Pan-Cancer Analysis of TLE3 Revealed Its Value in Tumor Microenvironment and Prognosis

Background Transducin-like enhancer of split 3 (TLE3), a member of the TLE gene family, is related to tumor genesis and progression. However, whether TLE3 played a crucial role in the whole pan-cancer remained unknown. Methods Comprehensive analysis of TCGA, GEO, and GTEx data with an online tool, and R language was performed to explore the relationship of TLE3 expression between prognosis, gene mutation, protein phosphorylation, DNA methylation, tumor microenvironment, and related pathways in 33 tumors. Results TLE3 was high-expressed in most tumors, and TLE3 expression and the prognosis of some tumor types were significantly correlated. The level of TLE3 expression in 33 cancer types was closely associated with DNA methylation. High-level phosphorylation sites of Tle3, such as S267 and S217, may promote cancers. In terms of the tumor microenvironment, TLE3 affected a wide variety of cancers, especially PRAD and LIHC, and TLE3 may act on them via immune-related pathways. Conclusions The current work provided the first comprehensive investigation of TLE3 in a pan-cancer study, highlighting the role of TLE3 in the tumor immune microenvironment, and also determined the potential of TLE3 as a prognostic, immunotherapy response, and diagnostic biomarker in many cancers. However, the present results were preliminary and required further validation as this study was based on bioinformatics analyses.


Introduction
Cancer is one of the most challenging diseases in clinical treatment, and its threat to human health is becoming increasingly serious [1]. Many oncogenic and tumor suppressor genes are implicated in cancer initiation and progression. Constantly developing techniques and bioinformatics databases would allow us to analyze more genes of interest in pan-cancer.
As a conserved family of corepressor proteins, the transducin-like enhancer of split (TLE) is present in multiple animals, including mice and humans. Te TLE protein family plays a key role in the entire life cycle of animals, including lateral inhibition, segmentation, sex determination, and eye and pancreatic development, and regulates basic processes such as metabolism through interacting with multiple pathways [2].
TLE3 as one of the TLE family proteins is a transcriptional inhibitory homolog of the Groucho protein [3], which is a part of the Drosophila Notch signaling cascade and may be expressed periodically during the M phase of the cell cycle [4]. It cooperates with the transcription factor HHEX to promote memory B cell development [5]. TLE3 has been shown to have inhibitory or promoting efects on diferent tumors [6,7]and is a predictive marker for response to taxane-containing regiments in breast and nonserious ovarian cancer [8,9]. However, TLE3 has previously been studied in only a few cancer types, its role in other types of cancer remains unclear, and the pathogenesis or impact on the survival of diferent cancers has not been elucidated.

Gene Ontology Analysis.
Genomic localization information for the TLE3 gene was obtained from the UCSC genome browser on humans (GRCh37/hg19) (https://genome.ucsc.edu/) [10]. Te analysis of TLE3 protein conserved functional domain in multiple species was analyzed by clicking the "HomoloGene" button of the NCBI(https://www.ncbi.nlm.nih.gov/) and then the "constraint-based multiple sequence alignment online tool." Te phylogenetic tree of TLE3 in various species was mapped by the "constraint-based multiple sequence alignment online tool." In addition, we accessed the data of the TLE3 mRNA expression in various cells and tissues under physiological conditions using an online HPA database (https://www.proteinatlas.org/ humanproteome/pathology).

mRNA and Protein Expression Analysis.
Te diferentially expressed TLE3 between various tumors and corresponding normal tissues were determined by clicking on the "Gene_de" button on TIMER2 web (https://timer. cistrome.org/). For some cancers without normal tissues information (e.g., LGG), the GEPIA2 web server (https:// gepia2.cancer-pku.cn/#analysis) [11] was applied, and we clicked the "Match TCGA normal and GTEx data" to acquire a chart of TLE3 expression contrast between these tumor tissues and the corresponding normal tissues using the GTEx database. Te TLE3 protein expression was obtained from the UALCAN portal (https:// ualcan.path.uab.edu/analysis-prot.html) using the "CPTAC" dataset [12]. Ten, we determined the levels of TLE3 protein expression between primary tumors and corresponding normal tissues. With the HPA database (HumanProteinAtlas), we acquired IHC (immunohistochemical) images to show the protein expression of TLE3 in tumors and corresponding normal tissues. Next, by the "Pathological Staging" module in GEPIA2, the diferent pathological (stages I, II, and III) stages of the tumor were obtained.

