The Yeast RAD7 and RAD16 Genes Are Required for Postincision Events during Nucleotide Excision Repair

In eukaryotes, nucleotide excision repair (NER) is a complex reaction requiring multiple proteins. In the yeastSaccharomyces cerevisiae, two of these proteins, Rad7 and Rad16, are specifically involved in the removal of lesions from transcriptionally silent regions of the genome in vivo. Extracts prepared from rad7 or rad16 mutant cells are deficient, but not totally defective, in both oligonucleotide excision and repair synthesis of damaged plasmid DNA. We show that these extracts are, however, fully proficient in the incision step of the NER reaction in vitro. Furthermore, using acdc9 mutant to trap incision intermediates, we demonstrate that rad7 and rad16 mutants are proficient in NER-dependent DNA incision in vivo. A purified protein complex containing both Rad7 and Rad16 proteins complements the oligonucleotide excision and repair synthesis defects inrad7 and rad16 mutant extracts. We conclude that the products of the RAD7 and RAD16 genes are involved in a postincision event(s) during NER in yeast.

In eukaryotes, nucleotide excision repair (NER) is a complex reaction requiring multiple proteins. In the yeast Saccharomyces cerevisiae, two of these proteins, Rad7 and Rad16, are specifically involved in the removal of lesions from transcriptionally silent regions of the genome in vivo. Extracts prepared from rad7 or rad16 mutant cells are deficient, but not totally defective, in both oligonucleotide excision and repair synthesis of damaged plasmid DNA. We show that these extracts are, however, fully proficient in the incision step of the NER reaction in vitro. Furthermore, using a cdc9 mutant to trap incision intermediates, we demonstrate that rad7 and rad16 mutants are proficient in NER-dependent DNA incision in vivo. A purified protein complex containing both Rad7 and Rad16 proteins complements the oligonucleotide excision and repair synthesis defects in rad7 and rad16 mutant extracts. We conclude that the products of the RAD7 and RAD16 genes are involved in a postincision event(s) during NER in yeast.
The yeast RAD7 and RAD16 genes belong to the RAD3 epistasis group of DNA damage-responsive genes (1,2). This epistasis group includes genes required for nucleotide excision repair (NER) 1 of DNA, a process by which multiple types of base damage are excised from the genome as oligonucleotide fragments, and the resulting single strand gaps are repaired by DNA synthesis (repair synthesis) and ligation (3). In contrast to the extreme sensitivity to UV light (and other DNA-damaging agents) conferred by complete inactivation of NER genes, such as RAD1, RAD2, RAD3, RAD4, RAD10, and RAD14, deletion of the RAD7 or RAD16 gene confers partial UV radiation sensitivity (1,2).
The precise role(s) of the RAD7 and RAD16 gene products in NER is not clear. The proteins are required for NER of transcriptionally repressed loci such as HML␣ and HMRa in vivo (4), and for the nontranscribed (coding) strand of transcriptionally active genes. However, unlike the other RAD genes mentioned above, the RAD7 and RAD16 genes are not required for NER of the transcribed (template) strand of such genes (5). The Rad7 and Rad16 proteins have been shown to stably interact both in vivo and in vitro (6,9). Rad16 is a member of the SWI2/SNF2 superfamily of proteins, several of which have been shown to be ATPases involved in chromatin remodeling (7). This observation, coupled with the requirement for the RAD7 and RAD16 genes for NER of transcriptionally repressed regions of the genome, has led to the notion that the Rad7 and Rad16 proteins may be subunits of a complex dedicated to the perturbation of chromatin structure in order to facilitate NER of transcriptionally silent regions of the genome and the nontranscribed strand of transcriptionally active genes (5).
It has been reported that a reconstituted in vitro system that supports damage-specific incision and oligonucleotide excision from purified plasmid DNA does not require the Rad7 and Rad16 proteins (8). Subsequent studies (9) reported that these proteins actually increase the efficiency of NER in vitro and offered the suggestion that the specific role of these proteins is to facilitate the recognition of damaged bases in nontranscribed regions of the genome in vivo (10). Contrary to the observation that the Rad7 and Rad16 proteins are not required for NER of plasmid DNA in a reconstituted in vitro system, we previously reported that they are indispensable for NER of plasmid DNA in yeast whole cell extracts (WCEs), using an assay that measures repair synthesis of DNA (6).
In view of the multiple and in some instances discordant roles proposed for the RAD7 and RAD16 products during NER of DNA, we have studied their requirement for specific events in this process in greater detail. We have investigated the possibility that when purified plasmid DNA is incubated with yeast WCEs, the DNA is assembled into nucleosomes. Such a result could explain the apparent contradiction between the requirement for Rad7 and Rad16 proteins for NER in WCEs and their dispensability in a reconstituted system using purified proteins, and it would also be consistent with the suggestion that the Rad7 and Rad16 proteins are required for the processing of nucleosome structures during NER. Additionally, we have utilized both in vitro and in vivo assays that allowed us to independently monitor the steps of damage-specific incision, oligonucleotide excision, and repair synthesis in wild-type and rad7 and rad16 mutants.
