Multi-Omics Pan-Cancer Bioinformatics Workflow for Evaluating PTDSS1 and PTDSS2 as Prognostic and Immune Biomarkers

Summary

This study conducted a comprehensive bioinformatics analysis using multiple public databases to evaluate the gene expression patterns, clinical correlations, survival prognosis, tumor stemness scores, and immune-related characteristics of PTDSS1 and PTDSS2 across various cancers. The findings suggest these genes may serve as potential biomarkers for cancer diagnosis, prognosis, and immunotherapy.

Abstract

The phosphatidylserine synthesis enzymes PTDSS1 and PTDSS2 play important roles in cellular processes, including membrane composition and signal transduction. Abnormal expressions of these enzymes has been associated with tumor-associated macrophage infiltration and poor survival outcomes in breast cancer. However, their comprehensive roles across different cancer types remain unclear. In this study, we performed a pan-cancer analysis to evaluate the diagnostic, prognostic, immune infiltration, and immunotherapeutic relevance of PTDSS1 and PTDSS2. Multiple public databases were used to analyze gene expression patterns, clinical correlations, survival outcomes, tumor stemness scores, and immune-related characteristics across various cancers. The results showed that PTDSS1 and PTDSS2 are widely expressed in human tissues and significantly upregulated in most tumor tissues compared with normal tissues. High expression of PTDSS1 or PTDSS2 was associated with poorer overall survival (OS) in several cancer types and showed significant correlations with clinical stage and tumor stemness scores. In addition, functional analyses suggested that these genes may contribute to tumorigenesis, immune regulation, and chemoresistance. Overall, the findings highlight the prognostic significance of PTDSS1 and PTDSS2 in multiple cancers and suggest that they may serve as potential biomarkers for cancer diagnosis and prognosis, as well as promising targets for immunotherapy.

Introduction

Although substantial advances have been achieved in oncology over the past several centuries, cancer remains one of the primary causes of death and disease burden worldwide^1,2. Current clinical management increasingly relies on multimodal treatment strategies, combining different therapeutic approaches to maximize efficacy while improving patient survival and overall well-being². In this context, cancer immunotherapy has rapidly evolved into a central pillar of modern oncology. Its advantages—including target selectivity, sustained therapeutic effects, applicability across multiple tumor types, resistance-mitigating capacity, and compatibility with other treatments—have driven widespread interest. Nevertheless, its clinical application remains constrained by several factors, including variable patient responsiveness, immune-related toxicities, acquired resistance, a high economic burden, and limitations in treatment monitoring. These challenges highlight the necessity for continued refinement and innovation in this field. At the same time, metabolic reprogramming has emerged as a hallmark of cancer, with lipid metabolism attracting increasing attention. Alterations in lipid acquisition, synthesis, and utilization provide essential support for tumor cell growth and survival and may also play a role in the development of therapeutic resistance^3,4. Reprogrammed lipid uptake and utilization support cancer cell proliferation and may contribute to therapeutic resistance⁵. In addition, lipid-related biomarkers have been associated with patient prognosis and treatment response^6,7. Despite increasing attention, the mechanisms of lipid metabolism in cancer and the development of effective lipid-targeted therapies remain insufficiently understood, warranting further investigation.

Phosphatidylserine synthases, PTDSS1 and PTDSS2, are key enzymes responsible for the biosynthesis of phosphatidylserine (PS)^8,9,10, one of the main anionic phospholipids in cell membranes¹¹, essential for maintaining membrane structure and function^8,12,13. These enzymes share moderate sequence similarity (~32%)¹³, and possess multiple transmembrane regions, mainly localized to the endoplasmic reticulum and mitochondria-associated membranes^14,15,16. PS is generated through a serine exchange reaction with existing phospholipids^8,9, with PTDSS1 preferentially utilizing phosphatidylcholine^17,18 and PTDSS2 using phosphatidylethanolamine^19,20. PTDSS1 is broadly expressed across tissues²¹, whereas PTDSS2 shows more restricted distribution, with higher levels in mouse brain neurons and testicular Sertoli cells¹⁹. Functionally, PTDSS2 is essential for normal testicular development, as about 10% of male knockout mice exhibit infertility and reduced testis size, a phenotype not observed in PTDSS1-deficient mice¹⁹. Although single-gene deletion of either enzyme does not impair viability, simultaneous loss leads to embryonic lethality^19,22, underscoring the importance of PS biosynthesis in cell survival. Recent evidence suggests that dysregulation of PTDSS1 and PTDSS2 in tumors is associated with altered tumor-associated macrophage infiltration and poorer survival in breast cancer patients²³. Collectively, these findings indicate that PTDSS1 and PTDSS2 are critical for membrane homeostasis, and their abnormal expression may contribute to disease progression, particularly in cancer.

