Glucose and Insulin Synergistically Activate Phosphatidylinositol 3-Kinase to Trigger Oscillations of Phosphatidylinositol 3,4,5-Trisphosphate in β-Cells*

In insulin-secreting β-cells, activation of phosphatidylinositol 3′-OH-kinase with resulting formation of phosphatidylinositol 3,4,5-trisphosphate (PIP3) has been implicated in the regulation of ion channels, insulin secretion, and gene transcription as well as in cell growth and survival, but the kinetics of PIP3 signals following physiological stimulation of insulin secretion is unknown. Using evanescent wave microscopy and a green fluorescent protein-tagged PIP3-binding protein domain for real-time monitoring of plasma membrane PIP3 concentration in single MIN6 β-cells, we now demonstrate that glucose stimulation of insulin secretion results in pronounced PIP3 oscillations via autocrine stimulation of insulin receptors. Glucose lacked effect when insulin secretion was prevented with the hyperpolarizing agent diazoxide, but the sugar dose dependently enhanced the PIP3 response to maximal insulin stimulation without affecting the rate of PIP3 degradation. We conclude that glucose is an important co-activator of phosphatidylinositol-3′-OH-kinase and that the plasma membrane PIP3 concentration in β-cells undergoes oscillations due to pulsatile release of insulin.

In insulin-secreting ␤-cells, activation of phosphatidylinositol 3-OH-kinase with resulting formation of phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) has been implicated in the regulation of ion channels, insulin secretion, and gene transcription as well as in cell growth and survival, but the kinetics of PIP 3 signals following physiological stimulation of insulin secretion is unknown. Using evanescent wave microscopy and a green fluorescent protein-tagged PIP 3 -binding protein domain for real-time monitoring of plasma membrane PIP 3 concentration in single MIN6 ␤-cells, we now demonstrate that glucose stimulation of insulin secretion results in pronounced PIP 3 oscillations via autocrine stimulation of insulin receptors. Glucose lacked effect when insulin secretion was prevented with the hyperpolarizing agent diazoxide, but the sugar dose dependently enhanced the PIP 3 response to maximal insulin stimulation without affecting the rate of PIP 3 degradation. We conclude that glucose is an important co-activator of phosphatidylinositol-3-OH-kinase and that the plasma membrane PIP 3 concentration in ␤-cells undergoes oscillations due to pulsatile release of insulin.
Insulin is secreted from pancreatic ␤-cells in response to an elevation of the plasma glucose concentration. The rapid uptake and metabolism of the sugar leads to an increase of the ATP/ADP ratio and closure of ATP-sensitive K ϩ channels in the plasma membrane. The resulting depolarization activates voltage-dependent Ca 2ϩ influx, which triggers exocytosis of the insulin secretory granules (1). The elevation of the cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] i ) 2 is often periodic, and coordination of [Ca 2ϩ ] i oscillations among ␤-cells is considered to account for the pulsatile release of insulin observed in healthy subjects (2).
In addition to endocrine stimulation of glucose uptake and storage in liver, muscle, and adipose tissue, insulin has autocrine effects on ␤-cells, regulating proliferation, survival, insulin synthesis, and secretion. Insulin has been reported to have both stimulatory (3)(4)(5)(6)(7) and inhibitory (8 -12) feedback effects on ␤-cells. Because some of the discordant observations may depend on different times allowed for insulin action, it is important to characterize the kinetics of insulin receptor-induced signaling events.
Glucose is known to activate PI3-kinase in ␤-cells, an effect that has been attributed to autocrine activation of insulin receptors by secreted insulin (13,14). However, little is known about the kinetics of this process, and it is unknown whether the pulsatile pattern of insulin release is reflected in the time course of PI3-kinase activation. In the present study, we used a PIP 3 -binding pleckstrin homology (PH) domain tagged with GFP in combination with evanescent wave microscopy for realtime monitoring of the plasma membrane PIP 3 concentration in individual MIN6 ␤-cells. It is demonstrated that glucose directly potentiates PI3-kinase activity in the presence of insulin and that this synergism, together with pulsatile release of insulin, results in pronounced oscillations of the plasma membrane PIP 3 concentration in glucose-stimulated ␤-cells.

