Chromium inhibits transcription from polycyclic aromatic hydrocarbon-inducible promoters by blocking the release of histone deacetylase and preventing the binding of p300 to chromatin.

Co-contamination with complex mixtures of carcinogenic metals, such as chromium, and polycyclic aromatic hydrocarbons is a common environmental problem with multiple biological consequences. Chromium exposure alters inducible gene expression, forms chromium-DNA adducts and chromium-DNA cross-links, and disrupts transcriptional activator-co-activator complexes. We have shown previously that exposure of mouse hepatoma Hepa-1 cells to chromate inhibits the induction of the Cyp1a1 and Nqo1 genes by dioxin. Here we have tested the hypothesis that chromium blocks gene expression by interfering with the assembly of productive transcriptional complexes at the promoter of inducible genes. To this end, we have studied the effects of chromium on the expression of genes induced by benzo[a]pyrene (B[a]P), another aryl hydrocarbon receptor agonist, and characterized the disruption of Cyp1a1 transcriptional induction by chromium. Gene expression profiling by using high density microarray analysis revealed that the inhibitory effect of chromium on B[a]P-dependent gene induction was generalized, affecting the induction of over 50 different genes involved in a variety of signaling transduction pathways. The inhibitory effect of chromium on Cyp1a1 transcription was found to depend on the presence of promoter-proximal sequences and not on the cis-acting enhancer sequences that bind the aryl hydrocarbon receptor-aryl hydrocarbon receptor nuclear translocator complex. By using transient reporter assays and chromatin immunoprecipitation analyses, we found that chromium prevented the B[a]P-dependent release of HDAC-1 from Cyp1a1 chromatin and blocked p300 recruitment. These results provide a mechanistic explanation for the observation that chromium inhibits inducible but not constitutive gene expression.

More often, the toxic agent is a complex mixture of chemical entities in numbers ranging from a few, such as may occur in occupational exposures, to several hundred, such as in cigarette smoke. Often, these mixtures include a combination of carcinogenic metals and polycyclic aromatic hydrocarbons, as for example those co-released from sources such as fossil fuel combustion or municipal waste incineration. Of these, chromium and B[a]P 1 are among the top 20 hazardous substances in the Agency for Toxic Substances and Disease Registry/ Environmental Protection Agency priority list (1) and are found as co-contaminants in more than 25% of the National Priority List of Superfund sites (2).
Chromium has been known for over 100 years to be a human carcinogen (3). Epidemiologically, the greatest risk of cancer from chromium exposure is associated with Cr(VI). Cr(VI) enters cells via the sulfate anion transporter system and is reduced to intermediate oxidation states, such as Cr(V) and Cr(IV), in the process of forming stable Cr(III) forms (4,5). Cr(III) is the most prevalent form of chromium in the environment and in biological tissues, but the vast majority of the evidence indicates that exposure to trivalent chromium does not induce tumors in animals (6).
The available experimental evidence indicates that chromium exposure has little or no effect on constitutive gene expression but alters inducible gene expression, possibly due to the formation of chromium-DNA adducts, chromium-DNA cross-links, or to the disruption of transcriptional activator-coactivator complexes (3,(7)(8)(9). For example, chromium was found to block the expression of metal-inducible metallothionein or hormone-inducible phosphoenolpyruvate carboxykinase without affecting the expression of housekeeping genes, such as ␤-actin or albumin (4,7,8,10,11). These observations have led to the hypothesis that the chromatin structure of inducible promoters, perhaps by virtue of being more open, may offer a better target for chromium binding than the more closed chromatin of constitutive promoters (12). Chromium-induced DNAprotein cross-links are found preferentially in nuclear matrix DNA (13), where many replication, repair, and transcription proteins associate, suggesting that cross-links between DNA and any or several of these proteins may be responsible for effectively blocking their function.
Generation of free radicals from the reduction of Cr(VI) to Cr(V) and Cr(IV) induces an increase of NFB DNA binding activity but do not cause a concomitant increase in NFB-dependent gene expression (14 -16), possibly because chromium blocks the binding of the p65 subunit to CBP/p300, whose association with p65 is essential for NFB-enhanced transcriptional activity (9). These data suggest that the intermediate oxidation states of chromium may be critical for its effect on gene expression. The molecular mechanism responsible for this effect is likely to involve the reduction process in the persistent stimulation of regulatory pathways affecting interactions of transcription factors with transcriptional co-regulators and chromatin remodeling factors, more so than the binding of the factors themselves to their cognate recognition sites.
Earlier studies from this laboratory have shown that exposure of mouse hepatoma Hepa-1 cells to chromate inhibits the induction of the Cyp1a1 and Nqo1 genes by dioxin, an AHR ligand (17). The AHR is a ligand-activated basic region helixloop-helix/Per-AHR-ARNT-Sim transcription factor that forms heterodimers with ARNT and binds to cis-acting AHR-response enhancer elements in the regulatory domains of target genes, such as Cyp1a1 and Nqo1, leading to changes in chromatin structure and activation of gene transcription (18). These changes also include the recruitment to the transcription machinery of associated co-regulator proteins. In recent years several transcriptional co-regulators, including the co-activators CBP/p300, SRC-1, RIP140, ERAP140, the chromatin remodeling factor BRG-1, and the co-repressor SMRT, have been shown to interact with AHR⅐ARNT complexes and play a role in transactivation (19 -26).
