Hepatic Glucagon Action Is Essential for Exercise-Induced Reversal of Mouse Fatty Liver

The Vanderbilt University Animal Care and Use Committee approved all procedures. BL6 mice were from Jackson Laboratories (Bar Harbor, ME). Gcgr^+/− mice are >98% BL6 confirmed by strain analysis (Charles Rivers, Indianapolis, IN) and were bred to generate gcgr^+/+ and gcgr^−/− littermates. Mice were housed in a temperature-controlled environment on a 12:12-h light-dark cycle (lights on 6:30 a.m. to 6:30 p.m.) and fed chow (5001; Purina Mills, St. Louis, MO) composed of 13, 58, and 28 kcal/g of fat, carbohydrate, and protein, respectively, or HFD (F3282; BioServ, Frenchtown, NJ) composed of 60, 24, and 16 kcal/g of fat, carbohydrate, and protein, respectively.

Male BL6 mice were fed chow or HFD beginning at 6 weeks. Liver fat was assessed using thin layer chromatography, MR, and histological techniques after 2, 4, 8, 12, and 20 weeks of HFD.

Liver fat was quantified using a 7T Varian Inova MR imager/spectrometer. Mice were anesthetized using isoflurane, positioned with the upper abdomen in the center of a 25 mm transmit/receive RF coil, and prepared for continuous respiratory monitoring. After introducing the animal to the magnet isocenter and tuning the coil, gradient-echo scout images were obtained. An ∼1-cm³ voxel was defined and shimmed to a line width of 25–30 Hz. T₁-weighted images were obtained using a 32-mm square field of view, 256 × 256 matrix, and fourteen 1-mm slices (0.125 × 0.125 × 1.0 mm resolution). The 14 slices were obtained in two packets of 7 slices; each had a 1-mm gap between slices, and the 2 slice packets were offset by 1 mm. Anesthesia was adjusted to maintain a respiratory period of ∼2 s and gate the MR acquisition. Gradient-echo images were acquired using two excitations with direct current offset correction and the same geometric parameters as the anatomical images, except that the acquired matrix size was 64 × 64 and the reconstructed matrix size was 128 × 128 (0.25 × 0.25 × 1.0 mm resolution). Complex images were obtained at three echo time values (3.09, 3.6065, and 4.123 ms), which correspond to in-phase, 180° out-of-phase, and in-phase fat-water acquisitions, respectively.

Gcgr^+/+ and gcgr^−/− mice were fed HFD from 6 to 12 weeks of age. At 12 weeks, liver fat was assessed using MR and mice were assigned to cages with an operable running wheel (RW group) or an inoperable running wheel (sedentary or SED group; Mini Mitter, Bend, OR) for 10 weeks. Wheel rotations were counted using a magnetic counter. Steel bars were used to render wheels inoperable. Body weight and wheel rotations were assessed weekly. Body composition was analyzed biweekly. Liver fat was assessed using MR during weeks 5 and 10. Blood was taken from the cut tail at weeks 0, 5, and 10 to measure glucose and insulin. Wheels in RW groups were fixed in place 2 days prior to killing to limit acute effects of exercise. Exercise stress tests were performed as described (20–22).

Gcgr^+/+ and gcgr^−/− mice were fed HFD from 6 to 12 weeks of age. At 12 weeks, liver fat was assessed using MR and mice were assigned to treadmill (EX) or SED groups. EX mice performed exercise (3 min at 10 m/min followed by 30 min at 20 m/min) 5 days per week for 6 weeks between 10:00 a.m. and 12:00 p.m. SED mice were placed in a nonmoving treadmill for 33 min 5 days per week for 6 weeks between 10:00 a.m. and 12:00 p.m. Liver fat was assessed using MR, and blood was taken from the cut tail to measure glucose and insulin at 3 and 6 weeks. On the last day of exercise, both EX and SED mice performed the Vo_2max protocol to control for the acute effect(s) of exercise. Mice were killed 2 days later. A cohort of chow-fed, chronically catheterized gcgr^+/+ and gcgr^−/− mice were also run until exhaustion using a previously described protocol (23) to investigate treadmill exercise-induced changes in catecholamines.

