Some alarming news (at least to this blog) is hidden in the editorial by Brian Shoichet in a recent Nature Chemistry on developments in drug discovery: the pharmaceutical companies are no longer interested in natural products as lead compounds as they are often only active as mixtures, and (...) not amenable to rapid synthetic optimization. This must be bad news for rainforest proprietors who are now stuck with their merchandise of medicinal flowers, plants , roots and whatever. Diversity oriented synthesis according to Shoichet also has its setbacks: the molecules this strategy spawns simple do not have any biological track record. Enter fragment-based lead discovery: decompose a large biomolecule into fragments and in one or more of them biological activity (binding to proteins) should be retained. For example, fragment analysis of the natural alkaloid cytisine led to the development of the artificial drug Varenicline.
The Shoichet editorial was only an appetizer for some actual fragment analysis by Over et al. (DOI) in the same issue. Here are the numbers. The initial population was about 1800,000 biomolecules who had their chains pruned, their fused ring decoupled and their spacers cleaved. This resulted in an initial set of 750,000 fragments and after eliminating the weak, toxic and feeble a set of 160,000. These fragments had 18,000 so-called Murcko fragments in common of which about 1500 were not already known in the ZINC database. After compensating for similarity (Tanimoto treatment) the final set was about 2000 clusters of fragments. These fragments were then analysed as MAP kinase inhibitors. Effective molecules were indeed found to fill up key protein voids.
Interesting thought: what would happen if you administer the complete set of fragments of a single biomolecule as a cocktail?