Glycolysis-Related Gene Analyses Indicate That DEPDC1 Promotes the Malignant Progression of Oral Squamous Cell Carcinoma via the WNT/β-Catenin Signaling Pathway

Increasing evidence suggests that aerobic glycolysis is related to the progression of oral squamous cell carcinoma (OSCC). Hence, we focused on glycolysis-related gene sets to screen for potential therapeutic targets for OSCC. The expression profiles of OSCC samples and normal controls were obtained from The Cancer Genome Atlas (TCGA). Then, the differentially expressed gene sets were selected from the official GSEA website following extraction of the differentially expressed core genes (DECGs). Subsequently, we tried to build a risk model on the basis of DECGs to predict the prognosis of OSCC patients via Cox regression analysis. Furthermore, crucial glycolysis-related genes were selected to explore their biological roles in OSCC. Two active glycolysis-related pathways were acquired and 66 DECGs were identified. Univariate Cox regression analysis showed that six genes, including HMMR, STC2, DDIT4, DEPDC1, SLC16A3, and AURKA, might be potential prognostic factors. Subsequently, a risk formula consisting of DEPDC1, DDIT4, and SLC16A3 was established on basis of the six molecules. Furthermore, DEPDC1 was proven to be related to advanced stage cancer and lymph node metastasis. Moreover, functional experiments suggested that DEPDC1 promoted the aerobic glycolysis, migration, and invasion of OSCC via the WNT/β-catenin pathway. The risk score according to glycolysis-related gene expression might be an independent prognostic factor in OSCC. In addition, DEPDC1 was identified as playing a carcinogenic role in OSCC progression, suggesting that DEPDC1 might be a novel biomarker and therapeutic target for OSCC.


Introduction
OSCC is one of the prevalent types of head and neck squamous cell carcinomas. Recently, OSCC has attracted increasing attention owing to its severe impact on patient quality of life. With advances in diagnosis and operative treatment, the prognosis and quality of life have significantly improved. However, for some advanced malignancies, the 5-year overall survival is just under 50% [1]. In addition, lymph node metastasis and local recurrence contribute to the poor prognosis. Studies have indicated that the overall survival of patients with early-stage OSCC could be improved to more than 85% [2]. Nevertheless, there are few early effective biomarkers and therapeutic targets for OSCC. Therefore, it is vital to explore the mechanism of OSCC progression and identify effective biomarkers to improve the overall survival for advanced-stage OSCC patients.
Cancer cells are characterized by infinite proliferation. Metabolism includes a variety of interconnected cellular chemical transformations that provide energy to sustain life [3]. 2

of 15
To meet their abundant energy demands, cancer cells show metabolic plasticity in the reprogramming of metabolic pathways such as glycolysis [4]. In 1956, Warburg indicated that cancer cells performed glycolysis to acquire energy even under aerobic conditions, which is called the Warburg effect or aerobic glycolysis [5]. Aerobic glycolysis leads to an increase in glucose uptake and accumulation of lactate in cancer cells. Meanwhile, it also speeds up the production of adenosine triphosphate (ATP) [6]. Studies have demonstrated that a faster energy supply through aerobic glycolysis contributed to the proliferation of cancer cells, and the acidic microenvironment formed by accumulation of lactate also facilitated cancer invasion and metastasis [7]. At present, many studies have shown that aerobic glycolysis plays a crucial role in a variety of cancers. For instance, the activation of aerobic glycolysis in pancreatic cancer cells was related to their growth [8]. In addition, accelerated glycolysis also promoted the metastasis of breast cancer [9]. In OSCC, previous studies have indicated that lactate dehydrogenase A, which catalyzes the conversion of L-lactate to pyruvate, could promote malignant progression and the epithelial-mesenchymal transition (EMT) by facilitating glycolysis [10]. Meanwhile, activation of aerobic glycolysis induced by protein disulfide isomerase family 6 promoted the progression of OSCC [11]. In addition, aerobic glycolysis was also relevant to the treatment of OSCC with compounds such as Tanshinone IIA [12] and Nobiletin [13]. Therefore, aerobic glycolysis may be significantly associated with tumorigenesis, progression, and drug therapy in OSCC. However, there are few studies which focus on glycolysis-related pathways and genes to explore their prognostic values.
In our study, we focused on glycolysis-related pathways and carried out Cox regression analysis to explore their prognostic value. As a result, the glycolysis-related biomarker DEP domain containing 1 (DEPDC1), which is a part of a transcription repressor complex and a newly discovered tumor-associated gene [14], was identified as being associated with the prognosis of OSCC. Moreover, the knockdown of DEPDC1 significantly inhibited migration, invasion, glucose uptake and accumulation of lactate. In addition, we demonstrated that DEPDC1 promoted OSCC progression via the WNT/β-catenin signaling pathway. These findings suggest that DEPDC1 may be a viable glycolysis-related biomarker and therapeutic target for OSCC.

