Epigenetic alteration by DNA-demethylating treatment restores apoptotic response to glucocorticoids in dexamethasone-resistant human malignant lymphoid cells

Brief exposure to the genomic DNA demethylating agent AZA restores the GC-dependent apoptotic response in each of three cell lines

We evaluated the contribution of the epigenetic state to GC resistance in three commonly used model systems of human lymphoid hematologic malignancies: 1) CEM-C1-15, a GC-resistant clone of the pediatric ALL cell line CCRF-CEM; 2) Molt-4, an uncloned T-cell derived pediatric ALL cell line; and 3) RPMI 8226, an uncloned myeloma line (B-cell lineage). The cells of each system contain functional GR; yet each is highly resistant to GC-evoked apoptosis [26, 27]. Initially, we treated each system with AZA for 24 h; then added Dex and followed the cultures over time. There was significant near-term restoration of sensitivity to Dex-driven apoptosis in each system. Thus, the % reduction in live cells treated with AZA 24 h followed by Dex for 72 h, compared to 72 h Dex only, was 87% for MOLT-4 (n = 4, each in triplicate), 74% for RPMI 8226 (n = 2, in triplicate) and 86% for CEM C1-15 (n = 3, in triplicate). Since AZA incorporation into replicating DNA results in multiple phenotypes, possibly not entirely random [28, 29], one could not expect every cell in the treated populations to convert to Dex sensitivity; so it is striking that in repeated experiments, such treatment always resulted in a near-term shift to sensitivity in a significant proportion of the population. In all three systems, phase-contrast visualization of the cells showed them to undergo the classic progression of apoptosis: cell shrinkage, nuclear condensation, karyorrhexis, and eventually, cell lysis. To estimate how many cells of each system were stably affected, we cloned each population after exposure to AZA or DMSO vehicle for 48 hours. About 1 month later, in a screening process, many clones were evaluated individually for Dex sensitivity. Because in some lymphoid cell systems Dex can arrest cell growth without causing apoptosis, we evaluated both repression of growth (total reduction in viable cells, and intact, yet apoptotic cells remaining at the end of the treatments. For this relatively high-throughput screen, we used trypan blue dye exclusion as a working definition of viability. Figures 1, 2 and 3 show results for CEM, MOLT-4 and RPMI 8226 cells, respectively. Panels 1A, 2A, and 3A show the data evaluating many individual clones as soon as clone outgrowth was sufficient for the assays. Each dot represents one clone, with % repression of growth on the x-axis, and % apoptotic (trypan blue positive) but still intact cells on the y-axis. While this analysis allows a rough estimate of the separation between growth retardation and ongoing apoptosis, it can underestimate the total quantity of cells lysed before the evaluations, since these are not “seen” by the instrument used. Nevertheless, the data in Figures 1A-3A show that in early clones from resistant cells pre-treated only with DMSO (open symbols), Dex caused varying degrees of growth repression with relatively few apoptotic cells to be found after 96 h. In contrast, in the clones derived from AZA treated cultures (closed symbols) Dex treatment often caused growth arrest plus a significant percentage of apoptotic cells at 96 h, suggesting ongoing apoptosis. Extensive growth arrest and Dex-dependent apoptosis generally were correlated in cells from all three lines pre-treated with AZA, consistent with our previous observations on the cell-cycle effects of Dex on sensitive CEM cells [30]. We conclude that the AZA effect is widespread, due to effects on many individual cells, and that the acute sensitization to Dex may sometimes persist.

