Endocrinology and Metabolism

Regulation of adipose branched-chain amino acid catabolism enzyme expression and cross-adipose amino acid flux in human obesity

Denise E. Lackey, Christopher J. Lynch, Kristine C. Olson, Rouzbeh Mostaedi, Mohamed Ali, William H. Smith, Fredrik Karpe, Sandy Humphreys, Daniel H. Bedinger, Tamara N. Dunn, Anthony P. Thomas, Pieter J. Oort, Dorothy A. Kieffer, Rajesh Amin, Ahmed Bettaieb, Fawaz G. Haj, Paska Permana, Tracy G. Anthony, Sean H. Adams


Elevated blood branched-chain amino acids (BCAA) are often associated with insulin resistance and type 2 diabetes, which might result from a reduced cellular utilization and/or incomplete BCAA oxidation. White adipose tissue (WAT) has become appreciated as a potential player in whole body BCAA metabolism. We tested if expression of the mitochondrial BCAA oxidation checkpoint, branched-chain α-ketoacid dehydrogenase (BCKD) complex, is reduced in obese WAT and regulated by metabolic signals. WAT BCKD protein (E1α subunit) was significantly reduced by 35–50% in various obesity models (fa/fa rats, db/db mice, diet-induced obese mice), and BCKD component transcripts significantly lower in subcutaneous (SC) adipocytes from obese vs. lean Pima Indians. Treatment of 3T3-L1 adipocytes or mice with peroxisome proliferator-activated receptor-γ agonists increased WAT BCAA catabolism enzyme mRNAs, whereas the nonmetabolizable glucose analog 2-deoxy-d-glucose had the opposite effect. The results support the hypothesis that suboptimal insulin action and/or perturbed metabolic signals in WAT, as would be seen with insulin resistance/type 2 diabetes, could impair WAT BCAA utilization. However, cross-tissue flux studies comparing lean vs. insulin-sensitive or insulin-resistant obese subjects revealed an unexpected negligible uptake of BCAA from human abdominal SC WAT. This suggests that SC WAT may not be an important contributor to blood BCAA phenotypes associated with insulin resistance in the overnight-fasted state. mRNA abundances for BCAA catabolic enzymes were markedly reduced in omental (but not SC) WAT of obese persons with metabolic syndrome compared with weight-matched healthy obese subjects, raising the possibility that visceral WAT contributes to the BCAA metabolic phenotype of metabolically compromised individuals.

  • bariatric
  • diabetes
  • hyperinsulinemia
  • mammalian target of rapamycin
  • protein

in the search for biomarkers that associate with or predict type 2 diabetes mellitus (T2DM), it has become appreciated that circulating concentrations of the branched-chain amino acids (BCAA; valine, leucine, isoleucine) are often increased in obese, insulin-resistant states and in T2DM. Higher fasting plasma BCAA concentrations were initially reported in obese subjects by Adibi and by Felig et al. (2, 12). Recent metabolomic studies found that plasma concentrations of BCAAs and large neutral amino acids are negatively correlated with insulin sensitivity in overweight and obese subjects (24), whereas the principal component that differentiated lean and obese individuals contained BCAA, methionine, phenylalanine, and tyrosine, with a linear relationship between plasma BCAA and homeostasis model assessment of insulin resistance (HOMA-IR) (36). Plasma concentrations of leucine and valine, as well as the leucine metabolite α-ketoisocaproate, were increased in obese female African-American T2DM subjects compared with age- and body mass index (BMI)-matched nondiabetic subjects, and plasma leucine significantly correlated with hemoglobin A1c, a marker that increases with poor blood sugar control (13). In the Framingham Offspring Study (47) and in a Finnish cohort (50), plasma BCAA and aromatic amino acid concentrations were predictive of future T2DM development, and decreased plasma BCAA concentrations correlated with reduced HOMA-IR after weight loss in a nondiabetic overweight and obese population (41). Similarly, gastric bypass surgery decreased plasma BCAA concentrations and associated metabolites (42), and to a greater extent than similar weight loss due to dietary intervention (30). A consistent finding across multiple investigations is that the methionine derivative α-hydroxybutyrate (α-HB, also termed 2-HB) and/or its precursor α-ketobutyrate (α-KB) are increased in blood of prediabetic, obese, or T2DM individuals (13, 20, 32, 39), and α-KB catabolism appears to be driven in part by the mitochondrial branched-chain α-ketoacid dehydrogenase (BCKD) complex (1).

It remains to be established whether obesity- or diabetes-associated perturbations in amino acid homeostasis arise from changes in cellular amino acid uptake from the blood, differences in protein turnover, and/or incomplete amino acid oxidative catabolism in one or more tissues. Regarding the latter, a logical site to study potential regulation is the mitochondrial BCKD complex. In most peripheral tissues, the mitochondrial isoform of branched-chain amino transferase (BCATm, encoded by BCAT2) catalyzes the first and reversible step in BCAA catabolism wherein they are deaminated to branched-chain α-ketoacids (BCKAs), leucine to α-ketoisocaproate, valine to α-ketoisovalerate, and isoleucine to α-keto-β-methylvalerate. In the next highly regulated and irreversible step, BCKAs are oxidized by the BCKD complex. This complex contains the components BCKD E1α, BCKD E1β, dihydrolipoamide branched-chain transacylase (DBT, E2 subunit), and dihydrolipoamide dehydrogenase (DLD, E3 subunit). The BCKD complex shares some components and is structurally related to pyruvate dehydrogenase (PDH). It also shares regulatory mechanisms with PDH, including inhibition through phosphorylation, increased NADH/NAD+, and possibly acetyl-CoA; the latter two factors are anticipated to be increased in the fasted or diabetic states, in tissues robustly oxidizing fatty acids (1). Interestingly, the gene encoding BCKD E1α, BCKDHA, was identified as a candidate gene associated with obesity traits and T2DM based on six of seven discovery methods (45). Less is known about the regulation of the FAD- and NAD-dependent catabolic enzymes downstream of BCKD, but conceivably these steps may also be modulated by mitochondrial redox status. Highlighting the importance of BCKD complex activity on circulating BCAA levels, knockout of BCKD kinase (a negative regulator of BCKD activity) led to significantly reduced blood BCAA (25), and knockout of PP2Cm phosphatase, an enzyme thought to dephosphorylate and thus activate BCKD, has the opposite effect with impaired BCKA clearance (33).