EXPERIMENTAL PROCEDURES
Materials-The PH domain from protein kinase B/Akt fused to the green fluorescent protein (GFP-PH Akt ) was used as a translocation biosensor for membrane PIP 3 concentration (21). This PH domain has a well documented binding preference for PIP 3 and phophatidylinositol 3,4-bisphosphate (PI3,4P 2 ) over other phosphoinositides and inositolpolyphosphates (22), and it was verified that receptor-induced changes in phosphatidylinositol 4,5-bisphosphate and inositol 1,4,5-trisphosphate did not induce GFP-PH Akt translocation. To confirm that the GFP-PH Akt responses reflected changes of PIP 3 rather than PI3,4P 2 , experiments were also performed with GFP-tagged GRP1 (gen-eral receptor for phosphoinositides-isoform 1), the PH domain of which only binds PIP 3 (22). Because of its strong nuclear localization, however, GRP1-GFP was used only in the experiments shown in Fig. 5B. The plasmids encoding the biosensor constructs as well as GFP targeted to the membrane via covalent lipid modification (GFP-CAAX) was generously provided by Professor Tobias Meyer, Stanford University. Diazoxide was a kind gift from Schering-Plough Int. (Kenilworth, NJ). Adrenaline, insulin, and LY294002 were from Sigma. Dulbecco's modified Eagle's medium and Lipofectamine 2000 were obtained from Invitrogen, and Fura Red was provided by Molecular Probes Invitrogen (Portland, OR).
Cell Culture and Transfection-MIN6 cells (passage 17-30) (23) were maintained in Dulbecco's modified Eagle's medium containing 4500 mg/liter glucose and supplemented with 15% fetal calf serum, 2 mM L-glutamine, 50 M 2-mercaptoethanol, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a 5% CO 2 humidified atmosphere. The cells were seeded onto poly-L-lysine-coated 25-mm glass coverslips to achieve 50% confluency on the day of transfection. Transient transfection was performed with 2 g of plasmid DNA and Lipofectamine 2000 in a ratio of 1:2.5 in 1 ml of Dulbecco's modified Eagle's medium during 4 h followed by repeated washing and further culture in Dulbecco's modified Eagle's medium for 12-24 h. Prior to experiments, the cells were transferred to a buffer containing 125 mM NaCl, 5.9 mM KCl, 1.28 mM CaCl 2 , 1.2 mM MgCl 2 , and 25 mM HEPES with pH adjusted to 7.40 with NaOH and incubated for 45 min at 37°C. In experiments with [Ca 2ϩ ] i measurements, the cells were preincubated in the presence of 10 M acetoxymethyl ester of the fluorescent Ca 2ϩ indicator Fura Red.
Fluorescence Microscopy-The subcellular localization of GFP-PH Akt was analyzed using a confocal system (Yokogawa CSU-10 spinning disk scanhead; Andor Technology, Belfast, Northern Ireland) attached to a Diaphot 200 microscope (Nikon, Kanagawa, Japan) equipped with a 60ϫ 1.40-NA objective. The 488-nm beam from an argon ion laser (Melles-Griot, Didam, The Netherlands) was coupled to the scanhead through an optical fiber (Point-Source, Southampton, UK). Fluorescence was detected at 520/35 nm using an Orca-AG camera (Hamamatsu Photonics, Hamamatsu City, Japan) under MetaFluor software control (Molecular Devices Corp., Downingtown, PA).
The plasma membrane concentration of GFP-PH Akt was recorded with an evanescent wave microscopy setup built around an Eclipse TE2000 microscope (Nikon) with a 60ϫ 1.45-NA objective as previously described (24). Selection of excitation and emission wavelengths was made with the following filters (center wavelength/half-bandwidth nm): GFP exc 488/10 nm, em 525/25 nm; Fura Red exc 488/10 nm and em 630 nm long pass. The fluorescence was registered by an Orca-ER camera (Hamamatsu) under MetaFluor software control (Molecular Devices). If not otherwise stated, images were acquired by 50 -100-ms exposure every 5 s. To avoid the potentially harmful laser light, the beam was blocked by an electronic shutter (Sutter Instruments, Novato, CA) between image captures.
