A Nucleoprotein Complex containing CCAAT/Enhancer-binding Protein β Interacts with an Insulin Response Sequence in the Insulin-like Growth Factor-binding Protein-1 Gene and Contributes to Insulin-regulated Gene Expression*

Highly related insulin response sequences (IRSs) mediate effects of insulin on the expression of multiple genes in the liver, including insulin-like growth factor binding protein-1 (IGFBP-1) and phosphoenolpyruvate carboxykinase (PEPCK). Gel shift studies reveal that oligonucleotide probes containing an IRS from the IGFBP-1 or PEPCK gene form a similar complex with hepatic nuclear proteins. Unlabeled competitors containing the IGFBP-1 or PEPCK IRS or a binding site for C/EBP proteins inhibit the formation of this complex. Antibody against C/EBPβ (but not other C/EBP proteins) supershifts this complex, and Western blotting of affinity purified proteins confirms that C/EBPβ is present in this complex. Studies with affinity purified and recombinant protein indicate that C/EBPβ does not interact directly with the IRS, but that other factors are required. Gel shift assays and reporter gene studies with constructs containing point mutations within the IRS reveal that the ability to interact with factors required for the formation of this complex correlates well with the ability of insulin to regulate promoter activity via this IRS (r = 0.849,p < 0.01). Replacing the IRS in reporter gene constructs with a C/EBP-binding site (but not an HNF-3/forkhead site or cAMP response element) maintains the effect of insulin on promoter activity. Together, these findings indicate that a nucleoprotein complex containing C/EBPβ interacts with IRSs from the IGFBP-1 and PEPCK genes in a sequence-specific fashion and may contribute to the ability of insulin to regulate gene expression.

Highly related insulin response sequences (IRSs) 1 have been found to mediate effects of insulin on the expression of a number of genes in the liver, including phosphoenolpyruvate carboxykinase, insulin-like growth factor binding protein-1 (IG-FBP-1), glucose-6-phosphatase, and apolipoprotein CIII (1)(2)(3)(4). This observation has suggested that insulin may exert effects on the expression of multiple hepatic genes through a common mechanism (5). Signaling via phosphatidylinositol 3Ј-kinase is critical for the ability of insulin to suppress activity of the IGFBP-1, PEPCK, and Glu-6-Pase promoters (6 -8) and we have reported that protein kinase B can mediate effects of insulin on basal IGFBP-1 promoter activity via an IRS downstream from phosphatidylinositol 3Ј-kinase (6). HNF-3/forkhead proteins can bind to oligonucleotide probes containing the insulin response sequences from the IGFBP-1 and PEPCK genes (9,10). However, these interactions also involve sequences flanking the IRS, indicating that HNF-3 proteins are not likely to mediate sequence-specific effects of insulin on promoter activity via an IRS (10,11). Recent reports have shown that another subgroup of forkhead/winged-helix transcription factors, including FKHR, FKHRL1, and AFX, can stimulate transcription through an IRS when they are overexpressed in cells, and that phosphorylation by protein kinase B can disrupt IRS-dependent transactivation by these proteins (12)(13)(14)(15)(16). However, studies to date have not demonstrated interaction between endogenous FKHR, FKHRL1, or AFX in nuclear extracts and an IRS in binding assays. Also, Hall et al. (17) have suggested that these forkhead family members may not interact with IRSs with the appropriate sequence-specificity to account for the ability of insulin to regulate promoter activity through an IRS in cells, indicating that other (as yet unidentified) factors also may contribute to the ability of insulin to suppress promoter function through an IRS (17).
CAAT/enhancer-binding proteins (C/EBPs) are enriched in the liver where they play an important role in the regulation of gene expression (18) and studies in knockout mouse models reveal that C/EBP proteins play a critical role in the regulation of hepatic glucose production during development and in fasting and insulin-deficient mice (19 -22). Cell culture studies have shown that C/EBP proteins also are critically involved in the differentiation of adipocytes (23)(24)(25), indicating that they play an important role in the integrated regulation of metabolic processes (26). We have previously reported that a nucleoprotein complex that interacts with an IRS in the IGFBP-1 promoter may involve members of the C/EBP family of transcription factors, based on gel shift studies with an unlabeled competitor containing a known C/EBP-binding site (11). We now report that a similar complex also interacts with the IRS in the PEPCK gene and that this complex contains C/EBP␤. Gel shift and reporter gene studies with constructs containing point mutations within the IRS indicate that this complex interacts with an IRS in a sequence-specific fashion that correlates well with the ability of insulin to regulate promoter activity. Also, replacing the IRS with a consensus C/EBP-binding site (but not an HNF-3-binding site or cAMP response element (CRE)) maintains the effect of insulin on promoter activity. Taken together, these studies indicate that a nucleoprotein complex containing C/EBP␤ can interact with IRSs in both the IGFBP-1 and PEPCK genes and may contribute to the ability of insulin to suppress promoter function.
