Exposure to Ambient Particulate Matter Induces a NASH-like Phenotype and Impairs Hepatic Glucose Metabolism in an Animal Model

Exposure of animals to ambient PM_2.5

Mice were exposed to concentrated ambient PM_2.5 or filtered air (FA) in a mobile trailer “Ohio’s Air Pollution Exposure System for the Interrogation of Systemic Effects (OASIS)-1” in Columbus, OH, where most of the PM_2.5 components are attributed to long-range transport (Supplemental figure 1) [1]. The concentrated PM_2.5 was generated using a versatile aerosol concentration enrichment system (VACES) as described previously [1, 11]. Mice were exposed to concentrated PM_2.5 at nominal 10 × ambient concentrations for 6 hours per day, 5 days per week for a total of 3 or 10 weeks, as detailed previously [1, 9]. The control (FA) mice in the experiment were exposed to an identical protocol with the exception of a high-efficiency particulate-air filter positioned in the inlet valve position to remove PM_2.5 particles in the filtered air stream.

For a full description of materials and methods, see Supplemental Information.

Mice after inhalation exposure to PM_2.5 develop a NASH-like phenotype

To elucidate in vivo effects of PM_2.5 exposure, male C57BL/6 mice were exposed to concentrated ambient PM_2.5 or filtered air (FA) for 3 or 10 weeks (5 days/week, 6 hours/day) in exposure chambers of “OASIS-1”, which was composed of the Midwestern regional background in Columbus, USA, where most of the PM_2.5 is attributed to long-range transport (Supplemental figure 1) [9]. It has been demonstrated that the distribution and size of concentrated PM_2.5 in the OASIS-1 exposure chamber air truly reflect that of non-concentrated PM_2.5 present in the ambient air [11, 12]. The “OASIS-1” system enables studies on animal models that recapitulate personal, short- or long-term exposure to environmentally relevant PM_2.5.

During the exposure time period, the mean daily ambient PM_2.5 concentration was 6.5μg/m³, while the mean concentration of PM_2.5 in the exposure chamber was 74.6μg/m³ [9]. The mice were exposed to concentrated PM_2.5 or FA for 6 hours per day, 5 days per week. After taking into account the unexposed time and weekends, the calculated average daily PM_2.5 concentration to which the mice were exposed was 11.6 μg/m³ [the annual average PM_2.5 National Ambient Air Quality Standard is 15.0 μg/m³ (epa.gov/air/criteria.html)]. Previously, our X-ray fluorescence spectroscopic analysis of PM_2.5 composition in the exposure chambers revealed that the major components of PM_2.5 complex are alkali metals, alkaline earth metals, transition metals, poor metals, non-metals, metalloid, and halogens [9].

To evaluate the impact of PM_2.5 exposure in liver pathology, we performed histological analyses with liver tissue sections of the mice exposed to PM_2.5 or FA for 3 or 10 weeks. Based on H&E staining of the liver cellular structure as well as Sirius-red and Masson’s trichrome staining of hepatic collagen deposition, we identified hepatic steatosis, lobular and portal inflammation, and perisinusoidal fibrosis in the liver of the mice exposed to PM_2.5 for 10 weeks (Figure 1A). In comparison, the mice exposed to PM_2.5 for 3 weeks displayed non-significant, subtle hepatic steatosis, lobular inflammation, and hepatocyte ballooning (Supplemental figure 2). Using the NAFLD grading and staging score system, we confirmed that the mice exposed to PM_2.5 for 10 weeks developed modest NASH, characterized by steatosis, hepatic inflammation, and perisinusoidal fibrosis (Figure 1B).

Exposure to PM_2.5 reduces hepatic glycogen storage and impairs glucose and insulin homeostasis

To assess the impact of PM_2.5 exposure in liver metabolism, we examined glucose and insulin homeostasis in the mice exposed to PM_2.5 or FA. Under normal conditions, glycogen storage is located in the area close to central veins in the liver tissue [13]. The mice exposed to PM_2.5, but not FA, for 10 weeks lost the normal distribution of glycogen storage around central veins, as indicated by Periodic-acid Schiff (PAS) staining of glycogens in liver tissue sections (Figure 2A). Both the glycogen staining and the enzymatic assessment of hepatic glycogen indicated that levels of glycogen in the PM_2.5-exposed mouse liver were reduced, compared to that in the FA-exposed mouse liver (Figure 2A–B), suggesting a significant negative effect of PM_2.5 exposure on hepatic glycogen storage. In comparison, the amounts and distribution of hepatic glycogen in the mice exposed to PM_2.5 for 3 weeks were similar to those in the FA-exposed control mice (Supplemental figure 3A and F).

