Insulin resistance and the metabolism of branched-chain amino acids in humans

Peripheral resistance to insulin action is the major mechanism causing the metabolic syndrome and eventually type 2 diabetes mellitus. The metabolic derangement associated with insulin resistance is extensive and not restricted to carbohydrates. The branched-chain amino acids (BCAAs) are particularly responsive to the inhibitory insulin action on amino acid release by skeletal muscle and their metabolism is profoundly altered in conditions featuring insulin resistance, insulin deficiency, or both. Obesity, the metabolic syndrome and diabetes mellitus display a gradual increase in the plasma concentration of BCAAs, from the obesity-related low-grade insulin-resistant state to the severe deficiency of insulin action in diabetes ketoacidosis. Obesity-associated hyperinsulinemia succeeds in maintaining near-normal or slightly elevated plasma concentration of BCAAs, despite the insulin-resistant state. The low circulating levels of insulin and/or the deeper insulin resistance occurring in diabetes mellitus are associated with more marked elevation in the plasma concentration of BCAAs. In diabetes ketoacidosis, the increase in plasma BCAAs is striking, returning to normal when adequate metabolic control is achieved. The metabolism of BCAAs is also disturbed in other situations typically featuring insulin resistance, including kidney and liver dysfunction. However, notwithstanding the insulin-resistant state, the plasma level of BCAAs in these conditions is lower than in healthy subjects, suggesting that these organs are involved in maintaining BCAAs blood concentration. The pathogenesis of the decreased BCAAs plasma level in kidney and liver dysfunction is unclear, but a decreased afflux of these amino acids into the blood stream has been observed.     

Liver BCATm transgenic mouse model reveals the important role of the liver in maintaining BCAA homeostasis

Unlike other amino acids, the branched-chain amino acids (BCAAs) largely bypass first-pass liver degradation due to a lack of hepatocyte expression of the mitochondrial branched-chain aminotransferase (BCATm). This sets up interorgan shuttling of BCAAs and liver–skeletal muscle cooperation in BCAA catabolism. To explore whether complete liver catabolism of BCAAs may impact BCAA shuttling in peripheral tissues, the BCATm gene was stably introduced into mouse liver. Two transgenic mouse lines with low and high hepatocyte expression of the BCATm transgene (LivTg-LE and LivTg-HE) were created and used to measure liver and plasma amino acid concentrations and determine whether the first two BCAA enzymatic steps in liver, skeletal muscle, heart and kidney were impacted. Expression of the hepatic BCATm transgene lowered the concentrations of hepatic BCAAs while enhancing the concentrations of some nonessential amino acids. Extrahepatic BCAA metabolic enzymes and plasma amino acids were largely unaffected, and no growth rate or body composition differences were observed in the transgenic animals as compared to wild-type mice. Feeding the transgenic animals a high-fat diet did not reverse the effect of the BCATm transgene on the hepatic BCAA catabolism, nor did the high-fat diet cause elevation in plasma BCAAs. However, the high-fat-diet-fed BCATm transgenic animals experienced attenuation in the mammalian target of rapamycin (mTOR) pathway in the liver and had impaired blood glucose tolerance. These results suggest that complete liver BCAA metabolism influences the regulation of glucose utilization during diet-induced obesity.     

Epinephrine-induced hyperpolarization of islet cells without KATP channels

ID-1101 (4-hydroxyisoleucine), an amino acid extracted from fenugreek seeds, exhibits an interesting glucose-dependent insulin-stimulating activity. The present study was undertaken to investigate a possible extrapancreatic effect of ID-1101 on insulin signaling and action besides its previously described insulinotropic action. Insulin-sensitizing effects of ID-1101 were investigated in rat in vivo by three different approaches: 1) using euglycemic hyperinsulinemic clamps in two different rat models of insulin resistance, i.e., Zucker fa/fa rats and rats fed a sucrose-lipid diet; 2) measuring liver and muscle phosphatidylinositol (PI) 3-kinase activity after an acute injection of ID-1101 in normal and insulin-resistant diabetic rats; and 3) after chronic treatment in two rat models of insulin resistance. Euglycemic hyperinsulinemic clamp experiments revealed that ID-1101 can improve insulin resistance through an increase of peripheral glucose utilization rate in sucrose-lipid-fed rats and by decreasing hepatic glucose production in Zucker fa/fa rats. Moreover, we demonstrated that a single injection of ID-1101 activates the PI 3-kinase activity in liver and muscle from normal rats but also in muscle from diabetic rats. Finally, chronic ID-1101 treatment significantly reduced insulinemia in type 2 diabetic rats and reduced the progression of hyperinsulinemia in insulin-resistant obese Zucker fa/fa rats. These findings clearly demonstrate that ID-1101 can reduce insulin resistance through activation of the early steps of insulin signaling in peripheral tissues and in liver. In summary, ID-1101, besides its insulinotropic effect, directly improves insulin sensitivity, making it a potentially very valuable therapeutic agent for diabetes treatment.     

