Introduction
Nasopharyngeal carcinoma (NPC) is a highly metastatic cancer that originates from the epithelial lining of the nasopharynx. It exhibits a marked ethnic and geographic distribution as a majority of the cases are reported in China, Southeast Asia and North Africa (1). Unfortunately, due to its non-specific early symptoms which are often misdiagnosed as minor illnesses such as a 'cold', NPC is often diagnosed at an advanced stage (2). Thus, the opportunity window for treating this disease has been seriously jeopardized. Furthermore, the role of surgery is limited and is reserved for recurrent or unresponsive tumors. Currently, the main treatment for NPC is radiotherapy, and for relatively advanced cases, combined radio-chemotherapy may increase the survival rate (3). Although advances in radiotherapeutic and comprehensive chemotherapeutic strategies have greatly improved the outcome of patients with primary NPC, the overall survival rate of advanced stage NPC patients is still unsatisfactory (4). Therefore, revealing the molecular mechanisms underlying the tumorigenesis of NPC is critical for developing novel therapeutic targets and treatment approaches.
Increasing evidence indicates the existence of cancer stem cells (CSCs) which are believed to play a critical role in the development of cancers (5–7). CD44, a member of the homing cell adhesion molecule family, is known to be overexpressed in ovarian, pulmonary, gastrointestinal and various other tumors (8–11). It is not only associated with proliferation, differentiation and invasion (12), but also is reported as a specific surface marker for various CSCs (13,14). A number of studies have shown that CD44 is overexpressed in NPC, and it is a surface marker of NPC CSCs (15–17). In our previous study, we demonstrated that CD44 was overexpressed in the human NPC SUNE-1 5–8F cell line, and these cells had stem cell-like characteristics in vitro (15).
B-cell-specific Moloney murine leukemia virus insertion site 1 (Bmi-1) is one member of the PcG family and acts as a transcription repressor, participating in the regulation of a series of biological processes, and is considered to be a stem-related gene (18). It has been reported that Bmi-1 is overexpressed in tumors of leukemia, stomach cancer, cervical cancer and head and neck squamous cell carcinoma (HNSCCs) tissues (19–22), and such upregulation is highly correlated with the maintenance of self-renewal and differentiation of CSCs. It has been noted that Bmi-1 expression was significantly higher in CD44+ NPC cells than in CD44− NPC cells (15). Collectively, Bmi-1 may play an important role in the maintenance of stem cell-like characteristics of CD44+ NPC cells, and targeting Bmi-1 in CD44+ NPC cells may serve as an important pathway for the prevention and treatment of NPC.
Although we previously confirmed that silencing of Bmi-1 resulted in CD44+ NPC CSC-LC sensitivity to radiotherapy (23), the role of Bmi-1 in NPC development has not been fully elucidated, and whether Bmi-1 can serve as a therapeutic target for NPC remains to be assessed. In the present study, scratch wound healing assay, together with Transwell migration and invasion assays were used to observe invasion and migration capacity; flow cytometry was used to analyze the cell apoptosis status; tumorigenesis in nude mice was used to assess tumorigenicity; and immunohistochemical techniques were used to detect the expression of CD44 in tumor tissues. Our results demonstrated that silencing of Bmi-1 significantly suppressed migration, invasion and tumorigenesis capabilities in vitro and increased apoptosis. Moreover, Bmi-1 silencing in CD44+ NPC cells resulted in impaired tumorigenicity and delayed onset of xenograft tumors in mice. These findings demonstrate that Bmi-1 plays a critical role in NPC development, and that Bmi-1 has potential clinical value in NPC therapy. The results presented herein provide initial data for future gene therapy of NPC.
Materials and methods
Cell culture
A stable Bmi-1-knockdown (KD) cell line was obtained by transfecting CD44+ cells with retroviral vector Bmi-1 short hairpin RNA (shRNA) (23). Negative control cells (NC) were transfected with an empty retroviral vector and the cells without transfection were used as the blank control (CON). CD44− cells were sorted according to previously reported procedures (15). All of the cells were cultured as previously described (23).
Cell proliferation as detected via CCK-8 assay
Cell growth was assessed by the Cell Counting Kit-8 (CCK-8; Dojindo, Japan) assay. The three groups of cells were seeded in 96-well plates at a density of 800 cells/well, which contained 100 µl serum-free Dulbecco's modified Eagle's medium (DMEM)/F12 (1:1) medium per well supplemented with 10 ng/ml basic fibroblast growth factor (bFGF), 20 ng/ml EGF, 5 µg/ml insulin, 100 U/ml penicillin and 100 µg/ml streptomycin (18 wells for each group). The wells without added cells but with medium were provided as a control simultaneously. All cells were cultured in a 5% CO2 humidified incubator at 37°C. A solution of 10 µl CCK-8 was added to each well on day 1, 3, 5 and 7, respectively, and incubated for another 2 h before measuring the absorbance at OD450 with a microplate reader (Bio-Tek Instruments, Winooski, VT, USA). The proliferation curve was plotted according to the mean absorbance.
