Pancreatic β-cell Raf-1 is required for glucose tolerance, insulin secretion, and insulin 2 transcription

Abstract

Regulation of glucose homeostasis by insulin depends on pancreatic β-cell growth, survival, and function. Raf-1 kinase is a major downstream target of several growth factors that promote proliferation and survival of many cell types, including the pancreatic β cells. We have previously reported that insulin protects β cells from apoptosis and promotes proliferation by activating Raf-1 signaling in cultured human islets, mouse islets, and MIN6 cells. As Raf-1 activity is critical for basal apoptosis and insulin secretion in vitro, we hypothesized that Raf-1 may play an important role in glucose homeostasis in vivo. To test this hypothesis, we utilized the Cre-loxP recombination system to obtain a pancreatic β-cell-specific ablation of Raf-1 kinase gene (RIPCre^+/+:Raf-1^flox/flox) and a complete set of littermate controls (RIPCre^+/+:Raf-1^wt/wt). RIPCre^+/+:Raf-1^flox/flox mice were viable, and no effects on weight gain were observed. RIPCre^+/+:Raf-1^flox/flox mice had increased fasting blood glucose levels and impaired glucose tolerance but normal insulin tolerance compared to littermate controls. Insulin secretion in vivo and in isolated islets was markedly impaired, but there was no apparent effect on the exocytosis machinery. However, islet insulin protein and insulin 2 mRNA, but not insulin 1 mRNA, were dramatically reduced in Raf-1-knockout mice. Analysis of insulin 2 knockout mice demonstrated that this reduction in mRNA was sufficient to impair in vivo insulin secretion. Our data further indicate that Raf-1 specifically and acutely regulates insulin 2 mRNA via negative action on Foxo1, which has been shown to selectively control the insulin 2 gene. This work provides the first direct evidence that Raf-1 signaling is essential for the regulation of basal insulin transcription and the supply of releasable insulin in vivo.—Alejandro, E. U., Lim, G. E., Mehran, A. E., Hu, X., Taghizadeh, F., Pelipeychenko, D., Baccarini, M., Johnson, J. D. Pancreatic β-cell Raf-1 is required for glucose tolerance, insulin secretion, and insulin 2 transcription.

Raf-1 is a multifunctional protein with serine and threonine kinase activity. Raf-1 is a critical target of many growth factors in various cell types, including pancreatic β cells (1–4). The Raf family has only recently been investigated in the β cell. We demonstrated using a small-molecule inhibitor and dominant-negative mutants that Raf-1 is essential for β-cell survival in vitro under serum-free culture conditions, suggesting that it mediates the protective signals from an autocrine/paracrine islet growth factor (2, 3). Indeed, we found that Raf-1 is involved in the antiapoptotic and mitogenic effects of insulin on cultured β cells (1, 3). Blocking Raf-1 signaling in transformed β-cell lines or in rat islets also reduced basal and glucose-induced insulin secretion (2, 5). ERK1/2, the canonical downstream target of Raf-1, has also been suggested to regulate insulin secretion (6, 7). There is a substantial, albeit controversial, literature suggesting that insulin can modulate the expression of its own gene (8–10). Given the proposed roles for ERK in insulin transcription (11–14) and the recent observation that an insulin receptor/ERK/B-Raf complex acts directly on insulin-regulated genes (15), it is also possible that Raf-1 might play a role in insulin synthesis. Interestingly, B-Raf, but not Raf-1, was recently shown to mediate the effects of glucose on ERK in MIN6 cells (16). In β cells, Raf-1 phosphorylates Bad at serine 112 and promotes its mitochondrial localization (3). Because Bad supports β-cell stimulus-secretion coupling via glucokinase (17), Raf-1 may also play a role in glucose signaling via Bad. Thus, evidence points to a pleiotropic and critical role for Raf-1 and its targets in the β cell. Despite its potential importance, the in vivo roles of β-cell Raf-1 remain unknown.

Whole-body Raf-1 gene deletion causes midgestational embryonic lethality (18), rendering analysis of glucose intolerance and adult islet function impossible in those knockout mice. To mitigate this problem, the Cre/loxP system of conditional gene ablation has been used to produce tissue-specific Raf-1 deletion in vivo (19, 20). This technology has already demonstrated that cardiac disruption of Raf-1 causes heart dysfunction and apoptosis (19). Here, we tested the role of Raf-1 in pancreatic β cells in vivo, by crossing a mouse line with a targeted Raf-1 allele with a mouse line expressing Cre recombinase driven by the insulin 2 promoter. We show that, compared to all littermate controls, mice lacking Raf-1 in their β cells (RIPCre^+/+: Raf-1^flox/flox) have impaired glucose tolerance due to reduced insulin expression and secretion. Our data demonstrate that Raf-1 plays important and unexpected roles in β-cell function in vivo.

