Nuclear Coactivator-62 kDa/Ski-interacting Protein Is a Nuclear Matrix-associated Coactivator That May Couple Vitamin D Receptor-mediated Transcription and RNA Splicing*

Nuclear coactivator-62 kDa/Ski-interacting protein (NCoA62/SKIP) is a putative vitamin D receptor (VDR) and nuclear receptor coactivator protein that is unrelated to other VDR coactivators such as those in the steroid receptor coactivator (SRC) family. The mechanism through which NCoA62/SKIP functions in VDR-activated transcription is unknown. In the present study, we identified a nuclear localization sequence in the COOH terminus of NCoA62/SKIP and showed that NCoA62/SKIP was targeted to nuclear matrix subdomains. Chromatin immunoprecipitation studies revealed that endogenous NCoA62/SKIP associated in a 1,25-dihydroxyvitamin D 3 -dependent manner with VDR target genes in ROS17/2.8 osteosarcoma cells. A cyclic pattern of promoter occupancy by VDR, SRC-1, and NCoA62/SKIP was observed, with NCoA62/SKIP enter-ing these promoter complexes after SRC-1. These studies provide strong support for the proposed role of NCoA62/SKIP as a VDR transcriptional coactivator, and they indicate that key mechanistic differences probably exist between NCoA62/SKIP and SRC coactivators. To explore potential

The vitamin D receptor (VDR) 1 is a nuclear receptor (NR) family member that mediates the biological actions of 1,25-dihydroxyvitamin D 3 (1,25-(OH) 2 D 3 ), the active hormone of the vitamin D endocrine system. VDR forms a heterodimer with retinoid X receptor (RXR) and binds to specific vitamin Dresponsive elements (VDREs) in the promoter region of target genes to regulate transcription by RNA polymerase II (1,2). Transcriptional activation through 1,25-(OH) 2 D 3 and VDR is enhanced by nuclear receptor coactivator proteins such as steroid receptor coactivators (SRCs) (3)(4)(5) and proteins of the vitamin D receptor-interacting protein (DRIP) complex (6,7). The SRCs and DRIPs utilize leucine-rich LXXLL motifs (8,9) to interact in a 1,25-(OH) 2 D 3 -dependent manner with a complementary hydrophobic cleft on the surface of the VDR ligand binding domain. This hydrophobic surface is composed of helices H3, H4, H5, and H12, and it is the ligand-dependent folding of H12 that creates the interaction cleft for SRC binding to NRs (10 -15). The H12 helix of the NRs contains the critical ligand-dependent activation function-2 domain that is essential for ligand activated transcription (16,17) and ligand-dependent SRC and DRIP interaction (3,6).
In addition to SRCs and DRIPs, other VDR-interacting proteins have been implicated as positive comodulator proteins in this process. Nuclear coactivator-62 kDa (NCoA62) is one such coactivator protein that was isolated as a VDR-interacting protein in a yeast two-hybrid screen (18). NCoA62 was independently isolated as a protein that interacts with the v-Ski oncoprotein, and it was termed Ski-interacting protein or SKIP (19). NCoA62/SKIP exhibits properties consistent with a NR coactivator in that it interacts directly with VDR and other NRs, and it enhances 1,25-(OH) 2 D 3 -, retinoic acid-, estrogen-, and glucocorticoid-activated transcription in transient expression assays. However, important differences exist between NCoA62/SKIP and SRCs. For example, NCoA62/SKIP lacks LXXLL type motifs, it interacts with VDR in an activation function-2-independent manner (18), and it interacts with VDR through domains that are distinct from the H3-5/H12 surface cleft (20). NCoA62/SKIP also interacts preferentially with the VDR-RXR heterodimer compared with VDR monomer or VDR homodimer (20).
The mechanism through which NCoA62/SKIP functions to augment VDR-mediated transcription is not well understood, but it probably involves cooperation with other coactivators. For example, NCoA62/SKIP, VDR, and SRC form a ternary complex in vitro, and NCoA62/SKIP and SRCs synergize to augment VDR-activated transcription in transient reporter gene assays (20). Studies using dominant negative inhibitors of each coactivator suggest that both NCoA62/SKIP and SRCs are needed for optimal VDR-mediated transcription in cell culture (20). Whereas these initial in vitro binding and heterologous expression studies strongly implicate a role for NCoA62/SKIP in nuclear receptor-mediated transcription, data showing that NCoA62/SKIP exists in functionally relevant transcription complexes on native vitamin D-responsive target gene promoters have not been reported.
Additional clues to the mechanism of NCoA62/SKIP in VDRand NR-mediated transcription comes from studies directed at characterizing protein components of the spliceosome as well as interchromatin granules, where many splicing factors reside (21)(22)(23). These proteomic approaches showed NCoA62/SKIP association with the splicing machinery. In addition, the Schizosaccharomyces pombe ortholog of NCoA62/SKIP was shown to interact with the small subunit of the essential splicing factor U2AF (spU2AF 23 ) (24). The identification of NCoA62/SKIP as a transcriptional coactivator and as a component of the spliceosome suggest that NCoA62/SKIP may function as a coupling factor that links NR-dependent transcription to RNA processing. This hypothesis complements recent studies indicating a tight coupling between NR-mediated transcription and RNA splicing (25,26) as well as the recent proposal that NR coactivators, including NCoA62/SKIP, may play integral roles in coupling NR-activated transcription to RNA splicing (25).
Accumulating evidence points to a close relationship between transcription regulation, RNA splicing, and the nuclear matrix (NM). The nuclear matrix is a complex network of proteins that is thought to provide a structural framework for organizing chromatin and for maintaining the overall size and shape of the nucleus (27). Beyond this critical architectural role, the nuclear matrix is emerging as a functional scaffold that may anchor components of diverse nuclear processes including DNA replication, RNA splicing, and RNA polymerase II-mediated transcription. For example, protein components representing sites of replication, transcription, and splicing as well as nascent RNA remain attached to the nuclear matrix following in situ extraction procedures (28 -31). Sites of active transcription are often associated with RNA splicing factors within discrete nuclear subdomains (32). Direct evidence for the importance of nuclear matrix targeting in transcription regulation is found in the Runx2/Cbfa1 osteoblast transcription factor. A nuclear matrix targeting signal within the Cbfa1 protein directs it to nuclear matrix-associated foci, and this targeting event is required for transactivation of the osteoblastspecific osteocalcin gene (33,34). Whereas the significance of the nuclear matrix in these nuclear processes is the subject of continued investigation, these observations support an important role for this scaffold in transcription and splicing events in the cell.
