miRNA-Mediated Low Expression of EPHX3 Is Associated with Poor Prognosis and Tumor Immune Infiltration in Head and Neck Squamous Cell Carcinomas

The aim of this study was to explore the regulatory role of epoxide hydrolase 3 (EPHX3) in head and neck squamous cell carcinoma (HNSCC) and to analyze its bioinformatic function, as well as, to screen and predict the miRNAs that can regulate EPHX3 expression in HNSCC. We examined the expression profile and prognostic potential of EPHX3 in TCGA and GTEX databases and performed functional enrichment analysis of EPHX3 using string database. Subsequently, we analyzed the regulatory role of miRNAs on EPHX3, including expression analysis, correlation analysis, and survival analysis. In addition, we also used TIMER to investigate the relationship among EPHX3 expression level, immune checkpoints, and immune infiltration in HNSCC. The results of data analysis after TGCA showed that EPHX3 is a key regulator of tumorigenesis in 13 cancers and can be used as a marker of poor prognosis in HNSCC patients. Bioinformatics analysis revealed that miR-4713-3p is a key miRNA of EPHX3 in HNSCC. Together, our findings indicate that EPHX3 exerts its anticancer effects by suppressing tumor immune checkpoint expression and immune cell infiltration. Overall, our data uncovered miRNA-mediated EPHX3 downregulation as a contributor to poor HNSCC prognosis and reduced tumor immune infiltration.


Introduction
Over the past 2020, the number of newly reported head and neck cancer cases worldwide has been approximately 740000 [1]. Head and neck squamous cell carcinoma (HNSCC) is the commonest cancer type affecting the head and neck region and may originate from the mucosal epithelium of the oral cavity, pharynx, or larynx [2]. It is the 7th most prevalent cancer worldwide at a mortality rate of 40-50% [3]. HNSCC is mainly treated using surgery, radiation, chemotherapy, or a combination of these. However, these treatments are often unsuccessful and significantly lower the patient's quality of life. In particular, salvage surgery is lengthy and mutilating for the patient [4]. Thus, effective HNSCC therapeutic targets or prognostic biomarkers are urgently needed.
Epoxide hydrolases are a small family of α/β hydrolase fold enzymes that break down chemically reactive xenobiotic epoxides and process endogenous epoxides that act as signaling molecules. EPHX3 is an epoxide hydrolase that contains a highly conserved 16-amino acid motif and is abundantly expressed in the proximal digestive system, bone marrow, lymphoid tissues, and skin [5]. EPHX3 expression has been associated with increased risk of some malignancies, such as gastric cancer [6], melanoma [7], and prostate cancer [8]. Hypermethylation in the promoter region was linked to a lower transcription rate. Recent findings that epoxyeicosatrienoic acids (EETs) can reawaken latent tumors and promote metastasis, presumably through their proangiogenic characteristics [9,10], imply that EPHX3 might have anticancer functions. However, there are no studies on the role of EPHX3 in HNSCC.
Here, we investigated EPHX3 expression levels and association with survival in various human cancers. We then subjected EPHX3 to functional enrichment analysis, followed by assessment of its regulatory miRNAs in HNSCC. Finally, we investigated the association between EPHX3 expression and immune infiltration, immune cell biomarkers, and immune checkpoints in HNSCC. Our data show that miRNA-mediated EPHX3 suppression correlates with poor prognosis and immune cell infiltration in HNSCC patients.

Materials and Methods
2.1. Data Sources and Analyses. About 33 human cancer type information data are contained within The Cancer Genome Atlas (TCGA) database in the Genomic Data Sharing (GDC) data portal [11]. We downloaded these tumor RNA SEQ data and matched normal tissue samples.
Using the R package (beanplot) to analyze the different expression of EPHX3 in pan-cancer, statistical analyses were done on R 4.0.3. Unless otherwise specified, only two groups of data were identified by rank sum test. p < 0:05 indicated statistically significant differences.