MATERIALS AND METHODS

Mice

Procedures were approved by the University of British Columbia Animal Care Committee. Mice harboring one allele of Cre-recombinase under the rat insulin 2 promoter (generously provided by Dr. Pedro Herrera, University of Geneva, Genvea, Switzerland) were bred with mice carrying a floxed Raf-1 exon 3 allele (18). Cre-mediated deletion of exon 3 results in a frameshift mutation, thereby generating Raf-1 mRNA with a premature stop codon. This would be expected to delete Raf-1 in pancreatic β cells and other insulin 2-expressing tissues, most notably the brain (refs. 21, 22 and http://www.brain-map.org). Male RIPCre^+/−:Raf-1^flox/wt and female RIPCre^−/−:Raf-1^flox/wt mice were bred to generate homozygous β-cell-knockout mice (RIPCre^+/+:Raf-1^flox/flox) and heterozygous β-cell Raf-1-knockout mice (RIPCre^+/+:Raf-1^flox/wt) and RIPCre^+/+:Raf-1^wt/wt littermate controls (see Fig. 1). Insulin 2 knockout mice on the C57BL/6J background were a generous gift from Professor Jacques Jami (INSERM, Paris, France).

Figure 1.

Generation of pancreatic β-cell-specific Raf-1-knockout mice. A) Differential expression of insulin receptor signaling and MAP kinase components between MIN6 cells and primary islets. Western blots are shown for insulin receptor (IR), B-Raf, and Pdx1 (Actin¹) loading control), as well as IGF1 receptor (IGF1R), Raf-1, ERK, and Bad (Actin² loading control). Quantitative comparisons between human and mouse cells are not valid because of possible differences in antibody-antigen affinities. B) Breeding scheme employed to generate experimental animals. C–E) Western blot analysis (C) and quantification of Raf-1 (D) and B-Raf (E) in isolated islets from 12-wk-old RIPCre^+/−:Raf-1^flox/flox (KO) and control RIPCre^+/−:Raf-1^wt/wt (WT) mice (n=3). *P < 0.05. F) Expression of Raf-1 in fat (WT 0.73±0.32 vs. KO 0.41±.02), liver (WT 2.09±0.52 vs. KO 2.21±0.45), and soleus muscle (WT 0.64±0.21 vs. KO 0.80±0.26) between WT and KO mice (n=3). G) Western blot analysis of Raf-1 expression in hypothalamus (WT 0.36±0.02 vs. KO 0.21±0.05) and whole brain (WT 0.58±0.02 vs. KO 0.54±0.02) between WT and KO mice (n=3).

Glucose homeostasis and insulin secretion analysis

Glucose and insulin tolerance tests were performed by intraperitoneal delivery of 2 g/kg glucose or 0.75 U/kg insulin (Humalog; Eli Lilly, Indianapolis, IN, USA) to mice after 6 h of fasting. Blood glucose was monitored for 2 h after glucose or insulin delivery. Blood samples for glucose-stimulated insulin secretion were collected after intraperitoneal delivery of glucose (2 g/kg) to mice after 6 h of food deprivation. Insulin levels were measured using a rat insulin ELISA kit (Crystal Chem, Downers Grove, IL, USA). Mouse islets were isolated using collagenase and filtration as described previously (3). Insulin secretion per islet was analyzed in vitro using the perifusion technique (23). Static incubation experiments were conducted in 25 mM glucose-containing DMEM supplemented with 10 μM ALLM (Calbiochem/EMD, Gibbstown, NJ, USA), which has been previously shown to potentiate insulin secretion in short-term experiments (24).

Quantitative real-time PCR

Mouse MIN6 insulinoma cells were cultured as described previously (2) in 25 mM glucose DMEM supplemented with 10% FBS and penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA). All experiments were performed with multiple passages to ensure reproducibility. Total RNA was isolated from MIN6 cells or mouse primary islets using the Qiagen RNeasy Plus kit (Qiagen, Mississauga, ON, Canada). cDNA was generated with the qScript cDNA synthesis kit (Quanta Biosciences, Gaithersburg, MD, USA). SYBR green primer pairs or Taqman primer and probe sets (Applied Biosystems, Carlsbad, CA, USA) were used to measure transcript levels. All qPCR experiments were performed with the PerfeCTa qPCR SuperMix, ROX, or PerfeCTa SYBR Green, Supermix, ROX (Quanta Biosciences) on a StepOnePlus Real-Time PCR System (Applied Biosystems). All values were normalized to β-actin via the 2^−ΔCt method.

Protein detection by immunoblot and radioimmunoassay

Mouse islets, skeletal muscle, fat, liver, brain, and hypothalamus were washed twice with ice-cold PBS prior to lysis with RIPA buffer (50 mM β-glycerol phosphate, 10 mM HEPES, 1% Triton X-100, 70 mM NaCl, 2 mM EGTA, 1 mM Na₃VO₄, and 1 mM NaF) supplemented with complete EDTA-free protease inhibitor cocktail (Roche Applied Science, Laval, QC, Canada). In separate experiments, cytoplasmic and nuclear fractions were collected using the PARIS kit (Ambion, Austin, TX, USA), as per the manufacturer's instructions. Whole-cell lysates were sonicated prior to protein quantification by the Bradford method. Membranes were probed with antibodies against Raf-1 (BD Biosciences, Franklin Lakes, NJ, USA), pERK1/2, ERK1/2, pFoxo1, Foxo1, SNAP25 (Cell Signaling Technology, Danvers, MA, USA), VAMP2 (Synaptic Solutions, Göttingen, Germany), β-tubulin, Syntaxin1 (Sigma-Aldrich, St. Louis, MO, USA), Syntaxin4 (Millipore, Billerica, MA, USA), Lamin A/C (Santa Cruz Biotechnology, Santa Cruz, CA, USA), and β-actin (Novus Biologicals, Littleton, CO, USA). Insulin content was measured from total mouse islet lysates by radioimmunoassay (Millipore) and normalized to overall protein concentration.

