Radioligand binding assays in the drug discovery process: potential pitfalls of high throughput screenings

Radioligand binding assays evaluating directly the ability of a drug to interact with a defined molecular target is part of the drug discovery process. The need for a high throughput rate in screening drugs is actually leading to simplified experimental schemes that increase the probability of false negative results. Special concern involves voltage-gated ion channel drug discovery where a great care is required in designing assays because of frequent multiplicity of (interacting) binding sites. To clearly illustrate this situation, three different assays used in the academic drug discovery program of the authors were selected because they are rich of intrinsic artifacts: (I) (20 mmol/l caffeine almost duplicated [3H]ryanodine binding (89% higher than control) to rat heart microsomes at 0.3 mumol/l free calcium but did not exert any effect when using a high (107 mumol/l) free calcium, as mostly used in ryanodine binding assays; (II) An agonist for the ionotropic glutamate receptor of the kainate type can distinctly affect [3H]kainate binding to chicken cerebellum membranes depending on its concentration: unlabelled kainic acid per se either stimulated about 30% (at 50-100 nmol/l), had no effect (at 200 nmol/l) or even progressively decreased (at 0.3-2 mumol/l) the binding of 5 nmol/l [3H]kainate, emphasizing the risk of using a single concentration for screening a drug; (III) in a classical [3H]flunitrazepam binding assay, the stimulatory effect of a GABA (gamma-aminobutyric acid) agonist was only observed when using extensively washed rat brain synaptosomes (10 mumol/l GABA increased flunitrazepam binding by 90%). On the other hand, the inhibitory effect of a GABA antagonist was only observed when using crude synaptosomes (10 mumol/l bicuculine reduced [3H]flunitrazepam binding by 40%). It can be concluded that carefully designed radioligand assays which can be performed in an academic laboratory are appropriate for screening a small number of drugs, especially if these are potential hits because of their rational design. Therefore, the low throughput rate could be partially balanced by a higher performance when compared to what is done in a robotic high throughput screening where simplification of assay conditions can lead to false negative results.