β-Aminoisobutyric acid (BAIBA), a thymine catabolite, increases fatty acid oxidation (FAO) in liver and reduces the gain of body fat mass in Swiss (lean) mice fed a standard chow. We determined whether BAIBA could prevent obesity and related metabolic disorders in different murine models. To this end, BAIBA (100 or 500 mg/kg/day) was administered for 4 months in mice totally deficient in leptin (ob/ob). BAIBA (100 mg/kg/day) was also given for 4 months in wild-type (+/+) mice and mice partially deficient in leptin (ob/+) fed a high-calorie (HC) diet. BAIBA did not limit obesity and hepatic steatosis in ob/ob mice, but reduced liver cytolysis and inflammation. In ob/+ mice fed the HC diet, BAIBA fully prevented, or limited, the gain of body fat, steatosis and necroinflammation, glucose intolerance, and hypertriglyceridemia. Plasma β-hydroxybutyrate was increased, whereas expression of carnitine palmitoyltransferase-1 was augmented in liver and white adipose tissue. Acetyl-CoA carboxylase was more phosphorylated, and de novo lipogenesis was less induced in liver. These favorable effects of BAIBA in ob/+ mice were associated with a restoration of plasma leptin levels. The reduction of body adiposity afforded by BAIBA was less marked in +/+ mice. Finally, BAIBA significantly stimulated the secretion of leptin in isolated ob/+ adipose cells, but not in +/+ cells. Thus, BAIBA could limit triglyceride accretion in tissues through a leptin-dependent stimulation of FAO. As partial leptin deficiency is not uncommon in the general population, supplementation with BAIBA may help to prevent diet-induced obesity and related metabolic disorders in low leptin secretors.
Increasing prevalence of obesity is a worldwide threat because it enhances the risk of various metabolic disorders and diseases such as insulin resistance, type 2 diabetes, hyperlipemia, coronary heart disease, steatosis and nonalcoholic steatohepatitis, and some cancers (1,2). Some antiobesity drugs are already marketed, but their efficacy is sometimes limited and their utilization can induce frequent side effects (3). Thus, there is an urgent need to develop effective and safe drugs to treat or prevent obesity.
Increasing fatty acid oxidation (FAO) appears to be a potentially interesting option for the management of obesity and some related metabolic disorders such as insulin resistance and nonalcoholic steatohepatitis (4,5). Recently, we found that the natural catabolite of thymine β-aminoisobutyric acid (BAIBA) increased mitochondrial FAO and reduced fat accretion in mice. Indeed, the administration of 100 mg/kg/day of BAIBA for 2 or 6 weeks stimulated FAO in liver mitochondria and increased plasma ketone bodies in Swiss (lean) mice fed a standard diet (6,7). Moreover, increased hepatic FAO was associated with a significant reduction of body fat mass and a trend toward lower hepatic triglycerides (7). However, when the same dose of BAIBA was administered to ob/ob mice, body fat mass and hepatic lipids were unchanged (7).
Ob/ob mice present total leptin deficiency due to missense leptin (ob) gene mutation (8). This complete lack of leptin induces a wide array of metabolic disorders such as morbid obesity, insulin resistance associated with type 2 diabetes, and hypercholesterolemia, which is however not associated with hypertriglyceridemia (7,8,9). Ob/ob mice also present liver damage reflected by a large increase (approximately five- to tenfold) in plasma alanine aminotransferase (ALT) activity (10,11). Indeed, ob/ob liver is characterized by massive steatosis associated with moderate necroinflammation and some apoptosis (10,11). In contrast, fibrosis is limited in ob/ob mice because leptin has a permissive effect on hepatic fibrogenesis (2,12).
The present study was carried out to know more about the metabolic effects of BAIBA, and, in particular, to fathom why it does not decrease body fat mass and liver triglycerides in ob/ob mice when administered at the dose of 100 mg/kg/day for 6 weeks (7). A first hypothesis could be that the dose and/or the length of administration of BAIBA were not sufficient. To test this possibility, ob/ob mice were treated for 4 months at the doses of 100 or 500 mg/kg/day. A second hypothesis could be that BAIBA could require leptin to afford its beneficial effect on fat accumulation. To test this hypothesis, BAIBA was administered at the dose of 100 mg/kg/day in wild-type (+/+) mice and C57BL-ob/+ mice (i.e., harboring only one allele of the ob gene) fed for 4 months with a high-fat, high-sucrose diet. Our data suggest that BAIBA could increase FAO and thus prevent fat accumulation through a leptin-dependent mechanism.
Methods and Procedures
Animals, diets, and treatment
A first series of experiments was performed on 5-week-old male C57BL/6J-ob/ob mice (henceforth referred to as ob/ob mice) purchased from Janvier (Le-Genest-St-Isle, France) and fed ad libitum a normal chow (A04 biscuits; UAR, Villemoisson-sur-Orge, France). This standard-calorie (SC) diet brings 2,900 kcal/kg of food and contains 3% fat, 60% complex carbohydrate (primarily starch), and 16% protein. After a week of acclimatization in the animal house, ob/ob mice were treated with 100 or 500 mg/kg/day of BAIBA (Sigma-Aldrich, Saint-Quentin Fallavier, France) for 4 months. BAIBA was administered in the drinking water, as previously described (6,7). In ob/ob mice, but not in ob/+ mice, we rapidly noticed that BAIBA slightly but significantly lowered the daily consumption of drinking water by 15%. Consequently, the concentration of BAIBA was adjusted in the drinking solution so that the daily intake of BAIBA was kept constant.
