Lipoyl domain-based mechanism for the integrated feedback control of the pyruvate dehydrogenase complex by enhancement of pyruvate dehydrogenase kinase activity.

To conserve carbohydrate reserves, the reaction of the pyruvate dehydrogenase complex (PDC) must be down-regulated when the citric acid cycle is provided sufficient acetyl-CoA. PDC activity is reduced primarily through increased phosphorylation of its pyruvate dehydrogenase (E1) component due to E1 kinase activity being markedly enhanced by elevated intramitochondrial NADH:NAD+ and acetyl-CoA:CoA ratios. A mechanism is evaluated in which enhanced kinase activity is facilitated by the build-up of the reduced and acetylated forms of the lipoyl moieties of the dihydrolipoyl acetyltransferase (E2) component through using NADH and acetyl-CoA in the reverse of the downstream reactions of the complex. Using a peptide substrate, kinase activity was stimulated by these products, ruling out the possibility kinase activity is increased due to changes in the reaction state of its substrate, E1 (thiamin pyrophosphate). Each E2 subunit contains two lipoyl domains, an NH2-terminal (L1) and the inward lipoyl domain (L2), which were individually produced in fully lipoylated forms by recombinant techniques. Although reduction and acetylation of the L1 domain or free lipoamide increased kinase activity, those modifications of the lipoate of the kinase-binding L2 domain gave much greater enhancements of kinase activity. The large stimulation of the kinase generated by acetyl-CoA only occurred upon addition of the transacetylase-catalyzing (lipoyl domain-free) inner core portion of E2 plus a reduced lipoate source, affirming that acetylation of this prosthetic group is an essential mechanistic step for acetyl-CoA enhancing kinase activity. Similarly, the lesser stimulation of kinase activity by just NADH required a lipoate source, supporting the need for lipoate reduction by E3 catalysis. Complete enzymatic delipoylation of PDC, the E2-kinase subcomplex, or recombinant L2 abolished the stimulatory effects of NADH and acetyl-CoA. Retention of a small portion of PDC lipoates lowered kinase activity but allowed stimulation of this residual kinase activity by these products. Reintroduction of lipoyl moieties, using lipoyl protein ligase, restored the capacity of the E2 core to support high kinase activity along with stimulation of that activity up to 3-fold by NADH and acetyl-CoA. As suggested by those results, the enhancement of kinase activity is very responsive to reductive acetylation with a half-maximal stimulation achieved with ∼20% of free L2 acetylated and, from an analysis of previous results, with acetylation of only 3-6 of the 60 L2 domains in intact PDC. Based on these findings, we suggest that kinase stimulation results from modification of the lipoate of an L2 domain that becomes specifically engaged in binding the kinase. In conclusion, kinase activity is attenuated through a substantial range in response to modest changes in the proportion of oxidized, reduced, and acetylated lipoyl moieties of the L2 domain of E2 produced by fluctuations in the NADH:NAD+ and acetyl-CoA:CoA ratios as translated by the rapid and reversible E3 and E2 reactions.

To conserve carbohydrate reserves, the reaction of the pyruvate dehydrogenase complex (PDC) must be down-regulated when the citric acid cycle is provided sufficient acetyl-CoA. PDC activity is reduced primarily through increased phosphorylation of its pyruvate dehydrogenase (E1) component due to E1 kinase activity being markedly enhanced by elevated intramitochondrial NADH:NAD ؉ and acetyl-CoA:CoA ratios. A mechanism is evaluated in which enhanced kinase activity is facilitated by the build-up of the reduced and acetylated forms of the lipoyl moieties of the dihydrolipoyl acetyltransferase (E2) component through using NADH and acetyl-CoA in the reverse of the downstream reactions of the complex. Using a peptide substrate, kinase activity was stimulated by these products, ruling out the possibility kinase activity is increased due to changes in the reaction state of its substrate, E1 (thiamin pyrophosphate). Each E2 subunit contains two lipoyl domains, an NH 2 -terminal (L1) and the inward lipoyl domain (L2), which were individually produced in fully lipoylated forms by recombinant techniques. Although reduction and acetylation of the L1 domain or free lipoamide increased kinase activity, those modifications of the lipoate of the kinase-binding L2 domain gave much greater enhancements of kinase activity. The large stimulation of the kinase generated by acetyl-CoA only occurred upon addition of the transacetylase-catalyzing (lipoyl domain-free) inner core portion of E2 plus a reduced lipoate source, affirming that acetylation of this prosthetic group is an essential mechanistic step for acetyl-CoA enhancing kinase activity. Similarly, the lesser stimulation of kinase activity by just NADH required a lipoate source, supporting the need for lipoate reduction by E3 catalysis.
Complete enzymatic delipoylation of PDC, the E2-kinase subcomplex, or recombinant L2 abolished the stimulatory effects of NADH and acetyl-CoA. Retention of a small portion of PDC lipoates lowered kinase activity but allowed stimulation of this residual kinase activity by these products. Reintroduction of lipoyl moieties, using lipoyl protein ligase, restored the capacity of the E2 core to support high kinase activity along with stimulation of that activity up to 3-fold by NADH and acetyl-CoA. As suggested by those results, the enhancement of kinase activity is very responsive to reductive acetylation with a half-maximal stimulation achieved with ϳ20% of free L2 acetylated and, from an analysis of previous results, with acetylation of only 3-6 of the 60 L2 domains in intact PDC. Based on these findings, we suggest that kinase stimulation results from modification of the lipoate of an L2 domain that becomes specifically engaged in binding the kinase. In conclusion, kinase activity is attenuated through a substantial range in response to modest changes in the proportion of oxidized, reduced, and acetylated lipoyl moieties of the L2 domain of E2 produced by fluctuations in the NADH: NAD ؉ and acetyl-CoA:CoA ratios as translated by the rapid and reversible E3 and E2 reactions.
In mammalian cells, the pyruvate dehydrogenase complex (PDC) 1 controls the oxidative utilization of glucose (1). Flux through this reaction results in a net depletion of body carbohydrate reserves. The activity of PDC must be reduced when fatty acids or ketone bodies are being preferentially used to provide 2-carbon units for oxidative energy production by citric acid cycle/oxidative phosphorylation systems, a routine situation in many organs. Furthermore, under conditions of starvation or diabetes, the activity of PDC is reduced to a minimal level to conserve carbohydrates essential for the brain and other specialized tissues/organs. To achieve its critical role in cellular fuel conservation, the PDC reaction is controlled primarily by a highly regulated phosphorylation/dephosphorylation cycle which is carried out by dedicated kinase and phosphatase components. Phosphorylation of the pyruvate dehydrogenase (E1) component inactivates the complex and dephosphorylation reactivates the complex.
