Branched chain amino acids (BCAAs: leucine, valine, isoleucine) have been under close scrutiny since the beginning of the 1970s because of their unique role in endogenous metabolism and their critical regulatory properties on protein metabolism. A notable property of this metabolism is its compartmentalization; in humans, at mealtime, while more than 50% of ingested amino acids are taken up by the liver, BCAAs essentially escape from the splanchnic territory and directly contribute to the postprandial activation of peripheral/muscle protein synthesis. It was initially thought that the primary role of these amino acids was restricted to the anabolic or anticatabolic properties of leucine. Indeed, this rapidly led, in the clinical setting, to the development of BCAA-enriched nutritional formulas ^[1] for the dietary management of cirrhotic patients (and hypothetically to treat hepatic encephalopathy, a neurological disorder due to a severe chronic liver disease) ^[1] or 2) for the stimulation of anabolism or slowdown of catabolism in injured patients ^[2]. Already at that time, there were some hints of possible interactions between BCAAs and insulin signaling. However, it was only around the years 2005–2010 that several studies suggested, on the one hand, a possible association between type 2 diabetes (T2D) and increased plasma BCAA levels and, on the other hand, a possible causal link between BCAAs and insulin resistance.

2. BCAA Metabolism in Insulin Resistance Situations

It should first be noted that studies on the effects of insulin resistance on BCAA metabolism can sometimes seem inconsistent. It is therefore important to consider the specificities of this metabolism in humans and in rodents, but also the conditions under which the studies were carried out, in the fasted or fed state. Furthermore, although it is well established that there can be significant discrepancies between gene or protein expression, enzymatic activities, and substrate fluxes in vivo, it seems that this is particularly true for the metabolism of AACR [66]^[3].

2.1. BCAA Metabolism and Insulin Resistance in Humans

It has been known for a long time that the peripheral utilization of amino acids in the postprandial period is closely related to the action of insulin. Already in the 1970s, Wahren et al. [5]^[4] showed, in insulin-dependent diabetic patients, in the absence of insulin, both a higher plasma concentration of BCAAs in the basal state and a greater increase in these concentrations in response to a protein meal. The measurement of leg postprandial arteriovenous amino acid differences showed that the postprandial increase in BCAA uptake was much more limited than in control subjects. In a study in 26 subjects, either overweight with impaired glucose tolerance (fasting glucose ≥ 5.6 mmol/L, 2 h glucose ≥ 7.8 mmol/L, or HOMA-IR > 2.0) or normal weight with normal glucose tolerance, mRNA-sequencing pathway analysis on subcutaneous white adipose tissue (WAT) and skeletal muscle biopsies indicated decreased post-prandial BCAA metabolism in both tissues, improved by a 12-week supervised exercise intervention [73]^[5]. In contrast, after an overnight fast, measurement of leucine turnover with 1-[¹³C]leucine or U[¹³C₆]leucine showed a 9 to 50% increase in leucine flux and a 20 to 50% increase in leucine oxidation in subjects of increasing insulin resistance (HOMA-IR: 2.57 to 4.89) and normal to morbidly increased weight (BMI: 24.4 to 38.5) [74,75,76]^[6][7][8]. This suggests that, over the course of the day, the increase in leucine oxidation during fasting fails to compensate for the defect in postprandial utilization, which could lead to increased fasting plasma levels. Note that, in the Glynn et al. study [74]^[6], neither plasma BCAAs nor leucine flux were affected after a 6-month training program including aerobic and whole-body resistance training, despite a significant improvement in insulin sensitivity (mean 54% increase in glucose infusion rate during a hyperinsulinemic-euglycemic clamp). Similarly, plasma BCAAs and related metabolites were not affected by a 12-week supervised intensive exercise intervention [73]^[5]. On the other hand, when leucine oxidation was measured during a 2-step hyperinsulinemic-euglycemic clamp in obese subjects, while it increased as a function of insulin level in all subjects, it was lower in DT2 patients compared to non-DT2 ones [77]^[9]. In the liver, the decrease in hepatic BCAA oxidation is such that it is exploited in a breath test using ¹³C-KIC to evaluate the defect in hepatocyte mitochondrial function [78]^[10]. Indeed, Grenier-Larouche et al. [79]^[11] showed that the association between BCKA and nonalcoholic fatty liver disease (NAFLD), the hepatic manifestation of insulin resistance and metabolic symdrome, is the result of the defect in hepatic BCKA metabolism. In WAT, studies showed either preserved or decreased expression of enzymes involved in BCAA catabolism in subcutaneous or visceral WAT in obese subjects compared to normal weight individuals or in insulin-resistant obese compared to non-insulin-resistant patients [52,74]^[6][12]. Bariatric surgery was associated with increased BCKD expression. It can be noted, however, that measurements of BCAAs and BCKA fluxes from arteriovenous differences in subcutaneous adipose tissues in humans do not show any difference with obesity and/or insulin resistance [78]^[10]; but, this is not surprising given the small amplitude of the variations and the inter-individual variability.

