August 1, 2009 (Vol. 29, No. 14)

New Tools Have Applications in Drug Repurposing and Investigation of Off-Target Effects

Drug discovery and development makes success out of failure—it takes hundreds of experiments to build a hypothesis that leads to a successful drug candidate. But all of these failures generate data that, when organized and applied appropriately, can help scientists not just to identify the right drug candidates to advance through development, but also to fail wrong drug candidates earlier. 

In particular, scientists have employed a number of computational techniques in an effort to make critical assessments or predictions about the druggability of target sites and candidate compounds before conducting more costly in vitro or in vivo experiments. These in silico techniques require exhaustive data and sophisticated, well-structured informatics tools to help scientists exploit the data and better understand mechanism of action, ADME performance, and toxicity.

Currently available scientific data facilitates global compound profiling, which provides a broad picture of how a molecule interacts in the pharmacology space and in later-stage ADME studies. These profiles help organizations plan and manage research programs by: alerting scientists to potential off-target interactions or toxicity issues; providing options for repurposing drugs; and enabling more definitive and informed go/no-go decisions.

In silico global compound profiling exploits in vitro and in vivo experimental information to produce a profile of properties for a given compound. It follows the same methodology as experimental profiling, in which a large set of compounds is screened for a particular activity and an active set further pared down (profiled) by specific parameters such as mechanism of action, selectivity, and ADME characteristics.

In silico profiles confer many advantages, including: fewer molecules synthesized, reduced costs; more targeted and less frequent testing and screening, reduced cell and animal use; and predictive, targeted R&D that enables faster experiments and more informed decisions.

The AurSCOPE knowledge databases, the foundation of Aureus’ in silico profiling platform, are organized around ligands, targets, and biological activities. Each element is described and organized into logical hierarchies according to experimental protocols appearing in the literature.

Taken as a whole, the information compiled by Aureus Pharma creates a global pharmacology space encompassing more than 50,000 publications and patents, 650,000 ligands, and 2.3 million biological activities. The AurPROFILER interface enables scientists to navigate this pharmacology space and conduct a variety of searches, including chemical and pharmacophoric similarity searches, to explore and simulate in silico selectivity profiles for targets, cell lines, or drugs/compounds.

AurPROFILER displays results as interactive heat maps to rapidly visualize, navigate, and filter results based on various parameters such as activity, species, bioassay protocols, publication type, and standard target classification hierarchies.

In silico compound profiling allows organizations to gain valuable information in order to organize and prioritize drug discovery and development programs. This tool is frequently used in repurposing drugs and investigating potential off-target effects.

Repurposing Drugs

Given the tremendous costs expended in drug development, organizations cannot afford to abandon investments outright. In silico profiling provides a low-cost opportunity to develop new strategies for failed candidates and currently marketed drugs, making it possible to capitalize on past work and enhance an overall drug portfolio.

Recently scientists at a major pharmaceutical firm used in silico profiling to develop a novel hit series of small molecule somatostatin 5 receptor antagonists based on astemizole. If they had used AurPROFILER, a consensus profile would have been obtained through a similarity search (chemical similarity index range, 0.7–1) using astemizole as the seed structure. AurPROFILER would have found 21 similar molecules in the AurSCOPE knowledge database across 33 targets (10 more targets are associated with astemizole alone).

When this consensus profile was expanded to show the full matrix, scientists would have seen an unexpected cluster of structures acting on somatostatin receptors (Figure 1).

While the consensus profile performed indicated only micromolar activity at SST5, the presence of any activity at all on this target was interesting because somatostatin receptors differ significantly from the other receptor families targeted by the seed structure. This finding was enough to prompt optimization efforts, resulting in the development of a new potent, highly selective ligand for somatostatin receptor.

Figure 1. The matrix profile generated from the search of astemizole-like compounds.

Off-Target Effects

As part of the investigation into potential off-target effects, scientists will often compare profiles of compounds similar to a potential drug candidate in order to identify interactions. In this article, we present an example exploring a hypothetical candidate structure to target the H3 receptor.

To determine other possible targets with which this structure might interact, the researcher needs to perform a chemical similarity search in AurSCOPE knowledge base. Figure 2 shows the matrix of the retrieved chemically similar compounds. Scanning the matrix, one notes that compounds similar to the candidate are indeed active on H3. In addition, two receptors in the sigma family, DAT and 5-HT1A, are also indicated in the matrix as potential off-targets These off-target interactions are all associated with CNS, hallucinogenic, or psychotropic effects.

In silico profiling, in this case, alerted scientists to a potential off-target concern with the candidate compound that can be strategically planned for as the candidate progresses through development. Scientists know in advance what effects to test for and can even use the information to modify the candidate to avoid potential off-target effects.

An array of useful techniques for speeding drug discovery and development decision making lie where chemistry, biology, and informatics converge. The most recent of these techniques is in silico global compound profiling. Armed with profiles that elucidate potential compound activities, mechanism of action, toxicities, or ADME effects, researchers can plan, manage, and conduct more informed and predictive discovery programs. The AurSCOPE GPS and associated AurPROFILER provide a capable datastructure and intuitive interface for exploring this area.

Figure 2. A matrix of chemically similar compounds for a hypothetical candidate structure

François Petitet (francois.petitet@aureus- is CSO at Aureus Pharma. Web:

Previous articleNeed for Novel Approaches to Treating HIV Still Exists
Next articleGene Variation that Increases Urinary Bladder Cancer Risk Discovered