Hypermethylation and down-regulation of DLEU2 in paediatric acute myeloid leukaemia independent of embedded tumour suppressor miR-15a/16-1

Abstract

Background: Acute Myeloid Leukaemia (AML) is a highly heterogeneous disease. Studies in adult AML have identified epigenetic changes, specifically DNA methylation, associated with leukaemia subtype, age of onset and patient survival which highlights this heterogeneity. However, only limited DNA methylation studies have elucidated any associations in paediatric AML.

Methods: We interrogated DNA methylation on a cohort of paediatric AML FAB subtype M5 patients using the Illumina HumanMethylation450 (HM450) BeadChip, identifying a number of target genes with p <0.01 and Δβ >0.4 between leukaemic and matched remission (n = 20 primary leukaemic, n = 13 matched remission). Amongst those genes identified, we interrogate DLEU2 methylation using locus-specific SEQUENOM MassARRAY® EpiTYPER® and an increased validation cohort (n = 28 primary leukaemic, n = 14 matched remission, n = 17 additional non-leukaemic and cell lines). Following methylation analysis, expression studies were undertaken utilising the same patient samples for singleplex TaqMan gene and miRNA assays and relative expression comparisons.

Results: We identified differential DNA methylation at the DLEU2 locus, encompassing the tumour suppressor microRNA miR-15a/16-1 cluster. A number of HM450 probes spanning the DLEU2/Alt1 Transcriptional Start Site showed increased levels of methylation in leukaemia (average over all probes >60%) compared to disease-free haematopoietic cells and patient remission samples (<24%) (p < 0.001). Interestingly, DLEU2 mRNA down-regulation in leukaemic patients (p < 0.05) was independent of the embedded mature miR-15a/16-1 expression. To assess prognostic significance of DLEU2 DNA methylation, we stratified paediatric AML patients by their methylation status. A subset of patients recorded methylation values for DLEU2 akin to non-leukaemic specimens, specifically patients with sole trisomy 8 and/or chromosome 11 abnormalities. These patients also showed similar miR-15a/16-1 expression to non-leukaemic samples, and potential improved disease prognosis.

Conclusions: The DLEU2 locus and embedded miRNA cluster miR-15a/16-1 is commonly deleted in adult cancers and shown to induce leukaemogenesis, however in paediatric AML we found the region to be transcriptionally repressed. In combination, our data highlights the utility of interrogating DNA methylation and microRNA in combination with underlying genetic status to provide novel insights into AML biology.

Figure 1

Regional interrogation of the DLEU2 gene including significantly differentially methylated probes identified by HM450 analysis. Bottom: Distribution of HM450 methylation probes across the DLEU2 region of 13q4. Genes located in this region include: DLEU1, DLEU2, miR15a/16-1 microRNA cluster, TRIM13, KCNRG and miR-3613.Top: HM450 probes identified as significantly differentially methylated between paediatric AML (FAB subtype M5) and non-leukaemic specimens have been plotted against genomic location. The leukaemic group (n = 16) refers to diagnostic bone marrow from paediatric patients. Non-leukaemic group (n = 11) consists of CD sorted cell populations (CD19+, CD33+, CD34+, CD45+) and patient remission specimens. Differential methylation is concentrated in the promoter region of the DLEU2/Alt1 long transcript, which also falls into the body region of DLEU1 within 3 CpG islands (Chr 13: 50, 690,000-50,708,000). Mean methylation β values and 95% CI are shown. Data for leukaemic samples are red and non-leukaemic samples in green. Methylation values range from 0 (0%, no detected methylation) to 1.0 (100% fully methylated). Significantly differentially methylated regions between leukaemic and non-leukaemic samples to p < 0.001 are highlighted with **.

Figure 2

Expression analysis of DLEU2 and miR-15/16 in paediatric AML. Here we interrogate the gene, miRNA and precursor miRNA expression in paediatric AML compared to non-leukaemic. This interrogation includes DLEU2 and embedded miR-15a/16-1 on chromosome 13q4. The leukaemic group refers to diagnostic bone marrow from paediatric patients. The non-leukaemic group consists of CD sorted cell populations (CD19+, CD33+, CD34+, CD45+) and patient remission specimens, and is represented by the dashed line at Y = 1. Fold Change (FC) is plotted using normalized data and the 2^-ΔΔCt method ± SD, and shows the fold change calculated from the means of each group. A. Gene expression including DLEU1, DLEU2 and TRIM13 in paediatric AML (n = 10) compared to non-leukaemic (n = 13) expression. DLEU2 shows a significant down-regulation in AML compared to non-leukaemic expression (0.07 FC, p = 0.014 represented by **), however there is no significant change in expression for TRIM13 or DLEU1. B. Mature microRNA expression, including primary precursor transcript (PRI) and alternate miRNA expression (*), from the miR-15a/16-1 miRNA cluster embedded within DLEU2 for paediatric AML (n = 28, including the 10 used in Figure  2A) compared to non-leukaemic specimens (n = 30, including the 13 used in Figure  2A). Here the miR-15a/16-1 PRI transcript is 3.03-fold higher in expression compared to non-leukaemic expression. Additionally, miR-16-1* (2.52 FC), miR-15a* (2.24 FC) and miR-15a (1.5 FC) also show increases in expression in paediatric AML. No significant change in expression was observed for miR-16 (0.94 FC) in paediatric AML compared to non-leukaemic expression.

