INTRODUCTION

DNA-protein complexes have been shown to be induced by a number of physical and chemical carcinogenic agents such as -radiation and UV light (1, 2), alkylating agents (3), formaldehyde (4), vinyl acetate (5), and metal compounds of chromate (6, 7, 8), nickel (9), cis- or trans-platinum (10) and vanidium(V) (11). Hexavalent chromium (Cr(VI)) compounds have been considered as potent human carcinogens and have been shown to cause different types of DNA damage including DNA-protein cross-linking in various cells and tissues (see Ref. 12 for a review). Interestingly, Cr(VI) does not bind to DNA or proteins in cell-free systems (13, 14). However, Cr(VI), which exists as an oxyanion at physiological pH, is readily transported into the cell through the cells' sulfate anion transport system (15, 16). Inside the cell, Cr(VI) is believed to be reduced by the cells' redox system to its biologically most stable form, chromium(III) (17, 18). Cr(III) binds to DNA as well as proteins in cell-free systems (19) and has high affinity for many other biological ligands (20). Cr(III), however, is poorly taken up into the cell and is considered to be noncarcinogenic (21). During the intracellular reduction of Cr(VI) to Cr(III), reactive species such as intermediate valance states of chromium and active oxygen species are generated (17, 22, 23), which may, in turn, initiate the carcinogenic process by altering the structure of DNA (24). Hydroxyl radicals (^·OH), which are generated during the cellular reduction of chromate (25) are also capable of causing DNA-protein cross-linking (26, 27) and are considered as the "ultimate agents" in chromate carcinogenesis (28). Cr(III) and the reactive intermediate states of chromium may also be considered as carcinogenic because ^·OH radicals are shown to be generated by redox cycling of Cr(III) (29), and DNA damage has been shown to be caused by intermediate valence states of chromium, such as Cr(V) (30).

Although chromate-induced DNA-protein complexes are implicated in chromate carcinogenicity, the mechanisms of their formation, composition, and biological significance are not well understood. It has been postulated that cross-linking of proteins to DNA could disrupt chromatin structure and the normal regulation of gene expression (31). This, in turn, could play a role in carcinogenesis in that deletion of DNA bases may result when portions of replicating DNA are buried under DNA-protein complexes (32). Such deletions to "tumor suppressor genes" (33) may give rise to loss or inactivation of the gene, leading to carcinogenesis. Furthermore, during normal regulation of gene expression, proteins, either alone or in cooperation with other proteins, reversibly interact with specific DNA sequences (34). Cross-linking of DNA with inappropriate proteins could disrupt the normal regulation of DNA-protein interactions, causing serious genetic consequences, including disruption in or alteration of gene expression. Therefore, it is necessary to know the identity of the proteins that participate in chromate-induced DNA-protein complexes and the nature of their interaction with DNA. Identification of proteins cross-linked to DNA may also assist in our understanding of chromatin structure and protein interactions, including the three-dimensional orientation of proteins around DNA.

In the present study, we have analyzed the proteins complexed to DNA by chromate treatment of MOLT4 cells. We have previously explained the reasons for the choice of this cell line (23). Additionally, an increased yield of DNA-protein cross-links and a faster reaction kinetics have been reported in MOLT4 cells as compared with other cells (11). We have also attempted to identify some of the proteins that cross-link with DNA after chromate exposure, by isolation and partial N-terminal sequencing of these proteins, as well as using antibodies to "candidate proteins," because inappropriate complexing of proteins of structural and/or functional importance to DNA rather than the DNA-protein complexes themselves may have importance in chromate carcinogenicity. Identification of such proteins is essential for better understanding of the potential consequences of DNA-protein cross-links and the three-dimensional orientation of proteins around DNA. The composition and stability of chromate-induced DNA-protein complexes and the effect of antioxidants on the formation of such complexes have also been reported here.

MATERIALS AND METHODS

Highest purity grade potassium chromate (K₂CrO₄) was purchased from J. T. Baker (Phillipsburg, NJ). A protein determination kit and all gel electrophoresis reagents were purchased from Bio-Rad. Polyvinylidene difluoride (PVDF)¹ membranes, [³H]thymidine, [³⁵S]methionine, ¹²⁵I-protein A, [⁵¹Cr]potassium chromate, and Aquasure LSC mixture were purchased from DuPont NEN. DNase-free RNase and proteinase K were purchased from Boehringer Mannheim. All other chemicals and enzymes were purchased from Sigma.

Human leukemic T-lymphocyte MOLT4 cells (ATCC CRL 1582) were purchased from American Type Culture Collection (Bethesda, MD) and were maintained in suspension at exponential growth phase in RPMI 1640 (HEPES-modified) medium supplemented with 10% heat-inactivated fetal bovine serum, 10 units of penicillin, and 10 µg/ml streptomycin solution as described before (22). Cellular DNA and proteins were radiolabeled with [³H]thymidine and [³⁵S]methionine (0.02 µCi/ml each), respectively, for ~24 h, in methionine-free RPMI 1640 medium. Radiolabeled cells were collected by centrifugation, washed three times in cold Saline A (5 mM NaHCO₃, 6 mM dextrose, 5 mM KCl, and 140 mM NaCl, pH 7.2), and resuspended in salts/glucose medium (SGM; 50 mM HEPES, 100 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 5 mM dextrose, pH 7.2) at a concentration of 10⁶ cells/ml. Potassium chloride (as control) or potassium chromate was added to the cell suspensions from freshly prepared stock solutions. Following treatment, cells were collected, and cytotoxicity was determined by exclusion of trypan blue as described previously (22).

