X-Ray Crystallography

The structures of the amino-terminal domains of apoE2, apoE3, and apoE4 in lipid-free forms have been determined, and all three structures adopt a four-helix-bundle motif (Figure 1). However, subtle differences in side-chain conformations and in salt-bridge arrangements among the isoforms affect their functions and characteristics. In addition, since apoE likely performs most, if not all, of its functions in a lipid-associated state, a major focus is to determine the influence of lipid binding on structure and function.

In collaboration with Dr. Bernhard Rupp (Macromolecular Crystallization Facility, Lawrence Livermore National Laboratory), we crystallized an amino-terminal fragment (residues 1–165) of apoE in three different crystal forms and solved their structures. Comparison of the structures revealed a large degree of conformational flexibility at the end of the molecule containing the loop (residues 80–91) connecting helices 2 and 3 and bends within helices 2 and 3. These bends, along with the flexible loop, suggest that this end of the molecule is flexible and likely represents the initiation site for lipid binding and the opening of the four-helix bundle as it reorganizes to bind lipid.

To test this model, we introduced single, double, and triple interhelical disulfide bonds to restrict the opening of the bundle in the amino-terminal domain fragment. Interaction of these mutants with dimyristoylphosphatidylcholine (DMPC) was assessed by vesicle disruption, turbidimetric clearing, and gel filtration assays. The results indicate that apoEïDMPC discoidal particles form in a series of steps. A triple disulfide mutant, in which all four helices were tethered, did not form complexes but could release encapsulated 5-(6)-carboxyfluorescein from DMPC vesicles, indicating that the initial interaction does not involve major reorganization of the helical bundle. After the initial interaction, the four-helix bundle opens to expose the hydrophobic faces of the amphipathic helices. In this step, helices 1 and 2 and helices 3 and 4 preferentially remain paired, since these disulfide-linked mutants bound to DMPC in a manner similar to that of the nonlinked control. In contrast, when helices 2 and 3 and/or helices 1 and 4 were paired, they bound poorly. However, all single and double helical pairings resulted in the formation of larger discs than were formed by the control, indicating that the helices undergo further reorganization after the initial opening of the four-helix bundle as the protein assumes its final lipid-bound conformation. In support of this rearrangement, reducing the disulfide bonds converted the large disulfide mutant discs to normal size.

The x-ray crystal studies indicated that the end of the bundle containing the flexible loop (residues 80–91) might initiate lipid binding. This loop contains three negatively charged residues, suggesting that an electrostatic interaction with lipid might be involved. This possibility was supported by three findings: (1) binding to DMPC was reduced by increasing ionic strength; (2) binding was reduced with a phospholipid with a negatively charged headgroup, dimyristoylphosphatidylglycerol (DMPG); and (3) neutralizing the negative charges in the flexible loop produced a mutant that bound to DMPG. These studies provide major new insights into how apoE binds lipid and the structural changes that occur with lipid interaction.

A major breakthrough in studying the interaction of apoE with lipid is the successful crystallization of apoE complexed with DMPC. This very exciting result opens for the first time the possibility of obtaining detailed structural information on protein–lipid complexes. This is important for apoE because high-affinity binding to LDL receptors requires lipid association.

ApoE4ïDMPC crystals displayed a fiber-like diffraction pattern with a unit cell spacing of 54 ≈ along the fiber axis and cell spacings of ~150 ≈ and ~300 ≈, respectively, for the two axes approximately perpendicular to the fiber axis. These findings are consistent with the idea that the apoEïDMPC discs (~150 ≈ in diameter and 55 ≈ thick) stack to form long, fiber-like rod structures. The stacking of the discs appears to be well defined along the fiber axis, as indicated by the resolution extending to about 7&ndash9 ≈ along this axis. The connection between the resulting rod-like fibers is less defined and extends to only ~15 ≈, indicating a weaker stacking interaction along the long cell axes where the sides of the discs touch each other. Recently, improved crystals have been obtained by substituting dipalmitoylphosphatidylcholine for DMPC. These crystals diffract to approximately 8 ≈ along all three axes.

 Tell a friend
 Printer friendly version