13 Dec 2005

What makes amyloid-β (Aβ) polymerize? Answering that question might reveal a way to prevent formation of toxic Aβ oligomers, protofibrils, and fibrils, and perhaps even slow or stop the progression of Alzheimer disease (AD). But there’s a snag. In vitro, it is very difficult to measure Aβ oligomerization in anything that approaches its natural environment, that is, the dense fatty tissue of the brain. That’s why many researchers have turned to “in silico” methods to model Aβ behavior. In the December 9 PNAS online, Eugene Stanley and colleagues at Boston University, together with David Teplow and colleagues at the University of California, Los Angeles, described computer modeling studies that offer insight into Aβ folding, particularly on how the so-called Dutch mutation may contribute to formation of Aβ fibrils. This variant, which has a glutamine instead of glutamic acid at position 22, forms fibrils much more readily than wild-type Aβ.

Thanks to rapidly increasing computer power, modeling of small peptides such as Aβ is getting progressively easier and more accurate. In this paper, first author Louis Cruz and colleagues report the use of molecular dynamics to estimate the interactions of all the atoms in a truncated version of Aβ comprising amino acids 21 to 30. These particular amino acids are involved in nucleating the oligomerization process (see, for example, ARF related news story) and, being resistant to proteolysis, they may play a crucial role in stabilizing long-lived oligomers or aggregates.

Cruz and colleagues modeled intramolecular bonds under various conditions, including water as solvent, in the presence of salt (which is more akin to natural conditions), and in water with added solutes (which effectively reduces the density of the solvent, as might be expected in the dense milieu of brain tissue). They also investigated the significance of the Dutch mutation. The scientists found that hydrophobic bonds and salt bridges between amino acids 28 (lysine) and 22 (glutamic acid) or 23 (asparagine) are important in stabilizing a loop that forms in a region extending from valine 24 to lysine 28. Salt increased the propensity of ionic bridges to form in this loop, whereas the addition of solutes led to more hydrophobic interactions. These hydrophobic bonds increased the likelihood that an α helix might form instead. Finally, in peptides containing the Dutch mutation, the loop is much less stable and appears more prone to unfold. Indeed, recent work from the Teplow lab suggests that partial unfolding of the 21-30 region of Aβ needs to occur before fibrils can form (see Lazo et al., 2005). Therefore, the poor stability of the valine 24 to lysine 28 loop in the Dutch Aβ21-30, “might explain the higher propensity of the Dutch peptide to form protofibrils and fibrils,” write the authors.—Tom Fagan