Increased Expression of Integrin Subunit α3 is Correlated with Worse Recurrence-Free Survival in Thyroid Cancer

Objective: To investigate the potential effect of integrin subunit α3 (ITGA3) on thyroid cancer (THCA). Materials and Methods: Based on bioinformatics databases, the expression levels of ITGA3 were rstly analyzed, followed by evaluating the prognostic signicance of ITGA3 in THCA patients. Cox regression analysis predicted the independent prognostic factors for THCA. Then, cBioportal and GSCA databases were applied to evaluate genetic alterations of ITGA3. Functional enrichment analysis was conducted using the R package. Finally, the upstream microRNAs of ITGA3 were determined. Results: ITGA3 gene and protein levels were higher in the THCA group than normal group (all P <0.05). And ITGA3 mRNA expression was signicantly related to cancer stage and histological subtype (all P <0.01). Moreover, high expression of ITGA3 was associated with poor recurrence-free survival (RFS) of THCA patients in all subgroups (all P <0.01). Cox regression analysis presented that ITGA3 overexpression was an independent prognostic factor for worse RFS in THCA patients. Besides, signicant relations were observed between ITGA3 and genetic alterations (FDR <0.01). Functional enrichment analysis indicated ECM-receptor interaction, and cell adhesion molecules were the shared regulatory pathways. We also found that ITGA3 might be the target gene of hsa-miR-3129, hsa-miR-181d, hsa-miR-181b, hsa-miR-199a, and hsa-miR-199b, which were all correlated with THCA patient prognosis. Conclusions: ITGA3 could be a reliable biomarker and promising therapeutic target for improving the diagnosis and clinical outcomes of THCA patients.


Introduction
Thyroid cancer (THCA) is the most common endocrine malignancy, which was secondary to cancers of breast, lung, rectal, and cervical in 2017 among female malignancies [1]. In China, there were approximately 90,000 THCA cases in 2015 [2]. It is estimated that the number of new THCA cases reached 12.9 per 100,000 populations annually in 2015 in the United States [3]. According to the statistics published by the American Cancer Society, about 12,150 and 32,130 new THCA cases occurred in males and females, respectively in 2021. Importantly, THCA would be responsible for 2,200 deaths including 1,150 females and 1,050 males [https://www.cancer.org/cancer/thyroid-cancer/about/keystatistics.html]. Currently, ne-needle aspiration biopsy of thyroid nodules is a paramount method for the diagnosis of THCA due to its high accuracy in distinguishing benign from malignant lesions. However, clinicians are faced with a big challenge in distinguishing malignant follicular lesions from benign lesions on biopsy [4][5][6]. In addition, despite the majority of THCA patients having a good prognosis, recurrences are observed in nearly 30% of the patients [7]. Therefore, it is urgent to develop novel biomarkers with good sensitivity and speci city to improve the diagnosis and prognosis of THCA patients.
Integrins are cell surface receptors for extracellular matrix (ECM) proteins, which are widely expressed in tissues and organs in the human body [8]. They represent a crucial role in the proliferation, migration, invasion, and metastasis of the cancer cells [9,10]. As a member of the integrin family, integrin subunit α3 (ITGA3) joins a β1 subunit to form an intact integrin and interacts with ECM proteins. Previous studies have demonstrated abnormal expression of ITGA3 in various cancers. Huang et al. showed that ITGA3 was highly expressed in intrahepatic cholangiocarcinoma, promoting cell proliferation and cell cycle progression [11]. The elevated levels of ITGA3 in head and neck cancer led to an unfavorable prognosis for the patients [12]. In addition, ITGA3 was directly regulated by miR-223, and silencing of ITGA3 signi cantly inhibited prostate cancer cell migration and invasion [13]. However, there have been few reports regarding the role of ITGA3 in THCA.
In this study, we aimed to explore the relationship between ITGA3 expression and the prognosis of THCA patients from the following aspects: the expression levels of ITGA3 were rstly analyzed through various bioinformatics databases. Then, the prognostic signi cance of ITGA3 mRNA expression in THCA was evaluated through Kaplan-Meier plotter and Cox regression analyses. By identifying the differentially expressed genes (DEGs), the gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) pathway analyses were performed, gene set enrichment analysis (GSEA) was conducted, and the upstream microRNAs (miRNAs) of ITGA3 were identi ed to reveal the underlying molecular mechanisms of abnormal ITGA3 in THCA.

