The role of chitinase-3-like protein 1 (CHI3L1) in ovarian cancer and the possible mechanisms were elucidated. CHI3L1 is a secreted glycoprotein and associated with inflammation, fibrosis, asthma, extracellular tissue remodeling and solid tumors. Our previous study showed CHI3L1 could be a potential prognostic biomarker for epithelial ovarian cancer and could protect cancer cells from apoptosis. Therefore, clinical data and quantitation of CHI3L1 of ovarian cancer patients, tumor spheroid formation, side-population assays, Aldefluor and apoptotic assays, ELISA, RT-PCR, immunoblotting and animal experiments were performed in two ovarian cancer cells lines, OVCAR3 and CA5171, and their CHI3L1-overexpressing and -knockdown transfectants. High expression of CHI3L1 was associated with poor outcome and chemoresistance in ovarian cancer patients. The mRNA expression of CHI3L1 in CA5171 ovarian cancer stem-like cells was 3-fold higher than in CA5171 parental cells. CHI3L1 promoted the properties of ovarian cancer stem-like cells including generating more and larger tumor spheroids and a higher percentage of ALDH+ in tumor cells and promoting resistance to cytotoxic drug-induced apoptosis. CHI3L1 could induce both the Akt (essential) and Erk signaling pathways, and then enhance expression of β-catenin followed by SOX2, and finally promote tumor spheroid formation and other properties of ovarian cancer stem-like cells. OVCAR3 CHI3L1-overexpressing transfectants were more tumorigenic in vivo, whereas CA5171 CHI3L1-knockdown transfectants were not tumorigenic in vivo. CHI3L1 critically enhances the properties of ovarian cancer stem-like cells. CHI3L1 or CHI3L1-regulated signaling pathways and molecules could be potential therapeutic targets in ovarian cancer.
Ovarian carcinoma is the leading cause of death among all gynecologic malignancies, with more than 70% of women presenting with advanced-stage disease (Cohen et al. 2014). The annual incidence rate of ovarian cancer in the United States was 12.2 per 100,000 and the death rate was 8.2 per 100,000 in 2007 (U.S. Cancer Statistics Working Group); these rates were respectively 9.12 and 3.09 in Taiwan (Chiang et al. 2013). The poor prognosis for these patients is the result of a lack of symptoms, difficulties in early diagnosis, insufficient accurate tumor markers and deficiency of knowledge about ovarian tumor cell biology (Pinke et al. 2008). The standard treatment for ovarian cancer is surgical tumor debulking, followed by platinum-containing chemotherapy (Agarwal & Kaye 2003, Nagourney et al. 2003). Conventional prognostic parameters for ovarian carcinoma are tumor stage, residual tumor volume after surgery, histologic type, differentiation grade and response to chemotherapy (van der Burg et al. 1988). However, these factors do not present a comprehensive picture of the tumor biology of ovarian cancer and are frequently interrelated. Thus, identifying the molecular pathways of ovarian cancer is extremely important for diagnosis and treatment of this disease.
Increasing evidence supports the hypothesis that a subpopulation of cancer cells with stem cell features lurks within cancers (Crea et al. 2009, Clevers 2011, Visvader 2011, Moncharmont et al. 2012, Pylväs-Eerola et al. 2016). The proportion of these cancer stem-like cells or cancer-initiating cells in tumors is generally less than 2% (Tang et al. 2007). Compared to other cells, these cancer stem-like cell populations are more tumorigenic, metastatic, invasive, prone to angiogenic switch and resistance to chemotherapy and radiotherapy (Bapat et al. 2005, Wu & Alman 2008, Crea et al. 2009, Ahmed et al. 2010, Visvader 2011, Moncharmont et al. 2012, Steg et al. 2012). The multipotent cancer stem-like cells with these properties drive histologic heterogeneity within the tumors (Park et al. 2010). The high incidence of disease recurrence may also contribute to the multidrug resistance and multiple histological phenotypes of the multipotency of stem cell-like properties (Raghavan et al. 2017). Chitinase-3-like protein 1 (CHI3L1 or YKL-40) is a secreted glycoprotein with a molecular weight of about 40 kDa. It is a member of the glycosyl hydrolase 18 family, lacks chitinase activity, and is produced by the various types of cells, including inflammatory cells and cancer cells (Libreros & Iragavarapu-Charyulu 2015, Wiley et al. 2015). Expression levels of CHI3L1 in serum or tissues positively correlate with various types of cancer, including ovary, lung, breast and brain (Johansen et al. 2004, Park et al. 2010, Ku et al. 2011, Chiang et al. 2015a), and with several inflammatory or autoimmune diseases, including coronary artery disease, diabetes mellitus and rheumatoid arthritis (Kazakova et al. 2013, Hansen et al. 2015, Wiley et al. 2015). CHI3L1 is hypothesized to play a role in cancer cell proliferation, migration, invasion and metastasis. Although CHI3L1 can be a marker associated with a poorer clinical outcome of different tumors, its biologic functions and possible mechanisms of action remain under investigation.
Our previous study identified that ovarian cancerous tissues expressed higher CHI3L1 than ovarian benign and normal tissues (Chiang et al. 2015a). Besides, CHI3L1 was associated with poor prognosis of advanced-stage ovarian cancer patients (Chiang et al. 2015a). To explore the possible biological functions of CHI3L1 in this cancer, we hypothesized that CHI3L1 could promote its cancer stem-like cell properties. We first found that RNA levels of CHI3L1 in CA5171 ovarian cancer stem-like cells were 3-fold higher than those in CA5171 parental cells. Moreover, CA5171 spheroid cells are more chemoresistant and prone to tumorigenesis than their parental cells. CHI3L1 promotes properties of ovarian cancer stem-like cells, including generating larger and more numbers of tumor spheroids, higher percentages of ALDH+ tumor cells, and resistance to cytotoxic drug-induced apoptosis. CHI3L1 could induce both of the Akt (essential) and Erk signaling pathways, and then enhances expression of β-catenin followed by SOX2, and finally promotes tumor spheroid formation and the various properties of ovarian cancer stem-like cells. In vivo, OVCAR3 CHI3L1-overexpressing transfectants were more tumorigenic, whereas CA5171 CHI3L1-knockdown transfectants were less tumorigenic. Our findings suggest that CHI3L1 and its related axis (CHI3L1-β-catenin-SOX2) may provide a promising target for the treatment of ovarian cancer.
