Deactivating Fatty Acids: Acyl-CoA Thioesterase-Mediated Control of Lipid Metabolism

Characteristics

Type I ACOTs belong to the α/β-hydrolase protein family, which also includes many lipases and esterases. These ACOT genes localize to narrow gene clusters in higher mammals and exhibit high sequence homology to each other and among species, suggestive of evolution by gene duplication from a common ancestor [10]. Acots 1–6 reside on mouse chromosome 12 D3, whereas in humans a cluster on chromosome 14q24.3 contains only 4 genes (ACOTs 1, 2, 4 and 6) of which 3 (ACOTs 1, 2 and 4) are transcribed and then translated to their respective full-length proteins [10–13]. The fourth gene (ACOT6) is translated from a downstream methionine in exon 2, generating a markedly smaller protein. Although ACOT6 contains a catalytic site, it is still unknown whether it possesses enzymatic activity [11]. Type I ACOTs are encoded by 3 exons and the crystal structure of ACOT2 reveals two major domains, an N-terminal β-sandwich domain and a C-terminal hydrolase domain, with the active site located in a hollow space in between the two domains [10, 11, 14, 15]. Type 1 ACOTs, like the other α/β-hydrolases, utilize a catalytic triad of a serine (located in a so-called nucleophilic elbow and acting as a nucleophile), a histidine and an aspartic acid [16]. Although the amino acid sequence nearby to the catalytic triad is highly conserved, lower sequence similarities in other parts of the C-terminal coding regions may contribute to the differences in substrate specificity and different targeting signals that mediate subcellular localization.

The K_m values for type I ACOTs are mostly in the low μM range for preferred substrates, with V_max values in the low umol/min/mg range. Even at high concentrations, free CoA does not inhibit their activities, suggesting that transcriptional regulation might be more important compared to regulation by end products [16]. Increased expression in response to PPARα activation suggests that the type 1 ACOTs may play important roles in fatty acid oxidation and energy homeostasis [17].

Metabolic functions

ACOT1 is a cytosolic enzyme with substrate selectivity for long chain (C12–C20) saturated-and monounsaturated acyl-CoAs. It is expressed mainly in liver, kidney, lung and heart in mouse [11, 16]. ACOT1 appears to modulate the cytosolic pool of long chain acyl-CoAs and FFA, potentially controlling ligand availability for the nuclear hormone receptors PPARα and HNF4α [17]. In mouse models of diabetes and sepsis-related cardiac-dysfunction, overexpression of ACOT1 reduced reactive oxidative species and improved cardiac function [18, 19].

ACOT2 is also selective for long chain fatty acyl-CoAs. An N-terminal mitochondrial localization sequence targets ACOT2 to the mitochondrial matrix in both rodents and humans [11, 20], suggesting that this enzyme regulates mitochondrial fatty acid oxidation [20]. Overexpression of ACOT2 in mouse liver, where it is normally expressed at low levels, increased hepatic utilization of fatty acids for oxidative phosphorylation in situ [21]. It was postulated that ACOT2 promotes fatty acid oxidation via a fatty acid export-activation-uptake cycle. This action may prevent build-up of acyl-CoAs in the mitochondrial matrix and also promote re-utilization of intra-mitochondrial CoA to support high rates of β-oxidation. ACOT2 also introduced a proton leak that may have contributed to some degree of uncoupling of respiration [21].

Mouse ACOTs 3–6 and human ACOT4 contain functional type 1 peroxisome targeting sequences at their C-termini. ACOT3 and ACOT5 exhibit substrate specificities for long-and medium chain acyl-CoAs, respectively. ACOT4 and ACOT6 show more narrow substrate specificities with ACOT4 hydrolyzing short dicarboxylic acyl-CoA esters, particularly succinyl-CoA, and ACOT6 hydrolyzing methyl-branched acyl-CoA esters, such as pristanoyl-CoA [12, 13, 22]. ACOT3 and ACOT5 are not present in human, however, human ACOT4 is active towards both short-chain dicarboxylyl-CoAs (e.g. succinyl-CoA) and medium- to long-chain acyl-CoAs and thereby combining activities of ACOTs 3–5 [11]. The functions of these ACOTs may be to promote peroxisomal β-oxidation by redirecting fatty acids and products after chain-shortening by the peroxisomal β-oxidation system to lengths that are suitable for export via specific membrane transporters. These ACOTs also provide a system for recycling of CoA within the peroxisome [5] (see Text Box).

