Interfacial Resistance and Preferential Internal Mixing of Phase-Separated Condensates

arXiv:2603.25447 · physics.bio-ph, cond-mat.soft, q-bio.SC · Submitted 2026-03-26 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: We're looking at the paper: "Interfacial Resistance and Preferential Internal Mixing of Phase-Separated Condensates".

Title and authors: Kai: So we're looking at the paper titled "Interfacial Resistance and Preferential Internal Mixing of Phase-Separated Condensates". It’s by Oihan Joyot, Zoé Ferrand, Fernando Muzzopappa, Pierre Weiss, and Fabian Erdel.

Mira: Yeah, it tackles the core problem of how molecules get stuck in these clusters without being able to move freely across the boundary.

Kai: Basically they introduce two main ways to describe this behavior: biased interfacial reflectivity and interfacial resistance. It’s about how strongly particles are reflected when they try to get in or out, and the kinetic rate at which they can cross that interface.

Lev: From a hardware standpoint, that means if we want to build a system where molecules stay put, we have to control both of those things precisely.

Kai: It’s about quantifying these two parameters experimentally; it’s been tricky for years. This paper puts together a theoretical and experimental framework to solve this problem by extending the half-FRAP approach.

The paper's summary: Mira: So, what the authors are actually doing here is taking that old half-FRAP method and upgrading it to explicitly account for interfacial resistance and biased reflectivity. It’s a mathematical model of diffusion in a spherical domain split into an inner sphere and an outer annulus.

Kai: They set up this diffusion equation where the concentration changes depend on different diffusion coefficients inside each part, like D in the inner sphere versus D out in the outer annulus.

Lev: And they introduce parameters like permeability kappa or one over kappa, which is that interfacial resistance we talked about before. They also have a partition coefficient gamma, which tells us how much of a molecule ends up on one side versus the other at the boundary.

Kai: They solve this using spectral decomposition, finding eigenvalues lambda k that solve a transcendental equation based on boundary conditions. Then they build a Green’s function from those eigenfunctions and eigenvalues.

Mira: The key takeaway here is how they quantify internal mixing with a parameter M. This M tells you the degree of internal mixing, specifically the residence time in the inner domain before escape compared to some free diffusion time.

Lev: So if M is between zero and one, it means there’s preferential internal mixing happening because of slow kinetics at the interface or strong suppression of diffusive escape in that outer domain.

The paper's improvements: Kai: They quantify this mixing degree using the parameter M, where M is related to the dimensionless resistance one over kappa star effective. This effectively tells you how many rounds of diffusive exploration particles make inside before they successfully escape.

Mira: Values of M between zero and one mean preferential internal mixing; it shows that the slow kinetics of crossing or the strong suppression of escape in the outer domain are what drive that internal mixing.

Lev: And they connect this to energetics too, showing a relationship between a theoretical dip depth in half-FRAP curves and the exchange cohesive energy N gamma E. The inflection point happens around an exchange cohesive energy of about two point five times the thermal energy k T.

Kai: That suggests that molecules in that interfacial layer have very few interactions per molecule compared to those deep inside the condensate.

Mira: Exactly, so they are more loosely connected at the boundary than in the core, and that's what drives this dip depth measurement we see in half-FRAP data.

Conclusion: Kai: So to wrap up, these results show that condensates with larger effective resistances—which means stronger intermolecular interactions—have a quasi-isolated pool of particles exhibiting preferential internal mixing.

Mira: And the dip depth in those half-FRAP curves is directly reporting on this preferential internal mixing behavior. This physical isolation can lead to localized biochemical memory and lets us selectively differentiate particle pools.

Lev: It also means higher interfacial reflectivities suppress those transient excursions into the surrounding medium, which is functionally relevant for sequestering molecules effectively.

Kai: They show that condensates at low salt concentrations exhibit a strong preference for internal mixing and are deep in the two-phase regime.

Mira: The paper concludes that the shape of the recovery curve and dip depth depends on this degree of internal mixing, which is controlled by the exchange cohesive energy.