Whether BCKD and other BCAA catabolic enzymes play a role in obesity- or insulin resistance-associated blood metabolite phenotypes remains to be established, and very little is known regarding tissue-specific regulators of BCKD expression, e.g., in white adipose tissue (WAT). Based on whole tissue estimates of BCKD complex activity, WAT leucine kinetics, and BCAA oxidation rates in tissue preparations, WAT could be an important contributor to regulation of whole body BCAA homeostasis (7, 37a, 38, 43). This is supported by a recent study demonstrating that gene expression of BCKD enzyme components was higher in perigonadal WAT compared with gastrocnemius skeletal muscle, and valine oxidation rate was higher in perigonadal WAT compared with gastrocnemius muscle explants (22). In that same study, adipose transplantation experiments demonstrated the potential importance of WAT in modulating systemic BCAA concentrations.

Most studies relating to BCKD and other pathway enzymes in WAT have been limited to genetic models of obesity. For instance, ob/ob mice and fa/fa Zucker rats had decreased WAT BCKD and BCATm protein levels in the fed state and decreased BCKD in the fasted state compared with their lean counterparts (42), suggesting that factors associated with obesity, diabetes, and/or impaired leptin signaling downregulate BCAA catabolism in adipose tissue. Little is known regarding the metabolic or endocrine factors that regulate gene and protein expression of BCAA-relevant enzymes in WAT. Expression of 18 genes related to BCAA catabolism were upregulated in WAT by administration of peroxisome proliferator-activated receptor (PPAR)-γ agonists in obese Zucker rats (23), and there was a positive correlation between glucose disposal rate and BCAA catabolism-related transcript levels in WAT from human subjects before and after treatment with thiazolidinediones (TZDs) (40). Whether BCAA-associated enzymes are direct targets of PPARγ or if TZD-related differences result from improved insulin or glucose homeostasis is not known.

Considering the paucity of information related to regulation of BCAA metabolism in WAT, we sought to answer the following questions. 1) Does polygenic diet-induced rodent obesity result in reduced BCAA catabolic enzyme gene expression in WAT, similar to prior observations in genetic models of obesity? 2) Are obesity-associated reductions in WAT BCAA enzyme gene expression also manifested in human obesity, and is this due to perturbations in metabolic health or obesity per se? And, following from this, 3) do maneuvers that improve adipose glucose utilization and insulin action (e.g., PPARγ agonism, reduced PTP1B activity) enhance BCKD complex and associated enzyme expression, and would impairment of glucose utilization and promotion of metabolic stress (2-deoxyglucose treatment) have the opposite effect? These studies extend prior observations through examination of expression of BCKD and a suite of BCAA-associated enzymes in lean and obese human adipocytes, and in subcutaneous (SC) and omental WAT depots from obese persons in poor metabolic health vs. so-called healthy obese. In complementary physiological studies, BCAA, BCKA, and other amino acid fluxes were also measured across human SC WAT to begin to understand how obesity and insulin resistance functionally impact net utilization of BCAA in adipose. The results support the concept that the BCAA catabolic pathway in adipose tissue is sensitive to changes in insulin action and metabolic signals and raise the possibility that visceral WAT depots are particularly important in this regard.



Murine 3T3-L1 fibroblasts and Dulbecco's modified Eagle medium with 4 mM l-glutamine were purchased from ATCC (Manassas, VA). Penicillin-streptomycin, Hank's balanced salt solution, and Ambion RiboPure RNA extraction kits were from Life Technologies (Carlsbad, CA). Recombinant human insulin, biotin, 3-isobutyl-1-methylxanthine, dexamethasone, 2-deoxy-d-glucose (2-DG), glacial acetic acid, Tris·HCl, EDTA, NaCl, and Triton X-100 were all purchased from Sigma-Aldrich (St. Louis, MO). FBS was from Atlas Biologicals (Fort Collins, CO). Calf serum and M-PER Mammalian Protein Extraction Reagent were from Thermo Fisher Scientific (Waltham, MA). Cell culture plates were purchased from BD Falcon (Franklin Lakes, NJ). Type 1 collagen from rat tail was from BD Biosciences (San Jose, CA).

Animal Studies

All animal protocols were approved by the University of California at Davis or the Auburn University [rosiglitazone (RSG) studies] Institutional Animal Care and Use Committee according to Animal Welfare Act guidelines.

Diet-induced obese mice.

The conditions for this study and the animal metabolic phenotypes have been described in detail elsewhere (44). Briefly, 4-wk-old male C57BL/6J mice were purchased from the Jackson Laboratory, individually housed under standard temperature (20–22°C) and light-dark cycle (12 h:12 h) conditions, and after a 1-wk acclimation period were randomly assigned to purified experimental diets containing 10, 45, or 60% energy from fat for 12 wk. At week 12, mice were briefly food deprived (between 3 and 8 h starting at 0600) before being deeply anesthetized via isoflurane inhalation (3% in O2). Retroperitoneal (RP) fat pads, epididymal fat pads, femoral SC fat pads, and liver were excised, weighed, and snap-frozen in liquid nitrogen. All tissues were stored at −80°C. A subset of samples was used for Western blot analysis (n = 6/group).

Zucker rats.

Ten-week-old inbred female obese Zucker rats (fa/fa; n = 10) and nonobese littermates (Fa/Fa; n = 10) were obtained from the breeding colony maintained at the University of California, Davis (courtesy of Dr. Judith Stern). Rats were housed under standard temperature (20–22°C) and light-dark cycle (14:10-h) conditions with free access to water and food (Formulab Diet 5008; Purina Mills) and then fed a purified control diet as previously described (35) for 2 wk before terminal blood and tissue collection. After overnight fast, rats were deeply anesthetized via isoflurane inhalation and killed by cervical dislocation before collection of RP WAT, liver, and hindlimb muscle (primarily biceps femoris) that was flash-frozen and stored at −80°C.

RSG-treated mice.

Male C57BL/6J and db/db mice were purchased at 9 wk old (Jackson Laboratories) and allowed to acclimate for 5 days before treatment. Mice were housed under standard temperature (20–22°C) and light-dark cycle (12:12-h) conditions with free access to water and standard chow (2018 Teklad Global 18% Protein Rodent Diet). Mice were treated with 10 mg/kg body wt rosiglitazone acetate by oral gavage daily for 14 days (n = 5/group). Animals were deeply anesthetized with 100 mg/kg ketamine hydrochloride and 10 mg/kg xylazine hydrochloride (ip), and adipose was collected and flash-frozen following decapitation.

PTP1B−/− mice.

