Abstract
BACKGROUND:
Bone morphogenetic proteins (BMPs) are a group of cancer-related proteins vital for development and progression of certain cancer types. Nevertheless, function of BMP family in pan-cancer was not detailedly researched.
OBJECTIVE:
Investigating expression pattern and prognostic value of the BMPs family (BMP1-8A and BMP8B) expression across multiple cancer types.
METHODS:
Our research integrated multi-omics data for exploring potential associations between BMPs expression and prognosis, clinicopathological characteristics, copy number or somatic mutations, immune characteristics, tumor microenvironment (TME), tumor mutation burden (TMB), microsatellite instability (MSI), immune checkpoint genes and drug sensitivity in The Cancer Genome Atlas (TCGA) tumors. Furthermore, association of BMPs expression and immunotherapy effectiveness was investigated in some confirmatory cohorts (GSE111636, GSE78220, GSE67501, GSE176307, IMvigor210 and mRNA sequencing data from currently undergoing TRUCE01 clinical research included), and biological function and potential signaling pathways of BMPs in bladder cancer (BCa) was explored via Gene Set Enrichment Analysis (GSEA). Eventually, immune infiltration analysis was done via BMPs expression, copy number or somatic mutations in BCa, as well as validation of the expression levels by reverse transcription-quantitative PCR and western blot, and in vitro functional experiments of BMP8A.
RESULTS:
Discoveries displayed BMPs expression was related to prognosis, clinicopathological characteristics, mutations, TME, TMB, MSI and immune checkpoint genes of TCGA tumors. Anticancer drug sensitivity analysis displayed BMPs were associated with various drug sensitivities. What’s more, it was discovered that expression level of certain BMP family members related to objective response to immunotherapy. By GSEA, we discovered multiple immune-associated functions and pathways were enriched. Immune infiltration analysis on BCa also displayed significant associations among BMPs copy number variations, mutation status and infiltration level of diverse immune cells. Furthermore, differential expression validation and in vitro phenotypic experiment indicated that BMP8A significantly promoted BCa cell proliferation, migration and invasion.
CONCLUSIONS:
Current results confirmed significance of both BMPs expression and genomic alteration in the prognosis and treatment of diverse cancer types, and suggested that BMPs may be vital for BCa and can possibly be utilized as biomarkers for immunotherapy.
Introduction
Worldwide, cancer is one primary public health issue and can be result from numerous factors [1]. To identify key tumor-related genes for increasing current awareness of cancer initiation, maintenance and progression is essential [2]. Large-sample, high-throughput and numerous cancer data should facilitate for identifying key genes involved in carcinogenesis. Certain medical public databases like Gene Expression Omnibus (GEO) and TCGA provide numerous omics data for diverse cancer types, making pan-cancer research among diverse omics possible [3, 4].
Tumor immunotherapy contains monoclonal antibody immune checkpoint inhibitors (ICIs), therapeutic antibodies, cancer vaccines, cell therapy and small molecule inhibitors [5]. Novel such as programmed death 1 (PD-1), programmed death ligand 1 (PD-L1, also known as CD274) and anticytotoxic T lymphocyte antigen 4 (CTLA-4) displayed great possibility for therapeutic efficacy in numerous tumors, non-small cell lung cancer, metastatic melanoma and bladder cancer (BCa) included [6, 7]. BCa is the commonest malignant tumor of urinary system, and which has been previously revealed that BCa progression relates to accumulation of numerous genetic and epigenetic alterations [8]. Despite ICIs displayed long-term durable response and tolerable safety profile in some clinical trials, 70–80% of patients may be unresponsive to ICIs [9]. With molecular biology and genomic analyses develop, gene alteration associated with tumorigenesis are utilized as potential prognostic biomarker [10]. Tumor immune microenvironment is vital for cancer development. Consequently, identifying effective biomarker and potential therapeutic target for BCa treatment and diagnosis is of great importance.
BMPs are secreted growth factors inducing bone and cartilage formation, and form part of TGF-
In our research, database research and thorough bioinformatic analysis were performed for investigating expression pattern and prognostic value of the BMPs family (BMP1-8A and BMP8B) expression across multiple cancer types. Additionally, we investigated association of BMPs expression and copy number or somatic mutations, TME, TMB, MSI, drug sensitivity and immune checkpoint genes in pan-cancer, and further analyzed immunotherapy effectiveness as markers of BMP family members from different immunotherapy cohorts. Additionally, correlation of BMPs expression pattern and signaling pathways, immune subtypes, clinicopathological characteristics, immune microenvironment and immune infiltrates was assessed in BCa, and reverse transcription-quantitative PCR (RT-qPCR) expression validation and functional experiments of BMP8A were carried out in vitro. The present findings could serve as a basis for the use of BMPs as prognostic biomarkers in BCa immunotherapy.