Scratch wound healing assays
For the wound healing assays, 5×105 cells were plated in 6-well dishes. Each group was provided with two parallel controls. Twenty-four hours after the cells reached 100% confluency, 2 ml/well serum-free RPMI-1640 culture medium was added to eliminate the effect of proliferation, and a mimicking scratch wound was made using a 200-µl pipette tip, followed by washing the cells gently with PBS to remove floating cells that were detached by scratching. The cells were maintained in a 5% CO2 humidified incubator at 37°C for 24 h, and images were captured at 0 and 24 h, respectively using an inverted microscope (magnification, ×200; Olympus, Japan). Image-Pro Plus 6.0 software was used to calculate the migration area.
Transwell cell migration and invasion assays
For the cell Transwell migration assay, 4×104 cells that were starved for 24 h in 200 µl serum-free RPMI-1640 medium were seeded on the basement membranes of Transwell chambers (8-µm pore size; Corning, USA). In the lower chamber, 700 µl RPMI-1640 with 10% FBS was added as a chemoattractant. After the cells were incubated for 24 h in a 5% CO2 humidified incubator at 37°C, the basement membranes were washed with PBS, and the non-migrated cells on the top surface of the thin basement membranes were scraped off with cotton swabs. The migrated cells adhering to the lower surface were fixed with methanol, stained with 0.1% crystal violet solution and counted under a microscope (×100) in five random and visual fields per well.
For the cell Transwell invasion assay, the procedure was similar to that of the cell migration assay above, except that the Transwell basement membranes were precoated with 40 µl extracellular matrix (ECM, 1 mg/ml) gel (Sigma, USA) and the cells were resuspended at a density of 2.5×105 cells/ml.
Apoptosis analysis
Cells were seeded in 6-well plates at 3×105 cells/well and harvested after incubation for 24 h. Single-cell suspensions were then prepared in 500 µl of binding buffer containing 5 µl 7-AAD dye, and incubated for 15 min in the dark at room temperature. Subsequently, 1 µl Annexin V-PE was added, and the cells were incubated for another 15 min in the dark before being analyzed 1 h later using flow cytometry (FCM).
Analysis of tumorigenesis in vivo in nude mice
A total of 5×103 cells (CD44− 5–8F, KD, NC or CON) for each animal in 0.2 ml RPMI-1640 medium was subcutaneously injected into the left forelimb of 4- to 6-week-old healthy male BALB/c-nu mice, purchased from the Experimental Animal Center of Wuhan University, China (weighing between 16–20 g). The mice were randomly divided into 4 groups with 4 mice in each group. All mice were maintained in a barrier facility with a SPF aseptic environment. After being observed for 4 weeks, the mice were sacrificed by cervical dislocation. All animal studies were conducted in accordance with the principles and procedures of the National Institutes of Health Guide for the Care of Laboratory Animals (permit no. SYXK2010-0057). All applicable international, national, and institutional guidelines for the care and use of animals were followed, and all efforts were made to minimize suffering. Tumor tissues were dissected and measured with a Vernier caliper to record the greatest length (a) and width (b), and the tumor volume was calculated via the formula V = 1/2 (a × b2). The tumor blocks were then fixed and made into tissue sections for immunohistochemical staining.
Immunohistochemical staining
Tumor tissues were embedded in paraffin, and sectioned at 4 µm to detect CD44 expression. After deparaffinization, rehydration and antigen retrieval, the tissues were incubated with 3% H2O2 for 10 min to eliminate endogenous peroxidase and blocked with 5% BSA for 20 min. The sections were then incubated with the primary anti-human CD44 antibody (1:100) at 4°C overnight and the secondary horseradish peroxidase-conjugated antibody (1:200) for 50 min at 4°C. PBS instead of the primary antibody was performed as a negative control. Finally, the sections were chromogenized by a DAB solution, counterstained with hematoxylin (Sigma), dehydrated, transparented with xylene, cemented with neutral gum and visualized by an inverted microscope. Image-Pro Plus 6.0 software was applied to analyze the immunohistochemical images. The average optical density value of each image (AIOD) = integrated optical density (IOD)/area (SUM). Both moderate and intensive stainings were judged as positive staining. When the section presented with <10% brown-colored granules, it was considered to be negative for protein expression, while >10% was considered as positive for protein expression: 10–25% (+); >25–50% (++); >50–75% (+++); >75% (++++).
