Functional role and therapeutic targeting of p21-activated kinase 4 in multiple myeloma

Multiple myeloma (MM) is a hematological malignancy characterized by proliferation of clonal plasma cells in the bone marrow (BM).¹  The introduction of novel agents including proteasome inhibitors and immunomodulatory agents alone or in combination has improved outcomes of MM patients.²  However, patients still relapse and ultimately succumb to this disease, providing the impetus to develop novel therapeutic modalities.³  Delineation of signaling pathways mediating MM cell growth, survival, and migration within the BM microenvironment can both enhance our understanding of disease pathogenesis and identify molecular targets for novel MM therapies.

The p21-activated kinase (PAK) family of serine/threonine kinases (STKs) comprises 6 mammalian proteins that are classified into group I (PAK1-3) and group II (PAK4-6) based on structural homology and regulatory function.⁴  Constitutive activation of PAK1 and 2, positively correlated with increased cell migration potential, has been demonstrated in myeloma cells. We here report high expression of total and phosphorylated (active) PAK4 in the majority of myeloma cell lines, and in all cases of asymptomatic and symptomatic myelomas tested.

As a key downstream effector of the K-Ras pathway and of the ρ-family of GTPases (ρ, Rac, and Cdc42), PAK4 is implicated in a number of intracellular processes, including cytoskeleton reorganization,⁵  embryonic development,⁶  as well as cell proliferation, survival, and motility.⁷  PAK4 is ubiquitously expressed at low levels in many tissues, including BM, and has been found to be overexpressed, genetically amplified, and/or point mutated in several cancer types.^8-16  In athymic mice, overexpression or constitutively active form of PAK4 leads to tumor formation, whereas its depletion inhibits tumorigenesis.⁹  Depletion of PAK4 negatively impacted the activation of NF-қB, extracellular signal-regulated kinase (ERK), and JNK pathways,¹⁷  while activating the ATM/Chk1/2/p53 pathway.¹⁸  Interestingly, PAK4 may also play a role in gene transcription pathways due to its ability to continuously cycle between the nucleus and the cytoplasm, allowing the modulation of nucleo-cyto trafficking of β-catenin.¹⁹ 

The relative high expression of PAK4 in myeloma and its involvement in major signaling pathways in cancer such as Ras, NF-κB, and Wnt/β-catenin suggests a possible role of PAK4 in myeloma pathogenesis. We here characterized growth and survival activity of PAK4 in myeloma cells and report the therapeutic potential of a novel PAK4 allosteric modulator (PAM).

Bone marrow mononuclear cells and primary MM cells were isolated using Ficoll-Hypaque density gradient sedimentation from BM aspirates MM patients following informed consent and institutional review board (Dana-Farber Cancer Institute) approval. The human myeloma cell lines (HMMCLs) were cultured in Roswell Park Memorial Institute 1640 medium (RPMI 1640; Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum.

Compounds were dissolved in dimethyl sulfoxide (DMSO) unless otherwise stated.

MM cell proliferation was measured by (³H)-thymidine (Perkin-Elmer, Boston, MA) incorporation assay, as previously described.²⁰  Cell viability was analyzed by CellTiter Glo (CTG; Promega). Study of caspases activity was performed using caspases 3-7, caspase 8, and caspase 9 Glo assay (Promega). Apoptosis was evaluated by flow cytometric analysis following Annexin V staining. Exvitech automated flow cytometry platform (Vivia Biotech, Madrid, Spain) was used to evaluate activity of KPT-9274 against primary myeloma cells in their microenvironment, as previously described.²¹  Briefly, BM was diluted with RPMI 1640 to seed 400 to 8000 live cells per well into 96-well plates previously prepared with increasing concentration of KPT-9274 (1 nM-10 µM) and DMSO (up to 0.5%) as vehicle and were incubated for 24 to 72 hours. Then, red cells were lysed with ammonium chloride lysis solution (20 mM KHCO₃, 310 mM NH₄Cl, 254 µM EDTA). The multiparametric flow cytometry was performed in the ExviTech platform using annexin V and CD138 monoclonal antibody (mAb; Becton Dickinson, San Jose, CA) to identify viable myeloma cells.

