The human ovarian cancer cell line A2780 was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and the authenticity was recently tested. A2780 cells were cultured in Dulbecco's modified Eagle's medium (DMEM)/H (HyClone Laboratories; GE Healthcare Life Sciences, Logan, UT, USA) with 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) and 1% penicillin/streptomycin (Beyotime Institute of Biotechnology, Shanghai, China). The primary tumor tissue mentioned in the present study was from surgical resection of one patient (female, 56 years old) categorized with stage III serous ovarian cancer in November 2017 in Hainan Branch of Chinese PLA General Hospital. The present study was approved by the Institutional Review Board of the Hainan Branch of Chinese PLA General Hospital, and informed consent was obtained from this patient.

The CD133⁺ OCSCs from the A2780 cell line were isolated by serum-free culture, as previously described (13). Briefly, after dissociation with trypsin (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA), single A2780 cells were cultured in low-attachment plates under stem cell conditions. The CD133⁺ OCSCs from patient samples were isolated by magnetic bead sorting (MACS) using the MACS system, as previously described (13). Briefly, single-cell suspensions were obtained by enzymatic dissociation of primary tumor samples, incubated with microbeads conjugated to anti-CD133/1, and sorted with a MACS column (both from Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). The sorted OCSCs were maintained in serum-free DMEM/F12 (HyClone Laboratories; GE Healthcare Life Sciences) supplemented with basic fibroblast growth factor (10 ng/ml), epithelial growth factor (20 ng/ml) (both from PeproTech, Inc., Rocky Hill, NJ, USA) and recombinant insulin (5 µg/ml; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). The sorted CD133⁻ cells and the A2780 cells were cultured in DMEM/H (HyClone Laboratories; GE Healthcare Life Sciences) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) and 1% penicillin/streptomycin (Beyotime Institute of Biotechnology).

These two patient cohorts involved in this study were derived from the GEO database in the NCBI website. The first patient cohort in Fig. 2D was obtained from GSE51373 (14), which compared the differences of gene expression between 13 patients sensitive to cisplatin and 10 patients resistant to cisplatin with stage III–IV high-grade serous ovarian cancer (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE51373). In addition, the other patient cohort in Fig. 5C was obtained from GSE13876 (15), which consisted of gene expression data from 157 patients with advanced-stage serous ovarian cancer; however, the control was not described in this study (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13876). All of the patients in the two databases mentioned in this study were diagnosed as advanced-stage serous ovarian cancer (III–IV stage), which was demonstrated by key clinical characteristics, histological type and stage. In addition, other clinical characteristics, such as age, may not be associated with gene expression and cisplatin efficacy in ovarian cancer (16).

A total of 5×10³ OCSCs were treated to obtain single cells and seeded into 96-well plates. Then, the OCSCs were treated with different concentrations (0, 10 or 20 µM) of cisplatin. After 48 h, the assay was performed by adding 10 µl of Cell Counting Kit-8 (CCK-8) solution (Beyotime Institute of Biotechnology) into each well, incubating the samples for 4 h and measuring the absorbance at 450 nm. The cell viability was calculated as follows: Cell viability = (Ab_exp-Ab_blank)/(Ab_con-Ab_blank) × 100%.

Digested A2780 cells or spheres of OCSCs were attached to coverslips or slides, respectively. After being fixed with 4% paraformaldehyde solution (Beyotime Institute of Biotechnology) for 30 min, the samples were blocked with 5% BSA (Beyotime Institute of Biotechnology) for 20 min at room temperature and incubated with mouse anti-S100B (1:200; cat. no. ab218513; Abcam, Cambridge, MA, USA) overnight at 4°C. After being washed with PBS, the samples were incubated with FITC-conjugated (1:200; cat. no. A0568; Beyotime Institute of Biotechnology) for 30 min at 37°C. After washing, the nuclei were stained with DAPI (Beyotime Institute of Biotechnology). Finally, the samples were mounted and then viewed under an Olympus confocal laser scanning microscope (Olympus FV3000; Olympus Corp., Tokyo, Japan).

Total RNA was isolated from OCSCs and non-OCSCs using TRIzol and the PureLink total RNA purification kit (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Then, reverse transcription and quantitative real-time PCR were performed with a cDNA Reverse Transcription kit and a SYBR Kit (both from Takara Biotechnology Co., Ltd., Dalian, China), respectively. The PCR amplification was performed with 40 cycles with the following protocol: Denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec, and lastly extension at 72°C for 1 min. The relative gene expression was calculated with the Cq value (17), and β-actin was the reference gene. The primers used herein were as follows: S100B, 5′-AGCTGGAGAAGGCCATGGTG-3′ (forward), and 5′-GAACTCGTGGCAGGCAGTAG-3′ (reverse); MDR-1, 5′-GCCTGGCAGCTGGAAGACAAATACACAAAATT-3′ (forward), and 5′-CAGACAGCAGCTGACAGTCCAAGAACAGGACT-3′ (reverse); MRP-1, 5′-CCTGCAGCAGAGAGGTCTTTTC-3′ (forward) and 5′-GGCATATAGGCCCTGCAGTTC-3′ (reverse); and β-actin, 5′-TGGAATCCTGTGGCATCCATGAAAC-3′ (forward) and 5′-TAAAACGCAGCTCAGTAACAGTCCG-3′ (reverse).

