¹Sealy Center for Molecular Science and Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galveston, Texas 77555, USA

²Department of Immunology and Microbiology, University of Texas Medical Branch, Galveston, Texas 77555, USA

³Reactor Institute, Kyoto University, Osaka 590-04, Japan

^aAuthor for correspondence

^bPresent address: The Cleveland Clinic Foundation Research Institute, 9500 Euclid Avenue, Cleveland, OH 44195

^cPresent address: Department of Immunology and Microbiology, University of Texas Medical Branch, Galveston, TX 77555, USA

O⁶-methylguanine-DNA methyltransferase (MGMT), a ubiquitous DNA repair protein, removes the mutagenic DNA adduct O⁶-alkylguanine, which is synthesized both endogenously and after exposure to alkylnitrosamines and alkylating antitumor drugs such as 2-chloroethyl-N-nitrosourea (CNU). The MGMT gene is highly regulated in mammalian cells and its overexpression, observed in many types of tumor cells, is often associated with cellular resistance to CNU. Dexamethasone, a synthetic glucocorticoid hormone, was found to increase MGMT expression in HeLa S3 cells, concomitant with their increased resistance to CNU. Two putative glucocorticoid responsive elements (GREs) were identified in the human MGMT (hMGMT) promoter. Transient expression of the luciferase reporter gene driven by an hMGMT promoter fragment containing these GREs was activated by dexamethasone. DNase I footprinting assays demonstrated the binding of glucocorticoid receptor to these sequences. In vitro transcription experiment showed that these DNA sequences are functional in glucocorticoid receptor signal-mediated activation of transcription. These results suggest glucocorticoid-mediated induction of the MGMT gene contributes to high level expression of MGMT.

CNU; DNA damage; DNA repair; glucocorticoid; gene regulation

Introduction

O⁶-methylguanine-DNA methyltransferase (MGMT), a ubiquitous DNA repair protein, is responsible for removal of O⁶-alkylguanine from DNA. This alkylated base adduct is generated in DNA either endogenously or after exposure to alkylating carcinogens and antitumor drugs, such as chemotherapeutic 2-chloroethyl-N-nitrosourea (CNU) derivatives (Mitra and Kaina, 1993; Pegg, 1990; Singer and Gruenberger, 1983). Although O⁶-methylguanine is a relatively minor component of methylated base adducts, it is believed to be the major mutagenic and carcinogenic lesion because its pairing with thymine, in addition to cytidine, during DNA replication, leading to a GC to AT transition mutation (Eadie et al., 1984; Snow et al., 1984). Moreover, the CNU derivatives yield O⁶-chloroethylguanine in DNA which ultimately lead to the formation of stable cytotoxic DNA crosslinks (Erickson et al., 1980; Ludlum, 1990). Repair of O⁶-alkylguanine adducts by MGMT is thus effective in protecting cells against the mutagenic, carcinogenic and cytotoxic effects of DNA alkylating agents (Brent et al., 1985; Robins et al., 1983; Schold et al., 1989).

Most normal cells and the majority of human tumor cells express the MGMT gene, although the level of expression varies widely in a tissue- and cell- type-specific manner (D'ambrosio et al., 1987; Sagher et al., 1989; Washington et al., 1989). A few untransformed rodent cells and a fraction of primary human tumor cells (-10%) express very low levels (Day et al., 1980; Mitra and Kaina, 1993; Myrnes et al., 1984; Pegg, 1990; Pieper et al., 1991; Yarosh et al., 1983); these MGMT-deficient (Mex^-) cell lines are very sensitive to N-nitrosomethyl-N-nitrosoureatype alkylating carcinogens and drugs. Conversely, some tumor cells express MGMT at a high level and this has profound clinical implications, because MGMT appears to be the only protein involved in removal of the O⁶-chloroethylguanine, which is the direct consequence of treatment with CNU derivatives. Thus, resistance to CNU derivatives often correlates with tumor MGMT levels (Brent et al., 1985; Kokkinakis et al., 1997; Pegg, 1990). This is also consistent with the observation that inactivation of MGMT by treatment with O⁶-benzylguanine significantly increases sensitivity of tumor cells to CNU derivatives (Liu et al., 1996; Pegg et al., 1995). Thus elucidation of the molecular mechanisms controlling MGMT expression may lead to development of drugs that specifically inhibit or enhance MGMT expression, and is of significant clinical importance. Earlier studies showed that MGMT expression can be enhanced by a variety of genotoxic agents in rodent and human cell lines (Fritz et al., 1991) and more recently studies have implicated altered chromatin organization and nucleosome positioning and aberrant silencing in the CpG island in the promoter region in regulating the expression of the gene (Patel et al., 1997; Watts et al., 1997). However, the cis-elements present in the promoter or the transacting factors responsible for the variable, cell-specific expression of MGMT are not known.

