β-Arrestin 1 Is Required for PAC1 Receptor-mediated Potentiation of Long-lasting ERK1/2 Activation by Glucose in Pancreatic β-Cells*

In pancreatic β-cells, the pituitary adenylate cyclase-activating polypeptide (PACAP) exerts a potent insulin secretory effect via PAC1 and VPAC receptors (Rs) through the Gαs/cAMP/protein kinase A pathway. Here, we investigated the mechanisms linking PAC1R to ERK1/2 activation in INS-1E β-cells and pancreatic islets. PACAP caused a transient (5 min) increase in ERK1/2 phosphorylation via PAC1Rs and promoted nuclear translocation of a fraction of cytosolic p-ERK1/2. Both protein kinase A- and Src-dependent pathways mediated this transient ERK1/2 activation. Moreover, PACAP potentiated glucose-induced long-lasting ERK1/2 activation. Blocking Ca2+ influx abolished glucose-induced ERK1/2 activation and PACAP potentiating effect. Glucose stimulation during KCl depolarization showed that, in addition to the triggering signal (rise in cytosolic [Ca2+]), the amplifying pathway was also involved in glucose-induced sustained ERK1/2 activation and was required for PACAP potentiation. The finding that at 30 min glucose-induced p-ERK1/2 was detected in both cytosol and nucleus while the potentiating effect of PACAP was only observed in the cytosol, suggested the involvement of the scaffold protein β-arrestin. Indeed, β-arrestin 1 (β-arr1) depletion (in β-arr1 knockout mouse islets or in INS-1E cells by siRNA) completely abolished PACAP potentiation of long-lasting ERK1/2 activation by glucose. Finally, PACAP potentiated glucose-induced CREB transcriptional activity and IRS-2 mRNA expression mainly via the ERK1/2 signaling pathway, and likewise, β-arr1 depletion reduced the PACAP potentiating effect on IRS-2 expression. These results establish for the first time that PACAP potentiates glucose-induced long-lasting ERK1/2 activation via a β-arr1-dependent pathway and thus provide new insights concerning the mechanisms of PACAP and glucose actions in pancreatic β-cells.

The neuropeptide, Pituitary Adenylate Cyclase-Activating Polypeptide (PACAP) 4 plays an important role in the regulation of pancreatic islet functions. PACAP is expressed in islet parasympathetic nerve terminals and strongly potentiates insulin secretion in a glucose-dependent manner both in vitro (1)(2)(3)(4)(5) and in vivo in rodent (6) and humans (7). The physiological functions of PACAP are mediated by three receptor subtypes that belong to the class II G-protein-coupled receptors (GPCRs): PAC 1 , VPAC 1 and VPAC 2 receptors (Rs). PAC 1 R is selective for PACAP, whereas VPAC 1 R and VPAC 2 R bind to PACAP and Vasoactive Intestinal Peptide (VIP) with equal high affinity (8). Class II GPCRs, which include also receptors for glucagon and the incretin glucagon-like peptide 1 (GLP-1), are coupled to the heterotrimeric G-protein Gs, which stimulates the adenylate cyclase (AC), (9). Besides its insulinotropic action, PACAP has been recently reported to exert long term beneficial effects on ␤-cell mass in various experimental mouse models of diabetes (10,11). However, the receptor(s) and the mechanism(s) involved in these long term effects are yet unknown.
Extracellular signal-regulated kinases 1 and 2 (ERK1/2) are important effectors of GPCRs and regulate cell growth, survival, and differentiation (12). In pancreatic ␤-cells, cross-talk between the cAMP-protein kinase A (PKA) and ERK1/2 signaling pathways has been reported (13)(14)(15)(16). PACAP, like glucagon and GLP-1, positively regulates ERK1/2 in ␤-cells (17,18). However, for PACAP, the receptor(s) and the mechanism(s) involved in ERK1/2 activation are yet unknown in ␤-cells. Depending on the receptor or the cellular context, GPCRs activate the Raf/MEK/ERK cascade through different pathways (19,20). Recent studies have shown that GPCRs activate ERK1/2 through G-protein-dependent and G-protein-independent pathways (19 -21). The G-protein-independent pathway occurs after G-protein uncoupling and results from the recruitment of scaffolding proteins, such as ␤-arrestins, which are cytosolic proteins initially identified for their role in GPCRs desensitization. Interestingly, G-protein-and ␤-arrestin-dependent activations of the ERK1/2 pathway are both temporally and spatially distincts in cells and thus modulate differently the functions of ERK1/2. Hence, whereas G-proteindependent activation of ERK1/2 signal is transient and nuclear, ERK1/2 activation through the ␤-arrestin-dependent pathway is usually sustained and retained in the cytosol.
