Branched-Chain Amino Acids Exacerbate Obesity-Related Hepatic Glucose and Lipid Metabolic Disorders via Attenuating Akt2 Signaling

Obesity and associated metabolic diseases, such as insulin resistance and type 2 diabetes mellitus (T2DM), have become major health problems worldwide (1). Obesity is associated with increased lipid storage in ectopic tissues, such as the liver, skeletal muscle, and kidney (2). Ectopic lipid accumulation in the liver is known as nonalcoholic fatty liver disease (NAFLD) and is closely associated with insulin resistance (3). NAFLD can be divided into simple steatosis and nonalcoholic steatohepatitis (NASH). As known, hepatic fat diminishes with the progression of NASH, which has been designated burned-out NASH (4). This phenomenon implicates altered hepatic lipid metabolism with the progression of NAFLD, but the associated mechanisms remain unknown.

The liver is a key organ in response to insulin, controlling glucose and lipid metabolism (5). Normally, insulin halts hepatic gluconeogenesis and promotes hepatic de novo lipid synthesis (6). Both of these insulin-mediated effects are exerted partly downstream of Akt2 (7,8). In the liver, insulin stimulates sterol-regulatory element binding protein (SREBP) activation via promoting Akt2 phosphorylation (9). Akt2 stimulates SREBP1-mediated enhanced expression of lipogenic genes, such as fatty acid synthase (FASN) and stearoyl–CoA desaturase-1 (SCD1) (10). Furthermore, the insulin/Akt2/Foxo1 signaling pathway controls hepatic glucose production (8). Under T2DM conditions, the liver usually exhibits selective insulin resistance: insulin fails to suppress gluconeogenesis but continues to activate lipogenesis, thereby contributing to hyperglycemia, hypertriglyceridemia, and fatty liver (6). This could be because the Foxo1 pathway remains insensitive, while the SREBP-1c pathway remains sensitive to insulin under such conditions (11). Animals with liver-specific knockout of Akt2 are protected from hepatic lipid accumulation in obese, insulin-resistant states (7). Importantly, the protection from hepatic lipid accumulation in the Akt2 knockout model is accompanied by reduced expression of SREBP1c and decreased de novo lipogenesis. Furthermore, liver insulin receptor-knockout mice manifest low liver and plasma triglyceride (TG) levels despite the presence of hyperglycemia and hyperinsulinemia (12). On the basis of these findings, it was proposed that when insulin-Akt2 signaling is impaired to extreme extent, the liver will display total insulin resistance: insulin will fail to halt gluconeogenesis and will not be able to promote lipogenesis.

Dietary carbohydrates and fats are causatively associated with the development of insulin resistance (11). Although the hazardous effects of high glucose and high fat (HF) on hepatic metabolic homeostasis are well recognized, the impact of high protein intake on obesity-associated hepatic metabolic disorders remains unclear. BCAAs (leucine, isoleucine, and valine) are a group of essential amino acids. Relatively abundant in food, they account for ∼15–25% of total protein intake (13). Recent studies suggest that BCAAs are associated with the development of T2DM (14) and NAFLD (11). Notably, elevated circulating BCAAs concentrations strongly predict the risk of T2DM (15). Targeting BCAA catabolism is a potential strategy to treat obesity-associated insulin resistance (16). Together, these clinical studies suggest that BCAAs may have a negative impact on hepatic metabolic homeostasis, particularly in insulin-resistant states. However, whether elevated BCAAs play a causative role in disturbing hepatic glucose and lipid metabolism remains unidentified.

Here, we seek to elucidate the role of BCAAs on hepatic glucose and lipid metabolism and on facilitating the development of insulin resistance into severe states and systemic glucose and lipid metabolic disorders, and to further determine the significance and clarify the underlying mechanisms.

All animal experiments were performed according to the National Institutes of Health guidelines on laboratory animals and were approved by the Air Force Military Medical University Animal Care and Use Committee. Adult male wild-type C57BL/6J mice were purchased from the Air Force Military Medical University Animal Center. All mice were raised in a temperature-controlled facility and maintained on the background of 26 ± 2°C and 55 ± 15% relative humidity with a constant 12-h light/dark cycle.