Materials and methods
Data acquisition
mRNA transcription expression, clinical, survival, mutation, immunophenotype and stem index evaluation (stemness score on the basis of DNA methylation/RNA expression) data of 33 cancer types in TCGA were acquired from University of California, Santa Cruz Xena database [15]. Microarray expression data of 4 GEO immunotherapy datasets [GSE111636 (
What’s more, in single-arm phase II clinical research on tislelizumab with low-dose nab-paclitaxel (named as ‘TRUCE-01’, registration number NCT04730219;
RNA-seq analysis on TRUCE-01
Total RNA was obtained from specimens via a RNeasy Mini Kit (Qiagen, Valencia, CA). RNA quality and quantity in libraries was controlled via BioAnalyzer 2100 system (Agilent Technologies). Sequencing libraries were produced via NEBNext UltraTM RNA Library Prep Kit (Illumina, USA) and adjusted to a final concentration of 4 nM measured via qPCR. Consequently, 150-bp paired-end reads were produced through sequencing for 150 cycles on Illumina HiSeq 4000 platform. Finally, HISAT2 aligner was employed for aligning raw reads to human reference genome GRCh38 and quantifying gene expression with default parameters.
Expression analysis on BMPs family in pan-cancer
Firstly, mRNA transcriptional data of normal samples were filtered out, while cancer tissues for 33 TCGA cancer types were kept for showing a box plot of expression levels of nine BMPs family members (BMP1-8A and BMP8B). Finally, differential expression levels of BMPs between normal and tumor tissues in 18 cancer types with over 5 normal adjacent normal tissue as control were analyzed by Wilcoxon test. And, ‘corrplot’ R package was employed for analyzing Pearson correlation of BMPs.
Clinical association analysis
Patients were grouped into high- and low-expression based on median expression level of every BMP family member. Kaplan-Meier (K-M) curves were created for displaying differences in patient survival, while univariate Cox proportional hazard regression analysis was used for obtaining hazard ratio of BMPs expression with respect to overall survival (OS), disease-free survival (DFS), disease-specific survival (DSS) and progression-free survival (PFS) in every TCGA tumor type. Multivariate Cox regression analysis of BMPs combined with clinical factors was further performed for identifying ones with independently prognostic value. Furthermore, by Wilcox rank-sum test, differential analysis was employed for detecting differences in BMPs expression levels at diverse stages and outcomes after initial treatment of the 33 cancer types.
Analyses of BMPs mutation frequency, effects of mutations on patient outcomes, and associations of pan-cancer BMPs mRNA expression with copy number variations (CNVs) or single nucleotide variations (SNVs)
cBioPortal for Cancer Genomics (
Correlation analysis on BMPs expression and immune subtypes, TME and stemness indices
Correlations of BMPs expression and the immune subtypes of pan-cancer and BCa were explored using TCGA database. Using the ESTIMATE algorithm, immune and stromal scores were obtained for every TCGA tumor sample. Next, pearson correlation test was done for identifying correlation of expression levels of BMPs and immune/stromal scores. For further analyzing correlation of BMPs and pan-cancer stemness, the pearson correlation test was done on the basis of gene expression and stemness indices analysis [18]. Finally, correlations of BMPs expression and DNA methylation-based stemness scores (DNAss)/RNA-based stemness scores (RNAss) of 33 TCGA tumors were obtained.
Correlation analysis of BMPs expression and TMB, MSI and expression of immune checkpoint genes
Correlations of TMB/MSI and BMPs expression in 33 TCGA tumors were assessed via pearson correlation test, which was also employed for evaluating correlations of the expression levels of immune checkpoint genes (including PD-L1, PD-1 and CTLA4) and BMPs expression. Discoveries were shown as heatmaps via ‘pheatmap’ package in R.
