Mechanistic Insights Into Oxidative Response of Heat Shock Factor 1 Condensates

Abstract

Heat shock factor 1 (Hsf1), a hub protein in the stress response and cell fate decisions, senses the strength, type, and duration of stress to balance cell survival and death through an unknown mechanism. Recently, changes in the physical property of Hsf1 condensates due to persistent stress have been suggested to trigger apoptosis, highlighting the importance of biological phase separation and transition in cell fate decisions. In this study, the mechanism underlying Hsf1 droplet formation and oxidative response was investigated through 3D refractive index imaging of the internal architecture, corroborated by molecular dynamics simulations and biophysical/biochemical experiments. We found that, in response to oxidative conditions, Hsf1 formed liquid condensates that suppressed its internal mobility. Furthermore, these conditions triggered the hyper-oligomerization of Hsf1, mediated by disulfide bonds and secondary structure stabilization, leading to the formation of dense core particles in the Hsf1 droplet. Collectively, these data demonstrate how the physical property of Hsf1 condensates undergoes an oxidative transition by sensing redox conditions to potentially drive cell fate decisions.

Figure 1

Oxidative response to lower internal-mobile state of Hsf1 droplets in vitro and in cells. (a) Confocal immunofluorescence images showing subcellular localization and foci formation of Hsf1 in HAP1 cells (scale bar, 10 μm). Cells were then costained with antibodies against HSP70 and 4′,6-diamidino-2-phenylindole (DAPI). Cells were treated with various stress conditions: 43 °C heat shock for 1 h, 1.0 mM H₂O₂ for 1 h, and 3.0 mM tert-butyl hydroperoxide (TBHP) for 1 h. (b) Number of Hsf1 foci per cell in the presence of H₂O₂. Cells were treated with the following H₂O₂ concentrations: 0, 0.05, 0.10, 0.15, 0.20, 0.25, and 0.50 mM for 2 h at 37 °C. Data are plotted as means ± s.e. of three or four independent experiments. (c) Differential interference contrast images of Hsf1 droplets in the presence of 10% (w/v) Ficoll 400. Left panels correspond to samples incubated with 10 mM DTT (+ Reductant) and the right panels correspond to samples in the absence of a reductant (− Reductant). Scale bar, 40 μm. (d) FRAP data demonstrating decreased internal-mobility of Hsf1 droplet under oxidative conditions. Bleaching event occurred at 0 s. Data are plotted as means ± s.d., with n = 7 independent experiments. (e) 3D RI images of Hsf1 droplets. Top and bottom panels correspond to the RI images of Hsf1 droplets in the presence or absence of 10 mM DTT (+, – Reductant), respectively. Areas with RI values greater than 1.353 are shown in yellow. (f) Quantitative analysis of RI images, measuring the mean RI of the Hsf1 droplets in the presence of 10 mM DTT (+ Reductant) and in the absence of reductant (− Reductant). (g) Scatter plot of the mean RIs of an Hsf1 droplet in the presence or absence of 10 mM DTT (+,– Reductant) with N = 16 (+ Reductant) and 10 (− Reductant). Each plot represents the mean RI of the average of the individual droplets. **, p < 0.01, statistical significance between the mean RI values in the absence or presence of the reductant.

