A Drug Repositioning Approach Reveals Ergotamine May Be a Potential Drug for the Treatment of Alzheimer’s Disease

Abstract

Background:

Alzheimer’s disease (AD) is a neurodegenerative disorder that is the most common form of dementia in the elderly. The drugs currently used to treat AD only have limited effects and are not able to cure the disease. Drug repositioning has increasingly become a promising approach to find potential drugs for diseases like AD.

Objective:

To screen potential drug candidates for AD based on the relationship between risk genes of AD and drugs.

Methods:

We collected the risk genes of AD and retrieved the information of known drugs from DrugBank. Then, the AD-related genes and the targets of each drug were mapped to the human protein-protein interaction network (PPIN) to represent AD and the drugs on the network. The network distances between each drug and AD were calculated to screen the drugs proximal to AD-related genes on PPIN, and the screened drug candidates were further analyzed by molecular docking and molecular dynamics simulations.

Results:

We compiled a list of 714 genes associated with AD. From 5,833 drugs used for human diseases, we identified 1,044 drugs that could be potentially used to treat AD. Then, amyloid-β (Aβ) protein, the key molecule involved in the pathogenesis of AD was selected as the target to further screen drugs that may inhibit Aβ aggregation by molecular docking. We found that ergotamine and RAF-265 could bind stably with Aβ. In further analysis by molecular dynamics simulations, both drugs exhibited reasonable stability.

Conclusions:

Our work indicated that ergotamine and RAF-265 may be potential candidates for treating AD.

Keywords

Alzheimer’s disease drug repositioning ergotamine molecular docking risk genes

INTRODUCTION

Alzheimer’s disease (AD) is the most common neurodegenerative disorder and the main cause of dementia and elderly disability. Currently, more than 50 million people suffer from AD or other dementia globally, and the number may reach 152 million by 2050, and the cost of treating and caring for AD patients may be as high as $9.12 trillion per year.1,2, 1,2 The rapid growing number of patient and the subsequent economic burden makes AD one of the most pressing health and financial issues of our time.

The clinical manifestations of AD patients are mainly reflected in memory dysfunction, progressive cognitive dysfunction and emotional dysfunction. The pathogenesis of AD is very complex, which mainly involves deposition of amyloid-β (Aβ) protein in cerebral cortex and hippocampus and neurofibrillary tangles formed by excessive phosphorylation of tau protein in brain nerve cells. 3 In addition, there are other pathogenic mechanisms of AD, including neuroinflammation, mitochondrial dysfunction, oxidative stress response and imbalance of intestinal bacteria AD group.

Until now, there is still no cure for AD, although some treatments that may change the progression of the disease are available. Also, there are some nonpharmacologic and pharmacologic interventions that may help treat AD symptoms. 4 Nonpharmacologic treatments include physical activity, memory and orientation exercises, as well as music- and art-based therapies that aim at reducing behavioral and psychological symptoms and maintaining or improving cognitive function and overall quality of life. In the past decades, a lot of resources have been devoted to developing drugs for AD. But so far, only a few drugs can slow the worsening of symptoms and none is able to cure or is appropriate for all individuals living with the disease. The US Food and Drug Administration (FDA)-approved drugs, including donepezil, rivastigmine, galantamine, memantine and memantine combined with donepezil, are aimed at improving symptoms of AD without affecting the brain changes underlying the symptoms or altering the course of the disease. Among these drugs, memantine works as a glutamate receptor antagonist that protects the brain from excessive levels of glutamate, while the others can slow the breakdown of acetylcholine in the brain by inhibiting acetylcholinesterase. Available evidence suggests that amyloid plaques generated by Aβ and neuronal fibrillary tangles composed of hyperphosphorylated tau proteins are the main factors causing neuronal cell death and brain tissue loss. 5 Thus, inhibiting the aggregation of Aβ or enhancing its clearance has become the focus of many studies on developing drugs for AD treatment. Two other drugs approved by the FDA in the past 20 years, aducanumab and lecanemab, are supposed to remove Aβ from the brain and slow cognitive and functional decline in patients at the early stage of AD.6 –8 However, there are still controversies over their efficacy or safety.9 –11

There are other treatments targeting the underlying biology of AD that are in the research pipeline. They usually address the brain changes associated with AD, including but not limited to tau accumulation, altered cell metabolism and inflammation. For example, donanemab, an experimental medication that targets Aβ, has finished phase 3 clinical trials and demonstrated a significant effect in halting the development of AD. 12 Gantenerumab, another drug under investigation and targeting the flexible N-terminal of Aβ and core amino acids, doesn’t show expected efficacy in eliminating Aβ and decreasing cognitive loss in AD patients during phase 3 clinical trials. 13

Since new drugs take a long time and very high costs to develop, and the rate of failure is high, drug repositioning that uses existing drugs to treat new diseases has drawn much attention. This approach has a number of advantages, as it can greatly reduces the time of drug research and development, effectively reduces the high cost of new drug research and development. 14 Many studies have tried to find new treatment for AD via the already-approved drugs. Drug repositioning techniques have identified 40% of the drug candidates in clinical trials studying potential new treatments for AD. 15 Specifically, a few of these drugs can adjust glutamate and neurotransmitter levels, offering potentially novel treatments for AD. 16 For instance, levetiracetam, a drug used for treating epilepsy by altering neurotransmitter release, can improve the spatial memory of AD patients with epilepsy in clinical trials.17,18, 17,18 Troriluzole, the prodrug of riluzole for treating amyotrophic lateral sclerosis, is able to improve cognitive function in AD by reducing glutamate levels through increasing glutamate re-uptake and decreasing its release. 19 Dapagliflozin, a drug used to treat type 2 diabetes, is found to be able to improve AD pathology in rat model of AD. 20 Thus, drug repositioning is promising in searching new drug candidatesfor AD.

