Insulin Receptor Substrate 1-induced Inhibition of Endoplasmic Reticulum Ca2+ Uptake in β-Cells

To understand the role of the insulin receptor pathway in β-cell function, we have generated stable β-cells (βIRS1-A) that overexpress by 2-fold the insulin receptor substrate-1 (IRS-1) and compared them to vector-expressing controls. IRS-1 overexpression dramatically increased basal cytosolic Ca2+ levels from 81 to 278 nm, but it did not affect Ca2+ response to glucose. Overexpression of the insulin receptor also caused an increase in cytosolic Ca2+. Increased cytosolic Ca2+ was due to inhibition of Ca2+ uptake by the endoplasmic reticulum, because endoplasmic reticulum Ca2+ uptake and content were reduced in βIRS1-A cells. Fractional insulin secretion was significantly increased 2-fold, and there was a decrease in βIRS1-A insulin content and insulin biosynthesis. Steady-state insulin mRNA levels and glucose-stimulated ATP were unchanged. High IRS-1 levels also reduced β-cell proliferation. These data demonstrate a direct link between the insulin receptor signaling pathway and the Ca2+-dependent pathways regulating insulin secretion of β-cells. We postulate that during regulated insulin secretion, released insulin binds the β-cell insulin receptor and activates IRS-1, thus further increasing cytosolic Ca2+ by reducing Ca2+ uptake. We suggest the existence of a novel pathway of autocrine regulation of intracellular Ca2+homeostasis and insulin secretion in the β-cell of the endocrine pancreas.

The insulin-secreting ␤-cell of the endocrine pancreas has a central role in regulating glucose homeostasis (1,2). It is now recognized that ␤-cell failure is a major contributing factor to type 2 diabetes mellitus, thus emphasizing the importance of elucidating the normal mechanisms of insulin secretion (3,4). Glucose oxidation by the ␤-cell is essential for insulin secretion. In particular, glucokinase, the first step in glycolysis, has been convincingly shown to be the ␤-cell glucose sensor (5). ␤-Cell metabolism of glucose results in an increase in the ATP/ADP ratio leading to closure of the K ATP channel, depolarization of the ␤-cell, and influx of extracellular Ca 2ϩ through voltage-de-pendent Ca 2ϩ channels. The subsequent increase in intracellular Ca 2ϩ then activates insulin exocytosis. The possibility of other signaling pathways involved in glucose-induced insulin secretion has also been suggested (6 -11).
Since the discovery of the insulin receptor in insulin-secreting ␤-cells by Rothenberg and colleagues (12,13), a rapidly growing body of evidence indicates that the insulin receptor signaling pathway is active in pancreatic ␤-cells (14) and plays an important role in ␤-cell regulation (4,(12)(13)(14)(15)(16)(17). Activation of the ␤-cell insulin receptor (IR) 1 results in rapid tyrosine phosphorylation of the IR ␤-subunit and the IR substrate proteins (12). Deletion of IR results in neonatal death in mice (18,19) and leprechaunism in humans (20). Mice with heterozygous null alleles of IR and insulin receptor substrate 1 (IRS-1) (IR/ IRS-1 ϩ/Ϫ ) exhibit hyperinsulinemia and ␤-cell hyperplasia and develop overt diabetes (21). Knockouts of the IRS-1 and IRS-2 produce different effects. Inactivation of IRS-1 (IRS-1 Ϫ/Ϫ ) leads to mild insulin resistance, hyperinsulinemia, and ␤-cell hyperplasia with no overt diabetes syndrome (4,17,22). In contrast, inactivation of IRS-2 (IRS-2 Ϫ/Ϫ ) results in ␤-cell failure and causes type 2 diabetes (17). The differential effects of IRS-1 and IRS-2 knockout indicate that the two major IR substrates mediate differential signals in ␤-cells, but the mechanisms accounting for such differential regulation and for IRS-1 function are still unknown.
Cellular Ca 2ϩ is a critical element in ␤-cell regulation. Rising intracellular Ca 2ϩ ([Ca 2ϩ ] i ) is an obligated step for glucose induced insulin secretion (1,23). Abnormal [Ca 2ϩ ] i is a common defect in both insulin-dependent type 1 diabetes and insulinindependent type 2 diabetes (24). Altered Ca 2ϩ metabolism has also been reported to affect ␤-cell function including insulin biosynthesis (25,26). The endoplasmic reticulum (ER) plays an important role in the regulation of intracellular Ca 2ϩ concentrations (27)(28)(29). Endoplasmic reticulum Ca 2ϩ -ATPase (SERCA) is the major calcium pump that sequestrates cytosol Ca 2ϩ into ER lumen (28,30). Thapsigargin, a nonphorboid tumor promoter, specifically inhibits ER Ca 2ϩ -ATPase activity (31). Addition of thapsigargin to pancreatic ␤-cells leads to elevated cytosol Ca 2ϩ concentration and enhanced short term glucose-stimulated insulin secretion (32). Recent data showed that IRSs may directly interact with ER Ca 2ϩ -ATPase (SERCA1 and SERCA2) in a tyrosine phosphorylation-dependent manner in muscle and heart (33). This finding suggests that insulin may via insulin receptor signaling pathway regulate ER Ca 2ϩ -ATPase activity, therefore influencing cellular Ca 2ϩ homeostasis. It is currently unknown whether insulin exerts any regulatory role in ␤-cell Ca 2ϩ homeostasis.