Survival
Analysis. OS (overall survival) and DFS (disease-free survival) data of the TLE3 gene were acquired in the "Survival Map" option of GEPIA2 [1], and the survival graphs were then produced in the "Survival Analysis" module. We then performed analysis and visualization using the uniformly standardized pan-cancer dataset according to the TCGA database in the R environment of survival and survminer package. We used multivariate Cox regression analysis to study the relationship between TLE3 expression and OS, DSS (disease-specifc survival), and PFI (progression-free interval). Results may vary due to diferent algorithms. To further validate the association between TLE3 expression and cancer survival, the Kaplan-Meier plotter (https://kmplot.com/analysis/) was used to analyze the OS, DMFS (distant metastasis-free survival), RFS (relapse-free survival), PPS (postprogression survival), FP (frst progression), DSS, and PFS (progression-free survival).

Genetic Alteration Analysis.
We obtained the mutations in the TLE3 gene using the cBioPortal web(https://www. cbioportal.org/) [13,14], including the frequency of mutations, mutation types, CNA (copy number alteration), mutation site information, and 3D protein structure maps of the TCGA tumors. Ten, patients with and without TLE3 gene mutations were compared for diferences in terms of overall, disease-free, progression-free, and disease-free survivals and Kaplan-Meier plots. Te results were shown by log-rank P-value.

DNA Methylation Analysis.
Based on the TCGA database, the UALCAN online tool was used to evaluate TLE3 gene promoter methylation levels between the primary tumors and corresponding normal tissues [15].TLE3 DNA methylation levels were determined through the MEXPRESS website. Based on the TGGA data, we used the R package to plot the association between TLE3 levels and DNA methyltransferases.

Protein Phosphorylation Analysis and Immunotherapeutic
Response of TLE3. Expression information of TLE3 protein phosphorylation was obtained from the UALCAN portal. We obtained the relationships between TLE3 mRNA expression and TMB, MSI analysis data from SangerBox (https://sangerbox.com/Tool). We integrated the data and log-transformed each expression value, which was then modifed using Microsoft PowerPoint software to generate images. We also used ggplot2 [version 3.3.3] to visualize the mismatch repair gene proteins (MLH1, MSH2, MSH6, PMS2, and EPCAM) based on the TGCA database.

2.7.
Immunocorrelation Analysis of TLE3. Immunocorrelation analysis of TLE3 was downloaded from TISIDB (https://cis.hku.hk/TISIDB/), a storage website integrating rich human cancer datasets in the TCGA database [16]. In addition, we also obtained the immune cell infltration of TLE3 in the MCPcounte algorithm from SangerBox.

TLE3-Related Gene Enrichment Analysis.
Te experimentally identifed TLE3 and binding proteins were available from the STRING website [17] (https://string-db.org/), and a protein-protein interaction (PPI) network map was generated. We exported short tabular text in TSV format and substituted it into Cytoscape 3.9.1 software to obtain the visual network diagram. Te frst 100 target genes associated with tle3 were obtained by selecting "TCGA + GTEx" in the "similar gene detection" function of GEPIA2. A scatter plot of the association between TLE3 and the frst 5 related genes was then generated, and a heat map of the correlation between these 5 genes and each tumor in diferent algorithms was obtained on the TIMER2 website. With jvenn (interactive Venn diagram viewer) [18], we compared TLE3 binding genes and interacting genes by cross-tabulation analysis to obtain the intersection. Analysis of functional enrichment for genes interacting with TLE3, including BP pre-mRNA processing and cytoskeleton assembly; typically contains a GH dipeptide 11-24 residues from its N-terminus and the WD dipeptide at its C-terminus and is 40 residues long, hence the name WD40; between GH and WD lies a conserved core; serves as a stable propeller-like platform to which proteins and bind either stably or reversibly; forms a propeller-like structure with several blades where each blade is composed of a four-stranded anti-parallel b-sheet; instances with few detectable copies are hypothesized to form larger structures by dimerization; each WD40 sequence repeat forms the first three strands of one blade and the last strand in the next blade; the last C-terminal WD40 repeat completes the blade structure of the first WD40 repeat to create the closed ring propeller-structure; residues on the top and bottom surface of the propeller are proposed to coordinate interactions with other proteins and/or smaII ligands; 7 copies of the repeat are present in this alignment. (biological process), CC (cellular component), and MF (molecular function), were performed using the "clus-terProfler" R package, and KEGG pathway (https:// www.kegg.jp/kegg/kegg1.html) was obtained from the DAVID database (https://david.ncifcrf.gov) [19]. After downloading the enrichment pathway analysis data from the DAVID database, the Cytoscape 3.9.1 software was inserted to obtain a visual network map using the enrichment map plugin. GSEA analysis of TLE3 data on BRCA and LIHC was obtained from LinkedOmics (http://linkedomics.org/) [20].

