Analysis and characterization of the novo centriole biogenesis in acentriolar drosophila sas-4 -/- cells

CRISTINA MARIA OLIVEIRA FERREIRA “ANALYSIS AND CHARACTERIZATION OF DE NOVO CENTRIOLE BIOGENESIS IN ACENTRIOLAR DROSOPHILA SAS-4 -/- CELLS” Dissertação de Candidatura ao grau de Mestre em Oncologia submetida ao Instituto de Ciências Abel Salazar da Universidade do Porto. Orientador – Dr. Helder Maiato Categoria – Professor Auxiliar Convidado Afiliação – Faculdade de Medicina da Universidade do Porto 2 3 “Life begins at the end of your comfort zone” Anonymous Para os meus Pais, Emília e Carlos À Olga Afonso 4 5 Agradecimentos Por vezes, pequenos elementos mudam o rumo das nossas vidas. No meu caso, esse elemento foi um simples envelope com um conjunto de cartas deixado, em meados de Outubro do ano 2011, na secretaria de um instituto científico localizado na cidade do Porto. O envelope dirigia-se ao investigador Hélder Maiato e não continha mais do que três cartas que forneciam informação sobre uma estudante de mestrado, licenciada em Biologia, e entusiasta em relação a ciência. As probabilidades de ser contactada eram reduzidas, mas nunca foram impedimento para o atrevimento. Dois anos passaram, e dentro das inúmeras aprendizagens, existe uma conclusão: a vida não avança, se a acompanhá-la não existir atrevimento. É inacreditável a quantidade de informação que absorvi e aprendi nestes últimos dois anos. Acredito agora mais que nunca, que a faculdade é simplesmente o ponto de partida e a base para um conhecimento generalizado. No entanto, é no laboratório onde se “talham” os cientistas. Esta tese de mestrado é o resultado de dois intensivos anos de trabalho laboratorial. Mas, acima de todo o trabalho cientifico elaborado, é o reflexo de muita persistência. Por sua vez, a persistência advém de um conjunto de pessoas que proporcionaram o ambiente ideal para que eu fosse capaz de ultrapassar os momentos mais complicados. Todos os elementos do laboratório do Hélder foram, sem excepções, elementos cruciais para o meu sucesso e bem-estar. Ana Pereira, António Pereira, Cristina Madureira, Danica Drpic, Elsa Logarinho, Filipe Sousa, Joana Macedo, Jorge Ferreira, Luísa Ferreira, Margarida Gomes, Marin Barisic, Martina Barisic, Nina Schweizer, Olga Afonso, Zaira Garcia e o próprio, Hélder Maiato. Em especial, agradeço à Cristina Madureira e à Zaira Garcia por todos os ensinamentos relativos à componente de biologia molecular. Obrigado à Ana Pereira por ter, sem dúvida, enriquecido o meu leque literário. Também agradeço aos elementos do laboratório do Cláudio Sunkel pela hospitalidade com que me receberam, especialmente à Tália Figueiredo, Carlos Conde e João Barbosa. 6 Agradeço à Zita Carvalho-Santos pelos valiosos comentários e discussões, principalmente em relação à análise por microscopia electrónica. E agradeço imenso ao Hélder pela oportunidade que me deu de desenvolver este trabalho científico como parte da minha tese de mestrado, mesmo tendo consciência da minha imensa inexperiência e falta de conhecimento. Foi, sem dúvida, o elemento crucial e influenciador das decisões que acabei por tomar ao longo destes últimos dois anos. No entanto, existe uma pessoa que, acima de todas, considero ser a base de toda a minha aprendizagem, orientação, e que estará sempre associada a todos os “passos” deste projecto. Essa pessoa é a Olga Afonso. É de facto louvável a atitude da Olga que, debatendo-se com o seu doutoramento, assumiu tal responsabilidade. É, devido a ela, que considero ter sido possível executar este projecto e, consequentemente, escrever esta tese. Se actualmente sou portadora de “algum” conhecimento científico, é inteiramente devido a ela. Muito obrigado, Olga, por toda a tua dedicação, carinho, paciência, orientação para comigo. Todos os méritos que, eventualmente, possam reconhecer a este trabalho, serão sempre partilhados contigo. Em especial agradeço, também, a – André Levi, Diogo Pedrosa, Gilda Carvalho, Jorge Correia de Castro, Maria Pedro, Patrícia, Rita Rocha, Tiago Ferreira - que sempre me apoiaram e incentivaram ao longo destes anos de amizade. Ao António Dias, Dennis Herrmann, Gabriela Fioreze, Jovin Jacobs, Lorenza Calcaterra e Tomás Cruz por tornarem os meus dias Lisboetas muito melhores. Ao Luís Pedro, quero agradecer todo o carinho, paciência e atenção ao longo destes anos, especialmente nos momentos mais delicados. À minha irmã, Ana Bárbara, que é um motivo de orgulho e admiração todos os dias. Por fim, Aos meus pais por toda a dedicação e afecto ao longo destes vinte e quatro anos. São voçês, sem dúvida, o “motor” da minha vida. Os meus “incentivadores”. Aqueles que, independentemente das circunstâncias, sempre acreditarão em mim. 7 Os elementos essenciais para a minha construção enquanto pessoa. Obrigado Pai e Mãe. 8 9 Abstract Centrioles, the key elements on centrosome biogenesis and function, are replicated during S phase of each cell cycle. Two main mechanisms of centriole assembly have been described: 1) a template or canonical, and 2) a de novo. While the template mechanism has been assigned as the major mode of centriole assembly in somatic cell divisions, the nature of the de novo mechanism of centriole assembly is less well understood, especially in proliferating somatic cells. Moreover, the function of the mother centriole as a template in the centriole duplication process has been recently brought into question (Rodrigues-Martins et al., 2007b). Here, we show that expression of DSas-4- GFP in somatic acentriolar DSas-4 mutant Drosophila cells led to de novo centriole formation. Mitotic DSas-4 rescued cells showed amplification and co-localization of de novo DSas-4-GFP foci with the centriolar and PCM proteins DSas-6, Ana1, D-PLP, Asl and Cnn, respectively. Live cell imaging analysis showed that the newly formed DSas-4- GFP foci nucleate discrete astral microtubule and mitotic spindle assembly seemed to occur through an outside-inside process. Moreover, we rescued the expression levels of the centriole-specific proteins Asl and DSas-6, suggestive of an activation of the centriole biogenesis pathway. Electron microscopy analysis revealed the presence of centriole-like structures as well as clouds of electron-dense material lacking centrioles. This study reinforces the notion that de novo and template mechanisms of centriole assembly may be variations of a common pathway based on the same molecular machinery. Centriole biogenesis is a template-free process in which the mother centriole may function as a platform for regulatory proteins involved in the centriole duplication process whereby offering an ideal environment for centriole duplication. Therefore, control of centriole number, spatial restriction along with a kinetic advantage might be the key factors that the mother centriole offers to the centriole biogenesis process. Key words: centrioles; centrosomes; de novo centriole biogenesis; DSas-4 protein; Pericentriolar material; Drosophila Sas-4 mutant cell line 16 INDEX 17 18 1. INTRODUCTION 23 2. DISSECTING CENTROSOMES AND CENTRIOLES 26 2.1. HISTORY AND INTER-RELATIONSHIP 26 2.2. ROLE IN SPINDLE ASSEMBLY DURING CELL DIVISION 27 2.3. ACENTRIOLAR ORGANISMS AND CELL LINES 29 3. THE CENTROSOME 31 3.1. FROM WHAT IS MADE A CENTROSOME ? 31 3.2. OVERVIEW OF CENTROSOME CYCLE 34 4. HOW TO BUILD A CENTRIOLE ? 37 4.1. CENTRIOLE ARCHITECTURE 37 4.2. CENTRIOLE ASSEMBLY PATHWAY: ACTION OF A CORE ANCESTRAL PROTEIN MODULE 40 5. SAS-4: A MULTIFACETED PROTEIN 45 6. CENTRIOLE BIOGENESIS: DIFFERENT ORIGINS, ONE GOAL 50 6.1. THE ROLE OF THE PARENTAL CENTRIOLE 52 OBJECTIVE 56 2. MATERIALS AND METHODS 60 3. RESULTS 67 PART 1 69 1. CHARACTERIZATION OF THE DROSOPHILA ACENTRIOLAR DSAS-4 -/- CELL LINE # 131 71 1.1. ACENTRIOLAR MITOTIC SPINDLE ASSEMBLY 71 1.2. RECRUITMENT OF PCM AND CENTRIOLAR PROTEINS IN DSAS-4 -/- CELLS 72 PART 2 79 2.1. TRANSFECTION OF DSAS-4 PROTEIN IN THE DSAS-4 -/- CELLS 81 2.1.1. DSAS-4-mRFP CONSTRUCT 81 2.1.2. DSAS-4-GFP CONSTRUCT 84 19 PART 3 89 3. ANALYSIS OF DE NOVO CENTRIOLE FORMATION IN DSAS-4 -/- RESCUED CELLS 91 3.1. RESCUE OF DSAS-4 -/- CELLS THROUGH DSAS-4 PROTEIN REINTRODUCTION 91 3.2. DE NOVO FORMED DSAS-4-GFP FOCI COLOCALIZE WITH CENTRIOLAR PROTEINS 92 3.3. AMPLIFICATION OF DSAS-4-GFP FOCI IN DSAS-4 -/- RESCUED CELLS 94 3.4. RECOVERY OF THE EXPRESSION LEVELS OF CENTRIOLAR PROTEINS IN DSAS-4 -/- 96 RESCUED CELLS 3.5. THE DE NOVO FORMED DSAS-4-GFP FOCI ARE ABLE TO NUCLEATE MICROTUBULES 97 AND TO ACCUMULATE CENTROSOMIN DURING MITOSIS 3.6. DE NOVO CENTRIOLE-LIKE STRUCTURES ARE FORMED IN DSAS-4 -/- RESCUED CELLS 100 4. DISCUSSION 104 DSAS-4 -/- CELLS SHOW ABNORMAL SPINDLE MORPHOLOGY ACENTRIOLAR DSAS-4 -/- CELLS RECRUIT PCM AND CENTRIOLAR COMPONENTS 107 TO THE POLES OF THE MITOTIC SPINDLE ECTOPIC DSAS-4 PROTEIN RESCUES DSAS-4 -/- ACENTRIOLAR PHENOTYPE 110 DE NOVO FORMED DSAS-4-GFP FOCI COLOCALIZE WITH CENTRIOLAR PROTEINS 112 AMPLIFICATION OF DSAS-4-GFP FOCI IN DSAS-4 -/- RESCUED CELLS 114 RECOVERY OF THE EXPRESSION LEVELS OF CENTRIOLAR PROTEINS IN DSAS-4 -/- 115 RESCUED CELLS THE DE NOVO FORMED DSAS-4-GFP FOCI ARE ABLE TO NUCLEATE MICROTUBULES 117 AND TO ACCUMULATE CENTROSOMIN DURIING MITOSIS DE NOVO CENTRIOLE-LIKE STRUCTURES ARE FORMED IN DSAS-4 -/- RESCUED CELLS 120 5. CONCLUSIONS 124 REFERENCES 131 106 20 21 Figure Index Introduction Figure 1 32 Figure 2 38 Figure 3 44 Figure 4 45 Results Figure 5 71 Figure 6 73 Figure 7 74 Figure 8 76 Figure 9 82 Figure 10 86 Figure 11 91 Figure 12 93 Figure 13 95 Figure 14 96 Figure 15 98 Figure 16 100 Figure 17 101 22 23 1. INTRODUCTION 24 25 1. Introduction "Omnis cellula e cellula" was postulated two centuries ago by the German physician Rudolf Virchow and it was the first approach defining that every cell arises from a preexisting parent cell. Back to our century, it is now known that Virchow's definition is undoubtedly correct and that cells need to proliferate to ensure progeny and the continuity of life (Rieder, 2006, pp.439). Cell division is the result of a complex and strictly regulated process composed by distinct stages integrated in the cell cycle. In eukaryotes, the major goal of the cell cycle is to reproduce a daughter cell inheriting the same diploid number of chromosomes from the mother cell in a highly dynamic process called Mitosis. After DNA replication in interphase, sister chromatids of each chromosome have to be segregated to assure the euploidy of the daughter cells during mitosis. Therefore, an important biological machine was developed by the cells in order to organize chromosomes during metaphase and to accurate segregate them in anaphase - the Mitotic Spindle. The mitotic spindle is a symmetrical, dynamic and bipolar structure made up of microtubules (MTs) that interact with chromosomes via their kinetochores. Its assembly is orchestrated by centrosomes which are defined as the major microtubule-organizing centre (MTOC) in animal cells. Apart from its role in cell division, the centrosome is also a very important organelle in cell shape, polarity, motility, signaling, protein trafficking and in cilia/flagella formation (Bettencourt-Dias, 2013; Nigg, 2007; Nigg and Raff, 2009) The advent of electron microscopy revealed the beautiful composition of the centrosome. This organelle is composed by a pair of barrel-shaped bodies called centrioles that are surrounded by an area of dense protein matrix termed the pericentriolar material (PCM) (Avidor-Reiss and Gopalakrishnan, 2013; Azimzadeh and Marshall, 2010; Debec et al., 2010; Nigg, 2007). As centrioles are at the basis of centrosome formation, it demands strict control of centriole number to ensure a correct number of centrosomes, and faithful chromosome segregation in mitosis. Indeed, it has been recognized the direct association between centriole number deregulation, centrosome overexpression and cancer (Nigg and Raff, 2009). Over the past decade, the mechanisms underlying centriole biology have been uncovered at a tremendous speed due to the development of more advanced imaging techniques and its combination with biochemical and cell biological approaches. A deeper knowledge of centriole structure, function and biogenesis is of major importance to a better understanding of centrosome biology and its role in disease. 32 centrosome integrity, function and in the maturation process at the onset of mitosis characterized by an enrichment of numerous PCM components into this organelle which influences its MT-nucleation ability. γ-tubulin is a conserved eukaryotic protein known for its major role in MT nucleation and thereby vital for centrosome function. It is present in a ring-shape structure, the γ-tubulin ring complex (γ-TuRC) that promotes MT polymerization and organization (reviewed by Kollman et al., 2011). Drosophila centrosomin (or its mammalian orthologue CDK5RAP2) is an essential component for the recruitment of many PCM factors, such as γ-tubulin and pericentrin, and promotes the cohesion between centrioles and the PCM network (Buchman et al., 2010; Choi et al., 2010; Fong et al., 2008; Goshima et al., 2007; Lucas and Raff, 2007; Megraw et al., 1999; Vaizel-Ohayon and Schejter, 1999). Its depletion completely prevents centrosome maturation (Dobbelaere et al., 2008; Muller et al., 2010) and it was suggested that its Polo (PLK1 in Figure 1. Centrosome structure. Schematic view of a typical animal centrosome, illustrating the mother and daughter centrioles that formed the centriole pair associated at each centrosome surrounded by a cloud of PCM proteins. Note that, specifically the mature “mother” centriole harbors distal and sub-distal appendages (adapted from Bettencourt-Dias and Glover, 2007). 33 mammalian cells) dependent-phosphorylation during mitosis initiates centrosome maturation in flies (Dobbelaere et al., 2008). Pericentrin and AKAP450 family make part of a group of proteins harboring a centrosomal targeting (PACT) domain and known for their role in docking and recruiting regulatory components involved in MT nucleation (e.g. γ- TuRC) and in PCM integrity (Bettencourt-Dias and Glover, 2007; Lawo et al., 2012; Mennella et al., 2012; Schatten, 2008). Drosophila pericentrin-like protein (D-PLP) is the only PACT domain protein identified in flies. d-plp Drosophila mutants, although viable, are severely uncoordinated and the sperm is nonmotile, suggesting an additional role in cilia and flagella formation (Martinez-Campos et al., 2004). Recently, the combination of 3D structured illumination microscopy (3D-SIM) and STORM (stochastic optical reconstruction microscopy) allowed a deeper look on centrosome architecture revealing a conserved high-order structure within PCM, as opposed to the traditional "amorphous" description. PCM organization is characterized by two overlapped layers: a radial and a concentric, in which proteins either are framed at specific sites in a layered fashion or extend outward from centriole wall. The PCM surface is composed by proteins in a toroidal arrangement linked to the outer wall of mother centriole such as SAS-4, followed by proteins extending out to the PCM periphery like SPD-2 (CEP192 in mammalian cells), Cnn and γ-tubulin. The radial distribution is strongly defined by D-PLP that binds its C-terminal PACT domain to the centriole wall and the N- terminal domain extends outwards in order to form a matrix of D-PLP extended fibrils that function as a scaffold for the recruitment of other PCM components. Interestingly, D-PLP fibrils formed an open gap (150-200 nm) on mother centrioles, coinciding with the position of daughter centriole assembly, which was not noticeable in the early stages of the cell cycle that precede centriole duplication (Fu and Glover, 2012; Lawo et al., 2012; Mennella et al., 2012; Sonnen et al., 2012). Centrioles are at the foundation of centrosomes. Thus, an interaction between the centriole pair and the surrounding PCM was expected to exist. A symbiotic relationship between PCM and centrioles has been reported by several studies. Only the mother centriole within the centriole pair is able to anchor MTs through the sub-distal appendages (Piel et al., 2000) and centrioles define centrosome size by the incorporation of Cnn into the PCM, which seems to be driven by the components Asterless (Asl) and Drosophila Spd-2 (DSpd-2) (Conduit et al., 2010). Furthermore, the levels of centriolar protein SAS-4 were reported to set centrosome size (Kirkham et al., 2003) and SAS-4 itself was found in complexes with some centrosomal proteins, such as Asl, Cnn and D-PLP, suggesting a scaffolding role to tether PCM components within the centrosome (Gopalakrishnan et al., 2011). Also, the interaction SAS-4/Tubulin was shown to control PCM recruitment depending on the tubulin guanine-bound state (discussed in chapter 5) (Gopalakrishnan 34 et al., 2012). In agreement with an influence of centrioles in PCM assembly, the transiently centriole disintegration in HeLa cells upon the microinjection of an antibody directly against glutamylated tubulin (GT335) led to PCM dispersion, suggesting that the presence of centrioles allow the correct segregation and organization of the pericentriolar material (Bobinnec et al., 1998). Alongside, the PCM also exhibits an influence over centrioles. It was demonstrated that overexpression of pericentrin in S-phase arrested CHO cells induced the formation of large PCM clouds containing variable number of centrioles (Loncarek et al., 2008). Moreover, depletion of PCM proteins in the model organism C. elegans resulted in failure of centriole assembly in ~50% of the time, and the centrioles who were formed failed to reach full size (Dammermann et al., 2004). Considering the combined data, there is a strong support to the idea of a symbiotic model in which the interaction between centriole and PCM is important to assemble structural and functional normal centrosomes during cell division. 