The Polycomb group (PcG) of genes is important for differentiation and cell-cycle regulation and is aberrantly expressed in several cancers. To analyse the role of deregulated PcG genes in acute myeloid leukaemia (AML), we determined by RQ-PCR the expression of the PcG genes BMI-1, MEL18, SCML2, YY1 and EZH2, and the downstream PcG targets HOXA4, HOXA9 and MEIS1 in diagnostic bone marrow samples from 126 AML patients. There was a general overexpression of the genes in AML patients compared to 20 healthy donors, except of HOXA4 and MEL18, which both displayed a wide range of expression levels within the AML subgroups. Among the AML patients with normal karyotype, a low HOXA4 level was associated with a shorter overall survival (P = 0.005). In addition, expression levels of MEL18 and EZH2 were significantly (P < 0.025) higher in patients with complex karyotype and lower in CBF-mutated patients. The t(8;21) vs. inv(16) positive patients showed significantly different expression of SCML2, BMI-1, YY1, HOXA9 and MEIS1 (P < or = 0.01). Comparisons between the PcG and PcG-regulated genes and a number of clinical and molecular data revealed correlations to genes involved in DNA methylation (DNMT1, DNMT3B), apoptosis (BAX, CASPASE 3) and multidrug-resistance (MDR1, MRP ) (P < 0.01). In conclusion, our data suggest that the role of PcG and PcG-regulated genes in leukaemogenesis varies between, as well as within karyotypic subgroups.

SummarySilencing of the putative tumour suppressor gene retinoic acid receptor b2 (RARb2) caused by aberrant promoter hypermethylation has been identified in several solid tumours. In order to evaluate the extent of RARb2 hypermethylation and transcription in acute myeloid leukaemia (AML) at diagnosis, 320 patients were investigated by bisulphite-denaturing gradient gel electrophoresis and mRNA transcription levels were analysed in 61 of these by quantitative real-time polymerase chain reaction. The results were compared with demographic-and molecular data from the patients. While RARb2 was unmethylated in 10/10 bone marrow and 7/7 blood samples from healthy individuals, the gene was hypermethylated in 43% of the AML patients. The RARb2 degree of promoter methylation differed between and within individuals, and the mRNA transcription levels of the gene varied inter-individually by a factor of 4000. A significant inverse correlation between promoter hypermethylation and gene expression could be established (t-test, P ¼ 0AE019). Comparison of methylation data with a series of other molecular alterations in the same patient materials revealed a correlation between hypermethylation of the RARb2 promoter and the presence of CBF-MYH11 fusion transcripts (P < 0AE01). Our data suggest that RARb2 promoter methylation is frequent in AML and may co-operate with the expression of CBF-MYH11 fusion transcripts in leukaemogenesis.

