Integrating Single-cell and Bulk RNA Sequencing Data Unveils Antigen Presentation and Process-related CAFs and Establishes a Predictive Signature in Prostate Cancer

Abstract

Abstract Background Cancer-associated fibroblasts (CAFs) are heterogeneous and can influence the progression of prostate cancer in multiple ways. However, their capacity to present and process antigens in PRAD has not been investigated. In this study, antigen presentation and process-related CAFs (APPCAFs) were identified using bioinformatics, and the clinical implications of APPCAFs-related signatures in PRAD were investigated. Methods SMART technology was used to sequence the transcriptome of primary CAFs isolated from patients undergoing different treatments. Differential expression genes (DEGs) screening was conducted. CD4 + T cell early activation assay was used to assess the activation degree of CD4 + T cell. The datasets of PRAD were obtained from The Cancer Genome Atlas (TCGA) database and NCBI Gene Expression Omnibus (GEO), and the list of antigen presentation and process-related genes was from the InnateDB database. Subsequently, APP-related CAFs were identified by non-negative matrix factorization (NMF) based on a single-cell seq (scRNA) matrix. GSVA functional enrichment analyses were performed to depict the biological functions. A risk signature based on APPCAF-related genes (APPCAFRS) was developed by least absolute shrinkage and selection operator (LASSO) regression analysis and the independence of the risk score as a prognostic factor was evaluated by univariate and multivariate Cox regression analyses. Furthermore, a biochemical recurrence-free survival (BCRFS) related nomogram was established, and immune-related characteristics were assessed using the ssGSEA function. The immune treatment response in PRAD was further analyzed by Tumor Immune Dysfunction and Exclusion (TIDE) tool. The expression levels of hub genes in APPCAFRS were verified in cell models. Results The functions and pathways of DEGs were significantly enriched in antigen processing and presentation processes, MHC class II protein complex and transport vesicle, MHC class II protein complex binding, and intestinal immune network for IgA production. Androgen withdrawal diminishes the activation effect of CAFs on T cells. NMF clustering of CAF was performed by APPRGs, and pseudotime analysis yielded the antigen presentation and process-related CAF subtype CTSK + MRC2 + CAF-C1. The CTSK + MRC2 + CAF-C1 exhibited ligand-receptor connections with epithelial cells and T cells. Additionally, we found a strong association between CTSK + MRC2 + CAF-C1 and inflammatory CAFs. Through the differential gene analysis of the CTSK + MRC2 + CAF-C1 and NoneAPP-CAF-C2 subgroups, 55 significant DEGs were identified, namely APPCAFRGs. Based on the expression profiles of APPCAFRGs, we divided the TCGA-PRAD cohort into two clusters using NMF consistent cluster analysis, with the genetic coefficient serving as the evaluation index. Four APPCAFRGs, including THBS2, DPT, COL5A1, and MARCKS were used to develop a prognostic signature that can predict BCR occurrence in PRAD patients. Subsequently, a nomogram with stability and accuracy in predicting BCR was constructed based on Gleason grade, PSA, T stage, and risk score. The analysis of immune infiltration showed a positive correlation between the abundance of resting CD4 + T memory cells, M1 macrophages, resting dendritic cells, and risk score. In addition, the mRNA expression levels of THBS2, DPT, COL5A1, and MARCKS in cell models were consistent with the results of bioinformatics analysis. Conclusions APPCAFRS based on four potential APPCAFRGs was developed and their interaction with the immune microenvironment may play a crucial role in the progression to castration resistance of PRAD. This novel approach provides valuable insights into the pathogenesis of PRAD and offers unexplored targets for future research.