Measurements of Insulin Secretion-MIN6 cells were seeded into 24-well plates and grown to ϳ70% confluency. Before measurements of insulin secretion the cells were preincubated for 90 min at 37°C and 5% CO 2 in Krebs Ringer Bicarbonate Hepes (KRBH) buffer (130 mM NaCl, 4.8 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 2.5 mM CaCl 2 , 10 mM HEPES, and 5 mM NaHCO 3 ) supplemented with 0.1% bovine serum albumin and 3 mM glucose and with pH adjusted to 7.40 with NaOH. Insulin secretion was then determined by further incubation for 30 min at 37°C and 5% CO 2 in KRBH buffer supplemented with various combinations of secretagogues and test substances. Insulin concentrations in the medium were determined using a mouse insulin enzyme-linked immunosorbent assay kit (Mercodia, Uppsala, Sweden). Total DNA was extracted from the cells and the concentration determined by spectrophotometry. Insulin release was normalized for total cellular DNA content and expressed in relation to the secretion obtained with 3 mM glucose.
Data and Statistical Analysis-Image analysis was performed using MetaFluor, MetaMorph (Universal Imaging), and ImageJ (W. S. Rasband, National Institutes of Health) software. Fluorescence intensities are expressed as changes relative to initial fluorescence after subtraction of background (⌬F/F 0 ). The Fura Red traces have been inverted to show increases of [Ca 2ϩ ] i as upward deflections. The SigmaStat software (SPSS Inc., Chicago, IL) was used for regression analysis and Igor Pro (Wavemetrics, Lake Oswego, OR) and Illustrator (Adobe Systems Inc., San José, CA) software for curve fitting and illustrations. All data are presented as mean values Ϯ S.E. Differences were statistically evaluated by two-tailed Student's t test. 3 Formation in Insulin-secreting MIN6 ␤-Cells via Activation of PI3-Kinase-Confocal microscopy imaging of GFP-PH Akt -expressing MIN6 ␤-cells maintained in basal medium containing 3 mM glucose demonstrated that the fluorescence was homogenously distributed in the cytoplasm with a slight enrichment in the nucleus (Fig. 1A). Stimulation of the cells with 100 nM insulin resulted in a pronounced translocation of the biosensor to the plasma membrane accompanied by a reduction of fluorescence intensity in the cytoplasm. To selectively image plasma membrane fluorescence we applied evanescent wave microscopy, which provides excitation restricted to a volume within ϳ100 nm of the plasma membrane. With this technique the insulin-induced GFP-PH Akt translocation is detected as an increase of fluorescence throughout the membrane adhering to the coverslip (Fig.  1B). The membrane fluorescence increased by 30 Ϯ 3% with half-maximal translocation at 45 Ϯ 5 s (n ϭ 43). After removal of the stimulus the fluorescence returned to base line with t1 ⁄ 2 ϭ 90 Ϯ 7 s (n ϭ 23; Fig. 1B). Consistent with the GFP-PH Akt translocation reflecting PI3-kinase-mediated formation of PIP 3 , there was no response when insulin was added in the presence of 100 M PI3-kinase inhibitor LY294002 but pronounced GFP-PH Akt translocation when the inhibitor was removed (Fig. 1C). The insulin-induced GFP-PH Akt translocation was dose dependent with half-maximal and maximal effects obtained at 454 pM and ϳ10 nM insulin, respectively (Fig.  1D). No further effect was observed with insulin concentrations as high as 10 M (data not shown). Insulin failed to alter the fluorescence recorded from cells expressing GFP alone in the cytoplasm (n ϭ 9, data not shown) or targeted to the plasma membrane (n ϭ 11, Fig. 1D), providing further support for the specificity of the GFP-PH Akt response. Together, these data demonstrate that MIN6 cells respond to insulin with dose-dependent activation of