The method used to isolate DNA-protein complexes was modified from our previously described method (35). Potassium chromate-treated and control cells were collected by centrifugation at 500 × g for 10 min and were washed three times in phosphate-buffered saline. The cells were lysed in 30 ml of 10 mM Tris-HCl containing 2% SDS, 1 mM PMSF (pH 7.5) by shaking on a platform shaker for 6 h at room temperature. The cell lysates were transferred into a tight-fitting homogenizer and given ten strokes. The samples were sedimented at 100,000 × g for 16 h at 18 °C, using a Beckman SW 27 rotor (Beckman Instruments, Fullerton, CA). The pellets were placed in 28 ml of 5 M urea containing 1 mM PMSF and rocked on a platform shaker at 4 °C for 6 h. The samples were again homogenized as above, and then SDS was added to a 2% final concentration. The DNA-protein complexes were isolated by ultracentrifugation as above and were rinsed in 10 mM Tris-HCl (pH 7.5) containing 1 mM PMSF and 2% SDS. The final pellets were resuspended in 10 mM Tris-HCl containing 1 mM PMSF (pH 7.5) in siliconized Eppendorf tubes by gentle pipetting and overnight rocking on a Nutator shaker at 4 °C. The DNA-protein complexes were precipitated in 70% acetone at 20 °C. The DNA-protein complexes were collected by centrifugation at 12,500 × g for 15 min at 4 °C using a Beckman microfuge, rinsed in 70% acetone, and resuspended in 1 ml of 10 mM Tris-HCl containing 1 mM PMSF (pH 7.5) by gentle pipetting or by shaking on a Nutator shaker for about 16 h at 4 °C. The DNA content and the purity of the samples were determined by measuring the absorbance at 260 and 280 nm (36). Both ³H and ³⁵S activities were determined by dissolving the samples in Aquasure Mixture (DuPont NEN) and counting in a Beckman LS 5800 Liquid Scintillation counter (Beckman Instruments, Inc., Irvine, CA). The protein content of cell lysates was determined by using Bio-Rad dye and bovine -globulin as standards (37).

Cells were collected by centrifugation at 500 × g for 10 min, after which cells were washed three times in phosphate-buffered saline and incubated for 15 min on ice in a cold hypotonic buffer (10 mM Tris-HCl, pH 7.5, containing 10 mM NaCl, 1.5 mM MgCl₂). Cells were collected by centrifugation at 300 × g for 5 min, resuspended in the above solution supplemented with 0.5% Nonidet P-40 and 1 mM PMSF, and were given 8-10 strokes in a loose fitting glass homogenizer. The nuclei were sedimented at 700 × g for 5 min at 4 °C; resuspended in 10 mM Tris-HCl containing 250 mM nuclease-free sucrose, 3 mM MgCl₂, and 1 mM PMSF (pH 7.5); and layered over a similar solution but containing 880 mM sucrose. Nuclei were subsequently collected by centrifugation for 10 min at 1000 × g at 4 °C using a swinging bucket rotor in an ICE Centra-7R refrigerated centrifuge. The nuclei were observed by phase contrast microscopy, Giemsa staining, and Acridine orange staining, and were found to be free from cytoplasmic contaminations (data not shown). The purified nuclei were used for isolation of DNA-protein complexes using the method described above.

DNA-protein complexes were analyzed by nonequilibrium pH gradient electrophoresis as described by O'Farrell et al. (38) This procedure does not allow nucleic acids to enter into the first dimension focusing gel. DNA-protein complexes containing 150 µg of DNA were acetone-precipitated or lyophilized (FTS Systems, Inc., Stone Ridge, NY) and solubilized in 30 µl of solubilizing buffer (9 M urea, 4% Nonidet P-40, 2% -mercaptoethanol, and 3% ampholines (Bio-Rad), pH range 3-10). Isoelectric focusing and nonequilibrium focusing were carried out in 200-µl capillary tubes (1.5-mm diameter, Fisher) containing 4% polyacrylamide and 2% ampholines (pH range 3-10). Second dimensional separation was carried out on 10 or 12% SDS-polyacrylamide gels by following the method of Laemmli (39) except that 1% -mercaptoethanol was used. In some cases, the DNA-protein complexes were digested by DNase I or RNase A and concentrated by acetone precipitation or lyophilization before analysis by two-dimensional gels. The gels were subjected to polychromatic silver staining by following the method of Sammons et al. (40).

Subcellular localization of the proteins complexed to DNA upon chromate treatment was determined by analyzing the cytoplasmic, nuclear, and nuclear matrix proteins of MOLT4 cells by two-dimensional gel electrophoresis. The protein fractions containing the membrane and cytoplasmic fractions were prepared by sedimentation of nuclei after hypotonic lysis of cells in the presence of Nonidet P-40, as described above (Nonidet P-40-soluble cytoplasmic material). The nuclear protein fraction was prepared from the SDS-soluble material of the isolated nuclei, or by sonication of the purified nuclei. The nuclear matrix fractions were prepared by digestion of the purified nuclei with DNase I (400 Kunitz units/ml) and RNase A (20 Kunitz units/ml), followed by extraction with 1.6 M NaCl, as described previously (41). The protein contents of different fractions were determined by the dye binding method, as described above. Thirty µg of protein from each fraction was acetone-precipitated and solubilized in 30 µl of the solubilizing buffer. Equilibrium or nonequilibrium focusing, separation in the second dimension on 10% polyacrylamide gels, and silver staining of samples were carried out as described above.

DNA-protein complexes containing 250 µg of DNA were analyzed by two-dimensional gel electrophoresis and were electroblotted onto PVDF membrane as described previously (42). The second dimensional gel was prerun in the presence of 1 mM sodium thioglycolate to protect the proteins from N-terminal blocking. Proteins were electroblotted onto PVDF membranes in a Bio-Rad Transblot apparatus using 10 mM CAPS and 10% high pressure liquid chromatography grade methanol (pH 11) as the electroblotting buffer at 50 V for 1 h at room temperature. Coomassie Brilliant Blue R-250 (Bio-Rad; 0.025% in 40% methanol) was used to visualize the proteins. Acetic acid was omitted from the staining and destaining solution, since it may cause N-terminal blocking. The protein band of interest was excised, and automated Edman degradation was performed using an Applied Biosystems 477A protein sequencer equipped with a 120 A analyzer (Applied Biosystems, Inc., Foster City, CA).