Glucose is the major physiological stimulus of pancreatic ␤-cells. It exerts a tight control on insulin secretion through its metabolism via two major, hierarchical signaling pathways: a triggering pathway, i.e. the ATP-sensitive K ϩ (K ϩ -ATP) channel-dependent pathway, and an amplifying pathway, i.e. the K ϩ -ATP channel-independent pathway (22). The triggering pathway results from oxidative glycolysis that increases the ATP/ADP ratio, leading to closure of K ϩ -ATP channels, and, subsequently, membrane depolarization, opening of voltagedependent Ca 2ϩ channels (VDCCs) and a rise in intracellular calcium ([Ca 2ϩ ] i ), which is the triggering signal. The amplifying pathway does not further increase [Ca 2ϩ ] i , but rather enhances the response to the triggering Ca 2ϩ signal. In pancreatic ␤-cells, glucose, at the physiological concentration range, stimulates ERK1/2 by a rise in [Ca 2ϩ ] i (15,17,18,(23)(24)(25). A major physiological relevance of the activation of ERK1/2 by glucose in ␤-cells is the stimulation of both transcription through phosphorylation of various transcriptional factors (26,27) and secretion (28) of insulin.
To further explore the role of PACAP in ␤-cells, we investigated the mechanisms linking PACAP receptors and ERK1/2 activation in the INS-1E cell line and mouse pancreatic islets, using pharmacological and molecular approaches. PACAP alone induces transient ERK1/2 activation via PAC 1 R through both PKA-and tyrosine kinase Src-dependent pathways, and promotes the nuclear translocation of a fraction of phosphorylated ERK1/2 (p-ERK1/2). In contrast, in the presence of glucose, PACAP potentiates the glucose-induced long-lasting ERK1/2 activation via a ␤-arrestin 1-dependent pathway in both INS-1E cells and pancreatic islets. In addition, activation of ERK1/2 through the ␤-arrestin 1-dependent pathway is involved in PACAP-induced insulin receptor substrate-2 (IRS-2) protein expression, which is crucial for ␤-cell function and survival.
Western Blotting-Cell lysates were clarified by centrifugation (15,000 ϫ g for 30 min at 4°C). Protein concentrations were determined by the bicinchoninic acid (BCA) method. Equal amounts of proteins were denatured in Laemmli sample buffer, separated through 10% SDS-PAGE, and transferred to nitrocellulose membranes. After blocking, membranes were probed with the appropriate antibody at 4°C overnight and then incubated with horseradish peroxidase-linked secondary antibody followed by enhanced chemiluminescence detection. Autoradiographs were digitized, and the band density was analyzed using ImageJ (National Institutes of Health, Bethesda, MD).
Silencing ␤-Arrestin 1 Expression Using RNA Interference-Three different 25-nucleotide stealth siRNA doublestranded duplexes were designed to specifically knock down rat ␤-arr1 expression. These siRNAs and a non-silencing RNA duplex (control siRNA), were used in deprotected and desalted forms. INS-1E cells were transiently transfected with 40 nM siRNA duplex using Lipofectamine TM 2000 according to the manufacturer's instructions. Briefly, INS-1E cells (400,000 cells per well) were seeded in 6-well/ plates in antibiotic-free culture medium 1 day before transfection. Then at ϳ40 -50% confluency, cells were transfected in Opti-MEM with Stealth TM RNAi-Lipofectamine TM 2000 complexes. Six hours after transfection, medium was replaced with complete RPMI medium. All assays were performed at least 70 h after siRNA transfection.
Subcellular Fractionation-After a 2-h preincubation in KRBH without glucose, INS-1E cells plated in 15-cm dishes (70 -80% confluency) were stimulated, or not, with glucose and/or PACAP for 5 or 30 min. Cells were then washed twice with ice-cold PBS and scraped in 1 ml of hypotonic buffer (10 mM HEPES, 10 mM NaCl, 1 mM KH 2 PO 4 , 5 mM NaHCO 3 , 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM EDTA, Proteases Inhibitors Mixture). After 15-min incubation, cells were Dounced 40 times on ice and centrifuged at 1000 ϫ g for 5 min at 4°C. Supernatants containing cytosol and membranes were collected for protein determination before denaturation in Laemmli buffer and Western blot analysis by SDS-PAGE. Pellets containing nuclei were Dounced 30 times in Tris-buffered saline-sucrose/EDTA buffer (10 mM Tris (pH 7.5), 300 mM sucrose, 1 mM EDTA (pH 8), 0.02% Nonidet P-40, Proteases Inhibitors Mixture), centrifuged at 2500 ϫ g for 5 min at 4°C, and pellets were rinsed twice in 1 ml of TSE buffer. Finally, pellets containing pure nuclei were dissolved in Tris-buffered saline-sucrose/EDTA buffer for protein determination before Western blot analysis by SDS-PAGE.