Pancreas morphology, β-cell mass, and apoptosis

Paraformaldehyde-fixed pancreas tissues were embedded into paraffin, and sections were obtained for analysis of β-cell mass and islet morphology. Sections were deparaffinized, rehydrated, and incubated with serum-free Protein Block (Dako, Burlington, ON, Canada). Sections were incubated overnight at 4°C with antibodies against insulin (Linco Research, St. Charles, MO, USA), glucagon (Sigma-Aldrich), pThr202/Tyr204 ERK1/2, or ERK1/2 (Cell Signaling Technology), followed by secondary antibodies conjugated to Texas Red or FITC (Jackson Immunoresearch, West Grove, PA, USA), or to AlexaFluor-647 (Invitrogen). DAPI-containing mounting medium (Vector Laboratories, Burlingame, CA, USA) was added to coverslips. Apoptosis was measured by TUNEL labeling of fragmented DNA in insulin-containing cells (Roche Applied Science). To assess β-cell proliferation, control and knockout mice were injected with BrDU prior to euthanasia, and the pancreas secretions were stained for BrDU incorporation (Roche Applied Science) and analyzed with insulin colabeling. All images were taken with identical magnifications using a ×40 objective on a Zeiss 200M inverted microscope (Carl Zeiss, Oberkochen, Germany) and SlideBook software (Intelligent Imaging Innovations, Denver, CO, USA). The area of the whole pancreas (DAPI area) and the β-cell area were used to approximate β-cell mass.

For live-cell imaging of apoptosis in vitro, mouse islets were dispersed (5000–10,000 cells/well) at 24 h after isolation and allowed to attach in a 96-well Viewplate (Packard, Meriden, CT, USA), and left to grow overnight before experiments. Treatments were prepared in medium containing 2.5 ng/μl of propidium iodide (PI; Sigma-Aldrich) and 5 mM glucose. PI incorporation was measured every hour using a high-throughput imaging system (Thermo/Cellomics, Pittsburgh, PA, USA), as detailed previously (25). Raf-1 inhibitor (Raf-1i; GW5074) was from Calbiochem/EMD.

Statistical analysis

Data were analyzed by Student's t test or ANOVA followed by post hoc analysis, where appropriate. In some experiments, Student's paired t test was used. Results were considered statistically significant at values of P < 0.05.

RESULTS

Protein levels of Raf-1 and associated proteins in primary β cells

Our previous data suggest important roles for Raf-1 in cultured primary β cells and transformed MIN6 cells (1–3). We hypothesized that cell lines might have altered growth factor signaling pathways relative to primary cells. As a step toward understanding whether the roles of Raf-1 and its associated proteins are distinct in primary β cells, we performed a side-by-side comparison of protein levels of Raf-1 and other key proteins in its associated pathways. Protein levels of insulin receptor, IGF-1 receptor, Raf-1, B-Raf, ERK, and Bad were all lower in primary islets when compared to MIN6 cells (Fig. 1A). Pdx1 protein levels were similar. It is not valid to directly compare levels between mouse and human islets due to the likelihood of differential antibody-antigen affinity. We can report a significant degree of heterogeneity between batches of human islets.

Generation of pancreatic β-cell-specific Raf-1-knockout mice

We have previously reported that insulin protects primary β cells from apoptosis and promotes β-cell proliferation via Raf-1 signaling in serum-free culture conditions (1, 3). To test the hypothesis that Raf-1 plays important in vivo roles in pancreatic β cells, we conditionally targeted the Raf-1 gene in β cells using a well-described transgenic mouse line expressing Cre recombinase driven by a 668-bp promoter fragment of the rat insulin 2 gene (RIP-Cre; ref. 22). Mice harboring the RIP-Cre transgene have been shown to have altered glucose homeostasis on some genetic backgrounds, depending on the number of backcrosses, as well as the age and sex of the mice (26, 27). Thus, we employed a labor-intensive breeding scheme that had the potential to provide all of the proper controls within individual litters (Fig. 1C). Male RIPCre^+/−:Raf-1^flox/wt mice were crossed with female RIPCre^−/−:Raf-1^flox/wt mice. Using this approach, we obtained Cre-only littermate controls (RIPCre^+/+:Raf-1^wt/wt) to control for the effects of the Cre transgene, controls containing only the floxed alleles (RIPCre^−/−:Raf-1^flox/flox), and controls that were wild-type at both loci (RIPCre^−/−:Raf-1^wt/wt), as well as a few heterozygous knockouts (RIPCre^+/+: Raf-1^wt/flox). This full set of littermate controls is required to confidently assess the role of Raf-1 gene deletion per se.

