Expression of Wild-type p53 Is Required for Efficient Global Genomic Nucleotide Excision Repair in UV-irradiated Human Fibroblasts*

We have shown previously that Li-Fraumeni syndrome fibroblasts homozygous for p53 mutations are deficient in the removal of UV-induced cyclobutane pyrimidine dimers from genomic DNA, but still proficient in the transcription-coupled repair pathway (Ford, J. M., and Hanawalt, P. C. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 8876–8880). We have now utilized monoclonal antibodies specific for cyclobutane pyrimidine dimers or 6-4 photoproducts, respectively, to measure their repair in UV-irradiated human fibroblasts. Cells homozygous for p53 mutations were deficient in the repair of both photoproducts, whereas cells heterozygous for mutant p53 exhibited normal repair of 6-4 photoproducts, but decreased initial rates of removal of cyclobutane pyrimidine dimers, compared with normal cells. The specificity of the effect of wild-type p53 on nucleotide excision repair was demonstrated in a p53 homozygous mutant cell line containing a tetracycline-regulated wild-type p53 gene. Wild-type p53 expression and activity were suppressed in the presence of tetracycline, whereas withdrawal of tetracycline resulted in the induction of p53 expression, cell cycle checkpoint activation, and DNA damage-induced apoptosis. The regulated expression of wild-type p53 resulted in the recovery of normal levels of repair of both cyclobutane pyrimidine dimers and 6-4 photoproducts in genomic DNA, but did not alter the transcription-coupled repair of cyclobutane pyrimidine dimers. Therefore, the wild-type p53 gene product is an important determinant of nucleotide excision repair activity in human cells.

The molecular mechanisms that regulate the responses of normal and neoplastic human cells to DNA damage are critical determinants for the biological end points of mutagenesis and cell death. The p53 tumor suppressor gene plays a central role in mediating the cellular responses to DNA damage in mammalian cells, as demonstrated by its ability to regulate cell cycle progression and programmed death in cells exposed to DNA damaging agents (1)(2)(3)(4).
A wide variety of endogenous and exogenous agents cause damage to DNA, and multiple enzymatic processes exist to repair these different lesions. The most versatile and ubiquitous mechanism for DNA repair is nucleotide excision repair (NER), 1 which operates to remove many types of lesions, including UV-induced cyclobutane pyrimidine dimers (CPDs) and 6-4 pyrimidine-pyrimidone photoproducts (hereafter referred to as 6-4 photoproducts). Such lesions may pose as structural blocks to transcription and replication, and they may also result in mutations if translesional replication occurs or if they are not repaired correctly. NER is heterogeneous within the mammalian genome, and CPDs are more efficiently removed from actively expressed genes than from unexpressed regions of the genome (5,6). Generally, the transcribed strand of an active gene is repaired more rapidly than the non-transcribed strand (7), a process termed transcription-coupled repair. In normal human cells, both CPDs and 6-4 photoproducts are removed from genomic, non-transcribed DNA through a pathway referred to as global genomic NER.
We and others have presented evidence for a direct role for p53 in NER, potentially unrelated to cell cycle and cell death events (8 -10). For example, the effects of mutations in the p53 gene on NER and cellular sensitivity to UV irradiation were examined in primary human skin fibroblasts from patients with the cancer prone disorder Li-Fraumeni syndrome (LFS), which were heterozygous for mutations in one allele of p53, and in derived sublines expressing only mutant p53 (8). The p53 homozygous mutant cells were severalfold more resistant to UV cytotoxicity and exhibited much less UV-induced apoptosis than did the primary p53 heterozygous mutant cells or than normal skin fibroblasts expressing only wild-type (wt) p53. However, the p53 homozygous mutant cells were deficient in the global genomic NER of CPDs compared with normal cells or to p53 heterozygous mutant cells. Similarly, the rate of repair of the non-transcribed strands of the actively expressed dihydrofolate reductase (DHFR) gene and the p53 gene in p53 homozygous mutant cells was decreased relative to that in normal fibroblasts, consistent with the decreased overall genomic repair level. However, the p53 homozygous mutant cells retained the ability to perform transcription-coupled repair. Thus, p53 homozygous mutant cells were deficient in global genomic NER of CPDs, whereas transcription-coupled repair remained intact.
To further characterize the effect of loss of wt p53 function on global genomic NER, we used a different method allowing for the measurement of removal of both CPDs and 6-4 photoproducts from genomic DNA of UV-irradiated cells. Whereas we previously used alkaline sucrose sedimentation analysis of T4 endonuclease V (TEV)-treated DNA to measure removal of CPDs, we have now utilized monoclonal antibodies which spe-cifically recognize CPDs and 6-4 photoproducts to measure their rate of removal from human fibroblasts that are wildtype, heterozygously mutant, or homozygously mutant for the p53 gene. Furthermore, to assess the role of the wt p53 gene product in modulating NER, analyses of both global and transcription-coupled repair have been performed in a LFS p53 homozygous mutant cell line containing a stably integrated tetracycline (Tet)-regulated wt p53 gene.
