I was a PhD student when I first read the Heald et al. article on the assembly of bipolar spindles around artificial chromosomes. Although it was known that mitosis in plants and meiosis in most animal species occurred in the absence of centrosomes, Heald et al. demonstrated that DNA-coated beads were sufficient to induce the formation of microtubule arrays that self-organized into bipolar spindles. Going back to this article, I was reminded of the power of simplified systems to address complex biological questions.

The cues that drive spindle bipolarity in somatic cells were thought to originate from the centrosomes and from centromeric DNA sequences that direct the assembly of kinetochores. To test the importance of centromeric DNA in this process, Heald et al. developed a simplified in vitro system in which magnetic beads were coated with plasmid DNA that did not contain centromeric DNA and therefore could not assemble kinetochores. When incubated with interphase Xenopus egg extracts, these beads assembled into chromatin that was able to replicate and promote the assembly of a functional nuclear envelope. In the presence of mitotic extracts and fluorescently labelled tubulin, spindle-like structures assembled from chromatin beads, demonstrating that neither centrosomes nor kinetochores are essential bipolarity cues.

Temporal analyses of this process uncovered three distinct phases: a first phase of nucleation, in which microtubule arrays extend radially; a second phase of coalescence, during which microtubules form compact bundles around chromatin; and a third phase of bipolarity establishment, characterized by the appearance of two focused poles that require the minus-end microtubule motor dynein. Fluorescently labelled microtubule seeds revealed that during nucleation microtubules display random polarity. By contrast, when spindles became bipolar, microtubule polarity was uniform with their minus ends leading towards the poles. The authors proposed that bipolarity arises from cytoplasmic asymmetry introduced by chromatin that favours microtubule nucleation, the intrinsic polarity of microtubules and the activity of motors leading to microtubule sorting into bipolar arrays of uniform polarity.

Heald et al. suggested that chromatin is not a conventional microtubule nucleation site but instead locally changes the cytoplasm in favour of microtubule nucleation. Interestingly, it turns out that this could be a more general characteristic of microtubule organizing centres (MTOCs) than anticipated. Indeed, as recently shown by Woodruff et al., centrosomes can locally increase tubulin concentration, thereby promoting microtubule nucleation. A closer look at how different MTOCs regulate local properties that regulate microtubule nucleation is likely to shed light on this fascinating mechanism.