Statistical analysis
Statistical analysis was performed by GraphPad Prism 5 software. When comparisons were made among groups, univariate repeated measure analysis of variance was used. Comparisons of two-sample data were made using independent samples t-test. Data are presented as the mean ± SD, with P<0.05 indicating statistical significance.
Results
The proliferative capability of CD44<sup>+</sup> NPC CSC-LCs is suppressed
The CCK-8 assay demonstrated that there was an increasing trend in the viability of all cells in the Bmi-1 KD, NC and CON group. Cell proliferation curves plotted based on the absorbance values on day 1, 3, 5 and 7 are shown in Fig. 1A. The curves indicated that the proliferative capability of the Bmi-1-KD cells was significantly decreased in vitro (P<0.001) when compared with the proliferation capability of the NC and CON cells, while no significant difference was observed between the NC and CON cells (P>0.05, Fig. 1B). Thus, gene silencing of Bmi-1 reduced the proliferation activity of the CD44+ NPC CSC-LCs.
Gene silencing of Bmi-1 suppresses the migration and invasion capacities of the CD44<sup>+</sup> NPC CSC-LCs
We next investigated whether Bmi-1 knockdown affects cell migration and invasion. To assess changes in cell migration, we performed a scratch wound healing assay. Confluent monolayers of Bmi-1 KD, NC and CON cells transfected with shRNA were scratched, and the migration of the cells into the free space was observed 24 h later under an inverted microscope. Obviously, in the NC and CON group, most of the scratched wounds were already close to fusion, while both sides of the cell fusion trend were inconspicuous in the Bmi-1 KD group (Fig. 2A and B and Table I). The migration area calculated by IPP 6.0 software demonstrated that the migration area of the Bmi-1-KD cells was significantly less than that of the CON and NC cells. The results preliminarily indicated that interference of the Bmi-1 gene evidently decreased the migration capacity of the CD44+ NPC CSC-LCs.
To further confirm this result, we performed a Transwell migration assay. The numbers of migrated cells that adhered to the lower surface of the Transwell chambers were counted under a microscope in five randomly chosen visual fields per well within the same area. Compared with the NC and CON groups, the migration of the Bmi-1-KD cells was significantly suppressed (P<0.05), but no significant difference was found between the NC and CON cells (P>0.05, Fig. 2C and Table II). These results further confirmed that silencing of Bmi-1 could decrease the CD44+ NPC CSC-LC migration in vitro.
To investigate the effects of Bmi-1 suppression on cell invasion, we monitored the numbers of cells which passed through the basement membrane coated with ECM gel in all groups. Strikingly, the number of invasive Bmi-1-KD cells decreased significantly when compared with the CON and NC groups (P<0.05). Moreover, the ability of invasion between the CON and NC group showed no difference (P>0.05, Fig. 2D and Table II). Therefore, our results demonstrated that suppression of Bmi-1 expression reduced the invasive ability of the CD44+ NPC CSC-LCs.
Gene silencing of Bmi-1 induces the apoptosis of the CD44<sup>+</sup> NPC CSC-LCs
It is well known that cell apoptosis plays a pivotal role in the occurrence and development of tumors. In the present study, the early apoptosis rates of the three groups of cells were analyzed by FCM analysis following Annexin V-PE/7-AAD staining. The results demonstrated that the percentage of early apoptotic cells (1.37±0.15%) in the Bmi-1-KD group was evidently higher than that in the NC group (0.73±0.12%) or CON group (0.54±0.11%) (Fig. 3 and Table III). The Bmi-1-KD group displayed an increase in the apoptosis rate when compared with that of the NC and CON groups, with statistical significance (P<0.001). No significant difference was noted between the NC and CON group (P>0.05) suggesting that the introduction of Bmi-1 shRNA had an obvious inductive effect on the apoptosis of the CD44+ NPC CSC-LCs.
