🔬 condensed matter

Liquid-Liquid Phase Separation in a Minimal Explicit-Solvent Lattice Model Mimicking Protein Solutions

This study employs a minimal explicit-solvent lattice model with quenched disorder to demonstrate how protein-solvent, protein-crowder, and protein-protein interactions collectively regulate the phase behavior, morphology, and stability of biomolecular condensates, offering design principles for stimuli-responsive systems.

Original authors: Siddhartha Roy, Rakesh S. Singh

Published 2026-05-26

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Siddhartha Roy, Rakesh S. Singh

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the inside of a cell not as a clear soup, but as a bustling, crowded city. In this city, there are three main types of "citizens":

Proteins: The workers, who can change their shape (like folding a piece of paper or unfolding it).
Solvent (Water): The space they move through, which we'll think of as the "air" or "room" they occupy.
Crowders: The static obstacles, like giant, unmoving statues or buildings that are stuck in place and take up space.

This paper is a computer simulation of how these citizens interact to form "condensates." Think of a condensate as a spontaneous gathering or a club meeting where proteins decide to huddle together, separating themselves from the rest of the room. The researchers wanted to understand what makes these gatherings form, break apart, or change their shape.

The Simulation Setup

The scientists built a simplified "grid world" (like a chessboard) to watch this happen.

The Proteins: They can be in two states: Native (folded up tight, like a compact ball) or Unfolded (loose and spread out). They can switch between these states depending on the temperature.
The Solvent: Instead of simulating every single water molecule (which would be too slow), they treated the water as "packets" of space, similar in size to the proteins.
The Crowders: These are "pinned" particles stuck to the grid. They don't move, but they interact with the proteins, acting like obstacles in the room.

What They Discovered

1. The Temperature Dance (Phase Behavior)

The researchers found that by changing how much the proteins "like" the water (protein-solvent interaction), they could make the proteins behave in three very different ways as the temperature changed:

The "Reentrant" Dance (The Loop): Imagine a group of people who start by huddling together in the cold (Phase 1). As it gets warmer, they get comfortable and spread out into the room (Phase 2). But if it gets too hot, they suddenly huddle together again, but this time they are in a different "outfit" (unfolded state) (Phase 3). Finally, if it gets extremely hot, they all scatter and mix completely (Phase 4). This is called "reentrant" behavior because they leave the crowd, come back, and then leave again.
The "Upper Critical" Story (UCST): In some scenarios, the proteins only huddle together when it's cold. As soon as it warms up, they mix perfectly and stay mixed forever.
The "Closed Loop" Story: Here, the proteins mix at low temperatures, huddle together in the middle temperatures, and then mix again at high temperatures. It's like a loop where they are only together for a specific "Goldilocks" zone of temperature.

The Secret Sauce: The paper explains that these weird loops happen because the proteins change their shape (folded vs. unfolded) as the temperature changes. The "folded" version likes to stick together at low temps, but the "unfolded" version likes to stick together at high temps. The competition between these two shapes creates the complex loops.

2. The Role of the "Crowders" (Static Obstacles)

When the researchers added the "crowders" (the stuck statues), the gatherings changed shape dramatically.

Partial Wetting: Sometimes, the protein huddle would only cover part of a statue, like a blanket draped over one side of a chair.
Full Wetting: Other times, the proteins would completely wrap around the statue, hiding it from view.
Segregation: In some cases, the proteins would form their own separate island, refusing to touch the statue at all, even though the statue was right there.

The key finding here is that how much the proteins like to touch the statues determines whether they wrap around them, ignore them, or only touch them partially.

3. The Binary Mixture (Two Types of Workers)

The scientists also simulated a room with two different types of proteins (let's call them Type A and Type B).

Without Crowders: Depending on how much A and B liked each other (and how much they liked the water), they formed different structures:
- Fully Wetted: Type B completely surrounded Type A.
- Partially Wetted: Type B only covered part of Type A.
- Segregated: Type A and Type B formed two separate, distinct islands that didn't touch.
- Associative: They mixed together into a single, happy blob.
With Crowders: Adding the statues made the picture even more complex. The statues could act as a "shield." If Type B proteins loved the statues, they would coat the statues, effectively blocking Type A proteins from getting close. This created new, complex shapes where one type of protein was trapped inside a layer of the other type.

The Main Takeaway

The paper concludes that the shape and stability of these protein gatherings aren't just about the proteins sticking to each other. They are a delicate balance of three things:

How proteins stick to each other.
How proteins interact with the water (solvent).
How proteins interact with the obstacles (crowders).

By tweaking these interactions (like changing the protein's "personality" or the environment's "temperature"), you can engineer these gatherings to be round, layered, segregated, or mixed. The study suggests that in real cells, the cell likely uses these exact interactions to control where and how these protein clubs form.

