Affinity purification and partial characterization of a yeast multiprotein complex for nucleotide excision repair using histidine-tagged Rad14 protein.

The nucleotide excision repair (NER) pathway of eukaryotes involves approximately 30 polypeptides. Reconstitution of this pathway with purified components is consistent with the sequential assembly of NER proteins at the DNA lesion. However, recent studies have suggested that NER proteins may be pre-assembled in a high molecular weight complex in the absence of DNA damage. To examine this model further, we have constructed a histidine-tagged version of the yeast DNA damage recognition protein Rad14. Affinity purification of this protein from yeast nuclear extracts resulted in the co-purification of Rad1, Rad7, Rad10, Rad16, Rad23, RPA, RPB1, and TFIIH proteins, whereas none of these proteins bound to the affinity resin in the absence of recombinant Rad14. Furthermore, many of the co-purifying proteins were present in approximately equimolar amounts. Co-elution of these proteins was also observed when the nuclear extract was fractionated by gel filtration, indicating that the NER proteins were associated in a complex with a molecular mass of >1000 kDa prior to affinity chromatography. The affinity purified NER complex catalyzed the incision of UV-irradiated DNA in an ATP-dependent reaction. We conclude that active high molecular weight complexes of NER proteins exist in undamaged yeast cells.

erate sensitivity to killing by ultraviolet (UV) radiation (5,6), suggesting that their polypeptide products may either increase the efficiency of NER or participate in a subpathway(s) of this process.
The characterization of cell-free extracts from mammalian (7) and yeast cells (8) capable of catalyzing repair synthesis of DNA associated with NER in vitro has led to important insights into the molecular mechanism of this process. By using this assay it was demonstrated that Rad3 protein functions in NER as a component of TFIIH, a multiprotein complex that is required for transcription initiation by RNA polymerase II (9). The identification of Rad3 as a subunit of this transcription factor provided an explanation for the essential function of the RAD3 gene (10). Additionally, the requirement for core TFIIH in NER predicted that genes encoding the other six subunits of this complex are also indispensable for this process. This has now been directly demonstrated for all seven proteins using a yeast cell-free system (2, 9 -13). 2 Interestingly, extracts prepared from rad7, rad16, or rad23 strains, which exhibit moderate sensitivity to cell killing by UV radiation (5,6), are also defective in repair synthesis in vitro (14). Recent studies have shown that rad7 and rad16 mutant extracts are proficient in DNA damage-dependent incision both in vitro and in vivo but fail to excise the oligonucleotide fragment containing the DNA lesion (15). Hence, the Rad7 and Rad16 proteins, which interact with each other (16), appear to be required for post-incision events during NER.
Distinct biochemical functions have been assigned to a number of the NER gene products. Rad14 protein recognizes and binds to UV radiation-induced DNA lesions, in particular (6 -4) photoproducts, suggesting a role in the recognition of base damage (17). It appears that the heterotrimeric protein complex Rpa, which is required for DNA replication, is also involved in DNA lesion recognition (18 -20). Rad3 and Ssl2 proteins, both of which are components of core TFIIH, are DNA helicases that appear to be involved in unwinding the DNA duplex in the vicinity of DNA lesions, thereby generating NER "bubbles" that include sites of base damage (12,21,22).
Finally, Rad2 and a heterodimeric complex of Rad1 and Rad10 proteins are DNA structure-specific endonucleases that are similar in that they cleave DNA at duplex/single-strand junctions but differ in terms of the polarity of the single strand that they cleave (23)(24)(25). These junctions are likely generated by TFIIH-mediated "bubble" formation. Rad2 protein cleaves the 5Ј single-strand at the junction and hence operates as a 3Ј endonuclease with respect to sites of base damage. Recipro-cally, Rad1/Rad10 protein cleaves the 3Ј single-strand at the junction and hence operates as a 5Ј endonuclease with respect to sites of base damage (23)(24)(25). The combined action of these two endonucleases is responsible for the dual incision events that are characteristic of NER in all eukaryotes studied and that result in the removal of DNA lesion-containing oligonucleotides 27-30 nucleotides in length (20,26). In addition to the proteins described above, a complex of Rad4 and Rad23 proteins is required for early events during NER (20); however, the precise roles of these proteins is not yet understood.
Recently, a relatively stable high molecular weight complex comprising core TFIIH and multiple NER subunits was identified and extensively purified from cell extracts prepared from yeast cells not exposed to exogenous DNA-damaging agents (11). These observations led to the suggestion that, in the yeast S. cerevisiae and possibly other eukaryotes, a NER complex is assembled in the absence of substrate base damage. In this initial study, the NER complex was called a repairosome although it was not shown to perform any of the reactions involved in NER. Here we define a repairosome as a stable complex of proteins that carry out the DNA damage recognition and incision reactions that are characteristic of NER. Recent studies have also provided evidence suggesting the existence of a high molecular weight complex of NER proteins in extracts from undamaged human cells that catalyze DNA damage-dependent DNA synthesis (27). Although it is plausible that such preformed multiprotein complexes may function to constantly monitor the genome for sites of base damage and catalyze NER in the presence of such damage, there is as yet no direct biochemical evidence for such functions by the repairosome. It has also not been established that yeast cells contain only a single multiprotein complex that comprises both TFIIH and NER polypeptides. Indeed, the primary question as to whether or not any NER complex is pre-assembled in yeast (and other eukaryotes) or is assembled in a step-wise process exclusively when cells are required to carry out NER is controversial (11,28) In initial studies that led to the elaboration of the repairosome hypothesis, extracts containing histidine (his)-tagged Tfb1 protein (one of the subunits of TFIIH) were fractionated by assaying for complementation of defective RNAP II transcription in vitro (11). To further test the hypothesis that a high molecular weight complex of proteins comprising both TFIIH and NER subunits exists in undamaged yeast cells, we have constructed a his-tagged derivative of the DNA damage recognition protein Rad14 (which is not required for RNAP II transcription) and expressed this recombinant protein in a yeast strain deleted of the chromosomal RAD14 gene. Here we present evidence that single step affinity chromatography of recombinant Rad14 protein from nuclear extracts results in the copurification of multiple other NER proteins, including TFIIH subunits, in a complex with a molecular mass Ͼ1000 kDa. Additionally, we show that fractions containing this complex incise UV-irradiated DNA in an ATP-dependent reaction.
