InteractiveFly: GeneBrief

SAK: Biological Overview | References

SAK/PLK4 is a distinct member of the polo-like kinase family. SAK^−/− mice die during embryogenesis, whereas SAK^+/− mice develop liver and lung tumors and SAK^+/− MEFs show mitotic abnormalities. However, the mechanism underlying these phenotypes is still not known. This study shows that downregulation of SAK in Drosophila cells, by mutation or RNAi, leads to loss of centrioles, the core structures of centrosomes. Such cells are able to undergo repeated rounds of cell division, but display broad disorganized mitotic spindle poles. SAK mutants lose their centrioles during the mitotic divisions preceding male meiosis but still produce cysts of 16 primary spermatocytes as in the wild-type. Mathematical modeling of the stereotyped cell divisions of spermatogenesis can account for such loss by defective centriole duplication. The majority of spermatids in SAK mutants lack centrioles and so are unable to make sperm axonemes. Finally, this study shows that depletion of SAK in human cells also prevents centriole duplication and gives rise to mitotic abnormalities. Thus SAK/PLK4 is necessary for centriole duplication both in Drosophila and human cells. Drosophila cells tolerate the lack of centrioles and undertake mitosis but cannot form basal bodies and hence flagella. Human cells depleted of SAK show error-prone mitosis, likely to underlie its tumor-suppressor role (Bettencourt-Dias, 2005).

Polo-like kinases (Plks) belong to a conserved family of mitotic serine-threonine protein kinases that play key roles in centrosome function and are misregulated in many human tumors. Two branches of the family have emerged in metazoans, and they are represented in Drosophila by Polo and SAK (also called Plk4) (Lowery, 2005). Whereas much is now known about the mitotic functions of the founder member of the family, Drosophila Polo, and its related mammalian counterparts (Plk1-3), the precise mitotic roles of SAK remain obscure. Conservation of the structures of the two types of Plks throughout evolution suggests that different roles may have been preserved. Both branches of the family have an amino-terminus kinase domain and a regulatory C-terminal domain that contains conserved polo boxes (PBs). However, whereas the two PB domains of mammalian Plk1 interact with each other to create a positively charged cleft able to bind phosphopeptides, the SAK PB forms an intermolecular homodimer in which different sequences are exposed (Lowery, 2005; Elia, 2003. Moreover, SAK has a second very divergent PB (cryptic PB) (Leung, 2002; Swallow, 2005) that does not bind to its conserved PB. The Plk1-3 group are more akin to the single Plks found in the yeasts, such as Cdc5 in S. cerevisiae, and expression of either Plk1 or Plk3, but not SAK, rescues the mitotic defects of temperature-sensitive cdc5-1 mutant cells (Bettencourt-Dias, 2005).

Drosophila Polo and its closest mammalian counterpart, Plk1, are associated with centrosomes, kinetochores, and the late-mitotic central spindle, reflecting their functions in centrosome maturation, in the metaphase-anaphase transition, and in cytokinesis. SAK also localizes to the centrosome, and SAK^−/− mice die shortly after gastrulation, showing a 20-fold increase in cell death (Leung, 2002; Hudson, 2001). It was recently reported that elderly SAK^+/− mice display a 15-times-higher incidence of spontaneous liver and lung cancers than their wild-type littermates (Ko, 2005). Similarly, the human SAK gene maps to a chromosome region, 4q28, that is frequently rearranged in hepatocellular carcinomas. Multipolar mitotic spindles were reported in livers and MEFS of SAK^+/− mice, suggesting that haploinsufficiency for tumor suppression may result from chromosome instability in the oncogenic pathway. This study addresses the underlying cause of these mitotic defects, showing that in both Drosophila and human cells, SAK is required for centriole duplication. It is therefore essential for centrosome integrity and thereby fidelity of the mitotic apparatus. Moreover, SAK is also required for development of axonemal structures, reflecting the dual nature of centrioles and basal bodies (Bettencourt-Dias, 2005).

To examine SAK function, more than 70% of SAK mRNA was depleted in cultured Drosophila cells. A 12-fold increase was observed in the percentage of mitotic cells that had no γ-tubulin at the poles (1.6%, controls versus 24.7%, SAK RNAi). Other proteins typically recruited to Drosophila centrosomes in mitosis were absent: CP190 and Cnn. In a recent survey of the cell-cycle function of all the protein kinases in Drosophila, loss of function of only one other kinase, Polo, led to loss of γ-tubulin from the centrosome, consistent with the known function of Polo/Plk1 in centrosome maturation. However, in contrast to the striking metaphase arrest following polo RNAi, there was no change in mitotic index or in the flow-cytometry profile of DNA content after SAK RNAi. Thus, in S2 cells, SAK is required for centrosome integrity but not for progression through the cell cycle or cell survival (Bettencourt-Dias, 2005).

