Research ArticleApoptosis

Cannabinoids Induce Pancreatic β-Cell Death by Directly Inhibiting Insulin Receptor Activation

+ See all authors and affiliations

Sci. Signal.  20 Mar 2012:
Vol. 5, Issue 216, pp. ra23
DOI: 10.1126/scisignal.2002519

Abstract

Cannabinoid 1 (CB1) receptors have been previously detected in pancreatic β cells, where they attenuate insulin action. We now report that CB1 receptors form a heteromeric complex with insulin receptors and the heterotrimeric guanosine triphosphate–binding protein α subunit Gαi. Gαi inhibited the kinase activity of the insulin receptor in β cells by directly binding to the activation loop in the tyrosine kinase domain of the receptor. Consequently, phosphorylation of proapoptotic protein Bad was reduced and its apoptotic activity was stimulated, leading to β-cell death. Pharmacological blockade or genetic deficiency of CB1 receptors enhanced insulin receptor signaling after injury, leading to reduced blood glucose concentrations and activation of Bad, which increased β-cell survival. These findings provide direct evidence of physical and functional interactions between CB1 and insulin receptors and suggest a mechanism whereby peripherally acting CB1 receptor antagonists improve insulin action in insulin-sensitive tissues independent of the other metabolic effects of CB1 receptors.

Introduction

Insulin secreted from pancreatic β cells activates a number of intracellular signaling pathways in virtually all mammalian cells, including β cells, that regulates not only energy homeostasis but also cellular proliferation and apoptosis. The actions of insulin are mediated by the insulin receptor, which is broadly distributed in normal tissues. The insulin receptor is composed of two extracellular α chains involved in ligand binding and two intracellular β chains that include the tyrosine kinase domain (1, 2). Insulin binding to the α chains induces a structural change that places the phosphorylation sites of one β chain within reach of the active site of the other β chain and facilitates autophosphorylation at Tyr1158, Tyr1162, and Tyr1163 in the activation loop of the β chains (3). Mutation of these tyrosine residues reduces insulin-stimulated autophosphorylation and kinase activity and results in a parallel loss of biological function (4, 5). The receptor also undergoes autophosphorylation at other tyrosine residues in the juxtamembrane region and the C-terminal tail (6, 7). Tyrosine phosphorylation increases the catalytic activity of the receptor and also serves as docking sites for downstream signaling proteins such as the insulin receptor substrates (IRSs) (8). A well-characterized signaling cascade that is activated by insulin is the IRS–phosphoinositide 3-kinase (PI3K)–AKT cascade, in which AKT is a critical mediator of insulin responses such as gene expression, protein synthesis, cell growth and survival, and glucose metabolism (8). AKT promotes cell survival and growth by phosphorylating the proapoptotic protein Bad (which results in inactivation of Bad) (9, 10), the transcriptional regulator FoxO (which results in inactivation of FoxO) (11, 12), and the cyclin-dependent kinase inhibitor p27 (which results in inactivation of p27) (1315). This is also true for pancreatic β cells because targeted mutations of genes in β cells that encode the insulin receptor and its downstream molecules such as IRS2, AKT, FoxO1, and p27 reduce β-cell growth or survival, resulting in age-dependent diabetes mellitus (1620). In addition, AKT-mediated phosphorylation of FoxO1 positively regulates insulin transcription, insulin secretion, and β-cell growth and survival by increasing the abundance of the pancreatic transcription factor pancreas/duodenum homeobox-1 (PDX-1) (16, 18).

Several groups have demonstrated that cannabinoid 1 (CB1) receptors and the necessary enzymes for catalyzing the biosynthesis and degradation of the endocannabinoids 2-arachidonoylglycerol (2-AG) and anandamide (AEA) are present in β cells of human and mouse islets (2124). We also confirmed in our previous study (25) that CB1 receptors are present in β cells and that β cells synthesize endocannabinoids in a glucose-dependent manner. The CB1 receptor is a heterotrimeric guanosine triphosphate (GTP)–binding protein (G protein)–coupled receptor that is activated by endocannabinoids, which are lipid transmitters synthesized “on demand” by Ca2+-dependent enzymes in the brain and the periphery (2527). Tetrahydrocannabinol, the main psychoactive compound in cannabis, is an exogenous ligand of CB1 receptors, which are distributed in several brain areas as well as hepatocytes (28) and muscle (29). Endocannabinoids induce cell cycle arrest and apoptosis by inhibiting the PI3K-AKT cascade in various cancer cells (3032). We and others have reported that endocannabinoids influence insulin action through regulation of insulin receptor signaling in insulin-sensitive tissues such as muscle, liver, and islets of Langerhans (25, 28, 33, 34). We now provide in-depth insight into the mechanisms by which the blockade of the CB1 receptor inhibits apoptosis in β cells. Previous research using CB1 receptor antagonists in animals had indicated that the resulting improvement in insulin action was due to weight loss (35), but we now provide evidence for direct crosstalk between CB1 and insulin receptors.

Results

Activation of CB1 receptors induces β-cell death in an insulin receptor–dependent manner

Activation of the CB1 receptor by the synthetic full agonist WIN55,212-2 or by endocannabinoids (AEA or 2-AG) decreased the viability (Fig. 1A), increased the cytotoxicity (Fig. 1B), and activated caspase-3 (Fig. 1C) in mouse insulinoma MIN6 cells in a dose-dependent manner. A similar but less pronounced response was seen in αTC1 glucagonoma cells, likely because CB1 receptors are less abundant in these cells than in β-cell lines (fig. S1). The CB1 receptor–mediated decrease in viability of MIN6 cells was reduced by Ac-DNLD-CHO, an inhibitor of caspase-3 and -7 (Fig. 1D).

