A fluorescent multi-domain protein reveals the unfolding mechanism of Hsp70

Proteins can acquire secondary and near-native tertiary structures co-translationally and, once released, may function, translocate, or assemble into oligomers. In the crowded cellular milieu, native proteins must remain sufficiently stable, yet stress can partially unfold proteins to expose hydrophobic residues, promoting misfolding and aggregation into diverse, often irreversible species. Misfolding and aggregation, with cytotoxic consequences notably in neurons, are intrinsic to protein physics. Cellular protein quality control relies on ATP-dependent chaperone families (Hsp70, Hsp60, Hsp90, Hsp100) and co-chaperones that prevent and actively reverse aggregation. Strong evidence indicates Hsp70s, Hsp100s and Hsp60s act as polypeptide-unfolding enzymes that convert stable misfolded and aggregated substrates into unstable, unfolded monomers, which can refold to native upon release. Hsp70 collaborates with Hsp60, Hsp90 and Hsp100 yet can function independently and serves as a central hub of protein repair machineries in prokaryotes and eukaryotes. However, mechanistic elucidation is hindered by the heterogeneity of non-native ensembles comprising aggregates of various sizes, misfolded monomers and transiently unfolded species. This study addresses these challenges using a designed multi-domain substrate to parse Hsp70 mechanisms.

Design of reporter substrate: Constructed MLucV, a monomeric ~120 kDa fusion protein comprising a mutant, stress-labile firefly luciferase core flanked N- and C-terminally by stress-resistant fluorescent proteins (mTFP1 donor and Venus acceptor). This architecture enables simultaneous monitoring of conformational states by FRET and enzymatic luciferase activity, in vitro and in vivo. MLucV cloning and expression were performed in E. coli; purification used Ni-NTA affinity and size-exclusion chromatography. Aggregate preparation: Generated reproducible small MLucV aggregates via two protocols: (i) urea denaturation (30 µM MLucV in 4 M urea, 5–10 min at 25 °C) followed by dilution to 0.4–1 µM (UMLucV); (ii) heat denaturation (0.4–0.5 µM MLucV, 23–35 min at 38–39 °C) followed by cooling (HSMLucV). Stability and solubility assessed by light scattering, high-speed centrifugation, and activity assays. Biophysical characterization: FRET spectroscopy (excitation 405 nm; emission 480–580 nm) with computation of ensemble FRET proximity ratio (FRET-PR) and normalization to native MLucV and separated fluorophores. Luciferase activity quantified via D-luciferin luminescence assays. SEC-RALS determined oligomer mass and stoichiometry. Negative-stain TEM visualized particle morphology and size distributions. Chaperone assays: Employed E. coli Hsp70 system DnaK-DnaJ-GrpE (KJE) at typical concentrations 4 µM DnaK, 1 µM DnaJ, 2 µM GrpE with 4 mM ATP unless specified. Order-of-addition protocols sequentially introduced DnaJ, DnaK, and GrpE to pre-formed aggregates in ATP to dissect steps. Compared conditions with ATP, ADP, or no nucleotide. Monitored time-resolved FRET and luciferase activity; TEM pre- and post-KJE incubation assessed particle disappearance. Non-equilibrium experiments at elevated temperature: Native MLucV incubated at 38 °C with KJE and varying ATP concentrations (0–6.4 mM) to test ATP dependence of native-state accumulation under denaturing conditions; repeated ATP pulses examined reversibility and persistence. Inclusion of ClpB tested necessity of co-disaggregase. Molecular dynamics (MD) simulations: Coarse-grained simulations (LAMMPS) modeled MLucV monomers in native, misfolded, urea-unfolded, and DnaK-bound states (with eight ADP-DnaK bound to luciferase segments), and aggregation of 12 misfolded MLucV monomers under shrinking confinement to form oligomers. Distance distributions between mTFP1 and Venus were computed; simulated dodecamers analyzed for geometry and diameter distributions. Binding site identification used established algorithms; rigid-body models for native domains and DnaK SBD/NBD were employed. In vivo measurements: MLucV expressed in E. coli W3110; whole-cell FRET recorded before and after 39 °C heat shock; luciferase activity measured from lysates to correlate misfolding with functional loss. Controls: Assessed fluorescence stability of individual mTFP1 and Venus under urea and heat; compared aggregation and solubility of native luciferase without FPs; evaluated BSA competition effects on refolding yields; GroEL/ES used as particle size controls in TEM.

