Activation of Serum Response Factor in the Depolarization Induction of Egr-1 Transcription in Pancreatic Islet β-Cells*

The results of the current studies define the major elements whereby glucose metabolism in islet β-cells leads to transcriptional activation of an early response gene in insulinoma cell lines and in rat islets. Glucose stimulation (2–20 mm) resulted in a 4-fold increase in Egr-1 mRNA at 30 min, as did the depolarizing agents KCl and tolbutamide. This response was inhibited by diazoxide and EGTA, indicating that β-cell depolarization and Ca2+ influx, respectively, are essential. Pharmacological inhibition of the Egr-1 induction by H89 (48%) and calmidazolium (35%), but not by mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1 and 2 or phosphatidylinositol 3-kinase inhibitors, implied that protein kinase A and Ca2+/calmodulin pathways are involved. Deletion mapping of the Egr-1 promoter revealed that the proximal −198 base pairs containing two serum response elements (SREs) and one cAMP-response element retained the depolarization response. Depolarization resulted in phosphorylation of cAMP-response element-binding protein, yet partial inhibition by a dominant negative cAMP-response element-binding protein, along with a robust response of a cAMP-response element-mutated Egr-1 promoter suggested the presence of a second Ca2+-responsive element. Depolarization activation of 5XSRE-LUC and serum response factor (SRF)-GAL4 constructs, along with activation of SRF-GAL4 by co-transfection with constitutively active calmodulin kinase IV and protein kinase A, and binding of Ser103-phosphorylated SRF in nuclear extracts, indicated that the SRE·SRF complexes contribute to the Ca2+-mediated transcriptional regulation of Egr-1. The results of the current experiments demonstrate for the first time SRE-dependent transcription and the role of SRF, a transcription factor known to be a major component of growth responses, in glucose-mediated transcriptional regulation in insulinoma cells.

Pancreatic islet ␤-cell mass can be regulated by multiple stimuli, including growth factors and nutrients. The ability to produce insulin is proportional to the islet mass that is defined by a balance between rates of proliferation and apoptosis. Several growth factors have been shown to stimulate pancreatic ␤-cell replication (1,2). During pregnancy, insulin resistance results in a marked increase in ␤-cell mass, and growth hormone and prolactin treatment have been shown to increase ␤-cell proliferation (3)(4)(5). Other growth factors, including insulin-like growth factor 1, hepatocyte growth factor, and nerve growth factor, have also been implicated in ␤-cell mitogenesis and differentiation (1, 6 -8). Nutrients including glucose and certain essential amino acids have been shown to increase ␤-cell proliferation and increase islet ␤-cell mass (1). In insulin resistance states, such as obesity for example, impaired insulin-mediated glucose disposal results in increased plasma glucose levels, and this mechanism is thought to be essential for the hyperinsulinemic response of ␤-cells.
Glucose induces pleiotropic effects in islet ␤-cells, and these effects are potentially mediated by activation of multiple intracellular signaling pathways. Glucose metabolism results in the alteration of ATP/ADP levels leading to closure of nucleotide regulated K ϩ channels (KATP). KATP channel inhibition results in depolarization and activation of voltage-dependent Ca 2ϩ channels. Ca 2ϩ influx has been implicated in the activation of calmodulin (CaM) 1 kinases (CamKs) II and IV, protein kinases A (PKA) and C, extracellular signal-regulated kinase mitogen-activated protein kinase, and phosphatidylinositol 3-kinase in islet ␤-cells (9 -12). The relationships between activation of these pathways and growth responses induced by glucose have not been defined.
Changes in gene expression profiles that result from the activation of these signaling pathways are responsible for the adaptation of ␤-cell mass to physiological and pathological states. The specific expression profile includes the induction of immediate early genes (IEGs). IEGs are rapidly activated by initiation of signal transduction pathways, and their activation does not require protein synthesis (13). The signaling pathways for induction of IEGs exhibit considerable stimulus and tissue specificity and in general involve activation of kinase/phosphatase cascades (13). Often, IEGs are transcription factors that in turn activate expression of downstream target genes generating distinct biological responses by inducing specific long term programs of gene expression (13). The IEG responses are not restricted to genes that control cell cycle reentry but also include genes that participate in other biological processes. IEGs are universally expressed, and the signal transduction pathways leading to activation of these genes have been extensively studied in numerous tissues.
