Cullin 4A-mediated Proteolysis of DDB2 Protein at DNA Damage Sites Regulates in Vivo Lesion Recognition by XPC*

Xeroderma pigmentosum (XP) complementation group E gene product, damaged DNA-binding protein 2 (DDB2), is a subunit of the DDB heterodimeric protein complex with high specificity for binding to a variety of DNA helix-distorting lesions. DDB is believed to play a role in the initial step of damage recognition in mammalian nucleotide excision repair (NER) of ultraviolet light (UV)-induced photolesions. It has been shown that DDB2 is rapidly degraded after cellular UV irradiation. However, the relevance of DDB2 degradation to its functionality in NER is still unknown. Here, we have provided evidence that Cullin 4A (CUL-4A), a key component of CUL-4A-based ubiquitin ligase, mediates DDB2 degradation at the damage sites and regulates the recruitment of XPC and the repair of cyclobutane pyrimidine dimers. We have shown that CUL-4A can be identified in a UV-responsive protein complex containing both DDB subunits. CUL-4A was visualized in localized UV-irradiated sites together with DDB2 and XPC. Degradation of DDB2 could be blocked by silencing CUL-4A using small interference RNA or by treating cells with proteasome inhibitor MG132. This blockage resulted in prolonged retention of DDB2 at the subnuclear DNA damage foci within micropore irradiated cells. Knock down of CUL-4A also decreased recruitment of the damage recognition factor, XPC, to the damaged foci and concomitantly reduced the removal of cyclobutane pyrimidine dimers from the entire genome. These results suggest that CUL-4A mediates the proteolytic degradation of DDB2 and that this degradation event, initiated at the lesion sites, regulates damage recognition by XPC during the early steps of NER.

The genome of all living organisms is constantly subjected to damage by many kinds of physical and chemical agents. Prompt repair of DNA damage enables cells to overcome genotoxicity and retain normal cellular functions. Nucleotide excision repair (NER) 3 is a versatile DNA repair pathway that eliminates a wide variety of helix-distorting DNA lesions, including UV-induced cyclobutane pyrimidine dimers (CPD) and 6 -4 pyrimidine-pyrimidone photoproducts . Mammalian NER consists of two distinct subpathways: global genomic repair (GGR), which operates throughout the genome, and transcription-coupled repair (TCR), which works on damage within transcribed DNA strands of transcriptionally active genes (1,2). Loss or impairment of NER is associated with several rare autosomal recessive genetic disorders, e.g. Xeroderma pigmentosum (XP) and Cockayne syndrome (3). XP syndrome is characterized by hypersensitivity to sunlight and a predisposition to skin cancer. Seven XP complementation groups have been identified, and all of the corresponding genes, from XPA to XPG, have been cloned (4).
Mutation in XPC results in the loss of GGR but not TCR, suggesting that the XPC gene encodes a damage recognition factor involved in GGR (5). The XPC protein binds in vivo to hHR23B, which is one of the two mammalian homologs of Saccharomyces cerevisiae Rad23 protein (6,7). Biochemical studies have revealed that XPC-hHR23B is a structure-specific DNA binding factor (8) and it is one of the six core factors (Replication protein A, XPA, XPC-hHR23B, XPG, ERCC1-XPF, Transcription factor II H) essential for damage recognition and dual incision in vitro (9). Accumulating evidence indicates that the XPC protein plays an essential role in the damage recognition process of GGR (10 -12). Like XPC cells, GGR is impaired in XP-E cells whereas TCR remains unaffected (13). XP-E cells lack the damaged DNA binding activity of the damaged DNA-binding protein (DDB), which is composed of the DDB1 (or p127) and the DDB2 (or p48) subunits (14). Mutation in the DDB2 gene is responsible for the phenotypic features of XP-E cells (15)(16)(17)(18)(19)(20). DDB has a much higher affinity and specificity for damaged DNA than XPC, especially with regards to binding to 6 -4PP (21)(22)(23)(24).
Despite its capability of binding to damaged DNA, DDB is not required for cell-free NER reconstituted in vitro (25)(26)(27)(28). However, GGR of CPD is profoundly reduced in XP-E cells, whereas the repair of 6 -4PP is only moderately impaired (16). Moreover, ectopic expression of human DDB2 in Chinese hamster cells enhances the removal of CPD from the genome and suppresses UV-induced mutagenesis (29). These findings suggest that DDB plays an important role in the recognition of CPD in vivo. DDB2, like XPC, accumulates at DNA damage sites immediately after UV irradiation (30,31). Recent evidence further indicates that DDB activates the recruitment of XPC to CPD in vivo (32). Besides acting on CPD, DDB mediates efficient targeting of XPC to 6 -4PP and accelerates its repair (33). Thus, DDB can be considered as the initial damage recognition factor for UV-induced photolesions.