Modulation of muscle insulin resistance by selective inhibition of GSK-3 in Zucker diabetic fatty rats

A role for elevated glycogen synthase kinase-3 (GSK-3) activity in the multifactorial etiology of insulin resistance is now emerging. However, the utility of specific GSK-3 inhibition in modulating insulin resistance of skeletal muscle glucose transport is not yet fully understood. Therefore, we assessed the effects of novel, selective organic inhibitors of GSK-3 (CT-98014 and CT-98023) on glucose transport in insulin-resistant muscles of Zucker diabetic fatty (ZDF) rats. Incubation of type IIb epitrochlearis and type I soleus muscles from ZDF rats with CT-98014 increased glycogen synthase activity (49 and 50%, respectively, P < 0.05) but did not alter basal glucose transport (2-deoxyglucose uptake). In contrast, CT-98014 significantly increased the stimulatory effects of both submaximal and maximal insulin concentrations in epitrochlearis (37 and 24%) and soleus (43 and 26%), and these effects were associated with increased cell-surface GLUT4 protein. Lithium enhanced glycogen synthase activity and both basal and insulin-stimulated glucose transport in muscles from ZDF rats. Acute oral administration (2 x 30 mg/kg) of CT-98023 to ZDF rats caused elevations in GSK-3 inhibitor concentrations in plasma and muscle. The glucose and insulin responses during a subsequent oral glucose tolerance test were reduced by 26 and 34%, respectively, in the GSK-3 inhibitor-treated animals. Thirty minutes after the final GSK-3 inhibitor treatment, insulin-stimulated glucose transport was significantly enhanced in epitrochlearis (57%) and soleus (43%). Two hours after the final treatment, insulin-mediated glucose transport was still significantly elevated (26%) only in the soleus. These results indicate that specific inhibition of GSK-3 enhances insulin action on glucose transport in skeletal muscle of the insulin-resistant ZDF rat. This unique approach may hold promise as a pharmacological treatment against insulin resistance of skeletal muscle glucose disposal.     

Hepatic glycogen metabolism in the db/db mouse

Summary Knowledge of the metabolic changes that occur in insulin-resistant type 2 diabetes is relatively lacking compared to insulin-deficient type 1 diabetes. This paper summarizes the importance of the C57BL/KsJ-db/db mouse as a model of type 2 diabetes, and illustrates the effects that insulin-deficient and insulin-resistant states have on hepatic glycogen metabolism. A longitudinal study of db/db mice of ages 2–15 weeks revealed that significant changes in certain parameters of hepatic glycogen metabolism occur during this period. The liver glycogen levels were similar between diabetic and control mice. However, glycogen particles from db/db mice were on average smaller in mass and had shorter exterior and interior chain lengths. Total phosphorylase and phosphorylase a activities were elevated in the genetically diabetic mice. This was primarily due to an increase in the amount of enzymic protein apparently the result of a decreased rate of degradation. It was not possible to find a consistent alteration in glycogen synthase activity in the db/db mice. Glycogen synthase and phosphorylase from diabetic liver revealed some changes in kinetic properties in the form of a decrease in V_max, and altered sensitivity to inhibitors like ATP. The altered glycogen structure in db/db mice may have contributed to changes in the activities and properties of glycogen synthase and phosphorylase. The exact role played by hormones (insulin and glucagon) in these changes is not clear but further studies should reveal their contributions. The db/db mouse provides a good model for type 2 diabetes and for fluctuating insulin and glucagon ratios. Its use should clarify the regulation of hepatic glycogen metabolism and other metabolic processes known to be controlled by these hormones. The other animal models of type 2 diabetes, ob/ob mouse and fatty Zucker (fa/fa) rat, show similar impairment of hepatic glycogen metabolism. The concentrations of glycogen metabolizing enzymes are high and in vitro studies indicate enhanced rate of glycogen synthesis and breakdown. However, streptozotocin-induced diabetic animals and BB rats which resemble insulin-deficient type 1 diabetes are characterized by decreased glycogen turnover as a result of reduction in the levels of glycogen metabolizing enzymes.     