Statistic Analysis.
We used Pearson's or Spearman's coefcient analysis to investigate the correlations between the variables. Continuous variables ftting a normal distribution between binary groups were compared using a t-test. Otherwise, the Mann-Whitney U test was used. Categorical variables were compared by the chi-squared test or Fisher's exact test. All the statistical data analyses were implemented using R software, version 3.6.3. Two-tailed P < 0.05 was considered statistically signifcant.

Gene Ontology Analysis.
Firstly, we discussed the basic information of TLE3 (NM_001105192.3 for mRNA or NP_001098662.1 for protein, Figure 1(a)) in humans. TLE3 is located at locus 70047790-70098176 on chromosome 15. Among animals including H. sapiens, P. troglodytes, and M. mulatta, the TLE3 protein structure is conserved and commonly made up of the TLE_N (pfam03920) and WD40 (cd00200) domains. Te data of the phylogenetic tree ( Figure 1(b)) shows the evolution relationship relation of the TLE3 protein in multiple species. We then explored the TLE3 expression in normal human tissues. According to the HPA, GTEx, and FANTOM5 consensus dataset, the top three TLE3 expressions were bone marrow, esophagus, and skin ( Figure 1(c)). TLE3 is expressed in all tissues, regions, and cell types, showing an enhanced RNA tissue specifcity. Analysis of TLE3 expression in various immune cells from the combination of HPA, Monaco, and Schmiedel datasets also showed enriched RNA blood cell type specifcity (Figure 1(d)). However, there was no information on blood protein concentration in plasma about TLE3.