Tumor IMmune Estimation Resource (TIMER)
TIMER 2.0 (
Analysis of drug sensitivity in pan-cancer
CellMiner database (
Evaluation on predictive capacity of BMPs expression in immunotherapy
The gene expression profiles and clinical information of five independent cohorts, including patients with BCa (GSE111636 and IMvigor210), melanoma (GSE78220), renal cell carcinoma (GSE67501) and our own sequencing data (TRUCE01) were downloaded for exploring predictive value of BMPs expression in immunotherapy. In our TRUCE01 study, group difference before and after treatment were compared via paired Wilcoxon test for BMPs. Discoveries were grouped into Respond and Non-respond.
Gene Set Enrichment Analysis (GSEA)
GSEA was done for identifying biological function and potential signaling pathways of BMPs in BCa via Molecular Signatures Database of KEGG gene sets.
Immune subtype, clinicopathological characteristics and TME, and analysis of BCa
For farther identifying association of BMPs expression and immune subtype and clinicopathological features of BCa, the Kruskal-Wallis test or Mann-Withney U test was employed for performing differential analyses. In addition, ‘ESTIMATE’ R package was utilized for analyzing TME to obtain estimate score profile. pearson correlation test was conducted for analyzing correlations of BMPs expression and immune score, stromal score, estimate score, DNAss, RNAss, TMB, and MSI in BCa.
Immune infiltration analysis of BCa
We firstly investigated the correlation of BMPs expression with immune infiltration level of 6 major immune cells (B cells, CD4
Cell culture
Human bladder cancer cell lines T24 and 253J-BV BCa cells were bought at Chinese Academy of Sciences Cell bank and cultured in the RPMI-1640 (BI Company) medium, supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin (Gibco) under 37∘C aired with 5% CO2.
Knockdown of BMP8A, RNA extraction, RT-qPCR and western blot validation
Of them, BMP8A were chosen for validation via qRT-PCR among 10 BCa patients with the age range being 38–84 years and the gender all being male (Data S2). The E.Z.N.A.™ Hp total RNA Kit (Omega Bio-Tek, Inc.) was employed for the extraction of total RNA from cell and tissue samples. Subsequently, total RNAs were used for reversing transcription via cDNA synthesis kit (Roche Diagnostic Co.). RT-qPCR was conducted via FastStart Universal SYBR Green Master Mix with ROX (Roche) on ABI 7900HT fast real-time PCR system (Applied Biosystems). Primers for BMP8A PCR and BMP8A siRNAs were synthesized from GenePharma (Suzhou, China) and presented as below.
The PCR primers sequences of BMP8A: forward, 5’-CACCCTTCTCATCTGGATCG-3’; reverse, 5’-CAGGAAGTAGGCACCGAGAG-3’. GAPDH was utilized as control gene; forward primer, 5’-CGGAGTCAACGGATTTGGTC-3’; reverse primer, 5’-TTCCCGTTCTCAGCCTTGAC-3’. siRNA sequences of BMP8A: siRNA1, sense 5’-GGGUGCAGUUAGCAUAUUATT-3’ and antisense 5’-UAAUAUGCUAACUGCACCCTT-3’; siRNA2, sense 5’-GUGUUAGGAGACAGGUAUUTT-3’ and antisense 5’-AAUACCUGUCUCCUAACACTT-3’; and negative control (NC), sense 5’-UUCUCCGAACGUGUCACGUTT-3’ and antisense 5’-ACGUGACACGUUCGGAGAATT-3’. siRNA2, which had the highest knockdown efficiency, as shown by RT-qPCR, was used for subsequent functional experiments. Every PCR experiment was done in triplicate. For western blot, a 1% protease inhibitor was added to the RIPA lysate for lysis of 5 paired pairs of carcinoma and adjacent tissues. 30
Cell Counting Kit (CCK)-8, clone formation, and Transwell migration and invasion assays
A total of 1.5
For Transwell assay, 4
Statistical analysis
The enumeration data are presented in percentage, and the metrical variables are expressed by median and range. The Wilcoxon signed ranks test was utilized for comparing two groups, while Kruskal-Wallis test was done for comparing over two groups. Pearson’s correlation coefficient test was employed for analyzing correlations. K-M curve analysis was done for evaluating difference in survival between high- and low-expression groups. Statistical significance was calculated between different groups via log-rank test. Independent prognostic factors of BMP members for OS, DFS, DSS and PFS in 33 cancer types were identified through univariate and multivariate Cox regression.
