Role of poly(ADP-ribose) polymerase-1 in the removal of UV-induced DNA lesions by nucleotide excision repair

Abstract

Among the earliest responses of mammalian cells to DNA damage is catalytic activation of a nuclear enzyme poly(ADP-ribose) polymerase-1 (PARP-1). Activated PARP-1 forms the polymers of ADP-ribose (pADPr or PAR) that posttranslationally modify its target proteins, such as PARP-1 and DNA repair–related proteins. Although this metabolism is known to be implicated in other repair pathways, here we show its role in the versatile nucleotide excision repair pathway (NER) that removes a variety of DNA damages including those induced by UV. We show that PARP inhibition or specific depletion of PARP-1 decreases the efficiency of removal of UV-induced DNA damage from human skin fibroblasts or mouse epidermis. Using NER-proficient and -deficient cells and in vitro PARP-1 assays, we show that damaged DNA-binding protein 2 (DDB2), a key lesion recognition protein of the global genomic subpathway of NER (GG-NER), associates with PARP-1 in the vicinity of UV-damaged chromatin, stimulates its catalytic activity, and is modified by pADPr. PARP inhibition abolishes UV-induced interaction of DDB2 with PARP-1 or xeroderma pigmentosum group C (XPC) and also decreases localization of XPC to UV-damaged DNA, which is a key step that leads to downstream events in GG-NER. Thus, PARP-1 collaborates with DDB2 to increase the efficiency of the lesion recognition step of GG-NER.

Mammalian cells respond very rapidly to different types of DNA damage by activation of an abundant and ubiquitous nuclear enzyme poly(ADP-ribose) polymerase-1 (PARP-1). The activated PARP-1 uses NAD⁺ to form polymers of ADP-ribose (pADPr or PAR) that modify PARP-1 itself and selected target proteins, such as histones and DNA repair proteins (1). This posttranslational modification, i.e., PARylation, has been implicated in cellular responses ranging from DNA repair to cell death. Among mammalian DNA repair pathways, PARP-1 has been implicated in base excision repair, homologous recombination, and nonhomologous end-joining pathways (2, 3), but we do not know its role in the most versatile nucleotide excision repair (NER) pathway that removes a wide variety of DNA lesions, including UV-induced thymine dimers (T-T) and other cyclobutane pyrimidine dimers (CPD), as well as 6–4 photoproducts (6-4PP) (4).

The core mammalian NER pathway uses more than 30 proteins to recognize the damaged site on DNA, remove 24- to 32-nucleotide-long single-stranded DNA containing the lesion, fill the gap using the nondamaged strand as a template, and finally ligate the nick (4). There are two subpathways of NER: the transcription-coupled NER (TC-NER) removes lesions from the actively transcribed strands of the genes and the global genomic NER (GG-NER) repairs lesions from the entire genome. These two pathways differ in the initial step of lesion recognition: TC-NER is initiated when elongating RNA polymerase II stalls at the lesion, whereas GG-NER is initiated when the lesion is recognized in the chromatin context by DDB2 (XPE), which through its participation in the UV-DDB-E3 ligase complex ubiquitinates and localizes the key GG-NER protein XPC to the damaged site (5–7).

Although the roles for different core NER proteins have been well characterized with bacterial and yeast model systems, we still cannot fully explain the accuracy and rapidity with which mammalian NER is targeted to a very few damaged bases that are surrounded by a large number of unmodified bases in chromatin. In this context, some of the posttranslational modifications, such as phosphorylation, acetylation, ubiquitination, and sumoylation, are known to help different steps of NER (8). Here, we examined whether PARylation that occurs rapidly after PARP-1 is activated by UV-induced DNA lesions (9) could be involved in improving the efficiency of mammalian NER. Earlier studies examined the effect of impaired PARP-1 function on mammalian NER but obtained contradictory results (10–12), because the repair of UV-induced CPD was unaffected by PARP inhibition in HeLa cells (10), whereas it was reduced in transdominantly PARP-1–inhibited CHO cells (11) or in the PARP-1–impaired triple negative breast cancer cell lines (12). In view of other confounding factors such as DNA repair defects in CHO or breast cancer cells, we used RNAi or PARP inhibition approaches in multiple mammalian NER-proficient models to show that PARP-1 is required for an efficient removal of UV-induced T-T and 6-4PP from genomic DNA. We also show that the catalytic activity of PARP-1 in collaboration with DDB2 leads to an improved function of DDB2 and XPC during the lesion recognition step of mammalian GG-NER.

