Sequence-encoded and composition-dependent protein-RNA interactions control multiphasic condensate morphologies

Multivalent protein-protein and protein-RNA interactions are the drivers of biological phase separation. Biomolecular condensates typically contain a dense network of multiple proteins and RNAs, and their competing molecular interactions play key roles in regulating the condensate composition and structure. Employing a ternary system comprising of a prion-like polypeptide (PLP), arginine-rich polypeptide (RRP), and RNA, we show that competition between the PLP and RNA for a single shared partner, the RRP, leads to RNA-induced demixing of PLP-RRP condensates into stable coexisting phases—homotypic PLP condensates and heterotypic RRP-RNA condensates. The morphology of these biphasic condensates (non-engulfing/ partial engulfing/ complete engulfing) is determined by the RNA-to-RRP stoichiometry and the hierarchy of intermolecular interactions, providing a glimpse of the broad range of multiphasic patterns that are accessible to these condensates. Our findings provide a minimal set of physical rules that govern the composition and spatial organization of multicomponent and multiphasic biomolecular condensates.

, main-text, but plotted against tyrosine concentrations in PLP and arginine-to-tyrosine ratio.The shaded green region represents the phase-separation regime and is drawn as a guide to the eye.(b) A plot of the ratio of LLPS concentration threshold for PLP with ligand ( ) and without ligand ( ) as a function of [RRP] concentration (ligand = RRP).The data points are estimated from the state diagram analysis as shown in Figure 1a, main-text.At a fixed concentration of PLP (any horizontal line across Figure 1a main-text), the ligand concentration required for LLPS was estimated as the mid-point between the LLPS (green) and No LLPS (black) transition points.Here, the error bars are equal to half of the range between the two points that define a transition between mixed regime (no LLPS) and phase separation regime (LLPS).The black points represent the data, while the red line is drawn as a guide to the eye.RRP = [RGRGG]5.and high (b) RNA-to-RRP mixing ratios.FUS FL accumulates at the surface of RRP-RNA condensates in both conditions due to its ability to interact with RRP chains (through PLP-RRP interactions) and RNA chains (through RBD-RNA interactions, see Fig. 5e, main-text).RNA is visualized as blue chains, RRP is visualized as red chains, and FUS FL is visualized as green chains.For both simulations,  =1.3 mg/ml,  =0.7 mg/ml and RNA-to-RRP ratio (wt/wt) of 0.5 (left) and 2.0 (right).

Figure
Figure S2.[RGRGG]5-PLP isothermal state diagram.(a) A state diagram for [RGRGG]5-PLP mixtures.This is identical data to Figure 1a, main-text, but plotted against tyrosine concentrations in PLP and arginine-to-tyrosine ratio.The shaded green region represents the phase-separation regime and is drawn as a guide to the eye.(b) A plot of the ratio of LLPS concentration threshold for PLP with ligand ( ) and without ligand ( ) as a function of [RRP] concentration (ligand = RRP).The data points are estimated from the state diagram analysis as shown in Figure1a, main-text.At a fixed concentration of PLP (any horizontal line across Figure1amain-text), the ligand concentration required for LLPS was estimated as the mid-point between the LLPS (green) and No LLPS (black) transition points.Here, the error bars are equal to half of the range between the two points that define a transition between mixed regime (no LLPS) and phase separation regime (LLPS).The black points represent the data, while the red line is drawn as a guide to the eye.RRP = [RGRGG]5.

Figure S3 .
Figure S3.Representative experimental data for FRAP experiments on PLP-RRP condensates.Time-lapse FRAP images (left) and the corresponding intensity time traces (right) for PLP-RRP condensates prepared at a fixed FUS PLP concentration of 280 µM and variable [RGRGG]5-to-FUS PLP ratios.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.The yellow dashed circle indicates the predetermined bleaching region.Scale bars are 5 µm.Bleaching occurs at t=3s.The microscopy images are representative of at least three FRAP events from different spots in the same sample.The FRAP experiment was performed utilizing ~1% (labeled: unlabeled ratio) Alexa488-labeled PLP.

