In-depth characterization of the microRNA transcriptome in a leukemia progression model

Abstract

MicroRNAs (miRNAs) have been shown to play important roles in physiological as well as multiple malignant processes, including acute myeloid leukemia (AML). In an effort to gain further insight into the role of miRNAs in AML, we have applied the Illumina massively parallel sequencing platform to carry out an in-depth analysis of the miRNA transcriptome in a murine leukemia progression model. This model simulates the stepwise conversion of a myeloid progenitor cell by an engineered overexpression of the nucleoporin 98 (NUP98)-homeobox HOXD13 fusion gene (ND13), to aggressive AML inducing cells upon transduction with the oncogenic collaborator Meis1. From this data set, we identified 307 miRNA/miRNA species in the ND13 cells and 306 miRNA/miRNA species in ND13+Meis1 cells, corresponding to 223 and 219 miRNA genes. Sequence counts varied between two and 136,558, indicating a remarkable expression range between the detected miRNA species. The large number of miRNAs expressed and the nature of differential expression suggest that leukemic progression as modeled here is dictated by the repertoire of shared, but differentially expressed miRNAs. Our finding of extensive sequence variations (isomiRs) for almost all miRNA and miRNA species adds additional complexity to the miRNA transcriptome. A stringent target prediction analysis coupled with in vitro target validation revealed the potential for miRNA-mediated release of oncogenes that facilitates leukemic progression from the preleukemic to leukemia inducing state. Finally, 55 novel miRNAs species were identified in our data set, adding further complexity to the emerging world of small RNAs.

Figure 1.

Overview of small RNA and miRNA gene expression in a preleukemic and leukemic cell model obtained by deep sequencing. (A) Breakdown of the proportions (in percent) of various classes of small RNAs detected by sequencing of the preleukemic ND13 library. The percentages are comparable to those found in the leukemic ND13+Meis1 library. Small RNAs belonging to the miRNA family constitute the majority (65.7% in the preleukemic and 66.2% in the leukemic cells). scRNA, small cytoplasmic RNA; snRNA, small nuclear RNA; snoRNA, small nucleolar RNA; rRNA, ribosomal RNA; tRNA, transferRNA; unknown, derived from unannotated/intergenic regions. (B) Distribution of miRNA genes expressed according to their sequence counts in the preleukemic (ND13) compared with leukemic (ND13+Meis1) cells. Shown are the numbers of unique miRNA genes plotted as a function of their expression levels as defined by a given range of sequence counts in the respective libraries of small RNAs. The total numbers of miRNA sequence counts were 1,240,570 and 1,030,414 for the preleukemic and leukemic libraries, respectively.

Figure 2.

Example of high frequency of miRNA sequence variation (isomiRs). Shown are the unique sequences and number of times this sequence was detected matching the pre-miRNA sequence of miR-181a. The most frequent occurring miR-181a sequence is not in accordance with the miRBase reference sequence. The three most common sequences were also detectable by linker-based cloning, as indicated in the figure. An example of a miR-181 isomiR not matching the genome is shown in the bottom part of the figure.

Figure 3.

Analysis of differentially-expressed miRNA genes in leukemic cells compared with preleukemic cells. (A) Distribution of differentially-expressed miRNA genes according to their fold changes. Shown are the number of miRNA genes whose expression was up-regulated (positive values) or down-regulated (negative values) in the leukemic cells as a function of the fold change. Only changes >1.5 and achieving a P-value of <0.05 were included. (B) Bubble plot depicting the abundance of selected miRNA/miRNA* species and their relative expression levels. The bubbles represent the sum of the most common sequence counts from both libraries for a miRNA/miRNA* species plotted as a function of fold difference between the leukemic versus preleukemic cells.

Figure 4.

Venn diagram of the predicted miRNA targets for the 19 most abundant miRNAs from each library and their shared targets with the Sanger Cancer Gene Census. The dark boxes indicate AML-specific oncogenes, whereas the gray box highlights a tumor suppressor gene targeted by miRNAs enriched in ND13+Meis1 cells.

Figure 5.

Dek-3′ UTR luciferase assays for miR-23a and miR-155. (A) Bar diagram demonstrating the binding of miR-23a and miR-155 to the 3′ UTR of the Dek oncogene. Dek23 comprises only binding sites for miR-23a, whereas Dek155 exhibits only predicted binding sites for miR-155. A nonbinding miRNA was used as negative control. *P < 0.05. (B) Schematic representation of the Dek 3′ UTR constructs and the predicted miRNA binding sites.