Peripheral Metabolic Effects of Ozone Exposure in Healthy and Diabetic Rats on Normal or High-Cholesterol Diet

Abstract

Epidemiological studies show that individuals with underlying diabetes and diet-associated ailments are more susceptible than healthy individuals to adverse health effects of air pollution. Exposure to air pollutants can induce metabolic stress and increase cardiometabolic disease risk. Using male Wistar and Wistar-derived Goto-Kakizaki (GK) rats, which exhibit a non-obese type-2 diabetes phenotype, we investigated whether two key metabolic stressors, type-2 diabetes and a high-cholesterol atherogenic diet, exacerbate ozone-induced metabolic effects. Rats were fed a normal control diet (ND) or high-cholesterol diet (HCD) for 12 weeks and then exposed to filtered air or 1.0-ppm ozone (6h/day) for 1 or 2 days. Metabolic responses were analyzed at the end of each day and after an 18-hour recovery period following the 2-day exposure. In GK rats, baseline hyperglycemia and glucose intolerance were exacerbated by HCD vs. ND and by ozone vs. air. HCD also resulted in higher insulin in ozone-exposed GK rats and circulating lipase, aspartate transaminase, and alanine transaminase in all groups (Wistar>GK). Histopathological effects induced by HCD in the liver, which included macrovesicular vacuolation and hepatocellular necrosis, were more severe in Wistar vs. GK rats. Liver gene expression in Wistar and GK rats fed ND showed numerous strain differences, including evidence of increased lipid metabolizing activity and ozone-induced alterations in glucose and lipid transporters, specifically in GK rats. Collectively, these findings indicate that peripheral metabolic alterations induced by diabetes and high-cholesterol diet can enhance susceptibility to the metabolic effects of inhaled pollutants.

INTRODUCTION

Twin epidemics of diabetes and obesity represent a widespread public health burden, globally and in the United States. According to the World Health Organization (WHO), the incidence of diabetes worldwide increased from 108 million in 1980 to 422 million in 2014 (WHO, 2020). Similarly, obesity rates have grown dramatically in recent decades, nearly tripling since 1975 and affecting 39% of adults over age 18 as of 2016. Even more concerning, nearly 30 million children under the age of 5 were overweight or obese as of 2019 (WHO 2020; Saeedi et al., 2019; 2020). Obesity and diabetes both contribute to cardiovascular disease and collectively drive a constellation of health effects referred to as cardiometabolic syndrome (Fernández-Rhodes et al., 2020). This condition is associated with chronic insulin resistance, diet-related lipid accumulation, systemic inflammation, and high blood pressure (McCracken et al., 2018). Risk factors include calorie-rich Western-type diets, sedentary lifestyle, and socioeconomic stressors, as well as common environmental exposures, such as air pollution. The pathophysiologic mechanisms by which obesity, diabetes, and environmental stressors interact are complex and poorly understood.

Epidemiological evidence has linked exposure to air pollutants with exacerbation of metabolic diseases, including diabetes and obesity, particularly among genetically predisposed individuals (Lim and Thurston, 2019; Yang et al., 2020; Furlong and Klimentidis, 2020). Interestingly, air pollution levels have also been associated with increased consumption of trans fats and other fast foods in adolescents (Chen et al., 2019), suggesting dietary, behavioral, and socioeconomic interactions. In contrast, experimental studies have shown that liver steatosis induced by a high fat diet was reduced in mice inhaling ambient particulate matter (Qiu et al., 2017), demonstrating the complexity of biological factors and particular exposures that may influence metabolic endpoints. It is likely that air pollutants may alter many different processes associated with insulin signaling, glucose uptake, and lipid metabolism (Alderete et al., 2018). A better understanding of these mechanisms, and how they are impacted by underlying genetic predisposition to diabetes with or without nutritional inadequacies, will be important in identifying and managing risk in susceptible populations.

A few studies of air pollution have examined the metabolic impact of underlying diabetes using experimental models. In a rat model of type-1 diabetes induced by streptozotocin, subchronic exposure of real-time ambient particulate matter was associated with increased cardiac and renal inflammatory changes (Yan et al., 2014). The same authors reported on enhancement of insulin resistance after exposure to ambient particulate matter in a rat model of obesity induced by high-fat diet (Yan et al., 2011). In mice, a high-fat diet led to potentiation of type-2 diabetes after exposure to particulate matter (Goettems-Fiorin et al., 2016). In another example, KK-Ay mice predisposed to type-2 diabetes had altered glucose tolerance after repeated exposure to ozone (Zhong et al., 2016). These studies highlight ways in which metabolic stress resulting from diabetes, unhealthy diet, and air pollution may combine to produce a unique susceptibility phenotype.

The Goto-Kakizaki (GK) rat is a Wistar-derived model based on phenotypic segregation of a glucose intolerance trait associated with non-obese diabetes mellitus (Galli, 1999; Kuwabara et al., 2017; Guest, 2019). Insulin insufficiency occurs very early in life due to a defect in pancreatic insulin production and transport, which eventually leads to the development of peripheral insulin resistance. This sequence differs from type-2 diabetes in humans, in which the pathogenesis begins with peripheral insulin resistance and eventually leads to impairment of insulin secretion in advanced disease (Guest, 2019). However, the physiological features, such as hyperglycemia, insulin resistance, and insulin insufficiency, are similar. The GK model was used in our previous studies for examining pulmonary and systemic effects of acrolein and smog (Snow et al., 2017; McGee et al., 2018). In this study, we hypothesized that a high-cholesterol diet (HCD) will lead to exacerbation of metabolic alterations in GK rats, owing to defects in insulin secretion and glucose metabolism. In addition, we predicted that GK rats fed HCD will be more susceptible to ozone-induced metabolic impairment and peripheral pathological changes in metabolically active organs. We fed normal diet (ND) or HCD to 4-week old healthy male Wistar and GK rats for 12 weeks and assessed their adiposity and glucose tolerance. We then exposed them to air or ozone and evaluated metabolic alterations, including changes in glucose and lipid metabolic processes, pathological alterations in the liver and adipose tissue, and gene expression changes in the liver.

