Functional genetic variant in the Kozak sequence of WW domain-containing oxidoreductase (WWOX) gene is associated with oral cancer risk

In Taiwan, oral cancer is the fourth leading cancer in males and is associated with exposure to environmental carcinogens. WW domain-containing oxidoreductase (WWOX), a tumor suppressor gene, is associated with the development of various cancers. We hypothesized that genetic variants of WWOX influence the susceptibility to oral cancer. Five polymorphisms of WWOX gene from 761 male patients with oral cancer and 1199 male cancer-free individuals were genotyped. We observed that individuals carrying the polymorphic allele of WWOX rs11545028 are more susceptible to oral cancer. Furthermore, patients with advanced-stage oral cancer were associated with a higher frequency of WWOX rs11545028 polymorphisms with the variant genotype TT than did patients with the wild-type gene. An additional integrated in silico analysis confirmed that rs11545028 affects WWOX expression, which significantly correlates with tumor expression and subsequently with tumor development and aggressiveness. In conclusion, genetic variants of WWOX contribute to the occurrence of oral cancer, and the findings regarding these biomarkers provided a prediction model for risk assessment.


INTRODUCTION
Oral cancer, a common malignant disease affecting the head and neck region, has a poor prognosis. The incidence of oral cancer is high in Asia, particularly in Taiwan, and it is the fourth common cancer in males [1] . Furthermore, oral cancer was ranked the fifth most common malignancy in Taiwan, annually accounting for more than 2600 deaths in both sexes. Oral squamous cell carcinoma (OSCC) is the most common cancer, accounting for approximately 90% of all oral cancers [2]. Characterizing multiple genetic alterations in OSCC is a critical problem in understanding tumor development and its association with environmental factors including tobacco smoking, alcohol consumption, betel-quid chewing, chronic inflammation, and viral infection [3][4][5][6].
The WW domain-containing oxidoreductase (WWOX) gene, a tumor suppressor gene, is located on chromosome 16q23 and encompasses the common fragile site FRA16D [7]. WWOX, which encodes a 414-amino acid protein, possesses 2 N-terminal WW domains and a high homology domain of the short-chain dehydrogenase/ reductase family [8][9][10]. WWOX is emerging as a tumor suppressor that is also involved in metabolic and

Research Paper
neurological disorders [11], In vivo studies have indicated that the WWOX gene is alternatively knocked out in mice, causing Leydig cell development failure in the testis and affecting normal prostate function [12]. However, several studies have reported a loss or downregulation of the WWOX protein and homozygous deletion within the WWOX locus in multiple malignant neoplasms such as lung cancer, pancreatic adenocarcinoma, oral cancer, ovarian cancer, and renal cell carcinoma [13][14][15][16][17][18][19][20][21].
Growing evidence emphasizes the importance of genetic variations, which induce cancer by affecting the functions of oncogenes and tumor suppressor genes or enzyme metabolism. The expression of certain genes may be affected by single-nucleotide polymorphisms (SNPs), which are the most common types of DNA sequence variation. Moreover, previous studies have reported the effect of WWOX gene polymorphisms on human cancer susceptibility, and they have indicated that genotypingrelated SNPs may efficiently predict the risk of cancers and other diseases [22][23][24]. Highly variable intronic and exonic polymorphisms were observed within WWOX in tumor cell lines [25]. In addition, studies have identified several SNPs in WWOX as potential risk factors for several cancers such as thyroid carcinomas, esophageal adenocarcinoma, pancreatic and ovarian cancer [22,[26][27][28]. Genome-wide scan analysis studies conducted on the rs1079635 which is in intron 7 of WWOX have also reported that this region demonstrated a strong association with prostate cancer susceptibility [29]. Nevertheless, although the effects of WWOX on functional analysis and phenotypic studies are adequately documented, the role of WWOX genetic polymorphism in the association between environmental carcinogens and OSCC and the clinicopathological characteristics of OSCC remain poorly investigated. In this study, we used a case-control study with 2 independent cohorts and analyzed 5 SNPs in WWOX in addition to investigating the associations between the SNPs and environmental factors. We further investigated the association between genetic factors and oral cancer clinicopathological characteristics. Table 2 shows the results of the statistical analysis of demographic characteristics. Significant differences were observed in the distribution of betel-quid chewing (p < 0.001), cigarette smoking (p < 0.001), and alcohol consumption (p < 0.001) between the controls and patients with OSCC. Table 3 shows genotype distributions and associations between oral cancer and WWOX gene polymorphisms. Alleles with the highest distribution frequency for rs11545028, rs12918952, rs3764340, rs73569323, and rs383362 polymorphisms of WWOX in both the controls and patients with OSCC were heterozygous for C/C, heterozygous for G/G, homozygous  Mann-Whitney U test or Fisher's exact test was used between healthy controls and patients with oral cancer. * p value < 0.05 as statistically significant.