A second series of experiments was performed on 5-week-old male C57BL/6J-+/+ mice (wild type, referred to as +/+ mice) and C57BL/6 -ob/+ mice (partially deficient in leptin, henceforth referred to as ob/+ mice) purchased from Janvier. After a week of acclimatization, both +/+ and ob/+ mice were fed ad libitum a high-fat, high-sucrose diet (referred to as high-calorie (HC) diet) purchased from SAFE (Augy, France). The HC diet brings 5,500 kcal/kg and includes 35% fat (primarily lard), 36% simple carbohydrate (mainly saccharose), and 19% protein. Ob/+ mice were treated or not treated with 100 mg/kg/day of BAIBA. Thus, three different groups of mice were studied for 4 months, namely +/+ fed the HC diet and ob/+ mice fed the HC diet, treated or not treated with BAIBA. To measure food consumption, the HC diet (which presents a pastry consistency) was manually transformed into pellets that were frozen until their utilization. Food consumption was then assessed every day during 6 consecutive weeks.
Our last experiment was performed on a few +/+ mice fed the SC or the HC diet for 4 months. Hence, three different groups of mice were studied (n = 6–8 per group), namely, +/+ mice fed the SC diet and +/+ mice fed the HC diet, treated or not treated with 100 mg/kg/day of BAIBA. All experiments were performed in agreement with national guidelines for the proper use of animals in biomedical research.
Blood and tissue sampling
Unless otherwise indicated, blood was drawn from the retroorbital sinus with heparinized capillary Pasteur pipettes. Blood was collected either in the postabsorptive state (henceforth referred to as the fed state) or after an overnight period of fast (henceforth referred to as the fasted state). In some experiments, blood was collected in +/+ and ob/+ mice before the onset of the investigations. All mice were sacrificed by cervical dislocation. Liver, epididymal white adipose tissue (eWAT) and interscapular brown adipose tissues (iBAT) were quickly removed and frozen in liquid nitrogen and kept at −80 °C until use. In some experiments, liver fragments were processed for histology and in situ detection of apoptosis.
Liver histology and in situ detection of apoptosis
To evaluate necroinflammation and fibrosis, liver fragments from fed animals were fixed in 10% neutral formalin and embedded in paraffin. Then, 5-μm sections were cut and then stained with specific dyes. Examination of the sections was performed by an experienced pathologist (A.A.-T.) without knowledge of the treatment. Necroinflammation was estimated after hematoxylin-eosin staining on 10 different fields at ×200 magnification, and semiquantified as 0 (no necroinflammation), 1 (mild necroinflammation), and 2 (moderate necroinflammation), depending of the number and the size of the inflammatory infiltrates (Figure 1). Portal and perisinusoidal fibrosis was evaluated respectively thanks to Masson's trichrome and picro-sirius red staining, and was scored 0 (no fibrosis), 1 (mild fibrosis), or 2 (moderate fibrosis), depending on the extent of collagen deposition around the portal triad and within the liver lobule. For the detection of neutral lipids, liver cryosections were stained with oil red O. Steatosis, evaluated as the percentage of hepatocytes containing vacuoles of fat, was assessed on 10 different fields at ×200 magnification (Figure 1). In situ detection of apoptosis was performed with the terminal deoxynucleotidyl transferase nick-end labeling assay using the TACS TdT kit (R&D Systems Europe, Lille, France), as previously described (11). The number of apoptotic nuclei was then counted on 20 different fields at ×400 magnification.
Plasma triglycerides, glucose, ALT, lactate dehydrogenase (LDH), total cholesterol, iron, ferritin, β-hydroxybutyrate, nonesterified fatty acids and total antioxidant status (expressed as Trolox equivalents) were measured on an automatic analyzer (Olympus AU400). Triglycerides, glucose, ALT, LDH, total cholesterol, iron and ferritin were measured with commercial kits (OSR6133, OSR6121, OSR6107, OSR6126, OSR6116, OSR6186, OSR61138, respectively) from Olympus Diagnostic (Rungis, France), whereas β-hydroxybutyrate, nonesterified fatty acids and total antioxidant status were measured with commercial kits (RB1007, FA115, and NX2332, respectively) from RANDOX Diagnostic (Montpellier, France). Insulin, leptin and adiponectin were measured using double antibody RIA kits (RI-13K, ML-82K, MADP-60HK, respectively) purchased from Linco Research (St. Charles, MO).
Assessment of leptin in eWAT and stomach
To measure leptin in adipose tissue, an eWAT fragment was homogenized in Krebs-Ringer buffer (100 mg/ml) with a protease inhibitor cocktail (1 μl/ml) (Sigma-Aldrich), as previously described (13). After centrifugation (10 min at 10,000 g), the infranatant was used to measure leptin with a mouse leptin ELISA kit (Crystal Chem, Downers Grove, IL). Assessment of leptin in the stomach was performed after 6 weeks of treatment. The stomach from fed animals was removed, and leptin was extracted from fundic epithelium scrapings, as previously described (14).
Assessment of body fat mass and lean mass
Fat mass and lean mass (the latter representing water and proteins) were determined by dual-energy X-ray absorptiometry using a Piximus apparatus (Lunar, Madison, WI), as previously described (7,15). Fed mice were first studied on the day before the onset of the investigations, and then at the end of the 4-month experiments. This allowed us to determine changes from baseline in each animal.
Lipids and triglycerides, assessment of fatty acid synthesis, and measurement of microsomal triglyceride transfer protein activity in liver
Hepatic total lipids and triglycerides were measured in fed mice as previously described (11). De novo fatty acid synthesis in liver was assessed in fed mice by using the method previously described by Stansbie et al. (16). Briefly, 150 μCi of 3H-H2O was injected intraperitoneally to mice in the postabsorptive state. Two hours later, blood was drawn to determine the 3H-H2O specific activity, and liver was quickly removed to extract fatty acids (16). After counting the radioactivity, the rate of de novo fatty acid synthesis was calculated as micromoles of 3H-incorporated into fatty acids per hour per gram of liver. Activity of microsomal triglyceride transfer protein (MTP) in liver was determined with a commercial kit (Roar Biomedical, New York, NY) based on the MTP-mediated transfer of a self-quenched fluorescent neutral lipid from the core of a donor particle to an acceptor particle, as previously described (17).
