Negative Interaction between the RelA(p65) Subunit of NF- (cid:107) B and the Progesterone Receptor*

Interactions between transcription factors are an im- portant means of regulating gene transcription. The present study describes the mutual repression of two transcription factors, the RelA(p65) subunit of NF- (cid:107) B and the progesterone receptor (PR). This trans -repres- sion is shown to occur independent of PR isoform, reporter construct, or cell type used. Together with the demonstration of an interaction between PR and RelA in vitro , these findings suggest that the mutual repres- sion is due to a direct interaction between these proteins. Furthermore, activation of NF- (cid:107) B by tumor necrosis factor- (cid:97) also results in repression of PR, while PR is able to repress tumor necrosis factor- (cid:97) -induced NF- (cid:107) B activity. Since NF- (cid:107) B-regulating cytokine receptors are expressed in progesterone target tissues, like breast and endometrium, the mutual repression of PR and RelA could play an important role in a wide variety of physiological processes in these tissues, including maintenance of pregnancy, immunosuppression, and tumorigenesis.

Interactions between transcription factors are an important means of regulating gene transcription. The present study describes the mutual repression of two transcription factors, the RelA(p65) subunit of NF-B and the progesterone receptor (PR). This trans-repression is shown to occur independent of PR isoform, reporter construct, or cell type used. Together with the demonstration of an interaction between PR and RelA in vitro, these findings suggest that the mutual repression is due to a direct interaction between these proteins. Furthermore, activation of NF-B by tumor necrosis factor-␣ also results in repression of PR, while PR is able to repress tumor necrosis factor-␣-induced NF-B activity. Since NF-B-regulating cytokine receptors are expressed in progesterone target tissues, like breast and endometrium, the mutual repression of PR and RelA could play an important role in a wide variety of physiological processes in these tissues, including maintenance of pregnancy, immunosuppression, and tumorigenesis.
The human progesterone receptor (PR) 1 belongs to the superfamily of steroid/thyroid hormone receptors (1)(2)(3). The transcription factors of this family share (at least) two structurally related functional domains, the DBD, which contains the socalled zinc finger motif, and the more C-terminally located HBD. Two transactivation domains have been mapped in the PR, of which one is located N-terminal to the DBD. This transactivation domain, named AF-1, functions autonomously, but the level of activity depends on the cell type and reporter construct used (4,5). The second transactivation domain AF-2 lies within the HBD, and its activity is strictly dependent on the presence of ligand (4). Detailed analysis of the function of these transactivation domains of the PR has also led to more insight in anti-hormone action. Anti-progestins bind to the receptor without activating AF-2, and the partial agonistic activity evoked by the anti-progestin RU486 is therefore the result of its stimulation of AF-1 activity (4,5).
Within the superfamily of steroid receptors, PR is unique in that it exists in two isoforms, named A and B (6), which differ in their N terminus. Differences in transcriptional activity between the PR B (amino acids 1-933) and PR A (amino acids 164 -933) isoforms have been observed, depending on the cell type and the reporter construct used (4,5).
Functioning of transcription factors can be modified by interplay with transcription factors of a different type, resulting in either an inhibitory or stimulatory effect. In particular, interactions between steroid receptors and members of the AP1 family of transcription factors have been studied extensively. AP1 family members, like c-Jun and c-Fos, and GR have been shown to repress each others functioning (7)(8)(9)(10)(11), but the actual mechanism of repression is currently controversial. In vitro, a direct interaction between GR and AP1 was shown, resulting in impaired DNA binding (7,9), while others failed to detect a direct interaction (10,11) or found the two proteins to interact without DNA binding being affected (8,12). It has clearly been demonstrated that the magnitude of the repression of AP1 by GR and PR is cell type-and promoter-specific (11,13,14), suggesting that intermediary proteins are likely to be involved, the expression of which can vary between different cell types. The transcriptional activity of PR was shown to be affected by c-Jun in a stimulatory or an inhibitory fashion, depending on the cell type (11), while c-Fos was shown to inhibit PR in all cases (11). Recently, GR (15)(16)(17)(18), and also estrogen receptor (19,20), have been shown to be inhibited by members of a second family of transcription factors, the NF-B family. These repressions were found to be mutual, and GR (15,17,18) and estrogen receptor (20) were shown to interact physically with NF-B proteins in vitro. Several studies have indicated that, in the case of GR, this interaction was shown to result in impaired DNA binding (15)(16)(17), but others failed to find this (21).
