Three-week-old female C57BL/6 mice were purchased in Beijing from Vital River Company and acclimated for 7 days after arrival. As per our previous protocol (Li et al., 2019; Jiang et al., 2020), the mice were randomly assigned to the filtered air (FA) and particulate matter (PM) groups in the real-ambient exposure system during winter in Shijiazhuang city (n = 10). Briefly, individually ventilated cages (IVCs) were linked directly to filtered air (FA group) or ambient outdoor air (PM group) after removing PM2.5 in ambient indoor air by three layers of high-efficiency particulate air (HEPA) filters. During the winter, the mice were exposed for 16 h per day (from 7:00 a.m. to 11:00 p.m.) for 4 or 8 weeks, and the mice were not restricted from eating or moving around. The experimental conditions were a 12-h daylight/12-h dark cycle. The room temperature was kept between 22 and 24°C. The experimental protocol was approved by the Animal Care and Use Committee of Qingdao University.

After 4 or 8 weeks of PM2.5 exposure, inferior vena cava blood collection, and the serum was collected after centrifugation. The liver, gWAT and sWAT were collected and immediately fixed in 4% PFA or frozen at −80°C.

The liver, gWAT and sWAT were fixed in 4% PFA, paraffin-embedded, sectioned, and stained with hematoxylin-eosin (H&E) to observe the histomorphology. Mitochondria-related genes, including CPT1 (catalog #ab128568; Abcam), CPT2 (catalog #ab181114; Abcam), PGC1α (catalog #2178s; Cell Signaling Technology, Danvers, MA) and COX4 (catalog #11967; Cell Signaling Technology, Danvers, MA), were examined by immunohistochemistry and immunofluorescence. Briefly, the paraffin-embedded tissues were sectioned, deparaffinized with xylene, dehydrated with ethanol, blocked with 10% FBS, incubated with primary and secondary antibodies, and stained with DAB substrate solution. Finally, under a microscope, the color of the antibody staining in the tissue sections was observed.

TRIzol reagent was utilized to extract total RNA from liver, gWAT, and sWAT tissues according to the user’s manual. Reverse transcription experiments were performed to obtain cDNA templates according to the instructions (Accurate Biology, Hunan), and further real-time quantitative PCR experiments were carried out to ascertain the expression of target genes using the QuantStudio Q7 RT–PCR system. The 2^−ΔΔCT approach was used to calculate the target molecule’s expression level relative to β-actin. In Supplementary Table 1, the primer sequences applied in the studies are listed.

According to the user’s manual, the liver was homogenized in ethanol, and lipids were extracted from the supernatant after centrifugation. TG, TC and FFA levels in the liver and serum were measured using Nanjing Jiancheng kits, and VLDL levels in the liver and serum were measured using enzyme-linked immunosorbent assay (ELISA) kits (Fankew, Shanghai). After incubation for the indicated time, the absorbance was detected by a microplate spectrophotometer, and the concentration was calculated based on a standard curve.

The metabolites in serum and liver were analyzed by extraction, computer detection and statistical analysis. First, 25 mg liver was added to 500 μL of extract solution (methanol:acetonitrile:water = 2:2:1) and 100 μL of serum sample was added to 400 μL of extract solution (methanol:acetonitrile = 1:1), both of which contained the isotopically labeled internal standard mixture. Mix well and sonicate for 10 min under ice-water bath conditions, noting that the liver needs to be processed by grinding. The supernatant solution was obtained by centrifugation at 12,000 rpm for 15 min at 4°C after 1 hour at −40°C. The target compounds were chromatographically separated on a Waters ACQUITY UPLC BEH Amide (2.1 mm × 100 mm, 1.7 μm) column using a Vanquish (Thermo Fisher Scientific) ultra-performance liquid chromatograph. The A-phase of liquid chromatography was aqueous containing 25 mmol/L ammonium acetate and 25 mmol/L ammonia, and the B-phase was acetonitrile. The data collection was performed on a QE HFX mass spectrometer (Orbitrap MS, Thermo) under the control of acquisition software (Xcalibur, Thermo). The raw data were also processed using ProteoWizard software.

