Keywords

1 Introduction

Oral squamous carcinoma (OSCC) is an alarming public health issue that originated from head and neck malignant tumors [1]. Among the oral malignancies its about 90% cases are detected as OSCC [2]. Recently, the number of oral cancer cases increased in Asia and Europe. In Bangladesh, lip and oral vesicle cancers are 2nd most (11.9%) loading cancer for males and 3rd most (6.5%) loading cancer for females [3]. In India, every 12 people are affected by OSCC from 100,000 people [4]. The average survival of oral cancer is more than five years. Beginning phase (I and II) oral malignant growth might be reparable by medical procedure or radiation treatment alone yet progressed tumors (stage III and IV) are the most part treated by a medical procedure followed by radiation treatment [5]. Patients who are reached cutting edge stage have possible change of survive more than 30 months, even those who are early accomplish total clinical abatement [6]. A basic figure is the absence of prognostic improvement is that a critical extent of cancer called asymptomatic injuries which are not analyzed or treated when they arrive at a high level stage. Early revelation of cancer is the best viable means to decrease death from this disease [7]. Also, most patients are prescribed to have radiotherapy and chemotherapy after medical procedure to diminish the chance of repeat. But clinical protection from chemotherapy actually prompts a helpless forecast of OSCC patients. In this analysis, two sets of data were examined to identify a molecular signature for the treatment of OSCC patients. More importantly, our study identified significant hub-genes which might be responsible for clinical resistance of OSCC and suggested candidate drugs might be a therapeutic target of OSCC.

2 Materials and methods

2.1 Conceptual Framework

To achieve the objective of this research, a logical framework was prepared through the organized steps and the methods were used in the analysis shown in Fig. 1.

Fig. 1.
figure 1

(Source Created by the authors)

Schematic diagram of this study

Full size image

2.2 Data Description

To identify the biomarkers of OSCC GSE111585 and GSE115119 data sets were downloaded from the NCBI with platforms GPL14715 and GPL16955, respectively. GSE111585 dataset containing six samples of SCC9 cells and 42152 genes which were partitioned into normal group and drug resistance group [8]. Dataset GSE115119 has four samples of CAL-27 cells and 42152 genes which were partitioned into normal group and drug-resistant group.

In this study, additionally we used the list of oncogenes and tumor-suppressor genes which gave a better validation of identified DEGs [9, 10].

2.3 Differentially Expressed Gene Identification Methods

In this analysis we used three different statistical methods to identify differentially expressed genes.

  1. (1)

    Analysis Of Variance (ANOVA) is a parametric method used to identify differentially expressed genes [11].

  2. (2)

    Linear Models for Microarray Data (Limma) is a popular method based on Bayesian approach to identifying differentially expressed genes of data arising from microarray experiments [12].

  3. (3)

    Significance Analysis of Microarrays (SAM) is an efficient technique used for analyzing microarray gene expression data. It was established in 2001 by Virginia Tusher et. al. [13]

2.4 Functional Annotation and Pathway Analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) used for finding biological information of biomarker genes. GO-terms are coordinated into three autonomous controlled ontologies: Biological process, Molecular function and Cellular component. KEGG is used for pathway analysis. We used WEB-based gene set analysis toolkit (WebGestalt) for the GO and KEGG enrichment analysis [14].

2.5 Protein-Protein Interaction (PPI) Network

We constructed a PPI network using the STRING that contains multiple proteins information. Then conceptualize the PPI networks using Cytoscape [15]. We selected a confidence score ≥ 0.4 and maximum interactors = 0 for cutoff criteria.

2.6 Reporter Transcription Factors (TFs) and miRNAs

We identified TFs DEGs interactions from HTRIdb [16]. TFs were targeted by the hub genes. For targeting miRNA prediction, we used miRBD to identify a potential miRNA-mRNA interaction network [17]. Then miRNA-mRNA regulatory network was constructed by Cytoscape.

2.7 Survival Analysis of Significance Biomarkers

We completed the prognostic analysis of hub-DEGs using SurvExpress [18]. In Kaplan-Meier plots, we considered log-rank P-value < 0.05 for checking the significant survival output.

2.8 Identification of Candidate Drugs

DGIdb database used to identify desired candidate drugs [19]. In this study, highest degrees of drug-genes network were considered to find candidate drugs for OSCC.

3 Results

3.1 Identification of Differentially Expressed Genes (DEGs)

The methods ANOVA, limma and SAM identified differentially expressed genes (DEGs). The result is shown in the following Table 1. There are 2935 and 2946 common DE genes identified by the three methods for GSE111585 and GSE115119 datasets, respectively (see Fig. 2 A and B). We identified 1380 genes were up-regulated and 1555 genes were down-regulated for cisplatin-safe OSCC from GSE111585 dataset compared with parental OSCC, as like as 2226 genes were up-regulated and 720 genes were down-regulated for cisplatin-safe OSCC from GSE115119 dataset compared with parental OSCC through cluster analysis using cutoff value (|log FC| > 1 and p-value <0.05.