Characteristics

In contrast to the type 1 ACOTs, type II enzymes share low degrees of sequence similarity but rather have common structural features. The defining structural motif of these ACOTs is the so-called HotDog domain [23], which is described as a seven-stranded antiparallel β-sheet ‘bun’ that wraps around a hydrophobic five-turn α-helical ‘sausage’, with a third layer comprising loops over the helix as ‘condiments’ [24]. Because they are identified based on structure, the group of type II enzymes has grown to consist of ACOTs 7–15, with the possibility of additional undiscovered members. Mouse ACOTs 9 and 10 share 96% identity, and Acot10 is only found in the mouse genome as an intron-less gene, and is most likely a retrogene [8, 25].

An additional distinctive feature of type II ACOTs is that their active sites are located at the dimer interface between two HotDog domains. Moreover, the dimers tend to self-associate, so that functional enzymes exist as oligomers, tetramers or trimers of dimers. Dimers may consist of tandem HotDog domains in the same protomer (e.g. ACOT7), or dimers of single HotDog domain protomers (e.g. ACOT13) [26–28]. Structural studies of ACOT7 have shown that the HotDog domains of the N-and C-termini need to associate to form the catalytic site of the enzyme. The dimerization of these units also creates a second (in)active site with potential regulatory function, which does not contain all the catalytic amino acids found in the active catalytic site. When the inactive site is mutated to contain the proper catalytic amino acids, the thioesterase activity of ACOT7 is increased [28].

In addition to tandem HotDog domains, ACOT11 and ACOT12 each contain a lipid-binding C-terminal steroidogenic acute regulatory transfer-related (START) domain, so they are also included in the START domain family of proteins as StarD14 and StarD15 [29]. This inclusion of a START-domain is an occurrence that is so far limited to mammalian ACOTs and is suggested to provide an additional level of intrinsic lipid-mediated regulation of the thioesterase activity [8, 23, 27, 30]. Interestingly, the single HotDog domain ACOT13 was originally identified as a binding partner of the minimal START domain protein phosphatidylcholine transfer protein (PC-TP; synonym StarD2) [31]. PC-TP increases the acyl-CoA thioesterase activity of ACOT13 in vitro, supporting the notion of START domains as regulatory units when contained with ACOTs (Figure 1) [3, 8, 31].

As with type 1 ACOTs, the type II enzymes are also localized to various cellular compartments in a wide range of tissues and are active towards a variety of acyl-CoA molecular species. Type II ACOTs have been more extensively studied using genetically engineered mice, so their likely biological roles are somewhat better appreciated.

Metabolic functions

ACOT7 was originally referred to as brain acyl-CoA hydrolase (BACH) due to its high expression and activity in the mammalian central nervous system [32]. ACOT7 accounts for the majority of the thioesterase activity in brain tissue, where it is highly expressed in the cytosol of neurons [33, 34]. Although high levels are found in brain and in testis, ACOT7 isoforms are found in other tissues as well. These isoforms are attributable to alternative splicing events that, in two instances, introduce targeting signals for mitochondrial and nuclear localization, respectively [34, 35]. ACOT7 is active towards a spectrum of long chain acyl-CoAs, including arachidonoyl-CoA and palmitoyl-CoA [28, 34]. Macrophages also expresses ACOT7, and decreased cellular levels of prostaglandins D2 and E3 upon overexpression of the enzyme support the likelihood that arachidonoyl-CoA is a preferred metabolic substrate for this enzyme [28].