The adipose tissue was ground, and water and methanol solution were added to the homogenate at a ratio of 2:1:4 to extract lipid components. To enable subsequent measurement, 250 mL of methyl tert-butyl ether (MTBE) comprising lipid standards was added after vortex mixing. The samples were bathed in 10°C water for 30 min (ultrasonic frequency 40 kHz, power 100 W). The organic phase was then separated by centrifugation at 10°C (1,400 × g) for 10 min. LC-MS analysis was performed using ultra-high pressure liquid chromatography combined with a Sciex Triple TOF 6600 liquid chromatograph. The Thermo Hypersil Gold C18 column (100 mm × 2.1 mm, 1.9 μm) was used. The mobile phase A was water-acetonitrile (6:4 v/v, containing 0.1% formic acid and 10 mmol ammonium formate), and the mobile phase B was acetonitrile-isopropanol (1:9 v/v, containing 0.1% formic acid and 1 mmol ammonium formate). Data analysis was performed using Peak view software to analyze lipid composition.

Total RNA was extracted from the liver, and the quality of the RNA was examined for concentration and degradation. mRNA was enriched with oligo-(dT) beads, and subsequently, cDNA was generated with poly (A) added to the 3′ end. After purification, sequencing adapters were added to the QiaQuick PCR kit, agarose gel electrophoresis was used to assess the size of the fragments, and PCR was performed to amplify the cDNA. After library construction, the quality inspection of the library was qualified, and PE150 sequencing was performed using the Illumina NovaSeq platform. KOBAS software was used to perform a KEGG enrichment analysis.

GraphPad Prism 8.3.0 software was used for the statistical analysis of the experimental data, and the unpaired two-tailed t-test was run. Error bars represent ± SEM, and statistical significance was considered when the p value was less than 0.05. *, p < 0.05, **, p < 0.01, and ***, p < 0.001. Regarding metabolomics analysis, after preprocessing the data, a set of observed potentially correlated variables was transformed into linearly uncorrelated variables by orthogonal transformation using principal component analysis (PCA). Using SIMCA software (V16.0.2, Sartorius Stedim Data Analytics AB, Umea, Sweden), the data were formatted by logarithmic (LOG) transformation plus centralization (CTR), and then automatic modeling analysis was performed to obtain scores indicating the first and second ranked principal components, respectively [PC (1) and PC (2)]. The orthogonal partial least squares-discriminant analysis (OPLS-DA) statistical method was also used to filter out orthogonal variables in metabolites that were not correlated with categorical variables, and non-orthogonal and orthogonal variables were analyzed separately. Thus, a more reliable letter of correlation between group differences of metabolites and experimental groups was obtained. Finally, SIMCA software was used to obtain the Variable Importance in the Projection (VIP) by logarithmic (LOG) transformation plus UV formatting of the data. VIP is the importance of the variables to the model, which describes the overall contribution of each variable to the model, and is usually set at a threshold value of VIP > 1, p value < 0.05.