Nuclear Extracts and Gel Shift Assay-H4IIE and HepG2 cells were grown to near confluency in polystyrene dishes and refed with Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 1 g/liter of fatty acid-free bovine serum albumin (Sigma) 24 h before harvest. Nuclear extracts were prepared from H4IIE hepatoma cells after the method of Dignam as reported (9), and nuclear extracts were prepared from HepG2 cells after Nonidet P-40 detergent lysis (27). Nuclear extracts were prepared from the livers of adult (200 -250 g) male rats after the method of Gorski, as previously reported (28). Protein content was measured by the Bradford dye-binding assay (Bio-Rad).
For gel shift and supershift studies, nuclear extracts were preincubated with unlabeled oligonucleotide competitors or GST-CHOP for 10 min or with antibody for 30 min at 22°C in 10 l of binding buffer containing 50 or 5 g/ml (Figs. 1B, 2B, and 3A) poly(dI-dC)⅐poly(dI-dC). Binding reactions were continued for an additional 30 min following the addition of ϳ1 ϫ 10 4 cpm end-labeled double-stranded oligonucleotide probes, then loaded for 6% native PAGE at 4°C, as previously reported (9). Gel shift conditions reported to optimize the binding of recombinant C/EBP to a probe containing the PEPCK IRS and flanking sequence (29) were used as noted (Fig. 6A). Gels were dried prior to autoradiography and/or phosphorimaging.
Double-stranded oligonucleotides containing binding sites for Sp1, Oct1, and CTF/NF-1 were purchased from Promega, and an oligonucleotide containing a consensus C/EBP-binding site was from Santa Cruz. An oligo containing an SRF site was provided by Dr. R. Prywes, and oligonucleotides containing the high affinity HNF-3 site of the rat transthyretin gene, the PEPCK IRS and flanking sequence (including residues 401-422 upstream from the cap site), or residues 85-122 upstream from the RNA cap site of the rat IGFBP-1 promoter were prepared as previously described (11). Other oligos used in gel shift studies are described in the text or in figures. Oligonucleotide probes were end-labeled with T4 polynucleotide kinase.
After dialysis, the salt concentration was adjusted to 0.1 M KCl based on conductivity, and proteins were loaded onto a heparin-agarose column (Sigma). The column was rinsed with low salt buffer, and proteins were eluted in 30 fractions with a linear gradient of 0.1-1 M KCl in 10 mM Hepes, pH 7.4, buffer containing 10% glycerol, protease inhibitors, and dithiothreitol. Binding activity in every other fraction was measured by gel shift assay using the ⌬IRS.1 oligo as a probe. Fractions forming the nucleoprotein complex containing C/EBP␤ were pooled, dialyzed against 0.1 M KCl, 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 2.5 mM MgCl 2 , 0.1% Nonidet P-40, and conductivity was adjusted to 0.1 M KCl prior to affinity chromatography.
Proteins were eluted with a 0.1-1 M KCl gradient in 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol and binding activity was monitored by gel shift assay. Following dialysis, proteins were re-loaded onto this column with 10 g/ml poly(dI-dC)⅐poly(dI-dC) to reduce nonspecific binding. Proteins were loaded for a third round of affinity chromatography without poly(dI-dC)⅐poly(dI-dC). Gel shift assays were performed with either the ⌬IRS.2 (round 1 and 3) or the ⌬IRS.1 oligo (round 2 of affinity chromatography) labeled as a probe, thereby ensuring that the purified proteins would interact both with the affinity matrix and with two different oligonucleotide probes containing an IRS.
Partially purified proteins were concentrated with a Centricon-10 concentrator (Amicon) and loaded for 4 -20% SDS-PAGE under reducing conditions, then transferred to nitrocellulose membranes for Western blotting with antibody against C/EBP␤. The membrane was probed with a second antibody tagged with horseradish peroxidase and the bound antibody was identified by enhanced chemiluminesence (Amersham Pharmacia Biotech).
Reporter Gene Constructs and Transient Transfection Studies-The SauI/HgaI fragment of the rat IGFBP-1 promoter was cloned into pGL2, and an NheI site was introduced immediately upstream of the insulin response element by site-directed mutagenesis, as previously described (6). The NheI/BamHI fragment containing the insulin response element was excised and replaced with double-stranded oligonucleotides containing a single IRS (⌬IRS.1 and ⌬IRS.2) with/without mutations of residues within the IRS (⌬IRS.1 m1-⌬IRS.1 m6, ⌬IRS.1M, and ⌬IRS.2M), or with oligonucleotides containing a consensus binding site for C/EBP (⌬C/EBP and ⌬C/EBP.2) or HNF-3 proteins (⌬HNF-3) or the CRE from the PEPCK promoter (⌬CRE), as shown in Tables I and  II. A functional or mutated IRS or C/EBP-binding site also was cloned into the polylinker region in pGL2 further upstream, as before (6). HepG2 hepatoma cells were transfected with calcium phosphate precipitates containing 3 g of reporter gene constructs, 2 g of a cytomegalovirus-driven expression vector for ␤-galactosidase, and 5 g of empty vector. Transfected cells were refed with serum-free medium with/ without 100 nM human insulin 18 h prior to lysis and analysis of luciferase activity (6).