3.2. Overview of Centrosome Cycle In active proliferating cells, the two functional centrosomes ensure accurate chromosome segregation through the formation of a bipolar mitotic spindle during mitosis. In order to be in harmony with cell cycle progression, the centrosome cycle must be highly coordinated with the DNA cycle. The Centrosome cycle depends on centriole duplication and behavior along the different phases of the cell cycle. The centrosome-centriole cycle can be divided into the following discrete and critical stages: (1) centriole disengagement; (2) centriole duplication and elongation; (3) centrosome maturation and (4) centrosome separation (Nigg, 2007) (Figure 3a). In interphase, a typical G1 centrosome harbors two centrioles that are structurally and functionally different - a mother/mature centriole and a daughter or immature centriole assembled during the previous cell cycle. Centriole duplication starts in late G1/early S transition with procentriole nucleation at right angle to the proximal end of each parental centriole. From S to mitosis, after the establishment of the basic centriolar structure, each procentriole starts to elongate reaching ~80% of its full-length during mitosis and eventually matures after a second cell cycle. A mature centriole differs from the younger one due to the acquisition of distal/sub-distal appendages and PCM enrichment. This final complete version of centriole has the ability to dock to the plasma membrane to promote ciliogenesis and to nucleate astral MTs to build the mitotic spindle. Thus, the assembly of a mature centriole able to promote and work as an independent and functional centrosome is a step-by-step process that needs two consecutive cell cycles (Avidor-Reiss and Gopalakrishnan, 2013; Azimzadeh and Marshall, 2010; Brito et al., 2012). The molecular mechanisms regulating centrosomecentriole cycle are for the most part poorly understood, nevertheless some experimental 35 works have already elucidated some essential regulators acting during the four stages of centrosome cycle. During metaphase, each centrosome defining one of the two spindle poles contains a pair of tightly associated parental-progeny centrioles. In spite of the unknown nature of that linker (S-M linker) (Nigg and Stearns, 2011), it is known that upon exit of M phase (or early G1 phase), the link between the two centrioles is lost in a process termed "centriole disengagement". Interestingly, it was shown in Drosophila that DSas-6-Ana2 complex might be involved in S-M linker functionality since their co-overexpression in spermatocytes formed short tubules (SAStubules) linking the inner region of the daughter centriole to the outer surface of the parental centriole, which is lost in meiosis I, coincident with centriole disengagement (Stevens et al., 2010b). It is now imperative to approach the licensing model based on two distinct rules that centrosome cycle must obey in order to promote a correct centriole/centrosome number over successive cell divisions: (1) centrosomes duplicate once and only once in each cell cycle (cell cycle control) and (2) only one progeny centriole must arise next to each parental centriole (copy number control) (Nigg, 2007). Centriole disengagement has been proposed to be a key event for centriole licensing, rendering both centrioles competent to duplicate in the following S phase. Like chromosomes, centriole separation also requires the activity of separase, a protease responsible for driving sister chromatids separation prior to anaphase, as well as the kinase PLK1 (Tsou and Stearns, 2006; Tsou et al., 2009). Centriole disengagement as a license model for centriole duplication is supported by a series of experimental works. Wong and Stearns (2003) in an elegant cell fusion assay have shown that when G2/S phase cells were fused, there was no centriole duplication, but when the fusion was between G1/G2-phase or G1/S phase cells, the G1 centrosomes replicate. Furthermore, Loncarek et al. (2008), by laser ablating the daughter centriole in S phase arrested cells, thereby artificially mimicking centriole disengagement, showed that resident mother centrioles were able to generate new daughter centrioles. Thus, both studies imply three major conclusions: first, centriole disengagement is a centriole-inherent permissive state to duplication; secondly, there is a centrosomeintrinsic-block to re-duplication which probably manifests itself in the already-duplicated and engaged centrioles (since G2/S-phase fusions did not lead to centrosome duplication, in spite of permissive cytoplasmic conditions in S phase) and third, the centrosome somehow "senses" the presence of an immature centriole, blocking re-duplication events. Coupled to DNA replication, the next stage in this sequence is centriole assembly in S phase in which each parental centriole "seeds" the growth of a new centriole (procentriole) which grows orthogonally in relation to the mother centriole. This process is controlled by a specific set of proteins discussed in detail in chapter 4. Nonetheless, the 36 cell cycle kinase CDK2 seems to be a critical factor to enhance centriole duplication (Hinchcliffe et al., 1999; Matsumoto et al., 1999; Meraldi et al., 1999). Formation of procentrioles coincides with an increase in CDK2 activity in S phase, and in complex either with cyclin A and E is essential to trigger centriole duplication through phosphorylation of nucleophosmin (NPM/B23) and Mps1 (Fisk and Winey, 2001; Okuda et al., 2000). However, the direct role of CDK2 in regulating centriole duplication is still controversial since it was shown that CDK2 activity is not essential for centriole duplication, but instead it seems to speed up procentriole formation (Duensing et al., 2006). In the following cell cycle stages (from G2 phase to mitosis) the two recently formed centrosomes must mature and separate in order to orchestrate the formation of a bipolar mitotic spindle. During late G2 phase, centrosome separation seems to require the disintegration of the linker (G1-G2 tether) that mediates centrosome cohesion by linking the proximal ends of the two parental or previously disengaged centrioles (but not the two centrioles within a pair) (Nigg and Stearns, 2011). This linker is assembled right after centriole disengagement and remains temporarily associated with centrioles throughout interphase to ensure centrosome cohesion and to avoid premature centrosome splitting prior to mitosis. Ultrastructural analysis revealed that rootletin, a conserved component of the ciliary rootlet, and C-Nap1 are key structural elements whose interaction is important for the functionality of this linker. Whereas C-Nap1 localization is restricted to the proximal ends of centrioles, rootletin forms fibers emanating from the proximal ends of both centrioles (Bahe et al., 2005; Mayor et al., 2000; Yang et al., 2006), suggesting that C- Nap1 functions as a "docking site" from where rootletin-based fibers attach and emanate. Upon entry into mitosis, this fibrous linker is disassembled through phosphorylation of C- Nap1 by the mitotic kinase Nek2 allowing the separation of the two independent centrosomes (each containing a pair of centrioles) and consequently mitotic spindle formation (Fry et al., 1998; Helps et al., 2000). Concomitantly with centrosome separation, the younger of the two parental centrioles (originated from the previous cell cycle) acquires distal and sub-distal appendages and enlarges its PCM with the recruitment of γ- TuRCs and other matrix components, thereby reaching full maturity and promoting centrosome maturation. 37 4. How to Build a Centriole ? 4.1. Centriole Architecture Centrioles are microtubule-based, cylindrical and evolutionary conserved eukaryotic cell structures that exhibit a distinct nine-fold symmetric radial array of stabilized microtubules. They are polarized along the proximo-distal axis with the base commonly referred as the proximal end and its tips as distal ends. The typical size for a human mature centriole is approximately 200 x 500 nm, although these measures vary between organisms, as it is the case of C. elegans and Drosophila centrioles that tend to be shorter (Gonczy, 2012; Pelletier et al., 2006). Nonetheless, the signature of centriole architecture is undoubtedly the conservation of the 9-fold symmetry across evolution. Cryo-electron tomography studies have brought new and instructive data concerning centriole/basal body structure (Guichard et al., 2010; Guichard et al., 2013; Li et al., 2012). How does the ninefold symmetry arise to build a highly ordered and complex structure as the centriole?. In Drosophila, unicellular algae, many protozoa and vertebrates, the basic structure of centrioles relies on microtubule blades displayed into a 9-fold radial array (Loncarek and Khodjakov, 2009). This symmetry is established by the cartwheel, which is a structure located in the very proximal end of the centriole and composed of a central hub with 20-25 nm in diameter and ~100 nm in height from which nine spokes radiate outwards (Gonczy, 2012; Guichard et al., 2010) (Figure 2). Each stroke ends in a structure so-called Pinhead which bridges the central hub with centriolar microtubules. The cartwheel is the central piece on centriole organization and the first intermediate of centriole structure to appear during the early events of centriole biogenesis. Interestingly, despite its presence in immature centrioles, the cartwheel disappears from mature centrioles in some organisms (e.g. humans). In most species, including human centrioles, nine sets of microtubule triplets decorate the outer surface of the centriole and are linked to the cartwheel through the stroke pinheads. From the inside out, each triplet comprises an A-microtubule, B-microtubule and a C-microtubule, in which only the A-microtubule is complete being formed by 13 protofilaments. The A-microtubule is oriented toward the center of the centriole and is connected to the pinhead of each stroke, B-microtubules attach to A-microtubules and consequently C-microtubules associate with neighboring B-microtubules. Moreover, A-microtubules can bind to C- microtubules from the previous triplet forming an A-C linker. This microtubule triplet pattern is only characteristic of the proximal-end of centrioles since the distal-end displays double microtubules (A- and B-microtubules) (Guichard et al., 2013; Li et al., 2012). 38 Figure 2. Centriole and Cartwheel architecture. (a,d) The ultrastructure of a resinembedded centriole and the cartwheel purified from human cells and Chlamydomonas reinhardtii. Part (a) shows the side view of a mature human centriole. The proximal and distal ends of the centriole are indicated. The arrow points to distal and subdistal appendages present on the sides of the distal part of the centriole. Part (d) shows a crosssection of the proximal part of a C. reinhardtii centriole. Note the central tube from which nine spokes emanate that radiate towards the vicinity of triplet microtubules. The A-,B- and C-microtubule are indicated and the arrow points to the A-C linker. (e) Schematic representation of a centriole and procentriole pair in a human cell. (f) Schematic representation of the cartwheel viewed from the proximal end (adapted figure from Pierre Gӧnczy, 2012). Interestingly, it was recently demonstrated that the pinhead structure plus the A-C linker are polarized structures that might be responsible for dictating the directionality of centriolar microtubule assembly, as well as the chirality of microtubule triplets (Guichard et al., 2013). An amazing exception to this cartwheel-based centriole structure is represented by C. elegans centrioles. Electron microscopy studies revealed that they do not comprise the cartwheel structure and, instead, follow a more simple architecture formed by a central tube on which nine sets of centriolar microtubules are directly attached (Pelletier et al., 2006). Additionally, the microtubule number can also vary between species. For example, C. elegans centrioles are made only by singlet 39 microtubules whereas Drosophila centrioles in embryos and most tissues are formed by doublet microtubules, and specifically in the spermatocytes of male germ line, centrioles do exhibit centriolar triplets of microtubules (Carvalho-Santos et al., 2010; Debec et al., 2010). In spite of these variations on centriole structure, the nine-fold symmetry is always maintained. A typical interphase G1 cell harbors one centrosome built from a pair of centrioles linked by a flexible connection. Through the following cell cycle stages, a series of events take place including centriole replication, elongation and maturation. A cryo-electron tomography analysis of centrosomes isolated from human lymphoblasts by Guichard and co-workers have revealed new insights about the initial structural events involved in centriole duplication. In the nascent procentriole, centriolar singlets A-microtubules were observed which seemed to be capped by a conical structure at their proximal or minus end resembling the γ-tubulin ring complex (γ-TuRC). This suggests, in association with other studies (Dammermann et al., 2008; Dammermann et al., 2004; Kleylein-Sohn et al., 2007), the involvement of γ-TuRC in nucleating each A-microtubule, which grows unidirectionally from the proximal to the distal (plus) end during centriole assembly. Accordingly, the distal end of A-microtubule is not closed and, instead, it showed outward curved extensions characteristic of growing microtubules. In contrast with the minus end of A-microtubule, B- and C- microtubules showed always open and outward curved extensions at both proximal and distal ends. Therefore, nucleation of these last microtubules (B and C) is not mediated by γ-TuRC, but it follows a template-dependent mechanism, in which A- and B-microtubules are the templates for a bidirectionally growth (from their plus and minus ends) of the B- and C-microtubules. Interestingly, the attachment of A-, B- and C-microtubules for the centriole wall formation occurs independently and without any specific order or position. As B- and C-microtubules reach the minus end of A-microtubules, their proximal ends become blunt and just the distalends continues to grow until completion of the microtubule triplet blades (Guichard et al., 2010). In a mature centriole, A-microtubules lose the closest conformation of the proximal end indicating that γ-TuRC is no longer necessary, being removed when the centriolar microtubule wall is fully developed. The composition of B- and C-microtubules is still unknown, but it seems that ε- and δ-tubulin are potential candidates to take in consideration, since their mutation in Chlamydomonas and Paramecium disrupts the MT triplet arrangement. Nevertheless these two tubulin isoforms are absent from D. melanogaster proteome suggesting an alternative mechanism for B- and C-microtubules nucleation (reviewed in Azimzadeh and Marshall (2010), Brito et al. (2012) and Gonczy et al. (2012)). The answer to this question awaits further experiments. 40 4.2. Centriole-assembly pathway: action of a core ancestral protein module Centrioles are structurally complex organelles, but surprisingly the mechanism orchestrating centriole assembly relies on a few and evolutionary conserved core of proteins. A long-standing question in centriole biology was focused on the players and the way they interacted to initiate and produce a centriole. The first cues concerning the molecular mechanism involved in centriole biogenesis came from studies in C. elegans embryos. Genetic analysis and RNAi-based screens revealed a hierarchical molecular cascade in which only five proteins are specifically required for centriole duplication: the coiled-coil proteins SPD-2, SAS-4, SAS-5, SAS-6 and the kinase ZYG-1 (Dammermann et al., 2004; Delattre et al., 2004; Kemp et al., 2004; Kirkham et al., 2003; Leidel and Gonczy, 2003; O'Connell et al., 2001). During centriole assembly, these proteins are sequentially recruited to centrioles (Delattre et al., 2006; Pelletier et al., 2006), where procentriole nucleation is triggered by the PCM protein SPD-2 that recruits ZYG-1 to the procentriole. The protein kinase ZYG-1 is responsible for the recruitment of the structural proteins SAS-5 and SAS-6, who physically interact (Leidel et al., 2005) to build the central tube and to recruit SAS-4, which induces the assembly of nine singlet microtubules to the outer wall of the emerging centriole. However, given that C. elegans centriole architecture is atypical and divergent compared with Drosophila and mammalian centrioles (Pelletier et al., 2006), it became imperative to address if the same protein module governing centriole duplication in worms was transversal to other organisms. An elegant study conducted by Kleylein-Sohn et al. (2007) unveiled how centriole biogenesis is governed in human cells. Taking advantage of centriole induction by PLK4 overexpression (Habedanck et al., 2005; Rodrigues-Martins et al., 2007b) in association with siRNA mediated depletion of individual centriolar proteins, a similar set of proteins associated with C. elegans centriole assembly were described to control this process in humans cells. Five proteins were identified to be essential for centriole duplication, specifically human SAS-6 (hSAS-6), CPAP (functional homologue of SAS-4/DSas-4 in C. elegans and Drosophila), CEP135, CP110 and γ-tubulin. A putative sequential centriole assembly pathway was described in which Polo-like kinase 4 (PLK4) is an upstream and key regulator essential to trigger procentriole assembly, and does not depend on the presence of any of the other proteins to localize to centrioles (Bettencourt-Dias et al., 2005; Habedanck et al., 