In this study, we systematically explored the relationship between AD and available drugs. The drugs used for human diseases and their targets were retrieved from the DrugBank database and were screened to identify the candidates for AD treatments by evaluating the drug-disease correlations through a network-based approach. Then, we further evaluated the selected drug candidates through molecular docking and molecular dynamics simulation, and identified the drugs that may have therapeutic effects on AD.

MATERIALS AND METHODS

Collection of AD-related genes and drugs

In this study, we mainly collected the gene sets of AD from three commonly used databases: Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg/), 21 OMIM (https://www.omim.org/), 22 and PheGenI (https://www.ncbi.nlm.nih.gov/gap/phegeni). 23 From KEGG, 384 genes included in the Alzheimer disease pathway (hsa05010) were retrieved. In OMIM, AD risk genes were identified by searching the keyword ‘Alzheimer disease’ and 157 genes were retrieved. In PheGenI, ‘Alzheimer Disease’ was used as the target phenotype and P-Value<1×10^–6 as the threshold to search genes related to AD, resulting a list of 376 risk genes. When these genes were pooled together, 714 AD unique genes were obtained (Supplementary Table 1).

In addition, AD-related genes were also retrieved from Agora (https://agora.adknowledgeportal.org/). Agora is a web application that hosts high-dimensional human transcriptomic, proteomic, and metabolomic evidence for genes associated with AD, and it contains a list of nascent drug targets for AD that are nominated by AD researchers. The drug targets from Agora were retrieved and combined with those 714 AD-related genes to screen the potential drugs.

Drugs and their targets were obtained from the DrugBank database. 24 Then, 5,833 drugs used to treat human diseases (i.e., those target one or more human genes), as well as their targets, were kept.

Constructing the human protein-protein interaction network (PPIN)

Human PPIN was used to assess the relationship between drugs and AD, which was built by combining the protein-protein interaction data from Human Reference Protein Interactome (HuRI; http://www.interactome-atlas.org/), 25 Protein Interaction Network Analysis (PINA; https://omics.bjcancer.org/pina/), 26 and Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; https://cn.string-db.org/) 27 databases. We collected 8,253 proteins and their 52,548 protein interaction pairs from HuRI, 16,183 proteins and their 166,778 protein interaction pairs from PINA, and 6,982 proteins and their 22,892 protein interaction pairs from STRING. As each of these datasets included some proteins and protein interaction pairs that were overlapped with other datasets, we eventually compiled a set of 18,514 proteins and 237,316 protein interacting pairs after removing the redundancy (Supplementary Figure 1), which was used to build the human PPIN (Supplementary Figure 2, Supplementary Table 2).

Enrichment analysis of AD risk genes

Functional enrichment analysis on the AD-related genes was performed using the plug-in ClueGO 28 in Cytoscape. 29 In a functionally organized network, ClueGO displays the non-redundant biological terms for sizable gene clusters. The statistically significant Gene Ontology (GO) terms and pathways were identified by adjusted p < 0.05. Then, the interaction network of the AD risk genes was analyzed to explore the pathological mechanisms of AD.

Screening potential drugs for AD treatment

In an earlier work, we proposed an algorithm to measure the drug-disease relationship by mapping drug targets and disease genes onto human PPIN, and using the path length between drug targets and disease modules to assess the proximity between drugs and diseases. 30 The method was adopted in this study to screen the AD drug candidates. First, AD risk genes were mapped to the human PPIN. Second, we calculated the mean minimum path length traveled by drug targets to the AD disease modules and adjacent genes (Eq. 1). Briefly, given G, the set of AD risk genes; T, the set of target genes of a drug; D, the node degree of AD risk genes (i.e., the number of edges in the human PPIN that are connected to each risk gene); distance d(g, t), the shortest path length between nodes g (AD risk gene in our case; g∈G) and t (drug target in our case; t∈T) in the human PPIN; d(G,T), the mean distance between the targets of the given drug and AD risk genes on the human PPIN. Then, we have: $d (G, T) = \frac{1}{| T |} \sum_{t \in T} \min_{g \in G} (d (g, t) + w)$ (1) where w, the weight of a target gene, is determined by two ways, i.e., if t is one of the risk genes of AD, then w = – ln(D + 1); otherwise, w = 0. By this way, the association of a drug and AD can be measured by its network distance to AD risk genes on the human PPIN. A large distance indicates that the targets of the drug are not close to the AD risk genes and the drug may have little impact on AD, thus less likely to be a drug for AD treatment.

