Additional binding sites for the pyruvate dehydrogenase kinase but not for protein X in the assembled core of the mammalian pyruvate dehydrogenase complex: Binding region for the kinase

ARCHIVES

OF BIOCHEMISTRY

AND

BIOPHYSICS

Vol. 296, No. 2, August 1, pp. 497-504, 1992

Additional Binding Sites for the Pyruvate Dehydrogenase Kinase but Not for Protein X in the Assembled Core of the Mammalian Pyruvate Dehydrogenase Complex: Binding Region for the Kinase Lin Li, Gary A. Radke, Kazuo Ono, and Thomas Department

of Biochemistry,

Kansas State University,

E. Roche’

Manhattan.

Kansas 66502

Received December 13, 1991, and in revised form April 3, 1992

A standard resolution of the bovine kidney pyruvate dehydrogenase complex yields a subcomplex composed of -60 dihydrolipoyl transacetylase (EZ) subunits, -6 protein X subunits, and -2 pyruvate dehydrogenase kinase heterodimers (KKb). Using a preparation of resolved kinase in which K, 4 Kb, EP-X-KCKt, subcomplex additionally bound at least 15 catalytic subunits of the kinase (K,) and a much lower level of Kb. The binding of K, to E2 greatly enhanced kinase activity even at high levels of bound kinase. Free protein X, functional in binding the E3 component, did not bind to E2-X-K,Kt, subcomplex. This pattern of binding K, but not protein X was unchanged either with a preparation of E2 oligomer greatly reduced in protein X or with subcomplex from which the lipoyl domain of protein X was selectively removed. The bound inner domain of protein X associated with the latter subcomplex did not exchange with free protein X. These data support the conclusion that E2 subunits bind the K, subunit of the kinase and suggest that the binding of the inner domain of protein X to the inner domain of the transacetylase occurs during the assembly of the oligomeric core. Selective release of a fragment of E2 subunits that contain the lipoyl domains (E2L fragment) releases the

1 To whom correspondence should be addressed. * Abbreviations used: PDC, pyruvate dehydrogenase complex; El, pyruvate dehydrogenase component; E2, dihydrolipoyl acetylatransferase component; E21, COOH-terminal, transacetylase-catalyzing, oligomerforming inner domain of E2; E2s, El-binding domain of E2; E2L,, NHterminal lipoyl domain; E2r,, , inner lipoyl domain of E2; E2L, fragment containing E2~r, E2Lz. and COOH-linked hinge regions of these domains; E3, dihydrolipoyl dehydrogenase component; K,, catalytic subunit of Ela kinase; Kb, basic subunit of the kinase; X, protein X or component X, XI, inner domain of protein X; McAb, monoclonal antibody; H, McAb heavy chain; L, McAb light chain; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; MOPS, 4-morpholinepropanesulfonic acid. 0003-9861/92 $5.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form

kinase (M. Rahmatullah et aE., 1990, J. Biol. Chem. 265, 14,512-14,517). Sucrose gradient centrifugationyielded an EaL-kinase fraction with an increased ratio of the kinase to E2L fragment. A monoclonal antibody specific for E2L was attached to a gel matrix. Binding of E2L fragment also led to specific binding of the kinase. Extensive washing did not reduce the level of bound kinase. Thus, the kinase is tightly bound by the lipoyl domain region of E2. 0 1992 Academic PWS, 1~.

The mammahan pyruvate dehydrogenase compiex (PDC)’ has a structural core composed of two lipoylbearing components-the dihydrolipoyl transacetylase (E2) and protein X. These components bind the other cardinal components (components required for the overall reaction catalyzed by the complex) as well as the regulatory enzymes (l-8). The E2 component contains a COOH-terminal inner domain (Ear) and an extended NHa-terminal structure (9) composed of three domains with connecting hinge regions between each of the four domains (10, 11). Connected to E2r is a domain (designated E2s) that binds the pyruvate dehydrogenase (El) component (4, B), then two lipoyl domains with the outermost (NH*-terminal) domain being designated E2L1 and the lipoyl domain between E2L1 and E2s domains being designated E2Lz. The protein X component also contains a lipoyl domain and an inner domain structure (XI) (4, 5,12, 13). Mammalian protein X binds the dihydrolipoyl dehydrogenase (E3) component (4, 6, 7). The X component of yeast PDC contains a linker region-connected three-domain structure, with a domain between the lipoyl domain and the inner domain that binds the E3 component of yeast PDC (13, 14). The pattern of proteolytic cleavage of the mammalian protein X (3) is consistent with such a three-domain structure. The inner domain of 491

Inc. reserved.

498

LI ET AL.

