Gene Silencing of Phogrin Unveils Its Essential Role in Glucose-Responsive Pancreatic β-Cell Growth

Anti-phogrin and anti–IA-2 antibodies were raised in rabbits against the luminal region of phogrin and the luminal region of IA-2, respectively, and were affinity-purified. The guinea pig anti-insulin antibody and the anti–α-tubulin and anti–β-actin mouse monoclonal antibodies were purchased from Sigma. Anti-adaptin, anti–carboxypeptidase E (anti-CPE), anti–calcium-dependent activator protein (anti-CAPS), anti-munc18, anti-dynamin, anti–clathrin heavy chain, anti–phosphatidylinositol 3-kinase (anti-PI3K), anti-Grb2, and anti–extracellular signal–related kinase (anti-ERK) mouse monoclonal antibodies and anti–insulin receptor β-subunit (anti-IRβ) rabbit polyclonal antibodies were from BD Biosciences. Anti-IRS1 and anti-STAT5 rabbit polyclonal antibodies were from Cell Signaling. Anti-IRS2 rabbit polyclonal and anti-epidermal growth factor receptor sheep polyclonal antibodies were from Upstate Biotech. Anti–IGF-IR, anti-IRβ, anti-syntrophin, and anti–5-bromo-2′-deoxyuridine (anti-BrdU) mouse monoclonal antibodies were from Lab Vision, Chemicon, Affinity Bioreagents, and Roche Diagnostics, respectively. Anti–vesicle-associated membrane protein (anti-VAMP2) and anti–green fluorescent protein (anti-GFP) rabbit polyclonal antibodies were from WAKO chemicals and MBL, respectively.

The RNA polymerase III H1 gene promoter was cloned to construct small interfering RNA expressing plasmid vector (pSUPER) according to the reference (35). Oligonucleotides (64-base) corresponding to sense target sequence, hairpin loop, and antisense target sequence were synthesized, annealed together, and then ligated into pSUPER vector. The target sense sequences were as follows: GGTCACCTTACAGAAGCTC (shPhogrin1) (mouse Phogrin mRNA, 244–264), GCCACAACTCACACTACAA (shPhogrin2) (mouse, 1,877–1,895; rat, 1,886–1,904), and GGATACATCCTCACAGGAA (shPhogrin3) (mouse, 1,492–1,510; rat, 1,501–1,519) for phogrin, and GTCTGTATTCAGGATGGCT (shIA2) (mouse, 163–181; rat, 172–190) for IA-2.

MIN6 cells before passage 25 were cultured in Dulbecco's modified Eagle's medium (DMEM) with 15% fetal bovine serum (FBS) and 50 μmol/l 2-mercaptoethanol. INS-1E cells were cultured in RPMI 1640 with 10% FBS, 10 mmol/l HEPES (pH 7.4), and 50 μmol/l 2-mercaptoethanol. Transfections were performed with Lipofectamin 2000 reagent (Invitrogen, Carlsbad, CA). MIN6 cells were transfected with pcDNA3 vector plus pSUPER plasmid, and stable clones were selected in the presence of G418 (36). Isolated colonies of the shPhogrin1-transfected cells were transferred to new culture dishes for propagation (a total of 192 clones: 72 clones in the 1st round, 120 clones in the 2nd round), but only two lines grew up to sufficient scales. Stable cell mixture transfected with empty pSUPER vector was designated as MshV, and individual clones with reduced phogrin expression were designated as MshP (MshP#33 and MshP#44). Insulin-producing cell lines from β-cell–specific insulin receptor–knockout (βIRKO) mice were established as described previously (37). Control cell lines (wild type) were from littermates of βIRKO mice. Mouse pancreatic islets were isolated and cultured as described previously (38).

H1-RNA promoter and the inserted DNA were cut by SmaI and HincII and then ligated into the promoterless cosmid pAdex vector (pAxcw; TaKaRa Biomedicals, Otsv, Japan). A full-length mouse phogrin cDNA was transferred to the pAdex vector (pAxCAwt). Viral production and propagation was generated using HEK 293 cells. Positive clones were selected by immunofluorescence analysis of MIN6 cells with anti-phogrin antibody. Purified adenoviruses were prepared by CsCl density gradient centrifugation.