The hMGMT gene, encoding an mRNA of 950 nucleotides, consists of 5 exons and spans more than 170 kilobase pair (kb; Nakatsu et al., 1993), and its 5'-regulatory promoter has been cloned and characterized for known regulatory sites (Harris et al., 1991). The DNA sequence in the 5' region preceding exon 1 is extremely GC-rich and lacks TATA as well as CAAT boxes. The presence of two non-consensus glucocorticoid response elements (GREs) separated by 20 base pairs (bp) and located 1 kb upstream of the transcription start site, suggested potential regulation of MGMT expression by glucocorticoids. Two AP-1 sites were also found in close proximity. Based on cis-sequences identified in many glucocorticoid-responsive genes, a GRE consensus sequence of 15 nucleotides was deduced; it contains two hexameric, partially palindromic half-sites, GGTACA and TGT^T/_CCT, separated by three nonspecific nucleotides. The candidate GREs in the hMGMT promoter have conserved the stronger, TGTCCT half-sites but are divergent from the consensus GRE in the other half site.

Glucocorticoids exert a multitude of physiological effects by modulating the expression of specific genes, mostly at the transcriptional level by binding to and activating intracellular glucocorticoid receptors (GR; Beato, 1989; McEwan et al., 1995; Tsai and O'Malley, 1994). The steroid-receptor complexes may interact with the GREs of glucocorticoid-responsive genes and/or with other transcription factors, e.g. AP-1 (Diamond et al., 1990; Saatcioglu et al., 1994). In addition to their use as anti-inflammatory agents, glucocorticoids are often used to reduce peritumoral edema, one of the commonest causes of neurologic deterioration, in brain tumor patients undergoing chemotherapy (Levin et al., 1997; Heiss et al., 1996; Yung, 1990). Modulation of DNA repair activity, in this case an increase, by these steroids may reduce the therapeutic efficiency of CNU-type alkylating drugs often used in therapy of CNS tumors. It is thus important to test whether glucocorticoids can induce MGMT in tumor cells. We have analysed the hMGMT regulatory domain and find that the hMGMT gene is induced by glucocorticoid mediated by the GREs.

Results

Dexamethasone-mediated activation of the hMGMT gene

We have characterized in HeLa S3 cells the role of glucocorticoid in h-MGMT mRNA expression, protein formation and bioactivity. HeLa S3 cells were stimulated with 0.01 M dexamethasone for 0 to 24 h and MGMT mRNA expression was analysed (Figure 1). Levels of MGMT mRNA started to increase at 12 h and continued to rise to a maximum up to 17 - 18 h, after which they remained steady until 24 h. Quantitation of MGMT mRNA: 18S rRNA ratios showed a maximal 2.5-fold increase. Simultaneous Western analysis of protein extracts from the same cell samples for induced MGMT protein synthesis demonstrated a twofold increase at 15 h after which the induced level was maintained for at least 24 h (Figure 2). Because glucocorticoid-dependent activation of genes requires expression of glucocorticoid receptor, we measured the level of GR in HeLa S3 cells by ligand binding assay. Approximately 15 - 18,000 GR molecules/cell were present in log-phase HeLa culture.

Correlation of MGMT activity with CNU resistance

An increase in hMGMT protein level might be expected to lead to a corresponding increase in hMGMT function. It is known that increased hMGMT activity can lead to resistance to alkylating agents. To confirm this in dexamethasone-treated HeLa S3 cells, we evaluated whether the elevated hMGMT protein level was correlated with resistance against such alkylating agents as BCNU in vivo. HeLa S3 cells were treated with dexamethasone (0.01, 0.05 or 0.1 M) for 16 h before treatment with differing concentrations of BCNU for 1 h in a serum-free medium. Cells were then grown in fresh medium for 9 - 12 days, and survival was tested by clonogenic assay. The dexamethasone treatment led to a significantly increased survival to challenge by BCNU. At 40 M, BCNU killed almost all of control HeLa S3 cells, but 10% of the dexamethasone treated cells showed resistance (Figure 3). This corresponds to a ~2 log increase in the number of clonogenic cells.

Deletion analysis for identification of GRE in hMGMT promoter

We evaluated whether the glucocorticoid response elements in the promoter of the hMGMT gene might be responsible for dexamethasone-mediated mRNA induction and protein synthesis. To test the transcriptional regulatory elements we constructed recombinant plasmids in which a luciferase reporter gene was placed under the control of various lengths of 5' flanking region of the hMGMT promoter and tested the luciferase activity in transient expression assays (Figure 4). Transfection experiments showed that pML-72/+24 was able to produce basal transcription in HeLa S3 cells; the larger promoter fragments showed a stepwise increase in level of expression (data not shown).