Rad1 and Rad10 proteins were purified as described previously (33). RPA was purified from the yeast strain RKY2275 that overexpresses each of the Rpa subunits, as described (34). Purified TFIIH containing his-tagged Tfb1 (35) and RNA polymerase II holoenzyme (36) were provided by Roger Kornberg. Recombinant XPG protein purified from baculovirus-infected insect cells was obtained from Rick Wood, ICRF, United Kingdom (37). Protein concentrations were measured by the Bradford assay (38) using bovine serum albumin as the standard.
Plasmid Constructs-A HindIII fragment from the plasmid pGEM4Z-SP6RAD14, which contains the RAD14 open reading frame (39), was subcloned into pQE-30 (Qiagen) to produce the plasmid pQE-hisRad14, in which a truncated version of Rad14 deleted of the aminoterminal 14 amino acids was expressed as a fusion protein with a poly(his) sequence at the amino terminus. An EcoRI fragment encoding his-tagged Rad14 was subcloned in the yeast plasmid pYES2 (Invitrogen) to produce plasmid pYES2-hisRad14, in which recombinant Rad14 was expressed from a GAL1-inducible promoter.
Cellular Sensitivity to UV Radiation-Plasmids pYES2 and pYES2-hisRad14 were transformed into the yeast strains SX46A (MATa ade2 his3-52 ura3-52 trp1-289) and SX46A ⌬RAD14::HIS3. Cells were grown at 30°C in the appropriate drop-out media with either 2% glucose or 1% galactose, 1% raffinose as the carbon source. When the A 600 of the cultures reached ϳ1.0 cells were collected by centrifugation and serially diluted in sterile water. Aliquots (100 l) of the dilutions were plated onto the appropriate minimal drop-out agar plates containing either glucose or galactose/raffinose as the carbon source. The plates were irradiated with 254 nm UV light (Sylvania Germicidal 15 watts) at a rate of 1.3 J/m 2 /s. Survival was scored after 3-4 days of incubation in the dark at 30°C.
Purification of Recombinant Rad14 Protein-The Escherichia coli strain M15 pREP4 (Qiagen) harboring pQE-hisRad14 was grown at either 30 or 37°C for preparation of soluble and insoluble recombinant Rad14, respectively. Expression of recombinant Rad14 in exponentially growing cultures was induced by the addition of isopropyl thiogalactoside to a final concentration of 2 mM. After further incubation for 3 h, cells were harvested by centrifugation, resuspended in 50 mM sodium phosphate (pH 7.8), 300 mM NaCl containing a mixture of protease inhibitors (33) and lysed by sonication. Recombinant Rad14 was purified from the clarified lysate by metal chelating affinity chromatography using Ni-NTA beads (Qiagen) and then by ion exchange chromatography using FPLC Mono Q and Resource S columns (Amersham Pharmacia Biotech). Approximately 0.4 mg of Rad14 that was Ͼ95% homogeneous was obtained from 1 liter of culture.
The majority of recombinant Rad14, which was present in the insoluble fraction, was purified by immobilized metal affinity chromatography according to the manufacturer's protocol (Qiagen). Recombinant Rad14 was eluted from the Ni-NTA beads with 8 M urea, 100 mM Na 2 H 2 PO 4 , 10 mM Tris-HCl (pH 4.5) and was purified to near homogeneity by electrophoresis through a preparative denaturing polyacrylamide gel (40). The position of Rad14 in the gel was identified by reverse staining with copper chloride using the Pro-Green Staining System according to the manufacturer's protocol (Integrated Separation Systems). Rad 14 was electroeluted from these gel slices using an Electro-Eluter 422 (Bio-Rad) as suggested by the manufacturer. Approximately 1.5 mg of Rad14 was obtained from a 1-liter culture.
Rad14 Antiserum-Mice were initially immunized with recombinant Rad14 purified from the insoluble fraction as described above, plus Freund's complete adjuvant. Subsequently the animals were immunized at 2-week intervals with the same antigen plus Freund's incomplete adjuvant and bled 10 days after each injection.
