Methylated Retroviral Integration Mutagenesis (MRIM) in Mice and Comparative Analysis in Human AML Identify Reduced PTP4A3 Expression As a Prognostic Indicator

Abstract 746

Acute myeloid leukemia (AML) is a complex disease driven by multiple cytogenetic abnormalities, aberrant expression levels and mutations of different genes. Frequent occurrence of chromosomal deletions suggests that haplo-insufficiencies contribute to the pathogenesis of AML. However, because deleted regions often harbor numerous genes, it remains difficult to pin point critical haplo-insufficient genes (HIGs) involved in the pathogenesis of AML. Retroviral insertion mutagenesis in mouse models has been used to discover novel genes involved in the development of different types of cancer. Only a small minority of identified genes in these screens has been classified as tumor suppressor genes or HIGs, based on disruption of coding sequences by proviral integrations. Gene therapy studies using murine leukemia virus based vectors have shown that epigenetic changes of long terminal repeats (LTRs) of integrated proviruses often result in silencing of therapeutic genes, and that preventing methylation of the CpG islands within LTRs overcomes this problem. Based on these observations, we hypothesized that methylation of viral sequences not only results in silencing of retroviral genes themselves but may also affect host genes located proximal to proviral integrations. Methylated LTRs located in proximity of promoter regions may thus identify genes that are deregulated leading to haplo-insufficiency.

To identify novel HIGs, genomic DNA of 6 murine leukemias induced by the Graffi 1.4 murine leukemia virus, was isolated and digested. Subsequently, methylated fragments were immunoprecipitated using methylated DNA immunoprecipitation (MeDIP), followed by inverse PCR to amplify fragments flanking methylated viral integrations (mVIS). Amplified fragments were labeled, fragmented and hybridized on murine promoter DNA arrays. DNA methylated viral integration sites were detected using a novel algorithm, hypergeometric analysis of tiling arrays (HAT) [1]. All integrations were confirmed by directed PCR and Sanger sequencing. Next, their methylation status was confirmed with methylation sensitive restriction analysis. We refer to this strategy as Methylated Virus Integration Mutagenesis (MRIM). In the murine leukemias, expression levels of genes identified by MRIM, relative to Tbp (TATA box binding protein) expression, were determined by quantitative PCR (qPCR). Survival analysis was performed in a cohort of 454 de novo AML patients under the age of 60 using a Cox-regression model. For this analysis, MAS5 normalised, log2 transformed PTP4A3 gene expression levels were used, determined by gene expression profiling.

Using MRIM, 6 genes that were flanked by a methylated LTR were identified. Of these genes, Ptp4a3 was transcriptionally down regulated relative to normal bone marrow. Ptp4a3 encodes a tyrosine phosphatase that is implicated in Flt3 signaling. Studies in a human AML patient cohort under the age of 60 (n=454) revealed that PTP4A3 expression is negatively correlated with prognostic outcome, both for overall survival (p-value <0.0001, HR = 1.269) and event-free survival (p-value <0.0001, HR = 1.261). Additional analysis showed that this negative correlation with survival was independent of other prognostic parameters as age, white blood cell count, cytogenetic risk and mutation status of NPM1, CEBPA and internal tandem duplication of FLT3 (FLT3ITD).

In conclusion, by mapping DNA methylated viral integration sites in murine leukemias induced by retroviral integration mutagenesis (MRIM), followed by expression level analysis, we identified Ptp4a3 as a candidate HIG. PTP4A3 has recently been proposed to have a role in drug-resistance in AMLs with FLT3ITD. This finding, together with the observation that high PTP4A3 expression negatively correlates with prognostic outcome, marks PTP4A3 as a potential therapeutic target in AML.

1. E. Taskesen et al., HAT: hypergeometric analysis of tiling-arrays with application to promoter-GeneChip data. BMC Bioinformatics, 2010. 11: p. 275.