The Polycomb group (PcG) of genes is important for differentiation and X-chromosome silencing. Recently much attention has been afforded to the role of its aberrant expression in cancer, especially in relation to the inactivation of tumor suppressor genes. We hypothesized that a deregulation in the expression profile may contribute to the development of acute myeloid leukaemia (AML). To address this, we determined the RNA levels by RQ-PCR in diagnostic bone marrow samples from 126 patients and 20 healthy donors to delineate their expression profile of the PcG genes BMI-1, MEL18, SCML2, YY1 and EZH2. To address the interplay with downstream targets of PcG proteins, we also determined the expression of HOXA4, HOXA9 and MEIS1. These data were compared not only to the demographic and clinical data of the patients, but also to a large number of molecular assays already performed in these patients (Olesen LH et al. Br.J. Haematol. 2005 131(4):457–467; Rethmeier et al. Br.J. Haematol. 2006 133(3):276–283.). At first we noticed a striking heterogeneity in the expression profiles of the AML patients (Fig. 1). We also observed that HOXA9, MEIS1, SCML2, YY1, BMI-1 and EZH2 were significantly (p≤0.003) higher expressed in the patients compared to the healthy donors. Moreover, when patients were analyzed according to the three cytogenetic prognostic groups (normal, core-binding factor positive and complex), the expression profile of patients with the t(8,21) aberration was characterized by a significantly decreased expression of HOXA9 and MEIS1 and a higher one of SCML2, YY1 and BMI-1 than AML patients in general (p<0.003). When evaluating the impact of cytogenetic subgrouping, the expression levels of MEL18 and EZH2 significantly (p< 0.025) reflected highest expression in patients with adverse prognosis and lowest expression with patients exhibiting the most favourable prognosis. While the expression levels of the genes in focus did not correlate to course of disease, we observed that a direct relationship between transcript levels of PcG and PcG-related on the one hand and the DNA methyl transferases (DNMT’s), apoptosis and multidrug-resistance genes (p<0.001) on the other. In conclusion, in this study, which is the first to systematically analyze a series of PcG genes and genes regulated by PcG, we failed to demonstrate a correlation to the clinical outcome of patients with AML. On the other hand, our data strongly suggest that these genes might be involved in the leukaemogenic process by virtue of their relations to DNA methylation (DNMT1, DNMT3B), apoptosis (BAX, CASPASE 3) and multidrug resistance (MDR1, MRP1). Figure 1. Expression profiles of PcG or PcG-regulated genes in AML patients and healthy controls. A. Gene expression profile of all 126 AML patients included (black lines) compared to 20 healthy donors. B. Patients with CBF aberrations, t(8,21), n =7, or inv(16), n =12. The expression is calculated as 2−ΔCt *100), where ΔCt = CtTG−CtCG, CtTG is the Ct value of the target gene, and CtCG is the mean Ct value of the two control genes (B2M and ABL). Figure Figure

Introduction: Treatment with Retinoic Acid (RA) has proved to be successful in acute myeloid leukemia (AML) patients harboring the t(15;17) aberration (Dubois et al. Blood 83;3264,1994). Promoter hypermethylation of the RA receptor Retinoic Acid Receptor β2 (RARβ2), a putative tumor-suppressor-gene, has been associated with gene silencing in several cancer types. Moreover, lack of RA response has been correlated to RARβ2 promoter hypermethylation (Sirchia et al. Cancer Research62;2455, 2002). Other studies have shown promoter hypermethylation of the this gene in t(15;17) patients and induction of demethylation, re-expression of the gene, and cell differentiation following treatment with RA (Croce et al. Science295;1079, 2002). Based on these findings we found it of interest to analyze the promoter methylation and the transcriptional status of the RARβ2 gene in a cohort of AML patients. Materials and Methods: Mononuclear cells from either peripheral blood (PB) and/or bone marrow (BM) from 229 AML patients were cryopreserved in DMSO and fetal calf serum. DNA and/or RNA were purified on a MagNa-Pure LC robot (Roche Diagnostic). Bisulfite treatment and Denaturing Gradient Gel Electrophoresis (Bisulfite-DGGE) for promoter hypermethylation analysis were performed as previously described (Aggerholm et al. Cancer Research59;436, 1999). Transcription levels were determined by real-time quantitative RT-PCR (RQ-PCR), which was performed in a 7700 Sequence Detector System (Applied Biosystems). Results: We first determined that promoter methylation of the gene was absent in BM from healthy individuals. In contrast, bisulfite-DGGE