Total protein was extracted with RIPA buffer supplemented with 1% phenylmethanesulfonyl fluoride (PMSF), and then a BCA kit (both from Beyotime Institute of Biotechnology) was used to examine the concentrations. For western blotting, 40 µg of protein were separated by 12% SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes. After blocking with 5% BSA (Beyotime Institute of Biotechnology) for 20 min at room temperature, the following primary antibodies were used to incubate the PVDF membranes overnight at 4°C: mouse anti-S100B (1:200; cat. no. ab218513; Abcam), mouse anti-β-actin (1:500; cat. no. AA128; Beyotime Institute of Biotechnology), rabbit anti-p53 (1:500; cat. no. 10442–1-AP; ProteinTech Group, Inc., Chicago, IL, USA), rabbit anti-p-p53 (1:500; cat. no. 9284; Cell Signaling Technology, Inc., Danvers, MA, USA), rabbit-anti-p21 (1:500; cat. no. ab109199) and rabbit-anti-MDM2 (1:500; cat. no. ab38618; both from Abcam). After washing with 0.1% TBS-Tween-20 (TBST) (1 ml Triton X-80 in 1 l TBS solution), a goat anti-mouse secondary antibody (1:5,000; cat. no. A0216) and a goat anti-rabbit secondary antibody (1:5,000; cat. no. A0239; both from Beyotime Institute of Biotechnology) were used to incubate the PVDF membranes for 30 min at room temperature. Then, the images were detected with an enhanced chemiluminescence (ECL) kit (Thermo Fisher Scientific, Inc.) using the Chemilmanger^™ 5500 system (Alpha Innotech Corporation, San Leandro, CA, USA). The gray values were calculated using ImageJ software (http://rsb.info.nih.gov//), and β-actin was used as the loading control for normalization.

S100B-shRNA and NT-shRNA lentiviruses were commercially obtained (Shanghai GeneChem Co., Ltd., Shanghai, China). Briefly, a stem-loop structure oligonucleotide containing S100B or p53-target sequences was cloned under the control of the human U6 promoter in lentiviral vectors, which also contained a GFP reporter. The sequences were as follows: NT-shRNA, 5′-CAATGATGGAGACGGCGCA-3′; S100B-shRNA, 5′-CTGCCACGAGTTCTTTGAA-3′; and shp53, 5′-GACTCCAGTGGTAATCTAC-3′. The lentivirus transfection was performed according to the manufacturer's protocol. Briefly, a total of 1×10⁴ OCSCs was treated to obtain single cells and seeded into 24-well plates with 100 µl culture medium. Then, 2×10⁵ IU of lentivirus and 0.5 µl Polybrene (Shanghai GeneChem Co., Ltd.) were added into the aforementioned 24-well plates, and the multiplicity of infection (MOI) value was 20. After 12 h, the medium containing the lentivirus was replaced with normal medium.

OCSCs transfected with shNT or shS100B were treated with cisplatin (10 µM) for 48 h. For the treatment of OCSCs with pifithrin-α (a p53 reversible inhibitor; Calbiochem; EMD Millipore, Billerica, MA, USA), the cells were treated with 20 µM pifithrin-α 4 h prior to cisplatin exposure. In addition, the cells were treated to obtain single cells and incubated with PE-conjugated Annexin V antibody and PI (1:500; 556570; BD Biosciences, Franklin Lakes, NJ, USA) at room temperature for 15 min. The cells were washed with phosphate-buffered saline (PBS) twice and determined by a flow cytometer (BD Biosciences). The data were analyzed by FlowJo software, and the Annexin V⁺PI⁺ OCSCs were recognized as ‘apoptotic cells’, and the Annexin V⁻PI⁻ OCSCs as ‘live cells’.

All quantitative data were expressed as the mean ± SEM. The independent samples t-test, ANOVA, Tukey's multiple comparison test and Pearson's correlation analysis were used for the statistical analysis. Differences were considered statistically significant when P<0.05. All statistical analyses were carried out with SPSS 18.0 software (SPSS, Inc., Chicago, IL, USA).

