Using Next-Generation Sequencing and Bioinformatic Methods to Predict New Genes That May Be Regulated by CD47 in Oral Squamous Cell Carcinoma

Oral squamous cell carcinoma (OSCC) is one of the most common cancers in the world, and the incidence and death rate of OSCC in men is twice that of women. CD47 is a ubiquitous cell surface transmembrane protein, also known as integrin-related protein (IAP). Previous studies have pointed out that CD47 can inhibit the growth of OSCC, but the detailed mechanism is not clear. This study aimed to explore the effect of CD47 gene expression profiles in OSCC. The OSCC cell lines, OECM-1 and OC-2, overexpressed CD47, and the expression profiles of mRNAs were analyzed through next-generation sequencing (NGS) with a bioinformatic approach. A total of 14 differentially expressed genes (DEGs) were listed. In addition, ingenuity pathway analysis (IPA) was used to analyze the molecular function (MF), biological process (BP), and cellular component (CC) network signaling. The human protein atlas (HPA) database was used to analyze gene expression and the survivability of human cancer. The results found that HSPA5, HYOU1, and PDIA4 were involved in the IPA network and when highly expressed, mediated the survivability of cancer. In addition, HSPA5 was positively and significantly correlated with CD47 expression (p < 0.0001) and induced by CD47-overexpression in the OECM-1 and OC-2 OSCC cancer cell lines. These findings provide important insights into possible new diagnostic strategies, including unfolded protein for OSCC-targeting CD47.


Introduction
Oral cancer is one of the most common cancers in the world. It is estimated that 657,000 new cases of oral and throat cancer are generated every year, and more than 330,000 people die. The incidence and death rate of oral cancer in men is twice that of women [1]. Normally, oral cancer is a group of head and neck cancers, formed on the mucosal surface of the lips, hard palate, posterior molar triangle, and mandible [2]. The most common oral cancer tissue type is squamous differentiated carcinoma, caused by epithelial mucosa, also known as oral squamous cell carcinoma (OSCC), which accounts for more than 90% of all oral cancers [3]. Oral cancer can be cured in its early stage by surgery, radiotherapy, and chemoradiation. However, most patients with OSCC already have advanced oral cancer when diagnosed. At this time, the treatment effect and prognosis are poor, and the mortality rate is high [4][5][6]. Studies have shown that the occurrence of oral cancer is related to multiple risk factors including tobacco use, drinking alcohol, betel chewing, human papillomavirus (HPV) [7], a weakened immune system, a lack of nutrition in the diet, and a poor lifestyle. In particular, betel chewing, excessive alcohol consumption, and smoking cause an increased incidence and mortality of OSCC [8,9].
CD47 is a ubiquitous cell surface transmembrane protein, also known as integrinrelated protein (IAP), which belongs to the immunoglobulin superfamily (IgSF). It can interact with a variety of cell surface receptors including integrins (such as αvβ3, αIIbβ3, and α2β1), thrombospondin (TSP-1), and signal regulatory protein α (SIRPα), for the signal initiation related pathway [10,11]. For example, CD47 is the ligand of SIRPα, which is a transmembrane receptor of the immunoglobulin (Ig) superfamily with extracellular immunoglobulin-like domains and structural antigen receptors. In bladder cancer, CD47 initiates an immune checkpoint through the SIRPα signaling pathway [12]. In addition, CD47 can also be used as a receptor of TSP-1, which is secreted by several non-hematopoietic cells such as platelets, monocytes, and macrophages. CD47 is involved in a variety of physiological functions including leukocyte adhesion and migration, T cell activation, apoptosis, and phagocytosis [13][14][15]. Related studies have shown that the expression of CD47 increases in migrating hematopoietic stem cells (HSCs), which may provide protection from an important feature of macrophage killing [16]. Therefore, the expression level of CD47 can predict the possibility of HSCs being swallowed during circulation. More evidence indicates that CD47 is a common mechanism found in human solid tumor cells [17][18][19]. Through these mechanisms, tumor cells can protect themselves from phagocytosis, which results in proliferation and metastasis [20].
During immunotherapy, we found that chimeric antigen receptor (CAR)-T cells that bind CD47 antigen specifically kill different types of cancer including ovarian and pancreatic cancer cells [21]. Interestingly, a previous study also found that the effect of radiotherapy can be activated by targeted immunotherapy, as silencing of CD47 and HER2 eliminates radioresistant breast cancer cells [22]. In addition, previous studies have found that high expression of CD47 was detected in OSCC cell lines and in vivo results have shown that CD47 was significantly higher in OSCC than in proximal normal tissues. Furthermore, the poor prognosis in patients with OSCC has been connected to the CD47-SIRPα signaling pathway [23].
In the present study, we examined gene expression profiles by NGS and associated these data with gene network signaling through bioinformatic analyses of OSCC cell lines. We found that unfolded proteins HSPA5, HYOU1, and PDIA4 are induced by CD47 within the OSCC IPA network.