To dissect the role of IRS-1 in ␤-cell function, we have overexpressed IRS-1 in an insulin-secreting ␤-cell line. We show that IRS-1 regulates ␤-cell Ca 2ϩ homeostasis, insulin biosynthesis, and ␤-cell proliferation and that elevated expression of IRS-1 induces abnormal Ca 2ϩ homeostasis and ␤-cell dysfunction.

MATERIALS AND METHODS
Cell Lines and Culture Media-The clonal mouse ␤-cell line ␤TC6-F7 and culture conditions were previously described (15,34). In brief, cells were maintained in high glucose Dulbecco's modified Eagle's medium (25 mM glucose; Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), 100 units/ml penicillin, 50 g/ml streptomycin and incubated at 37°C in a 10% CO 2 /90% air humidified incubator. The Chinese hamster ovary-T cell line was a kind gift from Dr. R. Roth (Stanford University School of Medicine, Stanford, CA) and was maintained in F-12 medium with 5% fetal bovine serum, 100 units/ml penicillin, 50 g/ml streptomycin at 37°C in a 5% CO 2 /95% air atmosphere.
Construction of the IRS-1 Expression Plasmid-The IRS-1 expression vector was constructed as follows. A 3.9-kilobase DNA fragment containing the mouse IRS-1 cDNA and the c-Myc epitope tag was excised from pQE31-mIRS1 (kindly provided by Dr. Ronald Kahn, Joslin Diabetes Center, Boston, MA) with EcoICRI and HindIII. The fragment was blunt-ended with Klenow fragment (New England Biolabs, Beverly, MA) and ligated into a SmaI digested vector pCI-Neo (Promega, Madson, WI). The resulting plasmid pCMV-IRS1 was purified with QIAGEN Maxi-purification Kit (Qiagen, Chatsworth, CA). Enzymes for recombinant DNA procedures were from Promega or New England Biolabs.
Transfection of ␤-Cells-Cells were transfected with cationic liposome reagent DMRIE-C (Life Technologies, Inc.) as described before (15). The transfected cells were selected with 500 g/ml neomycin (Geneticin, Life Technologies, Inc.) for 4 weeks, and the surviving colonies (defined as passage 4) were individually picked and transferred to 24-well plates. Then the IRS-1 and c-Myc protein levels were determined with anti-IRS1 polyclonal antibody (catalog no. 06-248, Upstate Biotechnology, Lake Placid, NY) and anti-c-Myc monoclonal antibody 9E10 (catalog no. sc-40, Santa Cruz Biotechnology, Santa Cruz, CA), respectively.
Immunoprecipitation and Western Analysis-Preparation of cell lysates, immunoprecipitation, and Western blotting were performed essentially as described previously (12,35). Tyrosine-phosphorylated proteins were detected with rabbit polyclonal anti-phosphotyrosine antibody K-18 (kindly provided by Dr. P. Rothenberg, University of Pennsylvania). A secondary antibody, rabbit anti mouse IgG (Sigma, catalog no. M7023), was used for c-Myc antibody immunoprecipitation and immunoblotting.
Measurement of Cellular Nucleotides-Cellular contents of nucleotides including ATP were measured by high performance liquid chromatography (HPLC) as described before (38). In brief, cells were washed with ice-cold phosphate-buffered saline and extracted with 1 ml of 5% trichloracetic acid. Cells were scraped, transferred to 1.5-ml Eppendorf tubes, incubated 5 min at ϩ4°C, and centrifuged at 14,000 ϫ g (10 min at ϩ4°C). The supernatant was collected into borosilicate 12 ϫ 75-mm tubes, extracted three times with 1 ml of diethyl ether, and centrifuged, and the ether phase was discarded. Samples (300 l) and known amounts of standard NAD, ADP, GDP, ATP, GTP, UTP, CTP, and TTP were analyzed on a Varian HPLC apparatus (Varian, Sugarland, Texas), using a 250 ϫ 4.6 mm Partisil SAX (10 m) column equipped with a guard column. The gradient of solvent A (1.25 MNaH 2 PO 4 , pH 3.8) over 80 min was as follows: 0 min, 100% water; 1-15 min, 1% A; 25 min, 28% A; 45 min, 32% A; 70 min, 50% A; 70 -80 min, 100% A; 80 -95 min, 100% water. The flow rate was 1.5 ml/min. Nucleotides were detected by UV spectrophotometry (340 nm) and quantitated by comparison to the corresponding standard curve run in parallel. The typical retention times were as follows : 8 min for NAD, 10 min for AMP, 35 min for ADP, 27 min for GDP, 32 min for UTP, 34.5 min for CTP, 34.8 min for TTP, 40 min for ATP, and 46 min for GTP.
