Centrosome Protein Centrin 2/Caltractin 1 Is Part of the Xeroderma Pigmentosum Group C Complex That Initiates Global Genome Nucleotide Excision Repair*

Nucleotide excision repair (NER) is carried out by xeroderma pigmentosum (XP) factors. Before the excision reaction, DNA damage is recognized by a complex originally thought to contain the XP group C responsible gene product (XPC) and the human homologue of Rad23 B (HR23B). Here, we show that centrin 2/caltrac-tin 1 (CEN2) is also a component of the XPC repair complex. We demonstrate that nearly all XPC complexes contain CEN2, that CEN2 interacts directly with XPC, and that CEN2, in cooperation with HR23B, stabilizes XPC, which stimulates XPC NER activity in vitro . CEN2 has been shown to play an important role in centrosome duplication. Thus, those findings suggest that the XPC-CEN2 interaction may reflect coupling of cell division and NER.

Centrin (caltractin) found in the centrosomes of a wide variety of organisms (1) is a member of the highly conserved superfamily of calcium binding EF-hand proteins. Centrin was first discovered in the flagellar apparatus of the unicellular green algae Chlamydomonas reinhardii, where it is directly responsible for the contraction of calcium-sensitive structures (2)(3)(4). Analysis of a centrin mutant of green algae suggested that centrin is important for the proper segregation of the flagellar apparatus during cell division and for accurate basal body duplication and separation (5). The counterpart of centrin in Saccharomyces cerevisiae, Cdc31, was shown to be necessary for the initiation of spindle pole body (SPB) 1 formation (6,7).
In human, at least three centrin isoforms have been identified: centrin 1/caltractin 2 (CEN1) (8), centrin2/caltractin1 (CEN2) (9), and centrin 3 (CEN3) (10). Among these centrin isoforms, CEN1 displays tissue-specific expression, whereas CEN2 is expressed ubiquitously (11). A recent study showed that the expression and distribution of these three isoforms varies during human ciliated cell differentiation and proliferation in vitro (12). Moreover, CEN3, the closest homologue of S. cerevisiae Cdc31, has been suggested to have a distinct function in centrosome duplication (13). On the other hand, CEN2 and its isoforms have been detected in the nuclear fraction, although all centrin isoforms co-exist in the centrosomes of animal cells (14). The nuclear function of centrins has not yet been explored.
Xeroderma pigmentosum (XP) is a hereditary disease characterized by photosensitivity, a high incidence of sunlight-induced skin cancer, and in some cases, neurological complications. XP patient cells can be classified into seven different genetic complementation groups (XP-A to XP-G), all of which are defective in nucleotide excision repair (NER) (reviewed in Ref. 15), and into one variant group, XP-V, which is defective in translesion DNA synthesis (16,17). NER is a versatile and universal DNA repair pathway that can eliminate most types of the lesions from DNA. NER reaction consists of four steps: 1) damage recognition, 2) excision of the damaged DNA by creating incisions on both sides of the lesion, 3) gap-filling by DNA polymerase activity, and 4) ligation (18). A group of recent studies revealed that XP gene products are involved in the early steps of the human NER reaction, before the gap-filling reaction begins (reviewed in Refs. 19 and 20). The DNA excision reaction can be reconstituted with six factors: XPA, the XPC complex, XPF-ERCC1, XPG, general transcription factor TFIIH, and replication protein A (RPA) on naked DNA (21,22) or minichromosomes (23). 24 -32 base oligonucleotides containing the lesion are excised from duplex DNA in the cell-free NER reaction.
The XPC complex acts as a key component of global genome nucleotide excision repair (GGR), a NER subpathway, by functioning as the initial damage detector (24). This complex was isolated as a heterodimeric complex and consists of the XPC gene product (XPC) and the human homologue of Rad23 B (HR23B) protein (25). Further biochemical analyses showed that the XPC protein by itself preferentially binds to damaged DNA and single-stranded DNA (26) and elicits the translocation of another NER factor, TFIIH, to damaged DNA (27). The second subunit of the XPC complex, HR23B stimulates the in vitro NER reaction only in the presence of XPC (28), and this stimulatory activity depends on the region of HR23B that mediates interaction with XPC (29). Further analyses showed that the association of HR23B with XPC was important for its ability to stimulate NER (30). However, the mechanism of NER stimulation by HR23B is still poorly understood (31).
In this article we report that the XPC complex contains another factor, CEN2, that is recruited via interaction with XPC. During the analyses of CEN2, we found that NER stimulatory activity of HR23B is due to stabilization of XPC protein.
We show that CEN2 and HR23B cooperatively stabilize XPC and thereby stimulate NER in vitro. Thus, we conclude that CEN2 is a novel NER factor and suspect that the XPC complex may be involved not only in DNA repair but also in cell division.