PI3-kinase and formation of PIP 3 in the plasma membrane. PIP 3 Formation Triggered by Endogenous Insulin Secretion in MIN6 ␤-Cells-It was next tested whether stimulation of endogenous insulin secretion resulted in activation of PI3-kinase. MIN6 cells expressing GFP-PH Akt and loaded with the fluorescent Ca 2ϩ indicator Fura Red were stimulated by raising the extracellular KCl concentration to 30 mM. The resulting depolarization triggered an almost instantaneous rise of the cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] i ) due to influx of the ion through voltage-dependent channels ( Fig. 2A), followed by a pronounced increase of GFP-PH Akt fluorescence. The time course of the GFP-PH Akt translocation was biphasic with an initial peak reaching 28 Ϯ 3% above base line (t1 ⁄ 2 ϭ 27 Ϯ 2 s, n ϭ 21), followed within 2 min by a sustained plateau 20 Ϯ 2% above the initial intensity (n ϭ 13; Fig. 2A). Both [Ca 2ϩ ] i and GFP-PH Akt fluorescence returned to the prestimulatory levels after normalization of the KCl concentration (t1 ⁄ 2 ϭ 47 Ϯ 5 s for GFP-PH Akt , n ϭ 8). There was no effect of KCl depolarization in control cells expressing cytoplasmic (n ϭ 9) or membrane-targeted GFP (n ϭ 7; data not shown). When GFP-PH Akt -expressing cells were depolarized with 30 mM KCl in a Ca 2ϩ -deficient medium containing 2 mM EGTA, there was neither rise of [Ca 2ϩ ] i nor translocation of GFP-PH Akt (Fig. 2B). However, reintroduction of 1.3 mM Ca 2ϩ immediately induced elevation   ] i and GFP-PH Akt translocation (peak 28 Ϯ 5%; plateau 18 Ϯ 4%, n ϭ 9), supporting the conclusion that Ca 2ϩ -triggered exocytosis of insulin secretory granules results in pronounced autocrine activation of PI3-kinase. The lack of GFP-PH Akt response in cells stimulated in the nominal absence of extracellular Ca 2ϩ was not due to an inhibition of the GFP-PH Akt translocation per se, because the cells readily responded to 100 nM exogenous insulin (18 Ϯ 2%, n ϭ 14; Fig. 2C).

Insulin Stimulates Dose-dependent PIP
The secretory responses of the cells were verified by conventional enzyme-linked immunosorbent assay detection of insulin release. Elevation of KCl to 30 mM induced an ϳ3.5-fold stimulation of insulin secretion, which was abolished (n ϭ 3; p Ͻ 0.001) by removal of Ca 2ϩ from the external medium (Fig. 2G).
Adrenaline Inhibits the Insulin Response Independent of [Ca 2ϩ ] i -To verify that the depolarization-induced PIP 3 formation was due to secreted insulin and not a direct effect of the voltage-dependent Ca 2ϩ influx, we stimulated the cells with 30 mM KCl in the presence of 10 M adrenaline, a known inhibitor of insulin secretion (25). Addition of 10 M adrenaline caused a small (6 Ϯ 1%, n ϭ 17) increase of plasma membrane GFP-PH Akt fluorescence, which probably did not reflect activation of PI3-kinase, as a similar increase was observed with membranetargeted GFP (data not shown). The presence of adrenaline prevented the KCl-induced translocation of GFP-PH Akt (Fig. 2, D  and E). In contrast, adrenaline had no effect on the depolarization-induced elevation of [Ca 2ϩ ] i . Thus, KCl induced a 20 Ϯ 1% (n ϭ 6) change in Fura Red fluorescence compared with 19 Ϯ 3% (n ϭ 6) in the absence of adrenaline (Fig. 2, D and F). Moreover, adrenaline did not prevent the GFP-PH Akt response to 100 nM exogenous insulin (22 Ϯ 2% fluorescence increase, n ϭ 13, Fig. 2E). Conventional enzyme-linked immunosorbent assay measurements of insulin secretion verified that 10 M adrenaline completely inhibited KCl-stimulated insulin release without effect on basal secretion (Fig. 2G). These data support the conclusion that the depolarization-and Ca 2ϩ -induced PIP 3 formation in insulin-secreting cells is due to secreted insulin.
