Protein-protein interaction networks can be highly sensitive to the membrane phase transition

Basic Information

• Paper ID: 2510.26949
• Title: Protein-protein interaction networks can be highly sensitive to the membrane phase transition
• Authors: Taylor Schaffner, Benjamin Machta (Yale University)
• Classification: physics.bio-ph, cond-mat.stat-mech, q-bio.SC
• Publication Date: November 3, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.26949

Abstract

This study investigates how protein-protein interaction (PPI) networks occurring in the plasma membrane are affected by membrane phase transitions. The authors propose a Monte Carlo simulation framework to study how changes in lipid domain size caused by membrane criticality perturbations lead to alterations in inter-component interaction rates. The research demonstrates that near the membrane phase transition critical point, the activity of simple PPI networks is highly sensitive to thermodynamic parameters. When protein interactions alter the partitioning of certain components, the system sometimes forms non-equilibrium domains (termed "pockets"), driven by both thermodynamic interactions and kinetic sorting. The study predicts that near the critical point, results from many different PPI networks will exhibit sensitive dependence on perturbations affecting critical behavior.

Research Background and Motivation

1. Research Questions

The core question addressed by this study is: How do the thermodynamic properties of the plasma membrane near the miscibility critical point influence the outcomes of protein-protein interaction networks occurring within it?

2. Problem Significance

• Heterogeneous membrane structure: The plasma membrane is no longer viewed as a uniform 2D solvent, but rather as a heterogeneous yet fluid structure composed of lipid domains (lipid rafts)
• Universality of critical behavior: Vesicles extracted from living cells exhibit critical behavior, suggesting that in vivo membranes may be near the critical point (with characteristic length scales ~20 nm)
• Biological relevance: Many signal transduction processes depend on membrane domains, such as the partitioning of activated kinases to ordered domains and inactive phosphatases to disordered domains in B lymphocytes

3. Limitations of Existing Methods

• Prior work focused on single proteins: Previous studies primarily explored effective forces between paired proteins in near-critical solvents, or regulation of individual ion channels
• Lack of network-level understanding: The systematic effects of membrane coupling on complex protein interaction networks remain unstudied
• Theory-experiment gap: While membrane thermodynamics are well understood, the mechanisms by which they affect PPI networks remain unclear

4. Research Motivation

• Experimental evidence support: Perturbations affecting critical distance (cholesterol depletion/loading, short-chain alcohols) influence multiple signaling pathways
• Component partitioning observations: Many PPI network components are known to partition into specific membrane domains
• Bidirectional regulation: Certain PPI networks produce downstream changes in domain size and composition (e.g., BCR activation increases ordered domain formation)

Core Contributions

1. Established theoretical model and simulation framework: Developed a Monte Carlo simulation framework (MemTropy) based on the 2D Ising model to study PPI networks in membranes near the critical point
2. Proposed graphical prediction method: Created a simple graphical approach for qualitatively predicting the effects of temperature changes on network activity through symbolic analysis of component partitioning and interaction pathways
3. Revealed size-dependent sensitivity: Demonstrated that larger protein components are more sensitive to critical temperature changes, with sensitivity increasing with protein radius
4. Discovered non-equilibrium "pocket" phenomenon: When interactions alter component partitioning (e.g., palmitoylation reactions), the system can form non-equilibrium domain structures driven by both thermodynamics and kinetics
5. Proposed hypothesis of membrane as global regulator: Predicted that cells can simultaneously control multiple PPI networks by regulating membrane composition near the critical point

Detailed Methods

Task Definition

Investigation of how protein interaction network activity depends on membrane thermodynamic parameters in environments near the liquid-liquid critical point, specifically:

• Inputs: Membrane rescaled temperature τ = T/Tc, magnetization m (ordered domain area fraction), component size, partitioning preferences
• Outputs: Target protein activity state (activation/inactivation ratio)
• Constraints: Membrane lipids follow conserved dynamics (model B universality class), protein diffusion satisfies detailed balance

