Transcriptional and protein expression levels of PTGDS in pan cancer.
Through integration and analysis of data from TCGA, GTEx, CPTAC, and HPA databases, we acquired a comprehensive understanding of PTGDS (ENSG00000107317) expression across various cancers. Transcriptionally, PTGDS exhibited statistically significant underexpression in 21 common malignancies, including LUAD, KIRC, and LIHC (Fig. 1A). Upon expanding the sample size of normal tissues, PTGDS was consistently underexpressed in 30 tumors, with OV and PAAD showing notable exceptions with high expression levels (Fig. 1B). Proteomic data from CPTAC revealed low expression of PTGDS in lung adenocarcinoma and breast cancer, albeit without statistical significance in other tumor types (Fig. 1C). Immunohistochemical analysis using HPA revealed a consistently low expression pattern of PTGDS in digestive tract tumors, including gastric cancer, liver cancer, kidney cancer, and colorectal cancer (Figure 1D). Additionally, PTGDS was predominantly expressed at low levels in lung cancer (Figure 1E, Figure S1A)..
Pan cancer analysis of clinical parameter correlations of PTGDS
Survival analysis spanning four domains—overall survival (OS), disease specific survival (DSS), disease free survival (DFS), and progression free survival (PFS)—emphasized the prognostic relevance of PTGDS across multiple cancers. Cox regression modeling identified high PTGDS expression as a risk factor for OS in KIPAN, KIRC, and STAD patients, yet it conferred protective effects in GBMLGG, LUAD, LGG, CESC, HNSC, and DLBC cohorts (Fig. 2A). Kaplan Meier (KM) survival curves further illustrated PTGDS's diagnostic potential across these cancer types (Fig. 2B). Moreover, PTGDS expression significantly correlated with DSS in various cancers, including KIPAN, KIRC, STAD, COAD, GBMLGG, LUAD, LGG, PRAD, DLBC, CESCH, and HNSC (Fig. 2C). Notably, elevated PTGDS expression was associated with increased risk for DFS in KIRP and KIPAN, while demonstrating protective effects in CHOL and BLCA (Fig. S2A). In terms of PFS, PTGDS was identified as a risk factor for KIPAN, STAD, and KIRP, while exhibiting protective effects in GBMLGG, LUAD, PRAD, DLBC, HNSC, MESO, CHOL, and LGG (Fig. 2D). The expression pattern of PTGDS appears to be associated with poor prognosis in kidney cancer, but with good prognosis in lung and brain tumors. This differential expression warrants further investigation.
We further investigated the association between PTGDS and various clinical and pathological parameters across different cancer types, including T (Tumor), N (Node), M (Metastasis) stages, overall stage, and age. PTGDS demonstrated associations with 12 tumor types, including LUAD, BRCA, STES, and KIRP (Fig. S2B). However, lymph node metastasis was linked to PTGDS in only six tumor types (Fig. S2C). Additionally, PTGDS expression closely correlated with the stages of LUAD, STES, KIRP, KIPAN, STAD, KIRC, and BLCA (Fig. S2D).
Genomic Diversity and Correlation with Tumor Stemness of PTGDS in Pan Cancer
We further investigated the role of PTGDS in tumorigenesis by examining its genomic variation across diverse cancer types. Using comprehensive copy number variation (CNA) data from the cBioPortal database and the extensive sample collection from the "Pan-Cancer Analysis of Whole Genomes" project by TCGA, we analyzed 2683 samples from 2565 patients. Our analysis revealed that diploidy is the predominant mutation type of PTGDS in pan-cancer, closely followed by shallow deletion (Fig. 3A). Notably, PTGDS mutations were most frequent in hepatobiliary cancer, followed by pancreatic cancer (Fig. 3B). Additionally, we identified COAD, ESCA, UCEC, and ACC as cancer types with notably high PTGDS mutation rates (Fig. 3C).
We also explored the genomic heterogeneity of PTGDS in pan-cancer and its correlation with tumor stemness. Guided by the established associations of tumor mutational burden (TMB) and microsatellite instability (MSI) with tumor aggressiveness and prognosis 22-24. our correlation analysis revealed a significant positive correlation between PTGDS and TMB in KIPAN, while a notable negative correlation was observed in 14 tumor types, including GBM, LUAD, and STES (Fig. 3D). Similarly, PTGDS showed a positive correlation with MSI in GBMLGG, but a significant negative correlation in 10 tumor types such as ESCA and SARC (Fig. 3E).
Given that tumor growth often depends on cancer stem cells 25, we examined the tumor stemness index (DNAss) derived from mRNA expression and methylation signatures 26. Spearman correlation analysis revealed a significant positive association between PTGDS and tumor stemness in three tumor types: KIRP, THCA, and UVM. Conversely, a significant negative correlation was observed in 18 tumor types, including GBMLGG, LUAD, COAD, and BRCA (Fig. 3F).
