Structure of the central core domain of TFIIEβ with a novel double‐stranded DNA‐binding surface

Masahiko Okuda, Yoshinori Watanabe, Hideyasu Okamura, Fumio Hanaoka, Yoshiaki Ohkuma, Yoshifumi Nishimura

Author Affiliations

  1. Masahiko Okuda1,
  2. Yoshinori Watanabe2,
  3. Hideyasu Okamura1,
  4. Fumio Hanaoka2,
  5. Yoshiaki Ohkuma2 and
  6. Yoshifumi Nishimura*,1,3
  1. 1 Graduate School of Integrated Science, Yokohama City University, 22‐2 Seto, Kanazawa‐ku, Yokohama, 236‐0027, Japan
  2. 2 Institute for Molecular and Cellular Biology, Osaka University, 1‐3 Yamada‐oka, Suita, Osaka, 565‐0871, Japan
  3. 3 Genomic Sciences Center (GSC), RIKEN (The Institute of Physical and Chemical Research), 2‐1 Hirosawa, Wako‐shi, Saitama, 351‐0198, Japan
  1. *Corresponding author. E-mail: nisimura{at}


Human general transcription factor TFIIE consists of two subunits, TFIIEα and TFIIEβ. Recently, TFIIEβ has been found to bind to the region where the promoter starts to open to be single‐stranded upon transcription initiation by RNA polymerase II. Here, the central core domain of human TFIIEβ (TFIIEβc) has been identified by a limited proteolysis. This solution structure has been determined by NMR. It consists of three helices with a β hairpin at the C–terminus, resembling the winged helix proteins. However, TFIIEβc shows a novel double‐stranded DNA‐binding activity where the DNA‐binding surface locates on the opposite side to the previously reported winged helix motif by forming a positively charged furrow. A model will be proposed that TFIIE stabilizes the preinitiation complex by binding not only to the general transcription factors together with RNA polymerase II but also to the promoter DNA, where double‐stranded DNA starts to open to be single‐stranded upon activation of the preinitiation complex.


Transcription initiation from eukaryotic protein‐coding genes is a complex process requiring RNA polymerase II (Pol II) and six general transcription factors (TFIIA, TFIIB, TFIID, TFIIE, TFIIF and TFIIH), which are necessary to assemble the preinitiation complex (PIC) at the promoter and initiate transcription (reviewed in Orphanides et al., 1996; Roeder, 1996). Analysis of the PIC assembly pathway using the isolated general transcription factors has indicated that the process is initiated by the binding of the TBP subunit of TFIID to the TATA box. Then the stepwise assembly of TFIIA, TFIIB, a complex of Pol II and TFIIF, TFIIE and TFIIH proceeds. TFIIH, which is recruited through a direct interaction with TFIIE, activates the PIC through its enzymatic activities, resulting in the open complex formation (Lu et al., 1991; Ohkuma and Roeder, 1994; Dvir et al., 1996). In addition to their key roles in the PIC formation and initiation, TFIIE and TFIIH play important roles in the transition from initiation to elongation (Goodrich and Tjian, 1994; Dvir et al., 1997; Kumar et al., 1998).

TFIIE is a heterotetramer containing two of each two subunits, TFIIEα and TFIIEβ, with molecular masses of 57 and 34 kDa, respectively (Ohkuma et al., 1990, 1991; Peterson et al., 1991; Sumimoto et al., 1991). TFIIE plays essential roles in TFIIH recruitment (Flores et al., 1992) and the regulation of TFIIH activities: a kinase activity that phosphorylates the C‐terminal domain (CTD) of the largest subunit of Pol II, a DNA‐dependent ATPase activity and a DNA helicase activity (Lu et al., 1991; Drapkin et al., 1994; Ohkuma and Roeder, 1994). In addition, TFIIE plays important roles in the open complex formation (Timmers, 1994; Holstege et al., 1995) and in the clearance of the initiation complex from the promoter (Goodrich and Tjian, 1994; Dvir et al., 1997). In the complex structure of yeast TFIIE with Pol II revealed by two‐dimensional crystallography, TFIIE interacts near the active center of Pol II (Leuther et al., 1996). A photo‐crosslinking study showed that TFIIEβ binds to the core promoter region of −14 and −2 upstream from the transcription initiation site (+1), while TFIIEα does not bind to DNA by itself (Robert et al., 1996). TFIIEα binds strongly to TBP and TFIIH, and weakly to Pol II and TFIIF, while TFIIEβ binds strongly to Pol II, TFIIB, Rap30 (TFIIFβ) and TFIIH, suggesting a plausible position of TFIIE in the PIC (reviewed in Ohkuma, 1997). However, a recent study has shown that TFIIEα could directly stimulate the TBP binding to the promoter in the absence of other general transcription factors and TFIIEβ could directly interact with TFIIA, suggesting new roles for TFIIE in the early stages of PIC formation (Yokomori et al., 1998). To understand the detailed molecular mechanism of PIC formation, several three‐dimensional structures of the general transcription factors have been determined (reviewed in Nikolov and Burley, 1997), for example, the core domain structures of TBP (Nikolov et al., 1992), TFIIB (Bagby et al., 1995; Zhu et al., 1996), TBP associated factors (TAFs) (Xie et al., 1996; Birck et al., 1998), TBP and TAF (Liu et al., 1998), and Rap30 (TFIIFβ) (Groft et al., 1998) in their DNA‐free states, in addition, the complex structures with DNA of the core domains of TBP (J.L.Kim et al., 1993; Y.Kim et al., 1993), TFIIB and TBP (Nikolov et al., 1995), and TFIIA and TBP (Geiger et al., 1996; Tan et al., 1996).

