The Cell Cycle Regulatory Factor TAF1 Stimulates Ribosomal DNA Transcription by Binding to the Activator UBF

Current Biology, Vol. 12, 2142–2146, December 23, 2002, 2002 Elsevier Science Ltd. All rights reserved.

PII S0960-9822(02)01389-1

The Cell Cycle Regulatory Factor TAF1 Stimulates Ribosomal DNA Transcription by Binding to the Activator UBF Chih-Yin Lin,2 JoAnn Tuan,2,3 Pierluigi Scalia, Tiffany Bui, and Lucio Comai1 Department of Molecular Microbiology and Immunology Keck School of Medicine University of Southern California Los Angeles, California, 90033

Summary Control of ribosome biogenesis is a potential mechanism for the regulation of cell size during growth [1, 2], and a key step in regulating ribosome production is ribosomal RNA synthesis by RNA polymerase I (Pol I) [3, 4]. In humans, Pol I transcription requires the upstream binding factor UBF and the selectivity factor SL1 to assemble coordinately on the promoter [5–7]. UBF is an HMG box-containing factor that binds to the rDNA promoter and activates Pol I transcription through its acidic carboxy-terminal tail [8, 9]. Using UBF (284–670) as bait in a yeast two-hybrid screen, we have identified an interaction between UBF and TAF1, a factor involved in the transcription of cell cycle and growth regulatory genes [10]. Coimmunoprecipitation and protein-protein interaction assays confirmed that TAF1 binds to UBF. Confocal microscopy showed that TAF1 colocalizes with UBF in Hela cells, and cell fractionation experiments provided further evidence that a portion of TAF1 is localized in the nucleolus, the organelle devoted to ribosomal DNA transcription. Cotransfection and in vitro transcription assays showed that TAF1 stimulates Pol I transcription in a dosage-dependent manner. Thus, TAF1 may be involved in the coordinate expression of Pol I- and Pol II-transcribed genes required for protein biosynthesis and cell cycle progression.

Results and Discussion The RNA Pol I Transcription Factor UBF Interacts with TAF1/CCG1 The region of UBF from amino acid 284 to 670 was chosen as bait in a yeast two-hybrid system for screening a human testis cDNA library. This region of UBF, which includes HMG boxes 3 and 4 and the region from amino acid 492 to 670 (referred as the X-domain), is required for optimal transcriptional activation [11]. UBF (284–670) lacks a transactivation domain (amino acids 670–765), and when fused to the Gal4DNA binding domain, it is transcriptionally inactive in yeast. A cDNA clone that interacted with UBF based on nutritional selection and ␤-galactosidase activity was selected for 1 2 3

Correspondence: [email protected] These authors contributed equally to this work. Present address: Amgen Inc., Thousand Oaks, California 91320.

further analysis (Figure 1A). The clone was sequenced, and a nucleotide database search revealed that it is identical to human TAF1 (previously known as TAFII250) [12, 13]. TAF1 is the largest subunit of the TFIID complex and is identical to a factor that has been implicated in the regulation of G1-to-S progression (CCG1) [14–16]. The clone isolated from the yeast two-hybrid screen encodes for the carboxy-terminal portion of TAF1, amino acids 1096–1872. Additional yeast two-hybrid experiments indicated that the region of UBF spanning HMG box 4 and the X-domain is sufficient for binding to TAF1 (Figure 1A). To obtain independent evidence for an interaction between TAF1 and UBF, we coinfected Sf9 cells with a baculovirus encoding for HA-tagged TAF1 and with baculoviruses expressing flag-tagged wild-type UBF (F-UBF) or the mutants UBFdx, UBFdb34, UBF381N, and UBF491N. The results of these experiments indicated that TAF1 coimmunoprecipitated with full-length UBF but not with a control protein (HCV-polymerase) (Figure 1B). In contrast, a UBF mutant (UBFdx) missing the X-domain failed to bind to TAF1, suggesting that this region of UBF is required for the interaction. Furthermore, a mutant form (UBFdb34) of UBF lacking the HMG boxes 3 and 4 bound poorly to TAF1. This finding was extended by the observation that the binding of UBF381N and UBF491N to TAF1 is weak compared to that of full-length UBF (compare lane 6 with 7 and 8). These results suggest that the region containing HMG boxes 3 and 4 contributes to the stable association between UBF and TAF1. A possible interpretation of these data is that the TAF1 binding domain overlaps with HMG box 4. Alternatively, HMG boxes 3 and 4 may provide the structural information necessary for efficient TAF1 binding. In a parallel set of experiments with TAF1 deletion mutants, we determined that the region of TAF1 spanning amino acids 1054–1425 is required for UBF binding (See Figure S1 in the Supplementary Material available with this article online). Taken together, these results show that TAF1 interacts with UBF and confirm the finding of the yeast two-hybrid screen.