Shortly after PDC was shown to be regulated by this interconversion (2), the capacity of fatty acids and ketone bodies to promote inactivation of PDC was demonstrated in intact tissues (3,4) and in studies with intact mitochondria (5,6). Studies with purified complex (7)(8)(9) and isolated mitochondria (10 -12) found that increases of the PDC reaction products promote an increase in the proportion of PDC in the phosphorylated (inactive) state. With purified complex, the activity of the kinase is greatly enhanced upon elevation of the NADH: NAD ϩ ratio and the acetyl-CoA:CoA ratio along with a reciprocal reduction in phosphatase activity as the NADH:NAD ϩ ratio is elevated (7).
Accordingly, the E1a kinase has a crucial role wherein it throttles down PDC activity in response to increases in the mitochondrial acetylation and reduction potentials. Not only are NADH and acetyl-CoA produced during mitochondrial oxidation of all fuels, they are direct products of the PDC reaction. This laboratory presented evidence that this control of kinase activity initially involves these product to substrate ratios being translated by competitive utilization in the downstream reactions of the complex which, in turn, adjusts the fraction of the complex's lipoyl prosthetic groups in the oxidized versus reduced versus acetylated forms (13)(14)(15)(16)(17). Specifically, our mechanism proposes that kinase down-regulates PDC activity due to NADH reacting in the reverse of the dihydrolipoyl dehydrogenase (E3) reaction and acetyl-CoA reacting in the reverse of the dihydrolipoyl acetyltransferase (E2) reaction. Typically a 60 -80% enhancement in kinase activity occurs upon lipoate reduction and up to a 3-fold enhancement following lipoate acetylation. The potential for understanding the molecular basis of this control has greatly improved with new insights into the structure of E2 subunits and the unusual nature of the association of the kinase with E2.
The E2 subunits of mammalian PDC have four domains connected by relatively large (2-3 kDa) and highly mobile linker regions Fig. 1 (18 -20). Sixty COOH-terminal inner domains (E2 I ) associate to form a dodecahedral inner core which catalyzes the transacetylation reaction. Each inner domain is connected to 3 globular domains by linker or hinge regions. The globular domains consist of two ϳ10-kDa lipoyl domains (E2 L1 and E2 L2 , or L1 and L2) and an E1-binding domain (E2 B ) located between the inner core and the lipoyl domain region.
The subunits of E1a kinases of PDC and the branched-chain ␣-keto acid dehydrogenase complex are related to procaryotic histidine kinases but not to the extramitochondrial serine and tyrosine kinases of eukaryotes (21)(22)(23). PDC kinase binds to the lipoyl domain region of E2 through an association that requires the lipoyl prosthetic group (24 -26). Using lipoylated and delipoylated forms of recombinant L1 and L2 of human PDC-E2 (27), the kinase was shown to bind preferentially to the lipoylated L2 ( Fig. 1) (26). The E1a kinase was also shown to interchange rapidly between L2 structures (26). To account for tight binding and rapid interchange, a dynamic "hand over hand" mechanism is proposed in which a dimeric kinase alternates between being bound to one and two L2 domains. This interchange and catalytic function of the kinase exert their combined effects in the limited space at the surface of the complex where the a kinase molecule and many E1 components are tightly bound to the mobile outer domains of E2. The capacity of continuously bound kinase to phosphorylate bound E1 components more rapidly than free kinase can phosphorylate free E1 is termed E2-activated kinase function.
The aim of this work is to determine the compulsory components, domains, and catalytic processes involved in kinase stimulation by NADH and acetyl-CoA, and to evaluate the relative capacities of the recombinant L1 and L2 in mediating the stimulation of kinase in the presence or absence of E2activated kinase function. We have found that stimulation occurs with a peptide substrate of the kinase; that a lipoyl source must be available for catalytic reduction by E3 or acetylation by E2 I ; that the L2 lipoyl domain is much more effective in mediating kinase stimulation than the L1 domain; and that the kinase is remarkably sensitive to the level of acetylation of L2. The "Discussion" integrates these observations and draws new mechanistic conclusions based on these and previous results.

EXPERIMENTAL PROCEDURES
Materials-Standard procedures were used to prepare: bovine kidney pyruvate dehydrogenase complex (28), the E2⅐X⅐K1K2 subcomplex (29), E1 component, the bilipoyl domain region (E2 L fragment) of bovine E2 (30). The inner core of bovine E2, E2 I , was prepared using trypsin treatment followed by pelleting through a sucrose layer in the presence of 5 g/ml soybean trypsin inhibitor (31). The individual lipoyl domains, L1(1-98) and L2(120 -233) of human E2 were expressed in Escherichia coli as fully lipoylated structures fused to glutathione S-transferase and purified to homogeneity as fusion protein free domains as described by Liu et al. (27). Lipoamidase from Enterococcus faecalis was prepared by modification of the procedure of Suzuki and Reed (32). Lipoate protein ligase of E. coli was overproduced in E. coli strain TM202 which contains the expression plasmid pTM70, and purified as described (33). The E3 component from porcine heart was from Sigma. Kinase pentadecapeptide substrate, YHGHS 1 MSDPGVS 2 YRT with the sequential sites of phosphorylation designated by S 1 and S 2 , was prepared in the Biotechnology core facility at Kansas State University and shown to give a single symmetric peak in reverse phase HPLC, to have the correct sequence by automated Edman degradation, and correct mass by FAB mass spectrometry. [␥-32 P]ATP, [1-14 C]acetyl-CoA, and [2-14 C]pyruvate were obtained from DuPont NEN.