Human TFIIEβ, consisting of 291 amino acids, is a basic protein with a pI value of 9.5 and contains several putative structural motifs as shown in Figure 1 (Sumimoto et al., 1991). By using a systematic series of deletion mutants of TFIIEβ, the C‐terminal two basic regions, basic helix–loop–helix and basic helix–loop regions, are found to be very important for the interactions with TFIIB, Rap30 (TFIIFβ), single‐stranded DNA, Drosophila transcriptional regulatory factor Krüppel, and TFIIEα (Okamoto et al., 1998). The roles of the other regions of TFIIEβ are not yet understood, although the central 227 amino acid region (residues 51–277) of TFIIEβ has been demonstrated to be necessary and sufficient for both basal and activated transcription (Okamoto et al., 1998). Here, we have identified a structural core domain of TFIIEβ (TFIIEβc) using proteolytic digestion, and the structure of TFIIEβc has been determined using NMR. Importantly, this domain was found to have a dominant negative effect on basal transcription and, therefore, has been assumed to be functionally very important (Okamoto et al., 1998). Through the three‐dimensional structure analysis, it is suggested that TFIIEβc plays a role in double‐stranded DNA binding of TFIIE, whose function might be important when the PIC is activated and the promoter starts opening to become single‐stranded.

Figure 1.

Sequence alignment of TFIIEβ. (A) Schematic diagram of putative structural motifs and characteristic sequences of human TFIIEβ. Ser‐rich, a serine‐rich sequence (residues 26–71); RAP30, a region similar to the Pol II binding region of Rap30 (TFIIFβ) (residues 79–111); LR, a leucine repeat motif (residues 145–163); σ3, a region similar to the bacterial σ factor subdomain 3 (residues 163–193); BR‐HLH, a basic region helix–loop–helix motif (residues 197–238); BR‐HL, a basic region helix–loop sequence (residues 258–291). (B) Sequence alignment of human TFIIEβ with other TFIIEβ. Residues identified in all five species are boxed in red, residues identified in four out of the five species are boxed in pink, and conserved residues are boxed in yellow. The asterisks indicate the residues that form the hydrophobic core of human TFIIEβc. The secondary structure of human TFIIEβc is indicated above the sequence. Three α helical regions are indicated by H1, H2 and H3, and three β‐stranded regions are indicated by S1, S2 and S3. ‘ht’ means a helical turn region.

Results and discussion

Structure determination

The TFIIEβ cDNAs have been cloned from human (Sumimoto et al., 1991), Xenopus (Ohkuma et al., 1992), Drosophila (Wang et al., 1997), Caenorhabditis elegans (S.Yamamoto, Y.Watanabe, P.J.van der Spek, T.Watanabe, H.Fujimoto, F.Hanaoka and Y.Ohkuma, in preparation) and Saccharomyces cerevisiae (Feaver et al., 1994), showing the amino acid sequence homology among these TFIIEβ proteins. To identify the structural domain of TFIIEβ, the human recombinant TFIIEβ was subjected to a limited proteolysis. One structural domain, consisting of the amino acid residues 66–146, and two structural domains consisting of the amino acid residues 74–152 and 211–244, have been identified by trypsin and chymotrypsin digestion, respectively. Here, the central core domain (designated as TFIIEβc) consisting of the amino acid residues 66–146, including a region putatively similar to the Pol II binding region of Rap30 (TFIIFβ), has been investigated (Figure 1). This human TFIIEβc was expressed in Escherichia coli BL21(DE3) pLysS. Purified TFIIEβc was subjected to conventional multi‐dimensional NMR measurements. Almost all signals of the main chain atoms, amide proton, amide nitrogen, α proton and α carbon for each amino acid of TFIIEβc could be assigned except for amide nitrogen signals of the three proline residues, Asn126 and the N‐terminal two residues, and amide proton signals of the three proline residues, Asn126 and the N‐terminal Ala66. The three‐dimensional structure of TFIIEβc was determined based on a total of 1184 distance restraints, as well as 38 dihedral angle restraints and 20 hydrogen bond restraints, using a four‐dimensional simulated annealing protocol (Nakai et al., 1993).