2005; Kleylein-Sohn et al., 2007). Although PLK4 is not an homologue of C. elegans ZYG-1, it seems that both proteins work in an analogous way to initiate centriole duplication in different organisms. Thus, in human cells, PLK4 activation on the surface of the parental centriole triggers procentriole assembly and is crucial for the recruitment of hSAS-6 and STIL (functional homologue of C. elegans SAS-5 or Drosophila Ana2) (Vulprecht et al., 2012), which in turn are needed for CPAP loading in the 41 procentriole (reviewed by Gonczy, 2012). CPAP was shown to be a substrate for PLK2 and its phosphorylation (CPAP residues 589-595) is critical for procentriole assembly as well as for its stabilization in the nascent centriole (Chang et al., 2010). Moreover, in agreement with Kleylein-Sohn et al. (2007), in which CEP135 was identified as a core component of centriole duplication, it was recently shown that CEP135 directly interacts with hSAS-6 and CPAP, indicating that these three proteins might regulate centriole biogenesis as a complex. It was also identified in its protein structure a MT-binding domain, elucidating a potential role for CEP135 in mediating CPAP-dependent centriolar microtubule assembly (Lin et al., 2013). Furthermore, it was shown that STIL N- terminal domain interacts and eventually recruits CPAP to the procentriole (Cottee et al., 2013; Vulprecht et al., 2012). Strikingly, as opposed to C. elegans and Drosophila, human STIL does not seem to form a stable complex with hSAS-6. Nevertheless, the siRNA- mediated depletion of STIL significantly decreased hSAS-6 levels at centrosome (Vulprecht et al., 2012), and STIL centriolar localization was also affected upon hSAS-6 depletion (Arquint et al., 2012), suggesting that STIL and SAS-6 might be partially interdependent for their localization at centrioles (Vulprecht et al., 2012). SPD-2 ortholog in humans, called CEP192, seems also to be required for centriole duplication in human cells (Zhu et al., 2008). CEP192 was recently found to cooperate with CEP152 for the centriolar recruitment of PLK4 during centriole duplication, indicating an important and direct role of this PCM component in human centriole biogenesis (Kim et al., 2013; Sonnen et al., 2013). As expected, the sequential model proposed in mammalian cells extends to D. melanogaster. In a genome-wide screen to dissect centriole duplication and centrosome maturation in Drosophila cells (S2R+), from among 119 centrosomal-related genes that were analyzed, only nine genes were identified to be directly involved in centriole assembly. From those nine genes, three coded for the well-known SAS-4, SAS-6 and SAK (Drosophila homologue of PLK4) proteins and another three genes coded for a set of proteins so-called Ana1, 2 and 3 (previously identified by Goshima et al., 2007). As emphasized previously, SAK (or PLK4 in human cells) is a master regulator in centriole assembly, whose depletion in Drosophila impairs centriole duplication and flagella formation (Bettencourt-Dias et al., 2005). Overexpression of Ana1 and Ana2 formed extra centrioles suggesting a potential role in centriole duplication in Drosophila cells (Dobbelaere et al., 2008). Latter, Ana2 was identified either as a conserved centriole duplication factor and as orthologue of STIL/SAS-5 based on weak sequence similarities (Stevens et al., 2010a), as well as Ana3 was described to be important for centriole structural integrity and cohesion, but not for centriole duplication. On the other hand, the precise function of Ana1 on centriole assembly is still not known. However, it has been 48 Accordingly with its PCM localization, the amount of SAS-4 was also shown to dictate centrosome size. Partial depletions of this protein (by varying the time between injection of SAS-4 dsRNA into hermaphrodites worms) led to incomplete centrosomes with less PCM and defects in MT nucleation capacity in C. elegans embryos (Kirkham et al., 2003). However, the last study did not clarify if the asymmetry in centrosome size was accomplished through a direct regulation of SAS-4 or by an indirect action through centriole formation. Gopalakrishnan and co-workers worked on the previous issue and in two recent papers (Gopalakrishnan et al., 2012; Gopalakrishnan et al., 2011) they were able to demystify the influence of SAS-4 on centrosome maturation. First, it was shown that SAS- 4 scaffolds cytoplasmic complexes of centrosomal proteins, including Cnn, D-PLP, βtubulin and Asl (S-CAP complexes) via its PN2-3 domain, and tethers them to a procentriole through its C-terminal domain allowing the establishment of a mature centrosome along the rest of the cell cycle. Using a cell free system in which "stripped centrosomes" 2 were created with high salt concentrations, it was demonstrated that in the presence of a recombinant SAS-4 protein, Asl and Cnn were able to bind to centrosomes stripped of PCM. In opposition, when SAS-4 was removed from the mixture leaving only Asl and Cnn, both proteins were not able to bind alone (Gopalakrishnan et al., 2011). In a second study, they reported and clarified the molecular mechanism by which SAS-4 regulates and tethers the S-CAP complexes into the centrosome. As the association between SAS-4 and tubulin was the first interaction to be reported (Gopalakrishnan et al., 2011; Hung et al., 2004; Hung et al., 2000), they analyzed how far this interaction could influence PCM recruitment. By combining several biochemical and genetic tools with a deep knowledge about the biochemical properties of the tubulin molecule, it was developed an unprecedented model on the SAS-4/tubulin interaction based on the activity of SAS-4 as a tubulin GAP 3 to control the GTP/GDP-bound state of tubulin. Briefly, it was shown that cytoplasmic free tubulin-GTP binds to SAS-4 and, since SAS-4 has GTPase activity, it is able to hydrolyze GTP into GDP. The tubulin-GTP/GDP inter-conversion becomes the molecular switch that regulates the SAS-4-dependent formation of the centrosomal complexes. SAS-4/tubulin-GDP positively promotes the assembly of stable S-CAP in the cytoplasm, whereas the association with tubulin-GTP negatively regulates the interaction between SAS-4 and centrosomal proteins (e.g. Cnn). It is also possible that MT depolymerization can contribute and enhance the formation of S-CAP complexes by releasing tubulin-GDP to the cytoplasm, which directly associates with SAS-4. After being tethered to the centrosome through the C-terminal 2 PCM components were removed. It is composed of centrioles and a salt-stable centrosome matrix. 3 GTPaseactivating proteins. 49 region of SAS-4, the tubulin GTP/GDP-bound status is again changed by the centrosome with the conversion of GDP into GTP, promoting the disassembly and release of SAS-4 and tubulin to the cytoplasm. Interestingly, the interaction SAS-4/tubulin controls either centrosome function and centriole length (Kitagawa et al., 2011a; Schmidt et al., 2009; Tang et al., 2009), but it seems not to have an influence on centrosome number. A potential role for CPAP in centrosome splitting was also described based on its self-interaction through the fifth coiled-coil domain (CC5) to form homodimers. In fact, the transition between monomeric and dimeric CPAP seems to be fundamental for a faithful cell division. In interphase, CPAP homodimerization seems to be essential to negatively regulate centrosome splitting. As cells progress through G2/M phases, centrosomes separate in order to build a bipolar spindle, and the homodimerization of CPAP is abolished by phosphorylation, blocking its self-interaction and promoting centrosome separation. Although it was not identified the kinase responsible for CPAP phosphorylation, neither the molecular mechanism by which centrosome splitting is inhibited, this study shows that CPAP dynamics during the cell cycle must be controlled to guarantee an accurate cell division (Zhao et al., 2010). Deregulations in centriole/centrosome cycle are linked to several human diseases including cancer, obesity and polycystic kidney disease (Nigg and Raff, 2009). Human neurodevelopmental disorder (autosomal recessive primary microcephaly, MCPH) is characterized by small brain size associated with mental retardation, and it has been linked with mutations in several loci coding for centrosomal proteins. Three specific mutations in locus MCPH6, encoding a modified version of CPAP, were shown to be associated with MCPH disorder (Bond et al., 2005). Specifically, two of those mutations encompass a missense mutation (E1235V) within the TCP domain and premature stop codons that result in a protein lacking the TCP domain. By combining siRNA of endogenous CPAP with the expression of GFP-fusions of CPAP-MCPH mutations in human cells (U2OS and HeLa), Kitagawa et al. (2011) demonstrated that impaired centriole biogenesis derived from CPAP mutations led to an increase in asymmetrical spindles associated with incomplete centrosomes presenting few astral MTs. Consequently, the asymmetry of the mitotic spindle led to a randomization and defects in spindle positioning. Since the correct spindle positioning during the first divisions of neuroepithelial progenitor cells is critical for a proper brain development, MCPH mutations, by affecting centriole assembly promote an incorrect spindle positioning orientation and consequently failure to produce the correct number of neurons during brain development (Kitagawa et al., 2011a). 50 6. Centriole Biogenesis: Different origins, one Goal Redundancy has always made part of nature. The discovery that genetic code has multiple codons which code for the same amino acid, allowed us to wonder until what extent this "feature" is used for other processes by the cell. Previously it was described how centrioles are inherited during cell division, in which the assembly of the new centriole during the transition G1/S occurs on a pre-existing centriole that serves as a template. This last centriole assembly mode is called canonical or template mechanism and suggests that the mother centriole somehow acts as a template providing a site or pattern for the assembly of the new centriole. But is the template pathway unique and exclusive for centriole assembly and inheritance?. In fact, this template mechanism has been challenged by peculiar situations, either naturally or experimentally, in which centrioles arise without a pre-existing centriole, or in another words, centrioles are formed de novo. While the templated/canonical pathway is the most common pathway for centriole propagation in somatic cells, the de novo mode of centriole assembly is frequently associated with the development of many organisms, including parthenogenetic species (e.g. Nasonia vitripenins, sea urchin, Spisula), the ameboflagellate Naegleria, gametogenesis in lower plants and fungi and, even in embryogenesis (reviewed in Debec et al., 2010, Marshall et al., 2001). Probably, the most intriguing and surprising case of de novo centriole formation in higher animals was reported during the early events of mouse development. As it was already described, a common feature during oogenesis is centriole loss (e.g. D. melanogaster). Nevertheless, there are cases (e.g. mouse) in which both parental and maternal centrioles are eliminated during gametogenesis. Thus, after fertilization, the very first embryonic divisions of the mouse zygote are acentrosomal, and only during blastomere stage of development (16 to 32 cells) centrioles assemble by the de novo pathway in a proper number. Interestingly, upon the first round of centriole assembly, those de novo formed centrioles duplicate and segregate by the canonical pathway thereafter (Szollosi et al., 1972). Another surprising case of centriole formation happens in differentiated ciliated epithelial cells. In mammals, multiciliated epithelium is found in the airways, the oviduct and the ventricular system of the brain. In order to give rise to hundreds of centrioles (that will ultimately be converted in cilia) per cell, ciliated epithelial cells developed a templatefree mechanism for centriole assembly based on the generation of hundreds of centrioles around a non-microtubule based/non-centriolar structure called deuterosome (Dawe et al., 2007; Vladar and Stearns, 2007). Although the first reports of de novo centriole formation were related to specific events as embryonic development or in differentiated/specialized cells, a series of experiments have proved that somatic cycling cells can also drive centriole biogenesis by 51 the de novo pathway. It is worth noting three studies (Khodjakov et al., 2002; La Terra et al., 2005; Uetake et al., 2007) conducted during the last decade in which the de novo centriole assembly was induced in cycling vertebrate somatic cells in culture, which would normally depend on a templated centriole production. Taking advantage of laser microsurgery to ablate the whole centrosome or only the mother centriole, both studies demonstrated that centrioles arise de novo in cultured transformed and non-transformed mammalian cells (e.g. CHO, HMEC, RPE1 and HeLa), thereby elucidating important features of this process. First, de novo centriole formation in S phase arrested cells is preceded by extensive clouds of typical electron-dense material without centrioles formed after 8-10h upon centrosome ablation; second, centrioles only appear approximately 24h after ablating the centrosome; third, de novo centrioles are able to mature, nucleate microtubules and replicate after completion of a second cell-cycle; fourth, the presence of a single centriole is sufficient to inhibit de novo centriole pathway. At last, centriole assembly is restricted to S phase, since G1-arrested cells for at least 72h did not form de novo centrioles as well as after centriole assembly in S phase there was not an increase in centriole number (centrin-GFP dots) throughout the cell cycle. Interestingly, in all cases of de novo centriole assembly, there was the formation of multiple centrioles (~2-14 per cell) exhibiting morphological intermediate stages, ranging from electron dense amorphous clouds with only two or three microtubule blades, open centriolar cylinders until more morphological normal centrioles. In addition to the laser ablation studies, the de novo centriole formation was also demonstrated in vivo in the unicellular green algae Chlamydomonas, in which vfl2 mutants that did not inherit centrioles due to a mutation on the centrin gene were also able to produce de novo centrioles (Marshall et al., 2001). In summary, these last studies claim the existence of two major centriole assembly modes, a template and a de novo pathway, proving that centrioles can be formed de novo in dividing cells that normally always contain template-formed centrioles and is not limited to specialized cases. Nevertheless, despite of these pioneer studies on centriole assembly, the de novo pathway is less well understood and documented than is the template pathway, and some questions remain to be addressed - "Is the de novo assembly pathway under the same cell cycle control as normal duplication?", "How far these two modes of centriole assembly are fundamentally different or represent variations of a common pathway?". To answer these questions, two recent studies (Peel et al., 2007; Rodrigues-Martins et al., 2007b) started to unveil the molecular mechanism underlying de novo centriole assembly and its relationship with the template or canonical pathway. Peel et al. (2007) have demonstrated that overexpressing GFP-fusions of three centriole assembly specific proteins (PLK4, 52 DSas-6 and DSas-4) in Drosophila unfertilized eggs induced large numbers of de novo centriole-like structures that organized MTs and co-stained either with centriolar and centrosomal markers such as D-PLP, Cnn and γ-tubulin. Accordingly, this last study was complemented by another one conducted by Rodrigues-Martins et al. (2007b) in which overexpression of PLK4/SAK in Drosophila unfertilized eggs and embryos led to amplification of the pre-existing centrioles inherited from the sperm (also demonstrated by Peel et al., 2007) whereas it triggered de novo assembly of multiple centrioles throughout the cytoplasm in unfertilized eggs. Strikingly, in contrast with Peel and co-workers (2007), the de novo formed centrioles after PLK4 overexpression in unfertilized Drosophila oocytes were able to enter new rounds of centriole duplication through the canonical mode. These two reports revealed that proteins typically associated with templated centriole biogenesis also act to trigger de novo centriole assembly, suggesting a unique and common set of regulatory proteins between de novo and templated assembly modes. As alternative sequences between three nucleotides of DNA can code for the same amino acid, the same protein module in a different context leads to the formation of centrioles. Therefore, centriole biogenesis appears as a potential redundant process in which there are two main "roads" based on the same molecular elements that culminate in one goal - centriole assembly. Nonetheless, it is also not possible to rule out the hypothesis of two common pathways that coexist but the template pathway normally suppresses the de novo pathway (Marshall et al., 2001; Nigg, 2007). Moreover, these last studies challenge the idea of the role played by the older centriole as a template on centriole biogenesis. Thus, "If centrioles can be formed de novo, why do new centrioles form next to preexisting ones in most proliferating somatic cells?". This question will be discussed in the following section. 