p85α deficiency protects β-cells from endoplasmic reticulum stress-induced apoptosis Jonathon N. Winnaya, Ercument Diriceb, Chong Wee Liewc, Rohit N. Kulkarnib, and C. Ronald Kahna,1 a Section on Integrative Physiology and Metabolism and bSection on Islet Cell and Regenerative Biology, Joslin Diabetes Center and Harvard Medical School, Boston, MA 02215; and cDepartment of Physiology and Biophysics, University of Illinois at Chicago, Chicago, IL 60612
Contributed by C. Ronald Kahn, December 6, 2013 (sent for review November 4, 2013)
In insulin resistant states such as type 2 diabetes, there is a high demand on the β-cell to synthesize and secrete insulin, which challenges the ability of the endoplasmic reticulum (ER) to synthesize and fold nascent proteins. This creates a state of ER stress that triggers a coordinated program referred to as the unfolded protein response (UPR) that attempts to restore ER homeostasis. We identified a role for the p85α regulatory subunit of PI3K to modulate the UPR by promoting the nuclear localization of X-box binding protein 1, a transcription factor central to the UPR. In the present study we demonstrate that reducing p85α expression in β-cells can markedly delay the onset and severity of the diabetic phenotype observed in Akita+/− mice, which express a mutant insulin molecule. This is due to a decrease in activation of ER stress-dependent apoptotic pathways and a preservation of β-cell mass and function. These data demonstrate that modulation of p85α can protect pancreatic β-cells from ER stress, pointing to a potentially therapeutic target in diabetic states.
I
n type 2 diabetes, insulin resistance in peripheral tissues and associated hyperglycemia represent a major challenge to the capacity of the β-cell to augment insulin synthesis and secretion (1). In many individuals this response eventually fails due to β-cell dysfunction, an increase in β-cell apoptosis, and an associated reduction in β-cell mass (2). One major mechanism contributing to this decline in β-cell mass is the development of endoplasmic reticulum (ER) stress (3). The ER is an important organelle that performs critical functions, including lipid biosynthesis, maintenance of intracellular calcium homeostasis, and the folding of integral membrane and secreted proteins. This system can be stressed when there are increases in client protein load or perturbations in the ER microenvironment that limit the ability of the ER to effectively fold proteins (4). The development of ER stress activates three signal transduction pathways that mediate the unfolded protein response (UPR): pancreatic EIf2-α kinase, activating transcription factor (ATF)6α, and inositol requiring 1α (IRE1α) (5–8). IRE1α oligomerizes in response to ER stress, leading to activation of intrinsic endonuclease activity that splices the x-box binding protein 1 (XBP-1) mRNA, creating a second ORF that encodes a transcriptionally active isoform of XBP-1 that induces a transcriptional program that restores ER homeostasis (8). The β-cell is particularly vulnerable to the development of ER stress, because it has a high secretory capacity and is subjected to a variety of metabolic disturbances that alter the ER microenvironment, such as hyperglycemia, hyperlipidemia, and inflammatory cytokines (9). A number of studies have linked the development of ER stress to β-cell dysfunction in type 2 diabetes, an increase in β-cell apoptosis, and a resultant decline in β-cell mass (10, 11). In the normal physiological context, acute activation of the UPR leads to the up-regulation of fundamental processes that restore ER homeostasis. In contrast, pathophysiological states that chronically activate the UPR lead to the activation of pathways that initiate apoptosis. We and others have recently characterized a role for the p85α regulatory subunit of PI3K in ER stress and the UPR (12, 13). In this role, p85α binds to and facilitates the nuclear translocation of XBP-1, thus participating in the induction of the UPR. In in vitro and in vivo model systems, we found that reducing the level
1192–1197 | PNAS | January 21, 2014 | vol. 111 | no. 3
