The centrosome has evolved in multicellular organisms from the basal body/axoneme of the unicellular ancestor (Azimzadeh and Bornens, 2004). It plays a major role in organizing the microtubule cytoskeleton in animal cells. During interphase, the centrosome organizes an astral array of microtubules (MTs) that participate in fundamental cellular functions such as intracellular trafficking, cell motility, cell adhesion and cell polarity. In proliferating cells, the centrosome starts duplicating just before, or at, the onset of S phase and the two newly formed centrosomes participate in the assembly and organization of the mitotic spindle, its orientation with respect to cortical cues, and the late events of cytokinesis.

The animal centrosome consists of a pair of centrioles linked together through their proximal regions by a matrix consisting in part of large coiled-coil proteins of the pericentrin family, which anchor other matrix components. The centrioles contain cylindrical arrays of triplet MTs organized with nine-fold radial symmetry and the proximal region is structurally similar to the basal bodies of cilia and flagella. In animals, centrioles retain the ability to act as basal bodies by templating the assembly at their distal end (the plus ends of the centriole MTs) either of a primary cilium or of beating cilia during ciliogenesis in specialized cells. Recent discoveries have revealed that cilia have crucial roles in an increasing number of cellular and developmental processes, establishing a link between dysfunctional cilia and several genetic diseases (for reviews, see Davis et al., 2006; Bisgrove and Yost, 2006; Dawe et al., 2007).

In post-mitotic cells, the centrosome contains a mature centriole called the mother centriole and an immature centriole assembled during the previous cell cycle, the daughter centriole, which is about 80% the length of the mother centriole (Chretien et al., 1997). Mother centrioles are distinguished by two sets of nine appendages at their distal ends (Paintrand et al., 1992), which are thought to be required for anchoring microtubules at the centriole and for docking of centrioles at the plasma membrane during ciliogenesis.

• Download figure
• Open in new tab
• Download powerpoint

Centrosome functions

The architecture of the microtubule array in differentiated cell types results not only from the dynamic behaviour of MTs but also from a balance between MT nucleation and MT-anchoring activities at the centrosome. During interphase, MTs are nucleated in the matrix associated with both mother and daughter centrioles, but only the mother centriole is able to anchor them on its associated sub-distal appendages (Piel et al., 2000). Microtubules are nucleated by the γ-tubulin ring complex (γ-TuRC). γ-Tubulin is present throughout the cell cycle in the matrix, close to the proximal walls of centrioles. Its levels increase dramatically prior to mitosis, concomitantly with the recruitment of MT-associated proteins required for mitotic spindle formation. This process, centrosome maturation, is under the control of the Polo-like and Aurora A kinases (for a review, see Blagden and Glover, 2003).

Following their nucleation by the γ-TuRC, MTs are either released into the cytoplasm or recaptured and anchored at the centrosome. Several different MT-anchoring mechanisms have been proposed. The subdistal appendages of the mother centriole are thought to be a major site for MT anchoring, and this activity requires ninein, a component of sub-distal appendages (Mogensen et al., 2000). In addition, ninein has been shown to interact with the γ-TuRC and thus also ensures a link with MT nucleation (Delgehyr et al., 2005). Other MT anchoring complexes are seemingly also present in the matrix, although preferentially associate with the mother centriole. Among them, the p150^glued subunit of the dynactin complex seems to play an important role in collaboration with the MT-associated protein EB1 (Askham et al., 2002).

A recently described complex containing the centrosome proteins CAP350 and FOP, and EB1, has also been proposed to play a role in anchoring MTs at the centrosome (Yan et al., 2006). In addition, centrosome anchoring capacity requires pericentrosomal satellites. The satellites are non-membranous granules of 70-100 nm composed of PCM1 protein, which binds to centrosome proteins such as centrin, ninein and pericentrin. The pericentrosomal localization of the satellites is MT- and dynein/dynactin-dependent (Kubo et al., 1999; Dammermann and Merdes, 2002). The BBS4 protein, one of the proteins involved in the Bardet-Biedl syndrome, a heterogeneous disease that results in part from defective ciliogenesis, has been shown to act as an adaptor protein between PCM1 and p150^glued (Kim et al., 2004).