Cytosolic free Ca 2ϩ Measurement-For cytosolic free Ca 2ϩ measurement, the cells were seeded on microscope glass coverslips (Fisher) coated with poly-D-lysine (Sigma), and grown for 2 days. Cells were then loaded with the calcium indicator fura-2 for 30 min at 37°C in 2 ml of KRB plus 0.1% BSA and supplemented with 2.0 M fura-2 acetoxymethylester (Molecular Probes, Eugene, OR) and 0.2 mg/ml pluronic F-127 (Molecular Probes). The coverslip with the loaded cells was then mounted in a perifusion chamber placed on the homeothermic platform of an inverted Zeiss microscope. The cells were superfused with Krebs-Ringer buffer (0.1% BSA) at 37°C at a flow rate of 1.5 ml/min. For experiments using KRB without Ca 2ϩ , CaCl 2 was omitted and 1 mM EGTA (Sigma) was added to chelate all Ca 2ϩ . Ca 2ϩ measurement is as follows and was also described in detail elsewhere (37,39). The microscope was used with a ϫ 40 oil objective. Fura-2 was successively excited at 334 and 380 nm by means of two narrow band-pass filters. The emitted fluorescence was filtered through a 520 nm filter, captured with an Attofluor CCD video camera at a resolution of 512 ϫ 480 pixels, digitized into 256 gray levels and analyzed with version 6.00 of the Attofluor RatioVision software (Atto Instruments, Rockville, MD). The concentration of Ca 2ϩ at each pixel was calculated by comparing the ratio of fluorescence to an in vitro two-point calibration curve. The Ca 2ϩ concentration is presented by averaging the values of all pixels of a cell body. Data were collected from at least 10 individual cells in each measurement at an interval of 4.5 s.
The endoplasmic reticulum Ca 2ϩ -ATPase inhibitor thapsigargin was purchased from Biomol (Plymouth Meeting, PA) and dissolved in dimethyl sulfoxide.
Measurement of Endoplasmic Reticulum Ca 2ϩ Uptake-Endoplasmic reticulum Ca 2ϩ uptake was measured as described (40,41). Briefly, the cells were trypsinized and washed twice with KRB without Ca 2ϩ or glucose (0.1% BSA). Then, they were permeabilized with 20 g/ml digitonin (Sigma) in KRB without Ca 2ϩ or glucose (0.1% BSA) at 37°C for 10 min. The cells were then incubated for 30 min in TES buffer (100 mM TES, pH 7.2, 100 mM KCl, 2.5 mM MgCl 2 , 0.2 mM EGTA, 5 Ci/ml of 45 Ca (ICN, Lisle, IL), 5 Ci/ml of 3 H 2 O (ICN) for diffusible space correction, 5 g/ml of ruthenium red, 0.1% BSA) at the indicated Ca 2ϩ concentration. The incubation medium was removed by aspiration, and the cells were solubilized with 100 l of 1 M NaOH and neutralized with 100 l of 1 M HCl. Radioactivity content of the endoplasmic reticulum was measured by liquid scintillation spectrometry. Calcium concentrations were titrated with a calcium electrode (Orion, Beverly, MA). Calculation of endoplasmic reticulum calcium uptake was as described before (41).
