Molecular mechanism and therapeutic implications of selinexor (KPT-330) in liposarcoma

Exportin-1 mediates nuclear export of multiple tumor suppressor and growth regulatory proteins. Aberrant expression of exportin-1 is noted in human malignancies, resulting in cytoplasmic mislocalization of its target proteins. We investigated the efficacy of selinexor against liposarcoma cells both in vitro and in vivo. Exportin-1 was highly expressed in liposarcoma samples and cell lines as determined by immunohistochemistry, western blot, and immunofluorescence assay. Knockdown of endogenous exportin-1 inhibited proliferation of liposarcoma cells. Selinexor also significantly decreased cell proliferation as well as induced cell cycle arrest and apoptosis of liposarcoma cells. The drug also significantly decreased tumor volumes and weights of liposarcoma xenografts. Importantly, selinexor inhibited insulin-like growth factor 1 (IGF1) activation of IGF-1R/AKT pathway through upregulation of insulin-like growth factor binding protein 5 (IGFBP5). Further, overexpression and knockdown experiments showed that IGFBP5 acts as a tumor suppressor and its expression was restored upon selinexor treatment of liposarcoma cells. Selinexor decreased aurora kinase A and B levels in these cells and inhibitors of these kinases suppressed the growth of the liposarcoma cells. Overall, our study showed that selinexor treatment restored tumor suppressive function of IGFBP5 and inhibited aurora kinase A and B in liposarcoma cells supporting the usefulness of selinexor as a potential therapeutic strategy for the treatment of this cancer.


Research Paper www.impactjournals.com/oncotarget
Transport of specific proteins between the nucleus and cytoplasm is a fundamental process for maintaining cell proliferation and apoptosis of normal and tumor tissues. Exportin-1 (XPO1) is a well-known nuclear export receptor responsible for transporting more than 220 cargo proteins as well as several RNA molecules from the nucleus to the cytoplasm [8][9][10]. XPO1 recognizes cargo proteins through a hydrophobic, leucine-rich nuclear export signal, which is dependent on the RanGTP/GDP axis [8][9][10]. XPO1 is the sole nuclear exporter of many tumor suppressive and growth-stimulatory proteins including p21, p27, p53, p73, STAT3, BRCA1, FOXO, CDKN1A, RB1, IkB, APC, NPM1 and Survivin [11][12][13][14]. XPO1 is up-regulated in different human malignancies such as leukemia [15], lung cancer [16], hepatocellular carcinoma [17], melanoma [18], as well as multiple myeloma [19]; and its overexpression is correlated with poor prognosis, resistance to chemotherapy and short survival [12], [15], [19]. Leptomycin B was the first well known natural XPO1 inhibitor that suppressed the growth of several human cancer cell lines [20]. However, this drug had significant toxicity and a narrow therapeutic window in preclinical animal models, as well as in phase 1 human clinical trial [21]. Recently, novel orally bioavailable small molecules known as Selective Inhibitors of Nuclear Export have been developed. These inhibitors specifically and reversibly bind to residue Cys528 in the cargobinding groove of XPO1. Selinexor (KPT-330) is the most advanced oral inhibitor of XPO1. Phase I/II human clinical trials have indicated that selinexor is well-tolerated and has a favorable outcome in patients with either relapsed or refractory acute myeloid leukemia (NCT01607892) and solid tumors (NCT01607905, NCT01896505) [22] (www.clinicaltrials.gov). In this current study, therapeutic potential of selinexor was examined against liposarcoma both in cell culture and in a murine xenograft model. Selinexor significantly inhibited cellular proliferation and induced cell cycle arrest and apoptosis of liposarcoma both in vitro and in vivo.