Dividing cells were examined in the CNS of SAK mutant larvae. Such cells showed an extremely similar phenotype to SAK RNAi cultured cells: a notable absence of γ-tubulin from the poles of spindles that were often disorganized and splayed. Meiotic spindles in testes of such mutants also often lacked γ-tubulin and Cnn at the poles. It was found that Drosophila SAK localized to centrosomes in both interphase and mitosis of wild-type cells, consistent with its having a function in ensuring centrosome integrity (Bettencourt-Dias, 2005).

In contrast to Polo-depleted cells, the poles of SAK-depleted cells were broad, similar to the mitotic figures of a Drosophila acentriolar cell line. This led to a determination of whether the innermost centrosomal structures, the centrioles, might be disrupted by depletion of SAK. It was found that the majority of mitotic cells that showed no γ-tubulin at the poles following SAK RNAi had no detectable pericentrin-like protein, normally present in both centrioles and the pericentriolar material (PCM). This was in contrast to polo or cnn RNAi cells, where D-PLP was still present in more than 75% of cells lacking γ-tubulin at both poles, reflecting the roles of Polo and Cnn in centrosome maturation but not centriolar integrity. In interphase, the majority of SAK RNAi cells had either zero or a single centrosome, whereas interphase polo or cnn RNAi cells did not show a significant change in centrosome number. The absence of D-PLP in a large proportion of interphase SAK RNAi cells suggested that they might be missing both centrioles and PCM. To confirm loss of centrioles after SAK RNAi, such cells were stained for the centriole marker centrin and then embedded and serially sectioned them for transmission electron microscopy (TEM). In SAK-depleted cells in which centrin staining was absent at the spindle poles, no centrioles were detected by TEM. Thus, the downregulation of SAK leads to loss of centrioles from affected cells, with no effect on the ability of those cells to proliferate (Bettencourt-Dias, 2005).

To determine whether loss of centrosomes also occurred in SAK mutants and still permit proliferation of diploid cells, the central nervous system of mutant larvae was examined. This revealed cells with two, one, or zero centrosomes, a phenotype identical to SAK RNAi. Moreover, the brains of mutant larvae were of normal size and the proportion of brain cells in mitosis was comparable to the wild-type, indicating no obvious defects in cell-cycle progression. SAK mutants were able to pupate, and adults eclosed from the pupae. However, the majority of adults were uncoordinated and died after getting stuck in the food. This phenotype is similar to D-PLP mutations that cause defects in basal bodies (Martinez-Campos, 2004), the centriolar-derived structures required for formation of cilia in neurons of type-I sensory organs that function in transduction of sensory stimuli. The lack of centriole markers in cells of the larval central nervous system of SAK mutants suggests that the uncoordinated adult phenotype is likely to reflect absence of basal bodies in sensory neurons (Bettencourt-Dias, 2005).

Whether the above defects could reflect a failure to assemble new centrosomes was examoned. This could account for the increased frequency of cells with only one centrosome in SAK mutants, if new centrosomes are not formed but cells continue to cycle. The prediction was that if cultured cells were allowed to continue dividing while being exposed to SAK dsRNA, the average number of centrosomes per cell would decrease with time. This was tested by repeatedly transfecting cells with SAK dsRNA at 4-day intervals. Successive transfections led to a progressive increase in the proportion of cells with no centrosomes, rising to greater than 93% of cells after 16 days. This dilution is consistent with SAK's having a role in centrosome assembly. Such a role was further substantiated by the finding that transfection of S2 cells with the active, but not inactive, kinase led to an increase in cells with more than two D-PLP foci. These foci are unlikely to result from aborted cell division because the majority of those cells had a single nuclei. These foci behaved as microtubule organizing centers clustering at the poles of mitotic spindles. Although additional experiments will be needed to verify the origin of the multiple foci, these results suggest that overexpression of SAK may result in multiple centrosomes (Bettencourt-Dias, 2005).

To determine whether the phenotypes observed in SAK mutant flies were associated with defective centrosome separation, abnormal centrosome inheritance, or problems in centriole duplication, spermatogenesis was examined in SAK mutant males. The germline has a stereotyped pattern of mitotic and meiotic divisions, a pattern that allows the cellular history of centrosomes and centrioles to be deduced. Additionally, their centrioles are approximately 10-fold longer than those found in other Drosophila cells and can be easily visualized by fluorescence of a GFP-PACT fusion protein harboring the centriole-targeting domain of D-PLP protein (Bettencourt-Dias, 2005).

