The remarkable structural and functional organization of the eukaryotic pyruvate dehydrogenase complexes

December 18, 2001
98 (26) 14802-14807


The three-dimensional reconstruction of the bovine kidney pyruvate dehydrogenase complex (Mr ≈ 7.8 × 106) comprising about 22 molecules of pyruvate dehydrogenase (E1) and about 6 molecules of dihydrolipoamide dehydrogenase (E3) with its binding protein associated with the 60-subunit dihydrolipoamide acetyltransferase (E2) core provides considerable insight into the structural and functional organization of the largest multienzyme complex known. The structure shows that potentially 60 centers for acetyl-CoA synthesis are organized in sets of three at each of the 20 vertices of the pentagonal dodecahedral core. These centers consist of three E1 molecules bound to one E2 trimer adjacent to an E3 molecule in each of 12 pentagonal openings. The E1 components are anchored to the E1-binding domain of the E2 subunits through an ≈50-Å-long linker. Three of these linkers emanate from the outside edges of the triangular base of the E2 trimer and form a cage around its base that may shelter the lipoyl domains and the E1 and E2 active sites. The docking of the atomic structures of E1 and the E1 binding and lipoyl domains of E2 in the electron microscopy map gives a good fit and indicates that the E1 active site is ≈95 Å above the base of the trimer. We propose that the lipoyl domains and its tether (swinging arm) rotate about the E1-binding domain of E2, which is centrally located 45–50 Å from the E1, E2, and E3 active sites, and that the highly flexible breathing core augments the transfer of intermediates between active sites.
The pyruvate dehydrogenase complex (PDC) serves as the link between glycolysis and the tricarboxylic acid cycle and generally has prominence in the description of these metabolic pathways because they serve as a major source of cellular energy. A central feature of PDCs is a 24-mer (Escherichia coli) or 60-mer (eukaryotes and some Gram-positive bacteria) core with the morphologies of a cube or pentagonal dodecahedron, respectively (14). The structures with the latter morphology comprise the largest (Mr ≈ 107) multienzyme complexes known. Even more remarkable than their exceptional size and morphology, these complexes encompass some of the most unusual features found in structural biology as described below.
The E2 core comprises the dihydrolipoamide acetyltransferase activity and is the only oligomeric enzyme complex known to be organized with the shape of a pentagonal dodecahedron. Moreover, the 250-Å-diameter dodecahedron has a very unusual feature: the tightly bound trimers at each of its 20 vertices seem to be interconnected by 30 flexible bridges enabling the core to “breathe,” as evidenced by an extraordinary size variability of 40 Å (17%) at room temperature. The breathing core apparently is a common feature in the phylogeny of the PDCs, suggesting that protein dynamics is an integral component of the function of these multienzyme complexes (5). Moreover, dodecahedral morphology of the core favors a synchronous or harmonious change in the length of the bridges that is related to the size variation and the function of the complex (5).
The E2 subunits are multidomain structures consisting of one, two, or three amino-terminal lipoyl domains, followed by a pyruvate dehydrogenase (E1) and/or dihydrolipoamide dehydrogenase (E3) binding domain, and then by catalytic and self-association domains (1–4; see Fig. 2C). Thus, the unique dodecahedral E2 core also serves as a scaffold about which the other components are organized. The yeast and mammalian PDCs require a binding protein (BP) to anchor E3 to the core, although in E. coli and Bacillus stearothermophilus PDCs, BP is not required.
The deposition of the BP⋅E3 is another unusual structural feature of the organization of the eukaryotic PDCs. We determined that the 60-subunit scaffold (inner core) of Saccharomyces cerevisiae PDC binds only 12 BP⋅E3 (6), and our structural studies revealed a paradigm for this mode of binding (7). A BP⋅E3 component is bound inside each of the 12 pentagonal openings of the scaffold so that the stoichiometry and arrangement of this component is determined by the geometric constraints of the underlying scaffold and not by the number of potential binding sites. Even though the E2 subunits are arranged with icosahedral symmetry, as a consequence of this mode of binding, the BP⋅E3 components are not (7).
Biochemical and structural studies have shown that the lipoyl and E1-binding domains comprising the N-terminal region of E2 are attached to the C-terminal catalytic and self-association domains by flexible linkers (14, 8). Consequently, the three-dimensional (3D) reconstructions of S. cerevisiae E2 and truncated E2 (tE2) (lacking the lipoyl and E1-binding domains) have very similar structures (5). Thus, residues 1 to ≈220 comprise the flexible part of the E2 subunit and, consequently, are not seen in the reconstruction, whereas residues ≈221–454 form the icosahedrally ordered scaffold. Because E1 is attached to the flexible N-terminal region of the E2 subunits, this component has been predicted to be poorly organized about the core, as depicted in a model for the B. stearothermophilus PDC (9). However, the present 3D structural study of bovine kidney PDC shows that the E1 components are remarkably well organized with icosahedral symmetry about the E2 scaffold.
Even though atomic resolution structures of most of the components of the PDC have been determined, our knowledge of their organization in this macromolecular complex is incomplete. This study, together with structural studies of the S. cerevisiae PDC and its subcomplexes, explains how the E1 components are arranged and provides a structural foundation for the functional organization of this most remarkably arranged multienzyme complex.