XPO1 expression in liposarcoma samples and cell lines and XPO1 silencing in liposarcoma cells
To determine the expression of endogenous XPO1 protein in liposarcoma patient samples, we first performed XPO1 staining on 20 well-differentiated liposarcoma, 13 dedifferentiated liposarcoma, 13 myxoid liposarcoma, 2 pleomorphic liposarcoma and benign lipoma tissue sections ( Figure 1A) and analyzed the staining levels by H-score method. A total of 58% of liposarcoma samples showed strong nuclear staining (H-score value > 199), 29% had moderate nuclear staining (H-score value > 99), and 13% had weak nuclear staining (H-score value 0 -99) (Supplementary Figure S1A). In contrast, very weak or negative immunoreactivity of XPO1 was observed in benign lipoma tissues ( Figure 1A). Western blot analysis showed XPO1 protein expression in liposarcoma cell lines of different histological subtypes (undifferentiated, SW872; well differentiated, T778;  dedifferentiated, LPS141, LP6; myxoid, MLS402; poorly differentiated, LISA-2; SA4) ( Figure 1B). Furthermore, immunofluorescence analysis revealed strong nuclear membrane localization of XPO1 protein in fixed, permeabilized LPS141, MLS402, SW872 and SA4 cells ( Figure 1C and Supplementary Figure S1B). In addition, XPO1 expression was examined in different subtypes of liposarcoma, using microarray database GSE21122 comprising 46 dedifferentiated liposarcoma, 23 pleomorphic liposarcoma, 20 myxoid liposarcoma samples and 9 normal fat samples. We observed that 90% of liposarcoma samples showed significantly (P < 0.01) higher expression of XPO1 compared to normal fat ( Figure 1D). These results demonstrated that XPO1 is prominently expressed in different histological subtypes of liposarcoma. To examine the biological role of XPO1 in liposarcoma, the gene was first suppressed using shRNA targeting to XPO1 resulted in 70-90% silencing of XPO1 protein in liposarcoma cells (LPS141, SW872, MLS402 and SA4) compared to scramble shRNA as shown by western blot analysis ( Figure 1E). This led to significant inhibition of cellular proliferation of these liposarcoma cells compared to scramble shRNA ( Figure 1F, Supplementary Figure S1C).

Inhibition of XPO1 decreased cellular growth of human liposarcoma cells
Next, efficacy of selinexor to inhibit XPO1 expression of LPS141, SW872, MLS402 and SA4 cells was examined after treating with increasing concentrations of selinexor (0-2000 nM, 24 h). Selinexor inhibited XPO1 protein levels in a dose-dependent fashion in all four liposarcoma cell lines at 24 h ( Figure 2A). However, selinexor treatment did not decrease XPO1 mRNA levels (data not shown) suggesting that the drug effected protein levels of XPO1. Further, a panel of liposarcoma cell lines representing different histological subtypes were treated with selinexor also caused a dose-dependent decrease in cell viability (IC50, ranged between 100-500 nM) ( Figure 2B) and also markedly inhibited the clonogenic capacity of liposarcoma cells in a dose-dependent manner ( Figure 2C and 2D).

Selinexor induced apoptosis and cell cycle arrest in liposarcoma cells
Liposarcoma cell lines were treated with increasing concentrations of selinexor (0-1000 nM) or diluent control for 24 h, and cell cycle distributions were determined by staining with propidium iodide (PI). Selinexor significantly lead to accumulation of cells in the G1 phase; and reduced cells in the S and G2M phase ( Figure 3A). Concomitantly, selinexor (1000 nM, 24 h) prominently decreased protein levels of cyclin B1, cyclin E, survivin and increased levels of p21, p27 and p53 ( Figure 3B). Flow cytometric analysis demonstrated a dose-dependent increase in the percentage of apoptotic cells (PI high /AV high ), and a parallel decrease in viable liposarcoma cells (PI low /AV low ) after culture with selinexor (0-2000 nM, 24 h) ( Figure 3C). Furthermore, western blot analysis showed that selinexor (1000 nM, 24 h) increased cleaved caspase-3 and cleaved PARP as well as increased BAX protein expression compared to vehicle control ( Figure 3B).

Selinexor inhibited growth of liposarcoma xenografts
LPS141 liposarcoma cells were injected subcutaneously into the right flank of NSG immunodeficient mice; after 14 days, tumors reached approximately 100 mm 3 in all the mice. The mice were randomly divide into two groups, and systemic treatment was begin with either selinexor (10 mg/kg dissolved in 100 μl of vehicle orally) or oral vehicle (100 µl) alone, 3 times per week for 4 weeks. Selinexor treatment resulted in significant decreased in tumor volumes and weights compared to the xenograft tumors in mice receiving vehicle alone ( Figure 4A and 4B). Western blot analysis of lysates of tumor tissue showed that selinexor decreased the protein levels of XPO1 and cyclin B1 and increased the levels of p21 and cleaved caspase 3 in the mice receiving selinexor compared with tumors in mice treated with vehicle control ( Figure 4C). Immunohistochemistry of these tumors showed significantly decreased number of Ki-67 (cell proliferation marker) and CD31 (blood vessels) positive cells and an increase in TUNEL-positive cells (increased apoptosis) in the selinexor treatment group compared to vehicle control group ( Figure 4D). Selinexor treatment caused no noticeable morbidity including no significant effect on body weight, numbers of total white blood cells, neutrophils, platelets, haemoglobin, as well as their serum levels of albumin, aspartate aminotransferase, alanine aminotransferase and creatinine in experimental versus control treated mice (Supplementary Table S2).

