Molecular Dissection of Centrosome Biogenesis and its Implication in Human Disease

by

Yi Liu

A thesis submitted in conformity with the requirements for the degree of Doctorate of Philosophy Graduate Department of Molecular Genetics University of Toronto

© Copyright by Yi Liu 2020 Molecular Dissection of Centrosome Biogenesis and its Implication in Human Disease

Yi Liu

Doctorate of Philosophy

Department of Molecular Genetics University of Toronto

2020 Abstract

Centrosomes function as the primary microtubule-organizing centers in animal cells and play instrumental roles in the control of cell signaling, motility, division and polarity. A typical centrosome consists of a centriole pair that recruits the microtubule-nucleating pericentriolar material, organizes mitotic spindle assembly, and templates the formation of cilia/flagella. In cycling cells, centriole number is tightly controlled and aberrations in this process are associated with a set of human disorders including cancer, microcephaly and ciliopathies. Although factors essential for centriole assembly, such as STIL and PLK4, have been identified, the underlying molecular mechanisms that drive this process are incompletely understood. Herein, I characterize

ANK2 and CEP85 as novel proteins that are essential for centriole formation. Firstly, I show that

ANK2 localizes to centrosomes and interacts with CEP120. Functionally, ANK2 is required for

CEP120 centrosomal localization and centriole duplication. ANK2 also plays a role in regulating microtubule stability and the distribution of centriolar satellites. Using protein proximity mapping and high-resolution structural methods, I identify CEP85 as a novel centriole duplication factor that directly interacts with STIL. Structure-guided mutational analyses indicate that this interaction is critical for centriole loading of STIL, robust PLK4 activation and efficient centriole assembly. In addition, I also find that CEP85 and STIL play important roles in cancer ii cell migration, probably through the ARP2/3 mediated actin organization. Taken together, my

PhD research advances the mechanistic understanding of centriole duplication and thus paves the way for new therapeutic approaches to treat human diseases with abnormal centrosomes.

iii

Acknowledgments

I would like to express my sincere gratitude to my supervisor, Dr. Laurence Pelletier for his valuable and constructive suggestions throughout my Ph.D. studies. I have benefited a lot from his great mentorship and endless support on both academic and personal levels to establish my scientific career.

Many thanks to my committee members, Dr. Jason Moffat and Dr. Jim Dennis for their guidance, encouragement and insightful comments on my research work to make me become a better scientist.

I am also grateful to all of our collaborators – Dr. Mark van Breugel‟s lab, Dr. Etienne Coyaud, Dr. Brian Raught, Dr. Megha Chandrashekhar, and Dr. Jason Moffat. My research would not be conducted so successfully without their help and contributions.

Many thanks to all members of the Pelletier lab, past and present and my friends at the LTRI. Special thanks to Dr. Gagan Gupta, Dr. Johnny Tkach, Dr. Yi Luo, Sally Cheung, and Dr. Fikret Gurkan Agircan for their help and suggestions on my research projects. Thanks to Monica for teaching me advanced microscopy skills. Also thanks Dr. Qiazhu Wu, Dr. Bahareh Adhami Mojarad, Dr. Ladan Gheiratmand, Dr. Mikhail Bashkurov, Dr. Christina Yeh, Dr. João Goncalves, Dr. David Comartin for being such amazing colleagues and giving me memorable experience.

It‟s my fortune to gratefully acknowledge the support of my family, mom, dad, uncle, aunt, our dog Jacob, cousins, and close friends. Thank you all for your love, encouragement and suggestions during my Ph.D.

Above all, I would like to sincerely thank Erika for her generous love and encouragement. In the past twelve years, Erika is my most enthusiastic cheerleader to motivate me to become a well- rounded person and to help me go through the hard time together. She has shared this entire amazing Ph.D. journey with me, and I couldn‟t have done it without you.

iv

Table of Contents

Acknowledgments ...... iv

Table of Contents ...... v

1 Chapter I: Introduction ...... 1

1.1 The Centrosome Structure ...... 1

1.1.1 Centriole Structure ...... 1

1.1.2 Pericentriolar Material ...... 3

1.1.3 Centriolar Satellites ...... 5

1.2 Centriole Biogenesis ...... 6

1.2.1 Establishing a New Centriole ...... 8

1.2.2 Defining Centriole Length ...... 14

1.3 Centriole Number Control ...... 17

1.3.1 Centriole Disengagement ...... 17

1.3.2 Centriole to Centrosome Conversion ...... 18

1.4 Cell-Cycle-Based Mechanisms ...... 19

1.4.1 De Novo Centriole Duplication Pathway ...... 20

1.5 Centriole Duplication & Cell Proliferation ...... 21

1.5.1 Centrosome Loss and Surveillance Pathway ...... 21

1.5.2 Sensing Centrosome Amplification ...... 22

1.6 The Centrosome in Health ...... 23

1.6.1 The Role of Interphase Centrosomes ...... 23

1.6.2 The Role of Mitotic Centrosomes ...... 24

1.6.3 The Role of Centrosomes in Ciliogenesis ...... 27

1.7 The Centrosome Defects in Disease ...... 28

1.7.1 Centrosomes and Cancer ...... 28

v

1.7.2 Centrosomes and Microcephaly ...... 32

1.8 Summary and Rationale of this Thesis ...... 35

2 Chapter II: Direct binding of CEP85 to STIL Ensures Robust PLK4 Activation and Efficient Centriole Assembly ...... 37

2.1 Statement of Contributions ...... 38

2.2 Summary ...... 39

2.3 Introduction ...... 40

2.4 Results ...... 43

2.4.1 CEP85 is a Regulator of Centriole Duplication ...... 43

2.4.2 CEP85 Is Required for STIL Localization and PLK4 Activation ...... 49

2.4.3 CEP85 and STIL Recruitment during Centriole Duplication ...... 54

2.4.4 CEP85 Interacts Directly With the N-terminal Domain of STIL ...... 57

2.4.5 Structural Characterisation of the STIL-CEP85 Interaction Region ...... 60

2.4.6 CEP85-STIL Interaction is required for STIL Localization and Centriole Duplication ...... 63

2.4.7 CEP85 and STIL Binding is Required for PLK4 Activation ...... 70

2.5 Discussion ...... 75

2.5.1 The Evolutionary Analysis of the CEP85-STIL Complex ...... 76

2.5.2 Structural and Functional Characterization of the CEP85-STIL Complex ...... 76

2.5.3 Placing CEP85 and STIL in the Centriole Duplication Pathway ...... 77

2.6 Materials and Methods ...... 82

3 Chapter III: Direct interaction between CEP85 and STIL mediates PLK4-driven directed migration ...... 90

3.1 Statement of Contribution ...... 91

3.2 Summary ...... 92

3.3 Introduction ...... 93

3.4 Results ...... 96

vi

3.4.1 CEP85 Regulates Directional Cell Migration ...... 96

3.4.2 The Binding of CEP85 to PLK4 but Not CEP192 is required for Cell Migration ...... 103

3.4.3 CEP85 and STIL Localize at the Cell Cortex ...... 107

3.4.4 CEP85 and STIL Regulate the Actin Cytoskeleton ...... 109

3.5 Discussion ...... 112

3.6 Materials and Methods ...... 115

4 Chapter IV: Functional characterization of ANK2 in centriole duplication ...... 120

4.1 Statement of Contributions ...... 121

4.2 Summary ...... 123

4.3 Introduction ...... 124

4.4 Results ...... 126

4.4.1 ANK2 is A Novel Centriole Duplication Factor ...... 126

4.4.2 ANK2 Localizes at Centrosomes ...... 128

4.4.3 ANK2 is Required for the Centriolar Localization of CEP120 ...... 130

4.4.4 ANK2 Controls the Dynamics of Microtubules ...... 132

4.5 Discussion ...... 137

4.6 Materials and Methods ...... 140

5 Chapter V: Conclusion and Future Directions ...... 144

5.1 The functions of CEP85 and STIL in PLK4 activation and centriole assembly ...... 145

5.2 A role for the CEP85-STIL complex in PLK4-driven directed cell migration ...... 147

5.3 The function of ANK2 in centriole assembly ...... 149

6 References ...... 151

7 Copyright Acknowledgements ...... 166

vii

List of Abbreviations

APC: Anaphase promoting complex AP-MS: Affinity purification followed by mass spectrometry BioID: Proximity dependent biotin identification cc: Coiled-coil Co-IP: Co-immunoprecipitation CDK: cyclin-dependent kinase DAPI: 4', 6-diamidino-2-phenylindole 3D-SIM: 3D-Structured illumination microscopy ITC: Isothermal titration calorimetry IF: Immunofluorescence GFP: Green fluorescent protein MT: Microtubule NTD: N-terminal domain NMR: Nuclear magnetic resonance spectroscopy PB: Polo box RNAi: RNA interference SEC: Size exclusion chromatography Tet: Tetracycline γ-TuRC: Gamma-tubulin Ring Complex MTOC: Microtubule-organizing MCPH: Microcephaly

viii 1

1 Chapter I: Introduction

1.1 The Centrosome Structure

In 1887, the German biologist Theodor Boveri first described the term „centrosome‟ as the dynamic center of the cell, which mediates the nuclear and cellular division (Scheer, 2014). The centrosome is further characterized as a membrane-free organelle that acts as the major microtubule-organizing center and the basal body for primary cilia formation in various present- day eukaryotic cells. It is now evident that centrosomes play primordial roles in many cellular processes including cell signaling, motility and the regulation of cell shape and polarity (Nigg and Holland, 2018, Arquint et al., 2014). The centrosome is composed of two orthogonally paired centrioles, surrounded by pericentriolar material (PCM), an electron-dense proteinaceous matrix that nucleates microtubules (Azimzadeh and Marshall, 2010). Surrounding centrosomes are centriolar satellites that mediate protein transport along microtubules to the centrosome or basal body (Hori and Toda, 2017b).

1.1.1 Centriole Structure

Centrioles are none membrane-bound cylindrical structures that recruit pericentriolar material (PCM) to constitute the centrosomes (Gönczy, 2012, Azimzadeh and Marshall, 2010). Centrosomes function as the primary microtubule-organizing center in interphase cells and direct the bipolar spindle assembly during mitosis. This machinery plays critical roles in the control of faithful chromosome segregation and precise cell division (Prosser and Pelletier, 2017). In quiescent cells, centrioles can dock at the plasma membrane and act as basal bodies to template the formation of cilia and flagella, which are hair-like cellular appendages that function in signaling and motility (Gonçalves and Pelletier, 2017). Despite those critical functions, centrioles are not present in all eukaryotic cells such as yeast and fungi (Carvalho-Santos et al., 2011). Those cells leverage the innovative spindle pole body (SPB) to organize microtubules and mitotic spindles, which is analogous to centrosomes in higher eukaryotes (Rüthnick and Schiebel,

2

2016). However, centrioles are absolutely essential for the formation of cilia and flagella (Carvalho-Santos et al., 2011).

The centriole structure is well characterized by its nine-fold symmetry, making the centriole one of the most protein-enriched subcellular organelles (Azimzadeh and Marshall, 2010, Gönczy, 2012). Although this unique radial symmetry is evolutionarily conserved, centriole size varies among different organisms and even different cell types within the same organism (Carvalho- Santos et al., 2011). In animal cells, centriole size ranges from 200 to 220 nm in diameter, and 150 to 500 nm in length (Greenan et al., 2018). Most centrioles are composed of nine symmetrically arranged microtubule triplets. Those centriolar microtubules are composed of 13 protofilaments (α and β tubulin heterodimers) with a diameter of ~25 nm, arranged side-by-side to form a cylindrical tube (Wang and Stearns, 2017). A single triplet microtubule is assembled by a complete microtubule (A-tubule) along with two partial microtubules (B- and C-tubules, respectively) (Wang and Stearns, 2017). Additionally, centriolar microtubules are more stable and highly modified by polyglutamylation when compared to cytoplasmic microtubules (Wang and Stearns, 2017).

Due to its unique nine-ness structure, the centriole is inherently polarized with three distinct regions, as shown in Figure 1.1: (i) the proximal region defined by microtubule triplets and the cartwheel structure where the new centriole is assembled, (ii) the distal part containing microtubule doublets and appendages that are required for cilia formation, and (iii) a proteinaceous linker that keeps the duplicated centrioles engaged (Gönczy, 2012, Azimzadeh and Marshall, 2010, Gönczy and Hatzopoulos, 2019). Notably, the cartwheel is a conserved structure critical for building the precise nine-ness architecture of centrioles. Pioneer studies using electron microscopy unveil the fine cartwheel structure of mammalian (triplet) and Drosophila (doublet) (Greenan et al., 2018, Guichard et al., 2012, Gonçalves and Pelletier, 2017). A single cartwheel is formed by a central ring with a diameter of around 22 nm, from which nine evenly arranged spokes emanate (Azimzadeh and Marshall, 2010). The spokes are approximately 50 nm in length between the ring and the bulging pinhead (Nigg and Holland, 2018). Each spoke binds to the A-tubule of the triplet microtubule, onto which is linked by the C-tubule of the neighboring triplet (Rüthnick and Schiebel, 2016). This sequential assembly of A-, B- and C- tubules is essential to establish the centriole as a conserved ninefold symmetry (Wang and Stearns, 2017). In human cells, cartwheels are present in newborn centrioles but quickly

3 disappear during early mitosis, a process that maintains centriole number and structure in cycling cells (Gönczy, 2012).

Figure 1.1. The ultrastructure of centrosomes from Winey and O’Toole, 2014. Electron micrographs reveal the 9-fold symmetric structures of centrioles in isolated centrosomes. The mother centriole has additional distal and subdistal appendage modifications, which are highlighted by the cross-section micrographs. The image is courtesy of Winey and O‟Toole, 2014.

1.1.2 Pericentriolar Material

Pericentriolar material (PCM) is an amorphous mass of proteins that surrounds the two centrioles, and plays an important role in microtubule nucleation (Gönczy, 2012, Azimzadeh and Marshall, 2010). The PCM is composed of various large coiled-coil proteins such as Pericentrin and the γ-tubulin ring complexes (γTuRCs) that are critical for the centrosome-based microtubule nucleation (Delaval and Doxsey, 2010). In cycling cells, the accumulation of PCM is originated from the proximal end of mother centrioles and significantly expands in size during mitosis to direct the proper spindle assembly (Delaval and Doxsey, 2010). The components of

4

PCM are in dynamic exchange with the cytoplasmic pool through either the centrosome-based microtubule transport or their surrounding centriolar satellites (Lerit et al., 2015). In animal cells, newly formed centrioles undergo their maturation process which requires successful passage through mitosis, in order to gain their abilities to nucleate PCM in the following cell cycle (Delaval and Doxsey, 2010). This process is termed centrosome maturation. It has been proposed that centriole length is correlated to the amount of PCM they can recruit. This idea is strongly supported by previous studies of SAS-4 on C. elegans PCM formation. In this model, SAS-4 depletion results in shortened centrioles along with a marked decrease in the PCM recruitment, whereas centrioles are over-elongated through SAS-4 overexpression, causing additional accumulation of PCM to increase their microtubule nucleation capacity (Galletta et al., 2016, Gopalakrishnan et al., 2011). Recent advances in proteomics and RNAi screening have unveiled most of the key PCM components including PLK1, CEP192, Pericentrin, Aurora-A, CDK5RAP2 and CPAP, which participate in the formation of PCM structure (Woodruff et al., 2014, Loncarek and Bettencourt-Dias, 2018). In addition, CEP120 was found to regulate preferential accumulation of PCM components to maintain PCM homeostasis during quiescence (Betleja et al., 2018). However, it remains to be fully understood about the intrinsic mechanisms that determine PCM shape and size. Among the outstanding questions are how cells define its boundary without surrounding it with membranes. Early electron micrograph identifies the intricate amorphous structure of PCM (Delaval and Doxsey, 2010). To this end, pioneer work using three dimensional structured illustration microscopy (3D-SIM) revealed that individual PCM components leverage a distinct radial distribution to form multiple discrete layers around interphase centrioles (Lawo et al., 2012). This unique spatial distribution is essential for the organization and assembly of PCM. Moreover, a recent structural study revealed that Centrosomin (Cnn), a key PCM protein in D. melanogaster, can self-tetramerize into well- ordered micron-scale structures, implying an intrinsic self-assembly mechanism for ectopic PCM network formation (Feng et al., 2017). Interestingly, an alternative idea is focused on the role of phase separation as a driving force for PCM assembly. Phase separation is an emerging biophysical concept that can be applied in many fundamental cellular processes including the assembly of membraneless organelles, signaling cascade activation and gene expression. An every-day example of this physical phenomenon is the separation of vinaigrette into distinct oil and vinegar liquid droplets. Similarly, cells can leverage this machinery to form condensates as distinct cellular compartments that enables the local enrichment of multiple or hundreds of

5 protein, DNA or RNA molecules to organize biological matter (Boeynaems et al., 2018). Notably, PCM is an amorphous protein matrix surrounding the centrioles, and this non- membrane binding property implies its potential abilities in phase separation mediated self- assembly (Delaval and Doxsey, 2010). In line with this idea, pioneering studies reveal that SPD- 5, a C. elegans PCM component and also functional homolog of Cnn, could form phase- separated spherical condensates, which is consistent with the conserved spherical morphology of centrosomes. Functionally, these condensates are able to effectively concentrate their ordered client PCM proteins including PLK1, SPD-2, and TPXL-1, and ZYG-9 through interactions with SPD-5. In turn, SPD5 condensates becomes stable and are sufficient for microtubule nucleation (Woodruff et al., 2017). Overall, this SPD-5 phase separation model provides a theoretical framework to explain PCM organization and growth.

1.1.3 Centriolar Satellites

Centriolar satellites are morphologically characterized as electron-dense nonmembranous granules with a diameter of 70-100nm, accumulated around centrosomes and ciliary basal bodies in a microtubule dependent manner (Hori and Toda, 2017b). Pericentriolar material 1 (PCM1), a 228 kDa coiled-coil protein originally identified as a centrosome associated auto-antigen, was characterized as the first marker of centriolar satellites in 1994 (Balczon et al., 1994), paving the way to understand the molecular architecture of these cytoplasmic apparatuses. Recent advances in proteomics and RNAi screening have identified most of the key components of centriolar satellites including BBS4, CEP131, AZI1, CCDC14, CEP290, SSX2IP, FGFR1OP, and OFD1, which contribute to the assembly and distribution of centriolar satellites (Quarantotti et al., 2019, Gheiratmand et al., 2019). Functionally, centriolar satellites can interact with microtubules and move along them in a dynein-dependent manner, which plays an essential role in the recruitment of centrosomal proteins including Centrin, Pericentrin, Ninein and SSX2IP (Dammermann and Merdes, 2002, Hori and Toda, 2017b). In this context, misregulation of the centriolar satellites components can result in defective centriole formation. For example, downregulation of SSX2IP significantly disrupts centriole duplication (Bärenz et al., 2013), whereas CCDC14 or CEP131 depletion leads to centriole amplification (Firat-Karalar et al., 2014). Further studies identify CEP131 as a novel PLK4 (a master regulator of centriole duplication) phosphorylation substrate

6 that plays an important in maintaining centriolar satellite integrity (Denu et al., 2019, Hori et al., 2016). However, recent work reveals that PCM1 depletion has no noticeable impact on the control of centriole number, implying that centriolar satellites may exert different regulations on centriole duplication upon their specific population (Odabasi et al., 2019). Moreover, centriolar satellites can also serve as a recruitment site for multiple ciliopathies associated proteins such as BBS419, CEP290, and OFD1 (Lopes et al., 2011). Interestingly, recent studies favor atypical crosstalk between centriolar satellites and autophagy. In this model, autophagy was reported to selectively degrade the population of OFD1 residing on the centriolar satellites, thereby promoting primary cilium biogenesis (Tang et al., 2013). Molecularly, PCM1 was further shown to directly bind to GABARAP and enhance its recruitment to centriolar satellites to stimulate autophagosome formation and mediate degradation in response to starvation (Joachim et al., 2017). Despite these recent progresses in centriolar satellites, future studies are required to further characterize their structural and functional properties upon different cellular and physiological states in order to explain their complex role in centrosome biology.

1.2 Centriole Biogenesis

In cycling cells, like DNA, each centriole is duplicated once per cell cycle to form two centrosomes, which enable the formation of bipolar spindles to mediate faithful chromosome segregation during mitosis (Gönczy, 2012). The morphological features of the centriole duplication cycle have been well studied by electron microscopy (Vorobjev and Chentsov Yu, 1982, Robbins et al., 1968, Kuriyama and Borisy, 1983, Dippell, 1968, Cavalier-Smith, 1974). At the start of this cycle, two centrioles are physically connected by a proteinaceous linker. In G1/S phase, each of the two linked centrioles initiate the process of duplication to assemble daughter centrioles from the proximal end of two preexisting mother centrioles. Throughout S and G2 phases, the nascent centrioles gradually elongate and remain engaged with their mother centrioles. At the G2/M transition, the PCM surrounding the centrioles dramatically increases in size and, remarkably, acquires the capacity for microtubule nucleation. This process is termed centrosome maturation. At the same time, the linker that connects the two mother centrioles is disassembled, allowing them to separate and mediate assembly of bipolar spindles during mitosis. At the exit from mitosis, each daughter cell inherits one centrosome that contains two

7 centrioles. In the following G1 phase, the daughter centriole detaches from its mother centriole to enable another round of centriole duplication, and this process is known as centriole disengagement (Fırat-Karalar and Stearns, 2014). Taken together, this entire process represents the canonical centriole duplication cycle in proliferating cells (Figure 1.2). Since the structure of the centriole is highly conserved, the existence of conventional molecular assembly machinery has been proposed and raised an important paradox.

8

Figure 1.2. Centriole duplication cycle. Centriole duplication starts at the G1-S transition, when a daughter centriole grows orthogonally from the proximal end of two preexisting mother centrioles. Later in the S and G2 phase, daughter centrioles elongate and constitute with each mother centriole as a mature centrosome. The two centrosomes are then separated to direct bipolar spindle pole assembly during mitosis. At the end of mitosis, each daughter cell has one centrosome containing two centrioles. In the flowing G1 phase, the tight association between the daughter centriole and its mother is dissembled in a process termed centriole disengagement, which allows a new round of centriole duplication.

1.2.1 Establishing a New Centriole

Work across different organisms uncovered five evolutionarily conserved core components with essential roles in centriole duplication, including PLK4 (ZYG-1 in C. elegans), CEP192 (SPD-2 in C. elegans and Spd-2 in Drosophila), SASS6 (SAS-6 in C. elegans), STIL (SAS-5 in C. elegans and Ana-2 in Drosophila) and CPAP (SAS-4 in C. elegans and Sas-4 in Drosophila) (Pelletier et al., 2004, Pelletier et al., 2006). Centriole duplication initiates at the G1/S transition when CEP152 and CEP192, surrounding the mother centrioles, cooperate to recruit PLK4 kinase through interacting with its Polo-box domain 1 and 2 (PB1 and PB2) (Hatch et al., 2010, Sonnen et al., 2013, Dzhindzhev et al., 2010). Subsequently, the PB3 domain of PLK4 interacts with STIL and further phosphorylates it in a conserved STAN domain to promote its direct binding to the C terminus of SASS6 (Ohta et al., 2018, Moyer et al., 2015, Ohta et al., 2014). SASS6 then self-oligomerizes through its N-terminal region to build a central, nine-fold-symmetric hub (termed cartwheel), which defines centriole symmetry and diameter. Additionally, PLK4 can phosphorylate STIL on a conserved site (S428), and this phosphorylation is dispensable for the binding of STIL to CPAP, linking the growing cartwheel to the outer microtubule wall of the newly formed centrioles (Moyer and Holland, 2019). Although this molecular pathway is remarkably conserved, we could still observe the additional complexity of centriole assembly in different organisms. For example, in C. elegans, SPD-2 alone is loaded first to mediate the hierarchical recruitment of ZYG-1, SAS-5/SAS-6 and SAS-4 to initiate centriole duplication (Pelletier et al., 2006), whereas Drosophila Spd-2 is not required for this process (Giansanti et al.,

9

2008). In contrast, Asterless (also known as Asl, human orthologue CEP152) acts as the only upstream factor that controls the centriole loading of SAK (human orthologue PLK4) and promotes efficient centriole assembly in Drosophila (Galletta et al., 2016).

In this recruitment cascade, PLK4 is a master regulator and the control of its kinase activity plays an important role. Initially, PLK4, short-lived kinase, was identified in the mouse as a functional homology of Drosophila SAK that is critical for embryo development (Hudson et al., 2001). In this model, depletion of PLK4 in mouse embryos disrupts the degradation of cyclin B, leading to severe mitotic delay/apoptosis and thus arresting embryos at E7.5 after gastrulation (Hudson et al., 2001). However, PLK4+/– mouse embryos can develop normally but with higher frequency to form liver and lung cancer (Ko et al., 2005). Subsequently, PLK4 has been characterized as a key centriole duplication factor in human cells (Kleylein-Sohn et al., 2007). It is known that abnormities in PLK4 activity levels cause catastrophic consequences for centriole assembly and is associated with a set of human disorders such as microcephaly and cancers (Nigg and Holland, 2018). Notably, too low PLK4 levels or activity impair centriole duplication, while high PLK4 levels or activity result in centriole amplification (Kleylein-Sohn et al., 2007). Therefore, the availability of PLK4 must be tightly controlled to maintain the fidelity of centriole duplication.

Polo-like kinases family members have a similar characteristic Polo-box (PB) located at the carboxyl terminus of the proteins. All of PLK1-5 contain two C-terminal PBs to mediate the binding of their phosphorylated targets and kinase activation, whereas PLK4 only acquires one (Sillibourne and Bornens, 2010). Interestingly, PLK4 possesses a large central crypto Polo-box (CPB) region that has weaker homology with the canonical Polo-box domain (Sillibourne and Bornens, 2010). Previous structural studies revealed the crystal structure of Drosophila PLK4 CPB at a 2.3 Å resolution (Slevin et al., 2012). This structural model suggests that two PLK4 PB1-PB2 motifs can form a butterfly-like side-by-side asymmetric homodimer, which is required for the recruitment of PLK4 to centrioles (Slevin et al., 2012). Along with its C-terminal PB domain, PLK4 acquires a distinct triple Polo-box architecture that enables its oligomerization and further promotes its trans-autophosphorylation (Slevin et al., 2012). The stability of PLK4 is regulated by the autophosphorylation of its regulatory sites creating a phosphodegron that targets PLK4 for degradation through the SCF-Slimb/βTrCP-E3 ubiquitin ligase (Klebba et al., 2015, Guderian et al., 2010). This unique self-destruction mechanism (Figure 1.3) plays a critical role in limiting a proper level of PLK4 and thus maintaining faithful centriole duplication.

10

At the onset of centriole formation, CEP192 and CEP152 have been proposed to function cooperatively to recruit PLK4 to mother centrioles (Figure 1.4A) (Sonnen et al., 2013). In this model, double depletion of CEP192 and CEP152 completely inhibits PLK4 loading to centrioles as well as centriole duplication (Sonnen et al., 2013). Moreover, the following studies suggest a hierarchical recruitment model that CEP192 and CEP152 competitively bind to the cryptic polo box (CPB) of PLK4 through their homologous N-terminal sequences containing acidic-α-helix and N/Q-rich motifs (Kim et al., 2013). Functionally, disrupting either the CEP192-PLK4 interaction or CEP152-PLK4 interaction is sufficient to inhibit PLK4 localization to centrioles and thus suppress centriole formation (Kim et al., 2013). Further crystal structure analyses suggest that CEP192 and CEP152 interact with the CPB of PLK4 in an opposite orientation and in a mutually exclusive manner (Park et al., 2014a). Here, CEP152 displays structural advantages to bind to PLK4 markedly better than CEP192, and therefore can effectively snatch CPB away from the established CEP192-PLK4 complex (Park et al., 2014a). In line with this notion, the 3D-SIM imaging analyses reveal that before the onset of centriole formation, the younger mother centriole surrounding CEP192 only is decorated with a PLK4 ring with an outer diameter of ~443 nm, however, along with CEP152 recruitment; this centriole exhibits a wider PLK4 ring with an outer diameter of ~590 nm, which locates at the boundary of the newly forming CEP152 toroid (Park et al., 2014a). Overall, these data demonstrate that an ordered binding of CEP192 and CEP152 to PLK4 enables a spatial-temporal regulation on the centriole recruitment of PLK4 to initiate centriole assembly.

Early in G1, PLK4 locates to the proximal end of mother centrioles to assume a ring-like structure (Kim et al., 2013). At the onset of centriole formation, this ring structure is converted into a single spot in order to restrict the target site for cartwheel assembly and maintain the fidelity of centriole biogenesis (Kim et al., 2013). Molecularly, autophosphorylation-mediated degradation of PLK4 has been shown to be essential for the restriction of PLK4 at a single site, suggesting that this symmetry breaking is based on PLK4 self-organization (Ohta et al., 2018). In line with this model, binding of STIL to PLK4 can stimulate autophosphorylation of PLK4 and therefore help to dissolve its ring structure, leaving a single focus of PLK4 that co-localizes with STIL where the new centriole is assembled (Ohta et al., 2018). Meanwhile, an alternative idea is focused on the role of phase separation as a driving force for the biased distribution of PLK4 (Yamamoto and Kitagawa, 2019). Accordingly, PLK4 has been shown to acquire its intrinsic

11 ability to form phase-separated condensates, which is regulated through its trans- autophosphorylation (Yamamoto and Kitagawa, 2019). This self-condensation property may trigger effective surface tension through coalescence or Ostwald ripening to shape the asymmetric localization of PLK4 around mother centrioles, eventually promoting the conversion of PLK4 ring into a dot in procentriole formation (Yamamoto and Kitagawa, 2019). However, a recent computational study proposes that the centriole-cytoplasmic competition of the PLK4- STIL complex mediates the PLK4 ring-to-dot conversion, which is contrary to the previous model on PLK4 self-organization (Leda et al., 2018). Therefore, future investigation should investigate the biophysical and biochemical properties of PLK4 and its associated factors, which will improve the mechanistic understanding of the spatial transition of PLK4 and its impact on centriole duplication.

Figure 1.3. Speculative models for PLK4 self-destruction. The Polo box (PB) domains are critical for PLK4 homodimerization. This dimerization promotes trans-autophosphorylation of PLK4, creating an extensive phosphodegron within each Downstream Regulatory Element (DRE) region. The SCFSlimb/β-TrCP-E3 ubiquitin ligase binds to the phosphodegron and thus ubiquitinates PLK4, targeting it for proteasomal degradation.

12

Alongside this ring-to-dot conversion, PLK4 is activated and subsequently recruits STIL to mediate centriole assembly (Figure 1.4A). The human STIL (SCL/TAL1 interrupting locus previously known as SIL) gene was first identified from a common chromosomal rearrangement associated with T cell acute lymphoblastic leukemia (Aplan et al., 1990). STIL is localized at the proximal end of nascent daughter centrioles and also connected with the proximal wall of mother centrioles through its interaction with PLK4 (Arquint and Nigg, 2016). Moreover, STIL levels in cells are tightly regulated in a cell cycle dependent manner, with initial centriole loading at the onset of centriole duplication (at the G1/S transition) and its subsequent destruction during late mitosis and the following G1 phase through the anaphase promoting complex/cyclosome (APC/C) proteasome pathway (Arquint and Nigg, 2014). Furthermore, the C-terminal KEN box region of STIL is conserved among many vertebrates that contain the critical motif mediating APC/C-dependent degradation (Arquint and Nigg, 2014). Recent studies reveal that mammalian SFI1 can cooperate with USP9X to regulate STIL stability during S-phase (Kodani et al., 2019). SFI1 is an evolutionarily conserved protein first identified in yeast, where it functions to regulate spindle pole body duplication (Rüthnick and Schiebel, 2016). The human homolog of SFI1 was found to bind and recruit USP9X, a deubiquitinating enzyme to the centrosomes (Kodani et al., 2019). Subsequently, these centrosomal pools of USP9X can deubiquitinate and stabilize STIL in order to promote centriole duplication (Kodani et al., 2019). In the physiological aspects, STIL was first shown to be essential for early vertebrate development. STIL knockout mice die at mid- gestation with marked growth retardation, defects in the developing neural fold and loss of left- right symmetry (Izraeli et al., 1999). Malfunctions in zebrafish STIL lead to mitotic spindle assembly defect and thus induce disseminated neuronal apoptosis (Pfaff et al., 2007). These studies indicate that loss-of-function of STIL can result in embryonic lethality in both mouse and zebrafish due to its role in mitotic regulation. Furthermore, human STIL was identified to share high sequence similarity with Drosophila Ana2 and C. elegans SAS-5, which have been identified as key centriole duplication factors. As expected, in human cells, depletion of STIL results in loss of centrosomes, whereas its over-expression triggers centriole amplification (Arquint and Nigg, 2016). Molecularly, in the centriole duplication pathway, the recruitment of STIL to centrioles requires PLK4 phosphorylation on its STAN motif to prime the binding site for SASS6 to further regulate cartwheel formation (Moyer et al., 2015, Ohta et al., 2014). However, in Drosophila, STAN motif phosphorylation is not necessary for the recruitment of Ana2 to the centrioles. Instead, PLK4 phosphorylates the ANST motif to promote Ana2's

13 recruitment (Dzhindzhev et al., 2017). The subsequent STAN motif phosphorylation by PLK4 is required to promote Ana2 binding to Sas6 to establish the cartwheel structure. This two-step phosphorylation of Ana2 by PLK4 is unique in Drosophila cells to control centriole assembly (Dzhindzhev et al., 2017). Moreover, the direct binding of STIL to PLK4 can activate its kinase activity, possibly through promoting self-phosphorylation of the activation loop of the kinase (Ohta et al., 2014, Moyer et al., 2015). Therefore, the activation of the PLK4-STIL module represents a key element in the initiation of centriole formation.

The PLK4-STIL module further recruits spindle assembly abnormal protein 6 (SASS6) to form the cartwheel structure, serving as a scaffold for new centriole assembly (Figure 1.4B) (Arquint and Nigg, 2016). The cartwheel is composed of a central ring from which nine evenly arranged spokes emanate. Each spoke binds to the A-tubule along with two partial tubules to build up the microtubule wall of new centrioles. Pioneering structural studies using cell free systems reveal that recombinant SASS6 self-associate into rod-shaped homodimers that directly associate with their N-terminal globular head domains to form oligomers comparable to the center of the cartwheel. Such circular oligomerization with nine SASS6 dimers imparts the universal nine-fold radial symmetry of centrioles. Functionally, depletion of SASS6 results in loss of centrosomes, whereas its over-expression triggers centriole amplification (Gönczy and Hatzopoulos, 2019). The protein level of SASS6 is essential for the cartwheel assembly and centriole duplication. In human cells, the cartwheel is a transient scaffold due to the spatial-temporal regulation of SASS6. It is now evident that SASS6 is targeted by the SCF-FBXW5 E3-ubiquitin ligase for degradation (Puklowski et al., 2011). In this model, PLK4 kinase activity plays an important role in the activity of SCF-FBXW5. FBXW5 is known as a cell-cycle-regulated protein with expression levels peaking at the G1/S transition when centriole duplication initiates (Puklowski et al., 2011). Then PLK4 phosphorylates FBXW5 and thus inhibits its ability to ubiquitinate SASS6, thereby stabilizing SASS6 to form the cartwheel structure (Puklowski et al., 2011). After initiating centriole assembly, active PLK4 catalyzes its own destruction through autophosphorylation. Thus, reduced PLK4 activity relieves the activity of the SCF–FBXW5 complex, thereby allowing the ubiquitination of SASS6 and blocking centriole re-duplication. In addition, ZYG-1 (the C. elegans orthologue of PLK4) directly phosphorylates SAS-6 at the site of Ser 123, which is essential for centriole duplication, although Ser 123 is not a conserved residue in human SASS6 (Lettman et al., 2013).