Our results show that loss of wt p53 activity results in a decrease in the global genomic NER of both CPDs and 6-4 photoproducts, but not transcription-coupled repair of CPDs, and that the expression of functional wt p53 protein in a p53 mutant human fibroblast cell line results in restoration of normal levels of global genomic repair of both photoproducts, as well as enhanced cell cycle arrest and apoptosis following UV irradiation.

EXPERIMENTAL PROCEDURES
Cell Lines-All human fibroblasts were maintained as exponentially growing cultures in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and incubated at 37°C in 5% CO 2 . LFS skin fibroblast cell lines (originally obtained from Dr. Michael Tainsky, M. D. Anderson Cancer Center, Houston, TX) from patient MDAH041 (hereafter termed 041 wt/mut) and MDAH087 (termed 087 wt/mut) were finite cell populations heterozygous for single base mutations in p53 at codons 184 and 248, respectively (11). Spontaneously immortalized derivatives of each LFS cell line homozygous for mutant p53 have been described previously, and are termed 041 mut and 087 mut, respectively (8). Early passage GM38 normal primary human diploid skin fibroblasts, which express only wild-type p53, were obtained from the NIGMS Human Genetic Mutant Cell Repository (Coriell Institute for Medical Research, Camden, NJ). Transformed GM2096-SV3 cells from a patient with xeroderma pigmentosum (XP) complementation group C were obtained from Dr. Dan Canaani (12). TR9-7 cells, provided by Dr. George Stark (Cleveland Clinic Foundation, Cleveland, OH), were constructed from LFS 041 mut cells into which a Tet-regulated system for expression of wt p53 was stably transfected (13). TR9-7 were subcloned, and a single clone demonstrating optimal regulation of wt p53 expression was selected for use in all experiments described, and termed 041 TR. This cell line was grown continuously in the presence of 600 g/ml G418 and 50 g/ml hygromycin to maintain selection pressure for the two stably integrated plasmid constructs containing the wt p53 cDNA together with the neomycin resistance gene, and the Tetregulated transactivator together with the hygromycin resistance gene, respectively. 041 TR cells were cultured in the continuous presence of 2 g/ml Tet when suppression of wt p53 expression was desired.
Western Blotting-For analysis of p53 and p21CIP1/WAF1 (hereafter termed p21) protein levels, total cellular protein was isolated by lysing cells in a buffer consisting of 1% Triton X-100, 50 mM Tris (-HCl, pH 7.5), 150 mM NaCl, 10 mM EDTA, 1 mM phenylmethanesulfonyl fluoride, 10 g/ml pepstatin, 5 g/ml leupeptin, and 100 g/ml aprotinin (Sigma). The lysed cells were centrifuged at 15,000 ϫ g for 15 min, and the supernatants collected. Protein concentrations were determined by the Bradford method, and equal amounts (75 g) of protein were separated by 12% SDS-polyacrylamide gel electrophoresis (14) and electroblotted to a nitrocellulose membrane (15). Immunoblotting was performed by incubating the membranes for 1 h with mouse monoclonal antibodies to human p53, diluted by 1/200 in phosphate buffered saline (PBS) (DO-1, Santa Cruz Biotechnology), and human p21, diluted by 1/1000 in PBS (Ab-1, Santa Cruz Biotechnology), followed by incubating the membranes for 1 h with horseradish peroxidase-conjugated antimouse secondary antibody diluted by 1/5000 in PBS. Proteins were detected using enhanced chemiluminescence and autoradiography according to the manufacturer's protocol (Amersham Corp.). The relative amounts of detected proteins were determined by scanning densitometry of autoradiographs, using NIH Image analysis software.
Global Genomic NER Immunoassay-The relative number of CPDs and 6-4 photoproducts in total genomic DNA from cells collected at various times following UV irradiation was determined using an immunoblot assay. Briefly, exponentially growing cells were prelabeled with [ 3 H]thymidine, washed with PBS, and irradiated with 10 J/m 2 UV using a 15-watt germicidal UV lamp delivering predominately 254 nm light. Cells were either processed immediately or incubated in growth medium for various periods. At appropriate times, cells were lysed in a solution containing 10 mM Tris (-HCl, pH 7.5), 1 mM EDTA, 0.5% SDS, 0.1 mg/ml proteinase K, and 0.1 mg/ml RNase, at 37°C. Genomic DNA was isolated by phenol extraction and ethanol precipitation, and DNA concentrations and specific radioactivity were determined. DNA was denatured by boiling for 5 min and placing on ice (since the monoclonal antibodies recognize UV photoproducts in single-stranded DNA), and equal amounts of genomic DNA from each sample in equal parts Tris-EDTA and 20 ϫ SSPE were fixed to a Hybond N ϩ nylon membrane (Amersham Corp.) in triplicate using a slot-blot apparatus (0.1 g of DNA/slot for detection of CPDs, 1 g of DNA/slot for detection of 6-4 photoproducts). Equal amounts of genomic DNA from unirradiated cells were also loaded for each experiment to control for nonspecific DNA binding of the monoclonal antibodies. The membrane was incubated for 1 h with mouse monoclonal antibodies specific for either CPDs (TDM-2) or 6-4 photoproducts (64M-2), at a dilution of 1/2000 in PBS (antibodies provided by Dr. Toshio Mori, Nara Medical University, Nara, Japan; Ref. 16). Horseradish peroxidase-conjugated secondary antibody at a dilution of 1/5000 in PBS, enhanced chemiluminescence (Amersham Corp.), and phosphorimager analysis (Bio-Rad model GS-363) were employed for detecting the primary antibodies. Following antibody detection, equal DNA loading to each slot of the membrane was confirmed by scintillation spectrophotometry of individual pieces cut from the membrane. Data from triplicate DNA samples from at least three different experiments for each cell line were averaged, and consistently exhibited less than 5% variation.
