Identification of dual inhibitors against BACE1 and GSK-3β for the treatment of Alzheimer’s disease: An in silico-based approach

doi:10.21203/rs.3.rs-2597691/v1

Download PDF

Research Article

Identification of dual inhibitors against BACE1 and GSK-3β for the treatment of Alzheimer’s disease: An in silico-based approach

https://doi.org/10.21203/rs.3.rs-2597691/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Alzheimer’s disease (AD) has no Food and Drug Administration (FDA) approved drugs that can halt neurodegeneration. To halt neurodegeneration, targeting the major pathways underlying in AD might be the potential strategy. One such pathways are Amyloid-β (Aβ) and tau pathways which are majorly regulated by β- amyloid precursor protein cleaving enzyme 1 (BACE1) and Glycogen synthase kinase (GSK-3β) respectively. BACE1 is a crucial enzyme involved in Aβ formation & generation and GSK-3β is one of the major kinases responsible for hyperphosphorylation of tau protein. The present study aimed to identify BACE1 and GSK-3β dual inhibitors by an in silico-based approach. Structure based virtual screening has been carried out by using 7,498 Zinc database molecules against BACE1 (PDB ID: 3TPP), and GSK-3β (PDB ID: 1Q3W). 3 hits against BACE1 have identified in Virtual screening with docking score ranging from -8.03 to -12.07 & glide energy ranging from -50.92 to -53.88 kcal/mol and 3 hits against GSK-3β with a docking score ranging from -10.59 to -10.85 & glide energy ranging from -40.5 to -49.7 kcal/mol. Based on amino acid interactions, docking score, and MMGBSA binding energy with BACE1 and GSK-3β active sites, hit molecule ZINC000034853956 with a docking score of -8.03 & -10.59 and glide energy of -53.27 & -49.7 Kcal/mol was selected. MMGBSA binding energy of ZINC000034853956 to BACE1 is -88.41 kcal/mol and to GSK-3β is -15.46 kcal/mol and found to have better affinity than other hit molecules. Further, Molecular dynamics (MD) simulation was performed for the same molecule and MD analysis revealed significant interaction with both the proteins. The present study successfully identified a hit molecule as a BACE1 and GSK-3β dual inhibitor for the treatment of AD.

BACE1

GSK-3β

dual inhibitor

molecular docking

neurodegeneration

Alzheimer’s disease (AD) affects an estimated 0.7 percent of the global population (51.6 million). From 1990 to 2019, the overall number of people affected has more than doubled. In comparison to Asia and Africa, the frequency is higher in high-income regions like Western Europe. However, the total number of people affected in South and East Asia, particularly China, Japan, and India, is substantial. By 2040, dementia-related fatalities are expected to rise from 2.4 million per year to 5.8 million per year (Javaid et al. 2021). During 2020 and 2021, when COVID-19 entered into category of the top ten causes of death, AD is reported to be the seventh-leading cause of death (“2022 Alzheimer’s disease facts and figures” 2022). Neurodegeneration is the main cause for overwhelming health, behavioral changes, socioeconomic problems, and disease progression of the patients suffering with AD (Avila and Hernández 2007). The primary pathological hallmarks responsible for neurodegeneration are senile plaques [SPs-extracellular aggregation of Amyloid-β (Aβ) plaques], and Neurofibrillary tangles (NFT’s-intracellular accumulation of tau protein) (Rampa et al. 2018). Unfortunately, none of the currently available anti-AD drugs can slow or stop neurodegeneration. As a result, there is a need for medications that can slow or stop neurodegeneration, i.e., therapies that target SPs and NFTs. In this context, multi-target directed ligands (MTDLs) capable of targeting Aβ and NFT neurodegenerative pathways could be potential therapeutic agents in the treatment of AD.

