The GENECARD database identifies genes associated with exercise, providing insights into their functions and relationships in various biological processes. Using the MetaboAnalyst platform, we conduct an integration study of these genes with metabolomics data.

We use DNA methylation and copy number variation (CNV) data from The Cancer Genome Atlas (TCGA) database for our pan-cancer study. After extracting CNV data for genes that overlap in different tumor types, the genes are classified as amplifications or deletions, and the rates of each are computed to find their frequencies in distinct tumor tissues. UALCAN (http://ualcan.path.uab.edu/analysis.html) is used to analyze the promoter DNA methylation levels of overlapping genes in both normal and malignant tissues. The “Gene Visualisation” feature of the MethSurv database provides DNA methylation patterns of various malignancies. Additionally, the R package “TCGAbiolinks” is used to retrieve mutation data in the Mutation Annotation Format (MAF), and the “maftools” R package is used to compute the TMB.

Using the “limma” R package, we perform differential expression analysis between tumor and normal samples from TCGA. P-value criteria and log2 fold change (log2fc) are used to identify genes with substantial variations in expression. Gene Set Enrichment Analysis (GSEA) was performed using the “clusterProfiler” package in R. This analysis involved formatting the data dimensions appropriately and visualizing the results with examples to highlight significant findings.

We assessed the predictive ability of overlapping gene expressions on patient survival outcomes using TCGA datasets, focusing on overall survival (OS). Survival analysis was conducted using the Kaplan-Meier method and log-rank test to evaluate survival across different cancer types. The “survival” and “survminer” R packages were utilized to generate survival curves. Additionally, the Cox proportional hazards model was employed to investigate the relationship between FGA and NOTCH3 expression and overall survival outcomes in pan-cancer patients, using the “forestplot” R package for visualization.

We employ markers from the CIBERSORTx website (https://cibersortx.stanford.edu/) and the main algorithm from CIBERSORT to assess the infiltration levels of 22 immune cell types in tumor samples. This analysis is facilitated by the CIBERSORT. R script. Heat maps are used to illustrate the findings of Spearman correlation analysis, which is used to investigate the link between single-gene data and immune infiltration matrix data across pan-cancer datasets. This is a specific technique used to evaluate immune cell populations, which helps to understand the role of tumor microenvironment in cancer progression.

Our methylation analysis specifically targets the TSS1500 region, which spans from −200 to −1,500 base pairs upstream of the transcription start site. We also focus on the TSS200 region, which is located within −200 base pairs upstream of the TSS. Additionally, we examine the first exon and the 5′untranslated region (5′UTR). Each sample’s methylation level is represented by the median value. In order to investigate the association between methylation levels and gene expression, we conduct a Spearman’s correlation analysis. The Spearman rank correlation coefficient, a non-parametric statistic, is used to assess the monotonic association between two variables, regardless of their distribution. The independent variable in this context is the methylation levels, while the dependent variable is the gene expression levels. The correlation between these two variables is evaluated by computing the Spearman rank correlation coefficient. In addition, the Wilcoxon rank-sum test is used to analyse the distribution of promoter methylation data between the tumor and normal groups.

We conduct Gene Ontology (GO) enrichment analysis on the chosen gene sets utilising the “clusterProfiler” tool. Furthermore, the Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analysis is performed using the identical software package. In order to evaluate the activity of a pathway, we utilise the GSVA software and analyse the data using four different methods: zscore, gsva, ssgsea, and plage. With the exception of the zscore parameter, the scores obtained from the other techniques are transformed into unitless Z-scores using the formula (x-μ)/σ. This is done to maintain uniformity among tumor data. The Wilcoxon Rank Sum Test is used to assess statistical disparities between tumor and normal tissues. The findings are visualised by creating box plots using the ggplot2 program.

The RAW 264.7 cell line is acquired from the Shanghai Cell Bank, which is a division of the Chinese Academy of Sciences. The cells are cultivated in RPMI-1640 medium (Gibco Invitrogen Co., San Diego, CA, United States) with the addition of 10% foetal bovine serum (Gibco BRL, Grand Island, NY, United States), 100 units/mL penicillin, and 0.1 mg/mL streptomycin. The culture conditions are upheld at a temperature of 37°C, with a humidity level of 95% and a CO₂ concentration of 5%. Docosahexaenoic acid (DHA) is dissolved in dimethyl sulfoxide (DMSO) to form a 1 millimolar (mM) concentrated solution, which is then kept at a temperature of −20°C. During the experiment, the stock solution is mixed with DMSO to provide a final concentration of 2 μM DHA and EPA for the RAW 264.7 cells. These cells are then subjected to a 48-h treatment. The DHA (D2534, purity >99%) and a standard for 37 fatty acids are obtained from Sigma (St. Louis, MO, United States), whereas eicosapentaenoic acid (EPA) is acquired from Sigma-Aldrich (St. Louis, MO, United States).

