Integrative transcriptomics analysis for uterine leiomyosarcoma identifies aberrant activation of cell cycle-dependent kinases and their potential therapeutic significance

Patients

Archival fresh-frozen tumor samples stored at the National Cancer Center Biobank (Tokyo, Japan) were used. Since 2011, there have been six patients with ULMS who underwent surgery without neoadjuvant therapy. The sarcoma and adjacent myometrium tissues of the six patients were obtained. Moreover, three patients with benign leiomyoma were included as controls. The study protocol was approved by the ethics committee at our institution (approval No. 2020-160). We obtained written informed consent from all patients.

RNA extraction and transcriptome analysis

Total RNA was extracted from six ULMS and three myoma samples using the miRNeasy Mini Kit (Qiagen, Hilden, Germany), and pair-end sequencing was carried out using a DNBSEQ-G400 (MGI Tech, Shenzhen, China) by Azenta (South Plainfield, NJ). From the sequencing data, expression levels for each gene were quantified by Kallisto (16). Then, the data were summarized using the tximport package (ver. 1.18.0) of R software (ver. 4.0.3) and Rstudio (RStudio, Boston, MA), and scaledTPM counts were used for further analysis. Excluding genes with low read coverage (maximum read count: < 100 reads), 3,070 differentially expressed genes (DEGs, |log2FC| > 1) between the ULMS and myoma samples were used for a heatmap analysis. The heatmap.2 function of the gplots package (ver. 3.1.0) was used after the data were converted to base 10 logarithms and z-scores. For volcano plots, the adjusted P values for each gene were calculated by the Wald test in DESeq2 (ver. 1.30.0) using data for 23,353 annotated genes. Subsequently, we performed pathway and upstream regulator analysis by Ingenuity Pathway Analysis (IPA, Qiagen) using the significant DEGs identified on the volcano plot.

NCBI GEO dataset

The three datasets, GSE36610 (12 ULMS and ten myometrium samples), GSE64763 (25 ULMS, 25 myomas, and 29 myometrium samples), and GSE68295 (three ULMS, three myomas, and three myometrium samples), were downloaded from the NCBI GEO database. The three datasets were microarray-based transcriptional profiles, and 10,641 common gene symbols were used for analysis (Supplementary Fig. S1A). The expression data were converted to z-scores, and 1,683 DEGs (the difference between the mean z-scores of ULMS and myometrium is greater than 1) were used to generate the heatmap and principal component analysis (PCA). The PCA was visualized using the prcomp and plot3d functions of the rgl package (ver. 0.100.54). For volcano plot, log2FC and adjusted P values for each gene were calculated for each dataset. Then, pathway analysis was performed by IPA using the common DEGs identified on the volcano plots.

Cell lines

Three ULMS-derived cell lines, SKN, SK-UT-1, and SK-LMS-1, were used. SKN was purchased from the Japanese Cancer Research Resources Bank (Osaka, Japan), and SK-UT-1 and SK-LMS-1 were purchased from the American Type Culture Collection (Manassas, VA). SKN cells were maintained in Ham’s F12 medium (Sigma-Aldrich, St. Louis, MO) containing 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA) and antibiotics. SK-UT-1 and SK-LMS-1 cells were maintained in MEM (Nacalai Tesque, Kyoto, Japan) containing 10% fetal bovine serum, 1 mM sodium pyruvate (Thermo Fisher Scientific), and antibiotics. The cell lines tested negative for mycoplasma contamination and were used between 5 and 40 passages for experiments.

Chemicals

All selective kinase inhibitors were purchased from Selleck (Houston, TX), and their putative targets are shown in Supplementally Table S1. Briefly, BI 2536 and volasertib (BI 6727) are PLK1 inhibitors. Prexasertib HCl (LY2606368) and PF-477736 are ATP-competitive CHEK1/2 inhibitors. Dinaciclib (SCH727965) and flavopiridol (Alvocidib) are pan-CDK inhibitors. AT9283 is a JAK2/3 and Aurora A/B inhibitor, and tozasertib (MK-0457) is a pan-Aurora inhibitor. JNJ-7706621 is a pan-CDK and potent Aurora A/B inhibitor, and BAY 1217389 is a TTK inhibitor. Pazopanib is an approved multi-target inhibitor. The drugs were dissolved in DMSO as stock solutions and further diluted in the culture medium for experiments. Moreover, cisplatin was purchased from Nichi-Iko Pharmaceutical (Toyama, Japan).

