Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Aug 27.
Published in final edited form as: Cell. 2015 Aug 27;162(5):961–973. doi: 10.1016/j.cell.2015.07.056

DNA-demethylating agents target colorectal cancer cells by inducing viral mimicry by endogenous transcripts

David Roulois 1, Helen Loo Yau 1,2, Rajat Singhania 1, Yadong Wang 1, Arnavaz Danesh 1, Shu Yi Shen 1, Han Han 4, Gangning Liang 4, Trevor J Pugh 1,2, Peter A Jones 4,5, Catherine O’Brien 1,3, Daniel D De Carvalho 1,2
PMCID: PMC4843502  NIHMSID: NIHMS774731  PMID: 26317465

Summary

DNA demethylating agents have shown clinical anti-tumor efficacy via an unknown mechanism of action. Using a combination of experimental and bioinformatics analysis in colorectal cancer cells, we demonstrate that low-dose 5-AZA-CdR targets colorectal cancer initiating cells (CICs) by inducing viral mimicry. This is associated with induction of dsRNAs derived at least in part from endogenous retroviral elements, activation of the MDA5/MAVS RNA recognition pathway and downstream activation of IRF7. Indeed, disruption of virus recognition pathways, by individually knocking down MDA5, MAVS or IRF7, inhibits the ability of 5-AZA-CdR to target colorectal CICs and significantly decreases 5-AZA-CdR long-term growth effects. Moreover, transfection of dsRNA into CICs can mimic the effects of 5-AZA-CdR. Together; our results represent a major shift in understanding the antitumor mechanisms of DNA-demethylating agents and highlight the MDA5/MAVS/IRF7 pathway as a potentially druggable target against CICs.

ETOC

Antitumor DNA demethylating agents act by inducing transcription of endogenous dsRNAs that activate the viral recognition and interferon response pathway This anti-viral response reduces the proliferation of colorectal cancer initiating cells.

INTRODUCTION

Many tumor types are organized as a cellular hierarchy sustained by a subpopulation of cancer-initiating cells (CICs) (Kreso and Dick, 2014). These CICs possess unique features, including long-term self-renewal, ability to initiate tumor growth in xenograft models and to differentiate into the bulk of the tumor mass (Kreso and Dick, 2014). Colorectal cancer follows this hierarchical model and contains CICs (Dalerba et al., 2007; O’Brien et al., 2012; O’Brien et al., 2007; Ricci-Vitiani et al., 2007). Colorectal CICs are believed to play a major role in tumor relapse and patient survival; suggesting that therapeutic strategies targeting this cell population would be highly beneficial to patient outcome (Kreso et al., 2014; Merlos-Suarez et al., 2011).

Recent works suggest that CICs can be targeted by epigenetic therapies, including experimental BMI-1 inhibitors in colorectal cancers and FDA-approved DNA methylation inhibitors such as 5-aza-2-deoxycytidine (5-AZA-CdR) in hematological malignancies (Kreso et al., 2014; Tsai et al., 2012). 5-AZA-CdR is a cytidine analog that traps DNA methyltransferases after incorporation into DNA, resulting in proteasomal degradation and global DNA demethylation (Kelly et al., 2010).

There is an ongoing debate about the molecular mechanisms underlying the clinical efficacy of 5-AZA-CdR and whether it represents a direct consequence of its effect on global demethylation (Issa and Kantarjian, 2009). Previous data suggests that promoter demethylation followed by gene re-activation of aberrantly methylated Tumor Suppressor Genes (TSGs) is a major mechanism of 5-AZA-CdR antitumor effects (Navada et al., 2014). This is in line with the finding that many TSG promoters are hypermethylated in cancer cells and that cancer cells become epigenetically addicted to DNA hypermethylation of a set of TSGs (De Carvalho et al., 2012). More recently, we demonstrated that gene-body DNA demethylation, followed by gene repression of oncogenic pathways may also play a role in cancer response to 5-AZA-CdR (Yang et al., 2014). However, the prolonged time to response observed in patients and the fact that global DNA methylation profiling, neither pretreatment nor during treatment, can predict response (Treppendahl et al., 2014) suggests that the major molecular mechanisms underlying the clinical efficacy of 5-AZA-CdR may lie beyond demethylation of TSG promoters and oncogene gene-bodies.

Here we describe that transient low dose exposure to 5-AZA-CdR targets colorectal CICs by inducing dsRNA expression, activation of the cytosolic pattern recognition receptor MDA5 and downstream activation of MAVS and IRF7. This will lead to a ‘viral mimicry’ state and is a major molecular mechanism for the anti-tumor effect mediated by transient low dose of 5-AZA-CdR.

RESULTS

Transient, low dose 5-AZA-CdR induces four different profiles of gene expression

We have recently shown that transient treatment of HCT116 colorectal cancer cells with low dose (0.3uM) of 5-AZA-CdR for 24 hours followed by cell culture in a drug-free medium has a long-term effect on population doubling time and colony formation ability (Yang et al., 2014). Here, we used the same experimental system to test the effects of transient treatment of 5-AZA-CdR on gene expression patterns. We monitored gene expression patterns before and after treatment up to 42 days after drug withdrawal. Using gene expression microarray and consensus-clustering to classify the most variable genes, we identified four patterns of gene expression (Figure 1A and Figure S1A–C). Group 1 and 2 are early response genes that are either down-regulated (Group 1) or up-regulated (Group 2) within 5 days of 5-AZA-CdR treatment, but return to the original expression level by the end of the follow-up period (42 days) (Figure 1A). Using DNA methylation BeadChips, we observed that Group 1 genes have a strong positive correlation between expression and DNA methylation (Figure 1B–C), with most of the DNA methylation changes occurring at gene bodies. Group 2 genes have a strong negative correlation between expression and DNA methylation (Figure 1B–C). Group 3 is a small and noisy group presenting no correlation between expression and DNA methylation (Figure 1A–C). Group 4 genes are late response genes where peak gene expression occurs 24 days after initial exposure to 5-AZA-CdR and higher gene expression levels are observed even 42 days after drug withdrawal (Figure 1A). This delayed activation and sustained gene expression is compelling, since most clinical responses to 5-AZA-CdR are delayed (Treppendahl et al., 2014). Group 4 genes have a weaker anti-correlation between expression and DNA methylation, with a fat tail around the weak and no correlation region (Correlation coefficient, r, between −0.5 and +0.5, Figure 1B–C), suggesting that several genes in this group are activated by 5-AZA-CdR in the absence of direct changes in DNA methylation of their promoters or coding regions. Most of the genes in this group have low DNA methylation levels pre-treatment (Figure 1C), suggesting that their activation by drug treatment does not involve promoter DNA demethylation.

Figure 1. Transient low dose treatment of 5-AZA-CdR induces durable and DNA demethylation-independent activation of a gene-set enriched for interferon responsive genes and RIG-1 pathway.

