A clash with a flexible loop won an inhibitor 12,000-fold selectivity for p38 delta

Potency and selectivity are not the same measurement, and the distance between the two gains says something about how this compound works. Potency against p38 delta went up 110-fold; selectivity went up 12,000-fold ^[3]. If selectivity here is the ratio of on-target to off-target binding, and both figures are quoted against the same earlier compounds, then binding to the off-targets fell by roughly 109-fold ^[16]. On that reading, most of the improvement came from the molecule getting worse at everything else. The release describes the comparison only as "previously available compounds" ^[3].

Nico Seidler synthesized several molecules in Stefan Laufer's lab at the University of Tubingen, and one was exceptionally selective for p38 delta. Why it was so selective stayed unclear until he moved to David Heppner's lab at Buffalo as a Fulbright fellow and solved the structure by X-ray crystallography ^[10]. "Typically, you design a molecule to fit the target. Here, we designed a molecule that makes the target adapt to it," Heppner said ^[9].

The clash is the part a shape-matching design would have engineered away. The nine-in-ten figure is Seidler's own characterization, not a count from a reported screen. "Nine times out of 10, this kind of bump would not result in binding. It would actually usually make things worse," Seidler said ^[6].

The transferable finding is a residue. Histidine 30 sits far from where the molecule binds, and it stabilizes the kink that lets the loop close around the compound ^[7]. A pocket-by-pocket comparison across the p38 family would never have flagged it. "It's rare to have a region of the protein so far from the binding site influence binding," Heppner said ^[8].

So the strategy question stays open. Aiming for induced fit requires knowing in advance which loop will kink. The evidence here is one compound whose structure was read after the fact ^[10], set against the authors' own estimate that most such clashes fail ^[6]. The narrower result is solid and useful. p38 kinases are notoriously hard for small molecules to bind ^[11], and until now there was no highly selective way to shut down p38 delta without touching its close relatives ^[12]. "Now that we have something selective for this target, we can better understand its biology and figure out where it might be best deployed therapeutically," Laufer said ^[13]. The work reported is structural and biochemical, funded by the National Institute of General Medical Sciences and the German Research Foundation. The account does not describe the compound tested in cells or animals ^[15][17].