A role for ACOT7 in neurological function was suggested by the discovery of low to absent levels of an isoform of ACOT7 in the hippocampus of patients with mesial temporal lobe epilepsy [36]. In keeping with this possibility, mice with conditional deletion of ACOT7 in neurons (Acot7^N−/−) exhibit age dependent neurodegeneration, with decreased cerebellar neuronal cell density giving rise to ataxia, as well as decreased motor coordination and locomotion. Biochemical phenotyping revealed that the loss of neuronal ACOT7 led to a decrease in brain thioesterase activity by >80%. Nevertheless, there were no changes in the tissue levels of long chain fatty acyl-CoAs, nor of free CoA. Interestingly, metabolic changes were observed only after overnight fasting and contrary to expectations, increases in long chain FFA were detected in the Acot7^N−/− mouse brain. Increased tissue levels of more complex lipids, such as sphingosines, phospholipids and triacylglycerols were also observed in fasting Acot7^N−/− mice, raising the possibility that ACOT7 might limit lipid accumulation in neurons during periods when FFA are abundant [33]. The relevance of this mouse model was underscored by the discovery of a patient with a homozygous frame shift deletion encompassing part of the first exon in ACOT7 during a screen for copy number variants in patients with undiagnosed neurodevelopment disorders [37]. This mutation was implicated as causal for the phenotype that was characterized by epilepsy, mild intellectual disability and abnormal behavior, similar to the Acot7^N−/− mouse [37].

The role of ACOT7 in non-neuronal tissues is less clear. Expression of ACOT7 is normally very low in pancreatic β-cells, but is increased in Zucker diabetic fatty rats, as well as in type 2 diabetic patients [38, 39]. Indeed, ACOT7 seems to contribute to β-cell dysfunction since overexpression in pancreatic β-cells lines and in a transgenic mouse model decreased insulin secretion in response to metabolic stimuli. It was suggested that ACOT7 induces β-cell ATP consumption, thereby decreasing metabolic sensitivity in the endocrine pancreas [40].

Interestingly, Acot7^N−/− mice exhibit additional metabolic abnormalities including increased adiposity, dyslipidemia and increased liver triglycerides after fasting, but the mechanistic basis for these effects remain unclear. To test for direct metabolic effects, an ACOT7 protein engineered to harbor gain-of-function mutations [28] was overexpressed in liver and adipose tissue [41]. Despite measurable increases in thioesterase activity, overexpression in liver did not alter the flux of fatty acids for oxidation or incorporation into complex lipids. Nor were any metabolic changes observed after the chronic metabolic stress of high fat feeding or an acute metabolic stress of 18h of food restriction. Similarly, there were no changes observed in body weight, adipose tissue mass or glucose tolerance after overexpression in adipose tissue. By contrast, increased thioesterase activity in brown adipose tissue (BAT) was associated with decreased capacity to maintain body temperature during an acute cold exposure at 4°C [41].

ACOT8 is a peroxisomal protein with broad substrate specificity, including medium-to long-chain and methyl branched acyl-CoAs, intermediates of β-oxidation and bile acid-CoAs [5] (see Text Box). The activity is inhibited by CoA and expression is upregulated by fasting and PPARα activation [42] and ACOT8 overexpression leads to peroxisomal proliferation in murine and human cell lines [43]. ACOT8 was first identified as an interacting partner of the HIV-1 Nef protein, a protein that is important for viral replication and infectivity [44, 45]. This interaction is thought to protect against proteasome degradation of Nef [46]. Interestingly, Nef increased the thioesterase activity of ACOT8 in vitro [44]. ACOT8 expression is increased in human hepatocellular carcinoma, and cell culture studies suggest a pathogenic role for the enzyme [47]. In lung adenocarcinoma, ACOT8 is upregulated in lymph node metastases and is associated with a poorer prognosis [48]. Whether ACOT8 contributes to tumorigenesis via its enzymatic activity or its capacity to induce peroxisomal proliferation is unclear.

ACOT9 is localized to mitochondria and is widely expressed, with high levels observed in mouse kidney, adipose tissue and brain, and very low expression in liver [25, 49]. ACOT9 is most active towards long-chain saturated acyl-CoAs, but also towards saturated and branched short-chain acyl-CoAs, many of which serve as intermediates in the pathways of amino acid degradation [25, 49]. These findings suggest that ACOT9 provides a regulatory connection, or metabolic crosstalk, between mitochondrial fatty acid and amino acid metabolism [25].