In this study, the transcriptional coactivator properties of NCoA62/SKIP and its potential interplay with the splicing machinery in the nucleus were investigated. We identified a nuclear localization signal at the C terminus of NCoA62/SKIP and show that nuclear NCoA62/SKIP is targeted to the nuclear matrix. Using chromatin immunoprecipitation strategies, we show that endogenous NCoA62/SKIP selectively interacts in a 1,25-(OH) 2 D 3 -induced manner with the vitamin D-responsive regions of native osteoblast promoters. It becomes associated with these complexes after VDR and SRC entry, suggesting it may function at more distal steps or through different mechanisms than the SRCs. We also provide evidence that NCoA62/ SKIP interacts predominantly with RNA splicing factors as well as with putative components of the nuclear matrix. Finally, a dominant negative NCoA62/SKIP selectively blocks splicing of a growth hormone minigene cassette, thus providing functional evidence that NCoA62/SKIP is needed for appropriate splicing in VDR-activated gene expression. Based on these studies, we propose that this dual functional role for NCoA62/ SKIP places it at the forefront of proteins that may serve to link transcription regulation by VDR and other NRs to RNA splicing and the spliceosome.

MATERIALS AND METHODS
Plasmid Constructs-A green fluorescent protein (GFP) fusion with the COOH terminus of NCoA62/SKIP (GFP-NCoA62 531-536) was constructed in the pEGFP-Cl vector (Clontech). Complementary oligonucleotides encoding the last 6 residues of hNCoA62/SKIP and the stop codon were annealed and cloned into the EcoRI and BamHI sites of pEGFP-Cl. The following primers were used, 5Ј-AATTCCAAGAAGAG-GAGGAAGGAATAGG-3Ј and 5Ј-GATCCCTATTCCTTCCTCCTCTTC-TTGG-3Ј. The construct was verified by DNA sequencing. To generate HA-tagged VDR, human VDR cDNA (encoding residues 4 -427) was ligated into EcoRI/EcoRV-digested pCDNA 3.1/HA-B (Invitrogen). GST-NCoA62 157-536 was expressed in the DH5-␣ strain of Escherichia coli and purified by glutathione-agarose affinity chromatography as described previously (20). pcDNA-HA-p53 and CMX-FLAG-14-3-3 were obtained from Dr. Y. C Yang and Dr. H. Y. Kao, respectively. The VDRE 4 TKGH minigene construct contains the human growth hormone gene under the control of the thymidine kinase promoter and four copies of a VDRE from the rat osteocalcin gene and was described previously (35). pSG5 (Stratagene) expression vectors for hVDR, hNCoA62/SKIP, dnNCoA62 87-342, and dnGRIP were described previously (20). Deletion fragments of NCoA62/SKIP were generated by polymerase chain reaction using SG5-NCoA62/SKIP as a template. Primers with unique restriction site sequences (EcoRI and BamHI) were used to amplify selected regions of the NCoA62/SKIP using Vent polymerase, and these were cloned into the CMV-FLAG expression plasmid (Sigma). All constructs were verified by DNA sequencing.
Cell Culture and Transfections-COS-7 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 5% charcoalstripped, heat-inactivated calf bovine serum for 4 days prior to transfection. COS-7 cells were transfected by standard calcium phosphate precipitation procedures as described previously (36). SaOS-2 cells were plated on 0.5% gelatin-coated coverslips in 6-well tissue culture plates. Cells were grown for ϳ24 h after plating in McCoy's 5A supplemented with 5% fetal bovine serum. ROS 17/2.8 cells were seeded in 6-well tissue culture plates at a density of 8 ϫ 10 4 cells/well. Cells were grown for ϳ24 h after plating in Ham's F-12 medium containing 5% fetal bovine serum.
Immunofluorescence Studies-Four-well Lab-Tek chamber slides (Nalge Nunc International, Rochester, NY) were used for immunostaining. Slides were pretreated with poly-L-lysine to promote cell attachment. Each chamber received 25,000 COS-7 cells in 0.3 ml of medium. Cells were transfected overnight with the indicated plasmids using calcium phosphate precipitation. Monolayers were washed with phosphate-buffered saline (PBS), medium was replaced, and proteins were allowed to express for an additional 24 h. Following a 10-min fixation in 4% paraformaldehyde, the cells were blocked and permeabilized using PBS containing 3% bovine serum albumin and 0.3% Triton X-100. Cells were incubated with primary antibody against the FLAG epitope (10 g/ml M5 antibody from Sigma) or against the HA epitope (20 g/ml anti-HA tag antibody from Upstate Biotechnology, Inc., Lake Placid, NY) for 2 h at room temperature. The cells were washed with PBS containing 0.3% Triton X-100 and further incubated with a 4 g/ml concentration of the appropriate fluorochrome-conjugated species-specific secondary antibody Alexa-Fluor 488 or 594 (Molecular Probes, Inc., Eugene, OR) for 1.5 h in the dark. Finally, cells were washed with PBS and mounted with ProLong Antifade reagent (Molecular Probes). Cell staining was visualized using fluorescence microscopy or confocal microscopy. Fluorescence micrographs were obtained with a Nikon (Tokyo, Japan) FX-Microphot microscope, and confocal images were from a Zeiss (Oberkochen, Germany) LSM 410 confocal laser-scanning microscope using an argon-krypton laser (excitation lines, 488 and 568 nm) and a ϫ 100 Plan-Neofluar oil objective. Confocal sections (0.5 m in thickness) were taken to localize cellular staining in several planes. To standardize the collection of images, z axis heights of cells were measured, and optical slices were collected through the middle of cells. After adjusting the appropriate exposure times for the brightest sam-ples, all images of cells were collected using the same magnification and exposure times.
Nuclear Matrix Localization of NCoA62/SKIP-Cells were grown on 0.5% gelatin-coated coverslips (for in situ immunofluorescence) or in 100-mm tissue culture plates (for biochemical fractionation). For whole cell preparations, cells were immediately fixed in 3.7% formaldehyde (for in situ immunofluorescence) or resuspended in SDS-PAGE loading buffer (for biochemical fractionations). For cytoskeletal (CSK) preparations, cells were extracted twice for 15 min on ice with CSK buffer (10 mM Pipes, pH 6.8, 300 mM sucrose, 100 mM NaCl, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100, 2 mM vanadyl ribonucleoside complex) and then fixed or solubilized in SDS-PAGE buffer as described above. For nuclear matrix intermediate filament (NMIF) preparations, following CSK extraction, cells were digested by the addition of digestion buffer (10 mM Pipes, pH 6.8, 300 mM sucrose, 50 mM NaCl, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100) containing 50 units/ml RNase-free DNase I (Roche Applied Science). After 30 min, chromatin and cytoskeletal proteins were removed by a 10-min incubation in digestion buffer supplemented with 0.25 M ammonium sulfate. This solution was removed and replaced with a fresh aliquot for an additional 15 min. Cells were either fixed or solubilized in SDS-PAGE buffer for in situ immunofluorescence or Western blot analysis, respectively.
Western Blots-Proteins were separated on 12% SDS-PAGE gels and transferred to polyvinylidene difluoride membrane followed by Western blot analysis. In brief, 5% milk in PBS containing 0.1% Tween 20 was used to block nonspecific binding. The blot was subsequently incubated with VDR antibody (anti-VDR C-20 rabbit polyclonal 1:2000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or anti-NCoA62 rabbit polyclonal antibody (1:10,000 dilution) and with the secondary antibody (peroxidase anti-rabbit 1:5000; Santa Cruz Biotechnology). After each antibody incubation, blots were extensively washed in PBS containing 0.1% Tween 20. For detection, the ECL kit (Amersham Biosciences) was used according to the directions of the manufacturer.
