3.1 Distinct immune microenvironment and HLA gene expression profiles in high and low PPAR groups in prostate cancer

The analysis of prostate cancer data from the TCGA database using ssGSEA stratified patients into high and low PPAR groups based on metabolic-related gene sets. Significant differences were observed in the immune microenvironment and HLA gene expression profiles between the two groups. Figure 1A shows a heatmap of various metabolic pathways, highlighting distinct clustering patterns and differential pathway activation. Specifically, fatty acid metabolism and PPAR signaling were notably different. The violin plots in Fig. 1B indicate that the high PPAR group had significantly lower ImmuneScore and ESTIMATEScore, suggesting a less immune-active tumor microenvironment. HLA-related gene expression, as depicted in Fig. 1C, revealed significant differences in several genes, including HLA-B, HLA-C, HLA-F, and HLA-G, implying alterations in antigen presentation and immune recognition in the high PPAR group. Additionally, Fig. 1D shows that certain immune cell types, such as T cells CD4 memory resting and T regulatory cells, had significantly different fractions between the two groups, with the high PPAR group exhibiting a higher fraction of immunosuppressive Tregs.

Fig. 1

Distinct immune microenvironment and HLA gene expression profiles in high and low PPAR groups in prostate cancer. A Heatmap showing the clustering patterns of various metabolic pathways in high and low PPAR groups. The color scale represents the enrichment scores of the pathways, with red indicating high enrichment and blue indicating low enrichment. B Violin plots comparing the ImmuneScore, StromalScore, and ESTIMATEScore between high and low PPAR groups. C Boxplots showing the expression levels of various HLA-related genes between high and low PPAR groups. D Boxplots illustrating the fractions of different immune cell types between high and low PPAR groups, highlighting significant differences in T cells CD4 memory resting and T regulatory cells

3.2 Prognostic model construction and survival analysis using univariate Cox regression, LASSO, and multivariate Cox regression

The prognostic model for prostate cancer was developed using univariate Cox regression analysis, LASSO regression, and multivariate Cox regression analysis. As shown in Fig. 2A, univariate Cox regression identified several significant predictors of survival, including ASPM, CENPF, and TOP2A, with their hazard ratios indicating varying levels of risk. The LASSO regression results in Fig. 2B show the partial likelihood deviance across different log(λ) values, with the optimal λ minimizing the deviance. The coefficients for the selected genes retained in the final LASSO model are depicted in Fig. 2C. Subsequently, multivariate Cox regression analysis refined the prognostic model by incorporating multiple significant genes, resulting in coefficients for key genes: BUD23 (0.0047), IQGAP3 (0.0531), SSPOP (0.0567), KRTAP5-1 (0.0710), NAALADL2-AS2 (0.0049), and AC069228.1 (0.0378). Survival analysis, as illustrated in Fig. 2D, shows Kaplan–Meier survival curves for high-risk and low-risk groups, indicating a significantly worse outcome for the high-risk group (p < 0.001). The ROC curves in Fig. 2E further validate the model’s predictive performance, with AUC values of 0.771 at 1 year, 0.738 at 3 years, and 0.701 at 5 years. These results highlight the model’s robust ability to stratify patients by risk and predict survival outcomes effectively.

Fig. 2

Prognostic model construction and survival analysis using univariate Cox regression, LASSO, and multivariate Cox regression. A Forest plot showing the hazard ratios and p-values from univariate Cox regression analysis for various genes, identifying significant predictors of survival. B LASSO regression plot displaying the partial likelihood deviance as a function of log(λ), with the optimal λ value indicated by the vertical dotted line. C Coefficients for the selected genes retained in the final LASSO model. D Kaplan–Meier survival curves for high-risk and low-risk groups, demonstrating a significantly worse outcome for the high-risk group. E ROC curves for the prognostic model at 1, 3, and 5 years, with corresponding AUC values indicating good predictive performance

