Translocation of a UV-damaged DNA binding protein into a tight association with chromatin after treatment of mammalian cells with UV light

The major DNA repair process that removes structurally unrelated lesions from DNA in both prokaryotic and eukaryotic cells is the NER pathway. Genetic and biochemical studies with mammalian cells implicate more than 20 proteins in the complete reaction, and defects in many of these proteins are associated with various inherited human diseases characterized by defective DNA repair, such as xeroderma pigmentosum (XP), Cockayne’s syndrome (CS), and trichothiodystrophy. The best characterized proteins correspond to genetic complementation groups of XP (A to G), and are involved in the early steps of NER (Friedberg et al., 1995). Protein-protein interaction studies and in vitro reconstitution experiments using damaged DNA have shown that XPA, a damage recognition protein, first forms a complex with RP-A (a major human single-stranded DNA-binding protein); other proteins are then recruited to the lesion, including TFIIH (a transcription factor containing DNA helicases XPB and XPD), XPC, and two DNA endonucleases, XPG and XPF/ERCC1. The result of this process is the removal of a damaged oligomer ∼30 nucleotides long. In subsequent steps, the resulting DNA gap is filled in by DNA polymerase and ligase in a PCNA (proliferating cell-nuclear antigen)-dependent reaction (Ma et al., 1995; Wood, 1996, and references therein).

It has been suggested that in addition to XPA, yet another protein, UV-DDB, has a role in the damage-recognition step of NER because of its high binding affinity for 6-4 PD in vitro (Hirschfeld et al., 1990; Treiber et al., 1992; Reardon et al., 1993; Protic and Levine, 1993). However, the exact role of UV-DDB in NER is still unknown despite the availability of its purified protein components (Abramić et al., 1991; Hwang and Chu, 1993; Keeney et al., 1993) and their corresponding cloned genes (Takao et al., 1993; Lee et al., 1995; Dualan et al., 1995; Hwang et al., 1996). UV-DDB has been purified from primate cells as a homodimer of 127 kDa subunits, and as a heterodimer of the 127 kDa protein and a 48 kDa protein (Abramić et al., 1991; Keeney et al., 1993). p127 appears to be a DNA-binding protein, while the function of p48 in the formation of the binding complex is still unresolved (Reardon et al., 1993). The DNA binding activity of the complex (UV-DDB activity) is absent in cells from only a subset of XP group E patients, although cells from all XP-E patients tested thus far show a 40-60% reduction in unscheduled DNA synthesis (UDS) and moderate sun-sensitivity (Chu and Chang, 1988; Hirschfeld et al., 1990; Kataoka and Fujiwara 1991; Keeney et al., 1992). Microinjection of the purified UV-DDB protein into XP-E fibroblasts that lack the binding activity restored UDS to the levels found in normal human cells. However, this biochemical complementation had no effect on UDS in cells from XP-E patients that demonstrate binding activity nor in cells from patients in other XP groups (Keeney et al., 1994). Although single-base mutations have been identified in the gene encoding the p48 subunit in cells from three XP-E patients who lack UV-DDB binding activity (Nichols et al., 1996), this result could simply reflect polymorphism and the molecular basis of XP group E requires further definition. Moreover, while UV-DDB is not essential for in vitro reconstituted NER (the core reaction) (Mu et al., 1995; Guzder et al., 1995), the protein can stimulate repair synthesis two-fold when added to the reaction in small amounts (Abussekhra et al., 1995). However, larger amounts can have an inhibitory effect on repair (Wood, 1996). We reported previously that mammalian cells show an early dose-dependent inhibition of the extractable UV-DDB activity after treatment with UV light (Protić et al., 1989; Hirschfeld et al., 1990). Such a UV response appears to be common to UV-damaged DNA binding activities in a diversity of vertebrates because a similar inhibition of extractable binding activities that recognize either cyclobutane pyrimidine dimers (CPD) or non-CPD lesions has been detected in cells from fish to primates (McLenigan et al., 1993). We also suggested that this inhibition is probably due to translocation of the UV-DDB from a loose to tight chromatin/UV-DDB/UV-damaged DNA structure; this tight structure would prevent release of UV-DDB, with the consequence that it could not be extracted under standard conditions for 0.3 M salt nuclear extract preparation (Hirschfeld et al., 1990; McLenigan et al., 1993). Since our previous data concerned only UV-DDB activity, and not the protein harboring the activity, we designed the present study to address both the activity and the protein. To test the hypothe-sis of UV-DDB binding to damaged DNA directly, and to learn more about the location and dynamics of UV-DDB after UV treatment in vivo, we have carried out an extensive nuclear fractionation analysis and are now able to demonstrate translocation of the p127 component of UV-DDB, as well as of the DNA repair and replication proteins XPA, RP-A and PCNA, from low-salt (loose association) to high-salt (tight association) chromatin following UV-irradiation of primate cells. Furthermore, we detected a specific interaction between UV-DDB/p127 and RP-A in vitro and in vivo, and this interaction appears to enhance binding of both proteins to UV-damaged DNA in vitro. Finally, we studied extractable UV-DDB activity over the course of 48 hours after UV treatment in a variety of normal and repair-deficient human cell lines, and found a positive correlation between the kinetics of reappearance of extractable UV-DDB activity after UV treatment and the capacity of these cell lines to repair 6-4 PD.