Model Architecture

1. Membrane Model

• 2D Ising model: Square lattice represents the membrane, with spins si = ±1 representing lipid components tending to partition into ordered/disordered phases
• Lattice spacing: ~1 nm (based on characteristic length and lipid size from cell-derived vesicles)
• Hamiltonian:

  Htot = Hmem({s}) + Σ Hint({s}∂, {b}i)
  

  where Hmem is the interaction energy between membrane lipids, and Hint includes interactions between inclusion boundary spins and adjacent membrane spins

2. Protein Model

• Disk-shaped inclusions: PPI components represented as disk-shaped inclusions coupled to the membrane through boundary spins {b}i
• Partitioning preferences: Implemented through coupling energy between boundary spins and neighboring membrane spins
• Size variation: Studied disks with radii r = 0, 1, 2, 3

3. Dynamics Algorithm

• Kawasaki algorithm: Spin exchange with neighbors (local conserved order parameter, model B universality class)
• Inclusion diffusion: Disk translation to adjacent lattice sites, obeying detailed balance
• Reaction kinetics: When specific components contact, state changes at fixed rates (e.g., phosphorylation)

4. PPI Reaction Schemes

Example basic reaction network:

A + TI → A + TA  (activation reaction)
I + TA → I + TI  (inactivation reaction)

where A is the activator, I is the inactivator, and TI/TA are inactive/active targets

Technical Innovations

1. Graphical Prediction Method

• Dual symbol system:
  • Upper symbols (purple +/-): Whether reactions increase or decrease target activity
  • Lower symbols (orange +/-): Whether interaction rates increase with temperature
  • Same-domain components: Orange symbol is negative (higher temperature, smaller domains, reduced contact)
  • Different-domain components: Orange symbol is positive (higher temperature, smaller domains, increased contact)
• Prediction rule: If all pathway symbol products are consistent, that product gives the sign of activity slope with respect to τ

2. Distinction from Baseline

• Traditional models: Treat membrane as uniform solvent, ignoring domain structure effects on interaction rates
• Model advantages:
  • Explicitly considers membrane thermodynamic fluctuations and domain structure
  • Couples spatial configuration with reaction kinetics
  • Captures enhanced sensitivity near critical point

3. Design Rationale

• Physical basis: Based on experimentally observed membrane critical behavior (2D Ising universality class)
• Biological relevance: Parameters like inclusion size and partitioning preferences correspond to real protein properties
• Computational efficiency: Lattice model enables efficient Monte Carlo sampling

Experimental Setup

Datasets

• Simulation system: Typical lattice size N×N (specific values not explicitly stated, estimated ~200×200 from figures)
• Component numbers: Protein quantities of each type vary by experimental design
• Temperature range: τ from ~0.8 to 1.2, covering the critical point region
• Magnetization: m from -0.4 to 0.4, exploring different domain area ratios

Evaluation Metrics

• Target activity fraction: fA = number of active targets / total targets
• Spatial correlation function: Radially averaged cross-correlation function G(r), quantifying spatial associations between components
• Sensitivity parameters:
  • Growth width Δτ: Obtained from logistic fitting, characterizing steepness of activity changes
  • Maximum slope: Maximum derivative of activity-temperature curve

Comparison Methods

• Different component sizes: r = 0 (point particles), 1, 2, 3
• Different partitioning schemes: Fixed partitioning vs. dynamically changing partitioning
• Different membrane compositions: Varying magnetization m

Implementation Details

• Equilibration: Non-local Kawasaki exchanges used to accelerate thermal equilibration
• Data collection: Activity measured in steady state
• Statistics: Multiple independent simulations for each parameter combination
• Fitting function: Generalized logistic function

  fA(τ) = (R-L)/2 * tanh((τ-τ0^fit)/(2Δτ)) + (R+L)/2
  

Experimental Results

Main Results

1. Domain Regulation of PPI Activity (Figure 2)

• Spatial organization changes: As τ increases, different-domain components (activator-target) contact more frequently, while same-domain components (inactivator-target) contact less
• Correlation function analysis: At τ = 0.9, targets highly correlated with inactivators; at τ = 1.1, target correlation with activators strengthens
• Activity changes: Target activity monotonically increases from ~0.3 at τ = 0.85 to ~0.6 at τ = 1.15
• Critical sensitivity: Maximum changes occur in τ = 1.0 ± 0.1 interval, with activity variance maximized near critical point

