Background
Ovarian cancer is the most lethal malignancy in gynecological carcinomas, with an overall 5-year survival rate of only approximately 30–40%. The main factors driving this poor prognosis is the high rates of recurrence and drug resistance observed in this disease [
1]. A growing number of recent studies have shown that CSCs are the source of relapse and chemoresistance of various cancers [
2],and may serve as target cells for cancer therapeutics [
3].
First identified in the hematopoietic system, CSCs have been successively identified in many different types of tumors (including ovarian cancer), by several putative markers, such as CD44, CD24, EpCAM, CD133, CD117, ALDH1 [
4‐
6]. The CSCs theory hypothesizes that tumors may arise from a subpopulation of tumor cells endowed with self-renewal, tumorigenic, and multilineage differentiation capacities [
7]. However, the mechanism by which CSCs maintain their stemness remains poorly understood.
Accumulating data show an intimate connection between autophagy and stemness [
8].Autophagy involves the sequestration of cytoplasmic components within a double-membraned vesicle, and subsequent fusion with lysosomes to generate autolysosomes, in which the autophagic cargo is degraded by acidic hydrolases and recycled for cells to survive under various stress conditions [
9]. Autophagy is essential to protect hematopoietic stem cells from metabolic stress and promotes survival in conditions of growth factor withdrawal and nutrient deprivation [
10]. Loss of autophagy impairs hematopoietic stem-cell self-renewal activity and regenerative potential [
11]. In addition, autophagy is also necessary for the maintenance of stemness of mesenchymal stem cells [
12], muscle stem cells [
13], and embryonic stem cells [
14].More recently, it has been revealed that autophagy was upregulated in some CSCs, such as breast CSCs [
15], pancreatic CSCs [
16], and liver CSCs [
17], and that compromising autophagy contributed to a decrease of self-renewal ability. However, both the mechanism by which autophagy maintains the stemness of CSCs and whether or not autophagy plays a critical role for ovarian CSCs (OCSCs) remain elusive.
In the present study, we enriched OCSCs from the ovarian cancer cell lines 3AO and SKOV3 using serum-free suspension culture system as we reported previously [
18,
19], and observed greater autophagic flux in OCSCs compared to the bulk population of ovarian cancer cells. Furthermore, we demonstrate that the self-renewal ability and chemoresistance of OCSCs were decreased when autophagy was inhibited by BafA1, CQ or ATG5 knockdown. Using a stem cell signaling-related PCR array, we screened molecules that participated in maintaining the stemness of OCSCs, and verified that FOXA2 served as a stemness regulator in autophagy, controlling the self-renewal of OCSCs. The aim of our study was to investigate the role of autophagy and involved molecules in maintaining the stemness of OCSCs and possible targets for ovarian cancer therapeutics.
Methods
Cell culture and reagents
Ovarian adenocarcinoma cell line 3AO was obtained from the Women’s Hospital, School of Medicine, Zhejiang University. SKOV3 was obtained from the American Type Culture Collection (ATCC, HTB-77). 3AO and SKOV3 were maintained in RPMI 1640 (Corning, Steuben County, NY, USA) and McCoy’s 5A medium (Gibco, Carlsbad, CA, USA) respectively, supplemented with 10% fetal bovine serum. To form spheres, the 3AO and SKOV3 cells were cultured at a density of 50,000 cells/ml in serum-free DMEM/F12 medium (Cellgro, Virginia, USA) composed of 10 ng/ml basic fibroblast growth factor (bFGF, Peprotech, Rocky Hill, NJ, USA) and 20 ng/ml epidermal growth factor (EGF, Peprotech), 1 mg/ML insulin (Sigma–Aldrich, St Louis, MO, USA), and 1 × B27 supplement (Life Technologies,Carlsbad, CA, USA) on the ultralow attachplates (Corning, Steuben County, NY, USA). All cells were maintained in a humidified incubator with 5% CO2 at 37 °C. Nutrient deprivation (starvation) was induced through incubating cells in Earle’s Balanced Salt Solution (EBSS, Sigma–Aldrich); BafilomycinA1(BafA1, Sigma–Aldrich) and Chloroquine (CQ, Sigma–Aldrich) were dissolved in DMSO as stock solution. Paclitaxel was sourced from Bristol-Myers Squibb PharmaceuticalsLtd.
