Centrioles Take Center Stage Wallace F

Review R487

Centrioles take center stage Wallace F. Marshall

Centrioles are among the most beautiful and Introduction mysterious of all cell organelles. Although the Centrioles are cylindrical structures found at the center of ultrastructure of centrioles has been studied in great the centrosome. Centrioles act as seeds to recruit micro- detail ever since the advent of electron microscopy, tubule nucleating material — referred to as pericentriolar these studies raised as many questions as they material — to give rise to a centrosome. Centrioles also act answered, and for a long time both the function and as structural templates to initiate the assembly of cilia and mode of duplication of centrioles remained flagella, and in this role they are referred to as basal controversial. It is now clear that centrioles play an bodies. Centrioles were an object of intense research in important role in cell division, although cells have the early days of cell biology, because of their central backup mechanisms for dividing if centrioles are location within the cell division machinery and their appar- missing. The recent identification of proteins ent self-replication. Early observations raised three central comprising the different ultrastructural features of questions of centriole biology that have remained unan- centrioles has proven that these are not just figments of swered to this day: what is the function of centrioles in cell the imagination but distinct components of a large and division; what is the molecular structure of centrioles; and complex protein machine. Finally, genetic and how do centrioles duplicate? biochemical studies have begun to identify the signals that regulate centriole duplication and coordinate the Recent data on the role of centrioles in centrosome assem- centriole cycle with the cell cycle. bly, cytokinesis and cell-cycle progression have sparked a resurgence of interest in these questions. Clearly, a Address: Department of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, Connecticut 06520, USA. genetic approach to centrioles will be essential. Unfortu- E-mail: [email protected] nately, yeast cells do not contain centrioles, but instead have morphologically distinct structures called spindle pole Current Biology 2001, 11:R487–R496 bodies. Despite tremendous advances in understanding 0960-9822/01/$ – see front matter spindle pole bodies, these studies have not yet answered © 2001 Elsevier Science Ltd. All rights reserved. the main questions of centriole biology, simply because it is unclear how spindle pole bodies are related to centrioles. Only two homologs of spindle pole body proteins have been found in animal genomes: centrin, the homolog of Cdc31, and kendrin, the homolog of Spc110 [1]. Until we identify more components of centrioles, we will lack the molecular Rosetta stone that would let us apply our knowledge of yeast spindle pole bodies to animal centrioles. Consequently, genetic approaches to centrioles have instead turned to genetic model organisms that have centrioles similar to those of mammalian cells, including Dro- sophila, Caenorhabditis elegans, and the yeast-like green alga Chlamydomonas.

In this review, I shall summarize the current state of our biochemical, genetic and ultrastructural understanding of centrioles, and how this information is leading to answers to the central questions about the function, composition and duplication of centrioles.

Centriole function The primary function of centrioles during cell division is to recruit microtubule nucleating factors into a discrete focus, the centrosome (Figure 1). Animal cells lacking centrioles do not form centrosomes, as judged by the accumulation of pericentriolar material into discrete microtubule organizing centers [2–5]. When centrioles are dissolved by injection of R488 Current Biology Vol 11 No 12

Figure 1 bipolar spindles in centriole-less cells form via an alternative, non-centrosomal pathway not normally used in somatic Normal centriole number cells [5]. In this case, microtubules nucleated by the chromosomes self-organize into a bipolar spindle with the aid Two centriole pairs of motor proteins [9,10]. This alternative spindle assembly Two centrosomes Astral bipolar spindle pathway is common in meiosis [3], consistent with the lack of both centrioles and centrosomes in meiotic cells. Thus, cells lacking centrioles do not form centrosomes, but can Proper mitotic segregation assemble spindles using this alternate pathway. The fact that cells without centrioles cannot form centrosomes, Not enough centrioles which are required to make astral microtubules, explains a long-standing observation that acentriolar spindles are One centriole pair always anastral [3,5,11–13]. Because astral microtubules One centrosome are thought to position the spindle during cytokinesis, we Monastral bipolar spindle would predict that cells lacking centrioles would show defects in spindle orientation and cytokinesis. This is Chromosome loss indeed the case.

One piece of evidence linking centrioles with cytokinesis No centrioles came from the bld2 mutation in the unicellular green alga No centrosomes Chlamydomonas. Mutant bld2 cells have severe defects in Anastral bipolar spindle centriole structure, so that instead of normal centrioles, bld2 cells only contain thin rings of singlet microtubules Chromosome loss [14]. Careful analysis of cell division in bld2 mutant strains revealed that, while spindle assembly and mitosis pro- ceeded without any problems, the spindle was often mis- Too many centrioles oriented with respect to the cleavage furrow, resulting in Multiple centriole pairs cytokinesis errors [15]. These relatively minor defects in Multiple centrosomes cell division were found with a partial loss-of-function Multipolar spindle allele of bld2. In contrast, a null allele of bld2 is lethal [16], suggesting that perhaps the structurally defective centriole formed by the original bld2 mutant cells retains some Chromosome loss essential function. As discussed below, centrioles are now Current Biology thought to contribute directly to cell-cycle progression, and this may explain the lethality in bld2 null mutants. Centriole function in cell division. Centrioles nucleate assembly of centrosomes by recruiting microtubule nucleating material into a discrete focus in the cell; centrioles (green), chromosomes (blue), Consistent with the results in algae, the removal or microtubule nucleating pericentriolar material (red), microtubules (black ablation of centrioles from mammalian cells also leads to lines). Cells with too few centrioles form either monopolar or monastral errors in cytokinesis [17–19]. These errors probably result bipolar spindles if just one centriole pair is present, and form anastral from an inability of the resulting anastral spindles to main- bipolar spindles if no centrioles are present. These abnormal spindles cannot be properly oriented during cytokinesis, leading to errors in tain the correct position as the cleavage furrow progresses. cytokinesis and chromosome segregation. Cells with too many It has long been known that, if mitotic spindles are dis- centrioles form multipolar spindles, again leading to chromosome placed within a cell during cytokinesis, they can induce segregation errors. Errors in centriole duplication thus lead to new cleavage-furrow formation in the new position chromosome loss and genomic instability. [20–22]. Failure of cytokinesis in acentriolar cells may result from an inability to anchor the spindle in one place long enough to establish or maintain a single cleavage antibodies recognizing a centriole-specific tubulin isoform, furrow. Moreover, completion of cleavage appears to coin- the centrosomes disperse, and only re-assemble when the cide with, and may be triggered by, the close approach and centrioles are restored [4], demonstrating a requirement for subsequent departure of the mother centriole from the centrioles in centrosome assembly and maintenance. midbody [18], implying that centrioles may have a second, more direct role in regulating cytokinesis. Cells lacking centrioles can still form bipolar spindles [5–8]. This was once thought to imply that centrioles are Further evidence that centriole-nucleated centrosomes posi- not needed to make centrosomes, but it turns out that tion spindles relative to the cleavage furrow in cytokinesis Review Centrioles R489

