arXiv:q-bio/0701037v1 [q-bio.BM] 24 Jan 2007
Role of electrostatic interactions in amyloid β-protein (Aβ) oligomer formation:
A discrete molecular dynamics study
S. Yun1∗, B. Urbanc1∗, L. Cruz1, G. Bitan2, D. B. Teplow2, and H. E. Stanley1
1Center for Polymer Studies, Department of Physics, Boston University, Boston, MA 02215
2Department of Neurology, David Geffen School of Medicine, Brain Research Institute and Molecular Biology Institute, Uni-
versity of California, Los Angeles, CA 90095
Corresponding authors. E-mail: email@example.com, firstname.lastname@example.org
Keywords: Alzheimer’s disease, amyloid β-protein, discrete molecular dynamics, four-bead
protein model, oligomer formation, electrostatic interaction.
Abbreviations: AD, Alzheimer’s disease; Aβ, amyloid β-protein; DMD, discrete molecular
dynamics; EIs, electrostatic interactions.
Pathological folding and oligomer formation of the amyloid β-protein (Aβ) are widely per-
ceived as central to Alzheimer’s disease (AD). Experimental approaches to study Aβ self-
assembly provide limited information because most relevant aggregates are quasi-stable and
inhomogeneous. We apply a discrete molecular dynamics (DMD) approach combined with
a four-bead protein model to study oligomer formation of Aβ. We address the differences
between the two most common Aβ alloforms, Aβ40 and Aβ42, which oligomerize differ-
ently in vitro. Our previous study showed that, despite simplifications, our DMD approach
accounts for the experimentally observed differences between Aβ40 and Aβ42 and yields
structural predictions amenable to in vitro testing. Here we study how the presence of elec-
trostatic interactions (EIs) between pairs of charged amino acids affects Aβ40 and Aβ42
oligomer formation. Our results indicate that EIs promote formation of larger oligomers in
both Aβ40 and Aβ42. Both Aβ40 and Aβ42 display a peak at trimers/tetramers, but Aβ42
displays additional peaks at nonamers and tetradecamers. EIs thus shift the oligomer size
distributions to larger oligomers. Nonetheless, the Aβ40 size distribution remains unimodal,
whereas the Aβ42 distribution is trimodal, as observed experimentally. We show that struc-
tural differences between Aβ40 and Aβ42 that already appear in the monomer folding, are
not affected by EIs. Aβ42 folded structure is characterized by a turn in the C-terminus that
is not present in Aβ40. We show that the same C-terminal region is also responsible for
the strongest intermolecular contacts in Aβ42 pentamers and larger oligomers. Our results
suggest that this C-terminal region plays a key role in the formation of Aβ42 oligomers and
the relative importance of this region increases in the presence of EIs. These results suggest
that inhibitors targeting the C-terminal region of Aβ42 oligomers may be able to prevent
oligomer formation or structurally modify the assemblies to reduce their toxicity.
Alzheimer’s disease (AD) is a progressive brain disorder, clinically characterized by the
accumulation of extracellular amyloid deposits composed of amyloid β-protein (Aβ), intra-
cellular neurofibrillary tangles, and neuronal loss. Recent research supports the hypothesis
that cerebral Aβ accumulation is the primary cause of neurotoxicity in AD . Accumulating
evidence suggests that Aβ oligomers and prefibrillar aggregates are the proximal effectors of
neurotoxicity in the early stages of AD [2, 3]. The predominant forms of Aβ found in brains
of AD patients are 40 amino acids long (Aβ40) and 42 amino acids long (Aβ42). Aβ42 is
linked particularly strongly with AD. Genetic studies have shown that autosomal dominant
forms of AD invariably involve increased production of Aβ or an increased Aβ42/Aβ40 con-
centration ratio . Aβ42 forms fibrils at significantly higher rates than does Aβ40 [5, 6]
and Aβ42 self-association produces structures that are more neurotoxic than those formed
by Aβ40 . Experimentally, there is a distinct difference in oligomerization pathways of
Aβ40 and Aβ42 . In vitro experiment using the techniques, photo-induced cross-linking of
unmodified proteins (PICUP), size-exclusion chromatography, dynamic light scattering, cir-
cular dichroism spectroscopy, and electron microscopy showed that Aβ exists as monomers,
dimers, trimers, tetramers, and larger oligomers in rapid equilibrium. The Aβ40 oligomer
size distribution comprises monomer, dimer, trimer, and tetramer, in similar amounts, and
few higher-order oligomers. The Aβ42 distribution is multimodal, displaying a prominent
peak of pentamers/hexamers and smaller peaks of dodecamers and octadecamers .
Detailed, quantitative analysis of the three-dimensional structures, energetics, and dy-
namics of oligomer formation are necessary steps toward a molecular understanding of Aβ
assembly and neurotoxicity. During the formation of fibrils, oligomers of different sizes co-
exist with monomers and larger aggregates such as protofibrils  and fibrils. The relative
amounts of each oligomer type are small, which makes determination of the structural prop-
erties of the oligomers difficult. Computer simulations, in contrast, are not subject to the
same kinds of problems, allowing small oligomers to be studied at full atomic resolution (for
recent reviews, see , , and ).
Conventional “all-atom” molecular dynamics (all-atom MD) simulations with explicit
solvent which take account of all the protein and solvent atoms give the most detailed
information. However, aggregation studies using all-atom MD with explicit solvent are
currently limited to either aggregation of small number of Aβ fragments such as three Aβ(16-
22) peptides  or stability studies of various Aβ dimers with predetermined structures [13,
14]. Tarus et al. used a protocol based on shape complementarity to determine the initial
Aβ10−35dimer structure and showed that the peptide dimers are stabilized primarily through
hydrophobic interactions . Huet et al. studied Aβ40 and Aβ42 dimers and their A21G
conformers, starting from their fibrillar conformations and found various possible topologies
of dimers in equilibrium . Keeping track of positions and velocities of all the atoms at
every time step is computationally expensive. Consequently, the times simulated by the
all-atom MD simulations are limited to few microseconds . However, protein folding
and aggregation usually occur on time scales larger than milliseconds. To overcome this
limitation, we use fast and efficient discrete molecular dynamics (DMD) simulations 
with a simplified four-bead protein model and implicit solvent. DMD is a simplified version
of MD using combination of square-well potentials. The DMD approach with a simplified
protein model and implicit solvent increases the efficiency of protein folding and aggregation
studies by a factor of ∼107compared to the all-atom MD .
The idea of applying the DMD approach to study protein folding was proposed in 1996
by Zhou et al. . Soon after, the method was combined with a one-bead protein model to
study folding of a model three-helix bundle protein [18, 19, 20, 21, 22]. In 2004, Peng et al.
used DMD with two-bead protein model to study aggregation of an ensemble of 28 Aβ40
peptides into a fibrillar structure . Smith and Hall introduced four-bead protein model
in combination with the DMD, and showed a cooperative transition of a polyalanine chain
into an α-helical conformation without any a priori knowledge of the native state [24, 25].
Using the four-bead protein model and hydrogen bond interactions in combination with the
DMD on a single 16-residue polyalanine chain, Ding et al. demonstrated a temperature-
induced conformational change from the α-helix to the β-hairpin conformation . Urbanc
et al. studied folding and dimer formation using DMD with the four-bead protein model,
and investigated stability of dimer conformations predicted by DMD approach using all-
atom MD simulations . Lam et al. used the same model to study the Aβ42 folding
and its temperature dependence . The results of Lam et al. were in a good qualitative
agreement with an all-atom study using implicit solvent  and, importantly, consistent
with the temperature dependence of Aβ secondary structure, experimentally determined by
Gursky and Aleshkov and Lim et al. [30, 31].
Recently, we studied oligomer formation using a four-bead model with backbone hydro-
gen bond interactions and the amino acid-specific hydropathic interactions, but no effective
EIs . We observed that dimers are the most abundant among the low molecular weight
Aβ40 oligomers and that the frequency of trimers and higher-order oligomers decreases
monotonically. In contrast, the Aβ42 oligomer size distribution was bimodal, with signifi-
cantly more pentamers than Aβ40. Multimodal and unimodal oligomer size distributions are
discriminating properties of Aβ42 and Aβ40, respectively, as observed in vitro by PICUP .