Expression analysis of ITGA3
We rst explored the mRNA expression of ITGA3 in pan-cancers through Gene Set Cancer Analysis (GSCA) database (http://bioinfo.life.hust.edu.cn/GSCA/#/). The GEPIA database (http://gepia.cancerpku.cn/) was used to analyze the mRNA expression of ITGA3 in THCA and normal samples by setting the parameters of |log2-fold change (FC)| cutoff =1 and P-value =0.01. Then, RNA-seq-HTseq-FPKM data for THCA samples based on GDC-TCGA were collected from UCSC Xena (https://xenabrowser.net/datapages/) database. These data were used to assess the ITGA3 mRNA expression in THCA and paired normal tissues. Moreover, the mRNA expression of ITGA3 in THCA based on clinical characteristics including age, gender, cancer stage, and histological subtype was analyzed via the UALCAN database (http://ualcan.path.uab.edu/). UALCAN is a comprehensive, user-friendly, and interactive web resource for analyzing cancer omics data.
Following this, the protein expression levels of ITGA3 in THCA and normal thyroid tissues were evaluated by the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/) in the "PATHOLOGY" column. The levels of expression were classi ed into not detected, low, medium, and high through the score system, including the intensity of staining (negative, weak, moderate, or strong) and the proportion of stained cells (<25%, 25-75%, or >75%). In this database, we also explored the subcellular distribution pattern of ITGA3 proteins in human cells. Three cell lines, A-431, U-2 OS, and U-251 MG, were included in the analysis of protein subcellular localization by immuno uorescence. The HPA database maps all the human proteins in cells, tissues, and organs using an integration of various omics technologies.

Survival analysis of ITGA3
Using the Kaplan-Meier plotter website (http://kmplot.com/), we examined the in uence of ITGA3 on the overall survival (OS) and the recurrence-free survival (RFS) on THCA patients by splitting patients with the best cutoff [14]. Then, the clinical information and related RNA-seq data were retrieved from cBioportal for Cancer Genomics (https://www.cbioportal.org/) by choosing "Thyroid carcinoma (TCGA, Firehose Legacy) and querying "ITGA3". Patients with complete survival information and ITGA3 expression data were enrolled in the study. We used "surv t" in R package to analyze the association of ITGA3 expression with clinical outcome of THCA patients according to age (<55/ ≥55), gender (female/male), and stage (1+2/3+4). Maxstat statistics was adopted to determine the optimal cutoff for dividing high and low ITGA3 expression samples. Further, univariate and multivariate Cox regression analyses were performed to predict the independent prognostic factors for RFS in THCA patients. Female was set as a reference level for gender, stage 1+2 for stage, and Asian for race.

Mutation, Copy Number Variation (CNV), and methylation analysis of ITGA3
We next focused on the potential mechanism of dysregulated ITGA3 expression in THCA and investigated the ITGA3 mutations in THCA through cBioPortal for Cancer Genomics in the "Mutations" column with a total of 397 samples. In addition, the association of ITGA3 mRNA expression with CNV, and DNA methylation were investigated, respectively through the GSCA database. GSCA is an integrated database for genomic and immunogenomic gene set cancer analysis.

Enrichment analysis
The R software limma package was employed to identify the DEGs between the low and high ITGA3 expression groups (cutoff value of 50%) in gene expression pro les of THCA. An absolute log2FC >1 and P-value <0.05 were used as the lter criterion for signi cantly DEGs. The signi cantly enriched GO terms and the KEGG pathways of these DEGs were investigated using the clusterPro ler in R package. P-value < 0.05 and false discovery rate (FDR) < 0.25 were considered statistically signi cant. Subsequently, GSEA was performed to elucidate the survival differences between the low and high ITGA3 expression groups. The gene set was permutated 1000 times and the expression level of ITGA3 was used as a phenotypic label. A nominal p-value <0.05 and an FDR q-value <0.25 were considered to be statistically signi cant.

Statistical analysis
All statistical analyses were performed by SPSS 23.0 (SPSS, Inc., Chicago, IL, USA) and R software. Paired t-test was used to assess the differential expression between THCA and paired normal tissues. Kaplan-Meier plotter curves and the log-rank test were performed to determine survival differences. The relationship between ITGA3 and CNV and DNA methylation was analyzed by the Spearman correlation test in GSCA. While the Pearson correlation test was utilized to evaluate the relationship between ITGA3 expression and miRNAs in starBase. P <0.05 was considered statistically signi cant.

