A Trail of Research from Lipoic Acid to α-Keto Acid Dehydrogenase Complexes

Abstract

pyruvate dehydrogenase α-ketoglutarate dehydrogenase binding protein truncated E2 In this article I shall retrace a trail of research that began with the isolation and characterization of a microbial growth factor and led to elucidation of the structure, function, and regulation of α-keto acid dehydrogenase complexes. The high points of this trail are presented below. This trail of discovery started in the spring of 1949, about 6 months after I joined the faculty of the Department of Chemistry at the University of Texas. At that time I started working on the isolation of a factor that replaced acetate in the growth medium for certain lactic acid bacteria. Research on the “acetate-replacing factor” was initiated by Esmond Snell and associates at the University of Wisconsin and then at the University of Texas. I inherited this project in the spring of 1949. We established that this factor is widely distributed in animal, plant, and microbial cells and that liver is a rich source. The factor is tightly bound to liver protein and is released by proteolysis or by acid hydrolysis. At that time pharmaceutical companies were processing large amounts of pork and beef liver to obtain extracts suitable for treatment of pernicious anemia. The active principle was shown later to be vitamin B12. Fresh liver was extracted with warm water, and the residual liver proteins and fatty material were dried and sold as an animal feed supplement. Arrangements were made with Eli Lilly and Co. to obtain liver residue, and we developed procedures for extracting and purifying the acetate-replacing factor. We progressed to the point of being able to process about 6 pounds of liver residue at a time. A 16,000–50,000-fold purification was achieved. In the late 1940s and early 1950s several other groups were trying to isolate factors that were similar to, if not identical with, the acetate-replacing factor. These factors included the “pyruvate oxidation factor” of I. C. Gunsalus and associates that was necessary for oxidation of pyruvate to acetate and carbon dioxide byStreptococcus faecalis cells grown in a synthetic medium. Gunsalus was also collaborating with Eli Lilly and Co. In the fall of 1950, the Lilly Research Laboratories merged the two separate collaborations to facilitate isolation of the acetate-replacing/pyruvate oxidation factor. The Lilly group adapted and scaled up isolation procedures developed by us. Instead of processing 6-pound batches of liver residue at a time, they were able (using commercial equipment) to process 250-pound batches. Concentrates of the factor that were 0.1–1% pure were sent to my laboratory for further processing. I obtained the first pale yellow crystals of the factor, about 3 mg, on or about March 15, 1951, a truly memorable occasion. It was partially characterized and given the trivial name α-lipoic acid (1Reed L.J. DeBusk B.G. Gunsalus I.C. Hornberger Jr., C.S. Crystalline α-lipoic acid: a catalytic agent associated with pyruvate dehydrogenase..Science. 1951; 114: 93-94Crossref PubMed Scopus (224) Google Scholar). The isolation involved a 300,000-fold purification. A total of ∼30 mg of crystalline lipoic acid was eventually isolated. We estimated that ∼10 tons of liver residue were processed to obtain this small amount of the pure substance. And to think that I was processing about 6 pounds of liver residue at a time, convinced that I would eventually isolate the pure material. We established that lipoic acid is a cyclic disulfide, either 6,8-, 5,8-, or 4,8-dithiooctanoic acid. That the correct structure is 6,8-dithiooctanoic acid (1,2-dithiolane-3-valeric acid) was established by synthesis of dl-lipoic acid, first achieved by E. L. R. Stokstad and associates at Lederle Laboratories. I was intrigued by this simple yet unique substance and wanted to know more about its biological function, i.e. with what and how does it function in living cells. Little did I know then that this trail would take me through five decades of research that has turned out to be a fascinating and rewarding chapter of modern biochemistry, elucidation of the mechanism of oxidative