2. Size-Dependent Sensitivity (Figure 3)

Fitting Parameters Table:

• Key finding: Δτ decreases with increasing r (0.0209→0.0150), indicating larger components are more sensitive to temperature changes
• Physical explanation: Inclusion energy contribution scales with size (Equation 1), larger disks produce stronger domain preference forces

3. Effects of Domain Area Fraction (Figure 4)

• Small components (r=0): Activity insensitive to magnetization m changes, primarily dependent on τ
• Large components (r=3):
  • Activity monotonically increases with m (activity ~0.2 at m = -0.4, ~0.8 at m = 0.4)
  • Can still form high-contrast domains at τ > 1 (unlike r=0)
  • Shows sensitivity to both m and τ
• Mechanism: Stronger solvent accumulation forces around larger inclusions maintain domain structure even above critical temperature

4. Non-equilibrium "Pocket" Formation (Figure 5)

• Occurrence conditions: When target partitioning changes with activity state (e.g., inactive state prefers disordered domain, active state prefers ordered domain)
• Statistical characteristics:
  • Average activity conforms to graphical predictions (increases with τ)
  • At τ < 1, rare simulations show anomalously high activity (outliers)
• Pocket structure:
  • Filled with active targets and activators
  • Excludes inactivators (despite having identical lipid interactions)
  • Results from combined thermodynamic and kinetic effects
• Formation mechanism:
  1. Activators in disordered domain activate targets
  2. Active targets prefer ordered domain, forming ordered domain "pocket" around them
  3. New inactive targets diffuse in and become activated, trapped in pocket
  4. Inactivators cannot cross disordered phase to enter pocket
• Enhancement factors: Pocket formation frequency significantly increases at m = -0.4 (tighter lipid budget)

Ablation Studies

Although not explicitly labeled as "ablation studies," systematic parameter variations provide equivalent analysis:

1. Component size effects: Comparison of systems with r = 0, 1, 2, 3
2. Partitioning scheme effects: Fixed partitioning vs. dynamically changing partitioning
3. Membrane composition effects: Comparison across different m values

Case Studies

Figure 5C presents three typical pocket formation cases:

• Case 1: Single small pocket, moderate activity increase
• Case 2: Large pocket, activity reaches ~0.8
• Case 3: Multiple coexisting pockets, activity near 1.0
• Time evolution: Activity-time trajectories show rapid activity increase after pocket formation with sustained levels

Experimental Findings

1. Critical point is sensitivity hotspot: All measured sensitivity metrics peak near τ ≈ 1
2. Size-sensitivity relationship: Δτ inversely proportional to r, maximum slope increases with r
3. Dual-parameter regulation: Large component systems sensitive to both τ and m, providing 2D regulatory space
4. Non-equilibrium state diversity: Dynamic partitioning systems exhibit bistable characteristics (normal domains vs. pockets)
5. Kinetic traps: Once formed, energy barriers prevent pocket dissociation, exhibiting memory effects

Related Work

1. Experimental Studies of Membrane Criticality

• Veatch et al. (2007, 2008): Critical fluctuations in synthetic GUVs and cell-derived vesicles conform to 2D Ising model
• Baumgart et al. (2007): Large-scale phase separation in living cell plasma membrane vesicles
• Burns et al. (2017): Miscibility transition temperature in zebrafish cell lines varies with growth temperature

2. Membrane-Mediated Interactions

• Machta et al. (2012): Critical Casimir forces in cell membranes
• Katira et al. (2016): Pre-transition effect-mediated assembly forces between transmembrane proteins
• Reynwar & Deserno (2008): Membrane composition-mediated protein-protein interactions

3. Single Protein Allosteric Regulation

• Kimchi et al. (2018): Ion channels allosterically regulated by membrane domains near demixing critical point
• Suma et al. (2024): Aggregation, cooperativity, and memory effects of ion channels in critical membranes

