Elimination of Negative Feedback Control Mechanisms Along the Insulin Signaling Pathway Improves β-Cell Function Under Stress

Supplementary material is also available in the online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0890/DC1.

Male C57BL/6J (age 9–10 weeks) mice were housed under standard light/dark conditions and were given access to food and water ad libitum. Experiments were approved by the Animal Care and Use Committee of the Weizmann institute of Science.

Chinese hamster ovary (CHO) cells that overexpress the insulin receptor (CHO-T cells) were grown in F-12 medium. Min6 insulin-secreting cells (26) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mmol/l glutamine. Both cell lines were supplemented with 10% FBS, 100 units/ml penicillin, and 100 μg/ml streptomycin.

Cytokine mixture (1X Cytomix) consisted of 3 nmol/l TNF-α, 3 nmol/l INF-γ, 1.5 nmol/l IL-1β, and FasL (1.25% v/v). Their biological activities were 10 units/ng (TNF-α, INF-γ) and 200 units/ng (IL-1β).

Murine islets were isolated as described (27). In brief, digested pancreata were filtered through 1,000-μm and 500-μm sieves, and islets >75 and <250 μm were handpicked under a stereoscope. Islets were cultured in suspension in RPMI 1,640 medium, 5 mmol/l glucose, 10% fetal calf serum (FCS), 50 units/ml penicillin, 50 μg/ml streptomycin, and 40 μg/ml Gentamicin. For Western blot analysis, islets (50–60 islets per treatment) were cultured in 35 mm Extracellular Matrix Protein (ECM) coated plates (Novamed, Jerusalem).

Myc-IRS-2^WT and Myc-IRS-2^5A were generated as we described (24). Site-directed mutagenesis of S303, 343, 362, 381, and S480 of IRS-2 into Ala was performed using a Quick-Change site-directed mutagenesis kit (Stratagene) according to manufacturer's instructions. To generate stable clones, CHO-T cells were transfected with the above constructs, together with pBabe-Puro, encoding puromycin resistance. After 24 h, the transfected cells were subjected to selection with 10 μg/ml puromycin.

Adenoviral constructs were generated as we described (24) according to the protocol provided by AdEasy vector system (Quantum). Briefly, cDNA encoding Myc-IRS-2 (WT or 5A) was ligated into the shuttle plasmid pAdTrack-CMV, which contains a green fluorescent protein (GFP) cassette driven by an independent promoter that serves as a tracing marker. The recombinant pAdTrack-CMV was cotransformed with pAdEasy-1 containing the adenovirus genome, into Escherichia coli strain BJ5183, where homologous recombination took place. Positive colonies were identified by restriction analysis. The recombinant pAdEasy-1-CMV-IRS-2 plasmids (WT or 5A) were transfected into HEK293 cells and viruses were amplified. Viruses were stored at −80°C at a viral titer of ∼10¹⁰ PFU/ml.

Murine islets were infected 24 h after isolation with adenoviral constructs (MOI 600) for the indicated times. Min6 cells were infected at MOI of 200 for 1.5 h in serum-free medium. Treatments were applied up to 72 h after infection.

CHO-T cells or murine islets were washed and harvested in buffer A (25 mmol/l Tris-HCl [pH 7.4], 10 mmol/l sodium orthovanadate, 10 mmol/l sodium pyrophosphate, 100 mmol/l sodium fluoride, 10 mmol/l EDTA, 10 mmol/l EGTA, and 1 mmol/l phenylmethylsulfonyl fluoride). Supernatants (12,000 g) of cell extracts (50–150 μg CHO-T cells; 15–30 μg murine islets) were resolved by SDS-PAGE and Western blotted with the indicated antibodies.

Approximately 100 islets embedded in 1% agarose gel were fixed for 16 h in 4% paraformaldehyde and were then transferred to PBS until being embedded in paraffin. Graft-bearing kidneys were formalin-fixed, and serial sections (5 μm each) were immunostained with the indicated antibodies as described (28).