PTGDS Association with Immune Infiltration and Checkpoints
We investigated the impact of PTGDS on the tumor microenvironment (TME) by examining its correlation with immune infiltration levels across various cancers (Fig. 4A-C). Analyzing PTGDS's associations with three immune scores, we observed consistent positive correlations. PTGDS was positively correlated with the StromalScore in 16 cancers, including STES, STAD, LUSC, THCA, and BLCA (Fig. 4A). Additionally, PTGDS expression in 15 cancers, such as LUAD, BRCA, STES, HNSC, and LUSC, significantly correlated with the ImmuneScore (Fig. 4B). Similarly, the ESTIMATEScore indicated a positive correlation between PTGDS expression and immune infiltration in CES, READ, BRCA, LUAD, and PAAD (Fig. 4C) (r > 0.3, p < 0.05). Further analysis of potential associations between PTGDS and 60 immune checkpoint genes revealed top correlations in THCA, GBMLGG, KIPAN, BRCA, and LUAD. PTGDS exhibited positive correlations with most immune checkpoint genes, except in GBMLGG (Fig. S3A). Moreover, we assessed PTGDS associations with 150 immune pathways and 44 RNA modification marker genes across different cancers. Notably, BRCA, KIPAN, LIHC, THCA, and LUAD emerged as the top five cancers closely related to immune regulatory genes. PTGDS displayed positive associations with immune regulatory genes in most cancers, except GBMLGG and LGG (Fig. S3B). These results suggest that PTGDS has a potential positive regulatory role in the TME.
PTGDS Interaction with Immune Cells
To evaluate the relationship between PTGDS expression and immune cell infiltration across various cancers, we utilized four established algorithms (TIMER, EPIC, QUANTISEQ, and Cibersort) to calculate a pan-cancer immune score. Our analysis revealed significant associations between PTGDS expression and immune cell infiltration in most cancers, particularly highlighting CD4+ T cells, macrophages, and B cells as closely linked to PTGDS expression levels (Fig. 4D-F, Fig. S3C). Additionally, data from the Human Protein Atlas (HPA) showcased PTGDS-specific expression in normal immune cells, notably in plasmacytoid dendritic cells and natural killer cells (Fig. S3D-F). Consistently, across all four algorithms, PTGDS demonstrated positive correlations with immune cell infiltration in malignant tumors such as LUAD, BRCA, and LUSC, contributing to the establishment of an immunosuppressive microenvironment (Fig. 4D-F, Fig. S3C). While slight variations were observed among the results of the four algorithms, PTGDS consistently exhibited robust positive or significant negative correlations with cancers such as STES, STAD, HNSC, COAD, and SKCM (Fig. 4D-F, Fig. S3C).
Single Cell Sequencing Reveals PTGDS Expression in Specific Cell Types
The findings suggest a potential specificity of PTGDS in lung cancer, prompting a focused investigation on LUAD. Utilizing single cell sequencing data from CancerSEA, we examined the correlation between PTGDS and functional status across 14 cancers. PTGDS showed predominantly negative correlations with DNA damage repair and cell cycle across most tumors, contrasting with positive associations with inflammation (Fig. 5A). Notably, in LUAD, PTGDS displayed a positive correlation with key biological behaviors, including angiogenesis and differentiation (Fig. 5A, B). Further analysis of the single cell atlas from the Human Protein Atlas (HPA) database unveiled predominant PTGDS expression in fibroblasts, endothelial cells, B cells, and T cells (Fig. 5C, D). Additionally, exploration of the normal human lung single cell atlas by Kyle et al., accessible via the UCSC Cell Browser, highlighted specific PTGDS expression in lipofibroblasts, adventitial fibroblasts, alveolar fibroblasts, natural killer cells, arteries, and veins (Fig. 5E, F).
Integrated Analysis of PTGDS Methylation and Functional Enrichment in LUAD
Methylation serves as a critical hallmark distinguishing tumor tissue from their normal counterparts, shaping the genetic landscape of cancer cells and impacting human longevity. Its potential in both tumor diagnosis and therapeutic interventions is substantial. Therefore, we embarked on a comprehensive investigation into the interplay between PTGDS methylation and tumor progression, focusing particularly on LUAD. While PTGDS displayed negative associations with immune regulatory genes in GBMLGG and LGG, it predominantly exhibited positive associations in most other cancer types. Notably, within LUAD, 27 RNA modifying genes, including DNMT3A, DNMT3B, and LRPPRC, were closely associated with PTGDS. Remarkably, the negative correlation of LRPPRC with PTGDS appeared consistent across various cancer types, indicating a robust relationship (Supplementary Fig. 4A).