Mice were fasted 3 h beginning at 6:00 a.m., and blood glucose was measured from the cut tail using a glucometer. Mice were killed using sodium pentobarbital, and blood was taken via cardiac puncture. Hormones and catecholamines were measured in the Vanderbilt Mouse Metabolic Phenotyping Center Hormone Assay and Analytical Resources Core (24,25). Plasma free fatty acids (FFAs), alanine aminotransferase (ALT), triglycerides (TGs), and cholesterol were measured using kits. Hepatic lipids were extracted using the method of Folch et al. (26). Extracts were filtered and recovered in the chloroform phase and separated by thin layer chromatography using Silica Gel 60 A plates developed in petroleum ether, ethyl ether, and acetic acid (80:20:1) and visualized by rhodamine 6G. Liver phospholipids (PLs), diglycerides (DGs), TGs, and cholesterol esters (CEs) were scraped from plates and methylated using BF3/methanol (27). Methylated FFAs were extracted and analyzed using an Agilent 7890 gas chromatograph equipped with flame ionization detectors and a capillary column (SP2380, 0.25 mm × 30 m, 0.25 µm film; Supelco, Bellefonte, PA). Helium was used as a carrier gas. Oven temperature increased from 160°C to 230°C at 4°C/min. FFA methyl esters were identified by comparison with retention times of known standards, and odd chain fatty acids were included to quantify concentration. Dipentadecanoyl phosphatidylcholine (C15:0), diheptadecanoin (C17:0), trieicosenoin (C20:1), and cholesteryl eicosenoate (C20:1) were used as standards. Liver fat was also assessed using Oil Red O staining graded by a blinded pathologist. Adenine nucleotides were measured using high-performance liquid chromatography (HPLC) (23). AMP-activated protein kinase (AMPK), peroxisome proliferator–activated receptor-α (PPARα), and fibroblast growth factor 21 (FGF21) protein content or gene expression were done as previously described (19,23).

Energy charge was calculated as [ATP + (ADP/2)/ATP + ADP + AMP]. Statistical comparisons were made using one-way ANOVA followed by the Fisher least significant differences test for post hoc comparisons or t tests. Data are presented as means ± SE. Statistical significance was defined as P < 0.05.

MR data were analyzed using Matlab 7.1 (The Mathworks, Inc., Natick, MA). Gradient-echo images were processed according to the three-point Dixon method (28).

At 6 weeks of age, BL6 mice were fed chow or HFD to assess fatty liver. Whole-body fat mass increased in mice fed HFD beginning at 4 weeks compared with the preceding time point and duration-matched chow controls (Fig. 1A). Whole-body fat mass was unchanged in chow mice over time (Fig. 1A). Whole-body lean mass in mice fed HFD progressively increased and was greater than chow controls at 4, 12, and 20 weeks (Fig. 1B). Whole-body lean mass also progressively increased in chow controls (Fig. 1B). Liver mass in mice fed HFD was unchanged at all points except 20 weeks and did not vary in chow mice (Fig. 1C). Hepatic TGs, DGs, and CEs in mice fed HFD increased between each measurement after 4 weeks and versus chow controls (Fig. 1D–F). In contrast, hepatic PLs in mice fed HFD were comparable to chow controls at 2, 4, and 8 weeks but were decreased at 12 and 20 weeks (Fig. 1G). Hepatic TGs, DGs, CEs, and PLs were unchanged in chow mice (Fig. 1D–G).

Blood glucose in BL6 mice fed HFD increased at 8 weeks versus earlier points and was elevated relative to chow controls at all points despite declining between 8 and 20 weeks (Fig. 1H). Blood glucose in chow controls was stable (Fig. 1H). Insulin, leptin, and ALT progressively increased versus preceding measurements and chow controls that did not change (Fig. 1I–K).

