Dynamical Structures of Hsp70 and Hsp70-Hsp40 Complexes

Abstract

Protein misfolding and aggregation are pathological events that place a significant amount of stress on the maintenance of protein homeostasis (proteostasis). For prevention and repair of protein misfolding and aggregation, cells are equipped with robust mechanisms that mainly rely on molecular chaperones. Two classes of molecular chaperones, heat shock protein 70 kDa (Hsp70) and Hsp40, recognize and bind to misfolded proteins, preventing their toxic biomolecular aggregation and enabling refolding or targeted degradation. Here, we review the current state of structural biology of Hsp70 and Hsp40-Hsp70 complexes and examine the link between their structures, dynamics, and functions. We highlight the power of nuclear magnetic resonance spectroscopy to untangle complex relationships behind molecular chaperones and their mechanism(s) of action.

FIGURE 1. The Hsp70 ATPase-substrate interaction cycle and the structures of Hsp70 and Hsp40

(A) (1) A J-protein (Hsp40) binds to a non-natively folded client protein via its C-terminal domain. The N-terminal J-domain of Hsp40 interacts with ATP-bound Hsp70 (2) and both the substrate and Hsp40 catalyze ATP hydrolysis, which leads to release of inorganic phosphate (P_i) and dissociation of Hsp40 from ADP-bound Hsp70. The client protein is transferred from Hsp40 to ADP-Hsp70 (3), which binds to substrate with a higher affinity than ATP-Hsp70. A nucleotide-exchange factor (NEF) binds to the nucleotide-binding domain (NBD) of ADP-Hsp70 (4) and promotes the exchange of ADP (5) for ATP (6). The NEF dissociates from ATP-Hsp70, and the client protein is released from ATP-Hsp70. If native-like contacts have not been established in the client protein, the J-protein will bind to exposed hydrophobic residues and the cycle will repeat. Image taken from (Kampinga and Craig, 2010) with permission from the Nature Publishing Group. (B) (Top) NMR-based structural model (PDB 2kho) of E. coli DnaK (1-605) bound to ADP and substrate peptide. The domains of DnaK are colored as follows: NBD (black), SBD (teal), interdomain linker (red). Specific regions of the NBD and SBD are annotated. (Bottom) X-ray crystal structure (PDB 4jne) of ATP-bound DnaK (T199A) containing truncation of loop_3,4 and removal of the disordered C-terminus. The domains are colored as above. Upon ATP binding, the interdomain linker adopted a β-strand conformation and docked onto a region of the NBD between sub-domains IA and IIA (inset). Note that text indicating the SBD sub-domains has been removed for clarity. (C) Structures of various Hsp40 proteins and their J-domains. (i) A combined EPR and X-ray crystallographic structural model of FL, dimeric DnaJ from T. thermophilus (PDB 4j80). The respective domains within DnaJ are colored as indicated and one monomer within the dimer is colored black for clarity. Helices I and IV have been colored orange to highlight their presence or absence in the ensuing J-domain structures. (ii) The solution structural ensemble of DnaJ₁₀₃ from E. coli (PDB 1bq0). This variant included helices I-IV in the J-domain, the proline-rich region, and a portion of the GF-linker; however, only the structure of the J-domain was solved by NMR. An inset displays the HPD motif as red sticks. (iii) A crystal structure of the auxilin J-domain from B. taurus, which comprised residues 810–910 (PDB 1nz6). Note that helix I is much shorter and helix IV is absent. (iv) A crystal structure of full-length HscB from E. coli (PDB 1fpo). The additional helices comprise the C-terminal domain, which binds to substrate protein.

