Nuclear Export of FoxO1 Is Associated with ERK Signaling in β-Cells Lacking Insulin Receptors*

The insulin/insulin-like growth factor (IGF) signaling pathway plays a critical role in the regulation of islet cell biology. However, the signaling pathway(s) utilized by insulin to directly modulate β-cells is unclear. To interrogate whether insulin exerts endocrine effects in regulating proteins in the insulin/IGF-1 signaling cascade in vivo in physiological states via the insulin receptor, we designed two experimental approaches: 1) glucose gavage and 2) hyperinsulinemic intravenous infusion, for studies in either β-cell specific insulin receptor knock-out (βIRKO) or control mice. Immunostaining of sections of pancreas (collected immediately after glucose gavage or insulin infusion) from controls showed significant increases in pAKT+, p-p70S6K+, and pERK+ β-cells and a significant decrease in % nuclear FoxO1+ β-cells compared with corresponding vehicle-treated groups. In contrast, in βIRKOs, we observed no significant changes in pAKT+ or p-p70S6K+ β-cells in either experiment; however, pERK+ β-cells were significantly increased, and an attenuated decrease in % nuclear FoxO1+ β cells was evident in response to glucose gavage or insulin infusion. Treatment of control and βIRKO β-cell lines with glucose or insulin showed significantly decreased % nuclear FoxO1+ β-cells suggesting direct effects. Furthermore, blocking MAPK signaling had virtually no effect on FoxO1 nuclear export in controls, in contrast to attenuated export in βIRKO β-cells. These data suggest insulin acts on β-cells in an endocrine manner in the normal situation; and that in β-cells lacking insulin receptors, insulin and glucose minimally activate the Akt pathway, while ERK phosphorylation and FoxO1 nuclear export occur independently of insulin signaling.

Multiple lines of evidence support the view that the pancreatic ␤-cell is an insulin/IGF-1 (insulin-like growth factor 1) 3 responsive tissue, and that insulin itself contributes to the regulation of insulin secretion, proliferation, and survival of pancreatic ␤-cells (1)(2)(3). For example, insulin and IGF-1 receptors and the major insulin receptor substrates (IRS-1 through IRS-4) are present and functional in mammalian ␤-cells (4 -7). Our laboratory has previously reported that mice with insulin receptor knock-out in ␤-cells (␤IRKO) exhibit defects in glucose-stimulated insulin secretion and age-dependent progressive glucose intolerance that resembles human T2D (8 -9). Poor cell cycle progression, nuclear restriction of FoxO1, and reduced expression of cell cycle proteins favoring growth arrest are key defects exhibited by islets derived from ␤IRKO mice (7)(8)(9). Although these observations are consistent with a role for insulin in the regulation of ␤-cell biology (reviewed in Ref.

3), it is still unclear how physiological changes in the levels of circulating (systemic) insulin directly impact downstream insulin signaling proteins in vivo.
This lack of clarity is, in part, related to the difficulty in interpreting data on potential independent effects of exogenous insulin versus endogenous insulin on ␤-cell biology in cultured ␤-cells and/or isolated islets due to continuous secretion of the hormone via the regulated and/or constitutive pathways. Furthermore, the lack of a suitable model precludes accurate estimates of local concentrations of insulin in close proximity to ␤-cells to assess its effects in the islet in a living organism. Whereas we and others have focused efforts to investigate the relevance of exogenous insulin on ␤-cell secretion in humans (10 -12) the immediate alterations that occur in signaling proteins in the ␤-cells in vivo continue to be poorly understood. To reverse this state of ignorance we designed two independent experiments to simulate physiological states, in order to specifically examine the signaling effects of exogenous versus endogenous insulin on ␤-cells in vivo in mice: 1) a glucose gavage to simulate a physiological example of glucose-induced endogenous insulin secretion, and 2) a hyperinsulinemic i.v. infusion to examine the endocrine effects of mildly elevated circulating insulin on ␤-cells. Both protocols were performed in control mice and compared with mice lacking insulin receptors in ␤-cells (␤IRKO) to dissect the role of the insulin receptor-mediated pathway in the regulation of downstream proteins. We report that, in the models we used, insulin acts in an endocrine manner to modulate ␤-cell signaling proteins in vivo in the normal state. Our results also indicate that, in ␤-cells lacking insulin receptors, insulin and glucose minimally activate the Akt pathway in ␤-cells, while ERK phosphorylation and nuclear export of FoxO1 occur independent of insulin signaling.

