FR van der Leij, Unit Life Sciences & Technology, University of Applied Sciences, Van Hall Larenstein and Noordelijke Hogeschool, Leeuwarden 8901 BV, the Netherlands. E-mail: email@example.com
Insulin sensitizers like metformin generally act through pathways triggered by adenosine monophosphate-activated protein kinase. Carnitine palmitoyltransferase 1 (CPT1) controls mitochondrial β-oxidation and is inhibited by malonyl-CoA, the product of acetyl-CoA carboxylase (ACC). The adenosine monophosphate-activated protein kinase-ACC-CPT1 axis tightly regulates mitochondrial long-chain fatty acid oxidation. Evidence indicates that ACC2, the isoform located in close proximity to CPT1, is the major regulator of CPT1 activity. ACC2 as well as CPT1 are therefore potential targets to treat components of the metabolic syndrome such as obesity and insulin resistance. Reversible inhibitors of the liver isoform of CPT1, developed to prevent ketoacidosis and hyperglycemia, have been found to be associated with side effects like hepatic steatosis. However, stimulation of systemic CPT1 activity may be an attractive means to accelerate peripheral fatty acid oxidation and hence improve insulin sensitivity. Stimulation of CPT1 can be achieved by elimination or inhibition of ACC2 activity and through activating transcription factors like peroxisome proliferator-activated receptors and their protein partners. The latter leads to enhanced CPT1 gene expression. Recent developments are discussed, including a recently identified CPT1 isoform, i.e. CPT1C. This protein is highly expressed in the brain and may provide a target for new tools to prevent obesity.
In view of the lack of successful behavioural measures to prevent or treat components of the metabolic syndrome such as obesity, additional strategies are needed to halt the diabetes epidemic. This review focuses on intervention in fatty acid oxidation (FAO) as potential means to prevent or treat obesity and insulin resistance. Most current pharmacological treatment options for diabetes, e.g. metformin, interfere with adenosine monophosphate-activated protein kinase (AMPK) (1) and other energy-sensing pathways. Drugs that act downstream of AMPK (Fig. 1) may therefore be potential modulators or (partial) substitutes for existing drugs.
Mitochondrial β-oxidation represents a crucial process in energy metabolism and is tightly regulated by interactions between the key enzymes carnitine palmitoyltransferase 1 (CPT1) (2) and acetyl-CoA carboxylase (ACC) 2 (3) via the intermediate malonyl-CoA (4) (Fig. 2). The β-oxidation process itself consists of four enzymatic steps (acyl-CoA dehydrogenase, 2-enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase and 3-ketoacyl-CoA thiolase) each of which are carried out by several enzymes with their own chain-length specificity. The flux of this pathway is primarily determined by CPT1 (2,5) as well as by substrate supply (5,6). It is evident that tissue specificity of the regulation of β-oxidation is important, as is the site of action of potential drugs. Paradoxically, two extreme measures to influence lipid oxidation have been advocated: either inhibition or stimulation as a means to prevent or treat symptoms of the metabolic syndrome.
The carnitine palmitoyltransferase system
During fasting or prolonged exercise, plasma glucose levels drop and ketone bodies are produced to provide important organs like the brain with sufficient amounts of energy. Ketone bodies (acetoacetate, β-hydroxybutyrate and acetone) in mitochondria of liver are generated from fatty acids and, to a lesser extent, from ketogenic amino acids (1,2). Long-chain fatty acids (LCFA) are released from triglycerides in adipose tissue under the control of lipases (7), and can then be taken up from blood into cells passively, or via transporter proteins. One example hereof is the fatty acid transporter CD36, the expression level of which is known to be increased in livers of patients with type 2 diabetes (8) leading to enhanced hepatic LCFA uptake. One of the actions of metformin is to dampen this effect, thereby improving the cellular lipid status.
