Exploring the functional role of palmitoylation genes in clear cell renal cell carcinoma and their impact on tumor microenvironment

Abstract

Objective

The current work focused on elucidating the regulatory mechanisms of palmitoylation-related genes (PRGs) in influencing the occurrence and development of clear cell renal cell carcinoma (ccRCC), and identifying possible biomarkers and therapeutic targets.

Methods

Integrative analyses were performed using multiple databases. Two-sample Mendelian randomization (MR) was first applied for identifying PRGs causally associated with ccRCC. Findings were then validated by summary-data-based Mendelian randomization (SMR), bioinformatics analyses, and experimental verification. Functional enrichment and molecular docking were used to predict target interactions, while single-cell transcriptomic analysis was performed for assessing the relations of key PRGs with the tumor microenvironment (TME). Mediation analysis further explored upstream and downstream regulatory mechanisms.

Results

In total, 22 PRGs associated with ccRCC were identified. Among these, ZDHHC18 emerged as a significant risk factor based on both MR and SMR results. Bioinformatics, molecular docking, and laboratory assays revealed that ZDHHC18 was upregulated and validated LYPLA2 as its downstream palmitoylation target. Pathway enrichment and single-cell analyses indicated that ZDHHC18 might shape immune heterogeneity in ccRCC by influencing multiple immune cell populations within the TME. Mediation analysis suggested that mitochondrial DNA copy number (mtDNA-cn) regulated ZDHHC18 expression, which in turn promoted ccRCC progression through four specific immune cell traits.

Conclusion

This study defines an mtDNA-cn-ZDHHC18-immune cell traits axis that mechanistically links mitochondrial function, palmitoylation, and immune modulation in ccRCC. These findings expand the current understanding of palmitoylation in renal cancer biology and identify ZDHHC18 as a potential metabolic-immune regulator and therapeutic target.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12863-026-01443-5.

Introduction

Clear cell renal cell carcinoma (ccRCC) has the highest prevalence among diverse subtypes of renal cell carcinoma (RCC) subtypes, taking up 70–80% of renal cancer cases [1, 2]. The malignancy poses a significant threat to both patients and the society. In advanced stages, the 5-year survival rate is < 10%, which, and the high treatment costs place a substantial burden on healthcare systems [3]. Current therapeutic strategies for ccRCC include surgical resection, targeted therapy, and immunotherapy. However, these approaches are often limited by high resistance rates and considerable interindividual variability in treatment efficacy [4, 5]. Consequently, identifying effective biomarkers and therapeutic targets is urgently needed to improve patient prognosis. This study aimed to address these gaps by exploring the molecular mechanisms underlying ccRCC, thereby underscoring the critical need for innovative research in this area.

The evolution of cancer therapy increasingly emphasizes biomarker-driven strategies, highlighting molecular signatures that can predict prognosis and therapeutic responsiveness. Within this framework, understanding novel regulatory layers, such as post-translational modifications (PTMs), may expand opportunities for mechanistic discovery and clinical translation. Among various PTMs—including phosphorylation, acetylation, and ubiquitination—palmitoylation, which indicates that palmitic acid is reversibly attached onto cysteine residues, uniquely integrates lipid metabolism with cell signaling and membrane dynamics [6–8]. In the context of cancer, aberrant palmitoylation is implicated in tumorigenesis and cancer progression [9–11]. Since PTMs are recognized as important contributors to ccRCC pathogenesis, palmitoylation has gradually become a focus of research in the field. Sun et al. revealed a new ZDHHC2-AGK axis affecting sensitivity of ccRCC cells to Sunitinib [12]. In a recent study, Xu et al. established a prognosis prediction model by incorporating palmitoylation-related genes (PRGs) of ccRCC, demonstrating significant predictive ability [13]. Palmitoylation in a majority of cancer-associated proteins is under the catalysis of ZDHHC proteins [14]. Although increasing evidence supports a role for palmitoylation in cancer, major gaps remain in our understanding of the specific enzymes involved and the associated regulatory mechanisms [15–17]. Moreover, the potential of targeting palmitoylation for antitumor therapy remains underexplored, underscoring the need for detailed research in this area.

Mitochondrial metabolism and signaling offer a compelling yet underinvestigated link to this regulation. Mitochondria play a vital role in energy production, apoptosis, and cellular metabolism [18]. Mitochondrial DNA copy number (mtDNA-cn) is a key determinant of energy metabolism and is related to diverse disorders [19]. Recently, mtDNA-cn is found to be important for cancer biology, suggesting that its alterations may affect tumorigenesis and progression [20, 21]. Previous studies indicate that mitochondrial functional status regulates the activity of key transcription factors through metabolic and oxidative signals [22]. Given that mitochondrial outputs can influence the activity of acyltransferases and metabolic enzymes via both transcriptional and post-translational mechanisms [23–25], it is plausible that fluctuations in mtDNA-cn may modulate protein palmitoylation by altering the expression or function of ZDHHC enzymes. However, this potential connection between mtDNA-cn and palmitoylation remains poorly characterized, representing a critical knowledge gap in our understanding of ccRCC metabolic plasticity and signaling adaptation.

