Glucose-dependent Insulinotropic Polypeptide Activates the Raf-Mek1/2-ERK1/2 Module via a Cyclic AMP/cAMP-dependent Protein Kinase/Rap1-mediated Pathway*

The gastrointestinal hormone, glucose-dependent insulinotropic polypeptide (GIP), is one of the most important regulators of insulin secretion following ingestion of a meal. GIP stimulates insulin secretion from the pancreatic β-cell via its G protein-coupled receptor activation of adenylyl cyclase and other signal transduction pathways, but there is little known regarding subsequent protein kinase pathways that are activated. A screening technique was used to determine the relative abundance of 75 protein kinases in CHO-K1 cells expressing the GIP receptor and in two pancreatic β-cell lines (βTC-3 and INS-1 (832/13) cells). This information was used to identify kinases that are potentially regulated following GIP stimulation, with a focus on GIP regulation of the ERK1/2 MAPK pathway. In CHO-K1 cells, GIP induced phosphorylation of Raf-1 (Ser-259), Mek1/2 (Ser-217/Ser-221), ERK1/2 (Thr-202 and Tyr-204), and p90 RSK (Ser-380) in a concentration-dependent manner. Activation of ERK1/2 was maximal at 4 min and was cAMP-dependent protein kinase-dependent and protein kinase C-independent. Studies using a β-cell line (INS-1 clone 832/13) corroborated these findings, and it was also demonstrated that the ERK1/2 module could be activated by GIP in the absence of glucose. Finally, we have shown that GIP regulation of the ERK1/2 module is via Rap1 but does not involve Gβγ subunits nor Src tyrosine kinase, and we propose that cAMP-based regulation occurs via B-Raf in both CHO-K1 and β-cells. These results establish the importance of GIP in the cellular regulation of the ERK1/2 module and identify a role for cAMP in coupling its G protein-coupled receptors to ERK1/2 activity in pancreatic β-cells.

The incretin hormone, glucose-dependent insulinotropic polypeptide (GIP), 1 is a potent stimulator of insulin secretion and an essential component of the enteroinsular axis in mammals (1)(2)(3). GIP transduces its biological effects by interacting with its G protein-coupled receptor (GPCR), a member of the class II GPCR superfamily that includes receptors for glucagon, glucagon-like peptide-1 (GLP-1), secretin, pituitary adenylyl cyclase-activating protein (PACAP), and vasoactive intestinal polypeptide (4 -7). All of these peptide hormone receptors are coupled to the production of cyclic AMP (cAMP) and have been shown to activate the mitogen-activated protein kinases (MAPK) ERK1 and -2 (extracellular signal-regulated kinase) (8 -12). GIP receptor activation has been shown to induce MAPK activity and was only shown recently (13,14) to activate ERK1/2 in a ␤-cell model (INS-1). However, elucidation of the mode of coupling between class II receptors and MAPK activation has been largely unexplored, and no distinct biological function has been definitively linked to its activation in ␤-cells.
At least five distinct mammalian MAPK signaling modules have been defined and include ERK1/2, p38 MAPK, Jun Nterminal kinase/stress-activated protein kinase (JNK/SAPK), ERK3, and ERK5 (BMK 1; big MAPK) (15)(16)(17)(18). These cytoplasmic protein-serine/threonine kinases are cellular regulators of numerous processes including gene transcription, differentiation, and proliferation (18). The core module of the MAPK pathways consists of three kinases; the kinases directly upstream of the MAPK are members of the Mek family, and these are activated by the Mek kinases (Mekk). Originally, the ERK1/2 module was shown to be coupled to growth factor activation of receptor-tyrosine kinases (RTKs). However coupling of GPCRs to this MAPK cascade has now been demonstrated for G s , G i , G q/11 , and G␤␥ classes of G proteins (19,20). The class II G s -coupled receptors for PACAP and vasoactive intestinal polypeptide have been shown to regulate ERK1/2 activity via the cAMP-dependent protein kinase A (PKA) pathway (20). At the time they were conducted, however, these studies could not fully explain the interaction of this pathway with the ERK1/2 module.
