Loss of the tumor suppressor NDRG2 has been implicated in the development of oral squamous cell carcinoma (OSCC), acting by modulating PI3K/AKT-mediated dephosphorylation of PTEN at S380/S382/T383 (STT). Here, we show that the majority of OSCC tumors with lymph node metastasis, a major prognostic factor, exhibit high levels of phosphorylated AKT-S473 and PTEN-STT and low levels of NDRG2 expression. In Ndrg2-deficient mice, which develop a wide range of tumors, we developed a model of OSCC by treatment with the tobacco surrogate 4-nitroquinoline-1-oxide (4-NQO). In this model, both the number and size of OSCC tumors were increased significantly by Ndrg2 deficiency, which also increased invasion of cervical lymph nodes. 4-NQO treatment of human OSCC cell lines exhibiting low NDRG2 expression induced epithelial–mesenchymal transition via activation of NF-κB signaling. Conversely, ectopic expression of NDRG2 reversed the EMT phenotype and inhibited NF-κB signaling via suppression of PTEN-STT and AKT-S473 phosphorylation. Our results show how NDRG2 expression serves as a critical determinant of the invasive and metastatic capacity of OSCC. Cancer Res; 77(9); 2363–74. ©2017 AACR.

Oral squamous cell carcinoma (OSCC) is the most common epithelial malignancy in the oral cavity and constitutes more than 90% of all oral malignancies. The number of OSCC patients has gradually increased as the number of elderly individuals in the Japanese population has increased. Epidemiological studies have shown that heavy tobacco smoking, high alcohol consumption and human papilloma virus infection are three major risk factors for OSCC development. Moreover, several molecular genetic changes have been reported to be associated with the development of OSCC, such as the activation of the Notch, cell-cycle and mitogenic signaling pathways and the inactivation of TP53 (1). Within mitogenic signaling, high frequent activator mutations in the PI3K/AKT signaling pathway were found in EGFR, HRAS, PIK3CA, or AKT1; however, either no mutations or a very low mutation or deletion rate in PTEN as a primary negative regulator have been reported in OSCC tumors.

Although the majority of OSCCs are reported to feature constitutive activation of AKT, the main reasons for this were not known until recent findings of next-generation sequence analysis became available (1). However, we have previously reported that N-myc downstream regulated gene-2 (NDRG2) is one of the important negative regulators of the PI3K/AKT signaling pathway and that the loss of NDRG2 expression is found in most cases of OSCC (2). Interestingly, positive phosphorylated-AKT staining was inversely correlated with negative NDRG2 staining in OSCC tumors with moderate to poor differentiation (2), which suggests that the loss of NDRG2 expression might be important for the development of OSCC with poor prognosis. Moreover, we recently found that NDRG2 is a novel PTEN-binding protein that recruits protein phosphatase 2A to dephosphorylated PTEN at its C-terminus (S380/S382/T383); after binding, the lipid-phosphatase activity of PTEN is maintained (3). Therefore, we suspected that the loss of NDRG2 expression might be involved in the tumorigenesis of OSCC through constitutive activation of the AKT pathway. Therefore, we next investigated whether the loss of NDRG2 might induce OSCC in a mouse model of Ndrg2 deficiency.

To determine the function of NDRG2 in the development of OSCC, the Ndrg2-deficient mice were treated with the well-known chemical carcinogen 4-Nitroquinoline 1-oxide (4-NQO) by adding it to their drinking water. This chemical was used in this study as a surrogate for tobacco exposure to induce the development of oral cancer (4, 5). The oral tumors that developed in the Ndrg2-deficient mice were larger and greater in number than those that developed in the wild-type control mice and were also a more pathologically invasive type of carcinoma. Moreover, cervical lymph node metastasis was frequently observed in Ndrg2-deficient mice, which suggests that the loss of Ndrg2 expression might be important for the development of metastasis. Therefore, using a murine model and human OSCC samples, we revealed and discussed the molecular mechanism of OSCC development and metastasis as it relates to the loss of NDRG2 expression.

Patient samples

Primary OSCC tumor samples that were subjected to IHC were obtained from patients during surgery. All patient samples were embedded in paraffin for IHC analysis and were used only after permission was received from the ethics committee of the Faculty of Medicine, University of Miyazaki.

Cell lines

Eight human OSCC cell lines (Ca922, HO-1-u-1, HSC2, HSC3, HSC4, SAS, HSQ89, and Sa-3) were bought from the RIKEN BioResource Center (Tsukuba, Japan; 2009). The HaCaT cell line, which is derived from human keratinocytes, was obtained from the Cell Line Service (2009) and the hOMK100 cell line, which is derived from human oral mucosa, was purchased from Cosmo Bio (2016); both were used as controls. After receiving, the cells were expanded at early passages and stocked in liquid nitrogen. The newly stocked vials were confirmed to be no contamination with mycoplasma by a PCR detection method (6), and were used in each experiment within passages for up to 2 months. The stocked cell lines were not further authenticated in our laboratory. hOMK100 cells were maintained in EpiLife medium (Thermo Fisher Scientific) and used at fewer than three passages, while the other cell lines were cultured in the appropriate media (RPMI-1640 or Dulbecco's Modified Eagle's Medium) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 μg/mL streptomycin. To establish the stable knockdown of NDRG2 expression in HaCaT, we used short hairpin RNA (shRNA) expression vector (RNAi-Ready-pSIREN-RetroQ-ZnGreen vector, Clontech) against NDRG2 as described previously (3).

Antibodies

A monoclonal anti–β-actin antibody produced in mouse (AC-15) was purchased from Sigma-Aldrich. The following antibodies were obtained from Cell Signaling Technology: rabbit monoclonal antibodies against human PTEN (138G6), phospho-AKT (Ser473; D9E), E-cadherin (24E10), vimentin (D21H3), S6 Ribosomal protein (5G10), phosphor-S6 Ribosomal protein (Ser240/244; D68F81), phospho-IKKα/β (Ser176/180; 16A6), phospho-GSK3β (Ser 9; D85E12) and GSK3β (27C10); rabbit polyclonal antibodies against human phospho-PTEN (Ser380/Thr382/383; 9554) and AKT (9272); a mouse monoclonal antibody against human phospho-IκBα (5A5). A goat polyclonal antibody against NDRG2 (E20), a mouse monoclonal antibody against Histone H1 (AE-4), and a rabbit polyclonal antibody against IκBα (C-21) were obtained from Santa Cruz Biotechnology. An anti-mouse monoclonal antibody to pan-keratin (ab8068) was purchased from Abcam.

Establishment of a mouse model of OSCC induced by 4-NQO treatment

Ndrg2-deficient mice were described elsewhere (3). In regard to the treatment with 4-nitroquinoline 1-oxide (4-NQO), drinking water containing 4-NQO (50 μg/mL) was available to the mice for 16 weeks or 30 weeks. 4-NQO is a chemical carcinogen that is known to selectively induce oral carcinogenesis (4, 5). After the period of carcinogen treatment, the mice were given pure drinking water for 2- to 4-week intervals before they were sacrificed.

RT-PCR

Total RNA was extracted from cells using TRIzol reagent (Invitrogen, Thermo Fisher Scientific) and 1 μg of total RNA was reverse transcribed to obtain first-strand cDNA using an RNA-PCR kit (Takara-Bio Inc.). The primers used were listed in Supplementary Materials and Methods.

Quantitative real-time PCR analysis

Quantitative real-time RT-PCR was performed with GeneAce SYBR qPCR Mix α (Nippon Gene) using a StepOne Real-time PCR System (Applied Biosystems). The amplification data were analyzed with StepOne software (Applied Biosystems), converted into cycle numbers at a set cycle threshold (Ct values) and quantified relative to a standard. The RNA extracted from OSCC cell lines was used as a standard in all experiments. To normalize the amounts of input cDNA, the relative amount of the generated product was separated by the relative amount of β-actin. All samples were analyzed in duplicate. The same primer pairs used for semiquantitative PCR analysis were used for real-time PCR.