Analysis of Strand-specific DNA Repair-Repair of CPDs was examined within the transcribed and non-transcribed strand of the 20kilobase pair KpnI restriction fragment spanning the central region of the DHFR gene, using methods previously described (7,8). Cells were UV-irradiated with 10 J/m 2 , lysed immediately for an initial sample or incubated in growth medium containing 5-bromodeoxyuridine (BrdUrd) to density label newly replicated DNA, and then lysed at various times. Density labeling was performed during repair periods to allow unreplicated DNA to be isolated by cesium chloride isopycnic density gradient sedimentation (17). The frequency of induction of CPDs and their rate of removal from the transcribed and non-transcribed strands of the human DHFR gene was measured by treating purified KpnI-digested DNA with TEV, and then quantifying the reappearance of the fulllength restriction fragments in DNA from cells allowed various times to remove CPDs from their DNA. KpnI-treated samples from each time point were treated or mock-treated with TEV, electrophoresed in parallel under denaturing conditions, Southern transferred to a membrane, and hybridized with 30 ϫ 10 6 cpm 32 P-labeled strand-specific RNA probes generated by transcription in vitro of the plasmid pGEM0.69EH (8). The ratio of full-length restriction fragments in the TEV treated and untreated samples was determined by phosphorimager analysis (Bio-Rad model GS-363), and was used to calculate the average number of CPDs (endonuclease-sensitive sites) per fragment using the Poisson expression.
Flow Cytometry-For analysis of cell cycle distribution and apoptosis, both floating and adherent cells (Ն1 ϫ 10 6 cells/time point) were collected and fixed in 70% ethanol in PBS, treated with 100 g/ml RNase, and stained with 20 g/ml propidium iodide. For each sample, at least 10,000 cells were analyzed for DNA content using a Coulter EPICS 753 flow cytometer (Coulter Electronics). The percentage of cells in sub-G 1 , G 1 , S, and G 2 /M was determined using an EASY2 computer system (Coulter Electronics).
Cytological Examination by Fluorescence Microscopy-To determine if UV treatment resulted in morphological changes characteristic of apoptosis, cultured cells were irradiated or not with 20 J/m 2 UV, incubated for 48 h, fixed in 70% ethanol in PBS, stained with the fluorescent DNA stain Hoeschst 33258 (0.5 g/ml), and examined microscopically. Cells were judged to be apoptotic if they displayed the following morphological characteristics: interphase cells that possessed hypercondensed marginated masses of DNA along the inner surface of their nuclear membranes, cell surface blebbing, and overall cell shrinkage due to reductions in cytoplasmic volume.

p53 Protein Levels and Transcriptional Activity in Normal and p53 Homozygous Mutant Human Fibroblast Cell Lines following UV Irradiation-
The effect of UV irradiation on p53 protein levels and p53-dependent transcriptional activity in normal human primary fibroblasts and in the two p53 homozygous mutant human fibroblast cell lines derived from LFS patients was examined by Western blot analysis of protein extracts from cells prior to, and at various times following, treatment with 20 J/m 2 UV irradiation, using monoclonal antibodies to p53 and to p21, which is transcriptionally regulated by wt p53 (18). In GM38 normal human fibroblast cells, wt p53 protein levels were induced by approximately 10-fold following UV irradiation, with peak levels occurring between 12 and 24 h (Fig. 1A). In addition, levels of p21 protein were also induced by approximately 8-fold following UV irradiation, reaching a maximum at 24 h. LFS 041 mut cells contain a frameshift mutation at codon 184 of the p53 gene, resulting in a truncated message (19). Western blotting analysis of protein extracts from these cells demonstrated no detectable p53 protein prior to or following UV irradiation, nor any induction of the p21 gene product (Fig. 1B). LFS 087 mut cells contain a missense Arg 3 Trp substitution mutation at codon 248 of the p53 gene (19). Western blotting analysis of protein extracts from these cells demonstrated constitutive overexpression of the mutant protein both prior to and following UV irradiation, but no induction of the p21 gene product following UV irradiation (Fig. 1C). Therefore, UV irradiation of normal human fibroblasts results in a strong and persistent induction of p53 protein levels and transcriptional activity. Furthermore, in human fibroblasts, UV induction of p21 expression appears to be dependent upon wt p53 activity.