We have studied the effects of chromium on the expression of genes induced by B[a]P, another AHR agonist. We have tested the hypothesis that blocking of AHR-dependent gene expression by chromium was not limited to Cyp1a1 or Nqo1 but that it affected globally a large number of genes induced by B[a]P. We have further tested the hypothesis that chromium blocked gene expression by interfering with the assembly of productive transcriptional complexes in the promoters of inducible genes. We find that chromium prevents the B[a]P-dependent release of HDAC-1 from chromatin, preventing the association of p300 with the Cyp1a1 transcriptional complex and hence blocking gene transcription.

EXPERIMENTAL PROCEDURES
Cell Culture and Chemical Treatments-Mouse hepatoma Hepa-1 (27) cells were grown in ␣-minimal essential medium (Invitrogen) supplemented with 5% fetal bovine serum, 100 g/ml penicillin, and 100 g/ml streptomycin. Cells were treated when grown at 80 -90% confluence. Hepa-AhRDTKLUC cells have been described previously (28); they were derived by stable integration into Hepa-1 cells of a pAhRDT-KLuc3 luciferase reporter plasmid containing the AHR-responsive domain of the mouse Cyp1a1 gene promoter (from Ϫ1100 to Ϫ869), harboring five AHR-responsive elements, fused to the HSV-1 thymidine kinase minimal promoter from Ϫ79 to ϩ53 (29) from which the Sp1binding site had been removed (Fig. 1). Potassium chromate and dichromate (K 2 CrO 4 and K 2 Cr 2 O 7 , referred to here simply as chromium) were freshly dissolved in sterile deionized water prior to use. Sodium butyrate was dissolved in serum-free ␣-minimal essential medium just prior to use. B[a]P was added to the cells in a final Me 2 SO volume of 0.1%. Detailed treatment procedures are given in the text and the figure legends.
Plasmid Constructs and Transfections-To assess AHR-dependent gene expression, we used the luciferase reporter plasmid p-1646Luc2 (28), containing the mouse Cyp1a1 gene promoter region from nucleotide -1646 to ϩ 57. Plasmids expressing the co-activator p300 were from Upstate Biotechnologies Inc.; human BRG-1 expression plasmid and co-regulator plasmids pCR3.1-hSRC-1A and pSG5-Tif 2 (SRC-2) were kindly provided by Dr. Erik Knudsen and Dr. Sohaib Khan (University of Cincinnati Medical Center), respectively. For transfection experiments, cells were plated in 24-well plates at a density of 4 ϫ 10 4 cells/well and transfected at 70 -80% confluence using LipofectAMINE Plus (Invitrogen). Briefly, 50 ng of reporter plasmid, 200 ng of pBluescript or co-regulator plasmids, and 25 ng of pCMV␤-gal plasmid were incubated with Plus reagent for 15 min in serum-and antibiotic-free medium. LipofectAMINE was added, and the mixture was incubated for an additional 15 min. The transfection was carried out for 3 h, and thereafter the medium was changed to normal culture medium containing 5% fetal bovine serum, and transfected cells were allowed to grow overnight before treatment. After treatment, cells were washed twice with PBS and lysed with 100 l of reporter lysis buffer (Promega, Madison, WI). Aliquots of 50 l of cell lysate were used to measure luciferase activity. Light units were determined immediately upon addition of 150 l of luciferase assay buffer (20 mM Tricine, 1.07 mM MgCO 2 , 2.67 mM MgSO 4 , 33.3 mM dithiothreitol, 14.8 mg coenzyme A, 530 M ATP, 0.1 mM EDTA, and 10 mg of luciferin), using a Wallac 1420 Victor plate reader. Luciferase measurements were normalized for transfection efficiency in transient transfections using ␤-galactosidase activity.
RNA Isolation and Real Time Reverse Transcriptase-PCR-Total RNA was isolated using TriReagent (Invitrogen) according to the manufacturer's instructions with additional purification steps applied to RNA samples used for microarray analysis. To verify RNA quality prior to labeling for microarray analyses, samples were analyzed using an Agilent 2100 Bioanalyzer. cDNA was synthesized by reverse transcription of 20 g of total RNA in a total volume of 30 l containing 1ϫ reverse transcriptase buffer, 2.5 M random hexamers, 0.25 mM dNTP, 0.01 M dithiothreitol, 20 units of RNasin and 200 units of SuperScript™ II RNase H Ϫ reverse transcriptase (Invitrogen). Samples were incubated at 42°C for 1 h, and the reverse transcriptase was inactivated by heating to 99°C for 5 min. For real time PCR amplification, 3 l of cDNA were amplified with mouse Cyp1a1 primers (forward primer, 5Ј-GCCTTCATTCTGGAGACCTTCC-3Ј; reverse primer, 5Ј-CAATG-GTCTCTCCGATGC-3Ј), giving a product of 280 bp between exon 5 and 7 of the mouse Cyp1a1 gene. Amplification of luciferase cDNA was with primers 5Ј-CCAACACCCCAACATCTTC-3Ј and 5Ј-CCACAAACA-CAACTCCTCC-3Ј, giving a product of 182 bp. ␤-Actin amplification of the same cDNA samples was used as an internal standard. Amplification was conducted in the Smart Cycler (Cepheid, Sunnyvale, CA) in a total volume of 25 l consisting of 1ϫ Brilliant™ SYBR® Green QPCR Master Mix (Stratagene) and 0.4 M mouse Cyp1a1 primers. The reaction mixtures were heated to 95°C for 10 min and immediately cycled 40 times through a 24-s denaturing step at 95°C, a 60-s annealing step at 55°C, and a 46-s elongation step at 72°C. Cycle threshold (C T ) of each sample was automatically determined to be the first cycle at which a significant increase in optical signal above an arbitrary base line set at 30 fluorescence units was detected. All determinations were done in triplicate. The values shown represent the C T ratios of experimental to control cells treated with Me 2 SO, normalized to the ␤-actin mRNA level in the same sample.
