The present study was approved by the Ethics Committee of The Third Affiliated Hospital of Kunming Medical University, and informed consent was obtained from all participants prior to enrollment. A total of 16 paired OTSCC tissues and adjacent non-tumor tissues (distance from the tumor of >2 cm) were collected from OTSCC patients who had undergone surgical resection at The Third Affiliated Hospital of Kunming Medical University between May 2010 and August 2010. The fresh tissues were immediately immersed in RNAlater (Ambion, Inc., Austin, TX, USA) after surgical resection, stored at 4°C overnight to allow thorough penetration of the tissue and then frozen at −80°C until RNA and protein extraction was performed. An additional 202 paraffin-embedded OTSCC samples, which were pathologically and clinically diagnosed between January 2000 and December 2005 at The Third Affiliated Hospital of Kunming Medical University, were collected for immunohistochemistry. These samples were from 112 males and 90 females, with a median age of 53 years (ranging from 21 to 78 years), and none of these patients received radiotherapy or chemotherapy prior to surgery. The clinicopathological features and BATF2 expression levels for the 202 patients are shown in Table I. The histological differentiation of the samples was determined according to the criteria of the World Health Organization. The tumor (T) classification, node (N) classification and clinical tumor-node-metastasis (TNM) stage were assessed according to the TNM classification of the American Joint Committee on Cancer (AJCC) (12).

A normal tongue epithelial cell line (NTEC1) was established by culturing normal tongue squamous epithelium from a non-tumor patient in keratinocyte/serum-free medium (Invitrogen Life Technologies, Carlsbad, CA, USA). OTSCC cell lines were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) (CAL-27, SCC-25 and SCC-9 cells) or from the Committee of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China) (TCA8113 and TSCCA cells). All cells were cultured in DMEM/F12 (Invitrogen Life Technologies) supplemented with 10% fetal bovine serum (HyClone, Logan, UT, USA), penicillin (100 U/ml) and streptomycin (100 U/ml) at 37°C in a humidified 5% CO₂ incubator.

Total RNA from the cell lines and human tissues was extracted using TRIzol reagent (Invitrogen Life Technologies). After reverse transcription of the total RNA, the first-strand cDNA was used as a template for detecting the expression of BATF2. qPCR was performed with an ABI PRISM 7900HT sequence detection system. The housekeeping gene GAPDH was used as an internal control to normalize the expression levels of BATF2. The primer sequences were: 5′-AGACCCCAAGGAGCAACA-3′ (sense), and 5′-CTTTTTCCAGAGACTCGTGCT-3′ (antisense) for BATF2, and 5′-CTCCTCCTGTTCGACAGTCAGC-3′ (sense), and 5′-CCCAATACGACCAAATCCGTT-3′ antisense for GAPDH. To ensure the reproducibility of the results, all experiments were repeated 3 times. The data were analyzed using the comparative threshold cycle (2^−ΔΔCT) method.

Western blot analysis was performed as previously described (10). When relevant, the blots were probed with an anti-BATF2 mouse monoclonal antibody (1:1,000 dilution; Abnova Corporation, Taiwan) and an anti-GAPDH mouse monoclonal antibody (1:1,000 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). The signals were detected using enhanced chemiluminescence (ECL) (Amersham Pharmacia Biotech, Piscataway, NJ, USA) according to the manufacturer’s suggested protocol.

In brief, paraffin-embedded tongue tissue specimens were cut into 4-μm sections and incubated at 65°C for 1 h. All sections were deparaffinized with xylene and rehydrated in a graded ethanol series. After treatment with 3% H₂O₂ for 15 min to block the endogenous peroxidase, the sections were submerged in EDTA antigen retrieval buffer (pH 8.0) and microwaved. Then, the sections were treated with normal goat serum for 30 min to reduce any nonspecific binding and were incubated with an anti-BATF2 mouse monoclonal antibody (1:1,000; Abnova Corporation) overnight at 4°C. After washing, the sections were incubated with a biotinylated anti-rabbit secondary antibody, followed by further incubation with streptavidin-horseradish peroxidase (both from Dako) at 37°C for 30 min. To produce a color reaction, diaminobenzidine (DAB) was used. For the negative controls, the antibody was replaced with normal goat serum. The immunostained samples were evaluated by two independent pathologists who did not have knowledge of the clinical data. According to previous studies (10,13), the intensity of the cells expressing BATF2 was scored as follows: 0 (no staining), 1 (weak staining, light yellow), 2 (moderate staining, yellow-brown) or 3 (strong staining, brown). The percentage of positive staining was scored as follows: 0 (no expression), 1 (1–25%), 2 (26–50%), 3 (51–75%) or 4 (>75%). The BATF2 expression level was calculated as the intensity score plus the proportion score and was divided into 4 grades: − (score of 0 and 1), + (score of 2 and 3), ++ (score of 4 and 5) and +++ (score of 6 and 7). In the present study, ‘−’ and ‘+’ represented no/low expression (score of 0–3), whereas ‘++’ and ‘+++’ were regarded as high expression (score of 4–7).