Monolayer cultures of TC-7 cells, a clone of the African green monkey kidney cell line CV-1, were grown at 37°C and 10% CO₂ in Dulbecco’s modified Eagle’s medium (MEM) supplemented with 10% fetal bovine serum and 20 mM-glutamine. For UV-irradiation, cells were washed twice with phosphate-buffered saline (PBS), precooled for 15 minutes at 4°C, and exposed to 254 nm light for 8 seconds (total energy of 27.2 J/m²). This dose of UV was chosen because, in our preliminary experiments, it caused complete inhibition of UV-DDB activity in 0.3 M-salt nuclear extracts from monkey cells immediately after UV treatment.

For kinetic studies, normal (GM00011, GM01652, GM05757, GM00037 and its SV40-transformed cell line, GM00637), CS-A (CS3BE), XP-A (XP12BE), and XP-C (XP9BE) human fibroblasts were obtained from the Human Genetic Mutant Repository, Camden, NJ. The XP-A cell line XP12R0 and its revertant XP129 were generous gifts from K. Valerie (MCV-VCU, Virginia) and J. Cleaver (UCSF, California), respectively. Cells were grown and UV-irradiated as above, except that medium was supplemented with 20% fetal bovine serum and 2× MEM amino acids solution, and the following UV doses were used: 12 J/m² (normal human fibroblasts and CS3BE), 6 J/m² (XP12BE, XP9BE and GM00637), and 3 J/m² (XP12R0 and XP129).

TC-7 cells were harvested by trypsinization immediately after irradiation (UV 0-hours) or mock treatment. Nuclei were prepared from 5-7×10⁸ cells as described previously (Tamiya-Koizumi et al., 1989). Briefly, cells were swollen in 2 volumes of hypotonic solution containing 5 mM MgCl₂, 1 mM NaHCO₃, 1 mM phenylmethylsulfonyl fluoride (PMSF), and disrupted in a Dounce homogenizer with 8-10 strokes of pestle A. The homogenate was immediately adjusted with 2 M sucrose to yield 0.25 M. Crude nuclei were collected by centrifugation at 800 g for 10 minutes. Following one wash with isotonic buffer, the nuclear pellet was resuspended in 50 volumes of 1.7 M sucrose, 5 mM MgCl₂, and sedimented through a 1.7 M sucrose layer at 52,000 g for 60 minutes. The purified nuclei were washed twice with 0.25 M sucrose, 5 mM MgCl₂. Nuclei obtained in this way were further fractionated according to the modified procedure of Smith and Berezney (1982). Nuclei were resuspended to 1.6 mg/ml DNA in 25 mM Tris-HCl, pH 7.4, 0.25 M sucrose, 5 mM MgCl₂, 1 mM PMSF, and digested with 50 units/ml bovine pancreatic DNAse I (USB, Cleveland, OH) at 4°C for 16 hours. The nuclei were spun for 15 minutes at 1,000 g, the supernatant was saved, and the pellet extracted 3 times with low-salt buffer (10 mM Tris-HCl, pH 7.4, 0.2 mM MgCl₂, 1 mM PMSF) to yield soluble bulk chromatin and an insoluble low-salt pellet. The pellet was consecutively extracted 3 times with the same buffer as above, but with increasing concentrations of NaCl. The resulting supernatants are referred to as 0.3 M, 0.5 M and 2.0 M, respectively. To obtain the nuclear matrix, the high-salt pellet was finally extracted with low-salt buffer, 1% Triton X-100 and washed twice with buffer only. Supernatants from 3 extractions were combined into a single sample. Extractions were done for 15 minutes at 4°C by dropping magnetic stirrers into Eppendorf tubes containing the samples. Centrifugations were for 15 minutes at 1,000 g after the low-salt and 0.3 M-salt extractions, at 1,500 g after the 0.5 M-salt extraction, and at 6,000 g after all subsequent extractions. The nuclear matrix was dispersed by sonication using a Hert System sonicator (Misonix Inc., Farmingdale, NY), at the maximum setting for 3 times at 20 second intervals in low-salt buffer supplemented with 1 mM EDTA, 1 mM DTT.