Western blotting (WB) was performed to delineate expression levels of total protein and phospho-specific isoforms using following antibodies: total PAK4 (Abcam 19007), p-PAK4 (Ser 474) (Santa Cruz Biotechnology), p-CREB (Ser 133) (Cell Signal), pERK (Thr202/Tyr204) (Cell Signal), pAKT (Ser 473) (Cell Signal), pMEK1/2 (Ser217/221) (Cell Signal). Glyceraldehyde-3-phosphatedehydrogenase (GAPDH) was used as loading control (Santa Cruz Biotechnology).

Immunohistochemical staining of normal and tumor tissue specimen sections of formalin-fixed, paraffin-embedded BM biopsies was prepared and processed for immunohistochemistry to detect PAK4 and p-PAK4 protein expression by using specific antibodies (Abcam, ab62509, and Santa Cruz, sc-135775, respectively).

The filter arrayed with antibodies (Hypromatrix) was incubated with the lysate of RPMI 8226 cells transfected with green fluorescent protein (GFP)–Pak4 expression plasmid. The antibody-antigen-Pak4 complex was detected by anti-GFP antibody conjugated with horseradish peroxidase.

Human TRIPZ PAK4 short hairpin RNA (shRNA) vectors were purchased from Thermo Scientific Bio (Tewksbury, MA). shRNA expression was induced adding 2.5 µg/mL doxycycline to the culturing media. The efficacy of the induction was confirmed by examining the cells microscopically for the presence of TurboRFP and by quantitative polymerase chain reaction (qPCR) after 72 hours of induction. Functional studies were performed as described above.

• Clone #23: Clone ID, V3THS_395103

• Clone #38: Clone ID, V3THS_335559

• Clone #84: Clone ID, V3THS_395101

• Clone #87: Clone ID, V3THS_395104

LentiORF clone of human PAK4 GFP tagged was purchased from Thermo Scientific Bio (PLOHS_100008678). MM cells were transduced in polybrene media (final concentration, 8 µg/mL) for 8 hours and selected by sorting GFP-positive cells.

Expression of human PAK4 transcript was determined using PAK4 TaqMan Gene Expression Assay, following manufacturer protocols (Applied Biosystems, Foster City, CA). Relative expression was calculated using the comparative δ δ (Ct) method.

All animal experiments were approved by and conform to the relevant regulatory standards of the Institutional Animal Care and Use Committee at the Dana-Farber Cancer Institute. The in vivo efficacy of KPT-9274 was tested in nude mice subcutaneously injected with either MM1S or OPM2 cell line. Following detection of tumor, mice were treated with either vehicle or KPT-9274 orally. Tumor growth was measured in 2 perpendicular dimensions using a caliper and the following formula: V = (a² × b)/2, where “a” is the width of the tumor (smaller diameter) and “b” is the length (larger diameter) as previously described.²² 

The pGL3 Luciferase plasmid containing the NF-қB responsive elements cloned upstream the firefly luciferase reporter gene and the Cignal SRE Reporter Assay Kit (Qiagen) were used to measure NF-қB and ERK activity, respectively. Transfections were performed by electroporation with Neon transfection system (Life Technologies), and Renilla luciferase from pCMV-RL was included to normalize expression of firefly luciferase.

Detailed methods can be found in the supplemental Methods, available on the Blood Web site.

Data were analyzed using unpaired Student t tests comparing 2 conditions or a 1-way analysis of variance with Bonferroni or Newman-Keuls correction for multiple comparisons using Graphpad software. P < .05 was considered significant. Data are presented as means, and error bars in the figures depict standard deviation.

