WHOLE GENOME AND TARGETED BISULFITE SEQUENCING REVEALS 3 DNA METHYLATION CLUSTERS AND NEW BIOLOGICALLY RELEVANT HYPERMETHYLATED GENES IN AML PATIENTS

Background:

In acute myeloid leukemia (AML), aberrant DNA methylation has been linked to the pathogenesis and progression of the disease. Changes in DNA methylation of promoters, or other regions, are studied primarily with respect to which pathways are involved in tumor transformation and their impact on prognosis.

Aims:

The aim was to profile DNA methylation changes in AML patients using methods of next-generation sequencing and to correlate these changes with gene expression for discovering biologically relevant genes affected by hypermethylation.

Methods:

Whole genome (3 AML, 2 controls) and targeted (14 AML, 1 control) bisulfite libraries were run on a HiSeqTM2000 sequencer (Illumina) using 105 bp paired-end sequencing reads. Validation of acquired methylation data was performed using 454 pyrosequencing and HumanMethylation27 Infinium arrays (Illumina). HumanHT-12 v4 Expression BeadChip (Illumina) was utilized for whole genome expression profiling. Expression data of selected genes were validated and extended to a larger number of examined patients by TaqMan real-time PCR.

Results:

Unsupervised hierarchical clustering of CpG methylation outside and/or inside CpG islands revealed three DNA methylation clusters. From the clinical and molecular characteristics, only CBF/MYH11+ patients clustered together. None of the other molecular abnormalities (i.e. DNMT3A mutations, MLL translocation, NPM1 mutations FTL3/ITD and CEPBα mutation) formed clusters and we did not observe an effect of clinical status of AML (de novo, secondary, AML with dysplastic changes or relapsed AML). A correlation between methylation and expression data enabled us to distinguish between aberrant DNA methylation with no effect on expression of downstream located genes (probably tissue-specific methylation) and biologically relevant DNA methylation (accompanied with changes of gene expression). ELAVL2 and CACNA1E had enormous differences in DNA methylation between AML samples and controls; however they are very probably genes displaying tissue-specific methylation changes, because we did not detect their expression even in healthy precursor cells. On the other hand, there were genes - most notably CHFR and PBX3, where DNA methylation changes were accompanied with a change in their expression.

We measured expression of CHRF and PBX3 genes in 123 AML samples at diagnosis. 20% of AML had down- and 22% up-regulated PBX3 expression, 5% of AML down- and 7% up-regulated CHFR expression. 454 pyrosequencing confirmed the role of DNA methylation in down- and up-regulation of PBX3 gene, hypomethylation (median methylation level 0.25, range 0.15 – 0.36) of a regulatory region located downstream of an annotated CpG island were connected with elevated levels of PBX3, whereas hypermethylation (median methylation level 0.51, range 0.31 – 0.98) with decreased levels of expression. Control samples displayed intermediate levels of methylation (median 0.35, range 0.19 – 0.51). Furthermore we observed correlation between PBX3 and HOXA9 expression, which is consistent with recently published data that PBX3 is an important cofactor of HOXA9 in leukemogenesis.

Summary / Conclusion:

In summary, we found new and biologically important genes that are influenced by differential DNA methylation and were able to distinguish 3 major DNA methylation clusters in AML patients.

This study was a part of the COST Action BM0801 (EuGESMA) and was supported by the Czech Ministry of Education, Youth and Sports OC10042 grant and institutional funding grant IHBT00023736.

Email address: hana.hajkova@uhkt.cz

Abstrakta v časopise Haematologica 2013, Suppl1

Online Program