Pulse-Chase Labeling of ␤-Cells and Determination of Insulin Biosynthesis-The rate of insulin biosynthesis was determined with [ 3 H]leucine pulse-chase labeling assay. The cells were seeded in 6-well culture dishes at 3 ϫ 10 5 cells/well and grown for 2 days in high glucose Dulbecco's modified Eagle's medium (Life Technologies, Inc.). The cells were then washed twice with KRB buffer and labeled with 40 Ci/ml of L-[3,4,5-3 H(N)]leucine (6660 GBq/mmol, NEN Life Science Products) in KRB buffer supplemented with 0.25% BSA and 0 or 16 mM glucose. After being labeled for 30 min, the cells were chased with 1 mM cold leucine for 30 min at the same glucose concentration as used in the labeling process. The cells were then extracted with 1 M acetic acid and sonicated. One hundred l of the clarified acetic acid extract were used for trichloroacetic acid precipitation as described (15) to determine the total cellular content of 3 H-labeled proteins. The 3 H-labeled insulin immunoreactive material was immunoprecipitated with 3 l of undiluted guinea pig anti-bovine insulin serum (ICN) and protein A-Sepharose CL-4B beads (Amersham Pharmacia Biotech). The radioactivity was determined by liquid scintillation counting. Normal guinea pig serum (Linco, St. Charles, MO) was used as background control. Recovery of insulin immunoreactive material was more than 90% under the assay condition as determined with 125 I-labeled insulin.
RNA Protection Assay-RNA protection assays were performed according to Ambion's protocol (Ambion, Austin, TX). Briefly, the cells were grown in culture medium for 2 days and then trypsinized and washed twice with Dulbecco's phosphate-buffered saline. To 4 ϫ 10 6 cells, 0.4 ml of Ambion's lysis/denaturation solution (direct protect kit, Ambion) were added to lyse the cells. An insulin probe was generated by in vitro transcription using pGEM-rPPI (gift of Dr. Christopher Rhodes, Joslin Diabetes Center, Boston, MA) as a template. For internal control, a probe for 18 S RNA was obtained by in vitro transcription using pT7RNA18S (Ambion) as a template. Labeling of the 18 S RNA probe was done with [ 32 P]UTP in the presence of 500 M cold UTP to lower the specific activity. The labeled transcripts were gel purified as described in Ambion's manual. The insulin probe was 430 bases in length and completely complementary to the insulin mRNA. The 18 S rRNA probe was 116 nucleotides in length, 80 of which are complementary to 18 S rRNA (Ambion). Hybridization, RNase and protease digestion, and separation detection were carried out as per Ambion's instructions. The results were quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Data Analysis-Data were analyzed by paired t test or one-way or two-way ANOVA; p Յ 0.05 was considered significant. Nonlinear regression and curve fitting were performed with PRISM 2.01 (GraphPad Software, San Diego, CA).

RESULTS
Overexpression of IRS-1 in ␤-Cells-We overexpressed IRS-1 in a clonal ␤-cell line ␤TC6-F7. The exogenous IRS-1 had a c-Myc tag at its C terminus and migrated slightly slower than the endogenous IRS-1 in SDS-gels (Fig. 1A, lanes 1, 3, and 4). ␤-IRS1-A, one of the 11 stable transfectants tested, expressed IRS-1 protein two times (199 Ϯ 36%) more than the controls (p ϭ 0.02) (Fig. 1B). Fifty percent of the increase was contributed by the exogenous IRS1-Myc; the other 50% was from the endogenous IRS-1. Addition of 100 nM insulin led to rapid tyrosine phosphorylation of proteins in the 160 -180 kDa range that co-migrated with the IRS-1 protein (Fig. 1C, upper gel, lanes 2, 4, and 6) (also see Ref. 12). A 120-kDa protein (p120) was also heavily tyrosine-phosphorylated in the ␤-cells. The abundance of p120 and its extent of tyrosine phosphorylation are relatively stable in the three ␤-cell lines ( Fig. 1C and data not shown), and it was therefore used to normalize the level of IRS-1 phosphorylation. Normalized level of IRS-1 tyrosine phosphorylation in the ␤IRS1-A cells was 2-fold higher than that in the control cells ( Fig. 1, C, upper gel, lane 4, and D). This is proportional to the elevated IRS-1 protein level in the ␤IRS1-A cells. These data demonstrated that the excess IRS-1 was tyrosine-phosphorylated to the same extent as the endogenous substrate upon insulin stimulation.
Down-regulation of Insulin Content and Secretion by Overexpression of IRS-1-Nine of the transfected clones that showed detectable c-Myc signal and increased IRS-1 levels, including ␤IRS1-A, were tested for their insulin content. They all exhibited lowered insulin content. Data from the ␤IRS1-A clone are shown in Fig. 2. The ␤IRS1-A cells had significantly lowered insulin content: 29.0 Ϯ 2.9 ng/10 5 cells versus 114 Ϯ 2.2 ng/10 5 cells (Neo control), p ϭ 0.0001 ( Fig. 2A). Net insulin secretion was also reduced 61% at 0 mM glucose (G0) and 58% at 15 mM glucose (G15) (Fig. 2B). Insulin secretion of both ␤IRS1-A and control cell lines was glucose-and extracellular calcium-dependent. Removal of extracellular Ca 2ϩ and addition of 1 mM EGTA completely abolished glucose-stimulated insulin secretion. Addition of either the phosphatidylinositol 3-kinase inhibitor wortmannin (100 nM) or the p70 ribosomal S6 protein kinase inhibitor rapamycin (100 nM) did not change insulin content and secretion of ␤IRS1-A cells (data not shown).
