Abstract
Hepatocellular carcinoma is highly resistant to chemotherapy. Research data supported that cancer stem cells (CSC) may be responsible for the chemoresistance and strategies that suppress CSCs stemness could also inhibit the drug resistance. In this study, we found that nuclear receptor binding protein 2 (NRBP2) expression was downregulated in the CD133+ hepatocellular carcinoma CSCs. Most adjacent noncancerous liver tissue analyzed expressed higher level of NRBP2 compared with cancerous tissue in hepatocellular carcinoma patients, and high NRBP2 expression indicated a better prognosis. Real-time PCR results showed that NRBP2 negatively correlated with stemness-related genes, including Oct3/4, Nanog, Notch1, Ep300, and CD133 mRNA expression. High NRBP2 expression in hepatocellular carcinoma cells downregulated CK19 protein expression, inhibited tumorsphere formation, and tumorigenesis ability, indicating that high NRBP2 expression restrains the hepatocellular carcinoma cell stemness. Overexpression of NRBP2 reduced the IC50 of sorafenib in hepatocellular carcinoma cells, and NRBP2 expression was negatively correlated with hepatocellular carcinoma cell resistance to the chemotherapy agents, including cisplatin and the Akt signaling inhibitor perifosine. Coimmunoprecipitation results showed that NRBP2 could bind with Annexin A2 (ANXA2) and inhibit ANXA2 expression. Coexpression of ANXA2 restored the chemoresistant ability in NRBP2-overexpressing hepatocellular carcinoma cells. Further analysis showed that NRBP2 downregulated Akt and its downstream signaling target Bad phosphorylation level. ANXA2 coexpression partially restored the Akt phosphorylation. Analysis of the expression of Bcl2 family proteins showed that NRBP2 may increase hepatocellular carcinoma cell chemosensitivity by regulating expression of survival proteins involved in the Akt and Bcl2 pathway. These results suggest that NRBP2 plays an important role in the tumor progression and chemotherapeutic resistance of hepatocellular carcinoma. Cancer Res; 76(23); 7059–71. ©2016 AACR.
Introduction
Hepatocellular carcinoma is a heterogeneous disease that is the second most common cause of cancer-related deaths worldwide (1). Curative resection or transplantation applies to only approximately 30% of patients and for advanced hepatocellular carcinoma patients, systemic chemotherapy are required (2). However, hepatocellular carcinoma patients frequently develop chemoresistance and fail in their treatment (3, 4).
The mechanisms by which tumor cells become chemoresistant are complex, including increased drug efflux, changes of anticancer drugs' targets, dysregulation of key signaling pathways like PI3K/Akt and STAT3 pathways, etc. (5, 6). Recent research suggests that hepatocellular carcinoma is organized by cancer stem cells (CSC), which act as a main player for chemoresistance. Several surface proteins are reported to be CSC markers in hepatocellular carcinoma, including CD133, EpCAM, CD90, OV6, CD44, etc. Strategies target hepatocellular carcinoma CSCs could also impair hepatocellular carcinoma chemoresistance (7). Liu and colleagues reported that HDAC inhibitors reduced hepatocellular carcinoma cell stemness. Knockdown of HDAC3 enhanced the sensitivity of hepatocellular carcinoma cells to sorafenib (8). Our previous research showed that forced differentiation of CD133+ hepatocellular carcinoma CSC inhibits these cells self-renew and reduced their chemoresistance (9). A comprehensive understanding of mechanisms regulating chemoresistance in hepatocellular carcinoma would increase the effectiveness of chemotherapy agents and improve prognosis.
Nuclear receptor binding protein 2 (NRBP2) shows 59% amino acid similarity to nuclear receptor binding protein 1 (NRBP1) and was first identified by the Mammalian Gene Collection (MGC) program (10, 11). NRBP2 knockdown in zebrafish embryos by splice-blocking morpholinos can lead to global delay and embryonic lethality (12). NRBP2 expression in the brain is associated with neuronal differentiation in both normal and malignant brain tissue (13). In this study, we demonstrated that NRBP2 expression was downregulated in the CD133+ hepatocellular carcinoma cells and involved in the regulation of hepatocellular carcinoma cell stemness-related gene expression. High NRBP2 expression increased hepatocellular carcinoma chemosensitivity to cisplatin and perifosine, which may be due to its regulation on Akt signaling pathway.
Materials and Methods
Cell culture
The human hepatocellular carcinoma cell line SMMC-7721 was obtained from the Cell Bank of the Institute of Biochemistry and Cell Biology (Shanghai, China) in 2009; MHCC-97L cells were kindly provided by the Liver Cancer Institute of Zhongshan Hospital at Fudan University (Shanghai, China) in 2008; PLC/PRF/5 was purchased from the ATCC in 2010; and Huh7 cells were obtained from the Riken Cell Bank in 2010; These cells lines were maintained in DMEM (Sigma-Aldrich) containing 10% FBS (Hyclone) at 37°C in 5% CO2. All cell lines were thawed fresh every 2 months and used within 20 passages. These cell lines were mycoplasma-free and authenticated by their examination of morphology and growth profile.