Glucose and other nutrients have been shown to transcriptionally activate several IEGs, including Egr-1 in insulinoma cell lines and pancreatic islets, in a Ca 2ϩ -dependent manner (14,15). Egr-1 is a zinc-finger DNA binding transcription factor that is expressed in multiple tissues and is induced by diverse stimuli of which the best studied examples have been the treatment with peptide mitogens (16 -18). Egr-1 mRNA has been shown to be increased by treatment with glucose, cAMP, and several glucoincretins in insulinoma cells and rat islets (14,15). The promoter elements involved in Egr-1 transcription by depolarization have not been defined. The genomic 5Ј-flanking sequence of the murine Egr-1 contains activator protein-1 (AP-1) sites, a cAMP-responsive element (CRE), and serum response elements (SRE). Of these elements only the CRE has been implicated in Ca 2ϩ induction of the glucagon gene in pancreatic ␤-cells (19). CRE was also essential for the induction of c-fos by the combination of glucose and glucoincretins in INS-1 cells (19). To our knowledge, nothing is known about SRE-dependent transcription of any gene in pancreatic islet ␤-cells.
In the current studies, we sought to define the mechanisms whereby glucose-induced depolarization and Ca 2ϩ influx activates Egr-1 transcription in insulinoma cells. We describe the signal transduction pathways, trans-activating factors, and the cis-regulatory elements involved in this process. We demonstrate that ␤-cell depolarization triggers a cascade of events in which Ca 2ϩ influx leads to the activation of PKA and Ca 2ϩ / CaM. The initiation of these signaling pathways leads to activation of CREB and SRF, a transcription factor involved in growth responses. This activation results in Egr-1 induction predominantly through five SRE elements in its proximal promoter. These studies for the first time link the potential role of SRE-dependent transcription as a mechanism for glucose effect in pancreatic ␤-cell mass.
Cell Culture Conditions-The MIN6 insulinoma cell line was obtained from Y. Oka (20) and was maintained in Dulbecco's modified Eagle's medium (DMEM) containing 25 mM glucose and supplemented with 15% heat-inactivated fetal bovine serum (FBS) in humidified 5% CO 2 , 95% air at 37°C. ␤TC6-F7 insulinoma cells, obtained from S. Efrat (Albert Einstein College of Medicine) were cultured in DMEM containing 25 mM glucose and supplemented with 15% horse serum and 2.5% FBS. ␤TC6-F7 and MIN6 cells were used between passages 16 -42 and 26 -35, respectively. For the analysis of expression, cells that were 70 -80% confluent were switched to regular DMEM containing 2 mM glucose for 20 -24 h (time zero of the experiments). The indicated agent was added directly to the medium, and cells were incubated as described in the figure legends followed by a 45-min incubation in the same medium containing the indicated agents. Pharmacologic inhibitors were added 60 min prior to stimulation of the cells with glucose or other agents, using the concentrations of inhibitors given in the figure legends.
Northern Blot Analysis and Reverse Transcriptase-PCR-Total cellular RNA was isolated by the acid guanidium thiocyanate-phenolchloroform method of Chomczynski and Sacchi (21). Twenty g of total RNA were fractionated by electrophoresis through a 1.1% agarose gel and transferred to a nylon membrane by the capillary blotting technique. RNA was cross-linked by UV irradiation. The cDNA probes were labeled with [ 32 P]dCTP using the High-Prime Labeling System (Roche Molecular Biochemicals). Prehybridization (2-4 h) and hybridization were carried out at 56°C in Church solution (22). Hybridized blots were washed once with 0.1ϫ SSC and 1% SDS at room temperature for 15 min and twice at 54°C for 15 min. The membranes were exposed overnight at Ϫ80°C using intensifying screens. Messenger RNA levels were quantitated by PhosphorImager scanning (Molecular Dynamics) and were normalized to those of ␤-actin. For reverse transcriptase-PCR, total RNA was reverse transcribed using the Superscript Preamplification System and random primers (Life Technologies, Inc.). PCR was done according to the manufacturer's protocol with addition of [ 32 P] dCTP to the PCR mixture. PCR products were separated on 5% polyacrylamide gels and exposed to x-ray films. Results were normalized to those of ␤-actin. The primers for Egr-1, Egr-3, and NGFI-C have been described previously (23,24).
Islets of Langerhans mRNA Analysis-Primary islets of Langerhans were isolated from adult male Harlan Sprague-Dawley rats by standard collagenase digestion (25). Pancreatic islets were cultured in RPMI medium containing 5 mM glucose for 48 h and stimulated by adding glucose to a final concentration of 25 mM for 45 min.
