Sei sulla pagina 1di 27

PC66CH28-Shea ARI 28 February 2015 14:17

ANNUAL
REVIEWS Further Computational Studies of
Click here for quick links to
Annual Reviews content online,
including:
Protein Aggregation: Methods
and Applications
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Other articles in this volume


Top cited articles
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

Top downloaded articles


Our comprehensive search
Alex Morriss-Andrews1 and Joan-Emma Shea1,2
1
Department of Physics and 2 Department of Chemistry, University of California, Santa Barbara,
California 93106; email: shea@chem.ucsb.edu

Annu. Rev. Phys. Chem. 2015. 66:64366 Keywords


First published online as a Review in Advance on amyloid brils, coarse-grained models, molecular dynamics simulations,
February 2, 2015
replica exchange molecular dynamics, enhanced sampling methods,
The Annual Review of Physical Chemistry is online at systematic coarse graining
physchem.annualreviews.org

This articles doi: Abstract


10.1146/annurev-physchem-040513-103738
Protein aggregation involves the self-assembly of normally soluble proteins
Copyright  c 2015 by Annual Reviews. into large supramolecular assemblies. The typical end product of aggregation
All rights reserved
is the amyloid bril, an extended structure enriched in -sheet content. The
aggregation process has been linked to a number of diseases, most notably
Alzheimers disease, but bril formation can also play a functional role in
certain organisms. This review focuses on theoretical studies of the process of
bril formation, with an emphasis on the computational models and methods
commonly used to tackle this problem.

643
PC66CH28-Shea ARI 28 February 2015 14:17

1. BACKGROUND
The computational study of protein aggregation has established itself as a mature and prolic
eld of research, with over 10,000 publications, dating back to the 1970s. The eld, however, has
recently burgeoned, propelled by greater computational resources, the application of enhanced
sampling algorithms, and the development of novel coarse-grained models. This review focuses
on these latest developments. We begin with an overview of protein aggregation, followed by
a description of computational models and methodologies, with a few selected applications as
illustrations.
Proteins are polymers of amino acids, synthesized on ribosomes and released as extended
chains. A class known as globular proteins folds to a specic three-dimensional structure, either
on their own or with the help of chaperone molecules. This folded state corresponds to the bio-
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

logically active, functional state. Other proteins are intrinsically disordered and only populate a
functional state once bound to a partner molecule. Proteins exist in a crowded, heterogeneous cel-
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

lular environment that can dramatically affect folding and association between proteins. Changes
in cellular condition (pH, temperature) or changes in the protein (mutation, posttranslational
modication, overexpression) can lead to misfolding or partial unfolding of a protein and subse-
quent self-assembly into aggregate structures (1). The aggregation process is often considered a
pathological one that depletes active proteins, and the formation of potentially toxic aggregate
species can indeed be harmful to the cell. Several diseases, including Alzheimers, Parkinsons,
and some forms of cancer, are closely linked to protein aggregation (2). Yet it is noteworthy that
aggregation is not problematic in all instances: Several organisms use it for functional purposes
(e.g., biolms in bacteria) (3).
The common end product of aggregation, seen in both pathological and physiological aggre-
gation processes, is the extended amyloid bril (100 nm long), highly enriched in content,
with a cross- structure. The latter involves an arrangement of sheets running parallel to the
bril axis, with perpendicular hydrogen bonds (4, 5) (see Figure 1).

a b

Figure 1
(a) Twisted morphology of a TTR(105115) bril. (b) Close-up view showing the molecular detail. Figure adapted from Reference 4,
with images created using the following structures: Electron Microscopy Data Bank, EMDB accession number EMD-2324, and
Protein Data Bank, http://www.pdb.org (PDB ID code 2m5n).

644 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

Mature
fibrils
Aggregation

Growth

Elongation or
Protofibrils dissociation at
Fragmentation
fibril end
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Secondary Association
nucleation
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

Native Small
state Nucleus
oligomers
Partially
unfolded
monomer

Nucleation Time

Figure 2
Schematic illustration of the bril nucleation growth mechanism, including secondary nucleation and
fragmentation processes.

Experimental studies of protein aggregation indicate a sigmoidal trace for the kinetics of bril
formation, as shown schematically in Figure 2. This sigmoidal shape has traditionally been as-
cribed to a standard nucleation growth mechanism, in which a partially folded monomer associates
with others to form a critical nucleus (the nucleation phase), at which point a small bril emerges
and elongates (the growth phase). The primary growth process is often attributed to bril-end
elongation by a dock-lock mechanism, in which the monomer rst binds to the edge of the growing
bril (the dock phase) and then rearranges its structure once bound (the lock phase). The process
is more complex in reality, with the formation of not only a host of on- and off-pathway oligomers,
but also secondary processes such as lateral growth, fragmentation, and association (69).
Protein aggregation is an attractive eld to theorists because theoretical and computational
challenges are coupled to a problem of real biological importance. The process of aggregation
involves length scales of one to hundreds of nanometers and timescales that can exceed hours
(Figure 3). As a result, the study of protein aggregation lends itself to a hierarchy of models, from
the quantum mechanical to the mesoscopic (Figure 4). This review focuses primarily on classical
molecular dynamics studies of atomistic and coarse-grained models (Figure 5).

2. ATOMISTIC MODELS
Atomistic models of both the protein and solvent offer the most detail but come at a large com-
putational cost. They are used primarily to study monomeric and very small oligomeric com-
plexes, as well as the stability of preformed bril models and their interaction with dyes or small
molecule or peptide inhibitors. Typically, the study of monomers and small oligomers needs to

www.annualreviews.org Computational Studies of Protein Aggregation 645


PC66CH28-Shea ARI 28 February 2015 14:17

PHE Protein
aggregation

Side-chain
rotations

1 ps 1 year

1 ns 1 month
Loop closure
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

1 s 1h
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

1 ms 1 min
1s Protein
folding
Helix
formation

Folding of hairpins

Figure 3
Illustration showing the contrasting breadth of timescales of protein rearrangement and assembly, from fast
side-chain rotations to slow protein aggregation.

be augmented by enhanced sampling methods, such as the replica exchange molecular dynamics
(REMD) and metadynamics methods described in Section 4.2. The convergence of simulations
is often problematic, with small proteins of 20 amino acids requiring on the order of several
hundred nanoseconds per replica (27).
Atomistic simulations have provided important insights into the structure of the early stages
of aggregation, at a resolution that surpasses experimental capabilities. The initial partially folded
aggregate-prone structure and prenucleus assemblies are transient, unstable species, difcult to
detect experimentally. Simulations have been particularly instrumental for the case of intrinsically
disordered peptides, capturing the transient secondary structure, which may provide clues about
the protein regions responsible for initiating aggregation. Not only have simulations been able
to study protein fragments, they also are now at the stage at which they can tackle full-length
proteins implicated in amyloid diseases, including the 4042-residue-long amyloid- (A) peptide
linked to Alzheimers disease and the islet amyloid polypeptide (IAPP) associated with type II
diabetes.
A powerful approach combines molecular dynamics simulations (typically with the replica
exchange sampling protocol described in Section 4.2) with nuclear magnetic resonance (NMR)
(molecular dynamics/NMR methodology). With this approach, important new information has
been obtained regarding structural differences arising at the monomeric level between the A40
and A42 alloforms. Experiments have shown that these peptides, which differ uniquely by the

646 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

Length

Continuum model

mm

Coarse-grained model

m Atomistic model
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

QM model
nm

Time

ps ns s ms s

Figure 4
The approximate timescales involved in different classes of molecular simulations: quantum mechanical
(QM) (10), atomistic (1114), coarse-grained (1517), and continuum models (18).

presence of two additional residues at the C terminus of the peptide, aggregate via different path-
ways (28, 29). Simulations were instrumental in complementing experimental studies by showing
that, although the peptides were considered intrinsically disordered, populating an ensemble of
diverse conformations, regions of local secondary structure could be identied (25, 30). In par-
ticular, simulations by Garcia and coworkers (25) demonstrated that both A40 and A42 have a
bend motif at residues V240K28, located near the loop region in the strand-loop-strand struc-
ture of the bril, which has been suggested as a nucleation site for monomer folding (31, 32)
(see Figure 5e). This bend is stabilized by a salt bridge between residues E22 and K28, which
is notable in light of familial Alzheimers mutations involving the E22 residue (33). Residues be-
longing to the central hydrophobic core (L17A21, a highly aggregate-prone region that forms
brils if excised from the protein) and to the I31V36 region adopt a -strand structure in both
A40 and A42. These same regions are found in a -strand structure in the context of the bril.
A42 further populates a hairpin in the C-terminal region (V39I41), also seen in simulations of
isolated fragments of the terminus (34). These simulations suggest that modeling the monomeric
structure, and identifying regions of transient secondary structure, can provide important clues
about the aggregation pathways and the role of point mutations in modulating aggregation. The
most important result from these simulations is the identication of possible aggregation-prone
structures among a diverse family of existing structures.
Similarly, in the case of the IAPP peptide, simulations on aggregating and nonaggregating
forms of the peptide have revealed signicant differences in monomeric structure, with aggre-
gating variants (e.g., human IAPP) populating both compact and extended conformations, and
nonaggregating variants (e.g., rat IAPP) exhibiting only compact structures (3537). Structures
generated from REMD or metadynamics simulations can serve as starting points for further sim-
ulations of the interaction of inhibitor molecules (e.g., EGCG) with amyloidogenic peptides.

www.annualreviews.org Computational Studies of Protein Aggregation 647


PC66CH28-Shea ARI 28 February 2015 14:17

a b c d e
N*
MARTINI MARTINI
water A40 A42
si
p i
b i

PRIME Monomer

CG
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

C(+) P P A()
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

X
OPEP PHE
X X ALA
Y Y
X
Y Y
X
Y Y YX E Binding of
E X PIB to fibrils
H H H
VAL
LYS

Simple Phenomenological Systematic Higher-resolution Atomistic


models models coarse graining models models

Figure 5
Different resolution models for the study of protein aggregation, from coarse grained to atomistic. (a) Simple models: the orientable
stick model (19) and the sphero-cylindrical model (20). (b) Phenomenological models: the lattice model (21), Caisch model (22), and
Shea model (23). (c) Systematic coarse graining: a coarse-grained polyalanine chain (24). (d ) High-resolution models: the MARTINI
model (15), PRIME model (17), and OPEP model (16). (e) Atomistic models of the Alzheimer amyloid- peptide: monomers from
replica exchange molecular dynamics simulations, adapted from Reference 25, and PIB bound to brils, adapted from Reference 26.

