Synthetic Evaluation of MicroRNA-1-3p Expression in Head and Neck Squamous Cell Carcinoma Based on Microarray Chips and MicroRNA Sequencing

Background MicroRNA-1-3p (miR-1-3p) exerts significant regulation in various tumor cells, but its molecular mechanisms in head and neck squamous cell carcinoma (HNSCC) are still ill defined. This study is aimed at detecting the expression of miR-1-3p in HNSCC and at determining its significant regulatory pathways. Methods Data were obtained from the Cancer Genome Atlas (TCGA), Gene Expression Omnibus (GEO), Oncomine, ArrayExpress, Sequence Read Archive (SRA) databases, and additional literature. Expression values of miR-1-3p in HNSCC were analyzed comprehensively. The R language software was employed to screen differentially expressed genes, and bioinformatics assessment was performed. One sequence dataset (HNSCC: n = 484; noncancer: n = 44) and 18 chip datasets (HNSCC: n = 656; noncancer: n = 199) were obtained. Results The expression of miR-1-3p in HNSCC was visibly decreased in compare with noncancerous tissues. There were distinct differences in tumor state (P = 0.0417), pathological stage (P = 0.0058), and T stage (P = 0.0044). Comprehensive analysis of sequence and chip data also indicated that miR-1-3p was lowly expressed in HNSCC. The diagnostic performance of miR-1-3p in HNSCC is reflected in the sensitivity and specificity of the collection, etc. Bioinformatics analysis showed the possible biological process, cellular component, molecular function, and KEGG pathways of miR-1-3p in HNSCC. And ITGB4 was a possible target of miR-1-3p. Conclusions miR-1-3p's low expression may facilitate tumorigenesis and evolution in HNSCC through signaling pathways. ITGB4 may be a key gene in targeting pathways but still needs verification through in vitro experiments.


Introduction
HNSCC is a human malignant neoplasm common in certain regions. The morbidity of HNSCC has improved dramatically in recent years, particularly among women. Each year, over 600,000 patients are diagnosed HNSCC globally, over half in the Asia-pacific regions [1]. Platinum-based chemotherapy, radiotherapy, and cetuximab are commonly used to treat recurrent/metastatic HNSCC in Asia [2]. Unfortunately, about 75% of these patients have local progression (60%) or metastasis (15%) upon first contact, and the median survival of patients with relapse and/or metastasis is only 6 months. Survival rates drop rapidly in patients who fail first-line treatment, often dying within three months [3]. Seeking antitumor therapies that are more effective, more scholars are paying close attention to moleculartargeted therapy. Recently, many studies on microRNAs (miRNAs) have been affirmed in the cancer field, and molecular-targeted therapy has become a burgeoning treatment for tumors [4].
MicroRNAs of approximately 22 nucleotides long are noncoding single-stranded RNAs, coded efficiently by endogenous genes [5]. They regulate the expression of posttranscription genes. Many studies have presented distinctions in the expression of some miRNAs in cancerous and noncancerous tissues [6], utilizing signaling pathways to control key genes affecting physiological and biochemical processes such as proliferation and differentiation in tumor cells. Notwithstanding, the clinical value and molecular functions of individual miR-NAs remain relatively unexplored. As a gene member, miRNA-1-3p targets different proteins or genes and affects the occurrence and development of gastric [7], colon [8], and breast carcinomas [9,10], among others [11][12][13]. However, there are few relevant articles about the expression of miR-1-3p in HNSCC. Only 8 research teams have explored the ontology of miR-1-3p on key genes that regulate HNSCC [1,[14][15][16][17][18][19][20]. Potential molecular mechanism regulation in HNSCC is still unclear, and new enrichment regulatory pathways need to be proposed.
In this study, a total of 1,140 cancer samples and 243 noncancer samples were collected based on sequencing, microarray, and literature data to probe the clinical significance and impact of miR-1-3p in HNSCC. Potential molecular mechanisms, including significant genes and enrichment paths of miR-1-3p in HNSCC, were summarized. ITGB4 was one of the target genes. The schematic of the research design is shown in Figure 1.

