Characterization of a Novel Calcium Response Element in the Glucagon Gene*

To maintain blood glucose levels within narrow limits, the synthesis and secretion of pancreatic islet hormones is controlled by a variety of extracellular signals. Depolarization-induced calcium influx into islet cells has been shown to stimulate glucagon gene transcription through the transcription factor cAMP response element-binding protein that binds to the glucagon cAMP response element. By transient transfection of glucagon-reporter fusion genes into islet cell lines, this study identified a second calcium response element in the glucagon gene (G2 element, from −165 to −200). Membrane depolarization was found to induce the binding of a nuclear complex with NFATp-like immunoreactivity to the G2 element. Consistent with nuclear translocation, a comigrating complex was found in cytosolic extracts of unstimulated cells, and the induction of nuclear protein binding was blocked by inhibition of calcineurin phosphatase activity by FK506. A mutational analysis of G2 function and nuclear protein binding as well as the effect of FK506 indicate that calcium responsiveness is conferred to the G2 element by NFATp functionally interacting with HNF-3β binding to a closely associated site. Transcription factors of the NFAT family are known to cooperate with AP-1 proteins in T cells for calcium-dependent activation of cytokine genes. This study shows a novel pairing of NFATp with the cell lineage-specific transcription factor HNF-3β in islet cells to form a novel calcium response element in the glucagon gene.


From the Department of Molecular Pharmacology, University of Göttingen, D-37070 Göttingen, Germany
To maintain blood glucose levels within narrow limits, the synthesis and secretion of pancreatic islet hormones is controlled by a variety of extracellular signals. Depolarization-induced calcium influx into islet cells has been shown to stimulate glucagon gene transcription through the transcription factor cAMP response element-binding protein that binds to the glucagon cAMP response element. By transient transfection of glucagon-reporter fusion genes into islet cell lines, this study identified a second calcium response element in the glucagon gene (G2 element, from ؊165 to ؊200). Membrane depolarization was found to induce the binding of a nuclear complex with NFATp-like immunoreactivity to the G2 element. Consistent with nuclear translocation, a comigrating complex was found in cytosolic extracts of unstimulated cells, and the induction of nuclear protein binding was blocked by inhibition of calcineurin phosphatase activity by FK506. A mutational analysis of G2 function and nuclear protein binding as well as the effect of FK506 indicate that calcium responsiveness is conferred to the G2 element by NFATp functionally interacting with HNF-3␤ binding to a closely associated site. Transcription factors of the NFAT family are known to cooperate with AP-1 proteins in T cells for calcium-dependent activation of cytokine genes. This study shows a novel pairing of NFATp with the cell lineage-specific transcription factor HNF-3␤ in islet cells to form a novel calcium response element in the glucagon gene.
Activation of gene transcription allows cells to adapt to changes in environmental conditions through a new pattern of expressed proteins. In cells that are electrically excitable, calcium is an important intracellular second messenger that directs the genomic response of the cell. Transcription factors that have been shown to mediate calcium-induced gene transcription include CREB 1 (1)(2)(3)(4), serum response factor (2,5), and C/EBP␤ (6).
Like neurons, endocrine cells of the pancreatic islets are electrically excitable and express L-type voltage-dependent calcium channels (7,8). By virtue of stimulation of glycogenolysis and gluconeogenesis in the liver, the islet hormone glucagon is an important regulator of blood glucose levels (9). Glucagonproducing islet cells show spontaneous electrical activity (10). Membrane electrical activity and calcium influx into glucagonproducing pancreatic islet cells is tightly controlled by extracellular messengers. Whereas L-arginine increases spike frequency (10), ␤-adrenergic cell-surface receptor stimulation by catecholamines through cAMP enhances the L-type calcium current, increasing the influx of calcium associated with each action potential (11). It is well known that membrane depolarization and calcium influx increase the cytosolic free calcium concentration, which stimulates hormone secretion by exocytosis (8,12). The question is whether, and if so, how, this calcium signal reaches also into the nucleus and regulates gene transcription.
Previous experiments have shown that membrane depolarization and calcium influx stimulate glucagon gene transcription (3). A mechanism involved has been characterized. Through a calcium-/calmodulin-dependent protein kinase calcium stimulates the phosphorylation of the transcription factor CREB on the same serine residue that is also phosphorylated by cAMP-dependent protein kinase A. CREB binds to a CRE in the 5Ј-flanking region of the glucagon gene and stimulates transcription (3,(13)(14)(15). The present study addressed the question whether there are additional calcium-responsive elements in the glucagon gene. By the results obtained, the G2 element is identified as a second calcium response element. The further characterization suggests that calcium responsiveness is conferred by the calcium/calcineurin-regulated transcription factor NFATp functionally synergizing with the cell-specifically expressed transcription factor HNF-3␤. NFAT family proteins are known to cooperate with newly synthetized AP-1 proteins in T cells (16 -19). This study shows a novel pairing of NFATp with the constitutively expressed cell-specific transcription factor HNF-3␤ forming a novel calcium response element in the glucagon gene.
