Without firm data on amyloid structure, prior efforts to block fibril formation used two basic approaches—screening for molecules that disrupt the process in solution, or devising compounds to interfere with the “sticky” portions of amyloidogenic proteins. “We tried a third approach,” Eisenberg said. “Starting with the atomic structure of a piece of tau protein, we asked what sort of molecule we could build that would cap a tau fiber so it won’t grow any further.”

Reaching the starting point was a feat in itself. Full-length tau does not crystallize, making it impossible to solve its atomic structure. Instead, the researchers focused on previously identified short peptides that drive protein self-aggregation. These short peptides can themselves grow into fibers. Furthermore, some of those hexapeptides formed crystals amenable to structural analysis, revealing an intriguing characteristic of amyloid fibrils: The side chains of hydrophobic amino acids on paired β-sheets interlocked tightly to form “steric zippers” (ARF related news story on Nelson et al., 2005). Tau protein also has these features (Sawaya et al., 2007).

For the current study, Stuart Sievers and colleagues focused on one such zipper—a hexapeptide (VQIVYK) that drives assembly of tau into pathological filaments (von Bergen et al., 2000). With this peptide as a template, Sievers worked with co-first author John Karanicolas of David Baker’s group at the University of Washington, Seattle, to design an inhibitor to cap elongating tau fibers. Howard Chang, Anni Zhao, and Lin Jiang of the Eisenberg lab also contributed equally to the work. The team used software developed in the Baker lab to identify candidate blockers, then showed that they work in solution to slow fibrillization of the template hexapeptide, as well as 150- and 128-residue tau fragments. To improve the inhibitors’ stability, researchers built them using non-natural D-amino acids. “Our bodies are filled with proteases trained to chew up [L-amino acid] peptides that are not normally in cells. That’s why we chose non-natural amino acids,” Eisenberg said. In a recent study, Aβ-binding D-enantiomeric peptides reduced plaque load and improved cognition in AD transgenic mice (Funke et al., 2010), suggesting D-amino acid peptides may be well suited for drug development.

“It’s an awesome protein engineering paper,” said Torleif Härd of the Swedish University of Agricultural Sciences, Uppsala. Dieter Willbold of the Institute of Structural Biology and Biophysics in Jülich, Germany, cited specificity as a key advantage of inhibitors designed using the structure-based approach (see full comment below). Mark Findeis of Satori Pharmaceuticals, Cambridge, Massachusetts, called the work “a really nice step forward,” adding that its methods could be used to design an inhibitor for “anything that has known structural data. You could apply it to models of other proteins, even if you didn’t have firm structural data like what’s available for tau,” he told ARF.

Sievers and colleagues used their computational method to build inhibitors not only for tau fibrils, but also for an amyloidogenic protein (prostatic acid phosphatase) that enhances human immunodeficiency virus (HIV) infection. The latter blocked formation of phosphatase fibrils and, importantly, prevented HIV infection in a functional assay.

The tau inhibitors, on the other hand, are not expected to work in a biological setting because they do not penetrate the brain, Eisenberg said. His group is currently using the inhibitors as tools for finding other tau fibrillization blockers that may be more brain penetrant.

In a PLoS Biology paper published earlier this week (Landau et al., 2011), Eisenberg and colleagues report the first atomic-level structures of small molecules bound to amyloid forms of tau and Aβ. “These pictures begin to map out what’s called the amyloid pharmacophore—a collection of sites on a protein that define how ligands bind to it. That’s a step to drug design,” Eisenberg said.

As for whether blocking fibrillization will be able to help AD patients, only time will tell. Several Aβ aggregation inhibitors have looked promising in Phase 2 trials. One (Elan Corporation’s scyllo-inositol) is heading toward Phase 3 (see ARF related news story). Another (tramiprosate, aka Neurochem’s Alzhemed) failed in a Phase 3 trial (see ARF related news story). Meanwhile, at least one study questions the whole premise by suggesting that some small-molecule amyloid inhibitors work by forming colloids that generally disrupt protein-protein interactions, even the good ones (ARF related news story on Feng et al., 2008).—Esther Landhuis.

Reference:
Sievers SA, Karanicolas J, Chang HW, Zhao A, Jiang L, Zirafi O, Stevens JT, Münch J, Baker D, Eisenberg D. Structure-based design of non-natural amino-acid inhibitors of amyloid fibril formation. Nature. 2011 June 15. Abstract