Fluorescent Labeling of Target cDNAs and High Density Microarray Hybridization-Labeling of cDNAs, preparation of microarrays, and hybridization reactions were performed by the University of Cincinnati Functional Genomics Core and are briefly described here. Fluorescence-FIG. 1. Schematic representation of the Cyp1a1 upstream regulatory sequences. The top diagram illustrates the upstream regulatory domain of the murine Cyp1a1 gene, from coordinates Ϫ1200 to the transcriptional start site at ϩ1. Closed symbols denote AHR-response element motifs. Placement of PCR primers to amplify enhancer and proximal promoter sequences is indicated by the arrows and the numerical coordinates of the 5Ј-most nucleotide of each primer. The bottom diagram shows the architecture of the pAhRDTKLUC plasmid.
labeled cDNAs were synthesized from 20 g of total RNA using an indirect amino allyl labeling method (30). The cDNA was synthesized by an oligo(dT)-primed, reverse transcriptase reaction, and the cDNA was labeled with monofunctional reactive cytidine-3 and cytidine-5 dyes (Cy3 and Cy5; Amersham Biosciences). Specific details of the labeling protocols may be found at microarray.uc.edu.
The hybridization probes were from arrayed mouse oligonucleotide microarrays derived from the Operon/Qiagen Verified Libraries currently containing 13,433 sequences from annotated mouse genes, affixed each in a 100-m diameter spot to polylysine-treated microscope slides. The hybridization targets were the paired Cy-3-and Cy-5-labeled control and test cDNAs, which were mixed in approximately equal proportions and applied to the microarray for hybridization under high stringency conditions. After hybridization and washing unhybridized targets, Cy3 (green) and Cy5 (red) fluorescent channels were simultaneously scanned with independent lasers at 10 m resolution. Each comparison was done in triplicate with flipped dye arrays to allow for the removal of gene-specific dye effects. Each comparison consisted of three microarray slides; in two slides, the cDNA was labeled with one fluorescent dye and in one it was labeled with the other dye. Overall, to eliminate labeling bias, cDNA from any one preparation was labeled an equal number of times with Cy3 as with Cy5.
Data Analysis and Normalization-Microarray hybridization data representing raw spot intensities generated by the GenePix software were analyzed to identify differentially expressed genes under different experimental conditions. Data normalization was performed in three steps for each microarray separately. First, channel-specific local background intensities were subtracted from the median intensity of each channel (Cy3 and Cy5). Second, background-adjusted intensities were log-transformed, and the differences (R) and averages (A) of log-transformed values were calculated as R ϭ log 2 (X1) Ϫ log 2 (X2) and A ϭ (log 2 (X1) ϩ log 2 (X2))/2, where X1 and X2 denote the Cy5 and Cy3 intensities after subtracting local backgrounds, respectively. Third, data centering was performed by fitting the array-specific local regression model of R as a function of A (31). The difference between the observed log ratio and the corresponding fitted value represented the normalized log-transformed gene expression ratio. Normalized log intensities for the two channels were then calculated by adding a half of the normalized ratio to A for the Cy5 channel and subtracting half of the normalized ratio from A for the Cy3 channel.
Identification of Differentially Expressed Genes-The statistical analysis was performed for each gene separately by fitting the following mixed effects linear model (32): where Y ijk corresponds to the normalized log intensity on the ith array (i ϭ 1, . . . , 15), with the jth treatment combination (j ϭ 1, . . . ,5), and labeled with the kth dye (k ϭ 1 for Cy5, and k ϭ 2 for Cy3). is the overall mean log intensity; A i is the effect of the ith array; S j is the effect of the jth treatment combination, and C k is the effect of the kth dye. Assumptions about model parameters were the same as described elsewhere (32), with array effects assumed to be random, and treatment and dye effects assumed to be fixed. Statistical significance of the differential expression between different treatment combinations, after adjusting for the array and dye effects, was assessed by calculating p values for corresponding linear contrasts. Multiple hypothesis testing adjustment was performed by calculating false discovery rate values (33,34). Data normalization and statistical analyses were performed using SAS statistical software package (SAS Institute Inc., Cary, NC).
Chromium Speciation-Hepa-1 cells were treated with 25 M K 2 Cr 2 O 7 and isolated immediately or after 1 h of incubation. Cells were rinsed twice with cold PBS and collected by scraping in deionized water. Collected cells were sonicated with four 15-s bursts and mixed with an equal volume of 100% ethanol. The precipitate was removed by centrifugation at 15,000 rpm, and the supernatant was flash-frozen on dry ice. Samples were thawed at 37°C and immediately chromatographed by high pressure liquid chromatography using modifications of methods developed by others (35). Briefly, a 25-cm Dionex CarboPac PA-100 anion exchange column was eluted with a mobile phase containing 50 mM (NH 4 ) 2 SO 4 , pH 9.2, at a flow rate of 1.5 ml/min. The effluent was monitored for chromium at m/z ϭ 53, using an Agilent 7500c ICP-MS (Agilent Technologies, Tokyo, Japan). Retention times for different chromium species were 10.8 min for Cr(VI), 2.1 min for Cr(III) complexed with EDTA, and Ͻ2 min for a broad peak of uncomplexed Cr(III).