Cell Culture and Transfection of DNA-The pancreatic islet cell line HIT-T15 (27) was grown in RPMI 1640 medium supplemented with 10% fetal calf serum, 5% horse serum, 100 units/ml penicillin, and 100 g/ml streptomycin. ␣TC2 cells (28,29) were grown in Dulbecco's modified Eagle's medium (4.5 g of glucose/liter) supplemented with 2.5% fetal calf serum, 15% horse serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were trypsinized and transfected in suspension by the DEAE-dextran method (3) with 2 g of indicator plasmid per 6-cm dish. Rous sarcoma virus-chloramphenicol acetyltransferase plasmid (0.4 g/6-cm dish) was added as a second reporter to check for transfection efficiency. When indicated, 0.7 g of pBK-CMV-NFATp-DBD was co-transfected per 6-cm dish; these co-transfections were done with a constant DNA concentration, which was maintained by adding Bluescript (Stratagene, Heidelberg, Germany). Cells were stimulated with high KCl (45 mM final concentration) or TPA (300 nM) for 6 h before harvest. FK506 (167 nM, unless noted otherwise) was added 1 h before stimulation. Cell extracts (3) were prepared 48 h after transfection. A chromatographic chloramphenicol acetyltransferase assay was performed as described previously (30). Thin layer chromatography plates were analyzed with a Fuji PhosphorImager. The luciferase assay was performed as described previously (3).
Cell Extracts-Nuclear extracts were prepared from ␣TC2 cells by the method of Schreiber et al. (31), except for the experiments shown in Fig. 3G. In the experiments shown in Fig. 3G, separate nuclear and cytosolic extracts were prepared as described previously (32)(33)(34).
Electrophoretic Mobility Shift Assay-Synthetic complementary oligonucleotides with 5Ј-GATC overhangs were annealed and labeled by a fill-in reaction with [␣-32 P]dCTP and Klenow enzyme. By using 15 g of protein from cell extracts, the electrophoretic mobility shift assay was performed as described by Klemsz et al. (35). In some binding reactions, a specific anti-NFATp antiserum (kindly provided by A. Rao, Harvard Medical School, Boston) was used. This antiserum (anti-67.1) is directed against the 67.1 peptide of murine NFATp; it does not cross-react with NFATc, NFAT3, or NFATx (36). Cell extracts were incubated with 1 l of a 1:10 dilution of preimmune serum or anti-NFATp antiserum in the reaction buffer with probe for 15 min at room temperature. Following a 15-min incubation on ice, the samples were loaded onto the gels and electrophoresed as described above.
Materials-A stock solution of TPA (1 mM) was prepared in dimethyl sulfoxide and further diluted in cell culture medium. FK506 (provided by Fujisawa) was dissolved in ethanol. Controls received the solvent only.

Identification of a Second Calcium-responsive Element in
the Glucagon Gene-To study the calcium responsiveness of the glucagon gene in the absence of a functional CRE, the CRE in the 5Ј-flanking region of the rat glucagon gene was removed by either 5Ј-deletion to Ϫ292 or by internal deletion of four bases within the CRE octamer motif, yielding the constructs Ϫ292GluLuc and Ϫ350(Ϫ297/Ϫ292)GluLuc, respectively (Fig.  1). The corresponding glucagon-reporter fusion genes were transiently transfected into two pancreatic islet cell lines, HIT and ␣TC2. These cell lines have been used previously to demonstrate that membrane depolarization and calcium influx in- duce glucagon gene transcription through the CRE of the glucagon gene (3,(13)(14)(15). HIT cells (27) express glucagon, although only at a low level (37,38), and have the advantage to respond to second messenger stimulation and to be well characterized with respect to electrical activity, voltage-dependent calcium channels, cytosolic calcium concentration, and secretion (see Ref. 3). Increases in extracellular potassium concentration to 40 mM have been shown to depolarize HIT cells with action potentials continuing at the peak of the depolarizing phase of the spontaneous activity; at the same time, high potassium initiates the influx of calcium through dihydropyridine-sensitive, L-type Ca 2ϩ channels in the cell membrane and elevates the cytosolic calcium concentration (39). Membrane depolarization was induced by elevating the potassium chloride concentration in the incubation medium from 5 to 45 mM. As shown in Fig. 1B, the high potassium-induced increase in glucagon gene transcription was decreased in HIT cells by 77% by the 4-base deletion in the CRE octamer motif, as has been reported previously (3). It was decreased by 46% when the CRE was removed by 5Ј-deletion to Ϫ292 (Fig. 1B). Basal reporter activity was not changed by these deletions (data not shown). Thus, although the depolarization-induced increase in glucagon gene transcription was reduced using both constructs, glucagon gene transcription did still respond to membrane depolarization, resulting in a 2.0-and 3.3-fold stimulation of transcription in the absence of the CRE, respectively (Fig. 1B). After transfection of Ϫ292GluLuc, the stimulation of transcription by high potassium was abolished when extracellular calcium was bound by 1.5 mM EGTA added to the medium (data not shown), suggesting that gene induction by membrane depolarization depends on calcium influx elevating intracellular calcium levels. The pancreatic ␣-like cell line ␣TC2 has been established from a glucagonoma arising in transgenic mice expressing the SV40 large T-antigen oncogene (28,29). These cells express predominantly glucagon in a rather uniform pattern (28). In ␣TC2 cells, the depolarization-induced stimulation of glucagon gene transcription was only slightly reduced by 5Ј-deletion to Ϫ292 or by the internal deletion of four bases within the CRE octamer (Fig. 1C). When compared with Ϫ350GluLuc, basal reporter activity of Ϫ350(Ϫ297/Ϫ292)Glu-Luc was not changed, and basal reporter activity of Ϫ292Glu-Luc was reduced to 38 Ϯ 7% (n ϭ 12). The results obtained in both pancreatic islet cell lines thus indicate that besides the CRE the rat glucagon gene 5Ј-flanking region contains at least one more calcium-responsive element within 292 bases in front of the transcription start site.