In addition to its function in MT organization, the centrosome could play a crucial role in cytokinesis. An increasing number of centrosomal proteins have been reported to participate in cytokinesis, such as centriolin, Cep55, CP110 and BBS6 (Gromley et al., 2005; Fabbro et al., 2005; Tsang et al., 2006; Kim et al., 2005). Other major transitions where centrosome activity could impinge on the cell division cycle are the G1-S and G2-M transitions (Jackman et al., 2003; Mikule et al., 2007; Uetake et al., 2007) (for a review, see Doxsey et al., 2005).

Centrosome duplication

Distinct phases of the centrosome cycle have been identified. Disorientation, or disengagement, corresponds to the loss of the duplicative orthogonal tight association between mother and daughter centrioles. Disengagement occurs during early G1 phase before completion of cytokinesis (Piel et al., 2001) and requires the activity of separase, a protease that also drives the separation of sister chromatids prior to anaphase (Tsou and Stearns, 2006). Whether separase acts directly or indirectly on the linkers between orthogonal centriole pairs, and the nature of those linkers, remains unknown.

The initiation of procentriole assembly appears to take place before or at the onset of S phase. This idea is supported by the early recruitment of centrin in the immediate vicinity of parental centrioles in human cells. Centrin proteins are ancient proteins that are associated with centriole/basal bodies in most eukaryotic species, sometimes forming a very intricate network – for example, in Chlamydomonas reinhardtii. The requirement for centrin proteins in the centrosome duplication process is not mechanistically understood, because different conclusions have been drawn from studies of different species. It is not absolute either, because the sequence and functions of centrin appear to have greatly diverged in the nematode Caenorhabditis elegans (Azimzadeh and Bornens, 2004).

In yeasts, centrin participates in the `half-bridge' characteristic of the centrosome/SPB and is clearly required for SPB duplication in both budding and fission yeasts. Remarkably, characterization of a centrin interactor called Sfi1p shows that the half-bridge-to-bridge transition that precedes the formation of the new SPB in budding yeast corresponds with the assembly of a new half-bridge that has a mirror image structure with respect to the other half-bridge (Kilmartin, 2003; Li et al., 2006). Thus the first event in the SPB duplication event is the duplication of the half-bridge, which connects the SPB to the nucleus. This early duplication could reflect the need to ensure that the daughter centrosome/SPB maintains or re-establishes an association with the dividing nucleus during cell division (see also Jaspersen et al., 2006). This could be a general feature in most species in which the nucleus–basal-body connection is crucial for cell polarity. Accordingly, a link between the centrosome and the nucleus has been conserved in many divergent organisms and the continuity of this link must be preserved during centrosome reproduction (Bornens and Azimzadeh, 2007).

The molecular mechanisms underlying centriole assembly have been best studied in C. elegans, in which five proteins essential for centriole duplication have been identified. After fertilization of the C. elegans embryo, SPD-2 is first recruited to the parental centriole and allows the recruitment of the kinase ZYG-1, which in turns allows the recruitment of the SAS-6–SAS-5 complex (Pelletier et al., 2006; Delattre et al., 2006). Recruitment of SAS-5 and SAS-6 is required for the formation of the central tube, a structure onto which singlet microtubules are subsequently assembled in an SAS-4-dependent manner (Pelletier et al., 2006). Although the centriolar structure is noticeably divergent in nematodes, this pathway is most likely to be conserved in other eukaryotes, because SAS-6 and SAS-4 have orthologs in other species. In particular, human SAS-6 and Drosophila SAS-4 and SAS-6 orthologs have been shown to be essential for centriole duplication (Leidel et al., 2005; Basto et al., 2006; Rodrigues-Martins et al., 2007). Human SAS-6 localizes to the centrosome and its overexpression triggers centrosome amplification, which suggests a crucial role in centriole assembly. In human cells, centriole duplication has also been shown to require centrobin, a centriole-associated protein that localizes asymmetrically to procentrioles and daughter centrioles (Zou et al., 2005).

