Biological condensates form percolated networks with molecular motion properties distinctly different from dilute solutions

  1. Zeyu Shen
  2. Bowen Jia
  3. Yang Xu
  4. Jonas Wessén
  5. Tanmoy Pal
  6. Hue Sun Chan
  7. Shengwang Du
  8. Mingjie Zhang  Is a corresponding author
  1. Division of Life Science, Hong Kong University of Science and Technology, ClearWater Bay, Kowloon, China
  2. Department of Biochemistry, University of Toronto, Canada
  3. Department of Physics, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, China
  4. Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, China
  5. Greater Bay Biomedical Innocenter, Shenzhen Bay Laboratory, China
  6. School of Life Sciences, Southern University of Science and Technology, China
6 figures, 1 video, 1 table and 1 additional file

Figures

Figure 1 with 1 supplement
Single-molecule imaging of phase separation on supported lipid bilayers.

(A) Schematic diagram showing phase separation of postsynaptic density (PSD) protein assembly on supported lipid bilayers (SLBs) (Zeng et al., 2018). (B) Upper panel: a TIRF image of Alexa 647 …

Figure 1—figure supplement 1
Determination of boundaries of the condensed phase throughout the imaging process and heterogeneous distribution of molecules in the condensed phase.

(A) Wide-field image merged with phase boundaries that were reconstructed from supper-resolution images (blue curves) showing no obvious phase boundary changes during the time duration of the …

Figure 2 with 4 supplements
Development of an adaptive single-molecule tracking algorithm for imaging single molecules in the condensed and dilute phases simultaneously.

(A) Flowchart of the adaptive single-molecule tracking algorithm. (B) Assignments of motion tracks of NR2B in both condensed and dilute phases in the postsynaptic density (PSD) condensates formed on …

Figure 2—figure supplement 1
Evaluation of the adaptive single-molecule tracking algorithm by simulation and by experiments.

(A) Determination of the optimized search range. Blue line shows the curve for the left-hand side of the equation Xe-X2=πσDt, and green line shows the range for the right-hand side of the same equation in …

Figure 2—figure supplement 2
Simulations of track assignment errors of phase separations with molecules in the condensed phase undergoing homogeneous free diffusions.

Simulated track assignment errors vs. maximum step limit/root mean square displacement (RMSD) ratios under different phase separation conditions. Different colors were used to distinguish different …

Figure 2—figure supplement 3
Simulations of diffusion coefficient errors of phase separations with molecules in the condensed phase undergoing homogeneous free diffusions.
Figure 2—figure supplement 4
Simulations of track assignment errors of phase separations with molecules in the condensed phase containing both confined and mobile states.

Simulated results of track assignment errors vs. maximum step limit/root mean square displacement (RMSD) ratios under different conditions. Different line colors were used to distinguish different …

Figure 3 with 3 supplements
Dynamic parameters and a diffusion model for an equilibrium state phase separation system.

(A) Displacement distribution of tracks in the (A1) condensed and (A2) dilute phases. Zoom-in view in (A1) shows detailed distribution of the distribution tail. Bin size of histogram is 5 nm. (B) …

Figure 3—figure supplement 1
Typical tracks of NR2B only tethered to supported lipid bilayer (SLB) or NR2B in 3D postsynaptic density (PSD) condensates.

(A) Representative tracks of NR2B only tethered to SLB, showing that the molecules undergo homogeneous diffusions on the membrane surface. The labeled color from red to black was used to distinguish …

Figure 3—figure supplement 2
No obvious hindrances against motions when molecules cross the phase boundaries.

(A) Phase boundary of the postsynaptic density (PSD) condensates determined by localization densities. The boundaries are shown by blue lines. Localizations are color-coded according to their local …

Figure 3—figure supplement 3
Correlation-based classification of molecular displacements without presuming simple diffusion.

(A) Conditional distribution of a collection of 2522 experimental trajectories with longer than 10 steps with each trajectory residing entirely within a model postsynaptic density (PSD) condensate. …

Figure 4 with 2 supplements
Immobilization of molecules by the large dynamic molecular network in the condensed phase of phase-separated systems.

(A) Schematic illustrations of the concept of stable molecular networks in the condensed phase formed by strongly favorable specific and multivalent interactions (left, blue) versus dynamic …

Figure 4—figure supplement 1
Binding between FUS prion-like domain (PrLD) is extremely weak.

(A) Fast protein liquid chromatography (FPLC)-coupled with static light scattering analysis showing the column behavior and measured molecular weight of the GB1-tagged PrLD at 100 μM (blue curve) …

Figure 4—figure supplement 2
Purified FUS prion-like domain (PrLD) takes more than 12 hr to form condensates.

(A) DIC images showing phase separation of FUS PrLD at different time point after removal of the GB1 tag by HRV-3C protease cleavage. Scale bar: 20 μm.

Figure 5 with 1 supplement
Simulated and experimental fluorescence recovery after photo-bleaching (FRAP) curves of the FUS prion-like domain (PrLD) systems.

(A) Schematic representation of the phase separation system for the present FRAP simulations. Simulations were conducted in a 20 μm × 8 μm box with periodic boundary conditions. Three regions of …

Figure 5—figure supplement 1
Representative confocal images showing fluorescence recovery after photo-bleaching (FRAP) experiments of the condensed droplets formed by prion-like domain (PrLD), PrLD-SAMME, or PrLD-SAMWT.

The region selected for photo-bleaching has size of 20 pixels or 1.95 μm in diameter. Photo-bleaching started at time point 0. Scale bar: 2 μm.

Author response image 1
Comparing fitting the experimental molecular displacements with a simple Brownian diffusion model (Figure 3—figure supplement 1C1) against fitting with the optimized HMM two-state diffusion model.

Best fit of the same experimental displacement distribution in condensed phase with a two-state HMM model. Red Curve is the HMM model fit obtained by non-linear least squares method using MATLAB. R2

Videos

Video 1
Raw image superimposed with phase boundary and tracks (steps length >5) in the NR2B + postsynaptic density (PSD) phase separation system on 2D supported lipid bilayer (SLB).

Red lines represent tracks in the condensed phase, green lines represent tracks in the dilute phase, and yellow lines represent tracks crossing phase boundaries. Molecules can be seen to switch …

Tables

Table 1
Simulation of phase separations with different input of diffusion parameters.

Monte Carlo method-based simulations of molecular diffusion on supported lipid bilayer (SLB) with experimental phase boundaries. A total of 50,000 molecules were included in each simulation, and …

Input parametersTheoretical resultsOutput results
Dm (μm2/s)Dd (μm2/s)Mobile ratioMobile state lifetime (s)Enrichment fold (EF)Enrichment fold (EF)
0.20.60.050.16058.9 ± 0.7
0.10.60.10.16058.6 ± 0.9
0.010.61-6057.0 ± 0.8
0.10.60.10.56057.9 ± 0.7
0.10.60.050.1120116.7 ± 1.9
0.20.60.10.13029.5 ± 0.3

Additional files

Download links