To assess the relevance of the association between drugs and AD, we used the same algorithm to generate reference distances for hypothetical drugs. In human PPIN, a randomly selected set of proteins, which has the same number of proteins as the drug targets, is used to represent a hypothetical drug. The network distance between the hypothetical drugs and AD risk genes on human PPIN was calculated using Eq. 1. The procedure was repeated 10,000 times to generate the reference distance between the hypothetical drugs and the AD. Since the targets of the hypothetical drugs are randomly selected from the PPIN, the distribution of distances of hypothetical drugs to the AD risk gene module provides a reasonable reference. In addition, as the number of drugs in DrugBank is relatively large, it is reasonable to assume that most of the drugs in DrugBank are unlikely to be effective for AD and they will have distances similar to the hypothetical drugs. Then, for the drugs and hypothetical drugs, kernel density estimation (KDE) based on the gaussian kernel was used to calculate the theoretical probability density for the distances. The kernel density curves depicted the distribution of network distances between the drugs (or hypothetical drugs) to AD. These potential drugs constitute a small proportion in DrugBank, as reflected in the kernel density plot, where smaller densities are more tightly linked to causal genes in the lower network distance intervals. Then, the two kernel density curves were compared to determine a threshold value to screen the potential drugs.

Molecular docking

Molecular docking was used to simulate the interaction between small compounds and the target proteins and to predict the affinities between drugs and proteins. 31 We utilized AutoDock Vina (version 1.1.2) 32 and AutoDockTools (version 1.5.7) 33 to perform molecular docking. From the Protein Data Bank (PDB), the structure of Aβ (PDB:1IYT) was retrieved. 34 We added hydrogen atoms to the protein via PyMOL (version 2.5.4) (obtained from https://www.schrodinger.com/pymol; PyMOL web site: https://pymol.org/). Then, the pdbqt file was generated by importing the modified protein conformations into AutoDockTools. The SDF files of the drug molecules were downloaded from PubChem 35 and imported into OpenBabel (version 3.1.1) 36 to make pdbqt files by adding hydrogen atoms. The neurotoxicity and aggregation of Aβ peptide have been demonstrated to be intimately associated with residues Y10-A21. 37 Consequently, the docking box was positioned in the docking pocket formed by these residues. AutoDock Vina was used to perform semi-flexible docking to calculate binding affinity. For every drug molecule, twenty potential interaction conformations were produced, the optimal conformation was chosen via the lowest binding energy. Additionally, we employed visual analysis tools PyMOL (http://www.pymol.org/) and Discovery Studio (https://www.3ds.com/products/biovia/discovery-studio) to show the results of molecular docking in the study.

Molecular dynamics simulation

Molecular dynamics (MD) can simulate the interactions of the atoms in a drug-protein complex and is widely used in drug repositioning. 38 In order to evaluate the stability of protein-ligand interactions, the drug molecules and proteins with the best conformations were ultimately chosen for MD simulations with Gromacs (version 5.1.4). 39 We generated topology files for proteins and drug molecules using the AMBER99SB protein force field and the GAFF force field via ACPYPE, respectively. 40 After that, the protein-ligand complexes were placed in the center of a cubic box, 1 nm from the box’s border, and solvated with simple point charge (SPC) water molecules. Additionally, we injected 3 Na+ to maintain the system’s electrical neutrality. Next, in order to bring the system to convergence and its lowest energy state, the energy of the protein-ligand complex was minimized using the steepest descent algorithm. The system was followed to isothermal (300 K) and isobaric (1 bar) equilibrium via NVT and NPT equilibrium, and then molecular dynamics simulations were run for 90 ns. Root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg) and the number of hydrogen bonds were calculated.

RESULTS

AD-related genes

From different resources, we have compiled a list of 714 genes potentially associated with the risk of AD. The pathogenic mechanisms of AD can be clarified by investigating the potential biological functions of these genes. Functional enrichment analysis on the AD genes showed that the GO biological process (BP) terms, including regulation of neurotransmitter secretion, regulation of interleukin-1 beta production, astrocyte differentiation and regulation of Aβ formation, were significantly enriched. Molecular function (MF) terms such as regulation of MAP kinase activity, positive regulation of kinase activity and regulation of protein kinase activity were enriched. Cellular component (CC) terms such as mitochondrial membrane, proteasome complex and synaptic membrane, were enriched (Fig. 1). Consistently, pathways related to neurodegenerative disorders, such as neurodegeneration, AD, prion diseases, and Parkinson disease, were enriched in the AD-related genes. Additionally, several pathways related to diabetes, such as Type II diabetes mellitus and diabetic cardiomyopathy, were also enriched. Pathways involved in neuroinflammation, such as apoptosis, Wnt signaling, TNF signaling, and T cell receptor signaling were also enriched. These results demonstrated that the collected risk genes of AD were diverse in function and were a relatively comprehensive coverage of genes involved in the pathogenesis of the disease.

Fig. 1

Results of functional enrichment analysis. In each panel, the X-axis shows the fraction of genes in the item as a percentage of all genes, and the Y-axis shows the items that are enriched for the genes. (A) BP terms; (B) MF terms; (C) CC terms; and (D) KEGG pathway enrichment terms.

