Progesterone-Dependent Deoxyribonucleic Acid Looping between RUSH/SMARCA3 and Egr-1 Mediates Repression by c-Rel

Abstract

Steroids regulate alternative splicing of RUSH/SMARCA3. The full-length, progesterone-dependent α-isoform and the 3′-truncated, estrogen-dependent β-isoform have identical DNA-binding domains, nuclear localization signals, and RING fingers. Transcription of RUSH/SMARCA3 is mediated by a bipartite progesterone receptor half-site/overlapping Y-box combination (−38/−26), where progesterone activation is attenuated by nuclear factor Y binding. Regulation also involves two GC-rich sequences in the proximal promoter (−162/+90) and a RUSH/SMARCA3 site (−616/−611) in the 5′-untranslated region. Isoform-specific binding to the RUSH/SMARCA3 site is dictated by the hormonal milieu, as is the availability of factors that bind to the distal GC-rich site (−131/−126), a composite binding site for Egr-1/specific protein-1/3/Myc-associated zinc finger protein/myeloid zinc finger-1/c-Rel, and the proximal GC-rich site (−62/−53), which binds only Sp1/3. TransSignal TF-TF interaction arrays, supershift assays, and chromatin immunoprecipitation analyses confirmed strong physical interactions between RUSH/Egr-1 and RUSH/c-Rel that were visualized with fluorescent microscopy. Higher-order, long-range interactions between RUSH and Egr-1/c-Rel derivative of the anisotropic flexibility of the intervening DNA sequence were shown by Chromosome Conformation Capture assay. Glutathione S-transferase pull-downs confirmed that the RING finger is the protein-binding domain, suggesting that the RUSH isoforms have equivalent potential for protein interactions. Transient transfection assays showed that the cooperative interaction between RUSH and Egr-1 mediates repression by c-Rel. Thus, progesterone-induced transcription is fine-tuned by isoform-specific autoregulation, in which newly synthesized RUSH-1α binds DNA and interacts physically with liganded Egr-1 in the proximal promoter via a DNA-looping mechanism to mediate repression by c-Rel. In the absence of progesterone induction, RUSH-1β replaces RUSH-1α binding, Egr-1 and c-Rel are unavailable as molecular ties, and DNA looping is disfavored.

THE DIVERSE EFFECTS of progesterone are mediated via nontranscription (extranuclear) and transcription (nuclear) signaling mechanisms (1). Transcription-based control involves the physical association of genes with their regulators (2,3). This includes the binding of progesterone receptor to response elements in gene targets, interactions with components of the basal transcription machinery, and the recruitment of coregulators. DNA looping (4), which allows distant regulatory regions to affect each other (5), may also be involved in fine-tuning progesterone-induced gene transcription. However, little is known about interactions between remote cis-acting elements and the promoters of progesterone-dependent target genes.

All of the models designed to explain the complex interactions among positive and negative regulatory elements and their distant promoters invoke protein-mediated DNA looping (6). First identified in the Escherichia coli ara operon (7), DNA looping is classified (4) as either energetic (short) or entropic (long). The physical forces that control loop formation, DNA elasticity for short loops and loss of entropy for long loops, is foundational to the classification schema. Loops as short as 60 bp in the lac operon (8) and 80 bp in nucleosome wrapping (9) contrast with loops as long as 180 kb in mating type switching in yeast (10). DNA loop formation in vivo is optimal at approximately 200 bp. Interest in long-range interactions led to the discovery of intrachromosomal looping between the mouse hemoglobin β-chain complex genes and their distal regulatory elements known as the locus control region. RNA tagging and recovery of associated proteins (TRAP) and Chromosome Conformation Capture (3C) assays were used to show that the close proximity of the locus control region and the actively transcribed β-globin genes was achieved by looping the intervening 50 kb of sequence (11). Chromatin immunoprecipitation (ChIP) combined with the 3C assay (ChIP-3C) provided evidence for androgen receptor-mediated DNA looping between the promoter and an upstream (4 kb) enhancer in the prostate-specific antigen (PSA) gene (12).

RUSH/SMARCA3, a progesterone target gene in rabbit endometrium, encodes a member of the switch/sucrose nonfermenting (SWI/SNF) family capable of genetic (sequence-selective DNA binding) and epigenetic (ATP-dependent DNA unwinding) regulation. Posttranscriptional processing produces a full-length, progesterone-dependent, α-isoform and a truncated, estrogen-dependent, β-isoform (13). Posttranslational processing (tyrosine phosphorylation) is required for RUSH-DNA binding (14). The isoforms have the selfsame DNA-binding domains, nuclear localization signals, and RING finger motifs. However, RUSH-1α has seven conserved DNA-dependent ATPase domains (I, Ia, and II–VI) compared with RUSH-1β, which has only the first four. Rabbit uteroglobin promoter binding proteins that are SWI/SNF-related helicases/ATPases (RUSH)-1α, the ortholog to human SWI/SNF-related matix-associated actin-dependent regulator of chromatin subfamily A member 3 (SMARCA3), is a nuclear effector of prolactin signals (15). It mediates the ability of prolactin to augment progesterone-dependent transcription of the rabbit uteroglobin (SCGB1A1) gene, the founding member of the secretoglobin (SCGB) gene family.