Figure 4 shows the induction in expression of luciferase activity in extracts from HeLa S3 cells transfected with the above constructs and stimulated with dexamethasone. A nearly fourfold increase in luciferase activity was observed with the construct p-954/+24ML. When the regulatory region was shortened at the 5' terminus by 88 nucleotides (p-866/+24ML), induction was reduced to twofold. Interestingly, the p-2500/+24ML and longer reporter constructs also showed only a 2-3-fold induction, suggesting the presence of negative upstream regulatory element(s). The level of induction from the p-954/+24ML was about 2.5-fold above that from p-72/+24ML, a fold of induction comparable to that of endogenous hMGMT mRNA or protein. A computer assisted search for sequences in the hMGMT promoter similar to the previously described consensus GRE revealed two putative half-site GREs, at positions -929 to -914 and -892 to -877. To analyse the regulatory property of these sequences, we made deletion constructs by removing either GRE-1 or both GRE-1 and GRE-2. Deletion of GRE-1 had very little effect on induction, but deletion of both GREs caused a significant reduction (compare p-954/+24ML with p-900/+24ML and p-866/+24ML). However, glucocortidal induction was not completely abolished by deletion of the two GREs. The residual induction (about 1.5-fold) suggests that additional active cis-regulatory elements are present within the remaining region of hMGMT promoter, or that there are some GRE-like sequences present in the vector itself giving rise to the background inductive activity. To better define the cis-acting sequences presumably mediating the effects of glucocorticoid on promoter activity, DNase I footprinting and in vitro functional analyses of the glucocorticoid receptor-binding sites were performed.

Footprinting of h-MGMT GREs with recombinant GR

A DNaseI protection assay on a 174 bp BamHI/ApaI fragment (-954 to -780) containing both GRE-1 and GRE-2 was performed with the DNA-binding domain of hGR. As shown in Figure 5, footprints were seen with both the GREs, -929 to -914 and -892 to -877 upstream of the transcriptional start site, arranged as imperfect palindromes separated by three nucleotides, typical of a functional GRE. As the concentration of GR-DNA binding domain was increased, GRE-2 was footprinted prior to GRE-1, and the protection patterns were also different.

GR-dependent activation of the hMGMT promoter in vitro

We tested the functionality of the putative MGMT GREs in an in vitro transcription assay. Tsai et al. (1990) have shown that pLov TATA, the parent plasmid used to generate pMGMT TATA gave only weak transactivation in the presence and absence of recombinant hGR, whereas there was strong increase in transcription with pPRE₂TATA. In this assay, in addition to pMGMTTATA, we also used pPRE₂TATA both for comparative purposes as well as to ensure the quality of the recombinant hGR preparation. pMGMTTATA, which contains two hexameric halfsites 20 bp apart, showed a fourfold stimulation (Figure 6a, lanes 3 and 4) when hGR was added, while pPRE₂TATA, which contain two copies of the palindromic GRE from the tyrosine aminotransferase (TAT) gene showed a tenfold stimulation (Figure 6a, lanes 1 and 2) upon addition of an identical quantity of hGR. The contribution of GRE-1 and GRE-2 to transcription of the reporter cassette was further analysed in detail (Figure 6c) by comparing the amount of hGR-dependent transcription from plasmids containing mutations in either GRE-1 (MW) or GRE-2 (WM) or both (MM). The results show that both GRE's contribute to GR-dependent activation. Relative to some threefold induction due to 372 bp hMGMT promoter fragment, the 62 bp fragment with WW, MW, WM and MM showed induction of 2.4-, 1.6-, 1.8- and 1.4-fold, respectively. Although the response was modest, our results were reproducible in multiple independent experiments.

Discussion

DNA repair activity in bacteria is often inducible in response to specific DNA damage (Friedberg et al., 1995). Although nonspecific induction of DNA repair genes has been observed in mammalian cells after exposure to a variety of genotoxic agents, the effect is often small and limited to very few cell lines. Our finding that a hormonally active steroid activates expression of a DNA repair gene, namely, MGMT, raises the question of how DNA repair in general is modulated in mammalian cells.

