Bulletin of Mathematical Biology (2019) 81:1943–1964

https://doi.org/10.1007/s11538-019-00583-3

A Mathematical Model for Amyloid-ˇ Aggregation

in the Presence of Metal Ions: A Timescale Analysis
for the Progress of Alzheimer Disease

Eda Asili1 · Shantia Yarahmadian1 · Hadi Khani2 · Meisam Sharify3

Received: 16 September 2017 / Accepted: 10 February 2019 / Published online: 26 February 2019
© Society for Mathematical Biology 2019

Abstract
The aggregation of amyloid-β (Aβ) proteins through their self-assembly into
oligomers, fibrils, or senile plaques is advocated as a key process of Alzheimer’s dis-
ease. Recent studies have revealed that metal ions play an essential role in modulating
the aggregation rate of amyloid-β (Aβ) into senile plaques because of high binding
affinity between Aβ proteins and metal ions. In this paper, we proposed a mathe-
matical model as a set of coupled kinetic equations that models the self-assembly of
amyloid-β (Aβ) proteins in the presence of metal ions. The numerical simulations
capture four timescales in the Aβ dynamics associated with three important events
which include the formation of the amyloid–metal complex, the homogeneous aggre-
gation of the amyloid–metal complexes, and the non-homogeneous aggregation of the
amyloid–metal complexes. The method of singular perturbation is used to identify
these timescales in the framework of slow–fast systems.

Keywords Alzheimer · Metal ions · Chemical reactions · Timescale analysis

B Shantia Yarahmadian
syarahmadian@math.msstate.edu
Eda Asili
ek343@msstate.edu
Hadi Khani
hk316@msstate.edu
Meisam Sharify
meisam.sharify@ipm.ir

1 Department of Mathematics and Statistics, Mississippi State University, Mississippi State,

MS 39762, USA
2 Department of Chemistry, Mississippi State University, Mississippi State, MS 39762, USA
3 School of Mathematics, Institute for Research in Fundamental Sciences (IPM),
P.O. Box:19395-5746, Tehran, Iran

123
1944 E. Asili et al.

1 Introduction

As one of the major health care crises facing the world today, Alzheimer’s disease (AD)
is the leading cause of dementia, associated with loss of memory, cognitive decline, and
behavioral and physical disability, ultimately leading to death (Mattson 2004; Kepp
2012). According to the World Alzheimer Report 2016, nearly 47 million people in the
world have Alzheimer’s or a related dementia, and this number is estimated to increase
to 131 million by 2050 (Prince et al. 2016). The total worldwide cost of Alzheimer’s
and dementia is estimated to be US$818 billion, and it will become a trillion dollar
disease by 2018 (Prince et al. 2016). In the USA, AD is one of the top 10 leading
causes of death (Hebert et al. 2013), and in 2015 an estimated 5.4 million Americans
of all ages had Alzheimer’s disease, and this number is expected to increase up to 16
million people by 2050 (ADS 2016). These statistics underline the importance of AD
treatment that may provide a significant opportunity for social and economic impact
in the twenty-first century.
Being the main hallmark of AD, Aβ monomers are formed via the enzymatic
cleavage of amyloid precursor protein (APP), which is an integral membrane protein
expressed in many tissues and concentrated in the synapses of neurons (Kepp 2012;
Jakob-Roetne and Jacobsen 2009; Hamley 2012; Rauk 2009). The main characteristic
pathology of AD, known as amyloid cascade hypothesis, is the aggregation of Aβ
peptides (two main Aβ forms, Aβ40 and Aβ42) in the form of oligomers and amyloid
fibrils in the brain (Karran et al. 2011). It is advocated that the aggregation of Aβ
monomers proceeds via a nucleation-dependent mechanism that is initiated by a slow
nucleation phase through the self-assembly of Aβ monomers into soluble prefibrillar
oligomers (known as nuclei) followed by a fast elongation of soluble small interme-
diates on the formed nuclei, leading to the formation of insoluble mature fibrils and
plaques (Jarrett and Lansbury 1993; Chiti and Dobson 2006; Wilson et al. 2008). The
aggregated Aβ forms are tightly associated with synaptic dysfunctions and neuronal
network perturbations that consequently lead to cognitive decline (Crimins et al. 2013;
Mucke and Selkoe 2012).
Studies based on the elementary analysis of senile plaques in the AD-affected brains
have revealed the presence of abnormally high concentrations of metal ions, including
copper (Cu) and zinc (Zn), with respect to healthy brain tissues (Lovell et al. 1998;
Miller et al. 2006). In addition to this observation, a large body of in vitro and in vivo
evidences have brought the scientific community into the conclusion that coordination
of Cu and Zn ions to Aβ intervene in the aggregation of Aβ, via different routes, mainly
by modulating Aβ aggregation (Cristóvão et al. 2016; Faller et al. 2014; Viles 2012).
Several mechanisms for such action have been proposed that nucleation step can be
either promoted or hindered by Cu and Zn ions; however, understanding the detailed
mechanisms by which Aβ aggregation is affected by Cu and Zn ions has been still
the subject of debate. The challenge is the great complexity of the Aβ aggregation
process itself and the vast dependency of in vitro metal ions-Aβ aggregation on the
experimental conditions (especially the metalAβ ratio) (Faller et al. 2013).
Theoretical kinetics models based on the rate laws have shown to be robust tools
for defining the chemical kinetics of a complex reaction network governing the growth

123
A Mathematical Model for Amyloid-β Aggregation in the… 1945

of filamentous protein structures and understanding the protein aggregation processes

(Cohen et al. 2013). Nevertheless, when a system with a pool of components with a
different degree of polymerization is the subject of a study, the components undergo an
inter-conversion through a complex assembly pathways, which makes it challenging
to derive the closed form expressions for the reaction rate of the system (Cohen et al.
2012). Back in 1950s–1970s, Oosawa et al. (Oosawa and Asakura 1975; Oosawa and
Kasai 1962; Oosawa et al. 1959) first derived an integrated rate law for the kinetic
analysis of the linear polymerization of actin and tubulin proteins by defining a homo-
geneous (i.e., primary) nucleation step followed by a filamentous growth through
monomer addition. More recently Knowles et al. (Knowles et al. 2009; Cohen et al.
2011a, b) extended the idea of heterogeneous (i.e., secondary) nucleation through the
analytical treatment of a set of coupled kinetic equations to generalize the Oosawa the-
ory by including the fragmentation of filaments and monomer-dependent secondary
nucleation pathways. While the basic kinetic equations describing the dynamic of Aβ
aggregation phenomena are well-established, the need for a mathematical model in the
form of a master equation that can describe the self-assembly of Aβ filaments in the
presence of metal ions is largely felt. The interest in elucidating the role of metal ions
in Aβ aggregation is of utmost importance to the metal chelation therapy as a potential
pharmacological option for the treatment of AD (Bolognin et al. 2009; Pithadia and
Lim 2012; Bica et al. 2009; Hegde et al. 2009). In this paper, we present a math-
ematical model, which consists of a set of coupled kinetic equations that describes
the filamentous growth of metal/Aβ-rich fibrils in a system consisting of both metal
ions and Aβ monomers. The proposed master equation considers four microscopic
processes (Fig. 1), which includes the reaction of metal ions (M) and Aβ monomers
to form the M–Aβ complexes, reaction between M–Aβ complexes themselves toward
the formation of nuclei, the monomeric addition of M–Aβ complexes to the growing
nuclei through the homogeneous elongation, and the addition of Aβ monomers to
the growing nuclei via the non-homogeneous elongation. Timescale analysis reveals a
multiple timescale dynamics consisting of four timescales related to four main events.
The first timescale represents the formation of nuclei through the self-assembly of
M–Aβ complexes, the second one describes nucleation process, the third one depicts
the homogeneous elongation of the filaments, and the fourth timescale is related to
the non-homogeneous elongation.

