Origin of kinetochore microtubules in Chinese hamster ovary cells

We have attempted to determine whether chromosomal microtubules arise by kinetochore nucleation or by attachment of pre-existing microtubules. The appearance of new microtubules was investigated in vivo on kinetochores to which microtubules had not previously been attached. The mitotic apparatus of Chinese hamsters ovary cells was reconstructed in three dimensions from 0.25 micrometer thick serial sections, and the location of chromosomes, kinetochore outer disks, centrioles, virus-like particles and microtubules determined. Central to the interpretation of these data is a synchronization scheme in which cells entered Colcemid arrest without forming mitotic microtubules. Cells were synchronized by the excess thymidine method and exposed to 0.3 microgram/ml Colcemid for 8 h. Electron microscopic examination showed that this Colcemid concentration eliminated all microtubules. Mitotic cells were collected by shaking off, and cell counts showed that over 95% of the cells were in interphase when treatment began and thus were arrested without the kinetochores having been previously attached to microtubules. Cells were then incubated in fresh medium and fixed for high voltage electron microscopy at intervals during recovery. -- In early stages of recovery, short microtubules were observed near and in contact with kinetochores and surrounding centrioles. Microtubules were associated with kinetochores facing away from centrosomes and far from any centrosomal microtubules, and thus were not of centrosomal origin. At a later stage of recovery, long parallel bundles of microtubules, terminating in the kinetochore outer disk, extended from kinetochores both toward and away from centrosomes. Because microtubules had never been attached to kinetochores, the possibility that kinetochore microtubules were initiated by microtubule stubs resistant to Colcemid was eliminated. Therefore we conclude that mammalian kinetochores can initiate microtubules in vivo, thus serving as microtubule organizing centers for the mitotic spindle, and that formation of kinetochore-microtubule bundles is not dependent on centrosomal activity.