CEM cells (Figure 1). Visual and Vi-cell evaluation of the CEM-C1-15 cell clones indicated that no clone in this experiment treated with DMSO alone reverted to GC-mediated apoptotic sensitivity. On the other hand, many clones from AZA-treated C1-15 cells showed high Dex sensitivity. Clone IV B9 (extreme upper right, Figure 1A) derived after AZA, showed a striking reversion to Dex-sensitivity, and it remained phenotypically stable over 4 months in continuous culture and after repeated freeze-thawing; therefore it was taken for further evaluation. Cytosine extension assay results (Figure 1B) indicated that clone IV B9 DNA was 4-fold hypomethylated relative to CEM-C1-15 cells. High throughput DNA sequencing showed regions of reduced methylation in DNA in all autosomes, relative to the resistant CEM C1-15 cells. From CEM C1-15, CEM C7-14 and revertant IV B9 cells, non-methylated and methylated DNA were separated and sequenced (Methods). The average coverages for non-overlapping 5,000 bp windows were calculated across all chromosomes, examining all DNA and also only unique gene locations. Coverage comparisons of C1-15 and C7-14 sequences enriched for methylated DNA showed that many C1-15 sequences contained regions- both repeat sequences and unique sequences- with high methylation that were not found in the C7-14 data. Similarly, a plot of sensitive revertant IV B9 against its resistant parent C1-15 revealed loss of methylation at many sites in IV B9.

MOLT-4 cells. Of the 3 cell systems, MOLT-4 showed the greatest acute response as a mass culture (Figure 2 and data cited in text). Clearly, there also is a shift to more GC response in the demethylated clones from MOLT-4 cells (Figure 2A). Upon extended culture, the Dex-sensitive phenotype of these clones proved to be unstable, however. The data in Figure 2B-2D therefore evaluate the Molt-4 mass culture following acute treatment with AZA and Dex. Test of the DNA after 48 hours of exposure to AZA confirmed demethylation (Figure 2B). Treatment of the cells with AZA for 24 hours followed by the addition of Dex resulted in a marked loss of cell viability over time (Figure 2C). After treatment with AZA only, there was a 20-25% reduction in viable cells (% culture viability, y-axis) Dex alone reduced viability by only about 5%. AZA followed by Dex, however, reduced viability in this experiment by 50%. These results support the conclusion that an epigenetic state is responsible, at least in part, for the GC resistance of many Molt-4 cells, though the AZA/Dex synergy could include additional mechanisms. Flow cytometry also indicated that the mass culture of MOLT-4 cells treated with AZA had undergone some cell loss (Figure 2D lower left panel and Table 1) However, there was a drastic reduction of intact cells after treatment with AZA followed by Dex, and an increase of cell debris, coupled with a reduction in total viable cells, resulting in a reduction in cells available for evaluation by flow cytometry. (Figure 2D, lower right panel. The scale for all histograms in Figure 2D is identical.) The cytometric evaluation also showed a shift of the residual population into G1, typical of the apoptotic effect of Dex in many lymphoid cell systems, e.g. CEM [30]. Since phase-contrast microscopy had shown these cells to go through the morphologic phases of apoptosis, we conclude that at this 96 hr time point, many cells in the culture had gone on to secondary necrosis/lysis.

RPMI cells (Figure 3) The mass culture of RPMI 8226 myeloma cells also became sensitive to Dex-induced lysis after treatment with AZA. Many outgrown clones showed significant growth arrest and ongoing apoptosis (Figure 3A). The most sensitive clone (II E6) was stable in phenotype and was kept for further analysis. In this clone, cytosine extension assay did not detect a statistically significant shift in total DNA methylation (Figure 3B). Exposure of clone II E6 to Dex revealed a significant loss of viable cells by 48 hr, comparable to CEM clone IV B9, and faster than CEM clones C1-6 and C7-14 (Compare Figures 1C and 3C). Flow cytometry also showed that under control conditions, clone II E6 sustained an ongoing low level of apoptosis, presumably a result of the earlier AZA treatment (lower left panel, Figure 3D and Table 1) After 96 hr in Dex, clone II E6 showed a near complete loss of viable cells, and only a few remaining cells had sub-diploid DNA content (Figure 3D Note, the scale of the Y-axis in all panels of 3D is identical, and see Table 1).