Protein-Protein Interaction (PPI) Network Functional
Enrichment Analysis. The interactions between proteins constitute a major component of cellular biochemical reaction networks. The STRING database (https://string-db.org/) is a database that searches online for interaction relationships between known proteins [12]. The 50 functional genes most closely related to EPHX3 were searched in STRING's "Multiple Proteins" module, requiring a minimum interaction score of 0.4 and imported into Cytoscape to construct a PPI network. Finally, we used the STRING database for functional enrichment analysis of genes associated with EPHX3 with p < 0:05 indicating statistical significance.
2.3. Analysis of the GEPIA Database. GEPIA (http://gepia .cancer-pku.cn/) is a newly developed web server for tumor/normal differential expression analysis and interactive analysis [13]. Analysis of the correlation between EPHX3 expression and survival in different cancer types was done using GEPIA, combining overall survival (OS) and disease-free survival (RFS). In addition, we also evaluated the correlation of EPHX3 expression with immune checkpoints in HNSCC utilizing the GEPIA database.
2.4. Prediction of miRNA Candidates. miRNAs are widely involved in the negative regulation of target genes. Here, we used TargetScan, TarBase, MIRDB, and miRmap to identify miRNAs that might modulate EPHX3 function and retained the ones that were in common between the different tools for subsequent analysis. TargetScan (http://www .targetscan.org/) looks for conserved 8mer, 7mer, and 6mer sites that match the seed region of each miRNA to predict biological targets for miRNAs [14]. TarBase (http://www .miRNA.gr/tarbase) is a database dedicated to indexing experimentally validated miRNA targets [15]. MiRDB (http://www.mirdb.org/) is a web-based database that predicts miRNA targets and provides functional annotations. MirTarget, a bioinformatics tool that was built by examining thousands of miRNA-target interactions from highthroughput sequencing experiments, predicted all of the targets in miRDB [16]. MiRmap (https://mirmap.ezlab.org/ app/) examines feature correlations and compares their prediction potential by utilizing high-throughput experimental data from immunopurification, transcriptomics, proteomics, and polyribosome isolation experiments [17]. These predicted miRNAs were considered as EPHX3 candidate miRNAs.
2.5. Exploration of the starBase Database. The starBase (https://starbase.sysu.edu.cn/) provides information on the interaction between miRNAs and various RNA molecules [18]. Next, we used starBase to evaluate the correlation of miRNA-EPHX3 expression in HNSCC and the expression levels of miRNA in HNSCC patients and healthy individuals.
2.6. Survival Analysis Using Kaplan-Meier Plotter. The Kaplan-Meier plotter (https://kmplot.com/analysis/) can evaluate the association of 54k genes (mRNA, miRNA, and proteins) and prognosis of 21 cancer types. The tool uses metaanalysis to identify and validate survival biomarkers [19] and can be used for miRNA survival analysis. The back-end database needs manual maintenance. GEO, EGA, and TCGA provided gene expression data as well as relapse-free and overall survival information. To determine the prognostic value of genes, patient samples were divided into two groups based on different quartiles of gene expression. For the purpose of study the relationship between the expression level of miR-4713-3p and the prognosis of patients with HNSCC, we conducted a KM plotter to analyze. Log-rank p < 0:05 was considered statistically significant.
2.7. Immunomodulation Analysis. TIMER (http://timer .cistrome.org/) [20] is a web tool that facilitates analysis of tumor-infiltrating immune cells in great detail. The xcell algorithm was used to investigate the relationship between EPHX3 expression and immune cell infiltration or immune checkpoint expression in HNSC on TIMER. p < 0:05 indicated statistical significance.

Cells and Main
Reagents. Nasopharyngeal carcinoma cell lines (CNE1 and CEN2) and nasopharyngeal epithelial cells (NP69) were kept at the Scientific Laboratory Centre of Hainan Medical University. 10% Gibco fetal bovine serum purchased from Life Technologies. Primary antibodies against EPHX3 and β-actin were acquired from Shanghai Absin Co., and the secondary antibodies labeled with horseradish peroxidase were purchased from Shanghai Biyuntian Co.
2.9. Western Blot to Detect the Expression Level of EPHX3 in Each Group of Cells. The cells were digested with trypsin, centrifuged, and diluted into cell suspension, inoculated into 6-well plates at 1 × 10 5 cells/well, incubated in 5% CO2 of    Journal of Oncology multiple groups, respectively; p < 0:05 was considered statistically significant.