It is important to note that different force elds can lead to somewhat different secondary
structure predictions, particularly in the case of intrinsically disordered peptides, such as A
and IAPP. Most force elds have been parameterized on the basis of folded motifs and may need
reoptimization to better account for the unfolded/partially folded nature of intrinsically disordered
peptides.
In addition to probing the early stages of aggregation, atomistic simulations have also been
instrumental in studying the structural characteristics of brils. Starting with coordinates obtained
from solid-state NMR, investigators have proposed and rened models for amyloid brils of A
and IAPP (3841). These bril structures have been used to gain insight into the binding and mode
of action of amyloid dyes and small molecule inhibitors (42, 43). For instance, in simulations,
thioavin T (ThT) and its derivative PIB would recognize and bind to the hydrophobic and
aromatic grooves formed on the -sheet surface of A40 and A42 brils, and in the case of
A42, there is an additional binding mode in the loop region of the bril (see Figure 5e) (26).
Further simulations with Congo Red revealed a new binding site, not seen for ThT and PIB, in
which the molecule bound to the edge of the bril (44), as seen in simulations involving anti-
inammatory drugs (45), possibly explaining the inhibitory role of Congo Red in blocking bril
extension.
Atomistic simulations have also probed the growth phase of bril formation (see Figure 2).
For instance, Bolhuis and coworkers (46) used transition path sampling (discussed in Section 4)
to study the mechanism of bril elongation. Their simulations conrmed the proposed dock-lock
mechanism (47), showing that the docked state was indeed an intermediate on the elongation

648 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

pathway and that the lock process could proceed by more than one pathway (either by the ini-
tial formation of hydrogen bonds, followed by side-chain reorientation, or vice versa). Another
example of the use of atomistic simulations to study elongation can be found in the work of Wang
and coworkers (48), who used REMD simulations to show that brils with a cross- structure
could be a template for the formation of additional structure in A monomers.

3. COARSE-GRAINED MODELS
The coarse-graining technique is well suited to the study of peptide aggregation. First, the min-
imum timescales and length scales required to capture the aggregation process are in general
signicantly higher than those for protein folding (Figures 3 and 4). Second, although atomistic
simulations can capture the initial stages of aggregation (up to roughly tetramers) and model pre-
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

formed small brils, the full assembly process from monomers to the bril is beyond the current
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

computational reach of all-atom simulations. Third, the structural similarity between amyloid
brils composed of different peptides seems to imply a degree of universality in the mechanism of
bril formation (6), lending support to the use of simplied peptide models that omit some molec-
ular details yet retain the essential physical elements governing aggregation. Fourth, the technique
works well for studying the interaction between peptide aggregation and other biomolecules in
the cellular milieu, such as membranes (4954) and other lipid structures (22, 55, 56). Finally,
these coarse-grained simulations are ideal to study systems that use a solid surface as a substrate
on which aggregates adsorb, a setup common in many experiments (e.g., atomic force microscopy)
(5764).
A wide spectrum of coarse graining is possible (65, 66). Some models are very lightly coarse
grained, keeping atomistic resolution for the backbone but coarse graining the side chain. At the
other extreme, coarse graining can be done on the molecular scale and beyond.

3.1. Lower-Resolution Models


Coarse graining comes with the trade-off of accuracy and computational efciency, and the degree
of coarse graining should depend on the smallest important length scale of the system of study. In
practice, however, it is nontrivial to determine a priori whether the small coarse-grained length
scales affect the physics of aggregation. This difculty is especially acute in aggregation because
of the extreme range of length scales and timescales involved (Figure 3).
Highly coarse-grained models sacrice sequence-level resolution. Representative models are
shown in Figure 5a. For instance, Barz & Urbanc (67) studied the general properties of aggrega-
tion using a minimalistic tetrahedral model with varying numbers of hydrophobic and hydrophilic
beads, whereas Auer et al. (68) developed a tube model, and Zhang & Muthukumar (69) devel-
oped a simple cuboid model capable of capturing the essence of the nucleation growth mechanism.
Wallin and colleagues (19) simulated more than 105 peptides, represented by sticks (with dened
orientations about their axes) placed on a cubic lattice, using Monte Carlo simulations. They ob-
served a nucleation growth mechanism in which the brils needed to form a sufcient number of
vertical layers before they could extend longitudinally. Low-resolution models have also been em-
ployed in conjunction with a dynamic Monte Carlo method, designed to capture kinetics within
the Monte Carlo framework by making the moves sufciently small that each one is physical.
Vacha and colleagues (20) studied a generic two-state peptide: one representing a soluble random
coil and the other a bril-prone -sheet conformation. They used an implicit water model and
simulated 600 peptides at concentrations ranging from 0.2 to 8.0 mM. Their system exhibited
brillar self-assembly after the formation of a critical nucleus, which they determined to be a

www.annualreviews.org Computational Studies of Protein Aggregation 649


PC66CH28-Shea ARI 28 February 2015 14:17

partially converted mixed aggregate. The aggregates conversion to a full -sheet structure was a
multistep process.
Efforts in coarse graining extend beyond the molecular length scale. Buehler and colleagues
(70) developed a very highly coarse-grained model that represents the bril itself as a chain of
coarse-grained beads. Their mesoscale model is designed to study the self-assembly of these
strands. Elastic parameters are obtained from implicit water all-atom simulations and used to
parameterize the coarse-grained model. The authors looked at the plaque assembly of 240 brils,
determining that, for sufcient length, adhesion forces between brils induce bending, which can
generate entangled/disordered plaques, ring-like geometries, and self-folded brils. Conversely,
shorter brils form ordered, rigid assemblies.
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

3.2. Phenomenological Models


Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

A higher level of resolution than is achievable with the models presented in the previous section
involves coarse graining over atomic length scales, with one or more beads representing an amino
acid. These simple physics-based models have been quite successful in elucidating which of a
proteins physical properties play a key role in the aggregation process. Instead of representing
the full spectrum of amino acid residues, these models typically use generic amino acid types, such
as charged, polar, hydrophobic, or neutral.
One of the simplest phenomenological models is the lattice model, which restricts the allowed
coordinates to a cubic lattice and typically employs a low level of resolution (a single bead) (71), al-
though it also allows for a multibead description of the amino acid (72). Simple lattice models, such
as the work by Li and coworkers (21), were capable of identifying aggregate-prone conformations
(the N conformation shown in Figure 5b).
Phenomenological models can also be off-lattice, of low to mid-resolution. Examples include
the Caisch model (22, 73, 74) (two beads per residue) and the Shea model (23, 75) (three beads per
residue). These phenomenological models focus on how the peptide -sheet propensity affects the
kinetics of bril formation from dimers to longer brils of tens of peptides. As initial congurations
often involve peptides scattered over a volume of thousands of cubed nanometers, they typically
employ implicit solvent models to avoid lling such a volume with explicit water.
The Caisch model (Figure 5b) represents the peptide as possessing one of two possible
congurations, a folded conformation and a -competent conformation, with a dihedral potential
designed to bias both to different degrees. The -sheet propensity parameter in this model controls
the depth of the potential energy well of the folded conformation (76). With this model, the most
amyloidogenic proteins exhibit features that are distinct from those of less amyloidogenic ones
(73). Fibril formation occurs rapidly along a single pathway following a smaller nucleus, without
the formation of intermediates such as micelles or protobrils. The bril growth rate depends far
more strongly on concentration. Any polymorphism in the brils is determined by external con-
ditions rather than being under kinetic control (77). Furthermore, brils are found to be cytotoxic
only during their growth phase. Fibril formation is accelerated by membranes and is not signif-
icantly decelerated by surfactants or accelerated by macromolecular crowding (22, 78). Proteins
falling into the highly amyloidogenic category include Phe-Phe, GNNQQNY, transthyretin,
and A40 . More weakly amyloidogenic proteins include A42 , Sup35, prion proteins, and
myoglobin.
The Shea peptide model uses three beads per residue: two for the backbone and one for the
side chain (Figure 5b). Similar to the Caisch model, the Shea model controls -sheet propensity
via a backbone dihedral potential. However, instead of modulating the relative well depths of -
competent and -protected conformations, it controls the resistance to backbone torques against

650 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

deviations from the preferred off-trans conguration of the side chains. Stiffer peptides are shown
to be more prone (23). Both models were able to distinguish between different aggregation
pathways, with peptides with high -sheet propensity forming brils via an ordered -sheet
nucleus, whereas peptides of lower -sheet propensity rst formed disordered oligomers from
which the structure then emerged. In addition to their use in studying the nucleation step, these
models could also be used in seeding simulations to study bril growth. An important outcome from
these simulations is the observation of both bril elongation (standard growth mechanisms) and
lateral bril growth (secondary growth mechanism) (75, 7981). The Shea group (49) studied the
aggregation of 32 short -sheet-prone peptides on the hydrophilic surface of a bilayer comprising
648 lipids in an implicit solvent. This work combines a coarse-grained peptide amyloid model (23)
with the Brannigan-Brown coarse-grained lipid bilayer model (82). They found that, similar to an
attractive solid surface (83), the membrane was biased toward -sheet morphologies. However,
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

unlike a solid surface, membrane undulations increased dynamic transitions between aggregate
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

structures and disrupted the formation of multiple bril layers parallel to the surface. Additionally,
the authors observed several effects on the membrane: reduced uctuations, increased bending
modulus, and a local ordering of lipid head groups to conform to optimal packing with the brils
hydrophilic residues. These effects were locally constrained to the position of the brils and did
not seed a large-scale phase transition in the membrane.

