Selective Binding of Steroid Hormone Receptors to Octamer Transcription Factors Determines Transcriptional Synergism at the Mouse Mammary Tumor Virus Promoter*

Transcriptional synergism between glucocorticoid receptor (GR) and octamer transcription factors 1 and 2 (Oct-1 and Oct-2) in the induction of mouse mammary tumor virus (MMTV) transcription has been proposed to be mediated through directed recruitment of the octamer factors to their binding sites in the viral long terminal repeat. This recruitment correlates with direct binding between the GR DNA binding domain and the POU domain of the octamer factors. In present study, in vitro experiments identified several nuclear hormone receptors to have the potential to bind to the POU domains of Oct-1 and Oct-2 through their DNA binding domains, suggesting that POU domain binding may be a property shared by many nuclear hormone receptors. However, physiologically relevant binding to the POU domain appeared to be a property restricted to only a few nuclear receptors as only GR, progesterone receptor (PR), and androgen receptor (AR), were found to interact physically and functionally with Oct-1 and Oct-2 in transfected cells. Thus GR, PR, and AR efficiently promoted the recruitment of Oct-2 to adjacent octamer motifs in the cell, whereas mineralocorticoid receptor (MR), estrogen receptor α, and retinoid X receptor failed to facilitate octamer factor DNA binding. For MMTV, although GR and MR both induced transcription efficiently, mutation of the promoter proximal octamer motifs strongly decreased GR-induced transcription without affecting the total level of reporter gene activity in response to MR. These results suggest that the configuration of the hormone response element within the MMTV long terminal repeat may promote a dependence for the glucocorticoid response upon the recruitment of octamer transcription factors to their response elements within the viral promoter.

The nuclear hormone receptor superfamily is distinguished by a striking conservation of the receptor DNA binding domains and highly redundant DNA sequence recognition. DNA binding by nuclear receptors is almost exclusively dependent upon the orientation and spacing of just two core DNA recognition sequences (1). Four steroid hormone receptors (glucocorticoid, mineralocorticoid, androgen, and progesterone) bind as homodimers to a palindrome of the sequence AGAACA with a 3-base pair spacing, whereas the estrogen receptors bind as dimers to a similar arrangement of the second core sequence; AGGTCA, and most other nuclear receptors recognize alternative arrangements of the same sequence (2).
Despite recognizing the same DNA sequence elements, glucocorticoid receptor (GR), 1 mineralocorticoid receptor (MR), progesterone receptor (PR), and androgen receptor (AR) differentially regulate the transcription of target genes in response to steroid (3)(4)(5)(6). For example, whereas the hormone response element of the mouse sex-limited protein gene is bound by GR and PR in addition to AR, sex-limited protein gene expression is specifically induced by androgens and is refractory to progestins and glucocorticoids (7). This specificity in responsiveness is determined by the specific interaction of AR with factors binding to auxiliary elements within the sex-limited protein gene enhancer (8).
The major sites of expression for mouse mammary tumor virus are the lactating mammary gland and the testis. Steroidinduced MMTV transcription is dependent upon a complex hormone response element (HRE) responsive to GR, AR, PR, and MR, as well as binding sites for nuclear factor 1 (NF1) and octamer transcription factors (5, 9 -13). The role of NF1 in steroid induction appears to be mediated indirectly, with NF1 occupancy of the viral promoter being primarily dependent upon the induction of changes in chromatin structure by the steroid receptor (14). By contrast, there is evidence that the octamer transcription factors participate in the steroid hormone response through a protein-protein interaction with GR and PR (15)(16)(17). Recently, we have suggested that a proteinprotein interaction between the GR DBD and the POU domains of Oct-1 and Oct-2 potentiates the glucocorticoid hormone responsiveness of MMTV by a recruitment mechanism that may facilitate or promote the binding of the octamer factors to their recognition sequences in the MMTV promoter (16). This sug-gestion is consistent with the cooperativity in DNA binding observed for GR and Oct-1 and Oct-2 that has been observed both in vitro and in vivo (14 -17). Binding between GR and Oct-1 and Oct-2 was found to be direct 2 and is likely to occur in vivo for the full-length proteins as well as for the isolated domains.