ACOT11, commonly referred to as Thioesterase superfamily member 1 (THEM1) or StARD14, was first described as brown fat-inducible thioesterase (BFIT), which is upregulated in BAT by cold ambient temperatures [50]. THEM1 hydrolyzes a broad range of substrates from acetyl-CoA to long chain fatty acyl-CoAs, with long chain acyl-CoAs being preferred in vitro [30, 51]. The lipid ligand of the START domain remains unknown, although the crystal structure suggests FFA may be accommodated within the lipid binding pocket [52]. Whereas the START domain exhibited binding to phosphatidylinositol 4-phosphate in a protein-lipid overlay assay, evidence is lacking for its specific binding within the lipid binding pocket [51]. Although not required for thioesterase activity, the START domain is essential for optimal enzyme activity. ATP stimulates, whereas ADP and CoA inhibit THEM1 activity, while each of these metabolites induce dimerization of the enzyme [30]. The cellular localization of THEM1 remains unclear. Studies in cell lines suggest that THEM1 is imported into mitochondria, whereas tissue fractionation studies in mouse BAT localize THEM1 to the cytosol, endoplasmic reticulum and nucleus [30, 51].

Due to its induction by cold exposure and the finding of increased mRNA expression in obesity-resistant mouse strains, it was suggested that THEM1 would promote energy expenditure [50]. In contrast to expectations, Them1^−/− mice exhibited increased energy expenditure and resistance to diet induced-obesity. Increased fatty acyl-CoAs in BAT and decreased thioesterase activity in BAT homogenates revealed that THEM1 functions as an ACOT in vivo [53]. The higher energy expenditure observed in Them1^−/− mice was explained by an enhanced cold thermogenesis, with THEM1 directly repressing lipolysis within BAT [54] (Figure 2).

In line with increased heat production and O₂ consumption, increased rates of fatty acid oxidation and glucose uptake were observed in BAT homogenates and primary brown adipocytes from Them1^−/− mice, respectively. Them1^−/− mice exhibit improved glucose homeostasis, which could be explained by the increased heat production by BAT. However, THEM1 is also upregulated by high fat feeding in liver and WAT, so additional studies are needed to ascertain the potential contributions of THEM1 to insulin resistance in these tissues.

ACOT12 was first purified as a cold labile acetyl-CoA hydrolase from the cytosol of rat liver [55]. It demonstrates substrate specificity for acetyl-CoA and to lesser extents, propionyl-, butyryl- and acetoacetyl-CoA. Like ACOT11, ACOT12 is activated by ATP and inhibited by ADP [55]. Although this ATP/ADP-dependent regulation was previously shown to influence oligomerization [56, 57], recent studies suggest that these nucleotides bind to a distinct site, and are not necessary for multimerization [27]. Interestingly, phosphatidic acids decrease the thioesterase activity of ACOT12 putatively by interaction with the START-domain, suggesting regulation that is linked to metabolic conditions within the cell [57]. ACOT12 expression is induced by metabolic stimuli including starvation and the administration of peroxisome proliferators, and is decreased by refeeding and insulin administration [57, 58]. Although ACOT12 harbors a putative peroxisomal targeting sequence at its C-terminus, only a minor fraction of the protein associates with peroxisomes [59, 60]. ACOT12 is primarily expressed in liver, kidney and small intestine [57], suggesting a possible role in gluconeogenesis, although its biological functions remain unknown.

ACOT13 (THEM2) is a small (~15 kDa) single HotDog domain thioesterase. The crystal structure of THEM2 reveals that the single HotDog structure self-associates as a homotetramer, with two homodimers stacked “back-to-back”. The physiological substrate for THEM2 appears to be long-chain fatty acyl-CoAs, although it is also active towards short-chain aromatic acyl-CoAs [26, 61, 62]. THEM2 is expressed in highly oxidative tissues, including liver, heart, kidney and BAT [31, 62, 63], and its expression is regulated by PPARα [62]. Although human THEM2 was demonstrated to co-localize to microtubules in cell culture [63], no influence of microtubules on mouse THEM2 activity was observed [62]. THEM2 is enriched in mouse mitochondria [62, 64] and liver mitochondria from Them2^−/− mice exhibit reduced thioesterase activity, further supporting the association of THEM2 with mitochondria [65]. Increased long chain fatty acyl-CoA and reduced FFA concentrations in livers of Them2^−/− mice confirm the function of THEM2 as an ACOT in vivo.