Tumorsphere formation
Hepatocellular carcinoma single-cell suspensions were prepared, and the suspension was added to ultralow-attachment multiwell plates (Costar) in serum-free chemically defined medium (CDM; ref. 9). Half volume of the flesh medium was changed every second day.
Real-time reverse transcription-PCR
Total RNA was extracted using TRIzol reagent (Invitrogen) and reverse transcribed using the PrimeScript RT Reagent Kit (Perfect Real Time; TaKaRa Biotechnology). The real-time PCR assay was performed as described previously (9). The primers for the target genes are listed in Supplementary Table S1.
Western blotting
Cell lysates were separated by SDS-PAGE and transferred onto nitrocellulose membranes. The membranes were incubated with a primary antibody overnight at 4°C and then probed with HRP-conjugated secondary antibodies. The immunoreactive blots were visualized using an enhanced chemiluminescence reagent (Pierce). The antibody information is listed in Supplementary Table S2.
Plasmid constructs, lentivirus production, and cell transfection
The full-length human NRBP2 ORF with an HA tag, the ANXA2 ORF with a FLAG tag, and the NRBP2 promoter sequence were generated and cloned into the lentiviral vector pWPLX (Addgene). Two siRNAs targeting NRBP2 and a generic negative control (NC) sequence were synthesized by GenePharma. Virus packaging and cell transfection were performed as described previously (14). Primers for cloning and the NRBP2 promoter sequence are provided in Supplementary Tables S3–S5.
Cell isolation by FACS or magnetic-activated cell sorting
Isolation and analysis of the hepatocellular carcinoma cells were done as described previously (9). For CSCs isolation, hepatocellular carcinoma cells were labeled directly with PE-conjugated anti-human CD133/1 antibody (AC133; Miltenyi Biotec), according to the manufacturer's instruction and sorted by FACS or the EasySep PE Selection Kit (StemCell Technologies), according to the manufacturer's instructions, and NRBP2high/low cells were sorted according to the GFP protein expression by FACS. The purity for sorted cells was evaluated by flow cytometry, and more than 90% of cells with viability determined by the Trypan blue staining were acceptable for the following experiments.
Plate colony formation assay
Two thousand cells were seeded and cultured in normal condition for 2 weeks, then fixed using 10% formaldehyde for 30 minutes. The cell colonies were stained using GIEMSA (Sigma-Aldrich) for 30 minutes. After washing, the cell colonies were quantified.
Drug sensitivity assays
The sensitivity of each cell population to chemotherapeutic drugs was measured by the 3-(4,5-dimethyltiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich) assay and each drug's half-maximal inhibitory concentration (IC50) were calculated as described previously (9).
Coimmunoprecipitation assay
Coimmunoprecipitation (co-IP) assays were performed as described previously (15). Briefly, cells were harvested in RIPA lysis buffer (Santa Cruz Biotechnology) for 50 minutes on ice and centrifuged at 5,000 × g for 15 minutes. Antibodies against HA-tag and Flag-tag with protein G agarose beads (Sigma-Aldrich) were added and incubated overnight at 4°C. After washing, the complexes were subjected to following Western Blot or mass spectrometric analysis.
Mass spectrometric analysis and protein identification
The co-IP proteins from anti-HA or IgG control groups were analyzed according to the procedures described previously (16). Briefly, the proteins resulting from the co-IP assay were digested with trypsin (25:1, w/w) at 37°C for 20 hours with gentle shaking. The tryptic peptides were analyzed using a NanoLC system (NanoLC-2D Ultra, Eksigent) equipped with Triple TOF 5600 mass spectrometer (AB SCIEX). Protein identification was performed via database searching against the Swiss-Prot Homo sapiens protein databases, with ProteinPilot 4.5 software (AB SCIEX).
Tumor xenograft models
Six- to 8-week-old male NOD/SCID mice were injected with hepatocellular carcinoma cells indicated subcutaneously in the armpit area on one side, and the control cells were injected on the other side. After observation for 6–8 weeks, the animals were sacrificed. Immediately after killing, xenograft tumors were weighed. All animal experimental protocols were approved by the Institutional Animal Care and Use Committee of the Shanghai Jiao Tong University School of Medicine (Shanghai, China) and the experiments were carried out in accordance with the approved guidelines.
IHC
All of 229 pairs of hepatocellular carcinoma tissue samples were obtained from the Qidong Liver Cancer Institute. Informed consent was obtained from all patients, and the study was approved by the Ethics Committee of Shanghai Jiao Tong University (Shanghai, China). Anti-NRBP2 polyclonal antibody (ab172866) was purchased from Abcam. The IHC and signal evaluation were performed according to our previously described procedures (9).