Analysis of the GR-Oct-1/2 interaction has demonstrated that the GR DBD is sufficient for Oct-1 and Oct-2 binding and that point mutations at Cys-500 and Leu-501 of the GR DBD abrogate the interaction between the full-length proteins (16). Localization of POU domain binding with the region of GR that is most highly conserved among nuclear hormone receptors (1) suggested that interaction with the POU domains of Oct-1 and Oct-2 may be a broadly conserved property of nuclear hormone receptors.
In the present study, we have evaluated the potential for other nuclear hormone receptors to interact with Oct-1 and Oct-2. Our results indicate that whereas similarities within the nuclear receptor DBD allowed several nuclear receptors to bind specifically to the octamer factor POU domains in vitro, interactions in transfected cells were limited to AR, PR, and GR. On the MMTV promoter AR, PR, and GR, but not MR, promoted the binding of Oct-2 to its recognition sequences. The inability of MR to promote octamer factor binding to the MMTV promoter correlated with a lack of transcriptional synergism between the MMTV HRE and octamer motifs in response to the activation of viral transcription by MR. Surprisingly, however, MR activated transcription from the wild-type MMTV promoter in CHO cells with efficiency similar to that of GR, whereas induction by GR alone was severely compromised from a promoter in which the octamer motifs were inactivated by site-directed mutagenesis. By contrast, GR and MR activated transcription similarly from a synthetic steroid response element. These results indicate that the requirement of the octamer motifs for GR-mediated induction may reflect a specific limitation in the configuration of the MMTV HRE that translates to a decrease in the transcriptional activation potential of GR in fibroblasts.
Tissue Culture and Transient Transfections-CHO-K1 cells (ATCC) were maintained in ␣-minimal essential medium supplemented with 10% fetal bovine serum. For chloramphenicol acetyltransferase (CAT) assays, cells were transfected with LipofectAMINE (Life Technologies, Inc.) (16). Briefly, CHO cells were plated onto 60-mm plates 24 h prior to transfection, and the DNA-LipofectAMINE complexes (10 l of Lipo-fectAMINE for the CAT assays and 20 l of LipofectAMINE for the coimmunoprecipitation assays) were incubated for 5 h on 60% confluent plates. The transfections were terminated by addition of fetal bovine serum to 10%.
The cells for the one-hybrid and immunoprecipitation assays were harvested 24 -48 h following transfection, whereas the cells used in the CAT assays with the full-length GR or MR were treated for 4 h with 10 Ϫ6 M dexamethasone (dex) or 10 Ϫ6 M aldosterone, respectively. To control for variation in transfection efficiency, CAT activities were standardized against ␤-galactosidase activity derived from the enzyme produced from pSV2-␤Gal (150 ng/plate) uniformly cotransfected. Each experiment was performed in duplicate a minimum of three independent times. Error bars represent the S.E. of the mean.
Transfections for in vivo footprinting were performed by calcium phosphate precipitation with CHO cells in 100-mm plates exactly as described previously (16). The DNA for the transfections included 1.5 g of footprinting template (pBSGalOCT, pBSGalIAP, or pMMTVCAT) and 1.5 g of pGalDBD/fusion protein expression plasmid, or plasmids encoding full-length nuclear receptors (GR, PR A , AR, or MR) supplemented with highly sheared salmon sperm DNA to 5 g. Cells were harvested 24 h following transfection. In the footprinting experiments cells expressing full-length GR, PR A , AR, or MR, cells were treated for 15 min with either 10 Ϫ6 M dex, R5020, dihydrotestosterone, and aldosterone, respectively.