Glucose and insulin tolerance tests
Intraperitoneal glucose tolerance test (IPGTT) was performed in mice after a 12-h overnight fast. At 10:00 am, D-glucose (2 g/kg body weight) was injected intraperitoneally to mice and blood was collected by tail bleeding at 0, 15, 30, 45, 60, 90, and 120 min for measurement of blood glucose by using a one-touch Accu-Check Glucometer (Roche, Paris, France). In one IPGTT experiment, small volume of blood was also drawn from the retroorbital sinus at 0, 30, 60, 90, and 120 min for subsequent measurement of plasma insulin by using a rat insulin ELISA kit from Crystal Chem. Intraperitoneal insulin tolerance test (IPITT) was performed in mice after a 5-h fast. Human insulin (Actrapid) purchased from Novo Nordisk (Chartres, France) was injected intraperitoneally (0.5 U/kg body weight), and blood was collected by tail bleeding with the timing detailed above for IPGTT. Areas under the curve during IPGTT and IPITT were then calculated by the linear trapezoidal method.
RNA isolation and real-time quantitative PCR analysis
Total hepatic RNA was isolated by acid guanidinium thiocyanate-phenol-chloroform extraction. Total RNA from iBAT and eWAT was extracted using the Lipid RNeasy Kit (Qiagen, Courtaboeuf, France). RNA integrity was assessed with the RNA 6000 Nano LabChip Kit (Agilent, Waldbronn, Germany). Real-time quantitative PCR was subsequently performed on selected genes expressed in liver, iBAT, and eWAT (Table 1). To this end, reverse transcription was performed with 2 μg of total RNA in a reaction buffer composed of 20 mmol/l Tris-HCl (pH 8.3), 375 mmol/l KCl, 15 mmol/l MgCl2, 10 mmol/l dithiothreitol, 0.5 mmol/l of each deoxynucleoside triphosphate, 250 ng of random primers, 2 U of RNase inhibitor, and 10 U of Moloney murine leukemia virus reverse transcriptase. The reaction was carried out at 37 °C for 50 min and the mixture was then heated at 70 °C for 15 min. Real-time quantitative PCR was subsequently performed on an aliquot (5 μl) of the reverse transcription reaction with 0.25 μmol/l of each primer (Table 1) and 10 μl of Master SYBR Green mix (Sigma-Aldrich), in a Chromo IV LightCycler apparatus (Biorad, Marnes-La-Coquette, France). The PCR conditions were 1 cycle at 94 °C for 3 min, followed by 40 cycles at 94 °C for 30s and 60 °C for 1 min. Amplification of specific transcripts was confirmed by melting curve profiles generated at the end of each run. PCR specificity was further ascertained with an agarose gel electrophoresis by checking the length of the PCR products. Expression of the mouse ribosomal protein S6 (S6) was used as reference, and the 2−ΔΔCt method was employed to express the relative expression of each selected gene.
Table 1. Sequences of primers used for real-time quantitative PCR
Extraction of liver proteins and western blot analysis
Frozen liver fragments (∼100 mg) were homogenized in a phosphate-buffered saline solution containing 0.1% triton and protease inhibitors. Homogenates were then centrifuged at 4,500 g at 4 °C to remove tissue debris. Protein content was measured in the supernatants by using the Lowry assay. In order to assess the hepatic expression of fatty acid synthase (FAS), total and phosphorylated acetyl-CoA carboxylase (ACC and pACC, respectively), manganese superoxide dismutase (MnSOD), and cytochrome P450 2E1 (CYP2E1) proteins (∼50 μg) underwent sodium dodecyl sulfate-polyacrylamide electrophoresis (8% polyacrylamide for FAS, pACC and ACC, 12% for MnSOD and CYP2E1), transfer to nitocellulose membrane (Hybond ECL; Amersham Biosciences, Orsay, France), and immunoblotting with rabbit polyclonal antibodies against FAS (Santa Cruz Biotechnology, Santa Cruz, CA), ACC and pACC (Upstate, Lake Placid, NY), MnSOD (Stressgen, Ann Arbor, MI), and CYP2E1 (Oxford Biomedical Research, Oxford, MI). Blots were incubated with appropriate secondary antibodies, and protein bands were revealed by enhanced chemiluminescence (Amersham Pharmacia Biotech, Orsay, France). To normalize protein loadings, blots were stripped and incubated with monoclonal mouse antibodies against β-actin (Sigma-Aldrich). Protein bands were quantified using a Helwett Packard Scanjet 4570c scanning unit and ImageMaster1D software (Amersham Pharmacia Biotech, Orsay, France).
Assessment of oxidative stress in liver
Reduced glutathione (GSH) levels were determined by a method adapted from Griffith, as previously described (11). In order to assess hepatic aconitase activity, frozen liver fragments (∼20 mg) were homogenized in 500 μl buffer containing 50 mmol/l Tris-HCl pH 7.4, 0.2 mmol/l sodium citrate, and 0.05 mmol/l MgCl2. Homogenates were then centrifuged at 800g at 4 °C for 10 min, and supernatants were then sonicated for 20 s. Aconitase activity was subsequently assessed on 200 μg proteins in the presence of 1 mmol/l sodium citrate, 1 mmol/l nicotinamide-adenine dinucleotide phosphate+, and 2 U of isocitrate dehydrogenase. Samples were preincubated at 37 °C for 5min, and aconitase activity (expressed as nmol of generated nicotinamide-adenine dinucleotide phosphate-oxidase/min/mg protein) was then assessed from the increased absorption measured at 340 nm for 5 min.
