The microarray and limited clinicopathological data from the Gene Expression Omnibus data set GSE30929 for LPS, a raw data set from the study of Gobble et al (15), were obtained from the National Center for Biotechnology Information website (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE30929). Only data of WD LPS and DD LPS were used for analysis. Gene expression data from GSE30929 were normalized using dChip (16,17) and filtered with a max/min threshold of ≥5, after which they were exported for further analysis. For another data set from the study of Singer et al (13), including 12 cases of DD LPS and 18 cases of WD LPS, only certain genes were available from www.cbio.mskcc.org/Public/Liposarcoma. Thus, no further filtering process was conducted. Both data sets are the Affymetrix U133A platform (Thermo Fisher Scientific, Inc.). In order to identify genes showing statistically significant, concordant differences of expressions between two biological states (e.g. phenotypes), Gene Set Enrichment Analysis (GSEA), a method developed by the Broad Institute, was performed using software downloaded from http://www.broadinstitute.org/gsea/index.jsp (18,19). The expression levels of individual genes were obtained using Z-score transformation, and the differences between different subtypes were then compared using a t-test. In the GSE30929 data set, data regarding patients' endpoint (distant recurrence) and follow-up time were obtained (ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE30nnn/GSE30929/matrix/). Therefore, distant recurrence-free survival (DRFS) was estimated via Kaplan-Meier analysis, and a log-rank test was used to compare the survival curves between groups.

The LPS cell lines, including LPS853 was kindly provided by Dr Fletcher JA (Department of Pathology, Brigham and Women's Hospital), while NDDLS-1 was kindly provided by Dr Ariizumi (Division of Orthopedic Surgery, Niigata University Graduate School of Medical and Dental Sciences). LPS853 cells were maintained in RPMI medium with 15% fetal bovine serum (HyClone; GE Healthcare Life Sciences), 100 µg/ml penicillin streptomycin (HyClone; GE Healthcare Life Sciences) and 2 mM L-glutamine solution (HyClone; GE Healthcare Life Sciences); NDDLS-1 cells were maintained in RPMI medium with 10% fetal bovine serum (HyClone; GE Healthcare Life Sciences), 100 µg/ml penicillin streptomycin (HyClone; GE Healthcare Life Sciences), 2 mM L-Glutamine solution (HyClone; GE Healthcare Life Sciences) and 60 µg/ml kanamycin (Thermo Fisher Scientific, Inc.). MLN8237, an AURKA-selective inhibitor, and chemotherapeutic agents doxorubicin, topotecan, cisplatin and taxol were purchased from Selleck Chemicals. For MLN8237, concentrations of 0, 0.02, 0.05, 0.1, 0.5 and 1 µM were used to evaluate mitotic inhibition, the cell cycle and cytotoxic effects; concentrations of 0, 0.4, 0.8, 1.2, 1.6 and 2 µM were used for the exploring synergistic effects with chemotherapeutic agents. The concentrations of chemotherapeutic agents were listed as follows: Cisplatin 0, 4, 8, 12, 16, 20 µM; topotecan 0, 4, 8, 12, 16, 20 µM; doxorubicin 0, 0.1, 0.2, 0.3, 0.4, 0.5 µM; and taxol 0, 6, 12, 18, 24, 30 µM. Nocodazole was purchased from Sigma-Aldrich (M1404-2MG; Merck KGaA). 0 µM inhibitor served as the control. The following antibodies were used for immunoblotting: β-Actin (cat. no. ab6276-100; 1:10,000; Abcam); Aurora A (cat. no. 3092; 1:1,000; Cell Signaling Technology, Inc.); anti-phosphorylated-Aurora A (Thr288; p-Aurora A; cat. no. 3079; 1:1,000; Cell Signaling Technology, Inc.); poly-(ADP-ribose) polymerase (PARP; cat. no. 9542; 1:1,000; Cell Signaling Technology, Inc.), Histone H3 (HisH3; cat. no. 9715; 1:1,000; Cell Signaling Technology, Inc.), anti-phosphorylated (p)HisH3 (Ser10; pHisH3; cat. no. 9701; 1:2,000; Cell Signaling Technology, Inc.), p70S6 kinase (cat. no. 9202; 1:1,000; Cell Signaling Technology, Inc.) and pp70S6 kinase (cat. no. 9205; 1:500; Cell Signaling Technology, Inc.). The secondary antibodies used were horse anti-mouse IgG (cat. no. 7076; 1:3,000) and goat anti-rabbit IgG (cat. no. 7074; 1:3,000), both from Cell Signaling Technology.

