Loss of Glucocorticoid Receptor Expression by DNA Methylation Prevents Glucocorticoid Induced Apoptosis in Human Small Cell Lung Cancer Cells

Human small cell lung cancer (SCLC) is highly aggressive, and quickly develops resistance to therapy. SCLC cells are typically insensitive to glucocorticoids due to impaired glucocorticoid receptor (GR) expression. This is important as we have previously shown that expression of a GR transgene induces cell death in-vitro, and inhibits tumor growth in-vivo. However, the underlying mechanism for loss of GR expression is unknown. The SCLC cell line, DMS79, has low GR expression, compared to non-SCLC cell lines and normal bronchial epithelial cells. Retroviral GR expression in DMS79 cells caused activation of the apoptotic pathway as evidenced by marked induction of caspase-3 activity. Methylation analysis of the GR promoter revealed some methylation in the 1D, and 1E promoters of the GR gene, however the ubiquitous constitutively active 1C promoter was heavily methylated. In the 1C promoter there was a highly significant increase in DNA methylation in a panel of 14 human SCLC cell lines compared to a mixed panel of GR expressing, and non-expressing cell lines, and to peripheral blood mononuclear cells. Furthermore, within the panel of SCLC cell lines there was a significant negative correlation seen between methylation of the 1C promoter, and GR protein expression. Reversal of GR gene methylation with DNA methyltransferase inhibition caused increased GR mRNA and protein expression in SCLC but not non-SCLC cells. This resulted in increased Gc sensitivity, decreased Bcl-2 expression and increased caspase-3 activity in SCLC cells. These data suggest that DNA methylation decreases GR gene expression in human SCLC cells, in a similar manner to that for conventional tumor suppressor genes.


Introduction
Changes in the epigenetic status of genes are important in many human diseases, but particularly in cancer [1]. Some of these are mediated by changes in the expression and activity of DNA methyltransferases, DMNT1, DNMT3A and DNMT3B [2,3]. Cancer cells characteristically have hypermethylated CpG islands associated with tumor suppressor genes [1], and DNMT1 is reported to be over-expressed in lung cancer patients who smoke [4]. Recent studies have identified that a tobacco smoke carcinogen, nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)1-butanone (also known as nicotine derived nitrosamine ketone; NNK) not only induces gene mutations, but also increases hypermethylation of multiple tumor suppressor gene promoters including cyclin dependent kinase inhibitor 2A (p16ink4a), death associated protein kinase 1 (dapk1), and retinoic acid receptor b (rarb) [4]. Smoking is the major risk factor for human small cell lung carcinoma, and smoking is itself associated with methylation of more than 20 tumor suppressor genes [5,6].
Many factors can affect glucocorticoid sensitivity, including genetic variation at the GR gene locus, and expression of interacting proteins [7][8][9][10]. We have previously shown that human SCLC cell lines are resistant to glucocorticoid (Gc) hormones and drugs and that this resistance is due to impaired GR expression [11,12]. Restoration of GR expression in the cells by transfection, or viral transduction is sufficient to restore Gc sensitivity [12]. However, more importantly, restoration of GR expression is also sufficient to induce apoptosis of SCLC cells both in-vitro [13], and also in a xenograft model [14]. This raises the possibility that GR is a novel tumor suppressor gene for SCLC, and that loss of GR expression is implicated in SCLC pathogenesis.
The GR gene (NR3C1) is a ubiquitously expressed, ligand activated transcription factor, a member of the nuclear receptor superfamily. There is a single copy gene on chromosome 5, but its structure is complex, and relatively little is known about how transcription is regulated from it [15]. Transcription is controlled by 9 promoters, each associated with an alternative transcription start site. Seven of these promoters are clustered together in a CpG island, a common feature of housekeeping genes [16]. All transcripts generate authentic full-length GR protein as the translation start site is located in the common exon 2.
The GR is activated by Gc hormones from the adrenal cortex, under tight control of the hypothalamic-pituitary-adrenal (HPA) axis. This axis is under feedback control at the hippocampus and hypothalamus through activation of GR protein expressed in these sites. Variability in rat HPA axis tone has been ascribed to changes in hippocampal GR expression, regulated by altered methylation of the rat GR exon 1-7 promoter [17]. This is located, as in the human, in the upstream CpG island. More recent studies have examined methylation status of a number of alternate human GR promoters in peripheral blood mononuclear cells [16]. These studies have revealed extensive, and variable methylation.
Hippocampal and hypothalamic GR plays a key role in the negative feedback control of the hypothalamic-pituitary-adrenal axis. Altered GR expression in the brain results in re-setting of the tone of this neuroendocrine axis and this is associated with altered methylation of nerve growth factor inducible-A (NGFI-A, or KROX, EGR1) binding site in the rat exon 1.7 promoter, homologous to the human exon 1F promoter [17][18][19]. This methylation controlled alteration in GR expression results in consequences for the whole organism, by regulating adrenal glucocorticoid production.
In this study we examined the pattern of GR promoter methylation in human SCLC cells, and showed that reversal of the methyl marks increased GR protein expression, and function. Moreover, we identified a negative correlation between GR 1C promoter methylation and GR protein expression across a panel of human SCLC cell lines, and human control cells.