In agreement with biochemical assessments, MR and histology found that liver fat in mice fed HFD increased over time and was elevated compared with chow controls at 8, 12, and 20 weeks (Fig. 2A–D). Hepatic fat assessed by MR and histology in chow controls were <5% at all points (Fig. 2A and B). Assessment of hepatic fat by MR correlated well with biochemical measurements (r² = 0.86). Histological assessment did not correlate as well (r² = 0.62) (Fig. 2E and F).

Gcgr^+/+ and gcgr^−/− mice were fed HFD for 6 weeks and subsequently housed with operable (RW) or inoperable (SED) running wheels for 10 weeks. Body weights were comparable prior to introduction of running wheels (Fig. 3A). Body weight increased in all groups, and there were no differences at any time point in gcgr^+/+ mice (Fig. 3A). In contrast, body weight was reduced in gcgr^−/− RW mice versus SED controls between 7 and 9 weeks (Fig. 3A). Total fat mass was similar between groups prior to introduction of running wheels and increased after 8 and 10 weeks in both SED groups (Fig. 3B). Total lean mass was unchanged in all groups (Fig. 3C). There were no differences in running wheel activity (532,432 ± 71,856 and 609,659 ± 59,592 revolutions at 10 weeks in gcgr^+/+ RW and gcgr^−/− RW mice, respectively).

Blood glucose was lower in gcgr^−/− mice compared with gcgr^+/+ littermates before introduction of running wheels (Fig. 3D). Glucose increased over time in SED groups (Fig. 3D). Glucose was unchanged over time in RW mice but was consistently higher in gcgr^+/+ RW compared with gcgr^−/− SED littermates (Fig. 3D). Insulin was also lower in gcgr^−/− mice compared with gcgr^+/+ littermates prior to training (Fig. 3E). Insulin also increased over time in SED groups (Fig. 3E). Leptin levels were lower in gcgr^−/− mice compared with gcgr^+/+ littermates after 6 weeks of HFD and increased in both SED groups (Fig. 3F). Leptin levels did not increase in RW mice and were lower than SED controls at 10 weeks (Fig. 3F). Plasma FFAs were lowered in gcgr^+/+ RW compared with gcgr^−/− SED littermates (Fig. 3G). Plasma FFAs did not differ between gcgr^−/− mice and were comparable to gcgr^+/+ SED mice (Fig. 3G). Plasma TGs and cholesterol at the final time point were comparable in all mice (Fig. 3H and I).

Liver fat assessed by MR after 6 weeks of HFD was similar between groups (Fig. 4A). Measures of liver fat at 5 and 10 weeks indicate progressive increases in SED mice (Fig. 4A). Liver fat was, however, reduced in gcgr^+/+ RW mice (Fig. 4A). It is remarkable that this effect was abolished in gcgr^−/− RW mice (Fig. 4A). Liver mass was reduced in gcgr^+/+ RW mice (Fig. 4B). In agreement with MR data, liver TGs and DGs at killing were similar in SED mice. Liver TGs and DGs were reduced in gcgr^+/+ RW mice, and there was again no effect of exercise in the absence of the glucagon receptor (Fig. 4C and D). Liver CEs and PLs were not different between groups (Fig. 4E and F).

Hepatic AMP increased in gcgr^+/+ RW mice versus gcgr^+/+ SED controls (Fig. 4G). Hepatic adenine nucleotides were, however, comparable in both gcgr^−/− groups and not different from gcgr^+/+ SED mice (Fig. 4G). Hepatic energy charges are shown in Fig. 4H. In contrast, gastrocnemius ATP was decreased, and AMP increased in RW groups versus SED controls (Fig. 4I). Muscle energy charges are noted in Fig. 4J. Hepatic p-AMPK^Thr172/AMPK and expression of AMPK-α1/-α2, PPAR-α, and FGF21 were elevated in gcgr^+/+ RW mice versus gcgr^+/+ and gcgr^−/− SED mice. It is striking that effects of exercise were again abolished in gcgr^−/− RW mice (Fig. 4K and L).

