INTRODUCTION

The classic model of steroid hormone action suggests that ligand-bound receptors dimerize and bind to specific sites (acceptor sites) on the chromatin/DNA to alter gene expression. In many genes, the steroid receptors (SR)¹ bind to specific cis-acting DNA sequences, referred to as hormone response elements (for reviews, see Refs. 1-3). These elements function as enhancer-like DNA sequences and are required for the steroid hormone regulation of transcription. However, not all steroid-responsive genes contain these elements, and several other pathways and participating nuclear proteins have now been identified. The SR regulation of gene transcription has now been shown to involve 1) interactions of the SR with transcription factors, such as the AP-1 complex (for review, see Refs. 2-4); 2) interactions of SR with accessory factors, which appear to function by assisting the binding of the receptor to the steroid response elements, or to transcription factors (for reviews see Refs. 4-6 and references therein); and finally, 3) interactions of SR with specific chromatin structures including the nuclear matrix (see Refs. 7-15 and references therein). In the latter, a class of high affinity binding sites (acceptor sites) for SRs in the nuclei, chromatin, and nuclear matrix DNA-binding proteins (e.g. acceptor proteins) has been reported by many laboratories for most of the steroid hormone receptor systems (for reviews, see Refs. 12, 15, and 16 and references therein).

Studies in many laboratories using a variety of steroid systems have localized the nuclear binding (acceptor) sites for SR to the nuclear matrix. These studies include estrogens and androgens (10, 17-20), mineralocorticoid (21), and progesterone (22, 23). Previous studies in our laboratory revealed that chromatin acceptor sites for the avian oviduct progesterone receptor (PR) were associated with the nuclear matrix as was the receptor-binding factor (RBF) (23-25). In short, the chromatin and nuclear matrix acceptor sites appear to be one and the same.

We have purified the nuclear matrix RBF (formerly termed a chromatin "acceptor" protein) based on its ability to generate specific, high affinity PR binding when renatured and bound to the avian genomic DNA (12, 22, 23, 26). Using immunohistochemical techniques with antibodies against the purified RBF, the RBF was localized to the nuclei of selected cell types in certain avian and rat tissues (27, 28). In similar studies, co-localizations of RBF and PR in the same cell types in the avian oviduct and rat reproductive organs were also shown. The full-length cDNA to RBF has been cloned and used to identify a 0.7-kilobase mRNA in specific avian tissues. The mRNA levels in these same tissues correlate with the RBF protein levels by Western blot analyses (28). A homologous RBF-like mRNA species was identified in human tissues.²

Earlier studies had indicated that specific DNA sequences were important when generating RBF/PR-binding sites. The PR-binding sites could only be generated when the purified RBF was bound to mammalian genomic DNA, but not when the RBF was complexed to prokaryotic, or lower eukaryotic genomic DNAs (29, 30). The PR-binding sites could also be generated with the genomic sequences of the c-myc gene but not those of the ovalbumin gene (31). Past studies in our laboratory demonstrated that the steady state levels of mRNA and the transcription rates of several nuclear proto-oncogenes, c-myc, c-fos, and c-jun, in the avian oviduct are rapidly regulated in vivo by Pg, glucocorticoids, and estrogen (24, 32-36). These nuclear proto-oncogenes have also been shown to be regulated by steroids in many other animal systems (37-43). In addition, Southern blot analyses revealed that the DNase-resistant nuclear matrix DNA had sequence homology with sequences in the 5-ends of the steroid-regulated c-myc and the c-jun proto-oncogenes but not to genomic sequences of some other genes in the avian oviduct, suggesting that nuclear matrix is attached to 5-flanking regions of those genes (23). Thus, a model is emerging, whereby the RBF-containing nuclear matrix is associated with c-myc and other proto-oncogene promoters and generates PR-binding sites (acceptor sites) through which steroids might regulate the transcription of these genes. It should be noted here that other nuclear matrix proteins have been reported to bind to promoter domains and alter gene expression (44, 45).

To challenge this model, it was necessary to determine whether RBF binds to a specific element in or near the same nuclear matrix attachment site in the c-myc 5-flanking domain. This paper demonstrates that the RBF does bind to a specific DNA element within this nuclear matrix attachment site in the 5-flanking domain of the avian c-myc gene. Further, the nuclear matrix is shown to attach to each side of this RBF-binding element. Primary sequence analyses of the RBF reveals structural motifs resembling other DNA-binding proteins. Preliminary studies show that the overexpression of RBF in transiently transfected human breast cancer cells inhibits reporter gene activity when the reporter gene construct contains the 1021-bp region of the c-myc promoter. The treatment of these cells with steroids ensures this inhibition. Thus, the RBF and steroid hormones may regulate c-myc gene promoter activity via this 1021-bp region of the promoter that contains the RBF element and nuclear matrix attachment sites.