Small interfering RNAs (siRNAs)

Silencer Select Pre-designed siRNAs for each gene (Thermo Fisher Scientific) were used, and the assay IDs were as follows; s448 (siPLK1 No. 1), s449 (siPLK1 No. 2), s503 (siCHEK1 No. 1), s504 (siCHEK1 No. 2), s22119 (siCHEK2 No. 1), and s22121 (siCHEK2 No. 2). Cells were transfected with 3 nM siRNA using Lipofectamine RNAi Max (Thermo Fisher Scientific).

Cell viability assay

Cells were seeded into 96-well plates. Immediately after attachment, the cells were treated with the inhibitors and incubated for 72 h. For siRNAs, cells were transfected with the siRNAs and incubated for 24, 48, and 72 h. Cell viability was assessed using the CellTiter-Glo 2.0 Cell Viability Assay (Promega, Madison, WI), and the luminescence measurements were taken 10 min after adding the reagent using a microplate reader (Molecular Devices, San Jose, CA). Viability was calculated with the percentage of untreated cells, and experiments were performed in triplicate and repeated three times. IC50 and drug dose-response curves were calculated in GraphPad Prism 7 (Version 7.0d, GraphPad Software, San Diego, CA). Synergy was determined using CompuSyn software (version 1.0) (http://www.combosyn.com/index.html).

Cell cycle assay

Cells were seeded in 6-well plates and grown to approximately 80% confluency. Then, cells were treated with selective inhibitors for 24 h. The cells were harvested following trypsinization, washed with 3% FBS/PBS, and fixed in cold 70% ethanol. The cells were then resuspended in 3% FBS/PBS and stained with ReadiDrop Propidium Iodide (Bio-Rad Laboratories, Hercules, CA). Cell analyzer EC800 (Sony, Tokyo, Japan) was used for the analysis. The resulting data were analyzed with FlowJo software (BD Biosciences). Experiments were performed in triplicate and repeated three times.

qRT-PCR

Paired ULMS and adjacent normal tissues were used, and total RNA was extracted as described previously. Total RNA was extracted from transfected cells, and cDNA was synthesized using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). The QuantiTect SYBR Green PCR Kit (Qiagen) or THUNDERBIRD SYBR qPCR Mix (Toyobo, Osaka, Japan) were used. Specific primers were synthesized by Fasmac (Kanagawa, Japan), and the primer sequences are shown in Supplementary Table S2. The amplification program was as follows: denaturation at 95°C for 10 min, followed by 40 amplification cycles of 95°C for 15 s and 60°C for 60 s. The amplified product was monitored by measuring SYBR Green I dye fluorescence intensity, and β-actin was used as a reference gene to normalize expression.

Western blot analysis

Cells were treated with prexasertib for 16 h, and then, total protein extracts were prepared using the M-PER Mammalian Protein Extraction Reagent (Thermo Fisher Scientific) containing the Halt Protease & Phosphatase Inhibitor Single-Use Cocktail (Thermo Fisher Scientific). After quantification, 10 μg of total protein was separated on Mini-PROTEAN TGX gels (4%–20%, Bio-Rad Laboratories) and transferred to PVDF membranes. The following primary antibodies were used: Chk1 (2G1D5) Mouse mAb #2360 (Cell Signaling Technology), Phospho-Chk1 (Ser296) Antibody #2349 (Cell Signaling Technology), and β-actin (C4) (MAB1501, Merck, Germany). Horseradish peroxidase-conjugated anti-mouse IgG (NA931) and horseradish peroxidase-conjugated anti-rabbit IgG (NA934) were purchased from GE Healthcare (Buckinghamshire, UK). Protein bands were visualized using ImmunoStar LD (Fujifilm Wako Pure Chemical, Osaka, Japan) and ImageQuant LAS-4000 (Fujifilm, Tokyo, Japan).

