Structural integrity of β-sheet assembly

The folding of a protein from a sequence of amino acids to a well-defined tertiary structure is one of the most studied and enigmatic events to take place in biological systems. Relatively recently, it has been established that some proteins and peptides are able to take on conformations other than their native fold to form long fibres known as amyloid. In vivo, these are associated with misfolding diseases, such as Alzheimer’s disease, Type 2 diabetes and the amyloidoses. In vitro, peptide assembly leads to amyloid-like fibres that have high stability, resistance to degradation and high tensile strength. Remarkably, despite the lack of any obvious sequence similarity between these fibrillogenic proteins and peptides, all amyloid fibrils share common structural characteristics and their underlying structure is known as ‘cross-β’. Nature is rich in β-sheet protein assemblies such as spider silk and other ‘useful’ amyloids such as curli from Escherichia coli, where the strength of fibrils is fundamental to their function. Identifying sequence determinants of protein assembly Numerous proteins and peptides are able to assemble to form β-sheet-rich amyloid fibrils associated with disease, from the 37-residue peptide IAPP (islet amyloid polypeptide) to the 253-residue prion protein (see [1,2] for reviews), although these proteins and peptides do not appear to share any obvious common sequence motifs. In addition, many non-disease-related sequences are able to assemble under optimal conditions to form amyloid-like fibrils, and it has been suggested that amyloid represents the lowest thermodynamic state achievable by a protein chain [1], and that any protein or peptide is capable of forming amyloid in vitro under particular environmental conditions [1]. This suggests that the backbone may be entirely responsible for the assembly process, and it is clear that the fibrils are stabilized by extensive hydrogen-bonding along the length of the β-sheets. However, the propensity for different sequences to form amyloid fibrils has been explored in detail, and a number of aggregation prediction algorithms have been produced [3,4] to try to predict the rate at which a particular sequence will assemble in an ordered β-sheet manner. In general, these algorithms assess the β-sheet propensity of residues, hydrophobicity and hydrophilicity, as well as the overall charge and charge distribution and the occurrence of aromatic residues [3–5]. Aromatic residues have been highlighted as contributing to the core structural stability of fibrils [5–7], although their presence is not necessary for assembly [8]. Designed short peptides are useful model systems for determining which residues are important in fibril formation [8]. Indeed, mutations in proteins are often the reason behind