Kinase Activity-E1a kinase activity was measured in duplicate or triplicate as the initial rate of [ 32 P]phosphate incorporation from [␥-32 P]ATP (250 to 500 cpm/pmol) into E1a tetramers in PDC or as the resolved E2-free component. Most assays were conducted in a final volume of 50 l of buffer A: 50 mM MOPS-K, 20 mM potassium phosphate, 60 mM KCl, 2 mM MgCl 2 , 0.4 mM EDTA, and 0.4 mM dithiothreitol (16,17). Unless otherwise indicated, assays used 25-30 g of PDC or 25 g of E1 and 1 g of E3. Some resolved E1 preparations have more E1a kinase; here such a preparation was a source of stable, E2-free kinase. 2 The specific activity of E1-kinase was 0.44 nmol/min/mg when assayed in 20 mM potassium phosphate in the absence of E2; this indicates kinase represents about 0.4% of the protein in this E1 fraction. Unless otherwise indicated, the levels for the following reactants were: NADH and NAD ϩ at 3:1 ratio with final concentrations of 0.6 mM NADH and 0.2 mM NAD ϩ and acetyl-CoA or pyruvate at 50 M, L1 or L2 at 16 M, dihydrolipoamide at 20 to 200 M, and E2 I at 2 g. Dihydrolipoamide was dissolved in H 2 O containing 0.2 mM EDTA by warming the solution to 60°C; it was critical that the stock solutions (0.2-2.0 mM) be maintained at room temperature (i.e. not placed on ice) to retain this ligand in solution. The latter was monitored with the E3 cycling reaction below. ATP was added last except in one experiment in which a concentrated fraction of delipoylated E2⅐X⅐K1K2 subcomplex components. E2 has four globular domains consisting of an NH 2terminal lipoyl domain, L1, an inner lipoyl domain, L2, an E1 binding domain, and a core-forming, transacetylase-catalyzing inner domain at the COOH-terminal end (18 -20). The kinase binds to the L2 domain by a domain-specific and lipoyl prosthetic group-requiring interaction (26). E1, an ␣ 2 ␤ 2 tetramer, binds to the B domain of E2 via its ␤-subunit (18).
was treated with ATP and lipoyl protein ligase and the subsequent kinase assays were performed without removal of residual ATP. In the latter series of assays, including those with parallel samples of untreated and delipoylated subcomplex, E1-kinase was added last to initiate kinase activity. For effective reductive acetylation of E2 lipoyl domains by E1, E1 was preincubated with 50 M TPP and diluted 10 or 20-fold into kinase assays lacking additional TPP. Kinase (and acetylation) assays conducted with CoA removal were performed using the conditions described under acetylation studies. All control assays were conducted in triplicate or quadruplicate and other assays at least in duplicate. In these and other assays the absolute deviations are shown. Reactions were terminated by spotting samples to dry trichloroacetic acid-containing Whatmann No. 3MM with E1 as the substrate or to Whatmann P81 phosphocellulose paper disks with the peptide substrate, followed by steps previously described (16,25).
Extent of Acetylation-The degree of acetylation of the lipoyl prosthetic groups from various sources of E2 lipoyl domains were measured using either the E1 reaction with [2-14 C]pyruvate as substrate (15,17) or E3/NADH reduction followed by acetylation with [1-14 C]acetyl-CoA (15)(16)(17). For acetylation with pyruvate, a concentrate of E1 ϩ 50 M TPP was prepared and used as described above. In some experiments, the acetylation of E2 or lipoyl domains using acetyl-CoA was driven to completion by converting the CoA product to succinyl-CoA with the ␣-ketoglutarate dehydrogenase complex in reaction mixtures containing 0.25 mM NAD ϩ and 0.25 mM NADH, 0.4 mM ␣-ketoglutarate and 50 M Ca 2ϩ (16). This gave a maximal acetylation of 16 M L1 or L2 with high acetyl-CoA (e.g. Ն24 M), whereas at lower levels of acetyl-CoA (e.g. 0.5-16 M), acetyl-CoA was completely used, producing an extent of acetylation of lipoyl domains equivalent to added acetyl-CoA. The acetylation of intact E2 subunits was measured by spotting samples on dry trichloroacetic acid-containing paper disks or on P81 phosphocellulose disks (as with the kinase assays); P81 disks were always used to measure acetylation of the L1 or L2 domains. The washing and counting procedures were as described previously (16,27).
Lipoamidase Treatment-Lipoyl groups were released from PDC, E2⅐X⅐K1K2, and isolated lipoyl domains by incubation with E. faecalis lipoamidase for 90 -210 min at 30°C or overnight at room temperature (24 -27). The incubation mixture contained 50 mM potassium phosphate (pH 7.5) and 1 mM EGTA and ratios of lipoyl-bearing structure to lipoamidase (w/w) 345:1 for PDC; of 180:1 for E2⅐X⅐K1K2; and of 100:1 for the human L2 domain. The extent of delipoylation was monitored by measuring either residual PDC activity or reconstituted PDC activity (below). A steep decline occurred only as the final 40% of lipoyl groups were removed from E2 and protein X. Final products were evaluated for their content of lipoyl groups by measuring the extents of acetylation by the CoA removal approach (described above). Delipoylation of isolated lipoyl domains was estimated from the loss of activity in the E3 cyclic assay. Free lipoate caused only a minor interference with the latter assay since lipoyl domains are much more efficient substrates 3 ; full delipoylation was judged to have occurred when the rate had dropped to that produced by an equivalent concentration of free lipoate. Delipoylation of L1 and L2 was also evaluated by the characteristic shift (mobility decrease) observed for these delipoylated domains in native gel electrophoresis (27). All delipoylated preparations were examined by SDS-PAGE to ensure that no subunit degradation had occurred. In experiments in which delipoylation was followed by ligase treatment (below), the delipoylated preparations were treated with 0.4 mM PMSF for 15 min at 30°C to terminate lipoamidase activity, and residual PMSF quenched with 0.5 mM dithiothreitol.
Ligase Procedure-Relipoylation was conducted by incubating delipoylated samples in a 25 mM potassium phosphate buffer, pH 7.5 containing lipoyl protein ligase, 50 M lipoic acid, 0.5 mM ATP and 2 mM MgCl 2 . The acetylation and E3 assays (described above) were used to show that lipoyl moieties had been incorporated. Usually, excess ATP was removed by dialysis versus 20 mM potassium phosphate buffer, pH 7.0.
PDC Activity Studies-The effects of delipoylation or relipoylation of PDC and E2⅐X⅐K1K2 samples were deduced from their PDC activities (34). For E2⅐X⅐K1K2 subcomplex, 4 g subcomplex was combined with 8 g of E1 and 4 g of E3 in 10 l and incubated for 60 s at 30°C, and a 6-l sample was added to a standard PDC assay. Since removal of lipoyl groups has a disproportionately low effect on PDC activity at early stages, delipoylation is analyzed more accurately by determining changes in acetylation capacity (above).
The fractional binding of TPP to E1 can be evaluated in PDC activity assays because bound TPP is continuously converted to the more tightly bound hydroxyethylidene-thiamin pyrophosphate intermediate, the substrate of the rate-limiting reductive acetylation step (35,36). The levels of TPP bound to E1 were evaluated by comparing the rates of PDC reaction obtained by diluting enzyme 200-fold as the last addition into assay mixtures lacking or containing TPP. Reconstituted PDC activity was measured with resolved E1 and excess E2⅐X⅐K1K2 subcomplex (10 g) and E3 (4 g) and limiting E1 (4 g). The fractional E1-TPP is defined as the ratio of the highest activity observed in the absence of TPP to that observed in the presence of TPP.
