Dissipating Excess Energy Stored in the Liver Is a Potential Treatment Strategy for Diabetes Associated With Obesity

Murine UCP1 cDNA (10) was provided by Professor Leslie P. Kozak (Pennington Biomedical Research Center). Murine liver carnitine palmitoyltransferase 1 (CPT1a) cDNA was obtained by RT-PCR with liver total RNA and primers designed from the reported sequence (GenBank accession no. NM_013495). Recombinant adenovirus, containing murine UCP1 (11) or CPT1a cDNA under the CAG promoter, was prepared as described previously (12). A recombinant adenovirus bearing the bacterial β-galactosidase gene (Adex1CAlacZ) (13) was used as a control.

Animal studies were conducted under protocols in accordance with the institutional guidelines for animal experiments at Tohoku University. Male C57BL/6N mice were housed individually and divided into high-fat diet (32% safflower oil, 33.1% casein, 17.6% sucrose, and 5.6% cellulose [14]) and standard diet (65% carbohydrate, 4% fat, and 24% protein) groups at 5 weeks of age, when body weights were 21.2 ± 0.25 g (means ± SE). Four weeks after separation, body weight–matched mice for each group received an injection of adenovirus via the tail vein. Viruses were administered intravenously at a dose of 2 × 10⁸ plaque-forming units. For pair-feeding experiments, after 4 weeks of high-fat diet, mice were allotted into three groups. Two groups of mice received an injection of UCP1 or LacZ adenovirus. After 24 h, mice in the third group received an injection of LacZ adenovirus. The latter LacZ mice were given their daily food allotments on the basis of the previous day’s consumption by UCP1 mice.

UCP1, acetyl-CoA carboxylase 1 (ACC 1), and insulin receptor antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The α-subunit of AMP-activated protein kinase (AMPK), phospho-AMPK (Thr172), and phospho-ACC (Ser79) antibodies were purchased from Cell Signaling Technology (Beverly, MA). Affinity-purified antibody against insulin receptor substrate 1 (IRS1) was prepared as described previously (15).

Tissue samples were prepared as previously described (9), and tissue protein extracts (250 μg of total protein) were boiled in Laemmli buffer that contained 10 mmol/l dithiothreitol and subjected to SDS-PAGE. The immunoblots were visualized with an enhanced chemiluminescence detection kit (Amersham, Buckinghamshire, U.K.).

Frozen livers were homogenized, and triglycerides were extracted with CHCl₃:CH₃OH (2:1, vol:vol), dried, and resuspended in 2-propanol (16). Triglyceride contents were measured using Lipidos liquid (TOYOBO, Osaka, Japan).

Oxygen consumption was measured with an O₂/CO₂ metabolism measuring system (model MK-5000RQ; Muromachikikai, Tokyo, Japan). Each mouse was kept unrestrained in a sealed chamber with an air flow of 0.5 l/min for 5 h at 25°C without food or water during the light cycle. Air was sampled every 3 min, and the consumed oxygen concentration (Vo₂) was calculated.

Livers as well as epididymal fat (white adipose tissue) and brown adipose tissues were removed and fixed with 10% formalin and embedded in paraffin. Tissue sections were stained with hematoxylin and eosin. Total adipocyte areas were traced manually and analyzed. Brown and white adipocyte areas were measured in 100 or more cells per mouse in each group.

Rectal temperature was measured with a Thermalert TH-5 (Physitemp, Clifton, NJ).

The ATP levels in liver homogenates were measured with a luciferase-luciferin system (17) by using an ATP determination kit (Molecular Probes, Eugene, OR).

Livers were homogenized, and aliquots of supernatant were incubated with anti-AMPK α-subunit antibody. AMPK activity in the immunoprecipitates was assessed as a function of SAMS peptide phosphorylation, as previously described (18).

Mice that were fasted for 16 h received an injection of 100 μl of normal saline (0.9% NaCl), with or without 10 units/kg body wt insulin, via the tail vein. Hindlimb muscles were removed 300 s later and immediately homogenated. After centrifugation, the resultant supernatants were used for immunoprecipitation with anti–insulin receptor or anti-IRS1 antibody. Immunoprecipitates were subjected to SDS-PAGE and then immunoblotted using anti-phosphotyrosine antibody (4G10) or individual antibodies as described previously (15).