3.3. MARTINI Force Field


More detailed representations of the protein are necessary to capture specic chemical properties.
The MARTINI force eld is the most widely used coarse-grained atom model at this resolution,
combining groups of (on average) four heavy atoms as a single bead (15, 51, 8489) (Figure 5d ).
The MARTINI water model similarly maps four water molecules per bead. For bonded interac-
tions, it matches to an all-atom reference, but nonbonded interactions are chosen based on the
chemistry of the united atoms (i.e., their charge, polarity, and hydrogen bonding capabilities) (90).
By selecting nonbonded potentials based on identiable chemical properties rather than system-
specic optimization to all-atom simulations (systematic coarse graining), the model is transferable
to many different molecules, including water, lipids, proteins, and (a work in progress) nucleic
acids. A known issue of the MARTINI model with respect to protein aggregation is its neglect
in the detail of the peptide backbone (90). The Tieleman group (85) has recently improved the
MARTINI potentials to better capture amyloid assembly.
Examples of the use of the MARTINI model to study bril formation include work by
Schitt and colleagues (88), who explored the effects of longitudinal versus lateral growth. They
employed a coarse-grained MARTINI model to represent 27 protobril fragments of amylin
(SNNFGAILSS), with each protobril consisting of 20 peptides (2029). The simulation fol-
lowed the merging of these protobrils. They found that the growth mechanism was temperature
dependent, preferring elongation over lateral growth at lower temperatures, a preference lost as
the temperature increased. Schatz and colleagues (86) simulated the self-assembly of peptide am-
phiphiles into bers, running a system of 400 peptides for 16 s using the MARTINI model in
explicit water. They found that the system rst formed spherical micelles, which arranged into
a three-dimensional network of micelles held by van der Waals interactions. The micelles then
merged to form an extended ber. In a high-throughput study, Tuttle and colleagues (89) used the
MARTINI model in explicit water to test the aggregation propensity of 400 different dipeptide
combinations. Each test consisted of 300 dipeptides run for 100 ns. Systems exhibiting a strong
aggregation propensity were then selected for more extensive simulations. This method can be
useful to rapidly nd aggregation-prone sequences.

www.annualreviews.org Computational Studies of Protein Aggregation 651


PC66CH28-Shea ARI 28 February 2015 14:17

Several recent computational studies have looked at the effect of membranes on protein aggre-
gation using coarse-grained models. Using a MARTINI model with 16 proteins, 7,000 lipids, and
coarse-grained water, Sansom and colleagues (54) showed that the morphology of transmembrane
protein aggregates depended on several factors. Hydrophobic mismatch (i.e., the size of the pro-
teins hydrophobic region relative to the thickness of the membranes hydrophobic core) can drive
protein aggregation. The protein class (helix versus barrel) and membrane curvature also affect
the aggregate morphology. Li & Gorfe (51) conducted a MARTINI model simulation with 32
H-Ras proteins aggregating on a lipid bilayer (7,320 lipids). Coarse-grained simulations of pep-
tide aggregation on small lipid micelles, comparable in size to the aggregates themselves, were
conducted by Hung & Yarovsky (56). These simulations used a MARTINI model with 125 lipids
and 27 apoC-II(6070) peptides, and 20,000 water beads. They showed a substantial reduction in
the aggregation rate for free lipids compared to bulk aggregation. The aggregate morphology was
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

strongly dependent on the local lipid environment: Greater hydrophobic contact with the lipids
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

resulted in elongated aggregate structures. Additionally, the presence of peptides disrupted lipid
assembly. Tieleman and colleagues (85) looked at the aggregation of 1, 8, and 64 octapeptides
[SNNFGAIL and (GV)4 ] at an explicit water-octane interface using an extension of the MAR-
TINI model. They found more extended morphologies at the interface than they observed in bulk
water. Adsorption was rapid, forming stretched conformers resembling strands.

3.4. The PRIME Model and Discontinuous Molecular Dynamics


A similar resolution model, PRIME (and the more recent PRIME20), has been employed exten-
sively by Hall and colleagues (17, 9196) to study the kinetics of bril formation (Figure 5d ).
They used discontinuous molecular dynamics (DMD), which allows for discontinuous breaks in
the energy functional by computing the reection/transmission of particles across the discontinu-
ity. PRIME reduces the number of particles to four per amino acid and is able to effectively capture
the energetics of hydrogen bonding using an efcient directional square-well hydrogen bond. The
parameterization philosophy for PRIME is to reduce the number of interaction parameters, such
that each parameter is physically meaningful, but to retain structural discrimination (17).
Hall and colleagues (91) studied the aggregation kinetics of 192 peptides (sequence KA14 K)
into a single bril, averaged over 30 independent simulations. Fibrils for this system formed by
the intermediate of amorphous aggregates that, given sufcient size, spontaneously ordered into
sheets (i.e., the nucleation growth mechanism), which subsequently extended by the addition of
monomers to the ends, one at a time (templated assembly). They also simulated the brillization of
48 tau fragment peptides (sequence VQIVYK) at varying temperatures (92). They contrasted two
hydrogen bond constraints: one favoring parallel sheets and one with no bias to parallel versus
antiparallel. The parallel-biased constraints were more consistent with X-ray crystallography (97).
At sufciently high temperatures, brils formed quickly by a templating mechanism.
Further simulations with the PRIME20 model were performed on seven different sequences,
each system using 48 peptides (93). Their brillization propensity agreed with experimental results
(98, 99) at sufcient temperature. Hall et al. hypothesized that the brillization transition temper-
ature in their model can predict a sequences bril propensity. In PRIME simulations, they studied
the bril formation of palindromic sequences: Syrian hamster prion protein SHaPrP(113120)
(AGAAAAGA), mouse prion protein MoPrP(111120) (VAGAAAAGAV), and eight variations on
these (94). Their analysis of how kinetic events inuence morphology demonstrated how sequences
with long stretches of hydrophobic residues decrease brillar order in preference of a disordered,
collapsed state. The Hall group (95) also employed the PRIME model to study the formation of
twisted A(1622) brils, implicated in Alzheimers disease. They found a temperature-dependent

652 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

mechanism similar to their later tau fragment study (i.e., a nucleation growth mechanism at low
temperature and templated assembly at high temperature) (92). They observed the formation of
structural details (e.g., intersheet distance, antiparallel sheets, side-chain interdigitation) con-
sistent with experiments (97, 100102).
The Hall group is not the only group to use DMD to study peptide aggregation. In a recent
study, Auer and colleagues (103) contrasted the contributions of kinetics and thermodynamics in
protein aggregation. They represented the proteins by a chain of hard spheres centered on the
C atoms, with sequence-dependent hydrogen bonding (104). They simulated 125 12-residue
peptides using DMD with an implicit solvent. They determined that kinetics were essential for
aggregate formation and suggested that kinetics allow amyloidogenic proteins to fold into their
native, aggregation-immune state, even when these simulations give the bril conformation as
being more thermodynamically stable. Urbanc and colleagues (105107) conducted a number of
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

recent studies on the oligomerization of A using coarse-grained DMD simulations, as well as


Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

highly coarse-grained DMD simulations of generic properties of aggregation using a tetrahedral


protein model (67). Additionally, Dokholyan et al. employed DMD simulations to study the ag-
gregation of A with a coarse-grained model with two beads per residue (108) and the aggregation
(109) and dimerization (110) of SOD1 with a two-bead and atomistic model, respectively.

3.5. The OPEP Model


OPEP is a high-resolution coarse-grained model that has a high degree of chemical specicity
without imposing articial constraints on the secondary structure (16) (Figure 5d ). Although
it represents the side chain as a single coarse-grained bead, the backbone is given full atomic
resolution for the heavy atoms.
Mousseau and colleagues (111) studied the dimerization of various alloforms of A using OPEP
with a combined temperature and Hamiltonian replica exchange methodology. They showed that
the dimerization propensity of the peptides is strongly affected by the Ile41 and Ala42 amino
acids and the salt bridge located at D23K28. Similarly, Derreumaux and colleagues looked at the
equilibrium structures of A(2535) trimers and hexamers (112) and have conducted studies of
the trimer structures of A(1742) (113) using the OPEP model with REMD. Additionally, they
saw how the latter structure is affected by ve different small molecule inhibitors. Their results
showed multiple binding modes of the drugs to the trimer and suggested that different drugs may
have varying efcacy at different stages of oligomerization.
The OPEP force eld was also used to study the early stages of oligomerization of the NNQQ
and GNNQQNY peptides derived from the yeast prion Sup35 (114, 115). For the GNNQQNY
peptide, Mousseau and colleagues (115) employed REMD with system sizes up to 20-mers. Larger
brils were able to form transiently but destabilized into globular/disordered forms. The authors
predicted a high degree of polymorphism of the sequence. Using conventional molecular dynamics
simulations, they also analyzed the oligomerization kinetics of the same sequence (116). They
found kinetics mostly consistent with classical nucleation theory, with a critical nucleus of four to
ve monomers. However, deviations from this theory occurred via rearrangements of the structure
postnucleation. They observed signicant polymorphism in the 20-mer.

4. DEVELOPMENTS IN COMPUTATIONAL METHODS


Table 1 presents a selection of computational techniques that have been (or could be) applied
to the study of protein aggregation. We discuss some of these techniques in more detail in the
subsequent sections.

www.annualreviews.org Computational Studies of Protein Aggregation 653


PC66CH28-Shea ARI 28 February 2015 14:17

Table 1 Selection of computational techniques for the study of protein aggregation


Computational
technique Method Description Reference(s)
Thermodynamic Metadynamics Enhanced sampling of free energy landscape using Method (117);
enhanced sampling collective variables; adaptive method; kinetics are lost applications
methods (118121)
Umbrella sampling Enhanced sampling of free energy landscape using Method (122);
collective variables; kinetics are lost applications
(123, 124)
Parallel tempering Enhanced sampling of free energy landscape using runs Method (125);
(temperature REMD) at multiple temperatures; kinetics are lost applications
(83, 112115)
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Replica exchange Enhanced sampling of free energy landscape using Method (125, 126);
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

molecular dynamics multiple temperatures (parallel tempering) and/or applications (83,


(REMD) varying Hamiltonian parameters; kinetics are lost 111115, 127, 128)
Replica exchange Replica minimization required for good sampling in Method and
statistical temperature REMD; adaptive method applications
molecular dynamics (129, 130)
Replica exchange Monte As for REMD, but for Monte Carlo sampling Method (131);
Carlo applications (132)
Kinetic enhanced Markov state model Enhanced sampling of kinetics using collective variables; Method (133, 134);
sampling methods launches short trajectories from congurations applications
requiring the greatest sampling; adaptive method (135, 136)
Free energy guided Enhanced sampling of kinetics using approximated free (137)
sampling energy surface; launches short trajectories in parallel
with Boltzmann distribution; adaptive method
WExplore Enhanced sampling of kinetics using multiple replicas in (160)
dynamically assigned tessellations of conguration
space; clones/merges replicas as needed; collective
variable method; adaptive method
Transition path Rare-event sampling method, well suited to study (138140)
sampling processes, such as protein folding and aggregation, that
have complex underlying energy landscapes; involves
generating a set of reactive trajectories between an
initial and nal state, typically followed by likelihood
maximization to obtain optimized reaction coordinates
String method Free energy landscape method to nd lowest barrier (141, 142)
path (mountain pass) between two states; parallel
runs are tied together in a string in state space to
prevent them from merging/separating; the string
converges to the transition path between the ends
Data/trajectory Secondary nucleation Analysis of trajectory/experimental brillization kinetic (2, 8, 9, 143)
analysis methods kinetic analysis data; involves tting to a master equation, including
secondary pathway terms
Normal mode analysis Trajectory analysis method using elastic network Method (144, 145);
models; used to study the low-frequency collective applications
motions of biomolecules (146, 147)
(Continued )