Following one- or two-dimensional gel electrophoresis, the proteins were electrophoretically transferred to nitrocellulose membrane by following a modification of the method of Burnette (43). Proteins were transferred to nitrocellulose membrane in 25 mM Tris, 100 mM glycine, and 10% methanol overnight at 200 mA. The nitrocellulose membrane sheet was blocked in 20 mM Tris-HCl containing 0.05% Tween 20, 1% NaCl, 0.05% NaN₃, and 4% nonfat dry milk (pH 7.5) for 1 h at room temperature. Then the blot was incubated with a 1:1000 dilution of the antibodies in blocking buffer for 3 h with shaking. The blot was washed in 20 mM Tris-HCl containing 0.05% Tween 20, 1% NaCl, 0.05% NaN₃ (pH 7.5) and was reacted with 2 µCi of ¹²⁵I-protein A for 3 h at room temperature. The unbound antibodies were removed by washing the blot in 20 mM Tris-HCl containing 0.05% Tween 20, 1% NaCl, 0.05% NaN₃ (pH 7.5) for five times (5 min each). The blot was blot-dried, and the antigens were detected by autoradiography of immunoblots using XRP-1 film (Eastman Kodak Co.).

DNA-protein complexes containing 10 µg of DNA from control and chromate-treated samples were fractioned on agarose gels (0.6% (w/v) in 40 mM Tris, 40 mM borate, 1 mM EDTA) (36) for ~16 h at 50 V on a model H4 horizontal gel electrophoresis apparatus (Life Technologies, Inc.). The gel was stained with ethidium bromide (0.2 µg/ml in TBE), and DNA was visualized by ethidium bromide fluorescence at 254 nm.

The nontoxic levels of the test compounds were determined by monitoring the effect of 0-100 µM -tocopherol succinate, 0-5 mM sodium ascorbate, 0-5 mM Tiron and 0-100 mM mannitol, on the growth of MOLT4 cells up to 96 h. Cells were treated with nontoxic levels (>98% viable) of -tocopherol succinate (25 µM), sodium ascorbate (1 mM), Tiron (1 mM), or mannitol (10 mM) for 16 h in complete RPMI at 37 °C before exposure to chromate in SGM for 3 h.

DNA-protein cross-links were detected by modification of previously published methods (44, 45, 46). Labeling of cells with [³H]thymidine and treatment with chromate were carried out as described above. Following treatment, cells were washed three times in ice-cold phosphate-buffered saline and were frozen at 80 °C for 12-16 h. The cells were thawed and were lysed in 20 mM Tris-HCl containing 1 mM PMSF and 2% SDS (pH 7.5). The cell lysates were briefly sonicated on ice using a Heat System sonicator/cell disrupter (model W 225 R, Ultrasonics, Inc., Plainview, NY) by giving 10 pulses at 50% duty cycle. The samples were incubated at 65 °C for 10 min, and KCl (in 20 mM Tris-HCl, pH 7.5) was added to 100 mM final concentration. The samples were chilled on ice for 10 min, and the K⁺-SDS precipitates formed were collected by centrifugation at 3000 × g for 10 min at 4 °C. The pellets were resuspended in 20 mM Tris-HCl containing 100 mM KCl. The samples were incubated at 65 °C for 10 min, cooled on ice, and collected by centrifugation as above. This shearing and washing step was repeated two more times. The final pellets were resuspended in water and the protein bound [³H]DNA was estimated by counting the samples in a Beckman LS 5800 Liquid Scintillation counter (Beckman Instruments, Inc., Irvine, CA) using Aquasure LSC Mixture (DuPont NEN). Trichloroacetic acid-insoluble material from the cell lysates was used to estimate the total [³H]DNA. The ratio of the percentage of K⁺-SDS-precipitable DNA in the treated cells to that in the control cells was used to estimate the DNA-protein cross-link coefficient (DPC coefficient). Unlike the SDS-urea method, which requires 5-6 days, the K⁺-SDS method is more sensitive, and results can be obtained in 1 day. Therefore, the K⁺-SDS method was used to monitor the effects of antioxidants where isolation and characterization of proteins were not required.

The intracellular ascorbate level of cells was estimated by using Folin phenol reagent as described before (22). The vitamin E level of cells was determined spectrophotometrically by using the method of Fabianek (47).

Cellular Uptake of ⁵¹CrO₄²

Cells, in SGM, were incubated with 1 µCi of K₂⁵¹CrO₄ at various concentrations for 2 h at 37 °C. Cells were collected by centrifugation and washed three times with ice-cold phosphate-buffered saline, and the cell number was counted in a Coulter counter (model ZM, Coulter Electronics, Inc., Hialeah, FL). The cellular uptake of chromate was determined by measuring the ⁵¹Cr activity in a Beckman  counter (Beckman Gamma 5500 equipped with a DP 5500 Data Reduction System, Beckman Instruments) and comparing with a standard curve generated by using 0-20 nM potassium ⁵¹chromate.

DNA-protein complexes containing 100 µg of DNA in 10 mM Tris-HCl, 1 mM PMSF (pH 7.5) were taken in siliconized microcentrifuge tubes. MgCl₂ was added to a 5 mM final concentration in samples treated with DNase I and RNase. DNase I (200 µg/ml), RNase A (40 µg/ml), proteinase K (2 mg/ml), EDTA (10-50 mM), thiourea (100 mM), or -mercaptoethanol (2%) was added and mixed, and the tubes were incubated at room temperature for 3 h. Either PMSF was omitted from the samples treated with proteinase K, or the samples were incubated at 4 °C for 24 h before treatment with proteinase K to allow inactivation of PMSF. SDS was then added to a final concentration of 0.5% to inhibit nonspecific cross-linking, and samples were centrifuged at 100,000 × g for 16 h at 18 °C. The supernatants were carefully removed, and the pellets were resuspended in 10 mM Tris-HCl (pH 7.5) by brief sonication. DNA and protein contents were determined from the ³H and ³⁵S specific activity, respectively, by liquid scintillation counting as described above.