6.1. The Role of the Parental Centriole Centriole duplication follows two simple rules: just one daughter centriole must arise per mother centriole once per cell cycle. Taking into consideration the previous reports, it is clear that de novo centriole assembly leads to a high and variable number of centrioles. Since supernumerary centrosomes are directly involved in multipolarity which in turn has been linked to genome instability, aneuploidy and ultimately cancer (Nigg and Raff, 2009), it is mandatory a strict control over centriole number. Given that de novo centriole assembly seems not to occur when a resident centriole is present, even under overexpression of centriolar proteins such as SAS-6 and PLK4 in embryos and somatic cells (Peel et al., 2007; Rodrigues-Martins et al., 2007b), it might be possible that the centriole itself has the ability to suppress de novo centriole pathway and consequently, control the number of centrioles per cell cycle. Thus, the 53 mother centriole appears as an intermediate that contribute to the precise spatial, temporal and numerical control over centriole duplication during cell cycle. La Terra et al. (2005) proposed a model in which the mother centriole contains a "docking site" at the proximal end that stabilizes and accelerates procentriole formation. In this regard, it might be possible that during centriole duplication in each cell cycle, de novo centriole formation may occur, but only those procentrioles that associate with the mature/mother centriole are stabilized and prone to follow the next stages of centriole assembly. In opposition to the "docking site" theory, the mother centriole might also inhibit de novo centriole formation elsewhere in the cell through the integration of cellular signals. It may be possible that these two last mechanisms are not mutually exclusive and both can be used to control centriole assembly. Although their nature has not yet been truly understood, it might explain why de novo centriole assembly is blocked in centriole-containing cells. Moreover, in all experimental systems previously described, either in Chlamydomonas mutant, unfertilized Drosophila oocytes or the laser ablation of centrosomes (Khodjakov et al., 2002; La Terra et al., 2005; Marshall et al., 2001; Rodrigues-Martins et al., 2007b), it was noticed that the kinetics of de novo centriole assembly was slower compared with the template pathway, which may imply the mother centriole as a platform that recruits and stabilizes centriole assembly promoting factors, hence offering a kinetic advantage to centriole formation by providing a surface or a "docking site" that favours the assembly reaction (Cunha-Ferreira et al., 2009; Nigg and Raff, 2009). Another alternative lies on the interface between PCM and centrioles. Since within PCM resides a variety of regulatory factors important for centriole assembly (Dammermann et al., 2004), it is possible to speculate that the well-defined cloud of PCM surrounding each centriole may enhance procentriole assembly by ensuring a local concentration of essential duplication factors, where protein-protein interactions and activation are promoted (Loncarek et al., 2008; Loncarek and Khodjakov, 2009; Nigg and Raff, 2009). This last approach depends largely on the symbiotic relationship between mature centrioles and their own PCM, whereby the absence of a resident centriole would promote the self-assembly of PCM clouds throughout the cytoplasm to eventually give rise to de novo centriole formation foci. In agreement with this point of view, La Terra et al. (2005) showed that de novo centriole formation in acentrosomal HeLa cells was preceded by the accumulation of small centrin-GFP aggregates in the cytoplasm, some of them disappeared within a few hours after their formation. Thus, it is possible that lack of centrosome or a resident centriole prevents the efficient transport of these small PCM pieces to the centrosome and, therefore prevents the spatial restriction of centrosome 54 assembly to the vicinity of the mother centriole. However, future work is needed to better understand these issues. In summary, the last studies showed that centriole assembly can be a templatefree process, regardless of the cell type. In spite of the last and diverse speculations about the origins and control of de novo centriole assembly, it should be kept in mind that these two pathways of centriole assembly seem to display common molecular requirements, thus they probably are not so different as we may think. Loncarek and Khodjakov in "Ab ovo or de novo? Mechanisms of centriole duplication" (2009) raised pertinent questions such as: "Is canonical centriole duplication a spatially-restricted and more stringently controlled version of centriole de novo assembly?; "If the template centriole duplication provides far more precise control over centriole numbers, then why do so many organisms employ centriole de novo formation, in such critical events as early embryogenesis ?". These questions of centriole biogenesis are begging for more answers and its comprehension might help to understand the dichotomy between extra centrosomes and cancer. 55 56 OBJECTIVE 57 64 Western Blotting and Inducible Expression of DSas-4-GFP For the induction of DSas-4-GFP expression in the stable cell lines, cells were incubated in M3 medium containing 1mM CuSO4 and collected after 24-72 hours of induction. For the parental Jupiter and DSas-4 -/- cells the same procedure was used. Cells were harvested by centrifugation and washed once with PBS. Protein extracts were obtained by resuspending the cell pellet in lysis buffer NP-40 containing 2mM DTT, 1mM PMSF and 1x Proteases Inhibitor at 4ºC for 30 min. Samples were centrifuged at 4ºC and the supernatant was collected for subsequent protein quantification with Bradford assay. Standard procedures were used for Western blotting. Gels differing in polyacrylamide concentration were used to resolve different proteins (DSas-4-GFP [7.5%], Asterless, Cnn, DSas-4 [8%] and DSas-6 [12%]). Primary antibodies were incubated overnight at 4ºC and secondary antibodies during 1-1.30 hours at room temperature in 5-10% milk. The primary antibodies used in western blotting were: rabbit anti-Asterless (1:1000), mouse anti-DSas-4 (1:1000) (all gift from Jordan Raff, University of Oxford, UK), rabbit anti-Cnn (1:2000) (gift from Alain Debec, Institute Jacques Monod, France), chicken anti- DSas-6 (1:500) (gift from Mónica Bettencourt-Dias, Gulbenkian Institute, Portugal), rabbit anti-GFP (1:100), mouse anti-α-Tubulin B512 (1:5000) (Sigma-Aldrich), goat anti-actin sc- 1616 (1:2000) (Santa Cruz Biotechnology). Appropriate secondary antibodies were used for detection using ECL (Pierce). Live cell imaging Drosophila DSas-4 -/- rescued cells were induced for 48h at 1mM CuSO4 and plated the day before live imaging in Ibidi glass bottom dishes (Ibidi) pre-coated with 0.25 mg/ml concanavalin A. Live-cell imaging was performed in a temperature-controlled Nikon TE2000 microscope equipped at the camera port with a modified Yokogawa CSU-X1 spinning-disc head (Solamere Technology), an ASI FW-1000 filter wheel followed by the Andor iXon+ DU-897E EM-CCD. The excitation optics is composed of two Coherent Sapphire lasers at 488nm and shuttering is performed by a Gooche&Housego R64040-150 acousto-optic tunable filter. Sample position is controlled via a SCAN-IM Marzhauser stage and a Physik Instrumente 541.ZSL piezo. A 100x Plan-Apo DIC CFI Nikon objective were used in all experiments, with a 2min time-lapse and 1 μm step image stacks. 65 Electron Microscopy EM was performed according to Lecland et al. (2013). 66 67 3. RESULTS 68 69 PART 1 70 71 PART 1 1. Characterization of the Drosophila Acentriolar DSas-4 -/- cell line # 131 1.1. Acentriolar Mitotic Spindle Assembly The acentriolar nature of DSas-4 -/- cells comes from the insertion of a P-element that truncates the DSas-4 protein, therefore impairing centriole biogenesis (Basto et al., 2006). This cell line was derived from dissociated DSas-4S2214 embryos (aged from 3 to 14 hours), also bearing the Jupiter-GFP protein in the mutant allele (Lecland et al., 2013). In order to study the overall spindle morphology of DSas-4 -/- acentriolar mitotic cells, an immunofluorescence protocol was used and cells were stained with α-tubulin and DAPI, for visualization of the mitotic spindle and chromosomes, respectively. Although, the DSas-4 -/- mutant cell line constitutively expresses the fusion protein Jupiter-GFP that marks the MT array of the mitotic spindle, the α-tubulin staining was required to improve the visualization of the microtubules. Importantly, for all the experiments the Drosophila Jupiter cell line containing centrosomes was used as control (Karpova et al., 2006). During mitosis, five types of acentriolar mitotic spindles were reported in a previous work characterizing this acentriolar cell line (Lecland et al., 2013) (Fig. 5, H). Accordingly, the acentriolar nature of the DSas-4 -/- cells can lead to wild-type like, barrel, dual, hyperpolarized or non-polarized mitotic spindle morphological types (Fig. 5, A-F). Specifically in this study, one more anastral spindle category (Undetermined) was included representing the cells that did not fit in any of the other categories. Figure 5. Characterization of spindle morphology in control and DSas-4-/- mutant cells. (A-F) Immunofluorescence on fixed cells. (A) Control cell exhibiting a polarized biconical mitotic spindle with astral MTs around the centrosome in both poles. (B-F) Representative acentriolar spindle 72 shapes identified in DSas4 -/- mutant cells. These ranged from more focused to non-polarized spindles. Scale bar, 5 μm. (G) Percentage of mitotic cells per category of spindle morphology in control cells: WT (85%) and Undetermined (15%); and in the acentriolar cell line: Barrel (24%); Dual (16.6%); Hyperpolarized (3.6%); WT-like (45%); Non-polarized (3.2%) and Undetermined (7.3%). A total number of control cells (n=85) from two independent experiments and DSas-4 -/- cells (n=438) from three independent experiments were analyzed. Error bars represent SD. (H) Categories of spindle morphology in control and acentriolar mitotic cells according Lecland et al., 2013. In the control cell line approximately 85% of cells exhibited the usual bipolar configuration with astral MTs surrounding each centrosome characteristic of a typical centrosomal mitotic spindle, compared with just 15% of cells that had an undetermined spindle shape, probably reflecting incomplete spindle assembly (Fig. 1, G). On the contrary and as mentioned before, the acentriolar DSas-4 -/- cells showed a variety of non-centrosomal spindle shapes. The barrel-shaped spindle is usually poorly polarized, with broad poles whereas the WT-like spindle exhibits a normal polarization. The dual spindle shows an asymmetry between poles, in which only one of the poles is focused. The hyperpolarized spindle shows an extreme polarization and the non-polarized spindle exhibits no focalization of the spindle extremities (Fig. 5, B-F). In this study, the predominant spindle morphology was the WT-like, which was present in approximately 50% of the mitotic cells, followed by the barrel (24%) and the dual (16%) shapes. The less frequent were the hyperpolarized and the non-polarized with 4% and 3% of cells respectively, and just 7% of cells had an undetermined spindle shape (Fig. 1, G). In all morphological spindle types no centrosome-like structures or astral MTs were detected at the poles. Despite undetectable astral MTs, these cells showed robust K fibers and interpolar MTs (Fig. 5, B-F). 1.2. Recruitment of PCM and Centriolar Proteins in DSas-4 -/- cells In order to analyze the localization of PCM and centriolar proteins in control and DSas4 -/- mutant cells, centrosomin and D-PLP antibodies were used. In control cells, D- PLP is present at the center of the centrosome and Cnn localizes to the centrosomes of metaphase cells (Fig. 6, upper panel). In DSas-4 -/- cells both proteins can be detected as disperse spots of variable size at the broad poles of the acentriolar mitotic spindles (Fig. 6, lower panel). In order to analyze the quantity of Cnn and D-PLP pool recruited to the acentriolar poles of DSas-4 -/- cells in comparison to control cells, fluorescence intensity 73 was measured for each protein, and co-localizations between both proteins was taken into account to avoid the measurement of unspecific dots. The mean fluorescence pixel intensity of Cnn and D-PLP at the centrosomes of control cells was 2.39 ± 1.21 and 1.47 ± 1.07 respectively, whereas DSas-4 -/- cells revealed a reduction in the fluorescence intensity of approximately 90 % for D-PLP and 70 % for Cnn (Fig. 6, B and C). Figure 6. Cnn and Drosophila Pericentrin-like (D-PLP) accumulate at mitotic spindle poles of acentriolar DSas-4 -/- cells. (A) Immunofluorescence on fixed cells. Control (upper panel) and DSas-4 -/- cells (lower panel) were stained for the centriolar and PCM markers D-PLP (red), Cnn (white), DNA (blue) and α-tubulin (green). The yellow arrowheads in the lower panel highlight D- PLP and Cnn co-localizations in DSas-4 -/- line. Scale bar, 10 μm. (B) D-PLP absolute fluorescence intensity levels with an average of 1.47 in control cells and 0.15 in DSas-4 -/- cells. (C) Cnn absolute fluorescence intensity levels with an average of 2.39 in control cells and 0.78 in DSas-4 -/- cells. Note that only the fluorescence intensity of Cnn and D-PLP co-localizations were measured and counted. Each dot in control cells represents one centrosome and in DSas-4 -/- cells represents foci of CNN and D-PLP accumulations. Cells with extra centrosomes in control line were not analyzed. Error bars represent SD and mean. Statistically significant differences (Student's t-test) from the respective controls are shown (p<0.0001); a total number of control (n=50) and DSas-4 -/- (n=66) mitotic cells were analyzed from one experiment. Since it was verified the presence of Cnn and D-PLP proteins at the poles of acentriolar mitotic spindles, we then investigated the localization of other centriolar components. The first analysis was performed with Asterless (Asl), an essential protein for centriole formation in Drosophila and also required for recruitment of PCM (Dzhindzhev et 80 81 Part 2 2. Transfection of DSas-4 protein in the DSas-4 -/- cells 2.1. Optimization of DSas-4 protein expression 2.1.1. DSas-4-mRFP construct The main goal of this study was to assess de novo centriole formation in somatic acentriolar DSas-4 -/- cells by SAS-4 protein reintroduction in an attempt to better understand the dichotomy between canonical/template and de novo centriole formation in dividing cells. In order to rescue the acentriolar phenotype of the mutant cell line, the first goal of this study was to successfully transfect the Drosophila SAS-4 protein in DSas-4 -/- cells. With this goal in mind, stable cell lines were generated carrying a full-length DSas-4- mRFP fusion protein under the control of the metallothionein promoter. The expression of DSas-4-mRFP protein was first examined and optimized with two approaches: 1) by transiently transfecting the DSas4-mRFP construct in Drosophila S2-U cells and 2) by stably transfecting the DSas4-mRFP construct in the control cell line previously used for comparison with the DSas4 mutant cell line. The analysis of the positively transfected cells was done by immunofluorescence on fixed cells. Since the Drosophila S2-U cells are easily transfected, they were firstly used in order to verify the functionality of the plasmid. After the transient transfection and induction of the DSas4-mRFP construct during 24h and 48h with the standard copper concentration used for Drosophila S2-U cells (0.5 mM), it was possible to detect the fusion DSas-4-mRFP signal at centrioles as two bright dots in a high proportion of S2-U cells (Fig. 9, E). Then, the conditions used for Drosophila S2-U cells were also applied to the control cell line used in this study. However the direct detection of DSas-4-mRFP signal at centrioles in the transfected control cell line was not successful (data not shown). Therefore, a protein expression optimization protocol was elaborated, and several conditions taking into account the induction time and copper concentration were set out (Fig. 9, A and B). A polyclonal antibody raised against mRFP (monomeric red fluorescent protein) was used in order to improve the detection of the exogenous DSas-4-mRFP fusion protein. The experimental scheme used for DSas-4-mRFP expression optimization is shown below (Fig. 9, B). As demonstrated in figure 9C and 9D, independently of the copper concentration (0.5 or 1 mM), the induction time (48, 72, 96 h) or the presence or not of the anti-mRFP antibody, the DSas-4-mRFP signal was not detected at centrioles in the majority of the control cells in the stable cell line through all stages of the cell cycle. Yet, in a small proportion of cells a faint DSas-4-mRFP signal was present at centrioles, as confirmed by co-localization with γ-tubulin 23C (Fig. 9, C and D, red arrowheads). 82 Nevertheless, either due to the unsuccessful detection of DSas-4-mRFP protein in majority of stable transfected control Jupiter cells or due to its faint presence, the created stable cell lines expressing DSas4-mRFP were not used for the purpose of this study. Interestingly, the expression of a full-length recombinant DSas-4-mCherry protein in control cells (Jupiter) also resulted in a failure to detect DSas-4 at centrosomes through fixed cell analysis similar to the DSas-4-mRFP construct (data not shown). 83 Figure 9. mRFP fusion of DSas-4 protein was not detected at centrioles in Jupiter control stable cell line. (A) and (B) Schematic outlines of the experimental strategy used for DSas-4- mRFP optimization. (A) Optimization conditions. (1) (+) and (-): with or without anti-mRFP antibody. (2) Copper sulfate (CuSO4) concentrations and timepoints. Note that each cooper concentration was tested individually in all timepoints. (B) Experimental scheme representing the four experimental stages executed for the analysis of the DSas-4-mRFP expression by immunofluorescence. 