Blood glucose was assayed with Antsense II (Horiba Industry, Kyoto, Japan). Serum insulin and leptin were determined with ELISA kits (Morinaga Institute of Biological Science, Yokohama, Japan). Serum adiponectin and tumor necrosis factor-α (TNF-α) concentrations were measured with an ELISA kit (Ohtsuka Pharmaceutical, Tokyo, Japan) and a TNF-α assay kit (Amersham Biosciences, Uppsala, Sweden), respectively. Serum total cholesterol, triglyceride, and free fatty acid concentrations were determined with a Cholescolor liquid, Lipidos liquid (TOYOBO), and NEFA C (Wako Pure Chemical, Osaka, Japan) kits, respectively.

Glucose tolerance tests were performed on fasted (10 h) mice. Mice were given oral glucose (2 g/kg body wt), and blood glucose was assayed immediately before and at 15, 30, 60, and 120 min after administration. Insulin tolerance tests were performed on fed mice. Mice received an injection of human regular insulin (0.75 units/kg body wt; Eli Lilly, Kobe, Japan) into the intraperitoneal space, and blood glucose was assayed immediately before and at 20, 40, 60, and 80 min after injection. Leptin tolerance tests were performed as reported previously (19) with slight modification. Fasted (12 h) mice received an injection of mouse leptin (7.2 mg/kg body wt; R&D Systems) into the intraperitoneal space, and food intake amounts for 12 h thereafter were determined. Ratios of food intake amounts to those of vehicle-injected mice were calculated.

Total RNA was isolated from 0.1 g of mouse hepatic tissue with ISOGEN (Wako Pure Chemical), and cDNA synthesis was performed with a Cloned AMV First Strand Synthesis Kit (Invitrogen, Rockville, MD) using 5 μg of total RNA. cDNA synthesized from total RNA was evaluated in a real-time PCR quantitative system (Light Cycler Quick System 350S; Roche Diagnostics, Mannheim, Germany). The relative amount of mRNA was calculated with glyceraldehyde-3-dehydrogenase mRNA as the invariant control. The primers used are described in Table 1.

All data were expressed as means ± SE. The statistical significance of differences was assessed by the unpaired t test and one-factor ANOVA.

Hepatic UCP1 expression increased energy expenditure and reduced body weight and blood glucose levels in mice that had high-fat diet–induced obesity and diabetes. C57BL/6 mice were on a high-fat diet for 4 weeks, resulting in diabetes associated with obesity. The UCP1 adenovirus vector (11) was then administered intravenously (UCP1 mice). Mice that were given the LacZ adenovirus were used as a control (LacZ mice). No significant alterations were observed in body weights (Fig. 1B), blood glucose levels (Fig. 1C), food intake amounts, body temperature, or plasma lipid parameters (data not shown) before versus after LacZ adenovirus administration. Systemic infusion of recombinant adenoviruses into mice through the tail vein primarily resulted in expression of transgenes in the liver, with no detectable expression in peripheral tissues such as muscle, fat, kidney, or brain (data not shown), as reported previously (20). As shown in Fig. 1A, immunoblotting revealed that ectopic UCP1 expression in the liver peaked on day 3. Maximal expression was maintained through day 8. After day 9, hepatic expression of UCP1 decreased, and very small amounts of UCP1 protein were detected on day 14 (Fig. 1A).

In UCP1 mice, body weight and blood glucose levels were markedly decreased (Fig. 1B and C) concomitantly with hepatic expression levels of UCP1. On day 7, body weights of UCP1 mice were significantly lower, by 13%, than those of control mice. After day 9, body weight and blood glucose levels began to increase as the expression of hepatic UCP1 declined. These findings indicate that hepatic UCP1 expression exerted therapeutic effects on diabetes associated with diet-induced obesity.