Wild-type C57BL/6J mice were randomly divided into four groups: a normal diet (ND)-fed group, an HF diet–fed group, an HF plus BCAA–fed (HF/BCAA) group, and an HF/paired group in which caloric intake was matched to that of the HF/BCAA group. All mice were fed for 16 weeks with free access to water or food. The body weight and food intake were monitored every 3 days and calculated as kcal/g/week. Adeno-associated virus serotype-9 (AAV9)-myr-Akt2 with a wide cytomegalovirus promoter was delivered through tail vein injection, and 1.2 × 10¹¹ vectors or genomes were administrated for every mouse. The sequence of myr is atggggagcagcaagagcaagcccaagtctaga.

Mice were euthanized by isoflurane. Blood was collected via the carotid artery. The tissues were collected after animal sacrifice and frozen immediately in liquid nitrogen. For fasted-refeeding studies, mice were fasted overnight and then refed for 6 h.

The paired-feeding protocols were performed as previously described (17). To conduct a paired-feeding assay, we first determined the total body weight of the HF/BCAA group (A g) and calculated the total calories consumed (B) each day (B = [total calories provided − calories in uneaten food]/day) and the calories consumed per day based on the BCAAs levels in the drinking water (C cal). We then determined the calories consumed per gram body weight (D cal/g) = (B [cal] + C [cal])/total body weight (A g). Next, we measured the total body weight (E g) of the HF/paired group and calculated the amount of HF diet food that should be administered to the HF/paired group (F cal) (F = D∗E) on the following day. The total food was divided into two portions and provided at separate times (8:00 a.m. and 6:00 p.m.) to avoid starvation. A 60% HF diet without BCAAs was provided to the HF/paired group.

The HF diet (60% of kcal from fat, D12492) was used to induce obesity-associated insulin resistance as previously reported (18), and an ND (10% of kcal from fat, D12450) was used as the ND that was fed to the control group. Both foods were purchased from Research Diets. BCAAs (4%) were dissolved into drinking water at a leucine-isoleucine-valine ratio of 2:1:1 (19).

Glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were performed as previously reported (20). Mice were injected intraperitoneally with human insulin (Eli Lilly) at 0.75 IU/g body wt or 20% glucose at 2.0 mg/g body weight after overnight food deprivation. Blood glucose was measured 0, 15, 30, 60, 90, and 120 min after injection.

The mice were fasted for 16 h before an intraperitoneal injection of sodium pyruvate (2 g/kg body wt). Plasma glucose was measured at 0, 15, 30, 60, 90, and 120 min after the injection, and blood samples were collected from the tail vein.

Primary hepatocytes were isolated from 8- to 10-week-old adult male WT C57BL/6J mice. Hepatocyte isolation was performed using a two-step collagenase perfusion method, as previously reported (21). Briefly, perfusion through the hepatic portal vein commenced successively with 40 mL PBS, 20 mL D-Hanks’ solution containing 0.1 mmol/L EDTA, and 40 mL 0.05% type IV collagenase solution. The liver was surgically removed, and primary hepatocytes were plated in collagen I-coated dishes. Primary hepatocyte viability, assessed by trypan blue exclusion, was >90%.

Adenovirus (Vector Biolabs) infection was performed 6 h after plating (multiplicity of infection 10). After 6 h of infection, the cells were washed twice with PBS and serum starved overnight before insulin stimulation. Nontargeting control and INSIG2 siRNA or Mul1 siRNA were transiently transfected into primary hepatocytes 6 h after plating with Lipofectamine 2000 (Invitrogen, Carlsbad, CA). At 24 h after transfection, the cells were serum starved overnight before insulin stimulation. The sequences of the INSIG2-siRNAs were as follows: CGGUGUUCGUGGGUAUAAATT (sense strand) and UUUAUACCCACGAACACCGTT (antisense strand). The sequences of the Mul1-siRNAs were as follows: GCGAGAGGCCCAAAGGCAUTT (sense strand) and AUGCCUUUGGGCCUCUCGCTT (antisense strand).

Glucose production tests in primary hepatocytes were performed as previously described (11). Hepatocytes were serum starved for 24 h and replaced with glucose production medium and incubated for 5 h with vehicle, 100 μmol/L 8-CPT–cAMP (Sigma-Aldrich), or cAMP plus 10 nmol/L insulin. Glucose levels in culture medium were determined by peroxidase-glucose oxidase assay (Sigma-Aldrich) and normalized to protein levels in the same well. For the Aktviii treatment group, Aktviii was added to serum-free medium 30 min before insulin treatment or vehicle. Additional cells were used to determine the expression of key genes controlling gluconeogenesis.