Overview of the structure

Figure 2A shows the 20 calculated structures of TFIIEβc and Figure 2B shows an averaged structure over 20 calculated structures. Structural statistics are summarized in Table I. With the exception of the N‐terminal nine residues (66–74) and the C‐terminal five residues (142–146), the backbone of TFIIEβc is well defined. The structural quality of the ensemble of the calculated structures was checked using the program PROCHECK‐NMR (Laskowski et al., 1996). Almost all residues except glycine fall into allowed regions: 83.2, 15.0 and 1.0% residues without glycine residues in the most favorable, additionally allowed and generously allowed regions, respectively. The average atomic root‐mean‐square (r.m.s.) deviations of the 20 simulated annealing (SA) structures about the mean coordinate positions are 0.41 (±0.04) Å for the backbone atoms and 0.88 (±0.05) Å for all heavy atoms.

Figure 2.

Stereo view of the solution structure of human TFIIEβc. (A) Best‐fit superposition of the ensemble of the final 20 NMR structures of human TFIIEβc. (B) The averaged structure of human TFIIEβc. The averaged structure over the final 20 NMR structures of human TFIIEβc is shown in a ribbon representation. These figures were made using the program MOLMOL (Koradi et al., 1996).

View this table:
Table 1. Structural statistics

The structure of TFIIEβc contains three helices and a C‐terminal β hairpin. The three helices, H1, H2 and H3 consist of Phe75–Gln90, Leu99–Glu105 and Leu113–Met120, respectively. A TXXE type capping box of α helix is found in the N‐terminal region of H2, by forming hydrogen bonds between Thr98 and Glu101 (Harper and Rose, 1993). The region of Thr121–Val125 has a helical turn structure (ht). The C‐terminal region has a β hairpin formed by a β strand with Ile130–Ile133 (S2 strand), a hairpin loop with Asp134 and Gly135, and a β strand with Lys136–Phe139 (S3 strand). Both S2 and S3 strands form a twisted anti‐parallel β sheet with Leu 97 (S1 strand). H1 is longer than H2 and H3. The three helices are maintained by a hydrophobic core, formed by Val77, Leu78 and Ile81 from H1, Leu99, Ile102 and Leu103 from H2, Trp118 and Leu119 from H3, and Ala123 and Leu124 from the helical turn as shown in Figure 3. In addition, the C‐terminal region of H1, consisting of Val82, Tyr84 and Met85, interacts with Leu97 from S1, Ile130 from S2, and Tyr137 and Phe139 from S3, extending the hydrophobic core. The amino acids that participate in the hydrophobic core are well conserved in the central core domains of different species as shown in Figure 1B, indicating that other central core domains of TFIIEβ from different species also hold a three‐dimensional structure similar to the human TFIIEβc structure.

Figure 3.

The hydrophobic core of TFIIEβc. The side chains of amino acids that form the hydrophobic core of TFIIEβc are shown. The hydrophobic side chains from the α‐helices, the β‐sheet and the helical turn are shown in cyan, light‐green and yellow, respectively. This figure was made using the program MOLMOL (Koradi et al., 1996).

Structural similarity to winged helix proteins implies the function of TFIIEβc

To identify the structural family of TFIIEβc, we have compared the TFIIEβc structure with other structures so far determined using the DALI program (Holm and Sander, 1993). Based on the Z score, which is the strength of similarity between two structures, the 13 highest similar structures to the TFIIEβc structure are identified by Z scores >3.9. It is interesting to find that all the 13 proteins with the highest Z scores are DNA‐binding proteins: the N‐terminal fragment of E.coli topoisomerase I (Z = 5.3; Lima et al., 1994); the DNA‐binding domain of the cell cycle transcription factor DP2 (Z = 5.3; Zheng et al., 1999); the DNA‐binding domain of HNF‐3γ (Z = 5.2; Clark et al., 1993); the DNA‐binding domain of catabolite gene activator protein (CAP) (Z = 5.1; Schultz et al., 1991); the globular domain of histone H5 (Z = 5.1; Ramakrishnan et al., 1993); the restriction endonuclease FokI (Z = 4.9; Wah et al., 1997); the DNA‐binding domain of Rap30 (TFIIFβ) (Z = 4.8; Groft et al., 1998); the DNA‐binding domain of OmpR (Z = 4.8; Kondo et al., 1997; Martinez‐Hackert and Stock, 1997); the transcription regulatory protein MotA fragment (Z = 4.6; Finnin et al., 1997); the DNA‐binding domain of LexA repressor (Z = 4.4; Fogh et al., 1994); the DNA‐binding domain of biotin repressor, BirA (Z = 4.1; Wilson et al., 1992); the DNA‐binding domain of arginine repressor (ArgR) (Z = 4.0; Ni et al., 1999); and the replication terminator protein (RTP) from Bacillus subtilis (Z = 3.9; Buissiere et al., 1995). It is well known that the DNA‐binding domains of DP2, HNF‐3γ, histone H5, Rap30 (TFIIFβ), OmpR, LexA, BirA, ArgR and RTP contain the winged helix motif, consisting of three helices and a C‐terminal β hairpin.