For kinetic studies, human cells were harvested 0, 1, 3, 6, 24, and 48 hours after irradiation or mock treatment. Cells were washed with ice-cold PBS, scraped from the culture plates, and collected by centrifugation. Cell pellets were kept frozen at −80°C until preparation of whole-cell extracts. They were then resuspended in an equal volume of lysis buffer: 20 mM Hepes, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 1 mM DTT and a protease inhibitors cocktail, freeze-thawed (dry ice/37°C water bath) 5 times, and incubated on wet ice for 60 minutes with occasional mixing. The samples were spun at 4°C for 10 minutes and the supernatant was used as a whole cell extract. Protein concentration was determined by the Bio-Rad microassay (Bio-Rad Laboratories, Hercules, CA) using bovine serum albumin (BSA) as the standard.

For immunoprecipitation studies, the nuclear extract from mock or UV 0-hour TC-7 cells was prepared as follows: cells were washed once with solution A (10 mM Hepes, pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 1 mM DTT, 0.3 mM PMSF, 40 ng/ml leupeptin, 40 ng/ml pepstatin, 40 ng/ml chymostatin), and homogenized in 2 volumes of the same solution. Nuclei, recovered by centrifugation, were incubated in lysis buffer (100 mM Tris-HCl, pH 8, 100 mM NaCl, 1% Nonident P-40, 5 mM MgCl₂, 0.3 mM PMSF, 40 ng/ml leupeptin, 40 ng/ml pepstatin, 40 ng/ml chymostatin) at 4°C for 15 minutes and digested with 100 units/ml DNAse I at 37°C for 1 hour. The lysate of 1.2×10⁷ nuclei/ml was adjusted to a final concentration of 1 mM EDTA and clarified by centrifugation at 3,000 rpm at 4°C. The resulting supernatant is referred to as nuclear lysate.

Nuclear lysates and purified UV-DDB fractions (Takao et al., 1993), solubilized in NENT (100 mM Tris-HCl, pH 8, 100 mM NaCl, 1% Nonident P-40, 1 mM EDTA, 0.3 mM PMSF, 40 ng/ml leupeptin, 40 ng/ml pepstatin, 40 ng/ml chymostatin), were incubated with anti-p127 or anti-RP-A antibodies and immune complexes were absorbed to Protein A Sepharose (Pharmacia Biotech Inc., Piscataway, NJ), and Protein G agarose (Life Technologies Inc., Gaithersburg, MD), respectively. The precipitates were washed four times with NENT, resuspended in electrophoresis sample buffer, and analyzed by SDS-PAGE.

Western analysis was done using chemiluminescent detection as recommended by the manufacturer (Tropix, Bedford, MA). The following antibodies and dilution factors were used: polyclonal anti-bodies anti-p127 (1:250) and anti-XPA (1:5,000; kindly provided by R. Legerski, UT, Texas); monoclonal antibodies anti-PCNA (1:100; clone PC10), anti-lamin B (1:100; clone 101-B7) and anti-actin (1:200; clone JLA20), all from Oncogene Science Inc., Cambridge, MA; anti-RP-A 34A and 70C (0.5 mg/ml, kindly provided by R. Wood, ICRF, England; Kenny et al., 1990); goat anti-rabbit IgG (H+L) AP-conjugate (1:20,000; Bio-Rad Laboratories, Hercules, CA); goat anti-mouse AP-conjugate (1:10,000; Novagen, Madison, WI); and goat anti-mouse IgM (1:15,000; Boehringer Mannheim Corp., Indianapolis, IN).

Polyclonal anti-p127 antibodies were affinity-purified by absorption to a PVDF strip containing the pure 40 kDa C-terminal domain of the recombinant p127 protein (Takao et al., 1993), and elution with 100 mM glycine, pH 2.5, 150 mM NaCl, followed by neutralization with 1 M Tris-HCl, pH 8.0.

The band-shift assay of UV-DDB activity was carried out with a UV-irradiated double-stranded 60mer DNA oligonucleotide (‘60/54’), and protein/DNA complexes were separated on a 5% nondenaturing poly-acrylamide gel as described previously (Hirschfeld et al., 1990; Protić and Levine, 1993).

Photographic images of bands generated by the band-shift assay and immunoblotting were measured by densitometry using the Image program, version 1.51 (National Technical Information Service, Springfield, VA) on an Apple Macintosh Quadra 800 computer, and normalized per total amount of protein per fraction.

