Dysregulation of the Imprinted DLK1-DIO3 Locus in Promyelocytic Leukemia

Abstract 3500

Small non-coding RNAs (sncRNAs) have emerged as important regulators of multiple cellular processes, with a growing body of evidence suggesting their dysregulation and possible pathogenic role in many malignancies, including acute myeloid leukemia (AML). Previous high throughput sncRNA transcriptome sequencing studies have focused on the 15–30 nucleotide fraction that primarily consists of microRNAs (miRNAs), thus excluding many larger, potentially important sncRNAs, such as small-nucleolar RNAs (snoRNAs) and piwiRNAs. To address this “sequencing gap”, we extended our transcriptome sequencing to include RNA species of 15–75 nucleotides in length. Here, we report the small RNA transcriptome sequencing of 28 cases of AML. As a control, we also sequenced the small RNA transcriptome in CD34+ progenitors (n = 4) and promyelocytes (n=2) from healthy donors. The most common class of sncRNA detected in all samples was small-nuclear/snoRNAs (∼40% of reads), followed by miRNAs (∼17% of reads), ribosomal/transfer RNAs (15% of reads), unannotated/miscellaneous sncRNAs (∼15% of reads), and piwiRNAs (∼10% of reads). Partek Genomics Suite was used to identify significantly dysregulated sncRNAs using a p-value<0.05, false discovery rate of 5%, and a minimum 2 fold-change from normal CD34+ cells. Compared to normal CD34+ cells, expression of 158 sncRNAs was significantly increased and 29 were significantly decreased in de novo AML samples.

The most striking example of dysregulation was observed in acute promyelocytic leukemia (M3 AML). The DLK1-DIO3 locus at 14q32.2 contains 41 snoRNA genes belonging to the SNORD112–114 family, in addition to a large number of miRNAs. Compared with normal CD34+ cells or non-M3 AML, 31 of the 41 snoRNAs in this locus were massively upregulated (6–2,000 fold) in M3 AML samples. Concordantly large increases (10–1,000 fold) of 20 different miRNAs in this locus were also observed in M3 AML. We extended these findings to an independent cohort of M3 and non-M3 AML cases (n = 187) for which RNA transcriptome sequencing was available (RJW and TJL, personal communication, on behalf of the TCGA AML analysis group). Analysis of this data confirmed massive up-regulation of sncRNAs in the DLK1-DIO3 locus that was restricted to M3 AML. We considered the possibility that the increase in sncRNAs in this locus was secondary to a promyelocyte-specific expression pattern in M3 AML cells, but expression of sncRNAs in normal promyelocytes was lower than that observed in normal CD34+ cells, making this hypothesis unlikely.

The DLK1-DIO3 locus is one of the best characterized imprinted regions in the human genome. The paternally-derived protein coding genes in this locus (DLK1, RTL1 and DIO3) showed no dysregulation in M3-AML in contrast to the aberrantly expressed, maternally-derived sncRNAs. While the function of the sncRNAs in this region is largely unknown, there is evidence they may contribute to stem cell pluripotency (Stadtfeld et al. Nature 2010, Liu et al. JBC 2010). Since imprinting at this locus is controlled by methylation of several differentially methylated regions (DMRs), we analyzed array-based methylation data from the TCGA cohort to determine if imprinting was disrupted. No difference in methylation status of the DMRs was observed between M3 and non-M3 AML. Moreover, based on expressed germline single nucleotide polymorphisms, mono-allelic expression of sncRNAs in this locus was preserved. Together, these data show that imprinting of the DKL1-DIO3 locus is not disrupted in M3-AML samples, and that dysregulation of the sncRNAs in this region occur through an imprinting-independent mechanism. In summary, extended small RNA transcriptome sequencing is a valuable new tool to analyze malignant cells; it successfully identified massive dysregulation of sncRNAs in the DKL1-DIO3 locus in M3 AML. The contribution of the aberrantly expressed sncRNAs within this locus to the pathogenesis of M3 AML will require additional study.