Eliminating ATP Binding in Specific ClpX Subunits Yields Functional ATP-Fueled Protein-Unfolding Machines

Abstract

AAA+ ring hexamers of ClpX utilize the energy of ATP binding and hydrolysis to unfold protein substrates and translocate the resulting denatured polypeptide into the degradation chamber of ClpP, an associated self-compartmentalized peptidase. Nucleotide-dependent conformational changes are necessary for ClpX binding to ClpP and to protein substrates as well as for allosteric machine function. ClpX functions asymmetrically. In crystal structures, for example, four loadable subunits are competent for nucleotide binding, whereas two unloadable subunits are not. Moreover, ATP hydrolysis in one subunit can power conformational changes in the entire ring and protein degradation. Subunit-subunit interactions are essential for ClpX machine function, but the communication mechanisms are poorly defined. For instance, it is not known whether the conformations of loadable and unloadable subunits remain fixed or interchange during the hundreds of cycles of ATP hydrolysis that are required for protein unfolding, translocation, and degradation. To probe subunit communication, we constructed covalently linked mutant hexamers in which the nucleotide affinity of specific subunits was dramatically reduced by mutations in the Walker A, sensor-II, or box II motifs and developed novel fluorescence assays to probe nucleotide-binding cooperativity as well as ATP binding to specific subunits. Strikingly, ClpX pseudo-hexamers bearing two opposed subunits with severe ATP-binding defects hydrolyze ATP at near normal rates and are able to unfold and translocate protein substrates. For some variants, machine function was retained despite sensor-II-dependent abrogation of the positive cooperativity of ATP binding. Because hexamers with two “permanent” unloadable subunits retain basic ClpX machine functionality, subunit switching between unloadable and loadable conformations does not appear to be required for protein unfolding or translocation. Our results are most consistent with probabilistic models of ATP binding and hydrolysis rather than strictly sequential models.