FIGURE 2. Conformational dynamics of the Hsp70 nucleotide-binding domain

(A) Alignment of Hsc70 nucleotide-binding domain (NBD) crystal structures in the presence or absence of various nucleotides. Here, the structures of isolated Hsc70 NBDs (residues 4–381) have been superimposed by alignment of their backbone atoms. Five structures were used in this alignment, comprising the apo (PDB ID 2qw9; 0.47 Å), ATP (PDB ID 1kax), ADP (PDB ID 2qwl; 0.32 Å), ADP-P_i (PDP ID 3hsc; 0.34 Å), and ADP-V_i (PDP ID 2qwm; 0.30 Å) states of Hsc70 NBD. (E. coli DnaK residues 1–37, 112–184, 363–383 IA; (38–111) IB; (185–227, 310–362) IIA; (228–309) IIB). (B) Refined structural models of the DnaK (T. thermophilus) NBD residues 1–381 with H^N RDCs in both AMPPNP- and ADP-bound states. Note the structural rearrangements as compared to Figure 2A. Figure adapted from (Bhattacharya et al., 2009) with permission from Elsevier. (C) Solution-state NMR spectra of the 40 kDa NBD of DnaK (T. thermophilus). Overlay of ¹H-¹⁵N TROSY-HSQC spectra of 400 μM apo- (blue) and ADP- NBD (black). Overlay of ¹H-¹⁵N TROSY-HSQC spectra of ADP- and AlF_x-NBD (green). AlF_x, which is an ATP analogue, suppressed peak doubling that was otherwise observed in the spectrum of ADP-NBD. Figure adapted from (Revington et al., 2004) with permission from the American Society for Biochemistry and Molecular Biology. See text for more details. (D) Interdomain linker-induced chemical shift perturbations (CSPs) mapped onto the ATP-bound NBD homology model of DnaK. Red spheres indicate residues with large CSPs. Green residues are those that displayed sensitivity to nucleotide, as measured by the observation of large differences in chemical shifts between ADP- and ATP-bound NBD. Space-filling model of the results depicted adjacently, with red, yellow, cyan, and grey residues indicating large, medium, or insignificant CSPs and no data, respectively. Figure adapted from (Zhuravleva and Gierasch, 2011) with permission from the National Academy of Sciences.

FIGURE 3. Conformational dynamics of the Hsp70 substrate-binding domain

(A) Solution-state NMR spectroscopy-derived structural ensembles of the isolated SBD of DnaK (E. coli) in the presence (top) and absence (bottom) of substrate peptide NRLLLTG. The β-strands are colored as follows: β1 red, β2 orange, β3 yellow, β4 green, β5 teal, β6 blue, β7 pink, and β8 purple. Functionally relevant loops 3,4 (L_3,4) and L_5,6 are indicated. See text for details regarding these loops. (B) Summary of EPR spectroscopy results that measured the accessible distances to the isolated SBD of DnaK (E. coli) in the presence of substrate. The cartoon depicts three models that are consistent with the distance restraints. Figure adapted from (Schlecht et al., 2011) with permission from the Nature Publishing Group. (C) A structural model of the DnaK (E. coli) SBD displaying significant CSPs (red) upon mutation of loop 3,4 (L_3,4) is shown. Note the large number of residues affected by this mutation. A region of a ¹H-¹⁵N TROSY-HSQC spectrum of ATP-bound DnaK (1-552) T199A/L542Y/L543E (black) overlaid upon the same protein with mutations in the loop 5,6 (L_5,6) region (red). Note that many of the resonances in the SBDβ either broaden into the noise or are reduced in intensity, indicative of enhanced μs–ms motions. A cartoon below depicts such motions in the L_5,6 mutant. Figure adapted from (Zhuravleva and Gierasch, 2015) with permission from the National Academy of Sciences, USA.

FIGURE 4. Detection of the allosterically active intermediate in the chaperone cycle of Hsp70

