Evaluation of competing J domain:Hsp70 complex models in light of existing mutational and NMR data

The work by Ahmad et al. (1) presented an NMR-based model for a bacterial DnaJ J domain:DnaK(Hsp70):ADP complex that differs from our crystal structure of a disulfide-linked bovine Hsc70:auxilin J domain complex (2). The work by Ahmad et al. (1) claimed that their model can better account for published mutational data, that their model is in better agreement with a previous NMR study of a bacterial DnaJ J domain:DnaK complex (3), and that the cross-link in our structure renders it irrelevant for understanding J domain:Hsp70 interactions or mechanism. However, bacterial DnaK and mammalian Hsc70 bind J proteins very differently. Bacterial DnaK interacts primarily with helix II of DnaJ (3), whereas bovine Hsc70 interacts primarily with helix III and the histidine-proline-aspartate (HPD) loop (4). In our structure, helix III and the HPD loop of the bovine auxilin J domain form the primary interaction surface with bovine Hsc70, and our structure is in excellent agreement with an NMR study of a bovine Hsc70:polyoma virus T-ag J domain complex (4), which is a more appropriate comparison than the bacterial complex. Our structure can better account for the mutational data: 8 of 9 DnaJ J domain residues with mutations that affect J function, 12 of 14 residues in viral T-ag J domains with mutations that affect J function, and 10 of 12 residues/regions in DnaK/Hsc70 with mutations that affect interaction with J correspond to residues that are close (<8 A) to the binding partner in our complex. In contrast, in the model by Ahmad et al. (1), only 1 of 9 functionally critical DnaJ residues, 1 of 12 DnaK residues/regions, and 0 of 14 viral T-ag J domain residues are within 15 A of residues in the binding partner that were spin-labeled to provide NMR distance constraints. The latter includes a spin-labeled DnaK residue (V210) that is in the center of the loop proposed in the model by Ahmad et al. (1) to form the binding surface for DnaJ. Cross-linking is frequently used to capture transient or dynamic complexes for crystallization, with several validated examples in the Protein Data Bank (i.e., 1RTD, 1KLN, and 1S6V). Evidence for the relevance of our structure is that (i) the J domain has ATPase stimulating effects on the cross-linked Hsc70 that exactly recapitulate the ATPase stimulation seen for the two uncross-linked species (2), (ii) the structure predicts the location of Hsc70 mutations that disrupt its ability to be stimulated by the J protein (2), (iii) the structure indicates how the J domain stimulates Hsc70 ATPase activity through conformational effects on the linker between the Hsc70 nucleotide and protein binding domains (2), a mechanism that is gaining increasing acceptance, and (iv) the structure is consistent with a cryo-EM structure of a clathrin cage:auxilin:Hsc70:ATP complex. Space and reference limitations preclude additional elaboration of these issues or listing of all relevant references here. However, interested readers can access a detailed discussion along with tables summarizing the NMR and mutational data and references in ref. 5.