Cancer–testis antigens (CTA), such as NY-ESO-1, MAGE-A1, and MAGE-A3, are immunogenic proteins encoded by genes, which are normally expressed only in male germ cells but are activated by ill-defined epigenetic mechanisms in human tumors, including lung cancers. Previously, we reported induction of these CTAs in cancer cells, but not normal cells, by DNA-demethylating agents and histone deacetylase inhibitors using clinically achievable exposure conditions. In the present study, we evaluated chromatin alterations associated with repression/activation of cancer–testis genes in lung cancer cells to further develop gene-induction regimens for cancer immunotherapy. Repression of NY-ESO-1, MAGE-A1, and MAGE-A3 coincided with DNA hypermethylation, recruitment, and binding of polycomb-group proteins, and histone heterochromatin modifications within the promoters of these genes. Derepression coincided with DNA demethylation, dissociation of polycomb proteins, and presence of euchromatin marks within the respective promoters. Short hairpin RNAs were used to inhibit several histone methyltransferases (KMT) and histone demethylases (KDM) that mediate histone methylation and repress gene expression. Knockdown of KMT6, KDM1, or KDM5B markedly enhanced deoxyazacytidine (DAC)-mediated activation of these cancer–testis genes in lung cancer cells. DZNep, a pharmacologic inhibitor of KMT6 expression, recapitulated the effects of KMT6 knockdown. Following DAC–DZNep exposure, lung cancer cells were specifically recognized and lysed by allogeneic lymphocytes expressing recombinant T-cell receptors recognizing NY-ESO-1 and MAGE-A3. Combining DNA-demethylating agents with compounds, such as DZNep, that modulate histone lysine methylation may provide a novel epigenetic strategy to augment cancer–testis gene expression as an adjunct to adoptive cancer immunotherapy. Cancer Res; 71(12); 4192–204. ©2011 AACR.
Cancer–testis antigens (CTA) are encoded by a unique class of genes [cancer–testis (CT) genes], normally expressed in germ cells or placenta, that are derepressed by epigenetic mechanisms in various human malignancies (1). Because they are typically expressed only in immune-privileged sites, CTAs induce humoral as well as cell-mediated immune responses when aberrantly expressed in somatic cells; as such, CTAs have emerged as highly attractive targets for cancer immunotherapy (2). Vaccines targeting CTAs such as NY-ESO-1, MAGE-A1, and MAGE-A3 induce antitumor immunity, and T cells expressing native or genetically engineered receptors recognizing these antigens mediate tumor regression in some cancer patients (3–5).
Approximately 50% of CT genes, including NY-ESO-1, MAGE-A1, and MAGE-A3, are located on the X chromosome (6). CT-X chromosome (CT-X) genes are normally expressed in spermatogonia, and typically comprise extended families associated with inverted DNA repeats (7). Relative to autosomal CT genes, CT-X genes are more frequently activated in cancer cells, and particular gene families appear to be derepressed in a tumor-specific manner. Although believed to be activated as a result of global DNA demethylation, the epigenetic mechanisms mediating coordinate derepression of CT genes during multistep carcinogenesis have not been fully elucidated (7–9).
Although NY-ESO-1, MAGE-A1, and MAGE-A3 are expressed in 25% to 40% of non–small cell lung cancers (NSCLC; ref. 10), immune responses to these CTAs are uncommon in lung cancer patients (11, 12) due, in part, to levels of antigen expression, which are below the threshold for immune recognition. Conceivably, upregulation of CTA expression by chromatin-remodeling agents can enhance immunogenicity of lung cancer cells, facilitating their eradication by endogenous immune mechanisms, or adoptively transferred T cells. Previously, we showed that the DNA-demethylating agent, 5-aza-2′-deoxycytidine (decitabine; DAC) and the histone deacetylase (HDAC) inhibitor depsipeptide (romidepsin; DP) mediate synergistic activation of CT-X gene expression in cultured lung cancer cells but not in normal epithelia or lymphoid cells (8). In addition, we reported that, following DAC or sequential DAC/DP exposure, lung cancer cells can be recognized by cytolytic T lymphocytes (CTL) expressing receptors specific for NY-ESO-1 or MAGE-A3 (13–15). Furthermore, we have shown upregulation of NY-ESO-1 and MAGE-A3 expression in primary lung cancers in patients receiving 72-hour continuous decitabine infusions (steady-state plasma concentrations ∼50–100 nmol/L; ref. 16; Schrump and colleagues, manuscript in preparation). Finally, we have shown that a CTA induced in tumor cells in vivo by systemic DAC administration can be effectively targeted by adoptively transferred CTL in immunocompetent mice (17). The present study was undertaken to comprehensively examine mechanisms regulating NY-ESO-1, MAGE-A1, and MAGE-A3 expression in lung cancer cells to further develop epigenetic strategies for human cancer immunotherapy.
