Bromodomain containing 4 (BRD4) as a potential prognostic marker in a pan-cancer analysis of human tumors

There is some evidence supporting an association between Bromodomain containing 4 (BRD4) and cancer, but no research using pan-cancer analysis has been conducted previously. We therefore investigated the oncogenic role of BRD4 in 33 tumors from the Gene Expression Omnibus and The Cancer Genome Atlas databases. BRD4 is highly expressed in many tumors, and the prognosis of certain cancers is vitally linked with BRD4 expression. BRD4 expression is associated with CD8+ T-cell infiltration levels in testicular germ cell tumors and head and neck squamous cell carcinomas, and we observed a positive relationship between BRD4 and Tcm (T central memory) and Th(T helper) cells, and a negative relationship pDC (plasmacytoid DC). BRD4 had negative associations with marophages, iDC, Treg, cytotoxic cells, and Th17 cells in multiple tumors. The top 100 genes that are most strongly related to BRD4 were identified, and enrichment analysis indicated that the biological process with the closest relationship was chromatin modifying enzymes, related pathways included the signaling pathways of intracellular receptor, cytokine Signaling in Immune system and regulation of TP53 activity through acetylation. BRD4 is related to biological cell behaviors such as DNA-templated transcription, regulation of histone modification, protein modification by small protein removal and mitotic sister chromatid segregation. As the first study to perform a pan-cancer analysis of BRD4, the present findings will improve the understanding of the oncogenic role of BRD4 in different tumors.


Introduction
The intricacy of tumors means that they require complex regulation. It is therefore necessary to analyze the genes relating to pan -cancer expression and determine the correlation between pre -and post-evaluations and the potential molecular mechanism [ 1]. The Cancer Genome Atlas (TCGA) public database and the Gene Expression Omnibus (GEO) project provide data on the functional genomics of different tumors [ 2 -4] , which we can use for pan-cancer analysis.
Bromodomain and extra-terminal domain( BET) family is a class of proteins that can specifically recognize acetylated lysine and regulate gene transcription, which plays an important role in the occurrence and development of many diseases. BRD4 is a member of bet family, which can bind with acet ylated histone or non histone proteins, and then regulate gene replication and transcription, affecting cell cycle, cell differentiation, apoptosis [ 5] .
Other research groups have suggested a functional association between the multifunctional BRD4 protein and the occurrence of ovarian [ 6 ] , lung [ 7 , 8 ] , and breast [ 9 , 1 0 ] cancers. The current evidence from animal and cell studies supports correlations between different cancer types and BRD4. Nevertheless, despite extensive clinical data, no evidence is available o n the pan-cancer associations between BRD4 and different tumor types. The present study is the first to use TCGA database and the GEO project to conduct a pan -cancer investigation of BRD4. Many aspects such as survival condition, gene expression, immune infiltration, genetic changes, and related cellular pathways are summarized in order to determine the possible molecular mechanisms of BRD4 in clinical prognoses or the pathogenesis of different cancers.
We hypothesized that BRD4 mutation alters its expression level, causes changes in the body's immune system, changes the expression levels of various immune cells, and further influences tumor prognosis and survival time, affecting some pathways in vivo. Our goal was to determine t hese changes, explore how BRD4 influences changes in immune cells and prognoses, and identify the affected molecular pathways in vivo to provide direction and guidance for clinical and drug treatments.

Analysis of gene expression
We used the website http://timer.cistrome.org/ to obtain TIMER2.0 (Tumor Immune Estimation Resource 2nd edition), entered BRD4 into the "Gene_DE" module, and observed the differences in BRD4 expression betwee n tumors of specific subtypes and adjacent normal tissues or different tumors in TCGA database. We did not analyze some highly restricted tissues such as TCGA diffuse large-B-cell lymphomas (DLBCs) and THYMs (thymomas). We used the website http://GEPIA2.cancer-pku.cn/#Analysis to obtain GEPIA2 (Gene Expression Profiling Interactive Analysis 2nd Edition) [ 11] and its "Expression-Analysis Box Plots" that can be used to create box plots of the expression differences between tumor tissues and the corresponding normal tissues from the GTEx (Genotype -Tissue Expression) database. We set the log2 relative change cutoff at 1 and a Pvalue cutoff of 0.01, expressed as "Match TCGA normal and the GTEx data." We also used GEPIA2 to obtain a violin plot of BRD4 expression in all TCGA tumors at different pathological stages (stage I to stage IV) using the "Pathological Stage Plot" module. Log2 TPM (transcripts per million) + 1 was used to transform expression data from the violin plot or box.