RNA-Seq Quantification
Differential gene expression (DEG) was analyzed by RNA-seq quantification following the protocol of Illumina (Genomics, Taipei, Taiwan). Briefly, Poly-T oligo-attached magnetic beads were used to purify the mRNA and synthesize double-strand cDNA, which was then sequenced on the Illumina NextSeq 500 platform, which obtained approximately 10 million reads for each sample. The gene expression profiles were analyzed and compared between the CD47-overexpressing and control OECM-1 and OC-2 oral cancer cell lines. All genes showing differential expression were further categorized using Gene Ontology (GO) and DEG report analysis.

Ingenuity Pathway Analysis
The IPA software (Ingenuity Systems, Redwood City, CA, USA) comprehensively explores DEGs to determine their biological function and signaling pathway. The DEGs were mapped to the IPA genetic network, then the score was calculated and sorted. The score was calculated by computing a statistic for each biological function according to a model that assigns random adjustment directions. The IPA software was used to conduct network analysis of CD47 downstream genes and to explore the underlying genetic networks.

Database for Annotation, Visualization and Integrated Discovery Analysis
DAVID analysis is a powerful tool for classifying genes through their function. It integrates data and calculates similarities using the global annotation map derived from multiple functional annotation databases including the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) biological processes.

Statistical Analysis
The gene expression was compared between cells with CD47-overexpression and normal cells. All statistical analyses were performed using the GraphPad Prism statistical software (GraphPad Software, San Diego, CA, USA).

Gene Expression Profiling and Ovexpression of CD47 in OSCC Cell Lies
Previous studies showed that CD47-overexpression may regulate the growth process of oral cancer [23]. In this study, we mainly focused on Asian patients suffering from OSCC. OECM-1 cells were derived from the surgical resection of a primary tumor from a Taiwanese patient with a history of betel quid chewing. OC-2 cells were derived from a primary tumor of the buccal mucosa of a Chinese patient with a habit of alcohol drinking, betel quid chewing, and cigarette smoking [24,25]. We used Illumina NextSeq 500 sequencing and bioinformatic studies to analyze the genes in the OECM-1 and OC-2 cells and whether they were regulated by CD47 ( Figure 1A). In addition, CD47 was overexpressed in OECM-1 and OC-2 cells ( Figure 1B). The RNA was extracted from the cells and the gene profile was analyzed by NGS. The flow chart shows the method of studying various gene expression changes with overexpressed CD47 in OECM-1 and OC-2 cells, through a gene expression array. The clusterProfiler tool was used to analyze bioinformatic approaches including gene ontology (GO) and DEGs. (B) Overexpression of CD47 was performed by transfecting cells with CD47 plasmid (1 µg), and CD47 expression was detected with qPCR and compared to the empty control. The data are presented as the mean ± standard deviation (SD) of three experiments. The data are presented as the mean ± SD of three experiments. * p < 0.05 vs. empty control; two-tailed Student's t-test. Figure 2A shows the differentially upregulated (right panel) and downregulated (left panel) gene expression in CD47-overexpressing OECM-1 and OC-2 oral cancer cell lines. Genes with-log10 (p value) and ≥5-fold and ≤0.2-fold changes were chosen for further analyses ( Figure 2B). Our results identified that the expression of 16 upregulated and 16 downregulated genes was found to be associated with CD47-overexpression in OECM-1 and OC-2 cells compared to the empty control ( Figure 2C) (Table 1).

