Cloning and Characterization of an Atypical Type IV P-type ATPase That Binds to the RING Motif of RUSH Transcription Factors*

RUSH proteins are SWI/SNF-related transcription factors with RING finger signatures near their COOH termini. Long suspected of mediating protein-protein interactions, the RING motif was used to clone a binding partner. The RING finger binding protein (RFBP) is a Type IV P-type ATPase, a putative phospholipid pump, with conserved sequences for two loop segments, an ATP-binding site, a phosphorylation domain, and transmembrane passes potentially involved in substrate binding and translocation. However, RFBP differs from all other Type IV P-type ATPases in three ways. It has only three of four highly conserved NH 2 -terminal trans- membrane passes, it is located in the inner nuclear membrane, and it binds the RING domain. Topographi-cally the orientation of the adjacent hydrophilic domains and the determinants of transport specificity are altered. As a result, the small, hydrophilic loop extends into the perinuclear space that is contiguous with the lumen of the endoplasmic reticulum. The large, conformationally flexible loop extends into the nucleoplasm to contact euchromatin. Competitive reverse transcriptase-polymerase chain reaction and high performance liquid chromatography analysis revealed that endometrial RFBP mRNA expression is hormonally regulated. The physical association of a hormone-dependent RING finger-binding protein with transcriptionally active chromatin supports the speculation that RFBP plays a role in the subnuclear trafficking of transcription factors with RING motifs. RUSH is an acronym for proteins with RING finger motifs that bind to products by agarose gel (1.5%) electrophoresis revealed a single 258-bp amplicon. A 1:100 dilution of this product was amplified again with gene-specific primers. This secondary PCR reaction was performed in a 100- m l PCR reaction with TaKaRa ExTaq DNA polymerase (3 units/100 m l), TaqStart antibody (0.66 m g/100 m l), deoxynucleoside triphosphates (0.2 m M each), and 0.4 m M each RFBP-specific primer. The RFBP-specific primers were forward, 5 9 -GTG CGT GGA CTC CCT ATG CTG CCC AG-3 , and reverse, 5 9 -GGA A total of 18 PCR cycles were performed as described above. The quality and size of the final reaction product was again verified by gel electro- phoresis. The PCR product was subcloned into pCR®II-TOPO, and closed circular DNA was purified by equilibrium centrifugation in CsCl- ethidium bromide gradients. The identity of the MIMIC competitor was confirmed by sequencing in both directions by the dideoxy chain termi- nation method. The pCR®II-TOPO vector containing the MIMIC competitor cDNA insert was linearized with the restriction enzyme Eco RV, and competitor RNA was transcribed. RNA was precipitated with isopropanol to prevent the coprecipitation of free nucleotides, resulting in a more accurate estimation of the RNA yield at 260 nm. Serial dilutions of competitor RNA (6, 3, 1.5, 0.6, 0.3, and 0.1 pg) were combined with poly(A) 1 RNA (1 ng) from HRE-H9 cells and reverse-transcribed with Moloney murine leukemia virus and random hexamers for 30 min at 42 °C. Products from this reaction were heated to 99 °C and cooled to 4 °C. Then 40 m l of a PCR mix containing Amplitaq gold DNA polymerase (1 units/50 m l) and 0.25 each DNASep an alkylated eluted in a 4-min of acetonitrile in M triethylammonium (pH at a of ml/min and products titration analysis show the preservation native a poly(A) Estimation similar to the confirming the precision of the single-tube with the denaturing HPLC quantification system. The repeated analysis n 5 of from HRE-H9 cells yielded a of 8.4%. The single tube analysis was then used to determine the relative expression native in

genes is accelerated or "RUSHed." Sequence alignment of homologs from rabbit, human, rodent, and plant (3) showed that all RUSH proteins contain the RING finger motif (C 3 HC 4 or RING-HC), which binds two zinc ions in a unique cross-braced system (4). RING fingers can be associated with other motifs to form larger, conserved domains such as the RING finger-B box-␣-helical coiled-coil (RBCC) domain (4). Some changes have occurred in the RING such that RING-H2 (C 3 H 2 C 3 ) designates a subclass in which a histidine residue was substituted for the fourth cysteine residue. The U-box (UFD2 homology domain) is a degenerate version of the RING motif that lacks the signature metal-chelating residues (5,6). Unlike the authentic RING finger that is stabilized by binding zinc ions, it has been inferred from the PROMODII model that the U-box is maintained by a system of salt bridges and hydrogen bonds (6).
It is generally accepted that the RING motif binds two zinc ions to form an integrated structural unit that mediates protein-protein interactions (4,(7)(8)(9). The search for functional binding partners ultimately led to the demonstration that the RING domain in one protein can associate with another RINGdomain to promote homo-or heterodimerization. Alternatively, the RING domain can also associate with non-RING domains (10). The recent discovery of the functional link between the ubiquitin-proteasome pathway and the RING-containing protein Cbl has attracted considerable attention (11,12). Although these studies support the idea that the RING motif is involved in ubiquitin-mediated proteolysis (11), Borden (9) speculates that the RING finger may serve as a scaffold for the evolution of different functions. Copps et al. (13) suggest that the RING motif may play a regulatory role in the ordered assembly of factors. Borden et al. (14) and Le et al. (15) used site-directed mutagenesis to show that the RING finger motif of the promyelocytic leukemia oncoprotein PML is necessary for the formation of speckled nuclear bodies also known as promyelocytic oncogenic domains. More recently, Cao et al. (16) show that mutations of the RING finger resulted in the retention of the RET finger protein in the cytoplasm even though the putative nuclear localization signal was intact. These studies suggest that the RING finger plays a role in nuclear targeting.
When the RING motif of the RUSH gene was used to clone a RING finger-binding protein (RFBP), 1 a new Type IV P-type ATPase was identified in the inner nuclear membrane. The topographical orientation of the molecule allows the large conformationally flexible loop portion of the protein to contact euchromatin. These physical affiliations support the specula-tion that RFBP and RUSH are functionally linked during transcription.