An initiator and downstream promoter element (Inr-DPE) combination directs accurate initiation of RUSH transcription (16). Binding of nuclear factor Y (NF-Y) to a progesterone response element (PRE) half-site/overlapping Y-box combination (−38/−26) in the proximal promoter (−162/+90) represses basal transcription, which is synonymous with estrous. Domestic rabbits experience continuous estrous with sexual activity balanced against seasonal bouts of reduced fertility. Follicles are present in the ovaries of sexually mature females, and mating induces ovulation. Progesterone-induced transcription of RUSH/SMARCA3 is mediated at the PRE half-site/overlapping Y-box combination where progesterone action is attenuated by NF-Y binding (16). Repression is also achieved via two GC-rich sites centered at −128 and −58 and a RUSH site at −616/−611 (16). The conservation of the Inr-DPE combination and the GC-rich sites in the putative core promoter of the human ortholog (16) suggested that transcription initiation is similarly conserved. However, the PRE/Y-box combination and the RUSH site are unique to the rabbit gene. Exhaustive analysis of this region (−712/+90) reveals progesterone-dependent DNA looping is an adjunct to progesterone induction.

RESULTS

A schematic (Fig. 1) of the 5′-flanking region of the RUSH/SMARCA3 gene shows the location of bona fide sites in the proximal promoter (−162/+90) that bind repressors (16). These include the NF-Y binding component of the PRE/Y-box combination (−38/−26) and two GC-rich [specific protein-1 (Sp1)] sites at −131/−126 and −62/−53. The RUSH site at −616/−611 is also shown.

Figure 1.

Organization of the 5′-Flanking Region (−712/+90) of the RUSH/SMARCA3 Gene

The 252-bp proximal promoter (−162/+90) contains an Inr element in a TATA-less region with a DPE (+29). Additional authentic regulatory elements include a progesterone receptor half-site (PRE) overlapping a NF-Y binding site (Y-box), Sp1 sites, and a RUSH binding site.

MatInspector Professional (Genomatix) analysis (17) of the RUSH site identified potential transcription factor binding sites for fork-head activin signal transducer-1 on the positive strand, and interferon regulatory factor-3 (IRF-3), serum response factor (SRF), Elk-1, and c-Myb on the negative strand with matrix similarity values of at least 0.80. Gel supershift assays with nuclear extract proteins from progesterone-treated rabbits and commercially available antibodies confirmed that the RUSH site is a solo binding site for RUSH and that none of the aforementioned transcription factors bind there (data not shown). Isoform-specific RUSH binding was demonstrated with gel shift/supershift assays (Fig. 2). RUSH-DNA complexes were supershifted with antibodies to human SMARCA3 in assays with nuclear extract proteins from progesterone-treated animals. Exclusive, site-specific binding of RUSH-1α was confirmed with α-specific antibodies (Fig. 2A). Assays with nuclear extract proteins from estrous animals and β-specific antibodies showed enhanced RUSH-1β-DNA retarded band intensity in the absence of a ternary supershift complex (Fig. 2B). Enhanced intensity after coincubation with β-specific antibodies but not after coincubation with α-specific antibodies confirmed exclusive binding of RUSH-1β to the same site.

Figure 2.

Validation of Hormone-Dependent, Isoform-Specific RUSH Binding to the −616/−611 Site

A, Supershift (SS) assays confirm that only RUSH-1α binds to the site when the source of RUSH is nuclear extract (NE, 5 μg) from progestational endometrium. Lane 1, Labeled RUSH/SMARCA3 probe (−626/−596) alone (PA); lane 2, labeled RUSH/SMARCA3 probe plus NE; lane 3, same as lane 2 plus goat antihuman SMARCA3 (10 μg); lane 4, same as lane 2 plus affinity-purified rabbit anti-RUSH-1α (10 μg). Supershifted bands are designated with an asterisk. B, Assays with nuclear extract proteins from estrous animals and increasing amounts (0–20 μg) of β-specific antibodies showed enhanced RUSH-β-DNA retarded band intensity in the absence of a ternary supershift (inset). There is no gel shift product with probe alone (PA). The graphed lines are the best fit to linear regression analyses. The slope of the line for β-antibody binding from an ANOVA table (F = 128.25) generated with GraphPad InStat version 3 for Macintosh is considered extremely significantly (P = 0.0003) different from the zero (no antibody) value. RUSH-1α binding is negligible with increasing amounts (0–20 μg) of α-specific antibodies.

The GC-rich sites centered at −128 and −58 were previously authenticated as Sp1 binding sites (16). Gel supershift assays with nuclear extract from progesterone-treated animals now confirm that Sp1 and Sp3 proteins bind to both sites (Fig. 3, A and B). Quantitative analysis of the supershift data showed that specific binding of Sp3 was 15- to 17-fold more abundant than Sp1 at each site. Sp1/3-specific binding was eliminated in gel shift assays with mutant oligonucleotide probes. Dissociation curves (Fig. 3C) indicate that the half-life for Sp binding at the distal [Sp1(A)] site that has a consensus (GGGGCGGGG) sequence was twice that of the half-life at the variant (GGGGCGGAG) proximal [Sp1(B)] site. Hormone-dependent differences in Sp1/3 binding were shown with gel shift/supershift assays with nuclear extract proteins from estrous rabbits. Only Sp3 binds to the Sp1(A) site, and protein binding to the Sp1(B) site was negligible (data not shown). Several reports confirm estrogen responsiveness is associated with estrogen receptor-α (ERα)/Sp1 interactions at GC-rich sites (18,19). To test putative ERα/Sp1 interactions in the RUSH promoter, recombinant ERα was added to nuclear extracts from progesterone-treated rabbits in gel shift assays with probes to each Sp1 site. The addition of ERα failed to increase either the mobility or the intensity of the gel shifts (data not shown). Similar results were obtained by the addition of either recombinant ERβ or recombinant TGFα, the negative control.