Sources of miR-1-3p Expression Data in HNSCC
2.1.1. MicroRNA-Seq Data. The miRNA sequence dataset was obtained from TCGA [21], including samples of cancerous and noncancerous tissues. The steps for data download were as follows: the site for UCSC-Xena (https://xena.ucsc.edu/) visited, checked "TCGA hub" in the "DATASETS" option, and selected "TCGA(HNSC)," which contains 25 datasets. In the new interface, selected "IlluminaHiseq (n = 529) TCGA hub" to download matrix files and gene annotation files. By matching the two, the expression value of miR-1-3p was determined. After deleting missing data, mature miR-1-3p expression data were extracted. Meanwhile, using "sangerbox," the corresponding clinical case data were downloaded for analysis.

Microarray Data.
Microarrays were filtered to evaluate miR-1-3p expression in Gene Expression Omnibus (GEO) [22]. The overall strategy for retrieval was OSCC OR HNSCC OR "head and neck" OR "nasopharynx." We adjusted search terms to achieve the best range. The search was restricted to "Series" in "Entry type" and "H. sapiens" in "organisms." All research contained in the chips followed these criteria: (1) the species is human, (2) the objective is tissue, (3) microarray results include required gene, (4) containing both HNSCC tissues and nontumor tissues, and (5) miR-1-3p expression is mature. Conversely, the exclusion criteria were (1) objects other than human, (2) serum sample or other liquid types, (3) the chip does not contain the required genes, (4) there is only the experimental group or the control group, (5) 10 15 2 BioMed Research International we downloaded the "SOFT formatted family file(s)" probe annotation files and "Series Matrix file(s)" gene probe expression Matrix file, to find the miR-1-3p corresponding expression data. For chip screening in ArrayExpress [23], Oncomine [24], and SRA [25], the methods were the same as above.

Literature Data.
We reviewed articles about gene expression in HNSCC in Chinese and foreign scientific research sites, including CNKI, Wanfang, Vip, PubMed, Web of Science, and EBSCO databases. We extracted these chips to supplement existing microarrays to acquire unabridged data.  (a, e) The total expression of miR-1-3p in HNSCC and nontumor tissue from the TCGA database. (b, f) The relationship between miR-1-3p expression and tumor status. (c, g) The relationship between miR-1-3p expression and pathologic stage. (d, h) The relationship between miR-1-3p expression and T stage. The differential expression of miR-1-3p in HNSCC was statistically significant, which was manifested in tissue, tumor status, pathological stage, and T stage. AUC: the area under the ROC curve; P value: T-test with two independent samples.

Comprehensive and Detailed
Analysis. Based on the sequencing data and the specific content of each microarray, the data were selected to make log2 processing or not. The mean expression level (mean) and standard deviation (SD) were calculated using SPSS 22.0. The "car" package of R was employed to draw violin plots to clarify that the miR-1-3p's expression was different in cancerous and noncancerous tissues. To evaluate the expression level of miR-1-3p comprehensively, a meta-analysis of continuous variables was conducted with Stata 12.0. When heterogeneity was small (I 2 < 50%), the fixed effect model was adopted for analysis. On the contrary, if heterogeneity was extensive (I 2 > 50%), a random-effect model was used, and we continued to perform sensitivity analysis to determine the sources of heterogeneity. After removing the chips that contributed to heterogeneity, we reanalyzed the results based on remaining data. If heterogeneity was less than 50%, the results were reliable. For statistical analysis, if the standard mean deviation ðSMDÞ < 0 and the 95% confidence interval (CI) did not cross the 0-point coordinate line, the gene was demonstrated to be significantly lower in HNSCC. An SMD > 0 meant that the research object was highly expressed in carcinoma. The ROC curves were plotted using SPSS, and the sROC curve was drawn using the Stata software to forecast the clinical significance of miR-1-3p in HNSCC. When the area under the curve was >0.5, this index had a certain diagnostic value for diseases. The AUC over 0.7 indicated a good diagnostic value. Publication bias was determined using Begg's funnel plot.
2.4. Bioinformatics Analytical Methodology. Gene annotation and pathway enrichment analysis of DEGs were conducted in DAVID v 6.8; top-ranked annotations and pathways were listed, and P < 0:05 indicated that the difference was statistically significant. The MCODE plugin in Cytoscape 3.4.0 is an APP that performs topological gathering on a fixed net to find dense connection areas. All differentially expressed genes enriched in five Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were introduced into the software. In addition, the most central modules in the PPI network were confirmed by MCODE.