Although the precise role of DDB in NER remains unclear, it has been known that DDB2 is regulated at the transcriptional level by tumor suppressor p53 (13) and at the posttranslational level via the ubiquitin (Ub)-proteasome system (34 -37). In eukaryotic cells, this pathway mediates a selective degradation of many cellular proteins (38). The target proteins are labeled by covalent attachment of multiple moieties of Ub, a highly conserved 76-residue eukaryotic protein. The poly(Ub)conjugated substrates are recognized and destroyed by the proteasome. The attachment of Ub to a protein substrate is preceded by a cascade reaction of three enzymes, namely, the E1 Ub-activating enzyme, the E2 Ub-conjugating enzyme, and the E3 Ub-protein ligase. Cullin 4A (CUL-4A)-based Ub ligase appears to be responsible for UV-induced DDB2 degradation. CUL-4A belongs to a cullin family of proteins that assemble into multiple E3 ligase complexes (39,40). It has been demonstrated that CUL-4A associates with DDB1 and overexpression of CUL-4A increases the ubiquitylation and the decay rate of DDB2 (35,42). The DDB-CUL-4A complex can directly ubiquitylate DDB2 in vitro (37). These results, together with immediate recruitment of DDB2 to DNA damage and rapid turnover of DDB2 after UV irradiation (19,36), suggest that the fate of DDB2 engaged in damage recognition is tightly controlled by Ub-mediated proteolysis.
Here, we have addressed the functionality (e.g. damage recognition and repair of UV-induced lesions) of DDB2 ubiquitylation and subsequent degradation mediated by CUL-4A-based E3 ligase. We present evidence that CUL-4A is an essential component for Ub-mediated proteolysis of DDB2 and is being physically recruited to DNA damage sites in chromatin. Degradation of DDB2 regulates the recruitment of XPC to DNA damage and subsequently the repair of UV-induced CPD. Our results support the idea that DDB2 degradation is integral to DNA damage handover from DDB to XPC during the initial steps of NER.

EXPERIMENTAL PROCEDURES
Cell Lines-The normal human fibroblasts (OSU-2) were established in culture in our laboratory as described by Venkatachalam et al. (43). HeLa cells stably expressing N-terminal FLAG-HA epitope-tagged DDB2 (HeLa-DDB2) were kindly provided by Dr. Yoshihiro Nakatani (Dana-Farber Cancer Institute, Boston, MA). HeLa-DDB2 cells stably expressing c-Myc epitope-tagged CUL-4A (HeLa-DCH) and V5-His epitope-tagged XPC (HeLa-XPC) were established in our laboratory. For expression of CUL-4A in HeLa-DDB2 cells, the Myc3-tagged CUL-4A expression construct (a gift from Dr. Yue Xiong, University of North Carolina) was introduced into the HeLa-DDB2 cells and the transfectants were selected with G418. For expression of V5-His-tagged XPC, the XPC cDNA were inserted into pcDNA3.1/V5-His vector (Invitrogen). The expression constructs were transfected into the Hela-DDB2 cells, and the stable transfectants were established through subcloning of the transfected cells. In both cases, the stably transfected cells were further subcloned by a single cell dilution and identified by Western blot analysis. The DNA transfection was performed using FuGENE 6 transfection reagents (Roche Applied Science) according to the manufacturer's recommendation. All cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, antibiotics, and 500 g/ml of G418 (for selection and maintenance of the transgenic cell lines) at 37°C in a humidified atmosphere of 5% CO 2 . For experiments to assess DNA damage and repair, the monolayer cells were grown to confluence and then incubated with serum-free medium for 24 h.
UV Irradiation-The cells ready for UV-C irradiation were washed twice with phosphate-buffered saline (PBS). The UV-C light of 254 nm was delivered from a germicidal lamp at a dose rate of 0.5 J/m 2 /s as measured by a Kettering model 65 radiometer (Cole-Palmer Instrument Co., Vernon Hills, IL). For local UV irradiation, the cells were grown for 24 -48 h to ϳ70% confluence on glass coverslips. The medium was aspirated, and the cells were washed with PBS. Prior to UV irradiation, an isopore polycarbonate filter (Millipore, Bedford, MA) with a pore size of 3, 5, or 8 m diameter, was placed on top of the cell monolayer. The filter-covered coverslips were irradiated with the desired UV doses. The filter was then gently removed, and the cells were processed immediately or maintained in a suitable medium for the desired period and processed thereafter.