Fig. 1

Activation of CB1 receptors induces caspase-3–mediated β-cell death in an insulin receptor–dependent manner. (A and B) Relative cell viability (A) and cytotoxicity (B) of αTC1 and MIN6 cells exposed to CB1 receptor agonists (WIN55,212-2, AEA, or 2-AG) in serum-free media (n = 3 independent experiments). (C) Activation of caspase-3 by WIN55,212-2 in MIN6 and αTC1 cells. αTC1 and MIN6 cells exposed to WIN55,212-2 in serum-free medium were subjected to immunostaining with anti-cleaved caspase-3 antibody. Scale bars, 50 μm. (D) Effect of the caspase-3 inhibitor Ac-DNLD-CHO on WIN55,212-2–induced decrease in MIN6 cell viability (n = 3 independent experiments). (E) Activation of caspase-3 by WIN55,212-2 in β cells from wild-type mice (βIRWT) and/or from mice with a β-cell–specific knockout of the insulin receptor (βIRKO) was assessed by immunostaining with anti-cleaved caspase-3 antibody. Scale bars, 50 μm. (F) Relative cell viability of βIRWT and βIRKO cells exposed to WIN55,212-2 in serum-free medium (n = 3 independent experiments). (G) Abundance of the indicated proteins in βIRWT and βIRKO cells exposed to WIN55,212-2. Relative densities for the indicated proteins are shown on the bottom (n = 3 independent experiments). p, phosphorylated. (H) Relative cell viability of βIRWT and βIRKO cells exposed to the CB1 receptor (CB1R) antagonists AM251, rimonabant, or SLV-319 (n = 3 independent experiments). Data represent the means ± SEM. *P < 0.05; **P < 0.01; n.s., not significant.

Because insulin receptor signaling is a key regulator of β-cell survival (1620) and because CB1 receptor agonism and antagonism influence insulin action (25, 33, 34), we next investigated the potential role of insulin receptors as mediators of CB1 receptor–controlled β-cell survival using β cells established from wild-type or β-cell–specific insulin receptor knockout mice (17, 36, 37). The effects of WIN55,212-2 on caspase-3 activation (Fig. 1E) and cell viability (Fig. 1F) were reduced in β cells from knockout mice compared to those from wild-type mice. WIN55,212-2 decreased, in a dose-dependent manner, the phosphorylation of IRS1/2 at Tyr612 and AKT at Ser473 in wild-type, but not insulin receptor–deficient, β cells (Fig. 1G). Lack of insulin receptor did not alter the abundance of the CB1 receptor (Fig. 1G). Consistently, the viability of wild-type, but not insulin receptor–deficient β cells, was increased by the CB1 receptor antagonists tested (AM251, rimonabant, and SLV-319) (Fig. 1H), and the effect of rimonabant was prevented by arachidonyl-2-chloroethylamide (ACEA), a selective CB1 receptor agonist (fig. S2A).

Activation of the CB1 receptor diminishes insulin-induced phosphorylation of Bad by inhibiting insulin receptor kinase activity

To test the effect of CB1 receptor activation on insulin-stimulated autophosphorylation of insulin receptor, we pretreated MIN6 cells with ACEA before addition of exogenous insulin. Consistent with our previous study (25), ACEA diminished exogenous insulin–stimulated autophosphorylation of the insulin receptor at Tyr1162 and Tyr1163 and phosphorylation of IRS1/2 and AKT (Fig. 2A). Knockdown of CB1 receptors by small interfering RNA (siRNA) (fig. S2B) abolished the ability of ACEA to inhibit insulin-stimulated phosphorylation of insulin receptor and AKT (fig. S2C). To further confirm whether the CB1 receptor interferes with insulin-stimulated autophosphorylation and kinase activity of insulin receptor, we carried out in vitro kinase assays with the recombinant p30 fragment of IRS1, which harbors the Tyr612 residue, and immune complexes of the β subunit of the insulin receptor after addition of exogenous insulin with vehicle or ACEA to wild-type β cells. We observed that activation of CB1 receptors inhibited insulin-stimulated autophosphorylation of insulin receptors, resulting in reduced kinase activity (Fig. 2B).

Fig. 2

Regulation of insulin-stimulated insulin receptor kinase activity and Bad phosphorylation by CB1 receptors in β cells. (A) Abundance of the indicated proteins in MIN6 cells exposed to insulin with ACEA or vehicle. Relative densities for the indicated proteins are shown on the right (n = 3 independent experiments). (B) In vitro kinase assay with IRβ immunoprecipitated from βIRWT cells exposed to insulin with ACEA or vehicle. Blots are representative of two independent experiments. IRS-p30, recombinant p30 fragment of IRS1. (C) Regulation of Bad function by AKT. (D) Regulation of insulin-induced Bad interaction with Bcl-xL or 14-3-3 by ACEA in βIRWT cells. Relative densities for the indicated proteins are shown on the right (n = 3 independent experiments). (E) Abundance of the indicated proteins in βIRKO cells exposed to ACEA after reconstitution with wild-type insulin receptor (IR-WT) or a mutant with Ala substitutions for Tyr1158, Tyr1162, and Tyr1163 (IR-3YA) (n = 3 independent experiments). (F) Bad abundance after siRNA transfection in βIRWT cells. (G) Effects of WIN55,212-2 on caspase-3 activation in βIRWT cells transfected with scrambled or Bad siRNA (n = 3 independent experiments). (H) Comparison of the viability of βIRWT cells transfected with scrambled or Bad siRNA (n = 3 independent experiments). (I) Effects of WIN55,212-2 on the viability of βIRWT cells transfected with the indicated siRNAs (n = 3 independent experiments). Data represent the means ± SEM. *P < 0.05; **P < 0.01; n.s., not significant.