Lysates from mouse liver tissue and primary hepatocytes were prepared. Total protein from liver tissue and primary hepatocytes was isolated using radioimmunoprecipitation assay lysis buffer (Beyotime, Shanghai, China) with protease and phosphatase inhibitors. The protein concentration was quantified using a bicinchoninic acid protein assay kit (Beyotime). The protein extracts were separated by SDS-PAGE, transferred to nitrocellulose membranes (Millipore), and incubated with the indicated antibodies (listed in Supplementary Table 1), followed by horseradish peroxidase–conjugated secondary antibody incubation. The antibodies were diluted 1:1,000 in Tris-buffered saline. The bands were detected using the Immobilon Western chemiluminescent horseradish peroxidase substrate (Millipore) and imaged using ChemiDoc Imaging Systems (Bio-Rad). β-Actin served as a control to normalize protein expression.

Cells in six-well plates were washed three times with PBS and lysed in ice-cold radioimmunoprecipitation assay lysis buffer with protease/phosphatase inhibitors. Lysates were incubated on ice for 30 min and then centrifuged at 12,000g for 15 min. The supernatants were incubated with primary antibodies for over 12 h at 4°C. Thereafter, a suspension of Dynabeads Protein G (Life Technologies) was added to the samples and incubated at 4°C for 2 h. The immunoprecipitated proteins were eluted, denatured, and boiled for 5 min for Western blotting.

For gene expression analyses, total RNA from primary hepatocytes and liver tissue was extracted using the TRIzol Reagent (Invitrogen). cDNA was synthesized using an RNA PCR kit (Takara Bio, Shiga, Japan) per the manufacturer’s instructions. SYBR Green-based quantitative real-time PCR was performed using an Applied Biosystems 7300 Real-Time PCR System. Triplicate samples were harvested for each treatment. Relative expression values were calculated as the ratio of target cDNA to β-actin. All data were analyzed using GraphPad Prism 7 software. Primer pair sequences are listed in Supplementary Table 1.

Statistical significance was assessed as follows: data sets with more than two groups were analyzed with ANOVA, followed by unpaired two-tailed t tests, and data sets for two groups were assessed with unpaired two-tailed t tests. P values of <0.05 were considered statistically significant. Data are presented as the mean ± SD. Western blot quantification was performed using Quantity One software.

The data sets generated during the current study are available from the corresponding author upon reasonable request. All noncommercially available resources generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To illustrate the effect of BCAAs on hepatic lipid and glucose metabolism and on ectopic lipid deposition, we fed mice the ND diet, HF diet, or HF/BCAA diet for 16 weeks. HF diet feeding successfully induced insulin resistance, as demonstrated by significant increases in body weight gain, fasted plasma glucose, and deteriorated glucose tolerance compared with the ND group (Supplementary Fig. 1A–C). Since BCAAs supplementation inhibited daily food intake (22), we included another HF/paired group. The body weight gain of the HF/BCAA group was markedly slower than that of the HF group (Supplementary Fig. 2A) but was nearly consistent with that of the HF/paired group. This was likely attributable to the lower caloric intake of the HF/BCAA group (Supplementary Fig. 2B). Analysis of hepatic BCAAs and BCKAs levels revealed significantly accumulated leucine, isoleucine, valine, and their catabolic metabolites α-ketoisocaproic acid and α-keto-β-methylvaleric acid in the HF/BCAA-fed mouse liver tissues (Supplementary Fig. 2C). BCAAs supplementation significantly reduced HF-induced hepatomegaly and decreased the liver-to-body weight ratios (Fig. 1A). The gross morphology of the liver indicated that BCAA reduced hepatic steatosis, as evaluated by hematoxylin and eosin and Oil Red O staining (Fig. 1B, left: upper and middle panels), but increased glycogen accumulation, as indicated by periodic acid Schiff (PAS) staining (Fig. 1B, left: lower panel). Consistently, the HF/BCAA-fed mice exhibited decreased hepatic TG content but increased glycogen content (Fig. 1C). In addition, the plasma LDL and TG levels were increased in the HF/BCAA-fed mice (Supplementary Fig. 2D), but the plasma cholesterol and HDL contents were not significantly changed (Supplementary Fig. 2E). The accumulation of muscular lipids (Fig. 1D), kidney lipids (Fig. 1E, upper and Fig. 1F, left), and urinary microalbuminuria (Fig. 1F, right) was substantially increased in the HF/BCAA mice. Renal tubule casts were evidently observed in the HF/BCAA mice (Fig. 1E, lower). These results suggest that BCAAs supplementation induces redistribution of lipids among the liver, muscle, and kidney and exacerbates renal injury in HF diet-induced obese (DIO) mice.