Materials and Methods
Cell lines and drug treatment conditions
All lung cancer lines were obtained from the American Type Culture Collection and were characterized and authenticated at the repository by methods including mycoplasma testing, DNA profiling, and cytogenetic analysis; these lines were used within 6 months of purchase for this study, validated in our laboratory by periodic human leukocyte antigen (HLA) typing, and cultured as described (8). Primary normal human bronchial epithelial (NHBE) cells, small airway epithelial cells (SAEC), and normal human dermal fibroblasts (NHDF) were purchased from Lonza, Inc., and cultured according to manufacturer's instructions. Immortalized human bronchial epithelial cells (HBEC) were generously provided by John D. Minna (University of Texas Southwestern, Dallas, TX) and cultured as described (18). DAC and trichostatin A (TSA) were purchased from Sigma Chemical Company. DZNep was provided by the Chemical Biology Laboratory, National Cancer Institute (NCI). DP was obtained from the Developmental Therapeutics Program, NCI. The effects of DAC and DZNep treatment on CT-X gene expression were determined after exposure to 0.1 μmol/L DAC or 0.5 to 5 μmol/L DZNep for 72 hours or concurrent DAC–DZNep (0.1:0.5 μmol/L) for 72 hours followed by exposure to normal media for 18 to 24 hours. DAC/DP and DAC/TSA treatments were carried out as described (8).
Real-time reverse transcription-PCR analysis
RNA was isolated using RNeasy Mini Kit (Qiagen). cDNAs were made using Reverse Transcription Kit (Bio-Rad). Quantitative reverse transcription (qRT)-PCR primers for CT-X genes and β-actin expression are listed in Supplementary Table S1.
Total cell proteins were extracted and immunoblotting was carried out as described previously (19) with minor modifications, using primary antibodies listed in Supplementary Table S2 including appropriate horseradish peroxidase–conjugated secondary antibodies, and SuperSignal West Pico Chemiluminescent Substrate (Pierce Biotechnology). Testis lysate (Abcam) was used as positive control for NY-ESO-1 and MAGE-A1. Lysates from HEK293 cells constitutively expressing MAGE-A3 were used as a positive control for this CTA (15).
NY-ESO-1, MAGE-A1, and MAGE-A3 expression in cultured cells was detected by immunofluorescence techniques using primary antibodies recognizing these CTAs (Supplementary Table S2) and visualized using fluorescein isothiocyanate (FITC)-labeled secondary antibodies (Supplementary Table S2) as described (20). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI).
CpG islands within the NY-ESO-1, MAGE-A1, and MAGE-A3 promoters were identified using an online CpG island search engine (21). Genomic DNA was isolated from drug-treated or control cells using the Qiagen DNeasy Kit. Bisulfite modification of DNA was done using the Qiagen EpiTect Bisulfite Kit. Pyrosequencing was done as described previously (18), using primers listed in Supplementary Table S1.
Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (ChIP) assay was conducted as described (22) with minor modifications. Briefly, DNA–protein complexes were cross-linked with formaldehyde at a final concentration of 1% for 15 minutes. Immune complexes were formed with either nonspecific immunoglobulin G (IgG) or ChIP-grade antibodies listed in Supplementary Table S2. DNA was eluted and purified from complexes, followed by PCR amplification of the NY-ESO-1, MAGE-A1, or MAGE-A3 promoters as previously described (8), using primers listed in Supplementary Table S1.