Experimentally-detected pentamer/hexamer Aβ42 oligomers were termed paranuclei. Ex-
istence of Aβ42 paranuclei and their homotypical assemblies, “oligo-paranuclei”, has been
independently confirmed by a combination of ion mobility and mass spectrometry . Im-
portantly, paranucleus-like assemblies have been detected in vivo in the form of dodecameric
assemblies termed ADDLs , globulomers , and Aβ⋆56 . In vitro studies showed
that oxidation of M35 blocks Aβ42 paranucleus formation . Aβ without oxidated M35
displays both accelerated [38, 39] and delayed  fibrillogenesis rate relative to wild type
Aβ. Analysis of intramolecular contacts in Aβ40 and Aβ42 pentamers in our in silico study
also showed that M35 forms contacts with I41 and A42 in Aβ42 , providing an expla-
nation of the above experimental results . In addition, our prior study indicated that
Aβ42 monomers but not Aβ40 monomers are characterized by a turn structure, centered
at G37-G38, and that this turn structure was more prominent in large oligomers . This
result is consistent with recent proteolysis results using Aβ40 and Aβ42 .
There is indirect in vitro as well as in silico evidence suggesting that EIs play a significant
role in both Aβ folding [41, 42, 43, 44] and Aβ fibril formation [45, 46, 47]. In the present
study, we follow the protocols of our previous study  using DMD and four-bead protein
model with amino acid-specific interactions  to elucidate the role of EIs between pairs of
charged amino acids (D, E, K, and R) on folding and oligomerization of Aβ40 and Aβ42.
For our simulation method, we use DMD simulations . The main simplification in this
method is to replace continuous interparticle potentials by a square-well or a combination
of square-well potentials. As a result, particles move along straight lines with constant
velocities until a pair of particles reaches a distance at which the interparticle potential is
discontinuous. A collision event then takes place during which the velocities and directions
of the particles are updated while preserving the total kinetic energy, momenta, and angular
momenta. Because DMD is event-driven, it is faster than all-atom MD. Our DMD approach
using coarse-grained protein models has been described in detail elsewhere .
Here we use a four-bead protein model with backbone hydrogen bonding, effective hy-
dropathic interactions and EIs. We use the four-bead model with hydrogen bonding, intro-
duced by Ding et al. , then further generalized by Urbanc et al.  to include amino
acid-specific hydropathic and electrostatic interactions. In the four-bead model, the back-
bone is represented by three beads, corresponding to the amide (N), the α-carbon (Cα),
and the carbonyl (C′) groups. Each side-chain is represented by one bead (Cβ). G, which
lacks a side-chain, has no Cβbead. As the carbonyl oxygen and the amide hydrogen are not
explicitly present, an effective backbone hydrogen bond is introduced between the nitrogen
atom Niof the i−th amino acid and the carbon atom Cjof the j −th amino acid. Because
the solvent is not explicitly present in our DMD approach, effective interactions between
the side-chain atoms are introduced to mimic the solvent effects. The relative strength of
hydropathic interactions between pairs of side chain beads is based on the Kyte-Doolittle
hydropathy scale . When two hydrophobic side chain beads are within the interaction
range of 0.75 nm, they interact through a one-step attractive potential. When two hy-
drophilic side chain beads are within the same interaction distance, they interact through a
one-step repulsive potential. In our model, the hydrophobic amino acids are A, C, F, L, M,
I, and V. The hydrophilic amino acids are D, E, H, K, N, Q, and R. The side chains of the
remaining amino acids G, P, S, T, W, and Y interact only through a hard-core repulsion.
The EIs are implemented by assigning a two-step potential with two interaction distances,
0.60 nm and 0.75 nm, as described elsewhere . When two beads with the same charge are
at the interaction distance, they interact through a positive (repulsive) two-step potential.
Two oppositely charged beads interact through a negative (attractive) two-step potential.
We set the potential energy of the hydrogen bond, EHB, which in proteins is typically in
the range 1−5 kcal/mol , to unit energy (EHB= 1). We set the potential energy of the
hydropathic interactions EHP = 0.3. Experimental free energy of salt bridge formation is
estimated to be in the range 0.7− 1.7 kcal/mol , thus we choose the potential energy of
EIs, ECH= 0.6. Using the unit of temperature EHB/kBwhere kBis Boltzmann’s constant,
we estimate that T = 0.15 is appropriate for simulating physiological temperatures. We
perform DMD simulations in the canonical ensemble (NVT) using the Berendsen thermostat
Because we treat the solvent in our DMD approach implicitly, the effective interactions
between the side-chain beads include not only protein-protein but also protein-solvent in-
teractions. Thus, there are no generic interaction parameters that would be independent
of the environment. Moreover, because different proteins may interact with the solvent in
different ways, the implicit effect of the solvent and thus the interaction parameters may
depend on the particular protein sequence. The complexity of protein-protein and protein-
solvent interactions represents a challenge in protein structure prediction where even the
most successful specialized models fail on certain targets . The question of how general
is a particular choice of interaction parameters in our DMD approach is a topics of future
Results and Discussion
We apply the four-bead model with hydrogen bonding and amino acid-specific interac-
tions due to hydropathy and charge and use DMD with implicit solvent to study Aβ40
and Aβ42 oligomer formation. Due to simplifications in protein description and implicit
solvent, our DMD approach is efficient enough to allow for a study of the whole process
starting from unfolded separated peptides to formation of quasi-stable Aβ oligomers with
well-defined size distributions. In our protein model, each side chain is replaced by at most
one bead, a significant simplification considering side-chain diversity. However, recent devel-
opments in understanding of protein folding and assembly show that despite the complexity
of the process as a whole, the underlying fundamental physics is simple [53, 54]. It is believed
that the patterns of hydrophobic and hydrophilic residues, rather than the highly specific
characters of the individual residues involved, play an important role [55, 56]. This is consis-
tent with our prior simulation results where we showed that amino acid-specific interactions
due to hydropathy itself are sufficient  for accounting for the experimentally observed 
oligomer size distribution differences between Aβ40 and Aβ42. Here, we apply the same
model, with the addition of Coulombic interactions between pairs of charged amino acids,
to study the effect of EIs on Aβ40 and Aβ42 oligomer formation.
The primary structure of Aβ42 is DAEFRHDSGYEVHHQKLVFFAEDVG
SNKGAIIGLMVGGVVIA. The primary structure of Aβ40 is identical, except that the last
two amino acids, I and A, are missing. We define the following peptide regions: (i) the
N-terminal region D1-K16 (NTR); (ii) the central hydrophobic cluster L17-A21 (CHC); (iii)
the turn A region E22-G29 (TRA); (iv) the mid-hydrophobic region A30-M35 (MHR); (v)
the turn B region V36-V39 (TRB); and (vi) the C-terminal region V40/V40-A42 (CTR).
The CTR of Aβ40 consists of only one amino acid, V40.
We simulate eight oligomerization trajectories for Aβ40 and Aβ42 each, starting from
spatially separated peptides. Each initial configuration consists of 32 Aβ40 (Aβ42) peptides
with a zero potential energy and with randomized spatial positions and randomized initial
velocities of atoms within a cubic box of side 25 nm. The molar concentration is ∼ 3.4 mM.
This initial setup follows the protocol of our prior publication . The concentration in
our simulation is 10 − 100 times higher than that studied experimentally . Lowering the
concentration is possible only at a high cost of efficiency of our approach. As shown in a
recent study by Nguyen and Hall , lowering the concentration may give rise to α-helical
aggregates at low temperatures, possibly altering the assembly pathways, a problem to be
addressed in future studies.
The energy is in our approach normalized to the potential energy of the hydrogen
bond EHB = 1. Temperature is expressed in units of energy and also normalized to
EHB. The maximal potential energy of the hydrophobic/hydrophilic interaction is set to
ECH = 0.6/EHB = 0.6. The N-terminal amine group and the C-terminal carboxyl group
Hydrogen bonding is the same for all amino acids and represents the basic interaction
needed to model the secondary structure, α-helix and β-strand, formation. When only
the hydrogen bond interactions are allowed (EHB= 1,EHP = 0, and ECH = 0), a single
planar β-sheet aggregate is formed [11, 27]. Thus, only hydrogen bond interaction is not
enough for description of spherical oligomers with only small amounts of secondary structure.
Recently, we introduced the effective hydrophobic/hydrophilic interactions which are amino
acid-specific to mimic the effect of aqueous solution . Using the hydrogen bonding and
effective hydropathic interactions but no EIs (EHB= 1,EHP= 0.3, and ECH= 0), we found
spherical Aβ aggregates with a dense hydrophobic core and with the hydrophilic N-termini
comprising the surface .