Figure 1

Open in a new tab

HCT116 cells were transiently treated with 0.3uM of 5-AZA-CdR for 24 hours (1 population doubling time). RNA was extracted before treatment and at day 5, 14, 24 and 42 after drug withdrawal and subjected to cDNA microarray. A) Expression profile for each of the four identified clusters. The solid line represents the median value for each group, while the dotted lines are the lower quartile (lower dotted line) and the upper quartile (upper dotted line) for each group. B) Kernel density plots showing the Pearson correlation (r) between gene expression and DNA methylation for each probe across the 42 days time-course experiment. The y-axis indicates the density of probes. C) Selected examples of genes from Group 1, Group 2, Group 3, and Group 4 are shown. Left y-axis represents the average expression level, and right y-axis represents the average DNA methylation level. The x-axis denotes time (in days) after 5-AZA-CdR withdrawal. Error bars represent the standard deviation D) Three most significant Canonical Pathways enriched at group 4 genes using Ingenuity Pathway Analysis. The significance value for each canonical pathway was calculated by Fisher’s exact test right-tailed. Left y-axis displays the -log of p-value. The orange points represent the ratio between the number of genes in Group 4 that belongs to that specific pathway by the total number of genes annotated to that specific pathway. E) Diagram of the canonical pathway entitled “Role of RIG1-like Receptors in Antiviral Innate Immunity”. Genes highlighted in pink belongs to Group 4. IPS-1 (MAVS) is a central node of the pathway, linking RIG-1/MDA5 receptors to IRF7. See also Figure S1.

Late-response genes are enriched for Interferon Responsive genes and the RIG1/MDA5 RNA sensing Pathway

To identify the most significant canonical pathways associated with the late responsive genes (Group 4), we performed Ingenuity Pathway Analysis. Using the Fisher’s right-tailed exact test we identified ‘Interferon Signaling’ (P value = 5.35−15, Figure S1D), ‘Activation of IRF by Cytosolic Pattern Recognition Receptors’ (P value = 2.37−12, Figure S4) and ‘Role of RIG-1-like receptors in Antiviral Innate Immunity’ (P value = 5−06, Figure 1E) as the top 3 enriched canonical pathways at the Ingenuity knowledge database (Figure 1D). Interestingly, recent work reporting activation of interferon responsive genes and the RIG1/MDA5 pathway after patient treatment with DNA methylation inhibitors at low doses (Wrangle et al., 2013) highlights the clinical relevance of our model system.

Although most of the genes in Group 4 are genes responsive to type I interferon (Figure 1D and Figure S1D), we could not detect any IFNα or IFNβ transcripts or protein as measured by qPCR and ELISA in HCT116, LIM1215 and HT29 colorectal cancer cells before or after transient low dose of 5-AZA-CdR treatment. Thus, suggests that the cytosolic pattern recognition receptors RIG1/MDA5 may be initiating the signaling cascade rather than type I interferon. To test this hypothesis, we took advantage of the fact that the RIG1/MDA5 signaling cascade depends on the adaptor molecule MAVS (mitochondrial antiviral signaling, also known as IPS1, CARDIF and VISA) (Barbalat et al., 2011). MAVS is an essential adaptor molecule that connects RIG1 and MDA5 activation to the downstream activation of IRF3 and IRF7 (Figure 1E). Previous work has established that MAVS knockdown blocks the RIG1/MDA5 signaling pathway (Barbalat et al., 2011), allowing us to test the effect of low dose 5-AZA-CdR treatment when the RIG1/MDA5 signaling pathway was impaired First, we treated the B16-Blue™ IFN-α/β Cells (InvivoGen) with 5-AZA-CdR for 24 hours. B16-Blue cells are stably transfected with a SEAP (Secreted embryonic alkaline phosphatase) reporter gene under the control of ISG54 promoter enhanced by a multimeric ISRE (Interferon-sensitive response element) and will respond to activation of STAT1, STAT3, IRF3, IRF7 and IRF9. Therefore, B16-Blue cells can detect bioactive type I IFNs by monitoring activation of STAT1/2 (Figure S1D); in addition they can detect activation of RIG1/MDA5 pathway by monitoring activation of IRF3/7 (Figure 1E).

We observed that upon transient low dose 5-AZA-CdR exposure, B16-Blue cells initiate a strong activation of the SEAP reporter gene (Figure 2A). This 5-AZA-CdR-mediated activation of the SEAP reporter gene was blocked by MAVS knockdown (Figure 2A and Figure S2A–B). This data highlights the importance of RIG1/MDA5 cytosolic receptors in inducing the activities of interferon-responsive promoters in response to 5-AZA-CdR treatment. To test this further, we incubated untreated B16-Blue cells with supernatant derived from wild-type B16 cells after 5-AZA-CdR treatment. The supernatant alone was not able to activate the SEAP reporter gene (Figure 2B), suggesting a lack of bioactive, type I IFNs in the supernatant of wild-type B16 cells treated with 5-AZA-CdR. However, wild-type B16 cells transfected with poly(I:C) (positive control), were able to release type I IFNs in the supernatant, as shown by SEAP activation on B16-Blue cells (Figure 2B). To test whether other cytokines could be playing a role in inducing the expression of interferon-responsive genes in response to 5-AZA-CdR, we treated the B16-Blue cells with 5-AZA-CdR in the presence or absence of Ruxolitinib (JAK1/JAK2 inhibitor) or CP-690550 (JAK3 inhibitor). In the presence of either Ruxolitinib or CP-690550, 5-AZA-CdR treatment was no longer able to activate the SEAP reporter gene (Figure 2C). Since we could not detect type I IFNs and the B16-Blue cells do not respond to IFN-γ due to the inactivation of the IFN-γ receptor, this data suggests a potential role of type III IFNs.

Figure 2. Transient low dose treatment of 5-AZA-CdR induces Type III Interferon production and activation of MAVS.

Figure 2

Open in a new tab

A) B16-Blue™ IFN-α/β Cells (InvivoGen) were transiently treated with 0.3uM of 5-AZA-CdR for 24 hours. Activity of the ISG54 promoter enhanced by a multimeric ISRE was determined by measuring secreted alkaline phosphatase (SEAP). SEAP activity was measured at 640 nm when the QUANTI-blue substrate was provided. The results presented are from three independent experiments. Error bars represent the SD of three independent experiments. B) Supernatants from B16 cells transiently treated with 5-AZA-CdR or transfected with 0.5ug/ml of poly(I:C) were added to wild-type B16-Blue™ IFN-α/β cells (InvivoGen), and the presence of bioactive IFNs-α/β were determined by measuring SEAP activity 24 hours after adding the supernatants. Error bars represent the SD of three independent experiments. C) B16-Blue™ IFN-α/β Cells (InvivoGen) were transiently treated with 5-AZA-CdR in the presence or absence of the JAK1/JAK2 inhibitor Ruxolitinib (1uM) or the JAK3 inhibitor CP-690550 (1uM). SEAP activity was measured at 640nm. D) LIM1215 colorectal cancer cell line was transiently treated with 5-AZA-CdR in the presence or absence of the JAK1/JAK2 inhibitor Ruxolitinib (1uM) or the JAK3 inhibitor CP-690550 (1uM). Gene expression of four selected interferon responsive genes was measured by real-time qPCR at day 5. E-F) LIM1215 with or without shRNA against MAVS were transiently treated with 5-AZA-CdR. Gene expression of the Type III interferon genes IL29 (E) and IL28a (F) was measured by real-time qPCR at day 5. G) Mitochondrial extracts were prepared from LIM1215 cells treated with 5-AZA-CdR for the indicated time, or transfected with Poly(I:C), and then aliquots of the extracts were analyzed by SDD-AGE. **** p<0.0001; ** p<0.01 (One-way ANOVA) See also Figure S2.

