GABP Mediates Insulin-increased Prolactin Gene Transcription*

The insulin-response element from the prolactin gene is identical to the Ets-binding site, and dominant-nega-tive Ets protein inhibits insulin-increased prolactin gene expression. Immunoblotting identified the Ets-re-lated transcription factor GABP in nuclear extracts from GH cells. Expression of GABP (cid:97) and GABP (cid:98) 1 squelches insulin-increased prolactin gene expression. GABP (cid:97) and GABP (cid:98) 1 bind the insulin-response element of the prolactin promoter, and anti-GABP (cid:97) and anti-GABP (cid:98) 1 antibodies supershift a species seen with nuclear extracts from GH cells. GABP (cid:97) immunoprecipitated from insulin-treated, 32 P-labeled GH cells was phosphorylated 3-fold more than GABP (cid:97) from control cells. There was no increase in phosphorylation of GABP (cid:98) in response to insulin. Mitogen-activated protein (MAP) kinase activity is increased 10-fold in insulin- treated GH4 cells. MAP kinase immunoprecipitated from control cells does not phosphorylate GABP (cid:97) while MAP kinase immunoprecipitated from insulin-treated cells shows substantial phosphorylation of GABP (cid:97) . These studies suggest that GABP mediates insulin-in- creased transcription of the prolactin gene. GABP may be regulated by MAP kinase phosphorylation. The activation of gene transcription by hormones that func-tion through protein-tyrosine kinase receptors is not well un-derstood in comparison with that mediated by other classes of hormones.

The activation of gene transcription by hormones that function through protein-tyrosine kinase receptors is not well understood in comparison with that mediated by other classes of hormones. The receptors for the steroid-thyroid hormones are transcription factors that are activated by hormone binding (1). G s protein-coupled receptors activate gene transcription following hormonal initiation of a cascade ending in the phosphorylation of CREB/ATF transcription factors (2). Recently, numerous cytokines have been shown to activate transcription via phosphorylation of cytosolic ISGF-3 proteins that then become nuclear localized transcription factors (3). The responsiveness of genes to each of these classes of hormones is dependent on the presence in the gene of the appropriate DNA sequence to which the activated transcription factor binds. Neither the hormone-responsive DNA element nor the transcription factors activated by protein-tyrosine kinase receptors are known.
Recently, we have identified an insulin-response element in the prolactin promoter that is identical to the binding site for the Ets-related transcription factors (4). This element also mediates the insulin sensitivity of the thymidine kinase and somatostatin promoters in both HeLa and GH4 cells and confers insulin responsiveness to the mammary tumor virus promoter when it is added to that promoter at Ϫ88. Further, the increase in the transcription of these genes in insulin-treated cells was inhibited by expression of a dominant-negative Ets protein (5). These studies identify the predominant Ets-related protein of GH4 cells, GABP␣, and suggest that GABP mediates insulinincreased prolactin gene expression. Phosphorylation of GABP by MAP 1  was from DuPont NEN. All enzymes and linkers were obtained from either New England Biolabs or from Boehringer Mannheim and, unless otherwise indicated, were used under conditions recommended by the suppliers. Oligonucleotides were purchased from Operon. Duplex poly(dI⅐dC) was obtained from Pharmacia Biotech Inc. Antibodies to MAP kinase (anti-Erk-1 and anti-Erk-2), antibody to the DNA binding domain of cEts-1 (pan-Ets), and horseradish peroxidase-conjugated goat anti-rabbit secondary antibody were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies directed against GABP␣ and GABP␤1 were the generous gift of Dr. S. L. McKnight (Tularik, South San Francisco, CA). Reagents used for gel electrophoresis were purchased from Fisher Scientific. Protein A agarose, acetyl-CoA, and silica gel plates were obtained from Sigma. Dulbecco's modified Eagle's medium containing 4.5 g/liter glucose (DMEM) was from Life Technologies, Inc., and iron-supplemented calf serum was obtained from Hyclone Laboratories. Triton X-100, reagents for enhanced chemiluminescence, and BCA reagent were from Pierce. All other reagents were of the highest purity available and were obtained from Sigma, Behring Diagnostics, Bio-Rad, Eastman, Fisher, or Boehringer Mannheim.
Plasmids-The construction of pPrl-CAT plasmids containing Ϫ173/ ϩ75 of prolactin 5Ј-flanking DNA was described (6). The human insulin expression vector, pRT3HIR2, was the gift of Dr. J. Whittaker (Stony Brook, NY). The plasmids CMV-GABP␣ and CMV-GABP␤ were provided by Dr. C. Thompson (Carnegie Institute, Baltimore, MD). The Elk-1 and SRF expression vectors were the gift of Dr. R. Treisman. The GST-GABP␣ expression plasmid was constructed by ligating a bluntended BamHI/XhoI fragment from the GABP␣ cDNA (plasmid F27, Dr. S. L. McKnight) into the blunt-ended EcoRI site of pGEX 2T (Pharmacia). The fusion protein consists of the GST protein (26 kDa) and GABP␣-(76 -336) (30 kDa) that contains the three potential MAP kinase phosphorylation sites.
