Cadmium Induces Conformational Modifications of Wild-type p53 and Suppresses p53 Response to DNA Damage in Cultured Cells*

The p53 tumor suppressor protein is a transcription factor that binds DNA in a sequence-specific manner through a protein domain stabilized by the coordination of zinc within a tetrahedral cluster of three cysteine residues and one histidine residue. We show that cadmium, a metal that binds thiols with high affinity and substitutes for zinc in the cysteinyl clusters of many proteins, inhibits the binding of recombinant, purified murine p53 to DNA. In human breast cancer MCF7 cells (expressing wild-type p53), exposure to cadmium (5–40 μm) disrupts native (wild-type) p53 conformation, inhibits DNA binding, and down-regulates transcriptional activation of a reporter gene. Cadmium at 10–30 μm impairs the p53 induction in response to DNA-damaging agents such as actinomycin D, methylmethane sulfonate, and hydrogen peroxide. Exposure to cadmium at 20 μm also suppresses the p53-dependent cell cycle arrest in G1 and G2/M phases induced by γ-irradiation. These observations indicate that cadmium at subtoxic levels impairs p53 function by inducing conformational changes in the wild-type protein. There is evidence that cadmium is carcinogenic to humans, in particular for lung and prostate, and cadmium is known to accumulate in several organs. This inhibition of p53 function could play a role in cadmium carcinogenicity.

The p53 protein is a tumor-suppressive transcription factor activated in response to multiple signals including radiation, genotoxic chemicals, hypoxia, depletion of ribonucleotides, and poisoning of the mitotic spindle. In most normal, nonexposed cells, p53 is a latent factor. Induction in response to stress involves nuclear accumulation (as a result of escape from mdm-2-mediated degradation and nuclear export) and conversion to an active form with high affinity for specific DNA sequences. Activation requires post-translational modifications at both the N and C terminus of the protein, including changes in phosphorylation, acetylation, and binding to heterologous proteins (1)(2)(3)(4)(5). Activated p53 controls several sets of genes to prevent the proliferation of cells under stress conditions. Genes transactivated by p53 include inhibitors of cell cycle progression in G 1 and G 2 (p21 waf-1 , 14-3-3, GADD 45), regulators of apoptosis (APO1-Fas/CD95, Bax-1, KILLER/DR5), and genes involved in the metabolism of reactive oxygen species (such as PIG-3, PIG-6, and PIG-12) that may play a role in induction of apoptosis (6,7). p53 also represses a number of promoters and modulates transcription, replication, and DNA repair through interaction with proteins such as RP-A and components of TFIID and TFIIH complexes (for recent reviews see . High affinity binding of p53 to specific DNA sequences is mediated by a conformation-sensitive structure in the central portion of the protein (residues 102-292) (12). The structure of the DNA-binding domain consists of two ␤-sheets supporting a loop-sheet-helix motif (that interacts with the major groove of DNA) and a loop-helix motif (L2/L3, that interacts with the minor groove). L2/L3 is stabilized by tetrahedric coordination of zinc by residues Cys 176 , His 179 , Cys 238 , and Cys 242 (13). Folded and unfolded forms of human wild-type p53 are distinguishable by their reactivity with the conformation-specific monoclonal antibodies PAb1620 (folded form, often termed "wild-type" conformation) and PAb240 (unfolded form, often termed "mutant" conformation).
The folding of the DNA-binding domain is sensitive to metal substitution and to oxido-reduction in vitro and in intact cells. Removal of zinc by chelation reversibly alters p53 conformation, with loss of DNA binding capacity (5, 14 -16). Furthermore, metals such as copper, cadmium, or mercury induce p53 to adopt a PAb240 ϩ phenotype in vitro (17;18). These observations raise the possibility that exposure to toxic metals and perturbation of the physiological metal supply may affect p53 function in vivo.
Metals such as cadmium, chromium, nickel, and arsenic are classified in group 1 of the International Agency for Research on Cancer categories of carcinogens (carcinogenic to humans; for reviews, see Refs. 19 -21). Cadmium is chemically close to zinc and binds with high affinity within the tetrahedral zincbinding domains of several metalloproteins in vitro (22)(23)(24). Cadmium is a widespread environmental pollutant that is also present in tobacco smoke (1-3 g/cigarette). Smoking, together with occupation, are the major sources of human exposure. Cadmium is absorbed by inhalation and ingestion and has a very long biological half-life (Ͼ25 years). Epidemiological studies have identified lung, prostate, and, to a lesser extent, kidney and stomach as primary targets for cadmium-induced tumorigenesis (21). In exposed industrial workers, cadmium accumulates in the kidneys (100 -400 g/g, wet weight) and liver (20 -100 g/g, wet weight), at levels that are 5-9 times higher than those of unexposed workers (25,26). The kidneys and liver express high levels of metallothioneins, a class of stress response proteins that bind and detoxify cadmium.
The mechanisms of cadmium carcinogenesis are poorly understood. In vitro, at concentrations between 0.1 and 10 mmol, cadmium is cytotoxic and induces radical-dependent DNA damage (27,28). However, compared with other carcinogenic metals, cadmium is a weak mutagen (29). At lower concentrations (1-100 mol), cadmium binds to proteins, decreases DNA repair (30,31), activates protein degradation, up-regulates cytokines and proto-oncogenes such as c-fos, c-jun, and c-myc (32,33), and induces the expression of metallothioneins (34). Thus, cadmium carcinogenicity may involve multiple factors, including up-regulation of mitogenic signals and interference with DNA repair (for a review, see Ref. 19).