Clonogenicity is lost in RPMI 8226 cells treated with AZA and Dex

The data of Figures 1 , 2 and 3 and Tables 1 and 2 demonstrate that in three cell systems, AZA treatment restores sensitivity to Dex-induced cell death. In an initial exploration of mechanisms, we examined the effects of the AZA treatment on some known important elements in Dex-dependent apoptotic pathways.

GC-stimulated events involving p38 MAPK , a feed- forward stimulator of GR action, are important for glucocorticoid-dependent apoptosis in all 3 cell systems

GC-stimulated p38 MAPK activation plays a major role in the glucocorticoid-dependent apoptotic mechanism in a variety of malignant lymphoid cell lines, including CEM, Molt-4 and RPMI 8226 [17, 18, 27, 31–33]. Elevated p38 relative to JNK and ERK MAPKs favors an apoptotic response in a variety of malignant lymphoid cells [32]. The GR is phosphorylated by p38 (and other kinases) at ser 211, which enhances GR transcriptional and apoptotic activities [17, 18, 27, 33]. Comparisons of the resistant parent CEM-C1-15, its demethylated clone IV B9, and the spontaneous revertant CEM-C1-6 revealed a strong increase in phosphorylated (i.e., activated) p38 MAPK in response to Dex stimulus in both IV B9 and C1-6 cells, and at best, a very slight increase in C1-15 cells. (Figure 4). Levels of total p38 MAPK protein remained unchanged, consistent with earlier reports [17, 18]. To test the importance of the GC-stimulated p38 activity for apoptosis, the demethylated, stable revertants CEM IV B9 and RPMI II E6 were treated with the p38 inhibitor SB203580 for 1 hour before the addition of 1 μM Dex, and viable cells were enumerated 48 h later.With the SB added to block p38, there was statistically significant protection from Dex-driven apoptosis for the SB-treated cells in both systems, with an increase in total viable cells after Dex (Figure 5). In the SB-treated cultures, there was also a statistically significant increase in the % of viable cells in the cell populations still present at the end of the experiment. This fraction increased from 33% (Dex only) to 62% (SB + Dex) (p = 0.001) for IV B9 cells, and in II E6 cells, from 48% to 68% (p = 0.003). These results are consistent with partial inhibition of the Dex-induced ongoing apoptosis. Also, in Molt-4 cells, the uncloned, mass culture system, treated acutely with AZA, followed by Dex with or without SB, modest, but statistically significant protection was seen when the p38 blocker was added: AZA alone reduced culture viability by 25%; AZA followed by Dex reduced viability by 60%. There was a statistically significant 30% gain in viable cells when SB was added to the Dex + AZA treatment: p = 0.02, n = 3. This partial inhibition of cell death in all three systems is as expected, since ser211 phosphorylation of the GR enhances, but is not essential for GR actions. The data above identify p38 as one possible AZA-sensitive node shared by all three cell systems in the known complex of pathways initiated by Dex that lead to apoptotic cell death. GR is known to affect p38 function and vice versa. We therefore further examined aspects of GR function. Both sufficient quantity and activity of the GR are known to be essential for it to drive lymphoid cell apoptosis.

GR functions are enhanced by differing mechanisms in CEM and RPMI 8226 cells

The p38 effects suggest one node of common action, highly relevant to GR activity, in the AZA-reverted cells of all three systems. However, considering the complex networks of interacting cellular signal systems, it is not surprising that additional, cell system-specific, non-overlapping effects influencing GR levels and activity were found. In the CEM cell system we observed: enhanced GR function, induction of GR protein and GR phosphorylation, and use of GR alternative promoters. Figure 6A shows that Dex induced GR protein in naturally sensitive clone C7-14 cells, in C1-6, the spontaneous revertant to sensitivity from C1-15 cells, and in the revertant clone IV-B9, obtained from C1-15 cells after AZA treatment. Dex did not induce GR protein in C1-15 cells. In C1-15 cells, there was lower basal GR ser211 phosphorylation and only weak induction, compared to the other three, Dex-sensitive cell clones. GR-driven transcriptional activity was consistent with these results. Cells were transfected with a GRE-driven luciferase reporter construct (Methods) and treated with Dex. There was highly increased GR transcriptional activity in both revertant clones C1-6 and IV B9 compared to C1-15 (Figure 6B). Revertant IV B9 showed at least 2-fold elevated activity over that of the spontaneous revertant C1-6, consistent with the cell death and GC dose-response data in Figure 1. It appears that the phosphorylation state of the GR at S211, total GR protein levels, and transactivation potential can be controlled by epigenetic mechanisms in this system.