This study systematically compares the pan-cancer effects of PTDSS1 and PTDSS2 on tumor expression, prognosis, stemness, immune regulation, genomic alterations, and treatment response. Unlike previous studies focusing on single cancer types or isolated lipid metabolism events, the present research integrates multi-omics data to explore potential links among phosphatidylserine biosynthesis, tumor immunity, and cancer progression. The analysis includes expression patterns, correlations with tumor characteristics and immune-related factors, genomic features, and drug sensitivity prediction. This study aims to deepen the understanding of the functional roles of PTDSS1 and PTDSS2 across different malignancies and to provide a theoretical foundation and practical guidance for the development of personalized treatment strategies.

Protocol

This study was based exclusively on publicly available datasets and did not involve direct human or animal participation; therefore, institutional ethical approval and informed consent were not required

1. PTDSS1 or PTDSS2 expression analysis

1. Assess PTDSS1 and PTDSS2 mRNA expression in normal human tissues using the Human Protein Atlas (HPA) database (https://www.proteinatlas.org).
2. Retrieve PTDSS1 and PTDSS2 gene expression profiles across multiple cancer types from the “Gene DE” module of the TIMER2 platform (http://timer.cistrome.org/).
3. Download RNA-seq expression data for normal and tumor tissues from the TCGA (http://cancergenome.nih.gov) and GTEx (http://commonfund.nih.gov/GTEx/) databases.

2. Pathological stage, survival, and tumor stemness analysis

1. Download the TCGA Pan-Cancer (PANCAN, N = 10,535, G = 60,499) dataset from the UCSC Xena platform (https://xenabrowser.net/).
2. Extract PTDSS1 and PTDSS2 expression data from all samples and exclude entries with zero expression values. Normalize the remaining expression data using the log2(x + 0.001) transformation.
3. Remove cancer types containing fewer than three samples, resulting in a final dataset comprising 37 tumor types. Perform differential expression analyses across clinical stages using R software.
4. Assess statistical significance between two groups using the unpaired Student’s t-test and among multiple groups using analysis of variance (ANOVA)²⁴.
5. Retrieve the TCGA Pan-Cancer dataset from the UCSC Xena platform (https://xenabrowser.net/). Extract PTDSS1 and PTDSS2 expression profiles and exclude samples with zero expression values.
6. Remove cases with follow-up durations shorter than 30 days. Normalize the remaining expression data using the log2 transformation with a pseudo-count of 0.001.
7. Exclude cancer types containing fewer than 10 samples, resulting in a final dataset comprising 39 cancer types with corresponding OS information. Perform survival analyses using the R package survival.
8. Construct Cox proportional hazards models using the coxph function to evaluate the associations between PTDSS1/PTDSS2 expression and patient prognosis across cancers. Assess survival differences using the log-rank test²⁴.
   NOTE: Due to the heterogeneity and incomplete availability of clinical annotation data across different tumor types in TCGA, additional clinical covariates (such as age, sex, and tumor stage) were not uniformly incorporated into the Cox regression analyses.
9. Obtain the TCGA Pan-Cancer dataset from the UCSC Xena platform (https://xenabrowser.net/). Extract PTDSS1 and PTDSS2 expression profiles and calculate tumor stemness indices based on DNA methylation features.
10. Integrate stemness scores with gene expression data for subsequent analyses. Exclude samples with zero expression values and normalize the remaining expression data using the log2 transformation.
11. Remove cancer types represented by fewer than three samples to improve analytical reliability. Perform Pearson correlation analyses to evaluate the association between PTDSS1/PTDSS2 expression and tumor stemness across different cancer types 24.
    NOTE: Samples with zero expression values were excluded to reduce the potential influence of technical noise and undefined values during log-transformation-based normalization. The same principle applies to subsequent steps.

3. Survival prognosis analysis

1. Analyze the associations between PTDSS1/PTDSS2 expression and tumor stage using the “Stage Plot” module in the Gene Expression Profiling Interactive Analysis (GEPIA) platform (http://gepia.cancer-pku.cn/).
2. Evaluate the prognostic significance of PTDSS1 and PTDSS2 across multiple cancer types using the KM Plotter database (https://kmplot.com/analysis/).
3. Stratify patients into high- and low-expression groups according to median expression values (cutoff-high = 50%, cutoff-low = 50%).
4. Generate OS significance maps and Kaplan–Meier survival curves for PTDSS1 and PTDSS2 across TCGA tumors using GEPIA2. Statistical significance is assessed using the log-rank test.

