The ESCC samples were collected from 104 patients who had not received chemotherapy or radiotherapy prior to surgery. For these patients, if preoperative ultrasonography or computed tomography (CT) did not show any enlarged cervical lymph nodes (minor axis <0.5 cm), patients underwent a subtotal esophagectomy with two-field lymphadenectomy through a right thoracotomy, followed by a laparotomy. If enlarged cervical lymph nodes were detected by preoperative ultrasonography or CT, patients underwent radical oesophagectomy with cervico-thoraco-abdominal three-field lymphadenectomy through a right thoracotomy, followed by a laparotomy and a cervical incision. The resection extent includes nodes along the cervical part of the esophagus and deep cervix. Tumor adjacent tissues were obtained 3 cm away from the tumor. Fifteen normal esophageal tissue samples were obtained from surgical resections of trauma patients. These normal esophageal tissue samples were obtained from upper endoscopy of trauma patients, which were performed by the same operator. These patients had a fish bone or a poultry bone stuck. These tissues were obtained postoperatively between 2010 and 2013 from the Gastrointestinal Center, Jiangyin People's Hospital, the Jiangyin Clinical College of Xuzhou Medical University (Jiangyin, China) as previously reported (17). All patients provided signed and informed consents for their tissue samples to be used for scientific research. The ethics approval for the study was obtained from the Jiangyin People's Hospital, the Jiangyin Clinical College of Xuzhou Medical University. All diagnoses were based on pathological and/or cytological evidence. The histological features of the specimens were evaluated by a senior pathologist according to the classification criteria from the World Health Organization (18). Tissue samples were obtained prior to chemotherapy and radiation therapy and were immediately fixed in 10% neutral buffered formalin prior to immunohistochemistry analysis.

The tissue samples were fixed in 10% neutral buffered formalin and embedded in paraffin. Three-micrometer thick paraffin sections were deparaffinized and heat-treated with citrate buffer (pH 6.0) for 7 min as an epitope retrieval protocol. Endogenous peroxidase was blocked with 3% hydrogen peroxide for 15 min at room temperature and tissue non-specific binding sites were blocked with skim milk powder at 4% applied for 30 min. Sections were then incubated with the AUF1 and GCH1 antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 1 h and mixed with skim milk powder at 2% again to reduce the non-specific staining. Biotinylated secondary antibody was then added for 30 min. IHC staining was visualized with substrate solution containing diaminobenzidine (DAB) and hydrogen peroxide. The counter-staining was performed with haematoxylin. All steps were performed at room temperature. Negative controls consisted of tissue sections undergoing similar staining procedures in the absence of the primary antibody. The criteria for scoring the stained sections were as follows: negative (0), <10% of the whole tissue section stained positive; weakly positive (+ 1), 10–25% of the whole tissue section stained positive; moderately positive (+ 2), 25–75% of the whole tissue section stained positive; and strongly positive (+ 3), 75% of the tissue section stained positive. Tissues with score 0 or 1 were considered as AUF1-negative, while with score 2 or 3 were considered as AUF1-positive.

The human esophageal cancer cell line Eca-109 was maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and antibiotics (Gibco, Grand Island, NY, USA). Cells were grown in a 37°C incubator with 5% CO₂.

The siRNA targeting AUF1 (siRNA-AUF1) and siRNA control (siRNA-NC) used in the present study were obtained from Santa Cruz Biotechnology. The cells were grown in a 6-well culture plate to 70–80% confluence and then transfected with either siRNA-NC, siRNA-AUF1 or siRNA-GCH1 (Santa Cruz Biotechnology) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. After 48 h, the cells were collected for western blot analysis.

Cell viability was evaluated using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay. Cells were seeded in 96-well plates. The next day, the cells were transfected with plasmids or siRNAs according to the experimental design. The cells were then incubated with 20 μl MTT (5 mg/ml) for 4 h. After the medium was removed, 100 μl DMSO was added and the optical density (OD) at 490 nm was measured using a microplate reader (Bio-Rad Laboratories, Hercules, CA, USA). The viability index was calculated as experimental OD value/control OD value. Three independent experiments were performed in quadruplicate.

Cells transfected with the indicated vectors were plated at low density (1,000 cells/6-cm plate), incubated for 10 days and fixed and stained with crystal violet. Foci and colonies containing >50 cells were counted using a microscope.