Therefore, a multi-omics integrated analysis combining two-sample Mendelian randomization (MR), summary-data-based MR (SMR), single-cell sequencing, and immune infiltration analyses was conducted in the present work to explore functions of palmitoylation-modified genes, with a focus on ZDHHC18, in ccRCC. This study aimed to elucidate the upstream and downstream molecular mechanisms through which ZDHHC18 affects the occurrence and development of ccRCC, thereby laying a foundation for discovering new biomarkers and formulating treatments, and addressing the urgent need for more effective interventions in diagnosing and treating advanced ccRCC.

Methods and materials

Data collection

The eQTL data of each gene were acquired from eQTLGen database (https://www.eqtlgen.org/cis-eqtls.html). These data consisted of cis-eQTLs derived from blood samples. PRGs were obtained from the study of Chen et al., and a detailed list is provided in Table S1 [26]. Genome Wide Association Study (GWAS) data for ccRCC were downloaded from the GWAS Catalog database (https://www.ebi.ac.uk/gwas/). These data originated from a study by Purdue et al., which included a total of 759,800 participants: 752,817 individuals of European ancestry, 3,526 African American or Afro-Caribbean participants, and 3,457 Hispanic or Latin American participants (accession number: GCST90320055) [27]. In addition, we collected GWAS data for mtDNA-cn from Qin et al.’s study (accession number: GCST90026372) [28]. Furthermore, GWAS summary data of each immune cell trait were extracted from the study by Orrù V et al. (accession numbers: GCST001391 to GCST002121) [29]. Patient data used for differential gene expression and immune infiltration analyses were derived from the KIRC cohort in TCGA database (https://cancergenome.nih.gov/). The gene expression matrix for ccRCC cells was collected from CCLE database (https://depmap.org/portal/data_page/?tab=allData). Single-cell analysis data were obtained from GEO database (https://www.ncbi.nlm.nih.gov/geo/), including a control group (accession number: GSE131685) and a tumor group (accession number: GSE121636, GSE171306) [30–32]. All single-cell analysis data used were derived from tissue samples. Some single-cell annotation data were obtained from TISCH database (http://tisch.comp-genomics.org/). In accordance with rigorous MR analysis requirements, all data were screened for single nucleotide polymorphisms (SNPs) to be used as instrumental variables (IVs) in line with the criteria below: p-value < 5e-8, clumping window of 10,000 kb, and r² < 0.1. Palindromic and ambiguous SNPs were excluded from the IVs used in the MR analysis.

Two-sample MR analysis

Two-sample MR analysis was carried out with “TwoSampleMR” package. This analysis consisted of five main parts: (1) cis-eQTL data corresponding to PRGs as the exposure and ccRCC as the outcome; (2) immune cell traits as the exposure while ccRCC as the outcome; (3) cis-eQTL data corresponding to target genes as the exposure while immune cell traits as the outcome; (4) mtDNA-cn data as the exposure and ccRCC as the outcome; and (5) mtDNA-cn as the exposure and cis-eQTL data corresponding to target genes as the outcome. To reduce bias caused by weak IVs, instruments with an F-value < 10 were removed prior to analysis. The primary analytical method was inverse variance weighted (IVW). Exposures for which IVW results could not be calculated were excluded. Pleiotropy was assessed using the MR-Egger intercept, heterogeneity was evaluated by Cochran’s Q, and the robustness of the results was examined through sensitivity analysis by the leave-one-out method. There was no sample overlap between the GWAS datasets serving as exposure and outcome.

SMR analysis

For SMR analysis, eQTL and GWAS summary data were used to evaluate those pleiotropic relationships of gene expression levels with the traits of interest. The HEIDI test was performed for assessing potential horizontal pleiotropy of co-localized signals. The null hypothesis of HEIDI test states no horizontal pleiotropy, and this indicates the situation where an individual genetic variant exerts a combined effect on multiple trait independently of the pathway related to the analyzed traits. Through SMR analysis combined with the HEIDI test, it is possible to determine whether SNPs influence the phenotype via protein levels or through other biological pathways. In this study, the smr-1.3.1-win.exe software tool was downloaded at the SMR website (https://yanglab.westlake.edu.cn/software/smr/#Overview) and used for SMR analysis with default parameters.

Interacting gene analysis and co-expression analysis

First, STAR-counts and relevant clinical data of ccRCC patients were obtained from TCGA database. Data in the transcripts per million (TPM) format were then extracted and normalized. Finally, samples with available RNA sequencing (RNAseq) and clinical data were retained, yielding altogether 532 samples for subsequent analyses. Candidate genes significantly associated with the target gene were screened using Pearson correlation analysis. Specifically, genes with a correlation coefficient satisfying p < 0.05 and |r| > 0.5 were considered significantly correlated, and result visualization was completed with the “circlize” package. Subsequently, a target gene interaction network was constructed by using STRING database (v12.0), with an interaction confidence threshold set to > 0.4 (Medium confidence) to extract protein-protein interaction (PPI) relationships. To further explore the functions of these interacting genes, functional enrichment analyses, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment, were performed using the “ClusterProfiler” package. Furthermore, correlation expression maps were drawn with “ggplot2” package for visualizing co-expression patterns of interacting genes.