Signals conveyed through G␣ s -coupled GPCRs are extremely diverse. Although these receptors generally influence cellular events through their regulation of adenylate cyclase, it is now evident that other proximal signals can mediate regulation of the ERK1/2 module. Depending on the cell type, such receptors have been shown to activate MAPK via PKA, protein kinase C (PKC), phosphatidylinositol 3-kinase, and/or RTKs (20,21). Furthermore, the prototypical GPCR, the ␤ 2 -adrenergic receptor, has even been reported to undergo a switch from G␣ s to G␣ i , with subsequent activation of ERK1/2 regulated by G␤␥ subunits (22,23). Finally, G␣ s has been shown to interact directly with, and activate, Src (24), and this protein-tyrosine kinase can regulate the ERK1/2 module due to its ability to transactivate RTKs and influence the activity of the small G protein Ras (via Grb2 and Sos) (23).
The cascades involving cAMP and MAPK are highly conserved, stringently regulated, and versatile signal transduction pathways essential to numerous physiological functions. The paradigm of the cAMP and PKA pathway now includes additional cAMP effectors such as cAMP GTPase exchange factors, and the small GTPases Rap1 and -2 (26 -29). These effectors form the basis for cAMP interaction with the MAPK module. Initial studies on the interaction of these two pathways documented an inhibitory effect of cAMP on ERK1/2 module activation by growth factors (30), but cAMP has since been shown to stimulate ERK1/2 activity in several cells types including 3T3-F442A adipocytes, ovarian granulosa cells, pituitary and neuronal cells, and the commonly used HEK and CHO-K1 cell lines (30). This ability of cAMP to activate ERK1/2 depends on which isoform of Raf (the Mekk) is expressed in the cell of interest (31). In this manner, cAMP has been shown to regulate either transient or sustained activation of ERK1/2 in the neuronal PC12 cell line (32).
In the present study, we sought to determine the mechanism by which GIP activates ERK1/2. In doing so, we have characterized a GIP receptor-transfected CHO-K1 (rGIP-15) cell line and two ␤-cell lines by screening for the relative expression of 75 different protein kinases. From this, we were able to conclude that at least four MAPK modules are potentially regulated by GIP and that the pancreatic ␤-cell lines, ␤TC-3 and INS-1 (832/13) cells, express 70 and 60% of the protein kinases examined, respectively, at similar levels to the rGIP-15 cells. For the first time, GIP is shown to influence the activity of all known kinases of the ERK1/2 module, including Raf-1, Mek1/2, ERK1/2, and p90 RSK in CHO-K1 cells and in the INS-1 (clone 832/13) ␤-cell line. This regulation is shown to occur via PKA and Rap1, thus identifying the mechanism by which GIP-stimulated cAMP production positively regulates ERK1/2.

EXPERIMENTAL PROCEDURES
Tissue Culture, Cell Transfections, and Materials-Chinese hamster ovary cells (CHO-K1), cultured in Dulbecco's modified Eagle's medium/ Ham's F-12 (Invitrogen) supplemented with 10% newborn calf serum (Cansera, Rexdale, Ontario, Canada), were stably transfected with the wild type rat GIP receptor as described previously (5). The CHO-K1 cell line obtained by pooling clones was termed rGIP-15 and has been shown previously (33) to express receptors at levels similar to high level expressing clones. In experiments targeted at investigating a role for Rap1 signaling, rGIP-15 clones were transiently transfected with plasmid DNA encoding the wild type Rap1b small GTPase or the empty vector (pCGN). Briefly, 80 -90% confluent monolayers in 6-well culture plates (BD Biosciences) were transfected using Lipofect2000 TM (Invitrogen) transfection reagent according to the manufacturer's protocol. Regular growth medium was replaced the following day, prior to cell starving overnight. The Rap1 constructs and vector were a kind gift from Dr. D. Altschuler (University of Pittsburgh), and the C terminus of ␤-adrenergic receptor kinase and the vector (pRK5) were provided by Dr. R. J. Lefkowitz (Duke University). Src constructs (wild type, Y527F, and Src RF) were all donated by Dr. J. Brugge (Harvard University). All transfections were performed concurrently with green fluorescent protein to ensure transfection and approximate efficiency (ϳ40% for CHO-K1 cells). When increasing amounts of construct DNA were transfected, empty vector was added to ensure the same amount of total DNA was used in all cases. Passages 20 -40 of rGIP-15 cells were used in these experiments.