A second cohort of gcgr^+/+ and gcgr^−/− littermate mice were fed HFD for 6 weeks and then performed 30 bouts of exercise over 6 weeks (5 days per week). Preintervention Vo_2max was comparable between genotypes and similarly increased with training (Fig. 5A). Body weight was similar between groups after 6 weeks of HFD (Fig. 5B). Body weight increased in SED mice and was higher than EX mice in each genotype after training (Fig. 5B). Total fat mass was similar between genotypes prior to training and increased at 6 weeks in SED groups compared with initial measurements (Fig. 5C). Total fat mass was unchanged in EX mice (Fig. 5C). Total lean mass was unchanged in all groups (Fig. 5D).

Blood glucose was lower in gcgr^−/− mice compared with gcgr^+/+ littermates after 6 weeks of HFD (Fig. 5E). Blood glucose increased over time in SED groups but was unchanged in EX mice (Fig. 5E). Insulin was lower in gcgr^−/− mice compared with gcgr^+/+ littermates after 6 weeks of HFD (Fig. 5F). Insulin was lower in EX mice versus SED littermates after training (Fig. 5F). Plasma FFAs, TGs, cholesterol, epinephrine, and norepinephrine measured at killing were comparable in all mice (Fig. 5G–K). Plasma catecholamines were similarly elevated in gcgr^+/+ and gcgr^−/− mice after exhaustive treadmill exercise (Fig. 5L and M).

Liver fat assessed by MR after 6 weeks of HFD was similar between genotypes (Fig. 6A). Repeat measures after 15 and 30 days of exercise indicate progressive increases in SED mice (Fig. 6A). In contrast, liver fat was reduced after 15 and 30 days in gcgr^+/+ EX mice compared with SED controls and earlier measurements (Fig. 6A). As in gcgr^−/− RW studies, liver fat was not reduced in gcgr^−/− EX mice (Fig. 6A). Liver mass was also reduced in gcgr^+/+ EX mice (Fig. 6B). Liver TGs and CEs were reduced in gcgr^+/+ EX mice compared with other groups (Fig. 6C and D). Liver PLs and DGs were similar in all groups (Fig. 6E and F).

Hepatic ATP was decreased and AMP increased in gcgr^+/+ EX mice compared with gcgr^+/+ SED controls (Fig. 6G). The effects of treadmill exercise on these variables were lost in the absence of the glucagon receptor (Fig. 6G). The corresponding hepatic energy charges are noted in Fig. 6H. Gastrocnemius ATP was decreased and AMP increased in both EX genotypes compared with SED (Fig. 6I). Corresponding muscle energy charges are shown in Fig. 6J. Hepatic p-AMPK^Thr172/AMPK and expression of AMPK-α1/-α2, PPAR-α, and FGF21 were also elevated in gcgr^+/+ EX mice versus SED controls regardless of genotype. These adaptations of exercise were lost in gcgr^−/− mice (Fig. 6K and L).

The current goal was to test whether exercise requires hepatic glucagon action to provoke reductions in HFD-induced fatty liver. Gcgr^+/+ and gcgr^−/− mice were fed HFD to induce moderate fatty liver, and exercise interventions were used in conjunction with MR to quantify effects on hepatic fat in vivo. The salient findings are that exercise-induced reductions in fatty liver 1) occur independently from changes in body weight and 2) require hepatic glucagon receptor activation.

The fatty liver time course was performed because insufficient data exist regarding HFD-induced fatty liver in mice. This is surprising because HFD in combination with inbred and/or genetically modified mice are used to examine metabolic disease and specific genes/pathways. Moreover, mouse models including those fed HFD are used to study fatty liver (29). Work in BL6 mice fed HFD up to 50 weeks illustrates obesity and metabolic impairment prior to nonalcoholic steatohepatitis (NASH) (30). These data were of limited utility, however, because the first reported time point was after 10 weeks of HFD. Our data illustrate that HFD increases liver TGs, DGs, and CEs as early as after 2 weeks. Striking elevations occur after 8–12 weeks, and further elevations are evident after 20 weeks. These findings, along with concomitant obesity, hyperglycemia, hyperinsulinemia, hyperleptinemia, and elevated ALT, indicate multiple metabolic impairments that would have likely worsened if mice were fed HFD for >20 weeks (30).

