Speaker
Description
The current model of mitotic spindle assembly proposes that microtubules nucleate at spindle poles and grow inward to capture chromosomes. However, recent structural studies reveal that spindles are composed of short microtubules that do not span the full pole-to-chromosome distance. It remains unclear how short, disconnected microtubules collectively generate and transmit the forces necessary to build a bipolar spindle. Using cryo-electron tomography to map microtubule polarity in intact human cells, we find that spindle microtubules form locally antiparallel dense regions with a consistent 8 nm wall-to-wall spacing. This spacing is too narrow for most molecular motors to fit between adjacent microtubules, ruling out direct motor crosslinking of the bundle interior. Instead, spacing scales inversely with local microtubule density, consistent with density-driven steric interactions, analogous to liquid crystal ordering. Motor perturbations combined with centriole depletion, which generated motor-active monopolar spindles, further revealed that the kinesin-5 Eg5 motor establishes local antiparallel overlap independently of spindle bipolarity, while a balance of Eg5 and dynein regulates microtubule density to maintain spindle architecture. Together, these findings challenge the pole-centric model and suggest a bottom-up, self-organized model in which motor-microtubule interactions within dense bundles generate forces that build bipolar spindles.
