Genes For Tfb2, Tfb3, and Tfb4 Subunits of Yeast Transcription/Repair Factor IIH

Genes for the Tfb2, Tfb3, and Tfb4 subunits of yeast RNA polymerase transcription factor IIH (TFIIH) are described. All three genes are essential for cell viability, and antibodies against Tfb3 specifically inhibit transcription in vitro. A C-terminal deletion of Tfb2 caused a defect in nucleotide excision repair, as shown by UV sensitivity of the mutant strain and loss of nucleotide excision repair activity in cell extracts (restored by the addition of purified TFIIH). An interaction between Tfb3 and the Kin28 subunit of TFIIH was detected by the two-hybrid approach, consistent with a role for Tfb3 in linking kinase and core domains of the factor. The deduced amino acid sequence of Tfb2 is similar to that of the 52-kDa subunit of human TFIIH, while Tfb3 is identified as a RING finger protein homologous to the 36-kDa subunit of murine CAK (cyclin-dependent kinase activatingkinase) and to the 32-kDa subunit of human TFIIH. Tfb4 is homologous to p34 of human TFIIH and is identified as the weakly associated 37-kDa subunit of the yeast factor. These and other findings reveal a one-to-one correspondence and high degree of sequence similarity between the entire set of yeast and human TFIIH polypeptides.

activity. In addition, TFIIH is also indispensable for another fundamental process, nucleotide excision repair of DNA damage, as well as appearing to relate in some way to cell cycle control.
Genes for six subunits of yeast TFIIH and eight subunits of the human protein have so far been described. In every case, the yeast and human polypeptides are homologous and appear to perform similar roles. Four polypeptides impart enzymatic activity; yeast (human) Ssl2 (XPB) and Rad3 (XPD) are ATPase/helicases, and Kin28 (MO15/CDK7) and Ccl1 (cyclin H) constitute a cyclin-dependent kinase/cyclin pair, with specificity for the C-terminal domain of RNA polymerase II. Yeast TFIIH can be dissociated into a core complex, which includes both ATPase/helicases, and a kinase complex, comprising the kinase and cyclin subunits, termed TFIIK (2). While the ATPase/helicases were the first components of TFIIH shown to be involved in DNA damage repair (3,4), additional subunits of the core complex prove to be required for repair as well (5)(6)(7)(8)(9). The human counterpart of TFIIK, known as CAK, was originally identified on the basis of its capacity to activate cyclindependent kinases involved in cell cycle control through phosphorylation of a key threonine residue in vitro (10,11). Genetic studies in yeast, while supporting the essential role of TFIIH in RNA polymerase II transcription and in phosphorylation of the C-terminal domain, have raised doubts as to the requirement of TFIIK for cyclin-dependent protein kinase function in vivo (12). Yeast CAK activity has been shown to reside in a distinct polypeptide (13,14).
We report here on the isolation and characterization of genes for two subunits of core TFIIH, Tfb2 and Tfb3. The results reveal an unexpected connection between the core protein and TFIIK, and together with a similar analysis of another subunit, termed Tfb4, complete the molecular description of TFIIH. Comparison with a companion study on the human factor (46) discloses an extraordinary degree of conservation between the yeast and human proteins.

EXPERIMENTAL PROCEDURES
TFIIH Purification and Microsequencing-TFIIH was purified to near homogeneity (2), transferred to polyvinylidene difluoride membrane (Bio-Rad), and stained with Ponceau-S (Sigma) as described (15). The bands corresponding to the 55-and 38-kDa subunits of core TFIIH were excised. Amino acid analysis, tryptic digestion, peptide isolation, and microsequencing were performed by W. Lane (Harvard Microchemistry Facility, Cambridge, MA).
Cloning of TFB2-The sequences of two tryptic peptides derived from p55 were as indicated (Fig. 1A). Degenerate oligonucleotide primers, designed on the basis of these sequences, were as follows, where R ϭ A/G, Y ϭ C/T, N ϭ A/C/T/G, and underlined nucleotides represent restriction sites.
