RNA-seq data based on The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov) was obtained and compiled from The Cancer Genome Atlas Breast Invasive Carcinoma (TCGA-BRCA) (breast invasive cancer) project STAR process. The data was extracted in the TPM format to remove non-clinical and duplicate information and then processed as log2 (value+1). Data was visualized using the ggplot2 package. We downloaded data from two Gene Expression Omnibus (GEO) datasets to obtain more clinical information related to breast cancer patients: GSE42568 (Clarke et al., 2013) (104 breast cancer and 17 normal breast biopsies) and GSE70947 (Barrett et al., 2013) (Age and estrogen-dependent inflammation in 148 breast adenocarcinoma and 148 normal breast tissue). All microarray data were called with the robust multichip average (RMA) method. We used the TIMER2.0 (http://timer.cistrome.org/) database for analyzing RNF157 levels among diverse cancers together with non-carcinoma samples (Li et al., 2020). A probability cutoff of 0.05 was applied.

We employed cBioPortal web to carry out genetic (https://www.cbioportal.org/) alteration analyses (Gao et al., 2013). In addition, a “curated set of non-redundant studies” (184 studies, 10528 samples) from the “query” section was selected, and “RNF157” was imported for queries regarding genetic alteration features related to RNF157. Using the “cancer types summary” module, the alteration frequency, mutation type, and copy number alteration (CAN) results in many cancer types were obtained. Moreover, data on mutated sites of RNF157 were observed in the “mutations” module. We also used the “comparison/s-urvival” module for obtaining information on the differences in the overall survival (OS) and the relapse-free survival of patients with/without RNF157 genetic alterations.

We carried out a bulk correlation analysis of RNF157 and all other molecular data in the TCGA database to find the biological pathways enriched by RNF157 using breast cancer transcriptome sequencing. The parametrization settings were |Cor| > 0.3, and p < 0.05. We then conducted an enrichment of the first 50 genes related to RNF157 in breast cancer using the R cluster Profiler package (Yu et al., 2012) for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (Liao et al., 2019). Statistical significance was set at the corrected p < 0.05. Finally, we utilized the ggplot2 package for creating a visual representation of the results obtained from the enrichment analysis.

Breast cancer cell lines namely MDA-MB-231, HS578T, UACC812, MCF7, and T47D were cultivated in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) (Gibco, Detroit, MI, USA). HCC1954 cells were cultured with RPMI1640 (Gibco, Detroit, MI, USA) supplemented with 10% FBS. Normal breast epithelial MCF10A cells were cultivated within MEBM (Lonza, Switzerland) that contained 10% FBS. Cells were incubated at 37°C and an atmosphere of 5% CO₂.

Total cellular protein extracts were obtained by lysing the cells with concentrated RIPA lysis buffer containing protease inhibitors. The extracted proteins were then separated using 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes. We added 5% skim milk to the membrane for 1h at room temperature to block non-specific binding. The membranes were then probed using primary antibodies against RNF157 (1:1000, #26189-1-AP, Abnova, Taiwan) and β-actin (1:5000, #3700T, Abcam, USA) for 24 h at 4°C. Western blots were observed and measured for intensity (Guo et al., 2015; Bai et al., 2019).

We used a total of 34 breast cancer tissues and 20 normal breast tissues for immunohistochemical (IHC) staining. The primary antibody against RNF157 (1:300 dilution, #bs-9226R; Bioss, China) was incubated with the samples at 4°C overnight. The sections were then treated with peroxidase-conjugated goat anti-rabbit secondary antibody (1:2,000 dilution, A0181; Beyotime, Shanghai, China) and incubated at room temperature for 2 h. Thereafter, 3,3′-Diaminobenzidine was added for IHC staining, followed by counterstaining with hematoxylin (Beyotime, Shanghai, China). The IHC staining results were analyzed and scored by two pathologists blinded to sample origin. The differences in RNF157 expression between breast cancer tissues and non-carcinoma counterparts were assessed based on the mean H-score.

We have presented a flowchart to demonstrate our analysis methodology (Fig. 1). The analysis of TCGA RNA-seq data in the TIMER database revealed a significant downregulation of RNF157 mRNA expression in breast cancer, glioblastoma multiforme, kidney chromophobe, lung squamous cell carcinoma, skin cutaneous melanoma, and thyroid carcinoma compared to their respective non-cancerous counterparts (Fig. 2A). We employed the TCGA database for assessing RNF157 mRNA expression among breast cancer cases compared to expression in non-carcinoma samples. We used the TCGA database to assess RNF157 mRNA expression in breast cancer cases compared to expression in non-cancerous samples. We found that RNF157 expression markedly decreased within breast cancer samples as compared with para-carcinoma samples (p < 0.01) (Fig. 2B). An identical expression pattern was also confirmed in breast cancer samples and paired paraneoplastic tissues (Fig. 2C). We further analyzed two other independent external GEO datasets, GSE42568 and GSE70947 were analyzed to determine RNF157 mRNA expression within breast cancer tissues and non-carcinoma counterparts. The RNF157 mRNA expression level in breast cancer samples remarkably decreased compared with unpaired and paired samples (p < 0.001) (Figs. 2D and 2E).

