Identification and Prediction of Promiscuous Aggregating Inhibitors among Known Drugs

Abstract

Some small molecules, often hits from screening, form aggregates in solution that inhibit many enzymes. In contrast, drugs are thought to act specifically. To investigate this assumption, 50 unrelated drugs were tested for promiscuous inhibition via aggregation. Each drug was tested against three unrelated model enzymes: beta-lactamase, chymotrypsin, and malate dehydrogenase, none of which are considered targets of these drugs. To be judged promiscuous, the drugs had to inhibit all three enzymes, do so in a time-dependent manner, be sensitive to detergent and to enzyme concentration, and form particles detectable by light scattering. Of the 50 drugs tested, 43 were nonpromiscuous by these criteria. Surprisingly, four of the drugs showed promiscuous, aggregation-based inhibition at concentrations below 100 microM: clotrimazole, benzyl benzoate, nicardipine, and delavirdine. Three other drugs also behaved as aggregation-based inhibitors, but only at high concentrations (about 400 microM). To investigate possible structure-activity relationships among promiscuous drugs, five analogues of the antifungal clotrimazole were studied. Three of these, miconazole, econazole, and sulconazole, were promiscuous but the other two, fluconazole and ketoconazole, were not. Using recursive partitioning, these experimental results were used to develop a model for predicting aggregate-based promiscuity. This model correctly classified 94% of 111 compounds-47 aggregators and 64 nonaggregators-that have been studied for this effect. To evaluate the model, it was used to predict the behavior of 75 drugs not previously investigated for aggregation. Several preliminary points emerge. Most drugs are not promiscuous, even at high concentrations. Nevertheless, at high enough concentrations (20-400 microM), some drugs can aggregate and act promiscuously, suggesting that aggregation may be common among small molecules at micromolar concentrations, at least in biochemical buffers.

Machine learning models identify molecules active against the Ebola virus in vitro

“…As stated before in perspectives by us 72 and others 2, 16, 20, 73 , the fact that approved drugs may be repurposed for other diseases should not be viewed as a negative aspect of the small molecules, belying undesirable target promiscuity 74 . Instead, we prefer to reference recently published crystallographic analyses 75 demonstrating that small molecules may bind multiple proteins in different types of binding sites and with distinct conformations to ultimately facilitate molecular repurposing.…”

Activation of Inhibitors by Sortase Triggers Irreversible Modification of the Active Site

“…1B). Structures of hit compounds were examined using custom sub-structure search routines in SARNavigator 1.2 (Tripos, St. Louis, MO) to "filter" reactive and promiscuous inhibitors (23)(24)(25)(26), known mutagens and genotoxics (27,28), and molecules lacking good physicochemical hit/lead characteristics (23,26,29,30). The alert molecules were separated into a salvage set, generating a set of 2,023 compounds (1.5% of initial) that were assigned lead-like activities.…”

Ensemble Methods for Classification in Cheminformatics

“…This data set is composed of 48 aggregators and 63 nonaggregators. 18 Since the training and the validation sets have no molecule in common, we checked the overlap by performing PCA in GC space.…”