Gene expression profiles of RBP7 in normal and cancer tissues
We utilized the HPA database to analyse the mRNA and protein expression profiles of RBP7 in human normal tissues and found that RBP7 expression varied significantly in different human tissues with relatively high expression in breast and adipose tissues (Figure 1A). Consistently, the protein expression of RBP7 was highly expressed in the breast and adipose tissues (Figure 1B). To assess the variation in RBP7 expression in pan-cancer, we performed analysis with the GEPIA online server from the TCGA and GTEx projects. Compared to the matched normal tissues, RBP7 was highly expressed in adrenocortical carcinoma (ACC), kidney renal clear cell carcinoma (KICH) and liver hepatocellular carcinoma (LIHC). In contrast, RBP7 expression was relatively lower in urothelial bladder carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and so on (Figure 1C).
To further verify the significance of RBP7 expression in cancers, the fluxionary expression of RBP7 in tumour and normal tissues was analysed by using ONCOMINE. We found that RBP7 was overexpressed in liver cancer and lymphoma, but decreased expression of RBP7 was found in brain and CNS cancer, BRCA, oesophageal cancer, head and neck cancer, leukaemia, and ovarian cancer (Figure S1A).
Prognostic value of RBP7 expression in BRCA
To investigate the role of RBP7 expression in BRCA, the UALCAN database was utilized to analyse the expression of RBP7 in 114 normal tissues and 1097 primary BRCA tissues. The TCGA database results revealed that RBP7 transcript levels were significantly reduced in patients with BRCA (Figure 2A), which was further validated by the GSE37751 dataset from the GEO database (Figure S1B). Then, we discovered that the protein expression of RBP7 was higher in normal tissues through CPTAC database (Figure 2B), which was consistent with the result of RBP7 expression in the TCGA database.
We subsequently utilized the transcriptomic sequencing data in the GEPIA database to assess the prognostic value of RBP7 in BRCA and found that a high level of expression of RBP7 was favourable to the prognosis of BRCA (Figure 2C). Survival analysis was performed with the GSE20685 (Figure 2D) and GSE42568 (Figure 2E) datasets from the GEO database, which showed that Shorter OS was observed in patients with lower expression of RBP7. Furthermore, we used FROGgeneV2 to confirm the association between RBP7 expression and survival rates of BRCA patients, and the results are consistent with the above (Figure 2F). These results reveal that there is a significant association between RBP7 expression and BRCA prognosis and that RBP7has a protective effect on BRCA prognosis.
RBP7 expression in different cells from BRCA tissues
It demonstrated that RBP7 exhibits cell type-specific expression in endothelial and epithelial cells in TISCH database (Figure 3A and 3B). A heat map of the gene module analysis depicts the expression of RBP7 in different cell types in BRCA datasets in which endothelial and epithelial cells are characterized by high expression of RBP7 (Figure S1C). Then, we utilized the immunohistochemistry (IHC) data detected by the HPA-034749 antibody from the HPA database to determine the protein expression of RBP7 in BRCA and adjacent/healthy tissues. The results showed that adipocytes were highly stained in normal breast tissues, while glandular and myoepithelial cells were mildly stained, mainly in the nucleus (Figure 3C). In the BRCA samples, the expression level of RBP7 in tumour cells was ranked as weak, moderate and strong, which was scored by the staining intensity in the pathological IHC (Figure 3D-3F). Interestingly, as a nuclear receptor, RBP7 was found to be mainly localized in the nucleus, indicating the vital role of RBP7 in the regulation of gene expression in epithelial cells of BRCA tissues.
RBP7 coexpression networks in BRCA
To in-depth knowledge the biological meaning of RBP7 in BRCA, RBP7 coexpression genes were analysed by the functional module of LinkedOmics. The top 50 significant genes that were positively and negatively correlated with RBP7 were selected as heat maps (Figure 4A and 4B) in which RBP7 displayed a strong positive related with the expression of FAM107A, GPIHBP1 and FXYD1. Remarkably, the top 50 negatively coexpressed genes had a high probability of being high-risk markers in BRCA, of which 15 genes had significantly high HRs (p value < 0.05). In contrast, there were no genes with high HRs (p< 0.05) in the top 50 positively coexpressed genes. These results further confirmed that RBP7 performs a protective role in the progression of BRCA (Figure 4C).
GO term annotation of biological processes showed that RBP7 coexpressed genes mainly participate in the adrenergic signalling pathway, excretion, endothelium development, regulation of transporter activity, cell communication by electrical coupling and G protein-coupled receptor signalling pathway, with inhibition of the biological processes including double-strand break repair, cargo loading into vesicle, DNA conformation change, ncRNA transcription and protein localization to chromosome (Figure 4D). KEGG pathway analysis showed that there was an enrichment in the regulation of lipolysis in adipocytes, PPAR signalling pathway, ovarian steroidogenesis, and drug metabolism (Figure 4E), indicating a widespread impact of RBP7 on the global transcriptome.
