Search

Background: Treatment with hypomethylating agents such as decitabine, which results in complete and partial remission rates of up to 50%, has become standard of care in older patients with acute myeloid leukemia (AML) who are no candidates for intensive chemotherapy. While clinical parameters such as reduced performance status, high leukocyte counts, elevated lactate dehydrogenase (LDH) levels as well as poor-risk cytogenetics are associated with lower response rates and shorter overall survival, there exists only limited data on molecular biomarkers that enables for selection of AML patients who are likely to benefit from epigenetic therapy. Shen et al. (JCO 2010) could show that distinct DNA methylation changes are prognostic for overall survival in myelodysplastic syndrome patients, however, response to decitabine therapy could not be predicted. Additionally, mutations in the epigenetic-modifier gene DNMT3A were recently reported to be associated with response and survival in AML patients treated with hypomethylating agents, however, while some of these studies revealed a positive association with outcome, other studies showed no association with response and overall survival (Metzeler et al., Leukemia 2012, DiNardo et al., Leuk Lymph 2014, Coombs et al., Blood 2015). Results: In order to contribute further knowledge regarding prognosis and response to hypomethylating therapy in AML patients, we investigated distinct genetic (FLT3-ITD, NPM1, DNMT3A) and epigenetic (estrogen receptor alpha (ERα), C/EBPα and OLIG2) aberrations in 87 AML patients from the recently published phase II decitabine trial (AML00331, Lübbert et al. Haematologica 2012) to identify potential molecular biomarkers. While FLT3-ITD and NPM1 mutational status were not associated with survival or response to therapy, in our cohort patients harboring DNMT3A R882 mutations showed shorter overall survival (hazard ratio (HR): 2.15, 95%-confidence interval (CI): 0.91-5.12, p=0.08). Promoter DNA methylation analyses using pyrosequencing revealed shorter overall survival of patients with higher levels of methylation of ERα (HR: 1.50, CI: 0.97-2.32, p=0.07) and OLIG2 CpG4 (HR: 1.52, CI: 0.96-2.41, p=0.08), while DNA methylation of C/EBPα showed no association with outcome. Importantly, in multivariate analyses adjusted for clinical baseline parameters, the impact of ERα and OLIG2 CpG4 methylation was conserved (HR: 1.74, CI: 1.03-2.96, p=0.04 and HR: 1.61, CI: 0.96-2.71, p=0.07, respectively) whereas the effect of DNMT3A R882 mutations was reduced to HR: 1.94 (CI: 0.79-4.73, p=0.15). In contrast, none of the investigated molecular markers was associated with response to treatment. Conclusion: In addition to the well established adverse prognostic clinical parameters such as patients' age, reduced performance status and elevated LDH levels, we provide further evidence that pretreatment genetic and especially epigenetic analyses including DNMT3A R882 mutation status, as well as ERα and OLIG2 CpG4 DNA methylation status may be potential molecular biomarkers in AML patients undergoing epigenetic therapy. Disclosures Lübbert: Celgene: Other: Travel Funding; Ratiopharm: Other: Study drug valproic acid; Janssen-Cilag: Other: Travel Funding, Research Funding.