In Situ Immunofluorescence for the Nuclear Matrix Studies-Detergent-extracted cell preparations and nuclear matrix intermediate filament preparations were incubated with PBS containing 0.5% bovine serum albumin to block nonspecific antibody binding. Anti-NCoA62 antibody was used at a 1:250 dilution and goat anti-rabbit IgG conjugated to fluorescein was used as the secondary antibody (Jackson Im-munoResearch; 1:300 dilution). After antibody incubations, cells were incubated with a 0.05 mg/ml concentration of the DNA counterstain 4Ј,6-diamidinophenylindole. Antibody incubations, washes, and DNA counterstaining were performed as described previously (37). Microscopic analysis was performed by using a Zeiss Axioplan 2 microscope and a charge-coupled device camera interfaced with the MetaMorph Imaging System (Universal Imaging, Media, PA).
Isolation of GST-NCoA62/SKIP-associated Proteins from HeLa Cell Nuclear Extracts-The GST and GST-NCoA62/SKIP fusion proteins were expressed in E. coli BL21 (DE3) and purified with glutathioneagarose as described (20). GST-NCoA62/SKIP was further purified by on a HiLoad Superdex 200 (16/60) column (Amersham Biosciences) equilibrated in 20 mM Tris-Cl, pH 7.9, 400 mM KCl, 0.5 mM EDTA, and 0.5 mM DTT. HeLa cell nuclear extracts (10 mg/ml), prepared by the method of Dignam (38), were kindly provided by Dr. C. M. Chiang. Twenty-five g of GST-NCoA62/SKIP or GST were incubated overnight at 4°C with 20 l of 50% glutathione-agarose in PBS in the presence or absence of 500 g of HeLa nuclear proteins in a final volume of 110 l in incubation buffer (20 mM Tris-HCl at pH 7.6, 50 mM KCl, 40 mM NaCl, 0.1 mM EDTA at pH 8.0, 0.1% Nonidet P-40, 10% glycerol, 1.5 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride, and 1ϫ protease inhibitor mixture (Sigma)). Subsequently, the beads were washed three times with 0.5 ml of washing buffer (20 mM Tris-HCl at pH 7.6, 50 mM NaCl, 0.2% Nonidet P-40). The beads were boiled in 80 l of SDS-PAGE loading buffer, and each sample was analyzed on a 10% SDS-polyacrylamide gel. Proteins were visualized with Coomassie Brilliant Blue or silver staining. The distinct proteins from HeLa cell nuclear extracts that specifically associated with GST-NCoA62/SKIP were isolated from the gel and subject to protein sequencing for identification by mass spectrometry at The Cleveland Clinic Foundation.
Mass Spectrometry-For the protein digestion, the bands were excised from a Coomassie Brilliant Blue-stained gel and washed/ destained in two aliquots of 50% ethanol, 5% acetic acid for 1 h each. The gel pieces were washed in 0.1 M ammonium bicarbonate, dehydrated in acetonitrile, reduced with DTT, and alkylated with iodoacetamide. The gel pieces were then dehydrated in acetonitrile, washed in 0.1 M ammonium bicarbonate, dehydrated again in acetonitrile, and dried in a Speed-Vac. In-gel trypsin proteolysis was overnight at room temperature. The peptides were extracted from the polyacrylamide in 50% acetonitrile, 5% formic acid and evaporated to Ͻ20 l for liquid chromatography-mass spectrometry analysis. A Finnigan LCQ-Deca ion trap mass spectrometer system with a Protana microelectrospray ion source interfaced to a self-packed 10 cm ϫ 75-m inner diameter Phenomenex Jupitor C18 reversed-phase capillary chromatography column was used for analysis. The data were analyzed by using all spectra collected in the experiment to search the NCBI nonredundant data base with the search program TurboSequest. All matching spectra were verified by manual interpretation. The interpretation process was also aided by additional searches using the programs Mascot and Fasta as needed. Some samples were also analyzed by matrix-assisted laser desorption/ionization (MALDI)-time-of-flight mass spectrometry. For these analyses, the digest was desalted using a Millipore C18 ZipTip column that was eluted in the MALDI matrix (5 mg/ml ␣-cyano-4hydroxycinnamic acid in 50% acetonitrile, 2.5% trifluoroacetic acid). Aliquots of the extract (1.5 l) were spotted on a sample plate and analyzed using a Micromass TofSpec 2E time-of-flight mass spectrometry system. Spectra were typically composed of the summed spectra of ϳ150 laser shots. The spectra were internally calibrated using trypsin autolysis peptides, giving mass accuracies that were generally better than 25 ppm.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assays were performed essentially as described (39) with several modifications. ROS 17/2.8 cells (five 150-mm plates for each time point) were cultured for 4 days prior to hormone addition in Dulbecco's modified Eagle's medium containing 5% charcoal-stripped, heat-inactivated calf bovine serum. After treatment of the cells with 1,25-(OH) 2 D 3 for the indicated times, 1% formaldehyde in cell culture medium was used to cross-link the cells for 10 min, and cross-linking was quenched with glycine. Cells were harvested and rinsed with PBS, and cell pellets were resuspended in 1 ml of cell lysis buffer (5 mM Pipes, pH 8.0, 85 mM KCl, 0.5% Nonidet P-40, 1 mM DTT, 0.25 mM phenylmethylsulfonyl fluoride, and 1 g/ml each of pepstatin, leupeptin, and aprotinin). Nuclei were collected and resuspended in 500 l of nuclear lysis buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 1% SDS, 1 mM DTT, 2.5 mM phenylmethylsulfonyl fluoride, and 1 g/ml each of pepstatin, leupeptin, and aprotinin). The chromatin samples were sonicated to yield DNA fragments between 300 and 2000 bp with an average length of about 500 bp. For each immunoprecipitation, 50 l of sheared chromatin was diluted to 1 ml with IP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.1, and 167 mM NaCl). Prior to chromatin immunoprecipitations, the chromatin solution was precleared with 80 l of protein A-Sepharose beads (Sigma) containing 240 g/ml of salmon sperm DNA and 240 g/ml of tRNA at 4°C for 30 min with rotation. The precleared chromatin was collected and incubated at 4°C overnight with 5 g of anti-VDR antibody 9A7 (Affinity Bioreagents Inc.), anti-SRC antibody (Affinity Bioreagents), anti-acetylated histone H4 antibody (Upstate Biotechnology), or a rabbit polyclonal antibody directed against a COOH-terminal peptide of hNCoA62/SKIP (antibody 57). Rabbit IgG (Sigma) was used as a negative control. The immune complexes were precipitated with 60 l of protein A-Sepharose beads (Sigma) at 4°C for 1 h. The supernatant from a control in the absence of antibody was used as the total chromatin input for later PCR analysis. The beads were then washed in the following order: 1 ml of IP dilution buffer, TSE-500 wash (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 500 mM NaCl), LiCl/detergent wash buffer (100 mM Tris-HCl, pH 8.1, 500 mM LiCl, 1% Nonidet P-40, 1% deoxycholic acid), and TE (10 mM Tris-HCl, pH 8.1, 1 mM EDTA, pH 8.0), twice in each buffer with a 10-min period between washes. The antibody-protein-DNA immunocomplexes were eluted twice with 250 l of elution buffer (1% SDS, 50 mM NaHCO 3 ). The sample was vortexed briefly and incubated at room temperature for 15 min with rotation. Formaldehyde cross-linking was reversed by heating at 65°C overnight with the addition of 5 M NaCl to a final concentration of 200 mM. All of the samples were digested at 45°C for 1 h with 10 g of RNase A and 20 g of proteinase K. The DNA were extracted by phenol-chloroform, ethanol-precipitated, and analyzed by PCR. The primer pair spanning Ϫ334 to Ϫ22, relative to the rat 24-OHase transcription start site, produced a 312-bp product. The primer sequences were 5Ј-TTGTGCAAGCGCAGCTTTGG-3Ј and 5Ј-CGTATTTATGGAAGCAGGGC-3Ј.
