Fig. 1 contains a comprehensive overview of the research workflow employed in the present study. First, GWAS summary data for different types of RCC were merged with summary data from blood-based eQTL, methylation QTL (mQTL) and protein QTL (pQTL) analyses, then the SMR method was used to identify candidate genes associated with ERS-related gene expression. To further validate these initial findings, Bayesian colocalization analysis were performed. Finally, cell experiments were conducted to corroborate the positive results in renal cancer progression.

A comprehensive search was conducted using the GeneCards database (https://www.genecards.org/) to identify relevant genes associated with ERS, employing the key word 'endoplasmic reticulum stress' as a guide. Following established protocols, only genes with relevance scores of ≥7 were considered for further analysis (16–18). In total, 1,193 ERS-related genes were identified through this process

The blood eQTL and mQTL data were procured utilizing methodologies that have been previously documented in the scientific literature (19–21). The eQTL summary data for genes related to ER stress was obtained from the eQTLGen consortium, a collaborative effort that gathered genetic data pertaining to gene expression from an extensive cohort of 31,684 individuals across 37 diverse datasets (19). Single-nucleotide variant (SNV)-gene pairs that had available data in multiple cohorts and a short genetic distance (≤1 Mb) were tested. Relevant information was obtained from the eQTLGen consortium webpage at https//www.eqtlgen.org/ciseqtls.html.

QTL datasets were selected based on the following: i) Matching European ancestry with RCC GWAS data; ii) sample size >1,000; and iii) rigorous quality control in the original studies. Exclusion criteria included: minor allele frequency <0.01, Hardy-Weinberg equilibrium P<0.0001, call rate <0.95 and platform-specific artifacts. To control for biases, population stratification (multidimensional scaling components), batch effects (normalization and probe matching) and cell type composition (surrogate variable analysis) were adjusted for. Linkage disequilibrium confounding were excluded using a Heterogeneity In Dependent Instruments (HEIDI) test (P>0.01) and Bayesian colocalization.

In a study of 35,559 individuals in Iceland, researchers used aptamers to measure plasma protein levels. This analysis identified 4,907 aptamers and associated genetic markers related to protein expression. From this data, 2,364 single-nucleotide polymorphisms (SNPs) potentially influencing the expression of 187 proteins were pinpointed which have been used in the present study (22).

To identify relevant instrumental variables for ERS-related genes, a significance threshold of P<5×10⁻⁸ was used. The mQTL summary data were originally from a comprehensive meta-analysis by Wu et al (23), and SNV data from the Yang work (23) were downloaded. This meta-analysis incorporated data from two European cohorts, the Lothian Birth Cohorts (n=1,366) and the Brisbane Systems Genetics Study (n=614) (23). The data were rigorously filtered, retaining only DNA methylation probes with a mQTL significance level of P<5×10⁻⁸ and SNVs located within 2 Mb of each DNAm probe. The datasets that were formatted in Biomedical Engineering Standard Data were retrieved from the Specified Medical Resource website (https://yanglab.westlake.edu.cn/software/smr/#mQTLsummarydata). The present study centered specifically on the mQTL summaries pertaining to ERS genes.

The dataset (24) comprised two comprehensive scans conducted by the IARC-Centre National de Genotypage, encompassing 11 studies from 18 countries. Specifically, it included 5,219 individuals diagnosed with RCC, consisting of 1,992 women and 3,227 men as well as 8,011 unaffected individuals (3,095 women and 4,916 men), all of European descent. The first scan used HumanHap 317k, 550 or 610Q arrays for genotyping, while the second scan used Omni5 and OmniExpress arrays (24).

The data used in the present research were obtained from publicly available databases that had already received ethical approval for their original research projects. Therefore, the present study did not require additional ethical clearance.

This analytical framework of the present study consisted of two main stages: SMR analysis and a subsequent sensitivity analysis. The SMR analysis formed the basis for the key findings, while the sensitivity analysis was performed to validate the robustness of these results. Detailed descriptions of both the SMR and sensitivity analyses can be found in the Table SI, Table SII, Table SIII, Table SIV, Table SV, Table SVI.