Although the net amount of insulin secreted by the IRS-1overproducing cells was reduced as mentioned above, glucosestimulated fractional insulin secretion (the ratio of secreted insulin/total insulin content, expressed as a percentage) was significantly increased in ␤IRS1-A cells (Fig. 2C). This increased fractional insulin secretion was glucose-dependent. At 0 mM glucose, fractional insulin secretion of the ␤IRS1-A cells (4.3 Ϯ 0.9%) was not significantly different from that of the Neo control (3.0 Ϯ 0.6%) (p Ͼ 0.5). At glucose concentrations above 1 mM (stimulatory glucose concentrations for this ␤-cell line), fractional insulin secretions were increased more than 2-fold compared with the Neo control (Fig. 2C) (p Ͻ 0.04). These data indicate that overexpression of IRS-1 may enhance the capacity of the ␤-cell to secrete insulin under stimulatory glucose concentrations. Neither wortmannin (100 nM) nor rapamycin (100 nM) affected glucose-stimulated fractional insulin secretion (data not shown).
Glucose Responsiveness of the ␤-Cells-To determine whether the reduced insulin content and secretion were due to reduced glucose sensitivity, we examined glucose responsiveness of the IRS-1-overproducing ␤-cells. Glucose strongly stimulates ␤-cell metabolism, and its effect is reflected by the MTT assay (36,37). As shown in Fig. 3A, addition of 15 mM glucose doubled MTT values in both the ␤IRS1-A and the Neo control cells. No significant difference was detected between the two cell lines.
Glucose metabolism also results in the production of ATP. An increased cellular ATP/ADP ratio is a critical event triggering glucose-stimulated insulin secretion. To determine whether IRS-1 overexpression affects the ability of the ␤-cell to produce ATP upon glucose stimulation, we measured cellular nucleotide contents using HPLC. As indicated in Fig. 3B, glucose increased cellular ATP content in both Neo control and ␤IRS1-A cells: 1.7 Ϯ 0.3-and 2.0 Ϯ 0.4-fold increase respectively. No statistical difference was detected (p Ͼ 0.05).
Glucose-stimulated insulin secretion was also measured at different glucose concentrations ranging from 0 to 30 mM. Insulin secretion at 30 mM glucose was set as the maximum. Insulin secretion at other glucose concentrations was expressed as a percentage of the maximum. The data were fitted to generate sigmoidal glucose-response curves for the ␤IRS1-A and Neo control cells (Fig. 3C). No significant difference was detected between the two cell lines (two-way ANOVA test, p Ͼ 0.28). These data indicate that overexpression of IRS-1 did not change ␤-cell glucose sensitivity or its glucose metabolism in general.
Down-regulation of Insulin Biosynthesis at Translational Level-IRS-1-overproducing cells exhibited a reduced rate of insulin biosynthesis both in the absence or the presence of 16 mM glucose. Addition of 16 mM glucose increased (pro)insulin as well as the total protein biosynthesis in ␤-cells (Fig. 4A).
Increased ␤-Cell Cytosol Ca 2ϩ Level-An increase in intracellular Ca 2ϩ is recognized as a key obligatory step in insulin secretion and has been implicated mechanistically in insulin release (1,23). Calcium content in ER has been shown to affect insulin biosynthesis. Reduced ER Ca 2ϩ content results in decreased rate of translation initiation (25). It is currently unknown whether insulin receptor signaling affects ␤-cell Ca 2ϩ homeostasis. To investigate that, we measured cytosolic Ca 2ϩ levels in ␤IRS1-A and control cells using fura-2 Ca 2ϩ indicator as described under "Materials and Methods." As shown in To determine whether this increased cytosolic [Ca 2ϩ ] is specific to insulin receptor signaling, we examined cytosolic Ca 2ϩ levels in the insulin receptor-overproducing cells (Fig. 5B). We have previously established ␤-cell lines overproducing the wild type IR (the IR-S2 cells) or the kinase-deficient IR (the AK-S2 cells) (15). Compared with control cells, expression level of IR was 4-fold and 2-fold higher in the IR-S2 and AK-S2 cells, respectively (15). Cytosolic [Ca 2ϩ ] in control cell lines ␤TC6-F7 and NEO was at 90 Ϯ 9 and 88 Ϯ 8 nM, respectively. The cells overexpressing the kinase-deficient insulin receptor (AK-S2) had a [Ca 2ϩ ] of 92 Ϯ 11 nM similar to the controls. The cells

FIG. 2. Insulin contents and glucose-stimulated insulin secretion.