A very high proportion of primary spermatocytes from SAK mutants had no centrioles at one or both spindle poles in meiosis I. When centrioles were absent, the spindle poles were broad and there were no astral microtubules, or very disorganized spindles were formed. Transverse TEM sections of wild-type sperm tails revealed the classic '9 + 2' axonemal microtubules that were missing from the majority of elongating SAK spermatids. Accordingly, the majority of sperm from SAK mutant testes were nonmotile, and males were sterile. Frequently, cells that had no axonemes had irregular size and numbers of mitochondrial derivatives per cell. This is usually associated with defective chromosome segregation and cytokinesis during meiosis. The lack of a strong spindle-assembly checkpoint in meiosis makes cells more likely to progress all the way through both meiotic divisions even in the presence of abnormalities. When axonemes were present, they appeared to be normal in structure (Bettencourt-Dias, 2005).

Wild-type primary spermatocytes enter meiosis I (MI) with a pair of centrioles at each spindle pole. Daughter cells inherit a pair of centrioles that are not duplicated in Drosophila. These centrioles separate, each generating a centrosome at the pole of the second meiotic spindle. Thus, each daughter spermatid inherits a single centriole. Failure to observe more than a single centriole in the products of meiosis II in the SAK mutant led to discarding of the possibility of defects in centriole separation (Bettencourt-Dias, 2005).

The cysts of 16 primary spermatocytes encapsulate the history of centriole duplication in the four preceding cell divisions. It was therefore asked whether the number of centrioles per cell in SAK mutant cysts could reflect abnormal centriole duplication. Mature cysts of primary spermatocytes in both wild-type and SAK mutant testes always contained 16 spermatocytes of comparable size, indicating success in the four rounds of premeiotic mitosis. However, whereas wild-type primary spermatocytes contained four centrioles (64 per cyst), the majority of spermatocytes from SAK mutants had no centrioles, and a smaller proportion of cells had intermediate numbers between one and four. The absence of cells having more than two centrosomes/four centrioles provided a second indication of the lack of defects in centrosome segregation to daughter cells. To test whether the distribution of centrioles observed resulted from a defect in centriole duplication, a mathematical model was generated to describe the centriole-duplication cycle and adjusted it for the four mitotic divisions of germ cells in a cyst. Multitype branching-process theory was used to evaluate the distribution of cells with a given number of centrioles. The model has some analogies with previous analytical work on plasmid copy number in bacteria. It assumes that centrosome separation and segregation is perfect, and that the variable number of centrioles in G2 after the four germline divisions is due to partially defective centriole duplication having a probability θ. Generating functions were built to follow the dynamics of the mean number of cells with a given number of centrioles. The proportion of cells with a given number of centrioles in G2 was evaluated after four cell divisions. Finally, a value of θ was found that best fit the empirical data. The function is very peaked around 0.55, fitting the empirical data very well and giving high confidence in the estimated duplication rate. Thus, it was possible to model the reduction in centriole number during the premeiotic divisions of SAK hypomorphic mutants solely by assuming reduced success of centriole duplication (Bettencourt-Dias, 2005).

Finally, it was asked whether the role of SAK in centriole duplication was conserved in human cells and SAK siRNA conditions were developed that reduced levels of SAK transcripts by more than 70%. Such depletion resulted in a 10-fold increase of HeLa cells with just one centriole. This was associated with an increased mitotic index and a higher-than-2-fold increase in apoptosis, leading to a decrease in cell number. A similar reduction was observed in centriole number after SAK RNAi in U2OS cells. It is known that centrioles continue to replicate when U2OS cells are blocked in S phase by treatment with the DNA polymerase α-inhibitor aphidicolin (AF) or the ribonucleotide reductase inhibitor hydoxyurea (HU). Following SAK RNAi, it was found that the number of cells accumulating supernumerary centrosomes in an S phase block is less than half that of control cells. Thus, downregulation of SAK reduces centriole duplication in cells blocked in S phase. There was no increase in the proportion of cells with a single centriole after SAK RNAi when cells were inhibited from dividing by treatment with AF or HU, a result consistent with inhibition of centriole duplication. It was therefore conclude that the human SAK kinase is required for centriole duplication in both HeLa and U2OS cells and for centriole reduplication in AF- or HU-treated U2OS cells (Bettencourt-Dias, 2005).

Reduced centriolar number was associated with a 6-fold increase in abnormal mitotic spindles 72 hr after SAK SiRNA in HeLa cells. These included monopolar and multipolar spindles. Similar defects were observed in U2OS cells. Curiously, in a few cases, recruitment of γ-tubulin to the acentriolar poles of multipolar spindles was seen, suggesting that acentriolar poles may organize some PCM in mammalian cells. No such cases were observed in Drosophila. In summary, spindle organization is affected in the absence of two canonical centrosomes having two centrioles, and this is likely to contribute to the reduced chromosome-transmission fidelity that has been suggested from observations of SAK-deficient mouse cells (Bettencourt-Dias, 2005).