All experiments were performed at least three times. Paired, two-tailed Student's t test was performed, and p values 0.05 were considered significant.

RESULTS

Exposure of MOLT4 cells to 0-2 mM potassium chromate in SGM for 2 h was found to have little cytotoxic effect, as assessed by trypan blue exclusion (viability was within 98 ± 2% of the control). The viability of cells treated with 200 µM chromate was not affected within 4 h of treatment. The viability of cells treated with 200 µM chromate for 16 h in SGM was decreased to 72 ± 3% of the control.

Cell exposure to 0-2 mM potassium chromate for 2 h resulted in a dose-dependent increase in the formation of DNA-protein complexes in MOLT4 cells. Cells treated with 200 µM chromate for 2 h had about 175% more DNA-protein complexes as compared with the control (Fig. 1). Chromate-induced DNA-protein complex formation in MOLT4 cells was also found to be time-dependent. As shown in Fig. 2, the DNA-protein complex formation increased in a time-dependent manner in cells treated with 200 µM chromate for different time periods, and after 24 h a 10-12-fold increase in the formation of DNA-protein complexes was observed as compared with the control cells.

DNA-protein complexes isolated from both the potassium chloride (control) or potassium chromate-treated cells were analyzed by nonequilibrium two-dimensional gel electrophoresis. DNA-protein complexes were loaded on the acidic end of the gel in order to avoid the entry of nucleic acids into the first dimensional focusing gels. Silver staining of two-dimensional gels of DNA-protein complexes isolated from control cells did not show any protein in the gel, indicating that the SDS/urea method used for isolation of DNA-protein complexes effectively dissociates the background DNA-protein complexes in the control cells (Fig. 3A).

Nonequilibrium two-dimensional gel electrophoresis of DNA-protein complexes isolated from whole cells or nuclei of control (potassium chloride) or chromate-treated MOLT4 cells. Chromate treatment of cells and isolation of DNA-protein complexes by the SDS/urea method was followed as described under "Materials and Methods." Each gel was loaded with DNA-protein complexes containing 150 µg of DNA. A, two-dimensional gel of DNA-protein complexes isolated from the nuclei of control cells. No proteins were detected on the gel of control DNA-protein cross-links, indicating that the SDS/urea extraction method dissociated most of the background DNA-protein complexes. B, two-dimensional gel of proteins dissociated from DNA-protein complexes (obtained from purified nuclei) generated by treatment of cells with 25 µM chromate for 16 h. The proteins a, b, c, and d were primarily cross-linked to DNA upon 25 µM chromate treatment of cells. C, same as B, except that cells were treated with 200 µM chromate for 16 h. The letters m, n, o, and p refer to proteins that were prevalent in the cytoplasmic protein fraction but cross-linked to DNA after 200 µM chromate exposure of intact cells. D, two-dimensional gel of proteins dissociated from DNA-protein complexes isolated from the nuclei of cells treated with 200 µM chromate for 4 h. As seen in this figure, identical proteins were cross-linked to DNA after 200 µM chromate treatment of cells for either 4 or 16 h. E, two-dimensional gel of proteins dissociated from DNA-protein complexes isolated from intact cells treated with 200 µM chromate for 16 h. As shown in this figure, identical proteins were cross-linked to DNA when DNA-protein cross-links were isolated from either the intact cells or the nuclei of cells exposed to chromate. The electrofocusing gel in C contained biolytes of pH range 3-10 and 8-10 in a ratio of 4:1. In all other gels biolytes of pH range 3-10 was used.

The proteins that were complexed to DNA, in cells exposed to 25 µM chromate for 16 h, and that were resistant to SDS/urea extraction are shown in Fig. 3B. Three acidic proteins (a, b, and c) and a basic protein, d, were primarily complexed to DNA upon chromate treatment. Analysis of the molecular weight and pI of these proteins showed that the protein a has a pI of 5.2-5.6 and a molecular mass of 74 kDa, the protein b has a pI of 5.2-5.4 and a molecular mass of 44 kDa, and the protein c has a pI of ~5.8 and molecular mass of 42 kDa, respectively. The protein d, on the other hand, is found to be a basic protein with a pI of 8.8-9.2 and a molecular mass of 51 kDa. The number of proteins cross-linked to DNA upon chromate exposure was found to be dependent on the dose of chromate, because cells treated with 200 µM chromate for 16 h had many other proteins, in addition to the above proteins, complexed to DNA (Fig. 3C). Since 25% of the cells were found to be killed by such treatments (200 µM chromate for 16 h), there was a reason to believe that the additional proteins cross-linked to DNA could be due to dead cells. That this was not the case was demonstrated by treating cells with 200 µM chromate for 4 h. This latter treatment did not affect cell viability but cross-linked the same proteins to DNA (Fig. 3D).

There is of course, a possibility that the cytoplasmic proteins might associate with DNA during the cell lysis. In order to lessen the likelihood of this subtle artifact, we analyzed DNA-protein complexes isolated from either whole cells or purified intact nuclei of cells treated with chromate. If cytoplasmic proteins become associated with DNA during cell lysis, additional proteins should appear in two-dimensional gels of whole cells as compared with that obtained from the nuclear fractions. That this was not the case is illustrated by the fact that identical proteins were found to be complexed to DNA when either whole cells or nuclei of cells treated with chromate were used as the starting material (Fig. 3, D and E). Hence, it appears likely that chromate induces the cross-linking of nuclear proteins to DNA.

Molecular mass and pI of the major proteins complexed to DNA upon 200 µM chromate treatment are listed in Table I. These proteins were not seen in two-dimensional gels of DNA-protein complexes isolated from untreated control cells, even when the initial DNA load was increased to visualize the background proteins. When the control material was digested with DNase I before analysis by two-dimensional gel, only trace amounts of some of the above proteins were observed (data not shown), but some of these proteins were also present as contaminants in DNase I (data not shown). Therefore, we assume that the traces of proteins observed in control cells are not from DNA-protein complexes but from the contaminant proteins present in DNase I. 