2.2. BCAA Metabolism in Experimental Models of Insulin Resistance

The defect in postprandial BCAA utilization is confirmed experimentally in models of insulin resistance with a more marked increase in post-prandial plasma BCAAs compared to controls [61]^[13]. Note that, in one study [52]^[12], the expression of the enzymes of BCAA catabolism in the fed state was not affected in the skeletal muscle but decreased in subcutaneous WAT; oxidation, only measured in WAT, was also decreased. In the fasted state, experimental studies do not demonstrate any changes in protein expression of BCATm, BCKD E1α subunit, BDK, and BCKD E1 phosphorylation ratio [61^[13][14],80], or BCKD activity [65]^[15] at the muscle level in models of genetic obesity (ob/ob mice, zucker rats) [61,80]^[13][14] or diet-induced (HF or HFHS) obesity [65,66]^[3][15]. In the liver, experimental data are discordant with, depending on the studies and models, normal or decreased expression of BCKD E1 subunit, normal or increased expression of BDK, and normal or increased BCKD E1 phosphorylation ratio [61,65,78]^[10][13][15]. However, both She et al. [61]^[13] and White et al. [81]^[16] observed a reduction in valine oxidation (using α-keto-[1-¹⁴C] isovalerate) in liver sample from ob/ob mice or zucker fatty rats, respectively. In WAT, studies are rather concordant. They globally observe a reduction in BCAA metabolism, whether in terms of enzyme expression or oxidation activity in subcutaneous or visceral adipose tissue in experimental models of genetic or diet-induced obesity [61,65,80,81]^{[13][14][15][16]}. Interestingly, treatment with thiazolidinediones in db/db mice was associated with an increase in protein expression of BCKD, suggesting a normalization of this metabolism. Taken together, based on the data of whole-body leucine turnover and of organ BCAA metabolism (enzyme activities or expression), it seems that, in fasting insulin-resistant individuals, BCAA metabolism is characterized by (i) increased muscle oxidation and (ii) decreased oxidation in adipose tissue and perhaps in the liver. This is further illustrated by the study by Neinast et al. [82]^[17] extrapolating the contribution of different organs to BCAA metabolism from the distribution of ¹³C after administration of ¹³C-labeled BCAAs in mice. These authors show in models of insulin resistance in fasting animals, a redirection of the oxidation of BCAAs to the benefit of the muscles at the expense of WAT, and to a lesser extent of the liver.