Results

PARP Inhibitors Delay Removal of UV-Induced DNA Lesions.

To explore the role of catalytic activation of PARP-1 in NER, we first examined the effect of PARP inhibitor PJ-34 on the efficiency of removal of UVC-induced T-T or 6-4PP from genomic DNA of two different human skin fibroblasts using a flow cytometry–based assay (13) (Fig. 1A). In this assay, the histograms for T-T or 6-4PP at early time points after irradiation (5–15 min) represent initial damage, and movement of histograms at later time points (6–63 h) toward untreated cells represents the extent of repair. In the SV-40–immortalized GMU6 human skin fibroblasts, a significant removal of T-T at 24 h was seen only in the normal but not in the PJ-34–treated cells (Fig. 1A, Left). The quantification of the average T-T signal confirmed that 43% of damage removed by 24 h from normal GMU6 cells was significantly more than 27% of the damage removed by PJ-34–treated cells (n = 4–7; P < 0.05). Next, we examined the effect of PJ-34 on the capacity of hTert-immortalized BJ-EH2 human foreskin fibroblasts (BJ-hTert) to remove UVC-induced T-T and 6-4PP lesions up to 63 and 6 h, respectively (Fig. 1A, Center and Right). Unlike normal BJ-hTert cells that removed all of the T-T signal by 63 h, PJ-34–treated cells removed significantly less (50%) damage, which was further confirmed by quantifying the average T-T signal from multiple assays (n = 6; P < 0.05). For removal of 6-4PP lesions, although the final repair by 6 h was not affected, the early phase of removal of damage at 1 h was significantly suppressed by PJ-34, and quantification of the signal confirmed that 45% of damage removed by normal BJ cells was significantly more than 20% of the damage removed by PJ-34–treated cells (n = 3; P < 0.01). Last, the removal of UVB-induced T-T from the epidermis of SKH-1 hairless mice was also reduced up to 48 h by the PARP inhibitor 1,5-dihydroxyisoquinoline (Fig. 1B). Thus, PARP inhibitors significantly decreased the efficiency of removal of UV-induced DNA photolesions in multiple models.

Fig. 1.

Impaired PARP-1 function delays removal of T-T and 6-4PP from genomic DNA. (A) PARP inhibition decreases repair of T-T and 6-4PP. The GMU6 and BJ-hTert fibroblasts were treated with 10 µM PJ-34 (or control) before UVC irradiation at 10 (for T-T) or 30 (for 6-4PP) J/m². Repair was monitored before (untreated: UT) or after irradiation at specified times. Representative histograms from one of four to six experiments with similar results are shown. (B) PARP inhibition delays removal of UV-induced DNA lesions from mouse epidermis. Mice were exposed to 1,600 J/m² UVB, and the PARP inhibitor DHQ was applied every 3 h to inhibit PARP activation up to 12 h. Skin was processed for immunohistological analysis of T-T. Data shown here are from one of four experiments with identical results. (C) PARP-1–depleted GMSiP cells are deficient in formation of pADPr after UV exposure. The chromatin-bound protein fractions from GMSiP and PARP-1–replete GMU6 cells at specified time after irradiation with 30 J/m² UVC were immunoblotted for PARylated proteins, as shown here for one of three experiments with identical results. Asterisk indicates nonspecific reaction with BSA, and Ponceau staining was a loading control. (D–F) PARP-1 depletion decreases repair of UV-induced T-T in nonsynchronized or in G1 synchronized human skin fibroblasts. GMU6 and GMSiP cells were nonsynchronized (D) or synchronized in G1 phase with 0.5 µM mimosine (E and F) before irradiation with 10 (D and E) or 100 (F) J/m² UVC. Repair was monitored at specified times by flow cytometry (D and E) or fluorescence microscopy (F). Data are from one of three to four experiments with similar results.

PARP-1 Depletion Decreases Efficiency of Removal of UV-Induced DNA Damage.