Association between WWOX single nucleotide polymorphisms and OSCC
cell differentiation, the distribution frequency of clinical statuses and WWOX genotype frequencies in patients with oral cancer were estimated. Regarding the genotypic frequency of the SNPs, WWOX rs11545028 demonstrated significant associations with clinical pathological variables in patients with OSCC. The results form Table 4 shown that WWOX rs11545028 gene polymorphism is associated with clinical stage (p= 0.030) and lymph node metastasis (p = 0.010), but no difference was observed in tumor size and cell differentiation (Table 4). Table 3: Odds ratio (OR) and 95% confidence interval (CI) of oral cancer associated with WWOX genotypic frequencies.
The odds ratio (OR) with their 95% confidence intervals were estimated by logistic regression models. * p value < 0.05 as statistically significant.

Functional analysis of the WWOX rs11545028 locus
We also investigated whether rs11545028 was associated with the differential expression of WWOX as a preliminary assessment of the putative functional role of the SNP. We obtained human WWOX from the NCBI gene database, selected its respective longest transcript, and defined its promoters as 1-kb upstream to 1-kb downstream of the predicted transcription start sites ( Figure 1A). Moreover, we identified the putative functional role of rs11545028, as indicated by the functional annotations in the ENCODE data. We determined that rs11545028 was situated at a locus with TF binding, histone modification patterns, DNase hypersensitivity, and CpG islands that were characterized as promoters or enhancers in several cell types ( Figure 1B). The effect of rs11545028 may be attributed to the suboptimal Kozak context surrounding the initiation codon of upstream open reading frames of human WWOX ( Figure 1C), which enables the modulation of initiation rates in response to the translational status. In addition, the GTEx database revealed a statistically significant downregulation of WWOX mRNA expression in the whole blood, muscle skeletal and esophagus mucosa of rs11545028-variant genotypes (CT or TT) compared with that of the WT homozygous CC genotype (p = 0.011, p = 0.0016 and p = 0.027, respectively) ( Figure 1D-1E).

Functional analysis of the WWOX rs11545028 locus in clinical sample
To determine the functional effect of the rs11545028 polymorphism on WWOX expression, we generated luciferase reporter vectors with either the rs11545028 C allele or the rs11545028 T allele. We used these vectors for transfection of HSC-3, OECM-1 and SCC-9 oral cancer cells lines. As shown in Figure 2A, the vectors with the rs11545028 T allele had significantly lower luciferase activity compared to the vectors with the rs11545028 C allele among these three cell lines (p<0.05). Furthermore, to realize correlation between the mRNA, protein level of WWOX and rs11545028 polymorphism, quantitative real time-PCR (qPCR) and Immunohistochemical (IHC) staining were used to analyze WWOX mRNA and protein expression in cancer tissue of 34 and 51 OSCC patients, respectively. We found that OSCC patient who carry C/T or T/T of rs11545028 polymorphism have significantly lower mRNA levels of WWOX compare to C/C genotype ( Figure 2B). Furthermore, when WWOX expression was classified into a two-tier grading system of weak (−/1+) ( Figure 2C) and strong (2+/3+) ( Figure  2D) WWOX staining, our analysis shown that specimens with rs11545028 C/C have higher WWOX expression, while specimens with rs11545028 C/T or T/T have lower WWOX expression (p=0.022) ( Figure 2E).
Overall, rs11545028 C to T substitution might affect the translational initiation and reduce WWOX mRNA and protein expression and the risk of oral cancer.