Results
Expression of BMP family in pan-cancer
A detailed flowchart of the present research is in Fig. 1. Firstly, to investigate expression of BMP family members in pan-cancer growth, mRNA expression levels of 9 BMP family members (BMP1-8A and BMP8B) in TCGA database were investigated (Fig. 2A and Fig. S1). The differential expression of every BMP family member in cancer tissues and the corresponding noncancerous normal tissues is detailed as Fig. S2. Results suggested that BMP1, BMP8A and BMP8B expression levels were higher in multiple cancer groups, while BMP2-7 expression levels were downregulated. A heat map was utilized for illustrating expression levels of BMP family members (Fig. 2B). Correlation between BMP family members was analyzed in Fig. 2C. BMP8A expression positively related to BMP8B expression (Pearson’s cor
Flow diagram of research on BMP family genes’ function on human cancer. Firstly, present study analyzed expression pattern and prognostic value of BMP family expression among 33 cancer types. Next, present study assessed correlation of BMPs expression with clinical traits, immunological properties, sensitivity to programmed death 1/programmed death ligand 1 immunotherapy or small molecule agents, and impact of BMPs genomic variation on survival and expression level of corresponding gene from pan-cancer dimension. What’s more, potential functional roles of BMP family in bladder cancer were assessed more specifically. Finally, reverse transcription-quantitative PCR expression validation and in vitro functional experiments related to BMP8A were performed. BMP, bone morphogenetic protein.
Patients suffering from tumors were divided into two groups via median expression level of each BMP family member. Firstly, gene expression profile data were used for investigating association of BMPs expression and prognosis among 33 TCGA tumor samples. K-M survival analysis demonstrated different expression levels of BMP family members greatly related to patient OS, DFS, DSS and PFS in multiple types of cancer (Table SI and Data S3). Secondly, univariate Cox regression analysis showed expression of BMP family members related to OS in many cancer types (Fig. 3A). Same method was used for analyzing DFS, DSS and PFS in 33 TCGA tumors, and univariate Cox regression revealed similar prognostic discoveries (Fig. 3B–D). Subsequently, multivariate Cox regression analyses of BMP family members combined with common clinical traits (i.e., age, gender and tumor staging) were performed for assessing whether individual of BMP genes was an independent prognostic predictor of OS, DFS, DSS and PFS (Tables 1–4). Among them, BMP1 can serve as a unfavorable prognostic indicator independent from other clinicopathological factors regarding OS, DFS, DSS and PFS in three or more varied types of tumors with (HR
Analysis on BMP family in terms of CNVs, somatic mutations and expression in pan-cancer
Mutations, CNVs and SNVs of BMPs family were analyzed using the cBioPortal tool in pan-cancer. Between all these alterations, a clear amplification pattern of the BMP family was estimated to be 15.95% in pan-cancer, which contained 2,683 samples in the ICGC/TCGA dataset (Fig. 4A). Furthermore, TCGA dataset suggested that mutation, amplification and deep deletion were the most important factors for alterations in different cancer groups, including melanoma, colorectal cancer, esophagogastric cancer, ovarian cancer, lung cancer, breast cancer and BCa (Fig. 4B and Fig. S3). As shown in Fig. 4C, the genetic alteration percentages of BMP family members for pan-cancer varied from 2.6 to 11% for individual
BMP family differential expression across pan-cancer. (A) Boxplot of BMP family expression in The Cancer Genome Atlas cancer tissues. (B) Heat map depicting BMPs family expression in cancer tissues and corresponding noncancerous normal tissues. (C) Pearson correlation of BMP family members. BMP, bone morphogenetic protein.
Association between bone morphogenetic protein family expression and clinical prognosis in pan-cancer. Forest plots were calculated via univariate Cox regression. (A) Overall survival. (B) Disease-free survival. (C) Disease-specific survival. (D) Progression-free survival.
Multivariate Cox-regression analysis on every BMP genes and common clinical characteristics for OS acorss diverse tumor types
Multivariate Cox-regression analysis of each BMP genes and common clinical characteristics for DFS acorss different tumor types
Multivariate Cox-regression analysis of each BMP genes and common clinical characteristics for DSS acorss different tumor types
Multivariate Cox-regression analysis of each BMP genes and common clinical characteristics for PFS acorss different tumor types
Use of the cBioPortal database to determine change of BMP family members in pan-cancer. (A) Change frequency of BMP family at overall level in pan-cancer. (B) BMP family members alteration level in diverse cancer types. (C) A visual display of genomic alteration on the basis of BMPs. (D) Kaplan-Meier curve analysis comparing overall survival in cases with and without BMP family members changes. BMP, bone morphogenetic protein.