Statistical analysis
The experimental data are presented as the mean ± SD and were analyzed using Student t test. Bivariate correlation analysis was performed using the Pearson correlation method. P < 0.05 was considered statistically significant.
Results
NRBP2 overexpression suppresses stemness-related gene expression in hepatocellular carcinoma cells
Our previous studies have shown that BMP4 treatment induces CD133+ hepatocellular carcinoma CSC differentiation (9). Here, we analyzed the differential gene expression in the CD133+ hepatocellular carcinoma cells forced differentiation via cDNA microarray and found that NRBP2 gene expression was upregulated. Real-time PCR showed that NRBP2 was highly expressed in the CD133− hepatocellular carcinoma cell population compared with its corresponding CD133+ cells in hepatocellular carcinoma cell lines (Supplementary Fig. S1A), indicated that NRBP2 may play a role in the hepatocellular carcinoma CSCs. We then analyzed NRBP2 mRNA expression in the Cancer Genome Atlas (TCGA) database. The results showed that hepatocellular carcinoma patients with a high NRBP2 mRNA expression had a better outcome (Supplementary Fig. S1B).
To further investigate the clinicopathologic role of NRBP2 in hepatocellular carcinoma progression, we determined the expression of NRBP2 in 229 pairs of hepatocellular carcinoma tissues and matched nontumorous liver tissues using immunohistochemical staining. Of the 229 cases, 159 cases (69.43%) had lower NRBP2 expression in hepatocellular carcinoma tissue compared with the corresponding nontumorous liver tissue, 67 cases (29.26%) had unchanged expression, and only 3 cases (1.31%) had a higher NRBP2 expression in tumor tissue (Fig. 1A and B). Patients with high NRBP2 expression showed a better prognosis (Supplementary Fig. S1C), and analysis of the relationship between NRBP2 expression and the clinicopathologic features demonstrated that hepatocellular carcinoma patients below the age of 50 years were more likely to have a low NRBP2 expression and NRBP2 was negatively associated with the histologic grade (Supplementary Table S6), indicating that loss of NRBP2 expression might contribute to hepatocellular carcinoma progression.
Western blotting assays showed that NRBP2 expression was markedly high in SMMC-7721, and which was low in PLC/PRF/5 and Huh7, with MHCC-97L showed a modest expression (Supplementary Fig. S1D and S1E). Then, NRBP2 gene with an N-terminal HA tag was overexpressed in PLC/PRF/5, Huh7 and MHCC-97L cells (PLC-HA-NRBP2, Huh7-HA-NRBP2, 97L-HA-NRBP2). Real-time PCR results showed that stemness-related genes, including Oct3/4, Nanog, Notch1, and tumorsphere-related Ep300 (7, 17, 18), and CD133 mRNA expression were downregulated in NRBP2-overexpressing cells compared with vector control (pWPXL) cells (Fig. 1C). Conversely, NRBP2 knockdown by shRNA led to upregulation of stemness-related genes and CD133 mRNA expression. (Fig. 1D). Cytokeratin 19 (CK19) is a biliary epithelial cell marker and acts as a progenitor cell marker in the liver (19). We analyzed CK19 expression in the NRBP2 overexpression and NRBP2 knockdown hepatocellular carcinoma cells and found that CK19 protein expression was negatively correlated with NRBP2 expression (Fig. 1E and F; Supplementary Fig. S1F), demonstrating that high NRBP2 expression may depress the expression of stemness-related genes in hepatocellular carcinoma cells.
Stemness-related genes are preferentially expressed in sorted NRBP2low cells
To investigate the gene regulation effect in a more natural context, hepatocellular carcinoma cells were stably transfected with an NRBP2-promoter construct driving GFP expression, which acted as an indicator of NRBP2 transcriptional activity. FACS analysis showed that NRBP2 was overexpressed in a subpopulation of hepatocellular carcinoma cells (data not shown). Then, hepatocellular carcinoma cells with high (NRBP2high) and low (NRBP2low) levels of NRBP2 expression were isolated on the basis of GFP expression by FACS sorting. Real-time PCR analysis showed that stemness-related genes (Oct3/4, Nanog, Notch1, Ep300), CK19, and CD133 expression were upregulated in the NRBP2low cells (Fig. 2A). As shown in Fig. 2B, Western blotting results confirmed that NRBP2 protein was upregulated in the NRBP2high cell population and which expressed lower levels of CK19 protein compared with the NRBP2low cells.
