The gene expression profiles of PC patients were downloaded from The Cancer Genome Atlas (TCGA) database¹, as well as the corresponding clinical and DNA methylation information. This included 499 PC samples and 52 adjacent normal tissues which were retrospectively studied. For validation, additional PRAD cohorts of 293 and 248 patients were obtained with the accession number GSE70770 and GSE116918 from the GEO database². Gene expression data and clinical data were downloaded.

Firstly, patients that lack information of T stage, N stage, PSA, Gleason score, residual tumor, SNHG10 expression, and survival status were excluded from following analysis. Then, 288 prostate cancer samples were screened from the TCGA data. These samples were randomly classified into the training cohort (n = 144) and the test cohort (n = 144). Subsequently, we developed the nomogram in the training cohort using cox regression analysis, and validated in the test cohort. R package “rms” was used to establish the nomogram for predict the probability of 3-year and 5-year survival. The ROC curve for nomogram was performed via “timeROC” R package. The calibration curve was conducted via “calibrate” function of “rms” R package. We performed the DCA analysis of survival outcome by “ggDCA” R package.

The Gene set enrichment analysis (GSEA) was downloaded from the GSEA website³. A customized Perl script⁴ was used to perform GSEA between high-SNHG10 and low-SNHG10 groups. According to the default statistical methods, an adjusted P-value < 0.05 was considered significant.

Single sample GSEA (ssGSEA) using the R package “GSVA” was applied to perform the immune infiltration analysis of PC. The infiltration levels of 24 immune cell types were determined based on gene expression profiles in the literature (Bindea et al., 2013; Hänzelmann et al., 2013). The correlation between SNHG10 and the infiltration levels of 24 immune cells was evaluated by Spearman’s test and the Wilcoxon rank sum test. In addition, TIMER website⁵ was used to further investigate the association between immune infiltration and the expression of SNHG10, which is a comprehensive resource for systematical analysis of immune infiltrates across diverse cancer types.

Human PC cell lines (VCaP, LNCaP, 22RV1, PC3, and DU145) and the human normal human prostate epithelial cell line (RWPE-1) were purchased from the American Type Culture Collection (ATCC⁶, United States). LNCaP, 22RV1, and DU145 cells were cultured in RPMI-1640 (Gibco, United States). PC3 cells were cultured in F12K medium (Gibco, Australia). VCaP and RWPE-1 cells were cultured in DMEM (Gibco, United States). All cells were cultured with 10% fetal bovine serum (Gibco, Australia) at 37°C in a 5% CO₂ atmosphere. SNHG10 siRNAs were synthesized by Hanbio (Shanghai, China). The sequences of the siRNAs were as follows: si-SNHG10: 5′-CAACCGCUUUGUUAGUUAATT-3′; si-NC: 5′-UUCUCCGAACGUGUCACGUTT-3′. For the silencing of SNHG10 in the DU145 and 22RV1 cells, the cells were transfected with 50 nM small interfering RNA targeting si-SNHG10 or negative control si-NC according to the manufacturer’s instructions of Lipofectamine 2000 (Invitrogen). Scrambled siRNAs were used as negative control.

Total RNA was isolated and reversely transcribed into cDNA using Trizol reagent (Thermo Fisher Scientific, Inc.) and TaKaRa Prime Script RT Reagent Kit (TaKaRa, China) based on the manufacturer’s protocols. RT-qPCR was performed using SYBR Premix Ex Taq (TaKaRa, China) according to standard methodology. The relative expression levels were calculated using the 2^–ΔΔCt method relative to GAPDH. The PCR primers used were as follows: GAPDH: 5′-AAAAGCATCACCCGGAGGAGAA-3′ (forward) and 5′-AAGGAAATGAATGGGCAGCCG-3′ (reverse); SNHG10: 5′-GTTGGTCTCTTGGGAGGTAG-3′ (forward) and 5′-CGCC ACGACGAACTGCATGC-3′ (reverse). All experiments were performed in triplicate.

Cell Counting Kit 8 (CCK-8) and colony formation assays were used to measure cell proliferation. For CCK-8 assays, PC cells were seeded into a 96-well plate (6000 cells per well). CCK-8 reagent (Hanbio, China) was added into wells at 24, 48, and 72 h. After the cells were incubated at 37°C for 2 h, the optical density (OD) was measured at 450 nm using a microplate reader. For colony formation, 2000 cells per well were seeded in 6-well plates in 3 mL medium per well. After approximately 2 weeks, cell colonies were fixed with 4% paraformaldehyde for 15 min. Then, the colonies were stained with 0.2% crystal violet (Solarbio, China) and counted using ImageJ software. Each experiment was repeated three times.