Immunoprecipitation of the DDB2 Complex and Western Blotting-The HeLa-DCH cells grown to confluence were UV irradiated at 25 J/m 2 and incubated for the indicated periods in fresh medium. The cells were washed twice with PBS and fixed with 1% formaldehyde (final concentration in PBS) at room temperature for 10 min, followed by addition of glycine to a final concentration of 125 mM and incubation for 5 min to quench the cross-linking. After two additional washes with cold PBS, the cells were scraped from dishes and collected in radioimmune precipitation assay buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS) containing 10 g/ml of leupeptin, 10 g/ml of pepstatin, and 1 mM phenylmethylsulfonyl fluoride. The cell extracts were made by sonication of the cell lysates on ice to break the DNA to a 100 -500-bp fragment. The DDB2 complex in cell extracts was purified by immunoprecipitation with anti-HA affinity matrix (Roche Applied Science) followed by recovery with anti-FLAG gel (Sigma). Briefly, the cell extracts containing ϳ2 mg of protein were incubated with 25 l of anti-HA affinity matrix in radioimmune precipitation buffer at 4°C overnight. The matrix beads were washed, and the bound proteins were eluted three times each with 25 l of elution buffer (20 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.1 mM EDTA) containing 1 mg/ml of HA peptide at room temperature for 10 min. The eluted proteins were pooled and diluted in radioimmune precipitation buffer and further immunoprecipitated with 25 l of anti-FLAG gel. The bound proteins were recovered by dissolving the immunoprecipitates in IP elution buffer (1% SDS, 0.1 M NaHCO 3 ). To reverse cross-link, the complex was incubated at 65°C for 5 h after adding 0.2 M NaCl to the eluents. For Western blotting, the cell extracts were made from either transfected or non-transfected cells by boiling the cell lysates for 10 min in a sample buffer (2% sodium dodecyl sulfate, 10% glycerol, 100 mM dithiothreitol, and a protease inhibitor mixture in 62 mM Tris-HCl, pH 6.8). The proteins were quantified and separated by SDS-PAGE, and the immunoblot analysis was performed as described earlier (43). For detecting DDB2 complex, Novex Tris-glycine gels (Invitrogen) were used for separating proteins under reduced or non-reduced conditions.
Quantitation of CPD and 6 -4PP Photolesions-OSU-2 cells, either siRNA transfected or mock transfected or untransfected, were maintained in fresh serum-free medium for 12 h before exposure to 20 J/m 2 dose of UV irradiation. At the indicated post-UV time, the cells were recovered for isolating genomic DNA. The initial formation of CPD and 6 -4PP and that remaining in DNA after cellular repair for varying times were quantitated using a non-competitive immuno-slot blot assay as described earlier (44). The damage levels were calculated by comparing the band intensities of the samples with UV-irradiated DNA standards run in parallel with all the blots. The total amount of DNA loaded on the nitrocellulose membrane was kept constant for each sample blotted.
Immunofluorescence-Immunofluorescence staining of the cells was conducted essentially according to the method established in our laboratory (45). The cells were grown on coverslips in 60-mm dishes, washed twice with cold PBS, UV irradiated, and then fixed with 2% paraformaldehyde in 0.5% Triton X-100/PBS at 4°C for 30 min, followed by three washes with PBS. For DNA denaturation, the cells were incubated in 2 N HCl for 5 min at 37°C. The coverslips were rinsed three time with PBS and blocked with 20% normal goat serum in wash-ing buffer (0.1% Triton X-100/PBS) at room temperature for 30 min. Primary rabbit anti-XPC, anti-CPD, mouse monoclonal anti-HA and anti-Myc antibodies as well as fluorescent (FITC or Texas Red)-conjugated secondary antibodies were all prepared in washing buffer containing 1.5% normal goat serum and layered on the coverslips for 1 h at room temperature. Following each antibody incubation step, the cells were washed with 0.1% Tween 20/PBS four times for 5 min each. After fluorescent staining, the coverslips were mounted in VectaShield antifade containing medium with 0.75 mg/ml of 4Ј, 6Ј-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA) as a DNA counterstain. Fluorescence images were obtained with a Nikon Fluorescence Microscope E80i (Nikon, Tokyo, Japan) fitted with appropriate filters for FITC, Texas Red, and DAPI. The digital images were then captured through automatic time exposures with a cooled CCD camera and processed with SPOT analysis software (Diagnostic Instruments, Sterling Heights, MI).
Antibodies-Rabbit anti-DDB2 antibody (DDB2-A) and anti-XPC antibody (XPC-2) were generated by immunizing rabbits with synthetic peptides and were affinity purified with the corresponding peptide (BioSource, Hopkinton, MA). Peptide KRPETQKTSEIVLRPRNKR matches to the N terminus of human DDB2 protein, whereas KTKREK-KAAASHLFPFEKL matches to the C terminus of human XPC protein.
Polyclonal anti-CUL-4A was a gift from Dr. Yue Xiong (University of North Carolina). Polyclonal anti-CPD was raised and characterized in our laboratory as previously described (44). Monoclonal anti-CPD and anti-6 -4PP antibodies were generously provided by Dr. Tsukasa Matsunaga (Kanazawa University, Japan). Monoclonal antibodies recognizing c-Myc, FLAG, or HA epitopes were purchased from Sigma or Roche Diagnostics, respectively. Antibodies against Actin and XPB were from Neomarkers (Fremont, CA) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Fluorescent-conjugated antibodies Alex Fluor 488 goat anti-mouse IgG2a and IgG1 were from Molecular Probes (Eugene, OR); Texas Red goat anti-rabbit IgG and FITC goat anti-rabbit IgG were from Santa Cruz Biotechnology.
