A Pancancer Study of PIEZO1 as a Prognosis and Immune Biomarker of Human Tumors

PIEZO1, a mechanosensitive ion channel protein, has been identified in the correlation between several cancers. However, the systematic pancancer study of PIEZO1 still lacks. We examined PIEZO1 across thirty-three types of cancers to explore its role in prognosis and immunological function for the first time. Based on the open databases TCGA, GTEx and CPTAC, PIEZO1 has been demonstrated to be differentially expressed in most cancers compared to adjacent normal tissues. The distinct correlation between PIEZO1 and prognosis of tumor patients was explored by GEPIA2. Genetic alteration of PIEZO1 in the TCGA tumors showed that mutation is the alteration which is linked to OS, DSS, DFS and PFS in some tumors. Alterations of protein phosphorylation levels were detected in some cancers based on the CPTAC dataset. PIEZO1 expression was linked with immune cell infiltration, such as endothelial cell and cancer-associated fibroblast. Finally, KEGG and GO enrichment analyses were applied to investigate the molecular mechanism of PIEZO1. Our first pancancer analysis illustrated the roles of PIEZO1 in different types of tumors.


Introduction
Cancer has the highest worldwide mortality rate of any disease [1]. Malignant tumors pose a severe threat to human health. Despite advances in diagnosis and treatment, cancers still cause the extremely high morbidity and mortality. Conducting a pancancer study of some certain genes and exploring its role in some malignant tumors and their corresponding cancers are highly warranted.
Piezo-type mechanosensitive ion channel component 1 (PIEZO1) characterized as the ion-conduction subunit of mechanically activated ion channel can initiate the intracellular Ca 2+ response [2]. PIEZO1 is widely distributed and expressed in various human organs and tissues, such as the cardiovascular system [3], brains [4], lungs [5], gastrointestinal tract [6] and bladder [7]. Moreover, recent relevant literature suggest that such expression of PIEZO1 could be different in various tumors, for example, the gastrointestinal system [8], urinary system [9], respiratory system [10] and reproductive system tumors [11]. e increasing evidences indicate that tumor immune microenvironment (TME) is strongly related to carcinogenesis and cancer progression [12]. e biomarkers associated with TEM indicating prognosis and survival may be crucial to immunotherapy.
It is commonly known that calcium signaling associated with immune cells plays a significant role in the expression of transcription factors and enzymes that regulate cancer development [13][14][15]. Numerous literature report that PIEZO1, recognized as a mechanical stress sensor in some immune cells, crucially relates to immune regulation [16][17][18]. e importance of PIEZO1 activity in immune cells has been confirmed which implies the potential contribution of PIEZO1 to cancer immunotherapies.
Despite above, there is no pancancer study exploring the correlation between PIEZO1 and different tumors, since most research studies involving PIEZO1 in tumors focus on one certain cancer. In this work, we adopt a diversity of databases such as e Cancer Genome Atlas (TCGA), Genotype Tissue-Expression (GTEx), cBioPortal, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to assess the expression levels of PIEZO1 and the link with survival in various tumors. We further assessed the potential correlations between PIEZO1 expression levels and genetic alteration, protein phosphorylation, immune infiltration and pathway enrichment in 33 types of cancers. Our results showed that PIEZO1 can be used to predict prognosis of different cancers, and it affected tumor infiltrating immune cells, which rises a pivotal role in tumor immunity.

Gene Expression Level Analysis.
e mRNA levels of PIEZO1 in a variety of cancers were analyzed by Tumor Immune Estimation Resource, version 2 (TIMER2). e differential expression levels of PIEZO1 were observed between the cancerous region and normal tissue for various types of tumors. e range was set as follows: log2 fold change (log2 FC) ≥ 1 or ≤ −1 and p value ≤ 0.05. e Gene Expression Profiling Interactive Analysis, version 2 (GEPIA2) was employed to analyze the PIEZO1 mRNA levels in some certain tumors with no corresponding normal tissues. Additionally, violin plots were applied to reveal the relationship between PIEZO1 and pathological stages of cancers by using the "Stage Plot" of GEPIA2.
UALCAN, an interactive, user-friendly, and synthetic online platform to analyze open-source TCGA data, was taken to analyze expression levels of protein with the Confirmatory/Discovery tool of the Clinical Proteomic Tumor Analysis Consortium (CPTAC). For its part, phosphoprotein levels of PIEZO1 were evaluated between tumor and normal tissues by CPTAC analysis.

