|Home | About | Journals | Submit | Contact Us | Français|
Resistance of human renal cell carcinoma (RCC) and melanoma to the apoptosis-inducing effects of interferons (IFNs) was postulated to result from epigenetic silencing of genes by DNA methylation, a common feature of human cancers. To reverse silencing, 5-AZA-deoxycytidine (5-AZA-dC) or selective depletion of DNA methyltransferase 1+ (DNMT1) by phosphorothioate oligonucleotide antisense (DNMT1 AS) were employed in cells resistant (<5% TUNEL +) to apoptosis induction by IFN-α2 and IFN-β (ACHN, SK-RC-45, A375). 5-AZA-dC and DNMT1 AS similarly depleted available DNMT1 protein and, at doses that did not cause apoptosis alone, resulted in apoptotic response to IFNs. The proapoptotic tumor suppressor RASSF1A was reactivated by DNMT1 inhibitors in all three cell lines. This was associated with demethylation of its promoter region. IFNs augmented RASSF1A protein expression after reactivation by DNMT1 inhibition. In IFN-sensitive WM9 melanoma cells expression of RASSF1A was constitutive, but also augmented by IFNs. RASSF1A siRNA reduced IFN-induced apoptosis in WM9 cells and in DNMT1 depleted ACHN cells. Conversely, lentiviral expression of RASSF1A, but not transduction with empty virus, enabled IFN-induced apoptosis. IFN induced TRAIL and TRAIL-neutralizing antibody inhibited apoptotic response to IFN in RASSF1A expressing ACHN cells. Accordingly RASSF1A markedly sensitized to recombinant TRAIL. Normal kidney epithelial cells, although expressing RASSF1A, did not undergo apoptosis in response to IFN or TRAIL, but had > 400 fold higher TRAIL decoy receptor 1 expression than transduced ACHN cells (real-time RT-PCR). Results identified RASSF1A as regulated by IFNs and participating in IFN-induced apoptosis at least in part by sensitization to TRAIL.
As a result of direct effects on cell function, IFNs play a critical role in host response to virus infection and cancer through influences on viral replication and on tumor cell survival, differentiation and motility (1–3). Activation of a signal cascade by IFNs culminates in induction of several hundred interferon stimulated genes (ISGs) that produce the cellular effects (4, 5). Resistance of cells to IFNs has been related not only to defects in specific components of the signaling cascade but also to homeostatic mechanisms that modulate receptor activation and transcriptional response (1–3). Epigenetic silencing of genes critical for effects of IFNs is a mechanism that also could influence cellular response and resistance. Although direct consequences for cell function have been little evaluated, gene profiling and expression studies indeed have identified genes involved in IFN pathways that are epigenetically silenced by hypermethylation of their 5′ regulatory regions (6–9).
Maintenance of DNA methylation in the promoter region of genes can result in heritable silencing of expression just as do mutational deletions (10). The degree of enzymatic redundancy in this process with the maintenance DNA methyltransferase 1 (DNMT1) and de novo DNMTs such as DNMT 3b (11) is still an unresolved issue but at least in a colon (12), a breast, and a lung cancer (13) cell line inhibition of DNMT1 by oligonucleotide antisense or siRNA was sufficient for re-expression of silenced genes. Epigenetic inactivation of genes that control DNA stability, cell proliferation, and apoptosis is integral to the neoplastic process (10). IFNs have increased expression of tumor suppressor genes such as p53 (14), that are frequently affected by mutational loss of function in cancer, and other ISGs that can be silenced by DNA methylation, such as XAF1, may have TSG function (5, 6). Nephrectomy specimens and resections of primary melanomas revealed a high frequency of silencing by promoter hypermethylation of the tumor suppressor gene RASSF1A in papillary (up to 100% of patients) (15) and clear cell RCC (up to 90%) (15, 16) as well as in melanomas (55%) (17). RASSF1A plays a role in mitosis control, it can furthermore interact with the proapoptotic kinase MST1 and the scaffolding protein CNK1 to induce apoptosis and has facilitated death receptor-induced Bax conformational change and apoptosis by relieving inhibition of MAP-1 (18–24). Apoptosis has been increasingly recognized as one of the mechanisms that may play a role in antitumor effects of IFNs (25). We postulated that epigenetic silencing of RASSF1A might confer resistance to apoptosis induction by IFNs in RCC and melanoma. Thus, the effects of the well-established potent DNA demethylating agent 5-AZA-dC were assessed in RCC and melanoma cell lines that were resistant to apoptosis induction by IFNs and did not express RASSF1A. The specific role of DNMT1 for gene silencing and IFN resistance was evaluated using a selective phosphorothioate antisense oligonucleotide inhibitor (DNMT1 AS) in the more readily transfectable RCC cells.
