A Tumor Suppressor microRNA Defines the Leukemic Hierarchy in Acute Myeloid Leukemia

In order to identify functionally relevant microRNAs (miRNA) in AML, we profiled global miRNA expression in a murine AML progression model based on Hoxa9 and Meis1 overexpression. We found miR-155 and miR-708 as the most significantly upregulated miRNA species in leukemic Hoxa9/Meis1 cells compared to the preleukemic Hoxa9/ctrl cells (both p<0.01). Subsequent analysis of various AML subtypes (CN-AML, t(11q23), t(8;21), t(15;17), n=38) showed significantly elevated levels of miR-155 and miR-708 in all tested samples (n=5-10) compared to total bone marrow from healthy donors, indicating potential oncogenic roles for these miRNAs.

We further investigated, in vivo, the role of miR-155 and miR-708 in AML by retroviral overexpression with Hoxa9 in murine bone marrow (mbm) cells followed by syngeneic transplantation. Overexpression of miR-155 in conjunction with Hoxa9 (Hoxa9/miR-155) caused a significantly accelerated onset of a myelomonocytic leukemia within 93 days (survival 93±23 days, p<0.0001), but still a less aggressive course of disease compared to mice transplanted with Hoxa9/Meis1 (survival 36±4 days, p<0.0001).

We then hypothesized that the combination of miR-155 and miR-708 could further replace the oncogenic potential of Meis1. Therefore, mbm cells were retrovirally transduced with Hoxa9, miR-155 and miR-708 (Hoxa9/155/708) or Hoxa9 and miR-708 (Hoxa9/miR-708) and functionally analyzed in vitro and in vivo. To our surprise, miR-708 abrogated the leukemogenic effect of Hoxa9, alone or in combination with Hoxa9/miR-155 in vivo (p=0.0117, p<0.0001, the mice were sacrificed after 197 days), with little or no engraftment.

To understand why miR-708, a potent tumor suppressor miRNA, is upregulated in the highly aggressive Hoxa9/Meis1 AML cells, we first showed in vivo that miR-708 does not impair the development of AML by Hoxa9/Meis1 cells. In order to explore the role of miR-708 in leukemia initiating cells (LIC), we FACS-sorted subpopulations enriched for LIC based on c-kit, Mac-1 and Gr-1 expression within the Hoxa9/Meis1 cells. Of the three subpopulations (c-kit⁺ Gr-1^- Mac-1^-, c-kit⁺ Gr-1⁺ Mac-1⁺, c-kit^- Gr-1⁺ Mac-1⁺) the transplanted c-kit⁺ Gr-1^- Mac-1^- cells caused a significantly shorter survival (survival 36 days, range: 32-37d) compared to the other sorted subpopulations (p=0.0072 and p=0.0021, respectively), whereas the c-kit^- Gr-1⁺ Mac-1⁺ subpopulation exhibited the longest survival (survival 44 days, range: 42-51d). Based on these findings, we measured miR-708 and miR-155 levels in the sorted Hoxa9/Meis1 subpopulations by qRT-PCR. Strikingly, no difference compared to bulk cells was detected for miR-155, highlighting its role as an oncogenic driver in all subpopulations. In contrast, miR-708 showed a distinct expression pattern with lower expression in the LIC-enriched c-kit⁺ Gr-1^- Mac-1^- subpopulation compared to bulk (p=0.032), allowing discernment of the LIC enriched population and highlighting its role as an orchestrator of the leukemic hierarchy. Adapted from these findings, we developed an AML 4-step model (Hoxa9/miR-708, Hoxa9/ctrl, Hoxa9/miR-155, Hoxa9/Meis1) based on four immortalized Hoxa9 based cell lines with completely different in vivo properties ranging from no engraftment to induction of highly aggressive AML.

Considering the central role of Hoxa9/Meis1 in AML, we demonstrate, for the first time, the concept of a tumor suppressor miRNA that stratifies the leukemic hierarchy driven by an oncogenic miRNA. Moreover, we present a new AML 4-step model as a novel tool that will allow detailed investigation of the impact of new oncogenes at every step of leukemogenesis.