CENTROSOME NUMBER HOMEOSTASIS:
LESSONS FROM CEP135 ISOFORM DYSREGULATION IN BREAST CANCER
By
DIVYA GANAPATHI SANKARAN
Bachelor of Technology, Anna University, India, 2013
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Cancer Biology Program
2019
This thesis for the Doctor of Philosophy degree by
Divya Ganapathi Sankaran
has been approved for the
Cancer Biology Program
By
Rytis Prekeris, Chair
Jeffrey Moore
David Bentley
Heide Ford
Mary Reyland
Chad G Pearson, Advisor
Date: 05/17/2019
ii Ganapathi Sankaran, Divya (Ph.D., Cancer Biology)
Centrosome Number Homeostasis: Lessons from CEP135 Isoform Dysregulation in
Breast Cancer
Thesis Directed by Associate Professor Chad G. Pearson
ABSTRACT
The centrosome, comprised of two centrioles surrounded by pericentriolar material, is the cell’s central microtubule organizing center. Centrosome duplication is coupled with the cell cycle such that centrosomes duplicate once in S phase. Loss of such coupling produces supernumerary centrosomes, a condition called centrosome amplification (CA). CA can promote hallmarks of tumorigenesis. In this thesis, I investigate the contribution of centriole overduplication to CA and its consequences on microtubule organization and genomic stability in breast cancer cells. CEP135, a centriole assembly protein, is dysregulated in some breast cancers. We previously identified a short isoform of CEP135, CEP135mini that represses centriole duplication.
CEP135mini represses centriole duplication by limiting the localization of essential proteins required for centriole duplication. Interestingly, the relative level of CEP135full to
CEP135mini (the CEP135full:mini ratio) is higher in centrosome amplified breast cancer cell lines. Specifically, I demonstrate that inducing expression of CEP135full increases CA. In contrast, elevating CEP135mini reduces centrosome number in breast cancer cells. I find that the CEP135 isoforms in vivo are generated by alternative polyadenylation. A directed genetic mutation near the CEP135mini alternative polyadenylation signal reduces the CEP135full:mini ratio and decreases CA, accordingly. Thus, dysregulation of the CEP135 isoforms can promote CA in breast cancer cells. Furthermore, in order to
iii characterize the consequences of CA in microtubule organization and chromosome segregation, we designed a semi-automatic algorithm using machine learning to detect and analyze centrosomes and microtubule organization. Using this algorithm, I find that centrosome amplified breast cancer cells have microtubule organization defects such as increased microtubule density. Moreover, dysregulation of CEP135 isoforms that promotes CA also increases the frequency of multipolar spindles, anaphase-lagging chromosomes, and micronuclei leading to chromosome segregation defects. I conclude that dysregulation of the CEP135 isoforms can promote centriole overduplication and results in chromosome segregation errors in breast cancer cells.
The form and content of this abstract are approved. I recommend its publication.
Approved: Chad G Pearson
iv ACKNOWLEDGEMENTS
I am very grateful to Dr. Chad G Pearson for all the life lessons I have acquired from
him over the past five years. He has taught me to be focused, to be thoughtful and to
never stop questioning. Chad has shown respect for me as a scientist through his
personal investment in my learning. I am very thankful for all his hard work and have
enjoyed being his student. I especially would like to thank him for giving me those
chances to bounce back when I struggled and for space he provided me to dig
intellectually deeper, grow and be successful.
I would also like to thank my lab members Alexander Stemm-Wolf, Kristin Dahl, Dr.
Nick Galati, Dr. Marisa Ruehle, Dr. Brian Bayless, Anthony Junker, and Adam Soh for creating a very collaborative and friendly environment for the last five years. My lab members have believed in me and encouraged me on my hardest days. I have learned
many things from all my lab members and they have been my home away from India.
I would like to dedicate my thesis to my husband Dr. Bharath Hariharan, my parents
Ganapathi Sankaran, Banumathi Ganapathi Sankaran and, my sister Dr.
Jayadurga Ganapathi Sankaran. My parents have worked very hard and selflessly for me. Bharath has motivated me relentlessly. They showed me how to work hard, taught
me to never look down upon failures and be very courageous throughout this journey.
I would also like to acknowledge my-in-laws Rama Hariharan and Hariharan, who
have been very kind and supported me throughout my graduate career.
Most importantly, I would also like to thank my committee members Dr. Jeffery Moore,
Dr. Rytis Prekeris, Dr. David Bentley, Dr. Mary Reyland and Dr. Heide Ford for all the great advice and the Cancer Biology Program for helping me pursue my interests.
v TABLE OF CONTENTS
CHAPTER
CENTROSOME NUMBER HOMEOSTASIS IN DEVELOPMENT AND DISEASE ...... 1
Introduction ...... 1
Centrioles, Pericentriolar Material, and Centrosome Self-Assembly ...... 2
Centrosome Number Homeostasis and its Loss during Development and Disease
...... 14
Centrosome Amplification in Cancer ...... 20
Centrosome Aberrations in Solid and Hematological Tumors ...... 21
Aggressive Breast Cancer Cells possess Increased Number of Centrioles and
Centrosome Amplification ...... 25
Causes of Centrosome Amplification ...... 27
Cell Cycle Aberrations ...... 27
Centriole Overduplication ...... 29
Consequences of Centrosome Amplification ...... 30
Genome Instability ...... 30
Centrosome Amplification in Tumorigenesis ...... 33
Thesis Outline ...... 38
Materials and Methods for Chapter I...... 39
Cell Culture ...... 39
Immunofluorescence ...... 39
vi Microscopy ...... 40
Centriole and Centrosome Number Counts ...... 40
Statistics and Biological Replicates ...... 41
CEP135 ISOFORM DYSREGULATION PROMOTES CENTROSOME AMPLIFICATION
IN BREAST CANCER CELLS ...... 42
Introduction ...... 42
Results ...... 48
Centriole Overduplication contributes to Centrosome Amplification in Breast Cancer
Cells ...... 48
The CEP135full:mini Ratio is Elevated in Centrosome Amplified Breast Cancer Cells
...... 52
Elevated CEP135full is Sufficient to Increase Centrosome Amplification in Breast
Cancer Cells...... 57
Elevated CEP135mini Limits the Centriolar Levels of Essential Duplication Factors 60
Elevated CEP135mini is Sufficient to Repress Centrosome Frequency in Breast
Cancer Cells...... 62
Mutations Affecting CEP135mini Alternative Polyadenylation Reduce the
CEP135full:mini Ratio and Centrosome Amplification ...... 64
Discussion ...... 71
Centriole Overduplication in Breast Cancer ...... 71
vii CEP135 Isoform Dysregulation Promotes Centriole Overduplication in Breast
Cancer ...... 72
CEP135mini in Repression of Centriole Duplication...... 73
Alternative Polyadenylation in CEP135 Isoform Regulation ...... 74
Materials and Methods for Chapter II ...... 75
Cell Culture ...... 75
Generation of mCherry-CEP135full-Tet and GFP-CEP135mini-Tet Cells ...... 76
Generation of 3’ UTR Mutant Cells ...... 77
Transfections...... 80
Immunofluorescence ...... 80
Microscopy ...... 81
Fluorescence Intensity Quantitation ...... 82
Centriole and Centrosome Number Counts ...... 82
Reverse-Transcription PCR and Quantitation ...... 83
3’RNA-Ligation Mediated RACE ...... 83
Statistics and Biological Replicates ...... 84
CEP135 ISOFORM DYSREGULATION AND CENTROSOME AMPLIFICATION IN
MICROTUBULE ORGANIZATION AND CHROMOSOME SEGREGATION ...... 86
Introduction ...... 86
Results ...... 91
viii An Image Processing Algorithm to Quantify Centrosome and Microtubule
Organization Defects ...... 91
A Semi-Automatic Machine Learning Algorithm Quantifies Centrosome and
Microtubule Organization Defects ...... 96
Amplified Centrosomes Possess Higher Microtubule Density ...... 102
Centrosome Amplified Breast Cancer Cells have Longer and Higher Density of
Microtubule Growing Ends ...... 106
Semi-Automatic Machine Learning Algorithm recapitulates the Pericentriolar
Defects in Amplified Centrosomes of Breast Cancer Cells ...... 109
Centrosome Amplification Promotes Chromosome Mis-Segregation in Breast
Cancer Cells...... 111
CEP135 Isoform Dysregulation Promotes Chromosome Mis-Segregation in Breast
Cancer Cells...... 113
Discussion ...... 116
Semi-Automatic Algorithm for Centrosome Detection ...... 116
PCM at Amplified Centrosomes ...... 117
Centrosome Amplification Affects Microtubule Organization ...... 118
CEP135 Isoform Dysregulation in Chromosome Segregation ...... 120
Materials and Methods for Chapter III ...... 121
Cell Culture ...... 121
ix Generation of Tetracycline Inducible mNeon-EB3-MCF10A and mNeon-EB3-MDA-
231 and mCherry-CEP135full-Tet or GFP-CEP135mini-Tet Cells ...... 122
Transfections...... 123
Immunofluorescence ...... 123
Microscopy ...... 124
Centriole and centrosome number counts ...... 125
Chromosome Mis-Segregation Counts ...... 125
Statistics and Biological Replicates ...... 125
CHAPTER IV ...... 127
CONCLUDING REMARKS AND FUTURE PROSPECTS ...... 127
CEP135 Isoform-Dependent Centriole Assembly and Mechanism Of Centriole
Number Control ...... 128
Cell Cycle Regulation of CEP135 Isoforms ...... 129
CEP135mini in Repression of Centriole Duplication...... 131
CEP135 Isoform Dysregulation in the Cause of Centrosome Amplification In Cancer
...... 132
Centriole Overduplication in Centrosome Amplification of Breast Cancers ...... 132
Alternative Isoforms and their Regulation in Centrosome Amplification ...... 135
Consequences of Centrosome Amplification ...... 138
Chromosome Mis-Segregation...... 138
x Microtubule Assembly and Organization ...... 139
REFERENCES ...... 141
xi LIST OF FIGURES
FIGURE
1.1: Centrosomes, centrioles, and PCM ………………………………………..….………...3
1.2: Centrosome duplication cycle ……...…………………………………….………………4
1.3: Cartwheel assembly ...……………………………………………………….……………5
1.4: Cross-section of the centriole at the proximal end ...……………………..……………7
1.5: Schematic of CEP135 ...…………………………………………………………………..8
1.6: Schematic of centriolar tubules ...……………………………………………………..…9
1.7: Distal and sub-distal appendages ...……………………………………………………10
1.8: PCM assembly ...…………………………………………………………………………11
1.9: Centrosome disjunction and centriole disengagement ..…………………………….13
1.10: PLK4 induced centriole overduplication ...…………..……………………………….16
1.11: Aggressive breast cancer cells exhibit increased centrosome amplification ...…..26
1.11.1: Supplemental, aggressive breast cancer cells exhibit increased centrosome amplification ...... 26
1.12: Centrosome amplification promotes chromosome segregation errors ...…………32
2.1: Schematic of CEP135 isoforms ...………………………………………………………43
2.2: Some alternative splicing models ..…………………………………………………….45
2.3: Some alternative polyadenylation models ...…………………………………………..46
2.4: Centriole overduplication in breast cancer cells that have centrosome amplification.
………………………………………………………………………………………………...…49
2.4.1: Supplemental, centriole overduplication in breast cancer cells with amplified centrosomes……………………………………………………………………………………51
xii 2.5: CEP135 isoform transcript and protein levels are altered in breast cancer cells….54
2.5.1: Supplemental, CEP135 transcript and protein levels are altered in breast cancer cells ...………………………………………………………………………………………..…56
2.6: Elevated CEP135full increases breast cancer cell centrosome amplification ...……58
2.6.1: Supplemental, elevated CEP135full expression increases centriole number and centrosome amplification in breast cancer cells .…………………………..……………..59
2.7: CEP135mini expression displaces SAS-6 and CPAP from centrosomes……………61
2.8: Elevated CEP135mini is sufficient to decrease centrosome amplification…..………63
2.8.1: Supplemental, elevated CEP135mini expression is sufficient to decrease centrosome amplification in breast cancer cells ..………..………………………………..64
2.9: CEP135mini is an alternatively polyadenylated isoform, and mutations near the
CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells ...…………………………………………………………………….……67
2.9.1: Supplemental, CEP135mini is an alternatively polyadenylated isoform, and mutations near the CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells ..…...…………………………...……………..69
3.1: An image processing algorithm to quantify centrosome and microtubule organization defects ….………………….……………………………………………………94
3.2: A semi-automatic machine learning algorithm that quantifies centrosome and microtubule organization defects………………………………………………………..….100
3.2.1: Supplemental, pipeline for estimating boundary strength based on the convolutional network output, p(x) …………………………………………………………102
xiii 3.3: Centrosome amplified breast cancer cells have pericentriolar and microtubule organization defects………………………………………………………………………….104
3.4: Centrosome amplified breast cancer cells have higher EB3 density and comet length………………………………………………………………………………..…………108
3.5: Semi-automatic machine learning algorithm recapitulates the image pericentriolar defects in amplified centrosomes of breast cancer cells…………………………………110
3.6: Centrosome amplification promotes chromosome missegregation in breast cancer cells……………………………………………………………………………………….……112
3.7: CEP135 isoform dysregulation is sufficient to promote multipolar mitosis, anaphase lagging chromosomes and formation of micronuclei in breast cancer cells……..…….114
xiv LIST OF ABBREVIATIONS
-TuRC- -Tubulin Ring Complex
AA: Amino acid
APC/C: Anaphase Promoting Complex/Cyclosome
BMP: Bone Morphogenic Protein
Bps: Basepairs
CA: Centrosome Amplification
CCD: Charge Coupled Device cDNA: complementary Deoxyribo Nucleic Acid
CEP135: Centrosomal Protein 135kDA
CMOS: Complementary metal–oxide–semiconductor
CRISPR: Clustered Regularly Interspaced Short Palindromic Repeats
DCIS: Ductal Carcinoma
DIC: Differential interference contrast
DNA: Deoxyribo Nucleic Acid
ECM: Extra Cellular Matrix
EMCCD: Electron Multiplying CCD
ER: Estrogen Receptor
GEFs: guanine nucleotide exchange factor
GTP: Guanosine Triphosphate
HCl: Hydro Chloric Acid
HPV: Human Papilloma Virus
HTLV: Human T-Lymphotropic Virus
xv kDa: Kilodaltons
MDA-231: MDA-MB-231 mRNA: messenger RNA nm: Nanometers nts: nucleotides
PBS: Phosphate Buffered Saline
PCM: Pericentriolar Material
PCR: Polymerase Chain Reaction
PLK: Polo-like Kinase
PR: Progesterone Receptor
RNA: Ribonucleic acid rtPCR: Reverse transcription PCR
SAS-6: Spindle assembly Abnormal protein-6
SIM: Structured Illumination Microscopy
Tet: Tetracycline
UTR: Untranslated Region
UV: Ultraviolet
WT: Wildtype
ZR751: ZR75.1
μm: Microns
xvi CHAPTER I
CENTROSOME NUMBER HOMEOSTASIS IN DEVELOPMENT AND DISEASE1
Introduction
Microtubules are highly dynamic tubulin polymers that perform important roles in a plethora of cellular processes including cell division, migration, intracellular trafficking, polarization and signaling (Desai and Mitchison, 1997). Microtubules are intrinsically polar, having dynamic plus ends that stochastically switch between the states of polymerization and depolymerization and a relatively less dynamic minus ends at the structures known as centrosomes (Allen and Borisy, 1974; Bergen and Borisy, 1980;
Walker et al., 1988; Desai and Mitchison, 1997). Centrosomes are the cell’s primary microtubule organizing centers (Brinkley, 1985). Edouard Van Beneden, Theodore
Boveri and Walter Flemming were some of the first cell biologists to spot centrosomes in the early nineteenth century (Boveri, 1888; Gall, 2004; Watt and Sever, 2008;
Bornens et al., 2014). The term centrosome is etymologically derived from the Latin term ‘centrum’ or center and Greek term ‘soma’ or cell body. Some of the most daring postulates were put forth based on the centrosome’s suggestive position at the cell center as the organizers of cell division, coordinators of cytokinesis and drivers of malignant transformation. Centrosomes fueled the formulation of the chromosome theory of heredity (Bornens and Gönczy, 2014).
1 Portions of this chapter are published with permission from our previously published article:
Ganapathi Sankaran, D, Stemm-Wolf, AJ, and Pearson, CG (2019). CEP135 isoform dysregulation promotes centrosome amplification in breast cancer cells. Molecular Biology of the Cell.
1 A century later centrosomes are now known to mediate the microtubule functions by nucleating the cytoplasmic microtubules during interphase and facilitating bipolar spindle formation during mitosis that is necessary for faithful chromosome segregation
(Nigg and Raff, 2009; Meraldi, 2016). Centrosomes are also important in the organization and positioning of the Golgi apparatus (Rios, 2014). They aid in directed cell migration by establishing the axis for directional movement (Hurtado et al., 2011).
Centrosomes also have other functions independent of their role in microtubule assembly. In certain cell types, cell cycle exit triggers the conversion of the centrosome into a basal body that nucleates the primary cilium (Sorokin, 1962). The primary cilia are signaling centers important in maintaining cell proliferation and differentiation (Fonte et al., 1971). Centrosomes also scaffold proteins that act as signaling platforms essential in cell cycle progression and DNA damage response (Arquint et al., 2014). Furthermore, centrosomes play a role in cell polarization by determining the plane for asymmetric cell divisions (Yamashita, 2009). Centrosome structure and numbers are reported to have gone awry during development and implicated in several human diseases highlighting the importance of centrosomes in cellular function and organ development (Nigg and
Raff, 2009; Chavali et al., 2014; Gonczy, 2015).
Centrioles, Pericentriolar Material, and Centrosome Self-Assembly
Centrioles and pericentriolar material:
Centrosomes mainly consist of a pair of orthogonal cylindrical centrioles embedded in a proteinaceous pericentriolar material (Figure 1.1) (Brinkley, 1985).
Centrioles are one of the most recognizable, remarkably conserved, protein-rich structures of biology. Nine triplets of microtubules are symmetrically organized into the
2 centriolar barrel that is approximately 250nm in diameter and 500nm in length (Li et al.,
2012). The nine-fold symmetry of centrioles is one of the highly conserved features of centrioles across evolution
(Carvalho-Santos et al., 2011). Centrioles The intrinsically polar nature of microtubules confers polarity Pericentriolar Material to centrioles. The minus end Figure 1.1: Centrosomes, centrioles and PCM. Centrosomes consist of pair of centrioles (red) in of microtubules defines the the pericentriolar material (green). proximal end of centrioles while the plus end of microtubules is the distal end of centrioles. At the proximal end, inside the microtubule scaffold, a well-defined structure known as the cartwheel exists in the centriolar lumen. The cartwheel, as the name suggests, consists of proteins that constitute the inner central hub with spokes emanating from it that connect to the pinhead hooks of the outer cartwheel. The cartwheel structure in the lumen defines the template and confers the nine-fold symmetry to the duplication of new centrioles (Hirono, 2014). At the proximal end of the centrioles, outside the microtubule scaffold, there are filaments that connect the two centrioles called the centriolar linkers (Agircan et al., 2014). On the contrary, there is luminal density in the distal end of centrioles and on the outside, there are structures known as distal and sub-distal appendages that are exclusively present only in the older centrioles (Winey et al., 2014). While the structure of centrioles is well-studied, the surrounding pericentriolar material (PCM) was considered an amorphous cloud of proteins until recently, due to limitations in methodology and its disorganized nature.
Super-resolution microscopy now reveals that centrioles form the basis of deposition of
3 the pericentriolar material and the PCM is an ordered structure consisting of scaffolds
that facilitate and organize microtubule assembly (Mennella et al., 2014; Woodruff et al.,
2014).
Centrosome self-assembly:
G1 phase cells have one centrosome with two centrioles that will each duplicate
once during S phase (Vorobjev and Chentsov, 1982; Piel et al., 2000). This produces
the two centrosomes with four centrioles that ensure bipolar spindle assembly and
Mitosis
G2 phase S phase G1 phase
Figure 1.2: Centrosome duplication cycle. Duplication of one centrosome with two centrioles (red) into two centrosomes and four centrioles through the centrosome duplication cycle. G1 phase centrosome duplicates during S phase and matures during G2 phase and segregates during mitosis. This allows the formation of the bipolar spindle and equal segregation of chromosomes (purple). faithful chromosome segregation during mitosis (Figure 1.2). The centrosome
duplication cycle dictates the self-assembly of the new daughter centriole and occurs in
consonance with the DNA replication cycle (Nigg and Stearns, 2011; Fırat-karalar and
Stearns, 2014). Centrosome self-assembly begins during the G1-S transition with
duplication of the cartwheel structure orthogonal and adjacent to the pre-existing
centriole. The self-assembly of the cartwheel structure is initiated by the redistribution of
4 the centriolar protein and a key polo-like kinase known as PLK4 from the circumference of the pre-existing mother centriole to a distinct site at the proximal end of the mother centriole (Figure 1.3). Cartwheel This redistribution occurs PLK4 to only one site per centriole to control the SAS6 CEP135 number of centrioles formed per cell cycle Figure 1.3: Cartwheel assembly. Cartwheel defines the template for duplication of new centrioles. Plk4 (blue), SAS6 (Kleylein-Sohn et al., (grayscale) and CEP135 (green) are some proteins in the formation of cartwheel structure. 2007). Centriolar proteins CEP152 and CEP192, which distribute symmetrically around the circumference of the centriole barrel, interact with PLK4 to cooperate in its redistribution and recruitment. This redistribution and recruitment to one distinct site on the proximal end of the centriole is dependent on the interactions between the positively charged polo- box domain of PLK4 and acidic regions of CEP192 and CEP152 (Kim et al., 2013;
Sonnen et al., 2013). However, it remains unknown how a particular site from a radially symmetric surface is chosen to define the origin for centriole duplication through PLK4.
Recent biophysical models suggest a synergistic activity-dependent retention of PLK4 and its binding partner STIL on the surface of the centriole that promotes spatial clustering of active PLK4-STIL. This leads to the emergence of a single PLK4 maximum from multiple PLK4 maxima on the radially symmetric surface to define the origin of centriole duplication (Ohta et al., 2014; Leda et al., 2018). While this model can
5 theoretically explain how a single origin of duplication is chosen, this model needs to be
experimentally investigated.
Inner cartwheel assembly:
Centriolar proteins SAS6 and STIL localize to this site during S phase and facilitate the formation of the inner hub and spokes of the cartwheel structure (Figure
1.3). The SAS6 molecule consists of a conserved N-terminus domain, a coiled-coil domain, and a non-conserved C-terminus domain. SAS6 molecules oligomerize such that nine homodimers assemble into a ring through an N to N terminus interaction that resembles the inner hub of the cartwheel (Figure 1.4). The interaction between the coiled-coil domains of nine homodimers allows the formation of the nine spokes of the cartwheel. This oligomerization property of nine SAS6 homodimers confers the nine-fold symmetry to the centrioles (Kitagawa et al., 2011; van Breugel et al., 2011; Gönczy,
2012). The nine-fold symmetry of centrioles is one of the highly conserved features of centrioles. However, it remains to be discovered as to how and why nine-fold symmetry is so highly conserved across evolution. To unravel the underlying function of nine-fold symmetry, it would be essential to understand what the nine-fold symmetry confers in comparison to a five- or seven- or other-symmetrical geometry. Thus, nine SAS-6 dimers oligomerize and load on to this active PLK4 site to form an internal hub and nine spokes of the cartwheel (Gopalakrishnan et al., 2010; van Breugel et al., 2011; Gönczy,
2012). One important question is how is the oligomerization so precisely regulated? It is possible the cellular levels of SAS-6 are regulated in such a fashion to form exactly one cartwheel or as-of-yet-unidentified factors regulate the SAS-6 oligomerization.
6 STIL is required for centriole duplication and is hypothesized to cross-link the spokes of SAS6 homodimer and promote the oligomerization of SAS-6 since its localization is observed close to the inter-spoke cross-links (Vulprecht et al., 2012;
Hirono, 2014; Ohta et al., 2014). However, STIL also binds directly to outer-cartwheel proteins such as CPAP (Tang et al., 2011). Hence the STIL cross-linking model and its precise localization at the cartwheel requires further investigation. This assembly of inner cartwheel during the G1-S transition initiates the formation of a new centriole.
The length of the spokes and assembly of the outer cartwheel is as important as the establishment of the central hub of the cartwheel in the formation of the centriolar structure. The highly conserved centriolar protein CEP135 acts as a pinhead hook that connects the spokes of SAS6 oligomers to the SAS6 outer cartwheel Hub N C N proteins CPAP and CEP135 microtubules (Figure
1.4) (Hiraki et Figure 1.4: Cross-section of the centriole at the proximal end. SAS6 (grayscale) forms the inner hub and spokes of the cartwheel. al., 2007; Lin et CEP135 (green) acts a pinhead hook that connects the spokes of SAS6 to outer triplet microtubules (red). al., 2013;
Hirono, 2014). CEP135 (Centrosomal protein 135kDA) is a coiled-coil protein of 1140 amino acids essential for the formation of centrioles in several species (Kilburn et al.,
2007; Kleylein-Sohn et al., 2007; Jerka-Dziadosz et al., 2010). The domain structure and function of CEP135 is remarkably conserved in several unicellular organisms to
7 chordates. The orthologue of CEP135 is Bld10p, which is extensively studied and characterized in other species including Chlamydomonas reinhardtii, Paramecium tetraurelia, and Tetrahymena thermophilia (Carvalho-Santos et al., 2010). Depletion of
CEP135 in many chordates has phenotypes varying from complete loss of centrioles to centriole structure and length defects (Hussain et al., 2012; Inanc et al., 2013; Lin et al.,
2013; Dahl et al., 2015). Similarly, the depletion of Bld10p in many unicellular organisms also leads to complete loss of centrioles(Matsuura et al., 2004; Jerka-Dziadosz et al.,
2010; Bayless et al., 2012). Bld10p localizes to the distal end of the spokes and controls the length of the spokes (Guichard et al., 2013). Deletion of the Bld10p C-terminus reduces the spoke length and pinhead size, suggesting that the C-terminus of Bld10p constitutes parts of the SAS-6 spoke shaft and pinhead (Figure 1.5) (Hiraki et al., 2007).
In conservation with this domain structure, the C-terminus of CEP135 also binds to
SAS-6 in human cells (Lin et al., 2013).
The crystal structure of the N-terminus of CEP135 was recently solved in human cells. The positively charged N-terminus CPAP binding SAS-6 binding of CEP135 in human cells forms an CEP135 N 1140 aa-134kDA C extensive homo-dimeric coiled-coil Microtubule binding structure with chains arranged in parallel. Figure 1.5: Schematic of CEP135 and its binding domains. The N terminus binds A basic 13 amino acids(aa) fragment in to microtubules (orange) and CPAP (blue). The C-terminus of CEP135 binds this N-terminus of CEP135 interacts with to SAS-6(green). the negatively charged outer surface of the triplet microtubules of centrioles (Kraatz et al., 2016). Furthermore, the N-terminus of CEP135 also binds to an outer cartwheel protein CPAP in human cells(Lin et al., 2013) (Figure 1.5). In conservation with its
8 domain structure, the N-terminus of Bld10p localizes near the triplet microtubules
(Jerka-Dziadosz et al., 2010). Furthermore, Bld10p mutants lack the entire triplet microtubules, suggesting that Bld10p may recruit the -tubulin ring complex essential for microtubule nucleation and organization (Leidel and Gönczy, 2003; Matsuura et al.,
2004). Thus CEP135 functions to tether the inner cartwheel proteins to the outer cartwheel structures.