RNF157 mRNA expression was also examined within diverse clinical categories based on the TCGA database to determine the correlation of RNF157 level with clinical characteristics of breast cancer cases. The relation of the RNF157 level with clinical characteristics in breast cancer cases is shown in Table 1. We found that the low expression of RNF157 was significantly related to the histological type (p < 0.001), progesterone receptor (PR) status (p < 0.001), ER status (p < 0.001), PAM50 (p < 0.001) and age (p < 0.001) of these patients (Figs. 2F–2I and Table 1). We then analyzed the correlation of RNF157 expression with the methylation status based on the UALCAN database to elucidate the mechanism underlying the aberrant down-regulation of RNF157 in breast cancer samples. The methylation level of RNF157 in breast cancer samples was evidently decreased compared to their para-carcinoma counterparts (Suppl. Fig. S1), indicating that the abnormally low expression of RNF157 in breast cancer may be related to significantly lower methylation levels.

We examined breast cancer tissues from 34 patients and 20 corresponding adjacent tissue samples using immunohistochemistry for exploring the possible role of RNF157 in breast carcinogenesis. The results demonstrated weaker positive staining for RNF157 in breast cancer samples than the non-carcinoma samples, as shown in Figs. 3A and 3B. We further investigated RNF157 protein levels among seven breast cancer cell lines (MCF10A, MCF7, T47D, UACC812, HCC1954, MDAMB231, HS578T) and compared them with that of human breast epithelial cells. We observed a decrease in the abundance of RNF157 protein in cancer cell lines, except for the MDAMB231 cell line (Fig. 3C).

To determine whether RNF157 expression is related to patient prognosis, we divided breast cancer cases derived from the TCGA database into two groups, the high and low RNF157 expression groups. The survival analysis based on the mean expression value of RNF157 was then performed. Overall survival analysis by the Kaplan Meier method suggested that low expression levels of RNF157 might predict poor OS (HR = 0.69, p = 5.7e-03) (Fig. 4A), distant metastasis-free survival (DMFS) (HR = 0.61, p = 2.9e-04) (Fig. 4B) and recurrence free survival (RFS) (HR = 0.7, p = 2.9e-06) (Fig. 4C).

We also analyzed the significance of RNF157 expression in predicting the prognosis of different breast cancer subgroups. The RNF157 down-regulation showed a positive relation with breast cancer patient cases at stage N3 (HR = 0.22, p = 0.016), with luminal B (LUMB) (HR = 0.44, p = 0.033) and with basal (HR = 0.45, p = 0.031) and were significantly associated with poor prognosis (Table 2).

The univariate logistic regression illustrated that the RNF157 level was the categorical dependent variable, which predicted the dismal prognosis-related clinicopathological features (Table 3). Further, the down-regulated RNF157 expression in breast cancer was positively related to the histological type (OR = 2.157 for infiltrating lobular carcinoma vs. infiltrating ductal carcinoma), the PR status (OR = 2.297 for indeterminate and positive vs. negative), the ER status (OR = 3.026 for indeterminate and positive vs. negative), and PAM50 (OR = 0.416 for LumA and luminal B (LUMB) and Her2 and basal vs. normal). All were significant (p < 0.01).

Univariate Cox analysis was applied to evaluate the factors influencing the OS. We found that a low expression RNF157 (HR = 0.640, 95% CI = 0.414–0.989, p = 0.045) was the predictive factor for worse OS. This was also seen for TNM stage (p < 0.001) and the histological type (p < 0.05) (Table 4). We then incorporated these factors were for multivariate Cox regression. It was observed that low RNF157 expression was an independent predictor of poor OS for breast cancer cases.

We next plotted the receiver operating characteristic (ROC) curve for investigating whether RNF157 expression could be used to distinguish breast cancer tissues and non-carcinoma counterparts. As shown in Fig. 4D, an area under the ROC curve (AUC) value of 0.831 was obtained, indicating RNF157 was a potential biomarker for diagnosing breast cancer.

We then constructed nomograms using the RNF157 expression and independent clinical risk factors based on multivariate Cox analysis to better predict the prognosis of breast cancer. The sum of the scores assigned to each clinical risk factor in the model was rescaled to a range of 1 to 100 using a point scale, depending on the degree of contribution of each clinical risk factor to the outcome variable (the magnitude of the regression coefficient). The total score was obtained by adding up the individual scores, and eventually, the survival probability for breast cancer patients at 1-, 3-, and 5-years was predicted by transforming the relation of total score with outcome event probability into a function (Fig. 4E). The calibration curve was close to the ideal curve, which suggested that the predictions were satisfactorily linear (Fig. 4F). Overall, the above data indicated that RNF157 may be a candidate prognostic biomarker of OS for breast cancer patients.