Primary culture of rat hepatocytes and assessment of FAO
Male Sprague-Dawley rats were anesthetized with pentobarbital, and hepatocytes were isolated by a two-step in situ perfusion, as previously described (18). Cell viability, estimated by Trypan blue exclusion, averaged 85–90%. Hepatocytes were cultured at 37 °C under a 5% CO2/ 95% air atmosphere in a Williams'E culture medium supplemented with 10% fetal calf serum, streptomycin (0.1 mg/ml), and penicillin (100 U/ml). After cell attachment (3 h), and for each additional day of culture, the medium was replaced by a new, serum-free Williams'E medium containing hydrocortisone (50 nmol/l) with or without BAIBA. Cultures were ended after 72 h of treatment and used for subsequent assessment of FAO. To this end, rat hepatocytes (3 × 106 cells) were seeded in 60-mm Petri dishes. The Williams'E culture medium was removed and replaced by a fresh medium containing 0.5 mmol/l L-carnitine and 10% fat-free bovine serum albumin (Sigma-Aldrich). [U-14C]palmitic acid (final concentration, 1 mmol/l; 0.05 μCi/ml) was added, and the reaction was carried out for 90 min at 37 °C. After the addition of perchloric acid (final concentration, 8%) and centrifugation at 4,000 g for 10 min, an aliquot of the supernatant was counted for 14C-labeled acid-soluble β-oxidation products, as previously described (19).
Primary mouse adipose tissue culture and leptin production
Isolation of mouse adipocytes from C57BL/6J-+/+ and C57BL/6-ob/+ mice was performed as previously described (20,21), with slight modifications. Briefly, the epididymal fat pads from 8 mice of each genotype were removed, pooled together, and disaggregated with collagenase type II (Sigma-Aldrich), in a pH 7.4 buffer containing 11 mmol/l D-glucose, 20 mmol/l HEPES, 119 mmol/l NaCl, 1.3 mmol/l MgCl2, 1 mmol/l NaH2PO4, 2.5 mmol/l KCl, and 2.5 mmol/l CaCl2. After incubation under continuous shaking for 1 h at 37 °C, the adipocytes were washed 3 times with the same buffer and then centrifuged at 1,500g for 5 min. The supernatant containing the adipocytes was then taken and diluted (1:1; vol:vol) with high-glucose (4.5 g/l) DMEM supplemented with 10% fetal calf serum. Adipocytes were then transferred into Corning CellBIND Surface 24-well plates (ATGC Biotechnologie, Marne la Vallée, France), and incubated at 37 °C under a 5% CO2/ 95% air atmosphere. After 48 h, culture medium was replaced with low-glucose (1 g/l) DMEM containing or not 3 mmol/l of BAIBA or propionic acid. This concentration was chosen, based on a recent study showing that 3 mmol/l of propionic acid increased leptin production in primary mouse adipose tissue culture (22). Finally, the culture medium was taken after 48 h for the determination of leptin levels, thanks to a mouse leptin ELISA kit (R&D Systems Europe, Lille, France). Total proteins were also measured in each well, so that the levels of secreted leptin were expressed as ng leptin per ml and per mg of protein.
Data are presented as means ± s.e.m. When several groups were compared, statistical analysis was performed by one-way ANOVA followed by a Fisher's test. In experiments with only two sets of data, the nonparametric Mann-Whitney test was used.
Effects of BAIBA in ob/ob mice
Ob/ob mice were treated for 4 months with 100 or 500 mg/kg/day of BAIBA. However, both doses of BAIBA did not reduce body adiposity and liver triglycerides (data not shown), thus extending our data published previously for the 100 mg/kg/day dosage (7). Moreover, BAIBA did not improve hepatic steatosis (Table 2). In contrast, we noticed lower necroinflammation in ob/ob mice treated with BAIBA for both doses, along with a strong reduction of the number of apoptotic nuclei in obese mice treated with 500 mg/kg/day (Table 2). Portal and perisinusoidal fibrosis was also alleviated with BAIBA (Table 2, Figure 2). Plasma levels of different biological parameters were virtually unchanged, although a significant decrease in plasma LDH was observed with the highest dose of BAIBA (Table 3). Interestingly, both doses of BAIBA increased liver GSH, though it was significant only with the 100 mg/kg/day dosage. Indeed, liver GSH was 69 ± 10, 92 ± 5 and 85 ± 4 nmol/mg protein (mean ± s.e.m. for 6–8 mice) in ob/ob controls and in ob/ob mice treated with 100 and 500 mg/kg/day of BAIBA. Finally, we performed glucose tolerance and insulin sensitivity tests in ob/ob mice treated with 100 mg/kg/day of BAIBA, but the profile of blood glucose during these tests was not improved with BAIBA (data not shown).
Table 2. Steatosis, necroinflammation, and fibrosis in obese ob/ob mice treated or not treated with 100 or 500 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Table 3. Plasma parameters in obese ob/ob mice treated or not treated with 100 or 500 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effect of BAIBA in ob/+ mice on body adiposity, leptin, and food consumption
When mice were fed with the HC diet, body fat mass was significantly increased in ob/+ mice when compared to +/+ mice, thus indicating that partial leptin deficiency greatly favored diet-induced obesity (Figure 3). BAIBA supplementation (100 mg/kg/day) for 4 months almost completely prevented in ob/+ mice the gain of body fat mass and partially protected against the loss of body lean mass (Figure 3). Plasma levels of leptin and its content in eWAT were also measured. As expected, ob/+ mice presented lower leptin levels in plasma and eWAT when compared to +/+ mice (Figure 4). The difference of plasma leptin levels between ob/+ and +/+ mice was even more marked when values were expressed per gram of body fat mass as determined by dual-energy X-ray absorptiometry (Figure 4). Importantly, BAIBA significantly augmented by 60% these relative levels of plasma leptin (Figure 4). Moreover, leptin content in eWAT was no longer significantly decreased (P = 0.10) in ob/+ mice treated with BAIBA (Figure 4).
Leptin content was also assessed in the stomach after 6 weeks of treatment (Figure 4). Interestingly, our results showed that leptin content in the fundic epithelium almost paralleled the levels found in plasma and eWAT. Indeed, BAIBA partially prevented the reduction of gastric leptin observed in ob/+ mice (Figure 4).