We then sorted NRBP2high and NRBP2low cells in MHCC-97L and cultured them for 12 days (Fig. 2C). The purity of the sorted NRBP2low cells was more than 99%. After culturing, the percentage of GFP-positive cells in the NRBP2low cell group returned to the level found before primary sorting. However, many GFP-positive cells remained in the sorted NRBP2high cell group. A secondary sorting was then applied in the NRBP2low cells, and sorted cells were cultured for another 7 days. We found that the GFP-positive cells percentage in the NRBP2low cell group returned again to the level before secondary sorting. A similar result could also be obtained in SMMC-7721 (Supplementary Fig. S2), suggesting that the NRBP2low cells could repopulate the heterogeneity in the hepatocellular carcinoma cell lines.
High NRBP2 expression inhibits the self-renewal capability and tumorigenesis of hepatocellular carcinoma cells
The proliferation ability of NRBP2high hepatocellular carcinoma cells was analyzed by flat-plate colony formation assay. We found that the overexpression of NRBP2 suppressed cell colony formation in PLC/PRF/5 and MHCC-97L cells (Fig. 3A); and as shown in Fig. 3B, there were more colonies in the NRBP2low cell group compared with the corresponding NRBP2high cell group in SMMC-7721 and MHCC-97L, which indicates that high NRBP2 expression inhibited hepatocellular carcinoma cell proliferation in vitro. Self-renewal capability is an important property of tumor stemness (20). We then investigated the self-renewal capability of the NRBP2-overexpressing hepatocellular carcinoma cells using a sphere formation assay. As shown in Fig. 3C, the tumorspheres obtained from NRBP2-overexpressing hepatocellular carcinoma cells were fewer in number and smaller in size than those control cells, indicating that the hepatocellular carcinoma cell stemness was impaired.
The tumorigenesis capacity of NRBP2-knockdown hepatocellular carcinoma cells was analyzed in an immunodeficient mouse xenograft model. We found that most xenografted tumors in the 97L-shNRBP2 group were larger than those in the 97L-shNC group (Fig. 3D and Supplementary Fig. S3A). Then NRBP2high and NRBP2low cells from SMMC-7721 cells were sorted and injected in the NOD/SCID mice subcutaneously. The results showed that most xenografted tumors in the 7721-NRBP2low group were larger than those in the 7721-NRBP2high group (Fig. 3D and Supplementary Fig. S3B). These results demonstrated that NRBP2low cells possess higher tumorigenesis property than NRBP2high cells, suggesting that CSCs may be enriched in the NRBP2low cell population.
NRBP2 enhances the chemotherapeutic sensitivity of hepatocellular carcinoma cells in vitro
Previous studies showed that tumor cell stemness is correlated with their chemoresistant ability. Considering that NRBP2high impaired hepatocellular carcinoma stemness, we then analyzed the relationship between NRBP2 expression and therapeutic agents' sensitivity. MTT assay showed that NRBP2 overexpression slightly but definitely reduced the IC50 of sorafenib in MHCC-97L and PLC/PRF/5 cells. Conversely, knockdown of NRBP2 slightly raised the IC50 of sorafenib in SMMC-7721 and MHCC-97L cells (Fig. 4A). The sorted NRBP2low cells were more resistant to sorafenib treatment compared with the NRBP2high cells from SMMC-7721 and MHCC-97L (Fig. 4B). Annexin V/7-AAD assay results showed that cisplatin treatment for 48 hours induced early and late apoptosis in the 97L-pWPXL cells, and the apoptosis rate was upregulated in the 97L-HA-NRBP2 cell group (Supplementary Fig. S4A). A similar result could be obtained after sorafenib treatment (Supplementary Fig. S4B), indicating NRBP2high hepatocellular carcinoma cells are more sensitive to apoptosis after drug treatments.
CellMiner is a web application that facilitates systems biology through the retrieval and integration of molecular and pharmacologic data sets for NCI-60 cell lines (21). Through the CellMiner system, we found that the sensitivity of many platinum compounds, such as oxaliplatin and carboplatin, and of some Akt signaling inhibitors, including perifosine and PX316, was significantly correlated with NRBP2 mRNA expression in NCI-60 cell lines (Supplementary Fig. S5). Next, the relationship between NRBP2 protein expression and the IC50 of the drugs (cisplatin and perifosine) were investigated in 8 hepatocellular carcinoma cells lines. Pearson correlation analysis results showed that NRBP2 expression was negatively correlated with the IC50 of cisplatin (P = 0.041) and perifosine (P = 0.009) in hepatocellular carcinoma cells (Fig. 4C). MTT assay results showed that NRBP2 overexpression reduced the perifosine IC50 in MHCC-97L and PLC/PRF/5 cells, and NRBP2 knockdown increased the perifosine IC50 in SMMC-7721 cells (Fig. 4D). A similar correlation was also obtained with sensitivity to cisplatin (Fig. 4E), thus indicating that high NRBP2 expression inhibits the chemotherapy resistance of hepatocellular carcinoma cells. Because perifosine was approved by the FDA for the phase I/II trials in the treatment of non–small cell lung cancer (NSCLC), we also analyzed the relationship between NRBP2 expression and drug IC50 in 5 NSCLC cell lines (Supplementary Fig. S6A and S6B). Pearson correlation analysis results displayed that NRBP2 expression was also negatively correlated with the IC50 of perifosine in NSCLC cells (Supplementary Fig. S6C).