of p85α by genetic ablation could blunt the UPR after induction of ER stress. In the present study we have explored the role of p85α in the ER stress response in pancreatic β-cells by developing a mouse with the p85α gene specifically deleted in the β-cell and then crossing it with Akita+/− mice that carry a mutant insulin molecule that prevents normal folding and secretion and induces ER stress (14). Whereas Akita+/− mice exhibited markedly elevated glucose levels and a concomitant decline in serum insulin in the fed and fasted states, reducing the level of p85α in β-cells in Akita mice resulted in a significant reduction in ER stress, with a restoration of islet mass due to suppression of β-cell apoptosis. Conversely, increasing XBP-1 expression in β-cells in vitro leads to an increase in the rate of β-cell apoptosis, and this effect is further augmented by overexpression of p85α. Thus, reducing p85α levels in β-cells can delay the decline in β-cell mass and function associated with chronic activation of the UPR, leading to a marked delay in the onset of hyperglycemia and hypoinsulinemia in this animal model. Results p85α Expression Is Reduced in Islets from β-KO and Akita/β-KO Animals. Previous studies have shown that p85α facilitates in-
duction of the UPR in response to chemical ER stressors (13). To examine the role of p85α in the development of ER stress in pancreatic β-cells in vivo, we crossed Akita+/−mice carrying a mutant insulin allele, which leads to abnormal proinsulin processing and folding, to mice with a β-cell–specific knockout of p85α (β-KO) to create Akita+/−/β-KO mice. Targeted deletion of p85α was confirmed by quantitative PCR on mRNA derived from isolated islets, which revealed a 63–67% reduction in p85α expression in β-KO and Akita/β-KO islets compared with control and Akita+/− mice (Fig. 1A). This decrease in p85α most likely Significance In type 2 diabetes, peripheral insulin resistance challenges the pancreatic β-cell to augment insulin synthesis and secretion. This compensatory mechanism eventually fails due to β-cell dysfunction, an increase in β-cell apoptosis, and a resultant decrease in β-cell mass. One mechanism contributing to this decline in β-cell mass is the development of endoplasmic reticulum (ER) stress. We have elucidated a role for the p85α regulatory subunit of PI3K as a modulator of ER stressdependent apoptosis in β-cells by virtue of its ability to positively regulate the proapoptotic actions of x-box binding protein 1. The ablation of p85α specifically in the β-cell in a mouse model of ER stress-dependent β-cell apoptosis delays the decline in β-cell mass and subsequent development of diabetes. Author contributions: J.N.W., E.D., C.W.L., R.N.K., and C.R.K. designed research; J.N.W., E.D., and C.W.L. performed research; J.N.W., R.N.K., and C.R.K. analyzed data; and J.N.W., R.N.K., and C.R.K. wrote the paper. The authors declare no conflict of interest. 1
To whom correspondence should be addressed. E-mail:
[email protected]. edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1322564111/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1322564111
B
C
D
E
F
Fig. 1. Assessment of glucose homeostasis at 6 wk of age. mRNA expression of (A) p85α or (B) p85β were measured from islets isolated from study group mice at 20 wk of age. (C) Fasting glucose was measured after a 16- to 18-h fast (n = 6–12). (D) Fasting insulin was measured from serum after a 16- to 18-h fast (n = 5–8). (E) Random fed glucose was measured between 8:00 and 10:00 AM (n = 5–7). (F) Random fed serum insulin levels. n = 5–6. *P < 0.05, **P < 0.01, ***P < 0.001.
reflects a nearly complete loss of p85α mRNA in β-cells that normally comprise ∼80% of islet cells. There was no difference in p85β mRNA among study groups (Fig. 1B). β-Cell–Specific Deficiency of p85α Improves Glucose Homeostasis in the Presence of the Akita Mutation. As previously reported (14),
by 6 wk of age, mice carrying the Akita allele displayed marked fasting hyperglycemia compared with controls (295 ± 48 vs. 112 ± 11 mg/dL, P < 0.001), whereas fasting glucose levels in β-KO were not statistically different from those in controls (Fig. 1C). Superposition of p85α β-cell deficiency on the mutant Akita allele prevented the development of marked hyperglycemia and restored glucose levels to normal (113 ± 10 mg/dL). This mirrored the changes observed in insulin levels. Thus, Akita+/− mice showed an 84% decrease (P < 0.05) in serum insulin compared with controls, whereas the Akita/β-KO animals exhibited normal insulin levels (Fig. 1D). In the fed state, Akita+/− mice displayed even more marked hyperglycemia (583 ± 47 mg/dL) compared with controls (131 ± 4 mg/dL). Superimposition of the KO of p85α on the Akita mutation also significantly reduced fed glucose levels compared with Akita+/− mice but did not completely prevent hyperglycemia (Fig. 1E). These mirrored the changes in insulin levels which were reduced in Akita+/− mice compared with controls (0.9 ± 0.1 vs. 2.4 ± 0.5 pg/dL; P < 0.05), whereas the Akita/β-KO mice had intermediate levels (1.6 ± 0.06 pg/dL) (Fig. 1F). Similar results on fed and fasted insulin and glucose levels were observed at 8 wk of age, when the hypoinsulinemia and hyperglycemia in the Akita +/− mice was even more marked (Fig. S1 A–D). Winnay et al.