Figure 4 shows the structural comparison of TFIIEβc with the DNA‐binding domains of DP2 and HNF‐3γ. It is apparent that the architecture of TFIIEβc is very close to those of the DNA‐binding domains. The DNA‐bound structure of the DNA‐binding domain of HNF‐3γ was established using X‐ray crystallography (Clark et al., 1993). The third helix of the domain, the recognition helix, locates in the major groove of DNA and two C‐terminal loops (wings) contact the phosphate backbone of DNA (Figure 4). The architecture of three helices and a C‐terminal β hairpin except the C‐terminal two loops of HNF‐3γ resembles that of TFIIEβc. The α carbon atoms of both domains can be superimposed with a 2.3 Å r.m.s. deviation for 58 residues. Recently the DNA‐complexed structure of the heterodimer of DP2 and E2F4 has been determined (Zheng et al., 1999). Both DNA‐binding domains of DP2 and E2F4 contain the winged helix motif, consisting of three α helices (H1, H2 and H3) and a β sheet. The structure of TFIIEβc is closer to that of DP2 (Z = 5.3) than that of E2F4 (Z = 2.8). Although the H2 and H3 helices of DP2 are much longer than the corresponding helices of TFIIEβc (Figure 4), the arrangement of three helices and a C‐terminal β hairpin of DP2 resembles that of TFIIEβc. DP2 and TFIIEβc can be superimposed with an r.m.s. deviation of 2.3 Å in the Cα positions of 59 residues. The DNA‐binding modes of DP2 and E2F4 are similar in each; the H3 helix of each domain recognizes a specific base sequence and the N‐terminal portion of the H2 (E2F4) or H1 (DP2) helix and a portion of each β sheet contacts the phosphate backbone of DNA (Figure 4). In addition, TFIIEβc, although containing a putative region similar to the Pol II binding region of Rap30, has a structure very similar to the DNA‐binding domain of Rap30 (Figure 4). Actually, TFIIEβc did not bind directly to Pol II (Y.Watanabe and Y.Ohkuma, unpublished data). The DNA‐binding ability of Rap30 has been well investigated and the DNA‐binding surface of Rap30 has been structurally mapped using NMR (Groft et al., 1998), indicating a DNA‐binding mode similar to other winged helix proteins (Figure 4). These results strongly suggest that TFIIEβc is a DNA‐binding domain like the winged helix proteins. However, it has been found that only the C‐terminal basic regions of TFIIEβ are responsible for the single‐stranded (ss)DNA binding; no double‐stranded (ds)DNA‐binding domain has been identified in TFIIEβ.

Figure 4.

Comparison of TFIIEβc with winged helix proteins. Comparison of TFIIEβc (A) with the DNA‐binding domains of DP2 (B), HNF‐3γ (C) and Rap30 (D). All four proteins have the same topology. In the DNA‐complexed structures of the DNA‐binding domains of DP2 (Zheng et al., 1999) and HNF‐3γ (Clark et al., 1993) amino acids that contact with DNA are shown in red. Based on the DNA titration experiment of Rap30 (Groft et al., 1998) amino acids whose signals change significantly are also shown in red. This figure was made using the program GRASP (Nicholls et al., 1991).

DNA‐binding ability of TFIIEβc

When the DNA‐binding activities of TFIIE subunits were studied, TFIIEβ could bind to both ssDNA and dsDNA by itself but TFIIEα could not bind directly and bound to both only when added together with TFIIEβ (Y.Ohkuma, unpublished results). However, the binding efficiencies of TFIIEβ were different between ssDNA and dsDNA, and the binding to ssDNA was stronger than to dsDNA. Since the ssDNA‐binding region of TFIIEβ has already been identified in the C‐terminal basic helix–loop region (Okamoto et al., 1998), the dsDNA‐binding regions were studied using the TFIIEβ internal deletion mutants mixed with dsDNA cellulose (Figure 5A). The dsDNA‐bound mutants were detected by anti‐TFIIEβ rabbit polyclonal antibody after Western blotting. Three mutants, Δ96–119, Δ117–153 and Δ257–291, bound more weakly to dsDNA than the other mutants (lanes 6, 7 and 12 versus lanes 1–5 and 8–11). Importantly, mutants Δ96–119 and Δ117–153 lack either the N‐ or C‐terminal region of the Rap30 homology region and mutant Δ257–291 lacks the C‐terminal basic helix–loop region (Figure 5C). To examine whether these regions in TFIIEβ bind to DNA, two domains of TFIIEβ, TFIIEβc (amino acid residues 66–146) and TFIIEβ245–291 (amino acid residues 245–291), were bacterially expressed and purified, and both DNA‐binding activities were tested (Figure 5B). As a result, TFIIEβc bound predominantly to dsDNA (lane 3 versus lane 2) and TFIIEβ245–291 bound to both ssDNA and dsDNA although the binding to ssDNA was stronger than that to dsDNA (lane 5 versus lane 6). Therefore, the dsDNA‐binding regions were localized in both the Rap30 homology region and the C‐terminal basic helix–loop region. Mutant Δ257–291 bound less effectively than the two mutants Δ96–119 and Δ117–153 to dsDNA (Figure 5A, lane 12 versus lanes 6 and 7). There are two possible reasons for this interpretation: one is that the mutant Δ257–291 possesses only the TFIIEβc region but the two mutants Δ96–119 and Δ117–153 possess half of the TFIIEβc region in addition to the TFIIEβ245–291 region; the other is that deletion of residues 257–291 causes structural changes that hinder dsDNA binding at the TFIIEβc region. Judging from these DNA–binding activities, the main dsDNA‐binding domain of TFIIEβ is likely to be the central core domain of TFIIEβ, that is, TFIIEβc.

