Overproduction, Purification, and Characterization of the XPC Subunit of the Human DNA Repair Excision Nuclease*

Xeroderma pigmentosum complementation group C gene (XPC) encodes a protein of 125 kDa which is present in a tight complex with a 58-kDa protein encoded by the human homolog of the yeast RAD23 gene, HHR23B (Masutani, C., Sugasawa, K., Yanagisawa, J., Sonoyama, T., Ui, M., Enomoto, T., Takio, K., Tanaka, K., van der Spek, P. J., Bootsma, D., Hoeijmakers, J. H. J., and Hanaoka, F. (1994) EMBO J. 13, 1831–1843). The XPC-HHR23B complex is required for excision of thymine dimers from DNA in a human excision nuclease system reconstituted from purified proteins. In order to under- stand the role of the XPC-HHR23B complex in excision repair, we have overexpressed each subunit alone and the heterodimer in heterologous systems, purified them, and characterized their biochemical properties. We find that both XPC and the heterodimer bind DNA with high affinity and UV-damaged DNA with slightly higher pref- erence. Surprisingly, we find that the XPC subunit alone is sufficient for reconstitution of the human excision nuclease and that the HHR23B subunit has no detectable effect on the excision activity of the reconstituted system. Nucleotide excision repair is a general repair system which is particularly suited for removing bulky DNA lesions such as thymine dimers and cisplatin-d(GpG) diadducts (Friedberg et al ., 1995; Sancar, 1996). A defect in excision repair causes xeroderma

Xeroderma pigmentosum complementation group C gene (XPC) encodes a protein of 125 kDa which is present in a tight complex with a 58-kDa protein encoded by the human homolog of the yeast RAD23 gene, HHR23B (Masutani, C., Sugasawa, K., Yanagisawa, J., Sonoyama, T., Ui, M., Enomoto, T., Takio, K., Tanaka, K., van der Spek, P. J., Bootsma, D., Hoeijmakers, J. H. J., and Hanaoka, F. (1994) EMBO J. 13, 1831-1843). The XPC-HHR23B complex is required for excision of thymine dimers from DNA in a human excision nuclease system reconstituted from purified proteins. In order to understand the role of the XPC-HHR23B complex in excision repair, we have overexpressed each subunit alone and the heterodimer in heterologous systems, purified them, and characterized their biochemical properties. We find that both XPC and the heterodimer bind DNA with high affinity and UV-damaged DNA with slightly higher preference. Surprisingly, we find that the XPC subunit alone is sufficient for reconstitution of the human excision nuclease and that the HHR23B subunit has no detectable effect on the excision activity of the reconstituted system.
Nucleotide excision repair is a general repair system which is particularly suited for removing bulky DNA lesions such as thymine dimers and cisplatin-d(GpG) diadducts (Friedberg et al., 1995;Sancar, 1996). A defect in excision repair causes xeroderma pigmentosum (XP) 1 in humans. XP patients are hypersensitive to sunlight and develop skin cancers at a young age and at high frequency; some patients, in addition, exhibit neurological symptoms (Cleaver and Kraemer, 1989). Proteins encoded by seven XP genes, XPA through XPG, are required for the dual incision (excision) step of nucleotide excision repair. In addition to the proteins encoded by the XP genes, the ERCC1 protein, the replication protein RPA, and the multimeric transcription factor TFIIH (two of the subunits are XPB and XPD) are required for the dual incision (Mu et al., , 1996Guzder et al., 1995b). Furthermore, upon purification of XPC complementing activity, it was found that the 125-kDa protein encoded by XPC (Legerski and Peterson, 1992;Masutani et al., 1994) was in a complex with a protein of 58 kDa which is highly homologous to the yeast excision repair protein encoded by the RAD23 gene (Watkins et al., 1993;Masutani et al., 1994). In humans there are two genes with sequence homology to RAD23, and these were named human homolog of RAD23 A and B (HHR23A and HHR23B), respectively; only the protein encoded by HHR23B is found in a complex with XPC protein (Masutani et al., 1994).
