Chromium-induced Cross-linking of Nuclear Proteins and DNA

The in vivo cross-linking of proteins to DNA in intact Novikoff ascites hepatoma cells exposed to the chromium salt K2Cr04 was studied. DNA-protein complexes were assayed by high speed centrifugation of cells solubilized in buffered 4% sodium dodecyl sulfate and by electrophoretic identification of proteins associated with DNA-containing pellets. Further evidence of DNA-protein complexes, not dissociable in this buffer, was obtained by CsCl gradient centrifugation. Time dependence experiments showed that detectable cross-linking occurred after cells were exposed to chromium salt for at least 4 h, and the amount of DNAprotein complexes increased with longer incubation times. Complex formation occurred only with chromium salt concentrations of 200 pM or greater, and maximal cross-linking was effected at 5 mM. Immunotransfer methodology employing antibodies to nuclear matrix fraction and lamins was used to identify some of the polypeptides comprising the cross-linked complexes. These studies indicated specificity of chromium-induced complex formation within the nuclear protein fractions assayed. Our results document the ability of chromate to produce specific DNA-protein cross-links in living cells.

The cells were harvested under sterile conditions and used in the cross-linking experiments.
Cross-linking Experiments-NAH cells were resuspended in 0.17 M NHX1 at 4 "C to lyse the contaminating erythrocytes. After brief incubation, the cells were washed twice with phosphate-buffered saline (0.15 M NaCl, 10 mM sodium phosphate, pH 7.2), resuspended in 20 volumes of Hank's balanced salt solution containing different concentrat,ions of K2Cr04, and incubated at 37 "C for the indicated time intervals. During incubation, the cells were gently shaken every 20 min. The viability of NAH cells was determined at each collection time by trypan blue exclusion assay. Next the cells were washed twice in phosphate-buffered saline a t 4 "C and solubilized in 4% sodium dodecyl sulfate, 50 mM Tris-HC1, pH 7.5 (4% SDS buffer). From this step on, all the solutions also contained 1 mM phenylmethylsulfonyl fluoride (PMSF). The solution was stirred slowly at room temperature for 4 h and then very gently homogenized in a glass homogenizer with loosely fitting Teflon pestle (11). After centrifugation at 100,000 X g for 16 h, the resulting DNA pellets were rinsed with 4% SDS buffer (for 30 s) and t.hen resuspended in 5 M urea (volumes equal to the original 4% SDS solution). After stirring a t 4°C for 3 h, the resulting solution was homogenized, SDS was added to a final concentration of 4%, and the solution was stirred slowly a t room temperature for 1 h. After centrifugation a t 100,000 X g for 16 h, the resulting pellets were rinsed with urea/SDS buffer (4% SDS, 50 mM Tris-HC1, pH 7.5,5 M urea) (or urea, 4% Sarkosyl buffer for CsCl gradient centrifugation) and then resuspended in 2 mM Tris-HCl buffer, pH 7.5. The suspension was sonicated (three 20-s bursts followed by intermittent cooling in ice), precipitated with ice-cold acetone, and resuspended in 2 mM Tris-HC1, 1 mM MgC12 buffer, pH 7.5. Next, DNase I was added (25 pg/ml, Worthington, 1,872 units/mg) and the samples were incubated at 37 "C for 1 h, in preparation for SDS-polyacrylamide gel electrophoresis.
Polyacrylamide Gel Electrophoresis-The nuclease-digested samples were made (final concentration) 2% SDS, 10% glycerol, 5% 2mercaptoethanol, 0.0625 M Tris-HC1, pH 6.8, boiled for 5 min, and electrophoresed as described (12) using 3.0% stacking gel and 7.5% running gel. The separated proteins were visualized either by using the silver staining technique (13) or by transferring the proteins to nitrocellulose sheets as described (14). To visualize the protein antigens, the nitrocellulose sheets were incubated with appropriate antiserum and stained by the peroxidase-antiperoxidase procedure described by Sternberger (15). To quantitate the immunoreactive staining, the densitometry method (16) was employed.
Antigens and Antisera-To identify a t least some of the proteins co-sedimenting with the DNA after incubation of NAH cells with K,Cr04, antisera to nuclear matrix and nuclear envelope proteins were elicited in rabbits.