This study has shown that depletion of SAK, but not Polo, leads to cells with a reduced number of centrioles. Centriole loss in a cycling population of cells can arise through defective centrosome duplication, abnormal separation of centrosomes at entry to mitosis, or abnormal centrosome segregation to the daughter cells in cytokinesis. In both human and Drosophila cells that there is no abnormal centrosome segregation in the absence of SAK because there is always loss but never gain of centrosomes/centrioles after cell division. Moreover, centriole reduplication in S phase is reduced after SAK knockdown in the absence of cell division in human cells. This analysis of centriole distribution in Drosophila SAK spermatids revealed no defects in centrosome separation. Moreover, the distribution of centrioles in cysts of 16 primary spermatocytes was consistent with a mathematical model assuming defects in centriole duplication. Together, these observations point to a conserved role of this member of the PLK family in centriole duplication. However, it cannot be entirely excluded that there is an additional function for SAK in mitosis until a complete loss-of-function mutant is isolated (Bettencourt-Dias, 2005).

Experiments in C. elegans have suggested five proteins to be important for centrosome duplication in embryogenesis, and these include one protein kinase, ZYG1 (O'Connell, 2001). Although ZYG-1 has only low sequence similarity to Drosophila SAK, it is the closest homolog in a BLAST search, and SAK may thus represent the ortholog of ZYG-1 in flies and vertebrates (Bettencourt-Dias, 2005).

What could be the role of SAK in centriole duplication? Structurally compromised centrioles have never been observed either in SAK mutants or after SAK depletion in either Drosophila or human tissue-culture cells. Moreover, overexpression of SAK, but not of the inactive kinase, leads to the formation of multiple D-PLP foci, suggestive of overduplication of centrioles. Together, these data suggest that SAK has a regulatory role in centriole duplication. ZYG-1 has been found to be high in the hierarchy of molecules necessary for centriole assembly and essential for the recruitment of SAS-6/SAS-5 to the centriole (Leidel, 2005), but it remains to be discovered whether this role has been conserved in Drosophila and vertebrates (Bettencourt-Dias, 2005).

The perdurance of centrosomal structures in SAK mutants studied here may result from some residual SAK function due to the hypomorphic nature of this allele and/or from remaining wild-type maternal SAK protein provided by the heterozygous mother. Such perdurance of maternal protein to the third larval-instar stage is a common feature of mitotic mutants in Drosophila. Nonetheless, SAK mutants provide the first opportunity to assess the consequences of the absence of centrioles and centrosomes upon the development of an organism. Our examination of the larval central nervous system suggests that as many as 72% of dividing cells have fewer centrosomes than expected, with 28% possessing no centrosomes at all. In cysts of primary spermatocytes, only 8% of the cells have the correct number of centrioles, whereas 68% have none. It has been proposed that centrosomes are not required for the formation of mitotic spindles, but do provide speed and fidelity to this process and so may be necessary for proper cell-cycle progression. it is conclude that in cultured Drosophila cells and in the whole organism, centrosomes are not essential for mitotic progression or cell survival. However, the absence of centrioles, and hence basal bodies, compromised both meiotic divisions and the formation of sperm axonemes. The only cilia and flagella known in the fly are found in the peripheral nervous system and in the male germline. Accordingly, it was found SAK mutants to be both uncoordinated and sterile (Bettencourt-Dias, 2005).

What could be the consequences of the lack of SAK in vertebrates? Mammalian cells appear to be more sensitive to depletion of SAK and the lack of centrosomes than Drosophila cells. There was a significant increase in abnormal mitoses, in mitotic index, and in apoptosis after depletion of SAK in human cells. Mitotic abnormalities and an increase in cell death are also observed after depletion of SAS-6, a protein involved in centriole replication, in U2OS cells and may thus be a general consequence of centriole loss in vertebrates. Mice homozygotes for SAK die very early during embryogenesis (Hudson, 2001), hindering study of the effects on cilia formation and development. However, these embryos showed increased cell death and a higher mitotic index. SAK^+/− MEFs also show abnormal spindles and chromosome segregation, and, significantly, SAK^+/− mice are more prone to develop cancer (Ko, 2005). Human Plk4 has been shown to be a key regulator of centriole duplication. Both gain- and loss-of-function experiments demonstrate that Plk4 is required - in cooperation with Cdk2, CP110 and Hs-SAS6 - for the precise reproduction of centrosomes during the cell cycle (Habedanck, 2005). The link between centrosome duplication, embryonic survival, and haploinsufficiency for tumor suppression is also seen with nucleophosmin, a regulator of centrosome duplication. Together, these studies reinforce the link between centrosome defects and oncogenesis. The sensitivity with which two tumor cell lines undergo aberrant mitosis and cell death after downregulation of the SAK kinase suggests that it may be a valuable target for cancer therapy (Bettencourt-Dias, 2005).