mRNA and Protein Expression Analysis.
We applied TIMER2 to explore the TLE3 mRNA expression across 33 cancers. As Figure 2(a) shows, the level of TLE3 mRNA expression in the 16 types of tumor tissues was higher than the corresponding control tissues. Normal tissue data from the GTEx dataset were used as a control, we further evaluated the diference in TLE3 mRNA expression between LGG control tissues and tumor tissues ( Figure 2(b), P < 0.05). However, no signifcant diferences were shown in other cancers ( Figure S1(a)). Proteins are the macromolecules that ultimately perform biological functions in the human body, and have expression closer to phenotypes, allowing for more realtime monitoring of disease progression. Terefore, based on the CPTAC dataset, we determined the TLE3 protein expression levels in 33 cancers. Te results showed that total protein expression of TLE3 was elevated in the primary tissues of breast cancer, clear cell RCC, lung adenocarcinoma, pancreatic adenocarcinoma, glioblastoma multiforme, hepatocellular carcinoma, ovarian cancer, and UCEC than in normal tissues. As shown in Figure 1(c), RNA levels (transcription levels) and protein levels (translation levels) of TLE3 were consistent in part, and the results of the two databases validated each other. In addition, to further verify the protein level of TLE3 expression, we compared the IHC results from the HPA database, which was also in line with gene expression data from the TCGA (Figure 2(d)). Normal breast, colon, Ovarian cancer * * 2 Clear cell RCC * * * 3 Breast cancer * * * 3    MESO  OV  PAAD  PCPG  ACC  BLCA  BRCA  CESC  CHOL  COAD  DLBC  ESCA  GBM  HNSC  KICH  KIRC  KIRP  LAML  LGG  LIHC  LUAD  LUSC  PRAD  READ  SARC  SKCM  STAD  TGCT  THCA  THYM  UCEC    Journal of Oncology lung, and prostate TLE3 IHC staining of tissues were negative or moderate, while that of tumor tissues were moderate or strong (Figure 2(e)). We also analyzed the relation between the mRNA expression of TLE3 and the pathological stages of cancer, OV showed signifcant diferences (Figure 2(e) P < 0.05), while others did not ( Figures S1(b)-S1(g)).
3.3. Survival Analysis. As TLE3 was overexpressed in pancancer, we further explored its prognostic profle. Low expression of TLE3 was connected to the poor OS of KIRC, whereas high TLE3 expression was associated with poor OS of MESO and poor DFS of ACC, according to the GEPIA2 database (Figure 3(a)). We then studied the relationship between TLE3 mRNA expression and survival of ACC, BLCA, BRCA, KIRC, KIRP, and UCS through multivariate Cox regression analysis ( Figure  3(b)). Moreover, survival data were analyzed using the Kaplan-Meier mapping tool, and a correlation was found between low TLE3 expression and poor prognosis of breast cancer in RFS, OS, and DMFS (Figure 3(c)). Additionally, low expression of TLE3 was associated with poor prognosis of FP and OS in lung cancer. By contrast, the high-expressed mRNA level of TLE3 was connected

Genetic Alteration Analysis.
Genetic mutations have an important infuence on cancer development, and these mutated genes may also be used as efective therapeutic targets. Terefore, we also observed the state of TLE3 genetic alteration in diferent tumor samples. Te frequency of alteration in TLE3 was the highest (>4%) in patients with   a "mutation" primary type of uterine tumor. It should be noted that all cases of genetically altered mesothelioma (more than 3% frequency) had amplifcation of TLE3 (Figure 4(a)). More information is shown in Figure 3(b), including the types, sites, and caseload of the mutant TLE3 gene. Te data demonstrated that the main type of TLE3 genetic alteration was missense mutation, and A477T alteration was detected in LGG, UCEC, STAD, and COAD, respectively, (Figure 4(b)), which had the capacity for inducing a frameshift mutation of the TLE3 gene, translating from A (Alanine) to T (Treonine) at the 477 sites of TLE3 protein. A477T site is shown in the 3D structure of the TLE3 protein (Figure 4(c)). Furthermore, the impact of TLE3 gene mutations on survival in pan-cancer was also studied. As indicated in Figure 4(d) from UCEC cases, we were unable to detect the association between alterations of TLE3 and OS, DSS, DFS, and PFS, compared with cases without altered TLE3.
To further study the TLE3 efect on immunological status in pan-cancers, we analyzed the association between its mRNA expression and the immunomodulators in 33 cancer types using the TISIDB database. As illustrated in Figure 9, our fndings revealed that TLE3 was negatively correlated with most of the immunomodulators in BRCA but positively correlated with LIHC. In addition, the MCPcounter algorithm was used to evaluate the immune infltration of TLE3 in diferent cancers to verify the above results. Te current results consistently showed that TLE3 was positively correlated with the immune infltration of LIHC ( Figure S5).
We also used the data from the TCGA database to evaluate the relationship between the TLE3 expression and 60 checkpoint genes ( Figure 10). Analysis of TLE3 correlation with checkpoint gene expression in pan cancers showed that Tle3 was closely linked to CD276. Moreover, the TLE3 expression was positively related to a large number of immune checkpoint genes in many cancers, especially KIRC,  Figure 11: (a) Available experimentally determined tle3-binding proteins obtained using the String tool. Based on the STRING tool, we obtained the available experimentally determined Tle3-binding proteins, (b) the top 100 genes associated with Tle3 in the TCGA project were obtained by the GEPIA2 method, and the expression correlation between Tle3 and the selected targeted genes, including THRAP3, SIN3A, RBM12, ARID1A, and ADNP, were analyzed, (c) the heatmap shows the detailed cancer types, (d) an intersection analysis between the Tle3-binding and correlated genes, (e) KEGG analysis based on the Tle3-binding and interacted genes, (f ) GO term enrichment analysis of Tle3-binding and interacted genes in molecular function, and (g) GSEA analysis of TLE3 showing in BRCA, LIHC using LinkedOmics data.
BRCA, and LAML, suggesting that TLE3 participated in the tumor immune response regulation through modulating immune checkpoint activity.