genes. The results of K-M curve and log-rank test displayed great statistical difference in OS in cases with and without BMP family gene alterations (Fig. 4D). Similarly, TCGA dataset suggested shallow deletions were more predominant relative to deep deletion for certain BMPs. What’s more, amplification and gain were more prevalent in patients suffering from pan-cancer, as seen in ICGC/TCGA Pan-Cancer Analysis of Whole Genomes Consortium, 2020 (Fig. 5). Also, non-mutated area was bigger relative to that of missense mutation. Limited truncations were found in ICGC/TCGA pan-cancer analysis, and further evaluation showed no significant differences between SNVs and BMPs expression levels (Fig. S4). Overall, these data indicated CNVs in pan-cancer can primarily result in BMPs overexpression.
(A–I) Analysis of association between expression of each bone morphogenetic protein family member and copy number alterations using International Cancer Genome Consortium/The Cancer Genome Atlas pan-cancer datasets (2,683 samples) from cBioPortal.
In general, tumor stage is the most critical factor associated with poor prognosis. Therefore, TCGA database was utilized for exploring association of BMPs family in pan-cancer clinical stage and patient response after initial treatment. discoveries proved BMPs expression levels in some tumor tissues were greatly different in diverse clinical stages. In addition, after initial treatment for several cancer types, there existed significant association of BMPs expression and outcome (Fig. S5).
Correlation between expression of BMP family members, and immune subtypes, TME and stemness indices
From Fig. 6A, expression levels of different BMP family members displayed significant difference in six immune subtypes in the research of David et al. [19] (C1 (wound healing), C2 (INF-r dominance), C3 (inflammation), C4 (lymphocyte depletion), C5 (immune quiet) as well as C6 (TGF
Correlation between expression of different BMP family genes, and TMB, MSI and immune checkpoint genes in pan-cancer
Association between TMB, MSI and BMP gene expression was examined in numerous cancer types. Discoveries indicated BMPs expression significantly related to TMB in different cancer types (Fig. 7A and Fig. S6). Correlation of BMPs gene expression and MSI was analysed in 33 cancer types, and BMPs expression greatly related to MSI in various cancer types (Fig. 7B and Fig. S7). Tumor immunotherapy is the focus of current research [22]. Therefore, immune checkpoint genes were extracted, and association between expression of different BMPs and these immune checkpoint genes was explored. Results displayed expression of different BMPs related greatly to expression of these immune checkpoint genes in various types of cancer (Fig. 6C and Fig. S8).
Drug sensitivity analysis on the BMP family
Based on drug-resistance role reported for the BMP family in cancer [23, 24], the present study further explored potential correlation of drug sensitivity and BMPs expression via CellMiner database. Discoveries displayed expression of different BMP genes greatly related to a variety of drug sensitivities. Notably, BMP8A expression positively related to drug sensitivity towards megestrol acetate, isotretinoin, and denileukin diftitox (trade name Ontak), while BMP3 expression negatively related to drug sensitivity towards ixazomib citrate, bortezomib, midostaurin and AT-13387. BMP4 expression also negatively related to drug sensitivity towards cisplatin and carboplatin. With their
Correlation of BMPs with immune subtypes, tumor microenvironment and stemness indices. (A) Correlation between expression of BMP family and immune subtypes in The Cancer Genome Atlas. (B) Different algorithms were utilized for exploring potential correlation between expression level of BMPs and infiltration level of CD8
Association between TMB, MSI, mRNA expression of immune checkpoints and expression levels of BMP family members in various tumors in The Cancer Genome Atlas database. (A and B) Correlation between expression of different BMPs and (A) TMB and (B) MSI. (C) Correlation between BMPs expression levels and mRNA expression of known immune checkpoints’ (like programmed death 1, programmed death ligand 1 and anticytotoxic T lymphocyte antigen 4) in multiple tumors. TMB, tumor mutation burden; MSI, microsatellite instability; BMP, bone morphogenetic protein.
Correlation between drug sensitivity and BMP family among different cancer types in The Cancer Genome Atlas. (A) Top 25 drugs significantly related to BMP family genes were obtained (
Expression level of BMP family genes in groups of patients with a different immunotherapy response status. (A) BMP family genes expression in different cohorts (including IMvigor210, GSE111636, GSE78220, GSE67501 and mRNA sequencing of our TRUCE-01 clinical cohort). (B–E) Changes in BMPs expression before and after immunotherapy (TRUCE-01) for bladder cancer. BMP, bone morphogenetic protein.