has recently been obtained through analysis of mutations progression, possibly by recruiting cell-cycle regulatory mol- in two Drosophila genes, asterless (asl) and centrosomin (cnn). ecules onto centrioles. Yet neither of these functions is In cells mutant for either of these genes, a functional cen- strictly essential for cell division. Instead, it appears that trosome fails to assemble around the centriole, as judged centrioles have evolved as a way of ensuring high fidelity by accumulation of centrosomal proteins such as γ-tubulin of chromosome segregation and spindle placement [29], by and CP190. As a result, such cells display a marked reduc- imposing spatial regulation on the inherently self-organiz- tion of astral microtubules at spindle poles [23,24]. Impor- ing mitotic apparatus. tantly, homozygous asl [25] mutant embryos, or cnn mutant embryos from homozygous cnn mutant female flies (which Centriole structure and composition will completely lack centrosomin protein), fail to develop A major limiting factor in investigating centriole function and show dramatic defects in cell division [24,26], confirm- is our current ignorance of the molecular composition of ing that centrosomes, and by extension centrioles, are essen- centrioles. A more detailed knowledge of centriole molec- tial for proper animal development. The importance of ular architecture would allow genetic and reverse genetic centrioles in animal development has been further sup- techniques to be brought to bear on investigations of ported by the finding that mutations in ZYG-1, a centro- centriole function. some-associated kinase that is required for centriole duplication in C. elegans, results in a complete failure of embry- Centriole core structure onic development [27]. Each centriole consists of a nine-fold symmetrical array of triplet microtubules, called blades (Figure 2). The distal The defects in development that are caused by centriole end contains the plus-ends of the microtubules, and tem- and centrosome mutations are probably not due to a simple plates the assembly of cilia and flagella when centrioles turn failure in spindle assembly, because bipolar spindles can into basal bodies. The other, proximal end of the centriole assemble without centrosomes. Examination of weak contains the ‘cartwheel’, a set of nine spokes connected to mutants of asl that survive through larval development a central axis. The proximal end is also the site of new cen- showed that spindle orientation was defective during the triole assembly. While a significant number of components asymmetric neuroblast divisions, consistent with a role for of the microtubule triplets have been identified, mainly as astral microtubules in spindle orientation [28]. In homozy- a byproduct of studies on flagellar doublet microtubules, gous cnn mutant offspring of heterozygous mothers, the much less is known about what molecules might compose maternal contribution of Cnn protein is sufficient to allow the interior of the centriole. development to adulthood. However, detailed analysis of cell division in these mutant flies revealed a reduction in The microtubule triplet blades contain α and β tubulin. astral microtubules that was accompanied by a significant Specific post-translational modifications of tubulin occur defect in asymmetric neuroblast division [26], similar to in the centriole blades. In particular, polyglutamylated the results with asl mutants. These results thus confirm tubulin is clearly very important for maintenance of centri- the importance of centriole-nucleated centrosomes in ori- ole integrity [4]. Additional protein components of the enting the spindle during cytokinesis, presumably via the triplet blades include Sp77, Sp83, Rib43 and tektin [30–32], astral microtubules. proteins of unknown function that were first identified as structural components of flagellar microtubule doublets. In addition to playing a role in cell division via nucleating These data suggest that a common set of underlying mole- astral microtubules, centriole-nucleated centrosomes have cular interactions establish the parallel microtubule doublets recently been found to play a role in regulating cell-cycle and triplets seen in centrioles and cilia. progression. Cells from which centrioles have been removed, either by microsurgery or laser ablation, progress Nothing whatsoever is known about the molecular compo- through mitosis but arrest in G1 of the following cell cycle sition of the cartwheel structure. γ-tubulin is located within and never reach S phase [17,19]. These studies imply that the lumen of the centriole barrel, near the proximal end centrioles can directly influence cell-cycle progression. This where the cartwheel is found [33]. Other proteins located may reflect the existence of a checkpoint that monitors cen- at the proximal end of the centriole include the NIMA triole copy number and arrests cell division in cells with (never in mitosis A)-related kinase Nek2 and the Nek2- too few centrioles, thus avoiding chromosome segregation interacting coiled-coil protein C-Nap1 [34]. (The founding errors in subsequent cell divisions. member of the NIMA family is a protein kinase that con- trols initiation of mitosis in Aspergillus nidulans.) Overex- Clearly, centrioles play an important role in cell division, pression of Nek2 [35], as well as interference with C-Nap1 both by forming the centrosome, thus giving rise to astral function either by antibody blocking or by expressing spindle poles that can be properly positioned relative to the dominant-negative constructs [36], causes centrosomes to cleavage furrow, and also by directly influencing cell-cycle split, suggesting that C-Nap1, under control of Nek2, R490 Current Biology Vol 11 No 12

Figure 2

Centriole structure. Schematic diagram Distal lumen showing main ultrastructural features of Vfl1 centrioles, indicating known protein Centrin constituents. The distal appendages function in membrane anchoring when the centriole acts as a basal body and migrates to the plasma membrane. The satellites are a focus of microtubule nucleation during interphase, but not mitosis. The proximal end of centriole is site of nucleation of new centrioles during Appendages centriole duplication, as well as the region of p210 attachment for several types of centriole- Cenexin? associated fibers. p52

Satellites PCM-1 Kendrin?