- MLucV enables discrimination of conformational states: Native, urea-unfolded, aggregated, and Hsp70-bound states show distinct FRET-PRs and activities. Native MLucV exhibits higher FRET-PR than separated fluorophores; 4 M urea abolishes activity above 3 M and reduces FRET-PR by ~27% relative to native while fluorophores remain intact. - Controlled small aggregates: Urea- or heat-pretreated MLucV forms discrete, soluble oligomers with high FRET-PR (~150–166% of native) and negligible light scattering, stable over >90 min with no spontaneous refolding. TEM shows roughly spherical particles with diameters ~22.3 ± 3.1 nm (UMLucV) and ~20.5 ± 3.3 nm (HSMLucV). SEC-RALS indicates average mass ~1,500 kDa corresponding to 12 ± 2 MLucV subunits. Aggregate size and FRET plateau across 0.4–4 µM protomer concentrations. - Aggregate architecture: Simulations of 12 misfolded MLucV monomers produce ~17 nm dodecameric particles with luciferase cores packed inside and FPs exposed on the surface, explaining elevated FRET and limited aggregate size by geometric constraints. - Dissection of Hsp70 mechanism: In ATP, DnaJ alone causes only minor (~5%) FRET decrease on UMLucV without activity recovery. Subsequent addition of excess DnaK triggers a dramatic FRET-PR decrease to below the urea-unfolded value, indicating extensive binding and stretching of polypeptides (up to 8 DnaK per monomer) beyond simply unfolded conformations; no activity during this bound, expanded state. Addition of GrpE releases substrates, rapidly recovering native luciferase activity up to ~70% and increasing FRET-PR towards native. Without DnaJ, DnaK fails to bind/unfold aggregates even with ATP and GrpE. TEM corroborates rapid disaggregation: ~55% particle loss within 5 min of KJE + ATP (from ~147 to ~67 particles per µm²), with remaining particles showing fuzzy edges. - Specificity for non-native substrates: Order-of-addition on native MLucV shows no significant FRET or activity changes upon DnaJ, DnaK, or GrpE addition in ATP, indicating selective targeting of misfolded species while sparing native conformers. - MD monomer analyses: Distance distributions between mTFP1 and Venus distinguish urea-unfolded and DnaK-expanded states (larger distances) from native/misfolded compact states. DnaK-bound monomers exhibit even larger distances than urea-unfolded, consistent with experimental very low FRET-PR during DnaK binding. - Non-equilibrium refolding at 38 °C: Native MLucV loses activity at ~10% min⁻¹, reaching >95% inactivation by 35 min. KJE without ATP does not alter denaturation. With KJE + ATP, native MLucV accumulates against equilibrium at ~2% min⁻¹; with 6.4 mM ATP, ~60% of initial activity is restored and sustained >1 h at 38 °C; with 0.4 mM ATP, <40% transient recovery followed by re-inactivation as ATP depletes. ATP pulses cause repeated disaggregation (FRET decrease) and reactivation; without ATP or with ADP, KJE does not prevent aggregation (FRET rises to ~130%) nor maintain activity, arguing against passive holdase function. - Chaperone dependence: DnaJ concentration strongly influences DnaK-mediated refolding yields; simulations indicate DnaJ can access misfolded luciferase segments on aggregate surfaces despite FP domains. - In vivo sensing: MLucV reports misfolding in E. coli during 39 °C heat shock by increased FRET and decreased activity, paralleling in vitro behavior.

The study resolves key steps by which Hsp70 (DnaK) with co-chaperones DnaJ and GrpE remodels stable protein aggregates. Using the MLucV reporter, the authors show that J-domain proteins selectively target DnaK to non-native substrates, enabling ATP-fueled binding that actively disaggregates and unfolds misfolded polypeptides into high-free-energy, highly extended intermediates. GrpE-driven nucleotide exchange and substrate release then allow spontaneous collapse into native states. This mechanistic sequence explains how Hsp70s can convert thermodynamically stable aggregates into metastable native proteins and accumulate native populations under conditions unfavorable to the native state. The data rebut a purely passive holdase model for Hsp70/Hsp40 under stress, emphasizing the requirement for iterative ATP-driven cycles that produce ultra-affinity and non-equilibrium stabilization. The Hsp70 system alone, without ClpB, functions as a bona fide disaggregase in this context. The approach bridges ensemble FRET/activity readouts with particle-level imaging and computational modeling, offering a comprehensive view that clarifies prior ambiguity arising from heterogeneous substrate ensembles.

This work introduces MLucV, a robust multi-domain fluorescent reporter that distinguishes native, unfolded, aggregated, and Hsp70-bound states, enabling precise dissection of Hsp70 mechanisms. The authors demonstrate that Hsp70 (with DnaJ and GrpE) actively disaggregates and unfolds stable aggregates using ATP, stretching substrates beyond urea-unfolded conformations before release and native refolding. The system can non-equilibriumly accumulate native protein at elevated temperatures adverse to the native state, excluding a primary role for passive holdase action. The findings establish Hsp70 as a stand-alone disaggregation and unfoldase machinery capable of converting stable aggregates into metastable, functional proteins. Future directions include: extending the reporter strategy to diverse client proteins and cellular compartments; quantifying chaperone stoichiometry and kinetics across substrates; dissecting nucleotide-state dynamics and ADP/ATP balance in vivo; and integrating additional chaperone systems (Hsp90, Hsp60, Hsp100) to map cooperative roles under varying stress conditions.

- Model substrate: Conclusions are based on an engineered MLucV fusion; while designed to minimally perturb function, generalizability to all substrates may vary. - Fluorophore effects: Small temperature-dependent changes in mTFP1/Venus emission at 39 °C cause up to ~10% underestimation of FRET ratios, potentially affecting precise quantification under heat stress. - Incomplete full recovery: Aggregate reactivation in order-of-addition experiments did not reach 100% within the measured timeframe; the authors note uncertainty between prolonged DnaK binding (increased ADP) versus slow reactivation, leaving full resolution to future studies. - Aggregate specificity: The discrete ~12-mer aggregates arise from MLucV’s architecture; natural aggregates are often more heterogeneous, which may influence chaperone action. - Nucleotide dependence: Non-equilibrium stabilization depends on sustained ATP; physiological ATP/ADP dynamics and their impact on cycles in vivo remain to be fully quantified.