Because the liver is a crucial organ controlling glucose and lipid metabolism in the body, we comprehensively analyzed hepatic glucose and lipid metabolism. We found that fatty acid β-oxidation showed no difference among the three groups (Supplementary Fig. 3A). Molecular analysis showed that BCAAs supplementation inhibited the expression and processing of SREBP1 and inhibited the expression of its targets, Fasn and Scd1 (Fig. 2A and B). In addition, hepatic genes regulating TG output were not significantly increased (Supplementary Fig. 3B). However, BCAAs supplementation enhanced the plasma free fatty acid (FFA) levels (Supplementary Fig. 3C) and modestly increased the TG output under fasting conditions (Supplementary Fig. 3D). In addition, BCAAs supplementation significantly enhanced fasting and feeding blood glucose levels (Fig. 2C), impaired pyruvate tolerance (Fig. 2D), inhibited Foxo1 phosphorylation (Fig. 2E), increased the relative amount of Foxo1 protein in the nucleus (Fig. 2F), and increased the expression levels of gfbp1 and Foxo1-targeted genes, including Pck1 and G6pc (Fig. 2E and G). Taken together, these data systematically illustrated that BCAAs supplementation inhibits hepatic lipid lipogenesis, promotes hepatic TG output, and enhances glycogen production in DIO mice.

Insulin promotes lipid deposition and inhibits gluconeogenesis via Akt2 signaling in hepatocytes (11). We thus wondered whether BCAAs supplementation influences hepatic insulin sensitivity. We found that BCAAs exacerbated HF-induced systemic insulin resistance, as shown by the GTT and ITT results (Fig. 3A and B). Moreover, the fasting plasma insulin levels were significantly elevated (Fig. 3C) in HF/BCAA-fed mice. These data suggest that BCAAs affect hepatic lipid and glucose metabolism by attenuating insulin-Akt2 signaling. Therefore, we detected the expression and phosphorylation levels of Akt1, Akt2, and Akt3 and found that BCAAs significantly reduced the protein expression and phosphorylation levels of hepatic Akt2 and Akt1 (Fig. 3D) but did not affect hepatic Akt2 mRNA levels (Fig. 3E), which demonstrated that BCAAs might affect hepatic Akt2 in a posttranslational manner. We then found that 10 mmol/L BCAAs or 0.5 mmol/L BCKAs alone induced Akt2 protein downregulation, while 0.5 mmol/L BCKAs did not affect the protein levels of Akt1 and Akt3. Moreover, in the presence of 0.25 mmol/L palmitate, BCAAs and BCKAs inhibited the Akt1, Akt2, and Akt3 protein levels in a dose-dependent manner in vitro (Fig. 3F) without affecting the Akt2 mRNA levels (Fig. 3G). Besides, we further confirmed that valine and its catabolite 2-ketoisovalerate are the primary components that induce Rictor and Akt2 downregulation (Supplementary Fig. 4A and B). Collectively, these data suggest that in DIO mice, BCAAs supplementation severely suppresses hepatic insulin signaling to Akt and induces hepatic Akt2 protein degradation. Akt2 is the isoform of Akt that exhibits the most sensitive response to BCAAs stimulation.

The ubiquitin-proteasome system–mediated protein degradation critically regulates hepatic Akt2 expression (23). Under HF treatment condition, because BCAAs and their metabolites BCKAs substantially accumulated in HF/BCAA-fed mouse livers and contributed to Akt2 downregulation, we used 3 mmol/L BCAAs/0.2 mmol/L BCKAs mixtures in cell experiments. As shown in Fig. 4A, both the E1 ubiquitin-activating enzyme inhibitor Pyr41 and the proteasome inhibitors MG132 and bortezomib (Supplementary Fig. 4C) blocked BCAAs/BCKAs-induced downregulation of Akt2, revealing that the ubiquitin-proteasome pathway mediates BCAAs-induced Akt2 degradation in hepatocytes. Consistently, an immunoprecipitation (IP) assay showed that BCAAs and BCKAs increased Akt2 ubiquitination (Fig. 4B). mTORC1 and mTORC2 are two key signaling pathways in response to amino acids (24). Previous studies reported that in the absence of mTORC2, nascent Akt cannot be phosphorylated at the threonine 450 site and is thereafter degraded by the ubiquitin-proteasome system (25). We found that BCAAs strongly activated p70s6k while moderately downregulating Rictor, an essential component of the mTORC2 protein complex (26) (Fig. 4C). In contrast, BCKAs significantly reduced Rictor protein levels but modestly activated p70s6k (Fig. 4C). It appears that BCAAs and BCKAs exert selective influences on mTORC1 and mTORC2, respectively. In the livers of HF/BCAA-fed mice, mTORC1 was activated and mTORC2 was downregulated (Fig. 4D). Therefore, the mixture of BCAAs and BCKAs can simultaneously activate mTORC1 and downregulate Rictor in vitro (Fig. 4E), mimicking the effect of accumulated BCAAs and BCKAs in vivo.