Cell migration assays were performed using 8-μm pore size chambers coated without Matrigel gel. A total of 5 x 10⁴ transfected cells were seeded into the upper chamber with serum-free medium and 20% FBS medium was added into the lower well. After incubation for 1 day, the cells were stained and observed under an optical microscope. Three random fields were analyzed for each sample. Cell invasion assays were performed using chambers with Matrigel gel. Other procedures were the same as above. All the assays were conducted three times independently.

For the datasets from TCGA database, statistical analyses were performed using R (v.3.6.3). The Wilcoxon rank sum test, Chi-square test, and Fisher exact test were used to estimate the association between SNHG10 and clinical pathologic characteristics. The Kaplan-Meier method was used to calculated PC patient survival rates. Univariate and multivariate cox analysis were performed to assess the relationship between clinical features and progression-free survival (PFS). For the data regarding the function of SNHG10, GraphPad Prism 7.01 was used for statistical analyses. The Student’s t-test evaluated the statistical significance between groups. The data was shown as mean ± SD from at least three independent experiments and P < 0.05 was considered statistically significant.

The workflow of this study is presented in Figure 1. To compare the expression of SNHG10 in PC and normal samples, we analyzed the expression of SNHG10 in 499 tumor tissues and 52 normal prostate tissues of TCGA data, and found that SNHG10 was over-expressed in PC tissues (P < 0.001) (Figure 2A). There were 52 pairs of cancer samples and matched adjacent normal samples in TCGA data. The expression of SNHG10 was also higher in cancer samples than matched adjacent normal samples (P < 0.01) (Figure 2B). In addition, we compared the expression of SNHG10 between normal and prostate cancer in GSE70770, and found that SNHG10 was over-expressed in cancer tissues (P < 0.001) (Figure 2C). Meanwhile, SNHG10 expression was significantly higher in PC cell lines (DU145, 22RV1, PC3, VCaP, and LNCaP) than the prostate cell line (RWPE-1) (Figure 2D). Additionally, Kaplan-Meier survival analysis was used to investigate the relationship of SNHG10 expression and overall survival (OS) or progression-free survival (PFS) in the PC patients of TCGA data. SNHG10 expression has great significance in predicting OS (P < 0.05) (Figure 2E). As shown in Figure 2F, SNHG10 expression was significantly associated with poor PFS of PC patients (P = 0.003). Additionally, the concordance-index (C-index) was calculated to evaluate SNGH10 expression in predicting the OS and PFS. The C-index for SNGH10 in predicting the OS was 0.619. The performance of SNHG10 for predicting the PFS showed a better prognostic power with a C-index of 0.727. Then, patients with high SNHG10 expression had a worse PFS compared with low expression group in GSE70770 (P < 0.001) (Figure 2G).

Overall, 499 PC patients with clinical parameters were classified into two subgroups according to the mean value of relative SNHG10 expression. We then analyzed the relationships between SNHG10 and clinical parameters, including T stage, N stage, M stage, Gleason score, primary therapy outcome, residual tumor, zone of origin, prostate specific antigen (PSA), age, and race. SNHG10 expression was significantly associated with T stage (P = 0.013), N stage (P = 0.005), Gleason score (P < 0.001), primary therapy outcome (P < 0.001), and race (P = 0.008), while patients of different ages, M stage, residual tumor, zone of origin, and PSA shown no significant difference (Table 1). To visualize and clarify these relationships, we divided patients into subgroups (T2 vs. T3+T4, N0 vs. N1, Gleason score 6+7+8 vs. Gleason score 9+10). As shown in Figures 3A–C, advanced T stage, lymph node, and Gleason score patients exhibited higher SNHG10 expression levels. Moreover, in the GSE116918, SNHG10 expression was significantly correlated with advanced T stage and Gleason score (Figures 3D,E). Then, ROC analysis showed that the SNHG10 could be used to differentiate PC patients from normal control with a cut-off of 2.31 which resulted in 59.9% for sensitivity and 90.2% for specificity (AUC = 0.805, C-index = 0.806, Youden index = 0.501) (Figure 3F). Besides, calibration curve of SNGH10 were coincident with the reference line, which indicated a high degree of credibility (Figure 3G).