EXPERIMENTAL PROCEDURES
Reagents, Antibodies, and Cells-cDNA Synthesis System Plus, cDNA cloning system-gt11, Sequenase Version 2.0 DNA sequencing kit, pGEX-2TK vector, glutathione-Sepharose 4B, protein A-Sepharose™CL-4B, isopropyl-␤-D-thiogalactopyranoside, and the following isotope were purchased from Amersham Pharmacia Biotech: [␣- 35 S]dATP, 1000 Ci/mmol, for sequencing. The SMART rapid amplification cDNA ends (RACE) and Marathon TM cDNA amplification kits, polymerase chain reaction (PCR) MIMIC construction kit, Rabbit genomic library, TaqStart antibody, Tth Start antibody, Advantage genomic PCR kit, and Advantage Tth polymerase mix were purchased from CLONTECH Laboratories, Inc. (Palo Alto, CA). TaKaRa ExTaq enzyme and 10ϫ LA PCR buffer were purchased from PanVera Corp. (Madison, WI). The GeneAmp RNA PCR core kit and Amplitaq gold DNA polymerase were purchased from PerkinElmer Life Sciences. The vector pSCREEN-1b(ϩ), thrombin cleavage capture kit, T7-Tag antibody alkaline phosphatase conjugate, S-protein alkaline phosphatase conjugate, and BLR(DE3)pLysS cells were purchased from Novagen (Madison, WI). The pCR™II and pCR®II-TOPO vectors are components of individual TA cloning kits, which were purchased from Invitrogen (San Diego, CA). MetaPhor-agarose was purchased from BioWhittaker Molecular Applications, Inc. (Rockland, ME). A 1-kb ladder was purchased from Promega (Madison, WI). The binary gradient HPLC system was purchased from Rainin Instrument Co., Inc. (Woburn, MA), and the DNASep columns containing alkylated polystyrene-divinylbenzene packing were purchased from Transgenomic, Inc. (San Jose, CA). Kodak X-Omat AR film was purchased from Eastman Kodak Co. Renaissance chemiluminescence reagents were purchased from PerkinElmer Life Sciences. Nitrocellulose transfer/immobilization membranes were purchased from Schleicher & Schuell. Tri reagent was purchased from Molecular Research Center, Inc. (Cincinnati, OH). LR White resin and goat anti-rabbit IgG conjugated to 6or 15-nm gold were purchased from Electron Microscopy Sciences (Ft. Washington, PA).
Rabbit antipeptide antibodies were made to a keyhole limpet hemocyanin peptide at Research Genetics (Huntsville, Alabama). Amino acids 663-678 were selected because they displayed strong antigenicity according to the PeptideStructure program (Genetics Computer Group Software, Madison, WI), and because they are unique to RFBP. They share only 3 of 16 amino acids common to any authentic Type IV P-type ATPase in the GenBank TM data base. The antipeptide antibody titer was determined with an enzyme-linked immunosorbent assay with free peptide on the solid phase (1 g/well), goat anti-rabbit IgG-horseradish peroxidase conjugate as the secondary antibody, and peroxidase dye. To reduce complications from nonspecific binding, antibodies were affinitypurified. Horseradish peroxidase-conjugated donkey anti-rabbit IgG was purchased from Amersham Pharmacia Biotech. Midland Certified Reagent Co. (Midland, TX) synthesized all PCR primers. Dr. J. Y. Chou, Human Genetics Branch, NICHD, National Institutes of Health (Bethesda, MD), provided HRE-H9 cells.
Cloning of the RING Motif-For cloning, the RING motif was amplified from 50 ng of a 1030-bp partial RUSH cDNA clone (2). The 50-l PCR reaction mix contained LA PCR buffer (1X), TaKaRa ExTaq DNA polymerase (2.5 units/50 l), TaqStart antibody (0.55 g/50 l), dNTPs (0.2 mM each), and primers (0.2 mM each). The forward primer (5Ј-GCG2AGCT1CA TGT GCT ATA TGC TTG G-3Ј) had a unique SacI site (bold), and the reverse primer (5Ј-GCA2AGCT1TC TGC ATA AAG GGC AC-3Ј) had a unique HindIII site (bold). A four-step, hot-start PCR reaction was performed. The conditions were as follows: 45 s at 95°C followed by 10 cycles of 94°C for 45 s, 39.5°C for 60 s, 72°C for 60 s, 10 cycles of 94°C for 45 s, 38.5°C for 60 s, 72°C for 60 s, 10 cycles of 94°C for 45 s, 55°C for 60 s, 72°C for 60 s, 20 cycles of 94°C for 45 s, 54°C for 60 s, 72°C for 60 s, and a final extension for 5 min at 72°C. Samples were rapidly cooled to 4°C. A single 142-bp PCR product was cloned into pCR TM II, excised with appropriate restriction enzymes, and directionally subcloned into the SacI-HindIII sites of the pSCREEN-1b(ϩ) expression vector. Insert orientation was confirmed by sequencing in both directions by the dideoxy chain termination method.
Recombinant protein with a T7 tag, a His tag, and an S tag at its NH 2 terminus was induced by exposure to isopropyl-␤-D-thiogalactopyranoside (1 mM) and purified by metal chelation chromatography. Briefly, recombinant His-tag protein was eluted under native conditions (1 M imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9), dialyzed against water to remove excess salt, and lyophilized. The protein concentration (1.05 mg/ml) was determined according to Lowry et al. (17) using bovine serum albumin as the standard. The leader sequence containing the T7 tag, and the His tag was removed by digestion with biotinylated thrombin. Biotinylated thrombin was separated from the remaining S-tag protein with streptavidin-agarose.
Cloning of the RING Finger Binding Partner-A gt11 cDNA expression library was prepared using mRNA isolated from HRE-H9 cells (18) and screened according to Macgregor et al. (19) to identify protein binding partners. Nitrocellulose filter replicas were prepared from gt11 recombinants (Ϸ3 ϫ 10 4 plaque-forming units/plate) and processed with 6 M guanidine-HCL denaturation-renaturation (20,21) to ensure proper protein folding. The filters were then screened with the recombinant RING finger protein (1 g/ml) with ZnSO 4 (10 M) to maintain functional conformation (22,23) and the T7-tag antibody alkaline phosphatase conjugate (1:10,000). A single, plaque-pure clone was rescreened with the thrombin-cleaved recombinant protein (1 g/ ml) in the presence of ZnSO 4 (10 M) and the S-protein alkaline phosphatase conjugate (1:5,000).