Materials and methods
Patients and specimens
From 2000 Jan to 2016 Jun, 113 women with advanced-stage (stages III and IV) ovarian carcinoma treated in our institute were enrolled. The Institutional Review Board of the hospital approved the study protocol and the patients provided informed signed consent after Jun, 2007. Tissue specimens were collected during surgery, immediately frozen in liquid nitrogen and stored at −70°C until analysis. The characteristics of disease were defined according to the system of International Federation of Gynecology and Obstetrics (FIGO; Prat 2014). The optimal debulking surgery was defined as when the maximum diameter of a residual tumor was <1 cm. After completion of the primary treatment, history taking, pelvic/rectal examination and regional lymph nodal palpation would be arranged every 3 months for 3 years, and every 6 months thereafter. Recurrence was recorded when tumor marker levels were greater than or equal to twofold the upper limit of normal in two consecutive tests with 2-week intervals, findings of imaging studies and aspiration cytology were abnormal or when there was histological confirmation from a tissue biopsy. The period from completion of the primary treatment until the diagnosis of disease recurrence was defined as disease-free survival (DFS). The time from the diagnosis of the disease until the date of death or last visit was defined as overall survival (OS).
Cancer cell lines
We used two OVCAR3 and CA5171 ovarian cancer cell lines in our studies. OVCAR3 cells purchased from American Type Culture Collection were maintained in RPMI-1640 medium supplemented with 0.01 mg/mL bovine insulin and 20% FBS and grown at 37°C under 5% CO2 (Chiang et al. 2015b). The molecular authentication of OVCAR3 cell line used in this study was confirmed by STRS analyses as showed in Supplementary Figs 1 and 2 (see section on supplementary data given at the end of this article). The other ovarian cancer cell line, CA5171, established in our lab, was cultured in RPMI-1640 supplemented with 20% FCS, 50 units/mL penicillin/streptomycin, 2 mM L-glutamine, 1 mM sodium pyruvate and 2 mM nonessential amino acids and grown at 37°C under 5% CO2 (Chiang et al. 2015b). The passage numbers of the cells used in the following experiments were from P3 to P5.
Tumor spheroid formation
To generate tumor spheroids, we seeded OVCAR3, OVCAR3/CHI3L1-C4, OVCAR3/CHI3L1-C9, CA5171, CA5171/shCHI3L1-S1 and CA5171/shCHI3L1-S2 cells under stem cell conditions as follows: serum-free minimal essential media (RPMI-1640) supplemented with 1% FBS in 6-well (1 × 106 cells/mL), 24-well (5 × 104 cells/mL), or 96-well (1 × 104 cells/mL) Ultra Low Attachment plates (Corning, Corning, NY, USA) and subsequent organization into tumor spheroids for at least 48 h (Sato et al. 2016).
Single CA5171 cell suspensions were stained with 5 µg/mL Hoechst 33342 dye (Sigma-Aldrich) for 45 min at 37°C. Propidium iodide was used to exclude nonviable cells. A total of 100 µM verapamil (Sigma-Aldrich) was then added to inhibit drug efflux pump proteins. The cells were analyzed using a MoFlo XDP cell sorter (Beckman Coulter; Chiba et al. 2006).
ALDH is a marker of stemness phenotypes. Aldefluor reagent (StemCell Technologies, Vancouver, Canada), a fluorescent non-toxic substrate for ALDH, was used to detect ALDH activity in spheroids. Spheroids were suspended in Aldefluor assay buffer containing the ALDH substrate and incubated for 45 min at 37°C. As a negative control, suspended spheroids in the same conditions were treated with diethylaminobenzaldehyde, a specific ALDH enzyme inhibitor. Propidium iodide was added for a 15 min before analysis on a BD FACSCalibur flow cytometer (Nakahata et al. 2015).
To evaluate the apoptosis of various tumor cell lines, their transfectants and their spheroids treated with cytotoxic drugs, cells were first incubated with 1 μM of paclitaxel or 300 μM of cisplatin for 48 h. Then cells were stained with FITC-conjugated annexin V and 7-AAD (BD Biosciences, San Jose, CA, USA), following the manufacturer’s instructions, and finally analyzed by flow cytometry as described earlier. Cells treated with DMSO were used as the negative control.
Xenograft animal model
Six to 8-week-old female SCID mice were purchased from the National Laboratory Animal Center (Taipei, Taiwan) and bred in the animal facility of the School of Medicine of National Taiwan University. All animal procedures were carried out according to approved protocols and in accordance with recommendations for the proper use and care of laboratory animals by the animal ethic committee of College of Medicine, National Taiwan University.
In a limit dilution assay, SCID mice (3 per group) were challenged with 1 × 106, 1 × 105 or 1 × 104 CA5171 parental and CA5171 cancer stem-like cells (cultured for 21 days) subcutaneously. Before being injected, cells were resuspended in a 100 µL 1:1 RPMI-1640 and PanaceaGel (Menicon Co., Ltd., Kasugai, JP) mixture. Tumor growth was monitored weekly by inspection and palpation, and the size of each tumor was measured and recorded.
To compare tumor growth of cancer stem-like cells with or without expression of CHI3L1, equal numbers of various cells (1 × 106, 1 × 105, or 1 × 104) were cultured to form spheroids under stem cell conditions for 48 h. Total spheroids of each group were then harvested and suspended in the single cell suspension. All cancer stem-like cells were resuspended in 100 µL 1:1 RPMI-1640 mixed with PanaceaGel (Menicon Co., Ltd.), injected subcutaneously into the right leg of SCID mice. Tumor growth was monitored and recorded twice per week.
Total mRNA of various tissue specimens, ovarian cancer cell lines and their transfectants was first isolated by TRIzol RNA isolation kit according to the manufacturer’s instructions. The Moloney murine leukemia virus reverse transcriptase kit was then used to generate all the cDNA template. To generate CHI3L1, the sense primer TGTGAAGGCGTCTCAAACAG and anti-sense primer AATTCGGCCTTCATTTCCTT were used in PCR for 30 cycles, as described previously (Chiang et al. 2015a). GAPDH was used as the housekeeping gene, generated with the sense primer ACCCAGAAGACTGTGGATGG and anti-sense primer TGCTGTA GCCAAATTCGTTG and used for 30 cycles at 58°C, as described (Chiang et al. 2015a). Analysis of PCR products was performed by electrophoresis (1% agarose gel with ethidium bromide) (Sigma) staining in TBE (a buffer mixture of Tris base, boric acid and EDTA) solution.
