2 February 2011. Amyloid sure likes to keep its secrets. Despite decades of research, scientists still do not know precisely what makes amyloidogenic molecules such as Aβ gang up and go bad. Two new papers yield some clues to the puzzle. In the January 26 Journal of Neuroscience, researchers led by Stewart Nuttall at Commonwealth Scientific and Industrial Research Organization in Parkville, Victoria, Australia, describe the first x-ray crystallographic structure of an oligomeric Aβ fragment—though not yet oligomers of the most pathogenic forms of Aβ. Intriguingly, the structure looks quite different from fibrillar Aβ. Its features jibe with computational models, although only further research will reveal whether the form occurs in vivo. Researchers led by Sheena Radford at the University of Leeds, U.K., focused on the question of what causes an amyloidogenic protein to begin sticking together, using β2-microglobulin as their test subject. In the January 21 Molecular Cell, the scientists detail how subtle changes in shape can tilt the molecule toward aggregation, changing a formerly well-behaved protein into a bad seed. The authors demonstrate a mechanism that allows a handful of these twisted molecules to corrupt a much larger mass of correctly folded proteins.
Much of scientists’ current understanding of Aβ structure rests on studies of fibrillar amyloid. One seminal nuclear magnetic resonance study led by Robert Tycko at the National Institutes of Health showed that Aβ fibrils have an ordered β-turn-β structure (see ARF related news story on Petkova et al., 2002). In recent years, however, the scientific spotlight has zoomed in on oligomers, now widely believed to be the most toxic form of Aβ. Oligomers have frustrated scientists’ efforts to crystallize them, because they appear never to hold still in solution, forming larger aggregates that spontaneously fall apart.
First author Victor Streltsov in Nuttall’s group got around these problems by making use of a single chain shark immunoglobulin to box in the Aβ oligomers as they formed. Streltsov and colleagues created chimeric proteins in which the immunoglobulin was bound to the p3 fragment of Aβ (residues 17 to 42, created by successive α and γ cleavage). As Aβ peptides began to stick together in solution, the attached antibody formed a cage around the amyloid fragment, trapping the oligomer and preventing further aggregation. The authors then crystallized this immobilized complex and analyzed its structure by x-ray. To their surprise, they saw not the classic β-hairpin shape known from fibrils, but instead a tetrameric, globular structure consisting of two connected loops.
“This paper is important because it finally shows that the structure of oligomers is not like a piece of a fibril,” said Brigita Urbanc of Drexel University in Philadelphia, Pennsylvania.
The general features of this oligomer match some recent predictions, Urbanc said. Computer modeling from her lab suggests that oligomers form compact shapes, with hydrophobic residues buried in the center and hydrophilic residues exposed, and predicted the key roles of residues identified by Streltsov and colleagues as central to tetramer formation (see Urbanc et al., 2004 and Urbanc et al., 2010). The observed structure also fits with spectroscopic data from Charles Glabe’s lab at the University of California in Irvine (see ARF related news story).
But does this structure occur naturally in brains? “That’s the million dollar question,” Urbanc said, pointing out that because of the presence of the shark proteins, the oligomer may form differently than in vivo.
Urbanc notes, however, that several details of the structure do fall in line with other studies. For example, Streltsov and colleagues show that the lysine 28 side chain can take on two different configurations, a finding detailed in several previous experimental and computational studies by Dave Teplow and colleagues at the University of California in Los Angeles (see Lazo et al., 2005; Borreguero et al., 2005; ARF related news story on Cruz et al., 2005; Baumketner et al., 2006; Grant et al., 2007; and Baumketner et al., 2008).
The choice of the p3 Aβ fragment raises additional questions about the pathogenic significance of this structure. The p3 fragment is found in amyloid plaques of Alzheimer’s disease and in the pre-amyloid lesions of Down's syndrome, but has typically been considered both harmless and non-amyloidogenic. A recent study led by Ratnesh Lal at the University of California in San Diego reported that p3 peptides can form channels in cell membranes, leading to calcium influx and toxicity (see ARF related news story on Jang et al., 2010). The fragment is also amyloidogenic, Lal contends, although it aggregates more slowly than full-length Aβ (see Pike et al., 1995). The data from the new paper are not conclusive on these points: The complex of Aβ and shark proteins was not toxic to cells, Streltsov and colleagues found, but the surrounding shark proteins may have blocked Aβ’s effects. Streltsov said that they are working on removing the oligomer from its protein cage to directly test its toxicity. One potential complication is that without the surrounding shark immunoglobulins, Aβ will be free to aggregate further, leading to a more heterogeneous mixture. The authors are also making oligomers of longer Aβ fragments inside scaffolding proteins, Streltsov said, and will analyze those structures by x-ray as well.
