P2RY2-AKT activation is a therapeutically actionable consequence of XPO1 inhibition in acute myeloid leukemia

Abstract

Selinexor is a first-in-class inhibitor of the nuclear exportin XPO1 that was recently approved by the US Food and Drug Administration for the treatment of multiple myeloma and diffuse large B-cell lymphoma. In relapsed/refractory acute myeloid leukemia (AML), selinexor has shown promising activity, suggesting that selinexor-based combination therapies may have clinical potential. Here, motivated by the hypothesis that selinexor's nuclear sequestration of diverse substrates imposes pleiotropic fitness effects on AML cells, we systematically catalog the pro- and anti-fitness consequences of selinexor treatment. We discover that selinexor activates PI3Kγ-dependent AKT signaling in AML by upregulating the purinergic receptor P2RY2. Inhibiting this axis potentiates the anti-leukemic effects of selinexor in AML cell lines, patient-derived primary cultures and multiple mouse models of AML. In a syngeneic, MLL-AF9-driven mouse model of AML, treatment with selinexor and ipatasertib outperforms both standard-of-care chemotherapy and chemotherapy with selinexor. Together, these findings establish drug-induced P2RY2-AKT signaling as an actionable consequence of XPO1 inhibition in AML.

Extended Data Fig. 1. CRISPR/Cas9 and RPPA analyses reveal signaling pathways modulated by Selinexor treatment

a) Scatterplot depicting replicate selinexor depletion gene scores from CRISPR/Cas9 drug-modifier screen. Screens conducted as n = 2 independent replicates with n = 5 sgRNAs per gene. b) Gene ontology (GO) analysis of selinexor “resister” genes; performed using Enrichr. c) Selinexor depletion gene scores ranked from most depleted to most enriched in the selinexor versus vehicle treated populations. Predicted genetic modifiers of selinexor sensitivity involved in G1/S cell cycle progression are annotated. d) Schematic relating G1/S cell cycle regulators to selinexor depletion gene scores and RPPA expression. Annotated as in Fig. 1d. e) GO analysis of selinexor “sensitizer” genes; performed using Enrichr. f) Schematic relating mTORC1 signaling to selinexor depletion gene scores and RPPA expression. Annotated as in Fig. 1d. g) Immunoblot depicting protein levels of phosphorylated and total S6K1 in five AML cell lines treated with DMSO or selinexor. B-actin included as loading control. Representative immunoblots of n = 2 independent experiments yielding similar results.

Extended Data Fig. 2. Activation of AKT signaling is a specific consequence of XPO1 inhibition

a) Immunoblot depicting protein levels of phosphorylated PRAS40, FOXO3a, and TSC2 following 24-hour treatment of OCI-AML2 cells with selinexor. b) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 following 24-hour treatment of OCI-AML2 and MOLM-13 cells with eltanexor. c) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2 cells treated with a panel of standard-of-care therapies for 24 hours. Representative immunoblots of n = 2-3 independent experiments yielding similar results. B-actin included as loading control.

Extended Data Fig. 3. Gating strategy and Gene Ontology analysis of FACS-based CRISPR/Cas9 screen

a) Scatterplot depicting gating strategy to isolate live sgRNA library transduced OCI-AML2 cells based on forward-scatter (FSC) and side-scatter (SSC). b) Scatterplot depicting gating strategy to isolate singlet sgRNA library transduced OCI-AML2 cells based on FSC. c) Gene ontology (GO) analysis table of scoring genes enriched in the bottom sort (FSGS of > 1.5). P-values calculated by Enrichr using Benjamini-Hochberg correction for multiple hypothesis testing.

Extended Data Fig. 4. Selinexor-induced AKT activation requires the P2RY2 purinergic receptor