GST Pull-down and Coimmunoprecipitation Assays-Recombinant GST, GSTOct-1POU, were produced in and purified from Escherichia coli (BL21 pLys S). Cells were grown to late log phase and induced to express the fusion proteins by addition of 0.4 mM isopropyl-1-thio-␤-Dgalactopyranoside for 2 h. The cells were harvested and lysed in lysis buffer (25 mM HEPES, pH 7.9, 100 mM KCl, 12% glycerol, 1.5 mM EDTA, 1 mM dithiothreitol, 0.15 mM phenylmethylsulfonyl fluoride, 0.1% Nonidet P-40) by sonication, and the recombinant proteins were purified from the bacterial lysates on glutathione-Sepharose (Amersham Pharmacia Biotech). Recombinant protein recovery was assessed by Coomassie Blue staining following SDS-polyacrylamide gel electrophoresis.
For immunoprecipitation experiments, nuclear extracts were prepared from transiently transfected CHO cells (33). Gal4 DBD fusion proteins were quantified by Western blot analysis with a Gal4 monoclonal antibody (sc-510, Santa Cruz Biotechnology). After equalizing the concentration of the fusion proteins in the extracts by the addition of control extract from untransfected cells, binding of in vitro translated Oct-2 to immunoprecipitated Gal4 fusion proteins was performed as described previously (16). The bound proteins were resolved by SDSpolyacrylamide gel electrophoresis and visualized by fluorography.
Footprinting of Nuclear Factor Binding to Transiently Transfected DNA Templates-Footprinting of nuclear factor binding to transiently transfected DNAs was performed as described in detail in Préfontaine et al. (16). Briefly, nuclei from cells transfected with pBSGalOCT or pBSGalIAP were digested with 100 units of XhoI and nuclei from cells transfected pMMTVCAT with 100 units of HindIII, together with 15 units of exonuclease (Life Technologies, Inc.) for 15 min at 30°C. After purification, the plasmid DNA was extended by linear PCR, using a 32 P-labeled T7 primer for pBSGalOCT and pBSGalIAP, or a 32 P-labeled primer from ϩ74 to ϩ52 of the MMTV LTR for pHCWT (16). The PCR products were resolved by denaturing sequencing gels and visualized by radiography.

Association of the GR, PR, and AR DBDs with Oct-1 and
Oct-2-To examine the potential for nuclear hormone receptors other than GR to bind to the POU domains of Oct-1 and -2, we tested the ability of several in vitro translated receptors to bind to the Oct-1 POU domain in a GST pull-down assay (Fig. 1). Somewhat surprisingly, all of the receptors tested in this experiment interacted with the Oct-1 POU domain (lanes 1-8). This included four steroid receptors (GR, ER␣, AR, and PR A ) (18 -21), a thyroid hormone receptor (TR␣2) (24), RAR and RXR isoforms (RAR␣Ј and RXR␣) (22,23), and even the Drosophila orphan receptor FTZ-F1␣ (25). For the steroid hormone receptors, binding was ligand-dependent, 3 and for GR binding was sensitive to the L501P substitution in the DBD (16). A similar profile of nuclear receptor binding was also observed with a GST fusion protein of the POU domain of Oct-2, but no significant binding was observed to GST alone. 3 Nuclear receptor binding to the POU domain appeared to be specific, as in vitro translated firefly luciferase was unable to bind to the POU domain (lane 9). We have also previously demonstrated that in vitro translated cAMP-response element-binding protein (34) is unable to interact specifically with the GR DBD under these binding conditions (16).
A closer examination of the binding of RAR␣Ј to the POU domain performed with C-terminally truncated RARs demonstrated that POU domain binding was lost upon truncation into the C terminus of the RAR DBD (Fig. 2). This was exactly consistent with the C-terminal boundary of the determinants of GR required for binding to the Oct-1 POU domain reported previously (16). Thus it appeared that at least several nuclear receptors had the potential to bind specifically to the POU domains of Oct-1 and Oct-2 and that this binding likely occurred through similar interactions involving the receptor DBDs.