ACHN, A375, HeLa (ATCC, Manassas, VA), SK-RC-45 (Memorial Sloan Kettering, NY), WM9 (Wistar Institute, PA), and normal kidney epithelial cells (isolated after review by a staff pathologist from abundant tissue of nephrectomies performed at the Cleveland Clinic Foundation, OH, and used in 2nd or 3rd passage) were cultured at 37°C in 5% CO2 using Minimum Essential Medium (or DMEM) (GIBCO, Invitrogen, Carlsbad, CA) with 0.1 mM non-essential amino acids, 1.0 mM pyruvate, 10% fetal bovine serum, penicillin G (50 U/ml), and streptomycin (50 μg/ml) and were regularly confirmed mycoplasma-free. NHEM normal human neonatal melanocytes (Cambrex, Baltimore, MD) were cultured under the same incubator conditions but in melanocyte media (Cambrex, Baltimore, MD) according to the supplier’s recommendations. Recombinant IFN-α2b (Schering-Plough, Kenilworth, NJ) and IFN-β1a (Serono, Rockland, MA) had specific activities of 2×108 units/mg protein. Apo2L / TRAIL was obtained from PeproTech (Rocky Hill, NJ), rabbit polyclonal TRAIL neutralizing antibody and control rabbit immunoglobulin from ProSci (Poway, CA).
To downregulate DNMT1, cells were transfected with MG98 (MethylGene, Quebec, Canada), a second-generation 4×4 2′O methyl (bold) phosphorothioate oligonucleotide antisense against the 3′ UTR of DNMT1 mRNA (5′-TTCATGTCAGCCAAGGCCAC-3′) or mismatch (MM) control oligonucleotide of similar sequence (5′- TTAATGTAACCTAAGGTCAA -3′) starting 1d after plating at cell concentrations that allowed at least doubling of number in 48 hr and optimal transfection efficiency (5000 and 15000 cells/cm 2 for SK-RC-45 and ACHN cells, respectively). Transfections were with 6.25 μg/ml Lipofectin in OptiMem (GIBCO, Invitrogen, Carlsbad, CA) over 4 hr. Before and after transfections cells were washed with PBS, every 2nd day cells were replated 4 hr after the end of the preceding transfection. 5-AZA-dC (Sigma-Aldrich, St Louis, Mo) was freshly thawed solution and diluted into media. Cells were replated 4 hr after 5-AZA-dC every 2d into complete media not containing 5-AZA-dC. siRNA transfections were performed like DNMT1 AS transfections but only once and with pre-incubation of siRNA 0.3 volume of the final amount of lipofectin to allow for formation of complexes. siRNAs were obtained from Dharmacon (option C). RASSF1A siRNA sequence was as published (23): (sense) 5′-GAC CUC UGU GGC GAC UUC A-3′, control siRNA sequence was: (sense) 5′-CAC GUC UCU CCC GAC UAG A-3′.
