1 Introduction

The incidence of Alzheimer’s disease (AD) tends to increase due to population growth, an aging population, and exposure to chemicals during urbanization and industrialization [1,2,3]. AD affects not only patients (e.g., their physiology, psychology, society, and finances), but also their families, caregivers, and society as a whole [4]. Therefore, drugs and improving the quality of life for people with AD were put at the top of the list of strategies to deal with the public health problem of AD.

Although there have been extensive studies on the pathogenetic and pathophysiology of AD (Aβ amyloid accumulation, tau hyperphosphorylation, mitochondrial dysfunction, central cholinergic damage, infection, gut microbes’ abnormal growth, and environmental and genetic factors) [2, 3, 5,6,7], the etiopathology of AD remains poorly understood. Several medicines, like donepezil, rivastigmine, tacrine, galantamine, memantine, and N-methyl-D-aspartate (NMDA) receptor antagonists, have been approved to treat AD, but they can only slow the disease's progression or lessen its symptoms. Furthermore, patients who used these medications reported various side effects (e.g., diarrhea, sickness, insomnia, and hepatotoxicity) [8]. On the other hand, poor drug pharmacokinetics and pharmacodynamics and poor blood–brain barrier (BBB) permeability are also known to be common causes of therapeutic failure [9]. Until now, no effective cures have been available. Due to the limitations of current medicine, identifying or developing new therapies for AD prevention and treatment is necessary and urgent. Natural products exert anti-inflammatory, anti-oxidant, anti-infective, and anti-apoptotic properties, which are promising alternative therapies for AD [10,11,12,13].

Abyssinones are flavonoids that have been prenylated and isolated from the plant Erythrina Abyssinica. These compounds have attracted a substantial amount of attention because Abyssinone-II has antioxidant, cytotoxic, and anticancer properties [14, 15]. In hormone-dependent cancer, Hatti et al. reported the role of Abyssinones as steroidogenesis modulators [16]. A recent study discovered that Abyssinone V-4’ methyl ether, which was extracted from Erythrina droogmansiana, has demonstrated anti-mammary tumor effects in mice and anti-breast cancer activity in MDA-MB-231 cells [17]. In addition, studies have demonstrated that Abyssinone I and II can trigger apoptosis in human cervical cancer cells via the mitochondrial pathway [14]. Though Abyssinone-I is extracted from natural sources, there has been little information on its biological activity except for its antioxidant and cytotoxic properties [14]. Therefore, further research is necessary to determine the effectiveness of these compounds in treating neurodegenerative diseases. Due to the rapid development of existing bioinformatics, publicly available databases offer big data to facilitate drug research. Therefore, integrating the data from these databases with the appropriate bioinformatical techniques can help clarify the biological properties of natural compounds. This study aimed to elucidate the molecular mechanisms of Abyssinone-I against AD.

2 Materials and methods

2.1 Drug likeness assessment

We first assessed the drug likeness of the studied compound. Abyssinone-I (C20H18O4, Fig. 1A) was obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/), and it was then submitted to SwissADME (www.swissadme.ch) to analyze the physicochemical properties, drug likeness, pharmacokinetics, and medicinal chemistry. In this study, drug likeness was evaluated following Lipinski’s rule [18].

Fig. 1
figure 1

Physicochemical properties of Abyssinone-I. A 2D structure of Abyssinone-I; B, C and D physicochemical properties of Abyssinone-I). HIA—Human Intestinal Absorption; BBB- blood–brain barrier; PGP—P-glycoprotein

Full size image

2.2 Target identification

Next, we identify the important target of Abyssinone-I by using Swiss Target Predictor (www.swisstargetprediction.ch) [19]. After that, we analyzed targets involved in the pathogenesis of AD using GeneCard (https://www.genecards.org) [20]. The JVenn tool (http://jvenn.toulouse.inra.fr/app/example.html) was used to classify the overlapping targets. Finally, these shared targets were assessed to categorize the protein class using the Panther classification system (http://www.pantherdb.org/) [21]. In this study, Homo sapiens was selected as a target organism when required.