654 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

Table 1 (Continued)
Computational
technique Method Description Reference(s)
Systematic coarse Relative entropy coarse Systematic coarse-graining method matching Method (148);
graining graining coarse-graining to all-atom congurational probability applications (24)
distributions for minimal information loss
Multiscale coarse Systematic coarse-graining method matching Method (149);
graining coarse-graining to all-atom momenta applications
(55, 150)
Iterative Boltzmann Systematic coarse-graining method matching Method (151);
inversion coarse-graining to all-atom Boltzmann distributions applications (152)
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Fast molecular Discontinuous molecular Molecular dynamics integrator allowing for Method (153);
dynamics dynamics discontinuous potentials applications
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

integrator (17, 9195, 103)

4.1. Systematic Coarse Graining


Systematic coarse graining provides a bottom-up methodology to obtain the optimum coarse-
grained potentials matching the behavior of all-atom simulations or experimental data. Recent
developments of this technique include the relative entropy method by Shell (148), the multiscale
coarse-graining method by Izvekov & Voth (149), and the iterative Boltzmann inversion method
by Muller-Plathe and colleagues (151). The relative entropy method uses information theory to
nd the potentials with minimal information loss in the congurational ensembles (148). The rel-
ative entropy is dened as the sum over the conguration space of Srel = PT ln(PT /PM ), where PT
is the probability of this conguration in the target (all-atom) ensemble, and PM is its probability
in the model (coarse-grained) ensemble. The relative entropy method minimizes Srel . Multiscale
coarse graining is a variational technique that best matches all-atom momenta in the coarse-
grained sites (149). The goal of this method is to minimize (in force eld parameter space) the
square of the difference between the momenta of the reference force eld and coarse-grained force
eld (154). The method accomplishes this by breaking the coarse-grained force eld into long-
range (Coulomb) and short-range (to be optimized) components. The short-ranged component
is approximated as a polynomial spline, which allows for an efcient least-squares optimization by
expressing it as an overdetermined system of linear equations. The iterative Boltzmann inversion
method is designed to reproduce all-atom Boltzmann statistics (151). The procedure begins with
an initial guess of the potential and a set of collective variables used to compare the reference
and coarse-grained force elds. For each probability distribution function p(n ), where n
denes the conguration of the system (e.g., bond lengths, dihedral angles), the estimated free
energy of the coarse-grained system at the i-th iteration [kB T ln( p i (n ))] will differ from the free
energy of the reference. The updated potentials become V i+1 (n ) = V i (n ) + kB T ln( p i (n )/ p(n )).
This iterative procedure converges to a potential at which p i (n ) p(n ).
One primary drawback to systematic coarse-grained methods is that they already necessitate
obtaining good all-atom statistics to parameterize the coarse-grained model (155). Additionally,
these potentials are system specic and in principle would require reparameterization of all inter-
actions if the system changed in any way. These methods are ideal for studying the self-assembly of
many identical molecules, as the inter- and intramolecular potentials are typically obtained from
a system of fewer molecules (assuming that the optimal potentials do not include direct many-
body interactions between several molecules). Many systematic coarse-grained methods have been
applied to aggregation (24, 55, 150, 152).

www.annualreviews.org Computational Studies of Protein Aggregation 655


PC66CH28-Shea ARI 28 February 2015 14:17

Carmichael & Shell (24) applied the relative entropy method to the self-assembly of polyalanine,
(Ala)15 . At least three beads per alanine were needed to capture the -helical and -hairpin
structures of the folded peptide. Simulating 25 copies of (Ala)15 , they found that the brillar order
emerged following the internal reorganization of a disordered intermediate.
The Voth group (150) applied the technique of multiscale coarse graining to a number of
protein aggregation systems. They studied the aggregation of 27 polyglutamine peptides and saw
an increase in the aggregation propensity with concentration and chain length. Additionally, they
studied how N-BAR proteins induce curvature changes in lipid vesicles (55). They simulated
vesicles 200300 nm in diameter with protein coverages ranging from 10% to 95%, immersed
in explicit solvent. They then mapped the coarse-grained lipid coordinates onto a mesoscopic
continuum model, with a eld variable describing the membranes protein composition. This
allowed the simulations to be extended to timescales comparable with experiment. The topology
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

of the spherical vesicle was dramatically altered into a tubular network. This change was associated
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

with the linear ordering of protein aggregates, which is believed to drive the formation of reticular
membrane structures in vivo.
Peter and colleagues (152) studied the aggregation of oligoalanine peptides, employing another
systematic coarse-graining method based on iterative Boltzmann inversion. They found that cer-
tain microscopic details lost in the coarse-graining process can be recovered by backmapping to
atomistic coordinates.

4.2. Thermodynamic Methods


Both atomistic and coarse-grained simulations combine well with enhanced sampling methods
to increase the sampling of Boltzmann-disfavored morphologies in order to compute free energy
proles more efciently. Some of these bias sampling on prespecied collective variables, such
as metadynamics (117) or umbrella sampling (122). Parallel tempering (or temperature replica
exchange) does not require the specication of collective variables; instead, it increases sampling
by exchanging trapped systems with a higher temperature (125).
Replica exchange enhances sampling by launching parallel simulations that each explore a
specic point in the parameter space. If this parameter is temperature, then the method is also
known as parallel tempering. In this method, parallel trajectories are launched, each with its own
parameter value (we discuss temperature for simplicity) (Figure 6). At regular intervals, each
trajectory is given an opportunity to swap with trajectories at neighboring temperatures according
to a Metropolis criterion that enforces the correct thermodynamics. The probability of swapping
replicas i and j is determined by the value of  = (i j )(U j U i ), where i = 1/kB T i , and Ui
is the potential energy of the current state of replica i. If  0, the swap always occurs; otherwise,
the swap probability is Pswap = exp(). The swapping procedure necessitates each parameter
having discontinuous trajectories, so correct kinetics are lost. Replica exchange can be employed
with both Monte Carlo and molecular dynamics simulations (125, 131). The replica exchange
methodology has been used in the atomistic simulations described earlier in this review, as well as
in a number of coarse-grained simulations (72, 83, 111115, 127, 128).
The replica exchange statistical temperature molecular dynamics algorithm, developed by
Keyes and colleagues (129), is an improvement on the usual temperature replica exchange al-
gorithm designed to minimize the number of replicas required for good sampling. The issue is
caused by the necessity of substantial energy overlap in neighboring replicas to maintain a high
temperature swap rate. It relies on the statistical temperature molecular dynamics algorithm (130)
to give a more even energy sampling and a self-adjusting weight. This method has been applied
to protein folding (129) but to our knowledge has not yet been applied to aggregation.

656 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

a Metadynamics

T1
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

b Replica exchange
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

T2

TN

Figure 6
Two commonly used simulation methods for the study of protein aggregation. (a) Metadynamics. Regions of
state space frequently visited ll up with Gaussian hills, biasing away from well-sampled states. (b) Replica
exchange molecular dynamics simulations. Parallel replicas swap to overcome free energy barriers.

Umbrella sampling is a technique in which a dened collective variable is held by a potential


well at a particular target value. Multiple trajectories are launched at varying target values such that
the statistics of neighboring umbrellas overlap. Umbrella sampling forces the system into regions
of state space that would otherwise have poor sampling. Because the bias potential is known, the
statistics of the unbiased potential can be recovered from the biased statistics. This is a useful
technique for determining the energy of the separation of monomers, as this separation can be
readily dened as a collective variable, as done by the Thirumalai group (123). Similarly, Davis &
Berkowitz (124) used this technique to bind A to a lipid bilayer and determined that the binding
promoted conversion to aggregate-prone conformations.
Metadynamics is another technique involving enhanced sampling over collective variables using
a biased potential to force the system to sample low-probability states (117). However, unlike
umbrella sampling, metadynamics is an adaptive method, automatically biasing congurations
away from those most visited to make the sampling more efcient. Similar to umbrella sampling,
the bias is accounted for to deduce the correct unbiased statistics of the system. More specically,
the method periodically adds a small Gaussian hill to the potential energy of the current region of
state space. Thus, regions that are frequently sampled are given a negative bias, eventually forcing
the system into rarer congurations. In the limit of t , the total potential energy will become

www.annualreviews.org Computational Studies of Protein Aggregation 657


PC66CH28-Shea ARI 28 February 2015 14:17

at as all the wells ll up with accumulated Gaussian hills. One must choose the extra parameters of
this method (hill height, width, and frequency) with care to ensure that the statistics do not depend
on them. This technique has been used to study congurations of amyloidogenic proteins (119
121, 156). A schematic comparison of replica exchange and metadynamics is given in Figure 6.

4.3. Kinetic Methods


The aforementioned methods lose kinetic information, by either biasing the sampling of collective
variables or giving discontinuous trajectories at constant temperature. Other sampling methods
preserve kinetics by launching many short, parallel trajectories, and it is the judicious choice of
how to launch these trajectories that enhances sampling.
The Markov state model (MSM) formalism has lately been applied to the study of biomolecular
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

simulation (133, 157, 158). This approach increases sampling by launching parallel trajectories.
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

The basic idea behind MSM is to bin sets of congurations in state space and model the system
as a set of Markovian transitions between these congurations, thereby generating a kinetic map
of transition probabilities between states. This method is well suited to sampling the kinetic
landscape because it adaptively selects starting congurations that require additional sampling.
It is important to follow up the MSM simulations with a verication that the transitions are
Markovian (history independent) to ensure self-consistency. It is also necessary to use a good state
space decomposition of the collective variables (collective variable binning) (159). MSM provides
a complementary approach to free energy methods such as replica exchange or metadynamics,
which lose kinetics in favor of sampling the energy landscape.
A related method is free energy guided sampling, developed by Zhou & Caisch (137). It dif-
fers from traditional MSM in that it uses an approximate free energy surface in place of collective
variables to bias the starting congurations. This method iteratively launches short trajectories in
parallel. It has two stages: exploration and renement. In the exploration stage, the initial trajecto-
ries are launched and the conguration space binned. An MSM model is constructed based on the
initial simulations, and a rough free energy prole around the starting conguration is generated.
This cycle is then restarted from free energy barriers farthest from the initial conguration. The
renement stage is designed to rene the calculated free energy surface. Trajectories are launched
from equally spaced initial congurations along the free energy surface, stopping when the calcu-
lated free energy has converged. The authors have applied this method to protein folding, but to
our knowledge, it has not yet been applied to aggregation.
The WExplore method developed by Dickson & Brooks (160) also biases the launching of
trajectories toward poorly sampled regions of conguration space, although unlike the MSM
methods it does not make the Markovian assumption for state transitions. It accomplishes this
using a weight for each trajectory with which it contributes to statistical averages. Sampling
regions are dened dynamically in conguration space, and trajectories are cloned and merged to
encourage even sampling across these regions. The sampling regions are dened using a distance
metric (e.g., RMSD) in a possibly high-dimensional space of order parameters and take the form of
Voronoi polyhedra. Unlike the original weighted ensemble algorithm (161), the sampling regions
can be dened in a hierarchical fashion, which allows for the balancing of computational effort
across multiple length scales. Boltzmann sampling is achieved by changing the weight of each
trajectory upon cloning steps (in which weights are split) and merging steps (in which weights are
added). Thus, replicas are forced to expand across the conguration space much faster than the
free energy surface would normally allow. This method has been applied to RNA conformational
dynamics (162) but would be well suited to protein aggregation as well.