Prognosis and Survival Analysis.
e "Survival Map" was employed to detect the overall survival (OS) and the diseasefree survival (DFS) of PIEZO1 in all the tumors in TCGA. e values for dividing groups with high and low expressions were defined as cutoff-high (50%) and cutoff-low (50%) values, respectively. Survival plots were obtained from the "survival analysis" module of GEPIA2. Among these, two curves were compared by the log-rank test. Meanwhile, univariate Cox regression analysis of PIEZO1 was performed in tumors where PIEZO1 was an independent prognostic indicator.

Genetic Alterations
Analysis. Data about PIEZO1 genetic mutation type, alteration frequency, mutated sites, and copy number were retrieved through cBioPortal [19]. e information of OS, DFS, and progression-free survival (PFS) differences between with and without PIEZO1 alteration in different tumors was obtained from the "Comparison" module. Kaplan-Meier plots with the log-rank p value were generated.

Gene Set Enrichment Analysis.
To study the potential function of PIEZO1 in pancancer, we divided the samples into high and low-expressed groups based on the PIEZO1 expression in each cancer type, consisting of the top 30% and bottom 30%. en, the gene set enrichment analysis (GSEA) (http://www.gsea-msigdb.org/gsea/index.jsp) was performed using the "clusterProfiler" R package. GSEA gene sets (H, C2, and C5 collections) were downloaded from the Molecular Signatures Database (v7.5.1). e FDR-adjusted p < 0.05 was considered statistically significant.

Immune Cell Infiltration Analysis.
We utilized the "Immune-Gene" unit of TIMER2 to discover the link between the levels of PIEZO1 expression and the infiltration of immune cells, such as endothelial cells, and cancer-associated fibroblasts. For this purpose, we evaluated different cancers.
e algorithms QUANTISEQ, XCELL, MCP-COUNTER, EPIC, TIMER, CIBERSORT, and CIBERSORT-ABS were applied to make immune infiltration estimations. P values and partial correlation (cor) values were obtained by using Spearman's rank correlation test, purity-adjusted. e heatmap and scatter plot were displayed as the outcome.

PIEZO1-Related Partners Enrichment Analysis.
PIEZO1-binding proteins were searched by the STRING website, using the query of "PIEZO1." en, we used Pearson correlation to analyze the first 100 PIEZO1-targeting genes which were based on the differential expression records of TCGA tumors and normal tissues in GEPIA2. e heatmap consisted of top 5 genes, which contains the partial correlation index, and p value was obtained using the "Gene_Corr" component of TIMER2 by the method of the purity-adjusted Spearman's rank correlation test. A Venn diagram viewer allowed the observation of the PIEZO1binding and interacting genes. Two sets of data were combined to perform KEGG pathway analysis.

Human Samples.
Human samples were obtained from subjects undergoing tumorectomy in the First Affiliated Hospital of Nanjing Medical University. Ethics approval was gained through the First Affiliated Hospital of Nanjing Medical University. e cancer tissues were harvested from the region of tumor. e adjacent normal tissues were collected more than 5 cm from cancerous tissues. Eight paired lung tumor tissues, stomach tumor tissues, and colon tumor tissues, as well as corresponding adjacent normal tissues were collected.

Immunohistochemical
Staining. All human sample slides were baked at 65°C for 2 hours and then were dewaxed and rehydrated. Slides were placed in a repair box filled with EDTA antigen repair buffer (pH 9.0, Servicebio, China, G1203), and antigen repair was conducted for 20 min at 100°C. After cooking, the slides were naturally cooled to room temperature. en, hydrogen peroxide was used for blocking endogenous peroxidase activity. en, blocking was performed with 10% goat serum for 1 hour at room temperature and incubated with primary antibodies PIEZO1 (Abcam, UK, ab128245, 1 : 200) diluted in blocking solution overnight at 4°C. e slides were incubated with HRP secondary antibody for 2 hours at room temperature. Finally, sections were detected by DAB staining. After stained with DAB, images of samples in the slides can be taken viewing using the microscope. Slides were washed three times with PBS between each step.