RBP7 methylation in BRCA
We displayed hierarchical clustering analysis of RBP7 mRNA expression related to DNA methylation by using the UCSC Cancer Genomics Browser. The results showed that RBP7 methylation mainly occurred in primary BRCA, and hypermethylation at the promoter region downregulated the mRNA expression of RBP7 (Figure 5A), indicating a potential correlation between transcription level and methylation on the promoter of RBP7.
Next, we performed methylation analysis with the Methsurv database and found that 4 methylation probes, namely, cg20413202, cg03406535, cg10796749 and cg14202757, in the promoter of RBP7 were highly methylated in BRCA (Figure 5B). According to this standard, we found that the β values of 3 of the 4 detected methylation probes, i.e., cg20413202 (β value=0.947), cg10796749 (β value=0.798) and cg14202757 (β value=0.694), were higher than 0.6, indicating that almost complete methylation occurred on the promoter of RBP7 (Figure S2A-D). Consistently, we found that most of the median β values of different clinical stages were also above 0.6 (Figure S2E-H), suggesting that promoter methylation leads to the possession of RBP7 gene transcription in BRCA.
Subsequently, we used the SMART App to verify the Spearman correlation between transcription level and DNA methylation of the probes of RBP7 in BRCA. RBP7 expression was significantly negatively correlated with the methylation probes cg03406535, cg27083689, cg18086187, cg03994053, cg27561954, cg15090005 and cg07224455 (Figure 5C). The aggregation plot for the 12 methylation probes performed that there was a significant negative correlation between RBP7 transcription levels and DNA methylation (Figure S1D).
Clinicopathological association of RBP7 and its prognostic value
BRCA is a complex disease with various morphological, clinical and molecular features. Molecular subtypes and optimal treatments for BRCA are usually based on immunohistochemical markers such as ER, progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2). We checked the relevance of RBP7 expression and different clinicopathological features by using the web-based tool bc-GenExMiner. We found that the expression of RBP7 was highly expressed in ER+ BRCA patients compared with ER-negative (ER-) patients (Figure 6A). RBP7 expression was significantly decreased in PR-negative compared with in PR-positive in BRCA patients (Figure 6B), whereas RBP7 expression was higher in HER2-negative BRCA patients than in HER2-positive patients (Figure 6C). To further determine the correlation of RBP7 expression and hormone receptors (HRs), we utilized DNA microarray data to perform correlation analysis of RBP7 expression with different combinations of ER and PR expression statuses, i.e., ER+/PR+, ER+/PR-, ER-/PR+ and ER-/PR-. The results showed that there were remarkable differences in RBP7 mRNA expression in some hormone receptor combinations, i.e., ER+/PR+ vs. ER-/PR-, p<0.01 and ER+/PR+ vs. ER+/PR-, p<0.01 (Figure 6D).
Furthermore, we performed survival analysis of BRCA patients with different ER/PRcombinations. Downregulated RBP7 expression was only significantly associated with poor prognosis in ER+/PR+ and ER+/PR- patients but not in ER-/PR+ and ER-/PR- BRCA patients (Figure 6E-6L). OSbrca was used to verify the prognostic value of RBP7 in these 4 groups with the TCGA database. Consistently, we discovered that down-regulation of RBP7 expression was significantly correlated with shorter OS in ER+/PR+ and ER+/PR- patients (Figure S3).
Potential regulatory mechanisms and target drugs of RBP7 in BRCA
To further explore the PPI network of RBP7 in BRCA, we used STRING to identify genes that may interact with RBP7. The interactions between RBP7 and proteins encoded by the functional genes, including SLC45A1, SLC25A33, UBE4B, TMEM201, NMNAT1, SOCS7, SOCS4, GPR157, PIK3R3 and POLR2G, are shown in Figure 7A. Interestingly, in this interacting network, PIK3R3, a regulatory subunit of phosphatidylinositol 3-kinase (PI3K), was an important factor in BRCA. Therefore, we further analysed the correlation between RBP7 and PIK3R3 in BRCA and found that the expression levels of RBP7 and PIK3R3 were negatively correlated (R = -0.231, p = 6.79e-15) (Figure 7B). Then, we acquired ER+ BRCA data from the TCGA database, which were quarterly ranked according to the expression level of RBP7. The high and low RBP7 expression groups were defined as the first and fourth quarters of ER+ BRCA data, respectively, and the differential genes (|log2(FC)|>1) between these two groups were analysed by using the Limma software package for display as a volcano plot (Figure 7C).
KEGG pathway analysis was conducted to investigate the functional implications of the DEGs, and it was found that several tumour-related pathways, such as the PPAR signalling pathway, regulation of lipolysis in adipocytes and tyrosine metabolism, were significantly enriched (Figure 7D). In addition, the PI3K-AKT signalling pathway, which is important in ER+ BRCA, was also enriched (Figure 7D).
Furthermore, we used the CARE database to analyse the association of the molecular alteration of RBP7 with drug efficacy and found that RBP7 was negatively associated with drug efficacy in the CCLE, CGP and CTRP databases (Figure 7E). Interestingly, we found that only one drug (nilotinib) was common among the resistant drugs from these 3 databases (Figure 7F). Finally, we visualized the binding site of RBP7 with nilotinib by SwissDock (Figure 7G).