RESULTS
Initial studies revealed that a 32 P-labeled double-stranded oligonucleotide probe containing the IGFBP-1 insulin response element (BP1) forms multiple nucleoprotein complexes with nuclear extracts prepared from rat H4IIE hepatoma cells (Fig.  1A, left panel). The formation of one complex (solid arrowhead) is inhibited by a 250-fold molar excess of an unlabeled oligonucleotide competitor containing a high affinity binding site for HNF-3/forkhead proteins (solid arrowhead), and we have shown that this complex contains HNF-3␤, based on supershift studies (9). An unlabeled oligo containing the IRS and flanking sequences from the PEPCK gene partially inhibits the formation of this complex, consistent with studies indicating that this region of the PEPCK gene weakly interacts with HNF-3/forkhead proteins (8,9).
In contrast, the formation of a complex with lower mobility (Fig. 1, solid arrow) is inhibited effectively by both the BP1 and PEPCK oligos, but not by oligos containing binding sites for HNF-3/forkhead proteins, DBP, Sp1, or Oct1. A complex with similar mobility is formed when the oligo containing the IRS and flanking sequences from the PEPCK promoter is used as a probe (Fig. 1A, right panel, open arrow), and the formation of this complex also is inhibited by competitors containing the IGFBP-1 insulin response element or PEPCK IRS, but not by other competitors. This indicates that this competition is specific and that HNF-3 proteins are not required for the formation of this complex.
A complex with similar mobility also is formed with the BP1 probe and nuclear proteins prepared from rat liver (Fig. 1B). As with extracts from H4IIE hepatoma cells, the formation of this complex is inhibited by an excess of oligo containing the PEPCK IRS, but not by a high affinity HNF-3/forkhead-binding site. The formation of this complex is enhanced with 1 g of nuclear proteins prepared from rat liver compared with 10 g of protein from H4IIE hepatoma cells, suggesting some factor(s) required for formation of this complex may be enriched in liver compared with H4IIE cells.
As shown in Fig. 2A, gel shift studies with the probe containing the IGFBP-1 insulin response element and nuclear proteins prepared from H4IIE cells (left panel) or rat liver (right panel) reveal that the formation of this complex (solid arrow) is inhibited by a 250-fold excess of the unlabeled BP1 oligo or an oligo containing a C/EBP-binding site, but not by other com-petitors, suggesting that C/EBP proteins may contribute to the formation of this complex. An excess of the C/EBP oligo (but not other competitors) also inhibits the formation of the corresponding complex that is formed with the probe containing the PEPCK IRS (not shown). As shown in Fig. 2B, polyclonal goat antibody against C/EBP␤ (but not C/EBP␣ or C/EBP␦) supershifts the complex that is formed by liver nuclear proteins and either the BP1 (left panel) or PEPCK (right panel) probe, indicating that C/EBP␤ is present in this complex.
The sequences flanking IRSs in the BP1 and PEPCK probes are not conserved (Fig. 1), suggesting that the formation of this complex involves direct interaction with IRSs in these probes. To explore this possibility, we created an oligonucleotide probe containing a single IRS (CAAAACA) which has been shifted upstream from its location in the BP1 probe to disrupt potential interactions with flanking sequences (⌬IRS.1), or where the IRS is returned to its initial location, but flanking sequences are modified (⌬IRS.2). As shown in Fig. 3A, the ⌬IRS.1 and BP1 probes form a complex with similar mobility. The formation of this complex is inhibited by an excess of unlabeled oligos containing an IRS (⌬IRS.1 and ⌬IRS.2), but not by oligos where the IRS is mutated (⌬IRS.1M and ⌬IRS.2M) (Fig. 3A), supporting the concept that interaction with the IRS is required for the formation of this complex. Supershift studies performed in the presence of an excess of the ⌬IRS.1M competitor reveal that the FIG. 1. Gel shift with oligonucleotide probes. A, H4IIE nuclear extracts. Nuclear proteins prepared from H4IIE hepatoma cells (10 g/lane) were preincubated with/without a 250-fold molar excess of unlabeled competitors prior to incubation with 32 P-labeled oligonucleotide probes containing IRSs and flanking sequences from the IGFBP-1 (BP1) (left) and PEPCK gene (right). Free probe (first lane in each panel) or binding reactions were loaded for 6% native gel electrophoresis. Unlabeled competitors contain the IGFBP-1 insulin response element or PEPCK IRS, or sequences known to bind HNF-3/forkhead proteins, DBP, Sp1, or Oct1. (Note: lanes 2-5 of the left panel were included in a previous publication (11) and are presented with permission of Elsevier Science.) B, nuclear extracts from H4IIE cells and rat liver. Nuclear proteins from H4IIE cells (10 g/lane) (left panel) or rat liver (1 g/lane) (right panel) were preincubated with a 250-fold excess of unlabeled oligonucleotide competitors prior to incubation with the 32 P-labeled probe containing the IGFBP-1 insulin response element. Nucleoprotein complexes and free probe were resolved by gel electrophoresis and identified by autoradiography, as in A.

