Data Normalization
Each array was normalized (centered) by quartile data normalization using the beadarray package in R bioconductor. Fig. 1A and Fig. 1B shows a boxplot of (before and after normalization) gene expression in each of the two groups in the E-MTAB-6728 dataset.
Identification of DEGs between obese patients and lean persons
To preliminarily understand the mechanism contributing to the obesity, 24 patients (adipocytes from 12 obese patients and 12 lean persons) were selected for subsequent analysis. A total of 876 DEGs (Log FC > 0.524 for up regulated genes, Log FC < -0.394 for down regulated genes, P < 0.05) were diagnosed (Table 2). Among them, 438 genes were up regulated and 438 genes were down regulated. Up regulated and down regulated gene expression heat map are shown in Fig. 2A and Fig.2B. Volcano map for DEGs is shown in Fig. 3.
Pathway enrichment analysis of DEGs
To preliminarily comprehend the function of DEGs, we submitted up regulated genes and down regulated genes, respectively, to the online software ToppCluster diagnose related pathways from different pathway databases (BIOCYC, KEGG, PID, REACTOME, GenMAPP, MSigDB C2 BIOCARTA, PantherDB and SMPDB). Pathway enrichment analysis results showed that up regulated genes were significantly enriched in thyroid hormone metabolism II (via conjugation and/or degradation), serotonin degradation, ECM-receptor interaction, focal adhesion, IL6-mediated signaling events, glypican 1 network, collagen formation, binding and uptake of ligands by scavenger receptors, steroid hormone metabolism, glycosaminoglycan degradation, genes encoding collagen proteins, genes encoding structural ECM glycoproteins, integrin signalling pathway, axon guidance mediated by Slit/Robo, hypertension, integrin signaling, suprofen pathway and mefanamic acid pathway (Table 3). Down regulated genes were mainly significantly enriched in the super pathway of methionine degradation, fatty acid beta-oxidation, ribosome, propanoate metabolism, FoxO family signaling, HIF-2-alpha transcription factor network, eukaryotic translation elongation, metabolism of amino acids and derivatives, selenoamino acid metabolism, nitrogen metabolism, CDK regulation of DNA replication, PTEN dependent cell cycle arrest and apoptosis, p38 MAPK pathway, serine glycine biosynthesis, glycine, serine and threonine metabolic, transulfuration of homocysteine metabolism, methionine metabolism and ketone body metabolism (Table 4).
Gene ontology (GO) enrichment analysis of DEGs
The results of GO enrichment analysis of up and down regulated genes in obese patients analyzed based on GO terms such as Biological Process (BP), Cellular Component (CC) and Molecular Function (MF) using ToppCluster are shown in Table 5 and Table 6. Up regulated genes were significantly enriched in BP, CC and MF related to blood vessel morphogenesis, cardiovascular system development, extracellular matrix, extracellular matrix component, growth factor binding and integrin binding. Down regulated genes were significantly enriched in BP, CC and MF related to organic acid biosynthetic process, cellular amide metabolic process, cytosolic small ribosomal subunit, ribosome, structural constituent of ribosome and transcription factor binding.
Integration of PPI network and module analysis
The mentha database and Cytoscape were used to construct a PPI network of up and down regulated genes. A total of 7271 nodes 16270 edges are included in the PPI networks of up regulated genes (Fig. 4I). The up regulated hub genes such as HSPA8, HSPA5, ERBB2, STAT3, YWHAH, SPTAN1, STEAP2, COL11A1, NEK6, COPG2 and NOTCH3 with highest node degree distribution, betweenness centrality, stress centrality, closeness centrality and low clustring coefficient are listed in Table 7, and statistical results and scatter plot for node degree distribution, betweenness centrality, stress centrality, closeness centrality and clustring coefficient are shown in Fig. 4IIA – 4IIB. These up regulated hub genes were mainly enriched in toxoplasmosis, Parkinson disease, focal adhesion, proteoglycans in cancer, PI3K-Akt signaling pathway, cytokine signaling in immune system, protein digestion and absorption and blood vessel morphogenesis. A total of 7276 nodes 19862 edges are included in the PPI networks of down regulated genes (Fig. 5I). The down regulated hub genes such as ESR1, FBL, CSNK2B, ARRB1, PCM1, RPS2, RPS3, PTEN, SIRT1, ZNF581, CKB, FBXO11, UBR2 and INTS6 with highest node degree distribution, betweenness centrality, stress centrality, closeness centrality and low clustring coefficient are listed in Table 7, and statistical results and scatter plot for node degree distribution, betweenness centrality, stress centrality, closeness centrality and clustring coefficient are shown in Fig. 5IIA – 5IIE. These down regulated hub genes were mainly enriched in regulation of nuclear SMAD2/3 signaling, the SARS-coronavirus life cycle, herpes simplex infection, macromolecule catabolic process, ribosome, cellular amide metabolic process, FoxO signaling pathway, AMPK signaling pathway, metabolic pathways, nucleolus, negative regulation of nucleic acid-templated transcription and nucleoplasm part.