Although XPC and HHR23B appear to be tightly bound and the final purification step for XPC yields these two proteins in 1:1 stoichiometry (Masutani et al., 1994;Aboussekhra et al., 1995;Mu et al., 1996), there was no genetic or biochemical data from mammalian systems indicating that HHR23B is involved in excision repair. To clarify the role of HHR23B in repair, we expressed XPC and HHR23B, separately and in a complex using baculovirus/insect cells and plasmid/Escherichia coli expression systems, purified these proteins, and characterized them. We found that XPC and the XPC-HHR23B complex bind to DNA nonspecifically and with relatively high affinity and to UV-damaged DNA with slightly higher affinity. Both forms of XPC are capable of complementing cell-free extracts of XP-C mutants and are active in reconstituting excision nuclease activity in a completely defined system. We conclude that with naked DNA under our experimental conditions HHR23B does not play a direct role in excision repair.

Plasmid Construction and Baculovirus Stock Establishment-Two
constructs were used for protein expression in insect cells (Fig. 1A). For p2Bac.XPC-HHR23B, a 1.8-kb SmaI-DraI fragment from pUC19.HHR23B (Masutani et al., 1994) was subcloned into the p2Bac vector (Invitrogen) at the PvuII site under the control of the PH promoter (p2Bac.HHR23B), followed by subcloning of the 3.6-kb NotI fragment from pBSIISK(ϩ).XPC (Masutani et al., 1994) at the NotI site under the control of the P10 promoter. For p2Bac.XPC, only the NotI fragment was subcloned into the p2Bac vector. Sf21 cells were transfected with either p2Bac.XPC-HHR23B or p2Bac.XPC using the Bacu-loGold Transfection Kit (Pharmingen). Recombinant virus stocks were established from single plaques and identified by polymerase chain reaction (PCR) amplification of viral DNA using a combination of p2Bac vector and gene-specific primers. Standard procedures (O'Reilly et al., 1994) were used for cell culture and infection by recombinant baculoviruses as well as for amplification and titering of the virus stocks.
To express recombinant proteins in bacterial cells, two constructs were used (Fig. 1B). For pHis.HHR23B, a 1.2-kb PCR fragment was subcloned into pQE-30 (Qiagen). The PCR primers were designed to introduce a 5Ј-BamHI site while simultaneously deleting the first two (Met-Gln) and substituting the 3rd and 4th amino acids (from Val-Thr to Gly-Ser) and to generate a 3Ј-SalI site beyond the stop codon. For pMBP.XPC, a 2.5-kb PCR fragment was subcloned into pMAL-c2 (New England Biolabs). Primers were designed to generate a blunt end at the first ATG codon of pXPC-3 (Legerski and Peterson, 1992) and to introduce a 3Ј-SalI site beyond the stop codon. The insert was in-frame with malE sequences (data not shown), but only a truncated fusion protein was produced. Therefore, the pMBP.XPC DNA was digested with Hin-dIII and SalI; the 1.3-kb fragment was gel-purified, and ends were filled in with the Klenow enzyme, recut with SalI, and subcloned into pMAL-c2 creating pMBP.XPC(445C), which encodes the malE gene fused in-frame to the 445 amino acids from the carboxyl terminus of XPC.
For purification of recombinant XPC, CFE (20 mg) was applied to a phosphocellulose P11 (Whatman) column (4 ml), washed with 40 ml of buffer B containing 0.15 M KCl, and bound proteins were eluted with buffer B containing 1.0 M KCl. The XPC fractions were pooled, adjusted to 0.6 M KCl by dilution with buffer B, and applied to a single-stranded DNA-cellulose (Sigma) column (5 ml). The column was washed with 15 ml of buffer B containing 0.6 M KCl, and bound proteins were eluted with buffer B containing 1.5 M KCl. The XPC fractions were pooled, dialyzed against buffer B containing 0.15 M KCl, and applied to a DEAE-agarose (Bio-Rad) column (1 ml). XPC does not bind to this column, and further purification is not achieved, but this chromatographic step removes contaminating DNA which interferes with subsequent analyses. The flow-through fractions containing XPC were pooled and stored at Ϫ80°C.