The effects of changes in NADH:NAD ϩ and acetyl-CoA:CoA ratios on PDC activity were analyzed in the absence of ATP in the same high K ϩ buffer (buffer A) used to evaluate the regulation of kinase activities rather than the standard 40 mM potassium phosphate buffer used in routine PDC activity assays. Pyruvate and TPP were introduced at saturating levels of 1.0 mM and 0.2 mM, respectively. When the relative amount NADH ϩ NAD ϩ were varied while retaining a total concentration of constant 0.5 mM, PDC activity was measured at 365 nm (⑀ ϭ 4.2 OD⅐mM Ϫ1 ⅐cm Ϫ1 rather than at 340 nm. Other conditions were as described in the legend to Fig. 8.

E1, E1-TPP, and Peptide
Substrates-The requirements for NADH and acetyl-CoA stimulation of E1 kinase activity were evaluated via phosphorylating various substrates using the E2⅐X⅐K1K2 subcomplex as a source of kinase activity. As in assays below, NAD ϩ and NADH were used at a 1:3 ratio to prevent inhibition of E3 by its conversion to the 4 e Ϫ -reduced state. At this ratio, kinase activity is near maximally stimulated by NADH and maximally stimulated by the combination of NADH and acetyl-CoA. We used an E1 preparation that gave Ͻ1% activity in a PDC reaction mixture lacking TPP. As shown in Table I, kinase activity was stimulated 50% by NADH and nearly 2-fold by the combination of NADH and acetyl-CoA in the phosphorylation of E1 whether or not E1 contained TPP cofactor. As expected (37), kinase activity was reduced by TPP. However, the stimulation has been due to reversing the inhibitory effect of TPP or activation via the formation of a reaction intermediate on E1 (e.g. hydroxyethylidenethiamin pyrophosphate formed from acetyl-lipoate by E2).
Table I also shows that E1 can be replaced by peptide substrate so the stimulations are not dependent on changes in E1 (38). 4 This is consistent with the observation of Reed et al. (40) that kinase stimulation occurs with tryptic digestion of PDC. Such PDC digestion produces a complex mixture of lipoyl domains, the catalytic inner core of E2 and the trypsin-resistant E3, all of which are required for kinase stimulation (see below).
Because some stimulation by acetyl-CoA is detected in the absence of NADH, the level of incorporation of covalently attached acetyl groups from [1-14 C]acetyl-CoA was determined in the presence or absence of unlabeled ATP under the conditions of kinase assays (Table I). Without NADH, exposure to submillimolar levels of dithiothreitol allows acetylation of a small portion of the lipoyl groups by acetyl-CoA. The low stimulation of kinase that occurs in conjunction with the low level of acetylation in the absence of NADH agrees with previous studies (15-17) (cf. "Discussion"). Thus, our results eliminate the possibility that kinase stimulation is due to changes in the form of TPP interacting with the E1 substrate but they are consistent with the participation of lipoyl prosthetic groups.
Pyruvate Stimulation-Stimulation by pyruvate was first reported by Cooper et al. (41). Support for this stimulation occurring via reductive-acetylation of lipoyl moieties stems from the findings that adding low levels of pyruvate to TPPcontaining PDC generates an increased kinase activity equivalent to that with NADH plus acetyl-CoA and that pyruvate is principally used in the reductive acetylation of oxidized lipoyl groups (13)(14)(15). This mechanism for kinase stimulation predicts some TPP bound to E1 and a lipoyl domain source are required. Using a resolved TPP-dependent E1 preparation, containing low levels of kinase activity but completely free of E2, the kinase activity was not significantly enhanced by pyruvate and/or lipoyl domain in the absence of TPP (Table II). Even in the presence of TPP, pyruvate failed to stimulate kinase activity with E1 unless lipoyl domains were added. Furthermore, a fragment of bovine E2 that contains both lipoyl domains (designated E2 L ) was effective. The results are thus consistent with the view that the enhancement of kinase activ-ity is mediated by the E1(TPP)-catalyzed acetylation of the E2 L fragment. Furthermore, a very low level of E2 L fragment (1 M) was effective. This is below the level that gave a small enhancement of the kinase (up to 40%) (25), suggesting a specific and possibly a stronger interaction of the kinase with acetylated-E2 L .
Lipoamidase Treatment of PDC-The proposed mechanism of lipoyl-mediated enhancement of kinase activity predicts that complete removal of lipoyl groups would abolish kinase stimulation. Lipoyl groups were removed by treatment of PDC, E2⅐X⅐K1K2 subcomplex, or isolated lipoyl domains with E. faecalis lipoamidase. The enhancement in kinase activity by the E2 component (in the absence of effectors) declines rapidly as the last portion of lipoyl groups is removed and the kinase tends to dissociate from E2 (24). Treating PDC for 150 min with lipoamidase (1 g/345 g of complex) produced a preparation of complex, PDC 1 ⅐D L , lacking Ͼ96% of the lipoyl groups, based on the maximal acetylation assay (see "Experimental Procedures"). Delipoylation reduced kinase activity from 6.0 to 2.4 nmol min Ϫ1 mg Ϫ1 PDC, i.e. it removed most of E2 enhancement of kinase function (Fig. 2, PDC 1 series). Surprisingly, PDC 1    could still be stimulated nearly 2-fold by NADH plus acetyl-CoA but the activity remained well below the fully lipoylated PDC 1 control (Fig. 2). The delipoylated PDC 1 ⅐D L incorporated Ͻ5 acetyl groups per molecule of complex (M r Ӎ 7.5 ϫ 10 6 ) under the same conditions. The retention of only a few lipoyl groups allowing effector stimulation can be explained by the lipoyl prosthetic group being a critical part of the high affinity binding site of the kinase and the kinase moving between lipoyl domains (25,26). This would allow the kinase to relocate to E2 subunits retaining lipoyl groups with the residual activity of this localized kinase still being enhanced by reductive acetylation of the remaining lipoyl groups (cf. "Discussion"). Using a 210-min lipoamidase treatment of PDC lowered the level of acetylation 1.15 Ϯ 0.3 in PDC 2 ⅐D L from 162.8 5 for the untreated PDC 2 control, and abolished the stimulations of kinase activity by NADH and/or acetyl-CoA relative to the control (Fig. 2). The low stimulation of control kinase of PDC 1 or PDC 2 by acetyl-CoA in the absence of NADH was associated with a low level of acetylation of lipoates reduced during storage of the complex in 0.5 mM dithiothreitol. SDS-PAGE analysis of the lipoamidase treated PDC 1 or PDC 2 showed no changes in the protein pattern indicating that there was no proteolysis during the lipoamidase treatments.