14

Taken together, it is well established that PLK4 needs to cooperate with STIL and SASS6 in space and time to precisely mediate centriole assembly. This PLK4-STIL-SASS6 module represents the core conserved elements at the very early stage of centriole biogenesis. However, the exact molecular basis for this collaborative action is not fully understood.

1.2.2 Defining Centriole Length

Along with the cartwheel formation, newborn centrioles start to elongate through the sequential assembly of tubule subunits to reach a certain length. Although the size of the centrioles is variable among different organisms, human centrioles achieve a conserved length 450–500 nm with a diameter of 200–250 nm (Azimzadeh and Marshall, 2010). In Drosophila and C. elegans, the cartwheel extends throughout the length of the newborn centriole in S phase, implying that the cartwheel dimension may be correlated to the centriole length (Gönczy and Hatzopoulos, 2019). However, human centrioles display a second phase of elongation that centriolar microtubule triplets extend around 300 nm past the cartwheel during G2/M, suggesting that human centrioles may have an additional regulatory mechanism for their length control (Gönczy and Hatzopoulos, 2019).

Centriole elongation is a dynamic process that depends on the rapid polymerization of tubulin dimers. At the molecular level, a set of microtubule associated components have been characterized to map the human centriole elongation pathway, including CPAP, CEP120, SPICE and CEP135 (Figure 1.4C) (Quarantotti et al., 2019). In this model, the CPAP-CEP120-SPICE module plays a central role in centriole length control. Depletion of any of them leads to shorter centrioles, while over-expression of CPAP or CEP120 results in a marked increase in centriole length (Comartin et al., 2013). Among those components, CPAP is most critical at the initial stage of centriole elongation that acts as downstream of SASS6 and upstream of CEP120, SPICE1 and CEP135 (Comartin et al., 2013). Structural and reconstitution analysis reveals that the N-terminal region of CPAP can directly interact with a site of β-tubulin exposed at the microtubule plus end, engaging in the control of longitudinal bond energy within the microtubule lattice (Zheng et al., 2016). This enables CPAP as a microtubule plus-end capping regulator to stabilize centriole microtubules, therefore limiting the growth of centriole microtubules (Zheng et al., 2016). In support of this notion, CPAP KR377EE mutant that is defective for microtubule

15 binding shows a marked decrease in microtubule stabilizing activity, negatively impacting centriole elongation (Tang et al., 2009). Furthermore, CEP120, SPICE and CEP135 play essential roles in CPAP-driven centriole elongation. CEP120 (also termed CCDC100) was first identified as a centriole component in a proteomic analysis using purified centrosomes (Andersen et al., 2003). Physiologically, silencing of CEP120 in the neocortex disrupts interkinetic nuclear migration and neural progenitor self-renewal, negatively affecting neocortical development (Xie et al., 2007). Molecularly, pioneering work reveals that CEP120 asymmetrically locates at the newborn centrioles through its C-terminal conserved coil-coil domain, and depletion of CEP120 impairs both centriole duplication and over-duplication (Mahjoub et al., 2010). Subsequent studies revealed that CEP120 plays a role in the control of centriole elongation. In detail, the N-terminal region of CEP120 can directly bind to microtubules, and over-expression of CEP120-K76A mutant (that is defective for this interaction) significantly suppresses the formation of elongated centrioles (Lin et al., 2013b). Moreover, CEP120 has been found to physically interact with both CPAP and SPICE, and this complex works downstream of SASS6, which is required for the precise loading of CEP135 to newborn centrioles and thus limit the growth of centrioles (Comartin et al., 2013). In this context, CEP135 can directly interact with SASS6 through its C-terminus and bind to microtubules through its N- terminus (Kraatz et al., 2016, Lin et al., 2013a). Depletion of CEP135 results in a loss of centriole microtubules, as well as a marked reduction in centriole length (Kraatz et al., 2016). Thus, CEP135 is proposed to function as a linker protein that directly connects the cartwheel structure to centriole microtubules, thereby stabilizing centriole elongation.

Alongside this regulatory cascade, γ-tubulin attaches to the pinheads of the cartwheel and is required for the nucleation of A-tubules through the γ-tubulin ring complex (γ-TuRC). The B- and C-tubules are nucleated by a γ-TuRC-independent mechanism, and C-tubules terminate before the end of the microtubule doublets formed by the A- and B-tubules at the distal ends (Azimzadeh and Marshall, 2010). To define the centriole length, CP110 and Cep97 form a cap- like structure at the distal parts of newly formed centrioles, which plays an important role in limiting the microtubule growth and thus modulating centriole elongation (Azimzadeh and Marshall, 2010).

16

Figure 1.4. Centriole duplication pathway. The step-wise assembly of daughter centriole has three continuous stages, which start with the initiation stage, the following cartwheel formation and the final step procentriole elongation. (A, B). At the molecular level, CEP192 cooperates with 152 to recruit PLK4 to the proximal end of mother centrioles, and then PLK4 phosphorylates STIL to promote its binding to SASS6 to form the cartwheel structure, which serves as the template for procentriole assembly. (B). CEP120 and

17

SPICE1 cooperate with CPAP to recruit CEP135 to mediate the centriole elongation process. Meanwhile, γ-tubulin is required for the A-tubules nucleation, and a cap structure containing CEP97 and CP110 forms at the distal end of daughter centrioles, which functions to limit the microtubule growth and control the centriole length. Moreover, Centrobin can directly interact with tubulins and play a critical role in stabilizing newly formed centrioles. At the onset of mitosis, the daughter centrioles reach its full length and complete this duplication cycle.

1.3 Centriole Number Control

In cycling cells, centrioles must be duplicated once per cell cycle to maintain the proper functions of the centrosome. The previous part of the introduction has summarized the roles of the core components in building the right centrioles. The following section centers on how cells precisely license the centriole duplication in space and time. From the current point of view, this fidelity is regulated by three main processes: centriole disengagement, centriole to centrosome conversion and cell-cycle-based mechanisms.

1.3.1 Centriole Disengagement

Centriole duplication initiates at G1-S transition, when daughter centrioles form perpendicularly from the proximal end of two existing mother centrioles that are engaged by the proteinaceous linker. Throughout S and G2 phases, the newly formed centrioles will gradually elongate and remain tightly connected with their mother centrioles. It is well established that this centriole engagement property can physically block the centriole assembly site and thus prevent unwanted re-duplication. At the G2/M transition, the linker that connects two mother centrioles is dissolved (Mardin and Schiebel, 2012). The separated centrosomes then mediate the assembly of bipolar spindles during mitosis. Following passage through mitosis, two newly formed cells each inherits one centrosome composed of a mother and a daughter centriole. In the following G1 phase, the daughter centriole detaches from its mother centriole and breaks their physical engagement (Figure 1.2). This disengagement process serves as the necessary first step to enable a new round of centriole duplication.

18

At the molecular level, PLK1 and the protease separase are involved in the control of centriole disengagement. In detail, the proteolytic activity of separase mediates the cleavage of centriole engagement, while PLK1 acts upstream to promote this regulation. It is known that depletion of either separase or PLK1 results in strong defects in both centriole separation and duplication (Tsou et al., 2009). Additionally, the PCM has been proposed as a potential protector to maintain the tight connection of the centriole pair. In support of this notion, the PCM component pericentrin (PCNT) was identified as a novel substrate of PLK1, and PLK1 phosphorylation of PCNT can promote separase-dependent cleavage of PCNT and further trigger centriole separation (Kim et al., 2015). Taken together, PLK1 activity is crucial for the control of centriole disengagement and separase facilitates this action upon mitotic exit.

1.3.2 Centriole to Centrosome Conversion

Daughter centrioles are assembled at G1/S transition, elongated in S and G2 phase. Although the newborn centrioles reach their full length in early mitosis, they have no inherent capacity to duplicate or organize the pericentriolar material (PCM). At the exit of mitosis, newly formed centrioles acquire the ability to recruit PCM in a PLK1-dependent manner, which renders daughter centrioles competent for motherhood (Nigg and Holland, 2018). This process is defined as the conversion of centrioles to centrosomes. To date, a set of conserved key components have been identified to establish a molecular network to mediate this conversion. In Drosophila and human cells, the sequential loading of CEP135, CEP295 (Ana1 in Drosophila) and CEP152 (Asterless in Drosophila) onto the new centriole walls, is critical for the centriole-to-centrosome conversion (Fu et al., 2015). In this context, CEP295 functions as a scaffold to directly interact with CEP135 and CEP152 to establish this molecular network (Fu et al., 2015). In addition, CEP295 has been found to directly interact with CEP192, and thus recruit CEP192 onto the daughter centriole wall, which presumably allows the daughter centrioles to organize microtubules, assemble PCM and duplicate new centrioles (Tsuchiya et al., 2016). Furthermore, CEP295 also plays a role in maintaining the stability of newly assembled centrioles, when their cartwheel structures are removed during mitosis (Izquierdo et al., 2014). Taken together, this centriole to centrosome conversion pathway ensures the formation of a functional centrosome. Further investigation would have to explain whether there are additional players in this pathway

19 and how the centriole to centrosome conversion is regulated in space and time to license centriole duplication.

1.4 Cell-Cycle-Based Mechanisms

Like DNA replication, centriole duplication is tightly licensed with cell cycle progression. Both events initiate at G1/S transition that depends on the activities of cyclin-dependent kinases (Cdk1 and CdK2). Cdk2 is well identified as a key player that promotes the G1/S transition, whereas Cdk1 has a different role in triggering entry into mitosis (Bashir and Pagano, 2005). Interestingly, in mammalian cells depleted of Cdk2, Cdk1 is capable of compensating this loss to promote entry into S phase, implying functional redundancy between these two cyclin-dependent kinases (Bashir and Pagano, 2005). In Xenopus, sea urchin zygotes and Chinese hamster ovary (CHO) cells, the activity of Cdk2-cyclin E complex was first reported to be essential for centriole duplication (Hinchcliffe et al., 1999). Interestingly, Ckd2 deficiency in mouse embryonic fibroblasts (MEFs) has no effect on normal centrosome duplication, but leads to defective centrosome over-duplication induced by Human Papillomavirus Type (HPV)-16 E7, implying a role of Cdk2 in licensing centrosomes for aberrant duplication (Duensing et al., 2006). Furthermore, Cdk1 has been proposed to play an inhibitory role in the control of centriole re- duplication. During early mitosis, Cdk1 binds to STIL in a phosphorylation independent manner, which prevents its interaction with PLK4 and thus inhibits centriole re-duplication. Following mitosis exit and Cdk1 inactivation, new synthesized STIL is released to interact with and activate PLK4 to initiate centriole assembly. Thus, Cdk 1 regulates the timely onset of PLK4-STIL dependent centriole duplication.

In addition, the stability of the PLK4-STIL-SASS6 centriole duplication module is precisely regulated in a cell cycle dependent manner (Arquint and Nigg, 2016). During late mitosis and G1 phase, STIL is degraded through the APC/C-E3 ubiquitin ligase (Arquint and Nigg, 2014) and is stabilized in S phase through its association with deubiquitinating enzyme USP9X to initiate centriole assembly. SASS6 levels are regulated by the SCF-FBXW5-E3 ubiquitin ligase. FBXW5 is targeted by the APC/C proteasome pathway for degradation during mitosis and G1. FBXW5 levels peak at the G1/S transition when centriole duplication initiates. Then PLK4 phosphorylates FBXW5 and thus inhibits its ability to ubiquitinate SASS6, thereby stabilizing

20

SASS6 to form the cartwheel structure. After initiating centriole assembly, active PLK4 catalyzes its own destruction through autophosphorylation. Thus, reduced PLK4 activity relieves the activity of the SCF–FBXW5 complex, thereby allowing the ubiquitination of SASS6 and blocking centriole re-duplication (Puklowski et al., 2011). Upon mitosis exit, Cdk1 inactivation allows PLK4 to interact with STIL in order to initiate centriole assembly, whereas Cdk1 is activated in mitosis and can bind to STIL via the same region to which PLK4 binds, thereby preventing the PLK4-STIL interaction and centriole re-duplication (Zitouni et al., 2016). Taken together, it is now evident that these multiple layers of cell-cycle based machinery cooperate in space and time to maintain the fidelity of centriole duplication.

1.4.1 De Novo Centriole Duplication Pathway

The majority of proliferating cells only duplicate two new centrioles per cell cycle, whereas vertebrate multi-ciliated epithelial cells that construct the oviduct, airway, ventricular and spinal cord tissues have to bypass this strict rule with centriole amplification (Loncarek and Bettencourt-Dias, 2018). For example, multi-ciliated cells leverage their specialized de novo centriole duplication pathway to rapidly duplicate centrioles in the cytoplasm without preexisting mother centrioles as templates. This machinery enables the dense formation of motile cilia, which is essential for luminal flow, mucus clearance and neuronal migration. At the molecule level, the CEP63 paralogue Deup1 is a key driving force in this duplication machinery. In this model, Deup1 is able to interact with CEP152 to recruit PLK4, further self-assembling deuterosomes in the cytoplasm to template massive centriole formation (Zhao et al., 2013). Although de novo-assembled centrioles can duplicate and ciliate, a recent electron microscopy study revealed that these centrioles are prone to structural errors, even in the presence of SAS-6 self-oligomerization, when compared to canonically duplicated centrioles (Wang et al., 2015). A classic example of de novo centriole duplication arises from mouse development during gametogenesis, when both male and female gametes eliminate their centrioles. In this scenario, the zygote can regain the normal number of centrioles by leveraging de novo centriole assembly (Loncarek et al., 2007). In somatic cells, the de novo pathway can only be activated when the resident centrioles are lost or experimentally removed from the cells (Loncarek et al., 2007). Therefore, it is of great interest to investigate the underlying mechanism by which cells

21 coordinate these two different centriole duplication pathways to maintain the correct number of centrosomes in different cell types.

1.5 Centriole Duplication & Cell Proliferation

As the primary microtubule-organizing center of animal cells, centrosomes play a primordial role in organizing the mitotic spindle assembly to precisely drive cell division. However, in the absence of centrosomes, specialized cell types are still able to assemble normal spindles using an alternative microtubule nucleation pathway. A classic example is the meiotic spindle formation in acentrosomal oocytes (mice, Xenopus and Drosophila) (Dumont and Desai, 2012). In this scenario, chromatin contributes to microtubule nucleation and stabilization probably through Ran GTPase pathway, and at the same time a set of microtubule motor proteins including dynein, Nod, Ncd, Xklp1 and Eg5 replace the role of centrosomes in sorting and organizing these microtubules to form mitotic spindles in a characteristic fusiform shape (Dumont and Desai, 2012). Such a collaborative action of chromatin and motor proteins allows the acentrosomal oocytes to achieve a faithful mitotic division. However, abnormities in centriole number are poorly tolerated in the majority of proliferating cells, resulting in either cell death or a cell cycle arrest. Here, the impact of centrosome loss and amplification on cell proliferation and survival will be discussed in the following section.

1.5.1 Centrosome Loss and Surveillance Pathway

In most mammalian cells, centrosomes are essential for normal cell proliferation and division. Centrosome loss in normal cells, through a failure of centriole duplication, is sensed by specialized mechanisms to prevent cell proliferation and result in robust cell cycle arrest. However, cancer cells can survive and circumvent this surveillance pathway by inhibiting p53 function, implying that the p53-dependent pathway plays a key role in this regulation (Lambrus and Holland, 2017). At the molecular level, recent genome-wide knockout screens identify a core signaling axis including USP28, 53BP1, p53 and p21 in response to centrosome loss caused growth arrest. Importantly, depleting any of those components enable the proliferation of cells without centrosomes (Lambrus et al., 2016, Fong et al., 2016). In this model, centrosome loss is

22 first sensed by USP28 and 53BP1. USP28 is known as a deubiquitinating enzyme that directly interacts with 53BP1 through its BRCT domains. Then 53BP1 directly binds to p53, which allows the USP28-mediated deubiquitination to stabilize and activate p53 (Fong et al., 2016, Lambrus et al., 2016). The cyclin-dependent kinase inhibitor p21 (also termed CDKN1A) functions downstream of p53 to trigger p53-mediated G1 arrest in response to centrosome loss (Lambrus et al., 2016, Fong et al., 2016). Although USP28, 53BP1and p53 have been found to participate in DNA damage repair, experimental data suggest that there is no obvious DNA damage in cells lacking centrosomes, implying that this module plays a DNA damage– independent role in this regulation (Fong et al., 2016, Lambrus et al., 2016). Therefore, the p53- dependent centrosome surveillance pathway has been proposed to protect against genome instability by preventing the growth of cells with too few centrosomes.

1.5.2 Sensing Centrosome Amplification

In healthy cycling cells, increased centrosome number, generally termed centrosome amplification can lead to the formation of multipolar spindles, chromosome segregation errors, inviable progeny and an increase in aneuploidy (Nigg and Holland, 2018). Notably, these abnormalities in healthy cells result in mitotic spindle assembly checkpoint (SAC) activation, which in turn activates the mitotic catastrophe pathway to induce cell death (Godinho and Pellman, 2014). In contrast, tumor cells can bypass the SAC through clustering their supernumerary centrosomes into functional bipolar spindles, thereby achieving seemingly normal division (Godinho and Pellman, 2014). However, such clustering mechanisms are error- prone, triggering merotelic chromosome attachments (e.g., chromosomes captured from microtubules emanating from the same spindle pole), which results in low rates of aneuploidy and genome instability that subsequently promote tumor formation (Godinho and Pellman, 2014). Like centrosome loss, cells with extra centrosomes also activate a p53-dependent cell cycle arrest. But this machinery is independent of USP28 and 53BP1, implying that there is an additional signaling cascade to activate p53 in response to centrosome amplification (Lambrus and Holland, 2017). In line with this notion, a recent study reveals an unexpected role of the PIDDosome in the activation of p53-dependent pathway upon centrosome amplification (Fava et al., 2017). Caspase-2 becomes selectively activated upon cytokinesis failure in a PIDDosome-

23 dependent manner, resulting in defective cleavage of MDM2 (a p53-specific E3 ubiquitin ligase) and thus stabilizing and activating p53 (Fava et al., 2017). Unlike p53 loss, depletion of caspase 2 is unable to allow the growth of cells with extra centrosomes, implying that there may be additional pathways involved in the p53 activation. Given that many tumor cells have amplified centrosomes, bypassing the inhibitory effect of this abnormality on cell proliferation seems to be a critical step to allow these cells to acquire their oncogenic properties, which in turn could serve as potential therapeutic approaches for cancer treatment.

1.6 The Centrosome in Health

To date, among those existing organisms on our planet, centrioles are present in the majority of eukaryotic species that form cilia or flagella, and have long been proposed to participate in a number of fundamental cellular properties including cell signaling, motility, division and polarity (Nigg and Holland, 2018). The following section of my thesis centers on the functional facets of such mysterious organelles.

1.6.1 The Role of Interphase Centrosomes

At the cellular level, the microtubule network plays a pivotal role in maintaining cell polarity, cell motility and division. Notably, interphase centrosomes are well recognized as the primary microtubule-organizing center (MTOC) in most animal cells. The centrosome consists of orthogonally paired centrioles surrounded by pericentriolar material (PCM). PCM harbors the γ- tubulin ring complex (γTuRCs) that enables centrosomes to establish the radial array of interphase microtubules with their minus ends embedded in PCM and plus ends extended toward the cell periphery. It is now evident that the centrosomal microtubule nucleation contributes to the correct organization and localization of Golgi membranes in the pericentriolar position (Rios, 2014). Therefore, the association between the centrosome and Golgi is required to establish and maintain proper cell polarization, which plays an important role in many cellular processes, such as molecule transport, cell differentiation, cell migration and activation of the immune response (Rios, 2014). For example, in response to directional movement stimulus, centrosomes are reoriented along with Golgi towards the cell front, which is required for the imbalanced

24 distributions of microtubules and molecular trafficking to the leading edge and thus direct cell migration (Luxton and Gundersen, 2011). Notably, alterations to the centrosome and Golgi interaction or their structures can disrupt this front-to-rear polarity and lead to defective cell movement. In agreement of this notion, during neuronal migration in the developing cortex, the centrosome and Golgi are repositioned toward the cortical plate to determine the site for neuronal axon outgrowth (Yanagida et al., 2012). Similarly, in cytotoxic T lymphocytes, these two apparatuses translocate to the immunological synapse to facilitate the delivery of lytic granules towards their targets (Stinchcombe and Griffiths, 2014). In addition, centrosomes are detached from Golgi in G2 phase to enable the mitotic Golgi fragmentation. Disruption of this separation has been found to prevent mitotic entry, suggesting that the centrosome-Golgi proximity plays an important role in normal cell cycle progression (Sütterlin et al., 2002).

Intriguingly, pioneering studies revealed that purified centrosomes are capable of directly promoting actin filament assembly through the nucleation-promoting factor WASH along with the ARP2/3 complex (Farina et al., 2016). Furthermore, the centriolar satellite component PCM1 is essential for the centrosome loading of the WASH and ARP2/3 complex to modulate centrosome-based actin organization (Farina et al., 2016). In support of this model, centrosome- associated ARP2/3 complex has been linked to the assembly of F-actin filaments in the immune synapse (Obino et al., 2016). Along with lymphocyte activation, ARP2/3 is partially removed from the centrosome. This separation results in a marked decrease in F-actin nucleation and thus enables the centrosome detachment from the nucleus to polarize at the immune synapse (Obino et al., 2016). Furthermore, these decreasing F-actin filaments has been found to enhance the centrosome-based microtubule nucleation, thereby promoting the polarization of centrosomes upon environmental stimuli (Obino et al., 2016). To this end, future investigation will be required to elucidate how the centrosome controls actin nucleation in space and time and what potential roles they would play in other cellular processes such as cell motility and division.

1.6.2 The Role of Mitotic Centrosomes

The key step of cell division is to evenly distribute genetic material (chromosomes) into the progeny cells. It is well established that mitotic spindles act as a primary cellular machine that physically conducts this task as cells grow and divide. Abnormalities in mitotic spindles lead to

25 chromosome segregation defects and trigger the formation of aneuploidy, contributing to a number of human disorders such as cancer and Down syndrome (Nigg and Raff, 2009).

As the primary microtubule-organizing center of cells, centrosomes have been proposed as a driving force for mitotic microtubule nucleation to maintain faithful chromosome segregation and cell division. At the onset of mitosis, centrosomal PCM expands dramatically around the centrioles (termed maturation) to increase the microtubule nucleating capacity and promote mitotic spindle assembly. At the molecular level, CEP192 functions upstream to establish a signaling cascade responsible for both centrosome maturation and bipolar spindle assembly. In this model, CEP192 can interact with Aurora A and PLK1, contributing to their centrosome loading in a Pericentrin-dependent manner. This regulation first triggers Aurora A activation through its autophosphorylation (Joukov et al., 2014). Active Aurora A then phosphorylates PLK1 in its T-loop to activate its kinase activity, and this in turn facilitates PLK1 phosphorylation on CEP192 to generate the binding site for γ-tubulin ring complex (γ-TuRC) (Joukov et al., 2014). Furthermore, this active cascade promotes the phosphorylation of NEDD1 to favor γ-TuRC with robust microtubule nucleation capacity in order to efficiently assemble mitotic spindles (Joukov et al., 2014). However, in some contexts, specialized cell types with centrosome loss are still capable of organizing bipolar spindles using alternative microtubule nucleation pathways involving the small GTPase Ran, chromatin and microtubule motor proteins such as Dynein (Dumont and Desai, 2012). Although centrosomes are not necessary, recent progress reveals that PLK4 and CEP152 play critical roles in the control of acentriolar spindle assembly. In mouse embryonic development, mitotic spindles are assembled normally in the absence of centrosomes. In this model, depletion of either PLK4 or its partner CEP152 disrupts mitotic spindle assembly, thereby resulting in cytokinesis failure and developmental arrest in mouse embryos (Coelho et al., 2013). However, it is not clear whether this regulation has any functional crosstalk with the Ran GTPase pathway for spindle formation. Intriguingly, in the mouse oocyte, PLK4 and Aurora A are functionally redundant to regulate microtubule growth and acentriolar spindle assembly in a Ran GTPase independent manner, reinforcing the existence of multiple independent mechanisms among different cell types (Bury et al., 2017). Intriguingly, an alternative model has been proposed on the role of phase separation in the formation of acentrosomal spindles. A recent work revealed that a set of centrosomal proteins (AKAP450, CEP170, and KIZ), centriolar satellite proteins (CEP72, PCM1, and LRRC36), microtubule

26 binding proteins (CAMSAP3 and KANSL3), microtubule nucleation and stability regulators (CHC17, chTOG, GTSE1, HAUS6, MCAK, MYO10, and TACC3), and dynein-associated proteins (HOOK3, NDE1, NDEL1, and SPDL1), are able to concentrate and form an amorphous cellular condensate throughout the entire spindle region (So et al., 2019). Proteins in this domain are highly dynamic and phase separated into a liquid-like meiotic spindle domain (LISD) that is essential for efficient acentrosomal spindle assembly (So et al., 2019). Depletion of Aurora A and its substrate TACC3 as well as the clathrin heavy chain CHC17 result in defective LISD formation (So et al., 2019). However, future investigation is needed to characterize the molecular and functional properties of the LISD-associated condensates and to improve the mechanistic understanding of phase separation-mediated spindle assembly.

It is now evident that the centrosome plays a pivotal role in the asymmetric division of neural stem cells and germline stem cells during early development (Saade et al., 2018). For example, in Drosophila male germ line stem cells (GSCs), the mother centrosomes accumulate more PCM to maintain a more robust microtubule array when compared to the daughter centrosomes. This asymmetry allows the older centrosomes to retain in the apical side connecting to the hub cells that serve as the stem cell niche, whereas the daughter centrosomes are forced to oppositely move toward the basal side of the GSCs. Such asymmetric centrosome behaviors maintain the proper mitotic spindle orientation to establish the apical-basal cell polarity. This machinery allows hub cells to secrete Upd as a key signaling ligand to activate the JAK-STAT pathway to retain stem cell identity (Herrera and Bach, 2019). Therefore, in this model, the mother centrosomes are retained in the germline stem cells, and the daughter centrosomes are displaced into the differentiating cells.

In Drosophila neural stem cells (also termed neuroblasts), this microtubule asymmetry is switched between mother and daughter centrioles. At the molecular level, recent works has revealed that upon mitotic exist, the older centrosomes maintain PLK4 kinase activity to phosphorylate Spd2, resulting in a marked reduction in PCM recruitment and microtubule nucleation (Gambarotto et al., 2019). Conversely, the younger centrosomes with inactive PLK4 acquire a more robust microtubule assay to anchor to the apical cell cortex, while the older centrosomes are displaced away to the basal side (Gambarotto et al., 2019). As a result, the younger centrosomes remain in the neural stem cells, whereas the older centrosome is inherited by the progenitor cells upon differentiation.

27

However, in the embryonic mouse neocortex, the mother centriole attaches to the ventricular zone and is preferentially inherited by radial glia progenitors, whereas the daughter centriole migrates outside the range of ventricular zone to stay in the differentiating cells. Molecularly, depletion of Ninein, a specific mother centriole subdistal appendage protein, impairs the preferential inheritance of the centrosomes and results in loss of progenitor pools in the ventricular zone in a spindle orientation independent manner (Wang et al., 2009). Taken together, these multiple lines of evidence raise the possibility that asymmetric centrosome inheritance and spindle orientation may function as decisive influence in balancing stem cell proliferation and differentiation during the embryonic development.

1.6.3 The Role of Centrosomes in Ciliogenesis

In most animal cells, centrosomes function as the primary microtubule organizing center in interphase cells and direct the bipolar spindle assembly during mitosis. It is now evident that centrosomes are not necessary for microtubule nucleation, mitotic spindle assembly or cell division in some cell types. In line with this point, centrioles are not present in some organisms populating our planet such as yeast, fungi and female germ cells of many species (Carvalho- Santos et al., 2011). However, centrioles are found in all eukaryotic cells with cilia or flagella, and are absolutely essential for their biogenesis (Carvalho-Santos et al., 2011). The distal appendages of mother centrioles have been proposed to anchor centrioles at the cell membrane to form basal bodies, which serve as templates for the formation of cilia (often termed cilium). The basal bodies then elongate the distal end of mother centrioles to establish a central cytoskeletal core termed the ciliary axoneme. The axoneme is composed of nine symmetrically arranged microtubule doublets, each of which contains one complete A-tubule and one incomplete B- tubule extended from centrioles. To form a fully functional apparatus, the axoneme is further coated with membrane lipids and proteins delivered from the Golgi and periciliary membrane, thereby assembling into the distinct ciliary membrane. Another important functional component is the ciliary transition zone (TZ), a specialized domain located at the proximal end of the cilium. The TZ has been proposed as a ciliary gate to regulate protein entry and exit from the cilium. In cultured mammalian cells, the cilium is readily detected upon exit from the cell cycle in response to starvation stimulus and then is resorbed following the cell cycle re-entry. Until then, there are

28 two major types of cilia: non-motile (primary) and motile. It is evident that both of their ciliary axonemes consist of nine microtubule doublets that extend from basal bodies, whereas motile cilia acquire an additional central pair of microtubules surrounded by outer and inner dynein arms that hydrolyze ATP for microtubule sliding to produce motility. In many vertebrates, motile cilia are present in the lungs, respiratory tracts and middle ear, lateral fallopian tubes and ventricular surface in the brain. The majority of these tissues harbor multiple motile cilia, which are formed by the previously described de novo centriole duplication pathway. Functionally, motile cilia can beat or wave rhythmically to remove the mucus and dirt from the airway, drive the cellular motion and propel fluid transport over epithelia. In contrast, primary cilia appear as single organelles that protrude from the cell surface, which plays a critical role in coordinating the activation of signal transduction cascades such as Hedgehog and Wnt signaling (Bettencourt- Dias et al., 2011). Taken together, it is now evident that both motile and primary cilia have been well recognized for their roles in development and physiology and their dysfunctions contribute to numerous human disorders, such as mental retardation, kidney diseases, retinal degeneration, obesity and diabetes (Bettencourt-Dias et al., 2011).

1.7 The Centrosome Defects in Disease

Given the pivotal role of centrosomes in many cellular processes, it is clear that abnormalities in centrosome number or structure result in a number of human disorders. In this section, I will discuss the centrosomes‟ functions in the development of cancer and microcephaly.

1.7.1 Centrosomes and Cancer

As discussed in previous sections, the centrosome plays an important role in the mitotic spindle assembly in most animal cells. It is evident that centrosome abnormalities can cause the formation of monopolar or multipolar spindles, leading to an increase in chromosome instability and formation of aneuploidy. These mitotic errors have been well recognized as the hallmarks of cancer. Indeed, centrosome abnormalities have been observed in a wide range of solid tumors including breast, colorectal, prostate and pancreatic cancers, and are associated with high tumor grade and low survival rate (Gönczy, 2015). These defects can be divided into structural or

29 numerical alterations, and the following section will explain the mechanism of how they contribute to the development of cancer.

Centrosome numerical defects mainly result from abnormal centrosome biogenesis. Centrosome loss, through defective centriole duplication, in healthy cells results in robust cell cycle arrest, whereas cancer cells can survive and circumvent this surveillance pathway by inhibiting p53 function (Godinho and Pellman, 2014). Although this abnormality may result in mitotic errors such as chromosome mis-segregation, its role in tumor formation remains elusive. In contrast, centrosome amplification is frequently observed in multiple human tumors, which triggers the formation of multipolar spindles, chromosome segregation defects, and an increase in aneuploidy (Godinho and Pellman, 2014). These defects in normal cells lead to activation of the mitotic spindle assembly checkpoint (SAC), thereby activating a so-called mitotic catastrophe pathway to cause cell death. Cancer cells bypass this mitotic catastrophe through clustering extra centrosomes, thereby achieving this seemingly normal bipolar division (Godinho and Pellman, 2014). In the past decades, remarkable progress has been made to unveil the molecular basis of centrosome clustering. It is now evident that this process primarily relies on a number of microtubule-based motors and microtubule-associated proteins (MAPs) that maintain mitotic spindle poles, including HSET (a microtubule plus-end motor), NUMA (a mitotic spindle associated MAP) and cytoplasmic dynein (a microtubule minus-end motor), KIF2C (centromere- associated kinesin), TACC3-chTOG complex (spindle microtubule stabilizers) (Leber et al., 2010). Genetic or chemical perturbation of these factors has been found to suppress centrosome clustering and induce apoptosis in cancer cells with extra centrosomes (Leber et al., 2010). In addition, the presence of E-cadherin in epithelial cells inhibits the efficient clustering of supernumerary centrosomes, while the loss of E-cadherin promotes cortical contractility to enable efficient centrosome movement, thereby facilitating HSET-mediated centrosome clustering (Rhys et al., 2018). In agreement with this notion, cancer cells with high-grade centrosome amplification often down-regulate E-cadherin to enhance cell proliferation and survival (Rhys et al., 2018). Along with this fitness advantage, centrosome clustering-mediated spindle assembly is error-prone (termed pseudo-bipolar spindles), causing merotelic chromosome attachments (e.g. chromosomes captured from microtubules emanating from the same spindle pole) that lead to low rates of chromosome instability (CIN) and aneuploidy (Godinho and Pellman, 2014). Such low enough frequency errors may not have the detrimental

30 effects of multipolar divisions, but can still contribute to genomic instability that might facilitate the initiation and progression of cancer (Godinho and Pellman, 2014). Despite these progresses, there is still uncertainty as to whether supernumerary centrosomes can alone drive tumorigenesis or if they are simply a passenger phenotype.

Centrosome structural defects can be classified as alterations to centriole size or the amount of PCM. To precisely define these defects, conventional electron microscopy is used to measure the centriole length which is normally around 500 nm in human cells, and three dimensional structured illustration microscopy (3D-SIM) helps to determine the spatial organization of PCM structure (Lawo et al., 2012). In cancer cells, these aberrations are often scored as a marked increase in centriole length as well as enhanced levels of PCM. In support of this notion, a recent study investigated the landscape of centrosome structural abnormities in 60 human cancer cell lines used by the National Cancer Institute (NCI-60) (Marteil et al., 2018). They found that 22 cell lines belonging to breast, skin and lung cancer displayed prominent centriole over- elongation, and this defect can over-activate centrosomes, resulting in enhanced levels of PCM and microtubule nucleation (Marteil et al., 2018). Moreover, these over-elongated centrioles have been shown to trigger centriole amplification (Marteil et al., 2018). Consequentially, over- elongated centrioles in these cancer cells lead to severe chromosome segregation errors, which may contribute to the development of cancer.

To examine the oncogenic properties of amplified centrosomes, inducible PLK4 overexpression system is widely used to trigger centrosome over-duplication. In the Drosophila brain, over- expression of SAK (human PLK4 homolog) is able to induce robust centrosome amplification. Unexpectedly, such gain of centrosomes is not sufficient to trigger large-scale genomic instability and fails to initiate tumorigeneisis (Nigg and Raff, 2009). However, transplantation of these brain cells with extra centrosomes in wild type hosts can trigger the formation of metastatic tumors (Nigg and Raff, 2009). Similarly, in mouse brain and epidermis with functional p53, PLK4 triggered centrosome amplification, leading to spindle orientation defects and aneuploidy, which was still not capable of promoting tumor formation (Vitre et al., 2015). In contrast, in p53 null mice, supernumerary centrosomes can accelerate the development of skin cancers and also promote the onset of lymphomas and sarcomas (Serçin et al., 2015, Coelho et al., 2015). Importantly, a recent study further validates that centrosome amplification by itself is sufficient to trigger aneuploidy and enhance the initiation of tumors in a mouse model of intestinal

31 neoplasia, when the p53 pathway is partially inactivated (Levine et al., 2017). Therefore, the p53 dependent pathway is proposed as a driving force to determine the oncogenic properties of centrosome amplification. In addition, centrosome amplification has been found to participate in the control of epithelial cell invasion. At the molecular level, extra centrosomes increase centrosomal microtubule nucleation that promotes the activation of small GTPase Rac1, thereby disrupting normal cell-cell adhesion and triggering cell invasion (Godinho et al., 2014). Overall, these findings demonstrate that centrosome amplification can confer advantageous small-scale genomic instability as well as increased tumor cell invasive behaviors, raising the possibility that extra centrosomes are able to promote cancer initiation and metastatic progression. Therefore, centrosome amplification has been proposed as a promising target for cancer treatment.

To test its potential as a therapeutic target, genetic or chemical perturbation of centrosome amplification has been explored, and PLK4, a master regulator of centriole duplication emerges as a suitable target. Pioneering work identified the first PLK4 inhibitor (termed CFI-400945) that strongly inhibits centriole duplication and triggers mitotic errors and apoptosis in breast cancer cell lines upon high concentration drug treatment (Mason et al., 2014). This PLK4 inhibitor also exhibits in vivo anti-tumor activity, which is currently under clinical trials for breast cancer treatment (Mason et al., 2014). However, CFI-400945 can also partially inhibit mitotic kinase Aurora B, which may contribute to genomic instability and cytokinesis failure. Such an off-target effect influences the interpretation of PLK4 inhibition in this anti-tumor activity. To develop a specific PLK4 inhibitor, a recent study utilized pan-Aurora kinase inhibitor VX-680 as a template to perform structural guide mutagenesis assays for the identification of novel PLK4 inhibitors (Wong et al., 2015). As a result, a compound (centrinone) was synthesized where methoxy substituent at the VX-680 C5 position, and centrinone exhibits over 1000-fold selectivity for PLK4 over Aurora A or B (Wong et al., 2015). As expected, this highly selective inhibitor enables the reversible depletion of centrosomes and has no obvious effect on the proliferation of multiple cancer cell lines, raising the possibility that centrinone may confer its therapeutic value to affect cancer invasion and metastasis caused by centrosome amplification (Wong et al., 2015). However, both CFI-400945 and centrinone treatment in healthy cells result in centrosome loss and cell cycle arrest, implying that they may not be an ideal anti-cancer strategy. In this context, inhibition of centrosome clustering is proposed as an alternative therapeutic approach against cancer cells with extra centrosomes. Towards this, recent work

32 nicely shows that disrupting the interaction of CPAP with microtubules can perturb centrosome clustering through increasing interphase microtubule nucleation (Mariappan et al., 2019). The following small molecule screen identified CCB02 as a potent and selective inhibitor against CPAP-Tubulin interaction (Mariappan et al., 2019). This chemical inhibition has been found to cause mitotic cancer cells to undergo centrosome declustering and subsequent cell death (Mariappan et al., 2019). More importantly, CCB02 treatment exhibits a significant in vivo anti- tumor activity, but has no adverse effect on healthy cells with normal centrosome numbers (Mariappan et al., 2019). These findings raise the possibility of developing a selective treatment for tumor cells with amplified centrosomes without harming healthy cells.

Despite these remarkable progresses in the past decades, we are still far from understanding the causes and consequences of centrosome abnormities in cancer. Future work will be required to investigate the molecular and genetic basis by which centrosome biogenesis is misregulated in cancer and explain how centrosome defects are leveraged to favour oncogenic properties. It will be of great interest, for example, to consider the additional mechanisms on tumor microenvironment, cancer stem cells, and epigenetic regulation. To achieve human biological readouts, primary patient-derived tumor cells can be used to generate organoids to characterize the oncogenic properties of centrosome defects and pave the way for new therapeutic approaches for cancer treatment.

1.7.2 Centrosomes and Microcephaly

Autosomal recessive primary microcephaly (MCPH) is a genetic neurodevelopmental disorder afflicting infants and children with a pronounced reduction of brain size, mental retardation and low life expectancy (Woods et al., 2005). Microcephaly is a rare birth defect with the estimated prevalence around 2-12 babies per 10,000 births. To diagnose microcephaly after birth, head circumference (HC) is a commonly used measurement to evaluate brain size. Patients with a HC three standard deviation (SD) less than average are considered to have microcephaly (Woods et al., 2005). Currently, there are no effective treatments or cures for this brain disorder. Molecularly, the disease is genetically heterogeneous, with at least twelve causative loci identified to date upon the growing amount of sequencing data for microcephaly patients. Remarkably, nearly eight of these mutated genes code for centriole duplication-related proteins,

33 including PLK4, STIL, CPAP, CEP135, CEP152, CEP63, CDK5RAP2, WDR62 and ASPM (Nigg and Holland, 2018), implying that abnormalities in centriole biogenesis may contribute to disease development. Given the major role of centrosomes in mitotic spindle assembly, centrosome loss or amplification results in severe mitotic errors, thereby activating mitotic spindle assembly checkpoint (SAC) and catastrophe pathway along with severe mitotic delay and apoptosis. These defects are thought to negatively affect the asymmetric division of neural progenitor cells (NPCs) to promote differentiation, resulting in a marked reduction in NPCs and neurons (Nigg and Holland, 2018). STIL is a key centriole duplication factor and mutations in STIL are associated with microcephaly. There are two STIL truncating mutations (p.Gln1239X and p.Val1219X) reported in microcephaly patients (Arquint and Nigg, 2014). These truncations delete the C-terminal KEN-box region that is targeted by APC/C-E3 ligase for degradation, resulting in a non-degradable version of STIL. Expressions of these mutants in U-2 OS cells can trigger centrosome over-duplication and induce the formation of aneuploidy, which is thought to be the major centrosomal root cause of microcephaly (Arquint and Nigg, 2014). In support of this notion, overexpression of PLK4 in mouse brains can induce the formation of supernumerary centrosomes and cause aneuploidy and apoptosis in neural progenitors, leading to significantly smaller brains (Marthiens et al., 2013). Moreover, amplified centrosomes have been observed in microcephaly patients with a truncating mutation in CEP135 (Hussain et al., 2012).

In addition, centrosome loss has been linked to the development of microcephaly. An elegant structural study revealed that the CPAP E1235V MCPH mutation leads to a remarkable decrease in the affinity of the interaction between CPAP and STIL (Cottee et al., 2013). Functionally, this mutation significantly perturbs the recruitment of CPAP to centrosomes, thereby causing defective centriole duplication (Cottee et al., 2013). Moreover, patient sequencing data reveals two loss-of-function mutations (F433L and R936S) in PLK4, a master regulator of centriole duplication (Martin et al., 2014). Expression of these mutants in Hela cells negatively affects PLK4‟s ability in centriole duplication (Martin et al., 2014). Consistently, a portion of patient derived fibroblasts carrying these mutations exhibit a significant decrease in centriole number alongside chromosome mis-segregation, cell cycle delay and apoptosis (Martin et al., 2014). It is well established that abnormal centriole number activates the p53-dependent surveillance pathway, which is recognized as a key mechanism for the loss of neural progenitor pools (Lambrus and Holland, 2017). In line with this notion, removal of p53 in centrosome deficient

34 mouse brains can bypass the surveillance pathway and rescue the loss of progenitors and neurons, therefore restoring the brain size (Insolera et al., 2014). Moreover, an alternative pathological mechanism centers on the control of mitotic spindle orientation. Abnormal spindle orientation has been linked to the perturbation of neural progenitor symmetric divisions, premature cell cycle exit and reduced neurogenesis (Lancaster and Knoblich, 2012). Depletion of ASPM and CDK5RAP2, two known microcephaly genes, has been found to cause severe mitotic spindle disorganization and led to loss of neuron-generating ability (Lizarraga et al., 2010, Gai et al., 2016), although their regulatory mechanisms remain unclear. Taken together, the centrosome number control, p53-dependent surveillance pathway and mitotic spindle orientation are well- recognized as the key centrosome root causes for the development of microcephaly, paving the way for its novel therapeutic treatments.

35

1.8 Summary and Rationale of this Thesis

Centrosomes act as the primary microtubule-organizing center in metazoan cells and are essential for a variety of cellular processes including cell signaling, division, shape and motility. The core of the centrosome is a pair of orthogonally arranged centrioles, which recruit pericentriolar material and template the formation of cilia/flagella. Given their essential cellular functions, centriole duplication must be precisely controlled to ensure a single round of duplication per cell cycle. Dysfunctions in this process are involved in a wide range of human disorders, including cancer, developmental brain diseases and ciliopathies. Despite their importance, the molecular mechanisms governing centriole duplication are not fully understood. My PhD research mainly focuses on studying the molecular basis of centrosome biogenesis and how abnormities in this process lead to human diseases. Towards this, our lab utilizes proximity- dependent biotinylation (BioID) to map the centrosome-cilium interface. With 58 bait proteins, we have generated a protein topology network comprising over 7000 interactions. This systematic profiling of proximity interactions provides a rich resource for a better understanding of centriole duplication, ciliogenesis, and centriolar satellite biogenesis. My PhD research is mainly built on this dataset to identify novel centrosome components and to further characterize their molecular functions in centriole assembly. In the first data chapter, through collaboration with the laboratory of Dr. Mark van Breugel‟s lab at the MRC in Cambridge UK, we analyze the BioID dataset to identify CEP85 as a novel interactor of STIL (a key centriole duplication factor), solve its atomic structure and reveal several conserved residues required for their interaction. Through targeted mutations, I demonstrate that the formation of the CEP85-STIL complex is essential for robust PLK4 activation and efficient centriole assembly in vivo. Thus, this work provides key structural and molecular insight into how mutations in centriole proteins may result in human diseases and illuminate, at the molecular level, how the CEP85-STIL interaction constitutes a crucial step in centriole duplication. In the second data chapter, I characterize the physiological relevance of the CEP85-STIL complex. To this end, my results reveal an unprecedented role of the CEP85-STIL complex in driving cancer cell motility through the control of ARP2/3-mediated actin organization. The major focus of my third data chapter is to determine the molecular mechanism by which CEP120 regulates the centriole length. Using BioID, I identify ANK2 as a novel interactor of CEP120, which is required for its centriole recruitment to mediate efficient centriole assembly. Altogether, the overarching goal of the thesis

36 presented here is to advance the mechanistic understanding of centriole biogenesis through the identification of proteins critical for this process and thus pave the way for new therapeutic approaches to treat human diseases with abnormal centrosomes.

37

2 Chapter II: Direct binding of CEP85 to STIL Ensures Robust PLK4 Activation and Efficient Centriole Assembly

38

2.1 Statement of Contributions

The data in this chapter has been published as an article in Nature Communications in 2018:

Direct binding of CEP85 to STIL ensures robust PLK4 activation and efficient centriole assembly

Yi Liu1,2, Gagan D. Gupta1, Deepak D. Barnabas3, Fikret G. Agircan1, Shahid Mehmood4, Di Wu4, Etienne Coyaud5, Christopher M. Johnson3, Stephen H. McLaughlin3, Antonina Andreeva3, Stefan M.V. Freund3, Carol V. Robinson4, Sally W.T. Cheung1, Brian Raught5,6, Laurence Pelletier1,2,&,#,*, Mark van Breugel3,#,*

1Lunenfeld-Tanenbaum Research Institute, University of Toronto, 600 University Avenue, Toronto M5G 1X5, Canada. 2Department of Molecular Genetics, University of Toronto, Toronto, Ontario, M5S 1A8, Canada. 3Medical Research Council-Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge CB2 0QH, UK. 4Department of Chemistry, University of Oxford, Oxford, UK 5Princess Margaret Cancer Centre, University Health Network, 101 College Street, Toronto, ON M5G 1L7, Canada. 6Department of Medical Biophysics, University of Toronto, Toronto, Ontario, M5G 1L7, Canada. * These authors contributed equally to this work.

Author contributions:

Protein purifications and the GST-based pull-down assays were conducted by M.V.B, the structural work, as well as the yeast-two-hybrid experiments by M.V.B. and D.D.B.. C.M.J. and S.H.M. performed the biophysical experiments (SEC-MALS, CD, AUC). S.F. conducted the NMR experiments; A.A. did the structural analyses and the evolutionary bioinformatics. The native mass-spectrometric experiments and cross-linking mass-spectrometric analyses were done by S.M. and C.V.R. and D.W., S.M. and C.V.R. respectively. Y.L. performed most of the cell biological experiments, imaging and in vivo characterization of the CEP85-STIL complex. Vectors for the STIL MT re-routing assay were made by D.D.B and F.G.A., and the assay was performed by D.D.B. and F.G.A., who also quantified the data. Creation of CEP85 cell lines, mass spectrometry and analyses were carried out by S.W.T.C., E.C., G.D.G., and B.R., G.D.G. assisted with the quantification of imaging data performed by Y.L. CRISPR-related reagents were prepared by J.M.T. The paper was written by Y.L., L.P., and M.V.B. with contributions from all authors. L.P. and M.V.B. directed the project.

39

2.2 Summary

The centrosome functions as the major microtubule-organizing center in animal cells and is composed of a single pair of centrioles that mark the site of pericentriolar material (PCM) recruitment to constitute centrosomes and are essential for the formation of cilia and flagella. Canonical centriole duplication occurs when PLK4 kinase is recruited around mother centrioles followed by the subsequent centriolar targeting of a conserved cascade of centriole duplication factors. The details of when and where PLK4 is activated and how the remaining components are recruited are currently incompletely understood. Combining protein proximity mapping with high-resolution structural methods, we identify CEP85 as a novel interactor of STIL that directly interacts with STIL through a highly conserved interaction interface involving a previously uncharacterized domain of STIL. Structure-guided mutational analyses in vivo reveal that this interaction is critical for efficient centriole loading of STIL, robust PLK4 activation and faithful daughter centriole assembly. Altogether, our results illuminate a molecular mechanism underpinning the spatiotemporal regulation of the early stages of centriole biogenesis.

40

2.3 Introduction

Centrosomes play an essential role in the organization of the cytoskeleton and chromosome segregation by participating in mitotic spindle assembly. Typically, one centrosome consists of two orthogonally arranged centrioles with a cylindrical array of microtubules that constitute their wall. When forming centrosomes, centrioles recruit, and give order to, the pericentriolar material (PCM), a proteinaceous matrix that nucleates microtubules. In addition, centrioles can dock at the plasma membrane and act as basal bodies to template the formation of cilia. Given their essential cellular roles, the centriole duplication cycle must be tightly regulated to ensure a single round of duplication per cell cycle. However, the mechanistic understanding of how new centrioles are assembled is far from concluded today.

Work across different organisms uncovered five evolutionarily conserved components with essential roles in centriole duplication: PLK4, CEP192, SASS6, STIL and CPAP in humans. Current knowledge indicates that the PLK4-STIL module plays a critical role in the initiation of centriole assembly. Molecularly, PLK4 regulates its activity through autophosphorylation to create a phosphodegron that is bound by the SCFSlimb/b-TrCP ubiquitin ligase for degradation (Arquint and Nigg, 2016). Active PLK4 phosphorylates STIL and recruits STIL to the mother centriole. In turn, binding of STIL activates PLK4 by promoting its autophosphorylation, which helps to trigger the downstream events of centriole formation (Ohta et al., 2014, Moyer et al., 2015). Despite these breakthroughs, key questions remain concerning how centrioles form at the right place and time, in particular what is not known is the molecular context in which this module is embedded and the nature of the mechanisms underpinning the tight control of the PLK4 kinase activity to ensure faithful initiation of centriole assembly.

To address these questions, I propose to leverage the proteomic analysis of those known centriole duplication factors in order to identify novel components. Using structural methods, biochemistry and super-resolution imaging, I will further determine their functional relationships with the PLK4-STIL module in the control of centriole assembly. The recently developed proximity-dependent biotinylation (BioID) is a unique and effective method to screen for the physiologically relevant protein interactions in living cells (Roux et al., 2013). In detail, BirA is an Escherichia coli biotin ligase with strong specificity for its substrate, whereas a mutant version (Arg118Gly, termed BirA*) has been shown to promiscuously biotinylate proteins in

41 close proximity of the enzyme (Roux et al., 2012). In BioID, BirA* is fused to the target proteins. When the fused proteins are expressed in cells in the presence of biotin, BirA* will trigger the biotinylation of the lysine residues of proteins proximal to the baits. Given that the covalent biotin linkage is stable under denaturing conditions, biotinylated proteins can be further identified by a combination of streptavidin affinity purification, followed by mass spectrometry (Roux et al., 2012). The key step of this method is that biotinylation occurs before solubilization, which helps to capture weak or transient interactions and overcome the insoluble nature of the centrosome baits. In 2015, our lab utilized BioID to map the centrosome-cilium interface and generate a protein topology network comprising over 7,000 interactions (Gupta et al., 2015). Analysis of interaction profiles coupled with high resolution phenotypic profiling implicates a number of protein modules in centriole duplication, ciliogenesis, and centriolar satellite biogenesis (Gupta et al., 2015). Based on this dataset, I identify CEP85 as a novel regulator that cooperates with PLK4 and STIL to modulate efficient centriole assembly.

CEP85 (also termed CCDC21) encodes a small protein of 762 amino acids with the predicted molecular weight of 85 kDa. To date, this gene has not been reported in any human diseases and its biological function is poorly understood. In 2011, a proteomic study unveiled CEP85 as a novel centrosome resident protein (Jakobsen et al., 2011). In this elegant work, purified centrosomes were subjected to mass spectrometry-based protein correlation profiling (PCP) for the identification of candidate proteins, and their centrosome localizations were further validated by microscopy (Jakobsen et al., 2011). In 2015, the molecular function of CEP85 was first indicated as a novel regulator of centrosome disjunction (Chen et al., 2015). That study validated the centrosome localization of CEP85 and identified its interaction with Nek2A through the middle coiled-coil domain (Chen et al., 2015). This interaction was found to antagonize Nek2A activity to regulate centrosome disjunction (Chen et al., 2015). However, two recent proteomic studies using CEP85 indicate that CEP85 is a significant potential interactor of PLK4, CEP192, CEP152, STIL, SASS6 and CPAP, which are the known centriole duplication factors (Gupta et al., 2015, Firat-Karalar et al., 2014). Furthermore, CEP85 has been shown to localize at the proximal end of mother centrioles, which is defined as the assembly site for the newborn centrioles (Chen et al., 2015). These findings led me to hypothesize CEP85 may play a role in centriole assembly.

42

To test this hypothesis, I combined the BioID, RNAi, CRISPR/Cas9, super-resolution imaging with structural and biochemical methods to characterize the biological function of CEP85. As a result, CEP85 was identified as a novel interactor of STIL that engages it across species through a highly conserved interaction interface involving a previously uncharacterized domain of STIL. Extensive structure-guided functional in vivo studies demonstrate unambiguously that this interaction is indispensable for efficient recruitment of STIL at the earliest stages of centriole duplication. More importantly, this, in turn, is crucial to ensure robust local PLK4 activation to drive the subsequent downstream events of centriole formation. Taken together, our findings elucidate the molecular basis behind a previously undescribed modulatory step during the most upstream events of centriole duplication.

43

2.4 Results

2.4.1 CEP85 is a Regulator of Centriole Duplication

To identify novel factors implicated in centriole biogenesis, we originally sought to map protein interactions of centriole components using proximity-dependent biotin identification, including the upstream centriole duplication factors STIL, SASS6, CEP152 and CEP63 (Gupta et al., 2015). Since CEP192 was not initially included as part of this study, we therefore submitted it to BioID exactly as performed previously (Figure 2.1A). The combined analysis of this proximal interaction network revealed that CEP85, a centrosomal protein so far only implicated in regulating centrosome disjunction through its interaction with Nek2A (Chen et al., 2015), displayed a prominent proximity signature with a number of centriole duplication factors (Figure 2.1A). Collectively, these data suggest a potential role of CEP85 in centriole duplication.

To directly test for this possibility, we therefore decided to investigate CEP85‟s function in centriole duplication. First, I depleted endogenous CEP85 in U-2 OS cells using siRNA and western blot experiments confirmed the knockdown efficiency (Figure 2.1D and G). Using CEP192 and Pericentrin (PCNT) as centrosome markers, I found a marked decrease in centrosome number in CEP85 RNAi-treated cells compared to control, and this defect could be rescued by the expression of an RNAi-resistant CEP85 transgene (Figure 2.1B-C). It has been indicated that up- and down-regulation of CEP85 affects centrosome separation at prometaphase, which may indirectly impair centriole duplication in the following cell cycle. To rule out this possibility, I modified the canonical centriole duplication assay to add Hydroxyurea (HU) after siRNA transfection to synchronize cells at S-phase for 24 hours, and next release them into G2 arrest by CDK1 inhibitor for another 18 hours before methanol fixation, thereby blocking the entry into mitosis. I performed this modified assay in U-2 OS cells stably expressing tetracycline (Tet) inducible siRNA-resistant human CEP85, and Centrin was used to label centrioles through imaging. Here, four centrioles are present upon completion of centriole duplication. My results indicate that depletion of CEP85 led to a significant decrease in the number of Centrin foci in G2 cells, a phenotype I could rescue by expressing an RNAi-resistant CEP85 (Figure 2.1E-F). Consistently, pronounced centriole duplication defects were also observed using PLK4 overexpression-induced centriole over-duplication assays (Figure 2.2A-B), S-phase arrest induced centriole over-duplication assays (Figure 2.2C-D). And transient CRISPR-mediated

44 disruption of the CEP85 genomic locus caused defects in centriole number comparable to CEP192 knockout (Figure 2.2E-G). Overall, these multiple lines of evidence indicate that CEP85 is a positive regulator of centriole duplication.

My next step was to determine at which step in centriole formation that CEP85 operates. To serve this aim, I depleted endogenous CEP85 in U-2 OS cells by siRNA. Subsequently, cells were enriched in S-phase through the addition of hydroxyurea and submitted to immunofluorescence (IF) analysis to assay for the recruitment of CEP192, PLK4, STIL, SASS6 and Centrin, which are known to play a role in centriole formation. The results indicated that CEP85 depletion resulted in increased centrosomal levels of PLK4 and at the same time to recruitment defects of STIL as well as downstream factors SASS6 and Centrin, while CEP192 levels were not significantly affected, when compared to the control (Figure 2.2H-I). To characterize the requirements for CEP85 localization to the centrioles, endogenous CEP192, CEP152, PLK4 and STIL were depleted in U-2 OS cells using the same method. Here, I observed that depletion of CEP192, CEP152 and PLK4 all resulted in a marked reduction in the centriolar loading of CEP85, while STIL depletion resulted in around 20% increase in CEP85 levels at centrioles (Figure 2.3A-B), implying a potential feedback regulation of CEP85 levels that requires future investigation. Collectively, these data indicate that CEP85 acts downstream of CEP192, CEP152 and PLK4 to play an important role in regulating the STIL localization and centriole assembly.

45

46

Figure 2.1. CEP85 is required for centriole duplication.

(A). Spring-embedded BioID network of centriole duplication factors. Shown are high- confidence interactors detected with two or more bait proteins. Bait, interactor and edges are colour-coded as indicated in the legend. Edge thickness is proportional to total peptide counts.

(B-D). U-2 OS cells expressing Tet-inducible FLAG or the siRNA-resistant FLAG-CEP85 transgene were transfected with control or CEP85 siRNA and induced with tetracycline for 72 h. Scale bar 10 μm, white boxes indicates the magnified region. (C). The graph shows the percentage of cells with the indicated centrosome numbers (n = 200/experiment, three independent experiments). (D). Western blot analysis of FLAG-CEP85 protein levels. α-tubulin served as a loading control.

(E-G). Impact of CEP85 depletion on centriole number. The G2-phase arrest assays (See the Materials and Methods) were performed in U-2 OS cells conditionally expressing FLAG or siRNA-resistant FLAG-CEP85, treated with control or CEP85 siRNA and induced with tetracycline for 72 h. Scale bar 10 μm, white boxes indicates the magnified region. (F). Bar graph, the number of centrioles per cell were counted. (n = 200/experiment, three independent experiments). (G). Western blot showing the levels of FLAG-CEP85 in control or CEP85 siRNA transfected cells. α-tubulin served as a loading control.

47

48

Figure 2.2. CEP85 is a regulator of centriole duplication.

(A, B). The PLK4 assays were performed as described in Materials and Methods. Scale bar 10 μm, white boxes indicates the magnified region. (B). The graph showing the percentage of cells with over four centrioles (n = 100/experiment, three independent experiments).

(C, D). The role of CEP85 in S-phase arrest induced centriole overduplication. The assays were performed as described in Materials and Methods. Scale bar 10 μm, white boxes indicates the magnified region. (D). The graph indicates the percentage of cells with over four centrioles (n = 100/experiment, three independent experiments).

(E-G). Western blot and IF analysis of lentivirus control (vector), lentivirus CRISPR CEP85 or CEP192 guide RNAs treated RPE-1 cells, labelled with the indicated antibodies. Selected images showing Centrin and γ-tubulin labelling. Scale bar 10 μm, white boxes indicate the magnified region. (G). The graph shows the percentage of cells with the indicated centriole numbers (n = 100/experiment, three independent experiments). Two-tailed t-test was performed for all p- values, all error bars represent S.D., and asterisks for p-values are **p<0.01 and *p<0.05.

(H, I). Determining at which stage CEP85 acts in the centriole duplication pathway using the PLK4-induced centriole overduplication assays (See the Materials and Methods). Scale bar 10 μm, white boxes indicates the magnified region. (I). The graph indicates the relative levels of CEP192, Myc-Plk4, STIL, SASS6 and Centrin at centrosomes (n = 100/experiment, three independent experiments). Two-tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p-values are **p<0.01 and *p<0.05.

49

2.4.2 CEP85 Is Required for STIL Localization and PLK4 Activation

Activation of PLK4 at centrioles plays a critical role in the initiation of centriole assembly. Active PLK4 phosphorylates itself, thereby creating a phospho-degron that results in PLK4 degradation in vivo (Arquint and Nigg, 2016). On this basis, increased PLK4 levels at centrioles under conditions of CEP85 depletion as shown in Figure 2.2H-I could therefore indicate a reduced activity of PLK4. To test this hypothesis, I depleted CEP85 in synchronized cells and assessed PLK4 activity by quantitative IF analysis using a phospho-specific PLK4 antibody that targets its auto-phosphorylated residue S305. Inhibition of PLK4 kinase activity under Centrinone treatment, a potent PLK4 inhibitor, demonstrated the specificity of the antibody in this assay (Figure 2.3C-D). Here, I observed that upon CEP85 depletion centrosomal levels of PLK4 were increased two-fold, whereas relative levels of active PLK4 (pS305/PLK4) were decreased (Figure 2.3E-G). This inactive pool of PLK4 appears to be stabilized as I consistently detected a marked increase in the total cellular levels of PLK4 upon CEP85 depletion (Figure 2.4 A-B). These data suggest that CEP85 is required for robust PLK4 activation.

Next, I ought to determine the effect of CEP85 depletion on STIL localization. STIL is a low abundance protein and there are few effective commercial antibodies worked well for IF staining. To overcome this limit, van Breugel‟s lab utilized their purified STIL-N recombinant proteins to generate a polyclonal anti-STIL antibody. To determine its specificity in IF staining, I pre- absorbed the STIL antibody with its immunizing antigen and the signal of STIL at centrioles was significantly reduced, thus validating its specificity (Figure 2.4D-E). To determine whether CEP85 depletion also affects the global levels of STIL in the absence of PLK4 overexpression, I knocked down CEP85 in U-2 OS cells using the same method. Transfected cells were treated with hydroxyurea to arrest them in S-phase prior to fixation and immunolabeling for STIL and PCNT to mark the position of centrioles. In line with the PLK4 overexpression data (Figure 2.2H-I), I found a marked reduction in the centrosomal level of STIL upon CEP85 depletion (Figure 2.4F-G). Defects in STIL recruitment coincided with a marked decrease in the total cellular levels of STIL (Figure 2.4A-C and H-I), suggesting that CEP85 depletion might reduce the total amount of bioavailable STIL. Previous work has revealed that STIL acts as an upstream regulator of PLK4, contributing to its robust activation. Taken together with my results, I hypothesize that CEP85 may play a direct role in the regulation STIL to mediate full activation of PLK4 kinase.

50

51

Figure 2.3. CEP85 is required for PLK4 activation.

(A, B). Factors required for the recruitment of CEP85 to centrioles. Centriolar levels of CEP85 during centriole duplication were assessed using the PLK4-induced centriole overduplication assays (See the Materials and Methods). Selected images of CEP85 and PCNT labelling are shown. Scale bar 10 μm, white boxes indicates the magnified region. (B). The graph indicates the relative intensity of CEP85 at centrosomes (n = 100/experiment, three independent experiments). Two-tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p- values are **p<0.01 and *p<0.05.

(C, D). The PLK4-induced centriole overduplication assay was performed as described in Materials and Methods. DMSO andcentrinone B (2500 nM) were added for 24 h berfore fixation. Cells were labelled with DAPI and the indicated antibodies. Scale bar 5 μm, white boxes indicate the magnified region. (D). The graph shows the relative levels of PLK4 pS305 at centrosomes (n = 100/experiment, three independent experiments.

(E-G). Examining the impact of CEP85 depletion on PLK4 activation using the PLK4-induced centriole overduplication assays (See the Materials and Methods). Selected images showing Myc-PLK4 and PLK4 pS305 labelling. Scale bar 10 μm, white boxes indicate the magnified region. (F, G). The graph indicates the levels of Myc-PLK4 and the relative ratio of pS305/PLK4 at centrosomes after depletion of endogenous CEP85. (n = 200/experiment, three independent experiments).

52

53

Figure 2.4. CEP85 is required for PLK4 activation and STIL localization to centrosomes.

(A-C). Western blot showing the levels of Myc-PLK4, CEP85 and STIL in control or CEP85 siRNA transfected cells. Cyclin A was used as a cell cycle marker and α-tubulin served as a loading control. (B, C). Quantification of protein levels shown in D (α-tubulin normalized, n = 2/experiment, five independent experiments).

(D, E). The anti-STIL antibody was pre-absorbed with the corresponding antigen before the IF experiments in U-2 OS cells. Selected images indicating STIL and Centrin labelling. Scale bar 1 μm. (d). Quantification showing the relative levels of STIL at centrosomes (n = 100/experiment, three independent experiments).

(F, G). IF analysis of STIL localization in control or CEP85-depleted cells. The S-phase arrest assays were performed as described in Materials and Methods, followed by labelling with DAPI and the indicated antibodies. Scale bar 10 μm, white boxes indicate the magnified region. (G). Quantification showing the relative levels of STIL at centrosomes in PCNA-positive cells (n = 200/experiment, three independent experiments).

(H, I). Western blot indicates the levels of STIL and CEP85 in control or CEP85 siRNA treated cells. Cyclin A and α-tubulin served as a cell cycle marker and loading control, respectively. (I). Quantification of the indicated protein levels shown in I (α-tubulin normalized, n = 2/experiment, five independent experiments). Two-tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p-values are **p<0.01 and *p<0.05.

54

2.4.3 CEP85 and STIL Recruitment during Centriole Duplication

Analysis of the cellular distribution of CEP85 in interphase revealed that it co-localized with CEP192 at the proximal end of mother centrioles, which is the site of new centriole assembly (Figure 2.5A). It is established that STIL is degraded in G1 and recruited again at the onset of centriole duplication during S-phase (Arquint et al., 2012). For this reason, I assessed the spatial distribution of CEP85 and STIL in S-phase. In this analysis, I found around 15% of centrioles with CEP85 present in the vicinity of STIL and about 85% of centrioles where CEP85 and STIL were found adjacent to but optically resolvable from each other (Figure 2.5B-C).

To further determine the spatial localization of CEP85 and STIL beyond the diffraction limit, I utilized 3D-SIM imaging and observed that CEP85 accumulated around mother centrioles, with a small fraction of cells harbouring a dot-like STIL pattern that overlapped with CEP85 (Figure 2.5D), implying that CEP85 may transiently associate with STIL at centrioles. To explore this possibility, I induced centriole over-duplication by overexpressing PLK4 in S-phase arrested cells and conducted IF analysis to assess the spatial accumulation of PLK4, STIL and CEP85 at early and later stages of centriole duplication. The 3D-SIM imaging data revealed that a proportion of CEP85 overlapped with STIL and PLK4 in a near ring-like pattern (Figure 2.5E-F). Following the centriole elongation, the STIL ring-like structure was resolved into a flower-like structure, however, CEP85 retained it localization in the vicinity of PLK4 (Figure 2.5E-F). Altogether, these data support the notion that the proportion of CEP85 that overlaps with STIL is the highest early on during new centriole assembly to mediate robust PLK4 activation, prior to centriole elongation.

55

56

Figure 2.5. CEP85 colocalizes with STIL at an early stage of centriole duplication.

(A). Determining the centriolar localization of CEP85. Cells were transfected with the GFP- CEP85 construct and stained with the indicated antibodies. Scale bar 0.5 μm. Schematic representation of CEP85 (green) localization on the proximal end of mother centrioles.

(B, C). IF analysis of CEP85 and STIL localization in U-2 OS cells after 24 h S-phase arrest using thymidine (1 mM). Selected images showing CEP85 and STIL labelling. (B). Bar graph, the percentage of centrioles with different localization patterns (n = 100/experiment, three independent experiments). Scale bar 0.5 μm.

(D). 3D-SIM micrographs of S-phase arrested (1 mM thymidine) U-2 OS cells stained with the indicated antibodies.

(E, F). 3D-SIM micrographs in U-2 OS Tet-inducible Myc-PLK4 cells after 24 h PLK4 induction and S-phase arrest using thymidine (1 mM), showing CEP85, STIL and Myc-PLK4 staining. PLK4 ring structures were encircled by white dashes. Scale bar 0.5 μm. (F). The graph indicates the fraction colocalization between CEP85 and STIL (n = 20/experiment, two independent experiments). All error bars represent S.D..

57

2.4.4 CEP85 Interacts Directly With the N-terminal Domain of STIL

Previous studies on the CPAP-STIL complex raised the possibility that the N-terminal domain (NTD) of STIL is located towards the periphery of the centriole cylinder and therefore might be close to the location of CEP85. To check whether CEP85 could associate with STIL, the van Breugel lab performed pull-down experiments with immobilized recombinant GST-STIL NTD and cell lysates expressing different FLAG-tagged CEP85 constructs. The results shown in Figure 2.6A-B demonstrated that CEP85 indeed was able to interact with the STIL NTD, which was dependent on the C-terminal region of CEP85. Moreover, yeast-two hybrid assays indicated that this region binds directly to the STIL NTD (Figure 2.6A and C). Further isothermal titration calorimetry (ITC) experiments (Figure 2.6D and E) and analytical ultracentrifugation (AUC) experiments (Figure 2.6F) with recombinant proteins allowed a fine-mapping of the binding region of CEP85 to a coiled-coil domain within this C-terminal region (cc4), depending on the experimental conditions, suggested a binding affinity of ~14-65 μM for this construct (Figure 2.6D-F). These ITC and AUC experiments were performed with chicken STIL NTD (66% identical to human STIL NTD, Figure 2.12B) since human STIL NTD was found to have an aggregation tendency at room temperature.

58

59

Figure 2.6. The N-terminal domain of STIL directly interacts with a C-terminal coiled-coil domain of CEP85.

(A). Domain overview of human STIL and CEP85. CR, conserved region; cc, coiled-coil; NTD, N-terminal domain. Constructs that were used in this work are indicated by red and green lines.

(B). STIL NTD and CEP85 interact. The C-terminal region of CEP85 is required for this interaction. Western blot showing a pull-down experiment with immobilized, recombinant GST or GST-human STIL NTD and lysates from tissue culture cells overexpressing the indicated 3xFLAG-tagged CEP85 constructs.

(C). The C-terminal region of CEP85 shows a yeast two-hybrid interaction with human STIL NTD. Yeast transformed with the indicated bait and prey constructs were plated on SC-Leu/-Trp plates (selecting for bait and prey plasmid, left) and on SC-Ura plates (selecting for Ura promoter activation, right).

(D). Recombinant chicken STIL NTD and human CEP85 cc4 directly interact with micromolar affinity. Typical ITC of chicken STIL NTD and human CEP85 cc4 binding at 25 °C. The resulting KD, ΔH and STIL NTD/CEP85 cc4 binding stoichiometry (N) as an average from a total of seven independent measurements are indicated (± standard deviation).

(E). Recombinant chicken STIL NTD and human CEP85 cc4 directly interact with micromolar affinity and form a 2:2 complex. Typical ITC binding isotherm for chicken STIL NTD titrated into human CEP85 cc4 at 10 °C. The resulting KD, ΔH and STIL NTD/CEP85 cc4 binding stoichiometry (N) as an average of three independent measurements are indicated (± standard deviation). In contrast, measurements at 25 °C (Figure 2.6D) apparently indicate that the binding is exothermic but these measurements might be distorted by the presence of low levels of unfolded CEP85 cc4 monomer in the ITC cell able to refold upon titration with the stabilising STIL NTD binding partner, thereby contributing a large exothermic heat effect. This effect could also affect the apparent stoichiometry of binding under these conditions that was lower than observed at 10 °C. Residue numbering is based on the used STIL homologues. Where applicable, the equivalent human STIL residue numbers are indicated in brackets.

60

(F). Sedimentation equilibrium analysis of the molecular weights of human CEP85 cc4 and chicken STIL NTD alone and in complex in solution. Filled circles show the sedimentation equilibrium profiles at different speeds, solid lines display the fits of the data using either an average mass model or a heterodimeric association model in which two monomers of chicken STIL NTD interact with a dimer of human CEP85 cc4. Derived molecular weights and binding affinity are indicated.

2.4.5 Structural Characterisation of the STIL-CEP85 Interaction Region

To further characterize the CEP85-STIL interaction, the higher-resolution information of this complex would be highly beneficial. To serve this aim, the van Breugel lab first revealed the structures of the STIL NTD (from Trichoplax adhaerens, 28 % identical to Homo sapiens STIL NTD, Figure 2.12B) and CEP85 cc4 (human) at a resolution of 2.1 and 1.7 Å respectively using X-ray crystallography (Figure 2.7A-B, Table1-2).

Subsequently, we also determined the structure of the complex between Trichoplax adhaerens STIL-NTD and human CEP85 cc4 using X-ray crystallography to a resolution of 4.6 Å (Figure 2.7A-B, Table1, 3 and 4). The complex structure shows a 2:2 binding ratio between one STIL NTD bound to each of the two coiled-coil strands of the parallel CEP85 cc4 dimer (Figure 2.7C). When compared to the unbound CEP85 coiled-coil structure (Figure 2.7B), we did not identify any density for the N-terminal half of the coiled-coil domain, probably due to its partial unfolding in the crystal. Consistently, both ITC binding studies and analytical ultracentrifugation with human CEP85 cc4 and chicken STIL NTD are best in agreement with a 2:2 binding stoichiometry between both proteins (Figure 2.6E-F). Collectively, these findings provide key structural insights into the further mechanistic characterization of the CEP85-STIL complex in centriole duplication.

61

62

Figure 2.7. STIL NTD interacts with CEP85 cc4 through a conserved interface.

(A). High-resolution structure of Trichoplax adhaerens STIL NTD. Left, as ribbon presentation, rainbow coloured from N- to C-terminus. α-helices (α), β-sheets (β) and loops (L) are labelled consecutively. Right, equivalent view as molecular surface coloured by CONSURF evolutionary conservation score (right) from unconserved (cyan) to highly conserved (burgundy), revealing the presence of two conserved patches on the surface of Trichoplax adhaerens STIL NTD. Ringed in blue are residues L47 and R50 (L64 and R67 in human STIL) that are part of patch 1 and are located at the interaction interface in the CEP85-STIL complex as shown in C.

(B). High resolution structure of human CEP85 cc4 as ribbon presentation (rainbow coloured from N- to C-terminus) (left) and as molecular surface coloured by CONSURF evolutionary conservation score (right). The dashed box indicates the CEP85 cc4 region with visible electron density in the CEP85-STIL complex (shown in C). Residues Q640 and E644 that are found at the interface of this complex are ringed in blue.

(C). Complex of Trichoplax adhaerens STIL NTD and human CEP85 cc4 at a resolution of 4.6 Å, as a ribbon presentation. The correct register of the CEP85 cc4 was obtained by introducing methionines into CEP85 cc4 and calculating phased anomalous difference maps from corresponding selenomethionine-derivate datasets. The human equivalents of the Trichoplax adhaerens STIL residues are indicated in brackets.

(D). Mutation of conserved interface residues in STIL NTD and CEP85 cc4 compromises binding affinity. ITC measurements of chicken STIL NTD titrated into human CEP85 cc4 at 25 °C. Excess heat observed for the mutant proteins approach that obtained when titrating wild- type (WT) STIL NTD into buffer alone as a non-binding control. Conserved residues mutated in chicken STIL NTD are L60A and R63A, equivalent to L64 and R67A in human STIL (in brackets) and L47 and R50 in Trichoplax adhaerens (Figure 2.7A).

63

2.4.6 CEP85-STIL Interaction is required for STIL Localization and Centriole Duplication

Our complex structure shows the presence of highly conserved residues in the binding interface between the STIL NTD and CEP85 cc4. When we introduced single alanine mutations of two of these residues either into the STIL NTD (L64A or R67A) or into CEP85 cc4 (Q640A or E644A), the van Breugel lab found that these mutations abolished the binding of the recombinant proteins in vitro as judged by ITC experiments (Figure 2.7D). In pull-down assays with cell lysates, binding of recombinant GST-STIL NTD to full length CEP85 was also strongly compromised in the corresponding mutants (Figure 2.8A and Figure 2.9A).

To further confirm that these observations are relevant for the interaction of the full length proteins in vivo, we artificially targeted mCherry-tagged CEP85 wild type, Q640A or E644A to microtubules by fusing them in frame to the microtubule-binding domain of MAP7 (3xFLAG- MAP717-282-mCherry-CEP85. MAP717-282 comprises MAP7‟s microtubule-binding domain) and then assessed whether GFP tagged STIL would follow CEP85 to the microtubule cytoskeleton. The results shown in Figure 2.8H-J indicate that in this assay, only wild-type CEP85 robustly recruited STIL to microtubules. In contrast, the CEP85 Q640A, E644A and double mutants were strongly compromised in their ability to re-route STIL, although wild-type and mutant proteins were expressed at similar levels (Figure 2.8H-J).

Next, I used these mutations to check whether the putative STIL NTD-CEP85 interaction is functionally important for CEP85‟s role in centriole duplication. To this end, I depleted endogenous CEP85 in U-2 OS cells and rescued it either with wild-type CEP85 or with the CEP85 Q640A, the E644A or the combined Q640A and E644A mutants. Firstly, I found that this panel of CEP85 mutants did not appear to negatively impact the overall levels of CEP85 or its ability to localize at centrosomes (Figure 2.8B-D). Subsequently, I assayed for the effect of these mutants on STIL recruitment to centrioles and for the ability of centrioles to duplicate (Figure 2.8E-G). As expected, these experiments indicated that in the absence of CEP85 both centriole duplication and STIL recruitment were perturbed, and the expression of CEP85 Q640A and E644A double mutant was not able to fully rescue both activities when compared to wild-type CEP85. These results suggest that a functional CEP85-STIL binding interface is required for both centriole duplication and STIL localization to centrosomes.

64

Conversely, I then asked whether similar defects in STIL recruitment and centriole duplication would also be observed with the STIL point mutations that are defective for CEP85 binding. First, I used the L64A and R67A double mutant of STIL and examined how it would affect STIL recruitment to centrioles in U-2 OS cells. The experiments shown in Figure 2.9B-C demonstrate that, compared to wild type STIL, this mutant displayed significantly reduced centriole localization, further hinting towards a role of the described CEP85-STIL interaction in facilitating STIL recruitment to centrioles. Subsequently, I assayed for the ability of the L64A, the R67A as well as the L64A + R67A STIL mutants to act in centriole duplication. To this end, I generated a STIL overexpression system in STIL CRISPR knock-out cells (that do not have centrioles) and assayed for the restoration of centriole formation in these cells (Figure 2.9D-E). My results suggested that both STIL L64A and R67A mutants were not able to fully restore centriole duplication to levels comparable to wild-type STIL (Figure 2.9D-E). Subsequently, I assessed the ability of these STIL mutants to rescue centriole duplication at lower STIL expression levels in cells that were siRNA depleted for STIL. Consistently, the expression of these STIL mutants could not rescue the centriole duplication defect in STIL depleted cells, when compared to wild-type STIL (Figure 2.9F). Together, my results indicate that the binding of CEP85 to STIL is essential for STIL localization to centrosomes and its ability in centriole assembly.

In addition, my previous data suggest that CEP85 depletion results in a marked reduction in total cellular levels of STIL, raising the possibility that the requirement for CEP85 could be bypassed by STIL overexpression. To assess this possibility, I depleted endogenous CEP85 in U-2 OS cells and tested the ability of wild-type STIL and the STIL L64A and R67A double mutant to rescue centriole duplication. My results suggest that expression of wild-type STIL and the STIL mutant were unable to rescue the centriole duplication defects in CEP85 depleted cells (Figure 2.10A-C). In line with this observation, I also observed that the overexpression of a non- degradable form of the STIL L64A+R67A double mutant (that are unable to interact with CEP85), harbouring the previously described Val1219X mutation, was unable to trigger centriole amplification to wild-type STIL levels (Figure 2.10D-F), thereby supporting a dual role for CEP85 in the recruitment of STIL to centrosomes and its stability.

65

66

Figure 2.8. The interaction between CEP85 and STIL is essential for centriole duplication.

(A). The interface residues described in Figure 2.7 are required for an efficient interaction of human STIL NTD and CEP85. Western blot showing a pull-down experiment with immobilized, recombinant GST or GST-human STIL NTD, and lysates from cells overexpressing 3xFLAG- tagged CEP85 (WT and mutants) as indicated.

(B, C). IF analysis of CEP85 localization. Tetracycline and hydroxyurea were added to induce the expression of FLAG-CEP85 transgenes and to arrest cells in S-phase for 24 h before fixation. Cells were labelled with the indicated antibodies. Scale bar 10 μm, white boxes indicate the magnified region. (C). Quantification showing the relative levels of FLAG-CEP85 (WT and mutants) at centrosomes. (D). Western blot analysis of the indicated protein levels. α-tubulin served as a loading control.

(E). Effect of CEP85 mutations on centriole duplication. U-2 OS cells conditionally expressing FLAG or the siRNA-resistant FLAG-CEP85 (WT and mutants) were treated with control or CEP85 siRNA and induced with tetracycline for 72 h. The G2-phase arrest assays were performed as described in Materials and Methods. Quantification showing the percentage of cells with the indicated centriole number (n = 100/experiment, three independent experiments).

(F, G). The role of CEP85 mutations in STIL localization. The S-phase arrest assays (See the Materials and Methods) were performed using U-2 OS cells expressing Tet-inducible FLAG or the siRNA-resistant FLAG-CEP85 (WT and mutants) and tetracycline was added for 72 h. Scale bar 10 μm, white boxes indicates the magnified region. (G). Quantification showing the relative levels of STIL at centrosomes (n = 100/experiment, three independent experiments).

(H, I). Confirming the interaction between CEP85 and STIL in vivo. U-2 OS STIL CRISPR KO cells were co-transfected with GFP-STIL and MAP7-mCherry-CEP85 (WT and mutants) for 24 h. Cells were fixed with 4% PFA and stained with DAPI. (I). The graph indicates the percentage of cells with STIL recruited to microtubules.

(J). Western blot shows the levels of GFP-STIL and MAP7-mCherry-CEP85 (WT and mutants). Two-tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p- values are **p<0.01 and *p<0.05.

67

68

Figure 2.9. The interaction between CEP85 and STIL is required for STIL localization and centriole assembly.

(A). The interface residues described in Figure 5 are required for an efficient interaction of human STIL NTD and CEP85. Western blot showing a pull-down experiment with immobilized, recombinant GST or GST-human STIL NTD (WT and mutants), and lysates from tissue culture cells overexpressing 3xFLAG-tagged CEP85 as indicated.

(B, C). The role of CEP85 binding in STIL centriolar localization. U-2 OS cells expressing Tet- inducible FLAG or the siRNA-resistant FLAG-STIL (WT and mutants) transgene were transfected with control or STIL siRNA for 72 h and tetracycline (2 μg/mL) was added for the final 48 h before fixation. The S-phase arrest assays were performed as described in Materials and Methods. Cells were labelled with DAPI and the indicated antibodies. Scale bar 10 μm, white boxes indicate the magnified region. (C). Quantification showing the relative levels of FLAG-STIL at centrosomes (n = 100/experiment, three independent experiments).

(D, E). Effect of STIL mutations on de novo centriole formation. STIL CRISPR knockout U-2 OS cells were transfected with FLAG-STIL (WT and mutants) for 72 h. Selected images showing STIL, Centrin and γ-tubulin labelling. Scale bar 10 μm. (E). The graph shows the percentage of cells with restored centrioles (n = 100/experiment, three independent experiments).

(F). Impact of STIL mutations on centriole duplication. U-2 OS cells conditionally expressing FLAG or siRNA-resistant FLAG-STIL (WT and mutants) were used to perform the S-phase arrest assays (See the Materials and Methods). Quantification showing the percentage of cells with the indicated centriole number. (n = 100/experiment, three independent experiments). Two- tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p- values are **p<0.01 and *p<0.05.

69

Figure 2.10. STIL overexpression does not bypass the requirement for CEP85 in centriole duplication.

(A). U-2 OS cells expressing Tet-inducible FLAG or the siRNA-resistant FLAG-STIL (WT and mutants) transgene were transfected with control or CEP85 siRNA for 72 h and tetracycline (2 μg/mL) was added for the final 48 h before fixation. The S-phase arrest assays were performed as described in Materials and Methods. Cells were labelled with DAPI and the indicated antibodies. Scale bar 10 μm, white boxes indicate the magnified region.

(B). Quantification showing the percentage of cells with four centrioles (n = 100/experiment, three independent experiments).

70

(C). Western blot showing the levels of CEP85 and STIL in control or CEP85 siRNA transfected cells. α-tubulin served as a loading control.

(D). U-2 OS cells were transfected with Myc-STIL Val1219X (WT and mutants) for 48 h. Selected images showing Myc-STIL and Centrin labelling. Scale bar 10 μm, white boxes indicate the magnified region.

(E). The graph shows the percentage of cells with over four centrioles (n = 100/experiment, three independent experiments).

(F). Western blot showing the levels of Myc-STIL Val1219X (WT and mutants). α-tubulin served as a loading control. Two-tailed t-test was performed for all p-values, all error bars represent S.D., and asterisks for p-values are **p<0.01 and *p<0.05.

2.4.7 CEP85 and STIL Binding is Required for PLK4 Activation

Given the recently identified role of STIL in PLK4 activation, I then asked whether the CEP85 point mutants also affect PLK4 activity in vivo. To assess this possibility, I utilized lentiviruses to stably express GFP-tagged wild-type CEP85 and the CEP85 Q640A + E644A mutant in PLK4 overexpressing cells and measured their effect on centriole amplification. Using this system, I observed around 70% of cells overexpressing GFP alone were able to over-duplicate their centrioles. Although the expression of wild-type CEP85 had a minor negative effect on centriole duplication, expression of similar levels of the CEP85 Q640A+ E644A mutant had a more pronounced suppressive effect on PLK4‟s ability to trigger centriole over-duplication (Figure 2.11A and B), which was accompanied by a marked increase in the total cellular level of PLK4 (Figure 2.11C and D). Subsequently, I synchronized cells and assayed PLK4 activity by immunofluorescence using a phospho-specific PLK4 antibody that recognizes its auto- phosphorylated residue S305. Consistent with my previous results of CEP85 depletions (Figure 2.3E-G), I found that expression of the CEP85 Q640A + E644A mutant also increased the centrosomal levels of PLK4 and at the same time resulted in decreased PLK4 activity levels as determined by a lowered ratio of S305P/PLK4 levels (Figure 2.11E-G). Collectively, our results suggest that the CEP85-STIL interaction is required to mediate efficient PLK4 kinase activation to mediate centriole duplication.

71

72

Figure 2.11. The interaction between CEP85 and STIL is essential for PLK4 activation and PLK4-induced centriole overduplication.

(A, B). Effect of CEP85 mutations on centriole overduplication. U-2 OS cells conditionally expressing Myc-PLK4 and constitutively expressing GFP-CEP85 (WT and mutant) were used to perform the PLK4-induced centriole overduplication assays. Cells were labelled with DAPI and the indicated antibodies. Scale bar 10 μm, white boxes indicate the magnified region. (B). Quantification showing the percentage of cells with over four centrioles (n = 100/experiment, three independent experiments).

(C). Western blot analysis indicating the Myc-PLK4 and GFP-CEP85 protein levels using the PLK4 assays described in A. Cyclin A was used as a cell cycle marker and α-tubulin served as a loading control.

(D). Quantification of protein levels shown in C (α-tubulin normalized, n = 2/experiment, six independent experiments).

(E). Effect of CEP85 mutations on PLK4 activation. The PLK4-induced centriole overduplication assays were performed as described in A. Selected images showing Myc-PLK4 and PLK4 pS305 labelling. Scale bar 10 μm, white boxes indicate the magnified region. (F, G) The graph indicates the relative levels of Myc-PLK4 and the relative ratio of pS305/PLK4 at centrosomes. (n = 100/experiment, three independent experiments).

(G). A model for how CEP85 operates in the centriole duplication pathway. CEP85 acts downstream of CEP192, CEP152 and PLK4 in centriole duplication. Direct binding of CEP85 to STIL stabilises STIL and facilitates its recruitment to centrosomes. This further facilitates robust PLK4 activation and subsequent centriole assembly. Two-tailed t-test was performed for all p- values, all error bars represent S.D., and asterisks for p-values are **p<0.01 and *p<0.05.

73

74

Figure 2.12. Evolutionary conservation of CEP85 and STIL NTD.

(A). Distribution of CEP85 and STIL across different representative metazoan organisms. CEP85 and STIL are shown as boxes, coloured in red and blue, respectively. The STIL NTD is shown as a dark blue box. Sas5 (STIL‟s functional homolog in C. elegans) is shown in a slightly different colour as its sequence does not have a detectable similarity to sequences of other STIL homologues. Insects such as flies, wasps, ants, butterflies, beetles and bees lack the STIL NTD. Their STIL domain organization and sequences are akin to Drosophila Ana2 (its STIL homologue). We were not able to identify CEP85 homologs in these organisms.

(B). Multiple sequence alignment of the STIL NTD homologues from Trichoplax adhaerens, Homo sapiens and Gallus gallus that were used in this study. The alignment is coloured according to the CLUSTAL coloring scheme, residue color intensity is based on conservation. The secondary structure elements are shown above the alignment. The two STIL residues mutated in this study are indicated with a red dot.

(C). Multiple sequence alignment of the CEP85 cc4 of CEP85 homologues from Homo sapiens, Gallus gallus and Trichoplax adhaerens. The alignment is coloured according to the CLUSTAL coloring scheme, residue color intensity is based on conservation. The two CEP85 residues mutated in this study are indicated with a red dot.

(D). Conservation of the STIL NTD: CEP85 interaction interface. The resolution of our Trichoplax adhaerens STIL NTD: human CEP85 cc4 complex structure was insufficient to resolve side-chains. We defined the interaction interface as constituted by residues whose Cα distances between STIL NTD and CEP85 cc4 do not exceed 8 Å. The residues defining the interaction interface according to this criterion are shown in ellipses and are coloured by CONSURF conservation scores. Residue pairs located within an 8 Å distance are indicated with black lines. STIL residue numbers are according to Trichoplax adhaerens STIL NTD. The human equivalents of these residues are shown smaller below the Trichoplax adhaerens residues.

75

2.5 Discussion

Faithful centriole duplication is crucial for genome stability and ciliogenesis. Alterations to this process result in many human disorders such as microcephaly and cancer. Despite recent advances into elucidating the molecular mechanisms of centriole assembly, many aspects of this process are still not fully understood. Towards this, the primary focus of my PhD research is to systematically identify the molecular basis underling the centriole number control. CEP85 has been shown to interact with Nek2A through its middle coiled-coil domain (cc1-2, Figure 2.6A) and, through this interaction, to influence centrosome disjunction by antagonizing Nek2A activity (Chen et al., 2015). Here, I extend the spectrum of its activities by showing that CEP85 plays an important role in centriole duplication. Molecularly, CEP85 has been found to directly interact with STIL to facilitate its centriole localization. This regulation allows the precise spatiotemporal control of PLK4 kinase activity and thus promotes robust new centriole assembly.

The most salient findings in this data chapter are as follows: (1) Through proximity-interaction mapping and quantitative high-resolution microscopy, I identify CEP85 as a new centriole duplication factor, which represents a long sought-after missing link in the early centriole assembly pathway in metazoans. (2) Using an array of biophysical measurements, biochemistry and precise protein interaction mapping tools, I demonstrate that CEP85 directly engages STIL across species through a previously uncharacterized N-terminal domain of STIL and a discrete coiled-coiled region of CEP85. (3) We solve the atomic structure of the CEP85-STIL interaction regions using X-ray crystallography and reveal a highly conserved interaction interface. Structure-guided mutagenesis coupled to extensive biophysical and structural characterizations, pull-down experiments and in vivo binding assays, allow us to design point mutations on both sides that impede the CEP85-STIL interaction. (4) Using these point mutations in functional in vivo assays, I demonstrate that an intact CEP85-STIL binding interface is necessary for efficient recruitment of STIL to centrioles and robust local PLK4 activation, which serves as a requisite for the initiation of new centriole assembly. However, many questions remain as to how CEP85 limits centriole duplication to only once per cell cycle and how exactly CEP85 works with STIL and PLK4 to regulate this process in space and time.

76

2.5.1 The Evolutionary Analysis of the CEP85-STIL Complex

The conservation of the CEP85-STIL interaction interface implies that this mechanism might be common among metazoans, with the possible exception of insects. A sequence analysis, based on the currently available protein sequences, showed that the STIL N-terminal domain and CEP85 are found in metazoan, placozoans and choanoflagelates but are absent in insects such as Drosophila (Figure 2.12A). Intriguingly, insects that do not possess CEP85 homologues have retained STIL homologues that lack its N-terminal domain (Figure 2.12A). Thus, in insects, the additional regulatory function conferred by the CEP85-STIL interaction is apparently not necessary for centriole duplication. This notion is also supported by the observation that centriole duplication in Drosophila does not rely on several critical interactions found to be essential in other metazoans.

2.5.2 Structural and Functional Characterization of the CEP85-STIL Complex

Multiple lines of evidence support the notion that the CEP85-STIL interaction we discovered is physiologically relevant. (1) The binding between both proteins is observed across different species; (2) Structurally, the interaction interface contains conserved residues on both sides; (3) Conserved interface residues on both sides, when mutated, perturb the CEP85-STIL interaction and result in centriole duplication defects; (4) CEP85 and STIL show a partial co-localization at the site of new centriole assembly and a functional CEP85-STIL interface is required for robust targeting of STIL to centrosomes; (5) In metazoan species in which CEP85 is absent, the corresponding STIL homologues also lack its N-terminal domain, hinting towards an evolutionarily important functional connection between both; (6) The in vitro binding affinities of STIL/CEP85 are within the range of physiologically relevant interactions.

Our structure of the CEP85-STIL complex showed a 2:2 complex between CEP85 cc4 and the STIL N-terminal domain. Our biophysical data demonstrates that the CEP85 cc4 needs to be dimeric to bind to the STIL N-terminal domain, but does not allow us to unambiguously determine whether only a single, or (as in the crystal of the complex) two STIL N-terminal domains can engage CEP85 at the same time. Further structural work, possibly with the full-

77 length proteins or larger complexes, should be able to address the question of the exact binding stoichiometry.

Our structure of the STIL N-terminal domain showed the presence of two conserved patches on its surface that are prime candidates for protein-protein interaction interfaces. In our complex structure, only one of these conserved regions is involved in engaging CEP85 cc4. Docking of our high-resolution structure of CEP85 cc4 into the complex revealed that the second conserved patch of the STIL N-terminal domain would be well placed to contact a region of high conservation in CEP85 cc4. However, NMR experiments show that this putative second contact region does not contribute significantly to CEP85 binding. Thus, it is possible that the STIL N- terminal domain binds other factors, as yet undiscovered, that might compete or synergize with CEP85 binding or be involved in CEP85 independent functions. Future experiments are required to explore these possibilities.

2.5.3 Placing CEP85 and STIL in the Centriole Duplication Pathway

Our localization data places CEP85 to the proximal end of mother centrioles (Figure 2.5A). I note that the distribution of CEP85 does not display the prototypical ring-like pattern displayed by PLK4, which is much akin to what has been described as a “honeynut” distribution by Chen and colleagues (Chen et al., 2015). Nevertheless, our quantitative immunofluorescence measurements of STIL localization during centriole duplication revealed that a fraction of STIL co-localizes with CEP85 early on in the centriole duplication process (Figure 2.5B-C). The available data is thus consistent with my preferred model whereby the association between CEP85 and STIL occurs transiently at the onset of centriole duplication downstream of CEP192 and upstream of SASS6, where it contributes to the regulations of STIL localization and PLK4 activity (see the model in Figure 2.11H).

The canonical view of STIL‟s role in centriole duplication places it at the early stage of centriole assembly, when PLK4 has been recruited and subsequently restricted to a single dot on one side of the mother centriole. STIL binding then activates PLK4, which phosphorylates STIL to promote its binding to SASS6 (Arquint and Nigg, 2016). In this context, my data further suggests that CEP85 facilitates and/or stabilizes STIL localization at centrioles by transiently binding it at

78 the site of new centriole assembly, which is functionally important for full PLK4 activation and centriole duplication. We currently do not understand whether STIL binding to CEP85 is mainly crucial for an efficient local enrichment of STIL at the place of centriole formation where it is stabilized or whether it also has a more direct, functional importance for PLK4 activity regulation and/or STIL phosphorylation to facilitate the subsequent steps of the centriole duplication pathway. Moreover, STIL regulation during centriole duplication might be subject to additional layers of complexities, including its association with APC/C E3 ubiquitin ligase and USP9X deubiquitin ligase to regulate its protein stability in a cell cycle-dependent manner (Arquint and Nigg, 2014, Kodani et al., 2019), and its direct interaction with both SASS6 and CPAP to form the cartwheel structures for centriole assembly (Cottee et al., 2013). On these bases, future investigation should test whether CEP85 could associate with those factors to determine the molecular and functional properties of STIL in centriole biogenesis. Taken together, our data firmly establish CEP85 as an important upstream regulator of centriole duplication, and its interaction with STIL contributes a facet to the tight spatiotemporal control of the upstream steps of centriole biogenesis.

79

Table 1. Data collection and refinement statistics.

SeMet Complex: SeMet T. adhaerens STIL1-348 - H.sapiens CEP85570-656 T. adhaerens STIL1-348 H.sapiens CEP85570-656 Data collection Space Group I222 P21 I4132 Cell dimensions a,b,c (Å) 37.9, 73.9, 128.1 66.0, 75.2, 68.5 268.7, 268.7, 268.7 , ,  (º) 90.0, 90.0, 90.0 90.0, 97.2, 90.0 90.0, 90.0, 90.0 36.98 – 1.67 37.62 - 2.09 95.00 – 4.60 Resolution (Å) (1.76 – 1.67)a (2.14 - 2.09) (5.14 – 4.60) Rmerge 0.202 (1.552) 0.160 (0.861) 0.272 (3.033) Rpim 0.063 (0.506) 0.077 (0.416) 0.043 (0.470) I/σI 8.6 (1.4) 7.8 (2.0) 13.2 (1.9) Completeness (%) 98.9 (92.2) 100.0 (100.0) 100.0 (100.0) Redundancy 11.1 (10.1) 5.2 (5.2) 39.9 (42.1) Refinement Resolution (Å) 36.98 – 1.67 37.62 - 2.09 95.00 – 4.60 No. reflections 21329 39724 9507 Rwork / Rfree 20.3 / 23.9 19.7 / 24.1 35.3 / 36.5 No. atoms Protein 1471 5247 1829 (Poly-Ala Model) Ligand/ion 22 (MPD) 12 (MES) 0 Water 227 294 0 B-factors Protein 22.7 28.5 271.8 Ligand/ion 28.5 (MPD) 36.4 (MES) N/A Water 33.9 28.5 N/A R.m.s. deviations Bond length (Å) 0.002 0.003 0.003 Bond angles (º) 0.538 0.573 0.896 Validation Fo,Fc correlation 0.96 0.94 0.86 Molprobity Score 1.1 (99th percentile) 1.2 (100th percentile) 1.3 (100th percentile) Molprobity Clashscore 3.0 2.4 2.7 Poor rotamers (%) 0.0 0.2 n/a (Poly-Ala Model) Ramachandran outliers (%) 0.0 0.0 0.3 Ramachandran favored (%) 100.0 96.6 95.9 PDB accession code 5OI7 5OI9 5OID

Footnote: Datasets were obtained from single crystals. a Values in parenthesis are for highest resolution shell.

80

Table 2. Data collection statistics of the two-wavelength MAD datasets of selenomethionine T. adhaerens STIL1-348.

SeMet T. adhaerens STIL1-348 Data collection Space Group P21 Peak Remote Cell dimensions 65.9, 75.1, 68.6 66.0, 75.2, 68.5 a,b,c (Å) 90.0, 97.0, 90.0 90.0, 97.2, 90.0 , ,  (º) Wavelength (Å) 0.97916 0.87290 75.18 - 2.09 37.59 - 2.09 Resolution (Å) (2.14 - 2.09) (2.14 - 2.09) Rmerge 0.167 (0.845) 0.205 (1.004) Rpim 0.103 (0.520) 0.088 (0.429) I/σI 7.8 (2.6) 8.0 (2.2) Completeness (%) 91.2 (90.9) 99.9 (99.3) Redundancy 3.4 (3.4) 6.3 (6.2)

Footnote: Datasets for each wavelength derive from single crystals

Table 3. Data collection statistics of the three-wavelength MAD datasets of the selenomethionine T. adhaerens STIL1-348 - H.sapiens CEP85570-656 complex.

Complex (SeMet) T. adhaerens STIL1-348 - H.sapiens CEP85570-656 Data collection Space Group I4132 Peak Inflection Remote Cell dimensions 269.2, 269.2, 269.2 269.9, 269.9, 269.9 268.3, 268.3, 268.3 a,b,c (Å) 90.0, 90.0, 90.0 90.0, 90.0, 90.0 90.0, 90.0, 90.0 , ,  (º) Wavelength (Å) 0.97924 0.97931 0.93927 49.13 - 5.00 49.13 - 5.00 49.13 - 5.00 Resolution (Å) (5.59 - 5.00) (5.59 - 5.00) (5.59 - 5.00) Rmerge 0.696 (18.568) 0.537 (10.903) 0.245 (4.926) Rpim 0.109 (2.938) 0.082 (1.672) 0.038 (0.751) I/σI 7.6 (0.6) 7.5 (0.6) 9.4 (1.2) Completeness (%) 99.9 (100.0) 99.9 (100.0) 99.9 (100.0) Redundancy 41.9 (40.4) 42.7 (42.7) 43.0 (43.3)

Footnote: Datasets for each wavelength derive from single crystals

81

Table 4. Data collection of the selenium peak wavelength datasets of the selenomethionine T. adhaerens STIL1-348 - H.sapiens CEP85570-656 T630M / H652M complexes.

Complex (SeMet) T. adhaerens STIL1-348 –

H.sapiens CEP85570-656 T630M / H652M Data collection Space Group I4132 Cep85570-656 T630M Cep85570-656 H652M

(Peak) (Peak) Cell dimensions 268.6, 268.6, 268.6 269.4, 269.4, 269.4 a,b,c (Å) 90.0, 90.0, 90.0 90.0, 90.0, 90.0 , ,  (º) Wavelength (Å) 0.97911 (Peak) 0. 97911 (Peak) 47.47 - 5.00 47.62 - 5.00 Resolution (Å) (5.59 - 5.00) (5.59 - 5.00) Rmerge 0.457 (5.283) 0.447 (6.548) Rpim 0.070 (0.830) 0.069 (1.017) I/σI 10.6 (1.6) 10.9 (1.2) Completeness (%) 99.9 (100.0) 99.9 (100.0) Redundancy 42.7 (42.0) 42.5 (41.8)

Footnote: Datasets derive from single crystals

82

2.6 Materials and Methods

Cell lines

U-2 OS cells were cultured in standard media and conditions and were purchased from ATCC. U-2 OS T-REx cells were grown in McCoy 5A medium supplemented with 10% FBS, 2mM GlutaMAX, zeocin (100 μg/ml) and blasticidin (3 μg/ml). U-2 OS T-REx cells with Tet- inducible Myc-tagged PLK4 were a kind gift from E. Nigg, and were maintained in 10% FBS, 2mM GlutaMAX and G418 (0.5 mg/mL). hTERT RPE-1 cells expressing FLAG-tagged Cas9 were grown in DMEM/F12 supplemented with 10% FBS, GlutaMAX and sodium bicarbonate, and were a kind gift from Daniel Durocher. All human cell lines were cultured in a humidified 5% CO2 atmosphere at 37°C. All cell lines were tested without mycoplasma contamination.

Cloning and plasmids

FLAG-tagged full-length STIL and CEP85 were cloned into pcDNA5/FRT/TO vector backbone (Life Technologies) using a tetracycline-inducible CMV promoter. The CEP85 (Q640A, E644A or Q640A and E644A) and STIL (L64A, R67A or L64A and R67A) mutants were generated using the QuickChange Site-Directed Mutagenesis protocol.

RNA interference

For siRNA-mediated depletion of STIL and CEP85, the following oligonucleotide sequences were used: human STIL siRNA23, 5'-GCUCCAAACAGUUUCUGCUGGAAU-3′and human CEP85 siRNA, 5'-CCUAGAGCAGGAAGUGGCUCAAGAA-3′. siRNAs against human PLK4 (M-005036-02-0005), CEP192 (L-032250-01-0005) and CEP152 (M-022241-01) were purchased from Dharmacon. The Luciferase GL2 Duplex non-targeting siRNA from Dharmacon was used as a negative control. All siRNAs were transfected using Lipofectamine RNAiMax (Invitrogen) according to the manufacturer‟s protocol.

83

Generation of stable cell lines

The human CEP85 siRNA (CCUAGAGCAGGAAGUGGCUCAAGAA) target sequencewas mutated (TCTCGAACAAGAGGTCGCCCAGGAG) and STIL siRNA (GCUCCAAACAGUUUCUGCUGGAAU) target sequence was mutated (ACTGCAGACCGTGTCCGCCGGCAA) using QuikChange Site-Directed Mutagenesis protocol and cloned into pCMV-TO/FRT-FLAG vector. U-2 OS T-REx cells were co-transfected with pOG44 (Flp-recombinase expression vector) and pCMV-TO/FRT-FLAG plasmid containing the coding sequence for siRNA resistant human CEP85 and STIL. Transfections were carried out with Lipofectamine 2000 (Invitrogen) according to the manufacturer‟s protocol. At 24 h post-transfection, cells were selected with 200 μg/ml hygromycin B.

CRISPR

Lentiviral CRISPR/Cas9-mediated gene targeting was used to deplete endogenous CEP85 and CEP192. To achieve complete gene knockout, two gRNAs targeting different exons of CEP85 were used, and their targeting sequences are as followed: CEP85 gRNA1 (CCATCTTAGAGCCAGCACAG) and CEP85 gRNA2 (AGCTGGGCCGTGTCTGGTGA). CEP192 gRNA sequence is GTGCTTAATCCAACTGACCGC. gRNAs were cloned into LentiCRSIPRv2 vector (gift of Feng Zhang, Addgene plasmid #52961). For lentivirus packaging, 293T cells (1.5x106) were transfected with 3.3 μg LCv2-CEP85 or CEP192 gRNA vector, 2.5μg psPAX2 and 1.7 μg pVSVG using Lipofectamine 3000 (Invitrogen) according to the manufacturer‟s protocol. 24 h post-transfection, the media was changed with high-serum media containing 30% FBS and virus was collected after 48 h. For lentiviral transduction, 100,000 cells were seeded on 6-well plates and infected with 1ml viral supernatant containing 8 μg/ml polybrene. 24 h post-infection, cells were split and selected in the presence of 10 μg/ml puromycin. Five days post-infection, cells were fixed with ice-cold methanol for immunofluorescence staining, and also collected for western blot experiments.

84

Immunofluorescence microscopy

Cells were fixed with ice-cold methanol at −20°C for 20 min and blocked in 1% BSA and 0.05% Tween-20 solution in PBS for one hour. Cells were incubated with the primary antibodies in blocking solution for 2 hours, washed 3x 5min in blocking solution and then incubated with fluorophore-conjugated secondary antibodies (Molecular Probes) and DAPI (0.1 μg/ml) in blocking solution for one hour. After a final wash 3x 5min in blocking solution, cells were inverted and mounted on glass slides with standard mounting solution (ProLong Gold antifade, Molecular Probes). Mouse polyclonal antibody used in this study is anti-CEP85 (H00064793- B01P, dilution 1:200; Abnova). Mouse monoclonal antibodies used in this study are anti-Centrin (clone 20H5, 04-1624, dilution 1:200; Millipore), anti-FLAG (F3165, dilution 1:1000; Sigma), mouse anti-PCNA (PC10, sc-56, dilution 1:200; Santa Cruz) and mouse anti-Pericentrin (ab28144, dilution 1:500; Abcam). Rabbit polyclonal antibodies used in this study are anti-PLK4 pSer305 (dilution 1:500; a kind gift from Tak W. Mak) and anti-CEP192 (A302-324A-1, dilution 1:1000; Bethyl Laboratories) and anti-STIL (dilution 1:1000). This STIL antibody was raised against human, recombinant STIL1-373 (containing the N-terminal and C-terminal linker GGS and ENLYFQ respectively) using Davids Biotechnologie GmbH, Regensburg, Germany and was affinity purified from the antiserum using covalently immobilised GST-human STIL1-390 (construct as described above) following standard protocols. Goat polyclonal antibodies used in this study are anti-γ-tubulin (sc-7396, dilution 1:200; Santa Cruz), anti-SASS6 (sc-82360, dilution 1:100; Santa Cruz) and anti-C-Myc (ab19234, 1:500; Abcam). Donkey anti-Rabbit Alexa 488/594/647 (Molecular Probes) were used in this study. Cells were imaged on a Deltavision Elite DV imaging system equipped with a sCMOS 2048x2048 pixels2 camera (GE Healthcare). Z stacks (0.2 μm apart) were used, and images were further deconvolved and maximum intensity projected using softWoRx (v6.0, Applied Precision).

Super resolution microscopy

Super resolution microscopy was essentially performed following standard procedures29, 30. Briefly, cells were imaged on a three-dimensional (3D) structured illumination microscope (OMX Blaze v4, GE Biosciences PA) equipped with 405, 445, 488, 514, 568 and 642 nm diode lasers, 4 high-speed sCMOS cameras (scientific CMOS, 2560*2560 pixels, manufactured by

85

PCO), and a ×60/1.42 NA planApochromat oil-immersion objective (Olympus). Multi-channel 3D-SIM image Z stacks were reconstructed, 3D aligned using calibrations based on a GE reference slide and 100 nm-diameter TetraSpeck Microspheres and maximum intensity projected using the softWoRx 6.0 software package (GE). The 3D-SIM modality of our Optical Microscopy eXperiment (OMX) provides an axial resolution of 340-380 nm depending on the imaging wavelength, thus sampling axially every 125nm satisfied the Nyquist criterion for oversampling (GE Biosciences PA).

Image analysis

Image quantifications were performed in MATLAB. For widefield quantifications, Z stacks were deconvolved and maximum intensity projected using the softWoRx (v6.0, Applied Precision). For 3D-SIM quantifications, we used reconstructed, maximum intensity projected, and aligned images. For each image set, the centrosome was detected as the composite region with the greatest integrated intensity across all channels. Within the composite region, we generated masks (using constant signal-to-noise and size thresholds) for each image channel specific for the labelled protein. Recruitment to the centrosome for a given label was quantified as the total pixel intensity of its masked region.

Western blots

Cells were collected, lysed in Laemmli sample buffer and digested with benzonase nuclease. Proteins were loaded to 8% SDS-PAGE gel for electrophoresis and transferred to a PVDF membrane (Immobilon-P, Millipore). Membranes were incubated with primary antibodies in TBST (TBS, 0.1% Tween-20) in 5% skim milk powder (BioShop). Blots were washed 3x10 mins in TBST, then incubated with secondary antibodies conjugated to HRP. Western blots were developed using SuperSignal reagents (Thermo Scientific). Mouse polyclonal antibody used in this study is anti-CEP85 (H00064793-B01P, dilution 1:500; Abnova). Mouse monoclonal antibody used in this study are anti-α-tubulin (Clone DM1A, T6199, dilution 1:15000; Sigma- Aldrich), anti-FLAG (F3165, dilution 1:1000; Sigma) and anti-Cyclin A (BF683, dilution 1:1000;

86

Cell Signaling). Rabbit polyclonal antibodies used in this study are anti-CEP192 (A302-324A-1, dilution 1:1000; Bethyl Laboratories) and anti-STIL (A302-441A, dilution 1:500; Bethyl Laboratories). Goat polyclonal antibody used in this study is anti-C-Myc (ab19234, dilution 1:500; Abcam).

BioID, mass spectrometry and analysis

BioID and mass spectrometry was performed as described (Gupta et al., 2015). The network in Supplementary Figure 1a was constructed from SAINT data were imported into Cytoscape 3.2.1 31, and with the Edge-Weighted Spring Embedded Layout. SAINT analysis of high confidence interactors to generate Supplementary Table 1 used the following parameters: TPP >0.9, unique peptides≥2, FDR ~1% 32. For each prey, peptide counts in each bait sample (two technical replicates of two biological replicates) were compared to those from 14 control runs.

Protein crystallization

SeMet T. adhaerens STIL1-348 crystals were obtained using the vapour diffusion method with sitting drops (1 μl of protein solution + 1 μl of reservoir solution) and a reservoir solution of 100 mM MES, pH 6.0, 3 % (v/v) MPD at 18 °C. Crystals were mounted after nine days in 100 mM MES, pH 6.0, 35 % (v/v) MPD and frozen in liquid nitrogen.

H. sapiens CEP85570-656 was crystallized at 18°C in sitting drops (1 μl of protein solution + 1 μl of reservoir solution) using the vapour diffusion method and a reservoir solution of 100 mM Na-HEPES, pH 7.7, 200 mM Na-Citrate, 36 % (v/v) MPD. Crystals were mounted after six days and frozen in liquid nitrogen.

The SeMet T. adhaerens STIL1-348 - H.sapiens CEP85570-656 complex was crystallized at 18 °C in sitting drops (1 μl of protein solution + 1 μl of reservoir solution) using the vapour diffusion method and a reservoir solution of 100 mM MES, pH 6.0, 1.45 M MgSO4, 2 mM DTT. Crystals were mounted after two days in 100 mM MES, pH 6.0, 1.45 M MgSO4, 25% (v/v) glycerol and frozen in liquid nitrogen.

87

The protein concentrations of the crystallized constructs were determined by the Bradford assay with BSA as a standard and were: 11.0 mg/ml (SeMet T. adhaerens STIL1-348), 49.9 mg/ml (H.sapiens CEP85570-656) and 32.8 mg/ml (SeMet T. adhaerens STIL1-348 + H. sapiens CEP85570-656).

Recombinant protein purification

DNA encoding T. adhaerens STIL1-348 or G. gallus STIL14-369 (WT and mutants) were cloned into a modified pRSETa vector (Invitrogen) containing two His-tagged lipoyl domains (Cottee et al., 2013). SeMet T. adhaerens STIL1-348 and G. gallus STIL14-369 were subsequently expressed in E.coli C41(DE3) either in supplemented M9 medium (van Breugel et al., 2014) (SeMet T. adhaerens STIL1-348) or in 2xTY medium (G. gallus STIL14-369) and purified by standard methods using NiNTA (Qiagen) beads. After elution, the eluates were dialysed against 50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 5 mM imidazole, pH 7.5, 10 mM β- mercapto-ethanol in the presence of TEV protease (kind gift of Mark Allen, MRC-LMB, Cambridge, UK) to cut off the His-tagged lipoyl tags and rebound to NiNTA Agarose (Qiagen) to remove the tags. The flow-through was further purified by size exclusion chromatography in 10 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM DTT (SeMet T. adhaerens STIL1-348) or by ion- exchange chromatography (HiTrap Q-FF, GE Healthcare) using linear salt gradients from 10 mM Tris-HCl, pH 8.0, 2 mM DTT to 10 mM Tris-HCl, pH 8.0, 2 mM DTT, 1 M NaCl (G. gallus STIL14-369).

DNA encoding H.sapiens CEP85570-656 or CEP85570-662 (WT and mutants) were cloned into a modified pOPTH vector (Ohashi et al., 2016) giving rise to a N-terminally His and MBP- tagged construct. Constructs were expressed in E. coli C41 (DE3) and purified by NiNTA (Qiagen) chromatography using standard methods. The eluates were dialysed against 50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 5 mM imidazole, pH 7.5 in the presence of TEV protease (kind gift of Mark Allen, MRC-LMB, Cambridge, UK) to cut off the His-MBP tag, rebound to NiNTA Agarose (Qiagen) to remove the tag and the flow-throughs further purified by size exclusion chromatography in 10 mM Tris-HCl, pH 7.4 (CEP85570-656) or pH 8.0 (CEP85570-662), 50 mM NaCl followed by ion-exchange chromatography (HiTrap Q-FF, GE Healthcare) using

88 linear salt gradients from 10 mM Tris-HCl, pH 8.0 to 10 mM Tris-HCl, pH 8.0, 1 M NaCl (CEP85570-662).

The SeMet T. adhaerens STIL1-348 - H.sapiens CEP85570-656 complex was purified by mixing the SeMet T. adhaerens STIL1-348 and the H.sapiens CEP85570-656 expressing cell cultures, followed by cell lysis, NiNTA purification, TEV cleavage and removal of the cut-off tags by rebinding to NiNTA resin as described above. The flow-through was dialysed against 10 mM Tris-HCl, pH 8.0, 2 mM DTT, NaCl added to 75 mM and the complex purified further by ion-exchange chromatography from 10 mM Tris-HCl, pH 8.0, 2 or 4 mM DTT to 10 mM Tris- HCl, pH 8.0, 1 M NaCl, 2 or 4 mM DTT (linear gradient) on a Mono-Q FPLC column (Pharmacia).

DNA encoding human STIL1-390 (WT and mutants) (including an N-terminal linker (SSRSNQTSLYKKAGSAAAPFT), were cloned into the BamH1 and EcoR1 sites of pGEX-6P1 (GE Healthcare). Constructs (and pGEX-6P1 alone) were expressed in E.coli Rosetta at 18°C and purified by Glutathione Sepharose 4B resin (GE Healthcare) and ion exchange chromatography (HiTrap Q HP column, GE Healthcare) using a linear salt gradient from 10 mM Tris-HCl, pH 8.5, 2 mM DTT to 10 mM Tris-HCl, pH 8.5, 1 M NaCl, 2 mM DTT. All proteins were concentrated, flash-frozen in liquid nitrogen and stored at -80°C.

GST-STIL NTD-3xFLAG-CEP85 pull-down assay

This assay was adapted from 18 ~30 μg of purified GST or GST-H.sapiens STIL1-390 were bound to Glutathione Sepharose 4B (GE Healthcare) beads in Lysis buffer (50 mM Bis-Tris- Propane, pH 7.2, 100 mM NaCl, 1 mM DTT, 0.1% (v/v) Nonidet-P40, supplemented with Complete Protease Inhibitor (EDTA free, Roche), washed and subsequently incubated (4 °C, 1 h) with centrifugationally cleared cell lysates (obtained by sonication) from Hek293 Trex Flpin cells (kind gift of Manu Hegde, MRC-LMB, Cambridge, UK) transfected for ~2 days by Fugene 6 or Fugene HD (Promega) with 3xFLAG human CEP85 constructs (cloned into a pcDNA3.1 derivative, kind gift of Manu Hegde, MRC-LMB, Cambridge, UK). After washing with Lysis buffer, beads were eluted (Lysis buffer, 100 mM Glutathione, pH 7.5) and eluates separated by SDS-PAGE before Western Blotting with anti-FLAG M2 mouse monoclonal antibody (Sigma).

89

Yeast Two Hybrid

Constructs were cloned into vector pENTR/D-TOPO and transferred into vectors pDEST32 (Bait, human STIL NTD) or pDEST22 (Prey, human CEP85 constructs) using site-specific recombination (Gateway LR Clonase II, Thermo Fisher Scientific). Bait and Prey plasmid combinations were co-transformed into yeast strain MaV203 (Thermo Fisher Scientific) and plated onto SC-Leu-Trp plates. Colonies were inoculated into SC-Leu-Trp medium, grown and spotted onto SC-Leu -Trp and SC-Ura plates. Plates were incubated for three days at 30°C.

Statistical methods

Two-tailed unpaired student t-tests were performed for all p-values. Individual p-values, experiment sample numbers and the number of replicates for statistical testing were indicated in corresponding figure legends. Unless otherwise mentioned, all error bars are S.D, and the asterisk placeholders for p-values are **p<0.01 and *p<0.05.

90

3 Chapter III: Direct interaction between CEP85 and STIL mediates PLK4-driven directed migration

91

3.1 Statement of Contribution

The data in this chapter has not been published. The manuscript is currently under peer-review in the Journal of Cell Science, on which I am the first author. I contributed to the conceptualization and 90% of the experimental investigation for the first submission and wrote the original draft. The lists below are the contributions from other authors.

Dr. Jaeyoun Kim performed the transwell cell migration assay, and quantified centrosome numbers and Golgi repositioning during directional cell migration.

Vaishali Sridhar conducted the flow cytometry analysis.

Dr. Megha Chandrashekhar drew the model for transwell migration assay.

Dr. Mark van Breugel carried out the yeast-two-hybrid experiments to confirm the direct interaction between CEP85 and CEP192.

92

3.2 Summary

PLK4 has emerged as a prime target for cancer therapeutics and its overexpression is frequently observed in various types of human cancer. Recent studies have further revealed an unexpected oncogenic activity of PLK4 in regulating cancer cell migration and invasion. However, the molecular basis behind PLK4‟s role in these processes still remains only partly understood. Our previous work demonstrated that an intact CEP85-STIL binding interface is necessary for robust PLK4 activation and centriole duplication. Here, I find that the CEP85-STIL complex plays an important role in the control of directed cell motility. Functional analyses indicate that the interactions of CEP85 with STIL and PLK4 are required for directed cell migration. Molecularly, I demonstrate that PLK4 mediates the localization of CEP85 and STIL at the cell cortex, and that depletion of CEP85 and STIL results in a marked reduction in ARP2 phosphorylation and defective actin reorganization, consequentially perturbing cell migration. Altogether, this work provides molecular insight into the critical role of the CEP85-STIL complex in modulating PLK4-dependent cancer cell migration.

93

3.3 Introduction

The directed migration of cells is a fundamental cellular process that is required for normal physiology and deregulated in disease, including cancer. This cellular property is a polarized cellular process that is associated with the formation of distinct protrusive frontal and retractive rear regions (Mayor and Etienne-Manneville, 2016). From the front to the rear, a dramatic remodeling of the cell cytoskeleton plays a pivotal role in the control of directed cell motility (Mayor and Etienne-Manneville, 2016). In detail, actin-dependent forces shape the cellular membranes and promote cell protrusion, adhesion, contraction, and retraction. Moreover, the front-rear polarization in migrating cells also relies on the regulation of microtubule dynamics, which is required for the development and/or maintenance of cell polarity (Mayor and Etienne- Manneville, 2016). In cancer, aberrant remodeling of the cell cytoskeleton leads to an unscheduled increase in the motile property of cancer cells, which promotes intravasation and contributes to the onset of a metastatic state (Stuelten et al., 2018). However, the molecular mechanism underlying how cells precisely coordinate cytoskeleton remodeling to direct their movements is not fully understood.

As the primary microtubule-organizing center of cells, the centrosome has long been recognized as a driving force in the control of cell motility. In migrating cells, centrosomes reposition alongside the Golgi apparatus to locate between the nucleus and the leading edge to establish the nuclear–centrosome–Golgi axis relative to the front-rear axis, which directs the microtubule nucleation and intracellular trafficking frontward (Luxton and Gundersen, 2011). It is now evident that this axis orientation is observed in many cell types and serves as a steering device for directed cell migration, most probably through the microtubule system. However, centrosomes have also been proposed to act as actin-organizing centers via Pericentriolar material 1 (PCM1) dependent recruitment of WASH and ARP2/3 complex, providing an additional mechanism for centrosome-based control of cell movement (Farina et al., 2016).

PLK4 is known as the most upstream regulator of centriole duplication and its kinase activity is required to initiate this process. It is well established that inhibiting PLK4 activity impairs centriole formation, whereas high levels of PLK4 activity result in centriole over-duplication (Arquint and Nigg, 2016). PLK4 can regulate its activity through autophosphorylation to create a phosphodegron that is bound by the SCFSlimb/b-TrCP ubiquitin ligase for degradation (Gönczy,

94

2012). Additionally, direct binding of STIL to PLK4 has been indicated to further stimulate its kinase activity, possibly through promoting PLK4 autophosphorylation (Moyer et al., 2015, Ohta et al., 2014). More recently, we have reported that interaction between CEP85 with STIL, which is essential for robust PLK4 activation and efficient centriole assembly (Liu et al., 2018).

The genetic amplification of PLK4, and its overexpression at the mRNA and at the protein levels are frequently observed in malignant cancers alongside a decreased survival of cancer patients (Maniswami et al., 2018). PLK4-drived centrosome over-duplication is sufficient to promote aneuploidy and initiate tumorigenesis and in mice, when the P53 pathway is partially inhibited (Levine et al., 2017). Moreover, excess PLK4 expression has also been linked to a distinct cellular-invasion mechanism through increased centrosome-based microtubule nucleation and RAC1 activity at the cortex (Godinho et al., 2014). Moreover, a recent study reveals unexpected oncogenic activity for PLK4 in controlling directional cancer cell migration (Kazazian et al., 2017). Molecularly, active PLK4 is detectable at the cell cortex, and both depletion of PLK4 and inhibition of PLK4 kinase activity significantly suppress directed cancer cell movement and invasion (Kazazian et al., 2017). PLK4 Polo-box1-Polo-box2 domain has been shown to physically interact with ARP2, and this interaction is required for the phosphorylation of ARP2 to regulate actin rearrangement in a RAC1 and CDC42 independent manner (Kazazian et al., 2017). Overall, these findings reveal the molecular basis of PLK4 in driving cancer cell motility. Despite remarkable progress on these mechanistic studies, it still remains enigmatic about how PLK4 mediates contextual control of actin remodeling and cell motility in space and time. In addition, CEP192, an upstream regulator of PLK4, also has been shown to participate in the control of directional cell migration through the control of microtubule dynamics.

In this work, I demonstrate that CEP85 and STIL participate in directed cell movement. Using structure-guided information of this complex, I find that disrupting the CEP85-STIL binding interface negatively affects cell migration. I further reveal that the central region of CEP85 is responsible for its interaction with PLK4, in a STIL-independent manner. Functionally, this interaction is essential for the control of both cell migration and centriole duplication. Interestingly, although CEP85 is also able to directly interact with CEP192, this interaction has no effect on both centriole assembly and cell motility control. Mechanistically, CEP85 and STIL can be recruited to the cell cortex through PLK4. Downregulation of either CEP85 or STIL significantly reduces phosphorylation of ARP2, thus impairing actin re-organization and cell

95 motility. Together, my findings reveal a novel role of the CEP85-STIL module in directed cell migration.

96

3.4 Results

3.4.1 CEP85 Regulates Directional Cell Migration

To determine whether there are additional components involved in PLK4-driven cell migration, I sought to assess the role of PLK4-associated centriole duplication factors in this process. To this end, I utilized siRNAs to deplete endogenous CEP192, CEP152, CEP85 and STIL in U-2 OS cells and examined the effect of their depletion on directional cell migration using conventional wound healing assays. Western blot analyses validated the efficient depletion of each of these proteins in U-2 OS cells (Figure 3.2A). My results indicate that the down-regulation of CEP85 in U-2 OS cells significantly delayed wound healing to levels akin to those observed upon PLK4 or CEP192 depletion (Figure 3.1A-B). In support of these observations, I found that the depletion of PLK4, CEP192 or CEP85 displayed a marked reduction in transwell cell migration (Figure 3.2B-D). STIL depletion also suppressed efficient wound closure and resulted in decreased transwell cell migration, albeit in a less pronounced manner (Figure 3.1A-B and Figure 3.2B-D). Intriguingly, although CEP152 is known to act as a scaffold for PLK4 activity (Dzhindzhev et al., 2010, Park et al., 2014b, Sonnen et al., 2013) and even though I could efficiently deplete it, no noticeable defects in wound healing could be observed, suggesting that it does not participate in this pathway. I next determined that the wound healing defects observed were not indirectly caused by impaired cell cycle progression, centrosome loss or orientation of the Golgi apparatus. My results suggest that depletion of PLK4, CEP192, CEP152, CEP85 and STIL, under the low- serum conditions used in the migration assays, resulted in centrosome loss in ~10% of cells (Figure 3.3A-B), no obvious defects in Golgi reorientation toward the leading edge in response to a scratch wound (Figure 3.3A and C) and cell cycle progression (Figure 3.4A-F). These phenotypes are unlikely to be large enough to account for the cell migration defects observed upon CEP85 and depletion. Our previous work demonstrated that CEP85 directly binds to STIL to ensure robust PLK4 kinase activation (Liu et al., 2018). On this basis, I aim to examine whether the CEP85-STIL interaction is also functionally important for PLK4-driven cell migration. To test it, I first depleted endogenous CEP85 or STIL in U-2 OS cells and assessed the ability of RNAi-resistant wild-type CEP85 or STIL or the CEP85 Q640A+E644A or STIL L64A+R67A mutants (that compromise the CEP85-STIL interaction) to rescue the cell migration defect (Liu et al., 2018). Indeed, the expression of these CEP85 or STIL mutants could not fully rescue the wound healing defect in CEP85 or STIL depleted cells, when compared to wild-type

97

CEP85 or STIL (Figure 3.1C-D). Together, these findings suggest that a functional CEP85-STIL binding interface is critical for the directed cell migration.

98

Figure 3.1. CEP85 and STIL are required for directional cell migration.

(A). Representative images from a scratch-wound healing assay of U-2 OS cells transfected with different siRNAs. Scale bar, 400 μm.

(B). Quantification of wound closure efficiency shown in Figure 3.1A as measured by the percentage of wound area recovered (mean SD, n = 3/experiment, three independent experiments).

(C, D). The graphs indicate the percentage of wound area closed in U-2 OS cells expressing Tet- inducible FLAG or the siRNA-resistant FLAG-CEP85 and STIL transgenes, treated with control,

CEP85 or STIL siRNA and induced with tetracycline for 72 h (mean SD, n = 3/experiment, three independent experiments).

99

Figure 3.2. The role of CEP85 and STIL in transwell cell migration.

(A). Western blot showing the levels of CEP192, CEP152, CEP85 and STIL in different siRNA transfected cells. α-tubulin was used as a loading control.

(B). Experimental setup of Boyden chamber transwell migration assay.

(C). 5 x 104 cells of indicated cell lines were seeded onto the top of a transwell insert. After 24 hours, the cells on the top of the insert were scraped off and the cells that migrated to the bottom were fixed, stained with crystal violet.

(D). Number of cells that migrated are plotted in the graph shown here. (mean SD, n = 100, three independent experiments**p < 0.01, *p <0.05).

100

101

Figure 3.3. Effect of CEP85 or STIL depletion on centrosome number and Golgi re- positioning.

(A). U-2 OS cells transfected with specified siRNAs were scratched and then healed for 6 h. Cells were subjected to immunocytochemistry with antibodies specific to GM130 (red), and PCNT (green). Actin filament and DNA were visualized with Phalloidin (cyan) and DAPI (violet). Arrows indicate axis of nucleus-centrosome-Golgi apparatus. Scale bar, 100 μm.

(B). The number of centrosomes was determined from PCNT signals. All data are represented as means SD (n >150, three independent experiments**p < 0.01, *p <0.05).

(C). Re-positioning of Centrosome and Golgi to leading edge was investigated 6 h after would healing. Dashed lines indicate the borders between cells and scratches. Axis of nucleus- centrosome-Golgi apparatus (arrows in Figure 3.3A) were counted and showed with radar graphs.

102

Figure 3.4. Analysis of cell cycle by flow cytometry.

(A-F). U2OS cells were treated with the specified siRNAs for 72 h followed by DAPI staining and flow cytometric analysis. Cell cycle distribution across G1, S and G2/M phases was modeled and quantified using ModFit software

103

3.4.2 The Binding of CEP85 to PLK4 but Not CEP192 is required for Cell Migration

Using BioID, we and others have previously identified PLK4 as a putative proximity interaction partner of CEP85 (Firat-Karalar et al., 2014, Liu et al., 2018). On these bases, I performed co-IP analysis with cell lysates expressing Myc-tagged PLK4 and FLAG-tagged CEP85 full length and domain truncation mutants (Figure 3.5A-B). I observed that both wild-type CEP85 and the CEP85 Q640A+E644A double mutant were able to interact with PLK4, and that treatment with centrinone B, a small molecule that perturbs PLK4 kinase activity (Wong et al., 2015), had no effect on this association (Figure 3.5A-B). These results indicate that the binding of CEP85 to PLK4 is independent of the CEP85-STIL interaction and PLK4 kinase activity. To determine which region in CEP85 is responsible for this interaction, I transiently expressed a set of CEP85 truncation mutants in U-2 OS cells and performed the co-IP experiments. The results show that the middle region of CEP85 is required for its interaction with PLK4 (Figure 3.5C-D). Interestingly, the CEP85 ∆M mutant that was defective for PLK4 binding exhibited significantly decreased centriole localization, implying that the CEP85-PLK4 interaction may facilitate the recruitment of CEP85 to centrioles (Figure 3.6B).

CEP192 has been shown to regulate directional cell movement, and our proteomic studies identify CEP192 as a potential proximity interaction partner of CEP85 (Firat-Karalar et al., 2014, Liu et al., 2018). These findings raise the possibility that CEP192 and CEP85 may work together in the control of cell migration. To test this hypothesis, my first step is to validate the CEP85- CEP192 interaction. In this regard, we performed the yeast-two-hybrid (Y2H) assay and found that the N-terminal region of CEP85 directly interacted with CEP192 (Figure 3.6A). However, this interaction is not essential for the localization of CEP85 to centrioles (Figure 3.6B). Functionally, I aim to assess the ability of CEP85 ∆N and ∆M mutants to rescue centriole duplication and cell migration defects in cells depleted of CEP85. My results demonstrate that the expression of CEP85 ∆M mutant could not efficiently rescue these phenotypes when compared to wild-type CEP85 and CEP85 ∆N mutant (Figure 3.5E-F). During directed cell migration, CEP192 has previously been shown to regulate microtubule dynamics (O‟Rourke et al., 2014). In line with this notion, Western blot analysis suggested that CEP192 depletion resulted in a marked increase in the levels of less dynamic acetylated microtubules, whereas depletion of CEP85, STIL or PLK4 had no noticeable effect (Figure 3.6C). Taken together, these

104 data indicate that the PLK4 interaction might be necessary for CEP85‟s ability to drive both centriole duplication and cell migration, independently of CEP192.

105

Figure 3.5. CEP85 interacts functionally with PLK4.

(A). Domain overview of human CEP85. cc coiled-coil. Vectors that were used in this work are shown by green lines.

(B). Detection of expressed FLAG-CEP85 WT or Q640A+E644A mutant coimmunoprecipitating with Myc-tagged PLK4 in U-2 OS cells with or without centrinone B treatment. An empty vector was used as a negative control.

(C, D). Detection of expressed FLAG-CEP85 fragments coimmunoprecipitating with Myc- PLK4 in U-2 OS cells. The middle region of CEP85 is required for its interaction with PLK4.

(E). U-2 OS cells expressing Tet-inducible FLAG or the siRNA-resistant FLAG-CEP85 transgene were transfected with control or CEP85 siRNA and induced with tetracycline for 72 h. The graphs indicate the percentage of cells with four centrioles (mean SD, n = 100, three independent experiments, **p < 0.01, *p <0.05).

(F). The graphs indicate the percentage of wound area closed in U-2 OS cells expressing Tet- inducible FLAG or the siRNA-resistant FLAG-CEP85 transgenes, treated with control, CEP85 siRNA and induced with tetracycline for 72 h. (mean SD, n = 3/experiment, three independent experiments, **p < 0.01, *p <0.05).

106

Figure 3.6. The cell cycle-dependent regulation of CEP85 and the role of CEP85-CEP192 interaction.

(A). The N-terminal region of CEP85 shows a yeast-two-hybrid interaction with CEP192. Yeast strain MaV203 (Invitrogen) transformed with the indicated bait and prey constructs (CEP192 in vector pDEST22, CEP85 constructs in vector pDEST32, pEXP22/RalGDS m2 in pDEST22 (Invitrogen)) were plated on SC-Leu/-Trp plates (selecting for bait and prey plasmid, top) and on SC-Ura plates (selecting for Ura promoter activation, bottom) and incubated for 3 days at 30°C.

107

(B). Representative images of U-2 OS cells expressing Tet-inducible GFP-CEP85 WT, ∆M and ∆N mutants, and induced with tetracycline for 24 h. Cells were stained with the indicated antibodies. Scale bar, 10 μm.

(C). Western blot showing the levels of acetylated tubulin in control, PLK4, CEP85, STIL or CEP192 siRNA transfected U-2 OS cells. α-tubulin was used as a loading control.

3.4.3 CEP85 and STIL Localize at the Cell Cortex

I next sought to determine if CEP85 and STIL can localize to the cell cortex. I was unable to reliably detect endogenous STIL at the cell cortex, likely since STIL is a low-abundance protein with the strong cytoplasmic background (Bauer et al., 2016). To address this limitation, I transiently expressed mCherry-STIL in U-2 OS cells harboring Tet-inducible wild-type GFP- CEP85. In order to better visualize the cell cortex, cells overexpressing GFP-tagged CEP85 and mCherry-tagged STIL were fixed with PFA without permeabilization prior to imaging. My results revealed that over-expressed wild-type CEP85 was able to co-localize with STIL at the cell cortex (Figure 3.6A-B). In contrast, expression of the CEP85 ∆M and Q640A+E644A mutants disrupted this localization, suggesting that the binding of CEP85 to both PLK4 and STIL plays a role in their cortical localization (Figure 3.7A-B).

To test this hypothesis, I further assessed the effect of knocking down PLK4 and CEP192 on the cortical localization of CEP85 and STIL in U-2 OS cells. My results indicate that depletion of PLK4, or treatment with centrinone B, but not CEP192 depletion, perturbed the localization of CEP85 and STIL at the cell cortex (Figure 3.7C-D). Collectively, my results demonstrate that PLK4 is an upstream regulator of CEP85 and STIL during cell migration.

108

109

Figure 3.7. CEP85 and STIL localization at the cell cortex.

(A). Representative images of U-2 OS cells expressing Tet-inducible GFP-CEP85 WT, ∆M and ∆N mutants, transfected with mCherry-STIL and induced with tetracycline for 24 h. Scale bar 20 μm.

(B). The graph shows the percentage of cells with the described localization pattern. (mean SD, n = 100, three independent experiments, **p < 0.01, *p <0.05).

(C). U-2 OS cells expressing GFP-CEP85 WT and mCherry-STIL were either treated with centrinone B (1 μM) for 24 h or transfected with control, PLK4 or CEP192 siRNA for 72 h. Scale bar 20 μm.

(D). The graph shows the percentage of cells with the described localization pattern. (mean SD, n = 100, three independent experiments, **p < 0.01, *p <0.05).

3.4.4 CEP85 and STIL Regulate the Actin Cytoskeleton

It has been indicated that PLK4 phosphorylates ARP2 to drive the assembly of the actin cytoskeleton (Kazazian et al., 2017). On this basis, I aim to examine whether CEP85 and STIL, that act to facilitate PLK4 activation (Liu et al., 2018), also contribute to this regulation. To test this hypothesis, I first used the canonical cell spreading assay in response to the stimulus of replating. My results demonstrate that the depletion of CEP85 and STIL negatively affected the cell spreading alongside a significant reduction in cell size, an effect that was comparable to PLK4 depletion or inhibition of its kinase activity (Figure 3.8A-B). To further explore the molecular mechanism, I sought to determine whether CEP85 and STIL impact PLK4-dependent phosphorylation of ARP2 (Kazazian et al., 2017). Using an antibody that specifically detects phosphorylated ARP2, I found that ARP2 phosphorylation at T237/T238 was significantly decreased in U-2 OS cells depleted of PLK4, CEP85 and STIL or treated with centrinone B, when compared to the control (Figure 3.8C-D). Together, these data demonstrate that the functional PLK4-CEP85-STIL module is critical for the activity of the ARP2/3 complex in the control of directed cell migration.

110

111

Figure 3.8. CEP85 and STIL regulate ARP2/3 mediated actin reorganization.

(A). Representative images of spreading assays (See the Materials and Methods) in control, PLK4, CEP85 or STIL siRNA transfected U-2 OS cells. Cells were stained with DAPI and Alexa Fluor 488 Phalloidin. Scale bar 20 μm.

(B). The violin plot of the cell area quantification shows the data distribution and its probability density (n = 100, three independent experiments).

(C). Western blot showing the levels of phospho-GFP-ARP2 and total GFP-ARP2 in control, PLK4, CEP85 or STIL siRNA treated U-2 OS cells, using a phospho-specific anti-ARP2 (phospho T237+T238) antibody. α-tubulin served as a loading control.

(D). Quantification of fold change in phospho-GFP ARP2 relative to total GFP-ARP2 (mean SD, n = 6, **p < 0.01, *p <0.05).

(E). A model for how PLK4-CEP85-STIL operates in the control of directional cell migration. CEP85 and STIL act downstream of PLK4 and their interaction facilitates PLK4 activation and subsequent ARP2 phosphorylation, which further regulates ARP2/3 mediated actin assembly and directional cell migration.

112

3.5 Discussion

The work described here identifies an unexpected function of a conserved centrosomal PLK4- CEP85-STIL module in directed cell migration. I find that PLK4 is required for the recruitment of CEP85 and STIL to the cell cortex, which is dependent on PLK4 kinase activity and its interaction with CEP85. In turn, the CEP85 and STIL interaction was previously reported to enable robust PLK4 kinase activation (Liu et al., 2018). This regulation further promotes ARP2 phosphorylation in order to drive ARP2/3 dependent actin re-organization and directed cell movement. Overall, my data suggest that the PLK4-CEP85-STIL module plays a pivotal role in the regulation of both centriole duplication and cell migration.

Recently, centrosomes have been proposed as actin-organizing centers through the ARP2/3- based actin nucleation (Farina et al., 2016). The ARP2/3 complex consists of two actin-related proteins (ARP2 and ARP3) and five accessory proteins, which function to stimulate actin brunching by creating nucleation sites on the exiting filaments (Rottner et al., 2010). The majority of studies on the ARP2/3-based regulations have been focused on the Wiskott-Aldrich syndrome (WASP) proteins. It is now evident that WASP proteins can bind to and activate the ARP2/3 complex to generate exclusively branched actin filaments (Rottner et al., 2010). However, a critical facet of how WASP activates this regulation remains elusive. Alternatively, an elegant study reveals that phosphorylation on residues T237, T238, and Y202 of the ARP2 subunit is critical for the activity of the ARP2/3 complex and the formation of lamellipodia (LeClaire et al., 2008). In this model, phosphorylation at these sites can stabilize the complex into a conformation that is permissive for robust activation, probably through the nucleation promoting factors (NSF) (LeClaire et al., 2008). In support of this notion, Nck-interacting kinase (NIK), a Ste20/MAP4K4 serine/threonine kinase, was found to interact with and phosphorylate the ARP2/3 complex to increase its actin nucleation activity and further promote membrane protrusions in response to EGF (LeClaire et al., 2015). In addition, more recent evidence suggests that PLK4 appears to specifically phosphorylate ARP2 at the T237 and T238 residues, and thus modulate the ARP2/3-driven cancer cell migration and invasion (Kazazian et al., 2017). On this basis, my studies further characterize the molecular mechanism and identify the CEP85- STIL complex as a novel modulator of PLK4-ARP2/3 mediated actin remodelling and cell migration. However, the key question remains to be investigated where PLK4 plays its roles in this regulation, since my data suggests that PLK4 did not show obvious localization at the cell

113 cortex. It is possible that PLK4 localization is transient and may not be detected upon immunofluorescence staining. Another possibility is that there may be additional factors that associated with PLK4 and function in the cytoplasm to perform this regulation, which requires future investigation. In addition, my results indicate that inhibition of PLK4 kinase activity disrupts the localization of CEP85 at the cell cortex but has no effect on the PLK4-CEP85 interaction, implying that PLK4 may phosphorylate CEP85 to direct its cortical localization. Overall, it is of particular interest to consider that the conserved PLK4-CEP85-STIL module identified here may not only provide a novel molecular framework of centrosome-based actin nucleation, but also raise the possibility for their non-centrosomal activities of proteins traditionally thought to be centrosome-resident in the control of actin assembly at the cell cortex.

Given the weak and transient nature of the CEP85-STIL complex, it will be interesting to utilize the high-sensitivity real-time imaging method to determine their functional associations in cell physiology. In this context, a combination of Förster resonance energy transfer (FRET) and fluorescence lifetime imaging microscopy (FLIM) ensures the efficient detection for transient or dynamic signal changes from probes, which will allow us to visualize and quantify the spatial- temporal interactions between CEP85 and STIL at the cell cortex and centrosomes. As a functional readout, we could also define the localization of those activated PLK4 that are relied on the CEP85-STIL interaction.

PLK4 has also been implicated in the control of exosome-WNT signalling stimulated cancer cell motility (Luo et al., 2019). In this model, CEP192 interacts with both PLK4 and AURKB. In response to exosome-WNT signaling simulation, DVL2 initiates the recruitment of the CEP192- PLK4-AURKB complex to cell protrusions (Luo et al., 2019). This promotes CEP192 localization to the cell cortex, PLK4 stability and AURKB nuclear exit. In turn, PLK4 and AURKB mediate switching of the formin DAAM1 for DAAM2 in cell protrusions, which then drives actin re-organization and non-directional cell migration (Luo et al., 2019). My data suggests that CEP85 can directly interact with CEP192. On these bases, future investigation should examine whether CEP85 could work with CEP192 to regulate the loading or activation of PLK4-AURKB module at the cell cortex to stimulate formin-dependent actin nucleation and to drive exosome-induced cancer cell motility. Given the important roles of PLK4 and AURKB in cancer development, it will be of great importance to use the mouse xenograft model to determine whether the disruption of the CEP85-STIL interaction constitutes to cancer invasions

114 and metastases in vivo, which may enable the development of a novel therapeutic approach for cancer treatment.

CEP192 acts as the most upstream factors in centriole duplication pathway, but also plays a role in regulating mitotic spindle assembly and cell motility. It has been shown that depletion of CEP192 results in a significant decrease in centrosome microtubule nucleation and a marked increase in both less dynamic acetylated microtubules and cell length/width, which lead to cell hyperpolarization and defective cell migration (O‟Rourke et al., 2014). My results indicate that the N-terminus of CEP85 can directly bind to CEP192. Moreover, CEP192 has been indicated as the upstream of PLK4 to regulate its recruitment to the cell cortex in order to maintain the molecular functions of PLK4 (Luo et al., 2019). These observations raise the possibility of CEP192 acting upstream of CEP85 in the control of both centriole duplication and cell motility. However, my data suggests that the CEP192-CEP85 interaction has no effect on these regulations in U-2 OS cells. Thus, future work should validate this notion using different types of human cells as well as to determine the molecular function of their interaction in other cellular processes such as mitotic spindle assembly.

It remains of importance to find out, mechanistically, how CEP85-STIL complex coordinates with PLK4 in activating ARP2/3 dependent actin nucleation, how this discrete module balances its centrosomal versus non-centrosomal regulation, and whether there are additional factors involved in this regulatory pathway. Taken together, my studies support the contextual roles of the PLK4-STIL module in the control of cancer cell motility via their interplay with CEP85. This also provides mechanistic insights into an unprecedented role of centrosomal components in actin nucleation that lies beyond the traditional view of centrosome and microtubule functions. Given the importance of the elevated level of PLK4 described in aggressive cancers, the CEP85- STIL binding interface might be a potential therapeutic target for treating cancers overexpressed PLK4.

115

3.6 Materials and Methods

Cell culture

U-2 OS cells were grown under standard conditions. U-2 OS T-REx cells were cultured in McCoy 5A medium with 10% FBS, 2mM GlutaMAX, zeocin (100 μg/ml) and blasticidin (3 μg/ml). U-2 OS T-REx cells with Tet-inducible Myc-tagged PLK4 were a kind gift from E. Nigg, and were maintained in 10% FBS, 2mM GlutaMAX and G418 (0.5 mg/mL). HCT116 cells were maintained in McCoy‟s 5A medium with L-glutamine (Life Technologies) supplemented with 10% FBS (Life Technologies) and 1% Penicillin/Streptomycin (Life Technologies). All human cell lines were cultured in a humidified 5% CO2 atmosphere at 37°C.

RNA interference

Luciferase duplex GL2 (CGUACGCGGAAUACTTCG) from Dharmacon was used as a control. The siRNAs against human PLK4 (M-005036-02-0005), CEP192 (L-032250-01-0005) and CEP152 (M-022241-01) were purchased from Dharmacon. CEP85 and STIL were depleted using the following siRNA oligonucleotide sequences: human CEP85 siRNA, 5'- CCUAGAGCAGGAAGUGGCUCAAGAA-3′ and human STIL siRNA, 5'- GCUCCAAAC AGUUUCUGCUGGAAU-3 ′ (Liu et al., 2018). Transfections were performed using Lipofectamine RNAiMax (Invitrogen) according to the manufacturer‟s protocol.

Cloning and stable cell lines

GFP or FLAG tagged CEP85 fragments were recombined into pcDNA5/FRT/TO vector backbone (Life Technologies) using a tetracycline-inducible CMV promoter. U-2 OS T-REx cells carrying full length CEP85 and STIL or CEP85 Q640A/E644A and STIL L64A/R67A mutants were generated in our previous study (Liu et al., 2018). U-2 OS T-REx cells were co- transfected with pOG44 (Flp-recombinase expression vector) and pCMV-TO/FRT-FLAG plasmid containing the coding sequence for siRNA resistant human CEP85 ∆M and ∆N transgenes. Transfections were carried out with Lipofectamine 2000 (Invitrogen) according to

116 the manufacturer‟s protocol. At 24 h post-transfection, cells were selected with 200 μg/ml hygromycin B.

Wound healing assay

U-2 OS cells were seeded into 96-well plates and transfected with the indicated siRNAs. At 24 h post transfection, cells were switched medium containing 0.5% FBS. At 48 h post transfection, a single scratch wound was generated using IncuCyte™ Wound Maker (Essen BioScience). Furthermore, IncuCyte™ live-cell imaging systems (Essen BioScience) were utilized to conduct the 96-well scratch wound cell migration assay according to the manufacturer‟s protocol. Imaging was taken at 1 h intervals up to 24 h using a 10X objective lens. Image analysis was performed using ImageJ to measure the area of healed wound at t=6 h, 12 h, 18 h and 24 h.

For measuring the axis of nucleus-centrosome-Golgi apparatus, cells were fixed 0 or 6 h after generating wounds and stained with DAPI and antibodies specific to Pericentrin and GM130. Axis of nucleus-centrosome-Golgi apparatus was determined from the middle of the nucleus to the middle point between centrosome and Golgi apparatus. Angles between leading edge and axis of nucleus-centrosome-Golgi apparatus were measured with Image J. Radar graphs were drawn in RStsudio (Ver 1.2.1335) with ggplot2 library.

Transwell assay

1.0 × 106 U-2 OS cells were seeded on 6-well plates and then indicated siRNAs were transfected next day. One day after siRNA transfection, cells were grown in media containing 0.5% FBS for 24 h. For migration assay, 5.0 × 104 cells were spread into the transwell (Millipore Sigma, CLS3422) with 0.1 ml of media containing 0.5% FBS and then the transwells were inserted into 24 wells containing 0.6 ml of media supplemented with 10% FBS. After 24 h cells were fixed with 4% PFA for 2 minutes and cold methanol for 10 minutes. After removing cells inside the transwell, cells were stained with 0.5% crystal violet for 20 minutes. Cells were imaged on a Deltavision Elite DV imaging system (GE Healthcare) equipped with a sCMOS 2048x2048 pixels2 camera (GE Healthcare).

117

Spreading assays

U-2 OS cells were seeded into 6-well plates and transfected with the indicated siRNAs. At 72 h post transfection, cells were trypsinized and replated onto 12-well plate with glass coverslips for another 6 h. Next, cells were fixed by 4% PFA and immunostained with Alexa Fluor 488 Phalloidin for labeling F-actin. Imaging were acquired on a Deltavision Elite DV imaging system equipped with a 60×/1.4 NA objective and a sCMOS 2048x2048 pixels2 camera (GE Healthcare). Z stacks (0.2 μm apart) were used, and images were deconvolved and projected using softWoRx (v6.0, Applied Precision). Quantification of cell area based on F-actin signal was performed using ImageJ.

Immunofluorescence microscopy

Centrin and PCNT staining were performed following standard protocols as previously described (Liu et al., 2018). For the Phalloidin staining, cells were fixed by 4% PFA at room temperature for 20 minutes and treated with permeabilization buffer (0.5 % Triton X-100 in PBS) for another 10 minutes. Cells were then blocked in 1% BSA in PBS for one hour and incubated with Alexa Fluor 488 Phalloidin and DAPI in blocking solution for another one hour. After a final wash three times with PBS for 5 minutes each, cells were inverted and mounted on glass slides with standard mounting solution (ProLong Gold antifade, Molecular Probes). Cells were imaged on a Deltavision Elite DV imaging system (GE Healthcare) equipped with a sCMOS 2048x2048 pixels2 camera (GE Healthcare). Z stacks (0.2 μm apart) were collected, and images were deconvolved and projected using softWoRx (v6.0, Applied Precision).

Co-Immunoprecipitation

The respective 293T or U-2 OS stable lines were seeded into 10-cm2 dishes and 24 h later transfected with 4 µg of plasmid DNA and incubated with tetracycline (2μg/ml). At 48 h post- transfection, transfected cells were washed with 1x PBS, then harvested and lysed immediately

118 in lysis buffer (50mM HEPES pH8; 100mM KCl; 2mM EDTA; 10% Glycerol; 0.1% NP-40; 1mM DTT; protease inhibitors (Roche)) for 30 minutes on ice. Lysates were frozen in dry ice for 5 minutes, then thawed and centrifuged for 20 minutes at 16,000xg at 4ºC. Cleared supernatant were incubated with anti-FLAG M2 Affinity Gel (Sigma-Aldrich) for 3 hours at 4ºC. A fraction of the protein extracts (Inputs) were saved before the incubation with the beads. After the incubation, the beads were pelleted and washed three times in lysis buffer. Samples (Inputs and IPs) were prepared by addition of Laemmli buffer and boiling at 95ºC for 5 minutes. Immunopurified proteins were analyzed by immunoblotting with the indicated antibodies.

Western blots

Cells were lysed in Laemmli sample supplemented with a phosphatase inhibitor cocktail and benzonase nuclease. For phospho-ARP2 T237+T238 detection, cells were treated with 5.0 mmol/L pervanadate for 10 minutes before lysis. Proteins were loaded to 8% SDS-PAGE gel for electrophoresis and then electroblotted onto PVDF membranes (Immobilon-P, Millipore). Membranes were incubated with primary antibodies in TBST (TBS, 0.1% Tween-20) in 5% skim milk powder or 5% BSA for the anti-ARP2 (phospho T237 + T238) antibody overnight at 4°C. Membranes were washed 3x10 minutes in TBST, and then incubated with secondary antibodies conjugated to HRP for 1 h at room temperature. Western blots were developed using SuperSignal reagents (Thermo Scientific).

Flow cytometry

U2OS cells were harvested 72h post-siRNA transfection from 10cm plates and washed with wash buffer (PBS+4% FBS). Fixation was performed using 4% paraformaldehyde for 10min at RT followed by permeabilization with 0.1% Triton-X for 15min at RT. Cells were washed and resuspended in wash buffer. DAPI was added at 1:100 final dilution and cells were incubated in the dark at RT for 15min. Flow cytometric analysis was performed on Fortessa X-20 (BD) using the UV laser for excitation and DAPI fluorescence was measured at 461nm. 50,000 events were

119 recorded for each sample. Doublet cell population was gated out and cell cycle distribution analysis was modeled using ModFit analysis and cells in G1, S and G2/M phases were quantified.

Yeast Two Hybrid

Constructs were cloned into vector pENTR/D-TOPO and transferred into vectors pDEST32 (Bait, human STIL NTD) or pDEST22 (Prey, human CEP85 constructs) using site-specific recombination (Gateway LR Clonase II, Thermo Fisher Scientific). Bait and Prey plasmid combinations were co-transformed into yeast strain MaV203 (Thermo Fisher Scientific) and plated onto SC-Leu-Trp plates. Colonies were inoculated into SC-Leu-Trp medium, grown and spotted onto SC-Leu -Trp and SC-Ura plates. Plates were incubated for three days at 30°C.

BioID, mass spectrometry and analysis

BioID and mass spectrometry was performed as described (Gupta et al., 2015). The network in Supplementary Figure 1a was constructed from SAINT data were imported into Cytoscape 3.2.1 31, and with the Edge-Weighted Spring Embedded Layout. SAINT analysis of high confidence interactors to generate Supplementary Table 1 used the following parameters: TPP >0.9, unique peptides≥2, FDR ~1% 32. For each prey, peptide counts in each bait sample (two technical replicates of two biological replicates) were compared to those from 14 control runs.

Statistical methods

Two-tailed unpaired student t-tests were performed for all p-values. Individual p-values, experiment sample numbers and the number of replicates for statistical testing were indicated in corresponding figure legends. Unless otherwise mentioned, all error bars are SD, and the asterisk placeholders for p-values are **p<0.01 and *p<0.05.

120

4 Chapter IV: Functional characterization of ANK2 in centriole duplication

121

4.1 Statement of Contributions

All the data in this chapter has been published as an article in Cell:

Gagan D. Gupta1,*, Étienne Coyaud2,*, João Gonçalves1,*, Bahareh A. Mojarad1,3,*, Yi Liu1,3, Qianzhu Wu1,3, Ladan Gheiratmand1, David Comartin1,3, Johnny M. Tkach1, Sally W.T. Cheung1, Mikhail Bashkurov1, Monica Hasegan1, James D. Knight1, Zhen-Yuan Lin1, Markus Schueler6,7, Friedhelm Hildebrandt6,7, Jason Moffat5, Anne-Claude Gingras1,3, Brian Raught2,4,§ and Laurence Pelletier1,3,§ 1Lunenfeld Tanenbaum Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, Ontario, M5G 1X5, Canada 2Princess Margaret Cancer Centre, University Health Network, 101 College Street, Toronto, ON M5G 1L7, Canada 3Department of Molecular Genetics, University of Toronto, Toronto, Ontario, M5S 1A8, Canada 4Department of Medical Biophysics, University of Toronto, Toronto, Ontario, M5G 1L7, Canada 5Donnelly Centre and Banting and Best Department of Medical Research, University of Toronto, 160 College Street, Toronto, ON M5S 1A8, Canada 6Division of Nephrology, Department of Medicine, Boston Children's Hospital, Harvard Medical School, 300 Longwood Avenue, Boston, MA, 02115, USA 7Howard Hughes Medical Institute, Chevy Chase, MD, 20815, USA *These authors contributed equally to this work.

Author contribution:

Creation of BioID cell lines for centriole and appendage components was carried out by B.A.M., S.W.T.C., and Q.W. BioID of TZ and satellite proteins was conceived and developed by J.G. E.C. and B.R. performed all BioID, MS, and related analyses. Z.-Y.L. and S.W.T.C. performed FLAG IP-MS. Network and computational analysis was carried out by E.C. and G.D.G., with suggestions from J.G. Co-IP validation work was executed by J.G. and B.A.M. J.G. and M.B. conducted the cilia screen. J.G. conceived and executed WRAP73 experiments. Prey localization was carried out by J.G., L.G., B.A.M., M.H., G.D.G., and J.M. Centriole over-duplication screen was performed by G.D.G. and Q.W. Centriolar satellite screen and analysis were conducted by G.D.G. Work on ANK2 was performed by Y.L. (Figure 4.1C-D and Figure 4.2-4.4) and D.C. (Figure 4.1A-B), and work on CEP128 was conducted by B.A.M. Analysis of RNAi screening data carried out by G.D.G. J.M.T. conducted all CRISPR experiments. A.-C.G. and J.K. assisted with MS data analysis, DotPlots, and data porting to ProHits Web. M.S. and F.H. sequenced

122 nephropathy patients. The paper was written by G.D.G., B.R., and L.P. with contributions from all authors. L.P and B.R. directed the project.

123

4.2 Summary

Centrosomes function as the primary microtubule organizing center (MTOC) in cells and are critical for a set of fundamental cellular and developmental processes. At the core of centrosomes are centrioles, which recruit pericentriolar material and template the formation of cilia/flagella. In order to fulfill their roles, the duplication of centrioles must be controlled temporally, spatially and numerically. Abnormalities in centriole duplication cause a remarkable range of human disorders, including cancer, microcephaly and ciliopathies. Despite the importance, the mechanisms governing centriole duplication are not fully understood. Thus, the aim of this project is to study the molecular mechanism by which CEP120 regulates centriole duplication. Using BioID, I identify ANK2 as a putative proximity interaction partner of CEP120. Functionally, ANK2 is able to localize at centrioles and plays an important role in centriole duplication. I find that depletion of ANK2 redistributes CEP120 on microtubules and reduces its recruitment to centrioles, thereby negatively affecting centriole assembly. In addition, a marked increase in microtubule stability is also observed in cells depleted of ANK2, which may contribute to the abnormal spatial distribution of centriolar satellites. Taken together, my results indicate that ANK2 functions as upstream of CEP120 to regulate centriole duplication, and also plays an important role in the control of microtubule dynamics.

124

4.3 Introduction

A finely regulated centriole duplication process is critical for faithful cell division. Abnormities in this process can lead to numerous human diseases, such as microcephaly and cancer. Despite recent progress into elucidating the molecular mechanisms of centriole assembly, many aspects of this process are still not fully understood. It is well established that the centriole assembly has three continues steps, which starts from the establishment of the initiation complex (CEP192- CEP152-PLK4), the following cartwheel assembly (STIL-SASS6), and the final stage centriole elongation (CPAP-CEP120-SPCIE1-CEP135) (Nigg and Holland, 2018). In this project, I aim to understand the molecular mechanism underlying centriole length control. Over-elongated centrioles have been found to trigger centriole amplification and to increase microtubule nucleation capacity along with defective chromosome segregation. Importantly, centriole over- elongation has been observed in multiple cancer cell lines, which is proposed as an additional centrosomal mechanism that contributes to cancer development.

In this project, I aim to determine the molecular mechanism by which CEP120 controls centriole elongation. Previous studies have characterized the biological functions of CEP120 in cell physiology and disease. During neocortical development, the silencing of CEP120 in the neocortex significantly inhibits interkinetic nuclear migration and neural progenitor self-renewal (Xie et al., 2007). Loss-of-function mutations in CEP120 have been linked to the development of Joubert Syndrome as well as complex ciliopathy phenotypes (Roosing et al., 2016, Joseph et al., 2018). Molecularly, CEP120 was first identified as a centrosome resident protein through a proteomic analysis of purified human centrosomes (Jakobsen et al., 2011). Functionally, CEP120 asymmetrically localizes at newly formed centrioles and depletion of CEP120 results in centriole assembly defect (Mahjoub et al., 2010). Further studies reveal that CEP120 cooperates with SPICE and CPAP to form a regulatory cascade for the recruitment of CEP135 to regulate centriole elongation (Lin et al., 2013b, Comartin et al., 2013). However, it still remains elusive how CEP120 cooperates with these factors to precisely control centriole elongation in space and time. To address this question, I aim to examine the published centrosome proteomic dataset to identify novel interactors of CEP120 (Gupta et al., 2015) and further characterize the molecular mechanism.

125

Our proteomic analysis using BioID has identified Ankyrin-B (ANK2) as a putative proximity interaction partner of CEP120. Canonical ANK2 is a 220 kDa microtubule-binding protein expressed in most tissues and cell types (Bennett and Chen, 2001). It consists of an N-terminal containing membrane-binding domain (MBD), a central region with the spectrin-binding domain (SBD), a death domain (DD) and a C-terminal regulatory domain (RD) (Bennett and Chen, 2001). The MBD domain contains 24 ANK repeats, which are multivalent and can accommodate multiple protein interactions (Bennett and Chen, 2001). The death domain in other proteins may function in activation of NF-kB to trigger cell death, however, this domain has no known role in ANK2 (Bennett and Chen, 2001). The spectrin-binding domain is reported to interact with spectrin, an actin-associated protein that provides the structural support to the plasma membrane. In addition, the SBD is also required of the binding of ANK2 to Dynactin 4 (P62), a key component of the Dynein-Dynactin motor complex that regulates microtubule-based transport (El Refaey and Mohler, 2017). Functionally, the interaction between spectrin and ANK2 plays an important role in normal cardiac physiology, and loss-of-function mutations in ANK2 have been linked to long QT syndrome, an autosomal dominant cardiac disorder (El Refaey and Mohler, 2017). Besides, studies of ANK2 knockout mice reveal the physiological role of ANK2 as a general adaptor for the Dynein-Dynactin complex to promote axonal transport of synaptic vesicles, which is required for normal axon length (Lorenzo et al., 2014).

Here, I identify ANK2 as a novel interactor of CEP120, which is required for efficient centriole assembly. Molecularly, I find that ANK2 is able to localize at centrioles through its N-terminal membrane-binding domain. Further analysis reveals that depletion of ANK2 reduces the centriole loading of CEP120 and leads to a redirstubution of CEP120 along cytoplasmic microtubules. Intriguingly, ANK2 depletion also results in a marked increase in microtubule stability, which may account for the defect in the spatial distribution of centriolar satellites. Overall, this work reveals the multipronged functions of ANK2 as a novel centriole duplication factor, a regulator of CEP120 localization and a modulator of microtubule dynamics. However, the mechanistic understanding of the diverse biological effects of MECP2 deficiency on cell physiology still remains enigmatic.

126

4.4 Results

4.4.1 ANK2 is A Novel Centriole Duplication Factor

To identify novel factors implicated in centriole elongation, we initially sought to map protein interactions of centriole components using proximity-dependent biotin identification. In this established proteomic dataset, we applied the hierarchical clustering algorithm in protein interaction networks and focused on individual groups sharing similar proximity profiles. To this end, we identified one distinct protein cluster comprising several known centriole elongation factors including CPAP, CEP120, SPICE1, and CEP135 (Figure 4.1A). Interestingly, a set of microtubule-associated proteins such as ANK2, MTUS1, MAP7, NAP1L1, MAP1S, and MAP9, were also clustered in this group, and we found that ANK2 appeared to interact exclusively with CEP120 in this module (Figure 4.1A). To better understand the role of this module in centriole functions, the previously described PLK4 over-expression induced centriole over-duplication assays were used to perform a small-scale RNAi screen to assess their effect in centriole amplification, centriole over-elongation, CEP120 mis-localization, tubulin glutamylation (Figure 4.1B). Notably, ANK2 was found to negatively affect all of these centriole functions (Figure 4.1B). Therefore, these findings raise the possibility that ANK2 may participate in CEP120- driven centriole assembly.

To test this hypothesis, I depleted endogenous ANK2 using two deconvoluted siRNAs in U-2 OS and western blot experiments indicated that the siRNA2 displayed higher knockdown efficiency. Therefore, ANK2 siRNA2 was used in the following functional assays. Here, my first step was to confirm the role of ANK2 in centriole duplication. To do this, I performed the genetic rescue experiments in U-2 OS cells harboring a Tet-inducible RNAi resistant GFP-ANK2 transgene. Cells were then submitted to immunofluorescence analysis and assayed for the centriole number using Centrin as a marker. Consistently, depletion of ANK2 strongly inhibited normal centriole duplication, while the expression of RNAi resistant ANK2 fully rescued this defect (Figure 4.1C). Further electron microscopy experiments revealed that the remaining centrioles in cells depleted of ANK2 were significantly over-elongated, when compared to the control (Figure 4.1D). These results suggest that ANK2 plays a critical role in regulating centriole duplication, probably through the CEP120-drvien centriole length control.

127

Figure 4.1. ANK2 is required for centriole duplication.

(A). Left: hierarchical clustering reveals modules with similar proximity profiles (dashed boxes). Peptide counts (MaxSpec) indicated for each interact. Right: mass spectrometry DotPlot of selected baits from the region of the CEP120/SPICE1 cluster. *SASS6 was not part of this interactor group and is included here as a control.

128

(B). Left: interactors (presented in the same order as in A were profiled for suppression of centriole amplification; see Figure 4A). Right: qualitative phenotypes related to MT organization (see legend and Experimental Procedures). Dashed purple (overduplication assay) or black (all other assays) lines represent 1.9 and 2 SD of the mean of negative control values (*p < 0.05, n > 100, three replicates), respectively. Negative control values for MT phenotypes were <2% in all cases (not shown).

(C). Effect of ANK2 depletion on centriole number. U-2 OS lines carrying Tet-inducible GFP or the siRNA-resistant GFP-ANK2 (GFP-ANK2*) transgenes were transfected with control or ANK2 siRNA for 72 hr. At 48 hr post-transfection, tetracycline and hydroxyurea were added to induce ANK2 expression and arrest cells in S-phase for 24 hr. Cells were fixed and labeled with anti-centrin, and the number of centrioles per cell was counted. Bar graph, percent cells with indicated centriole number (n > 300, three replicates, *p < 0.05, **p < 0.01).

(D). Left: electron micrographs of U-2 OS Tet-inducible Myc-PLK4 cells transfected with control (top) or ANK2 (bottom) siRNA for 72 hr. At 48 hr post-transfection, hydroxyurea and tetracycline were added for 24 hr to arrest cells in S-phase and induce centriole overduplication. Right: average centriole length (±SD) in each population (*p = 0.03, Student‟s t test, n > 15). Scale bar, 200 nm.

4.4.2 ANK2 Localizes at Centrosomes

Next, I sought to determine whether ANK2 is able to localize at centrosomes. My first step was to perform IF staining to examine the localization of endogenous ANK2. The monoclonal anti- ANK2 antibody has been validated for its specificity using western blot. The positive control indicated the presence of canonical ANK2, which was significantly down-regulated in ANK2 siRNA transfected cells (Figure 4.2A). However, the IF staining using this antibody displayed strong cytoplasmic background signal, which did not disappear upon ANK2 depletion. This makes it hard to detect the endogenous ANK2 localization at centrosomes. Alternatively, I over- expressed GFP tagged full length ANK2 and four ANK2 fragments (N1, N2, C1 and C2) in U-2 OS cells and examined their localization (Figure 4.2B). The results indicated that full length

129

ANK2 or N1 and N2 fragments were able to localize at centrosomes, suggesting that N-terminal MBD domain is responsible for this localization (Figure 4.2C).

130

Figure 4.2. The localization of ANK2 at centrosomes.

(A). Western blot analysis of ANK2 and a-tubulin levels in U-2 OS cells transfected with control or ANK2 siRNAs.

(B, C). Full-length (FL) and truncation constructs of GFP-ANK2 (top) were transfected into U-2 OS cells. IF microscopy was used to characterize localization at the centriole (boxed in white) with a marker (centrin). Right: Insets from top to bottom: GFP, centrin, pseudocolor merge. Scale bar, 10 µm, insets 3x.

4.4.3 ANK2 is Required for the Centriolar Localization of CEP120

Our BioID data suggest that ANK2 is a putative proximity interaction partner of CEP120. To confirm this interaction, I performed the co-IP experiments with cell lysates expressing FLAG- BirA*-tagged ANK2 to pull down endogenous CEP120. The data suggested that ANK2 was able to interact with CEP120 (Figure 4.3A), implying that these two proteins may work together to control centriole duplication. To test this hypothesis, I assessed the protein levels and localization of CEP120 in ANK2 siRNA transfected cells. The results indicated that ANK2 depletion significantly reduced the recruitment of CEP120 to centrioles (Figure 4.3B). Interestingly, I found that ANK2 depletion also caused the redistribution of CEP120 in the cytoplasm to form filament-like structures (Figure 4.3B). To further characterize this mis- localization phenotype, I used three-dimensional structured-illumination microscopy (3D-SIM) and performed the IF staining with anti-CEP120 and α -tubulin antibodies. The imaging data revealed that ANK2 depletion led to the redistribution of CEP120 along cytoplasmic microtubules (Figure 4.3C). Further western blot analyses demonstrate that ANK2 deficiency resulted in a marked increase in the total cellular levels of CEP120 (Figure 4.2A). Together, these data suggest that ANK2 plays a key role in the control of CEP120 localization to centrosomes, probably through a microtubule-dependent manner.

131

132

Figure 4.3. ANK2 is required for CEP120 localization to centrosomes.

(A). Western blots (as indicated) of input (left) or FLAG IP (right) conducted on lysates from 293 T-REx cells stably expressing FLAGBirA* (tag alone; control) or FLAGBirA*-ANK2, transfected with MYC-CEP120.

(B) Representative micrographs (left) of U-2 OS cells treated with control (top) or ANK2 (bottom) siRNA and labeled with antibodies to endogenous CEP120 (green). White arrowheads denote centrosomal CEP120 puncta. Cytoplasmic regions where CEP120 has relocalized are encircled by white dashes. Scale bar, 15 µm. Right: quantification of CEP120 levels at the centrosome (n > 300, three replicates, **p < 0.01, Student‟s t test). Grey region denotes 2 SD from the mean (red line) and pink region denotes the 95% confidence interval.

(C). 3D-SIM micrographs of control and ANK2-depleted U-2 OS Myc-PLK4 cells after PLK4 induction and S-phase arrest, labeled with the indicated antibodies and counterstained with DAPI (see Materials & Methods). Boxed inset shows centrosomal label as marked by CEP120 and its corresponding microtubule distribution in a single cell. Scale bar 5 µm, insets 1 µm.

4.4.4 ANK2 Controls the Dynamics of Microtubules

ANK2 has been linked to the control of microtubule architecture to regulate the transport of synaptic vehicles. On this basis, I aim to further assess the effect of ANK2 depletion on microtubule functions. My 3D-SIM imaging data suggested that ANK2 depletion increased the acetylation levels of microtubules. Consistently, western blot experiments showed a marked increase in the total cellular levels of microtubule acetylation. Notably, I found those CEP120 in ANK2 depleted cells were re-distributed on acetylated microtubules (Figure 4.4A). Acetylated microtubules have been considered to be as a marker as stable, long-lived microtubules; however, until recently, it still remains elusive about whether the longevity of these microtubules is the cause or the consequence of acetylation. To determine the role of ANK2 in microtubule stabilization, I performed the canonical microtubule depolymerization assays using the ice treatment, in order to determine the levels of stabilized microtubules. As expected, approximately 90% of control cells disassembled their microtubules upon the cold treatment,

133 whereas a large portion of ANK2 depleted cells still possessed intact microtubule structures (Figure 4.4C), supporting the notion of ANK2 as a regulator of microtubule stability.

Given that ANK2 is associated with the Dynein-Dynactin motor complex, it is possible that ANK2 also regulates the microtubule-based transport of centriolar satellites. To test this possibility, I performed 3D-SIM imaging and found that ANK2 depletion also redistributed centriolar satellites along with CEP120 on microtubules (Figure 4.4D). Centriolar satellites are small electron-dense granules that cluster in the vicinity of centrosomes. By using this morphological feature, I analyzed the ANK2 depleted cells using electron microscopy. Consistently, I found that these centriolar satellites were barely detected in the proximity of centrosomes upon ANK2 depletion, when compared to the control (Figure 4.4E). Taken together, these findings reveal that ANK2 is a microtubule regulator that controls the spatial distribution of centriolar satellites, which provides an additional mechanism to explain the role of ANK2 in CEP120 localization and centriole duplication.

134

135

136

Figure 4.4. ANK2 regulates the dynamics of microtubules.

(A). 3D-SIM micrographs of S-phase arrested, Myc-PLK4 overexpressing U-2 OS cells transfected with control and ANK2 siRNA and labeled with the indicated antibodies and counterstained with DAPI. Scale bar 5 µm, insets 4x.

(B). Top: Western blot analysis of a-tubulin and acetylated (Ac.) tubulin in U-2 OS cells transfected with control or ANK2 siRNAs. Bottom: densitometric quantitation of the blot.

(C). Micrographs of U-2 OS cells treated with control (NT siRNA) or ANK2 siRNA and subjected to cold shock (see timeline, top right), followed by fixation and labeling with the indicated antibodies. Percentage of cells with stabilized microtubules were quantified (n > 300, three replicates, **p < 0.01, *p < 0.05, Student‟s t test). Scale bar 10 µm.

(D). 3D-SIM micrographs of control and ANK2-depleted U-2 OS Myc-PLK4 cells after PLK4 induction and S-phase arrest, labeled with the indicated antibodies and counterstained with DAPI (see Supplemental Experimental Procedures). Boxed inset shows centrosomal label as marked by CEP120 and its corresponding PCM1 satellite distribution in a single cell. Scale bar 5 mm, insets 1 µm.

(E). Electron micrographs of U-2 OS Tet-inducible Myc-PLK4 cells transfected with control or ANK2 siRNA for 72 hr. At 48 hr post-transfection, hydroxyurea and tetracycline were added for 24 hr to arrest cells in S-phase and induce centriole overduplication. Right: average centriole length (±SD) in each population (*p = 0.03, Student‟s t test, n > 15). Scale bar, 500 nm.

137

4.5 Discussion

In the past decade, significant progress has been made to identify and map a list of conserved proteins into a regulatory pathway for centriole duplication. However, the current challenge is to investigate how these regulators work together to mediate the step-wise centriole assembly. To address this limitation, our lab utilized BioID to generate the first human centrosome-cilium protein interaction network, which provides rich information to systematically characterize novel centrosome components (Gupta et al., 2015). By leveraging this dataset, I identify ANK2 as a novel interactor of CEP120. Functionally, ANK2 has been found to localize at centrosomes through its N-terminal membrane-binding domain and depletion of ANK2 led to a marked decrease in the recruitment of CEP120 to centrioles. Moreover, I also observed that ANK2 plays a critical role in the control of microtubule stability, which may account for the defective localization of centriolar satellite in cells depleted of ANK2. Taken together, this study provides molecular insights into the role of ANK2 in both centriole assembly and microtubule regulation, and also improves our understanding of the CEP120-driven centriole length control. It is clear that the role of ANK2 is complex, however, no consensus has emerged that would synthesize the multiple functions ascribed to this protein into a unifying hypothesis to explain how ANK2 coordinates centriole duplication, CEP120 localization, microtubule stability, centriolar satellite distribution in space and time.

It has already been shown that the central domain is able to directly interact with Dynactin4. Dynactin4 is a core component of the Dynein-Dynactin motor complex, which is involved in the microtubule minus-end transport of cargos to centrosomes (Liu, 2017). Through this association, previous work shows that ANK2 can serve as a generic scaffold protein for Dynactin complex to mediate efficient transport of synaptic vesicles and thus maintain normal axonal length (Lorenzo et al., 2014). In line with this notion, my data suggest that ANK2 depletion redistributes both CEP120 and centriolar satellites along cytoplasmic microtubules, implying that the microtubule- based transport may be disrupted in ANK2 depleted cells. To this end, future research will be required to assess the effect of ANK2 deficiency on the functional and molecular properties of the Dynein-Dynactin complex, and to determine whether this complex directs the centriole loading of CEP120 or its associated factors such as CPAP, CEP135 and SPICE1.

138

The N-terminal membrane-binding domain of ANK2 contains 24 Ankyrin repeats, which are known as a multivalent region that favors multiple its interactions with a number of membrane proteins and microtubules (Bennett and Chen, 2001). My data reveals that the MBD domain is required to direct the recruitment of ANK2 to centrioles, and expression of this domain alone displays much stronger centriolar localization. These multiple lines of evidence raise the possibility that the Ankyrin repeat region may serve as a scaffold domain to facilitate protein interactions at centrosomes. Therefore, it will be of great interest to perform proteomic analysis to identify which centrosome components are associated with this region, and further characterize the molecular functions of these interactions.

Microtubules are linear polymers composed of α/β-tubulin heterodimers. The GTP bound to β- tubulin is hydrolyzed to GDP. This GTP hydrolysis weakens the binding capacity of tubulin to other molecules, thereby favoring the dynamic behaviors of microtubules and leading to depolymerization. Microtubule dynamics are critical for intracellular vesicle transportation and the integrity of centriolar satellites, which plays an important role in maintaining the proper functions of centrosomes (Hori and Toda, 2017a). Notably, HDAC6 is known as a microtubule- associated deacetylase, which displays the extensive co-localization with p150Glued, a core subunit of the Dynactin complex (Hubbert et al., 2002). Over-expression of HDAC6 has been found to result in global deacetylation of α-tubulin, whereas HDAC6 deficiency increases α- tubulin acetylation in assembled microtubules (Hubbert et al., 2002). Moreover, recent work reveals that inhibition of HDAC6 deacetylase activity promotes its interaction with microtubules and leads to a marked increase in microtubule stability, which is resistant to cold and nocodazole-induced depolymerizing conditions (Asthana et al., 2013). Functionally, HDAC6 is also involved in the control of primary cilia resorption, and depletion of HDAC6 has been shown to negatively affect centriole re-duplication (Yu et al., 2016). Intriguingly, my preliminary western blot experiments indicate that depletion of ANK2 causes a marked reduction in the total cellular levels of HDAC6. Therefore, these multiple lines of evidence raise the possibility that ANK2 may work with HDAC6 to modulate the dynamics of microtubules, which requires future investigation.

However, ANK2 depletion in post-mitotic neurons was found to have no obvious effect on microtubule post-translational modifications (Stephan et al., 2015). It is of particular interest in this context to investigate the Cell-type- and tissue-specific mechanisms of ANK2-driven

139 microtubule regulation and centriole duplication. Taken together, this study will not only expand the biological functions of ANK2 as a microtubule modulator, but also provide mechanistic understanding of the ANK2-CEP120 module in the control of centriole assembly.

140

4.6 Materials and Methods

Cell culture

U-2 OS cells were grown under standard conditions. U-2 OS T-REx cells were cultured in McCoy 5A medium with 10% FBS, 2mM GlutaMAX, zeocin (100 μg/ml) and blasticidin (3 μg/ml). U-2 OS T-REx cells with Tet-inducible Myc-tagged PLK4 were a kind gift from E. Nigg, and were maintained in 10% FBS, 2mM GlutaMAX and G418 (0.5 mg/mL). HCT116 cells were maintained in McCoy‟s 5A medium with L-glutamine (Life Technologies) supplemented with 10% FBS (Life Technologies) and 1% Penicillin/Streptomycin (Life Technologies). All human cell lines were cultured in a humidified 5% CO2 atmosphere at 37°C.

RNA interference

(Liu et al., 2018). All siRNA transfections were performed using the Lipofectamine RNAiMAX transfection reagent (Invitrogen) according to manufacturer‟s instructions. The Luciferase GL2 Duplex non-targeting siRNA from Dharmacon was used as a negative control. The human ANK2 siRNA (GCAUGUAGCAGCCAAGUAU) target sequence was mutated (CCACGTGGCCGCTAAATAC) using QuikChange Site-Directed Mutagenesis protocol and cloned into pCMV-TO/FRT-Emerald vector vector. Flp-In T-REx U-2 OS cells were co- transfected with pOG44 (Flp-recombinase expression vector) and a pCMV-TO/FRT-Emerald empty plasmid or a pCMV-TO/FRT-Emerald plasmid containing the coding sequence for siRNA1 resistant human ANK2. Transfections were performed with Lipofectamine 2000 (Invitrogen) according to manufacturer‟s instructions. After transfection, cells were selected with 200μg/ml hygromycin B. To silence ANK2, U-2 OS inducible GFP and GFP-ANK2 siRNA1 resistant stable cell lines (3×105 cells seeded in 12 well plates) were transfected with 30nmol of siRNA1 targeting ANK2. At 48 hours post-transfection, cells were induced with tetracycline (1μg/ml) for another 24 hours before methanol fixation to express GFP or GFP-tagged siRNA1 resistant human ANK2.

141

Immunofluorescence microscopy.

For immunofluorescence, cells were fixed with ice-cold methanol (10 minutes at −20°C), then blocked with 0.2% Fish Skin Gelatin (Sigma-Aldrich) in 1x PBS (20 minutes), incubated with the primary antibodies in blocking solution (1 hour), washed with blocking solution and incubated with fluorophore-conjugated secondary antibodies (Molecular Probes) and DAPI , the coverslips were mounted on glass slides by inverting them onto mounting solution (ProLong Gold antifade, Molecular Probes). Cells were imaged on a Deltavision Elite DV imaging system equipped with a sCMOS 2048x2048 pixels2 camera (GE Healthcare). Z stacks (0.2μm apart) were collected, deconvolved using softWoRx (v5.0, Applied Precision) and are shown as maximum intensity projections (pixel size 0.1064μm).

Super resolution microscopy

Super resolution was performed as described. Briefly, cells were imaged on a three-dimensional (3D) structured illumination microscope (OMX v3, Applied Precision) equipped with 405, 488 and 592.5nm diode lasers, electron multiplying CCD (charge-coupled device) cameras (Cascade II 512×512, Photometrics), and a ×60/1.42 NA planApochromat oil-immersion objective (Olympus). 3D-SIM image stacks (0.125μm apart) were reconstructed, aligned and maximum intensity projected using the softWoRx 5.0 software package (Applied Precision).

Electron microscopy

For thin-section EM, U-2 OS cells were grown on Aclar coverslips (Electron Microscopy Sciences), fixed for 1 hour in 2% glutaraldehyde in sodium cacodylate buffer, and post-fixed in 1% osmium tetroxide. Samples were dehydrated through a graded series of ethanol followed by propylene oxide and embedded in Embed 812 resin. Thin sections (90nm) were cut on an RMC MT6000 ultramicrotome, stained with 2% uranyl acetate in 70% methanol and then aqueous lead citrate. Samples were viewed on FEI Tecnai 20 transmission electron microscope (TEM).

142

Western blots

For Western blots, cells were collected, lysed in Laemmli buffer and treated with benzonase nuclease (Sigma-Aldrich). Proteins were subjected to SDS-PAGE and transferred to a PVDF membrane (Immobilon-P, Millipore). Membranes were incubated with primary antibodies in TBST (TBS, 0.1% Tween-20) in 5% skim milk powder (BioShop), supplemented with 2.5% BSA Fraction V (OmniPur) in the case of FLAG Western blots. Blots were washed 3x10 minutes in TBST, then incubated with secondary HRP-conjugated antibodies. Western blots were developed using SuperSignal reagents (Thermo Scientific).

Co-Immunoprecipitation

For co-immunoprecipitation of FLAG-BirA* fusions, the respective 293 stable lines were incubated with tetracycline (1μg/ml) for 24 hours then washed with 1x PBS, harvested and frozen at -80ºC or lysed immediately in lysis buffer (50mM HEPES pH8; 100mM KCl; 2mM EDTA; 10% Glycerol; 0.1% NP-40; 1mM DTT; protease inhibitors (Roche)) for 30 minutes on ice. Lysates were frozen in dry ice for 5 minutes, then thawed and centrifuged for 20 minutes at 16,000xg at 4ºC. Cleared lysates were incubated with anti-FLAG M2 Affinity Gel (Sigma- Aldrich) for a minimum of 3 hours at 4ºC. A fraction of the protein extracts (Inputs) were saved before the incubation with the beads. After the incubation, the beads were pelleted and washed with lysis buffer. Samples (Inputs and IPs) were prepared for SDS-PAGE by addition of Laemmli buffer and boiling. The proteins were transferred to PVDF membranes (Immobilon-P, Millipore) and probed with antibodies to detect the FLAG-BirA* fusions and endogenous proteins.

Statistical methods

All p-values are from two-tailed unpaired Student t-tests. Unless stated, all error bars are S.D. Individual p-values, experiment sample numbers and the number of replicates used for statistical

143 testing is reported in corresponding figure legends. Unless otherwise stated: ***p<0.001, **p<0.01, *p<0.05.

144

5 Chapter V: Conclusion and Future Directions

145

5.1 The functions of CEP85 and STIL in PLK4 activation and centriole assembly

Centrioles play essential roles in faithful cell division and the formation of cilia and flagella. A better understanding of the centriole duplication process is of utmost importance, since failures in this process can result in a wide range of disorders, such as cancer, developmental disorders and ciliopathies. Towards this, the primary focus of my PhD research is to characterize the molecular mechanism underlying the centriole number control. At the outset of my project, CEP85 is a resident centriole protein without known roles in centriole duplication. The first functional link of CEP85 in centrosome biology was published in 2015, on which CEP85 was shown to regulate mitotic centrosome separation through antagonizing Nek2A activity (Chen et al., 2015). However, our proteomic analysis using BioID reveals that CEP85 displays a prominent proximity signature with a number of centriole duplication factors such as STIL, CEP192, CEP152, SASS6 and CEP63 (Gupta et al., 2015). In line with this notion, previous BioID study using purified centrosomes also identifies CEP85 as a potential interactor of PLK4 and CEP192 (Firat-Karalar et al., 2014). These multiple lines of evidence motivate me to investigate the function of CEP85 function in centriole assembly. In collaboration with the laboratory of Dr. Mark van Breugel‟s lab at the MRC in Cambridge UK, we identify CEP85 as a novel interactor of STIL through a highly conserved interaction interface involving a previously uncharacterized domain of STIL. Extensive structure-guided functional studies demonstrate that this interaction is indispensable for efficient recruitment of STIL at the earliest stage of the centriole duplication. Most importantly, this, in turn, is crucial to ensure robust local PLK4 activation to drive the subsequent downstream events of centriole assembly. Thus, our findings elucidate the molecular basis behind a previously undescribed modulatory step during the most upstream events of centriole duplication.

Many questions remain as to how CEP85 works with STIL in the control of centriole assembly. Our structural data suggest that the CEP85-STIL complex showed a 2:2 complex between CEP85 cc4 and the STIL N-terminal domain. Moreover, our biophysical data indicate that the CEP85 cc4 needs to be dimeric to bind to the STIL N-terminal domain. In this context, it is possible that CEP85 may form a homodimer to promote its interaction with STIL and further control centriole assembly, which will be an important avenue of further investigation. Besides, future work also needs to test whether this dimerization determines the molecular and functional

146 properties of CEP85, such as the protein stability, centriolar localization, its interaction with STIL and the control of centriole duplication. In addition, further structural work, possibly with their full-length proteins or larger complexes such as PLK4 that directly interacts with STIL, may be able to address the question of the exact binding stoichiometry and help to explain the role of this complex in PLK4 activation and centriole duplication from a structural perspective.

The localization data places CEP85 to the proximal end of mother centrioles where it transiently associates with STIL at the onset of centriole duplication. However, the biochemistry data suggest that CEP85 depletion significantly reduces the total cellular levels of STIL. Thus, we currently do not understand whether the binding of CEP85 to STIL is to modulate the efficient local enrichment of STIL or to control the stabilization of STIL, thereby playing a role in centriole duplication. It has been shown that preventing PLK4-mediated STIL phosphorylation can increase the mobile fraction of centriolar STIL to negatively affect centriole assembly (Moyer et al., 2015). On this basis, it is important to further determine whether the binding of CEP85 to STIL is also required for the stable interaction of STIL with the centrioles. Future work may address this question by performing fluorescence recovery after photobleaching (FRAP) assays in cells expressing GFP-STIL transgenes. The live-cell Förster resonance energy transfer (FRET) imaging analysis can also be used to determine the dynamic associations of this weak/transient complex. Moreover, STIL regulation during centriole duplication might be subject to additional layers of complexities, including its association with APC/C E3 ubiquitin ligase and USP9X deubiquitin ligase to regulate its protein stability in a cell cycle-dependent manner (Arquint and Nigg, 2014, Kodani et al., 2019). It is of great interest to test whether CEP85 can associate with APC/C or USP9X to regulate the protein degradation of STIL. In addition, deregulation of the human STIL gene expression has been associated with primary microcephaly, although the pathological mechanism is not fully understood. In order to explore the physiological relevance of the CEP85-STIL complex, future work should center on characterizing the functions of their interaction in brain development using either mouse models or in vitro 3D brain organoids, thereby determining whether abnormities in their interaction could contribute to disease development. Overall, given the importance of centrioles in human health and disease, the identification of CEP85 as a novel centriole duplication factor and the elucidation, at the atomic level, of its function in regulating STIL localization and PLK4 kinase

147 activity can represent significant advances in centrosome biology and also pave the way for new therapeutic approaches for centrosome-associated human disorders.

5.2 A role for the CEP85-STIL complex in PLK4-driven directed cell migration

Metastasis is of paramount importance in the prognosis of cancer patients. Malignant cancer cells acquire the motile ability to invade the surrounding tissues and penetrate the lymphatic or vascular circulation to produce the secondary tumors, which is the major cause of mortality of most cancer patients. In the past decades, significant advances have been made to our understanding of the molecular and cellular basis of cancer cell motility. Centrosomes have long been appreciated for their role in the control of directional cell migration in a microtubule- and MTOC-dependent manner. PLK4, a key centriole duplication factor, has recently been proposed as a prime target for cancer treatment (Gönczy, 2015). High expression of PLK4 has been reported in a number of common human epithelial malignancies associated with a low survival rate. Mechanically, previous work reveals an unexpected oncogenic activity of PLK4 in regulating ARP2 phosphorylation and actin organization to promote directional migration and invasion in cancer cell lines (Kazazian et al., 2017). In addition, STIL has been found upregulated in multiple cancers of poor prognosis, including lung cancer, colon carcinoma, prostate adenocarcinoma, and ovarian cancers (Patwardhan et al., 2018). Molecularly, STIL over-expression is linked to supernumerary centrosomes and a high histopathological mitotic index in tumors (Patwardhan et al., 2018), although the role of STIL in this regulation has not been reached. Our previous work indicates that CEP85 directly interacts with STIL to regulate its localization to centrioles. This allows precise spatiotemporal control of PLK4 kinase activity to promote efficient centriole assembly (Liu et al., 2018). Collectively, these multiple lines of evidence point to the notion that the CEP85-STIL complex might play a role in the control of cell motility.

Here, my results unravel an atypical role of the centrosomal CEP85-STIL module in regulating cancer cell motility. Using structural information of this complex, I find that disrupting the CEP85-STIL binding interface negatively affects directed cell migration. The middle region of CEP85 is responsible for its interaction with PLK4 in a STIL independent manner. Functionally,

148 this interaction is essential for the control of both cell migration and centriole duplication. Moreover, CEP85 and STIL are recruited to the cell cortex through PLK4. And this localization is also dependent on PLK4 kinase activity. Down-regulation of either CEP85 or STIL significantly reduces phosphorylation of ARP2 at the T237/T238 activation sites, thus impairing actin organization and cell motility. Together, these findings reveal the molecular basis behind a previously undescribed modulatory step in the directed cell migration.

However, we currently do not fully understand how and where CEP85-STIL complex coordinates with PLK4 to modulate ARP2/3 dependent actin reorganization, and how this discrete module balances its centrosomal versus non-centrosomal regulation. It also remains elusive whether the cell cortex pools of the CEP85-STIL complex account for the control of cell motility. An additional complication is the transient nature and low stability of this module, which makes it notoriously hard to characterize their biological functions. In this regard, the FRET-FLIM analysis can not only help us to visualize and quantify how dynamic or transient the CEP85 and STIL interaction is, but also determine their spatial-temporal associations with PLK4 at the cell cortex and centrosomes, thereby improving the mechanistic understanding of this module in the control of cell migration.

In addition, our lab‟s recent publication makes an important conceptual advance in defining the atypical function of a CEP192-PLK4/AURKB module at the cell cortex to modulate the exosome-WNT signalling stimulated non-directional cell motility (Luo et al., 2019), which is requires neither centrosomes nor the microtubule network. Although my results suggest that CEP85 directly interacts with CEP192, this interaction is not essential for CEP85‟s ability in directed cell migration. On this basis, future studies should assess whether CEP85 works with CEP192 to mediate the non-directional cell motility, and whether the CEP85-STIL complex also plays a role in PLK4 activation in this model. Given the importance of the elevated level of PLK4 described in more aggressive cancers, the CEP85-STIL binding interface might be a potential therapeutic option for tumor cells with PLK4 over-expression. Therefore, it is necessary for us to leverage the mouse xenograft model to determine whether the CEP85-STIL complex contributes to cancer invasions and metastases in vivo, which may help to identify alternative targets for cancer treatment.

149

5.3 The function of ANK2 in centriole assembly

This data chapter addresses the molecular mechanisms of centriole length control. Understanding these mechanisms is of utmost importance considering the essential role of centrioles for faithful cell division and the formation of cilia and flagella. Over-elongated centrioles are found to trigger centriole amplification through both centriole fragmentation and ectopic procentriole formation, leading to defective chromosome segregation. This process has been proposed as an additional centrosomal mechanism that contributes to cancer development. In our previous studies, we elucidate the crucial aspects of centriole elongation by delineating its core assembly pathway (CEP120-SPICE-CPAP) as well as identifying novel regulators through proteomic analysis and high confidence RNAi screens (Lin et al., 2013b, Gupta et al., 2015). Based on these studies, I combine the protein proximity mapping, RNAi, super-resolution imaging as well as biochemical methods to identify ANK2 as a novel interactor of CEP120, which is required for centriole assembly. Molecularly, I show that ANK2 is able to localize at centrosomes through its N-terminal membrane-binding domain. Further analysis reveals that the depletion of ANK2 significantly reduces the recruitment of CEP120 to centrioles and at the same time leads to the redistribution of CEP120 along the cytoplasmic microtubules. Interestingly, I also observe that ANK2 plays a critical role in the control of microtubule dynamics and consequentially affects the distribution of centriolar satellites. Overall, this study provides novel insights into the role of ANK2 in both centriole duplication and microtubule dynamics, and also improves our understanding of the CEP120-driven centriole length control. However, the role of ANK2 is complex, and there is no clear consensus that could synthesize the multiple functions ascribed to this protein into a unifying hypothesis to explain how ANK2 modulates centriole assembly and microtubule dynamics in space and time. Future work should focus on the mechanistic understanding of the complex phenotype of ANK2 deficiency or overexpression in CEP120 localization, microtubule stability and centriolar satellite distribution. In this context, it will be of particular interest to perform the BioID analysis of ANK2 and further generate a protein topology network of ANK2, CEP120, CPAP, SPICE1 and CEP135. Systematic profiling of their proximity interactions combined with functional analyses will help to discover novel proteins involved in the ANK2-driven centriole assembly and also yield insight into the mechanisms by which ANK2 functions in this process. Additionally, in Figure 4.1A-B, we have identified a set of microtubule-associated proteins, including NAP1L1, CSPP1 and MAP7 that also play

150 important roles in regulating both CEP120 localization and PLK4 induced centriole over- duplication. These proteins should be further investigated to assess their requirement for centriole duplication in non-PLK4 overexpression systems, to validate their interaction with CEP120, SPICE CPAP, CEP135 or ANK2, and to determine their functions in the regulation of microtubule dynamics. Overall, this work will be seen as a valuable basis for the centrosome community by stimulating numerous further studies to dissect the molecular underpinnings of centriole length control and related human diseases.

151

6 References

ANDERSEN, J. S., WILKINSON, C. J., MAYOR, T., MORTENSEN, P., NIGG, E. A. & MANN, M. 2003. Proteomic characterization of the human centrosome by protein correlation profiling. Nature, 426, 570-574.

APLAN, P., LOMBARDI, D., GINSBERG, A., COSSMAN, J., BERTNESS, V. & KIRSCH, I. 1990. Disruption of the human SCL locus by "illegitimate" V-(D)-J recombinase activity. Science, 250, 1426-1429.

ARQUINT, C., GABRYJONCZYK, A. M. & NIGG, E. A. 2014. Centrosomes as signalling centres. Philos Trans R Soc Lond B Biol Sci, 369.

ARQUINT, C. & NIGG, ERICH A. 2014. STIL Microcephaly Mutations Interfere with APC/C- Mediated Degradation and Cause Centriole Amplification. Current Biology, 24, 351-360.

ARQUINT, C. & NIGG, E. A. 2016. The PLK4-STIL-SAS-6 module at the core of centriole duplication. Biochemical Society transactions, 44, 1253-1263.

ARQUINT, C., SONNEN, K. F., STIERHOF, Y.-D. & NIGG, E. A. 2012. Cell-cycle-regulated expression of STIL controls centriole number in human cells. Journal of Cell Science, 125, 1342-1352.

ASTHANA, J., KAPOOR, S., MOHAN, R. & PANDA, D. 2013. Inhibition of HDAC6 deacetylase activity increases its binding with microtubules and suppresses microtubule dynamic instability in MCF-7 cells. The Journal of biological chemistry, 288, 22516- 22526.

AZIMZADEH, J. & MARSHALL, W. F. 2010. Building the centriole. Current Biology 20, R816-25.

BALCZON, R., BAO, L. & ZIMMER, W. 1994. PCM-1, A 228-kD centrosome autoantigen with a distinct cell cycle distribution. The Journal of Cell Biology, 124, 783-793.

BÄRENZ, F., INOUE, D., YOKOYAMA, H., TEGHA-DUNGHU, J., FREISS, S., DRAEGER, S., MAYILO, D., CADO, I., MERKER, S., KLINGER, M., HOECKENDORF, B., PILZ, S., HUPFELD, K., STEINBEISSER, H., LORENZ, H., RUPPERT, T., WITTBRODT, J. & GRUSS, O. J. 2013. The centriolar satellite protein SSX2IP promotes centrosome maturation. The Journal of Cell Biology, 202, 81-95.

BASHIR, T. & PAGANO, M. 2005. Cdk1: the dominant sibling of Cdk2. Nature Cell Biology, 7, 779-781.

BAUER, M., CUBIZOLLES, F., SCHMIDT, A. & NIGG, E. A. 2016. Quantitative analysis of human centrosome architecture by targeted proteomics and fluorescence imaging. The EMBO Journal, 35, 2152-2166.

152

BENNETT, V. & CHEN, L. 2001. Ankyrins and cellular targeting of diverse membrane proteins to physiological sites. Current Opinion in Cell Biology, 13, 61-67.

BETLEJA, E., NANJUNDAPPA, R., CHENG, T. & MAHJOUB, M. R. 2018. A novel Cep120- dependent mechanism inhibits centriole maturation in quiescent cells. eLife, 7, e35439.

BETTENCOURT-DIAS, M., HILDEBRANDT, F., PELLMAN, D., WOODS, G. & GODINHO, S. A. 2011. Centrosomes and cilia in human disease. Trends in genetics : TIG, 27, 307- 315.

BOEYNAEMS, S., ALBERTI, S., FAWZI, N. L., MITTAG, T., POLYMENIDOU, M., ROUSSEAU, F., SCHYMKOWITZ, J., SHORTER, J., WOLOZIN, B., VAN DEN BOSCH, L., TOMPA, P. & FUXREITER, M. 2018. Protein Phase Separation: A New Phase in Cell Biology. Trends in Cell Biology, 28, 420-435.

BURY, L., COELHO, P. A., SIMEONE, A., FERRIES, S., EYERS, C. E., EYERS, P. A., ZERNICKA-GOETZ, M. & GLOVER, D. M. 2017. Plk4 and Aurora A cooperate in the initiation of acentriolar spindle assembly in mammalian oocytes. The Journal of Cell Biology, 216, 3571-3590.

CARVALHO-SANTOS, Z., AZIMZADEH, J., PEREIRA-LEAL, J. B. & BETTENCOURT- DIAS, M. 2011. Tracing the origins of centrioles, cilia, and flagella. The Journal of Cell Biology, 194, 165-175.

CAVALIER-SMITH, T. 1974. Basal Body and Flagellar Development During the Vegetative Cell Cycle and the Sexual Cycle of Chlamydomonas Reinhardii. Journal of Cell Science, 16, 529-556.

CHEN, C., TIAN, F., LU, L., WANG, Y., XIAO, Z., YU, C. & YU, X. 2015. Characterization of Cep85 - a new antagonist of Nek2A that is involved in the regulation of centrosome disjunction. Journal of cell science, 128, 3290-3303.

COELHO, P. A., BURY, L., SHAHBAZI, M. N., LIAKATH-ALI, K., TATE, P. H., WORMALD, S., HINDLEY, C. J., HUCH, M., ARCHER, J., SKARNES, W. C., ZERNICKA-GOETZ, M. & GLOVER, D. M. 2015. Over-expression of Plk4 induces centrosome amplification, loss of primary cilia and associated tissue hyperplasia in the mouse. Open biology, 5, 150209-150209.

COELHO, P. A., BURY, L., SHARIF, B., RIPARBELLI, M. G., FU, J., CALLAINI, G., GLOVER, D. M. & ZERNICKA-GOETZ, M. 2013. Spindle formation in the mouse embryo requires Plk4 in the absence of centrioles. Developmental cell, 27, 586-597.

COMARTIN, D., GUPTA, GAGAN D., FUSSNER, E., COYAUD, É., HASEGAN, M., ARCHINTI, M., CHEUNG, SALLY W. T., PINCHEV, D., LAWO, S., RAUGHT, B., BAZETT-JONES, D. P., LÜDERS, J. & PELLETIER, L. 2013. CEP120 and SPICE1 Cooperate with CPAP in Centriole Elongation. Current Biology, 23, 1360-1366.

COTTEE, M. A., MUSCHALIK, N., WONG, Y. L., JOHNSON, C. M., JOHNSON, S., ANDREEVA, A., OEGEMA, K., LEA, S. M., RAFF, J. W. & VAN BREUGEL, M.

153

2013. Crystal structures of the CPAP/STIL complex reveal its role in centriole assembly and human microcephaly. eLife, 2, e01071.

DAMMERMANN, A. & MERDES, A. 2002. Assembly of centrosomal proteins and microtubule organization depends on PCM-1. The Journal of cell biology, 159, 255-266.

DELAVAL, B. & DOXSEY, S. J. 2010. Pericentrin in cellular function and disease. The Journal of cell biology, 188, 181-190.

DENU, R. A., SASS, M. M., JOHNSON, J. M., POTTS, G. K., CHOUDHARY, A., COON, J. J. & BURKARD, M. E. 2019. Polo-like kinase 4 maintains centriolar satellite integrity by phosphorylation of centrosomal protein 131 (CEP131). Journal of Biological Chemistry.

DIPPELL, R. V. 1968. The development of basal bodies in paramecium. Proceedings of the National Academy of Sciences of the United States of America, 61, 461-468.

DUENSING, A., LIU, Y., TSENG, M., MALUMBRES, M., BARBACID, M. & DUENSING, S. 2006. Cyclin-dependent kinase 2 is dispensable for normal centrosome duplication but required for oncogene-induced centrosome overduplication. Oncogene, 25, 2943-2949.

DUMONT, J. & DESAI, A. 2012. Acentrosomal spindle assembly and chromosome segregation during oocyte meiosis. Trends in cell biology, 22, 241-249.

DZHINDZHEV, N. S., TZOLOVSKY, G., LIPINSZKI, Z., ABDELAZIZ, M., DEBSKI, J., DADLEZ, M. & GLOVER, D. M. 2017. Two-step phosphorylation of Ana2 by Plk4 is required for the sequential loading of Ana2 and Sas6 to initiate procentriole formation. Open biology, 7, 170247.

DZHINDZHEV, N. S., YU, Q. D., WEISKOPF, K., TZOLOVSKY, G., CUNHA-FERREIRA, I., RIPARBELLI, M., RODRIGUES-MARTINS, A., BETTENCOURT-DIAS, M., CALLAINI, G. & GLOVER, D. M. 2010. Asterless is a scaffold for the onset of centriole assembly. Nature, 467, 714.

EL REFAEY, M. M. & MOHLER, P. J. 2017. Ankyrins and Spectrins in Cardiovascular Biology and Disease. Frontiers in physiology, 8, 852-852.

FARINA, F., GAILLARD, J., GUÉRIN, C., COUTÉ, Y., SILLIBOURNE, J., BLANCHOIN, L. & THÉRY, M. 2016. The centrosome is an actin-organizing centre. Nature cell biology, 18, 65-75.

FAVA, L. L., SCHULER, F., SLADKY, V., HASCHKA, M. D., SORATROI, C., EITERER, L., DEMETZ, E., WEISS, G., GELEY, S., NIGG, E. A. & VILLUNGER, A. 2017. The PIDDosome activates p53 in response to supernumerary centrosomes. Genes & development, 31, 34-45.

FENG, Z., CABALLE, A., WAINMAN, A., JOHNSON, S., HAENSELE, A. F. M., COTTEE, M. A., CONDUIT, P. T., LEA, S. M. & RAFF, J. W. 2017. Structural Basis for Mitotic Centrosome Assembly in Flies. Cell, 169, 1078-1089.e13.

154

FIRAT-KARALAR, ELIF N., RAUNIYAR, N., YATES, JOHN R., III & STEARNS, T. 2014. Proximity Interactions among Centrosome Components Identify Regulators of Centriole Duplication. Current Biology, 24, 664-670.

FıRAT-KARALAR, E. N. & STEARNS, T. 2014. The centriole duplication cycle. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130460.

FONG, C. S., MAZO, G., DAS, T., GOODMAN, J., KIM, M., O'ROURKE, B. P., IZQUIERDO, D. & TSOU, M.-F. B. 2016. 53BP1 and USP28 mediate p53-dependent cell cycle arrest in response to centrosome loss and prolonged mitosis. eLife, 5, e16270.

FU, J., LIPINSZKI, Z., RANGONE, H., MIN, M., MYKURA, C., CHAO-CHU, J., SCHNEIDER, S., DZHINDZHEV, N. S., GOTTARDO, M., RIPARBELLI, M. G., CALLAINI, G. & GLOVER, D. M. 2015. Conserved molecular interactions in centriole- to-centrosome conversion. Nature Cell Biology, 18, 87.

GAI, M., BIANCHI, F. T., VAGNONI, C., VERNÌ, F., BONACCORSI, S., PASQUERO, S., BERTO, G. E., SGRÒ, F., CHIOTTO, A. M., ANNARATONE, L., SAPINO, A., BERGO, A., LANDSBERGER, N., BOND, J., HUTTNER, W. B. & DI CUNTO, F. 2016. ASPM and CITK regulate spindle orientation by affecting the dynamics of astral microtubules. EMBO reports, 17, 1396-1409.

GALLETTA, B. J., JACOBS, K. C., FAGERSTROM, C. J. & RUSAN, N. M. 2016. Asterless is required for centriole length control and sperm development. The Journal of Cell Biology, 213, 435-450.

GAMBAROTTO, D., PENNETIER, C., RYNIAWEC, J. M., BUSTER, D. W., GOGENDEAU, D., GOUPIL, A., NANO, M., SIMON, A., BLANC, D., RACINE, V., KIMATA, Y., ROGERS, G. C. & BASTO, R. 2019. Plk4 Regulates Centriole Asymmetry and Spindle Orientation in Neural Stem Cells. Developmental Cell, 50, 11-24.e10.

GHEIRATMAND, L., COYAUD, E., GUPTA, G. D., LAURENT, E. M., HASEGAN, M., PROSSER, S. L., GONÇALVES, J., RAUGHT, B. & PELLETIER, L. 2019. Spatial and proteomic profiling reveals centrosome-independent features of centriolar satellites. The EMBO journal, 38, e101109-e101109.

GIANSANTI, M. G., BUCCIARELLI, E., BONACCORSI, S. & GATTI, M. 2008. Drosophila SPD-2 Is an Essential Centriole Component Required for PCM Recruitment and Astral- Microtubule Nucleation. Current Biology, 18, 303-309.

GODINHO, S. A. & PELLMAN, D. 2014. Causes and consequences of centrosome abnormalities in cancer. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130467.

GODINHO, S. A., PICONE, R., BURUTE, M., DAGHER, R., SU, Y., LEUNG, C. T., POLYAK, K., BRUGGE, J. S., THÉRY, M. & PELLMAN, D. 2014. Oncogene-like induction of cellular invasion from centrosome amplification. Nature, 510, 167-171.

155

GONÇALVES, J. & PELLETIER, L. 2017. The Ciliary Transition Zone: Finding the Pieces and Assembling the Gate. Molecules and cells, 40, 243-253.

GÖNCZY, P. 2012. Towards a molecular architecture of centriole assembly. Nature Reviews Molecular Cell Biology, 13, 425.

GÖNCZY, P. 2015. Centrosomes and cancer: revisiting a long-standing relationship. Nature Reviews Cancer, 15, 639.

GÖNCZY, P. & HATZOPOULOS, G. N. 2019. Centriole assembly at a glance. Journal of Cell Science, 132, jcs228833.

GOPALAKRISHNAN, J., MENNELLA, V., BLACHON, S., ZHAI, B., SMITH, A. H., MEGRAW, T. L., NICASTRO, D., GYGI, S. P., AGARD, D. A. & AVIDOR-REISS, T. 2011. Sas-4 provides a scaffold for cytoplasmic complexes and tethers them in a centrosome. Nature Communications, 2, 359.

GREENAN, G. A., KESZTHELYI, B., VALE, R. D. & AGARD, D. A. 2018. Insights into centriole geometry revealed by cryotomography of doublet and triplet centrioles. eLife, 7, e36851.

GUDERIAN, G., WESTENDORF, J., ULDSCHMID, A. & NIGG, E. A. 2010. Plk4 trans- autophosphorylation regulates centriole number by controlling βTrCP-mediated degradation. Journal of Cell Science, 123, 2163-2169.

GUICHARD, P., DESFOSSES, A., MAHESHWARI, A., HACHET, V., DIETRICH, C., BRUNE, A., ISHIKAWA, T., SACHSE, C. & GÖNCZY, P. 2012. Cartwheel Architecture of Trichonympha Basal Body. Science, 337, 553-553.

GUPTA, GAGAN D., COYAUD, É., GONÇALVES, J., MOJARAD, BAHAREH A., LIU, Y., WU, Q., GHEIRATMAND, L., COMARTIN, D., TKACH, JOHNNY M., CHEUNG, SALLY W. T., BASHKUROV, M., HASEGAN, M., KNIGHT, JAMES D., LIN, Z.-Y., SCHUELER, M., HILDEBRANDT, F., MOFFAT, J., GINGRAS, A.-C., RAUGHT, B. & PELLETIER, L. 2015. A Dynamic Protein Interaction Landscape of the Human Centrosome-Cilium Interface. Cell, 163, 1484-1499.

HATCH, E. M., KULUKIAN, A., HOLLAND, A. J., CLEVELAND, D. W. & STEARNS, T. 2010. Cep152 interacts with Plk4 and is required for centriole duplication. The Journal of Cell Biology, 191, 721-729.

HERRERA, S. C. & BACH, E. A. 2019. JAK/STAT signaling in stem cells and regeneration: from Drosophila to vertebrates. Development, 146, dev167643.

HINCHCLIFFE, E. H., LI, C., THOMPSON, E. A., MALLER, J. L. & SLUDER, G. 1999. Requirement of Cdk2-Cyclin E Activity for Repeated Centrosome Reproduction in Xenopus Egg Extracts. Science, 283, 851-854.

156

HORI, A., BARNOUIN, K., SNIJDERS, A. P. & TODA, T. 2016. A non-canonical function of Plk4 in centriolar satellite integrity and ciliogenesis through PCM1 phosphorylation. EMBO reports, 17, 326-337.

HORI, A. & TODA, T. 2017a. Regulation of centriolar satellite integrity and its physiology. Cellular and Molecular Life Sciences, 74, 213-229.

HORI, A. & TODA, T. 2017b. Regulation of centriolar satellite integrity and its physiology. Cellular and molecular life sciences : CMLS, 74, 213-229.

HUBBERT, C., GUARDIOLA, A., SHAO, R., KAWAGUCHI, Y., ITO, A., NIXON, A., YOSHIDA, M., WANG, X.-F. & YAO, T.-P. 2002. HDAC6 is a microtubule-associated deacetylase. Nature, 417, 455-458.

HUDSON, J. W., KOZAROVA, A., CHEUNG, P., MACMILLAN, J. C., SWALLOW, C. J., CROSS, J. C. & DENNIS, J. W. 2001. Late mitotic failure in mice lacking Sak, a polo- like kinase. Current Biology, 11, 441-446.

HUSSAIN, MUHAMMAD S., BAIG, SHAHID M., NEUMANN, S., NÜRNBERG, G., FAROOQ, M., AHMAD, I., ALEF, T., HENNIES, HANS C., TECHNAU, M., ALTMÜLLER, J., FROMMOLT, P., THIELE, H., NOEGEL, ANGELIKA A. & NÜRNBERG, P. 2012. A Truncating Mutation of CEP135 Causes Primary Microcephaly and Disturbed Centrosomal Function. The American Journal of Human Genetics, 90, 871-878.

INSOLERA, R., BAZZI, H., SHAO, W., ANDERSON, K. V. & SHI, S.-H. 2014. Cortical neurogenesis in the absence of centrioles. Nature Neuroscience, 17, 1528.

IZQUIERDO, D., WANG, W.-J., URYU, K. & TSOU, M.-FU B. 2014. Stabilization of Cartwheel-less Centrioles for Duplication Requires CEP295-Mediated Centriole-to- Centrosome Conversion. Cell Reports, 8, 957-965.

IZRAELI, S., LOWE, L. A., BERTNESS, V. L., GOOD, D. J., DORWARD, D. W., KIRSCH, I. R. & KUEHN, M. R. 1999. The SIL gene is required for mouse embryonic axial development and left–right specification. Nature, 399, 691-694.

JAKOBSEN, L., VANSELOW, K., SKOGS, M., TOYODA, Y., LUNDBERG, E., POSER, I., FALKENBY, L. G., BENNETZEN, M., WESTENDORF, J., NIGG, E. A., UHLEN, M., HYMAN, A. A. & ANDERSEN, J. S. 2011. Novel asymmetrically localizing components of human centrosomes identified by complementary proteomics methods. The EMBO journal, 30, 1520-1535.

JOACHIM, J., RAZI, M., JUDITH, D., WIRTH, M., CALAMITA, E., ENCHEVA, V., DYNLACHT, B. D., SNIJDERS, A. P., O'REILLY, N., JEFFERIES, H. B. J. & TOOZE, S. A. 2017. Centriolar Satellites Control GABARAP Ubiquitination and GABARAP- Mediated Autophagy. Current biology : CB, 27, 2123-2136.e7.

JOSEPH, N., AL-JASSAR, C., JOHNSON, C. M., ANDREEVA, A., BARNABAS, D. D., FREUND, S. M. V., GERGELY, F. & VAN BREUGEL, M. 2018. Disease-Associated

157

Mutations in CEP120 Destabilize the Protein and Impair Ciliogenesis. Cell Reports, 23, 2805-2818.

JOUKOV, V., WALTER, JOHANNES C. & DE NICOLO, A. 2014. The Cep192-Organized Aurora A-Plk1 Cascade Is Essential for Centrosome Cycle and Bipolar Spindle Assembly. Molecular Cell, 55, 578-591.

KAZAZIAN, K., GO, C., WU, H., BRASHAVITSKAYA, O., XU, R., DENNIS, J. W., GINGRAS, A.-C. & SWALLOW, C. J. 2017. Plk4 Promotes Cancer Invasion and Metastasis through Arp2/3 Complex Regulation of the Actin Cytoskeleton. Cancer Research, 77, 434-447.

KIM, J., LEE, K. & RHEE, K. 2015. PLK1 regulation of PCNT cleavage ensures fidelity of centriole separation during mitotic exit. Nature communications, 6, 10076-10076.

KIM, T.-S., PARK, J.-E., SHUKLA, A., CHOI, S., MURUGAN, R. N., LEE, J. H., AHN, M., RHEE, K., BANG, J. K., KIM, B. Y., LONCAREK, J., ERIKSON, R. L. & LEE, K. S. 2013. Hierarchical recruitment of Plk4 and regulation of centriole biogenesis by two centrosomal scaffolds, Cep192 and Cep152. Proceedings of the National Academy of Sciences, 110, E4849-E4857.

KLEBBA, J. E., BUSTER, D. W., MCLAMARRAH, T. A., RUSAN, N. M. & ROGERS, G. C. 2015. Autoinhibition and relief mechanism for Polo-like kinase 4. Proceedings of the National Academy of Sciences, 112, E657-E666.

KLEYLEIN-SOHN, J., WESTENDORF, J., LE CLECH, M., HABEDANCK, R., STIERHOF, Y.-D. & NIGG, E. A. 2007. Plk4-Induced Centriole Biogenesis in Human Cells. Developmental Cell, 13, 190-202.

KO, M. A., ROSARIO, C. O., HUDSON, J. W., KULKARNI, S., POLLETT, A., DENNIS, J. W. & SWALLOW, C. J. 2005. Plk4 haploinsufficiency causes mitotic infidelity and carcinogenesis. Nature Genetics, 37, 883-888.

KODANI, A., MOYER, T., CHEN, A., HOLLAND, A., WALSH, C. A. & REITER, J. F. 2019. SFI1 promotes centriole duplication by recruiting USP9X to stabilize the microcephaly protein STIL. The Journal of Cell Biology, 218, 2185-2197.

KRAATZ, S., GUICHARD, P., OBBINENI, JAGAN M., OLIERIC, N., HATZOPOULOS, GEORGIOS N., HILBERT, M., SEN, I., MISSIMER, J., GÖNCZY, P. & STEINMETZ, MICHEL O. 2016. The Human Centriolar Protein CEP135 Contains a Two-Stranded Coiled-Coil Domain Critical for Microtubule Binding. Structure, 24, 1358-1371.

KURIYAMA, R. & BORISY, G. G. 1983. Cytasters induced within unfertilized sea-urchin eggs. Journal of Cell Science, 61, 175-189.

LAMBRUS, B. G., DAGGUBATI, V., UETAKE, Y., SCOTT, P. M., CLUTARIO, K. M., SLUDER, G. & HOLLAND, A. J. 2016. A USP28–53BP1–p53–p21 signaling axis arrests growth after centrosome loss or prolonged mitosis. The Journal of Cell Biology, 214, 143-153.

158

LAMBRUS, B. G. & HOLLAND, A. J. 2017. A New Mode of Mitotic Surveillance. Trends in Cell Biology, 27, 314-321.

LANCASTER, M. A. & KNOBLICH, J. A. 2012. Spindle orientation in mammalian cerebral cortical development. Current opinion in neurobiology, 22, 737-746.

LAWO, S., HASEGAN, M., GUPTA, G. D. & PELLETIER, L. 2012. Subdiffraction imaging of centrosomes reveals higher-order organizational features of pericentriolar material. Nature Cell Biology, 14, 1148.

LEBER, B., MAIER, B., FUCHS, F., CHI, J., RIFFEL, P., ANDERHUB, S., WAGNER, L., HO, A. D., SALISBURY, J. L., BOUTROS, M. & KRÄMER, A. 2010. Proteins Required for Centrosome Clustering in Cancer Cells. Science Translational Medicine, 2, 33ra38-33ra38.

LECLAIRE, L. L., 3RD, BAUMGARTNER, M., IWASA, J. H., MULLINS, R. D. & BARBER, D. L. 2008. Phosphorylation of the Arp2/3 complex is necessary to nucleate actin filaments. The Journal of cell biology, 182, 647-654.

LECLAIRE, L. L., RANA, M., BAUMGARTNER, M. & BARBER, D. L. 2015. The Nck- interacting kinase NIK increases Arp2/3 complex activity by phosphorylating the Arp2 subunit. The Journal of Cell Biology, 208, 161-170.

LEDA, M., HOLLAND, A. J. & GORYACHEV, A. B. 2018. Autoamplification and Competition Drive Symmetry Breaking: Initiation of Centriole Duplication by the PLK4- STIL Network. iScience, 8, 222-235.

LERIT, D. A., JORDAN, H. A., POULTON, J. S., FAGERSTROM, C. J., GALLETTA, B. J., PEIFER, M. & RUSAN, N. M. 2015. Interphase centrosome organization by the PLP- Cnn scaffold is required for centrosome function. The Journal of Cell Biology, 210, 79- 97.

LETTMAN, MOLLY M., WONG, YAO L., VISCARDI, V., NIESSEN, S., CHEN, S.-H., SHIAU, ANDREW K., ZHOU, H., DESAI, A. & OEGEMA, K. 2013. Direct Binding of SAS-6 to ZYG-1 Recruits SAS-6 to the Mother Centriole for Cartwheel Assembly. Developmental Cell, 25, 284-298.

LEVINE, M. S., BAKKER, B., BOECKX, B., MOYETT, J., LU, J., VITRE, B., SPIERINGS, D. C., LANSDORP, P. M., CLEVELAND, D. W., LAMBRECHTS, D., FOIJER, F. & HOLLAND, A. J. 2017. Centrosome Amplification Is Sufficient to Promote Spontaneous Tumorigenesis in Mammals. Developmental cell, 40, 313-322.e5.

LIN, Y.-C., CHANG, C.-W., HSU, W.-B., TANG, C.-J. C., LIN, Y.-N., CHOU, E.-J., WU, C.- T. & TANG, T. K. 2013a. Human microcephaly protein CEP135 binds to hSAS-6 and CPAP, and is required for centriole assembly. The EMBO Journal, 32, 1141-1154.

LIN, Y.-N., WU, C.-T., LIN, Y.-C., HSU, W.-B., TANG, C.-J. C., CHANG, C.-W. & TANG, T. K. 2013b. CEP120 interacts with CPAP and positively regulates centriole elongation. The Journal of Cell Biology, 202, 211-219.

159

LIU, J.-J. 2017. Regulation of dynein-dynactin-driven vesicular transport. Traffic, 18, 336-347.

LIU, Y., GUPTA, G. D., BARNABAS, D. D., AGIRCAN, F. G., MEHMOOD, S., WU, D., COYAUD, E., JOHNSON, C. M., MCLAUGHLIN, S. H., ANDREEVA, A., FREUND, S. M. V., ROBINSON, C. V., CHEUNG, S. W. T., RAUGHT, B., PELLETIER, L. & VAN BREUGEL, M. 2018. Direct binding of CEP85 to STIL ensures robust PLK4 activation and efficient centriole assembly. Nature Communications, 9, 1731.

LIZARRAGA, S. B., MARGOSSIAN, S. P., HARRIS, M. H., CAMPAGNA, D. R., HAN, A.- P., BLEVINS, S., MUDBHARY, R., BARKER, J. E., WALSH, C. A. & FLEMING, M. D. 2010. Cdk5rap2 regulates centrosome function and chromosome segregation in neuronal progenitors. Development (Cambridge, England), 137, 1907-1917.

LONCAREK, J. & BETTENCOURT-DIAS, M. 2018. Building the right centriole for each cell type. The Journal of Cell Biology, 217, 823-835.

LONCAREK, J., SLUDER, G. & KHODJAKOV, A. 2007. Centriole biogenesis: a tale of two pathways. Nature cell biology, 9, 736-738.

LOPES, C. A. M., PROSSER, S. L., ROMIO, L., HIRST, R. A., O'CALLAGHAN, C., WOOLF, A. S. & FRY, A. M. 2011. Centriolar satellites are assembly points for proteins implicated in human ciliopathies, including oral-facial-digital syndrome 1. Journal of cell science, 124, 600-612.

LORENZO, D. N., BADEA, A., DAVIS, J., HOSTETTLER, J., HE, J., ZHONG, G., ZHUANG, X. & BENNETT, V. 2014. A PIK3C3–Ankyrin-B–Dynactin pathway promotes axonal growth and multiorganelle transport. The Journal of Cell Biology, 207, 735-752.

LUO, Y., BARRIOS-RODILES, M., GUPTA, G. D., ZHANG, Y. Y., OGUNJIMI, A. A., BASHKUROV, M., TKACH, J. M., UNDERHILL, A. Q., ZHANG, L., BOURMOUM, M., WRANA, J. L. & PELLETIER, L. 2019. Atypical function of a centrosomal module in WNT signalling drives contextual cancer cell motility. Nature communications, 10, 2356-2356.

LUXTON, G. W. G. & GUNDERSEN, G. G. 2011. Orientation and function of the nuclear- centrosomal axis during cell migration. Current opinion in cell biology, 23, 579-588.

MAHJOUB, M. R., XIE, Z. & STEARNS, T. 2010. Cep120 is asymmetrically localized to the daughter centriole and is essential for centriole assembly. The Journal of cell biology, 191, 331-346.

MANISWAMI, R. R., PRASHANTH, S., KARANTH, A. V., KOUSHIK, S., GOVINDARAJ, H., MULLANGI, R., RAJAGOPAL, S. & JEGATHEESAN, S. K. 2018. PLK4: a link between centriole biogenesis and cancer. Expert Opinion on Therapeutic Targets, 22, 59- 73.

MARDIN, B. R. & SCHIEBEL, E. 2012. Breaking the ties that bind: New advances in centrosome biology. The Journal of Cell Biology, 197, 11-18.

160

MARIAPPAN, A., SONI, K., SCHORPP, K., ZHAO, F., MINAKAR, A., ZHENG, X., MANDAD, S., MACHELEIDT, I., RAMANI, A., KUBELKA, T., DAWIDOWSKI, M., GOLFMANN, K., WASON, A., YANG, C., SIMONS, J., SCHMALZ, H.-G., HYMAN, A. A., ANEJA, R., ULLRICH, R., URLAUB, H., ODENTHAL, M., BÜTTNER, R., LI, H., SATTLER, M., HADIAN, K. & GOPALAKRISHNAN, J. 2019. Inhibition of CPAP–tubulin interaction prevents proliferation of centrosome-amplified cancer cells. The EMBO Journal, 38, e99876.

MARTEIL, G., GUERRERO, A., VIEIRA, A. F., DE ALMEIDA, B. P., MACHADO, P., MENDONÇA, S., MESQUITA, M., VILLARREAL, B., FONSECA, I., FRANCIA, M. E., DORES, K., MARTINS, N. P., JANA, S. C., TRANFIELD, E. M., BARBOSA- MORAIS, N. L., PAREDES, J., PELLMAN, D., GODINHO, S. A. & BETTENCOURT- DIAS, M. 2018. Over-elongation of centrioles in cancer promotes centriole amplification and chromosome missegregation. Nature Communications, 9, 1258.

MARTHIENS, V., RUJANO, M. A., PENNETIER, C., TESSIER, S., PAUL-GILLOTEAUX, P. & BASTO, R. 2013. Centrosome amplification causes microcephaly. Nature Cell Biology, 15, 731.

MARTIN, C.-A., AHMAD, I., KLINGSEISEN, A., HUSSAIN, M. S., BICKNELL, L. S., LEITCH, A., NÜRNBERG, G., TOLIAT, M. R., MURRAY, J. E., HUNT, D., KHAN, F., ALI, Z., TINSCHERT, S., DING, J., KEITH, C., HARLEY, M. E., HEYN, P., MÜLLER, R., HOFFMANN, I., DAIRE, V. C., DOLLFUS, H., DUPUIS, L., BASHAMBOO, A., MCELREAVEY, K., KARIMINEJAD, A., MENDOZA- LONDONO, R., MOORE, A. T., SAGGAR, A., SCHLECHTER, C., WELEBER, R., THIELE, H., ALTMÜLLER, J., HÖHNE, W., HURLES, M. E., NOEGEL, A. A., BAIG, S. M., NÜRNBERG, P. & JACKSON, A. P. 2014. Mutations in PLK4, encoding a master regulator of centriole biogenesis, cause microcephaly, growth failure and retinopathy. Nature Genetics, 46, 1283.

MASON, JACQUELINE M., LIN, DAN C.-C., WEI, X., CHE, Y., YAO, Y., KIARASH, R., CESCON, DAVID W., FLETCHER, GRAHAM C., AWREY, DONALD E., BRAY, MARK R., PAN, G. & MAK, TAK W. 2014. Functional Characterization of CFI- 400945, a Polo-like Kinase 4 Inhibitor, as a Potential Anticancer Agent. Cancer Cell, 26, 163-176.

MAYOR, R. & ETIENNE-MANNEVILLE, S. 2016. The front and rear of collective cell migration. Nature Reviews Molecular Cell Biology, 17, 97.

MOYER, T. C., CLUTARIO, K. M., LAMBRUS, B. G., DAGGUBATI, V. & HOLLAND, A. J. 2015. Binding of STIL to Plk4 activates kinase activity to promote centriole assembly. The Journal of Cell Biology, 209, 863-878.

MOYER, T. C. & HOLLAND, A. J. 2019. PLK4 promotes centriole duplication by phosphorylating STIL to link the procentriole cartwheel to the microtubule wall. eLife, 8, e46054.

161

NIGG, E. A. & HOLLAND, A. J. 2018. Once and only once: mechanisms of centriole duplication and their deregulation in disease. Nature Reviews Molecular Cell Biology, 19, 297.

NIGG, E. A. & RAFF, J. W. 2009. Centrioles, Centrosomes, and Cilia in Health and Disease. Cell, 139, 663-678.

O‟ROURKE, B. P., GOMEZ-FERRERIA, M. A., BERK, R. H., HACKL, A. M. U., NICHOLAS, M. P., O‟ROURKE, S. C., PELLETIER, L. & SHARP, D. J. 2014. Cep192 Controls the Balance of Centrosome and Non-Centrosomal Microtubules during Interphase. PLOS ONE, 9, e101001.

OBINO, D., FARINA, F., MALBEC, O., SÁEZ, P. J., MAURIN, M., GAILLARD, J., DINGLI, F., LOEW, D., GAUTREAU, A., YUSEFF, M.-I., BLANCHOIN, L., THÉRY, M. & LENNON-DUMÉNIL, A.-M. 2016. Actin nucleation at the centrosome controls lymphocyte polarity. Nature communications, 7, 10969-10969.

ODABASI, E., GUL, S., KAVAKLI, I. H. & FIRAT-KARALAR, E. N. 2019. Centriolar satellites are required for efficient ciliogenesis and ciliary content regulation. EMBO reports, 20, e47723.

OHASHI, Y., SOLER, N., GARCÍA ORTEGÓN, M., ZHANG, L., KIRSTEN, M. L., PERISIC, O., MASSON, G. R., BURKE, J. E., JAKOBI, A. J., APOSTOLAKIS, A. A., JOHNSON, C. M., OHASHI, M., KTISTAKIS, N. T., SACHSE, C. & WILLIAMS, R. L. 2016. Characterization of Atg38 and NRBF2, a fifth subunit of the autophagic Vps34/PIK3C3 complex. Autophagy, 12, 2129-2144.

OHTA, M., ASHIKAWA, T., NOZAKI, Y., KOZUKA-HATA, H., GOTO, H., INAGAKI, M., OYAMA, M. & KITAGAWA, D. 2014. Direct interaction of Plk4 with STIL ensures formation of a single procentriole per parental centriole. Nature Communications, 5, 5267.

OHTA, M., WATANABE, K., ASHIKAWA, T., NOZAKI, Y., YOSHIBA, S., KIMURA, A. & KITAGAWA, D. 2018. Bimodal Binding of STIL to Plk4 Controls Proper Centriole Copy Number. Cell Reports, 23, 3160-3169.e4.

PARK, S.-Y., PARK, J.-E., KIM, T.-S., KIM, J. H., KWAK, M.-J., KU, B., TIAN, L., MURUGAN, R. N., AHN, M., KOMIYA, S., HOJO, H., KIM, N.-H., KIM, B. Y., BANG, J. K., ERIKSON, R. L., LEE, K. W., KIM, S. J., OH, B.-H., YANG, W. & LEE, K. S. 2014a. Molecular basis for unidirectional scaffold switching of human Plk4 in centriole biogenesis. Nature structural & molecular biology, 21, 696-703.

PARK, S.-Y., PARK, J.-E., KIM, T.-S., KIM, J. H., KWAK, M.-J., KU, B., TIAN, L., MURUGAN, R. N., AHN, M., KOMIYA, S., HOJO, H., KIM, N.-H., KIM, B. Y., BANG, J. K., ERIKSON, R. L., LEE, K. W., KIM, S. J., OH, B.-H., YANG, W. & LEE, K. S. 2014b. Molecular basis for unidirectional scaffold switching of human Plk4 in centriole biogenesis. Nature Structural &Amp; Molecular Biology, 21, 696.

162

PATWARDHAN, D., MANI, S., PASSEMARD, S., GRESSENS, P. & EL GHOUZZI, V. 2018. STIL balancing primary microcephaly and cancer. Cell death & disease, 9, 65-65.

PELLETIER, L., O‟TOOLE, E., SCHWAGER, A., HYMAN, A. A. & MÜLLER-REICHERT, T. 2006. Centriole assembly in Caenorhabditis elegans. Nature, 444, 619-623.

PELLETIER, L., ÖZLÜ, N., HANNAK, E., COWAN, C., HABERMANN, B., RUER, M., MÜLLER-REICHERT, T. & HYMAN, A. A. 2004. The Caenorhabditis elegans Centrosomal Protein SPD-2 Is Required for both Pericentriolar Material Recruitment and Centriole Duplication. Current Biology, 14, 863-873.

PFAFF, K. L., STRAUB, C. T., CHIANG, K., BEAR, D. M., ZHOU, Y. & ZON, L. I. 2007. The zebra fish cassiopeia mutant reveals that SIL is required for mitotic spindle organization. Molecular and cellular biology, 27, 5887-5897.

PROSSER, S. L. & PELLETIER, L. 2017. Mitotic spindle assembly in animal cells: a fine balancing act. Nature Reviews Molecular Cell Biology, 18, 187.

PUKLOWSKI, A., HOMSI, Y., KELLER, D., MAY, M., CHAUHAN, S., KOSSATZ, U., GRÜNWALD, V., KUBICKA, S., PICH, A., MANNS, M. P., HOFFMANN, I., GÖNCZY, P. & MALEK, N. P. 2011. The SCF–FBXW5 E3-ubiquitin ligase is regulated by PLK4 and targets HsSAS-6 to control centrosome duplication. Nature Cell Biology, 13, 1004.

QUARANTOTTI, V., CHEN, J.-X., TISCHER, J., GONZALEZ TEJEDO, C., PAPACHRISTOU, E. K., D'SANTOS, C. S., KILMARTIN, J. V., MILLER, M. L. & GERGELY, F. 2019. Centriolar satellites are acentriolar assemblies of centrosomal proteins. The EMBO Journal, 38, e101082.

RHYS, A. D., MONTEIRO, P., SMITH, C., VAGHELA, M., ARNANDIS, T., KATO, T., LEITINGER, B., SAHAI, E., MCAINSH, A., CHARRAS, G. & GODINHO, S. A. 2018. Loss of E-cadherin provides tolerance to centrosome amplification in epithelial cancer cells. The Journal of Cell Biology, 217, 195-209.

RIOS, R. M. 2014. The centrosome-Golgi apparatus nexus. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130462.

ROBBINS, E., JENTZSCH, G. & MICALI, A. 1968. THE CENTRIOLE CYCLE IN SYNCHRONIZED HELA CELLS. The Journal of Cell Biology, 36, 329-339.

ROOSING, S., ROMANI, M., ISRIE, M., ROSTI, R. O., MICALIZZI, A., MUSAEV, D., MAZZA, T., AL-GAZALI, L., ALTUNOGLU, U., BOLTSHAUSER, E., D'ARRIGO, S., DE KEERSMAECKER, B., KAYSERILI, H., BRANDENBERGER, S., KRAOUA, I., MARK, P. R., MCKANNA, T., VAN KEIRSBILCK, J., MOERMAN, P., PORETTI, A., PURI, R., VAN ESCH, H., GLEESON, J. G. & VALENTE, E. M. 2016. Mutations in CEP120 cause Joubert syndrome as well as complex ciliopathy phenotypes. Journal of Medical Genetics, 53, 608-615.

163

ROTTNER, K., HÄNISCH, J. & CAMPELLONE, K. G. 2010. WASH, WHAMM and JMY: regulation of Arp2/3 complex and beyond. Trends in Cell Biology, 20, 650-661.

ROUX, K. J., KIM, D. I. & BURKE, B. 2013. BioID: A Screen for Protein-Protein Interactions. Current Protocols in Protein Science, 74, 19.23.1-19.23.14.

ROUX, K. J., KIM, D. I., RAIDA, M. & BURKE, B. 2012. A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. The Journal of Cell Biology, 196, 801-810.

RÜTHNICK, D. & SCHIEBEL, E. 2016. Duplication of the Yeast Spindle Pole Body Once per Cell Cycle. Molecular and Cellular Biology, 36, 1324-1331.

SAADE, M., BLANCO-AMEIJEIRAS, J., GONZALEZ-GOBARTT, E. & MARTÍ, E. 2018. A centrosomal view of CNS growth. Development, 145, dev170613.

SCHEER, U. 2014. Historical roots of centrosome research: discovery of Boveri's microscope slides in Würzburg. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130469.

SERÇIN, Ö., LARSIMONT, J.-C., KARAMBELAS, A. E., MARTHIENS, V., MOERS, V., BOECKX, B., LE MERCIER, M., LAMBRECHTS, D., BASTO, R. & BLANPAIN, C. 2015. Transient PLK4 overexpression accelerates tumorigenesis in p53-deficient epidermis. Nature Cell Biology, 18, 100.

SILLIBOURNE, J. E. & BORNENS, M. 2010. Polo-like kinase 4: the odd one out of the family. Cell Division, 5, 25.

SLEVIN, L. K., NYE, J., PINKERTON, D. C., BUSTER, D. W., ROGERS, G. C. & SLEP, K. C. 2012. The structure of the plk4 cryptic polo box reveals two tandem polo boxes required for centriole duplication. Structure (London, England : 1993), 20, 1905-1917.

SO, C., SERES, K. B., STEYER, A. M., MÖNNICH, E., CLIFT, D., PEJKOVSKA, A., MÖBIUS, W. & SCHUH, M. 2019. A liquid-like spindle domain promotes acentrosomal spindle assembly in mammalian oocytes. Science, 364, eaat9557.

SONNEN, K. F., GABRYJONCZYK, A.-M., ANSELM, E., STIERHOF, Y.-D. & NIGG, E. A. 2013. Human Cep192 and Cep152 cooperate in Plk4 recruitment and centriole duplication. Journal of Cell Science, 126, 3223-3233.

STEPHAN, R., GOELLNER, B., MORENO, E., FRANK, C. A., HUGENSCHMIDT, T., GENOUD, C., ABERLE, H. & PIELAGE, J. 2015. Hierarchical microtubule organization controls axon caliber and transport and determines synaptic structure and stability. Developmental cell, 33, 5-21.

STINCHCOMBE, J. C. & GRIFFITHS, G. M. 2014. Communication, the centrosome and the immunological synapse. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130463.

164

STUELTEN, C. H., PARENT, C. A. & MONTELL, D. J. 2018. Cell motility in cancer invasion and metastasis: insights from simple model organisms. Nature Reviews Cancer, 18, 296.

SÜTTERLIN, C., HSU, P., MALLABIABARRENA, A. & MALHOTRA, V. 2002. Fragmentation and Dispersal of the Pericentriolar Golgi Complex Is Required for Entry into Mitosis in Mammalian Cells. Cell, 109, 359-369.

TANG, C.-J. C., FU, R.-H., WU, K.-S., HSU, W.-B. & TANG, T. K. 2009. CPAP is a cell-cycle regulated protein that controls centriole length. Nature Cell Biology, 11, 825.

TANG, Z., LIN, M. G., STOWE, T. R., CHEN, S., ZHU, M., STEARNS, T., FRANCO, B. & ZHONG, Q. 2013. Autophagy promotes primary ciliogenesis by removing OFD1 from centriolar satellites. Nature, 502, 254.

TSOU, M.-F. B., WANG, W.-J., GEORGE, K. A., URYU, K., STEARNS, T. & JALLEPALLI, P. V. 2009. Polo kinase and separase regulate the mitotic licensing of centriole duplication in human cells. Developmental cell, 17, 344-354.

TSUCHIYA, Y., YOSHIBA, S., GUPTA, A., WATANABE, K. & KITAGAWA, D. 2016. Cep295 is a conserved scaffold protein required for generation of a bona fide mother centriole. Nature Communications, 7, 12567.

VAN BREUGEL, M., WILCKEN, R., MCLAUGHLIN, S. H., RUTHERFORD, T. J. & JOHNSON, C. M. 2014. Structure of the SAS-6 cartwheel hub from Leishmania major. eLife, 3, e01812.

VITRE, B., HOLLAND, A. J., KULUKIAN, A., SHOSHANI, O., HIRAI, M., WANG, Y., MALDONADO, M., CHO, T., BOUBAKER, J., SWING, D. A., TESSAROLLO, L., EVANS, S. M., FUCHS, E. & CLEVELAND, D. W. 2015. Chronic centrosome amplification without tumorigenesis. Proceedings of the National Academy of Sciences, 112, E6321-E6330.

VOROBJEV, I. A. & CHENTSOV YU, S. 1982. Centrioles in the cell cycle. I. Epithelial cells. The Journal of cell biology, 93, 938-949.

WANG, J. T. & STEARNS, T. 2017. The ABCs of Centriole Architecture: The Form and Function of Triplet Microtubules. Cold Spring Harbor Symposia on Quantitative Biology, 82, 145-155.

WANG, W.-J., ACEHAN, D., KAO, C.-H., JANE, W.-N., URYU, K. & TSOU, M.-F. B. 2015. De novo centriole formation in human cells is error-prone and does not require SAS-6 self-assembly. eLife, 4, e10586.

WANG, X., TSAI, J.-W., IMAI, J. H., LIAN, W.-N., VALLEE, R. B. & SHI, S.-H. 2009. Asymmetric centrosome inheritance maintains neural progenitors in the neocortex. Nature, 461, 947-955.

WONG, Y. L., ANZOLA, J. V., DAVIS, R. L., YOON, M., MOTAMEDI, A., KROLL, A., SEO, C. P., HSIA, J. E., KIM, S. K., MITCHELL, J. W., MITCHELL, B. J., DESAI, A.,

165

GAHMAN, T. C., SHIAU, A. K. & OEGEMA, K. 2015. Reversible centriole depletion with an inhibitor of Polo-like kinase 4. Science, 348, 1155-1160.

WOODRUFF, J. B., FERREIRA GOMES, B., WIDLUND, P. O., MAHAMID, J., HONIGMANN, A. & HYMAN, A. A. 2017. The Centrosome Is a Selective Condensate that Nucleates Microtubules by Concentrating Tubulin. Cell, 169, 1066-1077.e10.

WOODRUFF, J. B., WUESEKE, O. & HYMAN, A. A. 2014. Pericentriolar material structure and dynamics. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 369, 20130459.

WOODS, C. G., BOND, J. & ENARD, W. 2005. Autosomal recessive primary microcephaly (MCPH): a review of clinical, molecular, and evolutionary findings. American journal of human genetics, 76, 717-728.

XIE, Z., MOY, L. Y., SANADA, K., ZHOU, Y., BUCHMAN, J. J. & TSAI, L.-H. 2007. Cep120 and TACCs Control Interkinetic Nuclear Migration and the Neural Progenitor Pool. Neuron, 56, 79-93.

YAMAMOTO, S. & KITAGAWA, D. 2019. Self-organization of Plk4 regulates symmetry breaking in centriole duplication. Nature Communications, 10, 1810.

YANAGIDA, M., MIYOSHI, R., TOYOKUNI, R., ZHU, Y. & MURAKAMI, F. 2012. Dynamics of the leading process, nucleus, and Golgi apparatus of migrating cortical interneurons in living mouse embryos. Proceedings of the National Academy of Sciences, 109, 16737-16742.

YU, F., RAN, J. & ZHOU, J. 2016. Ciliopathies: Does HDAC6 Represent a New Therapeutic Target? Trends in Pharmacological Sciences, 37, 114-119.

ZHAO, H., ZHU, L., ZHU, Y., CAO, J., LI, S., HUANG, Q., XU, T., HUANG, X., YAN, X. & ZHU, X. 2013. The Cep63 paralogue Deup1 enables massive de novo centriole biogenesis for vertebrate multiciliogenesis. Nature Cell Biology, 15, 1434.

ZHENG, X., RAMANI, A., SONI, K., GOTTARDO, M., ZHENG, S., MING GOOI, L., LI, W., FENG, S., MARIAPPAN, A., WASON, A., WIDLUND, P., POZNIAKOVSKY, A., POSER, I., DENG, H., OU, G., RIPARBELLI, M., GIULIANO, C., HYMAN, A. A., SATTLER, M., GOPALAKRISHNAN, J. & LI, H. 2016. Molecular basis for CPAP- tubulin interaction in controlling centriolar and ciliary length. Nature Communications, 7, 11874.

ZITOUNI, S., FRANCIA, MARIA E., LEAL, F., MONTENEGRO GOUVEIA, S., NABAIS, C., DUARTE, P., GILBERTO, S., BRITO, D., MOYER, T., KANDELS-LEWIS, S., OHTA, M., KITAGAWA, D., HOLLAND, ANDREW J., KARSENTI, E., LORCA, T., LINCE-FARIA, M. & BETTENCOURT-DIAS, M. 2016. CDK1 Prevents Unscheduled PLK4-STIL Complex Assembly in Centriole Biogenesis. Current Biology, 26, 1127- 1137.

166

7 Copyright Acknowledgements

Portions of the data described in this thesis are published in their original or modified form as a resource in Cell (2015) and a research article in Nature Communications (2018).

(1) Liu, Y, Gupta GD, Barnabas DD, Agircan FG, Mehmood S, Wu D, Coyaud E, Johnson CM, McLaughlin SH, Andreeva A, Freund SMV, Robinson CV, Cheung SWT, Raught B, Pelletier L, van Breugel M. (2018) Direct binding of CEP85 to STIL ensures robust PLK4 activation and efficient centriole assembly. Nature Communications. 9:1731. DOI: 10.1038/s41467-018-04122- x

(2) Gupta GD*, Coyaud É*, Gonçalves J*, Mojarad BA*, Liu Y, Wu Q, Gheiratmand L, Comartin D, Tkach JM, Cheung SW, Bashkurov M, Hasegan M, Knight JD, Lin ZY, Schueler M, Hildebrandt F, Moffat J, Gingras AC, Raught B, Pelletier L. (2015) A Dynamic Protein Interaction Landscape of the Human Centrosome-Cilium Interface. Cell. 163(6):1484-149. (*Co- first author). DOI: 10.1016/j.cell.2015.10.065