PARP inhibitors affect activity of all of the members of the PARP family; hence, we examined whether the effect of PARP inhibition on repair of UV-damaged DNA was caused by its effect on PARP-1 or on other PARPs. We used GMSiP human skin fibroblasts, in which PARP-1 has been stably and significantly depleted by shRNA without affecting the expression of PARP-2 (14). We assessed their capacity to (i) form PAR in response to UVC and (ii) repair the UVC-induced T-T lesions. In the pADPr immunoblot, a strong signal for heterogeneous bands of PARylated proteins above 116 kDa (15) could be seen in the matched PARP-1–replete GMU6 cells but not in the GMSiP cells (Fig. 1C), confirming that PARP-1 is the major, if not the only, producer of pADPr in UV-treated cells. Using flow cytometry assay, we noted a marked failure of GMSiP cells to remove T-T from 15 min to 24 h (Fig. 1D), which is in stark contrast to the significant repair seen in GMU6 cell (Fig. 1A, Left). To exclude the possible differences in the cell cycle phases influencing the repair capacity of these cells (13), we synchronized GMU6 and GMSiP cells in the G1 phase and compared their time course of removal of T-T up to 24 h by flow cytometry (Fig. 1E) and immunofluorescence microscopy (Fig. 1F). By both techniques, we observed that PARP-1–depleted cells were inefficient at removal of T-T damage. The quantification of T-T signal from flow-cytometry assays confirmed that, although GMU6 cells removed 54% of the initial damage, GMSiP cells barely removed any damage (∼2%; n = 3; P < 0.01). The immunofluorescence microscopy confirmed this trend because GMU6 cells removed 58% of the T-T signal per nuclei by 24 h compared with 15% damage removed by GMSiP cells (n > 125 nuclei; P < 0.01). Because PARP-1 depletion was sufficient to abolish UV-induced PAR synthesis and impair repair of UV-induced DNA damage, similar to that seen in PARP-inhibited cells, PARP-1 is likely to be the main PARP implicated in this repair process.

Characterization of UV-Induced Interaction Between PARP-1 and DDB2.

We showed earlier that PARP-1 rapidly binds to UV-damaged DNA in vitro or in UV-irradiated cells, and it is activated to form PAR within seconds after irradiation at the site of DNA damage (9). Because DDB2, the early GG-NER protein, is also known to translocate very rapidly at the site of UV-damaged DNA (16), we examined whether DDB2 and PARP-1 interact with each other in the vicinity of UV-damaged chromatin using cellular fractions that represent chromatin-bound proteins (Ch-fraction) rather than the whole cell extract in coimmunoprecipitation (co-IP) studies (Fig. S1A). The cell fractionation technique to isolate the Ch-fraction was validated by confirming the expected UV-induced relocalization of DDB2 to this fraction in both NER-proficient GM637 and NER-deficient XP-C cells (Fig. 2A, lanes 4, 8, and 12), although total cellular DDB2 levels remained unchanged before and after irradiation (Fig. 2A, lanes 1, 5, and 9). We further confirmed that UV irradiation promoted the recruitment of downstream NER proteins XPC and xeroderma pigmentosum group A (XPA) to the Ch-fraction of GM637 cells, whereas XP-C cells did not relocalize XPA to the Ch-fraction (Fig. 2A). For the IP studies, we used GMRSiP cells that express FLAG-tagged human PARP-1 (Fig. S1B). The IP of equal amounts of Ch-extracts (input, Fig. S1C) of these cells prepared before or 10 min after UVC irradiation with the PARP-1 antibody revealed a significant UV-induced association of DDB2 with PARP-1, which was confirmed in an inverse co-IP with DDB2 antibody (Fig. 2B, lanes 4 and 8). Mock IPs with control IgGs confirmed specificity of antibody-based IPs (Fig. S1D). To determine whether the interaction between these two proteins is direct or mediated via DNA, DDB2-IP was carried out in the presence of 200 µg/mL ethidium bromide to loosen the protein–DNA interactions. The failure of ethidium bromide to prevent co-IP of PARP-1 with DDB2 indicated a direct interaction between these two proteins (Fig. 2C; mock control: Fig. S1E).

Fig. 2.