To determine whether this in vitro result reflected the ability of nuclear hormone receptors to associate through their DBDs with the POU domains of octamer transcription factors in the cell, we examined the interaction of the nuclear receptor DBDs with Oct-2 in a mammalian one-hybrid assay (Fig. 3). In this assay, CHO cells were transfected with a CAT reporter gene whose transcription was under the control of an adenovirus E1B minimal promoter flanked by five DNA-binding sites for the yeast Gal4 protein (26). Cotransfection of a vector expressing Oct-2 had no effect on CAT activity from this promoter (lane 10), whereas expression of the Gal4 DBD protein alone had little effect. 4 However, coexpression of Oct-2 with the Gal-GR, -PR, and -AR fusion proteins induced CAT activity 3-6-fold (lanes 1, 3, and 4). For GR (lane 2), this activation was completely sensitive to the L501P mutation in the DBD that abrogated POU domain binding (16). Interestingly, the differences in the Oct-2-dependent CAT activity induced with the GR, PR, and AR DBDs did not simply reflect the level of expression of these fusion proteins in the cell (Fig. 4A) but rather suggested differences in the affinity of their association with Oct-2. By contrast, Gal-MR, -ER␣, -RXR␣, -RAR␣, and -FTZ-F1␣ DBD fusion proteins were completely unable to stimulate CAT activity (lanes 5-9). Therefore, whereas many nuclear receptors could associate with the POU domains of Oct-1/2 when the POU domains were present in vast excess to the receptors in vitro, only the GR, PR, and AR DBDs appeared to interact with the POU domain with sufficient affinity, or specificity, to be detectable under stringent binding conditions in the cell.
To extend our analysis of the binding of steroid receptors to the octamer factors, we tested the ability of the Gal4 steroid receptor DBD fusion proteins immunoprecipitated from nuclear extracts to bind to in vitro translated Oct-2 (Fig. 4). Western analysis with a Gal4 DBD antibody was used to verify that the binding assays were performed with similar amounts of each nuclear receptor-Gal4 DBD fusion protein (Fig. 4A). In these experiments, in vitro translated Oct-2 bound strongly to the GR, PR, and AR DBDs. However, little binding was ob- served with the Gal-MR and -ER␣ fusion proteins. (Fig. 4B). These results confirmed that under stringent binding conditions, only the DBDs of GR, PR, and AR associated with Oct-2.
GR, AR, and PR, but not MR, Promote the Binding of Oct-2 to the MMTV Promoter-GR, AR, PR, and MR bind to common DNA sequences (3). Previously, we demonstrated that GR promoted the binding of Oct-2 to octamer motifs in the MMTV promoter of transiently transfected plasmid DNAs and that the Gal-GR DBD fusion proteins promoted Oct-2 binding to plasmids containing single Gal4-binding sites and octamer motifs (16). To determine whether these properties also extended to PR and AR, we performed exonuclease footprinting assays on transiently transfected plasmid DNAs. The use of transiently transfected DNAs for these assays allows an examination of the DNA binding properties of transcription factors in the nucleus in the absence of an ordered chromatin structure. This is an advantage because it dissociates changes in DNA binding due to specific protein interactions from alterations that occur in response to changes in DNA accessibility resulting from the reorganization of chromatin (9).