A positive plaque was heated at 100°C for 10 min, then amplified in a 50-l PCR reaction mix containing LA PCR buffer (1ϫ), TaKaRa ExTaq DNA polymerase (2.5 units/50 l), TaqStart antibody (0.55 g/50 l), dNTPs (0.2 mM each), and gt11 primers (0.2 mM each). Forward (5Ј-GGT GGC GAC GAC TCC TGG AGC CCG-3Ј) and reverse (5Ј-TTG ACA CCA GAC CAA CTG GTA ATG-3Ј) gt11 primers flank the insert sequence in the vector. Hot-start PCR reactions were performed with the following conditions: 120 s at 95°C followed by 35 cycles of 94°C for 45 s, 54°C for 60 s, 72°C for 180 s, and a final extension for 10 min at 72°C. Test and control reactions were rapidly cooled to 4°C. A single 1584-bp PCR product was cloned into pCR TM II and sequenced in both directions by the dideoxy chain termination method.
Four different Marathon RACE reactions were performed to obtain overlapping cDNA clones as described under "Results" and in Fig. 1. Poly(A ϩ ) RNA (1 g) from HRE-H9 cells was used in the preparation of adaptor-ligated cDNA libraries. For each PCR reaction, 5 l of diluted cDNA library was mixed with 45 l of a PCR reaction mix containing LA PCR buffer (1ϫ), TaKaRa ExTaq DNA polymerase (2.5 units/50 l), TaqStart antibody (0.55 g/50 l), dNTPs (0.2 mM each), and primers (0.2 mM each). In the first reaction ( Fig. 1), RACE primer pair 1 consisted of a forward gene-specific primer (5Ј-GTG CGT GGA CTC CCT ATG CTG TTT CCC-3Ј) and a reverse adaptor primer (AP1; 5Ј-CCA TCC TAA TAC GAC TCA CTA TAG GGC-3Ј). A three-step, hot-start PCR reaction was performed with the following conditions: 60 s at 95°C followed by 5 cycles of 94°C for 30 s, 72°C for 60 s, 72°C for 180 s, 5 cycles of 94°C for 30 s, 71°C for 60 s, 72°C for 180 s, 25 cycles of 94°C for 30 s, 70°C for 60 s, 72°C for 180 s, and a final extension for 10 min at 72°C. Samples were rapidly cooled to 4°C. A single 1151-bp PCR product ( Fig. 1) was cloned into pCR®II-TOPO and sequenced in both directions by the dideoxy chain termination method.
In the second reaction ( Fig. 1), RACE primer pair 2 consisted of a degenerate forward primer (5Ј-GAY AAR ACN GGN ACN YTN ACN-3Ј, Y ϭ pyrimidine, R ϭ purine, and N ϭ A, T, G, or C) and a reverse gene-specific primer (5Ј-CGC TTT CTG CAG TGG TGC CAT ACG ACA GC-3Ј). A four-step, hot-start PCR reaction was performed with the following conditions: 30 s at 94°C followed by 5 cycles of 94°C for 5 s, 65°C for 240 s, 5 cycles of 94°C for 5 s, 60°C for 240 s, 5 cycles of 94°C for 5 s, 55°C for 240 s, 20 cycles of 94°C for 5 s, 50°C for 240 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. A single 1185-bp PCR product was cloned into pCR®II-TOPO and sequenced in both directions by the dideoxy chain termination method.
The third reaction actually consisted of a primary reaction followed by a nested secondary reaction (Fig. 1). In the primary reaction, the RACE primer pair consisted of the forward AP1 primer and a reverse, gene-specific primer (5Ј-CAG TGT GAC AGA GAC TGA CTG C-3Ј). The hot-start PCR reaction was performed with the following conditions: 30 s at 94°C followed by 30 cycles of 94°C for 5 s, 62.5°C for 240 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. Products from this reaction were used in a nested PCR reaction with RACE primer pair 3 ( Fig. 1), which consisted of forward adaptor primer 2 (AP2; 5Ј-ACT CAC TAT AGG GCT CGA GCG GC-3Ј) and a reverse gene-specific primer (5Ј-CCT TCT GAA GAA TCT GGT GTC GGT CC-3Ј). The hot start PCR reaction was performed with the following conditions: 30 s at 94°C followed by 25 cycles of 94°C for 5 s, 66.5°C for 240 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. A single 992-bp PCR product was cloned into pCR®II-TOPO and sequenced in both directions by the dideoxy chain termination method.
The fourth reaction also consisted of a primary reaction followed by a nested secondary reaction (Fig. 1). In the primary reaction, the RACE primer pair consisted of forward AP1 and a reverse gene-specific primer (5Ј-CAA GGC CAC CAC TTC GAA CAA CAT AAA CAG G-3Ј). A six-step, hot start PCR reaction was performed with the following conditions: 30 s at 94°C followed by 5 cycles of 94°C for 5 s, 70°C for 150 s, 5 cycles of 94°C for 5 s, 69°C for 150 s, 5 cycles of 94°C for 5 s, 68°C for 150 s, 5 cycles of 94°C for 5 s, 67°C for 150 s, 5 cycles of 94°C for 5 s, 66°C for 150 s, 15 cycles of 94°C for 5 s, 65°C for 150 s, and a final extension for 10 min at 68°C. Products from this reaction were used in a nested PCR reaction with RACE primer pair 4 ( Fig. 1), which consisted of forward AP2, and a reverse gene-specific primer (5Ј-GTC TCA ACC AAT CTT CGT ACC CCT G-3Ј). A six-step, hot-start PCR reaction was performed with the following conditions: 30 s at 94°C followed by 5 cycles of 94°C for 5 s, 68°C for 120 s, 5 cycles of 94°C for 5 s, 67°C for 120 s, 5 cycles of 94°C for 5 s, 66°C for 120 s, 5 cycles of 94°C for 5 s, 65°C for 120 s, 5 cycles of 94°C for 5 s, 64°C for 120 s, 10 cycles of 94°C for 5 s, 63°C for 120 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. A single 314-bp PCR product was cloned into pCR®II-TOPO and sequenced in both directions by the dideoxy chain termination method.
