Background
Metastasis is a major hallmark of cancer, and the major death cause of cancer patients [
1]. Many studies indicate that tumor heterogeneity is responsible for tumor metastasis, relapse and drug resistance of cancers including hepatocellular carcinoma (HCC) [
2]. Cancer stem cells are conceited as the cells within the tumors possessing the capacities of self-renewal and heterogeneous lineages differentiation [
3]. HCC cells sorted by CD133 [
4] and CD44 [
5] were regarded as a subpopulation of cells with stem cell properties [
6]. It is convinced that specifically targeting cancer stem-cells is a promising approach for cancer treatment based on a detailed understanding of this unique sub-group biological features distinct from its parental counterparts [
7]. Abnormally activated signal pathways and proteins in this subgroup could all be potential therapeutic targets [
8,
9].
Stroma factors from tumor cells or stromal cells are critical for tumor metastasis [
10]. Our previous studies have demonstrated OPN is a leading gene that promotes HCC metastasis [
11,
12]. Intracellular/nuclear OPN regulated epithelial-mesenchymal plasticity, which contributes to the increased population of cancer stem-like cells (CSCs), to enhance tumor metastasis [
13,
14]. In addition, OPN and its various cleavages in the tumor microenvironment serve as hematopoietic stem cell niche components that negatively regulates the anchorage and pool size of stem cells [
15,
16]. OPN is also highly expressed in cancer stem cells isolated from HCC cell lines [
17], and secreted OPN-CD44 signaling enhanced the phenotypes of cancer stem cells in glioma and colon cancer and promoted their aggressive tumor growth [
18,
19]. These indicate OPN contributes to a cancer stem-like phenotype.
Epigenetic regulation has an important contribution in the development of adult stem cells, and its important roles in silencing tumor repressors and differentiation-associated genes in precancerous cells have drawn much attention recently [
20]. Aberrant DNA methylation, the best studied epigenetic modification, has emerged as promising therapeutic targets to treat cancer. In ovarian cancer, epigenetic targeting agent can reprogram residual cancer stem-like cells [
21]. DNMT1 is the most abundant type of DNMTs which catalyzes DNA methylation. In tumor tissues or cancer stem cells, DNMT1 is highly up-regulated [
22,
23], and is required for maintaining the state of cancer stem cells [
24]. DNMT1 inhibitor, 5 Aza, an FDA-approved drug to treat myelodysplastic syndromes (MDS), can eradicate cancer stem cells of solid tumor by inducing cell apoptosis or differentiation [
25].
In the present study, we found that knockdown of OPN in CD133+/CD44+ cells inhibited sphere formation and migration through modulating the expression of DNMT1. Decreased levels of DNMT1 by OPN knockdown leaded to reduction in global DNA methylation, particularly in CpG island (CGI). In addition, CD133+/CD44+ subgroup from cell lines and HCC tissues with various OPN levels showed different sensitivities to 5 Aza. These results provide a potential significance of effective specific target therapy that epigenetic treatment is more effective for HCCs with high OPN expression.
Materials and methods
Magnetic activated cell sorting (MACS)
1 × 107 cells were prepared and incubated with CD133 microbeads for 30 min. 1-2 mL MACS running buffer was added and the sample was centrifugated at 2000 rpm, 3 min and resuspended in 500ul MACS running buffer. The cells were placed in the prepared LS column and the column was washed for 6 times. Cells were removed from the column with physical forces. Then the single positive cells were incubated with CD44 microbeads and repeat the steps as before.
Immunohistochemical staining was performed according to the previous work [
11].
Cell culture and primary cells isolation
Huh7 and Hep3B were purchased from Chinese Academy, Shanghai and cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Hyclone) with 10% FBS (Biowest). Primary cells were isolated from HCC tissues (from Huashan Hospital, Fudan University, Shanghai) by using IV type collagenase (Sigma) digested for 30 min and centrifugation for 1 min at 50 g.
Patients and clinical tissue samples
A total of 374 patients who underwent curative resection for HCC at the authors’ institute between 2005 and 2006, were enrolled in the present study. None of them received any preoperative adjuvant treatment. The clinical samples were collected from patients after obtaining informed consent in accordance with an established protocol approved by the Ethics Committee of Fudan University (Shanghai, China).