2.3 Enrichment analyses

We first proceeded by examining the gene–gene interactions. The gene–gene interactions were analyzed using GeneMANIA, a free Cytoscape plug-in (http://geneMANIA.org/plug-in/) [22]. Second, cytoscape software 3.9.1 (http://www.cytoscape.org/) and the CytoscapeClueGO plug-in version 2.5.8 were utilized to identify and visualize the biological processes of the shared targets. In this analysis, the GoBiologicalProcess databases were selected to attain the list of Cytoscape biological process plugins [11, 23,24,25]. After that, the CytoHubba plug-in was applied to obtain the Hub targets based on three key parameters (degree, closeness, and betweenness) [26]. To examine protein–protein interaction enrichment (PPIE), STRING (String version 11.5, available at https://string-db.org) and Metascape (available at https://metascape.org) were utilized. Only physical interactions were looked at in STRING (physical score > 0.132) [22]. In networks with between 3 and 500 proteins, the Molecular Complex Detection (MCODE) approach was performed to narrow down highly related network components [27, 28]. Additionally, ChIP-X Enrichment Analysis Version 3 (CHEA3) was carried out to identify the transcription factors responsible for the shared targets (https://maayanlab.cloud/CHEA3). Using Cytoscape, the integrated regulatory network was visualized using the top ten transcription factors as determined by the mean rank score [26, 29,30,31]. We further analyzed the cellular components and chromosomes related to AD etiology targeted by Abyssinone-I by using WebGestalt (www.webgestalt.org). In these analyses, over-representative analysis, gene ontology, chromosome location, and illumine humanht 12 v3 were selected. A volcano plot and a bar plot were used to visualize the outcomes [32]. Finally, we analyzed the interactions between miRNA and studied targets by using the MicroRNA ENrichment TURned NETwork (MIENTURNET, http://userver.bio.uniroma1.it/apps/mienturnet/) [33].

2.4 Biological activities and metabolisms

In this analysis, the biological activities and metabolisms of Abyssinone-I were investigated using PASS online (Prediction of Activity Spectra for Substances) [34]. We used "probable activity (Pa, probability to be active)" to assess the likelihood of common biological activities of the studied compound. A number of cytochrome (CYP) P450 isoforms' metabolic processes, including those of their substrates, inducers, and inhibitors, were also studied. Pa values are reported as percentages of probability (%) and range from 0.000 to 1.000.

2.5 Molecular docking

The key targets (PPARG, mTOR, EGFR, ESR1, and MAPK1) thought to be responsible for Abyssinone-I's effects against AD were utilized in this investigation's in-silico docking study using Autodock Vina [35]. In the present study, proteins exhibiting significant levels of x-ray crystallization were chosen. Prior to completing the analysis, the proteins were corrected using MOE software (version 2014). The PubChem database (https://www.ncbi.nlm.nih.gov/) was used to retrieve the Abyssinone-I structure, which was then converted to a PDB file format. Using the Avogadro software, the MMFF-94 force field, and the steepest descent algorithm, the structure of Abyssinone-I was then optimized over 5000 iterations using the MMFF-94 force field [36]. To obtain the target structures in PDB format, we accessed the RCSB website at http://www.rcsb.org/pdb. For this analysis, the PPARG structure that showed higher levels of x-ray crystallization was obtained. The key targets were prepared using MOE software (version 2014) prior to performing the analysis [37]. The protein–ligand complexes were cleared of water molecules and crystal ligands prior to docking. Hydrogen atoms and charges were inserted, while default values for other properties were used. The exhaustiveness setting was assigned a value of 8, and the energy range setting was assigned a value of 4. The AUTOGRID method is used to make a three-dimensional grid so that the total binding energy between abyssinone-I and the key target can be calculated. The grid is generated in the shape of a rectangle. The grid is structured with dimensions of 40 units on the x-axis, 40 units on the y-axis, and 40 units on the z-axis. According to the information provided in reference, the distance between each point on the grid is 0.375 Å [26]. Abyssinone-I's binding affinities to amino acids are estimated using the docking score. The visualization of the results was done using the BIOVIA Discovery Studio Visualizer 2016 program.

