Blinking membrane patterns induced by protein binding/unbinding

arXiv:2608.13016 · cond-mat.soft, nlin.PS, physics.bio-ph · Submitted 2026-08-13 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "Blinking membrane patterns induced by protein binding/unbinding".

Mira: Here's the gist of the paper. Blinking membrane patterns induced by protein binding/unbinding investigates spatiotemporal domain patterns caused by cyclic protein binding to one side of membranes, revealing that blinking domains can form even in tensionless membranes Mechanism of Blinking Domains The study utilizes a meshless membrane model with three states (one unbound and two bound) as an off-lattice version of an active Potts model The proteins have two states with different isotropic spontaneous curvatures of the same sign, and cyclic binding/unbinding occurs due to chemical potential differences, behaving as an off-lattice active Potts model This cyclic process induces blinking domains where convex domains of proteins with a higher spontaneous curvature grow and subsequently change to the other state with a lower spontaneous curvature, resulting in domain shrinkage, which repeats The membrane bending keeps the domain location unchanged, meaning these standing-wave-like patterns have not been obtained in either lattice or off-lattice active Potts models Equilibrium Phase Behavior In thermal equilibrium, hexagonal convex domains are formed by the competition between bending and surface tension energies, which are stably formed only under positive surface tension The simulation explores phase behaviors at reference states, where the bound state forms convex domains with a hexagonal alignment at intermediate chemical potential values The HD phase displays a large vertical span, whereas E0 and E2 exhibit small spans Both the E0–HD and HD–E2 transitions are of the first order, with two phases coexisting around the transition points through hysteresis Nonequilibrium Pattern Formation In nonequilibrium conditions, specifically when the chemical potential imbalance is positive (µ02 − µ01 − µ12 > 0), a novel dynamic mode called BD emerges at specific chemical potential values In this BD mode, the sizes of convex domains change cyclically as they grow toward their stable size, subsequently changing to the s = 1 state, and the resultant s = 1 domains shrink because they are unstable compared with unbound membranes The BD pattern is a stationary wave because the membrane deformation acts as a memory of the previous pattern, unlike lattice models where domain boundaries propagate Dynamic Mode Transitions The dynamic phase diagram at γ = 1 shows transitions between different modes, including the HD and BD patterns The HD phase can continuously change to the BD mode, as evidenced by the behavior in Fig.

Title and authors: Kai: So we're looking at this paper today about blinking membrane patterns induced by protein binding/unbinding. It's a study on how cyclic protein binding and unbinding can create these specific spatial domain patterns on membranes, and the authors are using a meshless membrane simulation to look at it.

Mira: It’s interesting because they found that these blinking domains can actually form even in tensionless membranes, which is a key point because most models we use usually require some surface tension to get those kinds of structures.

Lev: That’s what caught my attention immediately, because when you think about real experimental setups, getting stable patterns without some tension is often the hardest part of the puzzle.

Kai: Exactly. The paper explains that they model the proteins as having two states with different spontaneous curvatures, and this cyclic binding and unbinding happens because of differences in chemical potential between those states, which is treated like an off-lattice active Potts model according to them.

Mira: And what they found is that this cycle induces blinking domains where convex domains of proteins with a higher spontaneous curvature grow, then switch to the other state with a lower spontaneous curvature, causing the domain to shrink repeatedly.

Lev: So it’s like a kind of dynamic oscillation in the domain size, and the membrane bending itself doesn't change where that pattern is located, which they call standing-wave-like patterns.

Kai: Right. And that’s what makes it different from things we usually see in lattice or off-lattice active Potts models, because those models don't get this kind of standing wave behavior when you consider the membrane bending effect.

Mira: Now, if we look at the thermal equilibrium part, they say that hexagonal convex domains form when there is a competition between bending and surface tension energies, and those are only stably formed if you have positive surface tension.

Lev: That makes sense from a classical mechanics standpoint—you need some energy input to stabilize that structure against the natural tendency of the membrane to minimize its area.

Title and authors: Kai: But then they explore phase behaviors at reference states, where the bound state forms convex domains with a hexagonal alignment at intermediate chemical potential values, and they show that the HD phase has a large vertical span, while E zero and E two exhibit small spans.

Mira: The transitions between these phases are described as first order, meaning there's hysteresis around those transition points where you can have two phases coexisting.

Lev: Hysteresis is always a real headache when designing any physical system; it means the path you take to get from one state to another matters significantly for what you end up with.