(A) Methyl-TROSY spectra from ATP-bound DnaK₅₅₂ in the apo (grey) and substrate-bound (black) states. Overlaid upon these spectra are those recorded on the isolated NBD (red) and linker-bound NBD (blue). Substrate binding to the ATP-bound full-length protein modulates the conformational ensemble of the interdomain linker, as both linker-bound and -unbound states are populated, with chemical shifts between these states found in the isolated NBDs. (B) Residues with significant CSPs (yellow) between the NBDs of ATPγS-bound DnaK₅₅₂ and DnaK₅₅₂ (L390V) are mapped onto a homology model of ATP- and substrate-bound DnaK. (C) The methyl-TROSY spectrum of Ile-labeled DnaK₆₀₁(³⁸⁹DDD³⁹¹) bound to ATP and substrate (black) overlays nearly perfectly with the isolated ADP-bound NBD (red). This variant of DnaK₆₀₁ is allosterically defective. Note that additional resonances that are absent in the NBD spectrum have arisen from isoleucines in the SBD. (D) A schematic that depicts conformational transitions in DnaK. (1) The transition between domain-docked (ATP) and domain-undocked (ATP and substrate) ensembles, which corresponds to the grey and black resonances in Figure 3A. (2) The transition between linker-bound (ATP) and (3) linker-unbound (ATP and substrate) ensembles, which corresponds to the blue and red resonances in Figure 3A. In ATP- and substrate-bound FL DnaK, this transition occurs on a fast timescale, as only one resonance is observed. This resonance corresponds to the population-weighted average of the two linker conformations. Residues stabilized by binding of substrate, ATP, or both are colored red, blue, or yellow, respectively. Figure adapted from (Zhuravleva et al., 2012) with permission from Elsevier.

FIGURE 5. Conformational dynamics of Hsp40-Hsp70 complexes

(A) DnaJ₇₀ is shown in white, and its average position with respect to DnaK₆₀₁ (NBD in yellow, SBD in cyan) determined from MD simulations and PRE restraints with MTSL tags at the indicated residues in both DnaJ₇₀ and DnaK₆₀₁. Green residues were not in contact with either DnaJ₇₀ or DnaK₆₀₁ when an MTSL spin-label was placed at those positions in DnaK₆₀₁ or DnaJ₇₀, respectively. DnaJ₇₀ residues in blue were found to contact the red and orange residues in DnaK₆₀₁. (B) Overlaid snapshots from the 30 ns MD simulation of DnaJ₇₀ (white) bound to DnaK NBD (yellow). (C) A zoomed in region of the non-covalent complex indicating functional residues that were discussed in the text. The HPD motif is shown in green. Other residues have been colored as in A. Figure adapted from (Ahmad et al., 2011) with permission from the National Academy of Sciences, USA. (D) NMR spectroscopic results acquired from the interaction between unlabeled, ATPase-deficient HscA(T212V) and ¹⁵N-labeled HscB. HscB residues that were broadened upon addition of HscA(T212V) are colored red; residue that showed CSPs are colored blue. The nucleotide state of HscA(T212V) is indicated in the central boxes. Note that ATP-HscA(T212V) caused significant broadening in the both the J-domain and C-terminal domain (CTD) Figure adapted from (Kim et al., 2014) with permission from the American Chemical Society.

FIGURE 6. Evidence from X-ray crystallography and native mass spectrometry for dimeric Hsp70 and its multi-chaperone complexes

(A) The asymmetric unit of E. coli DnaK bound to ATP (PDB 4JNE) contains two molecules. Right, inter-domain contacts are indicated in circles. Figure adapted from (Sarbeng et al., 2015). (B) Native mass spectra of Hsp90 and Hop with Hsp70 from either E. coli (left) or Sf9 cells (right). The y-axis displays relative peak intensity and the x-axis is m/z; each protein and protein complex with a given mass m yields a Gaussian-shaped charge series. The authors assigned the peaks within each charge series to the indicated chaperone complexes. Note that Hsp70 from Sf9 cells yielded dimeric Hsp70 and Hsp90₂Hsp70₂Hop complexes. (C) A catalytic amount of Hsp40 mixed with E. coli Hsp70, Hsp90, Hop, and GR substrate protein led to the formation of a hexameric complex (indicated by rectangles). Other stoichiometries of chaperone-chaperone and chaperone-substrate complexes were observed as well. (D) Increasing the relative amount of Hsp70 from (C) led to the stable formation of the hexameric chaperone-substrate complex. (Cartoon) Comparative XLs were identified in the hexameric complex formed in (C) and (D). The lysine residues that formed XLs between each protein are annotated. Figure adapted from (Morgner et al., 2015) with permission from Elsevier.