Known elements within the enhancer region of the glucagon gene include the G2 element (20,40,41), the G3 element (30,40,(42)(43)(44), as well as a binding site for C/EBP proteins (CS) (23) (Fig. 2). To study their role, four copies of oligonucleotides containing the G2 element (from Ϫ200 to Ϫ165), the G3 element (from Ϫ274 to Ϫ234), or the CS element (from Ϫ241 to Ϫ212) were placed in front of the truncated viral thymidine kinase promoter of herpes simplex virus (from Ϫ81 to ϩ52) fused to the luciferase reporter gene (plasmid pT81Luc) (21). As shown in Fig. 2, the promoter alone did not respond to membrane depolarization in ␣TC2 cells. The G3 element and the binding site for C/EBP proteins were also inactive ( Fig. 2). However, the G2 element did confer depolarization responsiveness to the promoter (Fig. 2). The blocker of L-type, voltage-dependent calcium channels, diltiazem, inhibited the high potassium-induced transcriptional activation of the G2 element (data not shown). Depolarization responsiveness conferred by the G2 element to the nonresponsive thymidine kinase promoter was only somewhat less than that conferred by the glucagon CRE. When in separate experiments the plasmids 4xG2T81Luc and 4xGluCRET81Luc were transfected in parallel into ␣TC2 cells, membrane depolarization stimulated transcription to 182 Ϯ 14% of controls (n ϭ 6) through the G2 element and to 225 Ϯ 18% of controls (n ϭ 6) through the glucagon CRE. These data indicate that the G2 element is a calcium-responsive element of the glucagon gene.
Characterization of Protein Binding to the G2 Element-The G2 element has been shown before to confer cell-specific basal activity (20,40,41) and Ras/protein kinase C responsiveness (20) to the glucagon gene. The binding of two transcription factors is required for both of these activities in ␣TC2 cells, HNF-3␤ and an Ets-like protein (20). However, the precise role of HNF-3␤ in the islet-specific, basal transcriptional activity of G2 remains to be defined (20,41). To study the effect of membrane depolarization and calcium influx on nuclear protein binding to the G2 element, electrophoretic mobility shift assays were performed. Nuclear extracts were prepared from pancreatic islet ␣TC2 cells with or without stimulation by membrane depolarization. It has been shown previously that under basal nonstimulated conditions band h ( Fig. 3A) represents the binding of HNF-3␤, whereas several weak bands collectively labeled e (Fig. 3A) represent proteins with a binding specificity related to the Ets family of transcription factors (20). Membrane depolarization for 15 or 120 min induced the binding of a protein complex that migrated among the bands e (Fig. 3A, marked by an arrow). Whereas the intensities of some bands e varied somewhat between the experiments, the depolarization-induced nuclear protein binding was consistently observed (Fig.  3, A-F).
The binding specificity of the depolarization-induced protein complex was studied using G2 mutants. The depolarizationinduced nuclear protein complex did bind to labeled mutant 1 FIG. 3. Membrane depolarization induces protein binding to the G2 element adjacent to the binding of HNF-3␤ as revealed by the electrophoretic mobility shift assay. A, depolarization-induced nuclear protein binding. ␣TC2 cells were stimulated for 15 or 120 min with high potassium-induced membrane depolarization (KCl, 45 mM) or were left unstimulated. Nuclear extracts were prepared and incubated with labeled G2 oligonucleotide (wild type). It has been shown previously that under basal nonstimulated conditions band h represents the binding of HNF-3␤, whereas several weak bands collectively labeled e represent proteins with a binding specificity related to the Ets family of transcription factors (20). f, free probe; ns, nonspecific bands. The arrow indicates the depolarization-induced protein binding. The G2 wild-type sequence is presented at the top of the figure, and the bases mutated in mutants 1, 3, and 5 are indicated. The boxed sequence motifs show the homology to the HNF-3 consensus binding site (20, 45) as well as the GGAA motif, an essential purine-rich core of binding sites for members of the Ets family of transcription factors. Also boxed is a sequence, 5Ј-GGAAACTT-3Ј, which shares extensive homology with a sequence in the murine IL-2 gene that binds transcription (Fig. 3B, lanes 7 and 8) but not to labeled mutant 5 (Fig. 3B,  lanes 11 and 12). In mutant 1 and mutant 5, 5 and 2 bases, respectively, have been mutated that are required for the binding of HNF-3␤ and the Ets-like proteins, respectively (see Ref. 20; see also top of Fig. 3A). Consequently, band h was not formed when G2 mutant 1 was used as probe, and bands e were not detectable when G2 mutant 5 was used as probe (Fig. 3B). This indicates that the depolarization-induced nuclear complex binds to a site within G2 that is distinct from the HNF-3␤binding site; in contrast, the two guanine bases (noncoding strand) exchanged in mutant 5 are essential for binding (see top of Fig. 3A).