Gene silencing of Bmi-1 reduces the tumorigenicity of CD44<sup>+</sup> NPC CSC-LCs in vivo
The in vitro experiments showed that suppression of Bmi-1 expression inhibited the proliferation, migration and invasion of CD44+ NPC CSC-LCs. Based on this, we evaluated whether the downregulation of Bmi-1 expression could inhibit the tumorigenic capability of the CD44+ NPC CSC-LCs in vivo. Four weeks after the subcutaneous inoculation, the observation of tumor formation was carried out in the nude mice. None of the mice expired during the entire observation. In the CON and NC groups, small subcutaneous protrusions began to form (after subcutaneously inoculation on 6 and 8 days, respectively), and the subcutaneous tumors gradually clustered. After observation for 4 weeks, tumor formation was 4/4 in the CON group and 3/4 in the NC group. Tumor volumes were 1,351.68, 500.00, 444.36 and 51.64 mm3 in the CON group, with 626.69, 607.50 and 336.51 mm3 in the NC group. The tumor volume between the CON and NC group had no significant difference (P>0.05, Fig. 4 and Table IV). However, no protrusions appeared in the CD44− 5–8F group and KD group at the end of the observation. This suggests that CD44+ NPC cells displayed a stronger tumorigenic ability than CD44− NPC cells, while gene silencing of Bmi-1 weakened the tumorigenicity of the CD44+ NPC CSC-LCs in vivo.
Expression of CD44 in the tumor tissues
Xenografts were dissected, sectioned and paraffin-embedded after subcutaneous inoculation for 4 weeks. Immunohistochemical assay was used to detect CD44 expression on the cell membrane. As shown in Fig. 5 and Table V, CD44 was positive with a 50–75% (+++) expression in the tumor tissues formed by the CON and NC cells and there was no statistically significant difference between the two groups (P>0.05). This suggests that the CD44+ NPC CSC-LCs possessed stem-like cell characteristics with the potential to differentiate and Bmi-1 was necessary for the maintenance of the stem cell-like characteristics.
Discussion
CSCs are a specific subset of transformed cells which are biologically distinct from other cancer cell types. It has been proposed that CSCs are able to sustain primary tumor growth according to a hierarchical pattern and due to their ability to undergo unlimited self-renewal. Increasing evidence has shown that CSCs are responsible for tumor initiation, tumor recurrence and metastasis (5–7). Thus, designing agents that target CSCs may lead to more efficient cancer therapy.
Currently, CD44+ cells are most widely reported to possess stem cell properties in NPC. Lun et al observed that CD44+ NPC cells displayed a higher sphere formation rate than that of CD44− cells (16). Janisiewicz et al (17) also showed that CD44+ cells exhibited a stronger tumorigenic ability in a xenograft model. In addition, CD44+ cells dissociated from EB virus-related NPC patients were found to be associated with local treatment failure and decreased time to relapse. Therefore, targeting CD44+ NPC cells may provide new tools for the prevention and treatment of NPC. In a previous study, we enriched CD44+ NPC CSC-LCs from the human NPC SUNE-1 5–8F cell line and characterized their CSC properties (15). In the present study, we used CD44+ NPC CSC-LCs as an NPC CSC model to elucidate their tumor initiation mechanisms.
Bmi-1 overexpression was found to regulate stem cell self-renewal and malignant transformation in multiple carcinomas, including medulloblastoma stem cells (24) and prostate stem cells (25). Proctor et al (26) found that knockout of Bmi-1 reduced secondary and tertiary tumor sphere formation in primary human pancreatic cancer, inhibited the growth of primary pancreatic xenografts, and decreased the percentage of CSCs in tumor tissues. Yao et al (27) constructed a recombinant Bmi1-RNAi lentiviral expression system and stably transfected laryngeal cancer Hep-2 cells, decreasing the proliferation and invasion ability. Chen et al (28) silenced Bmi-1 expression in ALDH1+ primary HNSCC cell subsets by constructing Bmi-1 shRNA lentiviral vector. They found that Bmi-1 knockout distinctly inhibited the tumorigenicity of ALDH1+ HNSCC cells (29). In our previous studies, we identified that Bmi-1 is overexpressed in CD44+ NPC and silencing Bmi-1 increased the radiosensitivity of CD44+ NPC CSC-LCs (23). Nevertheless, the molecular mechanisms by which Bmi-1 affects CD44+ CSC-LCs in NPC have not been previously identified. In the present study, our results showed that knockdown of Bmi-1 by shRNA significantly decreased tumor proliferation, migration and invasion of CD44+ NPC CSC-LCs in vitro.
