Expression of the NH₂-Terminal Fragment of RasGAP in Pancreatic β-Cells Increases Their Resistance to Stresses and Protects Mice From Diabetes

The supplemental methods for cell culture, chemicals and antibodies, transgene detection by PCR, quantitative PCR, mouse islet isolation and dissociation, preparation of tissue sections and immunochemistry, insulin quantitation, Western blot analysis, Southern blot, nuclear protein extract preparation, and electromobility shift assay (EMSA) are found in the online appendix (available at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0104/DC1).

Apoptosis ex vivo was assessed by scoring the number of cells with pycnotic nuclei after Hoechst 33342 staining (20). Apoptosis in vivo was assessed by a terminal transferase dUTP nick-end labeling (TUNEL) assay (DeadEnd Fluorometric TUNEL system, catalog no. G3250; Promega, Basel, Switzerland) on islet paraffin sections as per the manufacturer's protocol.

All procedures on mice were performed according to the Swiss legislation for animal experimentation. Unless noted otherwise, the animals were used at an age of 8–12 weeks.

The transgenic construct (RIP-HA-N[D157A].xf3) bears fragment N of RasGAP under the control of the rat insulin promoter (RIP). It was obtained by ligation of a blunt-ended BamHI/SalI 1.4-kb fragment from plasmid HA-N(D157A).bs (22) with a blunt-ended XbaI/HindIII 4-kb fragment from RIP-vMos.xf3 plasmid. The correctness and functionality of the plasmid were controlled by sequencing and transfection into insulinoma cell lines. Finally, a BamHI 2.8-kb fragment from RIP-N.xf3 was microinjected into FVB/N oocytes at the transgenic animal facility of the University of Lausanne. Four independent RIP-N–expressing founders were obtained. Founders 1 and 2 were used in the experiments described here.

Blood glucose content of mice under feeding or fasting (16 h) conditions was determined with an Accu-Check Compact Plus glucometer (Roche Diagnostics). For the intraperitoneal glucose tolerance tests (IPGTTs), fasted (16 h) animals were injected intraperitoneally with 2 mg glucose per kilogram body weight. Blood glucose levels were determined from a blood drop taken after a short incision of the tail tip at increasing time intervals (−30, 0, 15, 30, 60, 90, 120, and 150 min) following glucose injection.

Type 1–like diabetes was induced by multiple low-dose streptozotocin injections. Briefly, 4-h–fasted female RIP-N mice were injected intraperitoneally with 50 mg streptozotocin per kg of mice. This procedure was repeated every day for a total period of 5 days. Streptozotocin was prepared and diluted in citrate buffer (pH 4.5) (sodium citrate 25 mmol/l, citric acid 23 mmol/l) just before injection. Control mice were injected with the citrate buffer alone. Blood glucose levels were assessed biweekly.

Islets were isolated from mice pancreas as described in the supplemental research design and methods section in the online appendix. The islets (200 per 100-mm dish in 10 ml culture medium) were incubated overnight at 37°C, 5% CO₂. The next day, the islets were hand-picked and cultured in Krebs-Ringer bicarbonate HEPES buffer (KRBH)-BSA (120 mmol/l NaCl, 4 mmol/l KH₂PO₄, 20 mmol/l HEPES, 1 mmol/l MgCl₂, 1 mmol/l CaCl₂, 5 mmol/l NAHCO₃, pH 7.4, with 0.5% BSA) at 37°C and 5% CO₂. The following day, well preserved and good-quality islets were again hand-picked and placed into 12-well plates (10 islets/well) in 1 ml KRBH-BSA containing 2.8 mmol/l glucose for 1 h. The islets were then transferred to new wells containing 2.8 or 20 mmol/l glucose with or without 10 nmol/l exendin-4 (catalog no. H-8370; Bachem) in 1 ml KRBH-BSA and incubated for 2 additional hours. The supernatant and islets were collected into separate tubes and placed on ice. The islets were lysed in 500 μl acid/ethanol (75% ethanol/1.5% concentrated HCl) and sonicated 15 sec (using a W-375 cell disruptor from Kontron equipped with a 3-mm tip). Insulin in the supernatant and extracted islets was measured using an radioimmunoassay kit (catalog no. RI-13K; Linco).

Unless stated otherwise, the statistical analyses were done with Microsoft Office Excel 2003 SP1 using the two-tailed unpaired Student t test. Significance is indicated by an asterisk when P < 0.05/n, where P is the probability derived from the t test analysis and n is the number of comparisons done (Bonferroni correction). All the other statistical analyses were performed with the SAS/STAT software (version 9.1.3; SAS Institute, Cary, NC).