1: 5Ј-ATATCTCGAGGATCCTTYTGNACYTCNCCNCC-3Ј 2: 5Ј-ATATTCTAGAAGCTTGGNGGNGARGTNCARAA-3Ј 3A: 5Ј-ATATCTCGAGGATCCCTRTTYTGNACYTGYTG-3Ј 3B: 5Ј-ATATCTCGAGGATCCCGRTTYTGNACYTGYTG-3Ј 4A: 5Ј-ATATTCTAGAAGCTTCARCARTGNCARAAYCG-3Ј 4B: 5Ј-ATATTCTAGAAGCTTCARCARGTNCARAAYAG-3Ј SEQUENCE 1 The polymerase chain reaction (PCR) was performed by the "touchdown" method as described (16) with the exception that the annealing temperature varied from 60 to 52°C. Primers 1 and a mixture of nucleotides 4A and 4B yielded a single major product of approximately 350 bp. This fragment was cloned into Bluescript SK ϩ (Stratagene) to give pBS/55PCR. Sequencing by the dideoxynucleotide chain termination method (Sequenase kit, United States Biochemical Corp.) revealed a 112-amino acid open reading frame (ORF). The deduced amino acid sequences adjacent to the primers matched the sequences of the tryptic peptides. The cloned PCR fragment was labeled by random priming (17) and used as a probe to screen a yeast genomic library prepared from strain BJ926 (18) as described (16). One positive clone was obtained (pBS/TFB2), which was sequenced and found to contain the entire ORF encoding Tfb2. The entire TFB2 gene was sequenced completely on both strands using a combination of custom designed primers and subcloning techniques.
Cloning of TFB3-The sequences of two tryptic peptides derived from p38 were as indicated (Fig. 1B). Degenerate oligonucleotide primers, designed on the basis of these sequences, were as follows, where R ϭ A/G, Y ϭ C/T, D ϭ A/G/T, H ϭ A/C/T, N ϭ A/C/T/G, and underlined nucleotides represent restriction sites: 1: 5Ј-ATATTCTAGAAGCTTCARATHTTYGAYGAYGT-3Ј 2: 5Ј-ATATCTCGAGGATCCTTRAARTANGGRTTRTT-3Ј 3: 5Ј-ATATTCTAGAAGCTTAAYAAYCCNTAYTTYAA-3Ј 4: 5Ј-ATATCTCGAGGATCCACRTCRTCRAADATYTG-3Ј SEQUENCE 2 PCR was performed as described above. Primers 1 and 2 yielded a single major product of approximately 500 bp. This fragment was cloned into Bluescript SK ϩ to give pBS/38PCR. Sequencing revealed a 170-residue ORF, with deduced amino acid sequences adjacent to the primers corresponding to the sequences of the tryptic peptides with the exception of two PCR-induced missense mutations. The cloned PCR fragment was used to screen a yeast genomic library as described above. Two positive clones were obtained, one of which (pBS/TFB3) was sequenced and found to contain the entire Tfb3 ORF. The TFB3 gene was completely sequenced on both strands using a combination of custom designed primers and subcloning techniques.
C-terminal Deletion of Tfb2-A deletion in TFB2 was made by digestion of pRS315/TFB2 with NdeI and MluI, followed by end filling and religation. In the resulting construct, pRS315/TFB2⌬, the 60 C-terminal residues of Tfb2 have been replaced by six amino acids derived from translation of downstream sequences. The C terminus of Tfb2⌬ is Leu 448 -Asp-Arg-Val-Ile-Thr-Arg-Val-Phe-Leu-Ala-Gln, where residues derived from Tfb2 are indicated in bold. The integrity of pRS315/TFB2⌬ was confirmed by sequencing. pRS315/TFB2 was made by cloning the 2.5-kb XbaI fragment (see below) into the corresponding site of pRS315 (19). To generate the TFB2 C-terminal deletion strain (and the isogenic wild type), pRS315/TFB2⌬ (pRS315/TFB2) was transformed into a haploid derivative of CRY3 containing the chromosomal disruption of TFB2 and pRS316/TFB2 (see below). The latter plasmid was subsequently cured by selection on media containing 5-fluoroorotic acid. Compared with the wild type, the deletion strain exhibited a slow growth phenotype at 30°C and was slightly temperature-sensitive at 37°C (data not shown).
Recombinant Tfb2, Tfb3, and Tfb4 -The TFB2 ORF was amplified by PCR with the primers 5Ј-ATATCCATGGGAAGTGACTATTCCCTG-AA-3Ј and 5Ј-ATATGGATCCTAGTGATGATGGTGGTGATGTTGTTT-CTTTTTCAACTT-3Ј, introducing NcoI and BamHI restriction sites (underlined) at the 5Ј and 3Ј ends of the ORF, respectively, and also introducing six histidine codons at the 3Ј end. The resulting NcoI-Ba-mHI fragment was cloned into pET-11d (Novagen). The TFB3 ORF was amplified by PCR with the primers 5Ј-ATATCATATGCATCACCATC-ACCATCACCTTATGGATGAGTATGAGGA-3Ј and 5Ј-ATATGGATCC-TTAAAGCTCCTCGGATATAA-3Ј, introducing NdeI and BamHI re-striction sites (underlined) at the 5Ј and 3Ј ends of the ORF, respectively, and also introducing six histidine codons at the 5Ј end. The resulting NdeI-BamHI fragment was cloned into pET-11a (Novagen). The TFB2 or TFB3 coding region in each expression plasmid was partially sequenced.