a) GSEA plots for gene ontologies enriched in RNA-seq datasets upon selinexor treatment. b) GSEA plots for gene ontologies depleted in RNA-seq datasets upon selinexor treatment. c) Comparison of differential gene expression analysis in OCI-AML2 and MOLM-13 cells treated +/− selinexor. The genes in red are upregulated in both cell lines when treated with selinexor whereas the genes in blue are downregulated in both cell lines. d) Selinexor gene signature representation in AML patients with low versus high P2RY2 expression. ON versus OFF selinexor signatures were assigned for each patient based on the ES z-score > 1 or < −1, respectively. The number of selinexor-ON patients in the P2RY2 high subset versus the P2RY2 low subset was compared; P-values computed using two-sided Fisher’s t-test. e) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2 and MOLM-13 cells treated with 50μM ATP for 48 hours. f) As in (e) but cells were treated with UTP. g) Relative extracellular ATP concentration in OCI-AML2 cells treated with selinexor versus DMSO control for 36 hours. P-values computed using Welch’s unpaired (two-sided) t-tests; data are presented as mean +/− s.e.m. for n = 6 biological replicates. h) Relative expression of P2RY2 in OCI-AML2 cells stably expressing indicated TetOn shRNA constructs following 48 hours of doxycycline treatment. i) Immunoblot depicting protein levels of total and phosphorylated AKT at T308 and S473 in OCI-AML2 cells expressing Cas9 and sgRNAs targeting GFP control or P2RY2 treated with selinexor or DMSO. B-actin included as loading control. j) Tide analysis of OCI-AML2 cells expressing sgP2RY2 to access knockout efficiency. . k) Immunoblot depicting protein levels of total and phosphorylated AKT at T308 and S473 in OCI-AML2 cells overexpressing P2RY2 or empty vector in OCI-AML2 cells. B-actin included as loading control. l) Relative expression of P2RY2 in OCI-AML2 cells stably expressing either empty vector or P2RY2 ORF. Extended Data Figure 4h, l P-values computed using multiple unpaired (two-sided) t-tests; data presented as mean +/−s.e.m. for n = 3 biological replicates. Extended Data Figure 4e,f,i,k Representative immunoblots of n=2-3 biologically independent experiments yielding similar results. B-actin included as loading control.

Extended Data Fig. 5. Isoform specific dependency on PI3Kγ for Selinexor-induced AKT activation

a) Immunoblot depicting protein levels of p110-gamma and p101 in OCI-AML2 cells harboring doxycycline-inducible shRNAs targeting PIK3CG (encoding for p110-gamma) and PIK3R5 (encoding for p101) versus scrambled control. B-actin included as loading control. b) Immunoblot depicting protein levels of phosphorylated AKT at T308 in OCI-AML2 cells harboring doxycycline-inducible shRNAs targeting PIK3CA, PIK3CB and PIK3CD versus scrambled control following treatment with selinexor for 36 hours. B-actin included as loading control. c) Immunoblot depicting protein levels of p110α, p110β and p110δ in OCI-AML2 cells harboring doxycycline-inducible shRNAs targeting PIK3CA, PIK3CB and PIK3CD versus scrambled control. B-actin included as loading control. d) Immunoblot depicting protein levels of total and phosphorylated AKT at T308 and S473 in OCI-AML2, MV4;11 and OCI-AML3 cells treated with BYL-719 (PI3K-alpha inhibitor), TGX-221 (PI3K-beta inhibitor), CAL-101 (PI3K-delta inhibitor) or IPI-549 (PI3K-gamma inhibitor) with or without selinexor for 36 hours. OCI-AML2 cells were treated with 500nM of PI3K inhibitors and MV;411 and OCI-AML3 cells were treated with 100nM of PI3K inhibitors. B-actin included as loading control. e) Immunoblot depicting protein levels of catalytic PI3K isoforms in OCI-AML2 cells treated with selinexor or DMSO. B-actin included as loading control. f) Immunoblot depicting protein levels of phospho-substrates of PKA or PKC in OCI-AML2 cells treated with selinexor or DMSO. B-actin included as loading control. Representative immunoblots of n=2-4 biologically independent experiments yielding similar results. B-actin included as loading control.

Extended Data Fig. 6. The activity and requirement of Ras for complete Selinexor-induced AKT activation