The time course data provide no evidence of hepatic microsteatosis, inflammation, or fibrosis after 2–12 weeks on HFD. Such features are characteristic of more severe hepatic dysfunction such as NASH. This conclusion is consistent with data on the Mouse Phenome Database (http://www.jax.org/phenome), arguments that a “second hit” or methionine- and choline-deficient diet is required for NASH in mice (29), or data showing that substitution of palm oil for lard in HFD is associated with NASH (31). Mice in our study did show features of NASH after 20 weeks of HFD (microsteatosis and elevated ALT). It is reasonable to conclude that HFD exposure for >20 weeks is necessary to produce features of NASH in BL6 mice. It is notable that insulin and leptin increase between 12 and 20 weeks on HFD. Studies beyond our focus are needed to investigate NASH and the relationship, if any, with hormonal changes and effect(s) of exercise. It is also important to clarify that these outcomes would likely vary using different diets and/or inbred strains. The current diet and strain were selected because it is a well-characterized model of metabolic disease. BL6 mice also exhibit moderate running wheel activity versus other strains (32) and have been studied using treadmill exercise (20–22).

The MR protocol was developed because it was valuable to perform repeated, noninvasive measures of hepatic fat. Biochemical analyses were considered the gold standard to validate the MR technique. These data show that MR is effective to quantify hepatic fat and can be successfully applied in vivo. This finding is in agreement with work in ob/ob mice or animals fed a methionine- and choline-deficient diet (33). It is also noteworthy that histological assessments were unreliable to quantify hepatic fat because they overestimate fat accumulation.

The intervention studies were designed to understand exercise and requirements for hepatic glucagon action as a modality to treat fatty liver. Work in rats shows that concurrent exercise and HFD limits development of fatty liver (8). Recent work in mice fed HFD also shows that 5 weeks of concurrent treadmill exercise during HFD prevents increases in body weight (34). It is speculated that the current intervention using 6 weeks of HFD to induce moderate fatty liver and then introducing exercise training is more consistent with treatment of the human condition. The 6-week time point was selected based on HFD time course data that indicates this duration is sufficient to increase liver lipid while limiting other metabolic defects including hyperinsulinemia. Additional studies are needed to test exercise interventions and/or requirements for hepatic glucagon action after more serious and/or numerous metabolic dysfunction(s) exist.

Voluntary and forced exercise paradigms were used because each has advantages/disadvantages. Running wheels heighten physical activity and are less stressful but can be variable. Forced exercise can define work but is associated with less volume and may be more stressful. Nonetheless, both paradigms improve insulin sensitivity, reduce hepatic fat, and alter liver gene expression (6,9–12). Our group has also shown that acute exercise increases plasma glucagon (20,23). The current findings in gcgr^+/+ mice emphasize that both strategies lower hepatic TGs, DGs, and CEs. It is interesting that MR data at the midpoint of each intervention reveal that reductions in hepatic fat occur before changes in body weight. This point is unexpectedly reinforced by findings that gcgr^+/+ RW mice exhibit reductions in liver fat despite no reduction in body weight. These data support human studies indicating that heightened physical activity reduces fatty liver independent of weight loss (2,35,36). It is likely that body weight would diverge if longer intervention periods were studied. However, exercise-induced weight loss would be expected to further reduce fatty liver.