We analyzed the NRBP2 promoter activity in hepatocellular carcinoma cells after cisplatin and sorafenib treatment. Interestingly, FACS results showed that cisplatin and sorafenib treatment for 48 hours significantly promoted the NRBP2 transcriptional activity in MHCC-97L and SMMC-7721 cells (Supplementary Fig. S7A). Western blot analysis also confirmed that sorafenib treatment could upregulate NRBP2 protein expression (Supplementary Fig. S7B and S7C), which indicates that there might be a positive-feedback regulation loop between the cytotoxic effects of chemotherapy agents and NRBP2 expression in hepatocellular carcinoma.
NRBP2 protein can bind to and regulate ANXA2 expression
NRBP is an adapter protein (22). To investigate the mechanism that NRBP2 regulates chemoresistance, we analyzed NRBP2-interacting proteins via co-IP and MS analysis in hepatocellular carcinoma cells. The initial screening identified a quantity of protein candidates in the 7721-HA-NRBP2 and 97L-HA-NRBP2 cells (Supplementary Table S7), and the ectopic expression of several genes and co-IP assays were performed in the following screening to further examine whether NRBP2 interacted with them. We found that NRBP2 could bind to ANXA2.
As shown in Fig. 5A, HA-NRBP2-pWPXL and FLAG-ANXA2-pWPXL were cotransfected in MHCC-97L cells, and co-IP analysis with either mouse anti-HA-tag or rabbit anti-FLAG-tag antibody followed by Western blot analysis results showed that NRBP2 interacted with ANXA2. The overexpression of NRBP2 inhibited ANXA2 protein expression (Fig. 5B), and NRBP2 knockdown promoted ANXA2 protein expression in hepatocellular carcinoma cells (Fig. 5C and D). In the sorted NRBP2high hepatocellular carcinoma cells, ANXA2 expression was downregulated compared with the corresponding NRBP2low cells (Fig. 5E), indicating that NRBP2 inhibits ANXA2 expression in hepatocellular carcinoma cells.
ANXA2 knockdown inhibits hepatocellular carcinoma cell stemness
To verify the function of ANXA2 downregulation in the hepatocellular carcinoma cells, siRNA oligonucleotides specifically targeting ANXA2 were synthesized, and the knockdown efficiency was confirmed. Real-time PCR analysis showed that RNAi-induced silencing of ANXA2 inhibited stemness-related genes, including Oct3/4, Nanog, Notch1, Ep300, and CSC marker CD133 mRNA expression (Supplementary Figs. S8A and S7B). ANXA2 knockdown promoted the chemosensitivity of MHCC-97L and PLC/PRF/5 cells to cisplatin treatment (Supplementary Fig. S8C); and in sorted 7721-NRBP2low cell population, ANXA2 knockdown had also decreased its chemoresistance to perifosine. A similar result could be obtained in 97L-NRBP2low cell population (Supplementary Fig. S8D). As shown in Supplementary Fig. S8E, the tumorspheres obtained from ANXA2 knockdown PLC/PRF/5 cells were fewer in number and smaller in size than those control cells; and in 97L-NRBP2low cell population, ANXA2 knockdown had also inhibited its tumorsphere formation ability, indicating that ANXA2 knockdown impaired the stemness properties of hepatocellular carcinoma cells.
ANXA2 coexpression restores chemoresistant ability of hepatocellular carcinoma cells
We analyzed the chemotherapy sensitivity of NRBP2-overexpressing hepatocellular carcinoma cells after ANXA2 coexpression. As shown in Fig. 6A and B, ANXA2 coexpression increased the IC50 of cisplatin and perifosine in PLC-HA-NRBP2 and 97L-HA-NRBP2 cells and which showed no significant difference compared with control group. The proliferation ability of these cells was also analyzed by flat-plate colony formation assay. ANXA2 overexpression promoted cell colony formation in PLC-HA-NRBP2 and 97L-HA-NRBP2 cells (Fig. 6C and D). Tumorsphere formation assays showed that in MHCC-97L cells, ANXA2 coexpression partially restored the spheres' size and number compared with the 97L-HA-NRBP2 cells, suggesting that ANXA2 coexpression could promote the self-renewal ability of NRBP2high cells (Fig. 6E).