Total Chromium Measurements-Hepa-1 cells were treated with 25 M chromate (K 2 CrO 4 ) or dichromate (K 2 Cr 2 O 7 ) for 4 h. Plates were rinsed twice with cold PBS, and cells were scraped from the plates in cold PBS and collected by centrifugation at 3,000 rpm for 2 min. Cell pellets were resuspended in 2 volumes of lysis buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 ) and Dounce-homogenized after addition of Nonidet P-40 to a final concentration of 0.2%. All procedures were carried out on ice. Cell lysates were centrifuged in an Eppendorf microcentrifuge at 3,000 rpm for 2 min. The supernatant and the pellet were frozen at Ϫ80°C and saved as the cytoplasmic and nuclear extracts, respectively. For analysis, samples were thawed, mixed with HNO 3 to a final concentration of 1%, and digested for 30 s in a Parr Microwave Digestion Bomb. Acid digests were analyzed for chromium concentration by monitoring m/z 53 using the Agilent 7500c ICP-MS and comparing to a chromium standard curve. Experiments were carried out in triplicate. For measurements of 51 Cr incorporation, 51 Cr as Na 2 CrO 4 (specific activity 425 Ci/mg; 15715 MBq/mg) was used (PerkinElmer Life Sciences). The final chromate concentration of carrier Na 2 CrO 4 was 40 M with 10% 51 Cr. Cells were incubated for 4 h and rinsed with cold PBS. Nuclear and cytoplasmic fractions were prepared as described above and then counted in a liquid scintillation counter.
Nuclear Run-off Experiments-Nuclear preparations were made from cells treated with Me 2 SO or 5 M B[a]P for 2 h or treated with 50 M chromium for 30 min prior to B[a]P addition. Nuclear run-offs were also done with nuclei from B[a]P-treated cells further treated with 50 M chromium during the run-off reaction. Nuclear preparations and run-off assays were conducted essentially as described (17) with the exception that 2 ϫ 10 7 nuclei and 0.2 mCi of [ 32 P]UTP (3000 Ci/mmol) were used for each reaction. Incubations were carried out for 30 min at room temperature. For those reactions in which chromium was added to the nuclei in vitro, K 2 Cr 2 O 7 was added before addition of [ 32 P]UTP to a final concentration of 1 mM. At the end of the run-off incubation, nuclei were centrifuged at 15,000 rpm for 10 s, and the supernatant was discarded. Purification of the labeled RNA and hybridization was carried out as described previously (17). Incorporation was quantitated using a PhosphorImager (Storm 860, Amersham Biosciences).
ChIP Assays-Protocols were based on procedures published by others (36,37) with minor modifications. Hepa-1 cells were grown to 90 -100% confluence (ϳ2-3 ϫ 10 7 cells) in 150-cm plates. Cells were treated with Me 2 SO vehicle or 5 M B[a]P for 1 h or pretreated with 50 M chromium for 0.5 h followed by either no other treatment or with 5 M B[a]P for an additional hour. Cross-linking was accomplished by a 10-min incubation at room temperature with formaldehyde added directly to the culture media to a final concentration of 1%. The reaction was stopped by adding glycine to a final concentration of 0.125 M. After rinsing twice with ice-cold PBS, cells were scraped from the dishes, pelleted, and washed again with PBS plus 0.5 mM phenylmethylsulfonyl fluoride. Cell pellets were resuspended in cell lysis buffer (5 mM PIPES, pH 8.0, 85 mM KCl, 0.5% Nonidet P-40, plus the protease inhibitors 0.5 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 5 g/ml aprotinin) and incubated on ice for 10 min. Cells were homogenized on ice in a Dounce homogenizer using the B-type pestle 20 times to aid in nuclei release. The nuclei were pelleted and resuspended in nuclei lysis buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 1% SDS plus protease inhibitors) and incubated on ice for 10 min. Chromatin was sonicated on ice with four 10-s bursts of 30 watts with a 30-s interval between bursts. Average length of the DNA was 600 bp. For immunoprecipitation, chromatin was first precleared for 15 min at 4°C with protein A-agarose saturated with bovine serum albumin and salmon sperm DNA (Upstate Biotechnology, Inc., Lake Placid, NY). The supernatant was divided equally among all samples and incubated overnight on a rotating platform at 4°C with antibodies (1 g/2 ϫ 10 7 cells) against AHR (Biomol), ARNT (a gift of Dr. Oliver Hankinson), p300 (Upstate Biotechnology, Inc.) and HDAC-1 (Upstate Biotechnology, Inc.), respectively. Protein A-agarose slurry (20 l) was added and incubated for 15 min at room temperature to allow it to bind to the antibody. The agarose beads were pelleted and washed twice with 1ϫ dialysis buffer (50 mM Tris-HCl, pH 8.0, 2 mM EDTA, 0.2% Sarkosyl) and sequentially four times with IP wash buffer (100 mM Tris-HCl, pH 9.0, 500 mM LiCl, 1% Nonidet P-40, 1% deoxycholic acid). Immune complexes were eluted from the beads with elution buffer (50 mM NaHCO 3 , 1% SDS). Cross-linking was reversed by heating the eluates at 67°C for 4 -5 h. To allow for quantitation of DNA recovery, a constant known amount of 32 P-labeled prokaryotic DNA was added to the eluates before DNA purification. The eluates were digested with proteinase K at 45°C for 1.5 h. DNA was extracted with phenol/chloroform/ isoamyl alcohol and precipitated with ethanol. Radioactivity in the dried DNA pellets was measured, and DNA was dissolved in TE buffer, adjusting for the recovery of radioactivity during the purification process. PCR amplification of the Cyp1a1 promoter was quantitated by inclusion of a small amount of [␣-32 P]dCTP during PCR amplification. Amplification of the distal promoter region was accomplished with the primer sets 5Ј-CTATCTCTTAAACCCCACCCCAA-3Ј and 5Ј-CTAAG-TATGGTGGAGGAAAGGGTG-3Ј, corresponding to the enhancer domain (Ϫ1141 to Ϫ784 bp, Fig. 1). The proximal promoter region was amplified with the set 5Ј-AATTTTTCCTCAAACCCCTCC-3Ј and 5Ј-AGGGACTGAAGTGAAGAGTG-3Ј (Ϫ291 to ϩ71 bp, Fig. 1). PCR products were separated in 10% polyacrylamide gels. After electrophoresis, gels were dried and exposed to x-ray film. DNA bands were visualized in a PhosphorImager (Storm 860, Amersham Biosciences) and quantitated using ImageQuant 5.2 software.