a) Immunoblot depicting active Ras following co-immunoprecipitation of Ras-GTP with GST-Raf1-Ras-binding domain (RBD) fusion proteins in a panel of selinexor-treated AML cell lines. Total Ras in input shown as control. b) Immunoblot depicting active Ras as in (c) in OCI-AML2 cells treated with AR-C 118925XX (2.5μM) and Selinexor (200nM), alone and in combination. Total Ras in input shown as control. c) Immunoblot depicting protein levels of immunoprecipitated GTP-bound Ras in OCI-AML2 cells with doxycycline-inducible shRNAs targeting P2RY2 versus scrambled shRNA control. Cells were exposed to doxycycline (75ng/mL) for 48 hours and treated with either vehicle or selinexor for 36 hours. Total Ras in input shown as control. d) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2 cells co-expressing doxycycline-inducible shRNAs against NRAS and KRAS versus scrambled shRNA control. Cells were exposed to doxycycline (75ng/mL) for 48 hours and treated with either vehicle or selinexor for 36 hours. e) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2 cells expressing doxycycline-inducible shRNAs against NRAS or KRAS. Cells were exposed to doxycycline (75ng/mL) for 48 hours and treated with either vehicle or selinexor for 36 hours. f) Relative 72h selinexor GI₅₀ and dose-response curves in OCI-AML2 cells expressing GFP or Luciferase control or activating constructs of Ras or AKT. P-values computed using one-way ANOVA with Tukey’s method for multiple comparisons; data are presented as mean +/−s.e.m. for n = 3 biological replicates. g) Genetic dependency of AML cell lines on XPO1 as defined by the DepMap dataset. P-values computed using unpaired (two-sided) t-test. Extended Data Figure 6a–e Representative immunoblots of n=2-4 biologically independent experiments yielding similar results. B-actin included as loading control.

Extended Data Fig. 7. The anti-leukemic effect of AKT inhibition combined with Selinexor in cell line models of AML

a) Bliss synergy analysis 2D plots for a panel of AML cell lines treated with a dilution series of selinexor and MK-2206. Delta scores indicate synergy (red) and antagonism (green) across the drug dilution matrix. b) Relative MK-2206 GI₅₀ values in OCI-AML2 cells harboring doxycycline (dox)-inducible shRNAs targeting XPO1 versus scrambled shRNA control. Cells treated with dox for 48 hours prior to incubation with MK-2206 drug-dilution series. Relative MK-2206 GI₅₀ value defined as (GI₅₀ MK-2206 -dox) / (GI₅₀ MK-2206 + dox). c) Time-to-progression assay of OCI-AML2 cells treated with 200nM selinexor, 5μM ipatasertib, or the two drugs in combination. d) Selinexor dose response curves in OCI-AML2 and MOLM-13 cell lines treated with everolimus (50nM or 100nM) in the background. e) Relative selinexor GI₅₀ values in OCI-AML2 cells harboring doxycycline-inducible shRNAs targeting P2RY2 versus scrambled shRNA control. Cells were pre-treated with dox for 48 hours prior to incubation with selinexor drug-dilution series. Relative selinexor sensitivity values defined as (GI₅₀ selinexor-dox) / (GI₅₀ selinexor + dox). . f) Relative selinexor GI₅₀ values in OCI-AML2 cells co-treated with 5μM AR-C or DMSO. g) Relative GI₅₀ values of selinexor in combination with PI3K-α/β/δ/γ-specific inhibitors across a panel of AML cell lines. Relative selinexor GI₅₀ value defined as (GI₅₀ selinexor + PI3K inhibitor) / (GI₅₀ selinexor alone). Background PI3K inhibitors dosed by cell line (OCI-AML2, 1μM; HL-60, 4 μM; MOLM-13, 2μM; MV4;11, 2μM; THP-1, 1μM). h) Immunoblot depicting protein levels of CASPASE 3 and cleaved CASPASE 3 in OCI-AML2 cells treated with selinexor (200nM), ipatasertib (5μM), or the combination. Vinculin shown as control. i) Drug-treated induction of annexin positivity in OCI-AML2 cells, as measured by flow cytometry, relative to baseline annexin positivity elicited with DMSO treatment. j) Immunoblot depicting protein levels of BAX in OCI-AML2 cells expressing hairpins targeting BAX or GFP control. B-actin shown as loading control. k) Bliss synergy landscape for two healthy donor cord blood-derived CD34+ cells treated with selinexor versus ipatasertib across a drug-dilution matrix. Extended Data Figure 7b–g, i data are presented as mean +/− s.e.m. for n = 3 biologically independent replicates. Extended Data Figure 7b,e–h,i P-values computed using multiple unpaired (two-sided) t-tests. Extended Data Figure 7h,j Representative immunoblots of n=2 biologically independent experiments yielding similar results