EXPERIMENTAL PROCEDURES

Hen oviduct was excised and the chromatin isolated. Chromosomal proteins were size-fractionated by molecular sieve chromatography as described previously (46). Fractions containing proteins in the 4-20-kDa size range were pooled and separated by preparative SDS-polyacrylamide gel electrophoresis in a Tris-Tricine buffer as described previously (28). The gel was transferred to six sheets of polyvinylidene difluoride membrane using a CAPS buffer system (10 mM CAPS + 10% methanol, pH 11.0). The sixth sheet was immunostained for RBF utilizing a RBF-specific monoclonal antibody (22, 46, 47). The five additional sheets were matched against the one with the visible immunoproduct. Bands encompassing RBF were excised, and the protein was eluted from the membrane in 70% (v/v) 2-propanol, 5% trifluoroacetic acid (v/v) overnight; dialyzed against 5% (v/v) acetonitrile/H₂O, 0.25 mM PMSF, and lyophilized. This protocol was tested and reported in detail (22, 28).

This method has been described in detail previously (22, 26, 48). Briefly, purified RBF was solubilized in 6 M guanidine-HCl in reconstitution buffer (10 mM Tris, 1 mM EDTA, 1 mM dithiothreitol, 0.2 mM PMSF) to a concentration of 250 ng/ml. This concentration of the RBF was found to be important for optimal protein renaturation and reconstitution of PR-binding sites on the DNA (26, 30, 48). It was also found in these current studies to be important for achieving the optimal DNA sequence-specific binding. The solution was loaded into 10-mm diameter dialysis tubing and dialyzed in a plastic cylinder filled with 200 ml of 6.0 M guanidine-HCl against 1 liter of reconstitution buffer at a replacement rate of 1 ml/min to a guanidine concentration of 0 molarity, while rocking at 4 °C. After 16 h, the protein solution was removed from the dialysis bags, homogenized, and used in the Southwestern blotting analysis.

Synthetic DNA corresponding to regions of the 5-c-myc gene were synthesized using PCR. However, the 1021-bp E fragment (see Fig. 1) was obtained by restriction digests of the c-myc plasmid with SmaI. The nuclear matrix DNA was isolated, with modifications, as described in Schuchard et al. (23). Briefly, nuclei, isolated from hen oviduct, were resuspended in buffer A (10 mM Tris-HCl, 0.2 mM MgCl₂, 0.1 mM PMSF, pH 7.5) and then centrifuged. The pellet was resuspended again in a buffer A and recentrifuged. The resulting pellet was resuspended in buffer B (10 mM Tris-HCl, 0.2 mM MgCl₂, 0.1 mM PMSF, and 2 M NaCl, pH 7.5) and centrifuged. After a second resuspension and centrifugation, the pellet was resuspended in TM buffer (50 mM Tris-HCl, 5 mM MgCl₂, 0.25 M sucrose, pH 7.5). This solution was subjected to extensive DNase I and RNase digestion. SDS and proteinase K were then added to 0.5% (w/v) and 50 mg/ml, respectively. The fractions were phenol-extracted and ethanol-precipitated. Matrix DNA was then gel-purified and fragments of 100-200 base pairs, representing the bulk of the residual/protected DNA, were excised. The DNA was dephosphorylated and the 5-end labeled with [-³²P]ATP.

The avian c-myc genomic clone was used as a template in standard PCR reactions. Primers were designed to yield DNA fragments representing different regions of the 5-upstream flanking domain of the c-myc gene. DNA was gel-purified and end-labeled with [-³²P]ATP by standard T4 kinase reactions. For the Southwestern blot analyses, the purified, reconstituted RBF (31 or 62 ng/slot) was diluted in reconstitution buffer and then slot blotted onto nitrocellulose under vacuum, using a slot blot apparatus. The RBF-MBP fusion protein was also slotted in some cases. The membrane was incubated with end-labeled DNA probes representative of nuclear matrix DNA or c-myc fragments of the upstream regions of the c-myc gene. Colipase C (50 ng; Sigma), a 10-kDa protein with a pI of 7.0, was also slot blotted and used as a negative control. In some cases, histones (histone type II-AS calf thymus; Sigma H7755) were blotted and used as a nonspecific DNA-binding protein. The nitrocellulose was cut into strips to be incubated with different probes. Blots were blocked for 1.5 h in standard Bowens buffer (50 mM Tris-Cl, pH 7.0, 0.1% bovine serum albumin, 0.1% Ficoll, 0.1% polyvinylpyrollidine), then probed with 3-5 × 10⁶ cpm end-labeled DNA probes, including 50 mM KCl and 1 mM CaCl₂ (23). The blots were washed with Bowens binding buffer and exposed to autoradiographic film.

