Abstract
Purpose :
As the second leading cause of blindness worldwide, there is a growing need for improved treatments. With only three therapies commercially available that target it, rho-associated protein kinase (ROCK) represents a key target for glaucoma drug discovery efforts. This study aims to identify lead-like candidate ROCK inhibitors using computational methods. Using netarsudil and its active metabolite (AR-13503) as benchmarks, the goal is to identify compounds that represent improvements over existing therapies.
Methods :
The project began with the identification of a set of ≈ 300 lead-like hit ROCK inhibitors from literature. These programs were put into the program AutoGrow4, which uses a genetic algorithm to evolve predicted ligands that are similar to the lead-like hits, but with improved docking scores. The compounds were subjected to filters for Dr. Hovhannes Gukasyan’s ophthalmic “rules of thumb”. Six short runs of the algorithm were performed, producing ≈5,000 novel compounds per generation. The top 1,500 scoring compounds and their Vina scores (mean -8.290 kcal/mol) were obtained for analysis. These compounds were taken into ADMET Predictor to calculate clogD (mean 2.922), TPSA (mean 66.196 Å^2), ΔGo/w (mean -18.909 kJ/mol), solubility (mean 2.944 M), and corneal permeability (mean 135.155 cm/s x 10^7). To determine compounds with optimal potency, the Vina score of AR-13503 (-8.6 kcal/mol) was used as a benchmark. Then, to determine compounds with optimal druggability, the corneal permeability of netarsudil (264.381 cm/s x 10^7) was used as a benchmark.
Results :
Applying these benchmarks for potency and druggability resulted in a group of 124 compounds (8.27% of the top 1,500 generated). Using the Vina scores and molecular properties of these compounds, five graphs were created to identify the optimal design space for ROCK inhibitors. To create the five graphs, docking score and each of the four molecular properties were plotted against corneal permeability.
Conclusions :
Analysis of the five graphs revealed five compounds as potential candidates. These compounds were deemed most optimal in terms of potency (mean Vina score = -9.1) and druggability (mean corneal permeability = 280.144). The next phase of this project will involve building an OCAT model of netarsudil in GastroPlus 9.8.3 to predict the ocular pharmacokinetics of the candidate-like leads.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.