To gain insight into the nature of the early inhibition of extractable UV-DDB activity after UV treatment, we fractionated nuclei from mock- and UV-treated TC-7 cells into seven components differing in salt and detergent solubility (Fig. 1A) and assayed the extractable binding activity of each fraction. The majority (∼70%) of the UV-DDB activity in mock-treated cells is present in low-salt and 0.3 M-salt supernatants (Fig. 1B, Unirradiated; lanes b and c). The rest of the activity is in less soluble complexes and could be traced all the way to the nuclear matrix which contains ∼1% of the total binding activity (lanes d to g). In contrast, the majority of the detectable UV-DDB activity in nuclei from UV-irradiated cells was present in the high-salt chromatin (Fig. 1B, UV-irradiated; lanes d and e), while ∼30% of the activity was in the low-salt fractions (lanes a-c). Because the total DNA binding activity of UV-DDB in UV-irradiated cells was only ∼20% of that found in mock-treated cells, the in vitro band-shift assay could not reveal the true redistribution of the UV-DDB protein within nuclei after UV.

This loss of UV-DDB activity was not due to leakage into the hypotonic cell extract, as we have seen with 5-10% of the total nuclear activity from mock-treated cells, because examination of the cytosolic fraction from UV-irradiated cells did not uncover any detectable UV-DDB activity (gels not shown). To eliminate the possibility that this inhibition was primarily the result of UV-DDB binding to lesions on genomic DNA present in the extracted fractions, we tested the extracts in a modified band shift assay performed at 37°C (which allows both association and dissociation of the protein complex with UV-damaged DNA). Under such conditions, UV-DDB activity in irradiated cells slightly increased, but was still inhibited about 70% (gel not shown). Moreover, the total amount of genomic DNA in each fraction does not positively correlate with the extent of inhibition in the same fraction. In contrast, the 2 M salt extract of irradiated cells contained most of the binding activity (Fig. 1B, UV-irradiated; lane e) and also contained the highest amount of genomic DNA (not shown). Several other approaches, including replacement of DNase I treatment of nuclei with sonication, and mixing of cytoplasmic and nuclear extracts from untreated and UV-treated cells, did not affect the inhibition of extractable UV-DDB activity by UV.

In DNaseI and Triton X-100 supernatants, from UV-irradiated or unirradiated cells, we detected two novel fast-migrating binding activities specific for UV-damaged DNA, but these activities have not been analyzed further (Fig. 1B, lanes a and f).

To test for UV-induced translocation of UV-DDB, we carried out western analyses of the same fractions that had been tested in the binding assay using anti-p127 antibodies (Fig. 2A). In mock-treated cells, the fractions most abundant in p127 were the low-salt and 0.3 M-salt supernatants, containing 76% of the total protein (Fig. 2C). The Triton X-100 supernatant and the nuclear matrix had <1% of the total protein (which was visible only after longer film exposure; not shown). Such a distribution of p127 reflects the binding activity of the UV-DDB in these nuclear fractions. In UV-treated cells, the p127 was again found in all nuclear fractions despite the fact that UV-DDB 20 activity was primarily detected in the 2 M fraction (Fig. 1B). However, the 2 M fraction from UV-treated cells (Fig. 2A, UV-irradiated, lane e) appears to be enriched in p127 because it has 3 to 4-fold more p127 (∼8% of the total nuclear p127; Fig. 2C) than the corresponding fraction from unirradiated cells. Although only a small portion of the total nuclear p127 was translocated into the 2 M-salt extractable structure, this result was reproducible (three experiments) for different nuclear preparations, and the shift in p127 distribution was observed only when a gradual increase of salt was applied in the procedure for fractionation of nuclei. In the same fractions, the distribution of two control proteins, a cytoskeletal protein (actin) and lamin B, did not change after UV-treatment (Fig. 2A and C). (Lamin B is a marker protein for the nuclear matrix, and our immunoblotting detection reveals 55%, 30%, and 10% of total cellular protein in the nuclear matrix, detergent soluble and 2 M salt fractions, respectively; Fig. 2C.)

To examine whether the other members of the NER pathway respond to UV as does UV-DDB (p127), we analyzed in parallel the same nuclear fractions with antibodies against XPA, RP-A (its two subunits, p34 and p70) and PCNA. We chose to test for RP-A and XPA as marker proteins for the recognition/incision step of NER (Robins et al., 1991; Coverley et al., 1991; He et al., 1995; Li et al., 1995), and PCNA for the repair synthesis step (Miura et al., 1992; Nichols and Sancar, 1992; Shivji et al., 1992). All three proteins showed a redistribution after UV (Fig. 2B and C). While the fraction most enriched in PCNA was the 0.5 M-salt supernatant, translocated RP-A was mainly detected in the 2 M-salt supernatant. Longer film exposure also revealed traces of RP-A in the detergent-soluble fraction and the nuclear matrix, but no difference in RP-A amounts was observed between the two treatments (not shown). Similarly to p127, RP-A and PCNA were enriched in the high-salt chromatin about 3-fold (Fig. 2B, lanes d and e), which corresponds to 14-20% of their total nuclear content (Fig. 2C).