Global Genomic DNA Repair in p53 wt, p53 Heterozygous Mutant, and p53 Homozygous Mutant Human Fibroblasts following UV Irradiation-The induction and repair of UV-induced CPDs and 6-4 photoproducts in genomic DNA was determined using an immunoblot assay and monoclonal antibodies specific to these lesions. Normal human diploid fibroblasts containing wt p53 efficiently repaired both CPDs and 6-4 photoproducts, removing a greater percentage of 6-4 photoproducts at earlier times than CPDs (Fig. 2), as expected from previous studies (20). As a negative control, repair in human fibroblasts mutant for the XP-C gene was also measured. The XP-C gene product is required for global genomic NER, but not for the transcription-coupled repair pathway (21); as expected, these cells showed very low levels of removal of both CPDs and 6-4 photoproducts (Fig. 2). Both LFS cell lines homozygous for p53 mutations showed a reduced rate and extent of repair of both UV-induced photoproducts compared with normal cells. Repair of CPDs in these cells was only slightly better than in the XP-C cells ( Fig. 2A), while repair of 6-4 photoproducts was substantially better than in the XP-C cells, but significantly lower than in normal cells (Fig. 2B). Both LFS cell lines heterozygous for mutant p53 exhibited normal repair of 6-4 photoproducts, but slightly decreased initial rates of removal of CPDs, compared with normal cells.
Characterization of Human 041 TR Fibroblasts-To determine the specificity of wt p53 for regulation of global genomic NER and apoptosis following UV irradiation, a derivative of the LFS 041 mut cell line was used, in which the expression of wt p53 could be regulated. In this cell line, 041 TR, suppression or induction of wt p53 activity was tightly controlled by treating cells with non-cytotoxic levels of Tet. In the presence of 2 g/ml Tet, p53 and p21 protein levels were nearly undetectable (Fig.  3A, 0 h lane), similar to the parental 041 cell line (Fig. 1B). However, following withdrawal of Tet, p53 protein levels increased by approximately 10-fold, reaching maximum levels at 24 h (Fig. 3A). p21 gene expression was also induced by approximately 12-fold, peaking at 24 -48 h. Therefore, wt p53 expression and activity were specifically induced in this cell line by withdrawal of Tet. Furthermore, in the presence of Tet, UV irradiation did not induce p53, and p21 levels increased only slightly, if at all (Fig. 3B). When 041 TR cells were UVirradiated 24 h following withdrawal of Tet, a 2-fold and 4-fold further increase in p53 and p21 protein levels occurred, respectively (Fig. 3C).
To further characterize the functional status of the wt p53 gene in this cell line, the effects of withdrawing Tet on other p53 dependent processes were examined. One of the most clearly defined roles of p53 is the inhibition of DNA replication following DNA damage. We therefore investigated DNA replication in UV-irradiated 041 TR cells. Cells were incubated with BrdUrd following UV irradiation (10 J/m 2 ). Newly replicated, density-labeled DNA was resolved from parental DNA by isopycnic cesium chloride gradients (17). As can be seen (Fig.  4), in the presence of Tet, a significant amount of DNA replication occurred within 24 h of UV irradiation, whereas follow- ing withdrawal of Tet and induction of wt p53, DNA replication following UV irradiation was completely blocked.
Flow cytometric analysis of propidium iodide-stained 041 TR cells was performed to determine the effects of Tet withdrawal and UV irradiation on cell cycle distribution. Following withdrawal of Tet and induction of wt p53, 041 TR cells exhibited a time-dependent increase in cells in both the G 1 and G 2 /M phases and a decrease in cells in S phase (Fig. 5A). A small increase in apoptosis was suggested by the increased sub-G 1 DNA fraction seen at 48 -72 h. In the presence of Tet, UVirradiated 041 TR cells did not undergo significant DNA damage-induced G 1 checkpoint arrest, inhibition of DNA replication, or apoptosis (Fig. 5B). In contrast, following induction of wt p53, UV-irradiated 041 TR cells showed an increase in cells in G 1 and G 2 /M phases from 0 to 48 h, as well as a dramatic increase in sub-G 1 apoptotic DNA degradation at 72 h (Fig. 5C).
The effect of wt p53 induction on UV-induced apoptosis in 041 TR cells was confirmed by morphologic examination of fluorescent stained cells. In the presence of Tet, 20 J/m 2 UV irradiation had little effect on the number of apoptotic cells seen 48 h later (Fig. 6). However, following withdrawal of Tet and induction of wt p53, a significant increase in the number of cells displaying morphologic alterations consistent with apoptosis was seen.
Therefore, the use of non-cytotoxic levels of Tet in 041 TR cells allows for the regulation of wt p53 levels, as well as regulation of several of the hallmark cellular functions of wt p53 following UV irradiation, including transcriptional activation of p53-responsive genes, induction of cell cycle checkpoints, inhibition of DNA replication following DNA damage, and DNA damage-induced apoptosis. Therefore, this cell line was appropriate for analysis of the effect of wt p53 expression and function on the cellular proc-  (17). Approximately 50,000 cpm of 3 H-labeled DNA in 6 ml of Tris-EDTA with cesium chloride at a density of 1.72 g/ml was loaded onto each gradient in 16 ϫ 76-mm polyclear centrifuge tubes, and centrifuged at 37,000 rpm for 36 h. Gradient fractions were collected, and the relative DNA content (fraction of total 3 H-labeled DNA) in each fraction determined by scintillation spectrophotometry.
essing of UV-induced DNA damage.
