Nuclear Factor I-mediated Repression of the Mouse Mammary Tumor Virus Promoter Is Abrogated by the Coactivators p300/CBP and SRC-1*

To better understand the function of nuclear factor I (NFI) proteins in transcription, we have used transient transfection assays to assess transcriptional modulation by NFI proteins on the NFI-dependent mouse mammary tumor virus (MMTV) promoter. Expression of NFI-C or NFI-X, but not NFI-A or NFI-B proteins, represses glucocorticoid induction of the MMTV promoter in HeLa cells. Repression is DNA binding-independent as a deletion construct expressing the NH2-terminal 160 residues of NFI-C represses but does not bind DNA. Repression by NFI-C is cell type-dependent and occurs in HeLa and COS-1 cells but not 293 or JEG-3 cells. NFI-C does not repress progesterone induction of the MMTV promoter in HeLa cells, suggesting that progesterone induction of the promoter differs mechanistically from glucocorticoid induction. NFI-C-mediated repression is alleviated by overexpression of glucocorticoid receptor (GR), suggesting that NFI-C represses the MMTV promoter by preventing GR function. However, repression by NFI-C occurs with only a subset of glucocorticoid-responsive promoters, as the chimeric NFIGREβ-gal promoter that is activated by GR is not repressed by NFI-C. Since the coactivator proteins p300/CBP, SRC-1A, and RAC3 had previously been shown to function at steroid hormone-responsive promoters, we asked whether they could influence NFI-C-mediated repression of MMTV expression. Expression of p300/CBP or SRC-1A alleviates repression by NFI-C, whereas RAC3 has no effect. This abrogation of NFI-C-mediated repression by p300/CBP and SRC-1A suggests that repression by NFI-C may occur by interference with coactivator function at the MMTV promoter.

encoded protein required for the efficient initiation of adenovirus (Ad) replication in vitro (1) and was later shown to be required for the expression of many cellular and viral genes. NFI-binding sites occur both in genes expressed in multiple tissues (2) and in genes expressed solely in brain (3), muscle (4), liver (5), mammary gland (6), and other differentiated cell types (7). NFI-binding sites are also found in the promoter regions of several viruses including the neurotropic JC virus (8), human papilloma virus type 16 (9), and the mouse mammary tumor virus (10). Mutational analysis indicates that these NFI sites are required for the proper expression of many tissue-specific and developmentally regulated genes (11,12).
Cloning of cDNAs encoding NFI proteins from several species (13)(14)(15) has identified a family of four genes (NFI-A, NFI-B, NFI-C, and NFI-X) that are highly conserved from chickens to humans. NFI proteins contain a highly conserved but gene-specific NH 2 -terminal 200 amino acid domain that mediates DNA binding, dimerization, and the initiation of Ad replication (16,17). NFI proteins bind to DNA as both homoand heterodimers and recognize the consensus binding site, TTGGC(N 5 )GCCAA with the same apparent affinity (18,19). However, considerable variation occurs within the COOH-terminal domains of the NFI proteins that likely encode distinct transcription modulation domains. Additional variation between NFI proteins is generated through differential splicing of transcripts from each of the four genes (20).
The existence of four conserved NFI genes in vertebrates that generate multiple alternatively spliced polypeptides argues in favor of diverse functions of individual NFI gene products. We have previously shown that NFI proteins representing each of the four murine NFI genes exhibit promoter-specific differences in their maximal transcriptional activation potentials due to differences in their COOH-terminal regions (21,22). The molecular basis for such differences in the transcriptional activation properties of NFI proteins is unknown. However, direct interactions between a human NFI-C isoform (CTF1) and components of the basal transcriptional machinery have been reported, with interactions dependent on a sequence motif related to the COOH-terminal heptapeptide repeat (CTD) of RNA polymerase II (23). This is unlikely to be the only mechanism by which NFI proteins activate transcription as some NFI proteins lacking a CTD repeat are potent activators in both yeast (24) and mammalian cells (21).
While it is widely accepted that NFI proteins differ in their activation potentials, their ability to repress transcription is poorly understood. Here, we show that NFI proteins exhibit cell type-and promoter-specific differences in their repression properties, with NFI-C and -X repressing the MMTV promoter in HeLa cells, while NFI-A and -B do not. In contrast to earlier studies, we show that this repression domain of NFI-C is contained within the NH 2 -terminal DNA binding domain, but repression does not require DNA binding activity. We also demonstrate that overexpression of p300/CBP or SRC-1A abrogates repression of MMTV expression, suggesting that NFI-C-mediated repression may occur by interference with coactivator function at the MMTV promoter.