Biological Process Analysis of Differentially Expressed Genes in CD47-Overexpressing Oral Squamous Cell Carcinoma
Simultaneously, we executed an intergroup comparison analysis of DEGs in each comparison, determined by RNA-seq by expectation-maximization (RSEM) and differential expression analysis at both gene and isoform level using RNA-seq data (EBSeq) (posterior probability of equal expression (PPEE) ≤ 0.05). In total, 14 genes were commonly identified as DEGs in OECM-1 and OC-2 oral cancer cells with CD47-overexpression ( Figure 3A). Furthermore, we used a heatmap to display the expression profiles of union sets ( Figure 3B,C).  In addition, we used IPA to analyze the DEGs regulated by CD47-overexpression in oral cancer for network analysis including molecular functions (MFs), biological processes (BPs), and cellular components (CCs). Then, we focused on the upregulated genes by CD47-overexpression in the OECM-1 and OC-2 cell lines. We found six genes (SDF2L1, HSPA5, CALMFR, HYOU1, SESN2, and PDIA4) involved in molecular functions, 13 genes (CHAC1, CTH, DDIT3, HERPUD1, HSPA5, SESN2, XBP1, SDF2L1, DERL3, ASNS, HYOU1, CALR, and PDIA4) involved in biological processes, and eight genes (HSPA5, SDF2L1, CALR, SLC3A2, SLC1A5, PDIA4, SLC1A4, and HYOU1) involved in cellular components ( Figure 4A). Furthermore, the results indicate that HSPA5, HYOU1, and PDIA4 genes appear in all three categories ( Figure 4B). To further explore the role of CD47, HSPA5, HYOU1, and PDIA4 genes in cancer, we used the human protein atlas (HPA) database [26] to analyze the expression of these target genes in cancer and their impact on the ability of cells to survive. We know that OSCC occupies a high proportion (90%) of head and neck squamous cell carcinomas [27]. Therefore, we selected head and neck squamous cell carcinoma samples from 499 patients for gene expression and survival analysis using the HPA database. The results showed that the fragments per kilobase per million (FPKM) of CD47, HSPA5, HYOU1, and PDIA4 was 22.19 ± 10.51, 224.2 ± 87.32, 25.35 ± 13.45, 84.83 ± 37.3, respectively. The increased expression of HSPA5 (cut off quantile 25%), HYOU1 (cut off quantile 75%), and PDIA4 (cut off quantile 25%) was significantly associated with survival rate in all 499 patients ( Figure 4C). In addition, HSPA5 was highly expressed in cancer and positively and significantly correlated with CD47 (R = 0.8432, p < 0.0001, Figure 4D).
To prove that HSPA5 was associated with CD47-overexpression in the OECM-1 and OC-2 cell lines, we used qPCR to quantify gene expression. We found that the expression of HSPA5 was induced by CD47-overexpression in the OECM-1 ( Figure 4E) and OC-2 ( Figure 4F) cell lines. We also verified the effect of HSPA5 overexpression in OSCC cell lines. HSPA5 plasmid was transfected into the cells and HSPA5 expression was analyzed by qPCR ( Figure 4G). In addition, cell viability assay revealed that HSPA5 increased cell growth in the OECM-1 and OC-2 cell lines ( Figure 4H,I). Next, we analyzed the CD47 expression in HSPA5-overexpressing OSCC cell lines. The results found that CD47 expression was induced by HSPA5-overexpression in both in cell lines ( Figure 4J). These data suggest that HSPA5, HYOU1 and PDIA4 were significantly associated with survival rate and HSPA5 is a suitable prognostic marker with CD47 in cancer survivability.

The DAVID Analyses of Differentially Expressed Genes following CD47-Overexpression
We also used the DAVID biological process analysis to analyze DEGs with CD47overexpressing OECM-1 and OC-2 cell lines from RNA sequencing. We concluded that the top 20 biological processes of the DEGs in the OECM-1 and OC-2 cell lines and the top five biological processes of these genes include responses to topologically incorrect proteins (13 genes), endoplasmic reticulum stress (14 genes), and unfolded proteins (12 genes), as well as cellular responses to topologically incorrect proteins (11 genes), endoplasmic reticulum unfolded protein responses (10 genes), and endoplasmic reticulum unfolded protein responses (10 genes) in OECM-1 cells. Meanwhile, in OC-2 cells, these genes include responses to viruses (41 genes), defense responses to other organisms (39 genes), defense responses to the virus (38 genes), the regulation of symbiosis, encompassing mutualism through parasitism (29 genes), and the regulation of viral processes (29 genes) ( Figure 5).

