Lipoic acid metabolism and mitochondrial redox regulation

Introduction

Lipoic acid (6,8-dithiooctanoic acid) was first identified by Reed and co-workers (1, 3) and Jukes and co-workers (2) in the 1950s, and over the last several decades much has been learned about the chemical properties and biological functions of this unique cofactor. The disulfide bond within the molecule provides a source of reductive potential that is required for catalysis by mitochondrial 2-ketoacid dehydrogenases and participates in stabilization and redox-dependent regulation of these multienzyme complexes. These functions make lipoic acid essential for cell growth, oxidation of carbohydrates, amino acids, and other fuels, and regulation of mitochondrial redox balance. Herein, we discuss the intricacies of lipoic acid synthesis in various organisms, the role lipoylation plays in 2-ketoacid dehydrogenase activities and regulation, and how these enzymes are influenced by reactive oxygen species through this covalently bound lipoic acid cofactor.

Lipoic acid synthesis

Lipoic acid (LA)² metabolism has been intensely investigated by Reed (1), Cronan (4, 5), and others through genetic and biochemical studies using prokaryotes and Saccharomyces cerevisiae (6). Homologous enzymes have been identified in plants (7, 8) and mammalian systems (9). That being said, distinct differences in LA metabolism among various organisms are relevant to the regulation of 2-ketoacid dehydrogenases. LA metabolism in Escherichia coli consists of a de novo biosynthetic pathway and a salvage pathway capable of using exogenous LA as a substrate. Both pathways lead to the covalent modification of the ϵ-amino group on conserved lysine residues within the lipoyl domain of the E2 subunits of 2-ketoacid dehydrogenase complexes (Fig. 1A) (4).

The biosynthetic pathway in E. coli is initiated by an octanoyltransferase enzyme (octanoyl-(ACP:protein)-protein N-octanoyltransferase), LipB, that transfers octanoic acid derived from mitochondrial fatty-acid synthesis type II (FASII) from the acyl carrier protein (ACP) to the lipoyl domain of a target enzyme. In the first step of LA biosynthesis, LipB transfers octanoate directly onto the lipoyl domain on the E2 subunit of 2-ketoacid dehydrogenases (Fig. 1A, reaction B1) (10–14). E. coli tolerates loss of this enzyme when exogenous lipoate is present through activity of the salvage pathway (discussed below), but in the absence of exogenous lipoate, growth is suppressed, and no lipoylation of pyruvate dehydrogenase (PDH-E2) or α-ketoglutarate dehydrogenase (OGDH-E2) is observed (15).

In the second step of LA biosynthesis, two sulfur atoms are inserted at the C6 and C8 positions by an iron–sulfur lipoic acid synthase enzyme, LipA (Fig. 1A, reaction B2) (15–17). This step is similar to biotin synthesis in that the reaction mechanism requires sulfur donation from a [4Fe-4S] cluster within the enzyme and reductive cleavage of S-adenosylmethionine (SAM) to generate 5′-deoxyadenosyl 5′-radicals that remove a hydrogen atom from C6 and C8 and facilitate insertion of the sulfur atoms (5, 18, 19). The requirements for this enzyme are similar to that of LipB in that E. coli can utilize the LA salvage pathway to compensate for loss of LipA in lipoate-containing media (15). Importantly, donation of sulfur from the [4Fe-4S] cluster within the enzyme originally suggested that LipA may be a self-sacrificing protein rather than performing true enzyme catalysis (5). This had been supported by the small amount of LA generated per molecule of LipA in E. coli, which could be enhanced by co-expression of LipA with iron–sulfur cluster proteins (5, 20). However, recently, the E. coli iron–sulfur cluster proteins NfuA and IscU were shown to reinstall the [4Fe-4S] cluster of LipA to facilitate additional turns of the enzyme (21). This mechanism is consistent with deficiencies in NFU1 resulting in phenotypes associated with lipoic acid deficiency and suggests that this mechanism may be conserved in mammals (22).

