Simplification of complex structures is one way medicinal chemists avoid lengthy (and risky) synthetic routes to the analogs of natural products. And it’s absolutely rational, if one can get the same pharmacological effect with a simple molecule, why bother making a complex 3D scaffold? One classic med-chem textbook example is simplification of morphine scaffold to pethidine that led eventually to the development of fentanyl.
After more than a century of collecting anecdotal examples of this approach, it’s finally time to hand it over to computers and move to something more interesting. So that’s exactly what Gisbert Schneider lab is trying to do at ETH Zurich. In their last Angewandte paper authors subjected the scaffold of (−)-englerine to an automated simplification.
The problem of simplification is two-fold. On the one hand the new molecule should maintain minimal features essential for the biological activity of the complex natural products. On the other hand, the synthesis of the simplified molecule should be, well, simple. The first part of the problem can be (and is) solved by now quite standard 3D pharmacophore alignment. It’s the second part that makes the analysis complicated.
To ensure the simplicity of synthesis, Schneider group uses their own algorithm of generating novel molecules: DOGS (Design of Genuine Structures). The software uses 83 formalized reaction schemes for constructing libraries of analogs. Of these reactions 34 represent various methods of making hetero- and carbocycles, 8 – variations of reductive amination, 5 – metal-catalyzed cross-coupling, 4 – amide coupling, 4 – Mitsunobu substitution. All in all, it’s “a medicinal chemistry-inspired method for the de novo design of drug-like compounds, placing special emphasis on the synthesizability of the designed molecules”. Yep, all the medicinal chemist’s favorite tools are in there but not much of something else.
The method (and number of similar ones) seems to produce quite some useful output. So avoiding natural product synthesis becomes even easier! Sure, the drug-like chemical space won’t become more diverse because of that, but who cares if the molecules will eventually do something useful?
The bigger question remains: what will ‘something more interesting’ be?