The NF-B family consists of DNA-binding proteins that share homology in an N-terminal region of 300 amino acids, termed the Rel homology region. Amongst others, this family includes RelA(p65), NFKB1(p50), and c-Rel (22,23). NF-B exists as a dimer, typically a heterodimer of RelA and NFKB1 (24), but homodimers as well as heterodimers of different composition are also possible. NF-B is present in an inactive form in the cytoplasm, where it is associated to an inhibitory protein, IB. Exposure of cells to a plethora of stimuli, including cytokines (TNF-␣ and IL-1), lipopolysaccharide, UV radiation, and oxidative stress, results in the dissociation of IB from the NF-B complex, probably through a phosphorylation event (25), upon which NF-B translocates to the nucleus. Subsequently, binding to specific DNA sequences and activation of transcription can occur. For some members of the NF-B family, like RelA (26) and c-Rel (27,28), C-terminally located transactivation functions have been found.
Since NF-B sites have been identified in the promoters of numerous genes that play a role in cell proliferation and immune response (22,23), and since progestins also play a role in these processes (29,30) we have investigated the effects of NF-B family members and PR on each others transcriptional activity. Here we show the RelA subunit to specifically inhibit the hormone-induced transactivation of PR, independent of receptor isoform, reporter construct, or cell type used. Furthermore, the repression is mutual, since PR is shown to repress the transcriptional activity of RelA. Both the DBD and the HBD of PR are shown to contribute to the repression. This trans-repression between RelA and PR could play an important role in a large variety of processes like maintenance of pregnancy, immunosuppression, and tumorigenesis.

EXPERIMENTAL PROCEDURES
Materials-A phenol red-free 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium (DF) was obtained from Life Technologies, Inc. FCS was purchased from Integro (Linz, Austria), and bovine serum albumin was from Sigma, recombinant human TNF-␣ was from Boehringer Mannheim. Trypsin and EDTA used for cell culture were obtained from Flow Laboratories (Irvine, UK). The progestin Org2058 was provided by Organon International (Oss, The Netherlands), the anti-progestins RU486 and ZK98299 were obtained from Roussel-Uclaf (Romainville, France) and Schering AG (Berlin, Germany), respectively. Monoclonal antibody against the AB region of PR (AB-52; 32) was a kind gift of Dr. K. B. Horwitz (Denver, CO), while monoclonal antibody against the E region (C262; 33) of PR was purchased from StressGen (Victoria, Canada). The polyclonal antibody against the N terminus of RelA (SC-109) was purchased from Santa-Cruz (Santa Cruz, CA). Dextran-coated charcoal-FCS was prepared by treatment of FCS with dextran-coated charcoal to remove steroids, as described previously (31). HeLa 229 and COS-1 cells were obtained from American Type Culture Collection (Rockville, MD); T47D cells were originally provided by Dr. R. L. Sutherland (Sydney, Australia). Cells were cultured in bicarbonate-buffered DF medium containing phenol red, supplemented with 7.5% FCS in a 7.5% CO 2 humidified atmosphere.