A Subpopulation of TAF1 Colocalizes and Is Associated with UBF in HeLa Cells To further examine the link between TAF1 and UBF, we asked whether the two proteins colocalize in human cells. HeLa cells were transfected with flag-TAF1, and the localization of the transfected TAF1 and endogenous UBF was analyzed by confocal immunofluorescence microscopy. UBF is found in discrete foci within the nucleus, typical of the pattern of localization of this nucleolar protein (Figure 2A, panel a⬘), whereas the pattern of TAF1 labeling is more evenly distributed throughout the nucleus (Figure 2A, panel a″) [17]. Importantly, merged images of both stainings indicate that a subpopulation of TAF1 colocalizes with UBF (Figure 2A, panel a⵮). An identical pattern of staining and colocalization between UBF and a portion of TAF1 is also observed with cryo-

Brief Communication 2143

Figure 1. TAF1 and UBF Interact with Each Other in the Yeast Two-Hybrid Assay and When Coexpressed in Insect Cells (A) Gal4 fusion proteins used in the yeast two-hybrid assays are shown with the results of the ␤-galactosidase assay and the growth in deficient media. Numbers in parentheses indicate amino acid residues. Vectors and the library used in the assays were purchased from Clontech Laboratories. pUA3-1 and pTD1-1 represent plasmids used as controls for positive interaction. PAS2-1 and pACT2 represent the GAL4 DNA binding domain and activation domain vectors, respectively. (B) Sf9 cells were coinfected with recombinant baculoviruses expressing HA-tagged TAF1 and flag-UBF, a series of flag-UBF mutants, or flag-HCV polymerase (HCV Pol). Forty-eight hours after infection, cell extracts were prepared and incubated with anti-flag affinity resin (Kodak) for immunoprecipitation of the bound proteins. Immunocomplexes were analyzed by SDS-PAGE, with subsequent immunoblotting and detection with anti-TAF1 antibodies (Santa Cruz Biotechnology). An aliquot of each extract was also used in Western blots with antibodies against TAF1 and the flag epitope for determination of protein expression. The amounts of extracts used in the immunoprecipitations were normalized for the same content of flag-tagged protein. (C) A schematic representation of UBF mutants used in the Sf9 coinfection experiments.

embedded thin sections of HeLa cells (Figure 2A, b⬘–b⵮). This pattern is specific for TAF1 because the staining of a transfected HA-tagged GRIP1 construct, a member of the p160 family of coactivators, does not overlap with UBF (our unpublished data). In addition, no specific pattern of fluorescence was observed when the primary antibody was omitted from the protocol (our unpub-

lished data). To provide additional evidence that a subpopulation of TAF1 is localized in the nucleolus, we isolated nuclear (soluble ⫹ chromatin), nucleolar, and nucleoplasmic fractions from Hela cells and determined the localization of TAF1, UBF, TAFI110, and Sp1 by immunoblot analysis. The results of these experiments show that a portion of TAF1 is found in the nucleolus,