Similarly to the B16 murine model, we observed that co-treatment of 5-AZA-CdR with Ruxolitinib or CP-690550 was sufficient to block the ability of 5-AZA-CdR to induce the expression of four interferon-responsive genes (Group 4) in the human colorectal cancer cell LIM1215 (Figure 2D), suggesting a potential role for type III IFNs. Indeed, we found that low dose 5-AZA-CdR treatment was able to significantly increase the expression of the type III IFN IL29 in LIM1215, HCT116, and HT29 human colorectal cancer cell lines (Figure 2E and Figure S2G–I). 5-AZA-CdR treatment also significantly increased the expression of the type III IFN IL28a in the LIM1215 cell line (Figure 2F and Figure S2H–J). Altogether, our data suggest that transient low dose 5-AZA-CdR treatment can induce the expression of type III, but not type I, IFNs, followed by a downstream activation of Interferon Stimulated Genes (ISGs) in a JAK-dependent manner.

Interestingly, when we knocked down MAVS (Figure S2C–F), we were able to repress the ability of 5-AZA-CdR treatment to induce type III IFNs (Figure 2E–F and Figure S2G–J), implicating an upstream role for the cytosolic pattern recognition receptors RIG1/MDA5 in initiating the signaling cascade induced by 5-AZA-Cd treatment that will culminate with type III IFNs and increase in expression of ISGs.

The ability of the mitochondrial protein MAVS to transduce the signal from the cytosolic pattern recognition receptors RIG1/MDA5 and activation of downstream IRFs to induce the expression of IFNs is dependent on the formation of very large prion-like aggregates of MAVS protein (Hou et al., 2011). To test whether the 5-AZA-CdR treatment is indeed inducing activation of RIG1/MDA5 and downstream signaling through MAVS, we isolated the mitochondrial fraction of LIM1215 treated with 5-AZA-CdR or Poly(I:C) (positive control) (Figure S2K) and performed a semi-denaturing detergent agarose gel electrophoresis (SDD-AGE), which was previously shown to detect the formation of MAVS prion-like aggregates (Hou et al., 2011). We observed a smear of SDS-resistant high-molecular weight MAVS aggregates in our positive control as well as the treated cells starting one day after 5-AZA-CdR treatment (Figure 2G), indicating a functional activation of MAVS.

Activation of late-response genes is dependent on MDA5/MAVS/IRF7 pathway activation

Next, we sought to investigate the individual contributions of the cytosolic pattern recognition receptors RIG1 and MDA5. We knocked-down RIG1 and MDA5 individually in LIM1215 cells using validated shRNAs (Figure 3A). MDA5 knockdown was sufficient to inhibit the activation of Group 4 genes after 5-AZA-CdR treatment, while RIG1 knockdown was able to partially reduce the activation of Group 4 genes (Figure 3A). This result suggests a major role for MDA5 in mediating the downstream activation of MAVS after 5-AZA-CdR treatment. Indeed, when we knocked-down MAVS (Figure S2C–F) we were also able to inhibit the activation of Group 4 genes after 5-AZA-CdR treatment (Figure 3B and Figure S3A–B). It is important to note that knockdown of MAVS had no effect on the ability of 5-AZA-CdR to induce global DNA demethylation (Figure S5E–G).

Figure 3. Activation of interferon responsive genes by transient low dose treatment of 5-AZA-CdR is dependent on MDA5/MAVS/IRF7 activation.

Figure 3

Open in a new tab

LIM1215 colorectal cancer cell line with or without shRNA against RIG1 (A), MDA5 (A), or MAVS (B) was transiently treated with 5-AZA-CdR. Gene expression of four selected interferon responsive genes was measured by real-time qPCR at day 5. C) Upstream Regulators Analysis using Ingenuity Pathway Analysis reveals a significant overlap between Group 4 genes and IRF7 direct target genes (p-value: 5.09e-15). D) Confocal microscopy of LIM1215 cells after treatment with transient low dose of 5-AZA-CdR. Total IRF7 is stained in green, and nuclei are stained in blue (DAPI). There is increased IRF7 expression and nuclear translocation in 5-AZA-CdR treated WT cells. E) LIM1215 colorectal cancer cell line with or without shRNA against IRF7 was transiently treated with 5-AZA-CdR. Gene expression of four selected interferon responsive genes was measured by real-time qPCR at day 5. Y-axis represents the relative expression (in % of control) compared to the wild-type untreated control. Error bars represent the SD of at least three independent experiments. ns, non-significant; * p<0.05; ** p<0.01; **** p<0.0001 (One-way ANOVA). See also Figure S3.

MAVS activation is known to activate several downstream transcription factors such as IRF1, IRF3 and IRF7 (Odendall et al., 2014; Seth et al., 2005). Next, we investigated which transcription factors are activated by MAVS in response to 5-AZA-CdR treatment. First, we performed IPA’s Upstream Regulators Analysis (Kramer et al., 2014) to identify transcription factors predicted to be activated based on the activation z-score of Group 4 genes. We found IRF7 as one of the top predicted transcription factors, with 20 known IRF7 direct target genes also being upregulated in Group 4 gene-set out of a total of 98 known IRF7 direct targets (p-value of overlap: 5.09−15, Figure 3C).

To validate that IRF7 was upregulated and activated, we performed confocal microscopy on the LIM1215 colorectal cancer cell line following 5-AZA-CdR treatment and observed increased protein levels of IRF7 and a robust nuclear translocation (Figure 3D, second row), indicating activation of this transcription factor. Moreover, in the presence of an impaired MDA5/MAVS pathway, by knockdown of MAVS, 5-AZA-CdR treatment is no longer sufficient to induce either IRF7 upregulation or activation, as measured by nuclear translocation (Figure 3D, fourth row). This data suggests that activation of interferon-responsive genes by 5-AZA-CdR treatment is mediated by MDA5/MAVS-dependent IRF7 activation, rather than by direct promoter DNA demethylation. Indeed, Group 4 genes downstream of IRF7 have a significantly weaker anti-correlation between DNA methylation and gene expression than the rest of Group 4 genes (Figure S3C, Kolmogorov-Smirnov test of Frequency distribution data, p-value=0.0004). Furthermore, we evaluated the DNA methylation levels of Group 4 IRF7 downstream genes (Figure 3C) in a set of 125 colorectal cancer samples from different subgroups and 29 adjacent normal samples (Hinoue et al., 2012). Consistent with our previous data, these genes are largely unmethylated in primary samples, independent of the tumor/normal status and independent of the CIMP (CpG Island Methylator Phenotype) status (Figure S3D). Moreover, 5-AZA-CdR treatment did not induce IRF1 protein expression (Figure S3E), nor did it increase the levels of IRF3 (Figure S3F), or induce IRF3 nuclear translocation (Figure S3F), suggesting a major role for IRF7. To confirm the role of IRF7, we knocked-down this transcription factor and observed that it was sufficient to inhibit the activation of Group 4 genes after 5-AZA-CdR treatment (Figure 3E).