Amino acid metabolism in several tissues of the obese Zucker rat as indicated by the tissue accumulation of α-amino[1-14C]isobutyrate

Measurements of the tissue accumulation of α-amino[1-¹⁴C]isobutyrate [1-¹⁴C]AIB) in lean (+/?) and obese (fa/fa) Zucker rats showed an augmented tissue/plasma ratio in the liver of the obese animals. In contrast, brown adipose tissue AIB accumulation was lower in the fa/fa animals. In response to a 24h starvation period AIB accumulation was significantly elevated in the liver and plasma of the lean animals and was unchanged in the liver of the fa/fa animals. The circulating concentration of alanine and branched-chain amino acids was elevated in the fa/fa animals as compared to their lean counterparts. These observations suggest that amino acid uptake is not involved in the impaired muscle development observed in the obese Zucker rat and that the ability of brown adipose tissue for amino acid utilization is decreased in the obese animals suggesting that this may partially explain the impaired thermoregulatory capacity observed in brown adipose tissue of obese Zucker rats.     

HIV-protease inhibitors induce expression of suppressor of cytokine signaling-1 in insulin-sensitive tissues and promote insulin resistance and type 2 diabetes mellitus

Insulin resistance, hyperglycemia, and type 2 diabetes are among the sequelae of metabolic syndromes that occur in 60-80% of human immunodeficiency virus (HIV)-positive patients treated with HIV-protease inhibitors (PIs). Studies to elucidate the molecular mechanism(s) contributing to these changes, however, have mainly focused on acute, in vitro actions of PIs. Here, we examined the chronic (7 wk) in vivo effects of the PI indinavir (IDV) in male Zucker diabetic fatty (fa/fa) (ZDF) rats. IDV exposure accelerated the diabetic state and dramatically exacerbated hyperglycemia and oral glucose intolerance in the ZDF rats, compared with vehicle-treated ZDF rats. Oligonucleotide gene array analyses revealed upregulation of suppressor of cytokine signaling-1 (SOCS-1) expression in insulin-sensitive tissues of IDV rats. SOCS-1 is a known inducer of insulin resistance and diabetes, and immunoblotting analyses revealed increases in SOCS-1 protein expression in adipose, skeletal muscle, and liver tissues of IDV-administered ZDF rats. This was associated with increases in the upstream regulator TNF-alpha and downstream effector sterol regulatory element-binding protein-1 and a decrease in IRS-2. IDV and other PIs currently in clinical use induced the SOCS-1 signaling cascade also in L6 myotubes and 3T3-L1 adipocytes exposed acutely to PIs under normal culturing conditions and in tissues from Zucker wild-type lean control rats administered PIs for 3 wk, suggesting an effect of these drugs even in the absence of background hyperglycemia/hyperlipidemia. Our findings therefore indicate that induction of the SOCS-1 signaling cascade by PIs could be an important contributing factor in the development of metabolic dysregulation associated with long-term exposures to HIV-PIs.     

Lipid-induced beta-cell dysfunction in vivo in models of progressive beta-cell failure

We determined the effect of 48-h elevation of plasma free fatty acids (FFA) on insulin secretion during hyperglycemic clamps in control female Wistar rats (group a) and in the following female rat models of progressive beta-cell dysfunction: lean Zucker diabetic fatty (ZDF) rats, both wild-type (group b) and heterozygous for the fa mutation in the leptin receptor gene (group c); obese (fa/fa) Zucker rats (nonprediabetic; group d); obese prediabetic (fa/fa) ZDF rats (group e); and obese (fa/fa) diabetic ZDF rats (group f). FFA induced insulin resistance in all groups but increased C-peptide levels (index of absolute insulin secretion) only in obese prediabetic ZDF rats. Insulin secretion corrected for insulin sensitivity using a hyperbolic or power relationship (disposition index or compensation index, respectively, both indexes of beta-cell function) was decreased by FFA. The decrease was greater in normoglycemic heterozygous lean ZDF rats than in Wistar controls. In obese "prediabetic" ZDF rats with mild hyperglycemia, the FFA-induced decrease in beta-cell function was no greater than that in obese Zucker rats. However, in overtly diabetic obese ZDF rats, FFA further impaired beta-cell function. In conclusion, 1) the FFA-induced impairment in beta-cell function is accentuated in the presence of a single copy of a mutated leptin receptor gene, independent of hyperglycemia. 2) In prediabetic ZDF rats with mild hyperglycemia, lipotoxicity is not accentuated, as the beta-cell mounts a partial compensatory response for FFA-induced insulin resistance. 3) This compensation is lost in diabetic rats with more marked hyperglycemia and loss of glucose sensing.     