decarboxylation of α-keto acids. Prior to the isolation of lipoic acid and its characterization as a cyclic disulfide, contributions from several laboratories had established the cofactor requirements for the oxidative decarboxylation of α-keto acids represented by the equation shown below. RCOCO2−+CoASH+NAD+→RCOSCoA+CO2+NADHIn addition to CoA and NAD+, thiamin diphosphate, a divalent metal ion, and protein-bound lipoic acid are required. A requirement for FAD was demonstrated later. The presence of a disulfide linkage in lipoic acid led Gunsalus to propose that lipoic acid underwent a cycle of reactions in α-keto acid oxidation comprising reductive acylation, acyl transfer, and electron transfer. Lipoic acid was visualized as functioning after thiamin diphosphate and before CoA and NAD+. Gunsalus, Lowell Hager, and associates obtained evidence for this proposal using lipoic acid and derivatives thereof in substrate amounts. However, the physiological reactions presumably involve catalytic amounts of protein-bound lipoic acid. Elucidation of the nature of the functional form of lipoic acid was essential for verification of the postulated reactions and for further clarification of mechanism. We decided to focus our attention initially on S. faecalis (strain 10C1), which had been grown on a lipoic acid-deficient medium. As shown by Gunsalus and associates these cells did not oxidize pyruvate or α-ketobutyrate unless lipoic acid was added to the preparations. However, attempts to activate cell-free extracts of the lipoic acid-deficient cells with lipoic acid or natural extracts containing complex forms of lipoic acid were unsuccessful. We discovered that cell-free extracts of the deficient cells could be activated by incubation with lipoic acid prior to addition of substrate and supplements. Approximately 30 min of preincubation were required for maximal activation. Activity was reduced only slightly by dialysis, suggesting that lipoic acid was converted to the enzymatically active, “bound” form during the incubation. That such is the case was shown by experiments with lipoic acid-35S2. By fractionation of the lipoic acid-deficient extracts, we determined the components required and the nature of the reactions involved in activation of the extract by lipoic acid. One fraction contained the apopyruvate oxidation system, and a second fraction contained a lipoic acid-activating system. A requirement for ATP was established, suggesting that lipoic acid was activated through its carboxyl group before incorporation into the apopyruvate oxidation system. Based on the results of Paul Berg and others at that time, demonstrating that acyl adenylates are produced by enzyme-catalyzed interaction of organic acids and ATP, we demonstrated that lipoic acid and ATP could be replaced by synthetic lipoyl adenylate. The lipoic acid-activating system, i.e. lipoate-protein ligase, was also detected inEscherichia coli and partially purified. During the course of these studies with cell-free extracts of lipoic acid-deficient S. faecalis, several amides of lipoic acid were synthesized for testing as possible antagonists of the lipoic acid-activating system. These amides did not inhibit activation of the cell-free extract by lipoic acid. On the contrary, these amides could replace lipoic acid, provided ATP was present. These observations suggested that hydrolysis of the amides occurred during incubation with the cell-free extract or the partially purified enzyme preparations. The hydrolytic enzyme, designated lipoyl X-hydrolase and later shown to be a lipoamidase, was purified about 100-fold. The availability of lipoamidase facilitated our preliminary studies in the late 1950s on the biosynthesis of lipoic acid in E. coli. Possible radioactive precursors were included in the growth medium. The PDH1 and KGDH complexes were purified and treated with lipoamidase. The released lipoic acid was extracted into benzene and its radioactivity was determined. We found that octanoate-1-14C was incorporated into lipoic acid as a unit, C-1 of lipoic acid corresponding to C-1 of octanoic acid. Recently, Michael Marletta, John Cronan, and associates have shown that lipoyl synthase, which contains an iron-sulfur