Originally published In Press as doi:10.1074/jbc.M403611200 on May 27, 2004
J. Biol. Chem., Vol. 279, Issue 31, 32968-32978, July 30, 2004
Cross-talk between Thiamin Diphosphate Binding and Phosphorylation Loop Conformation in Human Branched-chain 
-Keto Acid Decarboxylase/Dehydrogenase*
Jun Li
,
R. Max Wynn
,
Mischa Machius,
Jacinta L. Chuang,
Subramanian Karthikeyan,
Diana R. Tomchick, and
David T. Chuang¶
From the
Departments of Biochemistry and Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas 75390-9038
Received for publication, April 1, 2004
, and in revised form, May 24, 2004.
 |
ABSTRACT
|
|---|
The decarboxylase/dehydrogenase (E1b) component of the 4-megadalton human branched-chain 
-keto acid dehydrogenase (BCKD) metabolic machine is a thiamin diphosphate (ThDP)-dependent enzyme with a heterotetrameric cofactor-binding fold. The E1b component catalyzes the decarboxylation of -keto acids and the subsequent reductive acylation of the lipoic acid-bearing domain (LBD) from the 24-meric transacylase (E2b) core. In the present study, we show that the binding of cofactor ThDP to the E1b active site induces a disorder-to-order transition of the conserved phosphorylation loop carrying the two phosphorylation sites Ser292- and Ser302-, as deduced from the 1.80-1.85 Å apoE1b and holoE1b structures. The induced loop conformation is essential for the recognition of lipoylated LBD to initiate E1b-catalyzed reductive acylation. Alterations of invariant Arg287-, Asp295-, Tyr300-, and Arg301- that form a hydrogen-bonding network in the phosphorylation loop result in the disordering of the loop conformation as elucidated by limited proteolysis, accompanied by the impaired binding and diminished reductive acylation of lipoylated LBD. In contrast, kcat values for E1b-catalyzed decarboxylation of the -keto acid are higher in these E1b mutants than in wild-type E1b, with higher Km values for the substrate in the mutants. ThDP binding that orders the loop prevents phosphorylation of E1b by the BCKD kinase and averts the inactivation of wild-type E1b, but not the above mutants, by this covalent modification. Our results establish that the cross-talk between the bound ThDP and the phosphorylation loop conformation serves as a feed-forward switch for multiple reaction steps in the BCKD metabolic machine.
|
INTRODUCTION
|
|---|
The human branched-chain -keto acid dehydrogenase (BCKD)1 metabolic machine is a member of the highly conserved -keto acid dehydrogenase complexes in mitochondria. The BCKD complex catalyzes the oxidative decarboxylation of branched-chain -keto acids (BCKAs) derived from leucine, isoleucine, and valine (1) (Reaction 1),
Genetic defects in the BCKD complex result in the heritable Maple Syrup Urine Disease manifested by often-fetal ketoacidosis, encephalopathy, and mental retardation (2). The 4 x 106-dalton BCKD complex is organized around a cubic core comprising 24 lipoate-bearing dihydrolipoyl transacylase (E2b) subunits, associated with multiple copies of branched-chain -keto acid decarboxylase/dehydrogenase (E1b), dihydrolipoamide dehydrogenase (E3), the BCKD kinase, and the BCKD phosphatase. The kinase and the phosphatase regulate activity of the BCKD complex by a reversible phosphorylation (inactivation)/dephosphorylation (activation) cycle (3). Reaction steps catalyzed by the three enzyme components, analogous to those elucidated for the cognate pyruvate dehydrogenase complex (4), are as shown in Reactions 2-6.
 |
 |
 |
 |
 |
These reaction steps epitomize a classic case of substrate channeling in multienzyme complexes. The E1b component binds thiamin diphosphate (ThDP) and catalyzes the ThDP-mediated decarboxylation of -keto acids (Reaction 2), resulting in an enamine-ThDP intermediate (E1b-R-COH=ThDP), and the reductive transfer of the enamine to the lipoyl moiety that is covalently attached to E2b (abbreviated as E2b-[lipS2]) (Reaction 3). The lipoyl-bearing domain (LBD) carrying the S-acyldihydrolipoamide serves as a "swinging arm" (5) to transfer the acyl group from E1b to the E2b active site, where it is converted to acyl-CoA (Reaction 4). Finally, the E3 component with a tightly bound FAD moiety re-oxidizes the dihydrolipoyl residue on E2b (E2b-[lip(SH)2]) (Reaction 5) with NAD+ as the ultimate electron acceptor (Reaction 6). The overall reaction is the production of branched-chain acyl-CoA, CO2, and NADH from BCKAs (Reaction 1).
The E1b component of the human BCKD metabolic machine belongs to the superfamily of ThDP-dependent enzymes (6) with 20 known members to date. There are currently nine crystal structures for ThDP-dependent enzymes in the Protein Data Bank, including E1b components of the human (7) and Pseudomonas (8) BCKD complexes. The human and Pseudomonas E1b components are 2
2 heterotetramers with a distinct ThDP-binding fold. Two ThDP-binding sites are located in -' and '- subunit interfaces. The subunits bind to the diphosphate modules through residues in a conserved sequence motif (9) and involve a bridging Mg2+ ion (7), and the subunits bind to the pyrimidine modules of the cofactors. Moreover, a unique K+ ion-binding pocket in the human E1b subunit is crucial for the conformation of a cofactor-binding loop (7). The E1b ThDP-binding fold has also been observed recently in the pyruvate dehydrogenase (E1p) component of the human pyruvate dehydrogenase complex (10).
