Laboratory of Protein Dynamics and Signaling
NIH/Frederick National Laboratory for Cancer Research (FNLCR).
1050 Boyles Street, Frederick, MD 21702, USA
A centrosome is one of the most beautiful organelles in the cell, and the most enigmatic one. Both, its enigma and its beauty, are due to its core structure - a centriole, which is a perfectly symmetrical cylindrical assembly of nine microtubule triplets organized around a centrally positioned core (Fig. 1). The centrosome is a nonmembrane-bound organelle present in most animal cells. Its name "a central body" (soma from Latin for "body") echoes its usually central position within a cell. Mammalian differentiated non-dividing cells may contain various centrosome numbers. However, in typical animal cycling cell the number of centrosomes should be strictly restricted to only two centrosomes per cell. Numerical and structural centrosome abnormalities are the feature of human tumors, especially aggressive ones. Although the debate is still ongoing of whether the supernumerary centrosomes can directly lead to malignant transformation, there is increasing body of evidence leading toward this conclusion. This review describes the mechanisms that regulate centrosome number in mammalian cells. I will also try to direct the reader to many excellent publications that explore similar aspects of centrosome biology.
Figure 1. Electron micrograph and schematic presentation of a human centriole. Centrioles are built of nine radially organized microtubule triplets sourrounding a central core. Distal end of mature centriole also carries distal (D) and subdistal (SD) appendages. The centriole organizes pericentriolar material that can be seen as a denser amorphous material encompassing the centriole on electron micrographs.
Centrosome structure and function
A centrosome is comprised of microtubule-based cylindrical structure called a centriole and of surrounding electron-dense pericentriolar material (PCM) (Fig.1). The centrosome was discovered in 19th century but it was not until the advent of electron microscopy in 20th century that its beautiful intricate architecture was revealed (Alvey, 1985; Vorobjev and Chentsov, 1982; Kuriyama and Borisy, 1981).
The walls of the centrioles are symmetrically organized nine microtubule blades which in different organisms may contain microtubule triplets (higher animals), doublets (Drosophila melanogaster), or individual microtubule (C. elegans). The size of the centrioles is variable among species ranging from ~100x150nm in C. elegans to 200x500nm in mammalian cells. Centrioles by still unknown mechanism organize PCM around them and without a centriole PCM is unstable and disappears (Bobinnec et al., 1998), meaning that the number of centrioles determines the number of centrosomes within the cells. Recent proteomic studies performed on isolated centrosomes from several species have revealed that human centrosome is a complex of about hundred proteins, most highly conserved among the species (Andersen et al., 2003; Jakobsen et al., 2011) (for centrosome proteins database see: http://centrosome.dacya.ucm.es).
A centrosome is a dynamic organelle and its composition and function are intimately linked to the major events during cell cycle progression. A PCM component of the centrosomes serves as the main microtubule organizing center (MTOC) during interphase. In that capacity the centrosomes influence various microtubule-based processes such as cell shape, polarity and motility (Luders and Stearns, 2007; Bornens, 2012). In some cell types the centrosomes travel to the cell membrane where they serve as basal bodies for motile and primary cilia formation. Motile cilia are present in specialized differentiated cells in which they assemble in hundreds to produce fluid flow along epithelial surfaces (Stubbs et al., 2012). Primary cilium is typically present in one copy in many cells and serves as a mechano- and chemo-sensor and participates in many signaling and trafficking events during embryogenesis and in adult life (Hoyer-Fender, 2010; Beisson and Wright, 2003; Drummond, 2012). In preparation for mitosis two centrosomes increase their microtubule nucleating capacity to function as the poles of the mitotic spindle (Khodjakov and Rieder, 1999; Haren et al., 2009). During mitosis the PCM of the centrosomes nucleate most of the microtubules for the mitotic spindle. As the organizers of a bipolar mitotic spindle the centrosomes facilitate proper positioning and even segregation of the chromosomes, and ultimately contribute to genomic stability. That is why it is of high importance that a cycling cell contains only two centrosomes. Supernumerary centrosomes may lead to the formation of multipolar mitoses and to aneuploidy (Godinho et al., 2009; Lingle et al., 2005), which is a hallmark of most tumors. Since the number of active MTOCs depends on the number of centrioles, a centrosome number is controlled through the stringent regulation of centriole assembly.
