Inactivation of the 2-ketoglutarate and pyruvate dehydrogenase complexes of beef heart by branched chain keto acids.

Incubation of 2-ketoglutarate dehydrogenase complex with 2-ketoisovalerate, 2-keto-4-methylvalerate, or 2-keto-3-methylvalerate leads to the appearance of a lag phase and of a progressive loss of activity in subsequent measurements of the initial rate of oxidation of 2-ketoglutarate. In the case of 2-ketoisovalerate these effects are shown to be due to the formation of an isobutyryllipoate derivative of the enzyme, as a result of the very slow oxidation of 2-ketoisovalerate by the enzyme complex (Vmax congruent to 0.15% of that for 2-ketoglutarate). Incubation of the enzyme complex with 2-keto[14C]isovalerate or 2-keto[14C]glutarate results in comparable incorporation of radioactivity, amounting to 3.5 to 5.3 nmol of isobutyryl or succinyl residues per mg of protein in the complex. Isobutyryl residues are also incorporated in the enzyme during the simultaneous oxidation of both of these substrates. During the early phase of incubation of the complex with 2-ketoisovalerate the incorporation of isobutyryl residues is much faster than the loss of enzyme activity. This observation seems to support the suggestion that each 2-ketoglutarate decarboxylase subunit of the complex may catalyze the succinylation of more than one lipoate succinyltransferase subunit. Results are also presented showing the inactivation of pyruvate dehydrogenase complex on preincubation with 2-ketoisovalerate and of 2-ketoglutarate dehydrogenase complex with methylenecyclopropylpyruvate, the keto acid corresponding to the toxic amino acid hypoglycin. The relevance of covalent modifications of the two keto acid dehydrogenase complexes to the pathological manifestations of maple syrup urine disease are discussed.

Incubation of 2-ketoglutarate dehydrogenase complex with 2-ketoisovalerate, 2-keto-4-methylvalerate, or 2-keto-3-methylvalerate leads to the appearance of a lag phase and of a progressive loss of activity in subsequent measurements of the initial rate of oxidation of 2-ketoglutarate. In the case of 2-ketoisovalerate these effects are shown to be due to the formation of an isobutyryllipoate derivative of the enzyme, as a result of the very slow oxidation of 2-ketoisovalerate by the enzyme complex (Vmm = 0.15% of that for 2-ketoglutarate). Incubation of the enzyme complex with 2-ket~['~C]isovalerate or 2-ket0['~C]glutarate results in comparable incorporation of radioactivity, amounting to 3.5 to 5.3 m o l of isobutyryl or succinyl residues per mg of protein in the complex. Isobutyryl residues are also incorporated in the enzyme during the simultaneous oxidation of both of these substrates. During the early phase of incubation of the complex with 2-ketoisovalerate the incorporation of isobutyryl residues is much faster than the loss of enzyme activity. This observation seems to support the suggestion that each 2ketoglutarate decarboxylase subunit of the complex may catalyze the succinylation of more than one lipoate succinyltransferase subunit.
Results are also presented showing the inactivation of pyruvate dehydrogenase complex on preincubation with 2-ketoisovalerate and of 2-ketoglutarate dehydrogenase complex with methylenecyclopropylpyruvate, the keto acid corresponding to the toxic amino acid hypoglycin. The relevance of covalent modifications of the two keto acid dehydrogenase complexes to the pathological manifestations of maple syrup urine disease are discussed.
The branched chain keto acids, 2-ketoisovalerate, 2-keto-3methylvalerate, and 2-keto-4-methylvalerate (keto forms of the amino acids valine, isoleucine, and leucine, respectively), are catabolized in the mitochondrion, the initial step being decarboxylation by the branched chain keto acid dehydrogenase, a multienzyme complex. In maple syrup urine disease, a rare genetic disorder, this enzyme is defective and, therefore, branched chain keto acids accumulate (1,2). The plasma levels of these keto acids rise to the millimolar range (3, 4), * This work was supported by the Veterans Administration, by Grant HL-16251 from the National Institutes of Health, and by Grant PCM 81-19609 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and the level reached is a function of the remaining activity of the branched chain keto acid dehydrogenase of the patient.
In the heart tissue of normal rats the combined concentration of the three branched chain keto acids is about 4 IJM, while the plasma level is 40 IJM (5). One would expect the concentration in the heart tissue of patients afflicted with the disease to rise in response to the increased level of branched chain keto acids found in circulating fluids.
The increased excretion of 2-ketoglutarate in the urine of untreated patients with maple syrup urine disease (6, 7), the impaired utilization of pyruvate and of 2-ketoglutarate by tissues of such patients (6), and the inhibition of O2 uptake by rat brain slices in the presence of 2-ketoisovalerate (8), suggested many years ago that the inhibition of pyruvate and 2ketoglutarate dehydrogenase complexes by branched chain keto acids may be an important factor in the pathology of the disease (e.g. Patrick (6)). Using standard steady state kinetic methods (9) or measurements of ['4C]C02 released from carboxyl labeled substrates (10, 11), the inhibition of 2-ketoglutarate and pyruvate dehydrogenase complexes by branched chain keto acids was demonstrated, with Ki values in the millimolar range.
