Air pollution has become a severe environmental problem all over the world. In 2019, air pollution became the fourth leading global risk factor for death, surpassing other recognized risk factors for chronic diseases such as obesity, high cholesterol, and malnutrition (HEI, 2020). The annual mean guideline level of PM2.5 has been modified from 10 µg/m3 in 2005 AQG (WHO Air Quality Guidelines) to 5 µg/m3 in 2021 AQG (WHO, 2021b), indicating the urgency and necessity to investigate the adverse effects and related mechanism induced by PM2.5 exposure. This study explored the changes in the liver of obese mice at the transcriptome level under the cross-intervention of PM2.5 and metformin. Through the analysis of 12 hub genes, we discussed the pathophysiological and functional changes that may occur in the liver, especially the physiological processes and diseases associated with the liver, such as thyroid function, insulin resistance and lipid metabolism; NAFLD, HCC, even neurodegenerative diseases. The study could also provide some biomarkers for early screening of diseases caused by PM2.5.
GO function analysis (Fig. 1) showed that the most obvious term in molecular function is ‘protein binding’. Impairment of liver function may not only disturb liver metabolism, but also affect plasma protein binding, which in turn affects the distribution and removal of metabolites in the body (Verbeeck, 2008). What needs attention in cellular component was nucleus, cytosol and cytoplasm, which may be related to cytosol-nucleus traffic and colocalization in hepatocytes (Romanque et al., 2011). Transcription and metabolism were more important in biological processes. This may be because the liver uses a series of liver transcription factors to regulate the expression of genes involved in all aspects of lipid metabolism (including catabolism, transportation and synthesis) (Karagianni and Talianidis, 2015).
According to the KEGG pathway analysis, PM2.5 could affect the thyroid hormones signal pathway with most statistically different, even thyroid cancer (Fig. 2). Thyroid hormones (THs) had a significant effect on the anabolism of fatty acids and cholesterol in the liver, and direct regulate de no lipogenesis, tricarboxylic acid cycle (TCA), fatty acid β-oxidation, OXPHOS, lipolysis and lipophagy pathway, which mainly involves genes, such as acc1, me, fasn, thrsp, cpt1a, pdk4, mcad, ucp2, hmgcl, atgl (Sinha et al., 2018). It was reported that low thyroid function in population is associated with increased likelihood of chronic fibrotic diseases of the liver (Bano et al., 2020). Population studies found that prenatal exposure to PM2.5 can damage neonatal thyroid function (Ghassabian et al., 2019). Our research found that PM2.5 up-regulates the thyroid signaling pathway in the liver, which was also proved from the level of metabolic organs (Kim et al., 2020). The risk of NAFLD was inversely correlated with free thyroxine levels(Ritter et al., 2020). THs modulated the homeostasis of hepatic lipid metabolism by regulating lipoprotein, triglyceride (TAG) storage and cholesterol levels, which had a key effect on liver-related diseases, such as NAFLD and hypercholesterolemia(Martínez-Sánchez et al., 2017). And THs may modulate co-activators and co-repressors through the hypothalamic-pituitary-thyroid axis, thereby altering cholesterol metabolism in the liver (Ritter et al., 2020).
In addition, our results showed that the pathway with the most DEGs enrichment was the ‘metabolic pathways’ (Fig. 2). The liver played an important role in glucose homeostasis by controlling various pathways of glucose and lipid metabolism, including oxidation, gluconeogenesis and adipogenesis (Han et al., 2016). It also regulated other important metabolisms, including purines and pyrimidine synthesis, histidine catabolism, methionine recycling and formic acid utilization (Zaitsev et al., 2019). The results showed that the enriched related pathways were AMPK signaling pathway, PPAR signaling pathway, insulin resistance and hepatocellular carcinoma. These pathways were involved in the pathophysiological process of oxidative stress, inflammation, abnormal metabolism leading the accumulation of glucose and lipid in the liver (Xu et al., 2019). The excessive production and accumulation of hepatic lipid might induce liver fibrosis in further, which was in line with the evolution of NAFLD. Animal studies have confirmed that PM2.5 can induce excessive extracellular matrix accumulation in liver tissues and eventually lead to liver fibrosis, which was a foreshadow to liver cancer (Zheng et al., 2015). A prospective epidemiological study in U.S. shown that environmental PM2.5 exposure may be a risk factor for HCC (VoPham et al., 2018). So, the disturbance of metabolic pathways might be the first step of PM2.5-induced liver injury and the long-term hazardous hepatic effects of PM2.5 exposure could be overwhelming.