Two modes of centriole assembly
In typical cycling somatic cell the centrioles form by duplication of preexisting centrioles. In this canonical centriole duplication pathway, new centrioles assemble in association with a pre-existing mother centriole. There is another pathway where the centrioles can form in the absence of pre-existing centrioles. Such de novo centriole assembly pathway is suppressed by the presence of at least one centrosome within the cell. Interestingly however, de novo centriole assembly can be activated in vertebrate cells in the culture (such as HeLa or RPE-1) if all centrioles are destroyed by a laser micro beam or removed by microsurgery (Khodjakov et al., 2002; La Terra et al., 2005; Uetake et al., 2007; Loncarek et al., 2007). De novo mode of centriole assembly operates as a regular mode for centriole assembly under the conditions such as parthenogenesis in species where centrioles are normally contributed by sperm (Riparbelli and Callaini, 2003) or in the early embryology of species where both paternal and maternal gametes lose centrioles during gametogenesis (Manandhar et al., 2005). For instance, mouse zygote does not contain centrioles and initially divides without them. Then, during the blastomere stage of development the centrioles assemble in the proper number and continue their propagation by the canonical pathway. Centrioles also assemble de novo in differentiated epithelial cells during ciliogenesis. There, hundreds of centrioles form around amorphous granules composed of various centrosomal proteins (Dawe et al., 2007; Vladar and Stearns, 2007; Loncarek and Khodjakov, 2009).
Centriole duplication cycle
Centriole cycle and the progression through the cell cycle are closely linked. However, the two cycles are just coordinated in time but not dependent on each other; DNA cycle can be completed in the absence of centriole cycle. (Balczon et al., 1995; Loncarek et al., 2008; Loncarek et al., 2010) and conversely, the centriole cycle can continue in the absence of DNA synthesis.
In humans, the first centriole cycle begins upon fertilization when parentally inherited centriole (human oocyte lacks centrioles) duplicates to form the poles of the first mitotic spindle. (Schatten and Sun, 2010). Newly formed G1 cells contain two centrosomes with a single centriole which continue to reproduce via canonical centriole duplication cycle thereafter (Fig. 2). In G1 cell one centriole (a mother centriole) is at least one generation older than the other (a daughter) centriole. Older centriole carries subdistal and distal appendages and associates with slightly different set of PCM proteins (Jakobsen et al., 2011). Both centrioles can nucleate interphase microtubules while only the older one anchors them on its sub distal appendages. In addition, only older centriole can stabilize on the cell membrane via distal appendages to form a primary cilium. The two centrosomes in G1 can often be found apart from each other and may have different dynamics. Older centriole is typically associated with more abundant PCM (Piel et al., 2000). During G1 the formation of new the centrioles does not occur, presumably do to the low Cdk2/cyclinA and/or cdk2/cyclinE levels that have been proposed to be necessary for initiation of centriole duplication.
Figure 2. Centriole duplication and segregation cycle. Typical vertebrate cell is born with only two centrioles which duplicate during interphase and evenly segregate during mitosis. Centriole duplication follows two stringent rules which assure constant centriole through the generations of dividing cells; First: each centriole duplicates only once per cell cycle and second: only one daughter centriole forms per mother centriole.