In the present paper it will be shown that preincubation of these enzyme complexes with branched chain keto acids leads to the acylation of lipoate residues and that extensive inactivation ensues.

EXPERIMENTAL PROCEDURES
Materials-2-Ketoglutarate dehydrogenase and pyruvate dehydrogenase complexes were isolated from beef heart mitochondria as described by Stanley and Perham (12). The activity (at V,,,) of most samples of 2-ketoglutarate dehydrogenase complex ranged from 15 to 18 pmol of NADH min" mg protein", consistent with the highest specific activities reported (12,13). Preparations of 2-ketoglutarate dehydrogenase complex were 80 to 90% pure as judged by the intensity of staining by Coomassie blue on sodium dodecyl sulfate polyacrylamide gels after electrophoresis. Some preparations, as isolated, showed a lag phase in assay, which was abolished by preincubation with 2 mM NAD' and 0.2 m~ thiamin pyrophosphate at 30 "C prior to assay. This procedure also led to a variable (up to 40%) increase in the maximum rate in subsequent assays. Where necessary, these activating agents were removed by rapid centrifugation through Sephadex G-50 (14). Removal of the activators did not lower the specific activity. Preparations of the pyruvate dehydrogenase complex were approximately 9 0 8 pure with a specific activity of -12 pmol of NAD' reduced min" mg protein".
L-Amino acid oxidase (twice recrystallized) was prepared by the method of Wellner and Meister (15) and general fatty acyl CoA dehydrogenase from pig kidney was a gift of Dr. C. Thorpe (19). The sodium salts of 2-ketoisovalerate, 2-keto-3-methylvalerate, and 2keto-4-methylvalerate, and coenzyme A, isobutyryl coenzyme A, NAD', NADH, and Triton X-100 were from Sigma. Triton was purified as described (20). Methods-The activity of the 2-ketoglutarate dehydrogenase complex was assayed by following the reduction of NAD' at 340 nm (13), as described in the legend to Fig. 1. The same method was used for assay of the pyruvate dehydrogenase complex, except for the substitution of 2 mM pyruvate as substrate. 2-Ketoglutarate and pyruvate decarboxylase activities (also called "dehydrogenase" activities in the literature) were determined by following the reduction of 1 mM ferricyanide at 420 nm (21) by 2 m~ keto acid in 50 m~ potassium phosphate, pH 6.5, containing 2 m~ MgC12, 0.5 mM CaCl2, and 0.2 mM thiamin pyrophosphate at 30 "C. Other enzymes were assayed by published procedures, as follows: general fatty acyl-CoA dehydrogenase (22), lipoamide dehydrogenase (23), and NADH-5,5"dithiobis-(2nitrobenzoate) reductase (24). The latter reaction, catalyzed by lipoamide dehydrogenase, uses the enzyme-bound lipoic acid of lipoate acetyltransferase or lipoate succinyltransferase as a cofactor (24, 25).
Protein was determined by the method of Lowry et al. (26) in the presence of 0.1% (w/v) sodium dodecyl sulfate in order to prevent interference by Triton X-100 in the samples. Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate was per Laemmli (27), using 9% (w/v) polyacrylamide resolving gel and 5.5% stacking gel. Stopped flow measurements were made with an Aminco-Morrow apparatus interfaced with a Nova 2/4 minicomputer.
Radioactivity covalently bound to protein was determined as follows. The samples were precipitated with cold 10% (w/v) trichloroacetic acid onto Millipore filters, washed 4 times with cold 10% trichloroacetic acid, dried, dissolved in IO ml of Filtron X (National Diagnostics, Inc.), and counted in a liquid scintillation counter. In calculating the incorporation of isobutyryl and succinyl residues, allowance was made for the release of ['4C]C02 during decarboxylation of 2-ketoglutarate or 2-ketoisovalerate. To determine the lability of protein-bound succinyl or isobutyryl residues to performic acid, enzyme samples were applied to squares (2 X 2 cm) of Whatman No. 3 " filter paper and treated by the procedure of Pettit et al. (28). One set of control samples was counted directly following the washing procedure and the other was counted after incubation in a desiccator over a solution of 30% (w/v) H202-98% (v/v) formic acid, 1:20 (v/v), for 48 h and washing the papers to remove the succinic or isobutyric acid liberated.
The lability of protein-bound isobutyryl residues to CoA plus NAD' was determined as follows. 2-Ketoglutarate dehydrogenase complex (0.61 mg/ml) was incubated anaerobically with 2-keto[U- were applied at 4 "C to 0.8-ml columns of Sephadex G-50, fine, equilibrated in the same buffer, and centrifuged for 2.5 min, as described (14). The excluded fractions were pooled, incubated with 0.5 m~ CoA and 4 mM NAD' for 20 min at 30 "C, precipitated, filtered through Millipore fiiters, and the residues on the fiiters were counted. The catalytic activity of the enzyme in the overall reaction was monitored at each stage.