The phosphorylation of Ser292- and Ser302- (in the subunit) of human E1b by the BCKD kinase abolishes the overall activity (Reaction 1) of the human BCKD complex (11). In a crystal structure of human E1b refined to 1.85 Å, these two phosphorylation sites are located in a so-called phosphorylation loop region in the human E1b active site (12). As shown in Fig. 1A, the sequence for the phosphorylation loop region is highly conserved among heterotetrameric (22) E1 proteins from BCKD and pyruvate dehydrogenase complexes (12-14). The invariant His291- in the phosphorylation loop region coordinates to ThDP, and the mutation of this residue disrupts the binding of the cofactor (12, 14). The significance of sequence conservation for the phosphorylation loop region in E1b from Pseudomonas putida and E1p from the Gram-positive prokaryote Bacillus stearothermophilus is not entirely known, because these bacterial heterotetrameric E1 proteins are not regulated by reversible phosphorylation.
View larger version (46K):
[in this window]
[in a new window]
|
FIG. 1.
The conserved amino acid sequence and crystal structure of the human E1b phosphorylation loop region. A, sequence conservation in the phosphorylation loop regions of heterotetrameric E1 proteins. The sequence of the human E1b phosphorylation loop region (residues 286-315) was used as a query sequence in a gapped BLAST search. The numbering of the aligned amino acid residues is based on the subunit of human E1b. P1 and P2 refer to the two phosphorylation sites (Ser292- and Ser302-, respectively) in the subunit of human E1b. Invariant residues are indicated in blue. The secondary structural elements of the phosphorylation loop region of human E1b are indicated below the sequences. B, the active site in the human wild-type E1b, which binds cofactor ThDP, is formed between the (purple) and ' (cyan) subunits. The phosphorylation loop region that carries the two phosphorylation sites in the subunit is highlighted in red. Secondary structural elements in the phosphorylation loop consist of a loop (Arg286- to Ser302-), a short helix (Val303- to Lys311-), and a turn (Gln312- to His314-). In the wild-type E1b structure (Protein Data Bank code 1ols
[PDB]
), electron density is missing for residues Thr293-, Ser294-, and Ala299- and Ser302- to Asn307-. The missing residues were modeled according to the crystal structure of a H146A-' mutant where the entire loop has corresponding electron density. C, the hydrogen-bonding network in the phosphorylation loop region formed between four invariant residues (Arg287-, Asp295-, Tyr300-, and Arg301-). The number next to a hydrogen bond (dotted line) indicates the distance in Å between the two atoms forming the bond. Figure was generated using the programs Bobscript (39) and PovRay (Persistence of Vision, v3.02, POV Team, www.povray.org).
|
|
In the present study, we dissect the role of the phosphorylation loop region in E1b catalysis in the context of the BCKD metabolic machine. We focus on residues in this loop region that are invariant among heterotetrameric E1 proteins of -keto acid dehydrogenase complexes (Fig. 1A). Our results show that ThDP binding to apoE1b orders the phosphorylation loop region through the formation of a hydrogen-bonding network. The loop conformation is required for the recognition of E2b LBD domain to initiate E1b-catalyzed reductive acylation (Reaction 3). Moreover, this loop conformation prevents the inactivation of E1b through phosphorylation by the BCKD kinase. Thus, the cross-talk between cofactor binding and the phosphorylation loop region in E1b serves as a feed-forward switch for catalysis and subsequent substrate channeling in the human BCKD metabolic machine. This concerted regulatory mechanism through cofactor binding may have applications to other cofactor-dependent enzymes that are integral components of multienzyme complexes.
|
EXPERIMENTAL PROCEDURES
|
|---|
Constructs and Proteins—N-terminally and C-terminally His6-tagged wild-type and mutant E1b proteins were produced as described previously (15) with mutations introduced using the QuickChange site-directed mutagenesis kit provided by Stratagene. Wild-type and mutant apoE1b proteins were obtained by exhaustive dialysis in the presence of 0.2 mM EDTA. C-terminally His6-tagged LBD (residues 1-84 of the E2b subunit) was expressed (16) and lipoylated in vitro with DL-6,8-thioctic acid using LplA, a lipoic-protein ligase, as described earlier (17).
Enzyme Assays—The BCKD complex was reconstituted with E1b, lipoylated E2b (lip-E2), and E3 at a molar ratio of 12:1:55 as described previously (17). The oxidative decarboxylation of KIV catalyzed by the reconstituted complex (Reaction 1) was monitored by the increase of absorbance at 340 nm.
ThDP-mediated decarboxylation by the isolated E1b component (Reaction 2) was assayed at 30 °C with KIV as substrate in the presence of an artificial electron acceptor 2,6-dichlorophenolindophenol (DCPIP) as described previously (18). The oxidation of enamine-ThDP (RCOH=ThDP) by DCPIP converts enamine-ThDP to a free acid (RCOOH), which promotes the turnover of enamine-ThDP with concomitant release of the product acid from E1b. Reduction of DCPIP was monitored by the decrease in absorbance at 600 nm. The direct measurement for the rate of decarboxylation by E1b was carried out by incubating wild-type or mutant E1b (15 µg each) with 1 mM [1-14C]KIV, 1 mM Mg-ThDP, and 100 µM DCPIP for 10 min in 100 mM potassium phosphate buffer, pH 7.5, at 30 °C. The released CO2 was collected, and its radioactivity was counted as described previously (17).
The assay for reductive acylation of lip-LBD catalyzed by E1b (Reaction 3) was carried out as described previously with modifications (12). This assay is based on the irreversible decarboxylation step (Reaction 2) combined with the rate-limiting reductive acylation reaction, both catalyzed by E1b. The coupled reactions allow the use of [U-14C]KIV instead of enamine-ThDP as a substrate, with lip-LBD in lieu of E2b as the other. The assay was initiated by adding [U-14C]KIV (final concentration 2 mM) to the reaction mixture (final volume 250 µl) containing 16 µM lip-LBD and 0.16 µM E1b. After incubation at 22 °C for 45 s, the assay was terminated by adding trichloroacetic acid to a final 10% (v/v) concentration, followed by the addition of 0.2% (w/v) bovine serum albumin as a carrier. The precipitate was washed two times with 10% trichloroacetic acid to remove excess [U-14C]KIV and radiolabeled enamine-ThDP released from the denatured E1b protein. The washed pellet containing radiolabeled lip-LBD was resuspended in 50 µl of 8 M urea. To the suspension, 2 ml of scintillation mixture was added, and the radioactivity was determined. The nonspecific radioactivity trapped by nonlipoylated LBD and the E1b protein served as a blank.