Southern blot analysis using DNA/nuclear matrix was conducted according to the procedure described previously with some modifications (23). Equimolar amounts (10 pmol for 1021 and matrix; 20 pmol for I₁, I₂, H, J, K, I₆, I₈, and I₁₀, corresponding to upstream regions of the avian c-myc gene, as described previously above), were diluted in SSC buffer (0.15 M NaCl, 0.015 M sodium citrate, pH 7.0) were heat- and alkali-denatured and slot blotted onto nitrocellulose. Blots were baked, prehybridized in Denhardt's medium, then hybridized with the end-labeled nuclear matrix DNA, and washed and exposed to x-ray film.

The coding region of the RBF gene was cloned, in frame, into the pMAL-c2 vector and expressed as MBP as specified by the brochure "Protein Fusion and Purification System" accompanying the kit provided by New England BioLabs, Beverly, MA. After the sequence of the construct was verified, the vector was used to transform competent Escherichia coli whereby large scale amounts of the fusion protein were produced and purified by affinity chromatography as described in the kit brochure. The fusion protein migrated in SDS-polyacrylamide gel electrophoresis analysis as a 55-kDa protein. A new polyclonal antibody (pAb) was then prepared against the RBF-MBP fusion protein. Briefly, 1 mg od purified lyophilized protein was sent to Cocalico Biologicals, Inc. (Reamstown, PA), where two rabbits were inoculated and serums recovered. The pAb was then purified from the serum using the Affi-Gel Protein A MAPS II kit according to a brochure provided by Bio-Rad. Western blots showed that this polyclonal antibody (pAb 273) reacted against the RBF-MBP, as well as the purified reconstituted and factor Xa-cleaved RBF proteins.

Electrophoretic mobility gel shift assays (EMSA) were conducted, with modification to the original procedure, reported by Fried and Crothers (49-51). Briefly, 200-500 ng of RBF-MBP in binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1.0 mM dithiothreitol, 2.0 mM MgCl₂, 10% (v/v) glycerol) was incubated with 0.5 µg of poly(dI-dC) for 15 min at room temperature. The DNA probes were synthesized and labeled as above (500,000 cpm each) and added to the protein solution, which was incubated for 20 min at room temperature. The mixture was rocked with the RBF-MBP polyclonal antibody for 2 h at 4 °C. Protein-DNA complexes were resolved by gel electrophoresis on 6% (w/v) native acrylamide gels (75:1 acrylamide:bisacrylamide) at 26 V/cm in 0.5 × TBE buffer (0.05 M Tris borate, 0.001 M EDTA, pH 8.0) for 2.5 h at 4 °C. Gels were then dried and exposed to autoradiographic film.

The N-terminal amino acid sequence analysis of the RBF protein was determined by the Mayo Protein Core Facility (Mayo Clinic, Rochester, MN). Briefly, purified RBF protein (500 pmol) or its tryptic peptides were applied to a Polybrene-treated glass filter disk, and then microsequenced by the Edman degradation method on an Applied Biosystems (ABI) 476A Protein Sequencer using programmed cycles provided by ABI (Foster City, CA). The phenylthiohydantoin amino acids from each cycle were identified on-line by reverse phase high performance liquid chromatography on an ABI 120A PTH Analyzer.

The cDNA for RBF was obtained using polymerase chain reaction procedures, as described previously (28). Briefly, poly (A)⁺ mRNA from chicken oviduct was reverse transcribed using an oligo(dT) primer to obtain the first strand cDNA. The cDNA was used as a template in the PCR along with sense primer (oligonucleotides synthesized from the N-terminal sequence of RBF) and antisense primer (dT). One of the PCR primers also contained a T7 or SP6 phage promoter. The PCR-amplified DNA was transcribed and sequenced using RNA amplification with transcript sequencing (RAWTS) method described previously by Sarkar and Sommer (52). The cDNA sequence correlated with the amino acid sequence. Preliminary analysis of the primary and secondary structure of the RBF protein was performed by computer analysis using peptide structure and helical wheel algorithms contained in the sequence analysis software package, produced by the Genetics Computer Group (GCG) of the University of Wisconsin, Biotechnology Center, Madison, WI.

