Aberrant methylation and decreased expression of the TGF-β/Smad target gene FBXO32 in esophageal squamous cell carcinoma

Authors

  • Wei Guo PhD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Minghui Zhang MD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Supeng Shen PhD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Yanli Guo PhD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Gang Kuang MD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Zhibin Yang MD,

    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    Search for more papers by this author
  • Zhiming Dong PhD

    Corresponding author
    1. Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Shijiazhuang, Hebei, China
    • Corresponding author: Zhiming Dong, PhD, Laboratory of Pathology, Hebei Cancer Institute, the Fourth Hospital of Hebei Medical University, Jiankang Road 12, Shijiazhuang 050011, Hebei, China; Fax: (011) 86-311-86077634; dongzhiming2000@gmail.com

    Search for more papers by this author

  • We thank the patients and control individuals for taking part in this study.

Abstract

BACKGROUND

F-box protein 32 (FBXO32) (also known as atrogin-1), a member of the F-box protein family, has recently been identified as a transforming growth factor beta (TGF-β)/Smad target gene involved in regulating cell survival, and it may be transcriptionally silenced by epigenetic mechanisms in some kinds of carcinomas, yet its role in esophageal squamous cell carcinoma (ESCC) has not been defined.

METHODS

The role of FBXO32 in ESCC and the correlation of FBXO32 methylation with a series of pathologic parameters were studied in a large cohort of patients with ESCC.

RESULTS

Decreased messenger RNA (mRNA) expression and protein expression of FBXO32 were observed in esophageal cancer cell lines, and the silencing of FBXO32 could be reversed by treatment with 5-aza-2′-deoxycytidine or trichostatin A in the TE13 cell line. In addition, aberrant methylation of FBXO32 and histone deacetylation was capable of suppressing FBXO32 mRNA and protein expression in TE13 cells. Decreased mRNA and protein expression of FBXO32 was observed in ESCC tumor tissues and was associated with FBXO32 promoter methylation status. A positive correlation between FBXO32 and phosphorylated SMAD family members 2 and 3 expression and Smad4 protein expression also was observed in clinical specimens. FBXO32 methylation status and protein expression were independently associated with survival in patients with ESCC.

CONCLUSIONS

FBXO32 may be a functional tumor suppressor. Its inactivation through promoter methylation could play an important role in ESCC carcinogenesis, and reactivation of the FBXO32 gene may have therapeutic potential and might be used as a prognostic marker for patients with ESCC. Cancer 2014;120:2412–2423. © 2014 American Cancer Society.

INTRODUCTION

Esophageal cancer is the eighth most common malignancy and the sixth most common cause of cancer-related death worldwide.[1] Because its prevalence and death rate are continuously increasing, it has become a major health concern.[2] Esophageal squamous cell carcinoma (ESCC) is the predominant type of esophageal cancer, comprising almost 95% of cases. Because of the high incidence and poor survival, ESCC is 1 of the major health issues in China, especially in some counties bordering Henan, Hebei, and Shanxi Provinces. The precise molecular mechanisms of development and/or progression of ESCC remain unknown, although multiple genetic and epigenetic changes have been detected in ESCC.[3, 4] Therefore, additional elucidation of the molecular mechanisms involved in ESCC is urgently needed for more effective treatment.

F-box protein 32 (FBXO32) (also known as atrogin-1) is a member of the F-box protein family and constitutes 1 of the 4 subunits of the ubiquitin protein ligase complex involved in muscle atrophy.[5, 6] However, recent findings have demonstrated that FBXO32 may have a potential tumor suppressor function and that the v-akt murine thymoma viral oncogene (AKT) (protein kinase B) signaling pathway negatively regulates FBXO32.7 Furthermore, it has been reported that FBXO32 contributes to 3-Deazaneplanocin A (DZNep)-induced apoptosis in breast cancer cells, thus suggesting that it also may be involved in regulating cell survival.[8] More recently, it has been demonstrated that FBXO32 is transcriptionally silenced by epigenetic mechanisms in MCF-7 breast cancer cells and in a panel of ovarian cancer cell lines and that the methylation status of FBXO32 may predict survival in patients with ovarian cancer.[8, 9] Furthermore, Qin et al have identified FBXO32 as a transforming growth factor beta (TGF-)β/Smad target gene in ovarian surface epithelial cells using the chromatin immunoprecipitation microarray (ChIP-Chip) method.[10] The TGF-β signaling pathway plays an important role in the regulation of numerous effects on cell proliferation, differentiation, migration, and survival. The loss of the TGF-β response is well known for its contribution to oncogenesis and tumor progression.[11, 12]

Even with the potential importance of the FBXO32 gene in carcinogenesis, there has been very little investigation into the role of the FBXO32 gene in ESCC. In the current study, we attempted to ascertain the role of FBXO32 in ESCC and the correlation of FBXO32 methylation status to a series of pathologic parameters in a large sample of patients with ESCC to elucidate more information on the role of FBXO32 with regard to the pathogenesis of ESCC.