Experimental Procedures

Enzyme Preparation.

The bovine kidney PDC was purified to near homogeneity as described (10). Because variation in the number of E1 and BP⋅E3 components bound to the E2 scaffold may affect their deposition and spatial stability (see below) we characterized the bovine kidney PDC preparation used in this EM study with regard to these components. Its weight-average molecular weight was determined by light-scattering measurements to be about 7.8 × 106. On the basis of the known molecular weight of the complex and its component enzymes and the experimentally determined polypeptide chain ratios of E2:BP:E3 (6), we estimated that the subunit composition of the bovine kidney PDC is about 22 E1 tetramers, 60 E2 monomers, 12 BP monomers, and 6 E3 dimers.

Cryo-Electron Microscopy.

A 3-μl sample of the preparation (0.35 mg/ml containing 20 μg/ml of bacitracin) was deposited, blotted, and quick-frozen in liquid ethane on a glow-discharged carbon-coated holey grid. The vitrified samples were recorded at ≈1 μm under focus at ≈10 e/Å2 dose for image processing. A second exposure of ≈2–3 μm under focus was recorded for display (see Fig. 1) and as an aid in analyzing the images. The images were recorded on Kodak SO 163 film at a nominal magnification (×50,000) in a JEOL JEM 1200 electron microscope operated at 100 kV.
Figure 1
CryoEM images. Fields of images of frozen-hydrated bovine PDC (A) and S. cerevisiae tE2 (residues 221–454; B) and E2 (C). The E1 molecules appear equally distributed about the scaffold, and their association with its outside increases the diameter from ≈250 to ≈500 Å. The similarity between the tE2 and E2 images indicates that N-terminal domains (residues 1 to ≈220) are flexible.

Data Processing.

Micrographs were digitized with a Zeiss SCAI microdensitometer (Z/I Imaging, Huntsville, AL), using a step size of 2.8 Å per pixel at the specimen scale, and subsequently averaged to 5.6 Å per pixel for further processing. Data processing of the focal pair micrographs of PDC was performed on SGI Octane dual processor workstations with modular programs of the IMIRS program package (Y. Liang, E. Ke, and Z.H.Z., unpublished data; refs. 11 and 12). The defocus values of all micrographs were determined from the incoherently averaged Fourier transforms of particle images in each micrograph (13).
We used the focal pair method of orientation determination, refinement, and 3D reconstruction (14, 15) except that an additional step of particle-size evaluation was performed in the current reconstruction as described (5). The effective resolutions of the 3D maps were assessed by calculating the Fourier shell cross-correlation coefficients between two independent reconstructions. To evaluate the resolution differences between the core and E1 outer shell, the density spheres corresponding to the core and the E1 shell were extracted before the Fourier shell correlation calculation. Although the resolution of the core extended beyond 20 Å, the resolution of E1 outer shell was limited to about 35 Å. Therefore, the final reconstruction was calculated by merging 162 particle images with defocus values between 1 and 3 μm and data up to 30-Å resolution. The contrast transfer function of the microscope was corrected by using an amplitude contrast of 14% and an envelope-decaying factor of 1000 Å2 as described (16).

3D Visualization and Fitting of Atomic Structures.