Plasmid Constructs-The pSG5 expression vector containing human PR B has been described previously (34). To create an expression vector containing the human PR A isoform, PCR was performed with a primer containing the second ATG in the coding region of the human sequence and a Kozak sequence (5Ј-ggaattcgatatccacCATGAGCCGGTCCGGG-TGCAA-3Ј) and 5Ј-ggaattcGAGGCAGGATAGGCACGTGG-3Ј as reverse primer. This fragment was exchanged for the corresponding region in the pSG5-PR B construct using unique EcoRI and BalI sites. The ⌬E construct of PR B lacks amino acids 674 -933 (numbering refers to original cDNA (34,35), originally designated hPR5 (5), was a kind gift of Dr. Gronemeyer (Strasbourg, France). The ⌬AB1 construct was generated by cutting the PR B cDNA with BamHI and SauI and religating it in the presence of the oligonucleotide 5Ј-GGATCCTATCTCAACTAC-CTGAGG-3Ј, and it lacks amino acids 26 -537. The ⌬AB2 construct, lacking amino acids 26 -453, was generated by cutting with BamHI and HincII, filling-in, and religation. The ⌬AB3 construct, lacking amino acids 252-489, was generated by digestion with SacII and religation. The ⌬C construct of PR B was made by introducing two EcoRI sites into the original cDNA by site-directed mutagenesis (5Ј-CCTGAGGC-CGAATTCAGAAGC-3Ј and 5Ј-AGTCAGAGTTGTGAATTCACTGGAT-GCTGTTG-3Ј). As a consequence of the mutagenesis, amino acid 548 is changed from Asp to Asn and amino acid 650 from Ala to Ser. By cutting out the 300-base pair EcoRI fragment, amino acids 549 -649 are lacking.
The CMV-4 expression vectors containing the cDNAs encoding human RelA, NFKB1, and c-Rel have been described before (18). The ⌬TA1 construct of RelA, lacking amino acids 515-550 (numbering refers to original cDNA (37)), was made by cutting the cDNA at a unique SmaI site and ligating it into CMV-4.
GST-PR (amino acids 457-933) was made by cutting the PR B cDNA at an internal HincII site and a BamHI site in the pSG5 vector and, after filling-in, ligated into SmaI-cut PGEX-2T vector (Pharmacia, Uppsala, Sweden). GST-PR⌬C was cloned similarly, using pSG5-PR B ⌬C to isolate the HincII-BamHI fragment. To create GST-NFKB1, PCR was performed with 5Ј-tcccccgggcaccATGGCAGAAGATGATCC-3Ј as forward primer and T3 as reverse primer on the SK Ϫ -NFKB1 vector. This fragment was cut with SmaI and ligated into SmaI-cut PGEX-2T. For in vitro translation, the coding region of RelA was cut from the CMV4-RelA vector, using XbaI and HindIII sites outside the coding region, and cloned into XbaI-HindIII-cut pBluescript SK Ϫ .
Transient Transfections-Cells were cultured in six-well tissue culture plates in DFϩ, supplemented with 5% dextran-coated charcoal-FCS at a density of 2 ϫ 10 4 /cm 2 or 3 ϫ 10 4 /cm 2 for T47D cells. Cells were transfected by calcium phosphate co-precipitation using 5 g of CAT reporter or 2 g of luciferase reporter, 3 g of PDMLacZ plasmid, and 100 ng of eukaryotic expression plasmid pSG5 containing the PR cDNAs or CMV-4 plasmids containing the NF-B cDNAs. pBluescript SK Ϫ plasmid was added to obtain a total amount of 10 g of DNA/well. After 16 h, or 6 h in the case of T47D cells, the medium was refreshed and (anti-) hormones were added. Cells were harvested 24 h later and assayed for CAT activity (34,42) or luciferase activity (18) as described. Values were corrected for transfection efficiency by measuring ␤-galactosidase activity (43).