DISCUSSION
Several studies have suggested that chromosome 16q23 contains a tumor suppressor gene involved in multiple tumor types. WWOX was mapped to this region, and the loss of function of WWOX in cancer cells was associated with mucinous histologies and a poor prognosis, suggesting that WWOX suppresses tumor progression [30]. Recent studies have reported that WWOX polymorphism is associated with the susceptibility to several carcinomas including lung, breast, bladder, colorectal, and pancreatic cancers [31,32]. Five SNPs were included in a case-control study with 2 independent cohorts design. One of the SNPs (rs11545028) is located in exon 1 of WWOX. Our data reveal an increased risk of OSCC among patients with the WWOX polymorphic rs11545028 T/T compared with those with homozygous C/C. Only few studies have examined the functional role of rs11545028, and its functional importance has yet to be examined. An association of the risk of OSCC with the location of the analyzed variant is proposed.
In lung cancer, exonic polymorphisms within WWOX were revealed to exhibit a high incidence of the deletion of exon 6-8, which may result from amino acid changes and thus the loss of the tumor suppression function of WWOX. Furthermore, missense polymorphisms of WWOX, including Arg−314→His, Lys−182→Glu, Arg−120→Trp, and Thr−111→Ser, were detected in blood specimens from 15 and 34 patients with ovarian and colorectal cancers, respectively, but not in healthy participants [25]. By contrast, in our study, 2 of the missense SNPs located in exon 6-8 did not confer a risk of OSCC. In contrast to other tumors, the effect of missense polymorphisms in exon 6-8 did not alter the frequency of DNA strand breakage in OSCC, and this factor might be associated with the cancer type. Therefore, we suggested the presence of another major regulating mechanism associated with the downregulation of WWOX expression in OSCC.
Notably, we observed that the nonsense polymorphism rs11545028 C > T located in exon 1 conferred an increased risk of OSCC. Previous studies have reported that rs11545028 (C121T) in the data set indicated no significant difference between each tumor cell line and normal cell lines, even when the frequency of T/T in patients was lower. In this study, we determined whether the genetic variant rs11545028 C > T contributed to oral cancer susceptibility. Functional annotations from the ENCODE data indicate that rs11545028 is located in the region of an open chromatin, which probably corresponds to the promoters and CpG islands of WWOX. Several studies have documented the www.impactjournals.com/oncotarget importance of transcriptional regulation between WWOX polymorphisms and cancer risk [23,33]. Moreover, the common modifications of epigenetic changes in chromatin include DNA methylation, which has been considered a crucial mechanism underlying the inactivation of tumor suppressor genes as well as the loss of heterozygosity and mutation. An abnormal DNA methylation mainly occurs in the promoter region (CpG islands), which is associated with the transcriptional inactivation of tumor suppressor genes during tumor progression. The methylation rate of the WWOX promoter has been reported to be associated with the loss of WWOX expression in breast, lung, bladder, pancreatic, and prostate cancers [14,32,34]. The CpG methylation status in the WWOX promoter region was significantly higher in late-stage epithelial ovarian cancer tissues than in early stage epithelial ovarian cancer tissues [35]. In head and neck squamous cell carcinoma, it has also been reported that WWOX expression was decreased by miR-134 and promoter methylation [36,37]. Liu et al. showed miR-134 expression contributes to head and neck carcinogenesis by targeting the WWOX [36]. Moreover, Ekizoglu et al. reported that decreased WWOX expression in advanced-stage tumor samples or in tumors with OSCC was associated with methylation of the WWOX promoter region [37]. Pimenta et al. also shown that the WWOX gene alteration is an early genetic alteration and may contribute to oral carcinogenesis [38]. Nevertheless, the frequency of the methylation of the WWOX promoter was not explained in our study, and the methylation rate must be examined further. Critical evidence indicates the importance of the methylation of the WWOX promoter. The highest methylation at the CpG site (approximately 60%) was observed in the promoter region (−328 to −41 bp) and exon 1 (−27 to +334 bp) of WWOX [39]. Furthermore, an aberrant methylation of these WWOX regions may occur at the early stage of cancer and, more precisely, at the advanced stage of esophageal squamous cell carcinoma, thus coinciding with the rs11545028 (C121T) region. This observation suggests that WWOX methylation is a critical event in the development of OSCC and that silencing through WWOX methylation is a pivotal mechanism underlying WWOX inactivation.
In a previous study, rs11545028 was predicted to lie within the Kozak translation initiation site, which comprises 6−8 nucleotides surrounding the initiation codon [40]. Studies conducted on the optimal Kozak sequence at positions −3 and +4 have proposed a valuable method for determining gene expression. However, the Kozak sequence has been increasingly demonstrated to be capable of altering translational machinery in response to the regulation of gene expression [41,42]. A recent study revealed that an SNP located at position −1C/T in the Kozak sequence of CD40 was highly meaningful because the CD40 expression levels were significantly higher in −1C/C carriers than in −1C/T and −1T/T carriers [43]. Consistent with this observation, the less consensus Kozak sequence involving the nucleotide T at position −4 may markedly affect protein expression [44]. Figure  1C shows that the WWOX polymorphism rs11545028 at position −5 that involves the original consensus Kozak sequence contains the nucleotide C. Previous studies have demonstrated an association between the C allele of the Kozak polymorphism and gene expression both in vitro and in vivo. In the current study, the sequence containing the nucleotide C at position −5 more closely approximated the Kozak consensus, suggesting that the mRNA with the nucleotide T at 121 was associated with a markedly diminished efficiency. The GTEx database also revealed a significant drop in the WWOX mRNA expression in carriers of a genotype involving the variant T at rs11545028. In addition, we observed a high frequency of the homozygous 121 TT genotype and its combination with the heterozygous WWOX CT in patients, suggesting that changes in the translation initiation rate generally explains the differences in protein expression among the participants. The WWOX expression also suppresses tumor growth and induces cell apoptosis [45]. We observed that rs11545028 was associated with a higher risk of stage III and IV cancers, lymph node metastasis, and the cell differentiation grade. Overall, our findings suggest that the rs11545028 T allele reduced the translation initiation rate, which subsequently reduced the WWOX expression, thus contributing to a more aggressive phenotype in OSCC.
In conclusion, examining the complete medical information and conducting additional bioinformative analyses of a high number of patients provided comprehensive evidence of WWOX polymorphism in OSCC. Our results suggest that the WWOX polymorphic rs11545028 C/T in the suboptimal Kozak context is associated with clinical statuses and susceptibility to OSCC. The coeffects of WWOX polymorphism and environmental carcinogens markedly facilitate OSCC development. Overall, our analyses provide deeper insights into naturally occurring TIS variants. Comprehensive data on such types of variant are required for developing therapeutic approaches that can eventually ameliorate the clinical phenotype in patients harboring the corresponding lesions.