(a) (b) (c) (d)

ka kn kh knh
+ nc

Aβ monomers Metal ions Metal-Aβ complexes Nucleus Homogeneous fibril elongation Non-homogeneous fibril elongation

Fig. 1 Schematic representation of the microscopic processes of Aβ fibril formation and elongation in the
presence of metal ions: a soluble metal–Aβ complexes are formed by binding the soluble metal ions to
soluble Aβ monomers in an irreversible manner, b primary nucleation leads to the creation of a polymer
of length n c from soluble metal–Aβ complexes, c filaments grow homogeneously from both ends by the
addition of next metal–Aβ complexes in an irreversible manner, d filaments grow non-homogeneously from
both ends by the addition of next Aβ monomers in an irreversible manner

123
1946 E. Asili et al.

The paper is organized as follows: in Sect. 2 the mathematical model for studying
the effect of metal ions is developed, which leads to a closed system of kinetic
equations describing the dynamics of the chemical reactions. Section 3 discusses the
dynamics of the amyloid–metal aggregation. Our analysis leads to the determina-
tion of low-, average-, and high-metal regimes. Moreover, the analysis reveals three
critical parameters. Numerical simulations of the kinetic system are presented in
Sect. 4. For each of the low-, average-, and high-metal regimes and for a different
setting of the critical parameters, we provide the graphs via numerical simulations.
In Sect. 5, the timescale analysis of the system is provided by using singular per-
turbation theory. This analysis shows that there are four timescales for the kinetic
system.

2 Mathematical Model Formulation

In this section, we develop the mathematical model based on the law of the mass action
to study the effect of metal ions on the aggregation of the amyloid–metal complexes.
To be able to capture all possible events in a system consisting of Aβ and metal
ions, we first discuss the chemical reactions resulted from the mechanism described
in Fig. 1. These reactions yield to a master equation that describes the time evolution
of concentrations for Aβ–M complexes with a specific number of proteins and metal
ions. Next, the kinetic equations for the moments will be derived from the master
equations for the total number of aggregated complexes, bonded proteins, and bonded
metal ions. The kinetic equations also show the relation between the homogeneous
and non-homogeneous products, i.e., complexes with equal or not equal number of
metal ions and Aβ. Combining all of the kinetic equations for the moments together
will result in a closed equation system. Finally, by introducing new parameters, which
vary based on the initial conditions and the rate of chemical reactions, the dynamics
of the system will be fully analyzed.

2.1 Chemical Reactions

Aggregation of Aβ in the presence of metal ions is the essential theme that must
be taken into account in the Aβ-aggregation modeling. The fundamental processes
that enter the master equation are described in Fig. 1 and are listed as a set of
chemical reactions in Table 1. As represented by the first equation in Table 1, the
reaction of metal ions (M) and Aβ (A) generates the first product, A1 M1 com-
plexes, with the rate of ka . The second equation describes the reaction of A1 M1
complexes, which produces the nuclei A2 M2 with the rate of kn . The third reaction
shows the homogeneous aggregation process through the addition of A1 M1 -monomer
to the growing Ai Mi . This reaction accounts for all possible reactions between Ai Mi
and A1 M1 . The fourth reaction corresponds to the formation of non-homogeneous
aggregation by the sequential addition of Aβ monomers to the homogeneous com-
plexes, Ai Mi . Note that for any product resulted from the fourth chemical reaction,
i.e., Ai M j , the condition j ≤ i holds since the complexes that contribute in the

123
A Mathematical Model for Amyloid-β Aggregation in the… 1947

Table 1 Multistep aggregation process of Aβ in the presence of metal ions

Chemical reactions Reaction rate Explanation

ka
(1) A1 + M1 −→ A1 M1 ka Metal–Aβ complex formation
kn
(2) A1 M1 + A1 M1 −→ A2 M2 kn Nucleation step
kh
(3) Ai Mi + A1 M1 −→ Ai+1 Mi+1 kh Homogeneous aggregation; 2 ≤ i < ∞
knh
(4) Ai M j + A1 −→ Ai+1 M j knh Non-homogeneous aggregation;
2 ≤ i < ∞, 2 ≤ j ≤ i

non-homogeneous reaction have either the equal number of Aβ and metal ions or
larger number of Aβ than metal ions. Not to mention, A1 M1 complexes will not
directly participate in the fourth reaction, an assumption which is also considered
in the mathematical model developed by Cohen et al. (2011b) for the filamentous
growth of Aβ without considering the effect of metal ions. In our study, this can
be explained based on the argument of the overall charge, which is discussed in
Sect. 2.2.

2.2 Parameters Argument

The overall charge is an important parameter for aggregation because generally aggre-
gation is faster when the overall charge approaches zero. The net charge of Aβ is about
−3 at physiological pH (Kepp 2012). Metal ions (M n+ ) are positively charged, and
complexes of M n+ − Aβ would have a net charge of (n − 3) and would hence be
more prone to aggregation. In this study, a 2+ charge is assigned to metal ions (M 2+ )
since two well-studied metal ions of copper ion (Cu 2+ ) and zinc ion (Z n 2+ ) carry
a 2+ charge. The overall charge on the Cu 2+ − Aβ and Z n 2+ − Aβ complexes
is −1 (Faller et al. 2013). In our simulations, we consider kh < kn . This assump-
tion is true if we assume that Ai Mi complexes (for the case in which metal ions
are +2 charge) hold more negative charges (−i charge, where 2 < i < ∞) on
surface than that of monomeric A1 M1 complexes (−1 charge). The negative charge
on the Ai Mi and A1 M1 complexes causes an electrostatic repulsive force between
the complexes and leads to a decrease in the aggregation rate. With these consider-
ations in mind, we also considered knh < kn and kh . This assumption is true since
the electrostatic repulsive force between Aβ monomers (with −3 charge) and Ai M j
complexes (with a negative charge of −3i + 2 j, where 2 < i < ∞ and 2 < j < ∞)
are more than that of A1 M1 complexes (with −1 charge) with Ai Mi complexes. In
support of this claim, Ai M j complexes can also carry more negative charges on their
surfaces than Ai Mi complexes since i can be greater than j in case of Ai M j complexes.
In summary, there are four important events in the system which include the metal–Aβ
aggregation, nucleation, homogeneous aggregation and non-homogeneous aggre-
gation [see (1)]. In the modeling procedure, the reverse chemical reactions listed
in Table 1 are neglected due to high affinity constant for the metal–Aβ complex
formation.