Bad is a proapoptotic member of the Bcl-2 family that promotes cell death by displacing Bax from binding to Bcl-xL, and Bad activity is directly inhibited by AKT-mediated phosphorylation at Ser136 (9), which promotes binding of Bad to 14-3-3 protein instead of to Bcl-xL (Fig. 2C) (9, 10, 38). Thus, we examined the effect of CB1 receptor agonists on Bad activity in wild-type β cells. ACEA prevented the insulin-stimulated phosphorylation of Bad at Ser136, decreased the amount of Bad bound to 14-3-3, and increased the amount of Bad bound to Bcl-xL (Fig. 2D). To further confirm the inhibitory effects of the CB1 receptor on insulin-stimulated phosphorylation of Bad, we transfected Flag-tagged wild-type insulin receptor or an insulin receptor mutant with Ala substitutions for Tyr1158, Tyr1162, and Tyr1163 residues into insulin receptor–deficient β cells (Fig. 2E). Bad phosphorylation was detected in insulin receptor–deficient β cells reconstituted with wild-type insulin receptor, presumably due to endogenous insulin secretion, and treatment with ACEA reduced the phosphorylation of Bad in these cells. In contrast, phosphorylation of Bad was not altered in insulin receptor–deficient β cells reconstituted with the tyrosine phosphorylation–deficient insulin receptor mutant (Fig. 2E). Knockdown of Bad by siRNA (Fig. 2F) abolished the ability of WIN55,212-2 to activate caspase-3 (Fig. 2G), increased β-cell viability (Fig. 2H), and prevented the inhibitory actions of WIN55,212-2 on β-cell viability (Fig. 2I). Insulin receptor signaling was also higher in livers from CB1 receptor–null (CB1R−/−) mice (39) than in those from wild-type littermates, and WIN55,212-2 treatment decreased phosphorylation of insulin receptors in human hepatocarcinoma HepG2 cells (fig. S3, A and B). WIN55,212-2 treatment also increased cleaved caspase-3 and Bad activity and reduced cell viability (fig. S3, B to D). Similarly, viability of primary human hepatocytes was decreased by WIN55,212-2 treatment (fig. S3E). In sum, our results suggest that CB1 receptor signaling functions to inhibit insulin receptor signaling, including in non–insulin-secreting cells.

The CB1 receptor forms a heteromeric complex with the β subunit of the insulin receptor and Gαi

The CB1 receptor associated with the β subunit of the insulin receptor (IRβ) in wild-type β cells (Fig. 3A). This association was increased by treatment with ACEA (Fig. 3A) and decreased by treatment with exogenous insulin, an effect that was reversed by ACEA (Fig. 3B). In reconstituted insulin receptor–deficient β cells, the CB1 receptor showed greater association with the tyrosine phosphorylation–deficient mutant than with the wild-type receptor, an interaction that was increased by ACEA treatment (Fig. 3C). This result is consistent with the finding that exogenous insulin decreased IRβ association with CB1 receptor (Fig. 3B). Using a series of IRβ deletion mutants (Fig. 3D), we determined that the CB1 receptor primarily bound to the activation loop in the tyrosine kinase domain of IRβ (Fig. 3E).

Fig. 3

CB1 receptors interact with IRβ and Gαi. (A) Interaction of endogenous CB1 receptor with IRβ in βIRWT cells exposed to ACEA or vehicle. Relative density for IRβ is shown on the right (n = 3 independent experiments). (B) Interaction of endogenous CB1 receptor with IRβ in βIRWT cells exposed to insulin with or without ACEA. Relative density for IRβ is shown on the right (n = 3 independent experiments). (C) Increased interaction of the CB1 receptor with IRβ by substitution of Tyr1158, Tyr1162, and Tyr1163 with Ala. βIRKO cells transfected with empty vector, IR-WT, or IR-3YA were exposed to ACEA and subjected to coimmunoprecipitation assay. Relative density for IRβ is shown on the bottom (n = 2 independent experiments). (D) Schematic representation of Venus-tagged IRβ deletion mutants. (E) Domain-mapping studies in βIRKO cells transfected with Venus-IRβ deletion mutants. (F) Role of Gαi in the CB1 receptor–IRβ interaction. βIRKO cells transfected with scrambled or Gαi3 siRNA were reconstituted with IR-WT and subjected to coimmunoprecipitation with anti-CB1 receptor antibody. Blots are representative of two independent experiments. Data represent the means ± SEM. *P < 0.05.

Given that CB1 receptor–mediated activation in β-cell lines increased the activity of Gαi, which mediates the inhibitory effect of CB1 receptor activation on insulin-stimulated β-cell proliferation by its association with insulin receptor (25), we examined whether Gαi mediates the association of IRβ with CB1 receptor in wild-type β cells. We found that the CB1 receptor formed a heteromeric complex with IRβ and Gαi3, and siRNA-mediated silencing of Gαi3 reduced the association (Fig. 3F), suggesting that the CB1 receptor associates with IRβ through Gαi.

i mediates the inhibitory effects of CB1 receptors on insulin receptor kinase activity by direct binding to IRβ

Of the three subtypes of Gαi proteins, Gαi1 and Gαi3 were present mainly in β cells of both human and mouse, whereas Gαi2 was distributed mainly in α cells (fig. S4A). Thus, we examined the role of Gαi1 and Gαi3 on CB1 receptor–mediated inhibition of insulin receptor signaling. Gαi3 colocalized with insulin receptor at the cell membrane (Fig. 4A) and bound to the activation loop of IRβ (Fig. 4B). Using an in vitro binding assay, we found that glutathione S-transferase (GST)–IRβ pulled down Gαi1 and that more IRβ associated with active GTP-bound Gαi1 compared with inactive guanosine diphosphate (GDP)–bound Gαi1 (Fig. 4C), indicative of a direct and specific association of Gαi with IRβ. Because activation of CB1 receptors increased the activity of Gαi in β-cell lines (25), these results also suggest that Gαi mediates the inhibitory effect of CB1 receptor activation on insulin receptor kinase activity by its association with IRβ. Indeed, autophosphorylation and kinase activity of the insulin receptor were reduced by binding of recombinant GTP-bound Gαi1 in vitro (Fig. 4D). Overexpression of Gαi3 led to decreased tyrosine phosphorylation of the insulin receptor in β cells from knockout mice reconstituted with wild-type insulin receptor, but not those reconstituted with the tyrosine phosphorylation–deficient mutant (fig. S4B). Conversely, knockdown of Gαi3 by siRNA increased phosphorylation of the insulin receptor, AKT, and Bad in insulin receptor–deficient β cells reconstituted with wild-type receptor (fig. S4C) and abolished the ability of ACEA to inhibit insulin-stimulated phosphorylation of IRS1/2 and AKT in wild-type β cells (Fig. 4E). Moreover, knockdown of Gαi3 in wild-type β cells resulted in increased β-cell viability (Fig. 4F) and partially attenuated some of the inhibitory actions of ACEA (Fig. 4G).

