In our effort to design a soluble oligomer-forming mutant, we have previously predicted that the mutation of G37p will result in enhanced β hairpin formation and increased oligomer stability. Two-turn structures were obtained using NOE refinement. The occurrence of the first turn at V24-N27 is in agreement with the available SS-NMR fibril studies. The second turn consisting of residues V36-V39, absent in previous studies, is introduced here by the mutation. In the nucleated polymerization model fibrils are likely to grow by monomer addition 
. The two-turn structure of the mutant possibly depletes monomer content by diverting peptide into a stable oligomer formation.
Earlier NMR studies showed that Aβ monomers adopts a collapsed coil (mostly random) with a well-defined central hydrophobic cluster (L17-A21) and turn- or bend-like structures (D7-E11 and A21-S26) 
. The data presented here are consistent with these previous observations. Additionally, we have shown a β-turn in the C-terminal region of the peptide. Wuthrich et al.
have also studied the structure of the oxidized form of Met35ox
peptides in aqueous Tris-HCl buffered solutions at pH 6.4–8.2 
. They showed unstructured peptide strands punctuated by turns around S8-V12 and F20-V24 regions. Their 15
H} NOE data showed that the Aβ42
has reduced flexibility at the C-terminus relative to the Aβ40
suggesting insipient structure around this region, consistent with our hypothesis of a beta hairpin in the same region.
How does this structural data compare to previous SS-NMR data studies on fibril structure? Almost all fibrillar studies show a hairpin turn forming somewhere between residues 24 and 30. Examining three of the most recent and notable studies, we find three different sets of amino acids forming the turn location, although it is arguable that this may be attributed to the differences in information content for the various methods and their corresponding sensitivity to structural disorder 
. Ohman's 2006 study of residues 1–42 predict a turn consisting of residues 25–28 or GSNK in the sequence 
. Riek's predicted turn of the same chain is shifted two amino acids towards the N-terminus at residues 27–30 and sequence NKGA 
. Tycko's recent work concerning Aβ40
WT has found a turn at residues V24-N27 with sequence VGSN 
. This turn location is in agreement with our oligomer data. Moreover, our ensemble highly resembles an unconstrained MD study of the same chain, in which the ensemble was clustered and the most populated node was presented 
Our refined ensemble shows a less static ensemble than the SS-NMR fibril studies. The d-Pro induced turn at residues V36-V39 disrupts the inter-chain contacts present in the fibril models by changing the monomer topology. This leaves the VGS turn sequence to stabilize itself exclusively via intra-chain contacts, and we suggest this to be the reason we see greater flexibility in this region.
The AFM data for the mutant Aβ42
Nle35p37 peptide suggest that it forms predominantly low molecular weight species in solution. The 4 nm mean AFM particle height would be in the range of five- to eight-mer complexes by a statistical analysis performed by Lobanov et al.
on the radius of gyration of >3500 protein domains in the SCOP database 
. But notably, the domains in the Lobanov study were compact, folded, α and/or β containing proteins. The likelihood of the Aβ mutant peptide being in a loose, predominantly undefined structure, would possibly reduce the number of monomer units present per complex.
The data presented in this paper indicate the structurally disordered oligomeric assemblages of Aβ42WT and mutant differ in their propensity to form oligomers and fibrils. Aβ42WT peptide formed fibrils at the concentration of 0.4 mM at 10% DMSO/ PBS. The mutant preparation resulted entirely in low molecular weight entities. NMR studies on Aβ42Nle35p37 showed occurrence of two β-turns in the stretches V24-N27 and V36-V39.
Upon mixing Aβ42Nle35p37 mutant with Aβ42WT, Aβ42WT peptide is stabilized in solution suggesting a significant reduction in fibril formation. Presumably such reduced fibril formation is due to the engineered β-turn of the mutant (V36-V39) hindering the formation of the C-terminal β-turn (V24-A30) found in the fibril SS-NMR structure. Although our finding implies the existence of a stabilizing structure for the ADDLs in the mutant peptide, we were not able to detect any known secondary structure stretches, other than the two β-turns, by 1H-NMR and CD spectroscopy. This suggests that β-sheet or α-helix formation is not required for the ADDL stability. Finally, the ability of this mutant to inhibit the aggregation of WT Aβ peptide opens a door to another use for this mutant peptide, since a variant of this peptide or a small molecule peptide mimic could potentially serve as a means to inhibit Aβ aggregation.
How are these results useful in gaining insight into the nature of WT Aβ? While our NMR structural data of the mutant does not directly give structural data regarding the WT, the fact that the mutant mixed with the WT has slowed aggregation suggests that the C terminal beta hairpin presumably stabilized by the mutant does have structural relevance for understanding the nature of the aggregation of WT Aβ. Future work could either use the C terminal beta hairpin motif for small molecule drug discovery in order to find novel small molecule inhibitors of Aβ aggregation.