Effect of wt p53 Expression on Global Genomic NER and Transcription-coupled Repair in Human 041 TR Fibroblasts-The effect of wt p53 activity on the repair of CPDs and 6-4 photoproducts in overall genomic DNA, and the repair of CPDs in individual strands of the DHFR gene was examined in 041 TR cells in the presence or absence of Tet. In the presence of Tet, 041 TR cells removed less than 25% of CPDs from overall genomic DNA within 24 h (Fig. 7A), similar to the parental 041 mut cells ( Fig.  2A). However, when cells were irradiated 24 h following withdrawal of Tet, 041 TR cells removed ϳ70% of the CPDs from overall genomic DNA within 24 h after irradiation (Fig. 7A). When the cells were irradiated coincident with the removal of Tet, a similar response was observed (data not shown).
The removal of 6-4 photoproducts from overall genomic DNA following withdrawal of Tet occurred with kinetics similar to that for normal fibroblasts. In the presence of Tet, removal was slightly slower (for example, 55% versus 70% repair of 6-4 photoproducts by 2 h, Fig. 7B). Even in the presence of Tet, however, the 041 TR cells exhibited more efficient repair of 6-4 photoproducts than the parental 041 mut cells (55% versus 30% repair at 2 h, Fig. 2B). This probably results from low levels of wt p53 expression occurring in the 041 TR cells in the presence of Tet, which may selectively affect the removal of 6-4 photoproducts, but not alter the repair of CPDs.
We demonstrated previously that both the LFS 041 mut and 087 mut cells were also deficient in removing CPDs from the non-transcribed strand of the p53 and DHFR genes, consistent with their deficiency in global genomic NER. However, both these p53-deficient cell lines maintained the ability to preferentially repair the transcribed strand of these expressed genes in a manner identical to normal, p53 wt cells (8). To confirm that this selective reduction of CPD removal from the nontranscribed strand was due specifically to the wt p53 gene, repair of CPDs within individual strands of the human DHFR gene was analyzed in 041 TR cells in the presence or absence of Tet, using quantitative Southern hybridization and strandspecific RNA probes (Fig. 8). Following quantitation, it was clear that the presence or absence of Tet had no effect on repair of CPDs within the transcribed strand of DHFR (Fig. 9). However, the presence of Tet had a significant affect on the rate of removal of CPDs from the non-transcribed strand of DHFR. Specifically, in the presence of Tet, only ϳ30% of CPDs were removed from the non-transcribed strand of the DHFR gene by 24 h, an amount similar to that seen for the parental 041 mut cells (8). However, following withdrawal of Tet and induction of wt p53, repair of the non-transcribed strand occurred at a normal level, displaying a slightly decreased rate compared with the transcribed strand, as seen previously for many other normal primary human fibroblasts (8,22). Therefore, expression of wt p53 in functionally p53 null human fibroblasts results in recovery of normal levels of global genomic NER and repair of lesions within the non-transcribed strand of expressed genes, activities that we have previously shown to be defective in p53 mutant cells. Furthermore, this model confirms that levels of wt p53 do not affect the efficiency of transcription-coupled repair, which we have shown previously to be unaffected by loss of wt p53 in human fibroblasts. DISCUSSION These results provide strong evidence that the human wt p53 tumor suppressor gene is involved in the regulation of NER activity in human cells, in vivo. We have analyzed global genomic NER of UV-induced CPDs and 6-4 photoproducts, as well as transcription-coupled repair of CPDs from separate strands of the DHFR gene, in fully characterized human fibroblasts either wt or mutant for p53 function, and in human p53 mutant fibroblasts in which the expression of a wt p53 gene was regulated. Our findings confirm our initial description of the DNA repair-deficient phenotype associated with loss of wt p53 function (8), and demonstrate that the wt p53 gene product is specifically involved in the regulation of global genomic NER, but not transcription-coupled repair, of UV-induced DNA photoproducts.
We previously showed that mutations in the p53 gene resulted in a deficiency in global genomic DNA repair of CPDs in LFS 041 mut and 087 mut cells (8). To confirm these findings, and to characterize the effect of p53 mutations on global genomic NER more completely, we employed an immunoassay using monoclonal antibodies specific to each of the two major UV-induced DNA adducts, CPDs and 6-4 photoproducts. Our results with this immunoassay are fully consistent with those obtained previously for CPDs, using alkaline sucrose sedimen-  tation analysis. Specifically, we find that loss of wt p53 function is associated with a significant deficiency in the global genomic repair of CPDs, similar in extent to that seen in repair-deficient XP-C cells. In addition, we demonstrate that p53 homozygous mutant cells are also partially deficient in the repair of UVinduced 6-4 photoproducts, although not to the extent exhibited by XP-C cells. Also consistent with our previous findings, p53 heterozygous mutant cells displayed a small decrease in the initial rate of removal of CPDs from overall genomic DNA, although CPD repair was normal by 24 h and repair of 6-4 photoproducts was normal at all time points.
