TIP30 Interacts with an Estrogen Receptor (cid:1) -interacting Coactivator CIA and Regulates c- myc Transcription*

Deregulation of c- myc expression is implicated in the pathogenesis of many neoplasias. Estrogen receptor (cid:1) (ER (cid:1) ) can increase the rate of c- myc transcription through the recruitment of a variety of cofactors to the promoter, yet the precise roles of these cofactors in transcription and tumorigenesis are largely unknown. We show here that a putative tumor suppressor TIP30, also called CC3 or Htatip2, interacts with an ER (cid:1) -interacting coactivator CIA. Using chromatin immunoprecipitation assays, we demonstrate that TIP30 and CIA are distinct cofactors that are dynamically associated with the promoter and downstream regions of the c- myc gene in response to estrogen. Both TIP30 and CIA are recruited to the c- myc gene promoter by liganded ER (cid:1) in the second transcription cycle. TIP30 overexpression represses ER (cid:1) -mediated c- myc transcription, whereas TIP30 deficiency enhances c- myc transcription in both the absence and presence of estrogen. Ectopic

Estrogen plays an important role in the development and maintenance of the mammary glands, as well as various other tissues, and in numerous human diseases that include breast and endometrial cancer, cardiovascular disease, and osteoporosis (1)(2)(3)(4)(5). Most of the effects of estrogen are facilitated by estrogen receptor ␣ (ER␣), 1 which controls the expression of a number of hormone-responsive genes (5,6), including the c-myc gene, which is important for cell proliferation (7)(8)(9)(10). ER␣, like many other nuclear receptors, contains two intrinsic transcriptional activation domains, designated AF-1 and AF-2 (3,11,12). The function of AF-1 is estrogen-independent, whereas the activity of AF-2 is estrogen-dependent. AF-1 and AF-2 activities show promoter context and cell type specificity and can act synergistically to activate transcription (5).
Although the precise mechanisms by which ER␣ regulates gene expression are still not clearly defined, there is solid evidence suggesting that ER␣ executes its effects by directing cyclical and combinatorial recruitment of cofactors on promoters (13)(14)(15)(16)(17). Most ER␣-interacting coactivators identified thus far are also able to interact with many other members of the nuclear receptor superfamily and have generalized functions, hence affecting transcriptional activation mediated by a wide spectrum of nuclear receptors (17)(18)(19)(20)(21). More recently, Giguere and co-workers (22) identified a novel nuclear receptor coactivator, called CIA (coactivator independent of AF-2 function). CIA was shown to interact with ER␣ and ER␤ in a ligand-dependent manner but not with other members of the nuclear receptor family. Consistent with its binding specificity, CIA was found to potentiate transcriptional activation by the ER but not by other nuclear receptors (22). Nevertheless, the precise mechanism by which CIA enhances transcription remains unknown. CIA may represent a novel class of ligand-dependent ER coactivators that are independent of AF-2 function.
We previously purified a protein, TIP30 (Tat-interacting protein 30) that interacts specifically with the activation domain of Tat (23). TIP30 is identical to CC3, which is absent in highly metastatic human small cell lung carcinoma (24). TIP30/CC3 has been proposed to function as a metastasis suppressor via its ability to promote apoptosis and inhibit angiogenesis (24 -27). Consistent with this hypothesis, ectopic expression of TIP30 was found to elevate the expression of a subset of proapoptotic genes (27) and angiogenic inhibitors and to downregulate the expression of certain angiogenic stimulators (25). Moreover, deletion of one or both alleles of Tip30 results in spontaneous development of hepatocellular carcinomas and other tumors in mice at a relatively long latency (28). Reduced expression of TIP30 is observed in 33% of human hepatocellular carcinomas, and mutations in the Tip30 gene that caused the instability or abnormal cellular distributions of the TIP30 * This work was supported by American Cancer Society Research Scholar Grant for Beginning Investigators RSG0216501GMC, Department of Defense Breast Cancer Research Program Idea Award DAMD170210507, a University of Nebraska Medical Center start-up grant (to H. X.), and a National Institutes of Health grant (to R. G. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  protein were identified in some of the human hepatocellular carcinoma specimens (28). These data further suggest that TIP30 is a tumor suppressor.
In this study, we describe the cloning and characterization of a TIP30-interacting protein identical to CIA and provide evidence that TIP30 is an important regulator of ER␣-mediated c-myc transcription. This study defines a new pathway for regulating expression of the c-myc gene and possibly other ER␣ target genes that are involved in tumoriogenesis.

