Computational Modeling Studies of the
Beta-Amyloid Protein Binding to Develop
Drugs for the Treatment of Alzheimer’s Disease
Sanidhya D Tripathi1,2, Patrycja A Krakowiak1,2 and Jerry A Darsey2,3*
1Arkansas School for Mathematics, Sciences, and the Arts, Hot Springs, AR, USA
2Center for Molecular Design and Development, Little Rock, AR USA
3Department of Chemistry, Little Rock, AR USA
Submission: December 17, 2017; Published: December 21, 2017
*Corresponding author: Jerry A Darsey, Department of Chemistry & Center for Molecular Design and Development at the University of Arkansas at Little Rock, Little Rock, AR 72204, USA, Tel: 501 569-8828; Email: email@example.com
How to cite this article: Sanidhya D T, Patrycja A K, Jerry A D. Computational Modeling Studies of the Beta-Amyloid Protein Binding to Develop Drugs for
the Treatment of Alzheimer’s Disease. Curr Trends Biomedical Eng & Biosci. 2017; 10(5): 555799. DOI: 10.19080/CTBEB.2017.10.555799.
Alzheimer’s Disease (AD) is a progressive neurodegenerative disease that affects memory and other mental functions due to the accumulation of Amyloid-Beta (Aβ) plaques. The functions carried out by the Aβ protein leads to abnormalities in the suffering patient. If a molecule binds to the Aβ protein, it could lead to potential treatment. To propose new treatments for AD, novel molecules were identified that bind to the active site of Aβ. First, a correlation between reported
IC50 values of fourteen chemicals and their energy gap, dipole moment, and total energy were analyzed. The best correlation was discovered between 1/IC50 values and total energy (R2=0.9279). Then, each identified molecule was virtually modified and its new IC50 value was determined utilizing Gaussian 09, a molecular modeling program, which was used to predict total energies. Some of the most useful modifications included the addition of oxygen (-O), chlorine (-Cl), and fluorine (-F) atoms. The results of this study can be used as a tool to finding novel, medically relevant new molecules for other neurodegenerative diseases and assist researchers in developing new drugs to treat AD
Keywords: Alzheimer’s disease; Beta-Amyloid protein; Drug design; Protein misfolding; amyloid-β aggregation; Oxidative stress; Neuroinflammatory processes; Parkinson’s disease; Huntington’s disease; Amyotrophic lateral sclerosis (ALS), IC50; Computational molecular modelling; Density functional theory
Nearly six million people suffer from Alzheimer’s disease (AD) in the United States, and the number of patients with this multifaceted neurodegenerative disorder could rise to include as many as 16 million people by 2050 . Moreover, in 2017, this disease is estimated to cost $259 billion in global health care expenditures . Alzheimer’s disease is an irreversible neurological disorder that is characterized by memory loss, personality changes, and decline in thinking abilities. Though memory loss is the most widely recognized symptom of Alzheimer’s disease, many other complications, such as a range of behavioral and cognitive deteriorations, are also common. The cognitive symptoms include memory loss, confusion, and disorientation while the behavioral symptoms include depression, insomnia, and hallucinations . Due to the often-intense nature of the symptoms, AD patients can suffer in a multitude of ways and eventually become bedridden and dependent on custodial care. On average, death occurs nine years after diagnosis .
AD is characterized at a molecular level by factors such as protein misfolding and aggregation, oxidative stress, and neuroinflammatory processes. The complex pathogenesis has provided for only a limited development of effective therapeutic solutions to combat disease progression [3,4]. Currently, there are several symptomatic medications and treatments for AD. Some of these registered treatments include cholinesterase inhibitors, donepezil, galantamine, and rivastigmine [2,4,5]. Although these drugs are beneficial to some degree, they only provide temporary improvements in cognition and function of the diagnosed patient. Because there is still no efficient medication to prevent the deterioration of the neurodegenerative processes in AD patients, there remains an urgent need for novel drugs that may slow the neurodegenerative disease and potentially reverse its effects. Overall, no treatment with a strong disease modifying effect is currently available.
The primary suspects that are thought to cause Alzheimer’s disease are amyloid-beta (Aβ) and hyper-phosphorylated tau proteins. Agents inhibiting one or both forms of aggregation
are promising drug candidates for AD . Amyloid-β peptide
chains consist of 40 to 42 amino acids and are suspected to
contribute to the degenerative condition of AD. The amyloidbeta
precursor protein (APP) is found in many types of cell
membranes. The actions of the β-secretase and γ-secretase
release the Aβ proteins at an elevated rate into the cytoplasm.
