Blocking Vascular Niche E-Selectin Dampens AML Stem Cell Regeneration/Survival Potential In Vivo By Inhibiting MAPK/ERK and PI3K/AKT Signalling Pathways

We have previously shown vascular E (endothelial)-selectin to play a key role in niche-mediated chemo-resistance in Acute Myeloid Leukaemia (AML). Now we report that the cell surface glycosylation of AML blasts -and thus their E-selectin-binding potential- alters during therapy and queried whether these variations influence treatment outcome.

Using preclinical mouse models of 11q23-rearranged (MLL-AF9) monomyelocytic AML, we found that although AML blasts in untreated mice display a range of E-selectin-binding potentials, the blasts with highest E-selectin-binding potential dominate in bone marrow (BM) after 24hr cytarabine therapy. Indeed, the highest 10% of AML blasts for E-selectin-binding were >12-fold (p=0.0014) more likely to survive post-chemotherapy. These data raise the question whether E-selectin binding potential itself may prospectively identify AML blasts with heightened regenerative potential. An alternative explanation could be that high-binding AML blasts predominate after chemotherapy simply due to survival advantages mediated by vascular niche E-selectin interaction. To investigate this, BM AML blasts were FACs sorted from donor C57BL/6 mice based on E-selectin-binding potential (highest and lowest 10%) and transplanted into recipient mice (1,500 AML blasts/recipient, 8/gp). No significant differences in duration of disease-free survival was observed. Thus E-selectin-binding potential itself does not prognostically identify the most potent Leukemia Reconstituting Cells (LRC) that initiate relapse.

To determine instead whether the AML blasts that bind E-selectin can dominate during stress because of the survival advantage of interacting with E-selectin at the niche, an identical parallel experiment was performed, the only difference being that donor mice had E-selectin blocked (GMI-1271 100mg/kg BiD) for the last 48h prior to BM harvest, FACs sort of AML blasts and recipient mouse transplant. This time we observed a significant (2-fold) extension in disease-free survival in the recipients of high-binding AMLs (from donors treated with GMI-1271) compared to all other groups (p=0.012, median disease-free survival 33 vs. 63 days). Together these data indicate that administration of E-selectin antagonists, even as a single agent, may potentially improve patient outcomes - in cases where heightened E-selectin binding potential is observed.

Next we investigated the intracellular AML signaling pathways potentially dampened by E-selectin absence/blockade. Two common pathways used by malignant cells for survival/regeneration are the PI3K/AKT/mTOR/ NF-kB and RAS/MAPK/ERK pathways. We have already shown the absence (in Sele^-/- mice), or therapeutic blockade of E-selectin (with GMI-1271) in mice significantly dampens PI3K/AKT/NF-kB signaling in BM AML blasts in vivo. However, AML blasts can utilize alternative pathways such as RAS/MAPK/ERK for survival signaling as well, especially in the 30% of AMLs with NRAS mutations. So we determined whether MAPK/ERK signaling in AML could be similarly altered by E-selectin absence/blockade. Indeed MAPK/ERK phosphorylation was significantly reduced (2-fold) in BM AML blasts from host mice treated 24hr with GMI-1271 and in Sele^-/- hosts. Thus contact with vascular E-selectin induces a range of survival/regenerative signaling pathways within BM AML blasts that would be highly advantageous for the blast in times of stress.

In summary we show, (1) vascular niche E-selectin blockade by GMI-1271 dampens malignant AML reconstitution/survival potential in vivo when administered as sole agent alone, (2) That E-selectin blockade mediates these effects via dampening a range of intracellular survival/regeneration signalling pathways in the malignant cell, and finally (3) these data suggest E-selectin blockade may synergise with other specific pathway inhibitors to improve treatment outcomes - but only for malignant cells that are appropriately glycosylated to interact with E-selectin.