Upon cellular uptake, LCFA are activated by cytosolic acyl-CoA synthases (ACS; Fig. 2). The so-called carnitine shuttle enables activated LCFA to enter the mitochondria for entry in the β-oxidation spiral. This shuttle provides the net transport of LC acyl-CoA across the mitochondrial membranes (2) and is facilitated by three proteins: CPT1, carnitine acylcarnitine translocase and CPT2 (2,9). As a first step, CPT1 exchanges the CoA group of LC acyl-CoA for carnitine to form LC acylcarnitines (Fig. 3). The LC acylcarnitines are transported across the mitochondrial inner membrane by carnitine acylcarnitine translocase. Finally, inside the mitochondria, CPT2 releases the carnitine group in exchange for a CoA group and the carnitine shuttles back to the cytosol. In the mitochondrial matrix, the LC acyl-CoA are oxidized by the enzymes of the β-oxidation spiral (5).
Carnitine palmitoyltransferase 1
In mammals, three CPT1 isoforms exist that are encoded by different genes (2,9,10). The liver isoform (CPT1A) and the muscle isoform (CPT1B) are localized in the outer mitochondrial membrane and expose their active sites at the cytosolic face of the mitochondrion (11) (Figs 1 and 2). CPT1A, ∼88 kDa in size, represents the major hepatic isoform but is also expressed in heart, spleen, lung, kidney and adipose tissue (2,12). CPT1B has a predicted size of ∼88 kDa, but folds in such a way that its apparent size is about 82 kDa (6,12). In humans, this isoform is present in white adipose tissue, cardiac and skeletal muscle as well as in testis (6,12). Profound species- and gender-dependent differences in expression of CPT1A and CPT1B exist. For example, CPT1A is expressed in white adipose tissue of male mice but not of female mice. CPT1B is expressed in white adipose tissue of humans, female mice and male rats, but not in male mice and female rats (13). The cardiac expression patterns of CPT1A and CPT1B change during development (6,12). In general, heart CPT1A expression levels are high around birth and drop thereafter whereas CPT1B shows the opposite pattern (6,12,14). In the diabetic state, expression differences between the isoforms have been noted; i.e. increased CPT1A expression levels have been reported in the liver and the heart during diabetes as well as a small increase of CPT1A expression levels in the skeletal muscle (14). Hepatic expression of CPT1B is induced in vivo under specific circumstances (15). However, in view of the high sensitivity of CPT1B to inhibition by malonyl-CoA, such a hepatic induction seems not to be physiologically relevant.
A third isoform of CPT1 (CPT1C) is exclusively present in brain and testis (10). This enzyme has been discovered later than CPT1A and CPT1B and the function of this protein is not yet fully understood. It has been suggested that CPT1C regulates whole-body energy homeostasis (10) and, indeed, evidence from animal studies points towards that direction (16–18). Functional studies have been hampered by the fact that the enzyme is not active in mitochondria, although it is able to bind to the natural CPT1 inhibitor malonyl-CoA (10,16). It has recently been reported that CPT1C has a prevalent role in the endoplasmic reticulum (19). This finding warrants further investigation because it will be important to understand the contribution of each CPT1 isoform to brain energy metabolism and hypothalamic fuel sensing (18,20). CPT1A is also expressed in the brain and hypothalamic intervention directed against this isoform affects appetite control as well as glucose production (20). Therefore, brain CPT1A is important for glucose homeostasis and energy metabolism. Although sometimes stated otherwise (21), CPT1B is also expressed in important brain areas in mice (22), rats and humans (see e.g. the AceView database at http://www.ncbi.nlm.nih.gov, and its expression coincides with that of ACC2 (22)).
Malonyl-CoA, produced by condensation of two acetyl-CoA moieties as catalysed by ACC, is a natural inhibitor of CPT1 (2–4). The sensitivity for inhibition is different for CPT1A and CPT1B, the latter being about 100-fold more sensitive. Insulin and thyroid hormone can regulate the sensitivity of CPT1A for malonyl-CoA in the liver (23); however, the sensitivity of CPT1B is not altered by these hormones (24). Studies in ACC2-deficient mice (25) have indicated the existence of separate cellular malonyl-CoA pools. Malonyl-CoA produced by the cytosolic enzyme ACC1 is primarily used for lipogenesis, whereas the product of ACC2 mainly functions as CPT1 inhibitor (25). The latter mode of action provides major flux control over mitochondrial β-oxidation (5).