Lipoamidase and Lipoyl Protein Ligase Treatment of E 2 ⅐X⅐K1K2-The requirement for lipoate could be confirmed by reattaching the cofactor but E1 must be absent to avoid its phosphorylation in the ATP-dependent lipoyl-protein ligase reaction. E2⅐X⅐K1K2 was incubated with or without lipoamidase for 210 min and then lipoamidase-inactivated with PMSF as described under "Experimental Procedures" and the legend to Fig. 3. Acetylation per subcomplex was reduced from 165.3 to 1.89 per subcomplex using acetyl-CoA with CoA removal approach and reconstituted pyruvate dehydrogenase complex activity was decreased by Ͼ99%. Portions of the delipoylated and control subcomplexes were treated with lipoate, ATP, and lipoate protein ligase (as described under "Experimental Procedures") and dialyzed to remove ATP and lipoate. The acetylation capacity increased to 117.3 per subcomplex by 60 min, declined slightly thereafter, and remained below untreated subcomplex (Ͼ160 acetyl incorporated/subcomplex). In a parallel study, after lipoamidase and then lipoate protein ligase treatments, the E2⅐X⅐K1K2 subcomplex gave a reconstituted PDC activity of 30.6 mol/min/mg which was nearly as high as the untreated control (31.2 mol/min/mg). SDS-PAGE analysis did not detect any change in the profile for the subunits of the 5 This level of acetylation requires some biacetylation of lipoates. This tends to occur under conditions in which acetyl-CoA Ͼ Ͼ lipoyl domains of the complex, particularly when the CoA removal approach is used to prevent deacetylation. As acetylation is increased from a level expected to give primarily monoacetylated lipoates (some reduced and some biacetylated) to a level expected to give a much higher proportion of biacetylated lipoates, there is a concomitant decrease in the number of sites that can be alkylated (by N-[ 14 C]ethylmaleimide) supporting formation of biacetyl-lipoates (S. Rahmatullah and T. E. Roche, unpublished results).

FIG. 3. Effects on kinase activity and regulation of enzymatic removal and restoration of lipoyl groups of the E2⅐X⅐kinase subcomplex. The
E2⅐X⅐kinase subcomplex was treated with lipoamidase for 210 min, and the lipoamidase was inactivated with PMSF as described under "Experimental Procedures." Lipoyl groups were then restored to a portion of this preparation using E. coli lipoyl protein ligase and then substrates removed as described under "Experimental Procedures." The control sample of subcomplex was given parallel incubations. Kinase assays included 25 g of E1; pyridine nucleotides were added 60 s and acetyl-CoA 20 s prior to [␥-32 P]ATP, and the reactions were terminated after 60 s.

FIG. 4. Capacity of the L1 and L2
domains of E2 to support effector stimulation of kinase activity. Kinase assays were conducted with 25 g of E1 containing low kinase, and, as indicated, human recombinant L1 or L2 domains were included at 16 M. Effectors were added at the levels and times given in Fig.  3. The solid bar shows kinase activity in the absence of L1 or L2. Other conditions were as described under "Experimental Procedures." subcomplex. Delipoylation of the subcomplex reduced kinase activity and abolished the rate enhancing effects of NADH and acetyl-CoA (Fig. 3), but these were to a large extent restored following ligase treatment.
Lipoyl Domain Role and Requirement for Acetyltransferase Reaction-To investigate the relative effectiveness of E2's lipoyl domains, the capacity of recombinant human lipoyl domains (L1 and L2) in mediating kinase stimulation was evaluated with resolved E1 in the presence of E3 and in the presence or absence of lipoyl domain-free E2 I (Fig. 4). Kinase activity was increased slightly by NADH with the L1 domain but to a much larger extent by the L2 domain. In the absence of E2 I , acetyl-CoA had no effect, alone, nor did it enhance the effect of NADH alone. In the presence of acetyltransferasecatalyzing E2 I , kinase activity was stimulated by the combina-tion of NADH and acetyl-CoA with the L1 domain, but not to the extent achieved with just NADH with the L2 domain (Fig.  4). An even larger (nearly 3-fold) enhancement of kinase activity was observed when a combination of effectors was added with L2 in the presence of E2 I . Since there seems to be no possible role of E2 I other than catalyzing the transacetylation reaction, the results provide definitive support for the view that the stimulating effects of acetyl-CoA are mediated by lipoate acetylation. In the L2 ϩ E2 I series (Fig. 4), the enhancement by acetyl-CoA, alone, occurred with acetylation of L2 (below). The results establish a preferential role for L2 in mediating kinase stimulation in keeping with the preferential binding of the kinase to the L2 domain (26) (cf. "Discussion").
In parallel experiments, the levels of acetylation of lipoyl domains under the conditions used for studying kinase stimulation were evaluated. With the L2 domain at 20 s (time of ATP addition after acetyl-CoA addition) and at 80 s (the time the kinase reactions were terminated) 0.11 and 0.25 acetyl groups were incorporated per L2 in the presence of E2 I but in the absence of NADH. Whereas in the presence of NADH/NAD ϩ , slightly more than one acetyl group per L2 was incorporated at 20 and 80 s. Similar levels of acetylation occurred with L1. This indicates that there is a low level of biacetylation (i.e. at the 6and 8-positions of a lipoate) as previously observed with the recombinant L1 and L2 constructs (27).
Degree of Acetylation of L2 Versus the Extent of Stimulation of the Kinase-The above results suggest a progression in kinase stimulation with L2 acetylation. This relationship was evaluated studying kinase activity and acetylation, in parallel, under conditions of controlled acetylation by the acetyl-CoA being completely consumed (Fig. 5). Low concentrations of acetyl-CoA were incubated with the L2 domain in the presence of E2 I , E3, and a 1:1 ratio of NADH:NAD ϩ along with the ␣-ketoglutarate dehydrogenase complex plus ␣-ketoglutarate to convert CoA produced into succinyl-CoA. The measured extents of acetylation confirmed that, at all but the highest level (24 M), all the [1-14 C]acetyl-CoA was consumed in acetylating the L2 domain (16 M). The enhancement of kinase activity increased with the level of acetylation with a half-maximal stimulation observed when ϳ20% of the L2 were acetylated (Fig. 5). Though a low proportion of free L2, this is higher proportion of L2 acetylated than is required in the intact complex as explained under "Discussion." The decline in stimulation at high levels of acetylation is also consistent with the latter studies with intact complex (15)(16)(17).