In Vitro Binding Assays-Recombinant RelA in pBluescript SK Ϫ was transcribed and translated in vitro in reticulocyte lysate (Promega, Madison, WI) in the presence of [ 35 S]methionine according to the manufacturer's description. GST-fusion vectors were transformed into Escherichia coli BL21. Expression and purification with glutathione-coated Sepharose beads (Pharmacia) was performed as described previously (44). The fusion proteins loaded on Sepharose beads were subsequently incubated with in vitro translated protein in NETN (20 mM Tris (pH 7.5), 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, 1 mM dithiothreitol) containing protease inhibitors (0.5 mg/ml bactracin, 5 g/ml leupeptin, 5 g/ml pepstatin, 2 mM phenylmethylsulfonyl fluoride) for 2 h in the absence or presence of 1 M Org2058. Beads were washed 4 times with NETN, dried under vacuum, resuspended in sample buffer, and analyzed by SDS-polyacrylamide gel electrophoresis. Finally, gels were dried and, after amplification by fluorography (Amplify, Amersham Corp.), exposed at Ϫ70°C.
Western Blotting-COS-1 cells were transfected as described above. For hormone-dependent phosphorylation of PR, cells were treated with Org2058 (10 nM) for 4 h. Subsequently, cells were harvested immediately in SDS sample buffer. Samples were separated on 8% SDSpolyacrylamide gels and transferred onto Immobilon (Milipore, Bedford, MA). Blots were blocked with Blotto (phosphate-buffered saline containing 4% nonfat died milk powder and 0.05% Tween 20) for 1 h. All subsequent steps were carried out in Blotto/phosphate-buffered saline (1:1). Blots were probed with monoclonal antibodies AB-52 or C262 against PR or the SC-109 antibody against RelA. After vigorous washing, blots were incubated with peroxidase-conjugated antibodies (1: 10,000; Amersham Corp.). Subsequently, blots were washed again and immunoreactive bands were visualized with ECL (Amersham Corp.).
RNA Isolation and Northern Blotting-Total RNA was isolated by the acid-phenol method (45). Northern blotting and (pre-) hybridization with the fatty acid synthase cDNA (46) was carried out as described previously (47). For control hybridizations, a rat glyceraldehyde 3-phosphate dehydrogenase cDNA was used (48). Probes were labeled with [␣-32 P]dCTP using the Multiprime DNA labeling system (Amersham Corp.). Final (most stringent) washing was done in 0.1 ϫ SSC, 0.1% SDS at 68°C. Subsequently filters were exposed for autoradiography and quantified using a Molecular Dymanics PhosphorImager with Image Quant software.

Mutual Repression of PR and
RelA-To study the effects of NF-B on progesterone-mediated gene transcription, we transfected HeLa cells with a reporter construct containing two progesterone responsive elements (PREs) in front of the thymidine kinase promoter coupled to CAT, in combination with expression constructs expressing the A and B isoform of the human PR and expression constructs containing the RelA(p65) subunit of NF-B. As shown in Fig. 1A, RelA is clearly capable of repressing the transcriptional activity of PR A and PR B induced by the synthetic progestin Org2058 (10 nM). Only marginal repressive effects of RelA were observed in the absence of hormone (data not shown). The repression of activated PR is specific for RelA, since co-transfection with other NF-B family members, like NFKB1(p50) or c-Rel, did not affect the activity of PR A or PR B (Fig. 1B).
Next, transient transfections were performed with a reporter construct containing the MMTV promoter, which contains numerous PREs (38). As shown before (5), a marked difference between the activities of PR A and PR B is observed, with PR A being a weak activator of this reporter construct (Fig. 1C). Again, hormone-induced transactivation by both PR B and PR A is repressed by RelA (Fig. 1C). Similar results were obtained when a reporter construct was used in which two PREs are located directly upstream of a TATA box (data not shown). We conclude therefore that the RelA subunit of NF-B is capable of repressing both PR A and PR B , independent of the context of the PRE.
To examine whether PR is also able to repress RelA, HeLa cells were transfected with a reporter construct containing three NF-B sites from the human ICAM promoter (40), coupled to the thymidine kinase promoter in front of luciferase (21). Co-transfection with RelA expression vector (100 ng) resulted in a modest but significant activation of this construct (Fig. 2). When expression vectors containing PR A or PR B (1 g) were added, the RelA-mediated activity was already reduced in the absence of hormone, while hormone addition resulted in a further repression (Fig. 2). Similar results were obtained when 10-fold lower amounts of RelA and PR expression vectors were used (data not shown). Taken together, these results show that the repression between PR and RelA is mutual and occurs on a variety of promoters.
