The C-terminal Hinge Region of Lipoic Acid-bearing Domain of E2b Is Essential for Domain Interaction with Branched-chain α-Keto Acid Dehydrogenase Kinase*

The branched-chain α-keto acid dehydrogenase (BCKD) kinase (abbreviated as BCK) down-regulates activity of the mammalian mitochondrial BCKD complex by reversible phosphorylation of the decarboxylase (E1b) component of the complex. The binding of BCK to the holotransacylase (E2b) core of the BCKD complex results in the stimulation of BCK activity. Here we show that the lipoylated lipoic acid-bearing domain (lip-LBD) (residues 1–84) of E2b alone does not interact with BCK. However, lip-LBD constructs containing various lengths of the C-terminal hinge region of LBD are able to bind to BCK as measured by a newly developed solubility-based binding assay. Isothermal titration calorimetry measurements produced a dissociation constant of 8.06 × 10−6 m and binding enthalpy of –3.68 kcal/mol for the interaction of BCK with a construct containing lip-LBD and the Glu-Glu-Asp-Xaa-Xaa-Glu sequence of the C-terminal hinge region of LBD. These thermodynamic parameters are similar to those obtained for binding of BCK to a lipoylated di-domain construct, which harbors LBD, the entire hinge region, and the downstream subunit-binding domain of E2b. Our data establish that the C-terminal hinge region of LBD containing the above negatively charged residues is essential for the interaction between the lip-LBD construct and BCK.

The mammalian mitochondrial branched-chain ␣-keto acid dehydrogenase (BCKD) 1 complex catalyzes the rate-limiting step in the oxidation of branched-chain amino acids leucine, isoleucine, and valine (1). Genetic defects in this macromolecular multienzyme complex result in the accumulation of branched-chain amino acids leading to inheritable maple syrup urine disease. The clinical phenotype includes early onset and often fatal acidosis, neurological derangement, and mental retardation among survivors (1). The mammalian BCKD complex is organized around a cubic core of 24-meric dihydrolipoyl transacylase (E2b), to which multiple copies of branched-chain ␣-keto acid decarboxylase (E1b), dihydrolipoamide dehydrogenase (E3), BCKD kinase (abbreviated as BCK), and BCKD phosphatase are attached through ionic interactions (2)(3)(4). The activity of the BCKD complex is tightly regulated by a phosphorylation/dephosphorylation cycle in response to dietary and hormonal signals (5). Phosphorylation of Ser-292 and Ser-302 in the ␣ subunit of E1b by BCK results in the inactivation of the BCKD complex (6).
The structures of rat BCK in the apo-, ADP-bound, and ATP␥S-bound forms were recently determined (7). The BCK structures feature a nucleotide-binding (K) domain and a fourhelix bundle (B) domain, which are similar to modules found in bacterial protein-histidine kinases (8) and the related pyruvate dehydrogenase kinase 2 (PDK2) of the mitochondrial pyruvate dehydrogenase complex (PDC) (9). The K domain is highly conserved between BCK and members of the GHL ATPase family (10), with the presence of characteristic N1, G1, F, and G2 boxes and an "ATP lid." In BCK, the direct back-to-back interaction of two opposing K domains produces a dimeric structure in the crystal lattice. In contrast to the K domain, the functional significance of the extended B domain is unknown. Both ATPase (7) and phosphotransfer/kinase (11) activities of BCK are stimulated by lipoylated E2b. We have speculated that the E2b core binds to the B domain of BCK to promote the interaction between the B and K domains, resulting in the activation of BCK (7).
To approach this hypothesis, we have focused on the putative interaction of BCK with the N-terminal lipoic acid-bearing domain (LBD) of E2b, by analogy to the binding of the L2 domain of E2p to PDK (12). However, these studies have been hampered by the fact the lipoylated LBD (lip-LBD) alone (residues 1-84) cannot bind to BCK. Here we report that the previously overlooked C-terminal hinge region of LBD is essential for the interaction of lip-LBD with BCK. Constructs containing lip-LBD and various lengths of the C-terminal hinge region are able to bind to BCK to different degrees, as measured by a newly developed solubility-based assay and by isothermal titration calorimetry (ITC). The absence of relevant C-terminal hinge sequences may explain the previously observed weak binding of some PDK isoforms to the L2 domain of the PDC (13). More significantly, the inclusion of the C-terminal hinge region will facilitate investigations into the mechanism by which mitochondrial protein kinases, i.e. BCK and PDK isoforms, are regulated through interactions with their respective E2 cores of ␣-keto acid dehydrogenase complexes.