Quantitative real-time PCR (QRT-PCR)
The cDNA templates were synthesized as previous described (Chiang et al. 2015a). The primer Hs00609691_m1 (TaqMan Assays, Life Technologies Corporation) was used to detect the expression of CHI3L1 by a LightCycler Real-Time detection system (Roche Diagnostics) with 50 cycles of 10 s at 95°C, 10 s at 60°C and 10 s at 72°C. Internal control was performed using the LightCycler h-GAPDH housekeeping gene set (Roche Applied Science) with 50 cycles of 10 s at 95°C, 15 s at 55°C and 15 s at 72°C.
The expression of CHI3L1 was calculated by using the comparative 2−ΔΔCt method as described previously (Livak & Schmittgen 2001). The Ct value was counted by the number of cycles needed for amplification-generated fluorescence to reach a specific threshold of detection. The expression level of CHI3L1 in each sample was used to follow equation: Relative expression level of CHI3L1 = 2−ΔΔCt, ΔCt = Cttarget (CHI3L1) − Cthousekeeping (GAPDH).
Enzyme-linked immunosorbent assay
ELISA was performed to evaluate CHII3L1 concentrations in various cancer cell media. Cells (2.5 × 105 cells/mL) were seeded in a six-well culture plate and incubated for 48 h. Culture media were then collected and detected using ELISA kits (R&D Systems).
Immunoblotting assays were performed to detect various expression levels of CHI3L1, stemness-related proteins and signaling pathway proteins in OVCAR3, CA5171 and their transfectant cells. To measure alterations in stemness gene expression during spheroid formation, cancer cells (1 × 106 cells/mL) were suspended in six-well Ultra Low Attachment plates (Corning) and incubated for 48 h for spheroid formation. Spheroids were harvested and lysed in protein extraction buffer containing proteinase K inhibitor. The BCA Protein Assay Kit (Thermo Fisher Scientific) was used to quantify protein extracts. Equal amounts of protein were resolved on 10% SDS-PAGE gels, transferred to a PVDF/nylon membrane (EMD Millipore) and probed with Abs specific to α-tubulin (Abcam), β-catenin (Abcam), CHI3L1 (R&D Systems), Nanog (OriGene Technologies, Rockville, MD, USA) and SOX2 and Oct-4 (Cell Signaling). The membrane was then incubated with HRP-conjugated rabbit anti-mouse (Hycult Biotechnology, Uden, Netherlands) and goat anti-mouse (R&D Systems) secondary Abs. The ECL western blotting system (Perkin Elmer) was used to visualize specific bands, and ImageQuant 5.0 software (Molecular Dynamics Inc., Sunnyvale, CA, USA) was used to measure protein levels through densitometric analysis.
For signaling pathways involved in the spheroid formation, OVCAR3 tumor cells were seeded in six-well Ultra Low Attachment plates under stem cell conditions as described earlier and treated with various concentrations of recombinant human CHI3L1 protein (R&D Systems) for 30 min. To evaluate the phosphorylation of various proteins, Phospho-p44/42 MAPK, Phospho-Jnk, p44/42 MAPK, Akt, JNK (all from Cell Signaling) and phospho-Akt (Santa Cruz Biotechnology) were used as primary Abs, and HRP-conjugated rabbit anti-mouse (Hycult Biotechnology) was used as the secondary Ab.
In the signal transduction pathways blockade assays, OVCAR3 cells cultured in Ultra Low Attachment plates under stem cell conditions were first treated with the respective inhibitor (PD98059, 50 µM; LY294002, 50 µM; Sigma-Aldrich) for 2 h, followed by exposure to recombinant human CHI3L1 protein for 30 min. Cells were then extracted and immunoblotted to detect expression of the respective molecules in various signaling pathways, as previously described.
Generation of overexpressing and knockdown CHI3L1 in ovarian cancer cell lines
The lentivirus-packaging plasmids, PLKO_AS3w.puro, Δ8.91 and pMD.G were purchased from the National RNAi Core Facility (Academia Sinica, Taiwan). Full-length CHI3L1 was first generated by PCR using human placenta cDNA as the template and the following set of primers: sense, Hu-CHI3L1_S_NheI GGCTAGCACCATGGGTGTGAAGGCGTCTCAAAC; and anti-sense, (PstI) Hu-CHI3L1_AS_ PstI TAACTGCAGCTCAGCCTGGGACTCAGCA (PCR conditions). The results were cloned into the NheI/PstI sites of PLKO_AS3w.puro vector, designated as pLKO_AS3w CHI3L1.puro. To generate CHI3L1 lentivirus, 293T cells were co-transfected with three plasmids, pLKO_AS3w-CHI3L1.puro, Δ8.91 and pMD.G, in a total transfection volume of 6 mL of Opti-MEM medium containing Lipofectamine2000 reagent in 10 cm dishes and incubated at 37°C under 5% CO2 for 48 h (Chiang et al. 2015a).
To generate the CHI3L1-overexpressing OVCAR3 transfectants, OVCAR3 cells were infected by CHI3L1 lentivirus in culture medium with polybrene (10 µg/mL) and incubated at 37°C under 5% CO2 for 48 h (Chiang et al. 2015a). The PLKO_AS3w.puro was used to generate the blank lentivirus as the negative control. To select stable CHI3L1-overexpressing OVCAR3 transfectants, puromycin (3 µg/mL) was added to the culture medium and incubated at 37°C under 5% CO2 for more than 48 h. The puromycin-resistant CHI3L1-overexpressing OVCAR3 transfectants OVCAR3/CHI3L1-C4 and OVCAR3/CHI3L1-C9 were individually picked, expanded and used in the subsequent experiments.