Figure 5.

DNA‐binding experiments using DNA affinity chromatography. (A) Double‐stranded DNA binding of TFIIEβ. A series of internal deletion mutants of TFIIEβ as well as wild type [300 ng, see (C)] were incubated with double‐stranded DNA cellulose as described in Materials and methods. Bound mutants were detected by Western blotting with anti‐TFIIEβ antisera. TFIIEβ, either wild type or mutants, are indicated at the top. The sizes of the molecular mass markers are indicated on the left (in kilodaltons). (B) Comparison of the DNA‐binding activities between the core domain and the C‐terminal domain of TFIIEβ. The core domain TFIIEβc (IIEβc) and the C‐terminal domain TFIIEβ245–291(IIEβ245–291) were expressed in E.coli with a His6 tag at the N–terminus and purified through Ni‐NTA cellulose. Three hundred nanograms of both domains were incubated with either single‐stranded (s) or double‐stranded (d) DNA cellulose resin. ‘c’ is 10% input. Arrows on the right indicate the position of two polypeptides TFIIEβc (His6‐IIEβc) and TFIIEβ245–291(His6‐IIEβ245–291). The sizes of the molecular mass markers are indicated on the left (in kilodaltons). (C) Internal deletion mutants of TFIIEβ. All mutants were designed according to the putative structural motifs and characteristic sequences as shown below. The numbers on the left indicate the actual amino acid residues deleted and are used to denote each mutant (the deleted portion of TFIIEβ is shown as a gap).

To detect the dsDNA binding surface of TFIIEβc, we examined the chemical shift changes of backbone amide proton and nitrogen signals of TFIIEβc by adding dsDNA consisting of the −14 and −2 region of the adenovirus 2 major late (Ad2ML) promoter. By adding 0.5, 1.0, 1.5 and 2.0 molar ratios of dsDNA to TFIIEβc the chemical shifts of backbone amide proton and nitrogen signals changed gradually. In addition, we examined the chemical shift changes of the imino proton signals of dsDNA by adding TFIIEβc. Both spectral changes are explained by a fast exchange process. Figure 6A and B shows the chemical shift changes of amide proton and nitrogen signals for a 1:1 complex of TFIIEβc with dsDNA. It is apparent that Val77, Leu78, Asn83, Lys86, Glu105, Thr106, Gln107, Asp110, Ile111, Lys129, Lys140 and Lys142 showed significant chemical shift changes upon binding to DNA. These amino acids were mapped on the structure of TFIIEβc as shown in Figure 6C. The binding surface of TFIIEβc with dsDNA is identified in helix H1, the loop between helices H2 and H3, and the N‐ and C‐termini of the β hairpin, in contrast to the DNA‐binding surface in the winged helix proteins (Figure 4). This is consistent with the results from assays using the other regions of the same Ad2ML promoter as well as other promoters (data not shown).

Figure 6.

Chemical shift changes. (A) Overlay of the two‐dimensional 1H‐15N HSQC spectra of TFIIEβc without and with DNA. The spectrum of 1.0 mM TFIIEβc alone is shown in black and the spectrum after the addition of 1.0 mM dsDNA is shown in red. The residues that showed significant chemical shift changes are labeled. (B) Chemical shift changes of amide‐protons and nitrogens of the TFIIEβc backbone on the DNA titration experiment. The backbone signals of Ala66–Lys74, Gly76, Lys114 and Asn126 could not be observed in this NMR condition. The weak backbone signal of His108 was only observed in the DNA free form. Chemical shift changes after addition of 1.0 mM DNA into 1.0 mM [15N]TFIIEβc are plotted in black and red for the positive and negative changes, respectively. (C) Mapping of the residues that have chemical shift changes onto the TFIIEβc. The residues that showed relatively large chemical shift changes (absolute chemical shift differences >0.06 p.p.m. for NH, >0.50 p.p.m. for 15N) are marked in red on the backbone of TFIIEβc.