Comparison of Tissue Metal Concentrations in Zucker Lean, Zucker Obese, and Zucker Diabetic Fatty Rats and the Effects of Chromium Supplementation on Tissue Metal Concentrations

Diabetes results in several metabolic changes, including alterations in the transport, distribution, excretion, and accumulation of metals. While changes have been examined in several rat models of insulin resistance and diabetes, the metal ion concentrations in the tissues of Zucker lean, Zucker obese (an insulin resistance and early stage diabetes model), and Zucker diabetic fatty (ZDF, a type 2 diabetes model) have not previously been examined in detail. The concentration of Cu, Zn, Fe, Mg, and Ca were examined in the liver, kidney, heart and spleen, and Cr concentration in the liver and kidney of these rats were examined. Zucker obese rats have a reduction in the concentration of Cu, Zn, Fe, Mg in the liver compared to ZDF and/or lean Zucker rats, presumably as a result of the increased fat content of the liver of the obese rats. ZDF rats have increased concentrations of kidney Cu compared to the lean rats, while kidney Ca concentrations are increased in the Zucker obese rats. Spleen Fe concentrations are decreased in Zucker obese rats compared to the lean rats. No effects on metal concentrations in the heart were observed between the lean, obese, and ZDF rats, and no effects on Cr concentrations were identified. Cr(III) complexes have previously been shown to have beneficial effects on the signs of insulin resistance in Zucker obese and ZDF rats. The effects of daily gavage administration of chromium picolinate ([Cr(pic)₃]) (1 mg?Cr/kg body mass), CrCl₃ (1 mg?Cr/kg body mass), and Cr3 ([Cr₃O(propionate)₆(H₂O)₃]⁺) (33 μg and 1 mg?Cr/kg body mass) on metal concentrations in these tissues were examined. Treatment with CrCl₃ and Cr3, but not [Cr(pic)₃], at 1 mg?Cr/kg resulted in a statistically significant accumulation of Cr in the kidney of lean and obese but not ZDF rats but resulted in lowering the elevated levels of kidney Cu in ZDF rats, suggesting a beneficial effect on this symptom of type 2 diabetes.     

Free amino acid pools in some tissues of the pregnant rat

Plasma amino acid pools show important variations throughout the gestational period in the rat, with decreased values at mid-pregnancy and recovered levels before parturition. This decrease in amino acid levels at mid-pregnancy is mainly due to changes in the gluconeogenic amino acid group. During pregnancy, whole tissue amino acid pools increase in the liver but no changes appear in other studied tissues. However, individual gluconeogenic amino acid concentrations change significantly in skeletal muscle, skin and liver through the gestational period. Pregnancy may be understood as a challenge to the mother's metabolism but the pattern of tissue amino acid changes is not similar to that found in some stressant situations and it should be the net consequence of maternal adaptations to the increased metabolic needs.     

Amino acid catabolism by perfused rat hindquarter. The metabolic fates of valine.

Hindquarters from starved rats were perfused with plasma concentrations of amino acids, but without other added substrates. Release of amino acids was similar to that previously reported, but, if total amino acid changes were recorded, alanine and glutamine were not formed in excess of their occurrence in muscle proteins. In protein balance (excess insulin) there was no net formation of either alanine or glutamine, even though the branched-chain amino acids and methionine were consumed. If [U-14C]valine was present, radiolabelled 3-hydroxyisobutyrate and, to a lesser extent, 2-oxo-3-methylbutyrate accumulated and radiolabel was incorporated into citrate-cycle intermediates and metabolites closely associated with the citrate cycle (glutamine and glutamate, and, to a smaller extent, lactate and alanine). If a 2-chloro-4-methylvalerate was present to stimulate the branched-chain oxo acid dehydrogenase, flux through this step was accelerated, resulting in increased accumulation of 3-hydroxyisobutyrate, decreased accumulation of 2-oxo-3-methylbutyrate, and markedly increased incorporation of radiolabel (specific and total) into all measured metabolites formed after 3-hydroxyisobutyrate. It is concluded that: amino acid catabolism by skeletal muscle is confined to degradation of the branched-chain amino acids, methionine and those that are interconvertible with the citrate cycle; amino acid catabolism is relatively minor in supplying carbon for net synthesis of alanine and glutamine; and partial degradation products of the branched-chain amino acids are quantitatively significant substrates released from muscle for hepatic gluconeogenesis. For valine, 3-hydroxyisobutyrate appears to be quantitatively the most important intermediate released from muscle. A side path for inter-organ disposition of the branched-chain amino acids is proposed.     

