Three ideas motivate this project. First, bioactive small molecules often act on multiple targets- sometimes this """"""""polypharmacology"""""""" is key to their efficacy, sometimes it is the source of unwanted side-effects, rarely is it entirely absent. Second, these off-targets may be predicted comprehensively across all of pharmacological space. The last motivating idea is that drugs are so rare that when one is found every effort should be made to discover areas where it might be useful. Here we seek new indications for established drugs, focusing in Aim 1 on two specific disease areas, and in Aim 2 on developing a comprehensive map of drug off-target interactions to guide future studies. To do so we will use a chemoinformatic method, the Similarity Ensemble Approach (SEA), which predicts associations among targets based on the ligands that bind to them. In proof-of-concept studies, it has predicted 36 previously unknown """"""""off-targets"""""""" for 23 drugs, with confirmatory experimental affinities ranging from 1.2 to 14,000 nM.
The specific aims are:
Aim 1. To predict and test established drugs that act on targets in the areas of Multiple Sclerosis and cardiovascular disease. a. SEA predicts the molecular target for an investigational Multiple Sclerosis drug-until now, no good molecular target has been identified for this drug. This prediction will be tested experimentally in receptor-binding assays. b. If this drug is active at the receptor at relevant concentrations, we will subsequently screen for and design analogs that further optimize its activity for this target. Preliminary to second stage animal studies (to be conducted in a follow up project), we also will identify known receptor ligands that are dissimilar to the drug but that, because they bind to the same target, are expected to phenocopy its effect, establishing the relevance of this receptor for MS. c. In a pure drug repurposing effort, SEA suggests a novel receptor for an existing drug, and this receptor is a target for cardiovascular disease. This will be tested experimentally in ligand displacement assays. d. If confirmed, we will investigate optimization of affinity against this """"""""off-target"""""""" receptor by testing drug analogs.
Aim 2. To identify all addressable, high-likelihood off-targets for all FDA and worldwide drugs. The Similarity Ensemble Approach is model-free, and does not use particular structural or pharmacophore models, rather comparing all the chemical information in a drug or class of drugs against the same information in a set of ligands that have been established for a given target. SEA may thus be applied systematically and comprehensively, querying all target classes with all drugs and investigational drugs. We will therefore predict all likely off-targets for all 3665 FDA, worldwide and investigational drugs, across all targets for which ligands are known. a. To do so as comprehensively as possible, we will exploit the StARlite database of ligand-protein interactions;this database doubles the list of drug targets to 2,100 and doubles the number of ligands annotated for them to 455,000. This will provide a comprehensive view of drug off-target effects. b. We will correlate the new drug-target pairs in this map to known side effects for the drugs for which the new targets are relevant. In proof of concept studies, several of these will be tested in receptor binding assays. c. We also investigate those targets in the map that have been identified as current drug targets, testing several experimentally in proof-of-concept studies. This will be a first step in repurposing these drugs for new indications. Whereas these aims are ambitious, extensive preliminary results support their feasibility. Potent activity for these drugs would, in a longer-term project, be followed by animal efficacy studies for the disease. As these molecules are drugs already, animal efficacy would not long precede human trials.
Bioactive small molecules often act on multiple targets. Whether this is key to their efficacy or the origin of unwanted side effects, it can be predicted comprehensively. In this proposal we focus on drug repurposing in two specific disease areas, and more generally on developing a comprehensive map of drug off-target interactions to guide future studies.
| Irwin, John J; Gaskins, Garrett; Sterling, Teague et al. (2018) Predicted Biological Activity of Purchasable Chemical Space. J Chem Inf Model 58:148-164 |
| Ruderfer, Douglas M; Charney, Alexander W; Readhead, Ben et al. (2016) Polygenic overlap between schizophrenia risk and antipsychotic response: a genomic medicine approach. Lancet Psychiatry 3:350-7 |
| Lorberbaum, T; Nasir, M; Keiser, M J et al. (2015) Systems pharmacology augments drug safety surveillance. Clin Pharmacol Ther 97:151-8 |
| Lemieux, George A; Keiser, Michael J; Sassano, Maria F et al. (2013) In silico molecular comparisons of C. elegans and mammalian pharmacology identify distinct targets that regulate feeding. PLoS Biol 11:e1001712 |
| Lounkine, Eugen; Keiser, Michael J; Whitebread, Steven et al. (2012) Large-scale prediction and testing of drug activity on side-effect targets. Nature 486:361-7 |