MATERIALS AND METHODS

Cell cultures and treatments, cell proliferation assays, the effects of FBXO32 on regulating apoptosis, the transfection of small-interfering RNA (siRNA), and Western blot analyses all are described in the supporting materials (see online supporting information).

Patients and Specimens

All study patients were of ethnically homogeneous Han nationality and were residents of Hebei Province and its surrounding regions. Tumor samples and corresponding adjacent normal tissues were obtained from 132 patients with ESCC who were admitted as inpatients for surgical treatment at the Fourth Affiliated Hospital, Hebei Medical University between the years 2005 and 2007. The patients included 93 men and 39 women, and the mean patient age was 58.7 years (age range, 37-78 years). Individuals with at least 1 first-degree relative or at least 2 second-degree relatives who had esophageal/cardia/gastric cancer were defined as a having family history of upper gastrointestinal cancers (UGICs). Tumor samples and corresponding normal tissues were divided into 2 parallel parts: 1 part was frozen and stored at −80°C until DNA and RNA were extracted, and the other part was fixed in formalin and embedded in paraffin. Histologic tumor typing of the samples was carried out on the basis of resected specimens in the Department of Pathology at the same hospital. Information on clinicopathologic characteristics was available from hospital recordings and pathologic diagnoses. Recurrence and survival data were ascertained through the tumor registry and hospital chart review (Supporting Table 1; see online supporting information). The study was approved by the Ethics Committee of Hebei Cancer Institute, and informed consent was obtained from all participating patients.

Table 1. Immunohistochemical Staining Characteristics of F-Box Protein 32, Phosphorylated Smad2/3, and Smad4 in Patients With Esophageal Squamous Cell Carcinoma
  FBXO32 Methylation FrequencyFBXO32 Protein Expressionp-Smad2/3 Protein ExpressionSmad4 Protein Expression
GroupsTotal No.No. (%)PNo. (%)PNo. (%)PNo. (%)P
  1. Abbreviations:FBXO32, F-box protein 32; LN, lymph node; p-Smad2/3, phosphorylated SMAD family members 2 and 3; Smad4, SMAD family member 4; TNM, tumor, lymph node, metastasis; UGIC, upper gastrointestinal cancer.

Age         
<503118 (58.1) 8 (25.8) 11 (35.5) 13 (41.9) 
≥5010151 (50.5).46033 (32.7).47044 (43.6).42550 (49.5).460
Sex         
Men9350 (53.8) 26 (28) 38 (40.9) 42 (45.2) 
Women3919 (48.7).59615 (38.5).14217 (43.6).77221 (53.8).362
TNM stage         
I/II7231 (43) 29 (40.3) 37 (51.4) 41 (56.9) 
III/IV6038 (63.3).02012 (20).01218 (30).01322 (36.7).020
Pathologic tumor differentiation         
Well/moderate8035 (43.8) 28 (35) 39 (48.8) 46 (57.5) 
Poor5234 (65.4).01513 (25).22516 (30.8).04117 (32.7).005
Depth of invasion         
T1/T25120 (39.2) 22 (43.1) 27 (52.9) 31 (60.8) 
T3/T48149 (60.5).01719 (23.4).01728 (34.6).03732 (39.5).017
LN metastasis         
Negative: N02811 (39.3) 12 (42.9) 15 (53.6) 18 (64.3) 
Positive: N1/N2/N310458 (55.8).12129 (27.9).12940 (38.5).15045 (43.3).048
Distant metastasis or recurrence         
Negative7633 (43.4) 30 (39.5) 39 (51.3) 43 (56.6) 
Positive5636 (64.3).01811 (19.6).01516 (28.6).00920 (35.7).018
Family history of UGIC         
Negative7535 (46.7) 29 (38.7) 36 (48) 42 (56) 
Positive5734 (59.6).13912 (21).03019 (33.3).09021 (36.8).029

Stable Gene Transfections

To investigate the overexpression of FBXO32, exponentially grown TE13 cells cultured in 6-well plates were transfected with FBXO32 expression plasmid (pcDNA3.1-FBXO32) or with the relevant empty vector (pcDNA3.1-EV) as a control. The expression plasmid contained the full-length combinational DNA (cDNA) of FBXO32.

ChIP Assay

A ChIP assay was performed using the kit from Upstate Biotechnology (Lake Placid, NY) as described previously.[13] Antibodies against dimethylated histone H3 at lysine 4 (H3K4me2), dimethylated histone H3 at lysine 9 (H3K9me2), and acetylated histone H3 at lysine 9 (H3K9ac) (Millipore Corporation, Inc., Billerica, Mass) were used for immunoprecipitation. The following FBXO32specific primers were used to perform real-time polymerase chain reaction (PCR) analyses from immunoprecipitated DNA as well as input samples: forward, 5′-GGTCGATCCTGATAGCTCGG-3′; and reverse, 5′-GTGGAAACTTGAAGCGGTGC-3′. These primers amplified a 251-base pair (bp) region overlapping the cytosine-phosphate-guanine (CpG) island within the FBXO32 promoter. Amplifications were performed in triplicate, and enrichment was determined by comparing the results with input values.

