Induction of internucleosomal DNA fragmentation by carcinogenic chromate: relationship to DNA damage, genotoxicity, and inhibition of macromolecular synthesis.

Hexavalent chromium (Cr) compounds are respiratory carcinogens in humans and animals. Treatment of Chinese hamster ovary cells with 150 and 300 microM sodium chromate (Na2CrO4) for 2 hr decreased colony-forming efficiency by 46 and 92%, respectively. These treatments induced dose-dependent internucleosomal fragmentation of cellular DNA beyond 24 hr after chromate treatment. This fragmentation pattern is characteristic of apoptosis as a mechanism of cell death. These treatments also induced an immediate inhibition of macromolecular synthesis and delayed progression of cells through S-phase of the cell cycle. Cell growth (as evidenced by DNA synthesis) was inhibited for at least 4 days and transcription remained suppressed for at least 32 hr. Many of the cells that did progress to metaphase exhibited chromosome damage. Chromate caused the dose-dependent formation of DNA single-strand breaks and DNA-protein cross-links, but these were repaired 8 and 24 hr after removal of the treatment, respectively. In contrast, Cr-DNA adducts (up to 1/100 base-pairs) were extremely resistant to repair and were still detectable even 5 days after treatment. Compared with other regions of the genome, DNA-protein cross-links and Cr adducts were preferentially associated with the nuclear matrix DNA of treated cells, which was 4.5-fold enriched in actively transcribed genes. Chromium adducts, formed on DNA in vitro at a similar level to that detected in nuclear matrix DNA, arrested the progression of a DNA polymerase in a sequence-specific manner, possibly through the formation of DNA-DNA cross-links.(ABSTRACT TRUNCATED AT 250 WORDS)

Potentially, carcinogen toxicity may occur by one of two mechanisms: necrosis or apoptosis (16). These two types of cell death differ morphologically and biochemically. Necrotic cells exhibit an early characteristic cell swelling with the loss of the ionic gradient across the cell membrane and subsequent DNA degradation by released lysosomal enzymes (17). Necrotic cell death does not appear to be cell cycle related. In contrast, cells dying by apoptosis undergo shrinkage while the ionic gradient across the cell membrane is maintained until later stages of the apoptotic process (18). A characteristic feature of apoptosis is that DNA is fragmented internucleosomally before the loss of cell membrane integrity occurs. This is believed to result from the activation of a nuclear, Ca2+/Mg2+-dependent or pH-sensitive endonuclease, possibly DNase 11 (16,19). In many cases the process of apoptosis appears related to arrest of cells in a distinct phase of the cell cycle (18,20,21). Carcinogen-treated cells that do overcome the cell cycle block may begin the process of tumor formation, suggesting that apoptosis could serve as a method of escape from the process of multistage carcinogenesis initiated by genotoxic agents.
Recently, several carcinogenic and chemotherapeutic agents, including methylprednisolone, 4-hydroxytamoxifen, methotrexate, 5-fluorouracil, 5-fluorodeoxyuridine, etoposide, N-methyl-N'-nitro-N-nitrosoguanidine, cisplatin, y-irradiation, and nitrogen mustard, have been shown to induce apoptosis (18,(20)(21)(22)(23)(24). Recent studies in this laboratory have sought to understand the role of apoptosis in hexavalent chromium-induced toxicity and carcinogenicity. Treatment of Chinese hamster ovary (CHO) cells with toxic doses of sodium chromate was found to result in an immediate inhibition of cell growth, and in macromolecular synthesis; however, cell death was delayed for at least 24 hr. Between 24 and 72 hr post-treatment, considerable cell death occurred following an accumulation of cells in the S-phase of the cell cyde. Many ofthe cells that did progress to metaphase exhibited chromosome damage. Intemudeosomal fagmentation of cellular DNA was detected beyond 24 hr after chromate treatment. This fragmentation pattern is characteristic of apoptosis as a mechanism of cell death. Although treatment with 150 pM chromate induced several types of DNA damage, only chromium adducts remained unrepaired throughout the period when cell death occurred. This lesion was found to be preferentially associated with the nudear matrix, the putative site of replication and transcription in the cell (25). Chromium-DNA adducts formed on a synthetic DNA template in vitro inhibited progression of a DNA polymerase in a dose-dependent manner. These results suggest that the persistent chromium-DNA adducts may be responsible for the protracted cell cyde delay and transcriptional inhibition caused by chromate. They also suggest that potentially, escape from chromate-induced apoptosis may be one of the steps involved in chromate-induced carcinogenesis.

