CBFβ/MYH11 Hypomethylation Signature In Patients With Acute Myeloid Leukaemia

It has been shown that changes in DNA methylation pattern in patients with acute myeloid leukaemia (AML) may both reflect specific molecular abnormalities or characterize a group of patients without a known molecular aberration. Studying genes that are epigenetically deregulated in different groups of patients may contribute to a more detailed understanding of pathways involved in leukemic transformation.

For 15 samples (14 AMLs of diverse clinical and genetic background and pool of CD34+ control samples) SureSelect^XT Human Methyl-Seq system (Agilent) was used to interrogate DNA methylation of selected regions (84 megabases in total) of their genomes. Targeted bisulfite libraries were run on a HiSeq^TM2000 sequencer (Illumina) using 105 bp paired-end sequencing reads. Methylation status of candidate regions was verified using 454 bisulfite pyrosequencing (Roche). The expression levels of selected genes were assessed with TaqMan Gene Expression Assays (Applied Biosystems).

Targeted methylation data was correlated with previously obtained whole-genome bisulfite (WGBS) data for the cross-experiment (WGBS/target enrichment) samples (n=3), taking into account positions with minimal coverage of 10. There was a strong positive correlation (R≥0.97) for all three samples. Pairwise Pearson correlations were computed for all (targeted and WGBS) samples and formed the basis of a hierarchical clustering analysis. The cross-experiment replicates clustered together. From the clinical and molecular characteristics, only CBFβ/MYH11+ patients (n=2) clustered together in all clustering analysis performed (using CpG in/out CGIs or all CpG regardless of location). None of the other molecular abnormalities (i.e. DNMT3A mutations, MLL translocations, NPM1 mutations, FTL3/ITDs and CEPBα mutations) formed clusters. We did not observe any effect of clinical status of AML (de novo, secondary AML with dysplastic changes or relapsed AML). Based on this observation, we extracted genomic regions being uniquely differentially methylated for CBFβ/MYH11+ patients. There was a clear tendency towards hypomethylation in 125 regions out of 182 (69%) displaying lower DNA methylation levels compared to CD34+ pool of controls. All CBFβ/MYH11 differentially methylated regions (DMRs) were uploaded to GREAT (http://bejerano.stanford.edu/great/public/html) and enrichment for genes described as upregulated in inv(16) AML (CBFβ/MYH11+) was seen (reported in Valk et al., NEJM 2004, ID: VALK_AML_CLUSTER_9). This enrichment set comprised 10 genomic regions assigned for 6 genes – MN1, SPARC, ST18, DHRS3, FAM171A and BAHCC1.MN1, SPARC, ST18 and DHRS3 – were chosen for hypomethylation verification by 454 bisulfite pyrosequencing. Altogether 11 CBFβ/MYH11+ (inv(16)), 11 non-inv(16) AML M4, 6 other AML subtypes and 5 healthy control samples were examined by 454 pyrosequencing. Average levels of DNA methylation for selected regions were significantly lower for inv(16) versus non-inv(16) AML M4, other AML subtypes and healthy controls (P<0.0001). The 454 results support our theory about site specific CBFβ/MYH11 hypomethylation signature. MN1 gene was overexpressed in 82% of inv(16) AML versus 42% of non-inv(16) patients (χ² test, P=0.009). Interestingly, MN1 overexpression in non-inv(16) patients was not accompanied with MN1 hypomethylation suggesting different mechanisms of gene deregulation in various AML subtypes.

We found a new hypomethylation signature in CBFβ/MYH11+ patients. The described hypomethylated regions were assigned to genes known as being overexpressed in inv(16) patients. As for MN1 gene, we can speculate that different mechanism responsible for its overexpression in inv(16) versus non-inv(16) patients may explain discrepancy between its negative prognostic value in cytogenetically normal AML patients and generally good outcome of inv(16) patients.

This study was supported by the project (Ministry of Health, Czech Republic) for conceptual development of research organization (00023736, IHBT) and was a part of the COST Action BM0801 (EuGESMA).