RESULTS
Inactivation of 2-Ketoglutarate Dehydrogenase Complex by Incubation with Branched Chain Keto Acids-Preincubation of the 2-ketoglutarate dehydrogenase complex at 20 "C with 2 mM 2-ketoisovalerate or 2-keto-4-methylvalerate in the presence of thiamin pyrophosphate, Mg2+, and Ca2+ resulted in a lag phase in subsequent assays with 2-ketoglutarate as substrate (Fig. 1). This lag phase increased with the length of preincubation, and a progressive diminution of the maximal rate of 2-ketoglutarate oxidation was also observed (Fig. 2). Whereas preincubation with 2 mM 2-ketoisovalerate and 4 mM 2-keto-3-methylvalerate produced comparable decreases in the initial rate (Fig. 2 4 ) and the maximal rate of Z-ketoglutarate oxidation (Fig. 2B), on preincubation with 2 mM 2-keto-4-methylvalerate loss of initial activity was slower ( Fig. 2 A ) , but loss of maximal activity (Fig. 2B) was more extensive.
Mechanism of the Inactiuation-A sample of the complex (0.39 mg of protein ml") was incubated for 90 min with 2 mM 2-ketoisovalerate as described in Fig. 1. This led t o a 95% slower initial rate and an 87% reduction of the maximal rate of oxidation of 2-ketoglutarate in subsequent assays. Extensive dialysis against 50 mM Mops, pH 7.6, at 4 "C did not reactivate the enzyme, suggesting that a covalent modification of one of the proteins of the complex had occurred. Moreover, radioactivity became progressively incorporated into t h e complex when Z-ket~[U-'~C]isovalerate was used for inactivation.
The simplest interpretation of these observations is that the branched chain keto acids are decarboxylated to form a n adduct with the thiamin pyrophosphate bound noncovalently to 2-ketoglutarate decarboxylase and that the lipoate residue of lipoate succinyltransferase is then acylated to form a thioester. This hypothesis would predict that the 2-keto acids are substrates for the complex, provided that the enzyme could

2-Ketoglutarate Dehydrogenase and Branched Chain Keto Acids
transfer the acyl group from lipoate to CoA. The relative rate of oxidation of 2-ketoisovalerate by the complex, as measured by V,,, for the formation of NADH, was, in fact, 0.15% of the rate for 2-ketoglutarate, while V,,, for 2-keto-3-methylvalerate and 2-keto-4-methylvalerate were 0.04 and 0.06% of the rate of 2-ketoglutarate oxidation, respectively.
While it is difficult to rule out contamination of our preparations by traces of branched chain keto acid dehydrogenase complex, the polypeptides of that enzyme could not be detected following polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate.' Furthermore, the K,,, value obtained for 2-ketoisovalerate, 0.33 mM, is 10-fold higher than that reported for the branched chain keto acid dehydrogenase (13 to 40 p~ (29-31)). Additional evidence that 2ketoglutarate dehydrogenase complex catalyzes the oxidation of 2-ketoisovalerate will be found in the labeling experiments presented below.
Incorporation of Radioactivity during Processing of Labeled 2-Ketoisovalerate a n d Its Reversal-If the inactivation of the 2-ketoglutarate dehydrogenase complex by 2-ketoisovalerate (and the analogous reactions with other keto acids) was due to the formation of isobutyryllipoate, the resulting thioester bond should be labile to performic acid (32). T o test this hypothesis, the complex (1.55 mg of protein rn") was incubated anaerobically for 2 h at 4 "C with 0.285 mM 2-ket~[U-'~C]isovalerate in 100 m~ Mops buffer, pH 7.0, containing 2 mM MgC12, 0.5 mM CaC12, and 0.2 m~ thiamin pyrophosphate. Aliquots were removed at 1 and 2 h and radioactivity incorporated into the complex was determined before and after exposure to performic acid vapor, as described under "Methods." The samples not exposed to performic acid showed 3.5 and 4.6 nmol of isobutyryl residue incorporated per mg of complex a t 1 and 2 h, respectively. Performic acid removed 95% of the enzyme-bound radioactivity in each experiment, showing that a thioester linkage was, indeed, involved. Treatment of the modified enzyme with 0.5 mM CoA and 4 mM NAD' at 30 "C (see under "Methods") also liberated 95% of the protein-bound radioactivity, with an accompanying increase in the specific activity of the 2-ketoglutarate dehydrogenase complex from 10% of that of untreated samples before addition of CoA and NADf to 80% after 20 min of incubation. In the experiments of Fig. 3, enzyme inactivated with 2 mM 2-ketoisovalerate was incubated with NAD', CoA, and thiamin pyrophosphate, with restoration of full activity. Control experiments showed that of these additions, only CoA is effective in reversing the inactivation. The transfer of isobutyryl residues from lipoate to CoA with relief of inhibition, confirms that 2-ketoisovalerate is a substrate for the overall reaction of the complex. Incubation with CoA also abolishes the lag phase in assays. The lag phase produced by branched chain keto acids ( Fig. 1) is thus distinct from that seen in occasional samples of untreated enzyme, which is unaffected by CoA, but is abolished by preincubation with NAD' and thiamin pyrophosphate.