The 3D visualization was performed by using the IRIS EXPLORER (NAG, Downers Grove, IL) with custom designed modules (17). The maps were displayed at the same threshold by using a contour level such that the volume of the E1 tetramer matches the calculated volume based on the molecular mass, which corresponds to about 0.5 σ (standard deviation) above the mean density of the map.
The atomic coordinates of the crystal structures of the dodecahedral core of B. stearothermophilus tE2 (Protein Data Bank ID code 1B5S; ref. 18) and the Pseudomonas putida tetrameric E1 (Protein Data Bank ID code 1QS0; ref. 9), the NMR solution structures of the E2 lipoyl domain of B. stearothermophilus (Protein Data Bank ID code 1LAC; 19), and the E3-binding domain of E. coli E2 (Protein Data Bank ID code 1BBL; i.e., the putative E1-binding domain of E2; ref. 20) were downloaded directly from the Protein Data Bank. The areas of interest of the structures were rendered and exported to Open Inventor format by using either RIBBONS (21) or WEBLAB VIEWERPRO (Accelryes, San Diego, CA), or directly converted to an electron density map by using a Gaussian filter to a resolution similar to the EM structures (30 Å).


Cryo-Electron Microscopy.

The 22 E1 tetramers bound to E2 completely obscure the underlying scaffold and increase the apparent diameter of the molecule from ≈250 to ≈500 Å (Fig. 1). The images of tE2 (residues 221–454) and E2 of S. cerevisiae core are displayed for comparison (Fig. 1 B and C). They clearly show the characteristic views of the pentagonal dodecahedron-shaped core to which the E1 components are bound. The similarity between the tE2 and E2 images show that the N-terminal half of the subunit is extended and presumably flexible (5) and, consequently, is not seen in the images of the molecules. Similar cryo-electron microscopy (cryoEM) images were reported for the bovine kidney and heart PDCs and the E2⋅BP core (22). In both of these studies the E1 components seem equally distributed about the central core because no apparent discontinuity exists in the E1 distribution around the periphery of the images.

3D Reconstruction.

Considerable effort was made to reveal the connections (linkers) that bind E1 molecules to the E2 core because this structure would be expected to provide information that is fundamental to the appropriate docking of the E1 x-ray structure in the EM envelope (see below; Fig. 2). Initially, all our reconstructions of the S. cerevisiae and the bovine kidney PDCs displayed a shell of protein density surrounding the underlying core without revealing the associated linkers. The utilization of size-variation analysis to classify the images according to size (5) and the utilization of closer-to-focus images was an essential step in our effort to align the images with enough precision to reveal the connections (see Experimental Procedures). The inner linker between the E2 core and the E1-binding domain consists of ≈40 aa residues (2325). NMR studies of the inner linker of E. coli E2 indicate that it may be more constrained than the linker that joins the lipoyl domain to the E1-binding domain (outer linker; ref. 8; Fig. 2C). Because the inner linker is ≈50 Å in length (Fig. 2B) and a fully extended polypeptide chain of 40 residues is ≈140 Å (≈3.5 Å per residue), we predict this linker has significant secondary structure, which may be related to its purported stiffness (8). A portion of the inner linkers closest to the E2 trimer base are apparent in the 3D reconstruction of the human E2 core and are in the same position as seen in Fig. 3A, supporting the view that the inner linker is partially ordered (X. Yu, T. Roche, J.K.S., and Z.H.Z., unpublished data). In any event, it is apparent from this structure (Fig. 2B) that the display of the inner linker is related to the bound E1, because reconstructions of yeast (5) and human (X. Yu, T. Roche, J.K.S., and Z.H.Z., unpublished data) E2 do not reveal the extended linker seen in this structure.
Figure 2
3D reconstruction of bovine PDC (A and B) and diagrammatic representation of the structural domains of E2 subunit (C). Shaded-surface representation of 3-fold axes of symmetry of the 3D reconstruction of the bovine kidney PDC (A) and with the closest half removed to reveal the linker (blue) that binds E1 (yellow) to the E2 core (green; B). The inner linker is ≈50 Å in length. (C) The C-terminal self-association domain is responsible for the assembly of the dodecahedral scaffold to which E1 and BP⋅E3 bind. The N-terminal half of the E2 comprises the L1 and L2 lipoyl domains, and the E1-binding domain, and their associated linkers. The inner linker is revealed in the 3D structure of the PDC.
Figure 3
Shaded-surface representations of 3-fold view of E2 core. (A) E1 was removed from PDC to reveal the underlying E2 core (green) and its inner linkers (light blue). An E2 trimer is outlined by the yellow square, and the 5-, 3-, and 2-fold axes are indicated. The linkers change the shape of the trimer base so that it appears that they are connected to its corners. (B) Superposition of the cut-away cryoEM PDC structure and the x-ray structure of B. stearothermophilus tE2. The atomic model of the tE2 dodecahedral core was directly superimposed on the core of the cryoEM structure by aligning the icosahedral 5-, 3-, and 2-fold axes with those of the cryoEM structure. For clarity, only one of the 20 tE2 trimers of the atomic model is shown. The dark and light blue and green x-ray ribbon structure represent the three identical subunits that comprise the tE2 core. The three N-terminal linkers are directly opposite the N-terminal helix (H1) near the middle of the outside edge of the trimer base as indicated, and the N-terminal loop is directed toward the 3-fold axis. The bridge interconnects adjacent trimers to form the dodecahedral core.