Regular Reverse Transcriptase-PCR and Quantitative Real-Time Reverse Transcriptase-PCR Assays

Total RNA was extracted from cell lines treated or untreated with 5-aza-2′-deoxycytidine (5-Aza-dC), or trichostatin A (TSA), or calmodulin-binding protein (CBP); from the stable transfected TE13 clones; and from the frozen tumor sections and corresponding normal tissues with standard methods using a Trizol reagent (Invitrogen, Carlsbad, Calif). Two micrograms of RNA were used to synthesize cDNA with the advantage reverse transcriptase-PCR (RT-for-PCR) kit (Clontech, Palo Alto, Calif) with oligo (dT) priming as recommended in the protocol provided. cDNA from each sample was used for a regular RT-PCR and quantitative real-time RT-PCR template, and primers for FBXO32 were used as previously described.[9] The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene was used as an internal control. All primers and reaction conditions are listed in Supporting Table 2 (see online supporting information). For quantitative real-time RT-PCR, power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, Calif) was used as an amplification reaction mixture according to the manufacturer's instructions. The product of FBXO32 was normalized by using the product of GAPDH. All samples were run in triplicate.

Table 2. Association of F-Box Protein 32 Expression and Methylation Status in Patients With Esophageal Squamous Cell Carcinoma
 FBXO32 Methylationp-Smad2/3 ExpressionSmad4 Expression
FBXO32 Protein ExpressionMUP+P+P
  1. Abbreviations: −, negative; +, positive; FBXO32, F-box protein 32; M, methylated; p-Smad2/3, phosphorylated SMAD family members 2 and 3; Smad4, SMAD family member 4; U, unmethylated.

Positive1427 356 374 
Negative5536.0052071< .00052665< .0005

FBXO32 Luciferase Constructs

To explore the transcriptional regulation of FBXO32, three promoter reporter plasmids were constructed: FBXO32-F1, spanning from −851 to +200 bp; FBXO32-F2, spanning from −600 to +200 bp; and FBXO32-F3, spanning from −300 to +200 bp. The amplified fragments were inserted into the pGL3-basic vector (Promega, Madison, Wis) between the XhoI and HindIII sites. These recombination plasmids were then sequenced for confirmation.

Luciferase Assay

TE13 cells (1 × 105 cells per well) were seeded onto a 24-well dish 24 hours before transfection. In all, 200 ng of FBXO32 deletion constructs (FBXO32-F1, FBXO32-F2, and FBXO32-F3), pGL3-control vector construct (positive control), or pGL3-basic vector construct (negative control) were cotransfected with 10 ng of the herpes simplex virus-thymidine kinase promoter vector pRL-TK using Lipofectamine 2000 (Invitrogen). Luciferase activity was measured with the Dual-Luciferase Reporter Assay System (Promega). Promoter activity was expressed as the ratio of Firefly luciferase to Renilla luciferase activity, and each data point was determined in triplicate.

In Vitro DNA Methylation

Construct pGL3-F3 was methylated in vitro as described previously.[14] Then, 200 ng of mock or SSI-treated vector were transfected in TE13 cells. After 48 hours, luciferase and Renilla activity were assayed as described above.

Immunocytochemical and Immunohistochemical Staining

Immunohistochemistry was used to evaluate FBXO32 expression in four 5-Aza-dC-treated or untreated esophageal cancer cell lines. FBXO32, phosphorylated SMAD family members 2 and 3 (p-Smad2/3), and SMAD family member 4 (Smad4) expression was determined by immunostaining using the avidin-biotin complex immunoperoxidase method, which was performed on parallel histopathologic sections from paraffin-embedded tumor sections and corresponding normal tissues. Rabbit antihuman polyclonal antibody for FBXO32 (1:200 dilution; Abcam plc, Cambridge, United Kingdom), goat antihuman polyclonal antibody for p-Smad2/3 (1:200 dilution; Santa Cruz Biotechnology Inc., Santa Cruz, Calif), or mouse antihuman monoclonal antibody Smad4 (1:200 dilution; Santa Cruz Biotechnology Inc.) was applied to sections to detect protein expression of FBXO32, p-Smad2/3, or Smad4, respectively. For a negative control, the primary antibody was replaced with mouse immunoglobulin G. Slides with positive staining for FBXO32, p-Smad2/3, or Smad4 were used as positive controls.