Table I. Molecular mass and pI of the major proteins complexed to DNA upon chromate treatment of intact MOLT4 cells DNA-protein cross-links were isolated from MOLT4 cells treated with 200 µM potassium chromate for 16 h as described under "Materials and Methods." The DNA-protein cross-links containing 150-200 µg of DNA were analyzed in two-dimensional gels, and the molecular mass and the isoelectric point (pI) of major proteins resolved on the gel were determined. Data presented here are from results of this study and our previous study (Ref. 35). Molecular mass pI kDa 108 5.4 -5.8 98 5.2 -5.6 74 (a)a 5.2 -5.6 63 (m)a 5.2 -5.4 53 (n)a 5.2 51 (d)a 8.8 -9.2 49 (o)a 5.4 -5.8 44 (b)a 5.3 43 (p)a 6.0 -6.5 42 (c)a 5.8 40 4.8 -5.0 36 5.0 -5.2 36-38 (CNP)b 5.5 -7.2 29 6.8 25-28 (CNP) 7.0 -8.5 19 6.4 -6.8 16 (CNP) 5.6 -6.8 a  Proteins marked in Fig. 3C. The letters a, b, c, and d correspond to the proteins that cross-linked to DNA upon 25 µM chromate treatment of MOLT4 cells for 16 h. b  CNP, cluster of nuclear proteins.

To determine the subcellular localization of the four proteins complexed to DNA, cytoplasmic, nuclear, and nuclear matrix protein fractions were analyzed by two-dimensional gel electrophoresis. The purified nuclei were free from cytoplasmic contaminations (not shown). Proteins of similar molecular weight, isoelectric point, and coloration after polychromatic silver staining were assumed to be the same protein. Proteins b, and c were visualized and were found to correspond to proteins in the cytoplasmic fraction (Fig. 4A). Proteins a, and d were predominantly present in the nuclear fraction (Fig. 4B). Additional proteins found complexed to DNA (m, n, o, and p), upon treatment of cells with 200 µM chromate, were also present in the nuclear fraction, although they were predominantly present in the cytoplasmic fraction. Fig. 4C shows the two-dimensional resolution of nuclear matrix proteins of MOLT4 cells. As shown in this figure, all of the proteins cross-linked to DNA by 25 µM chromate treatment and a 63-kDa acidic protein (m) cross-linked to DNA by higher doses of chromate were found in this fraction. These results suggest that nuclear matrix proteins are the target for chromate-induced DNA-protein cross-linking.

Localization of major proteins cross-linking to DNA upon chromate treatment of cells, in the cytoplasmic, nuclear, and nuclear matrix protein fractions. Proteins from cytoplasmic fraction (A), nuclear fraction (B), and nuclear matrix fraction (C) containing 30 µg of protein were analyzed by nonequilibrium two-dimensional gel electrophoresis and were stained with silver stain. The proteins a, b, c, and d refer to the proteins that primarily cross-linked to DNA with 25 µM chromate treatment of intact cells for 16 h.

DNA-protein complexes were digested by DNase I or RNase A prior to analysis by two-dimensional gels to determine if nuclease digestion would dissociate any other protein from the complex that is not resolved under the gel electrophoretic conditions. The proteins resolved in two-dimensional gels without nuclease digestion of chromate-induced DNA-protein complexes are shown in Fig. 5A. Fig. 5, B and C, show the resolution of DNA-associated proteins following treatment with DNase I and RNase A, respectively. There was no significant difference in the resolution pattern of proteins dissociated from chromate-induced DNA-protein complexes with or without nuclease digestion. The difference in protein resolution pattern in DNase I-treated sample (Fig. 5B) was due to the presence of proteins in DNase I (labeled as D). Similarly, RNase A proteins are labeled as R in Fig. 5C. These results suggest that nuclease digestion is not required for the resolution of chromate-induced DNA-protein complexes in two-dimensional gels. The resistance of chromate-induced DNA-protein complexes to treatments such as 2% SDS and 5 M urea, but their resolution in two-dimensional gels indicates that these complexes are disruptable by the electrofocusing buffer, which contained 2% -mercaptoethanol and 9 M urea.

Effect of nuclease digestion on the resolution pattern of chromate-induced DNA-protein complexes. DNA-protein complexes (equivalent to 2-3 A₂₆₀ units) isolated from cells treated with 200 µM chromate for 16 h were incubated in the presence or absence of nucleases for 2 h at room temperature, concentrated, and analyzed by two-dimensional gels. A, two-dimensional gel of undigested DNA-protein complexes. B and C, two-dimensional gels of DNase I- (300 µg/ml) and RNase A- (200 µg/ml) digested DNA-protein complexes, respectively. The proteins a, b, c, and d refer to the proteins that primarily cross-linked to DNA with 25 µM chromate treatment of intact cells for 16 h. The gel in A was electrofocused for 1800 V-h. The gels in B and C were electrofocused for 2300 V-h. The proteins labeled as D and R are the proteins in DNase I and RNase A, respectively.