cluster, catalyzes the insertion of sulfur at C-6 and C-8 of octanoyl-acyl carrier protein to produce lipoyl-acyl carrier protein. Lipoamidase and lipoate-protein ligase proved to be invaluable in providing direct, unequivocal evidence of the involvement of protein-bound lipoic acid in the CoA- and NAD+-linked oxidative decarboxylation of pyruvate and α-ketoglutarate and in providing clarification of the mechanism of model reactions catalyzed by the pyruvate and α-ketoglutarate dehydrogenase complexes and components thereof. This evidence comprised a demonstration of inactivation and reactivation of the enzyme or enzyme complex accompanying, respectively, release and reincorporation of the lipoyl moiety. When E. coli (Crookes strain) was grown aerobically in the presence of lipoic acid-35S2, the radioactive substance was incorporated into the pyruvate and α-ketoglutarate dehydrogenation systems, due to the presence of lipoate-protein ligase. The availability of the highly purified complexes permitted rapid progress in the late 1950s in identification of the moiety to which lipoic acid is bound. The protein-bound radioactive lipoyl moiety was oxidized with performic acid, and the protein was partially hydrolyzed with 12 n hydrochloric acid (3 h at 105 °C). From the hydrolysates Hayao Nawa (2Nawa H. Brady W.T. Koike M. Reed L.J. Studies on the nature of protein-bound lipoic acid..J. Am. Chem. Soc. 1960; 82: 896-903Crossref Scopus (52) Google Scholar) isolated in good yield a ninhydrin-positive, radioactive conjugate, which was identified as ε-N-(6,8-disulfooctanoyl)-l-lysine by degradation and synthesis. The lipoyl moiety in the two complexes therefore is bound in amide linkage to the ε-amino group of a lysyl residue (Fig. 1). Prior to 1950 pyruvate and α-ketoglutarate oxidation had been studied mainly with particulate preparations that were unsuitable for detailed analysis. Solubilization of bacterial and animal α-keto acid oxidation systems in the early 1950s in the laboratories of Severo Ochoa and David Green was a significant advance. Seymour Korkes, Gunsalus, and Ochoa succeeded in separating the pyruvate oxidation system of anaerobically grown E. coli (strain ATCC 4157) into two components, designated Fraction A and Fraction B. Subsequently, Hager and Gunsalus found that extracts of aerobically grown E. coli (Crookes strain) contained 30–50 times the pyruvate oxidation activity of the anaerobically grown 4157 cells. Using lipoic acid and dihydrolipoic acid in substrate amounts they showed that Fraction A contained a lipoyl transacetylase and that Fraction B contained a lipoyl dehydrogenase. Richard Schweet and associates isolated a CoA- and NAD+-linked pyruvate oxidation system from pigeon breast muscle in a highly purified state, with an apparent molecular weight of about 4 million. D. R. Sanadi and associates isolated a CoA- and NAD+-linked α-ketoglutarate oxidation system from pig heart with an apparent molecular weight of 2 million. In my laboratory we developed mild procedures for purification of the pyruvate and α-ketoglutarate oxidation systems from E. coli (Crookes strain). By the late 1950s Masahiko Koike (3Koike M. Reed L.J. Carroll W.R. α-Keto acid dehydrogenation complexes. I. Purification and properties of pyruvate and α-ketoglutarate dehydrogenation complexes of Escherichia coli..J. Biol. Chem. 1960; 235: 1924-1930Abstract Full Text PDF PubMed Google Scholar) succeeded in isolating these enzyme systems as highly purified functional units with molecular weights in the millions. It was very exciting to see in the analytical ultracentrifuge of my friend and collaborator at NIH, William Carroll, a major symmetrical peak for each of the two highly purified preparations and that the boundary of the yellow color of the flavoprotein was associated with the main peak. The molecular weights of these multienzyme units were determined to be 4.8 and 2.4 million, respectively. By careful and persistent work over a period of several years, we dissected the pyruvate and α-ketoglutarate dehydrogenase complexes into their component enzymes, characterized them, and reassembled the large functional units from the isolated enzymes (4Koike M. Reed L.J. Carroll