Outer cartwheel and PCM assembly:
CPAP, an outer cartwheel protein binds to CEP135 through its C-terminus(Lin et al., 2013). CPAP can bind to tubulin dimers and, just as CEP135/Bld10p, is important in the organization of triplet microtubules(Cottee et al., 2013). There exists only one cartwheel in human cells, while many other species contain a stack of cartwheels at the proximal end of the centriole (Hirono, 2014). Outside the cartwheel, the centrioles are defined by the triplet microtubules. The triplet microtubules are one of the most recognizable structures of centrioles. Microtubules are tubules, typically made up of 13 linear protofilaments containing - and -tubulin heterodimers, assembled around a hollow core.
Three such tubules are called triplet microtubules.
While human centrioles contain triplet microtubules,
C certain species contain doublet and singlet B A microtubules (Winey et al., 2014). The triplet Figure 1.6: Schematic of microtubules are at a small angle to the centriolar microtubules (red). Cartwheel has triplet circumference of the cylinder. When triplet microtubules: A, B and C tubules. microtubules are viewed from the proximal end along the axial plane, they are twisted
9 anticlockwise (Guichard et al., 2013). The triplet microtubules of centrioles are comprised of the A-, B- and C- tubules (Figure 1.6). The A-tubule contains a complete microtubule with 13 - and -tubulin-containing protofilaments, while the B- and C- tubule contain 10 protofilament microtubules. The A-, B- and C-tubules are connected through linkers (Li et al., 2012). The length of the triplet microtubules or centrioles in human cells is around 500nm. The number of cartwheels in the centriolar stack is hypothesized to limit the elongation of the triplet microtubules. Another plausible model is that dilution of some cartwheel assembly factor from the proximal end is utilized as a molecular ruler to control the triplet microtubule elongation. The elongation of triplet microtubules at the distal end of centrioles occurs during late S phase or early G2 phase and is restricted by the capping proteins including CP110 (Tsang and Dynlacht,
2013). Unlike the proximal end of centrioles, the distal end of centrioles is poorly studied.
The distal end of centrioles consists of proteins such as centrin and is decorated with appendages and sub-appendages, outside the Distal appendage Older centriole microtubule scaffold (Figure 1.7). Distal appendages are Sub-distal appendage fibrous extensions emanating out from the distal end of the triplet microtubules (Paintrand et al., 1992). These fibrous extensions are added specifically to older centrioles. CEP164, CEP123, and FBF1 are some Figure 1.7: Schematic of distal distal appendage proteins. Sub-distal appendages and sub-distal appendages. Distal (blue) and sub-distal (orange) are triangular extensions attached laterally and appendages are shown on the older centriole. proximal to the distal appendages (Figure 1.7). CEP170 and Ninen are some of the sub-
10 distal appendage proteins. The distal appendages have a function in cilia formation, while the sub-distal appendages function to anchor centrosome-associated microtubules (Jana et al., 2014).
The assembly of centrioles dictates the duplication of the pericentriolar material.
Unlike the defined centriolar structure, the disorganized nature of PCM has made it challenging to understand the PCM structure, dynamics, and assembly. During S phase, a thin layer of PCM is deposited concomitantly around the new daughter centriole. This thin layer is composed of proteins known as the ‘centromatrix’ that constructs the PCM scaffold (Figure 1.8). Some of the core PCM proteins in the
Mitosis G1 phase S phase
Pericentrin Mother centriole γ- tubulin and other PCM proteins
Figure 1.8: PCM assembly and deposition through cell cycle. PCM proteins such as pericentrin (orange) and -tubulin (green) accumulate during G2-M transition. centromatrix include pericentrin and CDK5RAP2 (Gosti-testul et al., 1986). PCM proteins contain an abundance of coiled-coil domains that promote interactions between each other to facilitate a toroid organization during interphase. Pericentrin is organized as a toroid with a superficial nine-fold symmetry emanating from the centrioles and is hypothesized to serve as a molecular ruler for the deposition of the thin inner PCM layer
(Mennella et al., 2014). Coincident with the PCM scaffold deposition, the -tubulin ring complex (-TuRCs), composed of several subunits, spanning 25-30 nm in diameter, is
11 organized within in this thin layer. The -TuRCs along with CDK5RAP2 facilitate the nucleation and organization of microtubule protofilaments (Dictenberg et al., 2002; Fong et al., 2007). The 13 protofilaments of α- and β- heterodimers within a microtubule is established by the cap-like structure composed of -TuRC subunits in these scaffolds
(Stearns and Kirschner, 1994; Moritz et al., 1995; Zheng et al., 1995; Kollman et al.,
2011). However, -TuRC subunits alone are poor nucleators of microtubules in vitro
(Oegema et al., 1999). This suggests that other proteins in the PCM may promote the sequestration of α- and β- heterodimers. This may increase the local concentration of α- and β- heterodimers to promote the microtubule nucleation through -tubulin ring complexes. How PCM increases the local concentration of α- and β- heterodimers to promote nucleation through -TuRC is not clear. The microtubule nucleation and organization through the -TuRC and other PCM components allow the centrosomes to
be the microtubule organizing center of cells.
Centrosome disjunction and centriole disengagement:
During the G2 phase, the formation of two new daughter centrioles from the two pre-existing mother centrioles is complete. However, two pre-existing centrioles remain connected on their proximal end through linker proteins. Some of the major centriolar linker proteins include Nek2 kinase, C-Nap1, Rootletin and hSgo1(Bahe et al., 2005;
Yang et al., 2006; Agircan et al., 2014). C-Nap1 molecules are present at the base of the proximal end of centrioles and Rootletin fibers connect the two pre-existing mother centrioles. Centrosome maturation occurs before mitosis where the linker proteins between the two pre-existing mother centrioles begin to dissolve through a process known as ‘centrosome disjunction’ (Figure 1.9). The kinase Nek2 dissolves the linker
12 Mitosis Late Mitosis
G2 phase Centriole disengagement
Figure 1.9: Centrosome disjunction and centriole disengagement. Separation of centrosome occurs before mitosis through centrosome disjunction. Separation of mother and daughter centrioles occurs during telophase through centriole disengagement. proteins between the two pre-existing mother centrioles and promotes centrosome disjunction (Faragher and Fry, 2003). Centrosome disjunction is necessary for the two centrosomes to separate before mitosis. Nuclear envelope also breaks down before mitosis. This allows two centrosomes to move to the opposite end of cells and facilitates the formation of the bipolar spindle apparatus during the pro-metaphase transition.
Several motors including Eg5 and the nuclear envelope bound dynein drive centrosome separation(Mountain et al., 1999). However, each mother and its daughter centriole remain tightly associated or engaged with each other.
PCM expansion also occurs during this period through the phosphorylation of a key kinase known as PLK1. PLK1 phosphorylates pericentrin and CDK5RAP2 that promote the expansion through recruitment of more pericentriolar material (Lee and
Rhee, 2011). The organized nature of PCM during interphase becomes dis-organized during the metaphase-anaphase transition. The underlying mechanism that drives this dis-organization is poorly understood. Furthermore, the minimal components that are
13 necessary and sufficient to drive PCM deposition, organization and dis-organization are also not studied. Another kinase important in centrosome maturation is Aurora A kinase.
Loss of Aurora A kinase prevents deposition of the -TuRC in these scaffolds (Glover et al., 1995).
As the cell cycle progresses into late mitosis, PLK1 and Separase, a protein responsible for the separation of the sister chromatids during the metaphase-anaphase transition, dis-engage the tightly associated mother and daughter centrioles during the telophase (Tsou and Stearns, 2006b; Thein et al., 2007; Agircan et al., 2014). It is puzzling as to how an enzyme that is active at the metaphase to anaphase transition also promotes centriole disengagement during telophase. During telophase, the centrosomes also become dimmer and lose PCM from the centrosomes. It is hypothesized that de-phosphorylation of PCM proteins and the cortical forces during mitosis drive the disassembly of the PCM during mitosis (Decker et al., 2011). However, the underlying mechanisms that drive the assembly and disassembly require further investigation. Furthermore, the disengaged centrioles are connected to each other through a centriole linker in the succeeding cell cycle. However, little is known on how the centriole linkers are formed in the succeeding cell cycle.
Centrosome Number Homeostasis and its Loss during Development and Disease
Centrosome number control during the cell cycle:
The duplication of centrosomes is controlled such that every G1 phase cell has precisely one centrosome with two centrioles. A G1 phase centrosome duplicates only once during S phase such that every mitotic cell has two centrosomes. This is analogous to the DNA duplication cycle that fires the origin of replication only once
14 during S phase and segregates during mitosis. How such precise centrosome number homeostasis occurs in normal cells and how it is lost in disease is under extensive investigation and will be the focus of my thesis.
There are several regulatory steps that control the precise formation of one centriole and block the promiscuous re-duplication of centrioles. Additionally, such blocks must be relieved so the centrioles are licensed to duplicate the next cell cycle. At the level of the cell cycle, CDK2, the kinase that is active during the G1-S transition, is necessary for centriole duplication (Hinchcliffe et al., 1999). Centrosomes also have an intrinsic block to promiscuous reduplication. The intrinsic block prevents re-duplication of centrioles that previously duplicated. This was revealed by cell fusion experiments where cells from different cell cycle stages with duplicated and unduplicated centrosomes were fused together. It was observed that when centrosomes from a G2 cell that consisted of duplicated centrioles were fused with the cytoplasm of G1-S cells, these centrosomes were not able to reduplicate in the G1-S cytoplasm (Wong and
Stearns, 2003; Tsou and Stearns, 2006a, 2006b). This suggests that centrosomes block the reduplication of previously duplicated centrosomes. The blockage of re-duplication in previously duplicated centrioles occurs through the ‘engagement’ or orthogonal association of the mother and daughter centriole. This prevents its reduplication despite the presence of G1-S cytoplasm with all the necessary proteins to initiate centriole duplication. Disengagement in the cell cycle occurs during telophase and licenses centrioles to duplicate the next cell cycle (Wong and Stearns, 2003; Tsou and Stearns,
2006a, 2006b).
15 The first step in such number control is brought about by the accumulation of
PLK4 to a single site. The focal accumulation of PLK4 to a single site facilitates the formation of single daughter centriole from one mother centriole. Overexpression of
PLK4 leads to the formation of multiple pro-centrioles from a single centriole in the form of rosette structures (Habedanck et al., 2005; Kleylein-Sohn et al., 2007; Rodrigues-
Martins et al., 2007). Under these conditions, the rosette structure represents the
PLK4 induced Multi-cartwheels on a Centriole overduplication centriole rosettes single mother centriole
Figure 1.10: PLK4-induced centriole overduplication. Accumulation of multiple PLK4 (blue) foci around a single mother leads to formation of centriole rosettes (left panel). This promotes formation of multiple cartwheels (middle panel) and multiple daughter centrioles (right panel) on a single mother centriole. accumulation of multiple PLK4 foci instead of a single focal accumulation around the circumference of the centriolar barrel (Kleylein-Sohn et al., 2007) (Figure 1.10). The levels of PLK4 are important in the number control of centrioles and its regulation through the cell cycle is well-studied. PLK4 during the cell cycle is controlled by two known arms of regulation. PLK4 is a suicide kinase whose auto-phosphorylation initiates its own destruction. This promotes ubiquitin-mediated proteolysis through
SCFTrCP (Rogers et al., 2009; Holland et al., 2010, 2012b). The second arm of regulation involves p53 and stress-activated pathway that control the activity of PLK4
(Nakamura et al., 2013). These two arms of regulation suggest that PLK4 levels are
16 tightly regulated through the cell cycle to control the formation of one daughter centriole per mother centriole.
Several core centriolar proteins when overexpressed also lead to the formation of multiple procentrioles from a single mother centriole. Overexpression of SAS-6 and
STIL promotes the formation of excess pro-centrioles from a mother centriole (Leidel and Gönczy, 2003; Strnad et al., 2007; Arquint et al., 2012). SAS-6 levels are regulated through the cell cycle such that SAS-6 begins to accumulate at the end of the G1 phase and decreases in anaphase through proteasomal degradation. SAS6 and STIL have
KEN box motifs that are recognized by Cdh1 and target them to cyclosome-dependent degradation during anaphase (Strnad et al., 2007; Arquint and Nigg, 2014). Thus the levels of initiator proteins need to be tightly controlled to restrict the formation of two new centrioles during the cell cycle. Furthermore, the necessity of upstream kinases can be bypassed by the overexpression of the core centriolar proteins, suggesting overexpression of downstream initiator proteins is sufficient to misregulate the centriole duplication cycle.
Loss of centrosome number homeostasis in development and disease:
The number control in somatic cells is important in the regulated centrosome function. Several developmental and disease conditions lose such tight number control.
Centrosome amplification (CA) is a condition when cells have greater than two centrosomes. In contrast, when a cell has fewer than two centrosomes, it is deemed under-duplicated.
CA is observed during development in specialized cell types including multi- ciliated tracheal cells, olfactory neurons and ciliated sperms (Anderson and Brenner,
17 1971; Vladar and Stearns, 2007). The mammalian tracheal epithelial cells contain multi- ciliated cells that are important in the clearance of mucus. These multi-ciliary arrays are nucleated by hundreds of centrioles at their base. Hundreds of centrioles are generated by amplification of centrioles. Multiple centrioles in these cells are generated by the synergy of aberrant centriole duplication cycle and deuterosome-mediated centriole duplication (Vladar and Stearns, 2007; Dehring et al., 2013). The normal centriole duplication cycle mentioned above involves the formation of a daughter centriole utilizing the mother as the template centriole. The aberration of the centriole duplication cycle to generate hundreds of new centrioles occurs through a massive transcriptional upregulation of the core centriolar components including PLK4, SAS6, CEP135 and
CPAP (Hoh et al., 2012). A second de-novo non-templated centriole self-assembly cycle has been described. This involves the formation of new centrioles from non- descript electron-dense structures called the deuterosome (Dehring et al., 2013).
Olfactory sensory neurons are the main sensory cells in the olfactory epithelium.
The dendrites in these neurons stretch towards the nasal cavity that ends in a structure consisting of about 30 centrioles. These centrioles nucleate the cilia that project towards the epithelium. Studies on these structures suggest that about 30 centrioles assemble simultaneously, however, the underlying mechanisms remain unknown (Cuschieri and
Bannister, 1975).
CA in most of these developmental contexts involves the formation of multiple centrioles that nucleate the multi-ciliated arrays. However, outside the milieu of centriole amplification necessary for multi-ciliation, loss of centriole and centrosome number control has been negatively associated in various human maladies. CA is particularly
18 detrimental as it obliterates equal segregation of chromosomes between the daughter cells and promotes genome instability.
Microcephaly is a genetically heterogeneous neurodevelopmental disorder characterized by mental retardation and reduction in head circumference at birth that mainly affects the cerebral cortex region (Mahmood et al., 2011). It is caused by mutations in at least nine genes, of which five encode core centriolar components;
CEP135, CPAP, STIL, CEP152, and CDK5RAP2 (Gilmore and Walsh, 2013).
Mutations in these core centriolar components disrupt centriole biogenesis.
Microcephaly causing CEP135 mutations result in a truncated 324 amino acid protein that impairs centriole duplication. Microcephaly patient fibroblasts harboring the above
CEP135 mutations have a loss of centrosome number control (Hussain et al., 2012).
Similarly, CPAP microcephaly mutations disrupt its ability to bind to CEP135 and STIL and impair centriole duplication (Tang et al., 2011). In contrast, STIL microcephaly mutations prevent cell cycle-dependent degradation of the protein, thereby triggering centriole amplification (Arquint et al., 2014). Hence microcephaly patient fibroblasts have a heterogeneous centrosome number phenotype. The loss of centrosome number control leads to a mitotic delay in these microcephaly patients (Yang et al., 2008). The presence of prolonged mitosis causes cells to exit the cell cycle. It has been hypothesized that the delayed mitosis annihilates the proliferative capacity of these cells, possibly leading to reduced cerebral cortex volume in microcephaly.
Seckel syndrome and Meier-Gorlin syndrome define a group of genetic disorders that are characterized by pre-and post-natal growth defects including primary dwarfism
(Munnik et al., 2015). Primary dwarfism in these genetic disorders represents a
19 hypocellular condition where there exists a dramatic cell loss during fetal development
(Klingseisen and Jackson, 2011). Several genetic mutations that cause primary dwarfism are implicated in centrosome function (Rauch et al., 2008). Seckel derived patient fibroblasts and Meier-Gorlin syndrome exhibit CA. It has been hypothesized that
CA in these cells promotes genome instability leading to excess cell death (Kalay et al.,
2010). This explains the dramatic cell loss during fetal development and primary dwarfism in these genetic disorders.
Centrosome Amplification in Cancer
A likely model through which cancer arises is through multiple genetic assaults
(Vogelstein and Kinzler, 1993). Centrosome amplification (CA) is one such event that can drive several other genetic assaults through its effects on genome stability. In 1902,
Theodor Boveri proposed the bold idea that CA causes cancer. Boveri created eggs with multiple centrosomes and showed that those cells displayed multipolar mitoses
(Boveri, 1888; Scheer, 2014). These aneuploid progeny had different characteristics and he concluded that chromosomes transmit cellular traits, formulating the chromosome theory of heredity. However, Boveri did not work with cancer cells. His model was supported by some of his compeers Gino Galeotti and David von
Hansemann. They concluded that cancer cells had CA and abnormal mitotic figures, based on their observation on tumor histology (Hansemann, 1890; Galeotti, 1893). A century later the role of centrosomes and its causal association to cancer is still being investigated.
20 Centrosome Aberrations in Solid and Hematological Tumors
CA is observed in several types of hematological and solid tumors. Some of the common hematological tumors that exhibit centrosome defects include Hodgkin's and non-Hodgkin's lymphomas, acute and chronic myeloid leukemia, HTLV-1-associated adult T-cell leukemia and multiple myeloma (Krämer et al., 2005; Chan, 2011).
Acute myeloid leukemias are highly aneuploid. CA is observed in numerous hematopoietic disorders that evolve later into acute myeloid leukemias. It has been hypothesized that CA promotes the clonal evolution of pre-leukemic hematopoietic progenitors (Kearns et al., 2004; Neben et al., 2004). Similarly, chronic myeloid leukemia, which is characterized by the presence of BCR-ABL1 translocation generating a p210 fusion protein, also exhibit CA. Presence of CA in these BCR-ABL1 positive cells precedes chromosome instability. This suggests that CA is an early event in chronic myeloid leukemia patients (Giehl et al., 2005, 2007). Centrosome abnormalities are also frequently detected in B cell lineage lymphomas. Aggressive lymphomas such as diffuse large B-cell lymphoma, mantle cell lymphoma, Burkitt's lymphoma have more CA in comparison to the less aggressive lymphoma such as follicular lymphoma and marginal zone B-cell lymphomas. The proliferative/mitotic index also correlates with the degree of CA in these lymphomas(Krämer et al., 2005; Chan,
2011). These observations suggest that CA plays an important role in the evolution and progression of hematological tumors.
Solid tumors that exhibit centrosome abnormalities include cervical, colorectal, pancreatic, lung, prostate and breast tumors. Solid tumors have much higher CA than hematological tumors (Chan, 2011). CA has been detected in HPV positive cervical
21 dysplasia and invasive carcinoma. The mechanism underlying CA in cervical tumors is extensively studied. HPV encoded oncoprotein E7 generates CA and chromosome instability. E7 protein generates CA by interfering with the centrosome duplication cycle.
Overexpression of E7 promotes deregulation of CDK2 and PLK4. The upregulation of
PLK4 promotes the formation of multiple daughter centrioles from a single mother centriole. In transgenic mouse models, expression of the E7 protein is sufficient to invoke CA and invasive cervical lesions (Skyldberg et al., 2001; Duensing and Mu,
2002; Duensing et al., 2007). In colorectal carcinomas, CA is detected in low-grade carcinomas and associates with a higher grade invasive colorectal carcinomas. Some of the reported candidate oncogenes that are misregulated and correlate with CA in colorectal carcinomas include CEP135, Aurora A kinase and hSgo1(Ghadimi et al.,
2000). Similarly, CA is observed in prostate tumors progressing from low-grade prostate intraepithelial neoplasia to metastatic tumors. CA correlates with the chromosome instability observed in high-grade prostate tumors (Pihan et al., 2003a).
CA in breast cancers:
Breast cancer is the second most common cancer in the United States. It is the second leading cause of death and around 41,000 women will die from breast cancer in
2019. Breast cancers, analogous to other cancers, are classified based on the tissue of origin. Some subtypes of breast cancer, based on the origin of breast cancer, include ductal carcinoma, lobular carcinoma in situ, invasive ductal carcinoma and invasive lobular carcinoma. Ductal carcinoma in situ (DCIS) refers to a carcinoma that originates from the milk duct of the mammary glands and in situ refers to an original place. These carcinomas are non-invasive and rarely infiltrate from the tissue of origin. Similarly,
22 lobular carcinoma in situ refers to carcinomas of the lobules of the mammary gland and rarely infiltrate from the tissue of origin.
In contrast, invasive ductal carcinoma, as the name suggests, are infiltrating ductal carcinoma that originates at the milk ducts of the mammary gland. They are invasive and account for the most common subtype of breast cancer. Some subtypes of invasive ductal carcinoma include tubular carcinoma, medullary carcinoma, mucinous carcinoma, and papillary carcinoma. Similarly, invasive lobular carcinomas are infiltrating carcinomas of lobules in the mammary gland. They are invasive and the second most common subtype of breast cancer. Patients that exhibit ductal or lobular carcinoma in situ are at a higher risk to develop the invasive ductal or lobular carcinomas, suggesting that ductal or lobular carcinoma in situ are early precursors and later evolve into invasive carcinoma (Masood, 2016).
The idea that CA is an early event in breast tumorigenesis and can be important in the initiation of tumorigenesis comes from the data that CA is observed early in several DCIS patients. In these lesions, the degree of CA correlates to the advancement of these lesions (Lingle et al., 2002; Kronenwett et al., 2005; Liu et al.,
2009). These studies present correlative evidence that CA is a very early event in breast tumorigenesis.
The aggressiveness of breast cancer determines the prognosis and treatment course of breast cancer. Breast cancers are classified into three main molecular subtypes; hormone receptor positive, HER2 positive and triple-negative breast cancer.
Hormone receptor breast cancers test positive for the hormone receptors estrogen and/or progesterone receptors. Breast cancers that test positive for estrogen receptor
23 are known as (ER) positive breast cancers and those that are positive for progesterone receptors are known as (PR) positive breast cancers. Hormone receptor positive is the most common subtype of breast cancer. HER2 receptor positive subtype of breast cancer test positive for the HER2 receptor. HER2 receptor positive cancers can either be hormone receptor positive or negative. This subtype accounts for 20 percent of breast cancers. The above subtypes can also be classified as Luminal A and Luminal B breast cancers. Luminal A breast cancers are hormone receptor-positive (ER-positive and/or PR-positive), HER2 negative, low grade and slow growing breast cancers that have the best prognosis. Luminal B breast cancers are hormone receptor-positive (ER- positive and/or PR-positive), and/or HER2 negative that grow slightly faster than luminal
A breast cancers with a slightly worse prognosis.
Triple-negative breast cancers lack all of the above-mentioned receptors; ER- negative, PR-negative and HER2 receptor negative. These account for 15 percent of total breast cancers and are the most aggressive breast cancers. Similar to triple negative breast cancers are basal breast cancers that are hormone-independent and highly aggressive cancers. Often, breast cancers progress from a hormone-dependent to a hormone independent state making it hard to plan the treatment course (Dai et al.,
2015; Neve et al., 2006).
Seventy-five percent of the above breast tumors exhibit CA. However, the population that exhibits CA varies widely. The number of centrosomes per cell increase with the tumor grade, proliferation index and genome instability in breast cancer patient samples. Triple negative breast cancer patient samples have the highest CA (Lingle et al., 2002; Pihan et al., 2003b; Pannu et al., 2015; Denu et al., 2016).
24 This thesis focuses on CA in breast cancer cells. First, I measured the frequency of cells with CA in the cell lines representing the luminal and basal breast cancer molecular subtypes and validated the presence of CA in various breast cancer subtypes(Salisbury et al., 2004).
Aggressive Breast Cancer Cells possess Increased Number of Centrioles and
Centrosome Amplification
Cells with CA are conservatively defined as cells with numerically more than two centrosomes, regardless of their cell cycle stage (Figure 1.11A). By this conservative criterion, approximately 5% of the cell population of normal-like, MCF10A breast cells have CA. Less aggressive breast cancer cell lines have 7-13% of cells with CA, while more aggressive breast cancer cell lines exhibit 12-24% of cells with CA (Figure 1.11B and 2.5A; (Neve et al., 2006; Marteil et al., 2018).
In addition to having a higher percentage of cells with CA, the cells with CA in more aggressive breast cancer types also have more centrioles and centrosomes per cell (Figure 1.11C and 1.11.1A). The excess number of centrioles in these cells with CA suggests that centriole duplication contributes to CA in breast cancer cell lines.
These results strengthen previous findings that the percentage of cells with CA is greater in aggressive breast cancer cells (D’Assoro et al., 2002; Schneeweiss et al.,
2003; Guo et al., 2007; Denu et al., 2016; Marteil et al., 2018; Ganapathi Sankaran et al., 2019).
25 A MCF10A MCF7 ZR751 BT20 SUM159 BT-549 MDA-231
Centrioles (centrin) Centrioles
PCM ( ( PCM
γ
Non-amplified
-tubulin)
Amplified 18%
Numberof amplifiedcentrosomes
B 40% C 7.5 6.0 35% **** **** * *** *** *** 30% ** ** 6.0 25% 4.5 **** 20% ** 4.5 15% **** Cells with Cells 3.0 10% * 3.0 5%
centrosome amplification (%) amplification centrosome 1.5 1.5 0% centrioles amplified of Number MCF10A MCF7 ZR751 BT20 SUM159 BT549 MDA-231 MCF10A MCF7 ZR751 BT20 SUM159 BT549 MDA-231 Less Aggressive Highly Aggressive Breast cancer cell lines Breast cancer cell lines D Figure 1.11: Aggressive breast cancer cells exhibit increased CA. (A) Non-amplified and amplified centrioles and centrosomes in breast cancer cell subtypes. Centrioles (centrin, red) and PCM (γ-tubulin, green) are labeled. (B) Percentage of the cell population with cells exhibiting CA. Interquartile range ± highest and lowest observations. (C) Mean number of centrioles (red) and centrosomes (green) in breast cancer cells that have amplified centrosomes. Cell lines are classified from less aggressive to highly aggressive. Statistical tests compare to MCF10A cells. Fischer’s exactγ test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1μm. A 4.0 3.0 **** cell per of centrosomes Number **** 3.5 2.5 **** **** ****
3.0 2.0
2.5 1.5
Number of centrioles per cell cell per of centrioles Number
γ
2.0 1.0 MCF10A MCF7 ZR751 BT20 SUM159 BT549 MDA-231 Breast cancer cell lines FigureB 1.11.1: Supplemental, aggressive breast cancer cells exhibit increased CA. (A) Number of the total centrioles (red) and centrosomes (green) in all cell populations of breast cancer cell lines. Statistical tests compare to MCF10A cells. Fischer’s exact test and Mann-Whitney U test. ****p<0.0005. Scale bars, 1 μm.