658 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

The kinetics of bril formation can also be inferred through secondary nucleation data analysis
methods, such as those employed by Knowles and colleagues (2, 8, 9, 143). In this approach, one
represents the kinetics as a function of an order parameter dening the degree of brillization. The
data are t to a master equation that breaks down bril growth into several subprocesses. These
include elongation, fragmentation, nucleation, and, recently (9), end-to-end association. The au-
thors found that the lag phase cannot simply be described by the nucleation time of only the
primary pathway, highlighting the signicance of secondary nucleation. These analytical methods
can be used on both experimental data and simulation trajectories.
The normal mode analysis method (144, 145) is applied to an existing (typically all-atom)
trajectory. It breaks the protein into coarse-grained sites and deduces collective motions of the
biomolecule from the vibrational network of pairs of these sites. Eom and colleagues (146) em-
ployed this analysis on all-atom simulations to study elastic modes of an hIAPP bril (bending,
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

torsion, stretching) for various structural hierarchies (e.g., parallel/antiparallel sheets and bril
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

length). They showed how the bril structure affects its mechanical rigidity. The Buehler group
(147) studied the elastic properties of A brils using normal mode analysis. They found the
brils Youngs modulus to be consistent with experimental values.

5. CONCLUSIONS
Above we review models of different resolutions, as well as a selection of simulation methodologies
commonly used to study protein aggregation. We focus on the aggregation process itself, with
some mentions of aggregation on surfaces and membranes. In principle, all the methods and
models introduced here can be applied to the study of aggregation in a more cellular context. The
challenge lies in the increased complexity of the system: To even begin to describe the cellular
environment, one needs to take into account its many constituents (e.g., membranes, nucleic acids,
osmolytes). This effort has already begun in earnest, and we anticipate signicant advances in the
coming years in our understanding of how the cellular environment modulates the aggregation
process. Computational modeling shall continue to play a pivotal role in helping us understand
the nature of protein aggregation, providing an important complement to experimental studies.
The near-boundless complexity arising from the extreme many-body nature of the problem will
fuel the development of many more computational methods to come, ensuring the continued
signicance of biomolecular simulation to the eld of protein aggregation.

DISCLOSURE STATEMENT
The authors are not aware of any afliations, memberships, funding, or nancial holdings that
might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS
We acknowledge the support of a National Science Foundation (NSF) grant (MCB-1158577) and
the David and Lucile Packard Foundation. Additionally, we received support from the Center for
Scientic Computing from the CNSI, MRL, an NSF Materials Research Science and Engineering
Center (MRSEC) (DMR-1121053), and NSF CNS-0960316. This work was supported in part
by the MRSEC Program of the NSF under award DMR-1121053. This work used the Extreme
Science and Engineering Discovery Environment (XSEDE), which is supported by NSF grants

www.annualreviews.org Computational Studies of Protein Aggregation 659


PC66CH28-Shea ARI 28 February 2015 14:17

ACI-1053575 and TG-MCA05S027. We thank Andrij Baumketner, Scott Shell, Scott Carmichael,
Zach Levine, and Catie Carpenter for assistance with the gures.

LITERATURE CITED
1. Chiti F, Dobson C. 2006. Protein misfolding, functional amyloid, and human disease. Annu. Rev. Biochem.
75:33366
2. Knowles TP, Vendruscolo M, Dobson CM. 2014. The amyloid state and its association with protein
misfolding diseases. Nat. Rev. Mol. Cell Biol. 15:38496
3. Knowles TPJ, Buehler MJ. 2011. Nanomechanics of functional and pathological amyloid material. Nat.
Nanotechnol. 6:46979
4. Fitzpatrick AWP, Debelouchina GT, Bayro MJ, Clare DK, Caporini MA, et al. 2013. Atomic structure
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

and hierarchical assembly of a cross- amyloid bril. Proc. Natl. Acad. Sci. USA 110:546873
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

5. Sunde M, Serpell LC, Bartlam M, Fraser PE, Pepys MB, Blake CC. 1997. Common core structure of
amyloid brils by synchrotron X-ray diffraction. J. Mol. Biol. 273:72939
6. Thirumalai D, Klimov D, Dima R. 2003. Emerging ideas on the molecular basis of protein and peptide
aggregation. Curr. Opin. Struct. Biol. 13:14659
7. Straub J, Thirumalai D. 2011. Toward a molecular theory of early and late events in monomer to amyloid
bril formation. Annu. Rev. Phys. Chem. 62:43763
8. Knowles TP, Waudby CA, Devlin GL, Cohen SI, Aguzzi A, et al. 2009. An analytical solution to the
kinetics of breakable lament assembly. Science 326:153337
9. Michaels TC, Knowles TP. 2014. Role of lament annealing in the kinetics and thermodynamics of
nucleated polymerization. J. Chem. Phys. 140:214904
10. Friesner RA, Guallar V. 2005. Ab initio quantum chemical and mixed quantum mechanics/molecular
mechanics (QM/MM) methods for studying enzymatic catalysis. Annu. Rev. Phys. Chem. 56:389427
11. Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan S, Karplus M. 1983. CHARMM: a
program for macromolecular energy, minimization, and dynamics calculations. J. Comput. Chem. 4:187
217
12. Pearlman DA, Case DA, Caldwell JW, Ross WS, Cheatham TE III, et al. 1995. AMBER, a package of
computer programs for applying molecular mechanics, normal mode analysis, molecular dynamics and
free energy calculations to simulate the structural and energetic properties of molecules. Comput. Phys.
Commun. 91:141
13. Christen M, Hunenberger PH, Bakowies D, Baron R, Burgi R, et al. 2005. The GROMOS software for
biomolecular simulation: GROMOS05. J. Comput. Chem. 26:171951
14. Jorgensen WL, Tirado-Rives J. 1988. The OPLS [optimized potentials for liquid simulations] potential
functions for proteins, energy minimizations for crystals of cyclic peptides and crambin. J. Am. Chem.
Soc. 110:165766
15. Monticelli L, Kandasamy SK, Periole X, Larson RG, Tieleman DP, Marrink SJ. 2008. The MARTINI
coarse-grained force eld: extension to proteins. J. Chem. Theory Comput. 4:81934
16. Sterpone F, Melchionna S, Tuffery P, Pasquali S, Mousseau N, et al. 2014. The OPEP protein model:
from single molecules, amyloid formation, crowding and hydrodynamics to DNA/RNA systems. Chem.
Soc. Rev. 43:487193
17. Cheon M, Chang I, Hall C. 2010. Extending the prime model for protein aggregation to all 20 amino
acids. Proteins 78:295060
18. Kinjo AR, Takada S. 2003. Competition between protein folding and aggregation with molecular chap-
erones in crowded solutions: insight from mesoscopic simulations. Biophys. J. 85:352131
19. Irback A, Jonsson S, Linnemann N, Linse B, Wallin S. 2013. Aggregate geometry in amyloid bril
nucleation. Phys. Rev. Lett. 110:058101
20. Bieler NS, Knowles TP, Frenkel D, Vacha R. 2012. Connecting macroscopic observables and mi-
croscopic assembly events in amyloid formation using coarse grained simulations. PLoS Comput. Biol.
8:e1002692

660 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

21. Li M, Klimov D, Straub J, Thirumalai D. 2008. Probing the mechanisms of bril formation using lattice
models. J. Chem. Phys. 129:175101
22. Friedman R, Caisch A. 2011. Surfactant effects on amyloid aggregation kinetics. J. Mol. Biol. 414:30312
23. Bellesia G, Shea JE. 2007. Self-assembly of -sheet forming peptides into chiral brillar aggregates.
J. Chem. Phys. 126:245104
24. Carmichael SP, Shell MS. 2012. A new multiscale algorithm and its application to coarse-grained peptide
models for self-assembly. J. Phys. Chem. B 116:838393
25. Rosenman DJ, Connors CR, Chen W, Wang C, Garcia AE. 2013. A monomers transiently sample
oligomer and bril-like congurations: ensemble characterization using a combined MD/NMR ap-
proach. J. Mol. Biol. 425:333859
26. Wu C, Bowers MT, Shea JE. 2011. On the origin of the stronger binding of PIB over thioavin T to
protobrils of the Alzheimer amyloid- peptide: a molecular dynamics study. Biophys. J. 100:131624
27. Best RB, Mittal J. 2011. Free-energy landscape of the GB1 hairpin in all-atom explicit solvent simulations
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