A transgenic vector was constructed (see research design and methods) so as to encode an HA-tagged form of fragment N bearing the D157A mutation (preventing it from being cleaved by caspases) under the control of the RIP and regulatory sequences of the simian virus 40 (SV40) gene (Fig. 1,A). The construct was injected into FVB/N oocytes, and transgene-positive mice were identified by Southern blotting (Fig. 1,B). In total, four founder mice were obtained. The results presented here all include data from founder 1 (labeled mouse 5 in Fig. 1,B). When indicated, some experiments were also performed with mice derived from founder 2 (labeled mouse 28 in supplemental Fig. S1). By comparison, with the endogenous insulin promoters, it was estimated that founders 1 and 2 bore 12–15 and 1 copies of the transgene in their genome, respectively (Fig. 1 B and supplemental Fig. S1).

To determine the expression pattern of fragment N in the transgenic line, lysates from pancreatic islets, liver, brain, and spleen were analyzed by Western blotting using antibodies specific for the HA tag or for the NH₂-terminal part of RasGAP. Figure 1,C shows that fragment N was, as expected, only expressed in islet cells. Immunofluorescence analysis of both founders revealed that fragment N was restricted to the endocrine part of the pancreas (Fig. 1,D and supplemental Fig. S2A and B) and that the vast majority of fragment N–expressing cells corresponded to β-cells (i.e., insulin-containing cells) (Fig. 1 E).

In various cell types, fragment N, when ectopically expressed or when generated in response to mild stress, activates Akt (19,21,23). As adaptive mechanisms can take place in vivo, it was important to determine whether fragment N could induce a chronic Akt activity in islet cells in mice. Islets isolated from control and RIP-N mice were therefore analyzed for the presence of activated Akt. As shown in Fig. 2 A, there was a significant approximately threefold increase in Akt activity in islet cells from RIP-N compared with control islets. This indicates that fragment N can stimulate Akt on a long-term basis when expressed in vivo.

A potential important property of fragment N in the context of β-cell protection is its ability to block NF-κB activation. This property, however, had so far only been evidenced in cultured immortalized cell lines (19,21). As shown in Fig. 2,B, binding of nuclear factors to NFκB binding elements was markedly diminished in nuclear extracts from RIP-N mouse–isolated islet cells stimulated with cytokines compared with similarly treated islets isolated from control mice. Moreover, cytokine-induced expression of the transcript encoding inducible nitric oxide (iNOS) synthase, which participates in β-cells apoptosis (24) and the gene of which is an NFκB target (25), also appeared to be impaired in islets cells isolated from RIP mice compared with wild-type islets (Fig. 2 C).

These results indicate that fragment N regulates Akt and NFκB in β-cells in vivo in a manner similar to what has been described using cultured cell lines. As cytokines can induce apoptosis of β-cells via NFκB–mediated NO production (24), these results also suggest that the ability of fragment N to protect β-cells might rely, at least in part, on its capacity to target the NFκB–iNOS axis.

It was reported in transgenic models done using RIP-Cre mice that the RIP promoter can also be active in the brain (more specifically in the hypothalamus) (26,27). Immunohistochemical analysis, however, did not reveal the presence of fragment N in hypothalamic sections from adult RIP-N mice (supplemental Fig. S3). This indicates that the RIP-N transgene is not expressed in adult mouse brain or, if it is expressed, at levels that are much lower than those detected in the endocrine pancreas and that are under the sensitivity limit of our assay.

Expression of fragment N in insulinomas and islet cells leads to Akt activation (Fig. 1,F) (21). Since Akt signaling has the potential to stimulate cell survival and proliferation (28), and since transgenic mice expressing a constitutively active form of Akt (myr-Akt) show an increase in both β-cell size and total islet mass (15), the presence of fragment N in islets might affect the morphology and cellularity of the endocrine pancreas. However, neither the proportion of α- and β-cells (Fig. 3,A), nor the insulin content of the pancreas (Fig. 3,B), were affected by the presence of fragment N. Moreover, the size of the islets did not appear to be different in RIP-N transgenic mice compared with control mice (Fig. 3,C). Finally, the percentage of cells positive for the nuclear protein Ki67 that is preferentially expressed in dividing cells was similar in both types of mice (Fig. 3,D). These results indicate that fragment N does not favor β-cell proliferation in an in vivo setting and that it does not affect the normal development of the endocrine pancreas. Consistent with this notion is the observation that RIP-N mice did not develop insulinomas over an 18-month period (as assessed by a drop in glucose blood level and increased mortality) (Fig. 8).