Blades αβ-tubulin δ-tubulin Sp77 Sp83 Rib43 Tektins

Fibers sf-assemblin Centrin BAp95 BAp90

Cartwheel Composition unknown

Proximal lumen γ-tubulin Nek2 C-Nap1 Current Biology

maintains the linkage between the proximal ends of the One other putative centriole component needs to be two centrioles. mentioned, namely DNA. Based on a variety of biochemical, histochemical, and genetic experiments, it was once At the other end of the centriole, the distal lumen of the thought that centrioles might contain their own DNA- barrel contains the ‘EF-hand’ protein centrin [37,38] and based genomes, but subsequent work has absolutely ruled the coiled-coil protein Vfl1p, which localizes to the inner out this possibility [42]. wall of the distal lumen [39]. Other proteins located at the distal end include the microtubule-severing protein katanin Centriole accessory structures [40] and Fa1p, a protein involved in regulating detachment Centrioles bristle with more attached superstructures of flagella from basal bodies [41]. Interestingly, another than a World War II battleship. These include fibers, protein involved in flagellar detachment, Fa2p, has been protruding appendages and fuzzy balls. Although these found to be a homolog of Nek2 (L. Quarmby, personal structures were all seen by electron microscopy over communication), suggesting that NIMA family kinases may 30 years ago, their molecular composition is only now play multiple roles in centriole structure and function. being determined. Review Centrioles R491

A set of nine short appendages extend from the distal end A variety of different fibers extend outward from the of the centriole [43]. When a centriole becomes a basal proximal end of the centriole into the rest of the cell. body, these distal appendages — also known as transition These fibers have mainly been observed and studied in fibers — connect the centriole with the plasma mem- ciliates and algae, but are present in mammalian cells brane. Studies in algae have identified one distal appendage [43]. Studies in green algae have revealed the protein protein, p210, which shares homology with the clathrin components of three distinct types of centriole-associated coat assembly protein AP180 [44], consistent with a role fiber: the striated microtubule-associated fibers, com- in attachment to membranes. Electron micrographs have posed of the segmented coiled-coil protein SF-assemblin shown association of membrane vesicles with the distal [53]; the ominously named sinister fibers, composed of a end of centrioles in the process of becoming basal bodies different segmented coiled-coil protein BAp95 [54]; and [45], possibly via an interaction with the distal appendages. the nucleus–basal body connector fibers, composed of the Another likely component of the appendages is cenexin, EF-hand protein centrin [55–57]. Another structure associ- which localizes to the same general region as the distal ated with the proximal end of the centriole is a fibrous ring appendages [46] and is only found on the mother centri- of material aptly named the ‘halo’ [58,59], the function and ole, which is normally the only one to contain appendages composition of which are completely unknown. in mammalian cells [43]. (During centriole duplication, the pre-existing centriole is called the mother centriole, Centriole duplication and the new centriole that forms next to it is called its The complex structure of the centriole raises the question, daughter.) how are new centrioles assembled? Centrioles undergo a highly precise duplication, such that each centriole gives rise Fuzzy balls called satellites are found within the pericen- to exactly one new centriole per cell cycle. Indeed, centri- triolar material, and appear to be attached to the centriole oles are the only cellular structures besides chromosomes to barrel by striated stems [43]. The function of satellites is undergo such a precise discrete duplication. This duplica- unclear, although it has been observed that, in interphase tion process is necessary to maintain the exact number of but not mitotic cells, most of the microtubules emanating centrioles per cell; there is now strong evidence that abnor- from the centrosome have their minus ends embedded in malities in centriole number accompany, and probably con- the satellites [43]. So far only one protein, PCM-1, has been tribute directly to, tumor progression [60–64]. localized to the satellite [47]. Consistent with a role for the satellite in attaching cytoplasmic microtubules during Centrioles duplicate by a remarkable process whose mech- interphase, PCM-1 can interact with microtubules via cyto- anism is completely unknown. New centrioles form adja- plasmic dynein [47]. PCM-1 is present on centrosomes cent to, and at right angles to, pre-existing centrioles. The during G1, but dissociates from centrosomes during G2 and daughter centriole does not incorporate any part of the M phases [48], exactly when the satellites stop acting as mother centriole [65] and hence is not simply generated microtubule organizing sites. PCM-1 has recently been by a splitting process. Why new centrioles form only next found in a complex with kendrin [49], a coiled coil protein to old ones is one of the longest-running questions of cen- homologous to the yeast spindle-pole-body protein Spc110 triole biology. The most obvious model is that centrioles [1]. Spc110 is involved in recruiting microtubule nucleat- contain an essential template structure needed to produce ing material to the nuclear face of the spindle pole body, a new centriole, so that new centrioles simply cannot form suggesting that its homolog kendrin might be involved in except when nucleated by a pre-existing centriole. But microtubule nucleation by the satellites. de novo assembly of centrioles has been demonstrated by careful microscopic studies on parthenogenetic develop- The two centrioles present in the cell in interphase are ment of unfertilized oocytes [66,67], suggesting that pre- joined by various connecting fibers. In algae there are two existing centrioles are not strictly needed to make new fibers, the distal connecting fiber and the proximal con- centrioles. Other examples of de novo centriole assembly necting fiber. The distal connecting fiber contains the have been seen in certain lower plants [68], oomycetes protein centrin, while the proximal connection contains [69] and protists [70,71]. the protein BAp90 [50]. These fibers fail to assemble in vfl1 and vfl3 mutants of Chlamydomonas; consequently, If centrioles can form de novo, then why does centriole centriole segregation does not occur properly, leading to assembly normally start only next to pre-existing centri- variable numbers of centrioles per cell [51,52]. Evidently oles? Prions provide a possible analogy: prions normally are these fibers keep the centriole pair together until it is time switched to the infective conformation by pre-existing prion to separate them in a controlled fashion. Based on protein proteins, but under certain conditions they can attain the localization and mutant phenotypes, it is likely that Vfl1p altered conformation spontaneously, albeit at a very low and C-Nap1 are required to anchor the distal and proximal frequency. In this model, spontaneous de novo centriole connecting fibers, respectively, to the centrioles. assembly would normally be a very slow process that could R492 Current Biology Vol 11 No 12