The following cell lines—LNCaP (prostate cancer), HepG2 (liver cancer), A549 (lung adenocarcinoma), and HCT116 (colorectal cancer)—were cultured under standard conditions. Each cell line was maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂. The culture medium was replaced every 2–3 days, and cells were subcultured upon reaching 80%–90% confluence using 0.25% trypsin-EDTA for detachment. All cell lines were authenticated prior to experiments to ensure reliability and reproducibility of the results. For the colony formation assay, cells were seeded into 6-well plates at a density of 500–1,000 cells per well and allowed to adhere overnight. The cells were then cultured in DMEM supplemented with 10% FBS at 37°C in a humidified incubator with 5% CO₂. The medium was refreshed every 3–4 days. After 10–14 days, when visible colonies had formed, the medium was removed, and the cells were gently washed twice with phosphate-buffered saline (PBS). The colonies were fixed with 4% paraformaldehyde for 15 min at room temperature and subsequently stained with 0.5% crystal violet solution for 20 min. After washing away the residual stain with tap water, the plate was air-dried. After 7–10 days, colonies containing more than 50 cells were counted under a microscope, and the colony formation efficiency was calculated as the number of colonies divided by the number of cells seeded.

The RAW 264.7 cells were diluted to a concentration of 250,000 cells/mL. Luminol, a light-enhancing compound, and zymosan, which stimulates the production of ROS, were added to each well. The resulting chemiluminescent emission from the formation of ROS was then measured.

Immunofluorescence detection was performed as previously described (Im et al., 2019; Chen et al., 2024a). Cells were initially fixed with a 4% paraformaldehyde (PFA) solution for 15 min, followed by blocking for 1 h. The samples were then incubated overnight with the primary antibody at 4°C. On the second day, the samples were brought to room temperature (25°C) and stained with a fluorescently labeled secondary antibody for 1 h. After washing with PBS, a coverslip was mounted using an anti-fade mounting solution containing DAPI from Abcam. Fluorescence microscopy was then used to capture images.

A P-value of <0.05 was considered statistically significant, ensuring the reliability and accuracy of our findings regarding metabolites and pulmonary hypertension. Mean ± standard errors are reported for the data.

To address this inquiry, we utilized a two-sample MR methodology to evaluate the potential causal relationship between metabolites and the risk of PH. The intercept test conducted using MR Egger indicated no signs of horizontal pleiotropy (p > 0.05 for all instrumental variables), thereby reinforcing the reliability of our results. Additionally, the outcomes of the leave-one-out sensitivity analysis, employing a jackknife approach, are presented in Figure 1. Our analysis identified 57 metabolites with a significant causal association with the development of PH, using a significance threshold of 0.01 (Supplementary Figure S1). Among these, metabolites such as Glucuronide of piperine (P = 0.030, OR = 1.316, 95% CI = 1.026–1.689), N-lactoyl valine (P = 0.034, OR = 2.029, 95% CI = 1.052–3.911), and N-stearoyl-sphingadienine (d18:2/18:0) (P = 0.007, OR = 1.570, 95% CI = 1.126–2.189) were found to be positively associated with an increased risk of PH. Conversely, metabolites such as Cerotoylcarnitine (C26) (P = 0.027, OR = 0.688, 95% CI = 0.493–0.959), Docosatrienoate (22:3n6) (P = 0.041, OR = 0.642, 95% CI = 0.420–0.982), and 5-dodecenoylcarnitine (C12:1) (P = 0.036, OR = 0.545, 95% CI = 0.309–0.962) exhibited a protective effect against PH, showing negative correlations. Further reverse MR analysis assessed the impact of PH on these metabolites. The results suggested a negative correlation between PH and S-methylcysteine levels (P = 0.045, OR = 1.014, 95% CI = 1.000–1.028). The MR Forest Plot (Supplementary Figure S1) visually represents the odds ratios (ORs) and 95% confidence intervals (CIs) for each metabolite, with data analyzed across different MR techniques. The metabolites examined include a broad spectrum of biological compounds, such as amino acids, lipids, and carbohydrates. In the forest plot, horizontal lines representing the ORs and CIs indicate significant associations when they do not cross the null value (OR = 1). To ensure robust causal inference, we employed multiple MR methodologies, including Inverse-Variance Weighted (IVW), MR Egger, weighted median, weighted mode, and simple mode approaches, using genetic predictors derived from genome-wide association studies (GWAS). The IVW method combined the effect estimates of genetic instruments, MR Egger addressed directional pleiotropy, and the weighted median provided reliable estimates even when up to 50% of the data came from invalid instruments. These diverse approaches strengthen the robustness and validity of our findings. This comprehensive analysis underscores significant causal relationships between specific metabolites and the risk of PH, identifying potential therapeutic targets and biomarkers for disease prevention and management.