Immunoblotting was performed as previously described (20). In brief, cultured monolayer cells were rinsed in PBS and scraped into lysis buffer (RIPA Lysis and Extraction Buffer; cat. no. 89901; Thermo Fisher Scientific, Inc.) containing protease and a phosphatase inhibitor cocktail (1:100 dilution; Thermo Fisher Scientific, Inc.). Lysates were first incubated for 30 min at 4°C and then centrifuged for 30 min at 16,000 × g at 4°C. A Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Inc.) was employed to determine protein concentrations. Protein extracts (20-50 µg per lane) were then electrophoretically separated on SDS-PAGE (8-12% depending on the molecular weights of proteins), transferred to polyvinylidene difluoride membranes (PerkinElmer, Inc.), and immunoblotted with specific antibodies. For primary antibodies, the sample was incubated at 4°C overnight; for secondary antibody, the sample was incubated at room temperature for 1 h. Immunoreactive bands were detected through enhanced chemiluminescence (EMD Millipore) and exposed to X-ray films. Protein expression was quantified using ImageJ version 1.51j8 (National Institutes of Health).

TACS™ MTT assay (R&D Systems) was used for the cell cytotoxicity assay. In brief, ~2,000-20,000 cells per 100 µl per well were seeded in 96-well plates at 37°C overnight. The following day, reagents at different concentrations were added in triplicate. The plates were subsequently incubated for the 72 h at 37°C, incubated with 10 µl MTT reagent, and then incubated for a further 4 h at 37°C. Dimethyl sulfoxide 100 µl/well was added and mixed thoroughly to dissolve the blue crystals. A Vmax microplate reader (Molecular Devices, LLC) at wavelengths of 570 nm (test) and 650 nm was used to measure the absorbance. Cell survival was calculated through the following equation: % Survival=(mean experimental absorbance/mean control absorbance) ×100 (21).

The synergistic effect of the applied drug combination was measured through a combination index (CI), which was calculated using CalcuSyn software 2.1 (Premier Biosoft International) (22). CI >1 was defined as antagonism, CI=1 as additivity, and CI<1 as synergy; the experiment was performed in triplicate.

A trypan blue exclusion assay (23) was also employed for exploring the cytotoxic effects of MLN8237. LPS cells were plated at 5×10⁴ cells/well in 24-well plates; then, different concentrations of MLN8237 was added to each well the next day, and the plates were incubated at 37°C for 72 h. Subsequently, the cells were trypsinized and mixed with trypan blue at room temperature and observed immediately. Viable cells, recognized as cells that did not absorb trypan blue, were counted using a light microscope (magnification, ×100; cells were counted in the large, central gridded square at 1 mm², then multiplied by 1×10⁴ to estimate the number of cells per ml). All experiments were performed in triplicate.

After drug treatment, cells (1×10⁵ cells/well) were washed with 1X PBS once and resuspended in 100 µl staining solution (containing Annexin V fluorescein isothiocyanate (FITC) and propidium iodide in HEPES buffer, BD Pharmingen; BD Biosciences). The cells were then incubated at room temperature for 15 min and then diluted in 1X Annexin V-binding buffer (BD Biosciences). The percentages of apoptotic cells were then measured using flow cytometry (BD FACSCanto II, BD Biosciences; operation software: BD FACSDiva Software v8.0.2; analysis software: FlowJo 7.6).

For AURKA KD experiment, 5 ×10⁵ of LPS853 cells were transfected with either AURKA siRNA or non-targeting control pool using ON-TARGETplus system (GE Healthcare Dharmacon, Inc.) according to manufacturer's protocols (Target Sequence: UCGAAGAGAGUUAUUCAUA, CGGUAGGCCUGAUUGGGUU, UUCUUAGACUGUAUGGUUA, AAUAGGAACACGUGCUCUA; 10 µM); DharmaFECT 1 Transfection Reagent (cat. no. T-2001-03; GE Healthcare Dharmacon, Inc.) was employed for transfection. The cells were incubated for 48, 72 and 96 h along at 37°C with respective controls, and number of viable cells was measured using trypan blue exclusion assay, as described above. The remaining cells then were lysed and the KD efficiency of AURKA, as well as the protein levels of PARP were confirmed by western blotting, as aforementioned.

Cell cycle analysis was performed using flow cytometry, as previously described (9). After being trypsinized and washed twice with PBS, cells (1×10⁵ cells/ml) were fixed in 99% ethanol at -20°C overnight. The following day, the fixed cells were first washed twice with cold PBS, suspended in 420 µl PBS, added to 50 µl of 10 mg/ml RNase A (Sigma-Aldrich; Merck KGaA), and agitated at 37°C for 15 min. The cells were then maintained at room temperature for 1 h following the addition of 20 µl propidium iodide (0.2 mg/ml). Subsequently, flow cytometry was conducted using BD FACSCalibur (BD Biosciences) for relative DNA content based on red fluorescence levels. FACSDiva Software v8.0.2 was used to calculate the percentages of the cells in the different phases of the mitotic cell cycle.