We made parallel determination of the levels of hMGMT mRNA and protein as well as its activity in HeLa S3 cells treated with dexamethasone. As we have examined, HeLa S3 expresses a high level of GR. Therefore, dexamethasone-dependent activation of hMGMT gene should not be restricted by the level of GR. We found a 2 - 3-fold induction of both mRNA and protein levels at 15 h and beyond. As expected, the induction by dexamethasone was associated with significantly increased resistance of HeLa S3 cells to BCNU, a CNU derivative. Dexamethasone-dependent induction was also observed in MIA PaCa-2, a human pancreatic carcinoma cell line, and in IMR-90, a human diploid lung fibroblast and its transformed lines. Results similar to ours were recently reported in studies with rat hepatoma cell line by Grombacher et al. (1996), who found a fivefold increase in MGMT mRNA level after dexamethasone treatment and a similar induction in transient expression of a reporter gene under control of the hMGMT promoter. However, in that report no systematic analysis of the putative GRE(s) in MGMT promoters was performed. We here have identified and characterized the DNA sequences that confer the bulk of the glucocorticoid responsiveness to the hMGMT gene. The functional GREs we have identified are located about 929 bp upstream of the transcription start site. The functionality of these GREs was established by several independent criteria, namely, glucocorticoid-dependent induction of the luciferase reporter gene, DNase I footprinting assay, and in vitro transcription assays. DNase I footprinting assays confirmed the binding of the DNA binding domain of hGR to the putative sequences responsible for dexamethasone induction. GRE-2 seemed to show a stronger binding to GR than did GRE-1, and may be more important in the induction; however, further analysis will be needed to better define the individual roles of these two elements. The GREs were also found functional, though with a reduced efficiency compared to palindromic GRE, when placed upstream of a TATA-based promoter, as shown by hGR-dependent increase in transactivation from pMGMT TATA. Detailed analysis of wild-type and mutant MGMT GREs also confirmed a significant difference in functionality between the wild-type and mutant plasmids. Both GRE's appear to be involved in GR-dependent activation because mutation of either one significantly reduced the activation level. These studies strongly suggest that increased transcription significantly contributes to the overall increase in hMGMT mRNA level. It is interesting to note that, unlike most genes responsive of GRE-mediated regulation, the MGMT gene does not contain a TATA element. However, the function of its GREs is independent of the presence of this element. The delayed induction of MGMT by dexamethasone may reflect the requirement for synthesis of secondary mediator protein(s) involved in the hormonal induction, or that a longer time is needed before the increased synthesis reaches a level sufficient to be detected by the present methods.

The known cell type-specific variation in MGMT expression suggests that the MGMT promoter may be complex. We intend to identify the intermediates in the regulatory pathways in order to develop strategies for deliberate modulation of hMGMT gene expression, since it directly affects the biological outcome of exposure to many genotoxic alkylating agents and to antitumor drugs. The present study suggests that glucocorticoid-mediated activation of the MGMT gene may contribute to the overexpression of MGMT, often observed in tumor cells. Such a possibility could be directly tested and, if proven correct, will provide an opportunity for reversing MGMT overexpression by glucocorticoid antagonists. An important phenotype of some primary tumor cells is the near lack of expression of MGMT. These Mex^-/Mer^- cells are exquisitely sensitive to CNU-based drugs. The molecular basis for extinction of the MGMT gene is not known. It was recently suggested that altered chromatin organization and nucleosome positioning and aberrant silencing in the CpG island of the promoter region, surrounding of the MGMT gene is linked to Mex^- phenotype (Patel et al., 1997; Watts et al., 1997). However, the GRE sequences are located outside these CpG island regions and it appears that GREs in the MGMT promoter are involved in activation rather than repression. Finally, it should be stressed that downregulation of MGMT should result in increased therapeutic efficiency of CNU-based drugs. By the same token, the common practice of administration of glucocorticoid together with CNU derivatives for treatment of certain cancers such as primary and metastatic brain tumors, multiple myeloma etc. needs to be reconsidered because hormone-mediated upregulation of MGMT may reduce the therapeutic efficiency of the alkylating drug (Levin et al., 1997; Salmon and Cassady, 1997).

Material and methods

Cell lines and reagents

The HeLa S3 (ATCC CCL 2.2) cell line was used in all experiments. Fetal bovine serum, DMEM/F12 medium, Dulbecco's Phosphate buffered saline and penicillin-streptomycin were purchased from Life Technologies. The kit for radio-labeling DNA probe was obtained from Pharmacia Biotech., Inc. The Enhanced Chemiluminescence (ECL) Western blotting kit was purchased from Amersham Corporation. N,N'-bis(2-chloroethyl)-N-nitrosourea (BCNU) was obtained from Bristol Laboratories. Dexamethasone and other reagents were of analytical grade and were obtained from Sigma.

Cell culture

HeLa S3 cells were grown in Dulbecco's modified Eagle's Medium supplemented with F12 containing 5% (v/v) fetal bovine serum and 100 units/ml of penicillin/streptomycin. The cells were routinely grown in monolayer cultures at 37°C in 5% CO₂ and had a doubling time of about 20 h. By subculturing when the cells reached approximately half confluence, growth was kept logarithmic. For transient transfection and other experiments, cells were plated 24 h before treatment so that they were at approximately 50% confluency at the time of use.