Plasmids-Plasmids containing the rat Egr-1, Egr-2, Egr-3, and NGFI-C cDNAs were kindly provided by Dr. J. D. Milbrandt (Washington University School of Medicine). The plasmids containing the murine Egr-1 promoter linked to the luciferase vector pXP2 constructs were generously provided by Dr. David Cohen (Oregon Health Sciences University) and have been described in detail previously (26). The pmEgr-1200 contains 1.2 kilobase of 5Ј-flanking region of the mouse Egr-1 gene (GenBank TM accession number X12617). The pmEgr-510 construct included sequences Ϫ511 to Ϫ1013 bp; the pmEgrAP1 construct contains nucleotides Ϫ1 to Ϫ386 plus Ϫ886 to Ϫ1000 (from sequence X12617). Plasmids containing nucleotides Ϫ532 to ϩ100 (prEgr-532), Ϫ198 to ϩ100 (prEgr-198), and Ϫ109 to ϩ100 (prEgr-109) of the rat gene subcloned into the luciferase reporter plasmid pGL2 (Promega) were obtained from J. Milbrandt. The cis-reporter plasmid pSRE-luc (5XSREluc) contains the luciferase reporter gene driven by a basic promoter element (TATA box) joined to five tandem repeats of the c-fos SRE and CRE, respectively (Stratagene). The CREB-GAL4 system includes a trans-activator plasmid that expresses a fusion protein containing the activator domain of CREB fused to the DNA binding domain of GAL4 (residues 1-147) (CREB-GAL4) (Stratagene), cotransfected with a reporter gene containing a synthetic promoter with five tandem repeats of the yeast GAL4 binding sites that control expression of the luciferase gene pRC-LUC (Stratagene). The SRF-GAL4 fusion vector and a constitutively active form of CaMKIV (CaMKIVCT) were a gift from Dr. Michael Greenberg and have been described previously (27). The plasmid containing the catalytic unit of PKA (pFC-PKA) was purchased from Stratagene. The killer CREB (KCREB) construct was kindly provided by Dr. Richard Goodman and has been described previously (28). Briefly, KCREB contains an amino acid substitution within the leucine zipper that results in KCREB-CREB heterodimers with no DNA binding function. The pRL-TK control vector contains the herpes simplex virus thymidine kinase promoter upstream of the Renilla luciferase (Promega).
Transient Transfections-MIN6 cells were transfected by Lipo-fectAMINE and Plus Reagent (Life Technologies, Inc.) using the suggested amounts of DNA according to the manufacturer's protocol. Briefly, 1 ϫ 10 6 cells were plated in 6-well plates 4 days prior to transfection. Cells at about 80% confluence were transfected by mixing the indicated amount of DNA described in the figure legends, and a lipid mixture containing a 1:2 ratio of LipofectAMINE and Plus Reagent in 1 ml of OPTI-MEM medium (Life Technologies, Inc.). Following 4 h of incubation, 1 ml of DMEM containing 5 mM glucose and 2% serum was added to the cells. After 12 h, the medium⅐DNA complexes were replaced by preincubation medium containing DMEM with 5 mM glucose, 2% FBS, and the cells were left for 24 h. At the end of the 24 h the specific stimulating agent was added to the medium, and the cells were harvested 6 h later. For the overexpression experiments with pFC-PKA and CaMKIVCT, MIN6 cells were transfected as above, followed by incubation in DMEM containing 5 mM glucose and 2% FBS for 30 h until harvesting. Total DNA was maintained constant in all the transfection experiments by using the empty vector of the respective cDNA to be overexpressed. To correct for differences in transfection efficiencies, 1 ng of pRL-TK Renilla luciferase plasmid was simultaneously transfected. All results are normalized for transfection efficiency and expressed as the ratio of firefly to Renilla luciferase.
Nuclear extracts from MIN6 cells cultured under regular DMEM containing 25 mM glucose were prepared essentially as described previously (29), and protein content was determined using the Bio-Rad protein assay kit with bovine serum albumin as a standard. Lysates (5 g of protein) were incubated in a final volume of 20 l containing binding buffer (final concentrations, 4% glycerol, 1 mM MgCl 2 , 0.5 mM dithiothretinol, 50 mM NaCl, 10 mM Tris-HCl, pH 7.5), 200 ng of poly(dI-dC), and double-stranded ␥-32 P-oligonucleotide. Binding reactions proceeded at room temperature for 30 min and were analyzed by nondenaturing polyacrylamide electrophoresis followed by autoradiography. In assays where antibodies directed against SRF (1:10) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and SRF phosphorylated at Ser 103 (1:10) were tested, these were preincubated with the nuclear extracts for 20 min at room temperature (30).
Luciferase Assay-Cell lysis was performed using 400 l of passive lysis buffer (Promega). Firefly and Renilla luciferase were measured by the dual luciferase reporter assay system (Promega) using 20 l of cell lysate. Luciferase activity was measured in a Monolight 3010 luminometer.
