The Moment a Bubble Chooses a Side: RIKEN Rebuilds the Mechanics of Cellular Shape

A sphere acquires a direction

At first, the object under the microscope is almost offensively simple: a closed sphere only about the size of a cell. Its boundary is a lipid bilayer, the same basic architecture that separates living cells from the world. Suspended inside is a network of protein filaments. There is no nucleus, no genome, no metabolism, no sensory apparatus and no intention.

Then molecular motors begin to pull. The network contracts. The smooth outline breaks its symmetry, and one patch of membrane rounds outward like a small balloon emerging from a larger one. In the language of cell biology, it is a bleb. In the language of physics, a previously uniform object has selected a place where force, attachment and resistance no longer balance.

That small deformation is the center of a large biological question. Cells continually change shape as embryos develop, immune cells patrol tissue, wounds close, cancers invade, and one cell divides into two. The molecular inventory of those events is extensive, but an inventory is not an explanation. Knowing the proteins present does not reveal which physical relationships decide where a protrusion forms, how large it becomes, or whether the cell makes one front or several.

The international team led by Makito Miyazaki of the RIKEN Center for Integrative Medical Sciences and Taeyoon Kim of Purdue University chose to subtract almost everything. Their study, published in Science Advances, rebuilt a stripped-down actin cytoskeleton inside cell-sized liposomes and asked a severe question: how much of cellular shape can mechanics create by itself?

What the team put inside the artificial cell

A living animal cell has a thin, dynamic layer under its outer membrane called the actin cortex. It is made principally from actin filaments, myosin motors and proteins that organize or connect the network. The cortex supports the membrane, generates tension and helps the cell push, pull, squeeze and split.

RIKEN’s artificial cell began with a liposome: a closed vesicle made from a lipid bilayer. Into that vessel the team encapsulated purified actin filaments; myosin, a molecular motor that slides and contracts those filaments; α-actinin, which crosslinks actin and can connect it to the membrane; and ATP, the chemical energy source consumed by the machinery. ZIP kinase gradually phosphorylated and activated the myosin, imitating the controlled rise of motor activity in a cell.

One elegant piece of molecular engineering made the experiment unusually informative. The scientists added histidine tags to α-actinin and incorporated nickel–NTA lipids into the membrane. The interaction acted as an adjustable bridge between cytoskeleton and shell. Rather than merely asking whether the proteins could deform a vesicle, they could change the strength of force transmission in a systematic way.

As myosin activity rose, actin filaments pulled against one another. The network contracted; membrane and cytoskeleton exchanged force; blebs appeared spontaneously. A familiar piece of living-cell behavior had been reduced to purified parts sufficiently few that their roles could be disentangled.

These comparisons are only analogies; proteins and membranes are thermally active molecular materials, not miniature hardware. But the image captures the design problem. A motor can create force only if the network transmits it, and the network can reshape the membrane only if the two are coupled in the right way.

The three dials of shape

The study’s central achievement is not simply that the vesicles changed shape. Active vesicles have deformed before. It is the separation of three independently adjustable physical parameters and the connection of each to a different feature of blebbing.

The first result has a subtle shape. When membrane attachment was weak, deformation was small because contractile force could not be transferred effectively to the shell. As attachment strengthened, deformation grew. Yet stronger anchoring also meant that a larger contractile force was required before a bleb could begin. The coupling is therefore both transmission and restraint: it allows the network to move the membrane, while making local separation more difficult.

Changing the connectivity between actin filaments had a more limited effect on the overall amount of deformation. Its importance emerged elsewhere—in the way the system failed.

Peel or tear: both sides of a long argument were right

A bleb on a living cell expands when membrane locally loses the support of the actin cortex and internal pressure pushes the unsupported patch outward. Researchers have long debated the first event. Does the cortex detach from the membrane, leaving the membrane free to bulge? Or does the cortical network itself rupture under tension?

The minimal system showed both mechanisms. With relatively weak actin–membrane coupling, blebs tended to begin by detachment: the network locally peeled away from the membrane. With stronger coupling, the attachment could hold while the network itself gave way, producing rupture. Increasing actin-to-actin connectivity reduced rupture and favored detachment because a more integrated network was harder to tear.

The argument had persisted partly because living cells can use both routes. Their chemical signaling, membrane composition and cytoskeletal turnover vary across cell types and moments. By reducing the problem to two tunable mechanical relationships, the researchers converted apparently competing explanations into regions of one map.

This is more than a semantic reconciliation. It is a predictive framework. If scientists can estimate coupling and connectivity in a real cell, the model suggests which failure route should dominate. Conversely, a synthetic-cell designer can choose a regime rather than discovering it by accident.

The forgotten interior and the question of one front or many

Most textbook diagrams emphasize the dense actin cortex immediately beneath the cell membrane. Real cells also contain a sparser actin network throughout their interior. That interior mesh was easy to treat as background. The RIKEN–Purdue system made its role visible.