mammalian protein X associates with the inner domain of the transacetylase (3) and together these form an oligomer apparently composed of about 60 E2r and about 6 Xr domains. The location for binding of the XI domain to the dodecahedron structure formed by association of E2r domain has not been demonstrated. The initial process of resolution of the bovine kidney complex yields a subcomplex (M, N 4 X 106)that contains about two pyruvate dehydrogenase kinase dimers (K,K,) in addition to the E2 and X subunits. A further step can remove the kinase and some protein X (15, 16). Reassociation of the kinase with the E2-X subcomplex causes a large increase in kinase activity (15-18). Here we evaluate whether resolved kinase and resolved protein X (functional in the binding of the E3 component) can bind to the assembled EB-X-K,Kb subcomplex, the assembled E2 oligomer greatly reduced in the content of protein X (6), or the EB-Xr-K,(KJ subcomplex in which the lipoyl domain of protein X was selectively removed (5). The addition of resolved X to the latter XI-containing subcomplex not only allowed the possibility of binding additional protein X but the possible exchange of XI domains and protein X to be evaluated. The particular fraction of resolved kinase employed was one with a high K:Kb ratio, allowing the dependence of binding and activation of the catalytic subunits (K,) on the presence of the basic subunit (Kb) to be evaluated. Rahmatullah et al. (8) have shown that selective and complete release of an E2L fragment (composed of E2L1 and E2L2 plus the first and second hinge regions) also released the kinase subunits (K,K,). That suggests the binding site for the kinase resides in the E2L fragment and probably involves an association with one of the two lipoyl domains. Furthermore, assuming all E2L regions of E2 subunits are equivalent, this predicts that there should be a capacity for binding more than the two to three K,K,, dimers associated with the purified complex. Here we evaluate whether there is a higher binding capacity that explains the capacity of the subcomplex to greatly enhance the activity of a large number of kinase molecules (18). We eliminate the possibility (18) that activation and binding may have depended on the presence of protein X in the kinase fraction. Finally, we determine whether the kinase remains tightly associated with the E2L fragment. EXPERIMENTAL

PROCEDURES

Materials. Minor modifications of standard procedures were used to prepare the bovine kidney pyruvate dehydrogenase complex (19), the EZ-X-KKb subcomplex and El component (l), and the X, KK, fraction along with the E2-X subcomplex (15,16). The E2 oligomer with essentially all the protein X and kinase subunits removed was prepared by

3 After the minimum period required for cleavage of protein X, a portion (~40%) of the Kb subunit as well as a small portion of the E2 subunits (<6%) was cleaved (3). The E2-X-K,Kb used in many of the present studies had a low initial level of Kb.

the procedure of Powers-Greenwood et al. (6). The lipoyl domain of protein X was selectively removed and the EB-XI-K,(Kb) subcomplex prepared using protease Arg C (3).3 The EZL-K,Kb was selectively released from the EQ-X-K,Kb subcomplex using Clostridium histolyticum collagenase (8). [Y-~~P]ATP and [nC]ATP, and [1-‘?]acetyl-CoA were purchased from New England Nuclear. BASF Wyandotte Corp. supplied Pluronic F68 which is a block polymer with, on average, about 30 polypropylene glycol units flanked on each end by about 75 polyethylene glycol units. Binding of the kinase and protein X to E2 subcomplexes. Ef-containing subcomplexes were incubated with the X, K,K+, fraction at 22°C for 15 min in 50 mM Na-phosphate buffer (pH 7.5) containing 0.2 mM EDTA and 0.5 mM dithiothreitol in a total volume of 50 ~1. These fractions were then overlaid on a three-step sucrose gradient (20~~1 steps of 7.5, 10, and 15% w/v in 5 X 20-mm ultraclear tubes [Beckman]). Centrifugation was conducted at 130,OOOgfor 2 h in a custom designed Teflon holder placed at the bottom of the buckets of a SW 28 rotor as previously described (3, 6). After centrifugation, supernatants (110 ~1) were withdrawn and 55 ~1 of the above buffer was used to dissolve the pellets over 14-h period at 4°C. The protein concentrations of the supernatant and pellet fraction were determined in duplicate by the procedure of Fried et al. (33) using excitation and emission wavelengths of 330 and 418 nm, respectively. Bovine serum albumin was used as a protein standard. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed on samples of supernatants, pellets, and other samples using the Laemmli procedure (19). Bands were visualized with a silver stain by the procedure of Oakley et al. (20). Kinase activity, ATP binding, and transacetylase actiuity. Pyruvate dehydrogenase kinase activity was determined as the initial rate of incorporation of 32P-phosphoryl groups into El from [y-s’P]ATP (16,18). Assays were conducted at 30°C in 20 mM potassium phosphate (pH 7.0), 1 mM MgClr, 0.1 mM EDTA, and 2 mM dithiothreitol. The final two additions to assays were the kinase source followed by 60 to 120-s incubation and then [y-3ZP]ATP (1.5-3.0 X lo5 cpm/nmol) was added to a final concentration of 100 pM for the indicated times. Reactions were terminated and incorporation determined as previously described (21). Assays were conducted in duplicate. Binding of [“C]ATP (22) was conducted using a cold-trapping procedures as previously described (18). To kinase-containing fractions in 50 mM Mops-K buffer (pH 7.3) containing 60 mM KCl, 2 mM dithiothreitol, 1.5 mM MgCIz and 0.5 mM EDTA, [r4C]ATP (92,000 cpm/ nmol) was added for 90 s at 22°C in a total volume of 25 ~1. Then a 20pl aliquot was applied to a Millipore filter that had been presoaked with 5 mM ATP and washed with 20 mM potassium phosphate, pH 7.0, at 0°C. The filter was immediately washed with six 2-ml washes of 20 mM potassium phosphate at 0°C. After the membrane was dried, radioactivity was determined. Assays were performed in duplicate. Transacetylation activity was determined by the procedure of Butterworth et al. (23). Reactions were initiated at 22°C by addition of [lr4C]acetyl-CoA (6 X lo5 cpm/@mol) to a final concentration of 0.5 mM in a reaction mixture containing 25 mM potassium phosphate (pH 7.5), 0.5 mM dihydrolipoamide, 0.05 mM EDTA in 0.5 ml. One milliliter benzene was added after 120 s and the reaction mixture was vortexed for 30 s to stop the reaction and to extract the S-[I-r4C]acetyl dihydrolipoamide. Two 0.2-ml aliquots of the benzene layer were withdrawn and radioactivity was determined. Preparation, initial characterization, and utilization of monoclonal antibodies. Surh et al. (24) isolated monoclonal antibodies that reacted with E2 and characterized the specific epitope using fusion proteins containing different segments of the human E2. After repeated subcloning (leading to 0.2 designation after their original numbers) and preparation of ascites fluid by standard procedures (25), homogeneous monoclonal antibodies were prepared in milligram amounts by ammonium sulfate fractionation followed by desalting and chromatography using an 7.5 X loo-mm Bakerbond ABx column (26). The column was equilibrated and washed for 10 min at 0.7 ml/min with P-(N-morphololino)ethanesulfonic acid * Na (pH 5.0) and then eluted with linear gra-