Protein (20–40 μg) from whole cell lysates were probed for DNMT1 by polyclonal antibodies (MethylGene, Quebec, Canada), Stat1, Stat2, Stat3 by monoclonal antibody (mAB) (BD Transduction Laboratories, San Jose, CA), Caspase 3 by pAB (BIOMOL International L.P., Plymouth Meeting, PA), RASSF1A by mAB (eBioscience, San Diego, CA), MST1 by pAB (Cell Signaling, Beverly, MA), and actin by mAB (Sigma-Aldrich, St Louis, MO) after separation in 8–14% SDS-polyacrylamide gels and transfer to PVDF membranes. For detection of bound primary antibody PVD membranes were incubated with horseradish tagged goat anti-mouse or goat anti-rabbit antibody (Bio-Rad, Hercules, CA), followed after washing with TBST, by staining with enhanced chemiluminescence solution (Amersham, Piscataway, NJ). To compare relative RASSF1A expression densitometry was performed using BioRad Chemi Doc ™ XRS and Quantity One – 4.5.2 1D analysis software (BioRad). All signals were normalized to corresponding actin signals of stripped membranes.
For TUNEL assay, cells were harvested and processed according to the manufacturer’s instructions (BD Pharmingen, San Diego, CA). Apoptosis was confirmed with an assay for the activity of caspase 3 (BD Clontech, Palo Alto, CA), performed according to the manufacturers instructions, or caspase 3 cleavage detection by immunoblot (polyclonal caspase 3 antibody from Biomol) 48 h after IFNs. All results were confirmed with at least one additional analysis.
RNA was isolated using the Trizol (Invitrogen, Carlsbad, CA) and cDNA prepared with a superscript III first strand synthesis kit including a final Rnase H digestion step (Invitrogen) according to the manufacturer. For real-time RT-PCR, custom taqman expression primers (Applied Biosystems, Foster City, CA) were used according to the manufacturer instructions using ABI PRISM Sequence Detection Instrument 7700 (Applied Biosystems).
Genomic DNA (1 μg), harvested with a blood DNA mini kit (Quiagen, Valencia, CA), for bisulfite modification with the CpGenome kit (Chemicon International, Temecula, CA) according to the manufacturer’s instructions; 4μl of bisulfite modified DNA was used per 25 μl MSP reaction. Primers for RASSF1A MSP were as published (26). For PCR, methylated (M) primer pairs were denaturated at 95 °C for 5 min, followed by 35 cycles with a 1 min denaturation step, 30 sec annealing at 60 °C, and extension at 72 °C for 30 sec. Final extension after 35 cycles was at 72 °C for 4 min. For sequences specific for unmethylated (U) DNA, annealing was at 55 °C.
RT-PCR was with primers that amplified RASSF1 variants regulated by a promoter hypermethylated in cancer (19). Primers were 5′-AGC GTG CCA ACG CGC TGC GCA T-3′ (sense) and 5′-CAG GCT CGT CCA CGT TCG TGT C-3′ (antisense). Settings were 95°C - 4 min, (95°C-1min, 52°C-30sec, 72°C-30 sec for 30 to 35 cycles), 72°C-4min. GAPDH was amplified with the settings 95°C - 4 min, (95°C-45 sec, 55°C-30sec, 72°C-50 sec for 15 to 25 cycles), 72°C-4min. GAPDH primers were 5′-CAG ACC TAC TCA GGG ATT C-3′ (sense) and 5′-GAG CCA GAC GCT GCT TTG T-3′ (antisense). For sequencing of full length RASSF1 cDNA in DNMT1 AS treated ACHN cells RT-PCR with primers (5′ to 3′) CGC CCA GTC TGG ATC CTG (sense) and CTC AAT GCC TGC CTT ATT CTG (antisense) was performed using proofreading platinum Pfx polymerase (Invitrogen, Carlsbad, CA): denaturation at 95 °C for 4 min followed by 30 cycles at 95 °C for 45 sec, annealing at 58 °C for 30 sec, and extension at 68 °C for 3 min followed by final extension at 68 °C for 8 min. Products were cloned into Zero Blunt cloning vector and then into pcDNA4 for expansion. Sequencing of 4 independent bacterial clones all identified RASSF1A, (NM_007182), with a single nucleotide polymorphism at nucleotide 528 (T instead of G) leading to a conservative change at amino acid position 133 (serine for alanine). Before cloning into lentivirus all CpG sites 5′ of the translation start site were excluded by amplification with primers (5′ to 3′) GGA TCC ACC ATG TCG GGG (sense) and CTT CCG TCT GTC GTC CGC TAT AG (antisense) using proofreading platinum Pfx polymerase (Invitrogen): denaturation at 94 °C for 4 min followed by 25 cycles of denaturation at 94 °C for 30 sec, annealing at 63 °C for 30 sec, and extension at 68 °C for 2 min followed by final extension at 68 °C for 8 min.