Figure 3.

Demonstration that the Sp Sites in the RUSH Proximal Promoter Are Sp1/Sp3 Sites

Previously identified as Sp1 binding sites (16), both sites also bind Sp3. A, Supershift (SS) assays with nuclear extract (NE, 5 μg) from the endometrium of progesterone-treated rabbits confirm the specificity of Sp1/Sp3 binding to the Sp1(A) site. Lane 1, Labeled Sp1(A) probe (−139/−118) alone (PA); lane 2, labeled Sp1(A) probe plus NE; lane 3, same as lane 2 plus rabbit antihuman Sp1 (10 μg); lane 4, same as lane 2 plus rabbit antihuman Sp3 (10 μg); lane 5, labeled Sp1(A)m (mutant) probe (−139/−118) plus NE. Supershifted bands are designated with an asterisk. B, Supershift (SS) assays confirm the specificity of Sp1/Sp3 binding to the Sp1(B) site. Lane 1, labeled Sp1(B) probe (−70/−49) alone (PA); lane 2, labeled Sp1(B) probe plus NE; lane 3, same as lane 2 plus rabbit antihuman Sp1 (10 μg); lane 4, same as lane 2 plus rabbit antihuman Sp3 (10 μg).); lane 5, labeled Sp1(B)m probe (−70/−49) plus NE. Supershifted bands are designated with an asterisk. C, Dissociation curves for Sp1(A) and Sp1(B) sites with 10× cold competitor. Results are expressed as time-dependent changes in the amount of binding. The initial (T₀) value was set at 100. A graphical determination of the best-fit line to the data shows the off-rate was two times faster for the nonconsensus Sp1(B) site compared with the consensus Sp1(A) site. This result is representative of seven different experiments.

MatInspector Professional (Genomatix) analysis (17) of the Sp1(A) site identified putative Egr-1/Sp/MYC-associated zinc finger protein (MAZ)/myeloid zinc finger-1 (MZF1) sites on the positive strand, overlapping a putative c-Rel site on the negative strand (Fig. 4A) with matrix similarity values of at least 0.91. Supershift assays with nuclear extracts from progesterone-treated animals confirmed binding of all candidates (data not shown). Interest in the authenticity of this site as a composite binding site for multiple transcription factors was piqued by results from TransSignal TF-TF interaction arrays in which potential physical associations between RUSH/SMARCA3 and non-RUSH proteins were tested. Nuclear extract proteins from a progesterone-treated rabbit were incubated with a library of cis-elements. RUSH antibodies were used to co-immunoprecipitate RUSH-affiliated transcription factors with their corresponding cis-elements. The cis-elements were eluted and hybridized to arrays. Strong physical interactions occurred between RUSH and Egr-1 (Fig. 4B) and RUSH and c-Rel (Fig. 4C). In contrast, no RUSH/Sp1 or RUSH/NF-Y physical interactions were detected.

Figure 4.

Model and TransSignal Protein/DNA Array Analysis

A, A model derived from MatInspector analysis and supported by array data shows the complexity of the Sp1(A) site and the physical interactions between RUSH and two proteins that bind there. B, Sp1 and Egr-1 genes were spotted in duplicate. For each gene, the first row contained DNA spotted normally, the second row contained DNA diluted 10-fold. C, Single spots for NF-Y and c-Rel are shown. Diluted samples were not spotted on the DNA array membrane.

Supershift assays with nuclear extracts from progesterone-treated animals and a RUSH probe showed the physical affiliations between RUSH/Egr-1 (Fig. 5A) and RUSH/c-Rel (Fig. 5B) when RUSH was bound to DNA. Companion supershift assays with nuclear extracts from progesterone-treated animals and an Sp1(A)/c-Relm probe showed the physical affiliation between RUSH and Egr-1 at the composite Sp1(A) site in the absence of c-Rel binding (Fig. 5C). Supershift assays with nuclear extracts from progesterone-treated animals and the Sp1(A)/Egr-1m probe showed the physical affiliation between RUSH and c-Rel at the composite Sp1(A) site in the absence of Egr-1 binding (Fig. 5D). ChIP confirmed the specificity of RUSH-DNA binding as well as RUSH-protein interactions at the composite Sp1(A) site in transcriptionally active chromatin (Fig. 6).

Figure 5.