These proteins accumulate at high levels in neural cells; they are
internalized and folded into a β-pleated shape which results in
stacking that forms long aggregates known as senile plaques .
Acting as monomers, dimers, or multimers in cell membranes,
the Aβ proteins interfere with neurotransmission and memory
formation. As the protein aggregates continue to increase in size,
they become insoluble in the bloodstream. Scientists have begun
to study this problem and suggest treatment strategies, which
include inhibitors of the β-secretase and γ-secretase activities,
as drugs and physiological compounds to prevent amyloid-β
aggregation in the brain .
Much research in pursuit of novel compounds that have
efficient inhibitory properties of AD still needs to be completed.
For example, innovative techniques are needed to offset the
financial expenditures related to the traditional tissue culture
and model organism experiments. Though several vaccines are
in the process of being developed, a breakthrough discoveries
are needed to speed up this slow process. A successful drug
discovery process would not only revolutionize the treatment of
AD and improve the lives of countless people all over the world,
but it could also act as an example of how beneficial drugs for
other diseases could be discovered in the future. In conjunction
with previous research, drug candidates that bind to Aβ proteins
could be designed based on specific properties known to
enhance this binding. For example, chemical compounds that
have a low IC50 value (a measurement of the effectiveness of a
drug or inhibitorof binding to a biological molecule by IC50 % and
expressed in micromolar concentration, M) could be identified
and evaluated to better treat AD. As part of the described study,
it was proposed that there exists a correlation between the
IC50value and other notable parameters of a chemical compound
and its interaction with Aβ proteins. Using computational
molecular modeling and virtual simulations, molecules best
suited to bind and therefore inhibit Aβ proteins were identified
As shown in previous studies, determining a molecule to
bind to Aβ would result in a major step towards creating a drug
to help treat Alzheimer’s disease [8,9]. By determining specific
molecules that bind to the Aβ active site, a correlation can be found
between its binding affinities and energies. To achieve this goal,
computational molecular modeling was used in this research.
The goal of this study was to create a molecule that would bind
efficiently to the Aβ protein by finding correlations between
chemical properties when creating the modified molecule. The
plan of this research was to alter molecules that have been
previously identified as inhibitors of Aβ and then determine by
how much these modified molecules change the IC50 values of the
“parent” molecule. By collecting data such as the dipole moment
(the product of the separation of two ends of a dipole molecule
and the magnitude of the charges), highest occupied molecular
orbital (HOMO), lowest unoccupied molecular orbital (LUMO),
energy gap (value found by subtracting the HOMO value from
the LUMO value), and plotted vs. the IC50 value, a correlation
was determined and analyzed after the modifications were
incorporated. New, IC50 values were then predicted using the
line of best fit. The newly designed compounds with a lower
predicted IC50 values would be a potential new drug to treat
The research was initiated by locating and analyzing the
native structure of the Beta-Amyloid (Aβ) protein. After a detailed
study of their properties and relevance to Alzheimer’s disease,
21 compounds were identified in publications and patents that
had been tested as potential inhibitors of the Aβ protein .
However, the molecules mentioned in various sources were not
always described using the IUPAC naming system. Thus, the
correct name of these molecules had to be determined.
To accomplish this task, PubChem (https://pubchem.ncbi.
nlm.nih.gov/) was utilized as the database to find important
information about the molecules and their activities with respect
to Aβ protein. PubChem, a free public tool is maintained by the
National Center for Biotechnology Information (NCBI) Center at
the National Institutes of Health (NIH) and provides the IUPAC
(International Union of Pure and Applied Chemistry) name,
InChIKey (International Chemical Identifier), and chemical
structures of more than 92.1 million compounds. To enable the
extraction of this information from the PubChem database, the
OPSIN tool (Open Parser for Systematic IUPAC nomenclature),
freely available from the University of Cambridge, was used.
The InChIKey was identified for each chemical and was then
used to search for chemical properties on the PubChem website.
In ChIKey is a textual chemical identifier for compounds that
encodes chemical information and can be used to search
online databases. To extract all InChIKeys for the 22 identified
compounds, a Google Sheet was created with the first column
showing the original name of each chemical as cited in each
publication. Table 1 shows the compounds identified by
thoroughly searching the literature and their respective IC50
Using the computational molecular modeling program
(Gaussian 09), each chemical was altered by adding specific
functional groups and thus resulted in the generation of a
new group of compounds. Elements, bond types/sizes, and
chemical groups were altered using tools accessible to the user.