Fig. 4

i inhibits insulin receptor kinase activity by direct insulin receptor binding. (A) Colocalization of IRβ and Gαi3 in βIRWT cells. Scale bar, 20 μm. (B) Interaction of Gαi3 with Venus–IRβ-1 or -3 in βIRKO cells. (C) In vitro binding assay carried out with GST-IRβ and GDP- or GTP-bound Gαi1. Blots are representative of two independent experiments. (D) Effects of recombinant Gαi1 on the insulin receptor kinase activity. Relative densities for the indicated proteins are shown on the bottom (n = 2 independent experiments). (E) Phosphorylation of IRS1/2 and AKT in βIRWT cells exposed to insulin with ACEA or vehicle after transfection of the indicated siRNAs. Relative densities for the indicated proteins are shown on the bottom (n = 3 independent experiments). (F) Comparison of the viability between βIRWT cells transfected with scrambled or Gαi3 siRNA (n = 3 independent experiments). (G) Effects of WIN55,212-2 on the viability of βIRWT cells transfected with the indicated siRNAs (n = 3 independent experiments). Data represent the means ± SEM. *P < 0.05; **P < 0.01; n.s., not significant.

CB1 receptor blockade improves β-cell growth and survival in streptozotocin-treated mice

We next investigated whether CB1 receptor modulation might be beneficial to β-cell survival after injury. To examine the ability of β cells to regenerate after multiple injections of low-dose streptozotocin (STZ) in young adult animals, we injected the CB1 receptor antagonist AM251 beginning 1 day after terminating STZ treatment of 2-month-old CD1 mice (Fig. 5A). Multiple injections of low-dose STZ cause selective β-cell destruction, which in turn induces immune reactions against islets, and, over time, the remaining β cells attempt to survive and proliferate (40, 41). Mice treated with dimethyl sulfoxide (DMSO) for 3 weeks after termination of STZ treatment (STZ-DMSO) had high blood glucose concentrations (>300 mg/dl) resulting from low insulin concentrations, rendering the mice overtly diabetic, whereas blood glucose and insulin concentrations in the STZ-treated counterparts given AM251 (STZ-AM251) were less affected (Fig. 5, B and C). Islet architecture in STZ-DMSO mice was also disrupted and was accompanied by reduced insulin staining density (Fig. 5D, upper panel) and β-cell mass (Fig. 5D, lower panel) compared to that of non–STZ-treated mice. In contrast, the islet architecture of STZ-AM251 mice had a close-to-normal appearance and the β-cell mass was close to that of non–STZ-treated mice (Fig. 5D). As previously reported (42, 43), multiple injections of low-dose STZ induced caspase-3 activation in the STZ-treated mice, and the STZ-AM251 mice had lower caspase-3 activity compared to their DMSO-treated counterparts (Fig. 5E). We also observed ongoing β-cell proliferation in the STZ-AM251 mice, as evidenced by the increase in proliferating cell nuclear antigen (PCNA)–positive nuclei (Fig. 5F).

Fig. 5

The CB1 receptor antagonist AM251 improves β-cell mass in STZ-treated mice because of enhanced β-cell survival. (A) Experimental timeline for STZ (50 mg/kg) and AM251 (10 mg/kg) treatment in 2-month-old wild-type CD1 mice. (B and C) Random blood glucose (B) and plasma insulin (C) concentrations of DMSO- and AM251-injected mice (n = 5 animals per group) after STZ treatment. (D) Representative images for insulin in islets of DMSO- and AM251-injected mice after STZ treatment. Scale bars, 200 μm. The fraction of pancreatic tissue area covered by β cells is shown on the bottom (n = 3 to 5 animals per group). (E) Representative images for cleaved caspase-3 in islets of control and STZ-treated mice of cohorts in (D). Scale bars, 50 μm. Relative signal intensity for cleaved caspase-3 in islets is shown on the bottom (n = 3 animals per group). (F) Representative images for PCNA-positive β cells of STZ-treated mice of cohorts in (D). Arrows denote PCNA-positive cells. Scale bar, 50 μm. Quantification of PCNA-positive β cells is shown on the right (n = 3 animals per group). Data represent the means ± SEM. *P < 0.05; **P < 0.01.

We further confirmed the effects of CB1 receptor modulation on β-cell survival and proliferation by injecting a low dose of STZ into CB1R−/− mice (fig. S5A). Consistent with the results in the STZ-AM251 mice, deletion of the gene encoding the CB1 receptor also resulted in lower blood glucose and increased plasma insulin concentrations (fig. S5, B and C); it also improved islet architecture and increased insulin content and β-cell mass, resulting from enhanced β-cell survival and proliferation (fig. S5, D to F) compared to STZ-treated CB1R+/+ mice. The combination of these morphological and metabolic data suggests that normalization of blood glucose and insulin concentrations, islet architecture, and β-cell mass by CB1 receptor antagonism after a diabetes-inducing injury occurs as a result of increased β-cell survival and proliferation.