Immunofluorescence

After incubation with a prexasertib-containing medium for 24 h, cells were fixed with 4% paraformaldehyde. The cells were then treated with 0.3% Triton-X/Blocking One solution (Nacalai Tesque). Phospho-Histone H2A.X (Ser139) Antibody #2577 (Cell Signaling Technology, Danvers, MA), Alexa-Fluor 488 Goat anti-Rabbit IgG (H+L) Cross-Absorbed Secondary Antibody (Thermo Fisher Scientific), and Hoechst 33342, trihydrochloride, trihydrate (Thermo Fisher Scientific) were used for detection and staining, and images were captured using a BZ-X700 fluorescence microscope (Keyence, Osaka, Japan).

Animal studies

All mouse experiments were approved by the National Cancer Center Research Institute, Institute of Laboratory Animal Research (Number: T18-009). Four-week-old female BALB/C nude mice were used for the animal experiments, and 3.0 × 10⁶ SK-UT-1 cells were injected into the right flank of the mice. BI 2536 was dissolved in hydrochloric acid (0.1N) and diluted with 0.9% NaCl. Prexasertib was dissolved in a vehicle (5% DMSO + 40% PEG 300 + 5% Tween80 + ddH2O), and cisplatin was diluted with 0.9% NaCl. The drugs were administered intraperitoneally twice a week. Mice were monitored carefully, and tumor volume was calculated using the modified ellipsoid formula (Length × Width² × 0.5). Mice were sacrificed when the tumors reached a volume of 2,000 mm³.

Statistical analysis

Statistical analysis was performed with RStudio and R software (ver. 4.0.3). Welch’s t-test was used to determine the significance of differences between the means of two sets of data. Paired t-test was used to determine the significance of differences between the paired LMS and myometrium samples. Dunnett’s test was used for multiple comparisons with a control group using the multicomp package (ver. 1.4-17). Kaplan-Meier curves and a log-rank test were used for the analysis of the survival. A P value of less than 0.05 was considered statistically significant.

Data availability statement

The data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE185543.

Transcriptome analysis of clinical samples

Transcriptome analysis was performed using six ULMS and three myoma samples. The median age of the patients with ULMS was 59.5 (range, 53–79) years and all patients underwent surgery without neoadjuvant therapy. The heatmap showed that the gene expression profile of ULMS was quite different from that of myoma (Fig. 1A). The expression of 23,353 genes was compared by multivariate analysis. There were 387 significantly upregulated and 125 significantly downregulated genes in ULMS based on a cut-off of |log2FC| > 1 and an adjusted P-value < 0.05 (Fig. 1B and Supplementary Table S3 & S4). To assess the putative function of the 512 DEGs, pathway analysis was performed using the IPA software, which revealed that several pathways associated with the cell cycle and DNA damage checkpoint were significantly dysregulated. For example, these included “Kinetochore Metaphase Signaling Pathway (P = 5.01E-24),” “Mitotic Roles of Polo-Like Kinase (P = 1.58E-11),” and “Cell Cycle: G2/M DNA Damage Checkpoint Regulation (P = 2.51E-7)” (Fig.1C). In addition, upstream regulator analysis using IPA revealed that the function of CDK1, AURKB, PLK1, CHEK2, CHEK1, CDK2, and PRKDC was significantly activated in ULMS (Fig.1C and Table 1).

• Download figure
• Open in new tab

Fig. 1. The transcriptome analysis of uterine leiomyosarcomas (LMSs) and myomas

(A) The hierarchical clustering and heatmap showing 3,070 differentially expressed genes (DEGs) between the ULMSs and myomas. The DEGs were defined as an absolute log2 fold change exceeding 1. (B) The volcano plot showing significant DEGs between LMSs and myomas. The adjusted P values for each gene were calculated by the Wald test in DESeq2. (C) The top ten significantly dysregulated pathways and the graphical summary based on Ingenuity Pathway Analysis (IPA) for the significant DEGs. The orange and blue nodes represent the activated and inhibited genes or pathways, respectively. The orange arrows and blue inhibitory arrows indicate activation and suppression, respectively. (D) The principal component analysis (E) The hierarchical clustering and heatmap for GSE36610, GSE64763, and GSE68295 dataset. Each data was converted to z-scores and merged. Comparing the mean expression of ULMS and myometrium, 1,683 DEGs were used for analyses. (F) The Venn diagram showing the significantly upregulated and downregulated genes between ULMS and myometrium in each dataset. The name of seven predictively activated upstream regulators in ULMS is shown. (G) The top ten significantly dysregulated pathways for the three datasets. The IPA analysis was performed using the 282 significant DEGs in common.