Immunohistochemical staining was evaluated according to a previously reported scoring method.[15] The scores accounted for both the areas stained and the intensity of the staining. All slides were examined and scored by 3 independent observers who were blinded to the clinical data. All slides were reviewed concurrently by 3 experienced pathologists.

Methylation Analysis of FBXO32

Total DNA was isolated from 5-Aza-dC-treated and untreated cells using DNAzol (Invitrogen) according to the manufacturer's recommendation. Genomic DNA from tumor sections and corresponding normal tissues was isolated from flash-frozen tissues using a simplified Proteinase K digestion method. To examine the DNA methylation patterns, genomic DNA was treated with sodium bisulfite as described previously.[16] The methylation status of FBXO32 was then determined by bisulfite genomic sequencing and the bisulfite conversion-specific and methylation-specific PCR (BS-MSP) method, as described previously.[9, 17] For bisulfite genomic sequencing, primers were designed to recognize sodium bisulfite-converted DNA encompassing CpG islands 1 and 2 within the human FBXO32 gene CpG island of the promoter (from −624 to −106 bp). For BS-MSP, the promoter region (from −475 to −288 bp) of the FBXO32 gene was analyzed. The primers and reaction conditions are provided in Supporting Table 2 (see online supporting information). Genomic DNA, which was methylated in vitro using CpG methyltransferase (SssI) according to the manufacturer's directions (New England BioLabs, Beverly, Mass), was used as a positive control, and a water blank was used as a negative control. Reactions were performed in duplicate for each of the samples.

Statistical Analysis

Statistical analyses were performed using the SPSS statistical software package (version 13.0; SPSS Inc., Chicago, Ill). The results of RT-PCR, real-time RT-PCR, and Western blot analyses were expressed as the mean ± standard error. The means were compared using the Student t test. The chi-square test was used to compare the methylation and protein frequency. Correlations between variables were tested using Spearman correlation analysis. Kaplan-Meier survival curves were constructed, and the log-rank test or the Breslow test was used as needed for the univariate comparison of FBXO32, p-Smad2/3, and Smad4 expression and FBXO32 methylation status. A Cox multivariate test applied in a forward, stepwise manner was used to adjust for potentially confounding variables (eg disease stage and family history of UGIC) and to evaluate the role of FBXO32 as an independent predictor of patient prognosis. Two-sided tests were used to determine significance, and P values < .05 were regarded as statistically significant for all statistical tests.

RESULTS

Frequent Silencing of FBXO32 and Up-Regulation of the Gene by 5-Aza-dC Treatment in Esophageal Cancer Cell Lines

mRNA expression and protein expression of the FBXO32 gene were examined in 4 esophageal cancer cell lines to determine the expression status of FBXO32 in esophageal cancer. Figure 1A,B indicates that FBXO32 mRNA expression was remarkably reduced or silenced in the TE13, T.Tn, and Yes-2 cell lines. This finding was confirmed by the results from Western blot analysis and immunocytochemistry (Fig. 1C,D). However, treatment with 5-Aza-dC increased the level of mRNA and the expression of FBXO32 in the TE13, T.Tn, and Yes-2 cell lines, consistent with the notion that promoter methylation does repress the expression of FBXO32 in esophageal cancer cell lines.

Figure 1.

F-box protein 32 (FBXO32) expression is observed in 4 human esophageal cancer cell lines (TE1, TE13, T.Tn, and Yes-2). (A) FBXO32 messenger RNA (mRNA) expression is observed in esophageal cancer cell lines treated or untreated with 5-aza-2′-deoxycytidine (5-Aza-dC). GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase; −, negative; +, positive. (B) Relative expression of FBXO32 mRNA is illustrated in cancer cell lines treated or untreated with 5-Aza-dC. Asterisks indicate P < .05. (C) Protein expression of FBXO32, phosphorylated SMAD family members 2 and 3 (p-Smad2/3), and SMAD family member 4 (Smad4) is observed in cancer cell lines treated or untreated with 5-Aza-dC. (D) FBXO32 expression also was examined by immunocytochemistry in the same 4 esophageal cancer cell lines treated or untreated with 5-Aza-dC.

Up-Regulation of FBXO32 by 5-Aza-dC, or TSA, or CBP Treatment in the TE13 Cell Line

Treatment with 5-Aza-dC, or TSA, or CBP resulted in the up-regulation of FBXO32 in TE13 cells, and the combination of 5-Aza-dC and TSA resulted in significant up-regulation of FBXO32 in these cells, as illustrated in Figure 2A-C. Increased mRNA and protein expression of FBXO32 also was observed in stable transfected TE13 cells.

Figure 2.