Cooperation between PARP-1 and DDB2 at UVC-irradiated chromatin. (A) UVC-induced recruitment of NER proteins to chromatin-bound protein (Ch) fractions. Different cellular fractions (WC, whole cells; C, cytoplasm; Np, nucleoplasm; Ch, chromatin-bound) prepared from the NER-proficient GM637 and NER-deficient XP-C cells at 0 (unirradiated), 5, and 30 min after UVC irradiation were immunoblotted for DDB2, XPC, and XPA. (B) UV-induced association of PARP-1 and DDB2. The co-IPs were carried out with Ch-fractions (input) derived from control or UVC (30 J/m²) irradiated FLAG-PARP-1 expressing GMRSiP cells with antibodies to PARP-1 or DDB2. Input and IP eluates were immunoblotted for the presence of PARP-1/FLAG and DDB2. Results are from one of six experiments with identical results. (C) Interaction of PARP-1 and DDB2 is not mediated via DNA. The Ch-fractions (input) of GMRSiP cells prepared before or 10 min after exposure to 30 J/m² UVC were incubated with or without ethidium bromide before and during IP for DDB2, followed by immunoblotting for DDB2 and FLAG (PARP-1). Results are representative of two experiments with identical results. (D) PARP inhibitor disrupts co-IP of PARP-1 and DDB2. GMRSiP were treated with 10 µM PJ-34 (or control) before UVC irradiation at 30 J/m². The Ch-fractions (input) prepared at 10 min after irradiation were subjected to IP for FLAG followed by detection of PARP-1 and DDB2. Data are from one of three experiments with identical results. (E) PARP inhibition delays the departure of DDB2. The Ch-fractions isolated from GMU6 cells at specified times after UVC irradiation with 10 J/m² (or control) in the presence or absence of 10 µM PJ-34 were immunoblotted for DDB2. Ponceau staining was used as a loading control.

To determine the possible role of catalytic activation of PARP-1 in this interaction, the GMRSiP cells were irradiated in the presence or absence of the PARP inhibitor PJ-34. The FLAG-IP of Ch-fractions before or after irradiation confirmed UV-induced interaction of DDB2 and PARP-1 without the PARP inhibitor (Fig. 2D, DDB2, lane 4). Interestingly, although PJ-34 treatment did not prevent UV-induced accumulation of DDB2 in the input Ch-fraction before IP, it significantly suppressed its ability to associate with PARP-1 (Fig. 2D, DDB2, lanes 6 and 8). In an independent model of GMU6 cells expressing mycDDB2 (Fig. S2 A–C), we confirmed by local UVC irradiation (Fig. S2D) and co-IP of Ch-fractions with PARP-1 or myc antibodies (Fig. S2E) that the PARP inhibitor PJ-34 did not affect early accumulation of mycDDB2 but blocked its co-IP with PARP-1. We examined whether the PARP inhibitor that disrupts the interaction between DDB2 and PARP-1 also affects the departure of DDB2 from the damaged site, which is a necessary step for the continuation of GG-NER (17). Immunoblotting for DDB2 in the Ch-fractions isolated up to 2 h after irradiation revealed that DDB2 levels that accumulated from 5 to 15 min declined rapidly by 60 min in the normal cells but remained high until 120 min in PARP-inhibited cells (Fig. 2E). Thus, although the initial recruitment of DDB2 to UV-induced chromatin is independent of PARP-1, its subsequent direct association with PARP-1, as well as its eventual departure from the damaged site, is dependent on the catalytic activity of PARP-1 that would PARylate proteins near the damaged DNA.

DDB2 Stimulates Catalytic Activity of PARP-1 and Becomes a Target for PARylation.

The influence of catalytic activity of PARP-1 on DDB2 prompted us to examine whether DDB2 is PARylated in response to UV (Fig. 3 A and B). First, the pADPr-IP of GMU6-mycDDB2 cells pulled down more mycDDB2 after UVC exposure, indicating that it is either PARylated or interacts with other PARylated proteins (Fig. 3A). That DDB2 could be directly PARylated was tested in an in vitro Dot-blot assay, which showed that pADPr could bind to mycDDB2 (Fig. 3B, Upper). The binding of pADPr to its known acceptors PARP-1 and XPA (18) served as positive controls. To confirm that the recipient of pADPr was indeed the designated protein in the immunopurified mycDDB2 preparation, we ran a SouthWestern type of blot in which immunopurified mycDDB2 and purified GST-DDB2 were resolved on SDS/PAGE, blotted, and reacted with pADPr (Fig. 3B, Lower). The mycDDB2 and purified GST-DDB2 displayed one major pADPr-accepting band at 50 or 75 kDa, which corresponds to their respective bands in the DDB2 immunoblots. A strong signal for PARylated PARP-1 at 113 kDa was a positive control, and lack of pADPr binding by proteinase K (Fig. 3B) or DNase I and myc antibody (Fig. S3A) served as negative controls. Thus, DDB2 is an acceptor of pADPr.

