Hypermethylation of Metallothionein-I Promoter and Suppression of Its Induction in Cell Lines Overexpressing the Large Subunit of Ku Protein*

We have shown previously that the heavy metal-induced metallothionein-I (MT-I) gene expression is specifically repressed in a rat fibroblast cell line (Ku-80) overexpressing the 80-kDa subunit of Ku autoantigen but not in cell lines overexpressing the 70-kDa subunit or Ku heterodimer. Here, we explored the molecular mechanism of silencing of MT-I gene in Ku-80 cells. Genomic footprinting analysis revealed both basal and heavy metal-inducible binding at specific cis elements in the parental cell line (Rat-1). By contrast, MT-I promoter in Ku-80 cells was refractory to any transactivating factors, implying alteration of chromatin structure. Treatment of two clonal lines of Ku-80 cells with 5-azacytidine, a potent DNA demethylating agent, rendered MT-I gene inducible by heavy metals, suggesting that the gene is methylated in these cells. Bisulfite genomic sequencing revealed that all 21 CpG dinucleotides in MT-I immediate promoter were methylated in Ku-80 cells, whereas only four CpG dinucleotides were methylated in Rat-1 cells. Almost all methylated CpG dinucleotides were demethylated in Ku-80 cells after 5-azacytidine treatment. To our knowledge, this is the first report that describes hypermethylation of a specific gene promoter and its resultant silencing in response to overexpression of a cellular protein.

inhibit RNA polymerase I transcription in vitro, which was overcome by the addition of purified Ku protein to the reaction (6). Further study showed that Ku could physically and functionally interact with another RNA polymerase I transcription activator, CPBF (14), a protein that is structurally and functionally related to USF, a helix-loop-helix-zipper DNA binding factor (15,16). On the contrary, Ku purified from the growtharrested cells inhibited polymerase I transcription, which could be restored following addition of purified Ku from the control cells (11).
In the course of our study on the regulation of metallothionein (MT) 1 gene expression (for review, see Refs. 17 and 18) by heavy metals and oxidative stress, we observed abolishment of the induction of MT-I and MT-II (19) (the two predominant isoforms of MT) as a result of overexpression of the large subunit (p80) of Ku in rat fibroblast cells, Rat-1. Interestingly, overexpression of the small subunit (p70) of Ku or of the heterodimer (p70 and p80) did not inhibit MT induction by the heavy metals. Nuclear extracts from the p80-overexpressing cells contained a repressor activity that could block transcription from the MT-I promoter (19). The activities of two of the key transcription factors, Sp1 and MTF-1, that modulate MT-I transcription were not modified in the extracts from these cells. The present study was undertaken to determine the molecular mechanism by which MT-I expression is repressed in the p80 overexpressing cells. This study showed that the promoter hypermethylation is primarily responsible for the suppression of MT induction.

MATERIALS AND METHODS
Cell Cultures, Treatment with 5-Azacytidine, Heavy Metals, and Northern Blot Analysis-Rat-1 (parental cell line) as well as cell lines that overexpress p70 subunit (R70-15), p80 subunit (R80-1 and R80-6), and both subunits (R7080-6) were generously provided by Gloria Li, Memorial Sloan Kettering Institute. For convenience, overexpressing cell lines were also designated Ku-70, Ku-80, and Ku-7080, respectively. All experiments except that described in Fig. 2B were performed using the clonal isolate R80-1 (designated Ku-80 throughout the text). In the experiment described in Fig. 2B, R80-1 (Ku-80) and another clonal isolate, designated R80-6, were used. The culture conditions and treatment with the heavy metals were described earlier (19). For 5-AzaC treatment, Ku-80 cells at 25-30% confluency were grown in presence of increasing concentration of the analog for 72 h. RNA isolation and Northern blot analysis were performed as described (19).