MATERIALS AND METHODS

Animals and diets

Male Wistar and Wistar-derived Goto-Kakizaki (GK) rats were purchased from Charles River Laboratories, Inc. (Kingston, NY) at 3 weeks of age and maintained in an AAALAC-approved animal facility (12h light/dark cycle, 23±1°C). Rats were double-housed in polycarbonate cages containing hardwood chip bedding and acclimatized for 1 week prior to starting the dietary regimen. Animal use and care protocol was approved by the U.S. Environmental Protection Agency (EPA)’s Institutional Animal Care and Use Committee (IACUC).

Wistar and GK rats were each divided into two groups. A control group that continued to be fed the Purina 5001 rodent diet (Normal Diet, ND; Ralston Purina Laboratories, St. Louis, MO) used in our animal facility, while the diet-treatment group received a synthetic high-cholesterol atherogenic diet (HCD, Teklad Custom Research Diet TD.02028; Harlan Laboratories, Inc., Indianapolis, IN) (Figure 1). A brief description of diet composition is provided in Table 1. The dietary regimen started at 4 weeks of age and continued for over 12 weeks until the end of the experiment. Food and water were provided ad libitum unless otherwise stated.

Figure 1.

Schematic of experimental design showing timeline for dietary regimen, ozone exposures and assessment of endpoints.

Table 1.

The composition of normal and high-cholesterol atherogenic diets.

	Purina	Teklad
	5001	TD.02028
Content	Normal Diet (ND)	High-Cholesterol Diet (HCD)
Protein, % (wt/wt)	23.9	17.3
Carbohydrates, % (wt/wt)	48.7	46.9
Total fat, % (wt/wt)	5.7	21.2
Fiber, % (wt/wt)	5.3	---
Cholesterol, % (wt/wt)	0.0002	1.25
Metabolizable energy, kcal/g	3.02	4.5

The values for normal rodent diet Purina 5001 (Ralston Purina Laboratories, St. Louis, MO) are derived from https://www.labsupplytx.com/wp-content/uploads/2012/10/5001.pdf. The values for synthetic high-cholesterol atherogenic diet (Teklad Custom Research Diet- TD.02028; Harlan Laboratories, Inc., Indianapolis, IN) are derived from the diet information available at https://insights.envigo.com/hubfs/resources/data-sheets/02028.pdf.

Body weight and composition

Body weight and composition were assessed in both strains prior to starting of dietary regimen (week 0) and at weeks 4, 8, and 11.5 (rounded to 12 weeks throughout) of dietary treatment, prior to air or ozone exposure. A Bruker Minispec LF90 II TD-NMR body composition analyzer (Bruker Optics, Inc., Billerica, MA) was used to obtain relative percentages of lean body mass, body fat, and body fluid, as previously described (Gordon et al., 2017).

Glucose Tolerance Testing

Glucose tolerance testing (GTT) was performed prior to starting the dietary regimen (week 0), and at weeks 4, 8, and 12 of dietary treatment, each time after a 6-7 hour fast (Figure 1). Only the 2-day 18-hour recovery group underwent GTT on day 1 and day 2 after exposure to air or ozone and at the 18-hour recovery time point (prior to necropsy; see below). Blood glucose measurements were assessed with a glucose strip and a Bayer Contour glucometer using blood obtained by pricking the tip of tail. Rats were assessed at baseline and then immediately injected intraperitoneally with 20% pharmaceutical grade glucose solution to achieve a 2 g/kg dose (10 mL/kg). Glucose measurements were repeated at 30m, 60m, 90m and 120m. The glucose concentration data were plotted against time, and the trapezoidal method was used to derive the area under the curve (AUC) measurement.

Ozone exposure

After 12 weeks of receiving ND or HCD, Wistar and GK rats were exposed to clean air or 1.0 ppm ozone, 6 hours/day, for either one day or two consecutive days. One additional group of rats exposed for 2 days to ozone had an 18-hour recovery period prior to necropsy (18-hour recovery group). This latter group underwent GTT after day 1 and day 2 and again prior to necropsy (Figure 1). This air and ozone exposure paradigm allowed us to determine acute stress-mediated metabolic impairment through GTT on day 1, the peak of ozone-induced injury response on day 2 and potential reversal of these effects in the recovery group on day 3.

As shown in several of our previous studies (Snow et al., 2018; Gordon et al., 2017), ozone was generated using a silent arc discharge generator (OREC, Phoenix, AZ) and transported to Rochester-style “Hinners” chambers using mass flow controllers (Coastal Instruments Inc., Burgaw, NC). Chamber temperature, relative humidity, chamber air flow, and ozone concentrations were assessed continuously during exposure. The actual average chamber ozone concentration was 1.0 ± 0.02 ppm. The measured average air chamber temperature was 71.8 ± 0.5°F, relative humidity (RH) was 52.8 ± 2.7, and air flow was 265 ± 35 L/min (mean ± standard deviation). For the ozone chamber, temperature was 72.6 ± 0.6°F, RH was 51.7 ± 2.3, and air flow was 258 ± 2 L/min (mean ± standard deviation). These values are also reported in the companion paper (Snow et al., 2020).

Necropsy and sample collection

Necropsies were performed within a 2-hour period following day 1 or 2 of exposure, or after the 18-hour recovery period following GTT (Figure 1). Animals (n=6/group) were euthanized with an overdose of sodium pentobarbital (Virbac, AH, Inc., Fort Worth, TX; >200 mg/kg; intraperitoneally). Blood samples were collected from the abdominal aorta using serum separator vacutainers, centrifuged at 3500 × g for 10 min and serum aliquots were stored at −80°C for later analysis. Abdominal white adipose tissue and liver samples were fixed in formalin for pathology analysis. In addition, a small portion of liver from a right-middle lobe was frozen in liquid nitrogen and stored at −80°C for later RNA analysis.

Serum/plasma analysis of lipids and hormones

Serum concentrations of total cholesterol, high-density lipoprotein (HDL), triglyceride, lipase, and aspartate aminotransferase (AST, glutamic-oxaloacetic transaminase; SGOT) were analyzed using kits from TECO Diagnostics (Anaheim, CA). Low-density lipoprotein (LDL) and alanine aminotransferase (ALT, glutamic-pyruvic transaminase SGPT) values were analyzed using a kit from Thermo Fisher Scientific, Inc. (Middletown, VA). These assays were modified for the Konelab Arena 30 system (Thermo LabSystems, Espoo, Finland). Insulin, glucagon, and leptin levels were measured using rat-specific electrochemiluminescence assay kits (Meso Scale Discovery, Gaithersburg, MD) via manufacturer instructions.