An RBF open reading frame was cloned into an expression vector (pSG5; Stratagene, La Jolla, CA), which is regulated by SV40 enhancer element. In all experiments, the pSG5 expression vector itself was transfected as a control to determine the effects of the transfected plasmid alone. The reporter construct consisted of the c-myc gene E fragment (1021 bp) (see Fig. 1), cloned into the pGL2 basic (Promega) luciferase vector. The RBF construct was cotransfected along with the 1021 pGL2 reporter construct. A tk-CAT reporter construct was included in all transfections to access and normalize for variations in the transfection efficiency. The MCF-7 cells were cultured using Dulbecco's minimal essential medium (DMEM) containing 1 × antibiotic/antimycotic solution (100 µg of penicillin, 0.1 µg of streptomycin, and 0.25 µg of amphotericin B/ml (Sigma)), 10% (v/v) fetal bovine serum (Sigma), and 10⁸ M 17-estradiol (E₂) (Sigma). The E₂ was included to maintain optimal cell proliferation and high PR levels.

For the transfections, the MCF-7 cells were trypsinized, plated into 100-mm dishes, and grown for ~48 h to approximately 50% confluence in the above described (DMEM) medium containing charcoal-stripped serum containing 10⁸ M E₂. The charcoal stripping of the serum was performed to remove endogenous steroids. The cells were then co-transfected with the 1021 pGL2 reporter gene construct and either the pSG5-RBF expression construct or the pSG5 vector alone, using the LipofectAMINE reagent according to the procedure provided by Life Technologies, Inc. The assays contained experimentally optimized quantities of DNA (represented per plate as: 10 µg of pSG5 and/or RBF expression vectors along with 2 µg of the 1021 pGL2 reporter gene construct and 2.4 µg of tk-CAT). The various DNAs and LipofectAMINE were added individually to serum-free, antibiotic-free media, then mixed together and placed on the cells that had been washed twice in serum-free, antibiotic-free media. The optimal transfection incubation period was found to be 4 h at 37 °C. The transfection mixtures were then removed by aspiration and replaced with DMEM containing charcoal-stripped serum and 10⁸ M E₂. After 24 h, the transfected cells were then cultured in the same media containing 10⁸ M of both Pg (Sigma) and E₂. The 10⁸ M estrogen was maintained in all media to maintain high levels of PR. The Pg was added to generate functional/activated PR.

The transfected cells were then harvested into a solution containing 40 mM Tris, 150 mM NaCl, 1 mM EDTA, pH 7.4, and sedimented at 14,000 rpm for 30 s. The cell pellets were resuspended in 1 × reporter lysis buffer (according to the brochure on luciferase assay systems, Promega, Madison, WI), and the amount of protein calculated by Bradford analysis. Then 20 µg of the total protein was used for each luciferase assay and 50 µg of total protein was used in CAT assays. Relative levels of luciferase activity were assessed according to the brochure provided by Promega, Madison, WI, for the Promega luciferase assay system using a luminometer (Turner Luminometer Model 20, Sunnyvale, CA). These data were expressed in relative light units normalized against the amount of CAT activity per plate of cells. The CAT enzyme assays were conducted according to the method of Seed and Sheen (53), using the same extracts as for the luciferase assay. Briefly, equal amounts of cellular extracts (50 µg of protein each) were incubated with 0.1 µCi of [¹⁴C]chloramphenicol, 25 µg of n-butyl coenzyme A in a 37 °C water bath for 2-4 h. Following incubation, the reaction is terminated by adding 300 µl of 100% v/v xylene. The xylene phase was extracted two more times, each time with 100 µl of 0.25 M Tris (pH 8.0). Then 200 µl of the xylene phase was removed, mixed with 5.0 ml of scintillation fluid, and then counted in a liquid scintillation counter.

RESULTS

Previous data had shown that nuclear matrix DNA, which was protected from DNase and RNase digestion, hybridized to a specific 1021-bp region (the E domain) of the avian c-myc gene promoter (676 to +345 bp), supporting the presence of a nuclear matrix attachment region (MAR) in this 5-flanking domain (23). Since RBF has been identified as a nuclear matrix-associated protein, it was of interest to determine if the RBF protein itself would bind directly to the same E domain containing the MAR-like region of the c-myc promoter. As described under "Experimental Procedures" and outlined in Fig. 1, Southwestern blotting techniques were used whereby purified, renatured RBF protein was slot blotted onto nitrocellulose and probed in the early experiments with radiolabeled restriction enzyme fragments representing various stretches of the c-myc gene. In subsequent studies, the blots were probed with PCR-generated DNAs representing smaller domains of the c-myc promoter. Optimal assay conditions for the Southwestern blot technique were first determined using the early generated (A-E) fragments of the c-myc gene.