The initial analysis used a five-step method: i) SNPs as genetic factors, DNAm as the exposure and RCC as the outcome; ii) SNPs as genetic instruments, ER stress-related gene expression as the exposure and RCC as the outcome; iii) SNPs as genetic instruments, ER stress-related protein expression as the exposure and RCC as the outcome; iv) SNPs as genetic factors, DNAm as the exposure and ER stress gene expression as the outcome; and v) SNPs as genetic factors, ER stress-related gene expression as the exposure and protein expression as the outcome. Only findings satisfying the predefined significance thresholds (SMR P<0.05 and HEIDI P>0.01) from the first three steps were selected for further analysis.

To confirm potential causal relationships, the following criteria were used: i) An SMR P value of <0.05 in all SMR analyses; ii) P>0.01 for HEIDI in all SMR analyses; and iii) verification that the eQTL and mQTL or eQTL and pQTL matched the same gene symbols. Results were measured with odds ratios (ORs) determined by the equation OR=exp(β of SMR), where β represents the SMR regression coefficient (effect size), indicating the change in log odds of RCC per one standard deviation increase in the exposure trait, and exp denotes the exponential function (25).

A Bayesian colocalization analysis was performed to determine if a shared causal variant underlies GWAS summary data and different QTLs, specifically eQTLs, pQTLs and mQTLs. This analysis evaluated five primary hypotheses: H0, no association with either GWAS or QTL; H1, association with GWAS only; H2, association with QTL only; H3, association with both GWAS and QTL, but with distinct causal variants; and H4, association with both GWAS and QTL, with a shared causal variant. The posterior probability of each hypothesis was calculated. Following established protocols, all SNVs within 100 kb of the lead SNV associated with the probe were included in the colocalization analysis (26).

To perform colocalization analysis, all SNPs within a 100-kb window flanking the index SNP associated with the probe were identified. PPH4 of >0.7 was considered as strong evidence of overlap between the GWAS and QTL signals (27).

SMR software version 1.3.1 was used to perform the five-step SMR analysis and the HEIDI test. The ‘coloc’ R package (version 5.2.2) was used for the co-localization analysis (27). To account for multiple testing in the five-step SMR analysis, the Benjamini-Hochberg method was applied to correct for potentially incorrect results. P<0.05 was as defined statistical significance in the SMR dataset analysis.

One human proximal tubular epithelial cell line (HK-2; cat. no. GNHu47) and one human embryonic cell line (293T; cat. no. SCSP-502) were used as non-cancerous controls, along with two RCC cell lines (786-O; cat. no. TCHu186; Caki-1, cat. no. TCHu135). All cell lines were obtained from the China Center for Type Culture Collection and cultured in DMEM (Gibco; Thermo Fisher Scientific, Inc.; cat. no. 11875093) containing 10% FBS (Gibco; Thermo Fisher Scientific, Inc.; cat. no. 10099141C) and 1% penicillin-streptomycin-amphotericin B solution (Beyotime Biotechnology; cat. no. C0224-100ml) in an incubator at 37°C with 95% air and 5% CO₂. The cell lines were managed according to the guidelines outlined by the Chinese National Collection of Authenticated Cell Cultures. The HK-2 cell line served as a model of healthy kidney epithelial tissue, whereas the 786-O cell line is derived from ccRCC.

Confluency at the time of experiments was 80–95%. In total, three cell models were established: i) HK-2 cells (normal renal epithelium, control); ii) 293T cells treated with adriamycin (nephropathy model); and iii) 786-O cells (ccRCC). To induce nephropathy, 293T cells were seeded in 6-well plates (2×10⁵ cells/well) and treated with 2 µg/ml adriamycin for 24 h (24).

