Over the past 40 years, the incidence of obesity has increased in both developed and developing countries because of unbalanced diets, inadequate physical activity, chronic stress, and certain drugs and other environmental pollutants (Andersen et al., 2016; Yang et al., 2017; Sung et al., 2019; Tomiyama, 2019). According to the latest report of the World Health Organization, more than 1.9 million people were overweight in 2016, of which 6.5 million were obese (NCD Risk Factor Collaboration [NCD-RisC], 2017; WHO, 2018). Obesity can cause many chronic diseases, including cardiovascular and cerebrovascular diseases, diabetes, and some cancers of the colon, rectum, gastric cardia, liver, gallbladder, and pancreas (Arnold et al., 2016; Lauby-Secretan et al., 2016; Gadde et al., 2018). The most characteristic pathological changes associated with obesity are abnormal and excessive fat accumulation (WHO, 2018). Adipose tissue is widely distributed throughout the human body, and regulates metabolism by storing and releasing fatty acids and secreting adipokines (Giralt and Villarroya, 2013; Chait and den Hartigh, 2020). In obese people, this regulatory ability is disrupted, as is adipokine release (Munzberg and Myers, 2005; Leal and Mafra, 2013). Current treatment strategies for obesity includes inhibition of fat absorption, liposuction, and weight loss surgery (Yanovski and Yanovski, 2014; Apovian et al., 2015; Wolfe et al., 2016). However, these are symptomatic treatments rather than preventative measures controlling obesity development, and the number of obese individuals continues to increase annually. Therefore, it is important to identify biomarkers of obesity in adipose tissue and their related molecular mechanisms, to develop effective therapies to prevent obesity and its complications.

With the growing popularity of chip and sequencing technology, comprehensive bioinformatic analysis has become a promising method for exploring disease-related biomarkers and their molecular mechanisms (Sakharkar et al., 2019; Wan et al., 2020). For example, Zhang et al. found that the adipose tissues of insulin-resistant and non-insulin-resistant patients with obesity had differential expression of a number of genes, including matrix metallopeptidase 9, interleukin (IL)6, C-X-C motif chemokine ligand (CXCL) 8, C-C motif chemokine ligand 4, and CXCL10, which are enriched for several functions, including cytokine-cytokine receptor interactions, tumor necrosis factor (TNF) signaling, and pathways in cancer (Zhang et al., 2019). Ida et al. found that ST8 alpha-N-acetyl-neuraminide alpha-2,8-sialyltransferase 2 expression in omental fat was significantly associated with weight loss after Roux-en-Y gastric bypass (Hatoum et al., 2013). In addition, circadian clock and metabolic gene expression rhythms are reduced in the subcutaneous adipose tissue of obese patients with type 2 diabetes compared to lean controls (Stenvers et al., 2019). However, no large-scale studies have been performed to examine the differential expression of genes in the adipose tissues of obese and non-obese populations. In this study, we used an integrative bioinformatics approach to analyze immune cell differences and gene expression differences in adipose tissue from obese and non-obese patients. Moreover, we combined the analysis of differentially expressed genes (DEGs) with functional enhancement and protein-protein interaction (PPI) analysis. Finally, we identified the hub genes that may contribute to obesity.

Globally, the incidence of obesity is continuously increasing. The harms and economic losses caused by obesity are significant. In 2014, economic losses directly or indirectly caused by obesity were as high as $2.0 trillion (Woetzel et al., 2014). The main feature of obesity is the accumulation of excessive adipose tissue, which can induce several diseases via oxygen deprivation, abnormal cytokine secretion, and interference with glycolipid metabolism (Chait and den Hartigh, 2020). Therefore, it is important to explore gene expression changes and associated molecular mechanisms in the adipose tissue of obese individuals. As the different estrogen levels and neural underpinnings affect body adipose tissue levels and appetite, obesity rates are higher in women than in men in most countries and regions, with an average prevalence of 10 and 18% in men and women, respectively (Kroll et al., 2020). Because of this, many studies have examined male and female obesity separately (Garawi et al., 2014). However, other studies have shown that variants of certain genes, including FTO alpha-ketoglutarate dependent dioxygenase, retinoic acid induced 1, MAGE family member L2, and melanocortin 4 receptor contribute to obesity development (Fall and Ingelsson, 2014; Poitou et al., 2020). To identify genes and pathways that cause obesity in everyone, in this study we analyzed two datasets, GSE70353, containing samples from men in Northern Finland with an average age of 54, and GSE59034, containing samples from women in Sweden with an average age of 45. We analyzed their DEGs separately, then performed overlap analysis to obtain a list of DEGs that were meaningful to all. A total of 64 upregulated and 49 downregulated genes were identified. Upregulated DEGs were associated with many diseases, including liver cirrhosis, pulmonary fibrosis, asthma, hypertension, rheumatoid arthritis, and diabetes. This evidence and epidemiological data provide mutual verification of the harmful effects of obesity (Chooi et al., 2019). These DEGs may represent key targets for the treatment of obesity and its complications.