The aim of the present study is to explore the effects of EIs on oligomer formation of
Aβ40 and Aβ42. The question of how EIs affect the aggregation is intriguing because most
of the charged amino acids are at the N-part of the molecule: six of nine charged amino
acids are within the D1-K16 fragment as opposed to the hydrophobic residues which are
concentrated in the remaining fragment L17-V40/A42. Fig. 1 shows typical conformations of
a folded monomer, dimer, and pentamer of Aβ42 in the absence and presence of EIs. Similar
conformations are found in the case of Aβ40 (data not shown). We observe various topologies
at a fixed oligomer size, which is consistent with findings by Huet et al. . To gain more
quantitative insight into the oligomer formation and structure, we quantify the oligomer
size distributions, calculate the intra- and intermolecular contact maps, secondary structure
propensities, and Ramachandran plots for each Aβ40 and Aβ42 alloform separately.
Aβ40 and Aβ42 oligomer size distributions
All simulations are 10 million simulation steps long. Initially, all the oligomer size distri-
butions are peaked at monomers and the oligomer size distributions of Aβ40 and Aβ42 are
equivalent. The difference between Aβ40 and Aβ42 size distributions develops steadily with
increasing simulation time and at ∼ 6 million steps the difference between Aβ40 and Aβ42
oligomer size distributions becomes statistically significant as determined by applying the
χ2-test . When comparing oligomer size distributions of each alloform separately at 8.0,
8.5, 9.0, 9.5, and 10.0 million steps, we find that within this time window size distributions
do not differ significantly. However, the number of monomers and oligomers of all sizes
is variable. Each of the final oligomer size distributions is obtained by first average over
all 8 trajectories at a fixed simulation time, and then the resulting ensemble averages are
averaged over the simulation times of 8.0, 8.5, 9.0, 9.5, and 10.0 million steps.
We have shown previously that Aβ40 and Aβ42 oligomer size distributions in the absence
of EIs (ECH= 0) are significantly different (Fig. 2(a)) . Aβ40 and Aβ42 oligomer size
distributions in the presence of EIs (ECH = 0.6) are significantly shifted towards larger
oligomers, as shown in Fig. 2(b). Comparing the Aβ40 and Aβ42 oligomer size distributions
by applying the χ2-test, we conclude that in the presence of EIs, the distributions are
significantly different (p < 0.01).
In the presence of EIs, the average size of Aβ40 oligomers increases from 3.0 to 5.2
molecules, and the average size of Aβ42 oligomers increases from 3.7 to 6.2 molecules. These
results suggest that EIs facilitate aggregation. Aβ42 forms significantly more nonamers and
larger oligomers compared to Aβ40. The Aβ40 size distribution is unimodal with a peak at
tetramers. The Aβ42 distribution contains a trimer peak and two additional peaks, at n = 9
(nonamer) and n = 14 (tetradecamer), neither of which is present in the Aβ40 distribution.
A multimodal oligomer size distribution was observed experimentally with Aβ42, but not
with Aβ40 .
In our simulations, the N- and C-termini are uncharged, whereas in the experimental
studies, the N-terminus is positively charged (NH+
3) and the C-terminus is negatively
charged (COO−) [7, 42]. Observation of high-order oligomers in our simulations is con-
sistent with in vitro results in which the C-terminal carboxyl group was replaced by the
electrostatically neutral carboxamide, resulting in a greater abundance of high molecular
weight oligomers . Our simulation results, in combination with experimental findings,
thus suggest that inclusion of charged termini, in particular the C-terminal negative charge,
will moderate formation of Aβ oligomers. This hypothesis will be tested in future compu-
tational and experimental studies.
Secondary structure of Aβ monomers
We calculate the secondary structure propensities on each folded monomer separately
using the STRIDE program  and then average over different conformations to obtain the
average values of the α-helix, turn, and β-strand propensities per amino acid. At 1 million
(M) step, the potential energy of individual monomers is stabilized (data not shown), thus
we consider monomers to be in a folded state at 1M step.
Folded monomers do not have a significant amount of α-helix structure (data not shown).
Figs. 3 (a) and (b) show the turn propensity per amino acid for folded Aβ40 and Aβ42
monomers in the absence and presence of EIs. A dramatic effect of EIs on the turn propen-
sities in both alloforms is observed in the region A21-A30. In the absence of EIs this region is
characterized by two turns, the first at A21-V24 and the second at S26-G29. In the presence
of EIs, only a single turn within the region V24-G29 remains.
Figs. 3 (c) and (d) show the β-strand propensity per amino acid for folded Aβ40 and
Aβ42 monomers in the absence and presence of EIs. As a result of EIs in both alloforms, the
regions A21-D23 and K28-I31 show an increased β-strand propensity. In Aβ40 monomers
the regions A2-F4 and L34-G38 show a decreased β-strand propensity due to EIs. In Aβ42
monomers the regions R5-H6 and L34-V39 show a slightly decreased β-strand propensity
due to EIs. Notice that the β-strand propensity per amino acid is below 40% for Aβ40 and
below 30% for Aβ42. The number of turns and consequently also the number of β-strand
regions in the Aβ42 monomer (5) is bigger than in the Aβ40 monomer (4), indicating a more
compact structure of the Aβ42 monomer as compared to the Aβ40 monomer, a consequence
of a strongly hydrophobic CTR in Aβ42, which introduces an additional turn centered at
G37-G38. The average turn and β-strand contents of Aβ40 and Aβ42 folded monomers
are displayed in Table I. These contents are calculated from propensities per residue by
averaging over all residues in the the peptide. Table I shows that for both Aβ40 and Aβ42
the average turn content is in the range 43-45% while the average β-strand content is in
the range 10-12%. Neither the average turn nor the average β-strand content is strongly
affected by EIs.
The above results suggest that even in the presence of EIs, the Aβ monomer is a collapsed
coil with several turns and some β-strand but no α-helical structure, which is in agreement
with existing experimental studies [30, 41, 60]. The β-strand propensity of Aβ40 monomer
as shown in Fig. 3(c) is also consistent with a recent study of Aβ40 folding using a scanning
tunnelling microscopy that showed monomers folded into 3 or 4 domains with some β-strand
Intramolecular contacts of folded Aβ monomers
Here we discuss the effect of EIs on the intramolecular contacts among pairs of amino acids
of folded monomers. Initially, monomer peptides are in zero-potential energy (unfolded)
conformations. At 0.1M steps, over 60% of peptides (65.9% for Aβ40, 60.5% for Aβ42) are
folded. We describe the regions of the most important contacts between pairs of amino acids.
We first describe “short-range” contacts formed within the regions TRB, MHR, and TRA.
Then, we describe the “long-range” contacts between the regions CHC-CTR, CHC-MHR,
Previous results  showed that while Aβ40 and Aβ42 monomers both display strong
contacts within the TRA region, strong contact in the TRB region with a turn centered at
G37-G38 are characteristic of Aβ42 only. This in silico difference between Aβ40 and Aβ42
folding is consistent with experimental findings by Lazo et al. .
In Fig. 4, we compare the intramolecular contact maps of Aβ40 and Aβ42 in the presence
and absence of EIs. Fig. 4 shows the region containing the strongest contact V36-V39 as
reported in our previous study (rectangle 1 in (a) and (c)) . In Aβ40 the contacts
between the amino acid regions L34-V36 and V39-V40 are significantly weaker than similar
contacts between L34-V36 and V39-A42 in Aβ42. EIs do not affect contacts in the TRB
region (rectangle 1 in (b) and (d)). This result suggests that EIs do not alter the contacts
that contribute to differences between Aβ40 and Aβ42 folding in the CTR.
A few important contacts in both alloforms in the MHR, concentrated around the
strongest contact I31-L34, bring into proximity the two MHR regions A30-I32 and L34-
V36 and are not affected by EIs (rectangle 3 in (a)-(d)). The formation of these contacts
within the MHR is promoted by G33 because glycines are typically associated with a high
turn/loop propensity. Contacts between the CTR and MHR are present in both Aβ40 (rect-
angle 2 in (a)) and Aβ42 (rectangle 2 in (c)), but are significantly stronger in Aβ42. These
contacts are not affected significantly by EIs (rectangle 2 in (b) and (d)).
