Subunit interactions in yeast transcription/repair factor TFIIH. Requirement for Tfb3 subunit in nucleotide excision repair.

A yeast strain harboring a temperature-sensitive allele of TFB3 (tfb3(ts)), the 38-kDa subunit of the RNA polymerase II transcription/nucleotide excision repair factor TFIIH, was found to be sensitive to ultraviolet (UV) radiation and defective for nucleotide excision repair in vitro. Interestingly, tfb3(ts) failed to grow on medium containing caffeine. A comprehensive pairwise two-hybrid analysis between yeast TFIIH subunits identified novel interactions between Rad3 and Tfb3, Tfb4 and Ssl1, as well as Ssl2 and Tfb2. These interactions have facilitated a more complete model of the structure of TFIIH and the nucleotide excision repairosome.

The yeast transcription/nucleotide excision repair (NER) 1 factor TFIIH has been extensively purified and characterized (1). Comprised of a total of nine individual subunits, holoTFIIH is unique among RNA polymerase II (RNAP II) initiation factors in that it possesses enzymatic activity (2,3). The subunits Rad3 and Ssl2 endow holoTFIIH with bi-directional DNA helicase activity (3). TFIIH also has a kinase activity that phosphorylates the C-terminal domain of Rpb1, the largest subunit of yeast RNA polymerase II. Under some conditions holoTFIIH can dissociate into the seven-subunit coreTFIIH and the two subunit TFIIK subcomplexes (4). C-terminal domain kinase activity has been shown to reside in TFIIK (1,5). The recent isolation of TFB2, TFB3, and TFB4 completed the cloning of genes encoding subunits of yeast TFIIH (6). All TFIIH subunits are encoded by essential genes and have highly conserved counterparts in humans (6).
A requirement for yeast TFIIH in NER was first suggested by the identification of the well characterized DNA helicase and NER protein Rad3 as a subunit of core TFIIH (3). This requirement was subsequently demonstrated directly using an in vitro NER assay (7). To date, viable or conditional mutants of all the subunits of coreTFIIH except Tfb3 have been used to demonstrate a role for these polypeptides in NER (6 -9). In contrast to core TFIIH, a requirement for TFIIK in NER has not been demonstrated. In this study we report the generation of a yeast strain with a temperature-sensitive allele of TFB3 (tfb3 ts ). This strain is sensitive to ultraviolet (UV) radiation in vivo and defective for NER in vitro. We conclude that Tfb3, like all coreTFIIH subunits, is indispensable for NER.
A form of TFIIH has been identified that is associated with all of the other polypeptides known to be required for the early steps of NER in the absence of DNA damage. This large, preformed "super complex" is referred to as the nucleotide excision repairosome (4,10). A number of earlier studies used a variety of approaches to reveal interactions between TFIIH and/or repairosome subunits (see references in Table II). The cloning of genes encoding the Tfb2, Tfb3, and Tfb4 polypeptides has facilitated the inclusion of these subunits as well. We report here two-hybrid interactions between Ssl2 and Tfb2, Rad3 and Tfb3, and Tfb4 and Ssl1. Based on these interactions together with those previously known, we propose a more refined model for the structure of coreTFIIH and the repairosome.
Construction of tfb3 ts Strain-pRS313/TFB3 was mutagenized in vitro with hydroxylamine and transformed into a haploid derivative of YPH500 containing pRS316/TFB3 and a chromosomal TFB3::LEU2 disruption. pRS313/tfb3 ts was identified by its inability to grow in the presence of 5-fluoroorotic acid at the nonpermissive temperature. Sequencing revealed that a cysteine residue in the RING finger had been changed to tyrosine (C16Y). A more detailed description of the isolation of the tfb3 ts allele will be described elsewhere. 2 pRS315/tfb3 ts was made by subcloning the XhoI/SacI fragment from pRS313/tfb3 ts into the same sites of pRS315. pRS315/tfb3 ts was transformed into haploid derivative CRY3-TFB3::HIS3[pRS316/TFB3]. pRS316/TFB3 was subsequently cured from transformants by growth on medium containing 5-fluoroorotic acid.
