RNA sequencing data for 33 cancer types were retrieved from the Genomic Data Commons (GDC) data portal website (https://portal.gdc.cancer.gov/) and analyzed using R v4.0.3. Screening distinct mRNAs was performed with the R package ‘DESeq2’ (12); the threshold value was determined to be log2 fold change>1.0 and an adjusted P-value of 0.05 unless otherwise specified. With mRNAseq data from the TIMER database, we compared PIK3C2A expression in cancer tissues versus adjacent tissues (13).

The scRNA-seq data were analyzed using R (version 4.0.3) with the Seurat package (version 3.1.1). Filtering was performed on sample KIRC cells with < 200 and > 2500 genes and a mitochondrial gene percentage of >10%. The top 2000 most variable genes in the scRNA-seq data were used for principal components analysis (PCA), and the first 20 principal components were used for UMAP analysis. The FindClusters function divided all cells into 14 clusters, with resolution = 0.5.

Univariate Cox regression analysis was used to assess the relationship between PIK3C2A and the prognosis of malignancies. Based on the median expression value of PIK3C3A, KIRC survival curves were generated using the ‘survival’ R package. PIK3C2A was evaluated via multivariate Cox regression to determine its independence as a predictor of patient survival. Nomogram-based models for predicting patient survival were developed using the R package ‘rms’. The area under the time-dependent ROC curve (AUC) was generated using the ‘timeROC’ (version: 0.4) package to examine the sensitivity and specificity of PIK3C2A in survival prediction.

The human KIRC cell lines 769P, CAKI-1, 786-O, and ACHN cells were obtained from the cancer institute of Zhejiang Caner Hospital. KIRC cell lines were grown in RPMI-1640 medium (BasalMedia, China) containing 10% fetal bovine serum (FBS), 100 units of penicillin, and 50 units of streptomycin per ml.

PIK3C2A siRNA was purchased from Zorinbio Company (China); the PIK3C2A siRNA1 sequence was 5’-CCACUUAUGCUUUACCUUCUATT-3’, the PIK3C2A siRNA2 sequence was 5’-GCUAGUGUGAAGGUCUCCAUUTT-3’, and the PIK3C2A siRNA3 sequence was 5’-GAUGACAGUUUCGAGACUAAATT-3’. Lipofectamine 2000 was used to transfect KIRC cells with siRNA. Control cells were transfected with non-targeting siRNA.

Real-time PCR assay

The TRIzol Reagent (Sigma, St. Louis, MO) was used to isolate total RNA, and the PrimeScript RT Reagent Kit was used to reverse-transcribe this RNA into cDNA, per the manufacturer’s instructions (TaKaRa, Dalian, China). The mRNA expressions of PIK3C2A and GAPDH were measured using an RT-PCR quantitative PCR kit (TaKaRa, Dalian, China).

Cells were lysed in ice-cold RIPA buffer containing protease and phosphatase inhibitors. Proteins from each group were separated using SDS-PAGE and transferred in equal amounts to polyvinylidene fluoride (PVDF) membranes. The membranes were blocked with 5% fat-free milk for one hour and then treated overnight at 4°C with PIK3C2A Mouse mAb (ZENBIO, China). The membranes were then washed thrice in Tris-buffered saline with Tween-20 (TBST) and incubated at room temperature for 1 hour with a secondary antibody conjugated to HRP. Blots were examined with an ECL detection kit (NCM Biotech, China) and processed with Quantity One 1-D Analysis Software (Bio-Rad, San Francisco, CA, USA).

Colony formation tests were performed as described previously (14). Briefly, 3000 cells/well were seeded in 96-well plates, and cell proliferation was evaluated using a Cell Counting Kit-8 (CCK-8; NCM Biotech, China), measuring absorbance at 450 nm based on the instructions.