2.3. BCAA Metabolism in Brown/Beige Adipose Tissue

An interesting point highlighted by the above-mentioned work of Neinast et al. [82]^[17] as well as by that of Yoneshiro et al. [83]^[18] is the contribution of the brown/beige adipose tissue (BAT) in the oxidation of BCAAs. Long considered to be limited to hibernating mammals, various studies over the last two decades have highlighted the presence of a significant number of thermogenic adipocytes (brown or beige adipocytes) in humans and their potential contribution to energy metabolism. Indeed, Ouellet et al. [84]^[19] showed in healthy subjects exposed to cold, in conditions of limited shivering, an increase of up to 80% of resting energy expenditure, largely attributable to the brown/beige adipose tissue. If, until recently, it had been shown that this activation of BAT was associated with a significant oxidation of glucose and fatty acids, the work of Yoneshiro et al. [83]^[18] highlights an increased use also of BCAAs by the BAT in these conditions. In particular, these authors demonstrated an inverse relationship between BAT activity (assessed by ¹⁸F-Fluorodeoxyglucose positron emission tomography imaging) and plasma leucine and valine concentrations in healthy individuals. They further showed in mice the importance of BAT in the oxidation of BCAAs for thermogenesis and the maintenance of plasma concentrations of BCAAs. Finally, in a model of BCKD invalidation specifically in BAT, they showed that the defect in BCAA oxidation of the BAT was associated with weight gain and fat mass gain, increase in hepatic triglyceride content, and insulin resistance. Conversely, DIO in mice was associated with a decrease in BCAA catabolism in BAT [85]^[20]. Note that, in DIO mice, pharmacological stimulation of BCAA oxidation by 3,6-dichlorobenzo(b)thiophene-2-carboxylic acid (BT2) or by oral administration of some specific probiotics (Bacteroides dorei and Bacteroides vulgatus) improved BCAA catabolism by the BAT and mice metabolic status [85]^[20].

3. Do BCAAs Promote Insulin Resistance?

As already mentioned, leucine, through its interaction with Sestrin 2, promotes the activation of mTORC1, leading in particular to the activating phosphorylation of p70SK1 and inhibiting one of 4E-BP1, thus participating in the postprandial induction of protein synthesis. In essence, this system involves other regulatory elements or systems, for example, the availability of different substrates (e.g., via the level of amino acylation of tRNAs and General control nonderepressible 2 (GCN2)) or the level of restoration of protein pool. In terms of insulin signaling, through mTORC1 and p70S6K1, leucine induces a feedback inhibition via serine phosphorylation of IRS1. However, whether this feedback control causes insulin resistance is controversial. Alternatively, it has been proposed that it is not the BCAAs themselves but their metabolites that impair insulin sensitivity.