β- amyloid precursor protein cleaving enzyme 1 (BACE1) is an intriguing enzyme that cleaves amyloid precursor protein (APP) and responsible for Aβ-formation. It is also an aspartic acid protease which is majorly involved in the formation of myelin sheaths in both central and peripheral neurons (Fig. 1) (Hu et al. 2006, p. 1). GSK-3β is one of the key kinases involved in tau hyperphosphorylation. In addition, GSK-3β is known to be the link between Aβ and tau cascades and also regulates synaptic plasticity, inflammatory pathways, cell survival, axonal transport, cell cycle, neurogenesis, and microtubule dynamics impairment (Fig. 1). Thus, targeting of BACE1 and GSK-3β with MTDLs might offer multiple therapeutic benefits in the treatment of AD. The possible benefits of targeting BACE1 and GSK-3β with MTDLs are reduced Aβ formation & generation, tau hyperphosphorylation, cognitive deficits, and neuroinflammation (Fig. 1).

BACE1 is a type 1 transmembrane aspartic protease consisting of 501 amino acids and is related to the pepsin family. It consists of N-terminal domain (1–21), Pro-peptide (22–45), catalytic domain, cytosolic domain (48–421), and C-terminal domain (445–480) (Cole and Vassar 2008). BACE-1 has a conserved Asp32–Asp228 dyad in its active site (catalytic domain). The Asp dyad is protected by a flap, which is an anti-parallel hairpin loop of 11 residues (Val67-Glu77) (Hong and Tang 2004) (Fig. 2). BACE-1 is thought to be involved in catalysing the cleavage of substrate's peptide bond by acid–base mechanism of hydrolysis with the major involvement of Asp dyad and water molecule. Tyr 71, a flap residue, forms a hydrogen connection with amyloid precursor protein (APP), allowing the flap to move more freely (Fig. 2). This, in turn, allows substrate to enter the active site and ensures proper catalytic site conformation (Barman et al. 2011). The Asp dyad and flap are the important regions for BACE-1 inhibitor design. And other residues such as Asp32, Tyr71, Thr72, Asp228, Gly230, Thr231 were reported to be involved in binding with potent BACE1 inhibitors (Kumar et al. 2019).

GSK-3β has a tiny N lobe (35–138 residues) and a large C lobe (139–386 residues). The catalytic activity of GSK-3β is controlled by three regions: the P-loop (64–69), the activation loop (200–216), and the C-helix loop (96–104) (Buch et al. 2010). GSK-3β is activated by the residues Arg96, Arg180, and Lys205, and its activity is rendered by the residues Glu97 and Lys85 (Fig. 3). His179, Phe201, Met101, and Leu112 are all believed to play a role in activation state. In drug design, hinge domains (133–138) are crucial because they provide a platform for the ligand (Buch et al. 2010). Residues such as Arg141, Thr138, Gly65, Gly68, Lys85, Asp200, Ser 66, and Phe67 were crucial to inhibit the activity of GSK-3β.

Generally, strategy of designing dual inhibitors is under-explored and challenging for BACE1 and GSK-3β. Till now, only a few scaffolds (derivatives of Triazinone, Curcumin, and Naphthofuran) have been reported as dual inhibitors of BACE1 and GSK-3β (Fig. 4). Thus, the present study set the path towards the identification of novel and small molecule dual inhibitors of BACE1 and GSK-3β. In the present study, we have approached Molecular docking, MMGBSA energy calculation, and MD techniques to identify the potential hit molecules from the ZINC database library against BACE1 and GSK-3β.

2.1 In silico screening

Virtual screening workflow of Schrodinger software suite 2022-3 (Maestro, Schrödinger, LLC, New York, NY, 2022-3) was chosen for the screening of Zinc database molecules against the active sites of GSK-3β and BACE1.

2.1.1 Ligand preparation

A total of 7,498 ligands that belongs to categories such as Fda, World not Fda, and Investigational only drugs were downloaded in 2D SDF format. Ligprep module was used for the preparation of downloaded structures. The ligands were desalted, neutralized, and converted to low energy 3D structures, ionization, tautomeric, and possible stereoisomeric states were generated. To minimize the energy of generated 3D structures, OPLS 2005 force field was used.