The kinase Plk4/SAK is essential for centriole duplication in both human and Drosophila (Habedanck et al., 2005; Bettencourt-Dias et al., 2005) and has thus been proposed to be the functional equivalent of C. elegans ZYG-1. Plk4/SAK could be an assembly-limiting factor because overexpression of Plk4/SAK leads to centrosome amplification in both human and Drosophila (Habedanck et al., 2005; Rodrigues-Martins et al., 2007). Plk4/SAK appears to act directly upstream of the centriole assembly pathway. Indeed centriole amplification induced by SAK overexpression in Drosophila is suppressed when DSAS-4 or DSAS-6 are lacking (Rodrigues-Martins et al., 2007). SAK, DSAS-4 and DSAS-6 are required not only for canonical centriole duplication but also for de novo centriole assembly. A de novo assembly pathway for centriole assembly that is turned off when centrioles are present has been characterized in human cells and in the green algae Chlamydomonas (Khodjakov et al., 2000; Marshall et al., 2001; La Terra et al., 2005; Uetake et al., 2007). The fact that the same regulator (i.e. SAK) and the same downstream effectors are required for both the canonical and de novo pathways suggests that centriole biogenesis is a template-free process. The mother centriole in canonical duplication could thus be seen as a platform used to concentrate the components required for procentriole assembly (Rodrigues-Martins et al., 2007).

Canonical centrioles observed in most eukaryotic species are thought to assemble onto the cartwheel, a structure in the proximal region of the centriole that is the first nine-fold-symmetrical structure to appear during assembly. A mutation in BLD-10, the only component of the cartwheel identified to date, inhibits centriole assembly in Chlamydomonas (Matsuura et al., 2004). Whether BLD-10 has true orthologs in other eukaryotes remains to be elucidated. Whereas the cartwheel persists in mature basal bodies of ciliated protozoa, it is only transient in vertebrate proliferating cells, but the precise timing of cartwheel disassembly during G2-M is not known (Lemullois et al., 1988).

Procentriole elongation starts during late S phase; the centriole reaches full length during the following cell cycle. The mechanisms triggering centriole elongation are poorly understood but appear to require ϵ-tubulin in Chlamydomonas, because the ϵ-tubulin mutant BLD-2 forms short centrioles made of singlet MTs instead of triplets. ϵ-tubulin is conserved in mammals and has been proposed to be required for centriole duplication, although its precise function remains unclear (Dutcher, 2003).

During late G2 phase, centrosome separation allows the formation of a bipolar spindle. Centrosome separation is thought to require the disassembly of a fibrous linker that mediates centrosome cohesion by connecting the two centriole pairs (but not the two centrioles within a pair – see above). C-Nap1 is found at the proximal end of parental centrioles and is proposed to serve as a docking site for this linker. C-Nap1 interacts with rootletin, a conserved component of the ciliary rootlet. The ciliary rootlet is a cytoskeletal structure found in many ciliated cells that originates from the basal body and extends proximally toward the nucleus, providing structural support for the cilium (Yang et al., 2005). Rootletin is, however, also found in cells devoid of a ciliary rootlet, forming fibers that emanate from the proximal ends of centrioles. Centrosome cohesion is regulated during the cell cycle by phosphorylation of C-Nap1 and rootletin, which depends on the balance between NIMA-related kinase (Nek2) and protein phosphatase 1 (PP1) activities (Fry et al., 1998; Helps et al., 2000; Bahe et al., 2005; Yang et al., 2006). C-Nap1 and rootletin do not seem to form a continuous linker between the parental centrioles, and it is thus believed that other proteins are required for centrosome cohesion.

The complete maturation of the procentrioles into mother centrioles extends over one and a half cell cycles: it is completed only after two successive mitoses, culminating with the acquisition of distal and sub-distal appendages.

Centrosome duplication is tightly coupled to the cell cycle. In particular, it has been shown that the activity of the cell cycle kinase CDK2, in complex with cyclin E or cyclin A, is required for the initiation of centrosome duplication (Hinchcliffe et al., 1999; Meraldi et al., 1999). Intriguingly, cyclin E has a centrosome-binding domain essential for promoting S-phase entry in a CDK2-independent manner (Matsumoto and Maller, 2004).

In addition to the above-mentioned regulators, which must be activated in a cell-cycle-dependent manner to trigger centriole duplication, a mechanism that precludes centriole re-duplication has recently been characterized. In this `licensing model', Tsou and Stearns (Tsou and Stearns, 2006) propose that centriole re-duplication is prevented by temporal separation of licencing during anaphase, which would correspond to separase-dependent centriole disengagement, from centriole growth that requires S-phase-specific kinase activities.

How centrosome reproduction and cell division cycle are precisely coupled is still a matter of active research and has not yet led to a comprehensive picture that would fulfill Boveri's vision of the centrosome as the division organ coordinating karyokinesis and cytokinesis.