Binding Measurements Based on Tryptophan Fluorescence Quenching—Steady-state fluorescence quenching upon ThDP binding (19) to wild-type and mutant E1b was measured using a PerkinElmer Life Sciences LS50 B luminescence spectrometer in the photon counting mode. Fluorescence intensities were recorded at 25 °C using a 3-ml quartz cuvette at an excitation wavelength of 290 nm and an emission wavelength of 335 nm. Slit widths were set at 5 nm for excitation and 10 nm for emission. A 290 nm cut-off emission filter was installed to reduce light scattering effects. Protein concentrations for E1b and ThDP (A235 nm = 11,300 M-1 cm-1, pH > 7.0) were determined spectrophotometrically. The concentration for all protein samples was 0.2 µM (as heterotetramers) in 50 mM HEPES buffer, pH 7.5, 200 mM KCl, and 1 mM MgCl2. Fluorescence readings were corrected for dilution and inner filter effects using Equation 1 (20),
 | (Eq. 1) |
where, Fcorr is the corrected fluorescence intensity value; Fobs is the experimentally measured fluorescence intensity; V0 is the initial volume of the sample; V is the volume after adding ThDP; d is the path length of the cuvette; Aex is the absorption of the sample at the excitation wavelength; and Aem is the absorption of the sample at the emission wavelength. Three different readings were taken and averaged, and the experiment was repeated two times (n = 3). The binding data were fitted by nonlinear regression using the program KaleidaGraph (Synergy Software, Essex Junction, VT) according to Equation 2 describing a bimolecular reaction (21),
 | (Eq. 2) |
where 
F is the (corrected) fluorescence change; F0 is the fluorescence intensity prior to the addition of ThDP; Fmax is the maximal fluorescence change; Kd is the dissociation constant; and [ThDP] is the concentration of ThDP in the cuvette. The parameters determined by the fitting procedure were Fmax and Kd.
Binding of LBD to E1b Studied by Isothermal Titration Calorimetry—Human E1b and lip-LBD, both C-terminally His6-tagged, were dialyzed exhaustively against the same reservoir of 50 mM Tris buffer, pH 7.5, 50 mM KCl, 10 mM -mercaptoethanol, 5% (v/v) glycerol, and 0.2 mM EDTA to remove bound Mg2+ ions and ThDP. Immediately prior to ITC measurements, MgCl2 and ThDP stock solutions were added to both E1b and lip-LBD to a final concentration of 0.5 mM. A stock of 1.5 mM lip-LBD in the syringe was added in 8-µl increments to a reaction cell containing 1.8 ml of 75 µM human E1b (based on the heterodimer), and the resulting heat changes were measured at 22 °C in a Microcal (Northampton, MA) VP-ITC microcalorimeter. Binding isotherms were derived from heat changes resulting from the binding reactions as described previously (12). Curve fitting and the derivation of thermodynamic parameters were carried out with the ORIGIN software package provided by Microcal.
X-ray Crystallography of Wild-type and Mutant E1b Proteins—Crystals of wild-type and mutant human E1b proteins (C-terminally His6-tagged on the subunit) were grown at 22 °C via the vapor diffusion method and frozen in liquid nitrogen as described previously (12). For holoE1b crystals, Mn2+ ions could replace Mg2+ required for the binding of ThDP to E1b. The presence of Mn2+ ions in crystals resulted in improved x-ray diffraction qualities without affecting the catalytic properties (data not shown). Crystals obtained with this procedure exhibited the symmetry of space group P3121 with cell parameters of 
145 x 145 x 69 Å and contained one heterodimer per asymmetric unit. The crystals diffracted x-rays to a minimum Bragg spacing (dmin) of 1.8 Å. For x-ray diffraction, crystals were flash-cooled in liquid propane and kept at about 100 K during data collection at beamlines 19ID and 19BM (Advanced Photon Source (APS), Argonne National Laboratory, Argonne, IL). Data sets were processed with the HKL2000 package (22). Complete data processing statistics are listed in Table I (see A).
The E1b wild-type structure was determined by difference Fourier techniques using a previously determined E1b structure (Protein Data Bank code 1ols
[PDB]
) as an initial model. Refinement of the models was carried out with the program Refmac (23) of the CCP4 package (24) with a random subset of all data set aside for the calculation of free R factors. Manual adjustments to the models were carried out with the program O (25). The electron density clearly showed the presence of glycerol molecules in the crystal structure. After the refinement of the protein part was complete, solvent molecules were added where stereochemically reasonable. Single residues with little corresponding electron densities were included in the model, when they were flanked by residues with well defined electron density. Complete refinement statistics are listed in Table I (see B).
Limited Proteolysis of Wild-type and Mutant E1b Proteins—The digestion of 20 µg of apoE1b was carried out in a 50-µl volume in the presence and absence of ThDP for different times on ice in 50 mM potassium phosphate buffer, pH 7.5, 100 mM KCl, 1 mM MgCl2, 1 mM -mercaptoethanol, and 0.1 µg of chymotrypsin. The reaction was stopped by addition of 1 mM PMSF, and the digestion mixture was subjected to SDS-PAGE analysis in 12% gels. Protein bands were stained with Coomassie Blue or blotted to a polyvinylidene difluoride membrane for N-terminal sequencing.
