The authors injected material from human Alzheimer's patients into the brains of normal mice, and saw what appears to be the induction of amyloid pathlology. This didn't happen in control animals, got worse with time, and wasn't just noted at the point of injection. Their hypothesis is that Alzheimer's might be a prion-type disease of protein misfolding, and possibly capable of being spread by infectious particles.
On a very related note, I recently saw a talk from Zemer Gitai on an enzyme called CTP synthase that forms polymers, a property that's evolutionarily conserved from bacteria to yeast to fly to human. That's quite remarkable considering that it's a general metabolic enzyme involved in making nucleotides. And based on their high-throughput study of many enzymes, it seems like a lot more of them form these large-scale structures than people previously thought.