CB1 receptor blockade enhances insulin receptor signaling in β cells of STZ-treated mice

We next evaluated whether the increased β-cell survival and growth seen by CB1 receptor antagonism in STZ-injected mice were associated with changes in insulin receptor signaling. Phosphorylation of the insulin receptor, IRS1/2, AKT, and FoxO1 was significantly increased in AM251-treated mice compared with DMSO-treated mice (Fig. 6, A and B), and STZ-treated CB1R−/− mice also showed increased phosphorylation of the insulin receptor, IRS1/2, AKT, and FoxO1, compared with STZ-treated CB1R+/+ mice (fig. S6, A and B). Moreover, phosphorylation of the insulin receptor, IRS1/2, AKT, and FoxO1 and intra-islet insulin content were higher in STZ-treated CB1R−/− mice (although islet size was reduced) compared with STZ-treated CB1R+/+ mice (fig. S6C). Consistent with increased phosphorylation of AKT caused by CB1 receptor antagonism, AM251-treated mice had increased phosphorylation of Bad compared with DMSO-treated mice despite similar total abundances of Bad (Fig. 6C). The same effects were observed in STZ-treated CB1R−/− mice (fig. S6D).

Fig. 6

AM251 enhances insulin receptor signaling in β cells of STZ-treated mice. (A) Representative images for insulin receptor and phosphorylated insulin receptor in islets of DMSO- and AM251-injected mice after STZ treatment. Scale bars, 50 μm. Relative signal intensity for phosphorylated insulin receptor in islets is shown on the right (n = 3 animals per group). (B) Representative images for p-IRS1/2, p-AKT, and p-FoxO1 in islets of cohorts in (A). Scale bars, 50 μm. Relative signal intensity for the indicated proteins in islets is shown on the right (n = 3 animals per group). (C) Representative images for Bad and p-Bad in islets of cohorts in (A). Scale bars, 50 μm. Relative signal intensity for p-Bad in islets is shown on the right (n = 3 to 5 animals per group). (D) Representative images for insulin and p27 in islets of cohorts in (A). Boxed areas were magnified, shown on the last panel, for better visualization. Scale bars, 50 μm. Relative p27 intensity and percentage of nuclear p27-positive β cells in islets are shown on the bottom (n = 3 to 5 animals per group). Data represent the means ± SEM. *P < 0.05; **P < 0.01.

We examined the abundance and subcellular localization of p27 because AKT also regulates p27 activity by affecting both its abundance and its subcellular localization through the FoxO family (12, 44) and by direct phosphorylation (14, 15). Furthermore, accumulation of p27 in the nuclei of β cells contributes to β-cell failure during the development of diabetes (19), and p27−/− mice show reduced susceptibility to STZ-induced diabetes (45). Immunostaining of pancreatic sections from AM251-treated mice showed a significant decrease in both the total amount and nuclear localization of p27 in β cells, and most p27 was localized in the cytoplasm (Fig. 6D). The same effects were also observed in STZ-treated CB1R−/− mice (fig. S6E). The decrease in nuclear p27 most likely resulted from a decrease in the total abundance of p27 as well as from increased phosphorylation of p27 at Ser10 or Thr157 (fig. S7, A and B). These modifications modulate p27 cytoplasmic localization and inhibit its function (14, 15), and the nonphosphorylated form of p27 accumulates predominantly in the nucleus of β cells of IRS2- and leptin receptor–null mice, causing diabetes as a result of deficient β-cell mass and proliferation (19). Together, these results suggest that inhibition of CB1 receptor signaling promotes β-cell survival as well as proliferation after a diabetes-inducing injury by facilitating increased insulin receptor signaling.

AKT-mediated phosphorylation of FoxO1 in β cells results in the increased abundance of PDX-1 (16), a transcription factor that promotes insulin gene transcription and the increased abundance of glucose transporter 2 (GLUT2) and glucokinase, part of the glucose-sensing machinery of β cells. The abundance of PDX-1, GLUT2, and glucokinase was increased in β cells of AM251-treated mice compared with that of DMSO-treated mice (Fig. 7, A to C). A similar pattern was evident in pancreatic sections from STZ-injected CB1R−/− mice (fig. S8, A to C), suggesting that CB1 receptors could contribute to β-cell function and survival by regulating the abundance of PDX-1, GLUT2, and glucokinase (Fig. 7D).

Fig. 7

Increased abundance of PDX-1, GLUT2, and glucokinase in β cells of AM251-injected mice after STZ treatment. (A to C) Representative images for PDX-1 (right-hand images are ×4 magnifications) (A), GLUT2 (B), and glucokinase (C) in islets of cohorts in (A). Scale bar, 50 μm. Relative signal intensities in islets are shown on the bottom of each image. Data represent the means ± SEM from n = 3 to 5 animals per group. *P < 0.05; **P < 0.01. (D) Schematic unifying the regulation of insulin receptor signaling by endocannabinoids upon cell growth and survival. Binding of endocannabinoids to CB1 receptors activates the Gαi class of heterotrimeric proteins and increases the association between Gαi and the insulin receptor that counteracts the effects of insulin on autophosphorylation and kinase activity of the insulin receptor, resulting in decreased cell growth and survival. In pancreatic β cells, CB1 receptor signaling through Gαi leads to decreased PDX-1, GLUT2, and glucokinase abundance.

Discussion

There is growing interest in the role of endocannabinoids in the regulation of cell death and survival. Their proapoptotic and antiproliferative effects have been reported in various cancer cells and, at least in part, result from inhibition of the PI3K-AKT cascade (3032). Our data suggest a model for the direct regulation of insulin receptor activity by CB1 receptors in which physical and functional crosstalk between the CB1 receptors and the insulin receptors directly influences cell survival and growth (Fig. 7D). CB1 receptors form a heteromeric complex with insulin receptor and Gαi when activated, which in turn impairs autophosphorylation and kinase activity of the insulin receptor. Gαi mediates the formation of the complex as well as the inhibitory effects of CB1 receptors on insulin receptor kinase activity by directly binding to the activation loop of the insulin receptor. This leads to reduced AKT-mediated phosphorylation of the proapoptotic protein Bad that in turn paves the way to cell death. We also found that WIN55,212-2 treatment not only decreased insulin receptor phosphorylation, increased cleaved caspase-3 and Bad activity, and decreased cell viability in human hepatocarcinoma HepG2 cells but also decreased cell viability in primary human hepatocytes, suggesting that the effects of endocannabinoids are not unique to pancreatic β cells.