Molecular docking analysis

In this part, GPS-Palm Software (https://gpspalm.biocuckoo.cn/index.php) was employed to identify potential palmitoylation sites in the identified interacting genes. The corresponding gene sequences were collected from UniProt database (https://www.uniprot.org/). The three-dimensional (3D) structures of desired proteins were generated from these sequences using the SWISS-MODEL tool (https://swissmodel.expasy.org/). Finally, molecular docking was performed using HADDOCK v2.4 (https://rascar.science.uu.nl/) to assess potential binding affinities between the modification sites and the domains of the identified palmitoylated proteins.

Single-cell sequencing analysis

Prior to analysis, single-cell data were subjected to quality control (QC) and filtering based on several metrics, including the gene count in each cell, total gene count, and mitochondrial genome percentage. The data were then batch-normalized. After identifying highly variable genes, principal component analysis (PCA) was carried out for calculating significance of each principal component (PC), thereby determining the optimal number of PCs for clustering. Dimensionality reduction was carried out using UMAP and tSNE to project high-dimensional points onto a two-dimensional plane, enabling the distinction of different clusters. Finally, cell types were identified and annotated with marker genes for each cluster, and the corresponding cell proportions were calculated.

Immune infiltration analysis

After filtering the ccRCC patient data derived from the TCGA database, the ssGSEA algorithm was first utilized to evaluate immune infiltration levels of the target genes in key immune cell types. Potential immunotherapy responses were predicted with the TIDE algorithm. Data visualization was performed with the “ggplot2” package.

Mediation analysis

Based on the two-sample MR results, mediation analysis was performed separately for upstream and downstream pathways. In the downstream analysis, the total effect of ZDHHC18 on ccRCC was calculated and labeled as beta_all. Next, the effect of ZDHHC18 on immune cell traits was computed and denoted as beta1, while that of immune cell traits on ccRCC was denoted as beta2. The mediation effect (beta12) was then computed as beta12 = beta1 * beta2, and the direct effect (beta_dir) as beta_dir = beta_all-beta12. The mediation proportion (beta12_p) was computed as beta12_p = beta / beta_all * 100%. In the upstream analysis, the total effect of mtDNA-cn on ccRCC was determined (beta_all), along with the effect of mtDNA-cn on ZDHHC18 (beta1) and the effect of ZDHHC18 on ccRCC (beta2). beta12, beta_dir and beta12_p were then calculated in the same manner.

Cell lines and cell culture

HK-2 (human renal cortical proximal tubule epithelial cells), 786-O, 769-P, and OS-RC-2 (human renal cell carcinoma cells) were purchased from Pricella Biotechnology, China. Cells were routinely cultivated within 1640-medium containing 10% fetal bovine serum alongside 1% penicillin-streptomycin under 37 °C with 5% CO₂. To validate ZDHHC18-mediated palmitoylation, cells were subjected to transient transfection using plasmids expressing HA-tagged LYPLA2, Flag-tagged ZDHHC18 (Wild type, WT) or the corresponding catalytically inactive mutant (DHHS, where the cysteine residue in the DHHC motif is substituted with serine), using Lipofectamine 5000. To inhibit global palmitoylation, cells were exposed to 2-bromopalmitate (2-BP) treatment at 50 µM for 8 h prior to harvest. Ethidium bromide (EtBr) was used to deplete mtDNA. The medium containing EtBr was replaced once every 24 h for a total of 4 days. The materials utilized in basic experiments are detailed in Table S2.

Real time-quantitative polymerase chain reaction (RT-qPCR)

Total RNA extraction was completed with the RNA Rapid Extraction Kit following specific protocols. RNA concentration and quality were analyzed using a NanoDrop 2000 spectrophotometer (an A260/A280 ratio between 1.8 and 2.0 was considered acceptable). Reverse transcription was then performed using PrimeScript™ RT Master Mix. A 20 µL reaction system containing 1 µg of total RNA was prepared. The reaction conditions were as follows: reverse transcription at 37 ℃ for 15 min, and enzyme inactivation at 85 ℃ for 5 s. By utilizing TB Green™ Premix Ex Taq™, qPCR was performed by the SYBR Green method on a Bio-Rad CFX96 real-time fluorescence quantitative PCR instrument. Primer sequences are displayed in Table S3. Relative target gene expression levels were measured by the ΔΔCt approach, with β-actin as an endogenous control. Each assay was conducted in triplicate.

Western blotting

Cell samples were lysed with RIPA lysis buffer (including 1% protease inhibitor), while protein concentrations were determined by the BCA method. Later, 30 µg of total protein underwent SDS-PAGE for separation using a Page gel rapid preparation kit (12.5%) at a constant voltage of 80–120 V, before transfer to polyvinylidene fluoride (PVDF) membrane (constant current 250 mA, 50 min). After blocking with the rapid blocking solution for 30 min under ambient temperature, these PVDF membranes were precisely sectioned according to pre-determined molecular weight ranges using a clean surgical blade, guided by the pre-stained protein marker (25–70 kDa). These membranes then received primary antibody incubation (ZDHHC18, 1:1000, LYPLA2, 1:1000; GAPDH, 1:3000) overnight at 4 °C, rinsed thrice using TBST, and further probed with HRP-labeled secondary antibodies (1:5000) for 60 min under ambient temperature. Following three TBST washes, signals were developed with ECL chemiluminescence reagent. Band intensities were analyzed by ImageJ software. Target protein levels were determined based on the internal reference GAPDH, and each experiment was repeated three times.