Extended Data Fig. 8. The effect of Selinexor combined with AKT inhibition on mouse weight and hematopoietic cell compartment

a) Dose-escalation study assessing mouse weight in disease naïve C57BL/6 mice. Mice were treated until Day 4 with indicated treatment regimens (n=3 biologically independent mice). P-values calculated using a two-way ANOVA; data are presented as +/− mean s.d. b) Dose-escalation study assessing white blood cell count in disease naïve C57BL/6 mice. Mice were treated until Day 4 with indicated treatment regimens (n=3 biologically independent mice). P-values calculated using a two-way ANOVA; data are presented as +/− mean s.d. c) Immunoblot depicting protein level of Tp53 in DsRed+ MLL-AF9 cells from mice treated in vivo with Selinexor for indicated duration. Each timepoint representants an independent biologic replicate. Representative immunoblots of n=2 biologically independent experiments yielding similar results. Vinculin shown as loading control. d-n) Flow cytometric measurement of various hematopoietic cell types, each defined by gating strategies labeled on respective y-axes, in C57BL/6 mice treated with vehicle, chemotherapy (1mg/kg doxorubicin and 100mg/kg cytarabine), or the combination of 65mg/kg ipatasertib and 15mg/kg selinexor. Data are presented as mean +/− s.d. for n = 4 biologically independent replicates. Where indicated, P-values computed using unpaired (two-sided) t-test

Extended Data Fig. 9. Analysis of engraftment, tolerability, and efficacy of Selinexor combined with AKT inhibition in mouse models of AML

a) Confirmation of PDX engraftment by measurement of circulating human CD45+ cells transplanted into NOG-EXL mice (n=6 biologically independent mice per group); data are presented as mean +/− s.d. b) Bliss synergy analysis 2D plots for PDX model of AML treated with a dilution series of selinexor and ipatasertib. Delta scores indicate synergy (red) and antagonism (green) across drug dilution matrix. c) Weight of C57BL/6 mice treated with vehicle or selinexor (15mg/kg) plus ipatasertib (65mg/kg). d) Weight of C57BL/6 mice treated with vehicle or chemotherapy (doxorubicin 1mg/kg, cytarabine 100mg/kg). e) Weight of C57BL/6 mice treated with vehicle or selinexor (15mg/kg) plus chemotherapy (doxorubicin 1mg/kg, cytarabine 100mg/kg). f) Weight of C57BL/6 mice treated with vehicle or selinexor (7.5mg/kg) plus chemotherapy (doxorubicin 1mg/kg, cytarabine 100mg/kg). g) Measurement by flow cytometry of the proportion of DsRed+ MLL-AF9 cells taken from bone marrows (n=5 biologically independent mice per group) two days following treatment with vehicle, 15mg/kg selinexor plus 65mg/kg ipatasertib, chemotherapy (1mg/kg doxorubicin and 100mg/kg cytarabine), or 7.5mg/kg selinexor plus chemotherapy; P-values calculated using unpaired Mann-Whitney test; data are presented as mean +/− s.d. h) Measurement by flow cytometry of DsRed+ MLL-AF9 cells taken from bone marrows at the point of relapse observed in chemotherapy-treated mice (n=5 biologically independent mice per group except in the chemotherapy-treated subgroup in which n=4). P-values calculated using unpaired Mann-Whitney test; data are presented as mean +/− s.d. i) Selinexor dose response in OCI-AML2 and MOLM-13 cells treated with cytarabine (50nM) or daunorubicin (5nM) in the background; data are presented as mean +/− s.e.m. for n = 3 biologically independent experiments. Extended Data Figure 9c–f Dosing schedule shown at bottom left. (D) = mouse demise.

Extended Data Fig. 10. Efficacy of Selinexor combined with AKT inhibition on leukemia initiating cells

a) Leukemic burden of mice (n=3 biologically independent mice per group) prior to sorting and engraftment for extreme limiting dilution assay. P-values calculated using unpaired two-sided Welch’s test; data are presented as mean +/−s .d. b) FACS plots depicting gating strategy to isolate dsRed+ MLL-AF9 cells prior to engraftment for extreme limiting-dilution assay. c) Kaplan-Meier curves showing overall survival of secondary recipient mice (n=5 biologically independent mice) from each group in limiting dilution assay for determining LIC frequency. Statistical significance determined by log-rank (Mantel-Cox) test. n.s, not significant.

