Polypharmacology, Promiscuity and Pains

Polypharmacology, an emerging theme in Drug Discovery, is being increasingly implicated in altering the therapeutic efficacy of drugs. It is broadly accepted that drugs, perhaps often, trigger their therapeutic effects by interacting with multiple biological targets and thus perturbing multiple, interlinked or independent signaling pathways. This ability of a drug or small molecule to interact with multiple biological targets is known as compound promiscuity. 

In this post, I would like to provide my views on the compound promiscuity trends and the literature that sheds light on the same issue.

Drug-target networks suggests that a drug, on average, interacts with ~2 targets but the recent literature stresses that drugs might bind with ~ 2 to 7 targets [ref 1]. To perform a real time analysis, I chose the DrugBank drugs and analyzed the promiscuity profile of drugs categorized into different groups. Promiscuity rates were calculated based on the target annotations in DrugBank, including all the target categories (i.e. proteins, enzymes and transporters). The following plots depict the trend of promiscuity over different drug groups.

 

Plot I. Drug Group versus Rate of Promiscuity

Plot 2. Drug Group versus Percentage of drugs that are active against >= 2 and >5 targets

A recent Nature Comment surprised me as it explains how the academic researchers are deceived by the so-called promiscuous, assay-duping molecules [ref 2]. When the concept of promiscuity is extended from drugs to compound bioactivity data, complications arise due to various factors, incompleteness of data being the most important one. The Nature Comment proposes another factor that could mislead the researchers in identifying true hits. Hits identified in bioassays comprise artefacts, whose activity does not depend on the nature of interaction with a target protein. An artefact might give a false signal in the assay due to its subversive reactivity, which is different from an interaction of true drug. Such molecules are termed pan-assay interference compounds, or PAINS. These compounds are reported to have promising activity against a wide range of targets.

"Most PAINS function as reactive chemicals rather than discriminating drugs.", warn Baell J and Walters MA. A study conducted by Bajorath et al identified 6 compounds that showed activity against about one-third or more proteins they were screened against [ref 3]. Such behavior apparently pollutes the literature which proposes these pseudo-hits as promising lead molecules. To help the researchers further, Baell J et al have published a set of 26 substructures which are indicative of PAINS liability [ref 4]. These substructures can be assembled to use as a filter in screening processes to limit the dupe structures in the set of real hits. See the below figure to say hello to the PAINS.

Figure 1. The PAINS substructures reported by Baell J et al

However, a large prevalence of the PAINS substructures in the DrugBank approved drugs cannot be precluded. In fact, all the categories of drugs in the DrugBank except the Nutraceuticals have a decent number of drugs with the PAINS substructures. The structures in the above image were converted to SMILES using an in-house chemical structure recognition tool and further converted into SMARTS to efficiently match for their occurrence as substructures in the DrugBank drugs. For detailed statistics, see the plot below.

Plot 3. Total DrugBank drugs versus PAINS matching drugs

As examples, see below a couple DrugBank drugs that possess at least one PAINS substructures.

DB00409

DB04216

Bottom line of the post is that academic drug discovery researchers must be more careful in realising that the molecules displaying strong activity against multiple targets might not be the best hits, and that the PAINS hits must be ignored. So, the ideal way to go up through a fruitful way in drug discovery, according to Baell J and Walters MA, is apparently to follow the following three tips of PAINS-proof drug discovery.

• Learn disreputable structures

• Check the literature

• Assess assays

Hope the post was informative. Cheers!

References

1. Jalencas X and Mestres J: On the origins of drug polypharmacology. Med Chem Comm. 2013; 4(1): 80–87
2. Baell J and Walters M: Chemical con artists foil drug discovery. Nature. 2014; 513(7519): 481-3
3. Hu Y and Bajorath J: How promiscuous are pharmaceutically relevant compounds? A data-driven assessment. AAPS J. 2013; 15(1): 104-11.
4. Baell J and Holloway G: New Substructure Filters for Removal of Pan Assay Interference Compounds (PAINS) from Screening Libraries and for Their Exclusion in Bioassays. J Med Chem. 2010; 53(7): 2719-40