Apoptosis of Min6 cells (25,000 cells per well) and mouse islets (10 islets per well) was determined by Caspase-3/7 activity kit (Enzolyte-Caspase-3–RH110, AnaSpec Ltd.) according to the manufacturer instructions using fluorescent microplate reader Ex/Em = 496 nm/520 nm.

Islets were isolated and infected with adenoviral constructs as indicated. Groups of five islets were incubated for 1 h in Krebs-Ringer bicarbonate HEPES buffer (KRBH) at 37°C with 2.5 mmol/l glucose followed by incubation for 1 h in KRBH with 2.5 mmol/l or 20 mmol/l glucose. Insulin concentration in the culture mediums was determined using Mercodia mouse insulin ELISA kit.

Mice were made diabetic by an intraperitoneal injection of streptozotocin (STZ; 175 mg/kg) and were transplanted 5 days later when their fasting blood glucose levels were >400 mg/dl. Islets used for transplantation were isolated from naive mice. Islets either remained uninfected or were infected with adenoviral constructs expressing IRS2^WT or IRS2^5A (and GFP). Twenty-four h thereafter, 200 islets were washed, mounted on a 0.2-ml tip, and released into the renal subcapsular space of the kidneys of recipient mice through a puncture in the capsule, which was immediately sealed with 1-mm³ absorbable gelatin sponge (Surgifoam, Ethicon; Somerville, NJ).

Mice were fasted overnight with water access ad libitum. Mice were then injected intraperitoneally with glucose (1 mg/g body wt). Glucose levels were monitored using MediSense Optium Blood Glucose test strips (Abbott Laboratories, IL) on blood drawn at timed intervals from a tail vein.

Islet RNA was extracted using the PerfectPure RNA kit (5Prime, MD), and first-strand cDNA was generated by cDNA Reverse Transcription kit (Applied Biosystems, CA). Quantitative detection of specific mRNA transcripts was carried out by real-time PCR using ABI-PRISM 7900HT instrument (Applied Biosystems, CA). Data were normalized for the content of actin mRNA.

Data are presented as mean ± SEM. Data were analyzed by Student t test within a three-way ANOVA [factors were cytokine, treatment, and date of experiment (random factor)] at a minimum P < 0.05 threshold.

Mutations to Ala of inhibitory Ser sites of IRS-1 that conform to consensus PKC phosphorylation sites RXXS/TΨ (Ψ, mostly hydrophobic), located proximal to the PTB domain of IRS-1, improve insulin signaling in rat hepatoma Fao cells (24,25). To study whether similar mutations improve the function of IRS-2, five potential inhibitory serines (S303, S343, S362, S381, and S480) were mutated to Ala. To determine whether these sites are subjected to phosphorylation, tandem mass spectrometric analysis (MS/MS) was carried out. We found that S303 and S343 are subjected to phosphorylation in response to insulin treatment of Min6 β-cell line (not shown). Phosphorylation of the other three sites (S362, S381, S480) could not be determined due to low yields of peptides containing these residues.

The effects of mutation of inhibitory Ser sites on IRS-2 function were initially studied in CHO cells that coexpress the insulin receptor and either IRS-2^WT or IRS-2^5A. Both IRS-2^WT and IRS-2^5A underwent rapid insulin-induced Tyr-phosphorylation (Fig. 1,A, B), indicating that the 5A mutations did not affect the conformation of IRS-2^5A, leaving it as equipotent substrate to the insulin receptor. Still, IRS-2^5A better maintained its Tyr phosphorylated active conformation after 60 min of insulin treatment, compared with IRS-2^WT. These results suggest that mutation of potential inhibitory Ser sites protects IRS-2^5A from the action of insulin-stimulated IRS-2 Ser/Thr-kinases that otherwise induce the dissociation of IRS-2 from insulin receptor (22). Indeed, IRS-2^5A maintained stronger coupling with the insulin receptor, as evidenced by the higher amounts of insulin receptor associated with IRS-2^5A immunoprecipitates, compared with IRS-2^WT (Fig. 1 B).