EXPERIMENTAL PROCEDURES
Cell Culture and Cell-free Extracts-Lymphoblastoid cells (GM2248B, a kind gift of Dr. J. H. J. Hoeijmakers) from an XPC patient (XP3BE) were cultured at 37°C with 5% CO 2 in RPMI 1640 supplemented with 15% fetal bovine serum. For the in vitro NER assay, whole cell extracts were prepared as described previously (32,33). HeLa cells were cultured in spinner flasks and harvested for protein purification as described previously (25).
Protein Sequencing-Protein bands were excised from the Coomassie Brilliant Blue-stained gel and then digested with Achromobacter lyticus protease I (35) by an in-gel digestion method (36). Peptides extracted from the gel were separated by reverse phase high performance liquid chromatography using a Bakerbond octyl column (J. T. Baker (4.6 ϫ 250 mm)) with a linear gradient of 0 -60% acetonitrile in 0.1% trifluoroacetic acid for 1 h. Each peptide fragment was subjected to automated Edman degradation on a protein sequencer PPSQ-10 (Shimadzu). The amino acid sequences were ELGENLTDEELQEMIDEADRDG-DGEV and QEIREAFDLFDADGTGTIDVK. These two sequences are identical to deduced amino acid sequences of CEN2 but distinct from those of CEN1 and CEN3.
Glycerol Density Gradient Centrifugation-A portion (40 l) of the MonoQ fraction 22 was layered onto a 15-40% (v/v) glycerol gradient (1.2 ml) in buffer 2 (0.3 M KCl, 20 mM potassium phosphate (pH 7.5), 0.1 mM EDTA, 0.01% Triton X-100, 1 mM DTT, 0.25 mM PMSF, 0.2 g/ml aprotinin, 0.2 g/ml leupeptin, 0.1 g/ml antipain, and 50 M EGTA). Centrifugation was carried out in a Beckman TLS55 rotor at 194,000 ϫ g for 16 h at 2°C, and fractions (60 l) were collected from the top of the gradient. An identical gradient containing marker proteins was run at the same time. The markers were detected by SDS-PAGE followed by staining with Coomassie Brilliant Blue.
Construction of Plasmids for Protein Expression in Escherichia coli-The human CEN2-coding sequence was obtained by PCR with firststrand cDNA synthesized using HeLa poly(A)ϩ RNA as template. The sequences of the oligonucleotides used were 5Ј-GTACACGTCGGTTGC-CTAAC-3Ј and 5Ј-TTCTTCACGCTTGTGTGCTC-3Ј. The PCR products were further amplified by a second PCR using oligonucleotides 5Ј-CCTTTGACCATGGCCTCCAAC-3Ј and 5Ј-AGAGAATTCTGATCTTA-ATAGAGGC-3Ј, and then the second PCR products were cloned into pGEM-T Easy vector (Promega) to generate pGEM-CEN2. As a result of the second PCR, an NcoI site was introduced into the CEN2 cDNA that allowed the isolation of CEN2 cDNA as a 550-base pair fragment after NcoI digestion of pGEM-CEN2. The fragment was inserted into the NcoI site of E. coli expression vector, pET24d (Novagen), to generate pET24-CEN2. The stop codon in pET24-CEN2 was converted to an XhoI site by site-directed mutagenesis using the oligonucleotide 5Ј-GAGAAT-TCTGCTCGAGATAGAGGCTGGTC-3Ј. This was performed with the reagent Mutan-K (Takara Shuzo). The resulting plasmid was digested by XhoI and self-ligated to fuse the C terminus of CEN2 in-frame with eight amino acids (LEHHHHHH). This construct was designated pET24-CEN2H. All constructs were verified by DNA sequencing using an ALF-Red DNA sequencer (Amersham Pharmacia Biotech).