In order to determine the stoichiometry of incorporation of isobutyryl residues into the enzyme complex, the latter was incubated anaerobically with 0.3 mM 2-ket0[U-'~C]isovalerate in 100 mM Mops, pH 7.0, containing 2 mM MgC12, 0.5 mM CaCl', and 0.2 m~ thiamin pyrophosphate a t 4 "C. The extent of incorporation was in the range of 3.5 to 5.3 nmol of isobutyryl residue per mg of protein in the complex in several The two complexes may be distinguished by the presence of low molecular weight peptides, originating from the carboxylase component, in the branched chain keto arid dehydrogenase complex, which are not seen in the 2-ketoglutarate dehydrogenase complex. experiments. Increasing the concentration of 2-ketoisovalerate to 0.6 mM or the temperature to 25 "C increased the rate but not the extent of incorporation. The incorporation of succinyl residues from 2-keto['*C]glutarate in analogous studies was the same as that of isobutyryl residues for each preparation, although it occurred far more rapidly. Label originating from either substrate was over 95% released on exposure to performic acid vapor for 48 h. Hence, 2-ketoglutarate and 2ketoisovalerate seem to form thioester bonds with the same residues in the enzyme complex.
Calculation of the stoichiometry of incorporation per lipoic acid residue in the 2-ketoglutarate dehydrogenase complex is difficult, because no reliable values have been reported for the lipoate content of the enzyme from mammalian sources. The only analyses reported in the literature (33) used alkaline hydrolysis and a microbiological assay, which are not conducive to great accuracy. Using the data given above, and assuming that the subunit stoichiometry of the beef heart 2ketoglutarate dehydrogenase complex is decarboxy1ase:lipoate succiny1transferase:lipoamide dehydrogenase = 12:24:12 and the molecular weight 2,700,000 as reported (33), the stoichiometry of incorporation of isobutyryl (or succinyl) residues would be 0.4 to 0.6 mol per mol of lipoate succinyltransferase. If the mammalian enzyme complex, like the enzyme from Escherichia coli (34)(35)(36), contains 1 mol of lipoate per mol of lipoate succinyltransferase, as seems likely, the incorporation would be 0.4 to 0.6 mol of residue per mol of lipoic acid residue. We have, as yet, no explanation for the apparently substoichiometric labeling of the lipoate residues.
Other Evidence for the Proposed Mechanisms-If the mechanism of inactivation of the 2-ketoglutarate dehydrogenase complex by branched chain keto acids involves the formation of lipoate thioesters which can only slowly transfer their acyl moieties to CoA, it should be possible to inactivate the enzyme with isobutyryl-CoA and NADH, by reversal of the lipoate succinyltransferase and lipoamide dehydrogenase reactions. Fig. 4 shows that this is indeed the case.
If acylation of the lipoate residues of lipoate succinyltransferase were the cause of inactivation, only those reactions of the enzyme complex which involve lipoic acid should be inhibited. It was indeed found that the complete reaction of the complex, as measured by the reduction of NAD', and the reduction of 5,5'-dithiobis-(2-nitrobenzoate) by NADH, reac- The initial rate of NAD' reduction was measured on aliquots and is expressed as the per cent of the rate at the start of the incubation.
tions known to depend on the participation of lipoic acid, were >95% inactivated, but those activities in which the enzymebound lipoate plays no role (i.e. the reduction of Fe(CN)C3 by 2-ketoglutarate and the reduction of lipoamide by NADH) were unaffected. was incubated with 2-ket0['~C]isovalerate for 30 min under the conditions given above, and a sample was then rapidly mixed with an equal volume of 10 mM 2-ketoglutarate, containing all the other components used in the stopped flow assay. Aliquots precipitated with trichloroacetic acid and counted, as above, showed 2.8 f 1.5% (S.E.) of the radioactivity released within 20 s after mixing. Hence, it is valid to compare the kinetics of the labeling of the enzyme by isobutyryl residues with the rate of inactivation, as in Fig. 5. It may be noted that the loss of activity shows a lag not seen in the incorporation of isobutyryl residues. By plotting these parameters against each other (Fig. 6), it is evident that labeling of 20 and 40% of the available binding sites results in only 5 and 11% of inactivation, respectively. By the time 1 mol of isobutyryl residue was incorporated per mol of site, no reduction of NAD' could be seen within 10 s in stopped flow experiments, although over longer periods of assay (1 min) some reactivation was noted. Contamination by pyruvate dehydrogenase  complex (present a t about 2% of the concentration of 2-ketoglutarate dehydrogenase complex) was far too small to account for the discrepancy between the initial loss of activity and incorporation of label.

Incorporation of Isobutyryl Residues in the Presence of 2-Ketoglutarate and
CoA-The data presented so far suggest that the lipoate residues of the 2-ketoglutarate dehydrogenase complex are acylated by these branched chain keto acids and that the acyl residues formed are only slowly transferred to CoA. However, within the mitochondrion the natural substrate, 2-ketoglutarate, will compete with the branched chain keto acids and the presence of CoA will permit a degree of The experiment summarized in Table I was designed to test whether significant formation of isobutyryllipoate from 2-ketoisovalerate still occurs in the presence of substantial concentrations of 2-ketoglutarate (0.6 m~) and CoA (0.5 m~) .