Two-hybrid Analysis-Plasmids for two-hybrid analysis were constructed as follows: the NcoI/BamHI fragment from pET-11d/TFB2 (6) was subcloned into the same sites of pAS1-CYH2 and pACTII (14) to give pAS1-CYH2/TFB2 and pACTII/TFB2, respectively. The TFB2⌬ open reading frame was amplified by high fidelity polymerase chain reaction from 50 ng of XbaI-digested pRS315/TFB2⌬ (6) with primers 5Ј-ATATCCATGGGAAGTGACTATTCCCTGAA-3Ј and 5Ј-TATAGGA-TCCTTTAGGAAAAGCTTAGATAC-3Ј introducing NcoI and BamHI restriction sites (underlined) on the 5Ј and 3Ј ends of the open reading frame, respectively. The amplified fragment was digested with NcoI and BamHI and cloned into the same sites of pAS1-CYH2 and pACTII to give pAS1-CYH2/TFB2⌬ and pACTII/TFB2⌬. For pAS1-CYH2/TFB4, the TFB4 open reading frame was amplified by high fidelity PCR from yeast genomic DNA with primers 5Ј-ATATCATATGCACCATCACCA-TCACCATGATGCAATATCTGATCCAAC-3Ј and 5Ј-ATATGGATCCT-CATGGTTTCGTCACCTTCT-3Ј, introducing NdeI and BamHI sites (underlined) on the 5Ј and 3Ј ends of the amplified fragment, respectively. The PCR product was cloned directly into pCRII (Invitrogen) to give pCRII/TFB4. The NdeI/BamHI fragment was subcloned into the same sites of pAS1-CYH2 to give pAS1-CYH2/TFB4. For pACTII/TFB4, the PCR product used to construct pPW66R/TFB4 was digested with BamHI and cloned into the same site of pACTII to give pACTII/TFB4 (9). All other constructs for two-hybrid analysis listed in Table I have been previously described (6,11,12).
Other Methods-Growth of cells, preparation of whole cell extracts, measurement of in vitro NER activity, and quantification were as previously reported (8). Determination of UV radiation sensitivity has been described (9). Yeast transformations were performed by a standard lithium acetate protocol. YPD media contained 1% (w/v) yeast extract, 2% (w/v) bacto peptone, and 2% (w/v) dextrose. HoloTFIIH was prepared as previously reported (1).

RESULTS
The C-terminal Region of Tfb3 Is Essential for Viability-To investigate a possible role of Tfb3 protein in NER, we generated a yeast strain containing a viable mutant allele of TFB3. Previous work showed that deletion of a small region of the C terminus of either Ssl2, Tfb1, or Tfb2 resulted in UV radiation sensitivity and defective NER in vitro (6 -8, 16, 17). Unfortunately, deletion of either 22 (tfb3⌬1) or 46 (tfb3⌬2) amino acids from the C terminus of Tfb3 was lethal, indicating that this region is essential for Tfb3 function (data not shown).
We previously showed that when fused to the DNA-binding domain of Gal4, Tfb3 activates expression of a ␤-galactosidase reporter plasmid containing Gal4 binding sites (6). The mutant alleles tfb3⌬1 and tfb3⌬2 also activated transcription in this assay, indicating that the C terminus of Tfb3 protein is not required for this activity (data not shown). We also previously showed that Tfb3 interacts with the Kin28 subunit of TFIIK, leading to the speculation that Tfb3 activates transcription by direct recruitment of TFIIK (6). The observation that tfb3⌬1 can still interact with Kin28 is consistent with this model (data not shown); however, other possible models of Tfb3 transcriptional activation cannot be excluded.
A Conditional Mutation in TFB3 Renders Cells Sensitive to UV Radiation-In view of the inviability of TFB3 deletion mutants, we generated a conditional, temperature-sensitive (ts) allele of the gene. The tfb3 ts allele was isolated by standard yeast genetic techniques as described under "Experimental Procedures." As shown in Fig. 1A, the tfb3 ts strain failed to grow at 37°C but grew normally at 30°C. The isogenic wildtype strain grew at both temperatures (Fig. 1A). An increase in doubling time at the permissive temperature for the mutant strain (TFB3, 105 min; tfb3 ts , 180 min) is consistent with partial impairment of the essential function. When tested for UV radiation sensitivity, a hallmark of defective NER, the mutant was found to be more sensitive than the isogenic wild-type at 30°C (Fig. 1B). The magnitude of this effect was more pronounced at the semipermissive temperature of 33°C (Fig. 1B), supporting our conclusion that UV radiation sensitivity is the result of thermolability of the tfb3 ts protein.