2.6 Molecular dynamic simulation, principal component analysis, and binding energy calculation

In this study, a 100-ns molecular dynamic simulation (MDS) analysis of PPARG was carried out using the simulation tool Gromacs 5.1.4, which revealed associations and binding affinities with Abyssinone-I. The topology parameter files for the proteins were constructed using the GROMOS96 force field, while the ligand topology file was generated using the PRODRG2 online service [38]. The docked complexes were included in a cubic box with 1.0 periodic boundary conditions using the straightforward point-charge water model. The system was then neutralized by the addition of Couturier ions. The processed systems were subjected to energy reduction utilizing the steepest descent method and conjugate gradient reduction. For steady temperature (up to 300 K) and pressure (1 bar) stability at 5000 ps, respectively, the systems were modified using the NPT and NVT groups. Each component's temperature was controlled using the Parrinello-Rahman barostat. The stability, flexibility, and rigidity and compactness of Abyssinone-I were assessed via four key parameters: the root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration, and hydrogen bond interaction patterns. Principal component analysis (PCA) was performed using the condacolab package in Python version 3.0, Gromacs, and Pymol. PCA was executed on Colab, a cloud-based Jupyter Notebook platform that offers complimentary access to computational resources such as Graphics Processing Units (GPUs) and Tensor Processing Units (TPUs). The g_mmpbsa tool was utilized to measure the binding free energy of protein–ligand complexes. The binding free energy of the complex structure for the last 5 ns was determined using the Molecular Mechanics Poisson-Boltzmann Surface Area (MM-PBSA) method by g_mmpbsa. The potential energy of solvation was calculated, including the electrostatic and van der Waals interactions, the free energy of solvation, and the polar and non-polar solvation energies.

3 Results

3.1 Drug-likeness assessment

We first evaluate Abyssinone-I's druggability. As shown in Fig. 1B, the physicochemical properties of Abyssinone-I reached Lipinski’s rule of five. Abyssinone-I possessed the ideal range for solubility, lipophilicity, size, saturation, polarity, and flexibility (Fig 1C). Furthermore, the studied compound exerted high human gastrointestinal absorption and blood-brain barrier (BBB) permeation (Fig 1D). Therefore, Abyssinone-I can be a promising compound to optimize as an oral treatment for AD.

3.2 Overlapping target evaluation

Figure 2A shows 79 overlapping targets involved in the pathogenesis of AD and related to Abyssione-I. As shown in Fig. 2B, most of these targets (29/79, 37%) were implicated in the processes of protein-modifying enzymes, followed by metabolite interconversion enzymes (22/79, 28%), transmembrane signal receptors (15/79, 19%), and transporters (6/79, 8%). Most of the interactions between these targets were found to be physical interactions and co-expression (Fig. 2C).

Fig. 2
figure 2

Targets identification related to Alzheimer’s disease (AD) and Abyssinone-I. A Venn diagram showing overlapping targets between Abyssinone-I and AD (A). Bar chart showing functions of the shared targets (B). Network interaction between shared targets (C)

Full size image

3.3 Molecular mechanism analysis

We looked at the enrichment analysis for biological processes, cellular components, and PPIs to figure out how Abyssinone-I is involved in the cause of AD through shared targets. The most enriched biological processes involved in Abyssinone-I and AD were “response to amyloid-beta”, “protein autophosphorylation”, “protein tyrosine kinase activity”, “multicellular organismal response to stress”, “dopamine metabolic process”, and “peptidyl-serine modification” (Fig. 3A). The most enriched cellular components were the plasma membrane region, post-synapse, and dendrite (Fig. 3B). PPIE analysis highlighted “dopamine metabolic process”, “regulation of neurotransmitter levels”, and protein phosphorylation-related pathways, implicating that Abyssinone-I may combat AD by regulating these pathways (Fig. 3C and Table S1).

Fig. 3
figure 3

Enrichment analysis. Biological processes (A), cellular components (B), and protein–protein interactions (C) implicated in the etiology of Alzheimer’s disease and targeted by Abyssinone-I

Full size image

We next identify factors related to shared targets, including Hub targets, transcription factors, chromosomes, and miRNAs. As shown in Fig. 4A–C and Figure S1A–C, five Hub targets (PPARG, mTOR, EGFR, ESR1, and MAPK1), three transcription factors (PPARG (regulated 25/79 targets), MEF2B (regulated 17/79 targets), and MYBL2 (regulated 17/79 targets), and two chromosomes (chr9q22.2 and chr12q24.12) were listed as the top 10 most enriched terms.