Aliquots were removed, precipitated, and counted when 50, 90, and 100% of the CoA was used up as judged by the extent of NAD+ reduction. It is clear that accumulation of isobutyryllipoate still occurs despite the presence of 2-ketoglutarate and CoA. However, as the reaction proceeded toward completion, isobutyryllipoate did not accumulate further, possibly because of competing succinylation of the lipoate residues by NADH and CoA. The extent of incorporation observed during the course of the reaction, as monitored by NAD' reduction, was strongly dependent on the enzyme concentration, suggesting that the incorporation depends on the number of times the enzyme turns over.
Characteristics of the Inhibition of the 2-Ketoglutarate Dehydrogenase Complex in Steady State Assays-A number of investigators have reported on the kinetics of the inhibition of the 2-ketoglutarate dehydrogenase complex by branched chain keto acids (9-11) and have published varying K, values. There is also substantial disagreement on the characteristics of the inhibition. Thus, some authors claim (9, 10) that the inhibition by branched chain keto acids in initial rate measurements is mixed with respect to 2-ketoglutarate, whereas others (11) found it to be purely competitive with respect to 2-ketoglutarate. Lawlis and Roche (13) reported that the reaction of the enzyme complex with 2-ketoglutarate showed negative cooperativity, whereas others (9-11) did not find this. Such differences are unlikely to be due to the fact that the complex was isolated from different mammalian tissues in the various laboratories, especially in view of the demonstration (13) that the kinetic characteristics of the enzyme complex are qualitatively the same in different organs. It seemed necessary, therefore, to examine the kinetic characteristics of the inhibition by branched chain keto acids without preincubation with the enzyme.

Incorporation of isobutyryl residues into 2-ketoglutarate dehydrogenase complex during the simultaneous oxidation of 2-
ketoglutarate and 2-ketoisoualerate 2-Ketoglutarate dehydrogenase complex was incubated at 15 "C, pH 7.0, with 1 m~ 2-ket0['~C]isovalerate, 0.6 m~ 2-ketoglutarate, 0.2 mM thiamin pyrophosphate, 0.5 m~ CoA, and 0.2 m~ NAD'. The reduction of NAD+ was followed spectrophotometrically and samples were taken at intervals, precipitated with trichloroacetic acid onto Millipore filters, and counted. Note that at a concentration of 0.13 mg d" the enzyme must undergo twice as many turnovers to produce the same extent of NAD+ reduction as it would at a concentration of 0.26 mg ml". Since the CoA concentration was 0.5 m~, the reaction was complete when 0.5 m~ NAD+ had been reduced.
b T h e concentration of available sites was determined from the maximum amount of succinyl residues incorporated from 2-keto[14C] glutarate in parallel experiments.
During initial rate assays in the presence of inhibitor (the reaction being started with enzyme), the rate of reduction of NAD+ decreases much more in the course of the assay than in the control without inhibitor, particularly at high concentrations of the inhibitor. This is probably due to the rapid formation of isobutyryllipoate leading to quasi-irreversible inhibition. Under these conditions the steady state assumptions will be invalid. The failure to appreciate that these keto acids inhibit by covalent modification, as well as by competing at the 2-ketoglutarate-binding site, may explain some of the discrepancies in the literature.
Nevertheless, during the fist few seconds after the mixing of the reactants the absorbance change is sufficiently linear to suggest that the quasi-irreversible loss of activity is not yet a significant factor. From such initial rates the occurrence of negative cooperativity was readily confimed in the presence or absence of inhibitor (Fig. 7). Moreover, in the range of 0-20 mM 2-ketoisovalerate, the Sty for 2-ketoglutarate increased from 0.3 to 1.2 mM, whereas Vapp remained unchanged (Table  11). Thus, 2-ketoisovalerate appears to be a competitive inhibitor with respect to 2-ketoglutarate. 2-Keto-3-methylvalerate and 2-keto-4-methylvalerate were found to be considerably weaker inhibitors than 2-ketoisovalerate, yielding at 0 to 5 mM inhibitor a slight decrease in Vapp and only a marginal increase in S$Y. Although the presence of negative coopera-

Kinetic parameters for the inhibition of 2-ketoglutarate dehydrogenase by 2-ketoisoualerate
The kinetic parameters were estimated from Eadie-Hofstee plots (Fig. 7) of the initial rate of 2-ketoglutarate oxidation. tivity precludes the conventional calculation of Ki values, in our hands the inhibition of the purified enzyme complex from beef heart by these two keto acids is much less than has been reported by others (9-11).

Concentration
In other experiments, isobutyryl-CoA, the product of oxidative decarboxylation of 2-ketoisovalerate, was found to be a competitive inhibitor with respect to CoA, with an apparent Ki of 0.55 m~ for the range of concentrations of CoA over which substrate inhibition cannot be detected (37).