We have found no evidence for nucleosome formation in yeast WCEs that support robust NER of plasmid DNA. We show here that rad7 and rad16 mutants are fully proficient in damage-specific incision of DNA. In contrast, the postincision events of oligonucleotide excision and, as previously reported, repair synthesis (6) are severely deficient in rad7 and rad16 mutant extracts. Finally, we have purified a protein complex that includes Rad7 and Rad16 proteins, without overexpression of these proteins in yeast cells. This complex corrects both defective oligonucleotide excision and repair synthesis in vitro.
We conclude that the RAD7 and RAD16 gene products participate in postincision events during NER of transcriptionally silent DNA. The utility of such participation may be obviated in transcriptionally active DNA, perhaps as a direct consequence of transcription itself.
In Vitro NER-Yeast WCEs capable of supporting NER were prepared as described previously (11). NER was monitored by measuring DNA repair synthesis in pUC18 containing N-acetyl-2-aminofluorene (AAF) adducts. AAF-modified DNA was prepared by treating pUC18 with AAF and purifying the DNA on a 5-20% sucrose gradient as described (12). Plasmid DNA was recovered from extracts by restricting with HindIII enzyme, which linearizes the substrate and reveals a discrete band in a 1% agarose gel. To detect damage-specific incision, plasmid DNA was recovered from WCEs and electrophoresed through agarose without prior HindIII restriction. Undamaged plasmid DNA was analyzed similarly in independent incubations. In the standard NER assay, which measures repair synthesis, an undamaged DNA template is used as an internal control. Background incorporation of radioactivity (typically ϳ5% of that in the damaged plasmid) is subtracted from that incorporated in the damaged template. The DNA incision assay precludes the use of a quantitative internal control because a single nick changes the topology of the form I DNA. DNA was visualized in gels under UV light following staining with ethidium bromide. Repair synthesis was detected by autoradiography of dried agarose gels and quantitated by phosphorimaging as described (6).
Excision Repair Assay-An oligonucleotide of known sequence (5Ј-TATGATTGGGGATAAGAT*TG-3Ј) containing a single cyclobutane pyrimidine dimer near the 3Ј-end was kindly prepared by Dr. John-Stephen Taylor (Washington University, St. Louis, MO) (13,14). Plasmid substrates containing these oligonucleotides were constructed as described (15,16). Yeast WCEs were prepared and repair reactions were performed as described (17). Following the reactions, radiolabeled excision products were resolved on 10% polyacrylamide sequencing gels and analyzed by autoradiography.
In Vivo Incision of DNA at the HML␣ and MAT␣ Loci-Yeast strains were grown in YPD medium (yeast/peptone/dextrose) overnight at 23°C (permissive temperature for the cdc9 mutation). All strains were cdc9 ts and were either wild-type for NER (RAD ϩ ) or were rad2, rad7, or rad16 mutants. Cells were grown to mid-log phase, harvested by centrifugation, and resuspended to a density of 2 ϫ 10 7 cells/ml. Aliquots were incubated for 30 min at either 23 or 37°C (nonpermissive temperature for cdc9). Cell suspensions were placed in a Pyrex dish (14 ϫ 1 cm) to a depth of ϳ0.3 cm to ensure uniform radiation. Dishes were shaken while cell suspensions were irradiated at 254 nm (70 J/m 2 ) at a dose rate of 1 J/m 2 /s, using a germicidal lamp. Following irradiation, cells were pooled and kept at 23 or 37°C for 20 min before harvesting. Unirradiated control samples were harvested for each strain. Cells were lysed with zymolyase 1000T as described (18), and DNA was isolated as described (19). Purified DNA was resuspended in 10 mM Tris-HCl (pH 7.5), 1 mM EDTA and stored at 4°C. All procedures were carried out under safety lights to prevent enzymatic photoreactivation of pyrimidine dimers. Analysis of DNA incision at the MAT␣ and HML␣ loci was performed by Southern blot analysis as described (20). MAT␣ and HML␣ fragments were generated by digesting genomic DNA with HaeII enzyme, which yields a 3.6-kb MAT␣ fragment and a 4.0-kb HML␣ fragment. The probe was an internal fragment of the Y␣ region of MAT␣, which is strictly ␣-specific (see below) (21). DNA (5 g) was restricted with HaeII and incubated at 37°C for 2 h. The DNA was recovered after phenol-chloroform extraction and resuspended in 10 l of denaturing loading buffer (12.5% Ficoll, 5 mM EDTA, 0.125% bromphenol blue, 0.25 M NaOH). Aliquots from each HaeII restriction digest (with appropriate controls) were quantified by spectroscopy to ensure equal loading onto 1% alkaline agarose gels. Electrophoresis was performed overnight at 1.5 V/cm under denaturing conditions. DNA was vacuum-blotted to Genescreen plus (NEN) transfer membranes. The membranes were prehybridized for 1 h and then hybridized with a specific radiolabeled probe prepared by random primer labeling. MAT␣ and HML␣ bands were visualized by autoradiography.