Histopathological assessment of liver and adipose tissue

Formalin-fixed liver and visceral white adipose tissues from day 1 animals were paraffin-embedded, sectioned, and stained with hematoxylin and eosin (H&E) using standard histologic procedures. One section of each tissue type was evaluated for each animal in a blinded manner. Stained sections were evaluated via light microscopy by a board-certified pathologist using established pathologic criteria for inflammatory, degenerative, metaplastic, and proliferative changes (Thoolen et al., 2010; Greaves et al., 2013).

Targeted mRNA sequencing in liver

Total RNA was isolated from approximately 20 mg of liver for Wistar and GK rats fed ND and exposed to air or ozone for 1-day (n=6/group) using RNeasy mini kits (Qiagen, Valencia, CA). RNA yield was determined using Qubit fluorometric quantitation (Thermo Fisher Scientific Inc., Waltham, MA). Custom Illumina® TruSeq® Targeted RNA Expression kits were ordered from Illumina Inc. (San Diego, CA, USA) using a custom gene list selected by us that included genes involved in metabolic and pathological processes (published as a supplementary table in Snow et al., 2020). Targeted RNAseq libraries were prepared from total RNA using the Illumina TruSeq Targeted RNA Expression Guide 15034665 C protocol. Intact total RNA (50 ng per well) was used for first strand cDNA synthesis and followed the instructions for a 37-cycle PCR amplification. Resulting libraries were pooled based on tissue type, and the pooled library was quantified using Agilent Technologies High-sensitivity DNA assay chip and Qubit. On-board Clustering and sequencing of libraries was performed using Illumina MiSeq system and V3 kit. A 50-cycle sequencing protocol was performed at a final library concentration of 10 pM and 5% Phix spike-in. TruSeq_CRT_Manifest_TC0040455-CRT.txt was used by Miseq reporter as the manifest file for secondary analysis. Clusters passing filter and percent aligned sequences (>85%) were used as primary quality metrics.

Data analysis

GraphPad Prism v8.4.2 software (San Diego, CA) was used to graph and analyze body weight, glucose, serum total, HDL, and LDL cholesterol and triglycerides, injury markers, and hormonal. For baseline body weight, composition and GTT data two way ANOVA was used to determine interactive effect of diet and strain. For endpoints involving air and ozone exposure, three independent two- way ANOVA tests were performed for each time point to determine 1) ozone effect for a given strain/diet, 2) rat strain effect for given exposure/diet and 3) diet effect for given exposure/strain. Holm-Sidak’s multiple comparison test was used to stablish significant differences and a p-value <0.05 was considered statistically significant. The mRNAseq data were quintile-normalized, mean-centered, and visualized as Row z-scores in a heatmap. Hierarchical clustering using Ward linkage was performed from the Euclidean distances among these ozone-responsive endpoints. The analysis was performed using the packages of preprocess Core (version 1.36.0) and Complex Heatmap (version 1.12.0) in R statistical environment (version 3.3.2). Hierarchical clustering was performed using Average Linkage and Euclidean Distance Measurement Method. Principal component analysis was calculated for gene list using Cluster 3.0 (deHoon, 2004) and visualized using Sigma Plot 11.0. For histopathological data, differences in lesion incidence between groups were evaluated using a Fisher’s Exact Test.

RESULTS

Rat strain and diet effects on body weight and composition

Wistar rats had greater body weight compared to GK rats at 4 weeks of age, prior to start of the dietary regimen. Higher body weight for ND vs. HCD groups was observed at 8, and 12 weeks of dietary treatment in both strains (Figure 2A). The diet-induced increase in body weight after 12 weeks of dietary regimen (16 week of age) was more pronounced in GK compared to Wistar rats. Body fat percentage at the start of dietary regimen was not different between Wistar and GK rats. However, at 8, and 12 weeks of dietary regimen, GK rats gained significantly more fat mass than Wistar rats (HCD>ND). Body fat was modestly higher by ~2-3% at 4 and 8 weeks into dietary regimen in Wistar rats receiving ND vs. HCD, but at 12 weeks there was no significant difference between ND and HCD groups (Figure 2B).

Figure 2.

Body weight and composition of male Wistar and diabetic Goto-Kakizaki (GK) rats during a 12-week regimen with normal (ND) or high-cholesterol (HCD) diets. Body weights (A) and body fat content (B) were assessed at 4 weeks of age, prior to starting the dietary regimen (week 0), and then at 4, 8, and 12 weeks of dietary treatment using a Bruker Minispec LF90 II TD-NMR body composition analyzer (Bruker Optics, Inc., Billerica, MA). Values represent mean ± standard error of n=36 animals. Significant strain differences within the same dietary group for Wistars are shown by "†” (p ≤ 0.05) and for GK are shown by "‡” (p ≤ 0.05); significant diet effects within the same strain for Wistar are shown by “‡” (p ≤ 0.05) and for GK by “‡‡” (p ≤ 0.05).

Strain and diet effects on hyperglycemia and glucose intolerance

Wistar rats fed HCD did not exhibit hyperglycemia or glucose intolerance, as measured by clearance of injected glucose at baseline or after 4, 8, and 12 weeks of diet relative to ND. In contrast, GK rats showed evidence of hyperglycemia and glucose intolerance as early as 4 weeks of age, prior to the start of dietary regimen (Figure 3A–B). Both diets resulted in a further increase in hyperglycemia and glucose intolerance after 4, 8, and 12 weeks GK rats, with a greater effect in the HCD group (Figure 3C–H). The assessment of AUC confirmed these observations. These data suggest that GK rats are more susceptible to diet-induced impairment in glucose uptake by peripheral tissues.

Figure 3.

Glucose tolerance test (GTT) in Wistar and diabetic Goto-Kakizaki (GK) rats during a 12-week dietary regimen with normal (ND) or high-cholesterol (HCD) diets. GTT was performed on fasted rats at the start of the dietary regimen (week 0) and at 4, 8, and 12 weeks of dietary treatment (A, C, E, G). The respective area under the curve (AUC) data are shown in the right panel as bar graphs (B, D, F, H). Values represent mean ± standard error of n=12 animals. For curves (A, C, E, G), significant strain differences within the same dietary group are shown by "†” (p ≤ 0.05) for ND and "††” (p ≤ 0.05) for HCD; significant diet effects within the same strain are shown by “‡” (p ≤ 0.05) for Wistar and “‡‡” (p ≤ 0.05) for GK. For bar graphs (B, D, F, H), significant differences are shown by "†”, for Wistar vs. GK (for a given dietary group); and “‡”, for ND vs. HCD (for a given strain).