Insulin contents and secretion were assayed as described under "Materials and Methods." A, insulin content; B, glucose-stimulated insulin secretion. Open bars, 0 mM glucose; filled bars, 15 mM glucose. All data represent means Ϯ S.E. from at least two independent experiments each performed in triplicate. Cell number was determined by cell counting and measurement of cellular DNA content. **, p ϭ 0.0001. C, fractional insulin secretion. Secreted insulin at each glucose concentration was normalized with total cellular insulin content (secreted insulin/total insulin) and expressed as a percentage. The difference between the two fractional secretion curves is significant as analyzed with the ANOVA test (p Ͻ 0.05).
overproducing the wild type IR (IR-S2) had a significantly elevated Ca 2ϩ level: 135 Ϯ 16 nM (p Ͻ 0.01 compared with controls, a 50% increase). These data demonstrated that elevated cytosol [Ca 2ϩ ] is insulin receptor kinase-dependent.

Overexpression of IR and IRS-1 both increased cytosol [Ca 2ϩ ] in ␤-cells.
ER Ca 2ϩ -ATPase plays an important role in regulating cytosol [Ca 2ϩ ]. It sequestrates cytosolic Ca 2ϩ into endoplasmic reticulum in an ATP-dependent manner (44 -46) therefore lowering [Ca 2ϩ ] i . Thapsigargin is a specific inhibitor of the ER Ca 2ϩ -ATPase (31,47). Addition of thapsigargin to the ␤-cells prevented ER Ca 2ϩ uptake, therefore leading to elevated cytosol free [Ca 2ϩ ] (32). As shown in Fig. 4C, addition of 200 nM thapsigargin to the control ␤-cell ␤TC6-F7 led to an increase in [Ca 2ϩ ] i : from the basal level of 107 Ϯ 7 nM to 184 Ϯ 11 nM (p Ͻ 0.05) (Fig. 5, C and D). In the ␤IRS1-A cells, however, thapsigargin had no effect on cytosol free [Ca 2ϩ ]: 223 Ϯ 35 nM (basal) versus 194 Ϯ 36 nM (200 nM thapsigargin). Because addition of thapsigargin raised [Ca 2ϩ ] i to the same level as that caused by overexpression of IRS-1, these data indicated that altered ER Ca 2ϩ uptake may be the major cause of elevated [Ca 2ϩ ] i in IRS-1-overproducing cells.
Inhibition of Endoplasmic Reticulum Ca 2ϩ Uptake-To de-termine whether endoplasmic reticulum Ca 2ϩ uptake is affected by IRS-1, we directly measured ER Ca 2ϩ uptake with permeabilized ␤-cells. Calcium uptake by endoplasmic reticulum can be directly measured with radioactive 45 Ca 2ϩ . We assayed ER Ca 2ϩ uptake at two Ca 2ϩ concentrations: 100 nM (basal condition) and 500 nM (equivalent to glucose-stimulated ␤-cells). As shown in Fig. 6, Ca 2ϩ uptake in ␤IRS1-A was significantly reduced at both Ca 2ϩ concentrations compared with control. At 100 nM [Ca 2ϩ ], ER Ca 2ϩ uptake for the control and ␤IRS1-A cell was 4.28 Ϯ 0.89 and 2.77 Ϯ 0.22 nmol/mg of protein, respectively (a reduction of 35%) (p Ͻ 0.05). At 500 nM [Ca 2ϩ ], the uptake was 11.42 Ϯ 1.65 (control) and 7.15 Ϯ 0.90 (␤IRS1-A) nmol/mg of protein (p ϭ 0.01), a reduction of 37%. Similar ER Ca 2ϩ uptake results were also obtained from two additional IRS-1-overproducing clones (data not shown). These data clearly demonstrated that overexpression of IRS-1 reduced endoplasmic reticulum Ca 2ϩ uptake and therefore lowered ER Ca 2ϩ content. Inhibition of ␤-Cell Proliferation-The insulin receptor signaling pathway is also implicated in mitogenic regulation (49,50). To determine how overexpression of IRS-1 affects ␤-cells growth, we used the [ 3 H]thymidine incorporation assay (51) to assess ␤-cell proliferation (Fig. 7). ␤IRS1-A cells exhibited a   FIG. 3. Glucose responsiveness of the ␤-cells. A, MTT assay; B, cellular ATP content. Cellular ATP content was measured by HPLC as described under "Materials and Methods" and elsewhere (38). Open bars, 0 mM glucose; filled bars, 15 mM glucose. C, glucose-dependent insulin secretion. Insulin secretion at different glucose concentrations were normalized with the maximal insulin secretion at 30 mM glucose and expressed as a percentage of the maximum. Glucose response curves from the two cell lines were not significantly different (one-way ANOVA test, p Ͼ 0.28). All data represent means Ϯ S.E. from at least two independent experiments, each performed in triplicate.