Screening drugs for AD

Drugs and biologics can act by binding to genes, transcripts, proteins, metabolites, enzymes, histones, chaperones, and other molecular targets to exert their therapeutic effects. Small molecule drugs typically exert their effects by binding to one or several protein targets, 41 and the targets are usually directly or indirectly involved in the pathogenesis of the disease. We utilized the network distance between the targets of a drug and the AD-related genes to measure the relationship between the drug and AD, by which we were able to screen drug candidates for AD treatment. Of the 714 AD-related genes, 593 could be mapped to the human PPIN. For both drugs and hypothetical drugs, we calculated the distribution of network distance density profiles. At a distance of 0.75, the drug and potential drug density profiles dropped drastically, which was used as a threshold to screen potential drugs (Fig. 2). Eventually, 1044 drugs with a distance range of [–4.71, 0.75] were selected as potential AD drugs (Supplementary Table 3). In this list, six FDA-approved AD drugs, including aducanumab, lecanemab, donepezil, memantine, rivastigmine, and tacrine, were included, indicating a high reliability of the method. A negative network distance suggested that the drug had one or more targets that were also AD-related genes, and these particular targets were given the negative weight. A more negative weighting calculation result further suggested that the drug was associated with the AD disease modules more tightly.

Fig. 2

Association between drugs and AD risk genes. Network distance density plots show the distribution of distances to AD risk genes for all drugs. The blue represents the hypothetical drugs, while the red represents the drugs that was taken from the DrugBank database.

Molecular docking study

For the 1,044 drug candidates that were close to the AD-related genes in PPIN, 836 had valid molecular structures, which were further screened by examining their affinities with Aβ, the key molecule involved in the pathogenesis of AD (Supplementary Table 4). Ergotamine, RAF-265, and tucatinib were the AD drug candidates with the highest likelihood of targeting Aβ (Fig. 3; Table 1). Ergotamine had the highest affinity with Aβ, with a docking score of –7.3 kcal/mol. It could form hydrogen bonds with residues GLN15 and LYS16 of Aβ (length = 2.3ringA, length = 2.5ringA) and used residue PHE19 to form hydrophobic interactions. Tucatinib formed one hydrogen bond with residue GLN15 (length = 2.5ringA) and had a docking score of –7.2 kcal/mol. The hydrophobic contacts occurred through residue LEU17. With a docking score of –7.2 kcal/mol, RAF-265 established two hydrogen bonds with residue GLY9 and HIS13 (length = 2.4ringA, length = 2.4ringA, length = 2.7ringA). Additionally, the FDA-approved AD drugs displayed higher binding energies, including rivastigmine (–4.2 kcal/mol), tacrine (–5.4 kcal/mol), memantine (–4.4 kcal/mol), and donepezil (–5.7 kcal/mol). Consequently, in comparison to ergotamine, these drugs have comparatively low affinity for Aβ. In order to investigate the stability of the interaction of ergotamine, RAF-265, and tucatinib with Aβ, we further performed molecular dynamics simulations.

Table 1

Molecular docking results of top 3 potential drugs

Compound CID	Drug name	Distance	Binding energy (kcal/mol)	Binding site
8223	Ergotamine	0.53	–7.3	ASP7, GLN15, LYS16, PHE19
51039094	Tucatinib	–0.78	–7.2	VAL12, GLN15, LEU17
11656518	RAF-265	–0.74	–7.2	PHE4, GLY9, TYR10, HIS13, HIS14

Fig. 3

Molecular docking results of drugs. Black dashed lines represent hydrogen bonds, numerical values (ringA) denote hydrogen bond lengths, and blue amino acid residues denote binding sites. The drugs are (A) ergotamine, (B) RAF-265, and (C) tucatinib, respectively.

In addition, we expanded the AD-related gene set by including the potential drug targets for AD in Agora. From Agora, 947 nascent drug targets for AD nominated by AD researchers were retrieved and merged with the genes collected by us. After removing duplicated ones, we obtained a list of 1,550 genes, of which 1,390 could be mapped to the human PPIN. Based on these genes, 1,672 potential drugs for AD were screened. Of these, 1,040 were among the 1,044 drug candidates identified based on the 714 AD-related genes collected from KEGG, OMIM, and PheGenI. When the 632 (1672–1040) novel drug candidates were examined for their affinities with Aβ, all of them had binding energy higher than –7.2 kcal/mol. In other words, ergotamine, RAF-265 and tucatinib would still be identified with the inclusion of AD risk genes in Agora.

Molecular dynamics simulation study

The structural stability of the protein-ligand complex system was evaluated using molecular dynamics simulations, and the optimal drug molecule’s mechanism of binding to the protein was further examined. During the simulation of 90 ns, ergotamine was found to be the most stable drug molecule. The protein’s RMSD values in the ergotamine-Aβ system also indicated a stable binding mode (Fig. 4A). During 50–80 ns, the proteins fluctuated by less than 0.2 nm, suggesting that ergotamine may attach to Aβ in a stable and well-bound form. Conversely, in the RAF-265-Aβ system, the proteins’ RMSD values showed a notable rise in fluctuations within the first 40 ns, suggesting that the system was not stable enough within this period frame. However, the system reached a stable state at 45–70 ns, while the protein’s RMSD value fluctuations converged at 1.25 nm. In contrast, the tucatinib-Aβ complex’s RMSD fluctuated obviously between 40 and 60 ns and then became stabilized. According to these results, the ergotamine-Aβ complex appears to be more resilient than the other systems.

Fig. 4

Ergotamine, RAF-265 and tucatinib with Aβ complex system’s results for RMSD, Rg and RMSF from 90ns MD simulation. (A) RMSD values of backbone atoms of the Aβ; (B) RMSD values of backbone atoms of the drug molecules; (C) Radius of gyration plot of the drug molecule and Aβ complex system; and (D) RMSF of the drug molecule and Aβ complex system.