To assess whether the improvements in fed and fasted glucose and insulin levels in Akita/β-KO were associated with an improvement in peripheral glucose disposal, i.p. glucose tolerance tests were performed. As anticipated, Akita+/− mice displayed fasting hyperglycemia with a glucose excursion curve that was significantly higher at all time-points compared with controls and β-KO mice (Fig. S2 A and B). Akita/β-KO mice exhibited an intermediate response, with a statistically significant improvement in glycemia at the 0- and 15-min time points compared with Akita+/− mice (P < 0.05). A similar trend toward improvement in glucose tolerance in Akita/β-KO mice was also observed at the 8-wk time point compared with Akita +/− mice (Fig. S1 E and F). An in vivo assessment of glucose-stimulated insulin release at this time point revealed a severe defect in first phase insulin release in Akita+/− mice, whereas β-KO animals exhibited a partial defect compared with controls, which has been previously described (Fig. S1G). Akita/β-KO animals exhibited a similar, partial impairment in insulin secretion, indicating that the preservation of β-cell function in Akita/β-KO mice likely contributes to the relative improvement in glucose tolerance compared with Akita+/− mice. As Akita+/− mice age, the phenotype becomes progressively more severe. By 20 wk of age, the average fed glucose levels were >600 mg/dL (the limit of the glucose meter) and circulating insulin levels less than 0.52 ng/mL (Fig. S3 A and B). Again, superimposition of p85α deficiency resulted in a significant improvement in glycemia and a trend toward improvement in insulin levels in Akita/β-KO mice compared with Akita+/− animals. Moreover, when C-peptide levels were evaluated as a surrogate for insulin secretion, a clear reduction was observed in Akita+/− mice, which was partially rescued in Akita/β-KO mice (Fig. S3C). Collectively, these data show that the insulin deficiency and hyperglycemia observed in Akita+/− mice is significantly improved by p85α deficiency in β-cells, leading to a delay in the progression and reduction in the severity of the phenotype. Thus, β-cell– specific ablation of p85α preserves β-cell function and/or mass in the presence of the chronic ER stress imposed by the Akita mutation and significantly delays the decline in β-cell function and the development of hyperglycemia. β-cell Mass Is Protected and ER Stress Is Prevented in Akita/β-KO Mice. Pancreas weights were similar at 9 and 20 wk of age be-
tween the different study groups (Fig. S4 A and B); however, consistent with previous reports, hematoxylin and eosin staining or immunohistochemistry (IHC) performed with anti-insulin antibodies revealed that islets of Akita+/− mice at 9 and 20 wk of age were smaller than islets of control, β-KO, and Akita/β-KO mice (Fig. 2A and Fig. S5) (15). A quantitative assessment of average islet size confirmed the decrease observed in Akita+/− islets: average islet size was reduced compared with all other study groups (Fig. 2B). In contrast, average islet size and morphological appearance in Akita/β-KO animals were similar to those in controls. Quantification of β-cell area revealed an ∼70% reduction in Akita+/− mice compared with controls (0.12% ± 0.01% vs. 0.38% ± 0.01%, P < 0.001) (Fig. 2C) (15). In contrast, β-cell area in Akita/β-KO islets was statistically indistinguishable from controls. No difference in α-cell mass was observed between controls and Akita+/− or β-KO islets; however, there was a statistically significant increase in α-cell area in Akita/β-KO islets compared with controls (Fig. S6). As expected, electron microscopy of β-cells in the Akita+/− islets revealed a dramatic reduction in insulin secretory granules and the presence of markedly distended ER, a sign of ER stress, compared with controls (Fig. 3) (15). By contrast, the Akita/ β-KO β-cells, as well as the β-KO β-cells, showed normal ER structure and contained numerous insulin secretory granules. Likewise, there was an 83% decrease in insulin content in Akita+/− pancreata compared with control, which was reversed in the Akita/β-KO mice (Fig. S7A). Assessment of proinsulin/insulin ratios revealed a trend toward an increase in Akita+/− mice that was not observed in Akita/β-KO animals, a result suggestive of an PNAS | January 21, 2014 | vol. 111 | no. 3 | 1193
PHYSIOLOGY
A
A
B
apoptotic rates in β-KO islets compared with controls (P < 0.01) (16). Importantly, although Akita/β-KO islets had increased rates of apoptosis compared with controls, there was a marked reduction of apoptosis in the Akita/β-KO islets compared with Akita+/− islets (0.20 ± 0.01 vs. 0.66 ± 0.09, P < 0.05) (Fig. 4A). To determine whether p85α-deficient β-cells exhibited a similar resistance to pharmacologically induced, ER stress-dependent apoptosis, islets were isolated from control and β-KO animals, treated with tunicamycin in vitro, and apoptosis assessed by TUNEL assay. In control islets, treatment with tunicamycin led to an ∼58-fold increase in TUNEL-positive nuclei compared with vehicle-treated controls (Fig. 4B). In contrast, tunicamycin induced only a ∼6.2-fold increase in apoptosis in β-KO islets, indicating that p85α deficiency leads to a marked reduction in ER stress-dependent apoptosis (Fig. 4B; P < 0.001). Proliferation rates assessed by Ki67 staining in control, Akita+/−, and Akita/β-KO mice were statistically indistinguishable, and there was a modest increase in proliferation in β-KO islets compared with controls, as previously described (Fig. 4 C and D) (16). Thus, the preservation of β-cells in p85α-deficient islets in the face of heightened ER stress is mediated by the suppression of apoptosis rather than an increase in β-cell proliferation. Reduction in ER Stress, Oxidative Stress, and Proapoptotic Signals in Isolated Akita/β-KO Islets. To assess the relative levels of ER and
C
Fig. 2. Morphometric analysis of islets at 9 wk of age. (A) Hematoxylin and eosin staining and IHC using anti-insulin and anti-glucagon antibodies was performed on pancreas sections. Representative images are shown. (B) Islet area was determined using ImageJ, and average islet size is plotted. n = 4. (C) Total β-cell area as a percentage of total pancreas area. n = 3. *P < 0.05, **P < 0.01, ***P < 0.001.