Figure 1. Orthogonal Functional Genomic and Proteomic Analyses Nominate AKT Activation as an Actionable Consequence of Selinexor Treatment.

a) Experimental strategy for parallel assessment of cell-beneficial and cell-detrimental effects of nuclear export inhibition with the XPO1 inhibitor, selinexor. Pooled CRISPR-Cas9 screening in OCI-AML2 cells treated with selinexor reveals genetic modifiers of drug sensitivity. Reverse phase protein array (RPPA) analysis in selinexor treated OCI-AML2 cells reveals drug-responsive protein and phosho-protein expression. b) Selinexor depletion gene scores ranked from most depleted to most enriched in the selinexor versus vehicle treated populations. Scoring genes in the PI3K/AKT pathway are annotated as sensitizers (orange, depleted in the selinexor population) or resisters (blue, enriched in the selinexor population). Screens conducted as n = 2 independent replicates with n = 5 sgRNAs per gene. c) Volcano plot depicting differential expression for 160 RPPA probes following 48 hours of selinexor treatment relative to statistical significance in dataset. Annotated probes comprise pathways with decreased expression (blue) or increased expression (orange). RPPA expression analysis conducted as n = 3 independent experiments. P-values computed by multiple unpaired (two-sided) t-tests. d) Schematic relating PI3K/AKT pathway members to selinexor depletion gene scores and RPPA expression. Genes scoring as selinexor sensitizers are shaded in orange; genes scoring as selinexor resisters are shaded in blue; genes absent from library not shaded. Phosphorylated proteins with selinexor-induced increased (orange) RPPA expression are indicated.

Figure 2. Selinexor Activates PI3K/AKT Signaling in AML cells

a) Immunoblot depicting protein levels of phosphorylated and total PI3K/AKT pathway members following treatment of a panel of AML cell lines with selinexor. Selinexor was dosed at the following concentrations for each cell line: OCI-AML2 (200nM), MOLM-13 (75nM), MV;411 (50nM), HL-60 (300nM), OCI-AML3 (250nM). b) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2, OCI-AML3, and MOLM-13 cells treated with Selinexor for indicated duration. c) Immunoblot depicting protein levels of XPO1 and total and phosphorylated AKT at T308 in OCI-AML2 cells with doxycycline (dox) inducible shRNAs targeting XPO1 versus scrambled shRNA control. Cells were exposed to dox (75ng/mL) for indicated durations prior to collection. d) Immunoblot depicting protein levels of XPO1, total and phosphorylated AKT at T308 and S473 in OCI-AML2 cells with sgRNAs targeting XPO1 versus non-targeting controls. Representative immunoblots of n=3-5 biologically independent experiments yielding similar results. B-actin included as loading control.

Figure 3. FACS-Based CRISPR/Cas9 Screening Identifies Genetic Determinants of Selinexor-Induced AKT Activation.

a) FACS-based CRISPR/Cas9 screening strategy to identify genetic modifiers of AKT phosphorylation in selinexor-treated AML cells. sgRNA library transduced OCI-AML2 cells were treated with selinexor for 48 hours, fixed/permeabilized and stained with phosphorylated AKT T308 primary antibody followed by Alexa Flour 488 conjugated secondary antibody. Stained cells were then sorted according to phosphorylated AKT T308 expression into high-expressing cells (top sort) and low-expressing cells (bottom sort). Genomic DNA was extracted and sgRNA barcodes were amplified and indexed prior to deep sequencing. The FACS screen gene score (FSGS) enumerates genes whose sgRNA representatives were enriched in the top or bottom sorted populations. b) Histogram depicting distribution of phosphorylated AKT T308 expression in parental OCI-AML2 cells treated with vehicle or selinexor for 48 hours. c) Histogram depicting distribution of phosphorylated AKT T308 expression in sgRNA library transduced OCI-AML2 cells treated with vehicle or selinexor for 48 hours. Gates defining top (blue) and bottom (orange) sorted population in selinexor treated cells are indicated. d) Scatterplot depicting replicate FSGS values. Scoring genes enriched in the bottom sort (orange) or the top sort (blue) are annotated. LacZ, EGFP and luciferase targeting controls indicated in white. Screens conducted as n = 2 independent replicates with n = 4 sgRNAs per gene. e) Gene ontology (GO) analysis of scoring genes enriched in the bottom sort with a p-value < 5e-4. GO performed using Enrichr.