Authentication of RUSH/Egr-1/c-Rel Interactions by Gel Shift Assays

Supershift (SS) assays with nuclear extract (NE, 10 μg) from the endometrium of progesterone-treated rabbits confirmed RUSH/Egr-1 and RUSH/c-Rel form DNA-protein complexes. A, Lane 1, labeled RUSH probe (−626/−596) alone (PA); lane 2, labeled RUSH probe plus NE; lane 3, same as lane 2 plus mouse anti-Egr-1 (3 μg). B, Lane 1, labeled RUSH probe (−626/−596); lane 2, labeled RUSH probe plus NE; lane 3, same as lane 2 plus goat anti-c-Rel (3 μg). C, Lane 1, labeled Sp1(A)/c-Relm (mutant) probe (−139/−112) alone (PA); lane 2, labeled mutant probe plus NE; lane 3, same as lane 2 plus goat antihuman SMARCA3 (10 μg). D, Lane 1, labeled Sp1(A)/Egr-1m (mutant) probe (−139/−112) alone (PA); lane 2, labeled mutant probe plus NE; lane 3, same as lane 2 plus goat antihuman SMARCA3 (10 μg). Supershifted bands are designated with an asterisk.

Figure 6.

In Vivo Binding of RUSH to DNA and RUSH to Egr-1/c-Rel Was Confirmed by PCR Amplification of Products from ChIP Assays

Lane 1 shows the φX174/HaeIII size markers. Lane 2 shows the single 166-bp amplicons of the RUSH site resulting from IP with antibodies to SMARCA3 (50 μg). Lanes 3 and 4 contain negative IP (IgG) and PCR (water blank) controls, respectively. Lane 5 shows a ChIP specificity control, i.e. the absence of an amplification product from an unrelated region of the RUSH promoter. The inability to generate single 136-bp amplicons of the PRE/Y-box (−45/−15) after IP with antibodies to SMARCA3 also confirms that RUSH does not interact physically with either the progesterone receptor or NF-Y (Fig. 4). Lane 6 shows the single 153-bp amplicon of the Sp1(A) site resulting from IP with antibodies to SMARCA3. DNA was sonicated to an average fragment size of no more than 194 bp to ensure the RUSH and Sp1(A) sites could not co-immunoprecipitate.

A functional corollary of these findings is that RUSH mediates transcription via a connection between the RUSH site and the Sp1(A) site in the proximal promoter. Selective mutation of the RUSH/Egr-1/c-Rel sites individually or in combination showed that c-Rel is an effective repressor (P < 0.001) of progesterone-induced transcription in HRE-H9 cells (Fig. 7). Mutation of either the RUSH site or the Egr-1 site had a negligible effect (P > 0.05) on the repressive action of c-Rel. In contrast, cooperation between RUSH and Egr-1 was confirmed when the c-Rel site was mutated.

Figure 7.

RUSH and Egr-1 Cooperatively Activate the RUSH Gene

Promoter-reporter analysis with transiently transfected HRE-H9 cells shows progesterone (R5020)-induced transcriptional activation is repressed (P < 0.001) by c-Rel in the absence of RUSH/Egr-1 binding. Activation is dramatically increased (P < 0.001) by the cooperative interaction of RUSH and Egr-1 in the absence of c-Rel-DNA binding. Note that neither RUSH nor c-Rel alone is capable of increasing activation. Assays were performed four to six times with six replicates. Values are expressed as mean ± sem and analyzed with GraphPad InStat version 3 for Macintosh. The P value for the Kruskal-Wallis Test (KW = 114.58) was considered extremely significant (P < 0.0001) because the variation among the medians was significantly greater than expected by chance. Mean values were compared with Dunn's multiple comparisons test. Those with the same letter designation are not significantly (P > 0.05) different.

RUSH interactions with Egr-1 and c-Rel were biochemically confirmed by glutathione S-transferase (GST) pull-downs with the RING domain of RUSH (Fig. 8A) and nuclear extract from progesterone-treated animals. As shown in Fig. 8B, both Egr-1 and c-Rel were identified as RING-binding partners when GST pull-down products were resolved by SDS-PAGE and probed with relevant antibodies.

Figure 8.

The RING Domain in RUSH Is the Protein Interaction Site

A, The schematic of the C₃HC₄ RING-finger motif in RUSH shows a series of histidine and cysteine residues with the characteristic spacing that allows for the coordination of two zinc ions in a unique cross-brace structure. The first zinc binding site involves the tetrahedral ligation of one Zn ion by four cysteines, and the second site involves the tetrahedral ligation of one Zn ion by three cysteines and one histidine. B, GST-RING pull-down assays with Western blots for Erg-1 and c-Rel showed the RING finger is the protein interaction domain confirming the RUSH isoforms share the same potential for protein interactions. Molecular mass (kilodaltons) markers are provided.

To examine the intracellular distribution of native RUSH and its putative binding partners, HRE-H9 cells were immunolabeled after treatment with R5020 with or without prolactin to maximize RUSH expression/phosphorylation (14,15). As shown in Fig. 9, both c-Rel and RUSH were found in the cytoplasm and nuclei of HRE-H9 cells treated with R5020 plus prolactin. Colocalization of RUSH/c-Rel in the nucleus is detectable in the merged image after a 5-min treatment with prolactin. Egr-1 is expressed in the cytoplasm, but nuclear expression is negligible after a 5-min treatment with prolactin. However, after a 15-min treatment with prolactin, there is modest but detectable nuclear localization of Egr-1. Colocalization of RUSH/Egr-1 in the nucleus is visible in the merged image.