Effect of Branched Chain Keto Acids on Pyruvate Dehydrogenase
Complex-Incubation of the pyruvate dehydrogenase complex from beef heart with 2 m~ 2-ketoisovalerate prior to assay leads to extensive loss of activity in the overall reaction, whether initial rates of pyruvate oxidation or the maximal rate attained are measured (Fig. 8). The characteristics of the inactivation were found to be very similar to those observed with the 2-ketoglutarate dehydrogenase complex. Thus, the loss of activity was not reversed by dialysis and affected only those activities of the pyruvate dehydrogenase complex in which the lipoic acid component of lipoate acetyltransferase participates (i.e. the overall reduction of NAD' by pyruvate and the NADH-dependent reduction of 5,5'-dithiobis-(2-nitrobenzoate)), whereas the lipoamide dehydrogenase reaction and the reduction of Fe(CN)G? by pyruvate, catalyzed by the decarboxylase, were unaffected.
Incubation of the complex with 2 m~ 2-keto-4-methylvalerate or 4 mM ~~-2-keto-3-methylvderate under the conditions of the experiment in Fig. 8 produced no inactivation. As expected, these keto acids were also not perceptibly oxidized by the pyruvate dehydrogenase complex, while 2-ketoisovalerate was oxidized, as determined by NADH production, but only at 0.05% of the maximal rate of pyruvate oxidation even at 20 m~. Mops, pH 7.6, containing 2 mM MgC12, 0.5 mM CaC12, and 0.2 mM thiamin pyrophosphate at 20 "C, and aliquots were assayed at intervals by following the reduction of NAD'. The activity is expressed as the per cent of the rate given by untreated samples. 0, initial rate; 0, maximal rate a:tdined in the assay; M, activity of a control sample incubated without 2-ketoisovalerate. Note that in the latter the initial and maximal rates are the same. keto acids considered above. This compound had no effect on initial rates of the 2-ketoglutarate dehydrogenase complex at a concentration of 2 m~, but major and progressive inactivation was produced on preincubation with 1 mM methylenecyclopropylpyruvate prior to assay (Fig. 9). As in the case of branched chain keto acids (Fig. l), the partially inactivated enzyme shows a lag phase in assay. The lag phase was abolished and the activity regenerated by incubation of the inactivated enzyme complex for 5 min at 30 "C with the assay mixture (minus 2-ketoglutarate), showing that the adduct formed with the lipoate moiety can be slowly transferred to CoA. In a typical experiment after 2-h incubation with 1 mM methylenecyclopropylpyruvate the activity (initial rates) fell from 8.6 to 0.89 pmol of NADH produced per min per mg, but 5-min incubation with the assay mixture raised the activity to 7.6 pmol of NADH per min per mg.
The inactivation could also be reversed and the lag phase in assays abolished by 0.45 M hydroxylamine, which presumably cleaves the thioester in the inactivated enzyme, forming lipoic acid and methylenecyclopropylacetylhydroxamate.
That inactivation by this keto acid results from esterification of the lipoic acid residues of lipoate succinyltransferase, as has been concluded for the action of branched chain keto acids, is supported by the facts that only lipoate-dependent

2-Ketoglutarate Dehydrogenase and Branched Chain Keto Acids
reactions were affected by the inactivation, whereas the reduction of Fe(CN)6-3 by 2-ketoglutarate and the reactions of lipoamide dehydrogenase were not.
The oxidation of methylenecyclopropylpyruvate to methylenecyclopropylacetyl-CoA was too slow to be conveniently detected by NADH formation in the complete assay system (less 2-ketoglutarate). It could be readily followed, however, by measuring the suicide inhibition of the general fatty acyl-CoA dehydrogenase by methylenecyclopropylacetyl-CoA (38) (Fig. 10). Incubation of the pyruvate dehydrogenase complex with 1 mM methylenecyclopropylpyruvate led to no detectable inactivation, in agreement with the results of Kean and Pogson (39).

DISCUSSION
It is clear that the 2-ketoglutarate dehydrogenase complex from beef heart has a broader specificity for keto acids than has been previously reported for this enzyme from mammalian tissues, although a broad specificity has been noted in studies of the reverse reaction, i.e. the NADH-dependent acylation of lipoate residues (40). 2-Ketoglutarate dehydrogenase complex catalyzes the oxidative decarboxylation of 2-ketoisovalerate, 2-keto-3-methylvalerate, 2-keto-4-methylvalerate, and of methylenecyclopropylpyruvate. The failure of previous workers to detect the oxidation of 2-ketoisovalerate by the complex from pig heart (9, 30) may be due to the slow rate, which would be easily overlooked unless a sufficiently high concentration of enzyme were used. However, Kanzaki et al. (9) did note that the enzyme catalyzed the reduction of Fe(CN)s-", but not that of NAD', by 2-ketoisovalerate. We found that pyruvate dehydrogenase complex also catalyzes the oxidative decarboxylation of 2-ketoisovalerate, but the high activity toward 2-keto-4-methylvalerate previously reported (9) was not detected.
An important conclusion from this study is that the catalytic activity of the 2-ketoglutarate dehydrogenase complex toward keto acids is determined not only by the substrate site of the decarboxylase moiety, but also by the catalytic site of lipoate succinyltransferase for the transfer of acyl residues from lipoate to CoA. Thus, the rate of this latter reaction is dependent on the nature of the acyl group bound to the lipoate.
