A pan-cancer analysis of the FAT1 in human tumors

FAT atypical cadherin 1 (FAT1) is one of the most mutagenic genes in tumors, and several critical studies have revealed its role in tumors, although no pan-cancer studies are currently available. Therefore, we explored the potential oncogenic role of FAT1 in 33 tumors based on The Cancer Genome Atlas and Gene Expression Omibus datasets. We found that FAT1 was strongly expressed in most tumors and significantly correlated with their prognosis. Additionally, we analyzed the association of FAT1 with tumors from multiple perspectives, including single-cell sequencing, mutations, high tumor mutational burden, microsatellite instability, immune cell infiltration, and immune microenvironment. Our first pan-cancer study provided a relatively comprehensive understanding of the oncogenic role of FAT1 in tumors.

Survival prognosis analysis. We searched the relationship between FAT1 expression and pan-cancer patients prognosis using Sangerbox, which is based on the TCGA TARGET GTEx dataset. The TCGA prognosis study published in Cell journal 11 provided a high-quality prognostic data set of TCGA. TARGET follow-up data were obtained from UCSC as supplementary data. Additionally, the samples with no expression level and followup duration of less than 30 days were excluded, and the expression data were transformed using log2 (x + 0.001). Additionally, the cancers with less than ten samples were also excluded, and the pan-cancer expression data Table 1. Pan-cancer samples in TCGA database and tumor abbreviations. ↑ high expression; ↓ low expression; *P < 0.05; **P < 0.01; ***P < 0.001. Protein expression and protein-protein interaction (PPI). The Human Protein Atlas (https:// www. prote inatl as. org/) database was searched for information about FAT1 protein expression in 17 different types of tumor tissues and paracancerous tissues that had been validated by immunohistochemistry. We also explored the expression of FAT1 in different kinds of single cells. STRING (https:// string-db. org/) 17 is a known and predicted PPI database that was used to analyze the PPI network of FAT1 in this study. The cytoHubba plugin of Cytoscape was used to analyze the degree of each protein, and then the PPI network was drawn. We also searched the pathways and genes that interact with FAT1 in UCSC Genome Browser. The Gene Ontology and KEGG Pathway were analyzed in Metascape (https:// metas cape. org/) based on the interacting proteins in the PPI network.
Protein phosphorylation. PhosphoNET (http:// www. phosp honet. ca/) is the world's largest online repository of general and predictive information about human phosphorylation sites, their evolutionary conservation, the identity of protein kinases that may target these sites, and related phosphate sites. The verified and predicted phosphorylation sites were obtained by searching FAT1. The most significant phosphorylation sites and kinases involved in phosphorylation were predicted by calculating the Kinase Predictor V2 Score by the Kinexus P-Site Prediction algorithm.
Correlation analysis of marker genes. Studies 22,23 , and autophagy 24,25 . The correlation between the mRNA expression levels of FAT1 and these marker genes in pan-cancer was obtained by Spearman correlation analysis with the limma package. The results can predict the potential benefits of FAT1 in these processes and lay a bioinformatics foundation for further studying FAT1 in malignant tumors.

Data analysis.
Wilcox and T-tests were used for FAT1 differential expression analysis. Cox proportional hazard regression model and Kaplan-Meier methods were used for survival analysis, and Spearman's analysis www.nature.com/scientificreports/ was used for studying correlations. Except for special instructions, P < 0.05 demonstrated that the difference was statistically significant. *P < 0.05. **P < 0.01. ***P < 0.001.