The expression of F-box protein 32 (FBXO32) and its effects on the TE13 cell line are observed after treatment with 5-aza-2′-deoxycytidine (5-Aza-dC), trichostatin A (TSA), 5-Aza-dC/TSA, empty vector (pcDNA3.1-EV), stable transfected FBXO32 (pcDNA3.1-FBXO32), or calmodulin-binding protein (CBP). (A) Reverse transcriptase-polymerase chain reaction (RT-PCR) changes in FBXO32 gene expression are observed in treated and untreated TE13 cells. GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase. (B) Quantitative real-time RT-PCR results are illustrated. mRNA indicates messenger RNA. Asterisks indicate P < .05 compared with untreated TE13 cells. (C) FBXO32, phosphorylated SMAD family members 2 and 3 (p-Smad2/3), and SMAD family member 4 (Smad4) protein expression is observed in treated and untreated TE13 cells. (D-F) A real-time chromatin immunoprecipitation microarray was used to determine the enrichment of (D) dimethylated histone H3 at lysine 4 (H3K4me2), (E) dimethylated histone H3 at lysine 9 (H3K9me2), and (F) acetylated histone H3 at lysine 9 (H3K9ac) within the FBXO32 promoter in TE13 cells. Asterisks indicate P < .05 compared with untreated TE13 cells. (G) Results from a 3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide (MTT) assay are illustrated in the 4 cancer cell lines. Asterisks indicate P < .05 compared with untreated cells. (H) MTT assay results are illustrated in treated and untreated TE13 cells. Asterisks indicate P < .05 compared with untreated TE13 cells.

To determine the potential role of histone modifications in FBXO32 down-regulation in TE13 cells, the presence of active (H3K4me2, H3K9ac) and inactive (H3K9me2) histone modifications at the FBXO32 promoter was further examined by real-time ChIP TE13 cells (Fig. 2D-F). The repressive marker H3K9me2 was enriched more in TE13 cells than the active markers H3K4me2 and H3K9ac. After treatment with 5-Aza-dC, increased H3K4me2 and decreased H3K9me2 were detected. The onset of H3K9ac after TSA treatment also was confirmed.

Inhibition of Proliferation in Esophageal Cancer Cell Lines After Treatment With 5-Aza-dC, or TSA, or CBP

Figure 2G indicates that 5-Aza-dC treatment resulted in the significant inhibition of proliferation in all 4 cell lines. Treatment with 5-Aza-dC, or TSA, or CBP and stable transfection of FBXO32 also resulted in the significant inhibition of proliferation in the TE13 cell line (Fig. 2H).

FBXO32 Induces Apoptosis of an Esophageal Cancer Cell Line

Because previous studies have demonstrated that FBXO32 may play a critical role in regulating apoptosis in breast cancer and ovarian cancer,[8, 9] we further examined the role of FBXO32 in esophageal cancer cell line apoptosis. Results from a fluorescence-activated cell-sorting analysis demonstrated that transfection of FBXO32 enhanced apoptosis in TE13 cells compared with the vector control (apoptotic cell percentage [± standard error]: FBXO32, 10.28 ± 1.03; vector control, 2.39 ± 0.15; P = .012) (Supporting Fig. 1). Significant down-regulation of Survivin and up-regulation of Bax also were observed in FBXO32 stably transfected TE13 cells (Supporting Fig. 2). After siRNA for FBXO32 was transfected in TE1 cells, significantly decreased expression of FBXO32 and inhibition of apoptosis were observed, and these effects could be reversed in part by CBP treatment (Supporting Fig. 3) (see online supporting information).

Figure 3.

This schematic illustrates the structure of F-box protein 32 (FBXO32) cytosine-phosphate-guanine (CpG) islands. (A) Two CpG islands are depicted, and the methylation-specific polymerase chain reaction (MSP) region analyzed is indicated. (B) High-resolution mapping of the methylation status of every CpG site in the FBXO32 promoter was accomplished using bisulfite genomic sequencing (BGS) in 4 esophageal cancer cell lines (TE1, TE13, T.Tn, and Yes-2). A 495-base pair (bp) region spanning the promoter with 50 CpG sites was analyzed. Each CpG site is indicated in the top row as an individual number. The percentage of methylation was determined as the percentage of methylated cytosine from 8 to 10 sequenced colonies. The shading of circles for each CpG site represents the percentage of methylation. (C) Methylation status of the FBXO32 promoter was determined by bisulfite conversion-specific and methylation-specific polymerase chain reaction in various tumor cell lines with or without 5-aza-2′-deoxycytidine (5-Aza-dC) treatment. M indicates methylated; U, unmethylated; −, negative; +, positive. (D) Luciferase activity of the promoter constructs FBXO32-F1 (F1), FBXO32-F2 (F2), and FBXO32-F3 (F3) is illustrated. pGL3-control vector was used as a positive control (POS), and empty pGL3-basic vector was used as a negative control (EV). FBXO32-F3 had the highest relative luciferase activity. Asterisks indicate P < .05 compared with EV. (E) The promoter region from −300 to +200 base pairs was methylated in vitro, cloned into pGL3 vector, and luciferase activity was determined and compared with that of unmethylated FBXO32-F3. In vitro methylation of FBXO32-F3 led to a significant decrease in luciferase activity. The asterisk indicates P < .05.