To generate CHI3L1-knockdown CA5171 transfectants, CHI3L1 shRNA plasmid was purchased from the National RNAi Core Facility (Academia Sinica, Taiwan). The CHI3L1 shRNA lentivirus was generated by co-transfection with shRNA plasmid and Δ8.91 and pMD.G in Opti-MEM medium containing Lipofectamine2000 and incubated at 37°C under 5% CO2 for 48 h (Chiang et al. 2015a). The PLKO_AS3w.puro was used to generate the blank lentivirus as the negative control. CA5171 cancer cells were infected by CHI3L1 shRNA lentivirus in culture medium with polybrene (10 µg/mL) and incubated at 37°C under 5% CO2 for 48 h. For selecting stable CHI3L1-knockdown CA5171 transfectants, puromycin (3 µg/mL) was added to the culture medium and incubated at 37°C under 5% CO2 for more than 48 h. The puromycin-resistant CHI3L1-knockdown CA5171 transfectants, CA5171/shCHI3L1-S1 and CA5171/shCHI3L1-S2, were individually picked, expanded and used in the subsequent experiments.
Transfection of siRNA (small interfering RNA)
siRNA for SOX2 and CTNNB1 (β-catenin) was designed by and purchased from OriGene Technologies Inc. (Rockville, MD, USA). The duplex sequence of SOX2 was SR304528C-rGrGrUrUrGrArCrArCrCrGrUrUrGrGrUrArArUrUrUrArUAA, and the duplex sequence of CTNNB1 (β-catenin) was SR301063A-rGGAUCACAAGAUGGAAUUUAUCAAA. The negative control was the siRNA duplex carrying a 27-mer scrambled sequence from OriGene’s siRNA gene-specific products. OVCAR3 and CHI3L1-overexpressing OVCAR3 transfectants were seeded in 6-well culture plates (Corning) overnight and then transfected with siRNA SOX2 or siRNA CTNNB1 (β-catenin) in a final concentration of 100 nM using Lipofectamine 2000 transfection reagent (Invitrogen) and in a total of 2 mL of serum-free RPMI 1640 medium. After incubation at 37°C, 5% CO2 for 12 h, 2 mL of RPMI 1640 medium containing 20% FBS was added. All cells were harvested and prepared for the subsequent experiments.
SPSS version 15.0 (SPSS Inc) was used for data analysis. Statistical analyses were carried out using Kruskal–Wallis or one-way ANOVA test. Survival curves were generated using the Kaplan–Meier method, and differences in survival curves were calculated using the log-rank test. A P value <0.05 was considered to indicate statistical significance.
Ethics approval and consent
Animal handling and procedures were approved by the animal ethic committee of College of Medicine, National Taiwan University (20160416). The Institutional Review Board of the National Taiwan University Hospital approved the study entitled ‘Screening and identification of novel diagnostic and prognostic biomarkers on ovarian cancers’ (200706002R) to collect the human tissues after acquiring the informed consents.
High mRNA expression level of CHI3L1 correlated with chemoresistance and shorter DFS and OS
The clinicopathological characteristics of these cases are shown in Table 1. The mean follow-up duration was 39.0 months. The serous type was 75% (85/113) and tumor grade 3 was 92% (104/113) in all cases. The percentage of optimal debulking in all case was 50%. The median mRNA expression level of the CHI3L1 was 7.6 (range 0.0–668.0).
Clinico-pathologic characteristics of the ovarian carcinoma patients (n = 113).
|Characteristic||Patients (n = 113)|
|Age (<50/>50, year-old)||20/93|
|FIGO stage (III/IV, cases)||97/16|
|Histology (serous/clear cell/endometrioid/mix, cases)||85/15/11/2|
|Tumor grade (1/2/3, cases)||7/2/104|
|Debulking surgery (optimal/suboptimal, cases)||57/56|
|CHI3L1 mRNA expression level (median (range))||7.6 (0.00–668.0)|
|Disease recurrence (yes/no, cases)||86/27|
|Disease-related death (yes/no, cases)||53/60|
FIGO, International Federation of Gynecology and Obstetrics.
Chemoresistant patients had the higher median of CHI3L1 mRNA level (18.5 ± 16.3 vs 4.3 ± 2.5, Kruskal–Wallis test, P = 0.001, Fig. 1A) than chemosensitive patients. Besides, high CHI3L1 group (≥7.6) had both of shorter DFS (Fig. 1B) and OS (Fig. 1C) than low CHI3L1 group (log-rank test, P < 0.001).
Based on the clinical and outcome results, we hypothesized that the biologic role of CHI3L1 in ovarian cancer is to promote ovarian cancer stem cell formation.
Characteristics of CA5171 cancer stem-like cells in parental and spheroid cells
Representative figures of CA5175 parental cells and CA5171 spheroids are shown in Fig. 2A. Figure 2B shows representative results for side-population assays by flow cytometry. The percentages of CA5171 parental cells pumping out Hoechst33342 decreased from 2.8 to 0.7% under verapamil treatment. The estimated proportion of stem-like cells of the parental CA5171 line was around 2.1% by side-population assays.
CA5171 parental cells and spheroids expressing aldehyde dehydrogenase (ALDH) were further evaluated. As shown in Fig. 2C, 3.8% of CA5171 parental cells expressed ALDH, whereas 55% of CA5171 spheroid cells did so. These results indicated that percentages of cancer CA5171 stem-like cells increased by changing the culture media from regular medium to stem cell culture medium.
Spheroid cells are more chemoresistant to cytotoxic drugs and tumorigenesis than are their parental cells
The chemoresistant capabilities between CA5171 parental cells and spheroids were also evaluated. The IC50 for paclitaxel in CA5171 parental cells was around 0.11 μM (Fig. 2D) whereas for CA5171 spheroid cells, it was >1 μM. In addition, the IC50 values for cisplatin in CA5171 parental cells and their spheroid cells were 172.3 and 604.2 μM, respectively (Fig. 2D).
We also assessed in vivo tumorigenic capabilities. The subcutaneous tumors in CA5171 parental and spheroid cell-bearing mice are shown in Fig. 1E. Injection with 1 × 104 CA5171 parental cells yielded no tumorigenesis in the mice; however, injection with 1 × 104 CA5171 spheroid cells resulted in tumorigenesis in two out of six animals (Fig. 2E). These results indicate that CA5171 spheroid cells are more chemoresistant and prone to in vivo tumorigenesis than are CA5171 parental cells.