Preparation of Nuclei-Stationary phase cultures of yeast strains grown in uracil dropout media were diluted 1:20 in 8 liters of 2% Bacto-peptone, 1% yeast extract containing 1% galactose, 1% raffinose as the carbon source. When the cultures reached late logarithmic phase cells were harvested, washed with 0.1 M EDTA (pH 8.0), 1 mM DTT and resuspended in the same buffer at 10 ml/g wet weight of cells. Cells were resuspended in 1% yeast extract, 2% Bacto-peptone, 1 M sorbitol (1 ml/g wet weight of cells) containing yeast lytic enzyme (ICN, 1.4 mg/g wet weight of cells) and incubated with shaking at 30°C. When Ͼ80% of the cells had been converted to spheroplasts digestion was stopped, and spheroplasts were collected as described (8). Spheroplasts were resuspended in 50 mM Tris-HCl (pH 8.0), 20 mM KCl, 2 mM EDTA, 0.125 M spermidine, 0.05 M spermine, 1% thiodiglycol, 18% Ficoll plus a mixture of protease inhibitors (33) (5 ml/g wet weight of cells), and after Dounce homogenization the lysate was cleared by centrifugation at 7000 rpm for 10 min in a Sorvall SS34 rotor. The supernatant was centrifuged twice more at 7000 rpm for 5 min before a final spin at 15,000 rpm for 30 min which pelleted the nuclei. After removal of the supernatant the pellet of nuclei was flash-frozen in liquid nitrogen and stored at Ϫ80°C.
Affinity Purification of Rad14 and Associated Proteins-Nuclei were resuspended in 20 mM Tris acetate (pH 8.0), 300 mM potassium acetate, 20% glycerol, 5 mM ␤-mercaptoethanol and a mixture of protease inhibitors (buffer A), lysed by Dounce homogenization, and centrifuged at 15,000 rpm for 30 min. In some cases, MgCl 2 and DNase I were added to the cleared lysate to final concentrations of 10 mM and 20 g/ml, respectively, prior to incubation at 25°C for 30 min. The cleared lysate (25 mg) was incubated with rotation at 4°C with 2 ml of a 50% slurry of Ni 2ϩ -NTA agarose beads (Qiagen) for 1 h. This mixture was poured into a column, and the beads were washed extensively with buffer A containing 5 mM imidazole. Bound proteins were eluted in a stepwise manner with 20, 50, 100, and 200 mM imidazole in buffer A. Proteincontaining fractions from each eluate were detected using the Bradford assay (38). In some cases, the protein-containing fractions were pooled and concentrated by ultrafiltration using Centricon-10 (Amicon). The 20 mM eluates from the extracts with and without recombinant Rad14 contained 1.2 and 0.4 mg, respectively. The 50 mM eluates from the extracts with and without recombinant Rad14 contained 0.4 and 0.1 mg, respectively. Fractions were flash-frozen in liquid nitrogen and stored at Ϫ80°C.
Immunoblotting-Polypeptides were separated by denaturing gel electrophoresis and transferred electrophoretically to a nitrocellulose membrane. Following incubation with TBST (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.05% Tween 20) containing 5% dried milk for 30 min at room temperature, the membrane was incubated overnight at 4°C with the primary antibody in TBST containing 2% dried milk overnight. Antigen-antibody complexes were detected by enhanced chemiluminescence (Pierce) after incubation for 60 min at room temperature with the appropriate secondary antibody linked to horseradish peroxidase.
In the repairosome purification scheme described above Rad14-associated proteins were detected in both the 20 and 50 mM imidazole eluates from the Ni 2ϩ -NTA agarose beads. To facilitate analysis of the repairosome, the 20 mM elution step was omitted. Rad14 and the putative repairosome subunits were eluted by 50 mM imidazole. To determine the relative amounts of proteins that co-purify with recombinant Rad14, 5 g of the 50 mM imidazole eluate and varying amounts of a purified NER protein were separated by denaturing gel electrophoresis and detected by immunoblotting. The amount of chemiluminescent signal was quantitated by scanning of the x-ray film in a laser densitometer. After arbitrary densitometer units were plotted against protein concentration to construct a standard curve for each purified protein, the amount of that protein in the affinity purified fraction was estimated.
Analytical Gel Filtration-Aliquots (200 l) were applied to an FPLC Superose 6 HR10/30 column (Amersham Pharmacia Biotech) equilibrated with buffer A. Proteins were eluted with buffer A at a flow rate of 0.4 ml/min. Fractions (400 l) were collected, flash-frozen in liquid nitrogen, and stored at Ϫ80°C. NER proteins were detected by immunoblotting. The column was calibrated by determining the elution position of molecular mass standards (thyroglobulin 670,000, gamma globulin 158,000, ovalbumin 44,000, myoglobin 17,000, and vitamin B12 1,350, Bio-Rad) under identical chromatography conditions.
Nuclease Assays with DNA Substrates Containing Duplex/Singlestrand Junctions-Radiolabeled partial duplex DNA molecules with non-complementary 3Ј and 5Ј extensions at the same end (Y structures) were constructed by end-labeling one of the oligonucleotides with T4 polynucleotide kinase prior to annealing as described previously (29). These molecules were purified by electrophoresis through a preparative 15% polyacrylamide gel (29).