Western blot analysis

Cells were fixed with either NP-40 lysis buffer (50 mmol/L Tris–HCl, pH 8.0, 150 mmol/L NaCl, 5 mmol/L EDTA, 1% NP-40) supplemented with protease inhibitor cocktail (Sigma-Aldrich) and phosphatase inhibitors (PhosStop, Roche) or Laemmli SDS sample buffer (62.5 mmol/L Tris–HCl, pH 6.8, 2% SDS, 25% glycerol, 5% β-mercaptoethanol, and 0.01% bromophenol blue). The protein samples were separated by a 10% SDS-polyacrylamide gel and transferred to polyvinylidine difluoride membranes (Immobilon-P, Merck Millipore). After blocking with Tris-buffered saline (TBS)-Tween (0.1%) with either 5% bovine serum albumin (BSA) or 5% nonfat dried milk, membranes were incubated with each primary antibody described above, diluted with TBS containing 0.1% Tween 20 supplemented with either 5% nonfat dried milk or 5% BSA or in Can Get Signal buffer (Toyobo). Immunoreactive bands were visualized by a Lumi-light Plus kit (Roche Diagnostics). Band intensities on blots were quantified using NIH Image J software. All primary antibodies were used at a dilution of 1:1,000. For subcellular fractionation, cytoplasmic and nuclear protein extracts were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagent (Thermo Fisher Scientific). The efficiency of fractionation was confirmed by Western blot analysis for β-actin (cytoplasm) and histone H1 (nucleus).

Immunohistochemical and immunofluorescence staining

For the immunohistochemical staining of the pathological sections, the sections were treated with 0.1% hydrogen peroxide for 30 minutes to block endogenous peroxide activity. The sections were then blocked in 1% normal goat serum for 1 hour and incubated overnight with the primary antibodies described above overnight at 4°C. The slides were washed with TBST and then incubated with horseradish peroxidase–conjugated secondary antibodies for 1 hour. Peroxidase activity was detected with 3,3′-diaminobenzidine (Thermo Fisher Scientific) according to the manufacturer's protocol. The antibodies were diluted as follows: pan-keratin (1:10), p-PTEN (1:200), p-AKT (1:50), and NDRG2 (1:200). For immunofluorescence staining, the cells were fixed in 4% paraformaldehyde for 10 minutes at room temperature, washed with 0.1 mol/L glycine in TBS, and then permeabilized with 0.2% Triton X-100 for 10 minutes. After being washed with 0.1 mol/L glycine in TBS, the cells were blocked in 1% BSA in TBS for 1 hour, incubated with primary antibodies overnight at 4°C, washed three times with TBS containing 0.1% Tween 20, followed by incubation with Alexa Fluor-555, Alexa Fluor-647 or Alexa Fluor-488 anti-rabbit secondary antibodies (Molecular Probes, Thermo Fisher Scientific) at room temperature for 2 hours. The coverslips were washed three times in TBS containing 0.1% Tween 20, mounted on glass slides using an antifade reagent (Invitrogen, Thermo Fisher Scientific). The nuclei were counterstained with 4', 6-Diamidino-2-phenylindole (DAPI; Sigma-Aldrich), and the proteins were visualized using a confocal laser scanning microscope (Leica Microsystems).

Wound-healing assay

Cells were seeded in 12-well plates at a density of 50 × 103 cells/well in complete RPMI-1640 medium. The cells were cultured to confluence and serum-starved overnight in RPMI-1640 without FBS. A 1-mm wide scratch was made across the cell layer with a sterile 1,000-μL pipette tip. After being washed with phosphate-buffered saline, the cells were incubated with RPMI-1640 supplemented with 5% FBS with or without 4-NQO. The scratch wounds were photographed at 12 hours using an OLYMPUS C-5060 light microscope (Olympus) with a ×10 objective. All experimental conditions were tested in duplicate. The percentage of wound healing was evaluated. The moving distance was measured by ImageJ software (National Institutes of Health).

Invasion assay

A cell invasion assay was performed in modified Boyden chambers containing filter inserts with 8-μm pores (Kurabo) coated with Matrigel (BD Falcon). Cells in 300 μL of serum-free RPMI-1640 with or without 4-NQO were added to the upper chambers at a density of 2 × 105 cells per well and 500 μL of RPMI-1640 with 10% FBS was added to the lower chambers. After 24 hours of incubation at 37°C, cells in the upper chambers were removed with a cotton swab, and the filter was fixed and stained with 0.5% (w/v) crystal violet solution. Four randomly selected fields in each membrane were photographed under an OLYMPUS C-5060 light microscope (OLYMPUS) with a ×10 objective.

Luciferase reporter assay

Cells were transfected with pNF-κB-Luc (Agilent Technologies) and pRL-TK (Promega) vectors using HilyMax (Dojindo). The NF-κB transcriptional activity was measured with a luminometer (Turner Designs TD-20/20, Promega) using a dual luciferase assay kit (Promega) according to the manufacturer's instructions.

Statistical analysis

The in vivo statistical analysis of tumor incidence was performed using STATVIEW 5.0 software. The statistical significance was analyzed using a two-tailed Student t test, χ2 test and Bonferroni–Dunn multiple comparison tests. A value of P < 0.05 was considered statistically significant.

Low expression of NDRG2 in OSCC enhanced the phosphorylation of AKT through PTEN phosphorylation at the C-terminus (STT)

We previously reported that expression of NDRG2 was downregulated in most OSCC tumors due to promoter methylation, which resulted in enhanced PI3K/AKT signaling in OSCC, even though PTEN expression was maintained and no somatic mutations were found (2). Recently, we found that NDRG2 is a novel PTEN-binding protein that functions as a regulator of PTEN phosphatase activity. The loss of NDRG2 expression in various cancers enhanced the activation of the PI3K/AKT signaling pathway through enhanced phosphorylation of S380T382T383 (STT) in the C-terminus region of PTEN (3). Therefore, we initially determined the expression levels of NDRG2, PTEN, and AKT along with the phosphorylation status of AKT (S473) and PTEN (STT) in OSCC cell lines and primary OSCC tumors by Western blot analysis and immunohistochemistry. After testing two control cell lines (hoMK100 and HaCaT cells) and eight OSCC cell lines (Ca922, Ho1u1, HSC2, HSC3, HSC4, HSQ89, SAS, and Sa3), sustained expression of PTEN with high phosphorylation at STT as well as high phosphorylation of AKT (S473) was detected in all of the OSCC cell lines by Western blot analysis (Fig. 1A). Positive staining for phosphorylated AKT (S473) and PTEN (STT) as well as low or negative staining for NDRG2 were observed in serial sections of a typical OSCC tumor derived from a patient with OSCC (Fig. 1B). After 42 tumors from patients with OSCC were examined by immunohistochemistry, high negative rates of NDRG2 staining (81%) with high positive rates of p-AKT (90.5%) and p-PTEN (76.2%) were detected in the primary OSCC tumors (Table 1). Because the negative staining of NDRG2 was significantly associated with the positive staining of p-AKT and p-PTEN with statistical differences (P < 0.001; Supplementary Table S1), the loss of NDRG2 expression might enhance the expression of p-AKT and p-PTEN. In addition, a group of patients with OSCC whose tumors were negative for NDRG2 staining was significantly correlated with a group of patients with OSCC with lymph node metastasis (Table 1). This suggests that the loss of NDRG2 expression might be involved in the enhanced metastatic potential of OSCC.

Figure 1.

Identification of the expression of NDRG2, p-AKT (S473), and p-PTEN (STT indicated as Ser380/Thr382/Thr383) proteins in OSCC tissues. A, The protein expression of p-PTEN (STT), total PTEN, p-AKT (S473), total AKT, NDRG2, and β-actin (as a control) was determined in eight OSCC cell lines and two control cell lines (hoMK100 and HaCaT) using the specific antibodies described in Materials and Methods. B, An immunohistochemical analysis of NDRG2, p-AKT (S473), and p-PTEN (STT) was performed in 32 human OSCC specimens using the specific antibodies described in Materials and Methods (Table 1). H&E, hematoxylin and eosin staining.

Figure 1.