RNA Interference-siRNA oligonucleotides were obtained from Dharmacon (Lafayette, CO) in a purified and annealed duplex form. The sequences targeting human CUL-4A gene are 5Ј-GAACAGC-GAUCGUAAUCAAUU-3 (sense) and 5Ј-pUUGAUUACGAUCGCU-GUUCUU-3Ј (antisense). This set was chosen from four candidate siRNA through pilot experiments. siRNA transfection experiments were carried out using Lipofectamine transfection reagent according to the manufacturer's instructions. Briefly, the Lipofectamine 2000 and OPTI-MEM medium (Invitrogen) were mixed for 5 min and then incubated with siRNA for 20 min at room temperature. After addition of the proper amount of 10% fetal bovine serum/Dulbecco's modified Eagle's medium to the mixture, the siRNA-Lipofectamine mix was applied to the cell cultures. 48 h following siRNA transfection, the cells were used to analyze the expression of CUL-4A, DDB2, and other proteins. For DNA repair experiments, the transfected cells were maintained for 12 h in fresh serum-free medium after siRNA treatment. For immunofluorescent staining experiments, the siRNA transfection was performed with the cells grown on coverslips in 60-mm dishes.

CUL-4A Is an Essential Component for UV-induced DDB2 Degradation in
Vivo-It has been reported that the DDB2 protein undergoes a Ub-mediated proteolysis shortly after cellular UV irradiation (19,36). It has also been shown that DDB2 degradation is enhanced by overexpression of CUL-4A (35). We tested whether CUL-4A is an essential component for UV-induced degradation of DDB2 in vivo using RNA inter-ference-based targeting technology. As shown in Fig. 1A, HeLa-DDB2, stably expressing FLAG-HA-tagged DDB2, produced DDB2 at significant levels in the absence of UV irradiation (lanes 1 and 2). However, upon UV irradiation these cells exhibited a sharp decrease in the DDB2 levels within 2-8 h of exposure (lanes 3, 5, and 7). Moreover, the levels of XPB, a component of transcription factor II H, were unaffected by UV irradiation at all post-irradiation times. Although both of these factors, DDB2 and XPB, are involved in NER, these data indicated that UVinduced loss of DDB2 protein is highly specific and presumably related to its function in initiating NER. Treatment of the cells with the proteasome inhibitor MG132 showed a clear inhibition of the UV-induced loss of DDB2 in HeLa-DDB2 cells (lanes 4, 6, and 8), suggesting the involvement of the Ub-proteasome system in this process. Controls, mock transfected with nonspecific siRNA, showed prompt degradation of DDB2 as early as 0.5 h after UV irradiation (Fig. 1B). The lowest level of DDB2 was seen at 4 h, and it started to recover at 8 h following irradiation. In contrast, no obvious loss of DDB2 could be seen after UV irradiation in cells that were transfected with CUL-4A siRNA. Western blot analysis showed that CUL-4A siRNA was specific and effective in diminishing the targeted CUL-4A protein at all the experimental time points used for assessing the fate of repair factors (Fig. 1C). Thus, silencing CUL-4A in cells prevented UV-induced DDB2 degradation, which is in agreement with the potential role of Cul-4A in DDB2 clearance.