4. PTDSS1 or PTDSS2-related gene analysis

1. Obtain protein–protein interaction networks of PTDSS1 and PTDSS2 from the STRING database (https://cn.string-db.org/) by selecting experimentally validated interactions in Homo sapiens with the following parameters: minimum interaction score = 0.150 (low confidence), full network type, evidence-based edges, and a maximum number of interactors = 50.
2. Identify potential co-expressed genes using the “Similar Gene Detection” module in GEPIA2 (http://gepia2.cancer-pku.cn/#index), and select the top 100 related genes for further analysis.
3. Evaluate expression correlations between PTDSS1/PTDSS2 and the top 10 interacting proteins across tumor types using the “Gene_Corr” module in TIMER2.0 (http://timer.cistrome.org/), and visualize the results as heatmaps.
4. Retrieve Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotations through the KEGG REST API (https://www.kegg.jp/kegg/rest/keggapi.html) and use them as the reference background for functional enrichment analysis.
5. Perform gene set enrichment analysis using the R package clusterProfiler with gene set sizes ranging from 5 to 5000.
   NOTE: Statistical significance is defined as P < 0.05 and false discovery rate (FDR) < 0.25.
6. Conduct Gene Ontology (GO) enrichment analyses, including Biological Process (BP), Cellular Component (CC), and Molecular Function (MF), using the DAVID database. Visualize enriched pathways using the microbiome platform.

5. Genetic alteration analysis

1. Investigate the mutational profiles of PTDSS1 and PTDSS2 across multiple cancer types using the cBioPortal platform (http://www.cbioportal.org/) based on the TCGA Pan-Cancer Atlas Studies cohort.
2. Examine the frequency and alteration types of PTDSS1 and PTDSS2 using the “OncoPrint” and “Cancer Types Summary” modules. Use the “OncoPrint” module to visualize mutations, copy number variations, and gene expression alterations across samples.
3. Query PTDSS1 or PTDSS2 in the “Cancer Types Summary” and “Mutations” sections to obtain detailed information regarding alteration sites, mutation types, and their distribution among different cancer types.
4. Download the TCGA Pan-Cancer (PANCAN, N = 10,535, G = 60,499) dataset from the UCSC Xena database (https://xenabrowser.net/). Extract PTDSS1 and PTDSS2 expression data for each sample and obtain corresponding microsatellite instability (MSI) scores. Integrate gene expression and MSI data for downstream analyses.
5. Exclude samples with zero expression values and transform the remaining data using log2(x + 0.001) scale. Remove cancer types represented by fewer than three samples, resulting in a final dataset comprising expression and MSI data across 37 cancer types²⁴.
6. Obtain the TCGA Pan-Cancer dataset from the UCSC Xena database and extract PTDSS1 and PTDSS2 expression data for each sample. Download Level 4 Simple Nucleotide Variation data processed with MuTect2 from the GDC portal (https://portal.gdc.cancer.gov/).
7. Calculate tumor mutation burden (TMB) for each sample using the TMB function in the maftools R package and integrate TMB values with corresponding gene expression data.
8. Exclude samples with zero expression values and transform the remaining data using log2(x + 0.001) scale. Remove cancer types represented by fewer than three samples, resulting in a final dataset comprising 37 cancer types.

6. Immune regulatory genes, immune checkpoints, and RNA-modified genes analysis

1. Obtain the TCGA Pan-Cancer dataset from the UCSC database and extract PTDSS1 or PTDSS2 expression data together with 150 marker genes representing five immune-related pathways (chemokine: 41 genes; receptor: 18 genes; MHC: 21 genes; immunoinhibitor: 24 genes; immunostimulator: 46 genes) for each sample.
2. Exclude normal tissue samples and samples with zero expression values and apply a log2(x + 0.001) transformation to all expression data. Perform Pearson correlation analysis to assess associations between PTDSS1 or PTDSS2 and immune pathway marker genes.
3. Retrieve the TCGA Pan-Cancer dataset from the UCSC database and extract PTDSS1 or PTDSS2 expression data together with 60 genes involved in immune checkpoint pathways (inhibitory: 24 genes; stimulatory: 36 genes).
4. Remove normal samples and samples with zero expression values and transform all expression data using log2(x + 0.001). Conduct a Pearson correlation analysis to evaluate associations between PTDSS1 or PTDSS2 and immune checkpoint-related genes.
5. Download the TCGA Pan-Cancer dataset from the UCSC database and extract PTDSS1 or PTDSS2 expression data together with 44 genes associated with three types of RNA modifications (m1A: 10 genes; m5C: 13 genes; m6A: 21 genes) for each sample.
6. Exclude normal tissue samples and samples with zero expression values and apply a log2(x + 0.001) transformation to all expression data. Perform Pearson correlation analysis to evaluate the relationships between PTDSS1 or PTDSS2 and RNA modification-related genes.