Immunoblotting indicated that Raf-1 protein was markedly reduced by ∼80% in islets isolated from RIPCre^+/+:Raf-1^flox/flox mice compared to littermate control RIPCre^+/+:Raf-1^wt/wt mice (Fig. 1C, D). Since β cells contribute to ∼80% of islet cell protein, this reflected a near-complete deletion of Raf-1. We did not observe significant changes in B-Raf protein levels (Fig. 1C, E). Quantitative real-time PCR was performed on isolated RNA from RIPCre^+/+:Raf-1^flox/flox and control islets, and no changes in transcript levels of A-Raf or B-Raf were detected (A-Raf: RIPCre^+/+:Raf-1^wt/wt 0.031±0.006 vs. RIPCre^+/+:Raf-1^flox/flox 0.024±0.001; B-Raf: RIPCre^+/+:Raf-1^wt/wt 0.026±0.002 vs. RIPCre^+/+:Raf-1^flox/flox 0.023±0.001). Collectively, these data demonstrate that we were successful in generating pancreatic β-cell Raf-1-knockout mice without compensation from other Raf isoforms. Furthermore, we assessed whether Raf-1 deletion was specific to pancreatic islets by measuring Raf-1 protein levels in peripheral tissues, where no significant differences were detected (Fig. 1F, G).

We next examined the phosphorylation status of ERK, the canonical target of Raf-1, in our knockout islets. Interestingly, ERK phosphorylation at Thr202 and Tyr204 was not reduced following Raf-1 deletion in mouse islets (Fig. 2A). Moreover, cellular localization of phospho-ERK1/2 was comparable between RIPCre^+/+:Raf-1^flox/flox mice and littermate control RIPCre^+/+: Raf-1^wt/wt mice (Fig. 2B). No effects of Raf-1 deletion on total ERK levels were observed (data not shown). Thus, any phenotype in our model must be independent of reduced Raf-1-dependent ERK phosphorylation. These observations are reminiscent of the global Raf-1-knockout mice, which also did not show a decrease in ERK phosphorylation (18).

Figure 2.

Basal activation of ERK1/2 in β cells of RIPCre^+/+:Raf-1^flox/flox mice. A) Islets from 8-wk-old RIPCre^+/+:Raf-1^flox/flox and control RIPCre^+/+:Raf-1^wt/wt mice were isolated, lysed, and subjected to SDS-PAGE analysis for pERK1/2 and ERK (n=3–4). B) Pancreatic sections from 8-wk-old RIPCre^+/+:Raf-1^flox/flox and control RIPCre^+/+:Raf-1^wt/wt mice were stained for insulin (gray), pERK1/2 (green), and ERK1/2 (red). Nuclei were visualized with DAPI. Scale bars = 10 μm.

Body weight and glucose homeostasis of β-cell-specific Raf-1-knockout mice

Compared to all of the types of littermate controls, male and female RIPCre^+/+:Raf-1^flox/flox mice were equally viable and were grossly indistinguishable. RIPCre^+/+:Raf-1^flox/flox mice developed to adulthood, were fertile, and appeared to exhibit a normal life span. Because RIP-Cre has been shown to be active in the brain (28), with potential effects on food intake and obesity, we paid special attention to body weight. However, no significant differences in body weight between RIPCre^+/+:Raf-1^flox/flox and littermate control mice were observed in either males (not shown) or females (Fig. 3A). Thus, if even a small amount of Raf-1 was deleted in the brain by the RIP-Cre transgene, it was inconsequential to the regulation of body weight in these mice.

Figure 3.

Body weight and fasting blood glucose levels of RIPCre^+/−:Raf-1^flox/flox mice. Body weight (A) and 6 h fasting blood glucose levels (B) were measured at 8 wk of age (n=6). *P < 0.05 vs. littermate controls).

Next, we examined glucose homeostasis in RIPCre^+/+:Raf-1^flox/flox mice. Male RIPCre^+/+:Raf-1^flox/flox mice showed no significant differences in fasting blood glucose levels, although a nonsignificant trend toward glucose intolerance was observed (n=5, not shown). However, female RIPCre^+/+:Raf-1^flox/flox displayed elevated fasting blood glucose levels when compared to controls (Fig. 3B). Interestingly, this observation was most pronounced in younger mice (6–21 wk; not shown). Female mice were used for the remainder of the present study to examine the role of Raf-1 in pancreatic β cells and the mechanisms of the effects of Raf-1. Following an intraperitoneal glucose challenge, RIPCre^+/+:Raf-1^flox/flox mice displayed significant glucose intolerance at 8 wk of age (Fig. 4A). Intermediate glucose intolerance was observed even in heterozygous RIPCre^+/+:Raf-1^flox/wt mice, suggesting that these effects were dependent on gene dosage. These results demonstrate that β-cell Raf-1 is essential for normal glucose tolerance, particularly in female mice.

Figure 4.

Glucose tolerance of RIPCre^+/−:Raf-1^flox/flox mice. A) Four groups of female littermate mice (RIPCre^−/−:Raf-1^wt/wt, light gray triangles; RIPCre^−/−:Raf-1^wt/wt, open circles; RIPCre^−/−:Raf-1^flox/wt, dark gray circles; and RIPCre^+/−:Raf-1^flox/flox, solid circles) were subjected to glucose tolerance tests using intraperitoneal delivery of glucose (2 g/kg; 6 h fast) at 8 wk. Blood glucose level was monitored for 2 h after glucose delivery (n=6/group). *P < 0.05 for RIPCre^+/−:Raf-1^flox/flox vs. RIPCre^+/−:Raf-1^wt/wt mice). B) Area under the curve (AUC) was measured as the cumulative percentage of fasting blood glucose level (time 0) among all genotypes (n=6). *P <0.05 vs. other genotypes; ANOVA.

