An overview of the Evening Complex (EC), key features of ELF3, and the computational approaches taken to elucidate the molecular mechanisms of temperature sensing of the ELF3-PrD.

A. An illustration of the temperature-responsive mechanism of the evening complex. The left panel shows the intact EC repressing a growth gene target. Second to left illustrates the complex dissociating from DNA at high temperature. The next section describes the prion-like domain required for condensation and the poly-glutamine region responsible for tuning the sensitivity of the temperature response. The right-most cartoon illustrates the enhanced growth response characteristic of expanded poly-glutamine tracts. B. Helical propensity of hierarchical chain growth (HCG) ensembles by residue at four temperatures. The 7Q (wild-type) system is shown in the main panel with a region of interest of 7Q and 0Q shown in the inset. The light blue regions mark the locations of the poly-glutamine tracts and the yellow region highlights an aromatic helix of mechanistic importance. C. A structure of wild-type ELF3-PrD taken from a REST2 simulation. D. A charge fraction plot commonly used to classify IDPs based on sequence. The red dot indicates ELF3-PrD falls in region 1, classifying it as a weak polyampholite characterized by a globular or ’tadpole-like’ shape. E. An illustration of the hierarchical chain growth method used to produce ELF3-PrD ensembles at a variety of polyQ lengths and temperature conditions. F. A cartoon representation of the REST2 enhanced sampling simulation method. G. A snapshot from a coarse-grained Martini simulation used here to investigate temperature-dependent condensation propensity of ELF3-PrD variants.

REST2 simulations reveal local and global conformation changes in response to temperature.

A. Helical propensity of WT REST2 simulation at a range of temperatures. The variable polyQ tract is shaded in blue and the two polyQ tracts of consistent length are shaded in purple. A lengend denoting regions of interest is located at the top. B. Helical propensity of F527A REST2 simulation at a range of temperatures. The red line indicates the position of residue 527. C. Helical propensity of 0Q REST2 simulation at four temperatures. The gray shaded region indicates the removal of the variable polyQ tract. D. Top: Solvent-accessible surface area of WT and F527A systems taken from REST2 trajectories. E. The radius of gyration of the first 100 residues (black) and last 73 residues (grey) of ELF3-PrD. Top panel depicts and the WT and the bottom F527A, both at 290K.

Temperature-responsive breaking of aromatic interactions exposes sticky aromatic residues to the solvent.

A. The minimum distance between residue F527 and 𝐻𝑎𝑟𝑜 is plotted by temperature from the WT REST2 simulation. Representative low and high temperature structures are inside of a dashed blue or red box, respectively. F527 is in red and 𝐻𝑎𝑟𝑜 is in yellow. B. Free energy landscapes of WT (left) and F527A (right) with the X-axis defined as the minimum distance between residue 527 and 𝐻𝑎𝑟𝑜 and the y-axis defined as the radius of gyration. Each ensemble is represented at 290K and 320K. The blue box is the most populated region of conformation space at low temperature and the red box demarcates the emergent high temperature cluster. C. ELViM plots illustrating the conformation space of the full ensemble (left), the 290K ensemble (middle) and the 320K ensemble (right) where the color of each dot represents the minimum distance between F527 and 𝐻𝑎𝑟𝑜. Below are example clusters of conformers with the left representing a commonly populated area at low temperature and to the right a cluster only observed at increased temperature. D. Fold increase in SASA between the identified lowand high-temperature populations. Dotted red lines represent aromatic residues and the yellow-shaded region are residues of 𝐻𝑎𝑟𝑜.

Cluster formation and monomer conformation change are driven by similar aromatic residues.

A. The solvent-accessible surface area is shown for each HCG polyQ condition at 290K. The variable polyQ tract is highlighted in sky blue and the second and third polyQ tracts in a lighter blue. The black dashed line demarcates the average SASA value for the system. B. Per-residue hydration plot of ELF3-PrD WT at 290K (blue) and 405K (red). Each value represents the maximum height of the first peak of an RDF between the given residue and solvent oxygen atoms. The aromatic helix, 𝐻𝑎𝑟𝑜, is highlighted in yellow and the polyQ tracts in sky blue. C. A scatter plot relating change in hydrogen bonds between the variable polyQ tract and water with the change in hydrogen bonds between polyQ and protein from the wild-type REST2 simulation at 290K. Breaking of polyQ-water hydrogen bonds is correlated with an increase in polyQ-protein bonds.

Temperature-dependent conformation change enhances SASA of CG monomers and promotes cluster formation driven by central aromatic residues.

A. The intra-protein contact difference map between the 340K and 290K WT Martini condensate simulations. Values in red represent increases in contact value at 340K relative to 290K and blue represents a decrease. The entire 3 𝜇s trajectories were used to create these difference maps. B. Distance distributions between residue F527 and F458 at 290K (grey) and 340K (red) from WT CG condensate simulations. Distances are taken from intra-protein distances of each monomer. C. The solvent accessible surface area is plotted for all 100 monomers of both the 290K WT condensate simulation (black) and the 340K WT condensate simulation (red). D. The inter-protein contact difference map between the 340K and 290K WT Martini condensate simulations. As with Fig. 5A, contacts enhanced at high temperature are in red with those broken in blue. E. An illustration of the typical time evolution of an ELF3-PrD condensate simulation at 300K with simulation snapshots representing approximately 0 ns, 800 ns, and 3𝜇s, respectively.

Martini 3 simulations reveal temperature- and variant-dependent condensate stability and internal dynamics of ELF3-PrD.

Please check “Analysis of Martini Condensate Simulations” for more details of the methods A. Time evolution of the # of clusters at 290 K for 7Q (black), 0Q (red), 19Q (green), and F527A (orange); shaded regions show variability across independent replicas. Representative simulation snapshots illustrate coarsening from many small clusters to fewer larger assemblies. B. Condensate stability quantified as the mean size of the largest cluster (number of peptides) as a function of temperature for each variant (mean ± SEM calculated over the last 1 𝜇𝑠 of three independent replicas). The dashed line indicates the threshold used to define a condensed state (𝑁max > 50). C. Internal mobility within the condensed phase quantified by the effective diffusion coefficient 𝐷extracted from peptide mean-squared displacement analysis (mean ± SEM across three independent replicas), shown as a function of temperature for each variant. D. Desolvation of the variable polyQ region during condensation illustrated for 7Q at 300 K: the number of water contacts to the polyQ tract decreases (black) as the number of protein contacts increases (red) over time. E. Residue-resolved change in inter-protein contacts for wild-type (7Q) at 300 K relative to 290 K (ΔNet Contacts), highlighting regions spanning the aromatic helix (𝐻aro), F527, and the variable/additional polyQ segments. Aromatic residues are marked by dashed red lines.