NRBP2 inhibits Akt phosphorylation and regulates Bcl2 family protein expression via ANXA2
Recent studies show that ANXA2 could interact with ANXA2 receptor and activates the Akt signaling (23). We then investigated the Akt phosphorylation (pAkt) level in the hepatocellular carcinoma cells. Western blot analysis results showed that ANXA2 knockdown inhibited pAkt in the MHCC-97L cells (Supplementary Fig. S9); and pAkt level in the NRBP2high cells was lower than that in the NRBP2low cells (Fig. 7A). Previous studies showed that the chemoresistance of cancer cells is closely linked to the antiapoptotic properties of Akt signaling (24). We found that the antiapoptosis protein Bcl2 was upregulated in NRBP2low cells, whereas the proapoptosis protein Bax was downregulated. Bax/Bcl2 ratio could reflect the propensity or resistance to apoptosis (25). Then, Bax/Bcl2 ratio was calculated and the results displayed that the Bax/Bcl2 ratio was decreased in the NRBP2low cells (Fig. 7A). Stable knockdown of NRBP2 showed a similar result that pAkt was upregulated and Bax/Bcl2 ratio was depressed in 7721-shNRBP2 and 97L-shNRBP2 cells (Fig. 7B).
In PLC/PRF/5 cells, coexpression of ANXA2 partially restored Akt phosphorylation and Bcl2 protein expression to the normal level compared with PLC-HA-NRBP2 cells. It could also be noted that the Bax/Bcl2 ratio decreased to a certain level after the ANXA2 coexpression. Similar results were obtained in MHCC-97L cells (Fig. 7C). LY294002 is a specific inhibitor of PI3K/Akt signaling pathway. Western blot analysis results demonstrated that LY294002 treatment inhibited Akt phosphorylation in the 97L-shNC and 97L-shNRBP2 cells, and downregulated Bcl2 and upregulated Bax protein expression. However, NRBP2 knockdown induced ANXA2 upregulation was not significantly affected. Similar results were obtained in SMMC-7721 cells (Supplementary Fig. S10A), suggesting that NRBP2 may regulate Akt phosphorylation via ANXA2.
To further ascertain the relationship between NRBP2 expression and chemoresistance, hepatocellular carcinoma cells were incubated with different doses of sorafenib. Sorafenib could induce cancer cell apoptosis through the mitochondrial pathway, and cleavage of PARP is one of the early markers of chemotherapy-induced apoptosis (26, 27). As shown in Fig. 7D, Western blotting showed that MHCC-97L cells expressed cleaved-PARP protein after 8 μmol/L sorafenib treatment for 48 hours and 97L-HA-NRBP2 cells expressed more cleaved PARP. Bad is a downstream target of Akt signaling and its proapoptosis effect is inactivated by Akt-mediated phosphorylation (28). We found that the phosphor-Bad protein (pBad) was downregulated in 97L-HA-NRBP2 cells, and 97L-HA-NRBP2 cells expressed even less pBad than 97L-pWPXL cells after sorafenib treatment. The Bax/Bcl2 ratio was upregulated in the sorafenib-treated 97L-HA-NRBP2 cells, but which was not significantly changed in the 97L-pWPXL cells. The proapoptosis proteins Bid was also upregulated in sorafenib-treated cells, although NRBP2-overexpressing cells expressed more Bid protein. Phosphor-Bcl2 (pBcl2) upregulated in 97L-HA-NRBP2 cells after sorafenib treatment, but no significant change of pBcl2 expression could be seen in 97L-pWPXL cells. Antiapoptosis Bcl-xL protein expression was upregulated in 97L-pWPXL cells after sorafenib treatment, however, in 97L-HA-NRBP2 cells, Bcl-xL was downregulated (Supplementary Fig. S10B). Similar results were also seen in sorafenib-treated PLC/PRF/5 cells (Supplementary Fig. S10C). These results demonstrated that high NRBP2 expression regulated proteins expression involved in the Bcl2 survival pathway, which may render hepatocellular carcinoma cells more sensitive to chemotherapeutic agents.
Perifosine treatment inhibits stemness property of hepatocellular carcinoma cells
Akt signaling inhibitor perifosine has been reported to inhibit tumorsphere formation and repairing of DNA damage selectively in CSCs in breast cancer (29, 30). In hepatocellular carcinoma cells, real-time PCR results displayed that perifosine treatment inhibited stemness-related genes, including Oct3/4, Nanog, Notch1, Ep300 and CSCs marker CD133 mRNA expression (Supplementary Fig. S11A). We found that perifosine could inhibit the tumorspheres growth in 97L-pWPXL and especially in the 97L-HA-NRBP2 cells (Supplementary Fig. S11B). Then, CD133+ and CD133− PLC/PRF/5 cells were sorted, and tumorspheres were formed after cultured under serum-free medium for 2 weeks. Different dose of perifosine were added to the culture medium and these tumorspheres were treated for another 48 hours. We found that perifosine treatment inhibited CD133+ CSC-formed tumorsphere growth in a dose-dependent manner. Although tumorspheres from the CD133− cell group were fewer in number and smaller in size, its growth was also inhibited after perifosine treatment (Supplementary Fig. S11C). These results indicate that perifosine might be a CSC inhibitor in hepatocellular carcinoma.
