Caging helps proteins fold.

H ow do proteins fold spontaneously? The quest to answer this question has led to significant developments on theoretical, experimental, and computational fronts in the last decade (1–7). A combination of approaches has provided a detailed understanding of the nature of pathways and the transition states that the polypeptide chain encounters as it traverses the rugged energy landscape. Even as our understanding of in vitro protein folding at infinite dilution has advanced, it has become urgent to address two additional issues of biological interest. (i) In vivo folding is not always a spontaneous event. A subset of proteins may require molecular chaperones. In an illuminating article in this issue of PNAS, Takagi et al. (8) provide a detailed study of five model proteins of varying native state architecture confined to cylindrical nanopores, which are meant to mimic the cavity of GroEL. Of all the molecular chaperones, the GroEL GroES system from Escherichia coli, which assists folding of a fraction of cytosolic proteins, is the best understood. GroEL, a cylindrical barrel, consists of two heptameric rings stacked back-to-back giving rise to an unusual sevenfold symmetry about the axis of the cylinder (9). The annealing action of the chaperonin machinery (GroEL GroES) is complex and involves large allosteric domain movements in GroEL in response to binding of the substrate protein (SP), ATP, and GroES (10, 11). The presence of a central cavity has led to the proposal that GroEL merely offers a protective environment in which SP folds as it would in infinite dilution. Indeed, for an undetermined duration out of the total of 10 sec of the chaperonin cycle, SP experiences confinement in the cylindrical cavity, whose maximum volume is 175,000 Å3. However, the annealing action of GroEL is due to the changes in the inner lining of the cavity that are triggered by SP, ATP, and GroES binding, which implies that GroEL plays an active role in the rescue of SP. (ii) Even proteins that fold spontaneously in cells do so only in a milieu that contains other biological macromolecules that serve as crowding agents (12). The volume fraction of the crowding agents (ribosomes, RNA, polysaccharides, etc.) can be in the 20–30% range (13). To a first approximation, the role of crowding agents may be modeled as confining the protein to a restricted space (Fig. 1). Let us assume that the crowding agents merely exclude a fraction of the available volume to the polypeptide chain. If this is the case, then it is clear that the polypeptide chain would, with high probability, be found in a region free of the crowding agents; i.e., folding would take place in a restricted space (Fig. 1). To qualitatively understand in vivo folding it is important to develop a conceptual picture of how the principles that govern in vitro folding in infinite dilution change when other factors (molecular chaperones and crowding effects) are taken