After transfection, cells were harvested and lysed in RIPA lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) containing protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA) for 30 min at 4°C. Protein (40 μg) from each lysate was fractionated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA, USA). After blocking with 5% non-fat milk in PBS-Tween-20 for 1 h at room temperature, the membranes were blotted with AUF1 and GCH1 primary antibodies (Abcam, Cambridge, UK). GAPDH (Beyotime Institute of Biotechnology, Nantong, China) was used as a loading control. After washing four times with TBST, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibody for 2 h. The proteins were visualized using the enhanced chemiluminescence (ECL; Beyotime Institute of Biotechnology).

Microarray-based mRNA expression profiling was performed using the Roche-NimbleGen (135K array) array (Roche NimbleGen, Madison, WI, USA). The microarrays contained approximately 45,033 assay probes corresponding to all of the annotated human mRNA sequences (NCBI HG18). Total RNA labeling and hybridization were performed using standard conditions according to the manufacturer's instructions. Genes with fold change of 2 or greater were subsequently subjected to pathway analysis using Ingenuity Pathway Analysis (Qiagen, Redwood City, CA, USA).

The total RNA from the esophageal tissue samples was extracted with TRIzol (Invitrogen) and reverse transcribed to complementary DNA (cDNA) using an oligo(dT)₁₂ primer and SuperScript II (Invitrogen). The SYBR-Green dye (Takara Bio, Osaka, Japan) was used for the amplification of cDNA. The mRNA levels of transcripts were measured using real-time quantitative PCR in triplicate on a Prism 7500 real-time PCR machine (Applied Biosystems, Foster City, CA, USA). Primers specific for the genes are listed in Table I. PCR was performed in a 20 μl reaction mixture. The thermal cycling conditions included an initial denaturation step of 95°C for 20 sec, followed by 40 cycles at 95°C for 5 sec and annealing at 60°C for 30 sec. The relative mRNA expression levels were normalized to GAPDH.

Two luciferase reporters containing the full-length 3′UTR fragment of the GCH1 3′UTR (+1 to +1063, relative to the stop coden) was constructed by Shanghai Biobuy Co., Ltd. (Shanghai, China). Then, constructs each with a deletion of the two potential AU-rich elements in the GCH1 3′UTR were constructed. For the luciferase assay, the reporter vector with the GCH1 3′UTR was co-transfected with pRL-TK (Promega, Madison, WI, USA) to correct the transfection efficiency. Luciferase activity was measured with the Dual-luciferase reporter assay system (Promega). Promoter activities were expressed as the ratio of firefly luciferase to Renilla luciferase activities.

Total NO concentration in cultured cells was detected by measuring the concentration of nitrate and nitrite by modified Griess reaction method. After lysis of cells, a total Nitric Oxide assay kit (Beyotime Institute of Biotechnology) was used to detect NO. The optical densities at 540 nm wavelength were recorded using a microplate reader (BioTek Instruments, Winooski, VT, USA) and the concentrations of NO were calculated according to the standard curve.

Data are expressed as the means ± standard error of the mean (SEM) of at least three independent experiments. Standard error bars were included for all data points. The data were first analyzed with the Kolmogorov-Smirnov test for data distribution normality. The data were then analyzed using Student's t-test when only two groups were present or assessed by one-way analysis of variance (ANOVA) when more than two groups were compared. The P-values for t-tests were performed by 2-tailed t-tests. Correlation analysis of the mRNA expression data was performed using the Pearson test. Statistical analysis was performed using SPSS software (Release 19.0; SPSS, Inc., Chicago, IL, USA). Data were considered significant at P<0.05.

To investigate the expression of AUF1 in human ESCC tissues, we performed immunohistochemistry analysis on tissue section of normal esophageal tissue specimens, ESCC and tumor adjacent tissues. Fifteen paraffin-embedded normal esophageal tissues and 104 ESCC tissues, 54 of which had corresponding adjacent tissues were collected. The demographic features of patients with ESCC for the immunohistochemistry staining are summarized in Table II. Results from IHC staining showed that a generally stronger staining of AUF1 was observed in the nucleus of ESCC tissues, whereas very weak staining of AUF1 was observed in the normal samples (Fig. 1A). When the samples were scored by a pathologist, as described in Materials and methods, AUF1 expression was similar between normal esophageal tissues and tumor adjacent tissues, while increased significantly among ESCC tissues, and between tumor tissues and tumor adjacent tissues (P<0.05; Fig. 1B). All the above results indicate that AUF1 may be involved in the pathogenesis of ESCC.

Next, the relationship between clinicopathology of ESCC patients and AUF1 IHC staining score was analyzed. As shown in Table III, the results showed a non-significant correlation between AUF1 expression and age (P=0.610), gender (P=0.637) and lymphatic invasion (P=0.096) of ESCC patients. AUF1 expression correlated with ESCC stage (P=0.011) and marginally correlated with lymph node metastasis (P=0.055). These results suggest AUF1 as a characteristic of this malignancy.