Immunohistochemistry (IHC)

In this study, three ccRCC and matched adjacent non-carcinoma tissues were retrospectively collected from the First Affiliated Hospital of Chongqing Medical University, and the ethics committee approved the study protocols. All procedures were performed following the Declaration of Helsinki. Tissues were subjected to 24 h of 4% paraformaldehyde fixation under 4 °C, then dehydrated in gradient ethanol (30% to 100%) and embedded in paraffin. Section  (4 μm) were subsequently dewaxed in xylene and rehydrated with gradient alcohol solutions. Antigen retrieval was performed in 10 mM citrate buffer (pH 6.0) through microwave heating (100 °C, 10 min). Endogenous peroxidase activity was quenched with 3% H₂O₂ for 15 min. After blocking with 5% normal goat serum for 30 min under ambient temperature, sections received overnight primary antibody incubation under 4 °C (ZDHHC18, 1:1000). For negative control, the primary antibody was replaced with PBS. Immunoreactivity was detected using an HRP-conjugated secondary antibody and developed using 3,3′-diaminobenzidine (DAB) chromogen for 3 min. Sections were processed through hematoxylin counterstaining for 30 s, dehydration and mounting. Staining intensity was assessed using the IHC Profiler plug-in in Image J software.

Immunofluorescence analysis

Paraformaldehyde-fixed tissues were subjected to permeabilization using 0.1% Triton X-100, 5% BSA blocking, and overnight primary antibody incubation under 4 °C (ZDHHC18, 1:500; PD-1: 1:500; FOXP3: 1:100; CD11c, 1:500). Following PBS washes, samples underwent fluorochrome-conjugated secondary antibody incubation. After 5 min of DAPI treatment (1 µg/mL) for nuclear counterstaining, slides were mounted with ProLong antifade mountant.

Statistical methods

R (version 4.1.1, http://www.r-project.org) was adopted for statistical analysis. All software packages used for correlation analysis and visualization are indicated in the corresponding sections of the Methods. Forest plots were created using the grid and “forestploter2” package. Receiver operating characteristic (ROC) curves were generated with “pROC” package. A p-value < 0.05 suggested statistical significance.

Results

Screening of PRGs associated with ccRCC

First, all cis-eQTL genes were intersected with the list of PRGs, yielding 22 PRGs with complete cis-eQTL data (Fig. 1A) (Table 1). Based on the results of two-sample MR analysis, four of these genes were causally associated with ccRCC (p < 0.05). The results of the sensitivity analysis are presented in Figure S1. Heterogeneity and pleiotropy results are summarized in Table S4. The odds ratios (ORs) indicated ZDHHC14 and ZDHHC2 as protective factors, whereas ZDHHC18 and ZDHHC6 as risk factors (Fig. 1B).

Fig. 1.

Screening of palmitoylation genes associated with ccRCC. (A) Venn diagram showing the intersection of palmitoylation gene list and eQTL dataset genes; (B) Forest plot showing palmitoylation genes with causal relationship with ccRCC screened by two-sample MR analysis. (C) Forest plot showing SMR analysis results; (D) Expression of ZDHHC6 and ZDHHC18 in TCGA-KIRC cohort; (E) Relationship between ZDHHC18 expression and pathological T stage; (F) Relationship between ZDHHC18 expression and pathological N stage; (G) Relationship between ZDHHC18 expression and pathological M stage; (H) Relationship between ZDHHC18 expression and pathological grade. (I) Relationship between ZDHHC6 expression and pathological T stage; (J) Relationship between ZDHHC6 expression and pathological N stage; (K) Relationship between ZDHHC6 expression and pathological M stage; (L) Relationship between ZDHHC6 expression and pathological grade

Table 1.

Intersection genes between palmitoylation gene list and eQTL dataset genes

Genes
PPT1
PPT2
ZDHHC1
ZDHHC11
ZDHHC12
ZDHHC13
ZDHHC14
ZDHHC16
ZDHHC17
ZDHHC18
ZDHHC19
ZDHHC2
ZDHHC20
ZDHHC21
ZDHHC23
ZDHHC24
ZDHHC3
ZDHHC4
ZDHHC5
ZDHHC6
ZDHHC7
ZDHHC8

Verification of the screened genes by SMR analysis and bioinformatics analysis

SMR analysis was carried out on those four screened PRGs, and ZDHHC18 and ZDHHC6 showed significant risk effects (p < 0.05) (Fig. 1C). For these two genes, their expression in the KIRC cohort was analyzed using the TCGA database. Both genes showed significant differential expression: ZDHHC6 showed low expression and ZDHHC18 exhibited high expression (Fig. 1D). From further stratification analysis, ZDHHC18 expression markedly increased in advanced ccRCC (T3&T4, N1, M1, Stage III&IV) (Fig. 1E H), while ZDHHC6 showed a decreasing expression trend in advanced ccRCC (T3&T4, M1, III&IV stages) (Fig. 1I L). The above results suggest that these key PRGs may served as candidate biomarkers. Later, the ORs and p-values from both analyses were combined, and ZDHHC18, a cancer-promoting gene, was selected as the research target.