improvement in ER function and/or prohormone processing in Akita/β-KO islets (Fig. S7B). Thus, Akita+/− mice develop ultrastructural changes in β-cells indicative of ER stress, and this is associated with a loss of β-cell mass and a reduction in insulin content. All of these changes are ameliorated by p85α-deficiency. Rates of ER Stress-Dependent Apoptosis Are Decreased in p85αDeficient β-Cells. To assess whether the preservation of β-cell
area observed in Akita/β-KO animals was due to alterations in apoptosis or proliferation, the TUNEL assay or IHC using antiKi67 antibodies was performed. As expected, TUNEL staining revealed an 18.7-fold increase in the occurrence of apoptotic cells in Akita+/− islets compared with controls (0.66 ± 0.09 vs. 0.03 ± 0.01, P < 0.01) (Fig. 4A). Consistent with previous findings, there was also a statistically significant, 2.7-fold increase in
oxidative stress in study group islets, IHC was performed using anti-pIRE1α and anti-nitrotyrosine antibodies, respectively. This revealed significant increases in IRE1α phosphorylation and nitrotyrosine immunoreactivity in Akita+/− islets but not in Akita/β-KO islets, demonstrating that ER and oxidative stress is significantly attenuated in Akita/β-KO islets (Fig. S8). Moreover, gene expression measured in isolated islets revealed a significant increase in the ER stress-induced transcript for CHOP in Akita+/− islets, whereas no change was observed in the other study groups (Fig. 5A; P < 0.05). Oxidative stress-responsive genes including catalase, Nrf1, Nrf2, SOD2, and HMOX1 were significantly elevated in Akita+/− islets (Fig. 5B; P < 0.05). Although modest increases in catalase and SOD2 were observed in Akita/β-KO islets, the overall magnitude was significantly lower than that observed in Akita+/− islets. Collectively, these data indicate that p85α deficiency reduces the ER and oxidative stress imposed by the Akita allele. UPR-dependent activation of the NLRP3 inflammasome pathway promotes apoptosis under conditions of chronic ER stress (17, 18). The finding of preservation of β-cell area, a decrease in the rates of β-cell apoptosis, and a reduction in ER stress in Akita/ β-KO islets suggests that activation of the NLRP3-dependent apoptotic pathway may be reduced in p85α-deficient β-cells. In Akita+/− islets, there was a significant, 5.6-fold elevation in thioredoxin-interacting protein (TXNIP) gene expression, and this was returned to near normal in islets from Akita/β-KO animals (Fig. 5C). This was paralleled by changes in ATF5, a transcription factor responsible for the UPR-dependent transactivation of the TXNIP gene, which was up-regulated by 1.8-fold in Akita+/− islets and returned to normal in the Akita+/−/β-KO islets (Fig. 5C; P < 0.05). Likewise, IL-1β mRNA levels that are regulated by ER stress were increased twofold in Akita+/− islets and restored to normal in the Akita/β-KO islets (Fig. 5C) (17, 18). These data suggest that the NLRP3 inflammasome pathway is activated in
Fig. 3. Transmission electron microscopy (TEM) reveals a reduction in ER stress-associated abnormalities and restoration of insulin content in Akita/ β-KO islets. TEM was performed on islets isolated from study group animals at 20 wk of age. Representative images are shown.
1194 | www.pnas.org/cgi/doi/10.1073/pnas.1322564111
Winnay et al.
A
B
XBP-1 was over-expressed by adenoviral-mediated gene transduction, there was a dose-dependent increase of apoptosis as measured by the cleavage of caspase-3, and this was further enhanced after treatment with tunicamycin or thapsigargin (Fig. 6B). When p85α was overexpressed together with XBP-1, there was a further enhancement in caspase-3 cleavage with increasing p85α levels, indicating that p85α can potentiate XBP-1–dependent apoptosis (Fig. 6C and Fig. S9) (13). Thus, p85α is capable of modulating the level of XBP-1 and its proapoptotic action. Discussion We have demonstrated that in mice carrying a single copy of a mutant insulin allele, there is activation of the ER stress response pathways, leading to increased rates of β-cell apoptosis, a loss of β-cell mass, and development of marked hyperglycemia. Reduction of p85α in β-cells ameliorates ER stress, reduces the rate of β-cell apoptosis, preserves β-cell mass, and significantly
A
C B
D
Fig. 4. Measurement of proliferation and apoptosis at 9 wk of age. (A) Rates of apoptosis were determined by performing the TUNEL assay. n = 3 per group. (B) Isolated islets from control and β-KO animals were treated with vehicle or tunicamycin for 24 h before performing the TUNEL assay. (C) Proliferation of β-cells was determined by performing IHC using anti-insulin and anti-Ki67 antibodies. n = 3. *P < 0.05, **P < 0.01. (D) Quantification of proliferation rates was performed and is expressed as the percentage of insulin+/Ki67+ cells.
C
p85α Stabilizes XBP-1 and Promotes XBP-1-Dependent Apoptosis. We have previously shown that overexpression of p85α results in stabilization of XBP-1 protein levels (13). To determine whether a similar effect is observed in the β-cell, p85α was overexpressed in the MIN6 β-cell line in the presence of exogenously expressed XBP-1. A dose-dependent increase in p85α expression led to a concomitant increase in XBP-1 protein levels (Fig. 6A). When Winnay et al.
Fig. 5. Islet gene expression reveals a profile indicative of reduced ER and oxidative stress and activation of the NLRP3 inflammasome pathway in Akita/β-KO islets. Quantitative RT-PCR was performed with gene-specific primers to determine the relative mRNA expression of (A) ER stress, (B) oxidative stress, or (C) NLRP3 inflammasome-associated genes in isolated islets from mice at 20 wk of age. n = 3–5. Gene expression was normalized to TBP mRNA expression. *P < 0.05, **P < 0.01, ***P < 0.001.
PNAS | January 21, 2014 | vol. 111 | no. 3 | 1195
PHYSIOLOGY
islets of Akita+/− mice, and this is completely abrogated by p85α deficiency.