Glycolysis-Related Gene Sets in OSCC
The gene expression profiles and corresponding clinical data of 319 OSCC samples and 32 normal controls were obtained from TCGA. In addition, four glycolysis-related pathways including the biocarta-glycolysis-pathway, kegg-glycolysis-gluconeogenesis, hallmark-glycolysis, and reactome-glycolysis were gained from the official GSEA website. Bioinformatics analyses indicated that the gene sets hallmark-glycolysis ( Figure 1A), and reactome-glycolysis ( Figure 1B) were significantly activated in OSCC. To explore the prognostic value of the two gene sets, the core genes were screened out (Table 1). Furthermore, differential analyses indicated that there were 66 DEGs including 19 downregulated and 49 upregulated genes in OSCC ( Figure 1C,D, Tables 2 and 3).

DEPDC1 Was Identified as a Potential Biomarker in OSCC
To further understand the prognostic value of the three risk genes, survival analyses were carried out according to the expression of each gene. The results indicated that patients with high DEPDC1 expression had worse overall survival ( Figure 3A). Meanwhile, high expression of DEPDC1 was related to lymph node metastasis ( Figure 3B) and advanced stage ( Figure 3C). Therefore, DEPDC1 might be an advanced biomarker in OSCC. To further explore the potential mechanism of DEPDC1, we performed GSEA and showed that high expression of DEPDC1 was significantly relevant to the activation of the WNT/β-catenin signaling pathway ( Figure 3D,E). In addition, the carcinogenic roles of DEPDC1 in OSCC were further explored via GSEA ( Figure S3). To further understand the prognostic value of the three risk genes, survival analyses were carried out according to the expression of each gene. The results indicated that patients with high DEPDC1 expression had worse overall survival ( Figure 3A). Meanwhile, high expression of DEPDC1 was related to lymph node metastasis ( Figure 3B) and advanced stage ( Figure 3C). Therefore, DEPDC1 might be an advanced biomarker in OSCC. To further explore the potential mechanism of DEPDC1, we performed GSEA and showed that high expression of DEPDC1 was significantly relevant to the activation of the WNT/βcatenin signaling pathway ( Figure 3D,E). In addition, the carcinogenic roles of DEPDC1 in OSCC were further explored via GSEA ( Figure S3).

Knockdown of DEPDC1 Inhibited the Migration and Invasion of OSCC
To determine the tumorigenic role of DEPDC1 in vitro, siRNAs targeting DEPDC1 and their corresponding negative controls were transfected into SCC9 and SCC15. The efficiency of knockdown was identified by qRT-PCR and Western blot ( Figure S4A,B). As shown in the results, a decrease in DEPDC1 expression inhibited OSCC migration and invasion ( Figure 4A,B). Meanwhile, knockdown of DEPDC1 decreased the protein expression level of β-catenin and N-cadherin and enhanced the expression of E-cadherin ( Figure  4C).

Knockdown of DEPDC1 Inhibited the Migration and Invasion of OSCC
To determine the tumorigenic role of DEPDC1 in vitro, siRNAs targeting DEPDC1 and their corresponding negative controls were transfected into SCC9 and SCC15. The efficiency of knockdown was identified by qRT-PCR and Western blot ( Figure S4A,B). As shown in the results, a decrease in DEPDC1 expression inhibited OSCC migration and invasion ( Figure 4A,B). Meanwhile, knockdown of DEPDC1 decreased the protein expression level of β-catenin and N-cadherin and enhanced the expression of E-cadherin ( Figure 4C). knockdown of DEPDC1. Significant differences were considered at p < 0.05 *; p < 0.01 **; p < 0.001 ***; and p < 0.0001 ****.