The LA salvage pathway in E. coli consists of a single lipoyl-protein ligase enzyme, LplA, that first conjugates exogenous lipoic acid to an adenylate intermediate (lipoyl-AMP) followed by ligation to the lipoyl domain of E2 subunits and H-protein (Fig. 1A, reaction S1) (23–25). LplA can use both lipoic acid and octanoate to modify E2 subunits in intact cells, with the latter substrate requiring the activity of LipA to insert sulfur atoms and generate the lipoic acid moiety directly on target enzymes (23, 26). E. coli can tolerate the loss of this enzyme, but loss of LplA and LipB results in no lipoylated proteins, whereas loss of LplA and LipA results in the accumulation of octanoyl proteins (26). These data indicate that there are two distinct LA metabolism pathways in E. coli that ensure growth in both lipoate-containing and -deficient environments (11, 26).

LA metabolism in S. cerevisiae is different from E. coli primarily in that these two pathways are interdependent and cannot fully compensate for one another (27), and lipoylation of PDH-E2 (Lat1) and OGDH-E2 (Kgd2) requires initial lipoylation of the H-protein of the glycine cleavage system (Gcv3) (Fig. 1B, reaction 1) (6). This pathway is similar to Bacillus subtilis, where the octanoyltransferase in S. cerevisiae, Lip2, transfers octanoate (derived from FASII) to Gcv3, and this step is required for lipoylation of Lat1 and Kgd2 (6). Importantly, activity of the glycine cleavage system (GCS) is not required for downstream lipoylation events because loss of other GCS subunits has no impact on Lat1 or Kgd2 lipoylation (6). The lipoate synthase enzyme in S. cerevisiae is Lip5, which functions similarly to LipA in E. coli through iron–sulfur cluster-mediated insertion of the disulfide at C6 and C8 (Fig. 1B, reaction 2) (28). Yeast cannot tolerate the loss of Lip2 or Lip5 unless grown in medium containing ethanol and succinate, which bypasses Kgd2 in the TCA cycle (6).

The inability of Lip2 and Lip5 mutants to grow on media supplemented with lipoic acid was the first indication that an independent LA salvage pathway did not exist in S. cerevisiae (29). However, identification of an LplA homolog in yeast, Lip3, and the observation that lipoylation of Gcv3 was maintained in Lip3 mutants indicated that Lip3 functions downstream of Lip2 (Fig. 1B, reaction 3). In vitro studies have demonstrated that Lip3 is an octanoyltransferase, using octanoyl-CoA or octanoyl-Gcv3 as a substrate, but the enzyme lacks the ability to utilize lipoate plus ATP to generate the adenylate intermediate seen with LplA (27). Expression of E. coli LplA in yeast strains lacking Lip2 or Lip5 completely rescues growth in the presence of lipoate indicating that the lack of a true LA salvage pathway in S. cerevisiae is an aspect of the differential activities of LplA and Lip3 (27). Questions as to the true substrate of Lip3 in vivo are still unanswered from these studies. Does Lip3 transfer an octanoyl moiety from Gcv3 to Lat1 and Kgd2, where Lip5 then generates the lipoyl moiety, or does Lip3 act as a lipoyltransferase, transferring a lipoyl moiety from Gcv3? The maintenance of Gcv3 lipoylation in Lip3 mutant strains indicates that octanoyl-Gcv3 is a substrate for Lip5 and that either Lip2 does not transfer an octanoyl moiety from ACP to Lat1 and Kgd2 or that Lip5 cannot act on octanoyl-Lat1/Kgd2. This is particularly interesting because expression of yeast Lip3 in E. coli ΔlipB ΔlplA allows for growth in the presence of octanoate but not lipoate. Furthermore, in octanoate-containing media, E. coli ΔlipB ΔlplA expressing Lip3 has lipoylated PDH and OGDH (27). As well, lipoylation was enhanced in the presence of octanoate to a greater degree than with lipoate and on PDH more so than OGDH (27). These data suggest that in an E. coli system, Lip3 is an octanoyltransferase and that LipA can act on octanoyl-PDH/OGDH, but this has not been demonstrated in the native cellular environment of Lip3.