Identification of the expression of NDRG2, p-AKT (S473), and p-PTEN (STT indicated as Ser380/Thr382/Thr383) proteins in OSCC tissues. A, The protein expression of p-PTEN (STT), total PTEN, p-AKT (S473), total AKT, NDRG2, and β-actin (as a control) was determined in eight OSCC cell lines and two control cell lines (hoMK100 and HaCaT) using the specific antibodies described in Materials and Methods. B, An immunohistochemical analysis of NDRG2, p-AKT (S473), and p-PTEN (STT) was performed in 32 human OSCC specimens using the specific antibodies described in Materials and Methods (Table 1). H&E, hematoxylin and eosin staining.

Close modal
Table 1.

Expression of NDRG2, p-Akt, and p-PTEN in OSCC patients

NDRG2p-Aktp-PTEN
ParameterNNegative (%)Positive (%)PNegative (%)Positive (%)PNegative (%)Positive (%)P
Total 42 34 (81) 8 (19) P < 0.001 4 (9.5) 38 (90.5) P < 0.001 10 (23.8) 32 (76.2) P < 0.001 
Gender    0.65   0.448   0.209 
 Male 24 20 (83.3) 4 (16.7)  3 (12.5) 21 (87.5)  4 (16.7) 20 (83.3)  
 Female 18 14 (77.8) 4 (22.2)  1 (5.6) 17 (94.4)  6 (33.3) 12 (66.7)  
TNM stage    0.013   0.57   0.177 
 I–II 26 18 (69.2) 8 (30.8)  3 (11.5) 23 (88.5)  8 (30.8) 18 (69.2)  
 III–IV 16 16 (100) 0 (0)  1 (6.3) 15 (93.7)  2 (12.5) 14 (87.5)  
T stage    0.035   0.786   0.101 
 T1/T2 29 21 (72.4) 8 (27.6)  3 (10.3) 26 (89.7)  9 (31) 20 (69)  
 T3/T4 13 13 (100) 0 (0)  1 (7.7) 12 (92.3)  1 (7.7) 12 (92.3)  
Lymph nodes    0.321   0.812   0.439 
 N0 17 15 (88.2) 2 (11.8)  1 (5.9) 16 (94.1)  3 (17.6) 14 (82.4)  
 N+ 25 19 (76) 6 (24) P < 0.001 3 (12) 22 (88)  7 (28) 18 (72)  
NDRG2p-Aktp-PTEN
ParameterNNegative (%)Positive (%)PNegative (%)Positive (%)PNegative (%)Positive (%)P
Total 42 34 (81) 8 (19) P < 0.001 4 (9.5) 38 (90.5) P < 0.001 10 (23.8) 32 (76.2) P < 0.001 
Gender    0.65   0.448   0.209 
 Male 24 20 (83.3) 4 (16.7)  3 (12.5) 21 (87.5)  4 (16.7) 20 (83.3)  
 Female 18 14 (77.8) 4 (22.2)  1 (5.6) 17 (94.4)  6 (33.3) 12 (66.7)  
TNM stage    0.013   0.57   0.177 
 I–II 26 18 (69.2) 8 (30.8)  3 (11.5) 23 (88.5)  8 (30.8) 18 (69.2)  
 III–IV 16 16 (100) 0 (0)  1 (6.3) 15 (93.7)  2 (12.5) 14 (87.5)  
T stage    0.035   0.786   0.101 
 T1/T2 29 21 (72.4) 8 (27.6)  3 (10.3) 26 (89.7)  9 (31) 20 (69)  
 T3/T4 13 13 (100) 0 (0)  1 (7.7) 12 (92.3)  1 (7.7) 12 (92.3)  
Lymph nodes    0.321   0.812   0.439 
 N0 17 15 (88.2) 2 (11.8)  1 (5.9) 16 (94.1)  3 (17.6) 14 (82.4)  
 N+ 25 19 (76) 6 (24) P < 0.001 3 (12) 22 (88)  7 (28) 18 (72)  

Loss of Ndrg2 expression was involved in the development of SCC in Ndrg2-deficient mice after treatment with 4-NQO

Because the majority of OSCC tumors lose expression of NDRG2 but feature activation of the PI3K/AKT pathway, we next investigated whether the loss of NDRG2 expression might be involved in the development of OSCC tumorigenesis in vivo using Ndrg2-deficient mice. Because 4-NQO is one of many well-known chemical carcinogens that serves as a surrogate for tobacco exposure (4, 5). Sixteen wild-type mice and Ndrg2-deficient mice (21 Ndrg2+/− and 10 Ndrg2−/− mice) were treated with 50 μg/mL 4-NQO in their drinking water for 16 weeks, after which the mice were given normal water for four weeks (Fig. 2A). Then, 20 weeks after treatment with 4-NQO, oral tumors developed in approximately 50% of wild-type mice. The incidence of tumor formation in both the Ndrg2+/− and Ndrg2−/− mice was significantly increased and the development of tumors in Ndrg2-deficient mice was faster than that in wild-type mice (Fig. 2B and 2C). The number of tumors on the tongue and the tumor volumes in Ndrg2-deficient mice were also higher than those in wild-type mice; these differences were statistically significant (Fig. 2D). Based on the pathological findings, the incidence of all tumors, including papilloma and squamous cell carcinoma (SCC), in Ndrg2-deficient mice was higher than that of wild-type mice (Supplementary Table S2). In addition, the tumor incidence and multiplicity of invasive type-SCC in Ndrg2-deficient mice were significantly higher than those in wild-type mice (Supplementary Table S3). Furthermore, high expression of p-Akt and p-Pten was observed in the tumors that formed in Ndrg2-deficient mice as confirmed by low expression of Ndrg2 by immunohistochemistry using the specific antibodies described earlier (Fig. 2E). We also determined the lymph node metastasis rates in the mice that were treated with 4-NQO; however, metastasis to the submandibular lymph nodes, as indicated by immunohistochemical staining for keratin, was not observed in this experiment (Supplementary Fig. S1). Thus, these data suggest that the loss of NDRG2 expression might be involved in the development of OSCC and may particularly enhance the invasive capacity of OSCC cells.

Figure 2.

Induction of OSCC tumors in Ndrg2-deficient mice by treatment with 4-NQO in the drinking water. A, This scheme represents the protocol for the treatment of the mice with 4-NQO for 16 weeks with an observation period of an additional 4 weeks. B, Images of the murine oral cavity show tumor formation on the tongue of each mouse after the administration of 4-NQO to wild-type (WT) and Ndrg2-deficient mice (Ndrg2+/− and −/−). C, Time course of tumor incidence is presented after 4-NQO treatment in each mouse group. *, P < 0.01; **, P < 0.005. D, Comparisons of the number of tumors that developed (left) and the sizes of those tumors (right) between wild-type and Ndrg2-deficient mice are shown along with the statistical analysis. *, P < 0.01; **, P < 0.005. E, Tumors that developed on the tongue of each indicated mouse were immunostained using antibodies specific to Ndrg2, p-Akt (S473), and p-Pten (STT) and were also stained with hematoxylin and eosin (H&E).

Figure 2.

Induction of OSCC tumors in Ndrg2-deficient mice by treatment with 4-NQO in the drinking water. A, This scheme represents the protocol for the treatment of the mice with 4-NQO for 16 weeks with an observation period of an additional 4 weeks. B, Images of the murine oral cavity show tumor formation on the tongue of each mouse after the administration of 4-NQO to wild-type (WT) and Ndrg2-deficient mice (Ndrg2+/− and −/−). C, Time course of tumor incidence is presented after 4-NQO treatment in each mouse group. *, P < 0.01; **, P < 0.005. D, Comparisons of the number of tumors that developed (left) and the sizes of those tumors (right) between wild-type and Ndrg2-deficient mice are shown along with the statistical analysis. *, P < 0.01; **, P < 0.005. E, Tumors that developed on the tongue of each indicated mouse were immunostained using antibodies specific to Ndrg2, p-Akt (S473), and p-Pten (STT) and were also stained with hematoxylin and eosin (H&E).