In this study, we have examined the effects of cadmium on p53 protein conformation, DNA binding, and transcriptional activity. Using the breast carcinoma MCF7 cell line, which expresses high levels of wild-type p53, we show that cadmium at subtoxic concentrations (10 -30 M) perturbs the folding of p53, disrupts DNA binding, impairs p53 induction by DNAdamaging agents, inhibits transactivation of a reporter gene and of target genes such as p21 waf-1 , and prevents cell cycle arrest in response to ␥-irradiation. Based on these results, we propose that cadmium may inactivate wild-type p53 by altering metal-dependent folding and that this effect may contribute to cadmium carcinogenesis.

EXPERIMENTAL PROCEDURES
Purified Wild-type Recombinant p53-Murine p53 was produced in Sf9-infected cells using a baculovirus expression system and was purified in buffers depleted of ion transition metals ("metal-free buffers") as described previously (5). Metal-free buffers were prepared by incubation of DNA-binding reaction solutions with chelating resin (10% v/v) for 1 h at 4°C and used immediately (Chelex-100; Sigma).
Cell Culture and Treatment-The human breast carcinoma cell line MCF7, expressing high levels of wild-type p53, was cultured at 37°C under 10% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, and antibiotics (PAA, Linz, Austria). Murine 10.1 fibroblasts (p53-deficient) were cultured at 37°C under 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, and antibiotics (35). With the exception of the irradiation experiments, cells were plated in 9-cm Petri dishes and drug-treated at 70 -80% confluency.
A stock solution of CdCl 2 (0.5 M) was prepared in 10 mM Tris, pH 7.4, and diluted for cell treatment. In case of pretreatment with cadmium, cells were exposed to CdCl 2 (30 min or 2 h) before addition to the culture medium of actinomycin D (Act D; 1 2.5 ng/ml), methylmethane sulfonate (0.5 mM), or H 2 O 2 (100 M). All chemicals were from Sigma. Subconfluent MCF7 cells, plated in 75T flasks, were exposed to ionizing radiation by treatment in an irradiator ( 137 Cs) for an appropriate length of time to deliver the preselected dose of 5 Gy.
Cytoplasmic and Nuclear Protein Extractions-Cells were washed in PBS and collected by scraping. Cytoplasmic and nuclear extracts were prepared as described in Ref. 16. Briefly, cells were lysed in buffer A (20 mM HEPES (pH 7.6), 20% glycerol, 10 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM dithiothreitol, 0.1% Nonidet P-40). After centrifugation, supernatants were kept as cytoplasmic extracts. Nuclear proteins were obtained by extraction of the pellet in buffer B (same as buffer A but with 0.5 M instead of 10 mM NaCl) in the presence of a mixture of protease inhibitors: 0.5 mg/ml leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, 2 mg/ml aprotinin, 0.7 mg/ml pepstatin (all from Sigma). Protein contents were quantified by the Lowry method.
DNA-binding Assays-The protocols for DNA-binding assays were described earlier (16,37). Briefly, the p53 consensus binding sequence p53 CS (5Ј-GGACATGCCGGGCATGGTCC-3Ј) (38) and an oligonucleotide containing the Oct-1 binding sequence (underlined; 5Ј-GACCA-CCTGGGTAATTTGCATTTCTAAAATA-3Ј) were end-labeled with ϳ3000 Ci/mmol [␥-32 P]ATP (Amersham Pharmacia Biotech). DNA binding experiments were performed for 30 min at room temperature. Murine recombinant protein (100 ng) was incubated with 0.5 ng of labeled p53 CS (in 10 mM dithiothreitol, 5 g of bovine serum albumin, 140 mM NaCl, 20 mM HEPES (pH 7.6), 20% glycerol, 0.1% Nonidet P- 40) in the presence of herring sperm DNA (2.8 ng) as nonspecific competitor. A similar protocol was used for DNA-binding assays using nuclear extracts (40 g) with the following modifications; 1) the concentration of competitor (herring sperm) DNA was increased to 2.2 g; and 2) experiments were performed in the presence of 4 mM dithiothreitol. All experiments included PAb421 (100 ng) (OP03; Oncogene Science), a monoclonal antibody that stabilizes and supershifts p53-DNA complexes. With cellular extracts, no specific binding to DNA was detectable in the absence of PAb421. After incubation, DNA-p53 protein-PAb421 complexes were resolved on a 4% nondenaturating polyacrylamide gel electrophoresis in 1ϫ TBE for 2-3 h at 120 V. Gels were then fixed, dried, and exposed to Kodak x-ray films at Ϫ80°C. Control experiments using a mutant p53 consensus sequence, as well as competition experiments using unlabeled p53 CS , were performed to demonstrate the specificity of binding.
