12 June 2005. Prions, the infectious agent of mad cow and Creutzfeldt-Jakob disease in humans, are proteins that form self-propagating, infective fibrils called amyloids. Their amyloid structure is a common link to many other human disease-causing proteins, including the amyloid-β aggregates that play a role in Alzheimer disease. However, these structures have not thus far yielded to fine-scale structural analysis. Now, by using a variety of techniques from x-ray crystallography, NMR, and protein biophysical measurements, three different groups have each produced new insights into the structure of prion fibrils from yeast and a fungus. Their results, published in three papers in the June 9 issue of Nature, show a new protein fold at the heart of the prions that is necessary for their infectiveness, and also give clues as to how the prion fibrils nucleate, propagate, and assume different conformational states that give rise to different biological activities. Increasing understanding of what drives fibril formation, and what can stop it, could suggest new avenues for blocking the aggregation of not just prions, but also amyloid-β and other pathological proteins.
The first paper, from Rajamaran Krishnan and Susan Lindquist at the Whitehead Institute in Cambridge, Massachusetts, combines a variety of painstaking low-resolution structural analyses to piece together a picture of the arrangement of individual prion proteins within a fiber. Starting with a 250-amino-acid fragment of the prion protein Sup35 from yeast, Krishnan inserted cysteine residues at 37 different positions, then expressed and purified 37 different recombinant proteins, and used them to assemble fibers. By modifying cysteine with fluorescent probes, they determined that a large portion of the N-terminal region (residues 21-121) was involved in a cooperatively folded, solvent-inaccessible structure. Krishnan identified two small stretches of amino acids, the head regions (residues 25-38) and the tail (91-106) that formed the direct contacts between prion molecules, always in head-to-head or tail-to-tail manner. In contrast, complete separation of the residues in between the head and tail was essential for fiber formation. Previously, Todd Yeates and colleagues at the University of California Los Angeles showed that transthyretin, another amyloid-forming protein, is also assembled in head-to-head, tail-to-tail manner, suggesting that this type of conformation might be common among different types of amyloid (see ARF related news story).
The assembly of head and tail contacts provides a way for fibers to grow by continuously recruiting additional subunits, but how do they get started? Kinetic studies of fluorescence transfer showed that when soluble proteins were mixed together, the N-terminal region quickly became solvent-inaccessible, before fiber assembly began. The authors found that the first ordered contacts occurred in the head region, followed by the tail, and by introducing a charged residue that blocks the head-to-head interaction, they were able to prevent fibril assembly. Recent work suggests amyloid-β goes through a similar molten oligomer phase in the early stages of fibril assembly, before finding the right contacts to nucleate fiber assembly (Huang et al., 2004, Eakin et al., 2004), suggesting that identifying and blocking a head-like region in that protein could be a way to stop aggregation.
One mysterious feature of prions is how proteins with identical amino acid structure can propagate different, heritable conformations that produce different phenotypes or disease states. Strain conformation has also been implicated in the species specificity of prion infectivity (see ARF related news story). Krishnan and Lindquist showed that fibers formed at various temperatures, a condition known to produce different strains, had distinctive protein-protein contacts in the head and tail regions. Chemical cross-linking of residues in either the head or tail biased fibril production toward one strain or the other, showing that these interactions were critical for giving rise to strains.
In a second paper, researchers from David Eisenberg’s laboratory at UCLA and collaborators in Copenhagen zoomed in for a close-up view of the organization of the fibril core using x-ray crystallography. Previously, first author Rebecca Nelson had shown that an seven-amino-acid peptide (GNNQQNY) from near the N-terminus of Sup35 could form stable, self-seeding, and polymorphic fibril-like structures. What they already knew about the core amyloid structure was that it consisted of a cross-β sheet—β strands spaced roughly 5 angstroms apart and lying perpendicular to the fiber axis formed β-sheets when the strands hydrogen bonded in register and in parallel. This arrangement has been shown in Aβ fibrils, too (see see ARF related news story and Benzinger et al., 2000).
What they learned from their new high-resolution (less than 2 angstroms) data was that the β-sheets were organized into pairs, with their opposing hydrophobic side chains enmeshed so tightly that water was excluded. They called this interdigitated structure a steric zipper. The structure was extremely stable, and could grow by adding additional peptides to the stacks that made up each sheet. The researchers speculate that because the sheets were held together by van der Waals forces and not by specific H bonds, this could allow for multiple structures from a given sequence, and explain the phenomenon of strain polymorphisms.
In the steric zipper, the self-complementary nature of the peptides is critical for allowing fibril formation. High concentrations of such peptides could drive fibril initiation in cells (e.g., in cells with high Aβ), and according to the author’s thermodynamic analysis, once the zipper gets formed, it would be very hard to reverse. Nelson et al. suggest that disrupting or capping the steric zipper may be a winning strategy to interfere with amyloid formation for therapeutic purposes.
A third paper, this one from Roland Riek at the Salk Institute n La Jolla, California, and collaborators from France and Switzerland demonstrates that the core of paired β-sheets is the infectious agent of prion structure. Using the HET-s prion from the filamentous fungus Podospora anserine, the investigators did NMR structural determination to arrive at a model that agreed with Nelson et al. where four β strands of eight amino acids each formed a very stable, solvent-inaccessible structure composed of two opposing β-sheets. By introducing cysteine mutations, à la Krishna and Lindquist, they determined that all the β-sheets had one solvent-accessible side, and one inaccessible side. When they introduced proline residues into the β-strands, sheet formation was prevented, as was infectivity.
In an accompanying News and Views piece, Christopher Dobson aptly characterized this avalanche of new information as “hard-won structural knowledge" that will enhance our understanding of the complex phenomena of amyloid formation and function.—Pat McCaffrey.
Krishnan R, Lindquist SL. Structural insights into a yeast prion illuminate nucleation and strain diversity. Nature. 9 June 2005; 435:765-772. Abstract
Nelson R, Sawaya MR, Balbirnie M, Madsen AØ, Riekel C, Grothe R, Eisenberg D. Structure of the cross-beta spine of amyloid-like fibrils. Nature. 9 June 2005; 435:773-778. Abstract
Ritter C, Maddelein M, Siemer AB, Luhrs T, Ernst M, Meier BH, Saupe SJ, & Riek R. Correlation of structural elements and infectivity of the HET-s prion. Nature. 9 June 2005; 435:844-848. Abstract
Dobson CM. Structural biology: Prying into prions. Nature [News & Views]. 9 June 2005; 435:747-749. Abstract