1. Cells were plated and induced with 0.5 or 1 mM of CuSO4 to allow the expression of the transfected version of DSas-4 protein; 2. At each time point cells were collected and fixed; 3. Cells were processed for immunofluorescence analysis and the detection of exogenous DSas-4 protein was examined, using or not the anti-mRFP antibody, co-stained with γtubulin 23C. (C) Immunofluorescence analysis of Jupiter control cells with anti-mRFP antibody to detect DSas-4-mRFP fusion. Not all the conditions (according with the optimization plan) are shown (D) Immunofluorescence analysis of Jupiter control cells with direct detection of DSas-4- mRFP fusion protein. In (C) and (D) the red arrowheads point to a potential DSas-4-mRFP signal at centrioles (E) Immunofluorescence analysis of Drosophila S2-U cells transiently expressing the fusion protein DSas-4-mRFP (24h, 0.5 mM). Note that in this specific case the transfection signal 84 was detected without an anti-mRFP antibody. The yellow arrowheads highlight the DSas-4-mRFP signal at the centrosomes of an interphasic cell co-localizing with the centrosomal marker γ-tubulin 23C. In all experiments cells were stained for α-tubulin (green), DNA (blue), DSas-4-mRFP or antimRFP ab (red) and γ-tubulin (white). Scale bars, 5 μm 2.1.2. DSas-4-GFP construct Given that it was not possible to detect the fusion Drosophila Sas-4 protein signal in the control transfected stable cell line using the previous plasmid, a GFP-tagged protein version was examined. Therefore, a control Jupiter stable line carrying full-length DSas-4- GFP fusion protein under the control of the metallothionein promoter was generated. In order to optimize the expression conditions of the exogenous DSas-4-GFP protein, similar optimization parameters used for the detection of mRFP-tagged DSas-4 protein were tested. In this specific case, only three timepoints (24, 48 and 72h) for each copper concentration (0.5 and 1 mM) were examined. In order to evaluate the expression levels of DSas-4-GFP for each copper concentration and induction time, a comparative protein expression analysis by western blotting using an antibody raised against GFP (green fluorescent protein) was performed. It should be noted that, in spite of the presence of the endogenous DSas-4 protein in control cells, the anti-GFP antibody only detects the exogenous DSas-4 protein. As shown in figure 10A (right column), there was no presence of DSas-4-GFP protein expression in control and DSas-4 -/- parental lines. The absence of the GFP-tagged version of DSas-4 protein in the previous cells was completely expected since both cell lines were not transfected. In contrast, addition of 0.5 or 1 mM of CuSO4 to the medium, induced the expression of DSas-4-GFP fusion protein in the control cells, thereby confirming the successful establishment of a transfected control cell line stably expressing the exogenous DSas-4-GFP protein. A qualitative analysis comparing DSas-4-GFP expression levels between each condition revealed that DSas-4-GFP expression increased from 0.5 mM to 1 mM of CuSO4, suggesting that increased copper concentration influences the quantity of exogenous DSas-4-GFP protein being expressed. On the other side, the addition of 0.5 mM of copper did not alter significantly the DSas-4- GFP expression levels between the different time points (Fig. 10, A, middle column). In fact, DSas-4-GFP protein expression was higher at 1 mM of copper, reaching its maximum expression at 72h of induction (Fig. 10, A, left-hand column). Importantly, the western blot analysis also showed that control transfected cells that were not induced expressed basal levels of DSas-4-GFP protein (Fig. 10, A, without induction). Yet, the variability intrinsic to each experiment should be taken into account as repetition of the western blot in the previous conditions have also revealed that the basal DSas-4-GFP 85 pool from non-induced control transfected cells might not be present, as well as the expression levels of exogenous DSas-4-GFP protein can vary slightly between timepoints (but not cooper concentration) from different cell extracts (data not shown). In spite of the variability of the protein expression associated with each timepoint, the overall DSas-4- GFP expression levels were increased at 1 mM of copper concentration rather than at 0.5 mM in all western blot experiments. After confirming the expression of the DSas4-GFP protein in the control stable cell line the following parameters were taken into account to determine the proper conditions for analysis and visualization of DSas-4-GFP: (1) population doubling time from Jupiter and DSas-4 -/- cell lines, (2) western blotting analysis of DSas-4-GFP expression and (3) the presence of Jupiter-GFP protein that marks for the MT array in both cell lines. Relative to the first parameter, it was already shown by laser ablation experiments that de novo centriole formation in S-phase arrested vertebrate cycling cells (e.g. CHO) takes approximately 24h after centrosome ablation (Khodjakov et al., 2002), which is greater than the duration of a complete cell cycle (~18h for CHO cell line), suggesting that the kinetics for de novo centriole assembly in S-phase is slower than the normal cell cycle through the template mechanism. Thus, in order to avoid an analysis in which the timing for de novo centriole production would not be sufficient, the doubling population time was determined for control and acentriolar DSas-4 -/- cells (~36h for both cell lines, see material and methods). Regarding the expression profile of DSas4-GFP through western blot analysis the highest protein expression was observed with 1mM of CuSO4. Finally, by immunofluorescence analysis it was possible to visualize the direct DSas4-GFP signal over the Jupiter-GFP, constitutively expressed in this cell line (Fig. 10, B and C). It was previously shown that heterologous protein expression in Drosophila S2 cells increase progressively from lower (e.g. 24h) to high periods of induction (e.g. 72h) and fluorescence intensity is influenced by CuSO4 concentration (Santos et al., 2007). In agreement, recombinant DSas-4-GFP signal was also detected in the alternative optimization conditions, such as 24h and 48h at 0.5 mM of copper concentration, but in the majority of cells the fluorescence intensity of DSas-4-GFP signal was too faint to be reliable for future experiments (data not shown). 86 Figure 10. GFP fusion of DSas-4 protein is expressed and localizes to centrioles in stably transfected control Jupiter cells. (A) Immunoblotting of DSas-4-GFP expression levels according with CuSO4 concentration and induction time in asynchronously control and DSas-4 -/- nontransfected cell lines, as well as control transfected stable cells. Lysates were prepared at the indicated time points and were analyzed using an anti-GFP antibody. α-tubulin was used as a loading control. DSas-4-GFP molecular weight: 130 KDa. (B) Fixed cell analysis in control Jupiter cells through direct detection of DSas-4-GFP fusion protein at the optimized parameters (48h, 1mM CuSO4). The yellow arrowheads point to DSas-4-GFP signal at centrioles of a metaphase cell, which colocalizes with the centrosomal markers, γ-tubulin and D-PLP. Note that Jupiter-GFP is staining the mitotic spindle. Cells were stained for DNA (blue), γ-tubulin (red) and D-PLP (white). (C) Non-transfected control cell showing Jupiter-GFP signal staining to the MT array. Centrosomes were stained with Asl (red) and γ-tubulin (white). Scale bars in (B) and (C), 10 and 5 μm. Taking all these factors into account, it was established that the condition to be used during the entire study for DSas-4-GFP visualization and analysis would be 48h with induction of 1mM of CuSO4. These parameters allow the accomplishment of at least a first 87 cell cycle, as well as to yield enough DSas-4-GFP protein to guarantee its detection at centrioles over the strong Jupiter-GFP signal along the MT array. Additionally, it was confirmed that using a higher concentration of CuSO4 for the induction of the DSas-4-GFP protein did not affect cell viability neither disrupted DSas-4 localization, since it was possible to distinguish the direct DSas-4-GFP signal localizing to the centrioles throughout cell cycle (Fig. 10, compare B and C). 88 89 PART 3 96 colocalize with centriolar markers were not analyzed. A total number of control (n=90) and DSas-4 -/- R (n=48) cells were analyzed from the stainings showed in figure 12. 3.4. Recovery of the expression levels of centriolar proteins in DSas-4 -/- rescued cells To investigate a possible rescue in expression levels of specific centriolar and centrosomal proteins, a western blot analysis with DSas-4 -/- R cell extracts was performed using centriole and PCM-specific antibodies such as, Asl, DSas-6 and Cnn. In fact, it was verified a rescue, in the DSas-4 -/- R cell line, of the expression levels of the centriolar proteins DSas-6 and Asl, but not of the PCM component Cnn (Fig. 14). Figure 14A shows that Asl protein expression levels dramatically increased, after rescue of DSas-4, comparing with the protein levels in the mutant DSas-4 -/- cell line, which were completely absent. Interestingly, a qualitative analysis suggests an increase in Asl expression levels in DSas-4 -/- R cells similar to control cells. Furthermore, there was also an increase in the expression levels of the centriole structural protein DSas-6 in DSas-4 -/- R cells, although this increase was not as dramatic as with Asl (Fig. 14B). In opposition, a positive rescue in protein levels was not verified for the PCM protein Cnn, since the expression levels of this protein seemed not to change dramatically from the rescued DSas-4 -/- cell line to the mutant DSas-4 -/- cells (Fig. 14C). 97 Figure 14. Rescue of DSas-4 -/- mutant cells led to the recovery of centriolar protein levels, but not of the PCM protein Centrosomin. (A), (B) and (C) Immunoblotting of protein expression levels of Asterless, DSas-6 and Cnn, respectively, in asynchronous Jupiter control non-transfected, DSas-4 -/- non-transfected and DSas-4 -/- rescued stable cell lines. Lysates were prepared at the optimized parameters (1 mM CuSO4, 48 h), and were analyzed using an anti-Asl, DSas-6 and Cnn antibodies. α-tubulin (55 kDa) was used as a loading control for Asl and Cnn immunoblottings and actin(45 kDa) for DSas-6 immunobloting. 3.5. The de novo formed DSas-4-GFP foci are able to nucleate microtubules and to accumulate Centrosomin during mitosis Centrioles organize the two opposite poles of the mitotic spindle through the production of astral microtubules during the first stages of mitosis (Meunier et al. 2012). Given the evidence for a specificity of de novo DSas-4-GFP foci for centriolar markers (see part 3, 3.2.), it was addressed if microtubule organizing capacity was restored in DSas-4 -/- R cells, and therefore, if DSas-4-GFP foci could potentially nucleate microtubules during mitosis. Immunofluorescence analysis using α-tubulin to stain microtubules and the PCM protein Cnn suggests that DSas-4-GFP reintroduction led to the origin of microtubule fibers emanating from DSas-4-GFP foci in DSas-4 -/- R mitotic cells, although to a less extent than control cells (Fig. 15A, compare control and DSas-4 -/- R). The size and number of the produced microtubules was variable, as well as they were usually more discrete and in less number compared with the astral microtubules of control cells (Fig. 15A, see magnifications). Moreover, Cnn colocalized with DSas-4-GFP foci and its recruitment and localization was more restricted to a single dot at both poles in DSas-4 -/- R cells, comparing with the presence of multiple Cnn aggregates in the non-transfected DSas-4 -/- cell line (see part 1, 1.2.) (Fig. 15, A, lower panel). Given the previous evidence for MT nucleation capacity of DSas-4-GFP foci, it was investigated the relationship between the newly acquired microtubule capacity and mitotic spindle morphology in DSas-4 -/- R cells. Figure 15B shows that 83 % of control cells exhibit a polarized bipolar mitotic spindle (Fig. 15, A, left panel), and only 17 % present an undetermined spindle shape. In contrast, 50 % and 21 % of DSas-4 -/- R cells exhibited a WT-like and dual mitotic spindle, followed by 13 %, 2% and 15 % that showed a barrel, non-polarized and undetermined mitotic spindle morphology, respectively. Although the majority of rescued DSas-4 -/- cells exhibited a WT-like mitotic spindle, the overall 98 percentage of DSas-4 -/- R cells in each spindle morphology category did not significantly change in comparison with non-transfected DSas-4 -/- cells (compare Fig. 15B and Fig. 5B from part 1, 1.1.), in spite of the presence of astral microtubules in DSas-4 -/- R cell line. 99 Figure 15. de novo formed DSas-4-GFP foci are able to nucleate microtubules. (A) Immunofluorescence on fixed cells. Control and DSas-4 -/- R cells were stained for DNA (blue), αtubulin (green) and Cnn (white). Note that DSas-4-GFP signal was directly detected. Magnifications highlight astral microtubule fibers emanating from DSas-4-GFP foci at the mitotic poles. Scale bar, 5 μm. Higher magnifications 3x. (B) Percentage of mitotic cells per category of spindle morphology in control cells: WT (83%) and Undetermined (17%); DSas-4 -/- R cells: Barrel (13%); Dual (21%); WT-like (50%); Hyperpolarized (0%); Non-polarized (2%); Undetermined (15%). A total number of control (n=89) and DSas-4 -/- R (n=48) cells were analyzed from the stainings showed in figure 12. In order to address in living cells the functionality of the reported DSas-4-GFP foci, live cell imaging was performed. The analysis showed that DSas-4 -/- R cells were able to produce de novo DSas-4-GFP foci in interphase, from which discrete astral microtubules were generated at the beginning of mitosis (Fig. 16, A and B). Interestingly, and in agreement with the fixed cell analysis, the live imaging data suggests that some DSas-4 - /- R cells were able to produce a more focused spindle (Fig. 16, A) whereas, in opposition to control cells, the mitotic spindle morphology in other DSas-4 -/- R dividing cells was non-focused, in spite of the microtubule generation capacity from the DSas-4-GFP foci (Fig. 16, B). Moreover, DSas-4 -/- R cells also showed more than two DSas-4-GFP foci in interphase and, consequently, the mitotic cells were multipolar (Fig. 16, B). Surprisingly, the duration of mitosis was different between cells generating different spindle morphologies. A qualitative analysis suggests that cells with non-focused spindles were slower in producing the mitotic spindle whereas DSas-4 -/- R cells with a more focused spindle completed mitosis in half of the time (compare fig. 16, A and B). 100 Nevertheless, the live cell imaging analysis to DSas-4 -/- R cells should be repeated for more conclusive results. Figure 16. Live DSas-4 -/- rescued cells are able to generate astral MTs and produce a mitotic spindle from DSas-4-GFP foci. (A) and (B) live imaging of mitotic DSas-4 -/- cells expressing DSas-4-GFP and Jupiter-GFP. Cells seemed to progressively establish the mitotic spindle from discrete MTs bundles nucleated by DSas-4-GFP foci. The formed spindles might be nearly focused (A) or presenting a non-focused morphology (B). From left to right: interphase, prophase, metaphase and cytokinesis. The optimized parameters (1 mM CuSO4, 48 h) were used for the induction of DSas-4-GFP protein expression. Scale bar, 5 μm. Time is in minutes. A total of six movies were analyzed. 