In Vitro Mutagenesis-Because of the absence of an f1 origin in pmEgr-1200, the pmEgr1.2 64 CRE construct was made by subcloning the 1.2-kilobase insert from pmEgr1.2 into pBluescript SK(ϩ). Mutagenesis was performed in the pBluescript SK(ϩ) construct, and the mutated insert was then subcloned back into the original luciferase reporter vector PXP2. The CRE mutation was performed by oligonucleotide-directed mutagenesis without the phenotypic selection method (31). In brief, single-stranded uracil-containing DNA template was obtained by transforming CJ236 and growing overnight in TY medium with lambda phage followed by polyethylene glycol/NaCl and ethanol precipitation. A kinased oligonucleotide that produced a CRE mutation proven to be effective in other promoters were used: 5Ј-GGA TGG GAG GGC TCA TCA CAA CTC CGG GTC CTC (mutated nucleotides are underlined) (32). Double-stranded template was obtained by primer extension and T4 DNA ligase. Finally, Jm109 competent cells were transformed with 1 l of the primer extension reaction. The presence of the mutation was confirmed by sequencing.
Western Blot-Cells were lysed with buffer containing 1ϫ phosphatebuffered saline, 0.1% SDS, 0.01 M dithiothreitol, and half a tablet of "Complete" protease inhibitor mixture (Roche Molecular Biochemicals). After boiling, proteins were separated by electrophoresis though 10% polyacrylamide, 0.1% SDS gels and transferred to polyvinylidene difluoride membranes. Membranes were incubated overnight at room temperature in blocking buffer containing 0.2% I-Block (Tropix) and 1:1000 Tween 20. Subsequently, the membranes were hybridized at 4°C overnight in blocking buffer containing the CREB and the phospho-CREB antibody with the dilutions recommended by the manufacturer (New England Biolabs). After three washes at room temperature the membranes were incubated in secondary horseradish peroxidase antibody for 1 h. After washing for 1 h, immunodetection was performed with ECL Western blotting Detection System (Amersham Pharmacia Biotech) following the manufacturer's protocol.

Induction of Egr-1 mRNA by Glucose in ␤TC6-F7 and MIN6
Insulinoma Cell Lines and in Rat Islets-To establish a model system to study the transcriptional regulation of Egr-1 in pancreatic ␤-cells, insulinoma cell lines that have been shown to secrete insulin in response to glucose in the physiological range were investigated (20). Following preincubation of ␤TC6-F7 cells for 20 h in 2 mM glucose, the addition of glucose to a final concentration of 25 mM resulted in increased Egr-1 mRNA levels between 30 and 60 min (4-fold, p Ͻ 0.01) that returned to basal levels by 2 h (Fig. 1A). Similar responses to glucose stimulation were also noted in MIN6 insulinoma cells and in primary cultures of isolated rat pancreatic islets (data not shown), and these responses were shown to be glucose dose-dependent (Fig. 1B). The half-maximal response of Egr-1 transcription is in the same range as that observed for insulin secretion in this cell line (20). Egr-2 (NGFI-B or Krox 20) mRNA, but not Egr-3 mRNA, was also induced by glucose under the same conditions (data not shown). Mannose, the metabolizable epimer in position 2 of glucose, increased expression of Egr-1 mRNA, whereas nonmetabolizable sugars 2-deoxyglucose or 3-O-methylglucose had no effect (data not shown). These results are similar to those previously reported and confirm that glucose metabolism is required for Egr-1 induction (15).
Glucose metabolism is known to have several effects, the major one being depolarization through inhibition of KATP channels. Egr-1 mRNA was induced in insulinoma cells by tolbutamide and KCl, agents known to depolarize islet ␤-cells, with the same time course as that following glucose stimulation (data not shown), and this induction was Ca 2ϩ -dependent (Fig. 1C). To determine whether glucose induction of Egr-1 resulted solely from inactivation of KATP channels or additionally through increased glycolytic intermediates insulinoma cells (␤TC6-F7) were treated with glucose in the presence of diazoxide, an agent that hyperpolarizes the cell through activation of islet KATP channels. Treatment with this agent resulted in complete inhibition of Egr-1 mRNA induction (Fig.  1D), indicating that this response to glucose was solely dependent on depolarization and subsequent Ca 2ϩ influx. Activation of PKA by cAMP has been shown to be one of the major regulators of c-fos transcription in INS-1 insulinoma cells (19). In the current experiments, to evaluate the effect of PKA activation on Egr-1, MIN6 cells were treated with forskolin. This treatment resulted in similar induction of Egr-1 mRNA, and the response to both stimuli was greater than that to either alone (Fig. 1D).