PTP1B-floxed (PTP1Bfl/fl) mice were generated previously (6). Adiponectin (Adipoq)-Cre mice were obtained from Dr. E. Rosen (BIDMC/Harvard University). PTP1Bfl/fl were on a mixed 129Sv/J × C57Bl/6J background, and Adipoq-Cre mice were on a mixed FVB × C57Bl/6J background. Genotyping for the PTP1B floxed allele and presence of Cre was performed by PCR, using DNA extracted from tails (6). Mice were maintained on a 12:12-h light-dark cycle with free access to water and food (Purina Laboratory Chow no. 5001) at weaning and maintained on that diet for 26 wk. For tissue collection, fl/fl control and adipose-specific PTP1B−/− mice were fasted overnight, injected with saline (2/group) or 10 mU/g body wt insulin (3/group; for measurement of insulin-associated tissue phosphorylation targets, data not shown), and killed by cervical dislocation, and adipose depots were dissected. The acute insulin treatment did not impact BCKD protein expression, so data were combined within each genotype for statistical comparison (n = 5/genotype).

Human Studies

Studies were approved by the Tribal Council of the Gila River Indian Community and by the National Institute of Diabetes and Digestive and Kidney Diseases Institutional Review Board (Pima Indian studies), the University of California, Davis, Institutional Review Board (bariatric studies), or the Milton Keynes Research Ethics Committee (United Kingdom). All subjects provided written informed consent.

Pima Indian adipocytes.

Transcript expression levels of BCKD complex members were compared between isolated SC adipocytes derived from nonobese, nondiabetic and obese, nondiabetic adult Pima Indians derived from studies by Lee et al. (31), as archived in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) gene chip database (GEO accession no. GSE2508).

Healthy and metabolically compromised obese human subjects.

Women between 20 and 55 years of age and BMI 35–55 kg/m2 undergoing bariatric surgery at the University of California Davis Medical Center were recruited. Subjects with fasting glucose <5.55 mM (100 mg/dl), total insulin <19 μU/ml, triglycerides <150 mg/dl, high-density lipoprotein (HDL) >39 mg/dl, systolic blood pressure <140 mmHg, and diastolic blood pressure <90 mmHg and who were not taking medication for T2DM or dyslipidemia were considered metabolically healthy and free of metabolic syndrome (n = 10). Subjects with fasting glucose >5.55 mM or total insulin >19 μU/ml as well as dyslipidemia (triglyceride ≥150 mg/dl or HDL ≤39 mg/dl) or hypertension (systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg) who were not previously taking medication for T2DM or dyslipidemia were assigned to the metabolic syndrome group. Just before surgery, subjects underwent a 2- to 3-day clear liquid diet (gastric banding or gastric sleeve patients) or bowel prep (Roux-en-Y patients) with GoLYTELY (polyethylene glycol-electrolyte solution; Braintree Laboratories, Braintree, MA) followed by 1 g neomycin treatment. Induction of general anesthesia was accomplished with propofol (1.0–2.5 mg/kg) with/without the adjunct of lidocaine (0.2–1.5 mg/kg) and paralytic agents such as rocuronium (0.6 mg/kg), vecuronium (0.05–0.1 mg/kg), or succinylcholine (0.3–1.1 mg/kg). Maintenance anesthesia was sustained with desflurane (2.5–8.5%) or sevoflurane (0.5–3%). Omental and SC WAT surgical biopsies were flash-frozen immediately and stored at −80°C.

Human arteriovenous BCAA and BCKA analysis.

Amino acid and BCKA profiles were analyzed in the Lynch laboratory using arterial and venous plasma derived from lean insulin-sensitive, obese insulin-sensitive, and obese insulin-resistant subjects (Karpe laboratory). Human SC venous drainage was accessed by selective venous catheterization of a branch of the superficial epigastric as described previously (14). Arterialized venous blood was sampled by retrograde cannulation of a vein draining the hand, which was kept in a box with air temperature of 55–60°C. Following an overnight fast, and after a period of 30–60 min rest, heparinized blood samples were drawn simultaneously from the arterial or arterialized site and from the adipose venous catheter. Adipose tissue blood flow (ATBF) was monitored during the period of blood sampling by the washout of 133Xe (14). Blood samples were kept on ice and rapidly centrifuged at 4°C. Aliquots were stored at −80°C until analyzed.

For plasma BCAA analysis, a previously described method was used involving precolumn derivatization using Phenomenex EZ-fast reagent (48) that uses a solid-phase sorbent tip to separate derivatized “free” amino acids from proteins and does not involve an acid extraction step, employing multiple internal standards and an external standard curve for every amino acid. The extracted derivatized amino acids and standards were separated and analyzed by UPLC-MS using a Waters Synapt HDMS hybrid QTOF with Ion Mobility housed in the Pennsylvania State College of Medicine Macromolecular Core Facility. Data analysis was completed using the Waters Mass Lynx software. For plasma BCKA, acidified extracts were analyzed by HPLC following precolumn O-phenylenediamine (OPD) derivatization of acid-deproteinized samples as described (26). The OPD-derivatized standards or samples with internal standard (ketocaproic acid) were injected on a Shimadzu high-pressure liquid chromatograph (HPLC), separated by HPLC fitted with a fluorescence detector. The ketoacid peaks were quantified using LC Solutions software (Shimadzu, Columbia, MD).

3T3-L1 adipogenesis studies.

Changes in mRNA expression levels of BCAA catabolic enzymes during the course of fat cell differentiation and maturation, and the effects of treatment with the non-TZD PPARγ agonist GW-1929, were tested in samples archived from a 3T3-L1 adipogenesis model experiment described previously (37). In that study, GW-1929 (1 μg/ml, 0.1% by volume) was added for 18 h at different time points during the adipocyte maturation process or treated with vehicle (dimethyl sulfoxide; 0.1% by volume). To measure effects of PPARγ antagonism on BCAA-relevant enzyme gene expression, new experiments were conducted in mature (14 days postdifferentiation) 3T3-L1 adipocytes that were pretreated with vehicle or 10 μM PPARγ antagonist T0070907 (Cayman Chemicals) for 90 min and then treated with antagonist and varying doses of RSG for 24 h before harvesting.

Acute insulin treatment experiments and glucose utilization inhibitor studies were conducted using fully mature 3T3-L1 adipocytes (13–15 days postdifferentiation) treated with vehicle, insulin, or inhibitors as described for 48 h before harvesting for mRNA and protein preparation. 2-DG was used at 3 mM in the maintenance cell culture media, with untreated media serving as a vehicle control. Adipocyte studies were conducted using at least two independent cultures.