In contrast, the distribution of XPA, a protein exclusively localized in nuclei (Miyamoto et al., 1992), changed dramatically after UV. While in mock-treated cells, the DNase I supernatant contained all of the XPA, nuclear fractions from UV-irradiated cells had a more uniform distribution of the protein with ∼50% of the total XPA present in a 0.5 and 2 M salt extractable structure (Fig. 2B and C). However, the total amount of XPA protein in nuclei of UV-treated cells was ∼30% higher when compared to the amounts found in unirradiated cells (Fig. 2C). This difference may be due to leakage of the protein since a faint XPA signal was detected in the corresponding cytosol from unirradiated cells (not shown).

The similar pattern of redistribution after UV treatment found for p127 and RP-A (Fig. 2A,B and C) prompted us to examine whether these two proteins interact specifically under our experimental conditions. First, we did a western analysis of our UV-DDB fractions that had been eluted from UV-irradiated DNA cellulose during the process of UV-DDB purification (Takao et al., 1993) and found that they contain the p70 and p34 subunits of RP-A (Fig. 3A). (The presence of p11, the third RP-A subunit (Kenny et al., 1992), has not been tested here.) To eliminate the possibility that these proteins copurified only as a result of binding to the affinity column used for purification, we tested for a specific interaction between p127 and RP-A within the same fractions by coimmunoprecipitation analysis. Purified UV-DDB fractions were first incubated with affinity purified anti-p127, anti-p70 or anti-p34 antibodies; bound proteins in immunoprecipitates were then analyzed by SDS-PAGE; the same membrane was cut into three strips, and each strip was probed immunologically for a specific protein. As shown in Fig. 3B, p127 was coimmunoprecipitated with both p34 and p70 subunits of RP-A. Similarly, both RP-A subunits were detected as coimmunoprecipitates of p127.

The presence of a specific interaction between UV-DDB and RP-A was further confirmed by coimmunoprecipitation of p127 from the nuclear lysate using anti-RP-A antibodies (Fig. 3C). The amount of p127 coimmunoprecipitated with anti-p34 or anti-p70 antibodies was very low compared with the starting material, and did not increase even after cells received a high dose (100 J/m²) of UV (not shown), suggesting a constitutive interaction of the two proteins in the cell. When a similar analysis was performed using an anti-PCNA antibody, we did not detect either p127 in immunoprecipitates or PCNA in our UV-DDB purified fractions (not shown). Nonspecific trapping of immunoblotted proteins was not detected in any of the control lanes with Protein A or G only (Fig. 3B,C).

To test whether the specific interaction between UV-DDB and RP-A affects binding of UV-DDB to UV-damaged DNA, increasing amounts of purified RP-A were added to an RP-A-immunodepleted UV-DDB fraction and binding to UV-damaged DNA was assessed by the band-shift assay. Purified RP-A from HeLa cells bound to the UV-damaged DNA probe, and the complex migrated to about the same position on the gel as the UV-DDB/UV-DNA complex under our experimental conditions (Fig. 4, lanes 1 and 3). Addition of both proteins to the binding mixture enhanced the signal of the protein/UV-DNA complex about 5-fold and 3-fold when compared to UV-DDB or RP-A binding alone, respectively (Fig. 4, lanes 1-3). The enhanced binding is specific for UV-damaged DNA (lanes 14-16), and can be partly supershifted with anti-p70 antibodies (Fig. 4, lanes 5 and 6). Since the supershifted signal of the RP-A binding was about 2-fold greater when UV-DDB was present in the binding mixture, it appears that UV-DDB stimulates binding of RP-A to UV-damaged DNA. The residual (non-supershifted) binding was also enhanced (about 2.7-fold), suggesting coordinate stimulation of UV-DDB binding by RP-A. However, this interpretation is not certain because our anti-p127 antibodies do not supershift the UV-DDB (not shown) and therefore the enhanced DNA binding cannot be assigned to p127 alone.

In contrast to the above findings, the addition of E. coli single-stranded DNA binding protein (SSB) (USB, Cleveland, OH) or bovine serum albumin (BSA) (Sigma Chemical Co., St Louis, MO) to the binding mixture did not affect the binding of UV-DDB to UV-damaged DNA (Fig. 4, lanes 7-10), further suggesting the specificity of the UV-DDB/RP-A interaction in enhancing the binding of both proteins to UV-damaged DNA.