Cell Type Specificity of the Mutual Repression of PR and RelA-Since the mutual repression between PR and RelA could depend on cell type-specific factors, transient transfection assays were performed not only in HeLa cells but also in COS-1 cells and in the human breast tumor cell line T47D, which expresses high endogenous levels of both PR A and PR B in approximately equal amounts (data not shown). As was also shown in HeLa cells (Fig. 1), the transactivation potential of PR A and PR B is different in COS-1 cells (Table I). In both cell lines, however, RelA clearly inhibited the hormone-induced transcriptional activity of both PR A and PR B . RelA is also capable of repressing endogenous PR, as shown in T47D cells.
In addition, the effects of PR A and PR B on RelA-induced reporter activity were studied in the same cell lines. The activation by RelA was much greater in COS-1 cells than the modest activation in HeLa cells (Fig. 2), but also in these cells both PR isoforms inhibited RelA transactivation and progestin treatment resulted in a further repression of RelA-induced reporter activity (Table I). Furthermore, endogenous PR also represses RelA, as shown in T47D cells. Although the transactivation and repression potential of RelA and the PR isoforms differ between cell lines, the mutual repression between these proteins is observed clearly in all cell lines tested.
Domains of PR and RelA Involved in the Mutual Repression-To identify the regions of PR involved in the negative cross-talk with RelA, deletion constructs of PR were used which lack (part of) the AB region, the C region, or the E region (Fig.  3A). Expression of these PR deletion constructs was found to be approximately equal, as detected on Western blots, using monoclonal antibodies directed against the AB region (AB-52; The activity of PR B ⌬E, which functions independent of hormone since it lacks AF-2, was still repressed by RelA (Fig. 3B). RelA also repressed the activity of PR⌬AB1, which is strictly hormone-dependent since it only contains AF-2 (Fig. 3B). These findings suggest that not the two AFs of PR, but different domains are essential for the repression by RelA. To examine whether the same holds true for the repressor function of PR, the same PR constructs were used to examine their effect on RelA-induced activity of the 3xNF-B reporter. For this, COS-1 cells were used, in which co-transfection of RelA results in a stronger activation of this reporter (Table I) than in HeLa cells (Fig. 2). Unliganded PR A and PR B already repress RelA to some extent (Fig. 3C). This effect is lost when amino acids 26 -537 of the AB region of PR are deleted (PR⌬AB1; Fig. 3), but not when amino acids 26 -453 (PR⌬AB2) or 252-489 (PR B ⌬AB3) are deleted. Therefore, hormone-independent repression by PR requires a region between amino acids 489 and 537, but not AF-1, since this encompasses amino acids 455-546 (40). All PR constructs mentioned above lower the transcriptional activity of RelA even more upon hormone treatment (Fig. 3C). When the HBD was deleted (PR B ⌬E), the resultant receptor repressed RelA-induced transcription independent of hormone, to a level comparable with the hormone-independent repression by PR A or PR B (Fig. 3C), showing that the HBD is essential for hormone-dependent repression. Both the hormone-dependent and -independent repressive activity were lost when the DBD of PR was deleted (⌬C). These findings show that while (part of) the AB domain and the E domain are required for hormone-independent and -dependent repression, respectively, both types of repression require the C domain.
To examine whether the transactivation function of RelA is essential for the repression of PR, a deletion construct of RelA was used that lacks the 30 most C-terminal amino acids, which encode transactivation domain TA1 (26). Only a marginal difference between the repressive activity on PR A and PR B of this construct compared with the full-length RelA protein was observed (Fig. 3D). Similar results were obtained in COS-1 and T47D cells (data not shown). Therefore, TA1 is not essential for the repressive activity of RelA.