The Aberrant Promoter Methylation of FBXO32 Induces Silencing of FBXO32 Expression

The MethPrimer program[18] and the CpG Island Searcher[19] were used to determine whether the sequence of FBXO32 contains CpG islands. Two CpG islands were located in FBXO32 promoter and exon 1, as illustrated in Figure 3A. A bisulfite genomic sequencing assay revealed promoter hypermethylation of FBXO32 in the TE13, T.Tn, and Yes-2 cell lines (Fig. 3B). The results were further verified with a BS-MSP assay (Fig. 3C). Full methylation of the FBXO32 promoter was observed in TE13, T.Tn, and Yes-2 cells. After treatment with 5-Aza-dC, demethylation of the FBXO32 promoter was observed in these cells.

In Vitro Methylation of FBXO32 Leads to a Significant Decrease in Luciferase Activity

Three constructs (FBXO32-F1, FBXO32-F2, and FBXO32-F3) were designed for functional characterization of the FBXO32 promoter. FBXO32-F3 had the highest relative luciferase activity, providing a potential explanation for the importance of proximal promoter methylation of FBXO32 in the control of FBXO32 transcription (Fig. 3D). In vitro methylation of FBXO32-F3 led to a 90% decrease in luciferase activity (Fig. 3E), providing direct evidence for the role of methylation in this region.

Decreased Expression and Aberrant Methylation of FBXO32 in Clinical Specimens

The Student t test was used to compare mRNA expression between tumor sections and corresponding normal tissues. FBXO32 mRNA expression in ESCC tumor sections was reduced significantly compared with that in corresponding normal tissues (P = .001) (Fig. 4A,B). The pattern of immunohistochemical staining of FBXO32 was cytoplasmic (Fig. 4C). The chi-square test was used to compare the protein expression of FBXO32 between tumor sections and corresponding normal tissues. Positive protein expression of FBXO32 in tumor sections (41 of 123 samples; 31.1%) was significantly lower than that in corresponding normal tissues (106 of 132 samples; 80.3%; P = .000). FBXO32 protein expression was associated with TNM stage, depth of invasion, distant metastasis or recurrence, and UGIC family history (P < .05) (Table 1).

Figure 4.

Images depict messenger RNA (mRNA) expression, methylation status, and immunohistochemical staining of F-box protein 32 (FBXO32) in tissues. (A) Relative expression of FBXO32 mRNA is illustrated in corresponding normal tissues and tumor tissues. The asterisk indicates P < .05. (B) Blots show reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of FBXO32 in esophageal squamous cell carcinoma (ESCC) tumor tissues. GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase. (C) Immunohistochemical staining of FBXO32, phosphorylated SMAD family members 2 and 3 (p-Smad2/3), and SMAD family member 4 (Smad4) is observed in ESCC tumor tissues and normal tissues, including (a) positive staining of FBXO32 in ESCC tissue, (b) negative staining of FBXO32 in ESCC tissue, (c) positive staining of FBXO32 in normal tissues, (d) positive staining of p-Smad2/3 in ESCC tissue, (e) negative staining of p-Smad2/3 in ESCC tissue, (f) positive staining of p-Smad2/3 in normal tissue, (g) positive staining of Smad4 in ESCC tissue, (h) negative staining of Smad4 in ESCC tissue, and (i) positive staining of Smad4 in normal tissue (original magnification, ×400). (D) Methylation status of the FBXO32 regulatory cytosine-phosphate-guanine (CpG) site was determined with bisulfite conversion-specific and methylation-specific polymerase chain reaction (MSP) analysis in ESCC tumor tissues. U indicates unmethylated; M, methylated. (E) The relative expression of FBXO32 mRNA is illustrated in tumor tissues with different methylation status. Asterisks indicate P < .05 compared with the unmethylated group.

The methylation analysis was performed successfully in all specimens (Fig. 4D). The chi-square test was used to compare the methylation frequency between tumor sections and corresponding normal tissues. The methylation frequency of FBXO32 in tumor tissues (69 of 132 samples; 52.3%) was significantly greater than that in corresponding normal tissues (6 of 132 samples; 4.5%; P = .000). FBXO32 methylation was associated with TNM stage, pathologic differentiation, depth of invasion, and distant metastasis or recurrence (P < .05) (Table 1, Supporting Fig. 4) (see online supporting information).

