Centriole inheritance

Abstract

Early cell biologists perceived centrosomes to be permanent cellular structures. Centrosomes were observed to reproduce once each cycle and to orchestrate assembly of a transient mitotic apparatus that segregated chromosomes and a centrosome to each daughter at the completion of cell division. The molecular picture of centrosomes is now more complete. Centrosomes are composed of a pair of centrioles buried in a complex pericentriolar matrix. The bulk of microtubules in cells lie with one end buried in the pericentriolar matrix and the other extending outward into the cytoplasm. Centrioles recruit and organize pericentriolar material. As a result, centrioles dominate microtubule organization and spindle assembly in cells born with centrosomes. Centriole inheritance is ensured by cell cycle checkpoints that monitor the presence of centrosomes in G1 and spindle bipolarity in mitosis. In spite of these cell cycle controls, assembly of bipolar spindles in acentriolar cell extracts and cell proliferation in acentriolar mutants argue that neither centrioles nor centrosomes are essential structures. This review examines the recent advances in understanding centriole inheritance.

Introduction

Theodor Boveri (1862–1915) was among the first biologists to recognize the importance of centrosomes to inheritance of genetic information.1,2 Boveri and his contemporaries accepted that cells divide by binary fission and that chromosomes provide the genetic basis of inheritance. How the thread like chromosomes in the nucleus of the parental cell are transmitted to the two daughter cells was a mystery. Through careful observations with the crudest of microscopes by today's standards, but the keenest of insights, Boveri defined the centrosome (Fig. 1) as a pair of centrioles surrounded by a special material (pericentriolar matrix) that organized the surrounding cytoplasm and formed a “division center” (microtubule aster). Boveri argued that the centrosome was the “reproductive organ” of the cell, orchestrating assembly of a transient mitotic apparatus (spindle) that segregated bipartite chromosomes (sister chromatids) to each daughter cell. Central to his argument, centrosomes are permanent cellular structures that have the capacity to reproduce and divide once per cell cycle. He used simple tricks to generate cells with one or more centrosomes, finding a half-spindle organized by a single centrosome and multipolar spindles organized by a centrosome at each pole. He argued that the number of centrosomes determined the number of spindles poles. The Centrosome Theory formulated by Boveri met resistance; not all cells have centrosomes, but all cells do assemble spindles.

A landmark study provided proof of principle that centrosomes are not essential for spindle assembly.3 Xenopus eggs lack centrioles, but contain materials needed for many rounds of centrosome assembly. When chromatin coated beads are substituted for chromosomes, microtubules emerge from the beads, nucleated by members of a family of plus-end binding proteins that add tubulin at microtubule plus-ends, effectively pushing microtubule minus-ends outward. Microtubules are eventually bundled into spindle poles by microtubule dependent motor proteins. Spindle assembly in acentriolar Xenopus egg extracts suggests that cells have a default pathway to spindle assembly. The significance of these results are supported by studies showing that the components and molecular mechanisms that promote self-assembly of spindles in Xenopus extracts operate in vivo as well.4,5 The existence of a default pathway to spindle assembly begs the question of why centrosomes are conserved structures in virtually all animal cells.

Centrioles accumulate pericentriolar material and as a result, dominate microtubule organization in cells. Centrioles have a highly conserved nine-fold axis of symmetry (Fig. 1), most often formed by triplets of microtubules in a pinwheel. Centrioles are virtually identical to the core of basal bodies that organize the axonemes of cilia and flagella. Defects in basal body function have profound effects on cell signaling during development and human health,6 but space limitations preclude further discussion here on this important topic. This review examines the recent literature on inheritance of centrioles in light of recent debate on the need, or lack thereof, for centrioles in spindle assembly and completion of mitosis.

The Perceived Function of Centrioles in Mitosis

Microtubule organizing centers.