Generation of KMT6, KDM1, KDM5B, or SirT1 knockdown cells and KMT6-overexpressing stable cells
H841 cells were transduced with lentiviral short hairpin RNA (shRNA) vectors targeting KMT6, KDM1, KDM5B, and SirT1 or sham sequences (Sigma), or transfected with pCMV6-AC-GFP or pCMV6-AC-GFP-KMT6 (Origene). Target gene knockdown or overexpression was confirmed by RT-PCR and immunoblot. Stable transfectants (4 independent clones for each knockdown or overexpression) were isolated and expanded under puromycin (knockdowns) or G418 selection (overexpressors). Following reconfirmation of target gene knockdown or overexpression, individual clones were pooled for subsequent experiments.
Retroviral transduction of tumor cell lines with HLA-A*0201 and peripheral blood lymphocytes with T-cell receptor (TCR) genes against NY-ESO-1 or MAGE-A3 was carried out. H1299 or H841 lung cancer cells, SAECs, HBECs, or NHDFs were transduced with a retroviral vector expressing cDNA of HLA-A*0201 (23). H1299 and H841 cell lines stably expressing HLA-A*0201 were expanded under G418 selection. Peripheral blood lymphocytes (PBL) expressing HLA-A*0201–restricted TCRs recognizing NY-ESO-1 or MAGE-A3 were generated as described (14, 15).
Cytokine release assays
Drug-treated or control tumor cells with or without HLA-A*0201 expression were cocultured with untransduced or MAGE-A3 or NY-ESO-1 TCR-transduced lymphocytes; IFN-γ secretion in supernatants was measured by ELISA as described (14, 15).
Chromium release assays
The ability of NY-ESO-1- or MAGE-A3–specific TCR-transduced PBL to lyse HLA-A*0201+ lung cancer or normal lung cell targets was measured using chromium (51Cr) release assays. Briefly, after DAC, DZNep, or DAC–DZNep exposure, H2087 and H841-A*0201 or H1299-A*0201 and their respective controls were cocultured with effector cells, with subsequent analysis of 51Cr release as described (15, 23).
Expression profiles of CT-X genes in lung cancer cells
Preliminary qRT-PCR experiments were conducted to examine CT-X gene expression in cultured lung cancer cells as well as normal or immortalized respiratory epithelial cells (Supplementary Table S3). This analysis, which revealed heterogeneous CT-X gene expression in NSCLC and small cell lung cancer (SCLC) cells, but not normal respiratory epithelia, allowed us to choose several cell lines for further study (Fig. 1A; left). Relative to control testis cells, H1299 cells exhibit high-level expression of NY-ESO-1, MAGE-A1, and MAGE-A3. In contrast, H841 cells do not express NY-ESO-1, MAGE-A1, or MAGE-A3. A549 and Calu-6 cells exhibit moderate levels of MAGE-A3, but do not express NY-ESO-1 or MAGE-A1. NHBE cells and SAEC do not express any CT-X genes. Immunoblot analysis (Fig. 1A; right) confirmed results of these qRT-PCR experiments.
Chromatin structure relative to CT-X gene expression in lung cancer cells
Pyrosequencing and ChIP experiments were undertaken to examine DNA methylation and a variety of histone marks in lung cancer cells exhibiting differential CT-X gene expression. Pyrosequencing experiments (Fig. 1B) revealed that the NY-ESO-1 and MAGE-A1 promoters were hypermethylated in A549, Calu-6, and H841 lung cancer cells and NHBE cells; these promoters were hypomethylated in H1299 cells. In contrast, the MAGE-A3 promoter was hypermethylated in H841 and NHBE cells, partially methylated in A549 and Calu-6 cells, and demethylated in H1299 cells. These findings were consistent with results of previously described qRT-PCR and immunoblot experiments. Although demethylation of MAGE-A3 appeared to coincide with demethylation of D4Z4, no consistent relationship was evident with regard to NY-ESO-1, MAGE-A1, or MAGE-A3 promoter demethylation and global DNA demethylation assessed by pyrosequencing of NBL2, D4Z4, and LINE-1 repetitive DNA sequences, possibly attributable to incomplete analysis of these regions by pyrosequencing methods.
