Surfactants or scaffolds? RNAs of varying lengths control the thermodynamic stability of condensates differently

Abstract

Biomolecular condensates, thought to form via liquid-liquid phase separation of intracellular mixtures, are multicomponent systems that can include diverse types of proteins and RNAs. RNA is a critical modulator of RNA-protein condensate stability, as it induces an RNA concentration-dependent reentrant phase transition-increasing stability at low RNA concentrations and decreasing it at high concentrations. Beyond concentration, RNAs inside condensates can be heterogeneous in length, sequence, and structure. Here, we use multiscale simulations to understand how different RNA parameters interact with one another to modulate the properties of RNA-protein condensates. To do so, we perform residue/nucleotide resolution coarse-grained molecular dynamics simulations of multicomponent RNA-protein condensates containing RNAs of different lengths and concentrations, and either FUS or PR₂₅ proteins. Our simulations reveal that RNA length regulates the reentrant phase behavior of RNA-protein condensates: increasing RNA length sensitively rises the maximum value that the critical temperature of the mixture reaches, and the maximum concentration of RNA that the condensate can incorporate before beginning to become unstable. Strikingly, RNAs of different lengths are organized heterogeneously inside condensates, which allows them to enhance condensate stability via two distinct mechanisms: shorter RNA chains accumulate at the condensate's surface acting as natural biomolecular surfactants, while longer RNA chains concentrate inside the core to saturate their bonds and enhance the density of molecular connections in the condensate. Using a patchy particle model, we additionally demonstrate that the combined impact of RNA length and concentration on condensate properties is dictated by the valency, binding affinity, and polymer length of the various biomolecules involved. Our results postulate that diversity on RNA parameters within condensates allows RNAs to increase condensate stability by fulfilling two different criteria: maximizing enthalpic gain and minimizing interfacial free energy; hence, RNA diversity should be considered when assessing the impact of RNA on biomolecular condensates regulation.

Figure 1

Reentrant phase behavior driven by RNA is regulated by both concentration and length. (a) Residue resolution coarse-grained simulations with the Mpipi model (73) to investigate phase separation of RNA–protein mixtures. Coarse-grained representation of (full sequence) FUS, PR₂₅, and a 400-nt polyU RNA strand using the Mpipi model (73) in which each amino acid or nucleotide is represented by a single bead. Please note that the size of the beads depicted in this panel has been conveniently rescaled for visualization purposes. In FUS protein, beads of different colors indicate different protein domains. Direct coexistence simulations of polyU-PR₂₅ (top) and FUS-polyU (bottom) are also included. (b) Comparison of the predicted condensate densities as a function of temperature (renormalized by the highest critical temperature) for polyU–PR₂₅ mixtures composed by polyU strands of 400 nt (black symbols) and 20 nt (blue symbols) using direct coexistence simulations (solid circles) and bulk NpT simulations (solid squares). The estimated critical temperature of each system by both ensembles is depicted by empty symbols of the corresponding shape and color. Snapshots of a direct coexistence simulation and a bulk NpT simulation are included to illustrate the analogy between both ensembles when describing the system condensed phase. Continuous lines represent the phase diagram coexistence lines. (c) Normalized critical temperature of polyU-PR₂₅ mixtures as a function of the U/PR₂₅ mass ratio for different polyU strand lengths as indicated in the legend. (d) Normalized critical temperature of FUS-polyU mixtures as a function of the U/FUS mass ratio for different polyU strand lengths as indicated in the legend. Dashed lines connecting the critical temperatures as a function of concentration are included as a visual guide. While in (c) all temperatures have been normalized by the highest T at which phase separation was observed (T = 435 K), in (d) all temperatures have been normalized by the critical temperature of pure FUS (T_c,FUS = 365 K). Please note that higher critical temperatures in our model correspond to higher driving forces to undergo LLPS (i.e., lower saturation concentration). To see this figure in color, go online.