Expression and Purification of Recombinant CEN2 Proteins-Recombinant nontagged CEN2 protein was expressed and purified essentially according to the previous methods of Baron et al. (37) and Wiech et al. (38). Ten milliliters of a fresh, full-grown culture of E. coli BL21 (DE3) carrying pET24-CEN2 was inoculated into 1 liter of Super Broth medium (5 g of NaCl, 32 g of Bacto-trypton, and 20 g of yeast extract/liter (pH 7.0)) containing 50 g/ml of kanamycin. Cells were cultured at 30°C with vigorous shaking to an absorbance at 600 nm of 0.4, and then synthesis of the recombinant protein was induced by further incubation in the presence of 0.5 mM isopropylthioglucoside for 2 h at 30°C. The cell pellet was washed with ice-cold 10% glycerol, suspended in lysis buffer (50 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, 2 mM DTT, 0.5 mM PMSF, 0.4 g/ml aprotinin, 0.4 g/ml leupeptin, 0.2 g/ml antipain 0.1 mM EGTA, and 10 g/ml lysozyme), and disrupted by freezing and thawing. The supernatant (129 mg of protein), obtained by centrifugation, was loaded onto a HiTrap DEAE column (Amersham Pharmacia Biotech (5 ml)) that was equilibrated with buffer 4 (50 mM Tris, 34 mM HCl, and 1 mM DTT). The column was washed with the equilibration buffer, and the adsorbed proteins were eluted with 30 ml of a linear salt gradient of 0 -0.2 M NaCl in buffer 4. CEN2 eluted around 0.2 M NaCl was collected and dialyzed against buffer 4 containing 1.5 M NaCl, 5 mM CaCl 2 , and 10% glycerol. The dialysate was incubated in a water bath at 80°C for 10 min and chilled on ice. After centrifugation to remove denatured proteins, the supernatant (25.2 mg) was loaded onto a phenyl-Superose HR5/5 column (Amersham Pharmacia Biotech (1 ml)) equilibrated with the same buffer used for denaturation. The column was washed with the equilibration buffer, and the adsorbed proteins were eluted with buffer 5 (10 mM Tris, 6.8 mM HCl, 5 mM EGTA, 1 mM DTT, and 10% glycerol). Because a portion of CEN2 was recovered in the flow-through fraction, the fraction was applied to and eluted from the column three times. The eluates (2.6 mg) were combined, dialyzed against buffer 4, and loaded onto a MonoQ HR5/5 column (1 ml) equilibrated with buffer 4. After washing the column with buffer 4 containing 0.1 M NaCl, the adsorbed proteins were eluted with 10 ml of a linear salt gradient of 0.1-0.4 M NaCl. CEN2 (2.3 mg) eluted around 0.25 M NaCl was pooled and stored at Ϫ80°C.
Recombinant hexa-histidine-tagged CEN2 (CEN2-His) was expressed in E. coli cells transformed with pET24-CEN2H as described above. The cells were disrupted by sonication in buffer 6 (0.3 M NaCl, 50 mM Tris-HCl (pH 7.5), 0.5 mM PMSF, 0.1% Nonidet P-40, 0.4 g/ml aprotinin, 0.4 g/ml leupeptin, 0.2 g/ml antipain, and 0.1 mM EGTA) containing 10 g/ml lysozyme. The lysate was cleared by centrifugation, and the supernatant was loaded onto a nickel-chelating Sepharose column (Amersham Pharmacia Biotech (1 ml)) equilibrated with buffer 6 containing 20 mM imidazole. The column was washed with the equilibration buffer, and the adsorbed proteins were eluted with a linear gradient of 20 -250 mM imidazole. The protein-rich fractions were pooled and dialyzed against buffer 4. The dialysate was loaded onto a Mono Q HR5/5 column equilibrated with buffer 4. The column was washed with equilibration buffer, and the adsorbed proteins were eluted with a linear salt gradient of 0 -0.4 M NaCl. CEN2-His eluted around 0.25 M NaCl was pooled and stored at Ϫ80°C. Throughout the purification, column chromatographies were performed with fast protein liquid chromatography system, and the proteins were analyzed by SDS-PAGE.
Heat Inactivation Assay-To examine the stability of XPC protein, XPC was preincubated at various temperatures for 1 h in preincubation buffer (0.1 M NaCl, 25 mM Tris-HCl (pH 7.5), 10% glycerol, 0.01% Triton X-100, 1 mM DTT, and 0.25 mM PMSF) in the presence or absence of CEN2 and/or HR23B. The residual activity was analyzed by the XPCcomplementing assay and the damaged DNA binding assay. The XPC concentration during the preincubation was 12.5 nM and 50 nM for the XPC-complementing assay and the damaged DNA binding assay, respectively. In both cases, the molar ratios of CEN2 and HR23B were 10 times higher than that of XPC. In the 1-h preincubation, two tubes, the first one of which (tube A) contained XPC with or without CEN2 and/or HR23B and the second one of which (tube B) contained the proteins missing from the first tube, were incubated in a water bath. For the XPC-complementing assay, 2 l of solution from tube A was combined with 2 l of solution from tube B, added to the reaction mixture containing the SV40 minichromosome, and incubated at 30°C for 3 h under standard conditions. For the damaged DNA binding assay, 4 l of solution from tube A was combined with 4 l of solution from tube B, added to the reaction mixture containing damaged or undamaged DNA, and incubated at 30°C for 30 min. These procedures are summarized in Figs. 5A and 6A. At these amounts of XPC, both XPC-complementing activity and damaged DNA binding activity were dependent on XPC but not on CEN2 or HR23B, and the activity in both assays was not saturating but proportional to the amount of added XPC (data not shown). Thus, in both assays, the only limiting condition was the amount of XPC.