123
1948 E. Asili et al.

metal−Aβformation homogeneous aggregation

     
ka kn kh kh kh knh knh knh
A1 + M1 −→ A1 M1 −→ A2 M2 −→ A3 M3 −→ . . . −→ An Mn −→ An+1 Mn −→ . . . −→ Am Mn
     
nucleation non-homogeneous aggregation
(1)

2.3 Master Equation

Let f (t, i, j) be the time-dependent concentration of Ai M j complexes resulting from

the microscopic dynamics presented in Fig. 1 and Table 1. On the basis of the law of
mass action, the kinetic equation for the time evolution of f (t, i, j) is written as the
following master equation:

∂ f (t, i, j)  
 ∞
= ka m(t)l(t) − kn f (t, 1, 1)2 − kh f (t, 1, 1) f (t, k, k) δi,1 δ1, j
∂t
k=2
+ kn f (t, 1, 1) δi,2 δ2, j2
 
+ kh f (t, 1, 1) f (t, i − 1, j − 1)u 2 (i) − f (t, i, j)u 1 (i) δi, j
 
+ knh m(t) f (t, i − 1, j)u j (i)u 2 (i) − f (t, i, j)u 1 (i)
(2)
where m(t) and l(t) are the concentration of free Aβ proteins and free metal ions,
respectively. u k (i) is the step function, and δi,k is the Kronecker delta function, which
are defined as:

1 if i > k 1 if i = k
u k (i) = , δi,k = . (3)
0 if i ≤ k 0 else

∂ f (t,i, j)
∂t consists of four different terms:
(a) Metal–Aβ complex formation A1 M1

 ∞
∂ f (t, 1, 1)
= ka m(t)l(t) − kn f (t, 1, 1)2 − kh f (t, 1, 1) f (t, k, k) (4)
∂t
k=2

ka m(t)l(t) shows the formation of metal–amyloid complex (reaction 1 in Table 1).

−kn f (t, 1, 1)2 represents the nucleation step (reaction 2 in Table 1), and −kh f (t, 1, 1)
∞
i=2 f (t, i, j) accounts for the formation of homogeneous complexes (e.g., Ai Mi ,
reaction 3 in Table 1) for i ≥ 2.
(b) Nucleation (A2 M2 )

∂ f (t, 2, 2)
= kn f (t, 1, 1)2 − kh f (t, 1, 1) f (t, 2, 2) − knh m(t) f (t, 2, 2) (5)
∂t

123
A Mathematical Model for Amyloid-β Aggregation in the… 1949

The first term, +kn f (t, 1, 1)2 , describes the formation of A2 M2 nuclei, the second term
which is shown by −kh f (t, 1, 1) f (t, 2, 2), represents the reaction between A2 M2 and
A1 M1 to form A3 M3 , and the last term, −knh m(t) f (t, 2, 2), defines the consumption
of A2 M2 toward the formation of non-homogeneous A3 M2 complex.
(c) Homogeneous aggregation (Ai Mi , i > 2)

∂ f (t, i, i)
= kh f (t, 1, 1) f (t, i −1, i −1)−kh f (t, 1, 1) f (t, i, i)−knh m(t) f (t, i, i)
∂t
(6)
where the first term, +kh f (t, 1, 1) f (t, i − 1, i − 1), represents the formation of
Ai Mi from Ai−1 Mi−1 , the second term, −kh f (t, 1, 1) f (t, i, i), describes the reaction
between Ai Mi and A1 M1 to form a homogeneous Ai+1 Mi+1 complex, and the last
term, −knh m(t) f (t, i, i), accounts for the attachment of Aβ to Ai Mi , leading to the
formation of non-homogeneous complexes.
(d) Non-homogeneous aggregation (Ai M j , i > j, i > 2)

∂ f (t, i, j)
= knh m(t) f (t, i − 1, j) − knh m(t) f (t, i, j) (7)
∂t

The first term, knh m(t) f (t, i − 1, j), represents the formation of non-homogeneous
complexes, Ai M j , and the second term describes the attachment of Aβ to Ai M j to
the subsequent formation of other non-homogeneous complexes.
By taking these four parts into consideration, master equation (2) for i, j ≥ 1 is
derived. In the same manner, the time evolutions of m(t) and l(t) are described as
follows: ⎧
∂ 
∞  i
⎪
⎪ dm(t)
⎪
⎪ = − f (t, 1, 1) + i f (t, i, j)
⎪
⎨ dt ∂t
i=2 j=2
(8)
⎪
⎪ ∂  ∞ 
 i 
⎪
⎪ dl(t)
⎩ dt = − ∂t f (t, 1, 1) +
⎪ j f (t, i, j)
i=2 j=2

Equation (8) describes the variation of free metal ions and free Aβ proteins. Naturally,
the terms on the right-hand side of the first equation show the total amount of the
bonded proteins and in the second equations, the right-hand side show the total number
of bonded metal ions. The counting is performed over i and j, which are representing
the number of bonded proteins and metal ions in the complexes.

2.4 Closed System of Equations for the Aggregated Polymer Number, Mass and
Metal Ion Concentrations

The kinetic equations for the moments can be evaluated by taking the summation on i
and j on the both sides of (2). In this regard, we define P(t), M1 (t), and M2 (t) terms
as the total number of aggregated complexes f (t, i, j), the total number of bonded
proteins, and the total number of bonded metal ions, respectively, by the following
equations:

123
1950 E. Asili et al.

