Background
Gastric cancer (GC) is one of the most common malignant diseases worldwide [
1]. Although the worldwide morbidity and mortality due to gastric cancer declined in recent years, GC is still ranked as the second and third most common cancer in male and female population, respectively [
1]. Therefore, early diagnosis and effective treatment of gastric cancer remain the major research focuses of oncologists around the world. It is widely believed that a minority of tumor cells endowed with self-renewal and tumor-initiating properties, known as cancer stem cells (CSCs), play an important role in cancer development [
2‐
4]. Recent evidence suggests that CSCs are also key to tumor invasion, metastasis, disease recurrence, and resistance to chemotherapy [
5,
6]. In case of gastric cancer, the existence and pathological significance of gastric cancer stem cells (GCSCs) were firstly suggested by Takaishi et al. [
7]. Subsequently, further studies have confirmed the existence of GCSCs or stem-like cells in gastric cancer cell lines and gastric cancer tissues [
8‐
11]. However, the molecular mechanism underlying GCSC phenotype remains unclear.
Polycomb group (PcG) proteins, which regulate the expression of homeotic genes (HOX) via gene silencing, have been reportedly deregulated in an array of cancers [
12‐
16]. PcG proteins function at different sites in chromosomes by forming multi-protein complexes, known as polycomb repressive complexes (PRCs), resulting in the alteration of chromosomal structure and transcriptional repression of gene expression via epigenetic mechanisms [
12]. Several of these canonical and non-canonical PRCs have been documented to play important roles in embryonic development, cell proliferation and differentiation, malignant transformation, and maintaining properties of normal stem and cancer stem cells [
14‐
17]. Chromobox homolog 7 (CBX7), located on chromosome 22q13.1, encodes a chromobox protein and has been identified as a core PRC1 constituent [
18,
19]. As a member of PcG family proteins, CBX7 can function independently in the initiation and progression of various cancer types [
19‐
23]. However, the role of CBX7 in cancer development remains controversial. CBX7 may act as an oncogene or a tumor suppressor, depending on the cellular context and cancer types [
19‐
23]. While CBX7 knockout mice were reported to develop lung and liver tumors, we and others have clearly shown that CBX7 has oncogenic properties [
23]. In particular, CBX7 is overexpressed in gastric tumors. With respect to the oncogenic role of the PcG proteins, it has been reported that BMI1, EZH2, and Mel-18 were involved in the regulation of stem cells or CSC properties [
15‐
17,
24‐
26]. Notably, CBX7 has been also documented to positively regulate stem cell characteristics of prostate cells and hematopoietic stem cells [
27,
28]. Our previous study found that CBX7 is overexpressed and plays an oncogenic role in gastric cancer [
23]. Therefore, we speculated that CBX7 might also be involved in the regulation of stem cell properties of gastric cancer cells. In the current study, we found that CBX7 positively regulated GCSC phenotype. Further in-depth studies showed that CBX7 regulated GCSC phenotype via the downregulation of its known target p16 and the upregulation of microRNA-21 (miR-21) by virtue of the activation of AKT-NF-κB pathway.
Methods
Cellular reagents and methods
Five strains of human gastric cancer cell lines (MKN28, MKN45, SGC-7901, AGS, NCI-N87) were obtained from the Surgical Institution of Ruijin Hospital, and a human gastric cancer cell line HGC-27 was obtained from the Chinese Academy of Sciences, Shanghai Cell Institution. MNK28, MKN45, SGC-7901, and NCI-N87 cell lines were cultured in RPMI-1640, AGS cell line was cultured in F12, and HGC-27 cell line was cultured in DMEM. The culture mediums were supplemented with 10% fetal bovine serum (FBS) and antibiotics. Cell proliferation was assessed by Cell Counting Kit-8(CCK8) assay. For plate colony formation assay, cells were planted into six-well plates (500 cells/well) and cultured for 2 weeks. Colonies were washed with PBS for three times, then fixed with 4% paraformaldehyde for 10 min. These fixed colonies were stained with 0.1% crystal violet solution and counted. Finally, the crystal violet was washed away with 10% acetic acid, and 600-nm absorbance was assessed by a Microplate Reader (BIO-TEK Instruments Minneapolis, MN). Cell migration and invasion assays were performed using Transwell chamber as described [
29‐
31].