Resting oxygen consumption on day 3 was markedly increased, by 12%, in UCP1 mice compared with controls (Fig. 1D), whereas rectal temperature did not differ between the two (Fig. 1E). Thus, ectopic UCP1 in the liver, like endogenous UCP1 in brown adipocytes, promoted inefficient metabolism, thereby enhancing energy expenditure and leading to weight reduction. This effect, however, was not sufficient to raise whole-body temperature. In addition, hepatic UCP1 expression changed food intake. Whereas without hepatic UCP1 expression, food intake amounts in high-fat–fed mice were markedly increased compared with those in standard diet–fed lean mice (compare Figs. 1F and 5D), hepatic UCP1 expression reversed hyperphagia in mice with high-fat diet–induced obesity and diabetes (Fig. 1F). After day 8, concomitantly with the drop in hepatic UCP1 expression, hyperphagia was restored (Fig. 1F). In contrast, mice received an intravenous injection of adenovirus encoding CPT1a, another mitochondrial protein, did not show significantly altered food consumption (data not shown), suggesting that food intake suppression induced by hepatic UCP1 expression is not a nonspecific effect of expression of any of the hepatic mitochondrial proteins.

To eliminate any secondary effects of reduced food intake induced by hepatic UCP1 expression, we performed pair-feeding experiments. In contrast to UCP1 mice, pair-fed LacZ mice exhibited only slight decreases in body weights and blood glucose levels (of 3.1 and 6.9%, respectively, on day 7 after adenoviral administration). These results suggest that increased energy expenditure is an important mechanism underlying marked improvements of obesity and diabetes in UCP1 mice.

Hepatic and adipose fat accumulations were examined on day 7 after adenoviral gene delivery. In the high-fat–fed control mice, liver weight and triglyceride content were markedly increased compared with the standard chow–fed lean mice (compare Fig. 2A and B with Fig. 5E and F, respectively). Hepatic UCP1 expression significantly decreased liver weight (Fig. 2A) and triglyceride content (Fig. 2B) compared with LacZ mice, with high-fat feeding. It is interesting that hepatic UCP1 expression also decreased fat content in their adipose tissues. For example, epididymal fat weight was significantly decreased in UCP1 mice compared with that in controls (Fig. 2C). Thus, hepatic expression of UCP1 exerts not only local effects in the liver but also remote effects on metabolism in other tissues.

These results were confirmed by the histological findings. No apparent infiltration or structural change was observed in the livers of either LacZ mice or UCP1 mice, indicating the absence of adenovirus-induced liver damage (Fig. 2D). Whereas abundant lipid droplets were present in the livers of control mice, these lipid droplets were markedly diminished in UCP1 mouse livers, indicating marked improvement of fatty liver findings in response to UCP1 expression (Fig. 2D). Furthermore, the cell diameters in epididymal fat (Fig. 2E) and brown adipose (Fig. 2F) tissues were significantly decreased in UCP1 mice. Expression levels of endogenous UCP1 protein in brown adipocytes were similar in the two groups (Fig. 2G), suggesting that energy expenditure in brown adipocytes was not increased in UCP1 mice. These findings suggest that hepatic UCP1 expression promotes hydrolysis of triglycerides already stored in adipose tissues, leading to smaller adipocytes with the resultant fatty acids being mobilized and metabolized as a substrate for oxidation in the liver.

To elucidate the underlying mechanism whereby stored fat was decreased in the liver by hepatic UCP1 expression, we examined the expressions of proteins involved in lipid metabolism by quantitative RT-PCR. Significant reductions in the expressions of the lipogenic enzymes, including stearoyl-CoA desaturase-1 and fatty acid synthase, were observed in UCP1 mice (Fig. 3A). Sterol regulatory element binding protein 1c (SREBP1c) expression in the liver tended to be diminished. In contrast, hepatic expressions of enzymes involved in fatty acid oxidation tended to be increased. In particular, expressions of fatty acid transporter and UCP2 were significantly increased (Fig. 3B).

We further examined expression levels of key enzymes for hepatic glucose production. Hepatic phosphoenolpyruvate carboxykinase and glucose-6-phosphatase expressions were significantly decreased in UCP1 mice (Fig. 3C), suggesting a decrease to contribute to improvement of diabetes.