Upstream regulation of the carnitine palmitoyltransferase system
Insulin negatively regulates hepatic CPT1A activity via insulin growth factor I-receptor. Insulin is able to reverse the effects of diabetes on the sensitivity of CPT1 for malonyl-CoA and on CPT1 activity itself (23). The Ki value of malonyl-CoA-induced inhibition of CPT1 activity decreases when diabetic animals are injected with insulin, indicating an increase in malonyl-CoA sensitivity. Hepatic CPT1A activity is increased in the diabetic state and the ability of malonyl-CoA to inhibit this activity is reduced. Insulin is capable to reduce CPT1A mRNA levels; however, short-term treatment with insulin does not always change CPT1A activity (23). In ketotic rats, CPT1 activity is increased (26), which can be reduced by insulin treatment. Thyroid hormone is another regulator of CPT1A with profound effects on β-oxidation in the liver. In rats, thyroid hormone treatment caused a fivefold increase in hepatic mRNA expression levels of Cpt1a(27) while in hypothyroid rats a decrease of hepatic mRNA levels of Cpt1a was observed (28). These effects are mediated by the thyroid hormone receptor as an active heterodimer with the retinoid X receptor (27).
Transcriptional regulation of CPT1A involves several other transcription factors, including the peroxisomal proliferator-activated receptor gamma coactivator-1 (PGC-1). PGC-1α is capable to stimulate CPT1A expression in both liver and heart (29). PGC-1 acts via hepatocyte nuclear receptor 4, peroxisomal proliferator-activated receptor alpha (PPARα) and the glucocorticoid receptor. Although not consistently at significant levels, PPARα is also found to regulate CPT1A gene expression (15), and a peroxisome proliferator responsive element was identified in a conserved region of mammalian CPT1A(30). Other data suggest PPARα-independent pathways, at least in rodents (31,32). Therefore, the transcriptional regulation of CPT1A involves several independent processes.
In muscle, CPT1B expression is regulated by PPARα and retinoid X receptor and appears to be even more responsive to PPARβ/δ(33,34). Human CPT1B transcription is regulated via a conserved peroxisome proliferator responsive element, which is responsible for high responsiveness to cellular fatty acid accumulation in cardiac myocytes and myocyte tubules (33). An unexpected family of proteins that interact with PPARβ/δ in the regulation of CPT1B is the family of Krüppel-like transcription factors that are subject to modification by small ubiquitin-related proteins (34). This finding may open new roads for drug development to accelerate (skeletal) muscle β-oxidation. PGC1 is involved in the regulation of CPT1B in heart by acting together with myocyte enhancer factor 2 to activate gene transcription (35). Thus, subtle differences in the regulation of CPT1A and CPT1B expression exist, as represented by different (sub)sets of transcription factors involved. As PPARγ barely influences CPT1A and CPT1B expression, thiazolidinediones seems to have no direct effects at this level of regulation.
No studies on the transcriptional regulation of CPT1C have been reported thus far. However, CPT1C-deficient mouse models have been generated to evaluate the effects of disruption of this brain-specific gene (16,17). Results obtained with classic Cpt1c knockout mice showed decreased food intake and body weight. These mice were shown to be highly susceptible to diet-induced obesity upon feeding a high-fat (HF) diet (17), demonstrating the complexity of CPT1C involvement in body weight regulation. Thus, therapeutic agents to modulate hypothalamic CPT1C activity may be of interest for treatment of specific components of the metabolic syndrome.
Pharmacological inhibition of CPT1
The initial rationale to develop CPT1 inhibitors was to prevent ketoacidosis and hyperglycemia (36). Up to date, the treatment and prevention of ketoacidosis consist of proper insulin regimens whereas inhibition of CPT1 has never reached the status of a clinically applicable alternative. However, to reduce hyperglycemia due to (hepatic) insulin resistance, inhibition of CPT1 activity by etomoxir or tetradecyl glycidic acid (TDGA) has been tested. Whereas experiments in mice showed only marginal reductions of de novo glucose production upon TDGA treatment (37), clinical trials with etomoxir indeed showed an improvement of hyperglycemia (38). However, severe side effects like myocardial hypertrophy and hepatic steatosis put an end to the development of etomoxir and TDGA as therapeutic tools.