Total RNA was extracted using TRIzol reagent (Biosharp Life Sciences; cat. no. BS258A). Cells were lysed in 1 ml TRIzol, followed by the addition of 200 µl chloroform. The mixture was then centrifuged at 12,000 × g for 15 min at 4°C. The aqueous phase was collected, mixed with an equal volume of isopropanol, incubated at room temperature for 10 min and centrifuged at 12,000 × g for 10 min at 4°C. The resulting pellet was washed with 70% ethanol, air-dried and resuspended in RNase-free water.

RT was performed using the BeyoRT™ II First Strand cDNA Synthesis Kit (RNase H-) kit (Beyotime Biotechnology; cat. no. D7168S). qPCR was then conducted on an Eppendorf real-time PCR instrument using BeyoFast™ SYBR Green qPCR Mix (Beyotime Biotechnology; cat. no. D7260). The 20 µl reaction contained: 2X SuperMix 10 µl, forward and reverse primers (10 µM) 0.4 µl each, template 2 µl and water 6.8 µl. The thermocycling conditions were as follows: 95°C 3 min; 40 cycles of 95°C 30 sec, 55°C 20 sec and 72°C 20 sec. The relative expression level of target genes was normalized to β-actin as an internal control and calculated using the 2^−ΔΔCq method.

In total, 5 ER stress markers were detected: GRP78, PERK, IRE1α, activating transcription factor 6 (ATF6) and C/EBP homologous protein (CHOP). The primer sequences were as follows (5′-3′): GRP78 (also known as HSPA5) forward GAAAGGATGGTTAATGATGCTGAG and reverse GTCTTCAATGTCCGCATCCTG; PERK forward ACGATGAGACAGAGTTGCGAC and reverse ATCCAAGGCAGCAATTCTCCC; IRE1 (also known as ERN1) forward CCGAACGTGATCCGCTACTTCT and reverse CGCAAAGTCCTTCTGCTCCACA; ATF6 forward TCCTCGGTCAGTGGACTCTTA and reverse CTTGGGCTGAATTGAAGGTTTTG; CHOP forward GGAAACAGAGTGGTCATTCCC and reverse CTGCTTGAGCCGTTCATTCTC; β-actin forward GATGAGATTGGCATGGCTTT and reverse GTCACCTTCACCGTTCCAGT.

To validate the SMR findings, the expression levels of seven key genes [Polycystin 1 (PKD1), C-C Motif Chemokine Ligand 2 (CCL2), PTEN induced kinase 1 (PINK1), estrogen receptor 1 (ESR1), Advanced Glycosylation End-Product Specific Receptor (AGER), Insulin Receptor (INSR), and Dolichyl-Di-N-Acetylglucosamine Phosphodiesterase Alpha- And Beta-Subunits (DDOST)] were quantified. The primer sequences were as follows (5′-3′): PKD1 forward GCCAGAGCGCATAACGAA and reverse TACCCACCACTCACCTCC; CCL2 forward CCCAAGCAGAAGTGGGTTCA and reverse GTGTCTGGGGAAAGCTAGGG; PINK1 forward 5′-GCCTCATCGAGGAAAAACAGG and reverse GTCTCGTGTCCAACGGGTC; ESR1 forward CCCACTCAACAGCGTGTCTC and reverse CGTCGATTATCTGAATTTGGCCT; AGER forward GTGTCCTTCCCAACGGCTC and reverse ATTGCCTGGCACCGGAAAA; INSR forward AAAACGAGGCCCGAAGATTTC and reverse GAGCCCATAGACCCGGAAG; DDOST forward TACGCTCATCGTGGCTGACACT and reverse CCAGCACCAAAGGGTTATCAGG.

