Amphoterin includes a sequence motif which is homologous to the Alzheimer's beta-amyloid peptide (Abeta), forms amyloid fibrils in vitro, and binds avidly to Abeta.

Many of the proteins associated with amyloidoses have been found to share structural and sequence similarities, which are believed to be responsible for their capability to form amyloid fibrils. Interestingly, some proteins seem to be able to form amyloid-like fibrils although they are not associated with amyloidoses. This indicates that the ability to form amyloid fibrils may be a general property of a greater number of proteins not associated with these diseases. In the present work, we have searched for amyloidogenic consensus sequences in two current protein/peptide databases and show that many proteins share structures which can be predicted to form amyloid. One of these potentially amyloidogenic proteins is amphoterin (also known as HMG-1), involved in neuronal development and a ligand for the receptor for advanced glycation end products (RAGE). It contains an amyloidogenic peptide fragment which is highly homologous to the Alzheimer's amyloid beta-peptide. If enzymatically released from the native protein, it forms amyloid-like fibrils which are visible in electron microscopy, exhibit apple green birefringence under polarized light after Congo red staining, and increases thioflavin T fluorescence. This fragment also shows high affinity to Abeta as a free peptide or while part of the native protein. Our results support the hypothesis that the potential to form amyloid is a common characteristic of a number of proteins, independent of their relation to amyloidoses, and that this potential can be predicted based on the physicochemical properties of these proteins.