Conformation-specific Immunoprecipitation of p53-MCF7 cells were washed in PBS, lysed 10 min on ice in immunoprecipitation buffer (10 mM Tris-HCl, pH 7.6, 140 mM NaCl, 0.5% Nonidet P-40, with protease inhibitors as above), scraped, and kept on ice for another 10 min prior to 5-min centrifugation at 15,000 ϫ g at 4°C. Supernatants were precleared by incubation with 1 g of a non-anti-p53 antibody (PAb416, specific for large T antigen of Simian virus 40 (SV40), DP29, Oncogene Science) for 15 min at 4°C with shaking followed by incubation with 10% (w/v) of Staphylococcus aureus protein A suspension (50 l) (Sigma) for 15 min and by 5-min centrifugation (15,000 ϫ g) at 4°C. Supernatants were aliquoted for immunoprecipitation with monoclonal antibodies PAb1620 (specific for the wild-type, folded form, OP33), PAb240 (specific for the mutant, unfolded form, OP29), PAb421 (which recognizes both forms, OP03), and PAb416 (as negative control, DP29) (all from Oncogene Science). Immune complexes were collected using S. aureus protein A suspension and washed five times in immunoprecipitation buffer. Precipitates were then denaturated in Laemmli buffer and analyzed by Western blot experiments using the rabbit polyclonal anti-p53 antibody CM-1 (1:1000 dilution; Novocastra, Newcastle, UK) and peroxidase-conjugated goat anti-rabbit immunoglobulin G (250 ng/ ml, Pierce) as secondary antibody.
Transfections and ␤-Galactosidase Assays-Murine 10.1 cells were co-transfected by the calcium phosphate method with the reporter plasmid pRGC⌬FosLacZ, containing the p53 binding site located in the ribosomal gene cluster (RGC-␤gal) (39), and p53pcDNA, expressing full-length human p53 cDNA located under a cytomegalovirus promoter. Medium was changed 16 h later, and cells were further cultured for 20 h. Cytoplasmic protein extracts (100 g) of transfected cells were used to perform ␤-galactosidase assays.
Cell Cycle Analysis-Cell nuclei were collected and stained with propidium iodide, using the cycle TEST-PLUS DNA-staining kit, as specified by the manufacturer (Becton Dickinson). DNA contents of stained nuclei were measured on a FACSCalibur flow cytometer (Becton Dickinson). Data acquisition was performed using the CellQuest software. Cell cycle analysis and quantification of cell cycle phases was performed using the ModFit LT 2.0 software (Verity Software House, Inc.).
Statistical Evaluation-For autoradiograms, densitometric quantification was performed using a Bio-Rad imaging densitometer (GS-670) and Molecular Analyst software (Bio-Rad). The significance of observed differences was evaluated using the two-tailed t test. Probabilities of p Ͻ 0.05 were regarded as statistically significant.

Cadmium Inhibits Specific DNA Binding by Recombinant
Wild-type p53-Previous studies using in vitro translated murine p53 have shown that cadmium induces conformational changes in p53 with loss of the immunological, wild-type phenotype (reactive with PAb1620) (17). To better characterize the effect of cadmium on the p53 protein, we incubated purified, baculovirus-produced murine wild-type p53 with CdCl 2 or ZnCl 2 and analyzed its DNA-binding capacity by electrophoretic mobility shift assay (EMSA). Fig. 1 shows that divalent metal ions are required for DNA binding, since no activity was observed in buffers depleted of transition metal (treated with Chelex resin) (lane 1). Binding was restored in non-Chelex buffers, containing trace amounts of divalent metals (lane 2). Binding was increased when the protein was incubated in 20 M ZnCl 2 prior to EMSA (lane 3). In contrast, incubation with CdCl 2 (2-16 M) induced a dose-dependent inhibition of DNA binding activity (lanes 4 -7). This result confirms the role of zinc in DNA binding competence and shows that cadmium inhibits DNA binding, consistent with the hypothesis that it may compete with zinc in binding to reactive cysteines within the DNA-binding domain of p53.
Effect of CdCl 2 on Expression of p53 and of Metallothioneins-The MCF7 breast carcinoma cell line expresses high levels of wild-type p53 and is commonly used as a cellular model to assess p53 functions. Northern blot analysis of cells exposed to 10 M CdCl 2 showed a strong, time-dependent, increase in the expression of MT-IIA, a metal-binding protein specifically induced by cadmium. Increased mRNA expression was already detectable after 1 h and reached a plateau (50-fold) after 6 -12 h. In contrast, levels of p53 mRNA remained unchanged after up to 24 h of treatment (Fig. 2). These results indicate that cadmium was rapidly taken up by MCF7 cells and induced the expression of a metalloregulated gene but did not affect the level of p53 mRNA.
Down-regulation of p53 DNA Binding Activity in MCF7 Treated with CdCl 2 -Cadmium is a potent cytotoxic agent known to induce oxidative stress in cultured cells (40). Trypan blue exclusion tests showed that the percentage of surviving MCF7 cells after 24 h of culture in the presence of cadmium was 93, 50, and 20% at, respectively, 10, 20, and 30 M CdCl 2 . Cytotoxicity was time-dependent, and no significant cell death was observed at up to 8 h of incubation. To evaluate the long term effects of short exposures to cadmium, cells were incubated for 4 h with CdCl 2 and then further cultured for 20 h in CdCl 2 -free culture medium. Under these conditions, no significant cytotoxicity was shown at 10 M CdCl 2 , and up to 80% of cells survived with 20 M CdCl 2 . Nevertheless, treatment with 30 M CdCl 2 still induced a sharp drop in cell viability (by 75%) (data not shown). Unless otherwise stated, in all experiments reported here, cells were exposed to CdCl 2 for 4 h and immediately harvested in conditions were no significant decrease in viability was observed.