After 6 h from adenoviral infection, cells (∼1.5 × 10⁶ MIN6 cells/well, ∼2 × 10⁶ INS-1 cells/well) were seeded into a six-well plate for culture. The cells were extracted with acid ethanol for 15 h at 4°C. After clarification of the extracts by centrifugation, the insulin concentration was measured by radioimmunoassay (RIA) (Eiken Chemical, Tokyo, Japan). The infected cells were preincubated in modified Krebs-Ringer buffer (KRB) (39) containing 2 mmol/l glucose for 2 h and then incubated with KRB supplemented with glucose at 25 mmol/l (for MIN6) or 16.8 mmol/l (for INS-1) for 30 min. The cell number and the cellular protein content were measured in all secretion experiments. Similar cell number was maintained among all the samples. Thereby, normalized secreted insulin to either cell number, total cellular proteins, or total insulin content did not show any difference (data not shown).

After 6 h from adenoviral infection, MIN6 and INS-1E cells were seeded into six-well plates at 0.5 × 10⁶ and 1.5 × 10⁶ cells/well, respectively. At an indicated time, cells were collected, and the cell number was measured by a CyQUANT cell proliferation assay kit (Invitrogen). The proliferative activity of infected cells was detected by BrdU incorporation. Cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) to calculate the rate of BrdU-positive cells. DNA synthesis rates were measured by [³H]thymidine incorporation into islet cells. After 16 h from adenoviral infection, mouse islets were cultured for 48 h. [³H]thymidine was added at a final concentration of 1 mCi/ml to pools of 50 islets for an additional 24 h. The proteins and DNA were precipitated with ice-cold 10% trichloroacetic acid and solubilized in 0.3 N NaOH. Aliquots were counted in scintillation fluid and assayed for protein using the Bradford method.

Immunoprecipitation and immunoblot analyses were performed as described previously (36). For immunoprecipitation, MIN6 cells were extracted with lysis buffer (20 mmol/l Tris, pH 7.5, 150 mmol/l NaCl, 0.5% Nonidet P-40, 1 mmol/l EGTA, 0.5 mmol/l phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 10 μg/ml pepstatin). Band density was measured by densitometry, quantified using Gel plotting macros of NIH image 1.62 program, and normalized to an indicated sample in the identical membrane.

Subcellular fractionation was performed as described previously (39). Cells were suspended in buffer containing 250 mmol/l sucrose, 20 mmol/l HEPES (pH 7.4), 2 mmol/l MgCl₂, 2 mmol/l EGTA, and the protease inhibitors. The cells were homogenized for 40 strokes by the tight-fitting dounce homogenizer. The total homogenate was centrifuged at 700 × g for 15 min to pellet the nuclear and intact plasma membrane fraction. The resultant supernatant was then centrifuged at 12,000 × g for 20 min to separate the heavy organelle fraction, including the secretory granules, from the cytoplasmic materials. Fractionates were lysed, and equal proportions of each lysate were subjected to immunoprecipitation and immunoblotting analysis.

Quantitative RT-PCR analysis was performed as described previously (40). PCRs were performed using the following specific primers: phogrin, 5′-AGCCACGGTACCTTGTACAT-3′ and 5′-TTGTATGGCTCCAGCAACTG-3′ (237 bp); IRS2, 5′-TATCGCCATCGATGTGAGAGG-3′ and 5′-GTAGCGCTTCACTCTTTCACG-3′ or 5′-GCAGCACTTTACTCTTTCACC-3′ (322 bp). Amplified signals stained with ethidium bromide were quantified by ATTO cool saver system.

Results are given as the mean ± SE, except where indicated otherwise. Differences between groups were analyzed using Student's t test. P values <0.05 were considered statistically significant.

We developed a new antibody (MatN2) with a high specificity to phogrin by immunoblotting and immunostaining. To compare the phogrin expression in various endocrine cell lines, whole-cell extracts were immunoblotted with MatN2. Phogrin was highly expressed in two pancreatic β-cell lines, MIN6 and βHC9, whereas other cell lines, including rat INS-1, expressed a lower level of phogrin (supplementary Fig. S1A, available in an online appendix at http://dx.doi.org/10.2337/db08-0970). We further produced an antibody specific to IA-2 and obtained a similar result (supplementary Fig. S1B). However, we noted that IA-2 expression was inconstant and changeable by cell culture conditions (data not shown), which was consistent with the previous observations (32–34). From these results, we chose MIN6 (high expression type) and INS-1 (moderate expression type) for the knockdown study.