As the cycling cell progresses from G1 into S phase, each existing centriole (now called a mother centriole) begins to duplicate (Fig. 2). The first morphological manifestation of centriole duplication is the formation of a small procentriole structure near the proximal end of the mother centriole (Vorobjev and Chentsov, 1982; Kuriyama and Borisy, 1981; Loncarek and Khodjakov, 2009; Loncarek et al., 2008) (Fig. 2 and 3). Even very short procentrioles contain characteristic nine-fold symmetry and are orthogonally oriented to the wall of the mother centriole (Fig. 2 and 3). As the cell progresses through S phase the procentrioles elongate (now they are called daughter centrioles), remain orthogonal to the mother centriole, and inhabit the cloud of PCM organized by the mother. This structurally rigid complex between a mother and a daughter centriole is referred to as a "diplosome" and the centrioles are described as "engaged" (Fig. 2 and 3). During S-phase a daughter typically reaches about two-third of the length of the mature centriole (Vorobjev and Chentsov, 1982; Azimzadeh et al., 2009). In G2 phase the daughter centrioles further elongates and the distance between the mother and the daughter centriole slightly increases (Chrétien et al., 1997). Some centriolar proteins are accumulated at this stage at the daughter centrioles (Azimzadeh et al., 2009; Guarguaglini et al., 2005). By the time of mitosis, the daughter centrioles reach the full length and after anaphase onset the mother and the daughter centrioles lose their rigid orthogonal orientation, they disengage (Fig. 3). From this moment each centriole organizes its own cloud of PCM. During next cell cycle the daughter centrioles formed in previous cell cycle will finally acquire all of the biochemical and structural characteristics (for instance the appendages) of fully mature centrioles.
Procentriole assembly is mechanistically still poorly understood multistep process dependent on a relatively small group of core centrosomal proteins with conserved centriole assembly functions in wide range of organisms (Carvalho-Santos et al., 2011; Strnad et al., 2007; Strnad and Gönczy, 2008; Carvalho-Santos et al., 2010; Vulprecht et al., 2012). Protein kinase Plk4/SAK/Zyg-1 with its proposed substrate Cep152/Asl is the main master switch that initiates procentriole assembly (Fig. 3) (Hatch et al., 2010; Habedanck et al., 2005; Bettencourt-Dias et al., 2005; Kleylein-Sohn et al., 2007; Cizmecioglu et al., 2010). Procentriole assembly begins with the formation of a cartwheel, a nine-fold symmetric structure at the basis of the procentriole. Cartwheel protein Sas6 along with its binding partner Cep135/Bld10 are among the first proteins that accumulate to the procentriole. It has been shown that they cooperate in stabilization of nine-fold cartwheel symmetry (van Breugel et al., 2011; Kitagawa et al., 2011). Other structural centrosomal proteins like CPAP/Sas4 and Stil/Sas5/Ana2 also localize to the procentriole before the assembly of the centriolar microtubules. CPAP coprecipitates with gamma tubulin and regulates gamma tubulin-dependent nucleation of the centriolar microtubules (Tang et al., 2009). Elongating centriolar microtubules are stabilized by CP110 and Cep97 which forms a cap-like structure at the distal parts of centriole microtubules. All three proteins CPAP, CP110 and Cep97 are involved in the regulation of centriole length, but how is centriole length regulated is still not clear (Tang et al., 2009; Chen et al., 2002; Schmidt et al., 2009).
Figure 3. Centriole assembly is a multistep process that begins at early S and depends on a small group of centriolar proteins. Procentriole assembly is mediated by Plk4 kinase that together with Cep152 initiates the formation of the cartwheel, a structure that determines the future symmetry of the centriole. Sas6 and Cep135 are the two cartwheel proteins. CPAP promotes gamma tubulin-dependent nucleation of centriolar microtubules around the cartwheel. CP110 and Cep97 form a cap at distal part of the centriole and stabilize it. A cartwheel is lost from the newly formed centrioles after their disengagement from the mother centrioles at the end of mitosis. Several proteins such as Separase and Plk1 have been involved in disengagement. Note that mother centriole carries two sets of appendages. Appendages will be recruited to the daughter centriole during next cell cycle.
Control of centriole formation
Correct centriole number is maintained by the two levels of control during centriole duplication: numerical control assures that only one centriole forms in association with the mother centriole, and temporal control allows only one centriole duplication cycle during the cell cycle.