RASSF1A cDNA from DNMT1 AS treated cells was used for overexpression. The BamHI and EcoRV fragment of the RASSF1A ORF was subcloned into the corresponding sites of a modified self inactivating lentiviral vector LRV (6.9kb), with a blasticidin resistance marker (confirmed by sequencing) and was then transfected into 293T cells along with the packaging plasmids possessing the gag-pol, rrev, and VSV glycoprotein G for production and propagation of transducable Lenti-RASSF1A virus. Transductions into ACHN cells were 2–3 times at 20–30 MOI of Lenti-RASSF1A virus (10μg/ml of polybrene); 3 days later blasticidin (10μg/ml) was added. Colonies were pooled, expanded, and RASSF1A verified by western blot using RASSF1A mAB (eBioscience). Subsequently, the RASSF1A expressing cells were selected in medium containing blasticidin and sequence confirmation repeated.
ACHN and SK-RC-45 RCC and A375 melanoma cells, treated with IFN-α2 or IFN-β alone, were resistant to induction of apoptosis at doses up to 500 U/ml (<5% apoptotic cells on TUNEL, fig 1 A–C). To determine whether a potent DNMT1 inhibitor would influence apoptosis induction by IFNs, cells were treated with 5-AZA-dC for 2–6 days. After incorporation into DNA 5-AZA-dC covalently binds DNMT1 leading to reduction of available DNMT1 protein in cells and whole cell lysates (10, 12). Although 5-AZA-dC alone resulted in some apoptosis, apoptosis was markedly increased after the addition of IFN-α2 and even more so after IFN-β (fig 1 A–C).
In contrast, no augmentation of apoptosis by IFN-β was observed in normal human neonatal melanocytes (NHEM) or normal kidney epithelial cells (NKE) after pretreatment with 5-AZA-dC (D). In non-malignant cells 5-AZA-dC toxicity correlated with doubling times, which were approximately 4 days for NHEM, 3 days for NKE 01, and 2 days for NKE 02 (data not shown). To confirm reduction in DNMT1, cell lysates were subjected to western blot analysis. 5-AZA-dC markedly decreased DNMT1 protein (immunoblots in fig 1 A–D).
To determine whether DNMT1 depletion was sufficient for overcoming resistance to IFN-induced apoptosis ACHN cells were transfected daily with 40 nM DNMT1 AS. Nearly complete suppression of DNMT1 protein was achieved by day 4 in ACHN cells (fig. 2A). In contrast, DNMT1 AS did not influence expression of STAT1, STAT2, or STAT3 (fig 2A). This latter result was further confirmed by RT-PCR for STAT1 (data not shown) and cRNA array (data not shown).
In the absence of DNMT1, cell divisions should yield daughter cells with demethylated genes; therefore transfections were continued for an additional 2–4 days before apoptosis assessment. The duration of DNMT1 depletion correlated with sensitization of ACHN cells to even low doses (50 U/ml) of IFN-α2 or IFN–β (fig. 2A). Mismatch control oligonucleotide (MM) did not sensitize to IFN-induced programmed cell death (fig. 2A). Apoptosis was confirmed by immunoblotting for detection of cleaved (activated) caspase 3 (fig. 2A) and by caspase 3 activity assays (data not shown).
Similarly in SK-RC-45 cells, after 4 or 6 days of DNMT1 AS treatment complete suppression of DNMT protein expression was achieved with no effect of the MM (fig. 2B). Treatment with 500 U/ml IFN-α2 or IFN-β after DNMT1 depletion for 6 days increased the apoptotic fraction from 5.4 +/− 4.5 % to 23.4 +/− 12 % and 45 +/− 1% (mean +/− SD), respectively, whereas after MM IFNs caused apoptosis in less than 10% of cells (fig. 2B). Apoptosis was also confirmed by caspase 3 activity assays (data not shown).