26
γ Causes of Centrosome Amplification
CA is commonly detected in breast cancers. Numerous mechanisms are known to cause CA. Some of them include cell cycle aberrations, centriole overduplication, centrosome fragmentation, and de novo centriole assembly. Cell cycle aberrations include events such as cytokinesis failures and mitotic slippage (Fujiwara et al., 2005;
Ganem et al., 2007). Centriole overduplication involves the assembly of multiple daughter centrioles from a single mother centriole (Duensing, 2005; Kleylein-Sohn et al., 2007). In contrast, de novo centriole assembly involves the formation of multiple new centrioles without a mother centriole. Another mechanism entails the formation of multiple new centrioles through deuterosomes (Dehring et al., 2013). Centrosome fragmentation involves the fragmentation of the centriolar and pericentriolar material
(Godinho and Pellman, 2014). The relative contribution of the above mechanisms to CA in breast cancer cells is unclear. My thesis investigates the relative contribution of some of the above mechanisms to CA in breast cancer cells.
Cell Cycle Aberrations
Tetraploidization:
CA can occur through tetraploidization. Tetraploidy through genome instability that promotes CA leads to tumorigenesis in mice models (Fujiwara et al., 2005).
Analysis of about 5000 cancer genome sequences suggests that almost forty percent of human tumors undergo tetraploidy at least once (Zack et al., 2013). Tetraploidy can occur through three main mechanisms. Cytokinesis failures lead to the formation of tetraploid cells that occurs when daughter cells fail to separate the following mitosis.
27 Expression of short cyclin E isoform is often observed in cancer cells. Overexpression of the short cyclin E reduces the length of mitosis, accelerates mitotic exit and promotes cytokinesis failures (Wingate et al., 2009). Similarly, estrogen mediates overexpression of several centrosomal proteins including Aurora A kinase, that initiates dysplasia in the mammary gland (Li et al., 2004). Overexpression of Aurora A kinase promotes cytokinesis failures and tetraploidy in mouse models (Meraldi et al., 2002; Tong et al.,
2004). However, overexpression of Aurora A kinase and centrosomal proteins also impairs the regulation of the centrosome duplication cycle resulting in CA (Meraldi et al.,
2002). This suggests that Aurora A kinase plays a dual role to promote CA.
Mitotic slippage also results in tetraploid cells and accumulation of centrosomes.
It occurs when cells fail to undergo complete mitosis and exit mitosis. Microtubule stabilizing drugs such as taxol affect spindle assembly and activate the spindle assembly checkpoint. This activation promotes mitotic arrest. However, this is accompanied by mitotic slippage which promotes tetraploidization and CA (Godinho and
Pellman, 2014; Tsuda et al., 2017).
One other mechanism involves cell fusion. Cell fusion events occur either spontaneously or through certain viruses (Duelli et al., 2005, 2007). Finally, tetraploid cells can arise through endoreplication. Endoreplication is an event in which two rounds of DNA replication and centrosome duplication occurs without cell division (Krämer et al., 2004). These mechanisms promote CA in which there is an accumulation of both centrioles and pericentriolar material.
28 Centriole Overduplication
Centriole overduplication involves the formation of multiple daughter centrioles from single mother centriole. Every cell cycle entails the formation of precisely two new daughter centrioles from two mother centrioles. One major mechanism underlying centriole overduplication is dysregulation of the centrosome duplication cycle. The levels of core centriolar proteins need to be regulated to control the number of centrioles formed per cell cycle. Deregulation of ubiquitin regulators can be important in deregulation of the centrosome duplication cycle. TrCP, anaphase-promoting complex
(APC/C) and USP33 are some ubiquitin ligases that are implicated in the control of degradation of PLK1, SAS6, STIL and CP110 (Strnad et al., 2007; Cunha-Ferreira et al., 2009; Li et al., 2013). Loss of these ubiquitin ligases may promote the stabilization of the centriolar proteins, leading to centriole overduplication. Similarly, BRCA1 the canonical breast tumor suppressor, associates with its ring domain 1 (BARD1), a ubiquitin ligase that ubiquitinates several centrosomal proteins including -tubulin
(Starita et al., 2004). Loss of BRCA1 is often observed in breast cancer suggesting that it can be important in deregulating the centrosome duplication cycle in breast cancers.
The dysfunctional synergy between these pathways could be important in the assembly of multiple daughter centrioles from a single mother centriole. Centriole overduplication would result in the formation of multiple new daughter centrioles with a corresponding increase in the PCM. This is in contrast to pathways such as centrosome fragmentation that does not involve an increase in the number of centrioles with a corresponding increase in the PCM. These are distinct ways to identify centriole over- duplicated or cell cycle aberrated amplified centrosomes in comparison to centrosome
29 fragmented amplified centrosomes. I will use these distinct mechanisms to identify the underlying mechanism of CA in breast cancer cells.
Consequences of Centrosome Amplification
Centrosomes are the primary microtubule organizing center of most cells. CA is observed in several types of cancers and can be caused by mechanisms such as centriole overduplication, cell cycle aberrations, and other mechanisms. The presence of greater than two centrosomes impairs cellular function through several mechanisms.
Tumorigenesis in humans is a multistep process that occurs due to several genetic assaults. These genetic assaults allow cancer cells to breach the anti-cancer defense mechanisms that are integral to normal cellular functioning. Cancer cells acquire self- sufficiency and insensitivity to signaling pathways, promote tissue invasion and metastasis, and have highly unstable genomes (Hanahan and Weinberg, 2010). The consequences of CA in impairing normal cellular function, in the initiation and propagation of tumorigenesis is discussed below. The direct impact of centriole overduplication and CA on microtubule organization and genome instability is not clear and my thesis investigates the impact of CA in breast cancer cells.
Genome Instability
Bipolar spindle formation is facilitated by the presence of two centrosomes during mitosis, while CA promotes the formation of multipolar spindles during mitosis. One way that multipolar spindles promote genome instability is through multipolar spindle divisions. The cells that undergo multipolar spindle divisions are highly aneuploid with
30 compromised fitness. The loss of fitness affects the viability of these aneuploid cells.
Hence this poses a paradox as to how CA cells are observed in multiple passages in a culture and present throughout the evolution of tumors (Ganem et al., 2009). An alternative mechanism that explains this paradox has been put forth. Centrosomes in multipolar spindles often cluster to form bipolar spindles capable of producing viable daughter cells (Quintyne et al., 2005; Ganem et al., 2009; Godinho et al., 2009). This promotes bipolar division in clustered CA cells. Clustering of amplified centrosomes is known to occur through several microtubule-associated proteins. HSET/KIFC1, a minus-end-directed microtubule motor promotes cross-linking between antiparallel microtubules of two nearby centrosomes(Kwon et al., 2008). Several other proteins that promote tension between kinetochores and spindle microtubules are important for centrosome clustering. Components of the chromosome passenger complex as well as proteins involved in sister chromatid cohesion and kinetochore-microtubule attachments are involved in centrosome clustering. It is hypothesized that the forces that modulate the spindle formation and cortical contractility also control centrosome clustering (Leber,
2010; Rhys et al., 2018).
However, clustering of amplified centrosomes promotes erroneous merotelic microtubule-kinetochore attachments instead of accurate amphitelic kinetochore attachments to chromatids. Such merotelic orientation involves a single kinetochore that is attached to microtubules emanating from both the spindle poles. Such attachments satisfy the spindle assembly checkpoint and progress through into anaphase. However, they give rise to lagging chromosomes during anaphase (Figure 1.12) (Cimini et al.,
2001; Cimini, 2008; Khodjakov and Rieder, 2009).
31 Multipolar mitosis Lagging chromosomes
Nucleus Micronucleus Figure 1.12: Centrosome amplification promotes chromosome segregation errors. CA leads to formation of multipolar mitosis (left panel) and anaphase lagging chromosomes (middle panel). Anaphase lagging chromosomes promote formation of micronuclei (right panel) during interphase.
The anaphase lagging chromosomes are subsequently packaged into micronuclei in the following interphase (Ganem et al., 2009; Godinho et al., 2009;
Silkworth et al., 2009; Thompson et al., 2010; Thompson and Compton, 2011; Ly and
Cleveland, 2017). Micronuclei are sites for extensive chromosome rearrangements,
DNA damage, and varying DNA copy number. Such mutagenesis known as chromothripsis is reminiscent of cancer cells that have extensive chromosome rearrangements and aneuploidy. Several models have been proposed to explain the rearrangements in chromothripsis observed in cancer cells. Recent studies suggest that chromatids in micronuclei undergo fragmentation and reassembly. Such re-assemblies in micronuclei have multiple DNA segment losses. Furthermore, micronuclei fail to normally accumulate several DNA replication and repair proteins (Crasta et al., 2012;
Zhang et al., 2015). This promotes extensive mutagenesis and aneuploidy in cancer cells. Thus CA promotes extensive aneuploidy. The role of aneuploidy in tumor
32 progression is analogous to a double-edged sword (Rajagopalan and Lengauer, 2004;
Ganem et al., 2007). On the one hand, extremely high levels of aneuploidy function to suppress tumor progression through loss of cellular fitness (Schvartzman et al., 2010).
On the other hand, low levels of aneuploidy create a platform for mutagenesis and thereby promote tumorigenesis. Thus, CA promotes chromosome missegregation and aneuploidy.
Centrosome Amplification in Tumorigenesis
Tumor initiation:
CA occurs in early metaplasia and dysplasia that further evolves into invasive carcinomas, suggesting that CA is an early event in tumorigenesis (Lingle et al., 2002;
Kronenwett et al., 2005; Liu et al., 2009). Some of the first studies to test whether CA directly initiates tumorigenesis were not done until recently. When CA was induced through overexpression of PLK4, chronic induction to about 40-fold failed to initiate tumorigenesis (Kulukian et al., 2015; Vitre et al., 2015).
However, when PLK4 was induced to only 1.5 fold, it was sufficient to initiate tumorigenesis (Coelho et al., 2015; Serçin et al., 2016; Levine et al., 2017). There was a corresponding increase in levels of chromosome instability under these conditions.
This is corroboration with the idea that high levels of aneuploidy functions to suppress tumorigenesis. On the other hand, lower levels of aneuploidy promote the formation of tumors.
33 Cell migration and invasion:
Metastasis causes around 90% of human cancer deaths. Invasion is one of the early steps in the metastatic dissemination of cancers. It is a dynamic process where cancer cells release proteases that degrade and remodel the extracellular matrix
(ECM). This promotes the passage of cells through the stroma and entrance into new tissue. During the process of invasion, cancer cells also actively migrate by projecting lamellopodia that attach to the ECM and simultaneously break existing ECM contacts at their lagging edge to pull itself forward. These are some steps essential for cancer cells to invade into the nearby tissues (Friedl and Wolf, 2003).
CA triggers cellular invasion in vitro (Godinho et al., 2014; Pannu et al., 2015).
CA-induced invasion involves degradation of the surrounding ECM and CA cells have noticeable defects in cell-cell junction positioning and length. These phenotypes are similar to that of Rac1 induced cellular invasion. Rac1 activity is upregulated in cells with CA. Rac1 activity disrupts cell-cell junction and promotes cell invasion (Waterman-
Storer et al., 1999; Van Horck et al., 2001; Chang et al., 2008; Godinho et al., 2014).
The underlying mechanism that upregulates Rac1 in CA cells is not known. It is hypothesized that CA affects the surrounding microtubules that control Rac1 activity during the invasion. The release of GEFs (guanine exchange factors) through increased microtubule polymerization from a higher number of microtubules may promote Rac1 activity (Heck et al., 2012). Thus, CA can modulate cell invasion.
Directed cell migration is dependent on force generation that requires remodeling of focal adhesions. Focal adhesions are large protein complexes that connect the inner cytoskeleton to outer ECM (Parsons et al., 2010). Microtubules
34 control focal adhesion dynamics and turnover (Stehbens and Wittmann, 2012). Cancer cells with CA may have a higher number of microtubules (Godinho et al., 2014). This increase in a number of microtubules may affect the focal adhesions turn-over and thereby affect cell motility. Furthermore, CA cells with increased Rac1 activity have altered Arp2/3 mediated polymerization that also affects cell migration (Waterman-
Storer et al., 1999; Van Horck et al., 2001; Chang et al., 2008; Godinho et al., 2014).
Thus CA modulates cell migration.
However, CA is present only in a subset of cells (Figure 1.11). This leads to the question of does the subset of cells with CA affect the entire population. CA induces a paracrine-signaling axis that promotes a non-cell autonomous invasion. The paracrine signaling axis or extra centrosomes associated secretory phenotype promotes invasion through HER2 signaling activation in non-CA cells. However, how triple negative breast cancer cells that have very high CA but lack HER2 receptor would be affected by such paracrine signaling is not clear. Thus, the CA subset of cells has far-reaching consequences and alters the surrounding environment (Arnandis et al., 2018).
Loss of p53 and cell cycle arrest:
An association between CA and p53 loss is frequently observed in several types of cancers (Chan, 2011). CA is observed at a low frequency in wild-type p53 situations.
Loss of p53 facilitates a high frequency of CA (Lopes et al., 2018). In corroboration with this, highly aggressive cancers that have mutant p53 have high CA. However, loss of p53 alone is not sufficient to cause CA (Holland et al., 2012; Lambrus et al., 2015).
When CA is triggered to high levels in wild-type situations, p53 is stabilized which leads to cell cycle arrest (Lopes et al., 2018; Rivlin et al., 2011). Interestingly, both an
35 increase in centrosome number, as well as a decrease in centrosome number, lead to stabilization of p53 (Holland et al., 2012). This suggests that centrosomes number homeostasis and p53 are intricately linked. The underlying mechanism that triggers p53 stabilization during the loss of centrosome number control is still under investigation. It is hypothesized that when CA occurs, the formation of multipolar spindles leads to delayed mitosis. This delayed mitosis triggers a p53 response (Lambrus et al., 2015).
However, the underlying mechanism that triggers p53 upon a mitotic delay is unclear.
Cell and ciliary signaling:
Signaling activation is dependent on reaching a threshold concentration of signaling components. Centrosomes act as scaffolds and control the concentration of various signaling components/mediators such as members of the integrin, BMP, NF-κB and Wnt signaling pathways (Andersen et al., 2003). These pathways are extensively characterized for their contribution to tumorigenesis (Fielding et al., 2008; Itoh et al.,
2009; Kfoury et al., 2008). For example, CA promotes the accumulation of ubiquitin- ligases at centrosomes. Phosphorylated Smad1 mediates BMP signaling.
Phosphorylated Smad1 is targeted to centrosomes for degradation through ubiquitination. CA promotes the accumulation of ubiquitinated phospho-Smad1 at the centrosomes and promotes its degradation. This may impair the Smad1 mediated BMP signaling. Similarly, Diversin, an inhibitor of the Wnt pathway, functions when it localizes to centrosomes. CA may promote extended inhibition of the Wnt signaling pathway and compromise these signaling pathways (Fuentealba et al., 2007; Itoh et al., 2009).
In addition to the accumulation of more signaling molecules through CA, the presence of more centrioles can promote the formation of extra cilia. Formation of extra
36 cilia leads to reduced concentration of signaling molecules in cilia. The Shh pathway that signals through Smo on the cilia can be affected by the presence of extra cilia.
Extra cilia may dilute the ciliary concentration of Smo. This would result in altered ciliary signaling (Mahjoub and Stearns, 2014). Thus, on the one hand, when CA occurs there is an accumulation of more molecules that influences signaling pathways. Accumulation of extra centrioles may also promote the formation of extra cilia. On the other hand, extra cilia may dilute the ciliary concentration of signaling molecules and impair the balance of these signaling pathways, a condition often observed in cancer cells
(Hanahan and Weinberg, 2010). Thus CA can impair signaling pathways.
Asymmetric cell division and cell polarity:
Cancer stem cell theory hypothesizes that a subset of cancer cells with stem cell characteristics is responsible for tumor growth (Reya et al., 2001). Asymmetric cell division underlies the unequal segregation of cell fate determinants and is important in creating a balance of stem cells and differentiated cells. Impaired asymmetric cell division and segregation of cell fate determinants can promote loss of polarity, hyperproliferation of cancer stem cells and tumorigenesis. Centrosomes establish cell polarity through proper microtubule organization. Centrosome position and its surrounding microtubule organization dictate proper and polarized secretion of lytic granules. Accumulation of centrosomes impairs polarity and asymmetric division and is sufficient to initiate tumorigenesis (Caussinus et al., 2005). This may occur through hyper-proliferation of stem-like cells from impaired asymmetric cell division leading to tumorigenesis (Caussinus et al., 2005).
37 Thesis Outline
Taken together, centrosome number homeostasis is very tightly maintained to regulate normal cellular function. CA is detected in several developmental and disease contexts. The CA burden is very high in solid tumors like breast cancer. The underlying mechanism that generates CA in breast cancer cells is not clear. Centriole overduplication, cell cycle aberration, de novo centrosome formation, and centrosome fragmentation can all cause CA. My thesis investigates the relative contribution of centriole overduplication to CA in breast cancer cells. In Chapter II, I examine centriole overduplication in CA in breast cancer cells. Core centriole assembly proteins such as
PLK4, SAS-6, and STIL are tightly maintained along the cell cycle to prevent centriole overduplication in normal cells. How the core centriole assembly protein CEP135 is regulated in normal cells is not known. In this thesis, I investigate the regulation of
CEP135 in normal cells and ask whether dysregulation of CEP135 contributes to CA in breast cancer cells. CA can initiate and promote various hallmarks of cancer.
Centrosomes are microtubule organizing centers, yet how CA impacts microtubule organization is not well studied. In Chapter III, I investigate the consequences of CA to microtubule organization in breast cancer. To characterize the microtubule organization, we developed a computational tool that facilitates this analysis. Furthermore, CA promotes genome instability in cancer cells. I ask how centriole overduplication impacts genome instability through CEP135. In summary, I investigate the contribution of centriole overduplication to CA, the role the core centriole assembly protein CEP135 plays in this process, and the consequences of CA on microtubule organization and genomic stability in breast cancer cells.
38 Materials and Methods for Chapter I
Cell Culture
Mammalian tissue culture lines were all grown at 37°C and 5% CO2. MCF10A cells were grown with DMEM/F12 (Invitrogen #11330-032), 5% Horse Serum (Invitrogen
#16050-122), EGF (Invitogen #PHG0311), Hydrocortisone (Sigma #H-0888), Cholera toxin (Sigma #C-8052), Insulin (Sigma #I-1882) and Pen/Strep (Invitrogen #15070-063).
MDA-MB-231(MDA-231), BT-20 and MCF-7 cells were grown with DMEM (Invitrogen
#11965-092), Penicillin/Streptomycin (Invitrogen #15070-063) and 10% FBS (FBS;
Gemini Biosciences). ZR-751, BT-549 cells were grown with RPM1 (Invitrogen #11875-
093), 10% FBS (FBS; Gemini Biosciences) and Pen/Strep (Invitrogen #15070-063).
SUM-159.PT (SUM159) were grown with Hams media (Invitrogen #11765054),
Hydrocortisone (Sigma #H-0888), Penicillin-Streptomycin (Invitrogen #15070-063) and
10% FBS (FBS; Gemini Biosciences). Cells were passaged and sub-cultured using
Trypsin (Invitrogen #150901-046) when cultures reached 60-80% confluency.
Immunofluorescence
12 mm diameter coverslips were acid-washed and heated to 50°C in 100mM HCl for approximately 16 hours. This was followed by washes with water, 50%, 70%, and
95% ethanol for 30 minutes each. Coverslips were coated with type- 1 collagen (Sigma
# C9791), air-dried for 20 minutes in the laminar hood and exposed to UV light for cross-linking of collagen for 20 minutes. Cells of interest were then cultured on collagen-coated coverslips until 55-70% confluence.
For centrosome immunofluorescence, cells were fixed with 100% methanol at
−20°C for 8 minutes. Fixed cells were washed with PBS/Mg (1x PBS, 1mM MgCl2), and
39 then blocked with Knudsen Buffer (1x PBS, 0.5%BSA, 0.5% NP-40, 1mM MgCl2, 1mM
NaN3) for 1 hour. Cells were incubated overnight with primary antibodies diluted in
Knudsen Buffer at 4oC. Coverslips were washed with PBS three times in 5-minute intervals. Secondary antibodies and Hoechst 33258 (10μg ml−1, Sigma #B2261) were diluted in Knudsen buffer and incubated for 1 hour at room temperature. Coverslips were mounted using Citifluor (Ted Pella) and sealed with clear nail polish.
Microscopy
The fluorescence imaging utilized is identical to those described in Dahl et al,
2015. Briefly, images were acquired using a Nikon Ti (Nikon Instruments, Inc.) inverted microscope stand equipped with a 100X PlanApo DIC, NA 1.4 objective. Images were captured using an Andor iXon EMCCD 888E camera or an Andor Xyla 4.2 CMOS camera (Andor Technologies).
Nikon NIS Elements imaging software was used for image acquisition. Image acquisition times were constant within a given experiment and ranged from 50 to 400 msec, depending on the experiment. All images were acquired at room temperature.
Images presented in most of the figures are maximum-intensity projections of the complete z-stacks.
Centriole and Centrosome Number Counts
Cells were scored as amplified and non-amplified based on centrin and γ-tubulin staining (Dahl et al., 2015). Cells with greater than two γ-tubulin and four centrin foci were scored as amplified centrosomes. Non-amplified centrosomes possess both one or two γ-tubulin and two or four centrin foci.
40 Statistics and Biological Replicates
All center values represent means and error bars represent the standard error of the mean except for Figure 1.10 boxes which represent the mean and interquartile range while vertical lines indicate the highest and lowest observations. All the experiments were performed using at least three independent biological replicates.
Figure 1.10 A-C used >5 biological replicates, respectively. The number of cells used in each immunofluorescence experiment is as follows: Figure 1.10B: >100 cells per condition/700 cells, 1.10C: >100 cells per condition/700 cells. Fischer’s exact test,
Student’s two-tailed t-test, and Mann-Whitney U-test were used to assess statistical significance between means.
Fischer’s test was utilized to examine the significance of contingency when data were classified into two or more categories. Student’s two-tailed unpaired t-test was used to examine significance between two normal distributions (equal variance assumed). Normality tests were performed both on the raw data and meta-data extracted from the replicates of raw data. Shapiro-Wilk normality test and D'Agostino-
Pearson omnibus normality test was utilized to examine normality of data. Shapiro-Wilk normality test was used when the number of samples was less than eight. When the number of samples was greater than eight, the D'Agostino-Pearson omnibus normality test was used. Mann-Whitney u-test was utilized to examine the significance of non- normal distributions. Results were considered statistically significant with p-values less than 0.05. P-values were denoted on figures according the following values: * p<0.05, ** p<0.01, *** p<0.005 and **** p<0.0005.
41 CHAPTER II
CEP135 ISOFORM DYSREGULATION PROMOTES CENTROSOME
AMPLIFICATION IN BREAST CANCER CELLS2
Introduction
Centriole duplication begins during late G1 phase when PLK4 and STIL concentrate at an asymmetric site on the mother centriole wall (Fırat-karalar and
Stearns, 2014; Ohta et al., 2014). Core centriole assembly factors SAS-6, CEP135 and
CPAP then self-assemble at this site during S phase into a structure known as the cartwheel (Ohta et al., 2002; Carvalho-Santos et al., 2010; Lin et al., 2013; Fırat-karalar and Stearns, 2014; Hirono, 2014; Dahl et al., 2015). The levels of these core centriolar proteins are regulated to limit the formation of one new daughter centriole per mother centriole. SAS-6 and STIL levels are regulated through the cell cycle by proteasome- dependent degradation during mitosis. This mitotic degradation of SAS-6 and STIL promotes timely and controlled duplication of centrioles during S-phase. CEP135 has a unique and distinct regulation. At least two isoforms of CEP135 are transcribed from its single gene locus. The CEP135 gene locus encodes the above discussed full-length
1140 amino acid, 134KDa CEP135, CEP135full.CEP135full is necessary and promotes centriole duplication (Ohta et al., 2002; Kleylein-Sohn et al., 2007; Dahl et al., 2015).
2 Portions of this chapter are published with permission from our previously published articles:
1) Ganapathi Sankaran, D, Stemm-Wolf, AJ, and Pearson, CG (2019). CEP135 isoform dysregulation promotes centrosome amplification in breast cancer cells. Mol Biol Cell.
2) Dahl, KD, Ganapathi Sankaran, D, Bayless, BA, Pinter, ME, Galati, DF, Heasley, LR, Giddings, TH, and Pearson, CG (2015). A Short CEP135 Splice Isoform Controls Centriole Duplication. Curr Biol 25, 2591–2596.
42 The CEP135 C-terminus physically links SAS-6 spokes while the N-terminus connects microtubules and CPAP in the cartwheel (Figure 2.1).
There exists a short isoform of CEP135 known as CEP135mini. CEP135mini, a 249 amino acids or 29kDa protein has the same N-terminus as CEP135full but contains distinct C terminus (Figure 2.1). In contrast to CEP135full, CEP135mini contains only the complete microtubule binding domain. CPAP binding SAS-6 binding
Both of the CEP135 isoforms localize CEP135full N 1140 aa-134kDA C to centrosomes but have distinct Microtubule binding mini CEP135 N 249aa-29kDA C localization patterns at the Microtubule binding full centrosome. CEP135 localizes to Figure 2.1: Schematic of CEP135 isoforms. CEP135full binds to pinhead hook of the cartwheel microtubules and CPAP through the N- terminus. CEP135full’s C-terminus binds structure at the centrioles throughout to SAS-6. CEP135mini has the same N- terminus as that of CEP135full but has the cell cycle. In contrast, divergent C-domains.
CEP135mini’s localization pattern changes through the cell cycle. During the G1 phase of the cell cycle, CEP135mini localizes predominantly to the proximal end of the centriolar barrel whereas in the G2 phase it localizes both to centrioles and to the PCM (Dahl et al., 2015).
Interestingly, CEP135mini functions as a dominant negative and represses centriole duplication. The underlying mechanism through which CEP135mini represses centriole duplication is not known. A simple explanation is that CEP135mini functions as a dominant-negative molecule to remove CEP135full from the centriole. Overexpression of either CEP135 isoform does not disrupt the localization of the other, suggesting that
CEP135full is able to localize to centrioles even when CEP135mini levels are high. Hence,
43 the negative effect of CEP135mini on centriole duplication cannot be attributed to
CEP135mini disrupting the ability of CEP135full to localize at centrioles. An alternative possibility is that overexpressed CEP135mini disrupts centriole duplication by binding to
CEP135full and precluding its association with its binding partners (SAS-6 and CPAP).
Ectopic expression of CEP135 isoforms suggests CEP135mini is competent to associate with CEP135full(Dahl et al., 2015). However, whether CEP135mini’s expression indeed impairs CEP135full’s association with its binding partners (SAS-6 and CPAP) is not known.