Although less well-understood in mammalian systems, the LA biosynthetic pathway in mice and humans is carried out by an octanoyltransferase ortholog of LipB/Lip2 and a lipoic acid synthase ortholog of LipA/Lip5 known as LIPT2 and LIAS, respectively (Fig. 1C) (30–32). Deficiencies in either of these enzymes, as well as disruptions in mitochondrial FASII or iron–sulfur biogenesis, result in diminished lipoylation of PDH and OGDH and ultimately impaired mitochondrial function (30, 33, 34). LIPT1, the lipoyltransferase ortholog in mammals, is similar to LIP3 in S. cerevisiae in that it lacks the ability to generate an activated lipoyl-AMP intermediate and therefore is thought to be downstream of LIPT2 (9, 30, 35). There has been a report identifying a mammalian lipoic acid-activating enzyme that could activate exogenous lipoic acid (36); however, this function was ultimately attributed to the mitochondrial medium-chain acyl-CoA synthetase (ACSM1) (37, 38). This enzyme can utilize both the (R)- and (S)-enantiomers of LA and primarily uses GTP to activate the natural (R)-lipoic acid, but so far there has been no substantial evidence to support that this enzyme functions in LA metabolism in vivo (36). This is consistent with the inability for exogenous LA to rescue defects in cells derived from LIAS-deficient patients, embryonic lethality in LIAS-deficient mice, or to ameliorate symptoms in patients with this disease (22, 30, 32). Taken together, this suggests that mammalian LA metabolism is similar to S. cerevisiae where LIPT2 transfers octanoate from ACP to the H-protein of GCS, LIAS inserts sulfur atoms into the octanoyl group on H-protein, and LIPT1 transfers the lipoyl group from the H-protein to E2 subunits. It is unclear at present whether LIPT1 is an octanoyltransferase, a lipoyltransferase, or both, as its true in vivo substrate remains to be clarified.

Lipoic acid as a cofactor for mitochondrial 2-ketoacid dehydrogenases

As discussed above, the function of LA metabolism is to provide an essential, enzyme-bound cofactor for mitochondrial 2-ketoacid dehydrogenases and the GCS. The 2-ketoacid dehydrogenases include pyruvate dehydrogenase (PDH), α-ketoglutarate dehydrogenase (OGDH), branched-chain ketoacid dehydrogenase (BCKDH), and 2-oxoadipate dehydrogenase (OADH) (30, 39). These multienzyme complexes are composed of three independent subunits that undergo coupled reactions facilitated by the LA cofactor (Fig. 2) (40). The E1 subunit, which provides substrate specificity to the multienzyme complex, utilizes the cofactor thiamine pyrophosphate (TPP) to decarboxylate the substrate generating an acyl-TPP intermediate followed by reductive acylation of the lipoyl group on the E2 subunit. The E2 subunit functions as a dihydrolipoyl acyltransferase that transfers the acyl group to CoA, generating an acyl-CoA and dihydrolipoamide. The E3 subunit is a dihydrolipoamide dehydrogenase (DLD) using FAD to oxidize dihydrolipoamide, regenerating the disulfide bond for use in subsequent rounds of catalysis. The E3 subunit oxidizes FADH₂ back to FAD using NAD⁺ producing NADH in the process (40) (Fig. 2A). This activity is similar for PDH, OGDH, and BCKDH, where the E1 and E2 subunits are specific to the decarboxylation substrate and the E3 protein subunit is shared among all three complexes (4). The activity of 2-OADH has not been investigated to the degree of these other complexes, but elevated 2-oxoadipate has been reported in patients with DLD deficiency, suggesting that this complex uses the same DLD subunit shared by other 2-ketoacid dehydrogenases (41).