Figure
Figure S4.FUS PLP partition and apparent diffusion are altered with increasing RRP concentration.(a) Multicolor confocal fluorescence microscopy images for PLP-RRP condensates at a variable RRP-to-PLP mixing ratio.We note that the difference in size between PLP droplets and PLP-RRP droplets may be a consequence of RRP-PLP condensates having excess charge due to the presence of Arg residues.For all samples, PLP concentration is fixed at 280 µM and RRP ([RGRGG]5) concentration was varied.Scale bars represent 20 µm.~500 nM Alexa488-labeled PLP and ~500 nM Alexa594-labeled [RGRGG]5 were used for visualization.The reported images are representative of several replicates imaged from different spots in the same sample.(b) A plot showing PLP partition coefficient (K; n = 60 droplets per sample) and apparent diffusion coefficient (Dapp; n = 3 droplets per sample) as a function of RRP-to-PLP mixing ratio.Error bars represent ± 1 s.d and the central filled circle represents the mean value.(see the statistical analysis section and Fig. 1b&c, main-text).Source data are provided as a Source Data file.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure S5 .
Figure S5.State diagram analysis and Fluorescence Correlation Spectroscopy (FCS) for PLP-RNA mixture.(a) State diagram of PLP-RNA mixtures, showing that poly(rU) RNA does not affect PLP phase-separation.Shaded pink region denotes PLP homotypic phase separation regime (PLP saturation concentration: Csat ~ 240 µM).(b) The normalized autocorrelation curve for FUS PLP in the presence (red) and absence of RNA (black).The time scale at which the autocorrelation reaches zero is proportional to the diffusion time of the labeled molecules.PLP shows identical auto-correlation time-scale both in the presence and absence of RNA, indicating that PLP is not forming a complex with RNA (which would slow the diffusion and therefore would change the autocorrelation timescale).The sample contained [Alexa488-labeled FUS PLP ]= 50 nM (0.88 ng/ml) with 0.0 ng/ml RNA poly(rU) (black) and 7.1 ng/ml RNA poly(rU) (red).The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl.

Figure S6 .
Figure S6.An RNA oligomer, rU10 remains excluded from PLP condensates.Multicolor confocal fluorescence microscopy images and intensity profiles (across yellow dashed lines) (a) and partition coefficients box plot (b) showing that FAM-labeled rU10 is excluded out of PLP droplets at the tested poly(rU)-to-PLP ratio.PLP condensates were prepared at PLP = 280 μM (with ~ 1% Cy-5-labeled PLP) and varying poly(rU)-to-PLP ratio.The number of droplets (n) analyzed across different samples for partition is n = 35.Source data are provided as a Source Data file.The microscopy images in (a) are representative of two independent sample replicates.Scale bars represent 5 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5) and 150 mM NaCl.PLP = FUS PLP .

Figure
Figure S7.FUS RGG3 -poly(rU) isothermal state diagram.State diagram for FUS RGG3 -poly(rU) mixtures.This is identical data to Figure 1d, main-text, but plotted against arginine concentrations and nucleotide-to-arginine ratio.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.The shaded red region represents the phase-separation regime and is drawn as a guide to the eye.

Figure S8 .
Figure S8.Density profiles of RNA and RRP from MD simulations.Density profiles for the RRP (FUS RGG3 ) and poly(rU) RNA across RRP-RNA condensates from MD simulations at (a)  <  (b) and  >  .These profiles correspond to the MD configurations shown in Figure 1d in the main-text.For both simulations,  = 1.3 mg/ml and the RNA-to-RRP (wt/wt) ratio is 0.5 for (a) and 1.7 for (b).

Figure S9 .
Figure S9.Density profiles of RNA, RRP, and PLP from MD simulations.Equilibrium configurations from MD simulations and the corresponding density profiles for the RRP (FUS RGG3 ), PLP (FUS PLP ), and poly(rU) RNA across RRP-RNA condensates at (a)  <  and (b)  >  .The recruitment of PLP is enhanced at  <  with a visible localization of PLP chains on the surface of the RRP-RNA condensates.For both simulations,  = 1.3 mg/ml,  = 0.4 mg/ml and the RNA-to-RRP ratio (wt/wt) is 0.5 for (a) and 1.7 for (b).