Western Immunoblot Analysis-GH4 cells were harvested after 48 h with or without 1 g/ml insulin, and nuclei were prepared as described (7). A nuclear extract was prepared by disrupting the nuclei with 400 mM KCl in a buffer containing 15% glycerol, 25 mM Tris, pH 8, 10 mM ␤-mercaptoethanol, 0.5 mM EDTA, and 0.05% Triton X-100. SDS-polyacrylamide gel electrophoresis was performed using 12% gels (8). The proteins were then blotted to nitrocellulose membranes (Micron Separations) in Towbin's buffer (25 mM Trizma base, 192 mM glycine, and 20% methanol). Immunoblotting using enhanced chemiluminescence was performed as described by the manufacturer (Pierce).
Analysis of Prolactin Promoter Responsiveness Using Transient Transfection-Electroporation experiments and CAT assays were performed as described (9). GH4 cells were placed for 24 h in DMEM containing 10% hormone-depleted serum (9) and harvested with an EDTA solution, and 20 -40 ϫ 10 6 cells were used for each electroporation. Trypan blue exclusion before electroporation ranged from 95 to * This work was supported by National Institutes of Health Grant DK43365. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Present address: Mt. Sinai Medical Center, Asher Levy Place, New York, NY 10029.
99%. The voltage of the electroporation was 1550 V. This gives trypan blue exclusion of 70 -80% after electroporation. A Rous sarcoma virus-␤Gal expression plasmid was used to control for differences between electroporations as described (9). The transformed cells were then plated in multiwell dishes (Falcon Plastics) at 5 ϫ 10 6 cells/9-cm 2 tissue culture well in DMEM with 10% hormone-depleted serum. Cells were refed with DMEM with 10% hormone-depleted serum Ϯ insulin at 24 h. After 48 h, the flasks were washed three times with normal saline and frozen. The cells were harvested in hypotonic lysis buffer using a rubber policeman. CAT activity was assayed as described (10).
Assay of DNA-Protein Binding by Gel Electrophoresis-An oligonucleotide to the prolactin promoter sequence Ϫ106/Ϫ87 was prepared, purified on polyacrylamide gels, and end-labeled with [ 32 P]dCTP. The sequence of this oligonucleotide is 5Ј-TCTTAATGACGGAAATAGAT-3Ј. Labeled prolactin 5Ј-flanking DNA was then used in mobility shift experiments with unlabeled nuclear extracts performed as described (7). Two g of nuclear extract were incubated at 25°C for 30 min with 10,000 cpm (10 -20 fmol) of 32 P-labeled Prl Ϫ106/Ϫ87. The protein-DNA complexes were then analyzed by electrophoresis on a 4% polyacrylamide gel in Tris/acetate/EDTA buffer.
Phosphorylation of GABP␣ by MAP Kinase-GH4 cells were treated for 5 min with 1 g/ml insulin or were left untreated as controls. They were then washed and frozen at Ϫ70°C. The cells were lysed in a buffer consisting of 50 mM HEPES, pH 7.5, 1% Triton X-100, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 10% glycerol, 1 mM Na 3 VO 4 , 50 mM Na 4 P 2 O 7 , 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin. Immunoprecipitations were performed for 90 min at 4°C in this buffer using 200 g of protein. Antibody to MAP kinase was obtained from Santa Cruz, and 1 g was used for each immunoprecipitation. The kinase assay was performed with immunoprecipitated MAP kinase and 0.1 g of GST protein. The fusion protein consists of the GST protein (26 kDa) and GABP␣-(76 -336) (30 kDa) that contains the three potential MAP kinase phosphorylation sites. The assay was performed in a buffer containing 25 mM HEPES, pH 8.0, 50 M ATP, and 10 Ci of [ 32 P]ATP, 10 mM MgCl 2 , and 1 mM dithiothreitol (11). The supernatant was precipitated with 10% trichloroacetic acid and electrophoresed on a 10% SDS-polyacrylamide gel.
Labeling and Immunoprecipitation of GABP␣ and GABP␤-GH4 cells in 9-cm 2 tissue culture wells were incubated for 2 h with phosphate-free DMEM containing 10% dialyzed, charcoal-treated calf serum. They were then washed and incubated for 2 h in phosphate-free DMEM containing 10% dialyzed, charcoal-treated calf serum with 0.5 mCi/ml [ 32 P]H 3 PO 4 . Insulin was then added and the incubation was continued for 1 h. The cells were rapidly chilled, washed 3 times with ice-cold saline, and frozen at Ϫ70°C. The cells were then processed and immunoprecipitated as described above.