2.1.2 Protein preparation and receptor grid generation

Protein Preparation Wizard was used for preparing the proteins GSK-3β (PDB ID: 1Q3W) and BACE1 (PDB ID: 3TPP). Parameters such as missing atoms, bond orders, tautomer/ionization states, water orientations, and hydrogen bond networking were modified for pre-process of proteins. For constrained energy minimization, OPLS 3e 2005 force field was used. The prepared protein was used for receptor grid was generation. The size and positions of the grid box were defined using the centroid of the workspace ligands. SAVES v6.0 software was used to generate the report of Ramachandran plots of both proteins BACE1 and GSK-3β by using their PDB ID’s.

2.1.3 Molecular docking

Virtual screening work flow of Glide module of the software was used to carry out molecular docking. Prepared ligands were used as input structures. Then, ligands were prefiltered based on ADME properties by using Lipinski’s rule. The prefiltered ligands were used for docking against the generated receptor grid in three phases i.e., High throughput virtual screening (HTVS) phase, Standard precision (SP) phase, and Extra precision (XP) phase.

2.2 MMGBSA Binding

The binding energy of the hit molecules was generated using Prime module of the software i.e., molecular mechanics-generated Born surface area (MMGBSA).

2.3 Molecular dynamics (MD) simulation

MD was performed to study conformational stability and effective conformations between protein and ligand. Desmond module was used for MD simulation for 50ns and performed for one hit molecule having highest docking score against both proteins. A system was built comprising of TIP4P solvent model with dimensions of 10 Å ×10 Å ×10 Å and the charge was neutralized by adding opposite charge ions. Default settings of Desmond setup were used to minimize the energy of complex systems with 2000 iterations and threshold of 1 kcal/mol/ Å. NPT ensemble system with 1.0 bar pressure and 300 K temperature was used to carry out simulation for 50ns duration. For every 1000ps, the trajectory was recorded and the obtained data was used for calculations of root mean square fluctuation (RMSF) and root mean square deviation (RMSD) of ligand and proteins.

3.1 Molecular docking and MMGBSA binding

Protein structure validation was performed using the generated Ramachandran plots. In BACE1 structure most favourable region, additionally allowed region, and disallowed regions contains 90.3%, 9.4%, and 0% amino acid residues respectively. While in GSK-3β structure most favourable region, additionally allowed region, and disallowed regions contains 87.9%, 11.1%, and 1.0% amino acid residues respectively (Fig. 5). Thus, the protein structures can be reliably used for molecular docking. Further, a total of 7,498 zinc database molecules were used for docking against BACE1 (PDB ID: 3TPP), and GSK-3β (PDB ID: 1Q3W). The first phase of docking; HTVS phase have yielded 261 hits against BACE1, and 462 hits against GSK-3β. In the next phase; SP phase have yielded 24 hits against BACE1 and 51 hits against GSK-3β. The final phase of docking; XP phase have yielded 3 hits against BACE1 with a docking score ranging from − 8.03 to -12.07 & glide energy ranging from − 50.92 to -53.88 kcal/mol (Table 1) and 3 hits against GSK-3β with a docking score ranging from − 10.59 to -10.85 & glide energy ranging from − 40.5 to -49.7 (Table 2). ADME analysis of BACE1 and GSK-3β hits were given in Table 3. Hit molecules show properties in accordance with permitted limits of Jorgensen’s rules of 3 and Lipinski rules of 5. Therefore, above results show the acceptable ADME properties of all hits.

We have analysed the active site residues and important amino acid interactions of BACE1 and GSK-3β with their inhibitors 5HA (PDB ID: 3TPP) and Alsterpaullone (PDB ID: 1Q3W) respectively (Fig. 2&3). In our analysis of docking complexes, hit molecules of BACE1 such as ZINC000000004319 & ZINC000019364229 shows that they lack interaction with Asp228, Asp32, and Tyr71 which are important for inhibitory activity. Whereas, ZINC000034853956 form H-bond (Asp228, Lys224, Thr329, and Gly34), Pi-Pi stacking (Tyr71), and salt bridge (Asp228&Asp32) type interactions with BACE1 (Table 1& Fig. 6&7).

