Andrew P. Feinberg

Breast cancer, the most diagnosed cancer among women, is stratified by gene expression patterns into luminal and basal-like subtypes, and the latter is associated with worse prognosis^1-3. Recent advances in single-cell technologies have underscored the presence of intratumour heterogeneity, which poses substantial challenges in cancer treatment and achieving long-term disease-free survival^4-7. Here we extensively profile epigenetic heterogeneity in a cohort of treatment-naive human luminal tumours, which were obtained directly after surgical resection. By applying a single-cell epigenetic approach based on cytometry by time-of-flight, we identify a recurrent subpopulation of basal-like cells in luminal tumours with a conserved epigenetic signature across patients. These basal-like cells, also identified by deconvolution of whole-genome DNA methylation patterns, are characterized by high expression of aggressive oncogenic traits and downregulation of hormone receptor signalling. Direct comparison of single-cell proteomic and genomic profiling of monomethylation and trimethylation of histone 3 on lysine 4 revealed that these marks are key features associated with loss of luminal identity. Finally, we highlight the potential of concomitant single-molecule profiling of histone modifications on cell-free nucleosomes in blood as an epigenetic liquid biopsy approach for early diagnosis of breast cancer. Moreover, the epigenetic basal signature measured in blood is linked to prognosis, which enables the identification of patients at increased risk for disease relapse.

Disruption of the epigenetic landscape is of particular interest in acute myeloid leukemia (AML) due to its relatively low mutational burden and frequent occurrence of mutations in epigenetic regulators. In this study, we applied an information-theoretic analysis of methylation potential energy landscapes, capturing changes in mean methylation level and methylation entropy, to comprehensively analyze DNA methylation stochasticity in subtypes of AML defined by mutually exclusive genetic mutations. AML subtypes with CEBPA double mutation and those with IDH mutations were identified as distinctly high-entropy subtypes, marked by methylation disruption over a convergent set of genes. The analysis revealed a core program of epigenetic landscape disruption across all AML subtypes; discordant methylation stochasticity, transcriptional dysregulation, and altered chromatin accessibility converged on functionally important leukemic signatures. Demonstration of a relationship between methylation entropy and gene expression variability connected the disruption of the epigenetic landscape to transcription in AML. Finally, the hypomethylating drug decitabine led to reduction of DNA methylation entropy specifically in IDH2-mutant AML cells. Overall, this approach identified a convergent program of epigenetic dysregulation in leukemia, clarifying the contribution of specific genetic mutations to stochastic disruption of the epigenetic and transcriptional landscapes of AML.Significance:DNA methylation stochasticity, variable gene expression, and altered chromatin accessibility converge on key leukemic regulators across AML subtypes, suggesting a genotype-independent role of stochastic disruption of the epigenetic landscape in mediating leukemogenesis.

Figure S7: Transcriptional dysregulation recapitulates leukemia signatures.

HOMER transcription factor binding motif enrichment results for CPEL DMRs.

Gene set enrichment analysis of differential expression over the MSigDB C2 gene set collection.

Gene set enrichment analysis of differential expression over MSigDB Hallmark gene set collection.

Convergent differentially expressed genes, and differentially expressed genetic drivers of AML.

CPEL DMRs with associated dMML, dNME, and UC.

Figure S10: Methylation stochasticity relates to gene expression variability within individual AML samples.

Gene expression mean-adjusted variability, variance, and mean in each AML subtype.