Construction of His-tagged RAD7 Strain-A 6-His-tagged RAD7 strain was constructed with the RAD7 gene expressed from its own promoter on a centromeric plasmid. This strain was derived from strain YPH/TFB1:6-His (22), in which the deleted chromosomal TFB1 gene was replaced with the LEU2 marker. The strain carries a 6-His-tagged TFB1 gene driven by its own promoter on a centromeric vector. The chromosomal RAD7 gene was deleted using the one-step disruption technique (23). Strain YPH/TFB1:6-His was transformed with EcoRIdigested prad7::HIS. This plasmid was constructed by digesting prad7::LEU2 (5) with KpnI and HindIII, blunting the ends by filling in, and replacing the LEU2 gene with a 1.8-kb BamHI fragment containing HIS3. Generation of the strain YPH/TFB1:6-His, rad7 was confirmed by Southern analysis. Additionally, survival studies revealed a level of UV radiation sensitivity consistent with that expected for rad7 mutants. The centromeric plasmid pRS314, containing an untagged copy of TFB1 and the TRP1 marker (provided by Dr. W. John Feaver), was used to clone a 6-His-tagged RAD7 gene. This construct was introduced into strain YPH/TFB1:6-His, rad7. The 6-His-tagged RAD7 insert was generated by PCR. In addition to the RAD7 coding sequence, the insert includes the promoter region for RAD7. The 6-His tag was generated at the C terminus of the protein by including codons for histidine in the downstream primer immediately prior to the stop codon. Both primers included BamHI sites at their 5Ј-ends to facilitate cloning of the PCRgenerated product into the BamHI site of the pRS314-TFB1 vector. The sequence of the primers was as follows: upstream primer, 5ЈATATGG-ATCCTTTCTCCCCGACGATATAGGAATC 3Ј; downstream primer, 5ЈATATGGATCCTTAGTGATGGTGATGGTGATGTATACTGTCACTC-TGTCTCCCTAT3Ј. The 2.2-kb PCR product was cut with BamHI and cloned into the BamHI site of pRS314-TFB1-TRP1. Following cloning, the insert was sequenced to confirm the RAD7 sequence and the inframe addition of the 6-His-tag. Transformation of strain YPH/TFB1: 6-His, rad7 with pRS314-TFB1, RAD7:6-His resulted in complementation of UV radiation sensitivity. An untagged RAD7 strain was constructed in the same fashion. During final transformations, plasmid shuffling was employed to remove the 6-His-tagged TFB1:URA3 plasmid.
Production of RAD7 and RAD16 Antibodies-The RAD7 gene was isolated from pAS1-CYH2RAD7 and cloned into the NcoI/SalI site of pGEX-KG, creating an in-frame RAD7-glutathione S-transferase fusion gene (6). A 0.6-kb internal fragment of RAD16 that encodes 192 amino acids (from amino acids 475-667) was generated by PCR using the following primers: upstream primer, 5ЈCATGCCATGGATCTTTACA-GAAGT3Ј; downstream primer, 5ЈCCTGAAGCTTCACTGTTTGAA3Ј. The upstream primer included an NcoI site, whereas the downstream primer included a HindIII site to facilitate cloning of the PCR product in-frame into the pGEX-KG vector and allow expression of glutathione S-transferase-RAD16 fusion protein. E. coli DH5␣ cells were transfected with each of the constructs, and induction of the fusion genes was achieved by addition of 100 M isopropyl ␤-D-thiogalactopyranoside. Both fusion proteins were insoluble in E. coli. We therefore purified fusion proteins from inclusion bodies by affinity chromatography on glutathione agarose as described (24). After immobilizing fusion proteins on glutathione-agarose beads, we cleaved at the glycine linker regions between the Rad and the glutathione S-transferase moieties with thrombin (24) to isolate each Rad protein. Purified full-length Rad7 and truncated Rad16 proteins were used as antigens to generate polyclonal antisera in rabbits (Babco). Rad7 antibodies were purified from rabbit sera using antigens cross-linked to cyanogen bromideactivated sepharose-4B (Amersham Pharmacia Biotech).
Purification of a Protein Complex Containing Rad7 and Rad16 Proteins-A 10-liter culture of strain YPH/RAD7:6-His was grown in YPD medium until the cell density reached 10 8 cells/ml. Cell paste (ϳ100 g) was recovered following centrifugation. Cells were washed at 4°C, and all manipulations from this point were at 4°C. Cells were resuspended in 200 ml of extraction Buffer A (40 mM Hepes, pH 7.3, 350 mM NaCl, 0.1% Tween 20, 10% glycerol, pepstatin A (2 g/ml), leupeptin (2 g/ml), 1 mM phenylmethylsulfonyl fluoride, 0.097 units of trypsin inhibitor/ml of aprotinin). The cell suspension (300 ml) was disrupted in a Biospec bead beater apparatus using Zirconium beads (0.5 mm). Cells were disrupted with 20 cycles of agitation and intermittent cooling. After washing the beads, the extract (300 ml) was clarified by centrifugation at 100,000 ϫ g for 80 min and mixed with 10 ml of Sepharose CL-6B (Amersham Pharmacia Biotech) equilibrated with Buffer A. The mixture was rocked for 15 min and then centrifuged twice at 1000 ϫ g for 10 min to remove the resin. Unbound protein was added batchwise to 10 ml Ni 2ϩ -nitriloacetic acid agarose (Qiagen, Chatsworth, CA) for 3 h at 4°C. The suspension was transferred to a glass column (1.5 ϫ 20 cm). The flow-through was collected, and the column was washed with 50 ml of Buffer A supplemented with 1 mM ␤-mercaptoethanol (final concentration) and 10 mM imidazole. No Rad7 protein was detected by immunoblotting in either the flow-through fractions or after washing. Histagged Rad7 protein was eluted from the column with Buffer A plus 1 mM ␤-mercaptoethanol and 25, 50, or 100 mM imidazole. The peak of both Rad7 and Rad16 proteins eluted in 25-50 mM imidazole (see Fig.  2). The strain containing the untagged RAD7 gene was subjected to the same procedure in parallel. No binding of Rad7 ( Fig. 1) or Rad16 (data not shown) was observed in the nontagged strain. Fractions 1-6 from Ni 2ϩ -nitriloacetic acid were pooled (21 ml), dialyzed against Buffer A with 100 mM NaCl, 1 mM dithiothreitol for 2 h, and applied to a column of phosphocellulose (P11, Whatman, 1.5 ϫ 8 cm; 14-ml matrix). The column was washed and eluted with a 60-ml salt gradient (0.1-1.0 M NaCl) in Buffer A. Rad7 and Rad16 proteins co-eluted in fractions 7-10 at 480 mM NaCl (Fig. 2). Pooled P11 fractions (8 ml) were dialyzed against Buffer A with 0.1 M NaCl, 1 mM dithiothreitol and fractionated on DEAE-Sephacel (Amersham Pharmacia Biotech) (1.5 ϫ 8 cm; 14-ml matrix). Fractions were collected following elution with a 60-ml salt gradient from 0.1-1.0 M NaCl. Immunoblotting revealed co-fractionation of Rad7 and Rad16 proteins (Fig. 2). Peak fractions were pooled and dialyzed and further fractionated by fast protein liquid chromatography on a Mono-Q HR5/5 column using a 10-ml salt gradient (0.1-1.0 M NaCl). Once again, Rad7 and Rad16 proteins co-eluted (Fig. 2). Mono-Q fractions were pooled (1.5 ml) and concentrated in a Centricon-10 microconcentrator (Amicon) to a volume of 200 l. This was layered on a glycerol gradient (10 -30% glycerol). The gradient was centrifuged at 39,000 rpm for 16 h in a SW41 rotor. Fractions were collected, and Rad7 and Rad16 proteins were shown to co-elute (Fig. 2).

Analysis of Nucleosome Formation during Incubation of
Plasmid DNA in NER-proficient Extracts-In view of the notion that the RAD7 and RAD16 gene products may be required for the processing of chromatin during NER (4), we considered the possibility that chromatin or some other higher ordered DNA structure is generated during incubation of WCEs with purified plasmid DNA. We harvested plasmid DNA following incubation with yeast WCEs known to support robust NER and digested it with micrococcal nuclease. This enzyme cleaves nucleosomal DNA in linker regions, yielding a characteristic ladder of oligonucleosome fragments (25). In control experiments, we demonstrated efficient nucleosome formation following incubation of damaged plasmid DNA with Xenopus oocyte extracts (data not shown). However, we failed to observe a nucleosomal ladder in plasmid DNA incubated in yeast WCEs following mild micrococcal nuclease digestion conditions (data not shown). Hence, there is no evidence of nucleosome formation on the input plasmid DNA in these yeast WCEs.
