Vascular E-Selectin Protects Leukemia Cells from Chemotherapy By Directly Activating Pro-Survival NF-Kb Signalling - Therapeutic Blockade of E-Selectin Dampens NF-Kb Activation

The vascular adhesion molecule E-selectin is a key component of the Bone Marrow (BM) Haematopoietic Stem Cell (HSC) niche prompting HSC to proliferate at the expense of self-renewal (Winkler, Nat Med 2012).Only 3 - 5% of BM endothelial cells express E-selectin in steady state however E-selectin is greatly upregulated (5 - 10 fold) in BM of mice with acute myeloid leukemia (AML) raising the question; how do AML stem cells (LSC) respond to E-selectin at the vascular niche & does E-selectin signalling in AML & HSC differ?

Using models of murine AML generated by retroviral transduction of MLL-AF9 or AML1-ETO oncogenes, we found leukemic blasts rapidly upregulate E-selectin-binding upon oncogenic transformation. Furthermore E-selectin adhesion promoted LSC survival to cytarabine in vitro as well as in vivo. LSC survival to chemotherapy in wildtype compared to E-selectin knockout (E-/-) mice quantified by rigorous limiting-dilution transplantation assay of 1%, 0.1%, 0.01% femur BM demonstrated that E-selectin deletion increased sensitivity of LSC to high-dose cytarabine therapy ~11-fold (900mg/kg cytarabine n=6 donors &15 recipients/gp p=0.0037). Thus E-selectin is a critical vascular niche component mediating LSC chemo resistance. Importantly these findings could be replicated by administration of a potent small molecule glycomimetic E-selectin antagonist (GMI-1271) to wt mice. Furthermore treatment with GMI-1271 (40mg/kg bidaily) for 10 days in combination with standard mouse version of 7+3 induction chemotherapy (5 days cytarabine 100mg/kg; 3 days doxorubicin 1mg/kg) was able to significantly double mouse survival over chemotherapy alone (p=0.0054; no chemotherapy median survival 25 d, AraC/Dox alone 32 d, AraC/Dox plus GMI-1271 survival 41 d; n=8 mice/gp).

To understand mechanisms of this chemo-sensitisation, mice with MLL-AF9 monomyelocytic (11q23 translocation) or AML1-ETO granulocytic t(8;21) -induced AML were administered GMI-1271 or vehicle control for 5 days before sorting BM AML blasts for RNA sequencing. Analysis of differentially expressed transcripts by CuffDiff / DSeq2 revealed 170 RNAs differed following in vivo E-selectin blockade. KEGG pathway analysis indicated a pathway potentially dampened in AML blasts following GMI-1271 administration was PI3K - NF-kB signalling - raising the hypothesis that adhesion to E-selectin activates pro-survival NF-kB signalling in AML cells leading to enhanced chemoresistance.

Using two in vitro assays, we confirmed E-selectin to be unique among vascular adhesion molecules tested in being able to directly activate NF-kB. Activation of NF-kB was only observed upon E-selectin mediated adhesion & was not observed following adhesion to P-selectin, PECAM-1 or VCAM-1 using either myeloid NF-kB GFP reporter cell lines or by induction of p65 NF-kB (Ser 536). Importantly E-selectin mediated NF-kB activation was completely inhibited when E-selectin antagonist GMI-1271 added.

Assays repeated in presence of a specific NF-kB activation antagonist (BMS-345541 10uM 24hrs) demonstrated that NF-kB blockade alone reversed E-selectin-mediated chemoresistance in vitro. Upstream blockade of E-selectin by GMI-1271 not only inhibits NF-kB activation but also mobilizes LSC out of the protective BM niche & prevents re-entry thereby breaking the chemo resistance observed with these cells. A Phase I/II Clinical trial to study efficacy of GMI-1271 in combination with chemotherapy in AML patients (NCT02306291) is currently in progress