We analyzed expression of PAK4 and phosphorylated (p)-PAK4 Ser474 in different MM cell lines and peripheral blood mononuclear cells (PBMCs) by WB analysis. As seen in Figure 1A-B, PAK4 is highly phopshorylated in the majority of MM cell lines, but not in PBMCs from healthy donors. Protein expression correlated with messenger RNA (mRNA) expression in MM cell lines, as evaluated by RNA-seq (data not shown). We confirmed high expression of PAK4 and (p)-PAK4 in CD138⁺ MM cells from patients with SMM (n = 3) and MM (n = 9), whereas normal plasma cells were negative for both pPAK4 and total PAK4 (Figure 1C; supplemental Figure 1).

To determine the role of PAK4 in MM growth and survival, we suppressed its expression using an inducible knockdown system and observed significant inhibition of MM cell proliferation (Figure 1D-E) and induction of a robust apoptotic response proportional to the reduction in PAK4 levels (Figure 1F). The cells with the most significant reduction in PAK4 mRNA levels (pTRIPZ numbers 23 and 84) showed the greatest induction of apoptosis (39% and 45% of annexin V positive cells, respectively) compared with scrambled or untreated cells (≈10% of annexin V–positive cells). Activation of caspase 3, -8, and -9 was also observed in PAK4-depleted cells (Figure 1G). Importantly, the growth inhibitory effect of PAK4 silencing was confirmed in 2 additional MM cell lines using CRISPR knockdown studies (supplemental Figure 2A). To test whether inducible depletion of PAK4 in MM cells might affect their ability to form tumors in vivo, we treated mice with a doxycycline diet to induce PAK4 knockdown. PAK4 depletion after tumor development significantly inhibited MM tumor growth (supplemental Figure 2B).

Conversely, ectopic expression of PAK4 increased the proliferation rate in the RPMI 8226 MM cell line in vitro (Figure 2A-B). Importantly, overexpression of PAK4 increased the tumorigenic capacity of this cell line in vivo in a xenograft mouse model of human MM: all mice injected with PAK4 overexpressing cells developed tumors at an earlier time point than those receiving control cells, and tumors grew significantly more rapidly (Figure 2C). The growth-promoting effect of PAK4 overexpression was confirmed in 4 additional MM cell lines with intermediate to low levels of either PAK4 or p-PAK4 (Figure 2D-E).

Previous studies have shown an association between the PAKs and the RAF/MEK/ERK and PI3K/AKT signaling pathways. To assess the mechanism by which PAK4 contributes to MM cell proliferation, we examined the effect of ectopic PAK4 overexpression or RNA interference–mediated knockdown on RAF/MEK/ERK and PI3K/AKT signaling pathways.²³  PAK4 overexpression was associated with activation of NF-κB and MEK-ERK pathways and related induction of CREB (Figure 2F). Interestingly, we have observed a pleiotropic effect on AKT signaling pathway, with a significant activation observed in the H929 cells upon PAK4 overexpression. However, treatment of cells with MEK1 inhibitor U0126, but not the PI3K inhibitor LY294002, significantly reduced PAK4-induced MM growth (Figure 2G), suggesting a significant role of the MEK-ERK pathway in the oncogenic potential of PAK4 in MM. WB analysis confirmed significant impact of PAK4 on the MEK-ERK signaling pathway in response to inducible PAK4 depletion (Figure 2H). Finally, although we observed a significant impact on cytoskeleton-related proteins such as LIMK1 and vinculin in response to PAK4 overexpression and/or inhibition, adhesion and migration processes in MM cells were not perturbed by PAK4 dysregulation (data not shown).

From these observations, we conclude that PAK4 promotes MM cell proliferation both in vitro and in vivo via activation of MEK/ERK survival pathway, representing a potential new therapeutic target for MM.