The overall structure of the avian c-myc gene and the locations and sizes of the restriction enzyme-generated fragments are outlined in Fig. 1. As described in panel A, the PstI restriction fragments of the c-myc gene generated the A, B, and C fragments. The A fragment was subdivided by restriction enzymes into regions D, E, F, and G. The 1021-bp E fragment was in turn, subdivided into H through K regions. Finally, as shown in panel B, the smaller DNA fragments representing various regions of the I fragment were generated by PCR. Early studies using Southwestern blot analyses have revealed that RBF specifically bound to the A fragment and subsequently to the 1021-bp E fragment (676 to +345 bp) with no binding to the B, C, D, F, or G fragment of the gene (data not shown). More recent results, displayed in the inset portion of Fig. 1, indicate that RBF bound with the highest avidity to the 331-bp I fragment at 583 to 252 bp, with some binding to the 447-bp fragment K at 102 to +345 bp from the first exon. No binding was found with the H and J fragments. The specificity of the interactions between RBF and the DNA is supported by the failure of BSA to bind DNA, while, conversely, the basic histone proteins bound indiscriminately to all the DNA fragments. Subsequent studies focused on the I fragment since it displayed a greater level of RBF binding than the K fragment under the experimental conditions.

Southwestern blot analyses of RBF binding to the internal regions of the I fragment is shown in Fig. 2. Panel B of Fig. 1 gives the orientation and locales of the corresponding PCR fragments generated from the 331-bp I fragment of the c-myc gene promoter region. As shown in Fig. 2A, the RBF does not bind to the I₁ or the I₂ fragments. The RBF, however, does bind to the 152-bp I₃ fragments and to the successively shorter I₇ and I₈ fragments. The RBF protein did not bind to the other regions within the I fragment (data not shown). The specificity of the RBF binding is further supported by the fact that RBF binds optimally to I₈ fragment, while it does not bind to a 125-bp fragment of the human retinoblastoma gene promoter region which, by computer analysis, was shown to have 70% homology with the I₈ fragment. Fig. 2B shows additional Southwestern blot analyses demonstrating that the 54-bp I₁₀ fragment is the shortest fragment that binds the RBF. Deletions on either side of this 54-bp I₁₀ fragment (I₉ and I₁₁) result in a complete loss of RBF binding. The 10-kDa colipase C served as a negative control in these experiments. The sequence of the 54-bp I₁₀ (404 to 340) fragment of the c-myc gene that binds RBF is shown in Fig. 3. This fragment contains a 5-GC-rich region and a 3-AT-rich region. A GC-rich sequence that is located on the 3-end in the larger I₈ fragment was shown to be nonessential for RBF binding, since its deletion to achieve the I₁₀ fragment has no apparent effect on RBF binding. However, the 5-GC-rich region of I₁₀ is essential for RBF binding since its deletion to achieve the I₉ fragment results in the loss of RBF binding.

To demonstrate the specificity of the RBF bound to the I₁₀ element, electrophoretic mobility shift assays (EMSA) were performed. However, a larger protein complex containing RBF was needed to perform the EMSA. Therefore, a RBF-MBP fusion protein was constructed, expressed, and purified from E. coli. The purified RBF-MBP protein showed the same DNA sequence specificity in the Southwestern blots, as the native, reconstituted oviduct RBF. Fig. 4 shows that RBF-MBP fusion protein binds to the same c-myc regions as the native reconstituted RBF, binding to the I₈ and I₁₀ fragments, but not to the I₉ and I₁₁ fragments. The fusion protein and c-myc fragments used in Southwestern analyses were then subjected to EMSA. Fig. 5 shows that when the RBF-MBP fusion protein and the anti-RBF polyclonal antibody (pAb 273) complex are bound to the I₈ or I₁₀ c-myc DNA fragments, the protein DNA complex is supershifted. The specificity of the RBF-MBP/DNA binding is shown by the fact that a nonspecific antibody fails to cause a supershift and the MBP protein or antibody alone do not bind to the c-myc fragments (Fig. 5A). The specificity of RBF-MBP/DNA binding is further confirmed by the absence of RBF-MBP binding to the I₉ and I₁₁ probes, which do not bind the native reconstituted RBF in Southwestern blot analyses (Fig. 5B). Moreover, the specific DNA complexes that are visible with RBF-MBP/anti-RBF antibody were absent in reactions wherein the DNA/RBF-MBP/pAb complexes are incubated with protein-G beads followed by centrifugation to remove the beads and any bound Ab complexes before being resolved by the gel. These EMSA results demonstrate that the RBF-MBP protein specifically binds the c-myc gene I₁₀ fragment as does the native reconstituted RBF using Southwestern blot analysis.