Glucose Stimulates PIP 3 Formation in Insulin-secreting Cells-It was next determined how the plasma membrane PIP 3 concentration was affected by glucose. Elevation of the glucose concentration from 3 to 20 mM resulted in an initial lowering of [Ca 2ϩ ] i , known to reflect fueling of ␤-cell Ca 2ϩ -ATPases responsible for sequestration of the ion into the endoplasmic reticulum (26), followed after 2-3 min by a pronounced increase reflecting voltage-dependent Ca 2ϩ influx. The rise of [Ca 2ϩ ] i was followed after 23 Ϯ 2 s (n ϭ 18) by a pronounced increase of plasma membrane GFP-PH Akt fluorescence (Fig.  3A). The glucose-induced GFP-PH Akt translocation was biphasic with an initial peak reaching 48 Ϯ 3% (n ϭ 59) above base line, followed by a sustained increase at 31 Ϯ 2% (n ϭ 35) above initial fluorescence (Fig. 3A). After lowering of the glucose concentration to 3 mM the GFP-PH Akt fluorescence returned to base line with t1 ⁄ 2 ϭ 102 Ϯ 12 s (n ϭ 18).
Synergistic Effect of Glucose and Insulin on PIP 3 Formation-We also examined whether the glucose-induced formation of PIP 3 was entirely due to secreted insulin or whether glucose had an independent stimulatory effect. To this end MIN6 cells were stimulated with glucose in the presence of 250 M of the hyper-polarizing agent diazoxide, which prevents elevation of [Ca 2ϩ ] i by clamping the ␤-cell membrane potential close to the equilibrium potential for K ϩ . Elevation of the glucose concentration to 20 mM resulted in a small rise of GFP-PH Akt (Fig. 3B) with a magnitude equal to that seen in control cells expressing membrane-targeted GFP (7 Ϯ 1%, n ϭ 14). Removal of diazoxide resulted in additional (38 Ϯ 4%, n ϭ 40) increase of GFP-PH Akt fluorescence (Fig. 3B), not seen with membrane-GFP. The effects of glucose and diazoxide on plasma membrane PIP 3 concentration correlated well with their effects on insulin release (Fig. 3C). These experiments indicate that glucose has little effect on PIP 3 formation in the absence of secreted insulin.
To test whether glucose affected insulin-induced GFP-PH Akt translocation, MIN6 cells were stimulated with a maximally activating concentration of insulin (100 nM) in the presence of 250 M diazoxide. This resulted in a 32 Ϯ 1% (n ϭ 44) increase of GFP-PH Akt fluorescence (Fig. 4, A and B). Under these conditions the response to exogenous insulin is saturated and diazoxide should prevent depolarization and endogenous release of insulin. Nevertheless, elevation of the glucose concentration from 3 to 20 mM induced a rapid and sustained increase in GFP-PH Akt fluorescence reaching 70 Ϯ 6% (n ϭ 48) above the level observed with insulin alone (p Ͻ 0.001; Fig. 4, A and B). Under the same conditions, glucose had no effect on the fluorescence from membrane-targeted GFP (Fig. 4C). The amplifying effect of glucose showed a hyperbolic concentration dependence with half-maximal enhancement at 16 mM of the sugar (Fig. 4D). At 8 mM glucose there was a delay between increase of the sugar concentration and the rise of GFP-PH Akt fluorescence of approximately 1 min (Fig. 4E), but this lag decreased exponentially with increasing glucose concentration, being only 10 s at 30 mM of the sugar (Fig. 4E). The amplifying effect of glucose on PIP 3 formation was not associated with a change of [Ca 2ϩ ] i (data not shown) and could not be reproduced by the cAMP-elevating phosphodiesterase inhibitor 3-isobutylmethylxanthine (not shown). Moreover, the rate of PIP 3 degradation after inhibition of PI3kinase with 100 M LY294002 was unaffected by glucose (t1 ⁄ 2 ϭ 42 Ϯ 4 s, n ϭ 15 at 3 mM versus 40 Ϯ 5 s, n ϭ 18 at 20 mM, Fig. 4F), indicating that glucose acts synergistically with insulin to accelerate PI3-kinase-dependent PIP 3 formation.