TCGA database was used to identify the aberrant expression of SNHG10 across multiple cancers. As shown in the Figure 4A, SNHG10 expression was higher in multiple cancer types compared to the normal samples. In addition, SNHG10 expression and patient survival was further investigated in various cancer types to expand our analysis to a pan-cancer level. There are significant associations between SNHG10 expression and OS or PFS in adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), kidney renal clear cell carcinoma (KIRC), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD), and thymoma (THYM) (Figures 4B–I).

As shown in the Table 2, patients having complete clinical data were included in further Cox regression analysis. In the Cox univariate regression analysis, high expression of SNHG10, T stage, N stage, Gleason score, primary therapy outcome, residual tumor, and PSA were associated with PFS in PC patients. Multivariate Cox analysis further indicated that SNHG10 (P < 0.05) was an independent prognostic factor for PFS in PC patients, along with Gleason score and primary therapy outcome.

To allow clinical application of our findings, we construct the nomogram by TCGA data. The nomogram for predicting 3-year and 5-year PFS of PC was showed in Figure 5A. ROC curve, calibration, and discrimination were employed to evaluate the performance of the model. The C-index of the nomogram were 0.848 for the training cohort and 0.825 for the test cohort. The discriminative ability of the nomogram was measured using the 3-year and 5-year survival AUC values from time-dependent ROC curve. In the training cohort, the nomogram AUC values for 3-year and 5-year PFS were 0.84 and 0.883, respectively (Figure 5B). In addition, in the test cohort, the nomogram AUC values for 3-year and 5-year PFS were 0.847 and 0.878, respectively (Figure 5C). Moreover, the calibration plots in the training cohorts and test cohorts demonstrated that the nomogram-based predictive results were mostly consistent with the actual prognosis results (Figures 5D,E). DCA plots showed that our nomogram had great net benefits for predicting 3-year and 5-year PFS of patients both in the training cohorts and test cohorts (Figures 5F,G), demonstrating its application in guiding clinical decision for PC patients.

Using the ssGSEA method, we analyzed the relationship between SNHG10 expression and immune infiltration. These results suggested that SNHG10 expression was negatively associated with infiltration levels of neutrophils and T gamma delta (γδT) (P < 0.001) (Figure 6), and was positively correlated with that of natural killer (NK) CD56bright cells and plasmacytoid dendritic cells (pDC) (P < 0.001) (Supplementary Figure 1). In addition, the tool of TIMER was also used to identify the relationship between SNHG10 expression and tumor-infiltrated immune cells. There was a significant correlation between SNHG10 expression and different type of immune cells, including CD4+ T cells (P = 6.11e–5, cor = 0.195), CD8+ T cells (P = 3e−5, cor = −0.203), T cells regulatory (P = 2.78e−6, cor = 0.227), Neutrophils (P = 3.05e–2, cor = 0.106), and NK cells activated (P = 1.04e–2, cor = 0.125) (Figure 7A). Besides, different mutational forms of SNHG10 were associated with immune infiltrates of four leukocytes, which also revealed its influence on immune microenvironment (Figure 7B).

Gene set enrichment analysis was used to identify signaling pathways that were differentially activated between low and high SNHG10 expression groups. As shown in Figures 8A–F, there were six significant Kyoto Encyclopedia of Genes and Genomes pathways related with the high SNHG10 expression phenotype: oxidative phosphorylation, base excision repair, tyrosine metabolism, pyrimidine metabolism, RNA polymerase, and ribosome.

The above studies indicated that SNHG10 expression was distinctly up-regulated in PC tissues, and SNHG10 might influence the progression in PC. To further investigate the biological role of SNHG10 in PC, specific siRNA was used to construct DU145 and 22RV1 cells with stable knockdown of SNHG10 expression (Figure 9A). Our results demonstrated that, in DU145 and 22RV1 cells, cell growth was dramatically inhibited in SNHG10 knockdown cells compared with the control groups (Figure 9B). Next, Transwell assays showed that SNHG10 knockdown suppressed the migratory and invasive abilities of DU145 and 22RV1 cells (Figures 9C,D). SNHG10 inhibition also decreased tumor cell colony formation (Figure 9E). These results demonstrated that SNHG10 was involved in PC progression by inhibiting cell growth, migration, and invasion.