Boolean Implications Identify Wilms Tumor 1 Mutation as a Driver of DNA Hypermethylation in Acute Myeloid Leukemia

Transcription

1 Boolean Implications Identify Wilms Tumor 1 Mutation as a Driver of DNA Hypermethylation in Acute Myeloid Leukemia Subarna Sinha PhD Department of Computer Science Principal Investigator: David Dill Daniel Thomas MD PhD Department of Medicine, Hematology Division Stem Cell & Regenerative Medicine Institute Principal Investigator: Ravi Majeti Poster # 13

2 Aberrant Methylation in Acute Myeloid Leukemia Acute Myeloid Leukemia (AML) is a disease characterised by the accumulation myeloid precursor cells in the bone marrow that are blocked in their ability to differentiate into mature blood cells AML is associated with widespread deregulation of DNA methylation. Cell of origin? Stochastic? Genetic Mutation? Aberrant Methylation 1. Identify genetic drivers of aberrant methylation. 2. Find leads for a mutation-specific therapy. Methylation Red=hyper Green=hypo Figueroa et al 2010 DNMT3a TET2 IDH1 IDH2

3 Boolean Implications (IF THEN Rules) Attribute B Attribute A Four different implications: HIHI: IF A high, THEN B high HILO: IF A high, THEN B low LOHI: IF A low, THEN B high LOLO: if A low, THEN B low

4 Computational Pipeline TCGA AML samples Mutation Data Methylation Data 17 Recurrent mutations CpG site filtering= 285, 320 probes Discretize me values CpG site B Methylation DNMT3A mutation Generate Boolean Implications Count number of methylation HIHI and HILO Boolean implications for each mutation CpG site A Methylation IDH2 mutation

5 WT1 mutation AML is linked to hypermethylation Hyper- Hypo- Mixed Very few (<500) IDH WT CEBPA DNMT3A Cohesin No. of hypermethylation implications No. of hypomethylation implications RUNX IDH TET FLT NPM TP KIT KRAS 9 23 MT-CO NRAS PTPN U2AF Hypermethylation / Hypomethylation Hypermethylation Index

6 Distinct CpG sites and associated genes linked to hypermethylating mutations CpG sites Methylated Genes WT CEBPA IDH WT IDH2 * CEBPA *269

7 WT1 mutation induces hypermethylation in AML cells ins/dels WT1 NH 2 WT1mut COSMIC NH 2 P/Q rich P/Q rich ZF ZF ZF ZF COOH T2A GFP COOH Control WT1mutant transduced THP-1 stable cell-lines 10 passages 450K Beadchip array CpG me (β value) 0.0 to to to to to 1.0 Overlap with patients: 8.5E-34 Fisher s exact test

8 Mutant WT1 methylation signature is enriched for PRC2 target genes Patient samples with WT1 mut Gene Sets P-value Benporath ES with Benporath EED targets Benporath Suz12 targets Benporath PRC2 targets Mikkelsen MEF HCP with Mikkelsen Brain HCP with H3K4ME3 and Mikkelsen MCV6 HCP with Meissner Brain HCP with Mikkelsen NPC HCP with Meissner NPC HCP with H3K4ME3 and 1.6E E E E E E E E E E-25 THP1 cell-line with WT1mut Gene Sets P-value Benporath ES with Mikkelsen MEF HCP with Benporath EED targets Benporath SUZ12 targets Benporath PRC2 targets Mikkelsen Brain HCP with H3K4ME3 and Mikkelsen NPC HCP with Mikkelsen MCV6 HCP with Meissner NPC HCP with H3K4ME3 and Mikkelsen MEF HCP with 9.06E E E E E E E E E E Enrichment score (ES) H3K27me3 genes (Benporath) NES: log (p-value) PRC2 target genes (Benporath)

9 WT1 mutant AML shows aberrant repression of Polycomb repressor complex 2 targets transcriptional repression SUZ12 EZH2 EED K562 cells AML Normal H3K27me3 CHIP (ENCODE) PRC2 marked genes in adult hematopoiesis Repressed in WT1mut AML expressed repressed H3K27me3-marked genes normally up in mature myeloid cells Does WT1 mutation block myeloid differentiation?

10 Inhibition of PRC2 promotes differentiation in AML with WT1 mutation WT1mut+ NK AML, SU359 TF1+ GM 0% TF1+ EPO Isotype control WT1mut + GM 70% 0% 0.2% SSC CD117 in vitro differentiation WT1mut + EPO 9% CD45 CD34 Intracellular Fetal Hb (anti-human HbF-PE) % CD11b ** % CD ** % CD ** * p<0.01

11 Conclusions Mutation in WT1 is strongly linked to DNA hypermethylation in AML Introduction of mutant WT1 into wildtype cells induced the same pattern of DNA hypermethylation The pattern of methylation and gene expression is consistent with a differentiation block caused by WT1mut through dysregulated silencing of PRC2 targets Differentiation block in WT1mut AML can be overcome by EZH2 inhibition EZH2 inhibitors have activity in WT1mut AML Boolean implications are a useful data mining tool for large, heterogeneous cancer data sets WT1 mutation DNA hypermethylation and dysregulation of PRC2 targets differentiation block EZH2 inhibitor release of differentiation block RNA pol II SUZ12 EZH2 EED

12 Acknowledgements Dan Thomas David Dill Ravindra Majeti Sylvia Plevritis Stanford Centre for Cancer Systems Biology (CCB, NCI) Progenitor Cell Biology Consortium (PCBC, NHLBI) NHMRC CJ Martin Fellowship Hem/Oncology HSANZ Targetted Therapy Fellowship Andrew Gentles Andrew Feinberg Namyoung Jung