The binding specificity was further characterized in competition experiments using G2 wild type, mutant 1, mutant 5, and also mutant 3. In mutant 3, 4 bases have been mutated (see top of Fig. 3A). As shown in Fig. 3C, the binding of HNF-3␤ (band h) was competed away by mutants 5 and 3 but not by mutant 1, which carries mutations within the HNF-3␤-binding sequence. In contrast, mutant 1 competed for the binding of the depolarization-induced protein as well as did the wild-type sequence (Fig. 3C, compare lane 4 with lane 3 and lane 9 with lane 8), whereas mutant 3 and, even more so, mutant 5 were less effective competitors (Fig. 3C). This suggests that the depolarization-induced protein complex has critical contacts with bases that are mutated in mutants 3 and 5.
The mutations in mutants 3 and 5 alter a sequence within G2, 5Ј-GGAAACTT-3Ј, which shares extensive homology with a sequence in the murine IL-2 gene that binds transcription factors of the NFAT family (match of 7 bases out of 8) (18,19). In T cells, NFAT transcription factors have been shown to undergo a calcium-and calcineurin-dependent translocation from the cytosol to the nucleus (18,19,46). Therefore, a possible relationship between the depolarization-induced G2-binding protein complex and NFAT was studied. First, the oligonucleotide NFATcons, which contains a well characterized NFATbinding site, was used as probe and cold competitor in electrophoretic mobility shift assays. As shown in Fig. 3D, NFATcons specifically competed for the binding of the depolarization-induced nuclear protein to labeled G2 (lane 4). When NFATcons was used as probe, membrane depolarization induced the binding of a protein complex that comigrated with the depolarization-induced nuclear protein binding to labeled G2 (Fig. 3D, compare lane 7 with lanes 5 and 2; the depolarization-induced protein complexes are indicated by arrows); it also shown a binding specificity that was indistinguishable from that of the depolarization-induced nuclear binding to labeled G2, since the depolarization-induced binding to labeled NFATcons was competed away by G2 and NFATcons but not by G2 mutant 5 (Fig. 3D). In addition to the depolarization-induced nuclear protein, several additional complexes were detected with labeled NFATcons; these were not further characterized. The cross-competition between G2 and NFATcons supports the notion that the depolarization-induced nuclear complex on G2 has a binding specificity that is related to NFAT. Second, the effect of FK506 was studied. FK506 is an inhibitor of the calcium-/calmodulin-dependent protein phosphatase calcineurin (47). Previous studies have shown that FK506 and cyclosporin A effectively inhibit calcineurin activity also in pancreatic islet cells (14,15). Whereas membrane depolarization induced the binding of a protein complex to the G2 element in controls (Fig. 3E, lanes 5 and 6), treatment of the cells with FK506 (167 nM) blocked this effect (Fig. 3E, lanes 7  and 8), suggesting that the induction by membrane depolarization of nuclear protein binding to G2 depends on calcineurin phosphatase activity. Third, the effect of an antiserum directed specifically against NFATp was studied (donated by A. Rao, Boston, MA) (36). Whereas the addition of a preimmune serum had no effect, the addition of the anti-NFATp antiserum abolished specifically the depolarization-induced G2-binding protein (Fig. 3F, compare lane 6 with lanes 4 and 2), indicating that this protein complex possesses NFATp-like immunoreactivity. Finally, using another protocol, separate nuclear and cytosolic extracts were prepared and used in electrophoretic mobility shift assays. A G2-binding protein complex was found in cytosolic extracts from unstimulated cells that comigrated with the nuclear complex induced by membrane depolarization (Fig. 3G, compare lane 7 with lane 2). Concomitant with the induction of protein binding in nuclear extracts, membrane depolarization markedly decreased the binding of this complex in cytosolic extracts (Fig. 3G, compare lane 7 with lane 8), consistent with a depolarization-induced translocation from the cytosol to the nucleus. The binding of both the protein complex induced by membrane depolarization in nuclear extracts as well as the binding of the comigrating protein complex factors of the NFAT family (18). B, binding specificity as analyzed by using wild-type and mutant probes. Labeled G2, mutant 1 (m1) or mutant 5 (m5) were incubated with nuclear extracts prepared from ␣TC2 cells that had been stimulated for 120 min with high potassium-induced membrane depolarization (KCl, 45 mM) or had been left unstimulated. The arrow indicates the depolarization-induced protein complex. f, free probe; ns, nonspecific bands. C, binding specificity as analyzed in competition experiments. Labeled G2 (wild type) was incubated with nuclear extracts prepared from ␣TC2 cells that had been stimulated for 120 min with high potassium-induced membrane depolarization (KCl, 45 mM) or had been left unstimulated. Competitors were added at a 500-fold (lanes 3-6) or 175-fold (lanes 8 -11) molar excess as indicated. wt, wild type; m1, 3, or 5, mutants 1, 3, or 5. The arrow indicates the