When actin was localized primarily near the membrane, single blebs appeared more often. When the filament network extended through the vesicle interior, multiple blebs became more frequent. A single protrusion is a basic example of symmetry breaking: a sphere that had no preferred direction acquires one side different from the others. Multiple protrusions distribute that directional choice.

The team’s simulations supplied a mechanical explanation. Myosin near the cortex tends to generate force largely parallel to the membrane. Motors in the interior pull in more diverse directions. Those forces can restrain the expansion of an existing bleb, preventing one site from monopolizing the deformation and making new blebs elsewhere more likely.

This turns the interior cytoskeleton from filler into a control layer. A cell may tune polarity not only through signals painted onto its membrane, but through the geometry of force generation across its whole volume.

A digital twin that can see the invisible forces

The experiment could show when and where a bleb appeared, but many decisive events occur below the resolution of direct observation. The scientists therefore developed a large-scale agent-based simulator in which individual actin filaments, myosin motors and the membrane were represented explicitly.

The model reproduced the major experimental outputs: the magnitude of deformation, the switch between detachment and rupture, and the transition from single to multiple blebs. It also exposed quantities that are extremely difficult to measure inside a tiny vesicle—tension on individual filaments, the direction and size of motor forces, and the specific filaments likely to break in the following seconds.

Agreement between experiment and simulation matters because a convincing picture is not yet a mechanism. Many different invisible causes can produce a similar visible bulge. A model forced to reproduce several outcomes under changing conditions is a harder test. Here it gave the same three control variables explanatory power across distinct phenomena.

The simulation is not a complete virtual cell. It omits the vast signaling networks, membrane traffic, polymerization cycles, fluid flows and organelles of living systems. Its value lies in matching the scale of the question: a molecularly explicit account of how a contractile network deforms one boundary.

From a red-blood-cell calculation to a programmable membrane

The story behind the experiment begins a century before its publication. In 1925, Evert Gorter and François Grendel extracted lipids from red blood cells and spread them on water. Their measurements—historically important though methodologically imperfect—helped establish the idea that the cell membrane is a two-molecule-thick lipid layer.

In the mid-1960s, Alec Bangham and colleagues observed phospholipids spontaneously forming closed vesicles in water. These structures became known as liposomes. Their importance rests on a profound convenience of chemistry: lipid molecules carry water-loving heads and water-avoiding tails, so under the right conditions they assemble into bilayer shells without a builder placing each molecule.

A membrane compartment could now be made in the laboratory. But a compartment alone was only a bag. The next challenge was to give it internal organization, energy use, chemistry and mechanics.

Japanese research supplied several early steps toward the specific problem of shape. In 1992, Hidetake Miyata and Haruhiko Hotani showed that polymerizing actin inside liposomes could alter their morphology and produce actin bundles. In 1999, Miyata and colleagues reported protrusive growth from giant liposomes driven by actin polymerization. Those experiments established that the cytoskeleton could push against an artificial membrane from within.

Over the following decades, teams reconstructed membrane-bound actin cortices, used actomyosin tension to remodel vesicles, produced bleb-like deformations in active cytoskeletal vesicles, and assembled contractile rings inside cell-sized compartments. Miyazaki’s own work demonstrated in 2015 that spherical confinement can organize actomyosin into spontaneous contractile rings—an important clue that cell geometry is not merely a container for molecular machinery but an active participant in its organization.

Two roads to an artificial cell

“Artificial cell” covers projects with very different starting points. A top-down approach begins with a living organism and rewrites or removes its genetic instructions. In 2010, the J. Craig Venter Institute created a bacterial cell controlled by a chemically synthesized genome. In 2016, the same program produced JCVI-syn3.0, a self-replicating cell with 473 genes—149 of which then had no known function. The lesson was humbling: even a genome stripped toward a minimum retained biological machinery researchers could not explain.

The bottom-up approach begins with nonliving molecules and tries to assemble selected cellular functions. Lipids make a compartment. Cell-free systems express genes. Cytoskeletal proteins create structure and force. Signaling molecules connect detection to response. Each module can be studied before researchers try to integrate them.

RIKEN’s new work belongs firmly to the bottom-up tradition. Its artificial cell does not descend from a bacterium and has no minimal genome. Its power comes from the opposite move: it asks what a membrane plus a selected mechanical network can do without the rest of life.

The two roads answer complementary questions. Top-down work asks what a living cell can lose and remain alive. Bottom-up work asks what nonliving matter must gain before life-like functions emerge. Between them sits the central engineering challenge of synthetic biology: integration. A module that changes shape must eventually coordinate with energy regeneration, sensing, gene expression, membrane growth and division.