Numerical control: Several cell cycle kinases have been suggested to affect timely centriole duplication (Loncarek and Khodjakov, 2009; Strnad and Gönczy, 2008; Hinchcliffe and Sluder, 2001) but the activity of protein kinase Plk4 is directly implicated in numerical control of procentriole formation. Depletion of PLK4 inhibits centriole assembly, and Plk4 overexpression induces simultaneous formation of multiple procentrioles around a mother centriole (Bettencourt-Dias et al., 2005; Kleylein-Sohn et al., 2007). Clearly, Plk4 has a dose-dependent effect on procentriole formation. Plk4 levels are self-regulating through Plk4 interaction with SCFSlimb/βTrCP ubiquitin ligase complex and its subsequent degradation by the proteasome (Cunha-Ferreira et al., 2009; Rogers et al., 2009; Korzeniewski et al., 2009; Duensing et al., 2007). Plk4 autophosphorylation, in addition, increases its affinity for SCFSlimb/βTrCP and increases its degradation rate. How Plk4 initiate centriole duplication at the right rate, right place and right time is not yet clear. Similarly to PLK4, overexpression of structural proteins Sas6, CPAP/Sas4 and Sas5/Ana also induce the formation of multiple daughter centrioles (Strnad et al., 2007; Stevens et al., 2010; Kohlmaier et al., 2009). These proteins are normally degraded at the end of mitosis via another ubiquitin ligase complex APCcdh1. Numerical control of centriole duplication can be, thus, interpreted as a control of the litter size through a finely tuned equilibrium of several centrosomal proteins via their concerted protein recruitment, phosphorylation and degradation.
Temporal control: In order to maintain proper centrosome number only one round of centriole duplication should be allowed during a single cell cycle even if a cell cycle is temporarily stalled by a checkpoint. It has long been observed that some tumor cell lines do not obey that rule and if arrested in S phase by hydroxyurea (Balczon et al., 1995) or in G2 after DNA damage (Saladino et al., 2009) they produce supernumerary centrosomes. The question emerged what is awry with the centriole cycle in these cells? In 2003, Wong and Stearns fused the cells from different cell stages to demonstrate that already duplicated centrioles cannot further duplicate but that the single centrioles can duplicate within the same cytoplasm competent for duplication (Wong and Stearns, 2003). The authors hypothesized that the critical factor preventing reduplication of duplicated centrioles should be intrinsic to the centrosome and lies in their engagement. Engagement between a mother and a daughter established in early S after initiation of centriole duplication precludes a new round of duplication until the centrioles disengage during ensuing mitosis (Tsou and Stearns, 2006a; Tsou and Stearns, 2006b). This hypothesis received a direct confirmation in laser ablation experiments conducted in S-phase arrested HeLa cells which demonstrated that a new round of centriole duplication can be initiated in the same cell cycle after physical removal of the daughter centriole within the diplosome by the laser microbeam (Loncarek et al., 2008). Therefore, the engagement appears to be an intrinsic block to centriole reduplication within a single interphase. Disengagement is structurally defined as a loss of orthogonal orientation between a mother and a daughter centriole. However, the molecular mechanisms behind the process of centriole disengagement remain vague. It has been attractively suggested that centriole disengagement occurs due to the proteolytic degradation of a link protein between a mother and a daughter centriole by Separase at the end of mitosis (Tsou and Stearns, 2006b). Separase has been originally described as a proteolytic enzyme responsible for sister chromatide separation at the metaphase-anaphase transition. However, it appeared that the centrioles can still disengage in a separase-null cells, although with much slower rate (Tsou et al., 2009). Regardless, the search for the substrate for Separase at the centrosome continues after the findings that the disengagement can be inhibited by overexpression of shorter isoform of Shugoshin (sSgo1) (Wang et al., 2008). More recently, the role of mitotic kinase Plk1 has been implicated in centriole disengagement in both, mitosis (Tsou et al., 2009) and during prolong interphase (Loncarek et al., 2010). Unscheduled Plk1 activity in human cells arrested in S or G2 strongly promoted accumulation of maturation markers at the daughter centrioles and their disengagement. Once disengaged, the centrioles in these cells underwent new rounds of duplication. The phenotype was robust and consistent irrespective of cellular transformation (Loncarek et al., 2010). How Plk1 promotes centriole maturation and disengagement is not yet clear on molecular level. Plk1 activity is not required for initiation of centriole formation but it rather provides a synchronization mechanism between the cell and centriole cycle during later stages of cell cycle. Plk1 activity, normally low in G1 and S, rises in late G2 and peaks before mitosis. One possible scenario how Plk1 synchronizes the two cycles is that regulated Plk1 level assures that centrioles do not disengage prematurely during the cell cycle. Then, high Plk1 activity during late S/G2 "primes" the centrioles for disengagement in upcoming mitosis what allows the centriole cycle to be completed in synchrony with the cell division.