2.10.
Immunofluorescence. Cells were grown in 48-well plates. When grown to 70% confluence, the cells were transfected with PIEZO1 small interfering RNAs (siRNAs) (Applied GenePharma, China) using Lipofectamine 2000 (Invitrogen Life Technologies, USA, 11668-019). After 24 hours, cells were fixed with 4% paraformaldehyde for 20 min and the cell membrane was disrupted with 0.2% Triton X-100 (PBS configuration) for 15 min. Ki67 staining was performed by adding the primary antibody anti-Ki67 (Abcam, UK ab16667, 1 : 300) and incubated overnight at 4°C. e next day, cells were incubated with fluorescent secondary antibody for 2 hours at room temperature. After stained with DAPI for 10 minutes, the cells can be viewed using an inverted microscope. Between each step, cells were washed three times with PBS.
2.11. Scratch Assay. Cells were grown in 6-well plates. When grown to 100% confluence, a single scrape was made in the confluent monolayer using a 200 μl pipette tip. en, cells were cultured in DMEM with 2% FBS for 24 h after washing with PBS. e migration rate was determined by the scratch area at different time points and analyzed by the Image J software.

Statistical
Analysis. Data are shown as the mean-± standard error of mean (SEM). e GraphPad Prism software v5.0 (GraphPad Software, USA) was performed for statistical analyses. e unpaired Student's t-test was used for statistical analyses between two groups.

PIEZO1 Expression Analysis.
We first assessed the PIEZO1 expression levels in different cancers. e abbervation and full names of the 33 tumor types are given in Table 1 (CESC), and pancreatic adenocarcinoma (PAAD) showed no differential expression from tumor-adjacent tissues (Figure 1(a)). We further explored whether PIEZO1 was differentially expressed between tumor and normal tissues in the absence of data of normal tissue in TCGA. We found that cholangiocarcinoma (CHOL) (p < 0.05), GBM (p < 0.05), and PAAD (p < 0.05) showed higher expression in tumor tissues. No significant differences were detected between CESC, PCPG, PRAD, and their corresponding normal samples (Figure 1(b)).
We also analyzed the relationship between the PIEZO1 expression level and tumor pathological staging by GEPIA2. We found that PIEZO1 expression was changed in different pathological stages of a few tumor types, including KIRC (p � 0.0202), PAAD (p � 0.0063) and STAD (p � 0.0469) (Figure 1(d)).

PIEZO1 Expression in Tumor and Adjacent Nontumor
Tissues.
en, we detected the expression of PIEZO1 in LUAD, STAD and COAD, as well as their corresponding adjacent normal tissues by the method of IHC. Compared to adjacent normal tissues, the expression level of PIEZO1 was decreased in LUAD (p � 0.0389) (Figure 2(a)) and increased in STAD (p � 0.0596) (Figure 2(b)). ere was no significant difference between COAD and its adjacent normal tissue (Figure 2(c)).

Survival
Analysis. Next, we studied how PIEZO1 expression correlates with prognosis and used TCGA and GEO datasets to assess OS and DFS values. Samples were divided into high-expression and low-expression groups according to the PIEZO1 expression level. e results demonstrated that among cases with KIRC (p � 0.0006), brain lower grade glioma (LGG) (p � 0.0073), LIHC (p � 0.012), LUAD (p � 0.025) and LUSC (p � 0.048), those with low level of PIEZO1 had longer survival time (Figure 3(a)). e data DFS analysis revealed that high expression of PIEZO1 is linked to