Fig. 4
figure 4

Key molecular processes identification. Hub genes (A), transcription factors (B), and chromosomes (C) involved in the etiology of Alzheimer’s disease and targeted by Abyssinone-I

Full size image

In terms of miRNAs, hsa-miR-128-3p was found to interact with stemness-related targets four times (Fig. 5A). A network representation of key miRNAs (hsa-miR-128-3p, hsa-miR-124-3p, hsa-miR-16-5p, and hsa-miR-335-5p) and their hub targets was predicted in Fig. 5B. “Cell cycle”, “prolactin signaling pathway”, “PI3K-AKT-mTOR signaling pathway and therapeutic opportunities” and “Alzheimer’s disease”, and “tauopathy” were the important oncology processes targeted by Abyssinone-I.

Fig. 5
figure 5figure 5

The beneficial effects of Abyssinone-I against Alzheimer's disease are mediated via the structural interactions of hub targets, including PPARG (A), mTOR (B), EGRF (C), ESR1 (D), and MAPK1 (E). The functional residues that comprise binding pocket residues are denoted by three-letter amino acid codes, and the different kinds of interactions are denoted by different color schemes. For the donor and acceptor, respectively, in the H-bond, pink and green are displayed

Full size image

3.4 Molecular docking and stability of the protein–ligand complex

We next assessed the binding mode between five hub targets and Abyssinone-I. The binding affinity of these complexes ranged from −8.9 to −8.5 kcal/mol (Fig. 6A–E and Table 1). PPARG was found to have the lowest affinity energy (−8.9 kcal/mol), implying that PPARG plays critical roles in underlying processes combating AD. The binding energy between PPAGR and Abyssinone-I was identified via a Pi-anion interaction with Asp441 and a Pi-alkyl interaction with Arg443.

Fig. 6
figure 6figure 6

The beneficial effects of Abyssinone-I against Alzheimer's disease are mediated via the structural interactions of hub targets, including PPARG (A), mTOR (B), EGRF (C), ESR1 (D), and MAPK1 (E). The functional residues that comprise binding pocket residues are denoted by three-letter amino acid codes, and the different kinds of interactions are denoted by different color schemes. For the donor and acceptor, respectively, in the H-bond, pink and green are displayed

Full size image
Table 1 Binding sites of Abyssinone I with key amino acids and docking scores involved in the its protective effects on Alzheimer’s disease
Full size table

The RMSD graph demonstrates the stability of Abyssinone-I in PPARG. With an estimated backbone RMSD of 0.375 nm, the system and proteins are stable due to the low variation (Fig. 7A). The loop's various amino acid residues can be seen in the RMSF figure. However, these modifications did not significantly affect the average RMSF of 0.3 nm, and Abyssinone-I and PPARG were still able to achieve their conformations (Fig. 7B). In the simulation, 100 ns-long stable 2H-bonds were formed between Abyssinone-I and PPARG (Fig. 7C). Figure 7D demonstrates that PPARG is rigid and compact, and the small but consistent variation in its radius of gyration demonstrates the protein's stable folding. Based on the results obtained, it was predicted that the average supersurface area (SASA) value of the Abyssinone-I and PPARG complex combination would be 240 nm2 (Fig. 7E). It can be inferred from this prediction that the systems being examined did not exhibit any significant alterations in their conformational structure over the course of the simulation.

Fig. 7
figure 7

Simulation of molecular dynamics between Abyssinone-I and PPARG involved in the etiology of Alzheimer’s disease, including (A) root mean square deviation (RMSD), B root mean square fluctuation (RMSF), C number of hydrogen bonds, D Radius of gyration, and E the solvent accessible surface area (SASA)