Western blot analysis of p53 levels in cells exposed to CdCl 2 showed that cadmium induced a dose-dependent accumulation of p53 after 4 h of exposure, with a maximum increase of 2.3-fold at 20 M, followed by a decrease at higher doses (Fig.  3A, upper panel). Since no effect was seen on the p53 mRNA (Fig. 2), this accumulation may result from protein stabilization. DNA binding activity also showed a biphasic dose response, with first a slight increase at 10 M, followed by a marked inhibition at 20 M and above (Fig. 3A, middle panel). There was no significant change in the level of binding of Oct-1, a ubiquitous, constitutively expressed transcription factor used as a control (Fig. 3A, lower panel). These results indicate that cadmium exerted complex effects on p53 in intact cells, with accumulation of the protein (compared with untreated cells) and DNA binding inhibition (at concentrations of CdCl 2 equal to or greater than 20 M). At 10 M, cadmium increased both p53 levels and DNA binding activity, consistent with the notion that this metal may induce oxidative DNA damage. However, at higher concentrations (20 M and above), cadmium downregulated DNA binding activity even if p53 protein levels remained higher than in nontreated cells. Binding of the control Baculovirus-expressed murine wild-type p53 was purified in metal-free (Chelex-treated) buffers. Equal amounts of protein (100 ng) were incubated for 30 min at 37°C with zinc or cadmium chloride. Samples were then diluted 5-fold in EMSA reaction buffer, and DNAbinding reactions were performed using the 32 P-labeled synthetic oligonucleotide p53 CS , in the presence of PAb421 (which stabilizes and supershifts p53-DNA complexes (38)). Lane 1, all reactions were performed in metal-free buffers. Other lanes, reaction buffers were not Chelex-treated and thus contained traces of metals. Several retarded bands are detected, corresponding to different oligomers of p53 as described by Rainwater et al. (5). Filled arrowhead, free DNA probe. transcription factor Oct-1 shows that inhibition of p53 did not correspond to a general toxic effect on DNA-binding proteins. Similar results were obtained with other cell lines expressing wild-type p53 (3T3 mouse fibroblasts and A549 human lung carcinoma cells) (data not shown).
DNA binding of p53 in cell extracts requires the addition of PAb421, a monoclonal antibody that binds to the C terminus of the protein and stabilizes its activity (see "Experimental Procedures"). To determine if cadmium may inhibit DNA binding by interfering with the binding of PAb421, the p53 protein was immunoprecipitated from control and cadmium-treated cells, and p53 levels were determined by Western blotting. Quantitative analysis revealed variations of less than 10%, indicating that DNA binding inhibition by cadmium was not due to impaired reactivity of p53 with PAb421.
Nuclear Accumulation of p53 in Cells Exposed to CdCl 2 -To determine if inhibition of DNA binding activity by cadmium was the result of cytoplasmic sequestration, we analyzed p53 localization of in MCF7 cells by immunostaining. In several reports, it has been reported that growth stimulation of serumstarved MCF7 cells induces a protein synthesis-dependent cytoplasmic sequestration of p53 protein (41,42). However, this particular feature has not been found by others (43)(44)(45). In our experiments, p53 was found to localize essentially, if not exclusively, in the nucleus of exponentially growing MCF7 cells (Fig.  3B). In untreated cells, staining with the anti-p53 antibody PAb1801 was heterogeneous, with most cells showing a very low level of reactivity and a small number of cells clearly positive for nuclear staining. As positive control, exposure to Act D, a topoisomerase II inhibitor that is a strong inducer of p53, induced nuclear accumulation of p53 in almost every cell. After 4 h of exposure to 30 M CdCl 2 , p53 was also present in the nucleus of most cells. Therefore, the loss of DNA binding activity shown in Fig. 3A cannot be explained by cytoplasmic sequestration of p53 in cells exposed to cadmium.
Effects of CdCl 2 on p53 Protein Immunological Phenotype-To determine if conformational changes could account for down-regulation of p53 DNA binding activity by cadmium, we analyzed the reactivity of p53 in MCF7 cells using conformation-specific monoclonal antibodies PAb1620 and PAb240. In nontreated cells, p53 adopted a wild-type conformation, reactive with PAb1620 (Fig. 3C). In contrast, after 4 h of exposure to 30 M CdCl 2 , the protein had lost reactivity with PAb1620 and became reactive with PAb240. This change in conformation was dose-dependent, and a small increase in reactivity with PAb240 was already detectable at 10 M CdCl 2 (see Fig. 5C). This structural alteration, which correlates with loss of DNAbinding capacity (see Fig. 3A), suggests that cadmium alters p53 by disrupting its conformation.
