23 October 2002 21:30 GMT

by Richard Gaughan

San Diego - The results of high-throughput drug screening depend as much on the technology platform as on biochemistry. That's the "alarming conclusion" reached by Sean Wu, a research scientist specializing in new lead discovery at San Francisco-based company Exelixis.

Wu, speaking at the IBC LifeSciences Assay Development Conference here, described a nuclear receptor FXR antagonist assay developed using three technologies - ALPHAScreen (amplified luminescent proximity homogeneous assay), time resolved fluorescence (TRF), and time-resolved fluorescence energy transfer (TR-FRET). The assay was designed to screen an initial library of more than 42,000 single compounds for inhibition of FXR activity. FXR is a member of the nuclear receptor superfamily that plays a role in controlling dietary cholesterol.

Wu screened compounds singly and in five-compound mixtures. Each screening technique identified some compounds that were validated in a cell-based functional assay, but only one (ALPHAScreen) identified all the validated compounds.

This study, said Wu, "should raise concerns about all high-throughput screening."

Peter Knight, a postdoctoral researcher in the department of zoology at the University of British Columbia, commented that the obvious course of action is to evaluate different techniques for different assays.

"But if you have to do this for changes in compound libraries, reagents, and even the target biochemistry, it becomes problematic," said Knight. Still, he noted, it's intriguing that these kinds of differences should be found, "because the fundamental biological process is the same."

The three techniques considered in Wu's study identified 269 compounds of interest in the primary screening, but only 13 were "hits" in all three.

The results were similar at later confirmation and dose-response stages of the experiment. Only nine of the final 34 compounds confirmed in the cell-based functional assay had been identified by all three technologies. "This is a problem," said Wu, "because most people don't have time for a full competitive analysis for each assay they run, yet they must be aware of possible limitations."

Daniela Gabriel, leader of the Drug Discovery Center Laboratory for Novartis' Global Screening Operations Group in Basel, Switzerland, says that she would like to do competitive screening, or perhaps do primary screening with one technology and move to another for secondary screening.

"But that adds more time and more cost," she said, "which defeats the purpose of adopting high-throughput technology platforms in the first place." Her philosophy, which she says is shared by others involved in compound screening, is to choose the technology she trusts the most. "For example," she said, "ratioing methods are good."

Gabriel conducted her own comparative analysis, evaluating five different technologies - ALPHAScreen, homogenous time resolved fluorescence (HTRF), HitHunter (from DiscoveRX), fluorescence polarization (FP), and dissociation enhanced lance fluorescence immuno assay (DELFIA) - for their suitability in an intracellular cyclic AMP (cAMP) assay.

In addition to evaluating such practical elements as cost and ease-of-use, Gabriel performed an experimental comparison. Rather than conduct a complete comparative screening, she compared factors such as sensitivity, signal-to-noise, and "Z-prime" factor. Z-prime, or Z', is a metric relating the variability of positive and negative readings to the difference between the positive and negative levels. Usually a Z' below 0.5 is considered unacceptable. Gabriel's lab adopted a screening platform (HTRF) for the cAMP assay based on that set of performance metrics.

Wu is not certain that traditional performance metrics provide the complete picture. "There's no way to know ahead of time," he said, "if the platform you've selected is the best for the job."

Wei Zheng, a research fellow at Merck in North Wales, Philadelphia, would agree with that assessment. He developed a screening assay for measuring the function of voltage-gated ion channels in the cell membrane. The assay was difficult to develop, and many of the traditional performance metrics, such as signal-to-noise and Z' were relatively poor; yet the assay was very successful in terms of number of leads, the diversity of lead structures, and the quality of the leads generated.

"Even though the screening assay doesn't seem to pass most traditional criteria," said Zheng, "the leads generated will keep the chemists busy for years."

Wu believes the interaction of many factors determines which technology is appropriate for a particular screen. "For now," he said, "it's difficult to provide a recommendation. In the future we may understand enough about compound libraries and their interactions to suggest a specific technology for conducting a specific assay, but not yet."

Send us your comments for publication.

Sign up for BioMedNet News weekly email alerts.

Printer ready version
E-mail article to a friend

 

See also:
Novel detection strategies for drug discovery
[Review]
Ilkka A. Hemmilä and Pertti Hurskainen
Drug Discovery Today, 2002, 7:18:S150-S156

Lead compounds discovered from libraries
[Review]
Adam Golebiowski, Sean R. Klopfenstein and David E. Portlock
Current Opinion in Chemical Biology, 2001, 5:3:273-284

Guggulsterone: an old approach to a new problem
[Research News]
Christopher J. Sinal and Frank J. Gonzalez
Trends in Endocrinology and Metabolism, 2002, 13:7:275-276
 

Printer ready version
E-mail article to a friend