Enrichment Analysis of TLE3.
To elucidate the TLE3 gene molecular mechanism in tumorigenesis, a pathway enrichment analysis was performed after screening proteins bound to TLE3 targeting and those associated with expressed TLE3. Using the STRING tool, we detected 50 TLE3-binding proteins, all of which were experimentally validated. Te network map of proteins interacting with TLE3 is shown in Figure 11(a). We also performed a data visualization analysis, as shown in Figure S3, TLE1, TLE4, TLE2, and TLE3 in the TLE family were the most closely related. Te relationship between diferent genotypes and diferent subtypes, and genes was explored, and an interaction network was obtained ( Figure S6). Te top 100 genes associated with TLE3 expression were obtained from all tumor expression data from TCGA using the GEPIA2 tool. As shown in Figure 11(b), the expression level of TLE3 was positively correlated with that of THRAP3, SIN3A, RBM12, ARID1A, and ADNP genes (all P � 0). Figure 11(c) showed a positive relationship of TLE3 with the above fve genes in 33 cancers. Te intersection of the two groups was PAXIP1 (Figure 11(d)). We performed KEGG and GO enrichment analyses combining the two datasets. Figure 11(e) suggested KEGG data that the "Wnt signaling pathway,"" Notch signaling pathway," and "Human papillomavirus infection" might be related to the efect of TLE3 on tumor pathogenesis. Te GO data suggested that most of these genes were related to transcription factor binding ( Figure 11(f )). Te enrichment map obtained by Cytoscape Software also showed 23 enriched pathways, the top three with the highest enrichment were "Pathways in cancer," "Human papillomavirus infection," and "Wnt signaling pathway," "Pathways in cancer" was associated with 17 of these pathways ( Figure S4). In addition, we also studied BRCA and LIHC, respectively, two cancers with a strong correlation with TLE3. Furthermore, we also studied BRCA and LIHC, two cancers with a strong correlation with TLE3, separately, using the GSEA methods. We found that TLE3 was involved in immunerelated pathways (Figure 11(g)).