Current Biology 2144

tion, are found in the nucleolar fraction and are excluded from the nucleoplasm (lane 3). Because equal amounts of nucleoplasmic and nucleolar extracts were loaded on the gel, the seeming lower abundance of TAF1 in the nucleoplamic fraction is likely to be the result of TAF1 being less represented among the soluble nucleoplasmic proteins. Taken together, these findings indicate that a subpopulation of TAF1 is localized in the nucleolus and is likely to play a role that is distinct from its function in the TFIID complex. To further demonstrate that the nucleolar factor UBF interacts with TAF1 in Hela cells, we examined whether the two proteins can be coimmunoprecipitated from cell extracts. For this purpose, nuclear extracts from HeLa cells were incubated with antiserum against UBF or a control preimmune serum, and the immunocomplexes were then precipitated by incubation with protein A-agarose. Proteins bound to the antibody-affinity bead complex were resolved by SDS-PAGE and analyzed by Western blotting with affinity-purified antibodies against UBF, TAF1, and TAF4 (previously named TAFII130). The results of the immunoprecipitation assays shown in Figure 2C indicate that TAF1, but not TAF4, coimmunoprecipitates with UBF from a nuclear extract. In addition, neither UBF nor TAF1 were immunoprecipitated by the control preimmune serum.

Figure 2. A Subpopulation of TAF1 Colocalizes with UBF Confocal laser-scanning microscopy. Paraformaldehyde-fixed (a⬘– a⵮) or cryodissected (b⬘–b⵮; 8 ␮m sections) Hela cells were labeled with anti-UBF antibodies and TRITC-conjugated (a⬘) or FITC-conjugated (b⬘) anti-rabbit antibody, as well as with anti-flag antibodies and either FITC-conjugated (a″) or TRITC-conjugated (b″) antimouse antibodies. Merged images are shown in panels a⵮ and b⵮. Yellow color indicates colocalization of both labels. Identical results were obtained with transfected HA-tagged TAF1. (B) Nucleolar localization of TAF1. Forty-five or thirty (Sp1 immunoblot) ␮g of total nuclear (soluble fraction ⫹ chromatin; lane 1), nucleolar (lane 2), and nucleoplasmic (lane 3) extracts was resolved on SDS-PAGE, transferred to nitrocellulose, and analyzed with antibodies against UBF, TAF1, TAFI110, and Sp1 (see Supplementary Material for Experimental Procedures). (C) UBF interacts with TAF1 in human cells. Nuclear extracts from approximately 1 ⫻ 108 cells were incubated with either anti-UBF serum or a pre-immune serum. The resulting immunocomplexes were precipitated with protein A-sepharose, resolved by SDS-PAGE, and analyzed with antibodies against TAF1 (top) TAF4 (middle), and UBF (bottom). Input represents approximately 25 ␮g of nuclear extract.

the site of ribosomal DNA transcription (Figure 2B, lane 2). The presence of TAF1 in the nucleolar extract is not due to cross-contamination during the preparation of the extract because Sp1, an abundant nuclear factor, is absent from this fraction (lane 2). As expected, UBF and TAFI110, two factors involved in RNA pol I transcrip-

TAF1 Stimulates Pol I Transcription in Transfections and Cell-Free Transcription Assays To determine whether TAF1 functions in Pol I transcription, we transfected HeLa cells with a human rRNA promoter construct (prHu3CAT) together with a plasmid expressing either TAF1 or TAF4. Transfection of the promoter construct produces a 123 nucleotide primer extension fragment that corresponds to transcripts initiated at the Pol I transcription initiation site. In the presence of TAF1, transcription from the human rRNA promoter is stimulated in a dosage-dependent fashion (Figure 3A). The stimulation of Pol I transcription is specific for TAF1 because TAF4 failed to induce any significant increase in Pol I transcription. To further examine the functional relationship between TAF1 and Pol I transcription, we performed transcription assays with purified factors in a cell-free system. Recombinant TAF1 was purified from insect cells and was added to in vitro transcription reactions containing a purified RNA polymerase I fraction, SL1, and recombinant UBF. The results of the transcription assays indicate that the addition of increasing amounts of TAF1 results in an incremental increase in Pol I transcription (Figure 3B). The stimulation of transcription depends on the presence of UBF and SL1 because TAF1 did not activate transcription in the absence of either one of these two factors (lanes 4 and 5). Moreover, the addition of TFIID to the transcription reaction did not have any effect on Pol I transcription (Figure 3C, lanes 5 and 6), indicating that the TAF1 present in the TBPTAFII complex is not capable of stimulating Pol I transcription. To determine whether the interaction between UBF and TAF1 is necessary for stimulating Pol I transcription, we added TAF1 to the transcription reaction