Effects of Dihydrolipoamide and Acetyl-dihydrolipoamide and Requirement for a Lipoyl Source for NADH Stimulation of
Kinase Activity-Previously reported stimulation of PDC-kinase by dihydrolipoamide (13,14) could involve reduction of lipoyl moieties in the complex through E3 catalysis or disulfide exchange. We were unable to do experiments in the absence of E3 because of trace E3 in our E1 and resolved kinase preparations. (E3 is very active enzyme and only one round of turnover is needed to reduce the lipoate of L1 or L2.) With no lipoyl domain source present, E1-kinase was increased by dihydrolipoamide with a maximal stimulation at 50 M of 40 -55%. Previously Liu et al. (26) found L1 and L2 inhibited kinase activity at levels greater than 20 M. For direct comparison, effects of lipoate sources (dihydrolipoamide, L1, and L2) at 20 M, alone and in combination, were determined to evaluate the direct and supportive roles of dihydrolipoamide. Addition of 20 M dihydrolipoamide to E1-kinase without a lipoyl domain source or with L1 or L2 gave small but reproducible enhancements (15-40% ϮL1; 30 -60% ϩL2) of kinase activity (Fig. 6,  open bars). NADH did not stimulate in the absence of a lipoate source (first series, solid bar); indeed, the mixture of NADH ϩ NAD ϩ at 3:1 ratio seemed to inhibit kinase activity. This strongly supports the conclusion that E3 catalyzed reduction of a lipoate source is required. Only with L2 did NADH increase kinase activity beyond that with just dihydrolipoamide. In the presence of L2 (Fig. 6, third series), larger effects were observed with NADH than with dihydrolipoamide and the combination had no effect beyond NADH alone. These results indicate that the higher stimulation by dihydrolipoamide in the presence of L2 results from reduction of the lipoyl group of L2.
Dihydrolipoamide alone was effective in supporting acetyl-CoA stimulation. It was important that acetylation of dihydrolipoamide immediately precede the kinase assay since stimulation was not detected with stored preparations in which acetyl groups transfer from L-isomer to incorrect D-isomer and from 8-position to 6-position of the dihydrolipoamide. Dihydrolipoamide replaced NADH in supporting the higher acetyl-CoA stimulation with L1 and particularly with L2 (diagonally crosshatched bars). Acetyl-CoA stimulation again required E2 I (minus E2 I results not shown). For the marked stimulation by acetyl-CoA plus L2, dihydrolipoamide likely reduced enough L2 lipoates in the presence of E3 to allow L2-acetylation to an extent giving near maximal stimulation. Indeed, the 4.7-fold enhancement was greater than with NADH probably due to the extent of acetylation being somewhat lower, resulting in less diacetyl-L2. Although reduction and acetylation of the lipoate of L2 is clearly most effective, a direct stimulation of kinase activity is observable with dihydrolipoamide and acetyl-dihydrolipoamide and this limits mechanistic possibilities (cf. "Discussion").
Effect of Delipoylation and Relipoylation of L2 Domain-To test the importance of L2's lipoate while not exposing the kinase to lipoamidase and lipoate protein ligase, delipoylated and relipoylated preparations of L2 were made. Delipoylation of the L2 domain abolished its capacity to potentiate acetyl-CoA stimulation of the kinase and ligase treatment restored the full capacity for stimulation (Fig. 7). The previous lack of full recovery in the ligase treatment of delipoylated E2⅐X⅐K1K2 (Fig. 3) was probably due to loss or modification of some of the kinase during these treatments. The requirement for L2's lipoyl group and for E2 I catalysis seem to constitute incontrovertible evidence that kinase stimulation is mediated by acetylation most effectively via the L2 domain of E2.
Direct Product Inhibition of the PDC Reaction-The mechanism of product inhibition of PDC under low salt conditions (43) and the effects of higher ionic strengths on the PDC reaction have been characterized (44). However, at physiological K ϩ levels, the effects of varying NADH:NAD ϩ and acetyl-CoA:CoA ratios have been characterized for the kinase reaction (27) but not the PDC reaction. The results in Fig. 8 show that halfmaximal inhibition of PDC is observed at an NADH:NAD ϩ ratio of 0.42 (ࡗ) in the absence of acetyl-CoA; and at an acetyl-CoA:CoA ratio of 11.2 (f) in the absence of NADH and at 3.55 in combination with a fixed NADH:NAD ϩ ratio of 0.1 (Ç). The strongest direct product inhibition of PDC was found with a half-maximal effect at a low NADH:NAD ϩ ratio of 0.125 when the acetyl-CoA:CoA ratio was held at 1 (Fig. 8, E). However, near-maximal stimulation of the kinase to switch off bovine kidney PDC occurs with an NADH:NAD ϩ ratio of 0.1 at a 10-fold lower acetyl-CoA:CoA ratio. We conclude that much higher ratios of products to substrates are required for direct product inhibition of bovine kidney PDC than for speeding PDC inactivation by enhancing kinase activity (cf. "Discussion"). DISCUSSION Randle and co-workers (45,46) pointed out the importance of feedback control of PDC in satisfying metabolic needs and presented evidence for direct product inhibition of the PDC reaction. After regulatory interconversion of PDC between ac-FIG. 7. Effects of removal and restoration of the lipoyl groups of L2 on kinase stimulation by NADH and acetyl-CoA. The L2 domain was delipoylated with lipoamidase and relipoylated with lipoyl protein ligase using the conditions described under "Experimental Procedures." Kinase assays were conducted as described in the legend to Fig. 4 and under "Experimental Procedures," except that L2 was unchanged, delipoylated, or relipoylated, as indicated. The acetyl-CoA:CoA ratio was varied at a constant total pool of 0.2 mM in presence of 1.0 mM NAD ϩ (f) or at a 0.1 NADH:NAD ϩ ratio using 1.1 mM total pyridine nucleotides (Ç); the NADH:NAD ϩ ratio was varied with a constant total pool of 0.5 mM with 0.2 mM CoA (ࡗ) or at an acetyl-CoA:CoA ratio of 1 (E) using 0.1 mM of each. In each case 100% rates were the rate observed at a product to substrate ratio of zero for the varied ratio. Other conditions and assay procedures are described under "Experimental Procedures." tive and inactive forms (2) became known, Randle's laboratory contributed to evidence (cf. Introduction) that the products of PDC reaction influence the proportion of PDC in the active form, through diverse studies with purified complex isolated mitochondria, and intact tissues (1,8,9,47). In their studies with purified porcine heart PDC, no stimulation by acetyl-CoA beyond that of NADH was detected, and they speculated that enhanced kinase activity might be due to the removal of an inhibition of kinase activity caused by the absence of an interaction between oxidized lipoate and the E1 substrate (8,9). Although not the correct mechanism and detection of an effect of acetyl-CoA required use of higher K ϩ concentration (see below), the suggestion of a role for changes in the intermediate status of lipoyl groups was insightful as was their linking of the stimulatory effect of pyruvate (41) to the effect of products (9). This laboratory presented the initial and much subsequent evidence that reduction and acetylation of lipoyl prosthetic groups of PDC constituted essential steps in the operation of a sensitive signal translation process whereby increases in the NADH:NAD ϩ and acetyl-CoA:CoA ratios markedly enhance kinase activity (13)(14)(15)(16)(17). However, elucidation of the minimal requirements for the operation of this control could be accomplished only following a greatly enhanced understanding of the organization of the complex, the preparation of individual lipoyl domains of PDC-E2, and the availability of lipoamidase and lipoyl protein ligase, combined with the selective use of peptide substrate and free forms of lipoamide.