Because AUF1 expression was associated with ESCC progression, we sought to explore the role of AUF1 in esophageal cancer cell growth. Eca-109 cells were transfected with a siRNA control (siRNA-NC) or siRNA trageting AUF1 (siRNA-AUF1). As shown in Fig. 2A, the protein level of AUF1 was markedly reduced after transfection of siRNA-AUF1. Next, the effect of AUF1 on cell growth was measured with MTT and colony formation. The results showed that decreased AUF1 expression suppressed cell growth and focus formation (Fig. 2B and C). We next investigated whether reduced cell growth was associated with apoptosis in ESCC cells. As shown in Fig. 2E, silencing of AUF1 significantly increased apoptosis of Eca-109 cells. These results demonstrated that AUF1 modulates the proliferation and apoptotic cell death of ESCC cells.

Although many AUF1-regulated genes have been identified, the profile of AUF1-affected genes in ESCC cells remains elusive. To further analyze the underlying mechanisms responsible for AUF1-mediated effects, we screened gene expression between AUF1 silencing (siRNA-AUF1) and the control cells (siRNA-NC) by microarray. Twenty-four hours after the transfection, a total of 285 genes (138 upregulated and 147 downregulated genes) were identified with an expression differential of 2-fold or greater between the two groups (n=2; Fig. 3A and Table IV). The differentially expressed genes included Nalcn, MMP23B, E2F1, BEX2 and GCH1. We then validated the results from microarray analysis by RT-PCR. As shown in Fig. 3B, compared with control cells, several genes dysregulated in microarray data, such as Nalcn, UGP2, BEX2 and Prdm16 showed consistent changes by real-time PCR analysis. As expected, AUF1 appeared to have modulated the Eca-109 cells via complex mechanisms. Pathway analysis revealed that AUF1 affected multiple pathways, including MAPK signaling pathway, transcriptional regulation and adipocytokine signaling pathway (Fig. 3C). These results suggested that AUF1 might be a wide-spectrum regulator of gene expression in esophageal cancer cells.

Among AUF1 affected genes, we focused on GCH1, which catalyzes the formation of 5,6,7,8-tetrahydrobiopterin (BH4), an essential co-factor for all nitric oxide synthases (19,20). GCH1 has been reported to contribute to cancer progression (21). Sequence analysis of the GCH1 3′UTR identified two potential AU-rich elements (Fig. 4A), which were likely to be regulated by AUF1. We therefore investigated whether AUF1 modulated the expression of GCH1 and cancer progression through GCH1. The effect of AUF1 on the GCH1 3′UTR was also measured using luciferase constructs containing the GCH1 3′UTR with or without AUF1 binding sites (Fig. 4A). The results revealed that deletion of the ARE1 (pGL3-GCH1 3UTR Del1) showed significantly decreased luciferase activity compared with that of wild-type GCH1 3′UTR in Eca-109 cells (Fig. 4B). However, deletion of the ARE2 (pGL3-GCH1 3UTR Del2) showed similar luciferase activity as compared with that of wild-type GCH1 3′UTR in Eca-109 cells (Fig. 4B). Western blot analysis further confirmed that knockdown of AUF1 decreased the expression of GCH1 (Fig. 4C). Because GCH1 is an essential co-factor for all NOS, as expected, silencing of AUF1 significantly reduced cellular NO level. These results suggested that AUF1 regulated GCH1 through ARE1 region (Fig. 4D).

The above mentioned results led us to investigate whether the expression level of AUF1 correlated with that of GCH1 in tissue samples. Immunohistochemistry was used to detect the expression levels of GCH1 protein in 102 ESCC tissue samples. Representative IHC staining of both AUF1 and GCH1 in the same tissue is shown in Fig. 5A. Tissue samples with strong AUF1 staining in the nucleus exhibited positive GCH1 staining, whereas weak AUF1 expression showed negative GCH1 staining. Statistical analysis revealed that the protein levels of the two genes correlated with each other (Pearson r=0.251, P=0.019; Fig. 5B), indicating that the expression of AUF1 correlates with the expression of GCH1 in human ESCC tissues.

To explore the role of GCH1 in esophageal cancer cell growth, cells were first transfected with a siRNA-NC or siRNA-GCH1, and the resulting growth was measured with MTT and colony formation assay. The results showed that GCH1 silencing caused significantly decreased cell viability (75% of the control group). Moreover, knockdown of GCH1 attenuated the number and the size of colonies of Eca-109 cells. These results suggest that GCH1 promotes esophageal tumor growth in vitro. (Fig. 6).