∞ 
 i ∞ 
 i ∞ 
 i
P(t) = f (t, i, j), M1 (t) = i f (t, i, j), M2 (t) = j f (t, i, j)
i=2 j=2 i=2 j=2 i=2 j=2
(9)
Note that in (9), we only consider the homogeneous and non-homogeneous metal–
Aβ aggregates which start growing after the nucleation process; therefore, A1 M1
complexes are not counted in Eq. (9), and the summations are taken over i, j ≥ 2.
The reason is that A1 M1 s are participating only in the process of nucleation. They
are not considered as an aggregated filament (they are considered as seeds for the
nucleation process), therefore they are not being included in the counting process total
filaments. Moreover, since for any Ai M j complexes, i ≥ j, the summation over j
in M1 and M2 is bounded above by i. In order to discern between the homogeneous
and non-homogeneous aggregations, we introduce Ph (t) and Mh (t) as homogeneous
moments for the total number of aggregated homogeneous complexes and the bonded
proteins/metals, respectively:

∞
 ∞

Ph (t) = f (t, i, i), Mh (t) = i f (t, i, i) (10)
i=2 i=2

The relationship between the total number of homogeneous and non-homogeneous

aggregations is defined by Eq. (11):

∞ 
 i−1
P(t) = Ph (t) + f (t, i, j) (11)
i=3 j=2

Moreover, the relationship between the homogeneous and non-homogeneous moments

are written as:
∞ 
 i−1 ∞ 
 i−1
M1 (t) = Mh (t) + i f (t, i, j), M2 (t) = Mh (t) + j f (t, i, j) (12)
i=3 j=2 i=3 j=2

Theorem 2.1 The kinetic equations for the moments Ph , P, Mh , M1 , and M2 are
explicitly expressed as follows:
⎧
⎪ d Ph (t)
⎪
⎪ = kn f (t, 1, 1)2 − knh m(t)Ph (t)
⎪
⎪ dt
⎪
⎪
⎪
⎪ d P(t)
⎪
⎪ = kn f (t, 1, 1)2
⎪
⎪ dt
⎨
dMh (t)
= 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) − knh m(t)Mh (t) (13)
⎪
⎪ dt
⎪
⎪
⎪
⎪ dM1 (t)
⎪
⎪ = 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) + knh m(t)P(t)
⎪
⎪ dt
⎪
⎪
⎩ dM2 (t) = 2k f (t, 1, 1)2 + k f (t, 1, 1)P (t)
⎪
n h h
dt

123
A Mathematical Model for Amyloid-β Aggregation in the… 1951

Proof (13) can be derived by applying the summation over i and j to (2). Details of
calculations is provided in appendix. 

dm(t) dl(t)
Remark 2.1 Using (8), dt and dt are evaluated as
⎧  
⎪ dm(t) ∂ f (t, 1, 1) dM1 (t)
⎪
⎨ =− +
dt ∂t dt
  (14)
⎪ dl(t)
⎪ ∂ f (t, 1, 1) dM2 (t)
⎩ =− +
dt ∂t dt

Therefore, by substituting the value for Ph from (10) into (4) and adding the resulting
equation to the list of equations on (13), we get:
⎧ ∂ f (t, 1, 1)
⎪
⎪ = ka m(t)l(t) − kn f (t, 1, 1)2 − kh f (t, 1, 1)Ph (t)
⎪
⎪ ∂t
⎪
⎪
⎪ d Ph (t)
⎪
⎪
⎪ = kn f (t, 1, 1)2 − knh m(t)Ph (t)
⎪
⎪
⎪
⎪ dt
⎪
⎪ d P(t)
⎪
⎨ = kn f (t, 1, 1)2
dt (15)
⎪
⎪ dMh (t)
⎪
⎪ = 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) − knh m(t)Mh (t)
⎪
⎪ dt
⎪
⎪
⎪
⎪ dM1 (t)
⎪
⎪ = 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) + knh m(t)P(t)
⎪
⎪ dt
⎪
⎪
⎩ dM2 (t) = 2k f (t, 1, 1)2 + k f (t, 1, 1)P (t)
⎪
n h h
dt

3 Dynamics of the Amyloid–Metal Aggregation

Our model is founded upon chemical reaction network theory. Generally, in a closed
system of chemical reactions, the system leads to a point that either a stationary state or
an equilibrium point is attained (Turányi and Tomlin 2014). Representing the chemical
reactions in the form of a set of ordinary deferential equations provides a robust
mathematical tool in which some variables have derivatives of much larger magnitude
than those variables imposing a multiple timescales on the system. To analyze the
system and examine the time scales, we applied the quasi-steady-state hypothesis
(QSSA) to find the appropriate scales in the non-dimensionalization procedure. Then,
we use perturbation theory was used to evaluate the timescales.

3.1 The Quasi-Steady-State Approximation

Quasi-steady-state approximation (QSSA) states that which reactions are faster than
others, and thus, reach the steady-state quicker (Turányi and Tomlin 2014). In our
case, after the nucleation step, all of the intermediate-homogeneous complexes are
assumed to reach the equilibrium rapidly relative to other reactions. Therefore, the
QSSH value of Ph is evaluated by setting the derivative of Ph to zero.

123
1952 E. Asili et al.

Table 2 List of the model variables before and after non-dimensionalization

Variable description Unit Notation before Notation after
nondim. nondim.

Amyloid–metal complex, A1 M1 µM f (t, 1, 1) F1

Homogeneous filaments µM Ph P1
Total filaments µM P P2
Bonded proteins µM M1 1
Bonded metal ions µM M2 2
Free proteins µM m μ
Free metal ions µM l 
Time h t τ

Table 3 List of the model parameters before and after non-dimensionalization

Parameter description Unit Notation before Notation after

nondim. nondim.

Metal to protein ratio None ε None

Metal–Aβ formation rate (µM)−1 h−1 ka 1
Nucleation rate (µM)−1 h−1 kn κ1
Homogeneous aggregation rate (µM)−1 h−1 kh κ2
Non-homogeneous aggregation rate (µM)−1 h−1 knh κ3
Total proteins µM m tot 1
Total metal ions µM ltot ε

kn ltot 2
Ph∗ = (16)
knh m tot

In (16), ltot represents the total number of metal ions, m tot the total number of proteins.

Remark 3.1 The main reason that ltot and m tot are being used in scaling is the fact that
these values are the maximum values that f (t, 1, 1) and m(t) can reach, respectively.
It should be mentioned that f (t, 1, 1) and m(t) are time-dependent, while ltot and m tot
are fixed, and intuitively makes the scaling to be consistent with QSSA.

Tables 2 and 3 summarizing the model variables and parameters before and after non-
dimensionalization.

3.2 Non-dimensionalization

In the non-dimensionalization process P(t), M1 (t), and m(t) were scaled with the total
protein concentration, m tot , and f (t; 1; 1), M2 (t), and l(t) with the total concentration
of free metal ions, ltot . To identify the dimensionless variables, the following scaling
procedure is used.