Introduction: DNA methylation differences between normal and cancer tissue that result in differential expression of genes are a hallmark of acute myeloid leukemia (AML). DNA methylation mediated silencing of specific genes, especially transcription factors, can provide a growth advantage for malignant cells. Global DNA methylation analyses have not only led to a better understanding of AML subgroups and the impact of epigenetic aberrations in leukemogenesis, but also to the identification of new epigenetically regulated genes. We and others have recently identified the oligodendrocyte lineage transcription factor 2 (OLIG2) as differentially methylated in AML cell lines compared with normal bone marrow cells. Aim of the study: With the hypothesis that OLIG2, which is not expressed in normal hematopoiesis, may play a role in cancerogenesis as shown for acute lymphoblastic leukemia (Lin et al., Cancer Res. 2005) and malignant glioma (Mehta et al., Cancer Cell 2011), we sought to further dissect the impact of OLIG2 in AML, implementing functional studies and primary samples. Results: First, in a cohort of 93 AML patients, we could confirm previous results by Kröger et al. (Blood 2008) that OLIG2 is differentially methylated: using pyrosequencing, 37 patients (39.8%) showed methylation levels > 25% (range (r): 26-79%) in the 5 CpG containing amplicon of the OLIG2 promoter region, independent of cytogenetic subgroup. In a small subset of 13 patients where expression-data was available, an inverse correlation between OLIG2 DNA methylation and mRNA expression was significant (r2=0.55, p Interestingly, while CD 34+ cells from two healthy donors and 10 out of 12 AML patients where protein was available, showed no protein expression, OLIG2 was expressed in 2 patients, both bearing the translocation t(15;17). This corresponds well to OLIG2 expression of cell line NB-4, which also harbours t(15;17). Treatment of non-expressing cell lines PL-21 and U937 with 200 nM 5-aza-2'-deoxycytidine led to robust re-expression of OLIG2, both on mRNA and protein level, strongly implicating DNA methylation as a silencing mechanism in a subset of AML. To investigate the relationship between OLIG2 expression and AML cell growth we used a siRNA transient knock-down in OLIG2 expressing cell lines THP-1 and NB-4. While OLIG2 protein expression measured via densitometry could be strongly reduced to 38% and 45% from pre-treatment levels in THP-1 and NB-4 cells, respectively, no change on cell viability or cell growth was detected. However, stable over-expression of OLIG2 using the lentiviral-vector pLeGO-iG in Kasumi-1 cells, led to a significant growth-inhibition of 32.2% (r: 27.0-37.3%) after 5 days and a 47.7% (r: 30.7-64.6%) increase of apoptotic cells (Annexin-V-staining) as compared to control-vector transfected cells. This negative effect on cell proliferation supports our presumption that OLIG2 could act as a growth-regulator in a subgroup of AML. This could be caused by a direct interaction between OLIG2 and a cell cycle regulator or a transcription factor complex. Conclusion: We show that OLIG2 (I) is in part epigenetically regulated via DNA methylation in AML, resulting in an inverse correlation between DNA methylation and expression; (II) can be re-expressed upon demethylating treatment in cell lines, therefore making it an attractive biomarker to study in AML patients treated with demethylating agents; (III) shows antiproliferative activity in leukemia cell lines and thus should be further studied as a potential tumor suppressor in AML. Disclosures Lübbert: Cephalon / TEVA: Travel support Other.

The chromosomal translocation (8;21) results in the expression of the chimeric transcription factor AML1/ETO, the most frequent fusion gene in acute myeloid leukemia. The AML1/ETO fusion protein acts as a transcriptional repressor by mediating epigenetic silencing through recruitment of histone deacetylases. Recently, it was shown that it also mediates gene silencing by associating with DNA methyltransferase (Dnmt). We therefore hypothesized that cells expressing AML1/ETO might be preferentially sensitive to the effects of an inhibitor of Dnmt activity, and might provide a superior model for in vitro demethylation and reactivation of the promoter of the p15/INK4b gene (encoding a negative regulator of the cell cycle) that is frequently methylated and silenced in AML and MDS. The 3 myeloid cell lines Kasumi-1 cells (AML1/ETO-positive), KG-1, and KG-1a (both AML1-ETO-negative) are all bearing a heavily methylated p15/INK4b promoter. They were treated with 50 – 1000 nM 5-aza-2′-deoxycytidine (DAC) for three pulses of 24 hrs each. After 6 days, cell growth and viability were determined and FACS analysis performed after propidium iodide staining. Kasumi-1 