We produced three kinds of plasmids that direct the synthesis of shRNAs targeted against the phogrin sequence (shPh1, shPh2, and shPh3). Transient expression of each shRNA plasmid selectively reduced the production of phogrin–enhanced GFP (EGFP) but not the control EGFP (data not shown); however, endogeneous phogrin was not sufficiently decreased in MIN6 cells because DNA transfection is not efficient by conventional transfection methods. Thus, we initially established shRNA-expressing stable cell lines, termed MshP#33 and MshP#44 (see research design and methods). Immunoblot analysis with MatN2 confirmed silencing of the endogenous phogrin (Fig. 1A and E). Simultaneously, expression levels of various proteins were also examined, including IA-2 (28), α-adaptin (36), syntrophin (25), STAT5 (27), and CAPS (41), all of which have been reported to interact with IA-2 family proteins directly or indirectly. IA-2 protein levels in MshP#33 and MshP#44 were less than those in control cells; however, other secretory granule–resident proteins and proteins involved in the exo/endocytosis were not altered (Fig. 1A and B). Unexpectedly, we noted that insulin signaling pathway-composing proteins were significantly affected: IRS1 and IRS2 were nearly undetectable, and PI 3-kinase p85 and Grb2 were high (Fig. 1C and E). Other proteins were unaffected in expression (Fig. 1D).

Because proliferation of MshP#33 and MshP#44 was extremely slow, we found it difficult to keep these cells invariant throughout a number of experiments. Therefore, we generated an adenovirus expression system to achieve high efficiency shRNA expression. MIN6 cell lysates were prepared at 72 h after the infection with adenoviruses bearing shPh1, shPh2, shPh3, shIA2, or control vector. By immunoblotting, endogenous phogrin was reduced by 60∼90% in shPhogrin-expressing cells, and Ad-shPh3 was most effective (Fig. 2A). The knockdown effect was specific, because expression levels of IA-2 and α-tubulin were unaffected. Thus, reduction of IA-2 protein in MshP#33 and MshP#44 cells may have resulted from long-term cell cloning. Similarly, specific knockdown of IA-2 was accomplished by adenovirus shIA2 (Fig. 2A). Furthermore, the silencing effects of adenoviral shPhogrin were verified by immunostaining (Fig. 2B). Phogrin expression was reduced at 24 h after infection and most signals were disappeared at 72 h in the shPh3-expressing MIN6 cells (Fig. 2C). Insulin staining patterns were unchanged in these phogrin-knockdown cells.

To assess the involvement of phogrin in the β-cell secretory function, GSIS and total insulin content were determined in MIN6 cells. Knockdown of phogrin or IA-2 did not result in any notable effects on GSIS and insulin content (Fig. 2D). Similar results were shown for INS-1E cells, which displayed no significant difference in GSIS and insulin content between shPh3-expressing cells and control cells (Fig. 3A–C). Conversely, adenovirus-mediated phogrin overexpression did not affect GSIS and insulin content in INS-1E cells (Fig. 3E–G). These observations indicate that insulin secretion in the β-cell lines remained unaltered when phogrin or IA-2 was attenuated or increased.

Next, we determined the cell proliferation rate by measuring cell number. Silencing of phogrin or IA-2 reduced MIN6 cell growth in a time-dependent manner (Fig. 2E). No proapoptotic effects were observed at 72 h [5.8 (shVec), 6.7 (shPh1), 6.2 (shPh2), 5.7 (shPh3), and 6.8 (shIA2) % cell death rates]. The proliferation rates of shPh1, shPh2, shPh3, and shIA2-infected MIN6 cells, assessed at 72 h after infection, were 84.0, 77.3, 72.5, and 69.9%, respectively, compared with shVector-infected cells (P < 0.001). Moreover, the proliferation rate by BrdU incorporation per DAPI-positive cells revealed a 30% decrease of BrdU incorporation in shPh3-expressing cells (Fig. 2F). We further confirmed the similar inhibitory effect of phogrin knockdown on the proliferation of INS-1E cells (Fig. 3D). Consistently, adenoviral phogrin overexpression slightly enhanced the proliferation of INS-1E cells (Fig. 3H). MIN6 cells were not affected by phogrin overexpression (data not shown) perhaps because of a high expression of endogenous protein. To test the effects of Ad-shPhogrin on islet cell replication, mouse islets were incubated with [methyl-³H]thymidine. Phogrin knockdown caused a 36% decrease in [³H]thymidine incorporation into islet cell DNA (Fig. 4A).

