Data acquisition and gene expression analysis in KIRC

To investigate mRNA expression profiles of ST3Gals in KIRC tumor and non-tumor tissues, we analyzed several datasets from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases (Supplementary Tables 1 and Supplementary Table 2) with R package “DESeq2” or “limma” after normalized process, including TCGA-KIRC, GSE53757 [22], GSE68417 [23], GSE40435 [24], GSE46699 [25], GSE36895 [26], GSE71963 [27] and GSE76351. We assessed the levels of ST3Gals across various clinicopathological grades and stages of patients using TCGA-KIRC dataset.

Genomic mutation analyses

To examine alteration frequency of ST3Gals of KIRC patients, we accessed the TCGA-KIRC dataset through c-BioPortal (https://www.cbioportal.org/). The analysis also evaluated overall survival (OS) and disease-specific survival (DSS) with altered versus unaltered ST3Gals levels.

Survival analysis

Kaplan-Meier (K-M) method was utilized to analyze OS based on ST3Gals mRNA levels (high vs. low). Furthermore, we built a prognostic nomogram using “rms” package to forecast the impact of gene expressions and clinicopathological variables on OS. The performance of the prognostic risk model was evaluated using the Concordance index (C-index) to measure its predictive accuracy. Calibration curves were employed to confirm the precision of the model’s 1-year, 3-year, and 5-year survival forecasts by comparing them with actual patient outcomes.

Consensus clustering, tumor mutation burden (TMB), and immune infiltration analysis

After identifying ST3Gal1 and ST3Gal5 as key genes related to KIRC prognosis, we stratified patients into two clusters using “ConsensusClusterPlus” package [28]. For immune profile analysis, single-sample Gene Set Enrichment Analysis (ssGSEA) was conducted to assess the immune infiltration of 28 immune signatures using “GSVA”. Additionally, CIBERSORT was utilized to quantify the abundance of 22 different immune cells in KIRC patients. The ESTIMATE was applied to determine immune and stromal cells levels in KIRC tumor tissues using R package “estimate” [29]. Regarding TMB, we calculated the mutation frequency and types, depicted in waterfall plots created with “maftools” R package.

Therapeutic response prediction

We incorporated the tumor immune dysfunction and exclusion (TIDE) score and evaluated Merck18, IFN-γ, and PD-L1 levels from TIDE. The immunophenotypic score (IPS) was sourced from The Cancer Immunome Atlas (TCIA) (https://tcia.at/home), which includes scores for general immune phenotypes, as well as responses to CTLA-4 and PD-1 blockers, both individually and in combination.

Moreover, we employed “pRRophetic” to estimate drug response in two groups based on the half-maximal inhibitory concentration (IC50) data from the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org/). This approach provided insights into the likely efficacy of various cancer therapeutics on KIRC patient subgroups.

Immunohistochemistry (IHC) and hematoxylin and eosin (H&E) analysis

The ST3Gal1 and ST3Gal5 expression in tissue microarrays (Shanghai Outdo Biotech Company, China) of KIRC patients and paired normal tissues was detected by IHC according to our previous study [30]. Supplementary Table 3 showed the link between ST3Gal1/5 expression and clinicopathologic characteristics of KIRC patients.

The antibodies were used for IHC, such as ST3Gal5 (Proteintech, 14614-1-AP, 1:200), ST3Gal1 (Abcam, ab96129, 1:100), PCNA (Proteintech, 10205-2-AP, 1:200) and Ki67 (Cell Signaling Technology, 12202S, 1:200). The intensity was scored as 0 (no staining), 1 (weak staining), 2 (moderate staining) and 3 (strong staining). The proportion of positive cells was categorized into five categories: 0 (≤ 5%), 1 (6-25%), 2 (26-50%), 3 (51-75%) and 4 (> 75%). The final scores were determined by multiplying intensity and scores. The IHC scores were independently assessed by two blinded professional pathologists to minimize evaluation bias.