Close modal

A high incidence of cervical lymph node metastasis was observed in Ndrg2-deficient mice treated with 4-NQO for an extended 30 weeks

Cervical lymph node metastases are very common in patients with OSCC, and the lymphatic spread is associated with increased risk of locoregional recurrence. This significantly impairs the quality-of-life (QOL) of the patients due to lymph node dissection and can also alter the prognosis of the patients. Therefore, to determine whether the loss of Ndrg2 expression might be involved in cervical lymph node metastases, the duration of 4-NQO treatment of the Ndrg2-deficient mice was extended for 30 weeks, which is a timeline based on previous reports (Fig. 3A; ref. 7). The numbers of total tumors and SCC tumors in Ndrg2−/− mice were significantly higher than those in wild-type mice (Supplementary Table S4), and moreover, the number of invasive tumors in Ndrg2−/− mice was also higher than that in wild-type mice; these differences were statistically significant (Supplementary Table S5). Because the tumor burden of the Ndrg2−/− mice was not sufficient for the tumors to invade the oral cavity, including the lip, gingiva, and oral palate (Fig. 3B), we precisely assessed the tumor invasion of the cervical lymph nodes (Fig. 3C). In terms of cancer invasion to the cervical lymph nodes, 7 out of 12 Ndrg2−/− mice (58%) demonstrated metastasis to the cervical lymph nodes, although cervical lymph node metastasis was found in only 3 out of 9 wild-type mice (33%). Moreover, the most of lymph nodes (27 out of 34; 79%) in total 7 Ndrg2−/− mice contained tumor metastasis, although more than half of the lymph nodes showed no metastasis in wild-type mice (Fig. 3D), In Ndrg2−/− mice, the average number of metastasis lymph nodes per mouse was significantly increased compared with normal lymph nodes (Supplementary Table S6). Therefore, the potential metastatic ability to the cervical lymph nodes was possibly enhanced by the loss of Ndrg2 expression in OSCC.

Figure 3.

Induction of OSCC tumors with metastasis in Ndrg2-deficient mice by treatment with 4-NQO for longer periods of time until 30 weeks. A, This scheme represents a protocol for the treatment of mice with 4-NQO for 30 weeks with an observation period of an additional 2 weeks. B, Images show the oral tumors on the faces of the mice in each group. Arrows indicate tumors on the tongue and the arrowhead near the Ndrg2−/− mice indicates swelling around the mouth due to tumor cell infiltration. C, This image shows swelling of the cervical lymph node (white arrow). D, The swollen lymph nodes were immunohistochemically stained with an antibody against keratin, which illustrates metastasis in two sections.

Figure 3.

Induction of OSCC tumors with metastasis in Ndrg2-deficient mice by treatment with 4-NQO for longer periods of time until 30 weeks. A, This scheme represents a protocol for the treatment of mice with 4-NQO for 30 weeks with an observation period of an additional 2 weeks. B, Images show the oral tumors on the faces of the mice in each group. Arrows indicate tumors on the tongue and the arrowhead near the Ndrg2−/− mice indicates swelling around the mouth due to tumor cell infiltration. C, This image shows swelling of the cervical lymph node (white arrow). D, The swollen lymph nodes were immunohistochemically stained with an antibody against keratin, which illustrates metastasis in two sections.

Close modal

Enhanced activation of the PI3K/AKT signaling pathway by 4-NQO treatment in the development of SCC

Tobacco exposure is one of most important etiological factors in oral cancer tumorigenesis, and the tobacco carcinogens 4-NQO or NNK activate several signal transduction pathways, including AKT, during the early premalignancy stage of oral cancer development (8). Because mTOR inhibition by rapamycin has been shown to prevent the early onset of 4-NQO-induced SCC in a mouse model (9), we next investigated whether treatment with 4-NQO affects tumorigenesis and cell invasion capacity in OSCC cells through the activation of the PI3K/mTOR signaling pathways. After the treatment of the SAS/OSCC cell line with 1 μmol/L of 4-NQO, various types of signaling molecules involved in the PI3K/AKT pathway were determined at each indicated time point by the specific antibodies discussed earlier (Supplementary Fig. S2A). Five to 10 minutes after 4-NQO treatment, the levels of phosphorylated PTEN (STT), AKT (S473), GSK3β (S9), and S6 (S240/244) were gradually increased until 6 hours after treatment. Moreover, to investigate whether 4-NQO treatment affects the cell migration capacity, we measured the distance between the scratch edges, 12 hours after a scratch was generated on the glass slide containing the SAS/OSCC cells that were treated with or without 4-NQO. The 4-NQO treatment enhanced the cell migration capacity in a dose-dependent manner, and SAS/OSCC cells treated with the highest dose of 4-NQO migrated to fill the gap of the scratch (Supplementary Fig. S2B). Therefore, activation of the AKT signaling pathway and the cell migration capacity were enhanced after the treatment of OSCC cells with 4-NQO.

Because NDRG2 suppresses the AKT signaling pathway through the activation of PTEN phosphatase activity by dephosphorylation, we next determined whether forced NDRG2 expression suppresses the enhanced activation of the AKT signaling pathway induced by the treatment of OSCC cells with 4-NQO. Two cell lines, SAS/OSCC and HeLa/cervical cancer cell lines, presented low NDRG2 expression with sustained wild-type PTEN expression and were stably transfected with NDRG2 expression vector (2, 3). These cell lines were analyzed for the expression levels and the phosphorylation status of various AKT signaling molecules at each time point after treatment with 1 μmol/L 4-NQO (Fig. 4A). In mock control cells, the phosphorylation level of PTEN at STT was enhanced after 5 minutes of treatment with 4-NQO, while the phosphorylation of AKT, GSK3β and S6 was enhanced after 5 to 10 minutes of treatment. On the contrary, neither the phosphorylation status of PTEN, AKT, GSK3β and S6 was different before and after treatment with 4-NQO in the cell lines expressing NDRG2.

Figure 4.

Treatment with 4-NQO and the loss of NDRG2 expression acted together to enhance the migration and invasion capacity of OSCC cells. A, Using SAS/OSCC or HeLa cells transfected with the mock control or the NDRG2-expressing plasmid, the effects of 4-NQO treatment on the PI3K/AKT signaling pathway were determined at each time point by Western blot analysis using antibodies specific to NDRG2, Flag, phosphorylated PTEN (STT) with total PTEN, phosphorylated AKT (S473) with total AKT, phosphorylated GSK3β (S9) with total GSK3β, phosphorylated S6 (S240/244) with total S6, and β-actin as a control. B, Cell migration ability under four different conditions with/without expression of NDRG2 and with/without 4-NQO treatment in SAS/OSCC cells was determined by in vitro scratch assays. Images on the left show SAS cells under four different conditions on the glass slide; the black lines indicate the initial edge of the cells, while the red arrows indicate the migrated cells after the scratch was made. The bar graph (right) illustrates the relative migration rates of SAS cells under the four different conditions indicated under the bar. C, The images on the left show migrated SAS/OSCC cells under the four different conditions indicated in the figure; these cells were stained with crystal violet in a cell invasion assay using Boyden chambers (magnification, ×100). The bar graph on right side shows the number of migrated cells under the four different conditions. *, P < 0.05. D, A PI3K inhibitor (LY294002) suppressed the 4-NQO–induced activation of AKT signaling in three cell lines (SAS/OSCC, HSC3/OSCC, and HeLa cells). Using specific antibodies, the expression of phosphorylated AKT (S473) with total AKT, phosphorylated GSK3β (S9) with total GSK3β, phosphorylated S6 with total S6, and β-actin (as a control) was determined in three cell lines cultured under the four different conditions indicated in the figure. E, A PI3K inhibitor (LY294002) suppressed the 4-NQO–induced migration of SAS/OSCC cells. Images on the left side show SAS cells under the four different conditions on the glass slide; the black lines indicate the initial edge of the cells, while the red arrows indicate the migrated cells after the scratch was made. The bar graph (right side) illustrates the relative migration rates of SAS cells under the four different conditions indicated under the bar. *, P < 0.05. The data represent at least three experiments performed in triplicate.

Figure 4.