To further explore the dynamics of UV-DDB in NER, we studied the kinetics of its binding activity within 48 hours following UV exposure, in a variety of repair-deficient cell lines. Four normal human fibroblast lines (Fig. 5A) demonstrated a kinetic profile which correlates well with our previous results using monkey CV-1 cells (Hirschfeld et al., 1990); 60-80% of extractable UV-DDB activity was lost upon UV treatment but recovered to normal levels within 3 to 6 hours. A similarly normal profile of recovery of the activity was obtained with extracts from CS-A fibroblasts, a cell type that is defective in preferential repair of a subset of DNA lesions in transcriptionally active genes (van Hoffen et al., 1993; Henning et al., 1995); UV-DDB activity was inhibited 50% immediately after UV, but almost complete recovery occurred by 6 hours (Fig. 5B).

UV radiation is highly cytotoxic to cells from most XP groups due to low levels of their NER (Friedberg et al., 1995; Ma et al., 1995); therefore, instead of the 12 J/m² dose that was used for normal and CS-A fibroblasts, here we used a 2- to 4-fold lower fluence of UV. At these lower doses, normal cells showed slight or no inhibition of UV-DDB activity (not shown). XP-A cells, which have severe defects in repair either in the bulk DNA or in active genes (Miura et al., 1992; Evans et al., 1993; Wood, 1996), had 50% UV-DDB activity at UV 0-hours that declined to 30% by 6 hours post-treatment (Fig. 5B). With longer culturing times of XP12BE cells, we either saw no further change in this extractable activity or a very slow but incomplete recovery. The UV-DDB activity profile for XP-C fibroblasts, a cell type that is defective in repair of bulk DNA but proficient in repair of transcriptionally active genes (Venema et al., 1991; Evans et al., 1993; van Hoffen et al., 1995), was similar to that found for XP-A cells: UV-DDB activity was at near normal levels (∼80%) early after UV treatment, but then dropped dramatically to ∼10% by 6 hours. Further recovery of the activity was slow and incomplete by 48 hours (Fig. 5B).

We also tested XP129, a UV-resistant revertant of the SV40-immortalized XP-A cell line XP12RO (Cleaver et al., 1987). The XP12RO cells contain a termination codon, resulting in no detectable level of XPA protein, and no evidence of removal of 6-4 PD and CPD, although a very low level of repair synthesis was detected. The reversion of the parental mutation in XP129 allows the cells to make a reduced amount of full length XPA protein, and might also contribute to its altered substrate specificity; the revertant shows normal repair of 6-4 PD and normal repair synthesis (Jones et al., 1992; McDowell et al., 1993). In contrast to the delayed recovery of UV-DDB activity in the parental line, which has also been seen with XP12BE cell lines, the revertant’s activity profile is similar to that of the corresponding SV40-transformed normal cell line GM00637 (Fig. 5C). Even though recovery of UV-DDB activity in both XP-A cell lines, XP12RO and XP12BE, was delayed by 48 hours (Fig. 5B,C), about 60% of the activity was recovered in XP12RO cells (in contrast to 30% in XP12BE). We do not have an explanation for this difference between the two XP-A lines; although different mutations have been identified in the XPA gene in these two cell lines, neither line expresses XPA mRNA, and they both have negligible residual repair (States and Myrand, 1996; McDowell et al., 1993).

We have studied the nuclear dynamics of UV-DDB and several other NER and DNA replication proteins before and immediately after UV treatment of mammalian cells. Following irradiation, the p127 component of UV-DDB, as well as XPA, RP-A and PCNA, were translocated to different extents within the nuclei from low- to high-salt chromatin, suggesting that these proteins move to a tight chromatin association, presumably with UV-damaged DNA. In contrast to the dramatic translocation of XPA, which involved more than 50% of the total nuclear protein, only a small fraction of p127 was translocated to the high-salt chromatin (Fig. 2A,B,C), a result that could possibly be attributed to the different extent of involvement of these proteins in NER. XPA has binding affinity for a broad spectrum of DNA lesions (Jones and Wood, 1993), and is a key protein for assembling the NER complex (Matsuda et al., 1995; He et al., 1995; Li et al., 1995). So far, a role in assembly is the only known biological function of this protein. Therefore, it is to be expected that virtually all of this protein will be recruited to chromatin after UV damage, as is demonstrated in this study (Fig. 2). UV-DDB is an abundant cellular protein complex with high affinity for 6-4 PD in vitro (Abramić et al., 1991; Reardon et al., 1993), but the role of UV-DDB in NER has yet to be demonstrated clearly, even though the data here and elsewhere suggest that such a role is likely. At our UV doses, 6-4 PD represent only about 25% of the total UV-induced DNA lesions (Friedberg et al., 1995) and hence the requirement for the protein would be less in NER. The amount of p127 that we have observed to translocate into high-salt chromatin after UV (Fig. 2A) correlates with the results of Keeney et al. (1994), who restored the normal level of UDS in XP-E cells that lack UV-DDB activity by microinjecting less than 10% of the normal cellular content of UV-DDB. Park et al. (1996) reported that by using indirect immunofluorescence detection, the core NER protein XPG also shows dynamic regulation of its intranuclear distribution within 2 hours after a moderate UV-C dose (10 J/m²).

