Sequential combination of cisplatin with eugenol targets ovarian cancer stem cells through the Notch-Hes1 signalling pathway

SKOV3 and OV2774 (also known as MDAH-2774) cell lines were originally obtained from American Type Cell Culture (ATCC, Manassas, VA). Cells were grown in DMEM/F2 (1:1) (Gibco, MD, USA) containing 10% FBS and 1% antibiotics at 37 °C in a 5% CO2 incubator. Cells were regularly screened for mycoplasma contamination using MycoAlert Mycoplasma Detection Kits (Lonza).

Five weeks female Nu/J mice were used for in vivo experiments. The animal study was approved by the King Faisal Specialist Hospital and Research Centre and the animal study protocols were approved by the animal use committee (protocol/RAC#2170 034). Animals were injected with 1 × 10⁶ cells/mouse into the dorsal flank (10/group) and allowed to develop tumors. Once palpable, tumors were identified, mice were given intramuscular injection daily with eugenol (cat # E51719; Sigma, MO, USA) (50 mg/Kg), cisplatin (cat # 1134357, Sigma, MO, USA) (2 mg/Kg) and combination of both drugs for 21 days. Tumor growth and mouse body weight was recorded every alternate day. Mice were sacrificed at day 22, tumor samples were collected, frozen immediately for histological analysis or dissociated tumor cells were used for further in vitro and in vivo assays.

Please see Additional file 7: materials and methods for detailed description.

OV2774 and SKOV3 cells were transfected by dHes1d2-driven EGFP reporter plasmid [15] and cultured in the DMEM/F12 (1:1) complete medium. Hes1GFP⁺ and GFP⁻ cells were sorted using FACSAria cell sorter (BD Bioscience, USA). Hes1GFP⁺ cells were maintained and expanded as spheres in an ultra-low attachment 6-well plates (BD Bioscience), supplemented with 2% B27 supplements (Invitrogen, Carlsbad, CA), 20 ng/ml of epidermal growth factor (EGF, Sigma, St Louis, MD), 20 ng/ml basic fibroblast growth factor (bFGF, Sigma), 50 ng/ml hydrocortisone (Sigma), 2 μg/ml insulin, 4 μg/ml heparin (Stem Cell Technologies, BC, Canada) for 20 days. List of antibodies for flow cytometry is in the Additional file 8: Table S1.

FACS sorted OV2774 and SKOV3 cells were cultured in 6-well ultra-low attachment plates at a density of 5000 viable cells/well, supplemented with 0.4% BSA, 1% penicillin and streptomycin, B27, 20 ng/ml hEGF, 5 μg/ml insulin, 20 ng/ml FGF, 50 ng/ml hydrocortisone and 4 μg/ml heparin. Spheres were treated with cisplatin, eugenol and the combination of both. The number and size of spheres were viewed under microscope and counted the number of spheres every 3 days. To assess the effects of the drugs on secondary and tertiary sphere formation, cells were grown in the absence of drugs.

For statistical analysis and graphing we used the R-statistical software, (version 3.4.4). R packages: DRC for cell survival and cytotoxicity; ggplot2 for all graphing, CORRPLOT for correlation matrix, STATMOD for Extreme Limiting Dilution Assay (ELDA); ‘Survival’ and ‘survminer 0.3.0’ for tumor free survival analysis. All data presented in this study as the mean +/−SD. Triplicate results were analyzed by two-tailed Student t test. A p value of < 0.05 considered statistically significant. For multiple comparisons, 1 and 2-way analysis of variance (ANOVA) was performed using the R- Statistical software “ggplo2” package.

Note: Additional information is described in the Additional file 7: Materials and methods section.