EXPERIMENTAL PROCEDURES
Protein Purification and Chromatography-The HeLa cell line (HeLa-fTIP30) stably expressing FLAG-tagged TIP30 was obtained by the introduction of pCIN4-FLAG-tagged TIP30 into HeLa S3 and selection in a G418-containing medium (27). Nuclear extracts were prepared as described previously (29) with the following modifications. During purification procedures, HEPES-HCl, pH 7.9, was used in buffers instead of Tris-HCl. Nuclear pellets were resuspended in a low salt buffer containing sulfosuccinimidylpropionate (Pierce), transferred to a homogenizer, and mixed with six strokes of a loose pestle. The suspensions were then transferred to glass beakers and dispersed with a stirring bar. The high salt buffer was slowly added into the mixtures. The cross-linking reactions were incubated at 4°C for 2 h and then stopped by adding 1 M Tris-HCl, pH 8.0, to a final concentration of 20 mM. The extracts were centrifuged at 15000 rpm for 30 min and stored at Ϫ80°C. For coimmunoprecipitation, 100 ml of nuclear extract prepared from the HeLa-f:TIP30 cells was mixed with 200 l of anti-FLAG epitope immunoaffinity matrix M2 (Sigma) in BC450 containing 0.1% Nonidet P-40 and 0.45 M KCl. The matrix was washed extensively with BC500 plus 0.1% Nonidet P-40 and 0.5 M KCl and eluted with 1 ml of BC1000 containing 200 g/ml FLAG peptide and 1 M KCl. The eluted proteins were concentrated and resolved on SDS-PAGE and stained with silver or Coomassie Blue. The indicated bands were excised from the gel and digested with endoproteinase C. The resulting peptides were isolated by high pressure liquid chromatography and subjected to microsequencing analysis (Core Facility, Rockefeller University). Affinity chromatography was carried out as described previously (23).
Cloning of the CIA Gene-An expressed sequence tag clone (DKFZp434J208) was initially identified by data base search using two peptides (NMPQADAMVLVAR and DLRDFR) from amino acid sequencing and peptide mass data from mass spectrophotometer analysis of a TIP30-interacting protein (65 kDa). The expressed sequence tag clones (IMAGE: 1185534 and 1335209), containing similar cDNA sequences, were identified and purchased from ATCC and sequenced. Primers were designed for 5Ј-RACE and 3Ј-RACE PCRs according to the sequence of the insert cDNA. The HeLa cDNA library (Clontech) was used as a template for 5Ј-RACE and 3Ј-RACE. The experiments were carried out according to the manufacturer's instruction (Clontech). The resulting DNA fragments were subcloned into a pCR2.1 vector by PCR kits (Invitrogen). Several cDNA fragments were amplified and sequenced by the core facility at the University of Nebraska Medical Center, and the resulting sequences were used for searching the expressed sequence tag clones and known genes in GenBank TM . The 856-bp cDNA containing the translational start codon was amplified using PCR primers (primary primer, 5Ј-ATCTCTACTATGTGTGTG-GTCCCG; nest primer, 5Ј-CAAGTCTCGGGGGTCTCGAATGTC). The pCR2.1-CIA containing full-length CIA cDNA was constructed by subcloning a PCR-amplified cDNA from the HeLa cDNA library (Clontech) using specific primers (forward, 5Ј-GGAAGATCTATGAATACGGCTC-CATCAAGACCCAGC; reverse, 5Ј-GGAAGATCTCAGTAATGCCTCT-GGTAAGATCCCAT).
Plasmids and Antibodies-The pcDNA3.1-CIA plasmid was generated by inserting the CIA cDNA, excised from pCR2.1-CIA with BglII, into pcDNA3.1 at the BamHI site. For antibody production, the bacterial expression plasmid pREST-his-CIA was generated by inserting a BglII-DNA fragment released from pCR2.1-CIA into pRSET-B (Invitrogen). The recombinant His tag CIA protein was expressed in bacteria, purified as described previously, and used to generate rabbit anti-CIA antibodies (Covance Inc., Denver, PA). The pGL3-hu-Myc plasmid containing the human c-myc promoter was constructed by subcloning a 2540-bp HindIII-DNA fragment excised from pSV40CAT-hu-Myc plasmid (a gift from David Bentley, University of Colorado Health Science Center) into pGL3-basic vector (Promega) at the HindIII site. This DNA fragment extends 2.3 kb 5Ј of the P1 start site to a Nae1 site at ϩ50 relative to the P2 start site. Antigen-purified anti-TIP30 (23) and anti-CIA antibodies were purified as described previously (30). Anti-ER␣ (HC-20, Santa Cruz Biotechnology), anti-CBP (A22, Santa Cruz Biotechnology), and anti-RNA polymerase II (8WG16) antibodies were used for ChIP assays.