In Vivo Genomic Footprinting-In vivo methylation of cellular DNA by dimethyl sulfate (DMS) and the subsequent DNA extraction were done following the protocol of Mueller and Wold (20). The procedure of Ping et. al (21) was followed for ligation-mediated (LM)-PCR. MRE and MLTF/ARE sites (adeno major late transcription factor/antioxidant re-sponse element) were analyzed using one set of upper strand-and one set of lower strand-specific primers. The primers used to read the lower strand were: RMTI-5Ј-1, 5Ј-ACATGATGTTCCACACGTCAC; RMTI-5Ј-2, 5Ј-TGTTCCACACGTCACACGG; and RMTI-5Ј-3, 5Ј-CACACGTC-ACACGGGTCCTC. The annealing temperatures for this set of primers were 58, 61, and 64°C, respectively.
The PCR product was directly sequenced using fmol TM sequencing kit (Promega) with the S2 and A2 primers.

RESULTS
The MT-I Promoter in Ku-80 Cells Is Refractory to the Transacting Factors, as Revealed by In Vivo Genomic Footprinting-Previous study (19) in our laboratory demonstrated that Ku-80 cells (cells overexpressing the large subunit of the protein Ku) were unable to induce MT-I or MT-II in response to heavy metals such as CdSO 4 or ZnSO 4 . Under this condition, the parental Rat-1 cells, a fibroblast cell line, and Ku-70 cells (cells overexpressing the small subunit of Ku) or Ku-7080 cells (cells overproducing both subunits of Ku) could induce the gene by the toxic metals. This study suggested that the noninducibility of MT-I gene in Ku-80 cells was because of the existence of a repressor in the nuclear extract prepared from Ku-80 cells (19).
Alternatively, overexpression of p80 could result in altered chromatin structure that inhibits access of positive factors to the corresponding regulatory elements in the promoter. To address the latter issue, we took advantage of in vivo genomic footprinting (IVGF) to analyze and compare the state of MT-I promoter occupancy in Rat-1 and Ku-80 cells before and after exposure to heavy metals. This would allow us to identify the sequence-specific DNA-protein interactions that are unique to both cell lines, and potential alterations of these interactions specifically in the parental cells after exposure to heavy metals.
IVGF analysis of the MT-I promoter was performed by LM-PCR with primers designed to amplify the sequence from Ϫ226 to Ϫ16 bp, which encompasses all cis-acting elements involved in the basal and metal-induced expression of MT-I. Fig. 1A depicts the metal regulatory elements and other key elements of the rat MT-I promoter. Both Rat-1 and Ku-80 cells were treated with DMS before and after exposure to ZnSO 4 or CdSO 4 , and genomic DNA was extracted, cleaved with piperidine, followed by amplification of the promoter fragment by LM-PCR. Naked genomic DNA isolated from both Rat-1 and Ku-80 cells were treated in the same manner as intact cells and amplified to provide the genomic G-ladder. DNA-protein interactions can result in protection of G-residues from DMS reactivity that produce less intense bands (indicated by arrows in Fig. 1, B and C). Alternatively, such interactions can result in more intense bands of G-residues exhibiting enhanced DMS reactivity (indicated by asterisks in Fig. 1, B and C) when compared with the naked DNA ladder or untreated control. The DMS reactivity and cleavage pattern of both strands of naked genomic DNA from Rat-1 and Ku-80 cells were identical (Fig. 1, B and C).