Food consumption was slightly, but significantly, reduced in untreated ob/+ mice compared to +/+ mice, but this effect was no longer observed in ob/+ mice treated with BAIBA. Indeed, the daily food intake was 2.43 ± 0.05, 2.25 ± 0.06, and 2.50 ± 0.05 g per animal, respectively, in +/+ mice, untreated ob/+ mice, and ob/+ mice treated with BAIBA. Thus, the beneficial effect of BAIBA on body adiposity was not due to a reduction of food consumption.
Effect of BAIBA on glucose tolerance and insulin sensitivity
To assess glucose tolerance, IPGTT was performed before the onset of the treatment and thus after 4 months. The initial IPGTT indicated that ob/+ mice were slightly but significantly more glucose intolerant compared to +/+ mice (Figure 5). After 4 months of HC feeding, glucose intolerance was further augmented in ob/+ mice compared to +/+ mice (Figure 5). However, BAIBA partially prevented the aggravation of glucose intolerance in ob/+ mice (Figure 5). Plasma insulin measured during the IPGTT tended to be higher in ob/+ mice, whereas BAIBA prevented this effect (Figure 5). Finally, an IPITT was also performed in order to evaluate insulin sensitivity after 4 months, but it was only slightly altered in ob/+ mice (Table 4).
Table 4. Intraperitoneal insulin tolerance test (IPITT) in +/+ mice and ob/+ mice treated or not treated with 100 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effect of BAIBA in ob/+ mice on plasma biochemistry
Plasma levels of glucose, insulin, triglycerides, and total cholesterol were not different between ob/+ and +/+ mice before the treatment and in the fed state (Table 5). In ob/+ mice, initial plasma levels of adiponectin tended to be higher (P = 0.08), and the antioxidant status was slightly but significantly increased (Table 5). After 4 months of HC diet, plasma triglycerides were significantly increased by 38% in ob/+ mice in the fed state, and BAIBA partially prevented this augmentation (Figure 6). Total cholesterol was significantly increased in ob/+ mice by 18% and 30% in the fasted and the fed state, respectively (Figure 6). Plasma cholesterol was no longer augmented in the fasted state in ob/+ mice treated with BAIBA (Figure 6). BAIBA significantly increased plasma β-hydroxybutyrate in ob/+ mice in the fed state (Figure 6), suggesting increased mitochondrial FAO in liver. Glucose levels were similar in the three groups (Table 6). Insulin, adiponectin, and the antioxidant status were not significantly different among the different groups of mice in the fed state (Table 6). Ferritin was significantly increased in ob/+ mice, and BAIBA prevented this effect (Table 6). BAIBA slightly but significantly decreased iron in plasma (Table 6). Finally, nonesterified fatty acids were unchanged among the different groups of mice in the fasted state (Table 6).
Table 5. Plasma parameters in +/+ and ob/+ mice before the initiation of the investigations
Table 6. Plasma parameters in +/+ mice and ob/+ mice treated or not treated with 100 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effect of BAIBA in ob/+ mice on plasma liver enzymes, hepatic lipids, de novo lipogenesis, and histology
Plasma levels of ALT and LDH were measured before and after the 4-month experiment. Initial levels of ALT tended to be increased (P = 0.07) in ob/+ mice, whereas LDH levels were similar between ob/+ and +/+ mice (Figure 7). After 4 months of HC feeding, levels of ALT and LDH were significantly augmented in ob/+ mice, and BAIBA almost fully prevented this effect (Figure 7). Normalization of plasma liver enzymes suggested that BAIBA could limit liver injury. Indeed, BAIBA prevented the accretion of hepatic triglycerides in ob/+ mice (Figure 8). Moreover, a significant increase in de novo fatty acid synthesis was observed in untreated ob/+ mice, but this augmentation was no longer significant (P = 0.14) in mice treated with BAIBA (Figure 8). Liver histology was also performed in some animals as described in the Methods and Procedures section (Figure 1). Steatosis and necroinflammation were more marked in ob/+ mice when compared to +/+ mice, and BAIBA reduced the severity of these lesions in ob/+ mice (Table 7). When present, the inflammatory infiltrates were made predominantly of lymphocytes (Figure 1c,d), with a few macrophages in the portal area (data not shown). Finally, fibrosis was virtually absent, and apoptosis was not detected with the terminal deoxynucleotidyl transferase nick-end labeling assay, regardless of the groups of animals studied (data not shown).
Table 7. Percentage of hepatocytes with steatosis and score of necroinflammation in +/+ mice and ob/+ mice treated or not treated with 100 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effect of BAIBA in ob/+ mice on gene expression in liver and MTP activity
Next, the mRNA expression of several genes playing a major role in carbohydrate and lipid homeostasis was assessed by quantitative PCR in liver. The expression of genes involved in glycolysis (GK, L-type pyruvate kinase) and lipogenesis (ACC1, FAS, stearoyl-CoA desaturase-1, mitochondrial glycerol-3-phosphate acyltransferase) was significantly augmented in ob/+ mice when compared to +/+ mice, and BAIBA partially corrected this effect (Figure 9). In contrast, BAIBA did not downregulate peroxisome proliferator-activated receptor-γ mRNA levels, whereas the expression of the gluconeogenic enzyme phosphoenolpyruvate carboxykinase was unchanged among the three groups (data not shown). Interestingly, the mRNA expression of genes involved in FAO (peroxisome proliferator-activated receptor-α, liver carnitine palmitoyltransferase-1 (L-CPT1), and medium-chain acyl-CoA dehydrogenase) was significantly increased in ob/+ mice when compared to +/+ mice, and BAIBA further increased L-CPT1 expression (Figure 9). Finally, we assessed the activity of MTP, which plays a key role in apolipoprotein B lipidation and thus very low-density lipoprotein secretion (17). However, there was no difference of MTP activity in liver, whatever the groups of animals studied (data not shown).