Mice and MEFs-Generation of TIP30 null mice and MEFs was described previously (28). The genetic backgrounds of Tip30 ϩ/ϩ and Tip30 Ϫ/Ϫ mice were regarded to be identical because they were backcrossed 10 times with C57BL6J mice. MEF clones of each genotype were prepared from sibling embryos obtained by a heterozygous crossing. All animal experimentation was performed according to the National Institutes of Health guidelines.
Pull-down Assay-Pull-down assays were performed as described previously (22). [ 35 S]Methionine-labeled CIA and Bax proteins were generated by in vitro transcription and translation using the TNTcoupled reticulocyte lysate system (Promega). Fusion proteins were induced in DH5␣ Escherichia coli with 0.4 mM isopropyl-␤-D-thiogalactopyranoside for 2 h. The cells were sonicated 10 times using 30-s pulses in 1ϫ phosphate-buffered saline containing 1 mM phenylmethylsulfonyl fluoride and 1 mM dithiothreitol. The cleared lysates were bound to glutathione-Sepharose (Amersham Biosciences) for 15 min at 4°C. The protein-saturated beads were washed three times in 1ϫ phosphatebuffered saline with 1 M NaCl and stored in BC100 (20 mM Tris-HCl, pH 7.9, 1 mM EDTA, 1 mM dithiothreitol, 20% glycerol, 100 mM KCl, 0.1% Nonidet P-40). 20 l of GST or GST-TIP30 beads were incubated with 10 l of [ 35 S]methionine-labeled proteins in 10 l of BC135 buffer (20 mM Tris-HCl, pH7.9, 1 mM EDTA, 1 mM dithiothreitol, 20% glycerol, 0.1% Nonidet P-40, 135 mM KCl) for 2 h at 4°C. The binding reaction mixture was washed three times in BC135 buffer. The bound proteins were analyzed by SDS-PAGE followed by autoradiography of fixed and dried gels.
ChIP Assays-ChIP assays were performed as described previously (15). MCF-7 cells were treated with or without 10 Ϫ8 M E2 for 45 min and with formaldehyde to cross-link proteins to DNA. Soluble chromatin was prepared and immunoprecipitated with preimmune serum and protein A-Sepharose and then immunoprecipitated with anti-CIA, anti-TIP30 serum, or control anti-ER␣, anti-CBP, and anti-RNA polymerase II antibodies. The final DNA preparations were amplified using a pair of primers and analyzed in 1.5% agarose gel following ethidium bromide stain.
Transient Transfection-Transfection assays were performed as described previously (26). COS-1 cells in 24-well dishes or MEFs in 6-well dishes were transfected with plasmid DNA using LipofectAMINE according to the manufacturer's instruction (Invitrogen). In all experiments, a plasmid pRL-CMV for expressing Renilla luciferase was used as a control for transfection efficiency, and activities of firefly and Renilla luciferases were measured with Promega's dual luciferase reporter assay system, normalized, and expressed as relative luciferase light units. The vector pCMV-ER␣ for expressing human ER␣ and vector pNT7-TR␣ for human thyroid receptor ␣ (TR␣) were described previously (31). TRE-luciferase reporter was described previously (31). Gal-RXR expressing vector and Gal-luciferase reporter were described previously (19).
Histology and Immunohistochemistry-Mammary gland 4 was removed and fixed with 10% buffered-formalin and embedded in paraffin blocks. The sections were deparaffinized, rehydrated, and stained with hematoxilin and eosin. For immunohistochemistry staining, unstained sections were rehydrated and incubated overnight at 4°C with anti-Myc antibody (Upstate Biotechnology, Inc.), and staining was developed using VECTASTAIN Elite ABC kit (Vector Laboratories) according to the manufacturer's instructions and then counter-stained with hematoxilin and eosin. The number of Myc nuclear positive epithelial cells was divided by the total number of epithelial cells counted.