In our previous study, we isolated CD133⁺ OCSCs from A2780 cells and primary tumor tissues, which were demonstrated to harbor self-renewal properties and tumorigenicity (18). In the present study, we performed a CCK-8 assay to further examine the chemoresistance ability of CD133⁺ OCSCs and CD133⁻ non-OCSCs, which were obtained from A2780 cells and primary tumor tissues as previously described (13). After 48 h of treatment with different concentrations of cisplatin, the cell viability of OCSCs was significantly higher than that of non-OCSCs (Fig. 1A and B). Therefore, OCSCs were more resistant to cisplatin than were non-OCSCs.

Herein, we first verified the expression of S100B in OCSCs. Immunofluorescence staining revealed that S100B expression was higher in A2780-derived OCSCs than in non-OCSCs A2780 cells (Fig. 2A). In addition, S100B mRNA expression was much higher in OCSCs when compared with that in non-OCSCs (Fig. 2B). Moreover, S100B expression was also found to be 3–5-fold higher in A2780- and patient-derived OCSCs at the protein level (Fig. 2C). These findings revealed that S100B expression is higher in OCSCs. To further explore the relationship between S100B expression and chemoresistance in ovarian cancer patients, we analyzed S100B expression in a cohort containing 13 cisplatin-sensitive ovarian cancer samples and 10 cisplatin-resistant samples (GSE51373, high-grade serous ovarian cancer, III–IV stage). We found that S100B expression was significantly higher in cisplatin-resistant patients than in cisplatin-sensitive patients (Fig. 2D), which implied that S100B may be involved in chemoresistance to cisplatin in ovarian cancer patients.

To further investigate the role of S100B in OCSC resistance to cisplatin, we knocked down S100B in A2780- and patient-derived OCSCs by transfecting them with S100B shRNA. We found that S100B was clearly knocked down at both the mRNA and protein levels after S100B shRNA transfection (Fig. 3A and B). A CCK-8 assay revealed that S100B silencing led to a significant reduction of cell viability in A2780- and patient-derived OCSCs after treatment with different concentrations of cisplatin for 48 h (Fig. 3C and E). To further examine the effects of S100B knockdown on OCSC chemoresistance, flow cytometry was performed to determine the apoptosis of OCSCs after treatment with cisplatin. We found that the apoptosis of A2780-derived OCSCs transfected with shNT was not affected after treatment with cisplatin. However, S100B knockdown significantly increased the apoptosis of OCSCs induced by cisplatin (Fig. 3D). Moreover, the same results were obtained in patient-derived OCSCs when S100B was knocked down (Fig. 3F). These data revealed that S100B knockdown attenuated the chemoresistance of OCSCs to cisplatin.

It has been reported that S100B can inhibit p53 activation in many cancers. Moreover, p53 is a well-known tumor suppressor that can induce cell apoptosis. Thus, we speculated that S100B may promote OCSC chemoresistance through inhibition of p53. Clearly, S100B knockdown led to elevated expression of p53 and phosphorylated-p53 (p-p53) (Fig. 4A and B). In addition, S100B knockdown also significantly elevated the protein levels of two of p53 downstream effector genes, p21 and MDM2 (mouse double minute 2 homolog gene) (9), which further ascertained the activation of p53 (Fig. 4A and B). Furthermore, p53 knockdown could partially reverse the decreased chemoresistance caused by S100B silencing in A2780-derived OCSCs (Fig. 4C). In addition, we found that the increased apoptosis of OCSCs that was induced by S100B knockdown was also significantly reversed in patient-derived OCSCs when p53 was inhibited (Fig. 4D and E). These results indicated that the effect of S100B on OCSC chemoresistance may be mediated by a p53-dependent pathway.

Chemoresistance in various types of cancers has been demonstrated to be associated with the overexpression of multidrug resistance gene-1 (MDR1) and MDR-associated protein-1 (MRP1). In addition, p53 regulated MDR1 and MRP1 by modulating the activity of their promoters. Thus, we hypothesized that S100B promotes OCSC chemoresistance by regulating MDR1 and MRP1 expression via p53. To confirm our hypothesis, we first examined the expression of these two genes in patient-derived OCSCs when S100B was knocked down. We found that S100B silencing led to significantly decreased expression of MDR1 and MRP1 (Fig. 5A). Moreover, p53 knockdown partially rescued the decreases induced by S100B silencing in A2780-derived OCSCs (Fig. 5B). These data indicated that S100B may promote chemoresistance by indirectly regulating MDR1 and MRP1 expression.

To further study the influence of S100B on MDR1/MRP1 expression, we analyzed the correlations of S100B and MDR1/MRP1 expression in a cohort consisting of 157 ovarian cancer patients from the GEO database (GSE13876). Notably, S100B expression was positively correlated with MDR1 and MRP1 expression (Fig. 5C). These results further confirmed that S100B promoted MDR1 and MRP1 expression in ovarian cancer.