The RNA Binding Protein MSI2 Has Increased RNA Binding Activity in Leukemic Stem Cells Compared to Normal Hematopoietic Stem Cells

The RNA binding protein (RBP) MSI2 has been demonstrated to regulate the self-renewal activity and differentiation program of normal hematopoietic stem and progenitor cells (HSPCs), as well as leukemic stem cells (LSCs) (Kharas et al, Nature Medicine 2010, Park et al, 2013 Journal of Experimental Medicine, Park et al JCI 2015). Yet, the extent of MSI2 targets and their differential activity in these rare cells are not known. Standard techniques for identifying RBP targets, such as RNA-IP and CLIP, require large cell numbers (>10 million) and are technically challenging. Here we employed HyperTRIBE, a recently discovered method (McMahon et al. 2016 and Xu et al., 2017), to globally map the MSI2 targeting network in HSPCs and LSCs. In HyperTRIBE, the catalytic domain of the Drosophila ADAR (adenosine deaminase enzyme) is fused with an RBP. This fusion protein leaves a "fingerprint" on the RBP RNA targets by marking the binding sites with an editing A-to-G event. HyperTRIBE was originally developed in Drosophila, but it is not known whether it can be used as a tool for mapping RBP targets in mammalian systems.

To test this HyperTRIBE method, we first overexpressed MSI2-HyperTRIBE fusion, MSI2-ADAR dead catalytic mutant fusion control, or an empty vector control in the human AML cell line MOLM-13. Of the MSI2-HyperTRIBE identified targets, 75% overlapped with the iCLIP targets found in the NB4 cell line, suggesting that HyperTRIBE consistently identifies previously published direct MSI2 targets. We next employed MSI2-HyperTRIBE to determine MSI2 RNA targets in rare hematopoietic stem and progenitor cells. We overexpressed MSI2-HyperTRIBE or controls in murine LSKs in vivo for seven weeks and sorted long term HSCs (LT), short term HSCs (ST), multipotent progenitors (MPP2 and MPP4) for RNA-seq. Using 360 to 20,000 sorted cells and as low as 0.1ng of RNA, we were able to identify MSI2 targets in all four HSPC populations (856 in LT-HSC, 783 in ST-HSC, 659 in MPP2, and 661 in MPP4) for the first time. Homer de novo motif search revealed the previously established MSI2 binding motif in all four HSPC populations. Overlapping the targets in LT-HSC, ST-HSC, and MPPs uncovered the unique MSI2 targeting network in LT-HSCs. Gene set enrichment analysis by ENRICHR demonstrated that these targets have a strong signature matching with the hematopoietic stem cell and myelodysplastic syndrome gene sets. These data can be used to unravel the changes in MSI2 targeting program in different stages of stem and progenitor differentiation for the first time.

To decipher the differential regulation of MSI2 in normal HSPCs and LSCs, we performed HyperTRIBE in normal mouse LSKs and MLL-AF9 LSCs in vitro (RNA-sequencing was performed 48 hours after of transduction). Strikingly, there is a higher number of editing sites and target genes in LSCs compared to LSKs (2.5 times more editing sites and 1.4 times more genes) despite lower MSI2-HyperTRIBE expression in LSCs compared to LSKs and comparable endogenous MSI2 expression in both cell types. Overlapping the targets in these two cell types showed 82% of LSK targets were also found in LSCs. Furthermore, LSCs have three times more unique targets than LSKs. We observed, in most of the cases, an increase in the number of binding sites per MSI2 target in LSCs compared to LSKs, suggesting a higher affinity of MSI2 to its mRNA targets in LSCs. Of note, this phenomenon seems to be independent from the target mRNA abundance. Previously established targets such as Hoxa9 and Ikzf2 were found to be more efficiently edited and had more binding sites, indicating that MSI2 was more active in LSCs compared to LSKs. In summary, we demonstrated that MSI2 has higher activity at these genes in LSCs compared to LSKs. These data provide evidence for why leukemia cells are more dependent on MSI2 than HSPCs and suggest that expression of RBPs themselves can be distinct from their targeting activity. Moreover, our strategy provides a new platform for discovery of RBP targets in rare cells that can be generally applied to a wide range of dysregulated RBPs in cancer.