To determine whether mTORC1 or mTORC2 contributes to Akt2 degradation, we used rapamycin, a selective inhibitor of mTORC1, to inhibit mTORC1 signaling. As shown, blocking mTORC1 did not affect BCAAs/BCKAs-induced Akt2 downregulation but partially restored hepatocyte Akt phosphorylation (Supplementary Fig. 4D). We subsequently found that adenovirus-mediated Rictor overexpression significantly blocked BCAAs/BCKAs-induced Akt2 ubiquitination (Fig. 4F) and eliminated the inhibitory effects of BCAAs/BCKAs on the phosphorylation of Akt2 and Akt2 proteins (Fig. 4G). These data suggest that BCAAs and their products BCKAs induce Akt2 ubiquitination and degradation in an mTORC2-dependent manner and suppress Akt activation via mTORC1- and mTORC2-dependent pathways.

Ubiquitin-proteasome–dependent degradation of Akt2 protein is mediated by the consecutive actions of ubiquitin enzymes, and the E3 ubiquitin ligases determine the specificity of the system (27). Akt2 lysine ubiquitination is primarily mediated by BRCA1 (28), Mul1 (29), CHIP (30), TTC3 (31), TRAF6 (32), and Skp2 (33). However, the ubiquitin ligase enzyme that contributes to BCAAs/BCKAs-induced Akt2 degradation remains unknown. We found that BCAAs/BCKAs failed to enhance the combination of Akt2 with TRAF6 (Supplementary Fig. 5A). Co-IP experiments showed that Akt2 failed to bind to TTC3, CHIP, and BRCA1 in the presence of BCAAs/BCKAs (Supplementary Fig. 5B–E). However, we found that Mul1 expression was significantly upregulated by BCAAs/BCKAs treatment in hepatocytes (Fig. 5A). Rictor overexpression did not suppress the upregulation of Mul1 induced by BCAAs/BCKAs (Fig. 5B). Immunofluorescence assays showed that BCAAs/BCKAs promoted the colocalization of Akt2 and Mul1 in the nucleus and that this effect was reversed by Rictor overexpression (Fig. 5C). Co-IP also confirmed that BCAAs/BCKAs promoted Akt2 and Mul1 binding and that Rictor overexpression significantly inhibited the binding of these two proteins (Fig. 5D and E). Importantly, siRNA-mediated Mul1 knockdown significantly blocked BCAAs/BCKAs-induced Akt2 downregulation (Fig. 5F). Collectively, these data reveal that the E3 ligase Mul1 is required for BCAAs/BCKAs-mTORC2–induced Akt2 ubiquitination and degradation.

To determine whether attenuation of Akt signaling underlies the defect in lipogenesis and enhanced gluconeogenesis observed in the livers of HF/BCAA-fed mice, we used a membrane-targeted constitutively active allele of Akt2 (myr-Akt2) to directly restore hepatic Akt signaling. The construction of AAV9-myr-Akt2 is presented in Supplementary Fig. 6A. We found that AAV9-mediated Akt2 overexpression significantly restored the phosphorylated and total protein levels of hepatic Akt2 in HF/BCAA-fed mice (Fig. 6A and B). Notably, Akt2 overexpression upregulated hepatic TG content, enhanced HF-induced hepatic steatosis, as indicated by Oil Red O staining in response to the HF/BCAA diet (Fig. 6C), and clearly increased the mRNA and protein levels of SREBP1, SCD1, and FASN (Fig. 6D). These data reveal that Akt2 downregulation underlies BCAAs/BCKAs-mediated suppression of hepatic lipogenesis. In addition, Akt2 overexpression improved the HF/BCAA-impaired glucose tolerance, insulin tolerance (Supplementary Fig. 6B and C), and pyruvate tolerance (Supplementary Fig. 6D) and decreased the fasting and refeeding blood glucose levels (Supplementary Fig. 6E), which indicated that Akt2 overexpression ameliorated BCAAs supplementation-associated disorders of glucose homeostasis. Moreover, AAV-9–mediated Akt2 upregulation significantly suppressed hepatic glycogen accumulation (Fig. 6E) and decreased hepatic gfbp1 mRNA levels and Pck1 and G6pc expression in HF/BCAA-fed mice (Fig. 6F). These data suggest that weakened Akt2 signaling is responsible for the severely disordered glucose and lipid metabolism observed in the livers of HF/BCAA mice.