Discussion
NRBP2 shows sequence similarities to the NRBP1 gene, which regulates intestinal progenitor cell homeostasis and plays an important role in colon and breast tumor suppression (31, 32). Larsson and colleagues reported that NRBP2 protein expression is correlated to brain differentiation. In the adult mouse brain, NRBP2 expression is detected in CA3 pyramidal cells of the hippocampus, but not in CA1 and CA2 areas where adult neural stem cells reside. In addition, they found that in a pediatric medulloblastoma sample, most NRBP2-positive cells were also positive for the differentiated neuronal marker NF and did not colocalize with the CSC marker CD133 (13), indicating that NRBP2 may also play a role in tumorigenesis.
In this study, we found that NRBP2 was overexpressed in most noncancerous liver tissues analyzed. NRBP2 expression was negatively correlated with the histologic grade, and hepatocellular carcinoma patients with higher NRBP2 expression showed a better prognosis, suggesting NRBP2 plays an important role in the hepatocellular carcinoma progression. In hepatocellular carcinoma cell lines, CD133− hepatocellular carcinoma cells expressed a higher level of NRBP2 compared with corresponding CD133+ CSCs. CSCs are a rare population in cancer, which can self-renew and differentiate, and play a key role in the cancer progression and recurrence. According to the CSC model, tumorigenic CSCs and nontumorigenic cells are organized in a hierarchy in cancer and CSCs give rise to phenotypically diverse nontumorigenic cells. CSCs can be serially transplanted, re-establishing phenotypic heterogeneity with each passage (33, 34). FACS analysis results showed that NRBP2low cells could repopulate the heterogeneity in the hepatocellular carcinoma cell lines and xenograft tumors in the NOD/SCID mice were larger in the NRBP2low group than that in the NRBP2high group, suggesting that CSCs may be enriched in the NRBP2low cell population. Real-time PCR results displayed that stemness-related genes like Oct3/4, Nanog, Notch1, Ep300, and CD133 were upregulated in the NRBP2low and NRBP2 knockdown cells, and NRBP2 overexpression inhibited hepatocellular carcinoma cells self-renewal ability, demonstrated that high NRBP2 expression reduced the stemness of hepatocellular carcinoma cells.
Research data supported that tumor cell stemness is correlated with their chemoresistant ability. Several stemness-related genes could not only regulate the stemness maintenance, but also regulate tumor chemoresistance. Wang and colleagues reported that Oct4 overexpression enhances hepatocellular carcinoma cell resistance to chemotherapeutic drugs through a potential Oct4–Akt–ABCG2 pathway (35). Shan and colleagues found that Nanog regulates hepatocellular carcinoma CSCs self-renewal and Nanog-positive CSCs exhibit resistance to therapeutic agents (36). Park and colleagues showed that ectopic miR34a expression reduces breast cancer stemness and increases doxorubicin sensitivity by directly targeting Notch1 (37). Although the mechanism that NRBP2 inhibits stemness-related genes expression needs further study, it could be surmised that high NRBP2 expression may inhibit hepatocellular carcinoma chemoresistance.
Here, we demonstrated that NRBP2 overexpression could enhance the chemosensitivity of hepatocellular carcinoma cells to sorafenib and NRBP2 is negatively correlated with the IC50 of cisplatin and the Akt inhibitor perifosine. Constitutive activation of the Akt signaling pathway is critical for CSCs maintenance, and Akt signaling has been established as a major determinant of tumor cell chemoresistance (38, 39). In hepatocellular carcinoma, Dong and colleagues reported that Nogo-B receptor reduced p53 level through activation of PI3K/Akt/MDM2 pathway and increased the resistance of hepatocellular carcinoma cells to 5-FU (2). Hou and colleagues reported that tunicamycin suppressed several CSC markers expression and inactivated Akt signaling, and which potentiated the cytotoxicity of cisplatin (14). Ma and colleagues found that overactivation of Akt and Bcl2 survival pathway enabled CD133+ hepatocellular carcinoma CSCs to escape conventional chemotherapeutic agents (40). We found that Akt phosphorylation and its downstream target Bad phosphorylation levels were downregulated in the hepatocellular carcinoma cells with high NRBP2 expression. Bax/Bcl2 ratio was upregulated in NRBP2-overexpressing cells, indicating that these cells are more sensitive to apoptotic stimuli. After sorafenib treatment, the antiapoptosis pBad was further downregulated, but the proapoptosis Bid was further upregulated in the NRBP2-overexpressing cells, as well as the Bax/Bcl2 ratio, indicating that NRBP2 inhibited Akt and Bcl2 survival pathway in hepatocellular carcinoma cells. Considering that the early and late apoptosis rates were increased in NRBP2 overexpression hepatocellular carcinoma cells after cisplatin and sorafenib treatment, we thought that NRBP2 enhances the hepatocellular carcinoma chemosensitivity mainly by rendering hepatocellular carcinoma cells more sensitive to chemotherapeutic agent–induced apoptosis.