Figure 9.

Effects of Hormones on the Subcellular Localization of Endogenous RUSH, Egr-1, and c-Rel

Multicolor immunofluorescence confirms the physical affiliation of RUSH/SMARCA3 (green) with Egr-1 (red) and c-Rel (red) in the nuclei of hormone-treated HRE-H9 cells. Merged images are shown in yellow on the right. DAPI nuclear DNA staining is blue. In this temporal sequence of nuclear events, RUSH is affiliated with c-Rel and Egr-1 after 5 and 15 min exposure to prolactin (PRL), respectively. Results are representative of three independent experiments. Scale bar, 20 μm.

Evidence that RUSH interacts with Egr-1 and/or c-Rel through a DNA-looping mechanism was derived from the 3C assay. If the RUSH site is physically linked to the Sp1(A) site in the proximal promoter due to looping, fragments of the two regions generated by digestion with a restriction enzyme that excised a region of the intervening loop should have ligatable ends such that the smaller religated product can be amplified by PCR. When the cross-linked chromatin was digested with HaeIII (or its isoschizomer BsuRI) and ligated, PCR amplification produced a population of 444-bp amplicons (Fig. 10) that were 105 bp shorter than the original 549-bp loop. Excision of a 105-bp fragment from the 549-bp loop was confirmed by sequencing. In addition, a larger band contained PCR products that resulted from either incomplete digestion or an artifact of ligation, i.e. concatamerization of random HaeIII-ended mini-fragments.

Figure 10.

3C Assay Confirms DNA Looping

Protein-protein and protein-DNA cross-links (formaldehyde) were digested with HaeIII or its isoschizomer BsuRI. DNA fragments were ligated at dilute DNA concentrations (2.5 ng/μl) to favor intramolecular ligations. Cross-links were reversed (proteinase K digestion), and fragments were ligated. Reaction products were amplified by PCR. Populations of 444-bp amplicons from which a 105-bp DNA loop was excised from a 549-bp intervening DNA fragment are indicated with an asterisk. The negative PCR (water blank) control and φX174/HaeIII size markers are provided.

DISCUSSION

The expanded model of RUSH/SMARCA3 regulation (Fig. 11) integrates the previous findings (16) that basal (estrous) transcription is driven by an Inr/DPE combination and attenuated by transcription factor binding at RUSH/Sp1/Y-box sites, with the discovery that RUSH-1β binds the RUSH site and that Sp3 is the major repressor at the Sp1(A) site. NF-Y with its histone-like activity (20) binds the Y-box component of the bipartite PRE/Y-box element to repress basal and progesterone-induced transcription. Additional repressors of progesterone-induced transcription include c-Rel at the Sp1(A) site and Sp3 at the Sp1(B) site. Progesterone-induced transcription directs isoform-specific autoregulation in which RUSH-1β is replaced by newly synthesized RUSH-1α that is tethered to the promoter-proximal binding partners Egr-1 and c-Rel via its RING domain to foster DNA looping. Intersite cooperativity between RUSH and Egr-1 stabilizes a higher-order RUSH-promoter complex that challenges repression by c-Rel.

Figure 11.

DNA Looping in the Stepwise Activation of the RUSH/SMARCA3 Gene by Progesterone

This working model emphasizes isoform-specific autoregulation in which RUSH-1β is replaced by newly synthesized RUSH-1α. The α-isoform interacts physically with liganded Egr-1 in the proximal promoter via a DNA-looping mechanism to mediate repression by c-Rel. Progesterone regulates the availability of all of the key factors and therefore the process of DNA loop formation.

A spatiotemporal sequence for these nuclear events is provided by the immunofluorescence microscopy. The arrival of the repressor, c-Rel, precedes the arrival of Egr-1, which joins RUSH-1α to unite their noncontiguous binding sites via a DNA loop. General strategies for loop closure (21) can result from the interactions of two identical or nonidentical proteins bound to DNA, the binding of a bidentate protein to two remote DNA sites, the interactions of two proteins bound at distant sites via an adaptor protein, or the inclusion of an architectural protein to energetically aid DNA looping especially when the intervening DNA length is short, i.e. less than 200 bp. DNA loop formation by heterologous protein-protein interactions follows one of two trajectories, i.e. parallel or antiparallel, depending upon the mutual orientation of the DNA binding sites (21). The parallel DNA trajectory results in an O-shaped geometry whereas the antiparallel trajectory results in a U-shaped geometry. The resultant loop configuration directs additional protein-DNA interactions. Because of the tethering effect, DNA looping significantly increases protein associations at the lower-affinity site. In principle, the RUSH/Egr-1 interactions, which appear to result from the simple interaction of two nonidentical DNA-bound proteins, would produce a parallel DNA trajectory or wrapping mode because the binding sites are oriented in the same direction. However, the interaction with c-Rel might enable a parallel trajectory. The character of the loop geometry and the hierarchical cooperation among factors requires further testing. This will include the use of ChIP and real-time PCR analysis to profile spatial and temporal recruitment of the progesterone receptor and the factors involved in DNA loop formation. Additionally, it will be necessary to deplete the factors (RUSH, Egr-1, and c-Rel) individually, or in combination, to evaluate their effects on progesterone stimulation of the RUSH/SMARCA3 promoter.