Apart from the selection of candidate proteins based on similar molecular weights and isoelectric points, we have followed two different approaches to further characterize the proteins cross-linked to DNA by chromate. In one approach, proteins were partially sequenced from their N-terminal ends, and the amino acid sequence obtained was used to search for its homology in different protein data banks. So far we have not been successful in identifying the proteins a, c, and d (Fig. 3C) by following the above approach. The few amino acid sequences obtained by N-terminal sequencing of these proteins have not been found homologous to any proteins in the existing protein data banks (data not shown). However, using this method, a 43-kDa protein (labeled as p in Fig. 3C, p43, pI 6.0-6.5), which was predominantly detected in the cytoplasmic fraction but abundantly cross-linked to DNA, was identified as lectin. Thus, the N-terminal sequencing of p43 revealed six consecutive amino acids that have absolute homology with amino acid residues 24-29 of lectin Bra-3 (Fig. 6A). This sequence is also partially homologous to many glycoproteins and the human multidrug resistance protein 1. Based on the partial amino acid sequencing and the homology of the sequence, another 53-kDa protein (labeled as n in Fig. 3C) appears to be aminoglycoside nucleotidyltransferase. The seven amino acids sequenced from the N-terminal end (Fig. 6B) were found to have absolute homology with the amino acids 23-29 of aminoglycoside nucleotidyltransferase. The other approach used to identify the proteins was immunoblotting, using commercially available antibody to candidate proteins. Using an anti-actin polyclonal antibody (ICN Biochemicals, Inc.), the acidic 44-kDa protein (labeled as b in Fig. 3C) has been identified as -actin (Fig. 7). This protein is one of the most prevalent proteins complexed to DNA upon exposure of cells to chromate. The homology of the eight amino acids obtained by N-terminal sequencing of this protein with amino acids 19-26 of -actin (Fig. 6C) and its molecular weight and pI similar to that of -actin suggest that this protein is nuclear -actin.

During the biological reduction of Cr(VI), reactive species, such as chromium(V) and active oxygen species, are generated. Although oxygen radicals have been shown to cause DNA-protein cross-linking (26, 27) and several investigators have suggested ^·OH radicals as the "ultimate carcinogen" in the chromate-induced carcinogenic process (28), the role of oxygen radical species in chromate-induced DNA-protein cross-linking has not been reported. Therefore, the effect of antioxidants and free radical scavengers on chromate-induced DNA-protein cross-linking was investigated. Because differential uptake of chromate by cells would lead to alterations in chromate-induced DNA-protein cross-linking, the effect of antioxidants on cellular uptake of chromate was first investigated. As shown in Table II, the cellular uptake of chromate increased in a dose-dependent manner when MOLT4 cells were exposed to 5-15 nM of potassium ⁵¹chromate. Pretreatment of cells with -tocopherol succinate (25 µM), Tiron (1 mM), mannitol (10 mM), or ascorbate (1 mM) had no significant effect on the cellular uptake of Cr(VI). As shown in Fig. 8, pretreatment of cells with 25 µM -tocopherol succinate, an antioxidant, inhibited the chromate-induced DNA-protein cross-linking by about 50%. Pretreatment of cells with -tocopherol or -tocopherol acetate also had similar effects (data not shown). Cells treated with 25 µM -tocopherol succinate for 16 h in complete RPMI medium and exhaustively washed had an approximately 4-fold increase in cellular tocopherol level over the untreated controls (data not shown). This is consistent with the findings of Sugiyama et al. (48), who have shown a 10-fold increase in cellular -tocopherol level after a 24-h treatment with 25 µM vitamin E in Chinese hamster V 79 cells. Pretreatment of cells with 1 mM Tiron (a vitamin E analog) or 10 mM mannitol (^·OH scavenger) inhibited the chromate-induced DNA-protein cross-linking by 45 and 20% of control, respectively. Pretreatment of cells with 1 mM ascorbate, on the other hand, increased the chromate-induced DNA-protein cross-linking by about 150% of the control (Fig. 8). Cells treated with such levels of ascorbate increased the intracellular level of this vitamin by about 2-fold (data not shown). Since ascorbate had little effect on cellular uptake of chromate and pretreatment of control cells with ascorbate had trivial effects on the background level of DNA-protein complexes, the observed augmentation of chromate-induced DNA-protein complexing following this antioxidant treatment may, most probably, be due to the direct reduction of Cr(VI) by ascorbate, giving rise to an increased level of intracellular Cr(III).

Table II. Effect of antioxidants on the cellular uptake of 51CrO42 Cells, in logarithmic growth phase, were treated with different concentrations of K251CrO4 in SGM for 3 h at 37 °C. Cells were washed three times in ice-cold phosphate-buffered saline, and cellular 51CrO42 uptake was determined as detailed under "Materials and Methods." Each value is a mean of at least three different experiments ± S.D. Values not significantly different from control have p  0.05 (n = 3-5). 51CrO42 5 nM K251CrO4 10 nM K251CrO4 15 nM K251CrO4 pmol/106 cells Control 1.97  ± 0.94 5.18  ± 0.81 6.83  ± 0.73  -Tocopherol succinate (25 µM) 2.26  ± 0.26 4.86  ± 0.84 7.31  ± 0.89 Tiron (1 mM) 2.15  ± 0.65 5.37  ± 0.79 7.17  ± 0.82 Ascorbate (1 mM) 1.81  ± 0.50 5.29  ± 0.92 6.97  ± 0.67 Mannitol (10 mM) 2.07  ± 0.63 4.90  ± 0.68 7.07  ± 0.88

The stability of DNA-protein complexes was tested by monitoring the recovery of DNA and protein in the pellet following treatment of DNase I, RNase A, proteinase K, EDTA, -mercaptoethanol, or thiourea. As determined by agarose gel electrophoresis, the average size of DNA was approximately 7500 base pairs (not shown). The control samples (without any treatment) had almost 100% recovery of both DNA and protein in the pellet following ultracentrifugation, as determined by ³H and ³⁵S radioactivity, respectively. Treatment of DNA-protein complexes, isolated from both control and chromate-treated cells, with DNase I significantly reduced the recovery of ³H and ³⁵S in the pellet (Fig. 9). RNase A treatment of DNA-protein complexes did not interfere with recovery of DNA or protein. Proteinase K treatment dissociated most of the proteins from the DNA-protein complexes without affecting the recovery of DNA. These data indicate that chromate treatment induces the cross-linking of proteins to DNA and does not cause sedimentable protein aggregates and are consistent with a previous report (8) for chromate-induced DNA-protein complexes in cultured Chinese hamster ovary cells.