7. Immune infiltration analysis

1. Download the TCGA Pan-Cancer dataset from the UCSC Xena database and extract PTDSS1 and PTDSS2 expression data for each sample.
2. Exclude normal samples and samples with zero expression values. Normalize expression data using log2(x + 0.001) transformation and map gene identifiers to GeneSymbol format.
3. Calculate StromalScore, ImmuneScore, and ESTIMATE Score for each patient using the ESTIMATE R package, resulting in immune infiltration scores for 9,554 tumor samples across 39 cancer types.
4. Perform Pearson correlation analysis using the “corr.test” function in the psych R package to evaluate associations between gene expression and immune infiltration scores.
5. Estimate immune cell infiltration scores for B cells, CD4⁺ T cells, CD8⁺ T cells, neutrophils, macrophages, and dendritic cells using the TIMER algorithm implemented in the IOBR R package for 9,405 samples across 36 tumor types.
6. Conduct Pearson correlation analysis to assess relationships between PTDSS1/PTDSS2 expression and immune cell infiltration levels.

8. Drug sensitivity of PTDSS1 and PTDSS2 in pan-cancer

1. Download NCI-60 compound activity data and RNA-seq expression profiles from CellMiner.
2. Select FDA-approved or clinical trial compounds for subsequent analysis.
3. Perform drug sensitivity analysis of PTDSS1 and PTDSS2 in pan-cancer using R software.

Results

PTDSS1 or PTDSS2 mRNA expression in various human normal tissues

Using data from the HPA database, the mRNA expression profiles of PTDSS1 and PTDSS2 across a range of human tissues were examined. Analysis revealed notable tissue-specific expression patterns. PTDSS1 was highly expressed in the parathyroid gland, heart muscle, lymph nodes, tonsils, and bone marrow (Figure 1A), whereas PTDSS2 showed elevated expression in the parathyroid gland, testi...

Discussion

Cancer is a systemic disease characterized by the dysregulation of multiple systems and typically progresses through three stages: elimination, equilibrium, and escape²⁵. In the elimination stage, the immune system effectively eliminates newly formed cancer cells; during the equilibrium stage, cancer cells and immune cells are in a relative balance, preventing tumor expansion or metastasis; ultimately, in the escape stage, cancer cells evade immune surveillance and rapidly proliferate and metastas...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
cBioPortal platform	cBioPortal	http://www.cbioportal.org/	Platform for visualization and analysis of multidimensional cancer genomics data
GDC portal	Genomic Data Commons (GDC)	https://portal.gdc.cancer.gov/	Portal for downloading TCGA genomic and mutation datasets
GEPIA platform	GEPIA	http://gepia.cancer-pku.cn/	Web server for gene expression profiling and survival analysis based on TCGA and GTEx datasets
GEPIA2	GEPIA2	http://gepia2.cancer-pku.cn/#index	Updated web server for gene expression correlation and interactive analysis
GTEx Databases	Genotype-Tissue Expression (GTEx) Project	http://commonfund.nih.gov/GTEx/	Database containing gene expression profiles from normal human tissues
Human Protein Atlas (HPA) database	Human Protein Atlas	https://www.proteinatlas.org	Database for analyzing protein and RNA expression profiles in normal tissues and cancers
KEGG REST API	Kyoto Encyclopedia of Genes and Genomes (KEGG)	https://www.kegg.jp/kegg/rest/keggapi.html	Application programming interface for retrieving KEGG pathway annotations
KM Plotter tool	KMplot	https://kmplot.com/analysis/	Online tool for evaluating the prognostic significance of genes in multiple cancers
maftools R package	Bioconductor package maftools	version 2.8.05	R package used for tumor mutation burden and mutation analysis
psych R package	CRAN R package psych	version 2.1.6	R package used for correlation and psychological statistical analyses
R package	R Foundation for Statistical Computing	version 4.1.3	Statistical computing software used for drug sensitivity analysis
R package clusterProfiler	Bioconductor package clusterProfiler	version 3.14.3	R package used for KEGG and functional enrichment analyses
R package survival	CRAN R package survival	version 3.2-7	R package used for Cox regression and Kaplan–Meier survival analysis
R software	R Foundation for Statistical Computing	version 3.6.4	Statistical computing software used for bioinformatics and survival analyses
STRING database	STRING	https://cn.string-db.org/	Database for protein–protein interaction network analysis
TCGA Databases	The Cancer Genome Atlas (TCGA)	http://cancergenome.nih.gov	Large-scale cancer genomics database containing genomic, transcriptomic, and clinical cancer data
TIMER2.0	TIMER2.0	http://timer.cistrome.org/	Platform for immune cell infiltration estimation and correlation analysis in tumors
Tumor Immune Estimation Resource 2.0 (TIMER2) platform	TIMER2.0	http://timer.cistrome.org/	Web platform for systematic analysis of immune infiltration across diverse cancer types
UCSC Xena platform	UCSC Xena	https://xenabrowser.net/	Platform for visualization and analysis of TCGA and other public multi-omics datasets