Gene Expression Analysis

mRNA preparation and transcript abundance methods were carried out as described previously (37, 44). The quantitative real-time PCR assays utilized gene-specific TaqMan primers and FAM-MGB-labeled probes (TaqMan Gene Expression Assays; Applied Biosystems, Foster City, CA) and were analyzed in triplicate for each sample using an ABI 7900HT instrument. Reactions were carried out in a 384-well format containing the following in each well: cDNA corresponding to 20 ng of original total RNA, 1× specific primer probe mix, and 1× Master Mix (ABI Taqman Gene Expression Master Mix); cDNA was air-dried in each well before adding qPCR reagents to facilitate an 8 μl/well assay. Cycle conditions were 50°C for 2 min, 95°C for 10 min, and then 40 cycles of 95°C for 15 s/60°C for 1 min. Amplification cycle threshold number (Ct) of loading control mRNA (18S) for each sample was determined using Pre-Developed TaqMan Assay Reagent Eukaryotic 18S rRNA primer/probe (Hs99999909_m1) to correct for template-loading differences across all target genes (ΔCt = target gene Ct − reference gene Ct). Primer/probe Applied Biosystems identifiers for gene expression studies were Bckdha (Mm00476112_m1), Bckdhb (Mm01177077_m1), Dld (Mm00432831_m1), Dbt (Mm00501651_m1), Bcat1 (Mm00500289_m1), Bcat2 (Mm00802192_m1), Bckdk (Mm00437777_m1), Ppm1k (Mm00615792_m1), BCKDHA (Hs00958109_m1), BCKDHB (Hs00609053), BCAT1 (Hs00194075_m1), BCAT2 (Hs00154171_m1), and BCKDK (Hs00195380_m1).

Western Blots

Flash-frozen liver, WAT, and skeletal muscle were homogenized with mammalian protein extraction reagent (M-PER, 100 mg tissue/200 μl M-PER) containing 1× HALT protease and phosphatase inhibitors (Thermo Fisher SC). 3T3-L1 adipocytes were lysed with M-PER containing 1× HALT protease and phosphatase inhibitors and sonicated for 10 s. Lysates were centrifuged at 10,000 g for 10 min to clear. Protein concentrations were quantified using the bicinchoninic protein assay (Thermo Fisher Scientific). Two micrograms of total liver lysate and 10 μg of total WAT or 3T3-L1 cell lysate were separated on a 10% Bis-Tris gel using 1× MOPS running buffer (Invitrogen). The proteins were transferred to a polyvinylidene membrane (Bio-Rad) and immunoblotted with rabbit anti-BCKD E1α (DE1α) or anti-pS293-BCKD E1α polyclonal antibody (1:5,000 dilution) generated by the laboratory of She et al. (42), in 0.1% Tween 20 in PBS (PBS-T) supplemented with 3% BSA for 1 h at room temperature. Specific signal was detected with a horseradish peroxidase-conjugated secondary antibody (1:10,000 goat anti-rabbit horseradish peroxidase diluted in 3% BSA in PBS-T) using an Immun-Star WesternC Kit (Bio-Rad Laboratories, Hercules, CA). Blots were imaged using a ChemiDoc XRS+ Imaging System (Bio-Rad). β-Actin (1:2,000 anti-mouse; Sigma) or GAPDH (1:1,000 anti-mouse) was used as a loading control.

Statistical Analysis

Data were analyzed using Prism software version 6 (GraphPad Software, San Diego, CA). One-way and two-way ANOVA followed by Tukey's multiple-comparison posttest were used for studies involving multiple comparisons, unless otherwise specified, and Student's t-test was used for comparisons between only two groups. P < 0.05 was considered significant. Means ± SE are presented.


Confirmation That Obesity in Rodents is Associated With Decreased Total BCKD (E1α) Protein Abundance in WAT

We have extended results from previous studies that reported reduced BCKD expression in the WAT of male genetically obese rats and mice (42). Obese female fa/fa Zucker rats showed significant reduction in RP WAT total BCKD protein abundance (Fig. 1, A–C). Notably, there was reduced phosphorylated BCKD (pBCKD) and no significant difference in the pBCKD-to-total BCKD ratio compared with lean Fa/Fa littermates (Fig. 1C). There was no significant difference in total BCKD protein abundance or phosphorylation status in the hindlimb skeletal muscle or liver between the two groups (data not shown).

Fig. 1.

Branched-chain α-ketoacid dehydrogenase (BCKD) (E1α) protein abundance and phosphorylation (p) in retroperitoneal (RP) white adipose tissue (WAT) of female Zucker lean (Fa/Fa, n = 10) and obese (fa/fa, n = 9) rats (A–C), male C57BL/6J mice fed diets varying in fat energy for 12 wk (D–F), and in lean male C57BL/6J and obese db/db mice (G–I). Effects of treatment with rosiglitizone (RSG) are also shown for lean vs. db/db mice. BCKD protein abundance was normalized to β-actin or GAPDH as shown. Phosphorylated BCKD was normalized to total BCKD. All values are expressed as fold relative to the lean group. Ctl, control. Values are means ± SE; means without a common letter differ significantly, P < 0.001. **P < 0.01, ***P < 0.001, and ****P < 0.0001, statistically different from lean by Student's t-test.

WAT BCKD protein was also determined in polygenic rodent obesity models. As for obese Zucker rats, diet-induced obese (DIO) mice fed 45 or 60% kcal as fat had significantly reduced BCKD and pBCKD protein in RP WAT compared with 10% fat-fed lean mice, whereas there was a modest increase in the pBCKD-to-total BCKD ratio (Fig. 1, D–F). Reduced BCKD expression with 45 and 60% fat diets was also apparent in SC WAT (50.6 ± 0.13 and 35.0 ± 0.12% of control expression, respectively) and epididymal (50.0 ± 0.09 and 15.2 ± 0.03% of control expression, respectively) WAT depots. High-fat diet caused a 43% increase in liver total pBCKD protein only in 60% fat diet-fed mice (P < 0.05 vs. 10 and 45% fat groups), with no change in the pBCKC-to-total BCKD ratio (data not shown).