Cell-free NER Assay-For the XPC complementation assay, UVirradiated (200 J/m 2 ) simian virus 40 (SV40) minichromosome (0.3 g of DNA) was reacted with 80 g of GM2248B whole cell extract and/or other purified proteins (HeLa XPC-correcting fraction, XPC, CEN2, and HR23B) in the standard reaction mixture (39). After incubation at 30°C for 3 h, DNA was purified from the reaction mixture, linearized by EcoRI digestion, and then electrophoresed in a 1% agarose gel. The gel was dried and autoradiographed with Fuji RX x-ray films. DNA repair synthesis was quantified using the Fuji BAS2500 Bio-Imaging analyzer.
For the NER reaction with purified proteins, the standard reaction mixture (20 l) was as indicated above except that radiolabeled dCTP and crude cell extracts were omitted. The proteins used were 540 ng of RPA-His, 10 ng of XPA-His (23), 6 ng of XPF-ERCC1-His (23), 10 ng of XPG (23), and 880 ng of partially purified TFIIH, and/or the indicated amounts of recombinant proteins composing the XPC complex (28,29). After incubation at 30°C for 3 h, the DNA was purified from the reaction mixture, and the gaps on the DNA, generated by the excision of the damaged DNA, were filled in by gap-filling DNA synthesis with radioactive dNTPs (23).
DNA was purified again, linearized by EcoRI digestion, subjected to 1% agarose gel electrophoresis, and analyzed as described above.
The recombinant RPA trimeric complex carrying a hexa-histidine tag at the C-terminal end of the p32 subunit (RPA-His) was expressed in bacterial cells and purified by three column chromatographic steps (nickel-chelating Sepharose, single-stranded DNA-agarose, and MonoQ). The activity of RPA-His was indistinguishable from that of the nontagged form both in the in vitro NER assay and in the in vitro replication assay. 2 HeLa TFIIH was partially purified from HeLa nuclear extracts as far as the TSK-DEAE 5PW column step, corresponding to two column chromatographic steps before the final purification step (40). Using the previously purified fraction, we obtained results similar to those of the partially purified fraction (data not shown).
Damaged DNA Binding Assay-The damaged DNA binding assay was carried out in the standard reaction mixture containing 200 ng of pBS.XPC⌬ (500 J/m 2 UV-irradiated) and 200 ng of pHM14 (unirradiated) (24). After incubation at 30°C for 30 min, plasmid DNA bound to XPC was successively immunoprecipitated by protein A-Sepharose 4FF conjugated to anti-XPC antibody, washed with ice-cold immunoprecipitation wash buffer, eluted by 1% SDS treatment at 50°C for 1 h, purified, linearized by BamHI digestion, separated by 0.8% agarose gel electrophoresis, and blotted onto a Hybond-Nϩ membrane (Amersham Pharmacia Biotech) by alkali transfer (24). The blots were hybridized to 32 P-labeled pBluescript KS(ϩ) (Stratagene) sequences. Other Methods-SDS-PAGE was performed as described by Laemmli (41). For immunoblot analyses, proteins separated by SDS-PAGE were electrically transferred onto polyvinylidene difluoride membranes (Millipore Immobilon-P) and analyzed as described previously (23).
The anti-CEN2(rabbit) and anti-CSA antibodies were obtained by immunization of rabbits with recombinant CEN2 and histidine-tagged CSA, respectively, and purified by antigen-coupled affinity column chromatography from the antiserum. The CEN2(rat) antiserum was obtained by immunization of rats with recombinant CEN2. The anti-HR23B and anti-XPC affinity-purified antibodies were obtained as described previously (28).

XPC Forms a Trimeric Complex with CEN2 and HR23B-
The XPC complex was purified from HeLa nuclear extracts after four column chromatographic steps (Fig. 1A). In this 2 K. Tatsuta, T. Maekawa, and F. Hanaoka, unpublished observation. study, we obtained a fraction that was approximately 10 times more concentrated than previous preparations (25). The addition of the MonoQ fractions to XPC cell extracts showed significant recovery of NER synthesis, whereas the extracts alone did not show any NER synthesis (Fig. 1B). The elution profile of XPC-complementing activity (NER activity) correlated with the presence of XPC and HR23B proteins as detected by silver staining SDS-polyacrylamide gels containing samples from the elution peak (compare Fig. 1, B and C). However, several extra bands, which were not detected in the previous study, were observed in these fractions (Fig. 1C). Most of these could be accounted for XPC-derived degradation products and HR23A, another human homologue of S. cerevisiae Rad23, by immunoblot analyses (data not shown). However, protein sequencing analysis revealed that the 18-kDa protein was the centrosome protein centrin 2/caltractin 1 (CEN2).