The phosphorylation of downstream effectors of IRS-2 was studied as well. We found that, in response to insulin, p85α, the regulatory subunit of PI3K, better couples to IRS-2^5A compared with IRS-2^WT (Fig. 1,B). Accordingly, protein kinase B (PKB) was activated to a greater extent (Fig. 1,C), while ERK1/2 and S6K1 were also better activated, though to a lesser extent (Fig. 1 D,E).

Next, Min6 cells and murine islets were infected with adenoviral constructs expressing Myc-IRS-2^WT or Myc-IRS-2^5A. Expression of the IRS-2 proteins in Min6 cells results in ∼1.5–2.0-fold increase in the protein level compared with the endogenous IRS-2 protein (Fig. 2,A). Maximal expression of the IRS-2 proteins occurred 48 h after infection (Fig. 2,B and D); therefore, this time period was selected for all subsequent experiments. Calibration experiments were carried out throughout this study to ensure that IRS-2^5A and IRS-2^WT are expressed to a similar level in Min6 cells or isolated islets (Figs. 2 C–E).

To assess the yield of infected β-cells, we took advantage of the fact that the adenoviral constructs also express GFP driven by an independent promoter. As shown in Fig. 2,F, a significant fraction of β-cells, mainly those localized to the periphery of the islets, were stained for GFP. Approximately 80% of the total cells within representative islets sections were insulin-containing β-cells (Fig. 2 G). Of these, ∼50% were adenovirus-infected cells (GFP positive). The relatively high infection yield enabled us to detect the effects of the IRS-2 proteins at the level of whole islets. Still, the effects of both IRS-2^WT and IRS-2^5A were underestimated because not all β-cells incorporated these constructs.

To study the ability of IRS-2^5A to confer resistance from apoptosis induced by proinflammatory cytokines (29,30), Min6 cells were incubated for 24 h with a “Cytomix (1X)” composed of 3 nmol/l TNFα, 3 nmol/l INF-γ, 1.5 nmol/l IL-1β, and FasL (1.25% v/v). This treatment induced activation of Caspase-3 (Fig. 3,A). Introduction of either IRS-2^WT or IRS-2^5A into Min6 cells reduced apoptosis to a similar extent (∼30–40%). However, when the experiments were carried out in cultured murine islets, the differences between IRS-2^WT and IRS-2^5A became apparent (Fig. 3,B). Whereas IRS-2^WT failed to exert any protective effects, IRS-2^5A reduced ∼40% cytokine-induced apoptosis, compared with islets infected with an empty vector control. Somewhat lower protection (25%) was obtained in islets treated with cytokines without FasL (Fig. 3,C). Total Caspase activity in cytokine-treated noninfected (control) islets was ∼2-fold higher in the presence of FasL (Fig. 3,B versus 3C). This indicates that FasL is an important constituent in the induction of apoptosis. Also, cytokine-induced FasL secretion cannot be ruled out. Of note, infection per se of murine islets with adenoviral constructs might impede islet function to a certain extent (31). Therefore, Caspase-3 activity is higher in islets infected with an empty virus, compared with noninfected cells (Figs. 3 B and C). Still, in the absence of cytokines, islets infected with IRS-2^5A do not differ significantly from islets infected with an empty vector. Hence, IRS-2^5A does not provide protection from the adverse effects of infection.

The effects of IRS-2^WT versus IRS-2^5A on glucose-stimulated insulin secretion (GSIS) in isolated islets were evaluated next. Consistent with previous findings (32), overexpression of IRS-2^WT increased GSIS 1.5-fold in cultured islets, when compared with islets infected with an empty virus (Fig. 4). The effects of IRS-2^5A were ∼2.8-fold greater than those of IRS-2^WT. The stronger effects of IRS-2^5A were restricted to GSIS, as IRS-2^5A was equipotent to IRS-2^WT in elevating by ∼50% the glucose-independent insulin secretion, induced by a combination of l-arginine and KCl (Fig. 5). The decrease in GSIS in islets infected with empty adenoviral constructs (compared with noninfected cells) could be attributed to the adverse effects of infection per se on islet function (31).

