DNA Instructions Turn Passive Microtubule Sheets Into Self-Morphing Active Matter

Living cells are masters of self-transformation. They read genetic instructions, translate them into molecular motors and cross-linkers, and use those proteins to generate the forces that shape division, migration and morphogenesis. For decades, synthetic biologists and soft-matter physicists have tried to capture pieces of this magic in the laboratory, but a stubborn gap has remained: in virtually every reconstituted active-matter experiment, the concentrations of motors and cross-linkers are fixed the moment the sample is assembled. The material can be pushed around with light, chemical fuels or rapid mixing, but the regulatory logic always lives outside the material itself. A team led by Rochelle Silverman and Alexandra Tayar at the Weizmann Institute of Science, together with Vincent Noireaux’s group at the University of Minnesota and theorist Hillel Aharoni, has now closed that gap. Writing in Nature Materials, they report a microtubule-based active material whose mechanical behaviour is written in DNA and executed autonomously, without any external command after the experiment begins.

The heart of the platform is cell-free gene expression, or CFE, a cell-free transcription-translation system rebuilt from E. coli lysate that can synthesize proteins on demand outside any living cell. Rather than using CFE merely to make regulatory proteins that diffuse around a chamber, the researchers embedded the reaction directly inside a dense network of stabilized microtubules. DNA plasmids added to the mixture encode kinesin motors, and as the lysate churns out new kinesin, the motors cluster and attach to the microtubule scaffold. The newly synthesized proteins are not passive reporters of the gene circuit; they become the structural and force-generating components of the material itself. Activity, in other words, stops being a parameter chosen by the experimenter and becomes a dynamic variable governed by the underlying biochemical circuit.

The experimental choreography is striking to watch. At early times, when kinesin levels are still low, the microtubule sheet sits in the well as a mechanically quiescent, load-bearing gel with negligible internal motion. After roughly two hours, accumulated kinesin forms the multimeric clusters needed to slide adjacent filaments, and the flat sheet begins to deform. It first aligns, then buckles out of plane into a labyrinth of blister-like protrusions roughly 50 micrometres wide, spaced about 200 micrometres apart. About half an hour after buckling begins, the blisters fracture near their apexes, where deformation is largest, and the fragments reorganize into independently flowing active bundles that sustain turbulent-like extensile flows for up to eight hours, at speeds of roughly 0.3 to 0.5 micrometres per second. The material has, in effect, bootstrapped itself from a passive gel into self-mixing active turbulence using nothing but its own genetic instructions.

The buckling instability itself follows elegant physics. As kinesin accumulates, motor activity builds an in-plane compressive stress in the sheet. When that stress reaches a critical value scaling with the square root of the product of bending rigidity and gravitational loading, the flat sheet becomes unstable and wrinkles. The selected wavelength is set by the balance between the sheet’s bending rigidity and gravity, and the predicted value of roughly 360 micrometres, for a soft gel with an elastic modulus near one kilopascal and a thickness near ten micrometres, agrees well with the observed 150 to 200 micrometres. Crucially, this wavelength stayed essentially constant across all tested DNA concentrations, even though activation times varied enormously. DNA concentration, the authors conclude, does not simply tune how active the material becomes; it controls the trajectory through mechanical phase space by determining when successive active states become accessible.

That distinction became quantitative when the team varied the kinesin-encoding DNA from 0.1 to 13 nanomolar. Higher DNA concentrations advanced the onset of motion from about four hours down to about one hour and accelerated the rise to steady-state velocity, yet the maximal velocity saturated at roughly 0.5 micrometres per second above three nanomolar DNA, corresponding to an estimated tetrameric kinesin cluster concentration of about 0.7 micromolar. Notably, the onset of flow was not triggered at a fixed kinesin concentration: samples with less DNA accumulated more motor protein before flowing, while samples with more DNA flowed earlier at lower motor levels. The transition to buckling is therefore governed by the kinetics of motor production rather than a single critical threshold, and the mean blister width shrank fourfold, from about 80 to about 20 micrometres, as DNA concentration rose, following a power law consistent with an active length scale set by the competition between motor-generated stress and elastic resistance.