Effects of Pharmacological Agents on Glucose Induction of Egr-1 mRNA Expression-Specific pharmacological inhibitors were utilized to begin to define the signal transduction pathways involved in glucose and depolarization induction of Egr-1. The CaM and PKA pathways were assessed by the addition of the CaM inhibitor calmidazolium (10 M) or the PKA inhibitor H89 (20 M) (33,34). Calmidazolium and H89 independently reduced the glucose induction of Egr-1 mRNA to 70% (p Ͻ 0.01) and to 56% (p Ͻ 0.01), respectively (Table I). Simultaneous inhibition of the PKA and CaM pathways resulted in a combined suppression to 40% of control (p Ͻ 0.001). Glucose has been shown to activate the extracellular signal-regulated kinase 1/extracellular signal-regulated kinase 2 mitogen-activated protein kinase pathway in a Ca 2ϩ -dependent manner (9,10). The specific mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1 and 2 inhibitor PD98059 (100 M) used at concentrations known to inhibit these enzymes in other systems had no significant effect on Egr-1 mRNA induction by glucose (34 -37) (Table I). Staurosporin, a known inhibitor of protein kinase C, did not inhibit Egr-1 induction. Interestingly, both phosphatidylinositol 3-kinase inhibitors wortmannin and LY294002 resulted in small but significant (p Ͻ 0.01) enhancements of Egr-1 mRNA induction. Taken together, these results suggested that glucose induction of Egr-1 is a Ca 2ϩ -dependent process and that the Ca 2ϩ effects are mediated at least in part by PKA and Ca 2ϩ /CaM pathways.
Elements Present within Ϫ198 bp of the Egr-1 Promoter Maintain Ca 2ϩ Activation-To determine whether Egr-1 induction is mediated through transcriptional activation and whether the proximal promoter contains the elements required for this response, plasmids containing deletions of the 5Ј-flanking region of Egr-1 gene linked to a luciferase reporter were evaluated. The results of transient transfections in MIN6 cells with the indicated constructs is shown in Fig. 2A. The plasmid pmEgr-1200 containing 1200 bp of the 5Ј-flanking sequence of the mouse Egr-1 promoter was induced 4 -5-fold (p Ͻ 0.001) by KCl-stimulated depolarization. Treatment of cells with the phorbol ester PMA, a known activator of SRE-dependent transcription, also resulted in a significant induction (p Ͻ 0.01), similar to the results found by Northern blotting (data not shown). No response was obtained by exogenous insulin treat-ment at concentrations from 1 nM to 1 M (data not shown), suggesting that insulin signaling through the insulin receptor is not involved in depolarization activation. The plasmid pmEgr-510 had the 5Ј-end of the promoter containing the AP-1 elements deleted but still retained five SREs and a CRE. Transfection of this plasmid resulted in diminished basal activity relative to that of pmEgr-1200, suggesting that upstream elements may also play a role in the transcriptional regulation of this gene. This plasmid pmEgr-510 did retain the same fold induction to KCl and PMA, however. Basal expression and KCl and PMA induction, were all abolished when a construct lacking the SRE and CRE elements (pmEgrAP1) was employed. The results of these promoter deletion experiments indicated that the proximal Ϫ510 bp of the Egr-1 promoter contains the elements required for depolarization induction.
To further define the promoter elements required for induction, several additional deletion constructs of the Ϫ532 bp sequence derived from the rat Egr-1 promoter were tested (the proximal Ϫ532 bp of the mouse and rat Egr-1 genes are Ͼ93% conserved, with a 100% similarity of the CRE and SREs). As shown in Fig. 2B, prEgr-532 contains five SREs and one CRE. decrease in the basal activity, but similar fold responses to FK, KCl (2-2.2-fold respectively, p Ͻ 0.005), and KCl plus FK (3.3-fold, p ϭ 0.001), although the response to PMA was reduced by 60% (p ϭ 0.001). Deletion of the CRE element (prEgr-109) resulted in a further decrease in basal promoter activity, and a reduction in the fold responses to KCl and FK, suggesting that the CRE contributes to the depolarization activation of the Egr-1 promoter. Interestingly, the CRE-deleted promoter containing only the proximal two SREs still retained significant responses to KCL (1.5-fold, p ϭ 0.004), to forskolin (1.6-fold, p ϭ 0.004), and to both agents (3-fold, p Ͻ 0.01). The conclusions drawn from these promoter deletion experiments are: 1) elements required for the responses to each of the stimulatory agents are contained within the proximal Ϫ532 bp of the promoter, 2) elements present within Ϫ198 bp of the promoter respond to depolarization activation, and 3) that both the CRE and the SREs are likely to contribute to depolarization activation of the promoter.