In A375 melanoma cells only minimal reduction of DNMT1 protein resulted from non-toxic concentrations of DNMT1 AS (data not shown) whereas treatment with 200 nM 5-AZA-dC over 4 days reduced available DNMT1 protein and increased frequency of apoptotic cells from 1.5 +/− 1.1% in controls to 22.9 +/− 3% (mean +/− SD) after 100 U/ml IFN-β, while alone causing little apoptosis (4.7 +/−1.85% apoptotic cells) (fig. 1C).
Since the lowest non-toxic schedule of 5-AZA-dC resulted in less IFN-induced apoptosis than the DNMT1 AS in the RCC cell lines, DNMT1 AS was mostly used for subsequent experiments with these cells and subsequent experiments in A375 used 5-AZA-dC.
To determine whether RASSF1A might be involved in induction of apoptosis by IFNs after DNMT1 inhibition, its expression was assessed in the IFN resistant cell lines after both DNMT1 AS and 5-AZA-dC. Duration of DNMT1 depletion correlated with re-expression of RASSF1A mRNA, as shown in ACHN cells (fig. 3A). DNMT1 AS (40 nM over 6 days) or 5-AZA-dC (200 nM over 4 days) also reactivated RASSF1A expression in SK-RC-45 cells (fig. 3A). While it was not possible to deplete DNMT1 by AS at doses allowing for continuing cell divisions in A375 cells 5-AZA-dC led to reactivation of RASSF1A expression (fig. 3A). To confirm that RASSF1A expression was related to methylation status of its promoter region, methylation-specific PCR was undertaken. Reactivation of RASSF1A was associated with demethylation of a promoter CpG island in ACHN cells (fig. 3B).
To determine whether reactivation of RASSF1A transcription was followed by translation into protein and whether this was influenced by IFN treatment immunoblots were performed. Transfection with DNMT1 AS reactivated RASSF1A protein expression, an effect that was further augmented by IFNs (fig. 3C). Similar results occurred in A375 cells treated with 5-AZA-dC; after pretreatment with 5-AZA-dC, IFN-β induced RASSF1A protein (fig. 3C).
Hypothesizing that RASSF1A might be involved in IFN-induced apoptosis its expression was determined in a melanoma cell line (WM9) known to undergo programmed cell death upon treatment with IFN-β (27). In contrast to ACHN cells, RASSF1A mRNA and protein were constitutively expressed in WM9 cells (fig. 3A and C). Addition of IFN-β to WM-9 cells further augmented RASSF1A protein expression (fig. 3C). Thus, unless RASSF1A was silenced, IFNs could increase RASSF1A expression.
To further assess interactions between IFNs and RASSF1A, activation of MST1, an apoptotic partner molecule of RASSF1A, was also assessed. MST1 was cleaved (activated (28)) after IFNs in DNMT1 AS but not control oligonucleotide (MM) pretreated ACHN cells (fig. 3C). Thus not only did IFNs augment RASSF1A expression but the increased protein was associated with enhanced pro-apoptotic protein activation.
To determine whether RASSF1A played a role in apoptosis induction by IFNs ACHN cells depleted of DNMT1 by AS were transfected with RASSF1A siRNA followed by IFN treatment. Inhibition of RASSF1A by siRNA decreased IFN-induced apoptosis in DNMT1 AS pretreated cells from 63.9 +/− 9.19% in control siRNA treated cells to 35 +/− 4.1% (mean +/− SD), as determined by frequency of TUNEL positive cells (fig. 4A). Parallel assessment of RASSF1A protein expression by densitometry of western blots identified 67% reduction in DNMT1 AS treated cells and 61% reduction in DNMT1 AS and IFN beta treated cells by specific compared to control siRNA (fig. 4A). Absence of IFN induction by the control siRNA was assessed by real-time RT-PCR for the ISGs XAF1, IRF1, and USP18. None of these genes were upregulated as compared to treatment with lipofectin transfection reagent alone (0.6, 1.1, and 1.0 fold induction, respectively for each gene).