CEN2 co-sedimented with both XPC and HR23B proteins in a complex with a sedimentation co-efficient of 6.2 S during glycerol density gradient centrifugation ( Fig. 2A). Using hydrodynamic methods (42), Shivji et al. (43) determined that the XPC-complementing complex is 160 kDa, in good agreement with the calculated mass of these proteins (169 kDa).
To confirm that the XPC-HR23B-CEN2 complex is naturally present in the cells, we performed co-immunoprecipitation analyses. CEN2 was co-precipitated from HeLa cell extracts by resin containing anti-XPC antibody or by anti-HR23B antibody, but not by anti-CSA antibody (Fig. 2B). Quantitative analyses revealed that almost 100% of CEN2 in the cell extract was co-precipitated with the anti-XPC antibody. Thus, most of the CEN2 in this extract was complexed to XPC. The anti-HR23B resin precipitated lower amounts of CEN2 than did the anti-XPC resin (Fig. 2B). This was also true for the anti-HR23A antibody, which could also precipitate low but nonetheless detectable amounts of CEN2 (data not shown). These results are consistent with the observation that only a small proportion of the XPC complex in the final purification step contained HR23A (Fig. 1C).
CEN2 Directly Interacts with XPC-To understand XPC complex formation, we studied the properties of complexes reconstituted from each component that were purified as a recombinant protein (see "Experimental Procedures"). CEN2-His was bound to nickel-chelating Sepharose resin, as detected by immunoblotting (Fig. 3, A and B), whereas neither XPC nor HR23B bound to the resin. In the presence of CEN2-His, however, XPC was retained on nickel-chelating resin (Fig. 3A) but HR23B was not (Fig. 3B), indicating that CEN2 interacts with XPC but not HR23B. CEN2 has four EF-hand domains that are known to be important for calcium binding. However, the XPC binding activity of the recombinant CEN2, which was purified in the absence of calcium, was not affected by the presence of 5 mM calcium (data not shown).
NER Activities of the XPC Complex Components-From the above studies, we surmised that CEN2 might have some function in NER because it is present in virtually all of the purified XPC complexes. To directly test the role of CEN2 in NER, we purified recombinant nontagged CEN2 as described under "Experimental Procedures" (Fig. 4A). NER activity was measured on UV-irradiated SV40 minichromosomes in the cell-free NER assay reconstituted with purified NER factors (23). As shown in Fig. 4, B and C, in reactions containing all factors except the XPC complex, the addition of recombinant XPC alone resulted in weak NER activity, as reported previously (30). In the absence of XPC, CEN2 alone, HR23B alone, or a mixture of CEN2 and HR23B indicated no significant NER activity. However, HR23B stimulated the NER activity of XPC, as previously reported (30). On the other hand, the addition of CEN2 either alone or in combination with HR23B failed to produce any significant effect on XPC-dependent NER activity in vitro.
HR23B Stabilizes XPC-As shown above, HR23B has an FIG. 2. CEN2 is part of the XPC-correcting complex. A, MonoQ fraction 22 (Fig. 1) was subjected to glycerol density gradient centrifugation, and portions (16 l) of the collected fractions were analyzed by 8 -18% SDS-PAGE followed by silver staining. The sedimentation positions of marker proteins in a parallel gradient are indicated. A portion of the MonoQ fraction (Q) was subjected to SDS-PAGE as a control. The molecular weights were obtained using a 10-kDa ladder marker (Life Technologies, Inc.). BSA, bovine serum albumin. B, co-immunoprecipitation analyses of the XPC complex from HeLa whole cell extracts. HeLa whole cell extracts (200 g) were incubated with protein A beads conjugated with affinity-purified anti-XPC, anti-HR23B, or anti-CSA antibody. The precipitates were subjected to 14% SDS-PAGE followed by immunoblotting with anti-CEN2(rat) antibody. Anti-CSA antibody was used for a negative control of the affinity-purified antibody. Ten ng of recombinant CEN2 protein was loaded onto the gel as a positive control. The recombinant CEN2 was prepared under a calcium-free condition, and it was reported that the migration profile of endogenous CEN2 is not affected by calcium during electrophoresis (12). Therefore, the presence of a doublet band could not be due to a calcium binding state of CEN2. It is rather suggested that the upper band of the doublet is full-length CEN2 and that the lower band is a truncated version of CEN2 or another centrin isoform closely related to CEN2. Asterisks indicate the protein that cross-reacted with anti-CEN2(rat) serum but not with affinity-purified anti-CEN2(rabbit) (data not shown). Molecular weights were calculated from a set of prestained molecular weight standards (Life Technologies, Inc.). XPC-dependent NER stimulatory activity in defined conditions. Though the mechanism of NER stimulation by HR23B has not been well understood, several observations suggested that HR23B might function by stabilizing XPC. Previous studies showed that the XPC binding region of HR23B was necessary and sufficient for the stimulation of XPC-dependent NER activity in a semi-reconstituted reaction (29) and that the interaction of HR23B with XPC was important for the stimulation of NER by HR23B (30). Recently, we observed that the capacity of HR23B to stimulate XPC-dependent NER activity was greater during long incubations than short ones (data not shown). Taken together, these observations suggested that HR23B might function by protecting XPC protein from heat denaturation during the reaction.