Cells (BL21/DE3) harboring either the Tfb2 or Tfb3 expression constructs were grown at 37°C to an A 600 value of 0.6 and induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. Both recombinant Tfb2 and Tfb3 were expressed at a high level and found to be predominantly insoluble.
Following induction, the cultures (3 liters) were centrifuged and the cells resuspended in 50 ml of lysis buffer (10% glycerol, 20 mM HEPES, pH 7.6, 500 mM NaCl, 0.2% Tween 20, 2 mM imidazole, 5 mM ␤-mercaptoethanol) and lysed by sonication. Insoluble material was collected by centrifugation and resuspended in 100 ml of insoluble lysis buffer (8 M urea, 100 mM sodium phosphate [pH 8.0], 10 mM Tris, 2 mM imidazole, 5 mM ␤-mercaptoethanol) as described (15). The remaining insoluble material was removed by centrifugation and the supernatant bound in batch to 5 ml of Ni 2ϩ -nitrilotriacetic acid-agarose (QIAGEN) for 2 h at room temperature. The columns were washed with 25 ml of insoluble lysis buffer followed by 50 ml of insoluble lysis buffer containing 20 mM imidazole. Nearly homogeneous recombinant Tfb2 and Tfb3 were eluted with 15 ml of insoluble lysis buffer containing 500 mM imidazole. The yield for each protein was about 6 mg/liter of starting culture.
The ORF encoding Tfb4 was amplified by polymerase chain reaction from yeast genomic DNA with the primers 5Ј-CCGGGATCCTTCA-TATGCACCACCACCACCACCACGATGCAATATCTGATCCAACGT-TT-3Ј, and 5Ј-CCGGATCCGAATTCTCATGGTTTGCTCACCTTCTTTT-T-3Ј, introducing BamHI and EcoRI restriction sites (underlined) at the 5Ј and 3Ј of the ORF, respectively, and also introducing six histidine codons at the 5Ј end. The BamHI-EcoRI fragment was cloned into the corresponding sites of the bacterial expression plasmid pGEX-3X (Pharmacia Biotech Inc.), resulting in fusion to the gene encoding glutathione S-transferase. Recombinant glutathione S-transferase/Tfb4 protein was expressed in bacteria by growth at 37°C to an A 600 of 0.5 and then addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 0.5 mM. Cells were grown for 3 h after induction. Recombinant Tfb4 was purified from the insoluble fraction as described for Kin28 (15).
Preparation of Tfb2 and Tfb3 Antisera and Purification-Recombinant Tfb2, Tfb3, and glutathione S-transferase/Tfb4 were fractionated by SDS-PAGE and visualized by staining with aqueous 0.1% Coomassie Blue R-250. The protein-containing region was excised and used to inoculate rabbits (Berkeley Antibody Company, Richmond, CA). Antibody against Tfb2 and Tfb3 was purified on Protein A-Sepharose CL-4B columns (Sigma) as follows. 200 l of 1 M Tris-HCl (pH 8.0) were added to 2 ml of each antiserum or preimmune serum and subsequently applied to 400-l Protein A columns. Each column was washed with 4 ml of 100 mM Tris-HCl (pH 8.0) followed by 4 ml of 10 mM Tris-HCl (pH 8.0). Bound antibody was eluted with 100 mM glycine (pH 3.0) and immediately neutralized by the addition of 0.1 volume of 1 M Tris-HCl (pH 8.0). Protein containing fractions (500 l) were pooled to give fractions containing approximately 4 mg/ml IgG.
Southern, Northern, and Immunoblot Analysis-For Southern blots, 3 g of yeast genomic DNA from strain BJ926 were digested with the indicated restriction enzymes and analyzed as described previously (16). Northern analysis and probe preparation and hybridization conditions for both Southern and Northern blots were as described (16). Immunoblots were performed according to Chasman and Kornberg (20). Protein A-purified anti-Tfb2 and anti-Tfb3 antibodies were mixed and used at a final dilution of 1/500. The anti-Tfb4 antiserum was used at a final dilution of 1/1000. The secondary antibody/detection reagent was a goat anti-rabbit alkaline phosphatase conjugate (Bio-Rad). Silver staining was as described (21).