Expression analysis of ITGA3
Firstly, we analyzed the ITGA3 mRNA expression in various cancers through the GSCA database. As shown in Figure 1A, ITGA3 was highly expressed in 9 types of cancer including THCA (all P <0.05).
Quantitative analysis con rmed ITGA3 overexpression in THCA compared with normal thyroid tissue (P <0.05) ( Figure 1B). Besides, the THCA group had a signi cantly higher ITGA3 mRNA expression than that in the paired normal group (P <0.001) ( Figure 1C). Further, the mRNA expression of ITGA3 in THCA based on clinical characteristics including age, gender, cancer stage, and histological subtype were analyzed through the UALCAN database. There was no signi cant difference in age and gender (Figure 2A-2B), but cancer stage and histological subtype were signi cantly associated with ITGA3 expression (all P <0.01) ( Figure 2C-2D).
Next, the protein expression of ITGA3 in THCA and normal tissues were explored via the HPA database. Figure 1D showed that the protein staining of ITGA3 was high and the intensity was strong in THCA tissue; however, low staining and weak intensity were observed in normal tissue. Figure 1E presented the ITGA3 protein localization and immuno uorescence result, showing that the ITGA3 protein was mainly located at the plasma membrane in human tumor cells. Taken together, the above ndings indicated that ITGA3 was highly expressed in THCA and might be related to cancer development.

Survival analysis of ITGA3
To investigate the effect of ITGA3 on OS and RFS in THCA patients, the survival analysis was carried out using a Kaplan-Meier plotter. As exhibited in Figure 3A, ITGA3 had no signi cant impact on OS (P >0.05).
However, patients with higher ITGA3 mRNA expression had shorter RFS with a hazard ratio of 3.09 (P <0.01) ( Figure 3B). Further strati ed analysis showed that high ITGA3 expression correlated with poor RFS in all the subgroups ( Table 1).
Nevertheless, other clinical factors including age, gender, African-American, and Caucasian were not signi cantly related to patient prognosis. After integrating these factors into multivariate Cox regression analysis, stage 3+4 (P =0.027) and high expression of ITGA3 (P =0.002) were still independent prognostic factors for worse RFS in THCA patients (Table 2). These results revealed that ITGA3 played a vital role in the prognosis of THCA. Abbreviations: CI, con dence interval. Abbreviations: HR, hazard ratio; CI, con dence interval.

Mutation, CNV, and methylation analyses of ITGA3
Due to the independent prognostic value of ITGA3 for RFS in THCA patients, we further investigated the potential mechanism of dysregulated ITGA3. A mutation plot from the cBioportal for Cancer Genomics presented that the mutation sites of ITGA3 were D790N, and R874L ( Figure 4A). Besides, the mRNA expression of ITGA3 showed a signi cantly negative correlation with CNV (Cor. =-0.16, FDR =3.1e-03) ( Figure 4B). Similarly, the mRNA expression was negatively related to DNA methylation (Cor. =-0.42, FDR <0.001) ( Figure 4C). Therefore, the mutations, CNV, and DNA methylation might lead to the aberrant upregulation of ITGA3 in THCA.

Functional enrichment analysis
To elucidate the pathological role of ITGA3 in THCA, the GO and KEGG analysis of the DEGs between the low and high ITGA3 expression groups were performed using clusterPro le package. A total of 596 DEGs were generated from gene expression RNA-seq-HTseq-FPKM, including 198 upregulated and 398 downregulated DEGs as shown in the volcano plot ( Figure 5A) and the heat map ( Figure 5B). Functional enrichment analysis results presented that the major GO terms were biological adhesion, cell adhesion, extracellular region, vesicle, signal receptor binding, and MHC class II receptor activity ( Figure 5C). As for the KEGG pathway, they were mainly enriched in PI3K-Akt, cell adhesion molecules, MAPK signaling pathway, and ECM-receptor interaction ( Figure 5D).
Then, GSEA was conducted to gain further insight into the biological pathways involved in THCA. Among the top 5 pathways in high ITGA3 expression phenotype, ECM-receptor interaction, and cell adhesion molecules were the consistent pathways with KEGG analysis (Figure 6). Thus, high expression of ITGA3 contributed to the poor prognosis of THCA patients, which might be related to the activation of these pathways, especially ECM-receptor interaction, and cell adhesion molecules.