RESULTS AND DISCUSSION
Immunoblot analysis of GH4 cell nuclear extracts was performed to determine which of the Ets-related transcription factors might mediate the effects of insulin on prolactin gene expression (Fig. 1). A pan-Ets antibody was used to visualize Ets-related proteins in GH4 cell nuclear extracts. This antibody was raised against the conserved DNA binding domain of Ets-1, and it demonstrates broad cross-reactivity with Ets family proteins. One band of approximately 51 kDa was visible using either 1 or 3 g of nuclear extract. This is the same size as the previously identified Ets-related protein GABP␣. GABP␣ is a subunit of the heteromeric transcription factor, GABP. The other subunit is GABP␤, a notch-related protein (12). GABP was originally identified as the transcription factor that binds to a purine-rich cis-regulatory element required for VP16-mediated activation of herpes simplex virus immediate early gene (13). A separate set of filters was therefore analyzed using antibodies against GABP␣ or GABP␤1. One band, identical in size to that seen using the pan-Ets antibody, is seen with anti-GABP␣. Anti-GABP␤1 antibody reveals two bands. The lower, more intense band migrates with an apparent molecular mass of 43 kDa and thus likely represents GABP␤1. The levels of GABP␣ and GABP␤1 are not significantly different in nuclear extracts from control and insulin-treated cells (data not shown).
Experiments were performed to determine if GABP might be involved in the response of the prolactin gene to insulin. Cotransfection of small amounts of GABP␣ expression vector inhibited the insulin-induced increase in prolactin gene expression (Fig. 2) to 25% of that seen in control transfections (22fold) or in transfections with expression vector (21-fold). Cotransfection with an expression vector for GABP␤1 resulted in a 2-fold increase in the insulin-stimulated prolactin-CAT expression (45-fold). This suggests that levels of GABP␤1 may be limiting in GH cells. GABP␤1 was shown to increase the affinity of GABP␣ binding to DNA (12). Thus, it might be expected that a cotransfection with both GABP␣ and GABP␤1 would further affect insulin-increased prolactin-CAT expression. Coexpression of vectors for both GABP␣ with GABP␤1 completely eliminated any effect of insulin (Fig. 2) but has no significant effect on basal or EGF-increased prolactin gene expression. Basal expression of prolactin-CAT was not affected in these experiments (1.15 Ϯ 0.25% acetylation/10 g of protein in control cultures versus 1.07 Ϯ 0.44% acetylation/10 g of protein in GABP␣ ϩ GABP␤-cotransfected cells). Further, EGF-increased expression of prolactin-CAT was also not affected by cotrans- Similar results were seen previously with SRF and Elk-1 (14). The effects of Elk-1 and SRF were attributed to squelching caused by titration of limiting components or formation of non-functional transcription complexes (14). The squelching of prolactin gene expression due to overexpressed GABP appears to be specific since overexpression of SRF and Elk-1 has no effect on prolactin gene expression in GH cells (Fig. 2) although the Elk-1 binding site of c-Fos is similar to the insulin-response element of the prolactin gene.
Since expression of GABP was able to block insulin responsiveness of the prolactin promoter, the binding of GABP␣ and GABP␤1 to the insulin-response element of the prolactin gene was examined. Nuclear extracts from GH cells produced a characteristic mobility shift pattern (Fig. 3, lane 1). This gelshift pattern was identical with that previously shown to be inhibited by low levels of non-radioactive competitor (10). Gelmobility shift experiments performed with bacterially expressed GABP␣ showed three retarded bands with the prolactin insulin-response element (Fig. 3, lane 2). The upper band is a complex formed with a bacterially expressed protein since it was present in extracts from unprogrammed bacteria (not shown) The other two bands are complexes containing GABP␣. Similar complexes were previously reported to be formed between DNA and GABP␣ monomer and GABP␣ dimer (12). Only the bacterial protein band is seen with bacterially produced GABP␤1 (lane 3). This was expected since previous studies had shown that GABP␤1 is not a DNA binding protein (12). When GABP␣ and GABP␤1 were added together (lane 4), the bands corresponding to GABP␣ were no longer visible and an abundant, more slowly migrating band was seen. This band corresponds to an abundant band seen with nuclear extracts from GH cells. The identity of this band as consisting of GABP␣ and GABP␤1 is confirmed in the experiment shown in Fig. 3 (right). Since insulin receptor is a tyrosine-protein kinase that is activated by insulin binding, it is thought that activation of gene transcription by insulin may be the end product of a phosphorylation cascade. Therefore, we examined the phosphorylation of GABP in response to insulin in 32 P-labeled GH cells. The phosphorylation of GABP␣ was increased 3-fold in 1 h in insulin-treated cells as compared with control cells (Fig. 4A). GABP␤ co-immunoprecipitated with GABP␣ in this experiment shows no increase in response to insulin. Immunoprecipitation with anti-GABP␤1 confirms this observation. GABP␤1 phosphorylation was not significantly increased by insulin treatment (20% above control) while the co-immunoprecipitated GABP␣ is increased 3-fold by insulin.