Then, to validate our results, we used three GEO datasets, which were made up of the expression of 10,641 genes in 40 ULMS, 28 myomas, and 42 normal myometrium samples (Supplementary Fig. S1A). The PCA and heatmap analysis showed that the transcriptional profile of ULMS is different from that of myoma and normal myometrium (Fig.1D&1E). Then, comparing ULMS and myometrium samples in each dataset, we identified that 236 and 45 genes were commonly upregulated and downregulated in ULMS, respectively (Fig. 1F and Supplementary Fig. S1A). Moreover, the expression of the seven upstream regulators, which were identified in our cohorts, were confirmed to be upregulated in ULMS (Fig. 1F). Then, the IPA analysis using 281 DEGs validated the activation of “Kinetochore Metaphase Signaling Pathway (P = 2.58E-18)” and inhibition of “Cell Cycle: G2/M DNA Damage Checkpoint Regulation (P = 2.82E-10)” (Fig. 1G). Therefore, aberrant cell cycle regulation would be a dominant transcriptional feature of ULMS.

In vitro screening of selective inhibitors for the activated upstream regulators

We considered the upregulated and activated key regulators as potential therapeutic targets for ULMS. Hence, the anti-cancer effects of selective inhibitors for these genes were evaluated using three cell lines derived from ULMS. First, we evaluated the efficacy of pazopanib, which is an approved drug for malignant soft-tissue tumors. Cells were treated with pazopanib for 72 h, and the IC50 values for SK-UT-1, SK-LMS-1, and SKN cells were 30.7, 62.8, and 5.7 μM, respectively (Supplementary Fig. S2A). Then, we assessed the efficacy of selective inhibitors for the target genes, and most of them were highly effective compared with pazopanib (Fig. 2A&2B and Supplementary Fig. S2B). The inhibitors for PLK1 (BI 2536 and volasertib) exhibited cytotoxicity at a lower nanomolar concentration in SK-UT-1 and SK-LMS-1 cells (Fig. 2A). Moreover, CHEK1/2 inhibitor (prexasertib and PF-477736) also showed a higher sensitivity in SK-UT-1 cells, with an IC50 below 10 nM (Fig. 2B). On the other hand, both BI 2536 and prexasertib were less effective in SKN cells compared with SK-UT-1 cells, although their effect was at least ten times greater than that of pazopanib.

• Download figure
• Open in new tab

Fig. 2. The inhibitory effects of PLK1 or CHEK1/2 inhibitors

(A) The effect of PLK1 inhibitors; BI 2536 and volasertib. (B) The effect of CHEK1/2 inhibitors; prexasertib and PF-477736. Cells were treated with each inhibitor for 72 h. Red, green, and blue represent SK-UT-1, SK-LMS-1, and SKN, respectively. Experiments were performed in triplicate and repeated three times, and IC50 and drug dose-response curves were calculated in GraphPad Prism 7. (C) Cell-cycle distribution of BI 2536-treated cells. (D) Cell-cycle distribution of prexasertib-treated cells. Cells were treated with each concentration of inhibitor for 24 h. Cell-cycle distribution was calculated by FlowJo, and the percentage of cells was compared using Dunnett’s test. Error bars represent standard errors of the mean.

Then, we evaluated the effect of BI 2536 and prexasertib on the cell cycle. In the 24 hour-treatment, 10 nM BI 2536 showed little effect, but 100 nM BI 2536 considerably decreased the cell population in the G1 phase and increased that in the S and G2/M phase (Fig. 2C). Similarly, when treated with 100 nM prexasertib, the cell population in the G1 phase was considerably decreased, whereas that in the S and G2/M phases was considerably increased in all cell lines (Fig. 2D). In particular, SK-UT-1 cells were highly sensitive to prexasertib, and 10 nM prexasertib was enough to induce cell cycle arrest.