Evidence that PPARγ Activation and Adipocyte Glucose Homeostasis Influence BCAA Catabolic Enzyme Expression in Adipose

To begin to differentiate the influence of obesity from that of insulin resistance on BCKD complex and BCAA catabolic enzymes in adipose, models of altered WAT insulin sensitivity were examined. First, PPARγ agonist treatment was tested in lean C57BL/6J and obese db/db mice; the latter displayed reduced RP WAT BCKD protein (Fig. 1, G–I) as seen with obese Zucker rats and DIO mice. The TZD treatment led to a significant 38% reduction in fasting plasma glucose in the db/db mice (205 ± 18 to 128 ± 4 mg/dl, P < 0.01), with a smaller 20% effect in lean mice (144 ± 9 vs. 116 ± 14 mg/dl, not significant). Previously, it was reported that in vivo treatment of rats and humans with PPARγ agonists increases mRNA expression of BCKD complex and BCAA enzymes in WAT (23, 40). Consistent with this, treatment with 10 mg/kg 12-day oral gavage of the PPARγ agonist RSG significantly increased RP WAT BCKD protein abundance in lean mice, but this effect was not as robust in db/db mice (Fig. 1, G–I).

Potential cell-autonomous effects of PPARγ agonists have not been determined previously. Activation of PPARγ may directly induce BCKD complex genes, and/or these effects may require metabolic events downstream from PPARγ. To address these questions, we first took advantage of archived samples of 3T3-L1 preadipocytes and adipocytes treated with or without the specific and potent PPARγ agonist GW-1929 acutely (18 h) at different stages of adipocyte maturation (37). Expression of BCKD components Bckdha (E1α component), Bckdhb (E1β component), Dbt (E2 component), and Dld (E3 component) followed a similar pattern of expression, and all peaked midway through maturation, with expression slowly decreasing from the peak level by day 14 (Fig. 2). With respect to other enzymes in the BCAA catabolic pathway, mRNA expression of mitochondrial Bcat2 followed a similar pattern as that of the BCKD subunits; cytosolic Bcat1 decreased markedly throughout differentiation, whereas expression of the BCKD kinase Bckdk and the phosphatase Ppm1k did not change substantially during 3T3-L1 differentiation (Fig. 3). Expression of Bckdhb, Dbt, and Dld BCKD subunit mRNAs was significantly increased by treatment of adipocytes with the PPARγ agonist GW-1929, whereas Bckdha transcript remained unaffected (Fig. 2). This suggests that effects on the Bckdha gene may require longer-term activation of PPARγ or that additional factors not present in cell culture trigger TZD-induced gene and protein expression in vivo. Transcripts for Bcat2 and Ppm1k were significantly increased by GW-1929 treatment, with a significant effect of treatment (Fig. 3) and time × treatment interaction. However, expression of Bcat1 and Bckdk mRNAs was not significantly altered by acute GW-1929 treatment (Fig. 3).

Fig. 2.

mRNA expression profile of BCKD complex components during 3T3-L1 preadipocyte-to-adipocyte differentiation with or without acute (18 h) 10 μg/ml GW-1929 peroxisome proliferator-activated receptor (PPAR)-γ agonist treatment. Transcript levels for each gene are expressed relative to the mean expression for that gene on day 14 postdifferentiation and expressed as the means ± SE of n = 3 independent samples/day. *P < 0.05, **P < 0.01, and #P < 0.0001 indicates significantly different from control at the same time point by 2-way ANOVA followed by Bonferonni's post hoc test. All mRNA expression differs significantly by the time factor; treatment factor differs significantly in B–D, and there is a significant time × treatment interaction in B–D by 2-way ANOVA.

Fig. 3.

mRNA expression profile of branched-chain amino acid (BCAA)-associated genes during 3T3-L1 preadipocyte-to-adipocyte differentiation with or without acute (18 h) 10 μg/ml GW-1929 PPARγ agonist treatment. Transcript levels for each gene are expressed relative to the mean expression for that gene on day 14 postdifferentiation and expressed as means ± SE of n = 3 independent samples/day. #P < 0.0001 indicates significantly different from control at the same time point by 2-way ANOVA followed by Bonferonni's post hoc test. All mRNA expression differs significantly by the time factor.

New studies were conducted to test the specificity of PPARγ agonist effects on BCKD complex gene expression using 24 h RSG treatment (the same PPARγ agonist as that used in our mouse in vivo studies described above) with or without the PPARγ antagonist T0070907 in mature 3T3-L1 adipocytes. Consistent with results using the nonthiazolidinedione agonist GW-1929, Bckdhb, Dbt, and Dld transcript abundances were all significantly increased by RSG treatment, and PPARγ antagonist treatment blunted this response, whereas Bckdha was unaltered (Fig. 4). Increasing concentrations of RSG did not significantly change Bcat1, Bcat2, or Bckdk expression, and Ppm1k transcript abundance was increased only at the 100 nM agonist concentration (data not shown).

Fig. 4.

mRNA expression profile of BCKD complex components in mature 3T3-L1 adipocytes treated with rosiglitazone in the presence or absence of 10 μM PPARγ antagonist T0070907 for 24 h. Transcript levels for each gene are expressed relative to the mean expression for that gene treated without rosiglitazone or antagonist and expressed as means ± SE of n = 15 (untreated) or n = 8 (all other treatments) combined from 2 independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001, significantly different from unstimulated condition.

Because PPARγ activation increases insulin sensitivity, it is possible that improved insulin action and/or glucose utilization underlies PPARγ agonist effects on BCAA catabolism pathway gene expression and protein abundance in adipocytes. We took advantage of archived SC WAT samples derived from chow-fed adipose-specific PTP1B−/− mice. PTP1B is a physiological regulator of insulin signaling, glucose homeostasis, and adiposity (11), and adipose-specific PTP1B−/− mice exhibit increased insulin signaling in adipose tissue compared with control (fl/fl) mice (Bettaieb and Haj, unpublished observations). Total BCKD protein abundance in SC WAT was increased significantly in adipose-specific PTP1B−/− mice compared with fl/fl control mice (3.1 ± 0.7-fold of control expression; P < 0.05), whereas abundance was not changed in the liver (data not shown). The ratio of phosphorylated to total BCKD was not different between genotypes in either tissue (data not shown). These results support a role for insulin signaling, adipocyte glucose utilization, or other metabolic events associated with loss of PTP1B activity in maintaining or increasing expression of BCAA metabolic enzymes in WAT.

We next treated mature 3T3-L1 adipocytes with the glycolysis inhibitor 2-DG in the presence or absence of insulin for 48 h (timeframe shown in insulin treatment pilot studies to increase Bckdhb). 2-DG treatment significantly decreased Bckdha, Bckdhb, Dbt, Dld, Bcat2, and Ppm1k expression while markedly increasing Bcat1 expression regardless of insulin (Fig. 5). Only Bcat1 expression showed a significant insulin-by-2-DG treatment interaction. 2-DG treatment did not change Bckdk expression. These results indicate that altered glucose metabolism or metabolic stress associated with 2-DG can affect gene expression of BCAA metabolic enzymes in adipocytes. The effect of acute insulin on gene transcripts under these conditions was modest, with an increase in Bcdkhb and a reduction in Bcat1 levels.