Materials and Methods
CAl Cukue, Chroma e Treamnt, and CytouityAssys CHO AA8 cells were cultured as monolayers as described previously (13). Cells were treated with 150 or 300 PM Na2CrO4 for 2 hr in complete medium. Cytotoxicity Assays. Cytotoxicity was assayed by measuring reduction in plating efficiency relative to controls. Two hundred cells were seeded in 5 ml of basal media Eagle's onto 60 mm dishes and allowed to grow for 17 hr. Cultures were treated with Na2CrO4 for 2 hr. After treatment the medium was removed, cells were washed with PBS and then refed with 5 ml of fresh medium. Seven to 10 days later colonies were stained with crystal violet and counted. Experiments were performed in triplicate. There were three replicates per treatment group in each experiment.
DNA Fragmentation Analyis. Cellular DNA was analyzed by the method of Martikainen et al. (26). Cell pellets were suspended in 5 ml of 10 mM Tris-Cl pH 8.0, 0.5% (v/v) Triton X-100, 5 mM EDTA, 0.5% sodium dodecyl sulfate (SDS), and digested with 300 pglml proteinase K for 18 hr at 37°C. After digestion, the DNA was sequentially extracted with phe-nol, phenol/chloroform, chloroform, and ethanol precipitated. Purified DNA (10 pg) was electrophoresed on a 1.6% agarose gel. The gel was stained with 0.5 gml ethidium bromide in 10 mM Tris-Cl pH 8 (27). RNA snthesis was essed by incorporation of [ H]uridine into total cytoplasmic RNA or mRNA as described previously (13). GRP78 transcript levels were determined by Northern blot hybridization analysis as described by Manning et al. (13).

Preparaton ofNudei and Nudear
Matrix a on CHO cells were suspended in RSB buffer (10 mM Tris, pH 7.4, 10 mM NaCl, 3 mM MgCl2, 1 mM phenyl methyl sulfonyl fluoride [PMSF]) and incubated on ice for 20 min. Nonidet P-40 (Sigma Chemical Co., St Louis, MO) was added to a final concentration of 0.05% and the cells were homogenized with 10 to 12 strokes of a glass-teflon homogenizer. The nuclei were collected by centrifisgation (750 xg, 40C for 5 min), and washed twice with RSB containing 0.25 M sucrose. Finally, nudei were separated from whole cells by centrifugation at 75,000xg, 4°C for 1 hr through a cushion of 1.8 M sucrose in RSB. Nudear pellets were stored at -80°C until required. Chromatin fractionations were performed essentially as described by Obi

Analysis ofDNA Damage
Analysis of DNA damage by alkaline elution was performed by the method of Kohn et al. (29,30) as previously described by Patiemo et al. (10,13,31). DNA-protein cross-links were isolated from whole nudei and nudear matrix fractions were prepared from cells which had been prelabeled with 0.5 pCi/ml of trans- [35S]metabolic labeling reagent (a mixture of L- [35 ]methionine and L-35 S]cysteine; 1066 Ci/mM, ICN Biomedicals) for 24 to 48 hr prior to treatment with Na2CrO4. Samples were mixed with 15 vol of urea extraction buffer (8 M urea, 1% Triton X-100, and 10 mM Tris, pH 7.4) and rocked for 4 hr at room temperature. These mixtures were then applied to a Qiagen tip-500 column (Qiagen, Chatsworth, CA) which had been equilibrated with 10 ml of buffer A (0.75 M NaCl, 50 mM MOPS, pH 7.0, 15% ethanol, 0.15% Triton X-100, and 0.1 mM PMSF). The column was then washed with 30 ml of buffer B (1 M NaCl, 50  Chromium bound to DNA purified from whole cells; nudei and nudear matrix fractions were quantified by scintillation counting as described previously (13).