The opposite functions of the CEP135 isoforms in centriole duplication suggest that their centriolar levels are modulated through the cell cycle to promote centriole assembly only during the G1-S phase transition. Consistent with this hypothesis,
CEP135full transcript and protein levels are high during centriole duplication at the G1/S phase boundary. In corroboration with the role for CEP135mini in repressing centriole assembly, CEP135mini protein levels are lowest during G1/S when centrioles duplicate.
After centriole duplication, the centriolar levels of CEP135mini increase and peak at metaphase of mitosis presumably to prevent promiscuous re-duplication of centrioles
(Dahl et al., 2015). CEP135mini levels drop after metaphase presumably to license the next round of centriole duplication.
Taken together, the levels of the core centriolar proteins and the number of cartwheels formed at the mother centriole during the G1/S transition determines the number of daughter centrioles that will form (Carvalho-Santos et al., 2010; Hirono,
2014). In normal cells, centriole assembly factors are tightly regulated such that the two existing centrioles give rise to a total of two daughter centrioles. Many breast cancer
44 cells harbor excess centrioles, yet whether centriole assembly dysregulation is primarily responsible is not well understood.
The chromosomal locus of the centriole assembly factor CEP135 (4q12) is amplified and mutated in aggressive breast cancer patient samples, making CEP135 a candidate oncogene (Martinho et al., 2009; Yu et al., 2009; Johansson et al., 2011;
Tuupanen et al., 2014). The antagonistic functions of these two CEP135 isoforms suggest that their dysregulation may contribute to the increase in centriole numbers observed in breast cancer cells. I explored the level of CEP135 dysregulation in breast cancer cell lines and found that an abundance of the CEP135full isoform relative to the
CEP135mini isoform may contribute to centriole overduplication and CA.
Understanding the underlying mechanism that allows the formation of the
CEP135mini mRNA variants could provide insights as to how the CEP135 isoforms are regulated in breast cancer cells. The CEP135 gene contains 26 exons in which all the exons comprise the CEP135full mRNA resulting in 1140 amino acid protein. In contrast,
CEP135mini Exon skipping AS mRNA is
Alternative 5’ Splice Site comprised Selection of the first 6 Intron retention AS exons and at least part Common exons Alternating exons of intron 6 Figure 2.2: Some alternative splicing models: Exon skipping, alternative splice site selection and intron retention. where an in- frame stop codon terminates translation. This leads to a protein that is identical to the
45 first 233 amino acids of CEP135full but then harbors a unique 16 amino acid tail at its C- terminus (Dahl et al., 2015). CEP135mini mRNA was reported to be an alternative splice variant based on database reports (Dahl et al., 2015), however, the biogenesis of these isoforms had not been experimentally tested. Alternative transcription initiation, alternative splicing (AS) and alternative polyadenylation (APA) are some mechanisms that lead to diversity in mRNA variants. Alternative splicing involves the precise excision of introns and the concomitant joining of exons in the pre-mRNA to generate the mature mRNA. Some of the alternative splicing models that result in mRNA variants include exon skipping, alternative 5’ or 3’ splice site selection and intron retention (Matlin et al.,
2005) (Figure 2.2). The spliceosome, a macromolecular ribonucleoprotein machinery, assembles on the pre-mRNA to mediate alternative splicing in coordination with numerous splicing factors (Matera and Wang, 2014).
In contrast, alternative polyadenylation involves cleaving and polyadenylation of
Proximal poly(A) Distal poly(A) the pre-mRNA Tandem 3’UTR APA
at an Proximal poly(A) Distal poly(A) alternative poly Alternative terminal exon APA
(A) sites. Some Proximal poly(A) Distal poly(A)
Intronic APA of the common alternative Common exons Extended CDS Alternating exons 3’UTR polyadenylation Figure 2.3: Some alternative polyadenylation models: Tandem models include 3’UTR APA, Alternative terminal exon APA and intronic APA. tandem 3’ UTR APA, alternative terminal exon APA, internal exon APA and intronic
APA. Tandem 3’UTR APA isoforms have the same coding sequences, while alternative
46 terminal exon APA and intronic APA isoforms have different coding sequences.
Alternative terminal exon APA involves usage of alternative terminal exons and intronic
APA involves cleaving at intronic poly (A) site while extending the coding sequence of an internal exon and making it the terminal one (Elkon et al., 2013) (Figure 2.3). The 3’ cis-elements with the 3’ processing machinery control the alternative cleavage and polyadenylation of mRNA variants(Nunes et al., 2010; Di Giammartino et al., 2011).
Genome-wide gene expression studies and databases would normally distinguish between different mRNA biogenesis mechanisms by identifying the location of transcription initiation and termination. Transcription of CEP135full and CEP135mini mRNA initiates from the same promoter and both isoforms begin translation from the same start codon (genome.ucsc.edu and ensembl.org). However, the CEP135mini mRNA that reads intron 6 has a distinct C terminus and 3’ end. Intron 6 of CEP135 contains a genomically encoded poly(A) tract and is an internal priming candidate in studies that utilize oligo d(T) to map mRNA 3’ ends. Database annotations do not accurately reflect the transcription termination for mRNAs from such genes (Nam et al.,
2002). Thus, it is unclear where the CEP135mini mRNA ends and whether the
CEP135mini mRNA variant results from alternative splicing with intron 6 retention or alternative polyadenylation within intron 6.
I investigated the transcriptional termination of CEP135mini isoform and my data support a model that CEP135mini is generated by alternative polyadenylation and is affected by nucleotide sequences within intron 6.
47 Results
Centriole Overduplication contributes to Centrosome Amplification in Breast Cancer
Cells
To investigate whether centriole overduplication contributes to CA in breast cancer cells, I examined centrioles in amplified centrosomes. Centriole proteins
(CEP135, CPAP, CEP192, CEP152, and CEP170) were visualized to discern whether the centrioles were complete. Most amplified centrioles are mature and contain a complete complement of centriole proteins (Figure 2.4A and 2.4.1A). To quantify the frequency of centriole overduplication, I calculated the percentage of cells with more than two centriole foci containing SAS-6, which is present only in daughter centrioles until the cartwheel disassembles in late mitosis. Normally, G1 phase centrioles are devoid of SAS-6 foci while S, G2 and early mitotic phase cells have two SAS-6 foci
(Figure 2.4B). Compared to MCF10A cells, a larger proportion of ZR751 and MDA-231 cells have more than two daughter centrioles (Figure 2.4C). Additionally, the number of newly formed centrioles in MDA-231 cells with CA is greater than that for MCF10A and
ZR751 cells (Figure 2.4D and 2.4.1B). Thus, amplified centrosomes in breast cancer cells contain excess newly formed centrioles. I next measured the frequency of single mother centrioles giving rise to more than one daughter centriole. Of the MDA-231 cells that have more than two SAS-6 foci, 2% have multiple SAS-6 foci associated with a single mother centriole (Figure 2.4E and 2.4F). 98% of these cells have only one daughter centriole per mother centriole. These data suggest that the formation of multiple daughter centrioles from a single mother centriole is detectable but infrequent and that the amplified state is propagated by centriole assembly.
48 A
MDA-231 A SAS-6 CPAP CEP135 B SAS-6 phase
G1
Centrioles
Non-amplified Non-amplified
<2 new centrioles
C1AZ71MDA-231 ZR751 MCF10A
Early S
phase
Centrioles
Non-amplified Non-amplified
Late S phase
Amplified Amplified
Centrioles
>2 new centrioles
Amplified Amplified
Centrioles
CEP192 Centrin CEP192 Centrin
Multiple daughters Single daughter C DEFper mother per mother 15% *** 7 100%
6 entrin 12% **** C 75% 5 9% 4 50% 6% 3
CEP192 CEP192
Frequency of centrioles (%) 2 25%
3% SAS-6 positive)
formed per amplified cell amplified per formed 1 duplicating centrioles (%)
SAS-6 SAS-6 New ( New centrioles 0%
Ce lls with >2 ne (SAS-6 w positive) 0% 0 le le r 1 ip rs g e A 51 3 1 1 lt e in t 0 7 2 A 5 3 u t S h - 10 7 2 h g F1 ZR A F R - M g u C D Z A u a M M C D a d M M d
Figure 2.4: Centriole overduplication in breast cancer cells that have CA.
49 Figure 2.4: Centriole overduplication in breast cancer cells that have CA. (A) Non- amplified and amplified centrioles have a full complement of representative centriolar proteins. SAS-6 (left, grayscale), CPAP (middle, grayscale) and CEP135 (right, grayscale) are labeled relative to CEP192 (red) and centrin (green) and imaged using structured illumination microscopy (SIM). (B) New centriole assembly at non-amplified and amplified centrioles. Left panel, centrioles labeled for SAS-6 (grayscale), CEP192 (red) and centrin (green). Right panel, schematic of G1-, early, late S-phase and amplified centrioles. Arrows denote multiple new, SAS-6 positive procentrioles. (C) Percentage of the total cell population that have >2 new (SAS-6 positive) daughter centrioles. (D) Number of new centrioles (SAS-6 positive) assembled in cells that have amplified centrosomes. (E) Multiple and single daughter centrioles form from mother centrioles. Top panel, schematic of multiple and single daughter centrioles forming from mother centrioles. Bottom panel, representative images of centrioles labeled for SAS-6 (grayscale), CEP192 (red), and centrin (green) in S-phase MDA-231 cells that have over-duplicated centrioles. Arrows denote multiple daughter centriole assembly events at a single mother centriole. (F) Relative frequency of duplication of multiple daughter centrioles compared to a single daughter centriole from a single mother centriole. (C-D) Statistical tests compare to MCF10A cells. Mean±SEM. Fischer’s exact test and Mann- Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
50 Meg r e A SAS-6 CEP170
CPAP GT335 Amplified Centrosomes
CEP135 CEP170
MDA-231
CNAP CEP152
B 75% *
60%
45%
30%
Amplifiedcells with >2 15%
new(SAS-6 po sitive)centrio (%) les 0% A 1 1 0 5 3 1 7 -2 F R A C Z D M M
Figure 2.4.1: Supplemental, centriole overduplication in breast cancer cells with amplified centrosomes. (A) Both non-amplified and amplified centrioles contain a full complement of representative centriole proteins. Panels from top to bottom are centrioles stained for SAS-6, CEP170; CPAP, glutamylated tubulin (GT335); CEP135, CEP170; CNAP and CEP152. (B) Percentage of cells with amplified centrioles that show new centriole overduplication (based on SAS-6 foci). Mean±SEM. Statistical tests compare to MCF10A cells. Student’s t-test *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bar, 1 μm.
51 The CEP135full:mini Ratio is Elevated in Centrosome Amplified Breast Cancer Cells
It is not clear how centriole duplication is dysregulated to increase the frequency of cells with amplified centrosomes in the cell population. CEP135 is a centriole duplication factor whose isoforms, CEP135full and CEP135mini, perform opposing functions in controlling centriole assembly. CEP135full is important for cartwheel formation and promotes centriole assembly, while CEP135mini represses centriole assembly (Kleylein-Sohn et al., 2007; Dahl et al., 2015). Moreover, the chromosomal locus containing CEP135 (4q12) has an eight-fold copy gain in aggressive breast cancer patient datasets, suggesting that expression of CEP135 isoforms are elevated
(Finak et al., 2008; Martinho et al., 2009; Yu et al., 2009; Johansson et al., 2011; Glück et al., 2012; Tuupanen et al., 2014). CEP135full mRNA levels are also elevated in aggressive breast cancer cell lines (Figure 2.5.1A; (Neve et al., 2006; Barretina et al.,
2012)). This led us to ask how the two CEP135 mRNA isoforms with opposing functions are regulated in centrosome-amplified breast cancer cell lines.
The CEP135 gene contains 26 exons in which all the exons comprise the
CEP135full mRNA resulting in 1140 amino acid protein. In contrast, CEP135mini mRNA is comprised of the first 6 exons and at least part of intron 6 where an in-frame stop codon terminates translation. This leads to a protein that is identical to the first 233 amino acids of CEP135full but then harbors a unique 16 amino acid tail at its C-terminus (Dahl et al., 2015). To evaluate the levels of these two CEP135 isoforms in breast cancer cells, I detected each isoform by RT-PCR using primers directed to unique mRNA sequences (Figure 2.5A). CEP135full mRNA levels are elevated in MDA-231 cells relative to MCF10A cells. Although CEP135mini mRNA levels are also upregulated, it is to
52 a lesser degree than CEP135full mRNA levels (Figure 2.5A and 2.5B). Given the opposing functions of the two isoforms, I hypothesize that the relative levels of these two isoforms are important for regulating centrosome number in breast cancer cells. To examine this, the relative mRNA levels of the CEP135full and CEP135mini isoforms
(CEP135full:mini transcript ratio) were quantified in breast cancer cell lines (Figure 2.5.1B and 2.5.1C). The CEP135full:mini ratio is greater in aggressive breast cancer cell lines
(Figure 2.5C). Furthermore, the CEP135full:mini ratio correlates with the percentage of CA observed in these breast cancer cells (Pearson coefficient = 0.8745; Figure 2.4D).
To test whether the relative CEP135full and CEP135mini protein levels reflect the increased transcript ratios in breast cancer cells, CEP135full- and CEP135mini-specific antibodies were used to measure the fluorescence intensities of CEP135full and
CEP135mini at centrosomes. Consistent with transcript levels, CEP135full protein levels at centrosomes are elevated in MDA-231 cells relative to MCF10A cells (Figure 2.5E,
2.5G, 2.5.1D, and 2.5.1F). CEP135mini protein levels at centrosomes are reduced in
MDA-231 cells relative to MCF10A cells, despite a slight elevation in CEP135mini mRNA levels in MDA-231 cells (Figure 2.5F, 2.5G, 2.5.1E, and 2.5.1G). Surprisingly, the fluorescence intensity of both CEP135full and CEP135mini is lower in ZR751 cells relative to MCF10A cells (Figure 2.5E-G and 2.5.1D-G). However, consistent with the transcript ratios, the CEP135full:mini protein ratio is elevated in both ZR751 and MDA-231 cells compared to MCF10A cells (Figure 2.5G). These data indicate that the CEP135full:mini protein ratio is elevated at centrosomes in breast cancer cells.
53
Figure 2.5: CEP135 isoform transcript and protein levels are altered in breast cancer cells.
54
Figure 2.5: CEP135 isoform transcript and protein levels are altered in breast cancer cells. (A) Top panel, CEP135full, and CEP135mini genes. Red and green bars denote coding exons (Ex) of CEP135full and CEP135mini, respectively. Blue bars denote non- coding exons. Gray lines denote introns (In). F1 and R1 (red), F1 and R2 (green) denote forward and reverse primers for CEP135full and CEP135mini, respectively. Black pentagons represent stop codons for translation termination. Bottom panels, CEP135full, CEP135mini and control (GUSB) RT-PCR in breast cancer cells. (B) The relative CEP135full (red) and CEP135mini (green) transcript levels and the CEP135full:mini (black) ratio. (C) The CEP135full:mini transcript ratio in breast cancer cells. (D) Linear regression fit and Pearson coefficient (R) of the CEP135full:mini transcript ratio relative to the percentage of CA cells in the breast cancer cell populations. (E) CEP135full (red) and centrin (grayscale) at one centrosome of G2 phase cells. (F) CEP135mini (green) and centrin (grayscale) at one centrosome of G2 phase cells. (G) Centrosomal CEP135full (red) and CEP135mini (green) protein fluorescence intensities per centrosome of G2 phase cells and their ratios (black). (B, C, and G) Statistical tests compare to MCF10A cells. Mean±SEM. Student’s t-test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
55 CEP135full relative A transcript reads Invasiveness B MCF10A MCF7 BT20SUM159 ZR751 BT549 MDA-231
MDAMB415 full CEP135 full AU565 Cep135 Luminal ZR7530 min i BT474 CEP135 SKBR3 BT20 Basal A GUSB MDAMB453 Luminal BT549 Basal B 3.5 CEP135full MCF7 mini CAMA1 C CEP135 *** 3.0 CEP135full:mini ZR751 T47D 2.5 MDAMB361
Breast cancer cell lines cell cancer Breast MDAMB231 *** *** 2.0 *** MDAMB157 **** HCC1500 Basal B ** ** 1.5 *** MDAMB436 *
transcript levels 1.0
200 300 400 500 Relativenormalized CEP135 0.5 MCF10A MCF7 ZR751 BT20 SUM159 BT549 MDA-231 Breast cancer cell lines D E MCF10AZR751 MDA-231 MCF10AZR751 MDA-231
CEP135 full Centrin DNA CEP135 mini Centrin DNA
FH6 2.5 ** ** **** 5 2.0 4 1.5 ** ** 3 1.0
full 2 min i 1 0.5
CEP135 centrosome per
CEP135 centrosome per 0 0.0
Relative fluorescence intensity of intensity fluorescence Relative MCF10A ZR751 MDA-231 of intensity fluorescence Relative MCF10A ZR751 MDA-231
Figure 2.5.1: Supplemental, CEP135 transcript and protein levels are altered in breast cancer cells. (A) Comparison of CEP135full transcript reads and invasiveness in various breast cancer cell lines (Adapted from (Neve et al., 2006; Barretina et al., 2012)). (B) CEP135full, CEP135mini, and GUSB (control) RT-PCR in breast cancer cells. (C) CEP135full (red) and CEP135mini (green) transcript levels normalized to GUSB levels and represented relative to their corresponding transcript levels in MCF10A cells. The CEP135full:mini transcript ratio (black) in breast cancer cells is also represented relative to MCF10A cells. (D) CEP135full (red) and centrin (green) at centrosomes of G2 phase MCF10A, ZR751, and MDA-231 cells. Corresponding images in Figure 3E. (E) CEP135mini (green) and centrin (red) at centrosomes of G2 phase MCF10A, ZR751, and MDA-231 cells. Corresponding images in Figure 3F. (F) Quantification of CEP135full protein fluorescence intensity per G2 phase centrosome relative to MCF10A cells. (G) Quantification of CEP135mini protein fluorescence intensity per G2 phase centrosome relative to MCF10A cells. (C,F,G) Statistical tests compare to MCF10A cells. Mean±SEM. Student’s t-test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
56 Elevated CEP135full is Sufficient to Increase Centrosome Amplification in Breast Cancer
Cells
To examine the effects of elevated CEP135full expression in breast cancer cells, I engineered a stable tetracycline-inducible CEP135full MDA-231 cell line. Tetracycline treatment promotes the exogenous expression of fluorescently labeled mCherry-
CEP135full (hereafter mCh-CEP135full-Tet). I treated mCh-CEP135full-Tet MDA-231 cell lines with tetracycline for three days and quantified the number of centrosomes in these cells. The non-induced mCh-CEP135full-Tet cells had a reduced frequency of cells in the cell population with CA (14%) relative to wild-type MDA-231 cells (23%; Figure 2.6.1A).
I hypothesize that clonal selection of the mCh-CEP135full-Tet transfected MDA-231 cells selected for cells with reduced CA, thereby altering the homeostasis of CA. Tetracycline induces a 2.5-fold increase in mCh-CEP135full fluorescence intensity at centrosomes when compared to the non-induced cells (Figure 2.6A and 2.6.1B), and an approximately two-fold increase in the number of cells with CA (Figure 2.6D and 2.6E).
The number of centrioles and centrosomes within the population of cells with amplified centrosomes was also greater in the induced mCh-CEP135full-Tet cell line (Figure 2.6E).
This suggests that elevated mCh-CEP135full-Tet expression in breast cancer cells promotes new centrioles and centrosomes.
To measure the formation of new centrioles in the induced mCh-CEP135full-Tet cell line, I visualized mCh-CEP135full, SAS-6 and CPAP. Surprisingly, mCh-CEP135full localizes not only to the centriole proximal end but also decorates the walls of some of the centrioles (Figure 2.6B and 2.6.1C). Induced mCh-CEP135full-Tet cells have an increased number of SAS-6 positive centrioles relative to the non-induced cells (Figure
57 2.6C). Furthermore, multiple SAS-6 foci per mother centriole were observed (Figure
2.6.1D). However, it is not clear if all SAS-6 foci represent daughter centrioles or if overexpressed CEP135full can stabilize SAS-6 at mother centrioles. Regardless, these data suggest that elevated CEP135full levels in breast cancer cells are sufficient to
increase centriole number and CA.
A B C SAS-6 CPAP 15
full 3.0 **** 12
full
2.5 mCh- ****
markers
Centriole Centriole
CEP135 2.0 9
full 1.5 6 1.0 mCh-
CEP135
CEP192 3 0.5
intensity mCh-CEP135 intensity
Number of new (SAS-6 positive) (SAS-6 new of Number
e g
0.0 e g amplified centrioles formed per cell cell per formed centrioles amplified 0
Relative centrosomal fluorescence centrosomal Relative
M r e - + M r e - + Tetracycline induction Tetracycline induction mCh- 40% full -tubulin Mger e 9.0 9.0 D CEP135 Centrin γ Non-amplfied E
centrosome
35% **** Number of amplified 7.5 7.5
30% centrosomes
- 25% 6.0 6.0
20%
centrosome 4.5 4.5
centrioles centrioles
Cells with Cells
Amplified 15% 3.0 3.0
+ + 10% amplified of Number
Tetracycline induction Tetracycline
centrosome amplification (%) amplification centrosome 5% 1.5 1.5 - + - + Tetracycline induction Tetracycline induction F Figure 2.6: Supplemental, elevated CEP135full increases breast cancer cell CA. (A) Top panel, schematic and timeline of exogenous mCh-CEP135Full-Tet expression in
MDA-231γ cells. Bottom panel, relative centrosomal fluorescence intensity of mCh- CEP135full in non-induced and induced mCh-CEP135full-Tet MDA-231 cells. (B) SIM localization of mCh-CEP135full (red) and CEP192 (green) at centrioles in mCh- CEP135Full-Tet cells. (C) Left panel, amplified centrioles in mCh-CEP135 full-Tet cells. mCh- full-Tet CEP135 γ (red) cells were colocalized with SAS-6 (green) or CPAP (green). Right panel, mean number of new, daughter centrioles (SAS-6 positive) in non-induced and induced mCh-CEP135full-Tet CA MDA-231 cells. (D) Centrosomes in non-induced and induced mCh-CEP135 full-Tet cells. mCh-CEP135full-Tet (red) cells were stained for centrin (grayscale) and γ-tubulin (green). (E) Left panel, the percentage of cells with CA in non- full-Tet
inducedγ and induced mCh-CEP135 MDA-231 cells. Right panel, number of amplified centrioles (red) and centrosomes (green) in non-induced and tetracycline- induced mCh-CEP135full-Tet MDA-231 cells displaying CA. Mean±SEM. Fischer’s exact test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
58
A B Tetracycline induction of mCh-CEP135full Cell line CA Wild-type MDA-231 cells 23% Non-induced mCh-CEP135full-Tet 14% cells
Day 3
full C 15 D E full ****
f ull
mC h-
mCh-
12 CEP135
CEP135
9
SAS-6
6 CEP192
3
intensityo f mCh-CEP135
e g
e g
Relativenormalized fluorescence M r e
0 M r e - + Tetracycline induction
Figure 2.6.1: Supplemental, elevated CEP135full expression increases centriole number and CA in breast cancer cells. (A) CA frequency in control MDA-231 and non-induced mCh-CEP135Full-Tet MDA-231 cell lines. (B) Schematic of tetracycline-induced mCh- CEP135Full-Tet MDA-231 cells. (C) Relative normalized fluorescence intensity of mCh- CEP135full in non-induced and induced mCh-CEP135full-Tet MDA-231 cells. (C) mCh- CEP135full (red) localization relative to CEP192 (green). Arrows denote centrioles with mCh-CEP135full at the cylinder walls. (D) mCh-CEP135full (red) localization relative to SAS-6 (green). Arrows denote SAS-6 foci at the core cylinder. Mean±SEM. Mann- Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
59 Elevated CEP135mini Limits the Centriolar Levels of Essential Duplication Factors
CEP135full promotes centriole biogenesis and facilitates the assembly of the cartwheel. Centriole biogenesis is initiated by the assembly of the inner cartwheel that is, in part, dependent on SAS-6. The CEP135full C-terminus physically links SAS-6 spokes while the N-terminus connects microtubules and CPAP in the cartwheel.
CEP135mini represses centriole duplication. To establish whether CEP135mini affects the loading and assembly of centriole components that are fundamental to assembly, I determined the effect of CEP135full and CEP135mini expression on the localization of centrin, SAS-6, and CPAP. The expression of either CEP135 isoform had no effect on the centriolar levels of centrin. However, expression of CEP135full and CEP135mini causes a 43% and 44% decrease in SAS-6 levels at centrioles, respectively (Figure
2.7A). SAS-6 levels varied such that some populations of cells exhibited significant decreases in SAS-6 levels while others did not. I predict that this is related to the timing of CEP135 expression during the cell cycle. The reduced SAS-6 levels in CEP135full and CEP135mini expressing cells is consistent with the loss of SAS-6 protein from centrioles in Chlamydomonas Bld10 mutants and suggests that the interplay between
SAS-6 at the inner cartwheel and Bld10/CEP135full at the outer cartwheel is important for new centriole biogenesis. Moreover, CEP135mini promotes a conformation that blocks SAS-6 assembly. It remains to be discovered how CEP135mini promotes such an effect on the cartwheel and whether this is by directly affecting CEP135full’s function at the outer cartwheel. I next tested whether the CEP135 binding protein, CPAP, is also affected by CEP135 levels. CEP135mini expression causes a drastic reduction (90% decrease) in CPAP localization to centrioles (Figure 2.7B). CEP135full overexpression
60 also causes an intermediate loss in CPAP localization to centrioles and I predict that this is because the high levels of CEP135mini or CEP135full sequester CPAP to the cytoplasm. However, CEP135mini exhibits a more potent inhibitory effect on CPAP localization and this may promote its selective inhibition of centriole duplication.
Figure 2.7: CEP135mini expression displaces SAS-6 and CPAP from centrosomes. (A) Exogenous CEP135full and CEP135mini expression in RPE1 cells decreases SAS-6 localization to centrioles. The relative levels of SAS-6 were compared to untransfected cells. Exogenous CEP135full and CEP135mini expression causes a variable 43% and 44% (p<0.01) decrease in SAS-6 levels, respectively. Mean±SEM represents 6 separate experiments. Scale bar, 1 μm. (B) CEP135mini expression in RPE1 cells decreases CPAP localization to centrioles. Relative levels of CPAP were compared to untransfected cells. CEP135full and CEP135mini expression causes a 43% and 90% (p<0.01) decrease in CPAP levels. CEP135full expression causes an intermediate 42% decrease in CPAP levels. Mean±SEM represents 3 separate experiments. Scale bar, 1 μm.
61 Elevated CEP135mini is Sufficient to Repress Centrosome Frequency in Breast Cancer
Cells
CEP135mini represses centriole assembly and limits the localization of fundamental centriolar proteins including SAS-6 and CPAP (Dahl et al., 2015).
Moreover, CEP135mini levels are regulated in a cell cycle-dependent manner such that levels are lowest during G1/S when centrioles duplicate. CEP135mini levels increase through the rest of the cell cycle, peaking at metaphase of mitosis. I suggest this is to prevent promiscuous centriole duplication. Given that CEP135mini levels are reduced in breast cancer cells relative to CEP135full levels, I tested whether increasing CEP135mini could repress centriole overduplication. I expressed CEP135mini in MDA-231 cells using a stable tetracycline-inducible fluorescently labeled GFP-CEP135mini MDA-231 cell line
(hereafter GFP-CEP135mini-Tet) and measured the frequency of cells with CA.