W.R. α-Keto acid dehydrogenation complexes. IV. Resolution and reconstitution of the Escherichia coli pyruvate dehydrogenation complex..J. Biol. Chem. 1963; 238: 30-39Abstract Full Text PDF PubMed Google Scholar). We demonstrated that the individual enzymes are linked in the two complexes by non-covalent bonds and that by proper selection of experimental conditions the enzymes could be separated from one another without loss of enzymatic activity. We showed that each of these functional units is composed of multiple copies of three enzymes, a pyruvate and an α-ketoglutarate decarboxylase-dehydrogenase (E1), a dihydrolipoamide acetyltransferase and a succinyltransferase (E2), and a flavoprotein, dihydrolipoamide dehydrogenase (E3). These three enzymes, acting in sequence, catalyze the reactions shown in Fig.2. E1 catalyzes both the decarboxylation of the α-keto acid (Reaction 1) and the subsequent reductive acylation of the lipoyl moiety, which is covalently bound to E2 (Reaction 2).E2 catalyzes the acyl transfer to CoA (Reaction 3), and E3 catalyzes the reoxidation of the dihydrolipoyl moiety with NAD+ as the ultimate electron acceptor (Reactions 4 and 5). Binding experiments showed that the pyruvate dehydrogenase (E1) and the flavoprotein (E3) do not combine with each other, but each of these components does combine with the acetyltransferase (E2). The acetyltransferase serves a dual function, a catalytic function and a structural function,i.e. a scaffold for binding and localizingE1 and E3. In dilute acetic acid (0.83 m, pH 2.6) the acetyltransferase dissociated into inactive subunits with a molecular weight of about 70,000. Dilution of the acidic solution into suitable buffers resulted in restoration of enzymatic activity and the characteristic structure of the native acetyltransferase unit. The acetyltransferase appeared to be a self-assembling system. The two flavoproteins (from the PDH and KGDH complexes) were shown to be interchangeable with respect to both complex formation and function, and enzymatic, physical, and immunochemical data indicated that the two flavoproteins were very similar if not identical. It was evident that the lipoyl moiety undergoes a cycle of transformations, i.e. reductive acylation, acyl transfer, and electron transfer, involving three separate enzymes within a complex in which movement of the individual enzymes is restricted and from which intermediates do not dissociate. A possible molecular basis of these interactions emerged from our discovery that the lipoyl moiety is bound in amide linkage to the ε-amino group of a lysyl residue in the E2 component of the PDH and KGDH complexes. This linkage provides a flexible arm, about 14 Å in length (Fig. 1), conceivably permitting the lipoyl moiety to rotate among the active sites of E1, E2, andE3, i.e. a “swinging arm” active site coupling mechanism. Some 15 years later, spin label experiments by Richard Perham and Cees Veeger and their associates provided evidence that the lipoyllysyl residues are essentially free to rotate in the PDH complex. Our subsequent finding (see below) that the lipoyllysyl moiety is part of a “super arm,” i.e. a lipoyl domain, led to the proposal that movement of lipoyl domains as well as rotation of lipoyl moieties may provide the means to span the physical gaps between catalytic sites on E1,E2, and E3, as well as facilitate communication between lipoyl moieties. These were exciting times for us in the late 1950s and early 1960s. Our concept of the macromolecular organization of the PDH complex that emerged from these biochemical studies is that of an organized mosaic of enzymes in which each of the component enzymes is uniquely located to permit efficient coupling of the individual reactions catalyzed by these enzymes. This concept was confirmed and extended by electron microscopy studies conducted by my associate Robert Oliver. Electron micrographs of the E. coli PDH complex and its component enzymes negatively stained with phosphotungstate revealed that the complex had a polyhedral structure with a diameter of about 300 Å, that the acetyltransferase (E2) occupied the center of the polyhedron, and that the molecules of E1 andE3 were distributed on its