FIG. 2. C/EBP proteins in complexes formed with BP1 and PEPCK probes.
A, competitive binding studies. Nuclear proteins were prepared from rat H4IIE hepatoma cells (10 g) (left panel) or rat liver (2.5 g) (right panel) and incubated with unlabeled competitors and then the labeled BP1 probe in the presence of 50 g/ml poly(dI-dC) prior to electrophoresis and autoradiography, as before. (Note: lanes 1-4 of the left panel were included in a previous publication (9) and are presented with permission of Academic Press.) B, supershift studies with antibodies for C/EBP proteins. Nuclear proteins from rat liver (1 g/lane) were preincubated with/without the competitor containing a high affinity HNF-3-binding site and then with polyclonal antibodies specific for C/EBP␣, C/EBP␤, or C/EBP␦ prior to incubation with the labeled BP1 (left) and PEPCK (right) probes in the presence of 5 g/ml poly(dI-dC).
⌬IRS.1 probe forms a complex containing C/EBP␤ with nuclear proteins prepared from H4IIE or HepG2 hepatoma cells (10 g) or rat liver (1 g) (Fig. 3B). Studies with the ⌬IRS.2 probe yielded similar results (not shown). These results support the concept that the formation of this complex containing C/EBP␤ involves interaction with the IRS.
To characterize this complex further, we partially purified liver nuclear proteins forming this complex by chromatography with heparin-Sepharose and three rounds of DNA affinity chromatography using a matrix constructed with oligonucleotides containing an array of 3 IRSs and monitored binding activity in column eluates by gel shift assay (Fig. 4A). Goat polyclonal antibody against the C-terminal region of C/EBP␤ (2 g/lane) supershifts the major complex formed with partially purified proteins and the ⌬IRS.1 probe (Fig. 4B, left panel), similar to results with crude nuclear extracts. Studies with 2 l/lane rabbit polyclonal antiserum against the C-terminal region of C/EBP␤ (C/EBP␤ CT ) also supershifts this complex (Fig. 4B, right panel), although not as completely as 2 g of the purified goat antibody. Interestingly, antiserum against the N-terminal region of C/EBP␤ (C/EBP␤ NT ) does not disrupt or supershift this complex, suggesting that this region of C/EBP␤ may be masked. Western blotting indicates that 35-, 32-, and 14-kDa forms of C/EBP␤ (30) are present in these partially purified proteins (Fig. 4C).
As shown in Fig. 5A, competitive binding studies revealed that a 10-fold molar excess of the unlabeled ⌬IRS.1 oligo is sufficient to effectively inhibit the ability of partially purified proteins to form this complex with the ⌬IRS.1 probe (solid arrow). Higher titers of a competitor containing a consensus C/EBP-binding site (TTGCGCAA; ⌬C/EBP) are required to inhibit the ability of the ⌬IRS.1 probe to form this complex. This result indicates that some factor(s) required for the formation of this complex interacts preferentially with the IRS compared with a consensus C/EBP-binding site.
Also shown in Fig. 5A, partially purified proteins form a complex with higher mobility when the oligo containing a C/EBP site is used as a probe (open arrow), and the formation of this complex is competitively inhibited by a 10-fold excess of partially purified proteins were preincubated with/without goat polyclonal antibody (4 g/lane) against the C-terminal region of C/EBP␣, C/EBP␤, or C/EBP␦ prior to incubation with the ⌬IRS.2 probe, then loaded for 6% native gel electrophoresis as before. Right panel, partially purified proteins were incubated with/without goat polyclonal antibody (4 g/lane) against the C-terminal region of C/EBP␤, or with rabbit polyclonal antiserum (2 l/lane) against the C-terminal (C/EBP␤ CT ) or N-terminal (C/EBP␤ NT ) region of C/EBP␤, prior to incubation with the ⌬IRS.2 probe. C, Western blotting. Partially purified proteins were concentrated by centrifugation through a membrane with a 10,000 dalton cut-off, then loaded for SDS-PAGE under reducing conditions, and transferred to nitrocellullose for Western blotting with goat polyclonal antibody against the C-terminal region of C/EBP␤. the unlabeled ⌬C/EBP oligo, but not by the unlabeled ⌬IRS.1 oligo. These findings also support the concept that some other factor(s) in addition to C/EBP proteins contributes to the formation of the complex that is formed with the ⌬IRS.1 probe.