Survival prognostic analysis
GEPIA2 is an online tool for TCGA gene expression and survival analysis.
The GEPIA2 [ 1 1 ] "Survival Map" module was applied to all TCGA tumors to identify the DFS (disease-free survival) and OS (overall survival) due to high and low expression of BRD4. Expression thresholds were applied to the low (50%) and high (50%) cutoff values to divide i nto low-and high-expression cohorts of BRD4.
The log-rank test in the "Survival Analysis" module were used for hypothesis testing and the survival plots, respectively. TCGA data were then extracted, and a receiver operating characteristics (ROC) curve wa s plotted using the "Survival ROC" software package. In the ROC curve image, the abscissa and the vertical axis indicate the false-and true-positive rates, respectively. A larger area under the ROC curve (AUC) indicates greater prognostic accuracy.

Analysis of genetic alteration
We selected the "TCGA Pan-Cancer Atlas Studies" section on the cBioPortal website (https://www.cbioportal.org/ ) [12,13] , and entered " BRD4" to investigate the genetic alteration characteristics of BRD4. We obtained information on copynumber alterations (CNA), mutation types, and frequencies of all tumors in the "Cancer Type Summary" TCGA module. We used the "Comparison" module on TCGA cancer cases to obtain data on differences in OS, progression -free survival, disease-specific survival, and DFS rates with and without BRD4 gene changes.
Log-rank P values were used to construct the Kaplan -Meier graph, with a P value of <0.05 considered significant.

Analysis of immune infiltration
We chose the "Immune Gene" module from the TIMER2 web server to determine the relationship between all TCGA tumor immune infiltrations and BRD4 expression.CD8+ T cells immune infiltration data was obtained using the MCPCOUNTER, QUANTISEQ, CIBERSORT, CIBERS ORT-ABS, TIMER, EPIC, and XCELL algorithms. P values and sectional correlation values were obtained using Spearman's rank correlation test with purity adjustment. We used these data to construct scatter and maps. RNA -seq data and clinical data were then

Analysis of BRD4-related gene enrichment
We chose the organism "Homo sapiens" fro m the STRING [ 1 2 ] website (https://string-db.org/) and the single protein name "BRD4." The main parameters were then set as follows: "low confidence [0.150]" as the minimum interaction point, maximum number of displayed interaction factors ("no more than 50 interactors in the first shell"), meaning of the network edge ("evidence"), and active interaction sources ("experiments"). Furthermore, we obtained BRD4binding proteins that had been determined experimentally. We use the data of TCGA tumors and normal t issues to identify the 100 genes most strongly associated with BRD4 in the GEPIA2 "Similar Gene Detection" module. Selected genes and Pearson's correlation analysis of BRD4 paired genes were used in the "Correlation Analysis" GEPIA2 module. The dot plot used log2 TPM + 1, to determine correlation coefficients(R Values) and P values. The P values and partial correlation heat map data from the Spearman's rank correlation test were determined using selected genes in the TIMER2 "Gene_Corr " module after purity adjustment.
We also performed pathway analysis by combining the two data sets. We uploaded the lists of genes to the Metascape [ 1 3 ] (http://metascape.org/gp/index.htm l#/main/step1) website, and chose "Homosapiens (147)" and "Express Analysis." The flow chart of analysis is shown in Figure 7.

Results
We hypothesized that BRD4 mutation alters its expression level, alters the The Clinical Proteomic Tumor Analysis Consortium (CPTAC) integr ates genomic and proteomic data in order to identify and describe all proteins within tumor and normal tissues, and explores candidate proteins that can be used as tumor biomarkers. Data from the CPTAC data set indicated that BRD4 total protein expression was higher in breast cancer(P<0.01), colon cancer, lung adenocarcinoma, clear renal cell carcinoma, ovarian cancer and uterine corpus endometrial carcinoma than in normal tissues (P<0.001, Figure 1C), whereas it s expression was lower in Uterine Corpus Endometrial Carcinoma than in norm al tissues. (P<0.001, Figure 1C).