Altogether, our data suggest that transient low dose 5-AZA-CdR treatment can induce activation of interferon-responsive genes by activating the MDA5/MAVS/IRF7 signaling pathway in a process independent of promoter DNA demethylation of interferon-responsive genes.

5-AZA-CdR treatment induces an increase in dsRNAs

MDA5 is a cytosolic pattern-recognition receptor that recognizes dsRNAs associated with virus infections (Barbalat et al., 2011). Therefore, we investigated whether 5-AZA-CdR could be inducing a significant increase in dsRNAs that could explain the activation of MDA5/MAVS/IRF7 pathway. Using the J2 antibody, a gold standard for dsRNA detection (Weber et al., 2006), confocal microscopy showed a significant increase in dsRNA in LIM1215 cells starting 3 days after 5-AZA-CdR withdrawal (Figure 4A–D), which persisted for up to 10 days after drug withdrawal (Figure 4A–D). Moreover, the second highest enriched pathway, identified in Figure 1D, has two components: ‘Regulation of innate immune response by RNA sensing molecules’ and ‘Regulation of innate immune response by DNA sensing molecules’. Remarkably, only the genes belonging to the RNA sensing pathway were up-regulated by transient low dose 5-AZA-CdR treatment (Figure S4A–B).

Figure 4. Increased dsRNA by transient low dose treatment of 5-AZA-CdR.

Figure 4

Open in a new tab

A) Confocal microscopy of LIM1215 cells 3 days after treatment with transient low dose of 5-AZA-CdR. Total dsRNA is stained in red, and nuclei are stained in blue (DAPI). There is increased dsRNA expression in the cytoplasm of 5-AZA-CdR treated cells. B-D) Quantification of the dsRNA performed by Measuring Cell Fluorescence using ImageJ software 3 days (B), 5 days (C) or 10 days (D) after 5-AZA-CdR treatment. Corrected total cell fluorescence (CTCF) was calculated using the following formula. CTCF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings). **** p<0.0001, * p<0.05 (two-tailed T test). E) LIM1215 was transiently treated with 5-AZA-CdR. Expression level of ten selected ERVs was measured by real-time qPCR at day 5. **** p<0.0001; *** p<0.001; * p<0.05 (Two-way ANOVA) F) Total RNA was extracted from 5-AZA-CdR treated cells. RNA was then digested or not with 50ug/ml RNase A in high salt concentration (NaCl 0.35M) for 30 minute. Enrichment of dsRNA over ssRNA was then calculated by normalizing the delta Ct between RNAse A treated and non-treated of ERVs (dsRNA) against beta-actin (ssRNA). See also Figure S4.

It was recently reported that endogenous retrovirus (ERV) RNAs can trigger signaling by cytosolic pattern-recognition receptors and activate MAVS in mammals (Zeng et al., 2014). Therefore, we investigated whether ERVs could be the source of 5-AZA-CdR induced dsRNA and the trigger for the antiviral response. Indeed, we observed a very robust and significant increase in transcription of human ERVs from classes previously described to trigger a MAVS-mediated response (Zeng et al., 2014) (Figure 4E). Then, using an RNAse A protection assay under high salt concentration (see methods), we observed that ERVs transcripts were hundreds to thousands of times more resistant to RNAse A digestion than a ssRNA gene (beta actin) in 5-AZA-CdR treated cells (Figure 4F).

5-AZA-CdR effects on tumor cell growth are largely dependent on the MDA5/MAVS/IRF7 pathway

Transient low dose 5-AZA-CdR treatment can decrease tumor cell growth and reduce colony formation (Kelly et al., 2010; Tsai et al., 2012; Yang et al., 2014). In order to test whether these antitumor effects are mediated by activation of MDA5/MAVS/IRF7 pathway, we treated colorectal cancer cell lines with or without individual knockdowns of MDA5, MAVS, and IRF7. Population doubling time was monitored for up to 20 days and we observed that the 5-AZA-CdR treatment exerted a durable increase in the population doubling time in all the three colorectal cancer cell lines as expected (Figure 5A–C and Figure S5A–C). However, knockdown of MDA5, MAVS, or IRF7 was sufficient to render these cells insensitive to 5-AZA-CdR treatment (Figure 5A–C and Figure S5A–C), even though the drug could still induce global DNA demethylation in the knockdown cells (Figure S5E–G). Important to note, MAVS knockdown alone has no major effect on population doubling time (Figure S5D). Interestingly, RIG1 knockdown was not sufficient to render these cells insensitive to drug treatment (Figure S5A), suggesting, again, a major role for MDA5 rather than RIG1 in the response to 5-AZA-CdR treatment. This data suggests that the long-term effects on population doubling time are largely dependent on activation of the MDA5/MAVS/IRF7 pathway.

Figure 5. Transient 5-AZA-CdR treatment prolonged effect on cell growth and self-renew is dependent on MDA5/MAVS/IRF7 activation.

Figure 5

Open in a new tab

A-C) Population doubling time of LIM1215 with or without MDA5 (A), MAVS (B), and IRF7 (C) knockdown. The y-axis denotes population-doubling time (in hours) as a percentage of the vehicle-treated WT controls. The x-axis represents time (in days) after 5-AZA-CdR was withdrawal. Error bars represent the SD of three independent experiments. D) Schematic representation of the approach used to measure the sphere-initiating frequency after transient low dose 5-AZA-CdR treatment (see Experimental Procedures). E-F) Frequency of LIM1215 CICs before and after low dose transient 5-AZA-CdR treatment measured by in-vitro limiting dilution assay in wild-type (E) or shMAVS (F) cells. The y-axis denotes the confidence intervals (lower, estimate and upper) for CIC frequency. G) Schematic representation of the approach used to measure the cancer-initiating cells frequency in vivo after transient low dose 5-AZA-CdR treatment (see Experimental Procedures). H-I) Frequency of LIM1215 CICs before and after low dose transient 5-AZA-CdR treatment measured by in-vivo limiting dilution assay in wild-type (H) or shMAVS (I) cells. The y-axis denotes the confidence intervals (lower, estimate and upper) for CIC frequency. * p<0.05; ** p<0.01; *** p<0.001. Multiple t tests were used to test for difference in population doubling time. Pairwise chi-squared test was used to test for difference in CIC frequency. See also Figure S5.

5-AZA-CdR targets the cancer-initiating compartment by activation of MDA5/MAVS/IRF7 pathway

Next, we tested whether 5-AZA-CdR can target cancer-initiating cells (CICs), which are defined first and foremost by their abilities to self-renew. To calculate the sphere-initiating frequency, we used the in vitro assay for sphere initiation followed by in vivo limiting dilution assays (LDA), the gold standard assay for measuring self-renewal (Kreso et al., 2014; Mack et al., 2014). For this assay, we plated 1, 10, 100 or 1000 cells per well in a 96 well round bottom plate for suspension cells, after 4 weeks we counted the number of positive wells (i.e. with formation of a spheroid) (Figure 5D). Then we calculated the frequency of sphere-initiating cells using the previously described ELDA algorithm (Hu and Smyth, 2009). We used a pairwise chi-squared test to calculate whether the frequency of sphere-initiating cells were different between any two groups as previously described (Hu and Smyth, 2009; Kreso et al., 2014; Mack et al., 2014). Transient low dose 5-AZA-CdR treatment induced a significant 10-fold reduction in the frequency of sphere-initiating cells (Figure 5E–F). However, inhibition of MDA5/MAVS/IRF7 activation by MAVS knockdown significantly reduced the ability of 5-AZA-CdR to target sphere-initiating cells (Figure 5E–F, P-value = 0.000609). Thereby suggesting that the effect of 5-AZA-CdR on self-renewal in the CIC fraction was dependent on MDA5/MAVS/IRF7 activation. Indeed, MDA5, but not RIG1, knockdown completely inhibited the ability of 5-AZA-CdR to target sphere-initiating cells (Figure S5H–J).