Importantly, lipoylated 2-ketoacid dehydrogenases play a key role in carbon entry into the TCA cycle. Therefore, dysfunction in these enzymes may produce aberrant mitochondrial metabolism that can be deleterious (Fig. 3). Inborn errors have been reported in each of these enzyme complexes with mutations having been reported in all of the subunits.

Pyruvate dehydrogenase deficiency is rare, but hundreds of cases have been reported, most involving mutations in the X-linked gene encoding the E1-α1 subunit (PDHA1). Severe mutations are devastating in boys but cause a more variable spectrum of severity in females due to random X-inactivation. Broadly, symptoms of PDH deficiency include lactic acidosis, hypotonia, seizures, ataxia, and developmental delay. Treatment for PDH deficiency is limited; the use of a ketogenic diet and in some cases treatment with dichloroacetate or thiamine can provide improvement of clinical symptoms (42).

α-Ketoglutarate dehydrogenase deficiency has been reported less frequently than PDH deficiency, but the primary clinical manifestations are similar, including developmental delay, ataxia, hypotonia, and in some cases encephalopathy (42). The elevation of α-ketoglutarate levels may also be associated with an elevation in 2-hydroxyglutarate, a potential epigenetic regulator. It is currently unknown whether 2-hydroxyglutarate levels contribute to pathology in α-ketoglutarate dehydrogenase deficiency (43, 44).

Deficiency in the E1 or E2 subunit of branched-chain ketoacid dehydrogenase is also known as maple syrup urine disease. This disorder includes a classic presentation of a sweet, maple-syrup like odor and an elevation of plasma branched-chain amino acids and 2-ketoacids in urine. In the severe neonatal-onset form, patients display metabolic decompensation and neurological distress, which may be severe and result in neonatal coma. There are also acute and intermittent forms that may be late-onset and consist of recurrent metabolic decompensation episodes; as well, the chronic and progressive form presents with hypotonia and developmental delay. Treatment of maple syrup urine disease involves titrating dietary branched-chain amino acid content to avoid excessive exposure while maintaining sufficient levels to support normal development (42).

2-Oxoadipic aciduria has only been reported in about 20 individuals with half of them being asymptomatic. Those individuals that do show symptoms may display psychomotor retardation and hypotonia. Mutations in dehydrogenase E1 and transketolase domain containing 1 (DHTKD1) are thought to be causative in these cases (42).

Mutations have also been reported in DLD, LIAS, LIPT2, and LIPT1. These disorders have overlapping phenotypes, including lactic acidosis, developmental delay, and seizures. Collectively, dysfunction in these enzymes falls into a larger class of mitochondrial disease known as Leigh syndrome. LIAS patients may be distinguished from LIPT1 patients through presentation of nonketotic hyperglycinemia associated with dysfunction in the GCS. An excellent review of the clinical manifestations of these disorders was recently described by Mayr et al. (30). Critical to developing treatments for these disorders is intimate knowledge of the regulation of 2-ketoacid dehydrogenase complexes and how their activities may differ in various nutritional states and in individual tissues.

Lipoic acid and ROS generation from mitochondrial 2-ketoacid dehydrogenases

Concluding remarks

Lipoic acid is an often overlooked, essential cofactor for mitochondrial oxidative metabolism, which participates in catalysis and regulation of multiple enzyme complexes. Biosynthesis of LA links mitochondrial fatty-acid synthesis, SAM, and iron–sulfur cluster biosynthesis with TCA cycle oxidative capacity. This information is particularly important for diagnosis of inborn errors of metabolism because defects in lipoic acid metabolism can promote a myriad of clinical symptoms associated with dysfunction in a number of enzyme complexes (22, 30, 31, 84). Additionally, because lipoic acid plays a regulatory role, symptoms of some inborn errors may only manifest, or may worsen, in nutritional states associated with high flux through enzymes requiring lipoylation. The lack of an independent salvage pathway in humans abrogates the use of LA supplementation as a therapeutic option; thus, further investigation of how LA metabolism is regulated and functions in humans will be necessary to treat inborn errors of this pathway.