3.1. BCAA-Induced Insulin Resistance?

While some experimental works seemed to support an effector role of BCAAs in insulin resistance, the results are quite inconsistent. In the above-mentioned Newgard et al. study [55]^[21], Wistar rats were fed for 13 weeks a conventional (C) or a HF diet supplemented (150% higher content) or not with BCAAs; BCAA supplementation lead to a doubling of their plasma concentrations. HF/BCAA-fed rats had lower food intake and lower weight gain than HF fed rats but had similarly impaired glucose tolerance, while C/BCAA-fed animals exhibited weight gain and glucose tolerance similar to those of C-fed rats. The authors acknowledged the responsibility of the BCAA/HF-diet combination. These results contrast with, for example, those of Zhang et al. [86]^[22] in mice receiving, for 15 weeks, a doubled daily supply of leucine with either a control or a HF diet. They showed that, in HF-fed animals, a leucine-induced increase in energy expenditure was responsible for a reduction in weight gain and adiposity and was associated with improved glucose tolerance and lower insulin resistance. Indeed, leucine prevented the increase in blood glucose observed from the 4th week on the HF diet and improved insulin sensitivity throughout the study. In a more recent work [65]^[15], mice received an equivalent supplement of the 3 BCAAs (but half the amount of leucine) with an HF (12 weeks) or an HFHS (32 weeks) diet. Under these conditions, the BCAA supply did not affect glucose tolerance. A potential confounding factor could be the role of BCAAs in the regulation of food intake. In the Solon-Biet et al. study cited above [67]^[23], in which mice were fed a standard diet with variable BCAA intake, a double intake of BCAAs was associated with hyperphagia; this hyperphagia was related to reduced central serotonin synthesis and corrected by tryptophan supplementation. In addition, some of the differences between these studies, apart from the animal species, can be accounted for by different nutritional regimen with for example, according to the studies, quantitatively (40, 45, or 60% of the energy intake) and qualitatively (lard, milk fat, soybean oil) different lipid supplies or the supply of BCAAs individually or in mixture, in addition to the protein supply or in substitution of a part of the protein supply [55,67,86]^[21][22][23]. In humans, previous studies showed a decrease in glucose utilization during amino acid infusion. Hyperinsulinemic euglycemic clamp tests have since been used to try to clarify these observations. For example, Boden and Tappy [87]^[24] investigated glucose metabolism in healthy volunteers receiving an infusion of a parenteral amino acid formula, leading to a 5- to 6-fold increase in their plasma concentrations during hyperinsulinemic euglycemic clamps associated with somatostatin and glucagon (0.25 and 3 g/kg/min). No changes in glucose metabolism were observed. On the contrary, under similar conditions, Krebs et al. [43]^[25] showed a 25% reduction in whole body glucose disposal. In fact, at least in short-term experiments, the responsibility of the global protein intake rather than the specifically of BCAAs seems likely. Indeed, both Smith et al. [88]^[26] and Harris et al. [89]^[27] have compared in sedentary obese women during hyperinsulinemic euglycemic clamps the effect of a protein supply or of an isomolar amount of leucine. Ingestion of proteins, but not leucine, significantly decreased insulin-stimulated glucose disposal. However, the previous studies were based on short-term evaluation. In longer-term studies, researchers also find this difference between the effect of the global protein intake and the effect of BCAAs. The type of proteins administered is involved; indeed, Pal et al. [90]^[28] compared the effects of a 12-week supplementation with 27 g of casein, whey (rich in BCAAs), or glucose (control) in overweight patients and observed an improvement of the insulin sensitivity with the whey supplementation. More specifically, essential amino acids do not seem to affect insulin sensitivity by themselves [91]^[29]. Finally, Woo et al. [68]^[30] compared in a crossover study the effects of a 4-week supplementation of 20 g of BCAAs or rice protein (low in BCAAs) in 12 obese prediabetic subjects and showed a trend towards improved glucose utilization with BCAAs. Contrary to a possible BCAA-induced insulin resistance, several studies have demonstrated or suggested, particularly in the short term, an anabolic effect of BCAAs and more particularly of leucine. This notion, which has made them successful among sport enthusiasts in search of muscle gain, is based on the demonstration of an acute but transient activation of muscle protein