Effects of Glucose-stimulated Insulin Secretion on Signaling
Proteins in ␤-Cells in Vivo-We first investigated the effects of a physiological stimulus by a glucose gavage (Fig. 1A). Fasting glucose and insulin concentrations at time 0, as well as glycemia and insulin levels achieved 15 min after the gavages were comparable in control and ␤IRKO mice (Fig. 1, B and C). As expected, glucose levels were significantly increased in the glucose gavage groups compared with respective controls receiving saline gavage ( p Ͻ 0.01 for both groups versus respective controls; Fig. 1B). Similarly, circulating insulin levels were elevated, and in the upper physiological range (13,14), after the mice received glucose gavage compared with those receiving saline in both control and ␤IRKO mice ( p Ͻ 0.05 for both groups versus respective control; Fig. 1C).
To examine the effects of secreted insulin on the insulin signaling pathway in ␤-cells, we used immunohistochemistry to analyze pancreas sections collected immediately after the end of the glucose or saline gavage. The % of ␤-cells that were positive for pAKT in the basal state were significantly lower in the ␤IRKOs compared with controls ( p Ͻ 0.01, Fig. 2, A and B). We observed an increased % of ␤-cells that co-immunostained for pAKT in control mice after glucose gavage compared with controls with a saline gavage ( p ϭ 0.03, Fig. 2, A and B); in contrast to lack of significant changes in the ␤IRKOs ( p ϭ 0.30, Fig. 2, A  and B). Similarly, there was an increased % of ␤-cells that co-immunostained for p-p70S6K in control mice that underwent glucose gavage compared with controls ( p ϭ 0.04, Fig. 2, C and D), while no changes were detected in % of ␤-cells that co-immunostained for p-p70S6K in ␤IRKO mice ( p ϭ 0.364, Fig. 2, C and D). Interestingly, the % of ␤-cells that co-immunostained for pERK was enhanced in both control and ␤IRKO mice that underwent glucose gavage compared with saline-gavaged controls ( p ϭ 0.04 and p ϭ 0.02 respectively, Fig. 3, A and B). Consistently, we observed increased nuclear export of FoxO1 in both control and ␤IRKO mice that underwent glucose gavage compared with their respective controls, although the magnitude of change was much lower in the ␤IRKOs ( p Ͻ 0.01 for both comparisons, Fig. 3, C and D).
Effects of Exogenous Insulin Infusion on Signaling Proteins in ␤-Cells in Vivo-We next explored the relevance of circulating insulin using a modified hyperinsulinemic clamp (Fig. 4A). Fasting glucose and insulin concentrations, as well as glycemia and insulin levels achieved 15 min after saline or insulin infusion, were comparable in the two groups (time 0, Fig. 4, B and C). As expected, glucose levels were decreased in insulin infusion groups compared with those undergoing saline infusions (time 15, p Ͻ 0.05 for both groups versus respective controls; Fig. 4B). Furthermore, circulating insulin levels were significantly raised after insulin infusion compared with saline infusion in both control and ␤IRKO groups ( p Ͻ 0.01 for both groups versus respective control; Fig. 4C).