Sequential protein recruitment in C. elegans centriole formation

Formation of the microtubule-based centriole is a poorly understood process that is crucial for duplication of the centrosome, the principal microtubule-organizing center of animal cells. Five proteins have been identified as being essential for centriole formation in C. elegans: the kinase ZYG-1, as well as the coiled-coil proteins SAS-4, SAS-5, SAS-6, and SPD-2. The relationship between these proteins is incompletely understood, limiting understanding of how they contribute to centriole formation. This study established the order in which these five proteins are recruited to centrioles, and molecular epistasis experiments were conducted. SPD-2 is loaded first and is needed for the centriolar localization of the four other proteins. ZYG-1 recruitment is required thereafter for the remaining three proteins to localize to centrioles. SAS-5 and SAS-6 are recruited next and are needed for the presence of SAS-4, which is incorporated last. These results indicate in addition that the presence of SAS-5 and SAS-6 allows diminution of centriolar ZYG-1. Moreover, astral microtubules appear dispensable for the centriolar recruitment of all five proteins. Several of these proteins have homologs in other metazoans, and it is expected that the assembly pathway that stems from this work is conserved (Delattre, 2006).

Centrioles are minute cylindrical structures that contain nine sets of microtubules arranged in a radial fashion. At the onset of the centrosome duplication cycle, the two tightly apposed centrioles split slightly from one another. Each of these mother centrioles then seeds formation of a daughter centriole. Centriole formation has been described by ultrastructural analysis in vertebrate cells that primarily monitored the growth of the microtubules, which constitute the defining feature of centrioles. By contrast, the molecular tenets of this assembly process have remained elusive (Delattre, 2006).

The C. elegans embryo has proven well suited for investigating centrosome duplication. The five proteins known to be essential for centriole formation in this organism are enriched at centrioles and present at lower levels in the cytoplasm of early embryos. ZYG-1 is found at centrioles primarily during anaphase, whereas the four other proteins are centriolar throughout the cell cycle. Furthermore, SPD-2 is enriched in the PCM compared to the cytoplasm, possibly reflecting its additional role in PCM assembly (Delattre, 2006).

The order in which ZYG-1, SAS-4, SAS-5, SAS-6, and SPD-2 are recruited to centrioles was investigated. Experiments were designed that distinguish de novo centriolar recruitment from the prior presence of proteins at centrioles. Such experiments are rendered possible because the sperm contributes the sole pair of centrioles to the newly fertilized embryo. These two centrioles split slightly from one another, and each seeds the formation of a daughter centriole. Because the initial pair of centrioles is of paternal origin, one can assay specifically centriolar recruitment or exchange that occurs in the one-cell-stage embryo, provided the centrioles contributed by the sperm do not harbor the protein under scrutiny (Delattre, 2006).

Initially, SPD-2 was analyzed. Because GFP-SPD-2 is not present in sperm, in contrast to the endogenous protein, the time at which GFP-SPD-2 is first detected at centrioles was determined in one-cell-stage embryos. Double labeling was used with antibodies against SAS-4 to mark all centrioles and against GFP to detect GFP-SPD-2 recruitment. GFP-SPD-2 was first detected at centrioles during meiosis I. Even though endogenous SPD-2 is present in sperm, it is lost rapidly after fertilization in embryos depleted of SPD-2. Therefore, recruitment of the endogenous protein was studied and it was found that, as for GFP-SPD-2, SPD-2 is first detected at centrioles during meiosis I. Endogenous ZYG-1 is not present in sperm, which enabled assay of its recruitment after fertilization. ZYG-1 was also first detected at centrioles during meiosis I. In conducting these experiments, it was noted that the paternally contributed centrioles can be first distinguished a single entities during meiosis II, indicating that splitting has occurred by that time. Overall, it is concluded that SPD-2 and ZYG-1 centriolar recruitment initiates as early as meiosis I, prior to when splitting of the centriole pair can be observed by light microscopy (Delattre, 2006).

