The data of more than 30 types of cancer patients were downloaded from the GTEx database (https://gtexportal.org/) and TCGA database (https://portal.gdc.cancer.gov/). The GSE13507 from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) and the Kaplan-Meier plotter database (http://kmplot.com/analysis/) was adopted to verify the prognostic significance of PSMD2 in BCa. Differentially genes of PSMD2 high and low BCa samples were performed by using the “Limma” package, and setting the |log2(FC)|>1 and p. adj<0.05 as the threshold.

Human BCa cell lines (UC3, T24, J82) and normal bladder cell line (SV) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and cultivated as previously reported (14), the minimum essential medium (MEM) (Corning) was used to culture UC3 and J82 cell lines, and RPMI 1640 medium (Corning) was used to culture T24 and SV cell lines. These cells were cultured with medium and 10% sterilized fetal bovine serum (FBS, BI) at 5% CO2 and 37°C in a humidified atmosphere. Transfection cell experiments were performed by using the Polyplus transfection^® reagent (Proteintech), the siRNAs were synthesized in TsingKe (Hangzhou, China) and the sequences were listed in ( Table S1 ).

Total RNA was extracted by using TRIzol agentia (Takara), the PrimeScript RT reagent Kit (Takara) was used to perform reverse transcription, the SYBR Premix Ex Taq (Takara) was used to perform PCR in thermocycler, and the relative expression levels of PSMD2 were detected via qRT-PCR method as previous described (15), GAPDH was selected as the endogenous reference. All primers were included in Table S1 .

After the transfection procedure for 48 h, transfected cells were counted and conducted wound healing and colony formation assay, for colony formation assay, 500 treated cells per well were counted and cultured in 6-well plates, after cultured for 7-14 days, cells were fixed with methanol for 10 minutes and then stained with 0.3% crystal violet for 10 minutes, 6-well plates were washed and counted. For wound healing assay, the treated cells grew to 100% confluence in 6-well plates, and sterile tip was used to make a scratch on the plate to generate wounds. Cells were then maintained in serum-free medium for 24h and visualized using a phase-contrast microscopy, the ranges of wound healing were used to assess the ability of wound healing.

Cox analysis was adopted to explore the prognosis of PSMD2 in 33 types of tumor, including BCa. The forest plots were applied to exhibit the HR and P value of each character. With the help of the “rms” package, the nomogram was constructed to assess the OS of BCa patients through integrating PSMD2 expression and other clinicopathological factors, and the goal of a nomogram is to help predict individual BCa patient’s 1-year, 3-year and 5-year survival probability.

In order to investigate the possible roles and underlying mechanisms of PSMD2 in BCa, the “Cluster Profiler” was used to perform the GO and GSEA analyses (16) to find the GO and KEGG pathways of PSMD2 in BCa. For GSEA analysis, the permutation tests were performed 1,000 times, and p-values were adjusted for multiple testing by performing the Benjamini–Hochberg procedure, p < 0.05 and FDR < 0.05 was a significantly enriched pathway.

TIME is an important regulatory factor of tumor development, and tumor immune infiltrates are main component of TIME. To carry out a calculable evaluation of tumor immune infiltrates, the connections between PSMD2 and its top 6 coexpressed genes and TIME in BCa were firstly assessed via utilizing the ssGSEA algorithm (17). Then the BCa samples were separated into two PSMD2 groups based on median PSMD2 expression in TCGA-BLCA cohort, the xCell algorithm (18) was applied to evaluate the TIME component of this two PSMD2 groups via using the “immunedeconv” package and visualized by the “pheatmap” package. Meanwhile, the tumor immune dysfunction and exclusion (TIDE) (19) algorithm was adopted to assess the therapeutic effect of immunotherapy. TIDE uses a panel of gene expression signatures to evaluate the mechanisms of tumor immune escape, including dysfunction of tumor-infiltrating cytotoxic T lymphocytes (CTLS) and CTL rejection by immunosuppressive factors. The higher the TIDE score, the lower the patients’ sensitivity to immune checkpoint blocking therapy, and shorter survival after ICB treatment. Additionally, the correlation between PSMD2 expression with immune signature in pan-cancer (20) was also evaluated via utilizing data compiled by TCGA cohorts.