To examine whether exogenous insulin modulates the insulin signaling pathway of ␤-cells in an endocrine manner, we undertook immunohistochemistry of sections of pancreas harvested immediately after the end of the saline or insulin infusions. Similar to our observations in the gavage studies the % ␤-cells that were positive for pAKT in the basal state were significantly lower in the ␤IRKOs ( p Ͻ 0.001, Fig. 5, A and B). The number of ␤-cells that co-immunostained for pAKT in control mice that underwent insulin infusion was significantly higher compared with those undergoing saline infusion ( p ϭ 0.03, Fig.  5, A and B), while no significant differences were evident between the two treatments in the ␤IRKO group ( p ϭ 0.30, Fig.  5, A and B). Control mice that underwent an insulin infusion showed increased % of ␤-cells that co-immunostained for p-p70S6K ( p ϭ 0.04, Fig. 5, C and D), compared with a lack of significant change in ␤IRKOs ( p ϭ 0.364, Fig. 5, C and D). Both control and ␤IRKO groups exhibited an increase in ␤-cells that co-immunostained for pERK after the insulin infusion ( p ϭ 0.04 and p ϭ 0.02, respectively, Fig. 6, A and B), and the magnitude of change was somewhat greater in the ␤IRKOs.
The increase in nuclear export of FoxO1 was evident in both control and ␤IRKO mice that underwent insulin infusion compared with respective controls with the change in ␤IRKOs being smaller ( p Ͻ 0.01 for both comparisons, Fig. 6, C and D) and was similar to that seen in the gavage studies.
Effects of Exogenous Glucose versus Exogenous Insulin Stimulation on Nuclear versus Cytoplasmic Localization of FoxO1 in ␤-Cells in Vitro-To examine the independent effects of hyperglycemia versus insulin signaling directly on the localization of FoxO1, we stimulated control or ␤IRKO ␤-cell lines either with glucose or exogenous insulin. After starvation, ␤IRKO cells demonstrated a trend toward reduced proportion of nuclear FIGURE 1. Experimental design and glucose and insulin levels during gavage studies. A, schematic of Experiment 1. Control mice (n ϭ 6) or ␤IRKO mice (n ϭ 8) underwent saline or glucose gavage as described in "Experimental Procedures." Blood samples were collected at baseline and 15 min after the gavage. Pancreas was harvested and prepared for immunohistochemical analysis. B, glucose levels following saline gavage (red and black) and glucose gavage (blue and green) in controls (square) or ␤IRKOs (triangle). C, circulating insulin levels at baseline and following saline (red and black symbols) or glucose gavage (blue and green) in controls (square) or ␤IRKOs (triangle). *, p Ͻ 0.05; **, p Ͻ 0.01 compared with respective controls.
FoxO1 compared with control cells in the basal unstimulated condition (vehicle, Fig. 7, A and B). This difference reached statistical significance between groups in the presence of the PI3K inhibitor LY294002 ( p ϭ 0.03, LY294002, Fig. 7, A and B). Stimulation with 22.2 mM glucose for 30 min led to FoxO1 exclusion from the nucleus in ␤IRKO cells, but showed little effect in control cells ( p Ͻ 0.01, High Glucose in Fig. 7, A and B). In contrast, and as expected, insulin (10 nM) stimulation for 30 min exported FoxO1 from the nucleus into the cytosol in controls, but had virtually no effect in ␤-cells lacking insulin receptors (␤IRKO) ( p Ͻ 0.01, Insulin, Fig. 7, A and B). These results indicate that glucose stimulation is sufficient to exclude FoxO1 FIGURE 2. Alterations in pAKT and p-p70S6K in control or ␤IRKO mice following saline or glucose gavage. A, representative pictures of islets after saline or glucose gavage in control or ␤IRKO mice. Immunostaining for insulin (red) and pAKT (green) and DAPI (blue). 40ϫ objective, scale bar 100 m. 3ϫ magnified images are shown from the Merge panel. B, percentage of ␤-cells positive for p-AKT following saline (red bar) or glucose gavage (blue bar) in control or following saline (black bar) or glucose gavage (green bar) in ␤IRKO mice. C, representative pictures of islets after saline or glucose gavage in Control or ␤IRKO mice. Immunostaining for insulin (red), p-p70S6K (green), and DAPI (blue) immunostaining. 40ϫ objective, scale bar 100 m. 3ϫ magnified images are shown from the Merge panel. D, percentage of ␤-cells positive for p-p70S6K following following saline (red bar) or glucose gavage (blue bar) in control or following saline (black bar) or glucose gavage (green bar) in ␤IRKO mice. *, p Ͻ 0.05, compared with respective controls; **, p Ͻ 0.01 between saline control and saline ␤IRKO.