And hit molecules of GSK-3β such as and ZINC000022116608 & ZINC000022116612 revealed that they lack interaction with important hinge domain residue Asp133 which is crucial for inhibitory activity. Whereas, ZINC000034853956 forms H-bond (Asp133, Ile62, Val135), and salt bridge (Lys85) type interactions with GSK-3β (Table 2& Fig. 6&8). Above analysis revealed that ZINC000034853956 share interactions with important amino acid residues crucial for inhibitory activity which is similar to 5HA and Alsterpaullone (inhibitors of BACE1 and GSK-3β). And MMGBSA analysis revealed that the binding energy of 5HA to BACE1 is -41.32 kcal/mol, and Alsterpaullone to GSK-3β is -59.48 kcal/mol. While, ZINC000034853956 shows better binding energy when compared to other hit molecules. Binding energy is found be as ZINC000034853956 to BACE1 is -88.41 kcal/mol and to GSK-3β is -15.46 kcal/mol (Table 1&2).

Based on the above analysis, hit molecule ZINC000034853956 was selected for MD simulation.

Table 1

Molecular docking results of hit molecules against BACE1
ZINC ID	Docking score	Glide energy	Glide evdw	Glide ecoul	Glide emodel	MMGBSA dG bind	Residue interactions
ZINC ID	Docking score	Glide energy	Glide evdw	Glide ecoul	Glide emodel	MMGBSA dG bind	H-bond	Pi-Pi stacking	Salt bridge
ZINC000034853956	-8.03	-53.27	-38.55	-14.72	-80.01	-88.41	Gly34, Asp228, Lys224, Thr329	Asp228, Asp32	Tyr71
ZINC000000004319	-12.06	-53.88	-30.4	-23.4	-77.9	-76.81	Gln73, Arg235, Thr232, Gly230	Arg235	-
ZINC000019364229	-12.07	-50.92	-35.05	-15.87	-66.05	-73.97	Arg235, Thr231, Asp228	Asp228	-
5HA	-6.12	-60.44	-42.36	-18.08	-93.79	-41.32	Asp223, Lys224, Thr72, Tyr198	Trp197	-

Table 2

Molecular docking results of hit molecules against GSK-3β
ZINC ID	Docking score	Glide energy	Glide evdw	Glide ecoul	Glide emodel	MMGBSA dG bind	Residue interactions
ZINC ID	Docking score	Glide energy	Glide evdw	Glide ecoul	Glide emodel	MMGBSA dG bind	H-bond	Pi-Pi stacking	Salt bridge
ZINC000022116608	-10.85	-41.5	-26.8	-14.67	-53.11	17.93	Ile62, Asp200, Val135	Lys85	-
ZINC000022116612	-10.81	-40.5	-24.4	-16.07	-57.8	1.32	Ile62, Asp200, Val135	Lys85	-
ZINC000034853956	-10.59	-49.7	-32.7	-16.96	-68.2	-15.46	Ile62, Asp133, Val135	Lys85	-
Alsterpaullone	-9.9	-51.67	-38.60	-13.06	-78.02	-59.48	Lys85, Asp133, Val135	-	Lys85

Table 3

ADME properties of BACE1 and GSK-3β hits
ZINC ID	Stars	Mol.wt	DHB	AHB	logP	logS	SASA	Volume	PCaco	logBB	Met	HOA	%HOA	Rule of 5	Rule of 3
BACE1 hits
ZINC000034853956	0	368.43	4	7.2	1.86	-2.8	660.1	1185.9	54.3	-1.3	5	3	68.91	0	0
ZINC000000004319	0	328.4	4	5.45	2.88	-3.07	587.3	1110.8	58.31	-1.107	5	3	75.4	0	0
ZINC000019364229	0	388.8	1	7.7	1.16	-2.69	660.2	1209.0	18.65	0.11	6	2	56.5	0	1
GSK-3β hits
ZINC000022116608	0	305.76	4	3.75	1.49	-2.5	522.13	906.8	53.7	-0.63	5	3	66.6	0	0
ZINC000022116612	0	305.76	4	3.75	1.49	-2.4	519.12	905.5	53.7	-0.61	5	3	66.6	0	0
ZINC000034853956	0	368.43	4	7.2	1.85	-2.8	660.9	1185.4	54.14	-1.31	5	3	68.8	0	0

