We started studying the structures of yeast prions in 1998, in collaboration with R Wickner (NIDDK), focussing initially on Ure2p, a negative regulator of nitrogen catabolism. We showed that its N-terminal domain is responsible for prionogenesis, while the C-terminal domain which performs its regulatory function remains folded in filaments but is inactivated by a steric mechanism. In our amyloid backbone concept, the prion domains form the filament backbone and are surrounded by the C-terminal domains. In 2005, we published the parallel superpleated beta-structure model for the amyloid backbone. It envisages arrays of parallel beta-sheets generated by stacking monomers with planar beta-serpentine folds. Topologically similar structures are good candidates for certain other amyloid fibrils, including amylin. Ongoing work is aimed at testing and refining this model; exploring its range of applicability; and investigating fibril polymorphism. In FY07, we focused on two areas: ? ? (1) Fibrillation of the prion domain of the HET-s prion of the filamentous fungus Podospora anserine. This prion differs from Ure2p and Sup35p in being a gain-of function prion, and in not having a high concentration of Asn/Gln residues. We showed by electron diffraction that Het-s fibrils have a cross-beta structure, and performed mass-per-unit-length measurements by scanning transmission electron microscopy. The latter data show that it has an axial packing density of 1 subunit per 0.94 nm - half the density of Ure2p and Sup35p fibrils and in agreement with a published model, the stacked beta-solenoid. It follows that the respective amyloid architectures are basically different. We also investigated differences among the amyloids formed by the HET-s prion domain under various conditions in vitro. We distinguish two types formed at pH 7 from fibrils formed at pH 2 on morphological grounds. Unlike pH 7 fibrils, the pH 2 fibrils lack detectable infectivity. They also differ in ThT-binding, resistance to denaturants, assembly kinetics, secondary structure, and intrinsic fluorescence. Thus, the HET-s can form both infectious (prion) and non-infectious (non-prion) amyloids in vitro. Both contain 5 nm fibrils, either bundled or disordered (pH7) or as tightly twisted protofibrils (pH2). The altered properties of amyloid assembled at pH 2 may arise from a perturbation in the subunit fold or in fibrillar stacking.? ? (2) An extensive class of monomeric and oligomeric proteins have been demonstrated to have structures that has been proposed to occur, in polymeric form, in HET-s fibrils (see above) and some other amyloids. These proteins, which are mostly bacterial secreted proteins, have so-called beta-solenoid folds. They represent a source of insight into amyloid formation. We have developed bioinformatic methods to identify amino acid sequences conducive to beta-solenoid folds and to predict their three-dimensional structures, focussing on the passenger domains of autotransporter proteins from Type V bacterial secretion systems. In 2001, we proposed a beta-helical structure for the filamentous hemagglutinin (FHA) of B. pertussis. In 2005, a crystal structure for an FHA fragment was published. On comparing it with our model, we found very close agreement, confirming the prediction. The number of known autotransporters is now over 1000. To investigate the incidence of beta-solenoids among them, we carried out a sequence-based analysis and conclude that, despite wide sequence diversity, most of them have beta-solenoid domains that we classify into thirteen types based on properties of their beta-coils (repeat length, numbers and lengths of strands and turns, cross-sectional shape, etc) summarized in a coil template. As a further exercise in model-building from a coil template, we generated a type-T4 beta-solenoid for TibA of E. coli.