surface. The shape of the acetyltransferase indicated that it had a cubelike structure. The shape of the succinyltransferase component of the E. coliKGDH complex was very similar. These results, together with biochemical data, demonstrated that both E2s consist of 24 apparently identical polypeptide chains arranged as eight trimers (morphological subunits) at the vertices of a cube (Fig.3A). This proposed structure was confirmed later by x-ray diffraction analyses carried out by collaborators David DeRosier and Marvin Hackert demonstrating that both acyltransferases possess 432 molecular symmetry. Our interpretative model of the macromolecular organization of the E. coli PDH complex in the mid-1960s depicted 12E1 dimers and 6 E3 dimers arranged, respectively, on the 12 edges and in the 6 faces of the cubelike E2. All dihydrolipoamide acyltransferases possess a unique multidomain structure. This architectural feature was revealed initially in my laboratory in the late 1970s by limited proteolysis studies of the E. coli dihydrolipoamide acetyltransferase containing [2-3H]lipoyl moieties. Dennis Bleile (5Bleile D.M. Munk P. Oliver R.M. Reed L.J. Subunit structure of dihydrolipoyl transacetylase component of pyruvate dehydrogenase complex from Escherichia coli..Proc. Natl. Acad. Sci. U. S. A. 1979; 76: 4385-4389Crossref PubMed Scopus (82) Google Scholar) found that limited tryptic digestion at pH 7.0 and 4 °C cleaved theE2 subunits (Mr∼64,500) into two large fragments, an outer lipoyl-bearing domain and an inner catalytic and subunit binding domain. The latter fragment (Mr ∼29,600) had a compact structure, and it possessed the intersubunit binding sites of the acetyltransferase, the binding sites for E1 andE3, and the catalytic site for acetyl transfer. The assemblage of compact catalytic and subunit binding domains constitutes the inner core of the acetyltransferase, conferring the cubelike appearance of this E2 seen with the electron microscope. The other tryptic fragment (Mr ∼31,600), designated the lipoyl domain, contained the covalently bound lipoyl moieties and had an extended structure. We suggested that the two domains are connected by a trypsin-sensitive hinge region and that movement of lipoyl domains and not simply rotation of lipoyllysyl moieties may provide the means to span the physical gaps between catalytic sites on the complex. These early findings on the domain structure of dihydrolipoamide acyltransferases were confirmed and extended by studies involving molecular genetics, limited proteolysis, and proton NMR spectroscopy in the laboratories of John Guest and Richard Perham. Briefly, the amino-terminal part of the acyltransferases contains one, two, or three highly similar lipoyl domains, each of about 80 amino acid residues. The lipoyl domain (or domains) is followed by another structurally distinct segment that is involved in binding E3and/or E1. These domains are linked to each other and to the carboxyl-terminal part of the polypeptide chain (catalytic domain) by flexible segments (hinge regions) that are rich in alanine, proline, and charged amino acid residues. These segments are thought to provide flexibility to the lipoyl domains, facilitating active site coupling within the multienzyme complexes. In the late 1960s part of our research effort was directed toward isolation and characterization of the mammalian PDH and KGDH complexes, which are localized to mitochondria within the inner membrane-matrix compartment. Procedures were developed for preparation of mitochondria on a large scale from bovine kidney and heart (with the advice and assistance of my friend and colleague, Daniel Ziegler), and relatively mild procedures were developed to isolate the PDH and KGDH complexes from the mitochondrial extracts. In the course of attempts to stabilize these complexes in crude extracts of bovine kidney mitochondria, Tracy Linn observed that the PDH complex, but not the KGDH complex, underwent a time-dependent inactivation in the presence of ATP. A systematic investigation revealed that the bovine kidney and heart PDH complexes are regulated by a phosphorylation-dephosphorylation cycle (6Linn T.C. Pettit F.H. Reed L.J. α-Keto acid dehydrogenase