The Structure of Intrinsically Disordered Peptides Implicated in Amyloid Diseases: Insights from Fully Atomistic Simulations - Computational Modeling of Biological Systems - page 216

advances in NMR techniques are making some headway in this direction [ 12 ].
Examples of natively unfolded aggregating proteins include the Alzheimer amyloid-
“.A“/ protein implicated in Alzheimer's disease, the '-synuclein protein implicated
in Parkinson's disease, and the islet amyloid polypeptide (IAPP) implicated in
type II diabetes [ 13 ]. Numerous nonfibrillar aggregates (soluble oligomers, micellar
species, and amorphous aggregates) that can be on- or off-pathway to fibril
formation, and some of these aggregates that may possess toxic properties, have
been identified but not thoroughly structurally characterized [ 14 - 19 ]. Small soluble
oligomers are difficult to study experimentally, as they correspond to transient,
unstable species. Most experimental techniques do not possess the temporal and
spatial resolution to yield atomistically detailed information about oligomeric
species.
This chapter focuses on the use of fully atomistic simulations to probe the
very initial stage of aggregation of intrinsically disordered proteins: the monomeric
state. Simulations are uniquely poised to probe the structure of natively unfolded
proteins, as they tract individual protein conformations. We focus primarily on two
natively disordered peptides (the A“ peptide [ 20 ] and the IAPP peptide [ 21 , 22 ])
and review recent simulations on these proteins. Although amyloidogenic peptides
(e.g., A“ and IAPP) are defined as “natively disordered” by ensemble-averaging
experimental techniques such as CD and NMR, all-atom simulations actually reveal
that these on-average “natively unfolded” peptides in fact do have some partial
structure. In particular, the simulations that will be presented show that these
peptides either populate a small number of “-rich conformations that could serve
as direct precursors for the formation of amyloid fibrils or contain some structured
elements such as “-hairpin, short helix-coil-helix, salt bridges, and hydrophobic
cluster that may serve as nucleus for folding and oligomerization.

2 Simulation Approaches

The primary simulation technique to study the monomeric conformations of such
peptides is replica exchange molecular dynamics (REMD) simulation. Conventional
Monte Carlo (CMC) and molecular dynamics (CMD) sampling techniques per-
formed under constant temperature condition are prone to getting trapped in local
minima and are not suitable methods for a thorough exploration of conformational
space. The time required to overcome energy barriers grows exponentially with
the barrier height. At physiological temperatures, escape times can easily reach
scales that are inaccessible on current computers (seconds or larger). An incomplete
sampling of conformational space distorts the statistical picture of conformational
ensembles populated under a given set of conditions and can lead to incorrect
conclusions regarding both folding mechanisms and conformational preferences
of the peptides. A number of enhanced sampling schemes have been recently
developed to remedy this sampling problem and facilitate an escape from the
local energy minima. One of the most promising methods is the replica exchange