with different force elds: similarities and differences. Proteins 79:131828


Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

28. Bitan G, Kirkitadze M, Lomakin A, Vollers S, Benedek G, Teplow D. 2003. Amyloid -protein (A)
assembly: A40 and A42 oligomerize through distinct pathways. Proc. Natl. Acad. Sci. USA 100:33035
29. Bernstein SL, Dupuis NF, Lazo ND, Wyttenbach T, Condron MM, et al. 2009. Amyloid- protein
oligomerization and the importance of tetramers and dodecamers in the aetiology of Alzheimers disease.
Nat. Chem. 1:32631
30. Ball KA, Phillips AH, Wemmer DE, Head-Gordon T. 2013. Differences in -strand populations of
monomeric A40 and A42. Biophys. J. 104:271424
31. Lazo N, Grant M, Condron M, Rigby A, Teplow D. 2005. On the nucleation of amyloid -protein
monomer folding. Protein Sci. 14:158196
32. Murray MM, Krone MG, Bernstein SL, Baumketner A, Condron MM, et al. 2009. Amyloid -protein:
experiment and theory on the 2130 fragment. J. Phys. Chem. B 113:604146
33. Krone MG, Baumketner A, Bernstein SL, Wyttenbach T, Lazo ND, et al. 2008. Effects of familial
Alzheimers disease mutations on the folding nucleation of the amyloid -protein. J. Mol. Biol. 381:221
28
34. Wu C, Murray MM, Bernstein SL, Condron MM, Bitan G, et al. 2009. The structure of A42 C-terminal
fragments probed by a combined experimental and theoretical study. J. Mol. Biol. 387:492501
35. Wu C, Shea JE. 2013. Structural similarities and differences between amyloidogenic and non-
amyloidogenic islet amyloid polypeptide (IAPP) sequences and implications for the dual physiological
and pathological activities of these peptides. PLoS Comput. Biol. 9:e1003211
36. Reddy AS, Wang L, Singh S, Ling YL, Buchanan L, et al. 2010. Stable and metastable states of human
amylin in solution. Biophys. J. 99:220816
37. Dupuis NF, Wu C, Shea JE, Bowers M. 2009. Human islet amyloid polypeptide monomers form ordered
-hairpins: a possible direct amyloidogenic precursor. J. Am. Chem. Soc. 191:1828392
38. Miller Y, Ma B, Nussinov R. 2010. Polymorphism in Alzheimer A amyloid organization reects con-
formational selection in a rugged energy landscape. Chem. Rev. 110:482038
39. Zhao J, Yu X, Liang G, Zheng J. 2011. Heterogeneous triangular structures of human islet amyloid
polypeptide (amylin) with internal hydrophobic cavity and external wrapping morphology reveal the
polymorphic nature of amyloid brils. Biomacromolecules 12:178194
40. Wu C, Bowers M, Shea JE. 2010. Molecular structures of quiescently grown and brain-derived poly-
morphic brils of the Alzheimer amyloid A940 peptide: a comparison to agitated brils. PLoS Comput.
Biol. 6:e1000693
41. Buchete N, Hummer G. 2007. Structure and dynamics of parallel -sheets, hydrophobic core, and loops
in Alzheimers A brils. Biophys. J. 92:303239
42. Lemkul JA, Bevan DR. 2012. The role of molecular simulations in the development of inhibitors of
amyloid -peptide aggregation for the treatment of Alzheimers disease. ACS Chem. Neurosci. 3:84556
43. Ngo ST, Li MS. 2012. Curcumin binds to A140 peptides and brils stronger than ibuprofen and
naproxen. J. Phys. Chem. B 116:1016575
44. Wu C, Scott J, Shea JE. 2012. Binding of Congo Red to amyloid protobrils of the Alzheimer A940
peptide probed by molecular dynamics simulations. Biophys. J. 103:55057

www.annualreviews.org Computational Studies of Protein Aggregation 661


PC66CH28-Shea ARI 28 February 2015 14:17

45. Takeda T, Chang WE, Raman EP, Klimov DK. 2010. Binding of nonsteroidal anti-inammatory drugs
to A bril. Proteins 78:284960
46. Schor M, Vreede J, Bolhuis PG. 2012. Elucidating the locking mechanism of peptides onto growing
amyloid brils through transition path sampling. Biophys. J. 103:1296304
47. Esler W, Stimson E, Jennings J, Vinters H, Ghilardi J, et al. 2000. Alzheimers disease amyloid propa-
gation by a template-dependent dock-lock mechanism. Biochemistry 39:628895
48. Xi W, Li W, Wang W. 2012. Template induced conformational change of amyloid- monomer. J. Phys.
Chem. B 116:7398405
49. Morriss-Andrews A, Brown FLH, Shea JE. 2014. A coarse-grained model for peptide aggregation on a
membrane surface. J. Phys. Chem. B 118:842032
50. Simunovic M, Srivastava A, Voth GA. 2013. Linear aggregation of proteins on the membrane as a prelude
to membrane remodeling. Proc. Natl. Acad. Sci. USA 110:20396401
51. Li H, Gorfe AA. 2013. Aggregation of lipid-anchored full-length H-Ras in lipid bilayers: simulations
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

with the MARTINI force eld. PLoS ONE 8:e71018


Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

52. Pannuzzo M, Milardi D, Raudino A, Karttunen M, La Rosa C. 2013. Analytical model and multiscale
simulations of A peptide aggregation in lipid membranes: towards a unifying description of conforma-
tional transitions, oligomerization and membrane damage. Phys. Chem. Chem. Phys. 15:894051
53. Santo KP, Berkowitz ML. 2012. Difference between magainin-2 and melittin assemblies in phosphatidyl-
choline bilayers: results from coarse-grained simulations. J. Phys. Chem. B 116:302130
54. Parton DL, Klingelhoefer JW, Sansom MS. 2011. Aggregation of model membrane proteins, modulated
by hydrophobic mismatch, membrane curvature, and protein class. Biophys. J. 101:69199
55. Simunovic M, Mim C, Marlovits TC, Resch G, Unger VM, Voth GA. 2013. Protein-mediated trans-
formation of lipid vesicles into tubular networks. Biophys. J. 105:71119
56. Hung A, Yarovsky I. 2011. Inhibition of peptide aggregation by lipids: insights from coarse-grained
molecular simulations. J. Mol. Graph. Model. 29:597607
57. Keller A, Fritzsche M, Yu YP, Liu Q, Li YM, et al. 2011. Inuence of hydrophobicity on the surface-
catalyzed assembly of the islet amyloid polypeptide. ACS Nano 5:277078
58. Zhu M, Souillac P, Ionescu-Zanetti C, Carter S, Fink A. 2002. Surface-catalyzed amyloid bril formation.
J. Biol. Chem. 277:5091422
59. Kowalewski T, Holtzman DM. 1999. In situ atomic force microscopy study of Alzheimers -amyloid
peptide on different substrates: new insights into mechanism of -sheet formation. Proc. Natl. Acad. Sci.
USA 96:368893
60. Green JD, Goldsbury C, Kistler J, Cooper GJS, Aebi U. 2004. Human amylin oligomer growth and
bril elongation dene two distinct phases in amyloid formation. J. Biol. Chem. 279:1220612
61. Losic D, Martin LL, Aguilar MI, Small DH. 2006. -amyloid bril formation is promoted by step edges
of highly oriented pyrolytic graphite. Peptide Sci. 84:51926
62. Giacomelli CE, Norde W. 2003. Inuence of hydrophobic Teon particles on the structure of amyloid
-peptide. Biomacromolecules 4:171926
63. Ha C, Park CB. 2006. Ex situ atomic force microscopy analysis of -amyloid self-assembly and deposition
on a synthetic template. Langmuir 22:697785
64. OBrien EP, Ziv G, Haran G, Brooks BR, Thirumalai D. 2008. Effects of denaturants and osmolytes on
proteins are accurately predicted by the molecular transfer model. Proc. Natl. Acad. Sci. USA 105:134038
65. Wu C, Shea JE. 2011. Coarse-grained models for protein aggregation. Curr. Opin. Struct. Biol. 21:20920
66. Morriss-Andrews A, Shea JE. 2014. Simulations of protein aggregation: insights from atomistic and
coarse-grained models. J. Phys. Chem. Lett. 5:1899908
67. Barz B, Urbanc B. 2014. Minimal model of self-assembly: emergence of diversity and complexity.
J. Phys. Chem. B 118:376170
68. Auer S, Meersman F, Dobson C, Vendruscolo M. 2008. A generic mechanism of emergence of amyloid
protolaments from disordered oligomeric aggregates. PLoS Comput. Biol. 4:e1000222
69. Zhang J, Muthukumar M. 2009. Simulations of nucleation and elongation of amyloid brils. J. Chem.
Phys. 130:035102
70. Paparcone R, Cranford SW, Buehler MJ. 2011. Self-folding and aggregation of amyloid nanobrils.
Nanoscale 3:174855

662 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

71. Li M, Co N, Reddy G, Hu C, Straub J, Thirumalai D. 2010. Factors governing brillogenesis of


polypeptide chains revealed by lattice models. Phys. Rev. Lett. 105:218101
72. Ni R, Abeln S, Schor M, Stuart MAC, Bolhuis PG. 2013. Interplay between folding and assembly of
bril-forming polypeptides. Phys. Rev. Lett. 111:058101
73. Magno A, Pellarin R, Caisch A. 2012. Mechanisms and kinetics of amyloid aggregation investigated by
a phenomenological coarse-grained model. In Computational Modeling of Biological Systems: From Molecules
to Pathways, ed. NV Dokholyan, pp. 191214. New York: Wiley
74. Pellarin R, Caisch A. 2006. Interpreting the aggregation kinetics of amyloid peptides. J. Mol. Biol.
360:88292
75. Morriss-Andrews A, Bellesia G, Shea JE. 2012. -sheet propensity controls the kinetic pathways and
morphologies of seeded peptide aggregation. J. Chem. Phys. 137:145104
76. Pellarin R, Guarnera E, Caisch A. 2007. Pathways and intermediates of amyloid bril formation.
J. Mol. Biol. 374:91724
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