complexes. X. Regulation of the activity of the pyruvate dehydrogenase complex from beef kidney mitochondria by phosphorylation and dephosphorylation..Proc. Natl. Acad. Sci. U. S. A. 1969; 62: 234-241Crossref PubMed Scopus (514) Google Scholar). Phosphorylation and concomitant inactivation of the complex is catalyzed by an ATP-dependent kinase, which is tightly bound to the complex, and dephosphorylation and concomitant reactivation are catalyzed by a Mg2+-dependent phosphatase, which is loosely attached to the complex. It seemed curious at the time (1968) that inactivation of the PDH complex by phosphorylation had not been detected earlier. The explanation may lie in a remark by Henry Lardy after receiving a preprint of our paper on the phosphorylation and inactivation of the PDH complex. (This finding) “explains why we have never been able to get pyruvate to be oxidized in submitochondrial particles, because we invariably add ATP to keep things in the ‘optimum’ state.” This control mechanism was subsequently confirmed in the laboratories of Otto Wieland, Philip Randle, S. E. Severin, and other investigators with preparations of the PDH complex from other mammalian tissues and from pigeon breast muscle, plant tissue, andNeurospora crassa. In the early 1960s the three known examples of enzyme regulation by phosphorylation-dephosphorylation were phosphorylase, phosphorylase kinase, and glycogen synthase. Our results with the mammalian PDH complex indicated that this regulatory mechanism is more general than had been recognized previously. Edwin Krebs mentioned in the fourteenth Hopkins Memorial Lecture (1984) that in the early 1960s some investigators wondered whether “this was an esoteric type of control system restricted to one limited area of carbohydrate metabolism … with the finding from Lester Reed's laboratory that pyruvate dehydrogenase is regulated by phosphorylation-dephosphorylation the field broke out of the more restricted area.” Over a period of several years our group separated the bovine kidney and heart PDH complexes into their component enzymes (E1 (α2β2),E2, E3, PDH kinase, and PDH phosphatase) and characterized the individual enzymes (7Linn T.C. Pelley J.W. Pettit F.H. Hucho F. Randall D.D. Reed L.J. α-Keto acid dehydrogenase complexes. XV. Purification and properties of the component enzymes of the pyruvate dehydrogenase complexes from bovine kidney and heart..Arch. Biochem. Biophys. 1972; 148: 327-342Crossref PubMed Scopus (192) Google Scholar). We showed that the E1α subunit undergoes phosphorylation on three seryl residues. We were surprised by the appearance in the electron microscope of negatively stained preparations of the mammalian dihydrolipoamide acetyltransferase (E2). Its morphological subunits appeared to be located at the vertices of a pentagonal dodecahedron instead of at the vertices of a cube. Thus, our electron microscope studies revealed that there are two distinct polyhedral forms of E2, the cube and the pentagonal dodecahedron (Fig. 3). The former design, a cube, consists of 24 E2 subunits (8 trimers) arranged with octahedral (432) symmetry. This design is exhibited by the E2 components of the E. coli PDH and KGDH complexes and by the E2 components of the mammalian KGDH and branched chain α-keto acid dehydrogenase complexes. The latter design, a pentagonal dodecahedron, consists of 60E2 subunits (20 trimers) arranged with icosahedral (532) symmetry. E2 components of this morphology are found in the PDH complexes from mammalian and avian tissues, fungi, and the Gram-positive bacterium Bacillus stearothermophilus. A morphological unit consisting of threeE2 subunits appeared to be important in the assembly of both types of polyhedral forms. These conclusions were confirmed and extended by results from x-ray diffraction analysis by collaborators David DeRosier and Marvin Hackert and by Wim Hol and associates. The bovine heart PDH complex has a molecular weight of about 9.5 million. Its subunit composition is 60 E2subunits, ∼30 E1 tetramers (α2β2), and 12 E3dimers, which are positioned on the E2 core by 12 E3-binding protein (protein X) monomers (see below). We proposed that the E1 tetramers are located on the 30 edges and the E3 