Kinetworks TM KPKS 1.0 Western blotting analysis of the expression of 75 different protein kinases was initially performed by Kinexus Bioinformatics Corp., with 500 g of cell lysate protein following centrifugation (36) (www.kinexus.ca). The Kinetworks TM KPPS 1.1 screen of 31 different phosphorylation sites with phospho-site-specific antibodies was also performed by Kinexus with 300 g of cell lysate protein.
Quantification of the immunoreactive bands on the Kinetwork blots (trace quantity) with ECL detection was performed with a Bio-Rad FluroS Max Imager and Bio-Rad Quantity One software. This led to follow up studies in which the phosphorylation states of Raf-1, Mek1/2, p90 RSK, and Elk-1 were assessed. Phospho-Thr-202, Tyr-204-ERK1 (p-ERK1/2) was purchased from Santa Cruz Biotechnologies, and p-Raf (Ser-259), p-Mek (Ser-217 and Ser-221), p-p90 RSK (Ser-380), and p-Elk-1 (Ser-383) were obtained from Cell Signaling Technology (New England Biolabs). Total ERK1/2 was assessed using a C-terminal targeted antibody from Santa Cruz Biotechnologies. Briefly, cells were harvested and plated into 6-well plates 2 days prior to overnight serum starvation, and subsequent stimulation was performed on day 3. Cells were preincubated for 1 h at 37°C in modified Krebs-Ringer solution prior to the addition of agonists or pharmacological inhibitors. Following the elapsed stimulation period, cells were washed once with ice-cold KRBH and lysed on ice. Pharmacological inhibitors (H89, GF109203x, PD98059, U0126 (Calbiochem)) were added for 15 min prior to agonist addition and maintained in the presence of agonists. Protein samples (50 g of protein/well) were separated on a 13% SDS-polyacrylamide gel and transferred onto nitrocellulose (Bio-Rad) membranes. Probing of the membranes was performed with the above mentioned antibodies, and bands were visualized by enhanced chemiluminescence (Amersham Biosciences) using horseradish peroxidase-conjugated IgG secondary antibodies. For quantification of band density indicative of phosphorylation, films were analyzed using densitometric software (Eagle Eye (Stratagene, La Jolla, CA)).
Data Analysis-Data are expressed as means Ϯ S.E. with the number of individual experiments presented in the figure legends. All data were analyzed using the nonlinear regression analysis program PRISM (GraphPad, San Diego, CA), and significance was tested using the Student's t test and analysis of variance (ANOVA) with the Dunnett's multiple comparison test or the Newman-Keuls post test (p Ͻ 0.05) as indicated in figure legends.

Expression Profiling of Protein Kinases-
The expression levels of protein kinases are cell-specific. Our approach to delineating the signaling pathways responding to GIP receptor activation included an initial assessment of the various protein kinases expressed (Table I-III), with a comparison between a heterologous GIP receptor expression system (rGIP-15) and the two ␤-cell lines (␤TC-3 and INS-1 (832/13)). With respect to MAPK cascades, it is evident that four modules (Raf1/B 3 Mek6 3 p38␣, and ERK3) were all present in rGIP-15, ␤TC-3, and INS-1 cells (Tables I-III; ERK5 was not blotted for). Furthermore, a pathway regulating cell survival (3-phosphoinositide-dependent kinase 1 3 protein kinase B 3 glycogen synthase kinase 3), the multiple isoforms of PKC (␣, ␤, ␦, ⑀, ␥, , , , and ), and the ubiquitous cyclin-dependent kinases (kinases 1, 2, 4 -7, and 9) regulating cell cycle progression are also potential regulatory effector molecules for GIP in the cell lines (Tables I-III).