To isolate nuclear matrix, washed NAH cells were exposed to hypotonic shock in 10 mM Tris-HC1 buffer, pH 7.4, for 10 min and homogenized (4-5 strokes in a glass homogenizer with tightly fitting Teflon pestle), and the homogenate was made 10 mM in respect to MgC1, and centrifuged at 770 X g for 10 min. The pellets were resuspended in 2.2 M sucrose, 10 mM MgC12, homogenized as above, layered over 2.2 M sucrose, and centrifuged at 100,000 X g for 1 h. The pelleted nuclear material was resuspended by homogenization in 0.25 M sucrose, 50 mM Tris-HC1, 5 mM MgCl,, 1 mM PMSF, pH 7.4 (STM buffer), containing 0.04% Nonidet P-40 and centrifuged a t 1,000 X g for 10 min. The pellets were homogenized in STM buffer without the detergent, the DNA concentration was adjusted to 1.0-1.5 mg/ml, and 50 pg/ml of DNase I (1,872 units/mg) and 25 pg/ml RNase A (75 units/mg) were added. The mixture was incubated on ice for 1 h with occasional stirring, centrifuged a t 770 X g for 10 min, resuspended in 2 M NaCl, 10 mM Tris-HCI, p H 7.4, and incubated on ice for 15 min. Finally, the suspension was again centrifuged as above, and the pellet was used as immunogen (17) or dissolved in SDS buffer for polyacrylamide gel electrophoresis.
The nuclear pore-lamina complex was isolated exactly as described (18). This preparation was used to obtain antisera to nuclear lamins. The nuclear pore-lamina complex was electrophoresed in the presence of SDS, and the area containing lamins was excised, and, after homogenization with complete Freund adjuvant, used for immunization of adult male New Zealand rabbits. Details of this procedure and the characterization of antiserum which recognized lamins A and C are described elsewhere (19).
CsCl Equilibrium Density Gradient Centrifugation-The NAH cells were incubated with K2Cr04, solubilized in SDS buffer, and centrifuged as described. The resulting DNA pellets were resuspended in 5 M urea, 4% Sarkosyl solution, stirred slowly at room temperature for 4 h, and then gently homogenized in a glass homogenizer with loosely fitting Teflon pestle. The resulting suspension was centrifuged at 100,000 X g for 16 h. The final pellets were washed with 2 mM Tris-HCI, 1 mM PMSF, pH 7.5, resuspended by gentle homogenization in CsCl solution (1.7 g/ml), divided among six centrifuge tubes, and centrifuged a t 100,000 X g for 78 h. 0.5-ml fractions were collected from each gradient, and their density and absorbance a t 260 nm were determined. Corresponding fractions from each gradient were combined, dialyzed against 2 mM Tris-HC1, 0.1 mM PMSF, pH 7.5, lyophilized, and electrophoresed.
Incorporation of 51Cr(VZ)-NAH cells were incubated with 200 pCi of [51Cr]chromate in 1 mM K2Cr04 for up to 8 h. At time intervals, equal amounts of cells were collected the radioactivity of the medium was measured, and each aliquot of cells was washed four times in phosphate-buffered saline or until the radioactivity of the wash buffer was near the background. The cells were resuspended in the original aliquot volume, and their radioactivity was measured. Next, the cells were lysed in SDS buffer and processed as described. After the second 16-h spin (with urea/SDS buffer), the radioactivity of all the pellets and all the urea/SDS supernatants was determined.

RESULTS
As indicated by the trypan blue exclusion assay, the NAH cells tolerated the K2Cr04 very well. Even at the highest chromium concentration employed ( 5 mM), the average survival was the same for both the experiments and controls (87% after 8 h of incubation). Because of the relatively small amounts of protein which became bound to the DNA in cells exposed to K2Cr04 (especially at low concentrations and/or short incubation times), the sensitive silver staining procedure (13) was employed for the detection of proteins resolved by polyacrylamide gel electrophoresis. As can be seen in Fig. 1, where the concentration of K2Cr04 was held a t 1 mM, no detectable cross-linking could be seen until 4 h after the exposure of cells to K2Cr04. The number and the amounts of cross-linked proteins then increased with time until the end of the experiment at 8 h of incubation. By this time, the crosslinking became very extensive ( Fig. 1, lune 5), involving numerous bands over the entire resolution range of the gel. It is noteworthy that although the intensity of the individual bands increased with time, the resolution patterns remained essentially unchanged. There is little evidence for proteinprotein cross-linking although the disproportionate changes in staining intensities of several bands (e.g. approximate M, of 20,000 and 43,000) may suggest some rearrangements of chromatin structure during the incubation or changes in the AgNO3 stainability caused by the presence of protein-bound chromium. The corresponding controls, incubated from 1 to 8 h without K2Cr04 (Fig. 1, lunes 8-12), contained no detectable bands with the exception of the DNase I used for digestion of DNA. Essentially all the proteins visualized by AgN03 staining were also detectable in parallel gels stained with Coomassie Brilliant Blue (although at lower staining intensity; data not shown).