In Vi" Biing ofChrm ium (ll) to DNA Chromium (III) chloride (Sigma Chemical Co.) was dissolved in 10 mM Tris-Cl, pH 7.6, immediately before use. pSV2neoTS DNA (7 pg) at a concentration of 840 pM DNA-phosphate was treated with CrCl3 in a total volume of 25 pl for 16 hr at room temperature. For polymerase arrest assays, chromium-treated template DNA was denatured with 0.2 M NaOH at 370 C for 30 min. Unbound chromium was removed by passing samples over Chromaspin-30 columns in nuclease-free water (Clontech Laboratories Inc., Palo Alto, CA). To measure chromium binding, DNA was treated with 5lCrC13 (specific activity 200-900 Ci/g Cr; ICN Biomedicals). Following treatment, samples were diluted 50-fold in 10 mM Tris, pH 7.6, and aliquots were precipitated with an equal volume of 20% trichloroacetic acid (TCA) on ice for 5 min. The precipitated sample was applied to a GF/C glass microfiber filter (Whatman International Ltd., Maidstone, England) under vacuum. Filters were sequentially washed with 10% TCA (15 ml), 100% ethanol (5 ml), and air dried. The filters were scintillation counted using the tritium channel of a Beckman LS 7000 scintillation counter (Beckman Instruments, Inc., Fullerton, CA). To measure counting efficiency, aliquots of a dilution series of 51CrCl3 were also applied to filters, air dried, and scintillation counted. Data were expressed as mmole of 5ICr molecules bound per mole of DNA-phosphate.

Poymweraserr Assay
A synthetic 88-bp template was synthesized using a Cyclone Plus DNA Synthesizer (Millipore, Bedford, MA) ( Figure 7)) and cloned into the plasmid pSV2neo using standard techniques (32) at the site created by digestion of the vector with Eco RI and Bam HI restriction endonudeases to generate pSV2neoTS. A 25-base oligonudeotide primer [5'-CGTATCACGAGGC-CCTIT-l CGTCTTC-39 complementary to the region two bases upstream of the mutational target sequence was synthesized. Ten pmole of primer were end-labeled with 1 p1 of [Y-32P]-ATP (20 pmole, 170 pCi) (ICN Biochemicals) in a total volume of 20 pl containing 2 1pl of IOxT4 polynucleotide kinase buffer (400 mM Tris-Cl, pH 7.5, 100 mM MgCl2, 50 mM DTI), and 1 pl (8 units) of T4 polynucleotide kinase (Promega, Madison, WI) at 370C for 10 min. The reaction was stopped by incubating tubes at 900C for 2 min and the labeled primer was stored at -20°C. An aliquot of chromium-treated, denatured pSV2neoTS (2 pg in 7 pl) was combined with 2 pl of stop-assay reaction buffer (200 mM Tris-Cl, pH 7.5, 50 mM MgC12, 250 mM NaCl) and 1 pl of end-labeled primer (0.5 pmole). This mixture was heated to 65°C for 2 min and cooled slowly to below 300C. After annealing, each reaction was mixed with 1 pl of 0.1 M DTT, 2.5 pl of water and 2 pl of Sequenase Version 2.0 T7 polymerase (United States Biochemical, Cleveland, OH) diluted 1:8 in the manufacturer's enzyme dilution buffer (10 mM Tris-Cl, pH 7.5, 5 mM DTT, 0.5 mg/ml bovine serum albumin). The tubes were mixed well, briefly centrifuged, and 3.5 pl of each mixture was transferred to fresh tubes containing 2.5 pl of 160 pM each of dATP, dCTP, dTTP, and dGTP. These elongation reactions were incubated at 450 C for 5 min. Stop solution (4 pl of 95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol FF) was then added and the tubes heated to 78°C for 2 min and placed on ice.
Aliquots of each reaction (2 p1l) were analyzed on a 6% denaturing polyacrylamide sequencing gel. To determine the position of polymerase arrests, sequencing reactions of the DNA template were also analyzed on the same gel. Gels were run for approximately 2 hr at 2000 V in 1 xTBE buffer (0.045 M Tris-borate, 1 mM EDTA). Following electrophoresis, gels were wrapped in cellophane and exposed to X-ray films between two intensifying screens for 5 to 17 hr at -700 C.