Northern blot analysis

Total cellular RNA was isolated from HeLa S3 cells using RNAzol (Tel-Test, Inc.) based on the method of Chomczynski and Saachi (1987) and quantitated spectrophotometrically. For Northern analysis, RNA was denatured by heating for 15 min at 65°C in 0.5´MOPS pH 7.8 buffer containing 6.5% formaldehyde and 50% formamide and then fractionated by electrophoresis on 1% agarose gel containing 2.2M formaldehyde. The RNA was transferred onto a nylon membrane (Micron Separations, Incorporated) by capillary electrophoresis and then hybridized with ³²P-labeled probe for MGMT cDNA or 18S ribosomal RNA. The MGMT cDNA probe was an EcoRI fragment from pKT100 (Tano et al., 1990) and the 18S rRNA probe was a 1.15 kb BamHI - EcoRI fragment from p5B (Bowman et al., 1981). Both prehybridization and hybridization were carried out at 65°C in 6´SSC, 5´Denhardt's, 0.1% SDS. The membranes were subsequently washed twice with 1´SSC and 1% SDS for 30 min, and then with 0.1´SSC and 1% SDS at 50°C for required amount of time to reduce the background to an optimal level. The filters were exposed subsequently to either phosphor screen or Kodak-XAR film at -80°C with intensifying screens for appropriate periods of time so that the signals were in the linear range. Signals were quantified by using Imagequant (Molecular Dynamics) software or NIH Image (NIH) software, or both.

Western blot analysis

Immunoblotting analysis using polyclonal antibodies to hMGMT was performed on control or dexamethasone-treated HeLa S3 extracts. The monolayer cells were lysed by incubating with 400 l/dish of lysis buffer (50 mM Tris. HCl, pH 8, 150 mM NaCl, 0.1% SDS, 1% NP 40 and 50 g/ml PMSF) on ice for 20 min. Cells were then scraped with a cell lifter, transferred to a microcentrifuge tube, vortexed for 5 s and centrifuged at 4°C for 2 min at 15 000 g. Supernatants were transferred to a second tube, assayed for protein content (bicinchoninic acid procedure) and stored at -80°C. Samples (50 g protein/lane) were electrophoresed in a SDS-polyacrylamide gel and then electrophoretically transferred to a nitrocellulose membrane (Protran^TM, Schleicher & Schuell). The membranes were treated per manufacturer's instructions (ECL technique, Amersham). Polyclonal rabbit antisera against bacterially expressed human MGMT (Alpha Diagnostic International, SanAntonio, TX, USA) was used at 1 : 400 dilution. A hyperfilm (Amersham Corporation) was used to visualize the chemiluminescent bands. Exposure time was adjusted such that the signal intensity was in the linear range. Signals were quantified as described in case of Northern Blot analysis.

Construction of plasmids

The 11.3 kb hMGMT genomic fragment isolated from an EMBL clone was subcloned into pBluescript II SK+/- phagemid at the SalI site. The construct was named pBS1202. After establishing a restriction map, various segments of the promoter region which included the transcription start site were subcloned in the pGL2 basic (Promega) vector (Figure 4).

The 96 bp SmaI/XbaI fragment (containing the basal promoter region of hMGMT) was subcloned at the SmaI/NheI site of pGL2 basic to generate luciferase expression plasmid p-72/+24ML. The 599 bp PstI/XbaI fragment was first cloned into pUC19 at the PstI/XbaI site. It was then cut out with EcoRV/XbaI and subcloned at the SmaI/NheI site of pGL2 basic to generate p-575/+24ML. The 978 bp BamHI/XbaI fragment was first subcloned into pUC19. The resultant recombinant was called p1911. The KpnI/XbaI fragment from p1911 was cloned into KpnI/NheI site of pGL2 basic to generate p-954/+24ML. The 2.5 kb SstI/XbaI and the 3.5 kb KpnI/XbaI fragments were subcloned in SstI/NheI and KpnI/NheI site of pGL2 basic respectively to generate p-2500/+24 ML and p-3500/+24 ML. The 9.5 kb SalI/HindIII fragment was cloned in SmaI/HindIII site of pGL2 basic to generate p-9500/+250 ML.

A KpnI site was generated by site-directed mutagenesis between GRE-1 and GRE-2 in p1911. The KpnI/XbaI digested 924 bp fragment lacking the GRE-1 was then subcloned into the KpnI/NheI site of pGL2 basic to generate p-900/+24ML. Similarly another KpnI site was generated 11 bp downstream of the GRE-2 in p1911 and KpnI/XbaI digested 890 bp fragment lacking both the GREs was subcloned into KpnI/NheI site of pGL2 basic to construct pML-866/+24ML. Site-directed mutagenesis was carried out using Stratagene's double-stranded site-directed mutagenesis kit. The kit is based on a modification of the unique site elimination mutagenesis procedure described by Deng and Nickoloff (1992). The DNAs used for transfection were prepared by alkaline lysis procedure and then purified on CsCl-ethidium bromide density gradients.