Structural Organization of PDC.

The threshold (contour level of 0.5 σ, or standard deviation above the mean) of the reconstruction was set so that the EM envelope of E1 accommodates the x-ray structure of 2-oxoisovalerate dehydrogenase from P. putida (9). At this threshold the averaged protein density of E1 is approximately one-third that of E2, which agrees with the estimation of the E1 content from physicochemical measurements (22E1/60E2) with this PDC preparation.
The appearance of the E1 linker in the structure and the ≈35-Å resolution corresponding to the E1 shell shows that the flexibility of the E1 and its linker is largely constrained. In contrast, the proposal that the E1 component is not constrained by its string-like tether (9) would result in a random deposition of E1, so that the outer shell of protein would be featureless, if seen at all, in the 3D reconstruction with icosahedral symmetry imposed, and the Fourier shell correlation coefficient between independent reconstructions would show no significant correlation.
In this regard, it seems paradoxical that the E1 components are highly constrained when attached to a largely flexible tether. We propose that the E1 molecules exhibit weak but significant interaction with each other and, consequently, they are grouped together in small clusters above the underlying E2 trimers. This interaction may impart positional stability to the organization of the E1 molecules. The proposed interaction between the E1 molecules would be favored at the high local concentration (about 100 mg/ml) in the putative arrangement of the E1 molecules resulting from their attachment to the inner linkers of E2 core. In this regard, note the significant connections between adjacent E1 molecules on the 2-fold and 5-fold axes of the structure (Fig. 2A), which may be related to possible interactions between adjacent E1 molecules. In addition, a triangular feature seems to connect three E1 molecules ≈70 Å above the E2 trimer on the 3-fold axis. We believe that this component is not part of the E1 protein (see Fig. 4) but may be representative of the 40 unbound N-terminal linkers of E2. Consequently, we are assessing the effect of the number of E1 and BP⋅E3 molecules bound to the E2 core on their deposition in the PDC structure by in vitro assembling recombinantly expressed protein components (7).
Figure 4
Cut-away representation of the structural unit of bovine kidney PDC. (A) Top view of the cut-away structure comprising E2 trimer and its associated three E1 tetramers. (B) Side view of A. The three E1 tetramers, E2 trimer, and its linkers form a cage that encloses the E1 and E2 active sites. (C) Top view of wire-frame representation of the structure in A in which the atomic structures of P. putida E1 (light blue) and the B. stereothermophilus lipoyl domains (orange, denoted by arrows in D) and the putative E1-binding domain (bold red) are docked in the EM envelope of bovine E2. The E1 tetramer was docked in the cryoEM envelope so that the E1-binding domain of E2 and its inner linker is close to the 2-fold axis of E1 where the C-terminal domains of the β subunits meet (9). We propose that the 60 potential centers for acetyl-CoA synthesis are organized in sets of three comprising three E2 subunits and three E1 tetramers. (D) Side view of C in which the B. stearothermophilus tE2 x-ray structure is superimposed as described in Fig. 3. The docked ribbon structure of P. putida E1 (light blue) shows the spatial relationship of its crystallographic 2-fold axis to the 3-fold axis of B. stearothermophilus tE2. The E1 active sites are about 95 Å above the E2 catalytic site on the trimer base.

The Deposition of the Inner Linkers of E2.