Considering the Dex-dependent auto-induction of GR in these cells, and the correlation between such induction and apoptosis that has been reported [34], we evaluated the promoter-specific, basal and Dex-driven expression of the GR in revertant clone IV B9, compared to other relevant CEM clones. There are 11 untranslated exon 1 sequences at the 5’ end of the human GR [35], arising from specific human GR gene promoters. The promoter (s) for transcripts 1A and 1I are ~25 kbp upstream of exon 2, the first peptide coding exon and do not lie in a CpG island. The remaining promoters are proximal, located in the GC rich ~5 kb directly upstream from Exon 2 [36]. Hypothesizing that expression from 1A would be unaffected by AZA treatment, while transcription from one or more of the proximal promoters would be increased, we measured transcripts from GR exon 1A3, a T-cell selective promoter, and from exons 1D and 1C, two of the GR’s principal promoters in the proximal CpG-rich zone. Though transcription from promoter 1A3 is strongly induced in CEM C7 cells, this promoter is neither induced nor even expressed well in spontaneously resistant CEM clones ICR 27, C1-15, or in sensitive IV B9 cells. Dex-dependent induction of transcription did not occur from GR promoters 1C and 1D in resistant C1-15 cells, or in a spontaneously resistant CEM clone, ICR 27. This induction was seen, however, in the Dex-sensitive C1-15 revertant IV B9 (Figure 7). Limited pyrosequencing experiments could not confirm altered methylation of the proximal GR promoter. Pyrosequencing of a few short regions of the GR proximal promoter region showed a tendency for greater methylation at one or two positions in CEM C1-15 cells as compared to C7-14 cells (data not shown); however. pyrosequencing of the full 5000 bp promoter was beyond the resources available for this study. It will require further experiments in the future to clarify whether the GR promoter methylation state directly accounts for the shift in promoter usage or whether this is an effect secondary to alterations elsewhere in the genome.

RPMI 8226 myeloma cells: The GC-sensitive revertant II E6 has altered expression of several genes known to affect GR-driven transcription

We also compared expression from three GR promoters in Dex resistant RPMI 8226 cells and their stably sensitive clone II E6. Transcription from promoter 1A3 was virtually nil, whereas there clearly was transcription from 1D and 1C (Figure 7). None of the promoters was induced by Dex in these cells. However, other mechanisms known to bolster GR actions were altered, viz. the expression levels and GC inducibility of the genes for 3 of 5 coregulatory proteins known to affect GC actions in lymphoid cells [37–39]. Two, c-Myb and PGC-1a (PGC-1 alpha); act as GR coactivators; two as corepressors- PU.1 and Tel. Spi-B, the fifth coregulator, seems to interact with GR function, but without a clearly defined role as yet. None of these showed a clear pattern that seemed relevant to the CEM revertant (data not shown), but in the RPMI 8226 cell system, three of the five were altered so as to enhance GR activity (Figure 8). Similar basal levels of coactivator c-Myb were found in RPMI 8226 cells and its sensitive revertant II E6, but expression was significantly reduced by Dex only in the resistant 8226 cells (Figure 8 right hand bar graphs and inset). Coactivator PGC-1a also showed similar basal levels in parent and revertant but was strongly induced by Dex in the sensitive revertant. Corepressor PU.1 did not respond to Dex, but its level in the sensitive revertant cells was only about 1% of that in the resistant RPMI 8226 cells. Tel and Sp1B levels did not correlate with cellular response to Dex. Interestingly, the regulation of coactivators and corepressors plays a role in GR action in the pre-B-ALL 697 cell line [38, 39]. Thus, in two cell lines in the B-cell lineage (RPMI 8226 and 697) the regulation of coactivators and corepressors is important for GC action. In sum, expression levels and/or Dex regulation of three genes coregulatory for GR actions are altered in the Dex-sensitive RPMI 8226 revertant clone in a manner consistent with restoration of Dex sensitivity through enhanced GR actions.