To examine this possibility more directly, we designed a heat inactivation assay. As shown in Fig. 5A, XPC, CEN2, and HR23B were preincubated at various temperatures either alone or in various combinations. After preincubation, the proteins were combined and added to the XP-C whole cell extracts, and NER activity was measured (Fig. 5, B and C). When XPC was preincubated in the absence of either CEN2 and HR23B, ϳ75% activity was lost during a 1-h incubation at temperatures Ն37°C. With a preincubation in the presence of HR23B, XPC retained most of its activity at 37°C and showed a certain activity at higher temperatures. After a 1-h preincubation at 30°C, a small but reproducible increase in XPC activity was observed in the presence of HR23B. These observations strongly suggest that the NER stimulatory activity of HR23B is due mainly to its ability to stabilize XPC. CEN2 alone did not show a significant effect on XPC stabilization. During the 42°C preincubation, however, CEN2 increased the stability of XPC protein when HR23B was concomitantly present in the preincubation mixture, suggesting that CEN2 may help HR23B to stabilize XPC (see below).
To determine the reason for the loss of XPC function during the preincubation, we examined XPC binding to damaged DNA after preincubation (Fig. 6A). After preincubation, the components of the XPC complex were incubated with both UV-irradiated DNA and unirradiated DNA in reaction mixtures, and the DNA-XPC complexes were immunoprecipitated with resin conjugated to XPC antibodies (see "Experimental Procedures"). The DNA present in the precipitated complexes was purified and analyzed by Southern blotting (Fig. 6, B and C). The results showed that specific binding activity of XPC to damaged DNA was more resistant to preincubation at high temperatures in the presence of HR23B than in its absence. Moreover, CEN2 had a small but reproducible effect on XPC stabilization activity at 42°C when combined with HR23B, whereas CEN2 alone had no significant effect on XPC stabilization. Finally, there was good correlation between the heat inactivation profiles of XPC as assayed by the NER assay and the DNA binding assay (compare Fig. 5C and 6C). We could not exclude the possibility that highly denatured XPC protein was not co-precipitated as efficiently as native XPC by the anti-XPC antibody conjugated resin. Therefore, the precipitated DNA may not reflect the real DNA binding activity of XPC for damaged DNA. However, we examined the DNA binding affinity of XPC for either damaged or nondamaged DNA in a filter binding assay and found that DNA binding activity decreased with increasing preincubation temperature. We also observed that CEN2 and HR23B prevented heat inactivation of XPC as measured by a filter binding assay (data not shown). Thus, CEN2 and HR23B are important for the stability of XPC protein and, especially, for the preservation of the binding activity of XPC for damaged DNA, which is essential for NER. CEN2 Stabilizes XPC in Co-operation with HR23B-As shown in Figs. 5 and 6, CEN2 enhanced, albeit weakly, the ability of HR23B to stabilize XPC. To confirm this point, we performed the heat inactivation assay at different temperatures. As shown in Fig. 7, XPC preincubated only with HR23B Recombinant XPC and HB23B were purified as described previously (24,40), and CEN2 was purified as described under "Experimental Procedures." A, Coomassie Brilliant Blue staining of CEN2 (0.5 g). After incubation at 30°C for 3 h, DNA was purified, and the gaps generated by the excision of damaged DNA were labeled by T4 DNA polymerase with radioactive dNTPs. DNA was purified again, linearized by EcoRI digestion, and subjected to 1% agarose gel electrophoresis. B, an autoradiogram of the gel. C, incorporation of dCMP into SV40 minichromosomes was calculated, and the results are shown on the graph. Symbols: diamond, titration of XPC in the absence of both CEN2 and HR23B; triangle, in the presence of CEN2; square, in the presence of HR23B; circle, in the presence of both CEN2 and HR23B. Averages and experimental errors were calculated from two independent experiments. and lost more than 55% of its activity at 42°C, 89% of its activity at 44°C, and all of its activity at 46°C compared with the activity of the control incubated at 0°C. On the other hand, XPC preincubated with both CEN2 and HR23B retained nearly 69% of its activity at 42°C, 47% of its activity at 44°C, and 7% of its activity at 46°C compared with the activity of the control incubated at 0°C. These results further support our conclusion that CEN2 contributes to NER by binding to and stabilizing XPC.  30,37,42, and 50°C) for 1 h, and residual NER activity was measured by the XPC-complementing assay. XPC (12.5 nM) with or without 125 nM CEN2 and/or 125 nM HR23B were added to one tube, the omitted proteins were added to another tube, and both tubes were preincubated as a set. After preincubation, a portion (2 l) of solution from each tube was combined, and the residual NER activity of XPC was measured in the complementing assay. A, schematic representation for the heat inactivation and XPC-complementing assays. The XPC-complementing assay was carried out under standard conditions using UV-irradiated SV40 minichromosomes as described under "Experimental Procedures." B, an autoradiogram of the gel. C, incorporation of dCMP into SV40 minichromosomes was calculated, and the results are represented in graph form. Symbols: diamond, XPC was preincubated in the absence of both CEN2 and HR23B; triangle, in the presence of CEN2; square, in the presence of HR23B; circle, in the presence of both CEN2 and HR23B. Averages and experimental errors were calculated from two independent experiments.