Expectations of centrosome activity and the corresponding role historically ascribed to centrioles are rooted in microtubule dynamics and the perceived role of centrosomes in microtubule nucleation. Microtubules are closed tubes of ∼13 protofilaments of tubulin monomers, α/β-tubulin heterodimers, in head-to-tail polymers held together by noncovalent interactions.7 As a result microtubules constantly turn over in cells, changing both in length and composition. Tubulins are GTP binding proteins and polymerization into growing microtubules requires that both α- and β-tubulins are bound to GTP. Crystallography of α/β-tubulin bound to GTP shows a straight conformation that is thought to promote the longitudinal interactions necessary for polymerization.8 GTP bound to β-tubulin subsequently undergoes hydrolysis, resulting in a more unstable lattice.9 Tubulin self-assembles into microtubules in vitro; the rate at which new microtubules form, designated as nucleation, depends heavily on the concentration of tubulin, GTP and microtubule associated proteins (MAPs) that lower the concentration of tubulin needed for self-assembly of comparable amounts of polymer. Until recently the concentration of tubulin in cells was considered too low to support the level of tubulin polymer found in cells by self-assembly, requiring a mechanism to renew the microtubules lost to depolymerization. Because the bulk of microtubules are organized by centrosomes, microtubule nucleation was inferred to be the essential activity of the pericentriolar matrix.

The pericentriolar matrix.

γ-tubulin was among the first pericentriolar proteins identified.10 γ-tubulins show more structural diversity than α- and β-tubulins, but γ-tubulin is also ubiquitously expressed among eukaryotes and essential for viability. A portion of γ-tubulin in extracts of flies and vertebrates sediments in high molecular weight ring-like complexes, designated γTuRCs, with 5–6 other proteins (Table 1). The ring-like shape suggested that γTuRCs might provide a template for nucleation of microtubules.10 However, the constituents are not conserved across species (Table 1) and smaller complexes (γTuSCs) are also found in Drosophilae that do not have a ring-like structure. Crystals of γ-tubulin bound to a non-hydrolysable GTP analog showed a curved conformation that is most similar to α/β-tubulins incorporated into the walls of microtubules and in soluble monomers bound to GDP that cannot polymerize.8 Localization to microtubule minus ends has received considerable attention,11 but γ-tubulin is periodically positioned along the length of microtubules in fission yeast12 and in land plants where it is needed for assembly of branched cortical microtubule arrays.13,14

Initial studies suggested that γ-tubulin is universally essential for microtubule nucleation, but it is now widely acknowledged that alternative pathways exist, in large part because abundant microtubules persist in γ-tubulin deficient mutants and in mutants deficient in other γTuRC components.16–21 Additional proteins implicated in nucleation of microtubules include orthologs of XMAP215 in Xenopus, a conserved MAP that promotes dramatic growth at microtubule plus-ends.22–24 Loss of the ortholog ZYG-9 in Caenorhabditis reduces the number of centrosomal microtubules to a level comparable to loss of α-tubulin and more severely than loss of γ-tubulin.25 Reconstitution of microtubule asters with salt-stripped centrosomes requires XMAP215 in vertebrates.26 XMAP215 bound to beads nucleated microtubule asters in dilute solution of pure tubulin as well as in cell extracts,26 as expected of an authentic microtubule nucleator. In addition to γ-tubulin, glp-1 and ZYG-9, targeted RNAi screens in Caenorhabditis for genes involved in microtubule nucleation and growth identified proteins known to interact with ZYG-9, proteins previously identified in other pathways and novel proteins that have not yet been fully analyzed.25 Similar studies in vertebrate cells and flies have implicated additional proteins in centrosome function. An outstanding question in all cases is whether the perceived effect on microtubule nucleation is a direct or indirect effect on centrosome organization.

Centriole Assembly and Acquisition of Pericentriolar Material

Centriole function and assembly.

Centrioles organize pericentriolar components into a discrete focus. The mechanism is a mystery, but the effect is evident in the mouse blastocyst where centrioles do not appear until the second or third mitosis,27 coinciding with reorganization of pericentriolar components into the characteristic dense spherical mass of canonical centrosomes. At least some pericentriolar proteins are incorporated into centriole barrels; γ-tubulin is essential for centriole assembly as well as assembly of the central pair of microtubules that emerge from the transition zone in basal bodies (Fig. 3).28,29 Accumulation of γ-tubulin at centrioles in Caenorhabditis is mediated by another pericentriolar protein, the spindle defective protein SPD-5.30 RNAi inactivation of either SPD-5 or γ-tubulin reduced centriole assembly to ∼40–50% of the number expected.31 SPD-5 may serve only to recruit γ-tubulin since incorporation into centriole barrel has not been reported. An essential role for γ-tubulin in centriole assembly is significant because loss of γ-tubulin function disrupts centrosome organization across species. Severe γ-tubulin mutants generate malformed spindles, including monopolar structures that follow inward collapse of spindle poles in Caenorhabditis18,19 and fusion of nearby spindles in γ-tubulin deficient Drosophila embryos (Fig. 2).17,32 Thus, γ-tubulin function in microtubule organization and spindle bipolarity reflects, at least in part, its role in centriole assembly and the centriole's role in recruiting and organizing pericentriolar components.