Reagents

Unless stated otherwise all chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Cell culture and treatment

CEM cell clones were derived from the original cell line CEM-CCRF, which was developed from cells of a pediatric patient with T-cell ALL. CEM-C1-15 is a Dex-resistant sub-clone obtained without selective pressure from its parental resistant clone, CEM-C1. CEM-C1-6 is a spontaneous revertant to sensitive from CEM-C1 [27]. Clones CEM-IV B9 and RPMI II E6 were chosen from a subpopulation of CEM-C1-15 or RPMI 8226 cells treated with AZA. CEM-C7-14 is a subclone of CEM-C7; both are naturally Dex-sensitive. Molt-4 and RPMI 8226 cells were from the American Type Tissue Collection (ATCC, Manassas, VA). CEM cells were grown in RPMI 1640 (Cellgro, Mediatech Inc. Herndon, VA) supplemented with 5% fetal bovine serum (FBS) (Atlanta Biologicals, Lawrenceville, GA). Both Molt-4 and RPMI 8226 cells were propagated in RPMI 1640 supplemented with 10% FBS, 10 mM Hepes buffer (Cellgro), and 1 mM sodium pyruvate. All cells were grown in a tissue culture incubator in a humidified atmosphere at 37°C, 95% air / 5% CO₂ and subcultured regularly to ensure logarithmic growth. To evaluate Dex sensitivity, cells were treated with either ethanol vehicle (control) or concentrations of Dex ranging from 1 nM to 4 μM. For experiments examining the contribution of p38 MAPK, cells were pre-treated for 1 hour with DMSO vehicle or 1 μg/ml SB (p38 inhibitor) for CEM cells or 2 μg/ml SB for Molt-4 and RPMI 8226 cells before adding 1 μM Dex. Cells were enumerated and sized, with viability determined by Trypan blue exclusion assay, using a Vi-cell automated counter (Beckman Coulter, Fullerton, CA). For evaluation of acute demethylation, Molt-4 cells were exposed to 100 nM AZA for 24 hours prior to adding 1 μM Dex.

Derivation of clones and analysis of Dex sensitivityciv

Preliminary dose-response experiments established the optimal concentration of AZA for production of Dex sensitivity. Consequently, CEM-C1-15, Molt-4, or RPMI 8226 cells were incubated for 48 hours in 100 nM AZA or an equal volume of its DMSO vehicle for controls, after which time the cells were pelleted by centrifugation (Beckman Coulter, Allegra 6R table-top centrifuge) at 200 × g for 5 minutes at room temperature, their supernatant medium aspirated, and resuspended in 10 mls of sterile phosphate-buffered saline (PBS) (Cellgro). Cells were pelleted as before, resuspended in fresh growth medium without AZA, and allowed to recover for 5 days. The samples were cloned by serial dilution to single cells in RPMI 1640 supplemented with 10% FBS and an insulin-transferin-selenium supplement cocktail. Cloning medium for Molt-4 and RPMI cells also contained 10 mM Hepes and 1 mM sodium pyruvate. Approximately 1 month after cloning, cell populations from each clone were grown to about 10⁵ cells per ml and evaluated for Dex sensitivity by exposure ± 1 μM Dex for 96 hours. At that time, cells were counted and evaluated for Trypan Blue dye exclusion, size and morphology (Vi-cell, Beckman). Two calculations were performed: 1.% apoptosis in the residual whole-cell population = % Trypan Blue positive cells in Dex-treated population munus % TB positive cells in Vehicle-treated Control population; and 2. “growth represson” = (100-% Trypan Blue neg cells/total cells, Dex/Control).