As previously shown (33) and confirmed in Fig. 4, incubation of murine islets with a combination of cytokines inhibited (∼55%) islets function assessed by GSIS (None). Expression of IRS-2^5A in cytokine-treated islets improved GSIS ∼4-fold, (compared with empty vector, Fig. 4), while introduction of IRS-2^WT improved GSIS only 2.6-fold. In fact, GSIS in cytokine-treated islets expressing IRS-2^5A was 1.6-fold higher than GSIS in naïve islets. The effects of IRS-2^5A could not be attributed to alterations in cellular insulin content (not shown). These results support our hypothesis that IRS-2^5A is less prone to Ser phosphorylation by cytokine-activated IRS kinases than is IRS-2^WT and, as a result, can better protect pancreatic islets from their adverse effects.

IRS proteins are upstream activators of Akt that promote activation of Pdx1, a transcription factor that mediates GSIS (34,35). Therefore, we examined whether IRS-2^5A stimulates GSIS better than IRS-2^WT because of its effects on Pdx1. As shown in Fig. 6,A, expression of IRS-2^WT or IRS-2^5A in naïve islets increased Pdx1 expression 2- and 3.8-fold, respectively. These effects were specific as the IRS proteins did not affect actin gene transcription (not shown). Preincubation of murine islets with cytokines inhibited Pdx1 (but not actin) transcription in a time-dependent manner, reaching 80% inhibition by 24 h (Fig. 6,B). Still, IRS-2^5A, but not IRS-2^WT, maintained its potency to promote ∼2-fold Pdx1 transcription even in cytokine-treated islets (Fig. 6 C). These results suggest that IRS-2^5A could mediate its beneficial effects through promoting the expression of Pdx1.

To determine whether ex vivo introduction of IRS-2^5A into pancreatic islets prior to their transplantation improves the functionality of the transplanted β-cells, C57BL/6J mice were rendered diabetic by a single high-dose injection of STZ. Mice were transplanted five days later when their fasting blood glucose levels exceeded 400 mg/dl. Islets used for transplantation were isolated from healthy syngeneic mice and were infected with adenoviruses harboring either IRS2^WT, IRS2^5A, or empty GFP control. As shown in Fig. 7,A, the transplanted islets remained as a packed mass under the kidney capsule for 20 days after transplantation. Most transplanted cells retained their insulin content (Fig. 7,A, middle) and expressed the adenoviral constructs, revealed by their GFP staining (Fig. 7,A, left). As expected, the grafted islets exhibited an overlap between GFP expression and insulin-positive cells (Fig. 7 A, right).

The effects of islet transplantation were assessed by measuring the fasting blood glucose levels of the recipient diabetic mice. Blood glucose levels of healthy naïve mice were lower than 120 mg/dl (Fig. 7 B), while those of the diabetic mice exceeded 400 mg/dl. Diabetic mice transplanted with a subtherapeutic mass of 200 islets expressing an empty/control adenoviral construct exhibited high fasting blood glucose levels, three days after transplantation, while mice transplanted with 200 islets expressing either IRS-2^WT or IRS-2^5A showed normal fasting blood glucose levels. These results suggest that islets overexpressing either IRS-2^WT or IRS-2^5A are equipotent in restoring fasting normoglycemia in STZ-treated diabetic mice.