For purification of the XPC-HHR23B complex, a similar scheme was used with the following modifications. The CFE was dialyzed into buffer B containing 0.3 M KCl, 24 mg were applied to a phosphocellulose column (4 ml) which was washed with 25 ml of the same buffer, and bound proteins were eluted as described above. The XPC-HHR23Bcontaining fractions were pooled, adjusted to 0.3 M KCl by dilution with buffer B, and applied to a single-stranded DNA cellulose (Sigma) column (5 ml); this column was washed with 5 ml of buffer B containing 0.3 M KCl, and bound proteins were eluted with a step gradient of 0.3-1.5 M KCl in buffer B, with highly purified XPC-HHR23B eluting at 0.9 -1.2 M KCl. XPC-HHR23B fractions were pooled, dialyzed against buffer B containing 0.3 M KCl, and applied to a DEAE-agarose (Bio-Rad) column (3 ml) which was washed with buffer B containing 0.3 M KCl; fractions were pooled and stored at Ϫ80°C.
For purification of histidine-tagged HHR23B, E. coli strain DH5␣FЈlacI q was transformed with pHis.HHR23B, and cells were grown at 37°C to A 600 ϳ0.6 and induced for 4 h with isopropyl-1-thio-␤-D-galactopyranoside at 0.5 mM. The bacterial cells were collected, washed with phosphate-buffered saline, treated with lysozyme at 1 mg/ml, and disrupted by sonication (8 ϫ 15 s with a Branson Sonicator set at maximum power). Clarified extract was obtained by centrifugation at 35,000 rpm (Beckman Ti60 rotor), and His-HHR23B was purified by Ni-NTA chromatography (Qiagen) according to the manufacturer's directions. CFE (100 mg in 10 ml) was mixed with 10 ml of resin for 16 h at 4°C, and then the CFE-resin mixture was poured into a column (15 cm ϫ 1.5 cm). Following extensive washing with buffer C (50 mM sodium phosphate, pH 6, 0.3 M NaCl, 10% glycerol), bound proteins were eluted with a step gradient of 0.05 to 0.5 M imidazole in buffer C, with His-HHR23B eluting at 0.15-0.3 M imidazole. His-HHR23B-containing fractions were pooled, dialyzed into buffer A, and stored at Ϫ80°C.
For purification of the MBP-XPC(445C) fusion protein, the initial processing was as described above for His-HHR23B except the 4-h induction was with 0.3 mM isopropyl-1-thio-␤-D-galactopyranoside. Clarified extract (200 mg in 150 ml) was applied to an amylose column (5 ml, New England Biolabs), and the manufacturer's directions were followed for elution of fusion proteins with 10 mM maltose. MBP-XPC(445C)-containing fractions were pooled, dialyzed into buffer A, and stored at Ϫ80°C.
Antiserum Production and Immunoblotting-Two antigens, MBP-XPC(445C) and His-HHR23B, were used for the generation of anti-XPC and anti-HHR23B antisera. The bacterially expressed proteins were mixed with adjuvant (RIBI Immunochem Research, Inc.) and used to immunize rabbits at 2-3-week intervals (0.5 mg of protein per injection) with serum samples being collected after the second injection. These antisera react with XPC and HHR23B, respectively, in immunoblot analyses. For immunoblotting, proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Schleicher and Schuell, Protran) using standard procedures for electroblotting. Antigen-antibody complexes were detected using the alkaline phosphatase method (Promega) or enhanced chemiluminescence (Amersham).