We used the cBioPortal network for studying RNF157 genetic alterations in human pan-cancer samples for understanding mutational profile of RNF157 during carcinogenesis. We found that the highest RNF157 “amplification” alteration frequency appeared for patients with breast cancer. Further, the “mutations” type was the least alteration frequency of RNF157 with breast cancer (>4% alteration frequency) (Fig. 5A). The genetic alterations of RNF157 including types, sites, and case numbers have been shown in Fig. 5B. Missense mutation represented the major genetic alteration type of RNF157, and it was detected among 85 cases. Additionally, truncating mutations occurred in six patients. Splice mutations and fusions were detected only in four cases.

The relationship of genetic alterations in RNF157 with the clinical outcome of breast cancer was then analyzed. As shown in Figs. 5C and 5D, breast cancer cases with RNF157 alterations showed poor OS (p = 0.0437) and relapse-free survival (p = 1.202e-4). These results demonstrate that genetic alterations of RNF157 in breast cancer may be involved in altered RNF157 mRNA expression and poor clinical survival.

We analyzed the co-expression pattern of with other genes in breast cancer in the TCGA database for understanding its biological activity. We screened 25 positively correlated genes that displayed the greatest Spearman correlation coefficients along with 25 negatively correlated genes, which were illustrated by the heat map (Fig. 6A). To further understand the functional implications of these 50 genes that were most associated with RNF157 in breast cancer, functional enrichment analysis using GO and KEGG was done with the R cluster Profiler package. The results indicated that the MAPK signaling pathway and several microRNAs in breast cancer were enriched among these genes (Fig. 6B).

To confirm whether RNF157 in breast cancer is associated with the 245 genes that were shown to be related to the MAPK signaling pathway (https://pathcards.genecards.org/) (Belinky et al., 2015), we constructed another heat map (Fig. 6C). The RNF157 down-regulation in breast cancer patients showed significant correlation with these genes. Moreover, the relationship of RNF157 with MAPK signaling pathway-related genes in breast cancer was further examined. We found that MAPK signaling pathway-related genes MAP3K12, ACVR1B, MAP2K5, and MAP4K3 showed a positive correlation with RNF157 (r > 0.3, p < 0.001) (Figs. 6D–6G). Collectively, RNF157 showed a close association with MAPK signaling pathway regulation in breast cancer.

Immune cells of the adaptive and innate immune systems infiltrate into the tumor microenvironment (TME) and modulate cancer development (Seager et al., 2017). The immune infiltration in breast cancer showing diverse RNF157 expression was analyzed using the TIMER database to explore the link between RNF157 expression and tumor immune response. We found that natural killer (NK) cells, CD8 + T lymphocytes, CD56 bright cells, mast cells, and eosinophils exhibited remarkably decreased infiltration levels in breast cancer cases showing RNF157 down-regulation compared with those showing RNF157 up-regulation. On the contrary, activated dendritic cells (DCs), macrophages, NK CD56 dim cells, B cells, and regulatory T cells (Tregs) displayed significantly increased infiltration levels in breast cancer cases with RNF157 down-regulation compared to those showing RNF157 up-regulation. The infiltration levels of T cells and DC cells were not significantly different in both groups (Fig. 7A). Caption and used consistently thereafter. Accepted abbreviations for statistical parameters are: P, n, SD, SEM, df, ns, ANOVA, t. Naming of chemicals should follow that given in Chemical Abstracts Service.

The relation of RNF157 levels with immune infiltration in breast cancer was also examined. The RNF157 expression showed a positive relation with the infiltration level of mast cells (Fig. 7B, r = 0. 236, p < 0.001). However, RNF157 expression showed a negative relation with macrophage (Fig. 7C, r = 0.288, p < 0.001), B cells (Fig. 7D, r = 0. 279, p < 0.01), and Treg (Fig. 7E, r = 0.227, p < 0.01) infiltration levels.

The heat map displaying significant relations of the RNF157 level and with chemokines and chemokine receptors in breast cancer are shown in Figs. 8A and 8B. We synthetically examined the relation of the RNF157 level with chemokines and chemokine receptors to understand the relationship between the RNF157 level and the migration of immune cells. RNF157 expression showed a negative relation with CCL7 (r = −0.274, p = 3.05e-20), CCL20 (r = −0.28, p = 3.27e-21), and CXCL5 (r = −0.262, p = 1.4e-18) (Figs. 8C–8E). However, RNF157 was not related to other chemokines/chemokine receptors (−0.3 < r < 0.3). Collectively, RNF157 displayed a negative relation with chemokine levels and chemokine receptor levels in breast cancer.