4. Biological Function Studies

• Stone et al. (2017): Protein sorting supported by lipid phase-like domains in B lymphocyte plasma membranes
• Shelby et al. (2023): Membrane phase separation drives responsive assembly of receptor signaling domains
• Douglass & Vale (2005): Protein-protein networks create plasma membrane microdomains in T cells

Unique Contributions of This Work

• From single protein to networks: First systematic study of membrane criticality effects on complex PPI networks
• Quantitative prediction framework: Provides actionable graphical method and simulation tools
• Non-equilibrium phenomena: Discovers and explains pocket formation from dynamic partitioning

Conclusions and Discussion

Main Conclusions

1. Critical sensitivity is universal: PPI network outcomes near critical point are highly sensitive to membrane thermodynamic parameters (τ and m)
2. Size enhancement effect: Larger protein components produce stronger sensitivity, with Δτ reducing ~30%
3. Global regulation mechanism: Cells can simultaneously control multiple PPI networks by regulating membrane composition near critical point
4. Non-equilibrium domain formation: Dynamic partitioning mechanisms can produce "pocket" structures driven by both thermodynamics and kinetics
5. Biological relevance: Experimentally observed phenomena (e.g., cholesterol regulation of signaling, inflammation-induced changes in membrane criticality) can be explained by this model

Limitations

1. Model simplifications:
   • Uses 2D Ising model without hydrodynamic interactions
   • Disk-shaped inclusions are crude approximations of real protein shapes
   • Reaction rates set constant, not accounting for allosteric effects
2. Limited parameter space:
   • Primarily explores simple activation-inactivation networks
   • Limited component numbers and types
   • Lattice size constrained by computational resources
3. Randomness in pocket formation:
   • Formation frequency depends on initial conditions
   • Time scales to reach steady state are long
   • Quantitative prediction is difficult
4. Lack of experimental validation:
   • No direct experimental evidence for size-sensitivity relationship
   • Pocket phenomenon requires experimental confirmation
   • Quantitative comparisons need more precise parameter measurements

Future Directions

1. Self-organized criticality: Explore feedback mechanisms where PPI networks act as sensors maintaining membrane near critical point
2. Three-dimensional extension: Study similar effects of liquid-liquid phase separation in cell nucleus and cytoplasm on PPI networks
3. Complex networks: Simulate real signaling pathways (e.g., TCR, BCR signaling)
4. Experimental collaboration:
   • Design experiments measuring sensitivity of different-sized proteins
   • Search for experimental evidence of pocket phenomena
   • Validate simultaneous multi-pathway effects of membrane composition regulation
5. Theoretical development:
   • Develop analytical theory predicting sensitivity
   • Study energy landscapes of pocket stability
   • Explore memory effects and hysteresis phenomena

In-Depth Evaluation

Strengths

1. Methodological Innovation

• Interdisciplinary integration: Successfully combines statistical physics (Ising model, critical phenomena) with cell biology (signal transduction)
• Computational framework: Developed MemTropy framework with good scalability and open-source availability
• Graphical method: Concise prediction tool enabling qualitative conclusions without full simulation

2. Experimental Design Adequacy

• Systematic parameter scanning: Comprehensive exploration of key parameter spaces (τ, m, r)
• Multi-angle analysis: Combines spatial configuration, correlation functions, and activity measurements
• Statistical reliability: Multiple independent simulations with variance analysis showing robust results

3. Result Convincingness

• Quantitative fitting: Logistic function fitting provides comparable quantitative parameters (Δτ, τ0, etc.)
• Theory-simulation consistency: Graphical predictions highly consistent with simulation results
• Clear physical picture: Configuration snapshots and correlation functions intuitively demonstrate mechanisms

4. Writing Quality

• Clear logic: Progresses from simple to complex, building understanding systematically
• Excellent visualization: Well-designed figures with high information density
• Complete supplementary material: Detailed algorithm descriptions and data availability statements