Chromium Inhibits Expression of Many B[a]P-inducible
Genes-The available experimental evidence indicates that chromium exposure has little or no effect on constitutive gene expression but alters inducible gene expression, possibly due to the formation of chromium-DNA adducts, chromium-DNA cross-links, or to the disruption of transcriptional activator-coactivator complexes (7,9,10). Earlier studies from this laboratory have shown that chromium represses the induction of Cyp1a1 and Nqo1 mRNA by TCDD, disrupting the coordinate induction of phase I and phase II gene expression (17). To verify that this was not a unique effect of TCDD, we assessed the effects of chromium on gene transactivation by the Ah receptor using B[a]P, another AHR ligand, as the inducer. In addition, to test the hypothesis that the inhibitory effect of chromium was a generalized phenomenon affecting many inducible genes and not just specific to genes coding for drug-metabolizing enzymes, we determined global expression profiles of Hepa-1 cells treated with B[a]P or with chromium plus B[a]P, using a high density microarray analysis approach. More than 50 genes, including Cyp1a1 and Nqo1 among others, were upregulated by B[a]P alone, and their induction was markedly inhibited by addition of chromium prior to B[a]P (Table I). The pronounced inhibitory effect of chromium was partly relieved if the cells were first induced with B[a]P and chromium was not added until 1.5-2 h later. These data are in good agreement with previous results from this laboratory showing the partial relief of inhibition when addition of chromium was delayed relative to addition of agonist (17). The list of B[a]P-inducible genes inhibited by chromium includes genes involved in a variety of signaling transduction pathways, such as calcium-dependent regulation, receptor-associated kinases, transcription factors, cell cycle regulation, differentiation, and apoptosis in addition to genes involved in drug metabolism. It is evident that the inhibitory effect of chromium is well generalized, affecting the inducibility of more than just a few drug-metabolizing genes.
Significant Amounts of Cr(III) Are Found in the Nuclei-Our earlier observations on the differential effects of chromium inhibition depending on timing of addition had led us to hypothesize that chromium disrupts gene expression at an early step of transcription (17). If that were the case, it could be expected that extracellular chromium would enter the cells and be readily taken up into the nucleus. We used a combination of high pressure liquid chromatography and ICP-MS analyses to study chromium uptake, speciation, and distribution in Hepa-1 cells. Chromium uptake was very fast, Cr(VI) being very rapidly reduced to Cr(III) in the cells. By t ϭ 1 min, only 15% of the detectable chromium was still found as Cr(VI) (retention time ϭ 10.8 min), whereas by t ϭ 1 h Cr(VI) was less than 5% ( Fig. 2A). By 2 h, Cr(VI) was undetectable (data not shown).
Chromium distribution was determined by both uptake of Na 2 51 CrO 4 and ICP-MS detection of chromate or dichromate uptake. Both methods showed significant amounts of chromium present in both cytoplasmic and nuclear fractions after a 4-h incubation. Very similar levels and distribution patterns were found when chromate or dichromate (Cr 2 ) treatments were the source of chromium (Fig. 2B). These data indicate that a large fraction of the chromium administered to the cells enters the nucleus either as reduced Cr(III) or in the process of being reduced from Cr(VI) to Cr(III), and might cause its inhibitory effect on gene expression by interacting with the transcriptional machinery.
Chromium Inhibits B[a]P-induced Transcription-As indicated earlier, chromium blocks inducible gene expression at early times of addition more readily than at later times. It is reasonable to speculate that nuclear chromium, either as Cr(III) or as a reduction intermediate, will block transcription at an early step of initiation or elongation. To test this hypothesis, we used transcriptional run-off assays. We measured transcription rates of several B[a]P-inducible genes in the presence or absence of chromium added to cells prior to nuclear isolation or incubated with the nuclei during the run-off assay. The transcription rate of Cyp1a1, Gsta, and Gstp was induced severalfold by 5 M B[a]P relative to the transcription of ␤-actin, used to normalize transcription rates to the rate of an uninduced gene. Induction was inhibited by 50 M chromium, either when added as Cr(VI) to the cells prior to B[a]P addition or when added to nuclei from B[a]P-treated cells (Fig. 3). These results indicate that chromium blocks the transcription of inducible genes to a greater extent than the expression of the one housekeeping gene, ␤-actin, used to normalize the data.