123
A Mathematical Model for Amyloid-β Aggregation in the… 1953

f (t,1,1) Ph M1
F1 = ltot , P1 = Ph∗ , P2 = m tot ,
P
1 = m tot
M2
(17)
2 = ltot , μ= m tot ,
m
= ltot ,
l
τ = λt

where λ is the characteristic rate by which the timescales are identified. By substituting
the values of (17) and the value for Ph∗ and by setting the timescale λ = ka m tot ,
(the timescale associated with the fast relaxation to QSSA value), (13) is reduced
to: ⎧
⎪ dF1
⎪
⎪ = μ − κ1 εF12 − κ3 ε2 F1 P1
⎪
⎪ dτ
⎪
⎪  
⎪
⎪ d P1
⎪
⎪ = κ2 F1 2 − μP1
⎪
⎪ dτ
⎪
⎪
⎪
⎪ d P2
⎪
⎪ = κ1 ε2 F1 2
⎪
⎪ dτ
⎨
d1
= 2κ1 ε2 F1 2 + κ3 ε3 F1 P1 + κ2 μP2 (18)
⎪
⎪ dτ
⎪
⎪
⎪
⎪ d2
⎪
⎪ = 2κ1 ε2 F1 2 + κ3 ε3 F1 P1
⎪
⎪ dτ
⎪
⎪  
⎪
⎪ dμ
⎪
⎪ = − εμ + κ1 ε2 F12 + κ2 μP2
⎪
⎪ dτ
⎪ d
⎪  
⎪
⎩ = − μ + κ1 εF12
dτ
where ε, κ1 , κ2 , κ3 are the dimensionless parameters, which are defined as follows:

ltot kn knh kn kh
ε= , κ1 = , κ2 = , κ3 = . (19)
m tot ka ka ka knh

Remark 3.2 System (18) contains no reverse reactions and has a positive feedback,
therefore, all of the free proteins will be eventually converted to either homogeneous
and/or non-homogeneous metal–protein complexes. Intuitively, the dynamics of the
system is characterized by the ratio of Aβ to metal ions. As such, this results in three
different regimes comprising of a low (ε < 1), average (ε = 1), and high (ε > 1)
metal ion concentration. Besides ε, the first two reactions in Table 1 compete for the
production and consumption of A1 M1 . Therefore, κ1 identifies the dominant rate, and
depending on its value (κ1 < 1, κ1 = 1 or κ1 > 1), three different regimes can be
recognized respectively: the first reaction is dominant in the production of A1 M1 , both
first and second chemical reactions proceeds at the same rate and the second reaction
is predominant in the consumption of A1 M1 .

Remark 3.3 Although all possible cases can be mathematically considered, the results
of physiologically realistic cases with κ1 < 1 are mainly analyzed in our simulations. In
a similar way, the first and the fourth equations in Table 1 compete in the consumption
of A1 and κ2 value determines the dominant equation depending on whether κ2 <
1, κ2 = 1 or κ2 > 1. The next critical parameter represented by κ3 accounts for the
competition between the second and the third reaction toward the consumption of
A1 M1 . For a fixed ε, based on the values of κ1 , κ2 , and κ3 , 27 different subregimes can
be recognized by considering the rate of chemical reactions. This provides the overall

123
1954 E. Asili et al.

Table 4 Table of parameters

Parameters used in the Value
used in the simulation. Some the
simulation
existing experimental values are
taken from (Sarell 2010) ε 0.2, 1, 2
ka 3.4 × 106
kn 10−4
kh 8 × 104
knh 30
m tot 7.8 × 10−8
The others are assumed according to the parameters argument in 2.2

of 81 possible cases for (18) regarding variations of m tot , ltot and chemical reaction
rates.

4 Numerical Simulation of the Amyloid–Metal Aggregation

The numerical simulation of the system can be achieved by applying standard numer-
ical methods to (18) for the given values for chemical reaction rates and the initial
conditions. The numerical experiments were performed in MATLAB R2016a ode23
toolbox. The initial conditions are set to zero and in our analysis, we have used exper-
imental parameter values reported in Sarell (2010) for the case of aggregation of
Aβ in the presence of copper ions (Cu2+ ) into our model. Parameters values for the
simulations are summarized in Table 4. These results are depicted in Fig. 2.

4.1 Regime with Low Metal Ion Concentration

When the concentration of metal ions is low (ε < 1) in the system, they are used up
first in the formation of metal–Aβ complexes (A1 M1 ). Therefore, the level of metal
ions undergoes a fast decay to zero causing the first reaction to stop quickly. Aβ
monomers are consumed in both the formation of metal–protein complexes and non-
homogeneous products. Therefore, κ2 determines the rate of the system depending on
the value of κ1 and κ3 (see Fig. 2).

4.2 Regime with Average Metal Ion Concentration

In the case of ε = 1, since there are equal number of proteins and metal ions, depending
on the values of κ1 , κ2 , and κ3 . The major difference with the previous regime is that
there are cases where the steady-state metal ions is nonzero, i.e., the metal ions will
remain in the system.

123
A Mathematical Model for Amyloid-β Aggregation in the… 1955

5 10

4 8

3 6

2 4

1 2

0 0
10-6 10-4 10-2 100 102 104 106 100 101 102 103 104 105 106 107

5 5

4 4

3 3

2 2

1 1

0 0
100 101 102 103 104 105 106 107 100 101 102 103 104 105 106 107

5 10

4 8

3 6

2 4

1 2

0 0
100 101 102 103 104 105 106 107 100 101 102 103 104 105 106 107

5 10

4 8

3 6

2 4

1 2

0 0
10-6 10-4 10-2 100 102 104 106 10-6 10-4 10-2 100 102 104 106

Fig. 2 Growth kinetics of the first moments and free amyloid monomers and metal ions. The parameters
are ka = 3.4 × 106 , kn = 10−4 , kh = 10−6 , and knh = 3 × 104 . m tot = 5 µM and ltot = 10, 5, 1 for the
case of ε > 1, ε = 1 and ε < 1, respectively. Time is recorded in log scale (Color figure online)

123
1956 E. Asili et al.

Table 5 Table of dimensionless parameters, their values and the assumed value relative to the reference
small parameter ε. The parameters are m tot = 5 × 10−5 , ltot = 5 × 10−6 , ka = 3.4 × 106 , kn = 10−4 ,
kh = 8 × 104 , and knh = 30

Parameter Value Order of ε

ε 0.100 ε
κ1 2.9 × 10−11 O(ε 11 )
κ2 9 × 10−6 O(ε 5 )
κ3 7.8 × 10−8 O(ε 7 )

4.3 Regime with High Metal Ion Concentration

In the presence of high metal ions concentration i.e., ε > 1, because of fast absorption
of the free proteins in the reaction leads to the formation of metal–protein complex
and non-homogeneous complexes, some free metal ions will remain in the system. As
it is depicted in Fig. 2, in all of these regimes and their subregimes, the concentration
of the homogeneous middle terms, Ai Mi s, i ≥ 1, is alternating between decaying or
going into nonzero steady-state.

5 Timescale Analysis

We use singular perturbation theory to study the timescales of the metal–protein

aggregation according to the model. By observing the simulation results, it is clear
that there are several distinct timescales, beginning with a rapid increase and then a
rapid decline in A1 M1 . For applying singular perturbation theory we use the typi-
cal data values listed in Table 5. This analysis helps us in identifying the important
processes and time points in the metal–amyloid aggregation as well as revealing
the dominant and negligible mechanism. The analysis is provided in the limit case
ε 1.

