Helices 2 and 3 are the initiation sites in the PrP(C) → PrP(SC) transition.

It is established that prion protein is the sole causative agent in a number of diseases in humans and animals. However, the nature of conformational changes that the normal cellular form, PrP(C), undergoes in its conversion to a self-replicating state is still not fully understood. The ordered C-terminus of PrP(C) proteins has three helices (H1-H3). Here, we use statistical coupling analysis (SCA) to infer covariations at various locations using a family of evolutionarily related sequences and the response of mouse and human PrP(C)s to mechanical force to decipher the initiation sites for the transition from PrP(C) to an aggregation-prone PrP* state. Sequence-based SCA predicts that the clustered residues in nonmammals are localized in the stable core (near H1) of PrP(C), whereas in mammalian PrP(C), they are localized in frustrated helices H2 and H3 where most of the pathogenic mutations are found. Force-extension curves and free energy profiles as a function of extension of mouse and human PrP(C) in the absence of a disulfide (SS) bond between residues Cys179 and Cys214, generated by applying mechanical force to the ends of the molecule, show a sequence of unfolding events starting first with rupture of H2 and H3. This is followed by disruption of structure in two strands. Helix H1, stabilized by three salt bridges, resists substantial force before unfolding. Force extension profiles and the dynamics of rupture of tertiary contacts also show that even in the presence of an SS bond the instabilities in most of H3 and parts of H2 still determine the propensity to form the PrP* state. In mouse PrP(C) with an SS bond, there are ∼10 residues that retain their order even at high forces. Both SCA and single-molecule force simulations show that in the conversion from PrP(C) to PrP(SC) major conformational changes occur (at least initially) in H2 and H3, which because of their sequence compositions are frustrated in the helical state. Implications of our findings for the structural model for the scrapie form of PrP(C) are discussed.