news and views 84 nature structural biology • volume 10 number 2 • february 2003 Cells contain an intricate network of mol- ecular chaperones to ensure that protein folding proceeds smoothly 1,2 . Key players in this network are the Hsp60 class of chaperonins. The chaperonin GroEL from Escherichia coli, together with its co-chap- erone GroES, is the best studied member of the Hsp60 family of proteins. The GroEL–GroES complex, also known as GroELS, has proven to be an intriguingly complex and dynamic molecular machine that mediates protein folding in an ATP- dependent manner 3,4 . An elegant study by Weissman and collaborators 5 reported in the journal Cell now explores the bases and limits of the substrate spectrum of GroELS and provides stimulating insights into its evolutionary plasticity. GroELS substrate promiscuity The cylindrical GroEL complex is a homo- oligomer consisting of two heptameric rings stacked back-to-back, each contain- ing a cavity 6 . The co-chaperone GroES, a heptameric dome-like structure, interacts in an ATP-controlled fashion with one or both GroEL rings, thereby sealing the cavi- ties from the outside 7 . Each GroEL subunit consists of an equatorial ATPase domain, an apical domain that carries the hydrophobic interaction sites responsible for substrate and GroES binding, and a hinge-like intermediate domain, which, upon binding of GroES and ATP, mediates drastic conformational changes in the GroEL ring. These changes, in turn, alter the hydrophobicity and size of the folding cavity. The ATP-dependent interaction cycle of GroEL with both substrate and GroES has been extensively studied (for a review, see ref. 3). Fig. 1 illustrates the classical cis-folding cycle used for folding of substrates that are small enough to be encapsulated in the GroELS cavity. One of the most intriguing features of the GroELS machine is its substrate promiscuity. GroELS assists in the folding of a large variety of structurally and func- tionally unrelated proteins 8,9 , even those from heterologus sources. There are never- theless limits to this promiscuity. For example, actin and tubulin could not be folded by the GroELS system, although they do not exceed the apparent size con- traints (~58 kDa) 10 of the cis-folding cavity. One can envision four major features of GroELS that may define its substrate selec- tivity. First, GroEL does not recognize defined sequence motifs in its substrates but instead binds to surface-exposed hydrophobic patches. Second, the ATPase cycle governs the time a substrate spends encapsulated in the GroELS cavity, pro- tected against inappropriate interactions with other non-native proteins. The encapsulation time set by the ATP cycle may suffice for many (but not for all) sub- strates to reach conformations that are committed to fold to the native state within A folding machine for many but a master of none Annette Erbse, David A. Dougan and Bernd Bukau In vivo selection improves the folding efficiency of the GroELS chaperone toward a specific substrate. Optimizing efficiency, however, comes at the price of narrowed substrate specificity. ADP ADP ADP GroEL GroES ADP U or I kt →ADP ATP ∆t ≈15 sec ATP ADP+P i I kt N or I fc ATP I II III IV ATP ATP ATP cis folding ATP Fig. 1 Model of the GroELS-mediated cis-folding pathway of proteins (adapted from ref. 1). Unfolded protein (U) or a kinetically trapped folding intermediate (I kt ) binds the asymmetric GroEL–GroES complex by interacting with one or more of the hydrophobic binding sites (red) of the GroES- free GroEL ring. GroES dissociates from the trans ring of the tertiary GroEL–GroES trans complex (I) and binds, together with ATP, to the cis ring, thereby sealing the substrate binding cavity (II). The formation of the folding active tertiary cis complex induces drastic conformational changes in the GroEL ring. The substrate is released into the closed cavity and folding commences (III). The hydrophobic interaction sides (red) of GroEL are buried and the new cavity surface is more hydrophilic (green). ATP hydrolysis in the cis ring weakens the interaction between GroEL and GroES, prim- ing the system for GroES and substrate release (III). Binding of ATP to the trans ring can only take place after ATP hydrolysis in the cis ring. It triggers GroES and substrate release (IV), either as native protein (N), as a folding-committed intermediate (I fc ), or as aggregation-prone, kinetically-trapped intermediate (I kt ), which can be rebound by GroELS. The ATP hydrolysis rate thus limits the encapsulation time of the substrate to ~15 seconds. © 2003 Nature Publishing Group http://www.nature.com/naturestructuralbiology