For cell proliferation assessment, a 5-ethynyl-20-deoxyuridine (EdU) test kit was utilized (Beyotime Biotechnology, China). KIRC cells (2 × 10⁴ cells/well) were seeded in the glass bottom dish and cultured overnight. Afterward, the cells were incubated with 10 μM EdU buffer at 37°C for 2 hours, fixed with 4% formaldehyde for 15 minutes, permeabilized with 0.3% Triton X-100 for 25 minutes, and nuclear-stained with Hoechst. The cells were then examined using a fluorescence microscope. The ratio of EdU-positive cells (red cells) to all Hoechst-positive cells (blue cells) was defined as the EdU incorporation rate.

The TIMER database was utilized to evaluate the degree of relatedness between PIK3C2A and infiltrating immune cells, based on how certain markers show up in different tumors. Heat plots were used to show the correlation between PIK3C2A and each immune gene marker; PIK3C2A was plotted on the x-axis, immune markers on the y-axis, and the gene level was adjusted with log2 TPM. The correlation coefficient between PIK3C2A and immune markers was calculated using the Spearman method in GEPIA.

CIBERSORT is a deconvolution algorithm that can be used to describe the cell composition of complex tissues. Fluorescence-activated cell sorting (FACS) has confirmed that this protocol quantifies a terrific amount of specific cell types (15). Thereupon, CIBERSORT was applied to figure out the relative proportions of 22 infiltrating immune cell types in the PIK3C2A high-expression and low-expression groups, using the LM22 signature matrix and 1000 permutations. Immunedeconv, an R package that contains the CIBERSORT algorithm, was used to carry out this procedure (16).

The association of PIK3C2A with protein-coding genes (PCGs) was evaluated using Spearman correlation coefficients, and genes with a Spearman correlation coefficient greater than 0.50 were identified as PIK3C2A-associated PCGs. GO and KEGG pathway enrichment studies on the PIK3C2A-associated PCGs were then performed using the clusterProfiler program (17). P-values <0.05 were set as the threshold for GO/KEGG pathway enrichment analyses. The enrichment outcomes were visualized in R using the ggplot2 package.

The underlying signaling pathways associated with PIK3C2A in KIRC were identified by gene set enrichment analysis (FDR 0.25, adjusted p-value 0.05), based on the expression patterns between the high-expression and low-expression groups. This analysis was conducted with the clusterProfile application (17). The PIK3C2A-related pathway in the Molecular Signatures Database (MSigDB) was visualized using the R program ‘ggplot2’.

The TIMER database was employed to compare the levels of mRNA for PIK3C2A in various kinds of tumors and nearby tissues. Compared with normal controls, PIK3C2A expression was significantly lower in bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostate adenocarcinoma (PRAD), thyroid carcinoma (THCA), and (UCEC) ( Figure 1A ). However, PIK3C2A expression was significantly increased in cholangiocarcinoma (CHOL), liver hepatocellular carcinoma (LIHC), and stomach adenocarcinoma (STAD) ( Figure 1A ). In addition, we retrieved the RNA-seq and clinical data for 33 cancers from the GDC website. Univariate COX analysis revealed that PIK3C2A is a protective factor for overall survival (OS), disease-free survival (DFS), and progression-free survival (PFS) only in KIRC ( Figure 1B ). As a result, we selected KIRC to further investigate the role of PIK3C2A in tumor progression.

Based on the TCGA database, PIK3C2A mRNA expression level was investigated in KIRC. The outcomes exhibited significantly lower mRNA levels of PIK3C2A in KIRC tissues than in normal tissues ( Figure 2A ). These findings were validated in the GEO database; the results demonstrated lower PIK3C2A expression levels in KIRC tissues than in normal tissues ( Figures 2I, J ). According to the Human Protein Atlas database (18), the PIK3C2A protein level was significantly lower in KIRC tissues compared with adjacent non-tumor kidney tissues ( Figure 2K ). PIK3C2A protein expression level was further evaluated in eight pairs of KIRC and neighboring non-tumor kidney tissues by western blot. The results demonstrated lower PIK3C2A protein levels in KIRC tissues than in neighboring non-tumor kidney tissues ( Figure 2L ). Furthermore, we investigated the expression changes of PIK3C2A in KIRC samples with different clinicopathologic features; the results revealed no significant difference in PIK3C2A expression in KIRC samples with varying gender, age, and N stages ( Figure 2B–D ). However, PIK3C2A expression decreased with the progression of the T stage, M stage, histologic grade, and pathologic stage ( Figures 2E–H ).