dimers in the 12 faces of the E2 pentagonal dodecahedron. Hormonal regulation of the mammalian PDH complex is particularly fascinating because it involves signal transduction not only across the cell membrane but also across the inner mitochondrial membrane to target the PDH phosphatase and, consequently, the PDH complex, located in the mitochondrial matrix. It is now known that the major regulators of the phosphatase activity are Ca2+ and Mg2+, which involve the hormones epinephrine and insulin, respectively. In the early 1970s our group partially purified PDH phosphatase from bovine heart and kidney mitochondria and showed that it requires Mg2+ or Mn2+ for activity. Richard Denton, Philip Randle, and associates subsequently reported that Ca2+ stimulates the activity of the phosphatase in the presence of Mg2+. Flora Pettit and Thomas Roche in my group showed that Ca2+ mediates translocation of the phosphatase to the E2 component of the PDH complex, presumably in proximity to its substrate, phosphorylatedE1, thereby increasing the rate of dephosphorylation. This Ca2+-mediated translocation apparently is the molecular basis of the epinephrine-induced activation of PDH phosphatase observed by other investigators. In my laboratory in the early 1980s Martin Teague, Flora Pettit, and co-workers purified PDH phosphatase to near homogeneity and showed that it consists of a Mg2+-dependent and Ca2+-stimulated catalytic subunit (50 kDa; PDPc) and a flavoprotein of unknown function (100 kDa; later designated PDPr). Zahi Damuni showed that polyamines, particularly spermine, increase the sensitivity of PDH phosphatase to Mg2+. Richard Denton and associates subsequently showed that insulin stimulates the activity of PDH phosphatase in adipose tissue by increasing the sensitivity of the phosphatase to Mg2+. Spermine apparently mimics the insulin effect. The function of PDPr remained a mystery until Janet Lawson in the early 1990s cloned and expressed cDNA encoding PDPc. By comparing the properties of recombinant PDPc and the native PDH phosphatase heterodimer (PDPc bound to PDPr), we obtained insight into the function of PDPr. Jiangong Yan (8Yan J. Lawson J.E. Reed L.J. Role of the regulatory subunit of bovine pyruvate dehydrogenase phosphatase..Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 4953-4956Crossref PubMed Scopus (32) Google Scholar) found that PDPr decreases the sensitivity of PDPc to Mg2+ and that spermine increases the sensitivity of PDH phosphatase but not PDPc to Mg2+, apparently by interacting with PDPr. We interpreted these observations to indicate that PDPr blocks or distorts the Mg2+-binding site of PDPc and that spermine produces a conformational change in PDPr (allosteric effect) that reverses its inhibitory effect. These observations raise the intriguing prospect that an insulin-induced allosteric effect on PDPr may underlie its stimulation of PDH phosphatase activity. To gain further understanding of structure-function relationships in eukaryotic PDH complexes, we initiated in the late 1980s molecular genetics studies of the PDH complex in the yeastSaccharomyces cerevisiae. The genes encoding the five proteins comprising the complex (E1α,E1β, E2, BP, andE3) were cloned, sequenced, expressed, and disrupted. Studies on E3-binding protein (BP) confirmed and extended previous studies of Thomas Roche and of Gordon Lindsay and their associates with the protein X component of the bovine PDH complex. BP and E2 apparently evolved from a common ancestor. BP possesses an amino-terminal lipoyl domain, followed by an E3-binding domain, and then by a carboxyl-terminal domain that is involved in anchoring BP to the inner core of E2. The availability of recombinant BP,E3, and the inner core ofE2 (truncated E2(tE2)) provided an opportunity to elucidate the binding stoichiometry and localization of BP and BP-E3complex on tE2. In the 1990s cryoelectron microscopy and three-dimensional image reconstruction (three-dimensional electron microscopy) in collaboration with James Stoops and Timothy Baker and their associates revealed a unique structural organization of the tE2·BP-E3 complex (9Stoops J.K. Cheng R.H. Yazdi M.A. Maeng C-Y. Schroeter J.P. Klueppelberg U. Kolodziej