Reversibility of the Effects of CdCl 2 on p53 DNA Binding Activity-To determine if the effects of cadmium on p53 were reversible, MCF7 cells were first exposed to CdCl 2 for 4 h (30 M), and the culture medium was then replaced by fresh, CdCl 2 -free medium and further cultured for 12-24 h. Fig. 4 shows that removal of CdCl 2 from medium resulted in partial recovery of p53 DNA binding activity after 12 h (lane 4) or 24 h (lane 7). In contrast, DNA binding activity remained undetectable in cells that were continuously cultured in the presence of CdCl 2 for the same periods of time (lanes 5 and 8). It was noted that culture of MCF7 cells for 12 or 24 h in the presence of CdCl 2 at 30 M affected the survival MCF7 cells (see above). This cytotoxicity may explain the strong decrease of p53 protein levels seen in lanes 5 and 8 (Fig. 4B) and the concomitant absence of DNA binding activity.
Cadmium Prevents p53 Activation by DNA-damaging Agents-The results presented above suggest that cadmium may disrupt p53 protein conformation, thereby abrogating its tumor-suppressive functions in cells exposed to other forms of genotoxic stress. To test this hypothesis, we have analyzed the induction of p53 by Act D in the presence of CdCl 2 . MCF7 cells were exposed for 2 h to CdCl 2 (5-30 M) before the addition of Act D to the culture medium (2.5 ng/ml) and were further cultured in the presence of both agents for 4 h (Fig. 5). In the   FIG. 3. Effects of cadmium on p53 in MCF7 cells. A, down-regulation of p53 DNA binding activity. Subconfluent MCF7 cells were exposed for 4 h to 5-40 M CdCl 2 . As control, cells were exposed for 4 h to H 2 O 2 (100 M). Levels of p53 in nuclear extracts were analyzed by Western blot with DO-7 (30 g of proteins/lane; upper panel). DNA binding was analyzed by EMSA (40 g of protein/binding reaction). All p53 EMSA (middle panel) were performed in the presence of PAb421 as described in the legend to Fig. 1. Lower panel, DNA binding activity of Oct-1 as a control. n.s., complexes that do not contain p53, corresponding to an unidentified protein that binds to DNA in a non-sequencespecific manner. Filled arrowhead, DNA-p53-PAb421 complexes; open arrowhead, Oct-1-DNA complexes. B, nuclear localization of p53 in cells exposed to cadmium. Immunoperoxidase staining of p53 is shown in MCF7 cells labeled with the monoclonal antibody PAb1801 as follows: untreated cells, cells exposed to Act D (2.5 ng/ml), cells exposed to CdCl 2 (30 M), and control cells, without PAb1801. C, cadmium alters p53 conformation. Total extracts of MCF7 cells, treated or not by CdCl 2 (30 M for 4 h), were immunoprecipitated with monoclonal antibodies PAb240 (specific for mutant, unfolded conformation), PAb1620 (specific for wild-type, folded conformation), and PAb421 (which reacts with both conformers). PAb416 (specific for SV40 LT) was used as a negative control. Immunoprecipitates were analyzed by Western blot with the rabbit antibody CM-1. Black arrows, p53. A minor, p53-related band is also detected at 42 kDa.

FIG. 4.
Reversibility of the effects of cadmium on the DNA binding activity of p53. MCF7 cells were exposed to CdCl 2 for 4 h. The culture medium was then replaced by fresh medium, containing or not containing CdCl 2 . Cells were harvested 12 or 24 h after the medium change, and nuclear extracts were prepared. A, time scale of the experiment and sequence of the various treatments for each lane in the autoradiogram in B. In all lanes except lanes 1 and 2, the culture medium was removed after 4 h and replaced as indicated. Stars, time of cell harvest. B, levels of p53 were analyzed by Western blot, and p53 DNA binding activity was analyzed by EMSA (as described in the legend to Fig. 1). Only a portion of the autoradiogram, containing DNA-p53-PAb421 complexes, is shown. absence of CdCl 2 , Act D induced a 3-fold accumulation of p53 protein and a 10-fold activation of DNA-binding capacity. CdCl 2 induced a dose-dependent inhibition of the effect of Act D on p53 DNA binding activity (Fig. 5A), consistent with data presented in Figs. 3 and 4.
To examine whether cadmium at low, nontoxic doses could also impair p53 activation, MCF7 cells were pretreated for 30 min with 10 M CdCl 2 before the addition of Act D for 4 h. DNA binding activity was analyzed by EMSA and quantified by densitometric analysis of autoradiograms (Fig. 5B), and p53 conformation of was analyzed by immunoprecipitation (Fig.  5C). In the absence of Act D, CdCl 2 induced a small but significant increase in p53 DNA binding activity (2.28 Ϯ 0.26-fold).
In the presence of Act D, CdCl 2 (at 10 M) decreased by 40% the p53 activation triggered by Act D alone (Fig. 5B). This effect was correlated with a change in the conformation of a fraction of the p53 molecules as detected immunoprecipitation with conformational specific antibodies (Fig. 5C). CdCl 2 treatment at 10 M induced a small proportion of p53 to become reactive with PAb240 and this effect was more marked in cells exposed to both CdCl 2 and Act D. Overall, these data indicate that CdCl 2 at 10 M had a clear inhibitory effect on p53 activation by the DNA-damaging agent Act D. Similar results were observed with two other DNA-damaging agents, methylmethane sulfonate (an alkylating agent) and hydrogen peroxide (H 2 O 2 ) (data not shown).