Each well was loaded with 25 pg of sample (as DNA). The gel was stained with AgN03. Lane 11, control cells incubated for 8 h in chromium-free medium. The low molecular weight band present most prominently in lanes I and 2 represents DNase I (see "Materials and Methods"). A constant volume of DNase I was added to resuspended high speed pellets, regardless of the DNA concentration; hence the amount of enzyme showing up in each gel lane is variable, especially in lanes 1-3. cross-linking was time-dependent, we investigated the dependence of DNA-protein cross-linking on K2Cr04 concentration. Results of these experiments are shown in Fig. 2. All incubations were for 8 h, i.e. the time of maximum crosslinking with no significant decrease in cellular viability due to the metal. The first detectable indication of DNA-protein cross-linking appeared at a concentration of 200 p M K2Cr04 in the incubation medium. After that, the intensity of crosslinked protein bands appeared to increase up to the concentration of 5 mM K2CrO4, with a dramatic rise a t 5 mM K2Cr04 resulting in an obvious overload of the resolution system (all samples were layered onto gels a t a protein concentration equivalent to 25 wg of DNA in the pelleted material). As in the time experiments, the increasing concentration of K2Cr04 did not appear to change the qualitative patterns of the crosslinked proteins.
To explain bhe lack of detectable DNA-protein cross-linking at early times of incubation (1 and 2 h) and low concentrations of K2Cr04 (10-100 PM) a t 8 h of incubation, we next investigated the internalization of labeled chromate into the high speed DNA-protein pellets. As indicated in Fig. 3A, which illustrates an experiment using labeled chromate, 51Cr(VI) was taken up by the cells a t about the same rate throughout the entire incubation period (up to 8 h). Moreover, cellular uptake of the label was accompanied by a similar rate of disappearance of the label from the medium. Fig. 3B shows the appearance of radioactivity in the DNA pellets. There is relatively little 51Cr(VI) associated with the DNA pellet up to 2 h, but after the third hour of incubation, the amount of radioactivity increased dramatically. This may explain why we were not able to detect electrophoretically any DNA-protein cross-links after 1 and 2 h of incubation even though the rate of cellular uptake of the 51Cr(VI) was fastest between 1 and 2 h of incubation. Fig. 3B also shows a relatively small amount of radioactivity in the urea/SDS buffer, indicating that after the second 16-h spin, the high speed DNA pellets were almost free of unbound chromium.
Identity of Some Proteins Cross-linked to the DNA by K2Cr04-To gain some insight into the identity of at least some of the DNA-cross-linkedproteins, duplicate gels to those shown in Fig. 2 were transferred to nitrocellulose sheets and incubated with antisera to NAH lamins A and C or to a preparation of NAH nuclear matrix. Reactive antigens were then detected by peroxidase-antiperoxidase staining. Fig. 4 shows that both lamin C and, to a lesser extent, lamin A became cross-linked to the DNA in cells incubated with K2Cr04 concentrations of 200 PM or higher. Again, as in Fig.  2, no cross-linking could be observed a t chromium concentrations below 200 p~. The cross-linking reached maximum between 500 PM and 1 mM K2Cr04 and decreased significantly when the concentration of chromium was 5 mM.
That nuclear matrix proteins are heavily involved in the chromium-mediated DNA-protein cross-linking is shown in Fig. 5. The gel, duplicate to that in Fig. 2, was transferred to nitrocellulose sheets and reacted with antiserum to NAH nuclear matrix. Again, the first evidence of cross-linking could be detected at a    Fig. 2 was incubated with rabbit antiserum to NAH nuclear matrix. See Fig. 2 for K2CrO4 concentrations. Lane 10, nuclear matrix preparation (30 pg of protein). fractions for A260 measurement and CsCl density determination (Fig. 6A). Because of the small amounts of DNA-associated proteins, corresponding fractions from up to six parallel gradients were combined and further pooled, starting with only a single 0.5-ml fraction on top of the gradient. This fraction contained aggregated proteins in particular form and was not, therefore, mixed with other fractions. Hence, the individual 0.5-ml fraction pools represented all fractions 1 (pool l), fractions 2-4 (pool 2), fractions 5-12 (pool 3), fractions 13-16 (pool 4), fractions 17-22 (pool 5), and fractions 23-26 (pool 6). The pools were then electrophoresed, transferred to nitrocellulose sheets, and reacted with appropriate antisera. Densitometric scans of pools 1-6 reacted with antiserum to NAH lamins A and C are shown in Fig. 6B. Only pool 1 (insoluble "skin" on top of the gradient at the average density of 1.42 g/cm3) and pool 4 at the average density of 1.62 g/cm" contained proteins reactive with this antiserum. The free DNA peak was a t average density of 1.71 g/cm3 in pool 5. Pool 6 with A260 even higher than that of pool 5 most of the 0.5-ml fractions collected from the gradient; B, corresponding fractions from six identical gradients pooled as described under "Materials and Methods" to make a total of six pools. The six pools were dialyzed, treated with DNase I, lyophilized, electrophoresed, transferred to nitrocellulose, and reacted with polyclonal antibodies reactive with NAH lamins A and C. The relative amounts of antigen in each pool were determined by densitometric scanning of each of the six lanes on the nitrocellulose transfers. U , lamin A; W, lamin C. linked proteins reacted with antisera to NAH lamins A and C or to NAH nuclear matrix are shown in Figs. 7 and 8, respectively. Again, only pools 1 and 4 contained significant amounts of detectable antigens. The irregular band seen in most lanes in Fig. 8 a t approximately 64 kDa represents contamination of the system with human keratin (20).