In our first experiment, nuclei prepared from cells trans-fected with the Gal4-nuclear receptor fusion proteins, fulllength Oct-2, and a DNA template containing single Gal4 and Oct-2-binding sites were restricted and digested with exonuclease (Fig. 5). Occupancy of the octamer motif is revealed by a pause in the progress of the exonuclease adjacent to the octamer motif that is revealed by linear PCR (16,35). By contrast, the Gal4 DBD does not bind to DNA in a manner that provides a barrier to exonuclease digestion (16). As expected (16), Oct-2 binding to the octamer motif on the test construct was dependent upon coexpression of the Gal-GR fusion protein and was completely sensitive to the C500Y mutation in the GR DBD (lanes 2-12). In addition, Oct-2 binding to the octamer motif also was promoted by the presence of Gal-PR A and Gal-AR (lanes 13-16). By contrast, Gal-MR, Gal-ER␣, and Gal-RXR␣, all expressed at similar levels, were unable to promote detectable Oct-2 DNA binding (lanes [17][18][19][20][21][22]. Finally, the promotion of Oct-2 binding was dependent upon the presence of the octamer motif (lanes 23-28). One implication of these results was that on a natural promoter with common binding sites for GR, PR, AR, and MR, MR should differ from the other three receptors by being unable to promote the occupancy of octamer motifs. To test this hypothesis, we repeated our footprinting assay to examine the binding of Oct-2 to the MMTV promoter in response to steroid hormones (Fig. 6). The binding of Oct-2 to the MMTV promoter in this assay is reflected by a pause in exonuclease digestion at Ϫ49 (16,35). Under our experimental conditions, no binding to the octamer motif was detected in the absence of added steroid. Treatment of cells transfected to express full-length GR, PR, and AR with glucocorticoid agonist dex, the synthetic progestin R5020, and dihydrotestosterone, respectively, induced the binding of Oct-2 to the octamer motifs (lanes 1-10). By contrast, the treatment of cells expressing MR with aldosterone had no effect on the binding of Oct-2 to the MMTV promoter (lanes 12  and 13). In all instances, NF1 binding to the LTR, reflected by a pause at Ϫ75, occurred at similar levels, demonstrating that octamer factor binding directly reflected the hormonal status of the cells as opposed to the accessibility of the plasmid DNA.
Differential Dependence of the Hormonal Responsiveness of GR and MR on Octamer Motifs and HRE Configuration-The inability of MR to promote the binding of Oct-2 to the MMTV promoter suggested that the octamer motifs might be of minimal importance for the induction of MMTV transcription by MR. To examine this possibility, we compared the effect of mutating the octamer motifs on the GR and MR responsiveness of the MMTV promoter proximal regulatory region linked to the adenovirus E1B minimal promoter (Fig. 7A). In these experiments GR and MR were expressed from the same vector transfected in equal amounts.
In the absence of hormone or cotransfected receptor, mutation of the octamer motifs decreased hormone-independent MMTV transcription activated by endogenous octamer factors approximately 2-fold (lanes 1 and 2). In the presence of the octamer motifs GR induced transcription 25-30-fold (lanes 1  and 3). Strikingly, mutation of the MMTV octamer motifs almost completely eliminated the transcriptional response to GR in CHO cells (lanes 2 and 4). This was consistent with the proposal that the GR-mediated induction of transcription from the MMTV HRE was almost entirely dependent upon the adjacent octamer motifs (15,16). This was not a cell type-specific effect, as others have reported similar results with the MMTV LTR in HeLa cells (15), and we have obtained similar results in Cos7 cells. 3 This was also not a specific property of the rat GR as a similar result was obtained with human GR. 3 Somewhat unexpectedly, despite being unable to recruit octamer factors to the MMTV promoter, MR induced expression of the WT reporter gene almost as strongly as GR in CHO cells (20- fold, lanes 1 and 5). Moreover, for MR, mutation of the octamer motifs actually led to an increase in fold induction by hormone to approximately 50 times the level obtained in the absence of MR from the same template (lanes 2 and 6). Thus, not only did MR not interact synergistically with endogenous factors on the MMTV promoter proximal regulatory region, but the MMTV octamer motifs appear to be somewhat detrimental to the full transcriptional response by MR. Again this effect was neither cell type-specific nor restricted to the rat receptor as similar effects on MR-induced transcription were observed in Cos7 cells transfected with either rat or human MR. 3 Such a pronounced difference in transcriptional activation by GR and MR has not been observed previously. Moreover, our results also suggested that the steroid receptor-binding sites in the MMTV HRE are configured in a manner that render GR responsiveness almost entirely dependent upon the adjacent octamer motifs.
The difference in the transactivation potential of GR and MR appeared to be specific for, or dependent on, the MMTV HRE, as the two receptors activated transcription similarly from a reporter gene in which two copies of a consensus GR/MR response element had been cloned adjacent to a herpes simplex virus thymidine kinase promoter (Fig. 7B, lanes 1 and 3). By contrast neither receptor activated transcription from two estrogen response elements cloned in the same configuration (lanes 2 and 4). These results indicate that the difference in the induction of transcription by GR and MR from the MMTV promoter proximal regulatory region in the absence of octamer factor-binding sites reflects some inherent property of the HRE rather than a difference in the activation potential of the two receptors.