depolarization-induced protein binding. f, free probe; ns, nonspecific bands. D, cross-competition between G2 and an oligonucleotide containing a well characterized NFAT-binding site, NFATcons, for binding of the depolarization-induced nuclear protein. Labeled G2 or labeled NFATcons were incubated with nuclear extracts prepared from ␣TC2 cells that had been stimulated for 30 min with high potassium-induced membrane depolarization (KCl, 45 mM) or had been left unstimulated. Competitors were added at a 500-fold molar excess. m5, G2 mutant 5. The arrows indicate depolarization-induced protein binding. f, free probe; ns, nonspecific bands. The asterisk indicates HNF-3␤ binding to labeled G2. E, effect of FK506 on the induction of nuclear protein binding by membrane depolarization. Labeled G2 was incubated with nuclear extracts from ␣TC2 cells that had been stimulated with high potassium-induced membrane depolarization (KCl, 45 mM) for 120 min or had been left unstimulated. FK506 (167 nM) was added to the cultures 1 h prior to KCl. The arrow indicates the depolarization-induced protein binding. f, free probe; ns, nonspecific bands. F, the depolarization-induced nuclear G2-binding protein is recognized by a specific anti-NFATp antiserum. Labeled G2 was incubated with nuclear extracts prepared from ␣TC2 cells that had been stimulated for 60 min with high potassium-induced membrane depolarization (KCl, 45 mM) or had been left unstimulated. Preimmune serum (Pre) or anti-NFATp antiserum (Anti-NFATp) were added to the binding reaction as indicated on top of the lanes. The arrow indicates the depolarization-induced protein binding. The asterisk indicates a retarded band that appeared in the presence of anti-NFATp antiserum when added to the binding reaction with nuclear extracts from stimulated cells ("super-shifted band"). f, free probe; ns, nonspecific bands; h, HNF-3␤. G, evidence for depolarizationinduced translocation from the cytosol to the nucleus of a protein with NFATp-like immunoreactivity. Separate nuclear and cytosolic extracts were prepared from ␣TC2 cells stimulated or not with high potassium-induced membrane depolarization (KCl, 45 mM, for 120 min) and incubated with labeled G2 oligonucleotide. Preimmune serum (Pre) or anti-NFATp antiserum (Anti) as indicated on top of the lanes were added to the binding reaction. The arrow on the left indicates the depolarization-induced nuclear protein binding, whereas the arrow on the right indicates a comigrating complex in cytosolic extracts from unstimulated cells. Note that the depolarization-induced nuclear protein binding and the comigrating complex in cytosolic extracts from unstimulated cells are abolished by the anti-NFATp antiserum. The asterisk indicates a retarded band that appeared in the presence of anti-NFATp antiserum when added to the binding reaction with nuclear extracts from stimulated cells (super-shifted band). f, free probe; ns, nonspecific bands. in cytosolic extracts from unstimulated cells was not affected by the addition of a preimmune serum but was abolished by the addition of an anti-NFATp antiserum (Fig. 3G). Taken together, these results indicate that the depolarization-induced nuclear G2-binding protein is NFATp which in pancreatic islet cells undergoes a calcium-and calcineurin-dependent translocation from the cytosol to the nucleus in response to membrane depolarization.
The results obtained in the electrophoretic mobility shift assay indicate that NFATp is expressed in pancreatic islet ␣TC2 cells. This was further investigated using different approaches. In an RT-PCR analysis, a primer pair that is specific for NFATp generated from poly(A) ϩ RNA of ␣TC2 cells a product of the expected size (Fig. 4A). Subcloning and sequencing confirmed that this product is NFATp cDNA, indicating that NFATp is expressed at the RNA level. A more detailed analysis using primer pairs, which target the divergent C-terminal ends of NFATp isoforms (19,26), detected transcripts encoding the B and C isoforms but not the A isoform (not shown). To confirm further the expression of NFATp at the protein level Western blotting was used. NFAT proteins are known to migrate on SDS gels with an apparent molecular mass of 120 -140 kDa (19). As shown in Fig. 4B (upper panel), a protein with NFATplike immunoreactivity and an apparent molecular mass of about 120 kDa was detected in nuclear extracts from ␣TC2 cells that had been stimulated by membrane depolarization. No such band was detected in nuclear extracts from unstimulated cells (Fig. 4B, lane 1). As a control for the quality of the nuclear extracts, an antiserum directed against a constitutively nuclear protein, CREB, was also used and yielded bands of about 43 kDa with similar signal intensities in both extracts (Fig. 4B,  lower panel). Therefore, the results obtained in the electrophoretic mobility shift assay, in the RT-PCR analysis, and in Western blotting experiments concur that NFATp is expressed in these cells. When taken together, our data suggest that membrane depolarization induces in pancreatic islet cells through activation of calcineurin the nuclear translocation of NFATp, which binds to the G2 enhancer-like element of the glucagon gene.
