However, randomization is usually performed on a restricted region of target proteins, whereas the rest of it is left unchanged. Alternatively, a natural protein is used as a scaffold to engraft short random peptides. This approach can be defined as “directed randomization”, since randomization is confined to a certain region in order to achieve a novel—yet, chosen ‘a priori’—property. The novelty in our research is basically

different from “directed randomization” since it aims to explore the space of sequences of completely random Tanespimycin supplier proteins with no preconception as to what their properties might be: a “total randomization” approach. With our work, STI571 manufacturer find more using the technique of phage display, we were

able to produce large libraries of random de novo polypeptides and identify sequences for further structural investigation. These NBP has totally random sequences, except for a tri-peptide (PRG) which is the site of thrombin cleavage-based on the consideration that folded proteins were protected against such a digestion. Our data show that, very surprisingly, the frequency of fold in such libraries of never born proteins is very high, about 20% of the entire set. The determination of the optimal substrate (PRG) for thrombin cleavage was of particular importance. Furthermore, and most importantly for the general philosophy of the concept, protein folding appeared to protect the PRG site against thrombin digestion, in both the phage-linked form as well in the free protein used as control. This generalized

Morin Hydrate protocol for the selection of folded proteins by proteolysis guarantees an efficient digestion of unstructured protein sequences while folded proteins are not affected. This procedure can be applied both for protein stabilization or selection of stable variants derived from a mutant library of extant proteins and for the selection of folded and stable sequences from de novo totally random phage libraries based on their fold properties. The detailed structural study of each isolated protein is lengthy and complex and the characterization of purified samples is rate-limiting. In this preliminary phase, we present the partial characterization of few proteins, whereby the clones were chosen purely by a random procedure, which imparts a good degree of statistical validity despite their small number. In addition, the sequences have no putative conserved domains and no significant similarity with known protein sequences present in data banks. The sequences analysed in more detail appear to form globular, folded structures and, judging from the spectroscopic data (CD and fluorescence) and computer modelling they do not, at first sight, present peculiar structural features with respect to extant proteins.