Each molecule was analyzed using the optimization method,
which was applied to clean the molecular bonds and structures
utilizing the DFT (Density Functional Theory) 6-31G* basis set
(defined for atoms H to Zn and is a double-zeta polarized basis
set). The total energy (a.u.), dipole moment (Debye), and HOMOLUMO
(Highest Occupied Molecular Orbital-Lowest Unoccupied
Molecular Orbital) energy gap (Hartrees), were established and
recorded for each molecule.
Once the data from all the trials was recorded, binding
affinities, measured by the experimentally determined IC50
values of each of the 22 molecules. These values would serve
as a tool to later produce modified compounds and their IC50
values to be compared to the original unmodified molecule. The
data was entered into an Excel 2016 spread sheet and graphed
using the scatter plot tool. Plots included IC50 , log IC50 , and
1/ IC50 values vs. the various calculated parameters (the total
energy, dipole moment (Debye), and HOMO-LUMO energy gap).
From these plots, the strength of correlations were determined.
It should be noted that some data points were outliers and were
therefore removed from the data set since they were more than
two standard deviations away from the linear trendline. Plots
were included with and without the outliers (Figures 1-4).
Data points from the plots of the various Amyloid-Beta
(Aβ) inhibitors were used to identify characteristics that would
show the best correlations with IC50 values. The values of the
total energy, dipole moment, and energy gap of the promising
compounds were calculated using the Gaussian 09 software.
Numerous correlations were looked at including total energy vs.
log IC50 , total energy vs. 1/ IC50 , total energy vs. IC50 , dipole
moment vs. log IC50, dipole moment vs. 1/IC50, dipole moment
vs. IC50 , energy gap vs. log IC50 , energy gap vs. 1/ IC50 , and
energy gap vs. IC50 values. The analysis revealed that the most
highly correlated properties were total energy vs. 1/ IC50
values and total energy with log IC50 values (Figures 1-4). This
information was then used to find novel, modified molecules
and more efficient Aβ compounds that could become new
drugs to potentially reduce the negative physiological effects of
The log IC50 values of all the literature derived compounds
were used to determine a relationship between total energy and
IC50 values. The plot reveals that the log IC50 values for all. The
log IC50 values were used to determine a relationship between
total energy and IC50 values. The plot reveals that the log IC50
values are proportional to the total energy; this statement is
supported by the linear equation and R2 =0.8157 value of the
The 1/ IC50 values of all the literature derived compounds
were used to determine a relationship between total energy and
IC50 values and are plotted with respect to total energy that was
determined using the Gaussian 09 program (see Figure 3). The
plot depicts a correlation between the 1/ IC50 values and total
energy, with an R2=0.5076 value.
The data points in Figure 4, show the I IC50 values that
were obtained from previous experimental research after the
removal of the outliers; while the total energy associated with
each was obtained using the Gaussian 09 software. The 1/I
C50 values were used to determine a relationship between the
two parameters. The graph depicts how the 1/ IC50 values are
inversely proportional to the total energy, according to the
equation and R =0.9279 value of the trendline.
Another set of trials was conducted after the highest
correlation was determined i.e. total energy vs. 1/IC50 values).
To find novel molecules with lower IC50 values than those
mentioned in the literature, simple modifications had to be made
to accomplish this task. Therefore, the fourteen compounds
were modified by an addition of chemical groups/elements
to the existing structure and a new IC50 value was calculated
after the total energy value was outputted from the Gaussian
09 program. For this round of testing, each molecule was run
first under the optimization method, and then using the DFT
(Density Functional Theory) with the 6-31G* basis set. Table 2
shows the results of the new IC50 values after modifications had
After running the DFT with the 6-31G* basis set on all the
compounds, 3 compounds identified as C02, C13, C22 yield the
lowest IC50 (largest binding affinities) values in the sets tested.
Their chemical structures are show in order in Figure 5a-5c.