Tetracycline induction for three days produced an approximately 6.5-fold increase in centrosomal CEP135mini fluorescence intensity (Figures 2.8A, 2.8B, and
2.8.1A). While CA and centrosome number are modestly decreased, GFP-CEP135mini-
Tet expression resulted in an approximately 3-fold increase in the number of cells with underduplicated centrosomes and acentrosomal cells (Figures 2.8C and 2.8D).
Moreover, centrosomal γ-tubulin was reduced in GFP-CEP135mini-Tet expressing cells
(Figures 2.8E and 2.8.1B; (Dahl et al., 2015)). In summary, increased CEP135mini-Tet expression is sufficient to reduce centrosome number and γ-tubulin in breast cancer cells.
62 AB GFP- CEP135mini Centrin γ-tubulin Mgeer
mini 8
**** Tetracyclineinduction
6 -
centrosome 4 Non-amplified
cen trosomal
2 +
intensity ofGFP-CEP135
Relative 0 fluorescence
centrosome centrosome
- + Underduplicated Tetracycline induction Tetracycline induction C Underduplicated E - + centrosomes (%) D
Normal mini centrosomes (%) 125% cell per centrosome of Number 1.2
Amplified 4.0 3.0 GF P- centrosomes (%) CEP135 1.0 100% **** 3.5 0.8 2.5 75% ****
-tubulin 0.6 3.0 γ ** **
50% intensity -tubulinfluorescence 0.4 2.0 γ 2.5 25% 0.2 Percentageofcells *
Centrin
Relative Numberofcentrioles percell 0.0 0% 2.0 1.5 - + - + - + Tetracycline induction Tetracycline induction Tetracycline induction F Figure 2.8: Elevated CEP135mini is sufficient to decrease CA. (A) Top panel, schematic and timeline of tetracycline induction in GFP-CEP135mini-Tet MDA-231 cells. Bottom panel, the relative centrosome fluorescence intensity of GFP-CEP135mini in non-induced
γ and induced GFP-CEP135mini-Tet cells. (B) Non-amplified and underduplicated centrosomes in non-induced and induced GFP-CEP135mini-Tet cells. GFP-CEP135mini-Tet (green) cells stained for centrioles (centrin, grayscale) and PCM (γ tubulin, red). (C) Percentage of cells with underduplicated (gray), normal (black) and amplified (red) centrosomes in non-induced and induced GFP-CEP135mini-Tet cells. (D) The frequency of centrioles (red) and centrosomes (green) in the total population of non-induced and induced GFP-CEP135mini-Tet MDA-231 cells. (E) Left panel, non-amplified centrosomes in non-induced and induced GFP-CEP135mini-Tet cells. GFP-CEP135mini-Tet (green) cells stained for centrioles (centrin, grayscale) and PCM (γ tubulin, red). Right panel, the relative fluorescence intensity of γ-tubulin in non-inducedγ and tetracycline-induced GFP- CEP135γ mini-Tet cells. Mean±SEM. Fischer’s exact test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
63 A A B 3.0 Tetracycline induction of GFP-CEP135 mini 15 **** 2.5 i i min 12 2.0 **** 9 1.5
intensity
-tubulin -tubulin fluorescence
Day 3 γ 6 1.0
0.5 3 Relative
intensity GFP-CEP135 of
Relativenormalized fluorescence 0.0 0 - + - + Tetracycline induction Tetracycline induction
Figure 2.8.1: Supplemental, elevated CEP135mini expression is sufficient to decrease CA in breast cancer cells. (A) Left panel, schematic of GFP-CEP135mini-Tet MDA-231 cell induction. Right panel, quantification of the normalized fluorescence intensity of GFP- CEP135mini-Tet relative to non-induced cells. (B) Quantification of the relative fluorescence intensity of γ-tubulin in non-induced and induced GFP-CEP135mini-Tet MDA- 231 cells. Mean±SEM. Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
Mutations Affecting CEP135mini Alternative Polyadenylation Reduce the CEP135full:mini
Ratio and Centrosome Amplification
Dahl et al showed that CEP135mini mRNA includes at least part of intron 6 where
a translation termination codon is present (Figure 2.9A). Based on genome database
annotations, I previously reported CEP135mini to be an alternative splice isoform (Dahl et
al., 2015). However, the precise 3’ end of the CEP135mini mRNA was not investigated.
Intron six of CEP135 contains a genomically encoded poly(A) tract and is an internal
priming candidate in studies that utilize oligo d(T) to map mRNA 3’ ends. Database
annotations do not accurately reflect the transcription termination for RNAs from such
genes (Nam et al., 2002). Understanding the nature of the CEP135mini mRNA 3’ end
could inform how the CEP135 isoforms are formed and regulated. Two plausible models
64 for CEP135mini mRNA isoform generation are alternative splicing and alternative polyadenylation that both utilize overlapping machinery for RNA processing (Tian et al.,
2007). Alternative splicing would promote CEP135mini formation through intron 6 retention. Alternatively, a proximal non-canonical poly(A) signal within intron 6 may be used for transcriptional termination (Figure 2.9A). To distinguish between these models,
3’ RNA-ligation mediated RACE was performed on cytoplasmic RNA to enrich for mature messages. RT-PCR primers determined that CEP135mini’s mRNA terminates at a site between 580 nts and 839 nts downstream of the stop codon (Figure 2.9B).
Moreover, 3’READS+ data suggest there are three sites where poly(A) tracts are added in intron 6 (Hoque et al., 2013; Zheng et al., 2016). Amongst them, the first site is 802 bps downstream of the stop codon, which is consistent with my 3’ RACE data and suggests that CEP135mini’s 3’UTR terminates 802 bps downstream from the stop codon in intron 6 (Figure 2.9.1A). Two non-consensus poly(A) signals near this site are
AAUAUA and GAUAAA (Beaudoing, 2000). These results support the model that
CEP135mini is an alternatively polyadenylated CEP135 isoform.
Polyadenylation signal usage is often misregulated in cancers (DeRisi et al.,
1996; Mayr and Bartel, 2009). Altered utilization of CEP135 poly(A) signals would affect
CEP135full:mini transcript ratios in breast cancer cells. The usage of one poly(A) signal over another depends on both the relative strengths of the poly(A) signals and the juxtaposed sequences and trans-acting factors that bind to them (Moreira et al., 1995;
Nunes et al., 2010; Di Giammartino et al., 2011). CEP135full utilizes a distal, consensus poly(A) signal for transcription termination, whereas CEP135mini utilizes a proximal non- consensus poly(A) signal. I suggest that CEP135mini’s proximal non-consensus poly(A)
65 signal is weaker than CEP135full’s and is subject to regulation that controls the relative
CEP135 isoform levels.
To test whether poly(A) signals regulate the CEP135full:mini ratio, I used CRISPR-
Cas9 to attempt to insert an exogenous consensus poly(A) signal upstream of the endogenous CEP135mini non-consensus poly(A) signals in MDA-231 cells (Figure S6B;
(Levitt et al., 1989)). However, cells with these mutations were not recoverable. When screening for clones with mutations in the 3’UTR of CEP135mini, I identified a mutant
(CEP135mini-3’UTR Mutant) that exhibits an approximately 1.5-fold increase in
CEP135mini mRNA (Figure 2.9C, 2.9D, and 2.9.1C). This produced an approximately
40% decrease in CEP135full:mini ratio in MDA-231 cells and a corresponding increase in
CEP135mini protein levels (Figure 2.9D and 2.9E).
Consistent with the reduced CEP135full:mini ratio, the CEP135mini-3’UTR Mutant cells exhibit a significant increase in the number of cells with underduplicated centrosomes and a modest decrease in the number of cells in the population with CA
(Figure 2.9F and 2.9G). Furthermore, the CEP135mini-3’UTR Mutant cells exhibit a dramatic reduction in centrosomal γ-tubulin (Figure 2.9H, 2.9I and 2.9.1E). The
CEP135mini-3’UTR MDA-231 cells have an increased mitotic index and many of these cells have apolar mitoses (Figure 2.9.1D). This is likely due to the significant increase in underduplicated centrosomes. Overall, these results support a model in which the nucleotide sequences adjacent to the alternative poly(A) site influence the CEP135full:mini ratio and regulate centrosome number and function.
66 mini mini CEP135 isoforms: CEP135 specific CEP135 specific reverse primers ABEx-6 Ex-7 forward primer full In-6 CEP135 R 1 R 2 R3 F1
Ex-6 -Coding Exon(Ex) Ex- 6 Intron 6 Ex- 7 min i In-6 CEP135 - Non-Coding Exon 3’UTR? Stop codon - Stop codon R R R (3’ RLM-RACE) 1 2 3 - Intron(In) 436bps 580bps 839bps CEP135min i biogenesis models:
min i 1. ALTERNATIVE SPLICING 2. ALTERNATIVE POLYADENYLATION CEP135 3’UTR Intron retention Proximal poly (A) site selection Ex-6 Ex-7 Ex-6 In-6 In-6 Positive control (gDNA) 3’ UTR 3’ UTR Ex-6
min i 3’UTR full 3’ UTR Control CEP135 2.0 CEP135 C D CEP135mini min i full:mini 3’ UTR Mutant CEP135 CEP135 Insertion 1.5 ** 3’UTR MDA-231 Control Mutant
CEP135full 1.0
min i ****** ** CEP135 transcript0.5levels
GUSB Relative normalized CEP135 0.0 Control 3' UTR Mutant
mini E CEP135 Centrin Meg r e 2.0 CEP135min i F Underduplicated centrosomes (%) Centrin 125% Normal centrosomes (%) ** Amplified centrosomes (%)
1.5 100% ****
Con trol 75% 1.0
50%
0.5
Percentage of cells 25%
Relativecentrosomal
fluorescence intensity
UTRMutant
’
3 0.0 0% Control 3'UTR Mutant Control 3'UTR Mutant Control 3'UTR Mutant
1.2 G 4 4 HI me fcntrosomes cen of umber N Control 3’UTR Mutant 1.0
**** 0.8 3 3
p er cell cell er p 0.6
-tubulin
γ
intensity
percell
2 2 -tubulin fluorescence 0.4 * γ ****
Cen trin
Numberof centrioles 0.2
1 1 Normal centrosome Underduplicated Relative 0.0 Control 3'UTR Mutant centrosome Control 3'UTR Mutant Figure 2.9: CEP135mini is an alternatively polyadenylated isoform, and mutations near the CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells.
67
Figure 2.9: CEP135mini is an alternatively polyadenylated isoform, and mutations near the CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells. (A) Top panel, schematic of CEP135full and CEP135mini genes. Red and green bars denote coding exons (Ex) of CEP135full and CEP135mini, respectively. Blue bars denote non-coding exons. Gray lines denote introns (In). Black pentagons denote translation stop codons. Bottom panel, alternate models for CEP135mini mini transcript biogenesis. (B) CEP135 3’UTR specific forward primer (F1) and reverse mini primers (R1, R2 and R3) used to map the approximate end of the CEP135 3’UTR using 3’ RNA-ligated RACE. (C) Top panel, schematic of control and mutant 3’UTR CEP135mini transcripts. Red bar denotes the Cas9 mediated-insertion site in the 3’UTR. Bottom panel, CEP135full, CEP135mini and GUSB transcript levels in control and 3’UTR Mutant MDA-231 cells. (D) CEP135full, CEP135mini and CEP135full:mini transcript levels in control and 3’UTR Mutant MDA-231 cells. (E) Left panels, centrosomal CEP135mini protein in control and 3’UTR Mutant MDA-231 cells stained for CEP135mini (green) and centrin (grayscale). Right panel, quantification of CEP135mini and centrin fluorescence intensities in control and 3’UTR Mutant MDA-231 cells. (F) Percentage of cells with underduplicated (gray), normal (black) and amplified (red) centrosomes in control and 3’UTR Mutant MDA-231 cells. (G) Number of centrioles and centrosomes per cell in control and 3’UTR Mutant MDA-231 cells. (H) Normal and underduplicated centrosomes in control and 3’UTR Mutant MDA-231 cells stained for centrioles (centrin, red) and γ- tubulin (green). (I) Relative γ-tubulin fluorescence intensity in control and 3’UTR Mutant MDA-231 cells. Mean±SEM. Student’s t-test, Fischer’s exact test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1μm.
68 mini A CEP135 ’s poly(A) signal reads Ex-6 - Coding Exon(Ex) CEP135mini In-6 - Stop codon - Intron(In) Poly A Reads
Ex-6 In-6 Ex-6
802bps Poly A Reads from stop codon of CEP135mini AATATAGATGTAAC------56bps------GATAAA AATGGAGTTAAAGAAAATTGAC CA CEP135 non-consensus poly A signals
B CEP135mini’s 3’UTR REGION
PAM site CEP135 non-consensus poly A signals Homology Directed Repair Insert
Synthetic 5’-AATAAAA |T GATCTTTATTTTCATTA|GATCGTGTGTTGGTTTTTTTGTGTGT-3’ Poly A sequence: Consensus poly A signal C MDA-231 CEP135 Intron 6 Wildtype (WT) sequence sgRNA target sequence PAM 5’---AATATAGATGTAACTTAGGT GTTGG TATAACTAAATCTTTA--3’ 3’UTR Mutant CRISPR-Cas9 Mutant alleles Allele 1 5’---AATATAGATGTAACTTAGG-107bps insertion- T GTTGGTATAACTA--3’ Allele 2 5’---AATATAGATGTAACTTAGG- 74bps insertion- T GTTGGTATAACTA--3’ Allele 3 5’---AATATAGATGTAACTTAGGT-1bp insertion- GTTGGTATAACTAA--3’
Bipolar mitosis Apolar mitosis 1.2 D 15% E
12% 0.9
DNA 9% 0.6
6% intensity ****
-tubulin
γ γ 0.3
Mitoticindex (%) 3%
0% 0.0 Control 3’UTR Mutant Control 3'UTR Mutant Relative -tubulin fluorescence Control 3'UTR Mutant
Figure 2.9.1: Supplemental, CEP135mini is an alternatively polyadenylated isoform, and mutations near the CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells.
69 Figure 2.9.1: Supplemental, CEP135mini is an alternatively polyadenylated isoform, and mutations near the CEP135mini poly(A) signal reduce the CEP135full:mini ratio and centrosome number in breast cancer cells. (A) Top panel, schematic of the CEP135mini gene. Green bars denote coding exons (Ex) of CEP135mini. Gray lines denote introns (In). Black pentagons denote translation stop codons. Middle panel, poly(A) signal reads (red) in intron 6 of the CEP135 locus from the 3’READS+ dataset (Zheng et al., 2016). Bottom panel, magnified view of the first poly(A) signal read (red) that is 802bps downstream of the CEP135mini stop codon (pentagon). Underlined sequences represent CEP135mini’s predicted polyadenylation signals and cleavage sites. (B) Top panel, the predicted poly(A) signals AAUAUA and GAUAAA in CEP135mini’s 3’UTR region. Middle panel, 20 nucleotide Cas9 target region in CEP135’s intron 6 utilized for the design of gRNA along with the PAM site (blue). Bottom panel, Synthetic poly(A) signal attempted for homology-directed repair (HDR) knock-in into CEP135 intron 6. (C) Sequence of CEP135mini 3’UTR CRISPR-induced mutant alleles. (D) Left panel, bipolar mitosis in a control MDA-231 cell and apolar mitosis in a CEP135mini 3’UTR Mutant MDA-231 cell. Cells were stained for PCM (γ-tubulin, red) and chromosomes (Hoechst 33342, blue). Right panel, the percentage of cells in mitosis in control and 3’UTR Mutant MDA-231 cells. (E) Relative γ-tubulin fluorescence intensity in control and 3’UTR Mutant MDA-231 cells. Mean±SEM. Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
70 Discussion
Centriole overduplication provides a fundamental mechanism by which the frequency of amplified centrosomes increases in breast cancer cell lines. Centriole overduplication in breast cancer cells may be, in part, modulated by the regulated levels of two CEP135 isoforms: one that promotes centriole assembly and one that represses centriole assembly. These isoforms are generated by transcriptional termination either at the distal, canonical poly(A) signal or at a proximal non-canonical poly(A) signal. This suggests that precise control of transcription termination is required to prevent tumor- promoting events like centriole and centrosome overduplication.
Centriole Overduplication in Breast Cancer
The level of CA in my studies of cultured breast cancer cells is stable through multiple passages of each cell line. This suggests that despite abnormalities in centrosome number, homeostatic mechanisms exist that maintain a specific level of amplified centrioles and centrosomes in a cell population. This likely reflects a balance between the initiation and propagation of centriole overduplication and the loss or death of cells with CA. Centriole rosette-like structures containing multiple daughter centrioles surrounding a single mother were reported in primary malignancies, suggesting a high frequency of multiple daughter centriole overduplication from a single mother centriole
(Cosenza et al., 2017). I tested whether multiple daughter centrioles duplicate from a single mother centriole in breast cancer cells and observed a low frequency of these events. In the MDA-231 cell population that have more than two new centrioles, 2% of cells have multiple new centrioles from a single mother centriole. This represents a small fraction of the total population, which I estimated to be 0.3%, given that 13% of
71 the MDA-231 cells are overproducing new (SAS-6 positive) centrioles. This low frequency corresponds with the stable levels of CA observed through multiple cell passages of breast cancer cell lines, as a high frequency of centriole overduplication at each cell division would produce increasing frequencies of cells with centriole overduplication as cell cultures are passaged. Furthermore, I did not observe rosette formations. Almost all new centrioles form in a once-and-only-once event where each existing centriole forms only one new centriole to maintain either normal or excess numbers of centrioles. We, therefore, suggest that most centriole overduplication events simply maintain existing CA in breast cancer cells (Figure 2.4E and 2.4F). This indicates that breast cancer cells, despite having lost some aspects of centrosome number control, retain regulatory mechanisms that limit centriole duplication to a single daughter centriole for each mother centriole.
CEP135 Isoform Dysregulation Promotes Centriole Overduplication in Breast Cancer
Consistent with CEP135full’s function as a positive regulator of centriole duplication, CEP135full levels are elevated in breast cancer cell lines with increased frequencies of cells with CA (Figure 2.5). Furthermore, expression of CEP135full is associated with new centriole formation and CA in breast cancer cells (Figure 2.6). One model for CEP135full-induced centriole formation is that increases in CEP135full act to stabilize one of its interacting partners, SAS-6, at newly formed centrioles, promoting both centriole assembly and nascent centriole stability (Figure 2.6; (Matsuura et al.,
2004; Jerka-Dziadosz et al., 2010; Lin et al., 2013)). Moreover, CEP135full‘s function is not limited to centriole formation. CEP135full has a microtubule binding domain and affects microtubule stability and organization (Ohta et al., 2002; Carvalho-Santos et al.,
72 2012; Lin et al., 2013; Kraatz et al., 2016). Consistent with this, overexpressed
CEP135full decorates the centriole cylinder walls in addition to its conventional proximal- end localization (Figure 2.5B). Therefore, I hypothesize that increased CEP135full promotes ectopic procentriole formation by stabilizing additional SAS-6 foci.
In contrast to CEP135full, CEP135mini’s expression is not, or is minimally, upregulated in breast cancer cells. Consistent with CEP135mini’s function as a negative regulator of centriole duplication, increased expression of CEP135mini in breast cancer cells produces a modest decrease in the number of cells with CA and a significant increase in centrosome underduplication (Figure 2.8). This suggests that increased
CEP135mini levels in breast cancer cells repress centriole duplication leading to underduplicated centrosomes. It is interesting to consider how modulating CEP135mini could be used as a tool to reduce the number of cells with too many centrosomes in cancer. However, appropriate dosage routines of CEP135mini expression are required to limit the centrosome underduplication that I observed in these studies. Indeed, maintaining the fine balance between the CEP135 isoforms is critical for the homeostasis of centrosome numbers in breast cancer cells.
CEP135mini in Repression of Centriole Duplication
Elevated expression of CEP135mini limits centriole duplication and also limits localization of SAS-6 and CPAP to centrioles and represses centriole duplication
(Figure 2.8). It is not known how CEP135mini limits the localization of these proteins to centrioles. The N-terminus of CEP135full forms a two-stranded coiled-coil structure with chains in parallel that interacts with the negatively charged outer surface of microtubules and C-terminus of CPAP(Kraatz et al., 2016). The C-terminus of
73 CEP135full also has a second coiled-coil region that interacts with SAS-6. CEP135mini has a similar N-terminus to that of CEP135full. It is known that coiled-coil proteins can readily exchange chains between each other and form heterodimers. One plausible model as to how CEP135mini might limit SAS-6 and CPAP localization to centrioles is it may form a heterodimer through its interactions with coiled-coil regions both at the N- and C-termini of CEP135full and interfere with its interactions with its binding partners
SAS-6 and CPAP. Thus, the formation of a non-functional heterodimer can be a mechanism through which CEP135mini limits the localization of SAS-6 and CPAP to centrioles.
Alternative Polyadenylation in CEP135 Isoform Regulation
Two isoforms of CEP135 act antagonistically to control centriole duplication (Dahl et al., 2015). The CEP135 gene locus (chromosome locus 4q12) exhibits copy number gain in breast cancer patients predicting that both isoforms would be similarly upregulated and would maintain the CEP135full:mini ratio (Yu et al., 2009). However, both
CEP135 isoforms do not increase in the breast cancer cell lines examined. Inefficient, use of the non-consensus poly(A) signal is likely responsible for the low CEP135mini levels relative to CEP135full. However, a less efficient poly(A) signal on its own does not explain the increased ratio of CEP135full to CEP135mini in breast cancer cells. Additional levels of CEP135 isoform regulation must malfunction in breast cancer cells.
To study the direct impact of altered transcription termination, I engineered mutations near CEP135mini’s poly(A) signal to assess whether this would alter
CEP135mini levels. Attempted insertions of a consensus poly(A) signal did not yield viable cells, suggesting that high levels of CEP135mini are lethal. This is consistent with
74 my prior studies using the transient expression of CEP135mini (Dahl et al., 2015). An
MDA-231 cell line with inserted random DNA sequences near the poly(A) signal was isolated (Figure 2.9). The mutations in this cell line increase CEP135mini levels and reduce centrosome numbers. I hypothesize that the increased CEP135mini levels result from altered mRNA stability or changes to the strength or regulation of CEP135mini’s poly(A) signal. Either result is interesting and will be a target of future investigations to understand the mechanism by which the CEP135full:mini ratio is controlled. To my knowledge, this is the first demonstration that altered 3’-end formation in cancer cells regulates centrosome number. This is important because alternative splicing and polyadenylation are commonly dysregulated in cancer cells that exhibit CA (Mayr and
Bartel, 2009; David and Manley, 2010).
In summary, the dysregulation of CEP135 isoforms in breast cancer cells contributes to the loss of centrosome number homeostasis. Differential levels of
CEP135 isoforms are generated by the use of alternative polyadenylation signals.
These findings support the conclusion that alternative polyadenylation, which is commonly disrupted in cancers, regulates the assembly of macromolecular structures such as centrosomes.
Materials and Methods for Chapter II
Cell Culture
Breast cancer cell lines MCF10A, MCF7, ZR-75.1 (ZR751), BT-20, MDA-MB-231
(MDA-231), SUM159PT were obtained from University of Colorado Cancer Center
Tissue Culture Core and BT-549 were obtained from American Type Culture Collection.
Mammalian tissue culture lines were all grown at 37°C with 5% CO2. MCF10A cells
75 were received at passage 51 and were grown in DMEM/F12 (Invitrogen #11330-032),
5% Horse Serum (Invitrogen #16050-122), 20 ng ml-1 EGF (Invitrogen #PHG0311), 0.5 mg ml-1 Hydrocortisone (Sigma #H-0888), 100 ng ml-1 Cholera toxin (Sigma #C-8052),
10 μg ml-1 Insulin (Sigma #I-1882) and 1% Pen/Strep (Invitrogen #15070-063). MDA-
MB-231 cells were received at passage 15, BT-20 were received at passage 11, and
MCF-7 were received at passage 7. These lines as well as 293FT cells were grown in
DMEM (Invitrogen #11965-092), Pen/Strep (Invitrogen #15070-063) and 10% FBS
(FBS; Gemini Biosciences). ZR-75.1 (ZR751) cells were received at passage 51, ZR-
75.1 and BT-549 cells were grown in RPM1 (Invitrogen #11875-093), 10% FBS (FBS;
Gemini Biosciences) and Pen/Strep (Invitrogen #15070-063). SUM-159.PT (SUM159) were received at passage 10 and grown in Ham’s media (Invitrogen #11765054),
Hydrocortisone (Sigma #H-0888), Pen/Strep (Invitrogen #15070-063) and 10% FBS
(FBS; Gemini Biosciences). Cell lines were authenticated at the sources and tested negative for mycoplasma using the MycoAlert mycoplasma detection kit through the
University of Colorado Cancer Center Tissue Culture Core. Cells were passaged and sub-cultured using Trypsin (Invitrogen #150901-046) when cultures reached 60-80% confluency.
Generation of mCherry-CEP135full-Tet and GFP-CEP135mini-Tet Cells
The generation of tetracycline-inducible mCherry-CEP135full and GFP-CEP135mini constructs is described below. The mCherry-CEP135full fragment was obtained through
PCR with Phusion DNA polymerase of a pre-existing plasmid (pcDNA5-FRT-TO-
GCaMP-mCherry-CEP135full) using primers that have Nhe1 and Xho1 sites appended to them. This was cloned into the tetracycline-inducible construct pcw57.1 using Nhe1
76 and Sal1. The GFP-CEP135mini fragment was obtained by restriction digestion (using the enzymes Pme1 and Xho1) of a pre-existing construct (pcDNA5-FRT-TO-GFP-
CEP135mini). This was cloned into the tetracycline-inducible construct pcw57.1.
Lentivirus harboring tetracycline-inducible mCherry-CEP135full-Tet or GFP-
CEP135mini-Tet was made by transfection of 293FT cells. 293FT cells were plated in 6 cm dishes and allowed to reach 50%-70% confluency. Cells were then transfected with tetracycline-inducible mCherry-CEP135full-Tet or GFP-CEP135mini-Tet constructs, and second-generation lentivirus packaging plasmids (pMD2.G and psPAX2) using
Lipofectamine 2000 (Life Technologies # 11668019). 293FT media containing virus was harvested and MDA-231 cells were infected for 24-48 hours in the presence of 10 μg ml−1 (26.7 μM) polybrene. After a 24 hour recovery, transduced cells were selected with puromycin at 2 μg ml−1 (4.24 μM) and were flow sorted to isolate and plate single cells into 96 well plates. Such clones were cultured in 50% filtered conditioned media with
50% fresh media. mCh-CEP135full-tet and GFP-CEP135mini-tet cells were induced with tetracycline (Invitrogen #550205) at 2.5 μg ml−1 (5.63 μM).