Figure 7 shows the electrostatic potential surface of TFIIEβc. On the surface of TFIIEβc, there is a positive concave patch, formed by Lys86, Lys129 and Lys142, which might be a binding site for the phosphate backbone of DNA. Figure 7 shows that each of the winged helix proteins, the DNA‐binding domains of DP2, HNF‐3γ and Rap30 has a positive surface around the helix–turn–helix (HTH) motif, while the corresponding surface of TFIIEβc is rather negative (upper panel). In particular, H2, the first helix of the HTH motif of TFIIEβc, contains four negative amino acids in the amphipathic helix of Leu‐AspGlu‐Ile‐Leu‐AspGlu, whose sequence is highly conserved in other species (Figure 1B). The positive surface of TFIIEβc is observed on the opposite surface instead (lower panel). Therefore, the DNA‐binding modes of TFIIEβc revealed by the NMR experiment can be explained by the electrostatic potential surface of TFIIEβc.

Figure 7.

Electrostatic potential surfaces of TFIIEβc and winged helix proteins. The electrostatic potential surface of TFIIEβc (residues 71–143), the DNA‐binding domains of DP2 (Zheng et al., 1999), HNF‐3γ (Clark et al., 1993) and Rap30 (Groft et al., 1998). The molecular surfaces are colored according to electrostatic potential. Positive potential is shown in blue and negative potential is in red. The surface around the HTH motif of each protein is shown in the upper part with the same orientation as in Figure 4. The lower part of each protein is rotated 180° relative to the axis of each protein of the upper part. The electrostatic potential surfaces were made using the program GRASP (Nicholls et al., 1991).

Function of TFIIEβc in the preinitiation complex

Although TFIIEβc contains a region similar to the Pol II binding site of Rap30 (TFIIFβ), its structure is very close to the winged helix motif. The physiological Pol II binding region of TFIIEβ has been investigated using various deletion mutants of TFIIEβ, and the C‐terminal region, rather than the central core domain, has been identified to bind to Pol II (Y.Watanabe and Y.Ohkuma, unpublished results). Taking into account the various roles of TFIIEβ, the central core domain is likely to be the dsDNA binding region. Actually, TFIIEβc alone interacts with dsDNA, independently of nucleotide sequence (Figure 5B; data not shown). This relatively weak dsDNA binding ability of TFIIEβc seems to be essential for a specific function of TFIIE. It has been puzzling how PIC is activated and consequently promoter melting occurs. Recently, evidence has accumulated showing that TFIIE is a key factor for elucidation. For example, the previous general transcription factor‐promoter DNA photo‐crosslinking experiment has indicated that TFIIE actually binds to dsDNA between positions −14 and −2 upstream from the transcription start site (+1) including the region (around −10 to −8) where dsDNA starts to be single‐stranded upon promoter melting (Robert et al., 1996). Also, the two‐dimensional crystallography of yeast TFIIE (yTFIIE) with Pol II has demonstrated that yTFIIE binds to the active center of Pol II, which is located near the transcription initiation site (Leuther et al., 1996). We suggest here that the weak but significant binding ability of TFIIEβc to dsDNA is essential for the regulation of the position of the C‐terminal region of TFIIEβ onto DNA as well as Pol II, i.e. for the activation of PIC and promoter melting. In PIC formation, the correct position of TFIIE on the promoter DNA should be arranged by already preassembled general transcription factors, such as TFIID, TFIIA, TFIIB and TFIIF as well as Pol II. During this process, TFIIE will bind to dsDNA using the positively charged concave surface of TFIIEβc (Figure 7), and to Pol II near the active center (Leuther et al., 1996) at the C‐terminal Pol II binding region (Y.Watanabe and Y.Ohkuma, unpublished results). After PIC formation, the single‐stranded DNA‐binding ability of the C‐terminal basic region of TFIIEβ might be necessary for promoter melting. It is also possible that a special tertiary structure of DNA formed during promoter melting, rather than normal B form DNA, could be recognized tightly by TFIIEβc. Actually and intriguingly, a novel DNA bend was observed upon PIC formation at the position between −5 and +5 containing the +1 transcription start site (Robert et al., 1998). However, further examination is needed to unveil the exact role of TFIIEβ.

Materials and methods

Expression and purification of recombinant human TFIIEβ

Recombinant human TFIIEβ was expressed in E.coli BL21(DE3) pLysS by induction with isopropyl‐β‐d‐thiogalactopyranoside (IPTG) (Studier et al., 1990). For purification, soluble bacterial lysates were used. His6‐tagged human TFIIEβ (His6‐TFIIEβ) was purified through an Ni‐nitrilotriacetic acid (NTA) column (Qiagen) by eluting with 100 mM imidazole–HCl pH 7.9. The large scale preparation of TFIIEβ was as described before (Sumimoto et al., 1991), and resulted in >95% purity as judged by Coomassie Blue staining of an SDS–polyacrylamide gel.