To determine the mechanism underlying severe hepatic glucose and lipid metabolic disorders, we further analyzed the molecules downstream of Akt2 that control glucose and lipid metabolism, such as GSK3, LXR, mTORC1, and INSIGs signaling (regulating lipogenesis) (34–36) and Foxo1 signaling (regulating gluconeogenesis) (37).

In vitro, we used Akt inhibitor AktViii with 1, 2.5, 5, and 10 μmol/L to gradually weaken hepatocytes Akt signaling. The results showed that 1 μmol/L Aktviii clearly decreased the phosphorylated (p)-Foxo1 levels (Supplementary Fig. 7A), but significant suppression of processed SREBP1(p) was only observed at concentrations of 5 μmol/L Aktviii or higher (deeply inhibited Akt signaling) (Supplementary Fig. 7B). The relatively low sensitivity of the GSK3, LXR, and INSIGs pathways to Akt signaling compared with that of Foxo1 might explain the delayed inhibition of lipogenesis after the occurrence of enhanced gluconeogenesis in severely insulin-resistant hepatocytes (Supplementary Fig. 7B). In vivo, we detected no difference of hepatic p-GSK3α and p-GSK3β or LXRα and LXRβ among the three groups under fasted-refed conditions (Supplementary Fig. 7C). mTORC1 promotes lipogenesis (38) and is activated in HF/BCAA-fed animals, which indicates that this molecule cannot be responsible for the inhibition of hepatic SREBP1c in HF/BCAA mice. We subsequently analyzed the INSIGs signaling pathway and found that BCAAs upregulated the hepatic expression of INSIG2a (Fig. 7A and B) but not INSIG1 or INSIG2b in HF/BCAA-fed mice (Supplementary Fig. 7D and E). The upregulation of INSIG2a was completely reversed by Akt2 overexpression (Fig. 7C). These results suggest that BCAAs/BCKAs-induced Akt2 downregulation is a major cause of hepatic INSIG2a upregulation. In vitro, overexpression of INSIG2 eliminated Ad-Akt2 transfection–induced SREBP1(p) expression in the palmitate plus BCAAs/BCKAs-treated group (Fig. 7D). Knockdown of INSIG2 with siRNA reversed BCAAs/BCKAs-induced inhibition of lipogenesis (Fig. 7E). These results suggest that BCAAs/BCKAs suppress hepatic lipogenesis by interrupting Akt2-INSIG2a signaling.

Pertaining to gluconeogenesis-correlated mechanisms, we found that AAV9-mediated Akt2 upregulation significantly reversed BCAAs-induced Foxo1 activation, as indicated by increased Foxo1 phosphorylation (Fig. 7F). In vitro, we found that Foxo1 overexpression interrupted Ad-Akt2–mediated suppression of hepatic Pck1 and G6pc expression (Fig. 7G). Additionally, the Foxo1 inhibitor AS1842856 significantly blocked BCAAs/BCKAs-enhanced gluconeogenesis (Fig. 7H and I). These results suggest that BCAAs/BCKAs promote hepatic glycogenesis via blocking Akt2 signaling and subsequent activation of Foxo1 during obesity.

To determine whether hepatic glucose and lipid metabolic disorders contribute to increased kidney and muscular lipid accumulation and renal injury, we observed lipid deposition in muscle and kidney and renal function. Akt2 overexpression inhibited plasma FFA levels (Supplementary Fig. 8A) and inhibited hepatic VLDL-TG output (Fig. 8A) and TG production (Fig. 8B). Improved hepatic insulin resistance plus reduced plasma FFA levels might ameliorate hepatic VLDL-TG output; in addition, plasma LDL and TG levels were both significantly decreased (Fig. 8C). Microalbuminuria levels and renal lipid accumulation were evidently decreased (Fig. 8D and E). We also observed an amelioration of tubule casts (Fig. 8E, lower) after Akt2 overexpression. In addition, lipid deposition in skeletal muscles was also significantly improved (Fig. 8F). These results suggest that Akt2 overexpression improves blood lipid levels and ameliorates lipid accumulation and kidney injury.

