NRAS^G12V-Mediated Self-Renewal Signaling in Self-Renewing AML Stem Cells.

In acute myeloid leukemia (AML) standard therapies often induce complete remission, but patients frequently relapse and die of the disease. Leukemia stem cells (LSCs) have self-renewal potential and ability to recapitulate the disease. Our goal is to define the molecular mechanisms that allow AML to relapse. We have previously shown that activated NRAS (NRAS^G12V) facilitates self-renewal in the LSC-enriched subpopulation in a mouse model of AML (Mll-AF9/NRAS^G12V, Sachs et al. Blood 2014). We subsequently utilized single-cell RNA sequencing of the LSCs from this model to define and validate the only subset of the LSC-enriched population that can efficiently transplant leukemia in mice. We hypothesize that NRAS^G12V exerts a unique signaling profile that directs self-renewal in this subset of LSCs. Understanding these pathways at the single-cell level would enable us to design rational therapeutics that would prevent relapse in AML.

We used mass cytometry (CyTOF2) to define the signaling activation state of LSC subsets in our AML model. Similar to flow cytometry, mass cytometry provides quantitative measurements of cell-surface and intracellular proteins at the single-cell level. In addition, it can simultaneously and accurately measure over 40 proteins, allowing us to quantitate a panel of intracellular signaling molecules in well-defined immunophenotypic leukemia subpopulations. We previously reported that the LSC-enriched population in this leukemia model is Mac1^LowKit⁺Sca1⁺ (MKS) and subsequently showed that the self-renewing subset within the MKS population is MKSCD36^LowCD69^High. In contrast, the MKSCD36^HighCD69^Low population is incapable of transplanting leukemia in mice.

The MKS cells displayed elevated levels of activated signaling molecules relative to the non-MKS population. Comparing the MKS subsets to each other, we found that the self-renewing MKSCD36^LowCD69^High population displayed significantly higher levels of several signaling molecules including Myc, NF-kB, and β-catenin relative to MKSCD36^HighCD69^Low cells (which lack self-renewal capacity). We reasoned that self-renewal might be mediated through these signaling molecules uniquely elevated in MKSCD36^LowCD69^High cells. Next, we sought to define the global signaling activation network within individual MKS subsets to determine if the signaling cascades and dependencies vary between these populations. We used Bayesian network modeling (Sachs K et al. Science 2005) to compare the statistical relationships between these signaling molecules, at the single-cell level. Signaling molecules that impact the levels of downstream effectors can be inferred using this approach. Using this method, we found that the signaling activation network does not significantly vary between MKS subsets. These observations suggest that self-renewal may be driven by alteration in the levels of signaling intermediates rather than alternate signal transduction architecture. We previously found that NRAS^G12V-mediated signals drive self-renewal in this AML model (Sachs Z. et al. Blood 2014). We used this model to ask which of these self-renewal-associated signaling molecules might be NRAS^G12V-regulated. We abolished NRAS^G12V transgene expression in these mice and harvested leukemia cells 72 hours later (per our standard lab protocol). Using this approach, found that self-renewal-associated signaling molecules, including NF-kB and β-catenin, are significantly reduced after NRAS^G12V-withdrawal indicating that NRAS^G12V -dependent signaling likely leads to the increase in these signaling molecules.

In conclusion, we used mass cytometry analysis to identify the LSC self-renewal-associated signaling state in a murine model of AML and show that NRAS^G12V activates this signaling program. These data can be used to rationally design therapeutics such as small molecule inhibitors to target self-renewal-specific signaling and prevent relapse in AML.