To validate our in vitro sphere initiating results we measured self-renewal using an in vivo limiting dilution assay. NSG mice were injected subcutaneously with 10, 100, 1000, 10000, and 50000 cells following a transient in vitro exposure to low dose of 5-AZA-CdR or vehicle control. The presence or absence of tumors was assessed after four months so that the frequency of colorectal CICs could be calculated (Figure 5G). We observed that a transient low dose of 5-AZA-CdR significantly reduced colorectal CIC frequency in vivo (Figure 5H, P-value=0.0000155). However, blockage of MDA5/MAVS/IRF7 activation by MAVS knockdown rendered colorectal CICs completely insensitive to 5-AZA-CdR (Figure 5I).

We further investigated the potential of 5-AZA-CdR–mediated activation of MDA5/MAVS/IRF7 to target colorectal CICs using primary colorectal cancer cells isolated at the time of surgical resection and stored in our biobank. Colorectal cancer samples can be divided into three major molecular subgroups based on the CpG Island Methylator Phenotype: CIMPhigh, CIMPlow and non-CIMP (Hinoue et al., 2012). By matching the global DNA methylation profile of our primary samples against publicly available DNA methylation data from primary colorectal cancer samples (Hinoue et al., 2012), we were able to classify our samples into the three DNA methylation subgroups (CIMPhigh, CIMPlow and non-CIMP, Figure 6A). Primary colorectal cancer cells were grown in a defined medium without serum, allowing for enrichment of CICs as previously described (Kreso et al., 2014; O’Brien et al., 2012). Using the same in vitro assay for sphere initiation coupled with LDA (Figure 5D), we observed that 5-AZA-CdR treatment significantly reduced the frequency of sphere-initiating cells in the CIMPhigh (Figure 6B), CIMPlow (Figure 6D), and non-CIMP (Figure 6F) cells. However, 5-AZA-CdR-mediated depletion of sphere-initiating cells was completely rescued when MAVS was knocked-down in the CIMPhigh (Figure 6C and Figure S6A/D) and CIMPlow (Figure 6E and Figure S6B/D) cells, and partially rescued in the non-CIMP (Figure 6G and Figure S6C–D) cells. In agreement with our previous data, the ability of 5-AZA-CdR treatment to reduce the frequency of sphere-initiating cells in primary CRC samples was associated with a significant increase in dsRNA formation (Figure S6E–J).

Figure 6. Transient 5-AZA-CdR treatment reduces the frequency of primary colorectal CICs independent of CIMP status, and this activity is dependent on MAVS activation.

Figure 6

Open in a new tab

A) Colorectal patient-derived cells were classified accordingly their CpG Island Methylator Phenotype (CIMP) into CIMPhigh (Sample 92), CIMPlow (Sample 181) and non-CIMP (Sample 73). Primary tissue DNA methylation data and CIMP classification were obtained from (Hinoue et al., 2012). Heatmap shows the pairwise correlation using Spearman correlation method of each sample. B, D and F) Frequency of colorectal CICs before and after low dose transient 5-AZA-CdR treatment measured by in-vitro limiting dilution assay in MAVS wild type cells. Patient-derived cells obtained from a CIMPhigh (sample 92) (B), a CIMPlow (sample 181) (D) and a non-CIMP (sample 73) (F) were used. C, E and G) Frequency of colorectal CICs before and after low dose transient 5-AZA-CdR treatment measured by in-vitro limiting dilution assay in MAVS knocked-down (shMAVS) cells. Patient-derived cells obtained from a CIMPhigh (sample 92) (C), a CIMPlow (sample 181) (E) and a non-CIMP (sample 73) (G) were used. The y-axis denotes the confidence intervals (lower, estimate and upper) for CIC frequency. Pairwise chi-squared test was used to test for difference in CIC frequency. See also Figure S6.

Altogether, our data suggest that the majority of cell-intrinsic antitumor effects mediated by transient low dose 5-AZA-CdR treatment are mediated by activation of MDA5/MAVS/IRF7 pathway.

RIG1 and MDA5 are druggable candidates to target colorectal CICs

Since low dose 5-AZA-CdR treatment can target colorectal CICs by activation of the MDA5/MAVS/IRF7 pathway, we next investigated whether agonists to cytosolic pattern-recognition receptors alone are sufficient to reduce CIC frequency. Both RIG1 and MDA5 can recognize dsRNAs. RIG1 is preferentially activated by low molecular weight (LMW) dsRNAs while MDA5 is preferentially activated by high molecular weight (HMW) dsRNAs (Kato et al., 2008). We transfected three primary colorectal cancer cells belonging to different CIMP status with LMW or HMW poly(I:C). Using the same in vitro assay for sphere initiation coupled with LDA (Figure 5D), we observed that transfection of either LMW (RIG1 agonist) or HMW (MDA5 agonist) dsRNAs can significantly reduce the frequency of sphere-initiating cells in the CIMPhigh (Figure 7A), CIMPlow (Figure 7B), and non-CIMP (Figure 7C) samples. However, agonists for other innate pattern-recognition receptors, such as cGAMP or Flagellin were unable to significantly reduce the frequency of sphere-initiating cells in the CIMPhigh (Figure S7A), CIMPlow (Figure S7B), and non-CIMP (Figure S7C) samples, corroborating our gene expression analysis where we found enrichment only for Regulation of innate immune response by RNA sensing molecules (Figure S4).

Figure 7. RIG1 or MDA5 agonists reduce the frequency of primary colorectal CICs independent of CIMP status.

Figure 7

Open in a new tab

Frequency of colorectal CICs before and after transfection with 0.5 ug/ml of low molecular weight (LMW – RIG1 agonist) or high molecular weight (HMW – MDA5 agonist) poly(I:C) measured by in-vitro limiting dilution assay. Patient-derived cells obtained from a CIMPhigh (sample 92) (A), a CIMPlow (sample 181) (B), and a non-CIMP (sample 73) (C) were used. The y-axis denotes the confidence intervals (lower, estimate and upper) for CIC frequency. Pairwise chi-squared test was used to test for difference in CIC frequency. See also Figure S7.

This data suggest that the activation of RIG1 or MDA5 alone is sufficient to target colorectal CICs, highlighting the potential of these targets as druggable candidates to target colorectal CIC.

DISCUSSION

Recent clinical and experimental data have shown that low doses of 5-AZA-CdR induce sustained antitumor effects against solid cancers (Juergens et al., 2011; Tsai et al., 2012; Yang et al., 2014). However, the clinical response to low doses DNA demethylating agents in both solid tumors and MDS/AML usually take more than a month to occur (Ahuja et al., 2014; Juergens et al., 2011; Treppendahl et al., 2014). This delayed response has led many people in the field to speculate that non-cytotoxic mechanisms may be playing a role in patient response. In addition, it has been suggested that this type of epigenetic therapy may be specifically targeting the cancer-initiating cell compartment (Ahuja et al., 2014; Tsai et al., 2012). However, it remains unclear whether transient low doses of 5-AZA-CdR could be targeting CICs and whether it is a direct consequence of its epigenetic effects.