In the current study, we made several important observations. First, we demonstrated that BCAAs supplementation causes decreased liver lipid accumulation while aggravating muscle and kidney lipid deposition and renal injury in DIO mice. BCAAs supplementation enhanced gluconeogenesis, inhibited lipogenesis, and moderately augmented hepatic TG output, indicating that BCAAs inhibited hepatic lipid buffer and storage function and promoted hepatic lipid release. In addition, our previous study demonstrated that BCAAs stimulated adipocyte lipolysis (19), increasing circulating FFA levels in DIO mice, which might promote hepatic FFA delivery. Restoring hepatic Akt2 signaling reversed hepatic glucose and lipid metabolic disorders, inhibited the hepatic VLDL-TG output, decreased the plasma and hepatic lipids, and partially improved muscular and renal lipid deposition and renal function. A recent study demonstrated that 3-hydroxyisobutyrate, a catabolic intermediate of valine, activates endothelial fatty acid transport and promotes lipid accumulation in muscle (39). In this study, we found that BCAAs supplementation promoted lipid infiltration not only into muscular tissue but also into the kidney. The increased ectopic accumulation of lipids might be partially due to loss of the hepatic lipid buffer and storage function and the promotion of hepatic lipid release driven by BCAAs. Genetic evidence suggests that Akt2 is a major effector of insulin signaling for the induction of hepatic lipogenesis (7). Moreover, insulin signaling plays a vital role in regulation of VLDL synthesis and secretion (40). In DIO mice, BCAAs supplementation seriously inhibited hepatic Akt2 signaling, which might damage the liver’s ability to buffer blood lipids and resulted in hyperlipidemia. In addition, our previous study demonstrated that BCAAs stimulate adipocyte lipolysis (19), increasing circulating FFA levels, which might promote hepatic FFA delivery.

Second, we demonstrated that BCAAs supplementation significantly disrupts hepatic glucose and lipid metabolism and aggravates liver insulin resistance in DIO mice. Specifically, BCAAs inhibited lipogenesis via blocking insulin/Akt2/INSIG2a/SREBP1c signaling and reduced hepatic fat accumulation. Moreover, BCAAs increased hepatic glycogenesis and exacerbated obesity-associated hyperglycemia via insulin/Akt2/Foxo1 signaling. Akt2 signaling is upstream of the insulin-mediated regulation of lipogenesis and gluconeogenesis. Akt2-SREBP1 signaling mediates the insulin-stimulated hepatic expression of the lipogenic genes Fasn and SCD1 (7). Akt2-Foxo1 is involved in the insulin-mediated regulation of glucose production by suppressing the expression of G6pc and Pepck (41). BCAAs accumulation gravely inhibited hepatic Akt2 signaling and led to exacerbated insulin resistance, increased glucose production, and inhibited lipogenesis. Our data indicate that BCAAs supplementation in DIO mice leads to very serious liver insulin resistance: insulin fails not only to suppress gluconeogenesis but also to activate lipogenesis. For the first time, we systematically evaluated the effects of BCAAs on hepatic glucose and lipid metabolism. Given that BCAAs account for nearly 20% of total protein intake (13), the current study calls to attention that the combination of dietary BCAA and fat intake may cause severe disorders in hepatic glucose and lipid metabolism.

Third, we demonstrated that BCAAs block insulin-mediated Akt2 activation by inducing Akt2 ubiquitination and degradation. A previous study demonstrated that BCAAs interrupt insulin-induced Akt phosphorylation via mTORC1-mediated phosphorylation of insulin receptor substrate 1 at serine residues (14). BCAAs negatively regulate KLF15 through inhibited Akt signaling in mouse embryonic fibroblast cells (42). In the current study, we observed that BCAAs directly reduced total Akt2 expression and promoted Akt2 ubiquitin-dependent degradation. Furthermore, we also found that Mul1, an E3 ligase, was responsible for BCAAs-mediated Akt2 downregulation. These data provide a novel explanation for BCAAs-mediated hepatic insulin resistance and suggest that targeting BCAA metabolism might be an effective strategy for improving insulin-Akt signaling in insulin-resistant states such as NAFLD, T2DM, and metabolic syndrome.

