Analysis Of Molecular Mechanism Of Carotid Atherosclerosis With Microarray Analysis And Text Mining

Background:Carotid atherosclerotic cardiovascular disease is degenerative, proliferative andchronic systemic disease occurred in medium-sized arteries. It is characterized byvascular smooth muscle cell proliferation and lipid deposition in atheroscleroticplaques. In this paper, microarray analysis and text mining techniques were used toprovide novel insights into the molecular mechanism of atherosclerosis, thusproviding the basis for early diagnosis and treatment of the disease.Methods:In the current study, GSE28829microarray data package was downloaded fromthe GEO database, consisting of13early and16late human carotid atheroscleroticplaque tissue samples. Platform: GPL570[HG-U133_Plus₂] Affymetrix HumanGenome U133Plus2.0Array. Differentially expressed genes （DEGs） were screenedusing the LIMMA package in the R language. False discovery rate （FDR）<0.05and|logFC （fold change）|>1was set as the strict threshold.Principal component analysis （PCA） was performed for identified DEGs usingthe prcomp function in R language to analyze the differences in all samples withDEGs as variables.To investigate the co-expression of these identified DEGs, DEGs were dividedinto two groups: up-regulated and down-regulated DEGs groups. Based on theinformation from the COXPRESdb database, coexpression network of DEGs wasconstructed. The co-expression coefficient for each pair of genes was calculated. Onlypairs of DEGs whose co-expression coefficient was bigger than0.8were reserved.Gene ontology （GO） and KEGG （Kyoto Encyclopedia of Genes and Genomes）enrichment analysis were performed for DEGs involved in the co-expression networkusing the DAVID software based on the hypergeometric distribution. FDR <0.05wasset as the criterion. From the obtained DEGs, Differentially Co-expressed Genes （DCGs） wasselected using DCGL package in R （adjusted p-value<0.05） and a co-expressionnetwork was build. Functional analysis was performed for the DCGs, andDCGs-pathway network was built to analyze the pathways that might be affected bythe DCGs.With the text mining based on natural language processing, we obtainedatherosclerosis-related molecular interaction network. Atherosclerosis-relatedMammalian Phenotype （MP） were acquired based on Mouse Genome Informatics（MGI） database. Then MP-related mutated genes and atherosclerosis-relatedimportant genes were screened. A molecular interaction network was constructed,followed by network module analysis and functional annotation analysis.Result:1．With FDR <0.05and|logFC|>1as the strict cutoff,319DEGs were identifiedbetween early and late carotid atherosclerotic samples. Among these DEGs,267were up-regulated DEGs, and52were down-regulated DEGs.The result of PCAanalysis showed that these identified DEGs were characteristic genes, whichcould be used to identify early carotid atherosclerotic samples from late carotidatherosclerotic samples. In the co-expression network of DEGs,84pairs ofco-expressed DEGs and48nodes were acquired. All DEGs in the network wasup-regulated genes. Of all DEGs in the network, the DEG with highest degree wasTYROBP. Result of GO functional analysis displayed that a large number ofunregulated DEGs, which was associated with immune were enriched in latecarotid atherosclerosis. Moreover, all DEGs involved in the co-expressionnetwork were associated with immune. The most significant biological functionwas the immune response, which was enriched with a total of18DEGs. The resultof KEGG pathway enrichment analysis showed the most significant signalingpathway was antigen processing and presentation, which was enriched by6DEGs:IFI30, CTSS, HLA-DMB, HLA-DMA, CD74, HLA-DRA. Out of the6DEGs, HLA-DMB, HLA-DMA and HLA-DRA were enriched in immune function aswell.2．Overall,45DCGs were obtained and mainly involved in immune response, stressresponse and apoptosis. HOPX, IGHM, SLA, CD163and IGKV1-5affected agreat number of pathways.Cytokine-cytokine receptor interaction, cell adhesionmolecules （CAMs）, type I diabetes mellitus, focal adhesion and leukocytetrans-endothelial migration were associated with several DCGs.3．Through text mining,9human atherosclerosis-related mutated genes wereobtained. a molecular interaction network was built, and contained1,918nodesand4584arteriosclerosis-related molecular interaction relationship MAPK8,BCL2, LEP, IL10, NOS2A, MMP9, CCL2and CD44were important networknodes with a large number of the molecular interactions.Based on Molecularinteraction network,5clusters were obtained and they are significantly related tocomplement and coagulation cascades, ECM-receptor interaction, focal adhesion,nucleotide excision repair and lysosome pathways.Conclusions：Microarray analysis and text mining were employed to explore critical genes andpathways in the mechanism of carotid atherosclerosis, thus providing a base forfurther studies on the carotid atherosclerosis. Our findings confirmed roles of immunefunction, important pathways including antigen processing and presentation pathway,complement and coagulation cascades, ECM-receptor interaction, focal adhesion,nucleotide excision repair and lysosome pathways in the pathology of carotidatherosclerosis. HLA-DMB, HLA-DMA and HLA-DRA that enriched in bothimmune function and antigen processing and presentation pathway, important DCGs（HOPX, IGHM, SLA, CD163and IGKV1-5） and important network nodesincluding MAPK8, BCL2, LEP, IL10, NOS2A, MMP9, CCL2and CD44might bepotential gene markers to distinguish early carotid atherosclerosis from late carotidatherosclerosis and could serve as potential targets to develop novel medicines andtherapeutic approaches to treat carotid atherosclerosis. The findings from this study provide more information associated with the molecular mechanism of carotidatherosclerosis and contribute to a solid base for future studies over this disease.