Identification of the central core domain of TFIIEβ

Five hundred nanograms of His6‐TFIIEβ were partially digested by 0.25 ng of trypsin or α‐chymotrypsin for 4 h at 25°C. Fragments containing the His6‐tagged N‐terminus of TFIIEβ were separated through an Ni‐NTA column and the core domain was identified in the unbound fraction; 10% of total core domain was separated by SDS–PAGE, transferred to the PVDF membrane, and the peptide sequence was determined by the peptide sequencer (ABI 492), the rest of the sample was separated by ODS 120T reverse‐phase chromatography (TOSOH) and the separated core domain was either sequenced by the peptide sequencer (ABI 492) or weighed by mass spectrometry (MALDI/MASS, PerSeptive Biosystems). Following trypsin digestion, the peptide sequence ‘ALSGSSGYKF’ was determined by SDS–PAGE analysis and a little shorter sequence ‘ALSGSSG’ was determined by reverse‐phase chromatography. The measured molecular mass was 9274.3 Da. Thus, this domain was mapped from amino acid residue 66 to 146 and the calculated molecular mass of this region was 9271.90 Da. Following α‐chymotrypsin digestion, the peptide sequences ‘KFGVLAK’ and ‘SVDEEFQ’ were determined by SDS–PAGE analysis. The molecular masses were 9234.6 and 4061.2 Da, respectively. These correspond to amino acid residues 74–152 and 211–244. Taking these results into consideration, the region consisting of amino acid residues 66–146 was judged to be the TFIIEβ core domain.

Sample preparation

The nucleotide sequence of the central core domain, corresponding to amino acid residues 66–146 of human TFIIEβ, was amplified by the polymerase chain reaction (PCR) using two oligonucleotides, BS1‐1T (5′‐GACTGCATAGGCTTTGTCAGGAAGCTC‐3′) and BS1‐1B (5′‐CATCAGGATCCTTCTATCTCACGTTGTACTTGG‐3′), digested with NdeI and BamHI, and subcloned into the NdeI and BamHI restriction sites of the pET3a expression vector (Novagen). Therefore, the coding region contains methionine at the N‐terminus. The plasmid was transformed into E.coli BL21(DE3) pLysS (Novagen). The E.coli cells were grown at 30°C in LB broth medium or in M9 minimal media containing [15N]ammonium chloride with or without [13C]glucose. IPTG (1 mM) was added at OD600 = 0.4–0.5. After 5 h growth the cells were harvested. The cell pellet was resuspended in buffer A (20 mM Tris–HCl pH 7.0, 10% glycerol, 1 mM EDTA, 1 mM PMSF, 1 mM benzamidine and 0.1 M KCl). The cells were lysed by sonication on ice and centrifuged, and the supernatant was loaded onto the phosphocellulose (P11, Whatman) column, equilibrated with buffer A. The protein sample was eluted by a KCl linear gradient from 0.1 to 1 M. The peak fractions were pooled and the buffer was changed to buffer B (20 mM Tris–HCl pH 7.0, 10% glycerol, 1 mM EDTA and 0.1 M KCl). Then, the sample was loaded onto the S‐Sepharose (Amersham Pharmacia Biotech) column, equilibrated with buffer B, and was eluted by a KCl linear gradient from 0.1 to 0.5 M. The peak fractions were pooled. The sample was concentrated to get the volume below 3 ml using centriprep membrane with mol. wt 3000 cutoff, applied onto Superdex 30 (Amersham Pharmacia Biotech) equilibrated by 20 mM Tris–HCl pH 7.0, 10% glycerol, 1 mM EDTA and 1 M KCl, and the final sample fractions were collected.

NMR spectroscopy

Sample solution for NMR experiments in H2O is ∼200 μl of 1.0–2.0 mM protein concentration in 20 mM potassium phosphate buffer pH 6.0, 500 mM NaCl and 10% (v/v) of D2O was added. All NMR experiments were carried out at 27°C on either a Bruker AMX2‐500 or DMX‐600 spectrometer equipped with a triple‐resonance gradient probe. The sequential resonances of backbone were assigned using 3D CBCA (CO)NH, CBCANH experiments (Grzesiek and Bax, 1992). The side chain resonances were assigned using 2D DQF‐COSY (Rance et al., 1983), TOCSY (Bax and Davis, 1985), 3D HBHA(CO)NH (Grzesiek and Bax, 1992), 15N‐edited TOCSY‐HSQC (Marion et al., 1989) and HCCH‐TOCSY (Kay et al., 1993) experiments. Stereospecific assignments were obtained from a combination of 3D HNHB (Archer et al., 1991), 2D DQF‐COSY and 3D 15N‐edited NOESY‐HMQC (50 ms mixing time) (Kay et al., 1989) experiments. Distance information was obtained from 2D NOESY (Macura and Ernst, 1980), 3D 15N‐edited NOESY‐HMQC (50 and 150 ms mixing time), 3D 13C‐edited NOESY‐HSQC (Muhandiram et al., 1993) and 4D 13C, 13C‐edited HMQC‐NOESY‐HSQC (100 ms mixing time). The backbone torsion angles φ were obtained from 3D HNHA (Vuister and Bax, 1993) and 2D HMQC‐J (Kay and Bax, 1990) experiments. Spectra were processed using NMRPipe (Delaglio et al., 1995), and analyzed using the programs PIPP, CAPP and STAPP (Garrett et al., 1991).