Yeast Two-hybrid Analysis-Plasmids for the two-hybrid analyses shown in Table I were constructed as follows: pAS1-CYH2/KIN28 and pAS1-CYH2/TFB3 were made by subcloning the NdeI/BamHI fragments from either pET-11a/KIN28 (15) or pET-11a/TFB3 into the same sites in pAS1-CYH2 (22), a derivative of pAS1 (22) suitable for cycloheximide counterselection. pACTII/TFB3 was made by subcloning the NdeI (blunt)/BamHI fragment from pET-11a/TFB3 into the SmaI and BamHI sites of pACTII (22). pACTII/CCL1 was made by cloning the NdeI/BlpI (blunt) fragment from pET-20b/CCL1 (23) into the NdeI and SmaI sites of pACTII. To make pAS1-CYH2/TFG2, the TFG2 ORF was amplified by PCR as described (24) and cloned into the NcoI and BamHI sites of pAS1-CYH2. The integrity of the ORF at the fusion junction of each plasmid was confirmed by sequencing. pSE1112 contains the SNF1 gene fused to the GAL4 binding domain in pAS1 (provided by S. Elledge, Baylor College of Medicine, Houston, TX). pGAD/RAD3 was as described (25).
Disruptions of TFB2, TFB3, and TFB4 -To disrupt the TFB2 gene, the AatII (blunt)/NaeI fragment from pRS303 (19) containing the HIS3 gene was cloned between the EcoRV and ClaI (blunt) sites of pBS/TFB2/ Xba/2500/⌬XB to give pKO/TFB2. The former plasmid was constructed Tfb2, Tfb3, and Tfb4 Subunits of Yeast TFIIH by subcloning the approximately 2.5-kb XbaI fragment from pBS/TFB2 containing the entire TFB2 ORF into the XbaI site of Bluescript SK ϩ , followed by removal of the XhoI and BamHI sites in the polylinker by digestion, blunting, and religation. For the disruption of the TFB3 gene, the same HIS3 fragment was cloned between the ApaI (blunt) and ClaI (blunt) sites of pBS/TFB3/3500 to give pKO/TFB3. pBS/TFB3/3500 was constructed by subcloning the approximately 3.5-kb HindIII fragment (blunt) from pBS/TFB3 containing the entire TFB3 ORF into the KpnI (blunt) and EcoRV sites of Bluescript SK ϩ . Prior to transformation into the diploid strain CRY3 (MATa/␣ ade2-1 can1-100 his3-11 leu2-3 trp1-1 ura3-1) (27), pKO/TFB2 was digested with XbaI while pKO/ TFB3 was linearized by digestion with SmaI. To ensure that homologous recombination had occurred, genomic DNA was isolated and Southern blot analysis performed as described above on several His ϩ transformants. Strains carrying the correct null mutations were sporulated, and tetrads were dissected by micromanipulation (28) on YPD agar and replica plated onto synthetic media lacking histidine. Rescue of either the TFB2 or TFB3 disruption mutants was achieved by transformation with either pRS316/TFB2 or pRS316/TFB3. pRS316/TFB2 and pRS316/TFB3 were made by cloning either the 2.5-kb TFB2 XbaI or Residues comprising the RING finger are in bold. Residue numbers are indicated at right. The alignment was generated using the CLUSTAL V program (42) as described (15). C, comparison of the deduced amino acid sequence of Tfb4 (Y, lower line) and the 34-kDa subunit of human TFIIH (H, upper line). The alignment was generated as in A. The GenBank™/EMBL accession no. for TFB4 is Z49219. Percentiles given in the text were calculated as in A. the 3.5-kb TFB3 HindIII fragments (see above) into the corresponding sites of pRS316 (19).
For the disruption of TFB4, a cassette encoding the LEU2 marker was amplified by PCR from pRS305 (19) using primers which contained 17-mer oligonucleotide sequences for the amplification, flanked by 61mer sequences that were homologous to those immediately upstream and downstream from the TFB4 open reading frame, respectively. The sequences of the primers were: 5Ј-AGTTTTGAAATAAACAAGTCCTA-AAAGCACCTAAGGAAAATCGAAGAACACCCTGACAAAGATGCGG-CATCAGAGCAG-3Ј and 5Ј-TAAATTTCTGCTTGGAAAACCGGCCAT-GTCGGCGGCACATAAAAGTTCTATTTACCTTTAACTTACGCATCT-GTGCGG-3Ј. The product from the PCR amplification was used directly to transform diploid W303 (MATa/␣ ade2-1 trp1-1 can1-100 leu2, -3,  112 his3-11, 15 ura3) and cells plated on synthetic medium lacking leucine. To ensure that homologous recombination had occurred, PCR reactions with primers spanning the TFB4 gene, and with a primer from within the LEU2 gene together with one in the flanking sequence of TFB4, were performed on several Leu ϩ transformants. Strains carrying the correct mutation were sporulated and tetrads were dissected on YPD agar and replica plated onto synthetic media lacking leucine (28).