Cell culture and establishment of ST3Gal5 knockdown cells

Human KIRC cell lines 769-P, 786-O, and A-498 and human normal renal epithelial cell line HK-2 were purchased from Cell Bank of the Shanghai Life Science Institution, Chinese Academy of Sciences (Shanghai, China), cultured in RPMI 1640 medium (Gibco, Novato, CA, United States), MEM medium (KeyGEN BioTECH, China) and DMEM/F12 medium (KeyGEN BioTECH, China), respectively. The medium was supplemented with 10% fetal bovine serum (FBS, ExCell Bio, China) and 100 U/ml penicillin-streptomycin (Seven, Beijing, China) and maintained in a cell incubator at 37 °C with 5% CO2. All cell lines used in this study were confirmed to be mycoplasma negative every 3 months and authenticated by short tandem repeat (STR) profiling.

The pLKO.1 shRNA lentivirus system was used to generate shRNA lentivirus against human ST3Gal5 (ST3Gal5-KD), which was purchased from Applied Biological Materials (abm, Canada). 786-O cells at 30% confluency on a 6-cm dish were simultaneously transfected with ST3Gal5 knockdown lentivirus using 6 µg/ml polybrene (Sigma-Aldrich Chemical Co. India). ST3Gal5-KD cells were obtained by puromycin screening (1 µg/ml, Beijing Solarbio Science & Technology Co., Ltd.) in 786-O cells and obtained by puromycin screening (3 µg/ml) in 769-P cells.

Real-time quantitative PCR (RT-qPCR)

Total RNA from cells was extracted using RNAiso Plus reagent (TaKaRa, Japan). The RNA (1 µg) was reverse to cDNA, which was performed using SYBR Green Master Mix (Seven, Beijing, China). Relative expression of ST3Gal5 was calculated according to 2−ΔΔCT. Primers for ST3Gal5 (forward: 5’-CTGGGGCCGAGATAAGAACG-3’, reverse: 5’-CCAAAACCCGCCAAACTGAC-3’) and GAPDH (forward: 5’-TCCAAAATCAAGTGGGGCGA-3’, reverse: 5’-AAATGAGCCCCAGCCTTCTC-3’) were used.

Western blotting

Cells were treated with RIPA buffer for 30 min and centrifuged to obtain protein, whose concentration was measured using BCA Protein Assay Kit (Seven, Beijing, China). Equal quality protein samples were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred onto PVDF membranes (Millipore, USA). After incubation with 5% BSA for 2 hour (h), the membranes were incubated with anti-ST3Gal5 (1:1000, 14614-1-AP, Proteintech) or anti-GAPDH (1:10000, 10494-1-AP, Proteintech) at 4 °C overnight, followed by secondary antibody for 1 h, and visualized with an ECL system. Data quantification was performed by ImageJ software.

Cell Count Kit-8 (CCK-8) and colony formation assays

Cells were plated on 96-well plates with 2 × 103 cells per well. Then, 10 µl CCK-8 reagent (APExBIO Technology LLC, K1018, USA) was added to wells. The absorbance at 450 nm was measured using a microplate reader after incubating for 2 h. For colony formation assay, 0.5 × 103 cells per well were added to 6-well plates and incubated for 14 days. After fixed with 4% paraformaldehyde, cells were stained with 1% crystal violet. Each colony contained more than 50 cells.

Wound healing and transwell assay

Approximately 5 × 105 cells were planted to 6-well plates and allowed to proliferate to reach confluence. Subsequently, a 200 µl sterile pipette tip was employed to make a gap. Following an incubation period of either 6 h (786-O cells) or 23 h (769-P cells) in the medium without FBS, the same regions were subjected to photographic documentation.

For the migration assay, about 3 × 104 cells in 200 µl medium without FBS were added to upper compartment of a 24-well transwell chamber (8 μm pore size polycarbonate membrane, BIOFIL, China), with 550 µl medium with 10% FBS in lower compartment. After incubating for 24 h at 37℃ with 5% CO2, the upper compartment was washed with PBS and fixed, then stained with 1% crystal violet. After washing with PBS, the chambers were photographed. For invasion assay, the upper compartment was precoated with 40 µl matrigel and incubated for 2 h at 37℃, and about 3 × 104 cells were added to the upper compartment followed by the migration.