Next, SAS-5 was examined. In this case, both the endogenous protein and GFP-SAS-5 are present in sperm centrioles. Therefore, marked mating experiments were performed by crossing hermaphrodites expressing GFP-SAS-5 to wild-type males, the sperm of which provide centrioles not carrying the fusion protein. By contrast to the situation with SPD-2 and ZYG-1, it was found that GFP-SAS-5 is not present at centrioles during meiosis I. Instead, GFP-SAS-5 is first detected weakly at the end of meiosis II, with the centriolar signal becoming more robust thereafter. Because GFP-SAS-6 is not present in sperm, in contrast to the endogenous protein, it was possible to assess when the fusion protein is first recruited to centrioles after fertilization. It was found that centriolar GFP-SAS-6 is first detected shortly after meiosis II and more robustly soon thereafter, much like GFP-SAS-5. This is in line with the fact that SAS-5 and SAS-6 physically interact and are mutually dependent for their centriolar localization. Overall, it is concluded that SAS-5 and SAS-6 are recruited after SPD-2 and ZYG-1 (Delattre, 2006).

Because both SAS-4 and GFP-SAS-4 are present in sperm centrioles 3 and 4, marked mating experiments were conducted to investigate GFP-SAS-4 centriolar recruitment in one-cell-stage embryos. GFP-SAS-4 was incorporated progressively to centrioles during the first cell cycle, starting at the time of pronuclear formation. Taken together, these observations establish the following temporal sequence of recruitment to centrioles: first, SPD-2 and ZYG-1; second, SAS-5 and SAS-6; and third, SAS-4 (Delattre, 2006).

Next, whether this temporal sequence corresponds to related episatic interactions was assessed. In one extreme scenario, the five proteins could be recruited independently of one another. Alternatively, the proteins that are recruited early in the sequence may be needed for the presence of some that are recruited later. Whether SPD-2 is required for the centriolar recruitment of the other four proteins was investigated.ZYG-1, GFP-SAS-5, GFP-SAS-6, and GFP-SAS-4 all fail to be recruited to centrioles in spd-2(RNAi) embryos. Moreover, levels of SAS-4 on paternally contributed centrioles are diminished in spd-2(RNAi) embryos compared to the wild-type, as suggested by previous observations. Because SAS-4 is stably associated with the centriole in the wild-type, this indicates that SPD-2 also plays a role in maintaining SAS-4 after its incorporation into centrioles. In a converse set of experiments, it was found that SPD-2 distribution is not altered in zyg-1(RNAi), sas-5(RNAi), sas-6(RNAi), or sas-4(RNAi) embryos. Overall, it is concluded that SPD-2 controls centriolar recruitment of the four other proteins (Delattre, 2006).

ZYG-1, which is required for the presence of centriolar SAS-5 and SAS-6, which are themselves needed for GFP-SAS-4 recruitment, was investigated. In a converse set of experiments, ZYG-1 distribution was examined in embryos compromised for SAS-5, SAS-6, or SAS-4 function. In the wild-type, levels of ZYG-1 at centrioles are regulated across the cell cycle, with the signal being minimal during interphase and maximal during anaphase. ZYG-1 still localizes to centrioles in sas-5(RNAi) embryos as well as in sas-5(t2079) mutant embryos, in which SAS-5 and SAS-6 are not present at centrioles. ZYG-1 also localizes to centrioles in sas-6(RNAi) and sas-4(RNAi) embryos. Together, these observations establish that ZYG-1 acts upstream of SAS-5 and SAS-6 centriolar recruitment, which themselves act upstream of SAS-4 centriolar recruitment (Delattre, 2006).

In the course of these experiments it was discovered that ZYG-1 levels at centrioles remain high throughout the cell cycle in sas-5(t2079) mutant embryos and sas-6(RNAi) embryos. By contrast, levels of centriolar ZYG-1 still oscillate across the cell cycle in sas-4(RNAi) embryos, with the signal being minimal during interphase and maximal during anaphase. Together, these results indicate that SAS-5 and SAS-6 are required for the diminution of centriolar ZYG-1 during interphase. Because SAS-5 and SAS-6 are present in the cytoplasm but absent from centrioles in sas-5(t2079) mutant embryos, these results suggest in addition that this requirement reflects the presence or activity of centriolar SAS-5 and SAS-6 (Delattre, 2006).

It was of interest to place the recruitment of centriolar microtubules in the pathway that emerges from this study. However, their recruitment could not be assayed using GFP-β-tubulin, because the fusion protein is also incorporated in the remainder of the microtubule cytoskeleton, masking the specific centriolar signal. Therefore, the timing of centriolar microtubule recruitment relative to the five proteins discussed above is not known. Nevertheless, attempts were made to test whether astral microtubules are required for the recruitment of these proteins using RNAi against the alpha-tubulin gene tba-2 (Delattre, 2006).