Generation of 3’ UTR Mutant Cells
The 3’UTR of CEP135mini was edited using CRISPR/Cas9 mediated genome engineering in MDA-231 cells. The oligonucleotide donor sequence for homology- directed repair comprises a consensus poly(A) signal (Levitt et al., 1989) with 50bps of flanking homologous region. This oligonucleotide donor sequence was introduced to insert a consensus poly (A) sequence. The 20nt target region with the least off-targets utilized for sgRNA design was determined using the https://zlab.bio/guide-design- resources. The Cas9-EGFP containing the sgRNA was expressed from the pX458
77 construct. The sequences of sgRNA cloned into px458 and single-stranded consensus poly(A) signal donor sequences are shown in Figure 2.9.1B. pX458 and the single- stranded consensus poly(A) signal donor sequences were co-transfected into MDA-231 cells. Cas9-GFP positive cells were flow sorted to isolate and plate single cells into 96 well plates. Such clones were cultured in 50% filtered conditioned media with 50% fresh media. Clones were screened using PCR with primers flanking the predicted poly(A) signals. For the sequencing reaction, the PCR products were cloned into the plasmid pUC18. Below are the sequences of insertions observed in the three alleles of chromosome 4 in the 3’UTR Mutant MDA-231 cells.
Annotated genomic sequence (chromosome 4) in the 3’UTR of CEP135mini:
CACCGCATACTTTTTCTTTTCTGCATTGACTGCATTTTTTTTGAGTGATCTGCACACA
CAAATATAGATGTAACTTAGGGTGTTGGTATAACTAAATCTTTAAAGTGTTTTGAAG
ATTAGTTGGATAAAAATGGAGTTAAAGAAAATTGACCAGGCGTGGTG
Here, TATAGATGTAACTTAGGGTGTTGG represents gRNA and PAM site. The underlined sequences represent annotated genomic sequences in the 3’UTR of
CEP135mini. The blue sequences represent the mutations (insertions) at the target site.
Sequencing results of the PCR products representing the three alleles of chromosome 4 from the 3’UTR Mutant MDA-231 cells:
Allele 1:
TTTCTATTGATGTCTTGGCTAAATTTTGATTTGGGTAATTATAGGCTTGTTTTCTTAA
TATGCTTCTTGAAGGTTAAATTGGGTAACATGTTTTTCATTTTCTGCATTGACTGCAT
TTTTTTTGAGTGATCTGCACACACAAATATAGATGTAACTTAGGACGGTAAGCATAT
GATAGTCCATTTTAAAACATAATTTTAAAACTGCAAACTACCCAAGAAATTATTACTT
78 TCTACGTCACGTATTTTGTACTAATATCTTTGTGTTTTGTTGGTATAGCTAAATCTTT
AAAGTGTTTTGAAGATTAGTTGGATAAAAATGGAGTTAAAGAAAATTGACCAGGCGT
GGTGGCTCATGCCTGTAATCGCAGCACTTTGGGAGGCCGAGGCGGGTGGATCAC
TTGAGATCAGAAGTTTGAGACCATCCTGGCCAACATGGTGAAACCTCATCTCTACT
AAAAATACAAAAAATTAGCCGGGCGTAGTGGCGGGCGCCTGTAGTCCCAGCTACT
TGGGAGGCTGAGGCAGGAGAATGGCGTGAACCCG
Allele 2:
CTTTGTCTTTTCATAATTTATATTGATGTCTTGGATAAATTTTGATTTGGGTAATTATA
GGCTTGTTTTCCTAATATGCTTCTTGAAGGTTAAATTGGGTAACATGTTTTTCATTTT
CCGCATTGATTGCATTTTTTTTGAGTGATCTGCACACACAAATATAGATGTAACTTA
GGATGTATCCGCTCATGAGACAATAACCCTGATAAATGCTTCAATAATATTGAAAAA
GGAAGAGTATGAGTATTCATGTTGGTATAACTAAATCTTTAAAGTGTTTTGAAGATT
AGTTGGATAAAAATGGAGTTAAAGAAAATTGACCAGGCGTGGTGGCTCACGCCTGT
AATCGCAGCACTTTGGGAGGCCGAGGCGGGTGGATCACTTGAGATCAGAAGTTTG
AGACCATCCTGGCCAACATGGTGAAACCTCATCTCTACTAAAAATACAAAAAATTAG
CCGGGCGTAGTGGCGGGCGCCTGTAGTCCCAGCTACTTGGGAGGCTGAGGCAGG
AGAATGGCGGAACCCG
Allele 3:
TTTCTATTGATGTCTTGGCTAAATTTTGATTTGGGTAATTATAGGCTTGTTTTCTTAA
TATGCTTCTTGAAGGTTAAATTGGGTAACATGTTTTTCATTTTCTGCATTGACTGCAT
TTTTTTTGAGTGATCTGCACACACAAATATAGATGTAACTTAGGCTGTTGGTATAAC
TAAATCTTTAAAGTGTTTTGAAGATTAGTTGGATAAAAATGGAGTTAAAGAAAATTGA
CCAGGCGTGGTGGCTCATGCCTGTAATCGCAGCACTTTGGGAGGCCGAGGCGGG
79 TGGATCACTTGAGATCAGAAGTTTGAGACCATCCTGGCCAACATGGTGAAACCTCA
TCTCTACTAAAAATACAAAAAATTAGCCGGGCGTAGTGGCGGGCGCCTGTAGTCC
CAGCTACTTGGGAGGCTGAGGCAGGAGAATGGCGTGAACCCG
Transfections
MDA-231 cells at 50-80% confluence were transfected using Lipofectamine 2000
(Invitrogen # 11668019). Plasmid DNA and Plus reagent (Invitrogen # 11514015) were mixed at 1:1 and incubated for 5 minutes. This mixture was then combined with
Lipofectamine at a1:3 ratios. Complexes were diluted in Opti-MEM (Invitrogen
31985062). After a 4-hour incubation, the complexes were removed and the transfected cells were supplied with fresh media.
Immunofluorescence
12 mm diameter coverslips were acid-washed and heated to 50°C in 100mM HCl for 16 hours. This was followed by washes with water, 50%, 70%, and 95% ethanol for
30 minutes each. Coverslips were coated with Type-1 collagen (Sigma # C9791), air- dried for 20 minutes in the laminar hood and exposed to UV light for cross-linking of collagen for 20 minutes. Cells were cultured on collagen-coated coverslips to 55-70% confluence.
For centrosome immunofluorescence, cells were fixed with 100% methanol at
−20°C for 8 minutes. Fixed cells were washed with PBS/Mg (1x PBS, 1mM MgCl2), and then blocked with Knudsen Buffer (1x PBS, 0.5%BSA, 0.5% NP-40, 1mM MgCl2, 1mM
NaN3) for 1 hour. Cells were incubated overnight with primary antibodies diluted in
Knudsen Buffer at 4oC. Coverslips were washed with PBS three times in 5-minute intervals. Secondary antibodies and Hoechst 33258 (10μg ml−1, Sigma #B2261) were
80 diluted in Knudsen buffer and incubated for 1 hour at room temperature. Coverslips were mounted using Citifluor (Ted Pella) and sealed with clear nail polish. Coverslips for
SIM imaging were mounted using Prolong Gold (Life Technologies #P10144) and sealed with clear nail polish.
Antibodies used for immunofluorescence are - centrin (1:2,000; 20H5; Abcam),
--tubulin (1:1000; DQ-19; Sigma), -CEP135full (1:5,000; generous gift from Dr. T.K.
Tang), - CEP135mini(1:2,000; (Dahl et al., 2015)), -SAS-6 (1:2,000, Bethyl A301-
802A), -CPAP (1:350, Proteintech CENPJ 11517-1-AP) and -CEP192 (1:2,000 generous gift from Dr. Andrew Holland). Alexa-fluor secondary antibodies were diluted to 1:1,000 for all experiments (Molecular Probes).
Microscopy
Super-resolution imaging in Figure 2.4 was acquired using Nikon structured illumination microscopy (N-SIM) with a Nikon Ti2 (Nikon Instruments, Inc.; LU-N3-SIM) microscope equipped with a 100X SR Apo TIRF, NA 1.49 objective. Images were captured using a Hamamatsu ORCA-Flash 4.0 Digital CMOS camera (C13440).
The fluorescence imaging utilized for Figures is identical to those described in
(Dahl et al., 2015). Briefly, images were acquired using a Nikon TiE (Nikon Instruments,
Inc.) inverted microscope stand equipped with a 100X PlanApo DIC, NA 1.4 objective.
Images were captured using an Andor iXon EMCCD 888E camera or an Andor Xyla 4.2
CMOS camera (Andor Technologies). Images in Figure 2.5 were acquired using a
Swept Field Confocal system (Prairie Technologies / Nikon Instruments) on a Nikon Ti inverted microscope stand equipped with a 100X Plan Apo , NA 1.45 objective. Images were captured with an Andor Clara CCD camera (Andor Technologies).
81 Nikon NIS Elements imaging software was used for image acquisition. Image acquisition times were constant within a given experiment and ranged from 50 to 400 msec, depending on the experiment. All images were acquired at approximately 25ºC.
Images presented in most of the figures are maximum-intensity projections of the complete z-stacks.
Fluorescence Intensity Quantitation
Image analysis was performed using Python with the tiff file library for reading images and the NumPy library (https://pypi.org/project/tifffile/ and http://www.numpy.org/) for performing computations. A command-line python script was written and utilized for this analysis. The code is available at http://thepearsonlab.com/image-analysis-routines.html. The script was given the path of the image as a command-line argument. The script first showed the image in a graphical user interface. For fluorescence intensity analysis at the centrosome, a 15- pixel square box was centered on the centrosome. The total intensity in this box was computed and divided by the area of the box. The background value of the image was computed by identifying four boxes outside the centrosome and dividing the total intensity in these boxes by the total area of these boxes. The final intensity at the centrosome was computed after subtraction of the background value.
Centriole and Centrosome Number Counts
Cells were scored as amplified, non-amplified and underduplicated based on centrin and γ-tubulin staining (Dahl et al., 2015). Cells with greater than two γ-tubulin and four centrin foci were scored as amplified centrosomes. Non-amplified centrosomes
82 have both one or two γ-tubulin and two or four centrin foci. Underduplicated centrosomes contain one centrin focus.
Cells were scored for duplication of multiple daughter centrioles versus a single daughter centriole based on SAS-6, centrin and CEP192 staining. When a mother centriole (CEP192, centrin and no SAS-6) had greater than one SAS-6 focus, it was scored as duplicating multiple daughter centrioles. When a mother centriole formed a single SAS-6 focus it was quantified as a duplicating a single daughter centriole.
Reverse-Transcription PCR and Quantitation
Cytoplasmic RNA from cancer cells at 55%-70% confluence was harvested using
RNAeasy kit with DNAase I treatment (Qiagen # 74104). Equal amounts of RNA were utilized for reverse-transcription and cDNA amplification using One-Step rtPCR
(Invitrogen# 1257401). CEP135full transcripts were specifically detected using primers to exon 3 and exon 7 (CAAAATTATCTGCTGTGAAAGCTG and
CCAAAGCAACTGACAGTCG). CEP135mini transcripts were specifically detected using the above primer that anneals in exon 3 and a primer specific to CEP135mini in intron 6
(ACCTATCTCAATCCCTACTATGCAA). The absolute CEP135full and CEP135mini transcript levels were normalized to GUSB (CATTCCTATGCCATCGTGTGG and
GACACCGTGGAAATAGAAAGG).
3’RNA-Ligation Mediated RACE
Cytoplasmic RNA was harvested from cancer cells using the RNAeasy kit with
DNAase I treatment (Qiagen # 74104). Poly A+ RNA was enriched using poly (A) spin columns (NEB # S1560S). 5’ -Adenylation and phosphorylation of the adapter
(AAAGCGGCCGCAGTTGCATCGGATCATGCCCGGGCTCATATGC) was performed
83 using Mth RNA Ligase (NEB #M2611A) and T4 Polynucleotide Kinase (NEB #M0201S).
300pmol of 5’end of the adapter was phosphorylated in 5μl of 10X T4 PNK reaction buffer and 10mM ATP with 1μl of T4 PNK kinase. 100pmol of the phosphorylated adapter was adenylated in 2μl of 10X adenylation buffer and 1mM ATP with 1μl of Mth
RNA ligase.
Phosphorylated and adenylated adapter was ligated to poly A+ RNA using T4
RNA Ligase 2, truncated K227Q (NEB #M0351S). 20pmol (equal amounts) of poly A+
RNA and 5’ pre-adenylated adapter was incubated at 70οC for 2 minutes and cooled immediately to remove secondary structures and ligated in 2μl of 10X ligation buffer, 1μl of RNaseOUT (ThermoFischer Scientific #10777-019), 6 μl of PEG8000 with 1μl of T4
RNA ligase. cDNA was made using an adapter-specific reverse primer. To probe the
3’end of CEP135mini, primers throughout the intron were utilized
(CACTATTCATTGAGGTGAATAGTAG, GCTTGCTATGGTTATAATATAACTCTGC,
CCAACACCTAAGTTACATCTATATTTGTG) (Figure 2.9).
Statistics and Biological Replicates
All center values represent means and error bars represent the standard error of the mean. All the experiments were performed using at least three independent biological replicates. 2.5A-D used >4 biological replicates, respectively. The number of cells used in each immunofluorescence experiment is as follows: Figure 2.4C: >50 cells per condition/150 cells, 2.4D: >15 amplified cells per condition/50 amplified cells, 2.4E:
>15 amplified cells per condition/50 amplified cells. Figure 2.5E:> 50 G2 per condition/150 G2 cells, Figure 2.5F:> 50 G2 per condition/150 G2 cells, Figure 2.5G:>
50 G2 per condition/150 G2 cells. Figure 2.6A: >60 cells per condition/150 cells, 2.6C:
84 >20 cells per condition/40 amplified cells, 2.6E: >100 cells per condition/200 cells.
Figure 2.7A;>30 S-phase cells/90 cells, 2.7B:>30 S-phase cells/90 cells. Figure 2.8A:
>75 cells per condition/150 cells, 2.8C: >100 cells per condition/200 cells, 2.8D: >100 cells per condition/200 cells, 2.8E: >60 cells per condition/120 cells. Figure 2.9E: >60 cells per condition/120 cells, 2.9G:>60 cells per condition/120 cells, 2.9I:>60 cells per condition/120 cells. Fischer’s exact test, Student’s two-tailed t-test, and Mann-Whitney
U-test were used to assess statistical significance between means.
Fischer’s test was utilized to examine the significance of contingency when data were classified into two or more categories. Student’s two-tailed unpaired t-test was used to examine significance between two normal distributions (equal variance assumed). Normality tests were performed both on the raw data and meta-data extracted from the replicates of raw data. Shapiro-Wilk normality test and D'Agostino-
Pearson omnibus normality test was utilized to examine the normality of data. Shapiro-
Wilk normality test was used when the number of samples was less than eight. When the number of samples was greater than eight, the D'Agostino-Pearson omnibus normality test was used. Mann-Whitney u-test was utilized to examine the significance of non-normal distributions.
Results were considered statistically significant with p-values less than 0.05. P- values were denoted on figures according the following values: * p<0.05, ** p<0.01, *** p<0.005 and **** p<0.0005.
85 CHAPTER III
CEP135 ISOFORM DYSREGULATION AND CENTROSOME AMPLIFICATION IN
MICROTUBULE ORGANIZATION AND CHROMOSOME SEGREGATION3
Introduction
Microtubules perform important roles in various cellular processes. The interphase microtubule array is essential in cell migration, intracellular trafficking, and cell polarization while the mitotic microtubule array organizes the bipolar spindle and allows faithful chromosome segregation (Inoue’ and Salmon, 1995; Hyman and
Karsenti, 1996; Desai and Mitchison, 1997). Microtubules consist of tubulin polymers that are intrinsically polar with a minus end and plus end. Microtubule plus ends are highly dynamic that stochastically switch between the states of polymerization and depolymerization and have a relatively less dynamic minus ends at the structures known as centrosomes (Allen and Borisy, 1974; Bergen and Borisy, 1980; Walker et al.,
1988; Desai and Mitchison, 1997).
Centrosomes are the primary microtubule organizing centers of cells (Brinkley,
1985). They consist of a pair of centrioles surrounded by the pericentriolar material
(PCM). The PCM proteins, including -tubulin, CDK5RAP2, and Pericentrin, play an important role in microtubule assembly and organization. Pericentrin and CDK5RAP2 act as a scaffold and facilitate the recruitment -tubulin ring complex (-TuRC) in the
3 Portions of this chapter are published with permission from our previously published article and an article in preparation with our collaborator:
1) Ganapathi Sankaran, D., Hariharan, B., and Pearson, C. G. A quantitative approach to the analysis of the spatial distribution of centrosome and microtubule organization defects (In prep).
2) Ganapathi Sankaran, D, Stemm-Wolf, AJ, and Pearson, CG (2019). CEP135 isoform dysregulation promotes centrosome amplification in breast cancer cells. Mol Biol Cell.
86 PCM (Dictenberg et al., 2002; Fong et al., 2007; Farache et al., 2018). The -TuRC nucleates microtubules (Stearns and Kirschner, 1994; Moritz et al., 1995; Zheng et al.,
1995). Every cell has one centrosome with two centrioles during the G1 phase that duplicates during S phase to form two more daughter centrioles. With the duplication of the daughter centrioles, the organization of PCM as a toroid also occurs (Vorobjev and
Chentsov, 1982; Piel et al., 2000). Phosphorylation of several PCM proteins including
CDK5RAP2 and Pericentrin promote the expansion of PCM during the G2-M phase
(Haren et al., 2009; Lee and Rhee, 2011; Woodruff et al., 2015). The duplication of centrosomes during the cell cycle results in twice as much microtubule nucleation during interphase and allows the bipolar spindle formation during mitosis (Salaycik,
2005).
Loss of centrosome number homeostasis commonly occurs in cancer cells. CA is detected in cells from low-grade pre-malignant lesions to advanced metastatic cancers including breast cancers (Lingle et al., 2002; Pihan et al., 2003b; Martinho et al., 2009;
Denu et al., 2016; Lopes et al., 2018; Marteil et al., 2018). CA contributes to chromosome instability through the formation of multipolar spindles. Centrosomes in multipolar spindles often cluster to form bipolar spindles capable of producing viable daughter cells (Quintyne et al., 2005; Ganem et al., 2009; Godinho et al., 2009).
However, clustered multipolar spindles also produce merotelic microtubule-kinetochore attachments, giving rise to lagging chromosomes during anaphase that are subsequently packaged into micronuclei in the following interphase (Cimini, 2008;
Ganem et al., 2009; Silkworth et al., 2009; Thompson et al., 2010; Thompson and
Compton, 2011; Ly and Cleveland, 2017). This creates both aneuploidy and severe
87 chromosome rearrangements through chromothripsis that occurs in micronuclei (Crasta et al., 2012; Zhang et al., 2015). The CEP135full:mini ratio is elevated in breast cancer cells promotes CA. However, the impact of CEP135 isoform dysregulation on chromosome segregation is not known. I analyzed the formation of multipolar spindles, anaphase-lagging chromosomes and formation of micronuclei in breast cancer cells upon CEP135 isoform dysregulation.
Alterations to microtubule organization and dynamics from amplified centrosomes can explain such merotelic kinetochore attachments during mitosis (Ertych et al., 2014; Cosenza et al., 2017). Amplified centrosomes in mitotic spindle have higher microtubule density around them. However, whether and how the interphase microtubule network around the amplified centrosomes is affected is not known.
Furthermore, how PCM changes at the amplified centrosomes through the cell cycle in cancer cells is not known. A quantitative characterization of whether microtubule organization is altered upon CA is essential.
The microtubule network consists of individual or bundles of microtubule filaments, which can branch, with several microtubule binding proteins on them.
Individual microtubule filaments assemble from heterodimers of α- and β-tubulin on -
TuRC complexes (Stearns and Kirschner, 1994; Moritz et al., 1995; Zheng et al., 1995;
Kollman et al., 2011). -tubulin and other proteins can be also be recruited to the sides of a pre-existing microtubule to nucleate a branched microtubule(Petry et al., 2013;
Ishihara et al., 2014). α- and β- tubulin heterodimers bind to a nucleotide, guanosine triphosphate (GTP), that functions to regulate microtubule polymerization. The GTP hydrolysis results in alternate cycles of growth and shrinkage, a behavior known
88 as dynamic instability. The rate of microtubule growth or shrinkage is determined by the rate of tubulin addition relative to the rate of GTP hydrolysis. If the addition of new
GTP-bound tubulin molecules occurs more rapidly than GTP hydrolysis, the microtubule retains a GTP cap at its plus end and microtubule continues to grow. If the new GTP- bound tubulin is not added rapidly to the growing plus end, the GDP filled old microtubule lattice will disassociate (Bergen and Borisy, 1980; Mitchison and Kirschner,
1984; Walker et al., 1988; Tran et al., 1997; Bowne-Anderson et al., 2013).
Measurement of microtubule organization and dynamics characteristics has been possible by fluorescently labeling tubulin or microtubule-binding proteins such as end- binding protein EB1 or EB3. End-binding proteins track the growing end of microtubules.
These proteins bind to the outer microtubule surface, close to the GTP binding site.
They distinguish between the growing microtubule-end region and the older part of the microtubule by sensing a tubulin conformation that stabilizes the microtubule end and transforms with time into the GDP lattice (Maurer et al., 2012, 2014). This leads to the comet-like accumulation of EBs at the end region of growing microtubules. Tracking individual microtubules is clouded by its high density while counting EB3 comets is tenuous and requires quantitative computational approaches to be amenable.
Previously developed computational tools help in quantitation of EBs, however, these tools require extensive manual annotation (Applegate et al., 2011). In this article, we take a step towards automation of such analysis using machine learning.
Here, we investigate the microtubule organization in centrosome amplified breast cancer cells. To characterize microtubule organization defects, we developed a quantitative image-processing algorithm for the analysis of microtubule organization in
89 centrosome amplified breast cancer cells. I find that the levels of PCM proteins such as
-tubulin and pericentrin are elevated and altered in distribution in centrosome amplified breast cancer cells. Likewise, I observe an increase in microtubule density and EB3 density in centrosome amplified breast cancer cells. However, such image processing algorithms still require extensive manual input. Hence, we designed a computer vision algorithm for detecting centrosomes and quantifying centrosome and microtubule organization defects. Centrosomes can be hard to detect because of large variations in the fluorescence intensity and consequently the signal-to-noise ratio. As such, we found that simple image processing techniques often failed to detect these foci accurately.
We, therefore, trained convolutional networks4 to perform this detection.
Convolutional networks are machine learning-based models that can be trained to perform complex tasks, and that have been shown to produce accurate results on a variety of problems (LeCun et al., 1998). We hand-annotated a small dataset of centrosome images and used these as training images to train the convolutional networks. However, convolutional networks have millions of parameters that need to be estimated (He and Sun, 2015). Therefore, training these networks requires large labeled datasets, which are not available for the centrosome detection problem. Convolutional networks are also computationally expensive, making them challenging to apply out of the box in the large, microscopic images that we acquired. To address these challenges, we defined a new convolutional network architecture that was both
4 Convolutional networks: Convolutional networks are machine learning models comprised of sequences of convolution operations interspersed with subsampling operations. The filters of these convolutions are automatically estimated by the learning algorithm based on a training dataset consisting of pairs of images and the desired outputs.
90 parameter-efficient and computation-efficient. Using this algorithm, we automated and validated the detection of centrosomes and microtubule organization defects in breast cancer cells. This is a new and semi-automated tool available to detect centrosome and microtubule organization defects in cells.
Results
An Image Processing Algorithm to Quantify Centrosome and Microtubule Organization
Defects
We designed a tool with a graphical user interface (GUI) and an underlying image processing algorithm to help quantify centrosome and microtubule organization.
The tool calculates the density of centrosomal proteins, microtubules and EB3 foci in concentric rings around the centrosomes to analyze the spatial distribution at and around the centrosome (Figure 3.1A). A detailed description of the algorithm follows.
We first describe the analysis pipeline of PCM and microtubule density and then describe the analysis of EB3 foci. For the PCM microtubule density analysis, first, the tool asks the user to pick a cell for analysis and click on the centrosomes of the cell on the image. Next, the tool computes the centroid of the centriole locations and calculates the distance of each pixel from this centroid. Then, for each radius from r1 to rn identified by the user, it computes the total fluorescence intensity In of a disk of pixels that have a distance less than that radius from the centroid of centrosomes. The difference in total intensity In+1 – in between the n-th and (n+1)-th disks gives us the differential fluorescence intensity in the n-th concentric ring.
The concentric rings become larger as the radius increases, so it is necessary to normalize the differential intensity by the number of pixels considered for each ring. In
91 performing this normalization, we need to ensure that we only count pixels that fall within the cell. To identify these pixels, we used a simple low threshold on the microtubule fluorescence intensity; this was motivated by the fact that microtubules extend throughout the cell area. We verified for each cell manually that this process was correctly identifying the cellular area. Dividing the differential intensity by the differential cell area gives us the microtubule density (Figure 3.1B). We perform a similar analysis of the distribution of fluorescence intensity along centrosomes for PCM proteins, including -tubulin and pericentrin (3.3A-3.3F). In this case, we used smaller disks
(~0.065 m) that always land completely inside the cell. Therefore, we did not attempt to identify and calculate the cell area and simply divided by the difference in areas of consecutive disks to compute the average fluorescence intensity.
We were interested in understanding not just the distribution of the PCM proteins around the centrosomes, but also the distribution of microtubules around centrosomes.
We, therefore, designed an algorithm to understand the distribution of EB3 foci around the centrosomes (which in turn correspond to the growing ends of microtubules) (Figure
3.1C). To detect EB3 comets, median filtering5 was first performed to remove any noise.
Then, to identify pixels with a higher intensity than their neighbors, we performed high- pass filtering: a blurred version of the image (obtained by convolving6 the image with the
5 Median filter: A median filter replaces each pixel by the median value of its neighbors. This is especially useful for removing salt-and-pepper noise, where the noise takes the form of some pixels having very high or very low values of intensity.
6 Convolution: The convolution operation takes an image and a filter and produces a new image. The filter is represented as a 2D matrix of size k x k. The convolution operation then takes k x k neighborhoods in the image, multiplies it element-wise with the filter, and sums up the result to produce the output value for the center pixel.
92 box filter7) was subtracted from the image. Thresholding this difference gave a binary image which identified pixels that lay on EB3 comets. Finally, each connected component8 in this binary image was considered an EB3 comet. Analyzing the distribution of EB3 comets in concentric rings as above requires additional considerations since EB3 comets can straddle multiple rings.
To make sure each EB3 comet was counted exactly once, for every radius, we only counted the EB3 comet that both began and ended within that radius. For each radius rn, this gives us the number of comets En that completely fall within a distance of rn pixels from the centroid of centrosomes. We then compute the differential EB3 counts in the nth ring as the difference in comet counts En – En+1. As before, we divide this by the differential cell area to get the EB3 density (Figure 3.1D). The tools we have described here can be generalized to analyze the distribution of other proteins around centrosomes. However, the analysis still requires the user to click each individual centrosome. We next propose ways for automating this step too, to produce a complete semi-automated pipeline.
7 Box filter: A k x k box filter is a filter filled with a constant value of 1/k2. Convolving with this filter replaces every pixel by the average of its neighbors in a k x k neighborhood.