Genomic Cloning-An amplified rabbit genomic library in the EMBL3 SP6/T7 vector was screened with the 992-bp PCR product of the third PCR reaction described above and in Fig. 1. Nitrocellulose filter replicas were prepared from gt11 recombinants (Ϸ3 ϫ 10 4 plaqueforming units/plate) and screened with the random-prime-labeled cDNA (specific activity ϭ 3-6 ϫ 10 8 CPM/g). A genomic clone (Ϸ14 kb) was isolated and used as template in a PCR reaction. A region of the genomic DNA (150 ng) was amplified in a 50-l PCR reaction containing PCR buffer (1ϫ), magnesium acetate (1.1 mM), Tth DNA polymerase (0.1 units), TthStart antibody (0.01 g/l), dNTPs (0.2 mM each), and primers (0.2 mM each). The forward (5Ј-CAG GGG TAC GAA GAT TGG TTG AGA C-3Ј) and reverse (5Ј-CTG TCA GCG TAC CCG TTT TAT CTG TAA ACA C-3Ј) primers matched sequence in the cDNA. The hot-start PCR amplification reaction was performed as follows: 30 s at 94°C, followed by 5 cycles of 94°C for 5 s, 67°C for 360 s, 5 cycles of 94°C for 5 s, 66°C for 360 s, 5 cycles of 94°C for 5 s, 65°C for 360 s, 5 cycles of 94°C for 5 s, 64.5°C for 360 s, 20 cycles of 94°C for 5 s, 64°C for 360 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. A single 742-bp PCR product was cloned into pCR®II-TOPO and sequenced in both directions by the dideoxy chain termination method.
Subcellular Fractionation of Endometrium-Rabbit endometrium was homogenized (11 strokes) in 3 volumes (w/v) of buffer (10 mM Tris-HCl, pH 8.0, 100 mM KCl, 3 mM MgCl 2 , 250 mM sucrose, and 5 mM dithiothreitol) using a Potter-Elvejhem motor-driven Teflon pestle at 4°C. To test in vitro protein-protein interactions, Nonidet P-40 was added to an aliquot of the whole tissue homogenate to a final concentration of 1%. This detergent lysate was re-homogenized (5 strokes) and centrifuged at 2000 ϫ g to remove cellular debris. The supernatant fraction was used as a source of protein (28.5 g/l) in glutathione S-transferase (GST) pull-down assays. The remaining whole tissue homogenate was fractionated by centrifugation/ultracentrifugation to obtain nuclear (2000 ϫ g), mitochondrial (10,000 ϫ g), microsomal (165,000 ϫ g), and postmicrosomal supernatant fractions for Western analysis. Detergent lysates (1% Nonidet P-40) of the nuclear fraction were also prepared as a source of protein (22.5 g/l) in GST pull-down assays. The protein concentrations for all fractions were determined according to Lowry et al. (17) using bovine serum albumin as the standard.
GST Pull-down Assays and Immunoblotting-The RING motif was amplified from 50 ng of a 1030-bp partial RUSH cDNA clone (2), as described above with a forward primer (5Ј-GCG2GATC1CG AAG AAT GTG CT-3Ј) that had a unique BamHI site (bold) and a reverse primer (5Ј-GCG2AATT1CC ATG TAT ATC ATT-3Ј) that had a unique EcoRI site (bold). To accommodate the low melting temperature of this primer pair, a four-step touch-up PCR reaction was performed with a hot start. The conditions were as follows: 30 s at 94°C followed by 5 cycles of 94°C for 5 s, 35°C for 120 s, 5 cycles of 94°C for 5 s, 40°C for 120 s, 5 cycles of 94°C for 5 s, 45°C for 120 s, 5 cycles of 94°C for 5 s, 50°C for 120 s, and a final extension for 10 min at 68°C. Samples were rapidly cooled to 4°C. A single 161-bp PCR product was cloned into pCR TM II, excised with appropriate restriction enzymes, and directionally subcloned into the BamHI-EcoRI sites of the pGEX-2TK vector. Insert orientation was confirmed by sequencing in both directions by the dideoxy chain termination method. BLR(DE3)pLysS host cells were transformed with either the pGEX-2TK control or the pGEX-2TK-RING recombinant. Individual cultures were grown to an A 260 of 0.6, and isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 1 mM. After 3 h of additional growth, bacteria were pelleted by centrifugation at 5000 ϫ g for 10 min. Bacteria were resuspended in 1/10th volume of lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, leupeptin (1 g/ml), antipain (2 g/ml), benzamidine (10 g/ml), chymostatin (10 g/ml), pepstatin (10 g/ml), and phenylmethylsulfonyl fluoride (2 mM)). Bacteria were lysed on ice with mild sonication and centrifuged at 10,000 ϫ g for 10 min. Pellets were discarded, and supernatants (1 ml/each) were incubated with 25 l of glutathione-Sepharose (50% slurry in lysis buffer with 0.5% powdered milk) for 90 min at 4°C. GST and GST-RING were mixed with detergent lysates containing RFBP at 4°C for 90 min. Binding complexes were washed 3ϫ lysis buffer, fractionated by SDS/polyacrylamide gel electrophoresis in 12.5% gels and transferred to nitrocellulose membrane.
For coimmunoprecipitation assays, aliquots of whole tissue homogenates were incubated overnight at 4°C with anti-RFBP then incubated for 2 h at 4°C with a 50% slurry of protein A-Sepharose. Proteins were fractionated by SDS/polyacrylamide gel electrophoresis in 10% gels and transferred to nitrocellulose membrane. Membranes were processed exactly as for Western analysis except they were incubated with affinity-purified anti-RUSH-2 antibodies (1:100) as described (2).