The analysis of metabolic pathways and exercise-related genes in the context of PH provided significant insights into potential therapeutic targets. Figure 2A depicts the top 25 enriched metabolic pathways identified from metabolomics data of PH patients, analyzed using the MetaboAnalyst 6.0 platform. The pathways are ranked by p-value (left panel) and enrichment ratio (right panel). Pathways such as D-Arginine and D-Ornithine metabolism, starch and sucrose metabolism, and galactose metabolism exhibited significant associations with PH. The size and color of the dots represent the enrichment ratio and statistical significance, respectively, with higher ratios and more significant p-values highlighted in red. These findings suggest that disturbances in amino acid and carbohydrate metabolism play crucial roles in the pathophysiology of PH, potentially offering novel avenues for metabolic intervention. In Figure 2B, a network of exercise-related genes and their associations with various diseases was constructed using data from the GeneCards database. The network includes genes such as IL6, IL1B, COL1A1, COL5A1, IL10, IFNG, PRL, MPO, ACE, CCL2, REN, and F2, which are linked to a wide array of diseases, including inflammatory bowel disease, diabetes mellitus, rheumatoid arthritis, and Parkinson’s disease. These genes are known to play critical roles in inflammation, immune regulation, and metabolic processes. The network analysis underscores the relevance of these genes in the physiological response to exercise and their potential role in modulating the immune and metabolic dysregulation observed in PH. The identification of key metabolic pathways and gene-disease associations provides a foundation for future research aimed at developing exercise-based therapeutic strategies, which could modulate these metabolic and immune processes to improve outcomes for PH patients.

This study investigates the prognostic relevance of 12 overlapping genes (ACE, CCL2, COL1A1, COL5A1, F2, IFNG, IL1B, IL6, IL10, MPO, PRL, and REN) across different cancer types by analyzing their correlation with overall survival (OS). The hazard ratios (HRs) and 95% confidence intervals (CIs) for each gene’s expression were calculated, revealing important trends in cancer prognosis. As illustrated in Figure 3A, ACE expression is protective against kidney renal clear cell carcinoma (KIRC) and mesothelioma (MESO) but may present risks in uterine carcinosarcoma (UCS). Elevated ACE levels confer significant advantages for certain cancers while being detrimental to others. Regarding CCL2 (Figure 3B), increased expression correlates with higher risk in kidney renal papillary cell carcinoma (KIRP) and low-grade glioma (LGG), indicating its role as a negative prognostic marker for these malignancies. COL1A1 expression is associated with risk in various cancers (Figure 3C), including breast invasive carcinoma (BRCA), head and neck squamous cell carcinoma (HNSC), and stomach adenocarcinoma (STAD), whereas it provides protective effects in kidney chromophobe (KICH), highlighting its complex role depending on cancer type. For COL5A1 (Figure 3D), high expression is linked to increased risk in liver hepatocellular carcinoma (LIHC) and lung adenocarcinoma (LUAD) but offers protective benefits for KIRC, demonstrating its dual nature in cancer prognosis. The F2 gene (Figure 3E) predominantly exhibits protective effects across various cancers, although it poses risks in KIRC and bladder urothelial carcinoma (BLCA). Similarly, cancers like sarcoma (SARC) and LGG present adverse outcomes, reflecting the gene’s varied impact on different tumor types. In terms of IFNG (Figure 3F), sensitivity to expression levels is evident, as it shows protective trends in esophageal carcinoma (ESCA) and KIRC, yet poses risks in BRCA. A similar pattern is observed with IL1B (Figure 3G), which has crucial protective roles in KIRC and BRCA while being associated with significantly higher risks for stomach adenocarcinoma (STAD) and skin cutaneous melanoma (SKCM), despite its complex involvement in inflammatory pathways within tumors. For IL6 (Figure 3H), high expression indicates a poor prognosis in lung squamous cell carcinoma (LUSC) and BLCA, while demonstrating protective effects for KIRC, illustrating the dual aspects of cancer progression. IL10 (Figure 3I) correlates with increased risk in KIRP and pancreatic adenocarcinoma (PAAD), yet displays protective tendencies in LUSC, emphasizing its variable role across different tumor microenvironments. MPO (Figure 3J) becomes concerning, as elevated expression is associated with LIHC and LUSC, while providing protective benefits in BLCA, indicating its differing impacts based on tumor type. Lastly, PRL (Figure 3K) shows high expression linked to risks in colon adenocarcinoma (COAD) and BRCA, yet has favorable effects for KIRC, showcasing the diverse roles this gene plays in tumor biology. Finally, REN (Figure 3L) exhibited protective trends in cancers such as KIRC and kidney renal papillary cell carcinoma (KIRP), while being associated with increased risk in LGG and UCS, highlighting its potential as a prognostic marker.