Identification of a TIP30-interacting Protein and Cloning Its
Full-length Cognate cDNA-The discovery of roles of TIP30 in the control of expression of genes involved in apoptosis and angiogenesis indicates that TIP30 may interact with transcription factors to regulate gene expression. To identify and affinity purify cellular proteins that interact with TIP30, we first established HeLa cell lines that stably expressed FLAG-tagged wild type TIP30. After preparation of nuclei from these cells, nuclear proteins were cross-linked by mixing the nuclei with the water-soluble cross-linker sulfosuccinimidylpropionate. Sulfosuccinimidylpropionate-treated nuclear extracts were made and subjected to affinity purification with anti-FLAG M2 beads. The bound proteins were eluted with FLAG peptide and analyzed by SDS-polyacrylamide gel electrophoresis after cleaving the cross-linker. As shown in Fig. 1A, three proteins were coimmunopurified with TIP30 and designated TIP30-interacting proteins. These three proteins were not immunoprecipitated with control anti-hemagglutinin beads. Amino acid sequencing of one of these interacting proteins (65 kDa), designated TIP30-interacting protein 1 (TIPIF-1), yielded two peptide sequences (NMPQADAMVLVAR and DLRDFR). A HeLa cell-derived 2.2-kb cDNA encoding TIPIF-1 was cloned and sequenced (GenBank TM accession number AF470686). Sequence analysis revealed that it encodes a 587-amino acid protein with a tract of Arg-Asp residues in the amino-terminal region. The repeat Arg-Asp residues were previously identified in the splicing factor U1 70K and the RD protein, a component of the negative elongation factor NELF (32). TIPIF-1 also contains a receptor-binding motif known as the LXXLL motif (33,34) and seven repeats of a short sequence motif (RDLRD(H/ F)R) that is present in subunit 10 of mouse translation initiation factor 3. Following cloning of the TIPIF-1 cDNA, a data base search revealed identity with the estrogen receptor coactivator CIA (22). The reported CIA cDNA, which was isolated from a human fetal kidney cDNA library, contains a 620residue open reading frame lacking a translational stop codon at the 5Ј end. In contrast, the open reading frame in our TIFIF-1 cDNA starts at a methionine residue located immedi-ately after two in-frame stop codons, indicating that it encodes a full-length CIA. The sequence discrepancy in the 5Ј-untranslated region between TIPIF-1 and CIA may be due to alternative splicing products in HeLa and kidney cells, because the cDNA encoding CIA was cloned from the human fetal kidney cDNA library (22). We now refer to TIPIF-1 as CIA. To ascertain whether CIA interacts with TIP30, we tested the binding of TIP30 from nuclear extracts to a GST-CIA fusion protein. As shown in Fig. 1B, TIP30 was detected in the eluate from the GST-CIA affinity column but not in the eluate from the control GST column. In addition, we also tested the binding of in vitro translated CIA to GST-TIP30 using a pull-down assay. As shown in Fig. 1C, CIA bound GST-TIP30 but not GST control protein. By contrast, BAX protein, a pro-apoptotic protein used as a negative control was not bound by GST-TIP30. Collectively, these results indicate that CIA interacts with TIP30.
Estrogen Regulates TIP30 and CIA Occupancy of the c-myc Gene in Breast Cancer MCF-7 Cells-Transcription of the cmyc gene is controlled by transcription factors that interact with numerous positive and negative regulatory elements in the c-myc promoter regions (35). ER␣ can stimulate c-myc transcription by interacting with an estrogen-responsive element of the c-myc promoter (8). Using ChIP assays, a previous study has established that a number of ER␣-interacting coactivators are recruited to the promoters of endogenous estrogenresponsive target genes, including c-myc, following estrogen treatment (15). If CIA and TIP30 are specific cofactors for ER␣, they might also be associated with ER␣ on endogenous estrogen-responsive target genes. To test this possibility, we used the same ChIP assay (15) to determine whether CIA and/or TIP30 is recruited to the promoter of the c-myc gene in the estrogen-dependent human breast cancer cell line MCF-7. Fig.  2B shows that, as previously reported, the anti-ER␣ antibody precipitated a c-myc promoter fragment containing P1 and P2 promoters in the presence (lane 6) but not in the absence (lane 5) of E2. In contrast, the c-myc promoter was precipitated by anti-TIP30 antibody (lane 3) but much less so by anti-CIA antibody (lane 1) in the absence of E2. E2 increased the asso- Because previous studies demonstrated that ER␣ and coactivators (AIB1, p300, CBP, pCAF, and TRAP220) are assembled on the promoter during preinitiation but subsequently released during elongation (15), we next determined whether CIA and TIP30 occupy the c-myc gene during elongation. Four pairs of primers covering a region (ϩ2196 to ϩ2628) in exon 2 and a region (ϩ4330 to ϩ4685) in exon 3 were used for PCR amplification of the final DNA preparations. Binding of both TIP30 and CIA to these regions was observed in the absence of E2 (Fig. 2B, lanes 1 and 3). E2 induction resulted in increased CIA binding (lanes 1 versus 2) but decreased TIP30 binding (lanes 3 versus 4) to these regions. As a positive control, binding of the largest subunit of RNA polymerase II to the regions upstream and downstream of the initiation site was also observed in both the absence and presence of E2 (Fig. 2B, lanes 9  and 10). These results suggest that estrogen increases association of CIA with the coding regions of c-myc.