Using these tools, we have found that the marked enhancement of kinase activity by acetyl-CoA requires a lipoate source and its reduction, the catalytic domain of E2, and a peptide substrate. These requirements conclusively support a change in the kinase mediated by catalytically forming acetyl-dihydrolipoate. Accumulation of this intermediate explains the similarly strong stimulation of kinase activity by low pyruvate via E1(TPP) catalysis. The direct stimulation by dihydrolipoamide, its capacity to replace NADH in potentiating acetyl-CoA stimulation, and the complete lack of an effect of NADH on kinase activity in the absence of a lipoate source strongly favor NADH boosting kinase activity through its use in the E3 reaction. We have established that this intervention is abolished by complete removal of lipoates from intact E2 subunits or lipoyl domains. Of particular importance for understanding the operation and regulation of the kinase is our finding that the kinase-binding L2 domain is much more effective than E2's L1 domain or lipoamide in mediating kinase stimulation. The 3-fold increase in kinase activity generated by reductive acetylation of recombinant L2 is comparable to the change in kinase activity produced by acetylation of lipoyl moieties in the intact complex. Reduced L1 was only slightly more effective than free dihydrolipoamide when compared at 20 M, and 50 M dihydrolipoamide gave at least an equivalent stimulation, suggesting the structure of L1 does not contribute significantly to these effects. At higher levels L1 inhibits kinase activity (26). Thus, the L2 domain must have structural features that facilitate its dedicated roles in binding the kinase, producing enhanced E1 phosphorylation within the confines of the complex, and further increasing kinase activity upon reduction and acetylation of its lipoyl group. This raises the question of the linkage between these L2-supported actions of the kinase.
Detaching lipoates of the E2 60 core removes the capacity of the E2 core to give a severalfold increase in E1 phosphorylation (24). Via lipoyl-dependent binding to the L2 domain, rapid "hand over hand" interchange of a kinase dimer between lipoyl domains apparently eliminates constraints normally associated with binding that is as tight as exists between E2 and the kinase (25,26). It is significant that a lipoamidase treatment of PDC, which left only a few lipoates and caused close to full loss of this E2-enhanced kinase function, still allowed marked stimulation of the residual kinase activity upon acetylation of those few domains retaining lipoyl groups (PDC 1 series in Fig. 2). This result forcefully suggests that the kinase has moved to the limited number of lipoylated L2 domains and, furthermore, that their acetylation (during a short-lived dissociation and reassociation of the kinase) facilitated kinase stimulation. Based on this continued stimulation when very few E2 subunits retain lipoate, we hypothesize that maximal effector stimulation is mediated by an allosteric effect induced by a reductively acetylated L2 domain that becomes engaged in binding of the kinase (i.e. not by interaction of an acetylated prosthetic group on a neighboring lipoyl domain that is not engaged in binding the kinase). Such a highly specific interaction is further supported below.
Beside E1 catalyzed reductive-acetylation giving a nearly equivalent stimulation to acetyl-CoA (13)(14)(15), ␣-ketobutyrate (38) gives a lesser stimulation similar to that of propionyl-CoA at low levels of propionylation (16). In the absence of a lipoyl domain source, kinase phosphorylation of E1 is not stimulated at low pyruvate (Ref. 17, miniprint section); and here we show FIG. 9. Model of proposed steps in kinase stimulation. The proportion of L2 domains of the E 2 60 core having oxidized, reduced, or acetylated lipoates responds to changes in the NADH:NAD ϩ ratio and acetyl-CoA:CoA ratio via the rapid and reversible reactions catalyzed by E3 and E2, respectively. Our model proposes that a change from the kinase binding to one or two L2 containing only oxidized lipoate (nonstimulated K state) to interacting with an L2 containing a reduced or an acetylated lipoate results in the modified lipoates allosterically inducing conformational changes that generate the progressively more active K* or K** states, respectively. that TPP and a lipoyl domain source are needed to achieve pyruvate stimulation. These findings raise the question as to whether the E1 reaction or the downstream (E2,E3) reactions determine the relative proportion of lipoates in the oxidized, reduced and acetylated forms in the complex. Since E1 catalyzes the rate-limiting step (35,36), under most metabolic conditions, this distribution should be determined by the NADH:NAD ϩ and the acetyl-CoA:CoA ratios and kinase activity should primarily respond to their fluctuation.
With purified porcine liver and bovine kidney PDC, halfmaximal stimulations of kinase activities occur at ratios of NADH:NAD ϩ of 0.03 and 0.05, respectively, and, with 2 mM dithiothreitol as a reducing agent, at ratios of acetyl-CoA:CoA of 0.1 and 0.17, respectively (28). Even lower acetyl-CoA:CoA (Ͻ0.1) give half-maximal stimulation when a NADH:NAD ϩ ratio of 0.1 is provided to reduce lipoates (data not shown). At these low ratios, there is minimal product inhibition of the PDC reaction by NADH plus acetyl-CoA (cf. the 0.1 value on the Ç curve, Fig. 8). Significant direct product inhibition occurs with low NADH:NAD ϩ ratio when the acetyl-CoA:CoA ratio was held at a relatively high level of 1 (Fig. 8, E). Acetyl-CoA:CoA ratios have been generated at and above this range with isolated mitochondria and are proposed to contribute to direct PDC inhibition (48); depletion of CoA may also inhibit PDC due to buildup of fatty acyl-CoAs (49). While these probably contribute to a fine tuning role on a small portion of active PDC, enhanced phosphorylation should effectively throttle down PDC activity before onset of these extremes.