Gene expression profiling in selinexor treated liposarcoma cells
To determine global transcriptional consequences of XPO1 inhibition, gene expression profiling of LPS141 liposarcoma cells was performed after 12 hours treatment with either vehicle (DMSO) or selinexor using microarray. A total of 467 genes were down-regulated and 288 were up-regulated (FDR < 0.05) upon selinexor treatment of LPS141 cells as compared to vehicle treatment (Supplementary Figure S2A). Gene set enrichment analysis showed selinexor exposed cells had significant negative enrichment in cell cycle and aurora kinase pathways and positive enrichment in adipogenesis pathway (Supplementary Figures S2B-S2E). The fidelity of the microarray results was confirmed by further validation of 12 randomly selected genes through quantitative RT-PCR. Congruently with the microarray data, mRNA expression levels of cell cycle pathway genes (CCNB1, CCNB2, AURKA, AURKB, CDC25, TPX2) were significantly decreased, whereas mRNA levels of adipogenesis pathway genes (CEBPA, LPL, PPARG2, RB1, KLF6, IGFBP5, FBXW7, CEBPD, DDIT3) were significantly increased in LPS141 cells treated with selinexor compared to vehicle (Supplementary Figure S2F).

Inhibition of XPO1 induced cytotoxicity in liposarcoma cells by inducing insulin-like growth factor binding protein 5 (IGFBP5)
Insulin-like growth factor 1 receptor (IGF1R) is expressed in a wide range of tumors including liposarcoma; and IGF1R signaling is crucial for tumor formation and survival of malignant cells [23]. A recent study showed that the combination of IGF1R and CDK4 inhibitors synergistically reduced the cell proliferation of liposarcoma cells [24]. We found in our microarray data that two IGF-1 binding proteins (IGFBP5 and IGFBP6) were up-regulated in selinexor treated LPS141 cells (Supplementary Figure S2A). Many previous reports showed that IGFBP5 can act either as a tumor suppressor or oncogene in a tissue-specific context but nothing is known about the role of IGFBP5 in human liposarcoma [25][26][27]. RNA-sequencing analysis confirmed downregulation of IGFBP5 mRNA in liposarcoma cell lines compared to human adipose tissue (Supplementary Figure S3A, manuscript in preparation). Further, microarray data analysis (GSE21122) also showed downregulation of IGFBP5 compared to human normal fat samples (Supplementary Figure S3B). The up-regulation of IGFBP5 mRNA and protein upon treatment with selinexor was further verified both at the mRNA and protein levels in liposarcoma cell lines (Figures 5A and Supplementary Figure S3C). Interestingly, treatment of LPS141 and MLS402 cell lines with selinexor resulted in attenuation of phosphorylation of both IGF-1R and AKT after stimulation with IGF-1 ( Figure 5B).
We generated stable clones of LPS141, SW872 and MLS402 cells expressing either an empty vector or IGFBP5 expression vector ( Figure 5C). Overexpression of IGFBP5 in LPS141 and SW872 cells significantly (P < 0.001) inhibited cell growth in liquid culture ( Figure 5D and 5E), as well as clonogenic growth in soft agar (Supplementary Figure S3D and S3E) compared with empty vector. Silencing of IGFBP5 resulted in significant increase in the growth of LPS141 cells in liquid culture ( Figure 5F and 5G). Furthermore, knockdown of IGFBP5 in LPS141 and SW872 cells partially rescued the inhibitory effect of selinexor treatment ( Figure 5H and 5I).