Fourth, we demonstrated that BCAAs supplementation atop the HF diet caused an imbalance between mTORC1 and mTORC2 signaling. As expected, we observed that BCAAs activated hepatic mTORC1, but we also found that BCAAs/BCKAs directly reduced the expression of Rictor, an essential component of the mTORC2 complex. As reported, the mTORC2 complex can inhibit Akt degradation by inhibiting its ubiquitination (25). We also found that overexpression of Rictor reversed BCAAs-mediated Akt downregulation. These data provide a novel mechanism underlying BCAA-associated insulin resistance. BCAAs blocked insulin-mediated Akt phosphorylation through mTORC1-dependent insulin receptor substrate phosphorylation. On the other hand, BCAAs induced Rictor downregulation through mTORC2 inactivation, therefore leading to the ubiquitination and degradation of Akt2.

We noticed that leucine supplementation on the HF diet improves glucose tolerance and the plasma levels of total and LDL-cholesterol (18). There are some differences between this previous study and our experiments. In the previous study, the HF diet was supplemented with leucine alone, whereas we added 4% BCAAs (leucine-isoleucine-valine ratio = 2:1:1). Different components might play different roles on cellular metabolism. Another study reported that the valine-related catabolite 3- hydroxyisobutyrate drives vascular fatty acid transport and causes insulin resistance (39). The feeding time is also different.

Our findings are consistent with the previous studies that BCAAs facilitate insulin resistance (14). In addition, we noted that the glycogen and gluconeogenesis levels of the HF/paired group were reduced compared with those of the HF group. A previous study showed that BCAAs supplementation suppresses food intake (14). To dissect the interactions between caloric intake and diet composition as it relates to insulin sensitivity, we included the HF/paired-fed group in which the daily food consumption was matched to the HF/BCAA-fed mice. A lower calorie intake improved insulin resistance in the HF/paired group compared with the HF group. BCAAs supplementation seriously impaired hepatic Akt signaling and disrupted hepatic glucose and lipid metabolism.

In summary, our results demonstrate that BCAAs cause hepatic glucose and lipid metabolic disorders, which might inhibit hepatic lipid buffer and storage function and lead to ectopic lipid accumulation in muscle and kidney tissues. Mechanistically, in hepatocytes, BCAAs activate the mTORC1 complex and inhibit mTORC2 signaling. Activation of mTORC1 signaling blocks Akt activation. mTORC2 inhibition induces Akt2 ubiquitination and degradation via promoting the binding of Mul1 and Akt2. These data provide the first identification of BCAAs as critical negative regulators of hepatic Akt signaling by affecting both the activation and the expression of Akt2 in hepatocytes. The inhibitory effects of BCAAs on hepatic Akt signaling led to reduced hepatic lipogenesis and increased hepatic glycogenesis. Because BCAAs are abundant in proteins, our results call for caution regarding the ingestion of high-protein diets in obese individuals and those who have diabetes. BCAAs accumulation might play a pivotal causative role in facilitating systemic glucose and lipid metabolic disorders in T2DM. Targeting BCAA metabolism might help prevent the development of severe lipid and glucose metabolic disorders in patients with diabetes and prevent serious ectopic lipid accumulation and targeted organ injury in patients with severe diabetes.

Acknowledgments. The authors sincerely thank Lang Hu (Department of Cardiology) and Yi Liu (Department of Radiology), both at Xijing Hospital, Air Force Military Medical University, and Pan Chang (The Second Affiliated Hospital of Xi’an Medical University) for their helpful insights, discussions, and technological support.

Funding. This work was financially supported by the Program for National Science Funds of China (grant nos. 81730011, 81600683 and 81800326) and the National Key R&D Program of China (grant no. 2018YFA0107400).

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

Author Contributions. H.Z., F.Z., S.W., and L.T. developed the rationale for the study. H.Z., F.Z., S.W., and L.T. wrote, reviewed, and edited the manuscript. H.Z., D.S., and X.W. conducted the experiments and assisted with the data analysis. X.W., Y. Liu, and W.Y. contributed intellectually. X.Z., J.Z., F.Y., C.H., H.X., C.L., M.F., Y.C., K.L., Y. Li, and L.Z. provided technical support. L.T. supervised the research. H.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 29th Great Wall Congress of Cardiology (GW-ICC), Beijing, China, 11–14 October 2018, and published as Abstract GW29-e0692 in the Journal of the American College of Cardiology, 2018;72(16 Suppl. C):C83–C84.