UCP1 expression may activate AMPK as a result of decreased generation of ATP. AMPK activation reportedly decreases malonyl-CoA generation via inhibition of ACC (21), resulting in enhancement of fatty acid oxidation. Therefore, ATP levels and AMPK phosphorylation in the liver were examined in LacZ and UCP1 mice under ad libitum feeding conditions. Hepatic ATP concentrations in UCP1 mice were approximately half those in control mice (Fig. 3D) but still ∼2.3-fold those in standard diet–fed control mice. Hepatic AMPK activity was increased 1.6-fold in UCP1 mice compared with LacZ mice (Fig. 3E). The phosphorylation state of the α-subunit of AMPK in the liver was enhanced in UCP1 mice (Fig. 3F). Furthermore, resultant enhancement of hepatic ACC phosphorylation was observed (Fig. 3G). These findings suggest that AMPK activation induced by UCP1 expression plays an important role in the observed marked improvement of fatty liver findings via enhanced fatty acid oxidation.

The results of oral glucose tolerance (Fig. 4A) and insulin tolerance (Fig. 4B) tests on day 7 after adenoviral administration clearly showed that hepatic expression of UCP1 markedly improved glucose tolerance and insulin sensitivity in obese and diabetic mice. Improved insulin sensitivity in muscle was confirmed by enhanced insulin receptor and IRS1 phosphorylation (Fig. 4C) in response to insulin administration. Thus, hepatic UCP1 expression exerts a remote beneficial effect on insulin sensitivity in muscle.

In addition, plasma lipid parameters were decreased in UCP1 mice. Total plasma cholesterol levels tended to be decreased in UCP1 mice compared with controls, although the changes were not statistically significant (Fig. 4D). Plasma triglyceride and free fatty acid levels were significantly decreased in UCP1 mice (Fig. 4D). Thus, hepatic UCP1 expression also improved diet-induced dyslipidemia.

Serum insulin levels were markedly decreased, by 57% (Fig. 4E), in UCP1 mice, despite lower blood glucose levels (Fig. 1C), indicating marked improvement of systemic insulin sensitivity. Serum adiponectin and TNF-α levels were similar in these groups (Fig. 4F), suggesting that these adipocytokines are not involved in the improvement of insulin resistance in UCP1 mice. In contrast, serum leptin levels were significantly decreased, by 56%, in UCP1 mice compared with those in control mice (Fig. 4F) concomitantly with decreased food intake (Fig. 1F). In control mice that were fed a high-fat diet, marked hyperleptinemia was observed (serum leptin concentrations, standard diet–fed mice versus high-fat diet–fed mice: 0.48 ± 0.08 vs. 32 0.8 ± 4.6 ng/ml) despite increased food intake (compare Fig. 1F with Fig. 5D), indicating leptin resistance. The present results suggest that hepatic UCP1 expression improves hypothalamic leptin resistance in obese and diabetic mice. To directly test whether leptin sensitivity was improved, we performed leptin tolerance tests (Fig. 4G). Leptin was injected intraperitoneally into fasted mice, followed by measurement of 12-h food intakes. The food intake inhibition by leptin administration was far more profound in UCP1 mice than in LacZ mice. Thus, UCP1 mice responded strongly to leptin administration, clearly showing that hepatic UCP1 expression exerts a therapeutic effect on hypothalamic leptin resistance.

Hepatic UCP1 expression reduced body weight and blood glucose and lipid levels in obese and diabetic mice. These are very promising results suggesting that ectopic UCP1 expression may be useful in treating diabetic individuals who are obese. However, if this were also the case in lean individuals, then these individuals would become leaner, possibly even developing malnutrition and hypoglycemia. We therefore performed experiments with a similar design but used 9-week-old standard diet–fed lean mice, i.e., the same age as the high-fat–fed mice.

It is intriguing that although ectopic UCP1 expression levels in the liver were similar under high-fat and standard diet conditions (Fig. 5A), the resultant phenotypes were completely different. In standard diet–fed lean mice, hepatic UCP1 expression did not alter body weight (Fig. 5B), fasting blood glucose levels (Fig. 5C), or food intake amounts (Fig. 5D). In addition, hepatic weight (Fig. 5E), triglyceride content (Fig. 5F), and epididymal fat weight (Fig. 5G) were not changed. Thus, hepatic UCP1 expression did not exert significant effects on glucose metabolism or adiposity in lean mice.