A
B
C
Fig. 6. p85α stabilizes XBP-1 and regulates XBP-1–dependent apoptosis. (A) The MIN6 β-cell line was transduced with the indicated adenoviral constructs. Forty-eight hours later, whole-cell protein lysates were prepared, resolved by SDS/PAGE, and immunoblotted with the indicated antibodies. (B) The MIN6 β-cell line was transduced with the indicated adenoviral constructs. Twenty-four hours later, transduced cells were treated with vehicle, tunicamycin (5 μg/dL), or thapsigargin (200 nM) for an additional 24 h. Subsequently, whole-cell protein lysates were prepared, resolved by SDS/PAGE, and immunoblotted with the indicated antibodies. (C) The MIN6 β-cell line was transduced with the indicated adenoviral constructs. Forty-eight hours later, protein lysates were prepared, resolved by SDS/PAGE, and immunoblotted with the indicated antibodies.
delays the development of hyperglycemia and hypoinsulinemia. Mechanistically, p85α deficiency reduces the development of ER and oxidative stress that induces the CHOP and NLRP3 inflammasome-mediated, apoptotic pathways. These data reveal a role for p85α as an important mediator of the UPR-dependent apoptotic program in the setting of chronic ER stress. Although acute activation of the UPR under physiological conditions serves to restore ER homeostasis, irremediable ER stress leads to chronic activation of the UPR, resulting in the execution of proapoptotic programs. In humans, mutations in the insulin gene that lead to protein misfolding have also been shown to induce ER stress, and it is believed that this is associated with a decrease in β-cell mass, which can lead to the development of neonatal diabetes (19). Although the precise mechanism(s) governing this cell fate decision are poorly defined, several of the mediators of this apoptotic response have been identified, including CHOP, Tribbles-related protein 3, death receptor-5, c-jun N-terminal kinase (JNK), and TXNIP (14, 17, 18, 20). A variety of studies using CHOP-deficient mice have established a role for this transcription factor as a mediator of ER stress-dependent apoptosis in a number of disease models (21, 22). Consistent with our findings, when the Akita allele is introduced onto a CHOP-deficient mouse, the resultant animals exhibit a reduction in β-cell apoptosis and a delay in the progression of the Akita phenotype (21). In the Akita/β-cell p85α KO mice, we find that the elevation in ER stress normally observed in Akita+/− islets is markedly reduced, indicating that the induction of ER stress-dependent apoptotic pathways is largely prevented by p85α deficiency. Although reducing the level of p85α or CHOP delays the decline in β-cell function and mass, neither completely prevents the development of diabetes, owing 1196 | www.pnas.org/cgi/doi/10.1073/pnas.1322564111
to the robust activation of proapoptotic pathways and in part because there may be additional apoptotic pathways. In the present study we have provided evidence that the NLRP3 inflammasome-mediated apoptotic pathway is also activated in Akita+/− islets, and this is prevented or delayed when p85α is absent. Recent studies have identified TXNIP as an important node linking ER stress and cytokine-mediated apoptosis (17, 18). TXNIP is induced by ER stress and regulates the induction of IL-1β transcription and subsequent activation by the NLRP3 inflammasome, thereby initiating IL-1β–mediated apoptosis. TXNIP induction has been previously observed in islets from Akita+/− mice, and superimposing the Akita allele on a TXNIP-null background can also lead to a substantial decrease in β-cell apoptosis. We find that the increase in TXNIP expression in Akita+/− islets is paralleled by an increase in IL-1β mRNA levels and that both responses are blocked in Akita/β-KO islets, indicating that p85α also plays a role in the ER stress-dependent increase in NLRP3 inflammasome-mediated apoptosis. In addition to up-regulation of components of the NLRP3 inflammasome pathway, a gene expression signature indicative of oxidative stress is observed in Akita+/− but not Akita/β-KO islets. A critical role for reactive oxygen species (ROS) production in activation of the NLRP3 inflammasome is evident by the ability of ROS scavengers to prevent inflammasome activation in response to various NLRP3 agonists. Moreover, β-cell lines derived from Akita+/− mice have been shown to exhibit severe oxidative stress, as indicated by increased ROS, TXNIP induction, and protein tyrosine nitration, suggesting that oxidative stress may promote the activation of proapoptotic pathways that promote β-cell loss in Akita+/− islets. Although activation of the IRE1/XBP-1 pathway is critical for the acute cellular response to ER stress, constitutive activation of this arm of the UPR can induce apoptosis in β-cells. For example, hyperactivation of IRE1α leads to the recruitment of TNF receptor-associated factor 2, activation of JNK, and subsequent initiation of apoptosis. In the present study, we observed a significant increase in activation of IRE1α in Akita+/− but not in Akita/β-KO islets, suggesting that hyperactivation of this pathway may contribute to the observed increase in apoptotic rates in islets of Akita+/− mice. Moreover, because IRE1α has been shown to mediate the induction of TXNIP in response to ER stress, the observed hyperactivation of this pathway would also be expected to increase NLRP3 inflammasome-mediated apoptosis. Additional evidence suggesting that chronic activation of the IRE1/XBP-1 pathway may negatively influence β-cell fate comes from a study in which overexpression of XBP-1 in intact islets or dispersed β-cells led to increased rates of β-cell apoptosis (23). In the present study we also observed an increase in apoptosis in the MIN6 β-cell line upon XBP-1 overexpression. A robust increase in apoptosis after the pharmacological induction of ER stress also positively correlated with the extent of XBP-1 expression. Consistent with our hypothesis, XBP-1–dependent apoptosis was further increased in cells overexpressing both p85α and XBP-1, indicating that p85α potentiates or participates in XBP-1– dependent activation of apoptotic programs. Previous data establishing a role for p85α as a regulator of XBP-1 protein stability and nuclear translocation would predict that XBP-1 protein levels and nuclear translocation would be reduced in p85α-deficient β-cells (12, 13). Therefore, in the setting of p85α deficiency, potentially deleterious signaling through the IRE1/XBP-1 arm of the UPR that promotes apoptosis would be prevented, leading to enhanced survival. Our data demonstrate that p85α not only functions as a mediator of acute responses to ER stress by virtue of its ability to regulate the function of XBP-1, but also regulates cell fate decisions in the setting of chronic, irremediable ER stress (12, 13). The apparent lack of significant ER stress in p85α-deficient β-cells in β-KO mice suggests that physiological activation of the UPR pathways is sufficient to maintain normal ER homeostasis. In both in vitro and in vivo systems we have demonstrated that the nuclear translocation of XBP-1 is reduced, but not entirely Winnay et al.