DNA Incision in RAD ϩ and rad Mutant Extracts-In an effort to identify a specific molecular defect in rad7 and rad16 mutants, we next examined individual steps in the NER reaction in yeast WCEs. We analyzed damage-specific incision of plasmid DNA in RAD ϩ and in rad7, rad16, rad14, rad4, and ssl2 mutant extracts by monitoring the conversion of AAFdamaged form I DNA to the relaxed state. Extracts of the latter three mutants are known to be totally defective in damagespecific incision of DNA in vitro (8). When covalently closed circular (form I) AAF-treated plasmid DNA (Fig. 3A, lane 1) was incubated with 250 g of WCEs prepared from RAD ϩ strains, the DNA was converted to the relaxed state (form II) within 2 min of incubation (Fig. 3A, lanes 2-5), indicating incision of the DNA. When these experiments were repeated using a 10-fold reduced concentration of extract (25 g of protein), subsaturating conditions were achieved, and only partial conversion of form I to form II DNA was observed at 2 min (Fig.  3B). As anticipated, when damaged plasmid DNA was incubated with extracts from the incision-defective rad14, rad4 or ssl2 mutants, the DNA remained in the form I state (Fig. 4A,  lanes 4 -6). In contrast, incubation with 250 g of protein (Fig.  5A) or subsaturating amounts of protein (25 g) (Fig. 5B) from a rad7 mutant resulted in identical kinetics of conversion of form I to form II DNA as observed with the RAD ϩ strain (Fig.   FIG. 1. Western blot analysis showing specific binding of Histagged Rad7 protein to the Ni 2؉ -nitriloacetic acid matrix. Yeast WCEs were prepared from strains expressing the Rad7 protein with or without a C-terminal His 6 tag. Tagged Rad7 protein was eluted with 25 mM and 50 mM imidazole in Buffer A (lanes 1 and 3, respectively).  A and B). This result was confirmed in a direct comparison of wild-type, rad7, rad16, and various NER-defective mutants (Fig. 4A). The same results were obtained with rad16 mutant extracts (data not shown). Thus, rad7 and rad16 mutant extracts are as active as wild-type extracts with respect to damage-specific incision of DNA.

3,
Analysis of radiolabel incorporated by repair synthesis during incubation with wild-type (RAD ϩ ) extracts revealed increasing incorporation over a 2-h period in both form II DNA and in various topoisomers (Fig. 3C). The difference in the kinetics of repair synthesis and DNA incision reflects the fact that just a single nick in the form I substrate DNA converts it to the form II state, whereas the detection of radioactivity incorporated during repair synthesis is significantly less sensitive. Only background levels of repair synthesis were observed with extracts of rad4, rad14, and ssl2 mutants (Fig. 4B,  lanes 4 -6). Consistent with the results of previous studies (6), extracts of both rad7 and rad16 mutants supported markedly reduced (but clearly detectable) levels of repair synthesis in form II DNA (Figs. 4B and 5C). We have previously shown that the level of repair synthesis does not increase as a function of increasing the amount of WCEs (10). Hence, this observation is not the result of suboptimal incubation conditions. Furthermore, defective repair synthesis can be fully corrected by mixing rad7 or rad16 mutant extracts with extracts from other NER-defective mutants, such as rad14 or rad2 (10). Thus, the extracts are not nonspecifically inactivated for repair synthesis.
Conversion of form I to form II DNA was also observed when purified form I plasmid DNA not treated with AAF was incubated with extracts of RAD ϩ strains (data not shown). Our conclusion that the conversion of AAF-treated DNA to form II reflects damage-specific incision derives from the observation that such conversion was not observed in extracts of rad4, rad14, or ssl2 NER-defective mutants (Fig. 4A). It is well documented that "untreated" plasmid DNA contains background levels of base damage, typically oxidative damage, generated during purification of the DNA (26, 27). Because "native" DNA prepared under the most careful conditions is subject to such Both wild-type and rad7 or rad16 mutant extracts incise the DNA (lanes 1-3), whereas the incision-defective mutants rad4, rad14, and ssl2 (lanes 4 -6) do not. B, autoradiogram of A. Repair synthesis was observed in the wild-type extract (lane 1) but was defective in the rad4, rad14, and ssl2 mutant extracts (lanes 4 -6). Extracts of rad7 and rad16 mutants show residual repair synthesis (lanes 2 and 3). 5. A, DNA visualized by ethidium bromide staining. Lane 1, form I DNA prior to incubation with WCEs. Lanes 2-5, incubation with 250 g of rad7 mutant extract for 2, 20, 60, and 120 min, respectively, converts form I to form II DNA. B, incubation of rad7 mutant extract with 25 g of rad7 extract for 2, 20, and 40 min shows partial conversion of form I to form II DNA at 2 min. C, autoradiogram of A demonstrating residual incorporation of radiolabel into the substrate DNA associated with defective repair synthesis in rad7 mutant extracts.
perturbations (26,27), investigators wishing to avoid subtle topological changes in DNA and/or chromatin substrates during incubations in vitro should consider using extracts derived from cells that are defective in NER.
In summary, the results of the experiments described above indicate that rad7 and rad16 mutants are indeed defective in NER in yeast WCEs. However, this defect is not at the level of DNA incision, but in some postincision event(s).
rad7 and rad16 Mutants Are Proficient in Incision In Vivo-A number of in vivo studies have demonstrated that NER measured by the specific loss of pyrimidine dimers from the heterochromatic HML␣ locus is defective in rad7 and rad16 mutant strains (4). The results of the experiments with WCEs described above invited the possibility that the failure to observe a loss of UV-induced lesions from the HML␣ locus is similarly not the result of defective DNA incision. In wild-type living cells, once NER is initiated at any single lesion, the entire process is completed very rapidly. This precludes the facile detection of DNA incisions in the overall population of substrate DNA molecules. Hence, following UV irradiation of RAD ϩ cells, no significant change in the average molecular weight of genomic DNA is detected as a function of the time of incubation (28). Similar results were obtained when NER was examined at the level of individual genes by Southern hybridization with gene-specific probes (20). In such experiments, the presence of pyrimidine dimers in a defined DNA sequence can be detected by extracting the DNA and digesting it with dimerspecific enzymes, such as T4 endonuclease V (20). The persistence of dimers due to defective NER results in degradation of the DNA by the endonuclease and loss of a hybridization signal during subsequent Southern analysis (20). As NER approaches completion, substrate sites for the endonuclease probe (dimers) are lost, and the hybridization signal returns (20). However, in repair-proficient cells in which the DNA is not treated with a dimer-specific endonuclease, a hybridization signal of constant intensity is observed during the repair process (20). DNA incisions do not accumulate because the repair process at any given site is rapidly completed.