In the second paper, Radford and colleagues focused on what prompts amyloidogenic proteins to assume their toxic forms. The authors used β2-microglobulin, which is part of the Class I major histocompatibility complex (MHC-I) and forms amyloid deposits if not properly cleared by the kidney, as when people are on dialysis. These deposits contain a truncated form of β2-microglobulin called ΔN6 that lacks six N-terminal amino acids and is highly amyloidogenic. First author Timo Eichner began by analyzing the structure of ΔN6, finding that in the amyloid form, the peptide bond at proline32 was changed from a cis to a trans configuration. This small change led to widespread repacking of the hydrophobic core of the protein, altering surface charges and making the protein more sticky and aggregation-prone. The missing N-terminal region normally serves to lock the protein into its native structure and preserve the cis bond, the authors conclude. This is the first description of the atomic-resolution structure of ΔN6, points out Lila Gierasch of the University of Massachusetts, Amherst, in an accompanying commentary.
Eichner and colleagues next wondered how native β2-microglobulin might change into an aggregation-prone state. They found that adding even 1 percent ΔN6 to a β2-microglobulin solution catalyzed the assembly of all the native protein into amyloid fibrils over a period of several weeks. Amyloid conversion occurred by collision; when a misfolded protein bumped into a native protein in just the right orientation, Radford said, the native protein reorganized into the toxic conformation. NMR measurements revealed that the contact caused the AB-loop (residues 13-22) of β2-microglobulin to relax. This in turn displaced a strand from the protein’s β sandwich structure, allowing the proline32 bond to flip into the trans conformation. In ongoing studies, Radford said, they hope to further elucidate the exact mechanism by which ΔN6 promotes this change. The reaction occurred more quickly under mildly acidic conditions, which the authors traced to the effect of a histidine residue picking up a proton. This positive charge destabilized the protein and made the alternate, amyloidogenic shape more favorable.
“We learned that proteins can change their conformation on a collision, and that very subtle changes such as picking up a proton or isomerizing a single bond is enough to cause the conversion,” Radford said. Next, they would like to use the methods they have developed to look at later steps in the aggregation pathway, Radford said. Knowing the mechanisms behind aggregation might provide clues as to how to prevent it, Radford suggested, for example, by designing small molecules to inhibit the initial conformational change.
It is not clear if a similar mechanism occurs in Aβ aggregation, Radford said, as unlike β2-microglobulin, Aβ does not start as a folded protein. Nonetheless, the basic finding that subtle changes in shape can have dramatic effects on aggregation might well hold for Aβ, Radford speculated. In addition, the research suggests that, “Potentially any protein in a dangerous conformation could convert an innocuous protein,” Radford said. “Catalytic conversion could be a general phenomenon in amyloidosis.” If true, this might provide a mechanism for the emerging data that Aβ can spread in a prion-like way through the brain (see ARF related news story on Eisele et al., 2009). In other words, just as 1 percent ΔN6 sets off aggregation in a β2-microglobulin solution, a minor increase in the concentration of Aβ42 might trigger fibril formation in the brain.—Madolyn Bowman Rogers.
Streltsov VA, Varghese JN, Masters CL, Nuttall SD. Crystal structure of the amyloid-β p3 fragment provides a model for oligomer formation in Alzheimer’s disease. J Neurosci. 2011 Jan 26;31(4):1419-26. Abstract
Eichner T, Kalverda AP, Thompson GS, Homans SW, Radford SE. Conformational conversion during amyloid formation at atomic resolution. Mol Cell. 2011 Jan 21;41(2):161-72. Abstract
Gierasch LM. How one bad protein spoils the barrel: Structural details of β(2)-microglobulin amyloidogenicity. Mol Cell. 2011 Jan 21;41(2):129-31. Abstract