FAT1 expression.
According to the subcellular locations from COMPARTMENTS (https:// compa rtmen ts. jense nlab. org/), FAT1 protein is mainly expressed in the nucleus, plasma membrane, and extracellular regions of the cell [26][27][28] (Fig. 1A). We found that FAT1 was highly expressed in pan-cancer, except for low expression in some tumors (  www.nature.com/scientificreports/ expression and multi-database analysis. The analysis of FAT1 expression in different tumor stages revealed its significant difference in various tumors like ACC, COAD, ESCA, KICH, KIRP, LUAD, MESO, SKCM, STAD, and THCA ( Fig. 2A). Interestingly, there was no significant difference in FAT1 expression between SKCM tumor and normal tissues. FAT1 protein expression was significantly higher in clear cell renal cell carcinoma, colon cancer, and uterine corpus endometrial carcinoma and was significantly lower in breast cancer. There was no significant difference in FAT1 protein expression in lung adenocarcinoma and ovarian cancer (Fig. 2B). The noteworthy point is that, contrary to our findings, Nantana Kwaepila et al. 29 reported that FAT1 in breast cancer immunohistological expression data displayed high expression levels in human tumor samples, possibly due to sample specificity issues. High-or medium-intensity staining (Fig. 2C) was observed in the tumor tissues of glioma, head and neck, thyroid, colorectal, endometrial, liver, urothelial, lung, pancreatic cancers, and lymphomas, indicating that the FAT1 protein was significantly overexpressed in the tumor tissues of these diseases. We also found that FAT1 was highly expressed in many glandular epithelial cells, squamous epithelial cells, and specialized epithelial cells based on the summary of single-cell expression of FAT1. It was also highly expressed in astrocytes, smooth muscle cells, and other cells (Fig. 2D). Mutation analysis. FAT1 encoding gene is located at q35.2 of chromosome 4 (Fig. 4A). We analyzed the mutational information of FAT1 in pan-cancer using cBioportal and found that FAT1 was mutated in more than 10% of 10 tumors, including HNSC and UCEC (Fig. 4B). Missense and truncating were the two most common mutations of FAT1, which primarily affects extracellular and cytoplasmic regions. The PTMs results depicted that FAT1 proteins are primarily phosphorylated and ubiquitinated (Fig. 4C). The mutation survival analysis of the top four tumors with more than 15% mutation rate demonstrated that FAT1 mutations in UCEC predicted better prognosis for tumor patients, but FAT1 mutations in HNSC predicted poor survival (Fig. 4D). The TMB and immunotherapy dataset revealed that FAT1 mutations were observed in 10% of the samples, which is consistent with the findings from the pan-cancer analysis (Fig. 4E). Notably, the OS indicated that patients with FAT1 mutations had a better prognosis (Fig. 4F), suggesting that FAT1 mutations may be used as an indicator of patient's prognosis.

Survival analysis.
Immunocomprehensive analysis. Tumor therapy has advanced significantly owing to immunotherapy that targets immunological checkpoints 30 . It has been demonstrated that MSI and TMB 31 are biomarkers for predicting tumor immunotherapy and are crucial in directing the systemic treatment of tumors 32 . The analysis of sequencing data demonstrated significant correlations between FAT1 and MSI of several tumors, positive correlations in LUAD, LUSC, and TGCT, and negative correlations in BLCA, DLBC, HNSC, SKCM, and THCA (Fig. 5A). Additionally, FAT1 was also significantly correlated with TMB of many tumors; positively correlated in ACC, KIRC, LAML, PAAD, READ, STAD, and THYM, and negatively correlated in LGG and LUAD (Fig. 5B). The TME has a significant role in tumor occurrence and development, as well as in the evaluation of invasion, metastasis, and prognosis, thus pointing out the future direction for more effective immunotherapy [33][34][35] . Our study confirmed a significant correlation between the FAT1 expression and TME components in various tumors. Analysis was done to explore the correlation between the FAT1 expression and the Estimate score of the matrix score and immune score. The results revealed that FAT1 was positively correlated with BRCA, DLBC, LGG, OV, and PCPG but negatively correlated with GBM, KIRP, PAAD, PRAD, SARC, SKCM, STES, TGCT, THCA, and THYM (Table 2). Immune cells in the TME are essential for tumor regulation 36 . This study analyzed the relationship between FAT1 expression and 22 different types of immune cell infiltration in pan-cancer. The results demonstrated that FAT1 was significantly correlated with one or more immune cell infiltration in all tumor types except CHOL. There was a significant positive correlation with T-cells-CD4-memory-resting and macrophages-M1 of THYM and a significant negative correlation with Tregs (Fig. 5C).