Subsequently, we performed module analysis of the whole network by the PEWCC1 plug-in. 1798 modules were extracted from the PPI network (up regulated). The four most significant modules were selected (module 12 (49 nodes and 99 edges), module 17 (46 nodes and 110 edges), module 26 (41 nodes and 91 edges) and module 68 (68 had 14 nodes and 25 edges)) in this PPI network (Fig. 6A). Pathway and GO enrichment analysis showed that HSPA8, ERBB2, HSPA5, SPP1, YWHAH, STAT3, MSN and ITGB5 were included in module 12, module 17, module 26 and module 68, and participated in toxoplasmosis, focal adhesion, Parkinson disease, ECM-receptor interaction, negative regulation of multicellular organismal process, regulation of locomotion, cell migration and extracellular structure organization. Meanwhile, 2079 modules were extracted from the PPI network (down regulated). The four most significant modules were selected (module 15 (66 nodes and 754 edges), module 28 (46 nodes and 146 edges), module 60 (31 nodes and 146 edges) and module 86 (23 nodes and 67 edges)) in this PPI network (Fig. 6B). Pathway and GO enrichment analysis showed that ESR1, SRSF3, SIN3A, CSNK2A2, RPL14, KHDRBS1, SIRT1 and CSNK2B were included in module 15, module 28, module 60 and module 86, and participated in regulation of nuclear SMAD2/3 signaling, herpes simplex infection, signaling events mediated by HDAC Class I, metabolism of lipids and lipoproteins, cellular amide metabolic process, translation, organic acid biosynthetic process and transcription factor binding.
Construction of target genes - miRNA regulatory network
Up and down regulated target genes interacts with miRNA are shown Fig. 7A and Fig. 7B. The up regulated targeted genes (SOD2 (degree = 257, ex; hsa-mir-3144-3p ), CCND1 (degree = 251, ex; hsa-mir-7706), TUBB2A (degree = 193, ex; hsa-mir-7162-3p), CCND2 (degree = 179, ex; hsa-mir-5692c) and TMEM189 (degree = 146, ex; hsa-mir-548z)) are listed in Table 8. These up regulated target genes were mainly enriched in hypertension, focal adhesion and phagosome, Similarly, down regulated targeted genes (BTG2 (degree = 247, ex; hsa-mir-6075), TXNIP (degree = 228, ex; hsa-mir-3194-3p), MED28 (degree = 203, ex; hsa-mir-6861-5p), CNBP (degree = 197, ex; hsa-mir-4651) and MKNK2 (degree = 195, ex; hsa-mir-3650)) are listed in Table 8. These down regulated target genes were mainly enriched in cellular amide metabolic process, negative regulation of nucleic acid-templated transcription, nucleoplasm part, DNA-binding transcription factor activity and p38 MAPK pathway.
Construction of target genes - TF regulatory network
Up and down regulated target genes interacts with TFs are shown Fig. 8A and Fig. 8B. Up regulated targeted genes (YWHAH (degree =70, ex; MAZ), LYZ (degree = 62, ex; TFDP1), HP (degree = 60, ex; KLF9), TRAM2 (degree = 54, ex; KLF16) and CCND1 (degree = 51, ex; EZH2) are listed in Table 9. These up regulated target genes were mainly enriched in PI3K-Akt signaling pathway, neutrophil degranulation, amb2 Integrin signaling and focal adhesion. Similarly, top five down regulated targeted genes (EFNA1 (degree = 91, ex; TFDP1), MED16 (degree = 85, ex; MAZ), RWDD2A (degree = 82, ex; KDM5B), ADD3 (degree = 82, ex; SAP30) and AIP (degree = 82 , ex; PHF8) are listed in Table 9. These down regulated target genes were mainly enriched in HIF-2-alpha transcription factor network, fatty acid, triacylglycerol, and ketone body metabolism, chromosome and transcription factor binding.