RT2 profiler PCR arrays and qRT-PCR
We extracted total RNA from cells using TriZOL (Invitrogen, Carlsbad, CA, USA) according to themanufacturer’s instructions and further purified RNA using the RNeasy® MinElute™ Cleanup Kit (QIAGEN, Manchester, UK). RNA yield was then quantified via UV absorbance (NanoDrop® ND-1000) and quality was assessed by denaturing agarose gel electrophoresis to assess RNA, before generating cDNA by RT2 First Strand Kit (QIAGEN). Next, the cDNA is mixed with an appropriate RT2 SYBR Green Mastermix(QIAGEN, 330,601), aliquoted into the wells of the RT2 Profiler PCR Array (QIAGEN). Finally, relative expression was determined using the ∆∆CT method.
For qRT-PCR, RNA was extracted by TriZOL and reverse transcribed into cDNA using ReverTra Ace qPCR RT Kit (Toyobo, Japan). The subsequent quantitative RT-PCR was performed using Thunderbird SYBR qPCR Mix (Toyobo, Japan) and Applied Biosystems 7900HT fast real-time PCR System(Life Technologies, Carlsbad, CA, USA). GAPDH was used as an endogenous control for normalization. Primers sequences are listed in Additional file
1: Table S1.
Western blotting analysis
Primary antibodies to FOXA2, NES(Nestin), NANOG(Nanog), POU5F1(Oct4), ALDH1A1(Aldehyde Dehydrogenase1 family member A1) and IL8(Interleukin 8) were purchased from Abcam. Antibodies to LC3B, GAPDH were obtained from Sigma–Aldrich, ATG5 and SQSTM1(sequestosome 1)were purchased from Cell Signaling Technology. Proteins were loaded and separated on SDS–PAGE (15% for LC3B; 10% for other proteins), then transferred to PVDF membranes(0.45-um, Corning). After blocking in 5% nonfat dry milk in TBST, membranes were incubated with primary antibodies overnight at 4°C and subsequent secondary antibodies. Membranes were then washed with TBST, and treated with EZ-ECL kit (BI biological industries, Cromwell, CT, USA) to detect bands in Imagequant LAS400mini (GE Healthcare). All experiments were performed at least three times.
Flow Cytometry
To measure the expression of CD24 andCD44, cells were trypsinized into single cells, then washed twice with phosphate-buffered saline (PBS). Subsequently, 1 × 106cells of 3AO were labeled with 1 unit of phycoerythrin mouse anti-humanCD24 (BD Bioscience Pharmingen Inc., San Diego, CA, USA) and 1 × 106cells of SKOV3 were labeled with 1 unit of fluorescein isothiocyanate mouse anti-humanCD44 (BD Bioscience Pharmingen Inc.). Phycoerythrin Mouse IgG2a and Fluorescein isothiocyanate Mouse IgG2a were used as isotype controls (BD Bioscience Pharmingen Inc.). The surface expression of CD24 and CD44 were then assessed by flow cytometry (FCM, BD Bioscience Pharmingen Inc.) after incubation with antibody for 30 min at 4 °C in the dark.
Plasmids and shRNA transfection
The pEGFP(+)-FOXA2 plasmidwas constructed as follows. The ORF encoding FOXA2 was synthesized by Sangon Biotech (Shanghai, China), and inserted into the EcoRI/BamHI sites of a pEGFP-N1 vector. We designated the empty pEGFP-N1 vector as the mock control. The GFP-LC3plasmid was a gift from Dr. Hong-He Zhang at Departmentof Pathology & Pathophysiology, School of Medicine, Zhejiang University (Hangzhou, China). SKOV3 and 3AO cells were grown to 80% confluency before plasmid transfection. The ratio of DNA:X-tremeGENE HP DNA Transfection Reagent (Roche, Basel, Switzerland) was 1:2, using 2 mg DNA for each well of a 6-well plate. The transfection protocol followed the instructions of the manufacturer. SKOV3 and 3AO cells were diluted to 10% to 15% confluencyafter a 24-h transfection of plasmids, then selected with 400 mg/mlG418 (Sigma–Aldrich) for 10 days.