Next, we explored CNV, methylation, and TMB analysis of pan cancer overlapping genes to clarify their functional characteristics. Figure 4A shows the distribution of CNV rates among the overlapping genes across 20 different cancer types. Each bar in the graph represents the variation in CNV for individual genes across these cancers, with distinct colors indicating different cancer types. For instance, COL1A1 and COL5A1 showed significant CNV gains in cancers like BRCA and STAD, while ACE and IFNG exhibited frequent CNV losses in KIRC and LUAD. Figure 4B presents the differential expression analysis of these overlapping genes across various tumors. Notably, IL6 and IL10 were significantly upregulated in cancers such as PAAD and LIHC, suggesting their potential roles in tumor progression and inflammation. Figure 4C explores the correlation between CNV and gene expression in various malignancies. This figure reveals that genes like COL1A1 and COL5A1 display strong positive correlations between CNV and expression levels in cancers such as BRCA and STAD, indicating that CNV gains directly enhance the expression of these genes. Conversely, negative correlations were observed for IFNG in LUAD, suggesting that CNV losses are associated with reduced gene expression in certain contexts. Figure 4D delves into the relationship between promoter methylation and gene expression across different cancers. The data show that hypermethylation of genes like ACE and F2 is associated with decreased expression in tumors such as LUAD and KIRC, underscoring the role of epigenetic silencing in these cancers. On the other hand, IL1B and IL6 exhibited hypomethylation coupled with increased expression in cancers like colorectal adenocarcinoma (COAD) and BRCA, indicating that promoter demethylation may activate oncogenic pathways in these tumors. Figure 4E highlights the association between TMB and gene expression across various cancer types. This analysis reveals that genes such as IL6 and MPO show significant positive correlations with TMB in high-mutation cancers like LUSC and esophageal carcinoma (ESCA), suggesting that higher mutation loads might drive the overexpression of these inflammatory mediators. Conversely, genes like REN and PRL show a negative correlation with TMB in cancers like KIRC, implying that these genes might be suppressed in tumors with a high mutation burden. Figure 4F illustrates the δ-values of promoter methylation in tumors versus normal tissues for the overlapping genes across different cancers. The data indicate that COL5A1 and MPO exhibit significant hypermethylation in tumor tissues compared to normal tissues in cancers like STAD and BRCA, which is associated with their reduced expression. Conversely, IL10 and IFNG exhibit hypomethylation in pancreatic adenocarcinoma (PAAD) and lung squamous cell carcinoma (LUSC) tumors, which correlates positively with elevated mRNA levels. This suggests their potential role in fostering a pro-tumorigenic microenvironment. The GSEA was performed on a diverse set of cancer types, including cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), acute myeloid leukemia (LAML), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), ovarian serous cystadenocarcinoma (OV), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), skin cutaneous melanoma (SKCM), stomach adenocarcinoma (STAD), testicular germ cell tumors (TGCT), and uveal melanoma (UVM) (Supplementary Figure S2). A comprehensive analysis of thousands of pathways was conducted, encompassing critical cellular processes such as the canonical xenobiotic metabolism system, WNT beta-catenin signaling, unfolded protein response, and epithelial to mesenchymal transition (EMT), among others.