Luciferase Reporter Gene Assay-The transcriptional activity of CREB (cAMP-response element-binding protein) was evaluated using the PathDetect trans-reporting system, which consists in a fusion trans-activator plasmid expressing the activation domain of CREB fused with the yeast GAL4 DNA binding domain (pFA2-CREB) and a luciferase reporter plasmid containing a synthetic promoter with five tandem repeats of GAL4 binding sites that control expression of the luciferase gene (pFR-Luc). Transient transfections in INS-1E cells were carried out using Lipofectamine TM 2000. Briefly, INS-1E cells (250,000 cells per well) were seeded in 12-well plates with antibiotic-free culture medium 1 day before transfection. Cells at ϳ50 -60% confluency were transfected in Opti-MEM with the reporter (pFR-Luc) and the fusion transactivator (pFA2-CREB) plasmids. After 6-h transfection, medium was replaced with complete culture medium. The following day, after 2 h of quiescence in glucose-free KRBH buffer with or without 10 M U0126, transfected cells were incubated for 6 h as indicated in the legend of Fig. 7A. Following the treatment, cells were washed in PBS and harvested in active reporter lysis buffer for 15 min with shaking. Cell debris was removed by centrifugation (12,000 ϫ g for 2 min at 4°C), and supernatants were stored at Ϫ80°C until luciferase activity assay and protein determinations. Luciferase activity was measured using the enhanced Luciferase Assay kit and the ANALYST TM fluorometer (Molecular Devices). Protein concentrations were determined with the BCA method and used to normalize luciferase activity.
The selection of the appropriate housekeeping genes was performed using geNorm (31). The level of expression of each gene X was normalized to the geometric mean of the expression levels of three housekeeping genes R (B2M, Tbp, and Tubb2) according to the formula, X/geometric mean (R1, R2, and R3) ϭ 2 (Ct(X) Ϫ arithmetic mean (Ct(R1),Ct(R2),Ct(R3)) , where C t is the threshold cycle.
Expression of Data and Statistics-Data were presented as mean Ϯ S.E. of n independent experiments. Statistically significant differences between groups were assessed by Student's t test or by analysis of variance, followed by the Newman-Keuls test in the case of multiple comparisons. Differences were considered significant at p Ͻ 0.05.

PACAP via PAC 1 R Induces a Rapid and Transient Activation of ERK1/2 and Its Nuclear
Translocation-To investigate the mechanism(s) by which PACAP activates ERK1/2, we assessed ERK1/2 phosphorylation status by Western blotting with phospho-specific antibodies that recognize only Thr-202/Tyr-204phosphorylated ERK1/2 in the rat pancreatic ␤-cell line INS-1E. We firstly analyzed the effects of PACAP alone without glucose.
PACAP (100 nM) alone induced a rapid, monophasic, and transient phosphorylation of ERK1/2 that peaked at 5 min (by ϳ3.5-fold increase) and returned to basal values within 20 -30 min (Fig. 1A). Based on these results, we recorded the effect of PACAP at 5 min in the subsequent experiments. PACAP binds to PAC 1 Rs and VPACRs. To determine which receptor(s) mediated this stimulatory effect, we compared the effects of VIPandPACAP.PACAP(1pMto10nM)causedaconcentrationdependent phosphorylation of ERK1/2 ( Fig. 1B) with a maximal effect between 10 and 100 nM. In contrast, VIP (1 pM to 1 mM) was ineffective, which is consistent with the involvement of PAC 1 R alone in ERK1/2 activation. Spatiotemporal regulation of ERK1/2 activation plays an important role in determining the functions of these kinases (32). To investigate the subcellular distribution of p-ERK1/2, subcellular fractionation and immunofluorescence observed by confocal microscopy analysis were performed. In quiescent INS-1E cells, the majority of ERK1/2 was in the cytosolic fraction (Fig. 1C). However, there was a low but detectable amount of ERK2 in the nuclear fraction. Addition of PACAP induced an accumulation of both ERK1 and ERK2 in the nucleus and also an increase in p-ERK1/2 in both the cytosolic and nuclear fractions suggesting a translocation to the nucleus of a fraction of the cytosolic p-ERK1/2. No effect of PACAP was observed at 30 min (data not shown). In addition, although quiescent INS-1E cells exhibited no staining for p-ERK1/2 ( Fig. 1D), we found that ϳ14 Ϯ 3% of cells were responsive to PACAP and displayed p-ERK1/2 immunofluorescence in both the cytoplasm and the nucleus. Thus, PACAP induces transient ERK1/2 activation in both the cytosol and nucleus. PAC 1 R Transiently Activates ERK1/2 through MEKs, Gs/PKA, and Src kinase-ERK1 and ERK2 are phosphorylated on Thr-202 and Tyr-204 by ERK1/2 kinases (MEK1/2). To determine the intracellular mechanisms involved in PACAP activation of ERK1/2, we first evaluated the role of the upstream kinases MEK1/2 by using two selective inhibitors, PD98059 (20 M) and U0126 (10 M). They completely abolished the PACAP effect on ERK1/2 (supplemental Fig. S1, A and B), indicating the involvement of the upstream MEK1/2 in the PACAP-induced ERK1/2 activation.