Direct implication of this hypothesis is that unscheduled Plk1 activity often found in human tumors could directly promote centriole reduplication during prolonged cell cycle arrest. Additional time during arrest could offer the centrioles "primed" by high Plk1 activity a necessary time to disengage and to reduplicate their centrioles. Transient interphase arrest is indeed frequently induced in tumors by chemotherapy or irradiation. It is would be of the highest importance to investigate the behavior of the centrioles during these types of arrest as a function of Plk1 activity.
Most cellular organelles and structures are present in a cell in dozens, hundreds, or thousands of copies, and either losing or gaining one does not critically alter the fate of the cell and its progeny. In contrast, there are only two centrosomes in a cycling cell and a cell must precisely maintain that number, despite of a surplus of building blocks sufficient to, hypothetically, assemble many supplementary ones. On the other extreme, hundreds of centrioles assemble in non-dividing specialized cell types, such as in the cells of ciliated epithelium, upon their differentiation. Yet another aspect of centriole biology that is equally puzzling and strictly regulated is the reduction of centriole number, or their complete absence from the gametes of many species (including humans), and their subsequent re-appearance in early embryogenesis from out of the blue (as is the case in mice). Hence, it is obvious that there are, yet mostly uncovered, regulatory mechanisms which stringently regulate centriole number according to the requirements specific to the cell type, in order to ensure homeostasis and the integrity of the organism.
Centriole and centrosome cycle is attuned with cell cycle in different species and cell types. In humans particularly, the typical centriole duplication cycle coincides with the duplication of DNA, and culminates with their segregation into two nascent daughter cells in parallel with chromosome segregation at the end of mitosis. Within the centrosome, the process of centriole duplication is very stringently regulated by cell cycle regulators, assuring that the newly formed cell receives exactly two centrosomes. It is entirely counterintuitive to imagine how only one perfectly symmetrical cylinder assembles at a right angle in the vicinity of another perfectly symmetrical cylinder in a conservative fashion. At the same time, due to its self-assembly properties, the same structure can efficiently assemble in the absence of a preexisting structure.
Anomalies in centriole biology have a causative role in many human diseases. In ciliated cells the centrioles migrate to the cell cortex where they form cilia. Cilia are involved in many physiological processes and a list of ciliopathies in humans is growing daily (Zaghloul and Brugmann, 2011; Goetz and Anderson, 2010). Regulation of centriole assembly goes awry in almost all types of tumors resulting in centrosomal structural or/and numerical abnormalities. Cells containing supernumerary centrosomes suffer from extensive chromosome mi-segregation resulting in chromosomal instability (Godinho et al., 2009; Ganem et al., 2009). Recent data strongly indicates that experimentally perturbed centrosome numbers directly lead to cellular transformation (Basto et al., 2008). Thus, understanding the mechanisms that regulate centriole biogenesis is more than answering fascinating basic biological questions. It has a significant implication on health and especially in cancer biology. By uncovering the secrets of centriole biology, we can help to better understand the secrets of the human body and take substantial steps forward in understanding and possibly curing, of centriole/centrosome related illnesses.