Brief Communication 2145

in the presence of a UBF mutant (UBFdx) missing the X-domain. UBFdx is competent to activate Pol I transcription in an in vitro reconstituted transcription assay [11], but it does not interact with TAF1 (Figure 1A). As shown in Figure 3D, TAF1 failed to stimulate transcription in the presence of UBFdx (lanes 1–3), suggesting that the physical interaction between UBF and TAF1 is required for the stimulation of Pol I transcription. The finding that UBF, a Pol I-specific trans-acting factor, interacts with TAF1 is intriguing. TAF1 was originally identified as a component of the TFIID complex, a Pol II-specific factor incompetent of activating Pol I transcription (see Figure 3C). Although TAFs were originally thought to be exclusively subunits of the TFIID complex, recent studies have revealed that some TAFs are components of other cellular multiprotein complexes, such as TFTC, SAGA, and PCAF, which, unlike TFIID, do not contain TBP [18]. Although TAF1 is not a component of these TBP-free complexes, a significant amount of TAF1 is found in a TBP-free form in human cells (our unpublished data; [19]), and a recent study has shown that Drosophila TAF1 (dmTAF1) is a component of the Polycomb Repressive Complex 1 (PRC1) [20]. Taken together, these results suggest that TAF1 is a regulatory protein involved in several aspects of Pol I and II transcription. The involvement of TAF1 in the expression of genes transcribed by different nuclear polymerases is reminiscent of the TATA binding protein (TBP), which was originally thought to be solely a Pol II factor but was later also found to play an essential role in transcription by Pol I and III [21]. More recently, TFIIH, a basal Pol II transcription initiation factor, has been also implicated in Pol I transcription [22]. The role of TAF1 in the transcription process is not well understood, but is likely that the regulatory properties of this factor are linked to its intrinsic enzymatic activity (kinase, histone acetyltransferase (HAT), and ubiquitin activating-conjugating activities) [10]. Gene chip hybridization and differential display experiments indicated that TAF1 is required for the regulation of a subset of Pol II-transcribed genes (16%–18%), most of which are involved in G1-to-S progression and cell growth [23, 24]. Figure 3. Stimulation of RNA Polymerase I Transcription by TAF1 (A) The human ribosomal RNA gene promoter construct prHu3CAT (3 ␮g) was cotransfected with an empty pcDNA3 vector (bars 1 and 2; 2 and 4 ␮g, respectively), a plasmid expressing flag-TAF1 (bars 2 and 3; 2 ␮g (⫹) and 4 ␮g (⫹⫹), respectively), or a plasmid expressing HA-TAF4 (bars 4 and 5; 2 ␮g (⫹) and 4 ␮g (⫹⫹), respectively). Ribonuclease protection assays (RPA) for ␤-actin mRNA were used for normalizing the amount of RNA used in the primer extension assays (inset, top panel). The primer extension data were quantified with a Phosphoimager (Molecular Dynamics), and relative activities are shown as the mean ⫾ standard deviation. Data were obtained from five independent experiments, except for transfections with TAF4, which were performed twice. Expression of the transfected epitope-tagged proteins was determined by Western blot analysis with flag (TAF1) and HA (TAF4) antibodies (inset, bottom panel). (B) Increasing amounts of recombinant TAF1 (lanes 2 and 3, approximately 10 ng and 15 ng [0.8 ␮l and 1.2 ␮l], respectively; lanes 4 and 5, 1.2 ␮l each) were tested in reconstituted transcription assays in the presence of a purified RNA polymerase I fraction (lanes 1–5), purified SL1 (lanes 1–4), and recombinant UBF (approximately 5 ng; lanes 1–3 and 5). Transcription reactions were carried out as previously described [7, 9]. The amount of each recombinant protein used in the transcription assays was estimated from silver-