Role of aurora kinases in liposarcoma
Our microarray data showed that aurora kinase pathway was one of the most significantly enriched pathways in the down-regulated genes in liposarcoma cells upon selinexor treatment compared to vehicle exposure (Supplementary Figures S2A and S2C). Recently, several aurora kinase inhibitors have entered Phase I and II clinical trials against different types of cancers [28][29][30].  in preparation). In silico analysis also confirmed that aurora-A and aurora-B mRNAs were significantly upregulated in human liposarcoma microarray data (GSE21122) ( Figure 6A). Expression of aurora-A and aurora-B was silenced in liposarcoma cells using siRNA which resulted in a significant decreased in cell growth of LPS141 and MLS402 in liquid culture ( Figure  6B-6D, Supplementary Figure S4C). Selective inhibitors of aurora-A (MLN8237) and aurora-B (ZM447439) inhibited auto-phosphorylation of aurora-A and aurora-B and decreased growth in liquid (Supplementary Figure  S4D and S4E) as well as clonogenic growth ( Figure 6E) of LPS141 and MLS402 cells. Also, these drugs lowered the level of cyclin B1 (Supplementary Figure S4F) and markedly increased the protein expression of p53, p27 and p21 in LPS141 and MLS402 cells ( Figure 6F).

DISCUSSION
Liposarcoma is one of the most common soft tissue sarcoma causing substantial morbidity and mortality [3,31,32]. Inhibition of nucleo-cytoplasmic transport by natural and synthetic products is a therapeutic approach [33]. Recently, Selective Inhibitors of Nuclear Export compounds have been developed [17,22] and showing anti-tumor activity against various human malignancies in vitro and in murine human cancer xenograft models with minimal toxicities [15][16][17]34].
Present study shows prominent expression of XPO1 protein in different histological subtypes of liposarcoma in patient samples and cell lines compared to benign lipomas. Immunofluorescence experiment data confirmed that XPO1 is localized at the nuclear pore in liposarcoma cells. Also, microarray database GSE21122 indicated significant increased XPO1 mRNA expression in different subtypes of liposarcoma samples compared to normal fat. These data suggest that robust expression of XPO1 may have a potential role in the pathogenesis or progression of liposarcoma. Silencing of XPO1 suppressed the growth of several liposarcoma cell lines. Further, selinexor caused growth inhibition, cell-cycle arrest with upregulation of p21, p27, p53 and downregulation of cyclin B1, cyclin E and survivin. The drug increased apoptosis as shown by upregulation of cleaved caspase 3, cleaved PARP and BAX, as well as annexin V positive staining of liposarcoma cells. Well-differentiated and dedifferentiated liposarcomas have 12q13-15 amplification resulting in MDM2 overexpression. MDM2 can inactivate p53. We found that selinexor increased the expression p53 and its target genes (e.g., p21 and BAX) proteins in LPS141 liposarcoma cells (MDM2 amplification) without altering the expression of MDM2 ( Figure 3B and Supplementary Figure S5A). This observation is consistent with a previously published study [35]. Of note, selinexor exposure (≤ 2000 nM) did not affect p53, MDM2, CDKN1A mRNA levels suggesting that the drug post-transcriptionally increased the level of these proteins (Supplementary Figure S5B). Of interest, a recent study showed that selinexor induced cell cycle arrest and apoptosis of liposarcoma cells irrespective of their p53 expression or mutational status [35].
Importantly, selinexor (10 mg/kg orally, three times per week at a non-toxic dose) significantly inhibited growth of dedifferentiated liposarcoma (LPS141) xenografts in NSG mice associated with reduced cellular proliferation (Ki-67 staining) and an increase in apoptosis (Tunel staining) in the tumors. Recently, a phase 1 clinical trial of selinexor in patients with advanced solid tumors including liposarcoma [36,37] showed a favorable anti-tumor effect of selinexor in 78% (14 of 18) of the liposarcoma patients, with a well-tolerated dose of 35 mg/m 2 (approximately 60 mg flat dose) [37]. Taken together, these results strongly indicate that inhibition of XPO1 might be a valuable treatment approach for this disease.
Our microarray data analysis showed that IGFBP5 expression was markedly increased in liposarcoma cells after treatment with selinexor. IGFBP5 is the most conserved member of the IGFBPs family and is frequently deregulated in human malignancies such as neuroblastoma [25], osteosarcoma [27,38], breast [26,39,40] as well as head and neck squamous cell carcinoma [41]. IGFBP5 modulates cellular functions either in an IGF-dependent manner through IGF1R signaling or in an IGF-independent manner [42]. Notably, we found that selinexor increased expression of IGFBP5 both at the mRNA and protein levels in liposarcoma cells. We provide evidence that IGFBP5 can acts as a tumor suppressor in liposarcoma cells in an IGF-1 dependent manner; and selinexor can markedly up-regulate the expression of this tumor suppressor protein. In addition, microarray data showed significant enrichment of the  LPS141 and SW872 cells were transfected either with IGFBP5 siRNA or scramble siRNA. These cells were exposed to different concentration of selinexor for 48h, and growth inhibition was measured by MTT assay. Data represent mean ± SD; n = 4. **P ≤ 0.001. Data for G, H and I represent mean ± SD of three independent experiments done in triplicates. **P < 0.001 (Student's t-test). aurora kinase pathway in liposarcoma cells. Overexpression and gene amplification of aurora-A and aurora-B in liposarcoma have been reported to correlate with tumor grade and prognostic markers [43][44][45][46]. We showed that selinexor can significantly decrease levels of aurora-A and aurora-B mRNA. Also, inhibition of aurora kinases either through siRNA or selective inhibitors of aurora-A and aurora-B suppressed the cell growth of liposarcoma cells in both liquid culture and soft agar associated with decreased cyclin B1 and increased expression of tumor suppressor proteins, p53, p27, and p21.
Taken together, our data demonstrated that in preclinical studies, selinexor is a potent therapeutic agent against liposarcoma. This activity is probably mediated by a number of pathways as suggested by our studies. Clinical trials of this novel agent in liposarcoma are clearly warranted.