Reaction mixtures (20 l, final volume) contained 50 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , 5 mM DTT, labeled DNA substrate with a Y structure (250 pmol), and aliquots of fractions from the affinity column (1 g of protein). In some cases the affinity column fractions were preincubated with either the Rad1 monoclonal antibody, 3E3 (27), or purified mouse IgG for 10 min at 37°C. Reactions were incubated at 37°C for 40 min. Proteinase K (100 g) and SDS (2 l of a 1% solution) were added to deproteinize the reaction mixtures, and incubations were continued at 37°C for 10 min. Samples were electrophoresed through a non-denaturing 15% polyacrylamide gel. After drying, labeled oligonucleotides in the gel were detected by autoradiography.
Endonuclease Assays with UV-irradiated and Undamaged Plasmid DNA Substrates-Supercoiled plasmid DNA was irradiated with UV light at 1.3 J/m 2 /s. Reaction mixtures (15 l) contained 45 mM Hepes-KOH (pH 7.9), 8 mM MgCl 2 , 30 mM potassium acetate, 120 g/ml nuclease-free bovine serum albumin, 1.5 mM DTT, damaged or undamaged supercoiled plasmid DNA, and aliquots of fractions from the affinity columns as indicated. In some cases the affinity column fractions were preincubated with either the Rad1 monoclonal antibody, 3E3 (29), or purified mouse IgG for 10 min at 37°C. When indicated, an ATPregenerating system composed of 2 mM ATP, 30 mM creatine phosphate, and 200 ng of creatine kinase was added. After incubation at 37°C for 30 min proteinase K was added to a final concentration of 200 g/ml and incubation continued for a further 10 min. Samples were electrophoresed through 0.8% agarose gels in 20 mM Tris acetate (pH 7.5), 5 mM EDTA. Plasmid DNA was detected by staining with ethidium bromide (1.5 g/ml in distilled H 2 O). After destaining in 1-2 liters of distilled water gels were photographed through a red filter.

His-tagged Rad14 Protein Complements Both the UV Radiation Sensitivity and Defective NER of a rad14
Deletion Mutant-We constructed a version of the DNA damage recognition protein Rad14 (39) in which the amino-terminal 14 amino acids were replaced by a poly(his) sequence. A plasmid that expresses this version of Rad14 protein from a GAL1 promoter restored the UV radiation sensitivity of a RAD14 deletion mutant to wild type levels under both inducing conditions (galactose medium, Fig. 1A) and non-inducing conditions (glucose medium, Fig. 1B). To monitor the expression of Rad14 protein, we generated polyclonal antibodies to recombinant Rad14 purified from E. coli. In glucose medium, the cellular level of recombinant Rad14 was similar to that of native Rad14 in the wild type strain, whereas in galactose medium the cellular level of recombinant Rad14 was about 100-fold higher (Fig. 2). As expected, no immunoreactivity was detected with similarly prepared nuclear extracts from the ⌬RAD14 strain transfected with the plasmid vector (Fig. 2). Taken together these results demonstrate that the his-tagged version of Rad14 functions in NER in vivo.
Although the majority of recombinant his-tagged Rad14 protein expressed in E. coli is insoluble, we purified recombinant Rad14 from the soluble fraction to Ͼ95% homogeneity by affinity and ion exchange chromatography. Addition of purified recombinant Rad14 protein to cell-free extracts prepared from a ⌬RAD14 mutant restored NER activity as measured by DNA damage-dependent DNA synthesis (data not shown). Thus, his-tagged recombinant Rad14 is active both in vivo and in vitro.
Affinity Purification of Rad14 and Associated Proteins from Yeast Nuclear Extracts-When a nuclear extract from the ⌬RAD14 pYES2-hisRad14 strain was fractionated by immobilized metal affinity column chromatography, recombinant Rad14 was not detected in the flow-through fraction (Fig. 3A). However, substantial amounts of the protein were eluted in the 20 and 50 mM imidazole eluates (Fig. 3A). Smaller quantities were also detected in 100 and 200 mM imidazole eluates (data not shown). No immunoreactivity was detected when the nuclear extract from the ⌬RAD14 pYES2 strain was fractionated by the same procedure (Fig. 3A).
The repairosome complex described previously was isolated by affinity chromatography using his-tagged Tfb1 protein (11). We therefore asked whether Tfb1 co-purified with recombinant his-tagged Rad14. Nuclear extracts prepared from the ⌬RAD14 pYES2 and the ⌬RAD14 pYES2-hisRad14 strains contain equivalent amounts of Tfb1 protein (Fig. 3B). In the absence of recombinant Rad14 protein Tfb1 was detected in the column flow-through fraction but not in the 20 or 50 mM imidazole fractions. In contrast, Tfb1 protein was quantitatively bound by the affinity beads in the presence of recombinant Rad14 and eluted in the 20 and 50 mM imidazole eluates (Fig. 3B). The observations of Rad14-dependent retention of Tfb1 protein by the affinity resin and the co-elution of Rad14 and Tfb1 suggest that these proteins are associated in a complex.
Since TFIIH is required for transcription initiation by RNA polymerase II (RPB1), we asked whether this enzyme co-purified with recombinant Rad14. Some RPB1 was retained by the affinity beads in the presence of recombinant Rad14 and eluted in the 20 mM imidazole eluate (Fig. 3C). Unlike the situation with Tfb1, however, this represented only a small fraction of the RPB1 protein in the nuclear extract. To identify other members of this putative complex, we analyzed the same fractions for other proteins known or believed to be involved in NER. Since Tfb1 is one of seven polypeptides that constitute the core TFIIH subcomplex (2, 10), we anticipated that other subunits of this transcription factor would co-purify with Rad14 protein. In agreement with this notion we observed Rad14-dependent binding of Rad3 protein to the affinity beads (Fig. 3D).
