What question did this study set out to answer?

The research aims to uncover the role of fibroblasts and identify biomarkers related to ulcerative colitis (UC).

February 13, 2026Open Access

Combining Spatial Transcriptomics and Machine Learning Unravels Fibroblast-Related Biomarkers in Ulcerative Colitis

Key Points

The research aims to uncover the role of fibroblasts and identify biomarkers related to ulcerative colitis (UC).
Used single-cell RNA sequencing and cell-cell communication analysis to profile fibroblast interactions.
Applied weighted gene co-expression network analysis (WGCNA) to identify fibroblast-related gene modules.
Utilized machine learning models to predict diagnostic biomarkers and validate ANGPTL2 expression through various assays.
Conducted in vivo validation using a DSS-induced UC mouse model to assess ANGPTL2 expression and therapeutic response.
Identified fibroblasts as a key cell type in UC and revealed a strong correlation between fibroblast infiltration and the turquoise module in WGCNA.
Machine learning models indicated PHLDA1 and ANGPTL2 as high-accuracy diagnostic biomarkers (AUC 0.935 and 0.917).
ANGPTL2 was significantly upregulated in fibroblasts and serum from UC patients, supporting its potential as a non-invasive diagnostic indicator.
Spatial transcriptomics highlighted high ANGPTL2 expression and significant colocalization with fibroblasts across 13 distinct tissue regions.

Abstract

Ulcerative colitis (UC) is a chronic inflammatory bowel disease characterized by persistent mucosal inflammation. While fibroblasts contribute to tissue repair and fibrosis, their role in UC pathogenesis remains incompletely understood. Single-cell RNA sequencing (GSE114374, GSE231993) and cell-cell communication analysis were performed to profile fibroblast interactions. Weighted Gene Co-expression Network Analysis (WGCNA) identified fibroblast-related modules. Machine learning models predicted diagnostic biomarkers. IF, ELISA and WB assays were used to validate the expression of ANGPTL2 in UC samples. In vivo validation using a DSS-induced UC mouse model evaluated ANGPTL2 expression and therapeutic response to GDC-0152. Spatial transcriptomics (GSE189184) was used to analyze tissue regionalization, cell-type spatial distribution, and the localized expression of ANGPTL2 and its colocalization with fibroblasts. Single-cell analysis identified fibroblasts as a key cell type in UC. WGCNA revealed a strong correlation between the turquoise module and fibroblast infiltration, with genes in this module enriched in pathways such as cytokine-cytokine receptor interaction and TNF signaling. Machine learning models identified PHLDA1 and ANGPTL2 as potential biomarkers with high diagnostic accuracy (AUC 0.935 and 0.917, respectively). Furthermore, exprimental analyses confirmed that ANGPTL2 was significantly upregulated in fibroblasts from UC patients, serum ELISA experiments demonstrated that ANGPTL2 was also significantly elevated in the serum of UC patients, supporting its potential as a non-invasive diagnostic indicator. In vivo experiments confirmed ANGPTL2 upregulation in DSS-treated mice, reversed by GDC-0152. Spatial transcriptomics partitioned UC tissue into 13 distinct regions, revealing prominent fibroblast distribution and markedly high ANGPTL2 expression with significant spatial colocalization with fibroblasts. Fibroblasts are central to UC pathogenesis, contributing to inflammation and fibrosis. ANGPTL2 is a promising diagnostic biomarker and potential therapeutic target in UC.

Demander à l'IA

Bookmark

View Full Paper