Tarcity, and Effects on Cdl Growth
Treatment of CHO cells with 150 or 300 pM Na2CrO4 for 2 hr reduced colonyforming efficiency by 46 and 92%, respectively compared with untreated controls (not shown). Colonies in chromate-treated cultures were notably smaller than the corresponding untreated control colonies indicating that cell growth had been inhibited (not shown). Consistent with this observation, Figure 1 shows that immediately after treatment with sodium chromate the growth of CHO cells was suppressed. There was, however, no decrease in the number of adherent cells for at least 24 hr. Between days 1 and 3 after treatment the number of adherent cells in the 150-and 300-pM treatment groups decreased by 35 and 85%, respectively, indicating that considerable cell death occurred during this period. In both treatment groups greater than 95% of adherent and nonadherent cells excluded trypan blue up to at least 48 hr suggesting that extensive cell lysis had not occurred during this period. By 72 hr, none of the nonadherent cells excluded trypan blue, indicating that they were no longer viable. The number of adherent cells did not begin to increase until at least 5 days after treatment.
Analysis ofGenomic DNA Genomic DNA prepared from pooled unattached and adherent cells at various times after treatment with Na2CrO4 was analyzed by agarose gel electrophoresis (Figures 2, 3). DNA isolated from untreated cells was of high molecular weight (greater than 24 kb) and unfragmented at all times examined (Figures 2, 3). Up to 24 hr after chromate treatment, little DNA fagmentation was apparent ( Figure 2) even on highly overloaded gels (not shown). In contrast, dose-dependent internucleosomal DNA fragmentation, indicative of apoptosis, was detectable 48 hr after treatment of cells with either 150 or 300 pM chromate Cells were treated with 0, 150, or 300 pM chromate for 2 hr. Either immediately after this treatment (0 hr), 24 or 48 hr later, DNA was isolated from pooled adherent and nonadherent cells and analyzed on a 1.6% agarose gel. Molecular weight markers were a 123 bp ladder (Gibco, BRL, Gaithersburg, MD). These data are representative of five independent experiments. that expected to result from the internucleosomal DNA cleavage that occurs during apoptotic cell death (16). Seven days (168 hr) after chromate treatment, no DNA fragmentation was apparent, suggesting that the process of apoptosis was completed in these cell populations by this time (Figure 3). Figure 4 shows the residual DNA synthesis occurring in Na2CrO4-treated CHO cells expressed as a percentage of that occurring in logarithmic phase control cells. Treatment with 150 and 300 pM chromate for 2 hr immediately suppressed DNA synthesis to 30 and 10% of control values, respectively. This inhibition continued for at least 4 days in both treatment groups. By day 5, there was some recovery of DNA synthesis, to 70 and 50% of control values in the 150 and 300 pM treated cells, respectively. Figure 5 shows the effect of Na2CrO4 on the synthesis of cytoplasmic RNA and mRNA in CHO cells expressed as a percentage of that occurring in control cells. The extent of [3H]uridine incorporation into total cytoplasmic RNA and mRNA in a 1 hr pulse was measured up to 32 hr after treatment with Na2CrO4. Treatment with 150 and 300 pM chromate suppressed total RNA synthesis by approximately 50 to 60% and 75 to 90%, respectively compared with controls ( Figure 5). These concentrations also suppressed mRNA synthesis to approximately 30 and 20% of control levels, respectively ( Figure 5). Suppression of transcription of both total and mRNA synthesis occurred immediately after treatment (13) and this suppression persisted for at least 32 hr ( Figure 5).

Efets on Macrmolecuar Synthesis
Chromate was found to inhibit the induction of expression of the inducible GRP78 gene (33) by tunicamycin in a time-and concentration-dependent manner. CHO cells were exposed to 150 or 300 pM chromate for 2 hr. At various times after this treatment, cultures were incubated with tunicamycin for 8 hr to induce GRP78 gene expression. Increases in GRP78 mRNA levels were measured by Northern blot hybridization analysis. Immediately after treatment of cells with 150 and 300 pM chromate, induction of GRP78 expression was suppressed to approximately 10 and 2% of that occurring in nonchromate treated cells, respectively ( Figure 5, 8-hr column). In contrast, 24 hr after treatment with 150 pM chromate, there was a partial recovery of GRP78 inducibility to approximately 56% of that occurring in untreated cells ( Figure 5, 32hr column). There was, however, no induction of GRP78 expression in cultures treated with 300 pM chromate at this time.