ZDHHC18 expression in ccRCC cell lines

Analysis of the expression matrix of ccRCC cell lines from the CCLE database database revealed that, ZDHHC18 expression was upregulated in most ccRCC cell lines (Fig. 2A). Similarly, RT-qPCR and Western blotting assays showed that, ZDHHC18 expression remarkably increased in ccRCC cells relative to control cells (HK-2) (Fig. 2B, C). IHC results also demonstrated stronger staining intensity in ccRCC tissues relative to matched adjacent non-carcinoma tissues (Fig. 2D). The original Western blotting results are displayed in Figure S2.

Fig. 2.

Expression of ZDHHC18 in ccRCC. (A) CCLE database analysis results show the expression level of ZDHHC18 in different ccRCC cell lines; (B) RT-qPCR results of ZDHHC18 expression in different cell lines; (C) Western blotting results of ZDHHC18 expression in different cell lines; (D) IHC results (-: Negative; +: Low positive; ++: Positive; +++:High positive) (Scale bar: 100/20 µm)

Interaction gene screening results

Through Pearson correlation analysis, 2,157 candidate genes significantly associated with ZDHHC18 were obtained from the TCGA-KIRC cohort (Table S5). The top 10 genes positively and negatively correlated genes are visualized in Fig. 3A. Then, 10 genes interacting with ZDHHC18 were obtained through STRING analysis (Fig. 3B). These interacting genes were mainly enriched in pathways such as lipoprotein metabolic process (GO_BP), palmitoyl hydrolase activity (GO_MF), and Glycerophospholipid metabolism (KEGG), suggesting potential downstream mechanisms of ZDHHC18 (Fig. 3C). These interacting genes were further intersected with the candidate genes, yielding three highly correlated interacting genes (LYPLA2, FAM110A, and UPF1) (Fig. 3D). As discovered from correlation analysis, all three genes were significantly positively related to ZDHHC18 expression (p < 0.001), revealing their potential as downstream targets (Fig. 3E and G).

Fig. 3.

Screening of interacting genes of ZDHHC18. (A) Network diagram of the top 10 genes with positive/negative correlations; (B) PPI of ZDHHC18; (C) Functional enrichment analysis results of interacting genes; (D) Venn diagram of the intersection of correlation analysis results and PPI. (E) Correlation analysis results between ZDHHC18 and LYPLA2; (F) Correlation analysis results between ZDHHC18 and FAM110A; (G) Correlation analysis results between ZDHHC18 and UPF1

Molecular docking results and validation of ZDHHC18-mediated palmitoylation

GPS-Palm analysis revealed that all three interacting genes contained highly probable palmitoylation sites (scores > 0.9) (Table S6). Molecular docking was then conducted to evaluate the docking patterns of the three most likely modification sites with the DHHC domain of ZDHHC18 (Fig. 4A-C). All three genes showed good binding affinity, and the evaluation results are summarized in Table 2. Among them, LYPLA2 was considered the most likely target of ZDHHC18 due to its excellent HADDOCK score. Western blotting assays on transfected cell groups revealed that, compared with the negative control (NC) group (transfected with labeled LYPLA2 alone), the band in the ZDHHC18 wild-type (WT) group (co-transfected with labeled ZDHHC18) exhibited a slight upward shift. Conversely, the bands in the ZDHHC18 DHHS group and the 2-BP-treated ZDHHC18 WT group remained close to the baseline level of the NC group. No significant differences were observed in the total expression levels of LYPLA2 across any of the groups. These results provide protein-level confirmation that ZDHHC18 promotes the palmitoylation of LYPLA2 (Fig. 4D and E). The original Western blotting results are shown in Figure S3. To further evaluate the diagnostic potential of ZDHHC18 and its three interacting genes for ccRCC, ROC curves were generated for the TCGA-KIRC cohort (Fig. 4F-I). The results showed that regardless of which interacting gene was paired with ZDHHC18, the combined diagnostic efficacy was superior to that of ZDHHC18 alone (AUC: ZDHHC18, 0.875; ZDHHC18 + LYPLA2, 0.971; ZDHHC18 + FAM110A, 0.904; ZDHHC18 + UPF1, 0.967).

Fig. 4.

Multiplex analysis of interacting genes. (A) Binding pattern of ZDHHC18 domain and potential palmitoylation modification site of LYPLA2; (B) Binding pattern of ZDHHC18 domain and potential palmitoylation modification site of FAM110A; (C) Binding pattern of ZDHHC18 domain and potential palmitoylation modification site of UPF1; (D) Western blotting assays validate the ZDHHC18-mediated palmitoylation of LYPLA2; (E) ROC curve of ZDHHC18; (F) ROC curve of ZDHHC18 + LYPLA2; (G) ROC curve of ZDHHC18 + FAM110A; (H) ROC curve of ZDHHC18 + UPF1

Table 2.