The 3-fold view of the bovine kidney PDC showing the complex from which the outer E1 molecules have been removed gives the appearance that the linkers originate from the corners of the triangular-based E2 trimers (Fig. 3A). However, the ribbon presentation of the x-ray structure of B. stearothermophilus tE2 core superimposed in the cryoEM density map of bovine kidney PDC shows that the three linkers are located on the outside edge near the middle of the side of equilateral triangle-shaped base of the tE2 trimer directly opposite the N-terminal helix H1 (ref. 18; Fig. 3B). In the x-ray structure the N-terminal extension from the α-helix, which presumably represents the C-terminal end of the linker, lies across the base of the tE2 subunit and ends near the 3-fold axis of the trimer. In contrast, the corresponding extension of the N-terminal helix of E. coli truncated dihydrolipoamide succinyltransferase is also located near the outside edge of its triangle-shaped base and is directed outward from the base of the trimer as shown in the cryoEM map (Fig. 3). It was proposed that the differences in these two regions might result from crystal-packing forces (ref. 26; see the perspective in ref. 27). In any event, the EM envelope shows that the N-terminal linker departs from the outer edge of the trimer base and is directed toward the E1 complex on the outside of the core (Fig. 3).

Spatial Relationships of E1 Tetramer to the E2 Core.

The cut-away structure of the three pyruvate dehydrogenase tetramers (α2β2, Mr ≈ 154,000) bound to an E2 inner core trimer (Mr ≈ 81,000) shows that these trimers anchor in close proximity three E1 molecules, which are about six times the size of the underlying trimer (Fig. 4 A and B). The size comparison is based on the knowledge that the C-terminal half of the E2 subunits consisting of residues ≈221–454 (Mr ≈ 27,000) self-associate to form the tE2 trimer (2325), whereas the N-terminal half of the E2 comprises the extension with its E1-binding and two lipoyl domains (1–4; Fig. 2C). The x-ray structures of the tE2 trimers of the E. coli succinyltransferase (26), and Azotobacter vinelandii (28), B. stearothermophilus, and Enterococcus faecalis acetyltransferases (18) exhibit very extensive interactions between the adjacent subunits that comprise the trimer. These interactions may have an important role in holding together the much larger E1 molecules, which may be exposed to considerable forces resulting from the fluctuations of the solvent on the outside of the core (Fig. 4 A and B).


Putative Arrangement of E1, the Lipoyl, and E1-Binding Domains of E2.

The ribbon diagram of the x-ray structure of 2-oxoisovalerate dehydrogenase (E1) from P. putida (blue) fits nicely in the EM envelope of the bovine kidney E1 (Fig. 4 C and D). The envelope also accommodates two adjacent copies of the 80-residue lipoyl domain (orange) of B. stearothermophilus E2 (ref. 19; L1 and L2) so that the lysine residue to which the lipoyl moiety is attached is in the funnel-shaped opening leading to the channel of the E1 active site. The occupancy of both E1 active sites by the lipoyl domains and their lipoyl moieties may be favored by the high local concentration of six of these constituents in the vicinity of the two E1 active sites. As proposed previously, the cap over the 3-fold axis of the complex may be representative of either the unbound L1 or L2 domains. The two active sites reside ≈95 Å above the E2 triangular base below. The base of the E2 envelope also accommodates the 35-residue E1-binding domain represented by the E3-binding domain of the E. coli E2 (bold red), which resides ≈50Å above the E2 base. Even though the components generally fit well into the EM envelope of our low-resolution structure, these results obviously do not permit the location of the functional components at the atomic level. The exact orientation of the lipoyl and E1-binding domains are unknown because structures of these components bound to E1 have not been determined. Moreover, a small portion of one of the β-subunits is seen outside the envelope (Fig. 4 C and D), which may be related to small orientation variability of the E1 molecules on the outside of the E2 core. Nevertheless, the EM map provides considerable insight into the structural and functional organization of the complex.

Structure–Function Relationships.