Plasmids, retroviruses, and infection
Retroviral vectors (pGPU/Hygro and pEX-2/Hygro) expressing CBX7, CBX7 shRNA, and p16 shRNA were obtained from Shanghai GenePharma Co Ltd. The target sequence of CBX7 and p16 shRNA are shown as follows: CBX7, 5′-AGA ACC AGA GAG GCT CTG A-3′; p16 5′-GCC CAA CGC ACC GAA TAG TTA-3′.Other retroviral vectors including pSRα-mAkt (originated from Dr. N. Hay, UIC, Chicago, IL) and methods to produce retroviruses have been described before [
32]. Stable cell lines overexpressing CBX7 or other genes of interest were generated by infection of respective retroviruses or transfection of the plasmids. Successfully infected cells were sorted by flow cytometry and then cultured in medium supplemented with G418 or other antibiotics.
Clinical samples
Ninety-five paraffin-embedded primary site specimens of gastric cancer were obtained from the archives of the Fudan University Shanghai Cancer Center for immunohistochemical (IHC) analysis. The clinicopathological variables and survival data were obtained from medical records, and the disease stages were determined according to the 2010 UICC/AJCC gastric cancer TNM staging system. The study was approved by the Institutional Medical Ethics Committee of the Fudan University Shanghai Cancer Center. All clinical samples were collected with the informed consent of patients, and study protocols were in accordance with the ethical guidelines of the Declaration of Helsinki (1975).
Spheroid colony formation (SCF, Tumorisphere) assay was carried out as described previously [
11]. Briefly, cells were trypsinized and washed with PBS and then cultured in ultra-low-attachment plates containing serum-free medium DMEM/F12 supplemented with B27(1:50), N2(1:100), EGF (10 ng/ml), and bFGF (20 ng/ml). The medium was replaced every 5 days. Two weeks later, the number of spheres was counted and processed for protein extraction.
In vivo tumorigenesis
Gastric cancer cells were injected subcutaneously into the flanks or abdominal cavity of severe combined immunodeficiency (SCID) mice. After 10–12 weeks, mice were sacrificed by cervical dislocation. For the flank-injected mice, tumors were removed, and their weight and size were measured. All experiments concerning animals were approved by the Animal Care and Use Committee of the Fudan University.
Immunological reagents, Western blot, and immunohistochemical analysis
Antibodies against CBX7, Oct4, p16, and β-actin were purchased from Abcam. Antibodies against CD24 and CD44 were obtained from Abgent. Antibodies against pAkt (S473) and Akt were purchased from Cell Signaling Technology. The whole cell lysates were harvested using cell lysis buffer supplemented with protease inhibitor cocktail (Sigma), and Western blotting analysis was performed as described [
32]. Immunohistochemical (IHC) analysis to detect the expression of CBX7 or other markers was performed as described [
11,
23]. All slides were observed by two independent pathologists in a double-blinded fashion. The immunoreactive scoring method (IRS) was adopted to assess the result of immunohistochemistry. IRS value of more than 6 was regarded as high-expression group, while IRS value of less than 6 was regarded as low-expression group.
Chemo-sensitivity experiment
Cells were plated into 96-well plates (5000 cells/well) in triplicates and fed with PRMI-1640 medium containing 10% FBS, along with different concentrations of the chemotherapy reagent (Adriamycin (ADM), 5-fluorouracil (5-Fu), and no drug/solvent as control). The number of viable cells was assessed after 2 days of cultivation using the Cell Counting Kit-8 (CCK8) (Dojindo, Kamimashiki-gun Kumamoto, Japan), and the optical absorbance at wavelength 450 nm was detected using the Microplate Reader (BIO-TEK Instruments, Minneapolis, MN).
Statistics
All statistical analysis were done using SPSS 17.0 software package, and two-tailed P values of less than 0.05 were considered significant. In the set of IHC assay of paraffin-embedded tissue samples, the Pearson χ2 test was used to examine the correlation between CBX7 expression and clinicopathological parameters in primary gastric cancer, and paired t test was used to determine the difference of CBX7 expression between primary gastric cancer specimens and their corresponding paracancerous tissue. In in vitro experiments, data were described as mean ± SD and analyzed by Student’s t test.