The products of the RAD1 and RAD10 genes are known to form a highly stable heterodimeric complex that is responsible for incisions 5Ј to sites of base damage during NER (24,25). Both Rad1 and Rad10 proteins co-eluted with recombinant his-tagged Rad14 during affinity chromatography (Fig. 3D). Replication protein A (Rpa) is a single-strand DNA-binding protein initially identified as an essential factor for DNA replication (4,41,42). Subsequent studies have shown that this protein is also required for the reconstitution of the DNA damage recognition and incision steps of the eukaryotic NER pathway (1,20). The 34-kDa subunit of Rpa, which is encoded by the RFA2 gene (and presumably the 70-and 11-kDa subunits of Rpa), was also eluted from the affinity beads in the presence of recombinant Rad14 protein (Fig. 3D). The Rad7 and Rad16 proteins also form a stable heterodimeric complex (16). Recent studies have demonstrated that these proteins are required for post-incision events during NER (15). Both proteins co-fractionate with recombinant Rad14 (Fig. 3D), suggesting that they are also components of the repairosome. Rad23 has been purified as a component of a stable Rad23⅐Rad4 complex which is required for the reconstitution of NER with purified yeast proteins (20). Hence, the co-fractionation of Rad23 (and presumably Rad4) with recombinant Rad14 is consistent with the participation of this protein in NER (Fig. 3D).
Co-elution of the proteins discussed above was not observed using extracts lacking Rad14 protein. Thus we consider it unlikely that the co-fractionation of multiple NER proteins is purely fortuitous. It is conceivable, however, that the association of multiple NER proteins is mediated by their independent binding to DNA in nuclear extracts. To address this issue nuclear extracts were treated with DNase I prior to incubation with the affinity beads. This treatment had no effect on the Rad14-dependent binding of the proteins shown in Fig. 3 to the affinity column (data not shown).
It is possible that the co-purification of NER proteins described above is a consequence of Rad14 overexpression. To address this question, we performed the same fractionation procedure with nuclear extracts prepared from cells grown in media with glucose as the carbon source. Under these growth conditions, the cellular level of recombinant Rad14 is similar to ⌬RAD14::HIS3 pYES2-hisRad14 (filled squares) were grown at 30°C in appropriate drop-out media with either 1% galactose, 1% raffinose (A) or 2% glucose (B) as the carbon source. Cells were diluted, plated, and irradiated with UV light as described under "Experimental Procedures." Survival was scored after 3-4 days of incubation in the dark at 30°C.

FIG. 2. Analysis of the levels of recombinant and native Rad14 protein in yeast nuclear extracts by immunoblotting.
Yeast strains were grown in media with either glucose (glu) or galactose/ raffinose (gal) as the carbon source, and nuclear extracts were prepared as described under "Experimental Procedures." Rad14 in the nuclear extracts was detected by immunoblotting. 32 g of SX46A RAD14 (WT) nuclear extract from cells grown in either glucose or galactose medium. 35 and 4 g of SX46A ⌬RAD14::HIS3 pYES2 (⌬RAD14 pYes) nuclear extract from cells grown in glucose and galactose medium, respectively. 37 and 4 g of SX46A ⌬RAD14::HIS3 pYES2 (⌬RAD14 pYes) nuclear extract from cells grown in glucose and galactose medium, respectively. Purified recombinant Rad14 (0.4 g, lane C) was included as a positive control.

FIG. 3. Affinity purification of recombinant Rad14 and associated proteins from yeast nuclear extracts.
The yeast strains SX46A ⌬RAD14::HIS3 pYES2 and SX46A ⌬RAD14::HIS3 pYES2-hisRad14 were grown and nuclei prepared as described under "Experimental Procedures." The nuclei were lysed by Dounce homogenization, and the cleared lysates were fractionated by affinity chromatography as described under "Experimental Procedures." Equal amounts of protein from the two extracts in the cleared lysate (20 g, Load), pass-through fractions (15 g, Pass), 20 mM imidazole eluates (23 g, 20 mM), and 50 mM imidazole eluates (7 g, 50 mM) were separated by denaturing gel electrophoresis and transferred to nitrocellulose membranes. Antigen-antibody complexes were detected by enhanced chemiluminescence. Extracts were prepared from SX46A ⌬RAD14::HIS3 pYES2 (Ϫ) and SX46A ⌬RAD14::HIS3 pYES2-hisRad14, (ϩ) cells. A, immunoblot with Rad14 antiserum. B, immunoblot with Tfb1 antibody. C, immunoblot with RPB1 antibody. D, immunoblots with antibodies specific for the indicated NER proteins. E, immunoblot with PCNA antibody. that of native Rad14 in wild type cells (Fig. 2). In these experiments Rad23, Rad7, Tfb1, RPB1, and Rad10 were retained by the affinity beads in a Rad14-dependent manner (data not shown), indicating that the association of these proteins is not a consequence of the abnormally high levels of recombinant Rad14.