Fig. 3.

DDB2 stimulates PARP-1 activation and is a target for PARylation. (A) The pADPr IP was carried out with Ch-fractions isolated from GMU6-mycDDB2 cells before or 10 min after exposure to 10 J/m² UVC and immunoblotted for PARylated proteins and myc. (B) DDB2 binds to pADPr. (Upper) Dot-blot: Purified bovine PARP-1 and immunopurified XPA and DDB2 were spotted and reacted with pADPr before immunoprobing for pADPr. (Lower) SouthWestern blot: Immunopurified mycDDB2, purified GST-DDB2, purified bovine PARP-1, and proteinase K were resolved on SDS/PAGE, blotted, and reacted with pADPr followed by immunoprobing for pADPr, PARP-1, and DDB2. The results are representative of two experiments with similar results. (C and D) DDB2 stimulates catalytic activity of PARP-1 in response to UVC-damaged DNA. In vitro PARP-1 activation assays were performed with purified bovine PARP-1, immunopurified mycDDB2 (C), or different amounts of purified GST-DDB2 (D), control, or UV-irradiated covalently closed circular DNA, untagged or biotinylated NAD, and PJ-34, as specified. Aliquots were immunoblotted for PARP-1 (PARP-1), pADPr (10H or streptavidin), and DDB2 (myc or DDB2). Purified GST-DDB2 is undetectable by antibody at lower concentrations and shows minor degradation bands at higher levels. The results are representative of three experiments with similar results. (E) XP-E cells are inefficient at forming pADPr after UV exposure. GMU6 and XP-E cells were irradiated with 10 J/m² UVC, and whole cell extracts prepared at various times were immunoblotted for pADPr. Actin probing and Ponceau staining served as loading controls. (F) DDB2 stimulates catalytic activation of PARP-1 in vivo. XP-E cells were transiently transfected with myc-DDB2 expression vector for 24 h and treated with 10 µM PJ-34 (or control) before irradiation with 10 J/m² UVC. Samples were processed at 5 min for immunofluorescent visualization of mycDDB2 (red), pADPr (green), and DNA (DAPI-blue). The chart provides an average pADPr signal per unit nuclear area (mean ± SE) of XP-E cells that expressed the transfected mycDDB2 or not. The image and chart are derived from the data of multiple nuclei from different fields of two independent experiments.

We next examined whether interaction of DDB2 with PARP-1 has any influence on the activity of PARP-1. In an in vitro PARP-1 activation assay (15), we first observed using bovine PARP-1 and nonlabeled NAD (Fig. S3B) or immunopurified human FLAG-PARP-1 and biotinylated NAD (Fig. S3C), that the catalytic activation of PARP-1 by UV-damaged DNA was stimulated in the presence of mycDDB2, as seen from a heterogeneous smear above 113 kDa for PARylated PARP-1. Moreover, the effect of DDB2 on stimulating PARP-1 activation and formation of pADPr-modified PARP-1 was more pronounced with UV-damaged DNA than with the unirradiated DNA (Fig. 3C, lanes 2 and 3). Interestingly, the PARylation of mycDDB2 (50 kDa) was evident from the smearing of signal between 50 and 75 kDa in both pADPr and myc-blots in this reaction (lane 3). Controls show that PJ-34 suppressed PARP-1 activation (lanes 4–6), and mycDDB2 could not activate PARP-1 without DNA (lane 1). A dose–response for purified GST-DDB2 in the assay showed that ∼1/12th of the amount of GST-DDB2 gave an optimum activation of PARP-1, i.e., 67 pmol GST-DDB2 for 860 pmol PARP-1 per assay (Fig. 3D; Fig. S3D).