Chromate-Induced DNA Damage DNA damage induced in CHO cells by 2 hr of treatment with 150 or 300 pM Na2CrO4 was measured by alkaline elution. DNA single-strand breaks (13) and DNA-protein cross-links ( Figure 6A) were detected immediately after treatment of cells with chromate. The single-strand breaks were essentially repaired within 8 hr (13). DNA-protein cross-links were more persistent but were repaired in cells treated with 150 pM chromate by 24 hr after treatment ( Figure 6A). Cells treated with the 300 pM dose contained a much reduced level of cross-links at 24 hr compared with 0 hr, indicating that DNA repair had occurred. Figure 6B shows that chromium-DNA adducts were formed immediately after treatment of cells with [51Cr]-labeled Na2CrO4. After a slight reduction in adduct levels between 0 and 8 hr posttreatment, little further decrease occurred up to at least 32 hr ( Figure 6B).
Chromium-DNA adducts and DNAprotein cross-links isolated from the nuclei and nuclear matrix of CHO cells treated with 150 pM chromate were quantified. Immediately after treatment with chromate, chromium-DNA adducts were markedly enriched in the nuclear matrix (approximately 4-fold compared with total nuclear DNA) ( adducts were still enriched in nuclear matrix DNA but at a slightly decreased level compared with 0 hr. Consistent with the data presented in Figure 6B; however, there was little decrease in adduct levels in total nuclear DNA (Table 1). Chromiuminduced DNA-protein cross-links, which were stable to 8 M urea and 2% SDS, were also enriched in the nuclear matrix (Table 1). Consistent with the alkaline elution data presented in Figure 6A, the matrix-associated cross-links reached a°@  prior to the indicated times, cells were incubated in the presence of [3HJuridine to assess total cytoplasmic RNA synthesis and messenger RNA synthesis. Transcription occurring in chromate-treated cells was expressed as a percentage of that occurring in control cultures at the same time ± SEM of two to four independent experiments. Either immediately after treatment with chromate (0 hr) or 24 hr later, cells were incubated with 10 pg/ml tunicamycin for 8 hr to induce GRP78 expression. At the end of these incubation periods (8 and 32 hr, respectively), RNA was isolated and GRP78 transcript levels were assayed by Northern blot hybridization analysis. Results are expressed as a percentage of the increase in GRP78 mRNA levels occurring in nonchromate-treated control cells exposed to tunicamycin ± SEM of threedeterminations. Data reproduced with permission from Manining et al. (13). aThe presented data are the mean ± SE of two to four determinations in each case. bDNA-protein cross-links stable to extraction with urea and SDS were quantified as described in Materials and Methods. CResults were standardized to the value obtained with nonchromate-treated, unfractionated nuclei in each experiment. The level of protein bound to the nuclear matrix of untreated cells is also shown.

Effects on in Vitro Replication
To investigate the effect of chromium-DNA adducts on in vitro replication, a synthetic 88 bp DNA template was synthesized and cloned into the Eco Ri-Bam Hi site of the plasmid pSV2neo to form pSV2neoTS. The sequence of this template is shown in Figure 7 and includes a 24 bp sequence 67% enriched in AT residues, -a 17 bp sequence 100% enriched in GC residues, and a 17 bp self-complementary inverted repeat. These unique regions were included in the target sequence because each had been demonstrated to be particularly susceptible to chemical mutation (34). The plasmid pSV2neoTS was linearized by digestion with Not I to generate a discrete 105 bp full length elongation product in the polymerase arrest assay. Linearized plasmid was treated at a concentration of 840 PM with 0, 35, 105, 210, 420, or 840 pM CrCl3 for 16 hr at room temperature.