Optimum values: Stars (Number of properties or descriptor values that fall outside the 95% range of similar values for known drugs): 0–5; Mol Wt. (Molecular weight): 130–725; DHB (Hydrogen bond donors): 0–6; AHB (Hydrogen bond acceptors): 2 − 20; logP (Predicted octanol/water partition coefficient): 2.0–6.5; logS (Predicted aqueous solubility): -6.5–0.5;SASA (solvent accessible surface area): 300–1000 A³; Volume: 500–2000 A³; PCaco (Predicted apparent Caco-2 cell permeability in nm/s): <25 poor, > 500 great; logBB (Predicted brain/blood barrier partition coefficient): -3.0–1.2; Met (number of likely metabolic reactions): 1–8; HOA (human oral absorption): 1 low, 2 medium, 3 high; % HOA (% human oral absorption): >80% high, < 25% low; Number of violations of Lipinski’s Rule of 5 (The rules are: Mol. Wt. < 500, logPo/w < 5, DHB ≤ 5, AHB ≤ 10. Compounds that satisfy these rules are considered drug-Like): Maximum 4; Number of violations of Jorgensen’s Rule of 3 (The three rules are: QPlogS > -5.7, QPPCaco > 22 nm/s, # Primary Metabolites < 7. Compounds with fewer, and preferably no violation of these rules are more likely to be orally available: Maximum 3.

3.2 MD simulation

MD simulation of the molecule ZINC000034853956 was carried out to analyse the physical movements of atoms and interactions of the ligand-receptor complexes under physiological conditions. The simulation analysis of ZINC000034853956 and BACE1 complex (Fig. 9) shows ligand RMSD of 2.3 Å for first 15ns, then changed to 3.2 Å and became stable for remaining simulation period. The graph of protein RMSF shows that most of the BACE1 residues remain stable except for residues in the range of 165–175 & 315–325. Analysis of protein-ligand contacts shows that the molecule ZINC000034853956 interacts with almost 10 active site residues of BACE1. Types of interactions includes H-bonding, water bridges, hydrophobic, and ionic interactions. Among the above, H-bond interaction is the most prominent one throughout the simulation period by interacting with Asp32, Ser35, Asp228, Tyr198, and Thr329. Other residues which are involved in hydrophobic interactions and water bridges are Gln12, Leu30, Gly34, Pro70, Tyr71, Thr72, Gln73, Phe108, Trp115, Arg128, Val169, Asp223, Lys224, Ile226, Gly330, and Val332 (Fig. 9).

The simulation analysis of GSK-3β/ ZINC000034853956 complex (Fig. 10) shows a ligand RMSD of 1.2Å for first 5ns, then changes to 3 Å and then to 3.5Å and becomes stable for remaining simulation period. RMSD also shows that the ligand and protein are constantly and stably interacting with each other throughout the 50ns. The protein RMSF graph shows that most of the GSK-3β residues were not stable. Analysis of protein-ligand contacts shows that the molecule ZINC000034853956 interacts with almost 8 active site residues of GSK-3β. Types of interactions includes H-bonding, water bridges, hydrophobic, and ionic interactions. Among the above, H-bond interaction is the most prominent one throughout the period of 50ns by interacting with Ile62, Val135, Pro136, Glu137, Arg141, and Asp200. Other residues which are involved in hydrophobic interactions and water bridges but to a lesser extent are Asn64, Phe67, Val70, Lys85, Val87, Asp133, Tyr134, Pro136, Gln185, Asn186, and Leu188 (Fig. 10).

