Cotargeting of XPO1 Enhances the Antileukemic Activity of Midostaurin and Gilteritinib in Acute Myeloid Leukemia

Abstract

Acute myeloid leukemia (AML) is a hematopoietic stem-cell-derived leukemia with often successive derived driver mutations. Late onset acquisition of internal tandem duplication in FLT3 (FLT3-ITD) at a high variant allele frequency often contributes to full transformation to a highly proliferative, rapidly progressive disease with poor outcome. The FLT3-ITD mutation is targetable with approved FLT3 small molecule inhibitors, including midostaurin and gilteritinib. However, outside of patients receiving allogeneic transplant, most patients fail to respond or relapse, suggesting alternative approaches of therapy will be required. We employed genome-wide pooled CRISPR knockout screening as a method for large-scale identification of targets whose knockout produces a phenotypic effect that enhances the antitumor properties of FLT3 inhibitors. Among the candidate targets we identified the effect of XPO1 knockout to be synergistic with midostaurin treatment. Next, we validated the genetic finding with pharmacologic combination of the slowly reversible XPO1 inhibitor selinexor with midostaurin and gilteritinib in FLT3-ITD AML cell lines and primary patient samples. Lastly, we demonstrated improved survival with either combination therapy compared to its monotherapy components in an aggressive AML murine model, supporting further evaluation and rapid clinical translation of this combination strategy.

Keywords: AML; CRISPR-Cas9 screening; FLT3; XPO1; synergism.

Figure 1

Midostaurin CRISPR knockout screen. (a) Screen design. (b) Volcano scatter plot showing significant positive and negative selection results. (c) Changes in levels of the two most efficient sgRNAs targeting XPO1 in four replicate screens. (d) Enriched gene pathway analysis. (e) Schematic of nuclear pore complex (NPC) components with log-fold change (LFC) values from the screen indicated.

Figure 2

Genetic validation of XPO1 as a cotarget with FLT3. (a) Design of three guides targeting XPO1 used to knock down XPO1 expression in MOLM-13 cells. (b) Immunoblot analysis showing protein expression of XPO1 in MOLM-13 cells, XPO1 knockdown (KD) cell lines, and HG3 (positive control). (c) MOLM-13 parent cells and XPO1 knockdown cells were treated with 10 nM midostaurin or 8 nM gilteritinib for 48 h and proliferation changes compared. For each knockdown experiment, conditions were compared with each clone separately and then the results pooled together across clones to get an average effect (p < 0.001 where noted).

Figure 3

In vitro pharmacologic validation of XPO1 as a cotarget with FLT3. (a) MOLM-13 or MV4-11 cells were treated with a range of doses of midostaurin or gilteritinib plus selinexor for 48 h and proliferation changes measured. Regions of synergy were determined using highest single agent analysis shown here; mathematical synergy was calculated in Figure S3. (b) Similar proliferation and synergy analysis was applied to primary FLT3-ITD AML patient samples, AML1 and AML2, which were cocultured with HS5 stromal cells for 96 h. For each drug (n = 2 patient samples), blast cells were separated from stroma prior to development with MTS and results averaged prior to analysis against a highest single agent model shown here; mathematical synergy was calculated in Figure S4.

Figure 4

In vivo pharmacologic validation of XPO1 as a cotarget with FLT3. (a) Design of murine experiment where NSG mice were engrafted with MOLM-13 cells expressing luciferase and treated with vehicle, 50 mg/kg midostaurin daily, 30 mg/kg gilteritinib daily, 15 mg/kg selinexor twice weekly, or combinations of midostaurin or gilteritinib plus selinexor. (b) IVIS imaging of two mice per group and Kaplan-Meier analysis of survival of the entire cohort. (c) Estimated mean survival for each group with the 95% confidence interval (CI). (d) Representative images of spleens from each group ex vivo. (e) Differences in spleen weights between groups were estimated using analysis of variance (ANOVA) methods. p-values were adjusted for multiple comparisons within each drug combination using Holm’s procedure.