C/EBP proteins form hetero-or homodimers via a leucine zipper dimerization domain and interaction with the DNAbinding domain of both members of the dimer is required for binding to a canonical C/EBP-binding site (31). As shown in Fig. 5B, a recombinant GST fusion protein containing CHOP, a C/EBP family member which contains a leucine zipper but no functional DNA-binding domain (32), prevents partially purified proteins from forming a complex with the probe containing a consensus C/EBP-binding site (⌬C/EBP), but does not prevent the formation of the complex that is formed with the ⌬IRS.1 probe. This result also supports the concept that proteins outside the C/EBP family are involved in forming this complex with the IRS, and indicates that the formation of this complex prevents C/EBP␤ from interacting with other members of the C/EBP family members via its leucine zipper dimerization domain.
Gel shift studies performed under conditions previously found to optimize the binding of C/EBP proteins to the PEPCK probe (10) reveal that recombinant C/EBP␤ binds directly to probes containing a consensus C/EBP site or the PEPCK IRS, but does not bind a probe containing the IGFBP-1 insulin response element (Fig. 6A), consistent with previous studies (10). Similarly, full-length and truncated C/EBP␤ (C/EBP␤ 1-276 and C/EBP␤ 132-276 , respectively) also bind the probe containing a consensus C/EBP site (⌬C/EBP), but do not bind the ⌬IRS.1 probe under our standard binding conditions (Fig. 6B). Together, these results indicate that C/EBP proteins do not interact directly with IRSs in the IGFBP-1 promoter, and that another factor present in nuclear extracts is required for C/EBP␤ to form a complex with either the BP1 or ⌬IRS.1 probe.
Previous studies indicate that a complex containing a ϳ20-kDa protein and C/EBP␣ interacts with probes containing either the CRE or the IRS in the PEPCK promoter, and that this complex has ϳ30-fold greater affinity for the CRE compared with the PEPCK IRS (29). As shown in Fig. 7B, unlabeled oligos containing the PEPCK CRE (⌬CRE) or the CRE together with flanking sequences from the PEPCK gene (CRE) do not inhibit the formation of the complex containing C/EBP␤ that is formed with partially purified proteins and the ⌬IRS.1 probe. This indicates that formation of this complex does not require a factor that interacts preferentially with the PEPCK CRE.
To determine whether the complex containing C/EBP␤ might

FIG. 5. Interaction of partially purified proteins with oligos containing an IRS or consensus C/EBP-binding site.
A, competitive binding studies. Partially purified proteins were preincubated with/without unlabeled competitors prior to incubation with a 32 Plabeled probe containing a single IRS (⌬IRS.1) or a probe where the IRS has been replaced by a consensus C/EBP-binding site (TTGCGCAA; ⌬C/EBP). B, effect of recombinant GST-CHOP. Partially purified proteins (2 l) were preincubated with 500 ng of bacterially expressed recombinant GST-CHOP or GST, prior to incubation with the labeled ⌬C/EBP or ⌬IRS.1 probe, then loaded for native gel electrophoresis and autoradiography.

FIG. 6. Gel shift studies with recombinant C/EBP␤. A, gel shift
with poly(dG-dC)⅐poly(dG-dC). Initial gel shift studies with recombinant C/EBP␤ were performed under conditions which optimize interactions between C/EBP proteins and probes containing the PEPCK IRS and flanking sequences (10,29). Recombinant full-length C/EBP␤ (20 ng) was preincubated with/without a 250-fold excess of an unlabeled oligonucleotide competitor containing a consensus C/EBP-binding site, and then with probes containing the C/EBP site (C/EBP), the IGFBP-1 insulin response element (BP1), or the PEPCK IRS in the presence of poly(dG-dC)⅐poly(dG-dC) at 4°C, then loaded for native gel electrophoresis at 22°C. B, standard gel shift conditions. The labeled ⌬IRS.1 or ⌬C/EBP probe was incubated with recombinant full-length (C/ EBP␤ 1-276 ; left panel) or truncated (C/EBP␤ 132-276 ; right panel) C/EBP␤ in the presence of poly(dI-dC)⅐poly(dI-dC) under our standard conditions. Bound and free probe were resolved by native gel electrophoresis at 4°C, as in Figs. 1-5. contribute to the ability of insulin to inhibit promoter activity via an IRS, we examined the effect of altering individual residues within an IRS (CAAAACA) on the ability of unlabeled oligonucleotides to interact with factors required for the formation of this complex in gel shift assays, and the ability of insulin to suppress promoter activity in reporter gene studies. Previous studies have shown that altering residue 6 within the IRS (CAAAAgA) disrupts the ability of insulin to suppress promoter activity via an IRS in both the IGFBP-1 and PEPCK genes (6). Accordingly, we determined whether mutation of this residue also alters interactions with factors required for the formation of this complex. As shown in Fig. 8A, a 5-fold molar excess of the unlabeled ⌬IRS.1 oligo is sufficient to inhibit the ability of partially purified proteins to form this complex with the ⌬IRS.1 probe by 66%, based on PhosphorImaging. Replacing residue 6 of the IRS (⌬IRS.1m6) impairs the ability of a 5-fold excess of unlabeled oligo to interact with factors required for the formation of this complex and inhibit its formation in gel shift assays (Fig. 8A). At the same time, higher titers of the ⌬IRS.1m6 competitor were partially effective at inhibiting the formation of this complex (Fig. 8A), indicating that differences in binding activity in oligos containing point mutations within the IRS are most clear at low titers. Based on this result, we used a 5-fold molar excess of unlabeled competitors to examine the effects of altering other residues within the IRS on the ability to interact with factors required for the formation of this complex.