Figure 2

RNA-driven reentrant phase behavior of FUS and PR₂₅ polyU systems including mixtures with different RNA strand lengths. (a) Representation of PR₂₅, FUS, and two RNA strands of 50 and 400 nt each following the same color code and considerations discussed in Fig. 1a. Please note that the size of the beads depicted in this panel has been conveniently rescaled for visualization purposes. (b) Normalized critical temperature of polyU-PR₂₅ mixtures as a function of the U/PR₂₅ mass ratio for different polyU strand lengths as indicated in the legend. (c) Normalized critical temperature of FUS-polyU mixtures as a function of the U/FUS mass ratio for different polyU strand lengths as indicated in the legend. For the systems with mixed polyU lengths, each length represents half of the total polyU concentration. While in (b) all temperatures have been normalized by the highest T at which phase separation was observed (T = 425 K), in (c) all temperatures have been normalized by the critical temperature of pure FUS (T_c,FUS). To see this figure in color, go online.

Fgure 3

Structural condensate organization of RNA-binding proteins in the presence of long and short RNA strands. (a) PolyU–PR₂₅ mixture with a 0.8 U/PR₂₅ mass ratio where polyU strands are 50 and 400 nt long (each length contributing half to the total polyU concentration). Top: representative snapshot of a direct coexistence simulation of the system, where PR₂₅ molecules are colored in red and long and short RNAs are depicted in blue and cyan, respectively. Middle: density profile of PR₂₅ (red) and polyU RNA (black) along the long axis of the simulation box. Bottom: RNA density profile decomposed in 400-nt (blue) and 50-nt (cyan) polyU chains. (b) The same as in (a), but for a polyU–PR₂₅ mixture with a U/PR₂₅ mass ratio of 0.8 and polyU strands of 400 nt only. Note that we only show one density profile since here all RNAs are of equal length. (c) The same as in (a), but for a polyU–PR₂₅ mixture with a U/PR₂₅ mass ratio of 1.6, where polyU strands are also 50 and 400 nt long. (d) The same as in (b), but for a polyU-PR₂₅ mixture with a U/PR₂₅ mass ratio of 1.6 and polyU strands of 400 nt only. The temperature of systems shown in (a–d) was 0.9 with respect to their corresponding critical temperature. (e) RNA–FUS mixture at T/T_c,FUS = 0.98 and with a U/FUS mass ratio of 0.14, where polyU RNA strands are 50 and 400 nt long (each length contributing half to the total polyU concentration). Top: representative snapshot of a direct coexistence simulation. Middle: density profile of FUS (green) and RNA (black) along the long axis of the simulation box. Bottom: RNA density profile decomposed in 400-nt (blue) and 50-nt (cyan) polyU chains. (f) Surface tension for different polyU–PR₂₅ mixtures, all of them at a temperature of 0.85 with respect to its corresponding critical temperature for the system indicated in the legend. The high RNA concentration corresponds to a U/PR₂₅ mass ratio of 1.6, while the low RNA concentration corresponds to a U/PR₂₅ mass ratio of 0.8. To see this figure in color, go online.

Figure 4

Minimal coarse-grained model for protein LLPS. (a) Green and red spheres represent the excluded volume of scaffold and cognate proteins, respectively, while gray patches represent the binding sites of the proteins. Two different proteins are modeled: scaffold proteins, with three binding sites in a planar equidistant arrangement, and cognate proteins, with two binding sites in a polar arrangement. Blue spherical beads account for ∼5 nt each in the RNA model. Please note that, for visualization purposes, the size of the RNA beads has been scaled down. For further technical details on the model, please see the supporting material. (b) Normalized critical temperature of RNA–cognate protein mixtures as a function of the RNA/cognate protein ratio for different RNA strand lengths as indicated in the legend. (c) Normalized critical temperature of RNA–scaffold protein mixtures as a function of the RNA/scaffold protein ratio for different RNA strand lengths as indicated in the legend. While in (b) all temperatures have been normalized by the highest T (T^∗ = 0.11) at which phase separation was observed, in (c) all temperatures have been normalized by the critical temperature of the scaffold protein in the absence of RNA ($T_c^{\ast }$ = 0.09). To see this figure in color, go online.