not compete with normal templated assembly. However, In certain cells, de novo assembly appears to be exploited under certain circumstances, such as in oocytes loaded as a way of rapidly generating large numbers of centrioles. with high concentrations of centriole precursor proteins, During early embryogenesis in the wasp Nasonia, centrioles mass action might drive centriole assembly to occur spon- form de novo all around the cortex of the embryo, even in taneously even without pre-existing centrioles present. fertilized embryos that contain centrioles [76]. Evidently, in this case the suppression of de novo centriole assembly This hypothesis is supported by the fact that the number by pre-existing centrioles is not functioning. Another of centrioles formed de novo in artificially activated sea example is ciliogenesis in mammalian ciliated epithelia, urchin eggs is strongly dependent on the strength of acti- when centrioles form de novo in clusters that are spatially vation [72]. After the activation is ended, further centriole separated from the original centrioles inherited in the pre- assembly occurs by the templated mechanism, suggesting vious division [45], again suggesting that de novo assembly that once the driving stimulus is removed, de novo assem- can occur in cells that contain centrioles. bly reverts to being an inefficient process that cannot compete with templated assembly. These data suggest Coordinating the centriole cycle with the cell that de novo assembly might be a unique feature of certain cycle cell types that are developmentally primed to generate Centriole duplication is restricted not just spatially, through large numbers of centrioles. Further evidence that de novo the influence of pre-existing centrioles, but also temporally, assembly might only occur in specialized cells containing under the control of the cell-cycle engine. The complex massive stockpiles of precursor proteins has come from structure of the centriole assembles in a series of discrete experiments in which centrioles were removed from somatic steps (Figure 3), each occurring at a different stage of the cells. The resulting centriole-less cells never recovered cell cycle [43,58,59,77–80]. In this section, we will discuss centrioles [73,74], supporting the idea that de novo assem- the centriole duplication cycle, and its coordination with the bly might simply be too slow in somatic cells to compete nuclear division cycle, by breaking the process down into with templated assembly. several steps based on morphological data.

Experiments measuring the rate of de novo centriole Step 1: Initiation of daughter centriole assembly assembly in Chlamydomonas cells, however, have argued Assembly of new centrioles begins when cells enter S against this model. Mutants with defects in centriole seg- phase [78,79]. S phase is permissive for centriole duplica- regation frequently produce centriole-less progeny cells. tion, and multiple rounds of centriole duplication can occur In contrast to mammalian cells, which arrest in G1 after in S-phase-arrested cells [81]. The initiation of centriole centrioles are removed [17,19], centriole-less Chlamy- duplication during S phase requires the activity of cyclin- domonas cells continue to divide, and centrioles re-form dependent kinase 2 (Cdk2)–cyclin E in Xenopus eggs and de novo at a rate only about two-fold slower than templated early embryos [82,83] and Cdk2–cyclin A in somatic cells assembly [75]. This indicates that de novo assembly is fast [84], and recent work has identified nucleophosmin as the enough to compete with templated assembly, and there- downstream target of Cdk2–cyclin E in this process [85]. fore that pre-existing centrioles, in addition to catalyzing The step driven by Cdk2–cyclin E or A is probably the key new centriole assembly in their immediate vicinity, also control point in coordinating centriole duplication with the must negatively regulate de novo assembly elsewhere in cell cycle. Recently a novel centrosome-associated kinase, the cell [75]. ZYG-1, has been identified through a genetic screen in C. elegans. Zyg-1 mutants have a complete failure in centriole Experiments with cell-cycle arrest mutants showed that duplication, suggesting the ZYG-1 kinase plays a role in de novo centriole assembly occurs in S phase of the cell triggering centriole duplication [27]. cycle and cannot occur in G1-arrested cells [75]. This fact explains why de novo centriole assembly was not seen in Which part of the centriole assembles first? A structure the earlier experiments on mammalian cell fragments that called the generative disc has been reported as the first are missing centrioles [73]. Because centriole-less mam- morphologically recognizable step in centriole assembly in malian cells arrest in G1 [17,19], they never reach the Paramecium [86], and is a likely candidate for the precursor correct cell-cycle stage for de novo assembly to occur. The of centriole assembly. Identifying the molecular compo- lack of de novo assembly in such these cells reflects the nents of the generative disc is thus a high priority for cell-cycle-specific regulation of centriole assembly, rather understanding centriole duplication. The centriole satel- than the inherent ability of centrioles to form de novo. Thus lites have also been proposed to initiate centriole assembly we conclude that new centrioles normally form next to old [58]. However, satellites persist separately from the daugh- ones, not because they cannot form elsewhere, but because ter centriole after centriole duplication is complete [59]. this de novo assembly pathway is normally down-regulated Moreover, only the mother centriole of a pair has an associ- under the influence of pre-existing centrioles. ated satellite structure, yet both centrioles give rise to new Review Centrioles R493

Figure 3

Centriole duplication cycle; centrioles (green), chromosomes (blue), and pericentriolar material (red), much of which is sequestered in the nucleus during interphase. Numbers refer to steps of centriole duplication as listed Cdk2ÐcyclinE/A in the text. (1) Initiation of centriole duplication. Nucleophosmin (2) Formation of a disc-like pro-centriole of Centrin γ-tubulin nine singlet microtubules. (3) Conversion of 9 HFH4 Vfl3 singlet microtubules into triplets. 1 2 (4) Elongation of procentriole into full-length Uni1 centriole. (5) Recruitment of mitotic Vfl1,3 centromatrix and microtubule nucleating material onto centrioles to produce Bld2 centrosomes. (6) Separation of centriole pairs δ-tubulin to form the two poles of a bipolar spindle. 8 S 3 (7) Detachment of daughter centriole from its G1 mother, causing loss of perpendicular arrangement. (8) Disassembly of mitotic centrosome. (9) Interconversion between G2 centriole and basal body. Telophase The Cdc25string centriole 4 cycle Anaphase

M Pericentrin 5 γ Metaphase Prophase -TuRC NuMA 7 TPX2 fizzy CP60,190 Cdc20 Ran Asp Cnn 6 Centrin Aurora ARK1 family kinase Nek2 C-Nap1 Current Biology

daughter centrioles prior to division [77], suggesting the Step 2: Formation of a ring of nine singlet microtubules satellite is not needed to initiate centriole assembly. We The most obvious sign that centriole assembly has begun is conclude that the generative disc, and not the satellite, is the formation of a ring of nine singlet microtubules called the most likely precursor to centriole assembly. the procentriole [77]. How is a large nine-fold symmetric structure built? One way would be by packing interactions The molecules that are required for initiating centriole between large subunits, similar to the mechanism by which duplication remain unknown. In yeast, the centrin homolog viral capsids attain their various symmetry forms. Alterna- Cdc31 forms the half-bridge structure that appears to give tively, some small nine-fold rotationally symmetric structure, rise directly to a new spindle pole body, suggesting that possibly even just a single molecule, might self-assemble centrin might play a similar role in centriole duplication and then propagate its symmetry to the entire structure. [87]. Indeed, inhibition of centrin function has been shown Certain repeating sequences within an α helix can generate to cause defects in both templated [88,75] and de novo [89] a nine-fold symmetric arrangement of amino acids around centriole duplication. Other proteins involved in centriole the circumference of the helix [92]. Spoke structures, such as duplication are Vfl3p [75] and γ and η tubulin [90,91], those seen in the cartwheel at the proximal end of the centri- although it is difficult to distinguish between a block at oles, could bind to such a sequence motif and extend out to this stage versus a block at the singlet microtubule assembly the perimeter of the generative disc, directing assembly of stage (step 2). microtubule singlets in the correct positions. R494 Current Biology Vol 11 No 12