Materials and Methods
Morphological and Morphometric Analysis. Mice were anesthetized, pancreata rapidly dissected, weighed, fixed in 4% (vol/vol) paraformaldehyde solution, embedded in paraffin, sectioned, and stained as described below. β-Cell area was estimated by morphometric analysis as described previously and is expressed as insulin- or glucagon-positive area divided by total pancreas area (16). Detailed procedures for determination of apoptotic and proliferation rates are described in SI Materials and Methods. Electron Microscopy. Mouse islets were isolated from anesthetized animals using the intraductal collagenase method. Isolated islets were hand-picked and immediately placed into a fixative solution of 2.5% (vol/vol) glutaraldehyde in a 0.1-M phosphate buffer overnight at 4°, then washed and fixed in 2% (vol/vol) 0sO4. Sections (400–500 A°) were stained with saturated uranyl acetate and lead citrate before being photographed on a Philips 301 transmission electron microscope. Quantitative Measurement of Gene Expression. RNA was purified from isolated islets using an RNeasy kit according to the manufacturer’s directions (Qiagen). Reverse transcription was carried out using the High Capacity RNA-tocDNA kit according to the manufacturer’s directions (Applied Biosystems). cDNA was used as template to perform quantitative PCR using gene-specific primers and IQ SYBR Green Supermix (Bio-Rad). Quantitative PCR was carried out using a CFX384 Real-Time PCR Detection System (Bio-Rad). Results were obtained using the ΔΔCT method using Tata-binding protein (TBP) gene expression as an internal standard.
Creation and Genotyping of Mutant Mice. The floxed Pik3r1, Rip-Cre, and Akita mice have been previously described (15, 16). Mice were maintained on a 129Sv-c57BL/6 hybrid background. Animals were housed on a 12-h light/12-h dark cycle with ad libitum access to water and food (mouse diet F9; PMI Nutrition International). Genotyping for the Pik3r1 allele, RIP-CRE transgene, and Akita allele was performed using previously published methods (15, 16). All experiments were approved by and conducted in accordance with guidelines established by the Institutional Animal Care and Use Committee at the Joslin Diabetes Center.
Cell Culture and Immunoblot Analysis. MIN6 β-cells were used between passages 26 and 32 and grown in DMEM containing 20% (vol/vol) heat-inactivated FCS, 25 mM glucose, 100 mM β-mercaptoethanol, 100 IU/mL penicillin, and 100 μg/mL streptomycin in a humidified atmosphere at 37 °C with 5% CO2. Detailed procedures for protein isolation and immunoblot analysis are described in SI Materials and Methods.
Analytic Procedures. Fasted insulin and glucose levels were measured after a 16-h fast, and specimens collected for the determination of random fed values were obtained between 8:00 and 10:00 AM. Serum insulin levels were measured by the Specialized Assay Core at the Joslin Diabetes Center. Glucose levels were determined from whole venous blood using a glucometer (Infinity, US Diagnostics Inc.). Pancreatic insulin content was measured in acidethanol extracts of homogenized pancreas.
ACKNOWLEDGMENTS. We thank Gordon C. Weir for helpful discussions, and Christopher Cahill and the Advanced Microscopy Core at the Joslin Diabetes Center for help in performing electron microscopy studies. The Specialized Assay Core at the Joslin Diabetes Center performed determination of hormone levels. C.W.L. is supported by National Institutes of Health (NIH) Pathway to Independence Award R00 DK090210. This research was supported by NIH National Institute of Diabetes and Digestive and Kidney Grants DK67536 (to R.N.K.) and DK055545 (to C.R.K.).