To assess whether the difference in glucose tolerance could be attributed to changes in insulin sensitivity, perhaps related to Cre expression in the brain, we assessed insulin tolerance following an intraperitoneal injection of 0.75 U/kg insulin. No significant differences were detected between RIPCre^+/+:Raf-1^flox/flox mice and their littermate controls (Fig. 5). This result suggested that the glucose intolerance in RIPCre^+/+: Raf-1^flox/flox mice was likely due to defects in the pancreatic β cells, rather than effects on peripheral tissues, such as the brain.

Figure 5.

Insulin tolerance of RIPCre^+/−:Raf-1^flox/flox mice. A) Four groups of female littermate mice (RIPCre^−/−:Raf-1^wt/wt, light gray triangles; RIPCre^−/−:Raf-1^flox/flox, open triangles; RIPCre^+/−:Raf-1^wt/wt, open circles; and RIPCre^+/−:Raf-1^flox/flox, solid circles) were subjected to insulin tolerance tests using intraperitoneal delivery of insulin (0.75 U/kg; 6 h fasting) at 8 wk. Blood glucose was monitored for 1 h after insulin delivery. Blood glucose corrected from percentage of basal glucose levels (n=6/genotype). B) AUC was measured as the cumulative percentage of fasting blood glucose level (time 0; n=6/genotype).

Islet morphology, β-cell apoptosis, and β-cell proliferation

Because our results to this point suggested that pancreatic β cells were the source of the glucose intolerance in RIPCre^+/+:Raf-1^flox/flox mice and previous studies from our group and others have suggested antiapoptotic and proliferative roles for Raf-1 (1, 2, 4), we next examined islet architecture, β-cell mass, and β-cell apoptosis by fluorescence microscopy. No differences in islet morphology or islet size were evident (Fig. 6A) between RIPCre^+/+:Raf-1^flox/flox mice and control RIPCre^+/+: Raf-1^wt/wt mice. Raf-1 deletion in β cells had no significant effect on total pancreatic β-cell area measured at 8 wk of age (Fig. 6B). The morphology, relative size, and yield of islets isolated from RIPCre^+/+:Raf-1^flox/flox mice were also similar to those from controls (Fig. 6C). Analysis of β-cell proliferation by BrdU incorporation showed no differences between RIPCre^+/+:Raf-1^flox/flox mice and control RIPCre^+/+:Raf-1^wt/wt mice (data not shown). β-Cell apoptosis, as assessed by TUNEL⁺ cells, was statistically similar between groups (Fig. 6D). As our previous in vitro data showed that Raf-1 inhibition potentiates serum withdrawal-induced apoptosis, we repeated these inhibitor experiments in serum-containing culture conditions that are more comparable to the in vivo situation (Fig. 6E). Whereas the Raf-1 inhibitor GW5074 induced massive cell death in serum-free conditions, little effect was observed in the presence of serum. Together with our previous studies (2, 3), these observations suggest that Raf-1 is essential for β-cell survival when the sole source of growth factors (i.e., insulin) is the islet cells themselves. They also suggest that Raf-1 may play a more important role in β-cell survival under stressed conditions.

Figure 6.

Islet morphology and β-cell proliferation and apoptosis of RIPCre^+/−:Raf-1^flox/flox mice. A) Representative insulin (green) and glucagon (red) staining in islets of RIPCre^+/−:Raf-1^flox/flox, RIPCre^+/−:Raf-1^flox/wt, and RIPCre^−/−:Raf-1^wt/wt mice (original view, ×40). B) Quantification of β-cell area (%) as calculated by the percentage ratio of total islet area to pancreas area among RIPCre^+/−:Raf-1^flox/flox, RIPCre^+/−:Raf-1^flox/wt, and RIPCre^−/−:Raf-1^wt/wt mice (n=3). C) Light microscopy images (original view, ×4) of islets isolated from RIPCre^+/−:Raf-1^flox/flox and RIPCre^+/−:Raf-1^wt/wt. D) β-Cell apoptosis in RIPCre^+/−:Raf-1^flox/flox mice and control RIPCre^+/−:Raf-1^wt/wt mice, as assessed by TUNEL staining (original view, ×10; n=3). E) Real-time measurement of propidium iodide incorporation in dispersed WT mouse islet cells treated with Raf-1i (GW5074) with or without the presence of serum (n = 3). *P < 0.05 vs. control; Student's t test.