The ability of the transplanted islets to restore normoglycemia in diabetic mice after glucose injection was next evaluated. As shown in Fig. 7,C, STZ-treated diabetic mice, transplanted with suboptimal doses of naïve islets or islets infected with GFP-control vector, exhibited a diabetic pattern of glucose tolerance (>400 mg/dl). Mice transplanted with islets infected with IRS-2^WT remained significantly hyperglycemic (∼350 mg/dl) even 90 min after glucose injection. In contrast, mice transplanted with islets infected with IRS-2^5A exhibited near-normal glucose clearance rate, with blood glucose levels reaching normal level at 90 min. To evaluate the beneficial effects of IRS-2^5A on a longer time scale, STZ-induced diabetic mice were transplanted with a higher number (300) of islets expressing the different IRS-2 constructs, or an empty control virus. As shown in Fig. 8, islets expressing IRS-2^5A performed much better during the first 2–4 days posttransplantation as measured by postprandial blood glucose levels than islets expressing IRS-2^WT or an empty virus (control). By 8–90 days, all transplanted islets were equipotent in restoring postprandial normoglycemia. Hence, the beneficial effects of IRS-2^5A were most pronounced at the time when the transplanted islets faced the least vasculature. These results suggest that IRS-2 plays an important role in insulin secretion and islet function at the early stages immediately after transplantation.

In the present study, we provide evidence that elimination of negative feedback-control mechanisms along the insulin signaling pathway improves β-cells function under stress. Such elimination is based upon introduction into β-cells of IRS-2 proteins harboring mutations at five potential inhibitory Ser sites, the phosphorylation of which negatively regulates IRS protein function. The mutated IRS-2 proteins (IRS-2^5A) are less prone than native IRS-2 proteins to phosphorylation by IRS kinases activated by proinflammatory cytokines and other stress-inducers and can, therefore, better couple with their upstream activators (the insulin and IGF-1 receptors) and their downstream effectors. As a result, β-cells expressing IRS-2^5A can resist more efficiently the action of proinflammatory cytokines. This is evident in cultured islets that express IRS-2^5A, which secrete more insulin even when challenged with proinflammatory cytokines. Moreover, glucose disposal was better in diabetic mice transplanted with islets expressing IRS-2^5A compared with islets expressing IRS-2^WT or empty vector.

Several lines of evidence support the above conclusions. First, we showed that at least two of the five Ser sites, S303 and S343, are subjected to in vivo phosphorylation in response to insulin in Min6 cells. Second, when expressed in CHO-T cells, IRS-2^5A better maintains its Tyr-phosphorylated active conformation compared with IRS-2^WT. This can be attributed to the ability of IRS-2^5A to efficiently couple with the insulin and IGF-1 receptors, which otherwise dissociate from the IRS proteins once they are subjected to Ser phosphorylation by IRS kinases activated after prolonged insulin treatment (23). As a consequence, IRS-2^5A activates to a greater extent its downstream targets PI3K, Akt, ERK, and S6K1. In this respect, IRS-2^5A resembles IRS-1^7A, an IRS-1 mutated at seven potential inhibitory Ser sites, which better couples to the insulin receptor and better propagates insulin signaling (24,25).

The superior ability of IRS-2^5A to prevent β-cell apoptosis, induced by a combination of cytokines, suggests that the Ser sites mutated in IRS-2^5A serve as targets for cytokine-activated IRS kinases. Indeed, IL-1β, TNFα, and IFNγ promote activation of PKCδ (36), IKK (37), JNK (38), p38 MAPK, and ERK (39), which phosphorylate IRS proteins at “inhibitory” Ser sites (21,40). Cytokines are also known inhibitors of glucose-stimulated insulin secretion (41). Indeed, IRS-2^5A, unlike IRS-2^WT, exerts a significant protection and could almost eliminate the inhibitory effects of cytokines on GSIS.

The capability of IRS-2^5A to promote GSIS better than IRS-2^WT could be attributed to its ability to promote transcription of Pdx1, a transcription factor which is activated once its repressor FOXO1 is inhibited through Akt phosphorylation and nuclear exclusion (42,43). PDX1 promotes the expression of pro-insulin (44), Glut2 (45), and glucokinase (46), all contributing to increased insulin secretion. Indeed, we could show the greater potency of IRS-2^5A to increase transcription of Pdx1 in naïve islets and in islets treated with cytokines that inhibit Pdx1 gene transcription (34). This could be attributed to the ability of IRS-2^5A to couple more efficiently with p85α and activate Akt/PKB. These findings are consistent with observations that disruption of Pdx1 in murine β-cells reduces insulin secretion and causes progressive β-cell loss (34).