To examine the adverse effect of PM2.5 on adipose tissue, young mice were exposed to the real-ambient exposure system in Shijiazhuang during the winter of 2019. During the exposure period, the daily average ambient PM2.5 concentration reached 140.8 μg/m³ (Ji et al., 2022). After exposure for 4 or 8 weeks, the sWAT and gWAT were isolated and weighed. Interestingly, after 8 weeks of exposure, the gWAT and sWAT mass increased significantly in the PM group compared with the FA group (Figures 1A, B). Considering the close crosstalk between liver and adipose tissue (Scheja and Heeren, 2016), the liver weight was evaluated as well. Conversely, the liver weight was markedly reduced in the PM group compared with the FA group (Figure 1C). Because of the contribution of hypertrophic adipocytes to the growth of adipose tissue (Ghaben and Scherer, 2019), hematoxylin-eosin (H&E) staining was used to determine the adipocyte size in sWAT and gWAT. Consistently, after exposure for 8 weeks, the adipocytes of the sWAT in the PM group were considerably larger than those in the FA group, accompanied by a slightly increased adipocyte size in the gWAT in the PM groups (Figure 1D). Furthermore, the histomorphology of the liver was detected with H&E staining. However, the morphology was similar in the FA and PM groups (Figure 1D). Considering the pivotal role of fatty acids during the crosstalk of adipose tissue and liver (Fuchs et al., 2022), the levels of free fatty acids (FFAs) and very-low-density lipoproteins (VLDLs) in the liver and serum were measured by ELISA. We found a trend of higher levels of FFAs in the liver and lower levels in the serum in the PM group (Figure 1E, G). Importantly, the level of serum VLDL was strikingly increased in the PM group, accompanied by a decreasing trend in the liver, indicating the influx of VLDLs from hepatocytes to serum (Figure 1F, H). Taken together, our findings showed that chronic exposure to PM2.5 exacerbated the expansion of fat tissue.

It is generally known that lipid metabolism regulates the growth of adipose tissue (Samuel and Shulman, 2016). After exposure for 8 weeks, lipidomic analysis was performed in gWAT and sWAT. Based on PLS-DA, the lipid signature of gWAT and sWAT in the PM group was considerably distinct from that of the FA group (Figures 2A, D). According to volcano plots, some notably changed lipid species were observed in gWAT and sWAT (Figures 2B, E). To further identify the specific lipid species that were affected by PM2.5, all significantly altered lipid species were visualized by heatmaps. A total of 18 lipid species markedly changed in gWAT (Figure 2C), while nine species considerably changed in sWAT (Figure 2E). The lipid classes included diacylglycerol (DG), triacylglycerol (TG), acyl-carnitine coenzyme Q (CoQ), phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS), sphingomyelin (SM), phosphatidylethanol (PET), lyso-phosphatidylethanolamine (LPE), lyso-phosphatidylcholine (LPC) and lyso-phosphatidylglycerol (LPG) (Figure 2F).

The liver plays a crucial role in the expansion of adipose tissue by regulating lipid influx, and metabolites usually circulate in the bloodstream to realize crosstalk among tissues (Scheja and Heeren, 2016). Therefore, the metabolites in the liver and serum were examined with metabolomics analysis. Consistently, based on OPLS-DA, the metabolomic profiling of liver and serum in the PM group was notably distinct from that in the FA group (R²X = 0.36, R²Y = 0.989, Q² = 0.211 in Figure 2G, R²X = 0.407, R²Y = 0.882, Q² = 0.0422 in Figure 2J) (Figures 2G, J). Based on volcano and clustering heatmaps, when compared to the FA group, various metabolites in the liver and serum displayed significant differences in the PM group (Figures 2H, K). Consistently, the level of TG in the liver was increased after exposure to PM2.5 (Figure 2I). Glutathione is believed to be a key regulator in fat burning and storage (Rom et al., 2020). Glutathione was altered in the liver during PM2.5 exposure, accompanied by significant downregulation of the unsaturated fatty acid but-2-enoic acid (Figure 2I). At the same time, we also found that PM2.5 exposure had an effect on glucose metabolism, and its key metabolite glucose 6-phosphate was significantly reduced at the end of the exposure (Figure 2I). 12-Hydroxyeicosatetraenoic acid (12-HETE) and leukotriene B5 are considered inflammation markers (Maaløe et al., 2011; Cheng et al., 2019). Both of these metabolites notably increased in serum after PM exposure (Figure 2L). Taken together, these results revealed glycolipid metabolic remodeling in adipose tissue and liver when exposed to PM2.5.