Association Between FBXO32 Methylation Status and Expression

FBXO32 mRNA expression in ESCC tumor sections that had no detectable FBXO32 protein was significantly reduced compared with ESCC tumor tissues that had detectable FBXO32 protein (P = .002). For ESCC tumor sections in which the promoter of FBXO32 was methylated, FBXO32 mRNA expression was reduced significantly compared with that in ESCC tumor tissues without methylation of the gene (P = .01) (Fig. 4E). The correlation of FBXO32 methylation status and protein expression is provided in Table 2: a close correlation was noted between FBXO32 promoter methylation and loss of protein expression in the FBXO32 gene in ESCC (P = .005; chi-square test). However, immunohistochemical staining also revealed normal staining of FBXO32 in 14 tumor sections that had FBXO32 methylation.

The Role of FBXO32 Expression in Modulation of the TGF-β/Smad Signaling Pathway in ESCC

The protein expression of p-Smad2/3 and Smad4 was further examined in esophageal cancer cell lines and tissues. Figures 1C and 2C reveal decreased Smad4 protein expression in cancer cell lines and 5-Aza-dC or TSA or CBP treatment resulted in up-regulation of Smad4. As shown in Figure 4D, the pattern of immunohistochemical staining of p-Smad2/3 was nuclear staining, and Smad4 was nuclear and cytoplasmic staining. The chi-square test was used to compare the protein expression of p-Smad2/3 or Smad4 between tumor sections and corresponding normal tissues. Positive protein expression of p-Smad2/3 and Smad4 in tumor sections was significantly lower than that in corresponding normal tissues (P < .01). A positive correlation was noted between FBXO32 and protein expression of p-Smad2/3 and Smad4 (P = .000; chi-square test) (Table 2).

Survival Analysis of FBXO32, p-Smad2/3, and Smad4 in ESCC

The patients were followed for a minimum of 1.5 years (range, 1.5-6.5 years); and, for survivors, the median follow-up was 5.5 years. The Kaplan-Meier curves illustrated in Figure 5 indicate that FBXO32 methylation was inversely correlated with the survival of patients with ESCC; whereas FBXO32, p-Smad2/3, and Smad4 expression was positively correlated with survival in patients with ESCC (P < .05; log-rank test). For the patients with ESCC who had tumors with methylated FBXO32, the 5-year overall survival (OS) rate was 23% (median survival, 33 months; P = .001; log-rank test); whereas, for those with ESCC who had unmethylated FBXO32, the 5-year survival rate was 46% (median survival, not reached). For patients who had FBXO32-expressing ESCC tumors, the 5-year survival rate was 44% (median survival, not reached); and, for those who had FBXO32-negative ESCC tumors, the 5-year survival rate was 20% (median survival, 31 months; P = .001; log-rank test) (Supporting Table 3) (see online supporting information). Patients with ESCC who had stage III and IV disease, negative expression of FBXO32, and hypermethylation of FBXO32 were most likely to develop metastatic disease and also had the worse survival. Cox multivariate analysis demonstrated that FBXO32 methylation status, protein expression, TNM stage, and UGIC family history were independently associated with survival in patients with ESCC (Table 3).

Figure 5.

Kaplan-Meier univariate survival analysis according to the expression of F-box protein 32 (FBXO32), phosphorylated SMAD family members 2 and 3 (p-Smad2/3), and SMAD family member 4 (Smad4) in patients with esophageal squamous cell carcinoma (ESCC) revealed (A) a consistent, direct correlation between FBXO32 methylation and poor survival; (B) a direct correlation between negative FBXO32 expression and poor survival; (C) a direct correlation between negative p-Smad2/3 expression and poor survival; and (D) a direct correlation between negative Smad4 expression and poor survival. Poor survival also was observed (E) for patients with stage III and IV ESCC who had FBXO32 methylation and (F) for patients with stage III and IV ESCC who had negative FBXO32 expression.

Table 3. Cox Multivariate Analysis of Survival in Patients With Esophageal Squamous Cell Carcinoma
VariableßSEPOR (95% CI)
  1. Abbreviations: CI, confidence interval; FBXO32, F-box protein 32; LN, lymph node; OR, odds ratio; p-Smad2/3, phosphorylated SMAD family members 2 and 3; SE, standard error; Smad4, SMAD family member 4; TNM, tumor, lymph node, metastasis; UGIC, upper gastrointestinal cancer.

FBXO32 methylation0.6520.298.0291.919 (1.070-3.441)
FBXO32 expression1.0900.421.0102.973 (1.303-6.786)
p-Smad2/3 expression0.6570.435.1311.929 (0.823-4.524)
Smad4 expression0.3530.379.3521.423 (0.677-2.989)
TNM stage1.3060.360.0003.690 (1.823-7.468)
Depth of invasion0.5260.428.2191.692 (0.732-3.914)
LN metastasis1.4490.644.0244.261 (1.206-15.056)
Distant metastasis or recurrence0.6630.283.0191.940 (1.114-3.381)
Family history of UGIC0.6650.262.0111.945 (1.165-3.248)