Figure 3

Interrogation of paediatric AML by clinically defined cytogenetic and FAB subtypes alongside DLEU2 promoter methylation. Paediatric AML diagnostic cytogenetic and subtyping analyses were specifically investigated, including those with known gene abnormalities and CN-AML cases. We investigate here the DNA Methylation of DLEU2/Alt1 promoter region to 95% CI. DNA methylation values range from 0.0 (0% no detected methylation) to 1.0 (100% fully methylated). The leukaemic group refers to diagnostic bone marrow from paediatric patients. Non-leukaemic group consists of CD sorted cell populations (CD19+, CD33+, CD34+, CD45+) and patient remission specimens. A. DNA methylation for DLEU2 HM450 probes (cg12883980, cg20529344, cg5394800) according to cytogenetic type, showing heterogeneous DNA methylation outcomes. Of note, a subset of patients with observable chromosome 11 and trisomy 8 abnormalities have a reduced DNA methylation compared to all other AML abnormalities. B. A subset of DLEU2 DNA methylation results for paediatric AML chromosome 11 and trisomy 8 abnormality patients from Figure  3A fall into a range akin to non-leukaemic specimens. Grouping these patients, regardless of clinical subtyping (M5a, M5b or M1/M2/M4), forms an additional sub-group, denoted here as ‘t(11)/+8’. The DNA methylation values obtained for probe cg12883980 are depicted, as an indication of the methylation at the DLEU2 promoter. t(11)/+8 are not differentially methylated compared to non-leukaemic specimens (t(11)/+8 sub-group Mean Methylation (MM) = 0.38, non-leukaemic MM = 0.32. p = 0.22). M5a, M5b and M1/M2/M4 AMLs (with the removal of t(11)/+8 patients) have become increasingly hypermethylated compared to non-leukaemic methylation (M5a MM = 0.81, M5b MM = 0.78, M1/M2/M4 MM = 0.87. p < 0.0001), and significantly different from t(11)/+8 (M5a/b p < 0.05; M1/M2/M4 p < 0.0001).

Figure 4

Gene, primary, miRNA and miRNA* expression for paediatric AML defined through clinical classification and DLEU2 Methylation subtyping. Mature microRNA expression, primary precursor transcript (PRI) and alternate miRNA isoform expression (*) from the miR-15a/16-1 miRNA cluster embedded within DLEU2 for paediatric AML patients (n = 26: 12 M5a, 5 M5b, 4 M1/M2/M4, 5 t(11)/+8 sub-group) all compared to non-leukaemic specimens (n = 30). Leukaemic groups refer to diagnostic bone marrow from paediatric patients. Non-leukaemic group consists of CD sorted cell populations (CD19+, CD33+, CD34+, CD45+)and patient remission specimens. Linear Fold Change (FC) is plotted using normalized data and the 2^-ΔΔCt method ± SD, and shows the fold change calculated from the means of each group. DLEU2 gene expression is down-regulated in all subtypes. A. Fold change in expression comparing non-leukaemic to t(11)/+8 subtype. No significant differences in RNA expression are observed between non-leukaemic specimens and this subgroup. B. Fold change in expression comparing patients from subtype M5a to M1/M2/M4. DLEU2 is down-regulated in all subtypes (as previously described in Figure  2), and primary precursor for miR-15a also appears down-regulated (non-significant). M1/M2/M4 groupings do not show any mature miRNA expression changes (defined as >2-fold difference from non-leukaemic). Subtype M5a shows a 2.1-fold (±0.8 SD) increase in miR-15a* and a 2.81-fold (±0.4 SD) increase in miR-16-1*. C. FAB subtype M5b shows up-regulation of miR-15a PRI compared to non-leukaemic expression (20.29-fold (±0.6 SD) p < 0.001), with miR-15a* (4.87-fold ±1 SD) and miR-16-1* (9.86-fold ±1 SD) also up-regulated. M5b additionally shows a significant up-regulation of miR-15a PRI compared to t(11)/+8 sub-group samples (p < 0.05), and also from M5a and M1/M2/M4 sub-groups (p < 0.001).