Heavy metal-induced footprinting was observed in the lower 4 . Cells untreated or treated with ZnSO 4 (100 M) or CdSO 4 (30 M) were subjected to DMS treatment, followed by DNA extraction and piperidine cleavage (see "Materials and Methods"). The naked DNA from both wildtype and mutant cells were also treated with DMS and piperidine. An identical amount of DNA (2 g) from each sample was then subjected to LM-PCR with primers specific for the upper and the lower strands. A, schematic diagram of the MT-I promoter depicting cis elements that are relevant to heavy metal-induced expression. LM-PCR of the lower strand (B) and LM-PCR of the upper strand (C) of MT-I promoter in Rat-1 and Ku-80 cells. Lanes 2, 3, and 4 represent Rat-1 cells that are untreated or treated with ZnSO 4 (100 M) and CdSO 4 (30 M), respectively; lanes 6, 7, and 8 correspond to similar samples from Ku-80 cells, respectively; lanes 1 and 4 denote naked DNA from Rat-1 and Ku-80 cells, respectively. Arrows (3) indicate G-residues that are protected, whereas asterisks (*) represent hypersensitive G-residues. strand at MRE-d element where varying levels of protection of the G-residues spanning this element were detectable upon Zn 2ϩ or Cd 2ϩ treatment of the Rat-1 cells (Fig. 1B, lanes 3 and  4). Distinct footprinting of MRE-c element and the overlapping MRE-cЈ element were clearly detectable in the lower strand of MT-I promoter of Rat-1 cells, which were indistinguishable for Zn 2ϩ -and Cd 2ϩ -treated samples (Fig. 1B, lanes 3 and 4). The G-residues encompassing MRE-e, MRE-b, and MRE-a element of the lower strand of MT-I promoter from the heavy metaltreated Rat-1 cells were also protected (Fig. 1B). In addition, Zn 2ϩ -and Cd 2ϩ -induced footprinting was observed at the composite element MLTF/ARE in the lower strand of control Rat-1 cells when compared with the naked G-ladder (Fig. 1B). The G-residues in Sp1 binding element were constitutively protected (Fig. 1C, lane 2, double arrow), and some were hypersensitive in the upper strand of control Rat-1 cells (Fig. 1C, asterisk) before and after ZnSO 4 or CdSO 4 treatment. A very prominent DMS reactivity (hypersensitive residue) was observed at another Sp1 binding site overlapping with the MLTF/ ARE element in the upper strand of MT-I promoter from both Zn 2ϩ -and Cd 2ϩ -treated Rat-1 cells (Fig. 1C). On the contrary, none of the conspicuous constitutive and/or metal-induced footprinting observed in Rat-1 cells were visible in Ku-80 cells irrespective of metal ion treatment (Fig. 1, B and C). This observation indicates that overexpression of p80 protein results in altered chromatin structure. Such chromatin modification is likely to lead to a conformation of the MT-I promoter refractory to any positive factor binding, which could explain the lack of heavy metal-induced MT-I induction in Ku-80 cells.

MT-I Gene Is Activated by Various Inducers in Ku-80 Cells after 5-AzaC Treatment-
The failure of the positive factors to footprint on the promoter of other genes silenced because of methylation of CpG dinucleotides has been reported by many laboratories (23). Based on these published reports and our own IVGF data with MT-I gene (Fig. 1), we reasoned that MT-I promoter may be silenced in Ku-80 cells because of promoter methylation. To test this possibility, we determined the effect of 5-AzaC, a potent DNA demethylating agent (24) on MT-I expression. As treatment with 10 M 5-AzaC for 24 h did not result in induction of the gene in response to heavy metals (19), we first determined the concentration of the cytosine analog and duration of treatment that can activate MT-I gene in Ku-80 cells. Increase in 5-AzaC concentration from 10 to 20 M and time of incubation with the drug from 24 to 72 h resulted in significant expression of MT-I mRNA after CdSO 4 treatment ( Fig. 2A). These data showed that MT-I gene can indeed be induced in Ku-80 cells (R80-1 clonal line) by heavy metals after prolonged treatment with relatively high concentration of 5-AzaC. The MT-I promoter methylation in the clonal cell line used in the present study may be because of a critical chance event of methylation that contributed to further methylation. To rule out this possibility, we used another randomly selected clonal isolate (R80-6) of p80 overexpressing cells. This study showed that MT-I gene is repressed in these cells as well and that demethylation with 5-AzaC reactivated the promoter (Fig.  2B). The silencing of MT-I gene in the cells overexpressing the 80-kDa subunit is, therefore a specific event. After demethylation with 5-AzaC, MT-I gene could be activated by other inducers as well, e.g. zinc, dexamethasone, cycloheximide (data not shown). We then compared the level of induction of MT-I mRNA in Rat-1 cells following exposure to CdSO 4 to that of Ku-80 cells treated with 5-AzaC plus CdSO 4 (Fig. 2C). Under this condition, MT-I mRNA level in the 5-AzaC-treated Ku-80 cells reached almost the same level as that in Rat-1 cells. We could not detect any basal expression of MT-I in Ku-80 cells after 5-AzaC treatment. This data suggests that hypermethy-lation of MT-I gene plays a key role in its silencing in Ku-80 cells.