EPHX3 Expression in Cancers across the Board.
To understand the regulatory role of EPHX3 in cancer, we first examined its expression in 33 human cancer types on TCGA and found that EPHX3 is highly expressed in CESC, CHOL, COAD, GBM, LUAD, PCPG, READ, and THCA but poorly expressed in HNSC, BRCA, ESCA, KICH, KIRC, KIRP, and PRAD (Figure 1(a)). Next, analysis of EPHX3 expression in 33 cancers along with GETX data found that BLCA, CESC, CHOL, COAD, DLBC, GBM, LGG, LUAD, LUSC, OV, PAAD, PCPG, READ, STAD, TGCT, and THCA had high EPHX3 expression levels, while ACC, BRCA, ESCA, HNSC, KICH, LICH, PRAD, SKCM, and UCEC had low EPHX3 expression (Figure 1(b)). Studies have shown that patients with low EPHX3 expression in prostate cancer are more likely to relapse in their prognosis [8]. In Oral Squamous Cell Carcinoma, EPHX3 is closely related to lymph node involvement and secondary tumor events [21]. In addition, the downregulation of EPHX3 is related to the occurrence and development of adenoid cystic carcinoma of salivary gland [22]. These data indicate that EPHX3 is a critical regulator of tumorigenesis in the 13 cancers analyzed.

EPHX3's Prognosis Values in Oncogenesis.
Next, we examined the association between EPHX3 expression and survival in the 13 cancers. This analysis revealed that high EPHX3 levels correlate with poor overall survival (OS) in COAD patients, while in HNSC patients high EPHX3 levels correlate with favorable prognosis (Figure 2). However, high EPHX3 expression only correlated significantly with better disease free survival (DFS) in HNSCC patients but not in other cancer types ( Figure 3). Thus, EPHX3 can predict adverse outcomes for HNSCC based on OS and DFS.

EPHX3 Functional Enrichment Analysis.
To better understand the molecular function of EPHX3, the top 50 genes closely related to EPHX3 were identified on STRING database. We also performed protein-protein interaction network construction (PPI) (Figure 4(a)), gene ontology (GO) (Figure 4(b)), and Kyoto Encyclopedia of Genes and Genomes (KEGG) (Figure 4(c)) enrichment analysis. This   Journal of Oncology analysis showed that EPHX3 has a regulatory role in expression mainly through the spliceosome of mRNA.

Prediction of Anti-EPHX3
miRNAs. An analysis of whether EPHX3 is controlled by miRNA identified 14 miR-NAs that might target EPHX3 ( Figure 5(a)). miRNA expression generally correlates negatively with target gene expression. Our analysis showed that in HNSCC, EPHX3 significantly and inversely correlates with miR-4713-3p expression ( Figure 5(b)). Analysis of the expression of miR-4713-3p HNSCC, as well as its prognostic value, revealed that it is significantly elevated in HNSCC, and that its upregulation was linked to patient poor prognosis (Figures 5(c) and 5(d)). Together, these data highlight miR-4713-3p as the most plausible regulator of EPHX3 in HNSC.

Immune Cell Infiltration Analysis of EPHX3 and HNSC.
The relationship between EPHX3 and immune cell infiltration was investigated because immune cell plays an essential regulatory function in tumorigenesis. This analysis showed that EPHX3 expression negatively correlates with infiltration by naïve CD8+ T-cells, TH1 CD4+ T-cells, monocyte, macrophages, endothelial cells, and NK T-cells (Figures 6(a)-6(f)).