Kai: Then they move into nonequilibrium pattern formation, specifically when the chemical potential imbalance is positive, meaning the difference between mu two and mu one minus mu twelve is greater than zero <ref:2608.13016#pg2>.

Mira: In that nonequilibrium regime, they introduce a new dynamic mode called BD at specific chemical potential values. In this BD mode, the sizes of those convex domains change cyclically as they grow towards their stable size, then switch to the s equals one state, and those s equals one domains shrink because they are unstable compared to unbound membranes.

Lev: So the BD pattern is described as a stationary wave here because the membrane deformation acts like a memory of the previous pattern, which is a big departure from lattice models where you usually see domain boundaries propagating.

Kai: That memory effect seems crucial for understanding why these patterns can stand still on tensionless membranes, and that’s what makes this work so compelling for AI systems trying to predict behavior in those scenarios.

Mira: Moving onto the improvements they suggest, one key improvement is incorporating memory effects into AI systems so they can accurately predict domain behavior in tensionless membranes by accounting for previous states.

Lev: That would mean the AI needs a way to store and use past configurations, which adds complexity to the computational overhead you mentioned earlier.

Kai: Another suggestion is that improved AI systems should be able to distinguish between different pattern types by calculating a time fraction of three-state coexistence, using the criterion that BD emerges if that fraction is greater than zero point two.

Mira: That gives us a concrete metric to use; it’s not just looking at the final state, but how much time the system spends oscillating through those intermediate states.

Title and authors: Lev: I wonder how robust that three-state coexistence calculation is when you try to run it on real hardware where you have finite time steps and noise in your measurements.

Kai: The paper also suggests that improved AI systems could predict vesicle formation from any initial membrane condition at tensionless conditions, because they state that the BD mode is a robust dynamic mode that can emerge even when the mechanical surface tension gamma is zero.

Mira: That would be huge for biological modeling, since vesicles are fundamental structures in cell membranes and this suggests they might form under conditions where we usually expect instability.

Lev: If it truly emerges at gamma equals zero, then the overhead for running that simulation on hardware should be manageable if the system is stable enough to start in that regime.

Kai: Finally, the paper suggests improved AI systems should be able to model transitions between phases by analyzing how these things depend on chemical potentials and surface tensions, using their phase diagram to predict whether a system will enter one of the HD, BD, FD, or E one modes.

Mira: It sounds like the main implication is that we need AI capable of navigating this complex phase space defined by chemical potentials and tension to accurately predict which dynamic mode will dominate under different physical conditions.

Lev: So it’s pushing us toward building more sophisticated error-correction schemes that can handle these kinds of emergent, memory-dependent dynamics rather than just steady states.

Kai: It's definitely a shift from just looking at static structures to understanding the underlying active, dynamic processes that generate those structures.

Mira: And this whole paper on blinking membrane patterns induced by protein binding/unbinding gives us a new language for describing how localized molecular interactions can drive complex spatiotemporal patterns on surfaces.

Lev: It opens up new avenues for modeling active matter systems, which is where we see a lot of dynamic behavior in our error-correction research too.

Kai: Anyway, that covers what they found about the mechanism and what they think needs to be improved for computational prediction.

The paper's summary: Kai: So, basically, this paper is about how proteins sticking to and letting go of a membrane can create these specific blinking patterns on that surface.

Mira: Right, so they're looking at those cyclic binding and unbinding events and seeing how that drives the spatial pattern changes across the membrane.

Lev: What’s really interesting is their model. They use this meshless membrane thing, which is like an off-lattice version of an active Potts model to simulate it.

Kai: Exactly, they treat the proteins as having two states with different curvatures and they have them switching based on chemical potential differences between those states.

Mira: And that switching causes these convex domains to grow and then shrink as they cycle through the bound and unbound states.

Lev: It’s like a continuous oscillation in domain size, but here’s the kicker, the membrane bending itself keeps that pattern from moving around.

Kai: That means they get these standing-wave-like patterns that you don't usually see when you look at standard lattice or off-lattice active Potts models.

Mira: That’s a big deal because it shows this specific dynamic behavior can happen even on tensionless membranes, which is a condition we often struggle to model accurately.

Lev: And they looked at the equilibrium side too, where they said hexagonal convex domains form only when you have positive surface tension, which makes sense for classical energy minimization.