A priori, the lesser enhancement of kinase activity by just the conversion of lipoates from the oxidized to the reduced form could involve either removal of an inhibitory interaction of oxidized lipoates, an enhancement of kinase activity by gain of a positive allosteric interaction of the reduced form, or a change in the thiol disulfide state of a kinase subunit. Studies with E2-bound kinase found that the kinase is sensitive to thiol reagents (13,52), effects probably due to changes in the oxidation-reduction state of the lipoate of L2 domains. Although there is a small increase in alkylation of cysteines of the K1 subunit following treatment of E2⅐X⅐kinase-E3 with NADH, this was not rapidly reversed by excess NAD ϩ which opposes kinase stimulation. 6 Thus, a mediator role for changes in the thiol-disulfide status of the kinase is not indicated. Furthermore, our finding that free dihydrolipoamide and to a greater extent acetylated dihydrolipoamide directly stimulated kinase activity in the absence of a lipoyl domain source, eliminating the prospect that stimulation results from removal of an inhibitory effect by the oxidized form of lipoate. As modeled in Fig. 9, we conclude that kinase activity is enhanced through direct positive allosteric interactions of the reduced or the acetylated form of an L2 lipoate and we further suggest that L2 specificity derives from the L2 domain engaged in binding the kinase.
Consistent with stimulation occurring at low ratios of acetyl-CoA to CoA, half-maximal and near-maximal stimulation of kinase associated with the intact bovine kidney PDC are achieved with only 7-10 and 22-26 acetyl groups incorporated, respectively, per PDC (15)(16)(17). With the recent knowledge that there are 3 lipoyl domains, two on E2 and one on the E3binding protein giving a total of at least 126 lipoates per E2, 7 it seems likely that no more than 12 and possibly as few as 7 acetyl groups are incorporated per PDC into the kinase-binding L2 domain for near-maximal stimulation, and as few as 3 and no more than 6 are incorporated for half-maximal stimulation. 8 This constitutes a highly responsive regulatory mechanism in which the kinase must preferentially interact with acetylated L2 and suggests that full stimulation may result from only one subunit of a kinase dimer interacting with a reductively acetylated L2 domain. Consistent with this prospect, half-maximal stimulation occurred with free L2 when only 20% of the L2 are acetylated. Although a much higher portion was found than estimated (above) for the intact complex, very different conditions are operative at the outer surface of the complex where lipoyl domains are concentrated at Ն1 mM, making the localized L2 level at least 50-fold higher than the highest level of free L2 used in the present work. Interaction of the kinase with acetylated-L2 would be aided by steady state turnover in the overall PDC reaction rapidly changing the location of acetyl groups within the multienzyme cluster. Kinase shuffling between L2 domains may also contribute, but this is needed, regardless, since acetylation by acetyl-CoA requires an unhindered lipoate (50,51), and the kinase must interact after L2 acetylation. We conclude that increased reductive-acetylation is very effective in activating kinase due to this preferred interaction of acetylated L2 altering the structure of the kinase and thereby increasing kinase activity (Fig. 9).
In being stimulated, kinase kinetic properties must fundamentally change since stimulation occurs over a wide range of rates (e.g. even with a peptide substrate that is slowly phosphorylated) and in the absence of kinase movement on the surface of the E2 core. Previous studies demonstrated that stimulation of the kinase requires physiological levels of potassium salts (7,14,53). Increasing K ϩ from low (0 -10 mM) to higher levels (90 -120 mM) leads to a marked inhibition of kinase activity and a gain in the capacity for NADH and acetyl-CoA stimulation. Another property of the kinase that changes with increasing K ϩ ion is that ADP inhibition markedly increases (55). A potential mechanism is that interaction of the kinase with an L2 domain containing an acetyl-dihydrolipoate leads to a conformational change in the kinase that speeds up ADP dissociation, which is probably the rate-limiting step in the kinase reaction at elevated (physiological) K ϩ levels. Consistent with that prospect are the observation that acetylation causes a greater stimulation of ADP-inhibited kinase (13) and the evidence that kinase-ADP is a prominent reaction intermediate, since pyruvate effectively inhibits by binding to the kinase-ADP in the absence of free ADP (56). For this mechanism to be correct, it must operate in conjunction with the slow rate of phosphorylation of peptide substrate, a particularly useful system for testing this mechanism. Further studies will be needed to define the structural changes that attenuate the kinase through changes in L2's lipoyl group and to determine how the K1 and K2 subunits of the kinase vary in their regulatory control.
In conclusion, our results strongly support E3-catalyzed reduction and E2-catalyzed acetylation of lipoyl moiety of the kinase binding inner (L2) domain of E2 mediating a marked enhancement in kinase activity. The effectiveness in intact PDC of acetylating a low proportion of L2 domains in enhancing the phosphorylation of many bound E1 by a single tightly 6 J. Baker and T. E. Roche, unpublished results. 7 Maeng et al. (54) have presented evidence that there are 12 protein X (E3BP) in yeast PDC using an area densiometric analysis approach. Using a less accurate densiometric approach, Jilka et al. (42) determined that there are only 6 E3BP in mammalian PDC but that should be reinvestigated. 8 Even at fairly low levels of acetylation, both E2 and E3BP are acetylated (42). Furthermore, analysis by autoradiography of the acetylation of elastase-generated polypeptides of L1 and L2, that were identified on blots by L1-and L2-specific monoclonal antibodies (27), demonstrated that both lipoyl domains of E2 are acetylated well with low levels of acetylation (S. Rahmatullah and T. E. Roche, unpublished results). There appeared to be a somewhat higher acetylation of L2 but the complex distribution of L1-and L2-derived polypeptides made that judgment uncertain. bound kinase molecule is probably accomplished by inter-L2 domain movement of the kinase combined with a stronger interaction of the kinase with acetylated L2 domains. Several studies on kinase function and regulation are consistent with a change in a K ϩ -requiring process ultimately occurring in kinase stimulation. A candidate mechanism involves counteracting K ϩ -strengthened ADP binding to the kinase to speed up slow dissociation of this product from the active site of the kinase.