stained gels. An aliquot (3 ␮l) of the preparation of recombinant TAF1 used in the transcription reactions is shown in the inset (silverstained gel). (C) Transcription reactions were performed in the presence of TAF1 (lane 3, 20 ng), UBF (lanes 1–3, 5, and 6), or increasing amounts of a TFIID fraction (lanes 5 and 6). Purified RNA polymerase I and SL1 were added to each reaction. The TFIID fraction used in the reactions shown in lanes 5 and 6 contains approximately the same amount (lane 5) or twice the amount (lane 6) of TAF1 used in lane 3, as judged by Western blot analysis with TAF1 antibodies (not shown). This TFIID fraction was active in Pol II transcription assays in vitro (not shown). (D) Transcription reactions contained recombinant UBFdx (lanes 1–4 and 6; approximately 5 ng) and increasing amounts of recombinant TAF1 (lanes 1–3; approximately 5 ng, 10 ng, and 15 ng, respectively; lanes 4 and 5, 10 ng each). Recombinant UBF was used in the reaction shown in lane 7. Purified SL1 was omitted from the reaction shown in lane 4. Asterisks indicate the background level of activity. All the transcription experiments have been repeated several times with identical results.

Current Biology 2146

Based on these findings, it has been proposed that TAF1 plays a critical role in the coordinate control of cell growth and division [1, 2]. The transition between G1 and S is an important step in the cell’s life; it is here that a decision is made to increase the translation capacity in order to generate proteins that will ensure timely progression through the cell cycle. RNA polymerase I transcription controls ribosome biogenesis and a cell’s potential for growth. Thus, the finding that hsTAF1 regulates Pol I transcription provides further support for the idea that hsTAF1 may be specifically engaged in the regulation of genes, such as the ribosomal DNA genes, that are necessary to fulfill the cell’s requirement for growth (Figure S2 in the Supplementary Material). It is our hope that further dissection of the role of TAF1 in nucleolar transcription will reveal new details in the regulatory events controlling growth and cell cycle progression. Supplementary Material Supplementary Experimental Procedures and two supplementary figures are available with this article online at http://images.cellpress. com/supmat/supmatin.htm. Acknowledgments We would like to thank S. Navarro for technical support and W. Zhai for stimulating discussions. The generosity of E. Wang, N. Tanese, R. Dikstein, and M. Stallcup in providing reagents is especially appreciated. P.S. was supported by a fellowship from the AmericanItalian Cancer Foundation. This work is supported by grants from the National Institutes of Health and the American Cancer Society to L.C. Received: May 30, 2002 Revised: July 29, 2002 Accepted: October 4, 2002 Published: December 23, 2002 References 1. Polymenis, M., and Schmidt, E.V. (1999). Coordination of cell growth with cell division. Curr. Opin. Genet. Dev. 9, 76–80. 2. Neufeld, T.P., and Edgar, B.A. (1998). Connections between growth and the cell cycle. Curr. Opin. Cell Biol. 10, 784–790. 3. Reeder, R. (1999). Regulation of RNA polymerase I transcription in yeast and vertebrates. Prog. Nucleic Acid Res. Mol. Biol. 62, 293–327. 4. Grummt, I. (1999). Regulation of mammalian ribosomal gene transcription by RNA polymerase I. Prog. Nucleic Acid Res. Mol. Biol. 62, 109–154. 5. Bell, S.P., Learned, R.M., Jantzen, H.-M., and Tjian, R. (1988). Functional cooperativity between transcription factors UBF1 and SL1 mediates human ribosomal RNA synthesis. Science 241, 1192–1197. 6. Comai, L., Zomerdijk, J.C.B.M., Beckmann, H., Zhou, S., Admon, A., and Tjian, R. (1994). Reconstitution of transcription factor SL1: exclusive binding of TBP by SL1 or TFIID subunits. Science 266, 1966–1972. 7. Comai, L., Tanese, N., and Tjian, R. (1992). The TATA-binding protein and associated factors are integral components of the RNA polymerase I transcription factor, SL1. Cell 68, 965–976. 8. Jantzen, H.-M., Admon, A., Bell, S.P., and Tjian, R. (1990). Nucleolar transcription factor hUBF contains a DNA motif with homology to HMG proteins. Nature 344, 830–836. 9. Tuan, J.C., Zhai, W., and Comai, L. (1999). Recruitment of TATAbinding protein-TAFI complex SL1 to the human ribosomal DNA promoter is mediated by the carboxy-terminal activation domain of upstream binding factor (UBF) and is regulated by UBF phosphorylation. Mol. Cell. Biol. 19, 2872–2879.