77. Pellarin R, Schuetz P, Guarnera E, Caisch A. 2010. Amyloid bril polymorphism is under kinetic
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

control. J. Am. Chem. Soc. 132:1496070


78. Friedman R, Pellarin R, Caisch A. 2009. Amyloid aggregation on lipid bilayers and its impact on
membrane permeability. J. Mol. Biol. 387:40715
79. Bellesia G, Shea JE. 2009. Effect of -sheet propensity on peptide aggregation. J. Chem. Phys. 130:145103
80. Bellesia G, Shea JE. 2009. Diversity of kinetic pathways in amyloid bril formation. J. Chem. Phys.
131:111102
81. Morriss-Andrews A, Shea JE. 2012. Kinetic pathways to peptide aggregation on surfaces: the effects of
-sheet propensity and surface attraction. J. Chem. Phys. 136:065103
82. Brannigan G, Philips P, Brown F. 2005. Flexible lipid bilayers in implicit solvent. Phys. Rev. E 72:011915
83. Morriss-Andrews A, Bellesia G, Shea JE. 2011. Effects of surface interactions on peptide aggregate
morphology. J. Chem. Phys. 135:085102
84. Thota N, Luo Z, Hu Z, Jiang J. 2013. Self-assembly of amphiphilic peptide (AF)6H5K15: coarse-grained
molecular dynamics simulation. J. Phys. Chem. B 117:969098
85. Seo M, Rauscher S, Pomes R, Tieleman DP. 2012. Improving internal peptide dynamics in the coarse-
grained MARTINI model: toward large-scale simulations of amyloid- and elastin-like peptides. J. Chem.
Theory Comput. 8:177485
86. Lee OS, Cho V, Schatz GC. 2012. Modeling the self-assembly of peptide amphiphiles into bers using
coarse-grained molecular dynamics. Nano Lett. 12:490713
87. Guo C, Luo Y, Zhou R, Wei G. 2012. Probing the self-assembly mechanism of diphenylalanine-based
peptide nanovesicles and nanotubes. ACS Nano 6:390718
88. Srensen J, Periole X, Skeby KK, Marrink SJ, Schitt B. 2011. Protobrillar assembly toward the
formation of amyloid brils. J. Phys. Chem. Lett. 2:238590
89. Frederix PW, Ulijn RV, Hunt NT, Tuttle T. 2011. Virtual screening for dipeptide aggregation: toward
predictive tools for peptide self-assembly. J. Phys. Chem. Lett. 2:238084
90. Marrink SJ, Tieleman DP. 2013. Perspective on the Martini model. Chem. Soc. Rev. 42:680122
91. Phelps EM, Hall CK. 2012. Structural transitions and oligomerization along polyalanine bril formation
pathways from computer simulations. Proteins 80:158297
92. Cheon M, Chang I, Hall CK. 2012. Inuence of temperature on formation of perfect tau fragment brils
using PRIME20/DMD simulations. Protein Sci. 21:151427
93. Wagoner VA, Cheon M, Chang I, Hall CK. 2012. Fibrillization propensity for short designed hexapep-
tides predicted by computer simulation. J. Mol. Biol. 416:598609
94. Wagoner VA, Cheon M, Chang I, Hall CK. 2011. Computer simulation study of amyloid bril formation
by palindromic sequences in prion peptides. Proteins 79:213245
95. Cheon M, Chang I, Hall CK. 2011. Spontaneous formation of twisted A1622 brils in large-scale
molecular-dynamics simulations. Biophys. J. 101:2493501
96. Wagoner VA, Cheon M, Chang I, Hall CK. 2014. Impact of sequence on the molecular assembly of
short amyloid peptides. Proteins 82:146983
97. Sawaya MR, Sambashivan S, Nelson R, Ivanova MI, Sievers SA, et al. 2007. Atomic structures of amyloid
cross- spines reveal varied steric zippers. Nature 447:45357

www.annualreviews.org Computational Studies of Protein Aggregation 663


PC66CH28-Shea ARI 28 February 2015 14:17

98. de la Paz ML, Serrano L. 2004. Sequence determinants of amyloid bril formation. Proc. Natl. Acad. Sci.
USA 101:8792
99. de la Paz ML, Goldie K, Zurdo J, Lacroix E, Dobson CM, et al. 2002. De novo designed peptide-based
amyloid brils. Proc. Natl. Acad. Sci. USA 99:1605257
100. Mehta AK, Lu K, Childers WS, Liang Y, Dublin SN, et al. 2008. Facial symmetry in protein self-
assembly. J. Am. Chem. Soc. 130:982935
101. Balbach J, Ishii Y, Antzutkin O, Leapman R, Rizzo N, et al. 2000. Amyloid bril formation by A1622 ,
a seven-residue fragment of the Alzheimers -amyloid peptide, and structural characterization by solid
state NMR. Biochemistry 39:1374859
102. Tjernberg LO, Naslund J, Lindqvist F, Johansson J, Karlstrom AR, et al. 1996. Arrest of -amyloid
bril formation by a pentapeptide ligand. J. Biol. Chem. 271:854548
103. Ricchiuto P, Brukhno AV, Auer S. 2012. Protein aggregation: kinetics versus thermodynamics. J. Phys.
Chem. B 116:538490
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

104. Hoang TX, Trovato A, Seno F, Banavar JR, Maritan A. 2004. Geometry and symmetry presculpt the
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

free-energy landscape of proteins. Proc. Natl. Acad. Sci. USA 101:796064


105. Meral D, Urbanc B. 2013. Discrete molecular dynamics study of oligomer formation by N-terminally
truncated amyloid -protein. J. Mol. Biol. 425:226075
106. Urbanc B, Betnel M, Cruz L, Li H, Fradinger E, et al. 2011. Structural basis for AC-terminal fragments:
discrete molecular dynamics study. J. Mol. Biol. 410:31628
107. Urbanc B, Betnel M, Cruz L, Bitan G, Teplow D. 2010. Elucidation of amyloid -protein oligomeriza-
tion mechanisms: discrete molecular dynamics study. J. Am. Chem. Soc. 132:426680
108. Peng S, Ding F, Urbanc B, Buldyrev S, Cruz L, et al. 2004. Discrete molecular dynamics simulations of
peptide aggregation. Phys. Rev. E 69:041908
109. Ding F, Furukawa Y, Nukina N, Dokholyan NV. 2012. Local unfolding of Cu, Zn superoxide dismutase
monomer determines the morphology of brillar aggregates. J. Mol. Biol. 421:54860
110. Redler RL, Wilcox KC, Proctor EA, Fee L, Caplow M, Dokholyan NV. 2011. Glutathionylation at
Cys-111 induces dissociation of wild type and FALS mutant SOD1 dimers. Biochemistry 50:705766
111. Cote S, Laghaei R, Derreumaux P, Mousseau N. 2012. Distinct dimerization for various alloforms of
the amyloid- protein: A140 , A142 , and A140 (D23N). J. Phys. Chem. B 116:404355
112. Spill YG, Pasquali S, Derreumaux P. 2011. Impact of thermostats on folding and aggregation properties
of peptides using the optimized potential for efcient structure prediction coarse-grained model. J. Chem.
Theory Comput. 7:150210
113. Chebaro Y, Jiang P, Zang T, Mu Y, Nguyen PH, et al. 2012. Structures of A1742 trimers in isolation
and with ve small-molecule drugs using a hierarchical computational procedure. J. Phys. Chem. B
116:841222
114. Lu Y, Wei G, Derreumaux P. 2012. Structural, thermodynamical, and dynamical properties of oligomers
formed by the amyloid NNQQ peptide: insights from coarse-grained simulations. J. Chem. Phys.
137:025101
115. Nasica-Labouze J, Meli M, Derreumaux P, Colombo G, Mousseau N. 2011. A multiscale approach to
characterize the early aggregation steps of the amyloid-forming peptide GNNQQNY from the yeast
prion Sup-35. PLoS Comp. Biol. 7:e1002051
116. Nasica-Labouze J, Mousseau N. 2012. Kinetics of amyloid aggregation: a study of the GNNQQNY
prion sequence. PLoS Comput. Biol. 8:e1002782
117. Laio A, Parrinello M. 2002. Escaping free-energy minima. Proc. Natl. Acad. Sci. USA 99:1256266
118. Gronau G, Qin Z, Buehler MJ. 2013. Effect of sodium chloride on the structure and stability of spider
silks N-terminal protein domain. Biomater. Sci. 1:27684
119. Camilloni C, Schaal D, Schweimer K, Schwarzinger S, De Simone A. 2012. Energy landscape of the
prion protein helix 1 probed by metadynamics and NMR. Biophys. J. 102:15867
120. Rossetti G, Cossio P, Laio A, Carloni P. 2011. Conformations of the Huntingtin N-term in aqueous
solution from atomistic simulations. FEBS Lett. 585:308689
121. Wang H, Barreyro L, Provasi D, Djemil I, Torres-Arancivia C, et al. 2011. Molecular determinants and
thermodynamics of the amyloid precursor protein transmembrane domain implicated in Alzheimers
disease. J. Mol. Biol. 408:87995

664 Morriss-Andrews Shea


PC66CH28-Shea ARI 28 February 2015 14:17

122. Torrie GM, Valleau JP. 1977. Nonphysical sampling distributions in Monte Carlo free-energy estima-
tion: umbrella sampling. J. Comput. Phys. 23:18799
123. Rivera E, Straub J, Thirumalai D. 2009. Sequence and crowding effects in the aggregation of a 10-residue
fragment derived from islet amyloid polypeptide. Biophys. J. 96:455260
124. Davis C, Berkowitz M. 2009. Interaction between amyloid- (142) peptide and phospholipid bilayers:
a molecular dynamics study. Biophys. J. 96:78597
125. Sugita Y, Okamoto Y. 1999. Replica-exchange molecular dynamics method for protein folding. Chem.
Phys. Lett. 314:14151
126. Sugita Y, Kitao A, Okamoto Y. 2000. Multidimensional replica-exchange method for free-energy calcu-
lations. J. Chem. Phys. 113:604251
127. Fukunishi H, Watanabe O, Takada S. 2002. On the Hamiltonian replica exchange method for efcient
sampling of biomolecular systems: application to protein structure prediction. J. Chem. Phys. 116:9058
67
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