Cell Culture Transfection and Assays-The cell lines HeLa, COS-1, 293, and JEG-3 cells (American Type Culture Collection) were cultured in ␣-minimum Eagle's medium (Mediatech) containing 10% fetal bovine serum. Twenty-four hours prior to transfection, 2 ϫ 10 5 cells were plated onto 60-mm dishes and then transfected using calcium phosphate coprecipitation as described (31) Typically, each coprecipitation consisted of an NFI-dependent ␤-gal reporter construct (5 g), SV-40luciferase (pGL2-Control, Promega), internal control (2.5 g), and 2.5 g of various NFI effector constructs. When indicated, 0.5 g of phGR or phPR were cotransfected with pMMTV␤-gal and pNFIGRE␤-gal. Carrier DNA (pBSIIKSϩ) was added to a total of 15 g of transfected DNA per plate. Cells were incubated with CaPO 4 /DNA precipitate for 12-16 h, washed with PBS, and incubated for 24 h in culture media with or without 0.1 M dexamethasone or 10 nM progesterone (Sigma). Cells were harvested in 300 l/plate of Reporter Lysis buffer (Promega) with luciferase and ␤-gal assays performed as described (21). Transfections were performed in quadruplicate using duplicate precipitates for each point, and all results were confirmed by multiple independent experiments using at least two different CsCl-purified preparations of plasmid DNA.
Gel Mobility Shift Assays and HA Antibody Supershifts-For whole cell extracts cell monolayers were washed twice with ice-cold PBS and scraped into a 1.5-ml Eppendorf tube. Cells were pelleted by spinning at 1000 ϫ g for 5 min at 4°C; the supernatant was removed; and the pellet was resuspended in Lysis buffer (100 mM Tris, pH 7.4, 350 mM NaCl, 10% glycerol, 1% Nonidet P-40, 1 mM EDTA, 1 mM dithiothreitol, 1 mM EDTA, 10 g/ml leupeptin, 10 g/ml pepstatin, 10 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride) and incubated on ice for 15 min with occasional swirling. Cell debris was pelleted by spinning at 16,000 ϫ g for 10 min, with the supernatant removed and stored at Ϫ80°C. Gel mobility shift analysis were performed using the 32 Plabeled FIB-2.6 oligonucleotide as described previously (18). For antibody supershift analysis with the C125A anti-HA antibody (Boehringer Mannheim), binding reactions were carried out for 30 min on ice, followed by the addition of 1 l of the anti-HA antibody (200 g/ml) and further incubation for 20 min at room temperature. As a control, binding reactions not incubated with antibody were treated with 1 l of PBS and incubated for 20 min at room temperature. The DNA-protein complexes were resolved on a 6.5% polyacrylamide, 0.25% TBE gel, and analyzed using a Molecular Dynamics model 400 PhosphorImager.
Immunocytochemistry-To detect the intracellular location of the HA-tagged NFI proteins, HeLa cells were grown on coverslips, transfected with vectors expressing the NFI protein or control vectors, cultured for 48 h, and fixed in chilled methanol at Ϫ20°C for 15 min. Fixed cells were blocked for 1 h at room temperature with 3% bovine serum albumin, 0.1% Tween 20 in PBS (PBST), incubated with 2-3 g/ml anti-HA antibody in PBST for 45 min, washed in PBS, incubated for 45 min with fluorescein isothiocyanate-conjugated secondary antibody, and washed in PBS; the coverslips were mounted with Vectashield (Vector Laboratories); the cells were examined using a Nikon fluorescence microscope; and the images were captured with an Oncor imaging system (Cleveland Clinic Fluorescent Microscopy Core). For control staining, the specific antibody solution was replaced by PBST alone.
Western Blot Analysis-Cells transfected with the NFI effector plasmids and hormone receptor expression plasmids were pelleted, lysed directly in 1ϫ Laemmli buffer (32), boiled for 10 min, separated on a SDS-polyacrylamide gel, and electroblotted onto Immobilon-P membranes (Millipore). The blots were blocked and hybridized with either anti-HA antibody (Boehringer Mannheim) or anti-GR PA-510 antibody (Affinity Bioreagents) as per manufacturer's instructions. Proteins were detected by chemiluminescence (ECL, Amersham Pharmacia Biotech).