To determine whether PM_2.5 exposure can affect glucose homeostasis, we performed an intraperitoneal glucose tolerance test (IPGTT) with the animals after exposure to PM_2.5 or FA. Upon administration of glucose, the mice exposed to PM_2.5 for 10 weeks, but not 3 weeks, had higher levels of glucose remaining in the blood (Figure 2C; Supplemental figure 3G). Moreover, fasting blood glucose and insulin levels in the mice exposed to PM_2.5 for 10 weeks were much higher than those in the FA-exposed mice (Figure 2C–D). Homeostasis model assessment of insulin resistance (HOMA-IR) data indicated that the mice exposed to PM_2.5, but not FA, displayed significant insulin resistance (Figure 2E). Additionally, body weights of the mice after PM_2.5 exposure were slightly reduced, compared to that of the FA-exposed mice (Figure 2F; Supplemental figure 3H).

Exposure to PM_2.5 represses the signaling pathway mediated through the insulin receptor substrate 1 (IRS1)

We investigated potential signal transduction pathways in the liver through which PM_2.5 exposure disrupts glucose and insulin homeostasis. Glycogen synthesis is stimulated by the insulin receptor via activation of IRS1 [14]. It is known that phosphorylation of IRS1 at the residues Ser636 and/or Ser1101 inhibits IRS1-mediated insulin signaling through Akt kinase in regulating glycogen synthesis and blood glucose levels [14, 15]. Immunoblotting analysis indicated that phosphorylation of IRS1 at both Ser636 and Ser1101 was increased in the liver tissue of the mice exposed to PM_2.5, compared to that of the FA-exposed mice (Figure 2G), indicating PM_2.5-triggered inhibition of the IRS1 signaling. Consistently, phosphorylation of Akt, the downstream of IRS1 in regulating glycogen synthesis [16], was decreased in the liver of the mice exposed to PM_2.5, thus confirming the effect of PM_2.5 exposure on suppressing the IRS1-Akt signaling pathway (Supplemental figure 4A).

We assessed the effect of PM_2.5 on IRS1 signaling in hepatic stellate cells (HSC), a cell type in liver that interacts with resident macrophages to play critical roles in liver metabolism and fibrosis [17]. LX-2 is a human HSC cell line that retains the key features of HSC and has been used as an experimental tool in studying liver metabolism and fibrosis [18]. We cultured LX-2 cells in conditioned medium from the mouse macrophage cell line RAW264.7 exposed to 5 μg/ml of PM_2.5, a justified PM_2.5 concentration that can induce intracellular stress response and macrophage activation in vitro (Supplemental figure 5). Immunoblotting analysis showed that the phosphorylation of IRS1 at both Ser1101 and Ser636 was increased in LX-2 cells exposed to the PM_2.5-containing conditioned medium in a time-dependent manner (Supplemental figure 4B), thus confirming the suppression effect of PM_2.5 on the IRS1-mediated signaling.

PM_2.5 exposure leads to dysregulated lipid homeostasis and reduced expression of PPARγ and PPARα in the liver

We next examined the impact of PM_2.5 exposure in lipid metabolism. Consistent with the hepatic steatosis phenotype (Figure 1), accumulation of hepatic lipid droplets, as indicated by Oil-red O staining, was increased in the liver of mice exposed to PM_2.5 for 10 weeks (Figure 3A). The accumulation of hepatic lipid contents was further confirmed by the increased levels of hepatic triglycerides (TG) and cholesterol detected in the liver of the PM_2.5-exposed mice (Figure 3B–C). Moreover, levels of plasma TG and low/very low-density lipoproteins (LDL/VLDL) were elevated, while levels of plasma high-density lipoproteins (HDL) were not significantly changed in the mice exposed to PM_2.5 for 10 weeks (Figure 3D–E). However, no significant induction of liver enzymes aspartate amino transferase (AST) and alanine aminotransferase (ALT), the indicators of liver damage [19], was detected in the PM_2.5-exposed mice (Supplemental figure 6). Additionally, the mice exposed to PM_2.5 for 3 weeks only displayed marginal increase in hepatic steatosis (Supplemental figures 2 & 3A–C) and plasma TG levels (Supplemental figure 3D).

PPARs play important roles in the regulation of cellular differentiation, lipid and glucose metabolism, as well as inflammation [20]. In particular, levels of PPARγ and PPARα are inversely correlated with hepatic steatosis and glycogen storage in the progression of metabolic disease. Immunoblotting analysis showed that expression of two PPARγ isoforms, PPARγ1 and PPARγ2, were decreased in the livers of the mice exposed to PM_2.5 for 10 weeks, compared to that in the FA-exposed mice (Figure 3F). Expression of PPARα, a key regulator of fatty acid oxidation [20], was also decreased in the liver of PM_2.5-exposed mice (Figure 3F). Decreased expression of PPARα may contribute to hepatic steatosis in PM_2.5-exposed animals via down-regulation of fatty acid oxidation, an interesting question to be further investigated. Likely, down-regulation of PPARγ and PPARα, two key regulators of hepatic inflammation and metabolism, may partially account for the NASH-like phenotype observed in the PM_2.5-exposed animals.