Ozone effects on hyperglycemia and glucose intolerance

Previous work in our group has demonstrated that acute ozone exposure in healthy rats induces hyperglycemia and glucose intolerance, which are reversible upon discontinuation of treatment (Bass et al., 2013; Miller et al., 2015). In this study, we assessed the influence of underlying metabolic complications induced by HCD in Wistar and GK rats. On day 1, GK rats-exposed to ozone exhibited increased hyperglycemia compared to Wistar rats regardless of diet (Figure 4A–C). On day 2, ozone-induced hyperglycemia was evident only in GK fed ND. GK rats fed HCD were already highly hyperglycemic and effectiveness of ozone could not be accurately assessed using glucometer (Figure 4D–F). GK rats showed consistently higher glucose intolerance compared to Wistar rats regardless of diet or exposure. However, ozone-induced glucose intolerance in GK rats was not significantly different between ND and HCD on day 1 (Figure 4D–F). The ozone effects on blood glucose or glucose intolerance were not observed in Wistar rats at day 2, and Wistar or GK rats at the 18-hr recovery period (Figure 4G–I).

Figure 4.

Glucose tolerance test (GTT) in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) or high-cholesterol (HCD) diets and then exposed to air or ozone for 1 or 2 days. The 2-day group also included an 18-hour recovery time point. GTT was performed immediately following exposure on day 1, day 2, and the 18-hour recovery period. Line graphs show time course of blood glucose following glucose injection (A, D, G). Bar graphs show baseline glucose levels (B, E, H) and area under the curve (AUC) data (C, F, I). Values represent mean ± standard error of n=6 animals. Significant differences (p≤0.05) for bar graphs are shown by an “*”, for air vs. ozone (for a given strain or dietary group); “†”, for Wistar vs. GK (for a given exposure or dietary group); and “‡”, for ND vs. HCD (for a given strain or exposure condition).

Strain, diet, and ozone effects on circulating metabolic hormones

The GK rat model has a well-established genetic defect in the insulin production and secretion (Fakhrai-Rad et al., 2000; Nobrega et al., 2009). Accordingly, GK rats receiving ND in this study had generally lower insulin levels relative to Wistar rats at all time points (Figure 5A–C). HCD was associated with a generalized but highly variable increase in insulin in both strains. Ozone exposure at day 1 and 18-hour recovery period exacerbated this insulin increase in GK rats receiving HCD. This pattern was not observed with Wistar rats on HCD (Figure 5A–C). Serum glucagon levels generally appeared to be higher at baseline in GK rats relative to Wistar rats receiving ND; however, this trend also varied with HCD (Figure 5D–F). Leptin levels did not differ significantly between strains receiving ND and exposed to air. The HCD reduced leptin levels but only in Wistar rats regardless of ozone. As observed in a previous study (Miller et al., 2015), leptin levels on day 1 after ozone exposure increased in Wistar and GK rats fed ND, however, this increase was only evident in GK fed HCD (Figure 5G). No ozone-related changes in leptin were noted after day 2 or the 18-hour recovery period in either strain or dietary group (Figure 5H–I).

Figure 5.

Serum levels of metabolic hormones in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) and high-cholesterol diet (HCD) and exposed to air or ozone for 1 or 2 days. The 2-day exposure group also included an 18-hour recovery time point. Serum samples were collected within 2 hours following air or ozone exposure on day 1 (A, D, G), day 2 (B, E, H), and the 18-hour recovery period (C, F, I). Values represent mean ± standard error of n=6 animals. Significant differences (p≤0.05) are shown by “*”, for air vs. ozone (for a given strain or diet); "†”, for Wistar vs. GK (for a given exposure or diet); and “‡”,for ND vs. HCD (for a given strain or exposure).

Strain, diet, and ozone effects on serum lipids

We next measured circulating lipids as indicators of altered lipid homeostasis. Blood triglyceride levels were not significantly different between Wistar and GK rats fed ND or HCD and exposed to air at any time point (Figure 6A–C). On day 1, ozone exposure led to increases in triglycerides in GK rats receiving ND and HCD, but the effect in the HCD group was moderate (p=0.06). Baseline cholesterol levels between Wistar and GK receiving ND were not significantly different. As expected, circulating cholesterol levels increased in all rats receiving HCD at all times (Figure 6D–F). There were no significant ozone effects on cholesterol levels in either dietary group or strain. Concentrations of HDL were variable but higher in all Wistar and GK groups fed HCD (Figure 7A–C) with no consistent ozone exposure-related differences. In contrast, LDL was increased in all Wistar and GK rats receiving HCD (Figure 7D–F). Ozone exposure did not result in any significant group differences in LDL levels.

Figure 6.

Serum concentrations of triglycerides and cholesterol in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) and high-cholesterol (HCD) diets and exposed to air or ozone for 1 or 2 days. The 2-day group also included an 18-hour recovery time point. Serum samples were collected within 2 hours following exposure on day 1 (A, D), day 2 (B, E), and the 18-hour recovery period (C, F). Values represent mean ± standard error of n=6 animals. Significant differences (p≤0.05) are shown by “*”, for air vs. ozone (for a given strain or diet); "†”, for Wistar vs. GK (for a given exposure or diet); and “‡”, for ND vs. HCD (for a given strain or exposure).

Figure 7.

Serum levels of high-density (HDL) and low-density (LDL) lipoproteins in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) and high-cholesterol (HCD) and then exposed to air or ozone for 1 or 2 days. The 2-day group also included an 18-hour recovery time point. Serum samples were collected within 2 hours following exposure on day 1 (A, D), day 2 (B, E), and the 18-hour recovery period (C, F). Values represent mean ± standard error of n=6 animals. Significant differences (p≤0.05) are shown by the following: “*” for air vs. ozone exposure (within a given strain or diet); "†”, for Wistar vs. GK (within a given exposure or diet); and “‡”, for ND vs. HCD (within a given strain or exposure).