In wild-type mice, the basal apoptotic rate in islets is very low (< 0.5%; Fig. 6,B) or undetectable (15). In contrast, islets from transgenic mice expressing a constitutively active form of Akt show a marked increase in β-cell apoptosis (15). Despite the ability of fragment N to activate Akt (Fig. 2,A), there was no associated increase in the basal apoptotic rate in islets from RIP-N mice compared with the wild-type controls, either in vitro or in vivo (Fig. 4,A and first two bars of Fig. 6,B). Moreover, islets isolated from RIP-N mice were more resistant than those isolated from control mice when subjected to a variety of stress stimuli, including inflammatory cytokines, the free fatty acid palmitate, and high glucose concentrations (Fig. 4). These results demonstrate that fragment N efficiently protects pancreatic β-cells against various noxious conditions and stimuli, including some that are associated with the development of type 1 and type 2 diabetes (e.g., inflammatory cytokines and free fatty acids).

Transgenic mice expressing a nondegradable form of IκBα under the control of Pdx1 promoter, which drives its expression in the β-cells of the pancreas, display impaired glucose-induced insulin secretion (29). Fragment N, by blocking NFκB activity (19,21), could potentially similarly affect insulin secretion. However, fragment N expression in β-cells did not modify glycemia under nonfasted (Fig. 5,A) or fasted (Fig. 5,B, first points in the graphs) conditions. Moreover, the ability of the transgenic mice to metabolize glucose, assessed by IPGTTs, was not negatively affected by the presence of the transgene in β-cells (Fig. 5,B). Finally, islets isolated from control and RIP-N transgenic mice had a similar ability to secrete insulin in response to glucose and the gluco-incretin exendin-4 (Fig. 5 C). These results indicate that fragment N does not compromise the ability of β-cells to secrete insulin in response to augmented glucose levels.

Multiple low-dose streptozotocin injections in mice induce islet inflammation, ultimately leading to β-cell loss and diabetes (30,31). This model is thought to mimic the development of type 1 diabetes in humans (32). Using this protocol, it was found that RIP-N mice were resistant to diabetes induction compared with control mice (Fig. 6,A). Assessment of apoptosis by the TUNEL method showed that the percentage of β-cell apoptosis induced by streptozotocin in vivo was significantly reduced in the RIP-N mice compared with the wild-type controls (Fig. 6 B).

To further characterize the reduced sensitivity to streptozotocin-induced diabetes in RIP-N mice, pancreas sections from control and RIP-N mice treated or not with streptozotocin were prepared. The mice were killed 3 days after the last streptozotocin injection at a time were increased apoptosis can be detected in wild-type mice (see Fig. 6,B) but before the apparition of an overt diabetes (see Fig. 6,A). At this time, leukocytic infiltration can be visualized in the islets of wild-type streptozotocin-induced diabetic mice. This was accompanied by a loosening of the islet structure. This was not seen in RIP-N mice (Fig. 7,A). Additionally, insulin staining was reduced in the islets of wild-type streptozotocin-induced diabetic mice (Fig. 7,B). Moreover, there was almost no sign of insulinitis in the transgenic mice after the streptozotocin injections, while in similarly treated control mice, a strong insulinitis developed (Fig. 7,C). Finally, there was a significant reduction in CD3-positive leukocyte infiltration in RIP-N mice compared with the control mice after the streptozotocin treatment (Fig. 7 D). Taken together, these results indicate that streptozotocin induces less damage in RIP-N islets, which results in a weaker inflammatory response compared with wild-type islets.

In relatively young mice (8–12 weeks), glucose homeostasis and insulin secretion are unaffected by the presence of fragment N in β-cells (Fig. 5). To determine whether the transgene could nevertheless negatively affect the function of pancreatic β-cells on a longer-term basis, the glycemia of a cohort of female and male wild-type and RIP-N mice was followed for up to 130 weeks (Fig. 8,A). Males displayed higher glycemia values than females. There was also a significant decrease in glycemia as the mice aged. However, the glycemia between wild-type and RIP-N mice for a given sex was not statistically different. This indicates that fragment N does not negatively affect the function of the islets of Langerhans. Consistent with this is the observation that IPGTTs performed on very old animals did not reveal differences between wild-type and RIP-N mice (Fig. 8 B).

Constitutive expression of Akt in pancreatic β-cells increases the likelihood of insulinoma development, leading to a reduction of the lifespan expectancy of the mice (17). As fragment N activates Akt, it was relevant to check if fragment N would have a negative impact on the survival of the mice. Figure 8,C shows that this is not the case. While females lived significantly longer than males, the presence of the transgene did not affect the percent survival rate of the mice. Finally, there was no histological difference that could be evidenced on islets from very old wild-type and RIP-N mice (Fig. 8 D). Altogether, these results indicate that fragment N expressed in pancreatic β-cells displays no negative effect throughout the lifespan of mice.