First, we have tested the growth inhibitory effects of different concentrations of cisplatin and eugenol on two human ovarian cancer cell lines (OV2774; also known as MDAH-2774 and SKOV3) for different periods of time using the WST-1 assay. SKOV3 and OV2774 ovarian cell lines were extensively characterized previously [16, 17]. The growth inhibitory effect of cisplatin and eugenol alone were time- and concentration-dependent for both cell lines (Additional file 1: Figure S1A). The highest growth inhibition was observed by 72 h at 40 μM for cisplatin and 4 μM for eugenol (Additional file 1: Figure S1A). We then investigated the dose response of the combination of both drugs in two drug administration sequences, a) cisplatin (5, 10, 20, 30 and 40 μM) alone for 24 h followed by additional 48 h with eugenol (0.5, 1, 2, 3 and 4 μM) and, b) eugenol alone for 24 h followed by additional 48 h with cisplatin and cellular cytotoxicity and quantitative values of drug interaction combination index (CI) were determined using the method developed by Chou, 2006 [18]. In the sequence (a), the CI ranged from 0.971 to 0.081 for OV2774 cells and 0.956 to 0.183 for SKOV3 cells (Fig. 1a, Additional file 1: Figure S1B, Additional file 8: Tables S1A, S1B). In the sequence (b), the CI values for OV2774 cells was 0.834 for the combination doses of cisplatin 5 μM/eugenol 0.5 μM, and 1.192 for the combination doses cisplatin 20 μM/eugenol 2 μM. For SKOV3 cells, CI values ranged from 0.717 to 1.212 (Fig. 1a, Additional file 8: Table S1A, S1B). In the sequence (b), the CI values started to decline only at higher doses (cisplatin 30 μM)/eugenol 3 μM) and (cisplatin 40 μM/(eugenol 4 μM) (Fig. 1a, Additional file 1: Figure S1B, Additional file 8: Table S1B). These findings suggest that adding eugenol first at low concentrations generated antagonistic effects of the drugs, while adding cisplatin first followed by eugenol showed strong synergism.

To further test the effects of the combination, we have chosen two low-doses of cisplatin (5 and 10 μM), which killed between 20 and 30% cells (Additional file 1: Figure S1B). Therefore, when we combined cisplatin (5 and 10 μM) with a low dose of eugenol (1 μM), the proportion of cell death in OV2774 cells reached to 50 and 70% in a combination sequence (a) (Fig. 1b, top left panel; column 5 and 8 in bar graph). In contrast, when cells were treated with eugenol first followed by cisplatin (combination sequence b), cell death was 40 and 50% (Fig. 1b, top left panel; columns 4 and 7 in bar graph). Using similar combination sequence approach, we then tested if 2 μM of eugenol can further enhance the cell death with low-doses of cisplatin. In the sequence (b), when 2 μM of eugenol was added to 5 and 10 μM of cisplatin, cell death was enhanced to 51 and 62%, respectively (Fig. 1b, top right panel columns 4 and 7 in the bar graph). The percentage of cell death was further enhanced to 75 and 95% when cells were first treated with cisplatin followed by eugenol [combination sequence (a)] (Fig. 1b, top right panel; columns 5 and 8 in the bar graph). A similar response was seen with SKOV3 cells. Briefly, when cisplatin (5 and 10 μM) was combined with low-dose eugenol (1 μM), the proportion of cell death was 48 and 68% in a combination sequence (a) (Fig. 1b, bottom left panel, column 5 and 8 in the bar graph). However, in a combination sequence (b), the cell death was 35 and 52% respectively (Fig. 1b, bottom left panel, column 4 and 7 in the bar graph). The inhibition of cell death was more pronounced when 2 μM of eugenol was combined with 5 and 10 μM of cisplatin. The cell death was 35 and 48% (Fig. 1b, right panel, column 4 and 7 in the bar graph) and 55 and 90% respectively (Fig. 1b, bottom right panel, column 5 and 8 in the bar graph), when cells were first treated with cisplatin [combination sequence (a). Based on these results, we have chosen eugenol at 2 μM and cisplatin at 10 μM to test the pro-apoptotic effects of the combination using Annexin V/PI flow cytometry and immunoblotting. The synergistic effects were evident and the combination treatment increased the proportion of apoptotic cells to 87 and 82.5% in both OV2774 and SKOV3 cells (Fig. 1c). This was confirmed by showing clear increase in the level of cleaved-caspase 3 and cleaved PARP proteins, compared with single agents (Fig. 1d).