Step 3: Assembly of microtubule triplet blades Drosophila, the detachment of the daughter centriole from Once the singlets form, additional microtubules assemble the mother requires the activity of Cdc20fizzy, which alongside them, forming a ring of nine triplet micro- promotes exit from mitosis by activating the anaphase tubules. Generally, the microtubules form first doublets, promoting complex [80]. In Xenopus, proteasome inhibitors, and then triplets [86]. In Chlamydomonas bld2 mutants, as well as antibodies blocking SCF ubiquitin ligase sub- centriole development arrests at the beginning of this stage, units — named after their main components, Skp1, Cullin because bld2 mutant centrioles contain only singlet micro- and an F-box protein — prevent G1 centriole detachment tubules. The Chlamydomonas mutant uni3, defective in the [98]. These results imply that detachment involves ubiq- gene for δ tubulin, allows doublets to form, but prevents uitin-dependent proteolysis, perhaps to sever a physical them from assembling into triplets [93]. linkage between the mother and daughter centrioles.

Step 4: Elongation Step 8: Disassembly of mitotic centrosome After the triplet microtubules have initiated, the entire After mitosis, pericentriolar proteins are re-sequestered in structure elongates into the full-length centriole. Elongation the newly forming nuclei. Centrosome disassembly is normally takes place during mitosis [59], and experiments accompanied by dramatic changes in centriole ultrastruc- in Drosophila have shown that the phosphatase Cdc25String, ture. For example, the electron-dense halo seen around which regulates entry into mitosis, is required for centriole the mother centriole of each pair during mitosis disappears elongation [80]. after anaphase [43,78].

Step 5: Recruitment of the mitotic centrosome Step 9 (optional): Conversion into basal bodies When mitosis begins, proteins are recruited around the During G1 phase, centrioles can become basal bodies to centrioles to give rise to the mitotic centrosome. Many form cilia and flagella. Basal body assembly requires for- centrosomal proteins, such as NuMA [94], are sequestered mation of the transition region to nucleate flagellar assem- in the nucleus during interphase, which may explain the bly, and attachment to the cell surface. In mam-malian results of elegant experiments in which eggs of the ribbon- cells, conversion of centrioles into basal bodies appears to worm Cerebratulus were microsurgically cut to produce be under control of the forkhead transcription factor enucleated cell fragments. These fragments could be HFH-4 [99]. induced to assemble microtubule asters containing de novo assembled centrioles, but only if the microsurgery was per- Conclusions: deciphering the enigma formed after nuclear envelope breakdown [95]. Most of Centrioles are suddenly becoming less enigmatic. We are the centrosome proteins do not interact with centrioles now equipped with the molecular and genetic tools to take directly, but are instead recruited by a fibrous matrix, the studies of centriole biology beyond the limits of ultrastruc- ‘centromatrix’, which assembles onto the centrioles [96]. tural description. Genetic screens in Chlamydomonas, Drosophila and Caenorhabditis elegans are currently fleshing Step 6: Separation of centriole pairs out the genetic pathway of centriole assembly. At the Prior to mitosis, the centrosome must be split into two cen- same time, modern proteomic approaches are beginning to trosomes in order to form two spindle poles. Splitting is catalog the parts list of all centriolar proteins [100], and accompanied by a separation of the two mother–daughter cell-free extract systems have been developed in which centriole pairs from each other. This separation appears to centriole assembly and centrosome recruitment can be be triggered by phosphorylation of centrin during the reconstituted [101]. After having endured a century-long G2–M transition [97], suggesting that centrin-based con- bear market, centrioles are finally coming into their own. necting fibers may keep the original centriole pair connected after each member of the pair has formed its own References daughter centriole. Centriole pair separation can also be 1. Flory MR, Moser MJ, Monnat RJ, Davis TN: Identification of a human triggered by the kinase Nek2 [35], at least one target of centrosomal calmodulin-binding protein that shares homology with pericentrin. Proc Natl Acad Sci USA 2000, 97:5919-5923. which is the protein C-Nap1 located at the base of the cen- 2. Debec A, Detraves C, Montmory C, Geraud G, Wright M: Polar triole. It is important to distinguish this pre-mitotic split- organization of gamma-tubulin in acentriolar mitotic spindles of Drosophila melanogaster cells. J Cell Sci 1995, 108:2645-2653. ting event, in which pairs of centrioles separate from each 3. Matthies HJ, McDonald HB, Goldstein LS, Theurkauf WE: Anastral other, from the post-mitotic detachment (step 7), in which meiotic spindle morphogenesis: role of the non-claret the two centrioles within one pair separate from each other. disjunctional kinesin-like protein. J Cell Biol 1996, 134:455-464. 4. Bobinnec Y, Khodjakov A, Mir LM, Rieder CL, Eddé B, Bornens M: Centriole disassembly in vivo and its effect on centrosome Step 7: Detachment of the centriole pair structure and function in vertebrate cells. J Cell Biol 1998, As cells exit mitosis and enter G1 phase, the mother and 143:1575-1589. 5. de Saint Phalle B, Sullivan W: Spindle assembly and mitosis daughter centrioles detach from each other, losing without centrosomes in parthenogenetic Sciara embryos. J Cell their mutually perpendicular arrangement [43,78,79]. In Biol 1998, 141:1383-1391. Review Centrioles R495