PIK3C2A expression in diverse cell subpopulations of KIRC was evaluated using scRNA-seq data. The dataset included 3568 cells from the right KIRC tissue (KIRC1) and 3575 cells from the left KIRC tissue (KIRC2). The data were subjected to UMAP analysis following quality control. Cells in KIRC were classified into 14 distinct categories, including CD8+ cell, T helper cell, Treg cell, NKT cell, effector CD8+ memory T cell, tumor cell, macrophage, vascular endothelial cell, endothelial cell, neutrophil, mesangial cell, B cell, mast cell and collecting duct principal cell, according to data on previously reported markers ( Figure 3A ). PIK3C2A-expressing cells were displayed on the UMAP plot, revealing that PIK3C2A was primarily expressed in endothelial cells ( Figure 3B ). Furthermore, we assessed the expression levels of PIK3C2A in each subgroup and discovered that it is expressed in various immune cells ( Figure 3C ). PIK3C2A expression in tumor cells was dramatically downregulated in the right KIRC tissue (KIRC1) ( Figure 3C ).

The Kaplan-Meier curve confirmed that the OS rate of the high-expression subset was significantly higher than that of the low-expression subset (P-value=3.71e-05, Figure 4A ). Besides, the ROC curves were used to evaluate the predictive accuracy of PIK3C2A. PIK3C2A AUCs were 0.631, 0.632, and 0.644 at 1‐, 3‐ and 5‐year OS periods, respectively ( Figure 4B ). Similarly, the Kaplan Meier statistical analysis revealed that PIK3C2A may be a protective factor for the DFS rate (P-value=7.11e-06, Figure 4D ) and PFS rate (P-value=2.19e-04, Figure 4G ) of KIRC. The time-dependent ROC curves for DFS demonstrated that the 1-year, 3-year, and 5-year AUC values of PIK3C2A were 0.649%, 0.647%, and 0.678%, respectively ( Figure 4E ). The time-dependent ROC curves for PFS revealed a 1-year, 3-year, and 5-year AUC value of PIK3C2A of 0.601, 0.613, and 0.666, respectively ( Figure 4H ). These results demonstrated better predictive performance of PIK3C2A over time.

The relationship between PIK3C2A and the clinical properties of KIRC was investigated using the TCGA database. The results demonstrated a correlation of PIK3C2A with the T stage (p-value<0.001), M stage (p-value=0.004), pathologic stage (p-value<0.001), and histologic grade (p-value<0.001) ( Table 1 ). These findings confirmed that PIK3C2A expression decreases following the progression of the T stage, M stage, pathologic stage, and histologic grade of KIRC.

The independence of PIK3C2A from other clinicopathological characteristics of KIRC was determined using univariate and multivariate Cox regression analysis. Results from the univariate Cox regression analysis of KIRC patients revealed that PIK3C2A expression, T stage, N stage, M stage, histologic grade, and pathologic stage were all significant predictors of OS, DFS, and PFS ( Table 2 ). A multivariate Cox regression analysis revealed that PIK3C2A expression and M stage are independent prognostic variables for OS, DFS, and PFS in KIRC patients ( Table 2 ). In this view, PIK3C2A expression was suggested as a clinically promising biomarker for predicting patient outcomes in KIRC. Using the aforementioned variables, we developed nomograms to forecast the OS, DFS, and PFS of KIRC ( Figures 4C, F, I ). The C-index in the nomogram model for OS was 0.756, 0.846 for DFS, and 0.809 for PFS.