The central and most abundant contacts in folded monomers of both alloforms are formed
as a consequence of the formation of the turn involving the TRA region (rectangles 4 and 7
in (a)-(d)). The TRA region contains charged amino acids E22, D23, and K28, thus it is
expected that EIs will influence the contacts in this region. A strong contact A21-V24 in the
TRA region becomes weaker as a result of EIs (rectangle 7 in (a)-(d)), which is consistent
with the effect of EIs on the turn propensity in this region, changing a two-turn region into
a one-turn region. Formation of contacts within the TRA brings into proximity the CHC
and MHR (rectangles 5 in (a)-(d)). In both alloforms in the absence of EIs, the CHC region
makes contacts with the MHR with F19-I31 as the strongest contact (rectangles 5 in (a)
and (c)). EIs enhance the contacts within and around the TRA region in both alloforms,
making contacts between the regions L17-D23 and K28-I32 (rectangles 5 in (b) and (d))
stronger. This enhanced feature is a consequence of a salt bridge formation between the op-
positely charged D23 and K28. The TRA region was recently hypothesized to represent the
nucleation region of Aβ folding . This turn has been shown to be important in the fibril
structure [45, 46], suggesting that this region maintains conformational stability throughout
the folding and assembly of Aβ. Our results are consistent with this hypothesis as they show
that formation of contacts within the TRA region induces prominent contacts between the
CHC and MHR, resulting in the highest concentration of intramolecular contacts, involving
the TRA, CHC, and MHR.
In the absence of EIs, the MHR region A30-M35 makes contacts with both the CHC
(rectangle 5 in (a)-(d)) and CTR (rectangle 2 in (a)-(d)). These contacts do not change
significantly in the presence of EIs. The difference between Aβ40 and Aβ42 is that in Aβ40
contacts between the regions A30-I32 and L34-V36 are stronger than the contacts between
A30-I35 and V39-V40, while in Aβ42 the contacts between the regions A30-I35 and V39-A42
are dominant. This result suggests that in Aβ42 folding the CTR plays a prominent role,
while in Aβ40 the contacts within the MHR and between MHR and CHC regions are more
The contacts between the K16-F19 and E11-H14 become more pronounced in the pres-
ence of EIs due to the EI between the negatively charged E11 and positively charged K16
(rectangle 8 in (a)-(d)). A weaker group of contacts within the NTR between F4-H6 and
Y10-V12 is a result of a turn centered at D7-G9 and hydrophobic attraction F4-V12. These
contacts are very weak in the absence of EIs (rectangle 9 in (a) and (c)) but become stronger
in the presence of EIs due to salt bridge R5-E11 (rectangle 9 in (b) and (d)).
Long-range contacts between V39-V40 and CHC are observed in both Aβ40 and Aβ42 in
the absence of EIs (rectangle 6 in (a) and (c)). These contacts remain strong in the presence
of EIs (rectangle 6 in (b) and (d)). In Aβ42, these contacts are stronger than in Aβ40,
both in the absence and presence of EIs. Another region of long-range contacts is observed
in both alloforms between the K16-F20 and D1-F4 in the absence of EIs (rectangle 10
in (a) and (c)). These contacts become more pronounced in the presence of EIs due to
electrostatic attraction between the negatively charged D1 and E3 and positively charged
K16 (rectangle 10 in (b) and (d)). The long-range contacts between CTR and A2-F4, and
MHR and A2-F4 are also present in both Aβ40 and Aβ42 but are weaker than the others
and are not significantly influenced by EIs.
Time progression of Aβ folding events
Fig. 5 shows time evolution of Aβ40 and Aβ42 monomer folding events in the presence
of EIs. Initially, Aβ40 and Aβ42 monomers are in zero potential energy, random coil con-
formations. At 1k simulation steps, contacts are formed between L34-V36 and CTR in both
Aβ40 and Aβ42. However, only in Aβ42, these contacts are associated with a turn structure
in the TRB region as described in the previous section. At 2k steps, the contacts between
regions CHC and TRA, CHC and MHR, CHC and CTR develop in both Aβ40 and Aβ42.
These contacts are associated with a turn structure in the TRA region in both Aβ40 and
Aβ42. At 4k steps, contacts between NTR and CHC develop in Aβ40. At 8k steps, as the
contacts between NTR and CHC in Aβ40 are more pronounced, these contacts also emerge
in Aβ42. At 0.1M steps, the long-range contacts between NTR and CTR are formed in both
Aβ40 and Aβ42. Using the regions defined in Figs. 4 (b) and (d), the time progression of
contacts follows the numbering 1, 2, 3, ... 10, i.e., Aβ folding starts at the C-terminal and
progresses towards the N-terminal. In Aβ40, the turn structure in the TRA region is the
first structural element that is formed in the process of folding, supporting the hypothesis
of Lazo et al.  stating that the region 21-30 nucleates Aβfolding. However, in Aβ42 the
turn structure in the TRB region is formed before the formation of the turn structure in the
TRA region. This result suggests that in Aβ42 the TRB region nucleates the folding prior
to formation of contacts in the TRA region.
Secondary structure of Aβ pentamers and larger oligomers
In our previous work , we reported the secondary structure difference between Aβ40
and Aβ42 pentamers that can be found in the NTR and CTR. Aβ42 pentamers displayed an
increased β-strand propensity at the V39-I41, while Aβ40 pentamers showed an increased
β-strand propensity at the A2-F4. Our present data show that these differences remain
intact in the presence of EIs.
Pentamers and larger oligomers do not have any significant amount of α-helix structure
(data not shown). Figs. 6 (a) and (b) show the turn propensity per amino acid for Aβ40
and Aβ42 pentamers and larger oligomers in the absence and presence of EIs. EIs do not
affect the turn propensity significantly. In Aβ42, a slight increase in the turn propensity
due to EIs is found in the region R5-Y10.
Figs. 6 (c) and (d) show the β-strand propensity per amino acid for Aβ40 and Aβ42
pentamers and larger oligomers. In both alloforms, the β-strand propensity in the region
K28-I31 slightly increases and in the region L34-G38 decreases due to EIs. In the presence
of EIs, the β-strand propensity in the CHC increases in Aβ40, while it decreases in Aβ42
pentamers and larger oligomers.
We also calculate the average turn and β-strand contents within Aβ40 and Aβ42 pen-
tamers and larger oligomers in the absence and presence of EIs. The data is shown in Table
II. The average contents are calculated from propensities per residue by averaging over all
residues in the peptide. The average turn content is in the range 41-45% and the average
β-strand content is in the range 11-13%. There is no significant difference between the two
alloforms and no significant effect due to EIs.
These results show that pentamers and larger oligomers in our study have a globular
structure dominated by turns and loop and some β-strand propensity. EIs change the
relative β-strand propensities of some regions, but do not affect significantly the overall
Ramachandran plots of selected amino acids within the Aβ42 pentamers and higher
Because our protein model as well as the interactions are simplified, we tested Aβ42
oligomer conformers by calculating the Ramachandran plots. We selected the following 10
amino acids from different regions of the protein: D1, Y10, F19, E22, D23, S26, K28, M35,
I41, and A42.
Our results shown in Fig. 7 indicate that both in the absence and presence of EIs, the
most populated (Φ, Ψ) region corresponds to the β-sheet region. The exceptions are D1
and A42, the N- and C-terminal amino acids, due to an increased flexibility at the two
termini, and E22. Interestingly, E22 shows a substantially higher propensity to form a
right-handed alpha-helix. Our results show that EIs do not affect these plots in a significant
way. These results are in qualitative agreement with Aβ dimer analysis of Huet et al. who
studied Aβ dimer conformations by all-atom MD , suggesting that our four-bead model
yields relatively realistic set of Φ and Ψ angles and thus adequately accounts for the protein
Tertiary structure of pentamers and larger oligomers
The tertiary structure of Aβ molecules within pentamers and larger oligomers (Fig. 8)
is highly reminiscent of the structure of individual monomers (compare Figs. 4 and 8), sug-
gesting that no major refolding events are needed in monomers prior to oligomer formation.
However, there is less involvement of the N-terminal amino acids and more intramolecular
contacts involving the C-terminal amino acids in Aβ molecules comprising pentamers and
larger oligomers of both alloforms.
There are significant differences between Aβ40 and Aβ42 intramolecular contact maps
of pentamers and larger oligomers. The differences between Aβ40 and Aβ42 in the absence
of EIs have been described in our previous work  and can be observed comparing the
relative importance of the CHC and CTR: in Aβ42 the contacts of CTR with MHR and
CHC are dominant, while in Aβ40 the CHC plays a dominant role. In Aβ40 (Figs. 8(a)
and (b)) the contacts in regions marked by rectangles 1, 3, 4, and 5 get weaker due to
EIs, while the opposite is true in Aβ42 (Figs. 8(c) and (d)), where the contacts within the
rectangles 1, 2, 3, 4, and 5 get stronger. This effect of EIs on the intramolecular contacts
can only be observed in pentamers and larger oligomers and not in unassembled monomers.
Aβ42 pentamers and larger oligomers, in the presence of EIs, have significantly stronger
intramolecular contacts than Aβ40, suggesting that Aβ42 pentamers and larger oligomers
are intrinsically more stable than their Aβ40 counterparts.
