Session 3: Where have our explorations taken us so far?

Where have all the targets gone?

Joachim Pietzsch

In theory, the sequencing of the human genome was going to inundate researchers with drug targets. In reality, however, the situation is not as bountiful as first hoped.

In the golden age of drug discovery, there was not much ado about targets. Drugs were discovered by discussions between chemists, pharmacologists and physicians, and their pharmacological profile was to a large degree deduced from preclinical and clinical observations.

In the late eighties, however, the concept of 'targets' grew from the promise of genomics. By deciphering the sequence of the human genome, drug hunters hoped to become drunk with targets, as it was thought that hundreds, if not thousands, of previously hidden genes would prove to be linked with disease. "The number of potential drug targets may lie between 5,000 and 10,000", estimated Jürgen Drews in his widely-cited paper (Drews, J. Nature Biotechnol. 14, 1516-1518 (1996)) .

Almost ten years later, however, scientists have sobered up in their expectations. The biological information hidden in the genome proved to be more difficult to interpret than anticipated. Despite a deluge of genomic data, drug discovery is running short of targets.

"We have historically fewer innovative targets per year", said Christopher Lipinski, fomerly of Pfizer, showing that only 24 innovative drugs with new targets have been launched between 1994 and 2001 (TABLE 1).

		Table 1 | Few �innovator� targets are being discovered Only 24 new targets have been approved over the eight year period between 1994 and 2001. All drugs are small-molecule treatments apart from those denoted with an astrix, which are biological drugs.

"Many more druggable targets may have emerged in these eight years, but there are not enough druglike molecules to match them", Lipinski said. When high-throughput screening and other biology efforts fail to find target modulators for druggable targets, he added, isn't it then justified to say "It's the chemistry, stupid!"?

Partly, answered Andrew Hopkins from Pfizer. Yes, there are physicochemical constraints for new drug candidates, for example, in protein architecture and flexibility of the target and in its specific surface structure at the binding site. Yes, also because molecules also face pharmacokinetic limits which are determined by its molecular weight and other parameters as described in Lipinski's rule of five (see 'Exploring biological space'). But both of these issues can be addressed by the creativity of chemists.

It's the biological limits, however, that according to Hopkins are the biggest barriers to discovering druggable targets. In the universe of around 24,000 human genes, around 3,000 can be regarded as coding for druggable targets. On the other hand, large-scale mouse gene knockout studies showed that only 10% of knockouts had clinical effects, which implies that there are only around 2,500 therapeutically relevant genes in humans. The overlapping region between these two groups, that is, genes coding for therapeutically relevant, druggable targets is only around 200.

The main reason for such a low number of targets is the redundancy of biological networks. Blocking an enzyme or receptor doesn't necessarily guarantee a pathway will shut down - biological pathways often have compensatory mechanisms that maintain function. What is valuable for maintaining life, though, can be frustrating for drug hunters, as this is the main reason why many innovative drug compunds perform so well in a more simplistic in vitro environment, yet fail when they are tested in a more complex in vivo model.

Also, as Günter Wess from Aventis said: "Most of the molecules that are chemically optimized are biologically not relevant. We have moved away too far from the science of drug discovery to industrialization." Without a good starting point this approach fails.

Stuart Schreiber from Harvard University, on the other hand, blamed "the risk-adverse attitude of big pharma" for much of the trouble: "They are all going for the same targets, and then wonder why they cannot find any not more". A related remark came from Tom Blundell from the University of Cambridge, who pointed at the very narrow area of diseases big pharma is addressing in order to be profitable. It is likely therefore that only a narrow window of targets are being identified.

With more knowledge would come more power in finding targets, said some of the delegates. Given that the human genome sequence pointed to a greater complexity than the 'one-gene, one-protein' model, looking at the druggable proteome would be more informative than looking at the druggable genome, although what defines druggable varies in definition. Diseases need to be defined better, as understanding pathways will improve routes of attack. Also, drugs, by and large, only target active sites, and other modes of inhibition, such as allosteric modulators have not been investigated thoroughly.

Given all these gaps in knowledge, Schreiber questioned whether we had enough information to confidently say how many genes coding for therapeutically relevant, druggable targets are out there. The problem said Lipinski is more that in 2004, pharma has a problem finding drugs to target that are economically viable. In that case, said Schreiber, shouldn't academia and industry be combining forces to mine the genome better?