These results extend our understanding of RUSH/SMARCA3 function. RUSH, which exists as α- and β-isoforms, binds DNA in a sequence-selective manner. The α-isoform alters progesterone-dependent transcription of two different target genes. Acting alone, RUSH mediates the ability of prolactin to augment progesterone-dependent transcription of the uteroglobin (SCGB1A1) gene through direct binding to a response element in the proximal promoter. Acting synergistically with liganded Egr-1 in the proximal promoter of the RUSH gene, RUSH mediates repression by c-Rel through a DNA-looping mechanism. This regulatory mechanism is considered to be unique to the rabbit gene because the RUSH site is not conserved in the human ortholog. To the best of our knowledge, this is the first demonstration of the involvement of DNA looping in the regulation of a progesterone target gene.

The transcription factor c-Rel is a member of the Rel/NF-κB family that generally functions either as a homodimer or as a heterodimer with other family members (22). A known associate of Sp1 (23), c-Rel/Sp1 interactions in the RUSH promoter were not examined. In contrast, no interactions between RUSH and Sp1 or NF-Y or ERα and Sp1 were detected. Phosphorylation of c-Rel can affect its nuclear availability and DNA binding (24). In the RUSH promoter, the c-Rel effect involves DNA binding. Cell and promoter context are important to understanding Egr-1 function. For example, Egr-1 is activated by nongenomic estrogen action in MCF-7 cells (25,26). In rabbit endometrium, progesterone controls its binding availability.

It is not surprising that the RING domain is the protein interaction site because we previously used this domain to clone and characterize a novel binding partner (27). The real surprise was the fact that RUSH does not functionally interact with Sp1. Sp1 is known to recruit SWI/SNF family members to remodel chromatin. The carboxyl-terminal region of human SMARCA3 binds Sp1/3 (28). The final 85 amino acids that encode the last two motifs (V and VI) of the seven consecutive motifs (I, Ia, and II–VI) characteristic of ATPases and DNA helicases encompass the protein interaction site. Although the amino acid sequence is 95% identical between rabbit and human proteins, the native rabbit protein does not bind Sp1 as determined with the TF-TF interaction arrays. This means the full-length α-isoform and the β-isoform, which is truncated just after the RING domain and therefore devoid of motifs IV, V, and VI, are equipotent in their protein interactions via the RING motif.

Cooperative repression of c-Rel by RUSH/Egr-1 requires further investigation. An intriguing unanswered question is the relative effect of competitive binding of the different factors at the Sp1(A) site on the ratio of looped and nonlooped complexes, the stability of loops, and ultimately transcription. Egr-1/Sp1/3/MAZ/MZF1 all compete for the same site. Only c-Rel binding is independent of the dynamic interaction of the other factors at this location. Double-stranded DNA is a semiflexible polymer with a persistent length of about 150 bp (21). DNA segments that exceed the persistent length, such as the 549-bp segment between the RUSH and Sp1(A) sites, are relatively easy to bend (21) such that looping is controlled by the interactions of proteins with the DNA (4). In this study, progesterone controls DNA loop formation by controlling the availability of RUSH binding partners.

MATERIALS AND METHODS

Reagents and Tools

Staff at Midland Certified Reagent Co. (Midland, TX) synthesized custom oligonucleotides for gel shift assays and primers for PCR. TransSignal TF-TF interaction arrays (MA5010 and MA5011) were purchased from Panomics (Redwood City, CA). NuSieve GTG-agarose gel was purchased from Cambrex Bioscience Rockland, Inc. (Rockland, ME). QIAEX II was purchased from QIAGEN, Inc. (Valencia, CA). Promegestone (R5020) was purchased from NEN Life Science Products, Inc. (Boston, MA). Dr. A. F. Parlow, Scientific Director, National Hormone and Pituitary Program (Torrance, CA), provided the ovine prolactin (oPRL). Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) was the commercial source of consensus and mutant oligonucleotides as well as antibodies for Sp1 (sc-14027X), Sp3 (sc-644X), MAZ (sc-13485X), A-Myb (sc-9957X), c-Myb (sc-7874X), Elk-1 (sc-355X), FAST-1/2 (sc-14031X), IRF-3 (sc-9082X), and serum response factor (SRF; sc-13029X). Abgent (San Diego, CA) was the source of ZFP42 N-term (AP2051a) and ZFP42 C-term (AP2051b) antibodies. These are antibodies to MZF1, also known as zinc finger protein 42 (ZFP42). Antibodies to Egr-1 (H00001958-A01) were purchased from Abnova (Taiwan) Corp. (Taipei, Taiwan) and antibodies to c-Rel (AF2699) were purchased from R&D Systems (Minneapolis, MN). Invitrogen Corp. (Carlsbad, CA) was the source of Alexa-conjugated secondary antibodies, ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI), and recombinant human ERα (P2187) and ERβ (P2466). TGF-α (01-165) was purchased from Upstate (Lake Placid, NY). SMARCA3 antibodies (ab17984) were purchased from Abcam Inc. (Cambridge, MA). Affinity-purified RUSH-1α(β) and RUSH-1α antipeptide antibodies have been extensively characterized (13,15).

