Reverse Engineering the Evolutionary Logic of Cytoskeletal Dynamics

Living cells perform tasks by generating mechanical force through networks of protein filaments that are actively organized by molecular motors. Eukaryotic genomes encode diverse motor protein variants with distinct biochemical properties, which cells deploy for specific processes such as spindle formation during division or cilia for navigation. Despite extensive characterization of motor–filament systems by in vitro and in vivo approaches, how variations in genomic sequence and structure translate into filament organization dynamics remains poorly understood. Here we develop ActiveDROPS, a cell-free approach to reconstitute microtubule dynamics driven by genetically encoded kinesin motor variants in bacterial lysate droplets, enabling quantitative characterization of motor-induced cytoskeletal behavior in a cell-like context. We screened a library of uncharacterized kinesin-like sequences and identified 12 functional motors that can be grouped into three behavioral classes: “slow-sustained” (prolonged motion > 12 h, velocity 10–50 nm/s), “fast-burst” (brief motion < 2 h, velocity 50– 1000 nm/s), and “multiphase” defined by concentration-dependent transitions including nematic flows, chiral rotation, and global contraction. Through modular recombination of domains from “slow-sustained” and “fast-burst” motors, we engineered chimeras with combined “fast-sustained” dynamics (duration > 12 h, velocity > 100 nm/s). This revealed domain sets controlling microtubule motion lifespan and velocity, with computational analysis of AlphaFold-predicted structures suggesting 0 and L12 as the respective determinants. Incorporation of microtubule-associated proteins such as PRC1 and Tau, together with motor co-expression, further broadened the range of cytoskeletal dynamics. These results provide a framework to reverse-engineer how nature generates mechanical diversity, offering insight into how cells transform genetic variation into emergent behavior.