3.3 Immune infiltration analysis

The immune infiltration analysis revealed significant correlations between risk scores and various immune cell types in prostate cancer. As shown in Fig. 3A, the bubble plot summarizes the correlation coefficients of immune cell fractions with risk scores across different software tools, including XCELL, QUANTISEQ, MCPcounter, EPIC, and CIBERSORT. The plot indicates diverse immune cell types significantly associated with risk scores. Figures 3B to G provide scatter plots of specific immune cell types showing significant correlations with risk scores. Figure 3B depicts a positive correlation between B cell memory (CIBERSORT) and risk scores (R = 0.15, p = 0.00077). Similarly, Fig. 3C shows a significant correlation between cancer-associated fibroblasts (EPIC) and risk scores (R = 0.18, p = 9.2e−05). In Fig. 3D, macrophage M1 (CIBERSORT) also demonstrates a positive correlation with risk scores (R = 0.22, p = 1.3e−08), suggesting an enhanced presence of pro-inflammatory macrophages in high-risk patients. Figure 3E shows the correlation between macrophage M2 (XCELL) and risk scores (R = 0.13, p = 0.0033), highlighting the association with anti-inflammatory macrophages. Additionally, Figs. 3F and G show correlations of risk scores with NK cell activated (CIBERSORT) (R = 0.092, p = 0.041) and T cell CD4 + naive (XCELL) (R = 0.13, p = 0.0053), respectively. These findings indicate a complex interaction between the immune microenvironment and risk scores, with significant involvement of various immune cell types.

Fig. 3

Immune infiltration analysis. A Bubble plot summarizing the correlation coefficients of immune cell fractions with risk scores across different software tools, including XCELL, QUANTISEQ, MCPcounter, EPIC, and CIBERSORT. B–G Scatter plots showing significant correlations between risk scores and specific immune cell types: B B cell memory (CIBERSORT), C cancer-associated fibroblasts (EPIC), D macrophage M1 (CIBERSORT), E macrophage M2 (XCELL), F NK cell activated (CIBERSORT), and G T cell CD4 + naive (XCELL)

3.4 Expression and clinical correlation of BUD23 in prostate cancer

The analysis of BUD23 expression in prostate cancer revealed significant associations with clinical parameters. As shown in Fig. 4A, the distribution of pathological T stages varied between low and high-risk groups, with higher stages (T3b and T4) being more prevalent in the high-risk group. Figure 4B illustrates the distribution of nodal metastasis (N stages) between the groups, indicating a higher proportion of N1 stage in the high-risk group compared to the low-risk group. Figure 4C presents the expression levels of BUD23 in normal and tumor tissues, showing significantly higher expression in tumor tissues (p < 0.001). This elevated expression is further correlated with advanced pathological T stages (Fig. 4D), where T3 and T4 stages exhibit higher BUD23 expression compared to T2 stages (p < 0.01). Additionally, Fig. 4E shows that BUD23 expression is significantly higher in N1 stage compared to N0 stage (p < 0.05). Lastly, Fig. 4F demonstrates that higher Gleason scores (8–10) are associated with significantly increased BUD23 expression compared to lower Gleason scores (6–7) (p < 0.001). These results suggest that elevated BUD23 expression is associated with more aggressive prostate cancer features, including advanced pathological stages, nodal metastasis, and higher Gleason scores, indicating its potential role as a prognostic biomarker.

Fig. 4

Expression and clinical correlation of BUD23 in prostate cancer. A Stacked bar plots showing the distribution of pathological T stages in low and high-risk groups. B Stacked bar plots illustrating the distribution of nodal metastasis (N stages) between low and high-risk groups. C Violin plot comparing the expression levels of BUD23 between normal and tumor tissues. D–F Violin plots showing the correlation of BUD23 expression with (D) pathological T stages, (E) pathological N stages, and (F) Gleason scores, indicating higher BUD23 expression in more aggressive prostate cancer features