Correlation Analysis of EPHX3 and HNSCC Immune
Checkpoints. Important immunological checkpoints involved in tumor immune escape include CD274, IL1B, IL1A, PDCD1, PDCD1LG2, and SIRPA. These immune checkpoints are closely associated with HNSCC tumorigenesis (Figure 7(a)). Next, considering the potential tumor suppressor role of EPHX3 in HNSCC, we examined the association between EPHX3 and CD274, IL1B, IL1A,     (Figures 7(a)-7(e)), indicating that in HNSCC, EPHX3 may play an important role in the suppression of immune escape via these checkpoints.

Western Blot for EPHX3 Expression.
The results of Western blotting experiments of each group showed that EPHX3 expression level in NPC cells in the CEN1 and CEN2 groups was lower than that in the NP69 group (Figure 8), suggesting that EPHX3 may play an important role in NPC.

Discussion
Despite current advances in the diagnosis and treatment of HNSCC, typically patients are already advanced at diagnosis, and the 5-year overall survival rate is less than 50% [23]. The reasons for HNSCC mortality include local recurrence, cervical lymphatic metastasis, and treatment failure due to resistance to standard chemotherapy [24]. Thus, effective therapeutic targets or prognostic indicators for HNSCC are urgently needed. EPHX3 is expressed and regulated in various cancers. However, its role in HNSCC is poorly understood.
To determine EPHX3 expression in various cancers, we carried out a pan-cancer analysis of EPHX3 expression on TCGA and GETX datasets. In data analysis and cell experiments, we have confirmed the low expression of EPHX3 in nasopharyngeal carcinoma cells, suggesting that EPHX3 may play an important role in nasopharyngeal carcinoma. Next, analysis of OS and RFS data revealed that EPHX3 can be used as a prognosis marker for HNSCC and that  Journal of Oncology HNSCC patients with low levels of EPHX3 had worse prognosis. Studies on gastric [6] and prostate [25] cancer have shown that decreased EPHX3 expression promotes cancer recurrence while lowering patient survival.
To understand the molecular functional of EPHX3 better, we used STRING analysis to identify the top 50 genes closely associated with EPHX3 and subjected them to GO (BP) and KEGG enrichment analyses. This analyses showed that EPHX3 is mainly involved in the spliceosome pathway to regulate the expression of other genes. The spliceosome is made of five distinct ribonucleoprotein (RNP) subunits and numerous protein cofactors [26,27]. Impaired spliceosome function is implicated in cancer development mainly due to altered splicing of regulatory sequences on oncogenes [28,29] and alterations in the expression of mutant genes [30][31][32]. Splicing requires multiple proteins and RNA   Journal of Oncology interactions and is directed by many trans-acting proteins that are themselves regulated by posttranslational modifications and protein/RNA interactions. This allows manipulation of the spliceosome for therapeutic purposes [33]. However, the spliceosome processes need to be further explored.
Many studies show that immune infiltration affects the efficacy of chemoradiotherapy and immunotherapy, as well as cancer prognosis [39][40][41]. Our data show that in HNSC, EPHX3 negatively correlates with infiltration by various immune cells, including naive CD8+ T cells, TH1 CD4+ T cells, monocytes, macrophages, endothelial cells, and NK T cells. Additionally, EPHX3 had significant negative correlation with these infiltrating immune cells. These findings indicate that EPHX3 might have anticancer effects in HNSCC through regulation of immune cell infiltration.
Successful immunotherapy depends not only on the presence of a sufficient level of immune cell infiltration of the tumor microenvironment but also on the expression of immunological checkpoints [42]. Our analysis of the relationship between EPHX3 and immune checkpoints revealed that low level of EPHX3 was closely associated with CD274, IL1B, IL1A, PDCD1, PDCD1LG2, and    14 Journal of Oncology SIRPA in HNSCC, indicating that targeting EPHX3 may improve the efficacy of immunotherapy in HNSC.

Conclusions
In conclusion, our analyses indicate that poor EPHX3 expression in HNSC contributes to poor patient prognosis. Furthermore, we find that EPHX3 regulates gene expression mainly through spliceosome function. We show that in HNSC, EPHX3 is regulated by miR-4713 ( Figure 9). Our data also show that EPHX3 may exert its anticancer effects by reducing tumor immune cell infiltration and immune checkpoint expression. However, these results need further investigation.