Overall survival was calculated from the date of surgery until the day of death or the last date in the medical records on which the patient was reported to be alive. Disease-free survival was calculated from the date of surgery until the last follow-up date on which the patient was not found to have tumor recurrence. A paired samples t-test was used to compare the mRNA and protein expression of BATF2 in the paired OTSCC and adjacent non-tumor tissue samples. A χ² test was used to analyze the relationship between BATF2 expression and the clinicopathological features of the OTSCC patients. Survival curves were plotted using the Kaplan-Meier method and compared with the log-rank test. The Cox proportional hazard regression model was used for univariate and multivariate analyses to explore the influence of the clinicopathological variables and BATF2 expression on survival. SPSS 18.0 software (SPSS, Chicago, IL, USA) was used for all statistical analyses, and a P-value of <0.05 was considered to indicate a significant result.

To examine the expression levels of BATF2, qPCR and western blot analysis were conducted using the normal tongue epithelial cell line NTEC1 and the OTSCC cell lines SCC-9, SCC-25, CAL-27, TSCCA and TCA8113. qPCR revealed lower expression of BATF2 mRNA in all of the cancer cell lines when compared to the NTEC1 cell line (Fig. 1A). As shown in Fig. 1B, high expression of the BATF2 protein was observed in the NTEC1 cell line, whereas the expression of the BATF2 protein was significantly lower, i.e., at an undetectable level, in the OTSCC cell lines. These findings are consistent with those from our qPCR experiment. Thus, our data indicated that BATF2 was downregulated at both the mRNA and protein levels in the OTSCC cell lines.

To determine the expression of BATF2 in OTSCC tissues, qPCR and western blot analysis were performed using 16 OTSCC tissues and their matched adjacent non-tumor tissues. BATF2 mRNA was found to be significantly downregulated in 13 (81.25%, 13/16) of the OTSCC tissues when compared with their matched adjacent non-tumor tissues (P=0.001) (Fig. 2A and B). Similarly, the BATF2 protein was also significantly downregulated in 14 (87.5%, 14/16) of the tumor samples when compared with the adjacent non-tumor tissues from the same patients (P<0.001) (Fig. 2C).

To verify the relationship between BATF2 and the development of OTSCC, we examined BATF2 expression in 202 paraffin-embedded OTSCC samples and 30 adjacent non-tumor tissue specimens by immunohistochemistry. As a result, among the 30 non-tumor tissue specimens, 18 cases (60%) showed high expression (Fig. 3A and B), 8 cases (26.7%) showed low expression and 4 cases (13.3%) showed negative expression. However, among the 202 OTSCC samples, 20 cases (9.9%) showed negative expression (Fig. 3C and D), 104 cases (51.5%) showed low expression (Fig. 3E and F) and 78 (38.6%) cases showed high expression (Fig. 3G and H). As shown in Table I, the BATF2 expression level was significantly correlated with histological differentiation (P=0.002) but not with age, gender, tobacco history, T classification, N classification or clinical TNM stage.

The patients with no/low BATF2 expression had a significantly poorer 5-year overall survival rate and disease-free survival rate than those with high BATF2 expression (50.5 vs. 89.7%, P<0.001, Fig. 4A) (42.8 vs. 86.5%, P<0.001, Fig. 4B). Univariate Cox regression analyses demonstrated that clinical TNM stage, histological differentiation and BATF2 expression level were significantly associated with overall survival. Furthermore, multivariate Cox regression analyses revealed that the above mentioned 3 factors are independent predictors for the overall survival of OTSCC patients (P=0.000, 0.001 and 0.001, respectively; Table II). These findings indicate that a high level of BATF2 expression is a biomarker for better prognosis in patients with OTSCC.