CUL-4A Complexes with DDB and Is Recruited to DNA Damage Sites-Recent studies have identified a DDB-CUL-4A complex that possesses E3 Ub ligase activity (47). Biochemical analysis with purified DDB-CUL-4A complex also showed that the complex can directly ubiquitylate DDB2 in vitro (37). We hypothesized that by virtue of the DNA binding prowess of one of its associated components, CUL-4A would be physically recruited to DNA damage sites as a multimeric DDB-CUL-4A complex. This complex would there-

DDB2 Degradation and NER Regulation by Cullin 4A
fore respond to cellular UV irradiation like its physically associated component and the in vivo substrate, DDB2. To test this hypothesis, we first examined the DNA damage response of the complex after in vivo UV irradiation of cells. The DDB-CUL-4A complex was purified, from HeLa-DCH cells that express both HA-FLAG-tagged DDB2 and Myc3-tagged-CUL-4A, by two immunoprecipitation steps using anti-HA and anti-FLAG affinity matrices. For direct examination of the complex components, cellular proteins were cross-linked by formaldehyde treatment of cells before and at varying times after UV irradiation. A complex with a molecular mass larger than 220 kDa was detectable in Western blots using anti-HA (DDB2) antibody ( Fig. 2A). The same complex was also seen using anti-DDB1 as well as anti-Myc (CUL-4A) antibodies. The presence of Cul-4A in the complex was confirmed by direct binding of antibodies specific for native CUL-4A protein (data not shown). Interestingly, the strongest signal of the complex components was seen in the samples from unirradiated cells, and this was the case irrespective of which component-specific antibodies were used for detection. This high level of protein signal persisted at 0.5 h and then dramatically decreased at 2 and 4 h after cellular UV irradiation. Because the complex was pulled out through tagged DDB2, these results indicate that the DDB2 component is eliminated from the complex after irradiation. On the other hand, DDB1, DDB2, and CUL-4A were all detected at their corresponding molecular sizes after the cross-linking was reversed (Fig. 2B). Several distinct bands, larger than the size of the original DDB2, visualized only in 0.5-h samples were indications of the ubiquitinated forms of DDB2. In comparison to unirradiated control, lower levels of CUL-4A protein were detected in the samples from 0.5, 2, and 4 h post-irradiation, suggesting that CUL-4A may also be separated or eliminated from the complex after UV irradiation. It may also be noted that despite loading of the identical amounts of cross-linked or cross-link-reversed protein samples, signal intensity for both DDB1 and DDB2 was greater upon reversal of cross-links than that observed for components within cross-linked complexes. In essence, no significant loss of DDB1 and DDB2 could be demonstrated upon cross-link reversal. It is possible that the antibody-reactive epitopes are buried within the complexes and the reactivity to individual component proteins is more efficient. The resulting increase of signal intensity from DDB1 and DDB2 precludes the visualization of UV-induced decrease that is clearly demonstrable in the complexes seen in Fig.  2A. Furthermore, it is well recognized that DDB in cells can itself exist within other complexes besides that with CUL-4A.
To determine whether CUL-4A actually translocates to DNA damage sites, we employed the micropore UV irradiation technique, which can deliver UV damage to a localized area of the cell nucleus (45). The HeLa-DCH cells, stably expressing Myc3-tagged CUL-4A, were irradiated through a 5-m micropore filter and the cells were allowed to repair the damage for 15, 30, and 120 min. As shown in Fig. 2C, CUL-4A-specific signal (green) becomes intensified within subnuclear spots in UV-irradiated cells at 15 and 30 min after exposure. The same DNA damage-containing subnuclear spots were also visualized with anti-XPC antibodies (red), indicating the recruitment of repair factors for the assembly of productive NER machinery. Because subnuclear spots have been spatiotemporally characterized as localized damage spots (45), these results indicate that, like DDB2 and XPC, CUL-4A protein translocates to DNA-damaged sites immediately after UV irradiation. It may be noted that only some of the anti-XPC-visualized spots were also picked up by anti-Myc (CUL-4A) staining. In addition, at 2 h after micropore UV irradiation, some subnuclear spots were clearly detectable with anti-XPC staining, whereas none could be detected by anti-Myc (CUL-4A) staining. These results point to a temporal relationship according to which CUL-4A seems to get dislodged from the DNA damage and repair sites prior to releasing of XPC.
UV-induced DDB2 Elimination from the DNA Damage Sites Is Dependent on Cul-4A-Previous studies have described the translocation of DDB2 and its tight association with chromatin upon cellular UV irradiation (19,30,32). Here we have shown the recruitment of DDB2 as well as its elimination from the sites of DNA damage in a chromatin context within the native environment of an intact cell. We concomitantly visualized DDB2 as well as UV-induced CPD in the nuclei after micropore UV irradiation. Fig. 3A shows that at 30 min post-irradiation, DDB2 foci (green) could be visualized at the same CPD-detectable spots (red), indicating that DDB2 readily translocates to DNA damage sites (48). However, at 4 h, the DDB2 foci were no longer distinguishable despite the presence of multiple CPD foci. The calculated ratio of DDB2 foci versus CPD foci (DDB2/CPD) indicated that DDB2 translocates to most DNA damage sites soon after UV irradiation. The ratio was ϳ90% at 15 and 30 min post-irradiation and started to decrease dramatically after 2 h (Fig. 3D). The ratio dropped to Ͻ10% at 4 h, suggesting that DDB2 is either degraded at or dispersed from the damage sites. Because the fluorescent images were captured through automatic time exposures (expected to prominently reveal the fainter fluorescence), the overall intensity of DDB2 fluorescence at 4 h appears to be unaffected upon micropore UV irradiation as compared with unirradiated controls. Besides, it should be noted that the micropore irradiation only The DDB-CUL-4A complex was purified from UV-irradiated or unirradiated HeLa-DCH cells after fixation with 1% formaldehyde at the indicated post-irradiation times. The complex purification was carried out through two immunoprecipitation steps as described under "Experimental Procedures." The purified complex was analyzed either as cross-linked (A) or after reversal of cross-links (B). Immunoblotting was performed using indicated antibodies against individual components. C, the HeLa-DCH cells, grown on coverslips, were irradiated with 100 J/m 2 UV-C through a 5-m isopore filter and maintained in fresh medium for the periods indicated. After micropore UV irradiation, the cells were fixed with 2% paraformaldehyde. The Myc3-tagged CUL-4A and XPC were visualized by antibodies specific to each protein.