To investigate whether the repair deficiencies observed in these p53 homozygous mutant cells were due specifically to loss of wt p53, rather than to other, uncharacterized genetic changes occurring simultaneously in these inherently genetically unstable cells, we studied LFS 041 mut cells in which the expression of wt p53 could be regulated. The characteristics of these human 041 TR fibroblasts proved highly desirable for our studies for several reasons. Treatment with Tet resulted in the suppression of wt p53 gene expression and activity, even following UV irradiation. In addition, withdrawal of Tet resulted in the induction of wt p53 protein levels and functions to an extent and in a time course similar to that seen for endogenous wt p53 following DNA damage. Therefore, we employed these cells to further characterize the effect of wt p53 expression and function on NER. Utilizing an immunoblot assay for global genomic NER, we clearly demonstrated in 041 TR cells that the regulated expression of wt p53 in a p53 null background resulted in the restoration of normal levels of repair of both CPDs and 6-4 photoproducts, compared with their p53 homozygous mutant parental cells or to the same 041 TR cells in the presence of Tet. The enhanced level of 6-4 photoproduct repair in 041 TR cells in the presence of Tet compared with 041 mut cells suggests that only very low levels of wt p53 protein are required for this repair activity. The difference in the level of wt p53 function and biological activity required for restoration of normal global genomic repair of 6-4 photoproducts compared with CPDs suggests potential mechanisms for the role of p53 in NER, as discussed below.
The effect of wt p53 protein levels on repair of CPDs in the transcribed and non-transcribed strand of the human DHFR gene was also examined in 041 TR cells. In support of our earlier findings (8), wt p53 expression was not required for efficient transcription-coupled repair of CPDs. For example, the rate and extent of removal of CPDs from the transcribed strand of the DHFR gene was identical in 041 TR cells in the absence or presence of Tet, and similar to that seen in normal human fibroblasts. In contrast, the removal of CPDs from the non-transcribed strand of the DHFR gene was clearly deficient in 041 TR cells, similar to what we found for repair of the non-transcribed strand of both DHFR and p53 in the parental 041 mut cells (8). Following withdrawal of Tet from 041 TR cells, repair of CPDs in the non-transcribed strand of DHFR was restored to wild-type levels. This finding confirms the role of wt p53 in specifically regulating repair of lesions within genomic DNA, since repair of the non-transcribed strands of expressed genes generally reflects the overall capacity of cells to repair non-transcribed DNA sequences (21).
Coincident with our initial description of the specific NER deficiency found in human LFS fibroblasts, several other groups published data concerning the effect of p53 activity on NER in these same cells, and in other cells mutant for p53. In general, these studies are all consistent with our finding that loss of wt p53 function results in a deficiency in global genomic NER, but has no effect on transcription-coupled repair. For example, Fornace and colleagues (9,23) have presented several lines of evidence suggesting that alterations in p53 activity inhibit normal NER activity. Transient transfection of RKO human colon carcinoma cells containing wt p53 with vectors expressing genes the products of which suppress or disrupt wt p53 activity, such as a dominant negative mutant p53, human papilloma virus E6, SV40 T antigen, or Mdm-2, resulted in a reduction in the ability of cells to express a chloramphenicol FIG. 8. Southern blots of DNA from UV-irradiated human 041 TR fibroblasts probed for the transcribed and non-transcribed strand of the human DHFR gene. Repair of CPDs from the transcribed (A and B) and non-transcribed (C and D) strand of the DHFR gene was determined in 041 TR cells following UV irradiation with 10 J/m 2 , in the presence (A and C) or 24 h following withdrawal of 2 g/ml Tet (B and D). Samples (10 g) of unreplicated, purified DNA restricted with KpnI were treated (ϩ) or not treated (Ϫ) with TEV before electrophoresis through a 0.5% denaturing alkaline-agarose gel. The DNA was transferred to a nylon membrane and hybridized with 30 ϫ 10 6 cpm 32 P-labeled strand-specific RNA probe specific to either strand of a 20-kilobase pair KpnI fragment of the human DHFR gene. Following detection using phosphorimager analysis, each membrane was stripped and rehybridized using a probe for the other DNA strand.
FIG. 9. Time course for strand-specific removal of CPDs from the DHFR gene in human 041 TR fibroblasts. Quantitative Southern hybridization with strand-specific RNA probes to the DHFR gene was performed on DNA isolated from 041 TR cells following UV irradiation with 10 J/m 2 , as demonstrated in Fig. 8. The frequency of induction of CPDs and their rate of removal from the transcribed and nontranscribed strands of the human DHFR gene were determined by quantifying the reappearance of specific full-length restriction fragments in DNA from cells allowed various times to remove CPD from their DNA. The ratio of full-length restriction fragments in the enzyme treated and untreated samples was determined by phosphorimager analysis, and this was used to calculate the average number of CPDs per fragment. The points on the graph represent the average repair at each time point from two individual biological experiments, with each resulting Southern blot probed at least two times with an RNA probe specific for the transcribed (E, Ⅺ) or non-transcribed (q, f) strand of the human DHFR gene, in the presence (E, q) or absence (Ⅺ, f) of 2 g/ml Tet. acetyltransferase reporter gene contained within an exogenously UV-irradiated plasmid that had been introduced into these cells. In addition to this decrease in host-cell reactivation, the E6-transfected RKO cells also were deficient in unscheduled DNA synthesis following UV irradiation compared with the parental RKO cells (23).