γ-Tubulin was the first of an extended family of tubulin isoforms that proved to be associated with centriole and basal bodies. Almost a decade following discovery of γ-tubulin, genetic and/or genomic approaches identified δ-, ε-, η- and ζ-tubulins.29 In contrast to γ-tubulin, the δ-, ε-, η- and ζ-tubulins have limited distribution among species (Table 2). Although little is known about η-, and ζ-tubulin function, basal bodies in Chlamydomonas and Paramecium mutants deficient in δ- and ε-tubulins are shortened and missing the B- and/or C-tubules and ε-tubulin has been implicated in centriole assembly.29 Among other models for function,34 δ- and ε-tubulins may promote stability of centrioles/basal bodies. Stabilizing function is consistent with restored triplet microtubules in δ-tubulin deficient Chlamydomonas mutants that also express an altered form of α-tubulin. This mutant form of α-tubulin stabilizes microtubules and δ- or ε-tubulins may stabilize triplet microtubules in the same way.29 However, the precise role of any member of the extended tubulin family in centriole/basal body assembly is not yet known.

The centriole cycle and the spindle assembly checkpoint.

Centriole assembly is tightly controlled by the cell cycle, limiting the number of centrosomes and thereby promoting assembly of bipolar spindles. Each cell inherits a centriole pair at the completion of mitosis and cytokinesis of its parent (Fig. 4). Centrioles initially show an orthogonal orientation that is lost before assembly of a new centriole is initiated. A nascent centriole or procentriole appears during S-phase near the proximal end of the preexisting or parental centriole.35 Centriole assembly initiates with formation of central tube-like structure,36 followed by the appearance 9 singlet A microtubules in Caenorahditis36 and sequential addition B and C tubules in other species. Procentrioles continue to elongate during G2, but typically do not acquire a significant mass of pericentriolar material or associated microtubules until the next cell cycle. Centrioles remain tethered in most cells until the G2/M transition into mitosis when the connection is severed, allowing centrosomes to separate and form a bipolar spindle.35 If centriole duplication fails, the barren centrioles will separate and organize the poles of a bipolar spindle. A single centriole will be partitioned to each daughter by cytokinesis and these will organize a monopolar spindle in the following cell cycle (Fig. 4). In the absence of bipolar attachment of kinetochores, the spindle assembly checkpoint will arrest the cell cycle. Thus, the spindle assembly checkpoint is perceived as a backup mechanism to block proliferation of cells with defects in centriole duplication.

Genetic analysis of centriole assembly and duplication.

Initiation of centriole assembly requires a member of a conserved family of serine/threonine kinases, Plk-4 (polo kinase 4) in vertebrates and Sak in flies.38 Plk-4 is centrosome associated throughout the cell cycle, but phosphorylation by Cdk2-CyclinA/E near the G1/S phase transition is needed for centriole assembly.38 Overexpression of Plk-4 in cultured human cells generates several procentriole-like structures that encircle the proximal ends of the parental centriole,39,40 indicating that multiple procentrioles can form in association with a preexisting centriole. These results suggest that Plk-4/Sak is an essential positive regulator of centriole assembly.

Genetic studies in Caenorhabditis initiated a period of rapid advance in establishing the pathway of centriole assembly and the players involved. RNAi screens, chemical mutagenesis, and genomic approaches identified the novel zygotic defective ZYG-1 kinase, the spindle assembly proteins SAS-4, SAS-5, SAS-6 and the spindle defective protein SPD-2.41 Localization of SPD-2 is the essential first step and precedes appearance of ZYG-1.42 Caenorhabditis does not encode a Plk-4 homolog and it is not clear whether ZYG-1 provides the same or a different function. ZYG-1 activity is necessary for subsequent recruitment of SAS-5 and SAS-6 that are mutually dependent for localization to centrioles and necessary for subsequent recruitment of SAS-4. Formation of the central tube in forming centrioles requires SAS-5 and SAS-6 while assembly of nine singlet microtubules on the periphery of the central tube requires SAS-4.36 Homologs of SAS-5 in other species have not been reported, but homologs of SAS-6 and SAS-4 humans and flies are also needed for centriole assembly. Controlled expression experiments in human cells indicate that turnover of HsSas-6 protein by the proteasome is needed to limit assembly in S-phase to a single centriole.43