Figure 4.. PI3Kγ Promotes AKT Activation Upon Selinexor-Induced P2RY2 Upregulation.

a) Scatterplot depicting replicate-averaged FSGS compared to differential gene expression (DE) score with selinexor treatment. RNA-seq analysis of selinexor versus vehicle treated OCI-AML2 and MOLM-13 cells yielded DE for each cell line. Plotted DE score is the average DE across the two cell lines. Orange shaded region denotes genes that were both transcriptionally upregulated by selinexor treatment and yielded a high FSGS score in the FACS-based CRISPR/Cas9 screen. RNA-seq conducted as n = 3 biologically independent replicates. b) Relative expression of P2RY2 across a panel of selinexor-treated AML cell lines compared to DMSO control. P-values computed using multiple unpaired (two-sided) t-tests. Data are presented as mean +/−s.e.m. for n = 3 biologically independent replicates. c) Immunoblot depicting protein levels of total and phosphorylated AKT at T308 and S473 in OCI-AML2 cells with doxycycline-inducible shRNAs targeting P2RY2 versus scrambled shRNA control. Cells were exposed to doxycycline (75ng/mL) for 48 hours and treated with either vehicle or selinexor for 36 hours. d) Immunoblot depicting protein levels of phosphorylated AKT at T308 and S473 in OCI-AML2, MOLM-13, and MV4;11 cells following treatment of pertussis toxin (Ptx 100ng/mL), AR-C 118925XX (AR-C 2.5μM) or selinexor alone and in combination for 36 hours. e) Replicate-averaged FSGS ranked from most depleted to most enriched in bottom sort. Isoforms of PI3K are annotated. f) Selinexor depletion gene scores ranked from most depleted to most enriched in the selinexor versus vehicle treated populations. Isoforms of PI3K are annotated. g) Immunoblot depicting protein levels of total and phosphorylated AKT at T308 and S473 in OCI-AML2 cells with doxycycline-inducible shRNAs against PIK3CG and PIK3R5 versus scrambled shRNA control. Cells were exposed to doxycycline (75ng/mL) for 48 hours and treated with either vehicle or 200nM selinexor for 36 hours. Figure 4c,d,g Representative immunoblots of n=2-3 biologically independent experiments yielding similar results. B-actin included as loading control.

Figure 5. Inhibition of AKT Potentiates the Anti-Leukemic Effect of Selinexor.

a) Relative GI₅₀ values (GI₅₀ selinexor + AKT inhibitor / GI₅₀ selinexor alone) of selinexor in combination with AKT inhibitors across a panel of AML cell lines. Background AKT inhibitors dosed by cell line (OCI-AML2, 5μM; MOLM-13-3μM; MV4;11-3μM; HL-60-5μM; OCI-AML3-3μM; Kasumi-1-3μM; U937-3μM; THP-1,-5μM). b) Time-to-progression assay of OCI-AML2 cells treated with 200nM selinexor, 5μM MK-2206, or the two drugs in combination. Data are presented as mean for n = 3 biologically independent cell populations. c) Immunoblot depicting protein levels of cleaved CASPASE 3 and cleaved PARP across a panel of AML cell lines treated with selinexor (OCI-AML2, 200nM; MOLM-13, 75nM; MV4;11, 50nM; OCI-AML3, 250nM; HL-60, 300nM), MK-2206 dosed as in (a) or the combination. d) Percentage of cells staining annexin V+ across a panel of AML cell lines treated with selinexor, MK-2206 or the combination; relative to DMSO control. P-values computed using one-way ANOVA with Tukey’s method for multiple comparisons; data are presented as mean +/−s.e.m. for n = 3 biological replicates. e) Absolute selinexor GI₅₀ values in OCI-AML2 cells harboring shRNAs targeting BAX (shBAX_71) versus GFP control and treated with selinexor + DMSO or selinexor + MK-2206. f) Immunoblot depicting protein levels of cleaved PARP and phospho-BAD S136 in OCI-AML2 cells treated with selinexor (200nM), MK-2206 (5μM), or the combination. g) Bliss synergy scores for 32 primary patient samples treated with a drug-dilution matrix. Bliss values ≤ −5, > −5 and < 5, and ≥ 5 denote antagonism, additivity, and synergy, respectively. h) Methylcellulose colony formation of primary AML patient samples treated with selinexor (#11: 5nM, #17: 5nM) and/or ipatasertib (#11: 0.75μM, #17: 5μM) as indicated. i) Methylcellulose colony formation of normal cord-blood-derived CD34+ cells treated with selinexor (5nM) and/or ipatasertib (Sample #1-2: 0.75μM, Sample #3-4: 5μM) as indicated. Figure 5 a,e P-values computed using multiple unpaired (two-sided) t-tests; data are presented as mean +/−s.e.m. for n = 3 biologically independent replicates. Figure h,i 5 p-values calculated using nonparametric Mann-Whitney test; data are presented as mean +/− s.d. for n = 5 biologically independent replicates. Colonies were stained and counted at 14 days. Figure 5 c,f Representative immunoblots of n=2-3 biologically independent experiments yielding similar results. B-actin included as loading control.