Materials and Methods
Cell Culture and Maintenance A549 human lung epithelial carcinoma cells, HEK-293 human embryonic kidney cells, HeLa human cervical carcinoma cells and U20S human osteosarcoma cells (European Collection of Cell Cultures, Wiltshire, UK) were all cultured in DMEM (Invitrogen) supplemented with 10% fetal calf serum (FCS) (Invitrogen).
Non-small cell lung cancer lines NCI-H358 and -H727 (European Collection of Cell Cultures, Wiltshire, UK) and NCI-H23, -H441, -H1299 (American Type Culture Collection, USA) were grown in RPMI 1640 medium (Invitrogen) supplemented with 10% FCS and 10 mM HEPES as recommended by supplier.
Small Cell Lung Cancer (SCLC) cell lines used in this study were the 10'COR' cell lines Cor L24, Cor L27, Cor L31, Cor L32, Cor L42, Cor L47, Cor L51, Cor L88, Cor L99 and Cor L103. Also used were DMS 79, DMS 153, HC12 and HX 148. All of these cell lines were derived from patients with pathologically confirmed SCLC [20,21] with the exception of Cor L32, which was derived from a patient with poorly differentiated squamous carcinoma of the lung. This cell line did however exhibit the characteristics of SCLC [21]. All SCLC cell lines were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% FCS and 10 mM HEPES, as previously described [12].
Human peripheral blood mononuclear cells were obtained from healthy local donors, according to LREC approval 09/H1013/6. Normal human bronchial epithelium was harvested from lung resection specimens after surgery for lung cancer, with full, informed patient consent.

Treatments
Where appropriate, cells were treated with vehicle or 100 nM dexamethasone (Sigma-Aldrich) for 24 hours at 37uC, 5% C0 2 before lysis. In certain experiments, cells were incubated with vehicle or 5 mM 59Azadeoxycytidine (Sigma-Aldrich) for 72 hours or 5 days. After treating with 59Azadeoxycytidine some cells were treated for a further 72 hours with 100 nM dexamethasone.

Immunoblot analysis
Cells were lysed using 16 RIPA buffer (100 mM Tris base, 150 mM NaCl, 1% Igepal, 2.5% sodium deoxycholate, 1 mM EDTA).This buffer also contained protease inhibitors (Roche). Following centrifugation for 30 min at 10,000 g supernatants were harvested and assayed for protein content using a Bradford assay (Bio-Rad). Samples were then diluted in loading buffer [0.125 M TrisCl (pH 6.8), 0.1% sodium dodecyl sulphate (SDS), 20% glycerol, 0.2% b-mercaptoethanol, 0.001% bromophenolblue]. Cell extracts (50 mg) were then analysed by SDS-PAGE and western blotting as described [22,23]. Primary antibodies used were the mouse monoclonal anti-hGR (Clone 41, 1:2500), which binds in the N-terminal region of GR (BD Biosciences), rabbit polyclonal anti-hGR (P-20, 1:500) raised against a peptide mapping at the C-terminus of GRa (Santa Cruz) and the mouse monoclonal a-tubulin (1:5000) (Sigma-Aldrich). Secondary antibodies used were the horseradish peroxidase-conjugated anti-mouse (1:5000) and anti-rabbit (1:5000) (GE Healthcare). To quantify protein levels, densitometric analysis was carried out. Blots were scanned and the densitometry of each band was analysed using ImageJ software. The internal loading control (aTubulin) was also quantified and the results were normalised against these readings. This analysis was carried out on 3 separate blots for each experiment and the average calculated.

Reporter Gene Assays
Cells were co-transfected with 2 mg of a TAT3-luciferase construct [13] and 0.5 mg of a pCMV-Renilla Luciferase construct (to correct for transfection efficiency) using Fugene 6 (Roche). In some cases, cells were also transfected with 1 mg of a GR expression vector (pFUNC1-GR-eYFP) [13]. Cells were treated with dex as described before luciferase assays were carried out using the Dual Luciferase Reporter Assay System (Promega), as previously described [8].