These doses are lower than the intracellular concentration of chromium occurring in CHO cells treated with 150 jiM Na2CrO4 for 2 hr which was determined to be 5.2 mM (Xu J, Manning FCR, Patierno SR, unpublished). They are also comparable to the intracellular concentration of chromium occurring in cells treated with lead chromate for 24 hr (35). The selected treatments correspond to DNA-phosphate: Cr3+ ratios of 1:0, 1:0.04, 1:0.13, 1:0.25, 1:0.5, and 1: 1, respectively. After denaturation with NaOH and removal of unbound chromium by passage over Chromaspin-30 chromatography columns, the treated DNA was analyzed by polymerase arrest assay using Sequenase Version 2.0 T7 DNA polymerase. Few arrests occurred when untreated DNA was used as a template in this assay. The majority of synthesized DNA was of a size that corresponded to the expected 105 bp full-length fragment (Figure 7, control lane). In contrast, treatment of pSV2neoTS with CrCl3 resulted in the concentration-dependent synthesis of truncated DNA molecules and a reduction in the amount of full-length product (Figure 7, 35 to 840 FM lanes). Comparison of the size of these truncated molecules with the corresponding sequencing reactions (Figure 7; G, A, T, and C lanes) indicated that the polymerase arrests occurred in a sequence-specific manner, in the GCrich (100% GC) and palindromic (65% GC) sequences and to a lesser extent in the AT-rich region (27% GC). This effect was most pronounced after treatment with the lower concentrations of CrCl3 (Figure 7, 35 and 105 jiM lanes). The template sequence prior to the AT-rich region, in which a number of arrests occurred, is also enriched in GC (61%). At the highest concentrations of CrCl3 employed, almost every base was affected to some extent Volume 102, Supplement 3, September 1994 aDNA at a concentration of 840 p!yl DNA-phosphate was treated with the indicated concentrations of CrCI3 for 16 hr at room temperature. ±e data are the average ± the range of two independent experiments. (Figure 7, 420 and.840 pM lanes).Adentical polymerase arrests west detected when heat denaturation was substitmted for alkaline denat4fation in the:arrest assay (36). Treatment of pSV2n6oTS with concentrations of [5ICr]-labeled CrCl3 that produced polymerase arrests indicated that chromium-DNA adducts were formed on template DNA in a dose-dependent manner (Table 3).

Discussion
The mechanisms of toxicity and carcinogenicity of hexavalent chromium compounds are currently poorly understood.
Recently, however, several carcinogenic agents have been reported to induce apoptotic cell death in mammalian cells (18,(20)(21)(22)(23)(24). Studies in this laboratory have therefore investigated whether apoptosis could be a mechanism by which chromium (VI) toxicity is mediated.
Treatment of CHO cells with cytotoxic concentrations of Na2CrO4 for 2 hr immediately inhibited cell growth ( Figure 1) and DNA synthesis (Figure 4), however no immediate cell death, as judged by trypan blue exclusion, occurred. Rapid inhibition of DNA replication by chromate has previously been reported by Levis et al. (37). Bakke et al. (38) have previously reported that cells treated with chronic, high concentrations of chromate arrest in S-phase. Time After Oorrmate Treatment "i J In agreement with these observations, 24 hr after treatment with 150 pM chromate, cells were found to accumulate in the Sphase of the cell cycle (not shown). An Sphase delay has also been reported after treatment of cells with the toxic metal salts, CdCI2 and NiCI2 (39). In addition, nitrogen mustard blocks progression of cells through this phase of the cell cycle prior to the onset of apoptosis (21). Many of the cells that did progress to metaphase exhibited a high level of chromosome damage 24 hr after treatment with 150 pM chromate ( Table 2). The majority of the damage observed were chromatid lesions. These results are consistent with our previous finding that the relatively insoluble hexavalent chromium salt, lead AT-rich palindrome GC-rich chromate, is clastogenic to CHO cells (15). Metaphase chromosomes could not be obtained 24 hr after treatment with 300 pM chromate. This may be due to the extreme suppression of DNA synthesis produced by this dose (Figure 4). In agreement with this possibility, although cells treated with 150 ,uM chromate grew slightly in the 24 hr following treatment, no growth was detected after treatment with the 300 FM concentration (Figure 1) (13). A characteristic feature of apoptosis is that genomic DNA is fragmented intemucleosomally before the loss of cell membrane integrity occurs (18). The size of DNA isolated from chromate-treated CHO cells was therefore analyzed on agarose gels. Dose-dependent internucleosomal DNA fragmentation was detected 48 hr after treatment of cells with Na2CrO4 ( Figure 2). Beyond 24 hr, a large number of cells became detached from the culture dishes ( Figure 1). However, the membranes of these cells excluded trypan blue until between 48 and 72 hr after treatment. This is consistent with the known parameters of apoptotic cell death (16). By 72 hr after treatment with either 15 or 300 pM chromate, considerable internudeosomal DNA fragmentation was apparent (Figure 3), suggesting that a large number of cells were undergoing apoptosis at this time.
