(A) Monomer activation model, based on the originally proposed mechanism modified with our HX-MS data. In unstressed cells monomeric Hsf1 is in equilibrium between a closed, HR-C docked to HR-A/B, and open conformation, with HR-C dissociated from HR-A/B. Owing to high local concentration and electrostatic attraction the intramolecular association rate kon,i of the HR-C–HR-A/B interaction are very high as compared to the dissociation rate koff,i. Since only uncomplexed HR-A/B can trimerize and Hsf1 trimerization therefore depends on the concentration of the open conformation, at low temperatures, trimerization only occurs at high Hsf1 concentrations. Temperature-induced unfolding of HR-C in the docked or undocked state reduces the intramolecular association rates and/or increases the dissociation rate of the intramolecular HR-C–HR-A/B complex, thereby increasing the concentration of Hsf1 in the open conformation and allowing trimerization at low Hsf1 concentrations. (B) Dimer activation model. At low temperatures, HR-C is constitutively docked onto HR-A/B and monomeric Hsf1 transiently dimerizes through the free part of HR-A/B. Such transient dimerization may partially destabilize the HR-C–HR-A/B interaction. At high Hsf1 concentrations a third Hsf1 monomer could interact with a transient Hsf1 dimer to form a thermodynamically stable Hsf1 trimer with completely released HR-C even at low temperatures. Increasing temperatures lead to unfolding of HR-C in the dimeric Hsf1 species leading to stabilization of the Hsf1 dimer and increased probability of trimerization. Hsp90 might modulate the temperature response by stabilizing the dimeric Hsf1 species. (C) Estimation of the concentration dependence of the transition temperature of Hsf1. Data points are all the Tm values determined for 10 min incubation at elevated temperatures for Hsf1 wild type in the absence (black) or presence (green) of Hsp90 by HX-MS and by anisotropy. Black curve is a fit of the quadratic solution of the law of mass action of the monomer-dimer equilibrium, assuming that the fraction of dimer determines the Tm. This fit results in a Tm,M for the monomer of Hsf1 (extrapolation to 0 nM) of 53°C, the Tm,D for the dimer of 33°C, and a KD of the monomer-dimer equilibrium of 330 nM. Due to the sensitivity of the fit to data points at low Hsf1 concentrations, these are only rough estimates. The blue and red dotted lines are simulations using a lower value for KD (100 nM, blue) or KD (200 nM) and Tm,D (29°C, red) to simulate the effect of Hsp90. (D) Tentative model of the dimeric Hsf1 based on the recent crystal structure of the trimerization domain of C. thermophilum Skn7, which formed tetramers in two different crystal forms (PDB ID 5D5Y and 5D5Z, [Neudegger et al., 2016]). HR-A, HR-B and HR-C were homology modeled on the tetrameric Skn7 using I-TASSER (Roy et al., 2010; Zhang, 2008; Yang and Zhang, 2015; Yang et al., 2015). HR-C was positioned to accommodate interactions with HR-A and HR-B. The homology model is colored according to HX-MS data (Figure 1E). Residues of the heptad repeat involved in the tetramer interface are shown as sticks.