Cadmium Impairs Transactivation of Target Genes by p53-To determine if CdCl 2 may impair p53 protein function, we have co-transfected human wild-type p53 and the p53-dependent reporter construct pRGC⌬FosLacZ into the p53-null mouse fibroblast cell line BalbC10.1. Twenty h after transfection, CdCl 2 was added to the culture medium, and ␤-galactosidase activity was measured 12 h later. Fig. 6A shows that the addition of CdCl 2 at the noncytotoxic dose of 10 M reduced ␤-galactosidase activity by 65% in cells transfected with p53 (p Ͻ 0.002). In the absence of p53, CdCl 2 at 10 M had no detectable effects on basal ␤-galactosidase activity. A stronger decrease (85%) was seen with 30 M CdCl 2 . However, part of this effect may reflect the cytotoxicity of exposure for 20 h to CdCl 2 at 30 M. The expression of p21 waf-1 protein in MCF7 cells exposed to CdCl 2 was subsequently analyzed (Fig. 6B). Although p21 waf-1 is transcriptionally regulated by several factors other than p53, there is good evidence that p21 waf-1 expression in MCF7 cells is essentially p53-dependent (46). Levels of p21 waf-1 protein increased at 10 M CdCl 2 but decreased at 30 M. This observation is consistent with the biphasic effect shown in Fig. 3A. Preincubation of cells with CdCl 2 at 30 M prevented the induction of p21 by either of the DNA-damaging agents H 2 O 2 and methylmethane sulfonate. Moreover, the extent of p21 waf-1 induction was reduced by 49 Ϯ 21% in cells exposed to these agents in the presence of 10 M CdCl 2 (p Ͻ 0.001). Similar results were obtained for Mdm-2, another transcriptional target of p53 (data not shown).
Cadmium Inhibits p53-dependent G 1 Arrest after ␥-Irradiation-In MCF7 cells, activation of p53 by ionizing radiation induces in cell cycle arrest in both the G 1 and G 2 /M phases (47).
To determine whether the down-regulation of p53 activity by cadmium resulted in a disruption of cell cycle control, cells were exposed to cadmium prior to ␥-irradiation, and cell cycle distribution was analyzed by flow cytometry after labeling with propidium iodide (Fig. 7 and Table I). In nonirradiated cells, CdCl 2 did not alter cell cycle distribution at 10 or 20 M, but a marked increase in the sub-G 1 fraction corresponding to cell debris (7%) was observed at 30 M. However, even at the latter concentration, the overall distribution of cells in G 1 , S, and G 2 /M phases was essentially unchanged, suggesting that cad- Average and S.D. for autoradiograms of three independent experiments such as the one in A were analyzed. C, immunoprecipitation of p53 in cells exposed to CdCl 2 and Act D. MCF7 cells were exposed to 10 M CdCl 2 for 30 min prior to treatment with Act D. p53 was immunoprecipitated and analyzed as described in the legend to Fig. 3C. Filled arrowheads, p53. Several minor, slowly migrating bands are also detected. They may correspond to ubiquitinylated forms of p53 (64).

FIG. 6. Inhibition of p53 transactivation activity by cadmium.
A, effect of CdCl 2 on transactivation of a p53-dependent reporter gene. Mouse 10.1 cells were co-transfected with 5 g of pRGC⌬FosLacZ (RGC-␤gal) and 1 g of p53pcDNA (p53). Cells were then exposed for 12 h to CdCl 2 at concentrations as indicated, and ␤-galactosidase activity was measured in cytoplasmic extracts (100 g). Results show average and S.D. of three independent experiments. B, effect of CdCl 2 on the induction of p21 waf-1 . MCF7 cells were treated with CdCl 2 at concentrations as indicated for 30 min prior to the addition of methylmethane sulfonate (MMS; 0.5 mM) or H 2 O 2 (100 M) for 4 h. Nuclear extracts were used for analysis of p53 protein levels (by Western blot), p53 DNA binding activity (by EMSA), and p21 waf-1 protein levels (by Western blot). mium did not prevent cell cycle progression.
Irradiation at 5 Gy induced a cell cycle delay in G 1 , detected by the increase in G 1 phase (ϩ13%) and by the decrease in S phase (Ϫ45%). An increase in G 2 /M phases was also observed (ϩ50%) ( Table I). All of these effects were inhibited by the addition of CdCl 2 . With 10 M CdCl 2 , the changes in cell cycle distribution induced by irradiation were ϩ12% in G 1 phase, Ϫ33% in S phase, and ϩ18% in G 2 /M. With 20 M CdCl 2 (and 30 M), the cell cycle distribution of irradiated cells was identical to that of nonirradiated, CdCl 2 -treated cells (Table I).
These data indicate that MCF7 cells exposed to cadmium show impaired cell cycle arrest in response to ␥-radiation, a well known p53-dependent response, and that this effect is not a consequence of the toxicity of cadmium.