Figure
Figure S10.EWS PLP preferentially partitions into the surface of RRP-rich RRP-RNA condensates.Multicolor confocal fluorescence microscopy images (a) and partition coefficients box plot (b) showing that EWS PLP (labeled with Alexa488) is recruited into RNA-RRP [poly(rU)-FUS RGG3 ] droplets at low RNA-to-RRP ratio while at high RNA-to-RRP ratio, PLP partitioning significantly decreases.poly(rU)-FUS RGG3 condensates were prepared at FUS RGG3 =1 mg/ml (with ~ 1% Alexa594-labeled peptide) and varying poly(rU)-to-FUS RGG3 ratio.The number of droplets (n) analyzed across different samples for partition is n = 100.Source data are provided as a Source Data file.Scale bars represent 10 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure
Figure S11.FUS PLP shows preferential partitioning into RRP-rich RRP-RNA condensates.Multicolor confocal fluorescence microscopy images (a) and partition coefficient box plot (b) showing that FUS PLP is recruited into RNA-RRP [poly(rU)-FUS RGG3 ] droplets at low RNA-to-RRP ratio while at high RNA-to-RRP, PLP (labeled with Alexa488) partitioning significantly decreases.poly(rU)-FUS RGG3 condensates were prepared at FUS RGG3 = 1 mg/ml (with ~ 1% labeled:unlabeled ratio of Alexa594-FUS RGG3 ) and varying poly(rU)-to-FUS RGG3 ratio.The number of droplets (n) analyzed across different samples for partition coefficient calculation is n = 50.Source data are provided as a Source Data file.Scale bars represent 10 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure
Figure S12.BRG1 LCD shows preferential partitioning into RRP-rich RRP-RNA condensates.Multicolor confocal fluorescence microscopy images (a) and partition coefficient box plot (b) showing that BRG1 LCD is recruited into RNA-RRP [poly(rU)-FUS RGG3 ] droplets at low RNA-to-RRP ratio while at high RNA-to-RRP, BRG1 LCD (labeled with Alexa488) does not show any preferential partitioning.Poly(rU)-FUS RGG3 condensates were prepared at FUS RGG3 =1 mg/ml (with ~ 1% labeled:unlabeled Alexa594-FUS RGG3 ) and varying poly(rU)-to-FUS RGG3 ratio.The number of droplets (n) analyzed across different samples for partition coefficient calculation is n = 75.Source data are provided as a Source Data file.Scale bars represent 10 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure
Figure S13.RNA Pol II CTD preferentially partitions into RRP-rich RRP-RNA condensates.Multicolor confocal fluorescence microscopy images (a) and partition coefficients box plot (b) showing that Pol II CTD (labeled with Alexa488) is recruited into RRP-RNA [poly(rU)-FUS RGG3 ] droplets at low RNA-to-RRP ratio while at high RNA-to-RRP, Pol II CTD shows relatively lower partitioning.poly(rU)-FUS RGG3 condensates were prepared at FUS RGG3 = 1 mg/ml (with ~1% labeled:unlabeled Alexa594-FUS RGG3 ) and varying poly(rU)-to-FUS RGG3 ratio, as indicated.The number of droplets (n) analyzed across different samples for partition coefficient calculation is n = 45.Source data are provided as a Source Data file.Scale bars represent 10 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure S14 .
Figure S14.Surface versus core recruitment of PLP clients in RRP-RNA condensates.Box plot showing the ratio of surface and core intensity of PLP clients (EWS PLP , FUS PLP , and BRG1 LCD ) within RRP-RNA [poly(rU)-FUS RGG3 ] droplets at low RNA-to-RRP ratio.poly(rU)-FUS RGG3 condensates were prepared at FUS RGG3 = 1 mg/ml and poly(rU)-to-FUS RGG3 ratio of 0.2 (wt/wt).The number of droplets (n) analyzed across different samples is n = 70.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT. Source data are provided as a Source Data file.

