High-throughput screening (HTS) is now a mainstay of drug discovery and usually the starting point for most drug discovery projects. Industry usually has a lot of resources invested in HTS and therefore needs to be aware of false positives and false negatives that could hamper useful results and lead one down an erroneous path.
Among the many factors responsible for false positives in HTS, one of the most startling and important factors recently unearthed is the non-specific and potent inhibition of enzymes by aggregates of molecules occurring under typical assay conditions. These aggregates are large enough to be observed under a microscope and to be detected by dynamic light scattering. The aggregates adsorb enzyme molecules on their surface, and one of the best tests for detecting their presence is to re-run the enzyme assay under high detergent concentration. High detergent concentrations usually break up the aggregates and lead to a loss of potent inhibition. The phenomenon of aggregation-based inhibition was accidentally discovered by Brian Shoichet's group at UCSF and has been comprehensively explored by him and his students in a series of papers throughout the last decade, although much is still to be known about the exact physical nature of these aggregates. The reason why this has become a big deal is because it has been observed in an unusual number of cases, which leads to the suspicion that much effort might have been already expended in drug discovery campaigns in pursuing such false leads.
In a recent paper, Shoichet and Craik's groups at UCSF accidentally discovered aggregate-based inhibition in discovering inhibitors for the enzyme cruzain which is a part of the metabolic machinery of the parasite responsible for Chagas disease. The authors had started with an initial hit from a virtual screening campaign and were engaged in the usual process of modifying the hit based on SAR. The initial tinkering led to a series of oxadiazole inhibitors which exhibited potent inhibition of cruzain.
However, many of these molecules failed to show activity in cell-based assays. Such a discrepancy between enzyme and cell-based assays can be traced back to many reasons including poor permeability. But in this particular case, kinetic measurements hinted at aggregates of the oxadiazoles that were inhibiting the enzyme. At this stage it was also discovered that unlike the initial hits series, the oxadiazole series had been accidentally assayed under low detergent conditions. The molecules also inhibited another intensely studied enzyme in the Shoichet group- AmpC beta-lactamase. The quintessential test for aggregate-based inhibition, namely increasing the concentration of detergent (Triton in this case), also proved positive confirming the phenomenon. Interestingly the initial set of hit molecules also seemed to exhibit this phenomenon but only in case of AmpC lactamase and not in case of cruzain. In case of cruzain, experiments with differing detergent concentrations proved that the initial set of molecules were equally potent under both conditions, while the oxadiazoles lost activity under high detergent conditions, indicating divergent modes of inhibition between the two sets of molecules.
Finally, note that the aggregation-based inhibition would likely have not been discovered if the oxadiazole series had been assayed under the same low detergent condition as the initial hit series. What seemed like similar molecules turned out to behave very differently under different assay conditions. Sometimes mistakes can reward you with unexpected treasures, and similarity needs to be pried out from the eye of the experimenter. Never underestimate the importance of going wrong (of course revealed only in retrospect).
As the authors narrate, the moral of such studies should not be lost on medicinal chemists, who usually interpret high and low potency of related molecules based on local structural features like hydrogen bonding, electrostatics and hydrophobicity. Aggregation-based enzyme inhibition proves that chemists have to look beyond single molecule structural features toward supramolecular features of several molecules that are interacting with each other. Chemists regularly engaged in HTS campaigns might well keep this valuable piece of advice in mind. Scientific enumeration, it seems, has to always be done at several different levels.
Ferreira, R., Bryant, C., Ang, K., McKerrow, J., Shoichet, B., & Renslo, A. (2009). Divergent Modes of Enzyme Inhibition in a Homologous Structure−Activity Series Journal of Medicinal Chemistry DOI: 10.1021/jm9009229