Discussion
Based on previous studies, we know that CD47 is abundantly expressed in many tumors, and it plays an important role in macrophages, affecting the development of cancer [28,29]. In the current study, we identified that the differentially expressed genes in OSCC cells are enriched in the biological functions of CD47, including regulating the response of unfolded proteins by NGS and systematic bioinformatic analysis. The results showed that the unfolded proteins HSPA5, HYOU1, and PDIA4 were associated with survivability in cancer, and HSPA5 was positively and significantly correlated with CD47, and induced by CD47 in OSCC cells lines.
Heat shock protein family A (Hsp70) member 5 (HSPA5) is a member of the molecular chaperones, which regulates the response of unfolded proteins [30]. HSPA5 promotes cancer progression, drug resistance, and metastasis and is a poor prognosis marker [31,32]. Hypoxia and nutrient deprivation in the environment activate the UPR and upregulate HSPA5 expression, leading to the rapid growth of solid tumors [33]. In addition, knockout mouse models have demonstrated that HSPA5 is required for cell proliferation and is critical for embryonic cell growth and survival [34]. In vitro and in vivo studies found that silencing of HSPA5 expression in cancer cells inhibits tumor cell metastasis [35]. Therefore, we believe that HSPA5 plays an important role in cancer progression.
Hypoxia-upregulated protein 1 (HYOU1) belongs to the heat shock protein 70 family, which is overexpressed in many tumors. Highly expressed HYOU1 induces tumor invasion, results in a poor prognosis, and inhibits apoptosis [36][37][38]. A previous study found that heat shock protein modulation primary macrophages are polarized into classic (M1) macrophages and alternative (M2) macrophages. The Hsp70 family might be strictly involved in the effector stages of macrophage activity [39]. Previous research found that anti-CD47 treatment enhanced the phenotype of macrophages converted to M1 subtype, and produced a higher phagocytosis rate through M1 macrophages [40]. In addition, CD47 secreted significant levels of IL-10 via M2 macrophages differentiation to promote cancer cell migration [41]. According to the above results, we believe that CD47 plays an important role in the regulation of HSP. CD47 has the same ability to promote cancer progression [42], drug resistance [43], poor prognosis marker [44] and macrophage polarization [40] as HSP. In addition, CD47 as a transcription factor [45] can mediate gene expression. Our data also found that overexpression of CD47 induced the expression of HSPA5, and overexpression of HSPA5 induced the expression of CD47 in the OECM-1 and OC-2 cells. A previous study also demonstrated that inhibition of glucose-regulated protein 78 (GRP78), a member of the HSP70 family, downregulated CD47 expression in tumor cells [46]. Therefore, we believe that CD47 and HSPA5 have a feedback loop in OSCC cell lines, and suggest that CD47 mediates cancer progression through the HSPA5 signaling pathway in oral squamous cell carcinoma. In the future, we will further explore the mechanisms and regulatory capabilities of CD47 and HSPA5.
PDIA4, as a member of the protein disulfide isomerases (PDI) family, is a key factor for protein folding in the endoplasmic reticulum. A previous study found that PDIA4 participates in tumor progression by influencing cell apoptosis and DNA repair mechanisms [47]. Meanwhile, a recent study pointed out that PDIA4 may be regarded as a biomarker for the treatment of ovarian cancer [48]. Interestingly, PDIA4 has similar properties to HSP proteins, and both have the function of affecting protein folding. Protein folding is important for CD47 evasion of macrophage phagocytosis [49]. However, the detailed mechanism is unclear at present.

Conclusions
These results provide important insights into possible new diagnostic genes for OSCCtargeting CD47. Although the mechanism by which CD47 regulates unfolded proteins is still unclear, the detailed signaling pathway will be discussed in more depth in the future.