Figure 2

Enhanced oligomerization of Hsf1 through disulfide bonds under oxidative condition. (a) Oligomeric states of Hsf1 in redox buffers; 5 μM Hsf1 are incubated without reductant (−Reductant), with 10 mM DTT (+ Reductant), or in the presence of redox buffers (GSH: GSSG = 100:1, 3:1, 1:25) at 37 °C. Samples were quenched by addition of NEM and separated using SDS-PAGE on an 8% gel. (b) Quantification of the relative band intensities of the higher-order oligomer forms of Hsf1 compared to the band intensity of the – Reductant lane (white bar graphs), and the monomer forms of Hsf1 compared to the band intensity of the + Reductant lane (black bar graphs) in (a). Error bars correspond to the means ± s.d. of three independent experiments. (c) Domain organization, location of cysteine residues in Hsf1, and Hsf1 disorder prediction derived from the PONDR program. Following abbreviations are used: DBD, DNA-binding domain; LZ1–3, leucine-zipper domain-1-3; RegD, regulatory domain; LZ4, leucine-zipper domain-4; CTAD, C-terminal transactivation domain. (d) SEC-MALS of Hsf1 with 10 mM DTT (+ Reductant) and without reductant (− Reductant), and Hsf1 CS mutant without reductant (CS5 mutant–Reductant), injected at varying concentrations (black: 50 μM, dark gray: 20 μM, and gray: 10 μM). (e) Box and whisker plot of Hsf1 molecular mass under various conditions. (f) Differential interference contrast images of Hsf1 CS5 mutant droplets in the presence of 10% (w/v) Ficoll 400 without a reductant (− Reductant). Left panels correspond to Hsf1 WT, and the right panels correspond to Hsf1 CS5. Scale bar, 40 μm. (g) Quantification of the higher-order oligomer formation of various purified Hsf1 mutants. SDS-PAGE data (shown in Figure S7b) were used to compare the relative signal intensities of the higher-order oligomer forms in the Hsf1 mutant lanes with the signal intensity in the WT lane. Error bars correspond to the means ± s.d. of three independent experiments. n.d.: not detectable. (h) Close-up view of the structure around Cys153 with an electrostatic potential map of Hsf1, generated using the PyMol plugin (vacuum electrostatics).

Figure 3

Oxidative α-helix stabilization promotes the assembly of Hsf1. (a) SRCD spectra of 40 μM each of Hsf1 WT and CS5 in the presence of 18.75% (w/v) dextran 200. (b) Coarse-grained (CG) configurations of the Hsf1 structure derived from AlphaFold2, where the DBD region (residues 1–121) is excluded. MARTINI CG model relies mainly on a four-to-one mapping scheme; that is, on average, four heavy atoms and the associated hydrogen atoms are mapped to one bead. LZ1–3 (residue numbers: 121–207) of Hsf1 is colored orange, LZ4 (residue numbers: 380–411) is colored purple, and the rest is colored gray. (c) Time evolution of Hsf1_helix and Hsf1_flex in 150-mM KCl aqueous solution. K⁺ and Cl^– charged beads are shown as small dots in blue and green, respectively, and water beads represent the surface. (d,e) Close-up view of the internal structure of the Hsf1 flex (d) and helix (e) clusters, with water depicted by the surface model in cyan, showing that the Hsf1 cluster remained well hydrated. (f) Gyration radius of the Hsf1 cluster. Higher value for Hsf1_helix indicates that the cluster is elongated and nonspherical owing to the steric hindrance of the helices. Error bars correspond to the means ± s.d. of three independent simulations. (g) Snapshot of the Hsf1 cluster in the Hsf1_helix model at 1000 ns. The LZ1–3 (orange) and LZ4 (purple) helices are shown in a thickened licorice representation, whereas the others are shown in a line representation. (h) Analysis of Hsf1 diffusion in complex clusters. Short-term mean square displacements of the main-chain beads in the LZ and non-LZ domains were compared between the Hsf1_flex and Hsf1_helix models. Smaller slope of the LZ domain in the Hsf1 helix model indicates a lower diffusivity of the LZ helices compared to that of the coiled LZ domain in the Hsf1_flex and non-LZ domains. Error bars correspond to the means ± s.d. of three independent simulations. (i) Root-mean-square fluctuation for each amino acid residue of the Hsf1_flex (blue) and Hsf1_helix (red) models. Helical regions in LZ1–3 and LZ4 are highlighted in orange and purple, respectively. Error bars correspond to the means ± s.d. of three independent simulations.

Figure 4

Proposed transition mechanism of Hsf1 droplets. Hsf1 condensate undergoes a transition to lower internal-mobile state in a redox-dependent manner. Disulfide bond formation promotes higher-order oligomerization and stabilization of helix. These events enhance the assembly between oligomers, resulting in core particle formation.