Treatment with 4-NQO and the loss of NDRG2 expression acted together to enhance the migration and invasion capacity of OSCC cells. A, Using SAS/OSCC or HeLa cells transfected with the mock control or the NDRG2-expressing plasmid, the effects of 4-NQO treatment on the PI3K/AKT signaling pathway were determined at each time point by Western blot analysis using antibodies specific to NDRG2, Flag, phosphorylated PTEN (STT) with total PTEN, phosphorylated AKT (S473) with total AKT, phosphorylated GSK3β (S9) with total GSK3β, phosphorylated S6 (S240/244) with total S6, and β-actin as a control. B, Cell migration ability under four different conditions with/without expression of NDRG2 and with/without 4-NQO treatment in SAS/OSCC cells was determined by in vitro scratch assays. Images on the left show SAS cells under four different conditions on the glass slide; the black lines indicate the initial edge of the cells, while the red arrows indicate the migrated cells after the scratch was made. The bar graph (right) illustrates the relative migration rates of SAS cells under the four different conditions indicated under the bar. C, The images on the left show migrated SAS/OSCC cells under the four different conditions indicated in the figure; these cells were stained with crystal violet in a cell invasion assay using Boyden chambers (magnification, ×100). The bar graph on right side shows the number of migrated cells under the four different conditions. *, P < 0.05. D, A PI3K inhibitor (LY294002) suppressed the 4-NQO–induced activation of AKT signaling in three cell lines (SAS/OSCC, HSC3/OSCC, and HeLa cells). Using specific antibodies, the expression of phosphorylated AKT (S473) with total AKT, phosphorylated GSK3β (S9) with total GSK3β, phosphorylated S6 with total S6, and β-actin (as a control) was determined in three cell lines cultured under the four different conditions indicated in the figure. E, A PI3K inhibitor (LY294002) suppressed the 4-NQO–induced migration of SAS/OSCC cells. Images on the left side show SAS cells under the four different conditions on the glass slide; the black lines indicate the initial edge of the cells, while the red arrows indicate the migrated cells after the scratch was made. The bar graph (right side) illustrates the relative migration rates of SAS cells under the four different conditions indicated under the bar. *, P < 0.05. The data represent at least three experiments performed in triplicate.

Close modal

Moreover, to determine whether NDRG2 expression suppresses the enhanced cell migration capacity induced by the 4-NQO treatment, a cell scratch test was performed in SAS/OSCC cells with (SAS/NDRG2high) or without NDRG2 expression (SAS/NDRG2low). After treatment with 4-NQO, SAS/NDRG2low cells with mock transfection migrated to almost completely fill the scratched gap after 12 hours; however, SAS/NDRG2high cells could not move inside the gap within 12 hours after treatment with 4-NQO (Fig. 4B). To confirm whether cell invasion capacity in OSCC cells is also suppressed by NDRG2 expression, we next used a Boyden chamber invasion assay. After SAS/NDRG2low cells were treated with 4-NQO, the number of invading cells increased to approximately two times higher than that under no treatment with 4-NQO; however, the numbers of the invading cells were not different before and after treatment of SAS/NDRG2high cells with 4-NQO (Fig. 4C). We also investigated the reverse experiment through the suppression of NDRG2 expression in the HaCaT cell line. We established stable HaCaT cell lines expressing short hairpin RNA against NDRG2 (shNDRG2) or luciferase (shluc) as a control. The level of NDRG2 protein expression was significantly lower in HaCaT/shNDRG2 than in HaCaT/shluc. The phosphorylation rates of the PI3K/AKT signaling pathway (PTEN, AKT, GSK3β, and S6) were remarkably increased in HaCaT/shNDRG2 with the treatment of 4-NQO compared with HaCaT/shluc (Supplementary Fig. S2C). A wound-healing assay revealed that after treatment of 4-NQO, HaCaT/shluc recovered about 50% of the wound area, whereas HaCaT/shNDRG2 covered nearly 100% of the wound area (Supplementary Fig. S2D). Furthermore, the number of invading cells with 4-NQO treatment was increased in HaCaT/shNDRG2 compared with HaCaT/shluc (Supplementary Fig. S2E).

Because the level of phosphorylation of PI3K under the EGF receptor was enhanced by 4-NQO treatment (8), we used LY294002 as a specific inhibitor for PI3K to determine whether the activation status of the AKT signaling pathway induced by 4-NQO might be suppressed by LY294002 treatment in two OSCC (SAS and HSC3) cell lines and one cervical cancer (HeLa) cell line (Fig. 4D). Although 4-NQO treatment enhanced the phosphorylation status of AKT, GSK3β, and S6, treatment with 4-NQO with LY294002 suppressed the phosphorylation of AKT, GSK3β, and S6 in three cancer cell lines. Along with suppression of AKT signaling, the cell migration ability of SAS/NDRG2low cells was also suppressed by LY294002 treatment (Fig. 4E). Therefore, the activation of PI3K by EGF or other receptors is important in the ability of 4-NQO to activate the AKT signaling pathway, which results in enhanced cell migration and invasiveness of OSCC. The phosphorylation levels of the AKT signaling pathway components were significantly enhanced in mouse embryonic fibroblasts (MEF) derived from Ndrg2-deficient mice after 4-NQO treatment (Supplementary Fig. S2F). Additionally, forced NDRG2 expression in HSC3/OSCC cells suppressed their cell migration and invasion abilities after 4-NQO treatment (Supplementary Figs. S2G and H), which suggests that enhanced activation of PI3K by 4-NQO treatment and the loss of NDRG2 expression cooperate in the development of OSCC with enhanced metastasis.

EMT was important for enhanced metastasis after 4-NQO treatment and the loss of NDRG2 expression in OSCC

Epithelial–mesenchymal transition (EMT) is one of the important molecular mechanisms of metastasis, and the expression of NDRG2 was reported to abrogate EMT, which inhibits the invasion and migration of colorectal cancer cells (10). For these reasons, we further investigated whether the loss of NDRG2 expression in OSCC cells contributes to their metastatic potential through enhancement of EMT. Therefore, we initially treated the SAS/OSCC cells with 4-NQO to determine whether the EMT phenotype is enhanced by 4-NQO and whether its effect is suppressed by NDRG2 expression in OSCC cell lines. After treatment with 4-NQO for 24 hours, the classical epithelial morphology of SAS/NDRG2low cells, which formed densely packed colonies, changed to a mesenchymal, spindle-shaped morphology with more dispersed cell aggregates. However, the morphology of SAS/NDRG2high cells maintained their epithelial morphology and continued to form densely packed colonies (Fig. 5A). To confirm this result, we further investigated changes in the expression of genes related to EMT, such as the mesenchymal markers Snail, vimentin, and N-cadherin, and the epithelial marker E-cadherin. This was accomplished by semiquantitative RT-PCR, real-time PCR and immunofluorescence staining using antibodies specific to vimentin, E-cadherin, and NDRG2 (Figs. 5B and C; Supplementary Figs. S3A and B). The expression of mesenchymal markers (Snail, vimentin, and N-cadherin) was significantly enhanced and the expression of E-cadherin was decreased in SAS/NDRG2low cells at 24 hours after 4-NQO treatment. Moreover, the enhanced expression of Snail, vimentin, and N-cadherin along with suppression of E-cadherin expression after 4-NQO treatment was almost completely abrogated in SAS/NDRG2high cells. Conversely, the EMT phenotype was enhanced with 4-NQO treatment in HaCaT/shNDRG2 compared with HaCaT/shluc (Supplementary Fig. S3C). We also confirmed that the expression of mesenchymal markers, such as Snail, vimentin, and N-cadherin, was significantly enhanced and the epithelial marker E-cadherin was decreased in HaCaT/shNDRG2 after the treatment of 4-NQO compared with HaCaT/shluc by semiquantitative RT-PCR, and real-time PCR (Supplementary Figs. S3D and 3E). Therefore, NDRG2 expression might suppress metastatic events of OSCC through the inhibition of EMT progression.

Figure 5.