Summary of multiple analyses resluts for the three target genes

ID	R	PPI Score	HADDOCK Score
LYPLA2	0.495	0.457	−103.2
FAM110A	0.487	0.479	−31.1
UPF1	0.544	0.408	−24.1

Single-cell sequencing and immune infiltration analysis results

In this study, GSE131685 and GSE171306 were used for paired analysis based on the sequencing performed of the included cells. First, the data were first evaluated and filtered, with assessment of gene counts, total number of molecules, and proportion of mitochondrial genes per cell. The correlation between pre-filtering and post-filtering data was examined, and both exhibited a good trend (Figure S4A, B). After batch normalization, PCA was carried out on those highly variable genes to determine the number of significant PCs (Figure S4C, D). Subsequently, different cell populations were projected onto a two-dimensional plane via dimensionality reduction and annotated (Figure S4E, Fig. 5A). To visualize the proportions of immune cell populations, the percentage of each population was calculated. Relative to control group, B cells and Monocytes showed decreased proportions, while Macrophages, NK cells, and CD8 + T cells had remarkably increased proportions (Fig. 5B). Simultaneously, the ZDHHC18 distribution in the two cell groups significantly increased in CD8 + T cells, NK cells, Macrophages, epithelial cells and B cells (Fig. 5C and D). To further refine the expression pattern of ZDHHC18 in various immune cell subtypes, the single-cell sequencing dataset GSE121636, which only contains various types of immune cells, was also analyzed. The results showed that ZDHHC18 was expressed at high levels in dendritic cells (DCs), Treg, CD8 Tex and Tprolif cells (Fig. 5E). Together, these results not only support the high expression of ZDHHC18 in ccRCC tumor tissues but also demonstrate its potential impact on immune cells and the tumor microenvironment (TME) in ccRCC.

Fig. 5.

Analysis of single-cell sequencing results. (A) Cell clustering diagram (GSE131685 and GSE171306); (B) Cell proportion distribution and cell distribution in different groups; (C) Distribution of ZDHHC18 cell expression in the control group; (D) Distribution of ZDHHC18 cell expression in the tumor group; (E) Expression of ZDHHC18 in immune cells of ccRCC tissues (GSE121636)

Immune infiltration analysis based on the TCGA database further supported the above findings (Fig. 6A). ZDHHC18 was significantly differentially expressed in these key immune cell types. Correlation analysis further demonstrated strong associations with DCs, Macrophages, CD8 + T cells, Treg cells, and NK cells (Fig. 6B-F). Immunofluorescence results showed significant co-localization of ZDHHC18 with the following immune cell markers: PD-1 (CD8 + T cells), FOXP3 (Treg), and CD11c (DCs) (Fig. 6G). Correlation analysis also confirmed that ZDHHC18 expression was positively correlated with various activation markers of these immune cells (Fig. 7A and D). More interestingly, we observed that multiple immune checkpoint marker genes were significantly upregulated in the ZDHHC18 high-expression group (Fig. 7E). Moreover, when comparing immunotherapy response scores among ccRCC patient groups, the high ZDHHC18 expression group showed remarkably higher scores than the low-expression group (Fig. 7F). Together, these results further support the notion that ZDHHC18 promotes the development of ccRCC by affecting the TME and suggest that ZDHHC18 may serve as a marker for immunotherapy response.

Fig. 6.

Correlation analysis of ZDHHC18 immune infiltration. (A) ZDHHC18 expression in different immune cells groups; (B) Correlation analysis between ZDHHC18 and DC; (C) Correlation analysis between ZDHHC18 and Tregs; (D) Correlation analysis between ZDHHC18 and Macrophages; (E) Correlation analysis between ZDHHC18 and CD8 + T cell; (F) Correlation analysis between ZDHHC18 and NK cells; (G) Immunofluorescence co-localization results of ZDHHC18 with markers of CD8 + T cells, Tregs, and Macrophages (Blue: DAPI; Red: ZDHHC18; Yellow: PD-1; Green: FOXP3; Cyan: CD11c) (Scale bar: 100/20 µm)

Fig. 7.