Patient specimens
In 2007-2014, for the case group, we recruited 761 male patients at Chung Shan Medical University Hospital in Taichung and Changhua Christian Hospital in Changhua, Taiwan. For the control group, we randomly chose 1199 male non-cancer individuals from Taiwan Biobank and these control groups had neither selfreported history of cancer of any sites. For both groups, we administered a questionnaire to obtain information on their exposure to betel quid chewing, tobacco use, and alcohol consumption. Medical information of the patients, including TNM clinical staging, primary tumor size, lymph node involvement, and histologic grade, was obtained from their medical records. All participants provided written consent, and the Chung-Shan Medical University Hospital ethics committees approved the research protocol and informed consent was obtained from all subjects (CSMUH No: CS13214-1). All the methods applied in the study were carried out in accordance with the approved guidelines.

DNA extraction
DNA was extracted from buffy coats (white blood cells) using a QIAamp DNA blood mini kits (Qiagen, www.impactjournals.com/oncotarget Valencia, California) as described in detail previously [46]. DNA was dissolved in TE buffer and used as the template in polymerase chain reactions

SNP selection and genotyping
In this study, the selection of 5 well-characterized common polymorphisms from WWOX gene is based on their wide associations with the development of cancer (Table 1, Figure 1A) [23][24][25][26]33]. We included rs11545028 in the 5'UTR region. Rs12918952 and rs3764340, which are located in the exon of WWOX, were selected in this study since these 2 SNPs may result from amino acid changes and thus the loss of the tumor suppression function of WWOX [25]. The allelic discrimination of WWOX rs11545028, rs12918952, rs3764340, rs73569323, and rs383362 polymorphisms were assessed using an ABI StepOne TM Real-Time PCR System (Applied Biosystems, Foster City, CA) and analyzed using SDS v3.0 software (Applied Biosystems, Foster City, CA) as previously described [47].