CB1 receptor agonists have been reported to exert CB1 receptor–independent effects under certain conditions, but we believe this is unlikely in β cells because the receptor antagonist AM251 prevented the inhibitory effect of the agonist ACEA on β-cell proliferation and because siRNA directed against the CB1 receptor abolished the ability of its agonists to inhibit exogenous insulin–stimulated phosphorylation of the insulin receptor and β-cell proliferation (25). Additionally, the CB1 receptor antagonists that we tested increased the viability of β cells from wild-type mice in the nanomolar range (Fig. 1H), an effect that was prevented by ACEA (fig. S2A).

We and others have reported that endocannabinoids influence insulin action by influencing insulin receptor signaling in insulin-sensitive tissues (25, 33, 34). In skeletal muscle cells, insulin-stimulated phosphorylation of AKT was impaired by endocannabinoids and enhanced by CB1 receptor antagonists (33, 34), and CB1 receptors inhibited pancreatic β-cell proliferation in an insulin receptor–dependent manner (25). We had hypothesized that the close proximity of the CB1 receptors to insulin receptors would allow their involvement in influencing insulin receptor–mediated signaling (25), because these receptors as well as Gαi proteins are present within lipid rafts (4648), membrane microdomains with a distinct structural composition that appear to act as platforms for facilitating protein-protein interactions involved in intracellular signaling pathways (49). The association of the insulin receptor with the CB1 receptor was strengthened by a CB1 receptor agonist and by substitution of Tyr1158, Tyr1162, and Tyr1163 residues of the insulin receptor with Ala, but was attenuated by insulin, and the CB1 receptor primarily bound to the activation loop of IRβ, which contains the autophosphorylation sites (Tyr1158, Tyr1162, and Tyr1163). The association of the insulin and CB1 receptors is likely mediated by Gαi because knockdown of Gαi by siRNA diminished the association between these two receptors. Indeed, Gαi colocalized with the insulin receptor at the cell membrane and directly bound to the activation loop of IRβ, and more IRβ associated with active GTP-bound Gαi than with inactive GDP-bound Gαi. This observation is consistent with our previous finding that CB1 receptor agonism increased the amount of GTP-bound Gαi and its association with the insulin receptor (25). Together, our findings are indicative of physical and functional crosstalk between CB1 receptor and IRβ, with Gαi protein acting as an intermediary. Indeed, Gαi mediated the inhibitory effects of the CB1 receptor on autophosphorylation and kinase activity of the insulin receptor, as well as downstream signaling through direct binding to the activation loop of IRβ. The activation loop within the tyrosine kinase domain of the insulin receptor undergoes a major conformational change upon autophosphorylation of Tyr1158, Tyr1162, and Tyr1163, resulting in unrestricted access of adenosine triphosphate (ATP) and protein substrates to the kinase active site and stabilization of the conformation of the triple-phosphorylated activation loop (3). Therefore, these results further support our hypothesis that Gαi activated by the CB1 receptor directly associates with unphosphorylated insulin receptor at Tyr1158, Tyr1162, and Tyr1163, preventing a conformational change that secures the activation loop in a catalytically competent configuration upon ligand binding (25).

In mammals, the absolute number of β cells reflects a dynamic balance between β-cell growth and death. An inadequate expansion of β-cell mass to compensate for increased insulin demand, followed by the eventual loss of β cells due to apoptosis, is a hallmark of diabetes mellitus. This is most apparent in type 1 diabetes mellitus when ongoing autoimmunity causes destruction and consequent loss of β cells. Although β-cell mass is highly variable in human populations, declines in β-cell mass due to increased apoptosis have also been observed in patients with type 2 diabetes mellitus (50). A β-cell threshold seems to exist below which hyperglycemia will occur (51), and obese people with diabetes mellitus have reduced β-cell mass because of increased apoptotic rates (50, 52). Insulin acting through insulin receptors is a key growth and survival factor in most mammalian cells, including β cells (1620, 53, 54). From the data in this study, we propose that the endocannabinoid system that is intrinsic to islets (25) is a pathway by which β cells could influence their own survival and growth. The amounts of endocannabinoids are reported to be increased in both the circulation and the pancreas in diabetic and obese states (21, 22, 55, 56); additionally, increased endocannabinoid concentrations are associated with increased amounts of DAGLα (diacylglycerol lipase α, an enzyme involved in the synthesis of endocannabinoids) and decreased amounts of FAAH (fatty acid amide hydrolase, an enzyme involved in the degradation of endocannabinoids) in β cells (22). Thus, by impeding insulin receptor autophosphorylation in insulin-sensitive tissues in type 2 diabetes, increased endocannabinoid “tone” within islets (as a result of increased endocannabinoid synthesis, reduced degradation, or receptor abundance or activity) likely contributes to the lack of glucose responsiveness of β cells and the development of insulin resistance. Indeed, pharmacologic blockade of the CB1 receptor in obese fa/fa Zucker rats (57) and db/db mice (25) leads to decreased blood glucose concentrations and preserved β-cell mass. We showed in our previous study (25) that pharmacological blockade and genetic deficiency of the CB1 receptor in normal mice leads to decreased blood glucose concentrations and increased β-cell mass due to enhanced insulin receptor signaling. This is also true in STZ-treated mice. Thus, blockade of the CB1 receptor may also have effects in type 1 diabetes mellitus. These data suggest that in animal models of both type 1 and type 2 diabetes mellitus, endocannabinoid tone is increased in β cells and contributes to decreased β-cell mass. However, unlike in the pancreas of obese fa/fa Zucker rats and diet-induced obese mice (22, 55), it was previously observed that endocannabinoid or CB1 receptor abundance did not change in the pancreas of STZ-treated mice (22).