In this study, we conducted a comprehensive pan-cancer analysis of the relationship between the expression levels of key overlapping genes across various cancer types and immune cell infiltration, TMB, and MAF, highlighting the crucial roles of these overlapping genes. Using heatmaps, radial plots, and MAF heatmaps generated by CIBERSORT, we systematically evaluated these relationships. The heatmaps presented in Figures 5A–J show significant correlations between the expression levels of key genes (ACE, CCL2, COL1A1, COL5A1, F2, IFNG, IL1B, IL6, IL10, MPO, PRL, and REN) and immune cell infiltration levels, including B cells, CD8⁺ T cells, regulatory T cells (Tregs), natural killer (NK) cells, macrophages, immature dendritic cells (iDC), activated dendritic cells (aDC), and mast cells. Our findings provide strong evidence of the association between gene expression and immune cell infiltration from a pan-cancer perspective. To illustrate the expression patterns of these TMB-associated overlapping genes across cancer types, radial plot analysis revealed a bimodal distribution of TMB-related gene expression levels, suggesting their potential role as regulatory factors in cancer mutation processes (Figure 5K). Additionally, the heatmap in Figure 5L shows the MAF of these overlapping genes across different cancer types, with observed MAF differences potentially reflecting the heterogeneity of genetic alterations and selective pressures encountered in tumorigenesis. Overall, this study reveals critical insights into the genetic and immunological characteristics of cancer, suggests potential biomarkers for cancer detection and prognosis, and highlights molecular mechanisms that may drive tumor progression.

An integrated analysis of promoter methylation and its correlation with gene expression, alongside pathway activity and enrichment, can yield valuable insights into tumorigenesis across different cancers. Heatmaps presented in Figure 6A illustrate the varying levels of promoter methylation among multiple cancer types, with red indicating significantly higher methylation and blue indicating substantially lower methylation when comparing tumor tissues to normal tissues. In Figure 6B, we plot promoter methylation levels against mRNA expression, where red represents a positive correlation and blue indicates a negative correlation, highlighting the complex interplay between methylation and gene expression. Subsequent boxplots (Figures 6C–F) present pathway activity Z-scores in tumor tissues versus normal tissues across different cancers, with the Wilcoxon Rank Sum Test employed to assess statistical significance. These plots reveal significant differences in pathway activity between tumor and normal tissues, with distinct patterns observed in specific cancer types. Figure 6G offers a scatter plot of KEGG pathway enrichment analysis, where the color gradient from yellow to blue represents the level of significance, and the size of each dot reflects the proportion of enriched genes within each pathway. This analysis identifies critical pathways that are highly enriched in tumors, providing potential targets for therapeutic intervention. Furthermore, Figure 6H displays a radial plot of GO enrichment analysis.

This study evaluated the antioxidative and anti-inflammatory effects of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) on LPS-stimulated RAW 264.7 macrophages and various cancer cell lines, demonstrating a marked reduction in oxidative stress, inflammatory cytokine expression, and cancer cell proliferation. Flow cytometry analysis revealed that both DHA and EPA significantly reduced ROS levels in LPS-stimulated RAW 264.7 cells (Figure 7A). Compared to the LPS-only group, cells treated with DHA and EPA exhibited notably lower ROS levels. Immunofluorescence staining was employed to assess the expression of IL-6 and IL-1β in RAW 264.7 cells. In the DHA + LPS and EPA + LPS treatment groups, IL-6 expression was significantly reduced compared to the LPS-only group (Figure 7B). Similarly, DHA and EPA treatments decreased IL-1β expression, a key pro-inflammatory cytokine, as indicated by reduced red fluorescence (Figure 7C). These findings suggest that DHA and EPA effectively suppress LPS-induced inflammatory responses at the cellular level. The effects of DHA and EPA on cancer cell proliferation were evaluated using colony formation assays in various cancer cell lines (LNCaP, HCT116, HepG2, and A549). Compared to the untreated control group, cells treated with DHA and EPA showed a marked reduction in colony formation, indicating a significant inhibitory effect on cancer cell proliferation (Figure 7D). This finding implies that DHA and EPA may possess anticancer properties by inhibiting cell growth. Immunofluorescence analysis also examined the nuclear translocation of SUMO1, a protein involved in stress response pathways, under inflammatory conditions. In the LPS-stimulated group, there was a notable increase in SUMO1 nuclear localization (Figure 7E). Treatment with DHA and EPA resulted in reduced SUMO1 nuclear translocation, suggesting that these fatty acids may modulate stress-related post-translational modifications, further enhancing their anti-inflammatory effects. In conclusion, these findings indicate that DHA and EPA exert antioxidative and anti-inflammatory effects in LPS-stimulated RAW 264.7 cells by reducing ROS production and downregulating pro-inflammatory cytokine expression. Their ability to inhibit cancer cell proliferation and regulate SUMO1-associated nuclear translocation highlights their potential therapeutic applications in managing inflammation and cancer.