ChIP experiments were conducted to further investigate epigenetic phenomena associated with repression/activation of NY-ESO-1, MAGE-A1, and MAGE-A3 in lung cancer cells. As shown in Fig. 1C, the NY-ESO-1, MAGE-A1, and MAGE-A3 promoters in H1299 cells exhibited increased occupancy of RNA polymerase II (Pol II), enrichment of euchromatin/activation marks, such as H3K4Me2, H3K4Me3, H3K79Me2, total H3Ac, H3K9Ac, total H4Ac, and H4K16Ac, and decreased occupancy of SirT1 as well as polycomb repressor complex (PRC)-2 components (KMT6, EED, and SUZ-12), and the associated PRC-2–mediated repression mark, H3K27Me3. In contrast, RNA Pol II and histone-activation marks were markedly diminished, whereas SirT1, PRC-2 components, and H3K27Me3 levels were considerably higher within the NY-ESO-1 and MAGE-A1 promoters in A549, Calu-6, H841, and NHBE cells, all of which do not express these CT-X genes. Variable levels of RNA Pol II and activation/repression marks were present within the MAGE-A3 promoter in A549, Calu-6, H841, and NHBE cells, consistent with levels of expression of this CT-X gene in these cells. No consistent relationship was observed between activation/repression of these CT-X genes and H3K9Me3, previously considered to be a mark of stable, silenced heterochromatin (24), but more recently shown to coincide with RNA Pol II–mediated gene activation (25). Densitometry results of these ChIP experiments are summarized in Supplementary Table S4. Collectively, these experiments established that differential repression of NY-ESO-1, MAGE-A1, and MAGE-A3 in lung cancer cells is attributable to persistence of apparently normal heterochromatin structure within the promoters of these CT-X genes. Furthermore, levels of euchromatin marks, particularly H3K79Me2, appear to coincide with magnitude of CT-X gene derepression in lung cancer cells.
Effects of histone lysine methylation on CT-X gene expression
Our previous studies have shown that pharmacologic inhibition of histone lysine deacetylation enhances CT-X gene activation by DNA-demethylating agents (10). Therefore, additional experiments were conducted to ascertain whether modulation of histone lysine methylation alters CT-X gene expression in lung cancer cells. Briefly, lentiviral shRNA-transduction techniques were used to knock down LSD-1 (KDM1) and JARID1B (KDM5B) that mediate demethylation of mono-, di-, and trimethylated H3K4 (26, 27), or the histone lysine methyltransferase KMT6 that mediates trimethylation of H3K27 (28) in H841 cells; these cells were chosen for analysis because they do not express NY-ESO-1, MAGE-A1, or MAGE-A3. Preliminary qRT-PCR and immunoblot experiments showed approximately 50% to 70% reduction in target gene expression by the respective shRNAs relative to controls (Fig. 2A; left). Immunoblot analysis (Fig. 2A; right) showed decreased global H3K27Me3 and increased global H3K9Ac in KMT6 and SirT1 knockdowns, respectively, relative to control cells. Increased global levels of H3K4Me2 were evident in KDM1 and KDM5B knockdowns relative to control cells; interestingly, an increase in this activation mark was also observed in SirT1 knockdown cells. ChIP experiments revealed that global changes in these activation and repression marks tended to coincide with similar alterations and decreased occupancy of the respective histone modifiers in the NY-ESO-1, MAGE-A1, and MAGE-A3 promoters in knockdowns relative to control cells (Fig. 2B). Subsequent qRT-PCR experiments revealed that knockdown of KMT6, KDM1, or KDM5B alone was insufficient to activate NY-ESO-1, MAGE-A1, or MAGE-A3 in H841 lung cancer cells. However, knockdown of KMT6, KDM1, or KDM5B enhanced DAC-mediated induction of these CT-X genes approximately 3- to 11-fold in these cells (Fig. 2C); knockdown of KDM5B appeared to have the most effect with regard to potentiation of DAC-mediated CT-X gene activation in lung cancer cells. The effects of targeted modulation of histone lysine methylation appeared more pronounced than those observed following knockdown of the class III HDAC, SirT1.
Additional pyrosequencing experiments were conducted to ascertain whether modulation of histone lysine methylation affected DNA-methylation status of NY-ESO-1, MAGE-A1, and MAGE-A3 in DAC-treated and control lung cancer cells. Results of this analysis are depicted in Fig. 2D. Effects of histone methylation changes varied somewhat among the 3 CT-X genes. In general, the effects of KMT6, KDM1, or KDM5B knockdown on NY-ESO-1, MAGE-A1, or MAGE-A3 promoter methylation were modest, and did not directly coincide with magnitude of enhancement of DAC-mediated activation of these CT-X genes. A similar phenomenon was observed following knockdown of SirT1 in H841 cells.