For ER stress inhibition, cells were pretreated with 4-PBA (2 mM, 24 h, at 37°C in a 5% CO₂ atmosphere; MedChemExpress; cat. no. HY-A2081) prior to protein extraction. Total protein was extracted using RIPA lysis buffer (Beyotime Biotechnology; cat. no. P0013B) supplemented with protease and phosphatase inhibitors (Beyotime Biotechnology; cat. nos. P1005 and P1081). Protein concentration was then determined by BCA assay (Beyotime Biotechnology; cat. no. P0012). Equal amounts of protein (30 µg) were separated by 10% SDS-PAGE and transferred to PVDF membranes (MilliporeSigma; cat. no. IPVH00010). After blocking with 5% non-fat milk in TBST (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20, pH 7.6) for 1 h at room temperature, membranes were incubated overnight at 4°C with primary antibodies against phospho-IRE1α (Affinity Biosciences; cat. no. AF7150; 1:1,000), IRE1α (Cell Signalling Technology, Inc.; cat. no. 3294T; 1:1,000), PERK (Cell Signalling Technology, Inc.; cat. no. 3192S; 1:1,000), phospho-PERK (Affinity Biosciences; cat. no. DF7576; 1:1,000), GRP78 (Abcam; cat. no. ab108613; 1:1,000), CHOP (Cell Signalling Technology, Inc.; cat. no. 2895T; 1:1,000), eIF2α (Cell Signalling Technology, Inc.; cat. no. 5324T; 1:1,000), phospho-eIF2α (Cell Signalling Technology, Inc.; cat. no. 3398T; 1:1,000), ATF6 (Abcam; cat. no. ab227830; 1:1,000) and β-actin (Sigma-Aldrich; Merck KGaA; cat. no. A5316; 1:5,000). After washing with TBST, membranes were incubated with HRP-conjugated anti-rabbit (1:5,000; cat. no. 111-035-144; Jackson ImmunoResearch Laboratories; Inc.) and HRP-conjugated anti-mouse (1:5,000; cat. no. 115-035-146; Jackson ImmunoResearch Laboratories, Inc.) secondary antibodies for 1 h at room temperature. Protein bands were visualized using ECL (cat. no. 1705061; Bio-Rad Laboratories, Inc.) and captured by the ChemiDoc XRS+ system. Densitometric analysis was performed using ImageJ (version 2.16.1; Fiji distribution; National Institutes of Health).

Cells were fixed in 2.5% glutaraldehyde overnight at 4°C, post-fixed in 1% osmium tetroxide for 1 h at room temperature, dehydrated in a graded ethanol series (30, 50, 70, 80, 90, 95 and 100% ethanol, 15 min each at room temperature) and embedded in epoxy resin (polymerized at 60°C for 48 h). Ultrathin sections (70 nm) were stained with uranyl acetate for 15 min at room temperature and then with lead citrate for 8 min at room temperature. Images were captured using a JEOL JEM-1010 electron microscope at 80–120 kV. All TEM images were acquired at 500 nm.

To investigate the role of ERS in sunitinib resistance in ccRCC, sunitinib-sensitive and sunitinib-resistant cell lines were compared using two distinct methods: Western blot analysis of ERS-related protein expression and TEM for ultrastructural observation. Sunitinib-resistant cell lines were developed by exposing ccRCC cells to low doses of sunitinib (cat. no. SU 11248; MedChemExpress).

The methylation status of the genes was determined using MSP. PCR amplification was performed using 150–200 ng of treated DNA as a template, with two primer sets designed for the same target region. One set was specific to methylated DNA at cytosine-phosphate-guanine (CpG) sites and the other to unmethylated DNA. Each 20 µl PCR reaction contained 1 U of Hot start EX Taq DNA polymerase (Takara Bio), 2 µl of 10X EXtaq Buffer, 2 µl of dNTP mixture, 8 pmol each of forward and reverse primer and 1 µl of DNA template. The PCR conditions were as follows: Initial denaturation at 96°C for 3 min, followed by 40 cycles of denaturation at 96°C for 30 sec, annealing at 60°C for 30 sec and extension at 72°C for 30 sec, with a final extension at 72°C for 4 min. PCR products were visualized on either 2% agarose gels or 6% non-denaturing polyacrylamide gels. Primer sequences and expected product sizes are detailed in Table I.