3.5 Functional validation of BUD23 in prostate cancer cells

The functional role of BUD23 in prostate cancer cells was investigated using Western blotting (WB), quantitative PCR (qPCR), and CCK-8 assays. Figure 5A and B presents the WB results, showing that BUD23 expression was significantly knocked down in both PC-3 and LNCaP cells after transfection with sgBUD23-1 and sgBUD23-2, compared to the control sgROSA group. The qPCR analysis, shown in Fig. 5C and D, further confirmed the efficient knockdown of BUD23 expression at the mRNA level in PC-3 and LNCaP cells. Both sgBUD23-1 and sgBUD23-2 significantly reduced BUD23 expression compared to the control. The CCK-8 assay results, illustrated in Fig. 5E and F, demonstrate the impact of BUD23 knockdown on cell proliferation. In PC-3 cells (Fig. 5E), BUD23 knockdown resulted in a marked reduction in cell proliferation over a 6-day period. Similarly, in LNCaP cells (Fig. 5F), BUD23 knockdown significantly inhibited cell proliferation compared to the control group. The EdU proliferation assay was conducted to further assess the impact of BUD23 knockdown on cell proliferation in prostate cancer cells. The results are shown in Fig. 6, with images of PC-3 and LNCaP cells stained for EdU incorporation (green) and DAPI (blue) to label proliferating cells and nuclei, respectively. In the PC-3 cell line, EdU staining reveals a substantial decrease in proliferating cells in the sgBUD23-1 group compared to the control sgROSA group. This is evident from the reduced number of green-stained cells in the sgBUD23-1 group, indicating a significant inhibition of DNA synthesis and cell proliferation upon BUD23 knockdown. Similarly, in the LNCaP cell line, the sgBUD23-1 group shows a marked reduction in EdU-positive cells compared to the sgROSA control group. This reduction in green fluorescence further confirms that BUD23 knockdown significantly impairs the proliferative capacity of LNCaP cells (Fig. 6).

Fig. 5

Functional validation of BUD23 in prostate cancer cells. A, B Western blotting results showing BUD23 protein levels in A PC-3 and B LNCaP cells after transfection with sgBUD23-1 and sgBUD23-2, compared to the control sgROSA group. GAPDH is used as a loading control. C, D qPCR analysis confirming the knockdown efficiency of BUD23 at the mRNA level in C PC-3 and D LNCaP cells. E, F CCK-8 assay results showing the impact of BUD23 knockdown on cell proliferation in E PC-3 and F LNCaP cells over a 6-day period

Fig. 6

EdU proliferation assay. Images of PC-3 and LNCaP cells stained for EdU incorporation (green) and DAPI (blue) to label proliferating cells and nuclei, respectively. The images compare the control sgROSA group with the sgBUD23-1 knockdown group, showing a substantial decrease in EdU-positive cells in the BUD23 knockdown group, indicating reduced cell proliferation

3.6 Expression levels of PPAR-related proteins upon BUD23 knockdown

The expression levels of PPAR-related proteins were assessed in prostate cancer cells following BUD23 knockdown using Western blotting. Figure 7 displays the protein expression levels of PPAR-α, PPAR-β, and PPAR-γ, with GAPDH as a loading control. In the sgROSA control group, PPAR-α showed a higher expression compared to the sgBUD23-1 and sgBUD23-2 groups, indicating that BUD23 knockdown significantly reduces PPAR-α protein levels. Similarly, PPAR-β expression was also reduced in both sgBUD23-1 and sgBUD23-2 groups compared to the control. The most notable reduction was observed in PPAR-γ expression, where BUD23 knockdown via sgBUD23-1 and sgBUD23-2 resulted in a marked decrease in protein levels compared to the sgROSA control.

Fig. 7

Expression levels of PPAR-related proteins upon BUD23 knockdown. Western blotting results showing the protein expression levels of PPAR-α, PPAR-β, and PPAR-γ in prostate cancer cells following BUD23 knockdown using sgBUD23-1 and sgBUD23-2, compared to the control sgROSA group. GAPDH is used as a loading control. The results indicate significant reductions in the expression of PPAR-α, PPAR-β, and PPAR-γ in the BUD23 knockdown groups