Designing drugs that halt neurodegeneration of AD is challenging due to its multifactorial nature. Targeting crucial pathways such as Aβ and NFT pathways involved in neurodegeneration might be the key strategy to design drugs. BACE1 and GSK-3β are one such proteins that are majorly involved in Aβ and NFT pathways. Inhibiting both proteins might offer multiple therapeutic benefits such as reduced Aβ formation & generation, tau hyperphosphorylation, cognitive deficits, neuroinflammation, and ultimately halts neurodegeneration. Till date, no marketed drugs are available that target both pathways simultaneously. In the present study, an attempt is made to identify potential small molecule dual inhibitors that inhibit both targets (BACE1 and GSK-3β). Structure based virtual screening, and MD based in silico study has identified a novel and potential hit molecule from the ZINC database. Further, in vitro and in vivo studies needs to be carried out to identify exact molecular mechanism and therapeutic benefits in the treatment of AD.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Magham Sai Varshini, Dr. Praveen T.K., Dr. Diwakar Selvaraj, Aishwarya Reddy. The first draft of the manuscript was written by Magham Sai Varshini and Supervision, Review, and editing was done by Dr. Praveen T.K. Data analysis was done by Magham Sai Varshini, Aishwarya Reddy, and Dr. Diwakar Selvaraj. All authors read and approved the final manuscript.

Data Availability

All data generated or analysed during this study are included in this article.

Alzheimer’s disease facts and figures (2022).Alzheimer’s & Dementia, 18(4),700–789. https://doi.org/10.1002/alz.12638
Avila, J., & Hernández, F. (2007). GSK-3 inhibitors for Alzheimer’s disease. Expert Review of Neurotherapeutics, 7(11), 1527–1533. https://doi.org/10.1586/14737175.7.11.1527.
Barman, A., Schürer, S., & Prabhakar, R. (2011). Computational Modeling of Substrate Specificity and Catalysis of the β-Secretase (BACE1) Enzyme. Biochemistry, 50(20), 4337–4349. https://doi.org/10.1021/bi200081h.
Buch, I., Fishelovitch, D., London, N., Raveh, B., Wolfson, H. J., & Nussinov, R. (2010). Allosteric Regulation of Glycogen Synthase Kinase 3β: A Theoretical Study. Biochemistry, 49(51), 10890–10901. https://doi.org/10.1021/bi100822q.
Cole, S., & Vassar, R. (2008). BACE1 Structure and Function in Health and Alzheimers Disease. Current Alzheimer Research, 5(2), 100–120. https://doi.org/10.2174/156720508783954758.
Hong, L., & Tang, J. (2004). Flap Position of Free Memapsin 2 (β-Secretase), a Model for Flap Opening in Aspartic Protease Catalysis ^,. Biochemistry, 43(16), 4689–4695. https://doi.org/10.1021/bi0498252.
Hu, X., Hicks, C. W., He, W., Wong, P., Macklin, W. B., Trapp, B. D., & Yan, R. (2006). Bace1 modulates myelination in the central and peripheral nervous system. Nature Neuroscience, 9(12), 1520–1525. https://doi.org/10.1038/nn1797.
Javaid, S. F., Giebel, C., Khan, M. A., & Hashim, M. J. (2021). Epidemiology of Alzheimer’s disease and other dementias: rising global burden and forecasted trends. F1000Research, 10, 425. https://doi.org/10.12688/f1000research.50786.1
Kumar, A., Srivastava, G., Negi, A. S., & Sharma, A. (2019). Docking, molecular dynamics, binding energy-MM-PBSA studies of naphthofuran derivatives to identify potential dual inhibitors against BACE-1 and GSK-3β. Journal of Biomolecular Structure and Dynamics, 37(2), 275–290. https://doi.org/10.1080/07391102.2018.1426043.
Rampa, A., Gobbi, S., Di Martino, C., Belluti, R. M., F., & Bisi, A. (2018). Dual BACE-1/GSK-3β Inhibitors to Combat Alzheimer’s Disease: A Focused Review. Current Topics in Medicinal Chemistry, 17(31), 3361–3369. https://doi.org/10.2174/1568026618666180112161406.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Identification of dual inhibitors against BACE1 and GSK-3β for the treatment of Alzheimer’s disease: An in silico-based approach

Status:

Version 1

Abstract

Figures

1. Introduction

2. Methodology

2.1 In silico screening

2.1.1 Ligand preparation

2.1.2 Protein preparation and receptor grid generation

2.1.3 Molecular docking

2.2 MMGBSA Binding

2.3 Molecular dynamics (MD) simulation

3. Results And Discussion

3.1 Molecular docking and MMGBSA binding

3.2 MD simulation

4. Conclusion

Statements And Declarations

References

Additional Declarations

Status:

Version 1