Glucose-induced Oscillations of PIP 3 -When the MIN6 cells were exposed to elevated glucose concentrations during extended periods of time (Ͼ10 min), it became apparent that the initial peak of GFP-PH Akt translocation was followed by oscillations from an elevated level ( Fig. 5A) with amplitudes and frequencies averaging 20 Ϯ 1% change of membrane fluorescence and 0.224 Ϯ 0.015 min Ϫ1 (n ϭ 69), respectively. A similar response pattern was seen with GRP1-GFP as PIP 3 biosensor (Fig. 5B). Simultaneous recording of [Ca 2ϩ ] i and GFP-PH Akt demonstrated that each rise of PIP 3 was preceded by an increase of [Ca 2ϩ ] i (Fig. 5C). Occasionally, rapid transients of [Ca 2ϩ ] i were observed in the presence of elevated glucose concentrations, and these were also associated with transient increases of GFP-PH Akt fluorescence (Fig. 5D). More consistent rapid oscillations of both [Ca 2ϩ ] i and GFP-PH Akt were observed after stimulation of the cells with 20 mM of the K ϩ channel blocker tetraethylammonium ϩ (Fig. 5E). The delay observed between the peak of [Ca 2ϩ ] i and that of GFP-PH Akt translocation (20 Ϯ 2 s, n ϭ 18) reflects, at least in part, the time required for insulin receptor signal transduction. The lag between application of insulin and the first detectable GFP-PH Akt translocation averaged 14 Ϯ 2 s (n ϭ 34) (Fig. 5F). These data show that glucose stimulation of ␤-cells is associated with oscillations of PIP 3 in the plasma membrane.

DISCUSSION
Two important conclusions can be inferred from the present data. First, the autocrine activation of insulin receptors in ␤-cells with PI3-kinase-dependent formation of PIP 3 is markedly amplified by glucose. Direct control of PI3-kinase by a nutrient stimulus constitutes a novel regulatory principle for this ubiquitous signaling enzyme. The synergism between glucose and insulin enables PI3-kinase to serve as a coincidence detector, which adds a layer of complexity to the regulation of ␤-cell function by nutrients and receptor agonists. Second, our study provides the first demonstration that the plasma membrane PIP 3 concentration can undergo oscillations in response to physiological cell stimulation. The PIP 3 oscillations reflect pulsatile release of insulin from individual ␤-cells combined with the glucose-induced amplification of PI3-kinase activity and rely on a short half-life of the lipid messenger. Oscillations should expand the information content of the PIP 3 signal and contribute to the specificity and diversity of signaling downstream of PI3-kinase. In particular, the PIP 3 oscillations may partake in the generation of oscillatory electrical activity and pulsatile insulin release.
Glucose and Insulin Synergistically Activate PI3-Kinase-Insulin receptors were first demonstrated in ␤-cells in the early 1980s (27,28). Glucose-induced insulin secretion was later shown to activate insulin receptors as well as the downstream signaling proteins IRS-1 and PI3-kinase (13,14). Consistent with PI3-kinase being activated via secreted insulin, knock down of insulin receptors with small interfering RNA prevented glucose activation of PI3-kinase (3). We now demonstrate that glucose has a direct effect on PI3-kinase activity in the presence of insulin. Although insulin stimulation alone resulted in pronounced activation of the enzyme, PIP 3 formation was markedly enhanced after elevation of the glucose concentration. The larger effect by glucose cannot be explained by the insulin receptors being exposed to higher concentrations of the ligand, because the sugar also amplified the response to maximally activating concentrations of exogenous insulin and when insulin secretion was prevented with the ATP-sensitive K ϩ channel opener diazoxide. The effect of glucose is apparently due to accelerated PIP 3 production, because the sugar was without significant effect on PIP 3 degradation. Glucose stimulation of PI3-kinase activity required that the enzyme was already activated, because there was no significant stimulation of PIP 3 formation in the absence of exogenous insulin under conditions when endogenous secretion was prevented with diazoxide. Earlier studies have demonstrated that glucose is necessary for insulin-like growth factor I-mediated ␤-cell growth (29), although it is not excluded that the amplifying effect of glucose was mediated via glucose-induced insulin secretion.