It has been demonstrated that PCNA is required for repair synthesis, a late step in the NER process in eukaryotic cells (43). Remarkably, we did not observe co-fractionation of PCNA protein, either in the presence or absence of recombinant Rad14 (Fig. 3E). This result suggests that the yeast repairosome we have identified does not include components required for repair synthesis of DNA, an event that is in fact not specific for NER. Gratuitously, this result also indicates that the binding of Rad14 to the affinity beads does not facilitate nonspecific binding of other proteins in nuclear extracts. To support this notion further, we examined the binding of Rad51, a relatively abundant nuclear protein that is not involved in NER, to the affinity beads. As expected, we did not detect binding of this protein either in the presence or absence of recombinant Rad14 protein (data not shown).
Stoichiometry of the Putative Repairosome Components and Molecular Mass of the Repairosome Complex-Since NER proteins were consistently detected in both the 20 and 50 mM imidazole eluates, we omitted the 20 mM imidazole elution step in subsequent experiments. When the affinity column was eluted with a single step of 50 mM imidazole, the same NER proteins co-purified (data not shown). In quantitative immunoblotting experiments (Fig. 4) It is prudent to consider the possibility that the association of proteins with recombinant Rad14 is a biochemically specific but biologically irrelevant consequence of the binding of Rad14 to the affinity beads. To determine whether NER proteins are associated in the absence of affinity chromatography, we applied nuclear extracts directly to an FPLC Superose 6 gel filtration column. In Fig. 5 we show the elution profiles of Rad14 and several NER proteins that co-purified with recombinant Rad14 on the affinity column. In these experiments extracts were treated with DNase I prior to filtration. Similar results were obtained with extracts that were not treated with DNase I (data not shown). Recombinant Rad14 and other NER proteins co-eluted in fractions corresponding to a molecular mass of 1000 -1500 kDa, indicating that they are associated in a high molecular weight complex in the nuclear extract prior to the binding of recombinant Rad14 to the affinity beads.
DNA Structure-specific Endonuclease Activity Associated with the Repairosome-The results presented thus far support the notion that NER proteins exist in a high molecular weight complex in yeast cells not exposed to exogenous DNA damaging agents. For this complex to be designated as a functional repairosome, it should support some, if not all, of the reactions that transpire during the DNA damage recognition and incision steps of NER in vivo. We examined the ability of affinity purified Rad14 and associated proteins to specifically cleave Y-shaped DNA substrates that contain duplex/single-strand junctions known to be cleaved by the junction-specific Rad1/ Rad10 and Rad2 endonucleases (23)(24)(25). Fractions containing FIG. 4. Quantitation of the proteins that co-purify with recombinant Rad14. Proteins (5 g, S) that co-purified with recombinant Rad14 when the metal chelating column was eluted with single step of 50 mM imidazole and varying amounts of the indicated purified protein were separated by denaturing gel electrophoresis and transferred to nitrocellulose membranes. Antigen-antibody complexes were detected by immunoblotting with the indicated antibody and quantitated as described under "Experimental Procedures."

FIG. 5. Fractionation of yeast NER proteins in a nuclear extract by gel filtration.
An aliquot (200 l) of a nuclear extract from the yeast strain SX46A ⌬RAD14::HIS3 pYES2-hisRad14 was fractionated by gel filtration as described under "Experimental Procedures." Fractions (FXN) (400 l) were collected, flash-frozen in liquid nitrogen, and stored at Ϫ80°C. NER proteins were detected by immunoblotting. The column was calibrated by determining the elution position of molecular mass standards (Bio-Rad) under identical chromatography conditions. The void volume of this column corresponded to fraction 5. The largest molecular mass standard (thyroglobulin, molecular mass 670 kDa) eluted in fractions 18 -20. Rad14 and associated NER proteins cleaved a Y-shaped substrate radiolabeled at the duplex end to produce a partial linear duplex product (Fig. 6A). In contrast, equal amounts of protein from a comparable fraction purified from a ⌬RAD14 pYES2 nuclear extract did not contain junction-specific cleavage activity, although low levels of nonspecific nuclease activity were observed (Fig. 6A). Preincubation of the repairosome fraction with 6 g of Rad1 monoclonal antibody 3E3, which inhibits the Rad1/Rad10 endonuclease (29), reduced cleavage of the Y structure by ϳ80%, whereas purified mouse IgG had no inhibitory effect (data not shown). These results indicate that the Rad1 and Rad10 proteins that co-purify with recombinant Rad14 are catalytically active.
Cleavage of the radiolabeled Y-shaped substrate described above by either Rad1/Rad10 or Rad2 endonuclease generates products that are indistinguishable by gel electrophoresis. However, if the non-complementary 5Ј single-strand of the Y structure is end-labeled to produce the substrate shown in Fig.  5B, cleavage by the Rad2 endonuclease will produce a unique end-labeled single-strand oligonucleotide ϳ20 residues in length (23). In assays with such a substrate, fractions containing Rad14 and associated NER proteins generated both the partial linear duplex expected from cleavage by the Rad1/ Rad10 endonuclease and a novel product with increased electrophoretic mobility (Fig. 6B). This product was not observed in assays with a comparable fraction purified from a ⌬RAD14 pYES2 nuclear extract. However, it was detected in reactions with XPG protein, the human homolog of Rad2 protein (23,25,37,44) (Fig. 6B). Thus, we are led to the conclusion that the affinity purified fractions containing recombinant Rad14 possess Rad2 activity. Antibodies to Rad2 protein were not available to demonstrate specific inhibition of this activity. However, preincubation with 6 g of the Rad1 monoclonal antibody, 3E3, resulted in significant inhibition (ϳ80%) of the putative Rad2 activity (data not shown). Since the Rad1 antibody is known to inhibit the NER reaction catalyzed by yeast extracts (29), it is possible that binding of the antibody to the Rad1 component of the repairosome precludes the activity of the Rad2 endonuclease.