Cisplatin-induced CSC enrichment has been reported for OC cells [6]. Therefore, we assessed the effectiveness of the combination treatment on the sphere formation efficiency. While eugenol showed modest effect, cisplatin increased the number of spheres (Fig. 1e). Interestingly, the sphere formation efficiency was abolished by the combination treatment (Fig. 1e). This suggests that eugenol can suppress the cisplatin-related enrichment of OCSCs.

Studies have shown that Hes1, a downstream effector of the Notch signaling, is closely involved in cancer progression, chemoresistance, promotes self-renewal and tumor initiation in various cancers [19]. However, the role of Hes1 in OCSCs remains mostly unclear. Therefore, to evaluate the importance of the Notch-Hes1 pathway in OC, we first assessed the genetic alterations of Hes1, Hey1, Notch1, Notch2 and Notch3 from cBioportal for Cancer Genomics data portal. Hes1 locus is amplified in 2–25% in breast and 2–22% in ovarian cancer, followed by Hey1, Notch2 and Notch3 (Additional file 2: Figures S2A-S2E). To get more insight into the role of Hes1 in regulating OCSC functions and response to cisplatin, we have ectopically expressed human Hes1 in a promoter driven green fluorescence protein (GFP) reporter system pHes1d2EGFP [15] in OV2774 and SKOV3 cells (Fig. 2a). After transduction, Hes1GFP⁺ cells were sorted and then were cultured in CSC medium for 10 days (Fig. 2b). Spheres were enzymatically isolated and analyzed by flow cytometry to isolate Hes1GFP(+) and GFP(−) cells and identified approximately 74 and 63% of Hes1GFP⁺ cells in OV2774 and SKOV3 cells (Fig. 2c), however, the parental cells contained only 10 and 13% of Hes1⁺ cells, respectively (Fig. 2c). However. The expression of the Hes1 protein and mRNA in Hes1GFP⁺ and GFP⁻ cells were confirmed by Western blotting and qRT-PCR (Figs. 2d; e). The Hes1GFP⁺ cells grew slower in complete medium (medium supplemented with 10% FBS) (Fig. 2f) and showed higher sphere formation ability than GFP⁻ cells in supplemented CSC medium (Fig. 2g). This prompted us to assume that Hes1GFP⁺ cells are likely to be resistant to cisplatin than GFP⁻ cells. To test this, Hes1GFP⁺ and GFP⁻ cells were treated with increasing concentrations of cisplatin. Hes1GFP⁺ cells showed higher cell viability than GFP⁻ cells (Fig. 2h). Furthermore, the proportion of cleaved caspase-3 positive cells was higher in GFP⁻ cells relative to Hes1GFP⁺ cells when treated with cisplatin (10 μM) (Additional file 2: Figure S2F). Moreover, multi-drug resistance markers ABCG/MDR1, ABCG2 and ABCG5; CSC markers CD44, ALDH, EpCam and CD49f, and pluripotency markers Nanog, Oct3/4 and Sox2 were all highly expressed in Hes1GFP⁺ cells as compared to GFP⁻ cells (Fig. 2i). Additionally, Hes1GFP⁺ cells were more invasive than the parental and GFP⁻ cells (Additional file 2: Figure S2G). To confirm the in vitro results in animal model, Hes1GFP⁺ cells were isolated as described in Fig. 2a and implanted subcutaneously into nude mice (n = 5/group) and revealed the greater ability to form tumors by Hes1GFP⁺ compared to the GFP⁻ cells (Fig. 2j). These results indicate that increase in the expression of Hes1⁺ cells promote CSC traits, and resistance to cisplatin of OC cells.