Exogenous CHI3L1 promotes the properties of ovarian cancer stem-like cells
The roles and biologic functions of CHI3L1 in ovarian cancer stem-like cells were further investigated. Representative images of CHI3L1 mRNA expression levels in CA5171 parental cells and its spheroids are shown in Fig. 3A. The CA5171 spheroids expressed 3-folds of CHI3L1 mRNA higher than CA5171 parental cells (Fig. 3B). Figure 3C shows representative results for spheroids from OVCAR3 parental cells treated with recombinant human CHI3L1 protein. The average spheroid sizes (around 150–250 µm) of OVCAR3 parental cells treated with 1000 ng/mL of CHI3L1 were larger than with 100 ng/mL of CHI3L1 or PBS only (100–150 µm) (Fig. 3C).
In addition, the percentages of ALDH+ OVCAR3 cells treated with CHI3L1 increased in a dose-dependent manner (1.2% in PBS, 3.6% in 100 ng/mL and 4.3% in 1000 ng/mL; Fig. 3D). Protein levels of stemness-related genes including SOX2, NANOG, OCT4 and CTNNB1 (β-catenin) were analyzed by immunoblotting. As shown in Fig. 3E, the protein levels of SOX2 and CTNNB1 (β-catenin) of OVCAR3 parental cells increased in a dose-dependent manner when treated with recombinant CHI3L1. In contrast, protein levels of NANOG or OCT4 did not change under treatment with recombinant CHI3L1 (Fig. 3E).
Apoptotic assays of OVCAR3 cells treated with CHI3L1 were further evaluated. With culture of cells under stem cell conditions with or without CHI3L1, OVCAR3 cells treated with 1000 ng/mL of CHI3L1 had the lowest percentages of apoptotic cells (17.6 ± 0.1%) compared with those treated with 100 ng/mL of CHI3L1 (12.6 ± 0.2%) or PBS alone (8.3 ± 0.1%) (P < 0.001, one-way ANOVA) (Fig. 3F).
CHI3L1-overexpressing cancer cells generate the properties of cancer stem-like cells
The mRNA expression levels of CHI3L1 in OVCAR3/CHI3L1 transfectants and CA5171/shCHI3L1 transfectants are shown in Fig. 4A. The concentrations of CHI3L1 protein (ng/mL) in the culture media of OVCAR3/CHI3L1 transfectants (C4 (52.8 ± 3.7) and C9 (59.6 ± 2.4)) were higher than those of OVCA3/mock transfectants (5.5 ± 0.3) (P < 0.001, one-way ANOVA) (Fig. 4B). However, the concentrations of CHI3L1 protein (ng/mL) in the culture media of CA5171/shCHI3L1 transfectants (S1 (5.6 ± 0.6) and S2 (2.8 ± 0.5) were lower than those of CA5171/mock transfectants (25.0 ± 1.6) (P < 0.001, one-way ANOVA) (Fig. 4B).
The representative images of spheroid formation in different OVCAR3 transfectants and CA5171 transfectants are shown in Fig. 4C. The OVCAR3/CHI3L1-C4 (103.3 ± 10.6) and C9 (103.5 ± 7.4) transfectants had more tumor spheroids (spheroid size >100 µm was counted) than OVCAR3/mock (29.0 ± 3.5) transfectants (P < 0.001, one-way ANOVA; Fig. 4D). In contrast, the tumor spheroids of CA5171/shCHI3L1-S1 (24 ± 3.5) and S2 (28 ± 6.1) transfectants were fewer than CA5171/mock (71 ± 9.3) transfectants (P = 0.001, one-way ANOVA) (Fig. 4D).
Representative images of ALDH+ cells in various ovarian parental lines and their CHI3L1 transfectants detected by flow cytometric analyses are shown in Fig. 4E. The OVCAR3/CHI3L1 transfectants (C4 (16.4 ± 1.2%) and C9 (18.3 ± 1.0%)) had higher percentages of ALDH+ cells than OVCAR3/mock transfectants (1.2 ± 0.1%) (P = 0.017, one-way ANOVA; Fig. 4F). The CA5171/shCHI3L1 transfectants (S1 (25.6 ± 2.5%) and S2 cells (28.1 ± 2.3%)), however, had lower percentages of ALDH+ cells than the CA5171/mock transfectants (51.4 ± 0.7%) (P = 0.002, one-way ANOVA; Fig. 3F).
Immunoblotting analysis of various CHI3L1 transfectants for stemness-related genes was performed. The protein expression levels of SOX2 and β-catenin (CTNNB1) were higher in both OVCAR3/CHI3L1 transfectants (C4 and C9) than in OVCAR3/mock transfectants. In contrast, protein levels of SOX2 and β-catenin were lower in CA5171/shCHI3L1 transfectants (S1 and S2) than in CA5171/mock transfectants (Fig. 4G).
CHI3L1 stimulates both of Erk and Akt phosphorylation during tumor spheroid formation
We then further evaluated the signaling pathways involved in CHI3L1-related tumor spheroid formation. The representative images of immunoblotting results are shown in Fig. 4H. The phosphorylation of Erk1/2 and Akt increased in cells treated with recombinant CHI3L1 under stem cell conditions (Fig. 4H). However, the phosphorylation of JNK or cleavage of Notch1 did not change even under CHI3L1 treatment.
The CHI3L1-overexpressing OVCAR3 transfectants (OVCAR3/CHI3L1-C4 and -C9) also revealed increasing phosphorylation of Erk1/2 and Akt compared with the OVCAR3/mock transfectants (Fig. 4I). CHI3L1-downregulated CA5171 transfectants (CA5171/shCHI3L1-S1 and -S2), on the other hand, had less Erk1/2 or Akt phosphorylation than the CA5171/mock transfectants (Fig. 4I).
CHI3L1 could enhance tumor spheroid formation and characteristics of cancer stem-like cells through the Erk and Akt signaling pathways
To confirm the essential signaling pathway of CHI3L1 in tumor spheroid formation, specific pathway inhibitors such as LY294002 for Akt or PD98059 for Erk were further tested. As shown in Fig. 5A, Erk phosphorylation decreased when cells were cultured with recombinant human CHI3L1 and PD98059 (lane 3) or PD98059 alone (lane 5). In addition, Akt phosphorylation also decreased with cells cultured in recombinant human CHI3L1 and LY294002 (lane 4) or LY294002 alone (lane 6).