Effects of DZNep on CT-X gene expression
Additional experiments were conducted to ascertain whether pharmacologic agents in preclinical development could recapitulate the previously described effects of histone lysine methylation on DAC-mediated activation of CT-X genes. Our studies focused on DZNep, a novel inhibitor of PRC-2 expression (29). Briefly, lung cancer cells were cultured for 72 hours in normal media with or without DAC (0.1 μmol/L), DZNep (0.5 or 5 μmol/L), or concurrent DAC–DZNep (0.1:0.5 μmol/L) followed by analysis after 24 hours. Preliminary immunoblot experiments showed that DZNep mediated dose-dependent depletion of KMT6, EED, and SUZ-12 with concomitant reduction in global H3K27Me3 levels in H841 cells (Fig. 3A; left); qRT-PCR experiments revealed that DZNep mediated modest dose-dependent reductions in KMT6 and SUZ12 but not EED mRNA levels (Fig. 3A; right). Additional experiments showed that low-dose DZNep (0.5 μmol/L—approximately 1 log lower than the cytotoxic dose of this agent in cancer cells) mediated very modest activation of NY-ESO-1, MAGE-A1, and MAGE-A3 in H841 cells; in contrast, DZNep significantly enhanced DAC-mediated CT-X gene activation in these cells (Fig. 3B). Immunofluorescence experiments confirmed that DZNep enhanced DAC-mediated expression of NY-ESO-1, MAGE-A1, and MAGE-A3 in H841 cells (Fig. 3C). This phenomenon extended to other CT-X genes such as MAGE-A12 (Fig. 3B), and was observed in other lung cancer lines (Supplementary Table S5). The magnitude of enhancement of DAC-mediated derepression of CT-X genes in cancer cells by DZNep was markedly higher than that observed in SAECs (Fig. 3B) or NHBE cells (data not shown). Relative to normal SAECs, immortalized HBECs appeared more responsive to DAC and DZNep; however, DZNep did not appear to augment DAC-mediated CT-X gene activation in these cells. The magnitude of DAC–DZNep–mediated CT-X gene induction in lung cancer cells approximated or exceeded that observed following sequential DAC–DP or DAC–TSA treatment; addition of TSA or DP did not consistently improve CT-X gene activation mediated by low-dose DAC–DZNep (Supplementary Table S5).
Effects of DZNep on DNA methylation and H3K27Me3 within CT-X gene promoters
Pyrosequencing and ChIP analyses were conducted to further examine the mechanisms by which DZNep modulates CT-X gene expression in lung cancer cells. Results of these experiments are depicted in Fig. 4. NHBE and H1299 cells were used as positive and negative methylation controls, respectively. As anticipated, DAC-mediated activation of NY-ESO-1, MAGE-A1, and MAGE-A3 coincided with significant demethylation of the respective promoters. In contrast to what was observed following histone lysine methyltransferase knockdown (Fig. 2D), DZNep alone mediated a modest, but significant, demethylation of all 3 CT-X gene promoters (Fig. 4A); in combination with DAC, DZNep exhibited an additive demethylation effect in the NY-ESO-1 and MAGE-A3 promoters (Fig. 4A). The effects of DAC, DZNep, or DAC–DZNep on NY-ESO-1, MAGE-A1, and MAGE-A3 promoters coincided with similar effects on global DNA methylation assessed by pyrosequencing analysis of NBL2, D4Z4, and LINE-1 sequences (Fig. 4B). Subsequent ChIP experiments confirmed that DZNep decreased KMT6 and H3K27Me3 levels within the NY-ESO-1, MAGE-A1, and MAGE-A3 promoters (Fig. 4C); the magnitude of decrease in KMT6 and H3K27Me3 levels appeared to coincide with extent of demethylation and derepression of these promoters.
Effects of KMT6 overexpression on CT-X gene activation mediated by DZNep
Additional experiments were undertaken to specifically examine whether the effects of DZNep on DAC-mediated activation of CT-X genes were attributable, at least in part, to depletion of KMT6. Briefly, H841 cells stably expressing KMT6 were treated with DAC, DZNep, or DAC–DZNep as previously described. Immunoblot analysis (Fig. 5A; left) showed increased global levels of KMT6 and H3K27Me3 in KMT6-transfected H841 cells relative to vector controls. DZNep markedly depleted KMT6 and H3K27Me3 levels in KMT6 overexpressors, and qRT-PCR experiments showed a modest, but statistically insignificant, diminution of KMT6 expression by DZNep (Fig. 5A; right). Additional qRT-PCR experiments revealed that overexpression of KMT6 significantly attenuated the enhancement effect of DZNep on DAC-mediated induction of NY-ESO-1, MAGE-A1, or MAGE-A3 (Fig. 5B).