Since the RBF protein was found to be associated with the nuclear matrix, and the latter was shown to be attached to the same region (E fragment, which contains the RBF I₁₀ element) of the c-myc gene (23), the question arose as to whether the RBF binds to the same sequences as does the nuclear matrix. To answer this question, the same fragments of the c-myc gene 5-flanking region were examined for homology with the nuclear matrix DNA. Southern blots of the c-myc DNA fragments were performed using ³²P-end-labeled purified nuclear matrix DNA as a probe. As shown in Fig. 6, the fragments and, of course, the nuclear matrix DNA itself, displayed sequence homology to the nuclear matrix DNA fragment by Southern blotting, while the fragments which bind the RBF, I₈ and I₁₀, display little or no homology to the matrix DNA. The c-myc gene K region which RBF shows weak binding (Fig. 2), also showed no homology to matrix DNA. The fact that the nuclear matrix is bound to the I₁ and to the I₆/J regions, but not the I_8/10 fragments, suggests a dual attachment of the nuclear matrix, which flanks the RBF element located in the I_8/10 region.

To further test whether or not the RBF DNA element and the nuclear matrix DNA are distinct, Southwestern blot analyses for RBF binding were performed using ³²P-labeled I₃ and I₈ fragments, which bind RBF; two fragments, I₉ and H, which do not bind RBF, as well as the nuclear matrix DNA itself. The results in Fig. 7 again show that RBF does bind the avian c-myc, I₃ and I₈ fragments, but not the I₉ and H fragments, as shown earlier. Importantly, the RBF fails to bind the nuclear matrix DNA. These results support that the RBF-binding element, the specific I₁₀ region in the avian c-myc gene promoter, is distinct from the nuclear matrix-associated DNA. In fact, the sequences of the nuclear matrix attachment sites in the I₁ and the I₆/J regions flank the I₁₀ region containing the RBF element.

Since RBF appears to be a specific DNA-binding protein, the possible DNA-binding motifs of the RBF protein were then considered. Fig. 8 shows the primary sequence of RBF as determined by the nucleotide sequence of the cDNA and by direct amino acid sequencing of the RBF and its peptide fragments. Peptide structural analyses predicts an N-terminal domain with parallel and anti-parallel -sheet structure. The C-terminal region contains uninterrupted amphipathic leucine/isoleucine-rich domain, reminiscent of a leucine zipper domain for homo- and hetero-dimer interactions. One possible structure derived from this preliminary structural analysis suggests a homodimer complex at the C-terminal region, with a dual -sheet DNA-binding motif at the N-terminal region. As discussed later, the -sheet is a reported DNA-binding motif structure for transcription factors, repressors, and single-stranded binding proteins. Preliminary studies were performed to access whether the RBF protein can actually interact with the RBF-binding element to regulate transcription of the c-myc gene. The MCF-7 cells were plated into media containing charcoal-stripped serum containing 10⁸ M E₂ for 24 h, followed by treatment of both E₂ and Pg (10⁸ M) for an additional 24 h. These are optimal steroid concentrations for the generation of functional/activated progesterone receptors in these cells. The RBF expression construct, or only the control parent vector, was co-transfected into MCF-7 cells together with a reporter gene construct consisting of 1021 bp (E region) of the c-myc gene promoter cloned into the pGL2 basic luciferase reporter gene (Promega). As shown in Fig. 9, a 3-4-fold inhibition occurred only in the cells with the RBF expression construct. This inhibition required the presence of both E₂ and Pg. A nonsignificant and variable inhibition of the c-myc promoter was observed in the absence of the steroids. In any case, these results suggest that RBF protein can interact and inhibit the c-myc gene promoter activity using the domain containing the RBF element in the presence of activated PR.