Lentivirus vectors containing short hairpinRNAs against ATG5 and FOXA2, and scrambled control shRNA were obtained from Genechem(Shanghai, China). SKOV3 and 3AO cells were infected with the lentivirus vectors both at MOI 20 according to the manufacturer’s instructions. The expression of each protein was confirmed by Western blotting.
The sequences of shRNA are listed in Additional file
1: Table S1.
Immunofluorescence analysis
3AO and SKOV3 cells were transfected with GFP-LC3 overnight and then transferred to coverslips. After adhering to coverslips, the cells were treated with 50 nM BafA1 or not for 4 h. The images were collected by using aspinning disk confocal fluorescence microscope (on a systemcomposed of a CSU-X1 spinning disk from Yokogama, a IX81microscope from Olympus(Tokyo, Japan) and a IXON3 CCD from Andor) at 600 × magnification. The Metamorph offline 7.7.8.0 software package was used to count the amounts of GFP-LC3-II-positivepuncta in 50 GFP-positive cells for each group.
Transmission electron microscopy
TEM was conducted as previously described [
20]. 3AO cells and 3AO CSCs were treated with 50 nM BafA1 or not for 4 h and fixed overnight with 2.5% glutaraldehyde solution (Sigma-Aldrich), postfixed with 1%OsO4 and then dehydrated standards were embedded in fractionated ethanol in812 resin (Ted Pella,Redding, CA, USA). The thin sections were sectioned and stained with 2% uranyl acetate and detected with a Tecnai 10 transmission electron microscope (Philips, Amsterdam, NED).
Chemotherapy sensitivity assays
Sphere cells were seeded in ultra-low attachment 96-well plates at appropriate densities (9000 cells per well for SKOV3 and 8000 cells per well for 3AO), then exposed to paclitaxel at different final concentrations (0, 2, 5, 10, 20, 50 nM) for 48 h when autophagy was inhibited by ATG5 knockdown or treated with CQ (10 μM, 48 h).Each concentration was repeated in triplicate wells. Subsequently, Live cells was measured using CCK-8 kit (Dojindo laboratories, Kumamoto, Japan), and the absorbance was read by VarioskanFlash microplate reader (Thermo Scientific, Waltham, MA, USA). Cell viability was measured by ratio of absorbance value of paclitaxel treated/untreated wells in each group.
Statistical analysis
Statistical analysis of differences between the groups was performed by using Student’s t-test. Results were presented as mean ± SD and a P value of less than 0.05 was considered to be significant.
Discussion
Cancer stem cells can be isolated from cancer cell lines and enriched in the form of spheres through serum-free suspension culture system. In the current study, we successfully isolated and enriched OCSCs from parental 3AO and SKOV3 cells utilizing the same method as we previously reported [
18,
19]. There was however a difference in the surface markers expressed on 3AO and SKOV3 spheres. The former presented as CD24 negative while the latter were highly CD44 positive. Following our protocol, we were able to attain high percentages of OCSCs in spheres; 98.4% and 78.5% of CD24
− 3AO cells and CD44+ SKOV3 cells, respectively. Furthermore, some stemness regulators representing stem cell characteristics – including NES, NANOG, and POU5F1 – were all more highly expressed in CD24
− 3AO cells and CD44
+ SKOV3 cells than in parental cells, suggesting that induction of sphere formation by serum-free suspension culture of 3AO and SKOV3 does indeed enrich cancer stem cells, which can be further studied.