5.1 Singular Perturbation Theory

The singular perturbation theory reduces the system to simple solvable form with
the related timescale, which enables us to determine the slow, intermediate, and fast
mechanisms systematically. According to the rescaled parameters which are listed in
Table 5, we rescale (18) such that all parameters other than ε are of O(1). Therefore,
the new variables are written in terms of ε as follows: as:

κˆ1 ε11 = κ1 , κˆ2 ε5 = κ2 , κˆ3 ε7 = κ3 (20)

123
A Mathematical Model for Amyloid-β Aggregation in the… 1957

By substituting these values in (18), we get:

⎧
⎪ dF1
⎪
⎪ dτ = μ − κˆ1 ε F1 − κˆ3 ε F1 P1
12 2 9
⎪
⎪
⎪
⎪ d P1  
⎪
⎪
⎪
⎪ = κˆ ε 5
F 2
− μP
⎪
⎪ dτ
2 1 1
⎪
⎪
⎪
⎪ d P2
⎪
⎪ = κˆ1 ε13 F1 2
⎪
⎪ dτ
⎨
d1
= 2κˆ1 ε13 F1 2 + κˆ3 ε10 F1 P1 + κˆ2 ε5 μP2 (21)
⎪
⎪ dτ
⎪
⎪
⎪
⎪ d2
⎪
⎪ = 2κˆ1 ε13 F1 2 + κˆ3 ε10 F1 P1
⎪
⎪ dτ
⎪
⎪  
⎪
⎪ dμ
⎪
⎪ = − εμ + κˆ1 ε13 F12 + κˆ2 ε5 μP2
⎪
⎪ dτ
⎪
⎪  
⎪
⎩ d
= − μ + κˆ1 ε12 F12 .
dτ
5.1.1 First Timescale:  = O(")

The first fast time adjustment happens when t = O(ε). The variables related to this
timescale are written as follows

t = ετ ∗ , F1 = F1∗ , P1 = P1∗ , P2 = P2∗ , 1 = ∗1 , 2 = ∗2 , μ = μ∗ ,  = ∗

(22)
With the initial conditions
F1∗ = 0, P1∗ = 0, P2∗ = 0, ∗1 = 0, ∗2 = 0, μ∗ = 1, ∗ = 1 (23)
we expand all variables similar to
F1∗ (τ ∗ ) = F1∗0 (τ ∗ ) + εF1∗1 (τ ∗ ) + ε2 F1∗2 (τ ∗ ) + · · · (24)
As ε → 0, by solving (21) with the initial conditions, we get the following solutions
with the first non-trivial correction terms.
⎧ ∗
⎪ F1 = ετ ∗
⎪
⎪
⎪
⎪ ∗3
⎪ P1∗ = κˆ2 τ ε9
⎪
⎪
⎪
⎪ 3
⎪
⎪
⎪
⎪ κˆ τ ∗3
⎪
⎪ ∗
=
1
ε16
⎪
⎪ P2
⎨ 3
∗ 2κˆ1 τ ∗ 3 16 (25)
⎪
⎪  = ε
⎪
⎪
1
3
⎪
⎪
⎪
⎪ 2 κˆ1τ
∗3
⎪
⎪  ∗
= ε15
⎪
⎪ 2
⎪
⎪ 3
⎪
⎪ μ∗ = 1 − ε − ε2 τ ∗
⎪
⎪
⎩ ∗
 = 1 − ετ ∗ .

It should be noted that the first non-trivial leading solutions are not restricted to ε2 and
the non-trivial term can start from higher orders.

123
1958 E. Asili et al.

5.1.2 Second Timescale:  = O( "15 )

The second timesacle is identified by using

1
t= τ̄ , F1 = F̄1 , P1 = P̄1 , P2 = P̄2 , 1 =  ¯ 2 , μ = μ̄,  = ¯
¯ 1 , 2 = 
ε5
(26)
It is assumed that for this timescale there are no free metal ions. ¯0 and μ̄0 are the
leading order terms. ¯0 = 0 and μ̄0 = μ0 where μ0 is a constant. Concentration of
the homogeneous aggregates continue to drop and rise, while the concentration of the
non-homogeneous aggregates increase since it is the final product of the system. The
key quantity in this timescale is P̄1 , the solutions for this timescale are written as
⎧
⎪
⎪ F̄1 =1
⎪
⎪ κˆ2 −e−μ0 τ̄
⎪
⎪ P̄1 =
⎪
⎪ μ0
⎪
⎪
⎪ P̄
⎨ 2 = κˆ1 τ̄ ε8
¯1 = ( κˆμ0
2 κˆ3
τ̄ + μκˆ30 e−μ0 τ̄ )ε5 (27)
⎪
⎪
⎪
⎪ ¯2 = ( κˆμ2 κ0ˆ3 τ̄ + μ0 κˆ3 −μ0τ̄ 5
e )ε
⎪
⎪
⎪
⎪ κˆ1 κˆ2 μ0 τ̄ 2
⎪μ̄
⎪ = (− 2 − κˆ1 τ̄ )ε8
⎪
⎩¯
 =0

5.1.3 Third Timescale:  = O( "17 )

In the third timescale, homogeneous aggregation is completed and P̌1 , which represents
P1 in this timescale, reaches its maximum concentration. Using:
1
t= τ̌ , F1 = F̌1 , P1 = P̌1 , P2 = P̄2 , 1 =  ˇ 2 , μ = μ̌,  = ˇ
ˇ 1 , 2 = 
ε7
(28)
The concentration of μ̌ continues to decrease. We will consider that μ̌ = c2 , where c2
is a constant. The observation shows that free protein exists in the system to complete
the non-homogeneous aggregation. In this case the solutions are written as:
⎧ √ √
⎪
⎪ F̌1 = c2 − c2 ε
⎪
⎪
⎪
⎪ P̌1 → 1
⎪
⎪
⎪
⎪
⎪
⎪ P̌2 = κ̂1 c2 τ̌ ε6
⎪
⎪
⎪
⎨ √ c2 κ̂ τ̌ 2
ˇ 1 = (κ̂3 c2 τ̌ + 2 1 )ε3
 (29)
⎪
⎪ √
2
⎪
⎪ ˇ
 = κ̂ τ̌ ε 3
⎪
⎪ 2 3 c2
⎪
⎪
⎪
⎪ κ̂1 κ̂2 c22 τ̌ 2 3
⎪
⎪ μ̌ = c − c ε − c ε 2
− ε
⎪
⎪ 2 2 2
⎪
⎩
2
ˇ → 0

123
A Mathematical Model for Amyloid-β Aggregation in the… 1959

Note that c2 can be considered as the initial concentration of μ̌ at this timesacle. It is

ˇ Based on the assumptions,
not possible to derive the algebraic solutions for P̌1 and .
homogeneous aggregations are completed in this timescale, therefore all metal ions
will be used, which results in having P̌1 → 1 and ˇ → 0.

5.1.4 Fourth Timescale:  = O( "111 )

The last timescale is associated with the creation of final product P2 through the aggre-
gation of non-homogeneous complexes. There are no more free metal and protein ions
left in the system. Thus we assume that  = 0 and μ = 0. We assume that the system
reaches the steady-state since the reactions will not cause a considerable concentration
change in the system. The concentration of non-homogeneous complexes, bounded
proteins, and bounded metal ions reaches its maximum value. Thus, the solution can
be represented as

F̃1 → 0, P̃1 → 0, P̃2 → 1,  ˜ 2 → 1, μ̃ → 0, ˜ → 0.