On the other hand, it is known that [Ca 2ϩ ] i is critical to mediate ERK1/2 activation in pancreatic ␤-cells by physiological stimuli, i.e. glucose and GLP-1 (13,18,26). Furthermore, PACAP has been shown to increase [Ca 2ϩ ] i in a glucosedependent manner via opening of the VDCCs (1, 33). To determine whether Ca 2ϩ played a role in PACAP-induced ERK1/2 activation, we used the VDCCs blocker, nifedipine. In line with the observation that PACAP had no effect on [Ca 2ϩ ] i in the absence of glucose (1), the transient activation of ERK1/2 by PACAP was not significantly inhibited by nifedipine (supplemental Fig. S1, C and D). These results suggest that Ca 2ϩ influx via VDCCs is not essential in the PACAPinduced transient activation of ERK1/2. Equal amounts of total cellular lysates (30 g) were separated by SDS-PAGE and probed for p-ERK1/2 or total ERK1/2 as described under "Experimental Procedures." Representative blots are shown, and graphs illustrate the quantitative analyses of the increase of ERK2 activation over basal. Data are means Ϯ S.E. of 4 -6 independent experiments. C and D, subcellular distribution of activated ERK1/2 after PACAP stimulation. KRBH-quiescent cells were incubated with PACAP (100 nM) for 5 min. C, for subcellular fractionation, cytosolic and nuclear fractions were prepared as described under "Experimental Procedures." Equal amounts of nuclear and cytosolic fractions lysates (50 g) were separated by SDS-PAGE and probed for p-ERK1/2 or total ERK1/2. CREB and ␣/␤-tubulin were used as controls of nuclear and cytosolic fraction purity, respectively. Representative immunoblots of five independent experiments are shown. D, for immunofluorescence studies, cells were fixed, permeabilized, and immunostained for p-ERK1/2, and Hoechst was used to identify nuclear structure as described under "Experimental Procedures." Scale bar ϭ 10 m.
We and others have previously demonstrated that PAC 1 R induces cAMP production through AC activation in pancreatic islets cells (4,5,33), which in turn can lead to PKA activation. Because the cAMP/PKA pathway has been shown to activate the ERK1/2 signaling cascade in ␤-cells (13-16), we first attempted to evaluate the contribution of the G␣s/cAMP/PKA pathway in PACAP activation of ERK1/2 by using the PKA inhibitor H89. Western blot analysis with antibodies against phospho-PKA substrates showed that pretreatment with 10 M but not with 3 M was necessary to completely inhibit PACAP activation of PKA ( Fig. 2A). Moreover, 10 M H89 reduced PACAP activation of ERK1/2 by only ϳ50% (p Ͻ 0.001) after 5 min of stimulation (Fig. 2B). A similar inhibition was recorded with 3 M 4-cyano-3-methylisoquinoline, another selective PKA inhibitor (34). In addition, the specificity of H89 for PKA was demonstrated by the inability of 10 M H89 to alter EGFinduced ERK1/2 phosphorylation (data not shown). The partial suppression of the PACAP-induced ERK1/2 activation by the PKA inhibitors clearly indicates that PACAP activation of ERK1/2 requires PKA but suggests also the involvement of additional pathway(s). PAC 1 Rs have been shown to preferentially interact with G␣ s , but several studies have suggested that they could also interact with G␣ q (8). However, in line with the observation that PAC 1 Rs are not coupled to PLC in rat and mouse pancreatic islets (5), PACAP activation of ERK1/2 activity did not depend on the G␣ q /PLC/PKC pathway. Indeed, PACAP failed to increase inositol phosphate production, and pretreatment with GF109203X (10 M), a broad spectrum PKC inhibitor, had no significant effect on PACAP activation of ERK1/2 in INS-1E cells (data not shown).
Furthermore, several G␣ s -coupled receptors, notably the ␤ 2 -adrenergic receptor (35), were shown to promote ERK1/2 activation following a switch of coupling from G␣ s to G␣ i/o . Nevertheless, selective inhibition of G␣ i/o signaling by overnight treatment with pertussis toxin did not inhibit ERK1/2 activation by PACAP (supplemental Fig. S2, A and B). As a control, pertussis toxin efficiently blocked the inhibition of glucose-stimulated ERK1/2 activation by the G␣ i -coupled ␣ 2 -adrenergic receptor agonist UK14304 as recently reported (25). Taken together, these results suggest that the PAC 1 R mediates ERK1/2 phosphorylation in a G␣ s /PKA-but not a G␣ q -or a G␣ i/o -dependent manner.