Effect of BAIBA in ob/+ mice on pACC and FAS expression in liver
We also assessed hepatic levels of pACC, the inactivated form of ACC. The pACC/ACC ratio in liver was significantly decreased in ob/+ mice, and BAIBA partially restored this ratio, mostly by significantly (P < 0.05) increasing pACC (Figure 10). Moreover, BAIBA also partially prevented the augmentation of FAS protein in ob/+ mice (Figure 10).
Effects of BAIBA on primary culture of rat hepatocytes
Because BAIBA supplementation in mice augmented plasma β-hydroxybutyrate (Figure 6) as well as mRNA levels of L-CPT1 (Figure 9) and ACC phosphorylation (Figure 10) in liver, we asked whether this derivative could stimulate FAO directly on cultured rat hepatocytes. Plasma concentration of BAIBA was ∼300 ng/ml (i.e., 3 μmol/l) in mice treated with 100 mg/kg/day of BAIBA (15); therefore, rat hepatocytes were incubated for 72 h with 5 μmol/l of BAIBA. However, BAIBA failed to stimulate FAO in these in vitro conditions (data not shown). Moreover, FAO was unchanged with 50 and 100 μmol/l of BAIBA (data not shown). Thus, BAIBA is not able to stimulate directly FAO on hepatocytes even at high concentrations.
Assessment of oxidative stress in liver
Next, several parameters related to oxidative stress were measured in liver including GSH levels, aconitase activity, and the protein expression of MnSOD, CYP2E1, and Hsp70. GSH levels, MnSOD expression, and aconitase activity were unchanged whatever the groups of animals studied (Table 8). BAIBA significantly decreased Hsp70 expression in ob/+ mice when compared to +/+ mice (Table 8). CYP2E1 expression was augmented in ob/+ mice treated or not treated with BAIBA. Thus, there was no overt oxidative stress in liver, in particular in ob/+ mice fed the HC diet, despite a moderate increase in CYP2E1 expression.
Table 8. Parameters related to hepatic oxidative stress in +/+ mice and ob/+ mice treated or not treated with 100 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effect of BAIBA in ob/+ mice on CPT1 and UCP1 expression in adipose tissues
BAIBA increased mRNA levels of L-CPT1 in ob/+ mice fed the HC diet (Figure 9). Thus, we asked whether BAIBA could also augment CPT1 expression in WAT and BAT. In these tissues, M-CPT1 (Table 1) represents the main CPT1 isoform (23). M-CPT1 expression was significantly increased in iBAT and eWAT, by 19% and 54%, respectively, in ob/+ mice when compared to +/+ mice. In ob/+ mice, BAIBA restored M-CPT1 expression in iBAT, whereas it further enhanced significantly M-CPT1 expression in eWAT by 17%. Uncoupling protein-1 (UCP1) expression was also assessed in iBAT. UCP1 mRNA levels were moderately, but significantly, decreased by 14% in ob/+ mice fed the HC diet; however, BAIBA did not restore UCP1 expression (data not shown).
Effects of BAIBA in wild-type mice fed the HC diet
In another experiment, we asked whether BAIBA could prevent body fat accumulation due to calorie overconsumption in a few +/+ C57BL/6J mice. In this experiment, the gain of body fat mass was significantly increased in +/+ mice fed the HC diet when compared to +/+ mice fed a SC chow (Table 9). BAIBA supplementation tended (P = 0.09) to reduce adiposity in +/+ mice fed the HC diet (Table 9), but this effect was less marked when compared to ob/+ mice (Figure 3). Plasma leptin levels were significantly increased in +/+ mice fed the HC diet in comparison to +/+ mice fed the SC diet, as expected, and this augmentation was also observed when the values were expressed per gram of body fat mass (Table 9). Interestingly, whereas BAIBA did not augment the absolute values of plasma leptin, it enhanced by 31% (P = 0.09) the plasma leptin/body fat mass ratio (Table 9). However, this effect in +/+ mice was less marked when compared to ob/+ mice (Figure 4). Finally, plasma triglycerides, total cholesterol, and β-hydroxybutyrate were unchanged across all the animal groups studied (Table 9).
Table 9. Body and plasma parameters in +/+ mice fed the standard-calorie (SC) diet and +/+ mice fed the high-calorie (HC) and treated or not treated with 100 mg/kg/day of β-aminoisobutyric acid (BAIBA) for 4 months
Effects of BAIBA on leptin secretion in isolated adipose cells
The foregoing investigations suggested that BAIBA could augment leptin production differentially between +/+ and ob/+ mice, with a stimulating effect that seemed more marked in the latter group. Thus, in a last series of investigations, we asked whether BAIBA could differentially affect the secretion of leptin in primary cultured adipocytes isolated from +/+ and ob/+ mice. To this end, +/+ and ob/+ adipocytes were incubated for 48 h with 3 mmol/l of BAIBA and 3 mmol/l propionic acid, as positive control (22). Our results showed that the basal secretion of leptin was reduced by 42% in ob/+ adipose cells when compared to +/+ adipocytes (Figure 11). Moreover, whereas propionic acid and BAIBA had weak stimulating effects on leptin secretion in +/+ adipocytes, these compounds significantly augmented the levels of secreted leptin by 63% and 44%, respectively, in ob/+ adipocytes (Figure 11). However, propionic acid and BAIBA did not increase leptin secretion in ob/+ adipose cells after 24 h of incubation (data not shown). Finally, we also measured glycerol released in the culture medium. In cultured +/+ adipose cells, glycerol levels were 32 ± 1, 39 ± 4, and 37 ± 5 μmol/l, respectively, in basal conditions, and in propionate-treated and BAIBA-treated cultures (n = 6–8 values per group). In cultured ob/+ adipose cells, glycerol levels were 72 ± 16, 73 ± 11, and 121 ± 56 μmol/l, respectively, in basal conditions and in propionate-treated and BAIBA-treated cultures (n = 6–8 values per group). Thus, basal glycerol release was higher in ob/+ adipose cells when compared to +/+ mice, and BAIBA tended to enhance glycerol release in ob/+ adipocytes. Unfortunately, nonesterified fatty acid levels were too low to be accurately measured by our analytic method.