Overexpression of NFI-C and NFI-X Proteins Represses Glucocorticoid Induction of the MMTV-Long Terminal Repeat Promoter in HeLa
Cells-We previously reported the cloning of cDNAs representing each of the four murine NFI genes, analyzed their embryonic and postnatal expression patterns, and demonstrated promoter-specific differences in their relative activation potentials (21,22). Our previous studies showed that NFI proteins representing each of the four NFI genes differentially activated the MMTV promoter in human JEG-3 choriocarcinoma cells, with NFI-B activating expression ϳ12-fold, NFI-X activating ϳ10-fold, NFI-C ϳ6-fold, and NFI-A only ϳ2-fold. To determine whether the maximum activation potentials of these NFI isoforms differed in different cell types, we first examined their ability to modulate the MMTV promoter in human HeLa cells. HeLa cells had previously been used to demonstrate the requirement for an NFI-binding site for glucocorticoid induction of the MMTV promoter (33). Surprisingly, coexpression of NFI-C or NFI-X completely repressed glucocorticoid induction of the MMTV promoter in HeLa cells ( . Repression of MMTV expression by NFI-C and NFI-X appeared to be promoter-specific, since expression from the chimeric NFI␤-gal promoter, which contains a single NFI-binding site upstream of the Ad major late promoter, was not repressed by coexpression of any of the NFI proteins (Fig. 1B, lanes 10-13 versus lane 9). Previous studies had shown that this promoter requires NFI binding for maximal expression (22,25). Differences in repression by the four NFI gene products were not due to differences in expression since all four proteins were expressed at similar levels, as assessed by Western blot analysis of the HA-tagged proteins (Fig. 1C).
NFI-C Repression Domain Is Contained within the NH 2terminal 160 Amino Acids and Is DNA Binding-independent-To determine which domain(s) of the NFI-C and -X pro-teins mediated repressor function, we made COOH-terminal deletion constructs ( Fig. 2A). As NFI-C and NFI-X both repressed the MMTV promoter in HeLa cells and NFI-A and NFI-B did not, deletion analysis was performed on the NFI-B and NFI-C proteins as examples of the two classes. The COOHterminal deletion constructs express NFI-C proteins of 75 aa (NFI-C-75), 135 aa (NFI-C-130), 160 aa (NFI-C-160), and 240 aa (NFI-C-240). A plasmid containing the DNA binding domain NH 2 -terminal 235 residues of the NFI-B isoform (NFI-B-235) was used as a negative control. As expected, cotransfection of either the full-length NFI-B or NFI-B-235 expression plasmids failed to repress dexamethasone induction of the MMTV promoter in HeLa cells ( Previous studies showed that ϳ200 NH 2 -terminal residues of NFI proteins are required for DNA binding and dimerization (16,19,34). Since HeLa cells contain endogenous NFI proteins (18), anti-HA antibody supershifts were used to assess the DNA binding ability of the various transfected HA-tagged NFI proteins. As seen previously (18) (22)). However, no anti-HA-supershifted complexes were detected in extracts of cells transfected with the NFI-C-160 vector (Fig. 2C, lane 6 versus lane 5), indicating that the NFI-C-160 protein lacks DNA binding activity. The lack of DNA binding activity is not due to poor expression of NFI-C-160, since Western blots show that similar levels of NFI-B and NFI-C-160 proteins are expressed (Fig. 2D). Lack of DNA binding by NFI-C-160 was confirmed by transfection of the NFI-C-160 vector into JEG-3 cells which lack endogenous NFI proteins (data not shown). These data, together with previous SV-40 luciferase vector (pGL2), and 2.5 g of CMV control vector (lanes 1 and 2) or CMV vectors expressing NFI-A (lanes 3 and 4), NFI-B (lanes 5 and 6), NFI-C (lanes 7 and 8), or NFI-X (lanes 9 and 10). Transfected cells were cultured for 24 h in media in the absence or presence of 0.1 M dexamethasone (Ϫ or ϩDex). ␤-Galactosidase activity was normalized to luciferase activity levels, with the bars representing the mean and range of four measurements from duplicate transfections. B, HeLa cells were transfected with pMMTV␤-gal (5 g, lanes 1-8) or pNFI␤-gal (10 g, lanes 9 -13), phGR (0.5 g, lanes [5][6][7][8], and with 2. studies showing that ϳ200 NH 2 -terminal residues of NFI proteins are needed for DNA binding and dimerization (16,19,34), indicate that repression of the MMTV promoter by the NFI-C-160 protein occurs in the absence of NFI DNA binding or dimerization activity.