Interestingly, we found that cisplatin and sorafenib treatment could upregulate NRBP2 expression in hepatocellular carcinoma cells. The mechanism regulates NRBP2 expression is still unclear. Zhao and colleagues showed that titanium dioxide nanoparticles cause severe testicular oxidative damage or apoptosis and upregulated NRBP2 in the mouse testis (41). We thought that chemotherapeutic agents like cisplatin and sorafenib may promote NRBP2 expression by suppression of certain signaling pathway or by inducing apoptosis. In this study, NRBP2 in turn enhances the cytotoxic effects of these agents via regulating the Akt and Bcl2 survival pathway, which indicates a positive feedback regulation of chemotherapeutic agents on NRBP2 expression in hepatocellular carcinoma cells. On the other hand, sorafenib and cisplatin have also been reported to could target the CSCs in several types of cancers. Carra and colleagues reported that sorafenib selectively depletes human glioblastoma CSCs from primary cultures and downregulates the expression of stemness markers (42). Yu and colleagues showed that cisplatin suppresses metastasis and invasion of ovarian CSCs by targets the SDF1–CXCR4 axis (43), and we found that BMP4 treatment, which could induce CD133+ hepatocellular carcinoma CSC differentiation, promoted NRBP2 promoter activity in hepatocellular carcinoma cells (data not shown). We speculate that BMP4 treatment could induce hepatocellular carcinoma cell determination and differentiation, which then promotes NRBP2 expression directly or indirectly. Upregulation of NRBP2 expression may contribute to the BMP4-induced hepatocellular carcinoma CSC differentiation.
ANXA2 is a calcium-dependent phospholipid-binding protein and regulates a wide range of molecular and cellular processes (44). ANXA2 can activate the Akt pathway via its binding with the ANXA2 receptor (ANXA2R; ref. 23). In hepatocellular carcinoma, stable knockdown of ANXA2 inhibits hepatocellular carcinoma proliferation and enhances chemotherapeutic treatment (45–47). We found that NRBP2 may bind to ANXA2 in hepatocellular carcinoma cells. NRBP2 overexpression inhibited ANXA2 expression. Knockdown of ANXA2 had also inhibited the stemness-related genes expression, decreased the self-renewal and chemoresistance ability of hepatocellular carcinoma cells including in the NRBP2low hepatocellular carcinoma cells. NRBP2-induced ANXA2 downregulation contributed to the regulatory effects of NRBP2 on the Akt and Bcl2 signaling, the sensitivity of hepatocellular carcinoma cells to chemotherapy, and hepatocellular carcinoma cells' self-renewal ability. Previous reports demonstrated that NRBP could interact with Jab1 and Jab1 is a component of the COP9 signalosome complex, which play an important role in the ubiquitin proteasome system (22, 48), indicating that NRBP maybe also involved in the regulation of protein degradation. However, the mechanism that NRBP2 inhibits ANXA2 expression requires further investigation.
In summary, NRBP2 is involved in the regulation of stemness-related gene expression and self-renewal ability in hepatocellular carcinoma cells. High NRBP2 expression inhibits ANXA2 expression and then renders hepatocellular carcinoma cells sensitive to chemotherapy agents via the Akt and Bcl2 survival signaling pathway. Interestingly, we found that NRBP2 expression is negatively correlated with the chemoresistance of perifosine in hepatocellular carcinoma and NSCLC cell lines. Perifosine treatment inhibits the stemness property of hepatocellular carcinoma cells. These findings suggest that a combination of NRBP2 expression analysis and perifosine treatment may be a promising approach to overcome the chemoresistance of hepatocellular carcinoma.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: L. Zhang, M. Yao, J. Li
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Zhang, C. Ge, F. Zhao, Y. Zhang, X. Wang, M. Yao
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Zhang
Writing, review, and/or revision of the manuscript: L. Zhang, J. Li
Study supervision: J. Li
Grant Support
This work was supported in part by grants from the National Key Program for Basic Research of China (973; no. 2015CB553905 to J. Li), National Natural Science Foundation of China (no. 81301859 to L. Zhang; no. 81272438 and no. 81472726 to J. Li), Key Discipline and Specialty Foundation of Shanghai Municipal Commission of Health and Family Planning, the National Key Sci-Tech Special Project of China (no. 2013ZX10002-011 to J. Li), Innovation Program of Shanghai Municipal Education Commission (no. 13ZZ082 to J. Li), the Specialized Research Fund for the Doctoral Program of Higher Education of China (no. 20130073120012 to L. Zhang), and the SKLORG Research Foundation (no. 91-14-09, no. 91-15-03 to J. Li).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.