To circumvent undesired side effects outside the liver, reversible CPT1 inhibitors were developed for liver-specific inhibition, but little progress has been reported in this field. The last peer-reviewed publication of the development of such an agent dates back to 2003 (39). Treatment of db/db mice with the reversible CPT1A inhibitor ST1326 (40) caused a reduction of plasma glucose levels whereas insulin levels and other parameters like plasma triglycerides were not affected. Recent in vivo studies revealed an improvement of hyperglycemia upon 4 days of treatment with ST1326 followed by metformin treatment for 1 day. Clinical trials showed marginal improvements of the insulin values in diabetic patients (39,40). It has been concluded earlier that inhibition of CPT1 may be a dead-end story in terms of pharmacological feasibilities (41). Moreover, a liver-specific approach does not address the problem of hepatic steatosis, a potential prelude to insulin resistance and, eventually, type 2 diabetes.
Obici et al. used various forms of intervention to inhibit hypothalamic CPT1 expression and activity. Both brain-specific genetic deletion of CPT1A and pharmacological CPT1A inhibition in rats caused considerable decreases in food intake as well as reduction of endogenous glucose production (20). These findings indicate that central inhibition of FAO is of potential interest to prevent and/or treat obesity and insulin resistance.
Stimulation of fatty acid oxidation
Chronic systemic inhibition of CPT1 with etomoxir caused intracellular lipid accumulation and insulin resistance in rats (42). Therefore, one could reason that stimulation of systemic CPT1 activity would improve insulin sensitivity and accelerates peripheral FAO. Indeed, recent studies showed that increased CPT1B activity leads to improved insulin sensitivity in high-fat overfed rats (43). Although obesity is associated with increased lipid oxidation (44), accumulation of triglycerides in muscle tissue in type 2 diabetes has been observed (45). An induction of the oxidation rate in these patients may reduce the levels of triglycerides and improve peripheral insulin resistance. Whether successful stimulation can be achieved directly is questionable (see below). However, indirect CPT1 stimulation can be achieved in several ways (Table 1).
Table 1. Strategies to indirectly stimulate carnitine palmitoyltransferase 1
CPT1 activity also can be ‘stimulated’ by protection against damage caused by reactive oxygen species. The protective anti-oxidative compound astaxanthin improves muscle lipid metabolism during exercise by preventing oxidative CPT1 modifications (69).
Drugs that target co-activators (PGC1; Krüppel-like transcription factor 5)
Direct stimulation of carnitine palmitoyltransferase 1
CPT1 is a complex enzyme whose activity is regulated by malonyl-CoA through allosteric inhibition (9). C75 is a compound designed to mimic cerulenin, an inhibitor of fatty acid synthase (FAS). FAS catalyses the malonyl-CoA consuming step in lipogenesis that succeeds ACC1. Remarkably, C75 was found to act as a direct stimulator of CPT1 activity (46). Treatment with high doses of C75 increased peripheral FAO and energy utilization in obese mice. Although this part of action is difficult to distinguish from FAS inhibition, another cerulenin-mimic compound (C89b) was also found to stimulate CPT1 (47), and several biochemical studies confirmed CPT1 stimulation by C75. In 2006, however, it was discovered that low doses of C75 that are activated to C75-CoA act as inhibitor of CPT1 (48). These findings have recently been confirmed for C75-CoA action in the brain (21). Pharmacological stimulation of brain CPT1 in mice by C89b resulted in decreased body-weight gain and food intake without affecting fatty acid synthesis (47). Yet, Mera et al.(21) concluded that these phenomena are due to inhibition rather than stimulation of brain CPT1. It should be noted that CPT1B and CPT1C are not considered to be subject of hypothalamic C75-CoA inhibition (21); however, this discussion is not settled yet.
