The truth wins out in the end. Time and again we have fallen in love with spectacular molecular techniques of drug discovery, but Mother Nature is about the big picture–biology–which might as well be dark matter for how well we understand and control it.
And so it is with drug discovery. Ever since that mold flew in Alexander Fleming’s window, phenotypic assays have been helping us find drugs. Indeed, most drugs in use today were discovered based on what they did in some cell, animal or even human, rather than the way most of us in the business think about drug discovery–namely, find the drug for the protein target first.
So a “phenotype” is the observable biological effect that a drug has on a complex biological entity like a cell, organoid, animal or human. This is opposed to biochemical or biophysical effects, which, though they can be complex, are not nearly so as intact biological entities. These latter effects are usually in relation to an isolated pure protein, so any effect is immediately known to be directly due to the protein’s activity. Not so with phenotypes. When a drug-like chemical compound causes a cell to grow or die or move or something, it's not clear exactly which molecules in that cell the drug is acting on. Hence the reticence to use phenotypes as a way to discover drugs: Not knowing how they work seems like a black box, and pharmaceutical companies do NOT like groping around in the dark.
So the industry spent the last century doing the opposite: mapping those biological effects to individual proteins (drug targets) and then going hog wild on those proteins with biochemical and biophysical screens of millions of drug-like chemicals (“high throughput target based screens”) to find one that affected the activity of the protein drug target. Given the advances in molecular biology, biochemistry and chemistry over the past century, this part was all but assured to yield drug-like chemical compounds that did exactly what our colleages wanted with the protein drug target. But a funny thing happend on the way to FDA approval: Biology lurked back into the picture. Nearly all of these “perfect” drugs failed at animal or human testing.
Eventually, the industry reconsidered its priorities. Maybe taking on the scary black box was more cost-effective than the more reassuring target-based approach. Advances in cell biology, such as organoids, and animal models, such as zebrafish, solved one problem facing phenotype screens: scaling the screen to high throughout. And so we are now in the age of phenotypic drug discovery.
While that’s the story of the high-throughput phenotypic assay screen, the other part of the equation is the chemical library used in such screens. Overlooked by most is the fact that those libraries were designed for the target-based screens. Should they be optimized for phenotypic screens? How?
Because our historeceptomics technology is ideally suited to enhancing the efficiency of phenotypic screens by integrating both polypharmacology and target expression into compound evaluation, we developed a simple metric to score chemical libraries according to their suitability for phenotypic screens. Plainly put, the ideal drug-like chemicals in the screen are those with the LEAST polypharmacology, so that if they are active against the phenotype the list of possible targets on which they act is short–or even just one! The metric we published is a simple number that tells you how polypharmacologic your library is: lower is better. We will be presenting the work at a conference on phenotypic screens in New York City in June 2020: Be there and enter the future of drug discovery!