Phosphorylation of E1b by the BCKD Kinase—A mixture containing 1 µg of E1b protein, 0.75 µg of lip-E2b, and 0.1 µg of maltose-binding protein-tagged rat BCKD kinase (26) in 30 mM HEPES buffer, pH 7.35, 2 mM dithiothreitol, 1.5 mM MgCl2, 0.2 mM EGTA was preincubated at room temperature for 15 min. The phosphorylation reaction was initiated by adding 0.4 mM [
-32P]ATP (specific activity 0.5 µCi/nmol ATP) to the above reaction mixture. After 1 min, the reaction was terminated by adding a concentrated SDS-PAGE sample buffer (50 mM Tris-HCl, pH 6.8, 5% (w/v) SDS, 50% (v/v) glycerol, 500 mM dithiothreitol, 1% (w/v) bromphenol blue) with 1:5 dilution and immediately placed on ice. Samples were then heated at 65 °C for 10 min prior to loading onto a 12% SDS-PAGE gel. After electrophoresis, gels were stained with Coomassie Blue R-250 and vacuum-dried. Quantitative analysis of 32P incorporation to E1b was carried out by scanning a storage phosphor screen with a Typhoon 9200 Variable Mode Imager (Amersham Biosciences) using ImageQuant software.
|
RESULTS
|
|---|
Structure of the Phosphorylation Loop Region in the E1b Active Site—We defined previously (12) the phosphorylation loop region of human E1b to encompass residues Tyr286- to Gln312-. Electron density was not visible for residues Thr293-, Ser294-, and Ala299- and between Ser302- and Asn307- in wild-type human E1b structure. The phosphorylation loop is generally ordered to varying degrees in different crystals. There is no apparent correlation between the order of the phosphorylation loop and crystallization conditions or crystal handling procedures. For example, an H146A-' mutant features electron density for the entire loop, and this is presented for illustrative purposes in Fig. 1B in a superposition with the wild-type E1b structure. The conformation of the phosphorylation loop region consists of a loop that harbors the two phosphorylation sites (Ser292- and Ser302-), followed by a short helix (Val303- to Lys311-) and a turn (Gln312- to His314-). Fig. 1C shows the four invariant residues (Arg287-, Asp295-, Tyr300-, and Arg301-) in the human E1b active site, whose side chains are located within a distance of 3 Å from each other forming an intricate hydrogen-bonding network in the phosphorylation loop region.
Mutations in the Hydrogen-bonding Network Impede Catalysis of E1b and the BCKD Complex—To decipher the roles of these residues in catalysis, kinetic parameters of E1b mutants in the hydrogen-bonding network were determined. The decarboxylation activity of E1b according to Reaction 2 was measured with DCPIP as an electron acceptor. The conversion of R287A-, Y300A-, Y300F-, and R301A- mutations results in significantly higher Km values for the substrate KIV than the wild type (50.2 µM) (Table II). The highest increase (46-fold over the wild-type) was observed with the R287A- variant. In contrast, the D295A- mutant shows a Km for KIV similar to the wild type. In addition, Km values for ThDP for the mutants are 2-6-fold higher than for the wild type (6.6 µM). These kinetic parameters strongly suggest that the four residues forming the hydrogen-bonding network are critical for high affinity cofactor and/or substrate binding. Most interesting, kcat values for all five mutants are markedly elevated ranging from 2- to 3-fold over the wild type.
View this table:
[in this window]
[in a new window]
|
TABLE II
Kinetic parameters for the E1b-catalyzed decarboxylation reaction (Reaction 2) determined by the spectrophotometric DCPIP assay
|
|
The decarboxylation of KIV by wild-type and mutant E1b (Reaction 2) was also measured directly by a radiochemical assay using [1-14C]KIV as a substrate in the presence of DCPIP. At the saturating substrate concentration of 1 mM, wild-type and D295A- E1b show the decarboxylation rates of 9.9 and 28.8 nmol of CO2/min/nmol E1b, respectively (Fig. 2). These values are close to kcat values of 11.7 and 29.1 min-1, respectively, for wild-type and D295A- E1b measured by the spectrophotometric DCPIP assay (Table II). Rates of decarboxylation for [1-14C]KIV are also increased in the Y300F- and R301A- mutant E1b compared with the wild type. Because Km values for KIV of these mutants are elevated (Table II), the rates of decarboxylation measured at the nonsaturating 1 mM [1-14C]KIV concentration are lower than kcat values. The robust nonenzymatic decarboxylation of [1-14C]KIV at high substrate concentrations prevented the accurate measurement of kcat values for these two E1b variants. Nonetheless, the above results confirm that the rate of decarboxylation for the keto acid is significantly enhanced in the phosphorylation loop mutants over the wild type.
View larger version (11K):
[in this window]
[in a new window]
|
FIG. 2.
Direct measurements for the rate of decarboxylation by wild-type and mutant E1b. The wild-type (WT) or mutant E1b protein (15 µg each) was incubated for 10 min at 30 °C with 1 mM [1-14C]KIV, 1 mM Mg-ThDP, and 100 µM DCPIP. The released CO2 was collected and counted as described under "Experimental Procedures." Results are averages of triplicate measurements.
|
|
Effects of the above mutations on the overall activity, i.e. the oxidative decarboxylation of KIV catalyzed by the BCKD complex (Reaction 1), were also studied. The BCKD complex was reconstituted with wild-type or mutant E1b in the presence of lipoylated E2b and E3 at a molar ratio of 12:1:55. Y300A-, Y300F-, and R301A- mutants show 3-8-fold increased Km values for KIV. The D295A- variant shows the same Km value for KIV as the wild type (Table III), similar to that determined by the E1b decarboxylation assay (Reaction 2). It is striking that the BCKD complex reconstituted with the R287A- mutant shows no detectable overall activity, despite the robust decarboxylase activity (Reaction 2) of this E1b mutant. The kcat values for the other four mutants are significantly lower than for the wild type. The Km value of wild-type E1b for ThDP is 0.63 µM, which is an order of magnitude lower than that measured for E1b alone. The higher Km values for ThDP combined with reduced kcat for the reconstituted BCKD complex result in catalytic efficiencies for the four mutants D295A-, Y300A-, Y300F-, and R301A- that are 2 orders of magnitude lower than the wild type (Table III).