FIG. 6. Loss of damaged DNA binding activity by heat-inactivation. Various combinations of recombinant proteins of the XPC complex were preincubated at various temperatures (0, 30, 37, 42, and 50°C) for 1 h, and residual binding activity to damaged DNA was measured. XPC (50 nM) with or without 500 nM CEN2 and/or 500 nM HR23B were placed in one tube, and the omitted proteins were placed in a second tube. After preincubation, a portion (4 l) of solution from each tube was combined, and the residual amount of XPC binding activity to damaged DNA was measured. A, schematic representation of the heat inactivation and DNA binding assays. The damaged DNA binding assay was carried out under standard conditions (see "Experimental Procedures") using UV-irradiated (500 J/m 2 ) (UVϩ) or unirradiated (UVϪ) plasmid DNA. After incubation at 30°C for 30 min, plasmid DNA bound to XPC was co-immunoprecipitated with anti-XPC antibody. Plasmid DNA purified from the precipitates was linearized by BamHI digestion, subjected to 0.8% agarose gel electrophoresis, and analyzed by Southern blot. B, autoradiogram of the blot. C, the percentage amounts of DNA recovered by the anti-XPC antibody. Symbols: diamond, XPC was preincubated in the absence of both CEN2 and HR23B; triangle, in the presence of CEN2; square, in the presence of HR23B; circle, in the presence of both CEN2 and HR23B. Open symbols represent the percentage amounts of UV-irradiated plasmid bound to XPC, and solid symbols represent those of unirradiated DNA. Averages and experimental errors were calculated from two independent experiments.

DISCUSSION
Roles of CEN2 in the XPC Complex-In this study, we have demonstrated that CEN2 is a component of the XPC complex (Figs. 1 and 2) that directly interacts with XPC (Fig. 3). In contrast, we could detect neither XPC nor HR23B in a partially purified centrosome fraction from HeLa cells (data not shown). These observations strongly suggest that CEN2 has a role in NER in human nuclei that is distinct from its function in the centrosome. Actually, CEN2 contributed to XPC stability, thereby increasing the efficiency of NER (Fig. 5), whereas CEN2 did not show significant activity under standard conditions for in vitro NER. CEN2 stabilized XPC in vitro in cooperation with HR23B. Although the direct interaction was not observed between CEN2 and HR23B, it is possible that both proteins interact on XPC protein. It is also likely that the binding of CEN2 to XPC confers a proper interaction between XPC and HR23B. We suspect that CEN2 might be important for XPC complex formation in vivo, although CEN2 contributes relatively little to XPC stabilization in vitro (Fig. 5-7). Unprotected XPC is quite labile (Figs. 5 and 6) and susceptible to degradation during purification (Fig. 1C). Consequently, we believe that XPC is stabilized in human cells by virtue of association with CEN2 and HR23B.
Moreover, the complex reconstituted from the purified recombinant proteins showed lower activity in in vitro NER than the XPC complex purified from HeLa cells (data not shown). From our previous studies, we know this is mainly due to incomplete complex assembly in vitro (30). To examine the role of CEN2 in complex assembly, we purified XPC complexes from insect cells infected with recombinant baculoviruses expressing XPC and HR23B with or without CEN2 and measured XP-C complementation activity. The XPC complexes containing both CEN2 and HR23B showed slightly higher NER activity and solubility than XPC complexes containing HR23B alone. 3 These observations support the results shown in this paper and strongly suggest that CEN2 plays a role in XPC complex assembly within cells.