Oihan Joyot, Zoé Ferrand, Fernando Muzzopappa, Pierre Weiss, Fabian Erdel

Center for Integrative Biology (CBI), University of Toulouse, CNRS

physics.bio-ph, cond-mat.soft, q-bio.SC

Submitted: 2026-03-26

Updated: 2026-11-15

Comments: 22 pages, 9 figures

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 89/100

The gist: Biomolecular condensates organize biochemical processes by spatially concentrating molecules while allowing for dynamic exchange with their surroundings; however…

Key concepts

Interfacial Resistance (1/κ)
This parameter measures how difficult it is for molecules to cross the boundary between two regions within a condensate. It is strongly influenced by intermolecular interactions and decreases as salt concentration increases, suggesting electrostatic screening affects transport.
Preferential Internal Mixing (M)
This quantifies the degree to which particles tend to stay within the inner domain of a condensate before escaping. High mixing values indicate strong physical isolation due to slow kinetics at the interface, enabling localized biochemical memory.
Dip Depth in Half-FRAP Curves
The depth of this dip in recovery curves serves as a direct experimental readout for preferential internal mixing. It is quantitatively linked to the exchange cohesive energy, indicating that molecules near the interface have fewer interactions than those deep inside.
Reflective Bias (ρ)
This describes how much more a particle is reflected when encountering the interface from the inside compared to the outside. High reflective bias suggests strong confinement and suppression of transient excursions into the surrounding medium.

Terminology

Summary

Biomolecular condensates organize biochemical processes by spatially concentrating molecules while allowing for dynamic exchange with their surroundings; however, transport across their interface can be strongly attenuated, leading to enhanced retention and preferential internal mixing The authors present a theoretical and experimental framework to address this issue by extending the half-FRAP approach to explicitly consider interfacial resistance and biased reflectivity

Mathematical Model of Diffusion

The mathematical model considers a spherical domain decomposed into an inner sphere and an outer annulus, defining concentrations based on diffusion coefficients in each domain The diffusion equation is described by terms involving the concentration fields and the interface parameters like permeability κ (or interfacial resistance 1/κ) and partition coefficient Γ The solution is found using spectral decomposition, which involves finding eigenvalues λk that solve a transcendental equation derived from boundary conditions The Green's function is then expressed as a sum over these eigenfunctions and eigenvalues

Quantifying Internal Mixing

The degree of internal mixing is quantified by the parameter M defined by the residence time in the inner domain before escape, specifically Mdef = (τres − τdiff) − (τfree − τdiff) / ((τres − τdiff) + (τfree − τdiff)) This degree of mixing is related to the dimensionless resistance 1/κ∗ eff, which is interpreted as the number of rounds of diffusive exploration particles undergo in the inner domain before successfully escaping Values of 0 < M ≤ 1 correspond to preferential internal mixing, arising from slow kinetics of interfacial crossing or strong suppression of diffusive escape in the outer domain

Experimental Findings on Coacervates

Applying the framework to tunable coacervates composed of polylysine and hyaluronic acid reveals that their interfaces exhibit strongly biased reflectivity with ρ ≥ 50% and substantial resistance of 1/κ∗ ≥ 3, both driving preferential internal mixing These interfacial properties depend on salt concentration, linking transport to intermolecular interaction strength The reflective bias ρ is found to decrease with increasing salt concentrations, suggesting that particles are reflected more strongly from the inside than from the outside at higher concentrations

Link Between Energetics and Dynamics

The theoretical dip depth is shown to have a sigmoidal relationship with the exchange cohesive energy NΓE, with its inflection point occurring at an exchange cohesive energy of ∼ 2.5 kT This suggests that molecules in the interfacial layer are connected with very few interactions per molecule compared to those within the interior of the coacervate

Information Content and Identifiability

The Fisher Information Matrix (FIM) quantifies the sensitivity of observations to parameter perturbations, and its smallest eigenvalue λmin(FI(θ)) = 0 indicates that the parameter vector θ is not locally identifiable The information content of half-FRAP curves is higher than that of full-FRAP curves, with larger regions of identifiability despite higher noise levels A neural network trained on synthetic recovery curves can nearly reproduce a minimal variance estimator, suggesting practical near-optimal estimators can be constructed from theoretical recovery curves