The tfb3 ts Strain Is Defective for NER in Vitro-To rule out the possibility that tfb3 ts is indirectly sensitive to UV radiation as a result of defective transcription of DNA repair genes, we tested the ability of extracts from wild-type and tfb3 ts cells to perform NER in vitro (18,19). Whole cell extracts were prepared from each strain and incubated with two plasmid DNA substrates, one of which contained base damage. NER activity was measured as the amount of repair synthesis specifically occurring on the damaged substrate as described previously (18,19). In this assay tfb3 ts was almost completely defective for NER ( Fig. 2A, lanes 4 -5), whereas the wild-type extract possessed considerable activity ( Fig. 2A, lanes 1-3). To confirm that defective NER in tfb3 ts was specifically due to the mutation in the TFB3 gene, we tested the ability of purified holo-TFIIH (1) to complement the mutant extract. Core and holo-TFIIH have been shown to be equally active for NER in vitro (4). As is shown in Fig. 2B, the addition of purified holoTFIIH did indeed restore NER to the mutant extract. We conclude from these experiments that the Tfb3 subunit of yeast TFIIH is required for NER. FIG. 1. A, temperature sensitivity of strain tfb3 ts . Cells expressing either wild-type (TFB3) or tfb3 ts protein were streaked on YPD plates and grown at either 30 or 37°C for approximately three days. B, the tfb3 ts strain exhibits increased sensitivity to UV radiation. Strains were grown in YPD at 30°C and UV irradiated as described under "Experimental Procedures." After irradiation plates were incubated at either 30 or 33°C for 3-5 days until colonies were large enough to be counted. Complementation between NER-defective whole cell extracts from various NER mutant strains has previously been used to gain insights into protein-protein interactions (8). For example, tfb2 and ssl1 extracts fail to complement one another for defective NER activity in vitro, presumably due to the low rate of exchange of these TFIIH subunits (Fig. 3, lanes 1-3). In contrast, the tfb3 ts extract could be complemented by either rad14 or rad23 extracts (Fig. 3, lanes 4 -8), consistent with the notion that these proteins are not tightly associated. These results predict that tfb3 ts can not be complemented by extracts from either ssl1 or tfb2 mutant cells. However, surprisingly, the tfb3 ts extract could be complemented by either ssl1 or tfb2 mutant extracts (Fig. 3, lanes 9 -12). These results suggest that under the conditions of our in vitro NER assay, exchange of Tfb3 can occur.
The tfb3 ts Strain Exhibits Caffeine Sensitivity-In eukaryotic cells, caffeine is known to adversely affect genomic stability, at least in part by inhibiting DNA repair and/or overriding DNA damage checkpoint controls (20,21). In an effort to determine whether UV radiation sensitivity could be increased by inclusion of caffeine in the growth media, we were surprised to observe that tfb3 ts failed to grow in the presence of the drug, whereas the isogenic wild-type strain grew normally (Fig. 4). In contrast to tfb3 ts , an NER-defective C-terminal tfb2 deletion mutant (6) did not exhibit abnormal sensitivity to caffeine (Fig.  4). Thus caffeine sensitivity of tfb3 ts does not appear to be a direct or indirect result of defective NER. It also seems unlikely that caffeine sensitivity is the result of defective transcription of genes required for caffeine detoxification, since there is no evidence of a significant transcription defect in tfb3 ts at the permissive temperature. Consistent with this conclusion, tfb3 ts did not exhibit inositol auxotrophy at 30°C (data not shown), a phenotype commonly associated with transcription-defective mutants (22). Conceivably caffeine sensitivity is related to an as-yet unidentified function of Tfb3 and/or TFIIH.
Two-hybrid Interactions between Ssl2 and Tfb2, Rad3 and Tfb3, and Tfb4 and Ssl1-As mentioned above, a number of studies have revealed pair-wise interactions between yeast TFIIH subunits (see references given in Table II). The recent cloning of genes encoding Tfb2, Tfb3, and Tfb4 has allowed a more comprehensive analysis of interactions between subunits of this yeast transcription factor. Using the two-hybrid assay (23), we qualitatively tested the ability of Tfb2, Tfb3, and Tfb4 to interact with other subunits of TFIIH (Table I) of fusion proteins tested is given in Table I. Of all the interactions and controls tested, three gave significantly higher levels of ␤-galactosidase activity than either of the fusion proteins alone; Tfb3 and Rad3 (Fig. 5A), Ssl2 and Tfb2 (Fig. 5B), Tfb4 and Ssl1 (Fig. 5B). None of the fusion constructs giving these interactions behaved as "promiscuous interactors" (Fig. 5 and Table I). In the latter two cases an interaction was not observed with fusion proteins in the opposite orientation, a result not uncommon with this technique. DISCUSSION In this report we present evidence that Tfb3, like the other subunits of yeast transcription/repair factor TFIIH, is directly involved in NER. We base this conclusion on UV radiation FIG. 6. Proposed structure of yeast core TFIIH (A) and the nucleotide excision repairosome (B). Known pairwise interactions between TFIIH and repairosome subunits upon which these models are based are summarized in Table II.   5. A, two-hybrid interaction of Rad3 and Tfb3. B, two-hybrid interaction of Tfb4 and Ssl1 and SSL2 and TFB2. Patches of yeast transformed with the indicated Gal4 fusion expression plasmids were qualitatively assayed for ␤-galactosidase expression as described under "Experimental Procedures." sensitivity of cells and defective NER in extracts, using a strain harboring a temperature-sensitive allele of TFB3 (tfb3 ts ). We have previously shown that polyclonal antisera against Tfb3 can partially inhibit RNA polymerase II transcription in vitro (6). A different temperature-sensitive allele of TFB3 has been shown to be defective for transcription at the nonpermissive temperature (24). Taken together, these results indicate that the Tfb3 protein is required for both NER and RNA polymerase II transcription.
