Conjuring up needles as anticancer drugs

Ali Tavassoli and Stephen Benkovic at Pennsylvania State University thought that a good point of attack for the desired inhibitor would be the enzyme ATIC, which catalyzes the last two steps of the biosynthesis of purine.

This enzyme is active only if two identical units associate to form a dimer. The goal: to disrupt the interactions between these two units so that dimerization is blocked. “The inhibitor should be a small, cyclic peptide,” says Benkovic, “because these are particularly well suited as drugs.” But how to find a suitable peptide in the nearly infinite number of peptides theoretically possible? The answer: “By blindly producing as many of them as possible and then testing this library of substances for suitability–an extremely complex undertaking by conventional methods.

The researchers thus proceeded differently: They incorporated DNA fragments with a random sequence into bacteria. The random sequence is embedded in a DNA segment that codes for a specific type of protein, an intein. Using this set of instructions, the cells produce short peptides with the same–random–amino acid sequence that is integrated into the intein. Inteins have a special ability; they automatically cut the “extra” peptide sequence out of their center and cyclize it.

It is thus possible to produce several million cyclic peptides “in one go”. But how can active peptides be identified without complex tests? The cells include genes that allow them to survive in special antibiotic-containing media. A switching gene is also added. As long as a repressor protein is bound to the switch, the inserted survival genes are not read–the cells die. The trick is that the repressor is a hybrid of the DNA binding sequence and the dimeric target protein ATIC. Only when the cell contains an active dimerization inhibitor does the dimer split in half, releasing the repressor from the DNA and allowing the genes to we switched on. The cell then survives to form colonies whose DNA allows for the easy identification of the active peptides. Says Benkovic: “The selected cyclic peptides were, indeed, capable of inhibiting the target enzyme by blocking its dimerization.”