On comparing the kinase expression profiles of rGIP-15, ␤TC-3, and INS-1 cells, there were few major differences in expression levels of the ubiquitous kinases of the MAPK modules mentioned above. Differences were confined mainly to the relative expression level of the upstream Meks (Mek 1, 2, 4, 6, and 7; Mek3 was not blotted for), the Mekks (Mos, Cot, germinal center kinase, hematopoietic progenitor kinase, MstI), and the numerous PKC isoforms (Table I-III). The most reassuring finding was that 50 -70% of the protein kinases probed for were expressed at similar levels in the three cell lines (48% across all three; 59% for rGIP-15 versus INS-1; 61% for ␤TC-3 versus
GIP Regulates the ERK1/2 Module Upstream of Raf-1 via PKA-GIP has been shown previously (13,14) to activate ERK1/2, but the other kinases in this module were not examined, and no attempt was made to link ERK1/2 activation to GPCR effector coupling. Fig. 4A illustrates the concentrationdependent actions of GIP on phosphorylation of Ser-259 of Raf-1, Ser-217/221 of Mek1/2, and Ser-380 of p90 RSK, after a 5-min stimulation in rGIP-15 cells. There is a high correlation in responses among these three kinases, but we found no evidence for Elk-1 (Ser-383) phosphorylation after 5 min (n ϭ 3). Evidence supporting a role for Raf-1 and p90 RSK in responses to GIP in the ␤-cell (INS-1 clone 832/13) is shown in Fig. 4B, under 0 mM glucose conditions. Under these conditions the state of Ser-217/221 phosphorylation of Mek1/2 was already high, and further enhancement of activity by GIP could not be detected. However, the Mek1/2-specific inhibitors, PD98059 and U0126, both completely abolished forskolin, PMA, and GIP-mediated ERK1/2 phosphorylation, supporting a role for Mek1/2 in the cascade (Fig. 4C). Thus, GIP is able to regulate the Raf1 3 Mek1/2 3 ERK1/2 3 RSK module in both CHO-K1 and INS-1 ␤-cells.
The general PKC and PKA inhibitors, GF109203x (Bis) and H89, respectively, were used to evaluate the role of these protein-serine/threonine kinases in mediating GIP actions on the ERK module (Figs. 5 and 6). These studies were conducted in rGIP-15 cells using forskolin, PMA, and AA as agonists representing three potential activation pathways (PKA, PKC, and AA, respectively). PKC inhibition (2 M GF109203x) reduced basal ERK1/2 phosphorylation as well as significantly abrogating forskolin and PMA-induced ERK1/2 phosphorylation (p Ͻ 0.05, n ϭ 4). The specificity of GF109203x for PKC is demonstrated by the complete ablation of PMA effects. Fig. 5 clearly demonstrates that GIP-mediated activation of ERK1/2 can be PKC-independent.
GIP Activates ERK via Rap1 Independently of G␤␥ and Src-The above data reveal a positive effect of GIP on the Mek1/2 3 ERK1/2 3 RSK module, in contrast to its inhibitory influence on Raf-1 activity through promotion of Ser-259 phosphorylation (40 -42). It was hypothesized that another kinase must be responsible for the positive influence of GIP on ERK1/2 activation. Previous studies (30) have shown that the Raf isoform present in specific cell types determines whether cAMP affects ERK1/2 activation positively or negatively. From the protein kinase profile we were able to confirm the presence of an abundant amount of both Raf-1 and B-Raf in CHO-K1 (rGIP-15), ␤TC-3, and INS-1 (832/13) cells (Tables I-III and Fig. 7). B-Raf has been shown conclusively to regulate cAMP stimulation of ERK1/2 in PC12 cells (31). Thus, we transfected rGIP-15 cells with increasing amounts of the upstream regulator of B-Raf and the small GTPase Rap1 (43) and assessed GIPmediated ERK1/2 activation. Fig. 8A clearly shows that submaximal GIP concentrations were able to exert a greater influence on ERK1/2 phosphorylation in the presence of increased amounts of Rap1. This was without an effect on total ERK1/2 expression levels (Fig. 8A). Thus, we conclude that GIP regulation of the ERK1/2 module occurs upstream of B-Raf via PKA and Rap1.
