Having more choices is a wonderful luxury but can often end up making the decision process tougher. During the past decade, a growing number of novel assay methods have been developed to aid high-throughput screening of large compound libraries against the wealth of potential targets emerging from genomics. Because of the relatively broad applicability of several of these technologies, the choice of method for a specific assay is generally based on cost, speed, signal quality and ease of use — the assumption being that regardless of choice, similar, if not the same, hits should be generated.

But recently Xiang Wu, Matthew Sills and colleagues at Exelixis, Novartis and Pharmacopeia found that different assay technologies in fact identify different hits from a tyrosine kinase assay. One explanation for this variation is that the differences represent artefacts and the common hits functionally relevant compounds.

To explore this, Wu, Sills and another group of collaborators developed a nuclear receptor antagonist assay for the farnesoid X receptor (FXR) using three commonly used assay formats: AlphaScreen, time-resolved fluorescence (TRF) and time-resolved fluorescence resonance energy transfer (TR-FRET). In the Journal of Biomolecular Screening, they again report assay variation in screens against a random selection of 42,240 compounds from the Novartis synthetic compound library. After primary screening of single compounds and a confirmation experiment, 78, 25 and 43 compounds inhibited FXR binding using AlphaScreen, TRF and TR-FRET, respectively. Screening compounds as single entities, rather than mixtures, identified a greater number of hits and produced less false-positives and false-negatives, which shows that any gains from throughput and cost savings in screening mixtures of different compounds could be lost in terms of results output.

All the hits from the three screens were used in dose-response experiments and were evaluated in a cell-based secondary reporter gene assay for functional activity. Of the 128 compounds tested, 35 showed activity in the gene assay, 34, 11 and 15 of which were identified in the dose-response studies by AlphaScreen, TRF and TR-FRET, respectively. Nine of the 35 compounds were positively identified in all three screens; six were identified by AlphaScreen and TR-FRET, two by AlphaScreen and TRF, and 17 with AlphaScreen alone.

So, different and only partially overlapping sets of functional hits can be obtained from screening programmes. The fact that AlphaScreen produced the greatest number of functional antagonists in this assay does not necessarily mean that it will be the optimal method for other targets as there could be technology-specific interference with each screen. To find the best method, the authors suggest two approaches: either to use replicate determinations in the primary screening process, which should hopefully reduce the variability during this part of the process, or to use more than one assay for a specific target, an approach that should become more feasible owing to reduced screening times and cost reductions through miniaturization.