Shared pathogenic mechanisms linking obesity and idiopathic pulmonary fibrosis revealed by bioinformatics and in vivo validation

Abstract

Previous studies have suggested a potential correlation between obesity and idiopathic pulmonary fibrosis (IPF). This study aimed to elucidate pathogenic pathways connecting obesity and IPF and identify diagnostic biomarkers for obesity-related pulmonary fibrosis. Obesity and IPF datasets were obtained through the Gene Expression Omnibus (GEO) database. Differential expression analysis and weighted gene co-expression network analysis (WGCNA) were used to identify shared genes for obesity and IPF. Functional enrichment (GO/KEGG), protein-protein interaction (PPI) networks, and machine learning algorithms were applied to screen hub genes, validated by ROC curves. High-fat diet (HFD)-induced obese mice with bleomycin-induced pulmonary fibrosis underwent histological assessment and qRT-PCR validation. Molecular docking evaluated flavonoid binding to hub genes. We identified 128 shared genes between obesity and IPF, predominantly enriched in immune and inflammatory pathways. Machine learning prioritized three hub genes (NLRC4, SPI1, and NCF2), validated by ROC analysis (AUC > 0.7). In animal model, these genes exhibited significant upregulation, correlating with exacerbated fibrosis. Molecular docking highlighted strong binding affinities (-6.3 to -9.6 kcal/mol) between dietary flavonoids and hub targets. Immune-inflammatory dysregulation links obesity and IPF via NLRC4, SPI1, and NCF2. These genes serve as diagnostic biomarkers and therapeutic targets, with flavonoids showing intervention potential. Our findings advance mechanistic insights into obesity-related pulmonary fibrosis.

Keywords: Bioinformatic analysis; Flavonoids; Idiopathic pulmonary fibrosis; Machine learning; Obesity.

Fig. 1

The workflow chart of this study. (A) Flow chart of research design. (B) Schematic of mouse experiment. (C) Flow chart for identifying shared genes.

Fig. 2

Effects of HFD-induced obesity on BLM-triggered pulmonary fibrosis. (A) Masson staining of lung tissues (blue: collagen deposition). (B) Collagen volume fraction of lung tissues. (C) Ashcroft scores of lung tissues. Data expressed as mean ± SE, ***, p < 0.001, **, p < 0.01.

Fig. 3

Identification of differentially expressed genes (DEGs). (A) Heatmap presenting the top 30 obesity DEGs in GSE151839. (B) Volcano plot representing obesity DEGs in GSE151839. (C) Heatmap presenting the top 30 IPF DEGs in GSE28042. (D) Volcano plot representing IPF DEGs in GSE28042. (E) Venn diagram illustrating 5 overlapping DEGs between obesity and IPF.

Fig. 4

WGCNA analysis for screening key module genes for obesity and IPF. (A) Determination of the optimal soft thresholds for GSE151839. (B) Determination of the optimal soft thresholds for GSE28042. (C) Clustering dendrograms of genes in GSE151839, with different colors representing different modules. (D) Clustering dendrograms of genes in GSE28042, with different colors representing different modules. (E) Heatmap of the correlation between modules and obesity. Red color represents a positive correlation and blue color represents a negative correlation. (F) Heatmap of the correlation between modules and IPF. Red color represents a positive correlation and blue color represents a negative correlation. (G) Venn diagram illustrating 110 overlapping genes within the positively correlated modules. (H) Venn diagram illustrating 13 overlapping genes within the negatively correlated modules.

Fig. 5

Functional enrichment analysis of shared genes. (A) The bar graph of GO enrichment analysis. (B) The bar graph of KEGG enrichment analysis. (C) The Sankey diagram of KEGG enrichment analysis.

Fig. 6

Identification of candidate hub genes by PPI networks and machine learning. (A) The 10 genes were identified according to MCC algorithms from the CytoHubba plugin in Cytoscape. (B) AUC scores of machine learning combinations within the training and validation datasets. The genes within the optimal machine learning combination (first-ranked) were selected as candidate hub genes. (C) ROC curves for the GSE28042, GSE24206, and GSE53843 datasets.

Fig. 7

Identification of hub genes. (A) ROC curves of NLRC4, SPI1, SYK, NCF2, and TLR1 in the obesity dataset, respectively. (B) ROC curves of NLRC4, SPI1, SYK, NCF2, and TLR1 in the IPF dataset, respectively. (C) Expression of NLRC4, SPI1, SYK, NCF2, and TLR1 in the obesity dataset, respectively. (D) Expression of NLRC4, SPI1, SYK, NCF2, and TLR1 in the IPF dataset, respectively.

Fig. 8

Experimental verification of the mRNA expression levels of NLRC4, SPI1, and NCF2 in animal models. Quantitative Real-time PCR analysis showed the increased expressions of NLRC4, SPI1, and NCF2 in mice of obe + BLM group. *, p < 0.05; **, p < 0.01.

Fig. 9

Immune cell infiltration analysis. (A) The proportion of 22 immune cell types in each sample. (B) Boxplots showing the pattern of immune cell infiltration in the healthy control group and IPF group. (C) The correlations between immune cells. Red indicates a positive correlation, while blue indicates a negative correlation. (D-F) The correlations between the expression of three hub genes (NLRC4, SPI1, and NCF2) and immune cells. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

Fig. 10

Molecular docking analysis of flavonoid compounds targeting hub proteins. (A) Heatmap illustrating binding affinities (kcal/mol) between six dietary flavonoids (Luteolin, Naringenin, Kaempferol, Epicatechin, Daidzein, Peonidin) and three hub targets (NLRC4, SPI1, NCF2). (B) Representative 3D structural models of six dietary flavonoids and three hub targets. Hydrogen bonds and hydrophobic interactions are highlighted in dashed lines and surface shading, respectively.