Modelos mejorados de relación cuantitativa estructura-actividad (QSAR) para predecir la actividad de los ácidos...

There are ∼600,000 malaria-related deaths annually, predominantly due to Plasmodium falciparum infections. Malaria parasite drug resistance is driving the search for new antimalarials with novel modes of action to current options. Histone deacetylases (HDACs) modulate protein lysine acetylation, regulating a range of cellular processes and are prospective new antimalarial drug targets. To accelerate HDAC inhibitor discovery for malaria, we generated quantitative structure-activity relationship (QSAR) classification models B1-B9 using 572 hydroxamic acid-based structures and their paired P. falciparum 50 % growth inhibition (PfIC₅₀) values. Tripartite split models B5-B9 demonstrated high accuracy (AUC 0.953-0.984), similar to first-generation tripartite split models A5-A7 (AUC 0.961-0.976). External validation using 106 quisinostat-derived compounds from published sources demonstrated improved accuracy of B7 and B8 (>67 %) compared to A7 (∼5 %), the best first-generation model. Experimental assessment of the in vitro P. falciparum activity of 60 compounds selected from a virtual screen of >36,000 hydroxamic acid-containing structures from the ZINC20 and ChEMBL libraries demonstrated that model B7 was more accurate (∼81 %) than model B8 (∼68 %) and A7 (∼56 %). Moreover, the ability of these QSAR models to triage hydroxamic acid-based compounds with activity against malaria parasites was demonstrated by the identification of three potent antiplasmodial hits (PfIC₅₀ 8, 14 and 80 nM), two of which were shown to hyperacetylate P. falciparum histone H4 indicating deacetylase inhibition. These data demonstrate the potential of QSAR models to identify HDAC inhibitors for malaria drug discovery.