Insulin Alters Insulin/IGF1 Pathway in ␤-Cells
OCTOBER 7, 2016 • VOLUME 291 • NUMBER 41 from the nucleus in insulin receptor-deficient ␤-cells, while an insulin-mediated signal is necessary for a similar effect of glucose in controls. It is also possible that in this in vitro cell culture model the secreted insulin accumulated in the media acts via IGF-1 receptors to activate downstream signaling proteins that are common to the two ligands.
Effects of Inhibition of the ERK1/2 Pathway on Insulin-independent Export of FoxO1 in ␤-Cells in Vitro-To assess the involvement of insulin receptor-mediated signals in the regulation of FoxO1 localization in ␤IRKO cells, we next examined the effects of multiple inhibitors individually including: 1) the PI3K inhibitor, LY294002; 2) the insulin/IGF-1 receptor dual inhibitor, OSI-906; 3), the Akt1/2 inhibitor, MK-2206; or 4) the MEK1/2 inhibitor U0126, on nuclear export of FoxO1 induced by a combination of glucose and insulin stimulation. The combination of glucose (22.2 mM) and insulin (10 nM) stimulation significantly reduced the proportion of nuclear FoxO1 in both control and ␤IRKO ␤-cells (Fig. 8, A and B). While the % of cells with nuclear FoxO1 localization was elevated in both groups the change was lower in ␤IRKO ␤-cells as compared with control ␤-cells treated with the PI3K inhibitor (LY294002), the dual insulin/IGF-1 receptor inhibitor (OSI-906), or its downstream Akt inhibitor (MK-2206), respectively (Fig. 8, A and B, LY294002, OSI-906, MK-2206). However, blocking ERK1/2 signaling with a Mek1/2 inhibitor (U0126) had virtually no effect on FoxO1 nuclear export in control ␤-cells (compared with vehicle), in contrast to significantly attenuated FoxO1nuclear export in ␤IRKO cells ( p Ͻ 0.01, U0126, Fig. 8, A and  B). These data suggest that ERK-mediated FoxO1 nuclear export occurs partly independent of insulin signaling in ␤IRKO ␤-cells.
We confirmed the phosphorylation state of Akt and ERK in the presence of inhibitors described above. Stimulation of the cells with glucose or insulin promoted phosphorylation of Akt in both control and ␤IRKO cells, with the latter exhibiting reduced Akt phosphorylation both before and after stimulation ( Fig. 8C and supplemental Fig. S1). This Akt phosphorylation was blunted when the stimulation with both glucose and insulin was performed independently in the presence of the PI3K inhibitor, the dual insulin/IGF-1 receptor inhibitor, or the Akt inhibitor, but not the MEK1/2 inhibitor. ERK was activated in the basal state in ␤IRKO cells, while the increase in the ratio of ERK phosphorylation to total ERK level was smaller in ␤IRKO cells ( Fig. 8C and supplemental Fig. S1). The presence of the MEK1/2 inhibitor completely blocked ERK phosphorylation when the cells were stimulated with glucose and insulin in both control and ␤IRKO cells ( Fig. 8C and supplemental Fig. S1).