RNA Isolation and Analysis of Splicing by RT-PCR-COS-7 cells were grown in Dulbecco's modified Eagle's medium containing 5% of charcoal-stripped calf bovine serum in 100 mM plates. The cells were transfected by calcium phosphate precipitation with the indicated expression plasmids and the VDRE 4 TKGH growth hormone minigene cassette. The plates were treated with 10 Ϫ8 M 1,25-(OH) 2 D 3 for the indicated times, and total RNA was isolated using 1.2 ml of RNAzol B (Tel-Test) as recommended by the manufacturer. The RNA prepara-tions were DNase-digested to eliminate plasmid contamination, which would interfere with the RT-PCR analysis. The RNA (0.9 g) was incubated for 1 h at 37°C with RQ1 DNase (1 unit; Promega), AMV/TFl reaction buffer (Access RT-PCR System; Promega), and MgSO 4 (2.5 mM final concentration) followed by 20 min of DNase-heat inactivation at 70°C. This reaction was immediately used for RT-PCR using the Access RT-PCR system (Promega), following the manufacturer's instructions. The Access RT-PCR system was performed in a single tube and was used to avoid cross-contamination and variability. Forty cycles of PCR (30 s at 94°C, 1 min at 58°C, and 2 min at 68°C) were performed. The ratio between the different spliced products was not affected by performing 25-40 cycles (data not shown). Primers that spanned intron 1 of the GH minigene cassette were CAACTCCCCGAACCACTC and CTGGGGAGAAACCAGAGG. The RT-PCR products were separated by electrophoresis through 2% agarose gels and visualized by ethidium bromide staining.

RESULTS
Nuclear Localization of NCoA62/SKIP-NCoA62/SKIP is a nuclear protein that exhibits coactivator activity in VDR-and NR-mediated transcription (18). Using an affinity-purified polyclonal antibody directed against the COOH terminus of hNCoA62/SKIP, we observed that NCoA62/SKIP adopts a highly restricted nuclear expression profile (Fig. 1A). Even when it was overexpressed to high levels in COS-7 cells, cytoplasmic staining was not evident in this system. Native NCoA62/SKIP showed a similar nuclear restricted expression pattern in COS-7, ROS17/2.8, and HeLa cells (data not shown). Analysis of the primary sequence of hNCoA62 using the PSORT data base revealed two highly basic regions that could potentially serve as nuclear localization signals (Fig. 1B). One of these regions is in the NH 2 terminus (between residues 77 and 80), and another is in the extreme COOH terminus (between residues 531 and 536). To determine which, if either, of these putative nuclear localization signals was responsible for NCoA62/SKIP nuclear targeting, a series of FLAG-tagged NCoA62/SKIP deletion mutants were transiently expressed in COS-7 cells. Cells were immunostained with anti-FLAG antibody and examined using immunofluorescence and confocal microscopy. As shown in Fig 1B, elimination of the NH 2 -terminal half of NCoA62/SKIP, which contains the putative nuclear localization signal between residues 77 and 80 in the NCoA62/ SKIP 275-536 construct, resulted in a nuclear staining profile identical to intact, wild type NCoA62/SKIP. In contrast, elimination of the COOH-terminal half of the NCoA62/SKIP protein (NCoA62/SKIP 1-275) resulted in predominantly cytoplasmic localization. Indeed, smaller COOH-terminal deletions were similarly mislocalized to the cytoplasm, with little or no evidence of nuclear staining (NCoA62/SKIP 1-488, 1-513, and 1-530). The NCoA62/SKIP 1-530 mutant lacks only the last 6 residues from the basic region at the extreme COOH terminus, and this resulted in exclusive cytoplasmic staining (Fig. 1C). Thus, residues 531-536 were necessary for targeting NCoA62/ SKIP to the nucleus. To test whether these residues alone were sufficient for directing localization of a heterologous protein to FIG. 1. Identification of a nuclear localization sequence in NCoA62/SKIP. A, nuclear localization of FLAG-tagged NCoA62/SKIP. CMV-FLAG-NCoA62 was transfected into COS-7 cells. The cells were fixed and stained for FLAG expression using the anti-FLAG M2 antibody and a fluorescein-conjugated secondary antibody as described under "Materials and Methods." B, subcellular localization of various FLAG-tagged NCoA62/SKIP deletion mutants. Wild type NCoA62/SKIP 1-536 and the indicated deletion mutants were analyzed for nuclear (N) or cytoplasmic (C) localization as described for A. Putative nuclear localization sequences are indicated by black boxes. C, the extreme COOH terminus of NCoA62/SKIP is required for nuclear localization. Wild type, FLAG-tagged NCoA62/SKIP (left panel) or NCoA62/SKIP 1-530 (right panel) were expressed in COS-7 cells and visualized by confocal imaging as described for A. Note that removing those last six residues of NCoA62/SKIP resulted in an exclusively cytoplasmic expression pattern. D, the last six residues of NCoA62/SKIP are sufficient to target GFP to the nucleus. GFP and a GFP fusion with residues 531-536 of NCoA62/SKIP were transiently expressed in COS-7 cells. Subcellular localization was examined using confocal microscopy. the nucleus, residues 531-536 were fused to GFP. As is evident in Fig. 1D (right panel), the GFP-NCoA62 531-536 fusion was localized predominantly in the nucleus, whereas GFP was localized in both the nuclear and cytoplasmic compartments (Fig.  1D, left panel). Collectively, the data in Fig. 1 identify residues 531 KKRRKE 536 of NCoA62/SKIP as a nuclear localization sequence that is both necessary and sufficient to target NCoA62/ SKIP to the nuclear compartment of cells.