All CpG Elements within MT-I Promoter Are Methylated in Ku-80 Cells-To ensure that the activation of MT-I gene by various inducers after 5-AzaC treatment was indeed because of demethylation of the promoter, we performed bisulfite genomic sequencing before and after the drug treatment. Sequence analysis of MT-I promoter (Ϫ225 to ϩ1 bp with respect to transcription initiation site) that harbors the cis elements for most of the transacting factors revealed high density of potential methylatable CpG dinucleotides (twenty-one) spanning this region (Fig. 3A). Therefore, it was essential to compare the methylation status of MT-I promoter in Ku-80 cells with that in Ku-70 and Ku-7080 cell lines, which express the smaller subunit of Ku and both subunits together, respectively. The latter two cell lines can express MT-I mRNA at the same level as the parental cell line (Rat-1) in response to heavy metals (19). For this purpose, we performed bisulfite genomic sequencing of DNA isolated from (a) Rat-1, (b) Ku-80, (c) Ku-80 treated with 5-AzaC, (d) Ku-70, and (e) Ku-7080 cells. The bisulfite treatment of the genomic DNA resulted in complete conversion of unmethylated cytosines to uracils without affecting methylated cytosines. Subsequent PCR amplification and sequencing provided an accurate and consistent representation of mCpG sequences.
Completion of bisulfite conversion of the genomic DNA isolated from the five different cell samples were confirmed by Tsp509I digestion. The restriction site for Tsp509I (2AATT1) does not exist in MT-I promoter and is generated only after bisulfite conversion of unmethylated cytosine residues to uracils, which are amplified as thymine during PCR. The Tsp509I digestion analysis of PCR-amplified bisulfite-treated DNA from Ku-7080, Ku-70, Rat-1, and Ku-80 cells treated with 5-AzaC

FIG. 2. Northern blot analysis of MT-I expression in response to heavy metals in Ku-80 cells after treatment with 5-AzaC. A,
Ku-80 (R80-1) cells were exposed to increasing concentrations of 5-AzaC for 72 h followed by CdSO 4 (30 M) treatment, total RNA was isolated and subjected to Northern blot analysis with ␣-32 P-labeled, randomprimed, mouse MT-I cDNA as probe. B, Ku-80/R80-1 and R80-6 cells were exposed to 20 M 5-AzaC for 72 h, followed by treatment with 30 M CdSO 4 . Total RNA was isolated and subjected to Northern blot analysis as above. C, Rat-1 and Ku-80 cells were treated with 30 M CdSO 4 following treatment with 20 M 5-AzaC. RNA was isolated and subjected to Northern blot analysis as above. 18 S ribosomal RNA was used in all samples to confirm equal loading of RNA. and Ku-80 cells showed total cleavage of amplified DNA from all five samples (Fig. 3B), indicating completion of bisulfite reaction. To assess the existence of CpG methylation, we digested all the DNA samples with BstUI (CG1CG). The MT-I promoter contains the restriction site for this enzyme, which should be retained after bisulfite conversion if the cytosine residues were methylated. Only the amplified DNA from Ku-80 cells showed cleavage with BstUI, indicating the presence of methyl CpG dinucleotide in this cell line. The bisulfite conversion and restriction enzyme digestion of the amplified DNA were performed at least three times with different batches of chromosomal DNA.