Fig. 7 shows Ramachandran scattering plot on pentamers and larger oligomers of Aβ42.
As seen from contact map analysis, in the presence of EIs, D1s are more populated in β-sheet
region, which is the upper left corner.
Quaternary structure of pentamers and larger oligomers
Intermolecular contact maps indicate contacts among different Aβ molecules within an
oligomer that are most important in oligomer formation. Previously, we showed that in
Aβ40 pentamers, pairs of the CHC regions show the highest propensity to interact, whereas
in Aβ42 pentamers the most frequent contacts are between the CTR of one peptide and
the CHC and MHR of the other . That result indicated that the CTRs are critically
involved in aggregation of Aβ42 but not Aβ40.
Fig. 9 shows intermolecular contact maps of pentamers and larger oligomers of Aβ40 and
Aβ42 in the absence ((a) and (c)) and presence ((b) and (d)) of EIs. Perhaps the most
surprising overall observation is that the intermolecular contacts that involve the CHC, i.e.,
contacts between pairs of CHCs (rectangle 3 in (a)-(d)), between the CHC and MHR (rect-
angle 5 in (a)-(d)), and between the CHC and CTR (rectangle 6 in (a)-(d)), become weaker
as a consequence of EIs in both alloforms, but this weakening is more pronounced in Aβ40
oligomers. This weakening of the contacts involving the CHC due to EIs is surprising because
the CHC is surrounded by charged residues (K16, E22, and D23). Thus, we would expect
CHCs to interact pairwise in an anti-parallel fashion to maximize the the mutual attraction
involving hydrophobic residues by additional salt bridge formation and thus minimize the
free energy. Instead, our results show that EIs weaken the contacts between pairs of CHCs.
We also showed that EIs promote formation of larger oligomers in both Aβ40 and Aβ42.
These two results combined imply that weakening of the contacts between pairs of CHCs in
Aβ40 oligomers might actually indirectly promote aggregation into larger oligomers.
The only exception to the above observation is the region between D1-R5 and K16-D23
which is rather weak in both alloforms in the absence of EIs, but gets more pronounced in
particular in Aβ42 due to EIs (rectangle 7 in (a)-(d)).
Our results indicate important differences in the way EIs affect Aβ40 and Aβ42 oligomers.
In Aβ40 oligomers the intermolecular contacts between pairs of CTRs (rectangle 1 in (a)
and (b)), between pairs of MHRs (rectangle 2 in (a) and (b)), and between the CTR and
MHR (rectangle 4 in (a) and (b)) remain unaffected by EIs. In Aβ42 oligomers, on the other
hand, the intermolecular contacts in these same regions get stronger even though that part
of Aβ42 (MHR and CTR) is free of charge and thus EIs would not be expected to make a
difference. The strongest increase in the intermolecular contact intensity in Aβ42 oligomers
is between pairs of CTRs (rectangle 1 in (b) and (d)) and the second strongest is between the
CTR and MHR (rectangle 4 in (b) and (d)). Thus, in Aβ42 oligomers the contacts involving
the CHCs get weaker and the contacts involving the CTRs get stronger due to EIs, resulting
in a significantly larger oligomers. These results suggest that in Aβ42 the CTRs are most
important for intermolecular assembly into pentamers and larger oligomers. The lack of
strong intermolecular contacts involving CTRs in Aβ40 suggests that the CTRs are also
the main source of the differences between Aβ40 and Aβ42 oligomer formation. Recently,
the importance of the intermolecular CHC contacts in Aβ40 versus the intermolecular CTR
contacts in Aβ42 was observed experimentally by Maji et al. , in agreement with our
present in silico results, suggesting the biological relevance of our DMD approach which is
able to capture the essential differences between Aβ40 and Aβ42 oligomerization.
Intra and intermolecular hydrogen bonds in pentamers and larger oligomers
Here we address the question of how much hydrogen bonds contribute to intra- and in-
termolecular contacts in pentamers and larger oligomers. We first calculate the probabilities
for forming an intra- or intermolecular hydrogen bond per amino acid. The amino acids that
are most hydrogen bond active are shown in Tables III and IV. Our results show that even
for the amino acids that are most likely to form hydrogen bonds, probabilities are smaller
than 0.20. The sum of intra- and intermolecular probabilities per amino acid does not exceed
0.30/0.40, which is consistent with the β-strand propensity per amino acid (Fig. 6).
Fig. 10 shows the intramolecular hydrogen bond contacts of Aβ40 ((a) and (b)) and Aβ42
((c) and (d)) pentamers and larger oligomers in the absence ((a) and (c)) and presence ((b)
and (d)) of EIs. These intramolecular hydrogen bond maps are normalized to the highest
value of intramolecular hydrogen bond formation probability, which is < 0.09. The regions
with the highest amount of hydrogen bonds can be found between the regions K16-V24 and
K28-V40. In Aβ42 oligomers some additional hydrogen bonds are formed between the MHR
and CTR and between the CHC and CTR. EIs increase the hydrogen bond probabilities
within the TRA region and between the CHC and MHR due to salt bridge D23-K28. This
effect is more pronounced in Aβ40. Interestingly, the strongest intramolecular hydrogen
bond occurs in Aβ42 oligomers between F4 and V12, possibly stabilized by proximity of
oppositely charged R5 and E11. Why this same hydrogen bond is missing in Aβ40 oligomers
may be understood by observation that in Aβ40 the region A2-F4 forms a β-strand that is
in contact with the CHC and thus the charged NTR residues (E3 and R5) are interacting
with the charged residues K16 and E22, preventing R5-E11 from interacting and breaking
the F4-V12 hydrogen bond.
The intermolecular hydrogen bonds of Aβ40 ((a) and (b)) and Aβ42 ((c) and (d)) pen-
tamers and larger oligomers in the absence ((a) and (c)) and presence ((b) and (d)) of EIs
are presented in Fig. 11. These intermolecular contact maps are normalized to the highest
value of intermolecular hydrogen bond probability, which is < 0.04. The probability of in-
termolecular hydrogen bond formation is slightly higher in the regions where the contacts
are more pronounced. EIs do not influence the intermolecular hydrogen bond formation in
any significant way.
Our results show that the hydrogen bonds present in Aβ pentamers and larger oligomers
are not specific, indicating that oligomers are not characterized by any particular pattern
of hydrogen bonding. These findings suggest that hydrogen bonding is mostly a secondary
effect occurring as a consequence of hydrophobic contact formation in the regions CHC,
MHR, and CTR.
Because molecular dynamics approach to study proteins using all-atom representation
and explicit solvent is limited to time scales smaller than ∼ 10−6s, we use a simplified but
efficient DMD approach combined with a four-bead protein model and amino acid-specific
interactions that mimic the effects of a solvent . In our prior work we showed that this
approach yields biologically relevant results, which are consistent with existing experimental
findings on Aβ oligomer formation and have predictive power allowing for in vitro and further
in silico testing . In the present work we use the DMD approach to study the effects
of EIs on oligomer formation of Aβ40 and Aβ42. The role of electrostatic interactions,
in particular the salt bridge formation between negatively charged E22/D23 and positively
charged K28 was hypothesized to be important at early stages of folding as well as at later
stages of fibril formation. Thus, it is reasonable to expect that EIs may play an important
role at intermediate stages of oligomer formation.
We analyze the structure of folded Aβ40 and Aβ42 monomers in the presence and absence
of EIs. We show that independent of EIs the two alloforms display differences in folded
structure: in Aβ42 there is an additional turn centered at G37-G38 that is absent in Aβ40,
leading to an increased propensity to form β-strand in the CTR of only Aβ42.Aβ40
monomers also have an additional β-strand in the A2-F4 which is not present in Aβ42. Our
results demonstrate that the differences between the two alloforms are present already at
the stage of folding, prior to assembly. The existence of a turn structure centered at G37-
G38 is consistent with experimental findings by Lazo et al. who showed by using limited
proteolysis that Val39-Val40 in Aβ42 but not in Aβ40 monomer was protease resistant,
indicating that Aβ42 but not Aβ40 monomer was structured in the CTR region . Similar
was a conclusion of the solution NMR study on [Met(O)35]Aβ40 versus [Met(O)35]Aβ42
monomer structure by Riek et al. showing that G29-A42 region is less flexible and thus
more structured in Aβ42 than in Aβ40 . By measuring1Hα,13Cα, and13Cβchemical
shift indices of Aβ40 and Aβ42 Hou et al. recently showed that the C-terminus of Aβ42 but
not of Aβ40 monomer has a tendency to form β-sheet structure  which provides further
evidence that our simulation approach yields biologically relevant results consistent with in
Our results indicate that EIs stabilize a turn in the region D23-K28 by formation of a
D23-K28 salt bridge. A role for EIs in stabilizing this region has been postulated by Lazo
et al.  and further explored using a more complex united atom DMD model  and
all-atom MD in explicit  and implicit solvent . These studies show that Aβ folding
in the region A21-A30 is driven primarily by effective hydrophobic attraction between V24
and the butyl portion of K28, but that EIs help stabilize the region. In our model, due to its
simplicity, the side chains of V24 and K28 do not experience attractive interactions. Despite
the absence of this interaction, we still find this region to be the most structured in both
Aβ40 and Aβ42 monomers stabilized by D23-K28 salt bridge. The D23-K28 salt bridge was
suggested to stabilize the Aβ40 fibril structure by Petkova et al. . In addition, Sciarretta
et al. have shown an increase in the rate of Aβ40-Lactam (D23/K28) fibrillogenesis by 1000-
folds , providing additional experimental evidence supporting a critical role of D23-K28
salt bridge formation.