DNA Binding Assays-Several oligonucleotides were prepared for the DNA binding assays, but all substrates within a particular category were prepared in essentially the same manner. Approximately 50 pmol of each oligonucleotide were 5Ј-end-labeled with T4 polynucleotide kinase (New England Biolabs) and [␥-32 P]ATP (7000 Ci/mmol; ICN). Labeled oligonucleotides were either ethanol-precipitated alone (ss-DNA) or in the presence of unlabeled complementary oligomers (ds-DNA), and DNA was resuspended in annealing buffer (40 mM Tris, pH 7.4, 100 mM NaCl, 4 mM MgCl 2 ). To obtain ds-DNA substrates, the coprecipitated oligomers were heated at 90 -95°C for 5 min and slow cooled to Ͻ30°C (about 3 h). Both ss-and ds-DNA were separated from unincorporated [␥-32 P]ATP or nonhybridized complementary strands by electrophoresis on 6 -8% polyacrylamide gels, followed by electroelution, ethanol precipitation, and resuspension in annealing buffer. For preparing damaged DNA, the DNA was irradiated with 4 -5 kJ/m 2 using a germicidal lamp (254 nm). DNA was stored at 4°C in annealing buffer and diluted just prior to use in binding buffer. The substrates used (Table I) include (a) 64-nt labeled oligomer (ss-DNA), part of which was annealed to 61-mer (oligo 2) to generate a 60-bp duplex (ds-DNA) with a 1-base overhang at the 5Ј end and 4-base overhang at the 3Ј end, (b) 45-nt labeled oligomer (ss-DNA), an aliquot of which was annealed to 45-mer (oligo 4) to generate a 45-bp duplex (ds-DNA), (c) 30-nt labeled oligomer (ss-DNA), half of which was annealed to 30-mer (oligo 6) to generate a 26-bp duplex (ds-DNA), with 4-base overhangs at both the 5Ј and 3Ј ends, and (d) 20-nt labeled oligomer (ss-DNA), part of which was annealed to 20-mer (oligo 8) to generate a 20-bp duplex (ds-DNA).
For the binding assays, DNA (ϳ0.3 nM) was mixed with the indicated amounts of protein in 15 l of binding buffer (30 mM Hepes-KOH, pH 7.9, 4 mM potassium phosphate, pH 7.5, 100 mM KCl, 3.2 mM MgCl 2 , 0.4 mM DTT, 0.3 mM EDTA, 0.002% Triton X-100, 2 mM ATP, 2% glycerol, 1 g of bovine serum albumin). Following a 30-min incubation at 30°C, glycerol was added to 7%, and the protein-DNA complexes were separated by electrophoresis at a constant current of 25 mA at room temperature using 5% polyacrylamide gels containing acrylamide/methylenebisacrylamide at 30:1 in 50 mM Tris borate and 1 mM EDTA, pH 8.3. The gels were autoradiographed for visual inspection and quantitatively analyzed with an Ambis systems scanner. Repair Assays-The excision assay, utilizing internally labeled 140mer cholesterol-A substrate Mu et al., 1996) or a thymine dimer (TϽϾT) was used to assay repair activity for complementation of XPC cell-free extracts by rXPC and rXPC-HHR23B or for reconstitution of excision repair by either rXPC-HHR23B or rXPC Ϯ HHR23B. Briefly, for complementation, increasing amounts of CFE (60 -100 g) were mixed with a constant amount of either rXPC or rXPC-HHR23B and incubated for 60 min at 30°C in 25 l of excision buffer (35 mM Hepes-KOH, pH 7.9, 8 mM Tris-HCl, pH 7.5, 66 mM KCl, 32 mM NaCl, 5.6 mM MgCl 2 , 0.8 mM DTT, 0.4 mM EDTA, 2 mM ATP, 0.0004% Triton X-100, 20 M each of dATP, dCTP, dGTP, and dTTP, 0.2 mg/ml bovine serum albumin, and 2.9% glycerol). Then, the mixtures were incubated with Proteinase K for 15 min at 37°C, and the DNA was purified by phenol/phenol chloroform/ether extractions followed by ethanol precipitation, and the products were analyzed on 10% polyacrylamide sequencing gels. Similar conditions were used for the reconstitution experiments, except highly purified proteins were used in place of the mutant CFE and rXPC was compared with natural XPC (native, heterodimer purified from HeLa CFE) under optimal reaction conditions and under suboptimal conditions to test for the effect(s) of HHR23B, and the repair reaction was for 2.5 h at 30°C.