Potential Role for IGF-1 Receptor Signaling in ERK1/2-mediated FoxO1 Nuclear Export in ␤IRKO ␤-Cells-To investigate the roles of insulin receptor and IGF-1 receptor-mediated signaling in FoxO1 localization in control and ␤IRKO cells, we used the dual insulin/IGF-1 receptor inhibitor (OSI-906). The presence of the dual inhibitor led to nuclear accumulation of FoxO1 after stimulation with insulin or glucose in control cells (Fig. 9, A and B). Notably, FoxO1-nuclear export was significantly reduced by the blockade of both insulin and IGF-1 receptor in ␤IRKO cells in basal, insulin stimulated, or glucose-stimulated conditions (Fig. 9, A and B). These results suggest that in ␤-cells lacking insulin receptors the nuclear export of FoxO1 induced by glucose is secondary to IGF-1R-mediated signaling.
We validated that the inhibitor, OSI-906, completely blocked phosphorylation of insulin and IGF-1 receptor evoked by insulin or glucose stimulation in both control and ␤IRKO cells ( Fig.  9C and supplemental Fig. S2). Phosphorylation of Akt and ERK were also attenuated by the simultaneous inhibition of insulin and IGF-1 receptors in control cells. In ␤IRKO cells, treatment with OSI-906 significantly reduced phosphorylation levels of ERK in the basal state as well as stimulated conditions ( Fig. 9C and supplemental Fig. S2). Together, these data indicate that in addition to the effects of insulin receptor activation, IGF-1 receptor-mediated signaling, contributes to ERK phosphorylation and FoxO1 localization in ␤-cells.

Discussion
In this study, we designed in vivo experiments to directly investigate the ability of insulin to modulate proteins in the growth factor (insulin/IGF-1) signaling pathway in ␤-cells. We report that signaling proteins downstream of the insulin/IGF-1 receptors are activated by circulating insulin levels and support a potential endocrine role for the hormone in regulating ␤-cell biology. We also observed that ␤-cells in mice with conditional disruption of functional insulin receptors (e.g. ␤IRKO) exhibit minimal activation of AKT pathways in response to glucose or insulin stimulation, in contrast to ERK and FoxO1 proteins, which were phosphorylated independently of insulin receptor signaling.
The effects of glucose and insulin to regulate signaling proteins to regulate ␤-cell secretion and/or proliferation continues to be a significant area of research because of the technical difficulties in evaluating the direct effects of ambient insulin in a cell type that also secretes the hormone. While we and several other investigators have developed in vitro models (15,16,18,23,24) the interpretation of data from in vivo models is especially limited by the tedious timeline (hours) to isolate islets after an in vivo stimulation experiment that does not accurately reflect immediate changes occurring in protein expression in the ␤-cells. This is in contrast to signaling effects easily discernible in homogeneous tissues such as skeletal muscle, liver, or adipose each of which can be rapidly isolated (minutes) after injection of stimuli in vivo. Studies in cultured islets are also not entirely physiological because they do not maintain their normal anatomical blood supply. Using an in vivo glucose infusion model, Alonso et al. (17) recently reported that, surprisingly, the effects of glucose on ␤-cell proliferation do not require the insulin receptor. In contrast, Martinez et al. (18), also using a or ␤IRKO mice (n ϭ 8) underwent saline infusion or insulin infusion as described in "Experimental Procedures." Blood samples were collected at baseline and 15 min after infusion. Pancreas was harvested for immunohistochemical analysis. B, glucose levels following saline infusion (gray and white) and insulin infusion (yellow and orange) in control (square) or ␤IRKOs (triangle). C, circulating insulin levels at baseline and following saline infusion (gray and white) and insulin infusion (yellow and orange) in control (square) or ␤IRKOs (triangle). **, p Ͻ 0.01 and ***, p Ͻ 0.001 compared with respective controls. knockdown (80% decrease in insulin receptors) approach, reported that glucose effects on ␤-cells are mediated primarily through insulin acting upon its own receptor. Our own studies, using a ␤-cell line with a knock-out of the insulin receptor, clearly indicate that glucose effects on signaling proteins in ␤-cells require functional insulin receptors (15). In addition to the limitations of using isolated islets/cell lines discussed above an additional confounding factor contributing to the divergent conclusions of the different studies could be because Alonso and colleagues used ␤-cells with an incomplete (50%) knockdown of the insulin receptor in contrast to a "knock-out" model used in our studies and a higher KD (80%) in the studies by Martinez et al. To clarify these issues, we carefully designed experiments in this study to capture the alterations in signaling proteins in ␤-cells in response to changes in circulating levels of glucose and/or insulin in vivo by harvesting the pancreas immediately after ending the stimulus for immunohistochemical analyses.