˜ 1 → 1,  (30)

Since the equations cannot be solved algebraically, it is not possible to perform asymp-
totic analysis.

6 Speculations

The proposed model reveals that the ratio of metal ions to Aβs is a critical quantity
in the aggregation of metal–Aβ complexes. This fact has been confirmed by several
experimental studies such as Leal et al. (2012). There are different timescales related
to the three different regimes introduced in Sect. 4. By increasing the metal ions, there
is a rapid induced aggregation of metal–Aβ with no apparent lag-phase. Therefore, in
the low metal regime, it seems it is more effective to administer metal ions–inhibitor
drug or metal chelators such as EDTA (Ethylenediaminetetraacetic acid) in the first
timescale to slow down the progression of the disease. It should be added that the
method of chelation (or similar methods) is more effective in the early stages of the
disease. In general, the type of metal ions, their associated regime and timescale, and
the rate of drug administration are all influential factors in the treatment of AD. It
should be added that in the process of administration of different type of drugs for
the control of Aβ aggregations, all of these factors are participate and create a very
complicated problem in the drug administration and control, which requires extensive
studies.

7 Conclusion

The mathematical structure of current model is founded on the interactions of the metal
ions and Aβ proteins. The biochemical arguments of overall charge together with the
biological structure of the problem based on the previous studies in the literature
were the philosophical bases of this model. Naturally, the mathematical structure of

123
1960 E. Asili et al.

the model implies this in the process of derivation of the master equations. In this
paper, we have developed and studied a mathematical model for the time evolution
of the aggregation of Aβ in the presence of metal ions for the progress of Alzheimer
disease. The kinetic equations for the moments are derived from the master equations
governing the total number of aggregated complexes, bonded proteins, and bonded
metal ions in the form of a system of differential equations. The model reveals four
important events in the progress of AD, which consists of metal–Aβ aggregation,
nucleation, homogeneous aggregation and non-homogeneous aggregation. The model
also shows that the ratio of metal ion to the free Aβ is the critical parameter for the
identification of these regimes. In this regard, three regimes of low, average and high
metal content are identified and studied via the numerical simulations. It is identified
that in the high-metal regime, the metal ions in the steady-state will reach a nonzero
value steady-state, while in the average and low metal regimes this steady-state is
equal to zero. In the second part of the paper, by using singular perturbation theory,
four timescales associated with the different events in the process of Aβ aggregation
in the presence of metal ions are identified. The knowledge of the order of magnitude
of these timescales will help in controlling the progress of the AD by administrating
different forms of treatments.

Acknowledgements The authors would like to thank Henry Research Foundation in Mississippi State
University for the partial support of this project. The authors would like to thank Dr. Joe Hickey, MD, and
Hickey Wellness Center for showing us the importance of this subject and inspiring the second author to
start this project.

8 Appendix

Proof of Theorem 2.1.

• Evaluation of Ph (t)
Setting i = j in Eq. (2), applying the summation over i and using the definition of Ph
yields,
∞ ∞

d Ph (t)  
= kn f (t, 1, 1) + kh f (t, 1, 1)
2
f (t, i − 1, i − 1) − f (t, i, i)
dt
i=3 i=2
∞

− knh m(t) f (t, i, i).
i=2

∞ ∞
Since, f (t, i − 1, i − 1) = f (t, i, i), it is simplified to:
i=3 i=2

d Ph (t)
= kn f (t, 1, 1)2 − knh m(t)Ph (t)
dt

123
A Mathematical Model for Amyloid-β Aggregation in the… 1961

• Evaluation of P(t)

Observe that,
⎡ ⎤
∞ i−1  ∞  i−1  ∞  i−1
∂ f (t, i, j)
= knh m(t) ⎣ f (t, i − 1, j) − f (t, i, j)⎦
∂t
i=3 j=2 i=3 j=1 i=3 j=1
⎡ ⎛ ⎞⎤
∞
  i 
i−1
= knh m(t) ⎣ f (t, 2, 2) + ⎝ f (t, i, j) − f (t, i, j)⎠⎦
i=3 j=2 j=2

= knh m(t)Ph (t),

and since,

∞ i−1
d P(t) d Ph (t)   ∂ f (t, i, j)
= + ,
dt dt ∂t
i=3 j=2

it implies that,

d P(t)
= kn f (t, 1, 1)2 .
dt

• Evaluation of Mh (t)

∞ 
dMh (t)   ∞

= 2kn f (t, 1, 1) + kh f (t, 1, 1)
2
i f (t, i − 1, i − 1) − i f (t, i, i)
dt
i=3 i=2
∞

− knh m(t) i f (t, i, i)
i=2
(31)
Since,

∞
 ∞
 ∞

i f (t, i − 1, i − 1) = i f (t, i, i) + f (t, i, i),
i=3 i=2 i=2

we get:

dMh (t)
= 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) − knh m(t)Mh (t)
dt

123
1962 E. Asili et al.

• Evaluation of M1 (t)

Observe that,

⎡ ⎤
∞ i−1 ∞ i−1  ∞ i−1
∂ f (t, i, j)
i = kn m(t) ⎣ i f (t, i − 1, j) − i f (t, i, j)⎦
∂t
i=3 j=2 i=3 j=2 i=3 j=2
⎡ ⎤
∞ i−1 ∞ i−1 ∞ i−1
= kn m(t) ⎣ i f (t, i, j) + f (t, i, j) − i f (t, i, j)⎦
i=2 j=2 i=2 j=2 i=3 j=2

= kn m(t) [P(t) + Mh (t)] ,

and since,

∞ i−1
dM1 (t) dMh (t)   ∂ f (t, i, j)
= + i
dt dt ∂t
i=3 j=2

we get:

dM1 (t)
= 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t) + knh m(t)P(t)
dt

• Evaluation of M2 (t)

Observe that,
⎡ ⎤
∞ i−1  ∞ i−1  ∞ i−1
∂ f (t, i, j)
j = kn m(t) ⎣ j f (t, i − 1, j) − j f (t, i, j)⎦
∂t
i=3 j=2 i=3 j=2 i=3 j=2
⎡ ⎤
 ∞  i ∞ i−1
= kn m(t) ⎣ j f (t, i, j) − j f (t, i, j)⎦
i=2 j=2 i=3 j=2

