Disjunction of conjoined twins: Cdk1, Cdh1 and separation of centrosomes
© Crasta and Surana; licensee BioMed Central Ltd. 2006
Received: 12 June 2006
Accepted: 22 June 2006
Published: 22 June 2006
Skip to main content
© Crasta and Surana; licensee BioMed Central Ltd. 2006
Received: 12 June 2006
Accepted: 22 June 2006
Published: 22 June 2006
Accurate transmission of chromosomes from parent to progeny cell requires assembly of a bipolar spindle. Centrosomes (spindle pole body in yeast) are critical for the biogenesis of this complex mitotic apparatus since they confer bipolarity on the spindle and serve as the site of microtubule polymerization. In each division cycle, the centrosome is duplicated and the sister-centrosomes move away from each other, forming the two poles of the spindle. While the structure and the duplication of centrosomes have been investigated extensively, the understanding of the control of their segregation remains scant. Recent findings are beginning to yield insights into the regulation of centrosome segregation in yeast and its link to the mitotic kinase.
Segregation of chromosomes to the two opposite poles of a dividing cell is central to cellular reproduction. Fittingly, the mitotic spindle, which is primarily responsible for progressive separation of sister-chromatids during anaphase, is a tension-ridden, bipolar structure with centrosomes constituting the two poles. Centrosomes are also the microtubule organizing centers (MTOC) of a spindle . In a typical metaphase spindle, the two centrosomes are in a 'face-to-face' configuration, separated by a set of overlapping microtubules, emanating from each centrosome towards the other (pole-to-pole microtubules) . A second set of microtubules, called astral microtubules, radiate from each centrosome towards the cell cortex. A third set, termed kinetochore microtubules, emanate from the centrosomes and eventually attach to duplicated chromosomes at the kinetochores such that each sister kinetochore is linked to one pole that is directly facing it while the other is attached to the opposite pole. Due to the dynamic nature of the microtubules and the activities of the proteins associated with them (microtubule associated proteins or MAPS), the spindle poles (centrosomes) tend to move away from each other, causing the sister chromatids to be pulled in the opposite direction. In a metaphase spindle, this tendency is opposed by proteins known as cohesins that tether the sister chromatids together. Such opposing forces within the spindle are what make it a tension-ridden structure.
The centrosome is pivotal to the biogenesis of the mitotic spindle. In many animals, assembly of the first spindle in the fertilized egg is dependent on the MTOC (in the form of a basal body centriole) contributed by the sperm since oocyte centrosome degenerates sometime during oogenesis. Hence, during fertilization, the sperm contributes not only DNA but also MTOC for construction of a spindle . The incoming centriole then recruits maternal components that constitute the pericentriolar material (PCM) . Similarly, during division a cell inherits only one centrosome from its mother and has to build a complex spindle structure starting from this centrosome. In each cell cycle, centrosomes are duplicated and separated precisely to serve as two poles of the mitotic spindle, both acting as the organizing centers for nuclear and astral microtubules. It is quite astounding that a multi-protein structure like the centrosome is exactly duplicated without using a preexisting counterpart as a template, at least not in the sense that a template is used for copying DNA. We first take a brief look at this unique duplication process.
Some of the paradigms of centrosome duplication have been based on the exquisite work done on the spindle pole bodies (SPB), the centrosome-equivalent in yeast . At the structural level, SPBs do not bear much resemblance to centrosomes. In budding yeast Saccharomyces cerevisiae SPB is a cylindrical structure embedded in the nuclear envelope and comprises three distinct layers (Figure 1). The outer plaque faces the cytoplasm, contains proteins such as Tub4, Spc98, Spc97 and Spc72 and nucleates astral microtubules . The inner plaque faces the nucleoplasm and extends the nuclear microtubules that consist of both pole-pole microtubules and the kinetochore microtubules. The central plaque is embedded in the nuclear envelope and anchors both outer and inner plaques. Proteins such as Spc29, Spc42 and Cmd1 have been localized to this layer. The central plaque also contains an electron-dense structure known as the 'half-bridge'. The half-bridge is an important appendage because its distal end mediates the assembly of a new SPB. A number of proteins including Cdc31 (centrin homolog), Kar1 and Sfi1 (similar to human Sfi1) and Mps3 have been localized to this structure. The crystal structure of Sfi1-centrin filaments suggests that repeats in Sfi1 form a continuous filament to which centrin binds. Hence, multiple interactions between Sfi1 and centrins is proposed to gives the half-bridge its characteristic shape .
