Bennett acceptance ratio method to calculate the binding free energy of BACE1 inhibitors: Theoretical model and design of new ligands of the enzyme

• Published in: Chemical biology & drug design 2019-06, Vol.93 (6), p.1117-1128
• Main Authors: Gutiérrez, Margarita, Vallejos, Gabriel A., Cortés, Magdalena P., Bustos, Carlos
• Format: Article
• Language: English
• Subjects: Alzheimer's disease; Amyloid Precursor Protein Secretases - antagonists & inhibitors; Amyloid Precursor Protein Secretases - metabolism; Aspartic Acid Endopeptidases - antagonists & inhibitors; Aspartic Acid Endopeptidases - metabolism; BACE1 inhibitors; Bennett acceptance ratio; binding free energy; Drug Design; drugs design; Ligands; Models, Theoretical; Molecular Docking Simulation; Molecular Dynamics Simulation; Protease Inhibitors - metabolism; Protease Inhibitors - pharmacology; Reproducibility of Results
• ISSN: 1747-0277; 1747-0285
• DOI: 10.1111/cbdd.13456

In recent years, the design, development, and evaluation of several inhibitors of the BACE1 enzyme, as part of Alzheimer's treatment, have gathered the scientific community's interest. Here, a linear regression model was built using binding free energy calculations through the Bennett acceptance ratio method for 20 known inhibitors of the BACE1 enzyme, with a Pearson coefficient of R = 0.88 and R2 = 0.78. The validation of this model was verified employing eight additional random inhibitors, which also gave a linear correlation with R = 0.97 and R2 = 0.93. Furthermore, this linear regression model was also used for proposing the structure of four potential BACE1 inhibitors, and the most active of them gave a theoretical Kd = 10 nM. However, these molecules have not been synthesized yet. Our team used a total time of more than 800 ns for the Molecular Dynamics to carry out this study, and all the software used were freely available. In this work, the FEP‐BAR method was used to obtain an alternative linear regression model with 20 known BACE1 inhibitors. Statistical results (R = 0.88 and R2 = 0.78) proved that this model might be used as a predictive tool of new BACE1 inhibitors. The following figure shows the most active compound designed, which exhibit a KD = 10 nM. Besides, only freely available software packages were used, and the total Molecular Dynamics time was about 800 ns.