Verify the Expression of Target
Genes. Gene expression profile information of the 11 key genes in the sequence was imported into the cBioPortal website for genetic variation to verify the regulatory relationship between pivotal genes and miR-1-3p. In addition, the GraphPad Prism 8 software was used to verify the association between miR-1-3p and key genes.

Expression and ROC Curves of miR-1-3p in HNSCC in
Different Gene Chips. Eighteen pieces of qualified chips were filtered from GEO, ArrayExpress, Oncomine, and SRA databases (Table 2). We calculated the mean and the SD in microarrays (Table 3) and plotted violin diagrams ( Figure 3). We used adoptive data to calculate the AUC ( Figure 4). Foregoing chip results revealed that the miR-1-3p expression level in HNSCC was lower than in noncancer cells, in accordance with the results from the sequence.

Gene Enrichment Analysis of miR-1-3p in HNSCC.
According to the logFC and P value, 174 upregulated differentially expressed genes and 103 downregulated genes were screened. Given that miR-1-3p was lowly expressed in HNSCC, we chose 174 upregulated DEGs for GO annotation and KEGG enrichment by DAVID 6.8 (P < 0:05), whose top five enriched terms were summarized according to P value (Table 4). Leukocyte migration, positive regulation of cell proliferation, apoptotic process, cell adhesion, and inflammatory response were the five most conspicuous terms for biological process (BP). In the cellular component (CC), genes were major enriched in the cytoplasm, cytosol, nucleoplasm, extracellular exosome, and membrane. As for molecular function (MF), the coexpressed proteins were involved in protein binding, ATP binding, identical protein binding, receptor binding, and protein heterodimerization activity. For the KEGG pathway, the coexpressed genes were major gathered in pathways in cancer, proteoglycans in cancer, PI3K-Akt signaling pathway, focal adhesion, and micro-RNAs in cancer (Figure 7). In these genes, gene networks showed that KRAS, CD44, COL4A1, SHC1, CAV2, ITGB4, THBS1, SPP1, FLNA, FN1, and NRAS were closely connected in the pathways (Figure 8(a)).

Preliminary Prediction of ITGB4 as Target
Gene of miR-1-3p. The results of cBioPortal showed that the 11 target genes had different degrees of variation in HNSCC, reflected in a missense mutation, gene fusion, gene amplification, and gene deletion (Figure 8(b)). Only 2 HNSCC articles mentioned ITGB4 and showed that it may be the target of HNSCC, which can promote distant metastasis of tumors through the blood. Pearson correlation analysis showed that miR-1-3p had a correlation with ITGB4, which was statistically significant (P < 0:001) (Figure 8(c)). For this reason, this gene was selected for further verification, and special attention was focused on the expression and prognosis of ITGB4 in HNSCC (Figures 8(d)  and 8(e)).        Figure 6: The values of total sensitivity, total specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnosis rate (DOR), and 95% confidence interval are statistically significant. (a) The SROC curve of miR-1-3p expression based on 19 datasets. (b-f) The forest map showed the diagnostic performance of miR-1-3p in HNSCC: the sensitivity of the collection, the specificity of the collection, the positive likelihood ratio of the summary, the negative likelihood ratio of the summary, and the summary diagnostic ratio based on the qualified dataset. 16 BioMed Research International

Discussion
The downregulation of miR-1-3p in HNSCC was supported by 1,140 HNSCC and 243 noncancer tissue samples from TCGA, GEO, Oncomine, ArrayExpress, and SRA databases, which enhanced the dependability of our results. miR-1-3p's low expression could be associated with the burgeoning of HNSCC. To study the functional implication of miRNA-1 in HNSCC cells and identify new neoplastic paths, a more reliable set of target genes was obtained by integrating potential target genes composed of four parts: miR-1-transfected DEGs, sequence DEGs, chip DEGs, and the targets in the prediction tool. KEGG pathway analysis showed the most significant pathways were "the pathways in cancer," "proteoglycans in cancer," "PI3K-Akt signaling pathway," "focal adhesion," and "MicroRNAs in cancer." We confirmed pivotal targets of miR-1-3p. ITGB4 was one of the most important targets.

