Combining co-expression analysis and machine learning to explore the diagnostic value of complement and coagulation
Jingjing Feng1,2, Chengfang Xie1,2, Weiduo Jin1,2
1Department of Respiratory Medicine, Shangyu People's Hospital of Shaoxing, No. 517, Shimin Avenue, Shaoxing, Zhejiang, 312300, China.
Background:
Optimizing diagnostic strategies for asthma is crucial for alleviating its high global disease burden. In recent years, the complement and coagulation cascades (CCC), as a central hub of innate immunity and inflammatory responses, has gained increasing attention for its association with asthma.
Methods:
Asthma-related datasets (GSE67472, GSE4302, and GSE76262) were downloaded from the Gene Expression Omnibus database. Subsequently, weighted gene co-expression network analysis and four machine learning algorithms-Decision Tree, K-Nearest Neighbors, Random Forest (RF), and Gradient Boosting Machine-were employed to identify key genes and construct a diagnostic model. SERPINB2 expression was then detected in clinical samples. The functional mechanisms of the key genes were validated using Transforming Growth Factor Beta 1 (TGF-β1)-treated Human Bronchial Epithelial Cells (BEAS-2B) cells through Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR), western blot, CCK-8 assay, and Co-IP experiments.
Results:
This research identified seven key genes: Serpin Family B Member 2 (SERPINB2), Complement C3 (C3), Complement C6 (C6), Colony Stimulating Factor 2 Receptor Subunit Beta (CSF2RB), Death Associated Protein Kinase 1 (DAPK1), Solute Carrier Family 16 Member 6 (SLC16A6), and CD69 Molecule (CD69). The diagnostic model constructed based on the RF algorithm exhibited robust discriminatory performance in both the training and validation sets. qRT-PCR assays on clinical tissue samples revealed significantly upregulated SERPINB2 expression in asthmatic tissues. In vitro experiments further revealed that SERPINB2 expression was upregulated in TGF-β1-treated BEAS-2B cells. Knockdown of SERPINB2 suppressed the expression of proteins related to epithelial-mesenchymal transition, inflammatory responses, and airway remodeling. Co-immunoprecipitation (Co-IP) assays indicated an association between SERPINB2 and C3 at the protein level.
Conclusion:
The RF model constructed based on seven CCC-related candidate genes exhibited promising diagnostic value for asthma. SERPINB2 may participate in airway epithelial inflammation and remodeling and may be associated with complement C3 at the protein level. These findings provide novel evidence for further elucidating the potential roles and clinical implications of the CCC cascade in the initiation and progression of asthma.