DISCUSSION
The steroid receptors for glucocorticoids, mineralocorticoids, progestins, and androgens recognize the same DNA sequence elements in the cell (3), yet specific gene transcription is known in at least some cases to be selectively responsive to particular  1, 3, and 5), the other identical but with mutations inactivating the octamer motifs (lanes 2, 4, and 6). CAT activity from these cells was compared for their responsiveness to dex-stimulated GR (lanes 3 and 4) and aldosterone-stimulated MR (lanes 5 and 6) or activity from cells lacking exogenous receptor ( lanes  1 and 2). B, CAT activity derived from cells transfected with a TK-CAT reporter gene containing either two copies of a perfect HRE palindrome (lanes 1 and 3) or two copies of a consensus estrogen response element (lanes 2 and 4) in the presence of either GR (lanes 1 and 2) or MR (lanes 3 and 4) treated with dex or aldosterone.
hormones (27). Moreover, GR and MR differentially transmit transcriptional responses to corticosteroids in different tissues (36), despite both receptors displaying high affinities for glucocorticoids and aldosterone (4). The differential responsiveness of GR and MR is mediated at several levels. First, the two receptors are differentially expressed, particularly in the brain (36). Second, 11-hydroxysteroid dehydrogenase, which selectively metabolizes cortisol, but not aldosterone, also is differentially expressed (37). Third, differences in the GR and MR ligand binding domains and their completely divergent N termini are proposed to confer distinct transcriptional regulatory properties that are also expected to influence and even determine the specificity of transcriptional responses (27,38).
Our results demonstrate that the apparent inability of MR to bind to the POU domains of Oct-1 and Oct-2, and to promote the binding of Oct-2 to octamer motifs in the MMTV promoter proximal regulatory region, correlates directly with an absence of synergism between MR and octamer factors in transfection experiments. Interestingly, at the same time, our results also reveal a compensatory mechanism that appears to be encoded within the MMTV HRE that results in similar total hormonal responsiveness to GR and MR, at least in CHO cells. Thus in the context of the MMTV HRE, GR was only a weak activator of transcription in the absence of octamer motifs. By contrast GR and MR activated transcription similarly from synthetic perfectly palindromic response elements.
The structure and sequence of nuclear receptor DBDs have been highly conserved, and a number of nuclear hormone receptors have been demonstrated to interact synergistically with POU transcription factors (15, 17, 39 -41). Thus, it is perhaps not surprising that the DBDs of several nuclear receptors in addition to GR were able to bind to the POU domains of Oct-1/2 in GST pull-down assays. However, under the more stringent binding conditions of mammalian one-hybrid and immunoprecipitation experiments, only three receptor DBDs, those of GR, PR, and AR, interacted with the POU domain in a manner that was readily detectable in the cell. For at least one receptor, ER␣, our results are consistent with a previous report on the nature of its transcriptional synergism with the POU factor Pit-1 (42). In this study, whereas a weak interaction was noted between the ER␣ DBD and the Pit-1 POU domain, transcriptional synergism was entirely dependent upon a functional interaction between the N terminus of Pit-1 and the estrogen receptor LBD. Our results also highlight the importance of confirming in vitro binding results with bacterially expressed and in vitro translated proteins with more stringent assays that more directly reflect the potential for binding in vivo.
On the MMTV LTR, the promoter proximal octamer motifs are required for over 90% of the transcriptional responsiveness to glucocorticoids and progestins (15). This correlates directly with the promotion of the binding of Oct-2 to DNA that we observed on transiently transfected DNA templates. Furthermore, this effect appears to be directly dependent, at least in part, upon the binding of the GR or PR DBDs to the POU domain of the octamer factors. Our results also are consistent with in vitro studies that demonstrated that the addition of GR or PR to a binding assay decreased the concentration of Oct-1 required to saturate the MMTV octamer motifs (15). By comparison, the inability of MR to interact with Oct-2 and to promote its binding to octamer motifs adjacent to MR-binding sites on both the MMTV promoter and synthetic templates was exactly consistent with the inability of MR to interact synergistically with octamer factors to activate transcription from the MMTV promoter proximal regulatory region.