To investigate the effects of combination treatment on OCSCs, we have first analyzed the CD44 expression and ALDH activity in Hes1-GFP⁺ and GFP⁻ sorted OV2774 and SKOV3 cells. Hes1-GFP⁺ cells were grown in supplemented CSC medium to enrich the CSC activities and sorted as described in Fig. 2a. Immunoblot analysis from sorted Hes1-GFP⁺ and GFP⁻ populations showed that CD44 and ALDH1 expressions were distinguishable in both cell lines (Additional file 3: Figure S3A). We then analyzed the self-renewal capacity of sorted CD44⁺ and CD44⁻ cells. It revealed that CD44⁺ cells are more self-renewal capable than CD44⁻ cells (Additional file 3: Figure S3B). Furthermore, ALDH⁺ cells are more self-renewable capable than ALDH⁻ counterparts (Additional file 3: Figure S3C). In addition, the size and the number of spheres were significantly reduced in CD44⁻/ALDH⁻ fraction as compared to their CD44⁺/ALDH⁺ populations (Additional file 3: Figure S3D). When CD44⁺ and ALDH⁺ cells were both removed simultaneously by FACS, cells lost the sphere formation capacity (Additional file 3: Figure S3E, S3F). These results suggest that CD44⁺/ALDH⁺ subpopulations were the most efficient in sphere formation among Hes1-GFP⁺ cells.

Next, we hypothesized that combination treatment could specifically target CD44⁺/ALDH⁺ populations. To address this, sorted CD44^+/− and ALDH^+/− cells were treated with DMSO (control), cisplatin and eugenol either alone or in combination for 5 days, and then cell viability was assessed by the WST-1 assay. The combination treatment significantly inhibited the growth of CD44⁺/ALDH⁺ cells compared to single agents (Fig. 3a). Next, we assessed the effectiveness of combination treatment on self-renewal ability in CD44⁺/ALDH⁺ fractions. Cells were sorted, cultured (adherent culture) and treated for 72 h, trypsinized and cultured in the supplemented CSC medium for 10 days (sphere culture). While both control and eugenol treated cells were able to form spheres overtime (blue line: CD44⁺/ALDH⁺ yellow line: CD44⁻/ALDH⁻), treatment with cisplatin increased the number of spheres within the same time period (Fig. 3b). In contrast, the combination treatment started to reduce the ability of sphere formation as early as 3–4 days in both cell populations (Fig. 3b). We then assessed the effects of combination treatment in the changes of CD44 and ALDH fractions by flow cytometry. The combination treatment significantly reduced the fractions of CD44⁺ and ALDH⁺ subpopulations (Fig. 3c; d). While eugenol eliminated CD44⁺ population by only 6% in OV2774 cells (statistically not significant), the proportion was increased by cisplatin to 72.50%. However, the combination treatment strongly reduced the proportion of CD44⁺ cells to 10.5% (Fig. 3c). Likewise, cisplatin increased the proportion of CD44⁺ cells to 73.18% in SKOV3 cells, while co-treatment reduced this proportion to 11.5% (Fig. 3c). Similarly, the ALDH activity increased by cisplatin from 22.58% (control) to 40.70% in OV2774 cells and 15.75 to 41.99% in SKOV3 cells (Fig. 3d). In contrast, co-treatment significantly reduced ALDH activity to 4.23% in OV2774 cells and 3.33% in SKOV3 cells (Fig. 3d). These results suggest that combination treatment effectively reduced the CD44⁺ cells and ALDH activity in OCSC populations.

