Identification of Specific Hnrnps As Novel Therapeutic Targets and Responsive Indicators of KPT330 (selinexor) in Leukemia

Background: Activenuclear-cytoplasmic shuttling of proteins and RNAs, such as heterogeneous ribonucleoproteins (hnRNPs), is essential for the normal function and survival of eukaryotic cells and tumorigenesis (Dreyfuss et al. 1993 Annu Rev Biochem 62, 289; Gorlich and Mattaj 1996 Science 271, 1513). Up-regulation of exportin 1 (XPO1)/chromosomal maintenance 1 (CRM1), a member of the karyopherin-β family of nuclear export receptor proteins, has been implicated in solid and hematologic malignancies (Kau Kau et al. 2004).Selinexor (KPT-330) has been shown to be able block in vitro and in vivo XPO1/CRM1 functions and is currently in phase-II/IIb clinical trials for treatment of hematologic and solid tumors (Senapedis et al., 2014 Nat Rev Cancer 4, 106). However, the mechanisms underlying the selectivity and efficacy of selinexor are incompletely understood, and no biomarkers are currently available to predict clinical responses to selinexor in clinical settings. In this study, we focus on determining the effects of selinexor on the nuclear-cytoplasmic shuttling of hnRNPs, particularly hnRNPK and hnRNPA1, to elucidate the roles of the hnRNPs in the regulation of selectivity and efficacy of selinexor in myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML).

Method:We performed growth inhibition/killing assays, histopathologic evaluations, immunohistochemical studies, subcellular fraction western blotting, super-resolution stimulated emission depletion (STED) confocal microcopy and siRNA knockdown experiments.

Results: Our in vitro experiments demonstrate a marked increase in XPO1/CRM1 protein and decrease in TP53 in our azacitidine-resistant MDS/AML cell lines compared to our azacitidine-sensitive MDS/AML cell lines. Selinexor treatment efficiently blocks export of hnRNP K from nuclei and increased nuclear accumulation of hnRNPK and inhibits MDS/AML cell growth, while the protein levels of XPO1/CRM1 and TP53 remain unchanged. Our experiments using clinical bone marrow specimens show no significant difference in the total protein level or nuclear accumulation of XPO1/CRM1 between the normal control and MDS or AML bone marrow specimens. In contrast, a strong positive correlation between MDS/AML disease progression and hnRNPK protein accumulation is observed in those clinical specimens. We have extended our experiments to clinical bone marrow specimens from a small cohort in a clinical trial for selinexor in AML at the University of Chicago (NCT02573363). In our small cohort, 5 patients responded to selinexor, 4 patients did not respond and 1 had a partial response. All 5 responders show a striking decrease in their bone marrow blast percentage from their pre-treatment marrows (average blast percentage 37.4%) to their post-treatment (average blast percentage 1.8%). Non-responders show no such difference in pre and post-treatment blast percentage (56.3 and 57.1%, respectively). Importantly, our experiments demonstrate a marked difference in the protein accumulation and subcellular localization of hnRNPK and hnRNPA1, another member of the hnRNP family, between selinexor-responder and selinexor-non-responder bone marrow specimens. Specifically, selinexor responders had much higher levels of hnRNPK and hnRNPA1 proteins in their pre-treatment bone marrows than non-responders, despite the fact that the latter had higher bone marrow blast percentages on average. There is markedly reduced accumulation of hnRNPK and hnRNPA1 in the post-selinexor treatment bone marrow specimens from the responders, but not the non-responders, suggesting these hnRNPs as key therapeutic targets for selinexor in MDS and AML. In contrast, no significant change in XPO1/CRM1 protein levels is observed in the selinexor-responder vs. selinexor-non-responder bone marrow specimens.

Conclusion:Our data have revealed a novel drug-action mechanism by which selinexor impairs the nuclear-cytoplasmic shuttling of hnRNPK and hnRNPA1 in MDS and AML cells. Differential expression and localization of these hnRNPs in normal vs. MDS vs. AML cells may provide the rationale for the preferential killing of leukemia cells by selinexor. Our data also suggest the possibility to develop novel hnRNP-based biomarkers to predict the response to selinexor in clinical settings.