The accumulation of the Beta-Amyloid (Aβ) proteins, which
are derived from the Amyloid Precursor Protein (APP), can form
plaques within the human brain [6,11-15]. This ability to form
plaque can cause numerous health problems to the patient, which
culminate in Alzheimer’s disease. It was therefore beneficial to
identify molecules that could bind to these plaque molecules
marking them for excretion from the body. The purpose of the
current study was to investigate not only whether such potential
drug molecules exist, but also to design modifications that would
improve their binding capacity. To accomplish this goal, first, it
was important to establish if there is a correlation between the
literature-derived IC50 values of various chemicals and notable
properties including total energy, energy gap, and/or the dipole
moment. The property that correlated best with was the 1/IC50
values vs. total energy with an R2=0.9279. Therefore, total energy
was calculated for the proposed modifications to the original
compounds and the IC50 values for these were determined using
total energies and the linear regression equation in Figure 4. The
chemical modifications with lower determined IC50 values were
selected for further consideration. Any chemical that exhibited
lower IC50 values than its original parent molecule should be
tested experimentally for efficacy of binding Aβ plaques and its
ability to treat Alzheimer’s disease.
In the computational molecular modeling field of research,
having a reference chemical is a vital aspect of future
research. Based on the procedures of this current study, novel
compounds could be identified for various other diseases
such as Parkinson’s, Huntington’s, and Amyotrophic lateral
sclerosis (ALS). In addition to the distinguishing aspect of this
research that allows for modeling treatment for various notable
diseases, it also allows for far less time consuming laboratory
work since so much of the process is done computationally.
Because synthesizing novel chemicals and measuring their
IC50 values is both laborious and expensive, the computational
research reported here is invaluable to the future expansion of
pharmacology. Because scientists do not have to spend large
amounts of time synthesizing chemicals and testing them, the
computer-aided approach offers a much faster and efficient way
to obtain data critical to improving treatment of many deadly
diseases. Using this procedure, hundreds of molecules can be
analyzed and a far greater number of novel compounds can
be identified as promising medications. Ultimately, identifying
compounds with high binding affinities for the Aβ protein (with
lower IC50 values) that can act as inhibitors, can be applied
to Alzheimer’s disease research to develop new drugs to treat
The set of experiments described here reveal an opportunity
to design new molecular Amyloid Beta (Aβ) inhibitors using
computational molecular modeling techniques. The hypothesis
that new IC50 values of virtually altered chemicals could be
predicted by first finding a linear correlation between reported
IC50 values and a specific, calculable property of these molecules
was supported when a strong correlation identified between
Total Energy and 1/ IC50 values led to the discovery of lower
IC50 values of the modified molecules. More specifically, the linear
regression equation generated after graphing 1/ IC50 values vs.
Total Energy was used to predict how modifying the previously
reported chemicals could lower their predicted IC50 values
and so increase their binding capacities to Aβ. The data that
allowed for these predictions was generated using the quantum
mechanical Gaussian 09 software program. The significance
of this novel work is that it suggests new drug candidates to
treat Alzheimer’s disease (AD). One of the main causes of the
mental symptoms of AD are a buildup of Aβ plaques in the brain.
By proposing modifications to various chemicals that improve
binding and therefore inhibit Aβ, the experiments described
here allow for a set of novel chemicals to be tested in tissue
culture and animal models with Aβ buildup. The benefit of doing
the preliminary experiments virtually using a computer-aided
approach is that it is cheaper, faster and more efficient than
biological synthesis and testing. Finding leading drug candidates
for the treatment of AD, is critical for future researchers to be
able to develop effective medications to combat this debilitating
and deadly disease.
The presented research could also lead to investigations
of other protein misfolding diseases including Huntington’s
disease, Parkinson’s disease, and various Prion diseases. The
computational molecular modeling techniques described in
this manuscript could be used to identify molecules that bind
proteins associated with specific diseases, and then promote the
discovery of novel drugs by proposing modifications that could
make these molecules more efficient inhibitors, thus producing
better treatments for these neurodegenerative diseases.
These computationally driven findings could allow for quicker
discoveries of treatments that could improve and even save
millions of lives.
Sanidhya Tripathi (ST) carried out all the procedures. ST
and Jerry Darsey (JD) conceived of the study and performed the
statistical analysis. ST drafted the manuscript with the help of
PK and JD. JD and PK read and approved the final manuscript.
The authors would also like to thank the University of
Arkansas at Little Rock Center for Molecular Design and
Development and the Arkansas School for Mathematics, Sciences,
and the Arts for facilitating my research and providing the venue
to share these research findings.
Furthermore, the authors would like to thank Ms. Sumreen
Gul, a Ph.D. student, for help with running the Gaussian 09
software and giving one of the authors (ST) consistent feedback
and tips throughout the process. Finally, the author (ST) would
like to thank the parents Sanjay and Arati Tripathi for all the
encouragement and support.