The RMSF analysis on protein residue fluctuations in protein-ligand complex systems indicated changes in protein stability. Ergotamine exhibited a higher binding affinity than the other two systems. Additionally, all three systems displayed a similar fluctuation pattern. Ergotamine and RAF-265, on the other hand, docked to the Aβ peptide’s aggregation-driving segment, Y10-A21, with less fluctuation, and their bindings to Aβ decreased the protein’s ability to aggregate. The three drugs’ average RMSF values were 0.50 nm, 0.49 nm, and 0.59 nm (Fig. 4D), respectively, indicating their roles in Aβ inhibition. Among them, ergotamine interacted with Aβ at residues GLN15 and LYS16, with RMSF values of 0.33 nm and 0.46 nm, respectively. Where RAF-265 interacted with Aβ to form hydrogen bonds at residues GLY9 and HIS13, the RMSF values was 0.42 nm and 0.46 nm, respectively. Tucatinib’s RMSF value with residue GLN15 was 0.40 nm. Thus, binding of ergotamine with Aβ was more stable as it had the minimal protein residue fluctuation during contact with Aβ.

Furthermore, Radius of Gyration (Rg) was used to measure the compactness of the protein-ligand complexes formed by the three drugs and Aβ. The mean value of Rg for ergotamine, RAF-265, and tucatinib was determined to be 1.15 nm, 0.99 nm, and 1.04 nm, respectively (Fig. 4C), indicating the complex formed by ergotamine and Aβ was more solid than the other docking systems.

The quantity of hydrogen bonds was also examined in order to have a deeper understanding of the molecular stability. Evaluating the strength and stability of drug-protein interactions was aided by the estimation of hydrogen bond numbers. The ergotamine-Aβ system formed up to 6 hydrogen bonds and remained stable during the simulation period (Fig. 5A). Ergotamine exhibited the highest structural stability of the complex compared to the other drugs (Fig. 5A). During the period of simulation, the RAF-265-Aβ system had the maximum of 3 hydrogen bonds, with 2 hydrogen bonds formed mainly within 40–90 ns (Fig. 5B). The drug-protein binding strength was lower than that of the ergotamine-Aβ system. On the other hand, the tucatinib-Aβ system generated 1–2 hydrogen bonds (Fig. 5C), demonstrating that tucatinib’s interaction with Aβ was the weakest, in line with the outcomes of molecular docking. In a molecular complex formed by a drug and protein, the formation of multiple hydrogen bonds enhances the intermolecular interactions between the two molecules, thereby contributing to a more structurally stable molecular complex. During the period of simulation, ergotamine formed 2–6 hydrogen bonds with Aβ demonstrating stronger structural stability compared to other drugs that only formed 1–3 hydrogen bonds.

Fig. 5

Results of molecular dynamics simulation on the number of hydrogen bonds in the three AD potential drugs and Aβ complex system. The drugs are (A) ergotamine, (B) RAF-265, and (C) tucatinib, respectively. Y-axis value represents the number of hydrogen bonds, and X-axis indicates the time.

DISCUSSION

Among the brain diseases emerging in the world, AD had a higher incidence, 42 and was the most common form of dementia in the elderly. So far, more than 50 million people worldwide have been diagnosed with AD. The main symptoms of AD patients include memory and cognitive dysfunction. 5 It is known that the etiology of AD was complex, involving many factors such as genetics, environment and living habits. 43 In the vast majority of cases, the cause of AD cannot be determined. At present, there are limited treatment options available for patients with AD, and there also are many difficulties in developing drugs for the disease. With the drug repositioning method, we may be able to find new treatment for diseases from available drugs, which can effectively reduce cost of drug development and shorten the development time. 44 Molecular docking has increasing become an import tool for drug repurposing in recent years, and some powerful tools for molecular docking, including GOLD, FlexX, Molegro Virtual Docker, AutoDock Vina and AutoDockTools, have been developed. 45 In this study, AutoDockTools was adopted to generate the input files and then AutoDock Vina was used for molecular docking. AutoDockTools is a graphical user interface to AutoDock that can quickly and effectively process molecular structures. In addition, AutoDock Vina, a new version of AutoDock, is a widely used tool in high-throughput molecular docking. Its computing speed and effective energy scoring technique can significantly improve the prediction’s accuracy and efficiency of the binding mode. With AutoDockTools and AutoDock Vina, we were able to investigate the interactions between drugs and target in detail.

In this study, we adopted a two-step scheme to screen potential candidates for AD treatment from the known drugs. AD-related genes were collected from multiple resources, and the following functional enrichment analysis revealed the pathogenesis of AD. Biological processes such as regulation of neurotransmitter secretion and astrocyte differentiation were found to be related to AD. In the pathway enrichment analysis, oxidative phosphorylation, apoptosis and neurotrophin signaling pathway were significantly enriched. The results were consistent with previous analyses, 46 indicating the genes collected were closely related to the pathogenesis of AD.