Immunoelectron Microscopy-Nuclei from endometrium of progesterone-treated rabbits were isolated Ϯ 0.1% Triton X-100 on sucrose cushions containing protease inhibitors (24). Nuclear pellets were resuspended in phosphate-buffered paraformaldehyde (4%) and fixed for 30 min. Nuclei were dehydrated in a graded series of ethanol and embedded in LR white resin (50°C). Thin sections (600 -800 Å) were mounted on nickel grids, blocked with 5% normal goat serum, and incubated in antipeptide antibody (1:100 with 5% normal goat serum) overnight at 4°C. Grids were rinsed and incubated with goat antirabbit IgG conjugated to 6-or 15-nm gold (1:25 with 5% normal goat serum) for 2 h at 37°C. Grids were fixed with phosphate-buffered glutaraldehyde (3%) for 5 min, stained with uranyl acetate and lead citrate, and viewed with a Hitachi 600 electron microscope operated at 75 kV.
Animal Treatments-All studies were conducted in accord with 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 Texas Tech University Health Sciences Center. Adult New Zealand White rabbits (6 months of age) were housed for 3 weeks before experimentation. Seventeen estrous animals were divided into two groups. For group one, endometrium from six animals was pooled, and RNA was isolated using a cold precipitation method (25). For group two, the rabbits were used in three experimental subgroups (n ϭ 3 animals/subgroup; 5 animals in the progesterone treatment category). Animals were injected every 24 h with hormones and killed 24 h after the last injection. Treatments (2,24,26) included subcutaneous injections of progesterone (3 mg/kg/day for 5 days), prolactin (2 mg/day for 5 days), or prolactin (5 days) followed by progesterone (5 days). For group 2 animals, total RNA was isolated from individual endometrial samples in Tri-Reagent according to Chomczynski and Sacchi (27). RNA concentrations for all samples were determined spectrophotometrically (A 260 ). Electrophoretic fractionation and ethidium bromide staining confirmed the integrity of each sample. Two animals in the progesterone treatment category were used for subcellular fractionation of uterine endometrium and for nuclear isolation.
Competitive RT-PCR-The CLONTECH MIMIC system was used to develop a heterologous competitor RNA. Competitor construction required two PCR amplification reactions. In the primary reaction, composite primers consisting of RFBP-specific sequence contiguous with 20 bp of MIMIC-specific sequence were designed to hybridize to opposite strands of the MIMIC DNA template, a 576-bp BamHI/EcoRI fragment of the v-erbB gene. For each 50-l PCR reaction, MIMIC DNA template This secondary PCR reaction was performed in a 100-l PCR reaction with TaKaRa ExTaq DNA polymerase (3 units/100 l), TaqStart antibody (0.66 g/100 l), deoxynucleoside triphosphates (0.2 mM each), and 0.4 M each RFBP-specific primer. The RFBP-specific primers were forward, 5Ј-GTG CGT GGA CTC CCT ATG CTG TTT CCC AG-3Ј, and reverse, 5Ј-GGA GAG AGT CAA GAT GCT CCT GTC GTT GG-3Ј. A total of 18 PCR cycles were performed as described above. The quality and size of the final reaction product was again verified by gel electrophoresis. The PCR product was subcloned into pCR®II-TOPO, and closed circular DNA was purified by equilibrium centrifugation in CsClethidium bromide gradients. The identity of the MIMIC competitor was confirmed by sequencing in both directions by the dideoxy chain termination method.
The pCR®II-TOPO vector containing the MIMIC competitor cDNA insert was linearized with the restriction enzyme EcoRV, and competitor RNA was transcribed. RNA was precipitated with isopropanol to prevent the coprecipitation of free nucleotides, resulting in a more accurate estimation of the RNA yield at 260 nm. Serial dilutions of competitor RNA (6, 3, 1.5, 0.6, 0.3, and 0.1 pg) were combined with poly(A) ϩ RNA (1 ng) from HRE-H9 cells and reverse-transcribed with Moloney murine leukemia virus and random hexamers for 30 min at 42°C. Products from this reaction were heated to 99°C and cooled to 4°C. Then 40 l of a PCR mix containing Amplitaq gold DNA polymerase (1 units/50 l) and 0.25 M each RFBP-specific primer described above were added to each reaction tube. PCR cycle parameters included a hot start at 95°C for 2 min, 36 cycles of 45 s at 94°C, 45 s at 67°C, and 60 s at 72°C. Samples were subjected to a final extension at 72°C for 5 min before rapid cooling to 4°C.
Amplified products were quantified by means of a binary gradient HPLC system. Briefly, an aliquot (10 l) of each RT-PCR reaction was injected onto a DNASep column containing an alkylated polystyrenedivinylbenzene packing and eluted in a 4-min gradient of acetonitrile in 0.1 M triethylammonium acetate (pH 7.0) at a flow rate of 1 ml/min at room temperature. The amounts of native (166-bp) and competitor (258-bp) products were determined by on-line UV absorbance detection at 254 nm. This titration analysis was repeated 5 times to show the preservation of slope and linearity in the quantification of the native product amounts. Next, a single dose of competitor was mixed with a fixed amount of poly(A) ϩ RNA from the HRE-H9 cells. Estimation of the amount of native transcript gave a value similar to that from the titration analysis, confirming the precision of the single-tube measurement with the denaturing HPLC quantification system. The repeated analysis (n ϭ 10) of RNA from HRE-H9 cells yielded a coefficient of variation of 8.4%. The single tube analysis was then used to determine the relative expression of native RFBP in rabbit tissues.