Interfacial Properties of PLL-HA Coacervates

For PLL-HA coacervates, the reflective bias in the absence of MgCl2 amounted to ρ ∼ 90%, meaning particles encountering the interface from the inside are reflected much stronger than from the outside With increasing salt concentrations, ρ decreased and reached roughly 50% at 150 mM MgCl2 The interfacial resistance decreases with increasing salt concentrations, consistent with electrostatic interactions being screened Coacervates at all MgCl2 concentrations were preferentially internally mixed, with M ≥ 80% The dip depth in half-FRAP curves reports on preferential internal mixing

Conclusion on Dynamic Signatures

Condensates with larger effective resistances, linked to stronger intermolecular interactions, contain a quasi-isolated pool of particles that exhibits preferential internal mixing This physical isolation enables localized biochemical memory and can be used to selectively differentiate particle pools Higher interfacial reflectivities suppress transient excursions into the surrounding medium, which is functionally relevant for sequestering molecules The results suggest that coacervates at low salt concentrations exhibit a strong preference for internal mixing and are deep in the two-phase regime

Summary of Key Parameters

The parameters Din, Dout, κ, and Γ set the characteristic time and length scales of the system but do not control the shape of the recovery curve or dip depth By evaluating information content in both halves, all parameters can be obtained if the interface is permeable enough The effective diffusion coefficient Deff out is linked to the criterion for particle escape, and 1/κ∗ eff represents the number of rounds of diffusive exploration particles undergo before escaping These parameters are not independent when modeling associative phase separation, with relevant parameter space defined by ∆ ≥ 1, Γ ≥ 1 and ρ ≥ 0 The paper concludes that the shape of the recovery curve and dip depth depend on the degree of internal mixing controlled by exchange cohesive energy

Methodology for Experimental Validation

The study employed quantitative confocal imaging to determine partition coefficients and fluorescence correlation spectroscopy (FCS) to measure ratios of diffusion coefficients Half-FRAP experiments were used in conjunction with these known values to determine interfacial resistances, yielding a comprehensive description of molecular transport The analysis involved fitting the recovery curves to theoretical models derived from spectral decomposition and solving the associated transcendental equation for eigenvalues The study also used a Multi-Layer Perceptron to map noisy recovery curves to parameters, corroborating the Fisher information analysis

Materials and Methods

PLL-HA coacervates were prepared by mixing poly-lysine and hyaluronic acid solutions at a concentration of 10 mg/mL, followed by reconstitution with different amounts of MgCl2 For FRAP experiments, FITC-labeled PLL was spiked into unlabeled PLL at a ratio of 1:500 FCS experiments utilized a 63×/NA 1.2 oil immersion objective to measure autocorrelation curves FRAP measurements were performed by segmenting coacervates and extracting average intensities of the bleached half and non-bleached half

Appendix A: Functional Analysis

The functional framework involves solving a time-dependent diffusion equation in a semipermeable domain using spectral decomposition The problem is formulated in terms of weighted Hilbert spaces H and V to handle the discontinuity introduced by Γ ≠ 1 The variational formulation leads to a symmetric bilinear form a(u, v) that governs the evolution of the concentration field c(t) Spectral analysis confirms that the operator A is self-adjoint and possesses a purely discrete spectrum with eigenvalues λk → +∞

Appendix B: Reflective Bias of the Interface

The reflective bias ρ is derived from the unidirectional fluxes across the interface in the limit of a saturated interface, yielding ρdef = κout→in − κin→out / (κout→in + κin→out) This expression corresponds to the result reported previously for one-dimensional diffusion across a single interface The partition coefficient is determined by ρ and ∆ according to Γ = 1 + ρ / (1 − ρ)√∆

Appendix C: Escape from the Inner Domain

The mean residence time Tin(x0) for a particle starting at x0 in the inner domain is given by Tin(r0) = K1 = R squared / (6Din) + R / (3κ) + ΓR squared / (3Deff out) The term R 2/(6Din) represents the mean first passage time to reach the interface of the inner domain This time delay is related to the characteristic internal mixing time τmix ≈ R 2/(π / (2Din))

