Fragment-Based AI Optimization of Tranexamic Acid Derivatives Targeting the Kringle 1 Lysine-Binding Site of Human Plasminogen: An Integrated Computational Study for Postpartum Hemorrhage
Postpartum hemorrhage remains a leading cause of maternal mortality worldwide, frequently associated with excessive fibrinolysis mediated by plasminogen activation. Tranexamic acid (TXA) is widely used as an antifibrinolytic agent; however, optimization of its scaffold may improve binding characteristics and pharmacological properties. In this study, an integrated computational workflow combining AI-guided fragment-based design with structure-based drug design was employed to generate and evaluate tranexamic acid derivatives targeting the lysine-binding site of the human plasminogen kringle 1 domain (PDB ID: 1CEA), which is the structural site of action of lysine analogue antifibrinolytics. A total of twenty derivatives were generated and assessed using molecular docking, molecular dynamics simulation (500 ns), MM/GBSA binding free energy estimation, pharmacophore mapping, density functional theory (DFT), and ADMET prediction. Among the screened compounds, the lead derivative (R4) showed marginally more favorable predicted docking scores (−7.0 kcal/mol in PyRx and −6.8 kcal/mol in AutoDock Vina Extended) than TXA (−6.6 and −6.1 kcal/mol). Molecular dynamics simulation of the R4 complex indicated a persistent binding pose, with backbone RMSD stabilizing around 0.50 nm and ligand RMSD maintained near 0.06 nm throughout the trajectory. MM/GBSA analysis returned a small negative effective binding energy (ΔG_total = −3.1202 kcal/mol). DFT analysis provided complementary insight into electronic stability and charge distribution, while ADMET profiling indicated acceptable drug-likeness, although R4 was predicted to be bloodbrain barrier permeant, unlike TXA. Structurally, R4 retains the amine of TXA but not its carboxylate and therefore represents a partial mimetic of the parent scaffold. Simulation and binding energy analysis were performed for R4 only, so no comparative claim regarding dynamic stability is made. All results are computational predictions and require experimental validation.
Authors
- Sinan KARAGECILI (ORCID: https://orcid.org/0009-0006-8565-0257)
- Nouman Safdar Ali (ORCID: https://orcid.org/0009-0001-1979-7166)
- Deniz Inan
Institutions
- Twitter (United States) (US)
Publication Details
- Journal
- Journal of Computational Biophysics and Chemistry
- Published
- 2026-09-25
- DOI
- https://doi.org/10.1142/s2737416526501267
- Primary Topic
- Protease and Inhibitor Mechanisms
- Type
- article
- Field-Weighted Citation Impact
- 0.00