View this table:
[in this window]
[in a new window]
|
TABLE III
Kinetic parameters for the overall reaction (Reaction 1) catalyzed by the reconstituted wild-type and mutant BCKD complexes
The BCKD complex was reconstituted with wild-type and mutant E1b as indicated. The reduction rate of NAD+ was determined by monitoring the increase in absorbance at 340 nm as described under "Experimental Procedures."
|
|
Mutant E1b Proteins Bind Cofactor ThDP Poorly—Dissociation constants (Kd) for wild-type and mutant E1b proteins were obtained by measuring the quenching of fluorescence intensity as a function of the ThDP concentration (Table IV). For the wild-type E1b, a ThDP concentration range of 0-5 µM was used, and for the mutant E1b, a range of 0-300 µM was employed. The UV quenching was corrected for the inner filter effect due to free ThDP in the solution. The wild-type E1b shows a Kd value of 1.6 µM. R287A-, D295A-, Y300F-, and R301A- mutants display increased Kd values that are 2 orders of magnitude higher than the wild type. The Y300A- mutant exhibits a 36-fold higher Kd value (58.4 µM). The results confirm the reduced affinity of the E1b mutants for ThDP, as suggested by their increased Km values derived from E1b decarboxylation and the overall activity assays.
View this table:
[in this window]
[in a new window]
|
TABLE IV
Dissociation constants (Kd) for the binding of ThDP to wild-type and mutant E1b as determined by fluorescence quenching
|
|
E1b Mutations Impede Activity for Reductive Acylation—Effects of mutations in the phosphorylation loop region on reductive acylation of lip-LBD from E2b (Reaction 3) were deciphered. In the assay for reductive acylation, isolated recombinant lip-LBD instead of the natural lipoylated full-length E2b was used as substrate. The R287A- mutation essentially abolishes activity for reductive acylation catalyzed by E1b (Fig. 3). D295A- and Y300A- variants show 10% of wild-type activity (2.3 min-1), whereas R301A- and Y300F- mutants exhibit 30 and 40%, respectively, of wild-type activity. The data indicate that these residues are important for the reductive acylation of lip-LBD, with Arg287- playing the most critical role in this reaction.
Mutant E1b Proteins Have Low Affinity for lip-LBD—ITC was employed to measure the binding affinity of wild-type and mutant E1b for lip-LBD. Wild-type E1b binds lip-LBD with a Kd value of 20.3 µM and a binding enthalpy (H0) of -6.28 kcal/mol (Table V). The D295A- or Y300F- E1b also binds lip-LBD albeit with slightly lower affinity (Kd = 69.4 and 34.9 µM, respectively). There was no measurable heat evolution for the interaction of lip-LBD with the R287A- E1b, indicating that this mutant does not bind lip-LBD. The reduced binding affinities of mutant E1b for lip-LBD correlate with the decrease or absence (in the case of R287A-) of reductive acylation activity (Fig. 3).
View this table:
[in this window]
[in a new window]
|
TABLE V
Thermodynamic parameters for the interactions of lip-LBD with wild-type and mutant E1b as determined by isothermal titration calorimetry
|
|
ThDP Binding Orders Two Regions in the Active Site of ApoE1b—The conformational changes in apoE1b induced by ThDP binding were deciphered by solving the crystal structure of human apoE1b. The apoE1b structure at 1.8-Å resolution was superimposed with the holoE1b structure at 1.85-Å resolution. Fig. 4 shows that ThDP binding orders two major regions in the active site that are not ordered in the apoE1b structure. The first region corresponds to Arg287- to His314-, which harbors the phosphorylation loop region. The second region encompassing Gly223- to Thr288- is well ordered through the coordination of the invariant Asn222- to the Mn2+ ion used instead of the physiological Mg2+ in our crystals (see "Experimental Procedures"). It should be pointed out that the side chain of Tyr286- undergoes a rotation by about 90° between the apo- and holoE1b structures. Other regions that have clear electron density in holoE1b, as opposed to apoE1b, contain residues 6-13 and residues 26-29 of the subunit. The significance of these additional ordered regions in holoE1b is not clear at present. The conformation of the subunit is identical between apo- and holoE1b.
Mutations in the Hydrogen Bonding Network Result in Disordering of the Phosphorylation Loop Region—Fig. 5A shows the 2Fo - Fc electron density map of the active site in wild-type human holoE1b. Good electron density is visible in the phosphorylation loop region between Tyr286- and Arg301- except residues Thr293-, Ser294-, and Ala299-. The R287A- mutation results in the disordering of the conserved phosphorylation loop region (Ile288- to His314-), as indicated by the complete absence of electron density (Fig. 5B). The side chain of Tyr286- assumes a conformation similar to that in apoE1b, with the substituted residue Ala287- well ordered. Similar order-to-disorder transitions of the phosphorylation loop region and the altered side chain conformation involving Tyr286-, compared with wild-type holoE1b, are present in the D295A- (Fig. 5C) and Y300F- (Fig. 5D) mutants.