We gathered the drug information from DrugBank and retained 5,833 drugs with targets in human genome. AD genes were not distributed randomly on PPIN and they formed gene modules related to the disease. The drugs whose targets were close to the gene modules of AD on the human PPIN were potential candidates for the disease as their targets were more likely to act on the disease genes than genes far from the gene modules. A network algorithm was used to calculate the network distances between drugs and AD genes in the human PPIN. By this way, drugs could be used for AD treatments were more likely to be found close to the AD gene modules, while those with larger distances may be less likely to be effective for AD. We added weight w for specific targets during the calculation for potential drugs that affected the disease modules using a network distance algorithm. When mapping AD risk genes, the sub-network from human PPIN is referred to as AD PPIN. The node degree of AD risk genes in the AD PPIN was used to calculate w. Drug targets that are mapped to human PPIN occur in two cases. In the first one, the drug target is also an AD risk gene. The target, or AD risk gene, has a greater association with the disease modules. Thus, the special target is assigned a weight, which is calculated via taking the node degree of that risk gene in account. In the second case, where the target has no association to AD risk genes, network distance must be used to calculate the likelihood that the drug could affect the AD disease modules through the target. In this case, the target has a weight value of 0. Thus, the two types of targets are distinguished and possible drugs that could have an effect on the disease modules are acquired via offering the target a weight value. We systematically evaluated the shortest distance of the network between the drug and the AD genes, and setting a threshold of 0.75 based on the actual distance density profile, potential drugs were screened for possible use in the treatment of AD.

The targets of ergotamine and RAF-265 had relatively small distances to AD-related genes on PPIN, with a distance of 0.53 and –0.74, respectively. Molecular docking and molecular dynamics simulation analysis showed that they could bind to Aβ stably. Ergotamine, a drug used to treat acute migraine, stimulates 5-HT receptors, particularly 5-HT1B,47,48, 47,48 which have been found to have a decreased expression in APP-induced models. 49 Research has demonstrated that the prevention of Aβ-induced cytotoxicity is facilitated by 5-HT1B activation. 50 It also means that ergotamine may have therapeutic value in AD. Ergotamine has a strong binding affinity of –12.58 kcal/mol for histamine N-methyl transferase (HNMT) and, through its inhibition of HNMT, contributes to the enhancement of histaminergic neurotransmission in AD, according to the findings of computer simulation research. 51 In a recent study, it is found that ergotamine and several other alkaloids show favorable binding affinity with acetylcholinesterase, and is suggested to be a potential anti-AD compounds. 52 Furthermore, it has been reported that some derivative compounds of ergotamine, exhibit positive neuroprotective action in AD. Dihydroergotamine, for instance, can decrease the synthesis of Aβ by preventing γ-secretase activity. 53 In our study, ergotamine demonstrated a significant stability of the ergotamine-Aβ complex in addition to an excellent binding affinity for Aβ, which may be connected to its capacity to prevent Aβ aggregation. Consequently, an effective drug that may target Aβ is ergotamine. Moreover, ergotamines seem to have side effects that trigger vasospastic, and an acute overdose can have serious ischemic complications, 54 clinical trials still need to be explored to further investigate the clinical therapeutic efficacy for AD patients.

RAF-265 (CHIR-265), a Raf inhibitor, is currently being investigated for the treatment of melanoma.55,56, 55,56 Our findings indicate that RAF-265 has the ability to bind to the Y10-A21 aggregation driver segment of Aβ peptide, suggesting that it can prevent Aβ from aggregating and serve as an effective therapy for AD. Our findings align with those of earlier research. According to a prior study, RAF inhibitors were thought to be effective options for treating AD because of their capacities to reduce memory impairment and suppress APP expression in a mouse model of AD. 57 Previous studies have demonstrated that RAF inhibitors can shield cortical cells from Aβ damage, though the precise mechanism is still unknown. 58 As RAF-265 is an investigational drug, comprehensive testing and assessment of its toxicities are still in progress, and its therapeutic effect on AD needs to be tested in clinical trials.

However, there are some limitations with our work. First, we have not conducted an experimental study to validate the potential efficacy of the drugs we identified in AD treatment, although the available evidence suggest they may be promising candidates. Second, the method we used to screen the drug candidates depends on known AD risk genes, available PPIN, as well as the information of drugs and their targets. As research in the field of AD, human interactome and pharmacology continues, new risk genes of AD may be identified, more PPI data will be accumulated and novel drugs and targets are reported, our work should be also updated. In addition, biological networks are dynamic and continuously changing, while weight values in our method ours are static and is unable to measure such features of the biological systems. In the current study, we only examined the binding affinities of drugs to Aβ. It is likely that some of the novel drugs may play a function in AD treatment by interacting with other targets. Given the rich information included in Agora, it will be a valuable resource for discovering new drugs for AD. In the next step, we will explore how to fully utilize the data in Agora in our study.

In summary, based on a drug repositioning approach, we provided evidence that ergotamine and RAF-265 can be promising drug candidates for treating AD. Our work will not only improve our understanding of the mechanisms the disease, but also help us to explore new targets and drug candidates.

AUTHOR CONTRIBUTIONS

Qiuchen Wang (Investigation; Methodology; Writing – original draft); Mengjie Fu (Data curation; Software; Validation); Lihui Gao (Data curation; Methodology; Software); Xin Yuan (Software; Validation; Visualization); Ju Wang (Conceptualization; Resources; Supervision; Writing – review & editing).

Footnotes

ACKNOWLEDGMENTS

The results published here are in whole or in part based on data obtained from Agora, a platform initially developed by the NIA-funded AMP-AD consortium that shares evidence in support of AD target discovery. Agora is available at: doi: 10.57718/agora-adknowledgeportal.

FUNDING

This study was supported in part by grants from National Key Research and Development Program of China (No.2016YFC0906300).

CONFLICT OF INTEREST

The authors have no conflict of interest to report.

