Modeling complex patterns of differential DNA methylation that associate with gene expression changes.

Numerous genomic studies are underway to determine which genes are abnormally regulated by DNA methylation in disease. However, we have a poor understanding of how disease-specific methylation changes affect expression. We thus developed an integrative analysis tool, Methylation-based Gene Expression Classification (ME-Class), to explain specific variation in methylation that associates with expression change. This model captures the complexity of methylation changes around a gene promoter. Using 17 whole-genome bisulfite sequencing and RNA-seq datasets from different tissues from the Roadmap Epigenomics Project, ME-Class significantly outperforms standard methods using methylation to predict differential gene expression change. To demonstrate its utility, we used ME-Class to analyze 32 datasets from different hematopoietic cell types from the Blueprint Epigenome project. Expression-associated methylation changes were predominantly found when comparing cells from distantly related lineages, implying that changes in the cell's transcriptional program precede associated methylation changes. Training ME-Class on normal-tumor pairs from The Cancer Genome Atlas indicated that cancer-specific expression-associated methylation changes differ from tissue-specific changes. We further show that ME-Class can detect functionally relevant cancer-specific, expression-associated methylation changes that are reversed upon the removal of methylation. ME-Class is thus a powerful tool to identify genes that are dysregulated by DNA methylation in disease.

DNMT inhibitors reverse a specific signature of aberrant promoter DNA methylation and associated gene silencing in AML.

BACKGROUND: Myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) are neoplastic disorders of hematopoietic stem cells. DNA methyltransferase inhibitors, 5-azacytidine and 5-aza-2'-deoxycytidine (decitabine), benefit some MDS/AML patients. However, the role of DNA methyltransferase inhibitor-induced DNA hypomethylation in regulation of gene expression in AML is unclear. RESULTS: We compared the effects of 5-azacytidine on DNA methylation and gene expression using whole-genome single-nucleotide bisulfite-sequencing and RNA-sequencing in OCI-AML3 cells. For data analysis, we used an approach recently developed for discovery of differential patterns of DNA methylation associated with changes in gene expression, that is tailored to single-nucleotide bisulfite-sequencing data (Washington University Interpolated Methylation Signatures). Using this approach, we find that a subset of genes upregulated by 5-azacytidine are characterized by 5-azacytidine-induced signature methylation loss flanking the transcription start site. Many of these genes show increased methylation and decreased expression in OCI-AML3 cells compared to normal hematopoietic stem and progenitor cells. Moreover, these genes are preferentially upregulated by decitabine in human primary AML blasts, and control cell proliferation, death, and development. CONCLUSIONS: Our approach identifies a set of genes whose methylation and silencing in AML is reversed by DNA methyltransferase inhibitors. These genes are good candidates for direct regulation by DNA methyltransferase inhibitors, and their reactivation by DNA methyltransferase inhibitors may contribute to therapeutic activity.

Senescent cells harbour features of the cancer epigenome.

Altered DNA methylation and associated destabilization of genome integrity and function is a hallmark of cancer. Replicative senescence is a tumour suppressor process that imposes a limit on the proliferative potential of normal cells that all cancer cells must bypass. Here we show by whole-genome single-nucleotide bisulfite sequencing that replicative senescent human cells exhibit widespread DNA hypomethylation and focal hypermethylation. Hypomethylation occurs preferentially at gene-poor, late-replicating, lamin-associated domains and is linked to mislocalization of the maintenance DNA methyltransferase (DNMT1) in cells approaching senescence. Low-level gains of methylation are enriched in CpG islands, including at genes whose methylation and silencing is thought to promote cancer. Gains and losses of methylation in replicative senescence are thus qualitatively similar to those in cancer, and this 'reprogrammed' methylation landscape is largely retained when cells bypass senescence. Consequently, the DNA methylome of senescent cells might promote malignancy, if these cells escape the proliferative barrier.

Discovering high-resolution patterns of differential DNA methylation that correlate with gene expression changes.

Methylation of the CpG-rich region (CpG island) overlapping a gene's promoter is a generally accepted mechanism for silencing expression. While recent technological advances have enabled measurement of DNA methylation and expression changes genome-wide, only modest correlations between differential methylation at gene promoters and expression have been found. We hypothesize that stronger associations are not observed because existing analysis methods oversimplify their representation of the data and do not capture the diversity of existing methylation patterns. Recently, other patterns such as CpG island shore methylation and long partially hypomethylated domains have also been linked with gene silencing. Here, we detail a new approach for discovering differential methylation patterns associated with expression change using genome-wide high-resolution methylation data: we represent differential methylation as an interpolated curve, or signature, and then identify groups of genes with similarly shaped signatures and corresponding expression changes. Our technique uncovers a diverse set of patterns that are conserved across embryonic stem cell and cancer data sets. Overall, we find strong associations between these methylation patterns and expression. We further show that an extension of our method also outperforms other approaches by generating a longer list of genes with higher quality associations between differential methylation and expression.

DNMT gene expression and methylome in Marek's disease resistant and susceptible chickens prior to and following infection by MDV.

Marek's disease (MD) is characterized as a T cell lymphoma induced by a cell-associated α-herpesvirus, Marek's disease virus type 1 (MDV1). As with many viral infectious diseases, DNA methylation variations were observed in the progression of MD; these variations are thought to play an important role in host-virus interactions. We observed that DNA methyltransferase 3a (DNMT3a) and 3b (DNMT3b) were differentially expressed in chicken MD-resistant line 6 3 and MD-susceptible line 7 2 at 21 d after MDV infection. To better understand the role of methylation variation induced by MDV infection in both chicken lines, we mapped the genome-wide DNA methylation profiles in each line using Methyl-MAPS (methylation mapping analysis by paired-end sequencing). Collectively, the data sets collected in this study provide a more comprehensive picture of the chicken methylome. Overall, methylation levels were reduced in chickens from the resistant line 6 3 after MDV infection. We identified 11,512 infection-induced differential methylation regions (iDMRs). The number of iDMRs was larger in line 7 2 than in line 6 3, and most of iDMRs found in line 6 3 were overlapped with the iDMRs found in line 7 2. We further showed that in vitro methylation levels were associated with MDV replication, and found that MDV propagation in the infected cells was restricted by pharmacological inhibition of DNA methylation. Our results suggest that DNA methylation in the host may be associated with disease resistance or susceptibility. The methylation variations induced by viral infection may consequentially change the host transcriptome and result in diverse disease outcomes.

A finite-sample, distribution-free, probabilistic lower bound on mutual information.

For any memoryless communication channel with a binary-valued input and a one-dimensional real-valued output, we introduce a probabilistic lower bound on the mutual information given empirical observations on the channel. The bound is built on the Dvoretzky-Kiefer-Wolfowitz inequality and is distribution free. A quadratic time algorithm is described for computing the bound and its corresponding class-conditional distribution functions. We compare our approach to existing techniques and show the superiority of our bound to a method inspired by Fano's inequality where the continuous random variable is discretized.