Purification of Recombinant XPC and XPC-HHR23B-We
attempted to overproduce XPC and HHR23B in E. coli, but failed to express full-length XPC in E. coli either from its own initiation codon or in the form of a fusion protein with maltosebinding protein (MBP). Nevertheless, the MBP-XPC fusion protein containing the carboxyl-terminal 445 amino acids of XPC was expressed and used for generating XPC antibodies used in the present study.
To obtain functional recombinant XPC, we cloned the gene into the baculovirus expression vector p2Bac (Invitrogen) either alone or together with HHR23B. Additionally, HHR23B was cloned into an E. coli expression plasmid. Fig. 1 shows the plasmid constructs used in the current study. Full-length soluble proteins were expressed with these constructs. The XPC protein and XPC-HHR23B heterodimer were purified essentially as described by Masutani et al. (1994). The HHR23B FIG. 1. Plasmids for recombinant protein expression. Four constructs were used in this study. A, p2Bac.XPC and p2Bac.XPC-HHR23B were used for the expression and purification of recombinant proteins used in DNA binding and repair assays. B, protein expressed with pHis.HHR23B was used for both of these assays as well as for antisera production. Recombinant protein expressed with pMBP.XPC(445C) was used only for the generation of antisera. RecSeq indicates recombination sequences for integration into the viral genome; Amp is the ampicillin resistance gene, ␤-lactamase. protein was expressed in soluble form and in full-length in E. coli and was purified from E. coli by nickel-affinity chromatography. Fig. 2 shows the analyses of purified proteins by Coomassie Blue staining and Western blotting.
Expression of HHR23B in XPC Mutant Cell Lines-Since one of the goals of this study was the investigation of the properties of the XPC protein in the absence of HHR23B, we wished to know whether or not the XP-C mutant cell-free extracts contained HHR23B protein at normal or reduced levels. Cell-free extracts from a variety of cell lines were tested by Western blotting. Fig. 3A shows that XPC mutants have normal levels of HHR23B. In fact, during purification of XPC complementing activity from HeLa cells, we found that HHR23B is in vast molar excess over the XPC polypeptide and separates from the heterodimer readily by ion exchange chromatography. 2 In addition, our purified repair factors were tested for the presence of HHR23B to ascertain that factors other than the XPC fraction were free of HHR23B, and, hence, complementation experiments done with recombinant XPC reflected the activity of this subunit alone. Fig. 3B reveals that the repair factors purified from HeLa cells contained no detectable HHR23B (Ͻ0.2 ng per excision assay) and hence any activity arising from the addition of XPC to the reconstituted system could safely be ascribed to XPC alone.
DNA Binding Properties-The natural as well as the recom-binant XPC-HHR23B were purified through DNA-cellulose affinity chromatography and hence were known to be DNA-binding proteins. For quantitative analysis of DNA binding properties, we conducted electrophoretic mobility shift assays with various types of DNAs. The heterodimer (both recombinant and natural forms) binds to single-stranded and doublestranded DNA (60-bp duplex) with comparable affinities (Fig.  4A) and with a K D of ϳ5 ϫ 10 Ϫ9 M (calculated from the 50% binding point); heavily irradiated 60-bp duplex (about 3 photolesions per molecule) was bound with slightly higher affinity (K D ϳ4 ϫ 10 Ϫ9 M, Fig. 4B). The heterodimer bound to the 45-bp duplex with somewhat lower affinity and to a 26-bp duplex very weakly; no binding with 20-bp oligomer could be detected over the concentration range used (Fig. 4C); similar binding properties were observed when the probe was single-stranded DNA. The HHR23B subunit on its own has no affinity for DNA (data not shown). Interestingly, the XPC subunit binds the 60-bp duplex with approximately 2-fold higher affinity than the heterodimer (Fig. 4D), suggesting that in fact HHR23B reduces the intrinsic DNA binding affinity of XPC.