Table 1. Identified DE genes by using ANOVA, limma and SAM methods.
Full size table

The up-regulated genes of those two datasets were compared with the oncogenes and down-regulated genes were compared with the genes for finding DE genes that might be responsible for tumor. Venn diagram showed that common genes of up or down-regulated from the both datasets with oncogenes or tumor-suppressor genes (TSgenes) used to obtain DEGs in cisplatin-safe OSCC cells. Diagram showed that 27 up-regulated DEGs were intersected between up-regulated genes and the list of oncogenes (see Fig. 3A) and 25 down-regulated DEGs were common between down-regulated genes and the list of TS genes (see Fig. 3B).

3.2 Protein-Protein Interaction (PPI) Network for OSCC

To identify central hub protein from differentially expressed genes (DEGs), we construct a PPI network using STRING database and Cytoscape. Figure 4 shows the PPI network for 27 common up regulated DEGs (see Fig. 3A), and hub genes are identified using hub score that has the strong association among other genes. According to high degrees and between closeness, five genes (AR, ETS1, MET, PDGFB and VAV3) are identified.

3.3 GO and KEGG Pathway Analysis for Hub DEGs

According hub-DEGs, we performed GO and KEGG pathway analysis shown in Fig. 5. The hub-DEGs are mainly involved in positive regulation of endothelial cell migration in BP category (see Fig. 5A), basal plasma membrane in the CC category (see Fig. 5B), and superoxide − generating NADPH oxidase activator activity in MF category (see Fig. 5C). KEGG enrichment analysis showed that these hub-DEGs are involved with renal cell carcinoma, Human T − cell leukemia virus 1 infection and so on (see Fig. 5D).

Fig. 2.
figure 2

(Source Created by the authors)

Common DEGs and volcano plot of common DEGs. A & C for dataset GEO111585 and B & D for dataset GEO115119.

Full size image

3.4 Reporter Transcription Factors (TFs) and miRNAs for OSCC

We analyzed significant transcriptional regulatory components for OSCC. We detected TFs are HIF1A, MYC, FOXP3, E2F4, WT1, PURA, ZEB1 and USF2 that are closely targeted as hub DEGs (see Table 2). These hub DEGs are also closely integrated with tumor-related miRNAs. Cytoscape was used to build a miRNA-mRNA network with setting cutoff >70 shown in Fig. 6. Among these miRNAs, some of the miRNAs are connected with multiple genes. These miRNA’s are hsa-miR-377-3p, hsa-miR-499a-5p, hsa-miR-34c-5p, hsa-miR-326, hsa-miR-34a-5p, hsa-miR-449b-5p, hsa-miR-449a, hsa-miR-589-3p which are used for next prognostic biomarker analysis.

Fig. 3.
figure 3

(Source Created by the authors)

Common DEGs with oncogenes and TSgenes.

Full size image
Fig. 4.
figure 4

(Source Created by the authors)

PPI interaction network of hub DEGs and hub DEGs are highlighted from PPI network.

Full size image

3.5 Overall Survival Significance Biomarkers for OSCC

We also performed an overall survival analysis of hub DEGs, SurvExpress is used to estimate the overall survival probabilities of the hub-DEGs. Figure 7 represents overall survival analysis of hub-DEGs which is statistically significant.

3.6 Candidate Drugs Identification Through Drug Repositioning for OSCC

We also obtain novel drugs for hub-DEGs and TFs and also constructed a drug-genes network shown in Fig. 8. Which represents the drugs with has the highest degrees (IMATINIB, BROMOCRIPTINE, NIFEDIPINE, PIRETANIDE and TRIAMTERENE), and connected with multiple genes.

4 Discussion

OSCC is a significant general medical problem with restricted treatment procedures and once again look; foundational drug opposition has disturbed the present circumstance. In this study, significant bioinformatics instruments are utilized to observe 27 up-managed DEGs that are normal with Oncogenes and 25 down-directed DEGs which are normal with TS-genes (see Fig. 3). Cross-over DEGs are utilized to build a PPI organization (see Fig. 4). PPI network recognized as AR, ETS1, MET, PDGFB and VAV3 are the center qualities those are likewise measurably huge and statistically significant (see Fig. 7). AR is related with oral disease [20].