= kn m(t)Mh (t),

and since,

∞ i−1
dM2 (t) dMh (t)   ∂ f (t, i, j)
= + j ,
dt dt ∂t
i=3 j=2

we get,

dM2 (t)
= 2kn f (t, 1, 1)2 + kh f (t, 1, 1)Ph (t)
dt

123
A Mathematical Model for Amyloid-β Aggregation in the… 1963

References
ADS (2016) Alzheimers statistics. Alzheimers.net. http://www.alzheimers.net/resources/alzheimers-
statistics/
Bica L, Crouch PJ, Cappai R, White AR (2009) Metallo-complex activation of neuroprotective signalling
pathways as a therapeutic treatment for Alzheimer’s disease. Mol Biosyst 5:134–142
Bolognin S, Drago D, Messori L, Zatta P (2009) Chelation therapy for neurodegenerative diseases. Med
Res Rev 29(4):547–570
Chiti F, Dobson CM (2006) Protein misfolding, functional amyloid, and human disease. Annu Rev Biochem
75(1):333–366
Cohen SIA, Vendruscolo M, Dobson CM, Knowles TPJ (2011a) Nucleated polymerization with secondary
pathways. II. Determination of self-consistent solutions to growth processes described by non-linear
master equations. J Chem Phys 135(6):08B611
Cohen SIA, Vendruscolo M, Welland ME, Dobson CM, Terentjev EM, Knowles TPJ (2011b) Nucleated
polymerization with secondary pathways. I. Time evolution of the principal moments. J Chem Phys
135(6):08B615
Cohen SI, Vendruscolo M, Dobson CM, Knowles TP (2012) From macroscopic measurements to micro-
scopic mechanisms of protein aggregation. J Mol Biol 421(23):160–171 (Amyloid structure, function,
and molecular mechanisms (part I))
Cohen SIA, Vendruscolo M, Dobson CM, Knowles TPJ (2013) The kinetics and mechanisms of amyloid
formation. Wiley, Weinheim, pp 183–209
Crimins JL, Pooler A, Polydoro M, Luebke JI, Spires-Jones TL (2013) The intersection of amyloid beta
and tau in glutamatergic synaptic dysfunction and collapse in Alzheimer’s disease. Age Res Rev
12(3):757–763 (Synaptic Global positioning systems)
Cristóvão JS, Santos R, Gomes CM (2016) Metals and neuronal metal binding proteins implicated in
Alzheimer’s disease. Oxid Med Cell Longev 2016:9812178
Faller P, Hureau C, Berthoumieu O (2013) Role of metal ions in the self-assembly of the alzheimers
amyloid-β peptide. Inorg Chem 52(21):12193–12206
Faller P, Hureau C, La Penna G (2014) Metal ions and intrinsically disordered proteins and peptides: from
cu/zn amyloid-β to general principles. Acc Chem Res 47(8):2252–2259
Hamley IW (2012) The amyloid beta peptide: a chemists perspective. role in alzheimers and fibrillization.
Chem Rev 112(10):5147–5192
Hebert LE, Weuve J, Scherr PA, Evans DA (2013) Alzheimer disease in the united states (2010–2050)
estimated using the 2010 census. Neurology 80(19):1778–1783
Hegde ML, Bharathi P, Suram A, Venugopal C, Jagannathan R, Poddar P, Srinivas P, Sambamurti K, Rao
KJ, Scancar J, Messori L, Zecca L, Zatta P (2009) Challenges associated with metal chelation therapy
in Alzheimer’s disease. J Alzheimers Dis 17(3):457–468
Jakob-Roetne R, Jacobsen H (2009) Alzheimer’s disease: from pathology to therapeutic approaches. Angew
Chem Int Ed 48(17):3030–3059
Jarrett JT, Lansbury PTJ (1993) Seeding “one-dimensional crystallization”participating only in the process
of nucleation. They are of amyloid: a pathogenic mechanism in Alzheimer’s disease and scrapie. Cell
73(6):1055–1058
Karran E, Mercken M, Strooper BD (2011) The amyloid cascade hypothesis for Alzheimer’s disease: an
appraisal for the development of therapeutics. Nat Rev Drug Discov 10(9):698–712
Kepp KP (2012) Bioinorganic chemistry of Alzheimers disease. Chem Rev 112(10):5193–5239
Knowles TPJ, Waudby CA, Devlin GL, Cohen SIA, Aguzzi A, Vendruscolo M, Terentjev EM, Welland
ME, Dobson CM (2009) An analytical solution to the kinetics of breakable filament assembly. Science
326(5959):1533–1537
Leal SS, Botelho HM, Gomes CM (2012) Metal ions as modulators of protein conformation and misfolding
in neurodegeneration. Coord Chem Rev 256:2253–2270
Lovell MA, Robertson JD, Teesdale WJ, Campbell JL, Markesbery WR (1998) Copper, iron and zinc in
Alzheimer’s disease senile plaques. J Neurol Sci 158(1):47–52
Mattson MP (2004) Pathways towards and away from Alzheimer’s disease. Nature 430(7000):631–639
Miller LM, Wang Q, Telivala TP, Smith RJ, Lanzirotti A, Miklossy J (2006) Synchrotron-based infrared
and X-ray imaging shows focalized accumulation of cu and zn co-localized with β-amyloid deposits
in Alzheimers disease. J Struct Biol 155(1):30–37

123
1964 E. Asili et al.

Mucke L, Selkoe DJ (2012) Neurotoxicity of amyloid β-protein: synaptic and network dysfunction. Cold
Spring Harb Perspect Med 2(7):a006338
Oosawa F, Asakura S (1975) Thermodynamics of the polymerization of protein. Academic Press, London
Oosawa F, Kasai M (1962) A theory of linear and helical aggregations of macromolecules. J Mol Biol
4(1):10–21
Oosawa F, Asakura S, Hotta K, Imai N, Ooi T (1959) G-F transformation of actin as a fibrous condensation.
J Polym Sci 37(132):323–336
Pithadia AS, Lim MH (2012) Metal-associated amyloid-β species in Alzheimer’s disease. Curr Opin Chem
Biol 16:67–73 (Bioinorganic chemistry biocatalysis and biotransformation omics)
Prince M, Wimo A, Guerchet M, Ali G-C, Wu Y-T, Prina M (2016) World alzheimer report 2016: improving
healthcare for people living with dementia. Technical report. https://www.alz.co.uk/research/world-
report-2016
Rauk A (2009) The chemistry of Alzheimer’s disease. Chem Soc Rev 38:2698–2715
Sarell C (2010) The copper-amyloid-beta-peptide complex of Alzheimers disease: affinity, structure, fib-
ril formation and toxicity. Ph.D. thesis, School of Biological and Chemical Sciences, University of
London, Queen Mary, England
Turányi T, Tomlin AS (2014) Analysis of kinetic reaction mechanisms. Springer
Viles J H (2012) Metal ions and amyloid fiber formation in neurodegenerative diseases. Copper, zinc and
iron in Alzheimer’s, Parkinson’s and Prion diseases. Coord Chem Rev 256(1920):2271–2284 (Metal
Ions in neurodegenerative diseases)
Wilson MR, Yerbury JJ, Poon S (2008) Potential roles of abundant extracellular chaperones in the control
of amyloid formation and toxicity. Mol BioSyst 4:42–52

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

123