To further confirm the importance of RASSF1A for IFN-induced apoptosis, WM9 cells, sensitive to apoptosis induction by IFNs alone, were treated with IFN after depletion of RASSF1A by siRNA. Similar reduction of IFN-induced apoptosis by RASSF1A siRNA occurred in WM9 melanoma cells (from 60.45 +/− 3.18% in control siRNA treated cells to 39.2 +/− 3.11%, fig. 4B). RASSF1A protein was reduced by 65% at baseline and 57% in IFN treated cells by specific compared to control siRNA, as determined by densitometry (fig. 4B). Thus a limited suppression of RASSF1A partially inhibited apoptosis induced by IFN-β.
To further confirm the role of RASSF1A in overcoming resistance to IFN-induced apoptosis, independent of any other genes reactivated by DNMT1 depletion, forced expression of RASSF1A using a lentiviral construct (fig 5A) was undertaken. After transduction, the population of transduced cells was kept in selective antibiotic for 14 days before RASSF1A protein expression and sensitivity to IFN-induced apoptosis were assessed. Of RASSF1A transduced cells 16.83 +/− 0.98%% underwent apoptosis in response to 50 U/ml IFN-β, compared to 3.77 +/− 1.67% (mean +/− SD) of empty virus transfected cells (fig. 5B). Subcloning of stably transduced cells demonstrated that clones that expressed RASSF1A at levels comparable to the ones achieved by DNMT1 AS treatment (78, 234, and 134 % relative RASSF1A expression compared to DNMT1 AS by densitometry in clones 1.3, 1.7, and 1.8, respectively, fig. 5B) underwent apoptosis in response to 50 U/ml IFN-β (20 to 40% apoptotic cells, fig 5B). Thus RASSF1A alone could overcome resistance to IFN-induced apoptosis.
In WM9 cells Apo2L/TRAIL was known to be induced by IFN-β and to mediate IFN-induced apoptosis (27), which could be reduced by RASSF1A siRNA (fig. 4B). Since a recent report suggests that RASSF1A was required for death-receptor-induced Bax conformational change and apoptosis (18) we hypothesized that RASSF1A may overcome resistance to IFN-induced apoptosis by sensitization to Apo2L/TRAIL. As determined by real-time RT-PCR IFN-β (50 U/ml over 16 hr) induced Apo2L/TRAIL in ACHN cells (between 20 and 25 fold). Co-treatment with TRAIL-neutralizing antibody, but not control rabbit immunoglobulin, partially inhibited IFN-induced apoptosis in RASSF1A expressing ACHN cells (fig. 5C). Accordingly sensitivity of ACHN cells to Apo2L/TRAIL-induced apoptosis was markedly increased by forced RASSF1A expression (fig. 5D).
Normal kidney epithelial (NKE) cells expressed RASSF1A (fig. 3B), but even after 5-AZA-dC treatment remained resistant to IFN and Apo2L/TRAIL-induced apoptosis (fig. 1D, fig. 5D). Real-time RT-PCR identified greater than 400 fold higher TRAIL decoy receptor 1 (TRAIL DcR1) expression in NKE compared to transduced ACHN cells, while TRAIL receptor 1 and 2 (TRAIL R1, 2) expression was similar. 5-AZA-dC did not markedly the relative expression of pro-apoptotic (TRAIL R1, 2) to cell protective (TRAIL DcR1, 2) TRAIL receptors but modestly increased them all (table 1). This suggested that RASSF1A sensitized to IFN-induced apoptosis at least in part by sensitization to Apo2L/TRAIL and that strong expression of TRAIL decoy receptor 1 might protect certain RASSF1A expressing non-malignant cells from apoptosis induction by IFN or Apo2L/TRAIL.