Comparing the oligomer size distributions of Aβ40 and Aβ42 in the presence of EIs
with those obtained in the absence of EIs  reveals that EIs promote formation of larger
oligomers while maintaining a unimodal Aβ40 size distribution and a multimodal Aβ42
size distribution, as observed in vitro . In our simulations the N- and C-termini are
uncharged in contrast to most experimental studies with positively charged N- and negatively
charged C-termini. Our observation that EIs promote formation of larger oligomers is thus
consistent with results of the experimental study in which the C-terminal carboxyl group
was replaced by the electrostatically neutral carboxamide, resulting in a greater abundance
of high molecular weight oligomers .
It is critical to study the structural changes in oligomers due to EIs and understand which
structural changes are contributing to formation of larger oligomers in both Aβ40 and Aβ42.
Our results indicate that in Aβ40 pentamers and larger oligomers, EIs weaken intramolecular
interactions. In Aβ42, in contrast, the intramolecular contacts in the turn region D23-K28
are enhanced. Surprisingly, in both Aβ40 and Aβ42 oligomers, the intermolecular contacts
involving the CHC are significantly weaker in the presence of EIs. In addition, in Aβ42
oligomers, the contacts involving the CTR and MHR get stronger. These results, combined
with the fact that EIs promote larger oligomers, imply that the intermolecular interactions
between pairs of CHCs in an indirect way oppose the formation of larger oligomers, while
the interactions between pairs of CTRs, and to a smaller extent also pairs of MHRs, promote
formation of larger oligomers. Thus, therapeutic strategies using inhibitors that target the
CTR and MHR may prove successful in either inhibiting formation of toxic Aβ42 oligomers
or inducing structural modifications neutralizing their toxicity.
This work was supported by grants from the National Institutes of Health (AG023661, NS44147,
NS38328, AG18921, and AG027818), grant A04084 from the American Federation for Aging Re-
search, grant 2005/2E from the Larry L. Hillblom Foundation, an Alzheimer’s Association Zenith
Fellows award, and the Petroleum Research Fund. We are thankful to Stephen Bechtel, Jr. for a
1. Hardy, J., and D. J. Selkoe, 2002. The Amyloid Hypothesis of Alzheimer’s Disease: Progress
and Problems on the Road to Therapeutics. Science. 297:353–356.
2. Kirkitadze, M. D., G. Bitan, and D. B. Teplow, 2002. Paradigm shifts in Alzheimer’s disease
and other neurodegenerative disorders: the emerging role of oligomeric assemblies. J. Neurosci.
3. Klein, W. L., W. B. Stine, and D. B. Teplow, 2004. Small assemblies of unmodified amyloid
β-protein are the proximate neurotoxin in Alzheimer’s disease. Neurobiol. Aging. 25:569–580.
4. Sawamura, N., M. Morishima-Kawashima, H. Waki, K. Kobayashi, T. Kuramochi, M. P.
Frosch, K. Ding, M. Ito, T. W. Kim, R. E. Tanzi, F. Oyama, T. Tabira, S. Ando, and
Y. Ihara, 2000. Mutant presenilin 2 transgenic mice. A large increase in the levels of Aβ
42 is presumably associated with the low density membrane domain that contains decreased
levels of glycerophospholipids and sphingomyelin. J Biol Chem 275:27901–8.
5. Jarrett, J. T., E. P. Berger, and P. T. J. Lansbury, 1993. The carboxy terminus of the β amyloid
protein is critical for the seeding of amyloid formation: implications for the pathogenesis of
Alzheimer’s disease. Biochemistry 32:4693–7.
6. Jarrett, J. T., E. P. Berger, and P. T. J. Lansbury, 1993. The C-terminus of the β protein is
critical in amyloidogenesis. Ann N Y Acad Sci 695:144–8.
7. Bitan, G., M. D. Kirkitadze, A. Lomakin, S. S. Vollers, G. B. Benedek, and D. B. Teplow, 2003.
Amyloid β-protein (Aβ) assembly: Aβ40 and Aβ42 oligomerize through distinct pathways.
Proc. Natl. Acad. Sci. USA. 100:330–335.
8. Walsh, D. M., A. Lomakin, G. B. Benedek, M. M. Condron, and D. B. Teplow, 1997. Amyloid
β-protein fibrillogenesis - Detection of a protofibrillar intermediate. J. Biol. Chem. 272:22364–
9. Teplow, D. B., N. D. Lazo, G. Bitan, S. Bernstein, T. Wyttenbach, M. T. Bowers, A. Baumket-
ner, J.-E. Shea, B. Urbanc, L. Cruz, J. Borreguero, and H. E. Stanley, 2006. Elucidating amy-
loid β-protein folding and assembly: A multidisciplinary approach. Acc. Chem. Res. 39:635–
10. Urbanc, B., L. Cruz, D. B. Teplow, and H. E. Stanley, 2006.Computer simulations of
Alzheimer’s amyloid β-protein folding and assembly. Curr Alzheimer Res 3:493–504.
11. Urbanc, B., J. Borreguero, L. Cruz, and H. E. Stanley, 2006. Ab initio discrete molecular
dynamics approach to protein folding and aggregation. Methods. Enzymol. 412:314–338.
12. Klimov, D. K., and D. Thirumalai, 2003. Dissecting the assembly of Aβ16−22amyloid peptides
into antiparallel β sheets. Structure. 11:295–307.
13. Tarus, B., J. E. Straub, and D. Thirumalai, 2005. Probing the initial stage of aggregation of the
Aβ10−35-protein: assessing the propensity for peptide dimerization. J Mol Biol 345:1141–56.
14. Huet, A., and P. Derreumaux, 2006. Impact of the Mutation A21G (Flemish Variant) on
Alzheimer’s β-Amyloid Dimers by Molecular Dynamics Simulations. Biophys J 91:3829–40.
15. Rapaport, D. C., 1997. The art of molecular dynamics simulation. Cambridge University
16. Teplow, D., N. Lazo, G. Bitan, S. Bernstein, T. Wyttenbach, M. Bowers, A. Baumketner,
J.-E. Shea, B. Urbanc, L. Cruz, J. Borreguero, and H. E. Stanley, 2006. Elucidating Amyloid
β-Protein Folding and Assembly: A Multidisciplinary Approach. Acc Chem Res 39:635–645.
17. Zhou, Y., C. K. Hall, and M. Karplus, 1996. First-order disorder-to-order transition in an
isolated homopolymer model. Phys. Rev. Lett. 77:2822–2825.
18. Dokholyan, N. V., S. V. Buldyrev, H. E. Stanley, and E. I. Shakhnovich, 1998.Discrete
molecular dynamics studies of the folding of a protein-like model. Fold. Des. 3:577–587.
19. Dokholyan, N. V., S. V. Buldyrev, H. E. Stanley, and E. I. Shakhnovich, 2000. Identifying the
protein folding nucleus using molecular dynamics. J. Mol. Biol. 296:1183–1188.
20. Zhou, Y. Q., and M. Karplus, 1997. Folding thermodynamics of a model three-helix-bundle
protein. Proc. Natl. Acad. Sci. USA. 94:14429–14432.
21. Zhou, Y. Q., M. Karplus, J. M. Wichert, and C. K. Hall, 1997. Equilibrium thermodynamics
of homopolymers and clusters: Molecular dynamics and Monte Carlo simulations of systems
with square-well interactions. J. Chem. Phys. 107:10691–10708.