RT-PCR analysis and quantitative real-time PCR (qRT-PCR)
Total RNA was isolated from liposarcoma cell lines (LPS141, SW872, MLS402 and SA4) with the RNeasy mini kit (Qiagen GmbH, Hilden, Germany) according to manufacturer's protocol. 1 ug of total RNA was used for complementary DNA (cDNA) synthesis using one step cDNA synthesis kit (Life Technologies Inc., Gaithersburg, MD, USA). For quantitative real-time PCR amplification, KAPA SYBR Green was used. Real-time PCR was performed using KAPA SYBR master mix (KapaBiosystems, Woburn, MA, USA) on an ABI 7900 Fast real-time PCR system (Applied Biosystems, USA) following the supplier's protocol. Thermal conditions for real-time qRT-PCR were set at 95 0 C for 10 min for initial denaturation, followed by 40 cycles of PCR with denaturation at 95 0 C for 15 seconds and annealing/ extension at 60 0 C for 1 min. Primers used for qRT-PCR for selected genes are shown in Supplementary Table S1.

Cell proliferation assay (MTT assay)
Anti-proliferative effect of selinexor against liposarcoma cells was determined using colorimetric assays as described [48]. Briefly, 4 × 10 3 liposarcoma cells were seeded in quadruplets in 96-well plates either in the presence or absence of selinexor at 37°C in a CO 2 incubator. At the conclusion of the experiment, 20 μl of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich) was added to each well and incubated at 37°C in a CO 2 incubator for 2hr and then dissolved in 100 µl of stop solution (SDS-HCl). Absorbance was measured at 570 nm using a microplate reader (Infinite 200; Tecan, San Jose, CA, USA). Doseresponse curves were plotted to calculate half-maximal inhibitory concentrations (IC50) for selinexor by GraphPad Prism4 (Graph Pad Software, San Diego CA, USA).

Colony formation assay
Liposarcoma cells (1 × 10 3 ) were seeded into 6-well plates in triplicates in complete medium. After two days, media were supplemented with varying concentrations of selinexor. After 2 weeks, colonies were fixed with 5% glutaraldehyde and stained with crystal violet. For quantitative measurement, colonies were dissolved in 200 ul of DMSO and absorbance was measured at 570 nm using a microplate reader (Infinite 200; Tecan, San Jose, CA, USA).

Cell cycle analysis
To determine the effect of selinexor on cell cycle, liposarcoma cells were cultured with either diluent control (DMSO) or various concentrations of selinexor at indicated time points as described earlier [47]. Briefly, cells were trypsinized, washed with ice-cold PBS and fixed with 70% chilled ethanol. Cells were stained with PI solution (50 µg/ml PI, Triton X-100 (1%), 20 ug/ml DNase-free RNase A in PBS) for 30 min at 37°C in dark and analyzed by LSR-II flow cytometer (Becton-Dickinson, San Jose, CA, USA).

Annexin V and propidium iodide (Annexin V-PI) apoptosis analysis
Annexin V-PI staining was performed using flow cytometric analysis as previously described. Briefly, 1 × 10 6 cells were cultured with either diluents control (DMSO) or varying concentrations of selinexor in 6 well plates for 72 h and staining was performed using Apoptosis Detection Kit II (BD Biosciences, USA). Cells were washed twice with 1X ice-cold phosphate-buffered saline (PBS; Life technologies, USA), trypsinized and centrifuged. Cells were again washed with 1X PBS, resuspended in 1X binding buffer containing 5 ul of FITCI conjugated Annexin V and 5 ul of PI for 20 min in dark. The samples were analyzed using LSR-II flow cytometer (BD, San Jose, CA, USA).

