The pathological aggregation of proteins into amyloid fibrils is the hallmark for a variety of diseases. The fibrils are morphologically similar- long unbranched filaments displaying a characteristic cross-β structure- despite sometimes large sequence differences, and distinct pathological symptoms. While each amyloid disease is commonly linked to a main misfolded protein, progression of some amyloid diseases affects progression of others. In Alzheimer’s disease (AD), accumulation of amyloid (Aβ) in the brain is thought to occur at the earliest stages of the disease and drive pathogenesis. However, formation of neurofibrillary tangles containing another amyloid protein, tau, correlates much better with cognitive loss. Patients with AD also carry a significant risk for developing type-II diabetes, in which fibrils of human islet amyloid polypeptide (hIAPP) deposit in the kidneys.

The common amyloid aggregation motif has posed the hypothesis that one amyloid protein could directly nucleate another via cross-amyloid interactions. To elucidate the molecular basis for these interactions, we used a pioneer method, microcrystal electron diffraction, to determine atomic structures of potential cross-seeding regions of Aβ. We found that fibrils of structurally similar 11-residue segments from hIAPP and Aβ induce amyloid formation of both parent proteins through self- and cross-seeding. Structure-based inhibitors designed for the hIAPP segment were able to reduce cytotoxicity of both proteins. Additionally, we designed inhibitors for second potential cross-seeding region of Aβ. While reducing Aβ aggregation and toxicity as designed, the inhibitor also reduces Aβ-mediated tau aggregation, and moreover reduces aggregation and self-seeding of tau fibrils. Using mutagenesis and structural alignment, we find that tau and Aβ may contain similar interfaces. Fully understanding cross-amyloid interfaces could provide insight into disease progression and potentially also lead to therapeutic targets.