Discussion
Cancer stem cells (CSCs) belong to a class of functionally distinct subtype of cells in tumor cell population, with biological characteristics of self-renewal, multi-directional differentiation, high tumorigenicity, and resistance to radiotherapy and chemotherapy. These cells have been linked to tumor metastasis and disease recurrence. CSCs may form oncospheres in serum-free cultures supplemented with B27 and N2 supplements, as well as EGF and bFGF, making this technology widely used in the characterization of tumor cells with stem cell characteristics [
4,
37]. Gastric CSCs have been characterized from gastric cancer tissues and gastric cancer cell lines using this method by Han [
38] and Song et al. [
39]. Cancer stem cells and normal stem cells have many common characteristics such as self-renewal and multi-directional differentiation [
40]. Stem cell markers CD133, CD44, ABCG2, Oct4, Lgr5, and CD24 are also highly expressed in cancer stem cells. Using CD44 as a CSC marker protein, six GCSC lines were isolated by Takaishi et al. [
7]. It was also reported that MKN-45, MKN-74, and NCI-N87 gastric cancer cell lines contain a considerable number of CD44-positive cells. These cells form spheres in a high efficiency and lead to tumor formation when injected in to the gastric cavity and skin of the SCID (severe combined immunodeficiency) mice, indicating that CD44-positive gastric cancer cell subsets have potent in vivo tumor-initiating capacity. Zhang et al. [
9] reported that gastric cancer cells with positive expression of CD24 and CD44 exhibited stem cell-like properties, including self-renewal, multi-directional differentiation, and high tumorigenicity. Song et al. [
39] confirmed that, compared to adherent cells, stem-like sphere-forming cells isolated from HGC-27, MGC-803, and MKN-45 gastric cancer cell lines expressed significantly higher levels of CD44, CD24, and CD133 proteins.
In the present work and a previous study, we successfully isolated GCSC spheres from gastric cancer cell lines, MKN28 and SGC-7901 [
11,
25], and found that GCSCs express high levels of stem cell surface markers, including CD24, CD44, and OCT4. In addition, we found that GCSC spheres overexpress CBX7, suggesting that CBX7 may also participate in the regulation of GCSC characteristics. The function of CBX7 in malignant cancers is not clear and may depend on cellular context [
19‐
23]. High expression of CBX7 was observed in tumors of blood system and prostate cancer, suggesting an oncogenic role. However, in some other cancers, lower expression of CBX7 was observed. Previously, it was reported that CBX7 was highly expressed in gastric tumors and that it may have an oncogenic role similar to other PcG proteins [
23].
In the current study, we revealed that high expression of CBX7 in gastric carcinoma was positively correlated with lymph node metastasis and clinical stage. Functional studies showed that CBX7 could promote the growth of gastric cancer cells and colony formation, suggesting an oncogenic role in gastric carcinoma. In this regard, our experiments showed that overexpression of CBX7 promoted the formation of CSC-like spheres and facilitated the growth of in vivo subcutaneous transplanted tumor, cell invasion and migration, resistance to chemotherapy, and expression of stem cell markers, while downregulation of CBX7 could inhibit these oncogenic characteristics. Moreover, there was a positive correlation between the expression of CBX7 and stem cell markers OCT4 and CD133 in gastric carcinoma tissues. These results suggested that CBX7 positively regulates cancer stem cell characteristics of gastric cancer. Next, we investigated the potential mechanisms by which CBX7 regulates stem cell-like characteristics of gastric cancer cells. A previous study showed that CBX7 might inhibit the expression of p16 to prolong cellular life span [
41], promote oncogenic phenotypes in lymphoma [
33], and control the growth of normal and tumor-derived prostate cells [
27]. BMI1-mediated suppression of p16INK4a is also known to regulate stem cell phenotypes [
42]. Therefore, it is possible that CBX7 also regulates stem cell-like characteristics of gastric cancer via the repression of p16. Indeed, our studies strongly suggest that CBX7 regulates expression of p16 in p16-expressing gastric cancer cell lines SGC-7901 and AGS. Further functional reconstitution experiments found that p16 knockdown could partially restore the decreased cell proliferation, serum-free sphere formation, resistance to chemotherapy, and invasion and migration of cells caused by CBX7 gene silencing. Intriguingly, some reports suggested a direct involvement of p16 in the regulation of NF-κB pathway [
40,
43,
44] These findings highlight a potential interplay between CBX7-mediated repression of p16 and activation of AKT-NF-κB signaling. Indeed, we found that overexpression of mAKT could only partially restore the expression of p-p65 following CBX7 depletion in GC cells, suggesting a potential role of p16 in the repression of NF-κB pathway (Fig.