Further Assessment of the Role of the CRE Element and CREB Phosphorylation in Depolarization Induction of Egr-1-
Transcriptional activation of the c-fos gene by Ca 2ϩ and cAMP in neuronal cells is mediated by CREB phosphorylation of Ser 133 (13). To further define the process whereby depolarization activates the CRE of the Egr-1 promoter, CREB phosphorylation was assessed in MIN6 cells following depolarization. Western blotting was performed using an antibody specific for CREB phosphorylated at Ser 133 . As shown in Fig. 3A, both KCl and FK treatment induced CREB phosphorylation on Ser 133 within five minutes. High glucose resulted in an increase in CREB phosphorylation after 30 min. The PKA inhibitor H89 diminished both the glucose and the forskolin effect on CREB phosphorylation, although the magnitude of the inhibitory effect of H89 on forskolin may have been less, as there appeared to be a lower quantity of CREB protein. The total amount of CREB protein did not differ with any of the agents tested, as indicated by using an antibody to nonphosphorylated CREB. These results are consistent with glucose-induced depolarization and Ca 2ϩ activation of CREB, in part by a PKA-dependent mechanism.
To assess the transcriptional activity of CREB following depolarization, the induction of Egr-1 was assessed in the presence of a dominant inhibitor of CREB function (KCREB) (28). Co-transfection of the pmEgr-1200 construct along with KCREB at concentrations previously shown to inhibit CREB activation of transcription resulted in significant inhibition of the responses to glucose, KCl, and forskolin (66%, 16%, 60% and p ϭ 0.001, p ϭ 0.03, p Ͻ 0.02, respectively) (28, 31, 38) (Fig. 4A). The inhibition of the response to a constitutively active PKA (100%, p Ͻ 0.001 data not shown) demonstrated that KCREB was an effective inhibitor of CREB-mediated transcription.
The significant inhibition by KCREB of the glucose and KCl responses further suggested that CREB-dependent transcrip-  tional activation contributes to Egr-1 regulation. Yet the partial inhibition strongly suggested that other factors are involved in activation of Egr-1 transcription. To more carefully define the role of CRE/CREB in transcriptional induction of Egr-1 by depolarization, a mutant Egr-1 promoter was constructed (pmEgr-1200⌬CRE) by altering the CRE element without disturbing surrounding sequences and relative spacing. The glucose effect was reduced ϳ50% (p Ͻ 0.02) in the pmEgr-1200⌬CRE relative to the nonmutated promoter. Remarkably the KCl response of the CRE mutant was still greater than 5-fold (Fig. 4B) and inhibited only by 14% (p Ͻ 0.04), providing further evidence that other elements are involved in depolarization activation. Co-transfection of pmEgr-1200⌬CRE with a constitutively active PKA resulted in a response that was reduced by 63% relative to co-transfection with the nonmutated plasmid, indicating that PKA activates other elements in the promoter in addition to the CRE. The pmEgr-1200⌬CRE construct was also activated by cotransfection with a constitutively active form of CaMKIV, a known activator of both SRE-and CRE-mediated transcription (27).

Evaluation of the SRE Elements and SRF in Depolarization Induction of Egr-1 in ␤-Cells-
The results of the previous experiments indicated that in addition to CREB activation of the CRE, other elements of the Egr-1 promoter were involved. To assess the effect of depolarization on SRE-dependent transcription, a luciferase plasmid containing five tandem repeats of the SRE linked to a minimal promoter was evaluated (Fig. 4). Treatment with KCl resulted in a 3.2-fold response (p Ͻ 0.0001). Treatment with forskolin resulted in 1.5-fold (p ϭ 0.02) induction, and there was no augmentation of the KCl response by forskolin.