Plasma amino acid imbalance in alcoholic liver cirrhosis

Plasma amino acid concentrations and plasma glucagon and serum insulin levels were studied in male patients with compensated alcoholic and nonalcoholic liver cirrhosis. Age, nutritional status, and liver function tests were similar in both groups; none of the patients presented hepatic encephalopathy. Plasma valine and leucine concentrations were lower, and tyrosine, higher in alcoholic than nonalcoholic liver cirrhosis. As a result, the molar ratios of branched-chain amino acids (BCAA) to aromatic amino acids (AAA) were reduced markedly in this group. Although correlation coefficients comparing BCAA/AAA ratios and KICG in alcoholic and nonalcoholic liver cirrhosis were similar, a steeper regression line was observed in alcoholics. Plasma glucagon and proline levels were significantly higher in alcoholic than nonalcoholic liver cirrhosis, the former correlated with AAA concentrations only in alcoholic liver cirrhosis, but not with BCAA levels. These results indicated that alcoholic liver cirrhosis presented a more deranged plasma amino acid pattern than nonalcoholic, and the amino acid imbalances, except for depressed BCAA and elevated proline, were derived, in part, from the hyperglucagonemia.     

Effects of Insulin on Free Amino Acids in Plasma and the Role of the Amino Acid Metabolism in the Etiology of Diabetic Microangiopathy

To investigate if alterations of the amino acid metabolism may play a more important role in the etiology of diabetic microangiopathy than hitherto recognized, free amino acids in plasma were measured by means of high-performance liquid chromatography (HPLC) in healthy individuals (REF) and patients with insulin-dependent diabetes mellitus (IDDM) and non-insulin-dependent diabetes mellitus (NIDDM). Isoleucine and leucine in IDDM were within normal limits, whereas they were significantly higher in NIDDM (P < 0.01 and P < 0.001, respectively). This was not due to age differences. In order to evaluate the impact of insulin on amino acid metabolism, amino acids were also measured in pregnant women (PREG) undergoing glucose tolerance tests as a screening for pregnancy diabetes and in patients with polycystic ovary syndrome (PCO) undergoing euglycemic insulin clamp tests. Insulin considerably reduced the amino acid concentration. Isoleucine and leucine were particularly depressed. On the whole there was strong covariance between the three branched-chain amino acids, isoleucine, leucine, and valine (P < 0.0001). There was no covariance between amino acid and glucose or HbAlc concentrations, A protein meal strongly stimulated insulin production (+55 mIU/liter), whereas a galactose meal revealed only a minor increase in insulin response (+ 12 mIU/liter) in contrast to a tolerance test with the same amount of glucose (+ 67 mIU/liter). It is concluded that disturbed amino acid metabolism may be a more important causative factor in the etiology of diabetic microangiopathy than hitherto recognized and, in addition, that this may affect the therapeutic approach in both IDDM and NIDDM patients.     

Effects of hypoxia on plasma amino acids of fetal sheep

Summary. Secondary amino acid disturbances from circulatory responses during hypoxia may cause problems in interpreting plasma amino acid profiles of sick babies investigated for possible inherited defects. Systematic studies to characterise them are difficult in man. We investigated the effects of hypoxia on plasma amino acids by studying 9 late gestation fetal sheep in utero during 11 one hour episodes of moderately severe isocapnic hypoxia. In 6 experiments, maternal plasma amino acids were also monitored. Fourteen fetal plasma amino acids increased significantly, with the largest proportionate changes in alanine, valine, leucine, isoleucine, phenylalanine, tyrosine, ornithine and lysine. Maternal amino acids did not increase. Probable explanations were reflex peripheral vasoconstriction in skeletal muscle beds and decreased hepatic blood flow. The findings extend our knowledge of the fetal response to hypoxic stress, demonstrate the importance of skeletal muscle in branched-chain amino acid metabolism, and should help with interpretation of postnatal plasma amino acid disturbances. Received January 29, 1999, Accepted February 22, 1999     