6. Szollosi D, Calarco P, Donahue RP: Absence of centrioles in the 31. Norrander JM, deCathelineau AM, Brown JA, Porter ME, Linck RW: first and second meiotic spindles of mouse oocytes. J Cell Sci The Rib43a protein is associated with forming the specialized 1972, 11:521-541. protofilament ribbons of flagellar microtubules in 7. Brenner S, Branch A, Meredith S, Berns MW: The absence of Chlamydomonas. Mol Biol Cell 2000, 11:201-215. centrioles from spindle poles of rat kangaroo (PtK2) cells 32. Stephens RE, Lemieux NA: Tektins as structural determinants in undergoing meiotic-like reduction division in vitro. J Cell Biol basal bodies. Cell Motil Cytoskel 1998, 40:379-392. 1977, 72:368-379. 33. Fuller SD, Gowen BE, Reinsch S, Sawyer A, Buendia B, Wepf R, 8. Debec A, Szollosi A, Szollosi D: A Drosophila melanogaster cell line Karsenti E: The core of the mammalian centriole contains lacking centriole. Biol Cell 1982, 44:133-138. γ-tubulin. Curr Biol 1995, 5:1384-1393. 9. Waters JC, Salmon ED: Pathways of spindle assembly. Curr Opin 34. Fry AM, Mayor T, Meraldi P, Stierhof Y-D, Tanaka K, Nigg EA: C-Nap1, Cell Biol 1997, 9:37-43. a novel centrosomal coiled-coil protein and candidate substrate of 10. Merdes A, Cleveland DW: Pathways of spindle pole formation: the cell cycle-regulated protein kinase Nek2. J Cell Biol 1998, different mechanisms; conserved components. J Cell Biol 1997, 141:1563-1574. 138:953-956. 35. Fry AM, Meraldi P, Nigg EA: A centrosomal function for the human 11. Wilson EB: The Cell in Development and Heredity. New York: Nek2 protein kinase, a member of the NIMA family of cell cycle Macmillan; 1928:119,150-156,508-509. regulators. EMBO J 1998, 17:470-481. 12. Wilson PG, Fuller MT, Borisy GG: Monastral bipolar spindles: 36. Mayor T, Stierhof YD, Tanaka K, Fry AM, Nigg EA: The centrosomal implications for dynamic centrosome organization. J Cell Sci protein C-Nap1 is required for cell cycle-regulated centrosome cohesion. 151 1997, 110:451-464. J Cell Biol 2000, :837-846. 37. Lechtreck K-F, Grunow A: Evidence for a direct role of nascent 13. Khodjakov A, Cole RW, Oakley BR, Rieder CL: Centrosome- basal bodies during spindle pole initiation in the green alga independent mitotic spindle formation in vertebrates. Curr Biol Spermatozopsis similis. Protist 1999, 150:163-181. 2000, 10:59-67. 38. Paoletti A, Moudjou M, Paintrand M, Salisbury JL, Bornens M: Most of 14. Goodenough UW, St. Clair HS: BALD-2: a mutation affecting the centrin in animal cells is not centrosome-associated and formation of doublet and triplet sets of microtubules in centrosomal centrin is confined to the distal lumen of centrioles. Chlamydomonas reinhardtii. J Cell Biol 66: 1975, 480-491. J Cell Sci 1996, 109:3089-3102. Loss of spatial control of the 15. Ehler LL, Holmes JA, Dutcher SK: 39. Silflow CD, LaVoie M, Tam L-W, Tousey S, Sanders M, Wu W-C, mitotic spindle apparatus in a Chlamydomonas reinhardtii mutant Borodovsky M, Lefebvre PA: The Vfl1 protein in Chlamydomonas strain lacking basal bodies. Genetics 1995, 141:945-960. localizes in a rotationally asymmetric pattern at the distal ends of 16. Preble AM, Giddings TH, Dutcher SK: Extragenic bypass the basal bodies. J Cell Biol 2001, 153:63-74. suppressors of mutations in the essential gene BLD2 promote 40. Lohret TA, Zhao L, Quarmby LM: Cloning of Chlamydomonas p60 assembly of basal bodies with abnormal microtubules in katanin and localization to the site of outer doublet severing Chlamydomonas reinhardtii. Genetics 2001, 157:163-181. during deflagellation. Cell Motil Cytoskel 1999, 43:221-231. 17. Hinchcliffe EH, Miller FJ, Cham M, Khodjakov A, Sluder G: 41. Finst RJ, Kim PJ, Griffis ER, Quarmby LM: Fa1p is a 171 kDa protein Requirement of a centrosomal activity for cell cycle progression essential for axonemal microtubule severing in Chlamydomonas. through G1 into S phase. Science 2001, 291:1547-1550. J Cell Sci 2000, 113:1963-1971. 