Recognition of lung cancer cells by NY-ESO-1- and MAGE-A3-specific TCR-engineered T cells following DZNep exposure
Additional experiments were conducted to examine whether DZNep enhances immunogenicity of lung cancer cells. Briefly, H2087 lung cancer cells, which endogenously express HLA-A*0201, and SAECs (chosen because they proliferate faster than NHBE cells) and H841 cells transduced with HLA-A*0201 (SAEC-A2 and H841-A2, respectively) were exposed to NM, DAC, DZNep, or DAC–DZNep as previously described, and subsequently cocultured with TCR-engineered PBL recognizing NY-ESO-1 or MAGE-A3 in the context of HLA-A*0201. Representative results from 2 independent experiments conducted using PBL from 2 different patients are depicted in Fig. 6. For these experiments, H1299 and H1299-A2 cells served as negative and positive controls, respectively (Fig. 6A). As shown in Fig. 6B, increased IFN-γ release was observed following coculture of NY-ESO-1 and MAGE-A3 effector cells with H2087 and H841-A2 cells previously exposed to DAC, which was significantly augmented by concomitant exposure to DZNep (0.5 μmol/L). Very low-level cytokine release was observed following coculture of effector cells with DZNep-treated H2087 and H841-A2 targets. The magnitude of enhancement of DAC-mediated cytokine release by DZNep was more pronounced for MAGE-A3 relative to NY-ESO-1 effector cells; these results were consistent with qRT-PCR analysis of CT-X gene expression in target cells following drug treatment. Background levels of IFN-γ release were observed following coculture of effector cells with parental untreated H1299 cells, or drug-treated H841 cells lacking HLA-A*0201 expression. Effector cells did not recognize either drug-treated HLA-A*0201–transduced SAECs (Fig. 6B) or HBECs and NHDF (data not shown), presumably due to very low levels of NY-ESO-1 and MAGE-A3 induction in these cells by the treatment regimen (Supplementary Table S6).
Chromium release experiments were conducted to evaluate lysis of H841-A2 cells by MAGE-A3– or NY-ESO-1–specific effector cells. Representative results from 2 independent experiments conducted using PBL from 2 different donors are depicted in Fig. 7. H1299-A2 and parental H1299 cells served as positive and negative controls, respectively (Fig. 7A). Low-level lysis was observed following coculture of untransduced effector cells with H841 targets possibly due to nonspecific alloreactivity, recognition of tumor targets by endogenous T-cell receptors, presence of natural killer cells, and mild toxicity of the drug-treatment regimens (Fig. 7B). Compared with untreated controls, DAC-treated H841-A2 cells were more efficiently lysed by the effector cells (Fig. 7C). Interestingly, DZNep treatment also led to increased lysis of H841-A2 cells, whereas the percentage of specific lysis of H841-A2 cells treated with DAC exceeded that observed following treatment of target cells with DZNep for NY-ESO-1 effector cells, percent lysis following exposure of tumor targets to DZNep was comparable to that observed following treatment with DAC when tumor targets were cocultured with MAGE-A3 effector cells. Concurrent DAC–DZNep treatment of target cells markedly enhanced the percentage of specific lysis mediated by NY-ESO-1 or MAGE-A3 effector cells. The magnitude of lysis of DAC-, DZNep-, or DAC–DZNep-treated H841-A2 cells by MAGE-A3 effector cells exceeded that observed for NY-ESO-1 effector cells, possibly due to simultaneous upregulation of other MAGE-A genes such as MAGE-A12 encoding HLA-A*0201-restricted epitopes recognized by the genetically engineered MAGE-A3 TCR (15). Specific lysis of H841-A2 cells by NY-ESO-1 and MAGE-A3 effector cells corresponded with mRNA copy numbers (Supplementary Table S6) and IFN-γ release observed in coculture assays.