Reconstitution of Excision Nuclease with XPC-HHR23B and with the XPC Polypeptide-Although mutations in XPC confer sensitivity to DNA damage (Li et al., 1993), there is no genetic evidence from mammalian systems implicating HHR23B in repair. We wished to find out whether HHR23B plays a role in repair by conducting complementation and reconstitution experiments using the XPC polypeptide and the XPC-HHR23B FIG. 4. DNA binding by XPC and XPC-HHR23B. Assay conditions were as described under "Materials and Methods"; for A and B, the DNA concentration was 0.3 nM and protein concentrations were as indicated. A, binding of rXPC-HHR23B to single-stranded 64-mer (ss-DNA) and double-stranded 60-bp duplex (ds-DNA). Quantitative analyses of the data revealed that 50% binding was achieved with 4.8 nM and 5.2 nM XPC with single-and double-stranded 60-mers, respectively. B, binding of rXPC-HHR23B to undamaged (ϪUV) and damaged (ϩUV) double-stranded 60-bp duplex. For UV-DNA, 4 nM gave 50% binding. C, binding of rXPC-HHR23B to single-stranded oligomers (20-mer, 30-mer, 45-mer, and 64-mer) and double-stranded duplexes (20 bp, 26 bp, 45 bp, and 60 bp). DNA concentration was 0.3 nM and rXPC-HHR23B was at 10.5 nM. At this protein concentration, only 40% binding was observed with ss-or ds-45-mer, and 50% binding was estimated to be 13 nM. Less than 5% of 26-bp duplex was bound, and no binding was detected for 30-mer or 20-mer (ss-and ds-DNA). D, binding of rXPC and XPC-HHR23B to 60-bp duplex. DNA concentration was 0.3 nM and protein concentrations were as indicated; heterodimer used in this experiment was from its natural source (HeLa cells). As with recombinant heterodimer, 50% binding was at 5.2 nM while rXPC bound 50% substrate at 2 nM. For all panels, the faster migrating free DNA species is single-stranded DNA.
heterodimer. We found that both forms are active in complementing XP-C cell-free extracts (Fig. 5). This experiment, which reveals that recombinant XPC is active in excision repair, indicates that the XPC polypeptide made in the absence of HHR23B folds properly. However, these experiments do not show whether XPC can function without HHR23B because XPC cells appear to have normal levels of HHR23B protein (Fig. 3A) which could reconstitute the XPC-HHR23B heterodimer upon addition of XPC to the cell-free extract. To address this particular question, we conducted experiments with a defined system of purified excision repair proteins not containing detectable levels of HHR23B (Fig. 3B). Fig. 6 shows that XPC alone is sufficient to reconstitute the excision nuclease to the same extent as the heterodimer purified from its natural source (HeLa cells). Supplementing HHR23B to the defined excision repair system reconstituted with XPC alone did not alter the rate of excision.
A previous study (Mu et al., 1996) showed that a cholesterol substitutent attached with a shorter linker to the phosphodiester backbone (cholesterol-B) than the one used in the current work (cholesterol-A) was excised normally even in the absence of XPC-HHR23B. Therefore, it was conceivable that these excisions in the absence of HHR23B or even the XPC-HHR23B complex were peculiar to these unnatural substrates. Hence, we tested the classic substrate for human excision nuclease, cyclobutane thymine dimer, in our reconstituted system. As evident from Fig. 6, lanes 9 and 10, even the XPC polypeptide alone is sufficient for reconstitution and there is no detectable effect of HHR23B on excision under these experimental conditions.
Thus, we conclude that HHR23B does not play a direct role in excision repair but may modulate the XPC activity. Indeed, HHR23B may have a negative effect on both DNA binding and excision activity because we reproducibly observe better binding (Fig. 4D) and excision (Figs. 5 and 6) with rXPC alone compared to the heterodimer.