Putative homologs of SPD-2 in Caenorhabditis include DSpd-2 in Drosophila44,45 and Cep19246 in humans. Although localization indicates that each is a component of the pericentriolar matrix and centrioles, homology to SPD-2 is weak and limited to short segments of these long proteins that are dominated by coiled coil structure. The mutant phenotypes differ from SDP-2 and each other; RNAi inactivation of DSpd-2 in cultured cells showed mild effects on centrosome organization,44 but centriole duplication was not affected in somatic cells of DSpd-2 mutants.45 However, pericentriolar components fail to accumulate normally at the sperm centriole following fertilization of DSpd-2 deficient eggs.45 RNAi inactivation of Cep192 did not result in loss of centrioles tagged with centrin:GFP, but both spindle and centrosome organization was severely disrupted.46 These findings highlight the difficulty in establishing whether centrosomal proteins are true orthologs or proteins with similar, but distinct functions. The challenge is acerbated by use of RNAi methods and transposon insertion mutants that produce incomplete inactivation. Additional proteins have been implicated in centriole assembly, such as Cep135 in vertebrates40 and asterless in Drosophila47 that may not have orthologs in Caenorhabditis or humans.

Mortality of centrioles and basal bodies.

A widely held perception is that centrioles are effectively immortal; once assembled, centrioles are inherited in successive cell cycles as preformed protein structures. This perception is well founded in that centrioles are known to be highly stable to conditions that disassemble other microtubule arrays,48,49 but direct analysis of centriole longevity is limited. Two lines of evidence suggest that centriole structure in Caenorhabditis is dynamic, at least to a limited extent. One set of studies first showed that reduced expression of SAS-4, SAS-5 or SAS-6 generated shortened centrioles and reduced the corresponding size of the pericentriolar mass and microtubule aster,41 as if the aster of microtubules formed by centrosomes reflects centriole length and presumably stability. This is consistent with reduced levels of SAS-4:GFP in preassembled centrioles by reduced expression of SAS-5 or SAS-6 by RNAi, suggesting centriole assembly is reversible. The second set of studies used marked mating experiments in which sperm of wild-type males fertilized eggs expressing SAS-6:GFP. The results of this experiment showed that SAS-6 becomes associated with both newly formed centrioles as well as the preassembled parental centrioles. Taken together, these studies suggest that centrioles can disassemble, albeit perhaps more likely in centrioles with singlet microtubules or lacking δ-, ε-, η- and/or ζ-tubulins. In this context it is noteworthy that the absence of basal bodies in sensory cilia in Caenorhabditis50 could reflect disassembly of a preexisting basal body. Further, monastral bipolar spindles in which only one pole shows evidence of a centrosome44,51,52 could reflect loss of centrosome organization that is secondary to transient disassembly of centrioles.

Centriole Inheritance by Mitosis

Centrioles and cell cycle progression.

Centrioles assemble near the proximal end of a preexisting centriole that was built in an earlier cell cycle and inherited by mitosis and cell division. The concept of a preformed microtubule template or a centriole-localized genome that directs centriole assembly has been discarded for lack of experimental support.53,54 Yet, the nearly universal pattern of assembly suggests that the parental centriole contributes in some way to assembly of the new centriole. One of the first studies to address this question in cultured cells used microsurgical techniques to generate cytoplasts that contained centrioles, but lacked a nucleus and karyoplasts that contained a nucleus, but lacked centrioles (Fig. 5). The karyoplast remained viable, but new centrioles did not form,55 suggesting a preexisting centriole might be needed for new centriole assembly. Subsequent studies followed cell fragments with continuous microscopic observation,56,57 producing three important findings. First, acentriolar karyoplasts entered and completed mitosis, supporting in vitro studies showing centrosomes are not essential for assembly of bipolar spindles.58 Second, acentriolar cells successfully completed cytokinesis less often than cells with centrioles, suggesting a role for centrioles in the fidelity of cytokinesis. Third, acentriolar cells generated by cytokinesis arrested prior to entry into S phase, as if cells cannot exit G1 without a centriole in much the same way that cells cannot exit mitosis without a bipolar spindle. A centriole checkpoint offered an attractive method to block proliferation of acentriolar cells,59 but an outstanding question was whether removal of centrioles elicited a stress driven cell cycle arrest in response to loss of cell cycle regulators associated with centrosome60,61 or simply to stress induced by the mechanics of centrosome removal.