PDGFB is associated with pancreatic cancer risk [21]. MET is associated with the progression of OSCC [22]. ETS1 is also closely related to OSCC and consider as a potential biomarker of OSCC [23]. VAV3 have been identified as important molecules in tumor genesis, tumor growth and cell migration [22].After finding the core genes, GO and KEGG examination of hub genes were performed. In GO examination, BP positively associated with “positive regulation of endothelial cell migration”, “cellular response to antibiotic”, and so on (see Fig. 5A). CC enrichment showed that hub genes were importantly associated with “basal plasma membrane”, basal part of cell”, and so on (see Fig. 5B). MF analysis enriched in “superoxide − generating NADPH oxidase activator activity”, “phosphatidylinositol − 4, and so on (see Fig. 5C).

Fig. 5.
figure 5

(Source Created by the authors)

The GO (BP, CC and MF) and KEGG pathway enrichment analysis of hub-DEGs

Full size image
Table 2. Reporter Transcription Factors
Full size table

KEGG examination described that these hub genes were associated with “Renal cell carcinoma”, “Human T − cell leukemia virus 1 infection”, and so on (see Fig. 5D). The hub DEGs targeted as reporter TFs are HIF1A, MYC, FOXP3, E2F4, WT1, PURA, ZEB1 and USF2 (see Table 2).These genes promote tumor growth, development of cancer, and are strongly associated with OSCC.

Where HIF1A TFs as an oncogene engaged in OSCC multiplication and cisplatin-safe and may act as a novel remedial objectives for OSCC treatment [24]. It also promotes human coronavirus SARS-CoV-2 replication [25]. MYC is liable for regulated tumor genesis in OSCC [26]. MYC likewise prompts to the advancement of breast cancer. Foxp3 overexpression in cancer cells predicts unfortunate endurance in oral squamous cell carcinoma [27]. E2F4 is a statistical significance gene for OSCC [28]. WT1 can be used as epigenetic oral squamous cell carcinoma biomarkers [29]. Zeb1 is responsible for oral cancer cell invasion and its suppression by resveratrol [30]. USF2 is working as a risk factor for tumor genesis [31]. Hub-DEGs are also associated with miRNAs and are also involved in the cancer cell (see Fig. 6).

Fig. 6.
figure 6

(Source Created by the authors).

miRNA–mRNA network identification for hub-DEGs. Highlighted circles predicted the potential miRNAs that can regulate multiple hub-DEGs and miRNAs

Full size image

hsa-miR-495-3p can be used as a biomarker for the detection of OSCC [32]. Hsa-miR-155-5pis a potential oncogenic in oral squamous cell carcinoma [33]. Hsa-miR-454-3p has been involved in head and neck squamous cell carcinoma invasion [34]. Finally, we detected candidate drugs associated with hub-DEGs such as IMATINIB, BROMOCRIPTINE, NIFEDIPINE, PIRETANIDE and TRIAMTERENE which are mostly connected with multiple genes that have also highest degrees among the other drugs (see Fig. 8). IMATINIB is an oral cancer drug that specifically represses a few protein tyrosine kinases related with human malignancy. IMATINIB is an approved drug for oral cancer [35]. BROMOCRIPTINE is used in the supported medication of metastatic breast cancer and prostate cancer-related hyperprolactinemia [36]. NIFEDIPINE is also an indicator for carcinoma associated fibroblasts (CAFs) in OSCC [37]. PIRETANIDE has a significant effect on oral cancer [38]. TRIAMTERENE is a Photosensitizing antihypertensive drug to lip cancer in non-Hispanic males [39].

Fig. 7.
figure 7

(Source Created by the authors)

Overall survival analysis of hub-DEGs for OSCC.

Full size image
Fig. 8.
figure 8

(Source Created by the authors)

Drug-gene interaction network for OSCC. Potential drug with multiple connected hub-DEGs and TFs highlighted from the drug-gene network.

Full size image

5 Conclusion

The survival rate of OSCC patient is very low for its drug resistance. We analyzed cisplatin resistance of OSCC using bioinformatics tools and found hub-genes (AR, ETS1, MET, PDGFB and VAV3), TFs (HIF1A, MYC, FOXP3, E2F4, WT1, PURA, ZEB1 and USF2) and miRNA (hsa-miR-377-3p, hsa-miR-499a-5p, hsa-miR-34c-5p, hsa-miR-326, hsa-miR-34a-5p, hsa-miR-449b-5p, hsa-miR-449a and hsa-miR-589-3p) which may be responsible for cisplatin resistance of OSCC. Furthermore, we discovered candidate drugs (IMATINIB, BROMOCRIPTINE, NIFEDIPINE, PIRETANIDE and TRIAMTERENE) which might be a therapeutic target of OSCC. These molecular signatures are also needed further validation in wet lab.