To determine why hepatic UCP1 expression in lean mice did not significantly alter metabolic conditions, we measured basal energy expenditure and hepatic ATP contents. Hepatic UCP1 expression did not significantly change basal energy expenditure (Fig. 5H) or hepatic ATP levels (Fig. 5I), suggesting that UCP1 ectopically expressed in the liver is minimally involved in mitochondrial uncoupling, when surplus energy is not stored in the liver. Thus, hepatic UCP1 is likely to dissipate excess energy while having no effect on required energy. These characteristics are favorable in terms of therapeutic strategies for the metabolic syndrome.

In this study, after mice had developed obesity-associated diabetes, ectopically expressing UCP1 in the liver resulted in marked improvements in both disease conditions. UCP1 expression would be expected to decrease ATP generation in the liver and thus to activate hepatic AMPK. Indeed, ATP contents were decreased, and AMPK and ACC phosphorylations were increased. AMPK reportedly phosphorylates and inactivates ACC, resulting in a decrease in malonyl-CoA generation (21). Because malonyl-CoA is a negative regulator via suppression of CPT1, a rate-limiting enzyme for fatty acid oxidation (22), a decrease in malonyl-CoA generation is likely to enhance fatty acid oxidation to meet respiratory demands. Furthermore, hepatic expressions of lipogenic enzymes were decreased by UCP1 expression in the liver, which may be explained by AMPK activation and possible SREBP1 reduction in the liver; metformin reportedly activates AMPK and inhibits hepatic SREBP1 expression (23). Taken together, the results suggest that fatty acid synthesis was suppressed with concomitant enhancement of fatty acid oxidation, resulting in the marked decrease in hepatic triglyceride contents.

How might a change in hepatic lipid metabolism affect the energy balance of the entire body? It is noteworthy that the weight and/or cell sizes of epididymal fat and brown adipose tissues were markedly decreased by hepatic UCP1 expression in the present study. Inhibition of fat accumulation in adipose tissues was also observed in UCP1 and in UCP3 transgenic mice under the control of muscle-specific promoters (7,8). Mice lacking ACC2, which is predominantly expressed in the heart and muscle of wild-type mice, also markedly inhibited fat accumulation in their adipose tissues (24). In reports using transgenic models, muscle is a site of increasing energy expenditure, through mitochondrial uncoupling, which prevents obesity. In the present study, hepatic expression of UCP1 reduced fat contents, rather than inhibiting fat accumulation, not only in the liver but also in adipose tissues, indicating promotion of hydrolysis of triglycerides already stored in the adipose tissues. Thus, hepatic uncoupling is likely to convey signals to peripheral adipose tissues. These signals might involve an autonomic nerve network, because the hydrolysis of triglycerides stored in adipose tissues is controlled mainly by the cAMP-mediated pathway, including sympathetic nerve activation (25). Alternatively, a decline in serum fatty acid concentrations, observed in UCP1 mice, or some unknown factors secreted by the liver might trigger lipolysis in adipose tissues. Although more work is required to elucidate the mechanism underlying this remote effect, enhancement of hepatic uncoupling is likely to exert therapeutic, rather than preventive, effects on insulin resistance associated with obesity. Thus, the liver is a potential therapeutic target for diabetes with obesity. Furthermore, unraveling the underlying mechanism may lead to development of antiobesity pharmacological agents that promote lipolysis in adipose tissues.

The present results are also interesting with respect to appetite regulation. Transgenic mice overexpressing UCP3 in skeletal muscle are reportedly hyperphagic (8), whereas UCP1 transgenic mice show no changes in food intake (7). In these transgenic mice, UCPs are continuously overexpressed throughout life, including in the fetal stage. In contrast, the UCP was expressed after development of diabetes with obesity in the present study. In obese subjects, serum leptin levels are reportedly increased with an increment in adipose tissue mass (26,27). Despite increased serum leptin levels, neither appetite nor food intake was suppressed, but instead increased, which is explained by hypothalamic leptin resistance in obese subjects. Herein, control mice on a high-fat diet were hyperphagic compared with those on a standard diet, whereas serum leptin levels were markedly elevated in high-fat diet–fed mice, indicating the development of leptin resistance. It is interesting that hepatic UCP1 expression reversed hyperphagia in high-fat diet–fed mice. Leptin tolerance tests show marked improvement of hypothalamic leptin resistance in UCP1 mice, another remote effect of hepatic UCP1 expression. In addition, increased fatty acid oxidation might be involved in the decreased food intake, because administration of peroxisome proliferator–activated receptor (PPAR)-α agonists reportedly reduces food intake amounts, but not in mice deficient in PPAR-α (28). Furthermore, streptozotocin-induced hyperphagia was reportedly reversed by hepatic expression of protein phosphatase-1 (29), suggesting that altering hepatic metabolism modulates appetite. Vagal pathways from the liver to the brain mediate the fat-induced changes in hypothalamic neuropeptides and feeding behavior in diabetic rats (30). Taken together with these observations, through appetite modulation, the liver also holds promise as a target for treatment of diabetes with obesity.