In severely affected tba-2(RNAi) embryos, tubulin is detected only in paternally contributed centrioles and their immediate vicinity, as expected from the fact that RNAi does not target sperm under these experimental conditions. Interestingly, it was observed that the two paternally contributed centrioles split from one another in one-cell-stage tba-2(RNAi) embryos. Therefore, astral microtubules do not appear to be needed for centriole splitting at the onset of the duplication cycle in C. elegans embryos, as in vertebrate somatic cells. It was noted also that there are only two centrosomes in tba-2(RNAi) embryos, even after several cell cycles. In principle, these two centrosomes could each contain a pair of centrioles if just one round of centriole formation had occurred. However, since centrioles can split from one another in tba-2(RNAi) embryos, four centrosomes, each containing one centriole, would be expected in this scenario. As only two centrosomes are present, it appears instead that completion of daughter centriole formation is impaired and that each centrosome contains only one paternally contributed centriole in tba-2(RNAi) embryos. Therefore, centriole formation does not seem to occur in tba-2(RNAi) embryos. Similarly, centriole formation fails in vertebrate somatic cells treated with high doses of colcemid (Delattre, 2006).

It has been reported that GFP-SAS-5 and GFP-SAS-6 are recruited to centrioles in tba-2(RNAi) embryos. The same is true for GFP-SPD-2, as well as SPD-2, ZYG-1, and GFP-SAS-4. Although the possibility that residual tubulin contributes to the recruitment of these proteins, these results strongly suggest that SPD-2, ZYG-1, SAS-5, SAS-6, and SAS-4 can all be recruited independently of astral microtubules (Delattre, 2006).

This study has provided evidence for an emerging pathway for centriole formation in C. elegans. Together with earlier work, these findings lead a proposed sequence of events. First, SPD-2 is recruited to each mother centriole or to a closely associated structure. SPD-2 is needed for the centriolar recruitment of ZYG-1, which in turn is required for the remaining three proteins to localize to centrioles. SAS-5 and SAS-6 are recruited next and are needed for SAS-4 to be incorporated thereafter. Furthermore, these results suggest that assembly of centriolar microtubules occurs downstream of SAS-4 incorporation or in parallel to the entire pathway. In addition, the PCM components SPD-5 and gamma-tubulin play a partially redundant role in centriole formation, and it will be interesting to investigate their placement in this sequence. Overall, it is concluded that, like other assembly processes, centriole formation can be described as a series of consecutive steps that entails the sequential recruitment of at least five proteins that ensure formation of a daughter centriole next to each mother centriole once per cell cycle (Delattre, 2006).

SPD-2 is unique among the five proteins investigated in being also required for PCM assembly. In embryos depleted of SPD-2, the coiled-coil protein SPD-5 is not recruited to centrosomes, resulting in the absence of other PCM components, including the Aurora kinase AIR-1 and gamma-tubulin. Thus, SPD-2 lies upstream in the pathway for PCM assembly. Similarly, it was found that SPD-2 is the upstream-most component among the five proteins essential for centriole formation. Therefore, SPD-2 plays a pivotal role in coordinating assembly of the two principal constituents of centrosomes. Perhaps SPD-2 acts in a manner analogous to scaffold proteins in signaling networks, which serve to localize and modulate kinases and their substrates. In this scenario, ZYG-1 and its substrates may be brought together by SPD-2 during centriole formation. spd-2 and zyg-1 exhibit a strong genetic interaction, compatible with the two components having a close relationship. Interestingly, it was discovered that SAS-5 and SAS-are needed for the diminution of centriolar ZYG-1 during interphase. Whereas it remains to be determined whether diminution of centriolar ZYG-1 is important, it is tempting to speculate that this serves as a signal ensuring that SAS-5 and SAS-6 have been recruited before further steps can take place (Delattre, 2006).

Plk4-induced centriole biogenesis in human cells

Overexpression of Polo-like kinase 4 (Plk4) in human cells induces centrosome amplification through the simultaneous generation of multiple procentrioles adjoining each parental centriole. This has provided an opportunity for dissecting centriole assembly and characterizing assembly intermediates. Critical components were identified and ordered into an assembly pathway through siRNA and localized through immunoelectron microscopy. Plk4, hSas-6, CPAP, Cep135, gamma-tubulin, and CP110 are required at different stages of procentriole formation and in association with different centriolar structures. Remarkably, hSas-6 associates only transiently with nascent procentrioles, whereas Cep135 and CPAP formed a core structure within the proximal lumen of both parental and nascent centrioles. Finally, CP110 is recruited early and then associates with the growing distal tips, indicating that centrioles elongate through insertion of alpha-/beta-tubulin underneath a CP110 cap. Collectively, these data afford a comprehensive view of the assembly pathway underlying centriole biogenesis in human cells (Kleylein-Sohn, 2007)