DEPDC1 Promoted the Progression of OSCC via the WNT/β-Catenin Pathway
GSEA showed that high DEPDC1 expression was related to activation of the WNT/βcatenin pathway ( Figure 3D, E). To further determine the effect of DEPDC1 in OSCC, the overexpression vectors of DEPDC1 and their controls were transfected into OSCC cells ( Figure S4C,D), followed by treatment with XAV939 (Selleck Chemicals, Chengdu, China), an inhibitor of the WNT/β-catenin pathway, at a concentration of 10 μM. The knockdown of DEPDC1. Significant differences were considered at p < 0.05 *; p < 0.01 **; p < 0.001 ***; and p < 0.0001 ****.

DEPDC1 Promoted the Progression of OSCC via the WNT/β-Catenin Pathway
GSEA showed that high DEPDC1 expression was related to activation of the WNT/βcatenin pathway ( Figure 3D,E). To further determine the effect of DEPDC1 in OSCC, the overexpression vectors of DEPDC1 and their controls were transfected into OSCC cells ( Figure S4C,D), followed by treatment with XAV939 (Selleck Chemicals, Chengdu, China), an inhibitor of the WNT/β-catenin pathway, at a concentration of 10 µM. The results indicated that overexpression of DEPDC1 promoted the migration and invasion of OSCC. However, the migration and invasion abilities were reversed by XAV939 ( Figure 5A,B). Meanwhile, upregulation of DEPDC1 increased the protein levels of β-catenin and Ncadherin and decreased E-cadherin expression ( Figure 5C). The protein levels were also reversed by XAV939. results indicated that overexpression of DEPDC1 promoted the migration and invasion of OSCC. However, the migration and invasion abilities were reversed by XAV939 ( Figure  5A,B). Meanwhile, upregulation of DEPDC1 increased the protein levels of β-catenin and N-cadherin and decreased E-cadherin expression ( Figure 5C). The protein levels were also reversed by XAV939.

DEPDC1 Is Related to Glucose Uptake and Accumulation of Lactate in OSCC
To explore whether DEPDC1 influences aerobic glycolysis, we detected the glucose uptake and lactate production in OSCC cells after knockdown or overexpression of DE-PDC1. As shown in the figure, knockdown of DEPDC1 decreased the glucose uptake and accumulation of lactate ( Figure 6A). This phenomenon was reversed after overexpression of DEPDC1. Nevertheless, the activation of aerobic glycolysis by upregulated DEPDC1 was abolished following treatment with XAV939 ( Figure 6B). In addition, knockdown of DEPDC1 decreased the expression of glucose transporter 1 (Glut1), a major glucose transporter, while overexpression of DEPDC1 enhanced the Glut1 protein level, which was also abolished by XAV939 ( Figure 6C,D).

DEPDC1 Is Related to Glucose Uptake and Accumulation of Lactate in OSCC
To explore whether DEPDC1 influences aerobic glycolysis, we detected the glucose uptake and lactate production in OSCC cells after knockdown or overexpression of DEPDC1. As shown in the figure, knockdown of DEPDC1 decreased the glucose uptake and accumulation of lactate ( Figure 6A). This phenomenon was reversed after overexpression of DEPDC1. Nevertheless, the activation of aerobic glycolysis by upregulated DEPDC1 was abolished following treatment with XAV939 ( Figure 6B). In addition, knockdown of DEPDC1 decreased the expression of glucose transporter 1 (Glut1), a major glucose transporter, while overexpression of DEPDC1 enhanced the Glut1 protein level, which was also abolished by XAV939 ( Figure 6C,D).