18. Piel M, Nordberg J, Euteneuer U, Bornens M: Centrosome- 42. Marshall WF, Rosenbaum JL: Are there nucleic acids in the dependent exit of cytokinesis in animal cells. Science 2001, centrosome? In The Centrosome in Cell Replication and Early 291:1550-1553. Development. Edited by Palazzo RE, Schatten GP. San Diego: 19. Khodjakov A, Rieder CL: Centrosomes enhance the fidelity of Academic Press; 2000:187-205. cytokinesis in vertebrates and are required for cell cycle 43. Vorobjev IA, Chentsov YS: Centrioles in the cell cycle. I. Epithelial progression. J Cell Biol 2001, 153:237-242. cells. J Cell Biol 1982, 98:938-949. 20. Carlson JG: Microdissection studies of the dividing neuroblast of 44. Lechtreck K-F, Teltenkötter A, Grunow A: A 210 kDa protein is the grasshopper, Chortophaga viridifasciata. Chromosoma 1952, located in a membrane-microtubule linker at the distal end of 5:199-220. mature and nascent basal bodies. J Cell Sci 1999, 112:1633-1644. 21. Kawamura K: Studies on cytokinesis in neuroblast of the 45. Sorokin SP: Reconstruction of centriole formation and ciliogenesis grasshopper, Chortophaga viridifasciata (De Geer). I. Formation in mammalian lungs. J Cell Sci 1968, 3:207-230. and behavior of the mitotic apparatus. II. The role of the mitotic 46. Lange BMH, Gull K: A molecular marker for centriole maturation in apparatus in cytokinesis. Exp Cell Res 1960, 21:1-18. the mammalian cell cycle. J Cell Biol 1995, 130:919-927. 22. Rappaport R, Rappaport BN: Establishment of cleavage furrows by 47. Kubo A, Sasaki H, Yuba-Kubo A, Tsukita S, Shiina N: Centriolar the mitotic spindles. J Exp Zool 1974, 189:189-196. satellites: molecular characterization, ATP-dependent movement 23. Bonaccorsi S, Giansanti MG, Gatti M: Spindle assembly in towards centrioles and possible involvement in ciliogenesis. J Cell Drosophila neuroblasts and ganglion mother cells. Nat Cell Biol Biol 1999, 147:969-980. 2000, 2:54-56. 48. Balczon R, Bao L, Zimmer WE: PCM-1, a 228 kDa centrosome 24. Vaizel-Ohayon D, Schejter ED: Mutations in centrosomin reveal autoantigen with a distinct cell cycle distribution. J Cell Biol 1994, requirements for centrosomal function during early Drosophila 124:783-793. embryogenesis. Curr Biol 1999, 10:889-898. 49. Li Q, Hansen D, Killilea A, Joshi HC, Palazzo RE, Balczon R: 25. Bonaccorsi S, Giansanti MG, Gatti M: Spindle self-organization and Kendrin/pericentrin-B, a centrosome protein with homology to cytokinesis during male meiosis in asterless mutants of pericentrin that complexes with PCM-1. J Cell Sci 2001, Drosophila melanogaster. J Cell Biol 1998, 142:751-761. 114:797-809. A novel basal apparatus 26. Megraw TL, Kao L-R, Kaufman TC: Zygotic development without 50. Geimer S, Lechtreck K-F, Melkonian M: protein of 90 kD (BAp90) from the flagellate green alga functional mitotic centrosomes. Curr Biol 2001, 11:116-120. Spermatozopsis similis is a component of the proximal plates and 27. O’Connell KF, Caron C, Kopish KR, Hurd DD, Kemphues KJ, Li Y, identifies the d-(dexter) surface of the basal body. Protist 1998, The C. elegans zyg-1 gene encodes a novel regulator of White JG: 149:173-184. centrosome duplication with distinct maternal and paternal roles 51. Adams GMW, Wright RL, Jarvik JW: Defective temporal and spatial in the embryo. Cell 2001, 105: 547-558. control of flagellar assembly in a mutant of Chlamydomonas 28. Giansanti MG, Gatti M, Bonaccorsi S: The role of centrosomes and reinhardtii with variable flagellar number. J Cell Biol 1985, astral microtubules during asymmetric division of Drosophila 100:955-964. neuroblasts. Development 2001, 128:1137-1145. 52. Wright RL, Chojnacki B, Jarvik J: Abnormal basal-body number, 29. Murray AW: Centrioles at the checkpoint. Science 2001, location, and orientation in a striated fiber-defective mutant of 291:1499-1502. Chlamydomonas reinhardtii. J Cell Biol 1983, 96:1697-1707. 30. Hinchcliffe EH, Linck RW: Two proteins isolated from sea urchin 53. Lechtreck K-F, Melkonian M: Striated microtubule-associated fibers: sperm flagella: structural components common to the stable identification of assemblin, a novel 34 kD protein that forms microtubules of axonemes and centrioles. J Cell Sci 1998, paracrystals of 2-nm filaments in vitro. J Cell Biol 1991, 111:585-595. 115:705-716. R496 Current Biology Vol 11 No 12