Receptor tyrosine kinases and non-receptor tyrosine kinases also have been implicated in the ERK1/2 activation by GPCRs (12,19). We therefore explored whether activation of Src, a non-receptor tyrosine kinase, represented a critical link between PAC 1 R activation and ERK1/2 phosphorylation. Pretreatment with 10 M PP2, a selective inhibitor of the Src family of tyrosine kinases, reduced PACAP-induced ERK1/2 phosphorylation by ϳ50% (p Ͻ 0.01) (Fig. 2, C and D). This inhibition was specific, because pretreatment with PP3, a negative control for PP2, had no effect. Src can activate ERK1/2 by several mechanisms, notably by transactivation of EGFR (12,19). Here, EGFR transactivation was not involved, because PAC 1 R activation was insensitive to the EGFR intrinsic kinase inhibitor AG1478, which completely abolished EGF effects on ERK phosphorylation in parallel control experiments (supplemental Fig.  S2, C and D). Remarkably, PACAP activation of ERK1/2 was totally abolished by the co-pretreatment with H89 and PP2 (Fig.  2, E and F), indicating that PAC 1 R transiently stimulates ERK1/2 phosphorylation through both a PKA and a tyrosine kinase Src-dependent pathways.
PACAP Potentiates Glucose-induced Cytosolic Long-lasting ERK1/2 Activation-Glucose is the most potent physiological stimulus for pancreatic ␤-cells and is known to activate ERK1/2 at normal physiological concentrations (27). Therefore, we investigated ERK1/2 activation by PACAP (100 nM) in the presence of 8.3 mM glucose, a stimulatory concentration (Fig. 3, A  and B). At this concentration glucose induced a progressive and sustained stimulation of ERK1/2 as previously reported (17,18). In contrast to the rapid and transient activation observed in the absence of glucose, PACAP caused, in the presence of glucose, a biphasic ERK1/2 activation with the early peak at 5 min (ϳ4.4fold increase) followed by an additional sustained second phase lasting at least 3 h (ϳ1.8-fold increase). On the other hand, VIP remained ineffective. Thus, these data show that PACAP through PAC 1 R potentiates the long-lasting ERK1/2 activation by glucose. To investigate the mechanism involved in the Glucose induced a sustained increase of p-ERK1/2 in both cytosolic and nuclear fractions (Fig. 3C), as well as a translocation of a fraction of p-ERK1/2 in the nucleus as previously reported (16,23,24). PACAP amplified the glucose-induced p-ERK1/2 in both cytosolic and nuclear fractions at 5 min. In contrast, at 30 min, PACAP potentiated the glucose stimulation only in the cytosol. Similarly, at 30 min, PACAP mainly increased glucose-induced p-ERK1/2 immunofluorescence in the cytoplasm (Fig. 3D). Thus, PACAP alone induces both cytosolic and nuclear transient ERK1/2 activation, whereas in the presence of glucose, it potentiates the cytosolic and long-lasting ERK1/2 activation by glucose.
PACAP Potentiates the Amplifying Pathway of Glucose Involved in Long-lasting ERK1/2 Activation-Glucose exerts its control on ␤-cells via two major, hierarchical signaling pathways (22): the triggering pathway (K ϩ -ATP channel-dependent Ca 2ϩ influx with rise in [Ca 2ϩ ] i ) and the amplifying path-way (K ϩ -ATP channel-independent without further increase of [Ca 2ϩ ] i ). In pancreatic ␤-cells, glucose-mediated ERK1/2 activation depends on glucose metabolism and subsequently on Ca 2ϩ influx via VDCCs (15,17,18,23,24). Indeed, pretreatment with nifedipine, a VDCC blocker (Fig. 4, A and B) decreased glucose-induced ERK1/2 activation at both 5 and 30 min. Nifedipine markedly reduced PACAP potentiation of glucose-induced ERK1/2 activation at 30 min (ϳ70%, p Ͻ 0.01), whereas only slightly inhibited the PACAP response at 5 min (ϳ25%, p Ͻ 0.05). These results suggest that Ca 2ϩ influx generated by glucose metabolism (which corresponds to the triggering pathway) is required for glucose-induced ERK1/2 activation and, subsequently, for PACAP potentiation of the long-lasting ERK1/2 activation. In contrast, the early and transient ERK1/2 activation by PACAP is scarcely dependent on Ca 2ϩ influx in agreement with the results we obtained with PACAP alone (supplemental Fig. S1, C and D).