Figure
Figure S15.FUS PLP shows preferential partitioning into RRP-rich RRP-poly(phosphate) condensates.Multicolor confocal fluorescence microscopy images (a), intensity profile (b), and partition coefficient box plot (c) showing that FUS PLP is recruited into poly(P)-RRP [poly(P)-FUS RGG3 ] droplets at low poly(P)-to-RRP ratio while at high poly(P)-to-RRP, PLP (labeled with Alexa488) partitioning significantly decreases.poly(P)-FUS RGG3 condensates were prepared at FUS RGG3 = 1 mg/ml (with ~ 1% labeled:unlabeled ratio of Alexa594-FUS RGG3 ) and varying poly(P)to-FUS RGG3 ratio.(d) Box plot showing the ratio of surface and core intensity of FUS PLP within RRP-poly(P) [poly(P)-FUS RGG3 ] droplets at poly(P)-to-RRP ratio of 0.02 (wt/wt).(e) Turbidity at 350 nm for FUS RGG3 -poly(P) mixtures prepared at FUS RGG3 concentration of 1.0 mg/ml and variable poly(P) concentrations.The intensity profile in (b) is for poly(P)-to-RRP ratio of 0.02.The number of droplets (n) analyzed across different samples for box plots (c&d) is n = 50.Scale bars represent 10 µm.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT. Source data are provided as a Source Data file for c&d.

Figure
Figure S16.RNA mediated PLP-RRP demixing behavior is not impacted by sample volume change.(a) Time-lapse microscopy images after the addition of buffer to a sample containing PLP-RRP condensates showing that the PLP-RRP droplets are not affected by the volume change.The sample was prepared at [FUS PLP ] =250 µM and [FUS RGG3 ]=750 µM (2.6 mg/ml). 1 µL of buffer was added to the 4 µL droplet sample.(b) Time-lapse microscopy images after the addition of poly(rU) RNA to a sample containing PLP-RRP condensates showing the sequestering of RRP (i.e.FUS RGG3 ) from the PLP-RRP droplets.The sample was prepared at identical concentrations to those shown in (a).0.3 µL of poly(rU) stock solution was added to the 4 µL droplet sample, resulting in a 6.5 mg/ml final concentration of poly(rU) RNA.~ 500 nM Alexa488labeled FUS RGG3 was used for imaging.Scale bar = 20 µm.The microscopy images (a&b) are representative of two independent sample replicates.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure
Figure S22.A simple illustration of the fluid-interface modeling simulation and density profiles for MD simulations.(a) Time evolution of a cube of liquid with 50 mN/m surface tension using Surface Evolver.The minimization of interfacial energy leads to the transition from a cube to a spherical droplet, which is the geometry with the least surface energy.(b) Density profiles for the molecular dynamics simulation snapshots are shown in Figure 4f in the main text.For simulation details, see the legend of Figure 4f in the main text.

Figure S23 .
Figure S23.Representative FRAP images and intensity time traces for PLP-KRP condensates.Time-lapse FRAP images (left) and intensity time traces (right) for PLP-KRP condensates prepared at a fixed PLP concentration and variable KRP ([KGKGG]5) to PLP ratio.For all samples, FUS PLP concentration is fixed at 280 µM.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT. Scale bars represent 5µm.Bleaching occurs at t=3s.The microscopy images are representative of at least three FRAP events from different spots in the same sample.The imaging/FRAP assay was performed utilizing ~1% Alexa488labeled PLP (labeled: unlabeled ratio).

Figure S24 .
Figure S24.PLP does not partition into KRP-RNA condensates across all mixture compositions.Multicolor confocal fluorescence microscopy images (a) and partition coefficient box plot (b) of Alexa488-labeled FUS PLP in KRP-RNA droplets.KRP-RNA condensates were prepared at a fixed KRP ([KGKGG]5) concentration of 1 mg/ml and variable RNA [poly(rU)] to KRP ratio, as indicated.Scale bars represent 10 µm.The number of droplets (n) analyzed across different samples for partition is n = 150.Source data are provided as a Source Data file.The sample buffer contains 25 mM Tris-HCl (pH 7.5), 150 mM NaCl and 20 mM DTT.