In order to understand the changes in DEGs, we conducted a trend analysis and STC-GO analysis and obtained two reasonably interpretable trends (Fig. 3 and Fig. 4). As can be seen from the plot, the genes in Trend 6 increased after being exposure and decreased with metformin (Fig. 4A and C). The term ‘GO 0019322: pentose biosynthetic process’ has the most significant difference. Study observed an increasement in the oxidative branch of the pentose phosphate pathway and 13C incorporations suggestive of enhanced capacity for the de novo synthesis of fatty acids, which indicates an increase in insulin resistance (Reyes-Caballero et al., 2019). On the other hand, metformin can relieve the pentose phosphate pathway, inhibit gluconeogenesis and promote glycogen retention to reduce insulin resistance (Atangwho et al., 2014). Another term ‘ KO 04932: Non-alcoholic fatty liver disease (NAFLD)’ also involved gluconeogenesis, glycogen and insulin resistance, which was in accordance with the above research results. We also found other interesting pathways, such as ‘KO 05150: staphylococcus aureus infection’, ‘GO 0046942: carboxylic acid transport’, ‘GO 0007188: adenylate cyclase-modulating G protein-coupled receptor signaling pathway’. And there were also many articles confirmed that these terms are related with liver or PM2.5 exposure. For example, monocarboxylate transporter 1 (MCT1) expression was down-regulated in adipocytes of diabetic rats thus to impair the ability to transport lactic acid (Hajduch et al., 2000); metastasis and glycolysis could be induced by up-regulation of MCT1 expression and subsequently activating Wnt/β-catenin signaling pathway in HCC (Fan et al., 2018).
The expression of Trend 15 genes only increased significantly after PM2.5 exposure, and metformin intervention does not change the rising trend (Fig. 4B and D). The terms ‘WP 447: Adipogenesis genes’ and ‘GO 0046890: regulation of lipid biosynthetic process’ are noteworthy and need further study. PM2.5 exposure could increase the expression of genes related to lipid synthesis through different mechanism. For example, srebp1 was involved in regulating the expression of fasn, acc and scd1; exposure to PM2.5 resulted in increased expression of bmal1, rev-erbα and pparα, affecting circadian rhythm, liver triglyceride, free fatty acid levels, or fatty acid transport (Yan et al., 2020). It can be seen from the network diagram that these interesting gene groups we discussed are not adjacent. The other genes linking them are worthy of attention and further research.
In WGCNA analysis, the red module (mup6, mup8) is mainly related to lipid metabolism, oxidative stress and inflammation; The darkslateblue module (cd53, fcer1g, cd68, ctss, laptm5) is mainly related to cell activation, innate immune system and atherosclerosis; The darkmagenta module (sub1, snrpd2, zfp931, etohi1) is mainly related to transcriptional regulation, mRNA splicing and gene expression; The antiquewhite4 module (egln1) is mainly related to cellular oxygen sensor that catalyzes, under normoxic conditions (Fig. 6). More specifically, The cd68 was a surface marker for M1 macrophages and it was involved in liver damage such as inflammation, liver fibrosis and HCC (Liu et al., 2019). Shi et al. reported that PM2.5 up-regulated the expression of CD68 both in cell model and in lung tissues (Shi et al., 2019), indicating PM2.5 promoted the pro-inflammatory transformation of macrophage thus inducing tissue damage. The included cd68 and cd53 are related to liver inflammation and insulin sensitivity (Ehses et al., 2009). Inhibition of the expression of the cd family may be a therapeutic target for HCC. Additionally, PM2.5 exposure could increase the glucocorticoids in plasma by reducing the expression of glucocorticoid receptors in the hippocampus, thereby activating the inflammatory response and inducing neurotoxicity (Jia et al., 2018) cd68, ctss, laptm5, fcgr3a and cd53 were related to the regulation of microglia polarization and can detect out neuropathic pain early (Yu et al., 2020). On the other hand, PM2.5 has been confirmed in the population to cause neurodegenerative diseases such as Parkinson's syndrome, even if the concentration is lower than the current American national standard (Liu et al., 2016); animal experiments have shown that PM2.5 may aggravates Parkinson's disease via inhibition of autophagy and mitophagy pathway (Wang et al., 2021). It has also showed that ctss, cd53, igsf6, ptprc and laptm5 may be potential pathological target gene for the Parkinson’s syndrome, which is highly similar to our darkslateblue module (Cui et al., 2015). We can infer that darkslateblue module can be used as biomarker for neurodegenerative diseases such as Parkinson's disease. It is important that the genes in darkslateblue module increase after PM2.5 exposure, but decrease with metformin. These potential biomarkers might be helpful for the prediction and early screening of these related diseases.