Functional Analysis of NFATp and HNF-3␤ Binding to the G2 Element-To study the role of NFATp binding in calciuminduced activation of G2 transcriptional activity, transfection experiments were performed. As shown in Fig. 5, when nuclear FIG. 6. Mutational analysis of sequences within G2 required for depolarization responsiveness. Four copies of the G2 oligonucleotide with wild-type sequence (G2) or the G2 mutants 1, 3, or 5 (G2 m1, G2 m3, and G2 m5) were placed in front of the truncated viral thymidine kinase promoter of herpes simplex virus fused to the luciferase reporter gene. The plasmids were transfected into ␣TC2 cells. KCl, 45 mM; TPA, 300 nM. Luciferase activity is expressed as percentage of the mean value, in each experiment, of the activity measured in the respective control (no treatment). When compared with 4xG2T81Luc (100 Ϯ 6%), the basal activity was not changed by mutants 1 or 5 and was 677 Ϯ 65% for 4xG2 m3T81Luc, as has been reported previously (20). Values are mean Ϯ S.E. of three independent experiments, each done in duplicate. translocation of NFATp was blocked by FK506 (167 nM), the activation of G2 transcriptional activity by membrane depolarization and calcium influx was markedly reduced. This suggests that calcium-induced activation of G2 transcriptional activity depends on calcineurin phosphatase activity and NFATp binding. The regulatory sequences within G2 that are required for depolarization responsiveness were characterized. Four copies of G2 wild-type sequence or mutants 1, 3, and 5 were placed in front of the heterologous thymidine kinase promoter of herpes simplex virus (pT81Luc). In contrast to the G2 wildtype sequence, mutants 1, 3, and 5 failed to confer depolarization responsiveness (Fig. 6). Thus, the depolarization responsiveness of G2 was abolished by mutant 1 as well as by mutants 3 and 5, which interfere with the binding of HNF-3␤ (mutant 1) as well as with the depolarization-induced NFATp binding (mutants 3 and 5). Membrane depolarization was unable to stimulate transcription through four copies of a high affinity AP-1 site or a composite IL-2 NFAT:AP-1 site (data not shown). Taken together, the results of the present study suggest that the depolarization responsiveness of G2 may depend on the binding of both HNF-3␤ and NFATp. It is noteworthy that, in contrast to membrane depolarization, the protein kinase Cactivating phorbol ester TPA stimulated transcription through mutant 3 at least as well as through G2 wild-type sequence (Fig. 6), consistent with published data (20). This indicates that distinct proteins mediate protein kinase C and depolarization responsiveness of the G2 element. Whereas HNF-3␤ and an Ets-like protein have been shown to confer Ras and protein kinase C responsiveness (20), NFATp and HNF-3␤ are required for the stimulation of G2 activity by membrane depolarization and calcium influx.
Role of the G2 Element and NFATp in the Regulation of the Intact Glucagon Promoter by Membrane Depolarization-To study the role of the G2 element in depolarization-induced activation of the intact glucagon promoter, glucagon-reporter fusion genes were used that carry an internal deletion of the G2 element without (Ϫ350⌬GluLuc) or with an additional internal deletion inside the CRE (Ϫ350(2⌬)GluLuc). Deletion of the G2 element markedly reduced the depolarization-induced increase in transcriptional activity of the glucagon promoter in ␣TC2 cells (Fig. 7A, left panel), suggesting that G2 is required for depolarization responsiveness of the glucagon promoter. However, interpretation of the data is complicated by the marked decrease in basal activity by the G2 deletion (Fig. 7A, left  panel). Therefore, similar transfection experiments were performed in HIT cells. In this pancreatic islet ␤-cell line, the internal deletion of G2 caused only a slight decrease in basal activity (Fig. 7A, right panel). The internal deletion of G2 reduced the depolarization-induced activation of the glucagon promoter by 76% (Fig. 7A, right panel), suggesting again that G2 is required for depolarization responsiveness of the glucagon promoter. Similar results were obtained with the double deletion mutant (Fig. 7A, right panel). Depolarization responsiveness of the glucagon promoter was inhibited in HIT cells by the internal deletion of G2 (Fig. 7A, right panel) to the same degree (by about 75%) as by the deletion inside the CRE (Fig.  1B), suggesting that both G2 and the CRE are required for full activation of glucagon gene transcription by membrane depolarization; furthermore, this suggests that the depolarization responsiveness mediated by one element to the promoter depends in part on the other element.
To explore further the role of the G2-binding transcription factor NFATp for glucagon promoter activity, FK506 and the expression vector pBK-CMV-NFATpDBD were used. When nuclear translocation of NFATp was blocked by FK506, the activation of glucagon promoter activity by membrane depolarization and calcium influx was markedly reduced (Fig. 7B), suggesting that calcium-induced activation of the glucagon promoter depends on NFATp binding. In these experiments 292 base pairs of the glucagon promoter were used which lack the CRE, because the transactivation of serine 119-phosphorylated CREB is inhibited by FK506 (13)(14)(15). Although the minimal DNA-binding domain of NFATp is sufficient for cooperation with Fos and Jun proteins, it lacks an intrinsic transactivation domain (25) and, in the absence of interaction with AP-1 on composite NFAT-binding sites, is thus expected to inhibit NFAT transactivation when overexpressed by blocking NFATbinding sites (25). As shown in Fig. 7C, overexpression of the NFATp DNA-binding domain inhibited the depolarization-induced activation of glucagon gene transcription by 40%. The inhibition of depolarization-induced activation of the glucagon promoter by NFATpDBD was less than that by G2 deletion or FK506, which may be explained by a lower efficiency of NFAT-pDBD overexpression. Taken together, all these data suggest a role of the G2 element and of NFATp in the regulation of the intact glucagon promoter by membrane depolarization. DISCUSSION In this report we describe experiments that have led to the identification and characterization of a second calcium response element within the glucagon promoter. Evidence is presented that the transcription factor NFATp is expressed in glucagon-producing islet cells and is directed to the nucleus by calcium/calcineurin in response to membrane depolarization. NFATp is known to cooperate with AP-1 proteins in T cells (16 -19). By contrast, this study shows a novel pairing of NFATp with the cell-specific transcription factor HNF-3␤, building a novel calcium response element.