GraphPad Prism 8.0 and (version 3.6.3) R were used to perform the statistics. Wilcox test was applied to conduct comparisons among groups. Pearson analyses were utilized to evaluate associations of PSMD2 with others in BCa. P < 0.05 represented meaningful statistics.

Through analyzing the transcriptional data of the GTEx and TCGA database, the expression values of PSMD2 were firstly investigated in multiple cancers involving over 10000 cancer and normal tissue samples. As shown in Figure 1A , PSMD2 was remarkably overexpressed in most cancer types, including bladder cancer ( Figure 1B ). Meanwhile, PSMD2 was also found to be upregulated in 19 paired BCa and normal bladder tissue samples in the TCGA database ( Figure 1C ). Elevated PSMD2 expression was correlated with advanced tumor grade ( Figure 1D ) and pathologic stage ( Figure 1E ). Moreover, compared with papillary BCa, PSMD2 was significantly upregulated in non-papillary BCa ( Figure 1F ). To verify the expression level of PSMD2 in vitro, RT-qPCR was applied to detect relative PSMD2 transcriptional expression in BCa cells and normal bladder cell line, and the results revealed that PSMD2 is significantly upregulated in BCa cells ( Figure 1G ). These findings demonstrated that PSMD2 is overexpressed in BCa.

The prognostic values of PSMD2 in BCa were subsequently investigated in multiple datasets. Firstly, the TCGA-BLCA cohort (P=0.032) and GSE13507 dataset (P=0.005) both showed that BCa patients with higher expression level of PSMD2 showed more worse 5-year survial compared with patients with lower PSMD2 expression ( Figures 2A, B ), these were consistent with the Kaplan–Meier plotter database (P=0.0021; Figure 2C ). Additionally, the COX analysis was also adopted to investigate the prognosis of PSMD2 expression and other clinical parameters in BCa. As plotted in Figure 2D , the univariate analysis showed that the TNM stage (p < 0.0001), PSMD2 expression (P=0.01102) and age (p=3e-05) are related to the OS of BCa patients ( Figure 2D ), and multivariate analyses also demonstrated that the TNM stage, age and PSMD2 expression (p<0.05) are independent factors for predicting the OS of BCa patients ( Figure 2E ). Furthermore, a nomogram model involving PSMD2 expression and other clinical factors was also been constructed ( Figure 2F ), according to the expression level of PSMD2 in individual patient, and the clinical factors (such as patient’s age, gender, pathologic stage and histologic grade), doctor can calculate the total points of every clinical features, which would assist in evaluating the 1-year, 3-year and 5-year survival probability of BCa patients.

To explore possible roles and underlying functions of PSMD2 in BCa, differential analyses between PSMD2 high and low BCa samples was firstly performed, as plotted in Figure 3A , when setting the |log2(FC)|>1 and p. adj<0.05 as the threshold, 1096 genes were noted to be upregulated, and 3968 genes were downregulated. The Gene Ontology (GO) analysis revealed that the highly expressed PSMD2 related pathways are Negative regulation of interferon-gamma production, Cytokine-cytokine receptor interaction, Negative regulation of T cell proliferation, Cell-cell adhesion mediated by cadherin and et al. ( Figures 3B ). Moreover, GSEA analysis showed that the Cell cycle (NES=2.969), Toll-like receptor (NES=2.153), Antigen processing and presentation (NES=2.785), JAK-STAT (NES=2.008), P53 (NES=1.9) and MAPK signaling pathway (NES=1.6) ( Figures 3C–H ) are significantly centralized in overexpressed PSMD2 BCa samples. These findings revealed that the cell cycle and regulation of tumor immune infiltration pathways are tightly linked with the abnormally elevated PSMD2 in BCa.

Considering the elevated expression of PSMD2 may contribute to carcinogenic effect, in vitro assays were subsequently performed to verify the effects of PSMD2 in BCa. Firstly, BCa cells UC3 and T24 were transfected with PSMD2 small interference RNA, as plotted in Figures 4A, B , the expression level of PSMD2 was remarkably decreased compared to negative control T24 and UC3 cells. Then the si-NC and si-PSMD2 treated T24, UC3 cells were adopted to perform functional assays, including wound healing and colony formation assay. The results demonstrated that the silencing of PSMD2 remarkably weaken the colony formation efficiency ( Figure 4C ) and migration abilities ( Figures 4D, E ) of BCa cells. These findings revealed that the upregulated PSMD2 may be participate the progression of BCa.

