characterized by considerable cell stress and extensive tissue remodeling that is associated with apoptosis (17,40). In line with this, Bcl-xL–deficient β-cells were abnormally susceptible to death when subject to a synchronous cell stressor, low-dose STZ.

In vitro, Bcl-xL helped to protect β-cells against a variety of cell death inducers that activate different proapoptotic, BH3-only proteins (29). Cell death elicited by Thap-induced endoplasmic reticulum stress, γ-RAD, IL-1β plus IFNγ, or Fas death receptor signaling was enhanced in the absence of Bcl-xL, confirming it as a critical guardian of cell survival in these settings. Unexpectedly, Bcl-xL deficiency did not enhance cell death by the classic apoptosis initiator, Sts. Sts may cause a massive induction of several BH3-only proteins that rapidly overwhelms all antiapoptotic Bcl-2 family members in β-cells so that presence or absence of physiological levels of Bcl-xL makes little difference.

We also studied the effects of Bcl-2 overexpression in β-cells under the premise that Bcl-2 was a functional homologue of Bcl-xL. Bcl-2 overexpression reduced cell deaths induced by Sts and Fas ligand (see 35) but not by Thap or cytokines (IL-1β plus IFNγ). It is possible that in response to Thap or cytokines β-cells activate additional apoptotic death pathways not controlled by Bcl-2 family members, for example serine protease-mediated mechanisms that damage the mitochondrial membrane (41). Alternatively, Zhou et al. (6) found that overexpression of Bcl-xL itself did afford β-cells significant protection from Thap-induced killing. It may be that Bcl-2 and Bcl-xL can play different roles in the β-cell's response to different types of cytotoxic insults. An explanation for this comes from studies on mouse embryonic fibroblasts (42) which reported that although Bcl-2 and Bcl-xL can act as functional homologues under some situations, they in fact control the Bax/Bak molecules differently. Bcl-xL is able to restrain both Bak and Bax, whereas Bcl-2 can only restrain Bax and may therefore be a less effective inhibitor of apoptosis under certain conditions.

In conclusion, we have demonstrated that although Bcl-xL is not needed for β-cells to develop, it plays an important role in promoting islet cell survival when islets are exposed to a range of death stimuli. Our data indicate that Bcl-xL is critical for helping to protect β-cells against most, but not all, specific death signals. In addition, Bcl-xL and Bcl-2 are not functionally equivalent in β-cells. Whether Bcl-xL also functions to maintain islet integrity under unfavorable culture conditions will be important to examine for the prospects of clinical applications of islet manipulation in culture followed by subsequent transplantation into patients.

This work was supported by grants and fellowships from the Juvenile Diabetes Research Foundation, the National Health and Medical Research Council of Australia, the National Cancer Institute, the Leukemia and Lymphoma Society of America, and the Swiss National Science Foundation.

No potential conflicts of interest relevant to this article were reported.

We are very grateful to Dr. L. Hennighausen (National Institute of Diabetes and Digestive and Kidney Diseases and National Institutes of Health) for the Bcl-x(loxP) targeted mice, Dr. H. Thomas (St. Vincent's Institute, Australia) for the B6.RIP-Cre transgenic mice, and Apoxis/Topotarget (Lausanne, Switzerland) for the MegaFasL. We thank Rima Darwiche and Natalie Sanders for islet cell sorting, Dr. M. Yabal for MEFs, and Dr. M. Smeets for advice with Western blotting. Special thanks to Daniela Novembre-Cycon and Anthony Gomes for animal management and glucose testing. Additionally we thank Dr. D. Huang for helpful discussions.