LYN and CYBB are pivotal immune and inflammatory genes as diagnostic biomarkers in recurrent spontaneous abortion

Abstract

Recurrent spontaneous abortion (RSA) seriously affects women's reproductive health, and its pathogenesis is complex and varied. The aim of this study is to identify key molecular markers closely associated with RSA to rapidly and effectively predict the RSA, and to provide simple and practical indicators for clinical diagnosis and treatment.

Method: We obtained mRNA expression profiles from the GSE26787 and GSE165004 datasets of the Gene Expression Omnibus (GEO) database, immune-related genes (IRGs) from the ImmPort database (https://www.immport.org), and genes related to inflammatory response from the Molecular Signatures database. Different Inflammation- and immunity-related genes (DIIRGs) were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. Protein-protein interaction (PPI) networks were utilized to explore the connections between various DIIRGs. The candidate DIIRGs were analyzed by the least absolute shrinkage and selection operator (LASSO) and the multiple support vector machine recursive feature elimination (mSVM-RFE). The diagnostic ability of the candidate genes was verified using receiver operating characteristic (ROC) curves. The performance of the predictive model was evaluated using a Nomo plot. We further confirmed the expression levels and diagnostic value of key genes by performing immunohistochemistry (IHC) in clinical tissue samples. The compositional patterns of the infiltration of 22 immune cell types in RSA were analyzed via the CIBERSORT algorithm.

Result: We identified 403 differentially expressed genes (DEGs) and 7 DIIRGs between RSA endometrium and Non-RSA endometrium. GO analysis showed that DIIRGs were significantly enriched in positive regulation of cell-cell adhesion, inflammatory response to antigenic stimulus, and protein tyrosine kinase activity. KEGG enrichment analyses were performed mainly on Epithelial cell signaling in Helicobacter pylori infection, NOD-like receptor signaling pathway, and Ras signaling pathway. A predictive and diagnostic model composed of three genes (CYBB, LYN, and MET). The CYBB, LYN, and MET genes were identified as diagnostic biomarkers of RSA (AUC = 0.747, AUC = 0.751, AUC=0.703), and reduced levels of CYBB and LYN expression were found to correlate with RSA in clinical samples. In addition, immune microenvironment analysis showed that CYBB and MET were positively correlated with naïve B cells and negatively correlated with CD8 T cells, LYN and MET were positively correlated with M2 macrophages and negatively correlated with eosinophils, respectively (P < 0.05).

Conclusion: Inflammation-immunity is a key factor in the pathogenesis of RSA. CYBB and LYN are regarded as the crucial genes that constitute a model and contribute to inflammation-immunity throughout the occurrence and progression of RSA. These findings provide a new perspective on the diagnosis and pathogenesis of RSA.

Keywords: cybb; immunity; inflammation; lyn; recurrent spontaneous abortion.

Figure 1

Process diagram of this research.

Figure 2

Identification and function of DIIRGs. (A) The expression levels of the first 50 DEGs between RSA tissue and Non-RSA samples from the GEO database are visualized through heatmaps. The genes are named in the row annotations, and the column annotations, which are the sample IDs, are not shown in the plots. The color gradient, which goes from red to blue, represents the expression levels from high to low in the heatmaps. (B) 403 DEGs between RSA and Non-RSA are illustrated by the volcano plots. In these plots, genes that are upregulated are indicated by red dots, genes that are downregulated are denoted by green dots, and genes without differential expression are represented by black dots. (C) 7 DIIRGs are contained in the Venn diagram of the intersection of differential genes, inflammatory genes, and immune genes. (D) A circle plot for GO analysis of 7 DIIRGs is presented. (E) A bar graph for GO analysis of 7 DIIRGs is shown. (F) The DIIRGs are annotated by KEGG.

Figure 3

Association between DIIRGs and hub genes. (A) The PPI networks exploring five DIIRGs binding protein interactions are constructed by using the STRING tool. (B) Five hub genes are obtained through the MCC algorithm. (C) The expression levels of five hub genes between RSA tissue and Non-RSA samples are visualized through heatmaps. (D) Five hub genes between RSA tissue and Non-RSA samples are illustrated by the volcano plots.

Figure 4

Analyzing the correlation of DIIRGs. (A) DIIRGs’ co-expression network map is depicted. (B) The correlation of DIIRGs is plotted. (C) A scatter plot for some highly correlated DIIRGs is provided.

Figure 5

Construction of a prediction model for RSA. (A) A curve in the LASSO regression coefficient profiles of the 5 DIIRGs illustrates the changing course of each DIIRG. (B) The LASSO Cox regression model was employed to draw a plot of partial likelihood deviance versus log (l). (C) When k = 3, the curve of the total within the sum of the squared error curve under the corresponding cluster number k arrives at the “elbow point”. (D) At k = 3, the curve representing the average silhouette width for the corresponding cluster number k reaches its peak. (E) The Venn diagram shows the 3 diagnostic markers shared by the LASSO and SVM-RFE algorithms.

Figure 6

Additional analysis of three key DIIRGs. (A) The locations on the chromosome of three key DIIRGs. (B) RSA and Non-RSA can be clearly distinguished by principal component analysis using the three mentioned genes. (C) The relative expression levels of three key DIIRGs between RSA and Non-RSA are shown by the GSE26787+ GSE165004 datasets. (D) The performances of three key DIIRGs for predicting RSA in GSE26787+ GSE165004 datasets were verified by ROC curves. (E) Diagnostic Nomo plot related to three key DIIRGs. (F) Calibration curve of a model consisting of three key genes. **P < 0.01, ***P < 0.001.

Figure 7

Validation of the three essential DIIRGs. (A) Significantly low expression of LYN was found in RSA tissues when compared with Non-RSA specimens (Non-RSA specimens = 5; RSA = 6). (B) Significantly low expression of CYBB was found in RSA. (C) No Significant expression of MET was found between RSA and Non-RSA tissues. Representative images (×40 and ×200) of IHC staining for LYN, CYBB, and MET in 6 RSA and 5 Non-RSA specimens (high expression versus low expression); *P < 0.05. ns, no significance.

Figure 8

Distribution of immune cells is related to CYBB, LYN, and MET. (A) The distribution of 22 immune cell subtype proportions between RSA tissue and Non-RSA samples is illustrated by a bar plot. (B) Infiltrating immune cells in RSA are correlated with LYN, CYBB, and MET, and this correlation is analyzed.