Colocalization of NCoA62 and VDR in the Nucleus-In vitro assays indicate that NCoA62/SKIP interacts with VDR (18,20). However, it is unknown whether they form a complex in intact mammalian cells. As shown in Fig. 2A, confocal images of cells expressing wild-type FLAG-NCoA62/SKIP and HA-VDR indicate that both proteins are nuclear and that the majority of these expressed proteins colocalize in COS-7 cells. To extend this observation, coexpression studies with VDR and the NCoA62/SKIP 1-530 nuclear localization signal mutant were performed. As illustrated in Fig. 2B, the NCoA62/SKIP 1-530 mutant is expressed predominantly in the cytoplasm in cells that lack VDR. The arrow in the left panel of Fig. 2B indicates a cell in which the mutant NCoA62/SKIP 1-530 was localized in the cytoplasm in the absence of detectable HA-VDR expression (middle panel, Fig. 2B). Importantly, in cells that coexpressed the NCoA62/SKIP 1-530 mutant together with VDR (i.e. the remaining three cells in the representative field in Fig.  2B), the NCoA62/SKIP 1-530 mutant adopted a predominantly nuclear expression profile. This shuttling of the NCoA62/SKIP 1-530 mutant to the nucleus was selective for VDR expression, since expression of another nuclear protein, HA-P53, did not alter the cytoplasmic localization of NCoA62/SKIP 1-530 (middle row in Fig. 2C). Likewise, the expression of VDR did not alter the cytoplasmic localization of a FLAG-tagged 14-3-3 protein in this system (bottom row in Fig. 2C). Table I summarizes these coexpression studies. This quantitation indicated that 88% of cells expressing NCoA62 1-530 showed exclusive cytoplasmic staining, whereas 77% of cells expressing both NCoA62 1-530 and VDR showed nuclear staining of the NCoA62/SKIP derivative. Thus, these data provide strong evidence that VDR interacts with NCoA62/SKIP in intact cells and that this interaction with VDR recruits the cytoplasmic derivative of NCoA62/SKIP to the nuclear compartment.
Nuclear Matrix Localization of NCoA62/SKIP-During the course of these localization studies, we noted that native NCoA62/SKIP adopts a fine punctate subnuclear expression pattern in COS-7 cells as well as in several osteoblast cell lines (see Fig. 4 and data not shown). This suggested that NCoA62/ SKIP may associate with structural components within the nucleus. Therefore, biochemical fractionation studies were initiated to determine whether native NCoA62/SKIP copurified with nuclear matrix fractions. ROS17/2.8 rat osteosarcoma cells were subjected to nuclear matrix extraction protocols as described under "Materials and Methods." Individual extraction steps were analyzed by Western blots for the presence of NCoA62/SKIP (Fig. 3, left panel) or for VDR (Fig. 3, right  panel). As shown in Fig. 3 (left panel), NCoA62/SKIP was readily detected in ROS17/2.8 whole cell extracts (lane 1). Native NCoA62/SKIP was highly resistant to the extraction protocol, with only modest levels being present in the individual nuclear matrix extraction steps (lanes 2-5). The majority of native NCoA62/SKIP in ROS 17/2.8 cells was retained in the final nuclear matrix intermediate filament fraction (lane 6), indicating a tight association of NCoA62/SKIP with the nuclear matrix. In contrast, native VDR in ROS 17/2.8 cells was readily extracted in the early steps (Fig. 3, lanes 8 -10), with minor levels detected in the NMIF (lane 12). Thus, whereas NCoA62/ SKIP and VDR interacted in intact cells, they showed remark-ably distinct profiles with respect to nuclear matrix association in ROS 17/2.8 osteosarcoma cells. Similar tight association of NCoA62/SKIP with the nuclear matrix was observed in a variety of other cell lines including SaOS-2 cells and primary rat osteoblast cultures, although the percentage of NCoA62/SKIP in the NMIF varied from ϳ50 to 90% of the total detectable NCoA62/SKIP in the different cell lines.
To support these biochemical fractionation studies, in situ extraction and immunofluorescence studies were conducted in SaOS-2 cells. These cells were chosen because they are well preserved during the in situ extraction procedures, they stain intensely for endogenous nuclear NCoA62/SKIP, and over 50% of the NCoA62/SKIP remained associated with the NMIF in biochemical fractionation studies (data not shown). As shown in Fig. 4, endogenous NCoA62 in native SaOS-2 cells was

FIG. 2. Colocalization and interaction of NCoA62/SKIP and VDR in the nucleus.
A, colocalization of FLAG-NCoA62/SKIP with HA-tagged VDR in COS-7 cells. Immunostaining was as described under "Materials and Methods," and stained cells were visualized using confocal microscopy. B, VDR expression alters the cellular distribution of NCoA62/SKIP 1-530. FLAG-NCoA62/SKIP 1-530 was cotransfected with HA-tagged VDR in COS-7 cells, and the cells were processed as in A. The arrow denotes a cell expressing the NCoA62/SKIP 1-530 without VDR. Note its exclusive cytoplasmic staining as in A. The other three cells in this field express the mutant NCoA62/SKIP as well as HA-tagged VDR. Note the exclusive nuclear staining of the NCoA62/ SKIP 1-530 mutant in cells that express nuclear VDR. C, HA-VDR selectively targets FLAG-NCoA62/SKIP 1-530 into nucleus. FLAG-14-3-3 and HA-p53 were used as controls. Subcellular localization was examined with a confocal microscope. FLAG-tagged proteins are shown in green, HA-tagged proteins are shown in red, and merged images are shown in yellow. localized in punctate, finely distributed subnuclear foci (WC in Fig. 4, second row). These punctate foci remained associated to the final NMIF stage, where soluble cellular proteins, chromatin, DNA, and cytoskeletal proteins have been removed. Thus, the in situ extraction and biochemical extraction studies both indicated significant association of NCoA62/SKIP with the nuclear matrix.