Analysis of cell death

The Guidelines published in 2009 in Cell Death and Differentiation were followed: “…investigators need to combine at least two distinct methods,…and perform, whenever possible, long-term survival assays…” [43]. Accordingly, cells were evaluated in 4 ways- by phase contrast light microscopy, by vital dye exclusion and cell size- with on-screen visualization of cultures (Vi-cell, Beckman), by flow cytometry, and by multiday cell cultures following total and viable cells over time.

Clonogenicity

RPMI 8226 cells were incubated with 100 nM AZA or DMSO vehicle for 40 hours before addition of ethanol vehicle or 1 μM Dex for an additional 96 hours. Cells were then counted and immediately serially diluted in multi-well dishes to initial average densities of 1000, 100, 10, or single cells per well. After 2 weeks they were evaluated by Trypan blue exclusion assay for wells with live cells and viable cells per well enumerated.

Cytosine extension assay

The methylation state of cytosines within genomic DNA was analyzed using a cytosine extension method [59]. Cellular genomic DNA was extracted from logarithmically growing CEM-C1-15, CEM-IV B9, RPMI 8226, RPMI-II E6, or Molt-4 cells treated with DMSO vehicle or 100 nM AZA for 48 hours using a Qiagen DNeasy mini-prep kit (Qiagen, Valencia, CA). Residual RNA contamination was eliminated by digestion with 400 μg of RNase A. Purified DNA was quantified and 2 μg was digested overnight at 37°C using 8 units of the methylation-sensitive restriction enzyme Hpa II (New England Biolabs, Ipswich, MA). Extension was performed by combining 0.25 μg of digested DNA in the presence of ³H deoxycytidine triphosphate (dCTP) (53.0 Ci/mmol, Perkin Elmer, Waltham, MA), 0.25 units of TAQ polymerase (New England Biolabs), and buffer supplied with the enzyme for 1 hour at 56°C. Samples were subsequently placed on ice and duplicate 10 μl aliquots were bound onto Whatman DE-81 ion exchange filter papers (Whatman Inc., Florham Park, NJ) using a vacuum manifold. Filters were washed 3 times with 0.5 ml room temperature PBS, air dried, submerged in 3 mls of Scintiverse scintillation fluid (Fisher Scientific, Fair Lawn, NJ), vigorously vortex mixed for 15 seconds, allowed to recover for 1 hour, and evaluated using a Beckman scintillation counter (Beckman Coulter).

Propidium iodide staining and flow cytometry

CEM-C7-14, CEM-C1-6, CEM-C1-15, CEM-IV B9, RPMI 8226, and RPMI-II E6 cells were diluted to a starting density of 5×10⁴ viable cells/ml and treated with ethanol vehicle (control) or 1 μM Dex for 96 hours. Molt-4 cells were diluted to 5×10⁴ viable cells/ml and treated with ethanol/DMSO vehicle (control), 1 μM Dex alone, or pre-treated for 24 hours with 100 nM AZA after which time 1 μM Dex or ethanol vehicle was added for an additional 72 hours. Cell nuclear suspensions were prepared, stained by PI, and analyzed by flow cytometry in the UTMB Cytometry Core Lab as described [18].

Quantitative, real-time PCR of mRNA transcripts

TaqMan® real-time PCR assay and Taqman One-Step RT-PCR reagents (Applied Biosystems, Foster City, CA) were used to detect mRNA transcripts. The primers and Taqman probes used for hGR transcripts (1A3, 1C and 1D), c-Myb and PU.1 were described previously [52]. For measuring PGC-1a mRNA transcripts, 5′-AGAGACAAATGCACCTCCAAAAA-3′ (forward), 5′-6-FAM-AAGTCCCACACACAGTCGCAGTCACAA-TAMRA-3′ (Taqman probe) and 5′-AAAGTTGTTGGTTTGGCTTGTAAGT -3′ (reverse) were synthesized (Integrated DNA Technologies, Inc., San Diego, CA). The RT-qPCR reactions were measured on an ABI Prism 7900H Sequence Detection system (Applied Biosystems) and the results were normalized by endogenous 18S rRNA of each sample. The experiment was repeated and the results were pooled for statistical significance analysis (paired-sample t-test). Less than 5% (p < 0.05) of possible difference was considered significant.