Immunofluorescence (IF) assay

Cells were fixed with 4% paraformaldehyde and treated with 0.1% Triton X-100 to allow permeabilization. Subsequently, cells were blocked by an immunostaining blocking solution (Beyotime, China) and then applied with anti-ST3Gal5 (1:100, 14614-1-AP, Proteintech) and incubated overnight at 4℃. The second antibody labeled with TRITC was used at 37℃ for 1 h and the nucleus was stained with DAPI (Abbkine, China), and immunofluorescence images were taken using microscope.

Xenograft tumor model

In all, 1 × 107 of 769-P cells (negative control (NC) or ST3Gal5-KD) were respectively injected subcutaneously into the groin of athymic male BALB/c nude mice (4–6 weeks old) from Beijing Vital River Laboratory Animal Technology. All animal experiments were approved by the Animal Research and Care Committee of Dalian Medical University (No. AEE24064). Tumor size in mice groin was measured using a vernier caliper and volumes were calculated applying the formula: 1/2 × (length × width2).

Statistical analysis

Statistical analyses and visualization were conducted using R (version 4.1.3), SPSS (version 26), GraphPad Prism (version 9.5, La Jolla, CA, USA) and ImageJ (version 1.51j8). A multiple Student’s t-test was performed on Figs. 1A and B, 3C and D and 6F, Supplementary Fig. 3B, D. An unpaired Student’s t-test was performed on Figs. 2A and B, 3A, F, H, J and K and 6B, C and E, Supplementary Fig. 2D-F, 3 C, F. A paired Student’s t-test was performed on Supplementary Fig. 2A. A one-way ANOVA test was performed on Figs. 4A, B and D and 5A-C, Supplementary Fig. 1A, B, 4A, B. A two-way ANOVA test was performed on Fig. 4E. The log-rank test was utilized on Figs. 1D-G, 2K and 3B and G, Supplementary Fig. 1D-F. p < 0.05 was considered to be statistically significant.

Fig. 1

The mRNA expression and prognostic values of ST3Gals in KIRC and genetic mutations in ST3Gals genes for KIRC. The mRNA levels of ST3Gals in TCGA-KIRC datasets (A), and GEO datasets (GSE53757 and GSE68417) (B). (C) Total mutation rates of ST3Gals genes were observed in KIRC patients. K-M method comparing the OS (D) and DSS (E) rates in KIRC patients with and without genetic alterations of ST3Gals. The K-M analysis of OS based on the top 25% and bottom 25% of mRNA levels of ST3Gal1 (F) and ST3Gal5 (G). (H) The univariate and multivariate survival analysis of ST3Gal1 and ST3Gal5 of KIRC patients. (I) The 1-, 3-, and 5-year recurrent rate of KIRC patients after tumor resection could exactly be predicted by the nomogram. (J) The Calibration plots showed the comparison between predicted and actual OS for 1-, 3- and 5-year survival probabilities. (K) The predictive effect of prediction model, prediction model without ST3Gal5, and other clinical prognostic factors of KIRC on OS was evaluated by C-Index. p < 0.05 was statistically significant

Fig. 2

Consensus clustering of KIRC patients based on ST3Gal1 and ST3Gal5. The differential expression of ST3Gal1 (A) and ST3Gal5 (B) in KIRC and normal tissues by TCGA and GEO databases. (C) Correlation between ST3Gal1 and ST3Gal5 expression in TCGA-KIRC database. (D) Representative immunohistochemistry images of ST3Gal1 and ST3Gal5 in tissue microarrays of KIRC patients and paired normal tissues (n = 30). (E) Heatmap corresponding to the consensus matrix for k = 2 using consensus clustering. (F) The relative change in area under the CDF curve. (G) Consensus clustering CDF curve for K values ranging from 2 to 9. (H) PCA analysis of two clusters. (I) Heatmap showed that TCGA-KIRC patients were divided into two clusters according to the expression of ST3Gal1 and ST3Gal5. (J) The correlation of clusters, ST3Gal1/ST3Gal5 and OS. (K) OS analysis of two clusters. p < 0.05 was statistically significant