Spheroid-based migration assay on matrix protein
Flat-bottomed, 96-well plates (Corning) were coated with 0.1% gelatin (Sigma) in sterile water for 30 min at 37 °C. After removing the gelatin, 4-day spheroid was transferred into the plate cultured with 200ul/well medium supplemented with 2% FBS. Spheroids could adhere and migrate. Images were acquired at 0 h and 48 h using a stereomicroscope (Olympus, Shinjuku-ku, Tokyo, Japan) and analyzed by contrasting the area covered by migrating cells [
26].
iTRAQ assay
iTRAQ-8plex labeling reagents (AB Sciex) were added to the peptide samples. And then the peptides were fractionated on a waters UPLC using a C18 column. The fraction was separated by Nano-HPLC (Eksigent Technologies) on the secondary RP analytical column (Eksigent, C18, 3 μm, 150mmx75μm). Peptides were subsequently analyzed by the mass spectrometer (QTOF 5600). For the data processing of iTRAQ experiments, protein identification and iTRAQ 8 plex quantification were performed with ProteinPilot4.5 software (AB Sciex).
MeDIP-sequencing
Genomic DNA of cells was purified, sonicated and then denatured. Denatured DNA was incubated with 5mC antibody (Active Motif) at 4 °C overnight. DNA-antibody complexes were captured by protein A/G beads (Santa Cruze). Harvested DNA was purified and sequenced followed by standard Illumina protocols. Read sequences were mapped to the human genome (hg38) using ELAND v2 in the CASAVA (Illumina, v1.6) package.
LC-MS/MS analysis
LC-MS/MS analysis Genomic DNA (8 μg) from cultured cells was digested with DNA Degradase Plus (Zymo Research) at 37 °C for 3 h. The dideoxycytidine (TCI) was added as an internal control. The digested samples were then subjected to LC-MS/MS analysis using a ShimazuLC (LC-20AB pump) system coupled with TSQ-Vantage triple quadrupole mass spectrometer (Thermo). A C18 column (250 mm × 2.1 mm I.D., 3 μm particle size, ULTIMATE) was used. The mass spectrometer was optimized and set up in selected reaction monitoring (SRM) scan mode for monitoring the [M + H+] of 5-mC (m/z 242.1 → 126.1), 5-hmC (258.1 → 142.1), 5-fC (256.1 → 140.1), 5-caC (272.1 → 156.1) and dideoxycytidine (212.1 → 112.1). The Analyst Software was used for analysis.
Xenograft assay
For detection of the ability of tumor initiation, diluted cells mixed with matrigel were subcutaneously implanted into the 4–6 weeks-old NOD SCID mice. Tumor formation was monitored each week. To evaluate the rate of inhibition of 5 Aza, 30 mg/kg 5 Aza was intraperitoneally injected into the nude mice when they burdened a tumor in almost the same volume for five times a week.
Methylated DNA immunoprecipitation sequencing data analysis
Sequencing adapters were removed and low-quality bases (quality < 20) were trimmed from the 5′ and 3′ ends of reads using an in-house Perl script. The obtained clean reads were then mapped to the human reference genome (hg38) using the default parameters of the BWA program (version 0.7.7). Peaks of MeDIP-seq were identified by MACS2, then we merge the peak of CSCs-SCR and CSCs-shOPN samples. Static the fragments of each samples in enriched regions. Then we identified differentially methylated regions refer to a previously published method of differentially expressed genes [
27].
p values were adjusted by false discovery rate (FDR) for multiple tests. A threshold of FDR < 0.05 and fold change > 2 was applied.
Statistics analysis
All data are expressed as the mean ± standard deviation. Error bars represent ± standard deviation for triplicate experiments. The difference between groups was analyzed using Student t-test when comparing only two groups or one-way analysis of variance when comparing more than two groups. The level of significance was set at p < 0.05(*p < 0.05, **p < 0.01, ***p < 0.001).
Plasmids construction and transfection
The plasmids of OPN and DNMT1 were constructed as described in the Additional file
1.
Sphere formation
Sphere formation assay was assessed as detailed in the Additional file
1.
RNA isolation, Reverse-transcription, and quantitative PCR analysis
Quantitative PCR analysis (qPCR) were carried out as detailed in the Additional file
1.
Western blot assay
Whole cell lysis was produced by using RIPA buffer containing protease inhibitor. Western blot was performed as described in the Additional file
1.
Flow cytometry
Flow cytometry analysis was performed as described in the Additional file
1.
Dot blot assay
Genomic DNA was extracted by using phenol-chloroform and diluted into the same concentration. The detailed method was described in the Additional file
1.
Methylation-Specific PCR, MSP
Genomic DNA was isolated from the cells. MSP assay was done as described in the Additional file
1.
Discussion
High probabilities of metastatic relapse and drug resistance remain the major obstacle to further prolong the survival of patients with HCC [
33]. CD133+ and CD44+ were proposed to be markers of tumor-initiating cells (TICs) in liver cancers [
34]. Targeting TICs may provide a practical approach for cancer treatment [
35]. Many clinical trials and pre-clinical studies that specifically target TICs are under processing [
36]. Acyclic retinoid (ACR), a selective chemopreventive agent, selectively cleared MYCN+ liver CSCs [
37]. However, great breakthrough may be arduous to achieve for the dynamics and biomarkers uncertainty of HCC CSCs. Therefore, it is crucial to uncover the biology and phenotype regulation of CD133+/CD44+ subgroup in HCC.