Having found that PIK3C2A expression decreased in KIRC tissues, we analyzed the influence of PIK3C2A on the cellular biology of KIRC cells. Firstly, we assessed the relationship between PIK3C2A and tumor-related signal pathways using ssGSEA. According to the results, PIK3C2A expression was negatively correlated with tumor proliferation signature and DNA replication pathways (Spearman’s correlation value = -0.30 and -0.15, respectively, P<0.001; Figures 5A, B ). Subsequently, PIK3C2A expression in KIRC cell lines was blocked using small interfering RNAs (siRNAs). Real-time PCR and Western blotting analyses revealed that siRNA2 and siRNA3 could significantly decrease PIK3C2A expression; as such, they were selected for further functional tests ( Figures 5C, D ). CCK8 detection results demonstrated that inhibiting PIK3C2A expression significantly promoted the proliferation of 769P, CAKI-1, 786-O, and ACHN cells ( Figure 5E ). Inhibiting the expression of PIK3C2A significantly increased the colony-forming ability of KIRC cells ( Figure 5F ). The EdU assay revealed a significantly higher proportion of EdU-positive cells in the PIK3C2A knockdown groups than in the control groups ( Figure 5G ). These results confirmed that PIK3C2A knockdown resulted in significantly increased KIRC cell proliferation in vitro.

We split the samples of the TCGA dataset into high-expression and low-expression groups based on the median value of PIK3C2A expression to determine gene alterations caused by PIK3C2A modifications. Following that, DEGs were identified using the DESeq2 Bioconductor package, with cut-offs set at a log2foldchange (log2FC) >1 and an adjusted P-value (adj. P) < 0.05. The results revealed that 1039 DEGs were up-regulated in the high-expression group, while 2815 PCGs were down-regulated in the low-expression group ( Figure 6A ). These DEGs were evaluated using the ClusterProfiler package (version: 3.18.0) in R to unravel the underlying functions and pathways. Figures 6B, C depict the distribution of these DEGs across molecular biological roles (BP), functions (MF), cellular components (CC), and KEGG pathways. The Gene Set Enrichment Analysis (GSEA) was used to verify the potential pathways activated or inhibited due to PIK3C2A alterations in KIRC. The Reactome innate immune system was the most significantly enriched signaling pathway (NSE = -3.368, adjusted P-value=0.005; Figure 6D ), suggesting that PIK3C2A alterations may contribute to KIRC-related immunological dysfunction.

According to the analysis in the TIMER database, PIK3C2A was positively correlated with the infiltration degrees of B cells (R = 0.182, P = 8.92E–05), CD8+ T cells (R = 0.243, P = 2.65E–07), CD4+ T cells (R = 0.349, P =1.28E−14), macrophages (R = 0.421, P = 1.04E–20), neutrophils (R = 0.399, P = 5.87E–19), and dendritic cells (R = 0.255, P = 3.28E−08) in KIRC ( Figure 7A ). In addition, using the CIBERSORT algorithm, we evaluated the correlation between PIK3C2A status and 22 immune fractions, uncovering a widespread immune infiltration pattern of KIRC. The results revealed a higher infiltration degree of B cell native (adjusted p-value= 6.49E-10), B cell plasma (adjusted p-value=4.88E-02), T cell CD4+ memory resting (adjusted p-value = 8.92E-04), NK cell resting (adjusted p-value= 1.29E-02), Monocyte (adjusted p-value= 3.91E-08), Macrophage M1 (adjusted p-value= 1.69E-02), Macrophage M2 (adjusted p-value=4.64E-04), Myeloid dendritic cell resting (adjusted p-value= 3.84E-02), Mast cell activated (adjusted p-value= 1.62E-05) and Neutrophil (adjusted p-value = 2.72E-05) in the PIK3C2A high-expression group than in the PIK3C2A low-expression group ( Figure 7B ). On the contrary, the infiltration degree of B cell memory (adjusted p-value= 3.05E-10), T cell CD8+ (adjusted p-value= 9.10E-06), T cell follicular helper (adjusted p-value= 1.33E-10), T cell regulatory (adjusted p-value= 2.56E-15), T cell gamma delta (adjusted p-value=3.59E-03), NK cell activated (adjusted p-value= 2.48E-08), Macrophage M0 (adjusted p-value= 1.16E-06) and Mast cell resting (adjusted p-value= 1.65E-02) was lower in the PIK3C2A high-expression group than in the PIK3C2A low-expression group ( Figure 7B ). These results highlight a crucial role for PIK3C2A in regulating the immune infiltration microenvironment of KIRC.

