-
The Perseus and R software were used to screen DEPs in Shenzhen and Taiyuan samples. A total of 67 DEPs were screened in Shenzhen samples, of which 46 proteins were upregulated and 21 proteins were downregulated. Overall, 252 DEPs were screened in Taiyuan samples, of which 134 proteins were upregulated and 118 proteins were downregulated (Figure 1).
-
KEGG functional analysis is shown in Figure 2. The upregulated proteins in Shenzhen PM2.5 samples are mainly involved in signal pathways, including mineral absorption, renal cell carcinoma, and hypoxia-inducible factor (HIF)-1 (HIF-1 signaling pathway). The downregulated proteins in Shenzhen PM2.5 samples were mainly involved in signal pathways such as ubiquitin-mediated proteolysis (Hbi-1 signaling) and HIF-1 (HIF-1 signaling pathway). In Taiyuan PM2.5 samples, the upregulated proteins were mainly involved in signal pathways such as spliceosome and RNA transport, whereas the downregulated proteins were mainly involved in cell components such as ribosomes.
-
Following the exposure of HBE cells to the Shenzhen and Taiyuan samples, GO enrichment analysis was performed and their differential proteins are shown in Figures 3–5. At the biological process (BP) level, the upregulated proteins in the Shenzhen samples were mainly distributed in detoxification of copper ions, the copper ion stress response to copper ions, cellular response to zinc ions, and other biological processes. In Taiyuan samples, the upregulated proteins were mainly distributed in protein folding and cellular responses to metal ions; the downregulated proteins were mainly distributed in the negative regulation of protein translation and the regulation of protein polymerization, as well as other biological processes. In the Taiyuan samples, at the molecular function (MF) level, the upregulated proteins were mainly distributed in molecules that are functionally associated with peptide disulfide oxidoreductase activity and peptide disulfide oxidoreductase activity; the downregulated proteins were mainly distributed in MF, including hydrolase activity and actin filament binding. At the cellular component (CC) level, downregulated proteins in Shenzhen samples were mainly distributed on cell components such as the myelin sheath and axons. In Taiyuan samples, the upregulated proteins were mainly distributed in cell components such as mitochondrial protein complex, spliceosomal complex, and ribosomes. The downregulated proteins in Taiyuan samples were mainly distributed in the cytosol and cell components, including the cell cortex.
Figure 4. Molecular function analysis of GO for differentially expressed proteins. GO, Gene ontology.
Figure 5. Cellular component analysis of GO for differentially expressed proteins. GO, Gene ontology
GO-BP enrichment analysis provided no information regarding proteins enriched in SZ down; in the GO-MF analysis, no information was obtained regarding proteins enriched in SZ up; and in GO-CC analysis, no information was available regarding proteins enriched in SZ up and TY down.
-
The Venn diagram of the DEPs is shown in Figure 6. Among them, 33 proteins were upregulated only in the Shenzhen samples, 12 proteins were downregulated only in the Shenzhen samples, 109 proteins were upregulated only in the Taiyuan samples, 121 proteins were downregulated only in the Taiyuan samples, 13 proteins were upregulated both in Shenzhen and Taiyuan samples, and 9 proteins were downregulated proteins both in Shenzhen and Taiyuan samples.
-
The 4,297 proteins detected in the Shenzhen sample group, the Taiyuan sample group, and the control group were used to construct a WGCNA co-expression network module. The hierarchical clustering of the samples was analyzed using the flashClust function. The clustering results are shown in Figure 7. For the 4,297 proteins, the power value of 10, the lowest power value for the scale with an independence degree of up to 0.85, was selected to construct a hierarchical clustering tree (Figure 8).
Based on the principle of a certain gap between each module and a certain number of modules, we screened out the color dynamic hybrid TOM in Figure 7 for subsequent analysis. For the 4,297 detected proteins, correlation network analysis was performed. The results demonstrated 39 modules, of which 7 modules presented P valves < 0.05. By establishing the threshold of the correlation coefficient as 0.8, we obtained 3 modules, with each module represented by one color, and proteins in the same module demonstrating a higher degree of correlation (Figure 8).
-
KEGG enrichment analysis was performed for proteins in the eight significant co-expression modules (Figure 9). Among them, the protein in the green module was mainly concentrated in RNA transport and necroptosis pathways; the protein in the orange module was mainly concentrated in pathways such as necroptosis, mineral absorption, and protein processing in the endoplasmic reticulum; the protein in the pink module was mainly concentrated in protein processing in the endoplasmic reticulum, biosynthesis of amino acids, and fatty acid metabolism pathways.
-
The Cytoscape software was used to build the module, as well as to calculate the connectivity within the module. The intramodular connectivity was calculated for each gene by adding the connection strengths with other module genes and dividing this number by the maximum intramodular connectivity. Genes with high intramodular connectivity are considered as intramodular hub genes. In this study, we obtained nine hub genes, including ANXA2, DIABLO, RPL35A, CRABP2, FDPS, AIMP1, NDUFB1, CNN2, and BSG (Table 1). As shown in Figure 10A, the central genes of the green module are ANXA2, DIABLO, and RPL35A. As shown in Figure 10B, the central genes of the orange module are PARP1, CRABP2, FDPS, AIMP1, and NDUFB1. As shown in Figure 10C, the central genes of the pink module are CNN2, BSG, and FDPS. Combined with the differential proteins obtained from the Perseus software analysis and the hub protein in the module, we screened nine key proteins, including annexin, TNF receptor-associated protein 1 variant, diablo homolog, mitochondrial, 60S ribosomal protein L35a, cellular retinoic acid-binding protein 2, farnesyl pyrophosphate synthase, aminoacyl tRNA synthase complex-interacting multifunctional protein 1, NDUFB1 protein, calponin, and basigin (Table 1).
Protein serial number Protein name Hub genes Molecular weight (kD) A0A024R5Z7 Annexin ANXA2 38.58 Q9NR28 Diablo homolog, mitochondrial DIABLO 27.11 P18077 60S ribosomal protein L35a RPL35A 12.53 P29373 Cellular retinoic acid-binding protein 2 CRABP2 15.68 P14324 Farnesyl pyrophosphate synthase FDPS 48.24 Q12904 Aminoacyl tRNA synthase complex-interacting multifunctional protein 1 AIMP1 34.33 Q3MHU6 NDUFB1 protein NDUFB1 9.97 B4DUT8 Calponin CNN2 35.92 P35613 Basigin BSG 42.174 Note. WGCNA, Weighted correlation network analysis. Table 1. The hub genes and differentially expressed proteins from WGCNA analysis
Characteristics of Atmospheric Fine Particulate Matter (PM2.5) Induced Differentially Expressed Proteins Determined by Proteomics and Bioinformatics Analyses
doi: 10.3967/bes2020.077
- Received Date: 2020-02-11
- Accepted Date: 2020-05-01
-
Key words:
- PM2.5 /
- Proteomics /
- Bioinformatics /
- Differentially expressed proteins /
- Weighted correlation network analysis
Abstract:
Citation: | ZHENG Kai, CAI Ying, WANG Bing Yu, QIN Shuang Jian, LI Bo Ru, HUANG Hai Yan, QIN Xiao Yun, LONG Ding Xin, ZHANG Zhao Hui, XU Xin Yun. Characteristics of Atmospheric Fine Particulate Matter (PM2.5) Induced Differentially Expressed Proteins Determined by Proteomics and Bioinformatics Analyses[J]. Biomedical and Environmental Sciences, 2020, 33(8): 583-592. doi: 10.3967/bes2020.077 |