Spindle assembly checkpoint-dependent mitotic delay is required for cell division in absence of centrosomes.

The spindle assembly checkpoint (SAC) temporally regulates mitosis by preventing progression from metaphase to anaphase until all chromosomes are correctly attached to the mitotic spindle. Centrosomes refine the spatial organization of the mitotic spindle at the spindle poles. However, centrosome loss leads to elongated mitosis, suggesting that centrosomes also inform the temporal organization of mitosis in mammalian cells. Here, we find that the mitotic delay in acentrosomal cells is enforced by the SAC in a MPS1-dependent manner, and that a SAC-dependent mitotic delay is required for bipolar cell division to occur in acentrosomal cells. Although acentrosomal cells become polyploid, polyploidy is not sufficient to cause dependency on a SAC-mediated delay to complete cell division. Rather, the division failure in absence of MPS1 activity results from mitotic exit occurring before acentrosomal spindles can become bipolar. Furthermore, prevention of centrosome separation suffices to make cell division reliant on a SAC-dependent mitotic delay. Thus, centrosomes and their definition of two spindle poles early in mitosis provide a 'timely two-ness' that allows cell division to occur in absence of a SAC-dependent mitotic delay.

Growth disadvantage associated with centrosome amplification drives population-level centriole number homeostasis.

The centriole duplication cycle normally ensures that centriole number is maintained at two centrioles per G1 cell. However, some circumstances can result in an aberrant increase in centriole number-a phenotype that is particularly prevalent in several types of cancer. Following an artificial increase in centriole number without tetraploidization due to transient overexpression of the kinase PLK4, human cells return to a normal centriole number during the proliferation of the population. We examine the mechanisms responsible for this return to normal centriole number at the population level in human retinal pigment epithelial cells. We find that the return to normal centriole number in the population of induced cells cannot be explained by limited duplication of centrioles, instability of extra centrioles, or by grossly asymmetric segregation of extra centrioles in mitosis. However, cells with extra centrioles display heterogenous phenotypes including extended cell cycle arrest, longer interphase durations, and death, which overall results in a proliferative disadvantage relative to normal cells in the population. Although about half of cells with extra centrioles in a population were able to divide, the extent of the disadvantages conferred by other fates is sufficient to account for the observed rate of return to normal centriole number. These results suggest that only under conditions of positive selection for cells with extra centrioles, continuous generation of such centrioles, or alleviation of the disadvantageous growth phenotypes would they be maintained in a population.