Effects of Anti-progestins on the Mutual Repression of PR and RelA-Like progestins (11), the anti-progestin RU486 has been shown to be capable of inducing PR-mediated repression of AP1 (49), while it has antagonistic and partial agonistic activity with respect to PR-mediated transcription, depending on the cell type and reporter construct used (4,5). Therefore, we used this anti-progestin to examine the mutual repression between PR and RelA in more detail. As described before (5), RU486 behaved as a weak agonist on PR B when PRE 2 tkCAT is used as a reporter (Fig. 4A), while PR A was unaffected (Ref. 5; data not shown). Co-transfection of RelA resulted in repression of the transcriptional activity of the RU486-occupied PR B (Fig. 4A).
To investigate whether antagonist-occupied PR could also inhibit RelA-mediated transcription, transfections in COS-1 cells were performed with the 3xNF-B reporter described above. RU486 was able to induce PR A -or PR B -mediated repression of RelA (Fig. 4B), albeit to a lesser extent than the agonist Org2058 (Table I). The RU486-evoked repression was more easily observed when the PR⌬AB construct of PR is used (Fig.  4B), which lacks the hormone-independent repression observed with PR A and PR B (Fig. 3C). While RU486 only exhibits weak agonistic effects under certain specific conditions (Fig. 4A), the "pure" antagonist ZK98299 is unable to evoke PR-mediated transactivation (data not shown). Like RU486, this anti-pro-  gestin was able to repress RelA (Fig. 4B). Together with RelArepression by RU486-occupied PR⌬AB, which lacks AF-1 through which RU486 is thought to exert its agonistic action (5), these results indicate that the enhancer and repressor functions of the two PR isoforms are separate functions. This hypothesis is confirmed by our findings that the transcriptional activity of the two PR isoforms differs in HeLa and COS-1 cells, while the repressive effect on RelA is independent of receptorisoform and cell type (Table I).
Direct Physical Interaction between PR and RelA in Vitro-To investigate the possibility that RelA and PR repress each others transcriptional activity via a direct physical interaction, the PR cDNA (amino acids 457-933) and the PR cDNA lacking the C region (PR⌬C) were fused to GST. The complete coding region of NFKB1, a protein that can associate directly with RelA (24), was also fused to GST. Upon overexpression in E. coli, GST-fusions were purified with glutathione-coated agarose beads (44), and tested for their ability to bind in vitro translated, [ 35 S]methionine-labeled RelA. As shown in Fig. 5, the RelA protein could not be precipitated by GST alone (lane 2), while the GST-PR fusion protein clearly precipitated RelA (lane 3), indicating that the two proteins can bind directly to each other, while GST-PR⌬C hardly bound RelA (lane 5). In all cases, the additional presence of hormone had no effect (lanes 4 and 6). Bacterially expressed NFKB1 was clearly capable of interacting with in vitro translated RelA (lane 7). Therefore, we conclude that at least in vitro PR and RelA are capable of physically interacting with each other. Furthermore, the C region of PR, which was shown to be essential for the repression of RelA (Fig. 3C), is also essential for the in vitro interaction with PR.
Effects of RelA on PR Phosphorylation-Since PR becomes sequentially hyperphosphorylated upon hormone and DNA binding, reflected by reduced mobility in SDS-polyacrylamide gels (50,51), the effect of RelA on this process was examined. COS-1 cells were transiently transfected with PR A or PR B in the absence or presence of RelA. Subsequently, cells were treated with Org2058 for 4 h. Treatment with Org2058 results in reduced mobility of PR A and PR B , but co-transfection of RelA has no effect (Fig. 6), as was also true for RU486-treated cells (data not shown), indicating that hyperphosphorylation of PR, and therefore probably DNA binding, are not prevented by RelA.