To investigate the underlying molecular mechanism, the liver was subjected to transcriptomic analysis. KEGG enrichment analysis revealed the top 20 upregulated and downregulated signaling pathways in the liver. Intriguingly, the notably changed signaling pathways included metabolic pathways, fatty acid metabolism, fatty acid degradation, fatty acid elongation, biosynthesis of unsaturated fatty acids, oxidative phosphorylation (Figure 3A). These pathways are well known to be involved in regulating lipid metabolic homeostasis. Besides, in agreement with the results of the metabolomic analysis of this study, glycolysis/gluconeogenesis, and pentose and glucuronate interconversion pathways were remarkably altered (Figure 3A).

Acetyl-CoA carboxylase 1 (ACC1), acetyl-CoA carboxylase 2 (ACC2), fatty acid synthetase (FAS), sterol regulatory element-binding transcription factor 1 (Srebp1c), and stearoyl-coenzyme A desaturase 1 (SCD1) participate in the de novo lipogenesis (DNL) pathway. Peroxisome proliferator-activated receptor alpha (PPARα) plays a key role in lipid oxidation. Compared to the FA group, the mRNA levels of FAS and SCD1 were dramatically reduced (Figure 3B), whereas the mRNA levels of PPARα significantly increased in PM group after exposed for 8w (Figure 3C). Hydroxy acyl-CoA dehydrogenase (Hadh), acyl-CoA dehydrogenase (Acad), enoyl-CoA hydratase and 3-hydroxy acyl CoA dehydrogenase (Ehhadh) and cytochrome P450, family 4, subfamily a (Cyp4a) are involved in lipid oxidation (Hoek-van den Hil et al., 2013; Ipsen et al., 2018). These genes were found altered by transcriptome analysis, and were further validated with qPCR. Consistently, Hadh, Hadhb, Ehhadh, Acad long chain (Acadl) and Acad short chain (Acads) were greatly enhanced in the liver in the PM group (Figure 3C). In addition, the expression of metabolism-related genes was examined in gWAT and sWAT. The findings revealed that after exposure to PM2.5, the levels of FAS and SCD1 were considerably increased in gWAT and sWAT, accompanied by higher levels of ACC1 in sWAT (Figures 3D, E). These results suggested that PM2.5 exposure disturbed signaling pathways involved in metabolism in the liver and adipose tissue.

The primary function of mitochondria is maintaining energy homeostasis by breaking down lipids and other molecules (Thorp, 2019). Therefore, mitochondrial activity was evaluated after PM2.5 exposure. After 8 weeks of exposure, the mtDNA copy number were comparative in FA and PM groups in sWAT, gWAT and liver based on qPCR assays (Figure 4A). Carnitine palmitoyltransferase (CPT1, CPT2), as a mitochondrial enzyme, is critical in fatty acid oxidation (FAO) (Nassir and Ibdah, 2014). Peroxisome proliferator-activated receptor-gamma coactivator 1α (PGC1α) is considered a primary regulator of mitochondrial biosynthesis (Li et al., 2016; Jamwal et al., 2021). The levels of CPT1, CPT2, and PGC1α were investigated using qPCR and IHC staining in liver, gWAT, and sWAT. The levels of CPT2 and PGC1α were considerably increased in sWAT, and the mRNA expression of CPT1, CPT2 and PGC1α was considerably increased in the liver (Figures 4B–D). Consistently, the protein levels of these genes markedly increased in the gWAT, sWAT and liver after exposure for 8 weeks (Figure 4E). As a marker of mitochondria, cytochrome c oxidase subunit 4 (Cox4) was determined by immunofluorescence labeling. We found increased expression of Cox4 in gWAT and sWAT after exposure to PM2.5 (Figure 4E). In addition, the mRNA expression of Cox4 was significantly elevated in the liver after exposed to PM2.5 for 8 weeks (Figure 4D). These findings revealed that the number and activity of mitochondria increased during PM2.5 exposure.