Chromatin Immunoprecipitation Assays Identify NCoA62/ SKIP on Native Vitamin D-responsive Promoters in ROS17/2.8 Osteoblasts-The studies above indicate that NCoA62/SKIP and VDR interact in intact cells and that NCoA62/SKIP associates with the nuclear matrix. Accumulating evidence indicates that other promoter-binding transcriptional regulatory proteins are targeted to the nuclear matrix as well (34). Whereas previous in vitro interaction and cellular expression studies provide strong evidence that nuclear NCoA62/SKIP functions as an NR transcriptional coactivator, it is currently unknown whether endogenous NCoA62/SKIP assembles into relevant transcriptional complexes on native VDR-responsive promoters in target cells. Therefore, ChIP studies were initiated to monitor the presence and temporal entry of NCoA62/ SKIP onto the 1,25-(OH) 2 D 3 -responsive region of the native 24-hydroxylase promoter in ROS 17/2.8 osteosarcoma cells. These cells were chosen because they express a high level of VDR (Ͼ20,000 copies/cell) and they respond to 1,25-(OH) 2 D 3 by increasing the transcription of VDR target genes such as 24hydroxylase. In initial experiments, ROS 17/2.8 cells were treated with 10 Ϫ8 M 1,25-(OH) 2 D 3 for 1 h and subjected to ChIP assays as described under "Materials and Methods." Crosslinked extracts were immunoprecipitated with antibodies directed against VDR, SRC-1, NCoA62/SKIP, acetylated histone H4, or rabbit IgG. Following reversal of the cross-links, DNA was recovered and analyzed by PCR using primers designed to amplify the VDR-responsive region of the rat 24-hydroxylase promoter (primer set A) or a distal upstream, nonresponsive region (primer set B) as a control to demonstrate response element selectivity. As shown in Fig. 5A (lanes 8 -10), ChIP analysis clearly reflected a 1,25-(OH) 2 D 3 -dependent increase in the association of endogenous VDR, SRC-1, and NCoA62/SKIP with the VDREs in the native rat 24-hydroxylase promoter following a 1-h treatment with 1,25-(OH) 2 D 3 (compare upper panel in the absence of 1,25-(OH) 2 D 3 with the lower panel in the presence of 1,25-(OH) 2 D 3 ). The protein-DNA association was specific for the region containing the VDREs as evidenced by the lack of detectable product using a distal upstream primer pair to which VDR and associated factors would not be targeted (Fig. 5A, lanes 1-6). Moreover, the nonspecific rabbit IgG control (lanes 6 and 12) did not result in the amplification of either sequence. In addition to VDR and its associated factors, the acetylated state of histone H4 was also selectively enhanced in the VDRE-containing region of the target promoter following this 1-h treatment with 1,25-(OH) 2

D 3 (lane 11).
Having established that VDR, SRC, and NCoA62/SKIP selectively associate with the VDRE region of the 24-hydroxylase promoter, an expanded time course was performed to monitor the 1,25-(OH) 2 D 3 -induced temporal entry of these factors onto the VDREs. As illustrated in Fig. 5B, a 1,25-(OH) 2 D 3 -depend-    2 and 8), anti-SRC (lanes 3 and 9), anti-NCoA62 (lanes 4 and 10), anti-acetylated histone H4 (Ac-H4 in lanes 5 and 11), or a nonspecific IgG (lanes 6 and 12). Lanes 1 and 7 are 2% chromatin input controls. The relative positions of the primer set A and B (arrows) are illustrated at the bottom. PCR products were separated on a 2% agarose gel and visualized after ethidium bromide staining. B, temporal entry of VDR, SRC-1, and NCoA62/SKIP onto the native rat 24-hydroxylase VDREs by ChIP assay. For each time point, a single lysate preparation was assayed for recruitment of each of different factors. Cross-linked chromatin isolated from ROS17/2.8 cells was immunoprecipitated with anti-VDR, anti-SRC, anti-NCoA62/SKIP, anti-Ac-H4, and nonspecific antibody IgG at 12 different time points as indicated after 1,25-(OH) 2 D 3 treatment. The associated chromosomal DNA fragments were amplified with the primer set A spanning Ϫ334 to Ϫ22, which gives rise to a 312-bp PCR product. PCR products were separated on a 2% agarose gel and visualized after ethidium bromide staining. ent association of VDR with the 24-hydroxylase VDREs was dramatically evident as early as 15 min following exposure to hormone. VDR association appeared to cycle in this assay, similar to previous observations in other NR systems (40). The peak VDR levels occurred at 15-, 75-, and 135-min time points. SRC-1 association was also evident at the 15-min time point, but maximal accumulation was delayed relative to VDR. SRC-1 also cycled, showing two waves with peaks at 45 min and ϳ2 h. Strikingly, 1,25-(OH) 2 D 3 -dependent entrance of NCoA62/SKIP onto this promoter was markedly delayed relative to both VDR and SRC-1. It also displayed two cycles; however its initial entrance was not evident until 45 min, with an initial peak value at 1 h. These data strongly suggest that SRCs and NCoA62/SKIP may function at different stages of promoter regulation and probably through different mechanisms of action.
NCoA62/SKIP Interacts with Components of the Splicing Machinery-To address potential mechanistic differences between NCoA62/SKIP and SRCs, we purified and identified nuclear proteins that interact with NCoA62/SKIP. HeLa cell nuclear extracts were incubated with highly purified GST-NCoA62, and protein-protein complexes were purified with glutathioneagarose. Proteins bound to the GST-NCoA62 resin were analyzed by SDS-PAGE, and individual proteins were identified by mass spectrometry. Fig. 6 is a representative silver-stained gel with the individual protein bands and their identities indicated in Table II. Of particular note is the intensely stained doublet above the 183-kDa marker. Each of these bands was identified as single protein species; the lower band of the doublet is Prp8, and the upper band is U5 snRNP 200-kDa helicase. Both of these factors are key components of the spliceosome and are critical for proper splicing of RNA transcripts. Other components of the spliceosome machinery identified in this approach were Prp28 (P100 band) and a 116-kDa U5 snRNP component (P115). The majority of peptides contained within the intense band migrating near the 114-kDa marker were identified as matrin 3, a putative nuclear matrix protein. We also identified tubulin ␤ 5 as a selective NCoA62/SKIP-interacting protein by this approach. These latter two proteins represent potential sites, perhaps within the nuclear matrix, for binding or targeting NCoA62/SKIP to the nuclear matrix.
A Dominant Negative NCoA62/SKIP Interferes with RNA Processing of 1,25-(OH) 2 D 3 -responsive Growth Hormone Minigene Cassette-The data in Fig. 6 support previous studies showing that NCoA62/SKIP is present in global spliceosome complexes (21,23) and that the Prp45 ortholog of NCoA62/ SKIP is involved in RNA splicing in yeast (68). However, there are no functional data yet available to support a role for NCoA62/SKIP in RNA splicing in mammalian cells. To explore this potential role, we expressed a dominant negative inhibitor of NCoA62/SKIP to determine whether it could interfere with RNA splicing in a mammalian cell culture system. The dominant negative NCoA62/SKIP contains the highly conserved SNW domain from residues 87-342, and it potently inhibits VDR-activated reporter gene expression (20). A VDRE-driven growth hormone reporter gene cassette (VDRE 4 TKGH) containing five exons and five introns of the genomic sequence for human GH was used in this study. This minigene cassette and a VDR expression vector were transfected into COS-7 cells with or without the dominant negative NCoA62/SKIP expression vector (SG5-NCoA62/SKIP 87-342). Various times after the addition of 1,25-(OH) 2 D 3 , RNA was extracted, and RT-PCR was performed with the indicated primers to determine the levels of spliced and unspliced GH transcripts (Fig. 7). All of the products in this assay were dependent on the inclusion of reverse transcriptase (data not shown). A time-dependent increase in spliced transcripts occurred following 1,25-(OH) 2 D 3 treatment (lanes 1-4). Detectable levels of unspliced transcripts were also evident. In some experiments, these appeared as a doublet, presumably due to an altered secondary structure of the intron that resulted in a smaller product in the reverse transcriptase reaction. These unspliced transcripts varied only slightly in response to 1,25-(OH) 2 D 3 treatment (lanes 1-4). Importantly, expression of dominant negative NCoA62/SKIP (NCoA62/SKIP 87-342) resulted in a 1,25-(OH) 2 D 3 -dependent transient accumulation of the unspliced transcripts (lanes 5-8), first detected at 1 h, and a maximal accumulation at 4 h following 1,25-(OH) 2 D 3 treatment. The accumulation of unspliced transcripts was not apparent using a dominant-negative GRIP (SRC-2) expression vector (lanes 9 -12) or using a similar NCoA62 derivative that lacks dominant negative activity (NCoA62 87-275 in lanes 13 and 14). Moreover, coexpression of intact NCoA62/ SKIP efficiently rescued the dominant negative NCoA62/SKIPinduced splicing defect in this assay (Fig. 8). Collectively, these studies strongly suggest that disrupting native NCoA62/SKIP through the expression of the dominant-negative NCoA62/ SKIP derivative interferes with the proper splicing of transcripts from a 1,25-(OH) 2 D 3 -inducible reporter gene in mammalian cells.