Table SII. Correlation of sensitivity of commonly applied chemotherapeutic drugs for BCa and expression of BMP family members was also analyzed (
Immunotherapy induced by blockade of PD-L1/PD-1 is a breakthrough for cancer therapy [25]. Thus, we analyzed association between BMP family genes expression and effectiveness of immunotherapy in confirmatory cohorts of BCa (GSE111636 and IMvigor210), melanoma (GSE78220) and renal cell carcinoma (GSE67501). Patients were classified into high- or low-expression via median expression level of BMPs. It was found that expression level of BMP1/BMP8A negatively related to objective response to anti-PD-L1 treatment, and patients who responded to immunotherapy exhibited lower expression in Imvigor210 cohort. A similar result was seen for BMP1 in GSE111636 cohort (
GSEA analysis on BMP family members expression in BCa
Based on median expression levels of BMPs in BCa, samples were classified into high- or low-expression. For exploring potential molecular mechanisms underlying the action of BMP family members, GSEA was carried out to determine which KEGG or Hallmark pathways were enriched in high- and low-expression groups. Discoveries showed multiple tumor-related and immune-related pathways were differentially enriched in different BMPs in high- and low-expression groups in bladder tumors, ‘Cytokine cytokine receptor interaction’, ‘Epithelial mesenchymal transition’, ‘Angiogenesis’, ‘Neuroactive ligand receptor interaction’, ‘Calcium signaling pathway’, ‘JAK STAT signaling pathway’, ‘Inflammatory response’, ‘Interferon alpha/gamma response’, ‘Reactive oxygen species pathway’, ‘Natural killer cell mediated cytotoxicity’, ‘TGF-
Correlation between expression of BMP family, and immune subtypes and clinicopathological characteristics of BCa
The present study compared expression levels of BMPs between 5 immune subtypes (C1–C4, C6) in TCGA BCa cohort. Discoveries displayed expression levels of BMP family members, except for BMP4 and BMP8A, exhibited significant differences among the subtypes C1–C4 and C6 (Fig. 10A). Furthermore, association between BMPs expression and clinicopathological features of patients with BCa, age, sex, tumor grade, TNM stage and tumor subtypes (non-papillary or papillary) included, was explored (Fig. 10B–E and Fig. S11). Results demonstrated expression levels of certain BMP family members in bladder tumor tissues were greatly different among diverse tumor grades, stages subtypes. Notably, with an increase in tumor grade, tumor stage and non-papillary subtype, expression of BMP1, BMP6 and BMP8A/BMP8B was also increased (Fig. 10B–E).
Correlation between BMP family gene expression, and stemness indices and TME in BCa
From Fig. 11, only expression of BMP8B in patients with BCa was greatly positively correlated with BCa stemness indices (with RNA expression data) (RNAss) (
Furthermore, correlation between BMPs expression and infiltration of 6 major immune cells (including B, CD8
Correlation between expression of BMP family members and immune subtypes and clinicopathological characteristics. (A) Distribution of the expression levels for BMPs by immune subtypes: C1 (wound healing), C2 (IFN-
Correlation between bone morphogenetic proteins expression and Cancer stem cell index (including RNAss, and DNAss), Tumor microenvironment components (including StromaScore, ImmuneScore, ESTIMATEScore), tumor mutation burden and microsatellite instability based on The Cancer Genome Atlas database.
Associations of bone morphogenetic proteins expression with immune infiltration level of the 6 commonest immune cell types in bladder cancer based on the Tumor IMmune Estimation Resource database.
Association between BMPs family CNVs, mutations and immune infiltrations in BCa. (A) BMP genes CNVs in BCa. (B) BMP genes mutations in BCa. CNVs, copy number variations; BMP, bone morphogenetic protein; BCa, bladder cancer.