Centrosome/SPB duplication and its integration with other cell cycle events involve a host of structural and regulatory proteins. We will not discuss these components individually or the details of the functional interactions among them and how they contribute to the various stages of duplication process since a number of excellent reviews on this subject are already available in the existing literature [11, 14–17]. Instead, we will focus our attention on an event that occurs after duplication i.e. centrosome separation, without which a spindle cannot be assembled in its characteristic bipolar form even though cells can duplicate centrosomes successfully.
While centrosome duplication has been fairly well documented, the separation of duplicated centrosomes is poorly understood. In the case of vertebrate cells, the mother and daughter centrioles (and the duplicated centrosomes) appear to be tethered since centrioles remain paired in centrosome preparations. Electron microscopy studies show an electron dense material between the mother and daughter centrioles in isolated centrosomes. However, the molecular nature of this linker (intercentriolar linkage) remains unclear. (It should be noted that the linkage between the mother and daughter centrioles is what later, after duplication, becomes the linkage between the two pairs of centrioles i.e. centrosomes). While in vivo existence of a linker is not yet unequivocally proven, some of the proteins that regulate centrosome separation (or centriole disjunction) have been identified. Nek2A, a NIMA related kinase , forms a complex with the catalytic subunit of phosphatase PP1 and C-Nap1, a large coiled-coil protein (280 kDa) thought to provide a docking site for the linker. Nek2A is able to phosphorylate itself and C-Nap1, whereas PP1 can dephosphorylate both Nek2A and C-Nap1. It has been reported that inhibition of C-Nap1 activity or over-expression of Nek2A leads to premature separation of mother and daughter centrioles. Moreover, while co-expression of PP1 and Nek2A can prevent centriole separation, inhibition of PP1 activity promotes the separation. These observations suggest that mutual regulation of the C-Nap1, PP1 and Nek2A is important for the control of centrosome cohesion. Recently, a C-Nap1 interacting protein, called Rootletin, has been shown to participate in centriolar cohesion . It is also phosphorylated by Nek2 and is removed from centrosomes at the onset of mitosis. Like in the case of C-Nap1, inhibition of Rootletin results in centrosome splitting.
Motor proteins are known to play an essential role in bipolar spindle assembly . Eg5, a plus-end directed homotetrameric motor protein belonging to BimC (Kinesin-5) family is of particular importance in this context [21, 22]. It can both push apart pole-to-pole microtubules as well as recruit microtubules into bundles . Inhibition of Eg5 by a small molecule inhibitor, monastrol, prevents centrosome separation and leads to monopolar spindles suggesting that its functions are essential for biogenesis of a bipolar spindle . The association of Eg5 with spindle apparatus is regulated via phosphorylation by Cdk1 in human cells , suggesting a direct involvement of Cdk1 in centrosome separation. In addition to Cdk1, polo and aurora-A kinases are also implicated in centrosome separation. In both Xenopus and Drosophila, loss of aurora-A kinase activity results in a failure to separate centrosomes [26, 27]. In both Drosophila and human cells, centrosomes fail to separate in the absence of polo kinase [28, 29]. At least in vertebrate cells, it is possible that polo and aurora-A kinases may contribute to complete removal of C-Nap1 from centrosomes .
By the time budding yeast cells enter S phase, the SPBs are already duplicated and are linked by a complete bridge. At the end of S phase, the bridge is severed and the SPBs move away from each other, each carrying a half bridge, to constitute two poles of a short spindle. A number of cellular conditions can result in failure to separate SPBs and assembly of a bipolar spindle: 1) simultaneous deficiency of plus-end directed kinesin motors Cin8 (homolog of vertebrate Eg5) and Kip1 [31, 32], 2) deficiency of B-type cyclins Clb1, Clb2, Clb3 and Clb4 , 3) inability to dephosphorylate the conserved tyrosine 19 on Cdk1 (Cdc28) , and 4) over-expression of tyrosine kinase Swe1 (budding yeast homolog of Wee1) . These observations suggest that, like centrosomes, SPB separation also require microtubule-associated proteins (MAPs) such as Cin8 and Kip1, and Cdk1 activity. Although no additional proteins seem to have been clearly implicated in SPB separation, it should be noted that the inability to separate SPBs due to combined deficiency of Cin8 and Kip1 can be partially suppressed by deletion of a minus end-directed motor Kar3 . This has prompted the idea that SPB separation requires interplay of opposing forces and that additional regulators involved. Such antagonistic forces also profoundly influence the overall spindle dynamics that presumably influences the elongation and spindle length during mitosis. While the requirement for kinesin motor proteins Cin8 and Kip1 could be tentatively explained in terms of the need to move SPBs apart, the role of Cdk1 (and specifically its activation by Tyr 19 dephosphorylation) had not been elucidated.