Retroviral infections
pFUNC1-eYFP or pFUNC1-GR-eYFP were transfected into HEK293 for retroviral production using Fugene HD (Roche, UK), as a transfection reagent at a 3:2 reagent:DNA ratio. Infection of the retroviral particles into DMS 79 cells was performed as detailed before. Post infection, cells were cultured in normal growth media containing serum for 72 hours.
Images were collected on an Olympus BX51 upright microscope using a 606/1.40 UPlanApo objective and captured using a Coolsnap ES camera (Photometrics) through MetaVue Software (Molecular Devices). Specific band pass filter sets for DAPI, FITC and Texas red were used to prevent bleed through from one channel to the next. Images were processed and analysed using ImageJ (http://rsb.info.nih.gov/ij).

Sodium Bisulphite Sequencing
Genomic DNA was extracted from several of the cell lines used in this study, some of which had been treated with 59 Azadeoxycytidine (see results) using the DNeasy blood and tissue kit (Qiagen). Purified genomic DNA was then bisulphite converted using the Epitect Bisulphite Kit (Qiagen). This converted DNA was then used as a template in PCR reaction to amplify specific GR promoter regions. GR promoter regions amplified (and the primers used) were as follows: Promoter 1C (forward-59-AGGTGGATCCGGAAGGAGGTAGYGA-GAAAAGAAATT-39; reverse-59-AGGTGAATTCACACRA-ACTCRCAAAATAAAAAAAA-39).
All primers were supplied by MWG-Eurofins. PCR was carried out using Immolase Polymerase (Bioline), following the manufacturer's instructions. Cycling conditions were as follows: 95uC for 10 minutes, 30 cycles, at 95uC for 30 seconds, at 56uC for 45 seconds, and at 72uC for 30 seconds, this was followed by a final extension of 72uC for 10 minutes.
PCR products were electrophoresed on a poly-acrylamide gel and purified using a gel extraction kit (Qiagen). Ligations were then carried out using these PCR fragments using the pGEM-Teasy vector system (Promega) following the manufacturer's instructions. Following transformation into JM-109 competent bacteria (Promega), colonies were picked and incubated overnight in LB broth at 37uC. Mini-preps (Qiagen) were carried out on these cell suspensions, followed by a restriction digest using EcoRI (Roche) to confirm the presence of the correct insert. Plasmids containing the insert were then sent for sequencing using an SP16 primer at the DNA Sequencing Facility, University of Manchester. An initial study was carried out to analyse potential methylation and the effects of 59Azadeoxycytidine, using DNA from DMS 79 cells. In this case, one clone was analysed for each condition (each individual promoter, untreated or treated with 59Azadeoxycytidine). A more in depth study was carried out to analyse methylation of the 1C promoter in a panel of SCLC cell lines. In this case, 3 clones were analyzed per cell line.

Statistics
All statistical analysis was carried out using SPSS for Windows version 16. Further statistical analysis was carried out to compare

Results
Expression of GR is impaired in the human SCLC cell line, DMS79 GR protein was measured in the hSCLC cell line, DMS79 and compared with two Gc sensitive cells lines, HeLa and A549, and two Gc resistant cells lines, U2OS and HEK293. The SCLC cell line DMS79 expresses low levels of GR protein, similar to the U2OS cells, and clearly far less than the HeLa and A549 cells (Fig. 1a). Comparison with a panel of non-SCLC cell lines indicates that expression of GR is lower in the SCLC cells (Fig. 1b). GR protein was found to be expressed in normal human bronchial epithelium, harvested from lung tissue at resection of lung cancer (Fig. 1c).
GR migrated as two major species, and the relative abundance appeared to differ between cell types (Fig. 1a,b). This may reflect alternative splicing to the GRb isoform, or alternative translation start site usage [22][23][24]. To further characterize the species, immunoblots were repeated using a C terminal specific antibody that specifically recognizes GRa (Fig. 1d). The similar banding pattern observed with the two antibodies indicates both proteins share a common C terminal domain, suggesting that the protein diversity lies with alternative translation start site usage.

Restoration of GR expression induces apoptosis
In DMS79 cells expression of a GR-eYFP transgene by retroviral infection resulted in increased GR-eYFP protein (Fig. 2a) and apoptosis as evidenced by increased cleaved caspase-3 in the SCLC cells expressing the GR-eYFP (Fig. 2b). In these studies only a minority of cells are effectively transduced, hence the low abundance of expressed fusion protein in the pool of cells. In addition, expression of the transgene induced apoptosis, further reducing GR-eYFP concentration in pooled cell lysates [13,14].