Strain, diet, and ozone effects on circulating markers of cellular injury

We predicted that lipid-rich HCD would increase the release enzymes involved in liver injury. Both Wistar and GK rats receiving ND had low levels of serum lipase activity; however, activity increased significantly with HCD (Figure 8A–C). This effect was more pronounced in Wistar compared to GK rats. No significant ozone effects on serum lipase were observed in either strain or dietary group. Serum liver enzymes (ALT and AST) were also measured as indicators of liver cell injury following HCD and/or ozone exposure. Serum ALT activities were low in both strains receiving ND, while HCD resulted in variable increases at all three exposure times with inconsistent ozone effects (Figure 8D–F). Similarly, serum AST activity also variably increased in both strains on HCD with no consistent ozone exposure effects (Figure 8G–I).

Figure 8.

Serum levels of liver injury markers in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) or high-cholesterol (HCD) diets and then exposed to air or ozone for 1 or 2 days. The 2-day group also included an 18-hour recovery time point. Serum samples were collected within 2 hours following exposure on day 1 (A, D, G), day 2 (B, E, H), and the 18-hour recovery period (C, F, I). Values represent mean ± standard error of n=6 animals. Significant differences (p≤0.05) are shown by “*”, for air vs. ozone exposure (for a given strain or diet); “‡”, for Wistar vs. GK (within a given exposure or diet); and “‡”,for ND vs. HCD (within a given strain or exposure).

Histopathological findings in liver and adipose tissue

Liver and white adipose tissue were collected after day 1 of air or ozone exposure, fixed in neutral formalin, processed, sectioned, and stained with H&E. Sections were examined by light microscopy. Histological changes in the lung are reported in the companion paper in this issue of the journal (Snow et al., 2020). In the liver, HCD resulted in moderate to marked cytoplasmic vacuolation consistent with microvesicular steatosis (synonymous with fatty change or lipidosis) (Table 2; Figure 9, A and B). Associated changes included hepatocellular necrosis and mixed cell inflammation characterized by neutrophil, macrophage, and lymphocyte infiltrates. These effects were present in both Wistar and GK strains, although the severity of changes tended to be greater in Wistar rats. Across all groups, GK rats had modestly higher incidence of mononuclear cell aggregates in the liver (p=0.048), while Wistar rats were more likely to have a macrovesicular component to the vacuolar change (p<0.001). Ozone exposure did not alter diet-induced effects or result in additional histopathological effects in the liver.

Table 2.

Histological findings in the liver after acute ozone exposure in Wistar and diabetic Goto-Kakizaki (GK) rats receiving normal diet (ND) or high-cholesterol diet (HCD).

Histologic change:			Mononuclear cell aggregates	Periportal lymphocytic infiltrates	Hepatocellular necrosis	Cytoplasmic vacuolation, microvesicular	Cytoplasmic vacuolation, macrovesicular	Mixed cell inflammation	Focal bile duct proliferation	Periportal fibrosis
Rat model	Ozone	HCD	Incidence (severity score)	Incidence (severity score)	Incidence (severity score)	Incidence (severity score)	Incidence	Incidence (severity score)	Incidence (severity score)	Incidence
Wistar	n	n	3/6 (0.5)	0/6 (0.0)	0/6 (0.0)	0/6 (0.0)	2/6	0/6 (0.0)	0/6	0/6
GK	n	n	6/6 (1.0)	1/6 (0.2)	0/6 (0.0)	0/6 (0.0)	0/6	0/6 (0.0)	1/6	0/6
Wistar	y	n	2/6 (0.3)	1/6 (0.3)	0/6 (0.0)	0/6 (0.0)	1/6	0/6 (0.0)	1/6	1/6
GK	y	n	6/6 (1.0)	1/6 (0.2)	0/6 (0.0)	0/6 (0.0)	0/6	0/6 (0.0)	0/6	0/6
Wistar	n	y	6/6 (1.8)	2/6 (0.7)	6/6 (2.2)#^	6/6 (4.0)#	5/6*,^	6/6 (3.0)#	0/6	1/6
GK	n	y	6/6 (1.2)	1/6 (0.2)	1/6 (0.2)	6/6 (2.7)#	0/6	6/6 (1.7)#	1/6	0/6
Wistar	y	y	6/6 (1.8)	2/6 (0.3)	6/6 (1.8)#	6/6 (3.8)#	2/6	6/6 (2.2)#	2/6	1/6
GK	y	y	5/6 (1.3)	0/6 (0.0)	5/6 (1.3)#	6/6 (3.0)#	0/6	6/6 (1.3)#	0/6	0/6

Values represent incidence of each finding with the average severity score (across all animals in the group) in parentheses. Severity scores were based on a qualitative 0-4 scale assessment (0=absent, 1=minimal, 2=mild, 3=moderate, 4=severe). Mononuclear cell aggregates represent foci of lymphocytes, plasma cells, and macrophages (histiocytes). Cytoplasmic vacuolation (microvesicular and macrovesicular) was consistent with hepatocellular steatosis (fatty liver, lipidosis); however, in some cases the microvesicular pattern may represent glycogen change. Mixed cell inflammation included neutrophils, macrophages, and lymphocytes associated with areas of steatosis and necrosis. See Figure 9 for representative images. Significant group differences (p<0.05) based on a 2-tailed Fisher’s Exact Test were indicated by the following:

^“*”, air vs. ozone (within same strain and dietary groups);

^“#”, ND vs. HCD (within same strain and exposure groups);

^{“^”,} Wistar vs. GK (within same exposure and dietary groups).

Figure 9.

Representative images of histopathological findings in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal (ND) or high-cholesterol (HCD) diets and then exposed to air or ozone for 1 day (H&E stained slides). Select HCD effects in the liver: A) liver with microvesicular cytoplasmic vacuolation and mixed cell inflammation (GK fed HCD) and B) liver with microvesicular and macrovesicular cytoplasmic vacuolation, scattered hepatocyte necrosis, and mixed cell inflammation (Wistar fed HCD). Select adipose findings in GK rats: C) white adipose tissue with focus of mixed cell infiltrate (GK fed ND) and D) focal adipocyte atrophy (GK fed HCD). Objective magnification = 60x for image A, 20x for image B, and 40x for images C and D.