Inhibition of B[a]P-induced Gene Expression by Chromium Takes Place at the Level of the Proximal
Promoter-None of several antioxidants, such as catalase, superoxide dismutase, and 2,2,6,6-tetramethylpiperidinyl-N-oxyl, could reverse the inhibition by chromium of B[a]P-dependent gene expression induction (data not shown). On the other hand, the combined evidence of our speciation and nuclear run-off studies suggested that Cr(III) or a reduction species intermediate between Cr(VI) and Cr(III) was able to react with an element of the transcriptional machinery and block transcription. Because the effect resulted in transcription repression, we suspected that chromium interfered with the molecular interactions leading to assembly of the transcriptional complex on the Cyp1a1 promoter chromatin. To test this hypothesis, we used the Hepa-AhRDTKLUC cells that carry a stably integrated AhRDTK-Luc3 luciferase reporter plasmid. In this plasmid, luciferase expression is regulated by the AHR-responsive enhancer of the mouse Cyp1a1 gene promoter bearing five AHR-responsive elements. In this plasmid, the enhancer is fused to the HSV-1 thymidine kinase minimal promoter from which the Sp1-binding site had been removed. In these cells, mRNA expression of both luciferase and endogenous Cyp1a1 genes is induced by AHR ligands; however, the replacement of the Cyp1a1 proximal promoter (from position Ϫ869 to the transcription start site, see Fig. 1) with the HSV-1 tk minimal promoter facilitates the detection of Cyp1a1 expression changes resulting from molecular events occurring in the proximal promoter domain. To test the role of HDAC in repression by chromium of B[a]P gene induction, we compared Cyp1a1 and luciferase mRNA levels in cells treated with vehicle or with sodium butyrate, an HDAC inhibitor. Cyp1a1 mRNA levels were highly induced by 2 M BaP treatment, an effect that was practically abolished when cells were treated with 50 M chromium prior to B[a]P but not when B[a]P was added 2 h before chromium; 2 mM sodium butyrate further reversed the effect of chromium in the latter case (Fig. 4A). Expression of luciferase mRNA was induced by B[a]P as well, but was insensitive to chromium treatment (Fig. 4B), suggesting that control elements in the Cyp1a1 proximal promoter were responsible for the chromium-dependent repression of gene induction by B[a]P. In addition, pretreatment of the cells with 2 mM sodium butyrate also led to a further elevation of mRNA levels for both chromium plus B[a]P co-treatments (Fig. 4B). These results suggest that inhibition of B[a]P-inducible gene expression was likely to result from interference with chromatin remodeling processes associated with the assembly of transcription factors or the initiation of transcription at the proximal promoter.

Chromium Blocks B[a]P-dependent HDAC Release and Prevents Association of p300 with the Ah Receptor Complex at the
Cyp1a1 Promoter-Our previous work has shown that chromium does not block the association of AHR/ARNT heterodimers with their cognate binding AhRE sites in the Cyp1a1 enhancer (17). Hence, it seemed more likely that the transcrip-tional inhibitory effect of chromium could be exerted through interactions with chromatin occurring after enhanceosome assembly and engagement of the AHR/ARNT-responsive enhancer by the AHR complex. Specifically, given the relief of the chromium-induced block by HDAC inhibition, we hypothesized that chromium might interfere with the association of HAT with chromatin. If this were the case, we would expect that chromium treatment would also repress the superinduction of AHR/ARNT-dependent gene expression resulting from overexpression of the co-activators SRC-1, SRC-2, and p300, which recruit HAT to the trancriptional complex. To determine whether chromium would impair HAT recruitment, we used the p-1646Luc2 luciferase reporter, containing both the enhancer as well as the proximal promoter sequences of the mouse Cyp1a1 gene, in transient co-transfection assays with expression plasmids for SRC-1, SRC-2, p300 and with the chromatin remodeling BRG-1. This reporter will be expected to respond to chromium treatment in a similar manner as the Cell extracts were chromatographed in a Dionex anion exchange column, and the effluent was monitored for 53 Cr with an Agilent ICP-MS. Retention times for different chromium species were 10.8 min for Cr(VI), 2.1 min for Cr(III) complexed with EDTA, and Ͻ2 min for a broad peak of uncomplexed Cr(III). B, chromium distribution in cells was determined by either uptake of Na 2 51 CrO 4 measured by liquid scintillation counting or uptake of K 2 CrO 4 or K 2 Cr 2 O 7 measured by ICP-MS analysis. In each case, after incubation for 4 h with chromium, cells were separated into nuclear and cytoplasmic fractions. indigenous Cyp1a1 gene. All the co-activators tested increased basal luciferase expression levels in Me 2 SO-treated cells and increased B[a]P-induced levels from 2-to 6-fold over those in control cells co-transfected with an empty vector (Fig. 5). In all cases, pretreatment with 25 M chromium repressed superinduction, which was almost completely abolished by pretreatment with 50 M chromium (Fig. 5). These results indicated that recruitment of HAT by the co-activators and possibly the concomitant release of HDAC was impaired by chromium. To confirm this conclusion we used ChIP assays, focusing our analyses on the same two regulatory domains of the Cyp1a1 promoter, namely the cis-acting enhancer region, located between coordinates Ϫ1200 and Ϫ800 and containing five canonical AHR-responsive elements, and the proximal promoter