Identification of ORGs in OLF
The detailed workflow diagram of this study is depicted in Figure 1. After the raw data was processed by R software for background correction and data normalization (Figure 2A), a total of 920 DEGs, consisting of 532 up-regulated genes and 388 down-regulated genes were identified between OLF samples and normal controls (Figure 2B). The clustering heatmap showed that top forty DEGs can clearly distinguish OLF tissues from normal tissues (Figure 2C). After deleting overlapping genes, a total of 2051 obesity-related gene in Homo sapiens were obtained from above four databases (Figure 2D). After intersection of these 920 DEGs and 2051 ORGs, 99 ORDEGs in OLF were ultimately identified (Figure 2E). The two-dimensional PCA depicted a significant difference in these genes to allow further analysis (Figure 2F), including 54 up-regulated genes and 55 down-regulated genes (Figure 2G). The expression profile of top 70 ORDEGs was intuitively visualized through a circular heatmap (Figure 2H).
Analysis of Go Enrichment Functions
The biological functions of up-regulated and down-regulated genes were analyzed separately (Figure 3A, 3H). GO analysis identified that the up-regulated ORDEGs were significantly enriched in BP, including hormone secretion, hormone transport and peptide hormone secretion (Figure 3B, 3C), whereas down-regulated ORDEGs were mainly enriched in regulation of lipid metabolic process, response to lipopolysaccharide and response to molecule of bacterial origin (Figure 3I, 3J). In terms of CC, up-regulated genes were mainly involved in cation-transporting ATPase complex, ATPase dependent transmembrane transport complex and mediator complex (Figure 3D, 3E) while down-regulated ORDEGs were largely enriched in phosphatidylinositol 3-kinase complex, membrane raft and membrane microdomain (Figure 3K, 3L). Moreover, MF demonstrated that up-regulated genes were mainly related to neuropeptide hormone activity, receptor ligand activity and signaling receptor activator activity (Figure 3F, 3G), and down-regulated were mainly enriched in G protein-coupled receptor binding, phosphatase binding and nuclear receptor activity (Figure 3M, 3N). These genes could be related to multiple biological pathways orchestrating OLF pathogenesis.
Analysis of KEGG Enrichment Pathways
Analysis of lists of ORDEGs in terms of enriched biological pathways was also conducted separately for upregulated and downregulated genes. According to the enrichment analysis results of biological pathways, upregulated ORDEGs were mainly involved in collecting duct acid secretion, oxidative phosphorylation, adipocytokine signaling pathway, gastric acid secretion and cytokine-cytokine receptor interaction (Figure 4A, 4B, 4C). The top five KEGG terms among down-regulated ORDEGs were primarily associated with adipocytokine signaling pathway, mTOR signaling pathway, PPAR signaling pathway, insulin signaling pathway and JAK-STAT signaling pathway (Figure 4D, 4E, 4F). Interestingly, in upregulated and down-regulated ORDEGs, there is a common pathway, adipocytokine signaling pathway, which is implicated in multiple biological reactions (Figure 4G). Moreover, we also identified another two remarkable obesity‐related crosstalk pathways, mTOR signaling pathway (Figure 4H) and JAK-STAT signaling pathway (Figure 4I). Dysregulation of module cluster genes might therefore regulate OLF development through acting on these potential pathways, which were potential signatures for OLF.
PPI Network Construction and Hub Genes Selection
To systematically analyze the relationships between the common ORDEGs, we constructed a PPI network using the STRING database after removing unconnected nodes (Figure 5A). Cytoscape visualized the PPI network of ORDEGs, which consisted of 81 nodes and 313 edges (Figure 5B). A total of 14 hub nodes (degree ≥ 30), including AKT1, CCL2, CCL5, CXCL2, ICAM1, IL10, MYC, PTGS2, SAA1, SOCS1, SOCS3, STAT3, TNFRSF1B, VEGFA, were considered hub genes in the OLF DEG list (Figure 5C). Table 1 shows the detailed information and molecular functions of the key genes. In addition, the result of independence testing analysis suggested that all hub genes were significantly decreased in OLF samples except for CCL5 (Figure 5D). Correlation analysis among 14 genes was further conducted to investigate their whole interrelations (Figure 5E). The results showed that SOCS3 and CCL2 had the highest positive correlation with a spearman’s correlation coefficient of 0.99 (Figure 5F). STAT3 was positively correlated with SOCS3 (r = 0.98, Figure 5G) and VEGFA (r = 0.98, Figure 5H). VEGFA was also positively correlated with CXCL2 (r = 0.98, Figure 5I).
Analysis of the Functional Characteristics of Hub Genes
The analysis results from the GeneMANIA database showed that 14 hub genes and their co-expressed genes constitute a complex PPI network with co-expression of 65.62%, genetic interactions of 15.03%, physical interactions of 8.28%, shared protein domains of 3.99%, pathway of 3.86% and co-localization of 3.21%, whose functions were mainly associated with cellular response to biotic stimulus, regulation of inflammatory response, regulation of tyrosine phosphorylation of STAT protein (Figure 5J). Furthermore, Metascape functional annotation results revealed that hub genes were mainly enriched in response to lipopolysaccharide, positive regulation of cell adhesion, cellular response to biotic stimulus, T cell activation, receptor signaling pathway via JAK-STAT and receptor signaling pathway via STAT (Figure 6A, 6B). Meanwhile, ClueGO revealed that the most involved pathways were JAK-STAT signaling pathway, cytokine-cytokine receptor interaction, adipocytokine signaling pathway, and chemokine signaling pathway (Figure 6C, 6D, 6E). By this token, immune-related or inflammation-related biological responses and pathways were strongly linked to these hub genes in OLF. Therefore, further subgroup analysis was conducted to separately investigate possible immune functions of these genes, and T cell activation (55.56%), cellular response to interferon-gamma (22.22%), lymphocyte migration (11.11%) and mononuclear cell differentiation (11.11%) were identified as the potential immune responses involved (Figure 6F, 6G, 6H) .
Correlation Analysis of Hub Genes and Infiltrating Immune Cells
The above results demonstrated that the hub genes were also highly enriched in immune-related or inflammation-related responses and pathways. Our previous study had identified 14 types of OLF-related infiltrating immune cells (OIICs) using ssGSEA and xCell algorithm based on the gene expression matrix from the GSE106253 (Figure 7A). To explore the underlying mechanisms associated with these potential biomarkers, we estimated the correlations between these genes and infiltration of immune cell types in OLF samples. According to r > 0.90 and p < 0.001, 27 ORDEG-OIICs correlation pairs were screened, and cDCs were significantly associated with the most ORDEGs (Figure 7B). For example, STAT3, SOCS3 and VECFA, TNFRSF1B were negatively correlated with cDCs (r = −0.95; r = −0.96; r = −0.99; r = −0.95) (Figure 7C–F); MYC, IL10 and CCL2 were negatively correlated with memory B-cells (r = -0.97; r = -0.95; r = -0.95) (Figure 7G-I). In addition, CCL2 and MYC were positively correlated with preadipocytes (r = 0.96; r = 0.95) (Figure 7J-K); VEGFA and CXCL2 were positively correlated with NK CD56 bright cells (r = 0.96; r = 0.99) (Figure 7L-M). These results indicated a strong correlation between the seven ORDEGs and immune cells.