DISCUSSION
NCoA62/SKIP is a putative transcriptional coactivator for VDR that interacts selectively with the VDR-RXR heterodimer (20), with basal transcription factors such as TFIIB (41), and it augments VDR-mediated transcription in transient reporter gene expression studies (18,20). However, parallel studies also implicate NCoA62/SKIP in RNA splicing, since it was identified as a non-snRNP component of global spliceosome complexes as well as interchromatin granule complexes (21)(22)(23). Considerable evidence supports the concept of a functional coupling between RNA polymerase II-directed transcription and RNA processing events, including RNA capping, polyadenylation, and splicing (reviewed in Ref. 42). Putative candidates that may aid in the coupling process might be predicted  Table II. to associate with and alter the functional properties of proteins involved in both transcription and RNA processing. Indeed, one strong candidate is RNA polymerase II, which is known to recruit splicing factors through its carboxyl-terminal domain region (43,44). Recent studies have also indicated a potential coupling between transcriptional regulation by nuclear receptors and RNA splicing. For example, Auboeuf et al. (25) demonstrated that progesterone receptor-activated transcription influenced splicing decisions of alternatively spliced transcripts in a progesterone receptor and progesterone response element-dependent manner. They and others provided evidence implicating NR coactivators in this transcription-splicing coupling role (25,26). Our studies also support a dual role of the NCoA62/SKIP coactivator that is consistent with a transcription-splicing coupling factor in VDR-mediated gene expression. First, chromatin immunoprecipitation studies show that endogenous NCoA62/SKIP is recruited to native VDRresponsive promoters in a 1,25-(OH) 2 D 3 -dependent manner in osteoblast-like target cells, thus solidifying its coactivator role in the VDR-activated transcriptional process at target promoters. However, the major nuclear proteins interacting with NCoA62/SKIP are components of the U5 snRNP, a central snRNP involved in splicing of mRNA transcripts. Furthermore, NCoA62/SKIP is predominantly associated with the nuclear matrix, and it interacts with nuclear matrix constituents. This property is shared with numerous other splicing factors and transcriptional regulators (34,45). Finally, we provide functional evidence for a role of NCoA62/SKIP in the splicing mechanism by using a dominant-negative NCoA62/SKIP to transiently block splicing of 1,25-(OH) 2 D 3 -induced mRNA transcripts. Thus, our studies provide a framework to propose the hypothesis that NCoA62/SKIP is a candidate factor involved in coupling VDR-mediated transcription to RNA splicing.
Previous studies support a coactivator role for NCoA62/SKIP in 1,25-(OH) 2 D 3 -/VDR-activated transcription (18,20). However, these studies are largely based on in vitro interaction and reporter gene expression approaches, and thus, they lack a firm biological context to support this proposed role. The chromatin immunoprecipitation studies in Fig. 5 establish that NCoA62/ SKIP is physically recruited in a 1,25-(OH) 2 D 3 -dependent manner to the vitamin D-responsive regions of VDR target genes in osteoblasts. This selective recruitment is a key observation supporting the importance of NCoA62/SKIP in VDR-activated transcription in osteoblasts. As expected from previous studies (40), VDR entry to this region of the 24-hydroxylase promoter is rapid, occurring within 15 min of the 1,25-(OH) 2 D 3 addition to ROS17/2.8 osteoblasts. There is an apparent cycling of VDR on and off this promoter region that exhibits a periodicity of ϳ60 min over the time course of this experiment. This is consistent with previous ChIP studies examining ligand-dependent nuclear receptor recruitment to target gene promoters (40,46). SRC-1 is also rapidly recruited to this VDR-responsive promoter region, being first detected 15 min following the 1,25-(OH) 2 D 3 addition. The rapid entrance of SRCs probably reflects their role in an early step of the transactivation process, perhaps related to their chromatin remodeling or histone acetyltransferase activity (47,48). The early arrival of histone acetyltransferase proteins has been proposed as an initial step to open the promoter and permit greater accessibility of the tran-  scription machinery to the promoter template (40,49). We also note that the periodicity of SRC-1 is distinct from that of VDR, showing two phases of ϳ90-min duration. This may reflect SRC-1 replacement by other SRC family members or distinct coactivator complexes such as DRIP. For example, DRIPs have been shown to enter shortly after SRCs, leading to the proposal that DRIP functions in later distinct stages of the process, such as in RNA polymerase II recruitment (46). Remarkably, NCoA62/SKIP recruitment by 1,25-(OH) 2 D 3 is markedly delayed relative to SRC-1 (Fig. 5). NCoA62/SKIP demonstrates maximal 24-hydroxylase promoter occupancy at 60 min and at 135 min in two cycles. Clearly, this delay suggests a function for NCoA62/SKIP that is distinct from SRCs, a possibility that was first indicated by the synergism observed between NCoA62/SKIP and SRCs in VDR-activated transcription (20). The cyclical pattern of coactivator recruitment and the temporal differences between SRCs and NCoA62/SKIP are intriguing and will require further studies to sort out the mechanistic differences between these coactivators. Regardless, it is clear that endogenous NCoA62/SKIP is recruited in a 1,25-(OH) 2 D 3dependent manner to native promoters of vitamin D-responsive target genes in an osteoblast-like VDR target cell. Thus, NCoA62/SKIP is present in promoter complexes on native target genes in osteoblasts presumably through its interaction with the VDR-RXR heterodimer and through the formation of the VDR-SRC-NCoA62/SKIP ternary complex (20).