In visceral white adipose tissue, findings were limited to minimal mixed cell aggregates, adipocyte atrophy, and mast cell infiltrates (Table 3; Figure 9C and D). Mixed cell aggregates consisted of small foci of macrophages, lymphocytes, and/or neutrophils. Atrophic cells were smaller, often multilocular, and typically found in clusters. Minimal mast cell infiltrates were considered background. Incidence values of mixed cell aggregates and adipocyte atrophy was higher overall in GK vs. Wistar rats (p<0.05 for both, independent of diet or ozone exposure). Adipocyte atrophy incidence was also higher in HCD rats (p=0.001, independent of strain or ozone exposure). No effects of ozone exposure were observed.

Table 3.

Histological findings in visceral white adipose tissue after acute ozone exposure in Wistar and diabetic Goto-Kakizaki (GK) rats receiving normal diet (ND) or high-cholesterol diet (HCD).

Histologic change:			Mixed cell aggregates	Mast cell infiltrate	Adipocyte atrophy
Rat model	Ozone	HCD	Incidence (severity score)	Incidence (severity score)	Incidence (severity score)
Wistar	n	n	1/6 (0.2)	0/6 (0.0)	0/6 (0.0)
GK	n	n	1/6 (0.2)	1/6 (0.2)	0/6 (0.0)
Wistar	y	n	2/6 (0.3)	1/6 (0.2)	0/6 (0.0)
GK	y	n	4/5 (0.8)	0/5 (0.0)	0/5 (0.0)
Wistar	n	y	2/5 (0.4)	0/5 (0.0)	0/5 (0.0)
GK	n	y	6/6 (1.0)	1/6 (0.2)	4/6 (0.8)
Wistar	y	y	2/5 (0.4)	0/5 (0.0)	1/5 (0.2)
GK	y	y	4/6 (0.7)	0/6 (0.0)	4/6 (0.7)

Values represent incidence of each finding with the average severity score (across all animals in the group) in parentheses. Severity scores were based on a qualitative 0-4 scale (0=absent, 1=minimal, 2=mild, 3=moderate, 4=severe). Mixed cell aggregates were composed of foci containing macrophages, lymphocytes, and/or neutrophils. Atrophic adipocytes were smaller than normal cells, often multilocular, and typically found in clusters. See Figure 9 for representative images. Although trends were apparent, no significant group differences were detected by 2-tailed Fisher’s Exact Test between the following: air vs. ozone exposure (within same strain and dietary groups); ND vs. HCD (within same strain and exposure groups); and Wistar vs. GK (within same exposure and dietary groups).

Strain- and ozone-related changes in mRNA expression in the liver

We previously reported that ozone exposure in Wistar Kyoto rats markedly alters liver mRNA expression, including genes related to lipid, glucose and amino acid metabolism, and mitochondrial function, and that these changes were associated with increases in circulating stress hormones such as epinephrine and corticosterone (Miller et al., 2015; Henriquez et al., 2018). Here, we investigated whether Wistar and GK rats fed ND respond differently to ozone by assessing a targeted group of genes involved in metabolite transport, and glucose and lipid metabolic processes in the liver in 1 day exposure group. Several strain differences were observed in air-exposed rats. Notably, Gck mRNA was higher, while Irs1, Pparg. Pprgc1a, and Crat mRNAs were lower in GK vs. Wistar rats exposed to air (Figure 10). Differential responses to ozone were also observed between strains. In Wistar rats, ozone exposure resulted in higher Pik3r1, Socs3, Cpt2, and Insr expression and lower Hmgcs1, Fabp4, Akt2, Irs1, and Pparg expression compared to respective air exposure group. These genes are involved in lipid transport and metabolism (Figure 10). In GK rats, ozone exposure altered a greater number of genes and some in directionally different manner than Wistar rats. Ozone in GK rats increased expression of Pik3ca, Hmgcr, Acacb, Nr1h3, Gck, Hmgcs1, Pck1, and Srebf2, and decreased expression of Irs2, Ppargc1a, Crat, Acaca, Slc2a2, Pdk4, Lipe, Socs3, Cpt2, and Insr (Figure 10). Collectively, these changes indicate that diabetic GK rats exhibit differential response to ozone for genes involved in lipid and glucose metabolism and transport mechanisms (Table 4).

Figure 10.

Ozone-induced changes in liver mRNA expression in Wistar and diabetic Goto-Kakizaki (GK) rats fed normal diet (ND) and then exposed to air or ozone for 1 day. Data are presented as Principal Component Analysis (A) and heatmap (B). Gene expression was determined using Illumina mRNA sequencing for the custom panel of genes listed in Table 1 (n=6/group). Rats fed high-cholesterol diet (HCD) were not included in this analysis. The data were normalized and raw z-scores were calculated. Significant differences (p ≤ 0.05) are shown by “*”, for air vs. ozone exposure (within same strain); “†””, for Wistar vs. GK (within same exposure). Hierarchical clustering was performed using Average Linkage and Euclidean Distance Measurement.

Table 4.

Ozone effects on gene expression changes and their functional relevance.

Group Comparison	Expression: Up/down	Genes	Biological function/process
Wistar:ozone versus Wistar:air	↑	Socs3, Cpt2, Ppargc1a, Crat	Increased breakdown of glucose, fatty acid transport
	↓	Akt2, Nr1h2	Decreased glucose transport, changes in lipid homeostasis
GK:ozone versus GK:air	↑	Irs2, Ppargc1a, Crat, Acaca, Slc2a2, Pdk4, Lipe, Socs3, Cpt2	Increased glucose and fatty acids transport and energy metabolism, lipolysis and increased transcriptional activities for lipid homeostasis mechanisms
	↓	Pik3ca, Acacb, Nr1h3, Gck, Hmgcs1, Srebf2, Pparg	Inhibition of inflammation and hypoglycemia, decrease insulin effect, inhibition of cholesterol and fatty acid biosynthesis, lipid accumulation
GK:ozone versus Wistar:ozone	↑	Irs2, Ppargc1a, Crat, Acaca, Slc2a2, Nr1h2	Increased glucose uptake and fatty acids transport and lipid metabolism
	↓	Pik3ca, Gck, Hmgcs1, Pck1, Srebf2, Pparg	Inhibition of gluconeogenesis, glucose breakdown and cholesterol synthesis

^*Biological process/function were predicted based on the functional role of genes in metabolic processes.

DISCUSSION

Exposure to air pollutants induces metabolic stress and increases risk for cardiovascular disease (e.g. Bourdrel et al., 2017). To help manage this risk, it is important to identify key lifestyle factors and sub-populations of individuals for which air pollution may increase adverse health effects. The goal of this study was to better understand how ozone-induced metabolic alterations are impacted by underlying non-obese type-2 diabetes and a high-cholesterol atherogenic Western-type diet. We found that diabetic GK rats, with a genetic defect in pancreatic insulin secretion and peripheral insulin resistance, have high baseline body fat content, hyperglycemia, and glucose intolerance at 4 weeks of age relative to healthy Wistar rats. Treatment with HCD for up to 12 weeks exacerbated body fat accumulation, adipose tissue inflammation, hyperglycemia, and glucose intolerance in GK vs. Wistar rats; however, produced greater pathology in the livers of Wistar rats. HCD increased circulating insulin in both strains, and ozone exacerbated this effect specifically in GK rats, suggesting accelerated peripheral insulin resistance. Gene expression changes in the liver further indicated distinct responses to ozone in GK rats related to lipid and glucose metabolism and transport mechanisms (Table 4). These strain-related differences in metabolic response to HCD and ozone were apparent despite similar ozone-induced lung pathology in both strains (data provided in companion paper, Snow et al., 2021). Our results demonstrate that GK rats are more sensitive to systemic metabolic changes and adipose atrophy induced by HCD and/or ozone, while Wistar rats displayed greater HCD-induced liver pathology. These findings highlight how key lifestyle and host risk factors may collectively drive susceptibility to air pollutants.

The GK strain of rat has well-characterized defects in insulin production and transport from the pancreas, which lead to insulin resistance by 5 months of age (Cahová et al., 2012). Previously, we showed that ozone induces hyperglycemia and glucose intolerance in healthy (non-GK) rat strains (Bass et al., 2013; Miller et al., 2015), and that these effects are mediated by increased circulating stress hormones causing increased gluconeogenesis and inhibition of pancreatic insulin secretion (Miller et al., 2016). Here, we showed marked hyperglycemia and glucose intolerance in GK relative to Wistar rats regardless of dietary treatment that is associated with increases in insulin, but only in GK rats receiving HCD (Figure 4). This effect occurred despite lower insulin levels in GK vs. Wistar rats on ND, which has also been noted in other studies (Portha et al., 2012). Given the impaired mechanisms for pancreatic insulin secretion in the GK strain, and inhibition of pancreatic insulin secretion by ozone, it is likely that higher levels of ozone-induced insulin may relate to greater peripheral insulin resistance in animals fed HCD. Additional work is needed, however, to identify how HCD may increase circulating insulin and contribute to insulin resistance in GK rats. Similarly, the mechanism by which HCD drives greater lipid accumulation and glucose metabolic impairment is unclear. We speculate that this effect may relate to accelerated peripheral insulin resistance and/or hyperphagia noted in GK rats (Maekawa et al., 2006; Kuwabara et al., 2017).

A number of our recent studies have demonstrated the contribution of sympathetically-mediated release of catecholamines and corticosteroids in mediating hyperglycemia and glucose intolerance, lymphopenia, and pulmonary inflammation after an acute ozone exposure (Kodavanti 2019, Miller et al., 2016; Henriquez et al., 2018, 2019). Although we did not examine the role of stress hormones in this study, it is likely that differential susceptibility of Wistar and GK rats to ozone could be due to differences in the degree of stress hormone production or effectiveness and the contribution of HCD in mediating these effects. Both epinephrine and corticosterone have been shown to modulate glucose metabolism and beta cell insulin secretion in response to stress (Ito et al., 2017; Fine et al., 2018). We have shown that ozone exposure inhibits beta cell insulin secretion, due to increased circulating stress hormones and glucocorticoids, which led to gluconeogenesis and hyperglycemia (Miller et al., 2016; Henriquez et al., 2018). Because GK rats have impaired insulin secretion, ozone-induced insulin increases occur only in GK rats fed the HCD diet, suggesting contribution of stress hormones in mediating peripheral insulin resistance in given experimental conditions.

A number of morphological and biochemical abnormalities have been reported in the adipocytes of GK rats (Kanoh et al., 2001; Cariou et al., 2004; Movassat et al., 2008; Xue et al., 2011; Kampf et al., 2005). In the current study, we observed higher mixed cell infiltrates and adipocyte atrophy in the white adipose tissue of GK vs. Wistar rats, independent of diet and ozone exposure. These changes could be associated with tissue remodeling and ongoing disease progression (related to the diabetic phenotype of the GK model under HCD conditions), greater systemic inflammation (as described in our companion paper, Snow et al., 2021), and other alterations in lipid metabolism. Changes in lipid redistribution induced by HCD (e.g. between liver and white adipose tissue) may also differ between strains, leading to a greater impact of HCD on adipose tissues in GK rats and liver in Wistar rats. Differences in the activity of lipid transporters in liver and adipose tissue and in insulin secretion, which has been linked to impaired adipose lipolysis (Rasineni et al., 2020), may have further contributed to these strain differences.

We expected the HCD diet to increase circulating total cholesterol and LDL, which are risk factors for atherogenic changes in the vasculature (Afonso et al., 2018). Here we wanted to see if there were differences in the susceptibility of Wistar and GK rats for HCD-induced changes in circulating lipids. Unlike our previous study, where high coconut oil and olive oil (but not fish oil) diets led to increased triglycerides in Wistar Kyoto rats (Snow et al., 2018), HCD did not increase circulating triglycerides in Wistar or GK rats. However, the first day of ozone exposure did lead to higher triglycerides in GK rats (Figure 5), similar to acrolein exposure (Snow et al., 2017) suggesting that lipid redistribution might be differentially occurring in these strains. Hypercholesterolemia in Wistar relative to GK rats fed HCD may be associated with differential absorption of lipids and/or tissue distribution. These findings show strain differences in lipid metabolism, and liver versus adipose tissue involvement (Movassat et al., 2008).

To further investigate differential responses of Wistar and GK rats to diet and ozone, we assessed biomarkers of liver and pancreatic cell injury. Previous studies have shown that HCD is associated with increased serum concentrations of liver and pancreatic acinar cell enzymes in several species (e.g. Hunt et al., 1986; Hameed et al., 2015; Kanaki et al., 2017). HCD-related increases in serum AST and ALT correlated with observed pathological changes in liver (Figure 8). Although not significant, ozone exposure tended to exacerbate increases in AST and ALT, especially in GK rats on HCD, which may indicate ozone exacerbation of liver injury in metabolically susceptible GK rats. Marked ozone-induced changes in liver gene expression were also noted in GK rats on ND. Despite pancreatic beta cell abnormalities and elevated body fat in GK rats, they had low levels of circulating lipase when on ND, which did not change after ozone exposure. Interestingly, HCD increased lipase activity in Wistar but not GK rats, which suggests the potential contribution of liver lipase as these rats exhibited greater pathology.

Exposure to ozone provides a classic example of stress response (Kodavanti, 2019). Repeated exposure to ozone results in adaptation, also referred to as habituation or tolerance, such that most known adverse effects are reversible (Miller et al., 2016). One can postulate that the longevity of effects mediated by the stressor exposure may underlie greater susceptibility. Although ozone-induced changes in most of the metabolic indicators of HCD-fed GK rats were not necessarily bigger in amplitude then those in HCD-fed Wistars, some of the changes were more persistent such as increases in insulin and liver injury markers, suggesting that ozone may predispose these rats to greater metabolic disease. The mechanisms of this resiliency are likely complex and may involve impaired compensatory response at central and/or local tissue levels.

Glucose and lipid metabolic processes are regulated through transporters and metabolic enzymes in the liver. Given the marked changes in liver pathology associated with HCD, we focused on ND groups to more clearly assess liver mRNA responses to ozone in Wistar and GK rats (Figure 9). Perhaps the most notable baseline difference between strains was expression of glucokinase (Gck) gene, which was higher in GK vs. Wistar rats. Glucokinase converts glucose to glucose-6-phosphate and promotes fatty acid synthesis. Although in GK rats this enzyme is downregulated in hypothalamic arcuate nucleus (Tsumara et al., 2017), in human liver, increased glucokinase expression has been associated with fatty liver disease (Peter et al., 2011). Higher liver Gck mRNA in GK rats could relate to a reduction of cellular glucose uptake, as evidenced by inhibition of insulin receptor substrate 1 (Irs1), which is involved in insulin receptor-mediated glucose uptake (Eckstein et al., 2017). In contrast to Gck, baseline expression of the PPAR-gamma receptor (Pparg), which regulates glucose metabolic processes and fatty acid storage (Janani et al., 2014), was downregulated in GK vs. Wistar rats at baseline. This effect may relate to insulin insufficiency of GK rats that alter glucose and fatty acid metabolic processes (Karahashi et al., 2016). Some of these ozone-induced liver gene expression changes for markers of glucose transport in diabetic GK rats are contradictory to what is observed in human type 2 diabetics (Krause et al., 2020; Thorens 2015). Since GK rats have a defect in insulin release, much like what is noted in type 1 diabetes (Donga et al., 2013) and after ozone exposure in Wistar Kyoto rats (Miller et al., 2016), it is likely that ozone might exacerbate the diabetes type 1 phenotype in GK rats fed ND and accelerate insulin resistance when fed HCD as noted earlier.

Major strain differences in liver gene expression were observed following ozone exposure. GK rats exhibited highly exacerbated changes in glucose and lipid metabolism-related mRNAs in response to ozone relative to Wistar rats. Examples of genes with lower expression in ozone-exposed GK vs. Wistar rats included 3-Hydroxy-3-methylglutaryl-CoA synthase 1 (Hmgcs1), which is involved in cholesterol biosynthesis; nuclear receptor subfamily 1 group H member 3 (Nr1h3), which is involved in transcriptional regulation of genes involved in lipid homeostasis and innate immunity (Joseph et al., 2003); and sterol regulatory element binding transcription factor 2 (Srebf2), which is involved in cholesterol biosynthesis (Bommer and MacDougald, 2011). This pattern suggests downregulation of cholesterol synthesis pathways, consistent with our earlier findings in Wistar Kyoto rats after ozone exposure (Table 4; Miller et al., 2015). More importantly, ozone resulted in higher expression of several genes involved in solute transport, including Crat, Irs2, Slc2a2 (a facilitated glucose transporter), and Lipe (lipase E, involved in lipid breakdown) in GK relative to Wistar rats, suggesting increased lipid and glucose metabolic activity in the livers of GK rats that might promote increased utilization of substrates, changes in lipid transport and energy production. Additional studies are needed to determine whether more chronic ozone exposures accelerate insulin resistance in GK rats over time.

We should point out several limitations of this study. First, it was an experimental study using two different rat models, both of which may differ from humans in specific pathways of lipid and glucose metabolism. For the diabetic GK model in particular, insulin resistance may develop via different mechanisms compared to diabetes in humans. The ozone exposure used in this study was intended to induce an acute stress response, and therefore the concentrations used were high and not relevant to real-world ambient exposures in people. In a clinical study using 0.4 ppm ozone exposure during intermittent exercise, the delivered ozone dose to human lung was shown to be 4-5 times higher than what rodents will receive when exposed during rest and inactivity period (Hatch et al., 2014), suggesting the relevance of the dose to clinical studies. Lastly, we did not examine gene changes in the HCD group or in adipose tissue, and thus, were not able to examine pathway-level changes to relate to pathological effects.

In summary, diabetic GK rats were more sensitive to HCD-induced obesity, glucose intolerance and adipose tissue atrophy when compared to healthy Wistar rats. In contrast, HCD resulted in greater liver pathology and injury in Wistar vs. GK rats. Acute ozone exposure increased insulin levels but only in GK rats fed HCD, which may relate to accelerated insulin resistance. In response to ozone, GK rats on both diets showed higher triglyceride levels, greater changes in the glucose homeostasis, and elevated levels for liver gene markers of glucose and lipid metabolism. These changes are consistent with more active lipid metabolic processing in the livers of GK rats when compared to healthy Wistar rats. Enhanced lipid metabolism in livers of GK rats as a result of ozone exposure may lead to its differential tissue redistribution, which is consistent with higher fat accumulation than Wistar rats with or without HCD. This finding is in concordance with greater adiposity and associated histopathological changes in adipose tissue of GK rats. Our data support the epidemiological findings showing exacerbation of air pollutant-induced health effects in humans with underlying diet-induced obesity/diabetes.