DISCUSSION

FBXO32 protein, as a member of the muscle-specific F-box protein family, is highly expressed during muscle atrophy.[20] To date, several studies have linked this gene with cancer. FBXO32 was identified as a TGF-β/Smad target gene and was transcriptionally silenced by promoter hypermethylation in breast cancer cells and ovarian cancer cell lines; in addition, the methylation status of FBXO32 may predict survival in ovarian cancer.[8-10] However, the effect of FBXO32 in ESCC has not been previously reported. The current results suggest that aberrant methylation of FBXO32 and histone deacetylation may suppress the expression of FBXO32 mRNA and protein in esophageal cancer cells and further verify that histone H3 Lys-9 methylation, histone H3 Lys-9 acetylation, and H3 Lys-4 methylation are correlated or inversely correlated with FBXO32 expression. These findings are consistent with the proposed hypothesis that combinations of histone modifications at different residues act synergistically or antagonistically to affect gene expression.[21] Recent findings suggest that FBXO32 is a novel apoptosis regulator and is negatively regulated by a prosurvival signal.[7, 8] The current findings that CBP treatment results in the up-regulation of FBXO32 in TE13 and TE1 cells and in a significant down-regulation of Survivin and up-regulation of Bax in FBXO32 stably transfected TE13 cells further confirms that FBXO32 may act as an apoptosis regulator in esophageal cancer.

Recent studies have demonstrated that disruption of an upstream signaling pathway regulator may result in transcriptional repression of a downstream target gene through epigenetic mechanisms.[22, 23] It has been reported that dysregulation of TGF-β/Smad4 signaling leads to epigenetic silencing of its downstream targets, ADAM19 and RunX1T1, in ovarian cancer cells with impaired SMAD4 nuclear translocation.[23, 24] FBXO32 has been identified as a TGF-β/Smad target gene using the ChIP-Chip method, and dysregulation of TGF-β/Smad4 signaling may lead to aberrant DNA methylation of FBXO32 in ovarian cancer.[9, 10] Our current investigation of FBXO32 further supports this hypothesis. We observed decreased expression of FBXO32, p-Smad2/3, and Smad4 and a positive correlation between FBXO32 and p-Smad2/3, Smad4 expression in ESCC. In addition, we observed protein expression of Smad4 in both the cytoplasm and the nucleus, mainly in the nucleus in normal tissues and mainly in the cytoplasm in tumor samples, indicating the blocking of Smad4 nuclear translocation in ESCC. It is well accepted that nuclear proteins are synthesized in the cytoplasm and need to be imported through the nuclear pore complexes into the nucleus.[9] The effect of Smad4 nuclear translocation may have an important role in regulating target gene expression. Further investigations are needed to clarify the exact mechanisms and molecules that regulate the nuclear/cytoplasmic distribution of Smad4. Furthermore, the finding that 5-Aza-dC or TSA treatment can lead to the up-regulation of Smad4 suggests that decreased expression of Smad4 in ESCC may be caused in part by epigenetic mechanisms, as we observed in our previous study.[4]

We also observed a close correlation between FBXO32 promoter methylation and the loss of mRNA and protein expression of the gene in ESCC; however, immunohistochemical staining also produced positive staining for FBXO32 in some tumor samples that had FBXO32 methylation. Gene heterogenic methylation or an allele methylation may be an important reason. It has been demonstrated that DNA methylation, which suppresses gene expression mainly at transcriptional level, and the density of methylated CpG Island are related to the suppressed degree of transcription.[25] The observation that tumor tissues with both positive protein expression and hypermethylation of the gene had incomplete FBXO32 methylation indicates that the extent of promoter methylation of FBXO32 in these tumors was insufficient to suppress transcription of the gene. Some tumor tissues without methylation of FBXO32 demonstrated negative protein expression of the gene, indicating that the mechanisms of FBXO32 inactivation in ESCC can be attributed to several factors other than methylation.

In the current study, FBXO32 hypermethylation and the expression of FBXO32, p-Smad2/3, and Smad4 were correlated significantly and directly with survival in patients with ESCC. In multivariate analysis, the combination of tumor stage, FBXO32, methylation, and FBXO32 expression provided independent, predictive information about ESCC metastasis and poor survival. Patients with hypermethylation and down-expression of FBXO32 who had stage III and IV disease and a positive UGIC family history had the worst 5-year OS rate. Thus, FBXO32 may have the potential to serve as a useful new prognostic marker for ESCC.

In conclusion, the current results suggest that FBXO32 is epigenetically silenced in ESCC tumors with impaired TGF-β/Smad signaling and that FBXO32 promoter hypermethylation may be 1 of the mechanisms for inactivation of FBXO32 in ESCC. FBXO32 may be a functional tumor suppressor and may serve as a prognostic methylation biomarker for ESCC.

FUNDING SUPPORT

This study was supported by grants from the National Natural Science Foundation (no. 81101854).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary