The Mechanism of Centriole Inactivation in Starfish Oocytes

Joana Borrego Pinto Dissertation presented to obtain the Ph.D degree in Cell Biology Instituto de Tecnologia Química e Biológica | Universidade Nova de Lisboa Oeiras, May, 2015 The Mechanism of Centriole Inactivation in Starfish Oocytes PhD defense Joana Borrego Pinto Starting date: 1st September 2011 Defense date: 18th May 2015 Supervisors: Dr. Mónica Bettencourt-Dias, Instituto Gulbenkian de Ciência, Portugal Dr. Péter Lénárt, European Molecular Biology Laboratory, Germany Examiners: Dr. Edgar Gomes, Instituto de Medicina Molecular, Portugal Dr. Francesca Peri, European Molecular Biology Laboratory, Germany Dr. Isabelle Vernos, Centre de Regulació Genòmica, Spain Dr. Jan Ellenberg, European Molecular Biology Laboratory, Germany Dr. Marie-Hélene Verlhac, Collège de France, France 3 ACKNOWLEDGMENTS “Remember to look up at the stars and not down at your feet. Try to make sense of what you see and wonder about what makes the universe exist. Be curious. And however difficult life may seem, there is always something you can do and succeed at. It matters that you don’t just give up.” – Stephen Hawking First of all I would like to thank Péter Lénárt for accepting me as his first PhD student, and also for letting me work on centrioles. I had a great fun! I also have to thank him for all his support and for not complaining (too much) about all my singing and “loudness”. I would also like to thank Mónica Bettencourt-Dias for the centriole “bichinho” she transmits to everybody (it is very contagious!), and for all the help provided in this project as co-supervisor and TAC member. I would also like to thank my other TAC members, Jan Ellenberg and Francesca Peri, for all the helpful input in the project and for all their availability. I would also like to thank to all the people that contributed for this work, one way or another, it was great to work with you all: Julia König and Thomas MüllerReichert for the EM work, specially for all the hundred serial sections you had to do, Matthia Winter-Karreman and Yannick Schwab for the more serial sections and for all the happiness that you find in your work, Pedro Machado por ser o centriole “buddy” mais próximo, all lab members from Mónica Bettencourt-Dias for letting me participate in their journal club and for all the feedback, especialmente à Inês Bento por partilhar o mundo de eliminação de centrioles comigo, e pelas experiências tão “fixes” que fizemos em Woods Hole. Kreso! For taking such good care of the starfish! I would also like to thank all the former and current Lénárt lab members for all the great scientific conversations and non-scientific moments that we shared! You’ve created a great environment to work! Especially for all the “this-pear”s and funny (“S-W-A-T team, “’cause I am happy”, “that’s magenta”, “I had my first 4 experience...”, “I wanna steal your car/heart”, “...you know, in the streets of Paris”, “are we happy?”) moments! Special thanks to Philippe for being a “Rose” weirdo, Natalia for being the crazy mama, Johanna for all the marvelous cakes, Manuela for all the crazy Portuguese talk, Masha for being Masha, Ermin for breaking the centrifuges, Szilárd for being a Hungarian man trapped in an Irish body, Laura for showing her boyfriend around, Sara for being crrrazy, Imre for the craziness of being a physicist! A special thanks for Masashi Mori, my lab brother 2, and Kalman Somogyi, my lab papa, who both helped me so much! It was a pleasure to come to work everyday! I would like to thank the EMBL community, especially our 4th floor... that rocked! Specially Aastha, Kasia, Thibaut, Deepikaa Deepikaa, Iannaaaa, Ori, Marvin, Alessandra (not-Ginger Ale), Paolita, Gustavo, Oana, Aleksandrita, Ana Rita maluca, Serge, George Cryptic, Vaso, Karin, Simone, Nils Cooling System, Maria mama, Mendi, Dorito and Nicole (oh my, I hope I didn’t forget anyone!) for all the great moments: for all the cows, and “sereia à vista no portoooo”, for all the barbecues, for all the game nights, for all the flamingos, for all the movie nights, for all the smoothies and warm bretzels and humus and cheese and cakes, for all the chocolates, for all the Moscatel, for all the mentos, for all the grumpy cats, for all the ducks, for all the Muse concerts, for all the crazyyyy stuff, for all the dinners, for all the smiles! It was great to meet you all and spend so many cool moments with you! And always remember: “Um urso rosna quando um galho caiu sobre sua cabeça, mas ele estava sob o peso de uma árvore”. I also would like to thank all the Predocs, specially the year of 2011, for all the great fun, for the great community spirit and the great place to do science! Un petit salut a tous mes amis de Paris, notamment les kiiikiiiiissss qui sont toujours lá – c’est toujours beau la campagne et les singes et les petits pijamas – (vous me manquez téllement!), au Gomes lab qui m’a tellement bien accepté comme la petite étudiante de masteur! J’ai appris beaucoup avec vous! Au petit Bruno pour sa tranquilité, ao meu irmão de laboratório número 1 Daniel Osório e ao Edgar Gomes por me ter lançado na minha primeira investigação “à séria” e por continuar sempre disponível para me aturar. Et aussi a tous les autres (les cousins et les Liwerants) qui ne m’oublient pas et a tous mes amis Erasmus! Benno, Lana, Sammy, Ana Banana, Yvonne, JP thanks for still being around! 5 Um obrigada muito especial a todos os amigos que ficam, mesmo que a distância se alargue! Andreia Panada, Ana Sofia, Carina Men, Sara, Mariana, Melissa, Margarida, Bruno Afonso, Daniel, Neuza, Ana Zhu, Diana por continuarem aí, aconteça ou que acontecer! Porque vocês são lindos e continuam bem pertinho! Um muito obrigada a toda a minha família que não esquece de mim, embora a “pima” apanhe o avião tantas vezes e só volte esporadicamente. Obrigada especialmente ao Ruizinho que se preocupa mesmo quando finge bem, e à mãezinha que tanto lhe custa que a filhinha esteja tão longe: um muito obrigada por todo o teu apoio e carinho incondicional. Vou, mas trago-vos comigo. Finally, Konradino: what can I say? Thanks for all the help (work and life related). Thank you for being there, sometimes more than 500 miles away, and to be the “man who walks 500 more”. Thank you. Merci. Obrigada. You for sure make my life more colorful. 6 7 SUMMARY “We are drowning in information, while starving for wisdom. The world henceforth will be run by synthesizers, people able to put together the right information at the right time (...).” – Edward O. Wilson The centrosome is the major organizing center in a cell, composed by two centrioles, one mother and one daughter, and surrounded by a pericentriolar material, which nucleates microtubules. Centriole duplication and segregation is tightly coupled to cell cycle, which guarantees that centriole number is maintained over generations. During the somatic cell cycle, a pair of centrioles duplicates, after which each daughter cell receives a pair, forming a closed cycle. However, during fertilization, if both cells were to contribute with their pair of centrioles, gamete fusion would result in the double of the normal centriole number. Therefore, centriole number needs to be reduced during meiosis to prevent a surplus in centriole number that would lead to formation of a multipolar spindle in the zygote. Indeed, without exception, centrioles are actively eliminated in oocytes of all animal species, and only the sperm contributes with active centrioles to the zygote. The universality of centriole elimination in female meiosis demonstrates the essential nature of the process in animals. Nonetheless, its mechanisms remain poorly understood. The timing of centriole elimination varies between species; it may occur early in prophase I of meiosis or just before fertilization. In starfish oocytes (Patiria miniata), centrioles are present in fully grown oocytes, and are eliminated during the two consecutive meiotic divisions. I developed GFP markers to specifically label and follow centrioles during starfish meiosis by live imaging in order to visualize when the elimination occurs. I observed that at meiosis onset, two pairs of centrioles are present. One pair out of these is extruded into the first polar body at the end of meiosis I (MI), whereas one pair remains in the oocyte. No 14 15 TABLE OF CONTENTS “Às vezes ouço passar o vento; e só de ouvir o vento passar, vale a pena ter nascido” (Sometimes I hear the wind blowing; and just by listening to it, it is worth to be born) – Fernando Pessoa ACKNOWLEDGMENTS .................................................................................. 3 SUMMARY ....................................................................................................... 7 SUMÁRIO ....................................................................................................... 11 ABBREVIATIONS .......................................................................................... 23 1. INTRODUCTION ..................................................................................... 25 1.1. CENTROSOME ........................................................................................ 25 1.1.1. Historical outlook: the discovery of the centrosome ...................... 25 1.2. CENTROSOME FUNCTION ........................................................................ 26 1.2.1. The ancient function of the centrosome/basal body might have been to anchor flagella and cilia .................................................................. 26 1.2.2. Other centrosome functions in metazoan ...................................... 28 i. Centrosome function in spindle assembly ...................................... 28 ii. Other centrosomal functions .......................................................... 29 1.3. CENTROSOME STRUCTURE ..................................................................... 30 1.3.1. Centriole – a well conserved structure across eukaryotes ............ 31 1.3.2. PCM - constitution and formation .................................................. 33 1.3.3. Protein composition reflects phylogenetic relationships between eukaryotes .................................................................................................... 34 1.4. CONTROL OF CENTRIOLE NUMBER DURING CELL DIVISION ........................ 36 1.4.1. Cell division and DNA duplication ................................................. 36 1.4.2. Cell division and centriole duplication ........................................... 37 i. Centriole duplication is semi-conservative and occurs once per cell cycle.. ....................................................................................................... 37 16 ii. Centriole duplication: the onset and cartwheel formation ............... 38 iii. Centriole duplication: the centriole growth ..................................... 40 iv. Centriole and DNA segregation during mitosis .............................. 42 v. Exceptions to the rule: centriole number is not always conserved . 43 1.5. CENTRIOLE AGE: THE DIFFERENCE BETWEEN MOTHER AND DAUGHTER CENTRIOLE ......................................................................................................... 45 1.5.1. Mother and daughter centriole ....................................................... 45 1.5.2. Centriole maturation – the process ................................................ 46 1.5.3. The role of mother appendages in the anchoring to the plasma membrane ..................................................................................................... 48 i. Ciliogenesis ..................................................................................... 48 ii. Mother centriole movement during T-cell activation ....................... 49 1.5.4. Differential centrosome inheritance and asymmetric division ........ 50 1.6. CENTRIOLE ELIMINATION IN GAMETOGENESIS ........................................... 53 1.6.1. Meiosis overview ............................................................................ 54 1.6.2. Centrosome reduction during spermatogenesis ............................ 55 i. Sperm cells contain one pair of centrioles with different degrees of degeneration ............................................................................................. 56 ii. Sperm cells contain one single centriole with no degeneration and a proximal centriole-like ................................................................................ 57 iii. Sperm cells do not contain centrioles: complete degeneration ..... 57 1.6.3. Centriole elimination during oogenesis .......................................... 58 i. Centrioles are eliminated before prophase I: spindle assembly is acentriolar ................................................................................................. 60 ii. One single pair of centrioles is present at the beginning of meiosis: the sperm centriole intervenes .................................................................. 62 iii. Centrioles are eliminated at the end of meiosis: spindle assembly is centriolar ................................................................................................... 63 1.6.4. Scaling problems in oogenesis ...................................................... 63 i. Spindle positioning at the cell cortex ............................................... 63 ii. Spindle assembly ........................................................................... 66 1.7. CENTRIOLE ELIMINATION IN STARFISH OOCYTES ....................................... 68 1.7.1. Overview of starfish meiosis .......................................................... 68 17 1.7.2. Starfish oocyte as a model to study centriole elimination using molecular markers and live cell imaging ...................................................... 69 i. Centriole elimination in other model organisms .............................. 69 ii. Centriole elimination in starfish oocytes - what is known so far ..... 70 iii. Centriole elimination in starfish oocytes – the starting hypothesis 71 2. AIMS ........................................................................................................... 73 3. MATERIAL AND METHODS ..................................................................... 75 3.1. IDENTIFICATION OF HOMOLOGS OF CENTRIOLAR COMPONENTS IN STARFISH ......................................................................................................................... 75 3.2. FLUORESCENT MARKERS FOR LIVE IMAGING ............................................ 76 3.3. BIOLOGICAL MATERIAL ............................................................................ 77 3.3.1. Sperm and oocyte collection ......................................................... 77 3.3.2. Fertilization .................................................................................... 78 3.4. OOCYTE INJECTION ................................................................................ 78 3.5. DRUG TREATMENT ................................................................................. 80 3.6. CONFOCAL MICROSCOPY AND GENERAL IMAGE PROCESSING ................... 80 3.7. IMAGE ANALYSIS AND PROCESSING ......................................................... 81 3.7.1. Validation of mother and general centriole markers ...................... 82 i. Centriole detection .......................................................................... 82 ii. Quantification of overlap between mother and general centriole markers ..................................................................................................... 82 3.7.2. Quantification of centrioles extruded into polar bodies .................. 84 i. Vesicle autofluorescence subtraction from the Odf2-mEGFP channel ..................................................................................................... 84 ii. Mother centriole extrusion – quantification .................................... 86 3.7.3. 3D tracking of centrioles ................................................................ 86 i. Cell outline segmentation ............................................................... 86 ii. Centriole tracking over time and in 3D ........................................... 87 iii. Minimum distance between centriole and plasma membrane ...... 87 iv. Plotting minimum distances centriole-cell outline over time .......... 88 Calculation of centriole transport velocity ......................................... 88 3.7.4. Centrifugation experiments ........................................................... 89 i. Centriole movement quantification ................................................. 89 18 ii. Angle quantification ........................................................................ 90 3.7.5. MG-132 arrest and quantification ................................................... 90 i. MG-132 arrest in metaphase I ........................................................ 90 Cyclin B-mEGFP intensity quantification .......................................... 91 ii. MG-132 arrest in metaphase II ....................................................... 91 3.7.6. Electron microscopy ....................................................................... 92 i. High-pressure freezing and serial sectioning of the entire oocyte ... 92 Sample fixation ................................................................................. 92 Sample preparation ........................................................................... 92 Sample visualization: scanning for centrioles and tomography ........ 93 ii. Chemical fixation and sectioning around the PBI area ................... 93 Sample fixation ................................................................................. 93 Sample preparation ........................................................................... 93 X-ray ................................................................................................. 94 Sample visualization and tomography .............................................. 95 3.7.7. Morpholinos against mother centriole mRNA ................................. 95 i. Experimental details and image acquisition .................................... 96 ii. Quantification ................................................................................. 96 Acknowlegments: .............................................................................. 97 4. RESULTS .................................................................................................... 99 4.1. ESTABLISHMENT OF CENTRIOLE COMPOSITION AND LIVE CELL CENTRIOLAR MARKERS IN STARFISH ........................................................................................ 99 4.1.1. Identification of homologs of centriolar proteins ............................. 99 4.1.2. Establishment of centriolar markers ............................................. 102 4.1.3. Live cell centriolar markers are functional .................................... 104 4.2. STARFISH MEIOTIC SPINDLES ARE CENTRIOLAR ...................................... 107 4.2.1. Establishing live cell imaging conditions ...................................... 107 4.2.2. Live imaging with general centriole markers ................................ 110 4.3. MOTHER CENTRIOLES ARE EXTRUDED INTO THE POLAR BODIES .............. 113 4.3.1. Each centrosome consists of a mother and a daughter centriole 113 4.3.2. The mother centriole is specifically extruded into the second polar body ............................................................................................................ 115 19 4.4. EXTRUSION OF THE MOTHER CENTRIOLES IS ESSENTIAL FOR CENTRIOLE INACTIVATION .................................................................................................. 118 4.4.1. The single daughter centriole remaining in the egg does not contribute to the zygotic spindle ................................................................. 118 4.4.2. Mother centrioles artificially retained in the egg, remain active and contribute to the zygotic spindle ................................................................. 120 4.5. THE MOTHER CENTRIOLE IS SPECIFICALLY TRANSPORTED TO THE PLASMA MEMBRANE ...................................................................................................... 124 4.5.1. Tracking of the mother centriole reveals a two-step process ...... 124 4.5.2. Characterization of the mother centriole specific transport mechanism ................................................................................................. 127 i. Is the mother centriole specific transport driven by microtubules? 128 ii. Is the mother centriole transport driven by actin? ........................ 129 iii. Is the mother centriole transport dependent on the polar body I cytokinesis? ............................................................................................ 131 Mother centriole transport is independent of polar body cytokinesis ............................................................................................................ 133 iv. Mother centriole transport requires proximity to the nucleus ...... 135 4.6. THE MOTHER CENTRIOLE ANCHORS TO THE PLASMA MEMBRANE ............ 138 4.6.1. Actin and microtubules are not involved in centriole anchoring .. 138 i. Establishing the conditions to arrest oocytes in MII ...................... 138 ii. Mother centriole anchoring is independent of microtubules ........ 141 iii. Mother centriole anchoring is independent of dynamic actin ...... 143 4.6.2. Are the appendages connecting the mother centriole to the plasma membrane? ................................................................................................ 145 i. Visualization of centriole anchoring by electron microscopy ........ 145 ii. Perturbing mother appendages: an approach to understand centriole anchoring ................................................................................. 149 Acknowledgements: ....................................................................... 152 5. DISCUSSION ........................................................................................... 153 5.1. CENTROSOMES – AN EVERGREEN TOPIC FOR CELL BIOLOGY .................. 153 5.2. CENTRIOLE ELIMINATION AS A MECHANISM TO CONTROL CENTRIOLE NUMBER ....................................................................................................................... 154 20 5.3. THE MECHANISM OF CENTRIOLE ELIMINATION IN STARFISH OOCYTES ....... 155 5.3.1. Starting hypothesis ....................................................................... 155 5.3.2. Molecular characterization of centriolar proteins in starfish ......... 156 5.3.3. Live-imaging with centriole molecular markers ............................ 157 5.3.4. A mechanism for centriole elimination in starfish oocytes ............ 158 i. The mother centriole is specifically transported to the cell cortex . 159 Mother centriole transport is possibly dependent of a cytoplasmic gradient established upon nuclear envelope breakdown ..................... 161 ii. The mother centriole is anchored to the cell cortex ...................... 162 iii. Future directions: the mother centriole is directly anchored to the plasma membrane through mother appendages .................................... 164 5.3.5. Hypothesis: why only the mother centriole can move and anchor? .................................................................................................................... 165 i. Future directions: how to evidence the parallels of mother centriole transport between starfish oocytes and ciliated cells? ............................ 167 ii. Future directions: how to evidence the parallels of mother centriole anchoring between starfish oocytes, T-cells and ciliated cells? .............. 168 5.3.6. Hypothesis: how is the daughter centriole inactivated at the end of meiosis? ...................................................................................................... 168 i. How does the daughter centriole lose its microtubule nucleating activity? ................................................................................................... 170 ii. Are heterologous daughter centrioles equally inactivated at the end of meiosis? .............................................................................................. 171 5.3.7. A hypothetical model for the molecular mechanism of centriole elimination in starfish oocytes ..................................................................... 173 5.3.8. A general hypothesis: how to eliminate centrioles during female meiosis in animals ...................................................................................... 174 5.4. HYPOTHESIS: IMPORTANCE OF THE CENTROSOME DURING MEIOSIS ......... 176 5.4.1. Does an acentriolar vs. centriolar spindle correlate with an internal vs. external oocyte maturation? .................................................................. 176 5.5. HYPOTHESIS: A PREFERENTIAL INHERITANCE OF CENTROSOMES – LESSONS FOR STEM CELLS .............................................................................................. 178 6. APPENDIX ................................................................................................ 181 21 6.1. HOMOLOGS FOR CENTRIOLAR PROTEINS IN DIFFERENT ORGANISMS ....... 181 6.2. SUMMARY TABLE FOR CATEGORIES ....................................................... 182 6.3. PROTEIN ALIGNMENT ............................................................................ 182 6.3.1. Centrin-2 ...................................................................................... 182 6.3.2. Poc1 ............................................................................................ 183 6.3.3. Odf2 ............................................................................................. 184 6.3.4. Chibby ......................................................................................... 185 6.4. CHIBBY LOCALIZATION IN MII ................................................................ 186 REFERENCES ............................................................................................. 187 22 23 ABBREVIATIONS “Never memorize something that you can look up.” – Albert Einstein 1-MA: 1-methyladenine CytoD: cytochalasin D EM: electron microscopy LatB: Latrunculin B MI: meiosis I MII: meiosis II mGSC: male germ stem cell MTOC: microtubule organizing center NEBD: nuclear envelope breakdown PB/PBs: polar body/polar bodies PBI: first polar body PBII: second polar body PCM: pericentriolar material γ-TuRC: γ-tubulin ring complex 30 activation have been described for multiple systems (Debec et al., 2010; Doxsey, 2001; Piel et al., 2001). Moreover, centrosomes have a conserved role (from mammals to yeast) in spindle positioning: they nucleate astral microtubules, which are essential to correctly position the mitotic spindle (see fig. 1.3) (Doxsey, 2001). Importantly, during interphase, the centrosome is the major MTOC (see fig. 1.3). It is actively positioned at the center of the cell, and creates a microtubule cytoskeleton that defines the “cell’s coordinate system” for cell polarity and vesicle trafficking. Moreover, this active positioning defines the localization of other organelles, as the nucleus and the Golgi, to which the centrosome is attached (Bornens, 2012). 1.3. Centrosome structure As mentioned previously, a centrosome is composed of two orthogonally arranged centrioles, surrounded by a microtubule-nucleating matrix, the PCM (fig. 1.4). Next, I will provide a detailed description of each of these components. In this section, I will focus on the centriole structure and briefly discuss the formation of the PCM. A description of duplication and centriole assembly will be provided in later sections. I will also mention how centriole structure and molecular Figure 1.4: A centrosome is formed by two centrioles (one mother and one daughter centriole), surrounded by PCM. Only the mother centriole has two characteristic sets of appendages (distal and subdistal). 31 composition are so well conserved among organisms, and how this can provide insights into the evolution of this organelle. 1.3.1. Centriole – a well conserved structure across eukaryotes In the early 1950s, with the advent of electron microscopy (EM), the first centriole ultrastructures exposed the hidden complexity of the centriole. In one of these pioneering studies, Harven and Berhnard showed beautiful transversal and longitudinal sections of centrioles in different cells, revealing the well-conserved 9-fold symmetry of the centrioles and the parallel “tubules” that compose this hollow cylinder (Harven and Bernhard, 1956). Today it is well established that a centrosome has two centrioles, orthogonally arranged. The canonical centriole has 9-triplets of microtubules, which determine its symmetry, a diameter of 250 nm and a length that spans from 150 to 500 nm, depending on the cell type (fig. 1.5) ((Winey and O’Toole, 2014). Unrelated species such as mammals, paramecia, tetrahymena and clamydomona have microtubule triplets (CarvalhoSantos et al., 2010). However, there are exceptions to this rule: D. melanogaster and C. elegans feature microtubule doublets and singlets, respectively (fig. 1.5). Curiously, D. melanogaster sperm cells have triplets (Azimzadeh and Marshall, 2010; Winey and O’Toole, 2014). Figure 1.5: Each centriole that composes a centrosome is formed by 9-triplets of microtubules arranged in a cylindrical shape. Mother and daughter centrioles are represented – mother centriole has two sets of appendages. Each microtubule triplet is formed by A-, B-, C-tubules. The canonical centriole has 9 triplets of microtubules (as in the human centriole), but exceptions exist: D. melanogaster has 9 doublets, and C. elegans 9 single microtubules. Scale bar: 100 nm. Pictures adapted from (Brito, et al. 2012) – schematic, and (Winey and O’Toole, 2014) – EM pictures. 32 One centrosome has two different centrioles: one mother and one daughter (fig. 1.4 and 1.5). The mother centriole has two characteristic sets of appendages, distal and subdistal, which decorate one of its ends (fig. 1.4 and 1.5). The distal appendages are involved in ciliogenesis (see section 1.5) and are therefore conserved across eukaryotes. In contrast, the subdistal appendages are a new structure that only appeared in vertebrates. These appendages are involved in microtubule anchoring during interphase and they control basal body orientation during ciliogenesis. Similar structures, the basal feet, seem to take their function in other metazoans (Azimzadeh, 2014). Studies using different methodologies (immunogold labeling, super-resolution microscopy, immunofluorescence studies) mapped the localization of centriolar proteins within the centriole (Brito et al., 2012). These studies very clearly show that mother and daughter centrioles are different in terms of protein composition (fig. 1.6). Moreover, protein localization is conserved among species, with multiple homologs localizing consistently in multiple species. For simplicity, I will use the names of the human proteins for the rest of my thesis, but the equivalent homologs can be found in several different studies as (Brito et al., 2012) (see Appendix section 6.1). Figure 1.6: Ultrastructural localization of human centriolar markers in a fully mature centrosome. "(a) Electron micrograph of a longitudinal section of a centrosome isolated from human lymphoblastoma cells (KE37 cell line). (b) Schematic representation of the picture shown in (a). Several proteins are mother specific and compose the mother appendages (Cep164, Cep170, e-tubulin, EB1, Kif24, Ninein and Odf2). Conversely, Sas6 (in vertebrate cells) and Centrobin only localize to the daughter centriole. Figure and legend adapted from (Brito et al., 2012). 33 1.3.2. PCM - constitution and formation As mentioned above, a centrosome has two centrioles, which are surrounded by PCM. Boveri was the first to observe the PCM, and also the first to correctly characterize its main function. Indeed, the PCM works as the “centroplasm”, from where the microtubules are nucleated (reviewed by Scheer, 2014). Gould and Borisy experimentally showed for the first time how isolated PCM can nucleate microtubules in vitro (Gould and Borisy, 1977). Initially, PCM was merely described as an amorphous electron-dense cloud around the centrioles that would expand in size during mitosis. Nowadays, the PCM components start to be better known and described, and advances in super-resolution microscopy revealed the curious PCM organization (Fu and Glover, 2012; Lawo et al., 2012; Mennella et al., 2012; Sonnen et al., 2012). During interphase, the PCM is organized in well-defined concentric layers of proteins (as CEP192, CDK5RAP2, γ -tubulin) around the centrioles, which constitutes the PCM matrix (fig. 1.7). In this case, γ-tubulin, with a microtubule Figure 1.7: PCM organizes around the centrioles. The PCM is organized in two major layers: PCM fibers and PCM matrix (here represented separately to help visualization). PCM fibers comprise the elongated coiled-coil proteins pericentrin/pericentrin-like protein (PLP) and Cep152/Asl. PCM matrix contains: Cep192, CDK5RAP2, γ -tubulin. Picture modified from (Mennella et al., 2014). 34 nucleating function, localizes in the outer layers. Pericentrin and CEP152 constitute the PCM fibers and do not follow this concentric organization, but a more fiber-like organization (fig. 1.7) (Lawo et al., 2012; Mennella et al., 2012, 2014). Pericentrin appears to be at the top of the cascade that leads to PCM assembly, recruiting CDK5RAP2 and γ-tubulin (Mennella et al., 2014). Upon entry into mitosis, more PCM components are recruited, expanding the PCM “cloud”. Cell cycle kinases, such as PLK1 and Aurora-A, have a role in PCM expansion. PLK1 phosphorylates PCM components, such as Pericentrin, leading to the accumulation of more PCM components. Aurora-A acts more downstream adding microtubule-associated proteins to the structure. As phosphorylation leads to a PCM extension during mitosis, dephosphorylation of its components, in reverse, causes PCM dissolution at the end of mitosis (Mennella et al., 2014; Woodruff et al., 2014). Hence, the PCM assembly/disassembly is linked to the cell cycle, acting as the major factor to coordinate microtubule assembly. 1.3.3. Protein composition reflects phylogenetic relationships between eukaryotes As mention before (see section 1.2.), it is likely that the centrosome first emerged in evolution as a basal body, the cilia platform, whereas its MTOC function is likely to be a function acquired subsequently. Remarkably, centriole composition supports this idea, with a characteristic set of ancestral proteins, appearing in all species that form cilia at any point of their life (fig. 1.8). This core of proteins is conserved at least in four major eukaryotic groups: plantae (including species as Clamydomonas reinhardtii and Physcomitrella patens, a type of moss), excavata (Trypanossoma brucei, Naegleria gruberi), chromalveolata (Paramecium tetraurelia, Tetrahymena termophila) and holozoa (including Drosophila melanogaster, Caenorhabditis elegans and vertebrates) (fig. 1.8). A single species from the Fungi group, B. denbrobatidis also contains this set of ancestral proteins – fig. 1.8 (Carvalho-Santos et al., 2011; Hodges et al., 2010). 35 Among these ancestral set, several proteins (such as SAS6, SAS4, CEP135 and CEP164) are involved in the centriolar 9-fold symmetry, microtubule assembly and appendage formation, which suggests that these proteins emerged Figure 1.8: Distribution of centriolar and centrosomal proteins among eukaryotes. Protein homologs were identified for 45 eukaryotes (29 ciliated species (white) and 16 non-ciliated species (grey)). (+) Indicates presence of homologs. “Core” proteins are conserved ancestral centriolar proteins. “Centrosomal” proteins are associated with centrosomal functions. “Pole” proteins might have fulfilled a function in the ancestral spindle pole. “Controls” are proteins that are associated with general microtubule dynamics. “Ancestral” proteins are present among extant eukaryotes. “Holozoan” proteins have a restricted presence in holozoa (Metazoa and M. brevicollis). The asterisk indicates sequence drift of core and centrosomal proteins in C. elegans; divergent homologs known in the literature but not identified by their approach are highlighted with a pink border. Picture and legend adapted from (Hodges et al., 2010). 36 early in the eukaryotes and established the centriolar structure (Carvalho-Santos et al., 2011; Hodges et al., 2010). In holozoa, in which centrioles additionally serve as MTOCs, one can additionally find specific proteins associated with PCM assembly and microtubule nucleation, the holozoa set (fig. 1.8). This set includes proteins associated with PCM assembly and microtubule nucleation, as CEP192/Spd-2 (Azimzadeh, 2014; Carvalho-Santos et al., 2010; Hodges et al., 2010). Therefore the appearance of this group of proteins likely contributed to the “dual life” of the centrosome. Interestingly, PLK4, the master initiator of centriole duplication, appears also to be restricted to the holozoa group, while other PLK family members, such as PLK1, are identified in many other groups of the tree of life. This suggests that PLK4 possibly had its origin in the duplication of a PLK1-like ancestor, which became specialized in centriole duplication in the holozoa group (Carvalho-Santos et al., 2010). In summary, centrosomes/basal bodies are highly conserved structures with a conserved protein composition. This protein composition is consistent with the phylogenetic relationships established among organisms, which likely indicates that a centriolar structure was already present in their common ancestor. The study of molecular composition also allows inferring its ancestral and newly acquired functions during the evolutionary pathway. 1.4. Control of centriole number during cell division 1.4.1. Cell division and DNA duplication Cell division is the basis of cell continuity: cells duplicate and transmit their genetic information to the next generation. In unicellular organisms, cell division is reproduction; in multicellular organisms, cell division produces complexity, in which each cell is specialized for a specific task. 37 There are two types of cell divisions: to multiply their number, cells follow the mitotic process; to form gametes, and to reduce their chromosome content, cells follow the meiotic process (see section 1.6.1). Both processes are preceded by an interphase that prepares the cell for division, and comprises three distinctive phases: G1, S and G2. During G1, the cell grows in size, synthesizes multiple RNAs and proteins that are necessary for the following cell division. DNA duplication occurs during S-phase. DNA duplication is semiconservative, i.e. the parental DNA strands are used as a template for the synthesis of the new DNA strand. DNA has to be licensed at the end of the previous division cycle, by pre-replicative complexes, which localize at the regions of replication origin, and allows the strand to be duplicated. These complexes are displaced when duplication starts, which turns the DNA to an unlicensed state until the next cell cycle (Nishitani and Lygerou, 2002). This assures that DNA is only duplicated once per cell cycle, and guarantees maintenance of genomic content in the long term (Lodish, 2008). At last, the final preparations for cell division are made during G2-phase. After this, a cell enters either mitotic or meiotic division. 1.4.2. Cell division and centriole duplication i. Centriole+duplication+is+semi7conservative+and+occurs+once+ per+cell+cycle+ Although Boveri already described the centriole as an organelle able to duplicate, this duplication was first described morphologically using EM (Kochanski and Borisy, 1990; Robbins et al., 1968; Vorobjev and YuS, 1982). Briefly, the pair of centrioles enters a new cell division (G1) in a disengaged position, i.e. centrioles are still connected, but no longer in the orthogonal position. During S-phase, each pre-existing centriole duplicates, by “growing” a new centriole, called the “pro-centriole”, in an orthogonal position, re-establishing the characteristic orientation. Pro-centrioles then elongate during G2 until prometaphase. The centriole duplication resembles DNA duplication in four ways: i) it occurs during S-phase, ii) centriole duplication is semiconservative – the pre-existing 38 centrioles are not destroyed, but are instead used as a platform for the emerging centriole (Kochanski and Borisy, 1990). Then, iii) only licensed centrioles can duplicate. Centrioles are licensed during mitosis by PLK1 modification (Wang et al., 2011) and by disengagement through separase and also promoted by PLK1(Tsou and Stearns, 2006; Tsou et al., 2009). Finally, iv) a single site of origin ensures that centrioles can duplicate only once per cell cycle (concept adapted from Fırat-Karalar and Stearns, 2014). It is not yet understood how a single site of origin for centriole duplication is assigned. However, it is known that the levels of PLK4, the main regulator of centriole duplication, are tightly regulated during the cell cycle, which is essential to avoid centriole overduplication (Fırat-Karalar and Stearns, 2014). Other proteins, as SAS-6 and STIL, also stimulate centriole over-duplication and are therefore also tightly regulated (Azimzadeh and Marshall, 2010; Fırat-Karalar and Stearns, 2014). All this then assures that, similar to DNA, centrioles are duplicated only once per cell cycle, maintaining centriole number over generations. Moreover, at the beginning of a new cell division only the pre-existing centrioles, the mother centrioles, are licensed and therefore only these two can duplicate. The new centrioles, pro-centrioles or daughter centrioles, are blocked for duplication, because they are not yet licensed. In the next section, I will provide more details about the mechanism of centriole duplication, including the formation of the cartwheel (the base that confers the centriole its 9-fold symmetry), and how microtubules are then positioned around this cartwheel, and centrioles are elongated. ii. Centriole+duplication:+the+onset+and+cartwheel+formation++ At the beginning of a new cell cycle, a pair of disengaged centrioles starts its duplication during late G1/S-phase. Interestingly, a core of only five proteins, identified in C. elegans, is essential for centriole duplication: ZYG-1, SPD-2, SAS4, SAS-5 and SAS-6 (i.e. homologs of the corresponding human proteins: PLK4, CEP192, CPAP/CENJ, STIL, hSAS-6). While this core of five proteins is sufficient for centriole duplication in C. elegans, in D. melanogaster and human other factors are also essential – for example Asterless in the case of D. melanogaster, 39 and CEP152 and CEP135 in the case of human (Azimzadeh and Marshall, 2010; Hirono, 2014; Nigg and Raff, 2009; Strnad and Gönczy, 2008). Centriole duplication starts with the recruitment of PLK4 to the origin site, by CEP152 and CEP192, which then recruits SAS-6 (Fırat-Karalar and Stearns, 2014). Lettman MM. and colleagues described, that at least in C. elegans, the recruitment of SAS-6 would occur by direct interaction with the C. elegans PLK4 (ZYG-1) (fig. 1.9) (Lettman et al., 2013). The first morphological mark of a nascent daughter centriole is the presence of a cartwheel, formed by a central hub from which nine spokes radiate (fig. 1.9, 1.10 and 1.11). The cartwheel is evolutionarily conserved among eukaryotes and Figure 1.9: Model of the centriole duplication cycle for a human centriole. Briefly, (a) centriole assembly is triggered by Plk4, Cep152 and Cep135. (b) Cartwheel assembly by SAS6. (c) Centriole elongation starts, microtubules assemble around the cartwheel d) Daughter centriole elongation is completed. Centrosome separation, which allows the assembly of a bipolar spindle in mitosis, takes place in late G2-phase. (e) During mitosis, centrioles disengage and lose their orthogonal configuration, a process mediated by Plk1 and separase. (f) The daughter centriole is now a mother, upon full maturation. Centriole duplication and cell cycle stages are indicated at the top and bottom of the image, respectively. Key molecules are shown. Proteins represented in black indicate temporal and spatial localization during centriole assembly; proteins represented in red indicate moment of their displacement from the daughter centriole; proteins represented in green or orange indicate increasing or decreasing levels at the daughter centriole, respectively. Picture and legend adapted from (Brito et al., 2012). 46 (Paintrand et al., 1992; Vorobjev and YuS, 1982). Two sets of mother appendages, distal and sub-distal (fig. 1.14 inset 1 and 2, respectively), can be observed at the distal side of a mother centriole, following the conserved centriole 9-fold symmetry. However, distal and subdistal appendages carry out different functions. The distal appendages are essential during ciliogenesis and they mediate the centriole anchoring to the plasma membrane. The sub-distal appendages anchor microtubules during interphase and are important for cell polarity. The different functionality reflects the type of proteins that accumulate in each set of appendages. Sub-distal appendages accumulate microtubule-associated proteins as Ninein, CEP170 and ε-tubulin, while distal appendages accumulate proteins such as CEP164 and Odf2 (see protein localization in fig. 1.6, section 1.3.1) (Fu et al., 2015; Jana et al., 2014; Tateishi et al., 2013; Winey and O’Toole, 2014). Odf2 is one of the few proteins known to localize to both set of appendages and to be necessary for anchoring to the plasma membrane during ciliogenesis (Ishikawa et al., 2005). 1.5.2. Centriole maturation – the process At the start of a new cell cycle (G1), the daughter centriole is now licensed to duplicate, and eventually become a mother centriole. One could say that the centriole pair is now formed by a “grandmother”, which was already a mother in a previous cycle, and a “new” mother, which is now licensed to duplicate for the first time (fig. 1.15). Thus, the now licensed centriole goes through maturation, which involves the acquisition of PCM and mother appendages (Brito et al., 2012; Kong et al., 2014; Winey and O’Toole, 2014). By the end of the following mitosis, the two mothers (grandmother and new mother) have appendages, and each one of them also has a daughter that formed in S-phase. In other words, three generations of centrioles are present in the same cell, in a regular somatic cell division cycle (fig. 1.15). Note how the formation and elongation of a new centriole require only one cell cycle, whereas its maturation into a new mother only occurs during the following cell cycle. 47 The exact time of appendage assembly is likely to be dependent on the cell type (Kong et al., 2014). Still, PLK1 activity appears to be required throughout S and G2 of the centriole’s second cell cycle, and is essential for appendage formation in the new mother (Kong et al., 2014). First, appendage proteins start to accumulate at the distal part of the centrioles, and then eventually become appendage structures (Kong et al., 2014; Lange and Gull, 1995). Several studies show that protein accumulation at the mother appendages occurs in a hierarchical way (Ibi et al., 2011; Tanos et al., 2013; Tateishi et al., 2013). Odf2 accumulation in the new mother starts at G2/M transition (Kong et al., 2014; Lange and Gull, 1995) and in fact, Odf2 appears to be one of the most upstream components in the appendage assembly cascade. Consistently, its depletion completely eliminates the formation of either distal or subdistal mother appendages (Ishikawa et al., 2005). How Odf2 is recruited to the new mother centriole is unknown, but the protein 4.1R is likely involved, as upon depletion, Odf2 localization to the centriole is perturbed (Krauss et al., 2008). Odf2 is then required for the proper recruitment of several other appendage proteins as Ninein and CEP164 (Ibi et al., 2011; Tateishi et al., 2013). Figure 1.15: There are three generations of centrioles in a dividing cell. At G1, a disengaged centriole pair has a recently licensed “new” mother centriole and a “grandmother” centriole. During S-phase, this pair duplicates, and a new daughter centriole forms in an orthogonal position. The centrosome formed by the “new” mother and a daughter centriole is called the “younger” centrosome, whereas the “older” centrosome, contains the grandmother and a daughter centrioles. 48 As a consequence of the molecular changes outlined above, the centrioles need at least 1.5 cycles to become a functional mother. A daughter centriole, which just duplicated, cannot become a mother (Hoyer-Fender, 2010; Kong et al., 2014; Vorobjev and YuS, 1982), possibly because it needs to disengage first (Kong et al., 2014; Wang et al., 2011). 1.5.3. The role of mother appendages in the anchoring to the plasma membrane The distal mother appendages mediate the direct anchoring to the plasma membrane during ciliogenesis. However, recently another example of a direct anchoring to the plasma membrane mediated by appendages was described during T-cell activation. I will describe the two processes in the following sections. i. Ciliogenesis++ The mother centriole is required for cilia/flagella formation: it binds to the plasma membrane directly through the mother distal appendages, and once anchored, the mother centriole becomes a basal body, i.e. the platform from which cilia or flagella are formed. Cilia and flagella follow the same 9-symmetry as the mother centriole base, but instead have microtubule doublets. Motile cilia/flagella have normally a microtubule doublet in the center, which is not normally present when cilia are non-motile (Reiter et al., 2012). In order to become a cilia/flagella base, the mother centriole first has to migrate to the plasma membrane, where it then anchors. Details of this process are not entirely clear, but vesicular trafficking appears to be involved. Indeed, one of the current models for ciliogenesis assumes that vesicles dock to the distal appendages of mother centriole, while this is still located in the cytoplasm (fig. 1.16) (Sorokin, 1962, 1968; Sung and Leroux, 2013). The mother distal appendage proteins Odf2 and CEP164 are responsible for the interaction with Rab11 and Rab8 vesicles, respectively (Hehnly et al., 2012; Schmidt et al., 2012). More recently, Chibby was found to be involved in mediating the interaction between CEP164 and the Rab8 vesicles (Burke et al., 2014). Subsequently, the vesicles start to fuse with each other, forming a large vesicle, 49 called the ciliary vesicle, at the distal end of the mother centriole. Then, the mother centriole is transported to the plasma membrane along the vesicle transport pathway, where it anchors upon ciliary vesicle fusion with the plasma membrane (fig. 1.16) (Reiter et al., 2012; Sung and Leroux, 2013). How the mother centriole is transported to the plasma membrane is an unsolved question, but elements of the cytoskeleton are likely involved. The process is best described in multiciliated cells. For example, in multiciliated oviducts, defects in the centriole anchoring are observed upon depolymerization of actin, while microtubule depolymerization has no effect (Boisvieux-Ulrich et al., 1989, 1990; Dawe et al., 2007). It has been shown that the apical surface of multiciliated Xenopus embryonic cells are enriched with a dense meshwork of actin, which contributes to basal body spacing and docking, and coordination of cilia beating (Antoniades et al., 2014; Werner et al., 2011). Cytoplasmic microtubules also contribute to this process, by forming a network that polarizes locally the basal bodies (Werner et al., 2011). ii. Mother+centriole+movement+during+T7cell+activation+ Cytolytic immune cells such as cytotoxic T-lymphocytes kill infected cells by releasing lytic enzymes, which then induce apoptosis. Upon contact with the Figure 1.16: Model for ciliogenesis. (1) Mother centriole associates with Rab11 vesicles, which bind to the distal appendages. (2) and (3) Rab8 vesicles associated with the previous vesicles, forming the ciliarly vesicle. The mother centriole is transported to the plasma membrane, hitchhiking the secretory pathway. A pro-axoneme might start to extend while still inside the cell. (4) Vesicles fuse with the plasma membrane, and mother centriole remains anchored directly via its mother appendages. Cilium fully extends. Figure and legend modified from (Sung and Leroux, 2013) 50 target cell, an immunological synapse forms, and the T-cell undergoes polarization and re-organization of its microtubule cytoskeleton (fig. 1.17). This re-organization is accomplished by centrosome migration to the center of the immunological synapse. Cytolytic granules then move in a microtubule-minusend directed motion towards the contact site, and are delivered to the target cell (fig.1.17) (Stinchcombe et al., 2006, 2011). Microtubules and actin likely play a role in centrosome movement to the plasma membrane, mediated by dynein and formin respectively (Stinchcombe and Griffiths, 2014). Centrosome movement and association with the plasma membrane are very similar to cilia formation. Indeed, recently Stinchcombe and colleagues identified that in this case mother centriole connects to the plasma membrane through the mother appendages (personal communication). 1.5.4. Differential centrosome inheritance and asymmetric division As mentioned (see section 1.5.2), three generations of centrioles co-exist in a somatic cell. During metaphase, at one pole there is the older centrosome (which contains the “grandmother” centriole and its daughter), and at the other pole the younger centrosome (which contains the “new” mother and its daughter) Figure 1.17: Centrosome polarization in T-cells, after interaction with target. Centrosome is shown in red, T-cell in blue, and target cell in light brown. (i) When a T-cell meets its target, the T-cell’s centrosome moves towards the contact site. The microtubule network (black lines), including microtubule-associated organelles, such as secretory vesicles (yellow) and cytolytic secretory granules (orange) are reorganized. (ii) Tight centrosome localization at the plasma membrane aligns microtubules, which creates a flow of cytolytic secretory granules towards the contact site. Signalling pathways are activated at the contact site (red arrow). Figure and legend adapted from (Stinchcombe and Griffiths, 2014). 51 (fig. 1.15). Therefore, at the end of cell division, each daughter cell inherits one of these centrosomes. Due to this asymmetry in centrosome age, a cell division is in fact always asymmetric. In the standard definition of an asymmetric division two daughter cells are produced that differ regarding their cell fate and/or size. The prime example of this is the stem cell division: stem cells undergo asymmetric divisions and produce a daughter cell that will differentiate, while the other daughter cell maintains its stem cells status and pluripotency. Is there a link between centrosome age and cell fate? Centrosomes are important for spindle orientation and therefore for establishing the symmetry or asymmetry of cell division. Indeed, multiple studies show that centrosome age determines the division axis of stem cells, which consequently establishes which centrosome each cell inherits. The first report of this was in the male germ stem cell (mGSC) line of D. melanogaster. Hub cells are important in maintaining the mGSCs pluripotent environment (fig. 1.18) (Yamashita et al., 2007). The mGSC divides, maintaining the pluripotent cell close to the hub cells, while the other daughter cell proceeds into spermatogenesis. Interestingly, the older centrosome, due to its higher microtubule nucleating activity, is constrained to the adherens junction between the hub cell and the mGSC during interphase. Consequently, during spindle assembly, the younger centrosome moves distally, defining the future axis of cell division. As a result, the pluripotent stem daughter inherits the older centrosome, while the differentiating daughter inherits the younger centrosome (fig. 1.18) (Pelletier and Yamashita, 2012; Yamashita et al., 2007). However, not always the progenitor cell keeps the older centrosome. In the same organism, in dividing neuroblasts the pluripotent cell keeps the younger centrosome, while the differentiating daughter cell, which will give rise to the ganglion mother cell, receives the older centrosome (fig. 1.18). This asymmetry is established earlier, during interphase, when the neuroblast has a single pair of centrioles localized apically. The mother centriole rapidly loses its PCM and its apical localization, while the daughter centriole maintains its microtubule nucleating activity and remains connected apically, stabilized by the microtubule aster (fig. 1.18) (Conduit and Raff, 2010; Rebollo et al., 2007; Rusan and Peifer, 2007). Interestingly, Centrobin, a daughter centriole marker, is essential for 52 preserving PCM at the daughter centriole, conserving its MTOC potential during interphase (Conduit and Raff, 2010; Januschke et al., 2013). In the above cases, the connection of either the older or the younger centrosome, respectively in Drosophila mGSCs or neuroblasts, to the stem cell niche during interphase, determines the cell division axis, directing the stem cell into an asymmetric division program. However, from these examples, it is not clear whether there is a functional link between the type of centrosome that is Figure 1.18: Differential inheritance of centrosomes during stem cell division of D. melanogaster. (A) Male germinal stem cells (mGSC) (yellow) are associated with hub cells (blue). The older centrosome has a higher microtubule nucleating activity and remains confined to the adherens junction between these two cells. As a consequence, the older centrosome is preferentially inherited by the progenitor stem cell, whereas the goniablast (pink) receives the younger centrosome. (B) During interphase, the daughter centriole retains PCM and microtubule nucleating activity. Therefore it remains at the apical cortex. In contrast, the mother centriole loses PCM and microtubule nucleating activity. When the cell enters division, the progenitor cell (violet) inherits the younger centrosome, whereas the ganglion mother cell inherits the older centrosome (grey). 53 inherited and cell fate. In the mouse neural cortex, the apical progenitor cells (also known as radial glia progenitor cells) upon asymmetric division inherit the older centrosome (Wang et al., 2009). The older centrosome retains membrane components, reminiscent of the cilia, which help the progenitor cell to faster reform the cilia than the daughter cell with the younger centrosome. Thereby, Paridean and colleagues showed for the first time the functional importance of inheriting one specific centrosome: the two daughter cells sense and respond differently to the environment, which likely confines their cell fate into a progenitor type cell or into the neuronal differentiation pathway (Anderson and Stearns, 2009; Paridaen et al., 2013). 1.6. Centriole elimination in gametogenesis As explained in the above sections, centriole duplication and segregation maintains centriole number in somatic cells in a highly controlled manner and with high fidelity. However, deviations exist from the canonical centriole cycle. A prominent example is gametogenesis, where this process is essential for sexual reproduction of all animal species. During meiosis, oocytes and spermatozoa alter their centrosome activity and composition in a complementary manner. Spermatozoa lose their PCM components, but keep the centriolar structure because of their essential role in cilia formation, and consequently, sperm movement. In contrast, by the end of meiosis, all female gametes lose the centriole structure. Therefore after fertilization, the first embryonic centrioles are a sole paternal contribution. Yet, oocytes accumulate a vast reservoir of proteins necessary for the future embryonic development, and centrosomal proteins are not an exception. Intriguingly, upon fertilization, sperm reduced centrioles use this maternal centrosomal protein pool to fully recover its functionality (Fabritius et al., 2011; Manandhar et al., 2005) . Again, Boveri already provided us multiple hints about centriole elimination in the beginning of the 19th century. He was the first to describe the lack of centrioles in sea urchins eggs and provide evidence that the embryonic centriole originates from the sperm. Boveri also observed how centrosome number is 54 important during embryogenesis: an extra number of centrosomes that results from polyspermy (fertilization by more than one sperm) creates multipolar spindles leading to aneuploidy, with consequent defects in embryonic development (Maderspacher, 2008; Moritz and Sauer, 1996; Scheer, 2014). Following Boveri’s pioneering work, multiple studies attempted to characterize the centriole cycle during oocyte and sperm meiosis, referred to as oogenesis and spermatogenesis, respectively. In the next section I will provide a short summary of centriole reduction in spermatozoa, and will focus in more detail on the mechanism of centriole elimination in the oocyte. But first, I will just give a brief introduction to the major steps of meiosis. 1.6.1. Meiosis overview Meiosis was discovered by Edouard Van Beneden in 1883-84, using the horse roundworm Ascaris megalocephala (Hamoir, 1992). Meiosis involves the same steps as mitosis, but in this case the chromosomes still duplicate a single time in S-phase (4N), but undergo two successive divisions – Meiosis I (MI) and Meiosis II (MII). Hence, four daughter cells originate, each one of them with a haploid set of chromosomes (N). Both female and male gametocytes undergo meiosis, and male meiosis generates daughter cells of equivalent size (sperm cells or spermatozoa) - fig. 1.19; female meiosis is extremely asymmetric, originating the oocyte and polar bodies (PBs) – fig. 1.20 (Lodish, 2008). Briefly, the steps of meiosis are: MI starts with Prophase I, normally the longest phase of meiosis, which comprises different steps (only named for future reference): chromosomes get condensed at leptotene (from the greek leptonema, “thin treads”); then, at zygotene (greek zygonema, “paired threads”), each chromosome finds its “pair”, the other homologous chromosome. At the pachytene stage (pachynema, "thick threads"), crossing over occurs between the pair of homologous chromosomes. As a consequence, non-sister chromatids exchange genetic material, creating small new rearrangements in the DNA composition; at diplotene (diplonema, “two threads”), all crossing overs are established. Most oocytes are arrested at this stage, until fertilization or hormonal stimulation resume the process. Finally, prophase I is completed with diakinesis 55 (“moving through”), where chromosomes acquire their maximum condensation, and NEBD occurs (Lesch and Page, 2012). In metaphase I, the paired homologous chromosomes are aligned at the equatorial plate. In anaphase I, the pairs of homologous chromosomes separate, and each set of chromosomes migrates to its respective pole of the daughter cell. At telophase I, the two daughter cells are individualized, each one of them still diploid, since each chromosome still has two chromatids. A second division starts (MII), but no S-phase occurs and thus no duplication of DNA takes place. MII is similar to a mitotic division in terms of chromosome configuration and follows the same steps as MI. Thus, by telophase II, each cell receives a single set of chromatids and consequently four haploid cells are formed (fig. 1.19 and 1.20). 1.6.2. Centrosome reduction during spermatogenesis At the last stages of spermatogenesis, the primary spermatocyte (4N) undergoes meiosis, giving rise to four equivalent haploid spermatids (N) (fig. 1.19). These cells undergo a major differentiation, reducing all cell components to only those strictly indispensable for fertilization. DNA becomes highly condensed reducing nucleus size, the Golgi apparatus becomes the acrosomal cap and contains enzymes important for fertilization, centrioles organize the flagellar complex, and mitochondria accumulate at the neck of the spermatozoa and provide energy for the flagella movement (fig. 1.19) (Lodish, 2008). As mentioned, centrioles, which are essential for spermatozoa movement, are not eliminated in sperm cells, yet some degree of degeneration is observed. This process, termed centrosome reduction, involves loss of PCM proteins and microtubule nucleating activity, with some degree of centriole structure degradation. In some extreme cases, there is the complete elimination of the centriole and respective microtubule triplets. The extent of centriole degradation is variable between species; rodents are the only known species that completely eliminate their sperm centrioles (Manandhar et al., 2005). Moreover, differences also exist in the number of centrioles present in the sperm cells of different species. Next, I will provide an overview of the number and degeneration state of 62 Centriole elimination in this organism is delayed upon depletion of the germline helicase CGH-1, which associates with certain maternal mRNAs. As a consequence, centriole elimination is defective and oocytes still containing centrioles are fertilized, creating multipolar spindles in the embryos (MikeladzeDvali et al., 2012). Although a potential effector or mechanism was not identified, CGH-1 is the first protein known to be involved in centriole elimination. Taken together, although the species described above are not closely related, their centrioles are eliminated by the diplotene stage, during prophase I (fig. 1.21). Whether the underlying mechanisms are conserved is not known. ii. One+single+pair+of+centrioles+is+present+at+the+beginning+of+ meiosis:+the+sperm+centriole+intervenes++ The pulmonary snail Lymnea stagnalis has an alternative centriole cycle during meiosis – a single pair of centrioles is present at the beginning of MI. Consequently, the MI spindle organizes with a single centriole in each pole, with one of the centrioles being extruded into the PBI. The remaining centriole forms the MII spindle pole facing the outside of the cell, whereas the sperm basal body is positioned at the other spindle pole (Krioutchkova et al., 1994b). Although the formation of the MII spindle requires the sperm basal body’s intervention, the oocyte centriole pair is successfully extruded into the PBs and therefore eliminated from the egg cytoplasm. Figure 1.23: Schematic representation of a C. elegans gonad. The gonad can be subdivided into four regions: (1) proliferating germ cells; (2) germ cells in pachytene stage; (3) germ cells after pachytene; (4) germ cells in diplotene stage: centrioles are no longer visible at this stage. Note diakinesis oocytes are marked –1, –2, –3 prior to the spermatheca. Sheath cell nuclei are depicted in blue. Figure and legend adapted from (Mikeladze-Dvali et al., 2012). 63 iii. Centrioles+are+eliminated+at+the+end+of+meiosis:+spindle+ assembly+is+centriolar+ Not all oocytes organize an acentriolar spindle during meiosis. In fact, ultrastructure studies show that two pairs of centrioles are present at the beginning of meiosis in oocytes of echinoderms (such as sea urchin H. pulcherrimus, starfish P. pectinifera and arnurensis and sea-cucumber H. moebi) and bivalves (such as mussel M. edulis). Although only EM studies were performed in those species, it is described that during MI spindle formation, a pair of centrioles localizes at each pole of the MI spindle. The PBI is extruded with one pair of centrioles. Interestingly, during MII, single centrioles are observed at the MII spindle poles. A second PB (PBII) is extruded with one of the centrioles, and a single centriole remains in the mature egg (Kato et al., 1990; Longo and Anderson, 1969; Miyazaki et al., 2005; Nakashima and Kato, 2001). The remaining centriole is then eliminated, and the sperm provides the embryo with the first pair of centrioles (Saiki and Hamaguchi, 1998). How the remaining single centriole of the oocyte is eliminated remains an unsolved problem. In a next section, more information about the different centriolar behaviors observed in starfish oocytes will be detailed. The differences in timing of centriole elimination have direct consequences to the type of spindle assembly that is adopted: when centrioles are eliminated before NEBD, the spindle is acentriolar; when centrioles are maintained until the end of meiosis, centrioles organize the spindle. Although the type of spindle assembly, centriolar or acentriolar, has been identified for these species, we are far from understanding what dictates the timing of centriole elimination. 1.6.4. Scaling problems in oogenesis i. Spindle+positioning+at+the+cell+cortex+ Oocytes are among the biggest animal cells, which undergo highly asymmetric divisions in order to reduce their DNA content to a haploid set of chromosomes. Oocyte asymmetric division is functionally important because it preserves 64 nutrients and proteins, essential for early embryonic development. Moreover, only the mature egg is fertilizable: the PB cannot bind sperm due to its lack of microvilli, required for sperm entry (Brunet and Verlhac, 2011). Oocytes’ meiotic spindles are small and localize asymmetrically in close proximity to the oocyte plasma membrane, which ensures the extreme asymmetry of the division, so that just a small fraction of cytoplasm is lost during cell division (Chaigne et al., 2012; McNally, 2013). Although asymmetric spindle positioning occurs in all female oocytes, it can happen at different stages of meiosis. In many species, the entire nucleus moves to the animal pole before NEBD (e.g. starfish, sea cucumber, C. elegans, D. melanogaster), whereas in mouse oocytes, the nucleus remains in the center, and migration occurs only after spindle assembly (Fabritius et al., 2011; Miyazaki et al., 2005). Centrioles, which are retained until the end of meiosis in starfish and sea cucumber oocytes, localize close to the plasma membrane at the animal pole in the immature oocyte. The nucleus is then positioned to the animal pole in a microtubule-dependent mechanism. In starfish oocytes, this nuclear localization to the animal pole happens long before the resumption of meiosis (Miyazaki et al., 2000). In contrast, in sea cucumber, the long microtubules nucleated by the centrosome move the nucleus towards the animal pole, shortly after meiosis resumption (Miyazaki et al., 2005). Some studies suggest that centrioles, before elimination, are likely to position the oocyte nucleus in D. melanogaster. Indeed, it has been shown that the (multiple) centrioles grow microtubules, which push the nucleus from a central position until it reaches a lateral one (Zhao et al., 2012). In C. elegans, the nucleus is also localized to the cortex before NEBD, in a microtubule and Kinesin-1 dependent manner (Fabritius et al., 2011). In mouse oocytes, the mechanism of spindle localization to the plasma membrane is well characterized. Prior to meiotic resumption, the nucleus localizes to the center of the oocyte and only after NEBD and spindle formation, the MI spindle is transported towards the plasma membrane (Verlhac et al., 2000). In mouse oocytes, actin drives spindle transport (fig. 1.24 B). Indeed, the oocyte is filled up with an actin network with vesicles at the network’s branching points, which act as the organizing centers for this network. On the vesicle surface, actin nucleators (Spire-1 and -2 and Formin-2) and the motor protein 65 Myosin-Vb accumulate. Actin is polymerized from these vesicles, and Myosin-Vb generates pulling forces moving these vesicles towards each other. This altogether creates a highly dynamic actin network that serves as a substrate for spindle transport (Almonacid et al., 2014; Azoury et al., 2008; Holubcová et al., 2013; Schuh, 2011). Upon spindle formation, an actin cage forms surrounding the MI spindle, and Myosin-II localizes at spindle poles (fig. 1.24 B). The spindle is then transported to the closest cortex on the dynamic cytoplasmic actin network, driven by Myosin-II (Schuh and Ellenberg, 2008). Formin-2 regulation is likely associated with the symmetry-breaking event: before the resumption of meiosis, high levels of Formin-2 keep the nucleus at the center of the oocyte; upon resumption of meiosis, a sudden drop in the levels of Formin-2 changes the actin network organization, and consequently allows off-centering of the spindle (Azoury et al., 2008; Verlhac et al., 2000). As mentioned earlier, all oocytes position their spindle in close proximity to the plasma membrane before cell division. Moreover, the MI and MII spindle are Figure 1.24: Schematic representation of spindle positioning in a mouse oocyte (B) and respective comparison with somatic cell (A). (A) Somatic cell in metaphase of mitosis. Spindle is positioned by astral microtubules. Cortical F-actin anchors astral microtubules to the cortex and increases cell rigidity. (B) Mouse oocyte in metaphase I. A cytoplasmic and highly dynamic actin meshwork drives spindle positioning to the cell cortex. Cytoplasmic meshwork is nucleated from vesicles localizing at its branching points. By localizing at the spindle poles, myosin II likely drives this movement, by pulling on cytoplasmic F-actin (curved black arrows). Figure and legend adapted from (Almonacid et al., 2014). 66 orientated perpendicularly to the plasma membrane, a feature that is conserved in all oocytes. For example, in the worm C. elegans, the MI spindle lies first parallel to the cortex, and then rotates by dynein action, becoming perpendicular to the cortex. Although this rotation is less described in other species, spindles always have a perpendicular orientation before PB extrusion. This perpendicular orientation and the close proximity to the plasma membrane appear to be essential for correct PB extrusion (Fabritius et al., 2011). ii. Spindle+assembly++ In a mitotic cell, centrosomes organize the poles of the spindle, nucleating microtubule arrays that capture the chromosomes (fig. 1.24 A). In the presence of centrosomes, this centriolar spindle assembly pathway dominates. Nonetheless, chromosomes can also nucleate microtubules, contributing for spindle formation (Walczak and Heald, 2008). In the absence of centrosomes, chromosomes can promote the self-assembly of an acentriolar spindle. This was shown for the first time in Xenopus egg extracts: without centrioles, bipolar mitotic spindles are still assembled on sperm chromatin or even around DNA coated beads (Heald et al., 1996). In a simplified way, chromosomal microtubule nucleation depends on a Ran-GTP gradient, which is established around the chromosomes. Ran-GTP then activates several microtubule associated proteins, including TPX2, which recruits the Augmin complex that consequently recruits γTuRC, allowing microtubule nucleation from the chromosomes (Goshima et al., 2008; Gruss et al., 2001; Petry et al., 2013). Other proteins are also involved in the formation of the bipolar spindle: the kinesin Eg5 aligns the nascent microtubules, the motor XKlp1 pushes the microtubule away from the chromosomes, whereas dynein focuses the microtubule ends and contributes for spindle pole formation (Karsenti and Vernos, 2001; Walczak and Heald, 2008). Acentriolar spindle assembly mechanisms in live oocytes are not as well characterized as in the egg extract. However, in oocytes, acentriolar spindle assembly also occurs from non-centriolar MTOCs, and not only from the chromatin (Dumont and Desai, 2012). Some clarification of these processes arose from the identification of these non-centriolar MTOCs, particularly in D. melanogaster, frogs and mouse oocytes. 67 In frog oocytes, a disk-shaped MTOC (referred to as a transient microtubule array) assembles at the base of the nucleus, shortly after NEBD, and then migrates towards the animal pole. This MTOC is likely to collect the chromosomes and function as a precursor of the MI spindle, being subsequently remodeled into a bipolar spindle (Gard, 1992). In D. melanogaster, non-centriolar MTOCs appear de novo shortly before NEBD, organizing scattered microtubules asters, which then organize into a bipolar spindle. The kinesin Ncd localizes to the initial microtubules asters, and has a central role during MI spindle assembly, by re-shaping these microtubule asters into a bipolar spindle (Megraw and Kaufman, 2000; Sköld et al., 2005). Again, more data is available in mouse oocytes: several studies show the presence of multiple pericentrin and γ-tubulin-containing MTOCs, which form de novo shortly before NEBD (fig. 1.25) (Calarco, 2000; Carabatsos et al., 2000; Schuh and Ellenberg, 2007). These MTOCs are initially scattered throughout the oocyte, and converge towards each other and to the nuclear region after NEBD, forming a “sphere” of microtubules around the chromosomes (fig. 1.25 – I to III). This sphere of microtubules is progressively shaped into a MI bipolar spindle by Kinesin-5 and other molecular motors (fig. 1.25 – IV to VI) (Schuh and Ellenberg, 2007). Figure 1.25: Model of acentriolar spindle assembly in mouse oocytes. I-III Acentriolar MTOCs localize first dispersed in the oocyte’s cytoplasm. Upon NEBD, a “sphere” of microtubules organize around the chromosomes. IV-V: Acentriolar MTOCs start to cluster and a bipolar spindle is organized (VI), mediated by Kinesin-5. VI: Spindle further elongates. Other molecular motors are likely to be involved. Figure and legend adapted from (Schuh and Ellenberg, 2007). 68 As mentioned before, microtubule nucleation in frog extracts depends completely on the chromatin generated Ran-GTP gradients. However, in the formation of the MI spindle in mouse oocytes, Ran-GTP-dependent nucleation is not essential, yet contributes by accelerating the spindle assembly process (Dumont et al., 2007; Schuh and Ellenberg, 2007). Interestingly, the Ran-pathway is essential for MII spindle assembly, possibly by contributing speeding up the process, which occurs much faster than the MI spindle assembly (Dumont et al., 2007). In summary, one can consider that acentriolar assembly mechanisms are similar between egg extracts and meiotic oocytes: motor proteins are important for bipolarity, while chromosomes and associated Ran-GTP are a source of microtubule nucleation, yet not the only factor in meiotic oocytes. Even in cells forming acentriolar spindles, MTOCs are present. This suggests that either noncentriolar MTOCs or centrosomes typically accelerate the chromatin-mediated spindle assembly pathway during female meiosis. 1.7. Centriole elimination in starfish oocytes 1.7.1. Overview of starfish meiosis Starfish oocytes are large cells, with a diameter of 170 µm and when fully grown and arrested in Prophase I, contain a nucleus with 80 µm diameter (fig. 1.26 A). The large nucleus localizes closer to the animal pole, defining a clear axial asymmetry in the oocyte. Two pairs of centrosomes localize between the nucleus and the oocyte plasma membrane and actually hold the nucleus at the animal pole by long interphase microtubules (fig. 1.26 A) (Miyazaki et al., 2000). Shortly after meiosis resumption, NEBD occurs. Actin has a major role in the early events of meiosis: an actin shell, an Arp2/3-dependent structure composed of highly compacted branched actin, transiently accumulates at nuclear envelope and promotes fast fragmentation of the nuclear envelope (fig. 1.26 B) (Mori et al., 2014). Shortly after, an actin meshwork forms in the nuclear region, which collects the scattered chromosomes across the nucleus (fig. 1.26 C). This actin 69 meshwork is highly dynamic and contracts directionally towards the animal pole. Once the actin network delivers the chromosomes close to the animal pole, they are collected by the centrosomal microtubule asters, which then organize the MI spindle (Lénárt et al., 2005; Mori et al., 2011). The centrosomes organize the spindle (see section 1.6.3 iii) and the oocyte undergoes two consecutive meiotic divisions with the extrusion of two PBs. 1.7.2. Starfish oocyte as a model to study centriole elimination using molecular markers and live cell imaging i. Centriole+elimination+in+other+model+organisms+ Centriole elimination is a process characterized by a fast switch from a state including centrioles in the cytoplasm, to another without centrioles. Indeed, no “collapsing” structure has ever been observed. Moreover, it is likely a very fast process (Mikeladze-Dvali et al., 2012), which complicates its observation during the long prophase I, characteristic of all canonical model organisms (frogs, D. melanogaster, C. elegans, mouse). Figure 1.26: NEBD and chromosome congression during starfish meiosis. (A) Oocyte is arrested in Prophase I. (B) After hormone addition, the actin shell transiently accumulates at the nuclear envelope and promotes fast NEBD. (C) Actin meshwork forms inside the nuclear region, actin-rich regions organize around the chromosomes previously localized close to the nuclear envelope. (C) Actin meshwork contracts unidirectionally and chromosomes are transported towards the plasma membrane. 70 Consistently, so far no molecular mechanism has been defined for centriole elimination during female meiosis. For most model systems only the timing of centriole elimination is known, detected by the loss of centriolar proteins and subsequent disappearance of the centriolar structure. With the exception of the helicase CGH-1 in worms, no other protein was described to be involved in centriole elimination, and still its role remains unclear. ii. Centriole+elimination+in+starfish+oocytes+7+what+is+known+so+ far++ Starfish became a model system for meiosis research after the isolation and identification of 1-methyladenine (1-MA) (Kanatani et al., 1969), the hormone responsible for inducing resumption of meiosis, allowing experimentally controlled study of meiotic maturation. Starfish oocytes are a robust system that allows multiple and variable manipulations without affecting the viability of the oocytes. Centriole elimination was first investigated in this system more than two decades ago, because the transparency of the oocyte allowed the visualization of the process by phase-contrast microscopy. Centriole elimination was initially addressed from the point of parthenogenesis. Unfertilized oocytes rarely undergo parthenogenic development. However, Washitani-Nemoto and colleagues showed that parthenogenesis can be easily triggered in starfish oocytes if PB extrusion is suppressed (Washitani-Nemoto et al., 1994). At the time, a major hypothesis was that the presence of an active centrosome (here merely defined by a large aster visible by transmitted light microscopy) is key for parthenogenesis. A few years later, Saiki and Hamaguchi observed that only the centrioles extruded into the PBs retained a “replicative”, i.e. an aster forming capacity. They proposed for first time that starfish centrioles have an “intrinsic characteristic”, which makes them heterogeneous in their replication capability: the centriole, which remains in the mature egg is “non-replicative”, and as a consequence decays and is eliminated. In contrast, sperm derived centrioles, as well as centrioles derived from reintroduced PBs, are “replicative”, because they maintain their microtubule nucleating activity (Saiki and Hamaguchi, 1998). 71 Additionally, EM data indicated that at the beginning of meiosis, each centrosome contains a pair of centrioles (Kato et al., 1990). Therefore, out of these four centrioles three would be extruded into the two PBs, and the single centriole remaining in the mature egg would need to be a non-replicative centriole (fig. 1.27) (Tamura and Nemoto, 2001). Later, Uetake and colleagues further showed that upon PB suppression, two replicative centrioles remain in the egg, which cannot be eliminated. In fact, they can still duplicate either in parthenogenetic activated eggs or in fertilized eggs (Uetake et al., 2002; Zhang et al., 2004). By further transplantation experiments, it has been shown that these replicative centrioles are already “resistant” to elimination before meiosis resumption (Shirato et al., 2006). In conclusion, upon starfish oocyte meiosis, earlier studies established that two types of centrioles exist in the oocyte: replicative and non-replicative. The replicative centrioles are extruded into the PBs, whereas a non-replicative centriole remains in the mature egg, where it can be efficiently eliminated in contrast to the replicative centrioles forced to remain in the egg. iii. Centriole+elimination+in+starfish+oocytes+–+the+starting+ hypothesis+ In the above cited literature, a large variety of manipulations were performed, ranging from PB transplantation and PB suppression using either artificially Figure 1.27: Schematic representation of the centriole cycle in a starfish oocyte. (A) Two pairs of centrioles organize the first MI spindle. (B) One pair is extruded into the PBI. (C) A pair remains and single centriole localize at the MII spindle. (D) One centriole is extruded into the PBII. Two replicative centrioles are extruded into the two PBs. The remaining centriole is nonreplicative. Note that this cycle was never confirmed with molecular markers for centrioles. 78 maturation. Ovaries were then transferred to a Petri dish containing fresh filtered seawater, supplemented with acetylcholine (100 µM), to allow ovary contraction and oocyte release. Oocytes were then transferred with a Pasteur pipette to a Petri dish containing fresh filtered seawater, and kept at 14°C for up to two days. Sperm was obtained by puncturing a small hole on the dorsal side of the arm of a male starfish. Testis were then collected with forceps into an Eppendorf tube and kept “dry” at 4°C. 3.3.2. Fertilization Starfish fertilization is external, i.e. it occurs naturally in the sea, outside of the parent’s body; thus, embryos are easily obtained by mixing sperm with mature eggs. Nonetheless, to avoid polyspermy, sperm had first to be diluted in fresh filtered seawater (approximately 1:8000), which activates the sperm cells at the same time. Fertilization and embryo development can be followed in the same chambers used for microinjection up to one day, by which time embryos reach the early gastrula stage. Thereafter, the embryos become ciliated and swim away from the injection chambers. 3.4. Oocyte injection Immature oocytes were mounted into microinjection chambers as described in detail at http://mterasaki.us/panda/injection/. Briefly, the oocytes were mounted between two glass coverslips (one 22 x 22 mm square and a small cut coverslip) held together by double stick tape, referred to as coverslip chamber (fig. 3.1 A). Figure 3.1: (A) Coverslip chamber – where the oocytes are mounted. (B) Coverslip chamber is then mounted in the microinjection chamber – where the oocytes are injected and imaged. See side-view of coverslip chamber in fig. 3.2. 79 These coverslip chambers were then mounted in a U-shaped chamber and filled up with seawater, forming a microinjection chamber (fig. 3.1 B). This chamber keeps the oocytes in place during injection, overnight incubation and subsequent imaging. Moreover, to trigger oocyte maturation, the seawater just has to be exchanged by a 10 μM 1-methyladenine (1-MA) seawater solution. Microinjection was performed with a CellTram Oil manual injector using mercury-containing needles, as previously described at http://mterasaki.us/panda/injection/. All needles were pulled from Drummond glass capillaries using a Narishige PN-3 Glass Microelectrode Horizontal Needle Pipette Puller. Prior to injection, mRNAs and proteins were loaded into loading capillaries (Drummond) together with non-reactive silicon oil dimethylpolysiloxane (viscosity: 20 cts, Sigma), as described in http://mterasaki.us/panda/injection/. For each oocyte to be injected, the needle was front-loaded from the loading capillary, with oil, mRNA or protein, and oil. This non-reactive oil prevents mRNA mixing with the seawater and allows the identification of injected oocytes. The mercury localizes at the back (fig. 3.2), balances the pressure on the needle and helps controlling the movement of the injected liquid. Only the oocytes with the nucleus facing the top of the chamber (fig. 3.2) were injected. During imaging, this side of the chamber faces the objective and therefore imaging conditions are better, as the light has to travel less to the focus, minimizing scattering (fig. 3.2). Fluorescently labeled proteins were injected shortly before maturation and live cell imaging, whereas mRNAs encoding fluorescent markers were injected the day before to allow proteins to be expressed overnight. Figure 3.2: Microinjection: needle contains oil – mRNA/protein – oil – mercury. Only the oil and mRNA/protein is injected in the oocyte. Oocyte has a nucleus facing up and it is mounted in the microinjection chamber (side-view). 80 3.5. Drug treatment All the stock solutions were dissolved in DMSO. Before addition, the stock solutions were diluted in seawater and then applied to the oocytes. In order to be able to add drugs at specific time points during meiosis, while continuously imaging, I used a modification of the above described microinjection chamber (fig. 3.1). For this, the coverslip chamber was transferred to a small µ-Dish (Ibidi) with a window cut into the plastic bottom of the dish. These dishes are open at the top, which allows direct application of drug solutions, and are single-use, thus avoiding cross-contamination. I used the following drugs: cytochalasin D (cytoD) (Sigma, stock: 10mM, dilution: 10µM), latrunculin B (latB) (EMD Biosciences, stock: 1mM, dilution: 250nM), taxol (stock: 11µM, dilution: 11nM), Nocodazole (EMD Biosciences, stock: 10 mM, 3.3 µM dilution) and MG-132 (Calbiochem, stock: 50 mM, dilution: 250µM). LatB and cytoD are both actin-depolymerizing drugs, yet they act in a different way: latB sequesters actin monomers, whereas cytoD caps F-actin microfilaments. Both nocodazole and taxol act on microtubules, but have opposite effects: nocodazole depolymerizes microtubules and taxol stabilizes them. MG-132 is a proteasome inhibitor. Each drug has a different time of action: the effect of latB, nocodazole and taxol is visible in less than 2-5 min after drug addition. CytoD and MG-132 take 30 and 45 min, respectively, to act, requiring a longer preincubation of oocytes. 3.6. Confocal microscopy and general image processing All light microscopy was performed on a SP5II confocal microscope (Leica) equipped with a fast Z-focusing device (SuperZ Galvo stage) and using 40x HCX PL APO 1.10 NA (Leica) water immersion lens. Starfish oocytes were imaged in 3D and over time at room temperature (20°C). Scan speed was set for 700 or 1000 Hz, using a bidirectional scan and a line average of 3 or 4. Whereas the XY scaling and time resolution are variable 81 and provided for each figure, I used a Z-step of 1.5 µm for all the experiments, and exceptionally a step of 2 µm for imaging of entire embryos and oocytes. Each stack was taken with 20-25 Z-steps, depending on the orientation of the oocyte. Time resolution depended on the number of stacks and the type of experiment; but an average stack took 30-40 seconds to be acquired. Around 2-5 oocytes were imaged per chamber. During the fast 3D live imaging acquisition, to track the mother centriole transport, the imaging speed was improved in order to get a good description of such a fast process. In this case, one single oocyte was imaged per chamber, and the number of Z-steps reduced to 10-15. In this case, the imaging was performed with an approximate time resolution of 20 seconds. 3.7. Image analysis and processing For all data visualization, I used Fiji (http://fiji.sc/Fiji) and Imaris (http://www.bitplane.com/imaris). Fiji is an open source image analysis package, which was used for basic image processing and visualization, while Imaris was used for 3D visualization and 3D tracking. To reduce shot noise, a Gaussian blur filter (sigma value = 0.8) or Gaussian Blur 3D (X, Y and Z sigma value = 0.8) was applied to all images. Other image processing steps specific for each experiment are described in the appropriate section below. All processed figures were assembled in Adobe Illustrator. Some panels correspond either to single Z-slices or to maximum intensity projections, or are datasets rendered in 3D by Imaris, which is always noted in the legend. In Imaris, I created a 3D surface model of the oocyte outline using the cytoplasmic background intensities. Note that these models do not represent the real plasma membrane and are just an approximate representation of the cell contour. Time-lapse panels were generated in Fiji by projecting over time a rectangular ROI, using the “Make montage” function in Fiji. 82 3.7.1. Validation of mother and general centriole markers I developed an automatic method to validate our markers and confirm centrosome composition in starfish. In these experiments, oocytes were double injected with Poc1-mCherry as a general centriole marker and Odf2-mEGFP, as a mother centriole specific marker. Oocytes were matured and fertilized, as described above. After 24h, a Z-stack was acquired from a single layer of cells of an early gastrula epithelium (see schematic representation Results section 4.1.2 fig. 4.4 B). i. Centriole+detection++ Centriole detection was performed using the Imaris function “spot detection”, in which spots are automatically detected based on their diameter (here 0.3 µm), and intensity threshold. This function also saves the XYZ position of each spot. Other detection programs were tested as the “Mosaic” plugin (Sbalzarini and Koumoutsakos, 2005) for Fiji. Spots were also detected based on their diameter and intensity (selected radius: 2 pixels, percentile (i.e. which percentage of bright pixels are considered): 0.2%. Different measurements such as spot area and intensity (for each channel), and respective XYZ coordinates were recovered. Imaris was the method chosen for centriole detection (see below for the comparison between the two methods – fig. 3.4). ii. Quantification+of+overlap+between+mother+and+general+ centriole+markers+ We implemented a script in R (R Development Core Team, 2009) to automatically quantify the number of Odf2-mEGFP-labeled centrioles (“mother”) per pair of Poc1-mCherry-labeled centrioles (“general”). First, spot positions were independently loaded for each channel. To control against any shifts in the images, and to have an initial estimate of the spot colocalization, we plotted all spots in 3D using the “scatterplot3d” package (Ligges and Maechler) for R. 83 Matches between general-general spots and general-mother spots, i.e. colocalizing spots, were calculated using the Euclidean pairwise distance between all pairs of general and mother spots, and then solving the linear sum assignment problem (Papadimitriou and Steiglitz, 1982) using the “clue” package (Hornik, 2005) – fig. 3.3. Mother spots were only assigned to the closest general spots if their distance was less than a fixed threshold based on the cell diameter. It is important to define a pair of general spots as a single unit (i.e. two centrioles forming a centrosome), to be matched with a single mother spot. If not, one general spot of the pair would be matched with the mother spot, but the other would be categorized as an “unmatched” spot. Hence, we first described the respective pairs of general spots by finding the minimum distances between the pairwise distances. With this information, we calculated the minimum distance between the mother spots and the general spots (paired or not). Thus, the minimum distance directly relates with colocalization of the spots. All matches were organized in corresponding categories of matched or unmatched pairs, by counting the number of mother (m) spots (one, two or none) colocalizing with the number of general (g) spots (one, two – our biological pairs –, or none): g0m1, g1m0, g1m1, g2m0, g2m1 and g2m2. We tested which Figure 3.3: Chart for an example case of one embryo, showing colocalizing mother markers (m) and general (g) markers – in green; non-paired mother markers are represented in red; nonpaired general markers are shown in blue. 84 method for centriole detection was more sensitive (fig. 3.4). As Imaris was more sensitive in the identification of “g2m1” categories, “centrosome-like” configuration, it was the method selected for further centriole detection. The biological relevance of these categories is discussed in Results. All embryos and their respective centrosomes were categorized, and the total frequency of each category was calculated. Finally, we considered only the g2 category as biologically relevant and all others results were not considered for the final quantification (see in Results). 3.7.2. Quantification of centrioles extruded into polar bodies To determine the specific mother centriole extrusion during meiosis in starfish oocytes, I double injected oocytes with Odf2-mEGFP (or Chibby-mEGFP), and the spindle markers Poc1-mCherry or Cy3-Tubulin. i. Vesicle+autofluorescence+subtraction+from+the+Odf27mEGFP+ channel+ Data were acquired as a Z-stack and in four different channels: C1-C4. C1 contains the signal corresponding to Odf2-mEGFP matched to mEGFP emission profile, whereas C2 contains a red-shifted detection window to record the autofluorescence of cytoplasmic vesicles, taking advantage of the fact that autofluorescence has a much broader emission spectrum than mEGFP (fig. 3.5). Figure 3.4: Comparison between Mosaic and Imaris for an example case for one embryo. Note that both detect similar total number of pairs. However, Imaris is more sensitive detecting the category g2m1 vs. g1m1 (see blue squares). Therefore, it was the selected method for centriole detection. Categories are indicated (g) represents general, (m) indicates mother, with the respective number of spots found. A schematic representation of the categories is shown: green corresponds to mother, red corresponds to general. 85 In a second scan (using the sequential scan function of the Leica confocal software) C3 contains the signal corresponding to the spindle marker (Poc1mCherry or Cy3-Tubulin), and C4 records the transmitted light image. Recording the autofluorescence in a separate channel allowed me to subtract it from the mEGFP channel that contains both the specific signal of Odf2-mEGFP and autofluorescence of the cytoplasmic vesicles (fig. 3.5). This improved the images significantly, because it isolates the Odf2-mEGFP signal. Note that the cytoplasmic vesicles are much larger than centrioles and less bright; thus even without background subtraction, centrioles could still be unambiguously identified. I developed a macro in Fiji to automate this autofluorescence subtraction. First, all four channels were separated to individual channels, and C1 and C2 filtered by a “Gaussian blur” (sigma = 0.8). Then, a vesicle was selected using a matching circular ROI. The ROI at the same position was automatically selected in C2. The mean intensity values were obtained for this ROI for each channel, and the ratio between the two calculated. C2 window was then multiplied by this ratio to normalize the fluorescence levels between C1 and C2, followed by Figure 3.5: Vesicle autofluorescence subtraction from the Odf2-mEGFP channel – schematic representation. (A) Shows the different channels: C1 contains Odf2-mEGFP auto-fluorescence, C2 contains only the auto-fluorescence, C1-C2 shows only the Odf2-mEGFP signal. Arrowheads indicate Odf2-mEGFP signal. (B) Shows a representation of the detection range of the two channels. 86 subtraction of C2 intensity levels from C1. This subtraction was applied to all Zsteps of the multidimensional stack. All panels shown in figures are after subtraction of vesicle auto-fluorescence. ii. Mother+centriole+extrusion+–+quantification+ Odf2-mEGFP labeling was sufficient to unambiguously identify single mother centrioles, and it was additionally confirmed by co-localizing with the Poc1mCherry marker or the spindle pole labeled by Cy3-tubulin. For quantifying mother centriole localization at the metaphase II spindle, I counted the number of times an Odf2-mEGFP labeled centriole was observed at the MII spindle’s outer pole vs. inner pole. For quantifying mother centriole localization at the second polar body (PBII) stage, I counted the number of times Odf2-mEGFP centriole was extruded into PBII vs. the number of times that it remained inside the cytoplasm. Note that PBI always contains an Odf2-mEGFPand a non-labeled centriole. The same quantification was applied to oocytes injected with Chibby-mEGFP. 3.7.3. 3D tracking of centrioles To follow centriole transport, I developed a method to i) automatically segment out the cell outline based on the cytoplasmic background fluorescence, ii) detect centriole positions over time and in 3D, iii) calculate the minimum distance of centrioles to the plasma membrane over time and in 3D, and finally iv) plot minimum distances over time. Our 3D quantification intends to limit errors associated with different 3D oocyte and spindle orientations, which could affect measurements performed in 2D. This method was originally developed for Poc1-mEGFP signal, but works equally for EB3-mEGFP with a few changes that will be discussed in the appropriate section. i. Cell+outline+segmentation+ I developed a Fiji macro to automatically segment the cell outline and turn it into a set of 3D surface coordinates. An initial filtering step was performed using 87 the in-built plugin “Anisotropic Diffusion 2D” filter (number of iterations: 40, edge threshold height: 2), which efficiently filters out single pixel noise while preserving sharp edges. Next, additional Z slices were added by interpolation resulting in isotropic XYZ resolution. This interpolation “simulates” a plasma membrane closer to a real plasma membrane. This was followed by another filtering step (“3D Gaussian blur”, XYZ sigma=0.8), followed by an automatic thresholding (using Fiji’s “Mean” algorithm) that creates a mask, i.e. a binary 3D image of the cell outline (white: outside of cell, black: inside of the cell). The mask was filtered by the “Analyze particles” function, based on size (size (pixel2)=5000 to infinity). Finally, cell outlines coordinates were saved using the Fiji function (“Save XY coordinates...”), looping through Z and time to obtain all coordinates in a 4D stack. ii. Centriole+tracking+over+time+and+in+3D++ The filtered Z stack obtained in Fiji (after “3D Gaussian blur”) was loaded into Imaris to perform centriole tracking, using the automatic “spot detection” function plus the integrated tracking. Again, centrioles were detected based on intensity and size. Spots are then tracked using the “Autoregressive Motion” algorithm. Several parameters can be set, such as the maximum distance (µm) expected and the maximum gap size allowed between time points during the tracking. Tracks were further selected based on their duration. Spot detection was manually controlled and spots manually edited when automatic spot detection failed. The spots XYZ coordinates were exported to an Excel file. iii. Minimum+distance+between+centriole+and+plasma+membrane+ Using the 3D plasma membrane coordinates and the 3D centriole coordinates (both over time), the final step is to measure the distance between each centriole and the closest point in the plasma membrane (i.e. the minimum distance). For this, a script was written in Matlab (http://www.mathworks.com/products/matlab/). All cell outline XYZ coordinates were loaded into a single table, as well as all centriole coordinates. The minimum distance (in µm) between the two was calculated by searching for the minimum Euclidian distance among any possible 94 K4Fe(CN)6 (Merck) in 0.1 M cacodylate buffer. Secondary post-fixation was then performed with 1% OsO4 in 0.1M cacodylate buffer. The samples were stained with aqueous UA and gradually dehydrated in increasing concentration of ethanol in water (from 25% up to 100%). The sample was infiltrated with increasing concentration of Epon in ethanol (from 25% to 100%). Oocytes were then mounted in resin molds and left to polymerize for three to four days at 60oC. The resin blocks were finally trimmed using a trimming diamond knife (Diatome) to create a reference surface for future measurements. X7ray+ Before EM image acquisition, we performed Microscopic X-ray computed tomography (microCT) scanning in a Phoenix Nanotom m (GE Sensing & Inspection Technologies GmbH, Germany) operating under Phoenix datos|x 2 and xs control software (GE Sensing & Inspection Technologies GmbH, Germany). Resin-embedded samples were trimmed to a smaller volume (<1 mm3) and mounted as close as possible to the X-ray source, to obtain a higher resolution upon imaging. The microCT volume was reconstructed using Phoenix datos|x reconstruction software (GE Sensing & Inspection Technologies GmbH, Germany), and the volume was then processed using the VGStudio MAX software (Volume Graphics). The microCT datasets were then loaded in Amira (FEI company, http://www.fei.com/software/amira-3d-for-life-sciences/), and semi-automatically segmented using the “Labels” module. Then, 3D surface models were generated, which exposed the resin block, the oocyte and more importantly the protruding PBI (fig. 3.10). The distance between the PBI and the resin block was then measured in Amira, in order to determine how much material can be trimmed from the resin block until reaching the PBI. Therefore, this X-ray visualization allowed us to create a “distance map” of the PBI and section only the region around the PBI, instead of sectioning throughout the entire oocyte as performed before. Sections of 200nm were cut and collected onto grids for further EM analysis. 95 Sample+visualization+and+tomography+ Higher magnification images of the oocytes were acquired using the Biotwin electron microscope (120kV Transmission Electron Microscope, FEI company). Tomograms of the selected 200nm sections were then obtained with F30 (300 kV TEM, FEI company). The tomograms were then assembled and aligned in 3DMOD. 3.7.7. Morpholinos against mother centriole mRNA Morpholinos are small oligomers that bind target mRNA and block translation, consequently reducing protein levels (Wada et al., 2012). Morpholinos were designed as an antisense sequence for the 5’ end of Odf2, Chibby, and CEP164. As control, sense morpholinos were injected (table 1). Each morpholino sequence has a length of 20-30bp and includes the start codon of the target sequence. Each morpholino (synthetized by Gene Tools) was dissolved in water to a final concentration of 1mM. Figure 3.10: MicroCT datasets when loaded in Amira. Oocytes are shown in orange, embedded in the resin block (yellow). Left panels (A) and (B) show higher magnifications of protruding PBI. Higher magnifications insets of the PBI are equally shown. Matthia Winter-Karreman generated these 3D models. 96 Table 1: List of morpholinos tested. Antisense sequences are provided, except for the control (sense sequence provided). Morpholino (antisense) Sequence 5’ – 3’ Odf2 TGGTCATCCTCTACGAGATTTTCCA CEP164 AGCTGATCTCCCATCATCTTGATAT Chibby TGTTACTCGGGAGAAGTGGCATCTT Control sense GTTGGTCAATTCAAGATGCCACTTC i. Experimental+details+and+image+acquisition+ We performed morpholino depletion in oocytes and embryos. Morpholinos for our target genes were injected into oocytes, which were incubated for a total of three days at 14°C. The day before imaging, oocytes were injected with EB3mCherry mRNA and re-incubated at the same temperature. These experimental conditions were effective for CEP164 and Odf2, and for Chibby morpholino injection in embryos and oocytes, respectively. Other morpholino concentrations and incubation periods were tested for CEP164 and Odf2 in oocytes. The oocytes were live imaged during the process of meiosis. Spindle morphology and behavior were recorded using EB3-mEGFP marker. Embryos were observed at early gastrula stage, 24h after fertilization, using the microscope Zeiss Cellobserver. ii. Quantification++ Oocytes were analyzed and accessed according to the following phenotypes (or lack of): i) spindle anchoring defects and ii) PB extrusion defects. Embryos were accessed according to i) cell division defects, and ii) perturbation in cilia formation. 97 Acknowledgments:+ Kresimir (Kreso) Crnokic (Animal House, EMBL) for taking care of the starfish (and all the other animals from the marine facility). Kalman Somogyi (Lénárt lab) prepared the starfish cDNA collection (see section 3.1). Kalman also cloned Chibby-mEGFP (see section 3.2) and tested its localization. Moreover, he performed all morpholino experiments (see section 3.7.7.). Kasia Tarnawksa (Nédelec’s group, EMBL) provided the labeled tubulin used in this work (see section 3.2). Konrad Rudolph (Marioni’s group, at EMBL/EBI) wrote all the R scripts: i) in the validation of mother and general centriolar markers (see section 3.7.1 ii), and for ii) generating the mean curves for the MG-132 cyclinB-expressing oocytes (see section 3.7.5 i). Péter Lénárt helped with the Fiji macro to isolate the Odf2-mEGFP signal and the velocity measurements (see section 3.7.2 I and 3.7.3 iv). Philippe Bun (Lénárt’s group) helped with the Fiji macro for cell membrane segmentation (see section 3.7.3 i). Serge Dimitrieff (Nédelec’s group) generated the script in Matlab to calculate the distances between the centriole and the plasma membrane (see section 3.7.3 iii). Together with Julia König (Müller-Reichert’s group, at Dresden University of Technology) we immobilized starfish oocytes by high-pressure freezing, which she then processed and thoroughly serial sectioned (see section 3.7.6 i). Devrim Acehan (former member of the Electron Microscopy Core Facility, EMBL) helped me acquiring the tomograms for these samples (see section 3.7.6 i). Pedro Machado (Electron Microscopy Core Facility, EMBL) helped me with the 3D models generated from the tomograms (see section 3.7.6 i). Together with Matthia Winter-Karreman (Schwab’s group at EMBL) we fixed the oocytes by chemical fixation. Matthia performed all further processing, X-ray screening, sectioning and tomography of these samples (see section 3.7.6.ii). I would like to thank all for all the great help you provided. At last, I would like to thank all the actual Lénárt laboratory members, for all the shared equipment and reagents, and a special thanks to Masashi Mori, a former laboratory member, who taught me a lot about starfish oocytes. 98 99 4. RESULTS “The most exciting phrase to hear in science, the one that heralds the most discoveries, is not ‘Eureka!’ (I found it!) but ‘That’s funny…’” ― Isaac Asimov 4.1. Establishment of centriole composition and live cell centriolar markers in starfish 4.1.1. Identification of homologs of centriolar proteins Previous studies have analyzed the ultrastructure of centrioles in starfish, but the molecular composition was not known. The starfish centriole ultrastructure follows the conserved structure of metazoan centrioles with nine triplets of microtubules constituting the centriole’s cylinder (Kato et al., 1990). This conserved structure suggests that the centriole’s protein composition might also be conserved. To this end, an extensive phylogenetic comparison of centriole structure and molecular composition is already available for a large panel of species (Carvalho-Santos et al., 2010; Hodges et al., 2010). However, these studies did not include information on the molecular composition of starfish centrioles because no genomic or transcriptomic data was available. Recently, the Lénárt laboratory in collaboration with the laboratory of Prof. Takeo Kishimoto (Tokyo Institute of Technology, Tokyo, Japan) initiated a starfish transcriptome sequencing project for the two starfish species P. pectinifera and P. miniata. Using these datasets, I identified starfish homologs of centriolar proteins. The identification of the starfish homologs is obviously essential for further molecular analyses of centrosomal processes. 100 In my search I included centriolar proteins that have been analyzed in the two most comprehensive comparative studies (Carvalho-Santos et al., 2010, 2011; Hodges et al., 2010), as well as the functionally best described centriolar proteins in the literature. Additionally, since our central hypothesis is based on a specific behavior of mother centrioles, I have specifically focused on proteins that localize to the mother centriole-specific appendages. Strikingly, I was able to successfully identify homologs for all 25 centriolar proteins tested (fig. 4.1). First, I identified the homologue of the key component of the microtubule nucleating γTuRC complex, γ-tubulin and the key regulator of microtubule growth, the plus tip-localizing EB1 protein, both of which are widely conserved among eukaryotes (Raynaud-Messina and Merdes, 2007; Tirnauer and Bierer, 2000). I was able to additionally identify all other tubulin-family members in starfish: the components of the microtubule subunits, heterodimer forming αand β-tubulin, and also δand ε-tubulin proposed to be involved in the stabilization of the centriolar microtubule triplet (Winey and O’Toole, 2014). I also identified all the main proteins required for centriole duplication and centriole elongation: PLK4, CEP192, CPAP, SAS-6 and STIL. The same applies Figure 4.1: Identification of starfish homologs for centriolar markers. The homologs that were successfully identified are indicated by (+). Highly divergent proteins but still considered functional homologs are identified by a blue square. General and mother or daughter specific markers are indicated in the schematic representation at the bottom of the figure. Alternative names for other species are provided. 101 for the cartwheel stabilizer CEP135 and the protein CP110, which limits centriole growth. The components of the pericentriolar material (PCM) could also be identified (CEP192, CEP152 and Pericentrin), as well as PLK1, the mitotic kinase, important for PCM recruitment at mitosis (referred to as centrosome maturation). Additionally, Poc1 and Centrin-2 are highly conserved proteins among eukaryotes (Fourrage et al., 2010; Keller et al., 2009; Salisbury, 2007), and I found both to be present in starfish. Poc1 has important functions in centriole integrity (Venoux et al., 2012) and basal body stability (Pearson et al., 2009). Many studies suggest that Centrin-2 has a role in centriole duplication, yet its function is not completely understood (Salisbury, 2007). The binding partner of Centrin-2, hPoc5, involved in centriole elongation, was also identified. Furthermore, I also looked for protein homologs specific to the mother and daughter centriole. As mentioned before, the mother centriole has two sets of mother appendages. These include the subdistal appendage proteins Ninein and CEP170, and the distal appendage components CEP164. Chibby, which interacts with CEP164 and localizes to the distal appendages (Burke et al., 2014), was also found in starfish. Additionally, the ubiquitous appendage component Odf2 was identified. Centrobin is a specific daughter centriole marker and also present in starfish oocytes (see Introduction for more information about these proteins). Taken together, I was able to unambiguously identify starfish homologs of all tested structural and PCM components in the starfish transcriptome. Comparisons reveal that, consistent with its phylogenetic position, starfish centrioles have a “standard” deuterostome architecture, and therefore similar to human (H. sapiens) and sea urchin (S. purpuratus). In contrast, starfish centriole composition is quite distinct from protostome centrioles of the fruit fly (D. melanogaster) or the roundworm (C. elegans) (fig. 4.1). Notably, although many of the key centriolar proteins and their functions have been first identified in C. elegans, it has highly divergent centriolar composition, meaning that its proteins have reduced homology with other centriolar proteins and as a result fail to be detected in comparative studies (see blue squares in the fig. 4.1). This highly divergent set of centriole proteins in consistent with the highly divergent centriole ultrastructure also present in this organism (Hodges et al., 2010). 102 4.1.2. Establishment of centriolar markers Next I will detail the fluorescent protein centriolar markers that were cloned out of the set of homologs identified above. For each, alignment details and protein domain architecture will be provided. Two types of markers were generated: “general” molecular marker that labels both centrioles of the pair (used as a centriole reference), and also specific mother centriole markers that specifically identify mother centrioles. Centrin-2 and Poc1 were tested as general centriolar markers, as both have been broadly used in other species (Fourrage et al., 2010; Keller et al., 2009; Piel et al., 2000; White et al., 2000). Sequence alignments for these proteins show a high similarity between the protein sequences found in starfish and the other organisms analyzed (see Appendix section 6.3.1 and 6.3.2, respectively). Moreover, all the characteristic domains, described for these proteins in other species, can be detected. Starfish Centrin-2 contains four Ca2+-binding EF-hands domains (see fig. 4.2) and also contains at its C-terminal end the domain KKTSLY, which is characteristic of centriole-associated centrins (see fig. 4.2, orange domain) (Salisbury, 2007; Schiebel and Bornens, 1995). Starfish Poc1 contains seven WD40 repeats and the C-terminal coiled-coil which includes the conserved “Poc1” domain (see fig. 4.2, orange domain) (Fourrage et al., 2010; Keller et al., 2009) (see fig. 6.3 and 6.4 in Appendix for protein alignments). To generate a mother centriole specific marker I carefully analyzed the literature and selected a short list of potential markers to be initially cloned and tested: Odf2, Chibby, EB1 and ε-Tubulin. This list was based on i) validation as mother marker in previous studies, ii) sequence length compatible with amplification of the full cDNA sequence. This technical constraint excluded welldescribed mother centriole markers such as CEP164, CEP170 and Ninein (Chen et al., 2003; Graser et al., 2007; Guarguaglini et al., 2005; Lau et al., 2012; Ou et al., 2002; Wang et al., 2009), which are very large proteins (more than 1500 amino acids) and thus amplification from cDNA is technically more difficult. Odf2 is perhaps the best described mother centriole marker on this list, with multiple papers reporting its role as a mother appendage component (Chang et al., 2013; Ishikawa et al., 2005; Kunimoto et al., 2012; Lange and Gull, 1995; Nakagawa et al., 2001; Schweizer and Hoyer-Fender, 2009; Soung et al., 2006; Tateishi et al., 103 2013). Fewer studies report on EB1 and ε-Tubulin as mother centriole markers (Chang et al., 2002; Louie et al., 2004). Chibby was only recently characterized as a mother centriole marker (Burke et al., 2014; Steere et al., 2012) and therefore just recently included in this study. All of these markers were cloned and tested by expression of injected mRNA in starfish oocytes. Out of these, Odf2 and Chibby showed specific fluorescent labeling of the mother centrioles. Since a specific signal could not be detected, EB1 and ε-Tubulin were not further characterized. Odf2 was first reported in mammalian cells (Brohmann et al., 1997; Lange and Gull, 1995) and although phylogenetic analyses for this protein were never carefully performed, homologs for various species are annotated. However, these described Odf2 homologs are limited to vertebrates (frog, zebrafish, rat, mouse and monkey) and to the hemi-chordate acorn worm (Saccoglossus). Odf2 homologs were not previously reported in other phylogenetic groups. Indeed, my analysis identified Odf2 the first time in echinoderms, namely in starfish and the Figure 4.2: Construct design for cloned proteins. All mEGP were placed at the C-terminus of the protein. For each protein, the respective number of amino acids and characteristic domains are indicated. (A) Starfish Centrin-2: EF-h indicates EF-hands domains. In orange, a specific domain identified in centrins associated with centrioles. (B) Starfish Poc1: WD1–7 indicates the seven WD40 domains. The Poc1 domain (orange) can also be identified at the C-terminal end of the protein. CC indicates coiled-coil domains present in Poc1, Odf2 and Chibby. 110 timing, resulting in fully matured eggs that can be fertilized and develop into viable embryos. PB extrusion is preceded by a contraction wave, which causes a substantial movement of the oocyte. My imaging conditions also account for this, which guarantees that at all times the entire spindle is imaged. Importantly, the live data recapitulate all steps of spindle assembly previously described during starfish meiosis. In addition, it reveals all the delicate details and processes that have remained uncharacterized previously. Accordingly, my live cell data shows for the first time how the MI spindle assembles parallel to the plasma membrane and then rotates to assume a perpendicular position to the cortex before PB extrusion. This clearly differs from the immediate perpendicular orientation of the MII spindle assembly, which I could also show for the first time. Additionally, using EB3-mEGFP3 as a marker, this assay will allow quantification of microtubule dynamics during spindle assembly in future studies. 4.2.2. Live imaging with general centriole markers Although providing a good overview of microtubule dynamics and spindle assembly and disassembly during meiosis, EB3-3mEGFP3 imaging does not allow to directly investigate the presence of centrioles. EB3 foci are indicative of the presence of centrioles, but could also represent acentriolar microtubule organizing centers (MTOCs) that are seen in mouse oocytes, for example. Therefore, I used the above established imaging conditions and the centriole markers I described in 4.1, to follow the fate of individual centrioles throughout the meiotic divisions. First, I co-injected both general centriole markers Poc1mCherry and Centrin2-mEGFP and confirmed that both markers are able to efficiently label the four individual centrioles in starfish oocytes (fig. 4.8). While Centrin2-mEGFP labeling is restricted to the centrioles, Poc1 additionally labels microtubules (fig.4.8 and 4.9). This turned out to be an advantage, because this co-labeling of microtubules facilitates the identification of centrioles in the oocyte. Additionally, Centrin-2 often formed additional fluorescent aggregates, also described in other species that were difficult to unambiguously distinguish from the centrioles (fig.4.8). Therefore, fluorescently tagged Poc1 was then the preferred centriole marker used in further experiments. 111 Using these markers, I could determine centriole number at any given meiotic phase, which has not yet been shown in any previous studies of starfish oocytes. Thus, I could observe in a live oocyte that the poles of the MI spindle are organized in a similar manner to somatic cells with one pair of centrioles at each spindle pole (fig. 4.8 and 4.9, A and B). Upon PBI extrusion, the pair of centrioles proximal to the cortex is extruded with half of the chromosomes (fig. 4.8 and 4.9, E), whereas the other pair remains in the oocyte (fig. 4.8 and 4.9, C and D). At the end of MI, as no centriole duplication occurs between MI and MII, this pair splits and the single centrioles form the poles of the MII spindle (fig. 4.8 and 4.9, F and G). This configuration is very different from the MI spindle and mitotic spindles. At PBII extrusion, one centriole is extruded into the PBII (fig. 4.9, K), and a single centriole remains in the cytoplasm of the mature egg (fig. 4.9 J). Taken together, I could establish live cell imaging conditions to follow centrioles in starfish meiosis over 2 hours in 3D at 30 second time resolution without perturbing the process. Furthermore, using fluorescent centriolar protein markers, I could follow meiosis in live oocytes at a resolution that allowed me to unequivocally identify at all times all single centrioles present at that specific Figure 4.8: Two pairs of centrioles localize at the MI spindle, whereas single centrioles localize at the poles of the MII spindle. Dashed white line shows the outline of the oocyte. Insets show a higher magnification of centrioles. Pannels show a Z-projection of the acquired stacks. Scale bar: 10 µm. 112 stage. Consistent with previous electron EM studies, these observations show that in starfish oocytes the MI spindle is organized by a pair of centrioles at each pole, while the MII spindle has an unusual configuration with one single centriole at each pole. Consequently, in meiosis 3 out of 4 centrioles are extruded into the PBs, and only a single centriole remains in the mature egg. Figure 4.9: Poc1-mEGFP labels centrioles and the microtubules of the spindle. Arrowheads point at centrioles. Insets (A-M) show magnified centrioles for each time point. Movie starts 1h after 1-MA hormone addition. Z-stacks recorded every 30 seconds. Pannels show a Zprojection of the acquired stacks. Scale bar: 10 µm. A schematic representation of the centrioles organizing the meiotic spindles is shown. 113 4.3. Mother centrioles are extruded into the polar bodies 4.3.1. Each centrosome consists of a mother and a daughter centriole The above described tools and assays allowed me for the first time to label starfish centrioles with fluorescent markers and follow them live. Indeed, I could visualize single centrioles and show that starfish oocytes have two centrosomes, each of which is composed of a pair of centrioles at meiotic onset. It is well established in the literature that the centrosome is composed of one mother and one daughter centriole. However, until now no definitive data on centriole maturation state was available in starfish oocytes. To test if the normal configuration of the centrosome is observed in starfish oocytes, I co-expressed one general (Poc1-mCherry or Cy3-Tubulin) and one mother centriole specific (Odf2-mEGFP) marker in oocytes. I could visualize that during metaphase I, a single Odf2-mEGFP-labelled mother centriole localizes at each MI spindle pole, as expected (Fig. 4.10). Consequently, one Odf2-mEGFP labeled centriole is extruded into PBI along with one non-Odf2-mEGFP labeled centriole (Fig. 4.10). The pair that remains in the oocyte then separates, and the single centrioles organize the MII spindle. This is consistent with the data I obtained using only a general marker (as described in the previous section). However, using the specific Odf2-mEGFP-mother-labelling I observe clearly that the two spindle poles at MII are different in regard of centriole age, one containing mother centriole and the other the daughter (Fig. 4.10). It has to be noted that the mother markers used here cannot distinguish between the “older” mother centriole, i.e. the “grandmother”, and the “new” mother, and I will therefore consider that both poles of the MI spindle are equivalent. It is important to mention that the intensity of Odf2-mEGFP labeling is sometimes variable between the two mother centrioles, with one more intensely labeled than the other. However, this difference is likely to result from optical 114 effects, i.e. centrioles deeper in the cytoplasm appear dimmer due to light scattering. 4.3.2. Figure 4.10: Each starfish centrosome has one mother centriole. Movie starts 1h15 after 1-MA hormone addition. Z-stacks recorded every 38 seconds. Pannels show a Z-projection of the acquired stacks. Scale bar: 5 µm. Dashed white line indicates the outline of the oocyte. 115 4.3.2. The mother centriole is specifically extruded into the second polar body Intriguingly, I observed in all cases that the single Odf2-mEGFP-labeled mother centriole localizes at the MII spindle pole facing the plasma membrane, the outer pole, whereas the daughter centriole localizes at the pole towards the cell interior, the inner pole (n=17/17) (fig. 4.11 B). The 3D data can be difficult to visualize and interpret in 2D, therefore 3D visualization (using the software Imaris) was used to confirm all orientations ((fig. 4.11 A and B) As a consequence of its localization to the outer spindle pole, the mother centriole is extruded into the PBII (fig. 4.10). To test whether this is a specific mechanism to mother centriole, I expressed the mother marker Odf2-mEGFP together either with a general marker Poc1-mCherry or Cy3-tubulin, and counted how often the mother centriole is extruded into the PBII, in a large number of oocytes (fig. 4.12). Figure 4.11: The mother centriole always localizes to the outer pole of the MII spindle. (A) Zprojection, generated in Fiji. Scale bar: 10 µm. (B) 3D visualization of the same image, using Imaris. Note how the mother centriole clearly localizes to the outer pole in (B). The oocyte’s contour is shown in grey. Scale bar: 5 µm. 116 Figure 4.12. Mother centriole is always extruded into the PBs. (A) and (B) show an example of a double injection with the mother centriole marker Odf2-mEGFP and either Poc1-mCherry or Cy3Tubulin. (*) Indicates the remaining daughter centriole. (C) shows an example of a double injection with the mother centriole marker Chibby-mEGFP and EB3-mCherry3. All pannels show a Z-projection of the acquired stacks. Scale bar: 10 µm. (D) shows the number of cases observed for configuration (depicted below) of extrusion of mother centriole marker. 117 Strikingly, in all oocytes imaged (n=42), I found the Odf2-mEGFP labeled mother centriole in the PBII, whereas the Odf2-mEGFP negative daughter centriole always remained in the mature egg (fig. 4.12 A, B and D). Hence, these observations clearly show that in MII, the mother centriole is specifically extruded into the PBII. To confirm this pattern with another mother centriole marker, I tested ChibbymEGFP, recently described as a protein localizing to the distal mother appendages (Burke et al., 2014). Localization of Chibby-mEGFP follows exactly the same pattern as Odf2-mEGFP: one single centriole labeled per centriole pair at MI, an asymmetric localization to the outer MII spindle pole (see Appendix section 6.4), and the resulting extrusion of the Chibby-mEGFP-labeled mother centriole into PBII (n=12/12) (fig. 4.12 C and D). Taken together, I validated both Chibbyand Odf2-mEGFP as mother centriole markers, by which I was able to characterize centrosome composition and maturation state of centrioles in starfish oocytes. Most importantly, by following centrioles in live oocytes throughout the meiotic divisions, I could show that at MII, the spindle always orients with the mother centriole facing the cortex, and consequently the mother centriole is extruded into the PBII (fig. 4.10, 4.11 and 4.12, see schematic representation in fig. 4.13). Figure 4.13. Mother centrioles are always extruded into the PBs. One pair of centrioles (each with one mother and one daughter) organizes the MI spindle. During MII, the mother centriole localizes to the outer spindle pole and is consequently extruded into the PBII. A single daughter centriole remains in the mature egg. 118 Therefore, a clear asymmetry is established between the extruded and retained centrioles: the two mother centrioles are extruded into the PBs, whereas the single daughter centriole remains in the mature egg (fig. 4.13). The fact that I have observed this pattern in all oocytes imaged without exception (n>60) indicates that this is a tightly controlled process. 4.4. Extrusion of the mother centrioles is essential for centriole inactivation 4.4.1. The single daughter centriole remaining in the egg does not contribute to the zygotic spindle It is a pattern general to metazoa that the zygotic spindle is organized by the centrioles provided by the sperm, while the female centrioles are inactivated before the formation of the zygotic spindle. Specifically, starfish oocytes can be fertilized already during meiosis, shortly before the PBI extrusion. In this case, as previously shown (Kitajima and Hamaguchi, 2005), the sperm centrosomes stay “dormant” in the fertilized oocyte up to the completion of female meiosis. After completion of meiosis, the daughter centriole is eliminated shortly after the end of meiosis, and the pronucleus forms. The sperm centrosome aster rapidly grows and captures the female pronucleus, by pulling male and female pronuclei towards each other. The pronuclei then fuse and the first zygotic spindle is organized by the centrioles provided by the sperm (Zhang et al., 2004). Why do the centrioles not contribute to the zygotic spindle, and at what point are they eliminated? As described above, immature oocytes feature two pairs of duplicated centrioles of which two mother centrioles and one daughter centriole are extruded into the two PBs in the course of meiotic divisions, leaving a single daughter centriole in the mature egg. In EB3-mEGP3 injected oocytes, I could confirm that the same sequence of events occurs in oocytes, which have been 119 fertilized during meiosis. Indeed, the presence of sperm centrioles does not affect the fate of the oocyte’s centrioles – centriole extrusion still occurs (fig. 4.14). Importantly, by imaging fertilized oocytes live at the end of meiosis, I could observe how sperm aster extension is synchronous with the disassembly and the disappearance of the remaining daughter centriole aster (fig. 4.14, t=15.6 min). Thus, the daughter centriole does not participate in the formation of the zygotic spindle, and no sign of this centriole is detected later, as embryonic development progresses (fig. 4.14). In the literature it is assumed that the daughter centriole is eliminated at the end of oocyte meiosis, and my observations also point towards this direction. Shortly after completion of meiosis the daughter centriole loses its microtubule nucleation activity as detected by EB3-mEGFP3. Figure 4.14: The single daughter remaining in the mature egg does not participate in the embryonic spindle. Fertilization occurred after PBI extrusion. Movie starts 1h30 after 1-MA hormone addition. Z-stacks recorded every 79 seconds. Pannels show a Z-projection of the acquired stacks. Scale bar: 10 µm. See schematic representation below. 126 mother centriole. Additionally, I confirmed this by following the mother centriole transport in double EB3-mCherry3 and Odf2-mEGFP-labeled oocytes (fig. 4.19). These data directly confirms that only the mother centriole has the ability to be transported to the cell membrane, and no motion of the daughter centriole towards the cortex was ever observed. Figure 4.18: (A) Mother centriole transport occurs shortly after PBI extrusion. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Z-stacks recorded every 12 seconds. Movie starts after PBI extrusion. Scale bar: 5 µm. MC corresponds to mother centriole (B) Mother centriole is first transported towards the plasma membrane, where it then anchors. Chart shows distance measurements of the centriole to the plasma membrane recorded in 3D and over time. 127 Together, I conclude that a specific mechanism transports the mother centriole to the plasma membrane shortly after PBI extrusion. Thereafter, the mother centriole remains stably associated with the plasma membrane until the end of MII. At the same time the daughter centriole was never observed to move towards the plasma membrane but rather move into the cell interior through the elongation of the spindle (fig. 4.18 and 4.19). 4.5.2. Characterization of the mother centriole specific transport mechanism The specific and directional transport of the mother centriole towards the cell cortex strongly suggests that this process is mediated by the cytoskeleton. Therefore, to address the mechanisms of the specific transport of the mother centriole to the cell cortex, I used various inhibitors of the actin and microtubule cytoskeleton. However, these treatments turned out to be quite challenging, because centriole transport is a very quick process that occurs shortly after the end of MI, which in turn requires microtubules (telophase) as well as actin (cytokinesis) to be completed. I established protocols to add cytoskeletal inhibitors and thereby affect centriole transport as specifically as possible. Yet, due to the technical difficulties detailed above, these experiments are not yet fully conclusive. Figure 4.19: Time-lapse shows mother centriole being transported and anchoring to the plasma membrane. Movie starts after PBI extrusion. Z-stacks recorded every 30 seconds. Pannels show a Z-projection of the acquired stacks. Dashed line shows the cell countour. Scale bar: 10 µm. 128 i. Is+the+mother+centriole+specific+transport+driven+by+ microtubules?+ To test how microtubules contribute to centriole transport, I treated oocytes with the microtubule-depolymerizing drug, nocodazole, shortly after PBI extrusion in Poc1-mEGFP expressing oocytes (fig.4.20). Figure 4.20: (A) Microtubule depolymerization brings both centrioles close to the plasma membrane. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Movie starts after PBI extrusion. Z-stacks recorded every 20 seconds. Scale bar: 5 µm. MT corresponds to microtubules, PM to plasma membrane and ncdz to nocodazole. (B) Chart shows distance measurements of the centriole to the plasma membrane recorded in 3D and over time. Note how one of the centrioles the (presumed) mother centriole remains anchored to the plasma membrane, whereas the other ends up being lost from the Z-stack. Blue rectangle shows when the inhibitory drug starts to be active. 129 Recall that Poc1-mEGFP labels all centrioles and spindle microtubules, but since nocodazole only depolymerizes dynamic microtubules, centriolar microtubules are practically unaffected, as they are highly stable structures. Nocodazole effect is visible 2-4 minutes after addition: microtubules of the MI spindle quickly depolymerize resulting in a rapid inward collapse of the spindle. Unexpectedly, this collapse moves both centrioles near the plasma membrane (fig. 4.20). Thus, I was not able to conclude on the direct involvement of microtubules in the transport of the mother centriole to the plasma membrane: timed addition of nocodazole causes the spindle to collapse towards the cell cortex delivering both centrioles to the cell cortex. This however indicates that indeed the spindle is anchored to the cortex, which justifies why the spindle collapses towards the plasma membrane. Additionally, these results clearly show that by bringing any centriole close to the plasma membrane is not sufficient for stable anchoring: only the presumed mother centriole is able to anchor, while the presumed daughter centriole diffuses away (fig. 4.20). ii. Is+the+mother+centriole+transport+driven+by+actin?++ To test the involvement of actin in the mother centriole transport, I treated EB3-mEGFP3-expressing oocytes with latB, shortly after PBI extrusion, i.e. just before the beginning of the mother centriole specific transport to the plasma membrane. Note that although EB3-mEGFP labels equally mother and daughter centrioles, but they can be distinguished based on the microtubule nucleating at the end of meiosis: as shown above (see Results section 4.4.2 fig. 4.16) the daughter centriole is inactivated at the end of meiosis, losing its microtubule nucleating activity. In contrast, the mother centriole remains active and is able to nucleate microtubules in the mature egg. Therefore, by following Poc1-mEGFP or EB3-mEGFP3 label up to the end of meiosis, I can track back mother and daughter centrioles. 130 I observed that upon latB addition, the mother centriole (M2) is transported normally towards the cell membrane (fig. 4.21), where it properly anchors (n=6/6). Subsequently, as meiosis progresses, LatB-treated oocytes cannot extrude PBII, because the PBII extrusion is actin dependent. Therefore, the MII Figure 4.21: (A) Dynamic actin is not involved in the mother centriole transport. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Movie starts after PBI extrusion. Z-stacks recorded every 22 seconds. Scale bar: 5 µm. (B) Chart shows distance measurements of the centrioles to the plasma membrane. Transport phase indicated by the dashed lines in the graph. Blue rectangle shows when the inhibitory drug starts to be active. M2 and D2: mother and daughter centriole, respectively, from the MII spindle. 131 spindle disassembles, and the abnormally retained mother centriole (M2) preserves its microtubule nucleating activity in the mature egg’s cytoplasm (note how the daughter (D2) loses microtubule nucleating activity, whereas the mother is still active – fig. 4.21). These data together suggest that dynamic actin is not required for mother centriole transport to the cell membrane. However, similar to nocodazole, LatB sequesters actin monomers, and therefore rapidly affects dynamic or newly forming actin structures, however stable filaments are much more slowly affected. Thus, my protocol of treating oocytes with LatB for only a few minutes before centriole transport can exclude the possibility that the process is driven by dynamic actin structures, but cannot exclude the possibility of involvement of stable filaments. iii. Is+the+mother+centriole+transport+dependent+on+the+polar+ body+I+cytokinesis?+ In the previous section, I only considered oocytes in which PBI formation was completed just before latB addition, and in these oocytes mother centriole transport was normal. However, if latB was added just a little too early, oocytes failed to complete PBI cytokinesis, and the PBI collapses back into the oocyte (fig. 4.22). When I analyzed these oocytes to understand if transport still occurred, I found that, interestingly, the MII spindle does not organize in a perpendicular orientation. Instead, it forms parallel to the cell membrane (note that M2 and D2 are at almost the same distance from the membrane – fig. 4.22). Strikingly, the mother centriole is no longer transported to the plasma membrane and the entire MII spindle just sinks into the cytoplasm – fig. 4.22. I observed this same phenotype in 8 out of 9 oocytes in which PBI cytokinesis failed: no mother centriole transport occurs and the MII spindle forms parallel to the membrane. These data suggest that mother centriole transport may happen only if PBI formation is completed. Piel and colleagues showed that mother centriole moves towards the abscission site during cell division in mammalian cells. They hypothesized that this movement occurs because the mother centriole remains connected with the microtubules from the midbody (Piel et al., 2001). Indeed, a 132 similar mechanism could occur in starfish oocytes: the midbody formation during PBI cytokinesis would cause the mother centriole to move towards the abscission Figure 4.22: (A) PBI formation might be important for mother centriole transport. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Movie starts after PBI extrusion. Z-stacks recorded every 22 seconds. Scale bar: 5 µm. (B) Chart shows distance measurements of the centrioles to the plasma membrane. Blue rectangle shows when the inhibitory drug starts to act. M1 and D1 are respectively the mother and daughter centrioles from PBI. M2 and D2 are respectively the mother and daughter centrioles from the MII spindle. Both mother centrioles (M1 and M2) preserve MT nucleating activity at the end of meiosis. 