Amyloid Surprises

Many diseases, including Alzheimer’s and Parkinson’s, are characterized by the formation of clumps of long protein fibrils. Smaller intermediates that form first are suspected by some scientists to be the actual toxic form of the proteins that help cause these conditions. Stopping such intermediates from forming might thus be preventive or therapeutic, but little has been known about their structures.

Two recent studies provide the first glimpses of what two of those intermediates might look like, and another study probes fiber inhibition. All the findings are surprising.

In one study, David Eisenberg of the University of California, Los Angeles, and coworkers, collaborating with Charles G. Glabe of UC Irvine, focused on the protein αB-crystallin, which is found in the eye lens (Science, DOI: 10.1126/science.1213151). ­Fibrils formed by the protein are one source of cataracts. The researchers chose to study that protein because it forms fibrils more slowly than do proteins such as amyloid-β (Aβ), the peptide involved in Alzheimer’s disease.

They thought that the slower fibril formation would make it easier to capture an αB-crystallin oligomeric intermediate with A11, an antibody developed by Glabe that recognizes amyloid oligomers. Using a computer algorithm devised by Eisenberg’s team, they identified an 11-residue segment of αB-crystallin that forms amyloid structures recognized by A11.

In solution, the segment forms a hex­amer the researchers call cylindrin. It has six β-strands arranged in a cylindrical barrel. The strands are antiparallel, with neighboring strands running in opposite directions. In addition, they are out of register, meaning that they are shifted on their long axes relative to one another. (In antiparallel, in-register β-sheets, on the other hand, every other strand lines up exactly.)

“You can’t take an in-register sheet and form a cylindrin,” Eisenberg says. “If you roll up an in-register sheet, the side chains are all at the same level along the cylindrical axis. They would bump into each other.”

He and his coworkers find that for a cylindrin to form, a glycine must occupy a particular position in the strands and apolar residues like valine must hold the cylindrin together through hydrophobic interactions in the center of the barrel.

Cylindrins are a possible model for toxic amyloid species. “It’s a hypothesis at this point,” Eisenberg says, “but it’s a testable hypothesis.” His immediate goal is testing whether cylindrin really is the disease-causing form of the toxic intermediate. “If it is found to be the toxic structure, the next goal would be to block its formation,” he says.

“Purists will argue that the αB-crystallin peptide oligomers studied here are not biologically relevant,” comments Ronald B. Wetzel, an amyloid expert at the University of Pittsburgh. “However, the parallels to oligomeric aggregates of Aβ and other protein aggregation disease molecules seem inescapable.”

Robert Tycko, a biophysicist at the National Institute of Diabetes & Digestive & Kidney Diseases at the National Institutes of Health, and his coworkers found another amyloid intermediate made up of antiparallel β-sheets. They used solid-state nuclear magnetic resonance to determine the structure of these antiparallel β-sheets in an intermediate formed by the so-called Iowa mutant of amyloid-β (Proc. Natl. Acad. Sci. USA, DOI: 10.1073/pnas.1111305109).

When Tycko and Stephen C. Meredith of the University of Chicago first studied this mutant, they were puzzled by their NMR measurements. All of the other fibrils they’d studied contained only parallel β-sheets, and they thought parallelism “was a generic structural feature of amyloid fibrils,” Tycko says. “When we began to take solid-state NMR data on the Iowa mutant fibrils, the data didn’t make sense.”

The solid-state NMR measurements give them distance constraints between carbon-13 nuclei or between carbon-13 and nitrogen-15 nuclei, depending on the experiment. “It took us about two years to figure out that it was forming antiparallel structures as well as parallel structures,” Tycko says. The antiparallel structures were hard to find because, given enough time, antiparallel structures in the transient intermediates disappear and are replaced by parallel structures.

Postdoc Wei Qiang got around this problem by developing a procedure in which he used a mixture of antiparallel and parallel structures as a seed for fibril growth and filtered out the more rapidly growing parallel structures. That gave him a sample of antiparallel structures that was pure enough for NMR measurements.