Correlation analysis of ZDHHC18 and immune cell markers. (A) Correlation analysis of ZDHHC18 and DC immune activation markers (CD80, CD86, CD40, CCR7); (B) Correlation analysis of ZDHHC18 and Tregs immune activation markers (FOXP3, CTLA4, ICOS, IL10); (C) Correlation analysis of ZDHHC18 and Macrophages immune activation markers (TNF, CXCL9, NOS2, CD86); (D) Correlation analysis of ZDHHC18 and CD8 + T cell immune activation markers (IFNG, GZMB, PRF1, CD69); (E) Expression levels of common immune checkpoint genes in different groups (G1: ZDHHC18 low-expression group; G2: ZDHHC18 high-expression group); (F) Statistical analysis of immune response between low and high-expression ZDHHC18

ZDHHC18 affects ccRCC outcomes by influencing immune cell traits

Having established that ZDHHC18 affects ccRCC progression by influencing the TME, we next explored its downstream mechanism of action. According to the two-sample MR analysis results, 44 immune cell traits causally associated with ccRCC were selected (Fig. 8A) (Table S7). Among these, 11 immune cell traits potentially affected by ZDHHC18 were further screened (Fig. 8B). The sensitivity analysis results are presented in Figure S5, S6, and the heterogeneity and pleiotropy results are displayed in Table S4. To clarify the mediating effects of these immune cell traits, mediation analysis was also carried out. The mechanism diagram is presented in Fig. 8C. Through mediation analysis, four key immune cell traits were identified: Secreting Treg AC, CD28 + CD45RA+ CD8dim %T cell, FSC-A on myeloid DC, and CD45RA on resting Treg (Mediation effect ratios: 5.17%, 6.09%, 12.26%, and 5.12%, respectively) (Table 3). Among these, two traits belonged to the Treg panel, while the others were from the T cell panel and DC panel.

Fig. 8.

ZDHHC18 downstream and upstream analysis. (A) Forest plot showing immune cell traits with causal relationship with ccRCC screened by two-sample MR analysis; (B) Forest plot showing immune cell traits with causal relationship with ZDHHC18 screened by two-sample MR analysis; (C) Schematic diagram of mediation analysis mechanism: a is the total effect size of ZDHHC18 to ccRCC; c is the effect size of ZDHHC18 to immune cell traits; d is the effect size of immune cell traits to ccRCC; c*d is the indirect effect size mediated by immune cell traits; b is the direct effect size (b = a-c*d). (D) Forest plot showing the causal relationship among mtDNA-cn, ZDHHC18 and ccRCC by two-sample MR analysis; (E) Schematic diagram of mediation analysis mechanism: a is the total effect size of mtDNA-cn to ccRCC; c is the effect size of mtDNA-cn to ZDHHC18; d is the effect size of ZDHHC18 to ccRCC; c*d is the indirect effect size mediated by ZDHHC18; b is the direct effect size (b = a-c*d)

Table 3.

Mediation analysis between ZDHHC18 and ccRCC mediated by immune cell traits

ID	a	b	c	d	c*d	P
ZDHHC18-GCST90001492-ccRCC	0.1108	0.1051	-0.1815	-0.0316	0.0057	5.17%
ZDHHC18-GCST90001664-ccRCC	0.1108	0.1041	0.2101	0.0321	0.0067	6.09%
ZDHHC18-GCST90001963-ccRCC	0.1108	0.0972	-0.2053	-0.0662	0.0136	12.26%
ZDHHC18-GCST90002102-ccRCC	0.1108	0.1051	0.1650	0.0344	0.0057	5.12%

ZDHHC18 expression and its effect on ccRCC are regulated by mtDNA-cn

To explore those upstream mechanisms of ZDHHC18, we focused on mtDNA-cn as a key modulator. Two-sample MR analysis revealed that mtDNA-cn was a significant risk factor for both ZDHHC18 expression and ccRCC (p < 0.05) (Fig. 8D). Sensitivity analysis results are presented in Figure S7. We then proposed a mechanistic model involving mtDNA-cn, ZDHHC18, and ccRCC (Fig. 8E). Mediation analysis suggested that the mediation effect ratio of ZDHHC18 was 11.08% (Table 4), further completing the upstream and downstream regulatory axis of ZDHHC18 in ccRCC. To validate the association between mtDNA-cn and ZDHHC18, we first performed a correlation analysis. As a result, ZDHHC18 was significantly co-expressed with key genes involved in mtDNA-cn (TFAM, TWNK, POLG, SSBP1, OPA1) (p < 0.05) (Fig. 9A and E). Furthermore, RT-qPCR and Western blotting assays showed that ZDHHC18 expression changed correspondingly following mtDNA depletion induced by treatment with both low and high doses of EtBr (Fig. 9F and G). Original Western blotting results are displayed in Figure S8. These findings further support our hypothesis.

Table 4.

Mediation analysis between mtDNA copy number and ccRCC mediated by ZDHHC18

ID	a	b	c	d	c*d	P
mtDNA-ZDHHC18-ccRCC	0.2201	0.1957	0.2201	0.1108	0.0244	11.08%

Fig. 9.

