Portal de Periódicos da CAPES

YEAF1/RYBP and YAF-2 Are Functionally Distinct Members of a Cofactor Family for the YY1 and E4TF1/hGABP Transcription Factors

2002; Elsevier BV; Volume: 277; Issue: 25 Linguagem: Inglês

10.1074/jbc.m203060200

1083-351X

Chika Sawa, Tatsufumi Yoshikawa, Fumihiko Matsuda-Suzuki, Sophie Deléhouzée, Masahide Goto, Hajime Watanabe, Jun‐ichi Sawada, Kohsuke Kataoka, Hiroshi Handa,

14-3-3 protein interactions

The transcription factor hGABP/E4TF1 is a heterotetrameric complex composed of two DNA-binding subunits (hGABPα/E4TF1–60) and two transactivating subunits (hGABPβ/E4TF1–53). In order to understand the molecular mechanism of transcriptional regulation by hGABP, we searched for proteins that interact with the non-DNA-binding subunit, hGABPβ, using yeast two-hybrid screening. We identified a human cDNA encoding a protein related to YAF-2 (YY1-associated factor 2), which was previously isolated as an interacting partner of the Ying-Yang-1 (YY1) transcription factor. Reflecting this similarity, both YAF-2 and this novel protein (named YEAF1 for YY1- andE4TF1/hGABP-associatedfactor-1) interacted with hGABPβ and YY1in vitro and in vivo, indicating that YEAF1 and YAF-2 constitute a cofactor family for these two structurally distinct transcription factors. By using yeast three-hybrid assay, we demonstrated that hGABPβ and YY1 formed a complex only in the presence of YEAF1, indicating that YEAF1 is a bridging factor of these two transcription factors. These cofactors are functionally different in that YAF-2 positively regulates the transcriptional activity of hGABP but YEAF1 negatively regulates this activity. Also,YAF-2 mRNA is highly expressed in skeletal muscle, whereas YEAF1 mRNA is highly expressed in placenta. We speculate that the transcriptional activity of hGABP is in part regulated by the expression levels of these tissue-specific cofactors. These results provide a novel mechanism of transcriptional regulation by functionally distinct cofactor family members. The transcription factor hGABP/E4TF1 is a heterotetrameric complex composed of two DNA-binding subunits (hGABPα/E4TF1–60) and two transactivating subunits (hGABPβ/E4TF1–53). In order to understand the molecular mechanism of transcriptional regulation by hGABP, we searched for proteins that interact with the non-DNA-binding subunit, hGABPβ, using yeast two-hybrid screening. We identified a human cDNA encoding a protein related to YAF-2 (YY1-associated factor 2), which was previously isolated as an interacting partner of the Ying-Yang-1 (YY1) transcription factor. Reflecting this similarity, both YAF-2 and this novel protein (named YEAF1 for YY1- andE4TF1/hGABP-associatedfactor-1) interacted with hGABPβ and YY1in vitro and in vivo, indicating that YEAF1 and YAF-2 constitute a cofactor family for these two structurally distinct transcription factors. By using yeast three-hybrid assay, we demonstrated that hGABPβ and YY1 formed a complex only in the presence of YEAF1, indicating that YEAF1 is a bridging factor of these two transcription factors. These cofactors are functionally different in that YAF-2 positively regulates the transcriptional activity of hGABP but YEAF1 negatively regulates this activity. Also,YAF-2 mRNA is highly expressed in skeletal muscle, whereas YEAF1 mRNA is highly expressed in placenta. We speculate that the transcriptional activity of hGABP is in part regulated by the expression levels of these tissue-specific cofactors. These results provide a novel mechanism of transcriptional regulation by functionally distinct cofactor family members. adenovirus early 4 Ying-Yang-1 YY1-associated factor 2 human GA-binding protein YY1- and E4TF1/hGABP-associated factor-1 Ring1 and YY1 binding protein 3-aminotriazol Glutathione S-transferase retinoblastoma susceptibility gene conserved region, OR, original region surface plasmon resonance Polycomb group E4TF1 was originally purified from HeLa cells and was identified as one of the sequence-specific transcription factors that bind to and stimulate transcription from the adenovirus early 4 (E4)1 promoter (1Watanabe H. Imai T. Sharp P.A. Handa H. Mol. Cell. Biol. 1988; 8: 1290-1300Crossref PubMed Scopus (39) Google Scholar, 2Watanabe H. Wada T. Handa H. EMBO J. 1990; 9: 841-847Crossref PubMed Scopus (60) Google Scholar). Subsequent cDNA cloning and nucleotide sequence analyses (3Watanabe H. Sawada J.-i. Yano K. Yamaguchi K. Goto M. Handa H. Mol. Cell. Biol. 1993; 13: 1385-1391Crossref PubMed Scopus (91) Google Scholar, 4Sawada J.