DNA methylation is the major epigenetic mechanism silencing CT-X genes in normal somatic cells (30, 31). Whereas CT-X gene expression can be induced in cancer cells by DNA-demethylating agents (10) or simultaneous knockdown of DNMT1 and DNMT3b (32, 33), derepression of CT-X genes during malignant transformation cannot be attributed solely to global DNA demethylation. Transfected methylated MAGE-A1 transgenes do not undergo promoter demethylation, and unmethylated MAGE-A1 transgenes become methylated except for the 5′-region in cancer cells (34). Complex chromatin architecture including formation of double cruciform DNA (35) that potentially affects access of methyl-binding proteins, DNA methyltransferases, and transcription factors such as CTCF, BORIS, and SP1 (8, 32, 36) may contribute to coordinated repression/activation of CT-X genes within large inverted repeats (7).
In the present study, we sought to examine the feasibility of modulating histone lysine methylation as a strategy to enhance CT-X gene activation by DNA-demethylating agents under conditions potentially achievable in clinical settings (16). Our experiments showed that knockdown of KMT6, KDM1, and KDM5B significantly enhances DAC-mediated activation of NY-ESO-1 and several MAGE-A genes in lung cancer cells. Whereas knockdown of KMT6, KDM1, and KDM5B coincided with decreased occupancy of these histone lysine methyltransferases and their respective marks within the NY-ESO-1, MAGE-A1, and MAGE-A3 promoters, our data do not exclude the possibility that depletion of these histone modifiers facilitates CT-X gene activation via mechanisms independent of inhibition of methyltransferase activity (37).
Originally developed as an antiviral agent (38), DZNep has been shown to deplete KMT6, EED, and SUZ12 primarily via proteolytic mechanisms leading to growth arrest, differentiation, or apoptosis in cancer cells depending on histology and genotype (refs. 29, 39–42; Kemp and colleagues, unpublished data). Of particular interest, tumor-initiating cells appear exquisitely sensitive to DZNep, due to the critical role of polycomb proteins in maintenance of cancer stem cells (43). In addition to decreasing global H3K27Me3 levels, DZNep diminishes numerous other repressive and activation histone lysine methylation marks such as H3K9Me2 and H3K4Me3, respectively (44). DZNep reactivates genes silenced by polycomb mechanisms; however, despite the fact that DZNep exhibits mild DNA-demethylating effects, this agent is insufficient to derepress hypermethylated genes (44). Our analysis revealed that low-dose DZNep alone did not activate NY-ESO-1, MAGE-A1, or MAGE-A3 in lung cancer cells, but significantly enhanced DAC-mediated induction of these CT-X genes. Although our experiments suggested that enhancement of DAC-mediated CT-X gene induction by DZNep is attributable, in part, to depletion of KMT6, the precise mechanisms underlying this phenomenon have not been fully defined and are a focus of ongoing experiments.
Deciphering the mechanisms mediating derepression of CT-X genes in cancer cells may provide fundamental insights about malignant transformation and facilitate development of novel strategies for epigenetic therapy for cancer. Our observations that DZNep enhances DAC-mediated upregulation of NY-ESO-1 and MAGE-A family members, and markedly augments recognition and lysis of lung cancer cells by T cells specific for these CTAs, have direct translational implications about the development of gene-induction regimens for cancer immunotherapy. Our findings pertaining to the lack of CT-X gene induction in normal cells following DAC–DZNep exposure are consistent with our previously published data showing negligible activation of CT-X genes in SAECs or NHBE cells by DAC/DP (13, 14) or in normal tissues from lung cancer patients receiving these agents (16, 45). The fact that the magnitude of DAC–DZNep (as well as DAC/DP)-mediated CT-X gene induction is more pronounced in HBECs relative to SAECs, NHBE cells, or NHDFs, but less than in lung cancer cells with similar proliferation rates, suggests that global methylation changes associated with malignant transformation (18) contribute, in part, to the relative sensitivity of cancer cells to epigenetic treatment regimens. Although the mechanisms underlying this intriguing phenomenon remain elusive and are a focus of ongoing investigation, our data support further development of DZNep and other inhibitors of histone lysine methylation for cancer immunotherapy.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The authors thank Ms. Jan Pappas for assistance with the preparation of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received July 7, 2010.
- Revision received April 7, 2011.
- Accepted April 14, 2011.
- ©2011 American Association for Cancer Research.