Full size image

We conducted PCA to predict significant coordinated movements that occur during ligand binding. The initial eigenvectors are recognized as the primary factors that depict the general movements of the protein. We used the initial 100 eigenvectors to determine the highly correlated movements during the final 100 ns of the runs. We computed the eigenvalues by diagonalizing the covariance matrix of atomic fluctuations. Figure 8A displays the structures of the protein–ligand complex, including surface and non-surface. Figure 8B displays the graph of eigenvalues representing the variability in the protein–ligand combination. The analysis of 100 eigenvectors revealed that slight motion variations may not influence the behavior of Abyssinone-I and the PPARG complex. This suggests that the protein–ligand complexes are stable. Figure 8B illustrates that the average variance of this dimension was around 0.53 ∂^2. The atoms located at the protein's N-terminus and C-terminus that are involved in the principal component movement were measured. The atoms located in the middle and at the end exhibit minimal movement, providing supportive evidence for the stability of this intricate structure (Fig. 8C). Figure 8D displays the PCA of the three components. The binding free energy corresponds to the sum of all the non-covalent interactions. We employed the MM-PBSA method to determine the values for Abyssinone-I and the PPARG complex. The last 5 ns of the molecular dynamics’ trajectory were used to figure out the interaction energies. These included Van der Waal's energy, polar solvation energy, electrostatic energy, SASA energy, and binding energy (Table 2). Figure 9A–D shows binding energy, molecular mechanics potential energy, free energies of polarization, and nonpolar energy. The Abyssinone-I-PPARG complex exhibits a binding free energy of -212.451 kJmol-1. The data clearly demonstrate that Abyssinone-I exhibits a substantial binding affinity with PPARG.

Fig. 8
figure 8figure 8

Principle component analysis (PCA) of Abyssinone-I and the PPARG complex. The structures of the protein–ligand complex, including surface and non-surface (A). The eigenvalues display the variability in the protein–ligand combination (B). C Measurement of the atoms involved in the principal component movement. D PCA of three components related to Abyssinone-I and the PPARG complex

Full size image
Table 2 An energy landscape of ligand–protein complex
Full size table
Fig. 9
figure 9

Binding free energy of Abyssinone-I and the PPARG complex, including binding energy (A), molecular mechanics potential energy (B), free energies of polarization (C), and the nonpolar energy (D)

Full size image

3.5 Biological activities and metabolisms

We further analyzed the biological activities and metabolisms of Abyssinone-I. As shown in Fig. 10A, the main potential mechanisms against AD were antioxidative stress, anti-inflammatory, and anti-oxidant activities, membrane integrity agonists, and MAO inhibitors. Furthermore, Abyssinone-I acted as substrates, inhibitors, and inducers for various CYP450 enzymes (Fig. 10B).

Fig. 10
figure 10

Selected biological activities (A) and metabolisms (B) associated with the etiology of Alzheimer’s disease and targeted by Abyssinone-I

Full size image

4 Discussion

This study gave mechanistic insight into the potential effects of Abyssinone-I on AD. Overall, Abyssinone-I exerts various biological properties against AD (e.g., antioxidative stress, anti-inflammatory, anti-oxidant activities, membrane integrity agonists, and MAO inhibitors). Furthermore, Abyssinone-I possesses its therapeutic effects on AD via 79 targets, four miRNAs (hsa-miR-128-3p, hsa-miR-124-3p, hsa-miR-16-5p, and hsa-miR-335-5p), three transcription factors (PPARG, MEF2B, and MYBL2), and two chromosomes (chr9q22.2 and chr12q24.12). These targets were found to be linked with “response to amyloid-beta”, “protein autophosphorylation”, “protein tyrosine kinase activity”, “multicellular organismal response to stress”, “dopamine metabolic process”, and “peptidyl-serine modification”, while miRNAs were related to “cell cycle”, “prolactin signaling pathway”, “PI3K-AKT-mTOR signaling pathway and therapeutic opportunities”, “Alzheimer’s disease”, and “tauopathy”.

4.1 Shared targets

It has been known that one of the main causes of AD is serotonergic synapse dysregulation [39]. This study observed that Abyssinone-I combats AD by regulating 79 shared targets. Most of these targets were protein-modifying enzymes, metabolite interconversion enzymes, transmembrane signal receptors, and transporters. [7] found that these targets were involved in interactions between serotonergic, cholinergic, and neuroactive ligands and receptors. Coinciding with a cohort study of 136 AD patients and 183 healthy controls, these proteins were observed to be both upregulated and downregulated in AD patients compared with their counterparts [40]. These findings suggest that the effects of Abyssinone-I combat AD via modulating serotonergic synapse-related pathways.

