Identification of key genes in osteoarthritis using bioinformatics, principal component analysis and meta-analysis

Abstract

The present study aimed to identify key genes involved in osteoarthritis (OA). Based on a bioinformatics analysis of five gene expression profiling datasets (GSE55457, GSE55235, GSE82107, GSE12021 and GSE1919), differentially expressed genes (DEGs) in OA were identified. Subsequently, a protein-protein interaction (PPI) network was constructed and its topological structure was analyzed. In addition, key genes in OA were identified following a principal component analysis (PCA) based on the DEGs in the PPI network. Finally, the functions and pathways enriched by these key genes were also analyzed. The PPI network consisted of 241 nodes and 576 interactives, including a total of 171 upregulated DEGs [e.g., aspartylglucosaminidase (AGA), CD58 and CD86] and a total of 70 downregulated DEGs (e.g., acetyl-CoA carboxylase β and dihydropyrimidine dehydrogenase). The PPI network complied with an attribute of scale-free small-world network. After PCA, 47 key genes were identified, including β-1,4-galactosyltransferase-1 (B4GALT1), AGA, CD58, CD86, ezrin, and eukaryotic translation initiation factor 4 γ 1 (EIF4G1). Subsequently, the 47 key genes were identified to be enriched in 13 Gene Ontology (GO) terms and 2 Kyoto Encyclopedia of Genes and Genomes pathways, with the GO terms involving B4GALT1 including positive regulation of developmental processes, protein amino acid terminal glycosylation and protein amino acid terminal N-glycosylation. In addition, B4GALT1 and EIF4G1 were confirmed to be downregulated in OA samples compared with healthy controls, but only EIF4G1 was determined to be significantly downregulated in OA samples, as determined via a meta-analysis of the 5 abovementioned datasets. In conclusion, B4GALT1 and EIF4G1 were indicated to have significant roles in OA, and B4GALT1 may be involved in positive regulation of developmental processes, protein amino acid terminal glycosylation and protein amino acid terminal N-glycosylation. The present study may enhance the current understanding of the molecular mechanisms of OA and provide novel therapeutic targets.

Keywords: genomic meta-analysis; network; osteoarthritis; principal component analysis.

Figure 1

PCA biplots of QC measures for the five datasets included in the present meta-analysis. The abscissa indicates the first PC and the ordinate indicates the second PC in the PCA. QC, quality control; IQC, internal QC; EQC, external QC; CQCg, consistency QC of differentially expressed gene ranking; CQCp, consistency QC of enriched pathway ranking, AQCg, accuracy QC of gene detection; AQCp, accuracy QC of enriched pathway detection, PCA, principal component analysis.

Figure 2

Hierarchical cluster heatmap for the differentially expressed genes from the five datasets (GSE55457, GSE55235, GSE82107, GSE12021 and GSE1919). OA, osteoarthritis; Ctrl, control.

Figure 3

PPI network with a scale-free small world effect based on DEGs. (A) PPI network. Red indicates upregulated DEGs and green indicates downregulated DEGs. The color intensity indicates the logFC. The node size is proportional to its corresponding degree value and the names of hub genes are marked in dark blue. (B) The distribution of the degree of interaction of the nodes in the network. Scale-free feature means the degree of a small number of nodes are large, while that of most nodes are small. Small world effect means most nodes are not neighbors of one another, while they can be reached from any other nodes by a small number of steps. The x-axis displays the log2-transformed degree of interaction, while the y-axis indicates the numbers of the nodes. (C) Distribution of the path length of nodes in the network. FC, fold change; PPI, protein-protein interaction; DEG, differentially expressed gene.

Figure 4

Three-dimensional graph of sample distribution based on three principal components in the five datasets. (A) GSE55457, (B) GSE55235, (C) GSE82107, (D) GSE12021 and (E) GSE1919. Blue dots indicate the control samples and red triangles indicate osteoarthritis samples. PC, principal component.

Figure 5

Relative expression levels of B4GALT1 and EIF4G1 between osteoarthritis samples and control samples. (A) Forest plots of the expression values of the key genes. Squares indicate the outcome estimates for the corresponding study and the size of the square indicates the weight of the corresponding study. Horizontal lines and figures in parentheses represent the 95% CI. Diamonds indicate the pooled effect size with the corresponding 95% CI. (B) Relative expression of B4GALT1 and EIF4G1 mRNA vs. GAPDH in synovial tissues of a rat model of OA vs. control. ^*P<0.05; ^**P<0.01 vs. Control. Exp, expression; OR, odds ratio; CI, confidence interval; Pt, platform; PMID, PubMed unique identifier; B4GALT1, β-1,4-galactosyltransferase-1; EIF4G1, eukaryotic translation initiation factor 4 γ 1.