Fig. 5.

mRNA expression profile of BCKD complex components (A–D) and associated genes (E–H) in mature 3T3-L1 adipocytes treated with the glucose oxidation inhibitor 2-deoxy-d-glucose (2-DG) in the presence or absence of insulin for 48 h. Transcript levels for each gene are expressed relative to the mean expression for that gene treated without insulin or 2-DG and expressed as means ± SE of n = 8 combined from 2 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 indicates significantly different from control within insulin treatment status. mRNA expression differs significantly in B–E by insulin treatment. There is a significant 2-DG treatment × insulin treatment interaction in E by 2-way ANOVA.

Effect of Obesity and Metabolic Disease on BCKD and BCAA Catabolic Enzyme Gene Expression in Human Adipose

To determine if observations in rodent obesity models apply to humans, BCAA pathway transcript levels were assessed using data from gene chip analysis of isolated SC adipocytes as described by Lee et al. for lean and obese Pima Indians (31). The SC adipocyte BCKDHA gene expression was significantly decreased in obese subjects (Fig. 6A), a pattern also observed for BCKDHB and DLD (Fig. 6, B–C). Adipocyte mRNA expression levels of DBT, BCAT2, BCKDK, and PPM1K were not significantly changed by obesity status (data not shown).

Fig. 6.

BCKDHA (A), BCKDHB (B), and DLD (C) mRNA abundances in subcutaneous adipocytes isolated from nonobese and obese male and female Pima Indians (n = 10/group). Values are means ± SE. *P < 0.05, **P < 0.01, and ****P < 0.0001 by Student's t-test.

The basis for obesity-associated differences in expression of BCKD complex members and other BCAA-associated enzymes remains to be determined. Because obesity is often accompanied by metabolic dysfunction and adipocyte cell stress, it raises the possibility that factors beyond adiposity per se, e.g., metabolic, inflammation, or endocrine signals, drive WAT BCAA enzyme expression patterns. To evaluate this hypothesis, BCAA-associated enzyme gene expression in equally obese female bariatric surgery patients with or without metabolic syndrome were compared. The clinical characteristics of this cohort are provided in Table 1. In healthy obese women, omental WAT mRNA transcript abundances for BCAA catabolic enzymes BCKDA, BCKDB, BCAT2, and BCKDK were substantially higher compared with metabolic syndrome subjects (Fig. 7, A–D; DBT and DLD mRNAs did not differ, data not shown). Interestingly, there were no differences in BCAA catabolic enzyme mRNA abundances in SC WAT, and in healthy women expression levels of BCAA enzyme mRNAs were higher in omental than SC WAT. These results indicate that, in the context of obesity in women, phenomena associated with metabolic syndrome downregulate WAT BCAA pathway enzyme expression in visceral WAT.

View this table:
Table 1.

Characteristics of subjects used for obese healthy and metabolic syndrome WAT BCAA catabolic enzyme gene expression

Fig. 7.

BCKD complex components (A and B) and BCAA-associated (C and D) gene transcript abundances are altered in omental adipose tissue (but not subcutaneous WAT) of obese adult female subjects with metabolic syndrome relative to healthy weight-matched obese women. All values are expressed relative to healthy subject omental adipose tissue values. Values are means ± SE. *P < 0.05 and **P < 0.01, metabolic syndrome obese group differs statistically from healthy obese group by Student's t-test. There was a significant effect of WAT depot in BCKDHA (P < 0.05) and BCKDHB (P < 0.01) expression (2-way ANOVA).

Functional Assessment of BCAA Utilization in Subcutaneous WAT in Humans

Consistently observed reductions in WAT BCKD protein in obesity, and the apparent upregulation or maintenance of adipocyte BCKD components under conditions of increased insulin action and optimal glucose utilization, led to the hypothesis that insulin resistance impairs efficient BCAA catabolism in adipose. In proof-of-principle studies, arterial and venous BCAA and BCKA concentrations in lean insulin-sensitive, obese insulin-sensitive, and obese insulin-resistant subjects were measured to compare net SC WAT metabolite uptake and efflux (subject characteristics are provided in Table 2). As expected, venous nonesterified fatty acid and glycerol concentrations were significantly higher than arterial concentrations in all groups (P < 0.0001) and strongly correlated with one another (R2 = 0.74 and 0.91 for arterial and venous concentrations, respectively, P < 0.001). Consistent with previous reports (13, 20, 32, 36, 39), venous BCAA concentrations were highest in obese insulin-resistant subjects compared with either lean or obese insulin-sensitive subjects (Table 3). Arterial concentrations of BCAAs were also highest in obese insulin-resistant subjects. Generally speaking, obese insulin-sensitive subjects had concentrations intermediate between lean insulin-sensitive and obese insulin-resistant subjects. As shown in Table 3, most often values did not statistically differ in the obese groups using a conservative Tukey's multiple-comparison test that protects against type 1 error but can result in type 2 error. To account for this possibility, a post hoc Newman-Keuls multiple-comparisons test was used, and by this measure the obese insulin-resistant subjects' arterial valine and total arterial and venous total BCAA levels were significantly increased vs. the other two groups. Unexpectedly, most individuals displayed only negligible SC WAT uptake of BCAA, with arteriovenous differences essentially zero (Table 3). Results were highly variable, and neither the arteriovenous change in BCAA concentrations nor the uptake into the adipose differed significantly between the groups. For BCKAs (the immediate downstream metabolites of BCAA catabolism), there was evidence for net uptake, although minimal, into SC WAT in obese insulin-resistant subjects (Table 3).

View this table:
Table 2.

Characteristics of subjects used for cross-SC WAT BCAA and BCKA arteriovenous concentration analysis

View this table:
Table 3.

Cross-SC WAT BCAA and BCKA dynamics in adult humans

WAT arteriovenous differences in other amino acids across treatment groups are provided in Supplemental Table 1 (Supplemental data for this article may be found on the American Journal of Physiology: Endocrinology and Metabolism website.). Some notable observations include a robust uptake of glutamate by adipose tissue (and higher in obese), output of glutamine (only evident in obese insulin-resistant subjects), and indications that the insulin-resistant obese group had the highest absolute blood phenylalanine, tyrosine, methionine, and cystine concentrations.