Because glucose is a potent regulator of β-cell growth and both MIN6 and INS-1 cells are glucose responsive, we examined whether phogrin modulates the glucose-dependent mitogenesis in MIN6 cells. MIN6 growth was abolished by the phogrin knockdown in response to increasing glucose concentrations from 11 to 25 mmol/l, although the growth rate under the low-glucose culture (<5 mmol/l) was not influenced (Fig. 4B).

Previous reports suggested that insulin/IGF-I signaling (IIS) provides β-cell mass (6). To address the function of phogrin, we investigated IIS-associated activity and expression of growth-related molecules. Phogrin knockdown in MIN6 cells resulted in a marked reduction in IRS2 protein level (Fig. 5A), as seen in shPhogrin stable cell lines (Fig. 1). Reduction of the IRS2 level was evident in phogrin-knockdown cells >48 h after infection, whereas such reduction was undetected in shVector-infected cells even at 72 h (Fig. 5A). IRS2 reduction was partially dependent on the multiplicity of infection (MOI) number of adenoviruses (Fig. 5B). In addition, insulin receptor was slightly decreased in the cells having a higher degree of phogrin knockdown (Fig. 5A and B). On the other hand, the silencing of IA-2 protein caused a modest decrease in IRS2 protein but a marked reduction of an insulin receptor level (Fig. 5B). No significant changes were noted in other proteins, including IRS1 and other unrelated proteins, suggesting the specific reduction of IRS2 and insulin receptor. Similar reduction was also observed in INS-1E cells (Fig. 5C) and mouse islets (Fig. 5D) in a time-dependent manner. Furthermore, adenoviral phogrin overexpression caused a partial recovery of IRS2 level in shPhogrin stable cell lines (Fig. 5E), suggesting that phogrin specifically regulates IRS2 expression in β-cells.

Recent studies have shown that glucose in the physiological range (5–15 mmol/l) promotes IRS2 expression at the transcriptional level (42); however, chronic exposure to high glucose (>15 mmol/l) decreases IRS2 protein levels via proteasomal degradation in β-cells (43). We examined whether IRS2 protein is reduced either by transcriptional regulation or by protein modulation in the phogrin-knockdown cells. Phogrin mRNA was almost silenced at 24 h, and its decline was faster than that of protein, as expected (Fig. 6A and B). In contrast, IRS2 protein was decreased starting 48 h later, whereas IRS2 mRNA level was barely affected even at 72 h (Fig. 6A and B). This suggests that phogrin regulates the IRS2 expression at the protein level but not at the mRNA level. Consistently, IRS2 reduction was completely blocked by the proteasomal inhibitors lactacystin and MG-132 (Fig. 6C). Thus, IRS2 protein appears unstable and is rapidly degraded by the proteasome in the phogrin-knockdown cells. Inhibitors for autophagic/lysosomal protein degradation had virtually no effect on this event (Fig. 6C). Furthermore, IRS2 protein stability was assessed by treatment of the translational inhibitor cycloheximide (42). IRS2 protein levels in the phogrin-knockdown cells decayed more rapidly than those in control cells (Fig. 6D). Taken together, phogrin plays an essential role in the stability of IRS2 protein but not in its gene transcription.

