A unifying model of T-cell signaling protein condensates in reconstitution experiments
- Department of Chemical Engineering, Massachusetts Institute of Technology, United States
- Department of Chemistry, Stony Brook University, United States
- Department of Physics, Massachusetts Institute of Technology, United States
- Department of Chemistry, University of California, Berkeley, United States
- California Institute for Quantitative Biosciences, University of California, Berkeley, United States
- Institute for Medical Engineering and Science, Massachusetts Institute of Technology, United States
- Ragon Institute of Massachusetts General Hospital, Massachusetts Institute of Technology and Harvard University, United States
- Department of Chemistry, Massachusetts Institute of Technology, United States
Figures
Figure 1
Schematic representation of a T cell engaging with an antigen-presenting cell (APC) through the T-cell receptor (TCR) binding to peptide-MHC (pMHC).
Successful recognition of a foreign peptide triggers a signaling cascade that includes the formation of LAT condensates with the help of Grb2 and SOS1 as crosslinkers (inset). LAT condensates serve as hubs that incorporate various other signaling molecules. Left hand side is created with BioRender (modified further) and right hand side is also created with BioRender (modified further).
Figure 2 with 1 supplement
Reconstitution experiments on supported lipid bilayers (SLBs).
(a) shows the reconstitution experiment setup. LAT is attached to the SLB via a His-tag and pre-phosphorylated. Grb2 and SOS1 are added to the fluid domain above the SLB, triggering crosslinking and phase separation (Created with Biorender.com and further modified). (b) shows an experimental trajectory where LAT is fluorescently labeled. After an initial lag time without macroscopic density changes, LAT quickly phase separates, leading to an LAT-rich phase with LAT-depleted, non-circular inclusions. This behavior is confirmed by the relative probability of the LAT density, which changes from a distribution with a single peak in the homogeneous phase to a bimodal distribution upon phase separation. The figure was created from data described by Sun et al., 2022. Raw microscopy data are available on Zenodo (Groves et al., 2025).
Figure 2—video 1
Phase separation of fluorescently labeled LAT in reconstitution experiments, recorded at a rate of 1 frame per minute (62 frames total).
Macroscopic phase separation can only be observed after an initial lag time but proceeds rapidly thereafter, leading to a near-stationary morphology of non-circular, LAT-depleted inclusions in an LAT-rich phase. This movie was created from data described in Sun et al., 2022.
Figure 3
Different morphologies observed in reconstitution experiments, where LAT is fluorescently labeled.
The LAT-rich minority phase can form circular or non-circular patterns, and similar morphologies can be observed for an LAT-depleted minority phase (see Figure 11). In addition, the experiments can result in a bicontinuous phase. The figure was created from data described by Sun et al., 2022 (scale bars: 5 µm).
Figure 4
Illustration of the multiplicative split of the reaction kernels proposed by Moncho-Jordá et al., 2001.
The Brownian contribution provides the rate at which clusters of size and get within binding distance and describes the probability of the clusters to subsequently bind during this encounter (Created with Biorender.com and modified further).
Figure 5
Mean cluster size obtained from simulating Equations 1–9 for different values of the binding probability , plotted against non-dimensional time (left panel) and time rescaled by the characteristic time of dimer formation, (right panel).
The onset of a rapid increase in the average cluster size follows an initial lag time that is modulated by . Scaling of time by the characteristic time of dimer formation collapses all curves at early times in the right panel, indicating that the lag time originates from the initial formation of smaller aggregates.
Figure 6
Free energy density (a) and diffusion coefficient (b) for and .
The red dots indicate the minima of the free energy density . (a) shows that only a single minimum in the energy density exists for and two distinct minima for . If the LAT density is close to the minimum with the higher LAT density, the diffusivity is significantly reduced, as can be seen in (b). Otherwise, the diffusivity remains unaffected.
Figure 7
Bond formation in the homogeneous state before phase separation.