Insulin Alters Insulin/IGF1 Pathway in ␤-Cells
To evaluate insulin effects in vivo, we used glucose gavage as a physiological stimulus to increase insulin secretion. After gavage, both groups of animals showed a significant increase in blood glucose levels (16.6 -19.4 mmol/liter) and circulating levels of insulin, albeit, elevated to a relatively physiological range (ϳ0.17 nmol/liter). In controls, the enhanced phosphorylation of Akt, ERK and p70S6K and nuclear exclusion of FoxO1 could be ascribed to the combined effects of elevated glucose and insulin. In the ␤IRKOs, the lack of phosphorylation of Akt and p70S6K suggests that the effects in the controls are solely due to glucose, and are dependent on insulin and not IGF-1 receptors. In the basal state, however, the increase in phosphorylation of p70S6K in the ␤IRKO islets may reflect altered mTORC1 activity due to changes in protein translation (19). On the other hand, a persistent phosphorylation of ERK, and a small but significant effect on nuclear exclusion of FoxO1 even in mice

Insulin Alters Insulin/IGF1 Pathway in ␤-Cells
OCTOBER 7, 2016 • VOLUME 291 • NUMBER 41 lacking insulin receptors in ␤-cells indicate that these effects of glucose in the ␤IRKOs occur independent of the insulin receptor and potentially secondary to IGF-1 receptor-mediated signaling. Collectively these data are consistent with our previous observations of a lack of increase in pAkt, while pERK is enhanced in ␤IRKO cell lines in response to glucose stimulation (15).
Next, the studies using i.v. infusion provided supra-physiological levels of insulin (2.06 nmol/liter) in the presence of relatively low blood glucose (3.33 mmol/liter), compared with the levels observed in the gavage study. While the overall response in signaling proteins was similar in the ␤-cells in the two experimental groups of animals, the magnitude of the increase in phosphorylation of ERK in the ␤IRKOs was greater compared with the corresponding response in the gavage study. A more striking observation was the near total nuclear exclusion of FoxO1 in the ␤IRKOs. Since glucose levels were much lower and insulin levels were higher in the infusion, compared with gavage studies, one interpretation of these data is that insulin modulates downstream effects via IGF-1 receptors, which are reported to be up-regulated in ␤IRKO cells (18).