DISCUSSION
The propensity of chromium compounds to cross-link nuclear proteins to DNA is well documented in the literature.
Since only hexavalent chromium compounds can cross the cellular membrane while only the trivalent chromium reacts avidly with DNA and forms DNA-protein cross-links (8,9), reduction of hexavalent chromium must be the first step in cellular toxicity of chromium compounds. The presence of NADPH appears essential, and cellular microsomes or endoplasmic reticulum were shown to greatly increase the reduction reaction through the involvement of cellular cytochrome P-450 electron transport system (6).
In our experiments investigating the time and concentration dependence of chromium-induced DNA-protein crosslinks, we noted that incubation times of 1-2 h as well as likely contains RNA. The actual immunoblots of the cross-chromate concentrations of 10-100 p~ did not produce any  detectable cross-links ( Figs. 1 and 2). It also seemed that the cross-links appeared suddenly either after 4 h of incubation or at 200 PM K2Cr04. When we incubated the cells with ["Cr] chromate, we noted that the rate of uptake of chromium by the cells and the rate of disappearance of 51Cr(VI) from the medium was quite steady between 1 and 8 h of incubation (Fig. 3A). When we assayed the amount of radioactivity in the high speed DNA pellets after different incubation intervals, we noticed, however, that the amount after 1 and 2 h was very low as compared to the amount of radioactivity in DNA-protein pellets after 4 h of incubation. The amount of radioactivity in the DNA pellets increased greatly between the 3 and 4 h of incubation, and this high rate of chromium uptake continued up to 8 h (Fig. 3B). Although there may have been some DNA-protein cross-links formed after 2 h of incubation, their amount was too small to be detected in our system. Since the rate of chromium uptake by the cells is steady (Fig. 3A) and DNA-protein cross-link formation is not, it is likely that the hexavalent form of chromium in the cytoplasm is slowly reduced, by the NADPH-dependent chromate reductase, to its trivalent form up to 3 h of incubation. After this time, chromium seems to stimulate the cell to reduce chromate at a much faster rate. This may be the amount of time necessary for the cells to produce more chromate reductase. The lack of any detectable cross-links in cells incubated with 10-100 PM K2Cr04 can also be partially explained by the fact that chromium can be first coordinated to small organic ligands (21,22).
Even though the amount of proteins cross-linked to DNA increased with the increased concentration of chromate, at higher chromium concentrations (1 and 5 mM) the crosslinking amount decreased and some proteins, highly crosslinked at 500 PM to 1 mM, were not cross-linked at 5 mM. We interpreted this finding to mean that at higher metal concentrations chromium may have saturated all the sites on DNA and proteins, preventing the DNA-chromium-protein complexes from being formed. However, since not all the proteins appear affected to the same extent, other conditions, such as structural accessibility of the binding sites, chromating conformation, etc., may be also involved. This interpretation finds support in the fact that even at optimal chromate concentrations not all the nuclear proteins became crosslinked. Employment of a defined nuclear protein population, such as the nuclear matrix used in our experiments, together with antisera specific for selected antigens, makes it possible to follow the antigenic proteins during the. cross-linking events. It is clear that, with minor exceptions, the same proteins (antigens) are involved in the DNA-protein crosslinking, although their quantitative patterns may change with incubation time or chromate concentration. The use of specific antisera reacting with relatively few chromosomal proteins (as compared with the rather complex patterns of AgNOs-stained gels) also shows the virtual absence of detectable protein-protein cross-links. Unless involving the antigenantibody interaction site, such complexes would be easily detected by the antisera as new electrophoretic bands of increasing intensity with the time of incubation.
Although it remains to be determined how the observed cross-linking of chromosomal proteins to DNA correlates to the biological effects of chromium on living cells, such as transformation and mutagenesis, our experiments show that this process is rather selective, possibly depending on the primary structure of the cross-linked proteins, their proximity to DNA, their accessibility to the cross-linker and, perhaps most importantly, on the chromatin conformation. It is noteworthy that some of the proteins of nuclear matrix may actually participate in the regulation of DNA replication and RNA transcription and processing (23). It can be speculated that chromium, by altering the relationships of this nuclear structure to the DNA, may affect and modify any one of these processes.