The effects of IRS-2^5A on GSIS are specific, because it does not fare much better than IRS-2^WT when insulin secretion is triggered independent of glucose stimulation (by membrane depolarization induced by a combination of KCl and arginine). These results suggest that IRS-2^5A promotes insulin secretion mainly through its effects on PDX1. Indeed, PDX1 does not regulate glucose-independent insulin secretion, and the magnitude of insulin response to arginine stimulation is unchanged in Pdx-1^+/− mice (46). IRS-2^5A did not modify cytokine-induced iNOS and Fas mRNA expression (not shown), suggesting that it presumably acts downstream of these death inducers.

The ability of IRS-2^5A to improve islet functionality better than IRS-2^WT was also demonstrated in vivo. Because overexpression of IRS-2^WT improves islets functionality (32), a suboptimal mass of 200 islets was transplanted to enable comparison of the effects of IRS-2^5A versus IRS-2^WT. Interestingly, even with these suboptimal doses, transplanted islets expressing either IRS-2^WT or IRS-2^5A were capable of maintaining normal fasting blood glucose levels, suggesting that both IRS-2 proteins improve islets functionality when the challenge to the islets is moderate. The differences between IRS-2^WT and IRS-2^5A became apparent only when the transplanted islets were faced with a greater challenge in the form of a glucose load. Under these conditions, only mice transplanted with islets expressing IRS-2^5A behaved like normal mice three days after transplantation, while mice transplanted with islets expressing IRS-2^WT behaved almost like diabetic animals. When diabetic mice were transplanted with a higher number of islets, the beneficial effects of IRS-2^5A on postprandial blood glucose levels were most pronounced during the first 2–4 days after transplantation, a critical time when the transplanted islets were poorly vascularized. These results suggest that IRS-2^5A can better confer upon transplanted islets protection from the stressful environment they are facing immediately after transplantation.

Factors that negatively influence islet survival after transplantation include prolonged hypoxia during the revascularization process (47). The hypoxic conditions trigger stress-activated kinases to reduce β-cell survival and functionality. Being partially resistant to the action of these kinases, IRS-2^5A provides the transplanted islets with two advantages: it increases their intrinsic ability to perform GSIS by promoting the activity of Pdx1 and it protects them from external insults mediated by stress-activated kinases.

Different strategies were attempted to increase islet survival after transplantation [Reviewed in refs (48,49)], so far with limited success. Therefore, IRS-2^5A expression in β-cells ex vivo could be considered as a new mode for promoting β-cell survival. This strategy might be even more effective in human islets, where the β-cells localize to the islet periphery and, therefore, are more prone to gene transfer by viral infection. The ability of IRS-2^5A to confer upon β-cells protection from immunosuppressive drugs that otherwise induce insulin resistance is another aspect of relevance in this context. Still, the potential caveats of insulin signaling pathway being hyperactivated in cells expressing IRS-2^5A need to be addressed because sustained hyperactivation of IRS-2 might contribute to uncontrolled cell growth (50). Consequently, further studies are required before this technique becomes suitable for clinical implementation.

This work was supported by research grants from the Juvenile Diabetes Research Foundation International; The Israel Science Foundation (founded by the Israel Academy of Sciences and Humanities); d-CURE Israel; The Mitchel Kaplan Fund; and the MINERVA Foundation. Y.Z. is an incumbent of the Marte R. Gomez Professorial Chair.

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

D.G., R.I., G.S., M.A., and D.R. researched data. S.B.-H. and E.C.L. researched data and reviewed/edited the manuscript. R.I. researched data. Y.Z. wrote the manuscript.

We thank Michael Walker (Weizmann Institute, Rehovot, Israel) for his guidance in islet isolation; Hila Giladi (Hebrew University, Jerusalem) for her help with adenovirus generation; and Raya Eilam (Weizmann Institute, Rehovot, Israel) for her help with immunohistochemistry staining. We thank Sanford Sampson (Bar-Ilan University, Israel) for a critical review of this manuscript.