Apoptosis, which is the cause of β-cell death in patients with type 1 diabetes (33), might also participate in the loss of β-cell mass observed in type 2 diabetes (34,,–37). The notion, however, that there is a decrease in β-cell mass in type 2 diabetes has been controversial for a number of years. Nevertheless, if one considers those studies using well-preserved pancreases obtained from autopsies, it appears that there is a 3- to 10-fold increase in the rate of β-cell apoptosis in type 2 diabetic patients compared with control subjects (38). These results indicate that failure to compensate for insulin resistance could result from decreased β-cell mass mediated by apoptosis.

Understanding the pathways leading to β-cell death and β-cell protection might therefore be of crucial importance to find new approaches to treat diabetic patients. Procedures to block β-cell death could not only potentially inhibit the development of diabetes but might also be useful in the context of islet transplantation, where apoptosis has been shown to adversely affect the number of islets that can be implanted in patients. Here, we present an in vivo model, the RIP-N transgenic mice, where the NH₂-terminal fragment of RasGAP (called fragment N) effectively protects the pancreatic β-cell against apoptosis without affecting their ability to appropriately secrete insulin in physiological conditions and without favoring excessive proliferation.

Hyperlipidemia, a risk factor for the development of diabetes (39), can cause β-cell apoptosis (40,41). Islets isolated from RIP-N mice underwent less apoptosis in conditions mimicking hyperlipidemia (i.e., high concentrations of palmitate) compared with control islets. Hyperglicemia, which can lead to β-cell dysfunction and death (6), also induced less apoptosis in fragment N–expressing islets. Finally, RIP-N transgenic islets were more resistant to interleukin-1β–, tumor necrosis factor-α–, and interferon-γ–induced death. Interestingly, these inflammatory cytokines, known to be involved in the development of type 1 diabetes, have also been shown to be produced at high concentrations in diabetes-prone obese patients (42,43). These observations suggest that the protective signals elicited by fragment N can counteract prodiabetic conditions (e.g., hyperglycemia, hyperlipidemia, presence of inflammatory cytokines) that are deleterious for β-cells.

Much work on the cellular mechanisms controlling cell death and survival has been performed within the last few years. This knowledge has been used to manipulate β-cells in order to increase their survival capacities. One strategy was based on Akt because it is a potent antiapoptotic kinase in many cell types (13). Transgenic mice expressing an active form of Akt (myr-Akt) in β-cells in mice have larger β-cells and bigger islets (15) and this ultimately favors the development of insulinomas (17). These transgenic mice are resistant to experimentally induced diabetes, but, paradoxically, their β-cells have a much increased basal apoptotic rate (15). Conceivably, this higher apoptotic rate is compensated by increased β-cell renewal to maintain an adequate β-cell mass. As NFκB activation can induce β-cell death (6) and because Akt stimulates NFκB (44), the increased apoptosis response observed in myr-Akt–expressing β-cells likely results from the stimulation of the NFκB pathway. Indeed, prevention of NFκB activation using a dominant-negative IκB mutant allows mice to resist streptozotocin-induced diabetes (12,45). NFκB inhibition might, however, not always be protective against diabetes, as indicated by the increased susceptibility of nonobese diabetic (NOD) mice to develop diabetes when their β-cells express the dominant-negative IκB mutant (45). Additionally, it has been shown that expression of a dominant-negative IκB mutant in β-cells in mice can inhibit glucose-stimulated insulin secretion (29).

The RIP-N mice appear to lack some of the defects associated with the models described above. The activation of Akt by fragment N in the β-cells of RIP-N mice is not accompanied by an increased basal apoptotic rate, most likely because fragment N blocks Akt from stimulating NFκB. Moreover, in contrast to mice expressing an active form of Akt in β-cells (17), RIP-N mice do not display islet and β-cell hyperplasia, they do not develop insulinomas, and have a normal lifespan. Finally, RIP-N mice have no defect in glucose-induced insulin secretion and display normal glucose homeostasis, even in very old animals. Transgenic mice overexpressing proteins of the IAP (inhibitor of apoptosis) family specifically in the β-cells have been generated. Similarly to RIP-N mice, their β-cells are less susceptible to apoptosis, and this is not accompanied by alterations in islet morphology and function (46). It will therefore be important to define if there is a link between fragment N and IAPs that could explain the protective function of fragment N in β-cells. A detailed characterization at the molecular level of the pathways regulated by fragment N might ultimately lead to the identification of new strategies to preserve β-cell mass.

This work was supported by grants from the Juvenile Diabetes Research Foundation and from the Swiss National Science Foundation.

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

We thank Jean-Christophe Stehle from the Department of Pathology, Faculty of Biology and Medicine, University of Lausanne, for his help in performing immunohistochemistry experiments. We also thank Wanda Dolci for performing the in vitro insulin secretion experiments and Dr. Onur Boyman for the gift of the anti-CD3 antibody. We thank the Transgenic Animal Facility of the Faculty of Biology and Medicine and the University Hospital (University of Lausanne) for generation of the transgenic mice.