In order to facilitate the persistence of nicks associated with defective NER at a step following DNA incision, we utilized strains conditionally defective in the CDC9 gene, which encodes DNA ligase I (29). Fig. 6 shows the effect of the cdc9 ts mutation on NER at the MAT␣ and HML␣ loci in a RAD ϩ strain and in various rad mutants. When a UV-irradiated RAD ϩ , cdc9 strain was incubated at the nonpermissive temperature (Fig. 6A), we observed a loss of both MAT␣ and HML␣ signals following Southern hybridization, as expected if incisions persisted at these loci due to a failure of DNA ligation. No loss of signal was detected in unirradiated cells (Fig. 6A). Similarly, no loss of signal was observed at either locus in an irradiated or unirradiated rad2, cdc9 strain, consistent with defective incision in such a mutant. In contrast, proficient incision was detected at both the MAT␣ and HML␣ loci in rad7, cdc9 and rad16, cdc9 strains. To demonstrate that DNA ligation is indeed defective at the restrictive temperature, experiments were repeated at the permissive temperature for the cdc9 mutation. No incisions were detected in the MAT␣ and HML␣ loci in any of the strains examined (Fig. 6B). These observations indicate that incisions are generated at UV radiation-induced photoproducts at the MAT␣ and HML␣ loci both in RAD ϩ and rad7/rad16 mutant strains. These incisions are apparently rapidly religated under conditions permissive for DNA ligase I activity.
rad7 and rad16 Mutant Strains Are Defective in Oligonucleotide Excision in Vitro-Because rad7 and rad16 mutant extracts are proficient in DNA incision but defective in repair synthesis (6), we examined their ability to support the excision of oligonucleotide fragments generated by the bimodal incision reaction during NER. For these studies, we utilized a plasmid carrying a single pyrimidine dimer at a defined site on one of the DNA strands. The plasmid was radiolabeled close to the site of the dimer (see under "Materials and Methods"). Incubation of the plasmid with extracts of RAD ϩ cells resulted in the release of radiolabeled oligonucleotide fragments of the expected size (ϳ30 nucleotides) (Fig. 7, lane 5). No excision was detected in extracts of rad1 and rad14 mutant strains (Fig. 7,  lanes 1 and 2), consistent with the fact that these extracts do not support any damage-specific incision of DNA. Significantly reduced but clearly detectable oligonucleotide excision was observed in rad7 and rad16 mutant extracts (Fig. 7, lanes 3 and  4). Because these extracts are proficient for incision of damaged DNA, we conclude that as is the case in vivo (see above), incisions generated in vitro in rad7 and rad16 extracts are rapidly religated.

Complementation of Defective Repair Synthesis and Oligonucleotide Excision in rad7 and rad16 Mutant Extracts-Previous
studies have demonstrated that the Rad7 and Rad16 proteins are stably associated in vivo and in vitro (6,9). In order to avoid possible artifacts in the subunit composition of a complex that includes Rad7 and Rad16 proteins following overexpression of these proteins in yeast, we purified such a complex from a strain expressing endogenous Rad16 protein and His 6 -tagged Rad7 protein expressed from its normal promoter on a single copy plasmid. Rad7 and Rad16 proteins co-eluted through multiple distinct chromatographic steps (Fig. 2). We obtained ϳ30,000-fold enrichment of a Rad7/Rad16-containing complex with ϳ20% yield. Peak fractions from the final purification step (glycerol gradient sedimentation) (Fig. 2) coincided with a peak of activity that complemented defective repair synthesis in extracts of a rad7, rad16 double mutant (Figs. 2E and 8A). Addition of increasing amounts of this fraction resulted in increasing complementation of defective repair synthesis in rad7 and rad16 extracts (Fig. 8B, lanes 6 -8 and 9 -11, respec-FIG. 6. A, autoradiogram demonstrating gene-specific incision at 37°C (nonpermissive temperature for cdc9) following exposure of cells to UV radiation at the MAT␣ and HML␣ loci. Incisions are reflected by loss of the hybridization signal at the MAT␣ and HML␣ loci in the RAD ϩ , cdc9 and rad7 (or rad16), cdc9 mutants. No incisions were detected at either locus in a rad2, cdc9 mutant. Samples were loaded in pairs; one sample represents DNA from unirradiated cells (Ϫ), and the other represents DNA from cells irradiated with UV light (ϩ). B, autoradiogram showing the absence of gene-specific incision at the permissive temperature for cdc9 (23°C). tively) but had no effect on defective repair synthesis in rad2 mutant extracts (Fig. 8B, lanes 12-14). It should be noted that once again rad7 (Fig. 8B, lane 3) and rad16 (data not shown) mutant extracts showed residual capacity for repair synthesis compared with rad2 mutant extracts (Fig. 8B, lane 2) and that mixing rad7 and rad2 mutant extracts restored repair synthesis (Fig. 8B, lane 4). Complementation of defective oligonucleotide excision by the purified Rad7/Rad16-containing complex was also observed (data not shown).

