Long-range Regulation of Partially Folded Amyloidogenic Peptides

Abstract

Neurodegeneration involves abnormal aggregation of intrinsically disordered amyloidogenic peptides (IDPs), usually mediated by hydrophobic protein-protein interactions. There is mounting evidence that formation of α-helical intermediates is an early event during self-assembly of amyloid-β42 (Aβ42) and α-synuclein (αS) IDPs in Alzheimer's and Parkinson's disease pathogenesis, respectively. However, the driving force behind on-pathway molecular assembly of partially folded helical monomers into helical oligomers assembly remains unknown. Here, we employ extensive molecular dynamics simulations to sample the helical conformational sub-spaces of monomeric peptides of both Aβ42 and αS. Our computed free energies, population shifts, and dynamic cross-correlation network analyses reveal a common feature of long-range intra-peptide modulation of partial helical folds of the amyloidogenic central hydrophobic domains via concerted coupling with their charged terminal tails (N-terminus of Aβ42 and C-terminus of αS). The absence of such inter-domain fluctuations in both fully helical and completely unfolded (disordered) states suggests that long-range coupling regulates the dynamicity of partially folded helices, in both Aβ42 and αS peptides. The inter-domain coupling suggests a form of intra-molecular allosteric regulation of the aggregation trigger in partially folded helical monomers. This approach could be applied to study the broad range of amyloidogenic peptides, which could provide a new path to curbing pathogenic aggregation of partially folded conformers into oligomers, by inhibition of sites far from the hydrophobic core.

Figure 1

Helix-turn-helix molecular models solved by experimental NMR structures of (A) Aβ42 in 20% water/80% deuterated-hexafluoroisopropanol (PDB 1IYT) and (B) micelle-bound αS in aqueous solution with sodium dodecyl sulfate (PDB 1XQ8). The amino acid sequences are shown underneath each structure. The regions encompassing Helix 1 and Helix 2 are overlaid with transparent surfaces in both Aβ42 (S8–G25 and K28–G38) and αS (V3–V37 and K45–T92). The colour scheme for the subdomains in the structures and their sequences are, red: N-terminal region, green: C-terminal region, gray: central hydrophobic region (central hydrophobic cluster in Aβ42 or non-amyloid-β component in αS), and blue: turn region.

Figure 2

Free energy surfaces of fraction of helix content (%) and radius of gyration (R_g) for folded (A,C) and partially folded states (B,D) of Aβ42 and αS. The representative conformations corresponding to energy minima basins are marked. The N- and the C-terminus are represented as red and green spheres, respectively.

Figure 3

Computed contact maps (frequency, %) between residues in helically folded, partially folded helical, and unfolded states for (A) Aβ42, and (B) αS. A contact occurs if the minimum average distance between heavy atoms of each residue is within 5 Å (upper left triangle). Contact analyses performed with a 4 Å cut-off (lower right triangle) gave near-identical results. The respective domains in both the peptides are annotated in the map axes.

Figure 4

Consensus dynamic cross-correlation maps (DCCMs) with a correlation filter of 0.4 and above showing positively (red), negatively (blue) correlated and uncorrelated (white) fluctuating regions for the helically folded, partially folded and unfolded states of (A) Aβ42 and (B) αS.

Figure 5

Consensus networks of Aβ42 using a |c_ij | filter of 0.4, showing sub-optimal paths for propagation of disorder from (A) helically folded and (B) unfolded states, for full-length (source D1 – sink V42) and initial helical regions (source S8 – sink G38) in the folded structure. Network structures are overlaid on representative folded and disordered conformations with domains coloured by region: red = NTR, grey = CHC, blue = turn, and green = CTR. Path thickness reflects inter-residue (intra-domain) coupling strengths. Optimal paths are shown below each network, with residues coloured by hydropathy: white = hydrophobic, green = polar, red = negatively charged, and blue = positively charged. Broken lines between two nodes represent coupling including all residues in between.

Figure 6

Consensus correlation networks of αS using a |c_ij | filter of 0.4. Network paths for the (A) helically folded, and the (B) unfolded states. Legend is as for Fig. 5, except domains are coloured as red = NTR, grey = NAC and green = CTR, with different source-sink residue pairs annotating the full-length (source M1 – sink A140) and the initial helical regions (source S8 – sink G38).

Figure 7

Consensus networks of the partially folded helical states of (A) Aβ42, and (B) αS following the same scheme of source-sink residue pairs used in Figs. 5, 6.

Figure 8

Probability distribution of distances from the (A) centre of mass (COM) of D1 – A2 in the NTR or NTR_1-2 to the COM of the hydrophobic CHC (L17 – A21) or CHC_17-21 for Aβ42, and (B) from the COM of the hydrophobic NAC (Q62 – V95) or NAC_62-95 to the COM of Y133 – A140 in the CTR or CTR_133-140 of αS, for folded, partially folded and unfolded states. These regions are shown to be highly correlated in the partially folded helical states (Fig. 7).

Figure 9

Schematic of proposed mechanism of long-range (up to ~20 Å) regulation between the tip of the charged terminus (CT) and the central hydrophobic domain (CHD) of aggregation-prone partially folded helical states. This coupling is lost at distances >20 Å and is absent for the non-aggregating helically folded and the unfolded states.