Statistical analyses were performed using GraphPad Prism software (version 9.5; Dotmatics). Data are expressed as mean ± SEM. All experiments were repeated at least three times. Differences between two groups were analyzed using an unpaired two-tailed Student's t-test. For multiple comparisons, one-way or two-way ANOVA followed by Tukey-Kramer's multiple-comparison test was used. Two-way ANOVA was specifically used to analyze data involving two independent variables, such as cell type and drug treatment condition. P<0.05 was considered to indicate a statistically significant difference.

SMR was employed to investigate the relationship between SNVs, DNA methylation sites and ERS-related genes. Using blood mQTL data, 4,234 CpG sites associated with 1,193 ERS-related genes were identified, which were linked to 64,238 SNPs (Table SI).

Subsequently, the mQTL information for ERS genes was integrated with the GWAS results for RCC. This integration identified 3 DNA methylation probes (cg07431106, cg12446953 and cg08884395) corresponding to 3 ER stress-related genes (CCL2, PINK1 and ESR1), with SMR P<0.01 and HEIDI P<0.01 (Fig. 2). Causality estimates are expressed as β-values, representing the effect size for a one standard deviation change in ER stress-associated gene expression levels. Additionally, 2 protective associations between methylation levels and RCC (cg07431106, β=−0.18; cg12446953, β=−0.544) and 1 progression association suggesting increased risk of RCC (cg08884395, β=0.563) were observed.

Analysis of eQTL data revealed 985 eQTL probes linked to ERS genes, which were connected to 15,618 SNPs (Table SII). Integration of the eQTL data with the RCC GWAS data identified 834 ERS-related genes. One gene (PKD1) demonstrated a protective effect against RCC (Fig. 2), with a β-value of −1.44 (SMR, P< 0.001; HEIDI, P=0.836).

In total, 187 potential pQTL tools related to genetic variations were identified, suggesting a connection between the levels of 175 proteins and the risk of RCC (Table SIII). Specifically, integrating pQTL data with RCC GWAS summary data identified 3 protein probes (SeqId_4125_52, SeqId_3448_13 and SeqId_17161_1) corresponding to 3 ERS-related genes (AGER, INSR and DDOST). These probes exhibited significant associations with RCC risk (SMR, P<0.01; HEIDI, P>0.01; Fig. 2). Probe SeqId_17161_1 demonstrated the strongest cancer-promoting effect (β=1.71). Probe SeqId_4125_52 also promoted cancer progression (β=0.458), whereas probe SeqId_3448_13 acted as a protective factor against kidney cancer (β=−1.87). Sensitivity, heterogeneity and multi-effect analysis results are presented in Tables SIV, SV and SVI, respectively.

Bayesian co-localization analysis was used to eliminate potential confounding variables associated with RCC. The results indicated that 7 variants shared a causal relationship with RCC: 3 DNA methylation CpG sites (cg07431106, PPH4=0.786; cg12446953, PPH4=0.813; cg08884395, PPH4=0.915), 1 gene (PKD1, PPH4=0.944) and 3 proteins (AGER, PPH4=0.901; INSR, PPH4=0.875; DDOST, PPH4=0.738) (Fig. 3).

Given the well-known influence of gene methylation on gene expression, the potential association between DNA methylation and gene expression was investigated. Specifically, DNA methylation was used as the predictor variable and gene transcripts as the response variable. Using strict criteria (SMR FDR<0.05 and HEIDI P>0.01), specific regulatory connections were identified where the activity of 2 genes related to ER stress were influenced by 10 CpG sites associated with DNA methylation. For example, 2 methylation sites (cg01961278 and cg17183232) were significantly associated with PINK1 expression and positively correlated with it. For ESR1, 8 methylation sites (cg15626350, cg00601836, cg04063345, cg21157690, cg04211581, cg17264271, cg24764793 and cg06611115) exhibited varying regulatory effects on ESR1 expression (Tables SII and SIII).

Using the same screening criteria (SMR FDR<0.05 and HEIDI P>0.01), no regulatory relationships between ERS-related proteins and genes were identified (Tables SII and SIII).