RUSH-1β Antibody Preparation

Antipeptide antibodies to the unique portion of the C-terminal end of RUSH-1β, i.e. amino acids RFLSC, were prepared by staff at Research Genetics, Inc., currently known as Open Biosystems (Huntsville, AL). Briefly, the peptide TSSSK RFLSC (peptide A) was synthesized, conjugated to keyhole limpet hemocyanin carrier protein, and used to immunize two rabbits. Resultant antisera were tested by ELISA with free mitogen-activated peptide on the solid phase (1 μg/well), goat antirabbit IgG-horseradish peroxidase conjugate as the secondary antibody, and peroxidase dye to confirm the presence of antibodies that recognized the RFLSC peptide. Next, the peptide SNMEW TSSSK INALM (peptide B) was synthesized. Affinity columns of peptides A and B were prepared. Anti-TSSSK RFLSC antibodies were obtained by affinity purification of serum on column A. Application of the affinity-purified eluant to column B resulted in the flow-through of antibodies that required some portion of RFLSC for binding. In other words, anti-RFLSC antibodies did not bind to column B. The flow-through containing anti-RFLSC antibodies was concentrated. Binding to RFLSC peptide was verified by ELISA, and binding to RUSH-1β was verified by Western analysis. For gel supershift assays, pooled aliquots of these antibodies (800 μl) were dialyzed against two changes of 1 liter each of Tris-EDTA buffer (10 mm Tris-HCl, pH 8.0; 1 mm EDTA) at 4 C.

Animal Treatments

All studies with virgin, adult New Zealand White rabbits were conducted according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals, as reviewed and approved by the Animal Care and Use Committee at TTUHSC. Estrous animals were either used as controls or given sc injections of progesterone (3 mg/kg·d) for 5 d. Nuclei were isolated from the endometrium (29).

Transient Transfection Assays

Directional cloning of the RUSH construct (−712/+90) into the pGL3 vector has been described (16). RUSH, Egr-1, and c-Rel sites were mutated individually or in combination with the QuikChange site-directed mutagenesis kit and evaluated (15,16) in HRE-H9 cells. This SV-40 transformed rabbit uterine epithelial cell line was derived from the endometrium of human chorionic gonadotropin-treated pseudopregnant rabbits (30). Transfections were performed four to six times with six replicates (n = 23–36). Cells were treated (15,16) with or without R5020 (10⁻⁸ m) in dimethylsulfoxide (0.1% vol/vol) for 20 h. pRUSH-LUC expression was normalized to cotransfected pRL-TK-LUC expression to control for transfection efficiency. Multiple ratio data were ranked (31), and the ranks between the groups were analyzed by Kruskal-Wallis (nonparametric) ANOVA followed by Dunn's multiple comparisons test (P < 0.05 significance level) using GraphPad InStat version 3 for Macintosh (GraphPad Software, Inc., San Diego, CA).

Kits and Assays Previously Described by Us

Nuclear extract proteins (29) were used in gel shift assays (13,14,15,16,29). Oligonucleotide probes to the RUSH site (−626/−596), Sp1(A) site (−139/−118), and Sp1(B) site (−70/−49) as well as mutant probes, i.e. Sp1(A)/c-Relm (−139/−112, C→A change at position −117), Sp1(A)/Egr-1m (−139/−112, GGG CGG→CTG CAG change at −131/−126), and truncated-Sp1(A)m (−139/−118, GGG CGG→CTG CAG change at −131/−126), were ³²P labeled with [γ-³²P]ATP using polynucleotide kinase (50,000 cpm/ng) for binding reactions. Standard gel shift assays were used for off-rate determinations. A range from 10- to 75-fold molar excess of the competitor (binding site-specific oligonucleotide) was added to each binding reaction. The entire reaction was loaded at 0, 1, 2, 5, 10, 20, 30, 40, and 60 min after addition of the competitor. Relative band intensities were quantified (29) with a computer-assisted image analysis system (Bio-Image Visage 2000). ChIP from isolated nuclei (16) was modified such that DNA was sequentially cross-linked with disuccinimidyl glutarate (2 mm) and formaldehyde (2%) and then sonicated to an average fragment size of no more than 194 bp as verified by agarose gel (1.5%) electrophoresis. Interaction profiles for RUSH and 150 unique transcription factors were generated with TransSignal TF-TF interaction arrays (14,16). PCR products from the amplification of GST-RING, ChIP, and 3C assays were fractionated by electrophoresis on 2.5% NuSieve GTG-agarose gels, extracted with QIAEX II according to the manufacturer's instructions, and ligated to pCRII-TOPO (16). Insert orientation was confirmed by sequencing in both directions by the dideoxy chain termination method.

GST-RING

The RING motif was amplified from pET19b-RING with a forward primer (5′-GCG GAT CCG AAG AAT GTG CTA TAT-3′) that had a unique BamHI site and a reverse primer (5′-GCC TCG AGT CAT CCA TGT ATA TCA T-3′) that had a unique XhoI site. The reaction conditions were as follows: 30 sec at 94 C, followed by five cycles of 94 C for 5 sec and 63 C for 60 sec, five cycles of 94 C for 5 sec and 62.5 C for 60 sec, 5 cycles of 94 C for 5 sec and 62 C for 60 sec, five cycles of 94 C for 5 sec and 61.5 C for 60 sec, 15 cycles of 94 C for 5 sec and 61 C for 60 sec, and a final extension for 10 min at 68 C. Samples were rapidly cooled to 4 C and processed as described earlier. After cloning into pCRII-TOPO (see above) the RING motif was excised and directionally subcloned into the BamHI-XhoI sites of the pGEX-6p-1 vector. GST-RING pull-down assays were performed as described (27).