Next, we examined whether the interaction with DDB2 stimulates activity of PARP-1 in UV-irradiated cells. In the pADPr immunoblot, the signal for PARylated proteins in the first 60 min after UVC irradiation was extremely weak in DDB2-deficient XP-E cells compared with DDB2-replete GMU6 cells (Fig. 3E), although PARP-1 levels were comparable in both the cell types (Fig. S3E). This deficiency could be rescued in XP-E cells by transient transfection of the mycDDB2 expression vector, because despite being irradiated with UVC at the same time, the XP-E cells that did not express transfected mycDDB2 (nonred nuclei) showed a basal level of pADPr synthesis (green signal), which was significantly stimulated in the cells that expressed mycDDB2 (red nuclei) (Fig. 3F). The PJ-34 treatment abolished this additional pADPr synthesis. Collectively, these results indicate that DDB2 stimulates the catalytic activity of PARP-1 by UVC-damaged DNA and, in turn, PARP-1 PARylates DDB2.

XPC and the Interaction of PARP-1 with DDB2.

To determine whether XPC that is known to co-IP with DDB2 (5) plays a role in the PARP-1 and DDB2 interaction, we used XP-C cells because they could recruit DDB2 to the damaged DNA but not XPA and other downstream GG-NER proteins due to nonfunctional XPC (Fig. 2A). Because PARP-1 activation and pADPr synthesis played a key role in the DDB2–PARP-1 interaction, we first examined whether XP-C cells could make pADPr in response to UVC. The time course of the pADPr immunoblot revealed that the appearance and extent of PARylated proteins in UV-irradiated XP-C and NER-proficient GM637 cells was comparable (Fig. 4A). Using co-IP studies in XP-C cells for PARP-1 (Fig. 4B) or pADPr (Fig. 4C), we observed that UVC irradiation promoted an interaction between PARP-1 and DDB2 (Fig. 4B) and caused direct or indirect PARylation of DDB2 (Fig. 4C). Because these interactions were not as robust as in normal cells (Figs. 2 B and C and 3A), our results indicate that there may be an XPC-dependent and -independent interaction between PARP-1 and DDB2.

Fig. 4.

Effect of PARP inhibition on XPC-related events in response to UV. (A) UV-induced pADPr synthesis in XP-C cells. GM637 and XP-C cells were irradiated with 10 J/m² UVC, and whole cell extracts prepared at various time points were immunoblotted for pADPr. Actin was used as a loading control. (B and C) UV-induced interaction of PARP-1 with DDB2 and pADPr modification of DDB2 in XP-C cells. The Ch-fractions (input) derived from control or UVC (30 J/m²) irradiated XP-C cells were subjected to IP for PARP-1 or pADPr and immunoblotted for PARP-1, pADPr (10H), and DDB2. Results shown here are from one of three experiments with identical results. (D) PARP inhibitor disrupts UV-induced co-IP of DDB2 with XPC. GMU6 cells were treated with 10 µM PJ-34 (or control), and Ch-fractions prepared before or 10 min after UVC irradiation at 30 J/m² were subjected to IP for DDB2 followed by detection of XPC. Results are representative of two to three experiments with similar results. (E) Aberrant XPC localization in PARP-inhibited cells. GMU6 cells were locally irradiated with 100 J/m² UVC with or without PJ-34. Immunofluorescence microscopy allowed detection of XPC (green) and T-T (red) in DAPI-stained nuclei. The chart represents the percent of T-T spots, which are also positive for XPC, as pooled from three experiments, each in triplicate (n = 100–150 spots, data points are mean ± SE). (F) PARP inhibition decreases UVC-induced modification and stabilization of XPC. GMU6 cells treated with 10 µM PJ-34 (or control) were irradiated with 10 J/m² UVC, and Ch-fractions prepared at different time points were immunoblotted to detect unmodified and modified forms of XPC. Ponceau staining served as a loading control. Results shown here are from one of four experiments with identical results.