499

PYRUVATE DEHYDROGENASE KINASE BINDING SITE diem of 1 M Na-acetate (pH 7.0) (25). SDS-PAGE establishedthat the monoclonal antibodies were purified to homogeneity. Monoclonal antibodies were desalted into and stored in 50 mM Mops*Na, pH 7.0. McAbs were attached to Avid-Gel AX using standard procedures for NaIO, oxidation with the matrix hydrazide (27). Properties of several of these monoclonal antibodies on the function of the complex, kinase, and phosphatase will be described elsewhereP

X,Wb

(ug)

-E2-X-&Kb0.6 0.5 0.4 0.3 0.2 0.1

E2-X-K&, s

Pl

E? K&,-

P2 P3 s

X-K,& P

s

P

X,KcKB

z

1.0 0.5

I.0

k31

RESULTS

Binding of the kinase to different assemblages of the diWe evaluated whether adhydrolipoyl transacetylase. ditional molecules of kinase and of protein X in an X, K,K,, fraction (16) could bind to the E2-X-K,Ki, subcomplex, the E2 oligomer (prepared with most but not all the protein X removed), or the EB-Xi-K,KI, subcomplex (the outer, lipoyl-bearing domain of X removed). Kinase binding is considered first. The X, K,Kb fraction was deficient in the Kb subunit and contained E3 that was mostly inactive (16). We also employed an unusual E2-X-K,Kb preparation in which Kb < K, .5 These mixtures and control samples were pelleted through a sucrose gradient followed by determination of kinase activity, f14C]ATP binding capacity, E2 activity, and protein levels for the pellet and supernatant fractions as described under Experimental Procedures. Pelleting a mixture of 60 /*g X, K,Kb fraction combined with 30 pg EB-X-K,Kb subcomplex gave the results shown in Fig. 1. By comparison of the band intensities for three levels of the pellet fraction (PI, Ps, P3) to the various levels of the ES-X-K,Kb subcomplex (first six lanes), we conclude the pellet fraction has at least a sixfold increase in the ratio of intensity for the K, band relative to the E2 band. For instance, pellet fraction P3 (0.14 pg protein loaded to gel) has no more E2 and X than in 0.1 pg E2X-K,Ki, but contains at least as much K as sixfold higher level (0.6 pug) of ES-X-K,Kb. A more precise analysis based on activities is given below. Subcomplex combined with X, K,Kb fraction pelleted more rapidly than subcomplex alone (Fig. 1); the basis of this effect is not known but cannot be solely attributed to the increase in mass due to binding additional kinase molecules. Neither X nor kinase subunits pelleted as a consequence of gradient centrifugation of X, K,Kr, fraction (Fig. 1). The increase in binding of K, subunit to E2-X-K,Kb greatly exceeded the amount of Kb, indicating that Kb was not required for the association of the K, subunit. The binding of protein X (analyzed further below) was also much less than K,, demonstrating protein X was not required for K, binding. The fractions obtained in Fig. 1 were further analyzed for changes in kinase and E2. Table 1 shows [14C]ATP 4 C. L. Chang, K. Ono, G. A. Radke, W. Li, C. D. Surh, M. E. Gershwin, and T. E. Roche, manuscript in preparation. ’ SomeEZ-X-K,Kb preparations appearto have higher K, than Kbr but even when this is not the case, the K-to-K, ratio can be increased in preparation of X, K,Kt, fraction.

FIG. 1. SDS-PAGE analysis of the binding of kinase and protein X to the E2-X-K,Kb subcomplex. X, K,Kb fraction (60 pup) and E2-XK,Kb subcomplex (30 pg) incubated alone or together and then pellet and supernatant fractions prepared as described under Experimental Procedures. From left to right the first six lanes contained untreated E2-X-K,Kb subcomplex loaded at the indicated levels for reference purposes. The following four lanes contained 0.46 rg protein from 1.2 ~1 of supernatant fraction (S) and 0.58 pg (PI), 0.29 pg (PZ), and 0.14 pg (P3) protein samples from 0.6, 0.3, and 0.15 ~1, respectively, of the pellet fraction (P) from the mixture of X, K,Kb and E2-X-K,Kb. Then to the right are shown equivalent portions (1 ~1 supernatant and 0.5 ~1 pellet fraction) from the simultaneous centrifugation of the EB-X-K,Kb subcomplex (0.18 gg supernatant protein and 0.21 bg pellet protein) and of the X, K,Kb fraction (0.44 pg supernatant and -0.02 wg pellet). In the final three lanes untreated X, K,Kb fraction and pyruvate dehydrogenase complex were loaded at the indicated levels. Electrophoresis and staining were conducted as described under Experimental Procedures.