Vadilation of the correlation between ZDHHC18 and mtDNA-cn. (A) ZDHHC18 and TWNK; (B) ZDHHC18 and TFAM; (C) ZDHHC18 and POLG; (D) ZDHHC18 and SSBP1; (E) ZDHHC18 and OPA1; (F) RT-qPCR validate the impact of mtDNA levels on ZDHHC18 mRNA expression; (G) Western blotting validate the impact of mtDNA levels on ZDHHC18 protein expression. (EtBr Low: 50ng/mL; EtBr High: 100ng/mL)

Discussion

ccRCC is a prevalent and heterogeneous subtype of RCC characterized by prominent metabolic and immune dysregulation [1, 33]. Mitochondria act as central regulators of tumor metabolism and stress adaptation [34–37], and mtDNA-cn reflects mitochondrial functional status [38–40]. Our findings identified mtDNA-cn as an upstream factor influencing ZDHHC18, a palmitoyltransferase involved in signaling and membrane organization [41, 42]. Mediation analysis indicated that changes in mtDNA-cn affected ccRCC outcomes partly through ZDHHC18 (mediation effect ratios = 11.08%), suggesting that mitochondrial dysfunction contributes to oncogenesis via altered palmitoylation signaling [43–45]. This mechanistic link extends previous studies that focus mainly on the phenotypic or prognostic roles of ZDHHC18 [13, 46–48] and instead highlights its regulatory function in connecting mitochondrial activity with lipid-modification processes. With respect to modification targets, three interacting molecules—Lysophospholipase 2 (LYPLA2), Sequence similarity 110 family member A (FAM110A), and Up-frameshift protein 1 (UPF1)—were identified as potential downstream partners. These proteins have been validated to participate in lipid metabolism, cell cycle progression, and mRNA quality control, respectively [49–56]. Among them, LYPLA2 exhibited the strongest potential for palmitoylation, and this study experimentally verified that ZDHHC18 mediated the palmitoylation of LYPLA2.

Integrated multi-omics and single-cell analyses revealed that ZDHHC18 might coordinate metabolic and immune signaling pathways within the TME. These results are consistent with the view that palmitoylation can regulate protein interactions and cell signaling pathways, thereby affecting tumor behavior and immune evasion [57–59]. Regarding downstream mechanisms, ZDHHC18 expression was enriched in Tregs, CD8 + T cells, and DCs. Furthermore, four immune cell traits—Secreting Treg AC, CD28 + CD45RA+ CD8dim %T cell, FSC-A on myeloid DC, and CD45RA on resting Treg—emerged as potential mediators linking ZDHHC18 activity to patient prognosis in ccRCC. Among these, secretory Tregs have been shown to play a vital role in regulating immune responses and are causally associated with various diseases [60, 61]. The presence of CD28 + CD45RA+ CD8dim% T cells is associated with enhanced antitumor cell cytotoxicity, indicating their importance in antitumor immunity [62]. Furthermore, the analysis of FSC-A on myeloid DCs elucidated their role in antigen presentation and T cell activation, while the expression of CD45RA on resting Tregs highlights their regulatory function in maintaining immune homeostasis [63–65]. The positive correlation between ZDHHC18 expression and predicted immune response scores suggests that palmitoylation-driven lipid remodeling may stabilize immune receptors such as PD-1 and TCR, thereby enhancing immune cell activation and engagement. This phenomenon partly explains the associations observed in ZDHHC18 expression with multiple immune checkpoint genes. Consequently, elevated ZDHHC18 expression may promote a TME phenotype that is responsive to immunotherapeutic stimulation while facilitating immune adaptation through regulatory feedback mechanisms.

The present analytical framework—combining MR, SMR, bioinformatics, and single-cell sequencing —provides a comprehensive model linking mitochondrial regulation, palmitoylation, and immune modulation. From a translational perspective, the identified mtDNA-ZDHHC18-immune axis offers a conceptual foundation for biomarker development. Advances in ultra-sensitive molecular detection technologies now enable reliable capture of tumor-derived signals in circulation. Similar enrichment principles could eventually be adapted to quantify ZDHHC18-related genetic or epigenetic alterations for non-invasive disease monitoring, representing a promising future direction. However, several limitations should be noted. First, the MR analysis depended on blood-derived eQTL data, and could not completely reflect transcriptional modulation in renal tubular epithelium—the tissue of origin for ccRCC. Genetic variants influencing ZDHHC18 expression in blood may have different effects in kidney tissue; thus, future work using kidney-specific eQTL datasets, like those from the GTEx project, is needed. Our MR results therefore mainly reflect immune cell regulation of ZDHHC18, which is corroborated by its expression pattern in immune subtypes observed in the single-cell sequencing data. Second, transcriptomic data from large consortia such as TCGA can present batch effects and biological variability despite comprehensive normalization and cross-validation. It is necessary to consider these well-known limitations during the interpretation of multi-omics results.

Conclusion

In conclusion, this study elucidates the mechanism by which the PRG ZDHHC18 functions in ccRCC. ZDHHC18 is regulated by mtDNA-cn and affects the TME of ccRCC by modulating immune cell characteristics. Our results help better understand the effects of palmitoylation on ccRCC carcinogenesis and provide a theoretical basis and new insights for improving patient treatment and prognosis through targeted interventions.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization, JL, WT; Data curation: JL, HX; Formal analysis: JL, YB, and HC; Funding acquisition: WT; Methodology: JL, WT; Projection administration: WT; Software: JL, HX, and YB; Supervision: WT; Validation: JL, HX, and HC; Visualization: YB, HC; Writing-original draft: JL and HX; Writing-review and editing: All authors.

Funding

The research was supported by the Chongqing Key Projects for Technological Innovation and Application Development, China (cstc2019jscx-gksbx0007).

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

The study protocol was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University and conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.