To assess whether the differences in repression by the NFI-C COOH-terminal deletion proteins were due to differences in cellular localization, we examined the intracellular location of the NFI-C proteins by immuno-histochemistry. Mock-transfected cells stained with the anti-HA antibodies or cells stained with secondary antibody alone showed no fluorescence (Fig. 2E, panel D, and data not shown). Although both NFI-C-439 and NFIC-160 exhibited strong nuclear fluorescence (Fig. 2, panel  A versus panel B), cells expressing NFI-C1-160 exhibited some diffuse cytoplasmic staining not seen with NFI-C-439 (compare Fig. 2, panel A versus panel B). In contrast, cells expressing NFI-C-130 stained weakly in the nucleus but showed strong cytoplasmic staining with a punctate distribution (compare Fig. 2, panel C versus panels A and B), suggesting the presence of a nuclear localization or retention signal in residues 130 -160 of NFI-C.
To determine if the lack of repression observed with the NFI-C-130 construct was due to differences in nuclear localization, we made a chimeric construct containing residues 1-130 of NFI-C fused to residues 131-235 of NFI-B. Expression and nuclear localization of the chimeric protein was verified by DNA binding studies (data not shown). When cotransfected with the MMTV promoter in HeLa cells, the chimeric NFI construct failed to repress glucocorticoid induction (data not shown) suggesting that the differences in the repression properties of the NFI-C-160 and NFI-C-130 proteins are not due solely to differences in nuclear localization.
Overexpression of Glucocorticoid Receptor Alleviates NFI-Cmediated Repression in HeLa Cells-The ability of NFI-C to repress glucocorticoid induction of the MMTV promoter but not the chimeric NFI␤-gal reporter indicated that NFI-C might repress by inhibition of glucocorticoid receptor function. Since NFI-C-mediated repression is DNA binding-independent, it is unlikely that repression occurs by NFI and GR competing for binding to the MMTV promoter. Thus, repression likely involves some form of protein-protein interactions, either through direct interactions between NFI-C and GR or by NFI-C and GR competing for a limiting coactivator protein required for MMTV transcription (see "Discussion"). Either model would predict that repression by NFI-C might be overcome by overexpressing GR. To test this, we cotransfected increasing amounts of the hGR vector with pMMTV␤-gal, with or without the NFI-C expression plasmid (Fig. 3A). As seen above, HeLa cells cotransfected with 0.5 g of phGR, and pMMTV␤-gal exhibited strong dexamethasone-dependent MMTV promoter activity (Fig. 3A, lane 2 versus lane 1). In the absence of coexpressed NFI-C, increasing amounts of cotransfected phGR increased dexamethasone-dependent MMTV activity ϳ2-fold (lane 4 versus lane 2) and ϳ2.5-fold (lane 6 versus lane 2). MMTV promoter activity in cells transfected with 5.0 g of the hGR expression plasmid appeared to represent maximal promoter activation as further increasing the amount of phGR DNA transfected did not increase expression (data not shown). As seen above, repression of MMTV expression was observed when NFI-C was coexpressed together with either 0.5 or 1.5 g of phGR DNA (lanes 8 and 10 versus lanes 2 and 4, respectively). However, repression by NFI-C was partially abrogated when 5.0 g of phGR was cotransfected (lane 12 versus lanes 10 and 8). This ability of NFI-C to repress glucocorticoid induction of the MMTV promoter when sub-saturating amounts of the GR expression vector are used appears not to be due to NFI inhibiting GR expression, since Western blot analysis shows that glucocorticoid receptor levels are unaffected by expression of NFI-C (Fig. 3B, lane 3 5 g, lanes 1-6), or a CMV vector expressing NFI-C (2.5 g, lanes 7-12) with the indicated amounts of phGR. Cells were cultured in the absence or presence of 0.1 M dexamethasone (Ϫ or ϩDex) for 24 h. ␤-Galactosidase values were normalized to luciferase values and plotted as in Fig. 1. B, HeLa cells were mock-transfected (lane 1) or transfected with the indicated amounts of phGR and CMV control vector (2.5 g, lanes 2, 4, and 6), or the NFI-C expression vector (2.5 g, lanes 3, 5, and 7). Whole cell extracts were analyzed on a 5% SDS-polyacrylamide gel transferred to Immobilon-P membrane and probed with an anti-GR antibody (PA510, Affinity Bioreagents). Numbers indicate size markers (M r ϫ 10 Ϫ3 ); the arrow indicates glucocorticoid receptor, and the asterisk indicates a proteolytic fragment of GR.