Construction of luciferase reporter plasmids
A luciferase reporter plasmid encompassing the major allele (C) and minor allele (T) of rs11545028 in the promoter region of the WWOX gene was cloned into the pGL3-Enhancer Luciferase Reporter Vectors (Promega Corp., Madison, WI, USA), according to manufacturer instructions. The vectors were sequenced to confirm the orientation and integrity.

Transient transfections and luciferase assay
HSC-3 cells were purchased by the Japanese Collection of Research Bioresources Cell Bank (JCRB, Shinjuku, Japan) [48]. SCC-9 cells were purchased from and validated by the American Type Culture Collection (ATCC, Manassas, VA, USA). Both cell lines maintained in DMEM/F12 supplemented with 10% FBS, 400 ng/ml hydrocortisone and 0.1 mM non-essential amino acids (NEAA; Life Technologies). OECM-1 cells were obtained from Dr Meng's group where the cell line is originally established and authenticated and maintained in RPMI (Gibco) supplemented with 10% FBS [49]. All the cells were cultured and maintained at 37 °C in a 5% CO 2 and 95% air atmosphere. Each cell was seeded per well in a 24well plate, and each well was transfected with 0.75 μg of the vector DNA containing either the rs11545028 C allele or the rs11545028 T allele by using the Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA), according to manufacturer instructions. Cells were collected 48 h after transfection and analyzed for luciferase activity by using the Luciferase Reporter Assay System (Promega, Madison, WI, USA). All transfections were performed in duplicate and repeated three times.

RNA preparation, TaqMan quantitative real-time PCR
Total RNA was isolated from oral cancer tissues using RNeasy Mini Kit (Qiagen, Valencia, CA, USA). Quantitative real-time PCR analysis was performed using TaqMan one-step PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Total cDNA (2 μg) was added per 9 μl reactions with WWOX or GAPDH primers and TaqMan probes. The WWOX (Hs03044790_m1) and GAPDH (Hs99999905_m1) primers and probes were designed using commercial software (ABI PRISM Sequence Detection System; Applied Biosystems, Foster City, CA, USA) as previously described [50].

Immunohistochemistry
OSCC tissue microarray block slides were deparaffinised, as stated in our previous study [51]. The slides were incubated with 1:200 diluted anti-WWOX antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 60 min at room temperature. After thoroughly washing with PBS, the conventional streptavidinbiotin peroxidase method (LSAB Kit K675; Dako, Copenhagen, Denmark) using 3,3'-diaminobenzidine (DAB) was employed for assessing signal development. Two pathologists blinded to the clinical outcomes semiquantitatively assessed WWOX expression based on the staining intensity; they independently scored sections through light microscopy.

Statistical analysis
Mann-Whitney U-test and Fisher's exact test were used to compare the age differences and demographic characteristic distributions between the controls and patients with oral cancer. The odds ratio and 95% CIs of the association between the genotype frequencies and oral cancer risk and the clinical pathological characteristics were estimated using multiple logistic regression models. p < 0.05 was considered significant. The data were analyzed on SAS statistical software (Version 9.1, 2005; SAS Institute, Cary, NC).

Bioinformatics analysis
We used several semiautomated bioinformatics tools for assessing whether rs11545028 or its related genetic variants were associated with a putative function that might affect patient outcomes. HaploReg [52] v4 and Genotype-Tissue Expression (GTEx) [53] from the Encyclopedia of DNA Elements (ENCODE) [54] project were used for identifying the regulatory potential of candidate functional variants to examine factors of interest such as transcription factor (TF)-chromatin immunoprecipitation signals, DNase peaks, DNase footprints, and predicted DNA sequence motifs for TFs. The GTEx data were used for identifying the associations between the SNPs and whole blood-specific gene expression levels. Moreover, the publicly available cBioPortal for Cancer Genomics [55] and UCSC Cancer Genomics Browser [56] for hepatocellular adenocarcinomas were used for analyzing WWOX expression, DNA methylation, molecular features, and clinical outcomes.

CONFLICTS OF INTEREST
The authors declared no conflict of interest.

Novelty & impact statements
Genetic variants of WWOX contribute to the occurrence of oral cancer, and the findings regarding these biomarkers provided a prediction model for risk assessment.