-i. Goto M. Sawa C. Watanabe H. Handa H. EMBO J. 1994; 13: 1396-1402Crossref PubMed Scopus (46) Google Scholar) have revealed that it is a human homologue of the rat GA-binding protein (hGABP), which binds to the GA motif of the herpes simplex virus immediate early genes and stimulates their transcription (5LaMarco K. Thompson C.C. Byers B.P. Walton E.M. McKnight S.L. Science. 1991; 253: 789-792Crossref PubMed Scopus (259) Google Scholar,6Thompson C.C. Brown T.A. McKnight S.L. Science. 1991; 253: 762-768Crossref PubMed Scopus (320) Google Scholar). E4TF1/hGABP is a unique transcription factor in its subunit composition. It is a heterotetramer (α2β2) of two 60-kDa (E4TF1–60/hGABPα) and two 53-kDa (E4TF1–53/hGABPβ) subunits. Structural and biochemical analyses (3Watanabe H. Sawada J.-i. Yano K. Yamaguchi K. Goto M. Handa H. Mol. Cell. Biol. 1993; 13: 1385-1391Crossref PubMed Scopus (91) Google Scholar, 4Sawada J.-i. Goto M. Sawa C. Watanabe H. Handa H. EMBO J. 1994; 13: 1396-1402Crossref PubMed Scopus (46) Google Scholar, 7Sawa C. Goto M. Suzuki F. Watanabe H. Sawada J.-i. Handa H. Nucleic Acids Res. 1996; 24: 4954-4961Crossref PubMed Scopus (47) Google Scholar, 8Suzuki F. Goto M. Sawa C. Ito S. Watanabe H. Sawada J.-i. Handa H. J. Biol. Chem. 1998; 273: 29302-29308Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar) have revealed that hGABPα contains an Ets-related DNA-binding domain and can bind to the DNA sequence 5′-CGGAAGTG-3′. However, hGABPα alone is unable to activate transcription, and the transactivator function is conferred by the formation of a heterotetramer with hGABPβ. In contrast to hGABPα, hGABPβ alone cannot bind to DNA, but it forms a homodimer through its leucine zipper-like structure at the carboxyl terminus and forms a heterodimer with hGABPα through its Notch/ankyrin repeat motif at the amino terminus. The resultant α2β2 heterotetramer has the capacity to activate transcription. Another non-DNA-binding subunit of 47 kDa, E4TF1–47/GABPγ, is structurally identical to hGABPβ except that it differs at its carboxyl extremity and lacks the homodimerization domain. GABPγ retains the ability to form a heterodimer with hGABPα but the resultant αγ heterodimer cannot activate transcription. The subunit composition is therefore one mechanism for the regulation of the transcriptional activity of hGABP (4Sawada J.-i. Goto M. Sawa C. Watanabe H. Handa H. EMBO J. 1994; 13: 1396-1402Crossref PubMed Scopus (46) Google Scholar, 7Sawa C. Goto M. Suzuki F. Watanabe H. Sawada J.-i. Handa H. Nucleic Acids Res. 1996; 24: 4954-4961Crossref PubMed Scopus (47) Google Scholar, 8Suzuki F. Goto M. Sawa C. Ito S. Watanabe H. Sawada J.-i. Handa H. J. Biol. Chem. 1998; 273: 29302-29308Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). In addition to the adenovirus E4 gene and the herpes simplex virus immediate early genes, an increasing number of cellular genes have been found to be targets of E4TF1/hGABP. In accordance with the fact that E4TF1/hGABP is a ubiquitously expressed transcription factor, its targets include ubiquitously expressed genes such as the genes for cytochrome c oxidase subunits IV and Vb (9Virbasius J.V. Virbasius C.A. Scarpulla R.C. Genes Dev. 1993; 7: 380-392Crossref PubMed Scopus (240) Google Scholar), the ATP synthase β-subunit (10Villena J.A. Vinas O. Mampel T. Iglesias R. Giralt M. Villarroya F. Biochem. J. 1998; 331: 121-127Crossref PubMed Scopus (46) Google Scholar), ribosomal proteins L30 and L32 (11Genuario R.R. Kelley D.E. Perry R.P. Gene Expr. 1993; 3: 279-288PubMed Google Scholar), and the retinoblastoma tumor suppressor protein (12Sowa Y. Shiio Y. Fujita T. Matsumoto T. Okuyama Y. Kato D. Inoue J. Sawada J.-i. Goto M. Watanabe H. Handa H. Sakai T. Cancer Res. 1997; 57: 3145-3148PubMed Google Scholar, 13Savoysky E. Mizuno T. Sowa Y. Watanabe H. Sawada J.-i. Nomura H. Ohsugi Y. Handa H. Sakai T. Oncogene. 1994; 9: 1839-1846PubMed Google Scholar). However, some tissue-specific genes, such as male-specific steroid 16α-hydroxylase (14Yokomori N. Kobayashi R. Moore R. Sueyoshi T. Negishi M. Mol. Cell. Biol. 1995; 15: 5355-5362Crossref PubMed Scopus (81) Google Scholar), leukocyte-specific cell adhesion molecule CD18 (β2 integrin) (15Rosmarin A.G. Caprio D.G. Kirsch D.G. Handa H. Simkevich C.P. J. Biol. Chem. 1995; 270: 23627-23633Abstract Full Text Full Text PDF PubMed Scopus (88) Google Scholar), neutrophil elastase (16Nuchprayoon I. Simkevich C.P. Luo M. Friedman A.D. Rosmarin A.G. Blood. 1997; 89: 4546-4554Crossref PubMed Google Scholar), interleukin-2 (17Avots A. Hoffmeyer A. Flory E. Cimanis A. Rapp U.R. Serfling E. Mol. Cell. Biol. 1997; 17: 4381-4389Crossref PubMed Scopus (46) Google Scholar), utrophin (18Khurana T.S. Rosmarin A.G. Shang J. Krag T.O. Das S. Gammeltoft S. Mol. Biol. Cell. 1999; 10: 2075-2086Crossref PubMed Scopus (104) Google Scholar, 19Gramolini A.O. Angus L.M. Schaeffer L. Burton E.A. Tinsley J.M. Davies K.E. Changeux J.P. Jasmin B.J. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 3223-3227Crossref PubMed Scopus (119) Google Scholar), and nicotinic acetylcholine receptor subunits (20Schaeffer L. Duclert N. Huchet-Dymanus M. Changeux J.P. EMBO J. 1998; 17: 3078-3090Crossref PubMed Scopus (136) Google Scholar), have also been demonstrated to be regulated by E4TF1/hGABP. Despite accumulating evidence for both ubiquitous and tissue-specific gene regulation by E4TF1/hGABP, how this regulation is achieved is still unknown. One of the most likely control mechanisms of such transcriptional regulation is an interaction with other transcription factors. In fact, we and others have demonstrated that E4TF1/hGABP synergistically activates transcription through physical interaction with the ATF1, CREB (21Sawada J.-i. Simizu N. Suzuki F. Sawa C. Goto M. Hasegawa M. Imai T. Watanabe H. Handa H. J. Biol. Chem. 1999; 274: 35475-35482Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar), SP1 (22Rosmarin A.G. Luo M. Caprio D.G. Shang J. Simkevich C.P. J. Biol. Chem. 1998; 273: 13097-13103Abstract Full Text Full Text PDF PubMed Scopus (88) Google Scholar), SP3 (23Galvagni F. Capo S. Oliviero S. J. Mol. Biol. 2001; 306: 985-996Crossref PubMed Scopus (65) Google Scholar), and HCF transcription factors (24Vogel J.L. Kristie T.M. EMBO J. 2000; 19: 683-690Crossref PubMed Scopus (73) Google Scholar). In contrast, E4TF1/hGABP activity is inhibited by interaction with the mi transcription factor (MITF) in mast cells (25Morii E. Ogihara H. Oboki K. Sawa C. Sakuma T. Nomura S. Esko J.D. Handa H. Kitamura Y. Blood. 2001; 97: 3032-3039Crossref PubMed Scopus (31) Google Scholar). To understand the mechanism of transcriptional regulation by E4TF1/hGABP, it is important to clarify the regulatory cross-talk that occurs with other transcription factors and cofactors. To this end, we screened a cDNA expression library for genes whose products interact with the hGABPβ subunit, and we isolated a cofactor, YEAF1/RYBP. We demonstrate here that YEAF1/RYBP represses the transcriptional activity of E4TF1/hGABP, whereas its close relative, YAF-2 (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar), activates the activity of E4TF1/hGABP. These results provide the first example of a cofactor family with functionally distinct members. A yeast two-hybrid screen was performed using a modified version of the system of Fields and Song (27Fields S. Song O. Nature. 1989; 340: 245-246Crossref PubMed Scopus (4863) Google Scholar). Briefly, pBTM116/hGABPβ-(249–383), encoding a LexA-fusion protein with amino acids 249–383 of hGABPβ, was transformed into the yeast strain L40 by the lithium acetate method. The resultant strain was then transformed with a GAL4-activation domain fusion cDNA library constructed from mRNAs of HeLa cells (MATCHMAKER HL4000AA, CLONTECH). The transformants were plated onto a selective medium for histidine prototrophy (−Trp, −Leu, −Ura, −Lys, −His, and 10 mm 3-aminotriazole (3-AT)) and were incubated at 30 °C for 5 days. His+ colonies were then grown in a liquid selective (−Trp, −Leu) medium until theA 600 reached 1.0 to 1.2 and were further tested for β-galactosidase activity as described previously (13Savoysky E. Mizuno T. Sowa Y. Watanabe H. Sawada J.-i. Nomura H. Ohsugi Y. Handa H. Sakai T. Oncogene. 1994; 9: 1839-1846PubMed Google Scholar). Each GAL4 fusion prey plasmid was rescued from the 3-AT-resistant and β-galactosidase-positive yeast clones and transformed intoEscherichia coli (DH5α). For the interaction assay, the L40 yeast strain was transformed with an appropriate LexA fusion plasmid and GAL4-activation domain fusion plasmid (pGAD424; CLONTECH) and was plated onto a selective medium (−Trp, −Leu). Three independent transformed colonies were then assayed for 3-AT sensitivity and β-galactosidase activity. For three-hybrid assay, pBridge Three-Hybrid Vector (CLONTECH) was used to express GAL4-DNA-binding domain fusion and bridge proteins. The expression of the bridge protein can be inhibited by addition of methionine into the medium. LexA fusion or GAL4-activation domain fusion plasmids used for yeast two-hybrid interaction assays were constructed by insertion of PCR-amplified fragments into appropriate sites of pBTM116 (a generous gift of Dr. Hollenberg) or pGAD424 (CLONTECH), respectively. A human YY1 cDNA (28Shi Y. Seto E. Chang L.S. Shenk T. Cell. 1991; 67: 377-388Abstract Full Text PDF PubMed Scopus (819) Google Scholar) was a gift from Dr. T. Shenk. A human YAF-2 cDNA was obtained by screening a HeLa cDNA library using a YEAF1 cDNA probe. AnEcoRI-XhoI fragment containing a full-length YEAF1 open reading frame was excised from pGAD424/YEAF1 and inserted into EcoRI/XhoI-digested pGEX5X3 to make pGEX5X3/YEAF1. A glutathione S-transferase (GST)-YEAF1 protein was expressed in E. coli BL21 (DE3) and purified. GST-YEAF1 was then immobilized onto Sensor Chip CM5 (Biacore AB) (∼2500 resonance units) via an anti-GST antibody, and the chip was used to analyze the interaction using BIACORE2000 as described previously (8Suzuki F. Goto M. Sawa C. Ito S. Watanabe H. Sawada J.-i. Handa H. J. Biol. Chem. 1998; 273: 29302-29308Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). Recombinant hGABPβ and hGABPγ proteins expressed inE. coli were prepared as described previously (3Watanabe H. Sawada J.-i. Yano K. Yamaguchi K. Goto M. Handa H. Mol. Cell. Biol. 1993; 13: 1385-1391Crossref PubMed Scopus (91) Google Scholar). A mammalian expression plasmid for FLAG-tagged YEAF1 was constructed by inserting a FLAG sequence and YEAF1 cDNA fragment into a pCAGGS vector. Four micrograms of the resultant plasmid pCAGGS/FLAG-YEAF1 or of pCAGGS/FLAG, which expresses the FLAG peptide only, was transfected into 1 × 105 HeLa cells in a 100-mm dish using the Effectene reagent (Qiagen). Forty eight hours after transfection, cells were lysed with 500 μl of lysis buffer (20 mm HEPES, pH 7.9, 50 mm KCl, 10 μm ZnSO4, 1 mm phenylmethylsulfonyl fluoride, 1 μg/ml pepstatin, 1.0% Nonidet P-40). The total cell extract (400 μl) was incubated with 20 μl of anti-FLAG M2 antibody-conjugated resin (Sigma) for 8 h at 4 °C, and the resin was washed three times with wash buffer (20 mm HEPES, pH 7.9, 50 mm KCl, 10 μm ZnSO4, 1 mmphenylmethylsulfonyl fluoride, 1 μg/ml pepstatin, 0.01% Nonidet P-40). The precipitated complex was eluted by boiling the resin in 40 μl of SDS sample buffer. The eluate and the total cell extract were separated on a 10% SDS-polyacrylamide gel, transferred to a polyvinylidene difluoride membrane, and stained with monoclonal mouse anti-FLAG (Sigma), monoclonal mouse anti-hGABPβ, or polyclonal rabbit anti-YY1 (Santa Cruz Biotechnology) antibodies using ECL detection reagents (Amersham Biosciences). Transfection of Schneider's 2 (SL2)Drosophila melanogaster cells, the luciferase assay, and the β-galactosidase assay were performed as described previously (7Sawa C. Goto M. Suzuki F. Watanabe H. Sawada J.-i. Handa H. Nucleic Acids Res. 1996; 24: 4954-4961Crossref PubMed Scopus (47) Google Scholar). A luciferase reporter plasmid containing the human retinoblastoma susceptibility (Rb) gene promoter (pRB-luciferase) and expression plasmids for hGABPβ, hGABPγ, and hGABPα have also been described (12Sowa Y. Shiio Y. Fujita T. Matsumoto T. Okuyama Y. Kato D. Inoue J. Sawada J.-i. Goto M. Watanabe H. Handa H. Sakai T. Cancer Res. 1997; 57: 3145-3148PubMed Google Scholar). The expression plasmids in fly cells for YEAF1, YEAF1 (CR1OR), and YAF-2 (A5CΔP/YEAF1, A5CΔP/YEAF1 (CR1OR), and A5CΔP/YAF-2) were constructed by inserting each DNA fragment into an A5CΔP expression plasmid. A human multiple tissue Northern blot (CLONTECH) was hybridized with32P-labeled YEAF1, YAF-2, hGABPβ, and actin cDNA probes as recommended by the manufacturer. In order to understand the mechanism of transcriptional regulation by E4TF1/hGABP, it is important to clarify the regulatory cross-talk that occurs with other transcription factors and/or cofactors. We used the yeast two-hybrid method to screen a cDNA expression library for genes whose products interact with the non-DNA-binding subunit of E4TF1/hGABP (E4TF1-53/hGABP). We used an amino-terminally truncated form of hGABPβ as bait (LexA-hGABPβ-(249–383)) to avoid interaction with the DNA-binding subunit hGABPα. This fusion product retains the transactivator and self-association domains (see Fig. 2 A). By screening a HeLa cDNA library, we obtained 33 positive colonies out of ∼1 × 106 transformants. Isolation of the plasmids and subsequent restriction-digestion analysis of the inserted cDNAs revealed that all of these plasmids contained a single cDNA species of 1.1 kb in length. Nucleotide sequencing analysis revealed that this was a novel cDNA that showed similarity to the previously identified YY1-associated factor 2(YAF-2) (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar). By plaque hybridization using this partial cDNA fragment as a probe and 3′-rapid amplification of cDNA ends, we obtained a cDNA contig of 4765 bp (data not shown, GenBankTM accession number AB029551). The cDNA contig (designated as YY1 and E4TF1/hGABP associated factor-1 (YEAF1), see below) contained a putative poly(A) addition signal and poly(A) stretch at the 3′ end of the cDNA and two ATG codons at the 5′ end of its single open reading frame. We speculated that the second ATG codon might be the initiator codon, because it showed similarity to the putative initiator codon of YAF-2 (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar), and its surrounding sequence completely matches the Kozak's consensus sequence. Based on this information and the mRNA length predicted from a Northern blot analysis (5.0 kb, see below), we believe that we obtained nearly full-length cDNA. The YEAF1 protein contained no significant protein motif other than two potential zinc fingers at its amino-terminal region (Fig.1 A). As mentioned above, the protein was similar to the YAF-2 protein (Fig.1 A) reported previously (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar). YEAF1 and YAF-2 shared the highest homology at the amino-terminal domain containing the two zinc finger motifs and moderate similarity at the carboxyl-terminal domain, which we refer to as conserved region (CR) 1 and 2, respectively (Fig.1 B). YEAF1 also contains a unique region between CR1 and CR2 that is absent in YAF-2, and we refer to this as the YEAF1 original region (OR) (Fig. 1 B). Later on, YEAF1 was found to be highly homologous to mouse RYBP (Ring1 andYY1-binding protein) (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar), indicating that YEAF1 and RYBP are species homologues. Specific binding of YEAF1 to hGABPβ was verified by the yeast two-hybrid assay as monitored by 3-AT sensitivity and by β-galactosidase assay (Fig. 2,A and B). A GAL4AD-YEAF1 fusion protein interacted with a LexA fusion protein of full-length hGABPβ and an amino-terminally truncated hGABPβ-(249–383) that was used as bait for two-hybrid screening, indicating that the amino-terminal ankyrin repeat motifs required for interaction with hGABPα are dispensable for interaction with YEAF1. A more extensive amino-terminal deletion mutant, hGABPβ-(311–383), no longer associated with YEAF1 despite retaining the leucine zipper-like structure required for homodimerization and transactivation. These results indicate that a region of hGABPβ spanning amino acids 249–310 is required for association with YEAF1. Accordingly, hGABPγ, an alternatively spliced form of hGABPβ that retains this region but lacks the abilities to form homodimer and to transactivate, was able to interact with YEAF1. In contrast, the LexA fusion of the DNA-binding subunit hGABPα did not interact with GAL4AD-YEAF1, although it did interact with GAL4AD-hGABPβ (data not shown). Furthermore, we could not detect an interaction of YEAF1 with the unrelated ATF1 transcription factor (Fig.2 B) nor with the α- and β-subunits of casein kinase II (data not shown). These results indicate that YEAF1 specifically interacts with hGABPβ. We also found that YEAF1 interacts with YY1 (Fig. 2 B), as would be expected based on its similarity to YAF-2, which was originally identified as an interacting partner of YY1 (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar). Therefore, YEAF1 interacts with both hGABPβ and YY1. From these observations, we designated this protein YEAF1 (for YY1- andE4TF1/hGABP-associatedfactor-1). We next determined the domain of YEAF1 that is required for association with hGABPβ. A series of deletion mutants of YEAF1 were fused to GAL4AD and were assayed for interaction with LexA-hGABPβ in yeast (Fig.3 A). As was evident by both 3-AT sensitivity and by β-galactosidase assays (Fig. 3 B), the carboxyl-terminal conserved region 2 (CR2) of YEAF1 was necessary for interaction with hGABPβ. The fact that the CR2 of YEAF1 has sequence similarity with the corresponding region of YAF-2 prompted us to test the interaction of YAF-2 and hGABPβ. As expected, we were able to detect the interaction of the GAL4AD-YAF-2 and LexA-hGABPβ fusion proteins by a yeast two-hybrid assay (Fig. 3 B). Therefore, YEAF1 and YAF-2 constitute a protein family that specifically interacts with the hGABP transcription factor. We next analyzed the kinetics of the hGABPβ and YEAF1 interaction using surface plasmon resonance (SPR). A purified recombinant GST fusion protein with full-length YEAF1 (GST-YEAF1) was immobilized onto the sensor chip surface via a previously coupled anti-GST antibody. By injecting purified recombinant hGABPβ protein at various concentrations into the immobilized or control sensor chips, we measured real time SPR at the association and dissociation phases of interaction of hGABPβ and YEAF1. Fig.4 A shows the sensorgrams obtained by subtracting the background values. The specific interaction of hGABPβ and YEAF1 was dose-dependent, indicating that hGABPβ and YEAF1 bind directly in vitro. From these sensorgrams, the k d ,k a , and K D values of the hGABPβ/YEAF1 interaction were calculated (Fig. 4 B) and are summarized in Fig. 4 C, together with values previously obtained using hGABPβ and hGABPα (8Suzuki F. Goto M. Sawa C. Ito S. Watanabe H. Sawada J.