promoter, we transfected HeLa cells with pMMTV␤-gal and vectors expressing either progesterone or glucocorticoid receptor (Fig. 4). In the absence of cotransfected hPR, no progesteronedependent MMTV expression was seen (Fig. 4, lane 2  Expression of NFI-C Does Not Repress pNFIGRE␤-gal in HeLa Cells-Since expression of NFI-C repressed glucocorticoid induction of the MMTV promoter but had no effect on progesterone induction, we tested whether repression occurred with other glucocorticoid-dependent promoters. pNFIGRE␤-gal was made by the addition of two glucocorticoid response elements (GREs) immediately upstream of the NFI-binding site in pNFI␤-gal (22). We compared the activity of this promoter with both the MMTV promoter and the NFI promoter lacking GREs. As seen previously, MMTV expression was repressed by NFI-C at low GR levels but not at high GR levels (Fig. 5, lane 8 14). This lack of NFI-C-mediated repression, even in the absence of any coexpressed GR, indicates that NFI-C expression likely only affects a subset of GR-responsive promoters. The low level of GR-induced activity of NFIGRE promoter in the absence of coexpressed GR (Fig. 5, lane 14) differs from the absolute requirement for GR coexpression for MMTV activity (Fig. 1B, lane 6 versus lane 2), suggesting that the NFIGRE promoter is more sensitive than the MMTV promoter to low levels of endogenous GR present in HeLa cells.
NFI-C-mediated Repression Is Cell Type-specific-We recently reported that NFI-C activates the MMTV promoter in JEG-3 cells (22). Here, we have shown that NFI-C represses glucocorticoid-dependent MMTV expression in HeLa cells. To determine if NFI-C-mediated repression of the MMTV promoter was unique to HeLa cells, we performed transient transfections in both COS-1 and 293 cells (Fig. 6A). As with HeLa cells, cotransfection of phGR was required to detect dexamethasone-dependent MMTV activation in all cell types (data not shown). As reported previously, in JEG-3 cells (22) 10), demonstrating that repression of the MMTV promoter by NFI-C is cell type-specific. Differences in the transcription modulation properties of the NFI-C protein in these cell lines were not due to differences in NFI-C levels, as Western blot analysis showed similar levels of HA-tagged NFI-C proteins expressed in each cell line (Fig. 6B,  arrow). To ensure that the lack of repression in 293 cells was not due to higher hGR levels in these cells, we titrated the amount of hGR vector transfected and saw no repression by NFI-C, even at clearly submaximal hGR levels (Fig. 6C).
Coactivators p300/CBP and SRC-1A, but Not RAC3, Overcome NFI-C-mediated Repression of the MMTV Promoter in HeLa Cells-The inability of NFI-C to repress glucocorticoid induction of pNFIGRE␤-gal (Fig. 5), together with the cell type-specific repression by NFI-C (Fig. 6), argues against repression by direct interaction of NFI-C with glucocorticoid receptor. An alternate model would have NFI-C and GR competing for a limiting cofactor (in HeLa and COS-1 cells) which is required for dexamethasone induction of the MMTV promoter. To test this model we overexpressed various steroid receptor coactivator proteins with NFI-C and pMMTV␤-gal to ask if they could overcome NFI-C-mediated repression (Fig. 7). Cotransfection of the RAC3 expression plasmid had no effect on MMTV expression in either the absence or presence of NFI-C (Fig. 7, lane 4 versus lane 2 and lane 12 versus lane 10, respectively). However, expression of either p300 or SRC-1A increased MMTV promoter activity slightly (Ͻ2-fold) in the absence of NFI-C (Fig. 7, lanes 6 and 8 versus lane 2) and abrogated the repression of MMTV activity by NFI-C (Fig. 7,  lanes 14 and 16 versus lane 10). Alleviation of NFI-C-mediated repression was also seen with coexpression of CBP (not shown). This alleviation of NFI-C-mediated repression by both p300/ CBP and SRC-1A suggests several models for how these proteins may function with NFI-C at the MMTV promoter (see "Discussion").  [17][18][19][20], and CMV control vector (2.5 g, lanes 1, 2, 5, 6, 9, 10, 13, 14, 17, and 18) or the NFI-C expression vector (lanes 3, 4, 7, 8, 11, 12, 15, 16, 19, and 20). Cells were cultured for 24 h in the absence or presence of 10 nM progesterone (Ϫ or ϩ Pro) or 0.1 M dexamethasone (Ϫ or ϩDex). ␤-Galactosidase activity was normalized to luciferase levels and plotted as in Fig. 1.