Roles of HR23B in the XPC Complex-During the analyses of CEN2 function in in vitro NER, we found that HR23B stimulates NER mainly by stabilizing XPC and that CEN2 supports HR23B in this function. This is the first paper that clearly shows how HR23B stimulates in vitro NER by XPC-dependent manner, but we cannot rule out the possibility that HR23B has additional functions for NER in vivo (discussed below).
Studies in S. cerevisiae led to the idea that Rad23, the yeast counterpart of HR23B(A), has a function carried out by its N terminus ubiquitin-like (UbL) domain via interaction with proteasome. Previous studies showed that the UbL domains of Rad23 are important for cell survival after UV irradiation (44) and that Rad23 interacts with the 26 S proteasome through its UbL domain (45). Recently, two completely different models have been proposed to explain the function of the UbL region of HR23B. Russell et al. (46) show that the proteasome 19 S subunit plays a direct role in NER both in vitro and in vivo and that the ATPase activity, but not the proteolytic activity of the 19 S subunit, is essential for NER. They suggest that the molecular chaperone function, but not the proteolytic function, of the 19 S subunit is important for NER. However, Ortolan et al. (47) find that Rad23 can inhibit the degradation of proteasome substrates and show that a proteasome subunit mutation can partially rescue the UV sensitivity of a Rad23 deletion mutant. Taken together, they proposed that the UbL domain of Rad23 negatively regulates the degradation of repair factors after DNA damage, which may promote NER events. In neither report is there direct evidence of how an interaction between Rad23 and the proteasome acts to promote survival after UV irradiation. In human cells, a protein-protein interaction between HR23B and the 26S proteasome was observed (48), indicating that the above-mentioned mechanisms might be conserved in human cells.
Although most of the XPC-complementing complex contains CEN2 and HR23B, we also purified an XPC-HR23A-CEN2 complex (Fig. 1C) that was detected by co-immunoprecipitation (data not shown). HR23A was first identified as another human homologue of Rad23 by PCR methods (25) and was subsequently shown to be functionally redundant to HR23B in NER both in the cell-free system (30) and in mouse embryonic fibroblast cells derived from knock out mice. 4 Thus, we expect that HR23A can stabilize in a manner XPC similar to HR23B. However, the functional differences between these two isoforms and the mechanism that generates the alternative complex still remain unclear.
Centrosome Duplication and the XPC-HR23B-CEN2 Complex-Genetic studies in yeast suggest that XPC complex might have a role in centrosome duplication in human cells. In S. cerevisiae, mutations in the gene encoding a counterpart of human centrin, CDC31, were also isolated as DSK1, dominant suppressors of the Kar1-⌬17 allele, which produces defects in SPB duplication (49). Another dominant suppressor of Kar1-⌬17 is DSK2, which encodes a protein that is very similar to Rad23 within the N-terminal UbL domain. The double deletion mutant, ⌬rad23 ⌬dsk2 shows temperature-sensitive lethality 3  were preincubated with or without CEN2 (500 nM), and residual XPC binding activity to damaged DNA was measured. Preincubation was carried out at various temperatures as indicated. A, autoradiogram of the blot. B, the percentages of DNA recovered by the anti-XPC antibody. Symbols: square, XPC was preincubated in the presence of HR23B; circle, in the presence of both CEN2 and HR23B. Open symbols represent the percentage amounts of UV-irradiated plasmid bound to XPC, and solid symbols represent those of unirradiated DNA. Averages and experimental errors were calculated from two independent experiments. caused by defects in SPB duplication, although neither UbL protein is essential for cell growth (50). These observations suggest that RAD23 has a role that is redundant to DSK2 in SPB duplication and that RAD23 engages in an indirect genetic interaction with CDC31. Our findings appear to support the yeast genetic studies, and we suspect that Rad4 (the XPC counterpart in S. cerevisiae) interacts with Cdc31. Therefore, the Rad4 complex might have a role in SPB duplication in yeast.
In human cells, the intracellular localization of XPC changes dramatically during mitosis (51). During anaphase and telophase, the XPC protein becomes preferentially associated with chromatin, whereas it is distributed throughout the cell during metaphase. Thus, we can imagine that localization of XPC (Rad4 in yeast) complex is strictly regulated via CEN2 (Cdc31 in yeast) during mitosis and that the failure to regulate the Rad4 complex might cause some defect in SPB duplication in yeast. Further studies may show that XPC cancer-prone phenotypes are partly due to defects in the mechanism that couples cell division to NER and/or in the repair reaction involving the XPC-HR23B-CEN2 complex.