Following a protracted inhibition of growth, chromate-treated cells began to recover, as evidenced by increased DNA synthesis on day 5 after treatment ( Figure 4). This was followed by an increase in cell number between days 6 and 8 ( Figure 1). By day 7 after treatment, no DNA fragmentation was detected ( Figure 3) suggesting that cells were no longer undergoing apoptosis at this time.
Treatment of CHO cells with 150 and 300 pM Na2CrO4 produced an immediate inhibition of protein synthesis (40). Although the role of protein synthesis in apoptosis is unclear, the protein synthesis inhibitor cycloheximide has been reported to induce apoptosis in some instances (41). It is therefore possible that the inhibition ofprotein synthesis produced by chromate could be involved in the initiation of apoptosis.
The formation and repair of DNA damage oi chromate-treated CHO cells was examined. DNA single-strand breaks, DNA-protein cross-links, and chromium DNA adducts were formed in a concentration-dependent manner; subsequently each type of damage was repaired at a different rate. Single-strand breaks were rapidly repaired, essentially within 8 hr (13). The DNA-protein cross-links were more persis-tent than the strand breaks but were also repaired. By 24 hr no cross-links remained in cells treated with 150 pM chromate, while cells exposed to the 300 pM dose contained considerably lower levels of this type of damage at this time ( Figure 6A).
The most persistent form of damage was chromium-DNA adducts that remained unrepaired for at least 32 hr ( Figure 6B). This was the only type of damage to persist after the onset of cell death induced by both 150 and 300 pM chromate. In fact, adducts were still detectable in cells 5 days after treatment with the 150 pM dose (13). Potentially therefore, chromium-DNA adducts could be involved in the mechanism of chromate-induced toxicity or carcinogenicity. Alternatively, DNA single-strand breaks or DNA-protein cross-links, although repaired prior to the onset of cell death, may cause the induction of a signal that induces cells to undergo apoptosis. A further possibility is that free radicals, produced as a result of the intracellular reduction of hexavalent chromium (8,(42)(43)(44), may trigger apoptosis in affected cells.
Compared with total nuclear DNA, chromium-DNA adducts were found to be enriched in the nuclear matrix of CHO cells both immediately after treatment with 150 pM Na2CrO4 and 24 hr later (Table 1). Both these data and those presented in Figure 6B suggest that chromium-DNA adducts are very poorly repaired in CHO cells.
The nudear matrix is believed to be the site of replication in the cell (25). Potentially, chromium-DNA adducts associated with this fraction could interfere with replication and hence cause the inhibition of DNA synthesis and cellular growth observed after treatment of cells with Na2CrO4 (Figures 1, 4). One mechanism by which this may occur would be if chromium-DNA adducts inhibited the action of DNA polymerases. To test this possibility, the effect of chromium-DNA adducts on the in vitro replication of DNA by Sequenase Version 2.0 T7 DNA polymerase was examined. For these experiments a synthetic template containing GC-and AT-rich regions and a palindrome sequence (pSV2neoTS) was employed. A reduced trivalent form of chromium is believed to be the species which binds to cellular DNA (45). Treatment of pSV2neoTS with CrCl3 resulted in the dose-dependent formation of truncated molecules (Figure 7). After treatment with relatively low concentrations of CrCl3 (35,105, and 210 pM), these polymerase arrests occurred preferentially in the GC-rich regions of the template molecule. This is consistent with the known affinity of chromium binding for guanine residues (46). Notably, DNA treated in vitro with 35 pM CrCI3 contained a similar level of adducts (Table 3) to that detected in the nudear matrix of cells treated with 150 pM chromate (Table 1).