Assay for transiently expressed reporter gene

Transfection of HeLa S3 cells was carried out by electroporation (Izumi et al., 1996) because with this method the size of the vector does not significantly affect the transfection efficiency. About 1´10⁷ cells were trypsinized, washed twice and resuspended in 250 l of ice cold Dulbecco's Phosphate buffered saline. Luciferase expression plasmid (10 g), along with -galactosidase expression plasmid (5 g) under the control of SV40 promoter (pSV -gal; Promega), were mixed with the suspended cells in a 0.4 cm cuvette (BioRad), kept on ice for 10 min and transfected in a BioRad Gene Pulse electroporation unit at 960 F capacitance and 250 V. After 30 min of incubation on ice, the cells were plated in normal media in 100 mm tissue culture dishes. Dexamethasone was added to the medium 5 - 10 min after plating (the time giving maximal induction) and the cells were harvested 16 h after the shock. A preliminary time course study indicated that the level of expression is maximum at about 16 h and decreases significantly at 48 h. Luciferase and -galactosidase activity were measured using kits from Promega and the manufacturer-recommended protocols.

DNase I footprinting assay

The footprinting reaction was done essentially as described by Leblanc and Moss (1994). The BamHI/XbaI hMGMT promoter fragment (subcloned in pUC19) was digested with BamHI and ApaI and a 175 bp fragment containing the sequences of both GRE-1 and GRE-2 was isolated. This fragment was labeled with -³²P-dCTP (3000 Ci/mmol, DuPont-NEN) by filling in the 3' recessed BamHI end using FPLC pure DNA polymerase Klenow fragment (GIBCO - BRL) and then purified in a 5% non-denaturing PAG to remove unincorporated dNTPs. The binding reaction was performed in a total volume of 20 l containing 10 l of 2´binding buffer (20% glycerol, 2 mM EDTA, 4 mM MgCl₂, 8 mM DTT, 100 mM NaCl and 40 mM HEPES, pH 7.9), 0.5 l of 1 g/l poly[dI.dC], 2 ng of end-labeled DNA fragment (~10 000 c.p.m.) and varying amounts of bacterially expressed human GR DNA binding domain (GR-DBD, kindly provided by T Ma and EA Thompson). The reaction was incubated on ice for 20 min and then digested with 5 l of 0.002 kU/l DNase I (Sigma) for 2 min. The phenol chloroform-extracted and ethanol-precipitated DNA was electrophoresed in a 8% denaturing-polyacrylamide sequencing gel.

In vitro transcription

Transcriptional run-on synthesis was performed to test the transcriptional activation function of hMGMT GRE in heterologous promoter context. MGMT-GRE G-free cassette reporter plasmid pMGMTTATA (generated by inserting the 372 bp BamHI - PstI fragment containing putative MGMT GRE elements 1 and 2 into the BglII site of pLov TATA (Tsai et al., 1990; in the forward orientation) was used to test the function of MGMT GREs. For further characterization of the GRE-1 and GRE-2, 62-mer oligonucleotides containing mutation(s) in none, either one or both were designed and cloned in BglII site of pLovTATA in the forward orientation. The plasmid names include WW, WM, MW and MM to indicate wild-type, mutation in GRE-2, mutation in GRE-1 and mutations in both, respectively. The second half-site TGT of the GREs were changed to AAA in all of the mutant constructs. pLovTATA, containing only the ovalbumin TATA box, was used as a negative control and pPRE₂ TATA (Tsai et al., 1990), which contains two copies of the palindromic GRE from the tyrosine aminotransferase gene, was used for comparison of strength. pAdML200 was added to each tube as an internal control. The in vitro transcription reaction was performed essentially as described by Tsai et al. (1990) using HeLa S3 nuclear extract (Shapiro et al., 1988) supplemented with human glucocorticoid receptor (hGR, 100 ng/30 l reaction volume). pAdML200 (10 ng) along with 250 ng of pLovTATA, pMGMTTATA or pPRE₂ TATA were used in the reaction. This assay method generates one correctly initiated transcript and one non-specific read-through transcript (of a slightly larger size) for each template (Bagchi et al., 1990; Klein-Hitpass et al., 1990).

Acknowledgements

The authors like to thank Dr E Whorton for helping with the statistical analysis, Dr D Konkel for a careful editing of the manuscript and Ms Wanda Smith for expert secretarial assistance. This research was supported by US Public Health Service grants CA 31721 and ES 07572 (to SM)

Bagchi MK, Tsai SY, Weigel NL, Tsai MJ and O'Malley BW. (1990). J. Biol. Chem. 265, 5129-5134. MEDLINE

Beato M. (1989). Cell 56, 335-344. MEDLINE

Bowman LH, Rabin B and Schlessinger D. (1981). Nucleic Acids Res. 9, 4951-4966. MEDLINE

Brent TP, Houghton JA and Houghton PJ. (1985). Proc. Natl. Acad. Sci. USA 82, 2985-2989. MEDLINE

Chomczynski P and Sacchi N. (1987). Anal. Biochem. 162, 156-159. Article MEDLINE

D'Ambrosio SM, Samuel MJ, Dutta-Choudhury TA and Wani AA. (1987). Cancer Res. 47, 51-57. MEDLINE