Disorder-to-Order Transition of the Loop Is Confirmed by Limited Proteolysis—Wild-type and mutant E1b proteins were incubated with chymotrypsin at 0 °C for different lengths of time. After termination of the protease digestion with PMSF, reaction mixtures were analyzed by SDS-PAGE. A significant amount of the subunit from wild-type apoE1b is digested after 3 h with production of three major fragments: F1 (37 kDa), F2 (35 kDa), and F3 (10 kDa) (Fig. 6A). In the presence of 1 mM ThDP, the subunit of wild-type E1b is largely protected from proteolysis. An even higher degree of protection from proteolysis occurs in wild-type E1b in the presence of 1 mM ThDP and 2 mM KIV than with 1 mM ThDP alone (data not shown). Peptide sequencing of the three chymotryptic fragments shows N-terminal sequences of GGSHH for F1, FQSSL for F2, and RSVDE for F3, all from the subunit. The N-terminal residues of F1 and F2 fragments correspond to the original and digested N-terminal region of the recombinant subunit, respectively, with SSL in F2 representing the N terminus of native mature subunit. The N-terminal residues of F3 fragment are from the phosphorylation loop and correspond to residues 301-305 of the subunit (Fig. 1A). The above results support the notion that the loop in apoE1b is disordered and that ThDP binding to apoE1b promotes a disorder-to-order transition, rendering the loop resistant to proteolysis. A substitution of Tyr300- with alanine results in nearly complete protection from chymotryptic digestion due to the loss of the chymotrypsin recognition site (Fig. 6B). This result is consistent with the presence of a major chymotryptic cleavage site after Tyr300- in the phosphorylation loop region. The conversion of Y300F- mutation makes the apoform of this mutant significantly more susceptible to proteolysis than wild-type apoE1b, with a nearly complete digestion of the subunit at 3 h (Fig. 6C). ThDP binding to the apoY300F- mutant does not increase its resistance to chymotryptic digestion (Fig. 6C). Chymotrypsin cleaves peptides after a tyrosine or a phenylalanine residue with a similar efficiency (27). Therefore, the data indicate that the Y300F- mutation disrupts the hydrogen-bonding network, and abolishes the disorder-to-order transition induced by ThDP binding. The absence of protection from digestion by chymotrypsin in the presence of ThDP was also observed with R287A-, D295A-, and R301A- E1b mutants (data not shown).
ThDP Binding Inhibits Phosphorylation in Wild-type but Not Mutant E1b—The disorder-to-order transition of the phosphorylation loop region induced by ThDP binding inhibits the BCKD kinase-mediated incorporation of [-32P]phosphate from ATP to the subunit of wild-type E1b (Fig. 7A). The IC50 value of ThDP for inhibition of phosphorylation is 10 µM. Under the same conditions, the inhibition of phosphorylation by ThDP is not observed with the R287A-, D295A-, Y300A-, and R301A- mutants (Fig. 7A) as well as the H291A- variant (not shown). Fig. 7B shows that [-32P]phosphate is incorporated into the major phosphorylation site Ser292- of the S302A- variant (lane 1). The S302A- substitution abolishes the minor phosphorylation site Ser302- of human E1b. Substitutions of residues in the hydrogen-bonding network (R287A-, D295A-, Y300F-, and R301A-) in the S302A- E1b variant have no significant effects on the phosphorylation of Ser292- in the absence of ThDP. No 32P incorporation is detected with the S292A-/S302A- double mutant, which serves as a negative control. Similar results with respect to the inhibition of phosphorylation by ThDP (Fig. 7A) were obtained with E1b double mutants containing the S302A- substitution and one of the above mutations in the phosphorylation loop region (data not shown).
|
DISCUSSION
|
|---|
The present study focuses on the significance of the conserved phosphorylation loop region in the human E1b active site. According to their crystal structures, the ThDP-binding domains from human and Pseudomonas E1b, along with E1p proteins from eukaryotes and Gram-positive bacteria, form a distinct class of ThDP-dependent enzymes that feature the heterotetrameric 22 structure (pdb.weizmann.ac.il/scop/data). In the human E1b active site, the subunit provides amino acid ligands including Arg287- and His291- anchors for the diphosphate moiety of ThDP, whereas the ' subunit harbors a hydrophobic pocket to accommodate the aminopyrimidine and thiazolium rings of the cofactor. The latter subunit also contains the invariant electron-withdrawing catalytic residue Glu76-' crucial for activating ThDP for the decarboxylation of -keto acids (Reaction 2). In the present study, electron density for the phosphorylation loop region is absent in the human apoE1b crystal structure. The presence of electron density for the same region in the holoenzyme, therefore, corresponds to a disorder-to-order transition of the loop upon ThDP binding. Both apo- and holoenzymes were crystallized under identical conditions. In addition, no electron density for the phosphorylation loop region is visible in R287A- and D295A- holoE1b variants. The phosphorylation loop region is also disordered in the Y300F- holoE1b variant, indicating that the tyrosine with its hydroxyl group is essential for the maintenance of the hydrogen-bonding network. The conformational change associated with the aromatic side chain of Tyr286- appears to be important for efficient ThDP binding as it orients the neighboring Arg287- to adopt a position favorable for coordination to the cofactor. The Km for ThDP is increased by 1 order of magnitude in the Y286A- mutant compared with the wild type (data not shown). Taken together, the data indicate that an intact hydrogen-bonding network is essential for maintaining the ordered conformation of the phosphorylation loop region in the presence of bound ThDP.