Subsequently, leveraging data from the MethSurv platform, we identified eight methylation sites within the DNA sequence of PTGDS. Intriguingly, utilizing the MEXPRESS tool, we found that five methylation sites (cg18502630, cg02156769, cg13796381, cg13602921, cg13561390) were positively correlated with PTGDS expression levels (Supplementary Fig. 4B, C). Finally, our analysis unveiled higher PTGDS methylation levels in LUAD compared to normal tissues (Supplementary Fig. 4D).
To elucidate the underlying mechanisms, PTGDS underwent analysis using the STRING database. This analysis generated an interaction network between PTGDS and its related genes (Fig. 6A), subsequently subjected to KEGG and GO functional enrichment analyses to uncover potential pathways. These analyses revealed significant enrichment of PTGDS and its associated genes in arachidonic acid metabolism (Fig. 6B). Subsequent Gene Ontology Biological Process (GO, BP) terms highlighted pathways such as the cyclooxygenase pathway, prostaglandin biosynthetic process, and proteinoid biosynthetic process. Additionally, GO Cellular Component (GO, CC) terms suggested localization in the endoplasmic reticulum part and organelle sub compartment, with involvement in activities such as intramolecular oxidoreductase and isomerase (Fig. 6C-E).
Moreover, leveraging the TCGA LUAD cohort for Gene Set Enrichment Analysis (GSEA), we stratified the LUAD cohort into PTGDS high and low expression groups. Interestingly, the low expression group exhibited enrichment in pathways associated with autoimmune thyroid disease and hematopoietic cell lineage. Oncological features enriched in this group included Notch, KRAS, and VEGF pathways. Top Biological Process (BP) terms included adaptive immune response, regulation of cell activation, and B cell proliferation, while Cellular Component (CC) terms highlighted receptor complexes and plasma membrane protein complexes. Molecular Function (MF) terms mainly enriched in molecular transducer activity and immune receptor activity (Fig. 6F).
Potential Regulation of PTGDS by miR 3944 in LUAD
Investigating potential miRNAs regulating PTGDS, we analyzed LUAD miRNA data. Differential analysis between normal and tumor tissues revealed 364 differentially expressed miRNAs (DEMs), including 81 down regulated and 283 up regulated DEMs (Fig. 7A, B). Subsequently, we used the scan database to predict miRNAs targeting PTGDS and intersected the resulting DEM list, identifying 11 potential miRNAs (Fig. 7C). Survival analysis highlighted miR-3944, miR-296, and miR-542 for their diagnostic efficacy (Fig. 7D). Spearman's correlation analysis revealed miR-3944, miR-5090, miR-939, miR-6777, miR-3622, and miR-4758 associated with PTGDS, with miR-3944 showing diagnostic efficacy in LUAD and a negative correlation with PTGDS (Fig. 6E). These findings suggest a potential regulatory role of miR-3944 on PTGDS in LUAD.
Effects of PTGDS on Metabolism, Proliferation, Cell Cycle, and Apoptosis in A549 Cells
To delve into the biological function of PTGDS in LUAD, we employed the GSVA scoring algorithm to assess its correlation with various biological pathways. Our analysis revealed a positive association between PTGDS and several metabolism related pathways, including alpha linolenic acid metabolism, arachidonic acid metabolism, fatty acid degradation, glycerophospholipid metabolism, linoleic acid metabolism, and the citrate cycle (Fig. S5A). Additionally, PTGDS showed negative correlations with DNA repair, DNA replication, and the G2M checkpoint pathways, while positively correlating with apoptosis (Fig. S5B-D).
To investigate the impact of PTGDS overexpression, we transfected the GFP-PTGDS plasmid into A549 cells, which exhibit low PTGDS expression (Fig. 8A, Fig. S5E). This overexpression resulted in increased protein levels of PPARG, HADH, LDHA, and ACAT1, while levels of p-ACLY, ACC, and ACSL1 were decreased. These changes are consistent with the observed positive correlation with fatty acid degradation (Fig. 8B). Additionally, PTGDS overexpression influenced the expression of proteins related to glycolysis (Fig. 8C). Regarding cell cycle regulation, PTGDS overexpression inhibited the expression of CDC2 and CCND1, leading to an increase in cells in the G2 and S phases and a decrease in the G1 phase (Fig. 8D, E). The CCK-8 assay and colony formation experiments confirmed that this regulation was associated with reduced cell proliferation (Fig. 8F, G). Moreover, there was an increase in caspase-3 and caspase-9 proteins, although flow cytometry did not detect a change in the proportion of apoptotic cells (Fig. 8H, I).