To elucidate the phogrin function, an in vitro binding assay was first performed. Sodium vanadate-treated MIN6 cell extracts were incubated with bacterially expressed GST-phogrin, and bound proteins were detected by immunoblotting with anti–phospho-tyrosine antibodies. Specific bands ∼95, 100, and 130 kDa were detected in the membrane fraction to bind to GST-phogrin but not to GST alone (supplementary Fig. S2, available in the online appendix). The result indicates that tyrosine-phosphorylated transmembrane proteins are candidates as the phogrin-binding molecules. Coimmunoprecipitation experiments using MIN6 cell lysates revealed that both phogrin and IA2 immunoprecipitates contained a significant amount of insulin receptor β-subunit (IRβ) without its precursor (Fig. 7A; supplementary Fig. S3A, available in the online appendix). IRβ was not detected by the control antibodies. Furthermore, IRβ was barely found in the phogrin immunoprecipitates from the phogrin-knockdown cells (Fig. S3B). These results suggest that phogrin specifically interacts with the mature form of insulin receptor. Unexpectedly, IRS2 and IGF-IR were not coimmunoprecipitated with phogrin and IA-2 (Fig. 7A). When MIN6 cells were cultured in low-glucose condition, their interaction was detected at a low level by coimmunoprecipitation (Fig. 7B, 0-h point). Importantly, when cells were placed into the high-glucose culture, IRβ content increased dramatically in the phogrin immunoprecipitates, although there was no change in the phogrin and insulin receptor expression levels (Fig. 7B). Notably, IRS2 apparently increased during the high-glucose stimulation in MIN6 cells (Fig. 7B, right panel), as previously shown for primary rat β-cells (42). Blockade of secretory granule exocytosis by diazoxide, a potassium channel opener, completely prevented the coimmunoprecipitation of insulin receptor (Fig. 7C). Thus, the high-glucose culture promotes molecular interaction of phogrin and insulin receptor presumably via the translocation of phogrin into the plasma membrane. To confirm this idea, a conventional fractionation procedure was used. Under the low-glucose culture, phogrin and peripherally granule-associated Rab27a (39) were mainly distributed in the secretory granule fraction (Fig. 7D). The high-glucose stimulation induced significant redistributions of phogrin and Rab27a from secretory granule to the plasma membrane fraction, whereas the plasma membrane-associated insulin receptor and syntaxin 1 remained in the plasma membrane. Coimmunoprecipitation analyses using each fraction revealed that molecular interaction between phogrin and insulin receptor occurred at the plasma membrane under the high-glucose condition (Fig. 7D, right bottom panel). However, our results do not exclude a possibility that these proteins form a complex on secretory granule that attached to the plasma membrane (44).

Finally, we used two β-cell lines established from control (wild-type) and IRKO mice for proliferation assay. BrdU incorporation assay revealed that silencing of phogrin failed to show the growth retardation of IRKO β-cells, whereas a 33% decrease in growth rate was observed for wild-type β-cells (Fig. 8A). Furthermore, IRS2 protein degradation caused by phogrin knockdown was not observed in IRKO β-cells (Fig. 8B and C). This suggests that regulation of IRS2 stability and β-cell growth by phogrin depends on its interaction with insulin receptor.

Proliferation rates of phogrin-downregulated MIN6 stable cells (Fig. 1) and phogrin-overexpressed stable cells (S. Torii and T. Takeuchi, unpublished data) were reduced and elevated, respectively. Thus, phogrin appeared to be a positive regulator of β-cell growth. Because studies using stable cells seem inadequate for evaluation of primary functions of phogrin, we generated adenoviruses expressing shRNA. As expected, silencing of phogrin by Ad-shPhogrin caused a marked retardation of cell growth in both cultured cell lines and mouse islets (Figs. 2–4). However, GSIS stayed unaffected by phogrin or IA-2 knockdown (Figs. 2 and 3). Previous studies have shown that the deletion of IA-2 or phogrin gene or in combination in mice resulted in mild impairment of GSIS but did not affect β-cell mass (19–21). However, more recent studies have suggested that islets from double knockout mice failed to show any secretion defect (45), and β-cell regeneration was reduced in partially pancreatectomized ICA512 (IA-2) knockout mice (46). Therefore, it is possible that the alteration in GSIS in knockout mice is indirect and that β-cell mass is recovered by compensatory function of other genes. Our present results support the notion that the primary function of phogrin is a regulation of β-cell growth but not of insulin secretion.

Glucose is a potent mitogen for β-cells and regulates various cellular dynamics, including insulin secretion and nutritional metabolism (1). Although several signaling molecules such as PKC have been thought to be a mediator for glucose-induced β-cell growth (1), recent studies have indicated a novel pathway in which the autocrine/paracrine function of secreted insulin promotes β-cell proliferation (3,4). We demonstrated that β-cell growth retardation induced by phogrin knockdown was observed under the high-glucose culture (Fig. 4B). Because phogrin was found to form a complex with insulin receptor (Fig. 7) and because cell growth retardation by phogrin knockdown was not observed in insulin receptor–deficient cells (Fig. 8), insulin receptor is evidently a functional target of phogrin. Phogrin localizes to insulin-containing secretory granules and translocates to the plasma membrane whenever insulin exocytosis is induced (Fig. 7D) (36). Thus, interaction of phogrin with insulin receptor is coordinately coupled with the autocrine action of insulin. Their interaction was strikingly promoted by the high glucose-induced insulin secretion (Fig. 7B–D). In other words, phogrin regulates the glucose-induced β-cell growth through modulating the autocrine insulin signaling. Insulin secretion was never influenced by overexpression or repression of phogrin (Figs. 2 and 3), thereby, modulation of autocrine effects by phogrin is not intervened by extracellular insulin dosage.