Additional evidence supporting a role for p53 in global genomic NER activity has been provided by Mirzayans et al. (24), in an analysis of two LFS skin fibroblast cell lines of different origin than those used in the present study, and containing p53 mutations known to exert a dominant negative effect upon wt p53 DNA binding activity (25). Measurement of overall genomic NER was performed by alkaline sucrose gradient sedimentation analysis of DNA from UV-irradiated cells in which DNA polymerase ␦/⑀-dependent NER was pharmacologically inhibited. A reduced number of incomplete repair events resulting in single-strand DNA breaks (40 -75% of normal) were found in the LFS cell strains compared with normal diploid fibroblasts following UV irradiation.
Repair studies performed in p53 transgenic and knock-out mouse models also provide results consistent with our findings. For example, primary kerotinocytes in culture or harvested from irradiated transgenic mice carrying multiple copies of a mutant p53 allele or from mice in which p53 had been homozygously deleted exhibit decreased global genomic repair of both CPDs and 6-4 photoproducts as assayed using lesion-specific monoclonal antibodies and methods similar to those described in this paper (26,27).
In contrast to our findings, it has been suggested by several investigators that p53 activity may also effect the efficiency of transcription-coupled repair. Although this point remains controversial in the literature, to our knowledge none of the other studies reported have directly analyzed the effect of wt p53 expression upon transcription-coupled repair by measuring the removal of UV-irradiated DNA damage from the transcribed strand of an endogenous gene. Wang et al. (10) reported a decrease in the rate of removal of CPDs from the DHFR gene in primary human LFS fibroblasts, which were presumably heterozygous for p53 mutations, and suggested that this was due to decreased transcription-coupled repair. However, these investigators used a gene-specific DNA probe to DHFR, rather than strand-specific RNA probes as required to determine repair within individual strands of the gene, and thus their results necessarily represent an average of repair within the transcribed strand (transcription-coupled repair) and the nontranscribed strand (global genomic repair) of this gene. Therefore, these data are equally consistent with a deficiency in global genomic repair resulting in decreased removal of CPDs from the non-transcribed strand of DHFR, as we have observed. In fact, the authors note that no difference was detected in the recovery of RNA synthesis following UV irradiation, strongly suggesting that transcription-coupled repair was intact in these cells. Similar findings were also reported by Mirzayans et al. (24) in primary LFS skin fibroblasts containing the same p53 mutation studied by Wang et al., and utilizing a gene-specific probe to demonstrate decreased removal of CPDs from the c-myc gene.
Thus, neither of these studies directly measured the removal of lesions from the transcribed strand of an expressed gene (the definition of transcription-coupled repair); instead, they employed techniques measuring an average repair rate from both strands, which may reflect to varying degrees both overall and transcription-coupled repair. Thus, there is currently no evidence to challenge our conclusion that alterations in p53 activity have any significant effect on transcription-coupled repair. We would predict, however, that any gene whose expression was transcriptionally regulated by p53 would, in fact, display reduced transcription-coupled repair in p53 homozygous mutant cells. Careful analysis of strand-specific DNA repair within a representative endogenous gene sequence is required to clearly verify the effect of p53 mutations on global genomic NER versus transcription-coupled repair activities.
The issue of whether human p53 heterozygous mutant LFS skin fibroblasts exhibit a biologically relevant deficiency in NER remains unclear. We reported previously that primary LFS fibroblasts heterozygous for a base pair missense mutation at codon 248 (087 wt/mut cells) or a frameshift deletion mutation at codon 184 (041 wt/mut cells), both exhibit a slight decrease in global genomic repair of CPDs at early time points following UV irradiation, but display normal repair by 24 h, as assayed by the TEV-sensitive site assay (8). Further analysis of the repair of both CPDs and 6-4 photoproducts in these two cell lines by immunoblot assay also show a detectable difference in the initial rate of repair of CPDs, but normal levels of 6-4 photoproduct repair, compared with that in p53 wt cells. Additionally, the removal of CPDs from the non-transcribed strand of the DHFR gene in 041 wt/mut cells shows a slight decrease in the initial rate, but eventually normal overall extent of removal (data not shown). Tainsky and colleagues (28) have extensively characterized these same LFS p53 heterozygous mutant cells and suggested that they may demonstrate certain other characteristics suggestive of increased genetic instability, such as increased replicative errors in a shuttle vector resulting in enhanced mutation frequency of the reporter supF gene.