A previous report showed that both autophagosome formation and autophagic flux were enhanced in mammospheres derived from breast cancer cell lines and primary tumors, while the size and number of mammospheres were decreased when autophagy was blocked by 3-MA, BafA1, CQ, or when essential autophagy genes, like ATG7 and BECN1, were knocked down [
15]. Consistent results were also revealed in liver, pancreatic, and glioblastoma cancer stem cells [
17,
22,
23]. Here, we demonstrate that the amount of both ATG5 and LC3B-II was markedly increased in 3AO and SKOV3 spheres, concurrent with increased degradation of the autophagy cargo protein SQSTM1, in the presence of BafA1 and in the context of starvation induced by EBSS treatment, compared with bulk ovarian cancer cells. These findings were confirmed by counting GFP-LC3-II positive puncta by confocal microscopy and visualization of autophagic vacuoles in TEM, suggesting that autophagic flux is increased in OCSCs. Further, we found remarkably decreased sphere formation capacity and a reduced percentage of cells bearing stemness markers in 3AO and SKOV3 sphere cells after autophagy was impeded by CQ and ATG5 knockdown, suggesting that an increase in autophagy is essential for maintaining the self-renewal of OCSCs. We also found paclitaxel sensitivity was dramatically increased in sphere cells when autophagy was suppressed by CQ and ATG5 knockdown. Because our previous findings suggested the self-renewal ability of sphere cells were suppressed after sphere cells were treated with CQ or by ATG5 knockdown for 6 days. To exclude the effect of self-renewal, we measured cell viability only after 2 days that sphere cells were exposed to paclitaxel with CQ and shATG5 lentivirus or not. Moreover, as shown in Additional file
4: Figure S3c, CQ or ATG5 knockdown did not cause extra cell death in sphere cells, suggesting the increase of paclitaxel sensitivity was due to inhibition of autophagy.
Some previous studies reported that several autophagy related genes – such as BECN1(Beclin 1, autophagy-related), ATG4A, ATG4C [
15,
24,
25],ATM (Ataxia-Telangiectasia Mutated), and DDX53O (DEAD-box helicase 53) [
25,
26]– were involved in its regulation and contributed to maintaining the stemness ofCSCs. However, those genes are more associated with the regulation of autophagy rather than with the link between autophagy and stemness. To uncover the molecules that are involved in maintaining stemness through modulation of autophagy in OCSCs, we applied a stem cell signaling related PCR array to screen for the expression of genes implicated in these processes. Of 38 differentially expressed genes, FOXA2 was highly expressed in 3AO and SKOV3 spheres and selected for further study.
FOXA2, a pioneer transcription factor and one of the members of Fox (Forkhead box) family, plays a key role in embryo development and is a key marker of endoderm cells [
27]. Dysregulation of FOXA2 expression in fetal ventral mesencephalon-derived neural precursor cells appears to be associated with the loss of their potential to differentiate into dopaminergic neurons [
28].FOXA2 is also used in the identification of a progenitor population that gives rise primarily to ventricular cardiovascular cells [
29]. In addition, FOXA2 could maintain breast cancer stem cells, and downregulating it in these cells reduced the formation of spheres and the expression of CD44, ALDH1 [
30]. Taken together, FOXA2 is a stemness regulator in normal stem cells and cancer stem cells. In the current study, we found that the size and number of spheres derived from 3AO and SKOV3 cells, and the percentage of cells bearing cancer stem cell phenotype markers in spheres, were all depressed after FOXA2 was knocked down, suggesting that FOXA2 modulates self-renewal of OCSCs. Furthermore, inhibition of autophagy suppressed FOXA2 expression, and more importantly, FOXA2 could revert the impairment of self-renewal capacity induced by compromising autophagy, suggesting that autophagy drives OCSCs maintenance through FOXA2.
It seems impossible that autophagy target FOXA2 through direct degradation, because our results showed inhibition of autophagy synchronously decrease mRNA and protein level of FOXA2. A few molecules or pathways may affect FOXA2, for instance, FOXA2 may be the target gene of transcription factors CREB1[
31]and SOX17 [
32], and can be epigenetically suppressed by hypoacetylation of histone H3 and H4 and trimethylation of H3K27 on the FOXA2 promoter [
28]. Moreover, FOXA2 is a direct target gene of Hedgehog signaling pathway which involved in autophagy activation [
33,
34]. But whether autophagy employs Hedgehog or other signaling pathways to regulate FOXA2 need to be carefully explored.