Centrosome cycle and genome stability
Centrosomes duplicate once and only once per cell cycle. In brief, centriole duplication involves the formation of a single procentriole next to each parental centriole and its subsequent elongation. While the morphological changes that occur during centrosome duplication are well documented, our understanding of the molecular pathway responsible for the timely assembly of one and only one procentriole per parental centriole in each cell cycle is still far from being complete. In our group we have decided to tackle this question using reverse genetics in vertebrate cells. Our cell line of choice is the chicken B cell line, DT40, which is amenable to genetic manipulation due to its extremely high homologous recombination rates. Procentriole formation is driven by a small set of proteins: PLK-4, SAS-6, CPAP, CEP63, CEP135, STIL and CEP152. Of these, we disrupted CEP63, CEP135, CEP152 and STIL in chicken B cells to probe their relative contributions to centriole assembly. CEP63 mutant cells grow more slowly than control cells and display a high incidence of monopolar spindles as a result of abortive centrosome duplication cycles. CEP63 binds CEP152, an evolutionarily highly conserved protein. Using super-resolution microscopy we find that CEP63 and CEP152 form a discrete ring-shaped structure at the proximal end of parental centrioles, seemingly occupying a space near the centriole walls, a site implicated in both procentriole formation and centriole cohesion. Our study assigns a specific role for CEP63 in centrosome duplication, which is the centrosomal recruitment and organisation of CEP152, a protein implicated in the initiation of procentriole assembly (Sir et al., Nat Genet 2011; 43: 1147).
We next disrupted CEP135 in DT40 cells, but unexpectedly, CEP135 null cells display a relatively minor defect in centriole numbers. Electron microscopy studies however reveal aberrant structures in centrioles lacking CEP135. Therefore, CEP135 does not seem essential to assemble a centriole, but it could be important for its subsequent stability (Inanç, Puetz et al., Mol Biol Cell. 2013; Epub 17 July).
Finally, we generated cells that lack CEP152 or STIL. Characterisation of these cell lines revealed that they are devoid of intact centrioles and thus lack functional centrosomes. Both STIL- and CEP152-knockout cells grow slowly and exhibit a delay in mitosis. The most striking observation, however, is the increased frequency of chromosome missegregation in these cells and the resulting aneuploidy (Figure 1). Our results indicate that centrosomes have two major functions in proliferating cells: 1) they are important for normal mitotic timing and 2) they significantly increase mitotic fidelity, and thus minimise the occurrence of aneuploidy (Sir et al., J Cell Biol. 2013; 203: 747).
Figure 1: Growth defects and abnormal mitotic spindles in cells that lack functional centrosomes. (A) Graph shows reduced growth of CEP152-knockout cells. (B) Microscopy images show abnormal mitotic spindle morphology in a CEP152-knockout cell. Note that the mitotic spindle is symmetric and bipolar in the control cell, but is disorganised in the knockout cell. Also note the absence of centrosomal marker in the CEP152-KO cell. Centrosomal marker is in red, microtubules are in green and DNA is in blue in merged image.
Centrosome duplication is tightly controlled in normal cells to ensure that cells enter mitosis with two functional centrosomes. By contrast, centrosome amplification has been observed in many cancer types. Extra centrosomes cause multipolar spindle formation, which can potentially lead to catastrophic chromosome missegregation and cell death. Cancer cells with centrosome amplification circumvent this fate by assembling pseudo-bipolar spindles with the help of centrosome clustering. In collaboration with other groups, we have developed a small molecule inhibitor of HSET, a microtubule motor protein implicated in centrosome clustering (Watts et al., Chem Biol. 2013; 20: 1399). This inhibitor can indeed kill cells with supernumerary centrosomes, paving the way to selective targeting of tumour cells with aberrant centrosomes.
The centrosome duplication cycle is carefully coordinated with DNA replication, with many regulatory molecules shared between the two processes. However, to maintain normal centrosome numbers in cells, correct centrosome duplication must always be followed by equal segregation of the two centrosomes into the daughter cells. Centrosome segregation, just like chromosome segregation, is entirely dependent on a functional mitotic spindle. Centrosomes occupy a position at the spindle pole during mitosis, and the centrosomes are transmitted to the daughter cells by means of their attachment to spindle poles. Therefore a close association must be maintained between these structures throughout mitosis for normal centrosome segregation to occur. To understand how this is achieved, we need to elucidate the molecular complexes that build the centrosome-spindle pole interface. We have previously identified a role for the conserved protein CDK5RAP2 in contributing to this interface and hypothesised that CDK5RAP2 might act in a large protein complex. Indeed, to date several interacting proteins, including microtubule stabilising and cross-linking factors, have been discovered in our laboratory (Figure 2). Using genetic and biochemical models, we are currently investigating how the concerted efforts of these molecules ensure that centrosomes remain attached to spindle poles in the presence of complex spindle forces.
Figure 2: Chromosome missegregation is frequent in cells lacking centrosomes. (A) Control, STIL- and CEP152-knockout cells are shown during anaphase. Arrows mark lagging chromatids in STIL- and CEP152-knockout cells that lack centrosomes. Frequency of chromosome missegregation obtained from live cell analysis is indicated below for each genotype. Frequency of chromosome missegregation obtained from live cell analysis is indicated below for each genotype.