To assess the specificity of association of TIP30 and CIA with the c-myc gene, we examined whether TIP30 and CIA were assembled on the promoter of the cyclin B1 gene that is not directly regulated by ER␣ (21). As expected, RNA polymerase II is associated with the promoter (Fig. 2C, lanes 9 and 10), but ER␣ is not (Lanes 7 and 8). However, ER␣, TIP30, and CIA were not associated with the promoter in either the presence or the absence of E2 (Lanes 3-6). In addition, nonspecific antibodies from a preimmune rabbit serum did not precipitate the DNA elements of the c-myc gene (Fig. 2D). The protein levels of TIP30 and CIA in MCF-7 cells are not significantly changed after E2 treatment (Fig. 2E), suggesting that E2-regulated association of TIP30 and CIA with the c-myc gene is not due to the influence of TIP30 and CIA expression by E2. Therefore, associations of these proteins with the c-myc gene are specific.
To determine whether the occupancy of the c-myc gene by TIP30 and CIA requires elongating RNA polymerase II in estrogen-dependent and -independent transcription, MCF-7 cells were treated with ␣-amanitin, which specifically inhibits RNA polymerase II elongation (15,36), and then subjected to ChIP analysis. Consistent with the preceding observations, the occupancy of the c-myc promoter by TIP30 and CIA was not affected by ␣-amanitin (Fig. 2F). In contrast, the occupancy of the coding region of the c-myc gene by TIP30 and CIA was inhibited by ␣-amanitin (Fig. 2E). This result suggests that association of TIP30 and CIA with the coding regions of the c-myc gene depends on elongating RNA polymerase II.
Estrogen Dynamically Regulates Occupancy of the c-myc Gene by TIP30 and CIA-Occupancy of ER␣ and RNA polymerase II on the promoter was previously shown to peak at 30 -45 min in the first transcription cycle and at 120 -150 min in the second cycle following the addition of E2 (15). We therefore sought to determine the timing of TIP30 and CIA occupancy on the promoter. As was shown previously, ER␣ and CBP occupancy on the promoter peaked at 30 -60 min in the first transcription cycle and at 150 min in the second cycle upon E2 treatment (Fig. 3A). The results obtained with CIA were similar to those obtained with ER␣ and CBP. Interestingly, TIP30 was dissociated with the promoter at 60 min and reassociated with the promoter at 120 -150 min in the second cycle (Fig. 3A). Increased binding of TIP30 and CIA to exon 2 was observed at 150 min following E2 treatment (Fig. 3B). In contrast, much less CBP was detected to associate with exon 2 upon E2 induction (Fig. 3B). Therefore, these results suggest that estrogen increases the recruitment of CIA to but decreases dissociation of TIP30 from the c-myc promoter region during the first transcription cycle. Unlike the other known ER␣ coactivators, which are released during elongation, CIA shows an increased association with the coding regions following E2 treatment. However, in the second cycle of estrogen-induced transcription, both CIA and TIP30 are associated with the promoter as was observed for ER␣ and CBP. Prolonged estrogen treatment even further increases association of CIA and TIP30 with the coding region of the c-myc gene (Fig. 3B).
Ectopic Expression of CIA and TIP30 Inhibits ER␣-mediated Transcription-To test whether CIA and TIP30 function as coregulators of ER␣ in the transcription of the c-myc gene, we transfected COS-1 cells with vectors expressing ER␣, CIA, TIP30, and a kinase-defective mutant TIP30 (30) with a construct containing a luciferase reporter gene controlled by the c-myc promoter (Ϫ2.3 kb 5Ј of P1 start site to ϩ50 relative to P2 start site). As shown in Fig. 4A, ectopic TIP30 inhibited the E2-dependent transcriptional activity of ER␣ in COS-1 cells, whereas ectopic CIA had no significant effect on transcription. The kinase-defective mutant TIP30 (30) did not significantly affect ER␣-dependent transcription. Surprisingly and paradoxically, coexpression of ectopic CIA with ectopic TIP30 did not reverse but instead further potentiated the inhibitory effect of TIP30 (4-fold) in ER␣-mediated transcription.
