Identification of compounds to promote diabetic wound healing based on transcriptome signature

Abstract

Purpose: Diabetic wounds are characterized by delayed healing, and the resulting diabetic foot ulcer may lead to severe complications, including amputations and mortality. This study aimed to identify potential small molecule drug candidates that can enhance diabetic wound healing through integrating transcriptome signature and experimental validation strategies.

Method: Gene expression dataset (GSE147890) from a diabetic skin humanized mice model in the Gene Expression Omnibus database was analyzed to identify differentially expressed genes between diabetic and normal skin, as well as the wound edge at 24 h. The DEGs were integrated with wound-related genes from the Comparative Toxicogenomics Database to construct a diabetes-specific wound gene profile. Then, the expression signatures were analyzed using the ClusterProfiler package in R for Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Hub genes were identified through the String database and Cytoscope software. The Connectivity Map (CMap) was employed to predict compounds with potential therapeutic effects on diabetic wound healing. These predications were validated through in vitro and in vivo experiments.

Results: A total of 167 DEGs were identified between diabetic and normal wounds, with significant enrichment in biological processes related to the extracellular matrix and collagen. The top ten hub genes were predominantly associated with collagen synthesis and inflammatory responses. CMap analysis identified 12 small-molecule compounds, top four of which were further investigated. In vitro experiments demonstrated that two compounds promoted fibroblast proliferation. In vivo studies revealed that compound CG-930 enhanced early inflammatory responses and upregulated the Nod-like receptor signaling pathway, significantly improving wound healing in streptozotocin (STZ) -induced diabetic mice.

Conclusion: This study highlights the altered expression profiles associated with delayed diabetic wound healing, including reduced inflammation and collagen production. Further drug screening identified compound CG-930 as a novel therapeutic agent with significant potential to promote wound healing in diabetic conditions.

Keywords: bioinformatics; diabetic wound; drug discovery; mechanism; transcriptome.

Created by Biorender.com.

FIGURE 1

Predicting the compounds that promoting diabetic wound healing based on bioinformatics. (A) Screening for DEGs in the GSE147890 dataset. (B) Volcano plot of DEGs in diabetic skin/wound compared with normal skin/wound. (C) Venn diagram of DEGs in the GSE147890 dataset and genes associated with DFU disease in CTD database. (D) The top eight up- and downregulated genes. (E) GO enrichment results of DEGs. (F) KEGG enrichment results of DEGs. (G) PPI interaction network by Cytoscape. (H) The top 10 hub genes in the PPI screened by CytoHubba.

FIGURE 2

Screening for small molecule compounds based on CMap database. (A) Small molecule compounds screened by CMap. (B) The structure of (1) CG-930, (2) Y-27632, (3) NVP-AUY922, and (4) PU-H71.

FIGURE 3

Evaluating the compounds that promoting proliferation of fibroblasts. (A) The effects of compounds on the proliferative activity of normal fibroblasts. (B) The effects of compounds on the proliferative activity of high glucose-induced fibroblasts. (C) The effects of compounds on the proliferative activity of high lipid-induced fibroblasts. Data presented are individual values with means ± SEM from n = 3-6 for each group. Statistical analysis tested by one-way ANOVA. ####p < 0.0001 vs the NC group. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 vs the model group.

FIGURE 4

Validating the compounds that promoting diabetic wound healing in the mice. (A) A schematic diagram of experimental process. (B) The effect of CG-930 on the body weight. (C) Fasting blood glucose on day 7. (D) Representative images of wounds. (E) Relative wound area on day 0 to day 7. (F) Wound healing rate on day 7. (G) HE stain and Masson stain on the day 7. (H) Relative collagen area analysed by Masson staining on day 7. Data presented are individual values with means ± SEM from n = 6 for each group. Statistical analysis tested by one-way ANOVA and two-way ANOVA (figure B and E). ##p < 0.01 and ####p < 0.0001 vs the NC group. *p < 0.05 vs the DC group.

FIGURE 5

Study on mechanism of CG-930 promoting wound healing in T1D mice. (A) Correlation heat map of mRNA expression in wounds of NC, DC, and CG-930 (1 mg/kg) treatment groups at 24 h. (B) Volcano plot and KEGG dotplot of DEGs. (C) Venn diagram of DEGs in NLR signaling pathway between DC vs NC and CG-930 treatment vs DC. (D) PPI network map of all genes in the NLR signaling pathway. (E) Representative bands of Western blot. (F) Immunoblot analyses of TNF-α, NLRP3, p-JNK/JNK, p-NF-κB/NF-κB, and COL1A1. Data presented are individual values with means ± SEM from n = 5-6 for each group. Statistical analysis tested by one-way ANOVA. ##p < 0.01, ###p < 0.001, and ####p < 0.0001 vs the NC group. *p < 0.05, ***p < 0.001, and ****p < 0.0001 vs the DC group.