Weaknesses

1. Method Limitations

• Lattice model constraints: Discretization may affect critical behavior accuracy, especially with small characteristic lengths
• Lack of hydrodynamics: Real membrane hydrodynamic interactions may be important at cellular scales
• Reaction rate assumptions: Fixed reaction rates ignore conformational changes and allosteric effects

2. Experimental Setup Defects

• Finite size effects: Systematic study of finite-size effects on critical behavior lacking
• Insufficient pocket statistics: Rarity of pocket formation limits statistical samples
• Missing time scale analysis: Correspondence between simulation time and real biological time not explicitly established

3. Analysis Insufficiencies

• Missing quantitative predictions: Graphical method only gives qualitative trends, lacking quantitative prediction formulas
• Unexplored energy landscape: Systematic analysis of pocket energy barriers and stability lacking
• Weak universality arguments: Universality of conclusions across different network topologies not proven

4. Weak Biological Connections

• Vague parameter mapping: Correspondence between model parameters (e.g., coupling strength) and real biological quantities unclear
• Lack of experimental predictions: No quantitative experimental predictions directly testable
• Missing specific systems: No detailed simulation of any real signaling pathway

Impact

1. Contributions to the Field

• Theoretical framework: Provides new perspective for understanding biological functions of membrane criticality
• Methodology: MemTropy framework usable by other researchers for related problems
• Conceptual innovation: "Pocket" as non-equilibrium domain concept may inspire new experimental designs

2. Practical Value

• Drug development: Understanding membrane environment effects on signaling pathways may aid targeted therapy design
• Disease mechanisms: Functional consequences of membrane composition abnormalities (e.g., elevated cholesterol chemical potential in cancer)
• Synthetic biology: Design of membrane criticality-based biosensors and switches

3. Reproducibility

• Open-source code: Complete code and data available on GitLab
• Detailed methods: Algorithm details in supplementary material
• Clear parameters: Fitting parameters and confidence intervals fully reported
• Potential obstacles: Computational resource requirements may limit large-scale parameter sweep reproduction

Applicable Scenarios

1. Direct Applications

• Membrane protein signaling networks: B cell, T cell receptor signaling
• Membrane composition regulation studies: Cholesterol, fatty acid effects on signaling
• Temperature-sensitive processes: Membrane adaptation in ectothermic animals

2. Extended Applications

• Three-dimensional phase separation: Biomolecular condensates in cell nucleus and cytoplasm
• Interface processes: Condensate interactions with membranes, chromatin, and other surfaces
• Artificial membrane systems: Liposome drug delivery system design

3. Inapplicable Scenarios

• Systems far from critical point: Sensitivity will be substantially reduced
• Single protein behavior: Network effects insignificant
• Fast kinetic processes: Model assumes quasi-equilibrium state

References

Key Theoretical Foundations

1. Veatch et al. (2007, 2008): Experimental evidence of membrane critical fluctuations, PNAS and ACS Chem Biol
2. Machta et al. (2012): Critical Casimir forces in cell membranes, Phys Rev Lett
3. Hohenberg & Halperin (1977): Theory of dynamic critical phenomena, Rev Mod Phys

Biological Applications

4. Stone et al. (2017): Protein sorting by lipid phase domains in B lymphocytes, eLife
5. Shelby et al. (2023): Membrane phase separation drives receptor signaling domain assembly, Nature Chem Biol
6. Cammarota et al. (2020): Macrophage membrane lipid criticality reflects activation state, J Royal Soc Interface

Methodology

7. Kimchi et al. (2018): Allosteric regulation of ion channels, J Gen Physiol
8. Tasios et al. (2016): Critical Casimir interactions and colloidal self-assembly, J Chem Phys

Overall Assessment: This is a high-quality interdisciplinary theoretical research paper that successfully applies statistical physics critical phenomena theory to understanding protein-protein interaction networks in cell membranes. Main contributions lie in establishing a systematic theoretical framework and computational tools, revealing the potential of membrane criticality as a global regulatory mechanism. While quantitative prediction and experimental validation remain areas for improvement, this work opens new research directions at the intersection of membrane biophysics and cellular signal transduction, possessing significant theoretical value and potential application prospects.