To determine whether TIP30 and CIA also regulate transcription from other promoters, we performed transient-transfection experiments using the luciferase reporter gene construct containing the artificial promoter either with estrogen receptor-binding sites (EREs) or with thyroid receptor (TR) ␣-binding sites (31). As expected, CIA potentiated ER␣-mediated transcription from this promoter in the presence of E2 (Fig. 4B), but TIP30 inhibited the activity of ER␣, whereas a kinase-defective mutant TIP30 (30) did not. Consistent with the observations for c-myc promoter, the activity of ER␣ on the ERE-binding promoter was inhibited by ectopic expression of both TIP30 and CIA. In contrast to ER␣, the activity of TR␣ on the TRE-binding promoter was only slightly increased by cotransfection of TIP30 and/or CIA with TR␣ (Fig. 4C), suggesting that neither TIP30 nor CIA had a significant effect on TR-mediated transactivation.
Taken together, these data indicate that both CIA and TIP30 are important interacting factors in modulating the activity of ER␣, and CIA can cooperate with TIP30 to repress ER␣-mediated transcription when they are overexpressed. However, because Western blot analyses revealed that the levels of transiently expressed ectopic TIP30 and CIA were much higher than the levels of endogenous TIP30 and CIA in COS-1 cells (data not shown), it is possible that the effects of TIP30 and CIA on ER␣-mediated transcription observed here may not represent their physiological functions.
Loss of TIP30 Increases ER␣-mediated Transcription-To verify that the effect of TIP30 on ER␣-mediated transcription is not limited by its overexpression, we next examined whether the loss of TIP30 expression affected ER␣-mediated transcription by performing transient transfection assays in Tip30 ϩ/ϩ , Tip30 ϩ/Ϫ , and Tip30 Ϫ/Ϫ MEFs. Consistent with the preceding results, Fig. 5A shows that the loss of TIP30 results in an increased ER␣-mediated transcription from the artificial EREbinding promoter containing three EREs (left panel). Coexpression of either TIP30 alone or TIP30 and CIA in Tip30 Ϫ/Ϫ MEFs results in inhibition of ER␣-mediated transcription (right panel). In contrast, the loss of TIP30 had a minimal effect on ligand-dependent transactivation by Gal4-RXR␣ (AF2 domain of retinoid X receptor) (19), implying that TIP30 may preferentially inhibit the transcriptional activity of ER␣ (Fig. 5C). Expression of TIP30 did not completely restore ER␣ activity in Tip30 Ϫ/Ϫ MEFs. We reasoned that increased ER␣-mediated transcription was partly due to TIP30 deficiency in Tip30 Ϫ/Ϫ MEFs and partly due to other genetic changes that these cells may have undergone.
We also tested whether the complete absence of TIP30 in cells affects the transcriptional activity of ER␣ on the c-myc promoter. As expected, ER␣ showed a greater stimulation of the c-myc promoter in Tip30 Ϫ/Ϫ MEFs than in Tip30 ϩ/ϩ MEFs (16-fold; Fig. 5B, left panel) and Tip30 ϩ/Ϫ MEFs (3-fold lower) in the presence of E2. TIP30 loss also resulted in a 5-fold increase in estrogen-independent transcription from the c-myc prompter (Fig. 5B, left panel). This effect was due to the ab- sence of TIP30, because expression of TIP30 in Tip30 Ϫ/Ϫ cells resulted in a 42% inhibition in transcriptional activity of ER␣ (Fig. 5B, right panel). The finding that deletion of the Tip30 gene elicits higher transcription from the c-myc promoter indicates that the function of TIP30 is to repress both estrogenindependent and estrogen-dependent ER␣-mediated c-myc transcription.