Plasma BCAA concentrations typically increase in the context of obesity, insulin resistance, and T2DM and tend to decrease after weight loss in adult humans. The etiology of this phenomenon has not been elucidated, but it is possible that changes in adipocyte BCAA metabolism are involved since WAT might contribute to whole body BCAA homeostasis (7, 22, 38, 43). Considering the close association between systemic BCAA and insulin resistance phenotypes (reviewed in Ref. 1), maintenance of proper insulin action and metabolic health may be important to support efficient BCAA metabolism in adipose and other tissues. These questions prompted the current studies examining the regulation of BCKD complex components and BCAA-associated enzyme expression in human and rodent obesity and in cultured adipocytes.

Our results in monogenic and polygenic rodent models of obesity confirm previous studies showing decreased BCKD protein abundance in adipose of obese and insulin-resistant rats and mice (23, 40, 42). Both fa/fa obese Zucker rats and obese db/db mice showed decreased RP WAT BCKD protein abundance relative to their lean counterparts, and a moderate obesity-inducing diet (45% kcal from fat) and a more extreme diet (60% kcal from fat) decreased RP WAT BCKD protein abundance in DIO mice. We have extended these findings to humans, showing decreased BCKDHA, BCKDHB, and DLD transcript expression in SC adipocytes isolated from obese insulin-resistant compared with more insulin-sensitive lean male and female Pima Indians, and lower mRNA expression of some BCAA enzymes in omental WAT from persons with metabolic syndrome. Should protein patterns track with enzyme transcripts, the results indicate that obesity and metabolic dysfunction reduce the WAT BCAA catabolic machinery, at least in some human WAT depots.

It is possible that adiposity signals per se downregulate WAT BCAA enzyme expression. However, increased systemic BCAAs and α-hydroxybutyrate (upstream of BCKD complex) track degree of insulin resistance, worsening blood sugar control, and can serve as predictive tools for T2D in humans (13, 20, 36, 47), suggesting that metabolite and/or endocrine factors associated with metabolic homeostasis are more important than obesity in regulation of WAT BCAA pathway enzyme expression. Because metabolic dysfunction is often associated with excess adiposity, it could explain obesity-related reductions in BCAA enzyme expression. In support of this hypothesis, reduced levels of transcripts encoding mitochondrial BCAA catabolic enzymes BCKDHA, BCKDHB, and BCAT2 were observed in omental WAT from morbidly obese women with metabolic syndrome compared with metabolically healthy but equally obese counterparts. The findings indicate a role for impaired insulin action or other aspects of metabolic dysfunction in decreasing BCAA catabolic enzyme mRNA expression. Interestingly, differences in expression of these genes between the healthy vs. unhealthy obese groups were not apparent in SC WAT, and transcript levels were higher in omental vs. SC WAT, suggesting WAT depot/site-specific regulation of BCAA metabolism and a potentially important role for visceral WAT in modulating systemic BCAA. Because SC and omental WAT from lean healthy individuals were not available to compare with the obese subjects, it is not yet known if the BCAA enzyme expression levels in healthy obese subjects differ from normal-weight healthy individuals. Additional studies are warranted to examine this question and to determine depot-specific enzyme protein and enzyme activity levels, as well as site-dependent BCAA catabolism.

The results herein and reports that blood BCAA concentrations associate with insulin resistance and diabetes phenotypes suggest a role for insulin action, and/or a healthy metabolic or inflammatory phenotype, modulate adipocyte BCAA enzyme expression. We found, for instance, that, in both C57BL/6 and db/db mice, TZD treatment increased RP WAT BCKD protein abundance (albeit not robustly in db/db mice). Agonism of PPARγ in 3T3-L1 adipocytes increased transcript abundances of Bckdhb, Dld, Dbt, and Bcat2 during the course of 3T3-L1 adipocyte differentiation. These results are similar to those of Hsiao et al. who showed that adipose tissue from rats treated with PPARγ ligand had increased Bckdhb, Dbt, and Dld mRNA expression (23). The results are also reminiscent of the increase in WAT BCAA catabolic pathway transcripts after TZD treatment in humans (40). BCKD complex genes increased over the course of 3T3-L1 differentiation. That is consistent with previously published biochemical findings that BCKD complex activity and expression increase significantly during the differentiation of 3T3-L1 cells from fibroblasts to adipocytes (9, 16, 27). This pattern highlights again the potential importance of BCAA metabolism in mature adipocytes.

Earlier studies have shown that stimulation of rat adipose tissue explants with either leucine or insulin, together or alone, decreases the phosphorylation status and/or activity of BCKD (18, 19, 21, 34). We observed upregulation of BCAA-related pathways in chow-fed mice with adipose-specific knockout of PTP1B potentially contributing to their enhanced insulin signaling (and presumably, increased glucose uptake) in adipose; in these animals, there was significantly greater total BCKD protein abundance in SC WAT but no significant change in protein expression in the liver. These results support the hypothesis that suboptimal insulin action or glucose utilization plays a role in the reduced BCKD expression observed in obese rodent and human WAT. Consistent with this perspective, increased glucose disposal rate was related to increased expression of BCAT2, BCKDHA, BCKDHB, DBT, and DLD in human WAT before or after TZD treatment (40). An inhibitor of glucose oxidation (2-DG) diminished Bckdha, Bckdhb, Dbt, Dld, Bcat2, and Ppm1k transcript expression in the presence and absence of insulin in mature 3T3-L1 adipocytes. The effects were most apparent in the basal (no insulin) context, possibly because of insulin's effect to strongly activate glucose uptake, which may have blunted the effects of 2-DG. These results indicate that glucose oxidation could play an important role in promoting mitochondrial BCAA catabolism enzyme expression in adipocytes. Alternatively, because 2-DG elicits multiple effects in cells, including oxidative stress, inhibition of mammalian target of rapamycin, and reduced ATP levels (51), metabolic regulators beyond glucose utilization cannot be excluded as contributing to adipocyte BCAA enzyme regulation. In a murine model of transgenic overexpression of the glucose transporter GLUT-4 in adipocytes, expression of some BCAA catabolism enzymes was found to be reduced (22). Studies by Costeas and Chinsky in rat H4IIEC3 hepatoma cells indicated depression of E1α subunit mRNA and protein by insulin treatment (8, 10). Whether these contrasting results reflect specific nuances to the chronic GLUT-4 overexpression model or cell-type-specific effects remains to be determined, and the specific associations between insulin and glucose homeostasis and BCAA catabolism remain to be clarified. Because we observed only modest effects of acute insulin on expression of BCAA enzymes in cultured adipocytes compared with effects of PPARγ agonist or 2-DG treatments, it is speculated that primary regulation of expression derives from cellular metabolic signals and not a direct action of insulin on the genes per se.