Analysis of survival
Cancer cases were divided into high -and low-BRD4 expression groups, and TCGA and GEO data were mainly used, respectively, to investigate the correlations between BRD4 expression and the prognoses of different tumors. As shown in Figure  LGG(P=0.0024),was linked to high BRD4 expression ( Figure 2B).

Analysis of genetic alteration
We used TCGA cohort to analyze different tumor samples and their genetic alteration status with BRD4. As displayed in Figure 3A, the highest alteration frequency of BRD4 (>12%) was in patients with uterine corpus endometrial carcinoma with "mutation" as the primary type.
The dominant type of ovarian serous cystadenocarcinoma cases and all adrenocortical carcinoma 、 brain lower grade glioma 、 uveal melanoma 、 mesothelioma with genetic changes were specified as CNA "amplification" in   however, there were insufficient data on DFS to draw any conclusions.

Analysis of immune infiltration
As an essential part of the tumor microenvironment, the occurrence, development, and metastasis of cancer are closely related to tumor -infiltrating immune cells [20,21]. It has been reported that the tumor stromal microenvironment aims to regulate the effect of tumor -infiltrating immune cells [22,23].

Enrichment analysis of BRD4-correlated protein
To study the molecular mechanism of BRD4 during tumorigenesis, we conducted various pathway enrichment analyses to identify targeted BRD4combining proteins and their corresponding expression -related genes. We used previous experimental evidence to identify the top 50 BRD4-binding proteins on the STRING web. The network of interactions between these proteins is shown in Figure 6A. Combining the GEPIA2 tool with TCGA tumor expression data revealed the top 100 genes associated with BRD4 expression. The above two results were combined in Metascape to determine the Gene Ontology (GO) annotation results. The data in Figure 6D suggest that during tumor pathogenesis, chromatin modifying enzymes might be correlated with the effects of BRD4, as similarly suggested from a previous stud y [ 1 4 ] . This analysis also indicated that most of the above genes are related to biological cell behaviors such as DNA-templated transcription, regulation of histone modification, protein modification by small protein removal, DNA methylation or demethylation and mitotic sister chromatid segregatio n. This may be relevant to the signaling pathways of intracellular receptor, cytokine Signaling in Immune system and regulation of TP53 activity through acetylation ( Figure 6D).

Conclusion
This is the first research study to systematically evaluate the potential role of BRD4 in disease progression and prognosis in several types of cancer. The present finding indicates that BRD4 expression may regulate tumor prognosis by altering and regulating certain immune cells, which has positive relationships with Tcm and Th cells, and Tgd, and negative relationships with pDC, marophages, iDC, Treg, cytotoxic cells, and Th17 cells . chromatin modifying enzymes may be affected by BRD4, and BRD4 may be involved in the signaling pathways of intracellular receptor, cytokine Signaling in Immune system and regulation of TP53 activity through acetylation. It is therefore necessary to further investigate the diagnostic and therapeutic value of BRD4 in a variety of human cancers.

Research highlights
This is the first pan-cancer analysis of BRD4.
Novel effects of BRD4 on tumor prognosis and immune microenvironment have been revealed.
The relationship between the BRD4 protein and gene has been displayed.

Disclosure
The authors report no conflicts of interest in this work. Figure 1 Please see the Manuscript PDF le for the complete gure caption. shown. The receiver operating characteristics (ROC) curve between BRD4 and tumor prognosis was plotted based on data from TCGA database. (C) The areas under the ROC curves for adrenocortical carcinoma (ACC), brain lower grade glioma (LGG), colon adenocarcinoma (COAD), and head and neck squamous cell carcinoma (HNSC) were 0.848, 0.813, 0.805, and 0.821, respectively, indicating a high predictive value for tumor prognosis.

Figure 3
TCGA was used to obtain the mutation effect of BRD4 on different tumors via the cBioPortal. This gure displays (A) the alteration frequency in different mutation molds, (B) mutation sites, and (C) the potential links between mutation condition and versions of ACC survival curves, as obtained using the cBioPortal tool.

Figure 4
Links between BRD4 expression and CD8+ T immune cells. We used various algorithms to identify any links between BRD4 expression and immune cells. Within whole cancer types in TCGA project, we explored the expression level of BRD4 and the CD8+ T-cell in ltration status (A, B).   The ow chart of analysis