RESULTS
To test the hypothesis that the RING domain is involved in unique protein-protein interactions, the region of the RUSH cDNA that encodes the RING finger motif was obtained by PCR with primers that contained unique restriction enzyme sites. A single 142-bp PCR product was subcloned into pCR TM II, excised with appropriate restriction enzymes, and directionally subcloned into SacI-HindIII sites of the pSCREEN-1b(ϩ) expression vector. Insert orientation was confirmed by sequencing in both directions by the dideoxy chain termination method. The resultant recombinant protein with a T7 tag, a His tag, and an S tag at its NH 2 terminus was used in a novel cloning strategy (19) to identify protein binding partners. Filters with recombinant proteins from a gt-11 expression library prepared from HRE-H9 cells were denatured-renatured with guanidine hydrochloride to ensure proper protein folding (20,21). The filters were then screened with the recombinant RING finger protein in the presence of ZnSO 4 (10 M) to maintain functional conformation (22,23). The single phage clone (1.6-kb cDNA insert), depicted in Fig. 1, was identified with the T7-tag antibody alkaline phosphatase conjugate. Because the His tag is known to bind heavy metals, it was removed along with the T7 tag by enzymatic cleavage with thrombin. The truncated recombinant protein with its S tag was then used with the S-protein alkaline phosphatase conjugate to confirm the identity of the positive clone in a secondary screen.
The cDNA insert in the phage clone was amplified by PCR and subcloned into pCR TM II. Sequence analysis revealed the presence of a partial cDNA that contained a single open reading frame and a TAA stop codon but lacked an initiator codon. Northern analysis of rabbit endometrial mRNA with this clone revealed a strong 5.3-kb band. As shown in Fig. 1, RACE was used to extend the 5Ј-end of the cDNA and to obtain the entire 3Ј-end of the cDNA. The 5Ј region, which encodes the initiator methionine, is missing. Repeated attempts to obtain this sequence using 5Ј-RACE and SMART RACE met with no success.
The composite cDNA sequence (4286 bp) and the predicted amino acid sequence (1107 amino acids) for RFBP (GenBank TM accession number AF236061) are shown in Fig. 2. The incomplete protein has a predicted molecular mass of 126 kDa. A computer search of sequence data bases (SWISS-PROT ϩ PIR ϩ PRF and GenBank TM ) was conducted with the BLAST 2.0.12 program at the National Center for Biotechnology Information. This search revealed the predicted RFBP protein is 50% similar and 34% identical to Type IV P-type ATPases from Bos taurus (protein ID AAD03352), Mus musculus (protein ID AAB18627), and Saccharomyces cerevisiae (protein ID L01795). RFBP and the partial cDNA sequence for human KIAA0956 (GenBank TM accession number AB023173) are 92% identical using the BLAST algorithm for pairwise DNA-DNA sequence alignment, whereas the amino acid sequences are 93% identical and 95% similar. Such comparisons indicate that KIAA0956 is the human homolog of RFBP. A position-specific iterated BLAST (PSI-BLAST) program (28), which is sensitive to weak but biologically relevant sequence similarities, identified 597 hits. All of the hits were ATPases, and 289 of them had statistically significant expectation (e) values greater than the 0.001 threshold.
Although the coding sequence for RFBP is incomplete by ϳ18 amino acids, the ClustalW alignment (29) with known Type IV P-type ATPases indicates that it has all of the diagnostic structural features including consensus sequences for the conformationally flexible loop, seven of eight core segments, the ATP binding domain, the highly conserved phosphorylation site (DKTGT(L/I)T), and nine transmembrane domains potentially involved in substrate binding and translocation. Transmem- An Atypical Nuclear P-type ATPase Is a RFBP 3645 note that region D, which contains transmembrane domain 4, is absent from the RFBP cDNA sequence. However, to avoid confusion, the domains were numbered consecutively in Fig. 2.
All working models of membrane topology for P-type ATPases contain an even number of transmembrane passes, with four of them highly conserved at the NH 2 terminus (32,33). Thus, it was important to determine whether or not such an evolutionarily conserved domain that is missing from the cDNA was also missing from the gene. Primers that recognize cDNA sequence on either side of Region D in transmembrane domain 2 and core Region E were used in a PCR reaction with genomic DNA. Analysis of PCR reaction products by agarose gel electrophoresis is shown in Fig. 3A. Gel electrophoresis with MetaPhor agarose and ethidium bromide fluorescence allowed the visualization of a single 742-bp reaction product. Genomic sequence data provided in Fig. 3B has 100% identity with the corresponding region of the cDNA. These data show that region D, which contains transmembrane domain 4, is absent from the RFBP gene and confirm that RFBP is not a splice variant.
Western analysis with affinity-purified, antipeptide (IgG) antibodies prepared against RFBP amino acids 663-678 was used to identify a 128-kDa immunoreactive protein in nuclear fractions from endometrium (Fig. 4A). In vitro protein-protein interactions were verified by GST pull-down assays and immunoblotting. As shown in Fig. 4B, the RING domain peptide expressed as a GST fusion protein binds to native RFBP in a dose-dependent manner. Moreover, the RING domain peptide was able to bind RFBP isolated from nuclei (Fig. 4B). This observation indicated RFBP is the first Type IV P-type ATPase to be located in the nuclear membrane. Based on the ClustalW alignment, the hydrophilicity plot, and the immunoanalysis data, RFBP would span the nuclear membrane nine times leaving the NH 2 -and COOH-terminal ends on opposite sides of the bilayer. Each transmembrane domain is a 20 -30-amino acid residue coiled into a single ␣-helix that approximates the thickness of the polar region of the lipid bilayer (34). However, because the nuclear envelope is formed from two concentric membranes, it was not possible to deduce the membrane topology of RFBP from its primary structure alone. Because RFBP could be located in either the inner or the outer nuclear membrane, immunoelectron microscopy was used to show that RFBP is located in the inner nuclear membrane of nuclei isolated from rabbit endometrium (Fig. 5A). More specifically, immunogold labeling was used to show that the conformation-ally flexible loop extends into the nucleoplasm to contact euchromatin (Fig. 5B).
The extensive sequence identity of RFBP with known P-type ATPases from Plasmodium falciparum and B. taurus suggests the protein may be highly conserved. Therefore, competitive RT-PCR (Fig. 1) and an ion-pair reversed-phase HPLC product purification and detection system was used to show that RFBP is ubiquitous in its expression (Table I). The CLONTECH MIMIC system was used in the development of a heterologous competitor template. Native and mutant products were identified and quantified by denaturing HPLC, an analytical technique that preserves the accuracy of competitive RT-PCR beyond the log-linear phase of the reaction (2,(35)(36). A computer model of competitive PCR was used to show an 8-fold range in the amounts of RFBP, with some of the highest values recorded for organs of the reproductive tract.
