Mutations Give RNA Droplets Their Own Genes, Vacuoles and Movement

How simple can a cell be and still count as alive? For researchers trying to build life from the bottom up, the question hinges on a deceptively short checklist: a minimal cell must store genetic information, express it as a physical phenotype, keep that genotype physically linked to its compartment, and allow mutations to produce heritable, selectable variation. Building such a system from phospholipid membranes and proteins is daunting, with estimates suggesting that transcription and translation alone would require at least 150 genes, and membrane growth and division still lacking a sufficient rate for sustained cycles. A team of researchers from Heidelberg University, the Max Planck School Matter to Life and the University of Augsburg now reports in Nature Chemical Engineering that RNA droplets, assembled cotranscriptionally from DNA templates, satisfy all three core prerequisites for Darwinian evolution, and that a surprisingly small mutational load can endow these droplets with stable internal compartments and even micrometer-scale movement.

The system centers on an RNA nanostar, a single RNA strand that folds cotranscriptionally into four arms separated by single uracils, each arm terminating in a self-complementary kissing-loop motif. These kissing loops allow nanostars to bind one another and, at sufficient concentration, to condense into liquid droplets through liquid-liquid phase separation. Unlike most nucleic acid condensates, which assemble through nonspecific electrostatic interactions between polycations and polyanions, these droplets are held together by sequence-specific base pairing, which means the droplet’s material properties are written directly into its sequence. Two arms carry 25-base-pair duplex stems, a third carries a malachite green aptamer for fluorescence imaging, and the fourth carries a biotin aptamer for recruiting molecular targets. Crucially, the DNA template that encodes the nanostar remains in the reaction mixture as the droplets form.

The first question the team addressed was whether the droplets actually trap their own genes. By labeling the DNA template with fluorescein-labeled dUTP and imaging with confocal microscopy, the researchers found that the DNA is localized within the RNA droplets rather than dispersed in solution. The DNA is not distributed evenly; instead it forms hotspots one to three micrometers across, likely arising from sequence-specific RNA-RNA interactions combined with electrostatic repulsion between DNA and RNA. A plausible mechanism is cotranscriptional capture: nascent nanostar RNA folds and engages in multivalent interactions while still tethered to the DNA through the RNA polymerase, dragging the template into the condensate as it assembles. This establishes a physical genotype-phenotype link, the first prerequisite for an evolvable system.

With that link in place, the researchers asked whether mutations in the DNA would change the droplet phenotype. Exploiting the self-complementarity of the template, they used PCR to generate a pool of mutated genes, dubbed Mut DNA. The result was striking: transcription of Mut DNA consistently produced vacuolated droplets, hollow spheres containing stable internal cavities, whereas the unmutated wild-type template produced only homogeneous spherical droplets. Vacuoles appeared in 74 percent of Mut droplets with radii above 15 micrometers and in 94 percent of those above 25 micrometers, while no wild-type droplet ever showed a vacuole. Time-lapse imaging revealed that these cavities nucleate after roughly two hours of transcription, become clearly visible by six hours, and then persist and grow for at least 50 hours, remaining stable for days. This longevity distinguishes them from the short-lived vacuoles previously reported in DNA droplets, which lasted less than an hour.

Control experiments ruled out the obvious alternative explanations. Adding RNase inhibitors had no effect on vacuole formation, and wild-type droplets remained intact and vacuole-free even after 13 days at 37 degrees Celsius, indicating that enzymatic degradation was not responsible. Environmental conditions such as temperature and pH were held constant throughout, so the vacuole phenotype could be attributed to the mutations themselves. Gel electrophoresis of the PCR product revealed two additional bands beyond the main wild-type band: longer fragments, probably generated by self-priming of the template, and shorter fragments from spurious mispriming. When the team gel-purified each band and transcribed them separately or in combination, only preparations containing the longer nonspecific products produced vacuolated droplets on their own, while the shorter products caused vacuolation only when mixed with the main band. The nonspecific PCR products, mostly deletion and duplication mutations, were therefore the genetic cause of the hollow phenotype.

To show that this was not an artifact of one particular mutagenesis route, the researchers turned to substitution mutagenesis, the workhorse of directed-evolution experiments. Using dNTP analogs during PCR, they introduced point mutations into a second nanostar template while sparing the promoter region, controlling the mutation rate through the number of analog cycles. Four distinct phenotypes emerged depending on mutational load: normal droplets, vacuolated droplets, a percolating RNA network, and simple aggregates. Remarkably, a mutation rate of only about three percent, corresponding to roughly seven altered bases per nanostar template, was enough to trigger vacuole formation. An orthogonal error-prone PCR protocol using manganese ions to reduce the fidelity of Taq polymerase produced the same vacuolated phenotype, confirming that substitution mutations are a general route to increased phenotypic complexity in these droplets.