-i. Handa H. J. Biol. Chem. 1998; 273: 29302-29308Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar). The equilibrium dissociation constant (K D ) value of hGABPβ and YEAF1 was 8.0 × 10−9m, which is about 10 times higher than the K D value for the hGABPβ and hGABPα subunits. We also analyzed the interaction between hGABPγ and YEAF1 in a similar manner, and we obtained a similar K D value (3.7 × 10−9m) (Fig. 4 C and data not shown). To test whether YEAF1 forms a functional complex with hGABPβ or YY1 in vivo, we first constructed an expression plasmid containing FLAG-tagged full-length YEAF1 protein and transfected the plasmid into HeLa cells. Immunofluorescent staining of the transfected cells with an anti-FLAG antibody revealed that the FLAG-YEAF1 protein is predominantly localized in the nucleus (data not shown). Similar nuclear staining was observed when using COS-1 cells transfected with a full-length YEAF1 expression plasmid (without the tag) and anti-YEAF1 antiserum (data not shown). Nuclear localization of YEAF1 seems quite reasonable because both hGABP and YY1 are nuclear transcription factors. We then prepared nuclear extracts from the transfected cells and subjected them to immunoprecipitation using the anti-FLAG antibody. Aliquots of the precipitate were analyzed by immunoblotting with anti-FLAG, anti-hGABPβ, and anti-YY1 antisera. As shown in Fig.5, both hGABPβ and YY1 were detected in the anti-FLAG precipitate but not in the control precipitate, demonstrating that YEAF1 forms a complex with both hGABPβ and YY1 in the nucleus. The result shown above does not necessarily indicate that hGABP, YEAF1, and YY1 form a ternary complex. To test this possibility, we performed a yeast three-hybrid assay schematically show in Fig.6 A. In addition to the GAL4DB and GAL4AD fusion proteins, the third protein ("bridge" protein) is expressed in yeast to test a ternary complex formation. The expression of the bridge protein can be down-regulated by addition of methionine into the medium. The background β-galactosidase activity was relatively high because full-length hGABPβ was fused to the GAL4DB, and we could see no evidence for interaction of the GAL4DB-hGABPβ with the GAL4AD-YY1 (Fig. 6 B, lane 1). Co-expression of an intact YEAF1 protein resulted in significant β-galactosidase expression, and reduction of the YEAF1 expression level by addition of methionine resulted in decrease of the β-galactosidase activity (Fig.6 B, lanes 2 and 3). These results indicate that these three proteins form a ternary complex in yeast. We further showed that hGABPα, hGABPβ, and YEAF1 form a ternary complex (Fig. 6 B, lanes 4–6). These results together suggest that hGABPα, hGABPβ, YEAF1, and YY1 form a complex, and YEAF1 acts as a bridging factor of the hGABP and YY1 transcription factors. We next examined the effects of YEAF1 and YAF-2 on the biological activity of hGABP. To this end, we measured transcriptional activity of hGABP by transient transfection and a luciferase assay. We used a human retinoblastoma (Rb) gene promoter-luciferase construct (pRb-luciferase) as a reporter (Fig.7 A) and the D. melanogaster Schneider's line 2 (SL2) cell line as a recipient cell, because the Rb promoter contains an hGABP-binding site (12Sowa Y. Shiio Y. Fujita T. Matsumoto T. Okuyama Y. Kato D. Inoue J. Sawada J.-i. Goto M. Watanabe H. Handa H. Sakai T. Cancer Res. 1997; 57: 3145-3148PubMed Google Scholar, 13Savoysky E. Mizuno T. Sowa Y. Watanabe H. Sawada J.-i. Nomura H. Ohsugi Y. Handa H. Sakai T. Oncogene. 1994; 9: 1839-1846PubMed Google Scholar), and SL2 cells contain little endogenous hGABP-like activity (7Sawa C. Goto M. Suzuki F. Watanabe H. Sawada J.-i. Handa H. Nucleic Acids Res. 1996; 24: 4954-4961Crossref PubMed Scopus (47) Google Scholar). As we have reported previously (7Sawa C. Goto M. Suzuki F. Watanabe H. Sawada J.