In view of the fact that the ␤ 2 -adrenergic receptor was shown to switch from G␣ s to G␣ i coupling, and activate ERK1/2 via G␤␥ subunits (22), we investigated whether a similar pathway was involved in regulating downstream ERK1/2 signaling. An intervening peptide (C terminus of ␤-adrenergic receptor kinase) that we previously employed to investigate GIP signaling (38) was expressed in increasing amounts in rGIP-15 cells, but it was unable to reverse 1 nM GIP-mediated activation of ERK1/2 phosphorylation (Fig. 8), thus providing evidence against a role for G␤␥ signaling in ERK1/2 regulation. Recent findings have implicated G␣ s in the activation of the tyrosine kinase Src (24) and Src in the regulation of ERK1/2 (23). We found no evidence for Src involvement in the GIP activation of ERK1/2 using dominant negative and constitutively active constructs (Src Y527F and Src RF; n ϭ 3, data not shown) (44). This is consistent with the absence of detectable Src from our protein kinase profile data (Table III). DISCUSSION Several members of the glucagon/PACAP superfamily of peptide hormones have been identified as regulators of MAPK modules. PACAP, glucagon, and the incretin GLP-1 have been shown to positively regulate ERK1/2 in neuronal PC12 cells (8), HEK cells (9), and in ␤ and non-␤ cells, respectively (11,45). Both PACAP-and glucagon-mediated cAMP elevations were found to regulate these protein-serine/threonine kinases. Until recently, however, the role of cAMP in ERK1/2 regulation was controversial with some studies demonstrating positive effects and others negative effects. A recent study (9) on the glucagon receptor provided the first insight into how this superfamily of peptide hormones may regulate ERK1/2 activity in a cAMP-PKA-dependent manner. In the present study, we demonstrate that GIP can activate the ERK1/2 module via proximal cAMP-PKA-Rap1 activation, and we propose a role for B-Raf in the positive regulation of this module.
In attempting to delineate GIP receptor signaling pathways in CHO and ␤-cells, we have begun to map the expression of a wider range of protein kinases (Tables I-III). Although various protein kinases have been identified in pancreatic ␤-cells, few have been identified in a single cell simultaneously (46). We confirmed the presence of four intact MAPK modules (Raf1/B 3 Mek1/2 3 ERK1/2, Cot/Tpl-2 3 Mek4/7 3 SAPK, PAK␣ 3 Mek63 p38␣, and ERK3) in rGIP-15, ␤TC-3, and INS-1 (832/ 13) cells (Tables I-III). We did not assess the levels of the p38 MAPK activator Mek3, because we could not identify a reliable antibody that was commercially available. However, another p38 MAPK activator Mek6 was clearly detected in the ␤TC-3 and INS-1 (832/13) cells. By assessing the relative expression levels of the identified kinases, it is possible to generate predictions on regulatory pathways that may be particularly important in various cell types, and to select established cell lines that more closely resemble primary cells in the architecture of their signaling networks.
In most cell types, cAMP inhibits cell growth, and initial evidence detailing the molecular basis for this effect focused on cAMP inhibition of ERK1/2 activity. This was mapped to occur at the level of Raf-1, via PKA-mediated phosphorylation of Ser-43 (47), Ser-259 (48), and Ser-621 (49), resulting in enzyme inhibition. However, recent studies (30) have now highlighted a role for cAMP in positively regulating ERK1/2 activity in pituitary, ovarian, and neuronal cells. The basis for this interaction has been most extensively studied in the rat pheochromocytoma cell line, PC12 cells. These studies have culminated in a model where cAMP activates ERK1/2 when B-Raf is expressed in cells (30). Unlike Raf-1, which undergoes complex protein interactions to regulate its activity, B-Raf is mainly regulated by small GTPase proteins (40). The key regulator is the small G protein, Rap1, which is directly activated by elevated cAMP supports the notion that cAMP-mediated signaling via GIP can positively influence ERK1/2 activity.