8 Connected component analysis: Given a binary image, connected component analysis identifies contiguous regions of the image which have a `True’ value. This analysis is done by interpreting the image as a graph where True pixels are nodes and there are edges between every pair of neighboring pixels. Connected components in this graph correspond to contiguous regions in the image.
93 A B Centroid of centrioles
Cep192 Cep192
Amplifiedcentrosome
Non-amplified centrosome 2.5m
Microtubules,
varying radii varying
Amplifiedcentrosome
Non-amplified Non-amplified centrosome
circle of of circle
Area of cell under the the under of cell Area C D 1
High- Median pass filter Filtering
2 -tubu lin,
γ
circle of varying radii, varying of circle
mNeon-E B3,
EB3 foci under circle of interest interest of circle under foci EB3
3
Threshold
varying radii
Area of cell under the the under cell of Area
circle of
Figure 3.1: An image processing algorithm to quantify centrosome and microtubule organization defects.
94 Figure 3.1: An image processing algorithm to quantify centrosome and microtubule organization defects. (A)Top panel, Schematic of analysis of the spatial distribution of pericentriolar material at non-amplified (left) and amplified centrosomes (right). The region around the centroid of the centrioles is divided into concentric rings, and the density of peri-centriolar material is calculated in each ring. Bottom panel, non-amplified and amplified centrosomes in MDA-231 cells stained for centrioles (centrin, red) and γ- tubulin (green). (B) Top panel, Schematic of analysis of microtubule density for each ring. Middle left panel, Schematic for analysis of differential microtubule intensity, Middle right panel, centrosomes in MDA-231 cells stained for centrioles (CEP192, red) and microtubules (-tubulin, green). Bottom left panel, Schematic for analysis of the differential area, Bottom right panel, thresholded image (C) Pipeline for detecting EB3 foci. 1: original image, 2: processed image after applying a median filter and subtracting a blurred version of the same image, blurred using a box filter. Pixels on EB3 foci have high intensity and all other pixels have been zeroed out. 3. thresholded binary image. Connected components in this image are considered as EB3 foci. (D) Top panel, Schematic of analysis of EB3 density for each ring. Middle left panel, Schematic for analysis of differential EB3 counts, Middle right panel, centrosomes in tet-induced mNeon-EB3 MCF10A cells stained for centrioles (centrin, green) and γ-tubulin (red). Bottom left panel, Schematic for analysis of the differential area, Bottom right panel, thresholded image. Scale bars, 1 μm.
95 A Semi-Automatic Machine Learning Algorithm Quantifies Centrosome and Microtubule
Organization Defects
We designed a semi-automatic algorithm using machine learning to remove the need to click individual centrosomes by automatically detecting centrosomes. These foci can be hard to detect because of large variations in the signal-to-noise ratio. As such, we found that simple image processing techniques often failed to detect these foci accurately. We designed an alternative algorithm for detecting centrosomes that is adapted from existing work in object detection in the field of computer vision (Liu et al.,
2016). This prior work uses convolutional networks5 to assign a score to every location in the image. The score output at a particular location is interpreted as the probability of an object at that location. The convolutional network is trained using backpropagation on a dataset consisting of images with annotated locations of objects. We found that prior convolutional network architectures used in the computer vision community have a lot of parameters that need to be estimated during training. This, in turn, translates into a requirement for large amounts of labeled training data, which we do not have for our problem. These architectures are also expensive to run, requiring large amounts of memory and time, especially if we want to use them on large microscopic images.
To address these challenges, we designed a new convolutional network architecture that requires less training data (because it has fewer parameters that must be estimated) and less memory and time to run. This network takes as input an image which has been normalized so that the mean is 0 and the standard deviation is 1. It outputs a score for each pixel in the image that indicates the probability that the pixel is a centrosome. This score is the product of four terms. Two of these terms are noisy but
96 high-resolution signals obtained by applying a sigmoid function on the fluorescence intensity for centriolar and PCM markers. The third and fourth terms are low-resolution
(by a factor of 8) but accurate outputs from a small 4-layer fully convolutional network operating on each individual channel. All outputs are up-sampled using nearest- neighbor up-sampling to the size of the image and converted to a score between 0 and
1 using the sigmoid6 function (f(x) = 1/(1+e-x)) before being multiplied together. This multiplicative combination keeps the final output resolution high and accurate, but keeps the model small (fewer than 10,000 parameters) and efficient (fewer than 7 gigaflops).
To localize the centrosome in this output, we performed non-maxima suppression to identify peaks. This is standard practice in object detection methods in computer vision
(Viola and Jones, 2018). Concretely, the pixels in the image were considered sequentially in decreasing order of their score. The first (and thus highest scoring) pixel was declared as a centrosome. The subsequent pixels were declared as centrosomes if they were more than r=5 pixels away from all other previously declared centrosomes. r was chosen keeping in mind the resolution of the output produced by the convolutional network. This process was continued until the score of the centrosomes fell below a user-specified threshold, or the number of predicted spurious foci exceeded a large threshold (500), whichever came earlier. The full centrosome detection pipeline is shown in Figure 3.2A and some example detections are shown in Figure 3.2B.
To train the model, we annotated centrosomes on a small dataset of 10 images and used these as training images to train the convolutional networks. To evaluate this centrosome detection approach and make sure that it was indeed detecting centrosomes correctly, we annotated another image that the network did not see during
97 training. We then matched every predicted centrosome to the nearest hand-annotated centrosome, considering it a correct detection if it was within 5 pixels of a hand- annotated centrosome. However, if multiple predicted centrosomes matched the same human-annotated centrosome, only one was considered correct. We then measured the precision, or the fraction of predicted centrosomes that were deemed correct, and the recall, or the fraction of human-annotated centrosomes that were detected. An ideal algorithm would detect all centrosomes and only correct centrosomes, achieving precision and recall of 100%. We plotted how precision varies with recall as the score threshold for declaring a centrosome is reduced (Figure 3.2C). While not perfect, the centrosome detector maintains a precision of 75% even at high recall. We next evaluate the ability of the centrosome detector to accurately localize the centrosome (Figure
3.2D). We vary the distance threshold at which centrosome predictions are considered correct: detections farther from this threshold are considered incorrect. At each distance threshold, we compute the average precision or the precision averaged over multiple values of recall. We find that the centrosome detector maintains a high average precision even for stringent thresholds (a distance of 5 pixels corresponds to ~0.65 microns), indicating that the detector is accurate when it comes to localization. Finally, we asked how resilient the detector is to noise. We artificially added Gaussian noise to the image and computed the precision versus recall curve for the centrosomes identified on the noisy image (Figure 3.2E). It can be seen that in spite of a significant amount of noise, the centrosome detector is consistent in its ability to detect centrosomes.
However, centrosome detection alone is not sufficient. Our analysis requires that we analyze each cell separately. We therefore also need to identify cells and group the
98 detected centrosomes into cells. To do this, we created a pipeline for segmenting the cells in the image (Figure 3.2F). In the first step, a convolutional network identifies all pixels that fall inside a cell. This convolutional network is trained using a small dataset of 10 images where the cells have been annotated by hand. The output of this network can be interpreted as a probability p(x) for each pixel x that indicates whether it falls inside a cell. The next step is to group the pixels with a high p(x) into separate cells. To do this, we identify local peaks in this output as markers for possible cells and use a random walk segmenter (Grady, 2006) to segment the cell pixels by assigning each cell pixel to one of the markers. However, the random walk segmenter can over-segment and subdivide a single cell into multiple cells. We, therefore, estimate the strength of the boundary between the cells using the convolutional network output. We define this boundary strength as , where the summation is over all pixels on the 푛 boundary (Figure 3.2.1A).∑푖=1 (1We − then 푝(푥푖 )use)/푛 a user-defined threshold to merge cells that are separated by weak boundaries. We qualitatively compared the result of the semi- automatic segmentation approach to a human-annotated cell. It can be seen that the machine-generated segmentation does not accurately estimate cell boundaries, but it does capture most of the cellular area correctly (Figure 3.2G). Together, these results suggest that our centrosome detection and cell segmentation approach can indeed help speed up analysis, but may need manual intervention occasionally when predictions are incorrect. In what follows, I use these tools to analyze the distribution of pericentriolar proteins, microtubules and EB3 foci around centrosomes in cancer cells.
99 B Raw Processed A Convolutional Network on Image Image sigmoid Centriolar
Machine learned Machine
marker Processed Image centrosomes
sigmoid Centrin,
NMS Centrosomes
Chromosomes
-tubulin,
γ Raw Image sigmoid
Convolutional Network on sigmoid PCM marker Centrosome Detection -Noise +Noise CD E
-tubulin,
Centrin
125% γ 100% 125% No noise 100% Noise = 16 x standard deviations 80% 100% 75% 60% 75%
50% 40% 50%
(Precision) 25% 20% 25%
Correct detectionsCorrect (%)
Correct detections (%) 0% Average (%)Precision 0% 0% 0% 25% 50% 75% 100% 0 2 4 6 8 10 12 14 16 18 0% 20% 40% 60% 80% 100% Centrosomes detected (%) Distance threshold (pixels) Centrosomes detected (%) (Recall) Cell boundary Model F 1 4 G
Centrosomes
Centrin Centrin
Threshold Convolutional annotated Human
γ
and -tubulin, Network connected components
2 3
Random walk
segmenter Chromosomes,
Machine learned learned Machine
Figure 3.2: A semi-automatic machine learning algorithm that quantifies centrosome and microtubule organization defects.
100 Figure 3.2: A semi-automatic machine learning algorithm that quantifies centrosome and microtubule organization defects. (A) Machine learning algorithm for centrosome detection. Separate convolutional networks operating on centrin and pericentrin assign a score for each pixel indicating the likelihood that it is a centrosome. These are passed through a sigmoid function (f(x) = 1/(1+e-x)) and multiplied together with the intensities of the centrin and pericentrin channels, also after a sigmoid, to produce a final score for each pixel. Peaks in the final score are detected using non-max suppression and identified as centrosomes. (B) Left panel, Raw Image of centrosomes stained for centrioles (centrin, red) and γ-tubulin (green) Right panel, detected centrosomes (red). (C) Evaluation of the centrosome detector in terms of the fraction of detections that are deemed correct (precision) and the fraction of centrosomes detected (recall). The plot shows precision and recall as the detector is made less conservative by reducing the threshold score at which a centrosome is detected. (D) Variation of average precision (averaged over multiple recall values) as the criterion for a correct detection is made more lenient: the x-axis is the maximum distance between a true and a predicted centrosome at which the predicted centrosome is still considered correct. (E) Precision and recall values for a noisy image compared to the original image. (F) Pipeline for segmenting out individual cells. 1 is the original image, 2 is the output of the neural network that identifies pixels that fall within cells, 3 is the boundary map obtained from the random walk segmenter with estimated boundary strength, 4 is the final segmentation. (G): Comparison between predicted cell segmentation and human- annotated cells. Scale bars, 1 μm.
101 A
Random walk segmenter
Over-segmentation of cells
Cell boundary
Estimated boundary strength Figure 3.2.1: Supplemental, pipeline for estimating boundary strength based on the convolutional network output, p(x). The random walk segmenter over-segments cells (top right). We identify the boundary between the predicted cells (bottom left) and estimate the strength of the boundary (bottom) using 1-p(x) (bottom right).
Amplified Centrosomes Possess Higher Microtubule Density
I utilized the above-designed algorithm to analyze the distribution of PCM proteins in cancer cells. To ensure correct analysis, I first performed the analysis using the manual annotation of centrosomes (Figure 3.3) and then validated the results using the semi-automatic approach (Figure 3.5). Cancer cells have an increase in the number of centrioles (Marteil et al., 2018; Ganapathi Sankaran et al., 2019). Each centriole is surrounded by PCM proteins such as -tubulin and pericentrin, that forms a toroid structure (Fu and Glover, 2012; Mennella et al., 2014). Using the image processing analyses pipeline, I asked whether amplified centrosomes have altered levels and distribution of the pericentriolar proteins, -tubulin and pericentrin. I utilized breast cancer cells (MDA-231) that have approximately 23% of CA relative to normal-like breast cells (MCF10A) that have 5% of CA (Ganapathi Sankaran et al., 2019). Using the above image processing algorithm to analyze PCM density, I calculated the relative
102 intensity of -tubulin and pericentrin per unit area in normal versus amplified centrosomes of breast cancer cells.
I find that amplified centrosomes have a higher density of -tubulin relative to non-amplified centrosomes (Figure 3.3A). In comparison to the amplified centrosomes of MCF10A cells, the amplified centrosomes of MDA-231 cells have higher -tubulin density (Figure 3.3B and 3.3C). The relative -tubulin density diminishes outside the pericentriolar toroid similar to the non-amplified centrosomes (Figure 3.3B and 3.3C).
This suggests that while the density is elevated the distribution of -tubulin remained encompassed within the pericentriolar space of the amplified centrosomes. Consistent with the increase in density of -tubulin, pericentrin is also elevated in centrosome amplified breast cancer cells (Figure 3.3D, 3.3E and 3.3F). However, unlike the distribution of -tubulin, relative pericentrin density does not diminish immediately outside the pericentriolar toroid. This suggests that the pericentrin expands the pericentriolar toroid of amplified centrosomes (Figure 3.3F).
The increase in -tubulin in the amplified centrosomes suggests that it may promote the microtubule density around the amplified centrosomes. In order to ask whether cells with amplified centrosomes have altered microtubule density, I measured the microtubule fluorescence intensity per unit area using the above image processing algorithm. I find that microtubule density is higher in cells with CA in both MCF10A cells and MDA-231 cells (Figure 3.3G, 3.3H, and 3.3I). Overall, these data suggest that centrosome amplified breast cancer cells have an increase in -tubulin that may promote the microtubule nucleation and thereby increase the microtubule density.
103 A MCF10A MDA-231 B MCF10A C MDA-231 1.6 1.6 Non-amplified centrosomes
****2 **** 1.4 Amplified centrosomes
2 1.4 Non-amplified centrosomes m )
m ) Amplified centrosomes 1.2 1.2
1.0 1.0
centrosome
Non-amplified Centrin 0.8 0.8 0.6
0.6 -tubulin
γ -tubulin 0.4 γ 0.4
0.2 Intensity/Area (U/ 0.2
Intensity/Area (U/
-tubulin,
0.0 Relative 0.0 fluorescence
γ
Amplified
Relative 0.13fluorescence 0.39 0.65 0.91 1.17 1.43 1.69 1.95 2.21 2.47 0.130.390.650.911.171.431.691.952.212.47 centrosome Distance from the centroid of Distance from the centroid of centrioles ( m) centrioles ( m)
D MCF10A MDA-231 E MCF10AF MDA-231 1.6 1.6 Non-amplified centrosomes Non-amplified centrosomes 2 1.4 Amplified centrosomes 2 1.4 **** Amplified centrosomes m )
m ) ****
1.2 A 1.2 1.0 Centrin 1.0
centrosome Non-amplified 0.8 0.8
0.6 0.6 0.4 0.4 A 0.2 (U/Intensity/Area 0.2
Intensity/Area (U/
0.0
Amplified 0.0
Pericentrin,
Relativepericentrin fluorescence centrosome 0.130.39 0.650.91 1.171.43 1.69 1.952.21 2.47 0.130.390.650.911.171.431.691.952.212.47
Relativepericentrin fluorescence Distance from the centroid of Distance from the centroid of centrioles ( m) centrioles ( m) G MCF10A MDA-231 HI
MCF10A MDA-231
2
MCF10A Non-amplified centrosomes 2 1.6 Non-amplified centrosomes
1.6 MCF10A Amplified centrosomes Amplified centrosomes 1.4
centrosome ****
Non-amplified 1.4 **** 1.2
Centrosome 1.2 1.0 1.0 0.8 0.8
tubulin, 0.6 Microtubuledensity(U/ m0.6 )
Microtubuledensity(U/ m )
Amplified 0.5 1 1.3 2.6 3.9 5.2
- 0.5 1 1.3 2.6 3.9 5.2 centrosome Distance from centrosome ( m) Distance from centrosome ( m)
Figure 3.3: Centrosome amplified breast cancer cells have pericentriolar and microtubule organization defects.
104 Figure 3.3: Centrosome amplified breast cancer cells have pericentriolar and microtubule organization defects. (A) Left panel, non-amplified and amplified centrosomes in MCF10A cells stained for centrioles (centrin, red) and γ-tubulin (green). Right panel, non-amplified and amplified centrosomes in MDA-231 cells stained for centrioles (centrin, red) and γ-tubulin (green). (B) Relative γ-tubulin fluorescence density in non-amplified (black) and amplified (red) MCF10A cells. (C) Relative γ-tubulin fluorescence density in non-amplified (black) and amplified (red) MDA-231 cells. (D) Left panel, non-amplified and amplified centrosomes in MCF10A cells stained for centrioles (centrin, red) and pericentrin(cyan). Right panel, non-amplified and amplified centrosomes in MDA-231 cells stained for centrioles (centrin, red) and pericentrin (cyan). (E) Relative pericentrin fluorescence density in non-amplified (black) and amplified (red) MCF10A cells. (F) Relative pericentrin fluorescence density in non- amplified (black) and amplified (red) MDA-231 cells. (G) Left panel, non-amplified and amplified centrosomes in MCF10A cells stained for centrioles (CEP192, red) and microtubules (-tubulin, green). Right panel, non-amplified and amplified centrosomes in MDA-231 cells stained for centrioles (CEP192, red) and microtubules (-tubulin, green). (H) Relative microtubule fluorescence density in non-amplified (black) and amplified (red) MCF10A cells. (I) Relative microtubule fluorescence density in non- amplified (black) and amplified (red) MDA-231 cells. Mean±SEM. Wilcoxon test and Students t-test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
105 Centrosome Amplified Breast Cancer Cells have Longer and Higher Density of
Microtubule Growing Ends
To test whether increased -tubulin density indeed promotes nucleation and increases the growing ends of microtubules in centrosome amplified cancer cells; I created a tetracycline-inducible mNeon-EB3 tagged MCF10A and MDA-231 cell lines to track the growing ends of microtubules. We added tetracycline for 48 hours to express of mNeon tagged EB3 and measured the number of EB3 comets using the EB3 image processing algorithm. The EB3 image processing algorithm accurately detects and counts the number of EB3 foci per unit area (Figure 3.4B and 3.4E). I observed a 20% and 40% increase in the EB3 density around the amplified centrosomes relative to the non-amplified centrosomes in MCF10A and MDA-231 cells respectively. This data suggests that amplified centrosomes have higher EB3 density or higher density of growing ends of microtubules around them in comparison to the non-amplified centrosomes (Figure 3.4A, 3.4C, 3.4D, and 3.4F).
Furthermore, the number of EB3 comets per unit area is higher closer to the centrosomes in comparison to the periphery of the cells (Figure 3.4A and 3.4D). This is consistent with the data that microtubule density is higher closer to the centrosome relative to the periphery of the cells (Figure 3.3H and 3.3I). This suggests that growing microtubules are at a higher density closer to the centrosomes than the periphery of the cells. In comparison to the MCF10A cells, the EB3 density decreases dramatically towards the periphery of the cells in the MDA-231 cells (Figure 3.4C and 3.4F). This suggests that the MCF10A cells may have longer microtubules or increased microtubule branching relative to MDA-231 cells.
106 Surprisingly, the length of EB3 comets was altered in the centrosome amplified breast cancer cells relative to the non-amplified cells. The average comet length was
30% and 15% longer in centrosome amplified cells relative to their non-amplified
MCF10A and MDA-231 cells (Figure 3.4G, 3.4H, and 3.4I). EB proteins recognize the growing end of microtubules and sense the nucleotide conformational transitions
(Maurer et al., 2012, 2014). A longer comet may suggest a slower GTP hydrolysis at the growing microtubule ends. Increased microtubule assembly rates can also alter the length of EB3 comets. Hence, microtubules from amplified centrosomes may have higher microtubule assembly rates or slower GTP hydrolysis. Furthermore, in comparison to centrosome amplified MDA-231 cells, centrosome amplified MCF10A
EB3 comets have a much higher increase in comet length relative to their non-amplified counterparts. This suggests that the centrosome amplified MCF10A cells may have higher microtubule assembly rates relative to centrosome amplified MDA-231 cells
(Figure 3.4G, 3.4H, and 3.4I). However, the total EB3 density is lower in amplified
MCF10A cells in comparison to amplified MDA-231 cells (Figure 3.4C and 3.4F). This suggests that amplification in MCF10A cells may promote the formation of longer growing microtubules. In contrast, amplification in MDA-231 cells can promote a higher number of shorter growing microtubules. Overall, these data suggest that EB3 density and EB3 comet length are higher in centrosome amplified breast cancer cells relative to non-amplified cells.
107 A
Non-amplified Amplified Tet-induced mNeon-EB3 A B C MCF10A centrosomes centrosomes EB3 counts Area 1.8 2 Amplified centrosomes 1.6 Non-amplified centrosomes 1.4
-tubulin 1.2
γ *
1.0
MCF10A MCF10A 0.8
0.6 Tet-induced mNeon-EB3 mNeon-EB3 Tet-induced mN eon-EB3, density (counts/EB3 m ) 2.6 5.2 7.8 10.4 13 15.6 Distance from centrosome ( m)
Non-amplified Amplified centrosome EB3 counts Area DE centrosome F Tet-induced mNeon-EB3 MDA-231
G 2 1.8 Amplified centrosomes
* Centrin 1.6 Non-Amplified centrosomes
1.4
-tubulin, 1.2
γ
,
MD A-231 1.0
0.8
EB3 density (counts/ m ) 0.6
Tet-induced mNeon-EB3 Tet-induced mNeon-EB3 2.6 5.2 7.8 10.4 13 15.6 Distance from centrosome ( m)
G HI Tet-induced mNeon-EB3 Tet-induced mNeon-EB3 MCF10A 1.8 MCF10A 1.8 MDA-231 Non-Amplified *
centrosomes m) 1.7 m) 1.7
Amplified 1.6 1.6 centrosomes 1.5 1.5 MDA-231 1.4 **** 1.4 Non-Amplified centrosomes 1.3 1.3 1.2 1.2 Amplified centrosomes 1.1 1.1
EB3 comet length( 1.0 EB3 comet length(1.0 Non-Amplified Amplified Non-Amplified Amplified
Figure 3.4: Centrosome amplified breast cancer cells have higher EB3 density and comet length. (A) Non-amplified and amplified centrosomes in tet-induced mNeon-EB3 MCF10A cells stained for centrioles (centrin, green) and γ-tubulin (red). (B) Processed non-amplified tet-induced mNeon-EB3 MCF10A cells utilized for measurement of EB3 counts and area. (C) Relative EB3 counts per unit area in non-amplified (black) and amplified (red) MCF10A cells. (D) Non-amplified and amplified centrosomes in tet- induced mNeon-EB3 MDA-231 cells stained for centrioles (centrin, green) and γ-tubulin (red). (E) Processed amplified tet-induced mNeon-EB3 MDA-231 cells utilized for measurement of EB3 counts and area. (F) Relative EB3 counts per unit area in non- amplified (black) and amplified (red) MDA-231 cells. (G) Top panel, mNeon-EB3 comets in non-amplified and amplified MCF10A cells. Bottom panel, mNeon-EB3 comets is non-amplified and amplified MDA-231 cells. (H) Relative EB3 comet length in non- amplified (black) and amplified (red) MCF10A cells. (I) Relative EB3 comet length in non-amplified (black) and amplified (red) MDA-231 cells. Mean±SEM. Wilcoxon test and Students t-test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
108 Semi-Automatic Machine Learning Algorithm recapitulates the Pericentriolar Defects in
Amplified Centrosomes of Breast Cancer Cells
I found that the centrosome detection model and the cell segmentation model do not always give correct results on all images. We, therefore, created a GUI to allow us to correct any errors that the algorithm makes (Figure 3.5A). The GUI allows the user to click on new centrosomes, remove incorrectly detected foci, add new cell boundaries for cells that have been merged, merge cells and remove incorrectly machine annotated centrosomes that the algorithm erroneously over-segmented.
Using this GUI, I first asked whether I could recapitulate the pericentriolar defects observed in amplified centrosomes of breast cancer cells. In corroboration to the image processing algorithm, I find that amplified centrosomes indeed have a higher density of
-tubulin relative to non-amplified centrosomes (Figure 3.3B, 3.3C, 3.5B, and 3.5C).
While the density is elevated the distribution remained encompassed within the pericentriolar space of the amplified centrosomes. Consistent with the increase in density of pericentrin observed through the image processing algorithm, pericentrin is also elevated in centrosome amplified breast cancer cells (Figure 3.3E, 3.3F, 3.5D, and
3.5E). Furthermore, pericentrin also expands outside the pericentriolar toroid of centrosomes as observed through its quantitation (Figure 3.3F). These data suggest that the semi-automatic machine learning algorithm validates the organization defects at amplified centrosomes analyzed using the image processing algorithm.
109 A
B C MCF10A MDA-231 1.6 1.6 Non-amplified centrosomes 2 ****
2 **** 1.4 Amplified centrosomes 1.4 Non-amplified centrosomes m )
m )
Amplified centrosomes 1.2 1.2 1.0 1.0 0.8 0.8 0.6 0.6
-tubulin
-tubulin γ 0.4 γ 0.4 0.2
0.2 (U/intensity/Area (U/intensity/Area 0.0 0.0 0.130.390.650.911.171.431.691.952.212.47
Relative Relative fluorescence
Relative Relative 0.130.390.65fluorescence 0.911.171.431.691.95 2.212.47 Distance from the centroid of Distance from the centroid of centrioles ( m) centrioles ( m) D MCF10A E MDA-231 1.6 1.6 Non-amplified centrosomes Non-amplified centrosomes 1.4 2 1.4 Amplified centrosomes Amplified centrosomes 2 ****
m ) ****
m )
1.2 1.2
1.0 1.0
0.8 0.8
0.6 0.6 0.4 0.4
0.2 0.2
Intensity/Area (U/
0.0 (U/intensity/Area 0.0 0.13 0.39 0.65 0.91 1.17 1.43 1.69 1.95 2.21 2.47 0.13 0.39 0.65 0.91 1.17 1.43 1.69 1.95 2.21 2.47
Relativepericentrin fluorescence
Distance from the centroid of Relativepericentrin fluorescence Distance from the centroid of centrioles ( m) centrioles ( m) Figure 3.5: Semi-automatic machine learning algorithm recapitulates the image pericentriolar defects in amplified centrosomes of breast cancer cells. (A) Screenshot of GUI used for correcting the output of the automatic centrosome detection and cell segmentation system. The image is shown in the central viewport. (B) Relative γ-tubulin fluorescence density in non-amplified (black) and amplified (red) MCF10A cells. (C) Relative γ-tubulin fluorescence density in non-amplified (black) and amplified (red) MDA-231 cells. (D) Relative pericentrin fluorescence density in non-amplified (black) and amplified (red) MCF10A cells. (E) Relative pericentrin fluorescence density in non- amplified (black) and amplified (red) MDA-231 cells.
110 Centrosome Amplification Promotes Chromosome Mis-Segregation in Breast Cancer
Cells
Centrosome amplified breast cancer cells have increased pericentriolar material, higher microtubule growing ends and density. Such altered microtubule organization around amplified centrosomes can have important implications in several cellular functions.