-i. Handa H. Nucleic Acids Res. 1996; 24: 4954-4961Crossref PubMed Scopus (47) Google Scholar), transfection of an expression plasmid containing the DNA-binding subunit hGABPα alone had little effect on pRb-luciferase reporter activity (Fig. 7 B, lanes 1 and2), whereas co-expression of the hGABPβ subunit (but not the hGABPγ subunit) resulted in significant activation of luciferase activity (Fig. 7 B, lanes 5 and 8). Noticeably, co-transfection of an increasing amount of YEAF1 expression plasmid together with hGABPα and hGABPβ resulted in a dose-dependent decrease of luciferase activity (Fig.7 B, lanes 9 and 10), indicating that YEAF1 acts as a transcriptional co-repressor. The repressive effect was not observed when a YEAF1 mutant defective in hGABPβ binding (CR1OR, see Fig. 3 A) was used (Fig. 7 B, lanes 11 and 12). Instead, we reproducibly observed an increase of the luciferase activity. Possible explanation for this result will be discussed below. In contrast to YEAF1, the transcriptional activity of hGABP was enhanced by co-transfection of an increasing amount of YAF-2, indicating that YAF-2 acts as a transcriptional co-activator for hGABP (Fig. 7 C). Therefore, YEAF1 and YAF-2 exhibited opposite effects on the transcriptional activity of hGABP, despite being structurally related. We next examined mRNA expression of YEAF1, YAF-2 , andhGABPβ in various tissues using specific DNA probes. AYEAF1 mRNA of about 5 kb was detected in all the tissues we examined, with the highest expression level in placenta (Fig.8, top panel). Consistent with a previous report that YAF-2 is expressed in muscle cells (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar), YAF-2 mRNA expression was the highest in heart and skeletal muscle (Fig. 8, 2nd panel). As we had mentioned previously (8Suzuki F. Goto M. Sawa C. Ito S. Watanabe H. Sawada J.-i. Handa H. J. Biol. Chem. 1998; 273: 29302-29308Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar), hGABPβ mRNA was expressed ubiquitously (Fig. 8, 3rd panel). The expression level ofhGABPβ mRNA was the highest in placenta, heart, and skeletal muscle, where YEAF1 or YAF-2 mRNA were also highly expressed. Although more detailed expression profiling is necessary, these results support the idea that YEAF1 and YAF-2 are tissue-specific cofactors for the hGABP transcription factor. In this paper, we described the isolation and characterization of YEAF1, a novel interactor of the hGABP/E4TF1 transcription factor. We demonstrated that YEAF1 and its close relative YAF-2 also interact with another transcription factor YY1 and that they act as bridging factors for GABP and YY1. Despite their structural similarity, YEAF1 and YAF-2 were functionally distinct, in that YEAF1 negatively regulated the transcriptional activity of hGABP but YAF-2 positively regulated the activity. We demonstrated here that YEAF1 and YAF-2 constitute a family of cofactors for the hGABP and YY1 transcription factors. The first identified member, YAF-2, was originally isolated as a factor interacting with YY1 in yeast two-hybrid screening (26Kalenik J.L. Chen D. Bradley M.E. Chen S.J. Lee T.C. Nucleic Acids Res. 1997; 25: 843-849Crossref PubMed Scopus (85) Google Scholar). As expected from their sequence similarity, we showed that YEAF1 also interacts with YY1, and YAF-2 interacts with hGABPβ. We did not find any other family members of YEAF1 and YAF-2 in a search of the human genome using the GenBankTM data base, but we did identify homologous putative genes, CG12190 andC54H2.3, in the D. melanogaster andCaenorhabditis elegans genomes, respectively, 2C. Sawa, T. Yoshikawa, F. Matsuda-Suzuki, S. Deléhouzée, M. Goto, H. Watanabe, J.-i. Sawada, K. Kataoka, and H. Handa, unpublished observations. suggesting that they are evolutionarily conserved at least among multicellular organisms. YEAF1 is the human homologue of the mouse RYBP. RYBP was isolated as an interacting partner of Ring1A protein, a member of the Polycomb group (PcG) of proteins (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar). The PcG proteins form large complexes that are necessary for the maintenance of the transcriptionally repressed state of a number of genes. It has been shown that RYBP also interacts with another PcG protein, M33, a mouse homologue of Drosophila polycomb (Pc), and with YY1 (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar). YY1 has similarity to the Drosophila pleiohomeotic (pho) gene product (30Brown J.L. Mucci D. Whiteley M. Dirksen M.L. Kassis J.A. Mol. Cell. 1998; 1: 1057-1064Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar), and accumulating evidence suggests that YY1 is also involved in PcG function (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar, 31Satijn D.P. Hamer K.M. den Blaauwen J. Otte A.P. Mol. Cell. Biol. 2001; 21: 1360-1369Crossref PubMed Scopus (155) Google Scholar). Thus, RYBP is considered to be a component of PcG complexes and accordingly acts as a transcriptional repressor when fused with the DNA-binding domain of the GAL4 transcription factor (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar). The most plausible model of the repressor function of YEAF1/RYBP is that they recruit PcG complexes. This idea is supported by the recent report (32Trimarchi J.M. Fairchild B. Wen J. Lees