To show that gene-circuit architecture, not just expression level, can program mechanical behaviour, the researchers built three regulatory designs. The first was simple constitutive kinesin expression under a T7 promoter, which produced sustained active turbulence, with onset times tunable from about seven hours at low DNA to about 45 minutes at high DNA. The second added mechanical feedback by co-expressing PRC1, a natural microtubule-bundling protein, under a bacterial sigma-70 promoter. At low PRC1 DNA the material flowed as before; at intermediate concentrations buckling was followed by complete arrest; and at high concentrations the sheet pulsed, entering turbulence briefly before friction from excessive bundling overwhelmed the kinesin-generated stresses and froze the network. A single biochemical dial, the PRC1 DNA dose, thus switched the material between flowing, pulsed and arrested states.

The third and most spectacular circuit introduced delayed negative feedback, the same design principle that drives biological clocks. A kinesin-sigma-28 fusion protein activates expression of the bacteriophage lambda CI repressor, which in turn shuts off kinesin production, while the ssrA degradation tags on both proteins allow the endogenous ClpXP protease to clear them. Depending on the repressor DNA concentration, the material either reached steady turbulence, oscillated rhythmically with a period of about 30 minutes, or delivered a single pulse before arresting. The mechanical oscillations are a direct readout of the genetic oscillator: rising kinesin drives flow, delayed repression cuts it off, degradation resets the loop, and the cycle repeats. A material that pulses on its own schedule, encoded entirely in its DNA, is a genuinely new kind of synthetic object.

Tying everything together is a coarse-grained reaction-mechanics model that couples mRNA and protein dynamics to active flow. Transcription and translation rates, mRNA and protein lifetimes, and the regulatory function of each circuit feed into a mapping from motor and cross-linker abundance onto flow velocity. The model quantitatively reproduced all three experimental architectures, and its simulations revealed subtleties invisible to the microscope, such as the fact that velocity saturation in the constitutive case arises from the nonlinear motor-flow relationship rather than from any cessation of protein production, and that pulses and arrest in the PRC1 circuit emerge despite monotonically accumulating proteins, purely from mechanical coupling. The simulations also confirmed that without ClpXP-mediated degradation, the oscillator collapses back into the constitutive steady-flow solution, pinpointing protein turnover as the ingredient that unlocks oscillatory active matter.

Geometry supplies a second, complementary control layer. When the team confined the active sheet in 3D-printed channels between 100 and 500 micrometres wide, blister-like deformations aligned perpendicular to the channel axis, and their constrained length grew with channel width before saturating near 300 micrometres, while their width stayed near 50 micrometres, an intrinsic active length scale. Patterned ridges imposed anisotropic confinement that accelerated buckling and stretched deformations to millimetre scales along unconstrained directions, independent of ridge curvature. The broader implications reach toward self-morphing engineered materials, autonomous soft robots and artificial cells whose mechanics are programmed like software: gene circuits decide when distinct mechanical states emerge, geometry decides how they are spatially organized, and the material itself carries the instructions for its own transformation.

Subject of Research: Genetically programmed self-morphing microtubule-based active matter driven by cell-free gene expression

Article Title: Gene-driven self-morphing microtubule-based active matter

Article References: Silverman, R., Zerbib, E., Garenne, D., Aharoni, H., Noireaux, V., & Tayar, A. M. (2026). Gene-driven self-morphing microtubule-based active matter. Nature Materials. https://doi.org/10.1038/s41563-026-02767-4

Image Credits: AI Generated

DOI: 10.1038/s41563-026-02767-4

Keywords: active matter, microtubules, kinesin, cell-free gene expression, synthetic biology, gene circuits, molecular motors, self-assembly, buckling instability, PRC1, oscillators, bioinspired materials