54. Geimer S, Clees J, Melkonian M, Lechtreck K.-F: A novel 95 kD 80. Vidwans SJ, Wong ML, O’Farrell PH: Mitotic regulators govern protein is located in a linker between cytoplasmic microtubules progress through steps in the centrosome duplication cycle. J Cell and basal bodies in a green flagellate and forms striated Biol 1999, 147:1371-1377. filaments in vitro. J Cell Biol 1998, 140:1149-1158. 81. Balczon R, Bao L, Zimmer WE, Brown K, Zinkowski RP, Brinkley BR: 55. Salisbury JL, Baron A, Surek B, Melkonian M: Striated flagellar roots: Dissociation of centrosome replication events from cycles of DNA isolation and partial characterization of a calcium-modulated synthesis and mitotic division in hydroxyurea-arrested chinese contractile organelle. J Cell Biol 1984, 99:962-970. hamster ovary cells. J Cell Biol 1995, 130:105-115. 56. Wright RL, Salisbury J, Jarvik JW: A nucleus-basal body connector 82. Hinchcliffe EH, Li C, Thompson EA, Maller JL, Sluder G: Requirement in Chlamydomonas reinhardtii that may function in basal body of Cdk2-cyclin E activity for repeated centrosome reproduction in localization or segregation. J Cell Biol 1985, 101:1903-1912. Xenopus egg extracts. Science 1999, 283:851-854. 57. Wright RL, Adler SA, Spanier JG, Jarvik JW: Nucleus-basal body 83. Lacey KR, Jackson PK, Stearns T: Cyclin-dependent kinase control connector in Chlamydomonas: evidence for a role in basal body of centrosome duplication. Proc Natl Acad Sci USA 1999, segregation and against essential roles in mitosis or in 96:2817-2822. determining cell polarity. Cell Motil Cytoskel 1989, 14:516-526. 84. Meraldi P, Lukas J, Fry AM, Bartek J, Nigg EA: Centrosome 58. Rieder CL, Borisy GG: The centrosome cycle in PtK2 cells: duplication in mammalian somatic cells requires E2F and asymmetric distribution and structural changes in the Cdk2-cyclin A. Nat Cell Biol 1999, 1:88-93. pericentriolar material. Biol Cell 1982, 44:117-132. 85. Okuda M, Horn HF, Tarapore P, Tokuyama Y, Smulian AG, Chan PK, 59. Chrétien D, Buendia B, Fuller SD, Karsenti E: Reconstruction of the Knudsen ES, Hofmann IA, Snyder JD, Bove KE, Fukasawa K: centrosome cycle from cryoelectron micrographs. J Struct Biol Nucleophosmin/B23 is a target of CDK2/cyclin E in centrosome 1997, 120:117-133. duplication. Cell 2000, 103:127-140. 60. Brinkley BR, Goepfert TM: Supernumerary centrosomes and 86. Dippell RV: The development of basal bodies in Paramecium. cancer: Boveri’s hypothesis resurrected. Cell Motil Cytoskel 1998, Proc Natl Acad Sci USA 1968, 61:461-468. 41:281-288. 87. Adams I, Kilmartin JV: Spindle pole body duplication: a model for 61. Lingle WA, Lutz WH, Ingle JN, Maihle NJ, Salisbury JL: Centrosome centrosome duplication? Trends Cell Biol 2000, 10:329-335. hypertrophy in human breast tumors: implications for genomic 88. Middendorp S, Kuntziger T, Abraham Y, Holmes S, Bordes N, stability and cell polarity. Proc Natl Acad Sci USA 1998, Paintrand M, Paoletti A, Bornens M: A role for centrin 3 in 95:2950-2955. centrosome reproduction. J Cell Biol 2000, 148:405-416. 62. Pihan GA, Purohit A, Wallace J, Knecht H, Woda B, Quesenberry P, 89. Klink VP, Wolniak SM: Centrin is necessary for the formation of the Doxsey SJ: Centrosome defects and genetic instability in motile apparatus in spermatids of Marsilea. Mol Biol Cell 2001, malignant tumors. Cancer Res 1998, 58:3974-3985. 12:761-776. 63. Brinkley BR: Managing the centrosome numbers game: from 90. Ruiz F, Beisson J, Rossier J, Dupuis-Williams P: Basal body chaos to stability in cancer cell division. Trends Cell Biol 2001, duplication in Paramecium requires γ-tubulin. Curr Biol 1999, 11:18-21. 9:43-46. 64. Duensing S, Duensing A, Crum CP, Münger K: Human 91. Ruiz F, Krzywicka A, Klotz C, Keller A-M, Cohen J, Koll F, Balavoine G, papillomavirus type 16 E7 oncoprotein-induced abnormal Beisson J: The SM19 gene, required for duplication of basal centrosome synthesis is an early event in the evolving malignant bodies in Paramecium, encodes a novel tubulin, η-tubulin. phenotype. Cancer Res 2001, 61:2356-2360. Curr Biol 2000, 10:1451-1454. 65. Kochanski RS, Borisy GG: Mode of centriole duplication and 92. Satir P, Satir B: A model for ninefold symmetry in alpha keratin distribution. J Cell Biol 1990, 110:1599-1605. and cilia. J Theor Biol 1964, 7:123-128. 66. Miki-Noumura T: Studies on the de novo formation of centrioles: 93. Dutcher SK, Trabuco EC: The UNI3 gene is required for assembly aster formation in the activated eggs of sea urchin. J Cell Sci of basal bodies of Chlamydomonas and encodes δ-tubulin, a new 1977, 24:203-216. member of the tubulin superfamily. Mol Biol Cell 1998, 67. Palazzo RE, Vaisberg E, Cole RW, Rieder CL: Centriole duplication 9:1293-1308. in lysates of Spisula solidissima oocytes. Science 1992, 94. Nachury MV, Maresca TJ, Salmon WC, Waterman-Storer CM, 256:219-221. Heald R, Weis K: Importin beta is a mitotic target of the small 68. Mizukami I, Gall J: Centriole replication II. Sperm formation in the GTPase Ran in spindle assembly. Cell 2001, 104:95-106. fern, Marsilea, and the cycad, Zamia. J Cell Biol 1966, 29:97-111. 95. Yatsu N: The formation of centrosomes in enucleated egg- 69. Heath IB, Kaminskyj SGW, Bauchop T: Basal body loss during fragments. J Exp Zool 1905, 2:287-312. fungal zoospore encystment: evidence against centriole 96. Schnackenberg BJ, Palazzo RE: Identification and function of the autonomy. J Cell Sci 1986, 83:135-140. centrosome centromatrix. Biol Cell 1999, 91:429-438. 70. Fulton C, Dingle AD: Basal bodies, but not centrioles, in Naegleria. 97. Lutz W, Lingle WL, McCormick D, Greenwood TM, Salisbury JL: J Cell Biol 1971, 51:826-836. Phosphorylation of centrin during the cell cycle and its role in 71 Grimes GW: Morphological discontinuity of kinetosomes during centriole separation preceding centrosome duplication. J Biol the life cycle of Oxytricha fallax. J Cell Biol 1973, 57:229-232. Chem 2001, in press. 72. Just EE: Initiation of development in the egg of Arbacia. I. Effect of 98. Freed E, Lacey KR, Huie P, Lyapina SA, Deshaies RJ, Stearns T, hypertonic sea-water in producing membrane separation, Jackson PK: Components of an SCF ubiquitin ligase localize to the cleavage, and top-swimming plutei. Biol Bull 1922, 43:384-400. centrosome and regulate the centrosome duplication cycle. 73. Maniotis A, Schliwa M: Microsurgical removal of centrosomes Gen Dev 1999, 13:2242-2257. blocks cell reproduction and centriole generation in BSC-1 cells. 99. Brody SL, Yan XH, Wuerffel MK, Song SK, Shapiro SD: Ciliogenesis Cell 1991, 67:495-504. and left-right axis defects in forkhead factor HFH-4 null mice. 74. Sluder G, Miller FJ, Rieder CL: Reproductive capacity of sea urchin Am J Respir Cell Mol Biol 2000, 23:45-51. centrosomes without centrioles. Cell Motil Cytoskel 1989, 100. Lange BM, Bachi A, Wilm M, Gonzalez C: Hsp90 is a core 13:264-273. centrosomal component and is required at different stages of the 75. Marshall WF, Vucica Y, Rosenbaum JL: Kinetics and regulation of centrosome cycle in Drosophila and vertebrates. EMBO J 2000, de novo centriole assembly: implications for the mechanism of 19:1252-1262. centriole duplication. Curr Biol 2001, 11:308-317. 101. Suddith AW, Vaisberg EA, Kuznetsov SA, Steffen W, Rieder CL, 76. Tram U, Sullivan W: Reciprocal inheritance of centrosomes in the Palazzo RE: Centriole duplication, centrosome maturation and parthenogenetic Hymenopteran Nasonia vitripennis. Curr Biol spindle assembly in lysates of Spisula solidissima oocytes. 2000, 10:1412-1419. Methods Mol Biol 2001, 161:215-228. 77. Murray RG, Murray AS, Pizzo A: The fine structure of mitosis in rat thymic lymphocytes. J Cell Biol 1965, 26:601-619. 78. Robbins E, Jentzsch G, Micali A: The centriole cycle in synchronized HeLa cells. J Cell Biol 1968, 36:329-339. 79. Kuriyama R, Borisy GG: Centriole cycle in chinese hamster ovary cells as determined by whole-mount electron microscopy. J Cell Biol 1981, 91:814-821.