Why blebs matter far beyond an artificial vesicle

For years, membrane blebs were associated most visibly with apoptosis, the orderly program of cell death. Dying cells often develop dramatic surface bubbles. But blebs are not merely death throes. They contribute to cell division, development and migration, particularly when cells move through confined three-dimensional environments.

A crawling cell on a flat laboratory dish can extend broad actin-rich sheets called lamellipodia and grip the surface. Inside tissue, the landscape is different. Pores are narrow, resistance varies and adhesive footholds may be scarce. Some immune, embryonic and cancer cells use rounded, pressure-driven protrusions as part of amoeboid movement. Actomyosin contractility raises internal pressure; a local weakening of membrane–cortex support creates a bleb; a new cortex forms beneath it; and coordinated cycles can contribute to forward motion.

This flexibility is clinically important. Cancer cells can switch migration modes when blocked by a drug or constrained by their surroundings. Bleb-based motility can assist invasion without requiring the same extensive matrix degradation as some elongated forms of movement. Understanding the mechanics does not immediately produce an anticancer treatment, but it offers a cleaner framework for asking why a cell adopts one mode, where a protrusion appears and which physical link is vulnerable.

The same mechanics belongs to healthy biology. Immune surveillance, embryonic patterning and wound closure depend on cells finding paths and generating forces without tearing themselves apart. A design map derived from a minimal system can help identify which observations in living cells require biochemical signaling and which can arise directly from material properties.

The hardest word in the announcement is “toward”

RIKEN describes the work as progress toward artificial cells that move and divide. The direction is credible; the distance is substantial.

A bleb is not locomotion. For a protrusion to move a cell, its formation must be polarized, coupled to friction or adhesion against the environment, coordinated with rear contraction and repeated with feedback. The artificial vesicle in this study can break symmetry, but it does not sense a destination, steer around an obstacle or replenish the ATP it consumes.

A bleb is also not division. Division requires reliable placement of a constriction, coordination with membrane area and internal contents, closure without catastrophic leakage, and fair enough partitioning that both daughters continue functioning. Synthetic systems have reconstituted contractile rings and bacterial division proteins in vesicles, yet autonomous cycles of growth, genome replication and robust fission remain a frontier.

The team’s planned next step is therefore logical: add receptors and signaling systems to the mechanical core. In a living cell, chemistry tells mechanics when and where to act, while force feeds information back into chemistry. Reconstructing that loop would move the system from spontaneous deformation toward responsive behavior.

From synthetic biology to soft materials

An artificial cell need not become fully alive to become useful. A vesicle that senses a cue, changes shape and releases cargo could serve as a responsive delivery system. A population of deformable compartments could become a model tissue for studying how local forces produce larger structures. Molecular simulations calibrated to minimal experiments could improve predictions of how real cells respond to drugs or confinement.

The same principles may inform soft materials. Conventional machines separate actuator, frame, sensor and controller. Cells combine those roles in matter that reorganizes itself. An actomyosin network embedded in a membrane is both structure and motor. Change its attachment or geometry, and the behavior of the whole object changes.

That makes the three-parameter map a possible design language. Coupling controls how force reaches a boundary. Connectivity determines whether the network holds, tears or peels. Spatial distribution decides whether deformation concentrates or proliferates. The eventual materials may not use actin or myosin at all, but the relationships can carry over to active gels, microscopic robots and adaptive interfaces.

There are also safety and definition questions. A responsive soft material is not necessarily an organism; an engineered cell-derived system may be alive without being autonomous; and adding replication changes both capability and risk. Clear language matters because “artificial life” can outrun what an experiment has shown. In this case, the most accurate description is also the most intellectually interesting: researchers isolated a piece of cellular mechanics and discovered a compact set of rules inside it.

The value of building less

Living cells achieve reliability through intimidating complexity. Thousands of molecular species interact, compensate, amplify and repair. That complexity is the source of life’s power and a barrier to understanding it. Remove one protein from a cell and ten pathways may adapt. Observe a protrusion and dozens of signals may already have converged on the site.

Bottom-up reconstruction reverses the problem. It begins with a system too simple to live and adds only what is necessary for the phenomenon under study. Failure becomes informative. If the membrane does not move, force may not be transmitted. If the cortex tears, connectivity may be too weak relative to anchoring. If multiple blebs appear, the internal distribution of motors may be redirecting stress.

RIKEN’s artificial cell is compelling precisely because it is incomplete. Its smooth membrane carries no evolutionary history; its proteins cannot hide behind the redundancy of a living cytoplasm. In that stripped-down space, a century of membrane science and decades of cytoskeletal reconstitution converge on a visible event: a sphere develops a side.

Life has not been created in that moment. But one of life’s recurring gestures has been made legible. A cell changes shape not because any single molecule commands it, but because motors, filaments, anchors and a membrane negotiate where force can go. The next artificial cell will need chemistry that can conduct that negotiation in response to the world. RIKEN’s work provides something it did not have before: a mechanical grammar.