Structural characterization of human general transcription factor TFIIF in solution SATOKO AKASHI,1 SHINJIRO NAGAKURA,1 SEIJI YAMAMOTO,2 MASAHIKO OKUDA,1 YOSHIAKI OHKUMA,2,3 AND YOSHIFUMI NISHIMURA1 1International Graduate School of Arts and Sciences, Yokohama City University, Yokohama, Kanagawa 230-0045, Japan 2Graduate School of Pharmaceutical Sciences, Osaka University, Suita, Osaka 565-0871, Japan 3Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama 930-0194, Japan (RECEIVED September 20, 2007 FINAL REVISION November 16, 2007 ACCEPTED November 27, 2007) Abstract Human general transcription factor IIF (TFIIF), a component of the transcription pre-initiation complex (PIC) associated with RNA polymerase II (Pol II), was characterized by size-exclusion chromatography (SEC), electrospray ionization mass spectrometry (ESI-MS), and chemical cross-linking. Recombinant TFIIF, composed of an equimolar ratio of a and b subunits, was bacterially expressed, purified to homogeneity, and found to have a transcription activity similar to a natural one in the human in vitro transcription system. SEC of purified TFIIF, as previously reported, suggested that this protein has a size 200 kDa. In contrast, ESI-MS of the purified sample gave a molecular size of 87 kDa, indicating that TFIIF is an ab heterodimer, which was confirmed by matrix-assisted laser desorption/ionization (MALDI) MS of the cross-linked TFIIF components. Recent electron microscopy (EM) and photo- cross-linking studies showed that the yeast TFIIF homolog containing Tfg1 and Tfg2, corresponding to the human a and b subunits, exists as a heterodimer in the PIC, so the human TFIIF is also likely to exist as a heterodimer even in the PIC. In the yeast PIC, EM and photo-cross-linking studies showed different results for the mutual location of TFIIE and TFIIF along DNA. We have examined the direct interaction between human TFIIF and TFIIE by ESI-MS, SEC, and chemical cross-linking however, no direct interaction was observed, at least in solution. This is consistent with the previous photo-cross-linking observation that TFIIF and TFIIE flank DNA separately on both sides of the Pol II central cleft in the yeast PIC. Keywords: TFIIF size-exclusion chromatography (SEC) electrospray ionization mass spectrometry (ESI- MS) chemical cross-linking matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOFMS) TFIIE heterodimer Transcription processes in eukaryotes begin with the formation of a pre-initiation complex (PIC), composed of RNA polymerase II (Pol II) associated with five gen- eral transcription factors: TFIIB, TFIID, TFIIE, TFIIF, and TFIIH (Orphanides et al. 1996 Roeder 1996 Nikolov and Burley 1997). In general, the PIC assembly starts with the recognition of the TATA element by the TATA box-binding protein (TBP) of TFIID. Followed by the Reprint requests to: Satoko Akashi, International Graduate School of Arts and Sciences, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan e-mail: akashi@ tsurumi.yokohama-cu.ac.jp fax: 81-45-508-7362 or Yoshifumi Nishi- mura, International Graduate School of Arts and Sciences, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama, Kana- gawa 230-0045, Japan e-mail: nisimura@tsurumi.yokohama-cu.ac.jp fax: 81-45-508-7362. Article published online ahead of print. Article and publication date are at http://www.proteinscience.org/cgi/doi/10.1110/ps.073258108. Protein Science (2008), 17:389���400. Published by Cold Spring Harbor Laboratory Press. Copyright �� 2008 The Protein Society 389

attachment of TFIIB (Pinto et al. 1992 Bushnell et al. 2004), nonphosphorylated Pol II comes with TFIIF to the complex (Flores et al. 1991). After that, TFIIE and TFIIH are integrated to accomplish the PIC formation. TFIIH catalyzes ATP-dependent melting of the promoter DNA as well as phosphorylation of the C-terminal domain (CTD) of the largest subunit of Pol II, and then transcription by Pol II begins (Orphanides et al. 1996 Roeder 1996 Nikolov and Burley 1997 Ohkuma 1997). Human TFIIF consists of an equimolar ratio of TFIIFa (RAP74) and TFIIFb (RAP30) (Flores et al. 1990 Con- away and Conaway 1993). TFIIF comes with Pol II and delivers it to the TFIID���TFIIB���DNA complex on the promoter near the transcription initiation site. The TFIIF��� Pol II complex is required for the entry of TFIIE and TFIIH to the PIC since it provides the scaffold for them (Maxon et al. 1994). Until now, extensive mutational stud- ies have been carried out to identify the specific structural domains in TFIIF responsible for each of its functions in transcription. Several interactions between the specific structural domains have been characterized to date. With regard to