Lack of TIP30 Increases c-myc Expression in the Mammary Gland-To demonstrate that TIP30 negatively regulates expression of the c-myc gene in vivo, we utilized a semi-quantitative RT-PCR analysis to examine the level of c-myc mRNA in the mammary gland from virgin Tip30 Ϫ/Ϫ mice. As expected, a higher level of c-myc RNA was observed in mammary glands from Tip30 Ϫ/Ϫ mice relative to age-matched Tip30 ϩ/ϩ mice (Fig. 6A). In contrast, the level of cyclin B1 mRNA (Fig. 6A), whose promoter was not bound by ER␣, CIA, or TIP30 (Fig.  2C), was relatively normal in these Tip30 Ϫ/Ϫ mammary glands. We next used an immunohistochemical analysis to evaluate c-myc protein expression in the Tip30 Ϫ/Ϫ mammary glands. Punctate staining for c-Myc protein was detected in both-Tip30 ϩ/ϩ and Tip30 Ϫ/Ϫ mammary epithelial cells. However, 41% of the Tip30 Ϫ/Ϫ mammary epithelial cells had nuclear staining for c-myc, whereas 27% of the Tip30 ϩ/ϩ mammary epithelial cells had nuclear staining for c-myc (Fig. 6, B and C). We also examined ER␣ expression in these mammary glands using immunohistochemistry and Western blot analyses. The level of ER␣ and percentages of ER␣ nuclear staining in both Tip30 ϩ/ϩ and Tip30 Ϫ/Ϫ mammary epithelial cells were similar (data not shown), indicating that increased c-myc expression is not due to an increased expression of ER␣ in the mammary epithelium. These results demonstrate that a lack of TIP30 results in increased c-myc expression in mouse mammary glands. Taking these results together, we concluded that TIP30 is a repressor of ER␣-mediated transcription of the c-myc gene. DISCUSSION ER␣-mediated transcription requires coactivators that interact with ER␣ and enhance its transcriptional activity. Although most ER coactivators identified thus far associate with the AF-2 domain and enhance the transcriptional activity of many nuclear receptors, as well as nonreceptor activators (4,14,37), the unique coactivator CIA specifically enhances ER␣ activity independent of AF-2 function (22). In the present study, we have demonstrated the interaction between a puta- tive tumor suppressor TIP30 and CIA and subsequently cloned the full-length cDNA encoding CIA. Our biochemical and genetic data argue strongly that TIP30 is a repressor of ER␣mediated transcription of the c-myc gene. Specifically, TIP30 is coimmunoprecipitated with CIA from HeLa nuclear extracts and bound to CIA in vitro (Fig. 1). TIP30 and CIA are associated with the c-myc promoter in the absence of E2 and associated with the promoter and transcribed regions in the second transcription cycle in the presence of E2, suggesting that they are present in the same transcription complexes (Fig. 2). This is further supported by the results (Fig. 3A) that TIP30 and CIA cooperatively inhibit ER␣-mediated transcription in transient transfection assays. The fact that TIP30 and CIA were copurified in the presence of protein cross-linkers also indicates that they do not form a stable complex but may transiently interact with each other during the second transcription cycle. Finally, involvement of TIP30 in ER␣-mediated transcription is also supported by the observations that a lack of TIP30 results in enhanced ER␣-mediated transcription in transient transfection assays (Fig. 3) and increased c-myc expression in mammary glands (Fig. 4). Therefore, we conclude that interaction between TIP30 and CIA is biologically relevant, and TIP30 is a repressor of ER␣-mediated transcription of the c-myc gene.
Although CIA is not related to any known proteins, it contains a receptor-binding motif known as the LXXLL motif and a motif consisting of Arg-Asp repeats (22). Interestingly, the Arg-Asp repeats were previously found in a putative RNAbinding component (RD) of NELF (32) that interacts with RNA polymerase II and represses transcription. Because CIA, unlike other ER-interacting coactivators, is not released during elongation (Fig. 3), it may bind to RNA via the Arg-Asp repeats to regulate elongation. It is interesting in this regard that estrogen promotes occupancy of CIA on the promoter and the coding regions of the c-myc gene. The precedent for this type of assembly of transcription factors is observed in galactose activation of GAL1 and heat shock activation of HSP82 in yeast and hsp70 in HeLa cells. Elongation factors such as Spt5 (38) and Sug1/ Rpt6 (39,40) are associated with both promoters and open reading frames of transcribed genes in yeast. Transcription activation induces occupancy of those factors on the GAL1 promoter and the open reading frame. Therefore, we propose that CIA may be an elongation factor that is specifically involved in the transcription of ER␣-targeted genes in mammalian cells. TIP30 and CIA may represent a novel class of coregulators for ER␣.
The c-myc gene is controlled by numerous transcription factors through upstream sequences of its P1 and P2 start sites (7,35). Previous studies have established that ER␣ increases cmyc transcription through a noncanonical ERE element (8).