DISCUSSION
We have used the NFI-dependent MMTV promoter to assess the transcription modulation properties of NFI proteins from each of the four murine NFI genes (Nfi-a, -b, -c, and -x). In transiently transfected HeLa cells, expression of NFI-C or NFI-X represses glucocorticoid induction of the MMTV promoter, whereas expression of NFI-A or NFI-B does not (Fig.  1A). In contrast to all previously described functions of NFI proteins, this repression is DNA binding-independent, as expression of the non-DNA binding NH 2 -terminal 160 residues of NFI-C represses (Fig. 2B). Repression by NFI-C is overcome by overexpression of the glucocorticoid receptor (Fig. 3A), suggesting that repression may occur through interference with GR function. However, the chimeric pNFIGRE␤-gal reporter, which is also dependent on both NFI and GR for expression, is not repressed (Fig. 5), showing that repression occurs in only a subset of GR-dependent promoters. Repression of MMTV is cell type-specific, occurring in HeLa and COS-1 cells but not JEG-3 or 293 cells (Fig. 6A), suggesting cell type specificity in either NFI function or in the mechanism of GR activation of the MMTV promoter. Surprisingly, progesterone-dependent activation of the MMTV promoter is not repressed by NFI-C (Fig.  4), indicating that the mechanism of activation by progestins may differ from that of glucocorticoids. In addition, NFI-Cmediated repression is abrogated by coexpression of the receptor coactivators CBP/p300 and SRC-1A (Fig. 7), suggesting that repression may occur by interference with coactivator function at the MMTV promoter.
Previous studies showed that NFI proteins possess highly homologous NH 2 -terminal DNA binding/dimerization domains and more divergent COOH-terminal transactivation or repres-sion domains (14,16,36,37). For example, in JEG-3 choriocarcinoma cells, NFI proteins representing each of the four NFI genes differentially activate the MMTV promoter, and these differences in transactivation are mediated solely by their COOH-terminal domains (22). Here we report that in transfected HeLa cells, NFI proteins differentially repress the MMTV promoter, with NFI-C or NFI-X repressing glucocorticoid induction while NFI-A or NFI-B have no effect. Although transcriptional repression by NFI proteins has previously been reported (38,39), repression was dependent on residues within the COOH-terminal domains of NFI proteins and required DNA binding by NFI or NFI fusion proteins. In contrast, the repression of the MMTV promoter by NFI-C (and NFI-X) seen here is mediated by a 160-residue subdomain of the NH 2terminal domain that is incapable of binding DNA. The DNA binding independence of the repression suggests that the mechanism of this repression likely differs from that seen previously by COOH-terminal regions of NFI proteins. Within this 160residue repression domain there are 10 residues that differ between NFI-C and the non-repressing NFI-A and NFI-B proteins, and further studies are needed to identify the specific residues within NFI-C essential for this repression.
The MMTV promoter is one of the best studied models of steroid hormone-induced gene expression. Glucocorticoid induction of the MMTV promoter is dependent on binding sites for glucocorticoid receptor, NFI-and octamer-binding proteins (33,40). However, the NFI-binding site appears dispensable for progesterone induction of the MMTV promoter, since mutation of the NFI-binding site, which abolished glucocorticoid induction, did not affect progesterone induction of the promoter (41). Also, differences between the activation potentials of GR and PR on episomal and transiently transfected MMTV promoter constructs have been reported (42). These findings, together with our observed lack of NFI-C-mediated repression of PR activation of the MMTV promoter (Fig. 4), lend strong support to the hypothesis that hormone-dependent activation of the MMTV promoter occurs through at least two different pathways (43,44). Moreover, our observation that the GR-dependent NFIGRE promoter is not repressed by NFI-C suggests that glucocorticoid induction itself may occur through two different pathways, only one of which is sensitive to repression by NFI-C.