Because RUSH, the RFBP binding partner, is hormonally regulated in the endometrium (2, 37), it was important to determine whether hormone-dependent changes in the amount FIG. 5. Immunolabeling of ultra-thin sections. Immunolabeling was initiated with affinity-purified anti-RFBP antibodies and goat antirabbit IgG/15 nm gold (panel A) or goat anti-rabbit IgG/6-nm gold (panel B). Immunolabeling (arrows) showed RFBP associated with the inner nuclear envelope (panel A, 38,700ϫ) and euchromatin (panel B, 46,400ϫ). Immunolabeling was excluded from nucleoli. Disrupted outer nuclear membranes (O) and intact inner nuclear membranes (I) are shown. Immunostaining was negligible in negative controls. When the outer membrane was selectively removed with Triton X-100, immunolabeling of the inner membrane and euchromatin was consistent with data depicted here. of RFBP message corresponded to a similar pattern of change in RUSH message expression. Quantification of competitive RT-PCR reactions by ion-pair reversed-phase HPLC was used to show that message expression is increased (p Ͻ 0.05) in response to treatment with progesterone (Fig. 6). Prolactin plus progesterone further increased (p Ͻ 0.05) the amount of message over the value for progesterone alone. Without progesterone, the inclusion of prolactin in the treatment protocol resulted in RFBP levels comparable with estrous controls. This indicates that there is tight regulation between the expression of the RING finger containing RUSH protein and its binding partner. DISCUSSION P-type ATPases comprise an evolutionarily diverse superfamily of nearly 200 ATP hydrolysis-driven ion pumps (33,38). Structurally related to hydrolases (39), P-type ATPases are thought to share a common catalytic mechanism with these enzymes. Because P-type ATPases show very little similarity to each other (Ͻ15%), they were originally described as P1-ATPases (heavy metal pumps), P2-ATPases (nonheavy metal pumps and phospholipid translocases), and P3-ATPases (bacterial K ϩ pumps). However, when arranged phylogenetically according to substrate specificity and sequence alignment, Ptype ATPases can be divided into five main groups. The groups are designated Type I (heavy metal pumps), Type II (Ca 2ϩ , Na ϩ /K ϩ , and H ϩ /K ϩ pumps), Type III (H ϩ and Mg 2ϩ pumps), Type IV (phospholipid pumps), and Type V (no assigned substrate specificity). Type IV, the most recently discovered group is found only in eukaryotes and appears to be very distinct from the metal ion transporters that dominate this enzyme class. The absence of Type IV from prokaryotes supports the hypothesis that P-type ATPases are evolving at a variable rate and that Type IV enzymes have evolved more recently than their counterparts (33).
All working models of membrane topology for P-type ATPases contain 8 or 10 clearly defined transmembrane passes with four highly conserved spans at the NH 2 terminus (32)(33)40). Collectively these transmembrane helices anchor the protein in the lipid bilayer, orient the hydrophilic domains of the protein at the membrane surface, and assemble into channellike structures to regulate transport. As shown in Fig. 7, the hydrophilic portions of the protein, which include its NH 2 and COOH termini and two loop segments, are located on the same side of the plasma membrane and therefore in the same subcellular compartment (41). This topology preferentially exposes the small, strongly hydrophilic loop (Regions A-C) between transmembrane domains 2 and 3 and the large conformation-ally flexible loop (Regions D-H) between transmembrane domains 4 and 5 to the cytoplasm. Regions A-C have been assigned a role in energy transduction that may be as simple as maintaining the stability of the ATPase structure. Transmembrane domain 4 is thought to be directly involved in energy transduction. The sequences that link transmembrane domain 4 to the phosphorylation site, i.e. regions D and E, are two of the three most highly conserved regions in ATPases. Mutations in this region cause a significant reduction in the transport capabilities of H ϩ -ATPase (42). The conformationally flexible loop that accounts for 45% of the polypeptide contains highly conserved phosphorylation and ATP-binding sites. The function of the phosphorylation site in all P-type ATPases is the same, i.e. the transfer of ␥-phosphate from ATP to the aspartate residue in the phosphorylation site. In fact, P-type ATPases were named for the participation of a high energy aspartyl-phosphoryl enzyme intermediate in their catalytic cycle (38,43). Phosphorylation of the aspartate (D) residue in the invariant sequence DKTGT(L/I)T distinguishes P-type ATPases from V-type and F 0 F 1 -ATPases.
The newly identified RFBP differs from all described previously Type IV P-type ATPases in three important ways. The most striking feature of this putative phospholipid pump is an odd number of transmembrane domains resulting from the absence of Region D and transmembrane domain 4. Because of the highly conserved nature of the NH 2 terminus, the absence of the fourth membrane-embedded domain alters both the orientation of the adjacent hydrophilic domains and the determinants of transport specificity. Deletion of this region alters the conformational dynamic that functionally links the transport site in the membrane with the site of ATP hydrolysis and strongly suggests that RFBP is not involved in phospholipid transport. In the larger scheme of evolutionary advantage, altered substrate specificity requires a change in structural constraint resulting from a dramatic change in primary structure (33). Such a change in substrate specificity is likely to result from the absence of Region D and the PEGL motif, which is considered to play a critical role in energy transduction (44).
Our first thought was that the endometrial RFBP might be the product of an alternative splice event. Once considered to be the exception to the rule (one gene, one protein), alternative splicing is known to regulate ϳ5% of all genes (45). It is an economical way to produce different products from the same gene (46 -47). A survey on intron and exon length by Hawkins (48) indicated that although introns in vertebrates vary in size from 40 to greater than 3000 nucleotides, the largest single size range is 80 -90 nucleotides. Region D and its included transmembrane domain 4 are encoded by 72 nucleotides that could easily participate in the simplest of alternative splicing events, i.e. splice/don't splice. However, sequences from genomic clones confirmed that Region D and transmembrane domain 4 are absent from the RFBP gene. These results support a working model for membrane topology (Fig. 7) in which the NH 2 and COOH termini of RFBP are located on opposite sides of the membrane.