With a significant impact of PAK4 on MM cell growth, we have developed a novel class of orally bioavailable PAMs. Quantitative proteomics methods based on stable isotope labeling of amino acids in cells²⁴  revealed PAK4 (red circle) as the major target of KPT-7523 in the MS-751 cells (3-(4,5-dimethylthiazol-2-yl)-2,5-dimethyltetrazolium bromide; 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assay 50% inhibitory concentration [IC₅₀] = 30 nM) with minor targets identified as red dots in the graph (Figure 3A-B; Supplemental Table 1). Although KPT-7523 has a Michael acceptor, the analysis of the entire primary sequence by the mass spectrometry did not reveal any evidence of covalent binding of KPT-7523 to PAK4. We also confirmed binding of PAK4 to KPT-7523 resin in an additional 3 cell lines, MS-751, HeLa, and U-2 OS (Figure 3C). Although all 6 PAK family members are structurally similar and group II PAK members have an even greater functional similarity, KPT-7523 binds specifically to PAK4, with only a weak affinity to PAK5 (purified protein; Figure 3D and data not shown).

In order to identify the binding site of KPT-7523, PAK4 regulatory (1-290 amino acids [aa]) and kinase (291-591 aa) domains were produced and purified from Escherichia coli. Immunoprecipitation of the purified peptides with KPT-7523 revealed specific binding only to the PAK4 kinase domain (C-terminal; Figure 3E). In addition, immunoprecipitation in HEK293 cells transfected with FLAG-tagged full-length PAK4, PAK4 kinase domain, or PAK4 regulatory domain showed that KPT-7523 bound preferentially to full-length and KD-PAK4 but not to regulatory domain (supplemental Figure 3A), confirming the results obtained with purified proteins. Therefore, KPT-7523 binds specifically to the PAK4 kinase domain. We further confirmed binding of purified kinase domain or full-length PAK4 to KPT-7523 using isothermal calorimetry and surface plasmon resonance (supplemental Figure 3B-C). The free energy, enthalpy, and entropy suggest a favorable binding.

Structurally similar analogs of KPT-6604, KPT-7189, KPT-8752, and KPT-9274 were next tested against a panel of MM cell lines and PBMCs from healthy donors for their effect on growth and survival. As shown in supplemental Figure 4A for the OPM2 MM cell line, chosen as representative of a panel of 10 cell lines tested, all the compounds decreased MM cell growth and survival in a time- and dose-dependent manner while sparing healthy donor PBMCs, indicating a favorable therapeutic index. To confirm specificity of the effect, we also tested an inactive compound (KPT-9101) and observed no significant effect on both MM growth and survival (supplemental Figure 4B). Based on optimized properties of Absorption, Distribution, Metabolism, and Excretion/Pharmacokinetics, compound KPT-9274 (Figure 3F) was selected for additional preclinical studies as well as for human clinical trials (see clinicaltrials.gov; NCT02702492).

In order to establish the translational potential of targeting PAK4 in myeloma and the clinical applicability of KPT-9274, we evaluated the sensitivity of KPT-9274 in a large panel of human myeloma cell lines representing the different translocation-associated subgroups. Both myeloma cell survival and growth were significantly affected upon treatment with KPT-9274 in a time- and dose-dependent manner in MM cell lines (supplemental Figure 4C-D and data not shown). Importantly, IC₅₀ analysis at 48 hours after treatment showed a greater sensitivity to KPT-9274 in 18 of 23 myeloma cell lines compared with PBMC from healthy donors (Figure 4A). We confirmed significant antimyeloma effect of KPT-9274 in CD138-positive myeloma cells from patients (N = 4) (Figure 4B). Importantly, exogenous PAK4 expression partly rescued the impact of KPT-9274 on MM cell growth, whereas PAK4 depletion by inducible KD did not potentiate KPT-9274 activity against MM cell proliferation (supplemental Figure 4E-F).