Validation of hub genes
Immunohistochemical analysis demonstrated that the expression of STAT3, CORO1C, SERPINH1, MVP and ITGB5 were highly expressed in adipose tissues, whereas PCM1, SIRT1, EEF1G, PTEN and RPS2 were low expressed in adipose tissue (Fig. 9I). ROC curve analyses and the AUC were used to assess the discriminatory ability of ten DEGs (STAT3, CORO1C, SERPINH1, MVP, ITGB5, PCM1, SIRT1, EEF1G, PTEN and RPS2). The AUCs of all these four DEGs, including STAT3 (0.951), CORO1C (0.799), SERPINH1 (0.924), MVP (0.938), ITGB5 (0.938), PCM1 (0.826), SIRT1 (0.799), EEF1G (0.913), PTEN (0.833) and RPS2 (0.840), were more than 0.6 (Fig. 9II), which had great diagnostic value for obesity. Then, we also performed RT‐PCR to detect obesity patients and lean controls, finding that STAT3, CORO1C, SERPINH1, MVP and ITGB5 levels were elevated in adipose tissues of obesity patients compared with normal lean tissues, whereas PCM1, SIRT1, EEF1G, PTEN and RPS2 levels were decresed in adipose tissues of obesity patients compared with normal lean tissues (Fig. 9III).
Molecular docking studies
In the present research, the docking simulations are performed to knowidentify the active site conformation and major interactions responsible for complex stability with the ligand receptor. Designed novel molecules containing four membered more sensitive β-lactam ring, the four membered and performed docking studies using Sybyl X 2.1 drug design software. Molecules containing β-lactam ring is designed which is easily reacting group Fig.10, based on the structure of anti-obesity drug Orlistatfour membered ring Fig.11, has potent pancreatic lipase inhibitory activity.The molecules were designed based on the structure of the standard anti-obesity drug Orlistat. The one protein in each of three over expressed genes of ERBB2, its co-crystallised protein of PDB code 1MFL,HSPAB 8its co-crystallised protein of PDB code 5OOW and STAT 3its co-crystallised protein of PDB code of 3CWG respectively selected for docking studies. The investigation of designed molecules were performed to identify the potential molecule. The most of the designed molecules with respect to the standard anti-obesity drug Orlistat,obtained C-score greater than 5. The C-score greater than 5 are said to be an active, among total of 32 designed molecules few molecules have excellent good binding energy (C-score) greater than 7 respectively. The molecule ND 4, FU 5 and PF 5 obtained score of 7.242, 7.659&7.842with 1MFL and the molecules PM 6, ND 1, ND 3, ND 5, ND 6, PF 5 and PF 6obtained score of 7.5269,7.6271, 8.0824, 7.6595, 7.0792&7.2659with 3CWGand the molecules PM 4, PM 6, ND 1, ND 5, ND 6, PF 4, and PF obtained good binding score of 7.1631, 8.8312, 7.3781, 7.9872, 7.9567, 7.0213 & 7.0386with 5OOW respectively. The molecules found binding score 5-6 is PM1, PM 2 PM 3 PM 4 PM 5 PM 6 PM 7 PM 8 ND 1 ND 2 ND 3 ND 5 ND 6 ND 7 ND 8 FU 1 FU 2 FU 3 FU 4 FU 7 FU 8 PF 1 PF 2 PF 3 PF 4 PF 6 PF 7 PF 8 and standard Olistat (STD)with 1MFL and PM 2, PM 6, FU 17, FU 18, FU 19, FU 20, FU 23 PF 26 PF 27 PF 28 & PF 32 with 3CWG and PM1 PM 2 PM 3 PM 5 PM 7 PM 8 ND 2 ND 3 ND 4 ND 7, ND 8, FU 1, FU 2, FU 3, FU 4, FU 5, FU 6, FU 7, FU 8, PF 1, PF 2, PF 3, PF 6, PF 7 & PF 8 5OOW respectively. No molecules obtained binding score with less than 5 respectively, the values are depicted in Table 10. The molecule PF 5 has good binding score with all three proteins and ND 1, ND 3, ND 5 ND 6 obtained good binding score with 3CWG and 5OOW. The molecule ND 5has highest binding score and is very close with standard Olistat, the interaction with protein 5OOW and hydrogen bonding and other bonding interactions with amino acids are depicted by 3D (Fig. 12) and 2D (Fig. 13) figures.