Expression and Lipoylation of DD and LBD Constructs-C-terminally His 6 -tagged E2 domain constructs were transformed into BL-21 cells, and the expression at 30°C was induced by 0.75 mM isopropyl-␤-D-thiogalactopyranoside. Nickel-nitrilotriacetic acid-purified DD and LBD proteins were lipoylated in vitro with the bacterial enzyme LplA as described previously (15).
Solubility-based Binding Assays for His 6 -tagged BCK-N-terminally His 6 -tagged BCK (40 M, monomers) in 400 mM arginine was combined with lipoylated or unlipoylated DD or LBD constructs (70 M). Each of the mixtures was dialyzed at 4°C for 16 h against 50 mM Hepes buffer, pH 7.5, containing 50 mM KCl, 5 mM dithiothreitol, 1 mM MgCl 2 , and 5% glycerol. The dialyzed mixtures were clarified by centrifugation, and aliquots were analyzed by SDS-PAGE and Coomassie Blue staining.
Binding Studies of Maltose-binding Protein (MBP)-BCK with Lipoylated E2 Domain Constructs by Isothermal Titration Calorimetry-MBP-BCK and lip-DD or lip-LBD constructs were dialyzed exhaustively with three changes against the same reservoir of 50 mM potassium phosphate buffer, pH 7.5, 50 mM KCl, 20 mM ␤-mercaptoethanol, and 5% glycerol. To carry out ITC measurements, the reaction cell of a MicroCal (Northampton, MA) VP-ITC microcalorimeter containing 1.5 ml of 100 M MBP-BCK was equilibrated at 20°C. The stock of 500 M lip-DD or lip-LBD constructs in 10-l increments was injected into the cell, and heat changes due to binding were recorded. Binding isotherms derived from the raw data were used to calculate the standard free energy of binding (⌬G 0 ) according to the equation: ⌬G 0 ϭ ϪRTlnK a , where R is the gas constant, T is absolute temperature, and K a is the association constant. From binding isotherms, the number of binding sites (n) was obtained, and changes in enthalpy (⌬H 0 ) and entropy (⌬S 0 ) were calculated according to the equation: ⌬G 0 ϭ ⌬H 0 Ϫ T⌬S 0 . Curve-fitting and the derivation of thermodynamic parameters were facilitated using the MicroCal ORIGIN software package. The concentrations of MBP-BCK and E2 domain constructs were determined by A 280 nm using calculated extinction coefficients (in mg ml Ϫ1 cm Ϫ1 ) of 1.09 (for MBP-BCK), 1.07 (LBD2), 1.14 (LBD3), 1.16 (LBD2), and 0.74 (DD).
Other Methods-The construction and expression of His 6 -tagged BCK and MBP-BCK by co-transformation with the pGroESL plasmid overproducing chaperonins GroEL and GroES were described previously (7,11).

Interactions of His 6 -tagged BCK with Lipoylated
Di-domain -A DD (residues 1-167) construct comprising the LBD, a hinge region, and the SBD was expressed and lipoylated in vitro and designated as lip-DD. The N-terminally His 6 -tagged BCK is soluble to the extent of 2-3 mg/ml only in the presence of 300 mM arginine (7). However, the requirement of high arginine concentration for solubility was circumvented when N-terminally His 6 -tagged BCK was allowed to bind to lip-DD. Based on this finding, a solubility-based binding assay was developed. Lip-DD was combined with the N-terminally His 6tagged BCK in the presence of 400 mM arginine. The mixture was dialyzed to remove arginine because it interferes with the binding of lip-DD to BCK (data not shown). The solubility of His 6 -tagged BCK was indicated by its presence in the supernatant following dialysis and centrifugation. As shown in Fig.  1, in the presence of lip-DD, His 6 -tagged BCK remains in the supernatant after the removal of arginine (lane 2). The monomeric subunit stoichiometry of His 6 -tagged BCK to lip-DD is 1:2. The presence of excess lip-DD is necessary to prevent the precipitation of His 6 -tagged BCK during dialysis. In contrast, incubation of His 6 -tagged BCK with unlipoylated apo-DD does not render the former soluble in the absence of arginine (lane 3). These data confirm the finding that the covalently attached lipoic acid is essential for the binding of holo-E2p to PDK (16). In contrast, incubation with LBD (residues 1-84) alone, whether lipoylated (lane 4) or unlipoylated (lane 5), did not prevent the precipitation of His 6 -tagged BCK upon the removal of arginine. The results indicate that the sequence of LBD alone is not capable of binding to BCK.