Recall that the central core of the eukaryotic PDCs consists of 60 E2 subunits arranged in sets of three at the 20 vertices of a pentagonal dodecahedron. The trimers are cone-shaped with an outward-directed triangular base and are interconnected by 30 bridges (7). We propose that the E2 trimer building block with its three linked E1 tetramers and the adjacent E3 molecule in the pentagonal opening comprise the functional unit of the PDC (Fig. 5).
Figure 5
Cut-away model of the fully assembled PDC viewed on its 3-fold axis. The E3 homodimer (red) of A. vinelandii x-ray structure (32) filtered to 20Å resolution was docked into the pentagonal opening of the core (green) according to studies of the S. cerevisiae PDC (7). The BP⋅E3 components associated with the core are not revealed in the shaded-surface representation of the bovine kidney PDC at this threshold because only about 6 molecules of E3 are bound. A radial density plot analysis of the complex shows a peak of density inside the core which corresponds to the position of BP⋅E3 peak in S. cerevisiae tE2 (ref. 7; data not shown). The inner linkers (light blue) bind E1 (yellow) to the E2 scaffold (green). The E1-binding site on the E2 inner linker is located ≈50 Å above the scaffold as indicated by “*” and serves as the anchor for the lipoyl domains to pivot. The structure shows that the swinging arm pivots about a position that is ≈50 Å from the E1, E2, and E3 active sites.
Protein-engineering studies of the lipoyl domains in which the lipoyl-lysine residue is replaced with a glutamine residue showed that each of the three lipoyl domains of the E. coli E2 participates in catalysis, and that the activity of the PDC is independent of the position of the mutation in the L1–L3 sequence (8). It was also determined that the lipoyl domain greatly enhances the ability of lipoyl moiety to act as a substrate in the reductive acylation reaction at the active site of E1. The low Km value of 26 μM for the free lipoyl domain indicates that it has appreciable affinity for residues near the active site of E1 (8). Studies of mutations in the lipoyl domains revealed a specificity loop in which several residues in the lipoyl domain confer specificity on the reductive acylation reaction (2931). We have docked the lipoyl domains so that this specificity loop faces the E1 complex near its active site (Fig. 4 C and D).
Our EM map indicates that two lipoyl domains may occupy the two active sites of E1 (Fig. 4 C and D). The structure does not permit a determination whether both sites are occupied simultaneously by L1 and L2 or if one lipoyl domain is equally distributed between the two sites. It is also possible that a lipoyl domain from the E2 tether to which the E1 component is not bound may occupy the active site of E1. In this regard, there are on average one inner linker occupied by the E1 component and two that are not because in our PDC preparation, E1/E2 is about 1:3. As proposed above, the proximity of the lipoyl domains associated with the outer linkers lacking the E2 component to those in which E1 is bound may give rise to the cap over the 3-fold axis of the PDC.

A Model for the Dynamics of the Multifunctional PDC.

The view of an E1 complex with its two active sites occupied by the lipoyl domains (orange) shows that the active site to the left of the 2-fold axis of E1 faces into the cage, whereas the other is directed toward the outside of the structure (Fig. 4 C and D). It is possible that the orientation of the pivot permits equal access to both E1 active sites by the lipoyl domains. Another possibility is that the pivot is arranged in such a manner that the L2 and L1 domains bind in tandem to the inner and outer active sites, respectively. We favor the latter hypothesis, because the bulge in the envelope on the 2-fold axis may be related to density representing the tether connecting L1 and L2 (Fig. 4D). Moreover, if the cage provided by the inner linkers has a beneficial role in excluding large molecules in the cellular milieu that may interfere with the transfer of the acetyl moiety, the E1 active site facing the 3-fold axis of the E2 trimer (Fig. 4D) can best benefit from this arrangement.
The E1-binding domain (Fig. 4D) serves as a pivot or anchoring point for a “swinging arm” comprising the outer linker and the two lipoyl domains (Fig. 2C). This anchor is centrally located ≈50Å from the E1, E2, and E3 active sites. NMR studies show that the outer linker is highly flexible and its ≈60 residues can readily accommodate the distances required for the lipoyl domain to access the E1, E2, and E3 active sites (8). We propose that the swinging arm with its two N-terminal lipoyl domains begins the sequence of reactions at the E1 active site about 95 Å above the base formed by the three E2 subunits. After the reductive acylation reaction the arm swings to the E2 active site(s) on the trimer base below. Whether the three E2 acetyltransferase sites associated with the trimer base are equally accessible to the acetylated dihydrolipoyl moiety is unknown. The three linkers that surround the base of the E2 trimer form a cage that is covered by the accompanying E1 molecules (Fig. 4 A and B). As proposed previously, this cage-like structure serves as a shelter that is important in ensuring the successful transfer of the acetylated dihydrolipoyl moiety to the acetyltransferase site. On completing the acetyltransfer reaction, the lipoyl domain, with its reduced lipoyl moiety, swings over the E3 bone-shaped dimer in the pentagonal channel, formed by five E2 subunits, for the oxidation reaction near its outer end (ref. 7; Fig. 5). The pivot (E1-binding domain) is ≈50 Å from the E3 active site. In this scenario, the E3 active site serves lipoyl domains associated with five swinging arms surrounding the pentagonal opening. Because five equivalent positions for E3 exist in the channel, E3 rocks between them to promote access of the lipoyl moiety to the E3 active site. The large pentagonal channel can accommodate readily the movement of the outer end of the E3 molecule so that it can participate in the oxidation of dihydrolipoyl moieties associated with the five adjacent E2 subunits. The protein dynamics of the breathing core (5) may promote this movement and also augment the ≈95 Å transfer of lipoyl domains between the E1 and E2 active sites.