Our subcellular fractionation experiments indicate that phogrin interacts with insulin receptor on the plasma membrane (Fig. 7D). However, EGFP-tagged phogrin in MIN6 or PC12 cells did not spread to whole plasma membrane under the evanescent microscopy observation (44,47,48), suggesting that it remains on secretory granules during exocytotic events. On the other hand, the experimental result that phogrin antibody in culture medium accessed to the cell surface phogrin protein in MIN6 cells (49) suggests that phogrin positively reaches the cell surface. Thus, the plasma membrane fraction in our assay may contain the attached granules that had been connected to the plasma membrane by lipid bilayers merger. Because insulin receptor is reportedly distributed to the methyl-β-cyclodextrin–sensitive microdomains of the plasma membrane in HIT-T15 cells (50), it is possible that phogrin and insulin receptor colocalize at uncharacterized specific domains equivalent to the secretory granule targeting/fusion sites.

Our data indicate that phogrin interacts with insulin receptor and stabilizes IRS2 protein in β-cells. This hypothesis raises a question: What is the molecular mechanism of this interaction? Because phogrin and IA-2 have an inactive PTP domain in their cytoplasmic tail, it is possible that they bind to insulin receptor directly. In fact, the phogrin cytoplasmic fragment could bind to tyrosine-phosphorylated 95- to 100-kDa proteins in vitro (supplementary Fig. S2). Phogrin may regulate the insulin signaling through the tyrosine phosphorylation/dephosphorylation cycle. Insulin or IGF-I promotes degradation of IRS2 through PI 3-kinase, Akt, and mTOR signaling in adipocytes, hepatoma, and embryonic fibroblast cells (51). Also, chronic exposure (>8 h) to high glucose and/or IGF-I induces proteasomal degradation of IRS2 in β-cells (43). These proposed signaling pathways may be mediated by phogrin function for stabilization of IRS2.

Ablation of IRS2 in β-cells led to a decrease in β-cell mass and an increase in islet apoptosis (5). In contrast, IRS2 expression in mice prevented diabetes by promoting β-cell growth (52). Furthermore, IRS2 expression induced by glucose, GLP-1, and other signaling proteins has been shown to contribute to their regulatory functions for proliferation, indicating that IRS2 is a master regulator of β-cell growth (14). Our data further indicate that IRS2 protein is stabilized under the control of phogrin and insulin receptor interaction for its cell growth regulation. This is because silencing of phogrin resulted in a marked reduction of IRS2 without change in other protein levels. Furthermore, adenoviral overexpression of phogrin caused a partial recovery of IRS2 level in MshP#33 and MshP#44 cells (Fig. 5E). Thus, phogrin functions as a primary regulator of secreted insulin-mediated β-cell growth by stabilizing IRS2 protein.

To respond to hyperglycemia, β-cells proliferate and expand to compensate for increased insulin secretion demand (14). The present observations suggested a novel mechanism in which phogrin contributes to glucose-induced proliferation of β-cells via insulin receptor and IRS2. Interestingly, insulin receptor and IRS2 mRNA levels reportedly decreased in islets isolated from human type 2 diabetes (53). To prevent the decrease in insulin receptor and IRS2 functions, we suggest that phogrin and IA-2 in the β-cells are potential therapeutic targets for treating diabetes.

S.T. has received a grant-in-aid for Young Scientists from the Ministry of Education, Culture, Sports, Science, and Technology of Japan and support from the Uehara Memorial Foundation. R.N.K. has received National Institutes of Health Grant RO1-DK-67536. T.T. has received a grant of the Global Center of Excellence program from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

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

We thank Dr. P. Maechler (Centre Médical Universitaire) for providing INS-1E cells and Drs. R. Agami (The Netherlands Cancer Institute), M. Nishita (Kobe University), and S. Mizutani (Gunma University) for technical support. We also thank Dr. M. Hosaka, M. Kosaki, and M. Hosoi for secretarial and technical support.