Fragment-based approaches to lead discovery — in which low-molecular mass, weakly binding 'fragments' are transformed into highly potent larger molecules by various design strategies — have become increasingly popular in recent years. By taking a reverse 'deconstructive' approach and investigating the characteristics of fragments that might be derived from high-affinity lead compounds, two recent papers now provide insights that could be valuable for fragment-based lead discovery.

In the first paper, by Babaoglu and Shoichet in Nature Chemical Biology, the authors investigate the question of whether a higher-affinity inhibitor can always be parsed into component fragments that will still bind to the target protein in geometries corresponding to those observed in the original inhibitor.

Contrary to their expectations, when they deconstructed a previously known β-lactamase inhibitor into three fragments, none of them bound in a position that recapitulated its position in the larger molecule. As the authors note, their findings suggest that in some cases, good inhibitors might be missed by fragment-based approaches.

The second paper, from Hadjuk in the Journal of Medicinal Chemistry, considers the crucial issue of the impact of fragment potency on the likelihood of transforming an initial fragment into a more potent lead with appropriate drug-like properties. Hadjuk performed a retrospective analysis of 18 highly optimized inhibitors, in which the compounds were systematically deconstructed until the minimum binding elements could be identified. Analysing the potency changes that were observed as the leads were reduced in size revealed a nearly linear relationship between molecular mass and binding affinity over the entire range of sizes and potencies represented in the dataset.

On the basis of these data, Hadjuk suggests that one conclusion that can be drawn is that while following an ideal path from optimized fragment to optimized drug lead, every mass unit that is added to the initial fragment contributes equally and proportionally to the binding affinity. This has several potentially important implications for fragment-based lead discovery, as the author notes. First, the constancy in the binding efficiency (affinity/mass) could be used as a quantitative measure of effective fragment elaboration. Second, the data emphasize the importance of beginning with the most efficient fragment lead, regardless of absolute potency or molecular mass, because if efficiency (or inefficiency) remains constant during optimization, then beginning with an inefficient fragment will make optimization more challenging. Appropriate use of such relationships should increase the chances of success of future fragment-based lead discovery programs.