Centriole biogenesis in mammalian cells remains poorly understood, but substantial progress has recently been made in invertebrate organisms. In Caenorhabditis elegans, a protein kinase, Zyg-1, and four putative structural proteins, termed SPD-2, Sas-4, Sas-5, and Sas-6, are required for centriole duplication. Moreover, through elegant epistasis experiments and electron tomography the five proteins could be shown to assemble sequentially on nascent procentrioles. Independently, the protein kinase Plk4 (also known as Sak) has been identified as a key regulator of centriole duplication in both Drosophila and human cells. Although the two kinases lack obvious sequence homology, it is plausible that Plk4 represents a functional homolog of C. elegans Zyg-1. When overexpressed in unfertilized eggs of Drosophila, Plk4 (Sak) induces the de novo formation of centrioles, demonstrating that this kinase is able to induce centriole biogenesis even in the absence of pre-existing centrioles. Homologs of nematode Sas-4 and Sas-6 were also required for centriole biogenesis in Drosophila (see Drosophila Sas-4), and a requirement for Sas-6 was demonstrated for human cells, suggesting that fundamental aspects of centriole biogenesis have most likely been conserved during evolution (Kleylein-Sohn, 2007).

Overexpression of Plk4 in human cells causes the recruitment of electron-dense material onto the proximal walls of parental centrioles, suggesting that Plk4 is able to trigger procentriole formation. This study used a cell line allowing the temporally controlled expression of Plk4 to study the formation of centrioles in human cells. Plk4 is shown to trigger the simultaneous formation of multiple procentrioles around each pre-existing centriole. These multiple centrioles form during S phase and persist as flower-like structures throughout G2 and M phase before they disperse in response to disengagement during mitotic exit, giving rise to a typical centriole amplification phenotype. Through siRNA-mediated depletion of individual centrosomal proteins, several gene products important for Plk4-controlled centriole biogenesis have been identified and assigned individual proteins to distinct steps in the assembly pathway. Finally, these functional data have been correlated with morphological analyses using immunoelectron microscopy. Taken together, these results provide a first molecular analysis of centriole formation in human cells (Kleylein-Sohn, 2007).

Search PubMed for articles about Drosophila Sak

Bettencourt-Dias, M. et al. (2005). SAK/PLK4 is required for centriole duplication and flagella development, Curr. Biol. 15: 2199-2207. Medline abstract: 16326102

Delattre, M., Canard, C. and Gonczy, P. (2006). Sequential protein recruitment in C. elegans centriole formation. Curr. Biol. 16(18): 1844-9. Medline abstract: 16979563

Elia, A. E., et al. (2003). The molecular basis for phosphodependent substrate targeting and regulation of Plks by the Polo-box domain. Cell 115(1): 83-95. Medline abstract: 14532005

Habedanck, R., et al. (2005). The Polo kinase Plk4 functions in centriole duplication. Nat. Cell Biol. 7: 1140-1146. Medline abstract: 16244668

Hudson, J. W., et al. (2001). Late mitotic failure in mice lacking Sak, a polo-like kinase. Curr. Biol. 11(6): 441-6. Medline abstract: 11301255

Kleylein-Sohn, J., et al. (2007). Plk4-induced centriole biogenesis in human cells. Dev. Cell 13(2): 190-202. Medline abstract: 17681131

Ko, M. A., et al. (2005). Plk4 haploinsufficiency causes mitotic infidelity and carcinogenesis. Nat. Genet. 37(8): 883-8. Medline abstract: 16025114

Leidel, S., Delattre, M., Cerutti, L., Baumer, K. and Gonczy, P. (2005). SAS-6 defines a protein family required for centrosome duplication in C. elegans and in human cells. Nat. Cell Biol. 7(2): 115-25. Medline abstract: 15665853

Leung, G. C., et al. (2002). The Sak polo-box comprises a structural domain sufficient for mitotic subcellular localization. Nat. Struct. Biol. 9(10): 719-24. Medline abstract: 12352953

Lowery, D. M., Lim, D. and Yaffe, M. B. (2005). Structure and function of Polo-like kinases. Oncogene 24(2): 248-59. Medline abstract: 15640840

Martinez-Campos, M., Basto, R., Baker, J., Kernan, M. and Raff, J. W. (2004). The Drosophila pericentrin-like protein is essential for cilia/flagella function, but appears to be dispensable for mitosis. J. Cell Biol. 165(5): 673-83. Medline abstract: 15184400

O'Connell, K. F., et al. (2001). The C. elegans zyg-1 gene encodes a regulator of centrosome duplication with distinct maternal and paternal roles in the embryo. Cell 105(4): 547-58. Medline abstract: 11371350

Swallow, C. J., et al. (2005). Sak/Plk4 and mitotic fidelity. Oncogene 24(2): 306-12. Medline abstract: 15640847

date revised: 8 August 2008

Home page: The Interactive Fly © 2007 Thomas Brody, Ph.D.