We then tested the possible involvement of the amplifying pathway of glucose. To this aim, we used one classic experimental approach that consists to hold K ϩ -ATP channels open with diazoxide and to depolarize ␤-cells with KCl to induce Ca 2ϩ influx and a rise in [Ca 2ϩ ] i (22, 36). Under these conditions diazoxide prevents the K ϩ -ATP closure by glucose and therefore its triggering pathway. Treatment with 30 mM KCl and 250 M diazoxide induced only an early (5 min) increase in ERK1/2 phosphorylation (Fig.  4, C and D). Addition of 8.3 mM glucose did not significantly affect the early stimulation but in contrast, caused a clear ERK1/2 activation at 30 min (lane a versus b: ϳ3-fold increase, p Ͻ 0.001). PACAP caused early ERK1/2 activation independently of the presence of glucose, whereas the late sustained potentiation occurred only in the presence of the sugar (lane a versus c: ϳ2-fold increase of glucose response, p Ͻ 0.01, Fig. 4,  C and D). These results show that the glucose-induced longlasting ERK1/2 activation requires both the triggering pathway (i.e. a rise in [Ca 2ϩ ] i ), which is necessary but not sufficient, and the amplifying pathway. In addition, PACAP potentiates the amplifying pathway of glucose involved in long-lasting ERK1/2 activation.
␤-Arrestin 1 Is Required for PACAP to Potentiate Glucoseinduced Long-lasting ERK1/2 Activation-It is well documented that activation of ERK1/2 by several GPCRs can occur . C, for subcellular fractionation, cytosolic and nuclear fractions were prepared as described under "Experimental Procedures." Equal amounts of nuclear and cytosolic fractions lysates (50 g) were separated by SDS-PAGE and probed for p-ERK1/2 or total ERK1/2. CREB and ␣/␤-tubulin were used as controls of nuclear and cytosolic fraction purity, respectively. Representative immunoblots of five independent experiments are shown. D, for immunofluorescence studies, cells were fixed, permeabilized, and immunostained for p-ERK1/2, and Hoechst was used to identify nuclear structure as described under "Experimental Procedures." Scale bar ϭ 10 m.
Biological Relevance of ␤-Arrestin 1 Involvement in PACAP Potentiation of Glucose-induced Long-lasting ERK1/2 Activation in Mouse Pancreatic Islets-To strengthen our observations from the INS-1E cell line, experiments were also performed in pancreatic islets isolated from WT and ␤-arr1 KO mice. As shown in Fig. 6, raising glucose concentration from 1.1 to 16.7 mM induced at 30 min a clear stimulation of ERK1/2 (ϳ2-fold increase) in both WT and ␤-arr1 KO mouse islets. PACAP (100 nM) amplified the glucose-induced p-ERK1/2 (ϳ1.7-fold increase, p Ͻ 0.01) in WT mouse islets but had no effect in ␤-arr1 KO mouse islets. These results confirm the data obtained in the INS-1E cell line and give evidence that PACAP potentiation of glucose-induced long-lasting ERK1/2 activation via a ␤-arrestin 1-dependent pathway is physiologically relevant.
Functional Role of ␤-arr1-dependent Signaling in PACAP-induced IRS-2 Expression-In pancreatic ␤-cells, the transcription factor CREB can be activated by glucose and GLP-1 either directly through PKA or indirectly through ERK1/2 via p90RSK (37,38). In addition, it is known that IRS-2 is a CREB target gene, which is crucial for ␤-cell survival (39). Thus, we next investigated whether PACAP could increase CREB activity or IRS-2 mRNA levels via the ERK1/2 signaling pathways using U0126, a specific MEK1/2 inhibitor. To assess the ability of PACAP to enhance CREB transcriptional activity in INS-1E cells, we used a luciferase reporter system to determine CREB activity. As expected, treatment for 6 h with 8.3 mM glucose induced a ϳ9-fold increase in the activation of the CREB reporter luciferase activity and GLP-1 caused a ϳ3.5-fold amplification of this response (Fig. 7A). PACAP, similarly to GLP-1, caused an amplification of the glucose response (ϳ3-fold). U0126 (10 M) significantly decreased the response to glucose alone and with either PACAP or GLP-1 by 30% (p Ͻ 0.05), 50% (p Ͻ 0.01), and 45% (p Ͻ 0.01), respectively. These results demonstrate that, in the presence of glucose, PACAP induces CREB transcriptional activity mainly via the ERK1/2 pathway.

DISCUSSION
In pancreatic ␤-cells, PACAP is a parasympathetic neuropeptide, which potentiates insulin secretion in a glucose-dependent manner through PAC 1 Rs and VPACRs (33). These receptors are GPCRs mainly coupled to G s protein and AC activation resulting in strong activation of the cAMP/PKA signaling pathway. Several members of the glucagon/GLP-1 superfamily, including PACAP, have been shown to activate ERK1/2 in ␤-cells, but the mechanisms involved remained unclear. In this study, we provide novel information on the mechanism of PACAP action in pancreatic ␤-cells that highlights interactions between glucose, PAC 1 R, and ␤-arrestin. We demonstrate that, in the rat pancreatic cell line INS-1E, PACAP via PAC 1 R activates ERK1/2 by two temporally and spatially separate signaling pathways: (i) PACAP alone induces transient ERK1/2 activation through both PKA-and Src-dependent pathways (but ␤-arrestin 1-independent) and nuclear translocation of a fraction of p-ERK1/2; (ii) in the presence of glucose, PACAP potentiates glucose-induced cytosolic long-lasting ERK1/2 activation through a ␤-arrestin 1-dependent pathway, which biological relevance is confirmed in mouse pancreatic islets.