Functional enrichment analyses revealed that the upregulated genes were mainly involved in the enhancement of inflammatory signals, including hypoxia-inducible factor 1, chemokine, and TNF signaling pathways. The increased oxygen consumption and the increasing size of adipocytes in turn leads to the occurrence of adipose tissue hypoxia (Lee et al., 2014). This may further induce increased production of adipokines such as chemokines, which increase myeloid leukocyte and neutrophil migration leading to chronic inflammation of adipose tissue (Krinninger et al., 2014). Downregulated genes were mainly involved in the metabolism of various nutrients, including fructose and mannose, propanoate, and glycerophospholipid metabolism, as well as amino acid biosynthesis (Shi et al., 2020). This indicates decreased metabolic function and nutrient dysregulation in the adipose tissue of obese people (Rogero and Calder, 2018). We also used GSEA to verify and supplement the DEG enrichment results. Pathways positively related to obesity included inflammatory infection, the complement cascade, and the toll-like receptor signaling pathway, which indicated that the complement system and toll-like receptors are the main factors of inflammation in adipose tissue during obesity (Rogero and Calder, 2018). Pathways negatively related to obesity included adipocytokine signaling, insulin signaling, and fatty acid metabolism, which reflect a dysregulation of adipokines and insulin signaling in obese patients. As enrichment analysis indicated inflammation-related signal upregulation in the adipose tissue of obese people, we used the CIBERSORT algorithm to detect whether there were changes in the immune cell composition between tissues from the two groups. It revealed increases in M0, M1, and M2 macrophages in the tissues of obese people, which may be the main reasons of adipose tissue inflammation. It also showed decreases in T cells, including CD4, CD8, and γδ types. Hu et al. showed that T cells (especially γδ T cells) can promote sympathetic innervation by driving the expression of transforming growth factor β1 in parenchymal cells via IL17 receptor C (Hu et al., 2020). Interestingly, recent studies have also reported fewer circulating γδ T cells in obesity (Li et al., 2020), which reduces the ability to fight viral infections, leading to a poorer prognosis for obese people than for lean people after infections such as severe acute respiratory syndrome coronavirus 2 (Jouan et al., 2020). Strategies to increase γδ T cell activation in these patients may be a promising direction for future research.

To identify the most critical hub genes among the 113 DEGs, we constructed a PPI network to identify the most interconnected genes. The top 10 hub genes were all upregulated. CD53 is the most connected protein in the PPI network. It is mainly expressed on the membranes of immune cells, where it helps regulate many of their functions, including adhesion, migration, and cell fusion (Dunlock, 2020), and plays an important role in antigen presentation. Many studies have shown that CD53 is increased in obese and inflammatory tissues, and regulating its expression may be an effective treatment for obesity with complications (Nair et al., 2005; Zhao et al., 2020). PLEK is the major protein kinase C substrate phosphorylated in diabetic macrophages, where a 30% reduction in PLEK can inhibit TNFα secretion by 80% (Ding et al., 2007). CTSS can promote adipocyte differentiation by degrading fibronectin. In vitro studies have also shown that the inflammatory factors TNFα and IL1β can induce cathepsin secretion in adipose tissue (Hooton et al., 2012). The expression of cathepsin mRNA in the adipose tissue of obese subjects was two times higher than that of healthy-weight subjects and decreased after weight loss, which shows that cathepsins can be effective markers of obesity (Naour et al., 2010).

ITGB2 is positively correlated with obesity (Taleb et al., 2005; Pei et al., 2014; Imaizumi et al., 2018). It plays an important role in the immune response, and defects in ITGB2 cause leukocyte adhesion deficiency (Ng et al., 2015). TYROBP is strongly related to insulin resistance and is mainly involved in the regulation of immune responses and integrin-mediated signaling (Das et al., 2015). C3AR1 is a complement receptor that is expressed in both adipocytes and macrophages, and is increased by a high-fat diet (Mamane et al., 2009). C3AR1 activation can enhance lipolysis induced by adrenaline, which may be a good way to treat obesity (Cero et al., 2017). C1QB and C1QC belong to a family of complement proteins that are positively correlated with immune cell chemotaxis and can regulate the expression of the immune cells. Recent findings have revealed that C1Q-related gene knockouts can prevent obesity-related complications of Alzheimer’s disease (Graham et al., 2020). FCGR2B and FCER1G are immunoglobulin fragments that regulate the immune response, and increase the likelihood of obese patients developing allergic diseases (Mathur et al., 2013). Shu et al. (2018) showed that FCGR2B promotes lipid accumulation and gluconeogenesis in hepatocytes.

In summary, CD53, PLEK, CTSS, ITGB2, TYROBP, C1QB, C1QC, C3AR1, FCGR2B, and FCER1G were all highly expressed in the adipose tissue of obese people. They all contribute to obesity and its complications, primarily through immune inflammation and glucolipid metabolism-related effects. In addition, the problem of environmental pollution is becoming increasingly serious, and many fat-soluble chemicals can accumulate in adipose tissue and interfere with gene expression, leading to obesity. The CTD database indicated that cadmium chloride, parathion, tetrachlorodiben, zodioxin, and bisphenol A may cause obesity by interfering with the expression of these hub genes. Therefore, it is important to reduce exposure to these chemicals. A limitation of this study is that it is strictly based on bioinformatic analysis, and the results will require further verification in vitro and in vivo.

In conclusion, this is the first large-scale study to examine differential gene expression in the adipose tissues of obese and non-obese populations. Our study provides several potential biomarkers and related molecular mechanisms for obesity and its complications, and lays a foundation for the exploration of new targeted treatments.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

ZL wrote the article. LM designed the study. ZS and XS were responsible for data collection and analysis. WS and YZ were responsible for data analysis and mapping. JS and XY revised and critically reviewed the article. All authors have agreed on the journal to which the article has been submitted, agreed to be accountable for all aspects of the work criteria, and made a significant contribution to the work reported.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.