Ligand binding-induced conformational changes are common in cofactor-dependent enzymes. In human riboflavin kinase, FMN binding was shown to result in the ordering of the Flap II loop that in turn interacts with and orders the Flap I loop (28). The binding of ThDP to yeast transketolase, the homodimeric class of ThDP-dependent enzymes in which the and subunits are fused, orders two loops in the active site, which are disordered in the apoenzyme but ordered in the holoenzyme (29). The first loop in holotransketolase (residues 187-198) is involved in the binding of a Ca2+ ion that is in turn required for ThDP binding. Residues in this loop interact extensively with residues from the second loop (residues 383-393) of the neighboring subunit to stabilize the homodimeric structure. The ordering of these two loops in transketolase also closes off the metal ion and cofactor from the solvent (30). The ordering of the loop that interacts with the Mn2+ cation ion upon ThDP binding in human E1b (Fig. 4) is equivalent to the Ca2+-induced disorder-to-order transition in yeast transketolase. However, the stabilization of the hydrogen-bonding network in the phosphorylation loop region upon ThDP binding appears to be a unique property of the heterotetrameric human E1b as neither yeast transketolase nor homodimeric E1 proteins, e.g. E. coli E1p (31), feature an analogous phosphorylation loop. Within the heterotetrameric E1 proteins, the stabilization of the phosphorylation loop region upon ThDP binding likely is a general feature because of the high degree of sequence conservation (Fig. 1A).
The R287A- mutation in the phosphorylation loop causes an order-to-disorder transition of the loop because of the loss of both the Arg287- and the His291- interactions with ThDP (Fig. 5B). These conformational changes provide the structural basis for the markedly reduced affinity of mutant E1b for ThDP as indicated by nonphysiological increased Km (Table I) or Kd values (Table II) compared with wild-type E1b. Similar effects are observed when any one of the other three residues in the hydrogen-bonding network (Asp295-, Tyr300-, and Arg301-) is converted to alanine or phenylalanine (in the case of Tyr300-). It has been suggested that Arg267- in the heterotetrameric E1p from B. stearothermophilus, which is homologous to Arg287- in human E1b, interacts with the carboxylate group of -keto acid substrates (13). In E1b, Arg287- is 10 Å away from the reactive C-2 atom of the thiazolium ring and across from His146-' in the E1b active-site channel, making it unlikely for this residue to be involved in substrate binding and stabilization. Rather, the increased Km values of the R287A- mutant for substrates (Table II) over the wild type (when measured by E1b-catalyzed decarboxylation, Reaction 2) may be because of the reduced affinity of this mutant for ThDP associated with the disordering of the phosphorylation loop region.
An interesting aspect of our kinetic studies with the E1b mutants is their higher kcat values for the E1b-catalyzed decarboxylation reaction (Reaction 2) over the wild type. The mechanism for these increased kcat values is presently unknown. It has been suggested that the increased kcat value may be, in part, due to favorable entropic effects in the mutants through a more efficient removal of reaction products or increased accessibility of the electron acceptor DCPIP (13). Both mechanisms are likely promoted by the increased flexibility in the phosphorylation loop region. However, the accessibility and reduction of DCPIP are unlikely the rate-limiting steps, because similar increased decarboxylation rates with E1b mutants over the wild type were obtained in the presence of DCPIP or with the smaller ferricyanide, or in the absence of any artificial electron acceptor (data not shown). Similar results were obtained for the D295A- E1b mutant either by measuring the reduction of DCPIP spectrophotometrically (Table II) or the release of radiolabeled CO2 in the presence of DCPIP with [1-14C]KIV as a substrate (Fig. 2). These findings rule out the possibility that the increased kcat value for the mutants over the wild-type E1b is an artifact of the spectrophotometric DCPIP assay.
The Km value of wild-type E1b for ThDP is 10-fold higher when E1b decarboxylation activity was assayed with E1b alone (Table II) than when measured as a reconstituted BCKD complex (Table III). The results indicate that the binding of E2b induces conformational changes in E1b so as to increase its affinity for ThDP. The markedly decreased Km value for ThDP is not observed with E1b variants carrying mutations in the hydrogen-bonding network. Mutations in the conserved phosphorylation loop region do not affect binding of mutant E1p to the subunit-binding domain from the E2 component of pyruvate dehydrogenase complex from B. stearothermophilus (13). This result excludes the involvement of E1b binding to the subunit-binding domain of E2b in mediating the affinity of the BCKD complex for ThDP. Therefore, the absence of E2b-induced increase in affinity for ThDP in the hydrogen-bonding network mutants may be because of the order-to-disorder transition in the phosphorylation loop region. The data suggest that an ordered loop region maintained by a hydrogen-bonding network in the E1b active site functions as a mediator for the increased affinity of E1b for ThDP upon E2b binding.
The human E1b component is distinguishable from other ThDP-dependent enzymes, for example, yeast transketolase (29), in that it is both a decarboxylase (Reaction 2) and a dehydrogenase (Reaction 3). Mutations in the phosphorylation loop hydrogen-bonding network impede the dehydrogenase but not decarboxylase activity of these mutant human E1b proteins, because they are unable to or weakly interact with the lip-LBD from E2b, which serves as a substrate for Reaction 3. The loop conformation maintained by the hydrogen-bonding network appears to be essential for recognition by lip-LBD in the E1b active site. The binding of lip-LBD to the E1b active site initiates substrate channeling from the initial -keto acid to a branched-chain acyl-CoA (Reactions 2-4) in the BCKD metabolic machine. The assay for reductive acylation based on combined Reactions 2 and 3 is valid because the former reaction is irreversible and much faster than the latter. The significant residual activity of reductive acylation in the Y300F- and R301A- mutants may result from the fact that an unaltered residue in these two positions is still hydrogen-bonded to Arg287- such that the phosphorylation loop region is at least partially stabilized. The markedly lower rates of reductive acylation (Reaction 3) in the mutant E1b than the wild type, regardless of the elevated kcat values for E1b-medicated decarboxylation, are parallel to the decreased rates of the overall reaction (Reaction 1) catalyzed by the BCKD complex. A case in point, the lack of reductive acylation activity in the R287A- mutant E1b is consistent with the null overall decarboxylation activity of the mutant BCKD complex reconstituted with this E1b variant. These data corroborate the proposition that the reductive acylation of lip-LBD on E2b is rate-limiting in the overall reaction of -keto acid dehydrogenase complexes (13).