As shown in Fig. 8B, altering residues 3, 4, 5, or 6 within the IRS (⌬IRS.1m3-⌬IRS.1m6) also reduces the ability of a 5-fold molar excess of unlabeled oligos to interact with factors which are required for the formation of this complex. In contrast, altering residues 1, 2, or 7 or the IRS does not impair the ability of unlabeled oligos to inhibit the formation of this complex. As shown in Table I, altering residues 3, 4, 5, or 6 also reduces the ability of insulin to inhibit promoter activity via an IRS in reporter gene studies. As shown in Fig. 9, the ability of oligonucleotides containing point mutations within the IRS to interact with factors required for the formation of this complex correlates well with the ability of reporter gene constructs containing the same mutations to mediate inhibitory effects of insulin on promoter activity (r ϭ 0.849, p Ͻ 0.01). Together, these results indicate that this complex interacts with the IRS in a sequence-specific fashion consistent with the concept that this complex may contribute to the ability of insulin to inhibit promoter activity via an IRS.
Since C/EBP␤ is present in this complex, we next asked whether C/EBP proteins themselves might contribute to the ability of insulin to inhibit promoter activity. As shown in Table  II, the ⌬IRS.1 reporter gene construct contains a single IRS located ϳ100 bp upstream from the transcription initiation site, and insulin treatment reduces promoter activity in this construct by 55%. Substituting a single residue (⌬IRS.1m6) or several residues (⌬IRS.1M) within the IRS disrupts the ability of insulin to suppress promoter activity, indicating that this effect of insulin is mediated via the IRS, as previously reported (6). Placing a consensus C/EBP-binding site at the same location (⌬C/EBP) partially maintains the ability of insulin to inhibit promoter activity, similar to an IRS. Placing a consensus HNF-3 site (⌬HNF-3) or a CRE (⌬CRE) at this location does not maintain the ability of insulin to suppress promoter activity, demonstrating that this effect is specific. Control studies with FKHR expression and antisense vectors (12) confirm that FKHR forkhead proteins do not interact with this site and that phosphorylation of FKHR is not required for insulin to regulate promoter activity via this C/EBP-binding site (data not shown), indicating that the ability of insulin to regulate promoter activity via this C/EBP-binding site does not require interaction with FKHR-related forkhead proteins.