FIG. 4. Insulin biosynthesis.
A, glucose-stimulated insulin biosynthesis. Insulin biosynthesis was measured with [ 3 H]leucine pulse-chase assay. Insulin-specific radioactivity was immunoprecipitated with anti-insulin antibody (15). For each cell line, the rate of insulin-specific [ 3 H]leucine incorporation at 0 mM glucose was set as basal level. With 16 mM glucose stimulation, the rate of insulin biosynthesis increased 2-fold in both cell lines. B, fractional insulin biosynthesis (% of trichloroacetic acid). Insulin-specific 3 H-counts were divided by total trichloroacetic acid-precipitable 3 H-counts under each condition. Open bars 0 mM glucose; solid bars, 16 mM glucose. Data are presented as mean Ϯ S.E., n ϭ 7. *, p Ͻ 0.001. 32 Ϯ 4% (n ϭ 6) decrease in the rate of cell proliferation as measured by [ 3 H]thymidine incorporation (6,842 Ϯ 513 dpm/g DNA) compared with the passage-matched ␤TC6-F7 control cells (10,114 Ϯ 890 dpm/g DNA) (p ϭ 0.02) in the presence of 10% serum. In the absence of serum, the growth rate of ␤IRS1-A was reduced 40% compared with the control (1,895 Ϯ 167 dpm/g DNA, ␤IRS1-A versus 3,183 Ϯ 274 dpm/g DNA, control; n ϭ 6) (p ϭ 0.007). No increase in ␤-cell apoptosis was observed when measured with terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling assay (52) (data not shown). These data demonstrated that IRS-1 negatively regulates ␤-cell proliferation. DISCUSSION Earlier studies had suggested that the insulin receptor may be present in ␤-cells and that it could function to regulate insulin secretion. For many years, however, this concept was viewed as controversial in the absence of definitive evidence identifying the insulin receptor in the ␤-cells. Our group has recently demonstrated that the various components of the insulin receptor signaling pathway are indeed present in ␤-cells, including the insulin receptor and IRS-1 (12,13). Furthermore, we had shown that glucose-induced insulin secretion activates the ␤-cell surface insulin receptor tyrosine kinase and its intracellular signal transduction pathway and had proposed that this represented a novel autocrine mechanism for the regulation of ␤-cell function (12). However, the physiological role of this pathway in the ␤-cell has been difficult to elucidate because the ␤-cell is not a classical insulin target tissue, and there is scant evidence that insulin regulates its own secretion. Very recently, it has been shown that one role of the insulin receptor signaling pathway in ␤-cells is regulation of ␤-cell growth because disruption of IRS-2 leads to ␤-cell deficiency at birth and diabetes, and it has been proposed that IRS-2-dependent signaling pathways are involved in ␤-cell neogenesis, proliferation, and survival (17). In contrast, mice heterozygous for null alleles of the insulin receptor and IRS-1 become diabetic and develop ␤-cell hyperplasia (21). Other studies have also shown that insulin receptor signaling in the ␤-cell can regulate insulin gene transcription, as well as autoregulation of protein synthesis via PHAS-1 phosphorylation (14,16). Thus, the insulin receptor signaling pathway of the ␤-cell appears to have multiple physiological effects.
To identify the role of IRS-1 in insulin secreting ␤-cells, we overexpressed IRS-1 in a clonal ␤-cell line ␤TC6-F7. Our data demonstrates that IRS-1 is involved in regulating Ca 2ϩ homeostasis, insulin secretion, insulin biosynthesis, and ␤-cell proliferation and that elevated expression of IRS-1 induces ␤-cell failure. This is the first study to demonstrate that a 2-fold overexpression of IRS-1 in ␤-cells increases ␤-cell cytosol Ca 2ϩ levels and reduces ER Ca 2ϩ content. These findings are significant because the ER is one of the major compartments for intracellular Ca 2ϩ storage. Elegant earlier experiments have shown that the ER is actively involved in regulating intracellular Ca 2ϩ in the nanomolar range (27,28). In ␤-cells, it is widely established that glucose stimulation results in an increase in intracellular Ca 2ϩ , a required step for insulin secretion. Typically, basal Ca 2ϩ concentrations are in the 80 -100 nM range, and following glucose stimulation, they increase 3-5fold. Once the stimulation is removed, the ER sequesters excess cytosolic Ca 2ϩ , and the Ca 2ϩ levels return to baseline. The ER Ca 2ϩ -ATPase is the main pump responsible for Ca 2ϩ uptake into the ER. Its biochemical characteristics have been extensively described, and it is implicated in the regulation of intracellular Ca 2ϩ homeostasis. Recently, two isoforms of the Ca 2ϩ -ATPase, SERCA2 and SERCA3, have been localized to the islet. Furthermore, SERCA3 expression is reduced in the GK rat, a nonobese model of type 2 diabetes (44 -46).