De novo assembly of centrioles in acentriolar cultured cells.

The existence of a centriole checkpoint was examined in transformed Hela cells62 that show suppressed p53 and Rb function and compromised surveillance of the G1/S checkpoint as a result. Centrioles tagged with centrin:GFP at the poles of metaphase spindles were ablated with a laser and acentriolar cells were generated by cytokinesis as in previous experiments. Acentriolar Hela cells not only progressed through the cell cycle at the same pace as sibling control cells with centrosomes, but also formed centrioles de novo (Fig. 6). Surprisingly, subsequent studies in normal non-transformed cells also showed de novo assembly of centrioles.63 Cell cycle arrest in acentriolar cells generated in earlier works55–57 was shown to be a stress response elicited by removal of centrioles by laser ablation and microsurgery that was not evident until the G1/S transition in the following cell cycle. An implication of these findings is that a preformed template is not needed to form a new centriole or to progress through the G1/S checkpoint.

Centrioles assembled de novo were comparable to centrioles formed adjacent to a parental centriole in two important ways (Fig. 6). First, de novo assembled centrioles showed normal structure at the ultra-structural level. This is in contrast to centrioles assembled de novo in cells delayed in S phase, which included a mixture of normal and malformed centrioles.64,65 Malformed centrioles assembled de novo in S phase arrested cells may, directly or indirectly, reflect delay in S phase or the methods used to elicit S phase delay. Second, de novo assembled centrioles behaved like those in non-transformed cells (Fig. 6); pro/centrioles showed random excursions throughout the cell during the first cell cycle, showing minimal pericentriolar material and virtually no microtubules that could impede movement. Similar excursions of the newly formed centrioles are evident in some cell types where the parental and newly formed centrioles disjoin rather than remaining associated as a centriole pair.66 In both cases, wandering centrioles eventually coalesced and formed stationary groups that were surrounded by a mass pericentriolar material and a radial array of microtubules.

Centriole induced suppression of de novo assembly of new centrioles.

Following de novo assembly of centrioles during the first S-phase in acentriolar cells (Fig. 6, S₁), new centrioles did not appear in the cytoplasm during the subsequent S-phase (Fig. 6, S₂) and then new centrioles only appeared adjacent to preassembled parental centrioles,63 as if the presence of a centriole somehow suppressed the de novo assembly pathway. Subsequent studies have addressed this question in Drosophila, but the results raise new questions. Inactivation of Sak kinase, like Plk-4 in humans, reduced centriole number,67 consistent with a role in centriole assembly. Sak was overexpressed in an attempt to drive centriole assembly, using ectopic promoters to drive expression of GFP tagged proteins.68 Overexpression of Sak in larval brains generated extra foci of pericentriolar components as assayed by immunofluorescence analysis. However, these foci did not generate multipolar spindles as typical of cells with extra centrosomes. The presence of centrioles or centriole precursors was not tested by ultra-structural analysis and the possibility that the foci represent aggregates of pericentriolar material without assembly of centrioles cannot be ruled out. These results alone are consistent with the view that de novo assembly of new centrioles is suppressed in cells containing centrioles.