DISCUSSION
The yeast RAD23 gene belongs to the RAD3 epistasis group, whose members participate in excision repair. Null mutations of genes in this group, such as RAD1, completely abolish excision repair . In contrast, deletion of RAD23 increases the UV sensitivity of yeast cells only modestly (Perozzi and Prakash, 1986;Watkins et al., 1993) indicating that, unlike RAD1 and other members of the RAD3 epistasis group, RAD23 is not essential for excision. Thus, extrapolating from yeast genetics, one would expect that the same would be true in humans. However, even though the yeast and human excision repair systems are highly homologous Friedberg et al., 1995;Sancar, 1996), certain significant differences have been found. In particular, it was found that the 3Ј nick of the dual incision could occur in the absence of XPF-ERCC1 in humans (Mu et al., 1996) but requires the entire complement of excision repair proteins in yeast (Guzder et al., 1995b). Thus, a priori, it was not possible to predict whether XPC without HHR23B would be sufficient to reconstitute the human excision nuclease. Indeed, it was reported that it was not possible to separate XPC from HHR23B without losing the XPC correcting activity (Masutani et al., 1994). The results presented in this report indicate that the HHR23B polypeptide is not required for the excision repair function of the XPC protein and that HHR23B modulates XPC activity, perhaps by interacting with and masking the DNA binding domain of XPC. It is possible that the stringent conditions used to separate XPC from HHR23B in the previous study inactivated the XPC protein.
In light of our results, it is reasonable to ask whether HHR23B plays any role in repair. To answer this question, in addition to the yeast rad23 phenotype, one must take the following findings into account. First, in humans, two RAD23 homologs were found which exhibit 57% sequence identity to one another and 30 -34% sequence identity to RAD23. Second, of the two polypeptides, only HHR23B was found in a complex with XPC (Masutani et al., 1994). Third, RAD23 (Watkins et al., 1993) and HHR23 belong to the ubiquitin-fusion family of proteins in that the NH 2 -terminal 80 amino acids are highly homologous to ubiquitin (Masutani et al., 1994). The precise role of ubiquitin in these proteins is not known but is thought to function as a chaperone in assisting proper folding and assembly and thermostability (Finley et al., 1989;Garrett et al., 1994;Aso et al., 1995). Fourth, the Rad23 protein is not essential for, but promotes complex formation between TFIIH and Rad14 (XPA homolog) proteins (Guzder et al., 1995a). In humans TFIIH seems to be directly bound to XPA without the aid of other proteins  although the possibility that HHR23B stimulates XPA-TFIIH interaction has not been ruled out. Fifth, the yeast homolog of XPC, the Rad4 protein, is required for global excision repair  but not for excision repair of the transcribed strand of rDNA genes which are transcribed by RNA polymerase I (Verhage et al., 1996a). In contrast, in humans the template strand of the polymerase II-transcribed sequences is repaired in XP-C mutants at an essentially normal rate (Mellon et al., 1987;Kantor et al., 1990;Venema et al., 1990Venema et al., , 1991Evans et al., 1993). In fact, it has been demonstrated in vitro that for certain lesions, even in the absence of transcription, excision repair occurs without XPC-HHR23B (Mu et al., 1996). Thus, not only HHR23B but even XPC is dispensable for excision nuclease activity under certain conditions or with certain substrates.
The following model is consistent with most of these observations. The HHR23B subunit of the heterodimer may be involved in nucleosome disassembly or reorganization which makes DNA accessible to the XPC subunit and other factors of the excision nuclease system. Alternatively, HHR23B may modulate the activity of XPC and perhaps function as a chaperone molecule. Indeed, preliminary experiments suggest that HHR23B protects the XPC subunit from thermal inactivation as tested by the DNA binding assay. 3 The role of XPC may be to stabilize the local unwinding that is thought to occur in the preincision complex. Thus, with naked DNA (such as the 140mer substrate), there is no need for HHR23B. Similarly, in transcribed DNA where local unwinding is provided by the RNA polymerase stalled at a lesion or with certain lesions which intrinsically cause local unwinding, XPC is not needed and hence excision occurs at a near-normal rate in the absence of the XPC-HHR23B heterodimer. This model is in agreement with the finding of Miller et al. (1982) that yeast rad23 mutants removed pyrimidine dimers at about 60% of wild type levels. In contrast, McCready (1994) and Verhage et al. (1996b) using different methodologies did not find dimer removal from the genome overall or from transcriptionally active genes in rad23 mutants. Further studies are needed to establish the function of XPC, to determine whether there are bona fide differences between the roles of HHR23B in humans and RAD23 in yeast, and to reconcile these seemingly contradictory reports regarding the requirement of RAD23 for excision repair in yeast.