Subfigure (a) shows the bond density plotted against non-dimensional time for different values of the dissociation constant . The bond density increases rapidly initially and subsequently plateaus at a steady-state value that decreases with increasing . Subfigure (b) shows the time to reach , i.e. the threshold bond density required for phase separation. When approaches , defined in Equation 42, increases rapidly as the effects of bond dissociation become relevant. Lastly, Subfigure (c) shows for depending on the average number of binding sites per LAT molecule normalized by the case for each value of . We observe a several-fold increase of as the number of available binding sites decreases.
Figure 8
Simulation trajectory obtained for the parameters , , and .
The top row shows the LAT density while the bottom row shows the bond density . At early times, the bond density is small and no phase separation is observed. Between the second and third frame, the threshold bond density is reached and phase separation is observed. Subsequently, elongated, LAT-rich inclusions form in the LAT-depleted majority phase. Due to the small diffusion coefficient in the LAT-rich phase, these inclusions become close to stationary. Furthermore, the non-linear dependence on the binding contribution in Equation 29 leads to preferred binding in the LAT-rich phase, resulting in an increased bond density in that phase.
Figure 9
Examples of the different possible morphologies obtained from our simulations, closely resembling the morphologies obtained in experiments.
Figure 10
Phase diagrams of the observed morphologies at the end of each simulation for varying free energy parameters and , defined in Equations 33 and 34, respectively, at different initial densities .
The blue background indicates the domain of permissible parameter combinations based on Equations 22 and 24. Each marker represents a single simulation and the meaning of each marker is indicated in Figure 9.
Figure 11
With increasing density, the observed morphologies change from near circular LAT-rich inclusion to elongated LAT-depleted islands to LAT-depleted circular islands.
The figure is reproduced from data described by Huang et al., 2016. (scale bars: 5 μm).
Figure 12
Phase diagrams of the observed morphologies at the end of each simulation for varying free energy parameters and , defined in Equations 33 and 34, respectively, at different initial densities .
The blue background indicates the domain of permissible parameter combinations based on Equations 22 and 24. Each marker represents a single simulation and the meaning of each marker is indicated in Figure 9. The unbinding rate is twice the value of that used in Figure 10, i.e. .
Appendix 1—figure 1
Mean cluster size plotted against time scaled by the characteristic time of dimer formation .
We vary the parameters that are held fixed in the main text. The qualitative behavior of the aggregation process does not depend on the choice of these parameters. Specifically, is a good approximation of the characteristic onset time, irrespective of the parameter values considered here. The parameters that are held constant in each plot are set according to Table 1 of the main text.
Appendix 1—figure 2
Mean cluster size plotted against time, scaled by and , respectively.
We vary the parameters , , and to verify the scaling of in our simulations, showing that the crossover occurs at . The parameters that are held constant in each plot are set according to Table 1 of the main text.
Appendix 1—figure 3
Average aggregate size for different values of the fragmentation kernel.
At early times, the solution is unaffected by fragmentation. In contrast, at long times, the solution plateaus at a constant average aggregate size where aggregation and fragmentation are balanced.
Appendix 1—figure 4
Concentration of monomers, dimers, trimers, and tetramers for varying values of the fragmentation kernel.
Similarly to the average aggregate size, the early time behavior is unaffected by fragmentation. At long times, the concentrations remain elevated compared to the irreversible case.
Tables
Table 1
Parameters used to numerically solve the Smoluchowski aggregation model.
| 10−2 | |
| 1 | |
| 1 | |
| 10 pN µs/nm (Honerkamp-Smith et al., 2013) | |
| bs | 10−2 pN µs/(nm)3 (Anthony et al., 2022) |
| 2/3 (Meakin, 1987a) |
Table 2
Parameter values used in Equations 30–32 for the spatially resolved simulation results.
The value of is chosen in accordance with experimental observations (Huang et al., 2017a), the value of reflects the three tyrosine residues that preferentially bind Grb2 (Houtman et al., 2004) and is estimated based on previous theoretical findings (Nag et al., 2012). The remaining parameters are chosen to qualitatively reproduce experimental trajectories.
| 3 | 20 | 0.05 | 3 | 1 |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
A unifying model of T-cell signaling protein condensates in reconstitution experiments
eLife 15:e109567.
https://doi.org/10.7554/eLife.109567