analysis showed that 38% (87/229) of AML patients were intra- and interindividually heterogeneously methylated. While methylation was found to be present in all AML subtypes, (Table), patients harboring the inv(16) fusion transcript were significantly more often hypermethylated when compared to the other AML subgroups (χ2, p = 0.0003). Somewhat puzzling, RQ-PCR analysis of BM cells from 77 AML patients showed an increase in the transcription level of RARβ2 in 42% of the patients when these were compared to the transcription levels in BM from healthy individuals. Still, the group of inv(16)+ patients differed significantly from the other subgroups by being more often transcriptional inactive (χ2, p = 0.00004). Comparisons of the RARβ2 transcription levels in the methylated and unmethylated groups of patients revealed a repressive effect of methylation in spite of the general transcriptional up-regulation of the gene (student’s T-test, p = 0.006). Conclusion: Promoter hypermethylation of RARβ2 in AML patients is present in all subtypes. Moreover, a correlation between methylation and transcriptional expression could be demonstrated even though the overall transcription level of the gene was up-regulated. Thus, it seems that successful RA treatment in t(15,17)+ patients cannot be correlated to the promoter hypermethylation status of the RARβ2 gene. On the other hand, the demonstration of a higher frequency of hypermethylation and gene silencing in inv(16)+ patients could indicate that the fusion protein created from this lesion might be involved in the mechanisms behind such epigenetically changes. Distribution of methylated and unmethylated patients with different fusion transcripts Fusions Transcript Methylated Unmethylated t(8;21) 4 9 t(15;17) 3 4 inv (16) 14 4 dupMLL 5 7 Others 1 4 No fusion transcript 60 114 Total 87 142 Table 1

Acute leukemias are remarkably heterogeneous as evidenced by the increasing number of recurring genetic and epigenetic abnormalities, especially the presence or absence of balanced translocations, which have been shown to be of independent prognostic significance. Thus, In a recent single center study we analyzed 250 AML patients for a series a genetic alteration and found that balanced translocations (identified by a multiplex PCR reaction; Pallisgaard et al, BLOOD, 1998 and Olesen et al, Brit. J. Hem. in press) were present in 17% of the patients, FLT3 internal tandem duplication in 24%, and MLL partial tandem duplication in 4%. In addition, we delineated the presence of promoter hypermethylation for the p15 (71%), MDR1 (4%), E-cadherin (CDH1) (64%), and Estrogen receptor (ER) (40%) genes by bisulfite DGGE (Aggerholm, Cancer Res. 1999) as well as the abnormal expression by RQ-PCR of genes related to increased chemotherapy resistance (MDR1 and MRP1) as well as resistance to undergo apoptosis (FAS, Bcl2, BAX and CASPASE3). Here we have performed gene expression profiling focusing on patients negative for all these molecular lesions. Out of the 250 patients, 8 were determined to be molecularly negative. Global gene expression analysis (Affymetrix human genome chip (U133A)) was performed on the 6 samples, from which high-quality RNA could be harvested. Figure Figure As will be seen from the Figure, all samples could be easily distinguished from TEL/AML pre-B ALL samples, and also in 4/6 cases clustered differently from AML/ETO+ and CBF/MYH11+ AML groups. Interestingly, as seen from the dendrogram, the 6 samples separated in three clusters, one also containing a CBF/MYH11+ patient (4 cases), one consisting of one single patient, while the last patient clustered together with the AML/ETO cases. While these data suggest that the gene expression in the molecularly negative patients can be different from that in CBF leukemias, it gives no clues for the leukemogenetic events in these patients. To that end, we analyzed the gene expression data and found 24 genes to be more than 5-fold overexpressed and 26 genes to be underexpressed compared to CBF AMLs. Of special notice was the increased usage of the prostaglandin/leukotriene metabolism pathway in two patients with strikingly similar global gene expression (arrows 3 and 4 from right on Fig.) including the prostaglandin I2 (prostacyclin) synthase, and the prostaglandin-endoperoxide synthase 2 (prostaglandin G/H synthase and cyclooxygenase) genes, which were more than 10-fold increased. Also noteworthy was the upregulation of LR8 protein IQ motif containing GTPase activating protein 1 in the remaining 4 cases. Applied together with novel global cytogenetic techniques, these data will form a platform for the identification of leukemogenesis in AML patients with no demonstrable genetic aberrations.ßßß