Based on the results of the ChIP assays in Fig. 5, it is somewhat unexpected that the predominant proteins identified in the GST-NCoA62/SKIP interaction approach were unrelated to nuclear receptor associated factors or to general transcriptional processes. In fact, three prominent NCoA62/ SKIP-interacting bands of 270, 240, and 115 kDa were identified as key components of the spliceosome, a massive ribonucleoprotein complex consisting of more than 100 proteins and five small nuclear RNAs (U1, U2, U4/6, and U5 snRNAs) that functions to remove introns from premessenger RNA (reviewed in Ref. 50). The upper band of the Ͼ200-kDa doublet in Fig. 6 was identified as a single species, human Prp8. Prp8 is a highly conserved splicing factor that is predominantly associated with the U5 complex. Prp8 appears to function at the catalytic core of the spliceosome, where it contacts pre-mRNA at both the 5Ј and 3Ј splice sites (51,52). The lower band of the doublet is a predicted 240-kDa protein that is identical to the human U5 snRNP-specific 200-kDa helicase (NP_054733), with the exception that it contains a 325-amino acid NH 2 -terminal extension. This protein belongs to the superfamily II of helicases (53), and it contains two highly conserved DEXH domains that are characteristic of putative RNA-stimulated ATPases and RNA helicases. Consistent with a putative function as an ATP-dependent RNA helicase, the human U5 200-kDa protein is known to bind ATP in vitro (54) and to unwind RNA duplexes (55). The purified 115-kDa band is a mixture of several proteins, the two most abundant being matrin 3 (discussed below) and the 116-kDa component of the U5 complex. The 116-kDa U5 snRNP protein is a GTP-binding protein that is implicated in locating the 3Ј splice site prior to step 2 of splicing (56). Interestingly, these three major proteins (namely, Prp8, the 200-kDa U5 snRNP helicase, and the 116-kDa U5 snRNP protein) form a stable, chaotropic salt-resistant subcomplex that can be isolated from the U5 core particle (57). The apparent stoichiometric quantities of these three bands relative to one another in Fig. 6 indicates that NCoA62/SKIP may interact with this subcomplex with the U5 core particle. The observation that three of the most abundant proteins identified in this analysis are important components of the U5 snRNP agrees with the proposed model in which NCoA62/ SKIP is recruited to the spliceosome before the first catalytic step, it is present through both catalytic steps, and it exits in association with the spliced out intron (23).
In addition to splicing factors, several putative matrix components were also identified as NCoA62/SKIP-interacting proteins in HeLa cell nuclear extracts. These included tubulin, matrin 3, and scaffold attachment factor A (Table II). Matrin 3 was originally reported to be one of 12 major inner nuclear matrix structural proteins (58). It contains two tandem RNA recognition motifs, and it binds to several matrix or scaffold attachment regions (59). Recently, Zhang and Carmichael (60) isolated matrin 3 in a complex of proteins that selectively binds inosine-containing or hyperedited RNAs (i.e. mRNAs encoding defective proteins). They concluded that matrin 3 may play a role in anchoring hyperedited mRNAs to the nuclear matrix while allowing selectively edited mRNAs to be exported from the nucleus and translated into proteins. Scaffold attachment factor A is also an abundant component of both the nuclear scaffold and chromatin, and it also exists in heterogeneous nuclear ribonucleoprotein complexes (61,62). Previous experiments demonstrated that scaffold attachment factor A specifically binds to scaffold/matrix attachment regions of DNA and, thus, could be involved in higher order chromatin structure (63)(64)(65). The isolation of these nuclear matrix scaffold-type proteins as NCoA62/SKIP interactors suggests the possibility that they may play a role in tethering NCoA62/SKIP to the nuclear matrix.
Despite the increasing amount of evidence that NCoA62/ SKIP interacts with components of the spliceosome and that the yeast ortholog may function in RNA splicing (68), a functional demonstration for the involvement of NCoA62/SKIP in the splicing process in mammalian cells has not been previously reported. To address this, we examined the effect of a dnNCoA62/SKIP on the accumulation of RNA expressed from a 1,25-(OH) 2 D 3 -responsive GH minigene cassette. The dnNCoA62/SKIP was originally identified as the minimal domain of NCoA62/SKIP that interacts with the VDR/RXR heterodimer and interferes with hormone-mediated transcription (20). Thus, the mechanism of interference was FIG. 8. Intact NCoA62/SKIP rescues the accumulation of unspliced transcripts induced by the dominant negative NCoA62/SKIP 87-342. COS-7 cells were transfected with 7 g of the VDRE 4 TKGH minigene, 0.14 g of pSG5VDR, and 4.2 g of pSG5 (lanes 1-4), 4.2 g of pSG5dnNCoA62 87-342 (lanes 5-8), and 4.2 g of pSG5dnNCoA62 87-342 plus 1.0 g of pSG5NCoA62 wild type (lanes 9 -12). Cells were treated with 1,25-(OH) 2 D 3 for the indicated time, and RNA was isolated and subjected to RT-PCR. thought to involve a disruption of the interaction of endogenous NCoA62/SKIP with liganded VDR/RXR. This is analogous to the assumed mechanism of action of dnSRCs. These latter dominant negative proteins contain the LXXLL motifs or the RID domain, and they prevent interaction of the endogenous SRCs with VDR/RXR, thereby leading to a reduction in activated transcription (66,67). Using the two dominant negative proteins, dnNCoA62/SKIP and dnGRIP (dnSRC-2), we show in Fig. 7 that expression of the dnNCoA62/SKIP leads to a transient accumulation of unspliced RNA and a delay in the accumulation of fully spliced message. These data suggest that the NCoA62/SKIP derivative interferes with an important and as yet unknown step in the splicing mechanism. The effect observed with dnNCoA62/ SKIP is contrasted by that observed with dnGRIP. The expression of this latter protein completely blocks the 1,25-(OH) 2 D 3 -induced accumulation of both spliced and unprocessed RNA, suggesting that this family of proteins may act at an earlier stage in the process than does NCoA62/ SKIP. This is supported by the observation in Fig. 5 that NCoA62/SKIP enters into the transcriptional complex later than does SRC. These data provide further support for the concept that NCoA62/SKIP and SRCs function to enhance nuclear receptor-mediated transcription via distinct mechanisms and that one aspect of NCoA62/SKIP activity may involve mRNA splicing.
Taken together, our data are consistent with a coactivator role for NCoA62/SKIP in VDR-mediated transcription that may involve mRNA splicing. It is possible that these observations may simply reflect two disparate functions for nuclear NCoA62/SKIP, since we have no evidence as yet for a direct link between these two properties. However, these properties are consistent with those of a potential coupling factor linking VDR-and nuclear receptor-mediated transcription to mRNA splicing. Additional approaches are required to establish this link. There is little doubt that NCoA62/SKIP is targeted to promoter regions in response to 1,25-(OH) 2 D 3 treatment and that its temporal entrance into these promoter complexes indicates a role that is distinct from SRCs, perhaps in a later post-polymerase stage. As reagents are developed, it will be interesting to examine other U5 snRNP proteins (e.g. Prp8 or the helicase) to monitor their potential association with these promoter complexes. Such studies would address whether the entry of NCoA62/SKIP to promoter complexes is related to spliceosome interactions or other potential mechanisms.