1. Sachdeva MM, Stoffers DA (2009) Minireview: Meeting the demand for insulin: Molecular mechanisms of adaptive postnatal beta-cell mass expansion. Mol Endocrinol 23(6):747–758. 2. Butler AE, et al. (2003) Beta-cell deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 52(1):102–110. 3. Harding HP, Ron D (2002) Endoplasmic reticulum stress and the development of diabetes: A review. Diabetes 51(Suppl 3):S455–S461. 4. Görlach A, Klappa P, Kietzmann T (2006) The endoplasmic reticulum: Folding, calcium homeostasis, signaling, and redox control. Antioxid Redox Signal 8(9-10):1391–1418. 5. Harding HP, Zhang Y, Ron D (1999) Protein translation and folding are coupled by an endoplasmic-reticulum-resident kinase. Nature 397(6716):271–274. 6. Sidrauski C, Walter P (1997) The transmembrane kinase Ire1p is a site-specific endonuclease that initiates mRNA splicing in the unfolded protein response. Cell 90(6): 1031–1039. 7. Haze K, Yoshida H, Yanagi H, Yura T, Mori K (1999) Mammalian transcription factor ATF6 is synthesized as a transmembrane protein and activated by proteolysis in response to endoplasmic reticulum stress. Mol Biol Cell 10(11):3787–3799. 8. Lee K, et al. (2002) IRE1-mediated unconventional mRNA splicing and S2P-mediated ATF6 cleavage merge to regulate XBP1 in signaling the unfolded protein response. Genes Dev 16(4):452–466. 9. Allagnat F, et al. (2011) Mcl-1 downregulation by pro-inflammatory cytokines and palmitate is an early event contributing to β-cell apoptosis. Cell Death Differ 18(2): 328–337. 10. Laybutt DR, et al. (2007) Endoplasmic reticulum stress contributes to beta cell apoptosis in type 2 diabetes. Diabetologia 50(4):752–763. 11. Delépine M, et al. (2000) EIF2AK3, encoding translation initiation factor 2-alpha kinase 3, is mutated in patients with Wolcott-Rallison syndrome. Nat Genet 25(4): 406–409. 12. Park SW, et al. (2010) The regulatory subunits of PI3K, p85alpha and p85beta, interact with XBP-1 and increase its nuclear translocation. Nat Med 16(4):429–437.
13. Winnay JN, Boucher J, Mori MA, Ueki K, Kahn CR (2010) A regulatory subunit of phosphoinositide 3-kinase increases the nuclear accumulation of X-box-binding protein-1 to modulate the unfolded protein response. Nat Med 16(4):438–445. 14. Zinszner H, et al. (1998) CHOP is implicated in programmed cell death in response to impaired function of the endoplasmic reticulum. Genes Dev 12(7):982–995. 15. Yoshioka M, Kayo T, Ikeda T, Koizumi A (1997) A novel locus, Mody4, distal to D7Mit189 on chromosome 7 determines early-onset NIDDM in nonobese C57BL/6 (Akita) mutant mice. Diabetes 46(5):887–894. 16. Kaneko K, et al. (2010) Class IA phosphatidylinositol 3-kinase in pancreatic β cells controls insulin secretion by multiple mechanisms. Cell Metab 12(6):619–632. 17. Lerner AG, et al. (2012) IRE1α induces thioredoxin-interacting protein to activate the NLRP3 inflammasome and promote programmed cell death under irremediable ER stress. Cell Metab 16(2):250–264. 18. Oslowski CM, et al. (2012) Thioredoxin-interacting protein mediates ER stress-induced β cell death through initiation of the inflammasome. Cell Metab 16(2):265–273. 19. Colombo C, et al.; Early Onset Diabetes Study Group of the Italian Society of Pediatric Endocrinology and Diabetes (SIEDP) (2008) Seven mutations in the human insulin gene linked to permanent neonatal/infancy-onset diabetes mellitus. J Clin Invest 118(6):2148–2156. 20. Yamaguchi H, Wang HG (2004) CHOP is involved in endoplasmic reticulum stress-induced apoptosis by enhancing DR5 expression in human carcinoma cells. J Biol Chem 279(44):45495–45502. 21. Song B, Scheuner D, Ron D, Pennathur S, Kaufman RJ (2008) Chop deletion reduces oxidative stress, improves beta cell function, and promotes cell survival in multiple mouse models of diabetes. J Clin Invest 118(10):3378–3389. 22. Thorp E, et al. (2009) Reduced apoptosis and plaque necrosis in advanced atherosclerotic lesions of Apoe-/- and Ldlr-/- mice lacking CHOP. Cell Metab 9(5):474–481. 23. Allagnat F, et al. (2010) Sustained production of spliced X-box binding protein 1 (XBP1) induces pancreatic beta cell dysfunction and apoptosis. Diabetologia 53(6): 1120–1130.
Winnay et al.
PNAS | January 21, 2014 | vol. 111 | no. 3 | 1197
PHYSIOLOGY
prevented in the absence of p85α (13). The residual translocation in the absence of p85α may be mediated by p85β (12). Therefore, despite a partial impairment in the nuclear translocation of XBP-1 in response to physiological activation of the UPR in p85αdeficient β-cells, this residual response is probably sufficient to acutely maintain ER homeostasis. However, in the face of sustained ER stress, p85α deficiency results in a significant impairment of the IRE1/XBP-1 pathway that reduces or prevents the activation of proapoptotic programs. Collectively, our data support an important role for p85α as a mediator of ER stress-dependent apoptosis in the β-cell. Further experiments are needed to determine whether these effects are entirely dependent on the reduction of ER and oxidative stress observed in Akita/β-KO islets or whether p85α directly modulates components of these pathways. On the basis of our findings, p85α represents a target for the development of therapies to prevent the progressive pathogenesis of conditions in which ER stress-dependent apoptosis has been identified as a mechanism contributing to the development of diseases including inflammatory diseases, neurodegenerative disorders, cancer, and diabetes.