Elevating insulin in vivo can also lead to secondary factors that, in turn, have the potential to impact islet cells. For example, nutrient, neural and/or hormonal activation of growth factor signaling and the post-transcriptional modification of a variety of genes, miRNA or via the microbiome and peptides acting on the CNS (reviewed in Refs. 20 -22) all have the potential to indirectly regulate ␤-cell biology. Therefore, we also used ␤-cell lines derived from control or ␤IRKO mice to investigate the direct effects of glucose versus insulin in the context of FoxO1 translocation (8). The absence of a significant effect of high glucose or treatment with the PI3K inhibitor (LY294002) on the localization of FoxO1 in control ␤-cells suggests that either the effects of glucose on excluding FoxO1 from the nucleus are difficult to appreciate and/or the effects of insulin are dominant in control ␤-cells. Additional evidence for direct effects of insulin on ␤-cells was evident from treating ␤-cell  lines with inhibitors. Thus, consistent with the in vivo experiments, stimulation with a combination of (glucoseϩinsulin) promoted nuclear exclusion of FoxO1 in both control and ␤IRKO ␤-cell lines. Furthermore, inhibition of PI3K, IR, and IGF1R, or Akt1/2 with specific inhibitors in independent experiments, minimized translocation of FoxO1 from the nucleus induced by the combined treatment of (glucoseϩinsulin) in both control and ␤IRKO ␤-cells. These data suggest that the effects of glucose are reversed by blocking PI3K and Akt, which are known to be involved in the regulation of FoxO1 (1, 17). The ability of OSI-906, which blocks both insulin and IGF-1 receptors, to reverse the effects of the combination of (glucoseϩinsulin) suggests that the effects of glucose on FoxO1 are dependent, in part, on insulin/IGF-1 receptors. Although glucose has been shown to activate Akt, in MIN6 and INS-1 cells, for which FoxO1 is a known substrate (17,(23)(24)(25), the ability of OSI-906 to reverse glucose-induced nuclear export of FoxO1 to near basal levels in ␤IRKO cells, suggests that the effects of glucose are linked to IGF-1 receptor signaling.
Our observation of potential effects of glucose regulating ERK pathways that directly promotes FoxO1 export to allow transcriptional effects in ␤-cells could explain, in part, the 15-fold activation of MAP kinase in response to physiological concentrations of glucose (3-12 mM) in INS-1 ␤-cells (23). It is also evident from our studies that this effect is independent from the insulin signaling/AKT pathway, and is observed when ␤-cells are stimulated either by a glucose gavage or when circulating insulin is enhanced in an infusion setting. Of note, blocking ERK1/2 signaling attenuated FoxO1-nuclear export in ␤IRKOs, suggesting that ERK phosphorylation and FoxO1 nuclear export occur independently of insulin signaling. This is in agreement with previous data (18,26) that the effects of glucose on ERK activity occur in the absence of insulin signaling. While these data indicate that FoxO1 proteins are major targets of insulin action, they also support the possibility that FoxO1 can be activated independent of insulin signaling. The inhibition of insulin receptor and IGF-1 receptor signaling with OSI-906 resulted in the dephosphorylation and nuclear retention of ERK in ␤IRKO cells. Further studies are warranted to dissect the mechanism(s) of glucose versus insulin/IGF-1 in the regulation of specific target proteins involved in proliferation and/or secretory function in ␤-cells in dynamic and contextspecific states.

Experimental Procedures
Animals-12-week-old ␤-cell specific insulin receptor knock-out (␤IRKO) mice (expressing Cre recombinase on the rat insulin promoter or RIP-Cre) and backcrossed on the C57Bl/6 background for at least 15 generations (8) were studied along with control (C57Bl/6) mice. Mice were housed in controlled temperature (23°C) conditions with 12-hour light/dark cycle in the Animal Care Facility at Joslin Diabetes Center, Boston. All studies conducted and protocols used were approved by the Institutional Animal Care and Use Committee of the Joslin Diabetes Center and in accordance with National Institute of Health guidelines.
Study Design and Experimental Procedures-To investigate how insulin regulates downstream signaling proteins in vivo, we designed two experimental approaches. First, we simulated the effects of a physiological stimulus by performing either a glucose (GG) (1 g/kg) or saline gavage (SG) separately in ␤-cell specific insulin receptor knock-out (␤IRKO) or control mice (Experiment 1, n ϭ 3-4 each group, total 4 groups). Second, to explore the potential endocrine effects of circulating insulin we performed either a 15-min saline infusion or an insulin infusion (150 mUI/kg priming followed by 15 pmol⅐kg Ϫ1 ⅐min Ϫ1 insulin infusion rate) (Experiment 2, n ϭ 3-4 each group, total 4 groups). Plasma glucose and plasma insulin levels were measured at time 0 min (fasting) and at 15 min (Figs. 1A and 3A). Blood samples (ϳ2 l) were collected from the tail vein (27).