DISCUSSION
Zinc is essential for correct folding of wild-type p53. The DNA-binding domain contains a tetrahedrally coordinated zinc that stabilizes two loops at the DNA-binding surface of the protein (13). Several in vitro studies have shown that metal chelation abolishes binding of p53 to specific DNA (12,17,37,48). In addition, metal chelation increases oxidation of cysteines in p53, indicating that zinc binding is not purely structural but also controls the sensitivity of p53 to oxidation-reduction (17,49). Reduction of cysteines stimulates p53 DNAbinding (5,15,50), and Ref-1, a protein that regulates the redox state of several transcription factors, is a potent activator of p53 (51). In vitro, the conformation and DNA-binding capacity of p53 are altered by incubation with metals chemically close to zinc, such as cadmium and copper, but not with cobalt, magnesium, manganese, or iron (17,52). These observations have led to the suggestion that specific metals and redox factors may affect the fine tuning of p53 and participate in the physiological control of p53 functions (5,16,51,53).
We show here that sequence-specific DNA binding of p53 in vitro is decreased by cadmium in a dose-dependent manner. In MCF7 cells, transient exposure (4 h) to cadmium at 20 M and above induced a change in p53 conformation (to the unfolded, PAb240ϩ form) with loss of DNA binding and transcriptional activity. These results are consistent with this idea that cadmium perturbs the folding of p53 in a direct or indirect manner.
In the neuroblastoma cell line HT4, cadmium at concentrations up to 100 M, equal to or greater than those used here, has been shown to induce disruption of intracellular sulfhydryl homostasis and depletion of pools of GSH. These effects were accompanied by an increase in protein thiolation and ubiquitination (40). Extensive oxidative stress induced by cadmium may perhaps inactivate p53 by oxidation, a phenomenon that has been observed in cultured cells exposed to nitric oxide (14), to hydrogen peroxide (53), and to glutathione-depleting agents (54).
On the other hand, cadmium easily substitutes for zinc in several zinc-dependent DNA-binding proteins and inhibits many enzymes containing essential dithiols (55). For example, with the zinc-inducible transcription factor MTF-1, cadmium at 6 M partially abrogates activation of DNA binding by the addition of 30 M zinc and totally inactivates it at 60 M (56). MTF-1 contains six Cys 2 -His 2 zinc fingers, which are thought to bind cadmium with high affinity (56). Cadmium differs from zinc in that it has a higher affinity for thiols and has a larger atomic radius. The binding affinity of cadmium to cysteine thiolate clusters in zinc finger proteins is 2-3 orders of magnitude higher than that of zinc (22,24,57). Consistent with this notion, we found that the effect of cadmium on p53 was not reversed by the addition of excess ZnCl 2 (up to 25-fold) to cadmium-treated MCF7 cells (data not shown). Therefore, although our data do not provide a formal proof that cadmium can displace zinc from native p53, our results are consistent with the idea that cadmium perturbs the metal-dependent folding of the DNA-binding domain.
In MCF7 cells, cadmium exerts complex, biphasic effects on p53 protein levels and DNA binding activity. At low concentrations (up to 10 M), cadmium alone induces a small (2-3-fold) but reproducible accumulation of p53 protein, correlated with slightly enhanced DNA binding activity (2.28 Ϯ 0.26-fold). This effect may be due to p53 protein stabilization by low levels of oxidative DNA-damage induced by cadmium. However, increasing the concentration of cadmium does not result in higher levels of p53 protein activation. In contrast, it significantly decreases p53 DNA binding activity (at 20 M) and protein levels (at 40 M). Furthermore, inhibition of p53 activity correlates with a change in protein conformation, with loss of PAb1620 reactivity (wild type-specific) and acquisition of the PAb240-positive phenotype (Fig. 3C). Along with the observation that cadmium does not prevent p53 localization in the nucleus (Fig. 3B), these data indicate that cadmium inhibits FIG. 7. Effects of CdCl 2 on cell cycle distribution of MCF7 cells after ␥-irradiation. MCF7 cells were cultured for 4 h in the presence of CdCl 2 at the indicated concentrations before irradiation at 5 Gy. One h after irradiation, the medium was replaced with drug-free culture medium, and cells were cultured for another 11 h. Cells were then trypsinized, treated with RNase, labeled with propidium iodide, and analyzed with a Becton-Dickinson FACSCalibur flow cytometer using CellQuest software. Results are displayed as three-dimensional histograms of cell number versus DNA contents.

TABLE I Effects of CdCl 2 on cell cycle distribution of MCF7
cells after ␥-irradiation The cell cycle distribution of MCF7 cells analyzed in Fig. 7 was evaluated using the Modfit LT version 2.0 software for modeling cell cycle. Results are displayed as percentage of total number of cells. Controls are nonirradiated cells exposed to CdCl 2 at concentrations as indicated. 5 Gy corresponds to irradiated cells exposed to CdCl 2 as indicated (see also the legend to Fig. 7). p53 by turning it into an inactive, "mutant-like," form. Cadmium at 30 M induces total inhibition of p53 protein activation in response to DNA-damaging agents such as Act D, methylmethane sulfonate, or H 2 O 2 . This inhibition resulted in a loss of transcriptional activation of several p53 target genes including p21 waf-1 . Moreover, cadmium at noncytotoxic concentrations (10 M) is sufficient to significantly reduce (by about 40%) the extent of p53 induction by DNA-damaging agents and therefore to perturb the response of p53 to DNA damage.
The apparent contradiction between the effects of cadmium at 10 and 30 M may be resolved by considering that cadmium has two opposite effects on p53, with first protein stabilization as a result of generation of DNA damage by low doses of cadmium and, second, direct inhibition of p53 protein by metal substitution and conformational modifications at higher doses of cadmium. The level of p53 DNA binding activity detected in the presence of cadmium would thus depend upon a subtle balance between these two mechanisms.