Despite the growing body of evidence showing reduced expression of BCKD complex and other BCAA catabolism factors in obesity and insulin-resistant states, it has not been established if said changes equate to lower WAT BCAA utilization and catabolism. If so, this may explain, in part, the higher blood concentrations of BCAA in these states. In novel proof-of-principle studies using plasma samples from arteriovenous catheters across human SC WAT, we tested these concepts by determining cross-WAT BCAA, BCKA, and amino acid balance in lean and obese insulin-sensitive individuals vs. obese insulin-resistant subjects. With respect to non-BCAA amino acid patterns, we observed that human SC WAT is an avid consumer of glutamate and, in the obese insulin-resistant cohort, a potential net producer of glutamine. This is consistent with reports using microdialysis in rat inguinal WAT (28, 29) or using the human cross-SC WAT paradigm to measure these specific amino acids (15, 37a). The fate of SC WAT glutamate is not known, but, as an anaplerotic carbon source for the tricarboxylic acid cycle, it may serve as an adipocyte lipogenic substrate, as has been observed for liver (5).

As with other reports (13, 20, 32, 36, 39), both arterial and venous BCAA levels were elevated in obese insulin-resistant individuals compared with obese or lean insulin-sensitive subjects. Contrary to our hypothesis, net BCAA utilization and BCKA arteriovenous differences across SC WAT were negligible in absolute terms and did not differ much across groups. The results are qualitatively similar to those reported in a comparison of SC-WAT in lean and obese women (37a). Several possibilities could explain this outcome. First, the impact of obesity and insulin resistance on BCAA/BCKA flux may be very modest in human abdominal SC WAT depots in the overnight-fasted state. Furthermore, the variance stemming from person-to-person differences and measurement of very small cross-tissue metabolite fluxes make detection of group differences challenging. Second, it is possible that SC WAT BCKD is largely inactive in the fasted state compared with the fed/postprandial state; several reports showed that rodent WAT BCKD enzyme is phosphorylated and/or largely inactive under fasting conditions but can be dephosphorylated and activated in the fed state or when WAT is treated with insulin, leucine, and/or glucose (17, 19, 21, 27, 34, 46). Consistent with this, across our rodent models, phosphorylated WAT BCKD protein essentially tracked total BCKD levels. If this were also true in the overnight-fasted human subjects, it may help explain the minimal observed SC WAT utilization of BCAA. There is a markedly reduced postmeal SC ATBF in obesity (3, 4) that could also influence how SC WAT affects systemic BCAA, at least in the fed state. Clearly, the effect of fasting/feeding-associated factors on cross-adipose amino acid flux and BCKD phosphorylation in humans warrants further study. Third, it is possible that BCAA/BCKA dynamics across visceral WAT depots may be more important than SC WAT. We found that metabolic syndrome was associated with a marked reduction in omental WAT BCAA catabolic enzyme transcript abundances compared with metabolically healthy, equally obese individuals. Yet, no effect was observed in SC WAT, and BCAA enzyme transcript levels were much higher in omental WAT. Consistent with a potential visceral (central) WAT contribution to systemic BCAA homeostasis, Würtz and colleagues reported that increases in blood BCAA associated with insulin resistance were only seen in women with larger waist circumferences who by definition have greater central adiposity (49).

In conclusion, the results illustrate that metabolically compromised obesity is characterized by reduced BCAA catabolic enzyme expression in rodent and human WAT. PPARγ treatments and maneuvers designed to improve glucose utilization in adipocytes indicated that maintenance of a healthy metabolic profile supports higher expression of BCAA catabolic enzymes in adipocytes and WAT. Poor metabolic health in obese humans was characterized by substantially lower omental WAT expression of BCAA catabolic genes compared with equally obese healthy subjects. Thus, it is concluded that obesity, per se, is less important than metabolic, inflammatory, or endocrine regulators of WAT BCAA machinery expression in humans, and these effects on BCAA enzymes appear to be most prominant in visceral WAT.


This research was supported by the United States Department of Agriculture (USDA)-Agricultural Research Service (ARS) Intramural Project 5306-51530-019-00, a USDA-ARS Headquarters Postdoctoral Award (D. E. Lackey), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Grant R01-DK-078328 (S. H. Adams), and the National Dairy Council (grant administered by the Dairy Research Institute; S. H. Adams). Work in F. G. Haj's laboratory was supported by NIDDK Grant R01-DK-090492.


The USDA is an Equal Opportunity employer and provider. No conflicts of interest, financial or otherwise, are declared by the authors.


Author contributions: D.E.L., C.J.L., K.C.O., R.M., M.A., F.K., R.H.A., F.G.H., T.G.A., and S.H.A. conception and design of research; D.E.L., C.J.L., K.C.O., R.M., M.A., W.H.S., F.K., S.M.H., D.H.B., T.N.D., A.P.T., P.J.O., D.A.K., R.H.A., A.B., F.G.H., P.A.P., and S.H.A. performed experiments; D.E.L., C.J.L., K.C.O., F.K., S.M.H., D.H.B., T.N.D., P.J.O., P.A.P., and S.H.A. analyzed data; D.E.L., C.J.L., K.C.O., M.A., F.K., S.M.H., F.G.H., and S.H.A. interpreted results of experiments; D.E.L. and S.H.A. prepared figures; D.E.L. and S.H.A. drafted manuscript; D.E.L., C.J.L., K.C.O., R.M., M.A., W.H.S., F.K., S.M.H., D.H.B., T.N.D., A.P.T., R.H.A., A.B., F.G.H., P.A.P., T.G.A., and S.H.A. edited and revised manuscript; D.E.L., C.J.L., K.C.O., R.M., M.A., W.H.S., F.K., S.M.H., D.H.B., T.N.D., A.P.T., P.J.O., D.A.K., R.H.A., A.B., F.G.H., P.A.P., T.G.A., and S.H.A. approved final version of manuscript.


We thank Dr. Yoshiharu Shimomura for helpful discussions of BCAA metabolism and enzymes.

Current address for D. E. Lackey: Department of Medicine, University of California, San Diego, CA (e-mail: dlackey{at}ucsd.edu).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 37a.
  39. 38.
  40. 39.
  41. 40.
  42. 41.
  43. 42.
  44. 43.
  45. 44.
  46. 45.
  47. 46.
  48. 47.
  49. 48.
  50. 49.
  51. 50.
  52. 51.
View Abstract