Based on data from the TIMER and TCGA databases, we analyzed the correlations between PIK3C2A and a wide range of immunological indicators in KIRC to shed light on the potential links between this gene and invading immune cells. The PIK3C2A level was found to be closely correlated with 37 out of 57 immune cell markers in KIRC, after adjusting for the purity in the TIMER database ( Supplementary Table 1 ). The TCGA database analysis revealed that NOS2 and PTGS2 of the M1 phenotype ( Figure 7C ), CD163, VSIG4, and MS4A4A of the M2 phenotype ( Figure 7C ), IL10 of TAMs ( Figure 7C ), CD86 and CSF1R of monocytes ( Figure 7C ), CEACAM8 and ITGAM of neutrophils ( Figure 7C ) were significantly correlated with PIK3C2A expression in KIRC. The GEPIA database was used to determine whether the association between the two immune cell types and PIK3C2A expression in KIRC and comparable normal tissues was distinct. The analysis yielded comparable findings to TCGA results; the correlations between PIK3C2A expression and macrophages or neutrophils were quite robust, with 12 out of 14 immune cell markers; However, 2 out of 14 immune cell indicators correlated significantly with PIK3C2A expression in renal tissues ( Table 3 ). These data suggested a potential regulatory role of PIK3C2A on macrophage polarization and neutrophil infiltration in KIRC, but this activity was extremely poor in normal tissues.

Furthermore, there was a considerable variation in the correlation of PIK3C2A expression with B cells, natural killer cells, dendritic cells (DCs), T regulatory cells (Tregs), T helper cells (Th cells), and T cell exhaustion in KIRC or normal tissues. The results demonstrated that the markers of B cells, including CD19 and CD79A ( Figure 7C ), are negatively correlated with PIK3C2A expression. For natural killer cells, we found a positive correlation of PIK3C2A with KIR2DL1 and KIR2DL3 in KIRC but not in normal tissues ( Figure 7C ). PIK3C2A was highly relevant to DC infiltration in KIRC ( Figure 7C ). Further research is warranted to establish whether PIK3C2A has dual effects on cancer immunity. PIK3C2A was closely correlated with T cell exhaustion and Treg; PIK3C2A was positively linked to CCR8, STAT5B, and TGFB1, and negatively related to FOXP3, PDCD1, and LAG3 ( Figure 7C ). Also, we found a strong correlation between PIK3C2A and biomarkers of T helper cells, including Th1, Th2, Tfh, and Th17 cells ( Figure 7C ). These add to the evidence that PIK3C2A is strongly linked to immune infiltration in KIRC and may play a dualistic role in tumor immune responses.

The correlation of PIK3C2A with PCGs was evaluated using Spearman correlation analysis to determine the precise mechanism of PIK3C2A in KIRC carcinogenesis. The results revealed that 2345 PCGs and 466 PCGs were positively and negatively correlated with PIK3C2A, respectively (Spearman’s correlation coefficient>0.5, adjusted p-value<0.05); the top 50 positively correlated and negatively correlated PCGs are shown in Figures 8A, B , respectively. Functional enrichment analysis demonstrated that PIK3C2A correlated PCGs were primarily enriched in 245 gene ontology (GO) terms (adjust p-value<0.001; Supplementary Table 2 ) and 22 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (adjust p-value<0.001; Supplementary Table 3 ); the typical GO terms and KEGG pathways are displayed in Figures 8C, D , respectively.