Several studies have shown that side populations (SPs) exhibit CSC-like properties and are responsible for drug resistance and recurrence [20]. SPs are able to efflux DNA binding dye Hoechst 33342 out of the cell membrane and enriched with CSC populations in OC [21, 22]. We therefore sought to enrich and purify SPs from OV2774 and SKOV3 cells by FACS after verapamil treatment (verapamil is an inhibitor of several verapamil-sensitive ABC-transporters). In the SKOV3 cells, the proportion of SPs in the control cells was 8.17%, whereas this proportion was reduced to 1.07% when cells were treated with verapamil (Additional file 4: Figure S4A). When characterizing sorted SPs and non-SPs (NSPs), SPs were able to divide indefinitely, and formed cobblestone epithelial like colonies, highly tumorigenic when grown in agarose medium and positive for Hes1 (Fig. 4a). On the other hand, NSPs ceased proliferating within 3 weeks and were less tumorigenic and mostly negative for Hes1 (Fig. 4a). To analyze combination effect on these SPs, cells were treated with either cisplatin and/or eugenol alone or in combination, and then SP fractions were analyzed by FACS. The proportions of SPs were 6.78 and 8.58% for OV2774 and SKOV3 cells (Fig. 4b). While SPs were significantly unaffected by eugenol (6.14% for OV2774 and 7.31% for SKOV3), cisplatin treatment increased the proportion of SPs to 9.19% in OV2774, and 12.07% in SKOV3 cells (Fig. 4b). Interestingly, these inductions were reduced by cotreatment to 0.73% in OV2774 and 1.43% in SKOV3 cells. Next, we investigated the combination effect on the expression of several drug resistance, CSCs and pluripotency markers by qRT-PCR. While the effect of eugenol was minimal on ABCG1/MDR1, ABCG2 and ABCG5, these genes were significantly up-regulated when exposed to cisplatin. In contrast, these genes were sharply down-regulated by cotreatment (Fig. 4c). Similar results were obtained for the CSCs and the pluripotency cohort (Fig. 4c). Identical results were observed for all these genes in SKOV3 cells (Fig. 4c). These findings indicate that, SPs that are capable of self-renewal and drug resistance could effectively be targeted by the combination treatment. Furthermore, combination treatment significantly inhibited the invasion/migration abilities of both CD44⁺ and ALDH⁺ sub-populations compared to single agents (Fig. 4d, Additional file 4: Figure S4B). We then assessed the apoptotic response in the sorted CD44⁺ and ALDH⁺ cells and found that while single agents had only partial effects on these cells, the combination treatment significantly increased the proportion of apoptotic cells to 90% in both CD44⁺ and ALDH⁺ cells (Fig. 4e).

We then analyzed the correlation between drug resistance, CSC and pluripotency cohorts with Hes1 using results obtained from qRT-PCR and analyzed using “R” with CORRPLOT package. In most instances, Hes1 is correlated with all markers analyzed and “r” values ranging from 0.94 (Sox2 vs ABCG2) to 0.99 (ABCG2 vs Nanog) and so on. For example, Hes1 was highly positively correlated with ABCG2 (r = 1.0, p 0.001) and Nanog (r = 0.98, p 0.002) in OV2774 and SKOV3 cells (Additional file 4: Figure S4C, related Additional file 8: Table S4C-i-iv).

To explore the Notch-dependent cisplatin mediated drug resistance and CSC regulation, we investigated the possible implication of the Notch pathway in the cisplatin and eugenol treatment-dependent targeting of OCSCs. To explore this, sorted Hes1⁺ cells were allowed to grow in the CSC medium, and then were treated with cisplatin and eugenol either alone or in combination. Spheres were collected and the expression of the Notch pathway proteins was analyzed by immunostaining. The expression intensity of Hes1, CD44 and ALDH was significantly reduced in the combination-treated cells. While eugenol alone did not alter the expression intensity, cisplatin alone increased the expression intensity in both OV2774 and SKOV3 cells (Additional file 5: Figure S5A and S5B). In line with these results, the immunoblot results showed the levels of Hey1, Hes1, cleaved Notch1 and Jagged1 were all reduced by cotreatment as compared to single treatments (Fig. 5b). c-Myc is a direct downstream target of Notch1 and preferentially induces the Akt signaling [23, 24]. c-Myc and the phospho-Akt were downregulated by cotreatment without affecting the basal level of AKT (Fig. 5b). Additionally, we determined whether the combination affected the γ-secretase complex consisting of Prenisillin-2, Nicastrin, APH1alpha and PEN2. All γ-secretase complex proteins were downregulated by the cotreatment compared to single agents (Fig. 5b). Furthermore, adding DAPT, a γ-secretase complex inhibitor, further inhibited cell proliferation (Fig. 5c), self-renewal and sphere formation capacity (Fig. 5d), while induced apoptosis (as evidenced by Western blot and flow cytometry) and reduced Hes1 expression (Fig. 5e, f). Together, these results indicate that the combination treatment targets OCSCs by inducing apoptotic activity as well as through inhibition of the Notch signalling pathway.