The C-terminal Hinge Region of LBD Is Indispensable for the Interaction of LBD with His 6 -tagged BCK-The ability of lip-DD, but not lip-LBD, to maintain the solubility of His 6 -tagged BCK in the absence of arginine raises the question of whether the C-terminal SBD or the hinge sequence between LBD and the SBD is needed for binding. Therefore, solubility-based binding studies with His 6 -tagged BCK were carried out further with lip-LBD constructs containing various lengths of the Cterminal hinge region ( Fig. 2A). As shown in Fig. 2B Direct Binding Measurements by Isothermal Titration Calorimetry-We utilized ITC as a direct method to study the interaction of lip-LBD2 with BCK. Measurements of heat changes by ITC require relatively large quantities of proteins especially when the binding energy is weak. For binding studies by ITC, rat BCK was expressed in Escherichia coli as a MBP fusion. MBP-BCK, unlike His 6 -tagged BCK, is highly soluble (ϳ50 mg/ml) in the absence of arginine, and the presence of MBP has no effect on BCK activity and its interaction with lipoylated E2b (11). Fig. 3 (upper panel) shows the heat of titration with apo-LBD2 and lip-LBD2. The unchanged ITC isotherm with apo-LBD2 unequivocally demonstrates the absence of interactions with MBP-BCK (Fig. 3, lower panel). By comparison, lip-LBD2 shows a gradual rise in the isotherm, which is consistent with a weak binding to MBP-BCK. Because of the weak interaction, the titration is only ϳ80% saturated under the conditions used in these experiments. From the binding isotherm, a dissociation constant K D of 8.06 ϫ 10 Ϫ6 M, a binding energy ⌬G 0 of Ϫ6.84 kcal/mol, and a binding enthalpy ⌬H 0 of Ϫ3.68 kcal/mol were calculated (Fig. 3, lower  panel). The data represent direct evidence for the interaction between lip-LBD2 containing the C-terminal hinge region and MBP-BCK.
Thermodynamic Parameters for Interactions of LBD Constructs with MBP-BCK-The ITC method was further employed to characterize the interactions of MBP-BCK with LBD constructs containing various lengths of the C-terminal linker region. Table I shows that the interactions of lip-LBD2 with MBP-BCK is the most exothermic among the four LBD constructs studied. The binding enthalpy ⌬H 0 of Ϫ3.68 kcal/mol is comparable with that of Ϫ3.70 kcal/mol measured with lip-DD. Significantly lower enthalpies were obtained with lip-LBD3 and lip-LBD4. Smaller differences in dissociation constants K D were observed between lip-DD and lip-LBD2 through lip-LBD4. These thermodynamic parameters confirm the conclusions derived from the solubility-based binding assays that lip-LBD2 and lip-LBD3 interact with MBP-BCK to a degree similar to lip-DD. DISCUSSION A prominent feature of the mitochondrial kinases comprising BCK and PDK isoforms is the up-regulation through interaction with their respective E2 cores of ␣-keto acid dehydrogenase complexes (11,17). The lipoyl-containing domains of the E2 cores function both as an anchor and a modulator for activities of the mitochondrial kinases. Studies with the lipoylated inner L2 domain have been complicated by the fact that the domain alone does not interact as efficiently as holo-E2p with the PDK isoforms, except for PDK3 (13,18). In this communication, we show that lip-LBD alone does not bind to BCK; however, the inclusion of various lengths of the previously neglected C-terminal hinge region results in the interaction between the lip-LBD constructs and BCK. This information has facilitated the co-crystallization of BCK with lip-LBD constructs for the ongoing structural determination by x-ray diffraction (data not shown). The L2 domain of E2p used in binding studies contained either none (13) or a portion of the hinge region (12,19). The omission of certain important elements in the C-terminal hinge region of the L2 domain may explain the weaker interaction of these lipoylated constructs with PDK isoforms than with holo-E2.