The evidence presented in this paper indicates that the loss of activity of the two complexes caused by branched chain keto acids is more an inactivation than an inhibition. The initial effect of 2-ketoisovalerate on 2-ketoglutarate oxidation by the 2-ketoglutarate dehydrogenase complex is largely competitive inhibition toward 2-ketoglutarate. 2-Ketoisovalerate, being a substrate for the enzyme complex, is decarboxylated to yield an adduct with the thiamin pyrophosphate component of the complex. This adduct is then oxidized and transferred to the lipoate succinyltransferase component, forming a thioester with the lipoate moieties thereof. The difference between these events and the oxidation of natural substrates seems to be that whereas succinyllipoate reacts rapidly to form succinyl-CoA, regenerating thereby free lipoate, isobutyryllipoate reacts sluggishly with CoA, leaving the lipoate succinyltransferase in an unreactive (or inactive) state for a prolonged period.
Branched chain keto acids (and methylenecyclopropylpyruvate) thus act in some, but not all, respects like typical suicide inhibitors. Although a n inactive covalent adduct is formed with the enzyme as a result of the oxidation of the branched chain keto acid by the target enzyme complex, a process prevented by high concentrations of the natural substrate, adduct formation is only quasi-irreversible and the enzyme is slowly converted to the active form by reaction with free CoA.
The strong dependence of the extent of formation of isobutyryllipoate in the presence of substrates on enzyme concentration (Table I) has already been pointed out. A possible explanation is as follows. 2-Ketoglutarate and 2-ketoisovalerate compete for the first two reactions of the complex yielding thioesters with the lipoate residues. Succinyl groups are then transferred to CoA more rapidly than isobutyryl groups, freeing the lipoate residues to which succinyl groups were bound. This permits further competition between P-ketoglutarate and 2-ketoisovalerate and gradual accumulation of isobutyryllipoate. Since the lower its concentration the more times the enzyme must turn over to reduce the same amount of NADf, this mechanism predicts that the extent of inactivation would be inversely related to the enzyme concentration, as noted.
The observation that at low levels of incorporation the labeling of the complex by isobutyryl residues is more extensive than is the loss of activity (Figs. 5 and 6) requires an explanation. According to the accepted model (41), in the 2ketoglutarate dehydrogenase complex of E. coli, succinyl residues originating from 2-ketoglutarate form thioester linkages with the lipoate residues of lipoate succinyltransferase, which serve as swinging arms between the three enzymes of the complex. These lipoic acids are thought to transfer succinyl residues to other lipoic acid moieties in the core of the complex from E. coli (35). The general features of the model are usually thought to apply also to the mammalian enzyme complex.
One may visualize three possible mechanisms for the succinylation of lipoate residues. 1) Each 2-ketoglutarate decarboxylase subunit catalyzes the succinylation of a specific lipoate residue and the succinyl residue is then directly transferred to CoA. This model predicts that the loss of activity should be proportional to the fraction of lipoic acid modified, as was found for the inactivation of the 2-ketoglutarate dehydrogenase complex of E. coli by N-ethylmaleimide in the presence of 2-ketoglutarate (34). 2) Each 2-ketoglutarate decarboxylase subunit catalyzes the succinylation of a specific lipoid acid residue, but the succinyl group is obligatorily transferred to a second lipoic acid moiety prior to being passed on to CoA. This model requires that at any time inactivation be more extensive than incorporation of acyl residues derived from the inhibitor. 3) Each 2-ketoglutarate decarboxylase subunit can catalyze the succinylation of two or more lipoic acid residues bound to separate lipoate succinyltransferase chains. This model predicts that loss of activity be initially less extensive than the incorporation of isobutyryl residues, as found in Fig. 6 (43) and Berman et al. (44) in terms of the "lipoic acid takeover mechanism" summarized above (mechanism 3).
if two lipoate residues interact with each 2-ketoglutarate decarboxylase subunit. It remains to discuss the relevance of the finding that branched chain keto acids inactivate the 2-ketoglutarate and pyruvate dehydrogenase complexes to the pathology of maple syrup urine disease. Elevated plasma levels of branched chain keto acids are known to occur in uncontrolled diabetes, during elevated catabolism of protein, and especially in maple syrup urine disease (47,48). These high levels are known to be toxic, even when the catabolism of these keto acids is normal, but the effect is aggravated in patients with maple syrup urine disease, who have a defective or deficient branched chain keto acid dehydrogenase.