pyruvate dehydrogenase complex
pyruvate dehydrogenase
dihydrolipamide acetyltransferase
dihydrolipoamide dehydrogenase
truncated dihydrolipoamide acetyltransferase
binding protein
electron microscopy
cryo-electron microscopy


We thank Dr. Austin Riggs and Claire Riggs for determining the molecular mass of the bovine kidney PDC, Dr. Marvin Hackert and Dr. C. S. Raman for helpful discussion, and Imani Muhammad for secretarial support. This work was supported in part by the American Heart Association, Texas Affiliate Grant 98BG288 (to Z.H.Z.) and the Pew Charitable Trusts (to Z.H.Z.), U.S. Public Service Grants HL42886 (to J.K.S.) and GM06590 (to L.J.R.), and a grant from the Foundation for Research (to L.J.R.).


L J Reed, M L Hackert J Biol Chem 265, 8971–8974 (1990).
M S Patel, T E Roche FASEB J 4, 3224–3233 (1990).
J R Guest, S J Angier, G C Russell Ann NY Acad Sci 573, 76–99 (1989).
R N Perham Annu Rev Biochem 69, 961–1004 (2000).
Z H Zhou, W Liao, R H Cheng, J E Lawson, D B McCarthy, L J Reed, J K Stoops J Biol Chem 276, 21704–21713 (2001).
C Y Maeng, M A Yazdi, X D Niu, H Y Lee, L J Reed Biochemistry 33, 13801–13807 (1994).
J K Stoops, R H Cheng, M A Yazdi, C Y Maeng, J P Schroeter, U Klueppelberg, S J Kolodziej, T S Baker, L J Reed J Biol Chem 272, 5757–5764 (1997).
R N Perham Biochemistry 30, 8501–8512 (1991).
A Aevarsson, K Seger, S Turley, J R Sokatch, W G Hol Nat Struct Biol 6, 785–792 (1999).
F H Pettit, L J Reed Methods Enzymol 89, 376–386 (1982).
Z H Zhou, W Chiu, K Haskell, H Spears, J Jakana, F J Rixon, L R Scott Biophys J 74, 576–588 (1998).
O Johnson, V Govindan, Y Park, Z H Zhou Proceedings of the 4th International Conference on High Performance Computers (IEEE Computer Society Press, Los Alamitos, CA), pp. 517–521 (1997).
Z H Zhou, S Hardt, B Wang, M B Sherman, J Jakana, W Chiu J Struct Biol 116, 216–222 (1996).
S D Fuller Cell 48, 923–934 (1987).
Z H Zhou, J He, J Jakana, J D Tatman, F J Rixon, W Chiu Nat Struct Biol 2, 1026–1030 (1995).
Z H Zhou, D H Chen, J Jakana, F J Rixon, W Chiu J Virol 73, 3210–3218 (1999).
Z H Zhou, M Dougherty, J Jakana, J He, F J Rixon, W Chiu Science 288, 877–880 (2000).
T Izard, A Aevarsson, M D Allen, A H Westphal, R N Perham, A de Kok, W G Hol Proc Natl Acad Sci USA 96, 1240–1245 (1999).
F Dardel, A L Davis, E D Laue, R N Perham J Mol Biol 229, 1037–1048 (1993).
M A Robien, G M Clore, J G Omichinski, R N Perham, E Appella, K Sakaguchi, A M Gronenborn Biochemistry 31, 3463–3471 (1992).
M Carson J Appl Crystallogr 24, 958–961 (1991).
T Wagenknecht, R Grassucci, G A Radke, T E Roche J Biol Chem 266, 24650–24656 (1991).
J E Lawson, R H Behal, L J Reed Biochemistry 30, 2834–2839 (1991).
J E Lawson, X D Niu, L J Reed Biochemistry 30, 11249–11254 (1991).
M Rahmatullah, S Gopalakrishnan, G A Radke, T E Roche J Biol Chem 264, 1245–1251 (1989).
J E Knapp, D T Mitchell, M A Yazdi, S R Ernst, L J Reed, M L Hackert J Mol Biol 280, 655–668 (1998).
D J DeRosier Nature (London) 357, 196–197 (1992).
A Mattevi, G Obmolova, E Schulze, K H Kalk, A H Westphal, A de Kok, W G Hol Science 255, 1544–1550 (1992).
X Gong, T Peng, A Yakhnin, M Zolkiewski, J Quinn, S J Yeaman, T E Roche J Biol Chem 275, 13645–13653 (2000).
T E Roche, J C Baker, X Yan, Y Hiromasa, X Gong, T Peng, J Dong, A Turkan, S A Kasten Prog Nucleic Acid Res Mol Biol 70, 33–75 (2001).
S Liu, X Gong, X Yan, T Peng, J C Baker, L Li, P M Robben, S Ravindran, L A Andersson, A B Cole, T E Roche Arch Biochem Biophys 386, 123–135 (2001).
A J Schierbeek, M B Swarte, B W Dijkstra, G Vriend, R J Read, W G Hol, J Drenth, C Betzel J Mol Biol 206, 365–379 (1989).