22. Zhou, Y. Q., and M. Karplus, 1999. Folding of a model three-helix bundle protein: A thermo-
dynamic and kinetic analysis. J. Mol. Biol. 293:917–951.
23. Peng, S., F. Ding, B. Urbanc, S. V. Buldyrev, L. Cruz, H. E. Stanley, and N. V. Dokholyan,
2004. Discrete molecular dynamics simulations of peptide aggregation. Phys. Rev. E 69:041908.
24. Smith, A. V., and C. K. Hall, 2001. α-helix formation: Discontinuous molecular dynamics on
an intermediate-resolution protein model. Proteins: Struct. Func. & Genet. 44:344–360.
25. Smith, A. V., and C. K. Hall, 2001. Assembly of a tetrameric α-helical bundle: computer
simulations on an intermediate-resolution protein model. Proteins: Struct. Func. & Genet.
26. Ding, F., J. M. Borreguero, S. V. Buldyrey, H. E. Stanley, and N. V. Dokholyan, 2003. Mech-
anism for the α-helix to β-hairpin transition. Proteins: Struct. Func. & Genet. 53:220–228.
27. Urbanc, B., L. Cruz, F. Ding, D. Sammond, S. Khare, S. V. Buldyrev, H. E. Stanley, and N. V.
Dokholyan, 2004. Molecular dynamics simulation of amyloid β dimer formation. Biophys. J.
28. Lam, A., B. Urbanc, J. M. Borreguero, N. D. Lazo, D. B. Teplow, and H. E. Stanley, 2006.
Discrete Molecular Dynamics Study of Alzheimer Amyloid β-protein (Aβ) Folding. Proceedings
of the 2006 international conference on bioinformatics and computational biology 322–328.
29. Baumketner, A., S. L. Bernstein, T. Wyttenbach, G. Bitan, D. B. Teplow, M. T. Bowers, and
J.-E. Shea, 2006. Amyloid β-protein monomer structure: a computational and experimental
study. Protein Sci 15:420–8.
30. Gursky, O., and S. Aleshkov, 2000. Temperature-dependent β-sheet formation in β-amyloid
Aβ(1-40) peptide in water: uncoupling β-structure folding from aggregation. Biochim Biophys
31. Lim, K. H., H. H. Collver, Y. T. H. Le, P. Nagchowdhuri, and J. M. Kenney, 2007. Characteri-
zations of distinct amyloidogenic conformations of the Aβ (1-40) and (1-42) peptides. Biochem
Biophys Res Commun 353:443–9.
32. Urbanc, B., L. Cruz, S. Yun, S. V. Buldyrev, G. Bitan, D. B. Teplow, and H. E. Stanley, 2004.
In silico study of amyloid β-protein folding and oligomerization. Proc. Natl. Acad. Sci. USA.
33. Bernstein, S. L., T. Wyttenbach, A. Baumketner, J.-E. Shea, G. Bitan, D. B. Teplow, and
M. T. Bowers, 2005. Amyloid β-protein: monomer structure and early aggregation states of
Aβ42 and its Pro19 alloform. J Am Chem Soc 127:2075–84.
34. Gong, Y. S., L. Chang, K. L. Viola, P. N. Lacor, M. P. Lambert, C. E. Finch, G. A. Krafft,
and W. L. Klein, 2003. Alzheimer’s disease-affected brain: Presence of oligomeric Aβ ligands
(ADDLs) suggests a molecular basis for reversible memory loss. Proc. Natl. Acad. Sci. USA.
35. Barghorn, S., V. Nimmrich, A. Striebinger, C. Krantz, P. Keller, B. Janson, M. Bahr,
M. Schmidt, R. S. Bitner, J. Harlan, E. Barlow, U. Ebert, and H. Hillen, 2005. Globular amy-
loid β-peptide oligomer - a homogenous and stable neuropathological protein in Alzheimer’s
disease. J. Neurochem. 95:834–47.
36. Lesne, S., M. T. Koh, L. Kotilinek, R. Kayed, C. G. Glabe, A. Yang, M. Gallagher, and
K. H. Ashe, 2006. A specific amyloid-β protein assembly in the brain impairs memory. Nature
37. Bitan, G., B. Tarus, S. S. Vollers, H. A. Lashuel, M. M. Condron, J. E. Straub, and D. B.
Teplow, 2003. A molecular switch in amyloid assembly: Met35and amyloid β-protein oligomer-
ization. J Am Chem Soc 125:15359–65.
38. Seilheimer, B., B. Bohrmann, L. Bondolfi, F. Muller, D. Stuber, and H. Dobeli, 1997. The
toxicity of the Alzheimer’s β-amyloid peptide correlates with a distinct fiber morphology. J.
Struct. Biol. 119:59–71.
39. Snyder, S. W., U. S. Ladror, W. S. Wade, G. T. Wang, L. W. Barrett, E. D. Matayoshi, H. J.
Huffaker, G. A. Krafft, and T. F. Holzman, 1994. Amyloid-β aggregation: selective inhibition
of aggregation in mixtures of amyloid with different chain lengths. Biophys J 67:1216–28.
40. Watson, A. A., D. P. Fairlie, and D. J. Craik, 1998. Solution structure of methionine-oxidized
amyloid β-peptide (1-40). Does oxidation affect conformational switching?
41. Lazo, N. D., M. A. Grant, M. C. Condron, A. C. Rigby, and D. B. Teplow, 2005. On the
nucleation of amyloid β-protein monomer folding. Protein Sci 14:1581–96.
42. Bitan, G., S. Vollers, and D. B. Teplow, 2003. Elucidation of primary structure elements
controlling early amyloid β-protein oligomerization. J. Biol. Chem. 278:34882–34889.
43. Borreguero, J. M., B. Urbanc, N. D. Lazo, S. V. Buldyrev, D. B. Teplow, and H. E. Stanley,
2005. Folding events in the 21-30 region of amyloid β-protein (Aβ) studied in silico. Proc.
Natl. Acad. Sci. USA. 102:6015–6020.
44. Cruz, L., B. Urbanc, J. M. Borreguero, N. D. Lazo, D. B. Teplow, and H. E. Stanley, 2005.
Solvent and mutation effects on the nucleation of amyloid β-protein folding. Proc Natl Acad
Sci U S A 102:18258–63.
45. Petkova, A. T., W.-M. Yau, and R. Tycko, 2006. Experimental constraints on quaternary
structure in Alzheimer’s β-amyloid fibrils. Biochemistry 45:498–512.
46. Petkova, A. T., Y. Ishii, J. J. Balbach, O. N. Antzutkin, R. D. Leapman, F. Delaglio, and
R. Tycko, 2002. A structural model for Alzheimer’s β-amyloid fibrils based on experimental
constraints from solid state NMR. Proc. Natl. Acad. Sci. USA. 99:16742–16747.
47. Sciarretta, K. L., D. J. Gordon, A. T. Petkova, R. Tycko, and S. C. Meredith, 2005. Aβ40-
Lactam(D23/K28) models a conformation highly favorable for nucleation of amyloid. Bio-
48. Kyte, J., and R. F. Doolittle, 1982. A simple method for displaying the hydropathic character
of a protein. J. Mol. Biol. 157:105–132.
49. Sheu, S.-Y., D.-Y. Yang, H. L. Selzle, and E. W. Schlag, 2003. Energetics of hydrogen bonds
in peptides. Proc. Natl. Acad. Sci. USA. 100:12683–7.
50. Luisi, D. L., C. D. Snow, J.-J. Lin, Z. S. Hendsch, B. Tidor, and D. P. Raleigh, 2003. Surface
salt bridges, double-mutant cycles, and protein stability: an experimental and computational
analysis of the interaction of the Asp 23 side chain with the N-terminus of the N-terminal
domain of the ribosomal protein L9. Biochemistry 42:7050–60.
51. Berendsen, H. J. C., J. Poastma, W. V. Gunsteren, A. DiNola, and J. Haak, 1984. Molecular
dynamics with coupling to an external bath. J. Chem. Phys. 81:3684–3690.
52. Bradley, P., L. Malmstrom, B. Qian, J. Schonbrun, D. Chivian, D. E. Kim, J. Meiler, K. M. S.
Misura, and D. Baker, 2005. Free modeling with Rosetta in CASP6. Proteins 61 Suppl
53. Baker, D., 2000. A surprising simplicity to protein folding. Nature 405:39–42.
54. Dobson, C. M., 2004. Experimental investigation of protein folding and misfolding. Methods
55. Bowie, J. U., R. Luthy, and D. Eisenberg, 1991. A method to identify protein sequences that
fold into a known three-dimensional structure. Science 253:164–70.