Different results were obtained in acentriolar Drosophila eggs,68,69 where centrioles are inactivated and/or discarded during oogenesis.70 Although sufficient materials to assemble thousands of centrioles are provided to the eggs during oogenesis, centrioles do not form de novo. Rather, sperm at fertilization provides a founder centriole and centrioles duplicate once each cycle as in other cell types. High levels of Sak/Plk-4 as well as DSas-6 and DSas-4 over expression generated many masses in unfertilized eggs that were immunostained with antibodies against centrosomal proteins and many were associated with microtubules, consistent with de novo centriole assembly and centrosome organization. However, these foci did not appear to duplicate and increase in number during continuous live imaging.68 The relationship between these structures and similar microtubule asters that transiently form during oocyte activation17 is not known. Ultra-structural analysis was not performed on masses induced by overexpression of Sak or DSas-4, but transmission electron microscopy of masses generated by DSas-6 overexpression showed aberrant tube-like structures, but not recognizable centriole barrels.69 Taken together, these studies indicate that overexpression of Sak, DSas-4 and DSas-6 does not result in de novo assembly of normal centrioles, but may induce organization of centriole precursors and pericentriolar masses capable of organizing microtubules. Further advance may be afforded by development of an in vitro system for centriole assembly that allows controlled addition of one or multiple components.

Proliferation of acentriolar cells and the spindle assembly checkpoint.

Mutations in the DSas-4 gene block centriole duplication in Drosophila; ultra-structural analysis indicated the absence of identifiable centrioles in somatic cells of DSas-4 mutants.71 Spindle assembly was slowed, but mitosis continued without significant cell cycle delay. Development was remarkably normal, save for aberrant cilia and flagella that require a centriole/basal body for assembly. DSas-4 mutant animals are uncoordinated due to defects in sensory neurons and die shortly after birth. A plausible explanation offered for cell proliferation in DSas-4 mutants is that wild-type protein contributed by heterozygous females to unfertilized eggs sustains embryogenesis, but is gradually lost in sequential mitotic divisions during larval development. Gradual dilution of DSas-4 may allow mutant cells to gradually adjust to changes in centriole length and/or relocation of pericentriolar material. Similar explanations may apply to centrosomin72 and asterless,47,73 mutants in Drosophila. These results seem to confirm studies in vitro indicating that neither centrioles nor centrosomes are essential for spindle assembly, mitosis or cell cycle progression.58

The availability of acentriolar DSas-4 mutants allows tests of the role of centrioles and centrosomes in Drosophila development. The position of centrosomes in cells determines spindle positioning and the plane of cell division. Stem cells frequently utilize asymmetric positioning of spindles to allow self-renewal of one daughter as a stem cell and one daughter to differentiate. Partitioning of centrosomes during division of Drosophila germline stem cells was proposed to help keep one daughter as a stem cell and move the other toward differentiation.70,74 The role of centriole duplication and centrosomes in oogenesis and embryogenesis in Drosophila was examined in pupal ovaries of homozygous DSas-4 mutant females as well as in germline clones.75 The lack of centrioles and organized centrosomes had no effect on asymmetric division of germline stem cells or larval neuralblasts.75 These and other results in this work indicated that asymmetric inheritance of the parental centriole or asymmetric behavior of centrosomes is not essential features of stem cell divisions in Drosophila as previously suggested (refs. 76 and 77).

Perspectives

As genetic, molecular and cell biological studies continue to provide new insights into centriole inheritance, new questions emerge. If centrosomes are not essential for microtubule nucleation or spindle assembly, why are centrosomes and centrioles present in virtually all animal cells? If spindles can form without any centrioles or centrosomes, why does the presence of one centrosome generate monopolar spindles? Although canonical centrioles are not essential, do apparently acentriolar cells have unrecognizable centriole substructures that provide the function of centrioles without the structure? Does the involvement of pericentriolar proteins in centriole assembly indicate that centrioles reflect self-assembly of pericentriolar matrix? If so, what underlies recruitment and/or self-assembly of pericentriolar components? Further work may show whether function in assembly of cilia and flagella offers sufficient explanation for conservation or whether centrioles in animal cells reflect structures and functions that are not yet appreciated.

Figures and Tables

Figure 1 Centrosome structure. Pericentriolar material surrounds a pair of barrels-shaped centrioles. Minus ends of astral microtubules lie in pericentriolar matrix and plus ends extend outward. The designated plus-end of microtubules is more dynamic, frequently growing or shrinking. The opposing minus end is also dynamic although tubulin is lost, but not added in vivo.

Figure 2 Spindle fusions in γ-tubulin mutants. Spindles in embryos from wild-type (wt) and γ-Tub37C (γ-mat) mutants as indicated. Embryos were stained with antibodies against α-tubulin to visualize microtubules and a fluorescent chromatin dye to visualize chromosomes. Arrows indicate some sites of apparent spindle fusion. Magnification identical in all panels.