Nicking of UV-irradiated Supercoiled DNA by the Repairosome-In reconstitution experiments with purified NER proteins, specific nicking of UV-irradiated DNA is ATP-dependent (20). We asked whether the fraction from the affinity column that contains recombinant Rad14 and other NER proteins could perform the same reaction. ATP-dependent nicking of UV-irradiated plasmid DNA was indeed observed in assays with this fraction, whereas no nicking of undamaged or damaged plasmid DNA occurred in the absence of ATP (Fig. 7A,  lane 6). Similarly, no nicking was observed with an equivalent amount of protein from a comparable fraction purified from the nuclear extract lacking recombinant Rad14. Fractions containing the repairosome supported some nicking of undamaged plasmid DNA in the presence of ATP (Fig. 7A, lane 5) but to a significantly lower extent than that observed with UV-irradiated DNA.
Since the Rad1 monoclonal antibody 3E3 inhibits both the cleavage of Y structures by the Rad1/Rad10 endonuclease and in vitro NER by yeast cell-free extracts (29), we examined the effect of this antibody on the ATP-dependent nicking of UVirradiated plasmid DNA. Preincubation of an affinity-purified fraction containing Rad14 and associated NER proteins with antibody 3E3 resulted in Ͼ50% inhibition of endonuclease activity (Fig. 7B, lane 5). In contrast, preincubation of the same fraction with mouse IgG had no such effect (Fig. 7B, lane 6).
In the experiment shown in Fig. 8, we employed reaction conditions essentially identical to those described for the reconstituted NER reaction that includes the Rad7⅐Rad16 complex (16). UV-irradiated plasmid DNA containing approximately one UV photoproduct/molecule was rapidly cleaved such that ϳ50% of the substrate was converted to the nicked circular form in 2 min in the presence of 100 fmol of repairosome (assuming a molecular weight of 1.6 ϫ 10 6 from all the identified subunits) (Fig. 8, A and B). The rate of endonucleolytic cleavage was estimated to be 0.2 nicks/min/repairosome. After 30 min essentially all the supercoiled plasmid DNA was nicked (Fig. 8, A and B). Undamaged plasmid molecules were nicked at a similar initial rate (Fig. 8B). However, the reaction reached a plateau when ϳ30% of the substrate molecules were nicked (Fig. 8B). This reaction profile could be explained if a proportion of the putatively undamaged plasmid molecules contain some sort of DNA lesion that is recognized by the repairosome. In the NER reaction reconstituted from purified subcomplexes (16), the rate of cleavage of the UV-irradiated substrate was about 0.006 nicks/min/repairosome subcomplex. After subtracting the nicking of undamaged DNA, it appears that the affinity-purified repairosome is at least 10-fold more efficient at cleaving the UV-irradiated DNA substrate than the combination of purified NER subcomplexes (16). This may re- flect the higher specific activity of the preformed repairosome complex and/or the presence of additional factors in the affinity-purified fraction that enhance the rate of DNA damagedependent incision. Alternatively, it is possible that one or more of the subcomplexes may not have been purified in a fully active form, reducing the efficiency of the reconstituted reaction.

DISCUSSION
During purification of TFIIH containing a his-tagged Tfb1 subunit, it was observed that this RNA polymerase II basal transcription factor co-purified with NER proteins not involved in RNAP II transcription (11). Based on these results it was suggested that a large multiprotein complex dedicated to NER (nucleotide excision repairosome) exists in yeast cells, the assembly of which does not require prior exposure of cells to DNA damaging agents. Subsequent independent studies, which employed different column resins and different elution protocols during the initial purification steps, led to the conclusion that NER is mediated by the sequential assembly of subcomplexes at sites of base damage (28). If NER proteins are indeed associated in a large multiprotein complex prior to the time that cells sustain bulky base damage of the type that is a substrate for NER, it should be possible to identify such a complex by affinity purification regardless of which NER protein is tagged with poly-histidine residues, with the obvious caveat that the tagged protein is accessible for binding to appropriate affinity matrices.
By using a his-tagged version of the DNA damage recogni-tion protein Rad14, we have partially purified a multiprotein complex by a single affinity chromatography step. The NER proteins Rad1, Rad10, Rad7, Rad16, Rad23, the 34-kDa subunit of Rpa, the catalytic subunit of RNA polymerase II holoenzyme, RPB1, and at least two components of TFIIH bound to the affinity column exclusively in the presence of recombinant Rad14. These results confirm and extend previous observations and are consistent with other studies describing a high molecular weight complex including both TFIIH and NER proteins in yeast and human cell extracts (11,27). In contrast to this study, the Rad7 and Rad16 proteins were not detected in the repairosome that was affinity purified using his-tagged Tfb1 (11,13). This is presumably a consequence of the different purification schemes that were used to isolate the repairosome complexes.