Figure 6. Preclinical Efficacy of Combined XPO1 and AKT Inhibition in AML.

a, c-d) Kaplan–Meier survival curves of MLL-AF9 syngeneic (a), OCI-AML2 cell-line NSG xenograft (c) or patient-derived AML-engrafted NOG-EXL (d) mouse models treated with vehicle, 65mg/kg ipatasertib, 15mg/kg selinexor or the drug combination; n = 5 (a,d),n=10 (c), mice per cohort. b) FACS quantification of MLL-AF9 dsRed+ leukemic blast cells mouse bone marrow aspirates (n=5 biologically independent mice per group) following treatment with conditions indicated in (a). e) FACS quantification of human CD45⁺ leukemic blast cells from NOG-EXL mouse bone marrow aspirates (n=6 biologically independent mice per group) on day 28. Human CD45+ cells were injected on day 0; engraftment was confirmed on day 12; mice were treated day 13 through day 21 f) Kaplan–Meier survival curves of MLL-AF9 syngeneic mouse model of AML treated with vehicle, standard-of-care chemotherapy (1mg/kg doxorubicin and 100mg/kg cytarabine) or the combination of 65mg/kg ipatasertib plus 15mg/kg selinexor; n = 8 mice per cohort. g) FACS quantification of MLL-AF9 dsRed+ leukemic blast cells from mouse bone marrow aspirates following treatment with conditions indicated in (f); n = 3 biologically independent mice. h) FACS quantification of MLL-AF9 dsRed+ leukemic blast cells in spleen upon disease relapse following standard-of-care chemotherapy (1mg/kg doxorubicin and 100mg/kg cytarabine) versus the 65mg/kg ipatasertib plus 15mg/kg selinexor drug combination; n = 4 biologically independent mice in control and n= 3 selinexor + ipatasertib. i) FACS quantification of MLL-AF9 dsRed+ leukemic blast cells in bone marrow with conditions indicated in (h); n = 4 biologically independent mice in control and n= 3 selinexor + ipatasertib. j) Kaplan–Meier survival curves of MLL-AF9 syngeneic mouse model of AML treated with vehicle, standard-of-care chemotherapy (1mg/kg doxorubicin and 100mg/kg cytarabine), the combination of chemotherapy plus 7.5mg/kg selinexor, or the combination of 65mg/kg ipatasertib plus 15mg/kg Selinexor conducted with n = 5 mice per cohort. k) Limiting dilution assay performed on MLL-AF9 cells isolated from primary mice treated with either standard chemotherapy (100mg/kg cytarabine and 1mg/kg doxorubicin) or the combination of 15mg/kg selinexor and 65mg/kg ipatasertib for 24hrs and reinjected into secondary recipient mice with n=5 biologically independent replicates per cohort. l) Determination of leukemia-initiating cell (LIC) frequency with a 95% confidence interval in each group using extreme limiting dilution analysis (ELDA). Statistical significance determined by a two-sided chi-squared test. Figure 6 b,e, g–i Data are presented +/− s.d. P-values calculated using nonparametric Mann-Whitney test (b, e) or two-tailed Welch’s t-test (g-i). Figure 6 a,d,f,j Statistical significance determined by log-rank (Mantel-Cox) test. Duration treatments conditions depicted as colored bars.