The effect of PLK1 inhibition

According to the results of integrative transcriptome analysis and drug screening, PLK1 is the most attractive therapeutic target. In our cohort, the expression of PLK1 was significantly increased in ULMS compared with adjacent normal myometrium (P < 0.001, Fig. 3A). Thus, to assess the role of PLK1 in ULMS, we performed gene silencing experiments using siRNAs. Two siRNAs for PLK1 reduced the expression of PLK1 to about 30–40% and increased the number of the round shape cells (Fig. 3B&3C). The cell cycle analysis showed that siRNAs for PLK1 considerably decreased the cell population in the G1 phase and increased that in the S and G2/M phase in all cell lines (Fig. 3D). Therefore, PLK1 knockdown significantly decreased the cell proliferation (in SK-UT-1, SK-LMS-1, and SKN; P < 0.01, P < 0.001, and P < 0.01, respectively, Fig. 3E). In particular, the effect of 3 nM siRNA was almost the same as that of 100 nM BI 2536 and caused complete growth arrest in SK-LMS-1 cells (Fig. 3D&3E).

• Download figure
• Open in new tab

Fig. 3. The effects of PLK1 silencing

(A) The relative expression of PLK1 in paired ULMS and myometrium. The relative expression was compared using paired t-test. (B) Validation of PLK1 suppression following transfection with 3 nM of siRNA for PLK1 (siPLK1). (C) The representative images of siPLK1 transfected cells. Scale bars show 100 μm. (D) Cell-cycle distribution of siPLK1 transfected cells. Cell-cycle distribution was calculated by FlowJo, and the percentage of cells was compared using Dunnett’s test. (E) The proliferation of siPLK1 transfected cells. Cell viability was measured at 24, 48, and 72 h, and the luminescence was compared using Dunnett’s test. Error bars represent standard errors of the mean, *P < 0.05, **P < 0.01, and ***P < 0.001.

The effect of CHEK1 inhibition

In addition to PLK1 inhibition, CHEK1/2 inhibition is also a promising therapeutic strategy to impair DNA damage response. Both CHEK1 and CHEK2 were upregulated in ULMS compared with adjacent normal myometrium (P < 0.01 and P < 0.01). The fold change of CHEK1 is larger than that of CHEK2 (Fig. 4A). The siRNA-mediated downregulation of CHEK1 and CHEK2 was confirmed by qRT-PCR, but the effect of siRNAs was different depending on the cell types (Fig. 4B). In SK-UT-1 cells, siRNAs for CHEK1 significantly inhibited cell proliferation (P < 0.001), whereas siRNAs for CHEK2 showed no effect on proliferation (Fig. 4C). In SK-LMS-1 and SKN cells, siRNAs for both CHEK1 and CHEK2 slightly but significantly inhibited the proliferation (Fig. 4C). Moreover, prexasertib decreased the expression of pCHEK1(Ser296) in a dose-dependent manner (Fig. 4D). Therefore, CHEK1 was more responsible for the prexasertib-induced growth arrest.

• Download figure
• Open in new tab

Fig. 4. The effects of CHEK1 silencing and prexasertib

(A) The relative expression of CHEK1 and CHEK2 in paired ULMS and myometrium. The relative expression was compared using paired t-test. (B) Validation of CHEK1 or CHEK2 suppression following transfection with 3 nM of siRNA for CHEK1 (siCHEK1) or CHEK2 (siCHEK2). (C) The proliferation of siCHEK1 transfected cells. Cell viability was measured at 24, 48, and 72 h, and luminescence was compared using Dunnett’s test. (D) The expression of CHEK1 and pCHEK1(Ser296) protein in prexasertib-treated cells. Cells were treated with each concentration of prexasertib for 16 h. (E) Immunofluorescent of γH2AX in prexasertib-treated cells. Cells were treated with 0 or 100 nM prexasertib for 24 h, and the percentage of γH2AX cells was determined. Green and blue colors indicate γH2AX and Hoechst, respectively, and scale bars represent 10 μm. The percentage of γH2AX-positive cells was compared using Welch’s t-test. (F) The combination effect of prexasertib and cisplatin. Cells were treated with each drug concentration for 72 h, and the percentage of growth inhibition is shown relative to untreated controls. Drug synergy was analyzed using CompuSyn software. Error bars represent standard errors of the mean, *P < 0.05, **P < 0.01, and ***P < 0.001.