On the other hand, 5 hub targets (PPARG, mTOR, EGFR, ESR1, and MAPK1) were identified among 79 overlapped targets. These hub targets were also found to be linked with AD pathogenesis. For example, there are higher levels of the peroxisome proliferator-activated receptor gamma (PPARG) in the brains of people with AD, which regulates lipid, glucose, and energy metabolism [41]. Through molecular docking, it has been shown how Abyssinone-I and PPARG work together on a molecular level. The characterization of the binding of Abyssinone-I to PPARG involved the identification of two distinct types of interactions. These interactions collectively contribute to the compound's potential therapeutic efficacy in the treatment of Alzheimer's disease through its targeted action on PPARG. The first type of interaction, called a Pi-anion interaction, was found between Abyssinone-I and an essential amino acid residue, Asp441, in the binding site of PPARG. For this interaction to happen, the dispersed electron cloud of the aromatic ring system of Abyssinone-I must line up with the negatively charged area of Asp441. The Pi-anion interaction plays a crucial role in enhancing the stability of the ligand-receptor complex, thereby strengthening the binding affinity between Abyssinone-I and PPARG. The particular interaction described plays a crucial role in facilitating the initial identification and secure placement of Abyssinone-I within the binding pocket. The second interaction, which can be classified as a Pi-alkyl interaction, was observed between Abyssinone-I and Arg443, an additional crucial residue located within the binding site of PPARG. The Pi-alkyl interaction is a result of the close proximity between the aromatic portion of Abyssinone-I and the hydrophobic alkyl side chain of Arg443. The present interaction serves to enhance the stability of the ligand–protein complex through its contribution to the overall molecular cohesion and interaction between Abyssinone-I and PPARG. Furthermore, PPARG is a key regulator of glucose and lipid metabolism [42]. It has been linked to neuroinflammation and neurodegeneration, which means it could be a target for treating Alzheimer's disease [41]. By figuring out how Abyssinone-I binds to PPARG, scientists can get a better idea of how it could be used as a drug to change how PPARG works. This has the potential to contribute to the improvement of Alzheimer's disease pathology. In short, Abyssinone-I and PPARG interact molecularly through the Pi-anion's interaction with Asp441 and the Pi-alkyl's interaction with Arg443. These interactions give us an idea of how Abyssinone-I might work to fight Alzheimer's disease by specifically targeting and interacting with PPARG.

During the early AD stages, soluble amyloid-β induced the hyperactivity of the mammalian target of rapamycin (mTOR) and thus reduced autophagy induction. However, autophagosomes fail to integrate with lysosomes during the last stages of AD [43]. In in-vivo investigations, the epidermal growth factor receptor (EGFR) has been found to be a favorable target for treating Aβ-induced memory loss [44]. A meta-analysis of 8288 AD patients and healthy controls observed an association between AD and estrogen receptor 1 (ESR1) [45]. Also, the levels of mitogen-activated protein kinase phosphatase 1 (MAKP-1) were lower in the brain tissue of AD patients and an animal model of AD [46].

4.2 Biological processes

It has been known that tau hyperphosphorylation, amyloid-β accumulation, oxidative stress, and an altered dopaminergic system are risk factors for AD [47, 48]. In the enrichment analysis, “response to amyloid-beta”, “protein autophosphorylation”, “protein tyrosine kinase activity”, “multicellular organismal response to stress”, “dopamine metabolic process”, and “peptidyl-serine modification” were listed as the top six important biological processes. PPI enrichment analysis also indicated that “dopamine metabolic process”, “regulation of neurotransmitter levels”, and protein phosphorylation-related pathways were the core signaling pathways potentially regulated by Abyssinone-I. These findings highlight that dopamine metabolic processes and protein phosphorylation-related pathways are the most important in the processes by which Abyssinone-I combats AD.

