Deciphering the molecular signatures of neurodegeneration by predictive computational modelling

Amyloids are fibrillary aggregates identified in over 40 human diseases, including neurodegenerative diseases encompassing Alzheimer’s (AD), Parkinson’s (PD), Huntington’s and Prion diseases. Amyloids form by spontaneous self-assembly of monomeric precursor peptides known as intrinsically disordered proteins (IDPs). Experiments suggest that soluble low molecular weight oligomers formed in the early stages of assembly are toxic, and hence, most promising drug targets. However, experiments are insufficient to characterize oligomers due to their inherent polymorphic and short-lived nature. This thesis advances our mechanistic understanding of the formation of amyloid oligomers by delineating signature features of IDP monomers and ‘profibrillar’ oligomers through predictive computational modelling techniques employing atomic resolution molecular dynamics (MD) computer simulations. We additionally predict the assembly of non-aggregating low molecular weight oligomers. We first probe the molecular signatures of experimentally indicative non-aggregating folded a-helical conformers, and aggregation-prone partially folded a-helices of amyloid-b42 (Ab42) and a-synuclein (aS) IDPs implicated in AD and PD, respectively across a broad spectrum of physical models. We predict a common intra-peptide route to helix stabilization, showing that the terminal groups (N-terminal or NTR in Ab42 and C terminal or CTR in aS) frequently indulge in hydrophobic interactions with the central hydrophobic domains (CHDs) and secondary salt bridges with other domains. Lack of such short-range contacts during complete helix unfolding coupled with destabilized helices in terminal-deleted variants confer the aggregation protective role by terminal groups in folded helical conformers. Further, we reveal a shared feature of dynamic coupling between the partially folded helical regions of the CHD and the charged terminal ends (NTR in Ab42 and CTR in aS). Absence of such intra-peptide modulation in helically folded and unfolded states confer long-range allosteric regulation of the CHD by the termini that may render the partially folded helical states prone to initial oligomerization. Next, we design structural assemblies of experimentally uncharacterized aggregation-resistant low-weight aS tetramer. We model a de novo broken a-helical tetramer by reconstructing loop motif that optimizes packing of aS helical monomers. We show that monomers attain activated conformations during tetramer assembly, and familial missense mutations double the energy barrier to tetramerization, thus preserving the pool of aggregation-prone disordered monomers, and confirming the experimentally observed low tetramer:monomer ratios with mutants. In order to investigate the effect of helical continuity and periodicity, we model a de novo extended 11/3-helical tetramer. Broken a-helical tetramers show a more favourable assembly than the extended 11/3- helical tetramers, the ease of their interconversions diminishing with homologous E → K mutations. Additionally, rationally designing a series of broken a-helical multimers from dimers to octamers shows that tetramers have lowest activation energy, providing a rationale for the experimental observation that tetramers are the most populated oligomers. Finally, we investigate the molecular determinants of higher aggregation rate of Ab42 over Ab40 by simulating their profibrillar oligomers (dodecamers) on graphene water interface. Our data reveals that Ab dodecamers may facilitate a single layer growth along the graphene surface, with Ab42 presenting a more closed conformation with possibilities of unidirectional growth in Ab40, but not in Ab42. Oligomer height profiles on graphene indicate that dodecamers may be formed post mature fibril formation