PM_2.5 triggers the inflammatory pathways mediated through JNK, NF-κB, and TLR4

We assessed inflammatory responses in the livers of the animals exposed to PM_2.5. Immunohistochemical (IHC) staining of the macrophage activation marker F4/80 indicated that numbers of activated Kupffer cells (hepatic macrophages) and alveolar macrophages (pulmonary macrophages) were increased in the liver and lung of the mice exposed to PM_2.5 for 10 weeks, compared to those in the FA-exposed mice (Figure 4A). Expression of the mRNAs encoding pro-inflammatory cytokines IL1β, IL6, and TNFα was increased in the liver of the mice exposed to PM_2.5 (Figure 4B). PM_2.5-induced expression of the Il1β, Il6, and Tnfα genes was confirmed via in vitro experiment with RAW264.7 cells treated with PM_2.5 particles collected from the filters retrieved from OASIS-1 (Supplemental figure 7). Enzyme-linked immunosorbent assay (ELISA) indicated that plasma levels of TNFα, the major pro-inflammatory cytokine that is closely associated with NASH [21], were significantly increased in the mice exposed to PM_2.5 for 10 weeks (Figure 4C). Interestingly, although the mice exposed to PM_2.5 for 3 weeks exhibited marginal hepatic inflammation (Supplemental figure 2), plasma levels of TNFα were significantly increased in these mice (Supplemental figure 8B). Additionally, F4/80 IHC staining indicated a non-significant increase in numbers of activated alveolar macrophages in the lung of the mice after 3 weeks of PM_2.5 exposure (Supplemental figure 8D). Combining the results obtained from the mice exposed to PM_2.5 for 10 weeks, our studies suggest that systemic inflammation occurs from early on and is followed by lung and hepatic inflammation in the animals under PM_2.5 exposure.

To elucidate signal transduction pathways by which PM_2.5 stimulates expression of pro-inflammatory cytokines in the liver, we first examined JNK, a major inflammatory stress mediator that promotes pro-inflammatory cytokine expression by activating active-protein 1 (AP-1) [22]. JNK phosphorylation was increased in the liver of the mice exposed to PM_2.5 for 10 weeks (Figure 4D). Supporting the effect of PM_2.5 on JNK activation, levels of phosphorylated JNK were increased in RAW264.7 cells cultured with PM_2.5 (Supplemental figure 9A). Since AP-1 is the downstream trans-activator in regulating expression of pro-inflammatory cytokines under the JNK-mediated inflammatory response, we determined PM_2.5-triggered transcriptional activation of AP-1 in RAW264.7 cells. Upon PM_2.5 challenge, trans-activation of the AP-1 promoter, indicated by luciferase reporter analysis, was significantly increased in RAW264.7 cells (Supplemental figure 9B). Further, we tested whether JNK/AP-1 signaling is solely responsible for PM_2.5-triggered inflammation by using the JNK inhibitor SP600125 [23]. Suppression of JNK pathway in RAW264.7 macrophages by SP600125 did not significantly reduce induction of the IL6 gene triggered by PM_2.5 (Supplemental figure 9C), suggesting that PM_2.5 may trigger other inflammatory pathways to induce production of pro-inflammatory cytokines. Indeed, we found that PM_2.5 exposure can also activate NF-κB, a key mediator of the inflammatory response, in the liver. Activation of the NF-κB is initiated by signal-induced phosphorylation and degradation of NF-κB Inhibitor (IκB) [24]. Western blot analysis demonstrated that levels of phosphorylated IκB were increased in the liver of PM_2.5-exposed mice (Figure 4E), indicating activation of NF-κB pathway in the liver under PM_2.5 exposure. Furthermore, induction of Toll-like receptors (TLRs), the specific pattern recognition receptors that recognize structurally conserved molecules, can lead to activation of NF-κB and subsequent pro-inflammatory cytokine production [25]. Expression levels of TLR2 and TLR4 mRNAs were increased in the liver of mice exposed to PM_2.5 for 10 weeks (Supplemental figure 10A). Pre-treatment of the TLR4 signaling antagonist RP105 reduced expression levels of the pro-inflammatory cytokine genes IL6 and TNFα in RAW264.7 cells in response to PM_2.5 challenge (Supplemental figure 10B–D) [26], thus confirming the involvement of TLR4-mediated signaling in PM_2.5-triggered macrophage inflammation.

We have previously showed that reactive oxygen species (ROS) production is required for PM_2.5-induced ER-stress response in macrophages [9]. To further delineate the mechanism underlying PM_2.5-triggered inflammation, we tested whether PM_2.5-triggered inflammation depends on intracellular ROS. ROS production from NAPDH oxidase or mitochondria in macrophages was blocked by over-expressing dominant negative N17Rac1, the small GTPase component of NADPH oxidase, or manganese superoxide dismutase (Mn-SOD) using adenovirus-based expression system [9, 27]. Upon PM_2.5 challenge, expression levels of the genes encoding pro-inflammatory cytokines IL-6 and TNFα were significantly reduced in RAW264.7 macrophages expressing dominant negative N17Rac1 or Mn-SOD (Supplemental figure 11). These results indicate that ROS produced through NAPDH oxidase or mitochondria is critical for PM_2.5-triggered inflammation.