Because the PARP inhibitor disrupts the interaction between DDB2 and PARP-1 and blocks the departure of DDB2 from the UV-damaged DNA, we determined whether it would also affect the downstream actions of DDB2, such as its interaction with XPC or modification and stabilization of XPC to UV-damaged DNA (5, 6). Using DDB2-IP of GMU6 cells, we observed that PJ-34 significantly suppressed UV-induced association of DDB2 with XPC (Fig. 4D, lanes 6 and 8). Moreover, a slowly migrating heterogeneously modified XPC, often characterized as ubiquitinated XPC (5), was significantly increased in response to UV and suppressed in the PJ-34–treated cells (Fig. 4D, lanes 6 and 8). A possible consequence of this effect of PARP inhibition was evident on the recruitment and stabilization of XPC at the site of DNA damage after local UVC irradiation. In the normal GMU6 cells, the signal for XPC colocalized with subnuclear T-T by 10 min and sharply intensified by 30 min. In the PJ-34–treated cells, XPC was weakly localized with T-T at 10 min and failed to focus properly by 30 min (Fig. 4E). The quantification of XPC and T-T in the subnuclear spots revealed that PARP inhibition significantly reduced the colocalization of XPC with T-T (Fig. 4E). This influence of the catalytic function of PARP on the movement of XPC to the UV-damaged chromatin DNA was further confirmed by the effect of the PARP inhibitor on the presence of XPC in Ch-fractions of UV-irradiated GMU6 cells (Fig. 4F). In the normal cells, the XPC levels gradually declined from the peak values at 5–15 min up to 4 h, whereas in PARP-inhibited cells, XPC levels declined rapidly by 30–60 min. In addition, the band for modified XPC was significantly increased from 5 to 15 min in the normal cells but not in the PJ-34–treated cells (Fig. 4F). Thus, PARP inhibition disrupted the interaction of XPC with DDB2 and decreased its stabilization at the lesion site.

Discussion

We showed that, in response to UV irradiation, PARP-inhibited cells have (i) an impaired capacity to remove UV-induced DNA photolesions; (ii) a decreased level of interaction of DDB2 with XPC or PARP-1; (iii) an increased tendency for DDB2 to persist at the UV-damaged chromatin; and (iv) a decreased level of recruitment, modification, and localization of XPC to the damaged site. In addition, we show that (v) DDB2 and PARP-1 directly interact in the vicinity of the DNA lesion; (vi) DDB2 stimulates catalytic activity of PARP-1; and (vii) DDB2 is modified by PAR. Our results strongly indicate that PARP-1 is the principle player in the above responses, because cells specifically depleted of PARP-1 do not form detectable amounts of PARylated proteins in response to UV and are also inefficient at repair of UV-damaged genomic DNA. Our results are consistent with earlier reports that impaired PARP-1 function increases UV-induced skin cancer in mice (19), decreases cellular capacity to repair UV-induced DNA damage from viral reporter gene (20) or genomic DNA of CHO or triple negative breast cancer cells (11, 12), and decreases the clonogenicity in response to UV (20).

Our results together with previous reports suggest several possible ways in which PARP-1 can collaborate with DDB2 to increase the efficiency of GG-NER. (i) The PARylated PARP-1 or free PAR chains could serve as a scaffold on which PARylated DDB2 can interact with XPC, such as that suggested for their role with X-ray repair cross-complementing protein 1 (XRCC-1) in the base excision repair (21). Hence, PARP inhibitor or absence of PARP-1 could reduce participation of XPC in NER. (ii) The catalytic activation of PARP-1 and resultant PARylation of DDB2 could promote chromatin remodeling by DDB2. In fact, PARP inhibitor or PARP-1 depletion was recently shown to block chromatin remodeling of UV-damaged chromatin (7). In addition, the departure of DDB2 from the UV-induced lesion site was shown to be dependent on chromatin remodeling (22); hence, our result showing a delayed departure of DDB2 from the UV-damaged chromatin in PARP-inhibited cells supports an argument that catalytic function of PARP-1 plays a role in DDB2-mediated chromatin remodeling. This role of PARP-1 in chromatin remodelling would be in agreement with the reported role of PARylated proteins, such as amplified in liver cancer 1 (ALC-1) and aprataxin and PNKP like factor 1 (APLF-1) in stimulating their chromatin remodeling activity in DNA repair (23). (iii) The catalytic activity of PARP-1 and PAR formed locally around the lesion could be involved in the ubiquitination activity of the UV-DDB-E3 ligase complex, just as PARylation of targets has been shown to facilitate their modification by ubiquitin E3 ligase RNF146/Iduna (21). The ubiquitination of different proteins by UV-DDB-E3 ligase produces varying end results, e.g., ubiquitination of its own members DDB2, cullin 4A, and Rbx1 leads to a disengagement of the ligase complex from the damaged site; that of the histones H2A, H3, and H4 leads to chromatin remodeling (22, 24); and that of XPC helps in its recruitment and stabilization at the damaged site (24). Although a previous study has shown a chromatin destabilization effect of PARP inhibition (7), here we show its effect on reduced mobility of DDB2 and decreased modification and stabilization of XPC, all of which could be explained by inefficient ubiquitination activity of the UV-DDB-E3 ligase complex. (iv) Finally, a strong effect of PARP inhibition in suppression of the initial phase of repair of 6-4PP in BJ-hTert cells could be due to a specific influence of PARP-1 activation on NER at a given site of chromatin, because ubiquitination of XPC was recently shown to promote its relocalization from the intra- to the internucleosomal region to prioritize the repair at the latter site (6).