Insulin secretion defect in β-cell Raf-1-knockout islets

As RIPCre^+/+:Raf-1^flox/flox mice displayed glucose intolerance but normal insulin sensitivity, we examined insulin secretion from β cells in vivo. RIPCre^+/+:Raf-1^flox/flox mice and control mice were injected with 2 g/kg glucose. Basal and glucose-stimulated plasma insulin levels were markedly lower when compared to controls (Fig. 7A). To remove the influence of hormonal and neural stimuli, and to compare secretion on a per-islet basis, islets were isolated from RIPCre^+/+:Raf-1^flox/flox and control mice and subjected to perifusion analysis. Glucose-induced insulin secretion from RIPCre^+/+: Raf-1^flox/flox mice was strongly reduced in this assay. KCl-induced insulin secretion was also robustly inhibited, which points to either a distal exocytotic defect or a reduced amount of insulin available for release (Fig. 7B). Similar defects in glucose- and KCl-stimulated insulin secretion were also observed in heterozygous RIPCre^+/+:Raf-1^wt/flox mice, pointing out the critical importance of Raf-1 in these processes. Similar results were seen in preliminary studies with islets from male RIPCre^+/+:Raf-1^flox/flox mice (data not shown). We also analyzed the effect of acutely blocking Raf-1 activity on insulin secretion in vitro. Indeed, blocking Raf-1 action with GW5074 for 3 h inhibited insulin secretion in medium containing 25 mM glucose (Fig. 7C). Together, these data strongly suggest that Raf-1 plays an important role in glucose-stimulated insulin secretion.

Figure 7.

Insulin secretion defect in RIPCre^+/+:Raf-1^flox/flox mice. A) RIPCre^+/+:Raf-1^wt/wt (open circles) and RIPCre^+/+:Raf-1^flox/flox mice (solid circles) were deprived of food for 6 h before glucose-stimulated insulin secretion studies were performed by intraperitoneal delivery of glucose (2 g/kg) at 8 wk (n=3). *P < 0.05 vs. corresponding WT. B) Islets isolated from RIPCre^+/−:Raf-1^flox/flox (closed circles), RIPCre^+/−:Raf-1^flox/wt (gray circles), and RIPCre^+/−:Raf-1^wt/wt littermates (open circles) were perifused with Krebs-Ringer buffer containing 3 mM glucose. No differences were seen in absolute values of basal insulin secretion. Islets were exposed to 20 mM glucose (striped bar) and 30 mM KCl (gray bar). Values are normalized to the pretreatment levels of insulin secretion to compensate for uneven numbers of islets in each column (n=3 independent cultures from female mice, 6–8 wk of age). C) Insulin levels were measured in conditioned medium of MIN6 cells treated with Raf-1i (GW5074) for 3 h (n=3). *P < 0.05. D) Western blot analysis of SNAP25, Syntaxin-1, Syntaxin-4, and VAMP2 in isolated islets from RIPCre^+/+:Raf-1^flox/flox (KO) and RIPCre^+/+:Raf-1^wt/wt (WT) (n=3).

Next, we examined the molecular mechanisms underlying the defect in insulin secretion. Raf-1/ERK1/2 signaling has been proposed to regulate SNARE proteins (29–31), which would affect insulin secretion (32). No differences in the total protein expression of SNAP25, VAMP2, Syntaxin 1, or Syntaxin 4 were detected (Fig. 7D). While it is possible that Raf-1 deletion may have altered the subcellular localization of SNARE proteins or their post-translational modifications, these data and the magnitude of the secretory defect prompted us to examine the possibility that Raf-1 controls the amount of the insulin available for release.

Raf-1 is required for insulin synthesis

Another mechanism that could account for reduced insulin secretion would be a robust reduction in insulin synthesis. Indeed, RIPCre^+/+:Raf-1^flox/flox islets had a significant reduction in insulin content (Fig. 8A). To determine the mechanism of this effect, insulin 1 and insulin 2 transcript levels were measured by quantitative real-time PCR. A dramatic reduction (81±9%) in insulin 2 mRNA was observed in RIPCre^+/+:Raf-1^flox/flox islets (Fig. 8B), as expected from the fact that the majority of islet insulin protein is derived from the insulin 2 gene (33). No changes in insulin 1 levels were observed. Next we examined the gene expression of transcription factors that regulate the production of insulin. We observed a significant reduction in NeuroD1 mRNA in RIPCre^+/+:Raf-1^flox/flox islets compared to controls (Fig. 8C), but this is not likely to account for the dramatic reduction in insulin 2 mRNA (34). We did not detect any difference in Pdx1 or Maf-A mRNA levels between the RIPCre^+/+:Raf-1^flox/flox mice and controls. To corroborate our findings in mouse islets, MIN6 cells were treated for 24 h with the Raf-1 inhibitor, GW5074, and transcript levels were measured by qPCR. Inhibition of Raf-1 significantly reduced insulin 2, NeuroD1, and Maf-A gene expression under basal conditions (Fig. 8D–G); however, this inhibitory effect was blocked in high glucose conditions, suggesting that the positive effects of glucose on insulin gene transcription are not mediated by Raf-1. Indeed, it has recently been suggested that B-Raf is the dominant isoform in β-cell glucose signaling (16). Until recently, it was not clear whether any factors regulated the expression of insulin 1 and insulin 2 independently, despite reports that they have distinct spatial and temporal distribution patterns and different promoters (35). Recently, it was reported that the nuclear exclusion of the transcription factor Foxo1 is required for the expression of insulin 2, but not insulin 1 (36). Western blotting demonstrated that Raf-1 inhibition caused a 65% decrease in Foxo1 phosphorylation (Fig. 8H), which excludes Foxo1 from the nucleus and abrogates its negative activity on the insulin 2 gene (36). Interestingly, this effect was also absent under high glucose conditions. Cell fractionation studies in primary islets confirmed that nuclear Foxo1 levels were more than doubled in RIPCre^+/+:Raf-1^flox/flox islets (Fig. 8I), suggesting a role of Raf-1 in regulating Foxo1 translocation. Thus, Raf-1 may selectivity regulate insulin 2 via Foxo1 dephosphorylation and export of Foxo1 from the nucleus. Taken together, these data suggest that the decrease in insulin secretion following glucose and KCl challenges may be due to a failure in synthesis of mature insulin from the insulin 2 gene.