Discussion
Generally, cells depend on mitochondrial oxidative phosphorylation to produce energy for cellular processes. However, cancer cells always demand large amounts of energy for their fast proliferation and metastasis [15]. Therefore, cancer cells reprogram metabolic

Discussion
Generally, cells depend on mitochondrial oxidative phosphorylation to produce energy for cellular processes. However, cancer cells always demand large amounts of energy for their fast proliferation and metastasis [15]. Therefore, cancer cells reprogram metabolic pathways including aerobic glycolysis, glutamine metabolism, and fatty acid metabolism [4]. Among them, aerobic glycolysis, also called the "Warburg effect", is a common method of metabolic reprogramming [7]. This process leads to an increase in glucose uptake, accumulation of lactate and faster production of ATP, which plays a crucial role in tumorigenesis and tumor progression [6]. In addition, glycolysis helps cancer cells survive under cellular stress and accelerated aerobic glycolysis is related to the treatment resistance of cancers [16]. Hence, these mechanisms need to be further explored in OSCC, which might be helpful for improving prognosis.
Owing to the lack of effective early biomarkers, the 5-year overall survival of OSCC is still poor. Recently, a number of studies have indicated that aerobic glycolysis is associated with the proliferation [17], invasion [18], metastasis [19] and EMT [20] of OSCC. Therefore, it may be appropriate to explore glycolysis-related pathways to screen for potential biomarkers of OSCC. In our study, two active glycolysis-related pathways were filtered following extraction of 255 core genes and 66 DEGs. The hallmark-glycolysis gene set includes genes encoding proteins involved in glycolysis and gluconeogenesis, while reactome-glycolysis is related to glycolysis and canonical pathways. According to the DEGs from two gene sets, a Cox risk model including DDIT4, DEPDC1, and SLC16A3 was constructed, which predicted the prognosis of OSCC patients accurately. In addition, the risk score might be an independent prognostic factor, indicating that the three genes might play a crucial carcinogenic role in the progression of OSCC. DDIT4, located in the cytosol, is related to the response to hypoxia, which plays a vital role in aerobic glycolysis in the tumor microenvironment. Recently, Han et al. also showed that DDIT4 could predict the prognosis of OSCC patients [21]. In addition, DDIT4 is known to participate in a variety of cancers. In colorectal cancer, DDIT4 has been identified as an advanced stage and metastasis biomarker [22]. Overexpression of DDIT4 could accelerate the proliferation and tumorigenesis of gastric cancer. Generally, inhibition of DDIT4 inactivated P53 signaling pathways and mitogen-activated protein kinase (MAPK) activity, which was associated with drug resistance to 5-fluorouracil [23]. In addition, DDIT4 expression could be induced by nicotinamide nucleotide transhydrogenase antisense RNA 1 (NNT-AS1)/miR-496 axis in prostate cancer and played a carcinogenic role in proliferation and migration [24]. Similarly, SLC16A3, also known as monocarboxylate transporter 3 (MCT3) or MCT4, was associated with lactate and pyruvate transport across plasma membranes, which played a crucial role in diverse malignant tumors. For instance, SLC16A3, mediated by LINC00035/CCAAT enhancer binding protein beta (CEBPB), enhanced the glycolysis and promoted the development of ovarian cancer [25]. Levels of aberrant N6−methylation of adenosine (m6A) at SLC16A3 were also relevant to the efficacy of immunotherapy in melanoma [26]. In the present study, DEPDC1 was identified as a glycolysis-related biomarker via integrated bioinformatics. DEPDC1 is a newly identified cancer-associated gene that is part of a transcription repressor complex and related to cellular processes such as the cell cycle, transcription, mitosis and apoptosis [27]. Recently, DEPDC1 was identified as participating in tumorigenesis and cancer progression. Guo et al. indicated that DEPDC1 was overexpressed in OSCC tissues and related to overall survival [28]. They also indicated that DEPDC1 promoted the proliferation of OSCC via inhibiting cytochrome P450 family 27 subfamily B member 1 (CYP27B1) expression [28]. In addition, Qiu et al. showed that DEPDC1 could facilitate development and metastasis of OSCC [29]. In our study, we demonstrated that DEPDC1 might target the WNT/β−catenin signaling pathway, and then promoted the migration, invasion, and aerobic glycolysis of OSCC. Geng et al. also demonstrated that DEPDC1 could promote nephroblastoma progression through the WNT/β-catenin signaling pathway [30]. Moreover, high DEPDC1 expression was also shown in advanced stage cancer and lymph nodes metastasis, suggesting that DEPDC1 might be an advanced biomarker in OSCC. Unfortunately, we failed to validate these data in real patient samples. Generally, there may be a small difference between the database and real patients. Therefore, a further study should be done to clarify their role in OSCC. In addition, DEPDC1 is known to be associated with various biological processes of other cancers. For instance, DEPDC1 promoted cancer metastasis and differentiation via accelerating the cell cycle from G1 to S phase in gastric cancer [31]. In hepatocellular carcinoma, DEPDC1, as a metabolic gene related to glycolysis, has been validated as a contributor to the reconstruction of the tumor microenvironment [32]. In addition, high DEPDC1 expression mediated by LincRNA regulator of reprogramming (Linc-ROR) accelerated the development of hepatocellular carcinoma and angiogenesis [33]. Meanwhile, DEPDC1 also promoted chemotherapy resistance [34]. These results suggest that DEPDC1 might be a potential therapeutic target.
In our study, GSEA showed that DEPDC1 was relevant to various signaling pathways in OSCC. Recently, Zhao et al. demonstrated that DEPDC1 overexpression led to a significant promotion of proliferation by regulating the cell cycle in breast cancer, and that DEPDC1 was associated with a high activation of the PI3K-AKT-mTOR signaling pathway [35]. In addition, DEPDC1 could interact with E2F1 and promote its transcriptional activity in prostate cancer, resulting in the activation of the E2F signaling pathway to regulate the G1/S phase cell cycle transition and then increase cell proliferation [36]. In nasopharyngeal carcinoma, DEPDC1 knockdown downregulated various downstream targets, such as c−Myc, BCL2, MMP2 and MMP9, which participated in proliferation, tumorigenesis, and metastasis [37], suggesting that DEPDC1 might be a vital factor in these biological processes. In OSCC, higher activation of the PI3K-AKT-mTOR pathway was reported to accelerate progression by increasing aerobic glycolysis [38]. Similarly, the E2F signaling pathway was also highly active in OSCC tissues, and this high activation drove radioresistance in OSCC patients, resulting in worse recurrence-free survival of radiotherapy patients [39]. Moreover, many studies suggest that adipogenesis [40], fatty acid metabolism [41], the MYC pathway [42], and the MTORC1 pathway [43] play a crucial role in the initiation and progression of OSCC. In conclusion, all of these pathways were remarkably relevant to OSCC initiation, progression, and therapy resistance. DEPDC1 was identified as activating these pathways through GSEA. Therefore, DEPDC1 is expected to be a novel biomarker and therapeutic target in OSCC.