delivers a fraction of UV dose to cells. This low level irradiation would be insufficient to significantly deplete the DDB2 from cells to reveal appreciable intensity differences. Nevertheless, when the cells were treated with proteasome inhibitor MG132, DDB2 spots become clearly visible even at 4 h post-irradiation (Fig. 3B). MG132 treatment increased the DDB2/CPD ratio to more than three times at 4 h post-irradiation (Fig.  3D). The fact that DDB2 translocates to DNA damage sites very rapidly (48) and then promptly comes off points to the role of DDB2 proteolysis in the disappearance of DDB2 foci from DNA damage sites. Therefore, we further examined the effects of CUL-4A knock down on DDB2 elimination from the DNA damage sites. As illustrated in Fig. 3C, CUL-4A siRNA treatment did not affect the initial formation of DDB2 foci at damage sites (Fig. 3, C and D). The ratio of DDB2/CPD foci after CUL-4A silencing remained as high as 90% at 15 and 30 min postirradiation. Nevertheless, the siRNA treatment clearly halted the normal disappearance of DDB2 foci at 2 and 4 h post-irradiation (Fig. 3, C  and D). About 30% of DDB2 foci were still detectable at 4 h in CUL-4A siRNA-treated cells as compared with ϳ10% in mock-treated cells. The data strongly suggest that although CUL-4A is not required for the recruitment of DDB to DNA damage, it is clearly needed at the damage sites for the degradation of DDB2 and for the repair to move forward.
CUL-4A Knock Down Decreases XPC Recruitment and Affects GGR of CPD-Our previous studies, using a Ub-activating enzyme-defective thermosensitive cell line, have demonstrated that the Ub-proteasome system is required for efficient GGR and TCR (49). To investigate the functionality of DDB2 ubiquitylation and its subsequent degradation in NER, we next examined the effect of CUL-4A knock down on XPC recruitment. The spatiotemporal analysis of factor recruitment to damage sites was carried out after UV irradiation of cells at 40 J/m 2 through a 5-m micropore filter. Considering that the nucleus is mostly shielded by the filter, this UV dose is comparatively lower and therefore the overall lesion load produced would be quantitatively repaired. As shown in Fig. 4A, both XPC and CPD foci were visualized within the nucleus of UV-irradiated cells. The XPC/CPD ratio at 30 min post-irradiation was determined to be ϳ66% in controls without a CUL-4A siRNA treatment. However, the initial factor recruitment of damage sites was significantly affected as the ratio decreased to ϳ32% upon CUL-4A siRNAmediated knock down. The data suggest a clear involvement of CUL-4A-based E3 ligase in regulating the recruitment of XPC to UV-induced photolesions, presumably due to a lack of DDB2 clearance from damage sites.
We further examined the effect of CUL-4A knock down on GGR of UV-induced photolesions (Fig. 4, B-E). Both CPD and 6 -4PP lesions were quantitated by immuno-slot blot assay of the genomic DNA isolated at varying post-irradiation times from the CUL-4A siRNA-transfected or mock-transfected repair-proficient human fibroblasts (Fig. 4,  B and D). The mock-transfected cells exhibited the typical kinetics of CPD repair, i.e. the unrepaired CPD remaining in the genome were ϳ80, 65, and 46% of initial damage at 4, 8, and 24 h, respectively. In contrast, in CUL-4A siRNA-transfected cells, the unrepaired CPD was ϳ88, 78, and 63% at 4, 8, and 24 h, respectively (Fig. 4C). Obviously, the CUL-4A knock down decreased the efficiency of GGR at all the time points . Proteasome inhibition and CUL-4A knock down prevented the proteolysis of DDB2 and delayed the disappearance of DDB foci from DNA damage spots. A, the HeLa-DDB2 cells, grown on coverslips, were exposed to UV irradiation through micropore filters and processed as in Fig. 2C. Before immunostaining, the cells were briefly treated with 2 N HCl to denature the DNA. The DNA damage spots and DDB2 were visualized by antibodies against CDP and HA epitope, respectively. B, the UV-irradiated cells were kept in fresh medium for 15 min and then in the MG132-containing medium until the fixation step. C, the HeLa-DDB2 cells were transfected with CUL-4A siRNA before micropore UV irradiation. D, the ratio of DDB2 versus CPD foci was obtained from calculations based on the number of DDB2 and CPD foci in 200 individual cells from three different microscope fields.
tested. In contrast to CPD repair, the kinetics of 6 -4PP repair, which is known to occur at a faster rate than CPD, was identical in both siRNAand mock-transfected cells (Fig. 4E). Therefore, the GGR of 6 -4PP was not impacted by impairing the action of CUL-4A.