The most intriguing finding of the present study is that, despite similar UCP1 expression levels in mice on high-fat and standard diets, the resultant phenotypes were completely different. Hepatic UCP1 expression exerted no significant effects on food intake, weight change, or blood glucose levels in standard diet–fed lean mice. No alterations in energy expenditure or hepatic ATP contents were observed with hepatic UCP1 expression, indicating that, in the absence of a significant energy surplus, ectopic UCP1 has minimal effects on mitochondrial uncoupling. We performed similar experiments in a mildly obese and insulin-resistant model, 15% fat-fed mice. In these mice, hepatic UCP1 expression did not change body weight or food intake. Glucose tolerance and insulin sensitivity were significantly improved, but the effects were smaller (data not shown) than those in a more severely obese and insulin-resistant model, 32% fat-fed mice, reported here. Furthermore, under 32% high-fat–fed conditions in the present study, although hepatic UCP1 expression decreased ATP levels in the liver, the reduced ATP concentrations still exceeded those in standard diet–fed mice, suggesting that enhanced expression of UCPs in the liver does not itself produce an energy shortage. Taken together, hepatic UCP1 is likely to sense the metabolic state in the liver and function according to the degree of stored energy in the liver. In the reconstituted system, addition of fatty acids is indispensable for proton transport by UCP1 (31,32). Although the underlying mechanism has been widely debated (33,34), fatty acid cycling seems to be important for proton transport by UCP1 (35,36). Via such a mechanism, ectopic UCP1 activity in the liver may depend on the metabolic state, probably on the amount of stored fat in the liver. Thus, hepatic UCP1 seems to dissipate surplus energy but not to affect required energy. Therefore, the liver, in which intracellularly stored fat changes dramatically according to the energy balance, seems to be a good target tissue for enhanced expression of UCPs. This feature is of particular importance, as applied to therapeutic strategies for type 2 diabetes associated with obesity and insulin resistance.

Recently, it was reported that, using a transgenic technique, skeletal muscle expression of UCP1 in genetically obese mice lowers blood pressure (37), suggesting that uncoupling decreases the risk for atherosclerosis in patients with obesity and type 2 diabetes. In addition, uncoupling reportedly decreases the production of reactive oxygen species (38), although total oxygen consumption increases. A high mitochondrial electrochemical gradient is associated with the production of reactive oxygen species that may damage tissues, a possible cause of diabetes complications and atherosclerosis (39). Thus, the respiratory uncoupling increment in the liver may protect tissues from oxidative stress. Taken together with the results of the present study, enhancement of UCPs in the liver is a potential therapy for the metabolic syndrome via reductions in adiposity and blood glucose levels as well as possibly reactive oxygen species in obese and diabetic individuals.

This work was supported by a Grant-in-Aid for Scientific Research (B2, 15390282); a Grant-in-Aid for Exploratory Research (15659214); and Tohoku University 21st Century COE Program “Comprehensive Research and Education Center for Planning of Drug Development and Clinical Evaluation” to H.K. and a Grant-in-Aid for Scientific Research (13204062); and Tohoku University 21st Century COE Program “the Center for Innovative Therapeutic Development for Common Diseases” to Y.O. from the Ministry of Education, Science, Sports and Culture of Japan.

We thank Prof. Y. Moriyama (Okayama University) for helpful suggestions for measuring ATP. We also thank I. Sato and K. Kawamura for technical support.