Genome-Wide Identification of Aberrant Promoter Associated CpG Island Methylation in Acute Lymphoblastic Leukemia.

Aberrant DNA methylation is a common molecular feature of both pediatric and adult ALL. Specific methylation patterns predict for poor prognosis (Shen et al Blood 2004), and reactivation of epigenetically inactivated molecular pathways results in induction of leukemia cell death (Kuang et al. Oncogene 2007). Until now most studies of methylation in ALL have been based on arbitrary gene selection methods. To overcome this limitation and to study hundreds of promoter CpG islands simultaneously, we have developed a method that combines MCA (Methylated CpG Island Amplification) with either RDA (Representational Difference Analysis) or the Agilent Promoter Microarray platform. With these methods differentially methylated DNA treated with bisulfite is generated after mixing tester DNA (in our case DNA from de novo refractory Ph negative and MLL negative ALL patients) with driver DNA (normal B cell controls) and using specific restriction enzymes and several rounds of PCR. DNA fragments thus generated are either cloned (RDA) or labeled and spotted on the Agilent Array. Using this technology, that can potentially interrogate up to 17K promoters, we have identified 932 promoters targets of aberrant DNA methylation in poor risk ALL from patients that cannot be currently identified by standard molecular methods (Ph and MLL negative). The genes associated with these promoters are distributed through the human genome but an overrepresentation of methylated promoters located in chromosomes 3, 9, 11 and 19 was detected. Using molecular pathway clustering analysis, 404 of these genes are grouped together in 29 specific functional pathways. We have validated the methylation of 31 of these 923 genes by bisulfite pyrosequencing. Of these, 27 (87%) were confirmed to be hypermethylated in 23 human leukemia cell lines but not in normal controls (N=15). Methylation status analysis of these 27 genes allowed for the segregation of T cell versus B cell leukemia cell lines. Fifteen of these genes (GIPC2, RSPO1, MAGI1, CAST1, ADCY5, HSPA4L, OCLN, EFNA5, MSX2, GFPT2, GNA14, SALL1, MYO5B, ZNF382 and MN1) were also frequently hypermethylated in primary ALL samples. Expression analysis of 6 of these genes (GIPC2, MAGI1, ADCY5, HSPA4L, OCLN and GNA14) in leukemia cell lines further confirmed methylation associated gene silencing. Treatment of methylated/silenced cell lines with 5′-aza-2′-deoxycytidine and trichostatin A resulted in gene re-expression, further confirming the role of DNA methylation in their silencing. In summary, we have identified in excess of 900 targets of aberrant DNA methylation in ALL. The study of the epigenetically suppressed pathways represented by these genes should allow us to further understand the molecular pathogenesis of ALL and develop new prognostic biomarkers for patients with Ph and MLL negative disease.