Increased microsomal oxidation of hydroxyl radical scavenging agents and ethanol after chronic consumption of ethanol

The ability of the neonatal rat to oxidize the branched-chain amino acids leucine and valine and their corresponding keto acids was evaluated. In vivo, about 20% of orally administered labeled amino or keto acids were oxidized in 6 h, after which time little further oxidation occurred. In perfused neonatal liver the amino acids were oxidized at only 5-10% the rate of the keto acids. The oxidation of the keto acids showed a saturable dependence on concentration. The decarboxylation of ketoisocaproate (KIC) had a maximal rate of 40.1 +/- 1.6 mumol/h/g liver with an apparent Km of 0.27 +/- 0.03 mM, and decarboxylation of ketoisovalerate (KIV) had a maximal rate of 37.9 +/- 1.9 mumol/h/g liver and an apparent Km of 0.28 +/- 0.04 mM. KIC was ketogenic, producing mainly acetoacetate at a maximal rate of 44.5 +/- 1.6 mumol/h/g liver with an apparent Km of 0.27 +/- 0.03 mM. On the other hand, KIV was not gluconeogenic, although the perfused neonatal liver was able to produce glucose from lactate. During liver perfusion, KIV did not produce measurable quantities of either propionic or beta-aminoisobutyric acids, which are possible end products of KIV metabolism. Decanoic acid inhibited the decarboxylation of both keto acids to the same extent with a maximal effect at 0.4 mM fatty acid. At saturating levels, KIC was less ketogenic than decanoate. Inhibition of endogenous fatty acid oxidation by 2-tetradecylglycidic acid had no effect on keto acid oxidation. These data suggest that branched-chain amino acids derived from milk proteins are probably not quantitatively significant sources of either ketone bodies or glucose in the neonatal rat.     

Oxidation of branched-chain amino acids in skeletal muscle and liver of rat Effects of octanoate and energy state

The effect of octanoate on the oxidative decarboxylation of ¹⁴C-labeled amino acids has been studied in perfused hindquarter and liver of rat. Regulation of the branched-chain α-keto acid dehydrogenase has been further studied with α-[¹⁴C-1]ketoisovalerate in isolated rat muscle and liver mitochondria. (1) Octanoate has a stimulatory effect on the oxidation of branched-chain amino acids in perfused hindquarter. The oxidative decarboxylation of other amino acids are inhibited. Octanoate inhibits the oxidative decarboxylation of all amino acids in perfused liver. (2) The oxidation of valine is stimulated by octanoate and hexanoate also in isolated muscle mitochondria. The stimulatory effect is probably related to activation of the fatty acids since acyl-carnitines inhibit the oxidation. (3) The oxidation of α-ketoisovalerate in mitochondria is inhibited by competing substrates (pyruvate, α-ketoglutarate and succinate). This inhibition is counteracted by octanoate and ADP. (4) Low concentrations (1–5 μM) of 2,4-dinitrophenol (DNP) activates wheras higher concentrations inactivates the branched-chain α-keto acid dehydrogenase in intact but not in solubilized muscle mitochondria. The inactivation is counteracted by ATP, but is increased by octanoate. (5) The observations seem to suggest that the activation (like the inactivation) of branched-chain α-keto acid dehydrogenase in skeletal muscle is dependent on the mitochondrial energy state which therefore may regulate both activation and inactivation of the dehydrogenase.     

Measurement of alpha-keto acids in plasma using an amino acid analyzer

A method for measuring keto acid concentrations in physiological fluids using an amino acid analyzer was developed. After preliminary deproteinization and removal of amino acids, reduction with sodium cyanoborohydride at 105 degrees C resulted in efficient conversion of the keto acids to their corresponding amino acids. In applying the technique to plasma samples, the use of MeOH for deproteinization was necessary to avoid the large losses of keto acids that occurred during precipitation of proteins with perchloric acid. The method was used to follow plasma ketoisocaproate concentrations in rat plasma after administration of leucine, and was sufficiently sensitive to detect concomitant changes in other branched-chain keto acid concentrations.