8

8

Fig. 10 presents a model of the nuclear matrix attachment site with the intervening RBF acceptor site based on the data presented in this paper. The RBF/DNA complex, at the nuclear "acceptor site," 404 to 350 bp, is shown to be flanked by nuclear matrix attachment sites at 582 to 423 and at 310 to 252 bp. The nuclear matrix attachment extends into the J region (253 to 102 bp). The neighboring K fragment, which binds RBF but not the nuclear matrix, is also shown. Thus, while the RBF was reported to be associated with the isolated nuclear matrix (23), the protein appears to actually be bound to an intervening DNA region, as depicted in the figure, while the two nuclear matrix attachment sites flank this intervening element. Thus, the RBF acceptor site is made available by a dual nuclear matrix attachment. Interactions between steroid receptors and/or other factors with this RBF/DNA complex and may utilize the flanking nuclear matrix attachment sites to subsequently regulate the transcription of the c-myc gene expression.

DISCUSSION

Past studies have suggested an essential role for specific nuclear chromatin proteins in forming high affinity binding sites (acceptor sites) in the nuclei/chromatin of target tissues for a variety of steroid receptors (22, 23, 26, 54-66). In our laboratory, a specific "acceptor protein," RBF, was isolated based on its ability to generate saturable, high affinity binding sites on avian and mammalian genomic DNAs and characterized as a component of the chromatin nuclear matrix acceptor sites for the avian oviduct PR (22, 23, 25, 26, 29, 30, 67, 68). The RBF failed to generate these sites when using insect and prokaryotic genomic DNA.

The data presented in this paper address several aspects of the nuclear matrix and RBF, their structure, the steroid receptor interactions, and transcriptional regulation of the c-myc gene. First, the nuclear matrix-associated "RBF" binds to a unique, specific, DNA element within a MAR-like region in the 5-flanking domain of the c-myc proto-oncogene. Second, the nuclear matrix attachment sites are distinct from the specific DNA element for RBF, and actually flank this element. Third, the complete primary sequence of RBF is presented and shown to be unique with a structure similar to certain other DNA-binding proteins. Fourth, preliminary studies indicate that transient co-transfections of the RBF expression construct, together with c-myc promoter-reporter gene construct, results in the inhibition of the reporter gene in the presence of steroids. The latter results mimic earlier in vivo studies whereby the endogenous avian c-myc gene expression is rapidly (within 30 min of injection) inhibited by Pg and glucocorticoids (32, 33). In these studies, the combined treatments of estrogen and progesterone were found to be essential for this down-regulation of c-myc promoter. We speculate that the presence of the steroid generates functional PR required for the inhibition of the promoter activity wherein the estrogen induces PR levels in the MCF-7 cells and the Pg activates the PR. Further studies are required, however, to prove that the steroids and their receptors play a direct role in this inhibition. The RBF has been shown previously to bind to the genomic c-myc gene and generate specific binding sites for PR (31). Based on these data, as well as those described in this paper, a model was generated whereby the PR binds to the RBF-DNA element flanked by the nuclear matrix attachment sites in the c-myc promoter, which causes an inhibition of the c-myc gene transcription.

The computer-generated structure of RBF protein was based on its primary amino acid sequence. The analysis suggests a C-terminal end, which generates a homodimer via its amphipathic leucine-isoleucine-rich region. A similar leucine-isoleucine zipper-like structure has been shown for the human single-stranded DNA-binding protein (70). The primary sequence also predicts an N-terminal region with parallel and anti-parallel -sheet stretches. These -sheets are known DNA-binding motifs for some prokaryote and eukaryote transcription activators and repressors (71-76). Studies are under way to determine if RBF binds to its element via this -sheet structure. In general, RBF was found to be unique with some homology to several other proteins such as bovine, rat, and mouse F1/F0ATPase subunit E (77-79) and a rapidly UV-induced nuclear protein mRNA in Chinese hamster ovary cells (80).