binding capacity of the kinase, the specific activity of the kinase, and the transacetylation activity for the initial preparations and for the pellet and supernatant fractions. Based on the increase in the ratio of [14C]ATP binding sites to the E2 activity (and assuming an original M, of the E2-X-K,Kb subcomplex of 4.1 X 1O”),6 we estimate the number of associated kinase (K,) subunits increased from 2.1 to 17.7 k 1.3 per subcomplex. The kinase in the supernatant was very active. The maximum potential (if all added kinase subunits added were bound) was 28 kinase subunits per subcomplex. In an experiment in which the maximum potential was 10 kinase subunits per subcomplex, about 90% of the kinase subunits of this same preparation were bound (29). Based on a specific activity of 30 min.’ (i.e., 30 mol mini’ per mole [14C]ATP binding site (18)), we estimate a similar value of 17.6 +- 1.2 kinase subunits bound per

6 The M, of the ES-X-K,Kb subcomplex is based on the best available estimate of subunit M, values and their stoichiometrics (28). The M, of 4.1 X lo6 assumes the EZ-X-K,K,, subcomplex is composed of 60 59.5kDa E2 subunits, six 50-kDa protein X, two 46-kDa K,, two 43-kDa Kb, and 0.5 E3 dimers.

500

LI ET AL. TABLE

Pyruvate

Dehydrogenase

[14C]ATP bound”

Fraction

(rim01 mg-‘)

I

Kinase Content and Ratio to Oligomeric

Kinase specific activity* (nmol mini mg-‘)

E2-X-K,K,

0.50

11.9 (9.1)

X KK,

3.28

97.5

E2-X-K,% + X, I
2.84

84.9 (56.9)*

1.53

39.1

’ ATP binding to the kinase (15) measured with 180 pM [“C]ATP from duplicate assays were within +0.02 for E2-X-KKb and within

E2 Core

Transacetylase specific activitf (pm01 min-’ rng-‘)

ATP binding site per E2 oligomer“

4.4 0

2.1

2.9

17.7

n.d.’

under conditions described under Experimental Procedures. Average values +0.04 for all other measurements. Protein levels in assays were 12.2 pg E2-

X-K,Kb subcomplex, 4.0 ng X, K,Kb fraction, 7.8 pg pellet fraction, and 5.1 pg supernatant fraction. The pellet and supernatant were prepared as described in Fig. 1 and under Experimental Procedures. *Assays were conducted in the presence of 22 pg El and 5 ng E2-X subcomplex (except for the value shown in parentheses, in which E2-X was omitted from the assays). Activities were corrected for the low kinase activity contributed by these additions. The 50-nl kinase assay mixtures contained the following levels of protein from the indicated fractions; 5 pg EZ-X-K,Kb, 1.2 pg X, K,K,, 1.2 rg pellet fraction, or 0.51 pg supernatant fraction. After these fractions were added, the mixtures were incubated for 120 s at 30°C before the final addition of [Y-~*P]ATP. The reaction was terminated after 15 s (pellet fraction) or 30 s (other fractions). All duplicates gave values within 4% of each other. Other conditions were as described under Experimental Procedures. ’ E2 transacetylation activity was measured as described under Experimental Procedures for fractions added at the following levels: 6.1 pg E2X-K,Kb subcomplex, 10 pg X, K,K,,, or 4.1 pg pellet fraction. Duplicates gave values within 3% of each other. d The M, of EB-X-K,Kb, subcomplex was assumed to be 4.1 X 106; thus 1 mg of subcomplex was equivalent to 0.244 nmol of oligomeric E2. Then, 0.5 nmol ATP binding sites per 0.244 nmol E2 core implies 2.06 ATP binding sites in the initial subcomplex. The ATP binding sites per subcomplex in the pellet fraction equals 2.84 nmol ATP bound mg-’ pellet-’ * (4.4 nmol min-’ mg-i subcomplex-l/2.9 nmol mini’ mg-’ pellet-‘) +4.1 mg subcomplex nmol-’ subcomplex-’ = 17.7 ATP bound subcomplex -l. Considering the errors in various measurements, a range from 16.3 to 19.0 ATP bound subcomplex-’ is estimated.

’ Not determined.

Except for the value shown in parentheses the kinase activity measurements were made with extra E2-X subcomplex included in assay mixtures to ensure the -fivefold activation (15-18) of the kinase caused by the E2-X subcomplex (16, 18). A short (15 s) reaction time was used to minimize depletion of the prosubcomplex. in Table I,

tein substrate. Over the years we have observed a O-20% increase in the kinase activity of resolved E2-X-K,Kb

upon addition of kinase depleted E2-X subcomplex (activities corrected for kinase activity of E2-X plus El substrate above). In an experiment in which nine molecules of kinase were bound per E2-X-K,Kb (29), there was less than 25% increase in kinase specific activity (activity per ATP binding site) due to inclusion of extra E2-X subcomplex with the pellet fraction (data not shown). A 49% increase was observed when -18 molecules were bound (Table I). The latter may reflect an inability to measure true initial velocities or a somewhat diminished capacity ’ This value is calculated using M, of 4.1 X lo6 for EB-X-K,Kb subcomplex and the level of subcomplex in the pellet is estimated based on E2 activity (Table I). The deviation reflects protein and kinase measurements. Some additional uncertainty is introduced in using 30 mini for kinase specific activity. The X, KKb fraction used here had a specific activity of 29.7 min-’ (calculated from data in Table I) which is within experimental error of the average value of 30 min-‘. However, we have observed a range from 26-34 min-’ for the specific activity of different kinase preparations.