4-NQO treatment induced EMT, but the activation of EMT by 4-NQO was suppressed by high NDRG2 expression. A, Images show the morphology of SAS cells with or without NDRG2 expression before and 24 hours after the 4-NQO treatment. 4-NQO treatment induced morphological changes in the spindle-shaped cells; however, NDRG2 expression suppressed those morphological changes induced by 4-NQO treatment in SAS cells. B and C, mRNA expression of the mesenchymal genes Snail, TWIST1, vimentin, and N-cadherin was induced by 4-NQO treatment, while the mRNA expression of the epithelial gene E-cadherin was suppressed. After the introduction of the NDRG2 expression plasmid into SAS cells, the changes in gene expression by 4-NQO treatment were significantly suppressed. Semiquantitative RT-PCR was used to determine the expression of the indicated genes (B) and the relative expression rates and statistical significance of each gene were determined by quantitative RT-PCR (C). *, P < 0.05. The data represent at least three experiments performed in triplicate.

Figure 5.

4-NQO treatment induced EMT, but the activation of EMT by 4-NQO was suppressed by high NDRG2 expression. A, Images show the morphology of SAS cells with or without NDRG2 expression before and 24 hours after the 4-NQO treatment. 4-NQO treatment induced morphological changes in the spindle-shaped cells; however, NDRG2 expression suppressed those morphological changes induced by 4-NQO treatment in SAS cells. B and C, mRNA expression of the mesenchymal genes Snail, TWIST1, vimentin, and N-cadherin was induced by 4-NQO treatment, while the mRNA expression of the epithelial gene E-cadherin was suppressed. After the introduction of the NDRG2 expression plasmid into SAS cells, the changes in gene expression by 4-NQO treatment were significantly suppressed. Semiquantitative RT-PCR was used to determine the expression of the indicated genes (B) and the relative expression rates and statistical significance of each gene were determined by quantitative RT-PCR (C). *, P < 0.05. The data represent at least three experiments performed in triplicate.

Close modal

NDRG2 suppressed EMT through the inhibition of the AKT and NF-κB signaling pathways

Although TGFβ signaling has a predominant role in EMT, the convergence of other signaling pathways such as PI3K/AKT and NF-κB is also essential for EMT (11–13). Therefore, we next determined whether the 4-NQO treatment could modulate EMT signaling through the NF-κB and/or PI3K/AKT pathways. To investigate whether 4-NQO treatment modulates the NF-κB signaling pathway, SAS/OSCC and HSC3/OSCC cell lines were treated with 4-NQO and/or LY294002, a PI3K inhibitor, to determine the effects on the NF-κB pathway. The levels of phosphorylated IKKα/β and IκBα were increased by the 4-NQO treatments in the two cell lines, while the total IκBα protein was decreased. However, the enhanced level of phosphorylation induced by 4-NQO was almost completely suppressed by LY294002 treatment (Fig. 6A). Therefore, the activating effects of 4-NQO on NF-κB signaling were derived from the activation of PI3K/AKT. Moreover, forced expression of NDRG2 abrogated the effects of 4-NQO treatment, which increased the phosphorylation of IKKα/β and IκBα but decreased the levels of total IkBα protein (Fig. 6B). To confirm these results, we determined the level of active NF-κB according to the nuclear localization and expression level of the p65 protein, which is a transcription factor of the canonical NF-κB pathway, after 4-NQO treatment with or without forced expression of NDRG2. The level of p65 protein was increased in the nucleus at 2 hours after 4-NQO treatment along with increased NF-κB transcriptional activity; however, the forced expression of NDRG2 in OSCC cells abrogated the nuclear localization of p65 and the activation of NF-κB–mediated transcription (Supplementary Fig. S4A; Fig. 6C). Using MEFs derived from Ndgr2-deficient mice, we investigated the relationship between Ndrg2 expression and the effect of 4-NQO treatment on the NF-κB signaling pathway. The 4-NQO treatment enhanced NF-κB activity with increased phosphorylation of Ikkα/β and IκBα, suppressed total protein of IκBα, and increased NF-κB transcription activity in MEFs derived from Ndrg2-deficient mice compared with MEFs derived from control wild-type mice (Figs. 6D and E). Furthermore, the HaCaT/shNDRG2 cell line showed increased phosphorylation of IKKα/β and IκBα, decreased total IκBα levels, increased nuclear translocation of p65, and increased NF-κB transcriptional activity as compared with the HaCaT/shluc cell line (Supplementary Figs. S4B–S4D).

Figure 6.

4-NQO regulated EMT through the activation of NF-κB signaling. A, 4-NQO treatment activated NF-κB signaling, and the activation of NF-κB signaling by 4-NQO was suppressed by the PI3K inhibitor LY294002 in three cell lines (SAS/OSCC and HSC3/OSCC cells). The expression of phosphorylated IKKα/β, phosphorylated IκBα (S32/36) with total IκBα and β-actin (as a control) was determined by Western blot analysis using specific antibodies in two cell lines under four different conditions, as indicated. B, The SAS/OSCC cell line was treated with 4-NQO at each time point under low (mock) or high NDRG2 expression conditions, and the activation of NF-κB was determined by Western blot analysis. C, NF-κB transcriptional activity was determined by an NF-κB promoter assay using a luciferase reporter under the same condition as in Supplementary Fig. S4. *, P < 0.05. D, Using MEF cells derived from wild-type and Ndrg2−/− mice, we found that 4-NQO treatment activated NF-κB signaling in MEF cells derived from the Ndrg2−/− mice, but not in those derived from wild-type mice; the protein level of phosphorylated IKKα/β, phosphorylated IκBα (S32/36) with total IκBα, and β-actin (as a control) was determined by Western blot analysis. E, NF-κB transcriptional activity was determined by an NF-κB promoter assay using a luciferase promoter under the same condition as in E. *, P < 0.05. The data represent at least three experiments performed in triplicate. F, mRNA expression of the mesenchymal markers Snail, vimentin, and N-cadherin and the epithelial marker E-cadherin was determined by quantitative RT-PCR in eight different combined conditions: NDRG2 (+/−), 4-NQO treatment (+/−), and administration of the NF-κB inhibitor BAY 11-7082 (+/−). mRNA expression of Snail, vimentin, and N-cadherin was induced by 4-NQO treatment; however, NDRG2 expression or BAY 11-7082 treatment inhibited the expression of Snail, vimentin, and N-cadherin mRNA. In contrast, the expression of E-cadherin mRNA was suppressed by 4-NQO treatment; however, NDRG2 expression or BAY11-7082 treatment reversed the inhibition by 4-NQO treatment. *, P < 0.05. The data represent at least three experiments performed in triplicate. G, The same effect of BAY11-7082 was confirmed in E; cells under the same condition by Western blot analysis of E-cadherin, vimentin, and β-actin (as a control).

Figure 6.

4-NQO regulated EMT through the activation of NF-κB signaling. A, 4-NQO treatment activated NF-κB signaling, and the activation of NF-κB signaling by 4-NQO was suppressed by the PI3K inhibitor LY294002 in three cell lines (SAS/OSCC and HSC3/OSCC cells). The expression of phosphorylated IKKα/β, phosphorylated IκBα (S32/36) with total IκBα and β-actin (as a control) was determined by Western blot analysis using specific antibodies in two cell lines under four different conditions, as indicated. B, The SAS/OSCC cell line was treated with 4-NQO at each time point under low (mock) or high NDRG2 expression conditions, and the activation of NF-κB was determined by Western blot analysis. C, NF-κB transcriptional activity was determined by an NF-κB promoter assay using a luciferase reporter under the same condition as in Supplementary Fig. S4. *, P < 0.05. D, Using MEF cells derived from wild-type and Ndrg2−/− mice, we found that 4-NQO treatment activated NF-κB signaling in MEF cells derived from the Ndrg2−/− mice, but not in those derived from wild-type mice; the protein level of phosphorylated IKKα/β, phosphorylated IκBα (S32/36) with total IκBα, and β-actin (as a control) was determined by Western blot analysis. E, NF-κB transcriptional activity was determined by an NF-κB promoter assay using a luciferase promoter under the same condition as in E. *, P < 0.05. The data represent at least three experiments performed in triplicate. F, mRNA expression of the mesenchymal markers Snail, vimentin, and N-cadherin and the epithelial marker E-cadherin was determined by quantitative RT-PCR in eight different combined conditions: NDRG2 (+/−), 4-NQO treatment (+/−), and administration of the NF-κB inhibitor BAY 11-7082 (+/−). mRNA expression of Snail, vimentin, and N-cadherin was induced by 4-NQO treatment; however, NDRG2 expression or BAY 11-7082 treatment inhibited the expression of Snail, vimentin, and N-cadherin mRNA. In contrast, the expression of E-cadherin mRNA was suppressed by 4-NQO treatment; however, NDRG2 expression or BAY11-7082 treatment reversed the inhibition by 4-NQO treatment. *, P < 0.05. The data represent at least three experiments performed in triplicate. G, The same effect of BAY11-7082 was confirmed in E; cells under the same condition by Western blot analysis of E-cadherin, vimentin, and β-actin (as a control).