The association between BMP family CNVs, mutations and immune infiltration was investigated in BCa. Figure 13A shows significant associations between changes in the copy number of BMP family members and immune infiltrates in BCa. In particular, amplification of BMP3 and deletion of BMP4 related to substantially lower levels of immune infiltrates (CD8
BCa samples were then categorized according to BMPs mutation status, and different algorithms (i.e., TIMER, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, MCPCOUNTER, XCELL and EPIC) were employed for investigating their potential association with infiltration level of diverse immune cells. As shown in Fig. 13B, immune infiltration was predominantly found within BMP6-8A; however, these three family members related to distinct immune cell populations. Higher level mutations of BMP7 were enriched in CD8
Validation of the expression of BMP8A in BCa tissues and biological function of BMP8A in vitro
Biological functions of BMP8A in BCa. (A) BMP8A expression in 10 pairs of samples was analyzed by reverse transcription-quantitative PCR. (B) BMP8A expression in five pairs of BCa and adjacent noncancerous tissues detected by western blot. (C) Verification of knockdown efficiency of BMP8A in T24 and 253J-BV cells. (D) Cell Counting Kit-8 proliferation assay. (E) Clone formation assays for BMP8A knockdown in T24 and 253J-BV cells. (F) Cell migration and invasion were evaluated by Transwell assay. ∗P< 0.05, ∗∗P< 0.01, ∗∗∗P< 0.001, ∗∗∗∗P< 0.0001. BMP, bone morphogenetic protein; BCa, bladder cancer.
Finally, we analyzed BMP8A expression levels in BCa cell lines and tissue samples by RT-qPCR and western blot. From Fig. 14A and B, BMP8A expression was higher in bladder tumor tissues relative to normal tissues used as control (
BMPs have attracted large attention in cancer research, since they attend numerous biological processes like cancer development. Various TME factors strongly affecting tumorigenesis interact with BMPs like microRNAs, mutations or drug treatment [14]. A series of conflicting research have shown same BMP ligand can act diversely relying on cancer types, and its numerous members of BMP family appear to own different functions [26, 27]. In spite of the detection of changed BMPs expression among numerous cancer forms [12, 28] and its use as one marker for excellent prognosis in cancer therapy [29], there is still a lack of comprehensive and complete profiling of BMPs expression in pan-cancer and its prognostic value using data mining processes. We evaluated expression pattern of nine BMP family genes and conducted multi-omics analysis for understanding role of BMPs as prognostic marker genes or oncogenes across 33 cancer types, particularly BCa.
Detailed assessment of expression pattern of BMPs in many cancer types displayed BMP1, BMP8A and BMP8B expression was higher in numerous cancer groups relative to normal tissues, whereas BMP2-7 expression was downregulated in cancer. BMP1 expression was previously reported to be upregulated in osteosarcoma, and gastric, lung and kidney cancer [30], while BMP2, BMP3, BMP5 and BMP6 expression levels were found to be downregulated in breast cancer [26], which matches the current findings. In addition, the present results confirmed that not only BMP1-2, but also BMP4-7, BMP8A and BMP8B, showed differential expression in BCa samples. The present study provided a complete map of BMPs expression pattern across human tumors.
To explore role of BMP family members in pan-cancer, the current study investigated clinicopathological features and prognostic status of different BMP family members among 33 cancer types, and displayed potential prognostic value of BMPs depending on the cancer type by KM plotter, univariate Cox regression analysis. Moreover, numerous BMP family genes were identified to be independent prognostic factors associated with OS, DFS, DSS, or PFS via multivariate Cox regression analysis (Tables 1–4;
Cancer progression is affected by somatically acquired genetic, epigenetic, transcriptomic and proteomic alterations [29]. The somatic loss-of-function or gain-of-function alterations happen in specific genomic regions, indicating their potential inhibitory or carcinogenic roles separately [31]. Irreparable structural mutations in cells are primary reason for cancer in humans. Therefore, current study employed cBioPortal to display the mutations, CNVs and mutant mRNA expression. The discoveries showed amplification is the most frequent mutation type among different cancer types, and CNVs drive abnormal expression of BMP genes, which is in line with previously reported results [26]. Survival analysis on pan-cancer further indicated alterations in expression of BMP family genes greatly related to lower OS compared with that of unaltered BMPs (Fig. 4D). Therefore, these alterations are possibly vital for cancer progression and prognosis.