10. Wassarman, D.A., and Sauer, F. (2001). TAF(II)250: a transcription toolbox. J. Cell Sci. 114, 2895–2902. 11. Jantzen, H.M., Chow, A.M., King, D.S., and Tjian, R. (1992). Multiple domains of the RNA polymerase I activator hUBF interact with the TATA-binding protein complex hSL1 to mediate transcription. Genes Dev. 6, 1950–1963. 12. Hisatake, K., Hasegawa, S., Takada, R., Nakatani, Y., Horikoshi, M., and Roeder, R.G. (1993). The p250 subunit of native TATA box-binding factor TFIID is the cell-cyle regulatory protein CCG1. Nature 362, 179–181. 13. Ruppert, S., Wang, E.H., and Tjian, R. (1993). Cloning and expression of human TAFII250: a TBP-associated factor implicated in cell-cycle regulation. Nature 362, 175–179. 14. Hayashida, T., Sekiguchi, T., Noguchi, E., Sunamoto, H., Ohba, T., and Nishimoto, T. (1994). The CCG1/TAFII250 gene is mutated in thermosensitive G1 mutants of the BHK21 cell line derived from golden hamster. Gene 141, 267–270. 15. Sekiguchi, T., Nakashima, T., Hayashida, T., Kuraoka, A., Hashimoto, S., Tsuchida, N., Shibata, Y., Hunter, T., and Nishimoto, T. (1995). Apoptosis is induced in BHK cells by the tsBN462/ 13 mutation in the CCG1/TAFII250 subunit of the TFIID basal transcription factor. Exp. Cell Res. 218, 490–498. 16. Sekiguchi, T., Nohiro, Y., Nakamura, Y., Hisamoto, N., and Nishimoto, T. (1991). The human CCG1 gene, essential for progression of the G1 phase, encodes a 210-kilodalton nuclear DNAbinding protein. Mol. Cell. Biol. 11, 3317–3325. 17. Rabenstein, M.D., Zhou, S., Lis, J.T., and Tjian, R. (1999). TATA box-binding protein (TBP)-related factor 2 (TRF2), a third member of the TBP family. Proc. Natl. Acad. Sci. USA 96, 4791–4796. 18. Bell, B., and Tora, L. (1999). Regulation of gene expression by multiple forms of TFIID and other novel TAFII-containing complexes. Exp. Cell Res. 246, 11–19. 19. Bertolotti, A., Lutz, Y., Heard, D.J., Chambon, P., and Tora, L. (1996). hTAFII68, a novel RNA/ssDNA-binding protein with homology to the pro-oncoproteins TLS/FUS and EWS, is associated with both TFIID and RNA polymerase II. EMBO J. 15, 5022– 5031. 20. Saurin, A.J., Shao, Z., Erdjument-Bromage, H., Tempst, P., and Kingston, R.E. (2001). A Drosophila Polycomb group complex includes Zeste and dTAFII proteins. Nature 412, 655–660. 21. White, R.J., and Jackson, S.P. (1992). The TATA-binding protein: a central role in transcription by RNA polymerases I, II and III. Trends Genet. 8, 284–288. 22. Iben, S., Tschochner, H., Bier, M., Hoogstraten, D., Hozak, P., Egly, J.M., and Grummt, I. (2002). TFIIH plays an essential role in RNA polymerase I transcription. Cell 109, 297–306. 23. O’Brien, T., and Tjian, R. (2000). Different functional domains of TAFII250 modulate expression of distinct subsets of mammalian genes. Proc. Natl. Acad. Sci. USA 97, 2456–2461. 24. Walker, S.S., Shen, W.C., Reese, J.C., Apone, L.M., and Green, M.R. (1997). Yeast TAF(II)145 required for transcription of G1/S cyclin genes and regulated by the cellular growth state. Cell 90, 607–614.