The moderate level of UV radiation sensitivity of the tfb3 ts mutant does not correlate quantitatively with the severe defect in NER in vitro. Similar results were observed for a TFB2 C-terminal deletion mutant (6). It would appear that in these cases the partial NER defect in vivo is amplified in the in vitro assay. In contrast to our results, another group has reported that yeast cells carrying different mutant alleles of TFB3 (RIG2) are not sensitive to UV radiation, leading to the conclusion that Tfb3 protein is not required for NER (24). However, these mutants were not directly tested for NER activity in vitro. These investigators isolated TFB3 alleles in a screen designed to identify mutations synthetically lethal with a temperature-sensitive allele of KIN28. Perhaps the different way in which our mutant was isolated can explain our apparently contradictory results. Given the relatively tight association of Tfb3 with core TFIIH based on co-purification, we find it implausible that all of the other subunits of core TFIIH play a role in NER but not Tfb3.
Our observation that under the conditions of in vitro NER employed in these experiments some exchange of Tfb3 between TFIIH complexes can occur is intriguing. Under similar conditions, no exchange of Tfb1, Ssl1, or Tfb2 was observed (6,8). However, similar observations have been made with Ssl2 (7), and the selective loss of Ssl2 during purification suggests that Ssl2 is loosely associated with the other coreTFIIH subunits (3). In contrast, there is no evidence from purification of TFIIH that Tfb3 is more loosely associated than other TFIIH subunits. In addition, conditions that promote dissociation of TFIIK and Rad3 from holoTFIIH do not appear to cause significant loss of Tfb3 (4).
Human TFIIH exhibits the opposite behavior. Upon dissociation, MAT1, the ortholog of Tfb3, is identified as a component of CAK, the human counterpart of yeast TFIIK (25,26). We previously attributed this behavior to differences in relative affinities between subunits in the yeast and human TFIIH factors (6). Nonetheless, our observation that Tfb3 can undergo subunit exchange suggests that a form of Tfb3 associated with TFIIK in yeast remains to be identified.
The interactions between Rad3/Tfb3, Ssl2/Tfb2, and Tfb4/ Ssl1 reported here prompted us to propose a more detailed model of the subunit organization of yeast TFIIH and the repairosome (Fig. 6). A summary of the pairwise interactions upon which these models are based is given in Table II. It should be noted that additional interactions not listed in Table  II have been inferred based on extensive co-purification of various proteins. For example, replication protein A has been shown to be comprised of three subunits, Rpa1, Rpa2, and Rpa3 (27), overexpressed Rad2 can be purified as a component of TFIIH (28), and Rad1/Rad10/Rad14 can be isolated as a discrete complex (29) as can Rad7/Rad16/Abf1. 3 Additionally, there are necessarily additional protein-protein interactions that have yet to be identified. For example, Rad4 has been shown to co-immunoprecipitate with core TFIIH (12). However, it is not known through which TFIIH subunit(s) this interaction occurs. By two-hybrid analysis we failed to identify an interaction between Rad4 and any TFIIH subunit (Table I). 4 Additional work will be required to elucidate the remaining determinants of TFIIH/repairosome structure.

TABLE II
Interactions between yeast TFIIH and repairosome subunits Summary of known pairwise interactions between yeast TFIIH and repairosome subunits and the methods used to identify them. TH, two hybrid; CIP, co-immunoprecipitation; CP, co-purification; GST, glutathione S-transferase pull-down.