Maintenance of DNA methylation in the promoter region of genes can lead to heritable epigenetic silencing of expression with consequences similar to mutational deletions (10). Antitumor activity of IFNs, commonly used in the treatment of metastatic RCC and melanoma, depends on induction of gene expression in cancer cells, immune cells, and cells regulating angiogenesis(1). Hypothesizing that unsatisfactory response rates of melanoma and RCC to IFN (about 15%) are in part due to epigenetic silencing of genes, the role of a tumor suppressor gene that is frequently silenced in both malignancies (15–17) was examined in vitro. Treatment of three cell lines (2 renal and one melanoma) that were resistant to apoptosis induction by high doses of IFN-α2 and IFN-β (500 U/ml) with the DNA demethylating nucleoside analogue 5-AZA-dC overcame resistance to IFN-induced apoptosis (fig.1A–C) and reactivated expression of RASSF1A (fig. 3).
5-AZA-dC is a nucleoside analogue, that after incorporation into DNA, inhibits DNMT1 by covalent binding (10). DNMT1 is thus trapped and not available at the DNA replication fork to copy methylation patterns from mother to daughter strand resulting in demethylation upon cell division. The covalent binding of the 190 KD DNMT1 protein to DNA, however, also results in DNA damage and thus 5-AZA-dC may have effects in cells, independent of its DNA demethylating activity (29, 30). Sensitization to IFN-induced apoptosis was observed at 5-AZA-dC doses that did not result in apoptosis alone (fig 1) associated with reactivation of RASSF1A mRNA expression (fig. 3A). More importantly specific inhibition of DNMT1 by oligonucleotide antisense (DNMT1 AS) similarly reactivated RASSF1A and overcame resistance to apoptosis induction by IFNs (fig. 2–3). Compared to mismatch control oligonucleotide, transfection reagent alone, and media alone, DNMT1 AS did not induce ISGs (fig. 2A, and data not shown) suggesting that its effect on IFN resistance was due to DNMT1 depletion (fig. 2A) and associated DNA demethylation (fig. 3B). These results furthermore support that at least in RCC cells, which were more amenable to downregulation of DNMT1 by AS than studied melanoma cells, DNMT1 was critical for silencing of genes.
Containing a diacylglycerol and a rasGTP binding domain but no catalytic activity, RASSF1A has influenced function of binding partners, including the E1A-regulated transcription factor p120 (E4F) (31), the proapoptotic kinase MST1 (20, 22, 32), the scaffold protein CNK1 (22), and cdc20, an activator of anaphase promoting complex (24). Interaction of RASSF1A with cdc20 regulated mitotic progression (21, 24) and apoptosis resulted when RASSF1A was coexpressed with the scaffold protein CNK1 (22). RASSF1A directed MST1 to the cell membrane, where MST1 can be activated influencing apoptosis (20, 22). Since it was activated by IFNs after RASSF1A re-expression (fig. 3C), MST1 may be contributory to the apoptotic effects of IFNs. Interestingly IFNs also increased protein expression of RASSF1A after DNMT1 depletion and in cells with baseline expression (fig. 3C). The exact mechanism of RASSF1A protein regulation by IFNs will need to be studied in future experiments but real-time RT-PCR results suggest post-transcriptional events (data not shown).
In DNMT1 depleted ACHN cells, selective suppression of RASSF1A by siRNA reduced apoptosis in response to IFN from 63.9 +/− 9.19% to 35 +/− 4.1% (mean +/− SD) (fig. 4A). Rationalizing that cells sensitive to apoptosis induction by IFNs might express RASSF1A, WM9 cells (27) were studied. Without treatment RASSF1A was expressed, IFN increased expression (fig. 3A, C), and RASSF1A siRNA reduced IFN-β induced apoptosis from 60.45 +/− 3.18% to 39.2 +/− 3.11% accompanied by reduction of RASSF1A protein expression (fig. 4B). Conversely, in ACHN cells lentiviral expression of RASSF1A protein to levels comparable to the one achieved by DNA demethylation overcame resistance to IFN-induced apoptosis (fig. 5).