TFIIFa, interactions have been identified between TFIIFa (2���154) and TFIIFb (2���98) (Gaiser et al. 2000), TFIIFa (2���139) and TAF250 (Ruppert and Tjian 1995), TFIIFa (2���217) and DNA (Lei et al. 1998 Coulombe and Burton 1999), TFIIFa (356���517) and TFIIB (Fang and Burton 1996), and TFIIFa (363���517) and a Pol II/DNA complex (Wang and Burton 1995). For TFIIFb, interactions have been found between TFIIFb (2��� 98) and TFIIFa (2���154) (Gaiser et al. 2000), TFIIFb (1��� 152) and TFIIB (Fang and Burton 1996), and TFIIFb (160���249) and DNA (Tan et al. 1994). Since TFIIF par- ticipates in various crucial events in the processes of transcription initiation and the subsequent transition from initiation to elongation, it is of significance to character- ize the structure of the intact molecule. The intact TFIIF structure, however, has not been solved at the atomic level. This has long been hampered by the fact that TFIIFa was expressed in a shortened form due to premature termination, and that TFIIFb was instead expressed as an insoluble form. Therefore, only the structures of the C-terminal region of TFIIFa (Kamada et al. 2001) and the complex of interaction domains of TFIIFa and TFIIFb (Gaiser et al. 2000) have been determined so far by X-ray crystallography. Based on the complex structure, at least the molecular mechanism of the heterodimer formation between TFIIFa and TFIIFb has been well established (Gaiser et al. 2000). TFIIE, another general transcription factor, recruits TFIIH to the PIC and controls TFIIH helicase and kinase activities (Ohkuma and Roeder 1994). The tertiary struc- tures of only two functional domains, the core domains of TFIIEa and TFIIEb, have been determined (Okuda et al. 2000, 2004). It has long been believed that TFIIE functions as a heterotetramer composed of two a and two b subunits from its behavior in size-exclusion chromatography (SEC) (Ohkuma et al. 1990). However, we recently demonstrated that TFIIE exists as an ab heterodimer in solution by using electrospray ionization mass spectrometry (ESI-MS), analytical ultracentrifuga- tion, and small-angle X-ray scattering (SAXS) (Itoh et al. 2005). The rod-like molecular shape of TFIIE in solution had been misleading in overestimating its molecular size in SEC analysis, which was calibrated by globular pro- teins. Similarly, in the case of TFIIF, there was a dis- crepancy between the previous results of SEC analysis and glycerol gradient sedimentation experiments SEC both at high (1.0 M KCl) and low (0.1 M KCl) salt concentrations gave its molecular size to be 220 kDa, suggesting an a2b2 heterotetramer (Flores et al. 1990), while glycerol gradient sedimentation of TFIIF at a low salt concentration indicated a smaller size, 76 kDa (Flores et al. 1989). Based on the photo-cross-linking study (Forget et al. 2004), it was demonstrated that each subunit of both TFIIE and TFIIF interacts with two dis- tinct promoter regions, upstream and downstream, sug- gesting two copies of each TFIIE and TFIIF subunits within the PIC. So both TFIIE and TFIIF were assumed to be a heterotetramer in the PIC. However, Forget et al. (2004) could not rule out the possibility that these factors have two or more distinct domains held together by an unfolded linker, and that each heterodimer of TFIIE and TFIIF could cover two distinct promoter regions. Ac- tually, in the case of TFIIE, our SAXS analysis showed that TFIIE is a rod-like molecule that could cover the entire Pol II region (Itoh et al. 2005). Although the binding stoichiometry of human TFIIFa and TFIIFb has not definitely been identified until now, the yeast TFIIF homolog seems to bind Pol II as a heteromer composed of three subunits, two of which, Tfg1 and Tfg2 (human TFIIFa and TFIIFb subunits, respectively), are well con- served in human and yeast. Recent EM (Chung et al. 2003) and photo-cross-linking (Chen et al. 2007) studies of the yeast PIC showed that Tfg1 and Tfg2 form a probable heterodimer along the downstream DNA, but the mutual location of TFIIE and TFIIF along DNA is a little different between those results. To clarify the mutual location of TFIIE and TFIIF in the PIC, the association character between isolated TFIIE and TFIIF molecules would be helpful. Thus, for a better understanding of the TFIIF functions, we have prepared naturally soluble recombinant human TFIIF protein by co-expressing both TFIIF subunits in bacteria and investigated its biophysical characteristics. The molecular size of TFIIF was exam- ined by SEC and ESI-MS. These results were further confirmed by mass measurements of the cross-linked products of TFIIF. In addition, the interaction between Akashi et al. 390 Protein Science, vol. 17