PPI and functional enrichment analysis. PPI is significant in a tumor's molecular biological regulation
because of its versatility, specificity, and adaptability 17,37 . The PPI results based on Cytoscape analysis demonstrated that there was a significant interaction between MYC, EGFR, PIK3CA, TP53, TNF, RELA, NFKBIA, TJP1, JUN, PPARA, and FAT1 (Fig. 6A), and the degree of these interacting proteins in the network was more than 15. Second, PPI results from the UCSC tool depicted that FAT1 interacts with many transcription factors (ADNP, CBX5) and essential proteins such as EIF5B histones, AGL, and HSPA1A (Fig. 6B). The results of Gene Ontology and KEGG Pathway enrichment analysis (https:// www. kegg. jp/ kegg/ kegg1. html) 38 based on the proteins in the PPI network revealed that it was significantly enriched in cancer pathways and particularly related to transcription factor binding, intracellular receptor signaling pathway, regulation of epithelial cell proliferation, negative regulation of cell differentiation, epithelial cell development, and other pathways (Fig. 6C,D).

FAT1 protein phosphorylation.
Protein phosphorylation is recognized as the primary method of protein PTMs, and it is an effective way to regulate protein function under the catalysis of protein kinases. It is a fundamental cause of intracellular processes such as cell growth and development, signal transduction, and metabolism 39,40 . A database search revealed that 30 FAT1 proteins had been phosphorylated (Table 2). Second, data scoring predicted 45 sites most likely to be phosphorylated, and the protein kinases (Table S1)  www.nature.com/scientificreports/ to participate in the phosphorylation process were also predicted. The phosphorylation of the FAT1 protein is now better understood.

Correlation analysis of tumor marker genes. Groundbreaking research by Ievgenia Pastushenko et al. 1
demonstrated that FAT1 mutation promotes tumor initiation, progression, invasion, stemness, and metastasis by inducing hybrid EMT states in mouse and human skin squamous cell carcinoma. Chitrangda Srivastava et al. 41 found that FAT1 is a novel regulator of EMT in anoxic GBM, which suggested that it may be a viable therapeutic candidate. According to Xiaoling Hu et al. 42 , FAT1 destroys the MAPK/ERK pathway and participates in the EMT process of esophageal squamous cell carcinoma. Our analysis of the correlation between FAT1 and recognized EMT transcription factors and receptors demonstrated that FAT1 is associated with EMT marker genes in almost all tumors (Fig. 7), indicating that FAT1 may be involved in the EMT process in all tumors.
Mitogen-activated protein kinase 13 490 - Casein kinase II, alpha chain 115 - Casein kinase II, alpha' chain 103 20,068,231 Serine-threonineprotein kinase Nek4 291 20,068,231 Proto-oncogene serine-threonineprotein kinase Pim-1 202 -    www.nature.com/scientificreports/ presentation, cell migration, cell differentiation, and tumor invasion. Different immune cells are also involved in tumor regulation 46 . Our results depicted that FAT1 was significantly associated with one or more exosome markers in all tumors (Fig. S2). Immune cell marker genes analysis illustrated that FAT1 was significantly associated with a variety of immune cell markers in BRCA, DLBC, GBM, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, OV,  PCPG, PRAD, SARC, SKCM, STAD, TGCT, THCA, THYM, UCEC, and other tumors. Both DNA and RNA can be methylated. The family of DNA methyltransferases play a central role in epigenetic regulation 23 . RNA methylation modification accounts for more than 60% of all RNA modifications, of which m6A is an essential methylation modification in which writers, erasers, and readers play different roles, including regulation of tissue development, circadian rhythm, DNA damage response, sex determination, T cell homeostasis and tumorigenesis 22 . The correlation between FAT1 expression and methylation marker genes is displayed in Fig. S3. Our results demonstrated a significant correlation between FAT1 and multiple methylation marker genes in different tumors except for CHOL, ESCA, LAML, and UCS.
Hypoxia has recently been confirmed to play an essential role in tumor progression 19 . Tumor hypoxia is associated with increased invasion and metastasis and shows typical driving mutation characteristics. The correlation between FAT1 expression and tumor hypoxia marker genes is revealed in Fig. S4. The findings revealed a significant correlation between FAT1 and multiple hypoxia-related mutant genes in almost all tumors except for LAML.
The dynamic equilibrium of cells, tissues, and organisms depends heavily on the autophagy pathway, which is mediated by evolutionarily conserved ATGs 24 . ATGs mutations play an important role in various diseases, including cancer 25 . The correlation between FAT1 expression and autophagy-related marker genes is illustrated in Fig. S5.