synthesis. In the clinical setting, past studies have evaluated the possible benefit of BCAA supplementation in injured or cancer patients with, however, limited success, probably owing to methodological problems such as very inadequate parenteral nutrition mixtures not providing all the essential amino acids [2,22]^[2][31]. Since then, various studies have focused on the geriatric population with the aim of preventing or treating age-associated loss in muscle mass and function, i.e., sarcopenia. If short-term studies are globally positive, studies on prolonged periods are rather disappointing [28]^[32]. Interestingly, in healthy older adults (69.1 ± 1.1 yr) in a situation of prolonged bed rest (7 days of bed rest and 5 days of rehabilitation), leucine supplementation tended to preserve insulin sensitivity and muscle mitochondrial metabolism [92]^[33]. It should be noted that, because the various strategies taken individually have only limited effectiveness, a multimodal approach of sarcopenia is currently favored by combining leucine with another nutrient, such as vitamin D, or with resistance training exercise. For example, it has been shown in healthy elderly subjects that the intake of leucine and vitamin D for 6 weeks was associated with a significant increase in muscle protein synthesis and a gain in muscle mass [93]^[34]. The studies that evaluated insulin sensitivity in this context showed an improvement; however, this could have been expected given the improvement in body composition and in muscle mass, an important determinant of glucose disposal; therefore, it is difficult to relate it to a direct effect of leucine [94,95]^[35][36]. If BCAAs are responsible for insulin resistance, conversely, a reduction in plasma BCAA concentrations could be associated with an improvement in insulin sensitivity. It should be noted that the studies on this subject were carried out in the context of the evaluation of the effects of caloric restriction and, more specifically, protein restriction, with very significantly reduced protein intake, lower than 50% of standard levels. It is clearly established, mainly on the basis of experimental work, that protein restriction improves metabolic health and longevity. Part of the effects seems related, at the endocrine level, to the decrease in the secretion of IGF-1 and the increase in the secretion of FGF-21 [96]^[37], a fasting and metabolic stress-induced peptide hormone that regulates energy substrates utilization [97]^[38]. At the cellular level, this involves the response of the mTOR and GCN2 pathways to the decrease in the global or specific availability of certain amino acids [96]^[37]. In Solon-Biet et al.’s study [67]^[23], median lifespan was similar between control and BCAA-restricted (at 20 or 50% level of standard supply) mice, but the 20% group had lower hepatic steatosis and lower basal insulin level with normal basal glucose. Moreover, Yu et al. [98]^[39] showed that a 2/3 reduction in protein supply or specifically in BCAA supply was associated with an improvement in glucose tolerance and a slowing down of weight gain after 3 weeks. Surprisingly, only an equivalent reduction specifically in isoleucine supply reproduced similar effects, while leucine restriction had no effect. The authors showed that the effects of isoleucine restriction were independent of the activation of the mTORC1 and GCN2 pathways in the liver and were associated with increased FGF-21 secretion, activation of ketogenesis, and increased energy expenditure. This was related to a specific, isoleucine deprivation-associated increase in nuclear FOXA2 (Forkhead box protein A2), a transcription factor that activates lipid metabolism and ketogenesis and whose nuclear localization is blocked by insulin [99]^[40]. The effect of BCAA restriction has been reproduced in T2D patients receiving a one-week supply of proteins and amino acids (1 g/kg body weight) restricted (−60%) or not in BCAAs [100]^[41]; BCAA restriction was associated with an increase in the plasma concentration of FGF-21 and in the “oral glucose sensitivity index”, calculated during a “mixed meal tolerance test”. In summary, the results of the restriction studies show a specific response to BCAA deficiency that illustrates the importance of these amino acids in both protein synthesis and ketogenesis and the necessary adaptation of the metabolism for a better management of energy sources. Observations in excess situations are very contrasted beyond the simple postprandial regulation of substrate excess; the multiple confounding factors (animal species, nature of the associated diet, activity level of the animals, etc.) suggest that the direct role of BCAAs in insulin resistance is quite limited.