Liposarcoma xenograft murine model
Six-week-old male NSG mice were injected with LPS141 (2 × 10 6 ) cells suspensions mixed with matrigel (BD Biosciences) (1:1) in 200 μl of total volume and injected subcutaneously on the flank. Treatment was initiated 2 weeks after cell implantation. The mice were randomly divided into two groups (6 mice per group) to receive either vehicle control (0.6% w/v aqueous Pluronic F-68) or selinexor (10 mg/kg in 100 μl of the vehicle, 3 times per week). Tumor diameters were measured with a caliper, and tumor volumes were calculated by the formula (V = π/6 × Dl × Ds 2 ), where V is volume, Dl is the largest diameter, and Ds is the smallest diameter [49]. Mice were fed with Nutri-Cal (Tomlyn) thrice a week during treatment to provide good nutrition. All mice were sacrificed at the end of the study. Tumors were dissected and weighed. One-half of each tumor was fixed in 10% formaldehyde for immunohistochemistry and the other half was immediately kept in −80 0 C for subsequent RNA and protein analysis. We also analyzed peripheral blood for blood counts and serum chemistry after the 28-days of treatment (Supplementary Table S2). The animal studies were approved by the National University of Singapore (NUS) Institutional Animal Care and Use Committee (IACUC).

Immunohistochemistry
Tissue microarray slides were purchased from Super Bio Chips. Tumor xenografts were fixed in 10% formaldehyde, embedded in paraffin and cut into 4 μm thick sections. Immunohistochemistry was performed as described [47,49]. Briefly, endogenous peroxidase activity was blocked using 3% hydrogen peroxide for 10 min and antigen retrieval was done using retrieval buffer (pH 6). Slides were blocked with 10% goat serum. Antibodies against XPO1, CD31, Ki-67 and Tunel were used. Immunostaining was assessed by counting > 500 cells from 5 random fields of each specimen under ×400 magnification in the best-stained tumor area of each section as described previously [3,47].

Microarray analysis
Total RNA was extracted LPS141 cells after 12 hours treatment with either vehicle (DMSO) or selinexor using RNeasy mini kit (Qiagen, Valencia,CA) according to manufacturer's protocol for microarray experiments as described previously [47]. RNA Nano chip on Agilent Bioanalyzer 2100 (Agilent Technologies, Inc., Santa Clara, CA) was used to check integrity and quantity of RNA samples. cDNA and cRNA was synthesized using Illumina Total Prep RNA Amplification Kit (Illumina). Human gene HT-12 v4 Expression beads chip kits (47,000 probe spotted genes) from Illumina (Santa Clara, CA) were used.

RNA interference
To obtained knockdown of XPO1 in liposarcoma cell lines for in vitro studies, human XPO1 gene specific shRNAs and a non-targeting shRNA were cloned into a lentiviral vector. For stable knockdown, lentiviral particles were generated according to manufacturer's protocol. Cells were infected with lentivirus particles at a MOI of 20 with 8 μg/ml (Sigma-Aldrich) for 24 h, and stable cells were selected using puromycin for 1-2 weeks.
Human IGFBP5, aurora-A and aurora-B gene specific siRNA oligos (siGENOME) including scramble oligos were purchased from Dharmacon (Coralville, USA). LPS141, MLS402 and SW872 cells were transfected with 30 pmol siRNA using Lipofectamine RNAiMax (Life Technologies, USA) according to the manufacturer's protocol. After 48 hours of transfection, cells were used for qRT-PCR; western blotting confirmed knockdown. Each experiment was performed at least in triplicate on three different occasions.

Statistical analysis
For in vitro and in vivo experiments, we evaluated the statistical significance of the difference between two groups using two-tailed Student t-test. Asterisks in the figures represent significant differences between experimental groups in comparison to controls ((*p < 0.01, **p < 0.001, ***p < 0.0001). Data points in figures represent the means ± SD (standard deviation).

CONFLICTS OF INTEREST
S.S, E.B, and M.K are employees and stockholders of Karyopharm Therapeutics. The remaining authors have no competitive financial interest.