To study the status of all 21 CpG base pairs, we sequenced the amplified DNA from all five cell cultures (see Fig. 4 for schematic representation of the sequencing data). Interestingly, all 21 CpG base pairs (between Ϫ225 to ϩ1 site) in Ku-80 cells were methylated, whereas in Rat-1 cells only four methylated CpG dinucleotides were present in the same promoter stretch. It was logical to conceive that the additional CpG methylation of MT-I promoter in Ku-80 cells was a direct or indirect outcome of p80 overexpression in these cells. Interestingly, none of the CpG elements were methylated in Ku-70 or Ku-7080 cells, implying that methylation of MT-I promoter occurred in the cell line that overexpressed specifically p80 but not in cell lines overexpressing p70 homodimer or the p70/p80 heterodimer.
Sequencing of bisulfite-converted DNA from Ku-80 cells treated with 5-AzaC showed extensive demethylation of the MT-I promoter. This explains the ability of 5-AzaC-treated Ku-80 cells to express MT-I upon heavy metal treatment as observed in Northern blot analysis (Fig. 2). This particular batch of 5-AzaC-treated Ku-80 cells retained one methylated CpG at Ϫ112 site of the MT-I promoter, which did not interfere with the expression of the gene. DISCUSSION Although the role of Ku and associated DNA-PK has been well established in V-D-J recombination and double strand DNA break repair, the mechanism by which Ku regulates expression of various genes at the transcriptional level is still an enigma. Earlier studies showed that overexpression of the p70 subunit of Ku or the heterodimer represses induction of Hsp-70 protein (13,25). The p80 subunit of Ku can function as a receptor for somatostatin (9), whereas the small polypeptide (p70) can directly interact with the proto-oncogene p95 vav (26) and GCN5, a transcriptional adapter that has chromatin-modifying activity (27). The present study using different clonal isolates of Rat-1 cells overexpressing the large Ku subunit, has clearly demonstrated for the first time that overproduction of a single subunit of a cellular protein can hypermethylate and consequently silence the promoter of a specific gene. These results implicate that the individual subunits of Ku can play important regulatory roles in cellular functions. Although Ku exists predominantly as heterodimer, a certain population of this protein may also exist as homodimers which could exert their specific functions. The ratio of homodimer to heterodimer may vary with the cell type and under certain physiological conditions and disease states.
Because the Ku homodimers are generally unstable, they are degraded rapidly (2). Consequently, the amount of p80 in Ku-80 cells is considerably less than that in Ku-7080 cells where both p70 and p80 are overexpressed (19,25). The overproduction of any individual subunit under certain physiolog- ical conditions or in pathological states may, therefore, have gone undetected. We have recently demonstrated that MT-I promoter is hypermethylated in a rat hepatoma and a mouse lymphosarcoma cell line (relative to the parental cells) which resulted in the repression of MT-I expression in these cancer cells. 2 Chromatographic fractionation of the nuclear extracts from the hepatoma has frequently yielded fractions that contain variable amounts of Ku homodimers. 3 This observation is consistent with the potential role of Ku subunits in the methylation of some promoters, which results in their inactivation.
Previous study in our laboratory showed that only the cells that overproduce the large subunit of Ku fail to express metallothionein gene and that the suppression of MT-I transcription is at least, in part, because of the production of a repressor in these cells (19). This conclusion was reached by studying the in vitro transcription of MT-I promoter in nuclear extracts derived from the wild-type and mutant cells. This study did not suggest the role of promoter methylation along with the production of an active repressor in silencing the MT-I gene. The present data have clearly demonstrated that long term treatment of the cells with higher concentration of 5-AzaC was essential to remove the inhibitory methyl groups from methylated CpG residues in MT-I promoter.