Mature fibrils are generally long and straight, whereas the antiparallel fibrils are relatively short and curvy. That makes them excellent candidates as components of early-forming amyloid intermediates known as protofibrils.

Very little has been known about the molecular structure of protofibrils, Tycko says. But his group’s work suggests that some protofibrils may have antiparallel components similar to those in intermediates formed by the Iowa mutant of amyloid-β.

Both cylindrins and protofibrils appear to be detours on the route to formation of mature fibrils. “Which structures you see will depend on the details of your fibril growth conditions,” Tycko says.

Both intermediates can eventually form fibrils, but not by simple rearrangement. “The energy barrier is too high,” Tycko says. Instead, what likely happens is that individual strands leave the less stable intermediate and join a more stable growing fibril.

Like in the cylindrins, antiparallel β-sheets in oligomers of the Iowa mutant of amyloid-β are stabilized by hydrophobic interactions between strands. “You can get very similar hydrophobic cores with both antiparallel and parallel structures,” Tycko says. But the parallel structure is more stable than the antiparallel one, he notes.

In other recent amyloid work, Martin T. Zanni and Juan J. de Pablo of the University of Wisconsin, Madison; Daniel P. Raleigh of the State University of New York, Stony Brook; and coworkers studied the behavior of the rat peptide amylin, an inhibitor of human amylin fibril formation (Nat. Chem., DOI: 10.1038/nchem.1293). Rat amylin is not a strong inhibitor of amyloid fibril formation, and Zanni and coworkers wanted to better understand how it works so they could improve its inhibition ability.

They first studied the structure of human amylin, also called islet amyloid polypeptide, which forms β-sheet-rich fibrils in islet cells of patients with type 2 diabetes. The fibrils consist of one β-sheet at the amino, or N-terminal, end of the protein and another at the carboxy, C-terminal end.

In their structural study, they used two-dimensional infrared spectroscopy to measure frequency shifts between isotope-labeled and unlabeled amino acids in human amylin. Large shifts indicate that a particular residue is part of a highly ordered β-sheet.

Sequences of the rat and human peptides suggested that rat amylin would disrupt the C-terminal β-sheet of human amylin. Instead, Zanni and coworkers found that after the two peptides had equilibrated with one another for eight hours, the inhibitor disrupted human amylin’s N-terminal β-sheet but had little effect on its C-terminal β-sheet.

A bigger surprise came when they looked at the mixture after 24 hours. At that point, rat amylin no longer blocked human amylin β-sheet formation and instead formed a β-sheet of its own. Rat amylin had never before been observed to make amyloid β-sheets, so it was shocking that the supposed inhibitor had morphed into a potentially detrimental structure.

At this point, the researchers don’t know what’s happening on a molecular level, but Zanni has some ideas about why this unexpected structure forms.

The 2-D IR suggests that the rat amylin forms a “skin on the outside of the human fibers,” Zanni says. “We know that rat amylin prevents the formation of the outer β-sheets” of human amylin, he says. “But to do so, the inhibitor is being aligned by the very same structure that it’s supposed to inhibit. As a result, we hypothesize that it is templated into a conformation that it normally doesn’t adopt but that causes aggregation similar to the disease peptides.”

“The experiments by Raleigh, Zanni, and coworkers show that aggregation of rat amylin can be induced by human amylin, apparently through an unanticipated pathway in which the two peptides first aggregate together in an intermediate state and then aggregate separately in the final state,” Tycko says. “These results may have implications for biomedical phenomena in which aggregation of one disease-associated protein may catalyze or promote the aggregation of another disease-associated protein, thereby linking two distinct amyloid diseases.”

The understanding of mature fibrils has advanced tremendously in the past decade, Tycko says, but the understanding of intermediates has been more tenuous.

“These three new papers constitute progress toward a detailed understanding of intermediates that may be either significant cytotoxic entities in amyloid diseases or mechanistically important entities that can be targeted by amyloid inhibitors,” Tycko says.