One difference between the MMTV and NFIGRE promoters that could affect repression by NFI-C may be their ability to form phased nucleosomal arrays in vivo. The MMTV promoter contains precisely phased nucleosomes when present as autonwithout 0.1 M dexamethasone (ϩ or Ϫ Dex) for 24 h was plotted as in  2-5). Whole cell extracts were analyzed on a 7.5% SDS-polyacrylamide gel, transferred to Immobilon-P membrane, and probed with anti-HA antibodies. Numbers indicate size markers (M r ϫ 10 Ϫ3 ), and the arrow indicates the HA-tagged NFI-C polypeptide. Nonspecific bands are shown by the asterisk on the right. C, 293 cells were transfected with pMMTV␤-gal (5 g), pGL2 (2.5 g), various amounts of phGR (indicated below the figure), and 2.5 g of CMV control vector (lanes 1-6) or the NFI-C expression vector (lanes 7-12). ␤-Galactosidase and luciferase activity of extracts prepared from each cell line cultured with or without 0.1 M dexamethasone (ϩ or ϪDex) for 24 h was plotted as in Fig. 1.   FIG. 6. Repression of MMTV promoter activity by NFI-C is cell type-specific. A, cell lines (indicated above the figure) were transfected with pMMTV␤-gal (5 g), phGR (0.5 g), pGL2 (2.5 g) internal control plasmid, and 2.5 g of the CMV expression vectors (Control CMV, lanes 1 and 2, 5 and 6, 9 and 10, and 13 and 14; NFI-C, lanes 3  and 4, 7 and 8, 11 and 12, and 15 and 16). ␤-Galactosidase and luciferase activity of extracts prepared from each cell line cultured with or FIG. 7. Expression of coactivators p300 and SRC-1 abrogate NFI-C-mediated repression. Cells were transfected with pMMTV␤gal (5.0 g), phGR (0.5 g), pGL2 (2.5 g), and CMV control vector (2.5 g, lanes [1][2][3][4][5][6][7][8] or a CMV vector expressing NFI-C (2.5 g, lanes 9 -16) and cotransfected with CMV vectors expressing RAC3 (2.5 g, lanes 3, 4, 11 and 12), p300 (2.5 g, lanes 5, 6, 13, and 14), or SRC-1A (2.5 g, lanes 7, 8, 15, and 16). Cells were cultured in the absence or presence of 0.1 M dexamethasone (Ϫ or ϩDex) for 24 h. ␤-Galactosidase values were normalized to luciferase values and plotted as in Fig. 1. omously replicating episomal DNA (45). The spatial distribution of these nucleosomes was found to allow access of GR to HREs in the MMTV promoter, but to preclude access of NFI to its binding site (46). One highly regarded model for activation of MMTV transcription is that GR binds to the promoter in the presence of the phased nucleosomes and directly, or by recruitment of other proteins, initiates a chromatin remodeling event needed for NFI binding to the promoter and activation of transcription (47,48). However, nucleosome phasing has not been seen on transiently transfected MMTV templates. For example, the NFI-binding site on transiently transfected MMTV plasmids is occupied even in the absence of hormone, suggesting that the bulk fraction of transiently transfected plasmids does not contain properly positioned nucleosomes that can repress NFI binding (47). However, it is unclear what fraction of transiently transfected templates is actively transcribed in vivo, and thus measurements on the bulk population may not reflect the state of the transcriptionally active templates. Indeed, the findings that occupancy of the NFI site correlates directly with glucocorticoid induction of the MMTV promoter on episomal templates (47), together with the apparent requirement of the NFI site for glucocorticoid induction on both episomal and transiently transfected templates (33,47), suggests that NFI may function in a similar manner in GR induction on both types of template. Further studies of NFI-C repression of both episomal and transiently transfected templates will be needed to address this issue.
As noted above, the difference in the ability of NFI-C to repress glucocorticoid induction of the MMTV and NFIGRE promoters in HeLa cells could result from differences in chromatin structure between the two promoters. Consistent with this hypothesis is the apparent difference in sensitivity to intracellular GR levels of the MMTV and NFIGRE promoters. NFIGRE is partially activated by dexamethasone in the absence of cotransfected phGR (Fig. 4), whereas MMTV appears completely repressed (Fig. 5 versus Fig. 1B). Moreover, the basal expression level of NFIGRE in the absence of dexamethasone is higher than that of MMTV (Fig. 5), suggesting that the NFIGRE promoter may be more accessible to transcription factor binding even under basal conditions. Similarly, the apparent cell type-specific repression of MMTV by NFI-C may be due to differences in the extent or state of nucleosomal organization of the MMTV promoter between cell lines that are repressed (HeLa and COS-1) versus cell lines that are not repressed (JEG-3 and 293) (Fig. 6). Cell type-specific differences in NFI accessibility to DNA have previously been reported in human breast T47D and mouse C127 cells containing stably integrated copies of MMTV. The NFI-binding site of the MMTV promoter was constitutively occupied by NFI in the absence of dexamethasone in T47 cells, whereas in C127 cells the NFIbinding site was only occupied in the presence of dexamethasone (49). Studies using stably integrated or episomal MMTV templates may be needed to determine the role of chromatin structure in cell type-specific repression of the MMTV promoter by NFI-C.