The deleterious effects of branched chain keto acids on metabolism have been clearly demonstrated in studies in Williamson's laboratory with isolated hepatocytes from normal rats (49)(50)(51). Incubation with 1 mM 2-ketoisovalerate severely inhibited fluxes through the 2-ketoglutarate and pyruvate dehydrogenase complexes. Thus, flux between oxaloacetate and succinyl-CoA decreased by 97%, with a dramatic decline in the succinyl-CoA content and a rise in the concentration of isobutyryl-CoA, propionyl-CoA, methylmalonyl-CoA, and in the NADH/NAD' ratio, while flux through the pyruvate dehydrogenase complex diminished by 60% under these conditions. These marked effects cannot be explained by competitive inhibition of the two keto acid dehydrogenase complexes by 2-ketoisovalerate, but would not be unexpected in view of the covalent modification of the complexes as described in this paper. From our data, at 2 m~ 2-ketoisovalerate and in the presence of saturating NAD+ and CoA and 0.4 mM 2-ketoglutarate (within the range of mitochondrial concentrations estimated by LaNoue et al. (52)), the inhibition of the 2-ketoglutarate dehydrogenase complex is only 8% in initial rate studies. The effect of 1 m~ 2-ketoisovalerate on initial rates of the pyruvate dehydrogenase complex reaction at the expected prevailing concentration of pyruvate would be similarly negligible. Whether the increased concentrations of methylmalonyl-CoA and propionyl-CoA contribute significantly to the severe decline in flux through the 2-ketoglutarate dehydrogenase complex in the presence of 1 mM 2-ketoisovalerate (49, 50) cannot be evaluated, since the K, values of the enzyme for these compounds are not known. However, since the K, for isobutyryl-CoA, with respect to CoA, from our data is 0.55 m, whereas the K, for succinyl-CoA, with respect to CoA, is 6.9 PM (53) under comparable conditions, it is clear that the competitive inhibition by the increased level of isobutyryl-CoA would be more than compensated for by the drop in the level of succinyl-CoA. Product inhibition due to the increased NADH/NAD+ ratio (54) could contribute to the decrease in flux through the 2-ketoglutarate dehydrogenase complex. The severity of the decrease, however, suggests that the drop is largely caused by another factor, probably the incorporation of covalently bound isobutyryl residues into the enzyme. In contrast, inactivation of the pyruvate dehydrogenase complex by covalent modification may be less important, since the decrease in flux through that enzyme could be explained by the observed increase in the NADH/NAD+ and acetyl-CoA/CoA ratios together with inhibition by propionyl-CoA and isobutyryl-CoA (50,51).
In maple syrup urine disease, patients with severe cases have concentrations of 2 to 4 m~ 2-keto-4-methylvalerate, 1.9 mM 2-ketoisovalerate, and 1 to 1.5 m~ 2-keto-3-methylvalerate in the plasma (3), but lower plasma concentrations are stiU toxic (4). Assuming that the mitochondrial concentrations of these keto acids are comparable, from our data it seems likely that quasi-irreversible inactivation of the 2-ketoglutarate dehydrogenase complex would occur by oxidative decarboxyla-tion of the branched chain keto acids leading to the formation of thioesters with lipoate. The extent of such inactivation in vivo would be difficult to estimate without reliable data on the mitochondrial concentrations of target enzyme, P-ketoglutarate, the branched chain keto acids, and their products in the afflicted patients. Competition by the branched chain keto acids for the substrate site of the enzyme might further lower the rate of cycling of the enzyme to a small extent.
Inactivation of pyruvate dehydrogenase complex could also contribute to the overall toxicity of high levels of branched chain keto acids, although only 2-ketoisovalerate leads to progressive loss of activity, which may again reflect isobutyrylation of the lipoate residues. At comparable steady state concentrations of pyruvate and 2-ketoglutarate, however, the effects on the pyruvate dehydrogenase complex would be less than on the 2-ketoglutarate dehydrogenase complex, because the K, of the former enzyme for pyruvate is 20 PM, whereas s0.5 of the 2-ketoglutarate dehydrogenase complex for 2-ketoglutarate is 300 PM.
While the data suggest that covalent modification of the 2ketoglutarate dehydrogenase complex may be a major cause of the toxemia associated with maple syrup urine disease, our findings do not indicate that inhibition or inactivation of the enzyme by methylenecyclopropylpyruvate is of major importance in Jamaican vomiting sickness. The causative agent of this disease, hypoglycin, is transaminated in the body to methylenecyclopropylpyruvate. The latter compound and methylenecyclopropylacetate, the product of oxidative decarboxylation and hydrolysis, cause powerful inhibition of isovaleryl-CoA dehydrogenase and of the general fatty acyl-CoA and butyryl-CoA dehydrogenases on incubation with mitochondria (55)(56)(57). Methylenecyclopropylacetyl-CoA was, in fact, recently shown to be a suicide inhibitor of butyryl and general fatty acyl-CoA dehydrogenases (38,58). Collectively, these effects are thought to explain the hypoglycemia associated with the disease, since inhibition of these enzymes would reduce the concentration of acetyl-coA, an allosteric activator of pyruvate carboxylase, and, hence, interfere with gluconeogenesis.
Whereas the inactivation of isovaleryl-CoA, butyryl-CoA, and general fatty acyl-CoA dehydrogenase develops rapidly and at low concentrations of the appropriate inhibitor, inactivation of the 2-ketoglutarate dehydrogenase complex by methylenecyclopropylpyruvate is a slow process even at millimolar concentrations of the inhibitor (Fig. 9). Inactivation of the 2-ketoglutarate dehydrogenase complex is thus not likely to contribute significantly to the development of hypoglycemia. Since the oxidation of methylenecyclopropylpyruvate by this enzyme complex is also very slow, but the action of branched chain keto acid dehydrogenase on the compound would be expected to be more rapid in view of its similarity to branched chain keto acids, the 2-ketoglutarate dehydrogenase complex is also unlikely to be the predominant system for the metabolic conversion of methylenecyclopropylpyruvate to methylenecyclopropylacetyl-CoA.