DISCUSSION
Studies on NER in yeast have led to the definition of distinct subpathways of this process (30). Global genome repair is a process that operates on transcriptionally silent DNA, including the nontranscribed strands of transcriptionally active genes. Transcription-coupled repair is a process that operates exclusively on the transcribed strands of transcriptionally active genes (30). Although the two processes utilize the same general NER machinery, specific additional gene products have been implicated for each subpathway. The Rad7 and Rad16 proteins are implicated in the global subpathway because rad7 and rad16 mutants are defective in the removal of pyrimidine dimers from nontranscribed regions of the genome in vivo, whereas transcription-coupled repair operates normally in such strains (5). Similarly, the RAD26 gene product is implicated in transcription-coupled NER, because rad26 mutants are defective in this process but not in global genome repair (30).
The observation that rad7 and rad16 mutants are defective in NER of the transcriptionally repressed HML␣ locus but not the homologous transcriptionally active MAT␣ locus led to the suggestion that the Rad7 and Rad16 proteins participate in global genome repair by contributing to disruption of heterochromatic regions in order to allow access of the repair machinery to sites of base damage (4). This model has been indirectly reinforced by the observation that Rad16 protein is a member of the SWI2/SNF2 superfamily of DNA-dependent ATPases (7), a number of which are known subunits of chromatin remodeling complexes (31).
We previously reported a requirement for the Rad7 and Rad16 proteins for normal levels of repair synthesis of naked plasmid DNA in WCEs under conditions that are independent of RNA polymerase II transcription (6). In contrast, it has been reported that in a reconstituted system using purified NER proteins (also independent of transcription), both DNA incision and oligonucleotide excision transpire in the absence of Rad7 and Rad16 proteins (8), and that these proteins actually increase the efficiency of the recognition of base damage during NER (9). We were therefore led to examine NER in rad7 and rad16 mutant strains more closely. We investigated two possible explanations for defective repair synthesis in WCEs. First, we considered the possibility that following the addition of naked plasmid DNA to yeast WCEs, the DNA is assembled into chromatin, the processing of which specifically requires the Rad7 and Rad16 proteins during NER. Hence, defective repair synthesis in WCEs would reflect a defect in NER of chromatin that is obviated in a system reconstituted with purified proteins and naked DNA. However, the present studies provide no evidence for nucleosome assembly on plasmid DNA during incubation with yeast WCEs. This is not surprising, because it is well established that efficient nucleosome assembly is achieved with extracts from Xenopus oocytes or Drosophila embryos, which are highly enriched for histones (32), but not in yeast or human extracts. Indeed, this property of human extracts has been exploited to demonstrate a requirement for a distinct protein complex termed chromatin assembly factor 1 for efficient nucleosome assembly following repair synthesis in vitro (33). A similar protein complex encoded by the chromatin FIG. 8. A, in vitro complementation of repair synthesis in WCEs from a rad7, rad16 double mutant. Peak activity was observed in fraction 18. Reactions were performed using 5-l aliquots from each fraction (50 ng of protein in fraction 18). B, in vitro NER was performed using standard conditions with the various yeast WCEs indicated. Normal repair synthesis was observed in wild-type extracts (lane 1), and defective repair synthesis was observed in rad2 mutant extracts (lane 2). Residual repair synthesis was observed in rad7 (lane 3) extracts. Lane 4, complementation of repair synthesis by mixing rad2 and rad7 extracts. Complementation of repair synthesis was observed by addition of increasing amounts of purified Rad7/Rad16-containing complex (5, 50, and 75 ng or protein) in rad7 (lanes 6 -8) and rad16 (lanes 9 -11) extracts but not in rad2 (lanes 12-14) extracts.  rad1 and rad14 extracts (lanes 1 and 2), and markedly reduced levels of excision were observed in extracts of rad16 and rad7 mutants (lanes 3 and 4).
assembly complex genes has been identified in yeast (34).