The migration and invasion of cancer cells into the vascular system is a necessary step in tumor metastasis. Furthermore, the migratory and invasive abilities of tumor cells reflect the metastatic potential of the tumor. It has been found that Bmi-1 siRNA significantly weakened the migration of HeLa cells in vitro and in vivo (30). Yao et al (27) constructed a recombinant Bmi1-RNAi lentiviral expression system and stably transfected laryngeal cancer Hep-2 cells. After silencing Bmi-1 expression, the invasive ability of cells was markedly reduced. In addition, Chen et al (28) silenced Bmi-1 expression in ALDH1+ primary HNSCCs cell subsets by constructing the Bmi-1 shRNA lentiviral vector. They found that Bmi-1 knockout distinctly inhibited the invasiveness of ALDH1+ HNSCC cells (29). In a recent study, Zhu et al (31) reported that knockout of Bmi-1 in the human bladder CSC-LC side population (SP) cells induced suppression of migration in vitro. In the present study, for the first time, we exploited a scratch wound healing assay to evaluate the migration ability of CD44+ NPC CSC-LCs. Our study demonstrated that the downregulation of Bmi-1 by shRNA reduced the migration of CD44+ NPC CSC-LCs in vitro. We also saw similar results in the Transwell cell migration experiment. Moreover, the invasive ability of the cells was markedly reduced. Therefore, our results are consistent with the findings by other groups in different tumor types. Although further studies are still needed, these results provide compelling evidence that knockdown of Bmi-1 may contribute to decreased CD44+ NPC CSC-LC metastasis, and suggest that Bmi-1 is significantly correlated with the metastatic ability of CD44+ NPC CSC-LCs.
It is well known that cell apoptosis plays a pivotal role in the occurrence and development of tumors. Bmi-1 has been reported to be associated with the protection of cancer cells from apoptosis. Xu et al (32) found that Bmi-1 siRNA effectively inhibited cell proliferation and induced apoptosis in MCF-7 cells, and similar results have also been reported by other groups in different tumor types (33). However, the molecular mechanisms of the Bmi-1 function in tumor cells are not completely understood. A few possible mechanisms for Bmi-1-suppressed apoptosis have been proposed. Qin et al (34) reported that the expression of phospho-AKT and anti-apoptotic protein BCL-2 were downregulated in NPC cells when Bmi-1 expression was inhibited, whereas, the apoptosis inducer BAX was upregulated. It was also found that Bmi-1 knockout weakened NF-κB signaling and reduced VEGF-C expression (35). Some studies revealed that increased expression of Bmi-1 in primary human tumor cells led to the downregulation of the INK4a-ARF locus and thereby impacted the p14ARF-Mdm2-p53 pathway (36–38). Consistent with these reports, our previous results revealed that silencing of Bmi-1 resulted in upregulation of p14ARF, p16INK4a and p53 at the protein level (23). As expected, our apoptosis assay indicated that silencing of Bmi-1 induced cell apoptosis in the CD44+ NPC CSC-LCs. These results suggest that Bmi-1 may mediate cell apoptosis of CD44+ NPC CSC-LCs by suppression of the pINK4a-pARF-p53 pathway. In contrast, some studies demonstrated that there was no obvious correlation between Bmi-1 and p16INK4a (39). Yao et al (40) also reported no significant relationship among Bmi-1, p14ARF and p53 in gastric adenocarcinoma. These conflicting data indicate that different mechanisms may be involved in the development of cancers.
Given the role of Bmi-1 in CD44+ NPC CSC-LCs in vitro, we investigated whether the same effects appear in vivo. These observations were further confrmed by xenograft transplantation in nude mice, where Bmi-1 knockdown in CD44+ NPC CSC-LCs resulted in the failure to develop tumors. It is worth noting that in comparison with the results of proliferative capability and apoptosis, the tumorigenicity in nude mice (Fig. 4B) showed a marked difference between NC and KD cells. In combination with the results in vitro, we can conclude that the Bmi-1 gene plays an important role in the maintenance of stem cell-like characteristics of CD44+ NPC cells. We hypothesized that in Bmi-1-deficient CD44+ NPC cells, the stem cell-like characteristics including those that may be responsible for tumor escape are lost. Therefore, Bmi-1-deficient tumors could be recognized and eliminated by activating the immune system in vivo. However, we did not observe the effects of the immune system in vitro. To our knowledge, no explicit molecular mechanism has been proposed for this phenomenon, and further studies are still needed to investigate the possible mechanisms.
In conclusion, our results revealed that downregulation of Bmi-1 expression inhibited the proliferation, migration and invasion of CD44+ NPC CSC-LCs in vitro, followed by cell apoptosis. In addition, Bmi-1 achieved these functions to a great extent through the suppression of the pINK4a-pARF-p53 pathway. Bmi-1 silencing in CD44+ NPC cells also resulted in impaired tumorigenicity and delayed onset of xenograft tumors in vivo. These results provide important insights into the role of Bmi-1 in the occurrence and development of NPC. Based on our findings, regulation of Bmi-1 in CD44+ NPC CSC-LCs may provide a potential molecular target for NPC therapy, and targeted silencing of Bmi-1 by shRNA may have future clinical implications in NPC therapy.