We next examined the antimyeloma effect of KPT-9274 in the context of the myeloma BM milieu. Treatment with KPT-9274 suppressed MM–bone marrow stromal cells (BMSC) interaction-mediated growth of MM cells (Figure 4C), yet had no effect on the viability of BMSCs (data not shown). Importantly, we evaluated the effect of KPT-9274 in primary myeloma cells from 6 patients cultured in the presence of their respective BM microenvironment using an automated flow cytometry platform for functional drug screening. We observed a dose- and time-dependent effect of KPT-9274 against myeloma cells, whereas normal BM cells derived from the same patient resulted in resistance to the treatment (Figure 4D-E and data not shown).

Interestingly, IC₅₀ analysis showed a greater sensitivity to KPT-9274 in MM cell lines with t(4:14) or harboring FGFR3 gene mutations, such as KMS11 and OPM2 cells (Figure 5A; supplemental Figure 5), and in CD138-positive cells from a t4:14 MM patient with FGFR3/IGH fusion (IC₅₀ = 100 nM) (Figure 5B). KPT-9274 treatment had no significant effect on CD138-negative cells from the same patient. This superior sensitivity to KPT-9274 in t(4:14) myeloma cells was also observed in vivo in mice injected with the t(4:14) FGFR3-mutated OPM2 MM cell line, compared with humanized murine xenograft model of myeloma in which nude mice were subcutaneously injected with the MM1S MM cell line (Figure 5C-D).

Using a high-throughput array containing 400 antibodies against signal transduction proteins, we have identified FGFR3 as a binding partner of PAK4 in myeloma cells (Figure 5E). Based on this observation, we next investigated whether KPT-9274 treatment could disrupt this interaction and therefore specifically sensitize FGFR3-addicted cells. Protein binding analysis revealed disruption of PAK4 interaction with FGFR3 after KPT-9274 treatment (Figure 5F), suggesting that targeting PAK4 may have a significant impact especially in FGFR3-driven myeloma.

Finally, treatment with KPT-9274 triggered a dose- and time-dependent induction of apoptosis in myeloma cells cultured in the presence or absence of BMSC, as evaluated by flow cytometric analysis following Annexin V and propidium iodide (PI) staining (Figure 6A and data not shown). Both intrinsic and extrinsic apoptotic pathways through caspases -3/7, -8, and −9 activation and poly-ADP ribose polymerase cleavage were induced in MM cells by KPT-9274, as assessed by WB analysis (Figure 6B) and luminescence assays (Figure 6C). Caspases activation by KPT-9274 treatment was also confirmed in primary CD138⁺ patient MM cells (Figure 6D). Exposure to the pan-caspase inhibitor zVAD significantly decreased the induction of apoptosis by KPT-9274, suggesting a caspase-dependent apoptotic process (Figure 6E).

We next evaluated and observed a significant impact of KPT-9274 on activation of MEK/ERK, AKT, and NF-κB signaling pathways (Figure 7A). In order to establish a predominant pathway responsible for the antimyeloma activity of PAK4 inhibitor, we performed NF-қB and ERK promoter activity assay in both resistant and sensitive cell lines in the presence and absence of KPT-9274. We observed inhibition of NF-қB activity in all MM cell lines tested, including the resistant U266 cell line, upon KPT-9274 treatment (Figure 7B). Strikingly, ERK activity was significantly inhibited in the sensitive MM cell lines after treatment with KPT-9274, but not in the resistant U266 cells (Figure 7C).

Gene expression profiling in the OPM2 and KMS11 cell lines after KPT-9274 treatment confirmed a significant decrease of ERK target genes, as well as antiapoptotic and FGFR3-related genes (MCL1 and DUSP6 among the others) compared with untreated cells (Figure 7D) (Gene Expression Omnibus accession number GSE93745). These observations were also confirmed by qPCR and WB analysis in additional myeloma cell lines (data not shown). Interestingly, ingenuity pathway analysis confirmed the presence of 2 major gene network hubs: MEK (activation z-score −2.8) and MYC (activation z-score −2.2).