DATA AVAILABILITY

The data supporting the findings of this study are available within the article and/or its supplementary material.

The supplementary material is available in the electronic version of this article: .

References

Lynch

. World Alzheimer Report Attitudes to dementia, a global survey. Alzheimers Dementia 2020; 16: e038255.

, Gao

, Kwok

TCY

, et al. Derivation and validation of the cognitive impairment prediction model in older adults: A national cohort study. Front Aging Neurosci 2022; 14: 755005.

Han

, Hu

, Yu

, et al. Mafenide derivatives inhibit neuroinflammation in Alzheimer’s disease by regulating pyroptosis. J Cell Mol Med 2021; 25: 10534–10542.

Alzheimer’s Association. 2023 Alzheimer’s disease facts and figures. Alzheimers Dement 2023; 19: 1598–1695.

Khan

, Barve

, Kumar

. Recent advancements in pathogenesis, diagnostics and treatment of Alzheimer’s disease. Curr Neuropharmacol 2020; 18: 1106–1125.

Larkin

. Lecanemab gains FDA approval for early Alzheimer disease. JAMA 2023; 329: 363.

van Dyck

, Sabbagh

and Cohen

. Lecanemab in early Alzheimer’s disease. Reply. N Engl J Med 2023; 388: 1631–1632.

Dhillon

. Aducanumab: First approval. Drugs 2021; 81: 1437–1443.

Tagliapietra

. Aducanumab for the treatment of Alzheimer’s disease. Drugs Today (Barc) 2022; 58: 465–477.

10.

Alexander

, Emerson

and Kesselheim

. Evaluation of aducanumab for Alzheimer disease: Scientific evidence and regulatory review involving efficacy, safety, and futility. JAMA 2021; 325: 1717–1718.

11.

Reardon

. FDA approves Alzheimer’s drug lecanemab amid safety concerns. Nature 2023; 613: 227–228.

12.

Sims

, Zimmer

, Evans

, et al. Donanemab in early symptomatic Alzheimer disease: The TRAILBLAZER-ALZ 2 randomized clinical trial. JAMA 2023; 330: 512–527.

13.

Bateman

, Cummings

, Schobel

, et al. Gantenerumab: An anti-amyloid monoclonal antibody with potential disease-modifying effects in early Alzheimer’s disease. Alzheimers Res Ther 2022; 14: 178.

14.

, Visweswaran

and Batmanghelich

. Semi-supervised hierarchical drug embedding in hyperbolic space. J Chem Inf Model 2020; 60: 5647–5657.

15.

Athar

, Al Balushi

, Khan

. Recent advances on drug development and emerging therapeutic agents for Alzheimer’s disease. Mol Biol Rep 2021; 48: 5629–5645.

16.

Andersen

, Schousboe

and Verkhratsky

. Astrocyte energy and neurotransmitter metabolism in Alzheimer’s disease: Integration of the glutamate/GABA-glutamine cycle. Prog Neurobiol 2022; 217: 102331.

17.

Celdran de Castro

, Nascimento

, Beltran-Corbellini

, et al. Levetiracetam, from broad-spectrum use to precision prescription: A narrative review and expert opinion. Seizure 2023; 107: 121–131.

18.

Vossel

, Ranasinghe

, Beagle

, et al. Effect of levetiracetam on cognition in patients with Alzheimer disease with and without epileptiform activity: A randomized clinical trial. JAMA Neurol 2021; 78: 1345–1354.

19.

Al-Horani

. Riluzole and its prodrugs for the treatment of Alzheimer’s disease. Pharm Pat Anal 2023; 12: 79–85.

20.

Abd Elmaaboud

, Estfanous

, Atef

, et al. Dapagliflozin/hesperidin combination mitigates lipopolysaccharide-induced Alzheimer’s disease in rats. Pharmaceuticals (Basel) 2023; 16: 1370.

21.

Kanehisa

, Furumichi

, Sato

, et al. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res 2023; 51: D587–d592.

22.

Amberger

and Hamosh

. Searching Online Mendelian Inheritance in Man (OMIM): A knowledgebase of human genes and genetic phenotypes. Curr Protoc Bioinformatics 2017; 58: 1.2.1–1.2.12.

23.

Ramos

, Hoffman

, Junkins

, et al. Phenotype-Genotype Integrator (PheGenI): Synthesizing genome-wide association study (GWAS) data with existing genomic resources. Eur J Hum Genet 2014; 22: 144–147.

24.

Wishart

, Feunang

, Guo

, et al. DrugBank 5.0: A major update to the DrugBank database for 2018. Nucleic Acids Res 2018; 46: D1074–D1082.

25.

Luck

, Kim

, Lambourne

, et al. A reference map of the human binary protein interactome. Nature 2020; 580: 402–408.

26.

Cowley

, Pinese

, Kassahn

, et al. PINA v2.0: Mining interactome modules. Nucleic Acids Res 2012; 40: D862–D865.

27.

Szklarczyk

, Gable

, Nastou

, et al. The STRING database in Customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Res 2021; 49: D605–D612.

28.

Bindea

, Mlecnik

, Hackl

, et al. ClueGO: A Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 2009; 25: 1091–1093.

29.

Shannon

, Markiel

, Ozier

, et al. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res 2003; 13: 2498–2504.

30.