Genes for 55-and 38-kDa Subunits of Yeast TFIIH-Hexa-
histidine-tagged yeast TFIIH was purified to near homogeneity by a procedure involving affinity chromatography on Ni 2ϩnitrilotriacetic acid-agarose (2), resolved by SDS-PAGE, and transferred to a polyvinylidene difluoride membrane (29). Bands corresponding to the 55-and 38-kDa subunits were subjected to tryptic digestion, and two peptides from each subunit (underlined in Fig. 1) were sequenced by sequential Edman degradation. Degenerate oligonucleotide primers based on the peptide sequences were used to amplify fragments from yeast genomic DNA by PCR. The amplified fragments were apparently from protein coding regions, since the sequences of the fragments contained ORFs of 112 and 170 amino acids for 55-and 38-kDa subunits, respectively, and since each fragment formed a single band on hybridization to blots of poly(A) ϩ RNA (data not shown). The fragments were evidently derived from single-copy genes, since hybridization to blots of restriction enzyme-digested genomic DNA yielded single bands as well (data not shown). The fragments were used to screen a yeast genomic library, and genes encoding the 55-and 38-kDa subunits, designated TFB2 and TFB3 (transcription factor b subunits 2 and 3), respectively, were obtained. TFB2 contained an ORF of 1539 bp, encoding a polypeptide of 513 amino acids with a molecular mass of 58.6 kDa, while TFB3 contained an ORF of 963 bp, encoding a polypeptide of 321 amino acids with a molecular mass of 32.4 kDa (Fig. 1A), values consistent with the molecular masses of the subunits from SDS-PAGE. The sequences of all tryptic peptides were found within the ORFs (Fig. 1). Since the completion of this work, both TFB2 and TFB3 sequences have been deposited in GenBank by the yeast genome sequencing project. TFB2 is located on chromosome XVI (accession U43503) and TFB3 is located on chromosome IV (accession U33050).
The Tfb3 sequence contains a specialized Zn 2ϩ -binding motif known as a "RING finger" (30). In most occurrences of this motif, it is located near the N terminus and consists of one histidine and seven cysteine residues with a C3HC4 consensus. Tfb3 deviates from the consensus by the insertion of two amino acids (YK) between the last pair of cysteines. In the RING finger of the human Mdm2 protein, a threonine residue is substituted for the third cysteine (31), and it has been proposed that the hydroxyl of this threonine replaces the sulfhydryl of cysteine in the coordination of Zn 2ϩ . The hydroxyl of the tyrosine residue substituted in the RING finger of Tfb3 could coordinate the metal ion as well. With the exception of a number of potential phosphorylation and glycosylation sites (32), Tfb2 and Tfb3 did not contain any other sequence motifs.
Homology of Tfb2 and Tfb3 to Subunits of Human TFIIH and CAK-Searches of the Swiss-Prot and GenBank data bases with the Tfb2 sequence did not reveal significant similarity to any known polypeptide, but a comparison with unpublished sequences from J. M. Egly and colleagues disclosed a homolog in human TFIIH (BTF2). Tfb2 proved to be 63.9% similar and 39.7% identical in sequence to the 52-kDa subunit of the human protein (Fig. 2A). The homology extended over the entire length of the two proteins.
Data base searches with the Tfb3 sequence revealed significant similarity to proteins identified as components of CAK in mouse, human, frog, and starfish (Fig. 2B). The percent similarity/identity between Tfb3 and these proteins were 58.8/32.5 (human), 56.1/32.9 (frog), 55.4/34.2 (starfish), and 63.1/38.5 (mouse). Regions of homology were distributed throughout the proteins but were more pronounced in the N-terminal halves. The RING finger motif was present in all of the homologs, although none of the higher eukaryotic proteins contained the two amino acid insertion between the last pair of cysteines present in the yeast protein. Consistent with our isolation of Tfb3 as a subunit of yeast TFIIH, the human homolog has recently been identified as the 32-kDa subunit of human TFIIH (33).