To evaluate the efficacy of combination treatment on tumor growth and OCSCs in vivo, we utilized xenograft models. Mice carried xenografts from SKOV3 and OV2774 cells injected as single inoculums in the dorsal site of each mouse (n = 10/group; Fig. 6a). Inception of tumors was first confirmed, and then animals were randomized and treated with DMSO (control), eugenol, cisplatin or both for 3 weeks in a drug administration sequence described in Fig. 6a. Xenografts were monitored for treatment efficacy at the completion of treatment (n = 5/group) and disease progression (n = 5/group) 4 weeks after cessation of treatment. Tumor volume and weight were regressed in eugenol and cisplatin treatment group, but significant regression was observed in the cotreatment group (Fig. 6b; c). This indicates that eugenol strongly potentiates tumor growth inhibitory effect of cisplatin. Tumor sections were stained with H&E and were also immunostained for proliferation, disease specific and the Notch pathway markers. The cotreatment strongly reduced the expression of Ki-67 and PAX8 relative to the control and monotherapy groups (Fig. 6d; Additional file 6: Figure S6A). Similar effect was also observed on Hes1, Notch1 and Jagged1 (Fig. 6d; Additional file 6: Figure S6A). Furthermore, cleaved caspase-3 level was induced in cotreatment group as compared to controls (Fig. 6d; Additional file 6: Figure S6A). Progression-free survival is an important tool for evaluating the efficacy of a treatment and response to therapy. We therefore analyzed the tumor mass and the progression of the tumors 4 weeks after cessation of treatments (Fig. 6a). The treatment response was durable without gaining tumor mass in the co-therapy group 4 weeks post-treatment (Fig. 6e, f). In the cotreatment group, tumors were barely detectable in all mice, and only small foci of tumor was detected in one mouse (Fig. 6e, f). Tumor progression was observed in all monotherapy treated mice within 10 and 15 days after cessation of treatment except for the cotreatment group (Fig. 6g, h).

To assess the long-term effects of the cotreatment and the self-renewal capacity of OCSCs, equal number of dissociated unsorted cells from excised tumor xenografts were cultured for 3 weeks in a semi-solid agarose medium. While cells from control and eugenol treated xenografts grew robust colonies, cells from cisplatin-treated tumors had slower but steady growth and grew small colonies. On the other hand, no colonies were formed from tumors treated with combination (Fig. 7a, b). This indicates that cotreatment abolished the self-renewal capacity of OCSCs. Although, dissociated tumor cells from co-treated SKOV3 xenografts significantly reduced the proportion of CD44 population (4.97%) and ALDH (2.05%) activity in these tumors, these proportions remained higher in the controls and monotherapy treated tumors (Fig. 7c). Identical results were obtained for the tumors from OV2774 cells (Fig. 7c). To confirm these results in vivo, the dissociated cells from previously treated mice (refer to Fig. 6a) were re-implanted into mice subcutaneously (n = 5/group) and left for 16 weeks without treatment. While, tumor cells from untreated and monotherapy-treated mice regrew and progressed, only small foci of tumor (1 out of 5 mice) were detected in tumor cells from co-therapy treated mice group (Fig. 7d, e, f, g). In contrast, animals inoculated with tumor cells derived from cotherapy-treated animals showed significantly better tumor-free survival as compared to the control group (Fig. 7h, i).