3.6. de novo centriole-like structures are formed in DSas-4 -/- rescued cells The results so far suggested that expression of DSas-4-GFP, in the mutant background, is able to induce the formation of foci, that even though are able to recruit other centriolar and centrosomal proteins, don’t seem totally functional because spindle morphology is not fully recovered when compared to the DSas-4 mutant cells. To better understand the ultra structure of these Dsas-4-GFP foci, an electron microscopy analysis was performed (collaboration with Alain Debec, IJM, France). The data revealed that centriole-like structures were formed de novo in DSas-4 -/- cells expressing DSas-4-GFP recombinant protein. A qualitative analysis showed that some of these structures resembled morphologically recognizable centrioles with the presence of the inner central hub and the outer microtubule wall (Fig. 17, square boxes in A and B). A B 101 Strikingly, in one cell it was possible to observe a centriole-like structure surrounded by microtubule fibers (Fig. 17A, arrows). The EM analysis also revealed the presence of electron-dense material clouds that did not contain centriole-like structures (Fig. 17D). Nevertheless, electron microscopy analysis should be repeated in order to confirm and better characterize the de novo formed centriole-like structures in DSas-4- -/- R cells. Control A B DSas-4 -/- R C D 102 Figure 17. Centriole-like structures are formed in DSas-4 -/- R cells. Control non-transfected mitotic cell and DSas-4 -/- R cells. White arrow in control cell represents a canonical centrosome. (A) and (B) are examples of de novo formed centrioles that seem to be structurally normal in DSas-4 -/- R cells (square box). Arrows represent microtubule fibers that seem to emanate from the centriole-like structure. (C) and (D) arrowheads represent the presence of electron-dense material clouds without centriole-like structures. The optimized parameters (1 mM CuSO4, 48 h) were used for the induction of DSas-4-GFP protein in DSas-4 -/- R cells. Lateral images are higher magnifications of B and D, 4x. Scale bar, (B), (C) and (D), 500 nm, (A), 200nm. 103 104 4. DISCUSSION 105 112 The monomeric red fluorescent mCherry protein is highly photostable and thus more resistant to photobleaching in comparison with mRFP protein (Shaner et al., 2005). However, the establishment of stable transfected cell lines (Jupiter and DSas-4 -/-) using a single plasmid carrying the fusion protein DSas-4-mCherry as well as the resistance selection marker Blasticidin S also revealed similar results in terms of protein detection. The results from the protein expression optimization carried out for the DSas-4- mRFP fusion protein in control cells are in stark contrast with the successful direct detection of DSas-4-mRFP in S2 cells. The last case suggests that, although no plasmid sequence or western blotting analysis was performed in this study, the unsuccessful detection of DSas-4-mRFP (or mCherry) signals at centrioles of the Jupiter control cell line was probably not due to a defect in the nucleotide sequence of the plasmid or to a delocalization from centrioles of the recombinant protein, and it might be more related with the differences between cell lines in response to heterologous protein expression or due to protein folding and stability problems. Other alternative protocols that might have been used to more efficiently establish transfected stable cell lines in this study could have been the selection of transfected cells trough FACS (Fluorescence activated cell sorting) or cloning of a single transfected cell in order to originate a stable transfected cell line derived from a single cell, in which there is a more homogeneous recombinant protein expression and weeds out the possibility of growing drug-resistant cells. de novo formed DSas-4-GFP foci colocalize with centriolar proteins The successful establishment of a DSas-4 -/- cell line stably expressing DSas-4- GFP recombinant protein throughout the cell cycle allowed us to address the specificity of DSas-4-GFP foci using centriolar markers. In this study, the evidence of colocalization in the majority of DSas-4 -/- R cells between DSas-4-GFP foci and D-PLP, DSas-6, Asl and Ana1 at the mitotic poles suggests that DSas-4-GFP foci may not be just non-functional protein aggregates (Peel et al., 2007; Tang et al., 2009) due to the induction of DSas-4- GFP expression using the metallothionin promoter (pMT), (which does not allow a controlled protein expression) but rather, DSas-4-GFP foci may represent de novo centriole-like structures in dividing and rescued DSas-4 -/- cells. Furthermore, the previous proteins used to address DSas-4-GFP foci specificity have been used as a centriolar markers by several studies (Blachon et al., 2009; Dzhindzhev et al., 2010; Gopalakrishnan et al., 2012; Gopalakrishnan et al., 2013; Martinez-Campos et al., 2004; Peel et al., 2007; Rodrigues-Martins et al., 2007; Stevens et al., 2010). 113 In the last years, the de novo formation of centrioles has been adressed through different experimental approaches. From these studies, it is important to emphasize two studies in which de novo centriole-like structures were formed after overexpressing specific centriolar proteins in unfertilized Drosophila eggs. Rodrigues-Martins and coworkers (2007) have shown that SAK-overexpression in unfertilized eggs led to the formation of de novo centriole structures containing centriolar and PCM components such as DSas-6, γ-tubulin and Cnn. Interestingly, SAK-induced centriole biogenesis in acentriolar Drosophila S2 cells was dependent on the presence of the structural proteins DSas-4 and DSas-6. Moreover, Peel et al. (2007) have also shown that overexpression of DSas-6, SAK and specifically of DSas-4 induced the de novo formation of centriole-like structures in unfertilized Drosophila eggs that colocalized with specific markers such as D- PLP, Cnn and γ-tubulin. In agreement with the former study, it was also shown that extra centriole formation in larval brain cells through SAK-overexpression was also dependent on DSas-4 presence (Peel et al., 2007). The role of DSas-4 in centrosome biology has been thoroughly studied. In spite of its essential function in the centriole duplication process, it was shown that DSas-4 is a very important player in PCM recruitment. It promotes the assembly of cytoplasmic complexes (S-CAP) composed by proteins such as Cnn, D-PLP and Asl (Gopalakrishnan et al., 2012; Gopalakrishnan et al., 2013). This is complemented with the fact that its overexpression can originate acentriolar MTOCs (Dzhindzhev et al., 2010; Gopalakrishnan et al., 2013). Thus, it might be that the produced DSas-4-GFP foci containing D-PLP and Asl could merely be S-CAP not containing native centrioles. There is also some evidence that de novo centriole assembly is preceded by the formation of extensive clouds of electron-dense material that lacks centrioles, but contain PCM components such as γ-tubulin (Khodjakov et al., 2002; La Terra et al., 2005). Thus, along with the DSas-4-dependent formation of S-CAP, it is possible that DSas-4-GFP foci staining for centriolar markers, mainly D-PLP and Asl, may represent electron-dense material lacking centrioles. Nevertheless, and given the present data, there are several factors which favor the hypothesis that the de novo formed DSas-4-GFP foci in DSas-4 -/- R cells represent true centriole-like structures. First, the high percentage of colocalizations between centriolar markers and DSas-4-GFP foci in rescued DSas-4 -/- cells is indicative of DSas-4-GFP centriolar specificity. Second, the presence of DSas-6, a key protein in cartwheel assembly, points out the potential existence of a centriolar structure. Moreover, even though the protein fluorescence intensity levels for D-PLP, DSas-6, Ana1 and Asl were not measured in the DSas-4 -/- R cells (as it was performed for D-PLP and Asl in the mutant DSas-4 -/- cell line), the focused localization to a single dot of these centriolar 114 proteins at the mitotic poles of the majority DSas-4 -/- R cells is in contrast with the scattered localization or even absence verified for these markers in the mutant DSas-4 -/- cells. Thus, there is some evidence for a rescue of the acentriolar nature of DSas-4 -/- mutant cells after DSas-4-GFP expression, which supports the template-free mechanism in centriole biogenesis. It is worth noting that this quantitative analysis by immunofluorescence using centriolar markers should be repeated since the total number of mitotic DSas-4-/- R cells analysed per condition was not enough to perform a statistical analysis. Amplification of DSas-4-GFP foci in DSas-4 -/- rescued cells Centriole amplification has been directly linked to the de novo centriole formation in all studies that have adressed this mode of centriole assembly. Either by centrosome laser ablation or through the overexpression of centriole-specific proteins, it was shown the expression of extra centrioles in Drosophila and in mammalian cells (A. Rodrigues- Martins et al., 2007b; Khodjakov et al., 2002; La Terra et al., 2005; Peel et al., 2007; Uetake et al., 2007). In agreement with the previous studies, here it was shown that DSas-4 -/- R cells also formed more than two DSas-4-GFP foci per mitotic cell (54% of DSas-4 -/- R cells), whereas the majority of control mitotic cells only presented two foci (67%). Moreover, the localization of these foci in DSas-4 -/- R cells was variable. In some cells DSas-4-GFP foci were specifically localized in clusters at the mitotic poles or appeared randomly dispersed in the cytoplasm. In support of these results, Rodrigues-Martins et al. (2007) have shown two potential characteristics of de novo centriole assembly: (1) the ratio of cells with more than two centrosomes after SAK overexpression in acentrosomal Drosophila S2 cells was significantly increased compared with the control; (2) de novo formed centrosomes, in Drosophila unfertilized eggs overexpressing SAK, displayed scatterred and random positions compared with the clustered phenotype of de novo formed centrosomes in SAK overexpressing embryos. Moreover, Peel et al. (2007) showed that induction of de novo centriole-like structures in unfertilized Drosophila eggs through overexpression of DSas-6, DSas-4 and SAK led to the formation of supernumerary centrioles. Additionally, centrosome ablation studies in dividing mammalian cells, specifically in HeLa, RPE-1, HMEC and CHO culture cells, showed that de novo centriole assembly was always preceded by the formation of multiple centrioles within the cytoplasm of single 115 cells (from 2 up to 14 centriole-like structures) (Khodjakov et al., 2002; La Terra et al., 2005; Uetake et al., 2007). Although the present work was carried out in Drosophila somatic cells which are not comparable either to unfertilized Drosophila eggs or to mammalian culture cells, the evidence from the previous studies plus the data showed here support the idea that centriole amplification and the loss of spatial restriction might potentially be considered two hallmarks of de novo centriole biogenesis that are transversal to different organisms and cell types. In fact, it has been advocated that the mother centriole might appear as an intermediate responsible for the spatial and numerical control of centriole biogenesis. In the absence of the parental centriole, the spatial restriction and the numerical control over daughter centriole formation are lost. Recovery of the expression levels of centriolar proteins in DSas-4 -/- rescued cells In spite of the existence of two modes of centriole assembly, canonical/template and de novo, there is evidence that the molecular machinery underlying both forms of centriole biogenesis is the same (A. Rodrigues-Martins et al., 2007b; Peel et al., 2007). Since DSas-4 -/- cells are only mutant for DSas-4 protein and given the previous evidence for the potential formation of centriole-like structures after DSas-4-GFP expression in DSas-4 -/- R cells, it became imperative to address if the conserved centriole assembly pathway was activated through the recovery in the expression levels of specific proteins associated with the centriole biogenesis process. Asterless is a key element and upstream regulator of Drosophila centriole duplication (Dzhindzhev et al., 2010; Varmark et al., 2007). DSas-6 is a structural protein involved in the first steps of procentriole assemby and responsible for building the cartwheel in Drosophila and mammalian centrioles (Mennella et al., 2012; Rodrigues- Martins et al., 2007a; Van Breugel et al., 2011). As shown before, both proteins seemed to be either downregulated or delocalized in DSas-4 -/- mutant cell line (part 1, 1.2.), in which there was a decrease of approximately 70% of Asl levels at the centrosome and DSas-6 was absent in 100% of DSas-4 -/- cells. A comparative analysis of the protein expression pattern by western blot for Asl and DSas-6 revealed a rescue in the protein expression levels in the DSas-4 -/- R cells compared with the mutant DSas-4 -/- cell line. Although protein quantification was not performed, it is clear that there was a significant increase in the amount of protein between the DSas-4 transfected and non-transfected cell lines, mainly for Asl, in which there was almost a rescue in the expression levels similar to control cells. This recovery in 116 the expression levels of specific centriolar proteins might be suggestive of an activation of the centriole assembly pathway after DSas-4 expression, and therefore, representing the activation of de novo centriole biogenesis process. Interestingly, velocity sedimentation studies in Drosophila S2 cells overexpressing DSas-4 revealed that DSas-4-induced foci co-fractionate with several PCM components, including Asl (Gopalakrishnan et al., 2013). This evidence for a strong interaction between Asl and DSas-4 is in agreement with the dramatic increase verified for Asl levels in DSas- 4 -/- rescued cell line. As mentioned before, DSas-4 scaffolds cytoplasmic complexes during PCM formation and centrosome maturation. These S-CAP complexes include Asl and Cnn (Gopalakrishnan et al., 2012; Gopalakrishnan et al., 2013). Given this evidence, it is possible that the rescue in the protein expression levels of Asl can be associated with the formation of S-CAP complexes and are not directly related with the activation of the de novo centriole duplication process. Strikingly, in contrast to the recovery in Asl and DSas-6 protein levels between transfected and mutant DSas-4 cells, a transcriptomic analysis performed in the brain and imaginal disks of Drosophila DSas-4 and DSas-6 mutant flies (lacking centrosomes) did not reveal a significant differential expression between both mutants. Only in DSas-4 mutants there was a mis-regulation (up or down-regulation) in 14 genes compared to the WT background. However, these differentially regulated genes could not be clustered into any obvious functional pathway such as the centriole biogenesis pathway (Baumbach et al., 2012). Cnn is a PCM protein directly linked to PCM formation and, therefore responsible for controlling centrosome size (Conduit et al., 2010). Surprisingly, and in opposition to Asl and DSas-6 rescue, Cnn expression levels did not dramatically change between transfected and non-transfected DSas-4 cell lines. Given that an interaction between DSas-4, Asl and Cnn has been reported (Conduit et al., 2010; Gopalakrishnan et al., 2012; Gopalakrishnan et al., 2013), it would be expected a rescue in the Cnn expression levels similar to what was verified for Asl and DSas-6 in DSas-4 -/- R cells. The rescue in Cnn expression would be suggestive of either a recovery in centrosome functionality as a consequence of the de novo centriole assembly or the formation of S-CAP complexes without the production of centriolar structures. However, in contrast with Asl and DSas-6 biological functions, it is important to emphasize that Cnn is a PCM component and it is not directly associated with the centriole biogenesis process. For this reason, Cnn expression levels might not change dramatically in the presence or absence of centrioles, but its recruitment to centrosomes might be affected. Indeed, the characterization of acentriolar DSas-4 -/- mutant cell line for 117 centriolar and centrosomal proteins revealed that Cnn was detected as disperse dots of variable size at the mitotic poles, suggestive of centrosome delocalization. The de novo formed DSas-4-GFP foci are able to nucleate microtubules and to accumulate Centrosomin during mitosis The centrosome is the main microtubule organizing centre in animal cells. In the first stages of mitosis, the centrosome increases in size through the accumulation of PCM components and begins to define the mitotic spindle through the generation of astral microtubules according to an "outside-inside" process, in order to promote correct spindle positioning and accurate chromosome segregation (Meunier et al., 2012). Centrioles are key elements in this process because are focal points for the accumulation of pericentriolar material. In fact, Bobinnec and co-workers (1998) showed that centriole disintegration in HeLa cells led to PCM unstability and dispersion along with the loss of well-define MT asters. In this regard, it was adressed the functionality of DSas-4-GFP foci in DSas-4 -/- R cells for the production of astral microtubules, mitotic spindle morphology and the accumulation of the specific PCM component, Centrosomin. First, it was shown that the de novo DSas-4-GFP foci formed in mitotic DSas-4 rescued cells were able to nucleate astral microtubules, which is suggestive of a restoration in the microtubule organizing capacity. Indeed, the mitotic spindle in some DSas-4 -/- R cells seemed to derive from opposite and focused DSas-4-GFP positive poles. The evidence for astral microtubule nucleation in DSas-4 -/- R cells is a strong indicator of centrosome presence due to the de novo centriole assembly. Nonetheless, the microtubule fibers emanating from DSas-4-GFP foci in DSas-4 -/- R cells varied in shape as well as in number. These newly formed astral MTs were usually smaller and not so abundant comparing with the long and very well-defined MT asters emanating from the centrosomes of control cells. This last evidence suggests that the de novo formed DSas-4-GFP foci may represent structurally abnormal centrioles unable to fully recruit PCM and, therefore to proper restore centrosome function as MTOC. In agreement with this hypothesis, Peel et al. (2007) showed that, in unfertilized Drosophila eggs overexpressing recombinant centriole-specific proteins, only the very bright and larger structures were able to organize microtubules and to accumulate PCM markers such as Cnn and γ-tubulin. Although the previous study lacked electron microscopy analysis, it suggests the presence of abnormal de novo centrioles that are not able to organize and recruit PCM components in order to form a fully functional centrosome. 