SA-β-gal activity was detected using the Cellular Senescence Assay Kit (EMD Millipore), as described in the manufacturer's instructions. After being treated with MLN8237 at 37°C for 72 h, adherent LPS853 cells were fixed with 1X fix solution at room temperature, incubate for 10 min, and stained with X-gal in a staining solution at pH 6.0 at 37°C for 5 h. The cells were washed twice with 1X PBS. The percentage of SA-β-gal-positive cells (the number of positive cells relative to the total number of cells) was quantified by counting 100 cells in three randomly selected fields per dish with an OLYMPUS IX51 microscope (brightfield with phase contrast; magnification, ×200).

The expression of DcR2 was detected using flow cytometry (1×10⁵ cells/100 µl). LPS853 cells were first treated with MLN8237 at 37°C for 72 h, washed twice with 1X PBS, and incubated with Alexa Fluor^® 488-labeled anti-DcR2 (R&D Systems, Inc.) on a shaker at room temperature for 30 min. Cells were washed twice with 1X PBS and then resuspended in 1X PBS, the mean of the fluorescence intensity on the cell surface was measured by flow cytometry using FACSCalibur (BD Biosciences).

TRIzol^® Reagent (Thermo Fisher Scientific, Inc.) was used for RNA extraction from the cell lines, in accordance with the manufacturer's protocols. The SuperScript^® III First-Strand Synthesis System (Invitrogen Thermo Fisher Scientific, Inc.) was used for reverse transcription with 1 µg RNA (25°C for 10 min, 50°C for 50 min, 85°C for 5 min, hold at 4°C). The copy number for IL-1α, IL-6, IL-8, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was measured using RT-qPCR with Maxima SYBR^® Green/ROX qPCR Master Mix (Thermo Fisher Scientific, Inc.) and a LightCycler^® 480 System (Roche Diagnostics). The primer sequences used in the reaction were as follows: IL-1α (forward), 5′-CCG TGA GTT TCC CAG AAG AA-3′; IL-1α (reverse), 5′-ACTGCCCAAGATGAAGACCA-3′; IL-6 (forward), 5′-CATTTGTGGTTGGGTCAG G-3′; IL-6 (reverse), 5′-AGTGAGGAACAAGCCAGAGC-3′; IL-8 (forward) 5′-CAAGAGCCAGGAAGAAACCA-3′; IL-8 (reverse) 5′-AGCACTCCTTGGCAAAACTG-3′; GAPDH (forward) 5′-GCCAAGGTCATCCATGACAACT-3′; GAPDH (reverse), 5′-GAGGGGCCATCCACAGTCTT-3′ (24). The qPCR thermocycling conditions were as follows: 95°C for 5 min and then 45 cycles of 95°C for 30 sec, 55°C for 30 sec, and 72°C for 40 sec. The gene expression levels were calculated as described previously (25).

SPSS Statistics 22 (IBM Corp.) was used for statistical analysis. As aforementioned, survival analysis was estimated using the Kaplan-Meier method, and the log-rank test was conducted for survival curve comparison. A Student's t-test was used for comparing the difference of the expression levels of AURKA between two subtypes of LPS, and the expression levels of p-Aurora A in LPS cell lines before and after drug treatment. The difference in expression levels of pHisH3 (assessed by western blotting), the percentage of G2/M content (assessed by flow cytometry), percentage of cell survival (assessed by MTT assay), number of viable cells (assessed by trypan blue exclusion assay), percentage of apoptotic cells (assessed by apoptosis assay), and the senescence-related markers (SA-β-gal staining, flow cytometry of DcR2, and RT-qPCR of IL-1α, IL-6, and IL-8) between control and treated (all in triplicate) LPS cell lines were compared using one-way ANOVA, and Bonferroni test was used as post hoc test. Data was presented as the mean ± standard deviation; P<0.05 was considered to indicate a statistically significant difference.

A total of 92 arrays from GSE30929 were obtained, involving 40 cases of DD LPS and 52 cases of WD LPS (15). Microarray data were first normalized through dChip, and a total of 17,089 probes were obtained after filtering with a max/min threshold of ≥5. The expression values of LPS from GSE30929 and Singer et al (13) were then analyzed using GSEA software. Subsequently, a heatmap was generated from GESA using the top 50 genes of each phenotype (DD vs. WD LPS) in each set of tumors (Fig. S1). By comparing the top 50 genes that were upregulated in DD LPS in both sets of data, we identified 12 overlapping genes, namely nuclear and spindle associated protein 1, maternal embryonic leucine zipper kinase, AURKA, BAG family molecular chaperone regulator 2, CDC28 protein kinase regulatory subunit 2, kinesin family member 11, DNA topoisomerase II, Rac GTPase-activating protein 1, abnormal spindle-like microencephaly-associated protein, ribonuclo-side-disphosphate reductase subunit M2, retinoic acid induced 14 and ubiquitin-conjugating enzyme E2 S.