Both TFB2 and TFB3 Are Essential for Cell Viability-Diploid stains of yeast were made in which a portion of the coding region of one chromosomal copy of TFB2 or TFB3 was replaced Tfb2, Tfb3, and Tfb4 Subunits of Yeast TFIIH by HIS3 (Fig. 1). Gene disruption was confirmed by blot hybridization (data not shown). The strains were sporulated and tetrads were dissected. Only two spores from each tetrad formed colonies on rich media (data not shown). Plating on synthetic media lacking histidine revealed that colonies derived from all viable spores were his Ϫ . his ϩ /ura ϩ colonies were obtained if wild type TFB2 or wild type TFB3 on a centromeric URA3 plasmid was transformed into the corresponding disruption strain prior to sporulation and dissection (data not shown). We conclude that both TFB2 and TFB3 are essential for yeast cell proliferation consistent with a role in transcription.
Tfb2 Is Required for Nucleotide Excision Repair-As mentioned above, all previously characterized subunits of core yeast TFIIH (Ssl2, Rad3, Tfb1, and Ssl1) are required for NER (3,8,9). The finding that Tfb2 and Tfb3 are required for cell proliferation is consistent with an essential role in transcription initiation, but does not address possible roles in DNA repair. To this end, we generated a strain of yeast in which the 60 C-terminal residues of Tfb2 were deleted. Despite this large deletion, the strain was viable, and other than slow growth and slight temperature sensitivity, there was no apparent phenotype. Clearly the extreme C terminus of Tfb2 is not required for its essential function. To investigate a potential role in repair, the tfb2 deletion strain was exposed to UV radiation and found to be about 10-fold more sensitive than the isogenic wild type (Fig. 3A). This degree of sensitivity is similar to that observed with mutations in Tfb1 and Ssl1 (9) but less than that associated with mutations in other repair proteins such as Rad2 (34). UV sensitivity can be caused by a defect in NER. To test this possibility extracts were prepared from the deletion mutant and wild type strains and assayed for NER activity in vitro. The tfb2 mutant extract was completely inactive (Fig. 3B, compare  lanes 1 and 2). NER could be restored by the addition of purified TFIIH (Fig. 3B, lane 3), ruling out the possibility that lack of NER was a secondary consequence of a transcription defect, as the in vitro NER conditions do not support transcription or translation (8). We conclude that Tfb2 is required for NER in yeast. Additional work will be required to demonstrate a similar requirement for Tfb2 in transcription.
In addition to demonstrating the requirement of a particular factor for NER, complementation experiments with various mutant extracts can provide information about molecular associations. For example, tfb1 and ssl1 extracts do not complement one another for NER, consistent with the occurrence of these polypeptides in the same complex (TFIIH; Ref. 9). Similarly, the tfb2 mutant extract was not complemented by a tfb1 extract (Fig. 4, lane 5), but was complemented by a rad14 extract (Fig. 4, lane 6). This finding, together with copurification, provides evidence that Tfb2 is an integral component of TFIIH.
Recombinant Tfb2 and Tfb3-To further characterize Tfb2 and Tfb3, and to confirm the identity of the cloned genes, the proteins were expressed in bacteria and used for the production of polyclonal antisera. Recombinant Tfb2 and Tfb3 nearly comigrated with the 55-and 38-kDa subunits of TFIIH respectively, in SDS-PAGE and displayed similar silver-staining characteristics ( Fig. 5A; the recombinant proteins appeared slightly larger, migrating more slowly than the corresponding TFIIH subunits, due in all likelihood to the addition of hexahistidine tags for ease of purification from bacteria). Following transfer to nitrocellulose, the recombinant proteins and corresponding TFIIH subunits reacted to comparable extents with the anti -FIG. 3. A, UV radiation sensitivity of yeast tfb2 C-terminal deletion mutant. YPD cultures at 30°C in the stationary phase of growth were diluted in YPD medium and plated onto YPD plates. The uncovered plates were than irradiated with UV light at the indicated doses. After incubation at 30°C for 2 days, the surviving colonies were counted. Results are the averages for duplicate experiments. Survival rates are expressed relative to those of nonirradiated cells. TFB2 (filled circles), the wild type strain; tfb2 (open circles), the isogenic tfb2 mutant strain. B, deficient NER in tfb2 mutant extract and its complementation by TFIIH in vitro. NER assays in yeast cell-free extracts were performed at 26°C for 2 h as described previously (43,44). Standard repair reactions were carried out in 250 g of wild type (WT, lane 1) and its isogenic tfb2 mutant (lane 2) extracts. Complementation was achieved by the addition of 300 ng of purified holoTFIIH (2)  bodies against the recombinant proteins (Fig. 5B). These results confirm that the 55-and 38-kDa subunits of yeast TFIIH are encoded by the TFB2 and TFB3 genes.