The essential determinants in the C-terminal hinge region of E2b LBD for the binding of lip-LBD to BCK are presently uncertain. However, the inclusion of negatively charged residues Glu-Glu-Asp-Xaa-Xaa-Glu in lip-LBD2 and lip-LBD3 constructs results in binding efficiency similar to lip-DD, based on dissociation constants and binding enthalpies measured by ITC. The lip-DD construct contains the entire hinge region connecting LBD and SBD and therefore mimics holo-E2 regarding binding to BCK. The results with the lip-LBD constructs strongly suggest that the negative charged residues in the C-terminal hinge region are candidate determinants for interactions with BCK. In the BCK structure (7), a groove formed between helices BH4 and BH3 of the B domain is lined with positively charged residues that are suited to interact with the negatively charged residues present in the C-terminal hinge region of LBD, as suggested for the PDK2 structure (9). On the other hand, the covalently attached lipoic acid is also essential for the binding of lip-LBD constructs to BCK. Due to the aliphatic nature of the lipoic acid moiety and the presence of hydrophobic residues in the hinge region, there may be additional weak hydrophobic interactions between lip-LBD constructs and BCK. The C-terminal hinge region of LBD is located 26 Å away from the lipoylated Lys-44 in the recently determined human LBD solution structure (20). Both structural elements are on the same side of the ␤ barrel and repre- BCK does not form a stable complex with lip-DD or lip-LBD constructs to allow isolation by the conventional column chromatography or ultracentrifugation methods. We have therefore developed the solubility-based binding assay, which facilitates screening for the binding of domain constructs to BCK in a simple and straightforward fashion. The N-terminally His 6tagged BCK is soluble only in the presence of high concentrations of arginine. The amphipathic character of arginine presumably protects exposed hydrophobic interacting surfaces of the folded protein (21). The crux of the binding assay is based on the discovery that a molar excess of lip-DD can increase the solubility of N-terminally His 6 -tagged BCK from 2-3 mg/ml in the presence of 300 mM arginine to 20 mg/ml when arginine is replaced by lip-DD. However, arginine and lipoylated E2 domains are mutually exclusive in binding to His 6 -tagged BCK as indicated by the phosphotransfer/kinase activity assays (data not shown). The slow gradual removal of arginine by dialysis from His 6 -tagged BCK in the presence of lip-DD enables the latter to bind to BCK thereby maintaining its solubility. The molar excess of lip-DD or lip-LBD constructs in the binding assay ensures the optimal binding to His 6 -tagged BCK upon the removal of arginine.
The ITC method offers "real time" measurements of thermodynamic parameters for protein-ligand and protein-protein binding reactions under defined solution conditions. ITC is ideal for studies of weak interactions of BCK with lip-DD and lip-LBD constructs by taking advantage of the high solubility of MBP-BCK. The dissociation constants (K D ) in the 10 Ϫ6 M range derived from ITC measurements for lip-DD and lip-LBD2, lip-LBD3, and lip-LBD4 are consistent with the weak interactions of BCK with these E2b domain constructs. These values are four orders of magnitude higher than those estimated for the tight binding of E1p or E3 to the DD of the PDC from Bacillus stearothermophilius (22). On the other hand, the binding enthalpy obtained by ITC measurements appears to be a more accurate parameter than the dissociation constant in differentiating binding affinity among lip-LBD constructs ( Table I). The similar strongly exothermic binding enthalpies between lip-LBD2 and lip-DD indicate that the lip-LBD2 construct contains all the necessary determinants present in E2b for its interaction with BCK. Finally, for the fitting of the binding isotherm, a model based on one lip-LBD2 binding site per MBP-BCK monomer was used (Fig. 3). The best fit values generated a stoichiometry of n ϭ 0.5, indicating that due to the weak interactions only one of the two sites in the MBP-BCK homodimer is occupied by lip-LBD2 under non-saturating conditions. This binding stoichiometry is consistent with that deduced from packing analysis of the BCK/lip-LBD2 crystals (data not shown). a For these constructs, see the legend to Fig. 2 for sequences in the C-terminal hinge region.
Interaction of Lipoic Acid-bearing Domain with BCKD Kinase 36908