Background Diabetic foot ulcers (DFUs) represent a severe chronic complication of diabetes and are characterized by persistent impairment of wound healing, accompanied by defective angiogenesis, chronic inflammation, and dysregulated extracellular matrix remodeling. Although impaired angiogenesis is widely recognized as a key pathological feature of DFUs, its associated molecular alterations have not been systematically characterized at the transcriptomic and cellular levels. Methods In this study, bulk transcriptomic data were analyzed in combination with machine learning–based gene prioritization and single-cell RNA sequencing to investigate molecular features associated with angiogenesis impairment in DFUs. Differential expression analysis was performed using the GSE199939 and GSE134431 datasets, followed by GO and KEGG enrichment analyses. Angiogenesis-related genes were retrieved from the MSigDB HALLMARKANGIOGENESIS and GO: 0001525, and intersected with the DEGs to generate a candidate gene set. A LASSO logistic regression model was then constructed in the discovery cohort and evaluated in a replication cohort, yielding a five-gene signature consisting of APLN, ENG, FN1, SERPINA5, and TIMP1. Single-cell transcriptomic data were subsequently used to examine the cellular expression patterns of these feature genes. Results Among the five feature genes, FN1 and TIMP1 showed relatively clear expression localization at the single-cell level. Single-cell RNA sequencing analysis revealed that FN1 was mainly enriched in fibroblasts and stromal-related cell populations, including pericytes/smooth muscle cells, whereas TIMP1 exhibited a multicellular expression pattern, with relatively high expression in fibroblasts, inflammatory myeloid cells, macrophages, and proliferating cells. In vivo experiments further showed that TIMP1 and EGFR mRNA expression levels were significantly decreased, whereas MMP9 mRNA expression was significantly increased in wound tissues from the model group. FN1 mRNA showed a downward trend, although the difference did not reach statistical significance. Conclusion This integrative bioinformatic analysis provides an exploratory characterization of molecular features potentially related to restricted angiogenesis and impaired repair in DFUs and suggests that TIMP1 may represent a more robust candidate linked to proteolysis-related dysregulation, whereas FN1 may more likely reflect stromal extracellular matrix remodeling.
Hu et al. (Wed,) studied this question.