Loss-of-Function Mutations of MIR142 May Contribute to Leukemogenesis By Expanding the Pool of Myeloid-Biased HSCs

MicroRNA (miRNA) expression is sometimes dysregulated in acute myelogenous leukemia (AML), and this dysregulation has been suggested to play a role in leukemic transformation. However, somatic mutations in miRNAs are infrequent in AML. Through whole genome or exome sequencing of 200 cases of de novo AML, The Cancer Genome Atlas (NEJM 2013) identified recurring point mutations in MIR142. Heterozygous point mutations of MIR142 were identifiedin 3 cases, and bi-allelic mutations in 1 case (total incidence of 2%). All of these mutations localized to the "seed" sequence of miR-142-3p, which is critical for determining mRNA target specificity. Accordingly, we show that mutated miR-142 is unable to suppress several well-known targets of miR-142. Surprisingly, sequencing of small non-coding RNAs in AML cases carrying MIR142 mutations showed a selective loss of miR-142-5p expression. Indeed, we provide evidence that mutant miR-142-3p is selectively loaded into the RNA-induced silencing complex (RISC), resulting in degradation of miR-142-5p. Collectively, these data show that MIR142 mutations in AML likely disrupt both miRNA-142-5p and miRNA-142-3p function.

To model the effect of the loss of miR-142 on hematopoiesis, we analyzed Mir142^-/- mice. Prior studies in zebrafish showed that knockdown of miR-142-3p results in reduced hematopoietic stem cells (HSCs) and impaired myelopoiesis (Fan, Blood, 2014; Lu, Cell Research, 2013). Sun et al reported impaired T-cell responses in Mir142^-/- mice (Sun, JCI 2015). Here, we show that loss of miR-142 is associated with a modest increase in bone marrow and splenic neutrophils. Erythroid precursors in the bone marrow are significantly reduced with a corresponding increase in the spleen. Consistent with these data, granulocyte-macrophage progenitors (GMPs) in the bone marrow are significantly increased, while megakaryocyte-erythroid progenitors (MEP) are significantly decreased. While the total number of phenotypic HSCs (CD150⁺ CD48^- Kit⁺ Sca⁺ lineage^- cells) in the bone marrow is similar to control mice, a marked increase in the percentage of CD229^- myeloid-biased HSCs was observed in Mir142^-/- mice (69.4% ± 3.4) versus control mice (29.2% ± 3.3; P <0.001). Consistent with these findings, competitive repopulation assays show that the long-term repopulating activity and self-renewal capacity of Mir142^-/- HSCs is normal. However, lineage analysis of these mice revealed a strong myeloid bias. Together, these data suggest that miR-142 expression in HSCs normally inhibits commitment to the myeloid lineage.

To assess the hematopoietic cell-intrinsic leukemogenic potential of the loss of miR-142, we transplanted Mir142^-/- bone marrow into irradiated wild-type mice, and a tumor watch was established. No myeloproliferative disorder (MPD) was observed after one year of follow-up, suggesting that loss of Mir142 is not sufficient to induce AML in mice. All 4 human AML cases carrying MIR142 mutations also harbor mutations in IDH1/2. To assess the functional importance of this association, we transduced Mir142^-/- hematopoietic stem and progenitor cells (HSPCs) with a lentivirus expressing a canonical IDH2 mutation, R172H. These cells were then transplanted into irradiated mice and a tumor watch was established. Consistent with a prior report, expression of mutant IDH2 alone induced a MPD characterized by increased myeloid cells, anemia, and splenomegaly (Sasaki, Nature 2012). Surprisingly, the concomitant loss of Mir142 did not affect the latency or penetrance of MPD; the only significant difference observed was a more severe anemia.

Collectively, these data suggest that the loss-of-function mutations of MIR142 found in AML likely do not promote leukemogenesis by enhancing self-renewal capacity or inhibiting myeloid differentiation. Rather, our data suggest that these mutations promote leukemogenesis by expanding the pool of myeloid-biased HSCs, thereby increasing the likelihood of acquiring additional cooperating events, such as mutant IDH1/2. This model may explain the surprising lack of cooperativity between miR-142 loss and R172H IDH2, since these experimental mice were generated using a large number of transduced HSPCs.