The SRE is continuously occupied in vivo by SRF and Ets proteins of the ternary complex subfamily (39). Activated SRF induces transcription by binding to SREs and by recruiting other factors to the SRE complex. Phosphorylation of SRF at Ser 103 enhances the ability of SRF to bind the SRE (30). To demonstrate the presence of a nuclear factor binding to the SRE, nuclear extracts from MIN6 cells were incubated with a c-fos consensus SRE (Fig. 6). In the absence of nuclear extracts, band D represents migration of the free SRE oligonucleotide (lane P). At least three additional bands (A-C) were observed following incubation with nuclear extracts in the absence of competitors (lanes 1 and 7). They included two slower migrating bands (A and B) and a diffuse faster migrating complex (C). Bands A and B migrate similarly to bands identified by others as the SRE⅐SRF complex (band B) and the ternary complex (band A) composed of the SRE, SRF, and ELK-1 or another TCF  (40 -43). The identity of band C is not known and the intensity varies with cell type (43). Bands A and B disappeared when 10-, 25-, and 50-fold molar excess of oligonucleotide (lanes 2-4) containing the sequence of the most distal Egr-1 SRE was added. Similar results were obtained by competition with each of the five Egr-1 SREs (data not shown). Complexes A, B, and C were not evident when an excess of unlabeled oligonucleotide was present during the incubation with nuclear extracts (data not shown). In contrast, competition by a mutated SRE oligonucleotide did not appear to significantly alter bands A and B, suggesting specificity of binding (Fig. 5A, lanes 1, 5, and 6). To identify the identity of the proteins present in the SRE complex (bands A and B), nuclear extracts from MIN6 cells were preincubated with anti-SRF antibody (␣SRF). As shown in lane 8, bands A and B disappeared, and a more slowly migrating band F appeared when ␣SRF was added. To determine whether band F contains Ser 103 -phosphorylated SRF, the addition of a phospho-Ser 103 -specific ␣SRF antiserum was used. As shown in lane 9, a fraction of the phospho-SRF-containing protein⅐DNA complex in bands A and B was supershifted to band F, indicating that under chronic culture in 25 mM glucose a fraction of SRF is in an activated form. No supershift was observed when nonspecific antibodies were used (lanes 10 and 11).
To determine whether depolarization results in transcriptional activation of SRF, a trans-activator plasmid SRF-GAL4 containing amino acids 11-508 ligated downstream of the sequence encoding the DNA binding and the dimerization domain of GAL4 (44) was used. The SRF-GAL4 plasmid was cotransfected with a reporter gene containing a synthetic promoter with five tandem repeats of the yeast GAL4 binding site upstream of the luciferase gene. As shown in Fig. 7A, KCl and forskolin treatment activated the SRF-GAL4 gene (1.8-fold, p ϭ 0.002), and the effect was potentiated by the combination of both agents (8-fold, p Ͻ 0.001). To assess the transcriptional activation of CREB, a trans-activator plasmid that expresses a fusion protein containing the activator domain of CREB fused to the DNA binding domain of GAL4 (residues 1-147) was tested. This CREB-GAL4 construct responded ϳ2-fold to for-skolin and more than 10-fold to a constitutively active PKA or CaMKIV. Yet, in contrast to the activation of SRF-GAL4, there was no significant response of CREB-GAL4 to KCl (Fig. 7B). DISCUSSION The results of the current studies have now elucidated mechanisms whereby glucose induced depolarization and Ca 2ϩ influx regulates Egr-1 transcription in insulinoma cells. Failure of glucose to induce Egr-1 in the presence of diazoxide, an inhibitor of depolarization, conclusively showed that glucose metabolism in the absence of depolarization was not sufficient to activate Egr-1 transcription. Pharmacological inhibitor studies suggested that CaM and PKA pathways are involved in the glucose response. In addition, augmentation of the response by phosphatidylinositol 3-kinase inhibitors suggested that this pathway may also be involved, perhaps in an inhibitory fashion. Examination of the Egr-1 promoter indicated that: 1) the elements required for the response to depolarization are contained in the proximal Ϫ532 bp of the promoter containing five SREs and a CRE; 2) the distal three SREs in the Ϫ532 bp promoter contribute, but are not essential for depolarization induced transcriptional activation; and 3) the CRE also contributed but was not necessary for the depolarization response. Further promoter deletion experiments, transactivation assays, and gel shifts assays supported these conclusions and also demonstrated for the first time the activation of SRF in the depolarization response in pancreatic islet ␤-cells.