In order to test whether chromium is directly participating in the DNA-protein complexes, EDTA in excess was used as a chelating agent to examine whether chelation of chromium would disrupt the complex. As shown in Fig. 9, EDTA (50 mM) treatment of DNA-protein complexes (isolated from cells labeled with [³⁵S]methionine and [³H]thymidine) decreased the recovery of ³⁵S radioactivity without affecting the recovery of ³H activity. These results indicate that the decrease in ³⁵S activity was not due to fragmentation of DNA. Dissociation of ³⁵S or ⁵¹Cr increased with EDTA concentration up to 50 mM, and no further dissociation was observed beyond this concentration (data not shown). The maximum decrease in ³⁵S recovery after EDTA (50 mM) treatment was found to be approximately 18% of the control (Fig. 9). Since Cr(III) has high affinity for EDTA, the dissociation of some of the complexes by EDTA may, in part, be due to a chelatable form of chromium, such as Cr(III). Treatment of the complexes with -mercaptoethanol (2%) or thiourea (100 mM) decreased the recovery of proteins to about 50 and 55% of control, respectively, without affecting the recovery of DNA (Fig. 9), indicating the participation of sulfhydryl groups in chromate-induced DNA-protein complexes.

DISCUSSION

Although previous studies have shown that carcinogenic Cr(VI) induces DNA-protein cross-links, little is known about the characteristics of the chromate-induced DNA-protein cross-links. It has generally been believed that the reduced form of the carcinogenic Cr(VI), Cr(III), gives rise to DNA-protein cross-links by directly mediating the cross-linking between the cellular DNA and protein (49). This belief is mainly based on the fact that chromate-induced DNA-protein cross-links could be disrupted by EDTA, a chelator of Cr(III) but not of Cr(VI), and that the proteins in the DNA-protein complexes could be visualized on SDS-polyacrylamide gels without nuclease digestion (50). In these studies, chromate-induced DNA-protein complexes were only partially disrupted by EDTA. Again, the nature of the cross-link would not be explained by resolution of a protein on SDS-polyacrylamide gel electrophoresis if the same protein is cross-linked to DNA by different mechanisms, because a single protein could possess many different reactive groups to react with different cross-linking agents. Most other in vitro studies supporting Cr(III) as the cross-linking agent are based on the fact that reducing agents are required in the reaction mixtures for Cr(VI)-induced cross-linking of DNA and protein to take place (13, 51), a condition that generates reactive species capable of causing DNA-protein cross-linking (52). Therefore, the nature of chromate-induced DNA-protein cross-links is not fully resolved. We have previously reported that chromate induces an oxidative stress in the cells (22, 23) and that pretreatment of cells with vitamin E, an antioxidant, inhibits chromate-induced DNA-protein cross-linking (53). The results of the present study, for the first time, indicate that chromate-induced DNA-protein complexes may be formed by the generation of active oxygen species during the intracellular reduction of chromate.

In the present study, DNA-protein cross-linking increased in a dose- and time-dependent manner when MOLT4 cells were treated with chromate (Figs. 1 and 2) and did not attain a plateau under the present experimental conditions. The increase in the formation of DNA-protein complexes was not due to cell death, because short term exposure to chromate, which did not affect cell mortality, substantially increased DNA-protein cross-linking, and cross-linked similar proteins to DNA. A specific group of nuclear non-histone proteins seems to participate in chromate-induced DNA-protein complexes. These proteins must reside in close vicinity to DNA in relatively high concentrations, because a protein must reside in close proximity to DNA, and its reactive groups should be oriented such that they are able to interact with that of DNA in order for it to be cross-linked to DNA by any form of cross-linking agent. Present studies show that only four proteins (a, b, c, and d) were found to be primarily complexed to DNA, although several other proteins were seen in the nuclear protein fraction (Fig. 3B). Since chromate was required for the cross-linking of these proteins with DNA, it is apparent that their selective interaction with chromium was necessary for the cross-linking of these proteins to DNA. Other investigators have reported the association of a 45-kDa protein (similar in molecular weight and pI to protein b) to DNA by chromium (6, 50, 55) platinum (8), and ionizing radiation (56). The identity of proteins a, c, and d remains to be determined. Although histones constitute a substantial part of the chromatin, these basic proteins were not complexed to DNA upon chromate treatment. This is consistent with the findings of Miller et al. (50). Because Cr(III) has high affinity for sulfur-containing ligands and there is scarcity of cysteine residues among histones, it appears plausible that histones may not complex to DNA by chromate due to unavailability of appropriate ligands.

We have tentatively identified three different proteins that cross-link with DNA when cells are exposed to chromate. Their identity and the following plausible pathophysiological consequences are considered. (i) The 44-kDa acidic protein (protein b) could be nuclear -actin based on its identical molecular weight, isoelectric point, and homology with amino acids 19-26 of -actin. Furthermore, it positively reacted with an anti-actin polyclonal antibody. This is in accord with the finding that actin does cross-link with DNA in chromate-treated CHO cells (57). Actin is present in nucleus, nucleolus as well as the nuclear matrix (58). Actin has been shown to be associated with DNA replication, DNA repair, and RNA transcription (59, 60, 61). Hence, chromate-induced actin-DNA cross-linking may, at least in part, lead to altered gene expression, as has recently been shown for inducible genes (62). (ii) The homology of p43 (protein p) microsequence with amino acids 24-29 of lectin Bra-3 (Fig. 6) indicates that it could be a human lectin. Lectins have not been previously shown as DNA-binding proteins. Although lectin receptors have been found on the cytoplasmic surface of intracellular membranes such as the nuclear envelope and mitochondrial outer membrane, recent evidence indicates that lectin binding takes place on the noncytoplasmic surface of these organelles (63). Lectins are located in a wide variety of cells and cell membranes, and alteration in their levels has been reported upon malignant transformation (64). Lectins are shown to play important roles in the developmental processes (65). Lectins have also been shown to function as receptors (65) and mitogenic regulators (66). (iii) The seven N-terminal amino acids of the 49-kDa protein (labeled as n in Fig. 3C) were found to have absolute homology with aminoglycoside nucleotidyltransferase, a protein previously not known to bind to DNA. Because chromate-induced DNA-protein complexes predominantly occur in transcriptionally active DNA (62), it remains to be seen if these proteins are involved in the transcription process. Nonetheless, the cross-linking of these proteins to DNA could lead to serious physiological and genetic consequences.