Placing an IRS (but not a mutated sequence) 3 bp further downstream (⌬IRS.2 versus ⌬IRS.2M) or ϳ230 bp further upstream (⌬IRS.331 versus ⌬IRS.331.m6) to disrupt potential interactions with flanking sequences also maintains the ability FIG. 7. Supershift and competitive binding studies with partially purified proteins. A, supershift studies. Partially purified proteins were preincubated with specific antibodies against C/EBP proteins, ATF-1, -2, or -4, c-Jun, c-Fos, Rb p110, YY1, or NF␤ p50 or p65 prior to incubation with the ⌬IRS.1 probe. B, competitive binding studies with the PEPCK CRE. Partially purified proteins were preincubated with/without a 5-100-fold molar excess of unlabeled competitors prior to incubation with the 32 P-labeled ⌬IRS.1 probe. Competitors include ⌬IRS.1, an oligo where the IRS is replaced by the CRE from the PEPCK promoter (⌬CRE), and an oligo containing the CRE and its flanking sequences from the PEPCK promoter, as shown.  Table I. of insulin to suppress promoter function. In each case, placing a C/EBP-binding site at the same location also confers an effect of insulin on promoter activity (⌬C/EBP.2 and ⌬C/EBP.331), although not as effectively as an IRS. Together, these results indicate that C/EBP proteins may contribute to the ability of insulin to suppress promoter activity. DISCUSSION To understand specific mechanisms that may mediate effects of insulin on gene expression via an IRS, we examined nucleoprotein complexes that interact with oligonucleotide probes containing IRSs from either the IGFBP-1 or PEPCK gene. Previous studies have shown that recombinant C/EBP proteins can bind directly to oligonucleotide probes containing the PEPCK IRS, but not the IGFBP-1 insulin response element (10), and we confirmed this result. Direct binding of C/EBP proteins to the PEPCK IRS is thought to require interaction with flanking sequences and mutation of these sequences does not disrupt the effect of insulin (10). These observations suggested that C/EBP proteins are not likely to be involved in mediating the effect of insulin on promoter activity via an IRS (10). In contrast, we find that C/EBP␤ is present in a complex of proteins that interacts with residues within an IRS that are critical for the effect of insulin, and that sequences flanking the IRS are not essential for the formation of this complex. In addition, replacing the IRS with a consensus C/EBP-binding site maintains an effect of insulin on promoter activity. To our knowledge, these results provide the first report indicating that a nucleoprotein complex containing C/EBP␤ can interact with an IRS in a sequence-specific fashion and may contribute to the ability of insulin to suppress promoter activity. C/EBP proteins play an important role in the regulation of gene expression in the liver (18) and other insulin-responsive tissues (24,40). Studies in knockout mouse models show that both C/EBP␣ and C/EBP␤ contribute to the regulation of hepatic glucose production (20,21,41). C/EBP␣ contributes to cAMP-responsive expression of PEPCK (42), a rate-limiting enzyme which plays a critical role in the regulation of gluconeogenesis, and C/EBP␤ contributes to the regulation of the PEPCK promoter by thyroid hormone, glucocorticoids, and cAMP agonists (43,44). A recent report by Yeagley et al. (45) indicates that C/EBP proteins also may play an important role in mediating effects of insulin on cAMP stimulated activity of TABLE I Interaction with IRS in binding and reporter gene studies Gel shift studies were performed using partially purified nuclear proteins and the ⌬IRS.1 oligonucleotide probe containing a single IRS (CAAAACA). The ability of a 5-fold molar excess of unlabeled oligonucleotides where a single residue within the IRS was altered was evaluated (Fig. 7B) and the amount of radioactive probe bound in the presence (B) and absence (B 0 ) of competitor was measured by phosphorimaging. To assess the effect of altering the sequence of the IRS on the ability of insulin to suppress promoter activity, we performed transient transfection assays in HepG2 cells using a luciferase reporter gene construct which contains 320 bp of the IGFBP-1 promoter. The insulin response element was removed and replaced by a single IRS (CAAAACA) located Ϫ104 to Ϫ110 5Ј to the transcription initiation site (⌬IRS.1), or a modified IRS where a single residue has been altered (⌬IRS.1m1 and Ϫ ⌬IRS.1m6), as previously reported (6). Cells were treated with serum-free medium with/without 100 nM insulin for 18 h prior to lysis and analysis of luciferase activity, as before (6,11). (Note: the effect of insulin on promoter activity in HepG2 cells transfected with these reporter gene constructs was reported previously (11) and is presented in this table and in Fig. 9 Table I), and the ability of reporter gene constructs containing point mutations within an IRS to mediate effects insulin on promoter activity was assessed in transient transfection assays in HepG2 cells (Table I). The relationship between the effects of point mutations on binding activity and insulin-regulated promoter activity is shown. The significance of this relationship was examined by Pearson correlation analysis. the PEPCK promoter independent of the PEPCK IRS. Together, these findings indicate that C/EBP proteins play an important role in the multihormonal regulation of hepatic gene expression.
Competitive binding and functional studies with constructs containing point mutations within the IRS revealed that 4 residues (AAAC) are essential for interactions with factors required for the formation of this complex and for mediating effects of insulin on promoter activity. Recent studies by Hall et al. (17) indicate that several of these residues (AAC) also are critical for mediating effects via the IRS in the PEPCK promoter in cAMP/dexamethasone-stimulated H4IIE hepatoma cells, supporting the concept that similar factors may interact with the PEPCK IRS and contribute to the ability of insulin to regulate that gene. Of note, the consensus HNF-3-binding site (AAACAAACATT) we introduced into a reporter gene construct (⌬HNF-3) also contains this core sequence, but does not confer the ability of insulin to suppress promoter activity. We have reported that the ability of IRSs in the IGFBP-1 and PEPCK genes to bind HNF-3 proteins is relatively weak compared with a known high affinity HNF-3-binding site (9). Presumably, high affinity binding of HNF-3 proteins to this consensus sequence may prevent interaction with other factors that are required to mediate the effect of insulin. Based on the present study, we speculate that interactions with HNF-3 proteins must be relatively weak to permit IRSs to interact with other factors that are critical for the ability for insulin to regulate gene expression in liver-derived cells, and that variations in sequences flanking the AAAC core are important in determining the relative affinity with which IRSs interact with insulinresponsive and/or blocking factors.