showed the highest sensitivity to DAC treatment (growth inhibition at 500 nM DAC: Kasumi-1 74.28 %, KG-1 69.16 %, KG-1a 62.38 %). In addition, DAC treatment (500 nM) led to a stronger increase in the sub-G1 fraction in Kasumi-1 (30.46%) compared to KG-1a (20.84 %). Regional p15/INK4b promoter methylation was assessed quantitatively by bisulfite sequencing of ≤10 individual cloned alleles (containing 21 CpGs residues) for calculation of methylated CpG percentage. The p15/INK4b was highly methylated in all 3 cell lines (methylated CpGs Kasumi-1 95.2 %; KG-1 89.6 %; KG-1a 98.4 % ). In Kasumi-1 cells, treatment with DAC resulted in a striking, dose-dependent regional demethylation of the p15/INK4b promoter (demethylated CpGs at 200 nM of DAC: Kasumi-1 63.8 %, KG-1 48.9 %, KG-1a 9.3 %). No demethylating effect was achieved with equitoxic doses of cytarabine or melphalan. Effective demethylation of the p15/INK4b promoter was associated with p15/INK4b protein induction as determined by Western Blot. Simultaneous treatment with all-trans retinoic acid (ATRA) enhanced the effects of DAC treatment upon growth inhibition, but not upon p15/INK4b induction. U937 cells with ecdysone inducible AML1/ETO expression (Fliegauf et al, Oncogene 2004) were also treated with different doses of DAC. When AML1/ETO was induced, U937 cells showed a higher growth inhibition (U937 + AML1/ETO 38.6 %, U937 - AML1/ETO 18 % at 25 nM) and increase in Sub-G1 (U937 + AML1/ETO 18 %, U937 - AML1/ETO 10.55 % at 100 nM) after treatment with DAC. Our results imply that the growth-inhibitory and proapoptotic effect of DAC on leukemia cells is modulated by AML1/ETO protein (or its target genes). The greater accessibility of the p15/INK4b promoter to the demethylating effect of DAC in AML1/ETO expressing Kasumi-1 cells may also be due to differences in regional chromatin structure. With their differential sensitivity to DAC, the cell lines Kasumi-1 and KG-1a provide a model for the different responses of leukemic blasts to DAC.

Background: The translocation (8;21) is the most common chromosomal rearrangement in AML, resulting in the expression of the fusion protein AML1/ETO. We have developed an ecdysone-inducible U937 model, in which AML1/ETO is expressed in response to treatment with Ponasterone (Pon) A (Fliegauf et al, Oncogene 2004). This model system was used to determine the cellular effects of AML1/ETO and to identify its target genes in U937 cells. Methods: Effects of AML1/ETO expression upon cell growth, viability, cell cycle and apoptosis were analyzed by trypan blue exclusion, FACS analysis using propidium iodide and DiOC6 staining, DNA laddering and Western blot for PARP cleavage, respectively. The gene expression profile of U937 with and without conditional AML1/ETO expression was assessed using Affymetrix U133A microarrays. Wild-type U937 cells with and without PonA treatment as well as AML1/ETO-negative and AML1/ETO-positive myeloid cell lines served as controls. Northern and Western Blotting were used for validation of expression changes. Results: Induction of AML1/ETO expression in U937 resulted in reduced cell growth, G1 arrest and in apoptosis beginning 48–72 hours after PonA treatment. To investigate the underlying mechanisms, microarray analysis was performed. Expression profiles of AML1/ETO-positive and AML1/ETO-negative cell lines formed distinct clusters. Based on stringent criteria, 191 different genes were found upregulated, whereas 37 were downregulated upon expression of AML1/ETO in U937. The identified genes were screened for genes with known functions in cell cycle and apoptosis by automated and manual review and included 13 apoptosis-related genes. Among them, the CDK inhibitor p21/WAF/CIP1 was upregulated 19-fold upon induction of AML1/ETO, whereas the apoptosis regulator MCL-1 was induced 2.5-fold. Based on our criteria, no differential expression of other transcriptionally-controlled apoptosis regulators (such as BCL2, BAX, BAK1, BAD or c-flip) was noted. Northern and Western Blot analysis confirmed the strong induction of p21/WAF/CIP1 that paralleled the expression of AML1/ETO 10 hours after PonA treatment. Induction of p21/WAF/CIP1 was independent of the tumor suppressor protein p53 (Dou et al., Proc. Natl. Acad. Sci. 1995), and by Western blot, p53 was undetectable in U937. Northern Blot analysis revealed a higher expression of p21/WAF/CIP1 in the AML1/ETO-positive cell lines Kasumi-1 and SKNO-1 than in the AML1/ETO-negative cell lines HL-60, KG-1 and U937, supporting our finding that AML1/ETO may induce p21/WAF/CIP1. Conclusions: AML1/ETO expression resulted in increased expression of p21/WAF/CIP1, which might contribute to the observed growth arrest and induction of apoptosis caused by the conditional expression of AML1/ETO.