We have previously shown that PACAP and VIP act via both PAC 1 Rs and VPACRs in pancreatic ␤-cells (2,5,40). Both peptides are equipotent in increasing glucose-induced insulin secretion and cAMP production, but PACAP is more efficient than VIP on the latter (5). In the current study, we show that PACAP interacts only with PAC 1 R to stimulate the MEK/ ERK1/2 cascade. In pancreatic ␤-cells, PAC 1 R is highly expressed, and binding studies indicate that PAC 1 R prevails over VPACR (41). The functional role of PAC 1 R has been confirmed in PAC 1 R-deficient mice, and we have shown that PAC 1 Rs are required for an optimal glucose-stimulated insulin secretion (40).
It is acknowledged that GPCRs activate the Raf/MEK1/2/ ERK1/2 cascades through G-protein-and ␤-arrestin-dependent pathways (19 -21). The G-protein-dependent pathway is mediated by classic G-protein-stimulated production of their respective second messenger-dependent kinases (G␣ s via cAMP/PKA, G␣ q via PLC and PKC) or by other pathways, including non-receptor tyrosine kinases such as Src and/or receptor tyrosine kinases such as EGFRs. Because PAC 1 Rs are not coupled to PLC in rodent pancreatic islets (5) and in INS-1E cells (not shown), we rule out any implication of the G␣ q /PLC/ PKC. Indeed, PKC inhibition has no effect on the PACAP-stimulated ERK1/2 activation in INS-1E cells. In contrast to some G␣ s -coupled receptors, notably ␤ 2 -adrenergic receptor (35), the PAC 1 R-induced ERK1/2 activation is also independent from a pertussis toxin-sensitive G␣ i/o protein. On the other hand, we show that besides the G␣ s /cAMP/PKA pathway, PACAP activation of the ERK1/2 cascade also depends on the Src family of protein-tyrosine kinase. The involvement of Src is not due to transactivation of EGFR, because PACAP activation is insensitive to inhibition of the EGFR intrinsic kinase. Src has been shown to associate with either ␤-arrestin recruited to phosphorylated GPCRs by G-protein-coupled receptor kinases (43), or with the ␣ subunit of G s and G i (44), or directly with GPCRs (45). Important roles of Src have been reported in ERK1/2 activation by various GPCRs (46,47) through G-proteins and/or ␤-arrestindependent or -independent mechanisms. In our study, a ␤-arrestin 1dependent mechanism in the Src recruitment could be ruled out because ␤-arr1 depletion by siRNA did not affect PACAP-induced transient ERK1/2 activation (supplemental Fig. S3, A and B). Further studies are therefore required to determine the mechanisms involved in Src recruitment.
Pancreatic ␤-cells are tightly controlled by glucose, which acts via its oxidative metabolism by generating a triggering pathway that consists of K ϩ -ATP channels closure, depolarization, and rise in [Ca 2ϩ ] i , and an amplifying pathway that is characterized by increased efficacy of Ca 2ϩ (22). The amplifying pathway serves the action not only of glucose but also of other stimuli. Whereas the second messengers involved in the triggering pathway are well characterized, the underlying messengers involved in the amplifying pathway are still unknown (48). The fine mechanism of activation of ERK1/2 by glucose is still indeterminate (27). After confirming that glucose causes a slow, progressive, and sustained ERK1/2 phosphorylation (17,18,24), we show that calcium is essential, because this sustained ERK1/2 activation is prevented by inhibition of Ca 2ϩ influx through VDCCs in agreement with previous studies (17,18,23,24). However, we clearly show that calcium is not sufficient, because the sustained glucose action was not reproduced by KCl depolarization (which produces the triggering signal). Finally, we demonstrate that glucose-induced sustained ERK1/2 activation requires the amplifying pathway, because it is observed also during KCl depolarization even when glucose is unable to close K ϩ -ATP channels in the presence of diazoxide. Then we show that PACAP potentiation of glucose-induced long-lasting ERK1/2 activation requires not only Ca 2ϩ influx through VDCCs, as expected, but also the amplifying pathway of glucose. Thus, PACAP clearly potentiated the amplifying pathway of glucose-induced long-lasting ERK1/2 activation.