133 site. Therefore, mother centriole would only occur if PBI extrusion and midbody forms. This movement towards the membrane would then allow the mother centriole to be at a reachable distance to the plasma membrane to allow anchoring. Mother+centriole+transport+is+independent+of+polar+body+cytokinesis+ R. Matsuura and K. Chiba previously showed that upon gentle centrifugation of oocytes, centrioles maintain a cortical attachment at the animal pole of the oocyte, whereas the nucleus, which is less dense than the cytoplasm, moves away from the cortex (Matsuura and Chiba, 2004) (see Material and Methods section 3.7.4 fig. 3.7). Centrifuged oocytes still undergo meiosis: NEBD occurs normally, yet PB extrusion does not take place (Barakat et al., 1994; Matsuura and Chiba, 2004). As in these centrifuged oocytes no PB is formed, this constitutes a perfect system to address whether PB cytokinesis or the midbody is required for mother centriole transport to occur, as hypothesized above. Would the mother centriole still move to the plasma membrane in this manipulated system? Because I needed to cover a larger depth and a larger area during live imaging, I used EB3-mEGFP3 in these experiments. As before, I distinguished mother and daughter centrioles in retrospect by tracking back whether they are active or not at the end of meiosis. Interestingly, after maturation of these centrifuged oocytes, mother centrioles are still transported to the plasma membrane (n=9/9) (fig. 4.23 A and B – note how both mother centrioles (M1 and M2) localize to the plasma membrane). Moreover, mother centrioles localizing twice as deep in the cytoplasm, when compared to centrioles in non-centrifuged oocytes at the beginning of the transport, are still able to move to the cell membrane (compare the distances between fig. 4.18 and fig. 4.23). However, mother centriole transport is slower (0.4 µm/min) than in non-centrifuged oocytes. One idea would be that spindle elongation (in non-centrifuged oocytes) could accelerate the transport towards the plasma membrane. Once mother centrioles reach the plasma membrane, they also remain anchored. In contrast, daughter centrioles (D1 and D2) clearly 134 lack the potential to move to the plasma membrane, and they just move randomly in the cytoplasm (fig. 4.23). Taken together, these data indicate that mother centriole transport does not depend on PBI cytokinesis site, as mother centrioles (M1 and M2) clearly move to the membrane in centrifuged oocytes. This movement is also not distance dependent, as mother centrioles localizing deeper in the cytoplasm, can still move to the plasma membrane. On the other hand, this is clearly a specific Figure 4.23: (A) Mother centriole transport occurs in a centrifuged oocyte – they do not move as a consequence of midbody formation. “PM” stands for plasma membrane. 3D visualization using Imaris. Oocyte’s contour is shown in yellow. Movie starts after NEBD. Insets show magnifications of the centrioles’ localization. (*) Shows nuclear localization after centrifugation. Z-stacks recorded every 50 seconds. Scale bar: 50 µm. (B) Chart shows distance measurements of the centrioles to the plasma membrane. M1 and M2 are mother centrioles. D1 and D2 are daughter centrioles. In this case, classification as M1 or M2 is arbitrary (in contrast to fig. 4.22). The same applies for D1 and D2. 135 property of the mother centrioles, as no daughter centriole (D1 and D2) was ever transported to the plasma membrane. To reconcile these results with the observations described in the previous section (fig. 4.22), it is clear that the cytokinetic site is not a requirement for the specific transport of the mother centriole to the plasma membrane. However, a failed cytokinesis likely introduces an additional microtubule aster from the PBI that might prevent anchoring to occur. The 3D tracks indeed suggest this (fig. 4.22 B): when PBI collapses, the daughter centriole D1 (i.e. the daughter centriole from the PBI) has a descending trajectory, which is very similar to the also descending trajectories of the centrioles M2 and D2 (mother and daughter that form the MII spindle). This suggests that the descending microtubule asters of the collapsing PBI likely i) interrupt the formation of the MII spindle in a perpendicular orientation, which therefore becomes parallel to the plasma membrane, and ii) push the MII spindle down, spatially interfering with mother centriole transport to the plasma membrane, and consequent anchoring. iv. Mother+centriole+transport+requires+proximity+to+the+nucleus+ I observed that mother centrioles do not always move to the plasma membrane in centrifuged oocytes (fig. 4.24) (n=12). In these situations, both mother (M1 and M2) and daughter (D1 and D2) centrioles would remain in the cytoplasm and no directed movement was observed. This was rather intriguing: why would the mother centriole not move in some cases? Systematic analysis of multiple oocytes revealed that mother centriole movement was related with the centriole distance to the nucleus. Dependent on the initial positions of the oocytes in the coverslip chamber, centrifugation creates a pool of oocytes with variable distances between the new nuclear position and the centrioles (see Material and Methods section 3.7.4 Fig. 3.7 and 3.8). I characterized this distance between centrioles and nucleus by measuring the angle between the two; an angle of 180° between the two means that nucleus and centrosome are at opposite poles of the oocyte (see Material and Methods). 142 direction. Interestingly, this confirms my above observations (section 4.5.2.i): even when the daughter centriole localizes close to the plasma membrane (upon microtubule depolymerization) it cannot anchor. Figure 4.29: A) Mother centriole does not depend on microtubules to anchor to the plasma membrane. Note how astral microtubules grow from the inner pole, whereas in the outer pole, mother positioning is not affected. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Movie starts after drug addition. Z-stacks recorded every 40 and 44 seconds, for nocodazole and taxol, respectively. Scale bar: 5 µm. MC and DC correspond to mother centriole and daughter centriole, respectively. MT corresponds to microtubules (B) Chart shows distance measurements of the centrioles to the plasma membrane. Blue rectangle shows when the inhibitory drug starts to be active. 143 I also tested the effect of taxol treatment and consequent microtubule stabilization: taxol treatment induces spindle growth to the double of its normal length. Still the mother centriole does not detach from the membrane and its position remains stable over time (n=12/13) (fig. 4.29 A, second panel, and B). At the same time, the daughter centriole gets “pushed” deeper in the cytoplasm as spindle extends, and astral microtubules expand from the inner spindle pole. However, no growth of microtubules was observed from the outer pole. iii. Mother+centriole+anchoring+is+independent+of+dynamic+actin+ To also test whether actin is involved in the anchoring of the mother centriole, I performed similar experiments by treating arrested MII oocytes with actin depolymerizing drugs. Treatment either with Cytochalasin D (cytoD) or latB had no effect on the mother centriole anchoring (n=8/12, for both cases) (fig. 4.30), and for both cases the MII spindle remains attached to the plasma membrane. Interestingly, the MII spindle progressively shows a slanted orientation to the plasma membrane plane (approximately 45°), but still remains anchored (fig. 4.30 A and B, for latB and cytoD treatment). Curiously, the mother centriole moves in the plane of the plasma membrane, but never away from it. This suggests that actin, although not involved in the mother centriole anchoring, might stabilize the mother centriole position at the plasma membrane. Taken together, I conclude that microtubules and actin are not involved in anchoring the mother centriole to the plasma membrane. Moreover, taxol treatment clearly shows how the two centrioles at the two MII poles are different: astral microtubules grow beyond the inner pole, whereas the same does not occur in the outer pole. 144 Figure 4.30: A) Mother centriole does not depend on dynamic actin to anchor to the plasma membrane. 3D visualization using Imaris. Oocyte’s contour is shown in grey. Movie starts after drug addition. Z-stacks recorded every 40 and 37 seconds, for cytoD and latB, respectively. Scale bar: 5 µm. MC and DC correspond to mother centriole and daughter centriole, respectively. (B) Chart shows distance measurements of the centrioles to the plasma membrane. Blue rectangle shows when the inhibitory drug starts to be active. 145 4.6.2. Are the appendages connecting the mother centriole to the plasma membrane? As shown in the previous sections, actin and microtubules do not play a role in anchoring the mother centriole to the plasma membrane during MII. Indeed, such close connection between centriole and the plasma membrane has previously not been documented in cell division. However, cases exist in which the centriole appears in close proximity to the plasma membrane: during cilia and immunological synapse formation (Reiter et al., 2012; Stinchcombe and Griffiths, 2014; Sung and Leroux, 2013, personal comunication). In both cases, the mother centriole anchors to the plasma membrane through its appendages providing the foundation of growing cilia, and the basis to re-organize the microtubule cytoskeleton for the delivery of the cytotoxic granules to the immunological synapse region, respectively. Because of the similarity of these processes to our studied case, I wanted to test whether the same mechanism of direct anchoring through appendages to the plasma membrane also functions for the mother centriole in starfish oocytes. i. Visualization+of+centriole+anchoring+by+electron+microscopy+ Mother centriole appendages are best visualized by EM. Mother appendages can be observed as small rays radiating from the mother centriole and following its 9-fold symmetry (see Introduction section 1.5). Therefore, we performed EM to assess whether a direct linkage between mother centriole and plasma membrane exists. Starfish oocytes were immobilized by high-pressure freezing or by chemical fixation (see figure legend and Material and Methods 3.7.6 i and ii) preferentially at metaphase II, when a single mother centriole localizes to the outer spindle pole. This way we obtained several electron tomography datasets, which show the centriole structure and the relation between centriole and membrane at a much higher resolution than previously reported using thin section EM (Kato et al., 1990). We could visualize that consistent with our confocal data, centrioles localize in very close proximity to the membrane during MII as well as during MI (fig. 4.31 B and C, purple). In the reconstructed tomograms, the conserved 9-fold symmetry 146 and triplet organization can be nicely recognized (fig. 4.31 A), confirming previous reports in starfish oocytes (Kato et al., 1990). The tomograms also allowed measurement of centriole length of ≈320nm (n=4), and a diameter of ≈170nm (n=4), which are also consistent with previous studies (Kato et al., 1990). The best datasets that were obtained correspond to two oocytes both in early MII stage, in which two centrioles can nicely be identified inside the PBI (fig. 4.31 C and 4.32 A, centrioles in similar orientations are framed with the same colors - orange and blue). These datasets can be used to extrapolate how the centriole might be anchored while within the oocyte. Meanwhile, we are performing more fixations to obtain centrioles anchored at the MII spindle. Figure 4.31: Starfish centrioles ultrastructure: starfish centrioles localize in close proximity to the plasma membrane in MI and MII. (A) Starfish centrioles have 9 triplets of microtubules. (B) One pair of centrioles from the MI outer spindle pole localizes in close proximity to the plasma membrane. Note how the two centrioles localize in an orthogonal orientation to each other (see schematic representation). Single sections from two different tomograms, obtained from two consecutive serial sections. (A) and (B) obtained by high-pressure freezing. (C) Starfish in metaphase II. This sample was obtained by chemical fixation. In purple, shows the presumed mother centriole close to the plasma membrane. In orange and blue, two orthogonally oriented centrioles obtained from two consecutive serial sections of the PBI. Note how the perpendicular centriole (orange) is closely localized to the plasma membrane. Black dashed lines indicate the oocyte outline. Scale bar: 100 nm. 147 Figure 4.32: Ultrastructure of two centrioles contained in the same PBI. These samples were obtained by high-pressure freezing. (A) The perpendicular centriole inside the PBI is closely localized to the plasma membrane. (B) Same dataset as in (A), orange, but tomogram was rotated in a different orientation (see schematic representation, left). Orange arrows indicate points of connection between the centriole and the plasma membrane. Black dashed lines indicate the oocyte outline. Single sections from tomograms are shown. Scale bar: 100 nm. (C) 3D model of (A): each centriole is shown in the same corresponding colors orange and blue. Left panel shows the 3D model overlaid with the EM. Middle and left panels show two different rotations of the 3D model. 148 In these data sets, one can identify one centriole perpendicular and other parallel to the membrane (fig. 4.31 C and 4.32 A, orange and blue, respectively). This perpendicular orientation is typical of a mother centriole connected to the plasma membrane, acting as a basal body. Therefore this perpendicular centriole might correspond to the mother centriole. Indeed, this centriole clearly localizes in close proximity to the plasma membrane (fig. 4.31 C and 4.32 A, orange). Moreover, electron-dense connections are found between this centriole and the plasma membrane, which may correspond to mother appendages (fig. 4.32 B – arrows – this data set corresponds to a 3D rotation of the tomogram shown in fig. 4.32 A, orange). Multiple vesicles are found in close proximity to the presumed mother centriole (fig. 4.32 A and C, orange). Interestingly, in both data sets, the centriole in a parallel orientation localizes further away from the plasma membrane, and does not show any type of connection with the plasma membrane (fig. 4.31 C and 4.32 A). This is further evidenced by the 3D models (fig. 4.32 C, blue). Clearly, more samples will be required to elucidate the ultrastructure of centriole anchoring to the cell membrane during MII. As starfish oocytes are very large cells, sectioning through an entire oocyte proved to be a very challenging and time-consuming process. Therefore, I initiated a collaboration with Matthia Winter-Karreman, who recently introduced an innovative method for single cell of such large proportions: after chemical fixation, an X-ray tomography is performed before serial sectioning (see Material and Methods section 3.7.6.ii). This allows the identification of PB positioning and subsequent targeted sectioning of only the region of interest. Using this new technique, within a short time span, we obtained our first dataset (fig. 4.31 C), and hope to acquire more data in the near future. This data will establish the ultrastructure of mother centriole anchored to the plasma membrane at MII. In conclusion, the EM data reveals the close proximity of centrioles and the plasma membrane. I could visualize electron-dense connections between one of the centrioles and the cell membrane in the PBI, suggesting that a direct interaction via mother appendages is possible. Our new strategy that involves targeted sectioning using X-ray tomography promises to rapidly increase the 149 sample number and allow us to firmly establish the ultrastructure linking the centrioles to the plasma membrane in starfish oocytes. ii. Perturbing+mother+appendages:+an+approach+to+understand+ centriole+anchoring++ To complement the EM data, we performed a functional assay by morpholino knockdown to perturb appendage formation. In ciliogenesis, mother centriole anchors to the plasma membrane through its distal mother appendages (composed of Odf2 and CEP164), therefore we targeted the two distal mother appendage proteins which I previously identified in the starfish transcriptome: Odf2 and CEP164. Odf2 is one of the main upstream factors in the hierarchic process of mother appendage assembly (Ibi et al., 2011; Tateishi et al., 2013), and several proteins (Ninein, CEP164 and Chibby) depend on Odf2 to localize at the mother centriole (Ishikawa et al., 2005). In fact, depletion of Odf2 causes a complete inhibition of mother appendage formation, which directly affects cilia formation (Tateishi et al., 2013). Similarly, depletion of CEP164 in cultured cells impairs the formation of cilia (Graser et al., 2007). First, in order to test morpholino efficacy, we monitored the effect of morpholinos in embryos. The growing embryo requires active centriolar protein translation as cell division progresses and more centrioles are required. Therefore, if morpholino perturbation is effective, problems in cell division and cilia formation are expected. Indeed we observe a phenotype in embryos treated with morpholinos against Odf2 or CEP164 (fig. 4.33) (n=10/10 for each morpholino). These embryos have highly asymmetric cell divisions, problems in development and a reduction in the number of cilia, when compared to control embryos. This indicates that the morpholinos work in embryos and effectively block mRNA translation with consequences at phenotypic level. We then tested the effect of morpholino injection in oocytes; defects in spindle anchoring were monitored by EB3-mEGFP co-expression. However, both morpholinos (Odf2 and CEP164) did not show a phenotype in oocytes. The spindle still anchors normally to the plasma membrane and PBs form normally (data not shown). Possible explanations for the lack of phenotype could be: i) 150 oocytes have a high amount of stored mRNA, which complicates an effective mRNA inhibition, ii) morpholino efficiency strongly correlates with the turnover of the protein, i.e. the balance between protein production and degradation, which is the hardest factor to predict for candidate proteins. Proteins with a high turnover, as cyclin B (Wada et al., 2012), Dysferlin (Oulhen et al., 2014) and Mos (Tachibana et al., 2000), were shown to be effectively depleted upon morpholino injection in starfish oocytes. However, these proteins have a high turnover, which likely explains why morpholino treatment is so effective. Centrioles on the other hand are highly stable organelles, formed by proteins with a low turnover (Nigg, 2006). Therefore, even if the entire mRNA pool is inhibited upon morpholino injection, the mother appendages proteins might still remain, and be stable for several days or even months the time for which oocytes are normally stored in the mother’s body. This might explain the lack of a phenotype in the oocyte: mother appendage proteins would still be present and centriole would still anchor. In contrast, during embryonic development, mRNA has to be constantly translated in order to produce more centriolar proteins for the newly forming centrioles. This would explain the strong morpholino phenotype in the embryos. Figure 4.33: Odf2 morpholino affects the normal embryonic development. (A) Shows normal embryos upon injection of a control sense morpholino. See fig. 4.4 B for schematic representation. (B) Show mutant embryos after Odf2 morpholino injection. Cep164 morpholino show similar effects. Four examples are shown for each case. Scale bar: 10 µm. 151 We have recently started to test morpholino injection against Chibby. Chibby was recently described to interact with CEP164, binding to the distal appendages. Chibby depletion was shown to impair cilia formation and therefore we included it in our study (Burke et al., 2014; Enjolras et al., 2012). Upon morpholino injection, a phenotype is observed in the oocyte: in several instances the spindle fails to anchor, which directly leads to failure of PB extrusion (fig. 4.34 A). Indeed, when comparing to oocytes control (sense), a twoFigure 4.34: Chibby morpholino causes defects in spindle anchoring, and consequently PB extrusion. (A) Movie starts 1h after 1-MA hormone addition. Z-stacks recorded every 1 min (first panel) and 1min9sec (second panel). Pannels show a Z-projection of the acquired stacks. Scale bar: 10 µm. Dashed white line indicates the outline of the oocyte. (B) Quantification of oocytes injected with Chibby-antisense (n=45) morpholino or control (n=24)