of the subcomplex to activate the kinase when so many molecules of kinase are bound. The enhanced activity was consistent with the expectation that some redistribution of kinase molecules onto the added E2-X would allow more El molecules to be available to the kinase. An equilibrium distribution was probably not achieved in 120 s preincubation, but longer times did not increase the kinase activity observed. Furthermore, it should be emphasized that most of the -fivefold activation caused by association of the kinase with the subcomplex was observed with about 17 kinase (K,) subunits attached even in the absence of added E2-X, indicating a similar association of most if not all bound kinase molecules. As shown in Fig. 2, addition of X, K,Ki, fraction to the E2 oligomer and by E2-Xi-K,Kb subcomplex (lanes with + designation at the top) led to binding of extra kinase. Thus, removal of protein X or its outer domain did not affect binding of the kinase by E2 subunits. Binding of free protein X to the transacetylase cores. Upon comparison of intensities of E2 and protein

X bands (Fig. 1) in the series of untreated EB-X-K,Kb to those bands in the series (PI, PZ, P3) from the pellet fraction of subcomplex exposed to X, K,Kb, one finds that intensities for E2 and protein X for a given level of pellet are comparable to the same level of EB-X-K,Kb. For instance, Pl bands are close to those for 0.5 pg E2X-K,Kh and P2 bands are slightly above those for 0.2 pg

PYRUVATE EZ-X-K,Kb

E2- X1

E2

DEHYDROGENASE olig

FIG. 2. SDS-PAGE analysis of the binding of kinase and protein X to E2-X1 subcomplex and to protein X-depleted E2 oligomer (E2*). E2Xi subcomplex (20 pg) or E2 oligomer (20 pg) was incubated with or without 10 pg of X, K,Kb fraction and then fractionated as described under Experimental Procedures. SDS-PAGE was conducted on a set of samples equivalent to that shown in Fig. 1 but only pellet fractions at a similar protein level are shown here. From left to right, the first two lanes contain the indicated levels of untreated ES-X-KKi,. The remaining lanes also show the patterns for the indicated levels of protein from pellet fractions incubated with (+) or without (-) the X, K,Kb fraction.

of EB-X-K,K,,. This suggests lack of change in the Xto-E2 ratio, indicating little or no additional protein X was bound by the subcomplex. The protein X of this X, K,Ki, fraction bound 0.35 pg of E3 per microgram of X, K,Kb in a microplate binding assay (7). This estimate was based on an E3-specific activity of 210 pmol min-l mg-’ at 24°C. The lipoyl domain of resolved protein X was a substrate for reductive acetylation (16) catalyzed by the El component. The data in Fig. 2 allow us to distinguish whether the failure to significantly increase the level of protein X bound to the subcomplex reflected saturation of protein X binding sites or a lack of a reversible association of protein X. The experiment performed was similar to that in Fig. 1 with binding of protein X from the X, K,K,, fraction to E2 oligomer (depleted in protein X) or to the

KINASE

BINDING

SITE

501

EB-Xi--K,K,, subcomplex (lipoyl domain of protein X removed). While a set of data fully equivalent to Fig. 1 was generated, for simplicity sake, Fig. 2 presents only a single level of each pellet fraction and fractions of control E2X-K,Ki, at comparable levels. The E2 oligomer, depleted in protein X, failed to bind protein X and the E2-XiK,Kb did not bind protein X nor did the Xi domain exchange with free protein X. Thus, both protein X and its inner domain fragment bind very tightly to the assembled E2. It seems most likely that the binding of protein X occurs during the assembly of E2 subunits and that once protein X is integrated into the assembled core it does not readily dissociate. However, the possibility that the E2-binding region of resolved X (i.e., X’s inner domain (3)) has been selectively damaged by the resolution procedure cannot be eliminated (cf. Discussion). Binding of the kinase to the E2L fragment. Collagenase treatment of EB-X-K,Kb subcomplex releases an E2L fragment and the kinase subunits (8). Several approaches were used to evaluate whether the kinase remained bound to the E2L fragment. Gel filtration chromatography of this fraction through a 28-cm Sephacryl S-300 column gave a broad E2L peak with the kinase eluting at the front of the peak, but the E2L level exceeded the level of kinase (lane 1, Fig. 3). A portion of the E2L eluted after the kinase (lane 2, Fig. 3). Less than 20% of the kinase activity was recovered. There was a loss of kinase in proportion to the length of the column making the use of longer Sephacryl columns undesirable. This problem was even worse when Sephadex or Sepharose were employed. There was essentially complete loss of kinase when a Toyosoda TSK 3000 column was used with a variety of buffers. Even with Sephacryl columns, we have not been able to recover the majority of the kinase by inclusion of glycerol, salts, or pluronic F68 (most effective) in the elution buffers. That judgment was based on both kinase activity and the relative intensity of the stained kinase band to other bands following separation by SDS-PAGE (i.e., other light contaminants [e.g., E3 band] in E2i,-kinase sample were enriched relative to K, in the eluted fractions). The incomplete separation of E2L from a putative E2L-kinase complex may have resulted, in part, from aggregation of a portion of the E2L (cf. Discussion). The collagenase-generated E2i, fragment sediments very slowly in analytical ultracentrifugation experiments. This is due to a very high frictional coefficient as was demonstrated in studies on a somewhat smaller trypticgenerated E2, fragment (9). The kinase-containing E2L fraction was sedimented at 35,000 rpm in SW 5O/rotor for 18 h into a 5-20% linear sucrose gradient. The majority of E2L fragment remained at the top of the gradient (cf. sample lane 7, Fig. 3), whereas a portion of the E2L sedimented more rapidly in a fraction that contained the kinase subunits. Lane 3, Fig. 3, shows an example of such a preparation made from an EBi,-kinase fraction in which

502

LI ET AL.