region containing sequences immediately upstream of the transcription start site (Fig. 1), where the general transcriptional factors associated with RNA polymerase II bind. For ChIP assays, Hepa-1 cells were exposed to Me 2 SO vehicle, to 50 M chromium, to 5 M B[a]P, or to 50 M chromium followed 1 h later by 5 M B[a]P. All cultures were incubated at 37°C for 90 min, after which time chromatin was prepared and immunoprecipitated with antibodies to AHR, ARNT, p300, and HDAC-1. After cross-link reversal, DNA was purified and amplified by PCR using primers bracketing either the enhancer or the proximal promoter regions (Fig. 1). Immunoprecipitation profiles with anti-AHR and anti-ARNT antibodies were similar. As could be expected, binding of AHR and ARNT to en-hancer chromatin was highly induced by B[a]P treatment, as determined by the 10 -12-fold increase in enhancer sequences detected in the immunoprecipitates (Fig. 6A). In agreement with our previous electrophoretic mobility shift findings (17), much of the AHR/ARNT binding was still retained in chromium-pretreated cells. On the other hand, B[a]P-induced binding of the AHR⅐ARNT complex to promoter proximal sequences in the TATA box region was completely abolished by chromium pretreatment (Fig. 6A). Chromium alone had no apparent effect on AHR or ARNT binding to either enhancer or proximal promoter chromatin. These results suggest that chromium causes a transcriptional block by silencing the proximal promoter through interference with the recruitment of enhanceosome complexes to the proximal promoter, as described recently for Cyp1a1 (39,40). In agreement with this concept, immunoprecipitation with anti-p300 antibodies showed that p300 association with either enhancer or proximal promoter chromatin occurred only in B[a]P-treated cells and that a complete block of binding, to levels even below those in Me 2 SOtreated cells, took place in B[a]P-induced cells pretreated with chromium (Fig. 6B). Conversely, in uninduced cells, immunoprecipitation with anti-HDAC-1 antibodies showed a low level of HDAC-1 binding at the enhancer domain but a high level at the proximal promoter, which was completely eliminated by B[a]P treatment (Fig. 6B). Pretreatment with chromium abolished the release of HDAC from the proximal promoter, which showed levels comparable with those of uninduced cells. In the absence of B[a]P treatment, chromium alone more than doubled the level of promoter proximal chromatin that was bound to HDAC (Fig. 6B). These results strongly suggest that chromium-mediated inhibition of B[a]P-induced gene expression results from the chromium-dependent retention of HDAC at the proximal promoter chromatin and the prevention of p300 entry, in turn, blocking the association of the transcriptional AHR⅐ARNT complex with promoter proximal TATA box chromatin.

DISCUSSION
The results presented in this article show that inhibition of Ah receptor-dependent expression by chromium is a generalized phenomenon that extends to at least 50 B[a]P-inducible genes involved in a variety of cellular processes (Table I). Inhibition occurs at the transcriptional level at least for the three genes, Cyp1a1, Gsta, and Gstp, tested by run-off assays, suggesting that this might be the case for all the B[a]P-inducible genes whose induction was sensitive to inhibition by chromium. These results confirm and extend previous findings that suggest that chromium interferes with an early step of inducible gene transcription (4,7,8,10,11).
The evidence from our chromium distribution experiments indicates that significant amounts of chromium reach the nucleus and that B[a]P-induced transcription in isolated nuclei is also sensitive to chromium-dependent inhibition, as is the case in whole cells. Several lines of evidence suggest that the critical chromium-sensitive step occurs at the level of the proximal promoter and involves its interaction with the transcriptional assembly machinery. In stably transfected Hepa-AhRDTKLUC cells (Fig. 4), although the presence of the AHR-responsive Cyp1a1 enhancer in the integrated reporter plasmid makes luciferase expression to be inducible by B[a]P, the absence of the proximal promoter sequences make it insensitive to inhibition by chromium, unlike its effect on the endogenous Cyp1a1 gene. These experiments not only map the chromium-sensitive domain to the proximal Cyp1a1 promoter, but they also reveal a role for HDAC-1 in chromium-dependent inhibition of gene expression, which appears to be blocked in part by inhibition of HDAC with sodium butyrate. . Three independent ChIP assays were performed using antibodies against AHR, ARNT, p300, and HDAC-1 before DNA purification. Immune complexes were eluted from protein A-Sepharose, and prior to extraction, a constant known amount of 32 P-labeled prokaryotic DNA was added to the eluates to allow for quantitation of DNA recovery. Cyp1a1 promoter regions were amplified including a trace of [␣-32 P]dCTP in the PCR mix. Amplified DNAs were separated by 10% polyacrylamide gel, and the gels were exposed to x-ray film and to Phos-phorImager screens. DNA bands were quantified using ImageQuant 5.2. The position of the primers used for PCR amplification is shown in Fig. 1. A, immunoprecipitation with anti-AHR and anti-ARNT antibodies. The two upper panels show ethidium bromide staining of PCR-amplified input DNA from each of the four treatments to verify that the initial DNA amounts was the same in all four preparations. B, immunoprecipitation with anti-p300 and anti-HDAC-1 antibodies. Quantification of PCR bands is shown in the histograms below each data set.