RP-A and PCNA, besides their roles in NER (Coverley et al., 1991; Shivji et al., 1992 and 1995), are involved in other cell processes (Kenny et al., 1990; Hurwitz et al., 1990); thus, only a fraction of these proteins (14-20%) is moved into a tightly bound compartment after UV (Fig. 2). This result is consistent with previous reports that an increased association of PCNA with nuclei occurs after UV irradiation of nonreplicating DNA in normal human fibroblasts; only a small proportion of PCNA is tightly bound to the nucleus (Toschi and Bravo, 1988). The immunostaining of PCNA in various XP cell lines after UV irradiation correlates with their capacity for incision of damaged DNA (Miura et al., 1992; Aboussekhra and Wood, 1995). Taken together, these findings support the notion that only a fraction of PCNA and RP-A, as well as UV-DDB proteins, tightly associated with chromatin after UV, are involved in NER. The similar shift of p127 to that of RP-A and PCNA could be attributed to UV-DDB having an accessory role in NER (Aboussekhra et al., 1995), but another role(s) in other cell processes.

The increased levels of UV-DDB/p127, XPA, RP-A and PCNA associated with high-salt chromatin after UV treatment were due to redistribution of these proteins within the nuclei because the total amount of each nuclear protein, with the exception of XPA, was similar whether extracted before or after irradiation (Fig. 2C). Much data, especially from studies of in vitro interactions of eukaryotic NER proteins, suggest that a partial (He et al., 1995; Nagai et al., 1995) or complete ‘repairosome’ (the protein complex essential for NER) exists constitutively in the absence of damaged DNA (Svejstrup et al., 1995). Consistent with this evidence, our findings with the subset of proteins that likely participate in the formation of a repairosome in vivo suggest that upon irradiation of mammalian cells, such a repairosome becomes translocated from a loose to tight association with genomic DNA.

The distribution of UV-DDB activity within nuclear fractions (Fig. 1B) correlated with our initial observation (Protić et al., 1989; Hirschfeld et al., 1990) that only a trace of UV-DDB activity could be detected in low- and 0.3 M salt extracts after UV-treatment. Although our immunoblotting data confirmed that we recovered all p127 from the nuclei of irradiated cells (Fig. 2), 80% of control UV-DDB activity remained inhibited (Fig. 1B). At present, we cannot exclude a role for the p48 subunit of UV-DDB (Keeney et al., 1993, 1994; Nichols et al., 1996), nor a role for RP-A, since we found that UV-DDB and RP-A specifically interact in vivo (Fig. 3C). Further studies are needed to elucidate the mechanism of inhibition of UV-DDB activity, specifically related to UV, and its importance for NER. However, ionizing radiation that does not create high-affinity binding sites for UV-DDB had no effect on UV-DDB binding activity (V. Rapic-Otrin and M. Protić, unpublished data).

Here we have uncovered for the first time a specific interaction between the p127 protein component of UV-DDB and the p70 and p34 subunits of RP-A. These interactions were detected, in vitro and in vivo, by coimmunoprecipitation from UV-DDB purified fractions and nuclear lysates, respectively (Fig. 3B and C). The protein-protein interactions occur constitutively in vivo, and are not stimulated by UV treatment (Fig. 3C), which may account for the unenhanced level of copurified RP-A in the UV-DDB purified fraction from UV-irradiated cells, compared to mock treated cells (Fig. 3A). This is also consistent with the notion that NER may proceed by interactions of intermediate subassemblies. Most likely only a fraction of the nuclear p127 and RP-A (8 and 20%) that is translocated into a tight complex with chromatin after UV (Fig. 2), presumably involved in NER, is also involved in this protein-protein interaction. The very low level of coprecipitation of p127 with RP-A in vivo (Fig. 3C), and the fact that RP-A-immunodepletion does not immunoprecipitate all p127 from the UV-DDB purified fraction (this fraction still exhibits UV-DDB binding activity, Fig. 4, lane 1), support this notion. In fact, we cannot eliminate the possibility that the interaction between UV-DDB and RP-A is mediated by the p48 component, especially since the UV-DDB fractions used contain other protein bands (Fig. 3A).