Calcium influx through voltage-dependent calcium channels stimulates both glucagon secretion (9) and biosynthesis through gene transcription (3,13,15). The present study shows that, in addition to the glucagon CRE described previously (3,13,15), the glucagon G2 element is a second calcium response element of the glucagon gene. As is indicated by the effects of internal deletions of the CRE and G2, both elements are required for full depolarization responsiveness of the glucagon gene, similar to what has been shown for the calcium regulation of the c-fos gene by the c-fos CRE and the c-fos serum response element (5,48). Using primary hippocampal neurons, Bading et al. (49) demonstrated that calcium, depending on its mode of entry into neurons, activates distinct signaling pathways that lead to gene activation via different cis-acting regulatory elements. Furthermore, differential activation of transcription factors induced by calcium response amplitude and duration has been shown in B lymphocytes (50). Thus, it is possible that depending on the type of the calcium signal the G2 element and the glucagon CRE could subserve distinct functions in calcium signaling to the glucagon gene. The ability of membrane depolarization to potentiate cAMP-induced glucagon gene transcription in a synergistic fashion (3) may be physiologically important. Through synergistic interaction the calcium response of G2 and the CRE may participate in a cross-talk between intracellular signaling systems that in A cells integrates multiple stimuli to an appropriate transcrip-tional response of the glucagon gene in order to maintain glucose homeostasis.
NFATp was originally characterized as the preexisting cytoplasmic component of a transcription factor in T cell hybridomas implicated in the induction of several cytokine genes during the immune response (19). NFATp has been molecularly cloned (19,32) and is now known to be a member of a family of transcription factors that contain at least four members and share a Ϸ300-amino acid DNA-binding domain distantly related to the Rel homology region (18,19,51). The NFAT family of transcription factors is a target for the clinically important immunosuppressant drugs cyclosporin A and FK506, which inhibit the calcium-/calmodulin-dependent serine/threonine phosphatase calcineurin (47,52,53). In resting T cells, NFATp resides in the cytoplasm and is fully phosphorylated; following stimulation, it rapidly becomes dephosphorylated by calcineurin activated by an increase in cytosolic calcium concentration, and NFATp then translocates to the nucleus (19, 26, 46, 54 -56). Although some cells lack the requisite mechanisms for regulating NFATp subcellular localization (26), our results suggest that NFATp is expressed and regulated in glucagonproducing islet cells in a similar way as in T cells and binds to the glucagon gene. The G2 element within the glucagon promoter contains a typical NFAT-binding site (5Ј-GGAAACTT-3Ј). Using an electrophoretic mobility shift assay, a protein complex with NFATp-like immunoreactivity was found to bind to this site and was detected in cytosolic extracts from unstimulated cells and in nuclear extracts from cells stimulated by membrane depolarization, consistent with depolarization-induced translocation of NFATp from the cytosol to the nucleus. The mutations in G2 mutant 3 were less detrimental for NFATp binding than those of G2 mutant 5, consistent with the finding that the bases exchanged in G2 mutant 3 are less conserved (72-78%) than those of G2 mutant 5 (100%) in aligned sequences derived by PCR-based site selection of an optimal NFAT-binding site (57). Furthermore, NFAT makes sequence-specific major groove contacts in the 5Ј-half-site (58), which is altered in G2 mutant 5 but not in G2 mutant 3. The antiserum used is specific for NFATp (36) and completely blocked the depolarization-induced nuclear protein binding, providing no evidence that in addition to NFATp other members of the NFAT family of transcription factors may be involved. NFATp has been shown before to be expressed in a wide variety of tissues including pancreas (36,51,59). The expression of NFATp in pancreatic islet ␣TC2 cells was confirmed by RT-PCR of NFATp mRNA as well as by Western blotting. Inhibition of calcineurin phosphatase activity by FK506 blocked both the depolarization-caused induction of nuclear NFATp binding and G2-mediated transcription. The latter was also prevented by mutating the binding sites for NFATp or HNF-3␤ within G2. Taken together, these data suggest that the molecular mechanism by which glucagon gene transcription is induced by membrane depolarization and calcium influx in pancreatic islet cells through the G2 element may be consistent with a model in which calcium activates calcineurin phosphatase activity which then, probably by direct dephosphorylation, induces the translocation of NFATp from the cytosol to the nucleus where it cooperates with HNF-3␤ at the G2 element to stimulate transcription (Fig. 8).