J.A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 1519-1524Crossref PubMed Scopus (217) Google Scholar) that RYBP interacts with the repressor domain of E2F6, a distantly related member of the E2F transcription factors. E2F6 has been shown to form complexes with other PcG proteins such as Ring1, Bmi1, MEL-18, and Mph1. We demonstrated here that YEAF1 binds to hGABP and represses its transcriptional activity. Therefore, YEAF1/RYBP mediates the transcriptional repression of at least two sequence-specific DNA-binding factors, E2F6 and hGABP. YY1 may also be the target of YEAF1 repression because YY1 can act as a transcriptional repressor depending on the promoter context. It seems consistent that E2F6 (and possibly YY1) actively represses transcription by interacting with PcG complexes. In contrast to E2F6 and YY1, such an active repressor function or active repressor domain has never been assigned to hGABPβ. Although the YEAF1 interaction domain of hGABPβ overlaps with the transactivator domain, a more detailed domain analysis may reveal an active repressor domain. Alternatively, YEAF1 may not be able to recruit PcG to hGABP and may reduce the transcriptional activity of hGABP by competition with co-activators such as p300/CBP for binding to hGABP. These two possible mechanisms are not mutually exclusive, and further analysis is necessary to fully understand the molecular mechanism of transcriptional repression by YEAF1. Despite the similar affinity of YEAF1 and YAF-2 for hGABPβ, YAF-2 activated hGABP transcriptional activity and YEAF1 repressed it. As far as we know, YEAF1 and YAF-2 are the only structurally related transcriptional cofactors that have opposite functions. It has been reported that the carboxyl-terminal regions of RYBP/YEAF1, OR and CR2, are necessary for transcriptional repressor function. YAF-2 may lack the ability to repress transcription because it lacks the OR domain, and its CR2 region is relatively divergent. If this is the case, one possible explanation for the opposite functions of YEAF1 and YAF-2 is that YAF-2 lacks the ability to bind to PcG complexes. Alternatively, YAF-2 may have an intrinsic transactivation domain. It should be noted that co-expression with hGABP of a truncated YEAF1 protein (CR1OR) that lacks CR2 and the ability to bind to hGABP resulted in an increase of Rb promoter activity (Fig. 7). This mutant YEAF1 may compete for binding to endogenous YEAF1 with PcG complexes, which may lead to accumulation of non-functional PcG complexes and to the activation of transcriptional activity. We further showed that both hGABP, YEAF1, and YY1 form a ternary complex. As both hGABP and YY1 are sequence-specific DNA-binding proteins, YEAF1 (and probably YAF-2) should be able to bridge hGABP and YY1 when they bind to DNA. It is noteworthy that some promoters contain binding sites for both the hGABP and YY1 transcription factors. For example, the P6 promoter of the human B19 parvovirus contains adjacent hGABP- and YY1-binding sites (33Vassias I. Hazan U. Michel Y. Sawa C. Handa H. Gouya L. Morinet F. J. Biol. Chem. 1998; 273: 8287-8293Abstract Full Text Full Text PDF PubMed Scopus (23) Google Scholar). The activity of this promoter was activated by hGABP and repressed by YY1. Because YEAF1/RYBP is a component of PcG complexes (29Garcia E. Marcos-Gutierrez C. del Mar Lorente M. Moreno J.C. Vidal M. EMBO J. 1999; 18: 3404-3418Crossref PubMed Scopus (192) Google Scholar), binding of YY1 to the promoter may recruit YEAF1 and PcG complexes to hGABP, which may result in transcriptional repression. This may be not the case for cytochromec oxidase subunit genes (34Lenka N. Vijayasarathy C. Mullick J. Avadhani N.G. Prog. Nucleic Acids Res. Mol. Biol. 1998; 61: 309-344Crossref PubMed Google Scholar), whose promoter regions also contain binding sites for hGABP and YY1, but both of which in this case have been shown to be necessary for efficient transcription. One can speculate that bridging by YAF-2 may enhance transcription by hGABP and YY1. However, a more detailed analysis of the regulation of such promoters is necessary to understand the mechanism of positive and negative regulation by hGABP and YY1. Isolation and characterization of the functionally distinct bridging factors YEAF1 and YAF-2 described here may elucidate the molecular mechanisms of transcriptional regulation by hGABP and YY1, as well as by PcG complexes. We are grateful to S. M. Hollenberg for gifts of the yeast strains and the plasmid pBTM116 and its derivatives and to Dr. T. Yamamoto for assistance with the yeast procedures. We thank T. Shenk for the gift of YY1 cDNA. We also thank Drs. T. Wada and T. Imai for helpful discussions and advice.