Recent evidence has suggested requirement of RASSF1A for death receptor-induced Bax conformational change and apoptosis. RASSF1A enabled Bax apoptotic signaling by relieving an inactivating intramolecular conformation of a necessary partner molecule of Bax, BH3-like protein modulator of apoptosis-1 (MAP-1) (18). Apo2L/TRAIL was induced by IFN-β (50 U/ml) in ACHN cells, 20–25 fold as determined by real-time RT-PCR (data not shown), and TRAIL neutralizing antibody inhibited IFN-induced apoptosis of RASSF1A expressing cells (fig 5C). Accordingly RASSF1A markedly sensitized ACHN cells to Apo2L/TRAIL-induced cell death (fig. 5D).
On the other hand normal kidney epithelial (NKE) cells did not undergo programmed cell death in response to IFN or TRAIL despite expression of RASSF1A and no synergism of 5-AZA-dC with either drug was observed (fig. 1D, fig. 3A, fig. 4D). Real-time RT-PCR before and after 4 days of 5-AZA-dC treatment revealed greater than 400 fold higher TRAIL decoy receptor 1 expression in NKE compared to ACHN cells (table 1). TRAIL decoy receptors bind Apo2L/TRAIL without transmission of apoptotic signals into the cell (33). Thus RASSF1A overcame resistance to IFN-induced apoptosis at least in part by sensitization to Apo2L/TRAIL and strong TRAIL decoy receptor expression might protect certain non-malignant RASSF1A expressing cells from cell death induction by IFN or Apo2L/TRAIL.
Promoter hypermethylation of interferon stimulated genes (ISGs) including DAPK (mainly lymphoid malignancies), XAF1 (gastric), and IRF7 (fibrosarcoma), as well as of genes essential for IFN apoptotic signaling like TRAIL R1 and caspase 8 (lung), have been identified in cell lines and /or biopsy specimens (6–9). While not described as frequently hypermethylated in renal cancer, reactivation of such genes could have contributed to sensitization of ACHN cells to IFN-induced apoptosis after DNMT1 depletion. However, except for one ISG of unknown function, IFI27, no other ISGs or genes known to be essential for IFN apoptotic signaling (25), were increased in expression by DNMT1 AS treatment in ACHN cells, as determined by U133A Affymetrix cRNA array (data not shown). Although TRAIL R1 and TRAIL R2 were not represented on the array, quantitative RT-PCR did not identify evidence for their reactivation by DNMT1 AS (data not shown). However among the 137 genes increased at least twofold in DNMT1 AS over MM treated cells, 8 had roles in apoptosis and also may have contributed to overcoming resistance to IFN-induced programmed cell death by influencing apoptotic pathways (small GTPase ARHGDIB), inhibition of NF-κB (NFKBIA, IER3, C8FW), or other mechanisms (PHLDA1, NAC, STK17A, ASC) (data not shown).
IFN-α2 has increased survival of patients with metastatic RCC in randomized trials, albeit only for several weeks to months (1, 34). Prolongation of disease-free and possibly overall survival has resulted when IFN-α2 has been administered to melanoma patients for high risk primary disease (1). Resistance of RCC and melanoma cells to the apoptosis-inducing effects of IFN-α2 and IFN-β was overcome by inhibition of DNA methyltransferase 1 (DNMT1) (figs. 1, ,2).2). This was at least in part due to reactivation of the tumor suppressor gene RASSF1A (figs. 4, ,5),5), that is frequently silenced by DNA methylation in RCC and melanoma (15–17), and as shown herein up-regulated in expression by IFNs (fig. 3C). By targeting DNMT1 in RCC and melanoma, clinical antitumor effects of IFNs may be augmented through reactivation of silenced RASSF1A and possibly by reactivation of other heterogeneously silenced genes in IFN pathways.
Barbara Jacobs and Mamta Chawla-Sarkar contributed invaluable assistance and advice for initiation of these studies.
Supported in part by grants NIH R01 CA90914, funds from the Case Comprehensive Cancer Center, Cleveland, OH, and an unrestricted grant from MethylGene, Inc., Montreal, Quebec to E.C.B., and NIH R01 CA 90837 to D.W.L.