Appendix D: Integrals in the expressions for halfand full-FRAP curves

The radial integral Ir ln is given by a complex expression involving spherical Bessel functions and generalized hypergeometric functions For the isotropic mode, the expression simplifies to sin(pλ0n/DinR) (λ0n/Din) 3/2 − R cos(pλ0n/DinR) λ0n/Din The angular integrals I ba l and I nb a vanish for even l > 0 and adopt specific forms for odd l > 1 The normalization constant N ln is defined by a complex expression involving Bessel functions and the auxiliary function Ξl(z; zL)

Appendix E: Integrated full-FRAP curve for free diffusion

For the limiting case of free diffusion, i.e., Γ = 1, Din = Dout = D and κ → ∞, the closed-form solution for the full recovery curve is given by erf(u) − 1 / (√π / 3 u − 2 u 3) − e - u squared / (1/u − 2 u 3) This closed-form solution is also obtained from the spectral solution in Eq. p. 4

Where the paper says it

(D4)-(D5) show that full-FRAP curves can be obtained from half-FRAP curves via FRAPfull(t) = FRAPhalf,b(t) +FRAPhalf,nb(t)−1. (28) In the limit of free diffusion, the expression for full-FRAP simplifies to the following closed form (Appendix E): FRAPfree full(t) = 1 − c free full R √ Dt , (29) with c free full(u) = erf(u) − 1 √ π 3 u − 2 u 3 − e −u 2 1 u − 2 u 3 .

(26) by taking the limit where the radius of the outer domain L tends to infinity

Appendix F: Materials and Methods

PLL-HA coacervates were prepared using stock solutions of Poly-L-Lysine hydrobromide and Hyaluronic Acid sodium salt, reconstituted by mixing them at a mass ratio of 1:4 Partition coefficients were quantified by analyzing fluorescence intensity calibration curves derived from confocal microscopy images FCS experiments provided the ratio of diffusion coefficients ∆ via autocorrelation curves fitted to a function involving the beam waist FRAP experiments involved segmenting coacervates and calculating normalized recovery curves using the formula FRAPI B/NB = IB/NB(t) − IBG(t) / (IREF(t) − IBG(t)) + A

Discussion

Condensates with larger effective resistances, linked to stronger intermolecular interactions, contain a quasi-isolated pool of particles that exhibits preferential internal mixing The dip depth in halfFRAP curves reports on preferential internal mixing This physical isolation enables localized biochemical memory and can be used to selectively differentiate particle pools Higher interfacial reflectivities suppress transient excursions into the surrounding medium, which is functionally relevant for sequestering molecules The paper concludes that coacervates at low salt concentrations exhibit a strong preference for internal mixing and are deep in the two-phase regime

Figure 11 Interpretation

Condensates with larger effective resistances, which are linked to stronger intermolecular interactions, contain a quasi-isolated pool of particles that exhibits preferential internal mixing The dip depth in halfFRAP curves reports on preferential internal mixing

Figure 10 Link Between Dip Depth and Energetics

The theoretical dip depths are shown to depend on the degree of internal mixing, which is mainly controlled by the exchange cohesive energy The inflection point occurs at an exchange cohesive energy of ∼ 2.5 kT The relationship between these energies suggests that molecules in the interfacial layer are connected with very few interactions per molecule compared to those within the interior of the coacervate The dip depth is a direct readout of internal mixing

Figure 11 Condensates with Distinct Dynamic Signatures

Biomolecular condensates can display distinct dynamic properties distinguished by half-FRAP Condensates with larger effective resistances, which are linked to stronger intermolecular interactions, contain a quasi-isolated pool of particles that exhibits preferential internal mixing The dip depth in halfFRAP curves reports on preferential internal mixing

Figure 9 Interfacial Resistance Determined by Half-FRAP

The dimensionless interfacial resistance decreases with increasing salt concentrations, consistent with the interpretation that it arises from electrostatic interactions established by the molecules in the coacervates The effective resistance 1/κ∗ outer is smaller than the interfacial resistance and is independent of the salt concentration, indicating that particle escape is mainly limited by interfacial resistance Coacervates at all MgCl2 concentrations were preferentially internally mixed, with M ≥ 99% The exploration time in the inner domain is greater than 100-times in the absence of MgCl2 The exploration time decreases to about 4-times in the presence of 150 mM MgCl2