We then evaluated if a specific inhibitor of the Akt or Erk signaling pathway could inhibit the CHI3L1-regulated spheroid formation and the generation of stem-like cells. The representative figures of spheroid formation of various OVCAR3 and CA5171 cells and their respective CHI3L1 transfectants are shown in Fig. 5B and C. The spheroid sizes and numbers of OVCAR3 parental cells and OVCAR3/CHI3L1 transfectants decreased under treatment with PD98059 or LY294002 (Fig. 5B). The sizes and numbers of CA5171 parental cells and CA5171/shCHI3L1 transfectants also decreased under these treatments (Fig. 5C).
We then further investigated the percentages of ALDH+ cells. As shown in Fig. 5D, the percentages of ALDH+ OVCAR3 cells cultured with recombinant human CHI3L1 significantly decreased when pretreated with PD98059 (4.2 ± 0.1%, P < 0.001) or LY294002 (2.9 ± 0.1, P < 0.001) compared with the control group (6.4 ± 0.1%) (one-way ANOVA). The percentages of ALDH+ cells in OVCAR3 CHI3L1 transfectants were also lower when cells were pretreated with LY294002 (C4: 2.7 ± 0.3%; C9: 2.6 ± 0.3%) or PD98059 (C4: 8.0 ± 0.8%; C9: 8.6 ± 1.0%) compared with controls (C4: 15.7 ± 0.5%, P < 0.001; and C9: 16.5 ± 0.5%, P < 0.001) (one-way ANOVA) (Fig. 5E). In addition, the OVCAR3/CHI3L1 transfectants treated with LY294002 had lower percentages of ALDH+ cells than those treated with PD98059 (P < 0.001, one-way ANOVA).
We further evaluated if the inhibition of CHI3L1-regulated signaling pathways could enhance sensitivity to cytotoxic drugs. As shown in Fig. 5F, OVCAR3 parental cells cultured with recombinant human CHI3L1 (138.4 ± 3.7%) under stem cell conditions had higher percentages of viable cells than those treated with PBS (100.0 ± 4.0%) when subsequently treated with paclitaxel. However, the percentages of OVCAR3 parental cells treated with PD98059 (50.3 ± 0.6%) or LY294002 (8.7 ± 0.1%) were lower than the OVCAR3 parental cells treated with PBS, when subsequently cultured with paclitaxel (one-way ANOVA, P < 0.001) (Fig. 5G). The OVCAR3/CHI3L1 transfectants pretreated with LY294002 (C4: 12.5 ± 0.7%; C9: 11.3 ± 0.7%) also had significantly lower percentages of viable cells than those pretreated with PD98059 (C4: 117.9 ± 4.3%; C9: 107.9 ± 8.7%) (P < 0.001, one-way ANOVA) when cultured with paclitaxel.
CHI3L1 regulates expression of β-catenin and SOX2 stemness-related molecules through the Akt and Erk signaling pathways
We further evaluated which stemness-related molecules were regulated by CHI3L1 during tumor spheroid formation. As shown in Fig. 5H, the OVCAR3/CHI3L1 transfectants (C4 and C9) expressed higher levels of β-catenin and SOX2 molecules than OVCAR3 parental cells. The expression levels of Nanog or OCT4 did not differ between OVCAR3 parental cells and OVCAR3/CHI3L1 transfectants (Fig. 5H). PD98059 could only inhibit minor expression of SOX2, not β-catenin, in OVCAR3 parental cells and OVCAR3/CHI3L1 transfectants. LY294002, however, suppressed the expression of both β-catenin and SOX2 in OVCAR3 parental cells and OVCAR3/CHI3L1 transfectants. Our results revealed that CHI3L1 could regulate expression of β-catenin and SOX2 during spheroid formation, mainly through the Akt, not Erk, signaling pathway.
β-catenin was upstream of SOX2 among the CHI3L1-regulated tumor spheroid formation and cancer stem-like characteristics
We further assessed if CHI3L1 could regulate the formation and characteristics of cancer stem-like cells through β-catenin and SOX2. The β-catenin siRNA or SOX2 siRNA inhibited expression of their respective molecule in OVCAR3/CHI3L1 transfectants. Inhibition of β-catenin by β-catenin siRNA also suppressed the expression of SOX2 (Fig. 6A), but the inhibition of SOX2 by SOX2 siRNA did not influence the expression of β-catenin. As shown in Fig. 6B, the sizes and numbers of tumor spheroids of OVCAR3/CHI3L1 transfectants significantly decreased under treatment with β-catenin siRNA compared with those treated with scramble siRNA.
The representative flow cytometric figures of ALDH-expressing OVCAR3/CHI3L1 transfectants treated with various siRNA are shown in Fig. 6C. As shown in Fig. 6D, β-catenin siRNA (C4: 8.1 ± 1.6%; C9: 6.1 ± 1.6%) and SOX2 siRNA (C4: 7.8 ± 0.6%; C9: 6.6 ± 1.3%) decreased the percentages of ALDH+ OVCAR/CHI3L1 transfectants compared to scramble siRNA (C4: 18.9 ± 2.7%, P < 0.001; C9: 16.8 ± 1.5%, P < 0.001, one-way ANOVA). The percentages of ALDH+ cells were no different between OVCAR/CHI3L1 transfectants treated with β-catenin siRNA and SOX2 siRNA (P > 0.05, one-way ANOVA). The β-catenin siRNA (C4: 71.8 ± 2.9%; C9: 71.1 ± 0.9%) and SOX2 siRNA (C4: 72.3 ± 3.3%, C9: 74.8 ± 3.4%) also decreased the percentages of viable OVCAR3/CHI3L1 transfected cells treated with paclitaxel compared to scramble siRNA (C4: 94.7 ± 1.3%, P = 0.002; C9: 92.5 ± 3.3%, P = 0.002, one-way ANOVA) (Fig. 6E).