RP-A is primarily a single-stranded DNA-binding protein (Kenny et al., 1990), but it also binds to bulky lesions like cisplatin- (Clugston et al., 1992), AAAF- (He et al., 1995), and UV-damaged DNA (Burns et al., 1996). We also detected preferential binding of RP-A to UV-damaged versus undamaged DNA (Fig. 4, lanes 17-19). However, human RP-A purified from HeLa cells, which we used, has higher binding activity than an RP-A purified from E. coli; we obtained saturation of binding to a UV-damaged-DNA probe with 10 ng of RP-A (not shown), while 70-100 ng of an RP-A fusion protein were required (Burns et al., 1996). Although UV-DDB itself has high affinity for 6-4 PD, our results suggest that the combination of the two interacting proteins enhanced the binding to UV-damaged DNA shown by either protein alone (Fig. 4). At this point, we can only speculate that the RP-A/UV-DDB interaction leads to some conformational change in one or both proteins that results in its/their enhanced binding affinity for damaged DNA, which would be consistent with our earlier finding that the UV-DDB monomer is an inactive form of UV-DDB protein with respect to complexing with damaged DNA (Abramić et al., 1991; Takao et al., 1993). Substitution of human RP-A for SSB protein does not affect the binding of UV-DDB to UV-damaged DNA, further supporting the significance of the UV-DDB/RP-A interaction for enhancing the binding to UV-damaged DNA. Additional evidence favoring the role of this interaction in NER awaits a more detailed cell-free reconstruction analysis.

Several groups have shown that XPA binds specifically to the p70 and p34 subunits of RP-A, both in vitro and in vivo (Matsuda et al., 1995; He et al., 1995), although only the inter-action with p70 seems to be essential for NER (Li et al., 1995). The interaction enhances the otherwise low affinity of XPA for damaged DNA (Jones and Wood, 1993), and provides a mechanistic basis for the role of RP-A in the initial step of NER. This similarity of UV-DDB/RP-A to XPA/RP-A interactions supports our earlier hypothesis that UV-DDB has a role in the early damage-recognition step of NER.

Kazantsev et al. (1996), reported that a high concentration of RP-A can correct the in vitro repair defect of XP-E cell extracts, but found no RP-A mutations in an XP-E cell line that lacked UV-DDB binding activity. The authors propose that the XPE gene product acts as a molecular chaperone to aid in the assembly of RP-A with the other components of the DNA-bound repairosome; given a high enough level of RP-A, XPE would be unnecessary. The specific UV-DDB/RP-A interaction which we reported in this paper does not exclude this notion, but the fact that RP-A itself binds to 6-4 PD in vitro (Burns et al., 1996), could account for the observed complementation.

We have found previously that the kinetics of recovery of UV-DDB activity measured in cell extracts in vitro positively correlates with the kinetics of NER in vivo (Hirschfeld et al., 1990; McLenigan et al., 1993). Now we have extended this analysis in order to determine whether a correlation can be found between the kinetics of UV-DDB recovery and a specific type of NER. As seen before with monkey cells (Hirschfeld et al., 1990), normal human fibroblasts recover UV-DDB activity 3-6 hours after UV treatment, the time needed for repair of 6-4 PD in primate cells (Friedberg et al., 1995). In contrast, cells deficient in some aspect of NER showed two patterns of recovery of UV-DDB activity: XP-A and XP-C cells had delayed and incomplete recovery of the binding activity even 48 hours post-irradiation, while CS-A cells and an XPA revertant line, XP129, expressed a pattern similar to that of normal human fibroblasts.

The absence of a functional XPA protein alters the recovery of UV-DDB activity after UV (Fig. 5B and C), suggesting that UV-DDB might cooperate with XPA in recognizing 6-4 PD. Additional evidence may be found in the XP-A revertant (XP129) cells, which have acquired the ability to repair 6-4 PD in the genome overall and CPD to a certain extent, but only in the transcribed strand of active genes (Cleaver et al., 1987; Jones et al., 1992; Lommel and Hanawalt, 1993), and have normal recovery of UV-DDB activity (Fig. 5C). Furthermore, when XP-A is the limiting factor, as in XP129 cells, the presence of UV-DDB might promote the repair of 6-4 PD in preference to other photoproducts. The cooperative efforts of XPA and UV-DDB proteins may explain the greater efficiency at which 6-4 PD are repaired when both proteins are present in an in vitro reconstituted NER assay (Aboussekhra et al., 1995). As with XP-A cell lines, the fact that the absence of a functional XPC protein, which is involved in the repair of bulk genomic DNA and in the repair of the nontranscribed strand of active genes (Evans et al., 1993; Venema et al., 1991; van Hoffen et al., 1995), affects the normal profile of the UV-DDB binding recovery (Fig. 5B), offers indirect evidence that UV-DDB may be involved in repair of bulk DNA. Further evidence in support of UV-DDB involvement in efficient removal of 6-4 PD from the genome overall is the normal pattern of recovery of UV-DDB activity in CS-A cells, which is most likely a reflection of their normal repair of 6-4 lesions in the bulk genome (Parris and Kraemer, 1993).

We thank Dr Richard Wood for providing us with purified RP-A and for useful discussions throughout this work.