Table I Interfacial Properties of PLL-HA Coacervates

Table I lists parameters describing the interfaces of PLL-HA coacervates determined by a combination of fluorescence microscopy imaging, FCS, and half-FRAP The dimensionless interfacial resistance decreases with increasing salt concentrations, consistent with the interpretation that it arises from electrostatic interactions established by the molecules in the coacervates The degree of internal mixing M is found to be ≥ 99% at all MgCl2 concentrations The exploration time in the inner domain is greater than 100-times in the absence of MgCl2

Figure 6 Information Content and Parameter Identifiability

The information content of half-FRAP curves is higher than that of full-FRAP curves, with larger regions of identifiability despite higher noise levels The inner diffusion coefficient Din can be recovered across all the parameter space with half-FRAP, provided that the frame rate and duration of the experiment can be appropriately chosen The identifiability maps show a region extending from the top left corner where all four parameters can be recovered, coinciding with low effective resistances The relative standard deviations for parameters like Din can be recovered across all parameter space with half-FRAP

Figure 6b Neural Network Recovery

The trained MLP can nearly reproduce a minimal variance estimator, suggesting that the network can achieve a lower RMSE locally even when Fisher information is low Predictions based on half-FRAP curves are overall better than those based on full-FRAP curves The results suggest that it is possible to construct practical near-optimal estimators by training neural networks on theoretical recovery curves

Condensates with larger effective resistances, which are linked to stronger intermolecular interactions, contain a quasi-isolated pool of particles that exhibits preferential internal mixing The dip depth in halfFRAP curves reports on preferential internal mixing This physical isolation enables localized biochemical memory and can be used to selectively differentiate particle pools Higher interfacial reflectivities suppress transient excursions into the surrounding medium, which is functionally relevant for sequestering molecules The dip depth in halfFRAP curves reports on preferential internal mixing

The theoretical dip depths are shown to depend on the degree of internal mixing, which is mainly controlled by the exchange cohesive energy The inflection point occurs at an exchange cohesive energy of ∼ 2.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved parameter estimation for interfacial properties by utilizing Fisher information matrices to quantify sensitivity to parameter perturbations in integrated recovery curves. This allows for a lower bound on parameter uncertainty via the Cramér–Rao bound, providing a rigorous framework for inferring interfacial resistance and reflective bias from experimental data.

  2. Enhanced ability to predict experimental outcomes by training neural networks using projected principal components of recovery curves as input. This system can construct practical near-optimal estimators that may outperform unbiased estimators, especially in regions where Fisher information is low.

  3. Quantitative modeling of dynamic signatures by predicting the degree of internal mixing M from the dip depth observed in half-FRAP curves. This provides a direct readout linking experimental measurements to the physical isolation and preferential internal mixing behavior of biomolecular condensates.

  4. Design of tunable model systems for drug delivery by mapping macroscopic parameters like exchange cohesive energy to observable dynamic signatures such as dip depth, allowing for controlled manipulation of interfacial properties through salt concentration changes in PLL-HA coacervates.

Abstract

Biomolecular condensates spatially organize biochemical processes. They can selectively partition molecules and enrich them well above their concentration outside. Their function depends on the exchange of molecules with their surroundings, whose kinetics are determined not only by partitioning and diffusion but also by interfacial resistance. The latter can hinder molecules from crossing the condensate boundary, thereby increasing their residence times in the condensate and promoting repeated encounters with interaction partners therein. So far, it has proven difficult to measure the interfacial resistance of condensates experimentally. Here, we combine fluorescence recovery after photobleaching (FRAP) with fluorescence correlation spectroscopy (FCS) and calibrated fluorescence imaging to disentangle partitioning, interfacial resistance, and differences in diffusion inside and outside of condensates. Applying this approach to polyelectrolyte condensates across a range of salt concentrations, we find substantial, salt-dependent interfacial resistance, which causes molecules to undergo multiple rounds of internal mixing before escaping. DDX4-YFP condensates in living cells also exhibit substantial interfacial resistance, although they are more permeable than polyelectrolyte condensates with comparable partition coefficients. Our results indicate that interfacial resistance is a tunable property of condensates that can place them in distinct transport regimes, favoring distinct functions such as dynamic reaction crucibles or insulated molecular sinks.

Related papers