4.3 Transcription factors, miRNAs, and chromosomes

Transcription factors and miRNAs play important roles in regulating gene expression. This study found a link between AD, Abyssinone-I, and four miRNAs (hsa-miR-128-3p, hsa-miR-124-3p, hsa-miR-16-5p, and hsa-miR-335-5p). In accordance with this, various studies have reported the association between altered miRNAs and AD [3, 11, 30, 49, 50]. Three transcription factors (PPARG, MEF2B, and MYBL2) regulated the most targets related to AD and Abyssinone-I that were found. Coinciding with previous studies, these transcription factors were also reported to be implicated in the pathophysiology of AD [51,52,53]. This study also found an association between Abyssinone-I, AD, and two key chromosomes (chr9q22.2 and chr12q24.12). It has been known that chromosomes 9 and 12 have been reported to be linked with AD [54, 55]. These findings may partly bring a new understanding of AD and move treatment for AD closer.

4.4 Biological activities, physicochemical properties, and pharmacokinetics

Natural products exert multi-functional properties; therefore, they could become promising compounds as adjuvants or treatments for AD [10, 11, 13]. This analysis found that Abyssinone-I has several important properties that are needed to treat AD, such as antioxidative stress, anti-inflammatory, anti-oxidant activities, membrane integrity agonists, and MAO inhibitors. These biological activities are widely recognized as important factors that can ameliorate AD [10, 11, 13, 49, 56, 57]. These results were supported by the physical and chemical properties of Abyssinone-I, such as the fact that it is well absorbed by the GI tract and can pass through the BBB.

Although Abyssinone-I could be a promising compound for AD treatment, its bioavailability is still poor (a bioavailability score of 0.55). Furthermore, Abyssinone-I is a P-glycoprotein substrate, indicating it may be eliminated from the central nervous system by the P-glycoprotein. Based on its synthesis accessibility score of 3.93, Abyssinone-I is a chemical that should be suggested for synthesis (Figure S2). Rao et al. were able to synthesize Abyssinone I, which worked as an antioxidant and killed cells in living organisms [15].

Cytochrome P450 enzymes, mostly in the liver, play a role in drug metabolism, especially phase I reactions (hydrolysis, oxidation, and reduction) [58]. Abyssinone-I is an inducer or substrate for various cytochrome P450 enzymes (e.g., CYB, CYP2C, CYP3A, CAY1A1, CYP3A4, (Fig. 10B)). However, it is also an inhibitor for CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4, which are important enzymes implicated in the phase I reaction (Figure S2). These enzymes (e.g., CYP1A2, CYP2D6, and CYP3A4) are involved in the metabolism of various approved AD medications, including donepezil, tacrine, and galantamine. Given that AD frequently requires polytherapy (e.g., antiarrhythmics, antidepressants, analgesics, antiemetics, and neuroleptics) for both more effective treatment and the management of comorbidities, Abyssinone-I may cause drug-drug interactions. Given that AD frequently requires polytherapy (antidepressants, neuroleptics, antiarrhythmics, analgesics, and antiemetics) for both more effective treatment and the management of comorbidities, Abyssinone-I may cause drug-drug interactions [10, 13]. Additionally, CYP-related enzymes, including CYP2D6, CYP3A4, and CYP1A2, are involved in the metabolism of various standard AD medications, including tacrine, donepezil, and galantamine [59]. This suggests that additional investigation is needed to determine the most effective pharmaceutical dosage of this compound for the treatment of AD. To be used in the possible treatment of AD, therapeutic optimization is necessary.

4.5 Limitations

Our findings provide insight into the molecular processes behind Abyssinone's beneficial effects on AD. However, the present study has certain limitations. The reliability of the analysis depends on online databases (Swiss Target Predictor and GeneCard) [12, 23, 26, 60]. The sensitivities of the tested subjects, the route and period of exposure, as well as the dose–response relationship of Abyssinone-I exposure, were not analyzed [13, 23, 24, 31, 60]. The metabolism, multi-drug effects, interactions, and binding-specific structures of Abyssinone-I and other compounds were not explored due to a lack of high-quality databases [10, 11, 26, 30, 49]. No experiments were performed to validate our findings. Therefore, the results can only point to important molecular processes against AD that need to be studied further in vivo and in vitro. In conclusion, this study provides a mechanistic understanding of Abyssinone-I's ability to combat AD. To confirm our findings, more experiments will need to be done, with a focus on hub targets and other important molecular processes.