The PDH complex of B. stearothermophilus
self-assembles from a mixture of E1 heterotetramers and E3 dimers in the presence of the icosahedrally symmetric core formed by 60 E2 polypeptide chains. Typically a complex of ~42–48 E1 heterotetramers and 6–12 E3 dimers around one 60-mer E2 can be generated, but these values represent averages and the precise E1:E2:E3 stoichiometry is likely to vary from one complex to another (19
). E1 and E3 both bind tightly to the PSBD of the E2 chain (Kd
s of 0.33 and 0.58 nM, respectively) (37
), but the binding is mutually exclusive, and the distribution of these components about the E2 core is thus likely to have a major element of randomness to it. This intrinsic heterogeneity restricts the utility of approaches that involve averaging the structure of individual PDH complex assemblies that contain all of the constituent enzymes.
A key finding from the cryoelectron microscopic analysis of fully assembled E1E2 (14
) and E2E3 subcomplexes is that the E1 and E3 components are located as well defined radial shells with a clear gap of 75–90 Å separating them from the icosahedral inner core constructed of 60 E2 acetyltransferase domains ( and and Ref. 14
). The gap in the E2E3 subcomplex appears to be slightly smaller than that in the E1E2 subcomplex but that may be because of differences in the overall shapes of the E3-PSBD (21
) and E1-PSBD (7
) structures. Three sections of the E2 polypeptide must be located in this gap, although none of them makes any apparent contribution to the density map: the N-terminal lipoyl domain, which shuttles between the catalytic core and the outer shell, the linker region connecting the lipoyl domain to the binding domain, and the linker region connecting the binding domain to the catalytic core domain.
The structural mechanisms that contribute to the presence of the gap and maintain the size of the E1E2 and E2E3 subcomplexes and native PDH complexes are not yet clear. One possibility is that steric interactions between neighboring molecules reduce the possibility of large inward movements of the E1 tetramers and/or of the E3 dimers. It also seems plausible that the extended inner linker region (approximately between Ala167
may play a critical role in determining the distance of closest approach between the catalytic domain and the bound E1 and E3 enzyme complexes. No apparent secondary structure was detected in this stretch of amino acids by the program PredictProtein (38
), although all of the known secondary structure elements of E2 were correctly determined (data not shown). However, most of the X-Pro peptide bonds present in the inner linker appear to occupy the all-trans
rather than cis
), which could provide a certain level of rigidity to the chain of amino acids. Inspection of the structure of the assembled E2 core of 60 acetyltransferase domains (17
) suggests that the Gly204
residues of each of the three monomers that constitute a trimer at each vertex of the icosahedral core are separated by 42 Å, and the Gly204
residues in adjacent trimers are even further apart, making it unlikely that neighboring linker regions can form close associations to form a higher order structure. Further work by means of NMR spectroscopy and other techniques will be needed to analyze the structure and mechanistic relevance of the linker region in more detail.
The architecture of the B. stearothermophilus
PDH complex is likely to be broadly similar to that of the icosahedral PDH complexes of its eukaryotic counterparts. Thus cryoelectron microscopy of the PDH complex from bovine kidney (10
) and S. cerevisiae
) indicates that the E1 components form a radial shell ~50 Å above the central E2 core, although it is reported in the former paper that a high occupancy of the E1 binding sites can lead to an extension of the linker region from 50 to ~75 Å. Moreover, in these complexes, the E3 homodimers are bound to the E2 core by association with an additional component, an E3-binding protein. This appears to lead to the E3 being more intimately associated with the inner E2 core in the openings on the 12 5-fold faces of the dodecahedron (8
). Other cryoelectron microscopic (9
) and biochemical studies (40
) indicate a more exterior positioning of the E3 in the mammalian PDH complex. Thus, while following the same general pattern, the molecular architecture of PDH complexes from different sources may differ in detail.
We had previously suggested that the presence of a high surface concentration of the E1 heterotetramers limits the search space required for active site coupling, by allowing the lipoyl domain to shuttle back and forth across and within the gap between the E1 and E2 active sites (14
). A complete turnover of the complex requires the successive participation of an E1, E2, and E3 active site working on a given lipoyl domain. Given the high ratio (~4–8:1) of E1 to E3 in the assembled B. stearothermophilus
PDH complex (19
), and the apparently random placement of E1 and E3 over the surface of E2, it would clearly be desirable for the lipoyl domain to easily access the more limited number of E3 active sites. The positioning of E3 in an annular shell at about the same level as E1, rather than outside of the E1 protein layer, and with active sites accessible from the annular gap, is therefore likely to be important for the effective regeneration of the lipoyl moiety (). The lack of E1 and E3 protein density in the annular gap could also facilitate movement of the lipoyl domain that may have to sweep through considerable annular space before reaching an available E3 homodimer, especially in native PDH complexes with only a few E3 enzymes per complex. An individual lipoyl domain is predicted to migrate up to 140 Å away from its PSBD given the ~50 amino acid linker that separates the two domains (14
). In a native PDH complex, this movement would enable it to interact with at least nine peripherally bound enzymes. A hypothetical model of a native pyruvate dehydrogenase complex containing 60 E2, 50 E1, and 10 E3 () positioned about the E2 core using the best fitting coordinates derived from models of the E1E2 (14
) and E2E3 subcomplexes () indicates that there are no apparent steric constraints that might prevent the formation of such a complex. The arrangement of the enzymes is consistent with the hypothesis that the lipoyl domain primarily resides and functions inside the annular gap.
The possibility that the movement of the lipoyl domain is not completely free but instead may occur via constrained trajectories has been suggested previously (41
). Inspection of the outer surface of the icosahedral inner core of E2 reveals it to be predominantly positively charged, with a ring of positive charges near the entrance to each active site. In contrast, the lipoyl domain is predominantly negatively charged. Although the lysine residue to which the lipoyl group is attached is positively charged, this charge is neutralized by the amide bond formation that occurs upon attachment of the terminal amino group to the lipoyl domain. Thus, the movement of the lipoyl domain may not be random and could be guided, in part, by simple electrostatic interactions in the interior of this elegant molecular machine.