Kai: But then they explore nonequilibrium conditions where the chemical potential imbalance is positive, and that’s when this new dynamic mode called BD appears at specific chemical potential values.

Mira: In this BD mode, the domain sizes change cyclically as they grow towards a stable size, then switch to another state which shrinks because it’s unstable compared to being unbound.

Lev: So the BD pattern is stationary because the membrane deformation acts like a memory of what happened before, unlike lattice models where boundaries just keep moving around.

Kai: That memory effect seems key for why these patterns can stand still on tensionless membranes, which points toward how AI systems need to be built to predict this stuff accurately.

Mira: The authors suggest that better AI needs to incorporate that memory effect directly so it can predict domain behavior in tensionless membranes without needing surface tension to be present.

The paper's improvements: Tom: So, we’re looking at how they suggest making these kinds of predictions better by improving their AI models for this blinking membrane stuff.

Kai: Basically, they say that improved AI systems need to incorporate memory effects from previous states so they can predict domain behavior in tensionless membranes more accurately.

Mira: That makes sense because the paper showed that the membrane bending keeps the pattern location unchanged, so if you don't include that history in your AI, it won't work right.

Lev: I agree, and from an engineering standpoint, it means we need to build a system where past configurations are stored and used as input for future predictions instead of just looking at the current snapshot.

Kai: They also suggest that improved AI systems can distinguish between different pattern types by calculating a time fraction of three-state coexistence.

Mira: So, they propose using a threshold like BD if that fraction is greater than zero point two to tell blinking patterns apart from other modes.

Lev: That gives us a concrete metric to use for classification, which is better than just looking at the final state because it captures the dynamic process itself.

Kai: It’s also said that improved AI systems could predict vesicle formation from any initial membrane condition at tensionless conditions.

Mira: Since they showed that the BD mode is a robust dynamic mode that can emerge even when surface tension is zero, this implies vesicles might form under conditions we usually think are unstable.

Lev: If it truly emerges at zero surface tension, then running that simulation on real hardware should be manageable if the system is stable enough to start in that regime.

Kai: Finally, they suggest improved AI systems should model transitions between phases by analyzing how those things depend on chemical potentials and surface tensions using the phase diagram.

Mira: So, they want AI to use the whole phase diagram—the HD, BD, FD, and E one modes—to predict which dynamic mode will dominate based on the specific chemical potential values you start with.

Lev: It sounds like it’s pushing us toward building more sophisticated error-correction schemes that can handle these kinds of emergent, memory-dependent dynamics rather than just steady states.

Conclusion: Kai: So, to wrap up, this paper on "Blinking membrane patterns induced by protein binding/unbinding" shows that cyclic protein binding drives specific spatial domain patterns on membranes even when surface tension is zero.

Mira: It really highlights how these molecular interactions can create complex spatiotemporal structures on surfaces in ways we weren't expecting from simpler models.

Lev: The real impact here is showing that this dynamic mode, the BD pattern, can emerge robustly at zero surface tension, which changes how we think about vesicle formation under different physical constraints.

Kai: We also see suggestions for improving the AI tools used to predict these behaviors by incorporating memory effects and using metrics like three-state coexistence time fractions.

Mira: It means future AI models should be able to navigate the phase space of chemical potentials and surface tensions to predict whether a system will settle into that blinking mode or something else entirely.

Lev: For us, it’s about building error correction for hardware that can handle these kinds of memory-dependent dynamics instead of just steady states, which is where the real engineering challenge lies.

Kai: It’s fascinating how much detail they managed to get on the mechanism of these standing wave-like patterns without needing those traditional lattice assumptions.

Mira: And it gives us a new framework for describing how localized molecular binding can drive complex surface organization in active matter systems.

Lev: We should probably look at that BD mode more closely, especially since it’s linked to conditions where we usually expect instability.