We propose, therefore, that GIP stimulates ERK1/2 by influencing B-Raf activity in pancreatic ␤-cells. In support of this hypothesis, other GPCRs have also been shown to regulate ERK1/2 activity via B-Raf and Rap1. For example, the adenosine receptor (A 2A ), the prototypical ␤ 2 -adrenergic receptor, and the M 1 muscarinic receptor have all been found to stimulate ERK1/2 via Rap1 and B-Raf in CHO, HEK, and PC12 cells, respectively (51)(52)(53). These cells all express the Raf isoform B-Raf and are therefore able to activate ERK1/2 via cAMP signaling. Since Rap1 is able to increase GIP-mediated ERK1/2 activity, it is likely that PKA-dependent activation of Rap1 is able to regulate B-Raf and thereby activate Mek1/2 3 ERK1/2 3 p90RSK.
Neuronal and endocrine cells share many similar phenotypic features. Therefore, it may be expected that cAMP can positively influence ERK1/2 activity and cellular fate in both cell types. cAMP regulation of ERK1/2 activity in CHO cells has been controversial, with studies showing both activating and inhibitory affects (51,54,55). A previous report (13) investi-gating GIP signaling demonstrated an inhibitory effect of cAMP on ERK1/2 activity in CHO cells. In the present study, however, we report that cAMP is able to influence positively the ERK1/2 module in both CHO (rGIP-15) and ␤-cells (INS-1 these cell lines, together with the similar expression level of protein kinases profiled in our cell model versus ␤-cell lines, supports our model system as a means for elucidating GIP receptor signaling. It may seem counterintuitive for GIP to inhibit Raf-1 activity, while positively regulating downstream ERK1/2 via Rap1 activation (Figs. 5 and 7). However, similar results have been found for neuronal growth factor and ␤ 2 -adrenergic receptor signaling (32,52). The balance between Raf-1/B-Raf activation or inhibition represents a mechanism by which downstream kinetics of ERK1/2 activity may be regulated, thereby affording different physiological processes. For example, both neuronal growth factor and the ␤ 2 -adrenergic receptor are able to activate Ras and ERK1/2 rapidly and transiently via Raf-1 (32,52). However, sustained (prolonged) activation of the ERK1/2 module occurs through B-Raf in these systems. This attractive model may explain the molecular basis responsible for regulating differentiation (sustained ERK1/2 activation) versus proliferation (transient ERK1/2 activation) in PC12 cells. Although we provide evidence for extremely rapid kinetics of ERK1/2 activation, there may be prolonged effects mediated by GIP via Raf-1 in ␤-cells, which still need to be elucidated. Whereas GIP is able to regulate ␤-cell proliferation (14,56), the second incretin hormone, GLP-1, has been identified recently as both a proliferation and differentiation factor in ␤-cells and their precursor ductal cells (25,45,57,58). Thus, one could envision a similar control of cellular fate by the incretins at the level of the pancreas, as is the case for neuronal growth factor and neurons.
The GIP-stimulated cAMP-PKA-Rap1-ERK pathway identified here may be an important mechanism by which GIP regulates cellular proliferation/differentiation and/or gene transcription events in pancreatic ␤-cells. In elucidating this cascade, we have also identified ␤-cells as another example where cAMP can couple to activation of the ERK1/2 module. Extrapancreatic GIP target tissues (e.g. adipose tissue or skeletal muscle), which express B-Raf, may also present themselves as systems where cAMP signaling positively influences ERK module activation. Interestingly, the antagonism of Raf-1 by GIP may represent an important balance by which GIP can regulate transient and/or sustained ERK1/2 module activation.
Together with recent data identifying GIP as a growth factor for ␤-cells, our findings support a role for GIP in regulating the molecular events responsible for such biological functions.