Methylation status of GR promoter in SCLC cells
The decreased GR expression in SCLC cells may result from alteration in GR promoter methylation, or from an indirect mechanism. Therefore methylation status of the GR promoters in DMS79 cells was examined by bisulfite sequencing. We did not detect any DNA methylation in the promoters 1B, and 1F. The 1D and 1E promoter each had only a single methyl CpG, marked with an open box (Fig. 3a,b). In contrast the 1C promoter had four methyl CpGs (Fig. 3c). The observed methylated CpGs are located in potential transcription factor binding sites, (CpGs 6, 44 and 52), or lie close to such a site (CpG 69) (Fig. 3c).

SCLC cell lines have increased methylation of the GR 1C promoter
To determine if the observed changes in GR promoter methylation were found in other human SCLC cell lines, a panel of 14 hSCLC cells lines with varying GR expression were compared with two GR expressing cell lines, (A549 and HeLa) a non-expressing cell line, (U2OS), and peripheral blood mononuclear cells as a primary cell comparison (Fig. 4a,b). The beads represent CpG dinucleotides from position 1 to position 69. It is readily apparent that there is frequent methylation at the last two beads on the right, which represent CpG 68 and 69. This is present across the panel of SCLC cells, and is also seen in the control cells. However, there is also a marked increase in methylation at positions 1-67 across the SCLC panel, without any notable clustering. In contrast there is very little CpG methylation seen at positions 1-67 in the control cell panel (Fig. 4a,b).
Logistic regression was used to compare methylation between the SCLC panel, and all the control cells (Table 1). This revealed that there was a significant difference between groups with methylation at CpG 69 individually; and with CpG 68 and 69 in combination, and across the whole promoter region. A significant difference was also observed between groups for all CpGs excluding 68 and 69 (Table 1).

GR promoter methylation is correlated with GR expression
Across all the hSCLC cell lines there was a wide range of GR expression, although in all cases GR was less abundant than in the HeLa and A549 control cell lines (Fig. 5a, b). There was also a marked variation in the abundance of the two GR protein species across the cell lines examined, but for this analysis both GR protein isoforms were considered together. The relationship between GR 1C promoter methylation and GR protein expression was examined across the whole panel of hSCLC cell lines. There was a significant correlation between number of methylated CpGs, and the GR protein expression within the panel of hSCLC cell lines; r 2 = 0.54, and p = 0.01 (Fig. 5c).

Restoration of GR expression by reversal of DNA methylation
If the loss of GR expression in DMS79 cells resulted from DNA methylation, perhaps consequent to carcinogenesis, GR expression would be predicted to increase following reversal of methylation. After 72 hours of treatment with 59Azadeoxycytidine, GR mRNA levels increased dramatically in DMS79 cells when compared to HEK and A549 cells where no changes in expression were seen (Fig. 6a). Furthermore, after 72 hours, or after 5 days incubation with 59Azadeoxycytidine, there was a striking increase in GR protein (Fig. 6b). In contrast to the increase in GR expression seen in DMS79, there was decreased expression in U2OS cells, possibly due to toxicity (Fig. 6b).
The effect of 59Azadeoxycytidine was then compared in human SCLC cell lines and non-SCLC cell lines to determine if the regulation of GR expression by methylation was widespread in lung cancer. Three of the four SCLC cell lines showed augmented GR expression after 59Azadeoxycytidine treatment, in comparison to only one of the four non-SCLC cell lines (Fig. 6c).
The increase in GR expression seen in DMS79 cells is predicted to restore Gc sensitivity to these cells. Indeed, there was increased Gc transactivation of a simple reporter gene (Fig. 7a). The magnitude of Gc induction was considerably less than that seen with GR overexpression, but the latter does result in supraphysiological GR concentrations in target cells (Fig. 7a).
The increased GR expression in response to treatment with 59Azadeoxycytidine resulted in suppression of the pro-survival Bcl-2 gene in both HEK, and DMS79 cells (Fig. 7b). There was also increased cleaved caspase-3 activation in the SCLC cells but not the non-SCLC cells (Fig. 7c & 7d). It is important to note that after 5 days incubation with 59Azadeoxycytidine there was induction of cell death, possibly due to reacquisition of GR expression, or regulation of other survival genes.