Experiment 1-All mice were fasted overnight for 14 h followed by an oral saline or glucose (1 g/kg body wt.) gavage. Blood glucose was measured using an automatic glucometer (Glucometer Elite; Bayer) immediately before (time 0) and 15 min after the injection. Experiment 2-At least 4 days before infusion experiments mice were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg body wt.) and xylazine (10 mg/kg body wt.), and an indwelling catheter was inserted into the right internal jugular vein as previously described (28,29). On the day of the infusion experiment, a three-way connector was attached to the jugular vein catheter to deliver solutions intravenously (e.g. saline or insulin), and blood samples were obtained from the tail vein. The infusion experiments were performed in the awake state after an overnight fast in both controls and ␤IRKO mice. During the saline infusion, normal saline was infused for 15 min at a rate that matched the volume required during a hyperinsulinemic clamp. Consistent with a standard hyperinsulinemic euglycemic clamp experiment, a hyperinsulinemic i.v. infusion was conducted with a primed (150 mU/kg body wt) and continuous infusion of B28 Asp-insulin (Novolog TM , NovoNordisk, Bagsvaerd, Denmark) at a rate of 15 pmol⅐kg Ϫ1 ⅐min Ϫ1 to raise plasma insulin within the physiological range (ϳ200 pM) (30). Since the insulin infusion finished before a significant drop in glycemia, glucose was not infused during the infusion experiments.
Assays-Blood glucose was monitored using an automated glucose monitor (Glucometer Elite, Bayer), and plasma insulin levels were detected by ELISA (Insulin Mouse ELISA, Crystal Chem) at the Joslin Assay Core Facility.
Immunohistochemistry-At the end of each in vivo experiment, mice were anesthetized, and the pancreas was rapidly dissected (ϳ3 min), fixed in Z-fix, and then embedded in paraffin. Five-micrometer thin sections were stained using antibodies to p-AKT (Cell Signaling, cat. 37879); p-p70S6K (Cell Signaling, cat. 9204), p-ERK (Cell Signaling, cat. 4370), FoXO1 (Cell Signaling), or insulin (Cell Signaling, rabbit mono, cat. 3014) and appropriate secondary antibodies and counterstained with DAPI as the nuclear marker.
All islets in the pancreas sections were imaged at 200x (20ϫ objective) or 400ϫ (40ϫ objective) by a fluorescent microscope (Olympus BX-61; Olympus America, Melville, NY) equipped with a DP72 digital camera and the number of insulin-positive cells was manually counted in a blinded fashion by a single observer. Insulinϩ cells showing nuclear DAPI staining were scored as ␤-cells. Between 1,000 and 2,000 ␤-cell nuclei were counted for each pancreas section, and data were expressed as mean percentage of p-AKT, p-p70S6K, p-ERK, or pFoxO1ϩ ␤-cells for both control and ␤IRKO mice in each experiment.
Immunofluorescence-Cells were incubated on poly-L-lysine pre-coated cover slips during starvation and stimulation. Cells on coverslips were fixed with 1% formaldehyde and stained with FoxO1 antibody (Cell Signaling), followed by anti-rabbit IgG-Alexa594 (Jackson). Coverslips were treated with mounting media (DAKO), and cells were analyzed using a LSM710 confocal laser scanning microscope (Carl Zeiss). The nuclear localization of FoxO1 was determined by fluorescent intensity measured with Image J software. At least 200 cells per group were counted in each experiment.
Statistical Analysis-Statistical analysis was performed by Student's t test or ANOVA as appropriate. All values are expressed as means Ϯ S.E., and statistical significance was set at p Ͻ 0.05.