Recently, we showed that BAIBA reduced the gain of body fat mass in Swiss (lean) mice fed a standard chow, without altering food consumption (7). Moreover, BAIBA increased plasma ketone bodies and hepatic FAO in this murine model (6,7,15). However, administration for 6 weeks of 100 mg/kg/day of BAIBA to leptin-deficient ob/ob mice failed to reduce body adiposity and liver triglycerides and did not significantly augment plasma ketone bodies (7). In the present study, we formulated two hypotheses to explain why BAIBA has no beneficial effect on body fat mass and hepatic steatosis in ob/ob mice. A first hypothesis was that the dose and/or the length of administration of BAIBA were not sufficient to afford its beneficial effect. To test this possibility, ob/ob mice were treated for 4 months with BAIBA at doses of 100 or 500 mg/kg/day. Alternatively, BAIBA could need leptin to exert its beneficial effect on fat accumulation. To test this hypothesis, BAIBA was administered at the dose of 100 mg/kg/day in wild-type (+/+) mice and C57BL-ob/+ mice (i.e., harboring only one allele of the ob gene) fed for 4 months with a diet enriched in calorie.
Our results indicated that BAIBA was unable to limit body fat mass, hepatic steatosis, and glucose intolerance in ob/ob mice, even when administered for 4 months at the 500 mg/kg/day dosage (see Results section). In contrast, 100 mg/kg/day of BAIBA was able to fully prevent or limit most of the metabolic abnormalities observed in ob/+ mice fed a HC diet, including body fat mass (Figure 3), glucose intolerance (Figure 5), hypertriglyceridemia in the fed state (Figure 6), hepatic cytolysis (Figure 7), and steatosis (Figure 8, Table 7). More important, these beneficial effects in ob/+ mice were associated with higher plasma leptin levels and a lesser reduction of leptin content in eWAT and stomach (Figure 4). Furthermore, whereas BAIBA reduced body fat mass by 40% in ob/+ mice fed the HC diet (Figure 3), it lowered body adiposity by 27% in +/+ mice fed the same HC diet (see Results section). Thus, it appears that the favorable effect of BAIBA on body adiposity requires leptin and is optimal in the context of partial leptin deficiency.
An adequate leptin production in response to calorie overconsumption is needed to curb the expansion of body adiposity, thanks to leptin-induced reduction of calorie intake and increased energy expenditure (8,24,25). In the present study, the unrestricted production of leptin in +/+ mice fed the HC diet was associated with a limited gain of body fatness, whereas plasma triglycerides and total cholesterol were not significantly enhanced (Table 9). In contrast, the lower production and secretion of leptin in ob/+ mice (Figures 4 and 11) greatly favored body adiposity, dyslipidemia, glucose intolerance, and moderate steatohepatitis in the context of HC intake (Figures 3 and 5–8). BAIBA supplementation in ob/+ mice prevented diet-induced obesity and the related metabolic disorders most probably by restoring leptin production and secretion by the adipose tissue (Figures 4 and 11).
When +/+ mice were fed the HC diet, body fat mass increased but to a lower extent when compared to ob/+ mice (Figure 3), as pointed out previously. BAIBA supplementation in +/+ mice fed the HC diet tended to reduce by 27% the gain of body fat mass and to augment by 31% the plasma leptin/body fat mass ratio (Table 9). Thus, the lower ability of BAIBA to curb the gain of body fatness in +/+ mice (−27%) when compared to ob/+ mice (−40%) could be related to a lower stimulation of leptin production. Interestingly, the stimulating effect of BAIBA on leptin secretion in vitro was significant in ob/+ adipose cells but was weak in +/+ adipocytes (Figure 11).
Our previous investigations performed in RjOrl Swiss mice (that do not present leptin insufficiency) also showed that BAIBA reduced the gain of adiposity, even though these mice were fed a regular chow (7). Moreover, BAIBA supplementation enhanced by 11% the plasma leptin/body fat mass ratio, although these data in RjOrl Swiss mice were collected in the fasted state (7). Thus, although BAIBA affords optimal effects on leptin levels and body fatness in ob/+ mice, this derivative can also limit body adiposity in the context of adequate leptin expression.
Supplementation of BAIBA in ob/+ mice partially restored leptin content not only in eWAT, but also in the stomach (Figure 4). However, the latter effect was not associated with a reduction of food intake (see Results section). Although stomach-derived leptin can act as a satiety signal, it also influences nutrient absorption in the small intestine (26,27). Moreover, gastric leptin can be internalized by duodenal enterocytes and delivered to the blood circulation (28). Thanks to the portal blood flow, gastric leptin could be especially active on the liver (14), in particular through a direct action on lipid and carbohydrate homeostasis (29,30,31,32). Thus, although further investigations would be needed to determine why BAIBA did not reduce food intake in ob/+ mice, increased gastric leptin could have afforded beneficial effects on body fatness and related metabolic disorders through mechanism(s) not involving satiety.
The mechanism(s) whereby BAIBA limits leptin insufficiency in ob/+ mice is still unknown. A recent study showed the short-chain fatty acid propionate is able to stimulate leptin production in cultured mouse adipocytes through the G protein-coupled receptor GPR41 (22). Furthermore, the data indicated that butyrate and isobutyrate can activate GPR41, whereas hydroxybutyrate and γ-amino-n-butyrate were ineffective (22). Alternatively, some amino acids can augment leptin secretion from white adipocytes, especially valine and methionine which produce succinyl-CoA and acetyl-CoA (33). Increased secretion of leptin by these amino acids could be related to the stimulation of the tricarboxylic acid cycle (33). Interestingly, BAIBA is degraded within mitochondria into several metabolites including succinyl-CoA (see below). Thus, further investigations would be needed to determine if BAIBA is able to increase leptin secretion on isolated adipocytes and whether this could be due to GPR41 activation and/or to the stimulation of the tricarboxylic acid cycle.