Close modal

To confirm whether activation of NF-κB signaling is important for the enhanced expression of mesenchymal markers (TWIST, Snail, vimentin, and N-cadherin) and the downregulation of E-cadherin, BAY 11-7082, which is an NF-κB inhibitor, was administered to SAS/OSCC cells with or without forced NDRG2 expression. 4-NQO–induced EMT phenotype was suppressed by the treatment of BAY 11-7082 in SAS/NDRG2low cells (Supplementary Fig. S4E). The 4-NQO–induced enhancement of expression of mesenchymal markers (Snail, vimentin, and N-cadherin) and the downregulation of E-cadherin expression in SAS/NDRG2low cells was abrogated by the treatment of BAY 11-7082 (Fig. 6F and G). The suppressive effect of BAY 11-7082 on the action of 4-NQO in SAS/control cells was abrogated by the forced expression of NDRG2. Therefore, NF-κB activation might be an important signaling pathway in EMT progression and the expression level of NDRG2 may be one of the key factors in the progression of metastasis in OSCC.

In this study, the loss of NDRG2 expression not only promoted oral carcinogenesis in a chemically induced OSCC mouse model using 4-NQO, it also promoted metastasis of human OSCC. High expression of NDRG2 suppresses the PI3K/AKT and NF-κB signaling pathways through the dephosphorylation of PTEN at the C-terminal domain, which results in the suppression of EMT phenotypes in OSCC cells. Because a group of OSCC patients whose tumors were negative for NDRG2 according to immunohistochemistry was significantly correlated with a group of OSCC patients with lymph node metastasis, a low level of NDRG2 expression may contribute to enhanced tumor development with cervical metastasis in OSCC.

For metastasis to occur, cancer cells frequently change their gene expression programs to those of embryonic cells during fetal development, such as EMT and/or various signaling pathways that are induced by TGF-β signaling (14). On the contrary, because local inflammation, hypoxia, and/or infection are the major activators of metastasis and invasion, cigarette smoking, alcohol addiction, and/or viral infections such as HPV are the major risk factors for OSCC. Importantly, NF-κB signaling is established as a critical mediator in the response to the inflammation and infections in cases of OSCC (15). Because activation of the NF-κB pathway is required for the induction and maintenance of Ras-/PI3K/AKT- and TGFβ-dependent EMT (11, 12), in this article, we show that activation of the NF-κB pathway through the loss of NDRG2 expression is crucial for the activation of EMT in OSCC; this process might also be involved in the lymph node metastasis observed in OSCC. Therefore, both the 4-NQO treatment as an activator of PI3K/AKT and the loss of NDRG2 expression as an inhibitor of PTEN activation coordinately enhance the activation of the NF-κB pathway through PI3K/AKT, which results in enhanced carcinogenesis of OSCC with cervical metastasis. Because the loss of expression of NDRG2 might be induced by promoter methylation through activators of the NF-κB pathway (e.g., chronic inflammation and infection; paper in preparation), this suggests that constitutive activators of the NF-κB pathway, such as heavy tobacco use, alcoholism, and chronic infection, might be important risk factors for the development and metastasis of OSCC.

The loss of NDRG2 expression by promoter methylation has been frequently reported to involve tumorigenesis and enhanced metastasis in breast, prostate, hepatocellular, gastric, pancreatic, and colon cancers, among others (16–21). In regard to cancer metastasis, a few papers have reported that the loss of NDRG2 expression enhanced EMT in colorectal, gallbladder, and breast cancer cells (10, 21, 22); however, the direct molecular mechanism by which NDRG2 affects EMT has not yet been fully elucidated. We previously reported that NDRG2 can bind directly to PTEN and NIK to modulate their enzyme activity by phosphorylation via the recruitment of PP2A phosphatase with NDRG2; this in turn leads to the suppression of the PI3K/AKT and NF-κB signaling pathways (3, 23). Because the PI3K/AKT and NF-κB signaling pathways, along with TGFβ signaling, are important regulators of EMT, here we demonstrate the molecular mechanism of EMT progression in OSCC cells that have lost NDRG2 expression (Supplementary Fig. S5). When OSCC cells lose NDRG2 expression, PP2A cannot be recruited to PTEN, which results in a high phosphorylation status; this inactivates the lipid phosphatase activity of PTEN. Because 4-NQO enhances the phosphorylation of PI3K thorough the activation of some receptor-type tyrosine kinases, the activation of AKT is derived from the effects of both PI3K activation by 4-NQO and PTEN inactivation by loss of NDRG2 expression. Phosphorylated AKT activates the canonical NF-κB pathway and NF-κB activates the transcription of the EMT-related transcription factors TWIST and SNAIL, which inhibit the transcription of E-cadherin and activate the transcription of vimentin and N-cadherin. Therefore, most of the published data on the regulation of EMT by NDRG2 are explained in this scheme, with the exception of STAT3 regulation. Because NDRG2 suppresses the expression of SOCS3 (suppressor of cytokine signaling), the level of phosphorylation of STAT3 is enhanced to induce the transcription of SNAIL (21, 24); however, direct evidence as to how NDRG2 regulates the transcription of SOCS and how JAK/STAT signaling regulates SNAIL expression was not found. Therefore, in the future, we need to further analyze the relationship between NDRG2 expression and the JAK/STAT signaling pathway, as well as the relationship between SNAIL transcription and JAK/STAT signaling.

NDRG2 functions as a stress-induced gene that maintains homeostasis in the cellular environment after genotoxic stress that involves p53, hypoxic stress mediated by HIF1α, or inflammation mediated by the NF-κB signaling pathway (25–27). Particularly, according to the work presented in this article, the activation of NF-κB has an important function in the development of OSCC with cervical metastasis. The progression of OSCC might be dependent on the activation of a signal activator, such as 4-NQO, along with the suppression of a signal repressor, such as PTEN or NDRG2 inactivation. When we prevent the progression of OSCC carcinogenesis, the PI3K/AKT and NF-κB signaling pathways may be turned off by specific molecularly targeted drugs. Recently, we revealed that abnormal methylation of the NDRG2 promoter might be dependent on EZH1/EZH2 expression in leukemia cells (paper in preparation). Therefore, in the near future, we plan to identify the relationship between the expression of PRC2 methylation enzymes (e.g., EZH1/EZH2) and the progression and metastasis of OSCC. When we identify the molecular mechanism of abnormal methylation in OSCC, one goal would be the development of a novel therapeutic drug targeted at PRC2 methylation enzymes for the treatment of OSCC.

No potential conflicts of interest were disclosed.

Conception and design: T. Tamura, T. Ichikawa, K. Morishita

Development of methodology: T. Tamura, T. Ichikawa

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Tamura, T. Ichikawa, K. Yamamoto, K. Nagai, R. Yamaguchi, M. Futakuchi, K. Morishita

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Tamura, T. Ichikawa, Y. Tagawa, K. Yamamoto, R. Yamaguchi, K. Morishita

Writing, review, and/or revision of the manuscript: T. Tamura, T. Ichikawa, S. Nakahata, M. Futakuchi, K. Morishita

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Kondo, K. Nagai, R. Yamaguchi, Y. Yamashita, K. Morishita

Study supervision: Y. Kondo, T. Baba, K. Morishita

We thank Y. Motoyoshi for technical assistance and N. Ishigami for secretarial assistance. We thank all of the members of the Division of Tumor and Cellular Biochemistry and Oral and Maxillofacial Surgery at the University of Miyazaki for their helpful discussions and comments. Additionally, we gratefully thank all of the researchers who kindly provided us with important cell lines and materials.