Structure calculation

Interproton distance restraints derived from NOE intensities were grouped into three distance ranges, 1.9–3.0, 1.9–4.0 and 1.9–5.0 Å, corresponding to strong, medium and weak NOEs, respectively. An additional 0.5 Å was added to the upper limits for distances involving methyl groups to account for the higher apparent intensity of methyl resonances, and 2.0 Å pseudoatom correction was added to degenerate phenylalanine and tyrosine ring protons. Dihedral angles estimated from 3JHNα coupling constants were restrained to −90° < φ <−40° for 3JHNα ≤ 5.0 Hz and −160° < φ < −80° for 3JHNα ≥8.0 Hz. Hydrogen bond restraints in areas of regular secondary structure were introduced at the final stages of refinement. Structure calculations were performed using the four‐dimensional simulated‐annealing program EMBOSS (Nakai et al., 1993) on an Indigo 2 Impact Silicon Graphics workstation. A total of 100 simulated annealing structures were calculated, and 20 structures had no NOE violations larger than 0.3 Å and no dihedral angle violations greater than 1°. Structural statistics are summarized in Table I. The atomic coordinates of the 20 structures and the averaged structure have been deposited in the Protein Data Bank as 1D8J and 1D8K, respectively. Structures were analyzed and displayed using PROCHECK‐NMR (Laskowski et al., 1996), GRASP (Nicholls et al., 1991), MOLMOL (Koradi et al., 1996) and SYBYL (Tripos Inc., St Louis, MO).

Generation of polyclonal antibody against TFIIEβ

Two hundred micrograms (100 μl) of purified His6‐TFIIEβ (>99% pure) were mixed with the same volume (100 μl) of complete Freund's adjuvant (Difco) and injected into each rabbit. Two weeks after the first injection, a second injection was carried out with 100 μg (100 μl) of purified TFIIEβ in 100 μl of incomplete Freund's adjuvant (Difco). A third injection was carried out 2 weeks later using the same procedures as described for the second injection. Blood was collected 8 days after the third injection. We have checked by Western blotting that the generated antibody recognized all of the TFIIEβ deletion mutants used in this study (Y.Ohkuma, data not shown).

DNA‐binding assays by single‐ and double‐stranded DNA cellulose

Three hundred nanograms of each TFIIEβ mutant were incubated with 7 μl (packed volume) of either ss or ds DNA cellulose (Sigma) in a 500 μl reaction volume of buffer C [20 mM Tris–HCl pH 7.9 at 4°C, 0.5% EDTA, 20% (vol/vol) glycerol, 0.5 mM PMSF, 10 mM 2‐mercaptoethanol, 0.002% (vol/vol) Nonidet P‐40] containing 100 mM KCl (BC100) with 200 μg/ml bovine serum albumin for 4 h at 4°C with rotation. The resin was then washed twice with 500 μl of buffer C containing 200 mM KCl (BC200), once with 500 μl of BC100, boiled in SDS sample buffer, and analyzed by SDS–PAGE (15% acrylamide). Bound mutants were detected by Western blotting with anti‐TFIIEβ antisera (1:3000 dilution) as described above.

DNA titration

Titration buffer comprised 4.5 mM potassium phosphate pH 7.2 in H2O/D2O (10/1). A 13 bp oligonucleotide duplex sequence was 5′‐GCGCGTTCGTCCT‐3′, derived from Ad2ML promoter from −14 to −2. DNA titration was performed by adding 0.5 mM Ad2ML promoter into 1.0 mM [15N]TFIIEβc up to 2.0 mM. Chemical shift changes of backbone amide and nitrogen signals of TFIIEβc were monitored by 2D 1H‐15N HSQC spectrum (Kay et al., 1992).


We thank Kosuke Morikawa (BERI) for helpful discussion. This work was supported by grants from the Mitsubishi Foundation (to Y.N.), in part by Grants in Aid for Scientific Research on Priority Areas (06276103, 06276104, 17109102, 111154221 and 11358012 to Y.N., and 09249211 to Y.O.) from the Ministry of Education, Science and Culture of Japan and by the Core Research for Evolutional Science and Technology (CREST) from the Japan Science and Technology Corporation (to Y.O.). M.O. is a JSPS Research Fellow.