Genetic Alteration of PIEZO1 across Different Tumors.
e oncogenesis and progression of cancers are closely associated with genomic mutations [20]. We analyzed types and sites of genetic mutation of PIEZO1 in different cancers within the TCGA database. e frequency of PIEZO1 mutation (>10%) is the highest in patients with UCEC ( Figure 4(a)). In addition, the type, sites and number of the PIEZO1 genetic alteration were also discovered. e number of PIEZO1 genetic alteration is 185, which consists of 155 missense, 13 truncating, 2 inframe, 10 splice and 5 SV/fusion. R1943Q/W alteration was found in 2 cases of UCEC and 1 case of COAD ( Figure 4(b)). In addition, we found the relationship between genetic mutation of PIEZO1 and survival prognosis in different types of tumors. Cases of LIHC, with PIEZO1 alteration, showed good prognosis in OS (p � 0.009) and disease-specific survival (DSS) (p � 0.006). Also, cases of THCA, with PIEZO1 alteration, had good prognosis in OS (p � 0.031), DSS (p � 0.0001) and progression-free survival (PFS) (p � 0.001), while alteration of PIEZO1 in UCEC was associated with worse DSS (p � 0.032) and PFS (p � 0.028) compared with no alteration ( Figure 4(c)).
3.6. GSEA of PIEZO1 in Pancancer. GESA analysis was performed based on GO, KEGG, REACTOME, and HALLMARK ( Figure 6). In almost all cancers, we found that PIEZO1 expression was remarkably related to immune-related pathways, such as innate immune response, adaptive immune response, inflammatory response and some inflammatory cells, such as neutrophils, lymphocyte and leukocyte. In addition, ion transport and cation homeostasis were also related to PIEZO1 in pancancers. Activation of PIEZO1 can result in calcium influx, transducing mechanical forces into biochemical responses [21]. Calcium is a ubiquitous second messenger that is responsible for a diverse host of cellular functions that are necessary for successful cancer metastasis [22]. We also found that epithelial-mesenchymal transition (EMT) hallmark was significantly enriched in high-PIEZO1 subgroups in most cancers. EMT enhances migration and invasive potential which lead to increased cancer metastasis [23]. ese data revealed a potential association between PIEZO1 expression and immune activation in the tumor microenvironment (TME). In addition, the top 20 related pathways of GSEA were presented in the form of a mountain map (Supplementary Figure S1)

PIEZO1-Related Partners Enrichment Analysis.
To reveal the mechanism of the PIEZO1 molecular that links to the oncogenesis of tumors, PIEZO1-binding proteins and pathway enrichment analysis of PIEZO1-correlated genes were performed. 50 PIEZO1-binding proteins by utilizing the STRING tool constructed protein-protein interaction (PPI) networks (Figure 8(a)). en, we used the GEPIA2 tool to explore the top 100 PIEZO1-associating genes. As assessed in our data, the expression level of PIEZO1 was  (Figure 8(b)). e heatmap data revealed that the expression level of PIEZO1 was positively correlated with the above five genes in various cancers (Figure 8(c)). Transient receptor potential melastatin 4 (TRPM4), the only one common member in the intersection analysis of the PIEZO1-binding and correlated genes, was showed by the mean of Venn diagram (Figure 8(d)). en, KEGG and GO enrichment analyses were performed. e KEGG data suggested that "Inflammatory mediator regulation of TRP channels" seem to be associated with the role of PIEZO1 on tumor oncogenesis (Figure 8(e)). e results further showed that these genes are correlated with some activities, such as calcium ion transmembrane transporter activity calcium channel activity, divalent inorganic cation transmembrane transporter activity, cation channel activity and ion channel activity by analyzing GO enrichment data (Figure 8(f )).