Supporting Information Winnay et al. 10.1073/pnas.1322564111 SI Materials and Methods Morphological and Morphometric Analysis. Five-micrometer sec-
tions of paraffin-embedded pancreas were dewaxed using xylene, rehydrated through serial dilutions of ethyl alcohol, and subjected to antigen retrieval using 10 mM citrate (pH 6.1). Sections were stained with the indicated antibodies. For determination of apoptosis rates, islets were cultured for 24 h in RPMI 1640 supplemented with 10% (vol/vol) FBS and penicillin/streptomycin before treatment with tunicamycin (2.5 μg/ mL). Islets were fixed in 4% (vol/vol) buffered formalin and embedded in 1% agarose before embedding in paraffin. The TUNEL assay was performed according to the manufacturer’s protocol (DeadEnd Fluorometric TUNEL kit; Promega). Primary antibodies included guinea pig anti-insulin (DAKO), rabbit anti-glucagon (DAKO), rabbit anti-Ki67 (Abcam), rabbit
Winnay et al. www.pnas.org/cgi/content/short/1322564111
anti-nitrotyrosine (EMD Millipore), rabbit anti-nitrotyrosine (EMD Millipore), and rabbit anti-pIRE1α (Novus Biologicals). Fluorescently conjugated secondary antibodies were from Invitrogen. Immunoblot Analysis. MIN6 cells were lysed in RIPA buffer [50 mM Tris·HCl (pH 7.4), 150 mM NaCl, 0.25% deoxycholic acid, 1% Nonidet P-40, and 1 mM EDTA] and rotated for 10 min at 4° before centrifugation at 23,000 × g for 10 min. Soluble protein was isolated and protein concentrations determined by performing the BCA assay (Thermo Scientific). Normalized protein was resolved by SDS/PAGE, transferred to PVDF, and probed with anti-cleaved caspase-3 (Cell Signaling Technology), anti-FLAG (Sigma Aldrich), anti-HA, or anti-actin (Santa Cruz Biotechnology) antibodies.
1 of 5
Fig. S1. Assessment of glucose homeostasis at 8 wk of age. (A) Fasting glucose was measured from whole venous blood after a 16- to 18-h fast using a glucometer. n = 8–24. (B) Fasting insulin was measured from serum after a 16- to 18-h fast. n = 11–22. (C) Random fed glucose was measured from whole venous blood obtained between 8:00 and 10:00 AM. n = 10–17. (D) Random fed insulin was measured from serum. n = 10–13. (E and F) Glucose tolerance tests were performed on study group animals. n = 7–10. (G) Glucose-stimulated insulin release was assessed at the indicated time points after glucose administration. *P < 0.05, **P < 0.01, ***P < 0.001.
Fig. S2. Assessment of glucose tolerance at 8 wk of age. (A) Glucose tolerance was assessed by measuring glucose levels at the indicated time points after glucose administration and (B) area under the curve (AUC) analysis was performed. n = 6–8.
Winnay et al. www.pnas.org/cgi/content/short/1322564111
2 of 5
Fig. S3. Assessment of random fed glucose homeostasis and C-peptide at 20 wk of age. (A) Random fed glucose was measured from whole venous blood obtained between 8:00 and 10:00 AM. n = 6–20. (B) Random fed insulin or (C) C-peptide was measured from serum. n = 6–20. *P < 0.05, **P < 0.01, ***P < 0.001.
Fig. S4.
Measurement of pancreas weight. (A) Measurement of pancreas wet weight at 9 wk of age. n = 8–11. (B) Pancreas weight at 20 wk of age. n = 8–20.
Winnay et al. www.pnas.org/cgi/content/short/1322564111
3 of 5
Fig. S5. Islet morphology at 20 wk of age. Immunohistochemistry was performed using anti-insulin and anti-glucagon antibodies. ImageJ was used to merge images. Representative images are shown.
Fig. S6. Assessment of α-cell area at 9 wk of age. Total β-cell area as a percentage of total pancreas area. n = 3. *P < 0.05.
Fig. S7. Assessment of pancreatic insulin content and determination of proinsulin/insulin ratios at 20 wk of age. (A) Pancreatic insulin content was measured from acid:ethanol extracts. (B) Serum proinsulin and insulin levels were measured, and data are presented as the ratio of proinsulin to insulin. *P < 0.05.
Winnay et al. www.pnas.org/cgi/content/short/1322564111
4 of 5
Fig. S8. Evaluation of inositol requiring 1α (IRE1α) phosphorylation and oxidative stress at 20 wk of age. Immunohistochemistry was performed using antiinsulin, anti-pIRE1α, or anti-nitrotyrosine antibodies. Anti-insulin and anti-pIRE1α immunohistochemistry was performed on the same section, whereas antinitrotyrosine immunohistochemistry was performed on a serial section. Representative images are shown.
Fig. S9. Quantification of cleaved caspase-3. Cleaved caspase-3 was quantified from three independent experiments using ImageJ. **P < 0.01.
Winnay et al. www.pnas.org/cgi/content/short/1322564111
5 of 5