Our data show that low doses of 5-AZA-CdR can significantly reduce the frequency of colorectal CICs using enriched sphere cultures derived from human colorectal cancer at the time of surgical resection. Colorectal cancer is a heterogeneous disease that can be classified into three main subgroups based on the DNA methylation profiles: CIMPhigh, CIMPlow and non-CIMP (Hinoue et al., 2012). Remarkably, we found that low doses of 5-AZA-CdR can strongly reduce the frequency of colorectal CICs independent of the CIMP profile, suggesting that this effect is elicited without demethylation of aberrantly methylated CpG islands.

Our data suggest that the durable antitumor effect of low dose 5-AZA-CdR is actually mediated by induction of dsRNAs and activation of MDA5 RNA recognition receptor followed by downstream activation of MAVS, IRF7, type III IFNs and up-regulation of interferon responsive genes –independent of their promoter demethylation. Moreover, clinical trials in non-small cell lung cancer, breast cancer and colorectal cancer patients with low dose DNA demethylating agents also identified up-regulation of interferon responsive genes (Li et al., 2014; Wrangle et al., 2013), highlighting the clinical relevance of our results.

MDA5 is a pattern-recognition receptor (PRR) ubiquitously expressed in the cytoplasm of most human cells. MDA5 recognizes nucleic acids associated with viral infections (dsRNAs), and has two amino-terminal caspase recruitment domains (CARDs). The CARDs of MDA5, upon activation, can recruit the signaling adaptor protein MAVS (also known as IPS1, CARDIF or VISA), which resides in the outer mitochondrial membrane. MAVS induce signaling cascades that result in nuclear translocation of IRF7 and activation of antiviral response programs, such as immunogenic cell death for viral clearance (Barbalat et al., 2011). Recent data suggest that the activation of RIG1 and MDA5 in tumor cells can also induce immunogenic cell death in melanoma, AML, and pancreatic cancer models (Besch et al., 2009; Duewell et al., 2014; Jiang et al., 2011). Moreover, this antitumor effect of RIG1/MDA5 signaling seems to be independent of type I interferon (Besch et al., 2009), similar to our results with 5-AZA-CdR.

Our data suggest that the 5-AZA-CdR-mediated activation of MDA5/MAVS/IRF7 pathway is a result of dsRNA induction of endogenous retrovirus (ERVs). This is in line with previous analyses suggesting that 5-AZA-CdR treatment could potentially induce dsRNA by RNA polymerase III-driven bi-directional transcription of repetitive elements (Leonova et al., 2013) and with previous results showing that 5-AZA-CdR treatment can induce the expression of interferon-responsive genes and endogenous retrovirus (Karpf et al., 2004; Karpf et al., 1999). Therefore, low-dose 5-AZA-CdR treatment will ‘trick’ cancer cell to behave as a virus-infected cell, inducing an MDA5/MAVS/IRF7-dependent ‘viral mimicry’ state.

Altogether, our results showing that low dose 5-AZA-CdR targets colorectal CICs may explain the delayed response-time observed in patients undergoing clinical trials. Moreover, our results showing that the activation of MDA5/MAVS/IRF7 pathway by dsRNAs is the main mechanism responsible for the antitumor effects of low dose 5-AZA-CdR may explain the current absence of DNA methylation markers for patient response. Monitoring dsRNA formation or MDA5/MAVS/IRF7 pathway activation may thereby be a better predictor of patient response than monitoring DNA methylation of promoters or gene bodies. Furthermore, our results suggest that by targeting the colorectal CICs, low-dose 5-AZA-CdR treatment may improve patient response to debulking therapies, such as standard of care chemotherapy, known to spare the CICs (Wicha, 2014). Remarkably, our results identify RIG1 and MDA5 cytosolic pattern recognition receptors as druggable targets against colorectal CICs.

Experimental Procedures

See the Extended Experimental Procedures for detailed experimental procedures.

Chemical treatment and doubling time measurement

For each experiment, cells were plated 24 hours prior treatment. At day 0, cells were treated or mock treated with 5-AZA-CdR (0.3uM, sigma Aldrich). Cells were washed 24 hours after treatment and replenished with fresh media (without drug). Where indicated, Ruxolitinib (JAK1/JAK2 inhibitor) or CP-690550 (JAK3 inhibitor) was added at a concentration of 1ug/ml. At the subsequent time points cells were harvested, re-suspended, counted; a fraction of cells were re-plated for the subsequent point. Doubling time (in days) was estimated as follow:

  • Doubling time = A*LOG (2)/(LOG(B)-LOG(C))

A is the number of days since the cells were plated, B the total number of cells when the cells were harvested, and C the number of cells plated.

Total RNA isolation, dsRNA enrichment and real-time PCR

Total RNA was extracted using TRIzol reagent. Primers are listed in supplemental table 1. For dsRNA enrichment, RNA were treated or not for 30 minute with 50ug/ml RNaseA in high salt concentration (NaCl 0.35M) to prevent dsRNA degradation (life technology). After treatment RNase A was removed by RNA purification with TRIzol reagent. The thermal cycling protocol was 1 cycle 95 C° for 10 minutes, then 40 cycles 95 C° for 15 seconds, 60 C° for 1 minute, followed by a melting curve analysis. For each transcript, the efficiency of the PCR reaction was determined by the slope of the standard curve generated from a serial dilution. Each transcript level was normalized by the acidic ribosomal phosphoprotein P0 (RPLP0) housekeeping gene. For dsRNA enrichment, expression was normalized by beta actin (ssRNA).

Semi-denaturing Detergent Agarose Gel Electrophoresis

Semi-denaturing detergent agarose gel electrophoresis (SDD-AGE) was performed as previously described (Halfmann and Lindquist, 2008). Briefly, Mitochondria were isolated (Qproteome mitochondria isolation kit, Quiagen) from LIM1215 cells, resuspended in mitochondria buffer and diluted before loading on a 1.5% agarose gel with 4x sample buffer (2X TAE 20% glycerol 8% SDS, bromophenol blue). Electrophoresis was done at 4°C at a constant voltage of 100V in running buffer (1XTBE and 0.1% SDS) for 1 hour. Proteins were then transferred into a nitrocellulose membrane and MAVS protein was detected using anti-MAVS antibody (1/1000, Abcam).

Limiting Dilution Assay (LDA)

For in vitro LDAs, colorectal cancer cells treated or not with 5-AZA-CdR for one day, or transfected with 0.5ug/ml of poly(I:C) low and high molecular weight in LyoVec, (InvivoGen) or treated with flagellin 1ug/ml or transfected 0.5 ug/ml of cGAMP in LyoVec (Invivogen) for 3 days were dissociated into single cells. Cells were then seeded in 96-well plates at the indicated cell doses (1 cell, 10 cells/, 100 cells, and 1000 cells/well). SytoxBlue was used to exclude dead cells. For each cell dose, at least 18 wells were seeded with cells, and for the lower cell doses, at least 72 wells were plated. 4 weeks later, wells containing spheres were scored, and the number of positive wells was used to calculate the frequency of sphere-forming units using the Extreme Limiting Dilution Analysis (ELDA) software (http://bioinf.wehi.edu.au/software/elda/index.html), provided by the Walter and Eliza Hall Institute (Hu and Smyth, 2009).