Figure 3 Centriole and basal body structure. Amorphous material is the first indication of a forming centriole or basal body. (A–C) The A-, B- and C-tubules appear sequentially. (D) Appendages form at the distal end of the parental centriole. (E–G) The transition zone marks the end of C-tubules in basal bodies. B- and C-tubules extend into axonemes. (H) The central pair of microtubules emerges in the transition zone, but they are not present in cilia. (E–H) The plasma membrane surrounding basal body/axonemes is depicted as a circle. Modified from tomography of centriole/basal body structure in Chlamydomonas.33

Figure 4 The centriole cycle and the spindle assembly checkpoint. (A) Normal centriole cycle. (B1) Failure in centriole duplication, but centrosomes with a single centriole assemble a bipolar spindle. (B2) A single centrosome is partitioned to daughter cells that generate a monopolar spindle. Centriole pairs are connected by fibrillar material, designated by black dots. Parental centriole shows distal appendages (blue lines) that may be vestiges of centriole assembly since they are not essential for mitosis or viability.37 Pericentriolar matrix and microtubule asters are not depicted for simplicity.

Figure 5 Centriole checkpoint at the G1/S transition. Acentriolar cells were generated by microsurgery with the sharpened tip of glass pipette56 as shown here or by laser ablation.57 Karyoplasts complete mitosis and often cytokinesis, arresting before passing the G1/S phase transition. Mitosis in control cells is depicted at the top. For simplicity, asters of microtubules in interphase cells are not shown.

Figure 6 De novo assembly of centrioles. Following laser ablation of centrioles and completion of mitosis (M₀), centrin-GFP in presumptive procentrioles (small dots) form in the first S-phase (S₁). Pro/centrioles coalesce in G1₂ when centrioles acquire a robust pericentriolar mass and an aster of microtubules. Note that immature de novo assembled centrioles are depicted at the poles of a bipolar spindle at M₁ for simplicity, but multipolar spindles were also formed. Because immature centrioles are not associated with microtubules, positioning of immature de novo assembled centrioles at spindle poles in M₁ may reflect a microtubule-independent mechanism. Control cells retaining centrioles undergo normal cell cycle progression, but only G1₁-M₁ is depicted here.

Table 1 γ-tubulin containing complexesa

	Humans	Frogs	Flies	Budding Yeast	Fission Yeast	Nematodes
γTuSCb	TUBG1	Xgam	γTub23Cc	Tub4	Gtb1	tgb-1
			γTub37C
	GCP2	Xgrip110	Dgp84	Spc97	Alp4
	GCP3	Xgrip109	Dgp91	Spc98	Alp6	glp-1
gTuRCd	GCP4	Xgrip76	Dgp75	Gfh1
	GCP5	Xgrip133	Dgp128
	GCP6	Xgrip210	Dgp163	Alp16
peripherale	GCPWD	XNedd-1	Dgp71WD

a Components of γ-tubulin complexes have been reviewed recently.17

b Components corresponding to those in γTuSC are included in γTuRCs in vertebrates and flies.17

c Drosophila encodes two differentially expressed isoforms.22

d Yeast γTuRCs sediment from extracts of vertebrates. Both γTuRC and γTuSC sediment from extracts of fly embryos.

e Nonconserved peripheral component of γTuSCs in flies and vertebrates is required for localization of γ-tubulin to spindle microtubules.

Table 2 The extended tubulin superfamilya

	γ-tubulin	δ-tubulin	ε-tubulin	η-tubulin	ζ-tubulin
humans	+	+	+
frogs	+	+	+
flies	+
nematodes	+
ciliates	+	+	+	+
trypanosomes	+	+	+		+

a Additional tubulins identified by homology43 are not listed here.

Acknowledgements

I am grateful to the editor and reviewers for critical reading of the manuscript and helpful suggestions. I want to thank S. Nickerson and the Department of Animal and Dairy Science at the University of Georgia for defraying charges for color figures.

Note

This manuscript has been previously published: Centriole Wilson P. Chernoff Y. Inheritance. Protein-Based Inheritance 2007; Austin and New York Landes Bioscience and Kluwer Academic Press 119 - 133

 Google Scholar