The kinetics and subcellular localization of activated ERK1/2 are the major factors determining their cellular responses (32). Glucose has been previously shown to activate ERK1/2 in the cytosol and to cause their nuclear accumulation (23,24). Here, subcellular fractionation and immunofluorescence studies show that glucose leads to changes in the spatiotemporal dynamics of PACAP-mediated ERK1/2 activation. Indeed, PACAP induces early, transient ERK1/2 activation and nuclear translocation of a fraction of p-ERK1/2. Conversely, in the presence of glucose, it potentiates glucose-induced sustained ERK1/2 activation mainly in the cytosol. The spatiotemporal regulation of ERK1/2 activated by GPCRs is dependent on the pathway that mediated their activation, and ␤-arrestins appear to play a determinant role in the spatiotemporal feature of ERK1/2 signaling (19 -21). In the classic paradigm, G-proteinmediated ERK1/2 activation is rapid and transient and results in nuclear translocation. In contrast, ␤-arrestin-mediated ERK1/2 activation is slower, sustained, and sequestered in the cytosol. Here we show that ␤-arr1 is required for PACAP potentiation of the long-lasting ERK1/2 activation by glucose in both INS-1E cells and, more interestingly, in mouse pancreatic islets. In contrast, the early (5 min) ERK1/2 activation by PACAP alone or in the presence of glucose is insensitive to ␤-arr1 knockdown. In addition to their role as signaling scaffold proteins, ␤-arrestins were initially identified to be involved in the desensitization of GPCRs by facilitating their internalization by endocytosis. However, in our study ␤-arr1 depletion did not clearly enhance the early (5 min) ERK1/2 activation by PACAP suggesting that ␤-arr1 does not play a major role in the desensitization of PAC 1 R. GPCRs are divided into two classes according to the type of interaction they have with ␤-arrestins (49). Class A receptors exhibit a transient association with ␤-arrestins and dissociate rapidly after internalization. By contrast, class B receptors stay associated with ␤-arrestins after internalization promoting the formation of stable receptor-␤-arrestin-ERK1/2 complexes and thus sustained ERK1/2 activation. The stability of receptor-␤-arrestin complexes is differentially regulated depending on the type of G-protein-coupled receptor kinases that mediates the interaction of ␤-arrestin with the activated GPCRs, on the ubiquitinylation, or on Ser-412 phosphorylation of ␤-arrestin (20,46,50). The potential role of glucose in the control of such devices in the PAC 1 R-mediated sustained ERK1/2 activation should be addressed in future studies.
Finally, we assessed a functional role of ␤-arr1 in PACAP effect on ERK1/2 activation. In pancreatic ␤-cells, in addition to directly activate nuclear transcription factors, ERK1/2 can target numerous cytoplasmic substrates such as p90RSK (51), which after phosphorylation is translocated to the nucleus and subsequently activates transcription factors such as CREB (52). In pancreatic ␤-cells, CREB plays an essential role in the control of ␤-cell survival and growth by glucose or GLP-1 in part by regulating IRS-2 gene expression (39). In addition, we previously showed that ERK1/2 control CREB transcriptional activity (37). Here, we show that PACAP, like GLP-1, potentiates glucose-induced CREB transcriptional activity and increases IRS-2 mRNA largely via the MEKs/ERK1/2 cascade. In addition, both PACAP-and GLP-1-induced IRS-2 protein expression are markedly reduced by ␤-arr1 depletion. Thus, the ␤-arrestin 1-dependent pathway of ERK1/2 activation plays a key role in PAC 1 Rs and GLP-1Rs signaling leading to IRS-2 expression. Interestingly, it has been recently reported that ␤-arr1 modulates GLP-1 functions such as ERK1/2 and CREB activation and insulin secretion, in the pancreatic ␤-cell line INS-1 (53). However, in this study, basal IRS-2 protein level was directly down-regulated by ␤-arr1 silencing through an unknown mechanism. Because IRS-2 plays a pivotal role in the maintenance of mass and function in pancreatic ␤-cells (54,55), the ␤-arrestin 1-dependent pathway of PAC 1 R potentiation of glucose-induced sustained ERK1/2 activation could be relevant in the PAC 1 R action required for a normal glucose-stimulated insulin secretion (40) as well as in the long term protective effects of PACAP on ␤-cell mass reported in various mouse models of diabetes (10,11).
In summary, our study shows that PAC 1 R-mediated ERK1/2 activation is tightly controlled by glucose in the INS-1E cell line and pancreatic islets. So, independently of the presence of glucose, PACAP induces a rapid and transient ERK1/2 activation in both cytosol and nucleus. In contrast, in the presence of glucose, PACAP potentiates the glucose-induced long-lasting ERK1/2 activation in the cytosol. Moreover, we demonstrate that the ␤-arrestin 1-dependent pathway mediates PAC 1 R potentiation of both the cytosolic sustained ERK1/2 activation and IRS-2 expression induced by glucose. This emerging mechanism of GPCRs action, which is dependent on the amplifying pathway of glucose, could be important for the preservation of ␤-cell functional mass and may represent a new therapeutic approach for the treatment of diabetes.