The nature of chromate-induced DNA-protein complexes was analyzed by enzyme and chemical treatment of the complexes. Treatment of DNA-protein complexes isolated from control or chromate-treated cell nuclei with DNase I dissociated most of the proteins associated with DNA (Fig. 9), indicating that the sedimentable nature of the proteins is due to the association of proteins with the genomic DNA and not due to protein aggregation following chromate treatment or altered solubility of the metal-bound proteins. The small amount of DNA-protein complexes sedimented after DNase I digestion appears to be mostly in the form of stable chromium-nucleoprotein complexes. This is consistent with the findings of other investigators who have demonstrated the resistance of chromium-bound nucleoli to nuclease digestion (67) and have shown the cross-linking of nuclear matrix proteins to DNA by heavy metals and UV irradiation (55, 68). The resistance of the DNA-protein complexes to RNase A digestion indicates that chromate treatment does not induce the formation of RNA-protein complexes. The stability of the DNA-protein complexes was further assessed by monitoring the resistance of the complexes to EDTA treatment. Treatment with EDTA caused dissociation of only 18% of ³⁵S activity from the DNA-protein complexes. Because EDTA effectively chelates Cr(III) but poorly binds with the oxyanion of Cr(VI), EDTA-dissociable proteins from DNA-protein complexes could have been mediated by a chelatable form of chromium such as Cr(III). However, the majority of the chromate-induced DNA-protein complexes were resistant to EDTA treatment. These data suggest that the predominant form of DNA-protein cross-links caused by chromate may not involve the metal. Such cross-links could be generated by direct interaction between DNA and protein via formation of protein and/or DNA radicals produced during intracellular reduction of chromate. It is also possible that some of the Cr(III) in the cross-link is not accessible to EDTA chelation. The dissociation of some of the chromate-induced DNA-protein complexes by -mercaptoethanol or thiourea suggests that some of the cross-links involve sulfhydryl groups. Although Cr(III) is known to form complexes with sulfur-containing ligands, the -SH groups could directly be involved in the complex via thiyl radicals or disulfides that are generated during chromate-induced oxidative stress. The later contention is further supported by the fact that pretreatment of cells with antioxidants or free radical scavengers such as -tocopherol succinate, Tiron, and mannitol did inhibit the chromate-induced DNA-protein cross-linking. The observed increase in chromate-induced DNA-protein cross-linking following ascorbate treatment may, in part, be due to an increase in the reduction of Cr(VI), leading to the increased accumulation of intracellular Cr(III), which may, in turn, give rise to DNA-protein complexes. Intracellular Cr(III) is predominately generated by the reduction of Cr(VI), a process shown to generate oxygen free radicals (52, 69). Free radical generating systems such as ionizing radiation as well as Fenton type reactions have been shown to cause DNA-protein cross-linking (56, 69). Collectively, these results suggest that free radicals may, at least in part, be involved in chromate-induced DNA-protein cross-linking.

Free radical independent mechanisms may also play a role in some chromate-induced DNA-protein cross-linking, because the electrophoretic conditions would not disrupt the radical-induced covalent DNA-protein cross-links, and we were able to visualize the proteins cross-linked to DNA in two-dimensional gels without digesting DNA. Furthermore, nuclease digestion did not cause the appearance of additional proteins on two-dimensional gels. Visualization of proteins in two-dimensional gels without nuclease digestion of the complexes may, at least in part, be due to the reduction of sulfhydryl groups of proteins by -mercaptoethanol and the presence of a high concentration of urea in the sample buffer, leading to disruption of the complex. Such mechanisms have been shown to be the leading cause for dissociation of chromate- and cisplatin-induced DNA-protein complexes (55). Another possibility is that the same proteins are complexed to DNA via both the Cr(III) and oxidative mechanisms due to interaction of different amino acid residues with DNA. In that case, two-dimensional gel patterns of DNA-protein complexes may be the same with or without nuclease digestion. That this was the case was shown by digestion of the complexes with DNase I and RNase A (Fig. 6).

In view of the difficulties and limitations of the methods, it is not possible to propose a definite mechanism(s) of action of chromate in inducing DNA-protein complexes at this time. However, it is likely that chromate-induced DNA-protein complexes are formed via more than one mechanism and that at least some of the complexes are noncovalent in nature. Therefore, use of the term "cross-link" to indicate the association of proteins with DNA may not be appropriate, although we have used this term to express the association of DNA and protein, as has been used in the literature.

The results presented in this study indicate that chromate treatment of cells complexes a selected group of non-histone proteins to DNA. Actin, lectin, and aminoglycoside nucleotidyltransferase are among other proteins that appear to participate in chromate-induced DNA-protein complexes. The exact nature of the interaction between the DNA and protein remains to be determined. However, our results suggest both the participation of a chelatable form of chromium such as Cr(III) and the involvement of oxidative mechanisms in the process of chromate-induced DNA-protein cross-linking. Our results also suggest the involvement of sulfhydryl groups in chromate-induced DNA-protein cross-links. Although chromate-induced DNA-protein complexes are found to be resistant to treatments such as 2% SDS and 5 M urea, their reversibility in the gel electrophoretic conditions indicates that their association is in the form of noncovalent interactions. These characteristics of chromate-induced DNA-protein complexes suggest that it is possible to use chromium in studies involving chromatin structure as well as identification of proteins participating in DNA-protein interactions, specifically those that undergo transient interaction with DNA, such as transcription factors.