Negative Interaction between Cytokine and Progesterone Signaling-To determine whether the mutual repression between RelA and PR can also be observed when NF-B is activated by cytokines, several experiments were performed. First, the effects of TNF-␣ on PR activity were studied in HeLa and T47D cells. Gel retardation assays showed that TNF-␣ treatment of these cells results in binding of proteins to the NF-B site located in the ICAM-1 promoter (data not shown). When transient transfections were performed in HeLa cells, a clear inhibition of the hormone-induced transcriptional activity of PR A and PR B was observed when TNF-␣ was added (Table II). Transcriptional activity of PR was also inhibited by TNF-␣ in T47D (Table II), although less pronounced than in HeLa cells. This may be due to the relatively weak induction of NF-B, as assessed by measuring transactivation of a 3xNF-BtkLuc reporter (Table II). Alternatively, other TNF-␣ induced factors, like AP1 family members, could prevent effective repression of PR through their ability to bind to NF-B (17,52). The effect of PR A and PR B on TNF-␣-induced NF-B activity was also examined. Treatment with progestins resulted in a repression of the NF-B activity evoked by TNF-␣, both in HeLa and T47D cells (Table II). To study if the PR/NF-B interaction could have relevance for the in vivo situation, the influence of NF-B inducing reagents on the expression of a PR target gene was examined. To this end, T47D cells were treated with Org2058 (10 nM) for 16 h in the absence or presence of TNF-␣ (250 units/ml) or H 2 O 2 (150 M), and total RNA was isolated. Northern blot analysis showed that the induction of the progestin-induced fatty acid synthetase mRNA was decreased both by TNF-␣ and H 2 O 2 (Fig. 7A), with TNF-␣ treatment resulting in a 30% decrease and H 2 O 2 in a 60% decrease of induction (Fig. 7B). Although the reduction by TNF-␣ was relatively small, it is in line with the repression of PR mediated transactivation by TNF-␣ in these cells (Table II), which is probably caused by their inherent low sensitivity to TNF-␣. From these results, we conclude that TNF-␣-induced NF-B activity can repress PR functioning and vice versa and that this trans-repression is likely to be operational in vivo.

DISCUSSION
In the present study, we show mutual repression of the hormone-activated PR and the RelA(p65) subunit of NF-B. This mutual repression could be caused by several mechanisms. The first possibility is that RelA and PR are able to bind to their respective cognate DNA elements. However, treatment of T47D cells with progestins did not result in specific complex formation on the NF-B site from the ICAM-1 promoter. 2 Also in the case of the mutual inhibition of GR and AP1, no evidence was found for this mechanism (7, 9 -11). Second, PR and RelA could compete for common co-activators or transcription intermediary factors, a process referred to as transcriptional interference or squelching (53,54). This is unlikely, since we found the repression of PR by RelA to occur independent of receptor isoforms, transactivation functions of PR, reporter construct, and cell type and since repression of PR by RelA was independent of the main transactivation domain TA1 of RelA. A third possibility is that a direct interaction between the two proteins, resulting in a heterodimer, could account for the repression. In vitro association assays showed that the two proteins are indeed capable of a direct physical interaction. Such a complex could either (i) be unable to bind DNA or (ii) result in the formation of inactive complexes on the DNA by preventing the interaction with essential co-factors or the basal transcription machinery. Our data are in line with the second mechanism, since TNF-␣-induced DNA binding of NF-B was found to be unaffected by the presence of PR in gel retardation assays. 2 Furthermore, we also found the hormone-induced change in mobility of PR, which is caused by DNA-dependent phosphorylation events (51,52), to occur irrespective of the presence of RelA, indicating that binding of PR to DNA is not prevented by RelA. Similarly, GR was shown to interfere with AP1 activity without altering its DNA binding, as shown by in vivo footprinting (12).