Ablation of THEM2 protected mice against high fat diet-induced hepatic steatosis and impaired glucose homeostasis [65]. The phenotype of improved glucose homeostasis is shared with Pctp^−/− mice, supporting the metabolic relevance of THEM2-PC-TP interactions [66]. Mechanistic studies in cell culture systems revealed that the PC-TP-THEM2 complex suppresses insulin signaling by reducing the abundance and stability of insulin receptor substrate 2 and by interacting with tuberous sclerosis complex (TSC) 2 to stabilize the mTORC1-inhibiting TSC1-TSC2 complex. It was therefore postulated that PC-TP functions as an intracellular sensor for changes in membrane phosphatidylcholines that vary in response to metabolic stimuli such as fasting, that in turn activates THEM2 [67]. Studies in primary hepatocytes have demonstrated that the THEM2-PC-TP complex promotes fatty acid oxidation by mitochondria, [68], providing an increased flux of TCA cycle intermediates that supports gluconeogenesis under fasting conditions. In the setting of overnutrition, THEM2 instead promotes liponeogenesis by supplying the glycerolipid-pathway with substrates for ACSL1 and GPAT [68, 69] (Figure 2). THEM2 is also expressed in BAT [62], and increased adaptive thermogenesis is observed in Them2^−/− mice, supporting the hypothesis that THEM2 regulates energy substrate utilization in BAT [70].

ACOT14 (THEM4) was discovered as an interaction partner and inhibitor of the C-terminus of Akt, leading to its initial designation as the C-terminal modulator protein (CTMP) [71]. Bioinformatics and biochemical characterization of the recombinant protein revealed a HotDog domain with thioesterase activity towards medium and long chain fatty acids [72]. Like ACOT13, THEM4 forms a tetramer consisting of two homodimers [72, 73]. An N-terminal mitochondrial localization sequence targets the enzyme to mitochondria, where it can exit the organelle in response to apoptotic signals, potentially contributing to apoptosis by inhibiting Akt [74]. Another mechanism by which CTMP might sensitize the cell to apoptosis is by controlling mitochondrial dynamics [75]. Due to its inhibition of Akt activity, THEM4 may suppress insulin signaling and cancer cell growth [76, 77]. However depending on cell and cancer type, THEM4 has also been reported to activate Akt [78, 79]. Additional studies are needed to elucidate the regulatory effects that occur in response to THEM4-Akt interactions. It is notable that this interaction does not influence the thioesterase activity in vitro [72], and Them4^−/− mice only exhibit mild mitochondrial changes without evidence of altered lipid metabolism [75]. Consequently, the biological role for THEM4 as an acyl-CoA thioesterase remains unclear.

ACOT15 (THEM5) was included in the ACOT-family based on its sequence similarity to THEM4 [8]. It consists of a single HotDog domain that forms homodimers in solution, and potentially larger oligomers in vivo, where it localizes to the mitochondrial matrix [73]. THEM5 displays substrate specificity towards medium and long chain saturated and unsaturated acyl-CoAs, particularly linoleoyl-CoA (18:2-CoA). Its enzymatic function in vivo is evidenced by decreased contents of long chain fatty acids, especially linoleic acid, in mitochondria from livers of Them5^−/− mice. In addition, alterations in cardiolipin synthesis, with increased concentrations of the precursor monolysocardiolipin lead to elongation and enlargement of mitochondria, without associated changes in the mitochondrial DNA copy number [73]. Reduced respiratory capacity and β-oxidation in mitochondria were associated with the age-dependent development of hepatic steatosis, with increased fasting plasma FFA concentrations and hypoketosis [73]. Taken together, these findings illustrate a key role for THEM5 in regulating mitochondrial function.