128. Ostermeir K, Zacharias M. 2013. Advanced replica-exchange sampling to study the exibility and plas-
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

ticity of peptides and proteins. Biochim. Biophys. Acta 1834:84753


129. Kim J, Straub JE, Keyes T. 2012. Replica exchange statistical temperature molecular dynamics algorithm.
J. Phys. Chem. B 116:864653
130. Kim J, Straub JE, Keyes T. 2006. Statistical-temperature Monte Carlo and molecular dynamics algo-
rithms. Phys. Rev. Lett. 97:050601
131. Swendsen RH, Wang JS. 1986. Replica Monte Carlo simulation of spin-glasses. Phys. Rev. Lett. 57:26079
132. Enciso M, Rey A. 2012. Simple model for the simulation of peptide folding and aggregation with different
sequences. J. Chem. Phys. 136:215103
133. Swope WC, Pitera JW, Suits F. 2004. Describing protein folding kinetics by molecular dynamics simu-
lations. 1. Theory. J. Phys. Chem. B 108:657181
134. Park S, Pande VS. 2006. Validation of Markov state models using Shannons entropy. J. Chem. Phys.
124:054118
135. Swope WC, Pitera JW, Suits F, Pitman M, Eleftheriou M, et al. 2004. Describing protein folding kinetics
by molecular dynamics simulations. 2. Example applications to alanine dipeptide and a -hairpin peptide.
J. Phys. Chem. B 108:658294
136. Kelley NW, Vishal V, Krafft GA, Pande VS. 2008. Simulating oligomerization at experimental concen-
trations and long timescales: a Markov state model approach. J. Chem. Phys. 129:214707
137. Zhou T, Caisch A. 2012. Free energy guided sampling. J. Chem. Theory Comput. 8:213440
138. Dellago C, Bolhuis PG, Csajka FS, Chandler D. 1998. Transition path sampling and the calculation of
rate constants. J. Chem. Phys. 108:196477
139. Bolhuis PG, Chandler D, Dellago C, Geissler PL. 2002. Transition path sampling: throwing ropes over
rough mountain passes, in the dark. Annu. Rev. Phys. Chem. 53:291318
140. Peters B, Trout BL. 2006. Obtaining reaction coordinates by likelihood maximization. J. Chem. Phys.
125:054108
141. Maragliano L, Fischer A, Vanden-Eijnden E, Ciccotti G. 2006. String method in collective variables:
minimum free energy paths and isocommittor surfaces. J. Chem. Phys. 125:024106
142. E W, Ren W, Vanden-Eijnden E. 2002. String method for the study of rare events. Phys. Rev. B 66:052301
143. Cohen SI, Vendruscolo M, Dobson CM, Knowles TP. 2012. From macroscopic measurements to mi-
croscopic mechanisms of protein aggregation. J. Mol. Biol. 421:16071
144. Tirion MM. 1996. Large amplitude elastic motions in proteins from a single-parameter, atomic analysis.
Phys. Rev. Lett. 77:19058
145. Hinsen K. 1998. Analysis of domain motions by approximate normal mode calculations. Proteins 33:417
29
146. Yoon G, Kwak J, Kim JI, Na S, Eom K. 2011. Mechanical characterization of amyloid brils using
coarse-grained normal mode analysis. Adv. Funct. Mater. 21:345463
147. Xu Z, Paparcone R, Buehler MJ. 2010. Alzheimers A(140) amyloid brils feature size-dependent
mechanical properties. Biophys. J. 98:205362
148. Shell MS. 2008. The relative entropy is fundamental to multiscale and inverse thermodynamic problems.
J. Chem. Phys. 129:144108

www.annualreviews.org Computational Studies of Protein Aggregation 665


PC66CH28-Shea ARI 28 February 2015 14:17

149. Izvekov S, Voth GA. 2005. A multiscale coarse-graining method for biomolecular systems. J. Phys. Chem.
B 109:246973
150. Wang Y, Voth GA. 2010. Molecular dynamics simulations of polyglutamine aggregation using solvent-
free multiscale coarse-grained models. J. Phys. Chem. B 114:873543
151. Reith D, Putz M, Muller-Plathe F. 2003. Deriving effective mesoscale potentials from atomistic simu-
lations. J. Comput. Chem. 24:162436
152. Bezkorovaynaya O, Lukyanov A, Kremer K, Peter C. 2012. Multiscale simulation of small peptides:
consistent conformational sampling in atomistic and coarse-grained models. J. Comput. Chem. 33:937
49
153. Alder BJ, Wainwright T. 1959. Studies in molecular dynamics. I. General method. J. Chem. Phys. 31:459
66
154. Izvekov S, Parrinello M, Burnham CJ, Voth GA. 2004. Effective force elds for condensed phase sys-
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

tems from ab initio molecular dynamics simulation: a new method for force-matching. J. Chem. Phys.
120:10896913
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

155. Larini L, Shea JE. 2012. Coarse-grained modeling of simple molecules at different resolutions in the
absence of good sampling. J. Phys. Chem. B 116:833749
156. Chiu CC, Singh S, de Pablo JJ. 2013. Effect of proline mutations on the monomer conformations of
amylin. Biophys. J. 105:122735
157. Singhal N, Snow CD, Pande VS. 2004. Using path sampling to build better Markovian state models:
predicting the folding rate and mechanism of a tryptophan zipper beta hairpin. J. Chem. Phys. 121:41525
158. Prinz JH, Wu H, Sarich M, Keller B, Senne M, et al. 2011. Markov models of molecular kinetics:
generation and validation. J. Chem. Phys. 134:174105
159. Elmer SP, Park S, Pande VS. 2005. Foldamer dynamics expressed via Markov state models. II. State
space decomposition. J. Chem. Phys. 123:114903
160. Dickson A, Brooks CL III. 2014. WExplore: hierarchical exploration of high-dimensional spaces using
the weighted ensemble algorithm. J. Phys. Chem. B 118:353242
161. Huber GA, Kim S. 1996. Weighted-ensemble Brownian dynamics simulations for protein association
reactions. Biophys. J. 70:97110
162. Dickson A, Mustoe AM, Salmon L, Brooks CL III. 2014. Efcient in silico exploration of RNA inter-
helical conformations using Euler angles and WExplore. Nucl. Acids Res. 42:1212637

666 Morriss-Andrews Shea


PC66-FrontMatter ARI 4 March 2015 12:22

Annual Review of
Physical Chemistry
Contents Volume 66, 2015

Molecules in Motion: Chemical Reaction and Allied Dynamics in


Solution and Elsewhere
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

James T. Hynes p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

Crystal Structure and Prediction


Tejender S. Thakur, Ritesh Dubey, and Gautam R. Desiraju p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p21
Reaction Dynamics in Astrochemistry: Low-Temperature Pathways to
Polycyclic Aromatic Hydrocarbons in the Interstellar Medium
Ralf I. Kaiser, Dorian S.N. Parker, and Alexander M. Mebel p p p p p p p p p p p p p p p p p p p p p p p p p p p p p43
Coherence in Energy Transfer and Photosynthesis
Aurelia Chenu and Gregory D. Scholes p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p69
Ultrafast Dynamics of Electrons in Ammonia
Peter Vohringer p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p97
Dynamics of Bimolecular Reactions in Solution
Andrew J. Orr-Ewing p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 119
The Statistical Mechanics of Dynamic Pathways to Self-Assembly
Stephen Whitelam and Robert L. Jack p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 143
Reaction Dynamics at Liquid Interfaces
Ilan Benjamin p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 165
Quantitative Sum-Frequency Generation Vibrational Spectroscopy of
Molecular Surfaces and Interfaces: Lineshape, Polarization,
and Orientation
Hong-Fei Wang, Luis Velarde, Wei Gan, and Li Fu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 189
Mechanisms of Virus Assembly
Jason D. Perlmutter and Michael F. Hagan p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 217
Cold and Controlled Molecular Beams: Production and Applications
Justin Jankunas and Andreas Osterwalder p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 241
Spintronics and Chirality: Spin Selectivity in Electron Transport
Through Chiral Molecules
Ron Naaman and David H. Waldeck p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 263

v
PC66-FrontMatter ARI 4 March 2015 12:22

DFT: A Theory Full of Holes?


Aurora Pribram-Jones, David A. Gross, and Kieron Burke p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 283
Theoretical Description of Structural and Electronic Properties of
Organic Photovoltaic Materials
Andriy Zhugayevych and Sergei Tretiak p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 305
Advanced Physical Chemistry of Carbon Nanotubes
Jun Li and Gaind P. Pandey p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 331
Site-Specic Infrared Probes of Proteins
Jianqiang Ma, Ileana M. Pazos, Wenkai Zhang, Robert M. Culik, and Feng Gai p p p p p 357
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Biomolecular Damage Induced by Ionizing Radiation: The Direct and


Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

Indirect Effects of Low-Energy Electrons on DNA


Elahe Alizadeh, Thomas M. Orlando, and Leon Sanche p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 379
The Dynamics of Molecular Interactions and Chemical Reactions at
Metal Surfaces: Testing the Foundations of Theory
Kai Golibrzuch, Nils Bartels, Daniel J. Auerbach, and Alec M. Wodtke p p p p p p p p p p p p p p p p 399
Molecular Force Spectroscopy on Cells
Baoyu Liu, Wei Chen, and Cheng Zhu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 427
Mass Spectrometry of Protein Complexes: From Origins
to Applications
Shahid Mehmood, Timothy M. Allison, and Carol V. Robinson p p p p p p p p p p p p p p p p p p p p p p p p p p 453
Low-Temperature Kinetics and Dynamics with Coulomb Crystals
Brianna R. Heazlewood and Timothy P. Softley p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 475
Early Events of DNA Photodamage
Wolfgang J. Schreier, Peter Gilch, and Wolfgang Zinth p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 497
Physical Chemistry of Nanomedicine: Understanding the Complex
Behaviors of Nanoparticles in Vivo
Lucas A. Lane, Ximei Qian, Andrew M. Smith, and Shuming Nie p p p p p p p p p p p p p p p p p p p p p 521
Time-Domain Ab Initio Modeling of Photoinduced Dynamics at
Nanoscale Interfaces
Linjun Wang, Run Long, and Oleg V. Prezhdo p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 549
Toward Design Rules of Directional Janus Colloidal Assembly
Jie Zhang, Erik Luijten, and Steve Granick p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 581
Charge TransferMediated Singlet Fission
N. Monahan and X.-Y. Zhu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 601
Upconversion of Rare Earth Nanomaterials
Ling-Dong Sun, Hao Dong, Pei-Zhi Zhang, and Chun-Hua Yan p p p p p p p p p p p p p p p p p p p p p p 619

vi Contents
PC66-FrontMatter ARI 4 March 2015 12:22

Computational Studies of Protein Aggregation: Methods and


Applications
Alex Morriss-Andrews and Joan-Emma Shea p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 643
Experimental Implementations of Two-Dimensional Fourier
Transform Electronic Spectroscopy
Franklin D. Fuller and Jennifer P. Ogilvie p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 667
Electron Transfer Mechanisms of DNA Repair by Photolyase
Dongping Zhong p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 691
Vibrational Energy Transport in Molecules Studied by
Access provided by b-on: Universidade de Aveiro (UAveiro) on 06/02/16. For personal use only.

Relaxation-Assisted Two-Dimensional Infrared Spectroscopy


Natalia I. Rubtsova and Igor V. Rubtsov p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 717
Annu. Rev. Phys. Chem. 2015.66:643-666. Downloaded from www.annualreviews.org

Indexes

Cumulative Index of Contributing Authors, Volumes 6266 p p p p p p p p p p p p p p p p p p p p p p p p p p p 739


Cumulative Index of Article Titles, Volumes 6266 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 743

Errata

An online log of corrections to Annual Review of Physical Chemistry articles may be


found at http://www.annualreviews.org/errata/physchem

Contents vii

Potrebbero piacerti anche