Data Acquisition from TCGA
TCGA is a landmark cancer genomics program which contains gene expression profiles and clinical data for 33 cancers [44]. All of the OSCC data were downloaded from the TCGA database. A total of 351 samples including 319 OSCC samples and 32 normal controls were obtained. Two OSCC samples were excluded because of insufficient clinical data.

Gene Set Enrichment Analysis (GSEA)
A total of four glycolysis-related pathways (biocarta-glycolysis-pathway, keggglycolysis-gluconeogenesis, hallmark-glycolysis, reactome-glycolysis) were obtained from the GSEA website (http://www.gseamsigdb.org/gsea/, accessed on 30 September 2021). Subsequently, GSEA was used to screen the activated glycolysis-related gene sets in OSCC. Furthermore, the glycolysis-related genes were extracted from the activated glycolysisrelated gene sets and then we carried out differentially expressed analysis. In addition, GSEA was also used to investigate the potential pathways which were activated by DEPDC1 in OSCC.

Identification of Potential Prognostic Biomarker
Based on the differentially expressed genes (DEGs) from the glycolysis−related gene sets, univariate and multivariate Cox regression analyses were used to construct the Cox risk model and screen potential biomarkers to predict the prognosis of OSCC patients. Furthermore, the risk score of each patient was calculated on the basis of gene expression profiles, and independent prognosis analyses were carried out based on clinical parameters and risk score.

Survival Analysis
According to the risk score of each patient, OSCC samples were divided into high and low risk groups. Combined with the survival data in TCGA, survival analyses with the log rank p test were applied to verify the accuracy of the Cox model in R software with the survival package (Version R-4.2.1, Chengdu, China).

Cell Culture
The human OSCC cell lines SCC9 and SCC15 were purchased from ATCC. Cells were cultured in DMEM (Gibco, Cat#11995500TB, Chengdu, China) with 10% fetal bovine serum (ExCell Bio, Inc., Chengdu, China) and incubated at 37 • C with 5% CO 2 . The medium of all cells was exchanged about every 1-2 days according to the cell density.