Day RS, Ziolkowski CHJ, Seudiero DA, Meyer SA, Lubiniecki AS, Giardi AJ, Galloway SM and Bynum GD. (1980). Nature 288, 724-727. MEDLINE

Deng WP and Nickoloff JA. (1992). Anal. Biochem. 200, 81-88. MEDLINE

Diamond MI, Miner JN, Yoshinaga SK and Yamamoto KR. (1990). Science 249, 1266-1272. MEDLINE

Eadie JS, Conrad M, Toorchen D and Topal MD. (1984). Nature 308, 201-203. MEDLINE

Erickson LC, Laurent G, Sharkey NA and Kohn KW. (1980). Nature 288, 727-729. MEDLINE

Friedberg EC, Walker GC and Siede W. (1995). DNA Repair and Mutagenesis, . ASM Press, Washington, DC.,

Fritz G, Tano K, Mitra S and Kaina B. (1991). Mol. Cell. Biol. 11, 4660-4668. MEDLINE

Grombacher T, Mitra S and Kaina B. (1996). Carcinogenesis 17, 2329-2336. MEDLINE

Harris LC, Potter PM, Tano K, Shiota S, Mitra S and Brent TP. (1991). Nucleic Acids Res. 19, 6163-6167. MEDLINE

Heiss JD, Papavassilion E, Merrill MJ, Nieman L, Knightly JJ, Wallbridge S, Edwards NA and Oldfield EH. (1996). J. Clin. Invest. 98, 1400-1408. MEDLINE

Izumi T, Henner WD and Mitra S. (1996). Biochemistry 35, 14679-14683. MEDLINE

Klein-Hitpass L, Tsai SY, Weigel NL, Allan GF, Riley D, Rodriguez R, Schrader WT, Tsai MJ and O'Malley BW. (1990). Cell 60, 247-257. MEDLINE

Kokkinakis DM, Ahmed MM, Delgado R, Fruitwala MM, Mohiuddin M and Albores-Saavedra J. (1997). Cancer Res. 57, 5360-5368. MEDLINE

Leblanc B and Moss T. (1994). DNA-Protein Interactions: Principles and Protocols. Kneale GG (ed). Humana Press Inc: Totowa, NJ. , pp. 1-10.

Levin VA, Leibel SA and Gutin PH. (1997). Cancer: Principles and Practice of Oncology. DeVita Jr VT et al. (ed). Lippincott-Raven, Philadelphia, PA., pp. 2022-2082.

Liu L, Lee K, Wasan E and Gerson SL. (1996). Cancer Res. 56, 1880-1885. MEDLINE

Ludlum DB. (1990). Mutation Res. 233, 117-126. MEDLINE

McEwan IJ, Dahlman-Wright K, Amlot T, Ford J, Wright APH and Gustafsson J-A. (1995). Mutat. Res. 333, 15-22. MEDLINE

Mitra S and Kaina B. (1993). Prog. Nucl. Acids Res. Mol. Biol. 44, 109-141.

Myrnes B, Nordstrand K, Giercksky K, Sjunneskog C and Krokan H. (1984). Carcinogenesis 5, 1061-1064. MEDLINE

Nakatsu Y, Hattori K, Hayakawa H, Shimizu K and Sekiguchi M. (1993). Mutat. Res. 293, 119-132. MEDLINE

Patel SA, Graunke DM and Pieper RO. (1997). Mol. Cell. Biol. 17, 5813-5822. MEDLINE

Pegg AE. (1990). Cancer Res. 50, 6119-6129. MEDLINE

Pegg AE, Swenn K, Chae MY, Dolan ME and Moschel RC. (1995). Biochem. Pharmacol. 50, 1141-1148. MEDLINE

Pieper RO, Futscher BW, Dong Q, Ellis TM and Erickson LC. (1991). Cancer Commun. 2, 13-20.

Robins P, Harris AL, Goldsmith I and Lindahl T. (1983). Nucleic Acids Res. 11, 7743-7758. MEDLINE

Saatcioglu F, Claret F and Karin M. (1994). Sem. Cancer Biol. 5, 347-359.

Sagher D, Karrison T, Schwartz JL, Larson RA and Strauss B. (1989). Cancer Res. 49, 5339-5344. MEDLINE

Salmon SE and Cassady JR. (1989). Cancer: Principles and Practice of Oncology. DeVita Jr. VT, Hellman S and Rosenberg SA (eds). Lippincott-Raven, Philadelphia, PA., pp. 2344-2387.