Interestingly, however, the inability of MR binding to the octamer factors had little effect on the induction of MMTV transcription in response to stimulation of MR. Indeed, a mild reduction in the fold activation of transcription by MR was observed in the presence of the octamer motifs. By contrast, mutation of the octamer motifs, which eliminated the contribution of the octamer factors toward MMTV transcription (15,16), almost completely eliminated the induction of transcription by GR. Thus on the MMTV HRE, GR activated transcription 30 times less efficiently than MR in the absence of octamer motifs. This low transcriptional activation potential of GR was specific for the MMTV HRE, as both receptors induced transcription similarly from a promoter containing two copies of a perfect HRE palindrome. Our results with the synthetic HRE are also consistent with those of other groups that have reported a similar transcriptional regulatory potential for GR and MR (4,27,43). These results highlight that although GR and MR have similar transcriptional activation properties in many instances, the precise transcriptional activation potential of receptors may be allosterically regulated by the nature of the HRE (44,45). One alternative possibility in our experiments was that on the MMTV promoter proximal regulatory region, MR interacted more productively with NF1 than GR. It has been observed in vitro on chromatin-free DNA templates that GR and NF1 activate MMTV transcription independently and actually can compete for binding to the MMTV promoter in a way that is not observed on organized chromatin templates in vivo (14,46). Whether MR interacts differently with NF1 is not known. However, our footprinting of the promoter of the transiently transfected MMTV LTR construct indicates that occupancy of the MMTV NF1-binding site was not noticeably affected by steroid-induced MR or GR. Thus a preferential interaction between MR and NF1 would most likely occur subsequent to the binding of the two factors to the MMTV LTR.
In addition to GR-Oct-1/2 binding (16,47), there are several additional examples of the GR DBD entering into functional complexes with other transcription factors. Two of these interactions, with AP-1 and NFB, are conserved properties of several nuclear receptors (48 -53). All three appear to be mediated mainly through the second zinc finger of the receptor DBD. Although a direct comparison remains to be made, and AP-1 is known not to interact with MR (54), it appears that each interaction is mediated through determinants in the GR DBD that are differentially conserved in other receptors. For example, whereas receptor-Oct-1/2 binding was found to be restricted to GR, AR, and PR, NFB can also interact with ER␣ (55), and AP-1 can interact with RAR and TR (49,50). In addition, whereas rat GR truncated from the C terminus to amino acid 509 can still interact fully with AP-1 (56), the same truncation of GR reduces Oct-1/2 binding in a GST pull-down assay by over 80%. 3 Alignment of the rat GR and MR DBDs shows that there are just two non-conservative differences between the two receptors to the end of the second zinc finger, at Arg-496 and Tyr-497 of GR. Interestingly, substitution of these two amino acids interferes with GR binding to AP-1 (54), whereas mutations C500Y and L501P interfere with the binding of GR to Oct-1 and Oct-2. It will be interesting to compare directly the effect of the R496L and Y497Q for GR binding to Oct-1 and Oct-2. However, because truncation of GR from 525 to 509 also affects Oct-1 and Oct-2 binding in vitro, it will be important also to consider the several amino acid differences in this region that occur between GR and MR. Indeed, preliminary results suggest that determinants within these hinge regions of GR and MR may be crucial for distinguishing the binding of GR and MR to Oct-1 and Oct-2. Identification of a localized MR/GR substitution that eliminates the direct binding of GR to Oct-1 and Oct-2 may be expected to allow a detailed examination of the contribution of GR-Oct-1 and Oct-2 binding to transcriptional synergism between GR and Oct-1/2 in vivo.