Technical Summary: Liquid-Liquid Phase Separation in a Minimal Explicit-Solvent Lattice Model Mimicking Protein Solutions

Problem Statement
Biomolecular condensates, formed via liquid-liquid phase separation (LLPS), are critical to cellular function, yet their assembly and rational design principles remain incompletely understood. A significant gap exists in understanding how solvent properties and intracellular heterogeneity (crowding) modulate LLPS, particularly regarding the interplay between conformation-dependent protein states and environmental factors. While previous studies have utilized implicit-solvent models, these often fail to capture the specific mechanisms by which solvent modulates interactions through conformational changes. Furthermore, the intracellular environment is inherently heterogeneous and often out-of-equilibrium due to active processes, necessitating models that explicitly account for solvents and static disorder (crowders) to explain complex phase behaviors such as Upper Critical Solution Temperature (UCST), Lower Critical Solution Temperature (LCST), closed-loop, and reentrant transitions.

Methodology
The authors developed a minimal explicit-solvent lattice-gas (LG) model based on a Hamiltonian with quenched disorder.

System Representation: Proteins are modeled as particles with two internal conformational states: a non-degenerate ground state ("native," n) and a degenerate excited state ("unfolded," u). Solvents are represented as particles occupying lattice sites, defined as mesoscopic volume elements equivalent to the excluded volume of a protein unit. Static disorder (crowders) is implemented as clusters of pinned particles that interact with proteins and solvents but do not move.
Hamiltonian: The system energy includes terms for protein-protein interactions (dependent on conformational state), protein-solvent interactions, protein-crowder interactions, solvent-solvent interactions, and crowder-solvent interactions. Internal energy differences between conformational states are also included.
Simulation: Monte Carlo simulations in the canonical ensemble were employed. Dynamics included Kawasaki exchange (particle translation) and Metropolis excitation/de-excitation (conformational interconversion). Both equilibrium and non-equilibrium (driven) conditions were explored.
Parameters: The study systematically varied heterotypic interaction parameters (protein-solvent and protein-crowder) while keeping homotypic protein-protein and solvent-solvent interactions fixed. Both single-component protein systems and binary protein mixtures were analyzed.

Key Contributions and Results

Phase Behavior in Single-Component Systems:
- The model successfully reproduces diverse phase diagrams under varying protein-solvent interactions, including reentrant, UCST, and closed-loop (containing both LCST and UCST) behaviors.
- Reentrant Behavior: Observed when specific protein-solvent interaction strengths are tuned. The system transitions from a native-dominated demixed state (P1) to a mixed state (P2) at higher temperatures, then to an unfolded-dominated demixed state (P3), and finally to a homogeneous state (P4) at very high temperatures. This is driven by the abrupt shift in conformational population from native to unfolded states as temperature increases, altering the dominant interaction energy scale.
- Mechanism: The origin of these transitions is attributed to the competition between conformation-dependent protein-protein interactions and protein-solvent interactions. Increasing protein-solvent interaction strength generally lowers critical solution temperatures, enhancing solubility.
- Non-Equilibrium Effects: The study notes that non-equilibrium conditions (simulating active cellular noise) play a key role in shaping phase behavior, acting as regulatory parameters alongside solvent interactions.
Role of Quenched Disorder (Crowders):
- The introduction of static impurities (crowders) significantly alters condensate morphology without necessarily changing the fundamental topology of the phase diagram.
- Morphological Diversity: Depending on the relative strength of protein-crowder interactions ( $J_{cn}$ vs. $J_{cu}$ ), the model exhibits partial wetting, complete wetting, and segregated (de-wetted) morphologies.
- Screening Effects: Strong interactions between native proteins and crowders can "screen" the crowder surface, preventing unfolded proteins from wetting the cluster, leading to segregated aggregates.
Binary Protein Mixtures:
- In binary mixtures (Proteins A and B) without crowders, the system exhibits a wide range of morphologies: partially wetted, fully wetted, segregative, and associative.
- Sensitivity: These morphologies are delicately sensitive to protein-solvent interaction strengths. Increasing the interaction strength of one protein type with the solvent can shrink the stability regions of fully wetted and associative phases while expanding segregated phases.
- With Crowders: The presence of quenched disorder introduces a broader spectrum of complex morphologies (e.g., competitive wetting, selective de-wetting). The interplay between heterotypic protein-protein, protein-solvent, and protein-crowder interactions dictates whether the cluster is fully wetted, partially wetted, or if specific protein species are excluded from the cluster surface.

Significance and Claims
The paper claims that this minimal explicit-solvent model provides a robust framework for understanding how protein-solvent and protein-crowder interactions, in conjunction with protein-protein interactions, act as key regulatory parameters for modulating condensate morphology and stability.

Design Principles: The work underscores that tuning solvent properties (or protein sequence to alter solvent affinity) and accounting for intracellular heterogeneity are essential for designing stimuli-responsive condensates.
Mechanistic Insight: It elucidates that anomalous phase behaviors (like reentrance) arise from a delicate balance between conformational population shifts and interaction energy scales.
Future Directions: The authors suggest that these insights can guide future computational and experimental studies aimed at engineering specific biomolecular condensate morphologies. They propose that extending the model to include multiple internal states (e.g., via a q-state Potts model) and incorporating dynamical heterogeneities are logical next steps.

The study concludes that the interplay of these specific interactions within a multicomponent, heterogeneous medium is sufficient to generate the complex phase behaviors observed in biological systems, offering a pathway to rational design in biomolecular engineering.