MAP kinase activation was shown to be required for several types of insulin responses in numerous systems (15). Further, Elk-1, an Ets-related transcription factor, was shown to be activated by MAP kinase phosphorylation (16). Pointed-P2, an Ets-related protein from Drosophila, is phosphorylated by MAP kinase in the sevenless signal transduction pathway (17). Our studies 2 suggest that insulin activation of prolactin gene expression in GH cells is MAP kinase-dependent since all factors that inhibit insulin-increased prolactin gene expression also inhibit MAP kinase activation. Therefore, MAP kinase activated by insulin might phosphorylate GABP␣. The representative increase in MAP kinase activity in cell lysates from insulin-treated cells is shown (Fig. 4B) fore, GST-GABP␣ fusion protein containing the three MAP kinase phosphorylation sites was prepared and used in a kinase assay with MAP kinase immunoprecipitated from control or insulin-treated GH4 cells. GABP␣ was phosphorylated only by MAP kinase immunoprecipitated from insulin-treated cells (Fig. 4C). This shows that GABP␣ is a substrate for MAP kinase and suggests that MAP kinase phosphorylation of GABP may be functionally important.
GABP was shown to be important for enhancement of transcription of the herpes simplex virus immediate early gene, but its physiological role in uninfected cells is unknown. Our studies show that GABP mediates the insulin response of the prolactin gene. Since GABP is widely distributed, these results could be significant to understanding insulin regulation of other genes. Analysis of 22 insulin-responsive promoters has identified potential Ets-response elements in all of these. For some of these, the Ets-response element is in a region defined by deletion analysis to be important for the effects of insulin (5). The insulin-mediated increase in the transcription of all three genes, prolactin, somatostatin, and thymidine kinase, that we have studied is inhibited by dominant-negative Ets protein. This indicates that GABP may be implicated in the regulation of other insulin-responsive genes.
Ets-related transcription factors such as GABP are often found in large complexes with other transcription factors. For example, Ets-1 and Sp-1 interact to synergistically activate the human T-cell lymphotrophic virus long terminal repeat (18). Although this report demonstrates that GABP is necessary to the insulin effect, it may not be sufficient. The insulin responsiveness of the prolactin gene can be eliminated by mutation of two Ets motifs at Ϫ96/Ϫ87 and Ϫ76/Ϫ67 of the prolactin promoter (4). These mutations have little effect on basal prolactin gene transcription. However, mutation of Ϫ101/Ϫ92 of the prolactin promoter eliminates the effect of insulin and reduces basal prolactin gene expression by Ͼ100-fold. Clearly, another protein(s) interacts at this sequence and is important both for basal prolactin gene expression and the effect of insulin. It is likely that GABP is complexed with this protein(s) in the prolactin promoter and that this complex is important to the increase in prolactin gene expression seen in insulin-treated cells. FIG. 4. A, immunoprecipitation of GABP␣ and GABP␤ from 32 Plabeled GH4 cells. GH cells were labeled with 32 P and incubated with insulin for 1 h as described under "Experimental Procedures." Labeled proteins were then precipitated with an antibody to GABP␣ (lanes 1 and 2) or an antibody to GABP␤1 (lanes 3 and 4). Immunoprecipitations with control lysates are in lanes 1 and 3 while insulin-treated cell lysates are in lanes 2 and 4. The migration of GABP␣ and GABP␤ is indicated on the left while the migration of molecular weight markers is shown on the right. This experiment was repeated twice with similar results. B, insulin activation of MAP kinase. GH cells were transfected with 1 g of a vector expressing a human influenza hemagglutinintagged MAP kinase and with 5 g of pRT3HIR2. After a 24-h incubation in insulin-depleted serum containing medium, the cultures were incubated with insulin for 5 min or left untreated as controls. The cells were harvested and immunoprecipitated with anti-hemagglutinin antibody (Boehringer Mannheim) as described under "Experimental Procedures." The kinase activity of the immunoprecipitated MAP kinase was assayed as described under "Experimental Procedures" using myelin basic protein (Sigma) as a substrate.