Evidence demonstrating the significance of the nuclear matrix in steroid hormone action, including the location of specific nuclear binding sites for steroid receptors, has been accumulating (7-10, 17, 18, 21, 67, 81-84). Previous studies have reported that certain nuclear matrix proteins are bound to MAR-like AT-rich DNA sequences, similar to the RBF element, as described in this paper (95, 96). The functions and properties of these and other nuclear matrix proteins are only now being elucidated. Estrogen has recently reported to significantly increase the levels of several nuclear matrix-intermediate filaments in steroid-responsive human breast cells but not nonresponsive cells (69). These effects may reflect a coordinated response of steroid action on the nuclear matrix. Our laboratory reported that the chromatin acceptor sites and the nuclear matrix binding sites for the avian PR were the same (12, 67). In fact, the isolated RBF protein and PR nuclear binding were localized solely to the oviduct nuclear matrix (22, 25). The nuclear matrix is an organized structure responsible for anchoring topological domains, called closed loop domains, in interphase nuclei and play a role in DNA replication, mRNA synthesis, DNA loop attachment sites, and topoisomerase activity (85) (for recent reviews see Refs. 14 and 86). The total genomic 10 nm chromatin filaments are subdivided into 50,000 loops, each containing an average of 60,000 base pairs of DNA. These loops are tightly anchored to the DNA at specific matrix attachment regions (MARs), also termed scaffold attachment regions (SARs) (87-89). These MARs/SARs are DNase-resistant regions of variable length and sequence with AT-rich stretches and some containing topoisomerase II cleavage sites (88, 90). Recently, evidence for two classes of MARs with matrix protein complexes bound to them has been reported. There is a "constitutive class," which permanently anchors the loops and appears to be involved in DNA replication, and a "regulatory class," which is bound to the flanking domains of genes to regulate transcription as cis-acting regulatory elements (15, 87, 91-94). Our results implicate the latter "regulatory MARS" as another class of nuclear acceptor (binding) sites for SRs which functions in the steroid regulation of the c-myc gene transcription. In support of this model, many species of SRs, as well as their high affinity nuclear binding sites, have been localized to the nuclear matrix in a variety of animal systems (5, 8-10, 17, 18, 20, 21, 23, 67, 83-85).

In this paper we also present data supporting the theory that the nuclear matrix-associated RBF (acceptor protein) for the avian oviduct PR binds between two matrix attachment domains in the 5-promoter domain of the progesterone-regulated c-myc proto-oncogene. Since the nuclear matrix DNA represents those regions of DNA that are protected from DNase digestion of genomic DNA, the domains flanking the RBF element appear to be the domains actually protected from DNase action and thus bound by the nuclear matrix. The actual nuclear matrix attachment sites in these flanking regions have been localized to I₁ (and the 3-end of H) and to I₆ (and the 5-end of J). The RBF binds to an AT-rich domain located in an intervening sequence between these two nuclear matrix attachment sites. This novel dual nuclear matrix attachment structure is supported by the fact that not only does the RBF fail to bind to the nuclear matrix DNA, but the RBF element I₁₀ fails to recognize any sequences in nuclear matrix DNA. The RBF-binding element has a 25-bp 5-GC-rich domain and a 29-base AT-rich 3-domain, the latter containing a 19-bp homopurine/homopyrimidine tract.

Thus, these flanking domain regions of the avian c-myc proto-oncogene do not have the typical MAR-like sequences, but rather the intervening RBF element in this structure has the previously reported MAR-like primary sequence of both GC- and AT-rich regions. The K domain of the c-myc promoter, which also binds the RBF in Southwestern blot assays (Fig. 1), contains two AT-rich regions with 10-bp and 15-bp homopurine/homopyrimidine (poly(A)/poly T) stretches. However, the nuclear matrix does not appear to bind to this region. Whether the PR binds to these sites of RBF binding is not currently known. The DNA structures generated by the MAR-binding proteins include DNA bending (97-98), minor groove narrowing in the poly(A) tracts (100), and duplex unwinding/single strand generation, which can form cruciform structures (101, 102). The DNA duplex unwinding was found to be important for the generation of MAR sites, as well as the augmentation of gene expression in stable transformants (15, 102). The effects of the RBF on the structure of its 54-bp DNA element, with its poly(A)/poly(T) tract, and the site of interaction of PR, are currently under study.

In summary, the data presented in this paper support the model presented in Fig. 10, whereby the nuclear matrix is attached on either side of the RBF element. The nuclear matrix could play a role either in regulating the availability of the RBF/DNA complex for steroid receptor binding, or in regulating the steroid-induced gene transcription after steroid receptor binding has occurred. This model should readily allow the nuclear matrix to alter the structure of the intervening domain with its AT-rich polypurine-polypyrimidine structure, and maintain the latter in a unique structure poised for the interaction with steroid receptors and other transcription factors. The exact secondary/tertiary structures of the DNA element and the quaternary structure of the RBF-DNA complex are currently unknown. In any event, this nuclear matrix structure and steroid receptor binding (acceptor site) are novel chromatin structures and possible participants in the steroid regulation of the c-myc proto-oncogene expression. Further questions arise. 1) Does this nuclear matrix attachment structure in the c-myc gene 5-flanking domain occur in other gene promoters? 2) Is this structure utilized by the progesterone receptor as well as other steroid receptors in vivo and in other animal species? 3) Is this structure specific for "regulatory" MAR structures as opposed to "constitutive" MAR structures as per Bode et al. (15)?.