The experiments in the current study were designed to first determine the role of TIP30 and CIA in ER␣-mediated c-myc transcription. Our data clearly demonstrate that TIP30 and CIA are involved in the regulation of c-myc transcription. It should be emphasized that the inhibition of ER␣-mediated transcription by TIP30 is not caused by a nonspecific repression, because TIP30 overexpression or loss of expression failed to alter transactivation by other nuclear receptors under the same conditions. Overall, previous (22) and current studies have suggested complicated roles of TIP30 and CIA in ER␣mediated transcription, although it remains unclear precisely how TIP30 and CIA function in this process. However, based on our data, we hypothesize that TIP30 and CIA may act as repressors of c-myc transcription. In this scenario, the binding of TIP30 to the c-myc promoter in the absence of E2 might lead to phosphorylation of the CTD of RNA polymerase II (27) during the formation of preinitiation complexes. This in turn might destabilize the association of RNA polymerase II with coactivators such as AIB1, p300, pCAF, CBP, and TRAPs (15) at the initiation step. In this regard, an inhibitor of RNA polymerase II elongation, ␣-amanitin, (15,36), which inhibited the association of RNA polymerase II with the coding region of the c-myc gene, also abolished the association of TIP30 and CIA (Fig. 2F), suggesting that association of TIP30 and CIA with the coding region is RNA polymerase II-dependent. In addition, TIP30 was previously shown to phosphorylate the CTD in vitro (27), and the kinase-defective TIP30 mutant failed to repress ER␣mediated transcription (Fig. 4B). In the first transcription cycle, CIA may facilitate displacement of TIP30 from the promoter, possibly through a transient interaction with TIP30, and subsequently associate with elongation complexes. Alternatively, ER␣ and ER␣-recruited factors could also facilitate TIP30 displacement from the promoter, with the observed TIP30-CIA interactions being more relevant to their mutual presence in transcription elongation complexes in the absence FIG. 6. Loss of TIP30 increases c-myc expression in mammary glands. A, loss of TIP30 increases the levels of c-myc mRNA in mammary glands. RNA was isolated from mammary glands of 8-week-old virgin Tip30 ϩ/ϩ and Tip30 Ϫ/Ϫ mice. RT-PCR analysis was used to monitor expression of the c-myc, cyclin B1, and ␤-actin genes. B-D, loss of TIP30 increases c-myc-stained epithelial cells in mammary glands. Representative immunostaining for c-myc on mammary gland sections from 8-week-old Tip30 ϩ/ϩ (B) and Tip30 Ϫ/Ϫ virgin mice (C) are shown. The bar graph indicates the percentage of c-myc nuclear-positive epithelial cells (D). The results represent two 8-week-old virgin mice for each genotype. About 500 epithelial cells were counted in the mammary gland section for each animal. of estrogen as well as in the second cycle transcription complexes following estrogen treatment. In the second transcription cycle, CIA may cooperate with TIP30 to repress transcription. In support of this possibility, CIA was previously shown to inhibit transcription of a Gal-driven TK reporter when it was fused to a GAL4 DNA-binding domain, indicating that CIA may possess an intrinsic inhibitory function (22). Clearly, further studies will be required for testing this hypothesis.
Notwithstanding, we could not pinpoint the binding site for TIP30 on the c-myc promoter in the absence of estrogen because of the limitations in ChIP assays. Given the bewildering regulatory elements in the c-myc promoter and observations of RNA polymerase II holdback at the P2 promoter (35), future studies will focus on the effects of TIP30 and CIA on RNA polymerase II and the mechanism by which CIA and TIP30 regulate transcription.
The finding that TIP30 acts as a negative regulator in both unliganded and liganded ER␣-mediated c-myc expression led us to predict that TIP30 may regulate tumorigenesis and tissue development of ER␣-targeted organs (41,42). As described in a recent report (28), we observed that a group of C57B6/129Svj hybrid mice deficient in TIP30 spontaneously developed tumors in ER␣ target organs including the ovary, uterus, and liver but not in the mammary gland. Recently, we have extended these studies by examining mammogenesis and tumorigenesis of cohorts of C57B6 mice deficient in TIP30 and control wild type mice. We have observed that virgin female mice lacking TIP30 exhibited ductal hyperplasia and spontaneously developed tumors in the mammary gland compared with control wild type mice. 2 Because mild hyperplasia was described previously in the mammary gland of transgenic mice overexpressing c-myc (9,37,43), the ductal hyperplasia in Tip30 Ϫ/Ϫ mice could be partly due to elevated expression of the c-myc gene that was observed in their mammary glands (Fig. 6). This genetic evidence agrees well with a negative role of TIP30 in ER␣-mediated c-myc transcription. However, we do not exclude the possibility that TIP30 may also regulate other ER␣ target genes that may also contribute to the ductal hyperplasia. Future studies should identify additional TIP30 target genes in the mammary epithelial cells and illuminate the molecular basis for TIP30 loss-facilitated tumorigenesis.
In summary, we have identified a TIP30-interacting protein as the ER-interacting coactivator CIA and revealed that TIP30 and CIA directly regulate the transcription of the c-myc gene. Like other ER␣ cofactors (14,15), TIP30 and CIA are recruited by ER␣ on the c-myc promoter in a cyclic fashion. Concordant with its effect on ER␣-mediated c-myc transcription, loss of TIP30 increases c-myc expression in the mammary gland of mice. Our studies suggest that TIP30 and CIA represent a novel class of coregulators for regulating gene expression during mammogenesis and tumorigenesis.