Direct follow-up of this work could include generation of a specific β-cell CB1 receptor–null mouse. The predicted phenotype would be a mouse whose β cells would be more resistant to apoptosis. Mice that globally lack the CB1 receptor are more resistant to β-cell apoptosis and have larger islets (25). Another potential follow-up would be development of a peripherally acting CB1 receptor antagonist with poor brain penetrance to lessen psychiatric side effects that are potentially life-threatening. It might be useful therapies in type 1 and type 2 diabetes mellitus to influence insulin receptor activity and β-cell function in the remaining β cells.

Materials and Methods

Materials and reagents

Sources and dilutions of primary antibodies used in Western blotting, immunoprecipitation, and immunostaining are listed in table S1. AEA, 2-AG, WIN55,212-2, ACEA, and AM251 were obtained from Cayman Chemical. Rimonabant and SLV-319 were obtained from J. F. McElroy (Jenrin Discovery). The caspase-3 and -7 inhibitor Ac-DNLD-CHO was from Calbiochem. Recombinant IRβ (amino acids 941 to 1343) fused to GST, recombinant IRS1 consisting of the p30 fragment (IRS1-p30, amino acids 516 to 777), and recombinant Gαi1 were obtained from Calbiochem. Insulin and STZ were from Sigma. The human insulin receptor and Gαi3 complementary DNA (cDNA) were amplified by reverse transcription–polymerase chain reaction (RT-PCR) from a human pancreas total RNA (Stratagene), with oligo(dT) (18 base pairs) for the reverse transcription. The insulin receptor cDNA was incorporated into an mCerulean-N1 vector between 5′-Hind III and 3′-Age I sites for IR-cerulean with the cerulean epitope at its C terminus. The cerulean epitope of IR-cerulean was then replaced with a 3× Flag epitope between the 5′-Age I and 3′-Bsr GI sites to make Flag-tagged IR. The Flag-tagged insulin receptor mutant (IR-3YA) was generated from wild-type insulin receptor (IR-WT) with a QuikChange II XL site-directed mutagenesis kit (Stratagene). Three tyrosine residues at positions 1158, 1162, and 1163 of the wild-type insulin receptor were replaced with alanines. IRβ deletion mutants were amplified from IR-cerulean and cloned into mVenus-C1 vector. The Gαi3 cDNA was incorporated into an mVenus-C1 vector between 5′-Xho I and 3′-Eco RI sites for Venus-Gαi3 with the Venus epitope at its N terminus.

Cell culture, transfection, and cell viability assays

Wild-type and insulin receptor–deficient β cells were established from control and β-cell–specific insulin receptor knockout mice, respectively (17, 25, 36, 37). All β-cell line and αTC1 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum (FBS) (Invitrogen). Transfections of siRNAs (Santa Cruz) for Gαi3 and Bad and the expression vectors for Gαi3 and insulin receptor were carried out with Lipofectamine RNAiMAX or 2000 (Invitrogen). Scramble siRNA (Silencer Negative Control #1; Ambion) or empty vector was transfected as negative control. For exogenous insulin treatment, cells starved overnight in DMEM containing 2 mM glucose and 0.1% FBS were pretreated with CB1 receptor agonists for 15 min before insulin stimulation for 10 min with or without CB1 receptor agonists. For cell viability and cytotoxicity studies, cells were plated into 96-well plates and incubated for 2 days with complete medium. The viability and cytotoxicity of the cells were determined 48 hours after treatment with CB1 receptor agonists or antagonists in DMEM containing 0.1% FBS by means of the MultiTox-Fluor Multiplex Cytotoxicity Assay (Promega) according to the manufacturer’s instructions. To test the knockdown effects of Bad or Gαi on the cell viability, we treated β cells from wild-type mice transfected with the indicated siRNAs for 24 hours with WIN55,212-2 in the medium containing 0.1% FBS.

RNA isolation and quantitative real-time PCR

We prepared total RNA from isolated islets or cell lines with Trizol reagent (Invitrogen) according to the manufacturer’s instructions. After RT, the resulting materials were used for quantitative real-time PCR (qRT-PCR) amplification with gene-specific primer pairs and SYBR Green PCR master mix (Applied Biosystems).

In vitro binding assay

Recombinant Gαi1 (2 μg) was incubated with guanosine 5′-O-(3′-thiotriphosphate) (GTP-γ-S) or GDP for 90 min before incubation with 2 μg of GST or GST-IRβ and glutathione–Sepharose 4B beads (Amersham Pharmacia) in binding buffer (50 mM tris-HCl at pH 7.4, 150 mM NaCl, 0.5% Triton X-100, and 1 mM EDTA) containing protease and phosphatase inhibitor cocktails, after which the beads were extensively washed in the same buffer and the adsorbed proteins were subjected to Western blot analysis with the primary antibodies and with a horseradish peroxidase–conjugated secondary antibody. Blots were visualized by enhanced chemiluminescence (ECL, GE Health).

Immunoprecipitation

Cell lysates extracted with radioimmunoprecipitation assay (RIPA) buffer (50 mM tris-HCl at pH 7.4, 150 mM NaCl, 1% NP-40, 0.1% SDS, and 1 mM EDTA) containing protease and phosphatase inhibitor cocktails were incubated with the appropriate antibody overnight at 4°C and subsequently incubated with protein A/G beads for 3 hours. Beads were washed three times with RIPA buffer and subjected to Western blot analysis. Normal immunoglobulin G (IgG) was used as negative control. For domain-mapping studies, lysates of β cells from knockout mice transfected with IRβ deletion mutants were subjected to immunoprecipitation with normal rabbit IgG and anti–CB1 receptor or Gαi3 antibodies and Western blot analysis with anti–green fluorescent protein (GFP) antibody.