Insolubility of recombinant Tfb2 and Tfb3 prevented direct tests of functional activity. As an alternative, we assessed inhibition of transcription in vitro by Tfb2 and Tfb3 antibodies, as shown previously for antisera against the Tfb1 and Rad3 subunits of TFIIH (3). Purified Tfb3 antibodies specifically inhibited transcription (Fig. 6), whereas Tfb2 antibodies were without effect (data not shown). Addition of IgG purified from preimmune serum did not significantly inhibit transcription. At the highest level of Tfb3 antibodies tested the level of transcription was reduced to about 45% of the starting value. The relatively high level of residual activity was due, in part, to inhibition by Tfb3 antibodies of expression of only the smaller of two transcripts produced from the CYC1 promoter/G-less cassette fusion template used in this experiment (Fig. 6, top  panel). The selective inhibition of the smaller transcript is inconsistent with a nonspecific effect. This influence on transcription initiation might be due a steric effect of IgG binding to Tfb3, or it could indicate a direct role of TFIIH in the start site selection process. Additional work will be required to confirm a role for Tfb3 in transcription in vivo.
Tfb3 Interacts with Kin28 in Vivo-As mentioned above, yeast TFIIH can be dissociated into a seven subunit core complex including Tfb3, and a kinase complex comprising Kin28 and Ccl1, termed TFIIK. Similar dissociation of human TFIIH evidently occurs, except that MAT1, the human homolog of Tfb3, is recovered as part of the kinase complex (CAK) rather than the core protein (33,35,36). These findings raised the possibility that Tfb3 is involved in linking core and TFIIK. We therefore investigated interactions of Tfb3 with Kin28 and Ccl1 in vivo by the two-hybrid approach, which entails introduction of genes fused to Gal4 DNA-binding and activation domains in yeast (37). Interaction between the fusion proteins yields a hybrid activator whose activity can be monitored by induction of ␤-galactosidase activity from a suitable reporter. Interaction between Kin28 and Ccl1 was previously demonstrated in this way (38). We found evidence of an interaction between Tfb3 and Kin28, which was somewhat stronger than that of Kin28 with Ccl1 in this assay (Table I). Specificity was demonstrated by the failure of Tfb3 to interact with Tfg2, the 54-kDa subunit of yeast TFIIF (24), and by the failure of Kin28 to interact with the Rad3 subunit of TFIIH. We were unable to investigate interaction between Tfb3 and Ccl1 for lack of a suitable fusion of either protein to the Gal4 DNA-binding domain, as required for the two-hybrid approach. Tfb3 fused to the DNA-binding Either Protein A-purified anti-Tfb3 antibodies or antibodies purified from the preimmune serum were added to a reconstituted yeast in vitro transcription reaction. Reaction mixtures were incubated on ice for 10 min after which template, nucleotides, and transcription buffer were added. The amounts of antibodies added are given in micrograms. The inhibited lower transcript is indicated. Transcription was reconstituted with recombinant yeast TBP (70 ng) and a partially purified DE52 fraction (2.5 l) containing all the remaining basal factors (45). Reactions were performed using the CYCG Ϫ template as described (3). Specific transcripts (top panel) were quantified on a PhosphorImager (Molecular Dynamics). Transcription was plotted (lower panel) on the ordinate axis and the amount of antibody added was plotted on the abscissa (anti-Tfb3, diamonds; preimmune, squares). domain gave a significant effect on its own (Table I), while plasmids for expression of Ccl1 fused to the DNA-binding domain could not be constructed, perhaps due to their toxicity in bacteria (23). We conclude from these studies that Tfb3 interacts with Kin28 in vivo and that Tfb3 protein may therefore lie at the interface between core and TFIIK. Dissociation of core and TFIIK is accompanied by partial dissociation of Rad3 from the core as well (39). It is not known whether Rad3 and TFIIK dissociate independently or whether they interact in some way. Evidence for interaction comes from the human system in which a form of CAK was isolated, which included XPD, the human homolog of Rad3 (40). Recently, interaction of XPD with the MO15 subunit of CAK, the human homolog of Kin28, was demonstrated by the two-hybrid approach. 2 We performed the corresponding experiment in the yeast system but could detect no interaction between Rad3 and Kin28 ( Table I). The possibility of interaction between Rad3 and Ccl1 remains to be investigated.