Given the GO and GSEA analysis results of overexpressed PSMD2 in BCa, the functions of PSMD2 in TIME were further explored. Firstly, the top 6 positively co-expressed genes of PSMD2 were identified by using Pearson correlation analysis, the results showed that the EIF4G1 (r=0.880, p<0.001), tubulin alpha 1c (TUBA1C) (r=0.810, p<0.001), leucine rich repeat containing 42 (LRRC42) (r=0.790, p<0.001), histone acetyltransferase 1 (HAT1) (r=0.780, p<0.001), YKT6 (r=0.780, p<0.001), DNA cross-link repair 1B (DCLRE1B) (r=0.780, p<0.001) are the top 6 co-expressed genes of PSMD2 in TCGA-BLCA cohort ( Figures 5A–F ). The correlations of these 6 genes with PSMD2 in BCa were verified in TIMER database, which also indicated strong associations between PSMD2 and these 6 genes (p<0.05; Figures 5G–L ). Subsequently, the ssGSEA algorithm was utilized to assess the associations of PSMD2 with its top 6 coexpressed genes and TIME of BCa. The results demonstrated that there are tight links between PSMD2 and 6 coexpressed genes and TIME in BCa, especially the Th2 cells (P<0.01, Figures 6A–G ). These results implied PSMD2 might be involved in the regulation of TIME in BCa.

Subsequently, the BCa samples were separated into two PSMD2 groups based on its median transcriptional level, namely PSMD2 low and PSMD2 high group, and the xCell algorithm was applied to assess the TIME component of this two PSMD2 group BCa samples. As plotted in the Figure 7A , entirely different TIME was noted between the two PSMD2 groups. Specifically, the Macrophage (P<0.001, Figure 7B ), Myeloid Dendritic cell (P<0.05, Figure 7C ), and Th2 cell (P<0.001, Figure 7D ) were significantly enriched in PSMD2 high group, while the CD8+ T cell (P<0.001, Figure 7E ), NK T cell (P<0.05, Figure 7F ), and CD4+ Central Memory T cell (P<0.001, Figure 7G ) were remarkably decreased in PSMD2 high group. Moreover, immune escape markers such as PD-L1, TIGIT, CTLA4, HAVCR2, LAG3, and PD-L2 were remarkably elevated in PSMD2 group ( Figure 7H ). Additionally, the TIDE score was also calculated between two PSMD2 groups, and the results revealed in comparison with PSMD2 low group, the TIDE score is remarkably higher in PSMD2 high group ( Figure 7I ). These results implied that overexpressed PSMD2 might be involved in shaping TIME and related to immune escape in BCa.

Given that the vital roles of PSMD2 in BCa, and the expression levels of PSMD2 were widely elevated in the vast majority of cancers ( Figure 1A ), the roles of PSMD2 were subsequently explored in pan-cancer. The prognostic values of PSMD2 were firstly investigated in more than 30 types of cancer, and the results revealed that the highly expressed PSMD2 predicts unfavorable OS in twelve types of cancer ( Figure 8A ), including the BLCA (Bladder Urothelial Carcinoma), SKCM (Skin Cutaneous Melanoma), HNSC (Head and Neck squamous cell carcinoma), BRCA (Breast invasive carcinoma), LIHC (Liver hepatocellular carcinoma), KICH (Kidney Chromophobe), LAML (Acute Myeloid Leukemia), LGG (Brain Lower Grade Glioma), LUAD (Lung adenocarcinoma), MESO (Mesothelioma), PAAD (Pancreatic adenocarcinoma) and ACC (Adrenocortical carcinoma) (HR>1, P<0.05), while shows no prognostic values in other types of cancer (P>0.05) ( Figure 8A ). Furthermore, the roles of PSMD2 in TIME were also investigated in pan-cancer, as shown in Figure 8B , the PSMD2 expression was tightly linked to infiltrates of Th2 cells and is negatively related to CD8+ T cells and cytotoxic cells in the vast majority of cancers ( Figure 8B ). These findings implied that the overexpressed PSMD2 might be implicated in the immune escape in cancer.