Intriguingly, we have evidence that PAMs also inhibit the enzymatic activity of nicotinamide phosphoribosyltransferase (NAMPT; supplemental Figure 6A-B). We found that NAMPT forms a complex with the conserved PAK binding domain independently of Cdc42 activation, and treatment with KPT-9274 disrupted this interaction ( supplemental Figure 6C-D). Cea et al have shown that nicotinamide adenine dinucleotide (NAD) depletion decreased myeloma cell proliferation through ERK inhibition.²⁵  Our observations suggest that PAK4 and the NAMPT pathway may cooperate to regulate ERK signaling in myeloma, which can be efficiently targeted by KPT-9274.

ERKs can directly phosphorylate and therefore activate a set of transcription factors to regulate a diverse range of cellular processes.²⁶  By using a high throughput transcription factor DNA binding assay, we observed significant decreased binding to the DNA domains of established nuclear targets of activated ERK, including AP1, Ets, EGR1, and CREB, in myeloma cells upon treatment with KPT-9274 (Figure 7E). These results establish PAK4 as an oncogenic activator of MEK that induces phosphorylation and therefore induction of ERK1/2 signaling pathway. This intersection between PAK4 and the MEK/ERK signaling may ultimately impact expression and/or activity of downstream targets endowed with oncogenic potential in myeloma, such as MYC and Cyclin D2. Importantly, AKT and mTOR inhibitors were synergistic with KPT-9274, while the combination with the MEK inhibitor U0126 was at best ineffective or nearly antagonistic (Figure 7F; supplemental Figure 4G).

Altogether, this set of data establishes PAK4 as an oncogenic activator of MEK/ERK signaling pathway in myeloma cells that can be therapeutically targeted with KPT-9274.

STKs play an important role in cellular homeostasis and signaling through their ability to phosphorylate transcription factors, cell cycle regulators, and a vast array of cytoplasmic and nuclear effectors.²⁷  A number of STKs have been implicated in human cancers, including myeloma.²⁸  The PAK family of STKs has been associated with malignant transformation. Although PAK1 and PAK2 have been shown to be relevant in myeloma cell migration,²⁹  the role of PAK4 in MM remains unknown. The PAK4 gene lies within chromosome region 19q13.2 commonly amplified in 30% of MM patients.³⁰  We confirmed a correlation between copy number amplification and increased expression of PAK4 in 2 different myeloma patient datasets (data not shown). Moreover, high expression of total and phosphorylated PAK4 was also observed in the premalignant monoclonal gammopathy of undetermined significance and in the SMM stages, suggesting that its overexpression may be an early event in the pathogenesis of myeloma.

PAK4 is involved in a wide range of biological activities, which include protecting cells against apoptosis, activating cell survival pathways under stress, inhibiting cell adhesion, and promoting cell migration.^{6,7,17,19,31-33}  In this study, we show that myeloma cells with high PAK4 expression display sensitivity to conditional PAK4 knockdown, implying that an oncogene-addicted state exists in such cells. Moreover, we show that PAK4 promotes myeloma cell proliferation in vitro as well as in vivo in humanized murine models of myeloma through activation of antiapoptotic and survival signaling pathways, including the MEK/ERK pathway. This observation is further highlighted by sensitivity to a novel small molecule inhibitor of PAK4, KPT-9274, that has recently entered a phase 1 clinical trial in patients with solid malignancies or non-Hodgkin lymphoma (NCT02702492).

KPT-9274 effectively inhibits MM cell growth and survival, with no significant effect on normal PBMCs or bone marrow mononuclear cells, suggesting a potentially favorable therapeutic index. Inhibition of ERK by KPT-9274 also correlates with decreased DNA binding activity of ERK-dependent transcription factors AP1, ETS, CREB, and EGR, and decreased expression of ERK target genes such as CCND2, CCR1, and MYC. The observed lack of synergistic/additive effect of KPT-9274 in combination with MEK inhibitor U0126 (supplemental Figure 4G) further confirms ERK as one of the major pathway responsible for its activity in myeloma.