6a, c). These findings imply a complicated role of p16 in CBX7-intiated gastric cancer progression. In summary, our experimental results showed that CBX7 could regulate gastric cancer stem cell phenotype by inhibiting p16INK4a.
Given the facts that p16 knockdown can only partially restore GCSC characteristics in CBX7-deficient gastric cancer cells and that CBX7 can also regulate GCSC phenotype in p16-defcient MKN-28 gastric cancer cells, we speculate that CBX7 may regulate gastric cancer stem cell properties through pathways that are independent of p16. AKT and ERK signaling pathways are known to regulate cell proliferation and apoptosis, epithelial mesenchymal transition (EMT), tumor angiogenesis and metastasis, and resistance of chemotherapy through a series of downstream molecules [
45‐
48]. AKT is also a downstream effector of BMI1, which is a member of the PcG family and a well-known regulator of CSC phenotype. A study by Guo et al. found that BMI1 could promote breast cancer cell proliferation through the activation of AKT [
32]. Because both BMI1 and CBX7 are essential constituents of the PRC1, we speculated that CBX7 might function through AKT and EKR pathways. In this study, we found that overexpression of CBX7 in gastric carcinoma cells led to increased expression of phosphorylated AKT (pAKT) and phosphorylated EKR (pERK), while knockdown of CBX7 decreased the levels of pERK and pAKT, suggesting that CBX7 can regulate AKT and EKR pathways. Furthermore, expression of CBX7 positively correlates with the expression of pAKT in gastric carcinoma but not with the expression of pERK, suggesting that AKT may be an important downstream effector of CBX7 similar to BMI1. Hence, we further studied the role of AKT as a downstream effector of CBX7 and found that treatment with PI3K/Akt inhibitor LY294002 could partially inhibit enhanced gastric cancer cell proliferation, migration and invasion, and ability of sphere formation induced by high expression of CBX7, while exogenous high expression of constitutively active AKT (mAKT) can reverse decreased gastric cancer cell proliferation, migration and invasion, and ability of sphere formation due to CBX7 interference. These experimental results validated the hypothesis that CBX7 could regulate proliferation and stem cell-like properties of gastric cancer cells through the activation of Akt pathway. Further biochemical analyses suggested that CBX7 could activate AKT pathway via the repression of PTEN as reported for BMI1 [
49]. However, CBX7 does not appear to directly bind to PTEN promoter in gastric cancer cells (data not shown).
We previously reported that BMI1 could regulate stem cell properties of gastric cancer through the upregulation of miR-21 by activating AKT-NF-κB pathway. Because CBX7 can also activate AKT, we speculated that CBX7 might also upregulate miR-21 by regulating AKT-NF-κB pathway and that CBX7-intiated AKT-NF-κB-miR-21 pathway might contribute to GCSC phenotype. Indeed, we found that PI3K/Akt-specific inhibitor LY294002 could partially inhibit the upregulation of pP65 and the enhanced transcriptional activity of NF-κB caused by the upregulation of CBX7, while exogenous expression of pSRα-mAkt can partially restore the inhibition of pP65 by the downregulation of CBX7. Furthermore, PI3K/Akt-specific activator SC-79 can restore the transcriptional activity of NF-κB caused by the downregulation of CBX7. Along these lines, we also found that PI3K inhibitor INK1197 and NF-κB inhibitor PDTC treatment can inhibit the increase of miR-21 expression induced by CBX7 overexpression (Fig.
3e, f). These results suggested that CBX7 regulated miR-21 expression via AKT-NF-κB pathway, which in turn could possibly regulate stem cell-like properties of gastric cancer cells. Indeed, exogenous miR-21 overexpression can reverse the inhibition of stem cell-like properties (cell self-renewal, cell migration and resistance to chemotherapy) caused by CBX7 interference. These results suggested that CBX7 might regulate the stem cell-like characteristics of gastric cancer cells through miR-21. In summary, we propose that CBX7-AKT-NF-κB-miR-21 pathway is an important determinant of GCSC phenotype and that targeting this pathway may provide new therapeutic opportunities for gastric cancer treatment.