Availability of DDB2 and XPC Proteins Is Differentially Affected by UV-Once DDB2 binds to DNA damage and impinges on the initial repair event by being consumed through degradation, it cannot be available for the removal of any residual or newly induced lesions. To test that this actually happens, we designed a "two-hit experiment" to ascertain the participation of DDB in lesion repair. First, the entire nuclear DNA was subjected to UV-induced damage by global irradiation of cell monolayers. The cells were allowed to repair for 4 h so that DDB could bind, initiate repair, and consequently be degraded in the process. Then these pre-irradiated cell monolayers were UV irradiated a second time through micropore filters to generate localized DNA damage spots. The recruitment of DDB2 and XPC to these newly damaged subnuclear spots was examined by fluorescent immunostaining with cognate factor-specific antibodies. As shown in Fig. 5, global UV pre-irradiation at 20 J/m 2 significantly decreased the relative recruitment of DDB2 to ϳ30% of pre-irradiated controls. The percentage further decreased to ϳ10% in 40 J/m 2 pre-irradiated cells, indicating that global UV irradiation depleted DDB2 initially available for the repair of a certain level of damage. The recruitment of XPC to subnuclear damage spots generated by the second dose was also affected by the first global UV irradiation albeit in a qualitatively different manner than DDB2. For example, the recruitment of XPC remained constant at ϳ50 -55% for 20 and 40 J/m 2 , whereas DDB2 continued to decline as a function of dose. Because the fates of DDB2 and XPC after UV irradiation are differentially regulated by the Ub-proteasome system (50), these results support the idea that DNA repair-mediated degradation of DDB2 causes its depletion and, B-E, the transfected or mock-transfected fibroblasts were maintained overnight in serum-free medium and irradiated with UV-C at 20 J/m 2 . The cells were allowed to repair the damage for indicated times. The CPD (B and C) or 6 -4PP (D and E) in genomic DNA were detected by immunoslot blot assay using cognate damage-specific antibodies. The intensity of lesion-specific signals was determined by laser densitometric scanning, and the amount of damage was calculated upon comparing the band intensities with a standard reference run in parallel. The repair was expressed as the percentage of initial damage remaining in isolated cellular DNA. MAY 12, 2006 • VOLUME 281 • NUMBER 19 unlike XPC, prevents its recycling for the repair of additional photolesions.

DISCUSSION
Cellular exposure to UV irradiation is known to cause several changes of the DDB complex, e.g. prompt translocation from cytoplasm into nucleus, tight association with chromatin, and rapid degradation of its DDB2 protein component (19,51,52). These findings, together with recent identification of the DDB-CUL-4A E3 ligase complex, suggest that the complex could be functioning at the very early step of DNA damage recognition and repair. Here, we have provided evidence about how the two damage recognition factors, DDB2 and XPC, cooperate in early steps of GGR. Based on our accumulated data, it is posited that the clearance of DDB2 by Ub-mediated degradation facilitates the access of XPC and other NER factors to help complete the efficient repair of DNA lesions (37,50).
In this study, we first demonstrated that CUL-4A is an integral and functionally essential component for UV-induced DDB2 degradation in vivo and that DDB2 and CUL-4A are simultaneously recruited to damage spots following micropore UV irradiation of nuclear chromatin. The observation strongly suggested that damage-bound DDB2 is an in vivo substrate of the DDB-CUL-4A E3 ligase complex. It has been known that DDB-CUL-4A E3 ligase is inactivated through its association with regulatory COP9 signalosome (47). Because the neddylated forms of CUL-4A were specifically detected in the chromatin-bound fractions of the UV-irradiated cells, it is believed that the DDB-CUL-4A E3 complex is activated when signalosome leaves the complex upon cellular UV irradiation. Accordingly, Sugasawa et al. (50) and our laboratory (49,53) have recently found that another damage recognition factor, XPC, is ubiquitylated by DDB-CUL-4A E3 upon cellular UV irradiation. These studies help envisage that DDB-CUL-4A E3 could be physically recruited to the damage sites. Our present study shows not only the occurrence of CUL-4A in complex with DDB but also its translocation to localized DNA damage sites in response to irradiation (Fig.  2). CUL-4A foci were visualized at the DNA damage sites within 15 min after micropore UV irradiation. In addition, when the genomic DNA was recovered from chromatin immunoprecipitates of anti-DDB2 and anti-Myc (CUL-4A) antibodies, higher amounts of UV-induced CPD and 6 -4PP were detected in DDB2 or CUL-4A-bound DNA than in the bulk of the isolated DNA (data not shown). Furthermore, silencing CUL-4A did not affect the recruitment of DDB2 but exhibited a pronounced delay in dispersal of DDB2 from DNA damage sites (Fig. 3). Therefore, it seems that the association of DDB-CUL-4A E3 complex to DNA damage is mediated through DDB. It was shown that CUL-4A foci disappear at 2 h after micropore UV irradiation. This temporal sequence suggests that 6 -4PP is the major source of recruitment stimulus for the formation of CUL-4A foci, as most of the CPD remains essentially unrepaired at this time.