The N-terminal transactivation domain of NFATp has been shown to recruit the coactivators p300/CBP (60), which are also bound by CREB after phosphorylation in response to calciuminduced signaling (61). NFATp-CBP interactions may involve the CREB-binding domain and the cysteine-histidine-rich region 3 of CBP (60). Thus, cooperative recruitment of a common coactivator by CREB, binding to the CRE, and NFATp, binding together with HNF-3␤ to G2, is a potential explanation for the effects of the internal deletion of either the CRE or G2 on glucagon promoter activity in HIT cells, which indicate that depolarization responsiveness conferred to the glucagon promoter by G2 or the CRE depends in part on the other element.
Although NFATp appears to be regulated by calcium/calcineurin in both T cells and islet cells, there are significant differences. First, in T cells, NFATp is recruited by an inositol trisphosphate-induced release of calcium from intracellular stores and capacitative calcium entry (19,55), whereas in this study in islet cells NFATp was mobilized by membrane depolarization and calcium influx through voltage-dependent, Ltype calcium channels. This may be of general significance also in neurons (36,51). Second, this study shows a novel pairing of NFATp. An important component of transactivation by NFATp family proteins in T cells is their cooperation in the nucleus with Fos and Jun AP-1 proteins (16 -19). The interaction with AP-1 is essential for NFAT transcriptional activity. Since AP-1 is newly induced upon T cell receptor activation through separate signals, two signaling pathways are needed to trigger the IL-2 NFAT site for a full transcriptional response, Ras/protein kinase C for AP-1 induction and calcium/calcineurin for NFAT activation (19,47,52,53,62). In the IL-2 enhancer, a consensus recognition site for NFAT is closely juxtaposed with a nonconsensus AP-1 site (18,19). NFAT and AP-1 together form a highly stable cooperative complex (16 -19, 58, 63). A structural model has been proposed (58,63), wherein the so-called insert region of NFAT lies closest to the spacer region of c-Fos/c-Jun, allowing a single arginine residue in the c-Jun spacer (Arg-285) to make an especially important contact. This study shows a novel pairing of NFATp with HNF-3␤. Both proteins bind to closely associated sites within the G2 element. HNF-3␤ binds immediately 3Ј to the NFAT site and, thus, at a relative position occupied by AP-1 proteins in composite NFAT:AP-1 sites (19). The binding of NFATp or HNF-3␤ was not affected when the binding of HNF-3␤ (in labeled mutant 1) or NFATp (in labeled mutant 5), respectively, was eliminated, providing no evidence that, as with AP-1, the interaction between NFATp and HNF-3␤ involves cooperative binding; it appears to be based on a functional synergism that may or may not include a direct protein-protein interaction. HNF-3␤ is a member of the winged helix family of transcription factors (64). It is expressed early in development (64 -67). In adults, it is expressed in the liver (68), lungs, small intestine (68,69), exocrine pancreas (70), and pancreatic islets (20,41,70,71), being involved in cell-specific gene transcription in these endoderm-derived tissues. Instead of integration into Ras/protein kinase C pathways through AP-1 proteins (19,47,52,53,62), the interaction between NFATp and HNF-3␤ may not confer the requirement for an input from signaling pathways others than those regulating NFATp. Thus, the present study shows a new combinatorial usage of NFATp with HNF-3␤, which ties the calcium/calcineurin/NFAT pathway directly to cell-specific gene transcription.
The present findings expand the functions of the transcription factor HNF-3␤ to include a role in calcium-regulated gene transcription. HNF-3␤ was first defined as a transcription factor involved in developmental and cell-specific gene transcription (see above). Subsequently, it has been suggested that HNF-3␤ synergizes with the glucocorticoid receptor in the liver to confer glucocorticoid responsiveness to the tyrosine aminotransferase gene (72) and with an Ets-like transcription factor in islet cells to confer Ras and protein kinase C responsiveness to the glucagon gene (20). The results presented here suggest that HNF-3␤ potentiates transcriptional activation by the calcium-regulated transcription factor NFATp. Thus, HNF-3␤ is able to cooperate with several regulated transcription factors and in this way seems to function as a mediator of environmental signals in endoderm-derived tissues in which it is expressed.
There are several examples where multiple signaling pathways focus on a single DNA control element, although the mechanisms by which these signals are integrated or selected are different (1,2,73). The binding sites within G2 of calciumregulated NFATp and of the Ets-like proteins, probably regulated by protein kinase C and Ras (20), overlap, suggesting that these proteins compete for binding to G2 and thereby determine which signal is mediated through G2.
Diverse calcium response elements have already been described, including elements binding the transcription factors CREB (1-4), serum response factor (2,5), and C/EBP␤ (6). The calcium-responsive glucagon G2 element differs from these elements by DNA sequence, interacting proteins, and the lack of evidence for an involvement of calcium-/calmodulin-dependent protein kinases in signaling to this element. It also differs from composite NFAT-binding sites in the IL-2 and other cytokine genes (18,19) by a novel pairing of NFATp. In this sense, the depolarization-induced combination of NFATp with the cell lineage-specific transcription factor HNF-3␤ at the glucagon G2 element represents a novel calcium response element.