The two-step model of activation of the MMTV promoter where GR binding initiates chromatin remodeling, followed by NFI binding, predicts that proteins that affect nucleosome structure likely play an integral role at the promoter. One class of proteins that can modify chromatin structure is the coactivator proteins that have histone acetyltransferase activity. Hyperacetylation of core histones is a characteristic of gene activation, and histone deacetylation is associated with transcriptional repression (reviewed in Refs. 50 and 51). Biochemical and genetic studies have identified a number of histone acetyltransferase proteins including p300 (52), CBP (52), SRC-1A (53), P/CAF (54), and ACTR (55). Coexpression of ACTR and GR has been reported to enhance dexamethasonedependent activation of a transiently transfected MMTV reporter ϳ2-fold in HeLa cells (55). This finding, together with our observations that coexpression of p300/CBP or SRC-1A enhances GR-dependent MMTV expression ϳ2-fold (Fig. 7), suggests that these histone acetyltransferase proteins may function in the glucocorticoid induction of the MMTV promoter in HeLa cells. It is important to note that while p300/CBP and SRC-1A acetylate histones in vitro and effect expression in our assay system, whether histones are the most important target for MMTV activation in vivo is not clear. Among the reasons for this uncertainty are that other targets for p300/CBP acetylation are known (56,57) and that p300/CBP and SRC-1A possess both direct transcriptional activation and protein binding domains that map outside the domains encoding histone acetyltransferase activity (52,53,58). Knowing which of these functions are required for abrogation of NFI-C repression will be important in determining the precise target of NFI-C-mediated repression.
Although the molecular basis for repression of the MMTV promoter by the NFI-C subdomain is unknown, our data are consistent with at least the following two models: 1) NFI-C interacts with GR and establishes or stabilizes a repression complex composed of GR, NFI-C, and known or unknown corepressor proteins, or 2) NFI-C interacts directly with p300, SRC-1A, or another factor that associates with the known p300⅐SRC-1A complex, and represses transcription by preventing the ability of GR to recruit or activate these proteins at the promoter. Since NFI-C represses GR-induced expression on only a subset of GR-dependent promoters, models invoking a direct inhibition of either DNA binding or transactivation function of GR by NFI-C appear less likely. Corepressor molecules, including N-Cor, SMRT and others are known to function at some steroid hormone-dependent promoters and play an essential role in the repression of thyroid hormone and retinoid-dependent transcription (59,60). In the thyroid hormone system, unliganded RXR⅐TR heterodimers recruit multiprotein corepressor complexes that contain N-Cor, SMRT and thyroid hormone receptors, and histone deacetylases to the promoter and repress transcription (59,61). Although these corepressors are not known to function on glucocorticoid-dependent promoters, it is possible that binding of the NFI-C subdomain to GR could mimic the structure of unliganded retinoic acid receptor/thyroid hormone receptor dimers and mediate active repression of the MMTV promoter. However, we have failed to see direct interaction of GR with NFI-C in DNA binding assays or by immunoprecipitation (data not shown), reducing the likelihood of this model. A model where NFI-C interacts directly with a component of a p300/CBP coactivator complex and prevents its recruitment or function at the MMTV promoter is appealing for several reasons. Previous studies have demonstrated that microinjection of CBP-inactivating antibodies into HeLa cells represses glucocorticoid induction of the MMTV promoter in a dose-responsive manner (62), suggesting an integral role for CBP in MMTV expression. Our observation that coexpression of p300/ CBP or SRC-1A abrogates NFI-C-mediated MMTV repression is consistent with a model in which repression by NFI-C occurs through interference with coactivator function. A requirement for coactivator proteins in MMTV activation is also supported by the identification of GR⅐p300⅐SRC-1A complexes in vivo (63). However, whereas expression of SRC-1A or p300/CBP enhanced GR-dependent activity of the MMTV promoter in HeLa cells ϳ2-fold, the activity of NFIGRE was not affected (data not shown), suggesting that these coactivator proteins may func-tion (or be rate-limiting) at only a subset of GR-dependent promoters. Although we favor a model in which NFI-C represses MMTV expression by interacting with a component of the p300/CBP⅐SRC-1 complex, the direct target of NFI-C repression is unknown. Thus, it would be of great interest to assess the ability to abrogate NFI-C-mediated repression of p300/CBP and SRC-1 mutant proteins that are deficient in histone acetyltransferase activity, transcriptional activation capability, or fail to interact with GR or other proteins, thereby determining the detailed molecular mechanism by which NFI-C can repress the MMTV promoter.