3.2. BCAA Metabolites and Insulin Sensitivity

From metabolomic or epidemiological studies, some authors have hypothesized that it was probably not BCAAs themselves but their metabolites that could contribute to insulin resistance. Indeed, She et al. [101]^[42] observed that global invalidation of BCATm, which induces a blockade of BCAA degradation and an increase in their plasma concentrations, was associated with a reduced weight and fat mass gain, improved insulin sensitivity, and resistance to the obesogenic effect of a HF diet. In the same way, telmisartan, an angiotensin II receptor blocker with unique BCATm inhibitory properties, favors adipose tissue browning and improves body weight, glucose tolerance, and insulin sensitivity in HF-fed mice [102]^[43]. Of course, these data must be interpreted with caution as BCAAs/BCKAs are significant contributors to energy metabolism in situations where glucose/glycogen availability is limited and significant regulators of protein homeostasis. The blockade of their degradation must therefore be compensated by metabolic adaptation, as already mentioned in situations of severe protein or BCAA restriction. Nonetheless, the metabolism of BCAAs could be necessary to allow their negative effects on energy metabolism. Indeed, some metabolomic and epidemiological studies showed associations between plasma concentrations of certain metabolites and insulin resistance or the risk of developing insulin resistance. The presence of some of these metabolites, mainly BCKA, C5 and C3 acylcarnitines, and HIB, exacerbated in experimental models during concomitant administration of BCAAs and HF diet, has been interpreted as an indication of incomplete oxidation of BCAAs, and it has been suggested that these metabolites are potentially responsible for insulin resistance. Note that, as in the case of fatty acid oxidation and the appearance of even acyl-carnitines [103]^[44], increased BCAA derivative levels, particularly odd acyl-carnitines, may not be indicative of incomplete oxidation but may simply indicate an increase in metabolic flux. This is well illustrated by the study of energy substrate utilization during aerobic exercise as a function of glycogen availability. Under conditions of reduced glycogen availability, the muscle increases the oxidation of lipid substrates (with no change in exogenous carbohydrate utilization) but also of BCAAs from protein catabolism [104,105,106]^[45][46][47]. Metabolomic analysis showed an increase in fatty acid metabolites such as acyl-carnitine proportionately to the increase in lipid oxidation; similarly, the increase in BCAA metabolites was proportional to the increase in leucine oxidation [105]^[46]. In insulin-resistant subjects, BCAAs are metabolized more slowly in the postprandial period, but their oxidation increases in the post-absorptive period. Therefore, it does not seem surprising that the plasma concentrations of the reactive intermediates increase during these same periods. The negative effect of BCKAs on insulin function is suggested, for example, by the improvement of glucose tolerance and insulin sensitivity in ob/ob or DIO mice induced by increasing BCKD activity following BDK inhibition by BT2 [107]^[48]. In vitro work in C2C12 or L6 muscle cells gives some indication of a potential direct role of BCKAs in insulin resistance. In these cell models, BCKAs decrease Akt phosphorylation and thus insulin signaling and decrease glucose utilization [66,108]^[3][49]. More precisely, Biswas et al. [66]^[3] showed inhibition of insulin induced activating tyrosine phosphorylation of IRS1 (at very high KIC concentration) but upregulated protein translation signaling and synthesis. This is similar to the previously mentioned effect of leucine on the Akt-mTORC1 pathway, but this conclusion remains controversial. In addition, it has been shown that valsartan, an angiotensin II receptor blocker that displays BDK inhibitory properties [109]^[50], elicited only limited effects on insulin sensitivity in human. Indeed, the Navigator study on 9306 patients with impaired glucose tolerance and cardiovascular disease showed very limited effect of valsartan on glucose tolerance [110]^[51]. In fact, another mechanism may explain some of the experimental effects of BDK inhibitors. Indeed, the effect of BDK inhibition by BT2 is already observed after one week of treatment in Zucker rats and is associated with a shift in respiratory exchange ratio, suggesting decreased fatty acid storage and increased oxidation [111]^[52]. White et al. [111]^[52] showed that BDK and PPM1K also control ATP-citrate lyase phosphorylation, and thus de novo lipogenesis and fat storage. Interestingly, in a mouse model of nonalcoholic steatohepatitis (NASH), pharmacological inhibition of ATP-citrate lyase reduced hepatic steatosis, fibrosis, inflammation, blood glucose, and triglycerides [112]^[53], suggesting that the beneficial effect of BT2 may also be ascribed to this ATP-citrate lyase inhibition rather than to BCKD activation. Data in support of a role for acylcarnitines are quite limited [103]^[44]. In C2C12 myotubes, Aguer et al. [113]^[54] evaluated the effects of C4 acylcarnitines on insulin signaling and showed a significant reduction in Akt phosphorylation, which contrasts with the presumed effects of BCAAs that lead to hyperactivation of the Akt/mTORC1 pathway. On the contrary, and although this evidence is indirect, various studies show rather absent or inverse relationships. Experimentally, in HFHS- or HF-fed mice, BCAA supplementation is associated with increased C3 and C5 acyl-carnitine, with no impact on insulin resistance. From a clinical point of view, in the follow-up of patients with NAFLD, C5 acylcarnitine concentrations decreased with the progression of liver damage [114]^[55]. In the metabolic syndrome, Remchak et al. [115]^[56] characterized the chronotype of their patients and showed, in the early chronotype group (“early birds”) compared to the late chronotype group (“night owls”), a better insulin sensitivity and an increase in plasma concentrations of most acylcarnitines. The last candidate is HIB [103]^[44]. Note that HIB-dehydrogenase deficiency, a rare metabolic disease with inability to metabolize HIB, does not seem associated with severe insulin-resistance [116]^[57]. Moreover, efforts to improve insulin sensitivity by increasing the flux through the BCKD are a priori accompanied by an increase in the availability of HIB by simple flux effect. The hypothesis of a detrimental effect of HIB is based on the work of Jiang et al. [117]^[58], showing that conditioned-medium from C2C12 myoblasts overexpressing peroxisome proliferator-activated receptor-gamma coactivator 1α (PGC-1α) contained increased levels of HIB and that this HIB increased trans-cellular transport of fatty acids by HUVECs (human umbilical vein endothelial cells) for subsequent oxidation. However, in vivo, although mice supplemented with high doses of HIB displayed delayed glucose disposal, they presented a normal basal blood glucose. These data are all the more puzzling because PGC-1α activation has been associated with improved metabolic status. In this respect, Roberts et al. [118]^[59], in myocyte overexpressing PGC-1α, showed an increase in BAIBA release and that this BAIBA induced adipocyte browning. In addition, Hatazawa et al. [119]^[60] did not observe any increase in muscle HIB in PGC-1α transgenic mice but an increase in BAIBA content.