Figure 8.

β-Cell deletion of Raf-1 impairs insulin gene transcription and protein synthesis. A) Insulin content was measured from lysed RIPCre^−/−:Raf-1^flox/flox (WT) and RIPCre^+/+:Raf-1^flox/flox (KO) islets by radioimmunoassay and normalized to lysate protein concentration (n=3–4). B) Quantitative PCR was performed on isolated mRNA from WT and KO islets for insulin 1 and insulin 2. All data were normalized to levels of β-actin (n=4–6). C) Levels of NeuroD1, Pdx1, and Maf-A mRNA were measured from purified RNA from WT and KO islets by qPCR. Transcript levels were normalized to β-actin (n=4–6). D–G) MIN6 cells were treated with Raf-1i (GW5074) for 24 h in the presence of low (5 mM) and high (25 mM) glucose. Isolated RNA was subjected to qPCR for insulin 2 (D), NeuroD1 (E), Maf-A (F), and Pdx1 (G) (n=3). *P < 0.05; Student's t test. H) Western blot analysis of FoxO1 phosphorylation following Raf-1 inhibition under low (5 mM) and high (25 mM) glucose conditions (n=4/ group). *P < 0.05 vs. untreated low-glucose control. I) WT and KO islet lysates were subjected to nuclear fractionation followed by immunoblot analysis for Foxo1. Data are expressed as fold change over WT levels and normalized to β-tubulin (cytoplasmic) or Lamin A/C (nuclear); n = 3/group. *P < 0.05 vs. WT in each fraction.

One might question whether such a reduction in insulin 2 mRNA would be sufficient to cause an impairment of in vivo glucose-stimulated insulin release, given the multiple levels of transcriptional, translational, and exocytotic control in this process. We tested this hypothesis directly by performing identical intraperitoneal glucose-stimulated insulin secretion tests on age-matched insulin 2 knockout mice. Indeed, loss of the insulin 2 gene was sufficient to impair insulin secretion (Fig. 7A). The reduction in insulin release seen with 100% loss of insulin 2 mRNA was proportional to that seen in RIPCre^+/+:Raf-1^flox/flox mice with 80% of the normal insulin 2 gene expression. Thus, it seems likely that the majority of reduced insulin secretion in RIPCre^+/+:Raf-1^flox/flox mice can be accounted for by a reduction in insulin 2 gene expression, although this does not rule out additional, more modest, contributions by other molecular defects to the phenotype.

DISCUSSION

Defining the genes that control pancreatic β-cell function and fate is a major goal of diabetes research. The objective of the present study was to understand the in vivo role of Raf-1 kinase in pancreatic β cells. We determined that β-cell Raf-1 is essential for normal glucose homeostasis in mice. In particular, Raf-1 is required for the full expression of the insulin 2 gene and is, therefore, required for physiological glucose-stimulated insulin secretion. These investigations uncovered novel and expected roles for Raf-1 in pancreatic β cells.

We have previously demonstrated the importance of the Raf-1 pathway in regulating the survival and proliferation of cultured β cells (2, 4); notwithstanding, the in vivo role of this kinase in β cells had not been determined. In the present study, we show for the first time that Raf-1 is required for glucose-induced insulin secretion in vivo. These data are consistent with our previous findings in vitro (2), which were subsequently confirmed by others (5). There are several reasons to believe that the elevated fasting glucose and glucose intolerance in RIPCre^+/+:Raf-1^flox/flox mice were likely caused by a defect in β-cell function. RIPCre^+/+:Raf-1^flox/flox mice displayed normal insulin sensitivity and body weight, and their islets displayed a profound defect, even after isolation. The effects on insulin secretion and synthesis could be recapitulated in vitro using a rapidly acting small-molecule inhibitor. Our studies suggest that the major mechanism for this impaired insulin secretion is a robust down-regulation of insulin 2 gene transcription, although it is also possible that Raf-1 may play direct roles in exocytosis, perhaps by modulating Rokα- or RAC1-dependent cytoskeleton reorganization (5, 20, 37). We also demonstrate clearly that a reduction in insulin 2 gene expression is sufficient to impair glucose-stimulated insulin secretion in vivo. Our finding that the insulin 2 gene, but not the insulin 1 gene, was controlled by Raf-1 was unexpected, but not without precedent. Indeed, it has recently been shown that the insulin 1 gene is selectively controlled by NeuroD1 in adult β cells (34). The promoters for the duplicated insulin 1 gene and the ancestral insulin 2 (corresponding to human INSULIN) are similar, but they have several clear differences (35). Elegant recent work demonstrated that nuclear exclusion and inactivation of the Foxo1 transcription factor is required to maintain insulin 2, but not insulin 1 gene expression in rodent β cells (36). In the present study, loss of Raf-1 function was associated with decreased phosphorylation of Foxo1 and elevated levels of Foxo1 in the nucleus, providing a plausible mechanism for the specific inhibition of insulin 2 in mouse β cells lacking Raf-1.