Institute for Solid State Physics, University of Tokyo

cond-mat.soft, nlin.PS, physics.bio-ph

Submitted: 2026-08-13

Updated: 2026-10-16

Comments: 8 pages, 10 figures

Journal ref: H. Noguchi, Soft Matter 22, 7291-7298 (2026)

DOI: 10.1039/D6SM00785F

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 83/100

The gist: Blinking membrane patterns induced by protein binding/unbinding investigates spatiotemporal domain patterns caused by cyclic protein binding to one side of membranes…

Key concepts

Blinking Domains
These are dynamic regions on the membrane where protein domains repeatedly grow and then shrink due to cyclic binding and unbinding of proteins. This process creates a repeating pattern, similar to a blink, where convex domains change shape and size over time.
Active Potts Model (Off-lattice)
This is a mathematical model used to simulate the membrane dynamics. It treats the proteins as having two states with different spontaneous curvatures that interact based on chemical potential differences. This allows researchers to study how these protein interactions lead to domain formation and evolution.
BD Mode
The BD mode is a novel dynamic pattern that emerges under nonequilibrium conditions when there is an imbalance in chemical potentials. In this mode, convex domains grow toward a stable size before switching to the s=1 state and shrinking, acting as a stationary wave because the membrane bending maintains its location.
Memory Effect
The membrane's bending acts as a memory for previous patterns. This means that the physical deformation of the membrane influences where new domains form, allowing AI systems to better predict domain behavior by incorporating information about past states.

Terminology

Summary

Blinking membrane patterns induced by protein binding/unbinding investigates spatiotemporal domain patterns caused by cyclic protein binding to one side of membranes, revealing that blinking domains can form even in tensionless membranes

Mechanism of Blinking Domains

The study utilizes a meshless membrane model with three states (one unbound and two bound) as an off-lattice version of an active Potts model The proteins have two states with different isotropic spontaneous curvatures of the same sign, and cyclic binding/unbinding occurs due to chemical potential differences, behaving as an off-lattice active Potts model This cyclic process induces blinking domains where convex domains of proteins with a higher spontaneous curvature grow and subsequently change to the other state with a lower spontaneous curvature, resulting in domain shrinkage, which repeats The membrane bending keeps the domain location unchanged, meaning these standing-wave-like patterns have not been obtained in either lattice or off-lattice active Potts models

Equilibrium Phase Behavior

In thermal equilibrium, hexagonal convex domains are formed by the competition between bending and surface tension energies, which are stably formed only under positive surface tension The simulation explores phase behaviors at reference states, where the bound state forms convex domains with a hexagonal alignment at intermediate chemical potential values The HD phase displays a large vertical span, whereas E0 and E2 exhibit small spans Both the E0–HD and HD–E2 transitions are of the first order, with two phases coexisting around the transition points through hysteresis

Nonequilibrium Pattern Formation

In nonequilibrium conditions, specifically when the chemical potential imbalance is positive (µ02 − µ01 − µ12 > 0), a novel dynamic mode called BD emerges at specific chemical potential values In this BD mode, the sizes of convex domains change cyclically as they grow toward their stable size, subsequently changing to the s = 1 state, and the resultant s = 1 domains shrink because they are unstable compared with unbound membranes The BD pattern is a stationary wave because the membrane deformation acts as a memory of the previous pattern, unlike lattice models where domain boundaries propagate

Dynamic Mode Transitions

The dynamic phase diagram at γ = 1 shows transitions between different modes, including the HD and BD patterns The HD phase can continuously change to the BD mode, as evidenced by the behavior in Fig.

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 AI systems can accurately predict domain behavior in tensionless membranes by incorporating memory effects from previous states, as The membrane bending keeps the domain location unchanged is observed in blinking domains.

  2. Improved AI systems can distinguish between pattern types by calculating a time fraction of three-state coexistence, using the criterion "BD if f3co > 0.2" to identify blinking patterns versus other modes.

  3. Improved AI systems can predict vesicle formation from any initial membrane condition at tensionless conditions, as BD is a robust dynamic mode that can emerge even at gamma = 0 and vesiculation occurs from any of the initial membrane conditions.

  4. Improved AI systems can model transitions between phases by analyzing the dependence on chemical potentials and surface tensions, utilizing the phase diagram to predict whether a system will enter the HD, BD, FD, or E1 modes.

Abstract

Nonequilibrium membrane pattern formation is studied using meshless membrane simulation. Bound proteins are considered to have two states that generate different membrane spontaneous curvatures. Protein binding and unbinding occur cyclically owing to chemical potential differences, as an off-lattice active Potts model. It is found that this cyclic binding/unbinding can induce blinking domains, with oscillating size: convex domains of the proteins with a higher spontaneous curvature grow, and subsequently, the proteins change to the other state with a lower spontaneous curvature, resulting in domain shrinkage. These processes repeat. In thermal equilibrium, hexagonal convex domains are formed by the competition between bending and surface tension energies, so that they are stably formed only under positive surface tension. However, blinking domains can form even in tensionless membranes.

Related papers