Anoikis-related biomarkers PARP1 and SDCBP as diagnostic and therapeutic targets for asthma

Abstract

This study aims to explore the association between anoikis-related genes (ARGs) and asthma. The dataset GSE143303 for asthma were sourced from the GEO database, while ARGs were retrieved from the Harmonizome web portal and the GeneCards database. Differentially expressed genes (DEGs) identification and GO, KEGG enrichment analysis were performed to reveal potential biological pathways. To identify hub anoikis-related DEGs (hub ARDEGs), we employed WGCNA and machine learning methods including LASSO and Random Forest. Additionally, we constructed risk prediction nomogram model and ROC curves to evaluate the asthma diagnostic value of hub ARDEGs. SsGSEA immune infiltration analysis was used to analyze the role of hub ARDEGs in the asthma immune microenvironment. Finally, miRNAs and transcription factors (TFs) interacting with these hub ARDEGs were investigated. DEGs of ARGs between asthma and healthy controls, along with WGCNA, led to the identification of six ARDEGs. GO and KEGG analyses revealed that these ARDEGs were primarily involved in the apoptotic signaling pathway and adherens junctions. Machine learning methods further narrowed down the six ARDEGs to two hub ARDEGs: PARP1 and SDCBP, which were significantly upregulated in asthma and validated using the GSE147878 and experimental models. Based on these two hub ARDEGs, a risk prediction model for asthma was developed, demonstrating strong diagnostic potential and tissue specificity in endobronchial biopsies. Immune analysis revealed variations in immune cell infiltration within asthma samples correlated with hub ARDEGs. Additionally, the miRNA-TF-mRNA interaction network of hub ARDEGs highlights the complexity of the regulatory process. The process of anoikis, immune dysregulation, and asthma are closely interconnected. The anoikis-related biomarkers PARP1 and SDCBP may serve as diagnostic markers and therapeutic targets for asthma.

Keywords: Anoikis; Asthma; Biomarkers; Machine learning.

Fig. 1

The flowchart of this study.

Fig. 2

Screening of DEGs and asthma-related DEGs. (A) Volcano plot illustrating upregulated (red) and downregulated (blue) DEGs. (B) Heatmap displaying hierarchical clustering of DEGs between asthma and control groups. (C–E) Bubble plots representing GO enrichment analysis of DEGs, including Biological Processes (BP), Cellular Components (CC), and Molecular Functions (MF). (F) Bubble plot depicting KEGG enrichment analysis of DEGs. Figure (F) is cited from the source: www.kegg.jp/kegg/kegg1.html.

Fig. 3

Identification of gene modules associated with dilated asthma using WGCNA. (A) Selection of optimal soft thresholding power (β) based on a scale-free fit index (left) and mean connectivity (right) for varying powers. (B) Gene dendrogram illustrating modules associated with asthma, represented by distinct colors. (C) Correlation heatmap demonstrating relationships between gene modules and asthma status. (D,E) Scatter plots showing correlations between module membership (MM) and gene significance (GS) in the red and greenyellow modules.

Fig. 4

Correlation and enrichment analysis of ARDEGs. (A) Venn diagram depicting the intersection of DEGs, WGCNA genes, and Anoikis-related genes. (B) Correlation analysis of ARDEGs, *P < 0.05, **P < 0.01, ***P < 0.001. (C) Chromosomal distribution of six ARDEGs. (D–F) Bubble plots of GO enrichment analysis of ARDEGs, encompassing BP, CC, and MF. (G) Dot plot illustrating the top 10 KEGG entries for ARDEG enrichment. Figure (G) is cited from the source: www.kegg.jp/kegg/kegg1.html.

Fig. 5

Identification of hub ARDEGs by machine learning and Evaluation of gene expression. (A) LASSO coefficient analysis, with vertical dashed lines indicating the optimal lambda. (B) Fivefold cross-validation for selecting adjustment parameters in the LASSO model, with each curve representing a gene. (C) ROC curve demonstrating the LASSO model’s diagnostic performance. (D) Ranking of ARDEGs based on relative importance in the random forest model. (E) Error rate plotted against the number of random forest trees. (F) ROC curve of the random forest model. (G) Venn diagram combining results from LASSO and random forest algorithms to identify overlapping genes. (H) Expression analysis of PARP1 and SDCBP in the training dataset. (I) Expression analysis of PARP1 and SDCBP in the validation dataset GSE147878, ***P < 0.001.

Fig. 6

Validation of hub ARDEGs expressions in asthma mice. (A,B) RT-PCR analysis showing mRNA expression of PARP1 and SDCBP. (C–E) Representative western blot and statistical analysis of protein expression levels for PARP1 and SDCBP. Data represent three independent experiments with three mice per group *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 7

Hub ARDEGs risk prediction model. (A) Nomogram for asthma diagnosis based on PARP1 and SDCBP, where each gene corresponds to a score contributing to the overall risk prediction. In this nomogram, the ‘Points’ serve as a quantitative measure representing the scores assigned to each candidate gene, whereas the ‘Total Points’ reflect the overall score accumulated from all the listed genes. The predictive accuracy and clinical applicability of the nomogram were assessed in the training set. (B) Calibration curve assessing nomogram accuracy. (C) Decision curve analysis evaluating the net benefit of diagnostic decisions based on the nomogram. (D) ROC curves for the training set showing the diagnostic performance of PARP1 and SDCBP. (E) ROC curve for the training set showing the diagnostic performance of the nomogram. (F) ROC curves for the validation dataset GSE147878 showing the diagnostic performance of PARP1 and SDCBP. (G) ROC curves for the validation dataset GSE147878 showing the diagnostic performance of the nomogram.

Fig. 8

ssGSEA immune infiltration. (A) Heatmap depicting infiltration levels of 28 immune cells across samples. (B) Violin plots comparing immune cell infiltration between asthma and control samples. (C) Correlation Analysis between PARP1 and 28 Types of Immune Cells. (D) Correlation Analysis between SDCBP and 28 Types of Immune Cells. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 9

Single-gene GSEA enrichment analysis of Hub ARDEGs. GO and KEGG enrichment analyses for PARP1 and SDCBP using GSEA, displaying the top six enriched terms.

Fig. 10

The miRNA-TF-mRNA regulatory network. Visualization of target gene mRNAs (Red Circles), miRNAs (Blue V-Shapes) and TFs (Pink Squares).