A second possibility was that the Rad7 and Rad16 proteins may be involved in a specific stage of NER that precedes repair synthesis of DNA. We dissected the NER reaction using independent assays to examine individual events in this process. Extracts from rad7 or rad16 mutants are fully proficient in the incision of damaged DNA compared with wild-type strains. As already indicated, previous studies have shown that DNA incision (and oligonucleotide excision) can take place in a reconstituted in vitro system without added Rad7 and Rad16 protein (8). However, these studies did not exclude the possibility that such proteins may have been present as contaminants. Hence, the present studies provide a direct demonstration that rad7 and rad16 mutant extracts are proficient for DNA incision in vitro. This result was corroborated in vivo using a ts cdc9 mutant strain to trap incision intermediates during NER. Our studies additionally showed that rad7 and rad16 mutants are proficient for NER-dependent incision in regions such as HML␣, which were previously thought to be refractory to the NER machinery.
We confirmed an earlier observation (6) that rad7 and rad16 WCEs are defective in the repair synthesis step of the NER reaction. Additionally, we have provided a direct demonstration of a defect in oligonucleotide excision of base damage in these extracts. We have consistently observed that the defects in both oligonucleotide excision and repair synthesis in rad7 and rad16 mutants are not as complete as those observed in incision-defective mutants, such as rad1 or rad14. This is not the result of subsaturating reaction conditions, because previous studies have shown that deficient repair synthesis is independent of the protein concentration of WCEs (6). We estimate that the level of residual excision and resynthesis in rad7 and rad16 mutants to be ϳ10 -20% that of wild-type levels. This may explain the oligonucleotide excision observed in the reconstituted NER assay in the absence of added Rad7 and Rad16 proteins (8). Both defective repair synthesis and oligonucleotide excision can be complemented in vitro by a purified Rad7/ Rad16-containing protein complex.
A Rad7/Rad16 complex was previously isolated from cells in which both proteins were simultaneously overexpressed from strong promoters on multicopy plasmids (9). In order to avoid stoichiometric artifacts that might arise from such overexpression, including the possible exclusion of other relevant subunits, we purified such a complex without overexpression of Rad7 or Rad16. Preliminary analysis shows that the complex has a molecular mass of ϳ250 kDa (considerably larger than that expected for a complex with a 1:1 stoichiometry of Rad7 and Rad16 protein) and that it includes at least one polypeptide other than Rad7 or Rad16. 2 A more detailed characterization of this complex will be presented elsewhere.
In contrast to a previous suggestion (10) that the yeast Rad7 and Rad16 proteins participate in the recognition of base damage in transcriptionally silent DNA and stimulate DNA incision at such sites, we conclude that rad7 and rad16 mutants are proficient in the recognition and incision of sites of base damage in nontranscribed regions of DNA. However, such mutants cannot efficiently execute the subsequent steps of oligonucleotide excision and repair synthesis. Our results suggest that although the Rad7 and Rad16 proteins participate in postincision events during NER, they are not indispensable for this process. This may explain the fact that rad7 and rad16 mutants are only partially sensitive to DNA-damaging agents, such as UV radiation.
The specific role(s) of Rad7 and Rad16 in oligonucleotide excision (and possibly coupled repair synthesis) during NER of transcriptionally repressed regions of the genome and the nontranscribed strand of transcriptionally active genes, as well as the observation that such a role(s) is obviated during NER of the template strand of transcriptionally active genes, remains to be explained. All known members of the SWI/SNF superfamily of proteins are DNA-dependent ATPases for which the DNA dependence can vary. For example, the yeast and human SNF2 proteins (the ATPase subunits of the SWI/SNF complex) are stimulated by synthetic DNA with four-way junctions (31). This structure is believed to mimic crossover points where DNA enters and exits nucleosomes (31). On the other hand, imitation switch (ISWI) protein, the ATPase subunit of the nucleosome remodeling factor (NURF) complex (35), has a specific requirement for DNA in chromatin. Conceivably, the ATPase activity of Rad16 protein has a requirement for a particular DNA conformation generated by damage specific incision of the coding strand of transcriptionally active genes. Once this ATPase is activated, the Rad7/Rad16-containing complex may facilitate the remodeling of a protein-DNA structure (possibly including the NER machinery) that is uniquely associated with NER of transcriptionally repressed regions and the nontranscribed strand of transcriptionally active genes. This remodeling presumably facilitates oligonucleotide excision and repair synthesis. The dispensability of Rad7 and Rad16 for efficient oligonucleotide excision and repair synthesis of the template strand of transcriptionally active genes may be a consequence of a distinct DNA conformation that derives from transcription itself. Although our studies do not support the notion that purified plasmid DNA is organized into nucleosomes in yeast whole cell extracts, we cannot formally eliminate the possibility that a requirement for Rad7 and Rad16 proteins for oligonucleotide excision and repair synthesis during NER of plasmid DNA in such extracts reflects the presence of other protein-DNA complexes.
It is interesting to note that whereas the majority of the polypeptides involved in NER in yeast are highly conserved in humans, and the essential biochemistry of global genome repair in vitro appears to be identical in yeast and human extracts, human homologs of the yeast RAD7 and RAD16 genes have not been identified. It remains to be determined whether this reflects subtle differences in the biochemistry of NER and or RNA polymerase II transcription in lower and higher eukaryotes, or whether the human homologs have simply eluded identification to date.