For in vivo LDAs, LIM1215 cells were treated with and without 5-AZA-CdR for 24 hours. Following, single cells suspension was obtained and diluted serially to the desired cell doses. Cells were injected subcutaneously into the flanks of NSG mice. The number of tumors formed out of the number of sites injected was scored to determine the frequency of colorectal CICs calculated using the ELDA software.

Statistical analysis

LDA statistical analyzes were performed using a pairwise Chi-squared test within the ELDA software as previously described (Hu and Smyth, 2009). Pathway enrichment analysis and up-stream regulators statistical analysis were performed using a Fisher’s exact test right-tailed within the IPA software as previously described (Kramer et al., 2014).

The statistical analysis to compare the frequency distribution was performed using a Kolmogorov-Smirnov test of Frequency distribution data using GraphPad prism software. All other statistical analyzes were performed using GraphPad prism software.

Supplementary Material

1
2
3
4
5
6
7
8
9

HIGHLIGHTS.

  • 5-AZA-CdR induces formation of dsRNAs and activation of the MDA5/MAVS/IRF7 pathway

  • Anti-proliferative response to DNA demethylation is mediated by a viral mimicry

  • 5-AZA-CdR-mediated targeting of CICs is mainly mediated by a viral mimicry

  • MDA5/MAVS/IRF7 pathway is a potentially druggable target against colorectal cancer

Acknowledgments

Funding for this work was provided by Cancer Research Society (CRS19092) and by the Princess Margaret Cancer Foundation to DDC and by NCI 5R01CA082422, and Stand Up to Cancer to P.A.J.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Accession Numbers

All genome-wide DNA methylation data used in this study was deposited in the Gene Expression Omnibus (GEO) database under the accession number GSE62086.

Author contributions

D.R., G.L., P.A.J., C.O.B., and DDC conceived and designed the experiments. D.R., H.L.Y., Y.W., S.Y.S., H.H., and D.D.C. performed the experiments. R.S., A.D., T.J.P., and D.D.C. analyzed the data. D.R., and D.D.C wrote the paper.