18
BioMed Research International proliferation [1,34]. miR-1-3p was proven dysregulated in HNSCC [16,35]. Nohata et al. [15] demonstrated that miR-1 was downregulated in HNSCC samples and disclosed that transgelin 2 (TAGLN2) was directly adjusted by miR-1. Koshizuka et al. [1] verified gene expression in HNSCC and found that miR-1 was reduced clearly in HNSCC tissues. However, only eight studies have investigated miR-1 expression in HNSCC. Further studies are needed to determine the relationship between miR-1 and HNSCC. The 174 upregulated differentially expressed genes were screened from 39 datasets, three miR-1 transfection samples, and a prediction tool using the R language tool. Although the gene list was not large, screening was rigorous. The obtained difference genes were accurate and had a certain degree of persuasion. The KEGG enrichment pathway is involved in multifarious cancer processes. G proteincoupled receptors (GPCRs) play a vital role in signal transmission [36,37]. Targeted proteins recognize and bind to binding sites in eukaryotes and activate a series of signaling pathways, causing changes in tumor cell states, promoting tumor blood vessel regeneration, and participating in the occurrence and course of neoplasms [38][39][40][41]. Therefore, understanding the specific mechanism of GPCR involvement in malignant tumors and related target genes is conducive to providing new opportunities for cancer prevention and treatment [36,37]. Some studies have explored this pathway in bladder cancer [42,43], colorectal cancer [38], melanoma [44,45], endometrial cancer [46], lung cancer [47], renal cell carcinoma [48,49], and thyroid cancer [50,51]. Moreover, Koshizuka et al. found that miR-1 inhibited tumor growth by targeting growth factor receptors and participated in various signaling pathways, including the "pathways in cancer," which was consistent with our results [1].
As key genes corresponding to important action sites, which contained important signal pathway information, we focused on hub genes in further research. Eleven hub genes were enriched in KEGG pathways, most of which had been reported by a large number of previous studies on HNSCC. ITGB4 caught our attention. The results showed that ITGB4 was highly expressed in HNSCC and was harmful to patient prognosis (P < 0:001).
Integrin family, a family of cell adhesion receptors, is recognized to play a key role in malignant tumor metastasis [52]. As a component of the basement membrane, the expression levels of laminin-5 and its ligand were negatively correlated with tumor invasiveness, metastasis, and poor clinical prognosis [53][54][55]. ITGB4 encodes a receptor for laminin-5. Studies have shown that the decreased expression of ITGB4 and laminin-5 genes occurs during the progression of prostate intraepithelial neoplasia and the development of prostate cancer [56]. Meanwhile, ITGB4 can be used as a target site to form a lump in colorectal cancer [57], gastric cancer [58,59], prostate cancer [60][61][62], lung cancer [63], and other diseases [64,65] to regulate the progress of diseases. A single study explored the extracellular matrix-(ECM-) receptor interaction, and ITGB4 can be an underlying target for the diagnosis and treatment of HNSCC [66]. Through PCR analysis of oral squamous cell carcinoma data and assessment of pathological clinical parameters, Nagata et al. [67] found that ITGB4 could promote distant metastasis of tumors. ITGB4 is a good biological indicator of tumors. In this study, we used bioinformatics methods to speculate that ITGB4 gene can influence the disease course of HNSCC, and ITGB4 gene was correlated with miR-1-3p. In the treatment of HNSCC, this feature could be used to develop an inhibitor of ITGB4 for treatment of HNSCC.
Our research aimed to integrate microarrays and miRNA sequencing to study the expression and deep-

20
BioMed Research International seated mechanism of miR-1-3p in HNSCC. There are limitations in this research. First, data came from online databases; thus, larger clinical samples are needed for further experimental inquiry. Second, the targeting relationship between miR-1-3p and ITGB4 was preliminarily verified, but further experiments are needed before clinical application.

Conclusion
Our research confirms the downregulation of miR-1-3p in HNSCC, revealing that miR-1-3p can act on target genes, activate signaling pathways, and participate in the development of HNSCC. ITGB4 may be a novel biological target protein, which requires further experimentation.  Pearson correlation analysis showed that miR-1-3p was negatively correlated with ITGB4 (P < 0:0001). (d) ITGB4 expressed highly in HNSCC tumor tissues than in noncancer tissues. (e) Kaplan-Meier survival curve was used to analyze the ITGB4 expression data and evaluate its effects on the prognosis of HNSCC. ITGB4 had an apparent influence on the survival of HNSCC patients (P < 0:0001).

Data Availability
The datasets generated and analyzed in the present study are available from the corresponding author upon reasonable request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.