Recent advances in cell biology highlight the crucial role of biomolecular condensates in organizing cellular structures. A landmark study (Pombo-Garcia et al., Nature 2024) showed that the scaffold protein PATJ (P) undergoes liquid-liquid phase separation on the inner leaflet of the plasma membrane. This creates a “prewetted” domain that recruits and organizes ZO-1 (Z) proteins into ring-like tight junctions. While this discovery provides a compelling physical picture, it also raises deeper questions. Specifically, we still lack a quantitative understanding of several key aspects. These include the kinetic pathway from dispersed Z condensates to a continuous ring, how the membrane microenvironment (e.g., lipid composition) regulates assembly, and the synergistic interplay between protein diffusion, anchoring, and multivalent interactions. The absence of a predictive theoretical framework hinders both mechanistic insight and the potential to engineer such processes.
To address these challenges, we developed a novel multiscale strategy. Our approach uniquely integrates coarse-grained molecular dynamics (CG-MD) simulations with a reaction-diffusion (RD) theoretical model. The CG-MD simulations, using the Martini force field, explicitly model molecular interactions between P, Z, and the complex cytoplasmic domain membrane (CDM). Complementing this, our RD model is derived from an extended Ginzburg-Landau free-energy functional that incorporates Hill-type cooperative kinetics. This allows us to describe the macroscopic spatiotemporal evolution of P and Z concentrations on the 2D membrane surface. Together, this dual-scale approach bridges detailed molecular interactions with the emergence of tissue-scale organization.
The dynamic assembly of Z, orchestrated by P, is directly visualized in our CG-MD simulations, as shown in Figure 1. Panel (a) establishes the baseline: without P, Z proteins form small, dispersed condensates on the CDM. Introducing membrane-anchored P changes everything. When P is anchored to the membrane surface. Panels (b) (evolution of ring formation probability over time) through (c) capture the subsequent kinetic progression. First, P acts as a nucleation site for initial Z aggregation. These nascent clusters then extend and coalesce into elongated bands. Finally, the bands mature and close into continuous, circumferentially aligned rings (b-c). This visual sequence provides clear, molecule-level evidence. It unequivocally establishes P as the essential initiator and spatial director that drives the system from disorder into a precisely structured state.
The predictive power of our theoretical framework is further illustrated by the spatial concentration analysis in Figure 2. This figure shows the 2D distribution of relative concentrations for P and Z on the CDM at t = 200 min, a dynamically steady state. The left and right panels depict the concentration fields of P and Z, respectively, with color intensity indicating local concentration. A key observation is the strong spatial correlation: regions rich in P consistently overlap with domains of high Z concentration. This quantitative co-localization directly validates the prewetting and co-condensation mechanism proposed by our model. Moreover, the patterns exhibit a characteristic wavelength and connectivity that align well with the ring-like structures in Figure 1.
To extract general physical principles and make quantitative predictions, we solved our reaction-diffusion model analytically and numerically. This analysis yields several key insights. First, the relative diffusion coefficients of P and Z set a characteristic length scale for the phase separation. We find an optimal regime where matched diffusivities lead to efficient coupling and regular patterns. Second, the membrane anchoring strength of P has a non-monotonic effect. Intermediate anchoring maximizes P's dwell time on the membrane, promoting large, stable Z rings. In contrast, overly weak or strong anchoring leads to disorder or kinetic arrest. Third, and most critically, the cooperativity of the P-Z interaction—quantified by the Hill coefficient n-acts as a biological switch. The model predicts a sharp transition to a ring-dominated state when n exceeds a critical threshold (approximately 8.3). This provides a precise, testable criterion for continuous structure formation.
The consistency between our theory and simulations confirms the robustness of our framework. For instance, the characteristic patterns predicted by the RD model's linear stability analysis clearly appear in the simulated concentration fields of Figure 2. Similarly, predicted parameter thresholds, like that for the Hill coefficient, match the morphological transitions seen in the CG-MD snapshots of Figure 1. This agreement across scales confirms that our model captures the essential physics of the interface-driven assembly process.
In conclusion, this study moves beyond qualitative description to establish a quantitative, mechanistic framework for condensate-mediated protein assembly at interfaces. We have clarified the complete kinetic pathway, pinpointed the tripartite regulatory axis of diffusion, anchoring, and cooperativity, and rigorously validated the model through integrated simulation and theory. Our integrated model does more than deepen the understanding of tight junction formation. It serves as an extensible theoretical tool that offers a new physical perspective for exploring membrane-associated compartmentalization. Furthermore, it provides foundational design principles for engineering synthetic protein assemblies with defined architectures at cellular interfaces.