REFERENCES

  1. Ahuja N, Easwaran H, Baylin SB. Harnessing the potential of epigenetic therapy to target solid tumors. The Journal of clinical investigation. 2014;124:56–63. doi: 10.1172/JCI69736. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barbalat R, Ewald SE, Mouchess ML, Barton GM. Nucleic acid recognition by the innate immune system. Annual review of immunology. 2011;29:185–214. doi: 10.1146/annurev-immunol-031210-101340. [DOI] [PubMed] [Google Scholar]
  3. Besch R, Poeck H, Hohenauer T, Senft D, Hacker G, Berking C, Hornung V, Endres S, Ruzicka T, Rothenfusser S, et al. Proapoptotic signaling induced by RIG-I and MDA-5 results in type I interferon-independent apoptosis in human melanoma cells. The Journal of clinical investigation. 2009;119:2399–2411. doi: 10.1172/JCI37155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Dalerba P, Dylla SJ, Park IK, Liu R, Wang X, Cho RW, Hoey T, Gurney A, Huang EH, Simeone DM, et al. Phenotypic characterization of human colorectal cancer stem cells. Proceedings of the National Academy of Sciences of the United States of America. 2007;104:10158–10163. doi: 10.1073/pnas.0703478104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. De Carvalho DD, Sharma S, You JS, Su SF, Taberlay PC, Kelly TK, Yang X, Liang G, Jones PA. DNA methylation screening identifies driver epigenetic events of cancer cell survival. Cancer cell. 2012;21:655–667. doi: 10.1016/j.ccr.2012.03.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Duewell P, Steger A, Lohr H, Bourhis H, Hoelz H, Kirchleitner SV, Stieg MR, Grassmann S, Kobold S, Siveke JT, et al. RIG-I-like helicases induce immunogenic cell death of pancreatic cancer cells and sensitize tumors toward killing by CD8 T cells. Cell death and differentiation. 2014 doi: 10.1038/cdd.2014.96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Halfmann R, Lindquist S. Screening for amyloid aggregation by Semi-Denaturing Detergent-Agarose Gel Electrophoresis. Journal of visualized experiments : JoVE. 2008 doi: 10.3791/838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Hinoue T, Weisenberger DJ, Lange CP, Shen H, Byun HM, Van Den Berg D, Malik S, Pan F, Noushmehr H, van Dijk CM, et al. Genome-scale analysis of aberrant DNA methylation in colorectal cancer. Genome research. 2012;22:271–282. doi: 10.1101/gr.117523.110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Hou F, Sun L, Zheng H, Skaug B, Jiang QX, Chen ZJ. MAVS forms functional prion-like aggregates to activate and propagate antiviral innate immune response. Cell. 2011;146:448–461. doi: 10.1016/j.cell.2011.06.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Hu Y, Smyth GK. ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. Journal of immunological methods. 2009;347:70–78. doi: 10.1016/j.jim.2009.06.008. [DOI] [PubMed] [Google Scholar]
  11. Issa JP, Kantarjian HM. Targeting DNA methylation. Clinical cancer research : an official journal of the American Association for Cancer Research. 2009;15:3938–3946. doi: 10.1158/1078-0432.CCR-08-2783. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Jiang LJ, Zhang NN, Ding F, Li XY, Chen L, Zhang HX, Zhang W, Chen SJ, Wang ZG, Li JM, et al. RA-inducible gene-I induction augments STAT1 activation to inhibit leukemia cell proliferation. Proceedings of the National Academy of Sciences of the United States of America. 2011;108:1897–1902. doi: 10.1073/pnas.1019059108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Juergens RA, Wrangle J, Vendetti FP, Murphy SC, Zhao M, Coleman B, Sebree R, Rodgers K, Hooker CM, Franco N, et al. Combination epigenetic therapy has efficacy in patients with refractory advanced non-small cell lung cancer. Cancer discovery. 2011;1:598–607. doi: 10.1158/2159-8290.CD-11-0214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Karpf AR, Lasek AW, Ririe TO, Hanks AN, Grossman D, Jones DA. Limited gene activation in tumor and normal epithelial cells treated with the DNA methyltransferase inhibitor 5-aza-2’-deoxycytidine. Molecular pharmacology. 2004;65:18–27. doi: 10.1124/mol.65.1.18. [DOI] [PubMed] [Google Scholar]
  15. Karpf AR, Peterson PW, Rawlins JT, Dalley BK, Yang Q, Albertsen H, Jones DA. Inhibition of DNA methyltransferase stimulates the expression of signal transducer and activator of transcription 1, 2, and 3 genes in colon tumor cells. Proceedings of the National Academy of Sciences of the United States of America. 1999;96:14007–14012. doi: 10.1073/pnas.96.24.14007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kato H, Takeuchi O, Mikamo-Satoh E, Hirai R, Kawai T, Matsushita K, Hiiragi A, Dermody TS, Fujita T, Akira S. Length-dependent recognition of double-stranded ribonucleic acids by retinoic acid-inducible gene-I and melanoma differentiation-associated gene 5. The Journal of experimental medicine. 2008;205:1601–1610. doi: 10.1084/jem.20080091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kelly TK, De Carvalho DD, Jones PA. Epigenetic modifications as therapeutic targets. Nature biotechnology. 2010;28:1069–1078. doi: 10.1038/nbt.1678. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Kramer A, Green J, Pollard J, Jr, Tugendreich S. Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics. 2014;30:523–530. doi: 10.1093/bioinformatics/btt703. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kreso A, Dick JE. Evolution of the cancer stem cell model. Cell stem cell. 2014;14:275–291. doi: 10.1016/j.stem.2014.02.006. [DOI] [PubMed] [Google Scholar]
  20. Kreso A, van Galen P, Pedley NM, Lima-Fernandes E, Frelin C, Davis T, Cao L, Baiazitov R, Du W, Sydorenko N, et al. Self-renewal as a therapeutic target in human colorectal cancer. Nature medicine. 2014;20:29–36. doi: 10.1038/nm.3418. [DOI] [PubMed] [Google Scholar]
  21. Leonova KI, Brodsky L, Lipchick B, Pal M, Novototskaya L, Chenchik AA, Sen GC, Komarova EA, Gudkov AV. p53 cooperates with DNA methylation and a suicidal interferon response to maintain epigenetic silencing of repeats and noncoding RNAs. Proceedings of the National Academy of Sciences of the United States of America. 2013;110:E89–E98. doi: 10.1073/pnas.1216922110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Li H, Chiappinelli KB, Guzzetta AA, Easwaran H, Yen RW, Vatapalli R, Topper MJ, Luo J, Connolly RM, Azad NS, et al. Immune regulation by low doses of the DNA methyltransferase inhibitor 5-azacitidine in common human epithelial cancers. Oncotarget. 2014;5:587–598. doi: 10.18632/oncotarget.1782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Mack SC, Witt H, Piro RM, Gu L, Zuyderduyn S, Stutz AM, Wang X, Gallo M, Garzia L, Zayne K, et al. Epigenomic alterations define lethal CIMP-positive ependymomas of infancy. Nature. 2014;506:445–450. doi: 10.1038/nature13108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Merlos-Suarez A, Barriga FM, Jung P, Iglesias M, Cespedes MV, Rossell D, Sevillano M, Hernando-Momblona X, da Silva-Diz V, Munoz P, et al. The intestinal stem cell signature identifies colorectal cancer stem cells and predicts disease relapse. Cell stem cell. 2011;8:511–524. doi: 10.1016/j.stem.2011.02.020. [DOI] [PubMed] [Google Scholar]
  25. Navada SC, Steinmann J, Lubbert M, Silverman LR. Clinical development of demethylating agents in hematology. The Journal of clinical investigation. 2014;124:40–46. doi: 10.1172/JCI69739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. O’Brien CA, Kreso A, Ryan P, Hermans KG, Gibson L, Wang Y, Tsatsanis A, Gallinger S, Dick JE. ID1 and ID3 regulate the self-renewal capacity of human colon cancer-initiating cells through p21. Cancer cell. 2012;21:777–792. doi: 10.1016/j.ccr.2012.04.036. [DOI] [PubMed] [Google Scholar]
  27. O’Brien CA, Pollett A, Gallinger S, Dick JE. A human colon cancer cell capable of initiating tumour growth in immunodeficient mice. Nature. 2007;445:106–110. doi: 10.1038/nature05372. [DOI] [PubMed] [Google Scholar]
  28. Odendall C, Dixit E, Stavru F, Bierne H, Franz KM, Durbin AF, Boulant S, Gehrke L, Cossart P, Kagan JC. Diverse intracellular pathogens activate type III interferon expression from peroxisomes. Nature immunology. 2014;15:717–726. doi: 10.1038/ni.2915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ricci-Vitiani L, Lombardi DG, Pilozzi E, Biffoni M, Todaro M, Peschle C, De Maria R. Identification and expansion of human colon-cancer-initiating cells. Nature. 2007;445:111–115. doi: 10.1038/nature05384. [DOI] [PubMed] [Google Scholar]
  30. Seth RB, Sun L, Ea CK, Chen ZJ. Identification and characterization of MAVS, a mitochondrial antiviral signaling protein that activates NF-kappaB and IRF 3. Cell. 2005;122:669–682. doi: 10.1016/j.cell.2005.08.012. [DOI] [PubMed] [Google Scholar]
  31. Treppendahl MB, Kristensen LS, Gronbaek K. Predicting response to epigenetic therapy. The Journal of clinical investigation. 2014;124:47–55. doi: 10.1172/JCI69737. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Tsai HC, Li H, Van Neste L, Cai Y, Robert C, Rassool FV, Shin JJ, Harbom KM, Beaty R, Pappou E, et al. Transient low doses of DNA-demethylating agents exert durable antitumor effects on hematological and epithelial tumor cells. Cancer cell. 2012;21:430–446. doi: 10.1016/j.ccr.2011.12.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Weber F, Wagner V, Rasmussen SB, Hartmann R, Paludan SR. Double-stranded RNA is produced by positive-strand RNA viruses and DNA viruses but not in detectable amounts by negative-strand RNA viruses. Journal of virology. 2006;80:5059–5064. doi: 10.1128/JVI.80.10.5059-5064.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Wicha MS. Targeting self-renewal, an Achilles’ heel of cancer stem cells. Nature medicine. 2014;20:14–15. doi: 10.1038/nm.3434. [DOI] [PubMed] [Google Scholar]
  35. Wrangle J, Wang W, Koch A, Easwaran H, Mohammad HP, Vendetti F, Vancriekinge W, Demeyer T, Du Z, Parsana P, et al. Alterations of immune response of Non-Small Cell Lung Cancer with Azacytidine. Oncotarget. 2013;4:2067–2079. doi: 10.18632/oncotarget.1542. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Yang X, Han H, De Carvalho DD, Lay FD, Jones PA, Liang G. Gene Body Methylation Can Alter Gene Expression and Is a Therapeutic Target in Cancer. Cancer cell. 2014 doi: 10.1016/j.ccr.2014.07.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zeng M, Hu Z, Shi X, Li X, Zhan X, Li XD, Wang J, Choi JH, Wang KW, Purrington T, et al. MAVS, cGAS, and endogenous retroviruses in T-independent B cell responses. Science. 2014;346:1486–1492. doi: 10.1126/science.346.6216.1486. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS774731-supplement-1.pdf (105.2KB, pdf)
2
NIHMS774731-supplement-2.docx (23.4KB, docx)
3
NIHMS774731-supplement-3.pdf (237KB, pdf)
4
NIHMS774731-supplement-4.pdf (190.1KB, pdf)
5
NIHMS774731-supplement-5.pdf (966.9KB, pdf)
6
NIHMS774731-supplement-6.pdf (892.7KB, pdf)
7
NIHMS774731-supplement-7.pdf (58.7KB, pdf)
8
NIHMS774731-supplement-8.pdf (354.2KB, pdf)
9
NIHMS774731-supplement-9.pdf (31.9KB, pdf)

RESOURCES