In vitro kinase assay

For Fig. 2B, IRβ immune complexes from β cells from wild-type mice exposed to insulin with ACEA or vehicle were washed two times with RIPA buffer and once with tyrosine kinase buffer (50 mM Hepes at pH 7.4, 20 mM MgCl2, 0.1 mM MnCl2, and protease and phosphatase inhibitors) and then resuspended in tyrosine kinase buffer containing 25 μM ATP and IRS1-p30 (2 μg). The reaction mixtures were incubated for 30 min at 30°C, after which the reaction was terminated by the addition of SDS sample buffer. Samples were subjected to Western blot analysis. For Fig. 4D, IRβ immune complexes from insulin receptor–deficient β cells reconstituted with wild-type receptor or the tyrosine phosphorylation–deficient mutant were washed and incubated in tyrosine kinase buffer containing 25 μM ATP, IRS1-p30 (2 μg), and GTP-bound Gαi1 (2 μg) for 30 min at 30°C. Recombinant Gαi1 was incubated with GTP-γ-S for 90 min before addition.

Immunofluorescence

For detection of caspase-3 activation in αTC1 and β-cell lines, the cells were treated with WIN55,212-2, fixed, and incubated with anti–cleaved caspase-3 antibody, followed by secondary antibodies (Invitrogen) along with TO-PRO-3 (Invitrogen) for nuclear staining. For detection of endogenous IRβ and Gαi3 in wild-type β cells, cells were incubated with anti-IRβ and anti-Gαi3 antibodies, followed by secondary antibodies. Images were viewed with an LSM-710 confocal microscope (Carl Zeiss MicroImaging).

Animal experiments

All animal care and experimental procedures followed U.S. National Institutes of Health guidelines and were approved by the U.S. National Institute on Aging (NIA) Animal Care and Use Committee. CB1R−/− mice and their wild-type littermates were developed and backcrossed to a C57Bl/6J background, as previously described (39). For regeneration experiments, low-dose (50 mg/kg) STZ was administered by daily intraperitoneal injection into 2-month-old CD1 or CB1R−/− and CB1R+/+ mice (n = 5 per group) for 5 days. DMSO or AM251 (10 mg/kg) was then administered into CD1 mice by daily intraperitoneal injection without STZ. Three weeks after STZ withdrawal, pancreata and plasma were collected for the metabolic and morphological analyses. Blood glucose concentration was measured from tail-vein blood with a glucometer (Elite, Bayer Inc.), and plasma insulin was measured with rat/mouse insulin enzyme-linked immunosorbent assay (ELISA) kit (Crystal Chem). Pancreata were rapidly dissected, fixed in 4% paraformaldehyde (Sigma), immersed in 20% sucrose before freezing, and then sectioned at a thickness of 7 μm. After antigen unmasking, the slides were blocked with 5% bovine serum albumin (BSA)/phosphate-buffered saline (PBS) and incubated at 4°C with a specific primary antibody, followed by secondary antibodies along with TO-PRO-3, in some cases, for nuclear staining. Slides were viewed with an LSM-710 confocal microscope. Signal intensity and the number of nuclear p27- or PCNA-positive β cells were assessed with LSM Image Browser software (Carl Zeiss) or ImageJ software (http://rsb.info.nih.gov/ij/). For the analysis of β-cell and total pancreas area, digital images of multiple sections from three to five mice per group, separated by at least 200 μm from each section, at a magnification of ×10 were obtained, and the cross-sectional areas of pancreata and β cells (insulin-positive cells) were determined with LSM Image Browser software (Carl Zeiss). The relative cross-sectional area of β cells was determined by quantification of the cross-sectional area covered by insulin-positive cells divided by the cross-sectional area of total pancreas tissue. The number of cells that were positive for both insulin and nuclear p27 or PCNA was quantified as a percentage of the total number of insulin-positive cells in the sections.

Statistical analysis

Quantitative data are presented as means ± SEM. Differences between mean values were compared statistically by two-tailed Student’s t test or one-way analysis of variance (ANOVA) followed by Tukey’s post hoc comparison. Comparisons were performed with GraphPad Prism or SAS version 9.1. A P value of <0.05 was considered statistically significant.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/5/216/ra23/DC1

Fig. S1. CB1 receptor abundance in pancreatic β-cell lines.

Fig. S2. Effects of ACEA depend on the CB1 receptor.

Fig. S3. Effects of WIN55,212-2 on human hepatocarcinoma HepG2 cells and primary human hepatocytes.

Fig. S4. Effects of Gαi3 on insulin receptor signaling.

Fig. S5. Improved β-cell mass due to enhanced β-cell survival in STZ-treated CB1R−/− mice.

Fig. S6. Enhanced insulin signaling in β cells of STZ-treated CB1R−/− mice.

Fig. S7. Increased phosphorylation of p27 at Ser10 and Thr157 in islets of STZ-treated mice by CB1 receptor blockade.

Fig. S8. Increased abundance of PDX-1, GLUT2, and glucokinase in β cells of STZ-treated CB1R−/− mice.

Table S1. Details of the antibodies used for immunoblotting, immunoprecipitation, and immunofluorescence studies.

References and Notes

Acknowledgments: We are grateful to J. Pickel (National Institute of Mental Health Transgenic Core/NIH) for providing the CB1R−/− mice and to the animal facilities of NIA/NIH for carrying out the genotyping and husbandry. We also thank J. F. McElroy for providing reagents. Funding: This work was supported by the Intramural Research Program of the NIA/NIH. E.K.L. is supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (20110013116) and the Catholic Medical Center Research Foundation. R.N.K. is supported by NIH RO1 DK 67536 and 68721. Author contributions: W.K. designed and performed experiments and wrote/reviewed and edited the manuscript. Q.L. designed and performed some experiments and provided advice and reagents. Y.-K.S. and O.D.C. performed some experiments and analyzed the data. E.K.L. contributed to the design of some experiments, the interpretation of data, and the discussion. M.G. provided advice and reviewed the manuscript. R.N.K. provided advice and reagents and reviewed the manuscript. J.M.E. designed and performed experiments and wrote, reviewed, and edited the manuscript. Competing interests: Use of the wild-type (βIRWT) and insulin receptor–deficient (βIRKO) β-cell lines requires a material transfer agreement (MTA) from the Joslin Diabetes Center (R.N.K.). The other authors declare no conflicts of interest.
View Abstract

Related Content

Navigate This Article