To understand the mechanism, the team sequenced the mutated DNA with long-read nanopore sequencing and found that 94.7 percent of reads still matched the wild type, consistent with the low mutation threshold. The remaining 5.3 percent encoded various nanostar-like structures retaining at least one kissing loop, the most common being a four-armed nanostar with three functional kissing loops. The researchers reverse-engineered this dominant mutant as a defined nanostar, termed Mut RNA, and mapped the thermodynamic phase diagram of wild-type and mutant nanostars by measuring RNA concentrations inside and outside the condensates. Fitting the data with an effective free energy density revealed that phase separation is driven by adhesive interactions among the nanostars, that mutant-mutant interactions are less adhesive than wild-type-wild-type interactions, and that the mutant’s slightly larger molecular volume matches its slightly longer arm.

The phase diagram enabled a quantitative explanation of vacuole formation. As transcription proceeds, the ratio of mutant to wild-type nanostars is fixed by the mutant DNA fraction, but droplet growth outpaces internal diffusion, generating concentration gradients within the droplet. When these gradients push the local composition across the spinodal line, a local instability fires inside the droplet, creating a new internal interface, in other words, a vacuole. A kinetic model built on the measured phase diagram and transcription rates predicted that vacuoles can only form in droplets with radii above roughly 21.5 micrometers, because in smaller droplets diffusion erases the gradients, and only when the mutant fraction exceeds about two percent, because the mutant’s higher internal diffusivity favors crossing the spinodal. Both predictions matched the experiments. Imaging also showed no enrichment of mutant RNA at the water-condensate interface, ruling out a surfactant stabilization mechanism; the vacuoles’ long lifetime instead reflects the high viscosity of the RNA-rich phase, which makes vacuole coalescence vanishingly rare.

The most striking finding is that the vacuoles are not merely structural. Vacuolated droplets exhibit micrometer-scale movement over more than 50 hours, characterized by irregular, burst-like displacements. Tracking experiments showed significantly enhanced motility compared with wild-type droplets, and because moving droplets collide more often, the mutants fused more frequently and grew larger. The mechanism is internal mass redistribution: as vacuoles nucleate and grow, they displace dense RNA, shifting the droplet’s center of mass. Measuring the angle between vacuole displacement and droplet displacement confirmed this, with the most frequent angles falling between 174 and 180 degrees, meaning the vacuole and droplet center move in opposite directions. Unlike previously reported moving DNA droplets, the vacuoles remain stable throughout the trajectory and movement can occur repeatedly, without a persistent polarity or propulsion axis. Beyond motility, the vacuoles act as functional subcompartments: the dense RNA matrix excludes 50-nanometer lipid vesicles, but 50-kilodalton dextran, roughly 10 nanometers across, diffuses through the matrix and accumulates selectively inside the vacuoles, while the encoded biotin aptamer recruits biotinylated vesicles to the droplet surface as a shell.

Together, the results establish RNA droplets as self-contained, evolvable systems that satisfy the prerequisites of Darwinian evolution: retention of genetic material, mutation-driven phenotypic variability, and the emergence of selectable functional traits. The authors suggest that selection could, for example, favor increasingly hollow droplets, evolving toward a vesicle-like RNA condensate, and that future work could integrate ribozymes and aptamers to encode dynamic behavior as a starting point for open-ended evolution. Because sequence programmability can be extended with aptamers for specific molecular recruitment or with the catalytic abilities of ribozymes, the platform points toward evolvable materials with applications in biomedicine and biotechnology, from self-propelled microswimmers to drug delivery systems. It may also inform one of science’s oldest questions, offering a concrete experimental model for how genetic information, compartmentalization and selectable phenotypes could have first become coupled in an RNA world, when nonliving molecular assemblies took their first steps toward life.

Subject of Research: Genetic encoding, mutagenesis and evolvability of RNA nanostar droplets as protocell models

Article Title: Genetic encoding and mutagenesis of RNA droplet phenotypes

Article References: Verstraeten, W., Tran, M. P., Taskina, A., Jaiswal, P., Haugerud, I. S., Helbig, C., Hamberger, M., Green, E. W., Platten, M., Weber, C. A., & Göpfrich, K. (2026). Genetic encoding and mutagenesis of RNA droplet phenotypes. Nature Chemical Engineering. https://doi.org/10.1038/s44286-026-00454-4

Image Credits: AI Generated

DOI: 10.1038/s44286-026-00454-4

Keywords: RNA droplets, liquid-liquid phase separation, synthetic cells, protocells, mutagenesis, vacuoles, kissing-loop interactions, DNA template, phase separation, origins of life, cotranscriptional folding, directed evolution