Cell Transfection
The small interfering RNAs (siRNAs) and DEPDC1 expression vectors were designed and synthesized in TINGKE (Cheng, China). According to the lipofectamine 3000 protocol (Invitrogen, Cat# L3000-015, Chengdu, China), cells were seeded in 6-well or 12-well plates for 12-24 h at a density of nearly 70-90%, and approximately 2500 ng siRNAs or DEPDC1 overexpression vectors were transfected into OSCC cells. The transfection system would last for 8-12 h, after which the medium was exchanged with fresh medium. After incubation for 2-4 days, RNA and protein were collected for analysis. The siRNA sequence of DEPDC1 was as follows: sense 5 -3 : CGCAGCCCUCUUGCUAUUTT, antisense 5 -3 : AAUAGCAUAGAGGGCUGCGTT.

Transwell Assay
The protocol for the transwell assay has been described previously [46]. Briefly, 600 µL DMEM with 10% FBS was placed into the bottom wells, and about 5 × 10 4 cells resuspended in 200 µL serum-free medium were put in the upper chambers. In addition, the invasion assay was performed with 100 µL 1:8 diluted matrigel in the upper chamber. The device was cultured at 37 • C with 5% CO 2 for 24-72 h. Finally, the cells in the upper chamber were removed and the migration and invasion cells were stained with crystal violet.

Glucose Uptake and Lactate Detection
The cells of glucose uptake and lactate were detected using a detection kit (Cat#ADS-W-TDX002, ADS-W-T009-96, Chengdu, China) following the manufacturer's instructions. Briefly, the same amounts of cells, about 5 × 10 6 cells, were collected to obtain intracellular glucose and lactate. Cells were broken up by cell ultrasound equipment at 200 W for 3 s, with a 10 s pause between each cycle, for a total of 30 times, and then centrifuged at 12,000 rpm and 4 • C for 10 min to collect the supernatant. the absorbance of the supernatant at 450 nm (lactic acid production) and 520 nm (glucose uptake) were determined by a microplate reader. Higher absorbance meant a higher level of glucose uptake and accumulation of lactate.

Western Blotting
The Western blot methods have been elaborated upon [46]. In short, the proteins of the OSCC cell lines were extracted, separated by polyacrylamide gel electrophoresis (PAGE), and transferred to PVDF membranes which were then sealed in 5% skim milk for 1 h at room temperature. Subsequently, the membranes were incubated overnight with primary antibodies at 4 • C and second antibodies for 1 h at room temperature. Finally, ECL was used to analyze the protein level. The antibody information was as follows: GAPDH (Ray

Statiscal Analysis
SPSS software (IBM, Version 23.0, Armonk, NY, USA) was used to perform statistical analysis. The difference between two groups was investigated with a Student's t-test, and the difference in one factor among multiple groups was explored with a one-way ANOVA. Results were expressed as the mean ± SD (standard deviation). Significant differences were considered at p < 0.05 *; p < 0.01 **; p < 0.001 ***; and p < 0.0001 ****.

Conclusions
A total of two glycolysis-related gene sets with aberrant activity in OSCC were identified following extraction of the core genes. Subsequently, we tried to construct a Cox model which could predict the prognosis of OSCC patients more accurately, and found that the risk score according DEPDC1, DDIT4 and SLC16A3 expression profiles might be an independent prognostic factor, indicating that these three genes might play a crucial role in the prognosis of OSCC. Moreover, DEPDC1 was identified as accelerating aerobic glycolysis, migration, and invasion via the WNT/β-catenin signaling pathway in vitro. Meanwhile, DEPDC1 was reported to be relevant to various cancer associated pathways. Our findings, along with previous reports, suggest that DEPDC1 might be a promising novel biomarker and therapeutic target in OSCC.  Data Availability Statement: All the data presented in this study are available on request from the corresponding author. Publicly available data were obtained from official TCGA (https://portal.gdc. cancer.gov/, accessed on 30 September 2021) and GSEA websites (https://www.gsea-msigdb.org/ gsea/index.jsp, accessed on 30 September 2021).