Recent data obtained in this laboratory suggest that the CrCI3-induced polymerase arrests may result from the formation of DNA-DNA cross-links (36). A similar inhibition of polymerase action via the formation of chromium (III)-induced DNA-DNA cross-links has previously been reported by Snow and Xu (45). DNA-DNA cross-links formed in the nuclear matrix may therefore be a potential mechanism by which the chromate-induced inhibition of DNA synthesis is mediated.
As an alternative to the alkaline elution technique (29,30) we have developed a procedure to isolate and quantify DNAprotein cross-links that are stable to sequential extraction with urea and SDS (Xu J, Manning FCR, Patierno SR, unpublished). Using this procedure, chromiuminduced DNA-protein cross-links were detected predominantly in the nuclear matrix fraction immediately after treatment of cells with 150 pM Na2CrO4 (Table 1). Preferential cross-linking of nudear matrix proteins to DNA has also been reported by other workers (47,48). Consistent with the alkaline elution data presented in Figure  6A, the cross-links in the nuclear matrix reached a maximum level immediately after treatment with 150 pM Na2CrO4 but were essentially repaired 24 hr later ( Table 1).
Treatment of CHO cells with 150 or 300 pM Na2CrO4 suppressed both total RNA and mRNA synthesis for at least 32 hr ( Figure 5). In cells treated with the lower chromate concentration, DNAchromium adducts were the only damage which persisted over this entire period ( Figure 6B). In vitro, binding of chromium (III) has been found to interfere with the function of DNA templates, causing an increased number of initiation sites and the formation of truncated transcription products (49). The chromium adducts formed in CHO cells may therefore exert an inhibitory effect on transcription, possibly by interfering with the action of RNA polymerases.
Immediately after treatment with 150 pM chromate, induction of expression of the inducible gene GRP78 by tunicamycin was almost completely inhibited ( Figure 5). In contrast, by 24 hr after treatment a recovery in GRP78 induction occurred ( Figure 5). This recovery correlated with the repair of DNA-protein cross-links at this time ( Figure 6A; Table 1) but not with a general recovery of RNA synthesis ( Figure 5). Repair of single-strand breaks did not result in a recovery of GRP78 induction (13). These results are similar to those of Hamilton and Wetterhahn (47) who reported that dichromate inhibited induction of the inducible genes 5-aminolevinate synthase and cytochrome PB1 P450 in chick embryo liver in vivo and that this inhibition correlated with the presence of DNA-protein cross-links.
It has been suggested that transcription of inducible genes may be particularly sensitive to DNA-protein cross-link formation because these genes typically have complex regulatory regions and often require numerous DNA-protein interactions for proper gene expression (50). Potentially therefore, DNA-protein cross-links formed within the GRP78 promoter may inhibit the action of transacting factors required for gene induction. In fact, a number of transcription-factor binding sites, which are required for inducible gene expression, have been identified within the promoter region of the GRP78 gene (51)(52)(53)(54). DNAprotein cross-links occurred predominantly in the nuclear matrix fraction after treatment with 150 pM chromate (Table 1); and repair of these cross-links 24 hr after treatment correlated with recovery of GRP78 induction ( Figure 5; Table 1, respectively). It is therefore interesting that the nuclear matrix fraction was found to be enriched in GRP78 gene sequences compared with total nuclear DNA (Xu J, Manning FCR, Patierno SR, unpublished observations).
Only a partial recovery of GRP78 induction (56%) occurred when DNAprotein cross-links were repaired 24 hr after treatment with 150 pM chromate ( Figure 5). This suggests that chromium-DNA adducts, which were the only DNA damage still detectable at this time ( Figure  6B; Table 1), may contribute to the observed inhibition of GRP78 induction. Alternatively, this effect may be due to the large number of cells undergoing apoptosis beyond 24 hr (Figure 1). In cells treated with 300 pM chromate, no recovery of GRP78 induction occurred even 24 hr after treatment (Figure 5), probably because of the extreme toxicity of this dose.
The results presented in this study indicate that the toxicity of hexavalent chromium compounds can occur via apoptosis. Cells which do not undergo apoptosis may contain chromate-induced chromosome damage in genes associated with tumor formation such as oncogenes or tumor suppressor genes. Escape from chromate-induced apoptosis could therefore represent an early step in chromateinduced carcinogenesis.