Based on our results and previous work, we propose a model for the role of PARP-1 and DDB2 at the lesion recognition step of GG-NER (Fig. 5). Immediately after UV irradiation, PARP-1, due to its sheer abundance and known capacity to be rapidly activated in response to different types of DNA damage (8), is likely to be one of the first proteins to arrive at the lesion and be basally activated within seconds of DNA damage (step 1). The arrival of DDB2, which occurs in a similar time frame as PARP-1, will cause a stronger activation of PARP-1, which will result in PARylation of many proteins including PARP-1 and DDB2. This PARylation will foster a direct association of DDB2 and PARP-1 that resists separation by ethidium bromide during IP (step 2). Because both proteins are known to independently bind to DNA containing CPD lesions, it is likely that these proteins would interact with the damaged DNA and with each other (step 2). One possible scenario to explain the downstream events is that the PARylated DDB2 as part of the UV-DDB-E3 complex will remodel the chromatin (step 3A), ubiquitinate DDB2 to reduce its affinity for the lesion (step 3B), and ubiquitinate XPC to promote its stabilization at the damaged site (step 4), which will result in an efficient GG-NER (step 5). As indicated in the model, many of the steps in this model have been shown by us and others (7), implying the role for PARP-1 in collaboration with DDB2 at the lesion recognition step to improve the efficiency of mammalian GG-NER. It will be interesting to examine whether PARP-1 also plays other roles in NER.

Fig. 5.

Model for cooperation of PARP-1 and DDB2 in GG-NER. See Discussion for details. Based on the present data and previous study (*ref. 7), different NER-related end results that are susceptible to PARP inhibition are indicated in the box.

Materials and Methods

Full details are provided in SI Materials and Methods.

Cells, Clones, Transient Transfections, and Mice.

SV-40–immortalized GM637 and XP-C (GM15983) human skin fibroblasts and untransformed XP-E (GM01389) cells were obtained from Coriell. The hTert-immortalized human skin fibroblasts were a gift from W. Hahn, Dana Farber Cancer Institute, Boston. PARP-1–replete GMU6 and PARP-1–depleted GMSiP cells have been reported (14). Details for FLAG-hPARP-1–expressing GMRSiP or mycDDB2 and FLAG-XPA–expressing GMU6-derived clones are in SI Materials and Methods. XP-E cells were transfected with pcDNA3-mycDDB2 plasmid for 24 h before irradiation. Four-week-old female SKH-1 hairless mice were from Charles River.

UV Irradiation, Analyses of DNA Repair, and Recruitment of NER Factors.

Global and local UVC irradiation of cells has been reported (9). The repair kinetics assays for T-T and 6-4PP in cellular genomic DNA by flow cytometry and immunofluorescence microscopy and detection of NER proteins after local irradiation are detailed in SI Materials and Methods. SKH-1 hairless mice were irradiated with 1,600 J/m² of UVB, and immunohistological analysis was performed on paraffin sections of skin.

Immunopurification, Cell Fractionation, and co-IP.

FLAG-hPARP-1, mycDDB2, and FLAG-XPA were expressed in cells and immunopurified from whole cell extracts using suitable antibodies and beads. Purified human GST-DDB2 was purchased from Abnova. Cell fractionation protocol to extract chromatin-bound proteins (Ch-fraction) is detailed in SI Materials and Methods. IP was carried out with 100–300 µg protein of Ch-fractions.

In Vitro PARP-1 Activation Reactions, pADPr Dot-Blot, and SouthWestern Blot.

The pADPr immunodot-blot (18) and in vitro PARP-1 activation reaction (15) were performed as described. The SouthWestern blot method was derived from the Activity-Western blot of PARP-1 (15).