118 Moreover, it was previously shown that somatic 1182-4D acentriolar cells were able to recruit PCM components (e.g. γ-tubulin) to the polar regions. These PCM aggregates retained some MT nucleation capacity, although the resulting aster-like structures were 3- 4 fold smaller than wild-type cells (Moutinho-Pereira et al., 2009). Thus, it is possible that the MT asters seen in DSas-4 -/- R cells might just represent PCM aggregates with some MT nucleation capacity, but lacking centrioles. Interestingly, astral MTs emanating from PCM aggregates were not clearly discernible in the DSas-4 -/- mutant cells (see part 1, 1.1. and 1.2.). It will be interesting in the future to measure and compare the size of the MT asters between control and DSas-4 -/- R cells, in order to address whether there was a rescue in the astral MT formation. To further investigate the microtubule nucleating capacity of the de novo formed DSas-4-GFP foci, the recruitment of Cnn was analysed. As already mentioned, Cnn is highly recruited to centrosomes at the onset of mitosis and is a very important element for PCM organization and centrosome maturation (Conduit et al., 2010; Dobbelaere et al., 2008; Megraw et al., 1999; Vaizel-Ohayon et al., 1999). DSas-4-GFP foci in DSas-4 -/-R mitotic cells showed the presence of Cnn as a single and bright dot at the two opposite poles, whereas it was detected as small agregates of variable size at the mitotic poles of DSas-4 mutant background. This apparent strong and focalized Cnn presence at DSas-4- GFP foci is in contrast with the western blot analysis previoulsly described, in which there was not a recovery in Cnn expression levels after DSas-4-GFP expression in DSas-4 -/- R cells. However, it should be noted that the principle behind western blot is an analysis over the total amount of protein in the cell, and not the amount of protein in a particular cell compartment such as the centrosome. As previously claimed, and even though the fluorescence intensity protein levels of Cnn in DSas-4 -/- R cell line were not measured, it might be possible that Cnn expression is not altered between acentriolar and rescued DSas-4 -/- cells, but this protein is more efficiently recruited and concentrated at the poles due to the potential formation of de novo MTOCs in DSas-4 -/- R cells. In fact, since it was verified a decrease of approximately 70% in the Cnn protein fluorescence intensity levels in the mitotic poles of mutant DSas-4 cells, it would be interesting to investigate if there is a recovery of Cnn protein fluorescence intensity in the DSas-4 -/- R cell line. In a canonical centrosome, MT anchoring and organization is only performed by the mother centriole due to the presence of distal and sub-distal appendages and PCM enrichment, which are hallmarks of mature centrioles (Avidor-Reiss et al., 2013; Brito et al., 2012; Cunha-Ferreira et al., 2009; Debec et al., 2010). Because Drosophila centrioles do not show obvious appendages, one interesting possibility would be that DSas-4-GFP foci correspond to de novo centrioles that are able to mature only with the acquisition of PCM, and thus perform astral MT nucleating activity. In this study, recombinant DSas-4- 119 GFP expression was induced over a period of 48h and DSas-4 -/- cells have a doubling time of approximately 36h. Therefore, since centrioles require approximately two cell cycles from their initial formation to fully mature and, even though the kinetics of de novo centriole biogenesis is slower than the canonical pathway (A. Rodrigues-Martins et al., 2007b; Khodjakov et al., 2002; La Terra et al., 2005; Marshall et al., 2001), the previous conditions would, in principle, favor not only the de novo centriole formation, but also centriole maturation. This issue remains to be answered. Given there was some evidence for astral MT nucleation activity from DSas-4-GFP foci, the overall mitotic spindle morphology in DSas-4 -/- R cells was analysed, specifically the percentage of cells exhibiting a bipolar and focused mitotic spindle. Even though it was verified that approximately half of the rescued cells (50%) with DSas-4-GFP foci exhibited a WT-like mitotic spindle followed by the dual spindle morphology, the percentage of rescued cells exhibiting a WT-like spindle did not significantly increase comparing with the percentage of DSas-4 mutant cells exhibiting the same mitotic spindle morphology (45%). In fact, the percentage of cells in each spindle category (barrel, dual, non-polarized and hyperpolarized) was not significantly different between the transfected and mutant DSas-4 cell lines. The incapacity to rescue mitotic spindle morphology in DSas-4 -/- R cells might reflect incomplete centrosome maturation. Although a MT depolymerizing assay was not performed in order to identify sites of MT assembly, live cell imaging analysis of Jupiter-GFP in DSas-4 -/- R cell line suggests that mitotic cells containing DSas-4-GFP foci at the poles progressively established the mitotic spindle through discrete astral MT nucleation in an "outside-inside" process. Nevertheless, we cannot exclude that centrosome-independent mechanisms for mitotic spindle assembly are also involved. Interestingly, the spindle morphology of some dividing DSas-4 -/- R cells was still non-focused in spite of astral microtubule generation from DSas-4-GFP foci. This last evidence is in line with the imunofluorescence results for mitotic spindle morphology previously described, and reinforces the notion that some DSas-4-GFP foci might represent centriole-like structures that are not fully functional due to abnormalities in their structure or to different stages of centriole maturation that impair MT nucleation capacity. Additionally, it could also be that only one of the poles has formed de novo centrioles, explaining the assymetry in the spindle morphology of DSas-4 -/- R cells. In agreement with the DSas-4-GFP amplification verified by fixed cell analysis, live imaging of rescued DSas-4 -/- cells also showed the formation of multiple DSas-4-GFP foci within the cytoplasm of interphasic cells. Consequently, mitotic DSas-4 -/- R assembled multipolar spindles with more than two DSas-4-GFP spots from which discrete MT bundles emerged. La Terra et al. (2005) have also shown that multipolar spindles are 120 produced when de novo centriole assembly is induced in acentriolar HeLa cells. Moreover, the EM reconstruction revealed an association between de novo formed centrioles and the spindle poles. Therefore, it might be that DSas-4-GFP foci correspond to centriole-like structures that are able to mature and function as an MTOC. Furthermore, there is also some evidence for an increase in the length of mitosis that seems to be dependent on mitotic spindle morphology. DSas-4 -/- R cells presenting nearly focused spindles spent half of the time to complete mitosis comparing to cells with non-focused spindles. Indeed, it is known that cells without functional centrosomes spend more time (~ 2-3 times) in mitosis, especially in G2-M phases, in which a MT network needs to be generated and organized for chromosome segregation in metaphase. In the absence of centrioles, there is the lack of two specialized sources of microtubules at the poles and consequently, the formation of anastral spindles is mainly performed by mitotic chromatin in a potential kinetically slower “inside-outside” process (Basto et al., 2006; Lecland et al., 2013; Moutinho-Pereira et al., 2013; Sir et al., 2013). Although the previous studies did not address the relationship of anastral spindle configurations and mitotic duration, one might expect that cells with anastral non-focused spindle configurations take more time in mitosis due to the absence of focal points like centrosomes that guide mitosis in a more efficient way than the remaining MT assembly pathways. A more detailed analysis of the mitotic spindle geometry and mitotic timing should be performed. Furthermore, given that it is believed that centrosome accelarates mitotic spindle assembly, it would be interesting to compare the duration of mitosis between WT mitotic spindles and DSas-4 -/- rescued cells presenting nearly focused spindles, and assess if there is a rescue in mitotic timing. If that is the case, it would potentially represent the assembly of MTOCs due to the de novo centriole assembly. de novo centriole-like structures are formed in DSas-4 -/- rescued cells Drosophila centrioles are formed by the cartwheel, a structure composed by a central hub from which nine spokes radiate outward and are responsible for displaying the centriole 9-fold symmetry. This structure is connected to nine sets of microtubule doublets that decorate the outer surface of the centriole (Gonczy et al., 2012; Guichard et al., 2010; Guichard et al., 2013) (for a more complete description of centriole structure see introduction). An electron microscopy analysis became imperative to address whether DSas-4- GFP foci represented centrioles, since they incorporate PCM as well as centriolar markers and seemed to organize astral MTs to a certain degree. 121 The EM analysis of asynchronous DSas-4 -/- rescued cells revealed that DSas-4- GFP foci represent de novo assembled centrioles. These centrioles have developed into recognizable and structurally normal centrioles, apparently composed by the carthweel and a centriolar MT wall. In just one case, there was the presence of a centriole-like structure surrounded by microtubules Additionally, some cells showed clouds of electron-dense material lacking centrioles. This is in line with the hypothesis that some DSas-4-GFP foci might only be aggregates of PCM material (e.g. Asl, Cnn, D-PLP), since DSas-4 forms S-CAP complexes with pericentriolar proteins. Although there was no presence of clouds of electron-dense material surrounded by MTs in this EM analysis, it is known that PCM aggregates have some MT nucleation capacity, as previously described by Bobinnec et al. (1998) and Moutinho-Pereira et al. (2009), in which MT regrowth assays in vertebrate and Drosophila cells revealed the presence of small foci of PCM together with short MTs disseminated both within the cytoplasm and in the mitotic poles. Moreover, as already mentioned, DSas-4 overexpression can originate acentriolar MTOCs containing γ-tubulin (Dzhindzhev et al., 2010). Furthermore, these aggregates of pericentriolar material may also represent the earlier stages of de novo centriole biogenesis since this mode of centriole formation is potentially preceded by the production of clouds of PCM material that might provide an ideal environment to support centriole assembly (Khodjakov et al., 2002; La Terra et al., 2005). Thus, it is likely that some DSas-4 cells expressing DSas-4- GFP protein may lack a functional centrosome. Thus far, all studies approaching de novo centriole biogenesis either in Drosophila or mammalian cells, reported the presence of centriole-like structures in different stages of maturation (Bobinnec et al., 1998; Khodjakov et al., 2002; La Terra et al., 2005; Loncarek et al., 2008). These centriole-like structures included several types of structural anomalies such as incomplete cylinders, distorted centriolar walls and different cylinder lengths. It was also reported the presence of electron dense amorphous clouds that did not contain centrioles at their core (La Terra et al., 2005). Interestingly, only one study have reported the de novo formation of structurally normal centrioles (Rodrigues-Martins et al., 2007b). Additionally, the previous study along with time lapse microscopy analysis in HeLa cells (La Terra et al., 2005) have also showed the presence of procentrioles closely associated with the de novo assembled centrioles, which suggests a switch from de novo mode of assembly to rounds of canonical duplication. In opposition, Peel et al. (2007) showed that de novo formed centriole-like structures were not capable of undergoing rounds of template-mediated duplication. The EM analysis performed in this study revealed the absence of procentrioles next to the de novo formed centriole-like structures, even in the ones that were more 128 Another interesting hypothesis that might be claimed on the light of the previous arguments is based on the sequential and concomitantly occurrence of these two centriole biogenesis pathways during centriole duplication. It is possible that centriole biogenesis in S phase is first preceded by the de novo formation of centriolar precursors within PCM clouds throughout the cytoplasm, that is followed by the attachment to the proximal-end of the resident centriole of only one of those percursors. This last step would offer a kinetic advantage and restrict centriole number in order to promote fidelity in mitosis. One pertinent question that may arise considering the data from this study and from other studies is, why do the de novo assembled centrioles frequently exhibit a wide variety of intermediate morphological stages. One might expect that the mother centriole could act as a bona fide template ensuring the correct structural assembly of the new centriole. Indeed, the presence of the pre-existing centriole may partially ensure a correct centriole structure assembly and therefore, it may function as a monitoring mechanism of centriole morphology. Further studies will be required to clarify this issue. The evidence for a common set of proteins regulating the template and the de novo modes of centriole assembly has favor the hypothesis that these two modes are not more than variations of a common pathway (Gopalakrishnan et al., 2013; Loncarek et al., 2007). Moreover, both pathways seem to be strictly linked to each other since the de novo mode of centriole assembly never occurs when a mature centriole is present (Loncarek et al., 2008; La Terra et al., 2005). One might expect that if these two pathways were different, the de novo centriole formation should occur even in the presence of the mother centriole. Therefore, cells may take advantage of redundacy, so that there are two possible mechanisms or “roads” based on the same molecular elements working on the same goal – centriole assembly. Indeed, as previously claimed, it might be possible an interaction between these pathways, in which centriole assembly is first preceded by the de novo assembly of procentriole structures followed by the template mode, in which there is the attachment of one of those previously assembled structures into the mature centriole. This last speculative scenario may ensure a spatial and number restriction of centriole assembly. Another possibility is that, even though there are two choices of how to produce centrioles, cells may take advantage of the mother centriole as a platform for the new centriole assembly because it may offer a more parsimonious advantage by offering an ideal environment for the successful new centriole formation in S phase, while ensuring fidelity on centrosome biogenesis. In this regard, there could be a competition between both pathways, in which the templated mechanism ultimately inhibits the de novo pathway on each cell duplication cycle. 129 Strikingly, there is still one specific question that remains to be answered: "If template centriole duplication provides far more precise control over centriole numbers, then why do so many organisms employ centriole de novo formation, in such critical events as early embryogenesis ?" (Loncarek et al., 2009). Probably, the most striking example of precision on de novo centriole assembly is given during mouse development. It is known that centrioles arise de novo and in a proper number during the 16-cell blastomere stage, which is in contrast with all the existing reports addressing de novo centriole biogenesis in all model systems, including the present study. The nature of this difference remains a mystery, but it may lie on different concentrations of regulatory elements implicated in the centriole duplication pathway. Centrosome amplification is a well-known hallmark of cancer cells. The presence of extra centrosomes increases the occurrence of multipolar mitosis, which in turn might provide an ideal scenario for chromosome misegregation events followed by genomic instability (Nigg and Raff, 2009). Our observations that the de novo pathway promotes the formation of extra centrioles raises the possibility of a connection between this mode of centriole assembly and cancerous cells. Since centriole duplication is a very conserved process among vertebrates and invertebrates, the DSas-4 mutant cell line might provide an ideal system to address the previous issue. To conclude, it is worth noting that the present study is a collection of still very preliminary data that needs to be consolidated. Nonetheless, it provides the proof-of- principle of the de novo centriole biogenesis process on somatic cycling cells under normal physiological conditions Based on the expression “a picture is worth a thousand words”, this study gives very strong evidence for a template-free mechanism for centriole biogenesis. This study attempts to be a starting point or a groundwork to prompt more questions and to stimulate further studies. 