FIG. 3. Binding of the kinase to the E2,, fragment. SDS-PAGE patterns of silver-stained gels are shown for samples given the following treatments. Gel filtration chromatography on 0.7 X 28-cm Sephacryl S300 columns was conducted on 67 pg of EZL-I&Kb using 50 mM Naphosphate buffer (pH 7.0), 0.2 mM EDTA, 0.1 mM dithiothreitol with 0.45-ml fractions collected Lanes 1 and 2 show patterns for ZO-~1samples from fractions 16 and 19. The upper bands in lane 2 are from Clastridium listolyticum collagenase used to prepare the E2L fragment. Lane 3 shows the pattern for 0.5-ag sample of E2,-K,Kb (low Kb sample) prepared by the sucrose gradient procedure described in the text. A 7.5~rg sample of the fraction in lane 3 was mixed with 60 ~1 of Avid-Gel AX containing 0.4 mg/ml McAb 355.2 and incubated for 10 min with vortexing at 60s intervals. The gel was then pelleted repeatedly during six washes with 100 ~1 buffer (same as in gel filtration above but also containing 0.5 mg/ ml Pluronics F-68). Lane 4 shows SDS-PAGE pattern for lo-r1 sample of this gel. Lanes 5 and 6 show SDS-PAGE patterns following application of 45 ~1 of 0.22 mg/ml EPL-K,Kb fraction (prepared without gradient enrichment (8)) applied to 50 ~1 of Avid-Gel AX-McAb 355.2 or of similarly treated Avid-Gel AX lacking the McAb. The gels were washed six times as described above and 10-&l samples used. Lanes 7 show E2L (0.15 pg) sample prepared free of kinase by the gradient procedure and lanes 8 and 9 show binding of this E2t, to Avid-Gel AX-McAb 355.2 and lack of binding by Avid-Gel AX-McAb-95.1 (which binds FADcontaining subunit of the pyruvate dehydrogenase phosphatase). In the latter experiments 2 pg of E2L was applied to 40 ~1 of each gel. Other conditions were as described above. The heavy chains of McAbs are designated by an H (-90% of the heavy chains remain bound to the gel) and the light chain by an L in the figure.

K, B Kb (lane 1, Fig. 3). While it seems probable that most if not all of the E2r. fragment moving with the K, subunit is bound to it, binding was evaluated further with this gradient-enriched EBL-kinase fraction using an immunoaffinity approach. A monoclonal antibody (355.2), which is specific for the inner more COOH-terminal end of the inner lipoyl domain of the E2L fragment (24), was covalently attached to Avid-Gel AX (cf. Experimental Procedures). The E2Lkinase fraction was passed through samples of this gel matrix containing or lacking McAb 355.2 and then the gel was extensively washed with 50 InM Na-phosphate buffer (pH 7.3) containing 1 mg/ml Pluronic F-68. Lane

4 (Fig. 3) shows the capture of the E2L fraction (lane 3) that was enriched in kinase (and had K, > K,,) by the sucrose gradient procedure (above). There was full retention of kinase along with the E2L fragment. Lane 5 shows the retention by this anchored McAb of a typical E2LK,Kb fraction in which Kb _NK,. However, the control experiment, with Avid-Gel AX lacking the McAb, demonstrated nonspecific binding of a much lower level of kinase and an equivalent level of E2L (lane 6, Fig. 3). E2L prepared free of kinase by the gradient procedure above (lane 7) was bound by McAb 355.2 (lane 8) but gave no nonspecific binding to Avid-Gel AX lacking the McAb or containing a nonspecific McAb (lane 9). Thus the nonspecific retention of E2L in lane 6 apparently resulted from its specific association with nonspecifically bound kinase. Extensive washing of McAb bound EBL-kinase did not detectably reduce the level of bound kinase, indicating the association of the kinase and more specifically K, subunit to E2L is very tight. We obtained similar results when matrices containing protein G were used to precipitate McAb bound E2L. However, nonspecific binding of the kinase was worse with that system. DISCUSSION Upon removal of the El and E3 components from the bovine kidney complex, the pyruvate dehydrogenase kinase and protein X components remain avidly bound to the large oligomeric core formed by association of E2 subunits. Understanding the nature of the tight binding of the kinase constitutes an essential step toward elucidating how the activity of a kinase molecule is raised from a turnover rate of about 6 to about 30 min-l upon its binding to the core assemblage. We have addressed a variety of questions (below) concerning subunit and domain roles in kinase binding and activated kinase function. Protein X has a fundamental role in the organization of the complex since it binds the E3 component. Only protein X binds to the self-assembling inner domain of E2 subunits. The questions arises whether binding of protein X, unlike its dissociation, readily occurs with an already assembled core. That was pursued here with a free protein X fraction shown to retain other functions. The kinase studies will be considered first. Previously, we found that the E2-X subcomplex could enhance the activity of about 15 molecules of kinase from an X, K,K,, fraction and that tryptic cleavage of protein X and E2 subunits led to release of the kinase (18). At the time, we could not distinguish whether the kinase was bound by protein X or a specific site in the outer domain of the transacetylase (18). Subsequent studies (3,8) found that the kinase was released with production of the E2L fragment demonstrating E2 binds the kinase. Upon finding mammalian protein X had a role in the binding of the E3 component (4, 6-8), a secondary role in kinase binding seemed unlikely but was not eliminated. The