As was also shown for the repression of AP1 by GR (7-9), we found the C and E domains of PR to be essential for repression of RelA. With respect to the HBD, this domain is likely to function differently with respect to the enhancer and repressor functions of PR. The ability of this domain to enhance transcription is probably not essential for the repression of PR, since anti-progestin-occupied receptors, which are unable to evoke transactivation, still induced repression of RelA. Together with the cell type-specific differences between the two PR isoforms in transcriptional activity, but not in repressive function, these findings show that the ability to enhance or repress transcription are separate functions within the steroid receptor proteins.
Although the interactions of steroid receptors with transcription factors of the AP1 and NF-B families share several features, notable differences are also evident. The repressive action of c-Fos was mainly directed against AF-2 of PR (11), while we show RelA to inhibit the transactivation of a PR construct lacking this domain (PR B ⌬E). Second, the repression of AP1  proteins by PR was shown to be cell type-and promoter-specific (11), indicating that additional, promoter-specific proteins are involved also, the expression of which could be different in various cell types. While the transactivation potential of RelA and the PR isoforms was shown to differ, the repression of RelA by PR appeared to be independent of cell type and promoter context. Together with the association between PR and RelA in vitro, these data indicate that the mutual repression between PR and RelA could be due to a direct interaction between the proteins, without the additional involvement of other proteins.
Deletion analysis of RelA showed that the region corresponding to TA1 (amino acids 515-550; Ref. 26) is not required for the repression of PR, indicating that other regions of the protein are involved in the interaction with PR. Preliminary results indeed indicate that regions outside the TA1 domain of RelA are essential for repression of PR and GR. 3 This could explain why the NFKB1(p50) and c-Rel proteins, which are substantially different from RelA outside the Rel homology region, are unable to repress PR (this study) and GR (18).
The interaction between PR and RelA is potentially important in organs in which the PR together with NF-B-regulating cytokine receptors is expressed, like mammary gland, ovarium, and endometrium. During pregnancy, progesterone levels are high, and the presence of progesterone is essential for the maintenance of pregnancy (30). In endometrial cells, cytokines, which induce NF-B, like TNF-␣ and IL-1, and their receptors are expressed, as well as PR (55). Progesterone-induced decidualization of endometrial cells, which is thought to be important in maintaining pregnancy, is inhibited in vitro by TNF-␣ and IL-1 in endometrial cells (56,57). Second, the expression of IL-8, which is both a chemotactic factor for neutrophils and causes them to secrete lysosomal enzymes, is repressed by progesterone (58). Since these effects of IL-8 may, besides playing a role in inflammation, also be an early step in the initiation of labor (59), suppression of its secretion prevents premature birth. The recent finding that in mice that carry a null mutation of the PR, uterine inflammations occur frequently (60) is in line with this hypothesis. Recently, NF-B-induced expression of IL-8 was shown to be repressed by GR (16) in a mechanism similar to what we propose for PR. Another important function of progesterone during pregnancy is its immunosuppressive effect, to prevent activation of an immune response directed against the embryo (30). A number of genes that are important in the immune system have been shown to be regulated by NF-B (22,23). It is therefore possible that the immunosuppressive action of progesterone during pregnancy is partly due to the inhibition of the transcriptional activity of NF-B.
The negative cross-talk between RelA and PR could also play a role in cell proliferation, since both NF-B (23) and progestins (29) have been implicated in this process. Several lines of evidence suggest that constitutive activation of NF-B contributes to the malignant phenotype of tumor cells. First, a naturally occurring splice variant of RelA, named p65⌬, was shown to transform Rat-1 cells (61). Higgins et al. (62) have shown proliferation and tumorigenicity of several tumor cell lines, including the human breast tumor cell lines MCF7 and T47D, to be inhibited by antisense oligonucleotides to RelA. In addition, activation of NF-B through the disruption of IB␣ regulation, was shown to result in malignant transformation (63). In contrast to their different ability to transactivate, both progestins and anti-progestins can be used to treat breast tumors (64). Therefore, it seems reasonable to propose that trans-repression of NF-B (this study) and AP1 (11), which is induced by both types of ligand, could be relevant for tumor inhibition.