Our observations that thapsigargin, an ER Ca 2ϩ -ATPasespecific inhibitor, increased cytosolic Ca 2ϩ levels in the control cells, but not in the IRS-1-overproducing cells (Fig. 5, C and D) suggest that ER Ca 2ϩ -ATPase in the ␤IRS1-A cells could have been suppressed by IRS-1 overproduction. This is also strongly supported by the observed decrease in ␤IRS1-A ER Ca 2ϩ uptake (an indirect measurement of Ca 2ϩ -ATPase) and the fact that Ca 2ϩ release from the ER (induced by A23187, data not shown) is not changed in ␤IRS1-A cells. A possible explanation for IRS-1-induced inhibition of ER Ca 2ϩ -ATPase is based on an elegant study by Kahn and co-workers (33), who have identified SERCA1 in skeletal muscle and SERCA2 in cardiac muscle as novel IRS-1-and IRS-2-binding proteins. Importantly, this interaction is insulin-dependent (maximal at 100 nM insulin and at 5 min of stimulation), and requires tyrosine phosphorylation of IRS-1. Whether there is a similar interaction between IRS-1 and the ␤-cell ER Ca 2ϩ -ATPase still remains to be determined, although our data suggest that such an interaction could result in decreased ER Ca 2ϩ -ATPase activity, leading to reduced ER Ca 2ϩ uptake in ␤-cells and increased cytosolic Ca 2ϩ concentrations.
Our study provides a novel functional link between the IRS-1 signaling pathway and the stimulus-secretion pathway in ␤-cells, which we believe to be physiologically significant. We postulate that under basal conditions in the ␤-cell this pathway is not activated. However, once glucose or other secretagogues stimulate insulin secretion, the released insulin will feed back to the ␤-cell insulin receptor and activates the associated signal transduction pathway. Increased signaling results in IRS-1 tyrosine phosphorylation (12) and subsequent inhibition of ER Ca 2ϩ uptake as shown by this study. Decreased Ca 2ϩ fluxes into the ER can then increase cytosolic Ca 2ϩ and further facilitate the maintenance of increased Ca 2ϩ levels due to secretagogue-induced Ca 2ϩ influx from the extracellular space. We believe that our studies have identified a novel pathway of autocrine regulation of intracellular Ca 2ϩ homeostasis and insulin secretion in the ␤-cell of the endocrine pancreas.
Insulin secretion and cellular insulin content were significantly reduced in the cells overproducing IRS-1. This is most likely the consequence of reduced insulin biosynthesis. Elevated expression of IRS-1 inhibits insulin biosynthesis at the translational level as measured by the [ 3 H]leucine labeling assay. This inhibition is likely due to the reduced ER [Ca 2ϩ ] found in the IRS-1-overproducing cells. A similar finding that ER Ca 2ϩ affects insulin biosynthesis has been reported before (25,26). The insulin mRNA level in the IRS-1-overexpressing cells was similar to the controls as measured with RNA protection assay. It suggests that IRS-1 may not be involved in mediating activation of insulin gene transcription in ␤IRS1-A cells. A recent report suggests that IRS-2 may mediate signals for insulin gene transcription (16).
Augmented expression of IRS-1 also leads to inhibition of ␤-cell growth. ␤IRS1-A cells exhibited a 32 Ϯ 4% decrease in the rate of cell proliferation as measured by [ 3 H]thymidine incorporation compared with the control. These observations are supported by the finding that loss of IRS-1 function leads to ␤-cell hyperplasia and hyperinsulinemia in the IRS-1 knockout animals (21,48). ␤-Cell function is also regulated by the insulin receptor substrate IRS-2. Loss of IRS-2 function results in ␤-cell failure, including reduced ␤-cell growth and decreased insulin secretion. Deletion of IRS-2 leads to development of diabetes in transgenic mice (4,17). It appears that insulin receptor signaling differentially regulates ␤-cell function via different substrates. The net output of insulin receptor signaling in ␤-cells may therefore depend on the relative strength of different substrate signals. Abnormalities in the insulin receptor substrates, e.g. altered IRSs expression levels, may directly contribute to ␤-cell failure in diabetes.