The second, equally striking feature of RFBP is its exclusive presence in the nuclear membrane. Nicotera et al. (49) and Lanini et al. (50) identified a nuclear Type II P-type Ca 2ϩ -ATPase whose structure is that of a transporting pump. However, no Type IV P-type ATPases have been localized to the nuclear membrane. The use of immunoelectron microscopy to show RFBP is located in the inner nuclear membrane provides additional insights to its atypical structure. As shown in Fig. 7, the small hydrophilic loop (Regions A-C; 174 residues) extends into the perinuclear space that is contiguous with the lumen of the endoplasmic reticulum, and the large conformationally flexible loop (Regions E-H; 499 residues) extends into the nucleoplasm. Physical contact between the large loop and the euchromatin was confirmed with immunoelectron microscopy. A comparison (51)(52) of the three-dimensional map of the neurospora plasma membrane H ϩ -ATPase (53) with the map of the sarcoendoplasmic reticulum Ca 2ϩ -ATPase shows their transmembrane domains are similar, but their large loops differ substantially. The compact conformation of the Ca 2ϩ -ATPase contrasts with the relatively open conformation of the large cytoplasmic loop of the H ϩ -ATPase, which extends to a maximum height of 80 Å above the membrane surface and measures 85 ϫ 50 Å in cross-section (53). Such conformational differences can be attributed to major rigid body interdomain movements (52).
The third intriguing feature of RFBP is its ability to bind the RING finger, the common feature of a superfamily of nearly 200 otherwise nonhomologous proteins (4,7,14), many of which are transcription factors. Conventional approaches to understanding transcriptional regulation of target genes have been focused on the identification and characterization of promoter elements and their cognate binding proteins. However, it is biologically meaningful to investigate transcriptional activation and chromatin remodeling as interrelated processes within the context of nuclear architecture (54). Not only is intranuclear organization required for physiological control of normal gene expression, modifications in nuclear organization are linked to cancer and neurological disorders (55) and are likely to be linked to some forms of infertility. Multiple mechanisms are involved in the entry and trafficking of regulatory factors, and at least two signals are required, i.e. a nuclear localization signal and a nuclear matrix-targeting signal. Once inside the nucleus, proteins are targeted to unique, nonoverlapping sites where gene activation or suppression occurs. For example, gene regulatory factors (steroid receptors), chromatin-remodeling proteins (SWI/SNF), and processing factors (SC35) are found in discrete subnuclear foci, a unique subset of which is associated with RNA polymerase II (54,55). The punctate distribution of regulatory proteins, some of which are members of the RING finger family, has provided experimental opportunities for exploring the mechanism for compartmentalization within the nucleus (54). Moreover, RING domains are thought to be molecular scaffolds responsible for the assembly of protein complexes (9). Although the biochemical basis for the RFBP-RING-euchromatin interaction remains obscure, these physical affiliations support the speculation that RFBP is part of a mechanism that coordinates the spatial organization of regulatory proteins with RING domains within the nucleus.
The potential for hormones to regulate the expression of ATPases generally, and P-type ATPases specifically, remains relatively unexplored. A few studies consider the effects of hormones on ATPase activity. For example, the demonstration that physiologically relevant concentrations of estradiol increased the hydrolytic activity of rat cortical Ca 2ϩ -ATPase (56) supports the idea that this plasma membrane calcium pump is a nongenomic steroid target (56). Menkes-and Wilson-ATPases have been implicated in human disorders of copper homeostasis. In the case of the Menkes P-type ATPase, treatment of human breast carcinoma cells, PMC42, with a combination of estrogen, progesterone, and prolactin increased perinuclear (Golgi) and punctate (endosome) protein, as measured by indirect immunofluorescence (57). However, Northern analysis failed to show an effect of hormones on mRNA expression. By comparison, we used competitive RT-PCR and an ion-pair reversed-phase HPLC product purification and detection system to show that low abundance endometrial RFBP mRNA expres- FIG. 7. Comparative topology of a typical Type IV P-type ATPase and RFBP. The Type IV P-type ATPase is an integral component of the plasma membrane with both loops extending into the cytoplasm. Consensus sequences for the phosphorylation (P) site and the ATP binding site (ATP) are located in the conformationally flexible loop. Transmembrane domains are numbered 1-10, and Region D is highlighted. In contrast, RFBP is a component of the inner nuclear membrane, with the loops oriented on different sides of the membrane. Consensus sequences for the phosphorylation (P) site and the ATP binding site (ATP) are located in the conformationally flexible loop. Based on the cloning strategy, the conformationally flexible loop also contains the RING finger binding domain. The asterisk (*) designates the antigenic region that was identified by the Peptide Structure Program (Genetics Computer Group Software) and used for antipeptide antibody production. Although the fourth NH 2 -terminal transmembrane domain is missing, the domains have been numbered consecutively (1-9) to avoid confusion. sion is regulated by progesterone and that prolactin augments the progesterone-dependent increase in RFBP message. This is the first demonstration that hormones regulate the expression of a P-type ATPase in the uterus. The importance of ATPases in reproduction is underscored by the recent demonstration (58) that the ␣ isoform of the Na ϩ /K ϩ -ATPase plays a critical role in sperm motility and fertilization.
Only the Type IV P-type ATPase from B. taurus (protein ID AAD03352) has been authenticated at the biochemical level as a phospholipid pump (59). Sequence similarity between this known phospholipid transporter and RFBP supports the idea that RFBP may regulate nuclear phospholipid composition such that levels of intranuclear phospholipids are higher for active chromatin than for repressed chromatin (60). However, the unique structure of RFBP coupled to its ability to bind the RING domain and contact euchromatin supports the idea that it has a very different function. It is tempting to speculate that RFBP is part of a molecular mechanism that targets regulatory proteins, in this case transcription factors with RING finger motifs, to specific domains within the nucleus.