Abstract
Background: Hypermethylation of CpG-rich tumor suppressor gene promoters constitutes a prototypical cancer-specific epigenetic abnormality. Recently, methylome-wide studies have provided evidence of extensive aberrant DNA methylation also in gene bodies and intergenic regions; their functional role is still under debate. DNA hypomethylating agents (HMAs) such as 5-aza-CdR (DAC) and 5-aza-CR are thought to act particularly by reactivating hypermethylated promoters of epigenetically silenced genes, and this activity is enhanced by combining HMAs with histone deacetylase inhibitors (HDACi). However, transcriptome-wide studies have shown that HMAs can also downregulate mRNAs. Recently, Yang et al. (Cancer Cell 2014) showed that gene body methylation can be strikingly reversed by DAC in human colon cancer cells, with remarkable downregulation of genes involved in metabolic processes regulated by c-MYC, implicating gene body methylation as a drug target. Therefore, we asked whether HMA-induced reversal not only of promoter but also of gene body methylation occurs in AML cells, and whether gene body demethylation is preferentially associated with gene repression rather than activation. We further asked whether the synergism between HMAs and HDACi in gene reactivation might also be observed for gene repression.
Materials and Methods: U937 cells were treated with 3x24h pulses of DAC (50nM), alone or in combination with the HDACi panobinostat (Pano, 8nM) or Valproic acid (VPA, 1mM) for 24h with low-cytotoxic (>80% viability), <25% growth-inhibitory concentrations. Transcriptome and methylome profiles (biological triplicates) were generated using Affymetrix Human Gene 2.0 ST and Infinium HumanMethylation450 BeadChip arrays. Data were analyzed using R package RnBeads applying standard methods.
Results: Transcriptome analyses revealed that DAC, Pano or VPA alone induced 207, 267 and 179 transcripts, with downregulation of 28, 45 and 22 transcripts, respectively (FDR<0.01; ≥2-fold change). When DAC was combined with either Pano (D+P) or VPA (D+V), a striking synergistic effect was observed for upregulated transcripts (1069 and 680). Remarkably, a synergistic effect on downregulated transcripts was even stronger (667 and 233 transcripts). As expected, DAC treatment resulted in extensive DNA demethylation: 295.6K CpGs (89% of 331.8K CpGs hypermethylated >30% in untreated U937) became >10% demethylated (FDR<0.01), with demethylation of additional 8.7K CpGs (2.6%) and 5.7K CpGs (1.7%) with D+P and D+V, respectively.
Most of the differentially expressed transcripts mapped to demethylated CpGs located in both promoters and gene bodies. We focussed on up- and downregulated transcripts exclusively associated with either promoter- or gene body-demethylated CpGs. The ratio of transcripts mapping to promoter- vs. gene body-demethylated CpGs doubled from up- to downregulated transcripts (all 3 treatments). Thus, for upregulated transcripts these ratios were 1: 1.9 (DAC), 1: 3.5 (D+P) and 1: 3.4 (D+V); for downregulated transcripts 1: 4.0 (DAC), 1: 7.0 (D+P) and 1: 6.9 (D+V). Also, the combination treatments resulted in a clear shift to more transcripts mapping to gene body-demethylated CpGs than DAC treatment alone. Demethylated CpGs located in shore regions of gene bodies were particularly enriched in downregulated transcripts upon combination treatment, suggesting a possible mechanism of their regulation.
Of the 667 transcripts downregulated by D+P, the 227 gene-body demethylation-associated transcripts included various putative or bona fide oncogenes (such as c-MYC) and epigenetic modifiers. Six potential candidates were chosen for validation by RT-qPCR. Gene ontology analysis revealed an enrichment for genes involved in RNA processing and transcription.
Conclusions: The combination of a HMA with two different HDACi resulted not only in synergistic gene reactivation but also in massive, synergistic downregulation of genes including oncogenes, which was associated predominantly with gene body demethylation. Taken together, these findings have implications not only regarding the mechanisms of action of combined epigenetic treatment, but also for a better understanding of clinical responses in trials where this approach is tested, such as the randomized DECIDER AML trial (NCT00867672), by analyzing in vivo epigenetic drug effects on primary blasts.
Lübbert:Janssen: Honoraria, Other: travel support; Celgene: Honoraria, Other: travel support; TEVA: Other: travel support.
Author notes
Asterisk with author names denotes non-ASH members.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal