Funding agencies: Australian National Health and Medical Research Council's (NHMRC) project grant numbers: 585411, 1044288, and 102818.
Disclosure: The authors declare no conflict of interest.
Author contributions: F.H. designed and conducted the experiments, performed data analysis, and wrote the manuscript. V.B. supervised the exercise intervention. A.R.M. and E.E. assisted the experiments. M.M.Y. assessed liver histology. M.A.F., C.J.N., K.S.B.-A., and N.C.T. contributed intellectual input and reviewed and edited the manuscript. G.C.F. conceptualized and directed the study and reviewed and edited the manuscript.
Adipose inflammation and dysfunction underlie metabolic obesity. Exercise improves glycemic control and metabolic indices, but effects on adipose function and inflammation are less clear. Accordingly, it was hypothesized that exercise improves adipose morphometry to reduce adipose inflammation in hyperphagic obese mice.
Alms1 mutant foz/foz mice housed in pairs were fed an atherogenic or chow diet; half the cages were fitted with a computer-monitored wheel for voluntary exercise. Insulin-induced AKT-phosphorylation, adipocyte size distribution, and inflammatory recruitment were studied in visceral versus subcutaneous depots, and severity of fatty liver disease was determined.
Exercise prevented obesity and diabetes development in chow-fed foz/foz mice and delayed their onset in atherogenic-fed counterparts. Insulin-stimulated phospho-AKT levels in muscle were improved with exercise, but not in adipose or liver. Exercise suppressed adipose inflammatory recruitment, particularly in visceral adipose, associated with an increased number of small adipocyte subpopulations, and enhanced expression of beige adipocyte factor PRDM16 in subcutaneous fat. In atherogenic-fed foz/foz mice liver, exercise suppressed development of nonalcoholic steatohepatitis and related liver fibrosis.
Exercise confers metabo-protective effects in atherogenic-fed hyperphagic mice by preventing early onset of obesity and diabetes in association with enhanced muscle insulin sensitivity, improved adipose morphometry, and suppressed adipose and liver inflammation.
Adipose function is a key factor in metabolic health. Continuous disequilibrium in energy balance (intake vs. expenditure) can initiate adipose tissue overexpansion and impair its normal functioning [1]. A sedentary lifestyle leads to adipose dysfunction and metabolic disorders, but the mechanisms whereby physical activity protects against adipose dysfunction are less clear. PR-domain containing 16 (PRDM16) is a transcription co-regulator that causes “browning” of subcutaneous fat and increases body energy expenditure via mitochondria-rich beige adipocytes [2]. Healthy adipose tissue is therefore a “metabolic sink,” and subcutaneous adiposity may be beneficial to energy homeostasis [3]. In contrast, there is a well-characterized relationship between visceral obesity and metabolic comorbidities such as type 2 diabetes and nonalcoholic fatty liver disease (NAFLD) [3, 4]. This relationship has been attributed to the effects of nutrient oversupply in causing adipose stress and resultant inflammation, as well as the limited capacity of adipose for lipid partitioning [1, 5].
Obesity and diabetes are associated with NAFLD in at least 80% of cases [6]. Diabetes is particularly associated with severer forms of fatty liver disease, such as nonalcoholic steatohepatitis (NASH) and liver fibrosis [6, 7]. Interest therefore surrounds the metabolic milieu that contributes to the transition of simple steatosis to NASH and fibrosis. The earlier (“two hit”) concept of steatosis and NASH transition has been replaced by the idea that hepatocyte injury can be induced by lipotoxicity [7-9]. Hepatic injury is observed histologically as fatty, ballooned hepatocytes and liver necroinflammatory recruitment [7]. Suggestion of the term “liver lipotoxicity” for NASH cements the centrality of hepatic lipid partitioning in NASH pathogenesis [8-10]. Experimentally, hepatic macrophage accumulation has been associated with development of insulin resistance, while macrophages in adipose are associated with its persistence [11]. The capacity of adipose sites for lipid storage might play an indirect role in NASH pathogenesis [12], but whether adipose inflammation is also relevant to NASH pathogenesis remains unresolved.
In metabolically obese individuals, adipose inflammation is characterized by macrophages clustered around injured adipocytes as crown-like structures (CLSs) [13]. Identical CLSs are also prominent in NASH [14], but the mechanism(s) by which inflammatory cells are recruited to adipose in obesity/diabetes is still contentious [15, 16]. One idea is that rapid-onset over nutrition causes adipocyte hypertrophy rather than adaptive hyperplasia, and some adipocytes become stressed because of lipid engorgement [17]. This ultimately results in a reduced number of healthy small adipocytes [18], while large adipocytes activate chemokine/cytokine circuits to cause macrophage infiltration into the tissue [19].
Exercise lowers postprandial blood glucose levels with a concomitant decrease in serum insulin that slows progression to diabetes and improves glycemic control [20]. Resolution of hepatic steatosis and improved insulin sensitivity in extrahepatic sites have also been described [21]. Although limited in scope, lifestyle intervention studies in NAFLD infer that enhanced physical activity can improve liver histology [22, 23]. The present study was designed to clarify interactions between adipose function and inflammation, glycemic control, and pathogenesis of NASH and liver fibrosis. We used a mouse model that simulates the key predisposing factors to human NASH [24, 25], appetite dysregulation with overnutrition, diabetes, and metabolic syndrome, and the full spectrum of NAFLD from simple steatosis to fibrotic NASH depending on dietary composition [12]. Using exercise to ameliorate onset of insulin resistance, obesity, and diabetes, we studied detailed adipose morphometry coupled to indicators of differentiation and inflammation in order to dissect relationships between adipose compartments, tissue-specific insulin sensitivity, and liver pathology.
Alms1 mutant foz/foz mice are hyperphagic obese due to defective hypothalamic appetite dysregulation [12, 24, 25]. After weaning at 4 weeks age, groups (n = 8) of foz/foz NOD.B10 mice [bred from the original strain reported by Arsov et al. (24)] and wild-type littermates were fed either standard chow or an atherogenic diet ad libitum (4.78 kcal/g digestible energy; 23% fat, 45% carbohydrate, 0.19% cholesterol; Speciality Feeds, Glen Forrest, Australia). Mice were kept on 12-h light/dark cycle in the Australian National University (ANU) Medical School animal facility at the Canberra Hospital. Experimental procedures were approved by the ANU Animal Ethics Committee.
Half the cages were fitted with an exercise wheel (ASIFTB-PC; Able Scientific, Canning Vale, Australia) for voluntary exercise, and wheel rotations were recorded by a cycle computer (Bri2; Echowell, Taiwan). One week before sacrifice, we fasted mice 14-h, then measured glucose tolerance after intraperitoneal glucose injection (2 g/kg body weight) using a glucometer (Accu-Chek Advantage; Roche Diagnostics, Mannheim, Germany). At 16 weeks of age, mice were fasted 14-h, anesthetized (100 mg/kg ketamine, 16 mg/kg xylazine) and administered insulin (1 U/kg body weight; Eli Lilly, Indianapolis, IN) by intra-aortic injection. Liver, visceral (periovarian) and subcutaneous (lumbar) white adipose tissues (WATs), and gastrocnemius muscle were collected before and 3 min after insulin injection for further analyses.
Adipose and liver mRNA isolated by a combined protocol of TRI Reagent (Sigma-Aldrich, St. Louis, MO) and SV Total RNA Isolation System (Promega, Madison, WI) were measured with qRT-PCR analysis which was performed using iQ SYBR Green Supermix, conducting reactions in iQ5 thermal cycler (Bio-Rad Laboratories, Hercules, CA). Data normalization was performed using geometric mean of three housekeeping genes. Proteins isolated and estimated by Bradford assay [26], and then resolved with SDS-PAGE and phospho-/total blots (antibody information upon request) were visualized by chemiluminescence (Western Lightening Plus; Perkin-Elmer, Boston, MA). Protein bands were quantified by densitometry (MultiGauge; FujiFilm, Tokyo, Japan). Phosphorylation was expressed as phospho-/total protein ratio. Serum insulin levels were measured using a mouse insulin ELISA kit (Merck Millipore, Darmstadt, Germany). Serum alanine transaminase (ALT) was measured by colorimetric assay at hospital pathology department.
Formalin-fixed tissue samples were embedded in paraffin. Morphometry was performed on hematoxylin and eosin (H&E)-stained adipose sections (4 µm) using Leica Application Suite (LAS; Leica Microsystems, Wetzlar, Germany), using minimum of 10 fields (1 mm2). CLS numbers were obtained by analyzing 10 fields for each mouse (160× magnification) with results normalized to 100 adipocytes. H&E-stained liver sections were scored blinded by an experienced liver pathologist (MMY) according to the system devised for human NASH [27]. To quantify liver fibrosis, Masson's trichrome (POCD Healthcare, Sydney, Australia)-stained liver sections were analyzed with densitometry of collagen staining.
Data are presented as mean ± SEM. Protein/mRNA estimations were performed in duplicates. Significance of data was assessed by Prism 6 (GraphPad, La Jolla, CA) and SPSS Statistics 22 (IBM, New York, NY) softwares using the Student's t-test for single comparison and one-way or two-way analysis of variance (ANOVA), followed by Bonferroni's post hoc analysis. Group differences were considered significant when P < 0.05.
As measured by wheel rotations, genotype rather than dietary composition influenced physical activity. foz/foz mice tended to exercise less than wild-type mice which was significant on atherogenic diet (Table 1). Exercise increased food consumption in wild-type mice (Figure 1A,B). Interestingly, there was a nonsignificant trend for a reduction rather than increase in food intake in foz/foz mice with exercise, regardless of the type of diet (Figure 1A,B). The reason for this different food intake adaptation to exercise between the wild-type and foz/foz mice is now known.
From 4 weeks of age, foz/foz mice and wild-type (WT) littermates were housed without an exercise wheel for 12 weeks. (A,B) Compared with nonexercising mice (NO EX), food intake was higher in exercising (EX) wild-type mice on both diets, but less in exercising foz/foz mice. (C) Exercise corrected diet-induced weight gain in wild-type mice and prevented obesity development in chow (NC)-fed foz/foz mice, but (D) slowed rather than normalized weight gain in atherogenic (Ath) diet-fed foz/foz mice. (E) At 16 weeks, exercise caused a significant reduction in body weight in foz/foz mice, irrespective of diet. (F) Weight loss after overnight fasting was higher in all exercising groups. Data are mean ± SEM, with statistical analysis performed by two-way ANOVA (*P < 0.05 vs. sedentary control; †P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control).
| Genotype | Diet | n | Running time (h/day) | Average speed (km/h) | Distance (km/day) |
|---|---|---|---|---|---|
| |||||
| Wild-type | Chow | 8 | 3.75 ± 0.16 | 1.70 ± 0.18 | 6.47 ± 0.94 |
| foz/foz | Chow | 8 | 3.16 ± 0.20 | 1.83 ± 0.06 | 5.82 ± 0.48 |
| Wild-type | Atherogenic | 8 | 4.29 ± 0.24 | 2.13 ± 0.14 | 9.24 ± 0.94 |
| foz/foz | Atherogenic | 8 | 2.77 ± 0.37 | 1.66 ± 0.15 | 4.67 ± 0.85 |
Exercise mildly reduced weight gain in atherogenic-fed wild-type mice (Figure 1C,D). In chow-fed foz/foz mice, exercise prevented excess weight gain compared to nonexercising counterparts, which remained similar in weight to wild-type until ∼12 weeks of age (Figure 1C). Atherogenic feeding for 16 weeks strikingly increased body weight in foz/foz mice (Figure 1E). Exercise countered the sole genotype effect on body weight, and had significant but incomplete effect on reducing weight gain in atherogenic-fed foz/foz mice (Figure 1E). Overnight fasting weight loss (wheel removed) was significantly higher in all exercising groups, except chow-fed foz/foz mice (Figure 1F).
Physical activity did not alter gastrocnemius muscle mass in either wild-type or foz/foz mice, irrespective of diet (Figure 2A). Atherogenic diet and foz/foz genotype were each associated with increases in both visceral and subcutaneous WAT mass (P < 0.05). Exercise significantly reduced absolute visceral WAT in chow-fed foz/foz mice (Figure 2B) and subcutaneous WAT in atherogenic-fed foz/foz mice (Figure 2C) with similar trends of exercise in both adipose depots in all the other groups. Exercise improved muscle protein kinase B (AKT) phosphorylation in all groups except chow-fed wild-type (Figure 2D); thus postinsulin AKT-phosphorylation as a ratio of basal AKT increased with exercise (P < 0.05). However, exercise failed to enhance AKT-phosphorylation in either visceral or subcutaneous WATs (Figure 2E,F).
Tissue harvest was performed at 16 weeks of age after 12 weeks of diet, without exercise wheel provision. Tissue samples were collected before and 3 min after (contralateral side) intra-aortic insulin injection. (A) Exercise increased relative muscle mass in foz/foz mice on both diets. (B) Atherogenic (Ath) diet and foz/foz background were associated with increased visceral adipose weight, but exercise had minimal if any effect. (C) Exercise limited subcutaneous adipose expansion in foz/foz mice. Insulin-stimulated AKT-phosphorylation (as ratio of phospho-/total) was increased by exercise (D) in muscle, but not in (E) visceral (Vis) or (F) subcutaneous (Sub) WATs. (G) PPARγ, (H) GLUT4, and (I) PRDM16 mRNA expression in WAT was increased in foz/foz mice after 12 weeks of voluntary exercise. Data are mean ± SEM, by two-way ANOVA (*P < 0.05 vs. sedentary control; †P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control).
In wild-type mice, mRNA levels of peroxisome proliferator-activated receptor gamma (PPARγ), a transcription factor that regulates adipogenesis, adipose insulin sensitivity, and cell division [28], were similar in visceral and subcutaneous WAT and were not altered by exercise. However, in foz/foz mice, regular exercise increased PPARγ mRNA levels in WAT, a change that was highly significant in visceral WAT (Figure 2G). Levels of GLUT4 mRNA, the main insulin-regulated glucose uptake transporter in adipose, increased in a similar exercise-dependent pattern (Figure 2H). The expression of PRDM16, a transcription co-factor involved in promoting browning of adipocytes, was also higher in subcutaneous than visceral WAT in all groups; exercise countered the very low expression levels of PRDM16 mRNA in visceral adipose of foz/foz mice (P < 0.05; Figure 2I).
Fasting blood glucose (FBG) increased with atherogenic dietary feeding and foz/foz mutation, being highest in atherogenic-fed foz/foz mice (Figure 3A). Exercise failed to alter fasting hyperglycemia in any group (Figure 3A). However, the data for atherogenic-fed foz/foz mice appeared unduly influenced by two animals that developed severe diabetes (Figure 3B). In light of these ambiguous data, we established a small, separate cohort of atherogenic-fed foz/foz mice (n = 6/group) and monitored FBG over time (Figure 3F). The results indicate that substantial increase in FBG following a significant weight gain (Figure 3F; i) occurs abruptly during the 8th week of dietary feeding in nonexercising mice. All animals became prediabetic and this effect appears to be delayed in exercising atherogenic-fed foz/foz mice (Figure 3F; ii). Serum insulin (Figure 3F; iii) and the hepatic injury marker, ALT (Figure 3F; iv), levels were found less in exercising than sedentary atherogenic-fed foz/foz mice. Intraperitoneal glucose tolerance in chow-fed wild-type mice did not change with exercise. In chow-fed foz/foz mice, glucose tolerance was impaired, and this was substantially reversed by exercise (P < 0.05; Figure 3C,E). On the other hand, exercise wheel provision had little if any effect on impaired glucose tolerance in atherogenic-fed foz/foz mice (Figure 3D,E). Closer analysis revealed that this negative result was largely attributable to one cage in which animals were slow to start exercising (no rotations recorded in week 1) and overall use of the wheel was 2.5 weeks less; these are the two mice that developed diabetes (see Figure 3B) and NASH (see later).
(A) Fasting blood glucose (FBG) increased with atherogenic (Ath) feeding and foz/foz mutation; (B) exercise (EX) failed to improve FBG levels in atherogenic-fed foz/foz mice, largely because of two severely diabetic mice. Exercise was associated with a (C,E) striking improvement in intraperitoneal glucose tolerance in chow (NC)-fed foz/foz mice, and (D, E) moderate improvement in atherogenic-fed wild-type (WT) mice. (E) Glucose tolerance was severely impaired and did not improve with exercise in atherogenic-fed foz/foz mice, being partially influenced by the two diabetic mice (depicted in panel B). (F) Accordingly we set up a separate cohort of atherogenic-fed foz/foz female mice (n = 6) without an exercise wheel. (F-i) Similar to the main cohort, exercise significantly delayed weight gain. (F-ii) There was a break point at ∼8 weeks when the nonexercising (NO EX) mice developed significant hyperglycemia. (F-iii) Exercise appeared to delay this increase together with a strong trend in reduction of blood insulin levels (P = 0.1). (F-iv) Hepatocellular injury marker serum alanine transaminase (ALT) levels were found lower in exercising atherogenic-fed foz/foz mice. Data are mean ± SEM, with statistical analysis performed by two-way ANOVA (*P < 0.05 vs. sedentary control; †P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control).
In the present study, atherogenic dietary feeding increased mean adipocyte volume in the visceral compartment of wild-type mice (1.60 vs. 0.55 µm3 × 105, P < 0.05; Figure 4A), with similar changes in foz/foz mice on both diets (Figure 4A,B). In atherogenic-fed foz/foz mice, the size of enlarged visceral adipocytes did not alter with exercise. However, in exercising chow-fed foz/foz mice, adipocytes between 2000 and 5000 µm2 in size were a major population whereas adipocytes >9000 µm2 predominated in mice without an exercise wheel (P < 0.05; Figure 4C). The average volume of subcutaneous adipocytes also increased progressively according to diet and phenotype, the largest median volume being in nonexercising atherogenic-fed foz/foz mice. Exercise consistently reduced the mean volume of subcutaneous adipocytes in all groups (Figure 4B), and cell size distribution analysis revealed an increase in the subpopulation of small adipocytes in all exercising mice (Figure 4D).
Morphometry was performed on H&E-stained adipose sections (×160 magnification). (A) Average adipocyte volume in the visceral compartment increased with atherogenic (Ath) diet and foz/foz mutation; exercise did not cause a change. (B) The effect of exercise on subcutaneous adipocytes was more pronounced, and in addition, the average cell volume diminished in all groups. Adipocyte size distribution was assessed for visceral and subcutaneous WATs. (C) In exercising (EX) chow (NC)-fed foz/foz mice, adipocytes between 2000 and 5000 µm2 sizes were a major cell population in visceral pad, whereas adipocytes >9000 µm2 were predominant in mice without an exercise wheel. (D) A similar (more pronounced) pattern was noted for subcutaneous adipocyte size distribution. Data are mean ± SEM, with statistical analysis performed by two-way ANOVA (*P < 0.05 vs. sedentary control; †P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control).
In wild-type mice on either diet, there was no change in mRNA expression of the macrophage marker, cluster of differentiation 68 (CD68). Conversely, CD68 mRNA was abundant in visceral (and to a lesser extent subcutaneous) adipose of foz/foz mice irrespective of diet, an effect that was at least partially reversed by exercise (P < 0.05; Figure 5A). Levels of cluster of differentiation molecule 11b (CD11b) mRNA, a marker of inflammatory immune cells [29], were higher in nonexercising foz/foz mice adipose depots, except in chow-fed subcutaneous WAT (Figure 5B); exercise decreased CD11b transcript levels in all adipose tissues. Levels of the proinflammatory chemokine, monocyte chemoattractant protein 1 (MCP1) mRNA were likewise increased in foz/foz adipose, particularly in the visceral depot; this increase was limited by exercise (P < 0.05; Figure 5D).
(A) CD68 mRNA levels were high in foz/foz mice, particularly in those fed atherogenic (Ath) diet. This increase was more pronounced in visceral (Vis) than subcutaneous (Sub) WAT. The effects of atherogenic feeding on macrophage infiltration were ameliorated by exercise. A similar profile was found for (B) CD11b and (D) MCP1. (C) Crown-like structures (CLSs) are formed by a group of inflammatory macrophages surrounding an adipocyte. (E,F) CLS numbers (normalized to 100 adipocytes; ×160 magnification) were consistent with the other inflammatory markers, being highly abundant in inflamed visceral and subcutaneous WATs of foz/foz mice (red arrows); exercise (EX) reversed this effect in atherogenic-fed foz/foz visceral WAT. Data are mean ± SEM, by two-way ANOVA (*P < 0.05 vs. sedentary control; †P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control).
Contiguous clumps of macrophages, abutting a small adipocyte termed “crown-like structures” (CLSs, ×400 magnification; Figure 5C) are regarded as a marker of advanced WAT inflammation [13, 16]. As established by number of CLSs (normalized to 100 adipocytes), visceral WAT was highly inflamed in nonexercising atherogenic-fed foz/foz mice (Figure 5E,F); exercise decreased accumulation of CLSs at this site. Moderate inflammation was also observed in subcutaneous adipose of atherogenic-fed foz/foz mice (Figure 5E,F), and at this site as well exercise significantly diminished macrophage infiltration.
Liver histology remained normal in chow-fed wild-type mice (Table 2, Figure 6A). Atherogenic dietary feeding produced minor steatosis in two of eight wild-type mice but not in any exercising counterparts. In chow-fed foz/foz mice, numerous small fat droplets (microvesicular steatosis) were evident in hepatocytes with minor liver inflammation; exercise wheel provision prevented both steatosis and inflammatory recruitment in this group (Table 2; Figure 6A). As expected from previous studies [12, 30], atherogenic-fed foz/foz mice showed hepatomegaly with extensive steatosis, substantial inflammation and ballooned hepatocytes. The resultant median NAFLD Activity Score of 5 (range 4-6) indicated definite NASH in four instances [27], and borderline NASH in three other mice (Table 2). Exercise failed to prevent development of hepatomegaly in atherogenic-fed foz/foz mice (Figure 6B), but only three of eight mice provided with exercise wheel developed unequivocal NASH; including the two diabetic mice depicted in Figure 3B. Of the remainder, two developed borderline NASH and three showed only simple steatosis (Table 2, Figure 6A).
(A) Effects of atherogenic (Ath) diet, appetite defect (foz/foz), and exercise on liver histology in mice. H&E sections (×160 magnification). Atherogenic feeding caused steatosis in nonexercising (NO EX) wild-type (WT) mice, which was reversed by exercise. In foz/foz mice, steatosis is evident with normal chow (NC) and reversed by exercise, but NASH is present with atherogenic diet; exercise (EX) prevents NASH but simple steatosis is still present. Exercise failed (B) to prevent hepatomegaly development in atherogenic-fed foz/foz mice or (C) to influence insulin-stimulated AKT-phosphorylation in liver. (D-F) Collagen 1 positive tissue was less abundant in exercising atherogenic-fed foz/foz mice in comparison with sedentary counterparts. Correspondingly, (G) liver fibrosis markers collagen 1 and (H) alpha-smooth muscle actin (α-SMA) protein expression was lower in exercising atherogenic-fed foz/foz mice liver compared with nonexercising mice. Data are mean ± SEM. (†P < 0.05 vs. diet-matched control; ‡P < 0.05 vs. genotype-matched control by two-way ANOVA.)
| n | Steatosis score | Inflammation score | Ballooning score | NAS score# | Designation | |||
|---|---|---|---|---|---|---|---|---|
| ||||||||
| Normal chow | Wild-type | No exercise | 8 | 0 (−) | 0 (−) | 0 (−) | 0 (−) | Normal (8) |
| Exercise | 8 | 0 (−) | 0 (−) | 0 (−) | 0 (−) | Normal (8) | ||
| foz/foz | No exercise | 6 | 0.3 (0-1) | 0.17 (0-1) | 0 (−) | 0.4 (0-1) | Normal (4) | |
| Simple steatosis (1) | ||||||||
| Borderline NASH (1) | ||||||||
| Exercise | 8 | 0 (−) | 0 (−) | 0 (−) | 0 (−) | Normal (8) | ||
| Atherogenic diet | Wild-type | No exercise | 8 | 0.25 (0-1) | 0 (−) | 0 (−) | 0.25 (0-1) | Normal (6) |
| Simple steatosis (2) | ||||||||
| Exercise | 8 | 0 (−) | 0 (−) | 0 (−) | 0 (−) | Normal (8) | ||
| foz/foz | No exercise | 7 | 2.42 (2-3)†,‡ | 1.42 (1-2)†,‡ | 1.14 (1-2)†,‡ | 5 (4-6)†,‡ | Borderline NASH (3) | |
| Definite NASH (4) | ||||||||
| Exercise | 8 | 2 (1-3)†,‡ | 1.12 (1-2)†,‡ | 0.75 (0-2)†,‡,* | 3.87 (2-6)†,‡,* | Simple steatosis (3) | ||
| Borderline NASH (2) | ||||||||
| Definite NASH (3)a | ||||||||
We have previously reported development of substantial liver fibrosis in atherogenic-fed foz/foz mice after 24 weeks [12, 30]. Fibrosis is already evident at 16 weeks, as shown here by areas stained blue for collagen with Masson's trichrome (Figure 6D). By image analysis, fibrosis was significantly less in atherogenic-fed foz/foz mice provided with an exercise wheel than in nonexercising counterparts (P < 0.05; Figure 6E,F,D). The effects of exercise in ameliorating liver fibrosis were confirmed by decreased collagen 1A mRNA (not shown) and protein (P < 0.05; Figure 6G), and lower expression of alpha smooth muscle actin (α-SMA) (P < 0.05; Figure 6H).
Metabolic obesity results from interactions between lifestyle (environmental) factors that favor energy excess and a genetically predisposed host. In the present work, we studied the effects of exercise on complications of metabolic obesity in a mouse model that resembles the human condition by genetic predisposition to obesity and diabetes. The first new finding was that exercise promoted increased energy turnover sufficient to prevent both diet-induced and genetically-driven weight gain, but was less effective when these factors were together. In atherogenic-fed foz/foz mice, exercise slowed the rate of weight gain; however, at an apparent “set point,” use of the wheel declined and weight gain accelerated. Two individual animals that were slow to adopt regular exercise discontinued at an earlier time than six littermates; these two mice were the only ones to develop established diabetes by age 16 weeks. Despite this, the combined use of intraperitoneal glucose tolerance and tissue-specific AKT-phosphorylation after insulin administration, together with static measurements of serum insulin in the substudy, provide a clear picture of exercise-induced improvements in energy homeostasis.
Energy surplus requires WAT remodelling [17, 31]. Facing positive energy imbalance, adipocytes become lipid-engorged, and exhibit reduced glucose uptake and increased chemokine/cytokine secretion [32]. Conversely, healthy adipose contains a higher proportion of smaller adipocytes and shows no inflammation [33]. In the present study, both atherogenic diet and foz/foz background increased adipocyte volume, in both visceral and subcutaneous compartments. However, more detailed morphometric investigation showed a different size distribution of adipocytes: the percentage of cells between 2000 and 6000 µm2 (small to medium) was higher in exercising foz/foz mice, even though average adipocyte volume (spherical shape) did not change. Lipid-laden large adipocytes produce more pro-inflammatory chemokines/cytokines which regulate cellular trafficking [32, 33]. Atherogenic-fed foz/foz mice showed abundant immune cell infiltration, particularly into visceral WAT; exercise reversed this effect in association with lowered MCP1 production. Most of macrophages that infiltrate adipose assemble around very small (dying/injured) adipocytes in CLSs [13, 16]. In parallel with the morphometric and molecular findings, exercise substantially reduced CLS numbers in both visceral and subcutaneous WATs of atherogenic-fed foz/foz mice.
Improved muscle insulin sensitivity in exercising mice clearly contributed to greater energy consumption in these groups, as was evident by higher overnight fasting weight loss (energy utilization). Active muscle cells oxidize more energy and thereby diminish the systemic lipid burden in exercising mice. This energy is mostly provided by adipocytes because, for example, increasing adrenalin levels immediately before/during exercise activates hormone sensitive lipase (HSL) in adipose tissues. This causes increased lipolysis and energy release in nutrient forms from adipocytes. Autonomic activation of sympathetic nervous system may improve adipose metabolism as well. Nevertheless, direct effects of exercise on adipose tissue need further investigations.
Evidence of improved adipose function was indicated by changes in adipocyte size and expression of PPARγ which regulates glucose uptake and stimulates triglyceride storage [28, 34], with a corresponding increase of GLUT4 mRNA. Higher levels of PRDM16 with exercise, especially in subcutaneous WAT, refer to increased beige adipocytes which contain higher numbers of mitochondria [2, 35]. An important consequence of exercise-mediated improvement of adipocyte function for the development of other metabolic disorders, such as NAFLD, is that WAT sites become more responsive to circulating glucose and lipids, thereby reducing ectopic lipid deposition into liver and other cell types [21, 36].
Exercise has been shown to resolve steatosis in overweight humans with NAFLD [23, 37, 38], and this effect was reproduced in exercising chow-fed foz/foz and atherogenic-fed wild-type mice. In the present study, exercise-mediated suppression of adipose inflammation in foz/foz mice was also associated with other striking improvements in liver histology, including less inflammation and injury. The exceptions were two mice that failed to take up exercise early, discontinued it prematurely and developed diabetes. Exercise also consistently lowered liver fibrosis in atherogenic-fed foz/foz mice. In a transectional study, moderately vigorous (but not lesser) exercise was associated with improved liver indices and less severe liver fibrosis by an unclear mechanism [22]. The present study in a metabolic syndrome model of NASH with genetic predisposition to diabetes should prove suitable to clarify this protective mechanism. The possibilities include a direct relationship to reduced hepatocyte injury (by lipotoxicity) and liver inflammation, stabilization of hepatic stellate cells (shown by decreased α-SMA expression), and decreased levels of profibrogenic growth factors and other humoral mediators.
In conclusion, these data clarify the mechanism by which moderately vigorous exercise suppresses adipose inflammation in mice with disordered appetite control. Exercise prevented excessive weight gain in chow-fed foz/foz and atherogenic-fed wild-type mice and slowed weight gain in atherogenic-fed foz/foz mice in association with increased proportion of small-medium sized adipocytes. Exercise enhanced insulin signaling in muscle, but not in liver or adipose. Even in atherogenic-fed foz/foz mice, exercise suppressed the otherwise abundant CLSs of macrophages and inflammatory transcripts, particularly in visceral but also in subcutaneous WAT. In association with prevention of diabetes, exercise reduced steatosis and improved hepatocyte ballooning, liver inflammation (less NASH), and fibrosis in atherogenic-fed foz/foz mice. Thus, in appetite-dysregulated mice fed an energy-dense diet, exercise improved muscle insulin sensitivity to delay onset of diabetes, maintain adipose function, reduce adipose inflammation, and ameliorate NASH and hepatic fibrosis, the important liver complications of obesity and diabetes. In addition to the benefits of lifestyle modification for prevention and treatment of prediabetes and significant liver disease, these findings have implications for drug targets that could lead to similar benefits in nonexercising overweight humans.
The authors thank Canberra Hospital research office and animal house technicians for their highly skilled technical assistance. Dr. Bruce Shadbolt provided invaluable advice on statistical analyses. Part of this research was presented at the 74th Scientific Sessions of the American Diabetes Association, June 2014.