Intriguingly, we have evidence that PAMs also inhibit the enzymatic activity of NAMPT, by disrupting the interaction between the conserved PAK binding domain and NAMPT. We have previously shown that MM cells are remarkably dependent on NAMPT activity²⁵  and observed a significant cytotoxic activity against tumor cells via inhibition of ERK and mTOR pathways by FK866 (APO866), identifying NAD depletion as a promising antimyeloma therapy. Our observations suggest that PAK4 and the NAMPT (NAD) salvage pathway may cooperate to regulate ERK signaling. Therefore, this dual activity of KPT-9274 against PAK4 and NAMPT may provide a higher level of ERK pathway inhibition compared with drugs targeting individual pathways alone.

Finally, we have identified FGFR3, a commonly disrupted tyrosine kinase receptor,^34,35  and Grb2, an adaptor protein involved in RAS activation,³⁷  as novel PAK4 binding partners in myeloma and report disruption of this binding by KPT-9274. As a result, t(4;14)-positive MM cells expressing FGFR3 show greater sensitivity to PAK4 inhibition compared with MM cell lines without FGFR3 aberrations (eg, U266 and ANBL6). These observations are in line with the observed effect of KPT-9274 on MEK-ERK signaling, a pathway also induced by activating FGFR3 mutation.³⁶  However, FGFR3 downstream signaling pathways inducing oncogenesis and cancer cell growth are not yet fully delineated. A detailed investigation of cellular and molecular impact of FGFR3-PAK4 interaction may therefore provide new insights on FGFR3-mediated oncogenic signaling pathways and its therapeutic potential. Our findings indicate that FGFR3 may serve as a predictive marker of KPT-9274 sensitivity in MM patients, with the potential for broader application to other malignancies associated with dysregulation of FGFR3, including human bladder and cervical carcinomas.³⁸ 

In conclusion, our study sheds light on the oncogenic role of the STK PAK4 as survival and antiapoptotic factor in myeloma and shows that its inhibition with a new allosteric modulator, KPT-9274, represents a potential novel therapeutic intervention in MM, especially in high-risk t(4:14)-positive patients.

This work was supported by the Ron and Anita Wornick Fund, National Institutes of Health National Cancer Institute grants PO1-155258 and P50-100707 (N.C.M. and K.C.A.), Department of Veterans Affairs Merit Review Award 1 I01BX001584-01 (N.C.M.), and BA15/00035 and Cancer Research Innovation Spain grants (J.M.-L.).

Contribution: M.F. and N.C.M. designed the study and wrote the manuscript; M.F., J.M.-L., W.S., S.O., and R.L.B. performed the experiments; M.F. and J.M.-L. analyzed the data; N.A., R.S., A.R., and M.C. provided vital reagents and inputs to the study; Y.X. and A.G. performed protein binding assay; M.L. performed IHC staining; M.K.S. analyzed gene expression data; E.B. and S.S. designed KPT-9274 and its analogs; C.A. performed the PAK-NAMPT pull-down experiments; Y.L. and B.K. planned and designed the stable isotope labeling of amino acids in cells experiments; S.L., S.-L.W., and M.S. performed mass spectrometry runs and data analysis; and W.S., R.P., and K.C.A. provided critical evaluation of experimental data and manuscript.

Conflict-of-interest disclosure: W.S., E.B., S.S., C.A., and Y.L. are employees of Karyopharm Therapeutics Inc. N.C.M. serves on advisory boards to Millennium, Celgene, and Novartis. K.C.A. serves on advisory boards to Celgene, Millennium, BMS, and Gilead and holds equity ownership in Oncopep and Acetylon. The remaining authors declare no competing financial interests.

Correspondence: Nikhil C. Munshi, Dana-Farber Cancer Institute, 440 Brookline Ave, M230, Boston, MA 02115; e-mail: nikhil_munshi@dfci.harvard.edu.