It has been suggested that particular loci or types of mutations within the p53 gene may be associated with a more or less significant dominant negative effect on the activity of the remaining wt p53 protein (29). Whether such mutation-specific effects also correlate with NER activity remains unknown. As discussed above, a measurable decrease in the average removal of CPDs from both strands of the DHFR or c-myc gene, consistent with a significant deficiency in repair of the non-transcribed strand, has been observed in LFS p53 heterozygous fibroblasts mutant at codon 245 (10,24). Although the degree to which repair was deficient was greater than we found in two other primary LFS p53 heterozygous cell lines containing different p53 amino acid changes, the codon 245 base pair mutation is known to confer a dominant negative effect on wt p53 DNA binding activity (25,30). Therefore, certain mutations within the p53 gene may confer a stronger dominant negative effect on the remaining wt allele with regard to NER activity than we found in the two cell lines analyzed in our study. Confirmation of this hypothesis, however, requires a direct comparison of these various cell lines using assays that distinguish global genomic NER from transcription-coupled repair, and correlation with the effect of different p53 mutations on wt p53 biochemical activities. An alternative possibility is that a wt p53 gene dosage effect exists, and that varying degrees of wt versus mutant p53 gene expression result in altered global genomic NER.
The mechanism for the effect of wt p53 expression on global genomic NER activity is not yet understood. However, the multiple known biochemical activities associated with the wt p53 protein and the numerous molecular interactions in which wt p53 is involved allow for consideration of several possibilities. Most wt p53-associated responses so far characterized appear to be due to its ability to regulate the transcriptional activation of other downstream effector genes. Examples of potential relevance for NER include the p21 and GADD45 genes, both of which exhibit enhanced expression in a p53 dependent manner (31,32). Both p21 and GADD45 may alter in vitro assays of excision repair (33,34), and it has been suggested that this maybe due to the ability of these proteins to bind to proliferating cell nuclear antigen, a protein involved in both DNA synthesis and repair synthesis. However, the effect of either p21 or GADD45 on genomic repair processes has yet to be clearly defined.
It is likely that wt p53 transcriptionally regulates many other genes as yet unidentified, and it may potentially be involved in regulating expression of NER proteins or co-factors. We suggest that p53 may regulate the transcription or activity of gene products known to be directly involved in NER (for review, see Ref. 21). Of particular interest are those genes thought to be functionally involved in the recognition of lesions in genomic DNA, for accessibility of the repair complex to lesions in chromatin, or even for recruitment of damaged DNA to repair "factories" at the nuclear matrix (35). A precedent for this model is the phenotype of a partial revertant of XP-A, in which a single amino acid change within the mutant protein results in an increase in cellular XPA protein levels from 0 to 30% (36). Compared with the mutant XP-A cells, which are completely deficient in all NER activities, the XP129 revertant exhibits normal transcription-coupled repair of CPDs (37) and normal global genomic repair of 6-4 photoproducts, 2 but remains strikingly deficient in global genomic NER of CPDs (38). The XP129 repair phenotype is therefore similar to that seen in p53 mutant fibroblasts, suggesting that regulation of XPA protein levels by wt p53 may effect global genomic repair. Similar models may be hypothesized for other proteins putatively involved in DNA damage recognition or recruitment, such as the XP-C or XP-E gene products (21).
It is also known that the p53 protein may directly interact with other cellular proteins, and it has been suggested that such protein-protein interactions may be involved in the effect of p53 on NER. In fact, in vitro protein binding assays have shown that p53 is capable of binding to the TFIIH associated NER enzymes XP-B and XP-D, and may inhibit their helicase activities (10,39). Whether such protein interactions alter the NER activity or accessibility of the repair components of TFIIH for genomic DNA damage remains an untested hypothesis. The p53 protein can also directly bind to DNA at certain damage sites, including single strand breaks (40) and DNA insertiondeletion mismatches (41). However, no studies have yet suggested that p53 might recognize or bind to UV photoproducts or other bulky lesions processed by NER mechanisms. Finally, the results from cell-free repair assays strongly suggest that such a direct protein-protein or protein-DNA interaction of p53 is not involved in NER activity. Several groups have reported that the addition of recombinant wt p53 protein to cell-free in vitro assays measuring NER of a damaged DNA substrate has neither positive nor negative effects (39,42). It is important to note, however, that these studies do not address the cellular consequences of p53 activity on NER, since the downstream effects of p53 function (e.g. on transcriptional regulation of other genes) cannot be assessed in this type of experiment.
In summary, our results demonstrate that loss of wt p53 function in human skin fibroblasts is associated with a decrease in NER of overall and non-transcribed DNA, but does not affect preferential repair of the transcribed strands of expressed genes. Furthermore, the regulated expression of wt p53 in these p53 mutant fibroblasts results in restoration of a normal repair phenotype. While the specific mechanism by which wt p53 affects repair is not yet understood, these results suggest a novel cellular function for the p53 protein, in addition to its role in cell cycle control and apoptosis. We suggest that alterations in NER of UV-induced DNA damage associated with mutant p53 are due to a direct effect of p53, or other components of a p53-dependent pathway, on the regulation of the amount or activity of one or more repair proteins.