The acetyl-CoA carboxylase system
ACCs are important enzymes for both the lipogenic cascade and for regulation of mitochondrial β-oxidation. These enzymes catalyse the carboxylation of acetyl-CoA into malonyl-CoA. Malonyl-CoA is an intermediate in the synthesis of fatty acids and, as discussed above, serves as a natural inhibitor of mitochondrial β-oxidation (Fig. 4). Two isoforms of ACC, i.e. ACC1 and ACC2, have been identified and are encoded by separate genes. The notion that the two gene products have different functions became evident from studies in specific knockout mouse models (25,49,50) as well as from studies applying RNA interference technology (51). In fact, these data indicate that the two isoforms deliver malonyl-CoA into separate pools: a cytosolic pool used for lipogenesis and a subcytosolic pool – presumably in close proximity to the mitochondrial outer membrane – to inhibit CPT1.
ACC1 is a 265-kDa protein that is mainly expressed in liver and adipose tissue (52). The malonyl-CoA generated by this enzyme is preferably used for lipogenesis (25). ACC2 is a 280-kDa protein that, in comparison with ACC1, contains 114 additional amino acids at the N-terminus to anchor the protein onto the mitochondrial membrane (Fig. 1) (3,53). ACC2 is mainly expressed in muscle, but its expression in the liver and several other organs is also of significant importance (25).
Regulation of ACC enzyme activity and transcription
AMPK is a major regulator of ACC1 and ACC2 activities (Fig. 1). In situations in which energy is required, AMPK is activated by several protein kinases that phosphorylate AMPK at specific serine and thyrosine residues. Phosphorylated AMPK, in turn, is able to phosphorylate ACC to inactivate the latter (54) (Fig. 4). This cascade can be reversed by protein phosphatases. At metabolic level, citrate allosterically stimulates both isoforms of ACC (55) (Fig. 4). Thus, energy shortage and excess control ACC activity at short-term intervals.
Transcription factors that influence the expression of both isoforms of ACC provide regulation at the mid- and long-term. Sterol regulatory element binding protein 1c (SREBP1c) is an important transcription factor in this respect. After dietary intake of cholesterol and upon stimulation by insulin, ER-bound SREBP1c is cleaved and its activated form is transported into the nucleus where it promotes transcription of Acc1(56) and Acc2(57,58). Shimomura et al. were the first to show that insulin stimulates the expression of SREBP1c and its target genes (57).
The carbohydrate responsive element binding protein (ChREBP) is also involved in the regulation of Acc1 expression and most likely also of Acc2(59). Exposure of hepatocytes of wild-type mice to high glucose levels resulted in induction of mRNA levels of genes involved in lipogenesis, including those for ACC1 and FAS. High glucose failed to induce this response in hepatocytes of ChREBP−/− mice (59). Acc gene expression is responsive to fibrates and the induction of Acc2 coincides with Cpt1b responses in mouse liver (15). Balanced increases in gene expression may lead to higher sensitivities and quicker responses to subtle metabolic changes and in that way improve insulin sensitivity.
Genetic targeting of acetyl-CoA carboxylase in mice
Mouse models deficient in either one of the ACC isoforms have yielded a wealth of information about their physiological roles. These mouse models clearly indicated the existence of two functionally separate malonyl-CoA pools. Acc2−/− mice showed continuous FAO, reduced fat storage and improved insulin sensitivity, although they also showed a 30% increase in food intake compared with the intake of wild-type littermates (25). Moreover, lipogenesis in these mice was not affected, indicating that malonyl-CoA generated by mitochondrial ACC2 is not required for progression of fatty acid synthesis (3,25). Acc2−/− mice are protected from diet-induced obesity upon feeding a HF or a high-carbohydrate diet (25,50). Experiments performed in liver-specific Acc1 knockout mice showed a reduction in triglyceride accumulation in this organ without affecting glucose metabolism (60), which is another indication for the existence of distinct malonyl-CoA pools.