Figure
Figure S25.A tyrosine-variant of KRP restores PLP binding ability and stabilizes a shared fluid-fluid interface.(a) Isothermal state diagram of FUS PLP -[KGYGG]5 mixtures showing that [KGYGG]5 facilitates PLP phase-separation.The shaded green region shows the heterotypic phase separation regime for PLP-[KGYGG]5 mixtures while the shaded pink region denotes the PLP homotypic phase separation regime (saturation concentration ~240 µM).Both shaded regions are drawn as a guide to the eye.All samples were prepared in a 25 mM Tris-HCl, 150 mM NaCl and 20 mM DTT buffer.(b) Multicolor confocal fluorescence microscopy images and intensity profile for co-existing homotypic FUS PLP droplets and heterotypic [KGYGG]5-RNA droplets.Each type of droplet was separately prepared at initial concentrations of [FUS PLP ]=400 μM , [KGYGG]5=4 mg/ml and [poly(rU)]=0.4mg/ml and then mixed (1:1 vol/vol).The reported images are representative of several replicates imaged from different spots in the same sample.For imaging, 1% Alexa594-labeled [KGYGG]5 and Alexa488-labeled FUS PLP were used (labeled: unlabeled ratio).Scale bar represents 5 μm.

Figure S26 .
Figure S26.Full-length FUS (FUS FL ) partitions into RRP-RNA condensates across all mixture compositions.Multicolor confocal fluorescence microscopy images (a) and partition coefficient box plot (b) showing the partition of FUS FL (labeled with Alexa488) in RNA-RRP droplets at varying RNA-to-RRP ratio.RNA-RRP condensates were prepared at FUS RGG3 =1 mg/ml (labeled with Alexa594) and varying RNA [poly(rU)] to FUS RGG3 ratio.The number of droplets (n) analyzed across different samples for partition is n = 80.Source data are provided as a Source Data file.Scale bars represent 5 µm.The samples were prepared in a buffer containing 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 20 mM DTT. Compare this data with Figure 1h&k in the main-text.

Figure S27 .
Figure S27.Molecular dynamics simulation snapshots showing the partition of FUS FL in RRP-RNA droplets.Representative equilibrium configurations and corresponding density profiles obtained from molecular dynamics simulation of RRP-RNA condensates at both low (a)and high (b) RNA-to-RRP mixing ratios.FUS FL accumulates at the surface of RRP-RNA condensates in both conditions due to its ability to interact with RRP chains (through PLP-RRP interactions) and RNA chains (through RBD-RNA interactions, see Fig.5e, main-text).RNA is visualized as blue chains, RRP is visualized as red chains, and FUS FL is visualized as green chains.For both simulations,  =1.3 mg/ml,  =0.7 mg/ml and RNA-to-RRP ratio (wt/wt) of 0.5 (left) and 2.0 (right).

Figure
Figure S28.FUS FL condensates completely engulf RRP-RNA condensates across all mixture compositions.Multicolor confocal fluorescence microscopy images and intensity profiles for co-existing homotypic FUS FL droplets (red; Cy5-labeled FUS PLP ) and heterotypic RRP (green; Alexa594-labeled FUS RGG3 ) and poly(rU) RNA (blue; probed by SYTO13) condensates at different RNA-to-RRP ratio.Each type of droplet was separately prepared at [FUS FL ]= 21.3 µM, [FUS RGG3 ] = 1 mg/ml and varying RNA-to-RRP ratios, as indicated, and then mixed (1:1 vol/vol).All samples were made in a buffer containing 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 20 mM DTT.The microscopy images are representative of two independent sample replicates.All fluorescent probes were added at a 1% labeled: unlabeled ratio.All scale bars represent 5 µm.

Figure S29 .
Figure S29.The Janus-like architecture of FUS-KRP-RNA condensates.Multicolor confocal fluorescence and bright-field microscopy images and intensity profiles for Janus droplets formed by homotypic FUS FL droplets (blue) and heterotypic KRP-RNA condensates (red).Each type of droplet was separately prepared at initial concentrations of [FUS FL ]=22 μM , [KGKGG]5=4 mg/ml and RNA [poly(rU)]=3 mg/ml keeping [KGKGG]5 to poly(rU) ratio at 0.75 (wt/wt) and then mixed (1:1 vol/vol).For imaging, 500 nM Alexa594-labeled [KGKGG]5, 500 nM Alexa488-labeled FUS FL were used for (a); and 500 nM Alexa594-labeled [KGKGG]5, 500 nM Alexa488-labeled FUS PLP were used for (b).The arrow in (a) points to the line separating the two lobes of the Janus droplet (visible in the bright-field channel).All samples were prepared in a buffer containing 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 20 mM DTT.The reported images (a&b) are representative of several replicates imaged from different spots in the same sample.Scale bar represents 10 µm.Inset shows the zoomed-in appearance of a Janus droplet.