That motor proteins are required for separating SPBs seems intuitively obvious since the act of separation needs physical movement. Cin8, Kip1 and Ase1 can each induce SPB separation in cdc28 mutant cells . However, Ase1 does not exhibit any motor activity. What property, then, do these proteins share that allows SPB separation? The ability to recruit microtubules into bundles (bundling activity) is common to these three proteins. In the case of Cin8, bundling and motor activity can be separated by mutation: of certain amino acid residues: while R196K substitution eliminates the motor activity, R394A, H396A and E871A collectively abolish the bundling activity . Interestingly, ectopic expression of mutant cin8, defective in motor activity but proficient in bundling activity, is able to induce SPB separation and spindle formation in cdc28 mutants described above but bundling-defective mutant cin8 is unable to do so . Hence, while motor activity of Cin8 is dispensable for SPB separation, bundling activity is not (Figure 4). One implication of this conclusion is that the act of bundling, which involves conversion of interdigitating microtubules emanating from each SPB into an anti-parallel array of microtubules, generates enough shear force to break the bridge connecting the two SPBs.
It can be argued then, that the physical movement required for converting side-by-side SPBs into a short spindle is caused by the shearing force generated by bundling but not motor activity. Later in the cell cycle, when the short spindle extends into a long anaphase B spindle, motor activity plays a predominant role. Thus, two distinct activities of motor proteins like Cin8 can be considered important for two different stages of spindle biogenesis. It remains to be seen if bundling and motor activities are regulated differently during the cell cycle.
Despite little structural resemblance between centrosomes and SPBs, some broad parallels in the process of their duplication and separation can be seen. Both are multi-protein assemblies that cells must duplicate every division cycle, by using the preexisting mother structure, not as a template for copying but as a surface for seeding the assembly of a new one; in the case of centrosomes it is the proximal end of the mother centriole, whereas for the SPBs it is the distal end of the half-bridge that acts as a platform for seeding. As the SPB half-bridge lengthens and partially retracts during the duplication cycle, the fibrous material that connects the proximal ends of mother and daughter centrioles also shows lengthening and shortening during duplication. Human homologs of some of the proteins involved in SPB biogenesis, such as Cdc31 (centrin HsCen1, HsCen2, HsCen3), Sfi1 (HSfi1), Cmd1 (calmodulin Calm1, Calm2, Calm3) and Spc110 (kendrin) have also been found. Moreover, SPB biogenesis in yeast is drastically affected by over-expression of human centrin 3. Such parallels suggest that there may be a common building strategy that the construction of both centrosome and SPB follows .
Duplication of centrosomes and SPBs is followed by their separation; in both cases, it involves plus-end-directed kinesin motors and Cdk1 activity. As mentioned earlier, for centrosome separation (disjunction), Nek2A mediated displacement of C-Nap1 and the activity of Eg5 appear to be necessary. The involvement of Cdk1 is through phosphorylation of Eg5 which allows its binding to centrosomes. Unlike SPB separation in yeast where Cdk1's role is to stabilize Cin8, Cdk1-mediated stabilization of Eg5 has not been reported. However, Eg5 contains a KEN-box  which makes it a potential substrate for APCCdh1. Moreover, over-expression of Cdh1 prevents centrosome separation in human cells  suggesting a possible involvement of Cdh1 and Cdk1-mediated stabilization of Eg5 in centrosome separation. As for the requirement of motor versus bundling activity, recent findings clearly suggest that motor activity of Cin8 is dispensable for SPB separation while microtubule binding and bundling activities are essential . It has been reported that in Xenopus egg extracts, mutant version of Eg5 with ~6 fold reduced motor activity can fairly efficiently assemble bipolar spindles though the assembly process is slower suggesting that the motor activity of Eg5 may be dispensable. However, the importance of Eg5's bundling activity in centrosome separation is not known .
A number of key issues connected with centrosome duplication/separation such as the control of centrosome numbers, relationship between centrosome biogenesis and genomic instability, the nature of the seeding process that initiates duplication, details of the motor protein-mediated separation, identification of additional regulators and their roles to name a few, need to be resolved. Nonetheless, centrosomes/SPBs and the spindle they help to assemble, remain among the most beautifully symmetric cellular structures and still pose a challenge to be understood in detail.
We thank Hong Hwa Lim for her comments on the manuscript. This work was supported by Biomedical Research Council, Singapore.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.