It is formally possible that our results reflect the fortuitous independent binding to DNA of multiple proteins involved in NER. We consider this unlikely since treatment of the nuclear extract with DNase did not affect the behavior of these proteins in subsequent fractionation steps of either gel filtration or affinity chromatography. We also consider it unlikely that the binding of overexpressed recombinant Rad14 to the affinity column specifically promotes the association of NER (and possibly other) proteins in the nuclear extract for the following reasons: (i) when nuclear extracts were applied directly to a gel filtration column many of the NER proteins, which co-purified with recombinant Rad14 by affinity chromatography, co-eluted with Rad14 in fractions that correspond to a molecular mass of Ͼ1000 kDa; (ii) although the NER proteins that co-purify with recombinant Rad14 were present in approximately equimolar amounts, the relatively abundant nuclear proteins PCNA and Rad51, which are not involved in the DNA damage recognition and incision steps of NER (events that are specific to NER), were not retained by the affinity column in either the presence or absence of recombinant Rad14; (iii) a similar spectrum of NER proteins was retained by the affinity column in a Rad14dependent manner when recombinant Rad14 was expressed at the same level as the native protein.
A high molecular weight complex of NER proteins was initially detected after affinity purification of a his-tagged version of Tfb1 expressed at wild type levels (11). The putative repairosome containing his-tagged Tfb1 complemented the repair defect in extracts from several yeast strains mutated in NER genes (11). However, this complex was not investigated for its ability to directly support reactions involved in NER. Here we have demonstrated that the multiprotein complex partially purified by affinity chromatography of his-tagged Rad14 protein incises UV-irradiated DNA in an ATP-dependent reaction. Hence, we conclude that the high molecular complex of NER proteins constitutes a functional repairosome.
Unlike the complex of human proteins purified by XPA affinity chromatography (27), we failed to detect co-purification of PCNA with NER proteins. Thus, it is unlikely that the affinity purified complex of yeast NER proteins can support DNA damage-dependent DNA synthesis. In fact the yeast NER complex is similar to another high molecular weight transcription-repair complex isolated from human cells in that it contains the catalytic subunit of RNA polymerase II (45).
Although our results provide support for the notion that repairosomes exist in undamaged yeast cells, it is not known what proportion of NER proteins are present in such complexes. Nor can we eliminate the possibility that affinity chromatography identifies several independent complexes containing NER proteins, some of which contain additional polypeptides dedicated to other cellular responses to DNA damage. However, the relative ease of genetic manipulation in FIG. 7. Nicking of UV-irradiated and undamaged supercoiled DNA by the repairosome; effect of ATP. A, supercoiled plasmid DNA was irradiated with 300 J/m 2 of UV light from a germicidal lamp. Reactions (15 l) contained 500 ng of damaged (ϩ) or undamaged (Ϫ) supercoiled plasmid DNA and aliquots (1 g) of affinity purified fractions from the SX46A ⌬RAD14::HIS3 pYES2 (50Ϫ) and SX46A ⌬RAD14::HIS3 pYES2-hisRad14 (50ϩ) extracts. An ATP-regenerating system (ATP) was added as indicated. Incubations were performed as described under "Experimental Procedures." B, affinity purified fractions (50ϩ) were preincubated with the Rad1 monoclonal antibody 3E3 or purified mouse IgG as indicated, prior to the addition of the other reaction components including UV-irradiated DNA and the ATP-regenerating system. Similar assays were performed without addition of the affinity purified fraction (Ϫ) to demonstrate the absence of endonuclease activity in the antibody preparations. After electrophoresis through a 0.8% agarose gel plasmid DNA was detected by staining with ethidium bromide. The positions of supercoiled DNA (ccc) and nicked circular DNA (nc) are shown on the right. yeast make this an attractive organism to identify proteins that play important roles in the assembly and/or stability of multiprotein complexes. Finally, these results do not exclude the possibility that some, if not most, DNA lesions are in fact removed by the sequential assembly of NER subcomplexes at or near the site of DNA damage in vivo.
It seems reasonable to speculate that a preformed repairosome is dedicated to continual surveillance of the genome in eukaryotic cells for sites of base damage normally processed by NER. We suggest that when (and only when) such sites are encountered, the multiprotein complex undergoes specific conformational changes that strengthen the protein-DNA interactions. Additionally, these and/or further conformational changes may both facilitate localized unwinding of the DNA that flanks sites of base damage by the Rad3 and Ssl2 DNA helicases and position the Rad2 and Rad1/Rad10 endonucleases at the duplex/single-strand junctions generated by this unwinding, specifically on the damaged DNA strand.
Several observations suggest that the amount and/or composition of the nucleotide excision repairosome may change when yeast cells are exposed to UV radiation or other DNA-damaging agents that generate substrates for NER. For example, genes that encode several subunits of the repairosome are up-regulated under these conditions. Notable examples include RAD2 (46), RAD7 (47), RAD16 (48), and RAD23 (49). Additionally, recent studies in one of our laboratories (E. C. Friedberg) have demonstrated significant inhibition of RNAP II transcription in vitro in the presence of active NER (50). This inhibition can be relieved by supplementing in vitro reactions with purified holo-TFIIH, a form of TFIIH specifically required for transcription initiation. Hence, when cells are required to carry out NER, subunits of the repairosome such as TFIIH and Rpa which are indispensable for transcription and replication, respectively, may be preferentially sequestered by the repairosome. This would not only enhance the cellular repair capacity but would also inhibit transcription and replication until the genome is repaired.