Discussion
Epigenetic disorders are implicated in many human diseases including cancer. Cigarette smoke is the principal environmental agent promoting SCLC and recently, a cigarette smoke carcinogen, NNK, has been shown to exert a specific effect promoting DNMT1 expression, and activity [4]. Therefore this carcinogen may act by inducing epigenetic changes in key genes implicated in the pathogenesis of the cancer.
We have previously shown that the glucocorticoid receptor is effectively a tumor suppressor gene for SCLC, whose expression is inhibited [12]. We have extended the analysis in this study with further evidence that the GR expression is lower in a panel of SCLC cells compared to non-SCLC cells or normal bronchial epithelial cells. Reacquisition of GR expression, and glucocorticoid sensitivity in SCLC cells in vitro, or in xenograft tumors results in apoptosis [13,14]. GR is expressed in most cells and tissues, and exerts pleiotropic effects on cells. However, relatively little is known of how GR gene expression is regulated, as it has an unusually complex promoter structure. More recent studies in primary peripheral blood mononuclear cells defined highly individual patterns of GR promoter methylation [16]. This work also mapped potential, and proven transcription factor binding sites within the promoters, the majority of which contained CpG sites which could be methylated. This suggested that differential methylation of the human GR promoter is a mechanism to control promoter utilization, and so GR gene expression.
Three GR promoters were examined in the index cell line, DMS79, and all three contained methylated CpG sites. Further analysis was performed on the 1C promoter as it has been shown to be ubiquitously expressed in all tissues tested, including lung [15]. There was an obvious and highly significant increase in CpG methylation across a panel of hSCLC cell lines compared to the non-SCLC cell line A549. Importantly there was no increase in the two non-expressing control cell lines (U2OS and HEK293), indicating that 1C promoter methylation is not required to limit GR expression, but that other mechanisms may apply to regulate GR expression in different cell types. A broader comparison between hSCLC cells and a heterogeneous panel of GR expressing, and non-expressing cell lines also found a highly significant increase in methylated CpG sites in the hSCLC cells. It was striking that methylation of CpG68 and 69 was far higher across all the cell lines examined than positions ,68. CpG68 lies within a consensus SP-1 transcription factor binding element, and CpG69 lies immediately 39 to it. There was no other positional clustering of the methylated CpG sites, and statistical analysis did not identify any significant positional effect. Some of the methylated CpG sites at ,68 overlie consensus transcription factor binding sites (CpG6, 44 and 52), but CpGs1,2 and 3 did not.
Within the panel of hSCLC cells there was variation in GR protein expression, which was quantitated by immunoblot under non-ligand treated conditions. If GR promoter 1C methylation played a role in regulating GR expression a correlation would be expected between degree of methylation, and protein expression. Indeed, a significant negative correlation was seen.  Studies were aimed at discovering if DNA methylation could regulate GR expression in hSCLC cells, in a similar manner to that defined in the rat brain. Treatment with 59Azadeoxycytidine resulted in a striking induction in GR mRNA and protein in SCLC cells, although, as expected, there was heterogeneity amongst the cell lines profiled, with one cell line (COR L24) showing no induction. Two confounding issues in these studies are that use of 59azadeoxycytidine exerts a toxic effect on the SCLC cells, likely as a consequence of regulating multiple tumour suppressor genes and also that increased GR protein expression results in increased cell death in SCLC cells [13,14]. In contrast, augmented GR expression was only seen in one of four non-SCLC cell lines. Interestingly, in another GR deficient cell line, U2OS, there was no increase in GR expression, suggesting that loss of GR expression by promoter methylation may be specific to SCLC. Analysis of GR mRNA expression showed no induction in two further cell lines, suggesting a cell-type specific effect. The removal of methyl marks with 59Azadeoxycytidine resulted in restoration of glucocorticoid regulation of a simple reporter gene, confirming expression of a functional receptor and increased glucocorticoid effects. The increase in GR expression was accompanied by reacquisition of inhibition of Bcl-2, a key component of the antiapoptotic pathway and an increase in cleaved caspase-3, indicative of apoptosis. This suggests that removal of methyl marks from the GR results in apoptosis although changes in other tumor suppressor genes are also likely to play a role in this effect.
Taken together our data support a role for DNA methylation in regulating GR gene expression in human SCLC cells. This mechanism is proposed to explain the reduction in GR expression observed, and the consequent loss of Gc sensitivity. There is a clear survival advantage for the cancer cells from evading GR integrity, as restoring GR expression induces apoptosis. Indeed in our study, treatment with 59Azadeoxycytidine itself resulted in reacquisition of GR expression, which was followed by specific SCLC cell death. It seems possible that exposure to the tobacco smoke carcinogen NNK may play a role in silencing the GR gene by increasing DNMT1 mediated methylation of the GR promoters lying within the 59CpG island and thus reducing the role of GR in regulating apoptosis in the SCLC cells.