The finding that exercise requires hepatic glucagon receptor activation to lower hepatic fat content is a step toward understanding the benefits of regular exercise. Collectively, these data suggest that loss of hepatic glucagon action has specific effects to negate hepatic lipid-lowering effects of exercise but does not negatively affect other aspects of exercise interventions. Exercise-induced effects to attenuate HFD-induced increases in body weight, adiposity, blood glucose, insulin, and leptin remain intact in gcgr^−/− mice. It is important to note that both genotypes had similar pre- and postexercise Vo_2max and ran similar distances on running wheels. This indicates that a diminished aerobic capacity or exercise behavior in gcgr^−/− mice does not account for our results. We did not investigate if gcgr^−/− mice exhibit differences in running wheel speed or exercise duration. Such differences are possible but unlikely based on similar exercise-induced decrements in muscle energy charge indicating comparable stress. It is also noteworthy that exercise provoked a modest reduction in body weight in gcgr^−/− RW mice but not in gcgr^+/+ RW littermates. This finding is interesting considering the parallel reduction in fatty liver. However, our treadmill data in which body weight is similarly reduced in EX groups adds confidence to the current conclusions. It is important to the interpretation of these data to appreciate that gcgr^−/− mice are characterized by lower blood glucose, reduced plasma insulin, hyperglucagonemia, and elevated glucagon-like peptide 1 levels (19,23,37,38). We cannot exclude the possibility that there is an unpredictable interaction between these aspects of the gcgr^−/− mice and regular exercise that is not evident in gcgr^+/+ controls. It should be noted, however, that gcgr^−/− mice have been extensively characterized (19,23,37–39); these previous studies validate that effects on the liver are due to loss of glucagon signaling. Decreased hepatic energy charge, activation of AMPK, and increased expression of PPAR and FGF21 are also consistent with increased hepatic glucagon action (19,23,38). There were also no differences in plasma catecholamines, another cAMP-dependent protein kinase agonist in the liver, in response to exhaustive treadmill exercise or after training when comparing gcgr^+/+ and gcgr^−/− mice.

These experiments also provide additional mechanistic insight to understand how repeated bouts of exercise-stimulated hepatic glucagon action lower liver fat content. Postexercise findings in gcgr^+/+ mice show lowered hepatic energy state, increased p-AMPK^Thr172/AMPK, and elevated expression of AMPK-α1/-α2, PPAR-α, and FGF21 compared with SED controls and gcgr^−/− mice. AMPK, PPAR-α, and FGF21 are targets of hepatic glucagon action and key proteins involved in oxidative metabolism (19,23,38,40–43). Stimulation of these interrelated pathways suggests that repeated bouts of exercise-stimulated hepatic glucagon action heighten fat oxidation. Loss of hepatic glucagon action did not affect circulating TGs or cholesterol. These findings concur with work demonstrating that glucagon does not impact hepatic TG synthesis or secretion (38).

Taken together, the present results demonstrate that consuming HFD in BL6 mice provokes rapid and progressive fatty liver that is reversible by exercise in a body weight–independent but glucagon receptor–dependent manner. The fact that exercise lowered hepatic fat content without marked changes in body weight is important to highlight the potential benefits of physical activity on fatty liver even if weight reduction is not achieved. It is also important to consider that hepatic glucagon action in the context of type 2 diabetes is dysregulated and contributes to hyperglycemia. This defect is the basis for ongoing efforts to antagonize the receptor pharmacologically. Extension of the current data to this potential therapy is cautionary because they suggest that successful antagonism of hepatic glucagon action may impair glucagon’s “positive” effects to regulate fat oxidation.

This work was supported by National Institutes of Health grants R01-DK-050277 and U24-DK-59637.

No potential conflicts of interest relevant to this article were reported.

E.D.B. designed the experiments, researched data, and wrote the manuscript. D.G.L. and R.A.B. researched data. C.M.H. and R.S.L.-Y. researched data and reviewed and edited the manuscript. E.P.D. and S.E.L. researched data. L.L.S. researched data and reviewed and edited the manuscript. M.J.C. provided glucagon receptor null mice and reviewed and edited the manuscript. B.M.D. developed the magnetic resonance technique and reviewed and edited the manuscript. D.H.W. designed the experiments and reviewed and edited the manuscript.

Parts of this study were presented in abstract form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010.

The authors thank Bakula Trevedi, Eric Allen, and the other staff of the Vanderbilt Mouse Metabolic Phenotyping Center Hormone Assay Core.