PYRUVATE

DEHYDROGENASE

present work indicates tight kinase binding does not involve protein X. Here, we have further addressed the following questions concerning the binding of the catalytic subunit of the kinase (IQ. Does the Kb subunit contribute to E2 binding of K, or to the activated state of bound kinase? Is there a capacity for binding many more molecules of kinase than the two to three associated with the purified bovine complex? Does the kinase remain bound to the free bilipoyl domain-containing E2L fragment? Regarding a role for Kb, we had found that cleavage of Kb by protease Arg C did not release the kinase activity from its association with intact E2 subunits (3). However, this did not eliminate the possibility that a large fragment of Kb, present after cleavage by protease Arg C, was still facilitating this binding of K, to E2 or contributing to the activated state of bound kinase. In removing the kinase from the E2-X-K,Ks subcomplex we occasionally prepare X, K,Kb fractions in which K, 9 Kb. Using such a preparation, we have found that about 18 K, subunits can bind to the E2-X subcomplex with a much lower level of bound Kb subunits. Thus Kb is not required for binding of K, to E2 or for the large activation of the kinase. The finding that at least 15 molecules of kinase can bind to the oligomeric core suggests that there are equivalent kinase binding sites on a large number if not all E2 subunits. Thus, the low number of tightly bound kinase molecules associated with the purified complex probably reflects the relative level of synthesis of this regulatory enzyme. There are several possible explanations for the capacity of the ES-X-K,Kb subcomplex (containing initially 60 E2 and -2 K, per molecule of subcomplex) to maximally bind about 18 K,. Possible explanations include negative cooperativity or an association of K, subunit with more than one lipoyl domain region. The fact that, at lower levels of this same X, K,Kb fraction, virtually all the kinase was bound (29) indicates the limited binding did not reflect a problem with the resolved kinase. If some posttranslational modification of an E2 subunit is required for binding the kinase, then a substantial portion of the E2 subunits have undergone the required modification. Further studies are needed to elucidate the nature of the kinase binding site. Studies with recombinant lipoyl domains should facilitate this analysis (30). Since the kinase was released with the E2L fragment (8), the likely prospect was that the kinase remained bound to this region of E2. Since the E2L fragment has a high frictional coefficient (9), separation of EBL-kinase fraction from bulk E2L was likely to be more successful by gradient centrifugation than by gel filtration chromatography but both were attempted. In both cases the kinase eluted with the E2L fragment but the 1eveI of E2L exceeded the level of kinase in all gel filtration experiments and in some gradient separations. Particularly in the case of the gradient separation, portions of E2L bad a higher rate of sedimentation than expected suggesting the presence of E2L aggregates. That is consistent with

KINASE

BINDING

SITE

503

our studies using quasielastic light scattering which detected aggregates in the E2L fraction that could be removed by gel filtration chromatography.8 Using McAb 355.2, which is specific for the inner of the two lipoyl domain, we selectively captured E2L with this McAb anchored through the carbohydrate moieties of its F, region. Binding of E2L resulted in the retention of the kinase. These studies were complicated by some nonspecific binding of the kinase which also leads to nonspecific binding of E2L. However, E2L, free of kinase, did not undergo nonspecific binding. With the EBL-kinase anchored by the EBL-specific McAb, extensive washing (data not shown) did not appear to reduce the ratio of kinase to E2L indicating tight binding of the bovine kinase to E2L fragment. In continuing work, Ono et al.’ have found that the kinase bound to E2L fragment is rapidly activated (in 10 s) by addition of E2-X subcomplex but studies using physical separation technique do not show rapid transfer of the kinase from E2L to E2-X in the presence or absence of ATP. We are attempting to explain this apparent conflict. Activation is not achieved with E2iB-X subcomplex produced by collagenase removal of E2L fragment (8). This E2iB-X fully binds the El component. Thus the activation does not result simply from binding El to E2. Apparently the transfer of kinase from E2L to intact E2 or the capacity of kinase bound to E2i> to interact with intact E2 seems to be required. Since protein X was first characterized as a new component of the mammalian pyruvate dehydrogenase complex (12, 31), the nature of its tight association with the E2 component has not been understood. The lack of binding of protein X to the assembled E2 oligomer either results because protein X must bind during assembly or protein X was damaged during the reaction of E2-X-kinase subcomplex with the mercurial agent used in its release. That preparation procedure involves reacting reduced lipoyl moieties with p-hydroxymercuriphenyl sulfonate and, after isolation of the X, K,K,, fraction, exhaustive dialysis with dithiothreitol and EDTA to remove the mercurial agent (16). The protein X, which constituted less than 30% of the protein of this fraction, bound E3 efficiently (cf. Results). Thus, protein X was native in the region of its structure that binds E3. Its lipoyl domain was also in a native state since it served as an efficient substrate for El-catalyzed reductive acetylation reaction (16). That reaction requires structural integrity of the lipoyl domain and does not occur with noncognate lipoyl domains (e.g., 32). We could not establish whether the inner domain of protein X, which binds to E2’s inner domain (3), was in a native state since there is no other ‘T. E. Roche, S. L. Powers-Greenwood, W. F. Shi, W. B. Zhang, S. Z. Ren, E. D. Roche, D. J. Cox, and C. M. Sorensen, submitted for publication. ’ K. Ono, G. A. Radke, and T. E. Roche, unpublished observation.