PIEZO1 Deletion Reduced the Cell Proliferative
Capability. To further verify the function of PIEZO1, we used PIEZO1 siRNA to knockdown the expression of PIEZO1 in CFPAF cell line. Immunofluorescence staining of ki67 was performed for detecting cell proliferation. e data showed that knockdown of PIEZO1 markedly reduced pancreatic cancer cells proliferation ability (p � 0.0064) (Figure 9(a)). e cell migration was assessed by scratch assay.
e results showed that reduction of PIEZO1 can lower cell migration capacity than their control groups (p � 0.0420) (Figure 9(b)). In addition, deletion of PIEZO1 can downregulate the expression of TRPM4 (p � 0.0018) (Figure 9(c)).
is is not at par with our current findings. We further discovered that overexpression of PIEZO1 was correlated with worse DFS for patients with ACC, CHOL, GBM, LGG, LIHC and LUSC. Increased expression of PIEZO1 generally predicted poor OS and DFS, which indicated that PIEZO1 is a useful molecular as a prognostic biomarker for patients with tumors.
Cancer is the result of one or more genetic alterations. e way the gene has been altered may determine the prognosis of one cancer [32]. ese genetic alterations include gene mutation, structural variants, amplification, deep deletion and multiple alterations. PIEZO1 mostly underwent mutation alteration which is linked to OS, DSS, DFS and PFS in some tumors, such as LIHC, THCA and UCEC. Modification of the PIEZO1 gene could be a promising method to cure cancers. In addition, the PIEZO1 protein phosphorylation level was different between tumor and normal tissues, while different tumors shared diverse phosphorylation sites. ese results hint that it is important to clarify the phosphorylation sites of a certain cancer type. e tumor immune microenvironment (TME) is an integral component of tumor biology [33]. In addition to cancer cells, tumors exhibit another dimension of complexity. ey contain a repertoire of recruited, ostensibly normal cells, including smooth muscle cells, endothelial cells, fibroblasts, lymphocytes, macrophages and adipocytes, which significantly contributes to tumor progression [34]. Calcium signaling associated with immune cells plays a key role in regulating cancer development [13][14][15]. PIEZO1, a mechanosensitive ion channel protein, has identified its effect in the immune system by linking mechanical forces with immune regulation [16][17][18]. is study evidenced the relationship between the PIEZO1 expression level and the immune infiltration of endothelial cells and cancer-associated fibroblasts in most tumors. e confirmed role of PIEZO1 activity in immune cells raises the possibilities of utilizing PIEZO1 for cancer immunotherapies.
KEGG and GO analyses have revealed the molecular mechanisms of PIEZO1. e results showed that PIEZO1related genes including APRT, MFSD10, AGTRAP, MICALL2, and SYTL1 were mainly enriched in "Inflammatory mediator regulation of TRP channels." Our enrichment analyses demonstrated that PIEZO1 may exert its tumorigenic effects by calcium ion transmembrane transporter activity, calcium channel activity, divalent inorganic cation transmembrane transporter activity, cation channel activity, and ion channel activity.
Transient receptor potential melastatin 4 (TRPM4), the only one common member in the intersection analysis of the PIEZO1-binding and correlated genes, was shown by the mean of Venn diagram. TRPM4 is a nonselective cation channel conducting monovalent ions [35]. Intracellular Ca 2+ directly activates TRPM4 [36]. e protein expression levels of TRPM4 were reported to be increased in various tumors, such as colorectal cancer [37], large B cell lymphoma [38], prostate cancer [39], and so on. e TRPM4 channel also involved in regulating cancer cells to mesenchymal transition, migration and invasion [40]. PIEZO1 enables cells to  sense various mechanical forces, which induce the PIEZO1 channel from a closed state turn to an open state. e PIEZO1 channel can allow Ca 2+ to flow when it is open [41]. However, no study has yet tested the associations between PIEZO1 and TRPM4. Whether the flow of Ca 2+ caused by PIEZO1 channel affects TRPM4 is unknown. So, we used siRNA to knockdown PIEZO1 and detected the transcriptional level of TRPM4. e result shown that TRPM4 can be downregulated by the deletion of PIEZO1. Of cause, this is only a preliminary exploration. e relationship between the two needs further study.
In conclusion, the statistical correlations of PIEZO1 expression with clinical prognosis, protein phosphorylation and immune cell infiltration across multiple tumors were revealed in our first pancancer analysis in detail, which may promote better understanding of the effect of PIEZO1 in cancerogenesis and provide a novel perspective for cancer immunotherapy.

Conflicts of Interest
e authors declare that they have no conflicts of interest. Immunofluorescence of ki67 staining was performed, n � 3/group. Original magnification x200. Bar � 100 μm. Data are represented as mean ± SD. (b) Scratch assay used to assess migration capability, n � 3/group. Original magnification x100. Bar � 100 μm. Data are represented as mean ± SD. (c) e mRNA transcript level of TRPM4 detected by real-time PCR after deletion of PIEZO1 in CFPAC cell line, n � 3/group. e bar chart shown represents the mean ± SD. * P < 0.05, * * p < 0.01 versus the negative control group by Student's t-test.