CHI3L1 regulates the formation of cancer stem-like cells to enhance the tumorigenesis of ovarian cancer in vivo
We next investigated if OVCAR3/CHI3L1 transfectants could be more tumorigenic than OVCAR3 parental cells in vivo. As shown in Fig. 6F, when injected with 1 × 106 of the respective tumor cells, some of the mice challenged with the OVCAR3/CHI3L1 transfectants (4/8 for C4 and 3/8 for C9) showed tumorigenesis. However, none of the mice challenged with OVCAR3 parental cells did (Fig. 6F). With the injection of 1 × 106 or 1 × 105 CA5171/mock cells, four of eight and one of eight mice respectively showed tumorigenesis (Fig. 6G). As also shown in Fig. 6G, none of the mice showed tumorigenesis following injection with 1 × 106 CA5171/CHI3L1-S1 or CA5171/CHI3L1-S2 transfectants (CA5171/CHI3L1-knockdown transfectants).
Taken together, these results indicate that CHI3L1 can promote in vivo tumorigenesis by enhancing the quantity of cancer stem-like cells.
We investigated the biological functions of CHI3L1 and if it could be as a potential target in ovarian cancer. Chiang et al. reported that high CHI3L1 correlated with the poor outcome of ovarian cancer (Chiang et al. 2015a). However, the mechanism of CHI3L1 in cancer still needs to be identified. We first identified that ovarian cancer cells could generate spheroids during low-serum culture media. These spheroid cells are more chemoresistant and tumorigenic than their parental cells. Besides, cancer stem-like cells expressed higher CHI3L1 levels than parental cells. Both exogenous and endogenous CHI3L1 could promote the generation of cancer spheroids and properties of ovarian cancer stem-like cells including the expression of ALDH, stemness-related molecules SOX2 and β-catenin and drug-resistance. CHI3L1 could activate both the Akt (essential) and Erk signaling pathways during the generation of tumor spheroids and cancer stem-like cell properties. The blockade of CHI3L1 and molecules in its related signaling pathway could inhibit tumorigenesis in vivo.
Cancer stem cells play important roles in tumor growth and metastasis (Chabner & Murphy 2005, Horst et al. 2009, Wakamatsu et al. 2012), and metastasis is an intractable problem in cancer therapy. One hypothesis of cancer metastasis is that cancer stem-like cells separate first from the local tumor site and enter into the circulation as seeds that then land at various distant sites (Bapat et al. 2005, Clevers 2011). However, cancer cells cannot survive for long periods in single cell suspension (Mehlen & Puisieux 2006, Buchheit et al. 2014), whereas tumor spheroids have proved to be one form in which cancer cells survive longer and circulate within the blood vessels. To form a tumor spheroid, tumor cells need to become cancer stem-like cells (Weiswald et al. 2015), so the quantity of spheroids, including number and size, reflects the numbers of cancer stem-like cells. CHI3L1 promoted the formation of tumor spheroids in our study (Figs 2B and 3C), and earlier results have shown that ovarian cancer patients with the advanced-stage disease have higher CHI3L1 than those with the early stage disease (Weiswald et al. 2015). We hypothesize that ovarian cancer tumors highly expressing CHI3L1 could generate more tumor spheroids, resulting in disease progression and tumor metastasis.
Cancer stem cells also could underlie chemoresistance to cytotoxic drugs in the treatment of ovarian carcinoma (Weiswald et al. 2015). Chemotherapy is always used in metastatic disease, but chemoresistance has been a big challenge, and breaking through it has become a primary focus. Cancer stem-like cells are among the possible explanations for the generation of chemoresistance (Weiswald et al. 2015). Somasagara et al. reported that chemoresistance could be induced by regulating RAD6 gene, which was also associated with the stem cell signaling. RAD6 could activate TLS (PCNA-Ub) and FA/homologous recombination (FANCD2-Ub) to protect ovarian cancer cells from DNA damage caused by chemotherapy (Somasagara et al. 2017).
Patients with chemoresistance are reported to have higher CHI3L1 than those whose cancers are chemosensitive (e.g., in ovarian (Chiang et al. 2015a), glioblastoma (Ku et al. 2011, Weiswald et al. 2015) and lung cancer (Johansen et al. 2004)). CHI3L1 increased chemoresistance in ovarian cancer cells in our previous study and generated chemoresistance by boosting numbers of cancer stem-like cells in the current work (Fig. 5F and G; Chiang et al. 2015a). Thus, CHI3L1 could be a potential target for reversing chemoresistance by reducing the numbers of cancer stem-like cells.
The major challenges of current ovarian cancer treatments are the lack of good prognostic factors, initiation of tumor cells, chemoresistance and high recurrence rate. Cancer stem-like cell correlate with chemoresistance and recurrence (Crea et al. 2009, Ahmed et al. 2010, Steg et al. 2012, Jung et al. 2016, Pylväs-Eerola et al. 2016) and CHI3L1 is highly expressed in several human malignancies, including glioblastoma, lung cancer, hepatoma, breast and ovarian cancers (Johansen et al. 2004, Chiba et al. 2006, Park et al. 2010, Ku et al. 2011, Chiang et al. 2015a). Targeting cancer stem cells is an important and novel strategy for cancer therapy. These cells can be tracked through expression of specific surface markers like CD44, CD133 and ALDH (Monzani et al. 2007, Horst et al. 2009, Wakamatsu et al. 2012) through stemness genes such as Nanog, OCT-4 and SOX2 (Kitamura et al. 2013, Almozyan et al. 2017, Lee et al. 2017) or by the unique pattern of staining with certain dyes such as Hoechst 33342 (Goodell et al. 1996). SOX2 and β-catenin are two intermediate molecules to target in cancer stem-like cells by blocking the function of CHI3L1. Targeting the downstream molecules of CHI3L1 can be another strategy for inhibiting the formation of cancer stem-like cells. SOX2 and β-catenin were intermediate molecules in CHI3L1-regulated cancer stem-like cell formation in our study (Fig. 6B). Our results showed that blocking phosphor-Akt could suppress the expression level of SOX2 or β-catenin and impair spheroid formation in single cancer cell suspension. The major functions of SOX2 and β-catenin in generating spheroids, however, still need further study. The downstream pathway of SOX2 or β-catenin may offer more specific and novel targets for ovarian cancer therapy, as has already been discerned in other cancers (Garros-Regulez et al. 2016, Hwang et al. 2016). Recently, the CHI3L1 neutralization antibody, mAY, was shown to confer anti-angiogenic properties and promote apoptosis by inhibiting CHI3L1 (Faibish et al. 2011). He and colleagues reported that IL13Ra is the receptor for CHI3L1, offering IL13Ra another potential target for inhibiting CHI3L1 function (Faibish et al. 2011). Small molecule-based inhibition of the Akt or Erk pathway may be another strategy for impairing the function of CHI3L1. Further studies are needed to investigate these possibilities.