This work was funded in part by grant support from Japan Society for the Promotion of Science [Grant-in-Aid for Scientific Research (B; 25293081; K. Morishita), Young Scientists (B; 25860242; T. Ichikawa), Scientific Research (C; 15K08310; T. Ichikawa)], the Takeda Science Foundation (T. Ichikawa) and the Suzuken Memorial Foundation (T. Ichikawa).

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.

1.
Shu-Jen
C
,
Hsuan
L
,
Tzu-Chen
Y
. 
Ultra-deep targeted sequencing of advanced oral squamous cell carcinoma identifies a mutation-based prognostic gene signature
.
Oncotarget
2015
;
6
:
18066
80
.
2.
Furuta
H
,
Kondo
Y
,
Nakahata
S
,
Hamasaki
M
,
Sakoda
S
,
Morishita
K
. 
NDRG2 is a candidate tumor-suppressor for oral squamous-cell carcinoma
.
Biochem Biophys Res Commun
2010
;
391
:
1785
91
.
3.
Nakahata
S
,
Ichikawa
T
,
Maneesaay
P
,
Saito
Y
,
Nagai
K
,
Tamura
T
, et al
Loss of NDRG2 expression activates PI3K–AKT signalling via PTEN phosphorylation in ATLL and other cancers
.
Nat Commun
2014
;
5
:
3393
.
4.
Daneel
D
,
Christian
C
,
Ulrike
K
. 
Efficacy of quercetin against chemically induced murine oral squamous cell carcinoma
.
Oncol Lett
2015
;
10
:
2432
38
.
5.
Deepak
K
,
Milind
M
. 
4-Nitroquinoline-1-oxide induced experimental oral carcinogenesis
.
Oral Oncol
2006
;
42
:
655
67
.
6.
Ferreira
JB
,
Yamaguti
M
,
Marques
LM
,
Oliveira
RC
,
Neto
RL
,
Buzinhani
M
, et al
Detection of Mycoplasma pulmonis in laboratory rats and technicians
.
Zoonoses Public Health
2008
;
55
:
229
34
.
7.
Jing
L
,
Feixin
L
,
Dahai
Y
,
Haiyun
Q
,
Yiping
Y
. 
Development of a 4-nitroquinoline-1-oxide model of lymph node metastasis in oral squamous cell carcinoma
.
Oral Oncol
2013
;
49
:
299
305
.
8.
Cristiane H
S
,
Rogerio M
C
,
Aline C
A
,
Alfredo
M
,
Mark W
L
,
J Silvio
G
. 
PTEN deficiency contributes to the development and progression of head and neck cancer
.
Neoplasia
2013
;
15
:
461
71
.
9.
Czerninski
R
,
Amornphimoltham
P
,
Patel
V
,
Molinolo
AA
,
Gutkind
JS
. 
Targeting mammalian target of rapamycin by rapamycin prevents tumor progression in an oral-specific chemical carcinogenesis model
.
Cancer Prev Res
2009
;
2
:
27
36
.
10.
Shen
L
,
Qu
X
,
Zhang
J
. 
Tumor suppressor NDRG2 tips the balance of oncogenic TGF-β via EMT inhibition in colorectal cancer
.
Oncogenesis
2014
;
3
:
e86
.
11.
Papageorgis
P
. 
TGFβ signaling in tumor initiation, epithelial-to-mesenchymal transition, and metastasis
.
J Oncol
2015
;
2015
:
587193
.
12.
Saitoh
M
Epithelial-mesenchymal transition is regulated at post-transcriptional levels by transforming growth factor-β signaling during tumor progression
.
Cancer Sci
2015
;
106
:
481
8
.
13.
Murielle
M
,
Surinder K
B
. 
Frequent gene products and molecular pathways altered in prostate cancer– and metastasis-initiating cells and their progenies and novel promising multitargeted therapies
.
Mol Med
2011
;
17
:
949
64
.
14.
Singh
A
,
Settleman
J
. 
EMT, cancer stem cells and drug resistance: an emerging axis of evil in the war on cancer
.
Oncogene
2010
;
29
:
4741
51
.
15.
Nakayama
H
,
Ikebe
T
,
Beppu
M
,
Shirasuna
K
. 
High expression levels of nuclear factor κB, IκB kinase α and Akt kinase in squamous cell carcinoma of the oral cavity
.
Cancer
2001
;
92
:
3037
44
.
16.
Gao
L
,
Wu
GJ
,
Liu
XW
,
Zhang
R
,
Yu
L
,
Zhang
G
, et al
Suppression of invasion and metastasis of prostate cancer cells by overexpression of NDRG2 gene
.
Cancer Lett
2011
;
310
:
94
100
.
17.
Lee
DC
,
Kang
YK
,
Kim
WH
,
Jang
YJ
,
Kim
DJ
,
Park
IY
, et al
Functional and clinical evidence for NDRG2as a candidate suppressor of liver cancer metastasis
.
Cancer Res
2008
;
68
:
4210
.
18.
Mordalska
A
,
Latek
J
,
Ferenc
T
,
Pomorski
L
,
Gałecka
E
,
Zygmunt
A
, et al
Evaluation of NDRG2 gene expression in primary papillary thyroid carcinoma and in metastases of this neoplasm to regional lymph nodes
.
Thyroid Res
2010
;
3
:
6
.
19.
Wang
J
,
Yin
D
,
Xie
C
,
Zheng
T
,
Liang
Y
,
Hong
X
, et al
The iron chelator Dp44mT inhibits hepatocellular carcinoma metastasis via N-Myc downstream-regulated gene 2 (NDRG2)/gp130/STAT3 pathway
.
Oncotarget
2014
;
5
:
8478
91
.
20.
Kim
MJ
,
Kim
HS
,
Lee
SH
,
Yang
Y
,
Lee
MS
,
Lim
JS
. 
NDRG2 controls COX-2/PGE2-mediated breast cancer cell migration and invasion
.
Mol Cells
2014
;
37
:
759
65
.
21.
Kim
MJ
,
Lim
J
,
Yang
Y
,
Lee
MS
,
Lim
JS
. 
N-myc downstream-regulated gene 2 (NDRG2) suppresses the epithelial-mesenchymal transition (EMT) in breast cancer cells via STAT3/Snail signaling
.
Cancer Lett
2014
;
354
:
33
42
.
22.
Lee
DG
,
Lee
SH
,
Kim
JS
,
Park
J
,
Cho
YL
,
Kim
KS
, et al
Loss of NDRG2 promotes epithelial-mesenchymal transition of gallbladder carcinoma cells through MMP-19-mediated Slug expression
.
J Hepatol
2015
;
63
:
1429
39
.
23.
Ichikawa
T
,
Nakahata
S
,
Fujii
M
,
Iha
H
,
Morishita
K
. 
Loss of NDRG2 enhanced activation of the NF-κB pathway by PTEN and NIK phosphorylation for ATL and other cancer development
.
Sci Rep
2015
;
5
:
12841
.
24.
Lee
EB
,
Kim
A
,
Kang
K
,
Kim
H
,
Lim
JS
. 
NDRG2-mediated Modulation of SOCS3 and STAT3 Activity Inhibits IL-10 Production
.
Immune Netw
2010
;
10
:
219
29
.
25.
Wang
L
,
Liu
N
,
Yao
L
,
Li
F
,
Zhang
J
,
Deng
Y
, et al
NDRG2 is a new HIF-1 target gene necessary for hypoxia-induced apoptosis in A549 cells
.
Cell Physiol Biochem
2008
;
21
:
239
50
.
26.
Liu
N
,
Wang
L
,
Li
X
,
Yang
Q
,
Liu
X
,
Zhang
J
, et al
N-Myc downstream-regulated gene 2 is involved in p53-mediated apoptosis
.
Nucleic Acids Res
2008
;
36
:
5335
49
.
27.
Li
T
,
Hu
J
,
He
GH
,
Li
Y
,
Zhu
CC
,
Hou
WG
, et al
Up-regulation of NDRG2 through nuclear factor-kappa B is required for Leydig cell apoptosis in both human and murine infertile testes
.
Biochim Biophys Acta
2012
;
1822
:
301
13
.

Supplementary data