Schold SC, Brent TP, von Hofe E, Friedman HS, Mitra S, Bigner DD, Swenberg JA and Kleihues P. (1989). J. Neurosurg. 70, 573-577. MEDLINE

Shapiro DJ, Sharp PA, Wahli WW and Keller MJ. (1988). DNA 7, 47-55. MEDLINE

Singer B and Gruenberger D. (1983). Molecular Biology of Mutagens and Carcinogens. Plenum Press: New York.,

Snow ET, Foote RS and Mitra S. (1984). J. Biol. Chem. 259, 8095-8100. MEDLINE

Tano K, Shiota S, Collier J, Foote RS and Mitra S. (1990). Proc. Natl. Acad. Sci. USA 87, 686-690. MEDLINE

Tsai M-J and O'Malley BW. (1994). Annu. Rev. Biochem. 63, 451-486. MEDLINE

Tsai SY, Srinivasan G, Allan GF, Thompson EB, O'Malley BW and Tsai MY. (1990). J. Biol. Chem. 265, 17055-17061. MEDLINE

Washington WJ, Foote RS, Dunn WC, Generoso WM and Mitra S. (1989). Mech. Aging and Dev. 48, 43-52.

Watts GS, Pieper RO, Costello JF, Peng YM, Dalton WS and Futscher BW. (1997). Mol. Cell. Biol. 17, 5612-5619. MEDLINE

Yarosh DB, Rice M, Day III RS, Foote RS and Mitra S. (1983). Mutat. Res. 131, 27-36.

Yung WKA. (1990). Curr. Opin. Oncol. 2, 673-678. MEDLINE

Figure 1 Effect of dexamethasone on hMGMT mRNA expression. (a) Northern blot analysis; one of three representative experiments. (b) Quantitation of hMGMT mRNA using scanning densitometry. The induction of the hMGMT mRNA-fold change with the control level defined as 1.0, expressed as the mean±s.e. of the ratio of the optical density of the signal for hMGMT mRNA versus 18S rRNA

Figure 2 The effect of dexamethasone on hMGMT protein levels. (a) Western blot analysis; one of three representative experiments using specific polyclonal anti-hMGMT antisera. Lane 1, HeLa S3 extract, Control; Lanes 2 - 7, HeLa S3 extract, dexamethasone (0.01 M) treated for various time periods. (b) Quantitation of hMGMT protein. The hybridization signals, as revealed by enhanced chemiluminescence, were quantified by scanning densitometry. The induction of hMGMT protein was expressed as the mean±s.e. of the ratio of dexamethasone-treated versus control protein signals

Figure 3 Dexamethasone-mediated protection of HeLa S3 cells from BCNU toxicity. Survival was measured by clonogenic assay. Open circle, filled circle, filled triangle and filled square , , and represent 0 (control), 10 nM, 50 nM and 100 nM dexamethasone treated cells, respectively. Percent survival was expressed as the mean±s.e. of three independent experiments

Figure 4 Dexamethasone-mediated induction of luciferase activity expressed from a hMGMT promoter. Schematic representation of DNA constructs containing various lengths of hMGMT promoter linked to the luciferase gene (pGL2 basic) as described in Materials and methods, along with their specific induction by dexamethasone in transiently transfected HeLa S3 cells. Fold of induction was calculated as the ratio of the luciferase activities from induced (dexamethasone treated) vs control (ethanol treated) cells and was expressed as mean±s.e. of three to six independent experiments. Numbers in plasmids names refer to the position of the first and the last nucleotides of the hMGMT gene relative to the transcription start site (+1). The positions of restriction sites and of GRE-1 and GRE-2 are also indicated

Figure 5 DNase I footprinting of glucocorticoid receptor binding sites in the promoter of the hMGMT gene. (a) The 174 bp BamHI - ApaI fragment (-978 to -805), ³²P-labeled with klenow was incubated with increasing concentration of bacterially expressed GR DNA-binding domain, subsequently digested with DNase I and analysed in an 8% denaturing polyacrylamide gel. Lanes 1 - 5 show DNase I digestion (lower strand) in the presence of a 0, 100, 250, 500 or 1000-fold excess of GR DNA binding domain, respectively; and the lanes 6 and 7 represent G and G+A Maxam - Gilbert sequencing reactions, respectively on fragment. (b) Sequences protected against DNase I digestion by the glucocorticoid receptors are underlined. (c) Comparison of the consensus GRE with the two putative GREs present in the promoter of hMGMT gene. N represents any of the four bases

Figure 6 hMGMT-GRE mediated stimulation of specific in vitro transcription. (a) Transcription assay as described in Materials and methods. Lanes 1 and 2, control pPRE₂TATA, in the absence and presence of supplemented hGR, respectively; lanes 3 and 4, pMGMT TATA, in the absence and presence of supplemented hGR. (b) Schematic diagram of the reporter plasmids and supplemented hGR-dependent fold increase in pPRE₂TATA and pMGMTTATA transcription, normalized against pAdML200 transcripts, calculated from (a). (c) Bar diagram showing hGR-dependent increase in transcription of plovTATA containing wild-type and mutant GRE's, normalized and expressed as mean±s.e. of five to six independent experiments. The symbols of wild-type and mutated plasmids are described in Materials and methods