FIG. 5. Chromium blocks co-activator-mediated superinduction of B[a]Pinduced gene expression. Hepa-1 cells were co-transfected with the indicated plasmids and with the reporter p-1646Luc2 as described under "Experimental Procedures" together with the ␤-galactosidase-expression vector pCMV␤-gal. Transfected cells were allowed to grow overnight before treatment with 5 M B[a]P, with or without prior treatments for 2 h with 25 or 50 M chromium. Control cells were treated with an equal volume of Me 2 SO (DMSO) vehicle. After treatment, aliquots of cell lysates were used to measure luciferase and ␤-galactosidase activities. Luciferase measurements were normalized for transfection efficiency using ␤-galactosidase activities. The ordinate represents the fold induction of normalized luciferase activity relative to the activity in empty vector co-transfected, Me 2 SOtreated cells. The experiments were repeated three times with each determination done in triplicate, and the ordinate values represent the average Ϯ S.D. of a representative experiment.
Hexavalent chromium appears to be rapidly reduced by cells to stable trivalent chromium. Within our experimental parameters, we do not observe, nor are our experiments designed to detect, oxidative stress or DNA damage associated with chromium reduction through the reactive intermediates Cr(V) and Cr(IV). It is likely, however, that the intermediate oxidation states of chromium are critical factors not only for the generation of oxidative stress and DNA damage but also, within a more immediate time frame, for its effects on gene expression. Electron spin resonance and spin trapping measurements have shown that generation of free radicals from the reduction of Cr(VI) to Cr(V) and Cr(IV) induces an increase of NFB DNA binding activity in cultured human and rat cells (14 -16); however, NFB-dependent gene expression is not concomitantly increased, because chromium blocks the binding of the p65 NFB subunit to CBP/p300, a transcriptional co-activator with intrinsic HAT activity, whose association with p65 is essential for NFB-enhanced transcriptional activity (9). Furthermore, Cr(VI) has also been found to modify the transactivation potential of MTF-1 without affecting basal or inducible binding to metal-response elements (41). Similarly, the Ah receptor from Hepa-1 cells treated with chromium prior to TCDD induction retains its ability to bind to DNA in electrophoretic mobility shift assays (17) or to chromatin, as found in our present results by ChIP analyses. Hence, it is likely that chromium causes the persistent repression of regulatory pathways that affect the function of transcriptional co-regulators rather than the binding of transcription factors to their cognate recognition sites.
The AHR signal transduction pathway is modulated by several nuclear cofactors, including the co-activators CBP/p300, SRC-1, RIP140, ERAP140, the co-repressor SMRT, and chromatin remodeling factors such as BRG-1 (19 -26). In transient reporter assays, none of these relieved the effect of chromium on B[a]P-mediated expression, yet their effect in superinduction of B[a]P-induced gene expression was strongly blocked by chromium, indicating that chromium interferes with the recruitment or the binding of HAT to the transcriptional complex. ChIP assays confirmed this conclusion and revealed a critical role for HDAC-1 in the repression of B[a]P-inducible gene expression. The binding of AHR and ARNT to enhancer and to proximal promoter chromatin was reduced by chromium pretreatment to ϳ50% of the level in B[a]P-treated cells. In comparison, the interaction of HDAC-1 and p300 with chromatin was much more critically affected by chromium. p300 binding to either enhancer or proximal promoter chromatin was stimulated by 2-3-fold by B[a]P treatment, an effect that was completely blocked by chromium pretreatment prior to B[a]P induction. Conversely, HDAC-1 binding, which was very strong in the proximal promoter of uninduced or chromium-treated cells, was reduced to undetectable levels by B[a]P but remained strong when the cells were treated with chromium prior to B[a]P induction. These results lead us to conclude that chromium blocks the release of HDAC-1 from the Cyp1a1 proximal promoter, maintaining a state of histone deacetylation and transcriptional repression and preventing the recruitment of p300, with subsequent histone acetylation and transcriptional induction. Recent data have shown that HDAC-1 and HDAC-2 are recruited to Sp1/Sp3-binding sites through butyrate-response elements associated with Sp1/Sp3-binding sites (42). A computer analysis of the Cyp1a1 promoter reveals the presence of multiple potential Sp1/Sp3-binding sites, some of which have been shown to be functional and to cooperate with AHR and ARNT in the induction of the Cyp1a1 gene (19). It is attractive to speculate that chromium, known to cross-link DNA and proteins (43-46), may be able to cross-link HDAC to any one of a number of potential partners, including histone, bound Sp factors, or Sp-response DNA elements in the Cyp1a1 promoter. With the advent of new tools developed to map changes in histone acetylation status on a genome-wide basis (37,47), it will be possible to address this question.
Our ChIP analysis results suggest that in the process of gene activation, the AHR⅐ARNT complex bound at the enhancer domain makes contact with promoter proximal sequences. These data are consistent with a looping model of AHR/ARNTmediated Cyp1a1 trancriptional activation and agree with recent results reported by Tian and co-workers (40). It is noteworthy, however, that the basal expression level of this gene is maintained in a silent state not solely by the absence of a bound activated receptor complex but also by the additional repressive effect of HDAC-1 bound at the proximal chromatin. This unexpected finding is worthy of further investigation.