130 131 REFERENCES 132 133 Andersen, J.S., Wilkinson, C.J., Mayor, T., Mortensen, P., Nigg, E.A., and Mann, M. (2003). Proteomic characterization of the human centrosome by protein correlation profiling. Nature 426, 570-574. Arquint, C., Sonnen, K.F., Stierhof, Y.D., and Nigg, E.A. (2012). Cell-cycle-regulated expression of STIL controls centriole number in human cells. Journal of cell science 125, 1342-1352. Avidor-Reiss, T., and Gopalakrishnan, J. (2013). Building a centriole. Current opinion in cell biology 25, 72-77. Azimzadeh, J., and Marshall, W.F. (2010). Building the centriole. Current biology : CB 20, R816-825. Azimzadeh, J., Wong, M.L., Downhour, D.M., Sanchez Alvarado, A., and Marshall, W.F. (2012). Centrosome loss in the evolution of planarians. Science 335, 461-463. Bahe, S., Stierhof, Y.D., Wilkinson, C.J., Leiss, F., and Nigg, E.A. (2005). Rootletin forms centriole-associated filaments and functions in centrosome cohesion. The Journal of cell biology 171, 27-33. Bartolini, F., and Gundersen, G.G. (2006). Generation of noncentrosomal microtubule arrays. Journal of cell science 119, 4155-4163. Basto, R., Lau, J., Vinogradova, T., Gardiol, A., Woods, C.G., Khodjakov, A., and Raff, J.W. (2006). Flies without centrioles. Cell 125, 1375-1386. Bettencourt-Dias, M. (2013). Q&A: Who needs a centrosome? BMC biology 11, 28. Bettencourt-Dias, M., and Glover, D.M. (2007). Centrosome biogenesis and function: centrosomics brings new understanding. Nature reviews Molecular cell biology 8, 451- 463. Bettencourt-Dias, M., Rodrigues-Martins, A., Carpenter, L., Riparbelli, M., Lehmann, L., Gatt, M.K., Carmo, N., Balloux, F., Callaini, G., and Glover, D.M. (2005). SAK/PLK4 is required for centriole duplication and flagella development. Current biology : CB 15, 2199- 2207. Blachon, S., Cai, X., Roberts, K.A., Yang, K., Polyanovsky, A., Church, A., and Avidor- Reiss, T. (2009). A proximal centriole-like structure is present in Drosophila spermatids and can serve as a model to study centriole duplication. Genetics 182, 133-144. Blachon, S., Gopalakrishnan, J., Omori, Y., Polyanovsky, A., Church, A., Nicastro, D., Malicki, J., and Avidor-Reiss, T. (2008). Drosophila asterless and vertebrate Cep152 Are orthologs essential for centriole duplication. Genetics 180, 2081-2094. Bobinnec, Y., Khodjakov, A., Mir, L.M., Rieder, C.L., Edde, B., and Bornens, M. (1998). Centriole disassembly in vivo and its effect on centrosome structure and function in vertebrate cells. The Journal of cell biology 143, 1575-1589. Bond, J., Roberts, E., Springell, K., Lizarraga, S.B., Scott, S., Higgins, J., Hampshire, D.J., Morrison, E.E., Leal, G.F., Silva, E.O., et al. (2005). A centrosomal mechanism involving CDK5RAP2 and CENPJ controls brain size. Nature genetics 37, 353-355. Bornens, M., and Azimzadeh, J. (2007). Origin and evolution of the centrosome. Advances in experimental medicine and biology 607, 119-129. Brito, D.A., Gouveia, S.M., and Bettencourt-Dias, M. (2012). Deconstructing the centriole: structure and number control. Current opinion in cell biology 24, 4-13. Buchman, J.J., Tseng, H.C., Zhou, Y., Frank, C.L., Xie, Z.G., and Tsai, L.H. (2010). Cdk5rap2 Interacts with Pericentrin to Maintain the Neural Progenitor Pool in the Developing Neocortex. Neuron 66, 386-402. Carvalho-Santos, Z., Azimzadeh, J., Pereira-Leal, J.B., and Bettencourt-Dias, M. (2011). Evolution: Tracing the origins of centrioles, cilia, and flagella. The Journal of cell biology 194, 165-175. Carvalho-Santos, Z., Machado, P., Branco, P., Tavares-Cadete, F., Rodrigues-Martins, A., Pereira-Leal, J.B., and Bettencourt-Dias, M. (2010). Stepwise evolution of the centriole-assembly pathway. Journal of cell science 123, 1414-1426. Chang, J., Cizmecioglu, O., Hoffmann, I., and Rhee, K. (2010). PLK2 phosphorylation is critical for CPAP function in procentriole formation during the centrosome cycle. The EMBO journal 29, 2395-2406. 134 Choi, Y.K., Liu, P.F., Sze, S.K., Dai, C., and Qi, R.Z. (2010). CDK5RAP2 stimulates microtubule nucleation by the gamma-tubulin ring complex. Journal of Cell Biology 191, 1089-1095. Cizmecioglu, O., Arnold, M., Bahtz, R., Settele, F., Ehret, L., Haselmann-Weiss, U., Antony, C., and Hoffmann, I. (2010). Cep152 acts as a scaffold for recruitment of Plk4 and CPAP to the centrosome. The Journal of cell biology 191, 731-739. Conduit, P.T., Brunk, K., Dobbelaere, J., Dix, C.I., Lucas, E.P., and Raff, J.W. (2010). Centrioles regulate centrosome size by controlling the rate of Cnn incorporation into the PCM. Current biology : CB 20, 2178-2186. Cormier, A., Clement, M.J., Knossow, M., Lachkar, S., Savarin, P., Toma, F., Sobel, A., Gigant, B., and Curmi, P.A. (2009). The PN2-3 domain of centrosomal P4.1-associated protein implements a novel mechanism for tubulin sequestration. The Journal of biological chemistry 284, 6909-6917. Cottee, M.A., Muschalik, N., Wong, Y.L., Johnson, C.M., Johnson, S., Andreeva, A., Oegema, K., Lea, S.M., Raff, J.W., and van Breugel, M. (2013). Crystal structures of the CPAP/STIL complex reveal its role in centriole assembly and human microcephaly. eLife 2, e01071. Cottee, M.A., Raff, J.W., Lea, S.M., and Roque, H. (2011). SAS-6 oligomerization: the key to the centriole? Nature chemical biology 7, 650-653. Cunha-Ferreira, I., Bento, I., and Bettencourt-Dias, M. (2009). From zero to many: control of centriole number in development and disease. Traffic 10, 482-498. Dammermann, A., Maddox, P.S., Desai, A., and Oegema, K. (2008). SAS-4 is recruited to a dynamic structure in newly forming centrioles that is stabilized by the gamma-tubulin- mediated addition of centriolar microtubules. The Journal of cell biology 180, 771-785. Dammermann, A., Muller-Reichert, T., Pelletier, L., Habermann, B., Desai, A., and Oegema, K. (2004). Centriole assembly requires both centriolar and pericentriolar material proteins. Developmental cell 7, 815-829. Dawe, H.R., Farr, H., and Gull, K. (2007). Centriole/basal body morphogenesis and migration during ciliogenesis in animal cells. Journal of cell science 120, 7-15. Debec, A., Detraves, C., Montmory, C., Geraud, G., and Wright, M. (1995). Polar organization of gamma-tubulin in acentriolar mitotic spindles of Drosophila melanogaster cells. Journal of cell science 108 ( Pt 7), 2645-2653. Debec, A., Sullivan, W., and Bettencourt-Dias, M. (2010). Centrioles: active players or passengers during mitosis? Cellular and molecular life sciences : CMLS 67, 2173-2194. Debec, A., Szollosi, A, Szollosi, D. (1982). A Drosophila melanogaster cell line lacking centriole. Bio Cell 44, 133-138. Delattre, M., Canard, C., and Gonczy, P. (2006). Sequential protein recruitment in C. elegans centriole formation. Current biology : CB 16, 1844-1849. Delattre, M., Leidel, S., Wani, K., Baumer, K., Bamat, J., Schnabel, H., Feichtinger, R., Schnabel, R., and Gonczy, P. (2004). Centriolar SAS-5 is required for centrosome duplication in C. elegans. Nature cell biology 6, 656-664. Dix, C.I., and Raff, J.W. (2007). Drosophila Spd-2 recruits PCM to the sperm centriole, but is dispensable for centriole duplication. Current biology : CB 17, 1759-1764. Dobbelaere, J., Josue, F., Suijkerbuijk, S., Baum, B., Tapon, N., and Raff, J. (2008). A genome-wide RNAi screen to dissect centriole duplication and centrosome maturation in Drosophila. PLoS biology 6, e224. Duensing, A., Liu, Y., Tseng, M., Malumbres, M., Barbacid, M., and Duensing, S. (2006). Cyclin-dependent kinase 2 is dispensable for normal centrosome duplication but required for oncogene-induced centrosome overduplication. Oncogene 25, 2943-2949. Dzhindzhev, N.S., Yu, Q.D., Weiskopf, K., Tzolovsky, G., Cunha-Ferreira, I., Riparbelli, M., Rodrigues-Martins, A., Bettencourt-Dias, M., Callaini, G., and Glover, D.M. (2010). Asterless is a scaffold for the onset of centriole assembly. Nature 467, 714-718. Efimov A, Kharitonov A, Efimova N, Loncarek J, Miller PM, Andreyeva N, Gleeson P, Galjart N, Maia AR, McLeod IX, Yates JR, Maiato H, Khodjakov A, Akhmanova A, 135 Kaverina I. (2007). Asymmetric CLASP-dependent nucleation of noncentrosomal microtubules at the trans-Golgi network. Developmental cell 12, 917-30. Fisk, H.A., and Winey, M. (2001). The mouse Mps1p-like kinase regulates centrosome duplication. Cell 106, 95-104. Fong, K.W., Choi, Y.K., Rattner, J.B., and Qi, R.Z. (2008). CDK5RAP2 is a pericentriolar protein that functions in centrosomal attachment of the gamma-tubulin ring complex. Molecular biology of the cell 19, 115-125. Fry, A.M., Mayor, T., Meraldi, P., Stierhof, Y.D., Tanaka, K., and Nigg, E.A. (1998). C- Nap1, a novel centrosomal coiled-coil protein and candidate substrate of the cell cycleregulated protein kinase Nek2. The Journal of cell biology 141, 1563-1574. Fu, J., and Glover, D.M. (2012). Structured illumination of the interface between centriole and peri-centriolar material. Open biology 2, 120104. Gonczy, P. (2012). Towards a molecular architecture of centriole assembly. Nature reviews Molecular cell biology 13, 425-435. Gopalakrishnan, J., Chim, Y.C., Ha, A., Basiri, M.L., Lerit, D.A., Rusan, N.M., and Avidor- Reiss, T. (2012). Tubulin nucleotide status controls Sas-4-dependent pericentriolar material recruitment. Nature cell biology 14, 865-873. Gopalakrishnan, J., Mennella, V., Blachon, S., Zhai, B., Smith, A.H., Megraw, T.L., Nicastro, D., Gygi, S.P., Agard, D.A., and Avidor-Reiss, T. (2011). Sas-4 provides a scaffold for cytoplasmic complexes and tethers them in a centrosome. Nature communications 2, 359. Goshima, G., Wollman, R., Goodwin, S.S., Zhang, N., Scholey, J.M., Vale, R.D., and Stuurman, N. (2007). Genes required for mitotic spindle assembly in Drosophila S2 cells. Science 316, 417-421. Guichard, P., Chretien, D., Marco, S., and Tassin, A.M. (2010). Procentriole assembly revealed by cryo-electron tomography. The EMBO journal 29, 1565-1572. Guichard, P., Hachet, V., Majubu, N., Neves, A., Demurtas, D., Olieric, N., Fluckiger, I., Yamada, A., Kihara, K., Nishida, Y., et al. (2013). Native architecture of the centriole proximal region reveals features underlying its 9-fold radial symmetry. Current biology : CB 23, 1620-1628. Habedanck, R., Stierhof, Y.D., Wilkinson, C.J., and Nigg, E.A. (2005). The Polo kinase Plk4 functions in centriole duplication. Nature cell biology 7, 1140-1146. Hatzopoulos, G.N., Erat, M.C., Cutts, E., Rogala, K.B., Slater, L.M., Stansfeld, P.J., and Vakonakis, I. (2013). Structural analysis of the G-box domain of the microcephaly protein CPAP suggests a role in centriole architecture. Structure 21, 2069-2077. Helps, N.R., Luo, X., Barker, H.M., and Cohen, P.T. (2000). NIMA-related kinase 2 (Nek2), a cell-cycle-regulated protein kinase localized to centrosomes, is complexed to protein phosphatase 1. The Biochemical journal 349, 509-518. Hinchcliffe, E.H., Li, C., Thompson, E.A., Maller, J.L., and Sluder, G. (1999). Requirement of Cdk2-cyclin E activity for repeated centrosome reproduction in Xenopus egg extracts. Science 283, 851-854. Hinchcliffe, E.H., Miller, F.J., Cham, M., Khodjakov, A., and Sluder, G. (2001). Requirement of a centrosomal activity for cell cycle progression through G1 into S phase. Science 291, 1547-1550. Hodges, M.E., Scheumann, N., Wickstead, B., Langdale, J.A., and Gull, K. (2010). Reconstructing the evolutionary history of the centriole from protein components. Journal of cell science 123, 1407-1413. Hsu, W.B., Hung, L.Y., Tang, C.J., Su, C.L., Chang, Y., and Tang, T.K. (2008). Functional characterization of the microtubule-binding and -destabilizing domains of CPAP and d- SAS-4. Experimental cell research 314, 2591-2602. Hung, L.Y., Chen, H.L., Chang, C.W., Li, B.R., and Tang, T.K. (2004). Identification of a novel microtubule-destabilizing motif in CPAP that binds to tubulin heterodimers and inhibits microtubule assembly. Molecular biology of the cell 15, 2697-2706. 136 Hung, L.Y., Tang, C.J., and Tang, T.K. (2000). Protein 4.1 R-135 interacts with a novel centrosomal protein (CPAP) which is associated with the gamma-tubulin complex. Molecular and cellular biology 20, 7813-7825. Iwaki, T., and Castellino, F.J. (2008). A single plasmid transfection that offers a significant advantage associated with puromycin selection in Drosophila Schneider S2 cells expressing heterologous proteins. Cytotechnology 57, 45-49. Karpova, N., Bobinnec, Y., Fouix, S., Huitorel, P., and Debec, A. (2006). Jupiter, a new Drosophila protein associated with microtubules. Cell motility and the cytoskeleton 63, 301-312. Kemp, C.A., Kopish, K.R., Zipperlen, P., Ahringer, J., and O'Connell, K.F. (2004). Centrosome maturation and duplication in C. elegans require the coiled-coil protein SPD- 2. Developmental cell 6, 511-523. Khodjakov, A., Cole, R.W., Oakley, B.R., and Rieder, C.L. (2000). Centrosomeindependent mitotic spindle formation in vertebrates. Current biology : CB 10, 59-67. Khodjakov, A., Rieder, C.L., Sluder, G., Cassels, G., Sibon, O., and Wang, C.L. (2002). De novo formation of centrosomes in vertebrate cells arrested during S phase. The Journal of cell biology 158, 1171-1181. Kim, T.S., Park, J.E., Shukla, A., Choi, S., Murugan, R.N., Lee, J.H., Ahn, M., Rhee, K., Bang, J.K., Kim, B.Y., et al. (2013). Hierarchical recruitment of Plk4 and regulation of centriole biogenesis by two centrosomal scaffolds, Cep192 and Cep152. Proceedings of the National Academy of Sciences of the United States of America 110, E4849-4857. Kirkham, M., Muller-Reichert, T., Oegema, K., Grill, S., and Hyman, A.A. (2003). SAS-4 is a C. elegans centriolar protein that controls centrosome size. Cell 112, 575-587. Kitagawa, D., Kohlmaier, G., Keller, D., Strnad, P., Balestra, F.R., Fluckiger, I., and Gonczy, P. (2011a). Spindle positioning in human cells relies on proper centriole formation and on the microcephaly proteins CPAP and STIL. Journal of cell science 124, 3884-3893. Kitagawa, D., Vakonakis, I., Olieric, N., Hilbert, M., Keller, D., Olieric, V., Bortfeld, M., Erat, M.C., Fluckiger, I., Gonczy, P., et al. (2011b). Structural basis of the 9-fold symmetry of centrioles. Cell 144, 364-375. Kleylein-Sohn, J., Westendorf, J., Le Clech, M., Habedanck, R., Stierhof, Y.D., and Nigg, E.A. (2007). Plk4-induced centriole biogenesis in human cells. Developmental cell 13, 190-202. Kollman, J.M., Merdes, A., Mourey, L., and Agard, D.A. (2011). Microtubule nucleation by gamma-tubulin complexes. Nature reviews Molecular cell biology 12, 709-721. La Terra, S., English, C.N., Hergert, P., McEwen, B.F., Sluder, G., and Khodjakov, A. (2005). The de novo centriole assembly pathway in HeLa cells: cell cycle progression and centriole assembly/maturation. The Journal of cell biology 168, 713-722. Lawo, S., Hasegan, M., Gupta, G.D., and Pelletier, L. (2012). Subdiffraction imaging of centrosomes reveals higher-order organizational features of pericentriolar material. Nature cell biology 14, 1148-1158. Lecland, N., Debec, A., Delmas, A., Moutinho-Pereira, S., Malmanche, N., Bouissou, A., Dupre, C., Jourdan, A., Raynaud-Messina, B., Maiato, H., et al. (2013). Establishment and mitotic characterization of new Drosophila acentriolar cell lines from DSas-4 mutant. Biology open 2, 314-323. 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. Nature cell biology 7, 115-125. Leidel, S., and Gonczy, P. (2003). SAS-4 is essential for centrosome duplication in C elegans and is recruited to daughter centrioles once per cell cycle. Developmental cell 4, 431-439. Li, S., Fernandez, J.J., Marshall, W.F., and Agard, D.A. (2012). Three-dimensional structure of basal body triplet revealed by electron cryo-tomography. The EMBO journal 31, 552-562. 137 Lin, Y.C., Chang, C.W., Hsu, W.B., Tang, C.J., Lin, Y.N., Chou, E.J., Wu, C.T., and Tang, T.K. (2013). Human microcephaly protein CEP135 binds to hSAS-6 and CPAP, and is required for centriole assembly. The EMBO journal 32, 1141-1154. Loncarek, J., Hergert, P., Magidson, V., and Khodjakov, A. (2008). Control of daughter centriole formation by the pericentriolar material. Nature cell biology 10, 322-328. Loncarek, J., and Khodjakov, A. (2009). Ab ovo or de novo? Mechanisms of centriole duplication. Molecules and cells 27, 135-142. Lucas, E.P., and Raff, J.W. (2007). Maintaining the proper connection between the centrioles and the pericentriolar matrix requires Drosophila centrosomin. The Journal of cell biology 178, 725-732. Mahen, R., and Venkitaraman, A.R. (2012). Pattern formation in centrosome assembly. Current opinion in cell biology 24, 14-23. Mahoney, N.M., Goshima, G., Douglass, A.D., and Vale, R.D. (2006). Making microtubules and mitotic spindles in cells without functional centrosomes. Current biology : CB 16, 564-569. Marshall, W.F., Vucica, Y., and Rosenbaum, J.L. (2001). Kinetics and regulation of de novo centriole assembly. Implications for the mechanism of centriole duplication. Current biology : CB 11, 308-317. 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. The Journal of cell biology 165, 673-683. Matsumoto, Y., Hayashi, K., and Nishida, E. (1999). Cyclin-dependent kinase 2 (Cdk2) is required for centrosome duplication in mammalian cells. Current biology : CB 9, 429-432. Mayor, T., Stierhof, Y.D., Tanaka, K., Fry, A.M., and Nigg, E.A. (2000). The centrosomal protein C-Nap1 is required for cell cycle-regulated centrosome cohesion. The Journal of cell biology 151, 837-846. Megraw, T.L., Li, K., Kao, L.R., and Kaufman, T.C. (1999). The centrosomin protein is required for centrosome assembly and function during cleavage in Drosophila. Development 126, 2829-2839. Mennella, V., Keszthelyi, B., McDonald, K.L., Chhun, B., Kan, F., Rogers, G.C., Huang, B., and Agard, D.A. (2012). Subdiffraction-resolution fluorescence microscopy reveals a domain of the centrosome critical for pericentriolar material organization. Nature cell biology 14, 1159-1168. Meraldi, P., Lukas, J., Fry, A.M., Bartek, J., and Nigg, E.A. (1999). Centrosome duplication in mammalian somatic cells requires E2F and Cdk2-cyclin A. Nature cell biology 1, 88-93. Meunier, S., and Vernos, I. (2012). Microtubule assembly during mitosis - from distinct origins to distinct functions? Journal of cell science 125, 2805-2814. Moutinho-Pereira, S., Debec, A. and Maiato, H. (2009). Microtubule Cytoskeleton Remodeling by Acentriolar Microtubule-organizing Centers at the Entry and Exit from Mitosis in Drosophila Somatic cells. Molecular Biology of the Cell 20, 2796-2808. Moutinho-Pereira, S., Stuurman, N., Afonso, O., Hornsveld, M., Aguiar, P., Goshima, G., Vale, R.D., and Maiato, H. (2013). Genes involved in centrosome-independent mitotic spindle assembly in Drosophila S2 cells. Proceedings of the National Academy of Sciences of the United States of America 110, 19808-19813. Muller, H., Schmidt, D., Steinbrink, S., Mirgorodskaya, E., Lehmann, V., Habermann, K., Dreher, F., Gustavsson, N., Kessler, T., Lehrach, H., et al. (2010). Proteomic and functional analysis of the mitotic Drosophila centrosome. The EMBO journal 29, 3344- 3357. Nigg, E.A. (2007). Centrosome duplication: of rules and licenses. Trends in cell biology 17, 215-221. Nigg, E.A., and Raff, J.W. (2009). Centrioles, centrosomes, and cilia in health and disease. Cell 139, 663-678. Nigg, E.A., and Stearns, T. (2011). The centrosome cycle: Centriole biogenesis, duplication and inherent asymmetries. Nature cell biology 13, 1154-1160.