Immunochemical analysis

Logarithmically growing CEM-C7-14, CEM-C1-6, CEM-C1-15, and CEM-IV B9 cells were diluted to a starting density of 4×10⁵ viable cells/ml and treated for 16 hours with 100 nM Dex (GR blot), 1 μM Dex (p38 blot), or ethanol/DMSO vehicle (control). Evaluation of GR phospho-S211, total GR, phospho-p38, total p38, and actin by immunoblot was carried out as previously described [17, 18].

GR transactivation assay

Logarithmically growing CEM-C1-6, CEM-C1-15, or CEM-IV B9 cells were collected by centrifugation, washed with 10 ml of sterile 37°C PBS and recollected. The cells were resuspended to a density of 1×10⁷ viable cells/ml in serum-free 37°C RPMI 1640 containing 1.25% DMSO. Aliquots (400 μl) of the suspension were placed into 0.4 cm gap electroporation cuvettes (Bio-Rad, Hercules, CA) containing 15 μg of GRE-dependant mouse mammary tumor virus (MMTV) luciferase reporter vector phh-Luc (ATCC) prepared using a Qiagen maxi-prep kit (Qiagen). Cuvettes were electroporated using 975 μF and 270 V with a Gene Pulser II (Bio-Rad). Electroporated cells were diluted in 4 ml (per cuvette) of RPMI 1640 supplemented with 5% FBS and 1.25% DMSO and recultured. Twenty-four hours after electroporation, cellular debris was removed by filtering through a 70 μm nylon cell strainer (BD Biosciences (Falcon), Two Oak Park, Bedford, MA) and cells were pelleted, resuspended in fresh RPMI 1640 containing 5% FBS, and viable cells were quantified by hemacytometer using Trypan blue exclusion. The cultures were diluted to 2.5×10⁵ viable cells/ml in RPMI 1640 supplemented with 5% FBS, treated with ethanol vehicle (control) or 1 μM Dex, divided into 500 μl triplicate aliquots for each treatment, and placed in a 48-well tissue culture plate (Costar, Cambridge, MA). Twenty-four hours after Dex treatment, cells were pelleted and washed with PBS. Luciferase activity was evaluated by lysing cell pellets in reporter lysis buffer (Promega, Madison, WI), 1 round of freeze-thaw at -80°C to 37°C, and activation of enzymatic activity using luciferase substrate (Promega). Relative light units (RLU) were read using a Monolight luminometer (BD Biosciences). RLUs were normalized to μg of protein in the cellular lysate through use of a BCA protein assay (Pierce, Rockford, IL).

High Throughput DNA sequencing

Total DNA from cells of C1-15, C7-14, and post-demethylation clone C1-15 IV B9 was fragmented by sonication into pieces of about 200 bp. DNA preparations were obtained representing DNA enriched for non-methylated or methylated by an immunoprecipitation procedure according to manufacturer’s protocols. . These fractions were sequenced in 36 bp fragments (reads) using an Illumina GA IIx instrument, following the manufacturer’s protocols. Quality assessment, trimming, and filtration of low quality reads were performed using our Slim-Filter software application [60]. The resulting high quality 31 bp reads were mapped to the reference human genome, build 37.3, using our in-house software. The ratios of experimental to theoretical reads at each position were computed to “normalize” the results, allowing comparisons of samples. The relative methylation status of all autosomes was determined, and the coverage across the chromosomes was calculated using a non-overlapping window size of 5000 bp. The raw sequencing data have been deposited as FASTQ files in SRA data base, identifiable under PRJNA209577. For more specific data, please contact the Corresponding Author.

Pyrosequencing