Inhibition of p53 DNA binding activity by cadmium has important functional consequences in cultured cells. First, cadmium reduces p53-dependent transactivation of reporter or endogenous target genes. Second, cadmium prevents the cell cycle arrest induced by low doses of ␥-irradiation in MCF7 cells, suggesting that cadmium can effectively suppress p53 protein function. Cells exposed to cadmium thus behave in a manner analogous to p53-deficient cells that retain the capacity to proliferate after exposure to DNA-damaging agents. A similar hypothesis has been proposed in the case of excess production of nitric oxide, which also induces conformational and functional changes in wild-type p53 (14). Impairment of p53 function by cadmium may contribute to decrease the cell capacity to respond to the DNA damage induced by other carcinogens, thereby increasing the likelihood of acquiring mutations leading to cancer.
Cadmium is highly toxic in most biological systems and has a very long biological half-life (about 25 years in humans (20). Therefore, it is essential to consider whether the concentrations of cadmium used in our experiments are compatible with those that occur in target cells of exposed organisms. After exposure to cadmium, most of the intracellular pool of cadmium is bound to MTs, a class of inducible, metal-binding proteins that sequester cadmium and protect cells from its toxic effect. However, experiments with MT-I and -II knockout mice showed that cadmium also accumulates to high levels in the absence of MT. In MT-deficient mice, CdCl 2 injected subcutaneously at 30 g/kg accumulates in liver cells within 3-6 h at up to 20 -25 g/g of fresh tissue (58). These levels may correspond to intracellular concentrations 3-10-fold higher than those used in the present study. Although this dose of cadmium was toxic in MT-deficient mice, it produced only mild hepatotoxicity in control mice. Concentrations of cadmium of up to 25 M are well tolerated in many cultured cell lines (30). Therefore, the effects reported here are compatible with concentrations of cadmium that are not lethal and can occur in biological systems after acute or chronic exposure.
Alteration of p53 protein conformation by cadmium was described previously, using in vitro translated p53, with concentrations of CdCl 2 of 50 -100 M (17). In our study, we show that much lower concentrations of cadmium (10 -30 M) are able to alter p53 conformation and function in intact cells. This is the first report that a metal compound can inactivate p53 at doses compatible with biological effects.
In 1995, Zheng et al. (58) reported that cadmium could increase p53 mRNA levels in liver cells of mice injected with CdCl 2 . We did not observe such an effect in cultured MCF7 cells. It is important to note that induction of p53 mRNA was observed as a late event (after 6 -12 h) in mice receiving a high, hepatotoxic dose of cadmium. Therefore, it is possible that elevated p53 mRNA may represent a response to cell damage rather than a direct effect of cadmium on p53 gene expression.
The mechanism of p53 inactivation described here may account for some of the unexplained properties of cadmium as a carcinogen. Indeed, cadmium is a weak genotoxic agent compared with metals such as copper, iron, nickel, and chromium. Therefore, mechanisms other than direct genotoxicity have been proposed to explain cadmium carcinogenesis (59). Exposure to cadmium enhances the persistence of DNA lesions induced by mutagens such as benzo(a)pyrene and methylmethane sulfonate in human cells, suggesting that cadmium may inhibit DNA repair. Recently, Dally and Hartwig (30) have shown that cadmium, as well as nickel, inhibits the repair of DNA damage after irradiation. These authors propose that cadmium may either inactivate repair enzymes directly, for example by reaction with a histidine or cysteine residue, or compete with and displace essential metal ions, a hypothesis compatible with the results presented here. Inhibition of p53 function may explain the persistence of DNA lesions in cells exposed to both carcinogens and cadmium. According to this model, cadmium would not act as a conventional mutagen but rather as an indirect carcinogen that sensitizes cells to the genotoxic effects of other carcinogens by switching off essential components of cell cycle control and DNA repair pathways involving p53.
Cadmium exerts complex effects on the growth and survival of normal and cancer cells. The metal was shown to induce apoptosis or necrosis in some cells and tissues and to reduce the growth and metastasis of human lung carcinoma xenograft in nude mice (60). In contrast, cadmium was shown to inhibit apoptosis induced by DNA-damaging metals such as chromium (61). These observations suggest that the sensitivity to cadmium may vary from one cell type to another and that some cancer cells may be hypersensitive to cadmium. The cytotoxic impact of cadmium may be related to the cellular level of metallothioneins, which is frequently deregulated in cancer cells (62).
It would be naive to suggest that effects on p53 alone can explain all of the cadmium carcinogenicity. Indeed, it is likely that cadmium substitutes for zinc and alters the function of a number of other cellular proteins. For example, Cd 2ϩ (as well as a number of other metal ions) has been shown to alter the nucleotide selectivity of human DNA polymerase ␤ in vitro (63). In addition, factors such as competition between metals and interactions with metallothioneins should also be considered. However, we believe that our observations represent a important step in the understanding of the carcinogenic potential of cadmium. Moreover, these observations also provide a model system for determining how essential metals such as zinc or metal chelators may be used in preventive approaches to reduce cadmium carcinogenesis.