PCR Amplification from ChIP Products

Amplification of the RUSH binding site (16) was modified slightly so that it could be amplified in the same reaction as the Sp1(A) site and PRE/Y-Box. Forward (5′-CCC GAG GCC ACA GGT CCA TCG TC-3′) and reverse (5′-C ATT TCT TAG CGC GGG AAG ACT GTG AC-3′) primers that flank the region (−712/−543) that contains the RUSH site and forward (5′-CAC CAG AAA ACG CAG CAC CAT CGC AC-3′) and reverse (5′-CTC CGC CCC GCG CTG CGA TTC-3′) primers that flank the region (−206/−154) that contains the Sp1(A) site as well as forward (5′-GCA ACC GGG ACC CCC ACT G-3′) and reverse (5′-GAC CAG CGA CGC TCT GAC TG-3′) primers that flank the region (−98/+38) that contains the PRE/Y-box were used in a five-step PCR. Reaction conditions were as follows: 30 sec at 94 C, followed by five cycles of 94 C for 5 sec and 63 C for 60 sec, five cycles of 94 C for 5 sec and 63.5 C for 60 sec, five cycles of 94 C for 5 sec and 64 C for 60 sec, five cycles of 94 C for 5 sec and 64.5 C for 60 sec, 15 cycles of 94 C for 5 sec and 65 C for 60 sec, and a final extension for 10 min at 68 C. Samples were rapidly cooled to 4 C and processed as described earlier (16).

3C Assay

Aliquots of nuclei from a progesterone-treated rabbit (1 × 10⁶ nuclei, ∼15 μg) were fixed in buffered formaldehyde [1% formaldehyde, 0.01 m NaCl, 0.1 mm EDTA (pH 8), and 5 mm HEPES (pH 7.9)] for exactly 15 min at room temperature. Fixation was quenched with 0.125 m glycine. Nuclei were washed with PBS/Nonidet P-40 (0.5%) and collected by centrifugation. Nuclei were resuspended in lysis buffer [10 mm Tris-HCl (pH 8.0), 10 mm NaCl, 0.2% Nonidet P-40, and 0.5 mm dithiothreitol] plus a cocktail of protease inhibitors [leupeptin (1 μg/ml), antipain (2 μg/ml), benzamidine (10 μg/ml), chymostatin (10 μg/ml), pepstatin (10 μg/ml), and phenylmethylsulfonyl fluoride (2 mm)] and incubated for 90 min on ice. DNA was completely digested overnight at 37 C with HaeIII or its isoschizomer BsuRI. DNA was extensively diluted and ligated (2.5 ng/μl) overnight at 16 C. Cross-linking was reversed by proteinase K digestion (65 C overnight), and ligated DNA was purified (phenol/chloroform extraction and ethanol precipitation). The sequence between the RUSH and Sp1(A) sites was amplified from 300 ng ligated DNA with forward (5′-CCC GAT GAA GAT TTT CCA AAC TCG GTG TAT C-3′) and reverse (5′-CAG ACG TCG TCG CCG TTT CCT TC-3′) primers according to the following program: 60 sec at 95 C, followed by 35 cycles of 95 C for 60 sec and 65 C for 45 sec and one cycle of 95 C for 40 sec, 65 C for 45 sec, and 72 C for 480 sec. Samples were rapidly cooled to 4 C and processed as described earlier.

Immunofluorescence

HRE-H9 cells were grown on poly-d-lysine-coated coverslips and treated with or without R5020 (10⁻⁸ m) in dimethylsulfoxide (0.1% vol/vol) for 20 h with or without prolactin (10⁻⁸ m) for 0, 5, 10, and 15 min. Cells were fixed in cold methanol (−20 C) for 10 min followed by cold acetone (−20 C) for 1 min. Cells were rinsed in PBS and blocked in 10% normal chicken serum with 0.3% Triton X-100. Cells were incubated with primary antibodies (1:500), i.e. rabbit anti-SMARCA3, mouse anti-Egr-1, and goat anti-cRel, alone or in combination, overnight at 4 C. After two rinses in PBS, cells were incubated with secondary antibodies (1:400) for 60 min at room temperature in the dark. These included Alexa 488-conjugated chicken antirabbit IgG, Alexa 594-conjugated chicken antimouse IgG, and Alexa 594-conjugated chicken antigoat IgG. Cells were rinsed in PBS and mounted in ProLong Gold antifade reagent with DAPI. Controls included cells that were not treated with hormones and labeling reactions that excluded either primary or secondary antibodies. Immunofluorescence was observed with an inverted Olympus 1X71 microscope coupled to a CoolSNAPHQ digital camera (Roper Scientific, Inc., Tucson, AZ) that has high quantum efficiency in the near-infrared region of the spectrum. Digital pictures were acquired with the MetaMorph 6.3 (Molecular Devices, Sunnyvale, CA) image analysis computer program.