Whereas BAIBA and propionic acid enhanced leptin secretion in cultured adipose cells isolated from ob/+ mice (Figure 11), only BAIBA tended to increase glycerol release in the culture medium (see Results section), thus suggesting that BAIBA could favor lipolysis. Earlier in vitro investigations showed that propionic acid inhibited lipolysis induced by adrenalin or isoproterenol and promoted adipogenesis (34,35). Thus, the respective effects of BAIBA and propionic acid on leptin secretion and lipolysis could be divergent.
We confirmed herein our previous data showing that BAIBA is able to increase L-CPT1 expression and FAO in liver (6,7). Moreover, BAIBA enhanced the pACC/ACC ratio in this tissue (Figure 10), thus suggesting that BAIBA could decrease the levels malonyl-CoA, the endogenous inhibitor of CPT1. Interestingly, leptin seems to augment CPT1 expression and/or activity in different tissues including WAT and liver (36,37,38) and to phosphorylate hepatic ACC through an AMPK-independent pathway (29). In addition, leptin is able to reduce the hepatic expression of glycolytic and lipogenic enzymes such as GK, FAS, and stearoyl-CoA desaturase-1 (15,39,40). Thus, the beneficial effects of BAIBA on liver could be, at least in part, mediated by increased leptin signaling in this tissue, which could favor mitochondrial FAO and limit de novo lipogenesis. Finally, although this study and previous investigations (6,7) showed that BAIBA increased FAO in liver, further experiments would be needed to determine whether BAIBA could favor FAO in extra-hepatic tissues. As BAIBA augmented M-CPT1 expression in eWAT (see Results section), it is tempting to speculate that it could favor FAO in extra-hepatic tissues as well.
When administered to ob/ob mice, BAIBA did not reduce body adiposity and hepatic steatosis, yet it reduced necroinflammation (Table 2) as well as portal and perisinusoidal fibrosis (Figure 2). Furthermore, a strong reduction of apoptosis (Table 2) and a significant decrease in plasma LDH were observed in ob/ob mice treated with 500 mg of BAIBA (Table 3). Prevention of apoptosis in ob/ob mice is associated with lower levels of plasma liver enzymes (11). Interestingly, BAIBA increased GSH levels in ob/ob liver (see Results section). Although further investigations would be needed to determine the mechanism(s) whereby BAIBA can ameliorate necroinflammation, apoptosis, and fibrosis in ob/ob mice without ameliorating fatty liver, some data suggest that BAIBA may have some beneficial effects unrelated to leptin. Indeed, BAIBA is a partial agonist of the glycine receptor (41), and glycine appears to exert several protective actions in different tissues including the liver, in particular through anti-inflammatory and direct cytoprotective effects (42,43). Moreover, patients with dihydropyrimidine dehydrogenase or β-ureidopropionase deficiencies have low plasma concentrations of BAIBA and elevated urine levels of 8-hydroxydeoxyguanosine (44,45), thus suggesting that BAIBA could exert some systemic protections against oxidative stress.
BAIBA supplementation prevented the augmentation of plasma ferritin in ob/+ mice fed the HC diet (Table 6). Serum levels of ferritin can be increased in nonalcoholic fatty liver diseases, with higher levels in nonalcoholic steatohepatitis compared to simple steatosis (46,47). Increased ferritin could be independent of iron stores, and data suggest that it could be secondary to inflammation (47,48). Hence, reduction of ferritin levels in ob/+ mice treated with BAIBA could reflect lower necroinflammation (Table 7).
Total leptin deficiency is extremely rare in human (49). In contrast, partial leptin deficiency could be much more prevalent. Accordingly, it is estimated that 5–10% of obese human subjects are low leptin secretors (8,50). Because obesity itself upregulates leptin expression in WAT, the prevalence of partial leptin deficiency could be much greater in the general population, although data regarding this major issue are still lacking. Interestingly, a common promoter polymorphism (–2548G/A) in the leptin (ob) gene significantly influences leptin expression in adipose tissue and its plasma concentrations, although there are some discrepancies regarding the exact impact of each ob variant on these levels (50,51,52,53,54). Nevertheless, inherited low levels of leptin can increase the risk of being overweight (or obese) in mice (55) and humans (52,56,57). Thus, many individuals could be low leptin secretors and thus be at risk for obesity and other metabolic disorders linked to calorie overconsumption. Treatment with leptin could be an option for these individuals, but repeated subcutaneous injections of leptin can induce mild-to-moderate reactions at the administration site such as inflammation and pruritus (58). Alternatively, drugs that are able to enhance indirectly the plasma leptin levels could help prevent obesity in low leptin secretors, especially in the context of calorie overconsumption.
BAIBA is a natural catabolite of thymine, which is further degraded within mitochondria, especially in liver, into downstream catabolites such as propionyl-CoA, methylmalonyl-CoA, and succinyl-CoA (2,59). BAIBA is also a catabolite of the antiretroviral drugs stavudine (d4T) and zidovudine (AZT), but unlike these derivatives, it cannot impair mitochondrial DNA replication (6,7). BAIBA toxicity seems low, and plasma concentration of BAIBA associated with mouse mortality was reported to be ∼35 mmol/l (60). More important, plasma concentration was ∼3 μmol/l in RjOrl Swiss mice treated with 100 mg/kg/day of BAIBA (15), and in the present study, this derivative presented some beneficial effects in ob/ ob livers for the 500 mg/kg/day dosage. Thus, although further investigations should thoroughly evaluate BAIBA toxicity, this endogenous derivative may have a favorable safety profile that could be attractive for pharmacological usages.
K.B. has been the award winner of a Nestlé Nutrition grant in 2006. We are indebted to Yannis Ducourant for her help in animal care and to Véronique Descatoire for the primary culture of rat hepatocytes. We are grateful to André Bado for his advices regarding the assessment of leptin in the gastric mucosa and helpful discussion.