The defects in our model are likely mediated through pathways other than the canonical ERK cascade, as we observed no significant changes in basal ERK1/2 activity in RIPCre^+/+:Raf-1^flox/flox mice. This suggests that the defects in our model are mediated by pathways, which are either independent of Raf-1's positive kinase activity on ERK (38) or acting via other Raf-1 targets, such as Bad and Foxo1 (17). Our previous studies also pointed to important roles for non-ERK-dependent Raf-1 targets, such as Bad, in β cells (2, 3). Although ERK is a key downstream target of Raf-1 and a known regulator of insulin gene expression in β cells (13, 14), our results suggest that another Raf isoform or upstream kinase must control the effects of ERK on insulin transcription. Indeed, a recent study has demonstrated that glucose-induced ERK activation in β cells is primarily regulated by B-Raf or B-Raf:Raf-1 heterodimers (16), where Raf-1 may act to phosphorylate MEK1/2 or serve as a scaffold protein. In liver, a signaling complex including insulin receptor, B-Raf, and ERK acts on DNA to directly control insulin-sensitive gene expression (15). Indeed, the role of ERK as a positive or negative regulator of the insulin gene depends on the context of the experiment (13, 14), and as the defects in insulin 2 expression and insulin release are associated solely with loss of Raf-1 expression and activity, this further suggests the existence of an ERK-independent pathway. Additional studies are required to determine the precise mechanisms of signal coding in the Raf-ERK network.

In addition to playing a role in β-cell function, we had expected Raf-1 to play an important role in β-cell fate in nonstressed conditions in vivo. However, our data showing that RIPCre^+/+:Raf-1^flox/flox mice have normal β-cell mass, no significant increase in TUNEL⁺ cells, and no cell death after treatment with the Raf-1 inhibitor GW5074 in serum-containing conditions suggest that Raf-1 is essential for β-cell survival induced by endogenous growth factors produced by the β cells themselves (i.e., insulin) but not when β cells have access to a larger array of hormonal and growth factors. Under normal physiological conditions, it is possible that Raf-1-dependent autocrine insulin signaling (1, 3) is already maximally activated. It should also be noted that Raf-1 deletion tends to cause a compensatory increase in basal ERK activity in our model and others (18, 20), which might offset any prodeath signals that result from the loss of Raf-1 in vivo.

It should be noted that the Raf-1 pathway represents only one signaling node in the growth factor pathways modulating β-cell fate and function. Akt has been extensively studied as a possible downstream mediator of insulin, IGF-1, and other growth factors in pancreatic islets (39). While transgenic Akt1 overexpression in β cells led to increased β-cell mass (40, 41), genetic loss-of-function models have thus far failed to unequivocally confirm an essential role for Akt in β-cell survival in vivo. For example, Akt2-deficient mice show glucose intolerance, but also insulin resistance (42). Akt1-deficient mice exhibit placental hypotrophy, growth retardation, and reduced body weight (43). These studies show the relative importance of the different Akt isoforms in specific tissues, but the significance of each Akt in β-cell mass remains undefined. Interestingly, RIP-kdAkt mice lacking 80% of total Akt activity in β cells showed no increase in islet apoptosis and no loss of β-cell mass in normal or high-fat diet conditions (44). Instead, glucose-stimulated insulin secretion in RIP-kdAkt was significantly impaired, demonstrating an expected distal role for Akt kinase in insulin granule exocytosis (44). Recently, PI3-kinase has also been implicated in insulin exocytosis (45). Similar to the findings in the current study, these previous results point to unexpected roles of growth factor signaling genes in β-cell function in vivo.

Studies from our group and other groups have identified insulin signaling as a critical regulator of β-cell survival and proliferation in response to dietary stress (1, 46–49). We and others have also implicated insulin as a feedback regulator of insulin synthesis (8, 9). Together, our work and other studies strongly suggest that insulin and other local growth factors that activate the Raf-1 pathway play essential roles in regulating islet survival and function in vivo and in vitro. However, whether insulin stimulates or inhibits its own secretion remains controversial (8, 50, 51), and likely depends on the conditions, dose, and duration of insulin exposure (52).

In summary, our study defines the role of β-cell Raf-1 kinase in vivo and highlights its importance in glucose homeostasis and insulin synthesis. A greater understanding of Raf-1 function improves our knowledge of β-cell biology and may provide avenues to enhance β-cell function in diabetes. To the best of our knowledge, this is the first demonstration of a β-cell signaling protein that controls insulin secretion by controlling the amount of releasable insulin. This discovery adds an additional dimension to the multiple layers of regulation that control insulin release, glucose homeostasis, and diabetes at the level of the pancreatic β cell.