OPN had been reported to be highly expressed in cancer stem cells and its secreted form regulated the self-renewal of cancer stem cells genetically. Cancer stem cells are able to undergo self-renewing division and differentiate into other kinds of cells [
38]. With these properties of malignancy, cancer stem cells display phenotypical and functional heterogeneity accounting for therapeutic refractoriness and tumor dormancy [
39]. Cancer stem cells were often distinguished and sorted by relevant specific biomarkers. The heterogeneity of CSCs is universal. CD44 has been reported to be marker of CSCs in many tumors. Ectopic expression of CD44v6 in colorectal CSCs by activating the Wnt/β-catenin pathway, promotes migration and metastasis. However, in gastric cancer, CD44v8–10 enriched at the invasive front attenuates redox-stress-induced canonical Wnt activation [
40,
41]. The plasticity of CSCs depends on the cancer types and cell context. In this work, CD133+/CD44+ cells were approved to have the properties of cancer stem cells as reported. We found that OPN could enhance the self-renewal of CD133+/CD44+ cells via regulating DNMT1. Therefore, apart from the secreted form, intracellular OPN also contributes to the stemness of CD133+/CD44+ cells.
Aberrations in DNA methylation patterns contribute to cancer pathogenesis and malignancy, usually with a global DNA hypomethylation and local hypermethylation at specific tumor suppressor genes or differentiation-related genes. Abnormal DNA methylation at the 5 position of cytosine (5mC) is a well-known epigenetic feature of cancer [
42]. Meanwhile, epigenetic transcriptional regulation is also crucial in the development and maintenance of cancer stem cells. However, whether OPN affected DNA methylation remained unknown. Using MeDIP-seq, we found that OPN altered DNA methylation landscape of CD133+/CD44+ cells via regulating the expression of DNMT1, a key enzyme of DNA methylation. With aberrant DNA hypermethylation due to low expression of DNMT1, the expressions of some tumor suppressors, involving genes aberrantly methylated in HCC and those as drivers of other cancers, were significantly up-regulated, which could break down the balance of self-renewal and differentiation of CD133+/CD44+ cells. Considering that epigenetic alteration precedes gene expression, OPN-induced differentially methylated genes are speculated to have strong potential as epigenetic biomarkers for cancer diagnosis, prognostication and therapeutic interventions. Collectively, these demonstrate that OPN induces genome-wide methylome alterations. This may be a mechanism exploited by OPN to enhance the stemness of CD133+/CD44+ cells via methylome reprogramming to antagonize apoptosis or differentiation.
When TICs are eradicated thoroughly by targeting the abnormally activated signaling pathways or protein, tumor can be possibly cured [
43]. Targeting OPN can be a strategy for elimination of CD133+/CD44+ subgroup. Our lab has reported that neutralizing antibody to OPN can inhibit the in vitro invasion and in vivo lung metastasis of highly metastatic HCC cells. And microRNA against OPN led to an obvious inhibition of both in vitro invasion and in vivo lung metastasis of HCC-LM3 cells [
11]. But OPN-specific inhibitory compounds are not available yet. Thus, there is no ideal targeting therapeutic strategy. We have to explore an alternative strategy that indirectly targets OPN by the signaling cascade components, leading to OPN-dependent HCC stemness. DNMT1 can be an ideal subrogation of targeting OPN in terms of the epigenetic therapeutic strategy. We found that CD133+/CD44+ cells with high OPN expression up-regulated DNMT1, which made CD133+/CD44+ cells more sensitive to DNMT1 inhibitor, 5Aza. This work also provides an evidence that 5 Aza can be used to treat CD133+/CD44+ cells from solid tumor and HCC cells with high OPN expression. However, DNMT1 has been reported to be with multiple functions. High DNMT1 was related with a dismal prognosis of HCC patients. However, some else reports demonstrated that low DNMT1 contributed to the stemness of cancer stem cell-like cells and treatment with 5 Aza in HCC cells could increase the number of cancer stem cell-like cells [
44]. The real reason of the controversial results is not clear, and the possible mechanisms involved deserve further investigation. In a recent report, 5 Aza in combination with alendronate could reduce the dose of 5 Aza [
45], and the combination of 5 Aza and HDAC inhibitors result in a better curative effect in mouse model. Our findings at least in part support that 5 Aza can be used to treat HCC, especially for those with OPN over-expression.
Acknowledgements
We thank Lei Zhang and Guoquan Yan from Biomedical Core Facility, Fudan University, for technical support.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.