Additional molecular tests confirmed that ER stress may play a crucial role in ccRCC progression. Based on three cell models (ccRCC, human embryonic kidney-derived cells treated with adriamycin as a general cellular stress control and renal epithelial cell lines), the level of ER stress increased with disease progression (ccRCC cell lines > human embryonic kidney > renal epithelial cell lines). The levels of 5 ER stress-related molecules (GRP78, PERK, IRE1α, ATF6 and CHOP) were highest in the ccRCC cell line and lowest in the renal epithelial cell line (P<0.01; Fig. 4A). TEM analysis revealed progressive ultrastructural alterations characteristic of ER stress, including ER swelling, dilatation and vacuolization, which were most pronounced in 786-O cells, moderate in adriamycin-treated 293T cells, and minimal in HK-2 cells (Fig. 4A).

The potential importance of ER stress in ccRCC resistance to sunitinib was investigated by studying two sunitinib-resistant ccRCC cell lines [786-O (sunitinib) and Caki (sunitinib)] and comparing them with sensitive cell lines. Both resistant cell lines exhibited increased expression of 5 genes (GRP78, PERK, IRE1α, ATF6 and CHOP) related to ER stress. TEM analysis confirmed these findings by showing marked ER dilatation and swelling in sunitinib-resistant cells compared to their sensitive counterparts, indicating enhanced ER stress in the resistant phenotype (Fig. 4B and C).

To explore the role of ER stress in sunitinib resistance in ccRCC, the ER stress inhibitor 4-PBA was introduced and western blotting was used to assess 5 ER stress markers (GRP78, PERK, IRE1α, ATF6 and CHOP) in the sunitinib-resistant and sensitive ccRCC cell lines. The results showed that these ER stress markers were significantly higher in sunitinib-resistant ccRCC cell lines than in sensitive cell lines. After 4-PBA intervention, the half-maximal inhibitory concentration of sunitinib was markedly decreased in both sunitinib-resistant RCC cell lines; the IC₅₀ value of the 786-0 resistant strain decreased from 24.67 to 13.22 µM, and the IC₅₀ value of the Caki-1 resistant strain decreased from 16.79 to 9.76 µM (Fig. S1). Simultaneously, the expression of the ER stress-related proteins in resistant cell lines decreased markedly when treated with 4-PBA compared with the sunitinib-treated group,. Specifically, the phosphorylation of PERK and IRE1α, and downstream effectors such as CHOP, were markedly reduced when treated with 4-PBA compared with sunitinib-treated group. Additionally, the levels of GRP78 and ATF6 declined, indicating that 4-PBA effectively inhibited the ER stress response (Fig. 4D) when treated with 4-PBA compared with sunitinib-treated group.

The expression levels of 7 key genes, CCL2 (cg07431106), ESR1 (cg08884395), PINK1 (cg12446953), PKD1 (ENSG00000008710), AGER (SeqId_4125_52), INSR (SeqId_3448_13) and DDOST (SeqId_17161_1), selected through SMR analysis, were focused on in this analysis. Comparing the expression levels of these 7 genes across the three cell models revealed significant differences (P<0.05; Fig. 5A), indicating differential expression. Specifically, compared with normal renal HK-2 and 293T cells, the expression levels of all 7 molecules were upregulated in 786-O cells, with the highest expression levels observed in ccRCC cells, suggesting a positive correlation between the degree of ER stress and ccRCC. Furthermore, the methylation levels of three key methylation sites (cg07431106, cg08884395 and cg12446953) in the ccRCC (786-0 cell line), human embryonic kidney (293T cell line) and renal epithelial (HK2 cell line) cell models were analyzed. The results suggested that cg0743110 showed methylation characteristics in ccRCC cells but demethylation characteristics in nephropathy and renal epithelial cell models. Site cg08884395 showed methylation characteristics in the nephropathy model but demethylation characteristics in ccRCC or renal epithelial cell models. Site cg12446953 showed methylation characteristics in ccRCC or renal epithelial cell models but demethylation characteristics in the nephropathy model (Fig. 5B).