Information & Authors


Published in

Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 98 | No. 26
December 18, 2001
PubMed: 11752427


Submission history

Accepted: November 8, 2001
Published online: December 18, 2001
Published in issue: December 18, 2001


We thank Dr. Austin Riggs and Claire Riggs for determining the molecular mass of the bovine kidney PDC, Dr. Marvin Hackert and Dr. C. S. Raman for helpful discussion, and Imani Muhammad for secretarial support. This work was supported in part by the American Heart Association, Texas Affiliate Grant 98BG288 (to Z.H.Z.) and the Pew Charitable Trusts (to Z.H.Z.), U.S. Public Service Grants HL42886 (to J.K.S.) and GM06590 (to L.J.R.), and a grant from the Foundation for Research (to L.J.R.).



Z. Hong Zhou
Department of Pathology and Laboratory Medicine, University of Texas Medical School, Houston, TX 77030; and Department of Chemistry and Biochemistry, University of Texas, Austin, TX 78712
Diane B. McCarthy
Department of Pathology and Laboratory Medicine, University of Texas Medical School, Houston, TX 77030; and Department of Chemistry and Biochemistry, University of Texas, Austin, TX 78712
Catherine M. O'Connor§
Department of Pathology and Laboratory Medicine, University of Texas Medical School, Houston, TX 77030; and Department of Chemistry and Biochemistry, University of Texas, Austin, TX 78712
Lester J. Reed
Department of Pathology and Laboratory Medicine, University of Texas Medical School, Houston, TX 77030; and Department of Chemistry and Biochemistry, University of Texas, Austin, TX 78712
James K. Stoops
Department of Pathology and Laboratory Medicine, University of Texas Medical School, Houston, TX 77030; and Department of Chemistry and Biochemistry, University of Texas, Austin, TX 78712


Present address: Ciphergen Biosystems, Inc., Palo Alto, CA 94306.
Present address: Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720.
To whom reprint request should be addressed. E-mail: [email protected].
Contributed by Lester J. Reed

Metrics & Citations


Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.

Citation statements



If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by


    View Options

    View options

    PDF format

    Download this article as a PDF file


    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to access the full text.

    Single Article Purchase

    The remarkable structural and functional organization of the eukaryotic pyruvate dehydrogenase complexes
    Proceedings of the National Academy of Sciences
    • Vol. 98
    • No. 26
    • pp. 14745-15393







    Share article link

    Share on social media