Discussion
TLE3 properties in tumorigenesis have been found in addition to its dynamic functions in diferentiation and cell metabolism [23][24][25][26]. To elucidate the molecular mechanism of TLE3 in cancers, we comprehensively analyzed the molecular characteristics of its mRNA and protein expression, gene mutation, DNA methylation, protein phosphorylation, and the immune microenvironment in 33 cancers based on available data from TCGA, GEO, and GTEx. According to our results, the TLE3 protein had a conserved structure in various species, indicating that efects of TLE3 may exist with similar mechanisms between these species, hence, it could be viable to use mice for studying TLE3-related cancer in humans. TLE3 is wildly distributed in many tissue types and immune cells, especially in bone marrow and neutrophils, which may be associated with immunity, and this further encouraged us to study its immunoprognostic value in tumors.
Numerous researches have reported that TLE3 expression is upregulated in multiple cancers, including cervical cancer [27] and malignant meningiomas [28]. TLE3 alternatively spliced isoforms have been detected to be upregulated in prostate tumors [7,29], suggesting that the abnormal expression of TLE3 is strongly connected with cancer invasiveness. According to our results, TLE3 expression was high in most tumors with consistent results. Te data for genetic alterations from TCGA and protein alterations from CPTAC were not entirely consistent, which may be resulted from the fact that diferent acquisition and calculation of the data in the database would be afected by gene mutations, promoter methylation, and protein phosphorylation to some extent. TLE3 was high-expressed in the pathological stage I gene of OV and OV patients with high expression of TLE3 had a poor prognosis, suggesting that the gene played a role as an oncogene in OV, and the OV patients can be treated in the early stage or separately according to their pathological stage.
GEPIA2, Cox regression, and Kaplan-Meier analysis indicated that the prognosis of ACC with high TLE3 expression was poor. Besides, high concentrations of GRO homologous to the human gene TLE3 was found in ACC cells [30]. Cox regression and Kaplan-Meier analysis suggested that TLE3 was a protective factor for breast cancer. A recent study has shown that miR-3677 can accelerate cell proliferation, migration, and invasion by targeting TLE3 in human breast cancer [31]. Another study found that TLE3, a transcription corepressor recruited by FOXA2 to the ZEB2 promoter, inhibits the expression of the EMT-related transcription factor ZEB2, thereby suppressing the EMT of breast cancer cells [32]. Te results of diferent studies were consistent with our survival data. Tese fndings strongly supported that TLE3 may be a potential prognostic indicator and molecular therapeutic target in ACC and BRCA.
We explored the immune characteristics of TLE3 in the tumor microenvironment (TME), including tumorinfltrating lymphocytes, immunostimulatory factors, MHC molecules, chemokines, receptors, and immune checkpoints. Additionally, three immunotherapeutic biomarkers (TMB, MSI, and MMR) were found to be signifcantly associated with TLE3 in some cancers. Our results showed that the poor prognosis of ACC with high TLE3 expression was positively related to TMB and MSI. Immune checkpoint therapy (ICT) to enhance the antitumor immune response of T cells is a novel treatment for malignant tumor. CD276 is an immune checkpoint molecule in the epithelialmesenchymal transformation (EMT) pathway, which plays a crucial part in the cell proliferation, invasion, and migration of malignant tumors, and is a promising therapeutic target for tumors [33]. Here in this study, we also found that TLE3 was positively associated with many immune checkpoint molecules, indicating that it had the potential to become a novel immunotherapy target. It is worth noting that TLE3 was high-expressed in BRCA and negatively related with TMB, MSI, as well as most of the immune regulatory factors in BRCA. Terefore, we hypothesized that TLE3 played an immunosuppressive role in BRCA and may be an additional diagnostic marker and therapeutic target. TLE3 was positively associated with most of the immunomodulators of LIHC, indicating that LIHC was a candidate cancer type suitable for TLE3 immunotherapy. In conclusion, TLE3 played an important and dual role (both inhibition and promotion) in pan-cancer immune infltration.
Furthermore, the PPI map obtained by Cytoscape software suggested that TLE3 may play a role together with the TLE family in tumors. We performed a functional enrichment analysis of TLE3, and confrmed the potential efects of the "Wnt signaling pathway" and "Notch signaling pathway." It has been found that Wnt signaling plays a role in multiple immune cells, including in DCS, NK cells, T cells, B cells, and also macrophages [34], and the Wnt/betacatenin pathway can regulate the immune response against cancer [35]. Notch signaling also has critical functions in cancer, including the generation of blood vessels, the maintenance of cancer stem-cell-like cells, and tumor immunity [36]. Moreover, Enriched Map by Cytoscape Software also showed "Pathway in cancer," pointing to the close potential relationship between TLE3 and cancers. KEGG and GO enrichment and GSEA analysis of BRCA and LIHC also confrmed the correlation between TLE3 and the immune microenvironment, suggesting the signifcance of TLE3 in the immunity of a variety of cancers.
In brief, our study showed that TLE3 was high-expressed in the majority of cancers, had diferent prognostic efects in various tumor cases, and was associated with the progression of OV. Te TLE3 expression was also associated with TME. Te above data suggested that TLE3 was a valuable novel biomarker for prognostic and immunotherapy response assessment in several cancers.
However, our study has certain limitations. First of all, some of our results were obtained from the bioinformatics website as we did not have a complete grasp of the R language. Secondly, this manuscript was based on bioinformatics research, the results were inconsistent sometimes due to diferences in algorithms and databases. Terefore, the current fndings should be verifed experimentally, especially the role of TLE3 in BRCA and LIHC.

Data Availability
Te data used to support the fndings of this study are included in the article.

Conflicts of Interest
Te authors declare that they have no confict of interest.