The methylation of four CpG elements of twenty one such dinucleotides in the MT-I promoter of Rat-1 cells did not affect the ability to express MT-I in response to heavy metals. The persistence of DNA binding activity of MTF-1, the key transcription factor for MT-I induction, in vitro (gel shift assay) or ability to transactivate in vivo (transient transaction assay) (28) following methylation of some MRE sites may probably explain the capability of Rat-1 cells to express MT-I despite methylation of a few key cis elements. Interestingly, the DNA binding and transactivation property of the zinc finger protein Sp1 are also not inhibited upon methylation of its binding site (28). The differential methylation pattern of MT-I promoter in Rat-1 and Ku-80 cells leads us to postulate that the methylation of just a few key cis elements in Ku-80 cells alone may not cause MT-I gene silencing. Rather, extensive methylation of the promoter is likely to recruit the methyl CpG binding proteins along with repressors of transcription, like histone H1 or Sin3A (29) that leads to alteration in the chromatin structure and consequently gene silencing. The repressor detected in our earlier study in Ku-80 cell extract (19) is probably a methyl C binding protein (MeCP), as it has been shown earlier that MeCP can inhibit transcription in vitro from unmethylated as well as methylated promoters depending upon the assay conditions (30). Further, overexpression of MeCP2 is known to inhibit Sp1-activated transcription of human leukosialin gene, although methylation of Sp1 binding site does not interfere with its binding (31). Preliminary UV cross-linking studies have shown that the activity of a MeCP is significantly higher in nuclear extract from the Ku-80 cells compared with that from Rat-1 cells. Alternatively, it is conceivable that a repressor different from MeCP may function in concert with the methylated promoter to achieve repression of MT induction in Ku-80 cells.
It is unlikely that a repressor protein binds directly to the promoter (between Ϫ226 to ϩ1 bp) and alters the chromatin conformation, as no unique or new footprinting appeared in the MT-I promoter of Ku-80 cells relative to that of Rat-1 cells. We extended our footprinting analysis up to Ϫ350 bp of the promoter using another set of primers, which did not indicate any DNA/repressor interaction in Ku-80 cells (data not shown). The events occurring upstream of Ϫ350 bp cannot, however, be discerned from the present data. Based on these data, it is reasonable to conclude that overexpression of the 80-kDa subunit of Ku protein silences MT-I gene in vivo by methylation of CpG island. Methylated CpGs would then be bound by a specific MeCP(s), resulting in altered conformation of chromatin structure that is inaccessible to the transcription activators.
Ku-80 homodimer by itself or via a signaling mechanism might have initiated MT-I promoter methylation and subsequent silencing of the gene. Direct addition of recombinant p80 (expressed from histidine-tagged p80 in reticulocyte lysate) to the in vitro transcription system from Rat-1 cell extract did not inhibit MT-I promoter activity in vitro (19). These data do not, however, preclude the possibility of a direct role of p80 in the MT-I promoter hypermethylation in vivo. To test the possible activation of DNA methyl transferase in Ku-80 cells that could result in hypermethylation of MT-I promoter, we measured the enzyme activity in the whole cell extract prepared from the cell lines overexpressing different subunits of Ku. The activity of this enzyme in these cells was similar to that in the parental cell line (Rat-1). 4 These results clearly show that MT-I promoter methylation in Ku-80 cells was mediated by some mechanism(s) other than activation of DNA-methyl transferase.
Finally, the relative specificity of MT-I promoter methylation following overexpression of the 80-kDa subunit of Ku deserves comment. We have examined the expression of some growth suppressors (e.g. p16, p53) and housekeeping genes (e.g. Hsp70, heme oxygenase I) in Ku-80 cells. It is known that p16 and p53 promoters are hypermethylated and consequently silenced in many different types of cancers (32). None of these genes were silenced in the Ku-80 cells (data not presented). Furthermore, the MT-I promoter was inducible by heavy metals in the cells that overproduce the smaller subunit of Ku or both subunits together (19). Differential display of genes or related techniques should be performed to identify other genes that are suppressed in the Ku-80 cells. Clearly, overexpression of just the large subunit appears to trigger a signal transduction pathway that leads to hypermethylation of MT-I promoter and its silencing. The challenge for future studies is to identify the step(s) that leads to hypermethylation of a gene promoter as a result of p80 overexpression and the mechanism by which MT-I promoter and probably a few other promoters are susceptible to hypermethylation in Ku-80 cells.