The in vivo recruitment of DDB-CUL-4A E3 complex to DNA damage signifies the events that happen at the damage sites during early steps of damage recognition. Moreover, binding of this complex would provide the necessary transition from the damage recognition by DDB to that by XPC. The latter then initiates the assembly of remaining NER factors (12). In XPE cells lacking the DDB activity, GGR of CPD is known to be significantly impaired (13). This suggests the importance of DDB in NER in vivo, as DDB seems expendable for in vitro reconstituted NER (26,46). Accordingly, it has been shown that DDB2 activates the in vivo recruitment of XPC to CPD (32). On the other hand, DDB2 as well as XPC, unaided by DDB2, translocate to UV-irradiated sites containing 6 -4PP. It is possible that DDB provokes efficient targeting of NER machinery to both types of photolesions (33). This agrees with our hypothesis that the clearance of DDB2 by Ub-mediated degradation facilitates the access of XPC and other NER factors to DNA lesions. It should be emphasized that knock down of CUL-4A delayed the dispersal of DDB2 from the damage sites while it decreased the recruitment of XPC to DNA damage and subsequently reduced the removal of CPD from the genome (Figs. 3 and 4). These results are consistent with previous published observations that knock down of CSN5, a critical component of COP9, leads to defect of GGR and TCR (47). While COP9 is a negative regulator of associated E3 ligase in vitro, it would be required for the same E3 activity in vivo (41). Therefore, interpretation of the requirement of CSN5 for GGR suggests the importance of the ligase activity of DDB-CUL-4A E3 complex in GGR. Our siRNA experiments specifically targeted CUL-4A-based E3 ligase and therefore provided direct evidence that the ligase activity of DDB-CUL-4A E3 complex is involved in GGR. In these experiments, the decrease in XPC recruitment by CUL-4A knock down further suggested that the ligase activity of DDB-CUL-4A E3 complex is required for an efficient recruitment of XPC to DNA damage. Sugasawa et al. (50) recently reported that the ubiquitylation of DDB2 alters the DNA binding property of DDB in vitro, suggesting that the ubiquitylation regulates the departure of DDB from damage DNA. Our results, however, favor the suggestion that both ubiquitylation and subsequent proteasomal degradation of DDB2 are responsible for the clearance of DDB, allowing XPC to load at damage sites. In support of this view, we have previously shown that treatment of cells with proteasome inhibitors compromises the repair of UV-induced CPD and also inhibits the UV-induced XPC recruitment to DNA damage sites (49).
Interestingly, the fates of DDB2 and XPC are quite different, although both of them appear to be the substrates of DDB-CUL-4A E3 complex (50,53). In our two-hit experiments (Fig. 5), 40 J/m 2 global UV irradiation (first hit) diminished the DDB2 recruitment ϳ90% to damage that was freshly produced by micropore UV irradiation (second hit). This indicates that the damage loading following the first exposure depletes most of the available DDB activity. Nevertheless, at the same damage load, 50% XPC recruitment was observed at the freshly induced damaged sites. The difference between the availability of XPC and that of the DDB2 strongly suggests that XPC is re-used for the repair of persistent damage. Because DDB2 is required for in vivo XPC recruitment and it is FIGURE 5. Two-hit experiments show differential effects of damage load on the recruitment of DDB2 and XPC to freshly induced DNA damage sites. HeLa-DDB2 cells on coverslips were globally irradiated with UV-C at indicated UV doses and maintained for 4 h in fresh medium to allow the first phase of repair. A second cycle of 100 J/m 2 UV was delivered to the cells through a 5-m micropore filter. The cells were incubated for another 30 min, fixed, and double immunostained with anti-HA (DDB2) and anti-XPC antibodies. The relative recruitment of DDB2 and XPC was calculated based on the numbers of DDB2 or XPC foci in 200 cells from at least three microscopic fields.
degraded after the global UV irradiation, it is reasonable to assume that 6 -4PP presents a major stimulus for the recruitment of XPC to newly formed damage sites.
In summary, we have examined the CUL-4A-mediated proteolysis of DDB2 at the DNA damage sites and explored the functional relevance of such DDB2 degradation to GGR with regard to XPC recruitment and removal of UV-induced photolesions. It should be recognized, however, that DDB-CUL-4A E3 complex could regulate NER through substrates other than DDB2. For example, XPC has been recently identified as one of the key DNA repair-related substrates of the complex (50). Clearly, regulation of NER by the Ub-proteasome system remains an exciting research area for further exploration.