S.J. Baker T.S. Reed L.J. On the unique structural organization of the Saccharomyces cerevisiae pyruvate dehydrogenase complex..J. Biol. Chem. 1997; 272: 5757-5764Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar). As revealed previously by x-ray crystallography (Wim Hol and associates) and by three-dimensional electron microscopy, the 60-mer tE2consists of 20 cone-shaped trimers arranged at the vertices of a pentagonal dodecahedron (Fig. 4). The 20 trimers are connected by 30 bridges to form an empty cagelike structure, with the tip of the trimer directed toward the center of the cage. Our results showed that BP binds near the tips of theE2 trimers within the central cavity and anchors an E3 dimer inside each of the 12 pentagonal faces. The unusual finding that the geometric constraints of the tE2 scaffold determine the extent and disposition of BP-E3 binding provides a satisfactory explanation of the observation that tE2 binds a maximum of 12 copies of BP-E3 complex. Three-dimensional electron microscopy of wild-type PDH complex from S. cerevisiae andE2·E1 subcomplex thereof shows that the disposition of E1tetramers around the E2 core is also restricted. Research on the structural organization of S. cerevisiae PDH complex is continuing. Recent three-dimensional electron microscopy studies reveal that individual molecules of tE2 exhibit an unusual size variability of 20% (10Zhou Z.H. Liao W. Cheng R.H. Lawson J.E. McCarthy D.B. Reed L.J. Stoops J.K. Direct evidence for the size and conformational variability of the pyruvate dehydrogenase complex revealed by three-dimensional electron microscopy..J. Biol. Chem. 2001; 276: 21704-21713Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar). We have proposed that expansion and contraction of the 60-mer core is thermally driven and that protein dynamics is an integral component of the function of the PDH multienzyme complex. I hope these reflections have given some appreciation of the thrill and excitement I have experienced in establishing this trail of research from lipoic acid to the structure, function, and regulation of the α-keto acid dehydrogenase complexes. I have been accompanied in the various stages of this journey by excellent associates, including undergraduate, graduate, and postdoctoral students, technicians, and members of the senior staff of the Biochemical Institute, and by collaborators at other universities and institutes. I am pleased to acknowledge the Clayton Foundation for Research and the National Institutes of Health for generous financial support. Denton, R. M., McCormack, J. G., Midgley, P. J. W., and Rutter, G. A. (1987) Hormonal regulation of fluxes through pyruvate dehydrogenase and the citric acid cycle in mammalian tissues. Biochem. Soc. Symp.54, 127–143 Guest, J. R., Angier, S. J., and Russell, G. C. (1989) Structure, expression, and protein engineering of the pyruvate dehydrogenase complex of Escherichia coli. Ann. N. Y. Acad. Sci.573, 76–99 Gunsalus, I. C. (1954) Group transfer and acyl-generating functions of lipoic acid derivatives. in The Mechanism of Enzyme Action (McElroy, W. D., and Glass, B., eds) pp. 545–580, Johns Hopkins University Press, Baltimore Perham, R. N. (2000) Swinging arms and swinging domains in multifunctional enzymes: catalytic machines for multiple reactions.Annu. Rev. Biochem.69, 961–1004 Randle, P. J. (1986) Fuel selection in animals. Biochem. Soc. Symp.14, 799–806 Reed, L. J. (1957) The chemistry and function of lipoic acid.Adv. Enzymol.18, 319–347 Reed, L. J. (1974) Multienzyme complexes. Acc. Chem. Res.7, 40–46 Roche, T. E., and Patel, M. S. (eds) (1989) Alpha-keto acid dehydrogenase complexes: organization, regulation, and biomedical ramifications. Ann. N. Y. Acad. Sci.573,1–462 Roche, T. E., Liu, S., Ravindran, S., Baker, J. C., and Wang, L. (1996) Role of the E2 core in the dominant mechanisms of regulatory control of mammalian pyruvate dehydrogenase complex. inAlpha-Keto Acid Dehydrogenase Complexes (Patel, M. S., Roche, T. E., and Harris, R. A., eds) pp. 33–52, Birkhaüser Verlag, Basel, Switzerland Wieland, O. H. (1983) The mammalian pyruvate dehydrogenase complex: structure and regulation. Rev. Physiol. Biochem. Pharmacol.96, 123–170