The phosphorylation of mammalian E1b by the BCKD kinase results in the inactivation of the BCKD complex (32, 33), and the binding of ThDP to E1b inhibits its phosphorylation (34). In the present study, apoE1b shows a maximum level of phosphorylation, but ThDP cannot impede the phosphorylation of mutant E1b proteins with a disordered phosphorylation loop region caused by the disruption of the hydrogen-bonding network. Therefore, a disordered phosphorylation loop appears to be required for E1b phosphorylation by the BCKD kinase. Our results support a recent proteome-wide predictor analysis that showed that protein phosphorylation sites predominantly occur within intrinsically disordered regions (35). The regulation of E1b phosphorylation by ThDP may be physiologically relevant. The intracellular concentration of ThDP was estimated to be around 30 µM, with > 90% of the cofactor in the bound form and much of this in mitochondria (36). However, the extent to which holoE1b is present in a tissue-specific manner is unknown. Mitochondria are capable of high affinity uptake of ThDP with a Km of 0.4-0.6 µM (37). Therefore, the inhibition of E1b phosphorylation with IC50 at 10 µM can take place in this organelle when cells are exposed to high thiamin concentrations. During the enhanced BCKA oxidation in response to nutritional and hormonal stimuli, an elevated mitochondrial ThDP concentration can activate E1b by increasing the proportion of functional holoE1b and concomitantly prevent its inactivation by phosphorylation. These concerted mechanisms ensure that the mitochondrial E1b is locked in the active form to meet these physiological demands.
We report the following findings that are in direct contrast to those reported previously (14). First, the R287A-/S302A- double substitutions do not significantly reduce the level of 32P incorporation into Ser292-, as compared with the S302A- single mutant that serves as a positive control for site 1 phosphorylation (Fig. 7B). Second, the inhibitory effect of ThDP on phosphorylation is abolished in the D295A- E1b variant, similar to other E1b mutants harboring substitutions in the hydrogen-bonding network. Finally, the D295A- E1b variant shows 25% wild-type activity for the reconstituted BCKD complex, in contrast with its complete absence reported previously (14). The reasons for these differences are not entirely obvious. However, our data do not support the suggestion that Arg287- is important for recognition by the BCKD kinase (14). A synthetic oligopeptide in which the equivalent arginine residue in the conserved phosphorylation loop region is not included can be phosphorylated by the related pyruvate dehydrogenase kinase (38).
In conclusion, we demonstrate that the conserved phosphorylation loop region in the E1b active site plays a pivotal role in substrate channeling by binding to the lip-LBD domain of the BCKD metabolic machine. For lip-LBD recognition, the phosphorylation loop region requires a specific conformation that is maintained through the following: 1) the bound cofactor ThDP with Arg287- and His291- as the key anchors, and 2) an intricate hydrogen-bonding network involving Arg287-, Asp295-, Tyr300-, and Arg301-. A perturbation of either component results in the disordering of the phosphorylation loop region and abolishes (in the case of R287A- E1b) or interferes with the recognition of the phosphorylation loop region by lip-LBD. The net consequence is the interruption of substrate channeling in the BCKD machine. On the other hand, ThDP binding to apoE1b orders the phosphorylation loop region by promoting the formation of the hydrogen-bonding network. The induced loop conformation in the E1b active site in turn prevents the phosphorylation of Ser292- by the BCKD kinase to maintain E1b in the fully active state. Our in vitro data indicate that the binding of cofactor ThDP to E1b may function as a feed-forward switch for substrate channeling in the BCKD metabolic machine by ordering the phosphorylation loop region in the E1b active site. This mechanism may serve as a paradigm for E1 components of mammalian -keto acid dehydrogenase complexes in particular and other cofactor-dependent metabolic machines in general.
|
FOOTNOTES
|
|---|
The atomic coordinates and structure factors (codes 1V1R
[PDB]
, 1V11
[PDB]
, 1V1M
[PDB]
, and 1V16
[PDB]
) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).
* This work was supported by Grants DK-26758 and DK-62306 from the National Institutes of Health and Grant I-1286 from the Welch 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. 
¶ To whom correspondence should be addressed: Dept. of Biochemistry, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75390-9038. Tel.: 214-648-2457; Fax: 214-648-8856; E-mail: david.chuang{at}utsouthwestern.edu.
1 The abbreviations used are: BCKD, branched-chain -keto acid dehydrogenase; BCKA, branched-chain -keto acid; DCPIP, 2,6-dichlorophenolindophenol; E1b, branched-chain -keto acid decarboxylase/dehydrogenase; E2b, dihydrolipoyl transacylase; E1p, E1 component of pyruvate dehydrogenase complex; E3, dihydrolipoamide dehydrogenase; ITC, isothermal titration calorimetry; KIV, -ketoisovalerate; LBD, lipoic acid-bearing domain; lip-LBD, lipoylated lipoyl-bearing domain; ThDP, thiamin diphosphate; PMSF, phenylmethylsulfonyl fluoride.
|
REFERENCES
|
|---|
- Chuang, D. T., Chuang, J. L., Wynn, R. M., and Song, J.-L. (2001) in Encyclopedia of Molecular Medicine (Creighton, T. E., ed) Vol. 5, pp. 393-396, John Wiley & Sons, Inc., New York
TOP
ABSTRACT
INTRODUCTION
) of wild-type human holoE1b discloses that the phosphorylation loop region between Tyr286-
FQSSL. The arrow indicates the chymotryptic cleavage site. The underlined sequences, from the N to the C terminus, represent the His6 tag and the tobacco-etch virus protease recognition site, respectively, in the expression vector. The SSL sequence (in boldface) in fragment F2 represents the N terminus of the native mature E1