Similar to Cpt1a−/− and Cpt1b−/− knockout mice (61,62), Acc1−/− knockout mice are not viable and die at embryonic day 8.5 (49), underscoring the importance of lipogenesis in early embryonic development. Mao et al.(60) showed that – under normal feeding conditions – liver-specific deletion of Acc1 does not cause any health problem. Hepatic triglyceride content was reduced in these mice upon feeding a fat-free diet without affecting FAO and glucose homeostasis. However, upon feeding a HF diet the liver-specific absence of ACC1 did not prevent obesity and insulin resistance (60).
Pharmacological inhibition of acetyl-CoA carboxylase
In view of the different functions of the two putative malonyl-CoA pools, inhibition of either one or both ACC isoforms can be of interest for treatment of obesity and/or insulin resistance. Inhibition of ACC1 would decrease the rate of lipogenesis and inhibition of ACC2 would stimulate mitochondrial β-oxidation. The beneficial effects of such strategies were shown in the antisense oligonucleotide (ASO) study by Savage et al.(51). In these studies reduction of mRNA of both ACC isoforms using ASO technology showed that diet-induced hepatic steatosis and hepatic insulin resistance could be reversed in rats (51). Therefore, clinical applications of ASOs against ACC2 expression may be a future promise to treat insulin resistance; yet, several technical issues concerning the formulation of the compounds need to be solved.
One of the earliest developed chemical ACC inhibitors is 5-(tetradecloxy)-2-fuoric acid. This compound induces the hepatic accumulation of glucose as well as ketogenesis and β-oxidation (63). The CoA-ester of 5-(tetradecloxy)-2-fuoric acid is able to inhibit both ACC isoforms and causes a decrease in malonyl-CoA levels, which leads to the stimulation of β-oxidation and a reduction in fatty acid synthesis (63). Other more recently developed compounds are CP-640186 (64) and Soraphen (65). CP-640186 is a reversible non-selective ACC inhibitor. It has similar affinity for both ACC isoforms (64) and rats treated with this compound showed a reduction in fatty acid synthesis and triglyceride synthesis. Malonyl-CoA levels of lipogenic and oxidative tissues were decreased and whole-body lipid oxidation was increased in these animals (64). However, experiments in diabetic ob/ob mice showed an increase in plasma glucose and triglyceride levels while glucose tolerance was worsened in these mice (66). Experiments with CP-640186 in mice with diet-induced insulin resistance showed reduced body-weight gain and improved peripheral insulin sensitivity (67).
Soraphen has been suggested to be a more potent ACC inhibitor compared with CP-640186 (68). It binds to the biotin carboxylase domain of ACC and thereby disrupts the oligomerization needed for activity. In experiments using mice fed a HF diet, treatment with Soraphen (50 or 100 mg kg−1 d−1) caused a reduction in body-weight gain and a major improvement in peripheral insulin sensitivity. We also observed about 70% reduction of de novo lipogenesis (67). Results of these studies show that indeed, ACC is an interesting drug target to improve insulin sensitivity.
Direct intervention in the CPT system, or indirect intervention via the ACC system, potentially has beneficial effects for the prevention and treatment of obesity and other aspects of the metabolic syndrome. However, awareness of potential side effects remains a major issue. For that reason, CPT1 inhibition may not be a fruitful choice as it leads to lipid accumulation and insulin resistance.
Inhibition of the ACC system, however, has shown its potential in several experimental models without obvious side effects. Although various issues need to be resolved, the application of tools such as ASOs seems promising. Potential other means to indirectly promote systemic CPT1 activity via transcriptional mechanisms include the application of PPARβ/δ agonists and the prevention of damage to CPT1 by counteracting reactive oxygen species.
Recent work has pointed to crucial roles of hypothalamic ACC and CPT1 in control of food intake and energy homeostasis. The contribution and precise mode of action of each hypothalamic CPT1 isoform has not been fully established. Yet, CPT1C may provide a new and unique target for intervention, as it appears to be active in the endoplasmic reticulum rather than in mitochondria.