To confirm the prexasertib-induced DNA damage, immunocytochemistry for phospho-H2AX was performed. In SK-UT-1 and SKN cells, a 24-hour exposure to 100 nM prexasertib caused structural abnormalities in the nucleus (Fig. 4E). Moreover, in all cell lines, the percentage of γH2AX-positive cells was significantly increased by prexasertib treatment (in SK-UT-1, SK-LMS-1, and SKN; P < 0.001, P < 0.001, and P < 0.05, respectively, Fig. 4E). In addition, increasing DNA damage is expected to enhance the effect of prexasertib, and therefore, we assessed the combination effect of prexasertib and cisplatin. Cells were treated with a combination of various concentrations of prexasertib and cisplatin, and the synergy was determined using CompuSyn software. As a result, cisplatin synergistically enhanced the effect of prexasertib (Fig. 4F).

In vivo efficacy of PLK1 and CHEK1 inhibition

Finally, we investigated the in vivo efficacy of the inhibitors. SK-UT-1 tumor-bearing mice were treated with either BI 2536 (20 mg/kg or 30 mg/kg) or saline for two weeks after implantation. The mice treated with BI 2536 monotherapy exhibited marked tumor regression (P < 0.001, Fig. 5A and Supplementary Fig. S3A). The toxicity was well tolerated, and no mice died due to treatment. When the tumor volume of the control mice reached 2,000 mm³, all the mice were sacrificed. The mean tumor weight of the high-dose of BI 2536, low-dose of BI 2536, and control groups were 0.53 g, 0.93 g, and 2.24 g, respectively. Hence, BI 2536 monotherapy significantly decreased the tumor weight in a dose-dependent manner (low-dose and high-dose; P < 0.01 and P < 0.01, respectively, Fig. 5B).

• Download figure
• Open in new tab

Fig. 5. In vivo efficacy of BI 2536 or prexasertib

(A) Estimated tumor volume of SK-UT-1 tumor-bearing mice treated with either BI 2536 monotherapy or saline (n = 6 per group). High-dose (30 mg/kg), low-dose (20 mg/kg) of BI 2536, or saline was intraperitoneally administered twice a week for four weeks. (B) The representative images of tumors and the mean tumor volume of SK-UT-1 tumor-bearing mice treated with either BI 2536 or saline. The mice were sacrificed when the tumors of the control mice reached a volume of 2,000 mm³. (C) Estimated tumor volume and Kaplan-Meier plot of SK-UT-1 tumor-bearing mice treated with prexasertib and cisplatin in combination (n = 7 per group). Prexasertib (3 mg/kg), cisplatin (3 mg/kg), the combination of prexasertib (3 mg/kg) plus cisplatin (3 mg/kg), or vehicle (DMSO) was intraperitoneally administered twice a week for four weeks. The mice were sacrificed when the tumors reached a volume of 2,000 mm³. The tumor volume and weight were compared using Welch’s t-test, and survival was compared by a log-rank test. (D) The representative images of tumors and the mean tumor volume of SK-UT-1 tumor-bearing mice treated with The prexasertib and cisplatin combination therapy. The tumor weight was compared using Dunnett’s test. The scale bars represent 1 cm. The error bars represent thr standard errors of the mean. *P < 0.05, **P < 0.01, and ***P < 0.001 (compared with control mice). †P < 0.05 and ††P < 0.01 (compared with cisplatin-treated mice), N.S., not significant.

Combination therapy is an effective strategy for increasing therapeutic effects and reducing adverse events. We assessed the anti-cancer effect of the combination of prexasertib (3 mg/kg) with cisplatin (3 mg/kg). The prexasertib monotherapy showed no considerable tumor regression effect, whereas cisplatin monotherapy significantly prolonged survival compared with the DMSO treatment (P < 0.001, Fig. 5C). When combined with prexasertib and cisplatin, the anti-cancer effect was enhanced, and marked growth inhibition was observed. Thus, compared with DMSO treatment, the combination therapy significantly reduced the tumor volume (P < 0.05) and prolonged survival (P < 0.001, Fig. 5C and Supplementary Fig. S3B). In the combination therapy group, the tumors of three mice did not reach a volume of 2,000 mm³ within 12 weeks. The tumor weights of sacrificed mice were not significantly different between the groups, indicating that the experiment was conducted fairly (Fig. 5C). Moreover, the survival period in the combination group was significantly longer compared with that in the cisplatin monotherapy group (p < 0.05, Fig. 5C&5D). Therefore, BI 2536 monotherapy and prexasertib plus cisplatin combination therapy caused significant tumor regression in a ULMS mouse model.