Identification of TFB4 -The cloning of TFB2 and TFB3 revealed that every subunit of human TFIIH had a corresponding subunit in the yeast factor with the exception of one, p34 (6). A search of the yeast data base for proteins similar to human p34 yielded an open reading frame (GenBank™/EMBL accession no. Z49219) encoding a protein with an expected molecular mass of 37.2 kDa, whose sequence was 32.6% identical and 58.8% similar to p34 ( Fig. 2C and Table II). As the presence of a 37-kDa protein had previously been noted in some preparations of yeast TFIIH (2), the open reading frame was cloned and its product was expressed in bacteria. Antiserum against the recombinant protein was raised in rabbits and tested for reactivity toward the subunits of essentially homogeneous yeast TFIIH (from the Mono S step of purification) by immunoblotting. Immune (Fig. 7A) but not preimmune serum (data not shown) specifically recognized a protein with the same mobility as the 37-kDa protein evident from silver staining of the same TFIIH preparation. Moreover, during chromatography of the TFIIH preparation, immune reactivity precisely coeluted with Kin28, with the other subunits of TFIIH (Fig. 7B), and with C-terminal domain kinase activity. We conclude that the ORF, which we term the TFB4 (transcription factor b subunit 4), is the yeast homolog of the human BTF2 p34 gene.
It may be noted that Tfb4 protein was apparently absent from some previous preparations of TFIIH (compare Fig. 7A in this report with Fig. 3 in Ref. 2). Tfb4 may tend to dissociate and be lost during purification of TFIIH, as has been observed for Ssl2 (2,3). Alternatively, silver staining of Tfb4 may be ineffective and erratic, as has been found for Ssl1 and Tfb3 (3).
To test the requirement for TFB4 in yeast, a diploid strain was made in which the entire gene was replaced with LEU2. The strain was induced to sporulate, and the resulting tetrads were dissected. Only two spores from each tetrad formed colonies, and these were leu Ϫ (data not shown), demonstrating that TFB4 is essential for yeast cell growth. DISCUSSION This work, together with a related study on the human homolog of Tfb2 (46), completes the molecular definition of TFIIH. Due to the complexity, scarcity, and fragile nature of the complex, the effort has consumed the better part of a decade. A basis has been laid for elucidation of the roles of the protein in transcription and other fundamental processes.
The parallel pursuit of yeast and human TFIIH has disclosed a remarkable degree of evolutionary conservation. With genes for all subunits apparently now available in both systems, a perfect congruity has emerged. Every component of one system 2 J.-M. Egly, personal communication.
FIG. 7. The 37-kDa subunit of yeast TFIIH cross-reacts with anti-Tfb4 antibodies and co-purifies with Kin28 in highly purified TFIIH. A, purified holoTFIIH (60 l of Mono S fraction 46 (2) was analyzed in a 10% SDS-polyacrylamide gel and visualized by staining with silver (silver), or transferred to nitrocellulose and detected with anti-Tfb4 antibody (Tfb4 ab). The subunits of holoTFIIH are indicated at the left. Protein bands migrating slower than Ssl2 are the only apparent contaminants of the fraction. B, consecutive fractions (60 l) from Mono S (2) were analyzed in a 10% SDS-polyacrylamide gel and visualized by staining with silver (top panel) or transferred to nitrocellulose and probed with anti-Tfb4 antibodies (panel labeled Tfb4) or with affinity-purified anti-Kin28 antibodies (panel labeled Kin28). has a counterpart, significantly similar in amino acid sequence, in the other (Table II). The main difference so far noted between yeast and human TFIIH concerns the relative strengths of interaction among the subunits and consequent patterns of dissociation. Ssl2 tends to be lost, while Tfb3 is retained by the yeast core protein; the human homologs of these proteins, XPB and MAT1, respectively, exhibit the reverse behavior. This difference may be of no functional significance, but together with the identification of MAT1 as a component of the human kinase complex (CAK), it suggested a role for Tfb3 in linking the core protein and kinase complex. Support for this possibility came from the demonstration of Tfb3 interaction with the yeast kinase Kin28 in a two-hybrid analysis.
The homology of Tfb3 with MAT1 strengthens the case for CAK as a component of TFIIH. The homology extends to the occurrence in both proteins of a RING finger motif, deviating in Tfb3 by the insertion of two amino acids between the last pair of cysteines of the motif. While the functional significance of this insertion is unknown, it was not an artifact of cloning or sequence determination, since amplification by PCR of genomic DNA from several yeast strains yielded the same sequence, and since the sequence from the yeast genome project was identical as well.
The RING finger motif has been found in many proteins, including some products of yeast RAD genes, whose mutation confers a UV-sensitive phenotype. A yeast strain containing a truncated form of Tfb2 is sensitive to UV and is completely defective for nucleotide excision DNA repair in vitro. Comparable analyses of Tfb3 and Tfb4 are in progress.