504

LI ET AL.

known function to be evaluated. It is very interesting that coexpression of yeast E2 and yeast protein X under conditions that led to excess production of protein X did not result in an increased level of incorporation of X into yeast complex (14). Lawson et al. (14) did report that mixing of dilute E2 with crude fractions containing protein X led to reconstitution of some PDC activity. That clearly supports binding of a few protein X molecules, but the possibility of some partial assembly or disassembly of dilute E2 under these conditions was not eliminated. The basis for the binding of only about 6 protein X to 60 subunits of E2 remains an interesting problem since the lowest symmetry element in the dodecahedron structure formed by assembled E2 is 12 faces. We have found that the catalytic subunit of the kinase is specifically and tightly bound to the lipoyl domain region of E2 subunits. Since the activity of only a couple molecules of kinase can rapidly inactivate the bovine kidney complex, it is not surprising that about this level of kinase is found in the isolated complex. The finding of an expanded capacity to bind the kinase suggests that low number results from control in the production of kinase rather than an inherent limitation in the number of binding sites. The structural feature that limits the assembled E2 oligomer to binding only one kinase per three subunits needs to be determined and may give new insights into the nature of kinase binding and function. As in the case of protein X binding which appears to be limited to five to six per core, these results suggest some form of existing or induced asymmetry in the assembled core. ACKNOWLEDGMENTS We thank Connie Schmidt for help in preparation of the manuscript. This work was supported by National Institutes of Health Grant DK8320 and by Kansas State Agriculture Experiment Station Contribution 92298J.

REFERENCES 1. Linn, T. C., Pelley, J. W., Pettit, F. H., Hucho, F., Randall, D. D., and Reed, L. S. (1972) Arch. Biochem. Biophys. 148,327-342. 2. Pettit, F. H., Rocbe, T. E., and Reed, L. J. (1972) Bioc&rrz. Biophys. Res. Commun. 49, 563-571. 3. Rahmatullah, M., Gopalakrishnan, S., Radke, G. A., and Roche, T. E. (1989) J. Biol. Chem. 264, 12451251. 4. Rahmatullah, M., Gopalakrishnan, S., Andrews, P. C., Chang, C. L., Radke, G. A., and Roche, T. E. (1989) J. Biol. Chem. 264, 2221-2227. 5. Rahmatullah, 8110.

M., and Roche, T. E. (1988) J. Bial. &em. 263,8106-

6. Powers-Greenwood, S. L., Rahmatullah, M., Radke, G. A., and Roche, T. E. (1989) J. Biol. Chem. 264, 3655-3657. 7. Gopalakrishnan, S., Rahmatullah, M., Radke, G. A., PowersGreenwood, S. L., and Roche, T. E. (1989) B&hem. Biophys. Res. Commun. 160,715-721. 8. Rahmatullah, M., Radke, G. A., Powers-Greenwood, S. L., Andrews, P. C., and Roche, T. E. (1990) J. Biol. Chcm. 265, 14,512-14,517. 9. Bleile, D. M., Hackert, M. L., Pettit, F. H., and Reed, L. J. (1981) J. Biol. Chem. 256, 514-519. 10. Coppel, R. L., McNeilage, L. J., Surh, C. D., Van de Water, J., Spithill, T. W., Whittingham, S., and Gershwin, M. E. (1988) Z+OC. Natl. Acad. Sci. USA 85, 7317-7321. T. J., Ho, L., Wexler, I. D., Pons, G., Liu, T. C., and 11. Thekkumkara, Patel, M. S. (1988) FEBS Lett. 240, 45-58. 12. Jilka, J. M., Rahmatullah, M., Kazemi, M., and Roche, T. E. (1986) J. Biol. Chem. 261,1858-1867. 13. Nu, Y-D., Stoops, J. K., and Reed, L. J. (1990) Biochemistry 29, 8614-8619. 14. Lawson, J. E., Behal, R. H., and Reed, L. J. (1991) Biochemistry 30,2834-2839. 15. Stepp, L. R., Pettit, F. H., Yeaman, S. J., and Reed, L. J. (1983) J. Biol. Chem. 258,9454-9458. M., and Roche, T. E. (1987) J. Biol. Chem. 262, 16. Rahmatullah, 10,265-10,271. 17. Hucho, F., Randall, D. D., Roche, T. E., Burgett, M. W., Pelley, J. W., and Reed, L. J. (1972) Arch. Biochem. Biophys. 157, 328340. 18. Rahmatullah, M., Jilka, J. M., Radke, G. A., and Roche, T. E. (1986) J. Biol. Chem. 261,6515-6523. 19. Laemmli, U. K. (1970) Nature 227,680-685. 20. Oakley, B. R., Kirsch, D. R., and Morris, N. R. (1980) Anal. Biochem.

105,361-353. M., and Roche, T. E. (1985) J. Biol. Chem. 260, 21. Rahmatullah, 10,146-10,152. 22. Pratt, M. L., and Roche, T. E. (1979) J. Biol. Chem. 254, 71917196. P. J., Tsai, C. S., Eley, M. H., Roche, T. E., and Reed, 23. Butterworth, L. J. (1975) J. Biol. Chem. 250, 1921-1925. A., and Gershwin, M. E. (1990) J. Zm24. Surh, C. D., Ahmed-Ansari, munol. 144,2647-2652. 25. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Sidman, J. G., Smith, J. A., and Struhl, K. (1988) Current Protocols in Immunology, Suppl. 1, Chap. II, Breene/Wiley-Interscience, New York. 26. Nau, D. R. (1989) Biochromatography 4,4-18. 27. Cress, M. C., and Ngo, T. T. (1989) Am. Biotechnol. Lab. 7, 16-19. 28. Patek M., and Roche, T. E. (1990) FASEB J. 4, 3224-3233. 29. Li, L. (1988) M. S. thesis, Kansas State University, Manhattan, KS. 30. Liu, S., and Roche, T. E. (1992) FASEB J. 6, A460. 31. DeMarcucci, D. L., and Lindsay, J. G. (1985) Eur. J. Biochem. 149, 641-648. 32. Perham, R. N., and Packman, L. C. (1989) Ann. Reu. N. Y. Acad. Sci. 573, l-20. 33. Fried, V. A., Ando, M. E., and Bell, A. J. (1985) Anal. Biochem.

146,271-276.