Control of PI3-kinase activity by glucose may be an important regulatory mechanism, particularly in the ␤-cells. With the high sensitivity of ␤-cells to insulin, the concentration of insulin reached even at low glucose concentrations in the narrow intercellular space within the islet should result in significant constitutive PI3-kinase activity. Glucose regulation of PI3-kinase would thus extend the dynamic range of enzyme activity beyond that achieved by receptor stimulation alone. A similar mechanism may be important for enhancing the response to growth factors and other PI3-kinase activators selectively under glucose-stimulated conditions. Further studies are warranted to clarify the mechanisms underlying direct glucose acti- vation of PI3-kinase. It is possible that ATP generated by glucose metabolism directly stimulates PI3-kinase activity. Another possibility is that glucose-induced generation of reactive oxygen species (30) leads to PI3-kinase activation via receptor tyrosine kinases (31,32).
Glucose-induced PIP 3 Oscillations-The glucose-induced oscillations of PIP 3 concentration in the plasma membrane are probably the result of pulsatile insulin release combined with glucose amplification of PI3-kinase activity and the observed short half-life (ϳ40 s) of the lipid in the membrane. Insulin released from the pancreas (33) and from individual islets is pulsatile (2). Measurements of [Ca 2ϩ ] i (34) and of Zn 2ϩ co-released from the insulin secretory granules (35) indicate that isolated ␤-cells also show pulsatile secretory activity. The present observations provide further evidence that insulin is released in pulses from individual cells. Pulsatile insulin secretion is considered important to prevent desensitization or down-regulation of the receptors in the target tissues (36), and deterioration of the regular plasma insulin oscillations is observed in patients with type-2 diabetes (37) as well as in their close relatives (38).
The significance of the PIP 3 oscillations is unclear. However, the lipid is a putative mediator of PI3-kinase activation of the ATP-sensitive K ϩ channel (10,18). Oscillations of PIP 3 may therefore not only be a consequence of pulsatile insulin secretion but via feedback effects on glucose-induced oscillations of the membrane potential and [Ca 2ϩ ] i also take part in the generation of pulsatile hormone release (10). The time course of PIP 3 oscillations with the rise of PIP 3 coinciding with decrease of [Ca 2ϩ ] i is consistent with PIP 3 being involved in such a regulatory circuit.
It is well established that different time courses of messenger signals, including PIP 3 , may induce different downstream cellular responses. For example, in adipose cells the efficiency of insulin and platelet-derived growth factor to induce membrane insertion of the GLUT4 glucose transporter depends on the amplitude and duration of the PIP 3 signals induced by the different receptors (21). Oscillatory signaling has not previously been described for PIP 3 but is well established for Ca 2ϩ and cAMP. Recent observations in our laboratory have shown that cAMP oscillations restrict the activity of protein kinase A to the cytoplasm (39). Other studies indicate that the frequency and amplitude of Ca 2ϩ oscillations affect the specificity and efficiency of gene transcription (40). Future studies will clarify whether PIP 3 oscillations contribute to selective activation of downstream processes in an analogous manner.
In conclusion, the present study provides new insights into the signaling downstream of the insulin receptor. Glucose is identified as an important co-activator of PI3-kinase, and glucose-induced insulin secretion is found to be associated with PIP 3 oscillations in individual insulin-secreting cells. Dysregulation of the delicate balance between PIP 3 production and degradation may contribute to the disturbed insulin secretory pattern in type-2 diabetes.