Discussion
FAT atypical cadherin 1 (FAT1) encodes protocadherin, one of the most frequently mutated genes in human cancers. Studies conducted over the past 20 years have demonstrated that FAT1 regulates multiple signaling pathways, including Wnt/β-catenin, Hippo and MAPK/ ERK, to affect the proliferation, migration, and invasion of various tumor cells 1,47 . In this study, FAT1 was found to have a mutation rate of > 10% in more than ten tumors. It has not been established whether FAT1 plays a role in the pathogenesis of different tumors through some common molecular mechanisms; therefore, we performed a comprehensive pan-cancer analysis of FAT1 based on data in TCGA, CPTAC, and GEO from an overall tumor perspective, focusing on gene expression, gene alterations, protein phosphorylation, and prognosis.
FAT1 is highly expressed in most tumors. However, the prognostic survival analysis of the FAT1 gene data suggested different conclusions for different tumors. In the present study, we combined different prognostic analysis methods and data. Finally, we revealed that high FAT1 expression predicted a worse prognosis in all tumors except STAD, STES, ESCA, KIRP, UCS, UCEC, READ, and TGCT. This is according to certain recent studies 47 .
According to previous studies, FAT1 is mutated in a variety of tumors. This study analyzed the relationship between FAT1 mutations and patient survival in the four tumors with the highest mutation rates. We revealed that FAT1 mutations in UCEC predicted a better prognosis for tumor patients, whereas they predicted poor survival in HNSC patients. According to the OS, patients with FAT1 mutation had a better prognosis, as revealed by the mutation data from TMB and immunotherapy (MSKCC, NatGenet 2019) 13 . Furthermore, we found that FAT1 was significantly correlated with TMB in ACC, KIRC, LAML, PAAD, READ, STAD, THYM, LGG, and LUAD tumors. This result suggested that FAT1 mutations may serve as immunotherapeutic markers for these tumors and may be useful for guiding novel immunotherapies.
The PTMs regulate the function of most eukaryotic proteins 48 , and protein phosphorylation is very closely associated with tumors 49 . In this study, the database statistically predicted the most phosphorylated sites and protein kinases of FAT1 protein to explore potential associations between FAT1 and tumors. Recent studies have demonstrated that deletion of FAT1 in cutaneous squamous cell carcinoma accelerates tumor initiation, malignant progression and promotes the EMT phenotype 1 . We found a significant association between FAT1 mRNA level and EMT phenotype-related marker genes in almost all tumor types, implying that FAT1 may play a role in the EMT phenotype in other tumors. Studies analyzing the correlation between FAT1 and markers genes such as exosomal and immune, methylation, hypoxia, and autophagy have produced a number of additional useful findings. These phenomena need to be confirmed by further research.
For the first time, this study provided evidence for a potential correlation between FAT1 expression and MSI or TMB in all TCGA tumors. We also integrated information on FAT1 binding components and FAT1 expressionrelated genes in all tumors. We performed enrichment analyses to identify the potential effects of FAT1 in tumor pathways, reverse regulation of cell differentiation, transcription factor binding, and epithelial cell development.
FAT1 has been reported to have immunotherapeutic potential in several tumors 50,51 . We applied the immunological deconvolution method to observe the statistical correlation between FAT1 expression and the infiltration level of 22 immune cells in pan-cancer. FAT1 may play a function in the immunotherapy of THYM as a result of the highly substantial association between THYM and many immune cells in THYM. Additionally, we found that FAT1 expression was significantly and positively correlated with immune scores of BRCA, DLBC, LGG, OV, and PCPG based on the calculated stromal, immune, and ESTIMATE scores for each patient in the TCGA pan-cancer dataset of 9530 tumor samples from a total of 39 tumor types. These findings can provide research directions for tumor immunotherapy based on FAT1.
Our first pan-cancer analysis of FAT1 demonstrated a statistical correlation between FAT1 expression and clinical prognosis, protein phosphorylation, MSI, TMB, TME, and immune cell infiltration, which helped comprehend the role of FAT1 in tumorigenesis from the perspective of clinical tumor samples. Moreover, FAT1 mutations were also found to be closely associated with immunotherapy and may develop into tumor immunotherapy