Previous research revealed patients with cancer could be divided into 6 immune infiltration subtypes [32]. The present study found that BMP family members were differentially expressed in different immune subtypes of pan-cancer and BCa. The TME mainly includes tumor, stromal and immune cells, and extracellular matrix [33], which are vital for tumor growth and development, therapeutic resistance and clinical outcome [34, 35, 36]. The current study analyzed pearman correlation of BMP family gene expression and tumor stemness indices among 33 cancer types. Discoveries displayed the higher the expression of BMPs, the higher the content of immune and stromal cells, and the lower the content of tumor cells. Expression of BMP8B in patients with BCa significantly positively related to BCa stemness indices (with RNA expression data), whereas expression of BMP1, BMP2 and BMP4-8A greatly negatively related to BCa stemness indices (RNAss), displaying BMPs are possible vital for stemness maintenance. The higher the expression of BMP8B, the more features of tumor BCa stem cells and the lower tumor BCa differentiation, while the higher the expression of BMP1, BMP2 and BMP4-8A, the fewer the characteristics of tumor BCa stem cells and the higher tumor BCa differentiation. It has been confirmed that tumor-related stromal cells attend progression and invasion of numerous cancer types [37]. Elimination of immune cells in the TME and resistance to chemotherapy were influenced by TGF-
ICIs block interaction of immune checkpoints and their ligands, break immune tolerance, boost immune cell activity and enhance the immune clearance of tumor cells, thus inhibiting occurrence and growth of tumors [39]. TMB is an independent biomarker identified in various tumor immunotherapies and can be utilized for predicting immunotherapy efficacy [40, 41]. Similarly, MSI has gained widespread attention as a promising predictive biomarker for immunotherapy efficacy, particularly in colorectal and lung cancers [42, 43]. Farther analysis on correlation of BMPs, TMB and MSI was done in present study, and the results demonstrated that BMPs expression greatly related to TMB and MSI in certain cancer types. What’s more, expression of BMPs greatly related to expression of multiple immune checkpoint genes (PD-1, PD-L1 and CTLA4 included) among most cancer types. Above discoveries powerfully suggested BMPs expression may influence response to immune checkpoint therapy among patients suffering from cancer, which would shed new light on immunotherapy prognosis.
Tislelizumab is a fresh humanized monoclonal antibody that acts as an inhibitor of PD-1, and displays a predictable and manageable safety/tolerability profile among patients suffering from PD-L1
Conclusion
In summary, the current study confirmed that BMP family genes were differentially expressed in numerous cancer types, and expression of BMPs related to prognosis, clinicopathological characteristics, mutations, immune subtypes, TME, tumor drug sensitivity, immune cell infiltration, immunotherapy indicators (TMB and MSI) and immune checkpoint gene expression in 33 cancer types, including BCa. Moreover, expression level of certain BMP family genes related to objective response to immunotherapy. Both BMPs expressions and genomic alterations is of importance for the evaluation of prognosis and immune characteristics. By GSEA, multiple functions and pathways related to immunity were enriched. The results of RT-qPCR and western blot analysis and in vitro experiments confirmed that BMP8A functioned as an oncogene in BCa. Therefore, BMPs play important roles in human cancer and may be potential biomarkers for prognosis and immunotherapy, as well as drug candidates in BCa.
Funding
The present research gained support from Tianjin Municipal Health Industry Key Project (grant no. TJWJ2022XK014), Scientific Research Project of Tianjin Municipal Education Commission (grant no. 2022ZD069), Youth Fund of Tianjin Medical University Second Hospital (grant no. 2022ydey15), Tianjin Institute of Urology Talent Funding Program (grant no. MYSRC202310), Tianjin Health Science and Technology Project (grant no. ZC20119), and The Talents Cultivated Project of Department of Urology, the Second Hospital of Tianjin Medical University (grant no. MNRC202313).
Availability of data and materials
We displayed all the main data via tables in supplementary or additional files. Datasets supporting results of our paper are available in public database from TCGA and NCBI-GEO (GSE111636, GSE78220, GSE67501, GSE176307, IMvigor210).
The raw sequence data of TRUCE01 are being uploaded to Genome Sequence Archive (GSA) database (
Author contributions
HH and CS designed the present study. HH and CC collected and verified the raw data. YZ, CC and YL wrote the manuscript. CS screened database and collected data. CS and ZZ conducted bioinformatic analyses. HH, CS, ZW and YQ revised the manuscript. GZ offered critical comments. Authors all approved the submitted version.
Ethics approval and consent to participate
Our research was conducted complying with Declaration of Helsinki and obtianed approval from Ethics Committee of the Second Affiliated Hospital of Tianjin Medical University (approval no. KY2021K003). All patients offered written informed consent for participation.
Patient consent for publication
Patients offered written informed consent for publication.
Supplementary data
The supplementary files are available to download from http://dx.doi.org/10.3233/THC-232004.
Footnotes
Acknowledgments
Not applicable.
Conflict of interest
Authors declare none competing interests exist.