Selective Inhibition of Nuclear Cytoplasmic Transport As a New Treatment Paradigm in Myelofibrosis

Myelofibrosis (MF) is a hematopoietic stem cell neoplasm characterized by constitutive activation of JAK/STAT signaling due to mutations in JAK2, calreticulin or MPL. Many MF patients suffer from severe constitutional symptoms and have reduced life expectancy due to cytopenias, progression to acute myeloid leukemia and thromboembolic events. JAK kinase inhibitors such as ruxolitinib (RUX) reduce MF symptoms, but like all other drugs used in MF, are not curative, with persistence of mutant cells and prompt symptom rebound upon discontinuation. This defines a clinical need to identify strategies capable of inducing more profound and durable responses in MF.

To identify previously unrecognized molecular vulnerabilities in MF, we infected HEL cells (homozygous for JAK2^V617F) with a barcoded lentiviral shRNA library targeting ~5,000 human signal transduction genes, with 5-6 shRNAs/gene (Cellecta Human Module 1). Conditions were optimized to achieve a multitude of infection (MOI) of ~1. Barcode abundance was compared between days 0 and 9 after infection by next generation sequencing. Candidates were selected based on ≥ 15-fold reduction of abundance by ≥ 2 shRNAs targeting the same gene, similar to Khorashad et al. [Blood. 2015;125(11):1772-81]. Amongst the genes meeting these criteria, nuclear cytoplasmic transport (NCT) was significantly enriched, with RAN and RANBP2 amongst the top genes, suggesting that HEL cells may be highly dependent on NCT. For confirmation, HEL cells were stably transduced with doxycycline (DOX)-inducible shRAN. After 72 hours DOX-induced knockdown of RAN reduced viable cells by 77.3±5.5% and colony formation by 82.8±1.3% and dramatically increased apoptosis (uninduced: ~10% vs. induced: ~50%). Similar results were observed in SET-2 cells (heterozygous for JAK2^V617F). We next cultured HEL and SET-2 cells with graded concentrations of the KPT-330 (selinexor, Karyopharm), an inhibitor of CRM-1, the core component of NCT, or RUX as a comparison. Selinexor was five-fold more potent than RUX against HEL cells (IC₅₀: 98nM for KPT vs. 536 nM for RUX) and as potent as RUX in SET-2 cells (IC₅₀:~100 nM). Importantly, RUX-resistant HEL cells (IC₅₀:24µM) were highly sensitive to inhibition of NCT by knockdown of RAN or selinexor (IC₅₀:160nM). Selinexor also selectively inhibited colony formation by primary MF vs. cord blood (CB) CD34⁺ cells (IC₅₀:93nM for MF vs. 203nM for CB). Lastly, selinexor enhanced RUX-induced growth inhibition and apoptosis in primary MF CD34⁺cells cultured ex vivo for 72h (including both JAK2 mutation positive and negative MF samples, n=3 for each, and RUX resistant patient samples, n=6). Nuclear:cytoplasmic fractionation of HEL cells revealed that the expression and nuclear localization of the tumor suppressors FoxO3A and APC, but not of PP2A and nucleophosmin (NPM) were significantly increased upon knockdown of RAN, which may contribute to the increased apoptosis following NCT inhibition.

To determine the in vivo effects of selinexor in MF, we induced MPN in Balb/c mice by transplanting donor marrow infected with JAK2^V617F for three weeks, and then treated mice (n=13/group) with vehicle, selinexor (initial dose 20 mg/kg, 3x weekly, orally) or RUX (initial dose 50 mg/kg twice daily, orally) or the combination of RUX plus selinexor for up to 4 weeks. Combination treatment significantly reduced white blood cell counts and normalized spleen size. Compared to vehicle, selinexor alone significantly reduced GFP⁺cells in the spleen, and this effect was further enhanced with the combination treatment. Histopathology revealed that combination treatment restored splenic architecture, while bone marrow fibrosis was not significantly altered by selinexor or the combination. Mice in all groups, including the combined vehicle controls, experienced considerable weight loss, suggesting that toxicity may be partially due to high dose and frequent drug administration. Experiments with the next generation NCT inhibitor KPT-8602 [Etchin et al., Leukemia, 2016 Jun 24] are underway.

In summary, our results suggest that MF cells are exquisitely dependent on NCT, and that NCT inhibition alone or in combination with RUX may reduce JAK2^V617F allelic burden. This identifies NCT as a prime therapeutic target in MF. A phase I clinical trial of selinexor in refractory MF is in preparation.