56. Finkelstein, A. V., and B. A. Reva, 1991. A search for the most stable folds of protein chains.
57. Nguyen, H. D., and C. K. Hall, 2006. Spontaneous Fibril Formation by Polyalanines; Discon-
tinuous Molecular Dynamics Simulations. J Am Chem Soc 128:1890–1901.
58. Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flanney, 1992. Numerical Recipes
in FORTRAN: The Art of Scientific Computing, second edition. Cambridge University Press.
59. Heinig, M., and D. Frishman, 2004. STRIDE: a web server for secondary structure assignment
from known atomic coordinates of proteins. Nucleic Acids Res 32:W500–2.
60. Zhang, S., K. Iwata, M. J. Lachenmann, J. W. Peng, S. Li, E. R. Stimson, Y. Lu, A. M.
Felix, J. E. Maggio, and J. P. Lee, 2000. The Alzheimer’s peptide Aβ adopts a collapsed coil
structure in water. J Struct Biol 130:130–41.
61. Losic, D., L. L. Martin, A. Mechler, M.-I. Aguilar, and D. H. Small, 2006. High resolution scan-
ning tunnelling microscopy of the β-amyloid protein (Aβ1-40) of Alzheimer’s disease suggests
a novel mechanism of oligomer assembly. J Struct Biol 155:104–10.
62. Maji, S. K., J. J. Amsden, K. J. Rothschild, M. M. Condron, and D. B. Teplow, 2005. Con-
formational dynamics of amyloid β-protein assembly probed using intrinsic fluorescence. Bio-
63. Riek, R., P. Guntert, H. Dobeli, B. Wipf, and K. Wuthrich, 2001. NMR studies in aqueous
solution fail to identify significant conformational differences between the monomeric forms
of two Alzheimer peptides with widely different plaque-competence, Aβ(1-40)(ox)and Aβ(1-
42)(ox). Eur. J. Biochem. 268:5930–5936.
64. Hou, L., H. Shao, Y. Zhang, H. Li, N. K. Menon, E. B. Neuhaus, J. M. Brewer, I. L. Byeon,
D. G. Ray, M. P. Vitek, T. Iwashita, R. A. Makula, A. B. Przybyla, and M. G. Zagorski,
2004. Solution NMR studies of the Aβ(1-40) and Aβ(1-42) peptides establish that the Met35
oxidation state affects the mechanism of amyloid formation. J Am Chem Soc 126:1992–2005.
65. Humphrey, W., A. Dalke, and K. Schulten, 1996. VMD: visual molecular dynamics. J. Mol.
Figure 1: Representative conformations of a monomer, dimer, and pentamer of Aβ42 in
the absence (ECH = 0) and presence (ECH = 0.6) of EIs. A monomer conformation (a)
in the absence and (b) presence of EIs. A dimer conformation (c) in the absence and (d)
presence of EIs. A pentamer conformation (e) in the absence and (f) presence of EIs. Yellow
arrows correspond to the β-strand structure, turns are represented by light blue tubes and
random coil-like part are represented by gray tubes. The N-terminal D1 is marked as a
red sphere, and the C-terminal A42 is marked as a blue sphere. I31, I32, and I41, the most
hydrophobic residues, are represented as green spheres. This figure is generated by the VMD
Figure 2: Oligomer size distributions of Aβ40 and Aβ42 at (a) ECH= 0 and (b) ECH=
0.6. All size distributions are averages over the time frames at 8, 8.5, 9, 9.5, and 10 million
Figure 3: The effect of EIs on turn and β-strand propensities per residue in folded Aβ
monomers in the absence and presence of EIs. Turn propensities of (a) Aβ40 and (b) Aβ42
monomers. β-strand propensities of (c) Aβ40 and (d) Aβ42 monomers.
Figure 4: Intramolecular contact maps of folded Aβ40 and Aβ42 monomers at ECH= 0
(left column) and ECH = 0.6 (right column). The strength of the contact map is color-
coded following the rainbow scheme: from blue (no contacts) to red (the largest number
of contacts). Each contact map is an average of over 100 monomer conformations after 1
million simulation steps.
Figure 5: Detailed time evolution of intramolecular contact formation during Aβ40 (left
column) and Aβ42 folding (right column). The strength of the contact map is color-coded
as in Fig. 4. Each contact map is an average of over 100 monomer conformations.
Figure 6: The effect of EIs on turn and β-strand propensities per residue within Aβ
pentamers and larger oligomers in the absence and presence of EIs. Turn propensities of (a)
Aβ40 and (b) Aβ42 pentamers and larger oligomers. β-strand propensities of (c) Aβ40 and
(d) Aβ42 pentamers and larger oligomers.
Figure 7: Ramachandran plots of Aβ42 pentamers and larger oligomers for selected
residues D1, Y10, F19, E22, D23, S26, K28, M35, I41, and A42 in the absence and presence
of EIs. Horizontal and vertical axes correspond to the angles Φ and Ψ, respectively, both
varying from -180◦to 180◦. Each plot contains ∼640 points corresponding to Aβ42 pen-
tamers to decamers obtained at 8, 8.5, 9, 9.5, and 10 million simulation steps. Ramachandran
plots are generated using the VMD software package .
Figure 8: Intramolecular contact maps of Aβ40 and Aβ42 pentamers and larger oligomers
at ECH = 0 and ECH = 0.6. Each contact map is an average of over 50 conformations
obtained at 8, 8.5, 9, 9.5, and 10 million simulation steps.
Figure 9: Intermolecular contact maps of Aβ40 and Aβ42 pentamers and larger oligomers
at ECH= 0 and ECH= 0.6. Each contact map is an average of over 50 conformations at 8,
8.5, 9, 9.5, and 10 million simulation steps.
Figure 10: Intramolecular hydrogen bond maps of Aβ40 and Aβ42 pentamers and larger
oligomers at ECH= 0 and ECH= 0.6. Each map is an average of over 50 conformations at
8, 8.5, 9, 9.5, and 10 million simulation steps.
Figure 11: Intermolecular hydrogen bond maps of Aβ40 and Aβ42 pentamers and larger
oligomers at ECH= 0 and ECH= 0.6. Each map is an average of over 50 conformations at
8, 8.5, 9, 9.5, and 10 million simulation steps.
Table 1: Average turn and β-strand propensities per residue with standard errors within
folded Aβ40 and Aβ42 monomers. Each value is an average of over 100 monomer confor-
mations after 1 million simulation steps.
Table 2: Average turn and β-strand propensities per residue with standard errors within
Aβ40 and Aβ42 pentamers and larger oligomers. Each value is an average of over 50 con-
formations at 8, 8.5, 9, 9.5, and 10 million simulation steps.
Table 3: Average hydrogen bond propensities per residue, showing the five most frequent
residues involved in intramolecular hydrogen bonding within Aβ40 and Aβ42 pentamers and
larger oligomers. Each value is an average of over 50 conformations at 8, 8.5, 9, 9.5, and 10
million simulation steps.
Table 4: Average hydrogen bond propensity per residue, showing the five most frequent
amino acids involved in intermolecular hydrogen bonding within Aβ40 and Aβ42 pentamers
and larger oligomers. Each value is an average of over 50 conformations at 8, 8.5, 9, 9.5,
and 10 million simulation steps.
810 12 14 16 18 20
8 10 12 14 16 18 20
Aβ 40 (ECH = 0)
Aβ 40 (ECH = 0.6)
Aβ 42 (ECH = 0)
Aβ 42 (ECH = 0.6)
Aβ 40 (ECH = 0)
Aβ 40 (ECH = 0.6)
Aβ 42 (ECH = 0)
Aβ 42 (ECH = 0.6)
Aβ 40 (ECH = 0)
Aβ 40 (ECH = 0.6)
Aβ 42 (ECH = 0)
Aβ 42 (ECH = 0.6)
Aβ 40 (ECH = 0)
Aβ 40 (ECH = 0.6)
Aβ 42 (ECH = 0)
Aβ 42 (ECH = 0.6)
0.44 ± 0.04
0.11 ± 0.02
0.43 ± 0.04
0.12 ± 0.03
0.48 ± 0.04
0.11 ± 0.03
0.50 ± 0.05
0.10 ± 0.03
0.44 ± 0.02
0.13 ± 0.02
0.45 ± 0.02
0.11 ± 0.01
0.42 ± 0.02
0.13 ± 0.01
0.45 ± 0.02
0.11 ± 0.01
0.17 A30 0.17
0.16 G29 0.15
0.15 E11 0.15
Aβ40Aβ42 Download full-text