Biological membranes function as an essential barrier between living cells and their environments. The membrane associated peptides (MAPs) interact with membrane either to facilitate the energy and molecules exchange between the environments and cytoplasm (e.g. cell-penetrating peptide), or to disturb the membrane and cause deadly membrane leakage (e.g. amyloid peptides and antimicrobial peptides). The structures and activity of these peptides are essential to understand the membrane association mechanisms and to screen the drug candidates. However, the poor atomic details of MAPs membrane-bound structures and their complicated interactions with cell membranes leads to the difficulty to better understand their biological roles.
As the structures of MAPs are the prerequisite, in this dissertation, the structure prediction and screening of MAPs were firstly performed in Chapter II, Chapter III, and Chapter IV. We selected amyloid peptides as they usually form complicated polymorphic oligomeric structures, which are the most toxic species. Misfolding and self-assembly of human islet amyloid polypeptide (hIAPP, one of amyloid peptides which belong to MAPs) monomer into polymorphic amyloid oligomers is pathologically linked to type II diabetes. We developed a structures-screening program base on GBMV implicit-solvent evaluation and a structure population evaluation program by Monte Carlo simulation to search the aggregated structures of hIAPP with dominant populations. After the structure search, the stacking-sandwich model and wrapping-cord model were proposed to describe polymorphic structures of hIAPP oligomers, and all-atom molecular dynamics simulations were used to examine the structure, dynamics, and association of the self-assembled hIAPP oligomers. Seven oligomers from the stacking-sandwich model and three oligomers from the wrapping-cord model were determined by their high structural stability with favorable peptide-peptide interactions, although all of them displayed completely different structures in symmetry and beta-sheet packing. These oligomeric structures can also serve as templates to present double- and triple-stranded helical fibrils via peptide elongation, explaining the polymorphism of amyloid oligomers and fibrils.
Base on the predicted oligomeric structures, the mechanisms of amyloid toxicity can be studied. The leaking pore mechanism is more and more widely accepted, in which the amyloid peptides form an ion channel-like but unregulated pore structures. We further investigated the dynamic structures, ion conductivity, and membrane interactions of hIAPP pores in the DOPC bilayer using molecular dynamics simulations (Chapter V and Chapter VI). In the simulated lipid environments, a series of annular-like hIAPP structures with different sizes and topologies were compatible with the doughnut-like images obtained by atomic force microscopy (AFM) and with those of modeled channels for Abeta, K3 peptide, and antimicrobial peptide PG-1, suggesting that loosely-associated beta-structure motifs can be a general feature of toxic, unregulated channels.
Base on the previous works of peptide aggregates, adsoption on membrane/artificial surfaces, and ion-leakage activity, the process how MAPs adsorb on membrane and further penetrate across the membrane is evaluated by the transmembrane potential mean force (PMF). Antimicrobial peptides (AMPs) are selected to be studied due to their simple aggregated structures and short length. We constructed an effect platform including adaptive biasing force (ABF) method which accelerates the membrane penetration process, umbrella sampling method which effectively generates trans-membrane PMFs, and MARTINI coarse-grained force field to measure the free energy required to transfer the AMPs from bulk water phase to water-membrane interface, and further to bilayer interior (Chapter VII). The results implied that biological activity (i.e. antimicrobial or cytolitic activity) appeared to be closely related to the trans-membrane ability indicated by the PMF profiles and the PMFs are instructive index to identify the activity of either existing or designed AMPs. This work provides a useful computational tool to effectively search polymorphic structures of MAPs, to better understand the mechanism and energetics of membrane insertion of MAPs and to rational design new effective MAPs or related inhibitors.
Though the amyloid toxicity leaking pore mechanism was revealed by our simulations, how the pores formed from amyloid monomers or transient oligomers, and how these monomers or oligomers dynamically adsord on membrane and finally imbedded in membrane are still unknown. Due to the complicated components of cell membrane, it is better to simplify the interactions between amyloid-membrane to amyloid-artificial surfaces. Thus, in the last part of the dissertation, we further presented a series of exploratory molecular dynamics (MD) simulations to study the early adsorption and conformational change of Abeta oligomers from dimer to hexamer on three different self-assembled monolayers (SAMs) terminated with -CH3, -OH, and -COOH groups (Chapter VIII). Within the timescale of MD simulations, the conformation, orientation, and adsorption of Abeta oligomers on the SAMs was determined by complex interplay among the size of Abeta oligomers, the surface chemistry of the SAMs, and the structure and dynamics of interfacial waters. Energetic analysis of Abeta adsorption on the SAMs reveals that Abeta adsorption on the SAMs is a net outcome of different competitions between dominant hydrophobic Abeta-CH3-SAM interactions and weak CH3-SAM-water interactions, between dominant electrostatic Abeta-COOH-SAM interactions and strong COOH-SAM-water interactions, and between comparable hydrophobic and electrostatic Abeta-COOH-SAM interactions and strong OH-SAM-water interactions. Atomic force microscopy images also confirmed that all SAMs can induce the adsorption and polymerization of Abeta oligomers. Structural analysis of Abeta oligomers on the SAMs shows a dramatic increase in structural stability and beta-sheet content from dimer to trimer, suggesting that Abeta trimer could act as seeds for Abeta polymerization on the SAMs.