Discussion
THCA accounts for more than 95% of endocrine cancer, and most are well-differentiated arising from follicular epithelial cells [15]. Although THCA patients have better clinical outcomes, a number of patients still develop local recurrence with or without cervical lymph node metastasis or distant metastases [16].
With the development of molecular biology technology, researches on THCA have also been progressed [17]. Previous studies have demonstrated that high expression of COX-2 was related to the pathological type of THCA, and its overexpression contributed to poor prognosis [18,19]. Fu et al. con rmed that TFAP2B overexpression accelerated tumor growth and progression of THCA via the COX-2 signaling pathway [16]. Besides, PTPN2 upregulated by in ammatory response or oxidative stress promoted the THCA development [20]. However, little is known about the expression of ITGA3 and prognostic value in THCA.
In this study, we found that ITGA3 was highly expressed at both gene and protein levels in the THCA group compared with the normal group. ITGA3 mRNA expression was signi cantly related to the cancer stage and histological subtype. It's worth noting that the high expression of ITGA3 had an intimate relation with worse RFS. Cox regression analysis exhibited that the upregulated expression of ITGA3 served as an independent prognostic factor for worse RFS of THCA patients. Of note, ITGA3 overexpression predicted poor clinical outcomes in all subgroups. Thus, ITGA3 might be a novel and robust prognostic biomarker for THCA.
Tumor-initiating genetic events have become a primary focus in THCA initiation in the recent decade [21,22]. By analyzing the TCGA data, we found that mutations, CNV, and methylation might result in the aberrant elevation of ITGA3 expression in THCA. In addition, GO and KEGG analysis of DEGs based on ITGA3 expression levels showed that ITGA3 regulated the development of THCA by participating in multiple pathways including PI3K-Akt, cell adhesion molecules, MAPK signaling pathway, and ECMreceptor interaction. This nding further indicated that genetic alterations activate intracellular signaling pathways such as PI3K-Akt and MAPK pathways, which were implicated in THCA cell survival and proliferation [23,24]. To further reveal the pathological function of ITGA3 in THCA, we performed GSEA to analyze the genome dataset of THCA samples with high and low expression groups of ITGA3. The results showed that ECM-receptor interaction and cell adhesion molecules were the consistent pathways with KEGG analysis. Integrins are involved in mediating multiple intracellular signals via interaction with the ECM and participate in the attachment of cells to the ECM by the formation of cell adhesion complexes [25]. It has been reported that integrins played a regulatory role in angiogenesis, tumor growth, and invasion through interaction with the ECM in glioblastomas [26]. Accumulated evidence exhibited that integrins are expressed in stem cells and promote cell migration through ECM stimulation [27,28]. The inhibition of ITGA3 led to a decrease in the expression of cancer stem cell markers, indicating that the knockdown of ITGA3 impairs breast cancer cell stemness [29]. Similarly, THCA originates from stem cells [30], and hence it is reasonable to infer that ITGA3 overexpression might upregulate the expression of THCA stem cell markers, and promote cancer progression via the ECM-receptor interaction pathway. Our study provided new insight into the development of THCA.
We next explored the potential miRNAs targeting ITGA3 mRNA expression by using DINAN, miRDB, miRWALK, and Targetscan databases. Through Venn analysis, has-mir-181b, has-mir-181d, has-mir-3129, has-mir-199b, and has-mir-199a were identi ed as consistent miRNAs from the four databases. As short endogenous non-coding molecules, miRNAs played predominant roles in the occurrence of many tumors [7,31]. It has been demonstrated that miR-3666 was related to THCA cell proliferation, and miR-146b-5p was involved in cell invasion and migration of different THCA cells [32,33]. This research revealed that has-mir-181b, has-mir-181d, has-mir-3129, has-mir-199b, and has-mir-199a were all associated with the prognosis of THCA patients. Moreover, the correlation analysis exhibited that ITGA3 had a signi cant relationship with the consistent miRNAs except for has-mir-3129. For the rst time, we linked the expression of ITGA3 to the ve consistent miRNAs, which provided a better understanding of the underlying mechanisms by which ITGA3 affected THCA progression. However, in vitro and in vivo experiments are required to validate these results if conditions are available in the future.
In conclusion, our study showed that ITGA3 was highly expressed in THCA and high expression of ITGA3 led to worse RFS of THCA patients. Besides, ITGA3 might affect the progression of THCA by regulating the ECM-receptor interaction pathway. ITGA3 might serve as a vital predictor and potential target for THCA treatment.

Declarations
Availability of data and materials: The dataset used and/or analyzed during the current study is available from the corresponding author on reasonable request.