Figure 7 illustrates a hypothesized pathway based on our findings in this and our previous study. When ovarian cancer cells initiate and grow, chronic inflammation arises in the tumor microenvironment. Cancer cells and/or inflammatory cells secrete abundant CHI3L1 in ascites or extracellular matrix. CHI3L1 then regulates cancer cells to form tumor spheroids by the expression of stemness-related β-catenin and then SOX-2 through activation of the Akt (essential) and Erk signaling pathways. These tumor spheroids then increase chemoresistance and the survival of tumor cells, finally migrating to other sites to establish metastases.
In conclusion, cancer stem-like cells are associated with tumor metastasis, resistance to chemotherapy and radiotherapy and tumor recurrence in many studies (Chabner & Murphy 2005, Chiba et al. 2006, Crea et al. 2009, Moncharmont et al. 2012, Raghavan et al. 2017). Targeting these cells is thus an important and novel strategy to decrease the possibility of tumor metastasis, chemo- or radioresistance and recurrence and to extend patient survival. Our findings suggest that CHI3L1 and its related axis (CHI3L1-β-catenin-SOX2) may provide a promising target for the treatment of ovarian cancer.
This is linked to the online version of the paper at https://doi.org/10.1530/ERC-18-0300.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This work was supported in part by the 2nd, 3rd and 7th core laboratory facility of the Department of Medical Research of National Taiwan University Hospital. This study was granted by the Ministry of Science and Technology of Taiwan (MOST 104-2314-B-002-210, MOST 106-2314-B-002 -191 -MY2, and MOST 106-2314-B-002-194-MY2) and National Taiwan University Hospital (NTUH 105-S3097, 106-N3741, and 107-S3842). The funding agency had no role in study design, data collection and analysis, the decision to publish or the preparation of the manuscript.
AlmozyanSColakDMansourFAlaiyaAAl-HaraziOQattanAAl-MohannaFAl-AlwanMGhebehH 2017 PD-L1 promotes OCT4 and Nanog expression in breast cancer stem cells by sustaining PI3K/AKT pathway activation. International Journal of Cancer 141 1402–1412. (https://doi.org/10.1002/ijc.30834)
ChiangYCChenCAChiangCJHsuTHLinMCYouSLChengWFLaiMS 2013 Trends in incidence and survival outcome of epithelial ovarian cancer: 30-year national population-based registry in Taiwan. Journal of Gynecologic Oncology 24 342–351. (https://doi.org/10.3802/jgo.2013.24.4.342)
HorstDScheelSKLiebmannSNeumannJMaatzSKirchnerTJungA 2009 The cancer stem cell marker CD133 has high prognostic impact but unknown functional relevance for the metastasis of human colon cancer. Journal of Pathology 219 427–434. (https://doi.org/10.1002/path.2597)
HwangSYDengXByunSLeeCLeeSJSuhHZhangJKangQZhangTWestoverKD 2016 Direct targeting of beta-catenin by a small molecule stimulates proteasomal degradation and suppresses oncogenic Wnt/beta-catenin signaling. Cell Reports 16 28–36. (https://doi.org/10.1016/j.celrep.2016.05.071)
KitamuraHTorigoeTHirohashiYAsanumaHInoueRNishidaSTanakaTFukutaFMasumoriNSatoN 2013 Prognostic impact of the expression of ALDH1 and SOX2 in urothelial cancer of the upper urinary tract. Modern Pathology 26 117–24 (https://doi.org/10.1038/modpathol.2012.139)
KuBMLeeYKRyuJJeongJYChoiJEunKMShinHYKimDGHwangEMYooJC 2011 CHI3L1 (YKL-40) is expressed in human gliomas and regulates the invasion, growth and survival of glioma cells. International Journal of Cancer 128 1316–1326. (https://doi.org/10.1002/ijc.25466)
MonzaniEFacchettiFGalmozziECorsiniEBenettiACavazzinCGrittiAPiccininiAPorroDSantinamiM 2007 Melanoma contains CD133 and ABCG2 positive cells with enhanced tumourigenic potential. European Journal of Cancer 43 935–946. (https://doi.org/10.1016/j.ejca.2007.01.017)
NagourneyRABrewerCARadeckiSKidderWASommersBLEvansSSMinorDRDiSaiaPJ 2003 Phase II trial of gemcitabine plus cisplatin repeating doublet therapy in previously treated, relapsed ovarian cancer patients. Gynecologic Oncology 88 35–39. (https://doi.org/10.1006/gyno.2002.6855)
RaghavanSMehtaPWardMRBregenzerMEFleckEMATanLMcLeanKBuckanovichRJMehtaG 2017 Personalized medicine-based approach to model patterns of chemoresistance and tumor recurrence using ovarian cancer stem cell spheroids. Clinical Cancer Research 23 6934–6945. (https://doi.org/10.1158/1078-0432.CCR-17-0133)
SatoMKawanaKAdachiKFujimotoAYoshidaMNakamuraHNishidaHInoueTTaguchiATakahashiJ 2016 Spheroid cancer stem cells display reprogrammed metabolism and obtain energy by actively running the tricarboxylic acid (TCA) cycle. Oncotarget 7 33297–33305. (https://doi.org/10.18632/oncotarget.8947)
SomasagaraRRSpencerSMTripathiKClarkDWManiCMadeira da SilvaLScaliciJKothayerHWestwellADRocconiRP 2017 RAD6 promotes DNA repair and stem cell signaling in ovarian cancer and is a promising therapeutic target to prevent and treat acquired chemoresistance. Oncogene 36 6680–6690. (https://doi.org/10.1038/onc.2017.279)
WakamatsuYSakamotoNOoHZNaitoYUraokaNAnamiKSentaniKOueNYasuiW 2012 Expression of cancer stem cell markers ALDH1, CD44 and CD133 in primary tumor and lymph node metastasis of gastric cancer. Pathology International 62 112–9. (https://doi.org/10.1111/j.1440-1827.2011.02760.x)