Notably, the top five gene sets enriched in DD LPS in both sets of data were all involved in cell cycle regulation (Tables I and II). Among the 12 overlapping genes, AURKA is a well-known gene involved in cell cycle regulation (26). Thus, we selected this gene for further analysis. As presented in Fig. 1, in both sets of tumors, AURKA was significantly upregulated in DD LPS in comparison with WD LPS. Survival data are available in GSE30929. We then compared the DRFS data of cases with low AURKA expression (Z-score <0) vs. those with high AURKA expression in DD LPS. As shown in Fig. 1C, among the 40 cases of DD LPS, patients with high expression of AURKA in tumors (n=27) had significantly worse DRFS than those with low expression (n=13).

AURKA-KD was performed in LPS853 cells using siRNA against AURKA with appropriate controls. As presented in Fig. 2A, silencing of AURKA was demonstrated at the protein level using anti-AURKA antibody. AURKA-KD was also associated with increased cleavage form of PARP, indicating an apoptosis-inducing effect (Fig. 2A). Furthermore, compared with the control, AURKA-KD significantly decreased number of viable cells (as determined by a trypan blue exclusion assay) in LPS853 cells (Fig. 2B). The results indicated that AURKA could act as a potential therapeutic target in LPS cells.

We used an in vitro model to determine the potential of AURKA as a therapeutic target in LPS. MLN8237 is a potent inhibitor of AURKA that reduces the activity of AURKA in a variety of cancers (9,27,28). We examined whether MLN8237 inhibits AURKA phosphorylation and induces mitotic arrest in LPS cells. As shown in Fig. 3, immunoblotting revealed significantly decreased AURKA phosphorylation levels in nocodazole-stimulated and MLN8237-treated (treated with nocodazole 400 nM for 8 h then added MLN8237 for 4 h) LPS853 and NDDLS-1 compared with the control (0 µM inhibitor) (Fig. 3A and B). In addition, the levels of pHisH3, an indicator of mitotic arrest, were significantly increased in both cell lines, in comparison with control, after treatment with MLN8237 at 0.5 µM for 8 h; LPS853 cells exhibited a significant increase in pHis3 expression in response to 1 µM MLN8237 (Fig. 3C). Flow cytometry analysis of DNA content in LPS853 and NDDLS-1 cells treated with MLN8237 demonstrated that this compound caused a significant accumulation of cells at G2/M DNA content, in comparison with control, in the two LPS cell lines (Figs. 3D and S2). These studies indicated that MLN8237 could target AURKA and induce mitotic arrest in LPS cells.

We examined whether MLN8237 could exhibit cytotoxic activity against LPS cell lines. MLN8237 exerted significant cytotoxic effects against NDDLS-1 and LPS853 cells as assessed through an MTT assay (Fig. 4A) or via a trypan blue exclusion assay (Fig. 4B). A significant induction of apoptosis in LPS cell lines was also detected by costaining with propidium iodide and FITC-labeled Annexin V (both early and late apoptotic cells are included as indicated by total cell number with positive Annexin V-FITC, Figs. 4C and S3).

As LPS exhibited some sensitivity to chemotherapy, we next explored the possible synergistic effects of treatment with both MLN8237 and chemotherapy. As presented in Figs. 5 and 6, except for cisplatin, MLN8237 exhibited significant synergistic effect (combination index<1.0) with all other chemotherapeutic agents against LPS cell lines LPS853 (Fig. 5) and NDDLS-1 (Fig. 6) at least at two different combination dose levels. These results indicate that MLN8237 can act synergistically with some chemotherapeutic agents, and provide future chances for combination therapy in LPS.

AURKA inhibition may induce senescence in cells. To examine whether senescence occurred in LPS cells following treatment with MLN8237, a SA-β-gal assay was performed in LPS853 cells after being treated with MLN8237 for 72 h. The size of LPS cells became larger and the shape went flatter after MLN8237 treatment. In addition, MLN8237 treatment significantly increased SA-β-gal activity in LPS cells compared with the control (Fig. 7A and B). The levels of pp70S6 kinase remained steady, indicating sustained mTOR activity (Fig. 7C). Furthermore, administration of MLN8237 significantly increased the expression of DcR2 (Fig. 7D), a well-known senescence biomarker, as well as the expression of IL-1α (Fig. 7E), IL-6 (Fig. 7F) and IL-8 (Fig. 7G), cyto-kines associated with the senescence-associated secretory phenotype (SASP) (29), in LPS853 cells. Of note, the levels of IL-1α and IL-8 did not significantly increase in response to 0.1 µM MLN8237. Collectively, these results demonstrate that MLN8237 treatment induces senescence in LPS cells.