Peng

, Yuan

, Xin

, et al. Screening novel drug candidates for Alzheimer’s disease by an integrated network and transcriptome analysis. Bioinformatics 2020; 36: 4626–4632.

31.

Kong

, Wang

, et al. Bilirubin induces pain desensitization in cholestasis by activating 5-hydroxytryptamine 3A receptor in spinal cord. Front Cell Dev Biol 2021; 9: 13.

32.

Eberhardt

, Santos-Martins

, Tillack

, et al. AutoDock Vina 1.2.0: New docking methods, expanded force field, and python bindings. J Chem Inf Model 2021; 61: 3891–3898.

33.

Morris

, Huey

, Lindstrom

, et al. AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem 2009; 30: 2785–2791.

34.

Burley

, Berman

, Kleywegt

, et al. Protein Data Bank (PDB): The single global macromolecular structure archive. Methods Mol Biol 2017; 1607: 627–641.

35.

Kim

, Chen

, Cheng

, et al. PubChem in 2021: New data content and improved web interfaces. Nucleic Acids Res 2021; 49: D1388–D1395.

36.

O’Boyle

, Banck

, James

, et al. Open Babel: An open chemical toolbox. J Cheminform 2011; 3: 33.

37.

Zhou

, Wen

, Lei

, et al. Molecular dynamics study of conformation transition from helix to sheet of Abeta42 peptide. J Mol Graph Model 2021; 109: 108027.

38.

, Xu

, Li

, et al. Application of molecular dynamics simulation in biomedicine. Chem Biol Drug Des 2022; 99: 789–800.

39.

Páll

, Zhmurov

, Bauer

, et al. Heterogeneous parallelization and acceleration of molecular dynamics simulations in GROMACS. J Chem Phys 2020; 153: 134110.

40.

Sousa da Silva

and Vranken

. ACPYPE – AnteChamber PYthon Parser interfacE. BMC Res Notes 2012; 5: 367.

41.

Dubach

, Vinegoni

, Mazitschek

, et al. In vivo imaging of specific drug-target binding at subcellular resolution. Nat Commun 2014; 5: 3946.

42.

Wang

, Shao

, Hao

, et al. Identify connectome between genotypes and brain network phenotypes via deep self-reconstruction sparse canonical correlation analysis. Bioinformatics 2022; 38: 2323–2332.

43.

Coleman

and Mastroeni

. A call for new approaches to Alzheimer’s disease research. Neurobiol Aging 2017; 57: iii–iv.

44.

Pushpakom

, Iorio

, Eyers

, et al. Drug repurposing: Progress, challenges and recommendations. Nat Rev Drug Discov 2019; 18: 41–58.

45.

Torres

PHM

, Sodero

ACR

, Jofily

, et al. Key topics in molecular docking for drug design. Int J Mol Sci 2019; 20: 4574.

46.

, Xin

, Hu

, et al. Analyzing the genes related to Alzheimer’s disease via a network and pathway-based approach. Alzheimers Res Ther 2017; 9: 29.

47.

Jacob

, Braekow

, Schwarz

, et al. Ergotamine stimulates human 5-HT(4)-serotonin receptors and human H(2)-histamine receptors in the heart. Int J Mol Sci 2023; 24: 4749.

48.

Dahlof

, Maassen Van Den Brink

. Dihydroergotamine, ergotamine, methysergide and sumatriptan – basic science in relation to migraine treatment. Headache 2012; 52: 707–714.

49.

Tajeddinn

, Persson

, Maioli

, et al. 5-HT1B and other related serotonergic proteins are altered in APPswe mutation. Neurosci Lett 2015; 594: 137–143.

50.

Yang

, Zhang

, Yu

, et al. A novel 5-HT(1B) receptor agonist of herbal compounds and one of the therapeutic uses for Alzheimer’s disease. Front Pharmacol 2021; 12: 735876.

51.

Jiménez

, Żołek

, Hernández Perez

, et al. Drug repurposing to inhibit histamine N-methyl transferase. . Molecules 2023; 28: 576.

52.

Singh

, Rani

, Kaur

, et al. Exploring the therapeutic potential of alkaloids in Alzheimer’s disease management. Cent Nerv Syst Agents Med Chem 2024; 24: 206–218.

53.

Lei

, Yu

, Niu

, et al. The FDA-approved natural product dihydroergocristine reduces the production of the Alzheimer’s disease amyloid-beta peptides. Sci Rep 2015; 5: 16541.

54.

Roberto

, Raschi

, Piccinni

, et al. Adverse cardiovascular events associated with triptans and ergotamines for treatment of migraine: Systematic review of observational studies. Cephalalgia 2015; 35: 118–131.

55.

James

, Ruggeri

, Armstrong

, et al. CEP- A novel orally active BRAF(V600E) inhibitor with selective cellular and in vivo antitumor activity. Mol Cancer Ther 2012; 11: 930–941.

56.

Grimaldi

, Cassidy

, Leachmann

, et al. Novel approaches in melanoma prevention and therapy. Cancer Treat Res 2014; 159: 443–455.

57.

Burgess

and Echeverria

. Raf inhibitors as therapeutic agents against neurodegenerative diseases. CNS Neurol Disord Drug Targets 2010; 9: 120–127.

58.

Echeverria

, Burgess

, Gamble-George

, et al. Raf inhibition protects cortical cells against beta-amyloid toxicity. Neurosci Lett 2008; 444: 92–96.