In this context, it is important to note that this core sequence also is conserved in a recently identified consensus binding sequence for FKHR-related forkhead proteins (including FKHR, FKHRL1, AFX, and DAF-16) (46), which are thought to contribute to effects of insulin on gene expression (12)(13)(14)(15)(16). Taken together with the results of the present study, these observations suggest the interesting possibility that multiple factors, including forkhead proteins and the complex containing C/EBP␤, may interact with this core AAAC sequence and contribute to the ability of insulin to regulate promoter activity via an IRS. This concept also is suggested by previous studies indicating that IRSs from the apolipoprotein CIII and tyrosine aminotransferase genes also can interact with factors required for the formation of this C/EBP␤-containing complex (11), but that another IRS from the IGFBP-1 promoter (IRSB) does not (11).
Previous studies have shown that C/EBP␤ may interact with and form nucleoprotein complexes with a number of factors, including ATF-4 (CREB-2), c-Jun, and c-Fos, Rb p110, and NF␤ proteins and the glucocorticoid receptor. However, antibodies against these factors fail to disrupt or supershift the C/EBP␤-containing complex which interacts with the IRS. FKHR-related forkhead transcription factors are expressed in the liver and can interact directly with IRSs in the IGFBP-1 and PEPCK genes (12,14,15,47,48) and it is interesting to speculate that C/EBP␤ may interact with these forkhead proteins to form a complex with the IRS. Unlabeled competitors containing the high affinity HNF-3-binding site from the transthyretin gene have been reported to inhibit the binding of recombinant FKHR to the IGFBP-1 insulin response element (47). However, a competitor containing this HNF-3-binding site failed to inhibit the formation of this C/EBP␤ containing complex ( Figs. 1 and 2). Also, antibodies against HNF-3/forkhead proteins, FKHR and FKHRL1 (which also are expressed in liver cells), also failed to detect forkhead proteins in this com-plex, suggesting that forkhead proteins are not required for the formation of this complex. However, this negative results does not entirely exclude this possibility, since epitopes required for recognition by these antibodies may be masked by the formation of this complex. At the same time, the observation that a consensus C/EBP-binding site mediates effects of insulin on promoter activity similar to an IRS indicates that C/EBP proteins may contribute directly to the ability of insulin to suppress promoter activity via a mechanism that is independent of forkhead proteins, and studies reported in an accompanying paper (48) support this concept.
At the same time, it is important to note that a C/EBP site is only partially as effective in mediating an effect of insulin on promoter activity compared with a recognized IRS (Table II). In related studies, we found that insulin suppresses transactivation by C/EBP␤, but not C/EBP␣ when these proteins are expressed in HepG2 cells (48). Since both C/EBP␣ and C/EBP␤ are expressed in HepG2 cells (49,50), C/EBP␣ may limit the ability of insulin to suppress promoter activity through a consensus C/EBP-binding site (which can bind hetero-and homodimers of C/EBP proteins containing C/EBP␣ and/or C/EBP␤) compared with an IRS (which interacts with a complex containing C/EBP␤, but not C/EBP␣). Alternatively, other factors, including forkhead proteins, may be required to mediate the full effect of insulin on promoter activity via an IRS in HepG2 cells.
In this context, it is interesting to note that the ability of probes containing an IRS to form this complex is enhanced in nuclear extracts prepared from rat liver compared with extracts prepared from H4IIE and HepG2 hepatoma cells. Since levels of C/EBP proteins are reduced in hepatoma cells lines compared with normal liver (49), 2 nucleoprotein complexes containing C/EBP␤ may play a more prominent role in mediating effects of insulin on hepatic gene expression in vivo than might be apparent in studies performed in these transformed cells.
In summary, the results of the present study indicate that a nucleoprotein complex containing C/EBP␤ can interact with IRSs in the IGFBP-1 and PEPCK genes. Based on competitive binding data and studies with recombinant proteins, it appears that C/EBP␤ does not interact directly with the IRS in the IGFBP-1 promoter, but that some other factor(s) is required. The ability of the IRS to interact with factors required for the formation of this complex correlates well with the ability of an IRS to mediate inhibitory effects of insulin on promoter activity, and replacing the IRS with a consensus C/EBP-binding site maintains the effect of insulin. Additional studies are required to identify the other factor(s) required for the formation of this complex, and to examine the relative role that this nucleoprotein complex and C/EBP proteins play in mediating effects of insulin on hepatic gene expression.