We considered whether the glucose-induced rapid transcriptional activation of Egr-1 in insulinoma cells could be secondary to glucose-induced insulin secretion, i.e. an autocrine/paracrine effect. A prominent anabolic property of insulin is its effect on gene transcription. Acting through the insulin receptor, insulin has been shown to activate IEG transcription through a Ras/Raf/mitogen-activated protein kinase-dependent pathway (45)(46)(47). This appears to be an unlikely mechanism for glucose induction of Egr-1 transcription in insulinoma cells; however, as KCl activation of Egr-1 transcription was not inhibited by inhibitors of mitogen-activated protein kinase or phosphatidylinositol 3-kinase (Table I), known mediators of insulin signaling pathways for gene transcription. Further, experiments evaluating directly the effects of insulin on MIN6 cells showed no activation of Egr-1 transcription when exogenous insulin was added at concentrations as high as 1 M. That this insulin signaling pathway is intact in these insulinoma cells has been demonstrated by activation of Akt/PkB in response to exogenous insulin in this concentration range. 2 Because the depolarization-induced activation of Egr-1 transcription occurred as early as 30 min and was independent of new protein synthesis, we assessed the role of phosphorylation of transcription factors known to activate early gene transcription in other cells. For example, in neuronal cells, depolarization induction of c-fos results for the most part through Ca 2ϩ activation of CREB (48,49). We initially suspected that depolarization induction of Egr-1 in pancreatic islet ␤-cells would be because of a similar Ca 2ϩ activation of CREB. In fact, earlier studies of depolarization-induced gene transcription in hamster insulinoma (HIT) cells emphasized the role of activated CREB interacting with the CRE of the glucagon gene promoter (50). The current observations demonstrated that, like depolarization activation of the glucagon gene, CRE/CREB is also involved in depolarization-mediated transcriptional induction of Egr-1. The results of the promoter deletions and mutation of the CRE element supported this idea. Depolarization resulted in phosphorylation of CREB, and transfection of insulinoma cells with a dominant negative CREB had a significant inhibitory effect. It should be noted, however, that the small inhibition of the depolarization response by the dominant negative KCREB and the robust response of the mutated CRE construct is not surprising because the Egr-1 promoter contains five SREs in addition to the CRE. These results diverge to those described for the c-fos gene in which the CRE is a potent mediator of transcriptional activation by Ca 2ϩ signaling (48,49). The sum of these findings, along with the demonstration of binding of CREB to the Egr-1 promoter CRE (data not shown), indicated that Ca 2ϩ -mediated CREB-dependent transcriptional activation contributes to depolarization induction of Egr-1, although other elements in the promoter can mediate this response.
The results of the current experiments demonstrate for the first time the role of the SRE/SRF in depolarization induction of transcription in insulinoma cells. Susini et al. (14) evaluated glucose and cAMP induction of the c-fos gene in INS-1 insulinoma cells. In their study, induction of the c-fos promoter was dependent on an intact CRE and not altered following mutation of the SRE. This study differed from the present one in that c-fos was not induced by glucose in the absence of cAMP nor was the response to KCl-induced depolarization assessed. In addition, the c-fos promoter has only one SRE, whereas the Egr-1 promoter has five SREs. In the present studies, the responses of the mutated CRE plasmid pmEgr-1200⌬CRE and 5XSRE-LUC to depolarization demonstrated that the SREs also serve as Ca 2ϩ -response elements in this model. Similar conclusions have been made for the c-fos promoter in PC12 cells, where membrane depolarization rapidly induces the phosphorylation of SRF at Ser 103 , and this phosphorylation enhances the ability of SRF to bind the SRE (30). The SRE consensus sequence, CC(A/T) 6 GG, is bound in vitro and in vivo by SRF protein. The demonstration of transcriptional activation of SRF by the SRF-GAL4 assay, as well as that of phosphorylated SRF bound to the SRE (Fig. 6), indicated that a similar mechanism exists in insulinoma cells. The present studies do not define how Ca 2ϩ /CaM activates SRF. Whereas both CaMKII and CaMKIV can phosphorylate SRF in vitro, CaMKIV is the best candidate because of its nuclear localization (51). Demonstration that transfection of insulinoma cells with a constitutively active CaMKIV also induced the activation of SRF (see Fig. 7) suggested that this kinase is likely to mediate Ca 2ϩ /CaM activation of transcription.
These findings demonstrate the effect of glucose on SRE/SRFdependent transcription in pancreatic islet ␤-cells. The SRF is a transcription factor early defined as one of many mediating mitogenic responses and regulating fibroblast proliferation in response to growth factors (52). Activated SRF induces transcription by binding to SREs and by recruiting accessory proteins (ternary complex factors). These accessory proteins can potentiate the transcriptional response of the SRF. However, the contribution of transcription factors that interact with SRF in SRE-dependent transcription by glucose in ␤-cells remains undefined. Prolonged exposure to glucose is a well recognized stimulus to ␤-cell hypertrophy and hyperplasia (1). The current results suggest that following exposure to glucose, ensuing ␤-cell depolarization and Ca 2ϩ induction of SRE-mediated transcription, as well as CRE-mediated transcription, could represent important mechanisms that regulate the normal morphological and physiological changes of pancreatic islets and the changes occurring in pathological conditions such as diabetes.