I examined the functional consequences of CA on chromosome segregation in normal-like (MCF10A), less aggressive (ZR751) and highly aggressive (MDA-231) breast cancer cells, respectively. Aggressive breast cancer cells with CA exhibit an increased percentage of cells with multipolar mitoses (Figure 3.6A-3.6C, (Salisbury et al., 2004)). Furthermore, centrosomes in multipolar spindles often cluster to form bipolar spindles capable of producing viable daughter cells (Quintyne et al., 2005; Ganem et al., 2009; Godinho et al., 2009).
However, clustered multipolar spindles also produce merotelic microtubule- kinetochore attachments, giving rise to lagging chromosomes during anaphase that are subsequently packaged into micronuclei in the following interphase (Cimini, 2008;
Ganem et al., 2009; Silkworth et al., 2009; Thompson et al., 2010; Thompson and
Compton, 2011; Ly and Cleveland, 2017). Similarly, I observe more cells with anaphase-lagging chromosomes and micronuclei in cell populations that had greater
CA (Figure 3.6D and 3.6E). Thus, elevated CA is associated with chromosome missegregation in breast cancer cells.
111 A Bipolar mitosis Multipolar mitosis Clustered Unclustered 12% *** *
9%
Microtubules Microtubules 6%
DNA DNA
Cells with with Cells CREST 3%
multipolar mitoses (%) mitoses multipolar 0%
-tubulin -tubulin
γ MCF10A ZR751 MDA-231 Monopolar Bipolar Tripolar B C Monopolar mitosis Tripolar mitosis 15% Tetrapolar mitosis Pentapolar mitosis >Pentapolar mitosis 12% ***
9%
DNA
CREST 6%
Microtubules
Cells in mitosis (%) in mitosis Cells 3%
-tubulin
γ 0% Tetrapolar Pentapolar Decapolar MCF10A ZR751 MDA-231 D Anaphase Lagging ChromosomesMicronucleiE 25% 15%
20% * 12% *** * 15% 9%
D NA
10% DNA DNA 6%
Cells with Cells
CREST
C REST
Microtubules 5% (%) micronuclei 3% *** *
Cells with anaphase with Cells 0% -tubulin 0%
lagging chromosomes (%) chromosomes lagging γ MCF10A ZR751 MDA-231 MCF10A ZR751 MDA-231
Figure 3.6: CA promotes chromosome missegregation in breast cancer cells. (A) Left panels, bipolar and multipolar mitoses in breast cancer cells. PCM (γ-tubulin, red), microtubules (α-tubulin, green), DNA (Hoechst 33342, blue) and kinetochores (- CREST; grayscale) are labeled. Right panel, the percentage of multipolar cells in the mitotic cell population. (B) Monopolar, bipolar and multipolar mitoses in MDA-231 cells. Cells stained for PCM (γ-tubulin; red), microtubules (α-tubulin; green), DNA (Hoechst 33342; blue) and kinetochores (CREST; grayscale). (C) Percentage of mitotic cells exhibiting monopolar, bipolar and multipolar mitosis. (D) Left panel, anaphase lagging chromosomes in breast cancer cells. Cells are labeled as in (A). Arrow and insets denote lagging chromosomes. Right panel, the percentage of late anaphase cells with lagging. (E) Left panel, micronuclei in breast cancer cells. Cells were stained for DNA (Hoechst 33342; blue) and kinetochores (CREST; white). Arrow and inset denote a micronucleus. Right panel, the percentage of cells with micronuclei. Mean±SEM. Statistical tests compare to MCF10A cells. Fischer’s exact test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1μm.
112 CEP135 Isoform Dysregulation Promotes Chromosome Mis-Segregation in Breast
Cancer Cells
CEP135full expression is elevated in breast cancer cells and is sufficient to increase the number of centrosomes. I asked whether CEP135full dysregulation disrupts mitosis. I utilized the stable tetracycline-inducible CEP135full MDA-231 cell line. I treated mCh-CEP135full-Tet MDA-231 cell lines with tetracycline for three days and quantified chromosome segregation defects in these cells. Upon mCh-CEP135full-Tet induction, cells exhibit substantial cell death and an approximately 5-fold increase in multipolar mitoses (Figure 3.7A). Accordingly, mCh-CEP135full-Tet expressing cells have an increased incidence of anaphase lagging chromosomes and micronuclei (Figure 3.7B,
3.7C). Thus, elevated CEP135full is sufficient to increase chromosome missegregation.
CEP135mini expression reduces centrosome number in breast cancer cells. I next examined the effect of CEP135mini overexpression on mitosis. I utilized a tetracycline- inducible fluorescently labeled GFP-CEP135mini-Tet cell line and measured chromosome segregation defects. Surprisingly, tetracycline-induced GFP-CEP135mini-Tet cells have the same frequency of multipolar divisions as non-induced cells (Figures 3.7D).
However, more tetracycline-induced GFP-CEP135mini-Tet cells exhibit apolar mitotic divisions compared to non-induced controls, likely resulting from CEP135mini-induced centriole underduplication (Figure 3.7D). Elevated GFP-CEP135mini-Tet also increases the number of cells with anaphase lagging chromosomes (Figure 3.7E). Consistent with the increased lagging chromosomes, more GFP-CEP135mini-Tet expressing cells have micronuclei (Figure 3.7F). Thus, elevated CEP135mini expression in breast cancer cells disrupts normal mitotic chromosome segregation.
113 A Bipolar mitosis Multipolar mitosis 18% Clustered Unclustered ** ** 15%
12%
-tubulin -tubulin
γ
DNA DNA f ul l 9%
Cells with Cells 6%
CREST CREST
3%
multipolar mitosis (%) mitosis multipolar
mCh-CEP135 full full γ-tubulin mCh-CEP135 mCh-CEP135 0% - + - Tetracycline +Tetracycline Tetracycline induction
B Anaphase Lagging Chromosomes Micronuclei 50% C 30%
40% * 25% ****
-tubulin 20%
γ 30% CREST
full full
DNA full 15% 20%
DNA Cells with Cells 10%
CREST 10% (%) micronuclei 5%
Cells with anaphase with Cells
lagging chromosomes (%) chromosomes lagging
mCh-CEP135 mCh-CEP135 0% 0% - + - + Tetracycline induction Tetracycline induction
D Bipolar mitosis Multipolar mitosis Apolar mitosis 8% Multipolar mitosis (%) Apolar mitosis (%)
6%
-tubulin
γ
DNA
mini mini 4%
Mitosis(%) 2%
CREST
0% GFP-CEP135 - + - + - Tetracycline +Tetracycline Tetracycline induction
Anaphase Lagging Chromosomes Micronuclei E 21% F 21%
18% 18%
15% 15% ***
-tubulin
γ -tubulin 12% 12%
γ
DNA
mini mini
DNA
9% 9%
mini mini
Cellswith 6% 6%
micro nuclei (%)
CREST
3% 3%
Cellswith anaphase
CREST
0% 0% lagging chro mosomes(%) - + - + Tetracycline induction GFP-CEP135 Tetracycline induction GFP-CEP135
Figure 3.7: CEP135 isoform dysregulation is sufficient to promote multipolar mitosis, anaphase lagging chromosomes, and formation of micronuclei in breast cancer cells.
114 Figure 3.7: CEP135 isoform dysregulation is sufficient to promote multipolar mitosis, anaphase lagging chromosomes, and formation of micronuclei in breast cancer cells. (A) Left panels, bipolar and multipolar mitoses in non-induced and tetracycline-induced mCh-CEP135full-Tet MDA-231 cells stained for PCM (γ-tubulin, green), DNA (Hoechst 33342, blue) and kinetochores (-CREST, grayscale). Right panel, percentage of mitotic cells with >2 poles in non-induced and induced mCh-CEP135full-Tet cells. (B) Left panel, anaphase lagging chromosomes in induced mCh-CEP135full-Tet (red) MDA-231 cells stained for DNA (Hoechst 33342, blue) and kinetochores (-CREST, grayscale). Arrow and inset denote a lagging chromosome. Right panel, percentage of late anaphase cells with lagging chromosomes in non-induced and induced mCh-CEP135full- Tet MDA-231 late mitotic cells. (C) Left panel, micronuclei in induced mCh-CEP135full-Tet (red) MDA-231 cells stained for DNA (Hoechst 33342, blue) and kinetochores (- CREST, grayscale). Arrow and inset denote a micronucleus. Right panel, percentage of interphase cells with micronuclei in non-induced and induced mCh-CEP135full-Tet MDA- 231 cells. (D) Left panels, bipolar, multipolar and apolar mitoses in non-induced and induced GFP-CEP135mini-Tet MDA-231 cells stained for PCM (γ-tubulin, red), DNA (Hoechst 33342, blue) and kinetochores (-CREST, grayscale). Right panels, percentage of multipolar and apolar mitoses in non-induced and induced GFP- CEP135mini-Tet MDA-231 cells. (E) Left panel, anaphase lagging chromosomes in induced GFP-CEP135mini-Tet (green) MDA-231 cells stained for PCM (γ-tubulin, red), DNA (Hoechst 33342, blue) and kinetochores (-CREST, grayscale). Arrow and inset denote a lagging chromosome. Right panel, percentage of late anaphase cells with lagging chromosomes in non-induced and induced GFP-CEP135mini-Tet MDA-231 cells. (F) Left panel, micronuclei in induced GFP-CEP135mini-Tet MDA-231 cells stained for DNA (Hoechst 33342, blue) and kinetochores (-CREST, grayscale). Arrow and inset denote a micronucleus. Right panel, percentage of interphase cells with micronuclei in non-induced and induced GFP-CEP135mini-Tet MDA-231 cells. Mean±SEM. Fischer’s exact test and Mann-Whitney U test. *p<0.05, **p<0.01, ***p<0.005 and ****p<0.0005. Scale bars, 1 μm.
115 Discussion
Semi-Automatic Algorithm for Centrosome Detection
We designed a semi-automatic machine learning algorithm that can significantly speed up analysis and aid in removing bias during analysis since it removes the need for the user to manually identify every centrosome. At the same time, we found that having the ability to correct the algorithm’s predictions was crucial to ensure the correctness of the analysis. An interesting aspect of these corrections which we did not explore in the current work is the ability to use these corrections to further train the underlying machine learning model. This is especially beneficial when collecting enough annotations for training the model is expensive and time-consuming.
While correcting the algorithm’s predictions, we also found several common modes of error. First, unlike natural images, the dynamic range of fluorescence images can vary significantly and naïve training of convolutional networks does not generalize across such variations. We addressed this by normalizing each image independently to have zero mean and unit variance. Second, we found that while the algorithm often correctly identified the rough location of the centrosomes, it often did not correctly estimate their number. We thus had to correct its predictions as to which cells exhibited
CA. Accurate counting has not been addressed in the computer vision literature and deserves further attention. Finally, we found that the algorithm often merged together nearby cells or estimated their boundary incorrectly. In such cases, providing an interface where a user can quickly correct this segmentation proved challenging and requires new insights both at the end of computer vision techniques and on the interface design.
116 PCM at Amplified Centrosomes
Taxol, a microtubule stabilizing drug has been used in chemotherapy for the treatment of breast cancers for the last couple of decades (Pazdur et al., 1993; Schiff and Horwitz, 2006). However, it has been unclear whether there are underlying differences in the microtubule population between normal and transformed breast cells.
To begin with, I investigated the levels and the distribution of PCM proteins -tubulin and pericentrin, nucleators of microtubules, using our image processing and machine learning algorithm in breast cancer cells. The PCM proteins are organized as a toroid around the centrioles during interphase that intensifies and accumulate more PCM proteins during mitosis (Mennella et al., 2012; Woodruff et al., 2014, 2015; Fry et al.,
2017). I find that -tubulin and pericentrin are at a higher density at interphase amplified centrosomes in breast cancer cells (Figure 3.3A-3.3F). This increase in density at the amplified centrosomes during interphase in breast cancer cells is reminiscent of the
PCM proteins accumulation during mitosis in normal cells.
At the same time during mitosis in normal cells, the PCM also enlarges in size
(Woodruff et al., 2014). Similarly, I find that the distribution of pericentrin also expands at the interphase amplified centrosomes (Figure 3.3D-3.3F). How this untimely expansion occurs and impacts the cellular function is not known. Phosphorylation of
CDK5RAP2 and pericentrin has been implicated in the expansion of PCM during mitosis
(Haren et al., 2009; Lee and Rhee, 2011; Woodruff et al., 2015). The untimely expansion of pericentrin during interphase at amplified centrosomes can suggest that phosphorylation of PCM proteins may be misregulated at amplified centrosomes in breast cancer cells. Interestingly, the accumulation of additional PCM proteins and the
117 expansion of the PCM toroid involves a microtubule-dependent flaring mechanism
(Conduit et al., 2010). Such a flaring mechanism involves the flux of PCM proteins from the inner to the outer regions of PCM toroid during mitosis. The untimely expansion of
PCM proteins at amplified centrosomes can suggest a disruption of microtubule- dependent flaring mechanism in breast cancer cells. Furthermore, expansions of PCM may interfere with the microtubule-motor-dependent trafficking pathways that have been implicated in disrupted ciliary biogenesis (Galati et al., 2018). Overall, the accumulation of more PCM proteins in amplified centrosomes can disrupt several microtubule-related processes.
Centrosome Amplification Affects Microtubule Organization
Next, I asked whether interphase cytoplasmic microtubule network is altered in centrosome amplified breast cancer cells. The density of microtubules is elevated at
MCF10A amplified centrosomes relative to the non-amplified cells. The microtubule density at the MCF10A non-amplified centrosomes decreases dramatically towards the periphery of cells. In contrast, MCF10A amplified centrosomes have a higher microtubule density at the periphery of cells (Figure 3.3G-3.3I). Some plausible explanations to an increased microtubule density from MCF10A amplified centrosomes at the periphery of cells include the presence of longer microtubules or increased branching of the microtubule network (Petry et al., 2013; Ishihara et al., 2014). In corroboration to the increase in microtubule density at MCF10A amplified centrosomes,
MDA-231 cells also have elevated microtubule density at amplified centrosomes.
However, in contrast to the elevated peripheral microtubule density at MCF10A amplified centrosomes, peripheral microtubule density at MDA-231 amplified
118 centrosomes decreases drastically (Figure 3.3H and 3.3I). An increase in microtubule density around amplified centrosomes can suggest an increased microtubule assembly.
This would suggest an elevated tubulin concentration to facilitate this increased assembly. It is not known whether the cytoplasmic tubulin concentrations are indeed altered in cancer cells to facilitate this increased microtubule assembly. Furthermore, an increased microtubule density can lead to increased trafficking on additional microtubules (Caviston and Holzbaur, 2006). Increased trafficking through such altered cytoskeletal architecture has been implicated in the disruption of important cellular processes such as cell migration and cell polarization (Siegrist and Doe, 2007; Bouchet and Akhmanova, 2017).
An increase in microtubule density can be due to increased microtubule assembly rates. This is consistent with the observation that some colorectal cancer cells have higher microtubule assembly rates (Ertych et al., 2014). However, whether the centrosome amplified population has higher microtubule assembly rates was not investigated in this study (Ertych et al., 2014). I investigated whether EB3 density is altered in amplified and non-amplified breast cancer cells. I find that the EB3 density is elevated in centrosome amplified breast cancer cells relative to the non-amplified cells.
Furthermore, EB3 density dramatically decreases towards the periphery of centrosome amplified MDA-231 cells relative to amplified MCF10A cells (Figure 3.4A-3.4I). This is consistent with the observation that microtubule density decreases towards the periphery of amplified MDA-231 cells relative to amplified MCF10A cells (Figure 3.3G-
3.3I). An increased EB3 density around amplified centrosomes suggests an increase in the density of microtubule growing ends. Increased microtubule assembly rates can be
119 a plausible mechanism for the increased density of microtubule growing ends at amplified centrosomes (Matov et al., 2011). Increased microtubule assembly rates can also alter the length of EB3 comets. Consistent with this idea, I observe that cells with amplified centrosomes have longer EB3 comets relative to the non-amplified centrosomes (Figure 3.4G-3.4I). An increase in EB3 comet length can occur due to an increased microtubule assembly rates or reduction in GTP hydrolysis rate (Maurer et al.,
2012). These underlying mechanisms remain to be explored and whether these are direct effects of amplification of centrosomes is also not known. Indeed, the increase in
EB3 comet length and density at amplified centrosomes reflect intrinsic differences in microtubules from non-amplified and amplified cells.
CEP135 Isoform Dysregulation in Chromosome Segregation
CA promotes chromosome missegregation through the formation of multipolar spindles, anaphase lagging chromosomes, and micronuclei. I investigated the consequences of CEP135 isoform dysregulation on chromosome segregation.
CEP135full overexpression promotes centriole overduplication and CA in cancer cells. In corroboration with the increase in the CA, CEP135full overexpression leads to the formation of multipolar spindles. However, the frequency of cells with anaphase lagging chromosomes is even further elevated relative to the frequency of multipolar spindles formed. CEP135full has a microtubule binding domain and I hypothesize that the increase in anaphase lagging chromosomes in CEP135full overexpressed cells may be mediated by altered microtubule dynamics upon CEP135full overexpression (Figures
3.7A-3.7C).
120 CEP135mini overexpression did not change the formation of multipolar spindles, however, led to the formation of apolar mitotic cells. The fate of these apolar cells along the cell cycle is not clear. Surprisingly, elevated CEP135mini levels increase anaphase lagging chromosomes and micronuclei, two phenotypes commonly associated with cells that have amplified centrosomes (Figures 3.7D-3.7F). The severe loss of γ-tubulin in elevated CEP135mini cells disrupts centrosome microtubule nucleation (data not shown) and may thereby disrupt mitotic progression. Thus CEP135 isoform dysregulation is sufficient to promote chromosome segregation errors in cancer cells.
Materials and Methods for Chapter III
Cell Culture
Breast cancer cell lines MCF10A, ZR-75.1 (ZR751), and MDA-MB-231 (MDA-
231) were obtained from the University of Colorado Cancer Center Tissue Culture Core.
Mammalian tissue culture lines were all grown at 37°C with 5% CO2. MCF10A cells were received at passage 51 and were grown in DMEM/F12 (Invitrogen #11330-032),
5% Horse Serum (Invitrogen #16050-122), 20 ng ml-1 EGF (Invitrogen #PHG0311), 0.5 mg ml-1 Hydrocortisone (Sigma #H-0888), 100 ng ml-1 Cholera toxin (Sigma #C-8052),
10 μg ml-1 Insulin (Sigma #I-1882) and 1% Pen/Strep (Invitrogen #15070-063). MDA-
MB-231 cells were received at passage 15 were grown in DMEM (Invitrogen #11965-
092), Pen/Strep (Invitrogen #15070-063) and 10% FBS (FBS; Gemini Biosciences). ZR-
75.1 (ZR751) cells were received at passage 51, ZR-75.1 cells were grown in RPM1
(Invitrogen #11875-093), 10% FBS (FBS; Gemini Biosciences) and Pen/Strep
(Invitrogen #15070-063). Cell lines were authenticated at the sources and tested negative for mycoplasma using the MycoAlert mycoplasma detection kit through the
121 University of Colorado Cancer Center Tissue Culture Core. Cells were passaged and sub-cultured using Trypsin (Invitrogen #150901-046) when cultures reached 60-80% confluency.
Generation of Tetracycline Inducible mNeon-EB3-MCF10A and mNeon-EB3-MDA-231 and mCherry-CEP135full-Tet or GFP-CEP135mini-Tet Cells
The generation of tetracycline-inducible mNeon-EB3 construct is described below. The EB3-mNeon (C-terminal fusion) fragment was obtained through
PCR with Phusion DNA polymerase of a pre-existing plasmid using primers that have
Nhe1 and Xma1 sites appended to them. This was cloned into the tetracycline-inducible construct pcw57.1 using the enzymes Nhe1 and Age1.
Lentivirus harboring tetracycline-inducible mNeon-EB3-MCF10A, MDA-231, mCherry-CEP135full-Tet, and GFP-CEP135mini-Tet were made by transfection of 293FT cells. 293FT cells were plated in 6 cm dishes and allowed to reach 50%-70% confluency. Cells were then transfected with tetracycline-inducible mCherry-CEP135full-
Tet, GFP-CEP135mini-Tet, tetracycline-inducible EB3-mNeon constructs, and second- generation lentivirus packaging plasmids (pMD2.G and psPAX2) using Lipofectamine
2000 (Life Technologies # 11668019). 293FT media containing virus was harvested and
MDA-231 cells were infected for 24-48 hours in the presence of 10 μg ml−1 (26.7 μM) polybrene. After a 24 hour recovery, transduced cells were selected with puromycin at 2
μg ml−1 (4.24 μM) and were flow sorted to isolate and plate single cells into 96 well plates. Such clones were cultured in 50% filtered conditioned media with 50% fresh media. mCh-CEP135full-tet and GFP-CEP135mini-tet cells were induced with tetracycline
(Invitrogen #550205) at 2.5 μg ml−1 (5.63 μM).
122 Transfections
MCF10A and MDA-231 cells at 50-80% confluence were transfected using
Lipofectamine 2000 (Invitrogen # 11668019). Plasmid DNA and Plus reagent (Invitrogen
# 11514015) were mixed at 1:1 and incubated for 5 minutes. This mixture was then combined with Lipofectamine at a1:3 ratio. Complexes were diluted in Opti-MEM
(Invitrogen 31985062). After a 4-hour incubation, the complexes were removed and the transfected cells were supplied with fresh media.
Immunofluorescence
12 mm diameter coverslips were acid-washed and heated to 50°C in 100mM HCl for 16 hours. This was followed by washes with water, 50%, 70%, and 95% ethanol for
30 minutes each. Coverslips were coated with Type-1 collagen (Sigma # C9791), air- dried for 20 minutes in the laminar hood and exposed to UV light for cross-linking of collagen for 20 minutes. Cells were cultured on collagen-coated coverslips to 55-70% confluence.
For centrosome immunofluorescence, cells were fixed with 100% methanol at
−20°C for 8 minutes. Mitotic cells in Figure 3.6 were fixed with 4% paraformaldehyde
(Electron Microscopy Sciences) for 4 minutes followed by 100% methanol at −20°C for
4 minutes. Mitotic cells in Figure 3.7 and Figure 3.7 were fixed with 100% methanol at
−20°C for 8 minutes to preserve the mCherry-CEP135full and GFP-CEP135mini fluorescence.
Fixed cells were washed with PBS/Mg (1x PBS, 1mM MgCl2), and then blocked with Knudsen Buffer (1x PBS, 0.5%BSA, 0.5% NP-40, 1mM MgCl2, 1mM NaN3) for 1 hour. Cells were incubated overnight with primary antibodies diluted in Knudsen Buffer
123 at 4oC. Coverslips were washed with PBS three times in 5-minute intervals. Secondary antibodies and Hoechst 33258 (10μg ml−1, Sigma #B2261) were diluted in Knudsen buffer and incubated for 1 hour at room temperature. Coverslips were mounted using
Citifluor (Ted Pella) and sealed with clear nail polish.
Antibodies used for immunofluorescence are - centrin (1:2,000; 20H5; Abcam),
--tubulin (1:1000; DQ-19; Sigma), --tubulin (1:500; DM1A; Sigma), -Centromere derived from human calcinosis, Raynaud's phenomenon, Esophageal dysmotility,
Sclerodactyly, and Telangiectasia (CREST) patient serum (1:2,000; generous gift from
Dr. Jennifer DeLuca), -CPAP (1:350, Proteintech CENPJ 11517-1-AP) and -CEP192
(1:2,000 generous gift from Dr. Andrew Holland). Alexa-fluor secondary antibodies were diluted to 1:1,000 for all experiments (Molecular Probes).
Microscopy
The fluorescence imaging utilized for Figures is identical to those described in
Dahl et al, 2015. Briefly, images were acquired using a Nikon TiE (Nikon Instruments,
Inc.) inverted microscope stand equipped with a 100X PlanApo DIC, NA 1.4 objective.
Images were captured using an Andor iXon EMCCD 888E camera or an Andor Xyla 4.2
CMOS camera (Andor Technologies). Images in Figure 3.3 were acquired using a
Swept Field Confocal system (Prairie Technologies / Nikon Instruments) on a Nikon Ti inverted microscope stand equipped with a 100X Plan Apo , NA 1.45 objective. Images were captured with an Andor Clara CCD camera (Andor Technologies).
Nikon NIS Elements imaging software was used for image acquisition. Image acquisition times were constant within a given experiment and ranged from 50 to 400 msec, depending on the experiment. All images were acquired at approximately 25ºC.
124 Images presented in most of the figures are maximum-intensity projections of the complete z-stacks. Exceptions include certain mitotic images that are constructed from selected z-planes to clearly distinguish kinetochores and lagging chromosomes.
Centriole and centrosome number counts
Cells were scored as amplified, non-amplified and underduplicated based on centrin and γ-tubulin staining (Dahl et al., 2015). Cells with greater than two γ-tubulin and four centrin foci were scored as amplified centrosomes. Non-amplified centrosomes have both one or two γ-tubulin and two or four centrin foci. Underduplicated centrosomes contain one centrin focus.
Chromosome Mis-Segregation Counts
Mitotic cells that contain one, two or greater than two poles labeled by γ-tubulin and microtubules (α-tubulin) were considered as mono-, bi- and multi-polar respectively.
Cells were considered to have lagging chromosomes when they had a kinetochore- positive chromosome in between the two completely separated anaphase chromosome masses.
Statistics and Biological Replicates
All center values represent means and error bars represent the standard error of the mean. All the experiments were performed using at least three independent biological replicates.
The number of cells used in each immunofluorescence experiment is as follows:
3.3A:40 cells per condition/80 cells per experiment. 3.3D: 40 cells per condition/80 cells
125 per experiment. 3.3G: 30 cells per condition/ 60 cells experiment. 3.4A: 40 cells per experiment. 3.4D: 40 cells per experiment. 3.4G: 10 cells per experiment.
Fischer’s exact test, Student’s two-tailed t-test, Mann-Whitney U-test, and
Wilcoxon tests were used to assess statistical significance between means. Fischer’s test was utilized to examine the significance of contingency when data were classified into two or more categories. Student’s two-tailed unpaired t-test was used to examine significance between two normal distributions (equal variance assumed).
Normality tests were performed both on the raw data and meta-data extracted from the replicates of raw data. Shapiro-Wilk normality test and D'Agostino-Pearson omnibus normality test was utilized to examine the normality of data. Shapiro-Wilk normality test was used when the number of samples was less than eight. When the number of samples was greater than eight, the D'Agostino-Pearson omnibus normality test was used. Mann-Whitney u-test was utilized to examine the significance of non- normal unpaired distributions. Wilcoxon test was used utilized to examine the significance of non-normal paired distributions.
Results were considered statistically significant with p-values less than 0.05. P- values were denoted on figures according the following values: * p<0.05, ** p<0.01, *** p<0.005 and **** p<0.0005.
126 CHAPTER IV
CONCLUDING REMARKS AND FUTURE PROSPECTS
In this thesis, I have examined the contribution of centriole overduplication to CA and its consequences on microtubule organization and genomic stability in breast cancer cells. I find a few instances of formation of multiple daughter centrioles from a mother centriole. These instances suggest that centriole overduplication can initiate CA in cancer cells. However, in most instances I find that centriole overduplication propagates the existing CA. The underlying mechanism that supports such propagation of centriole overduplication is not known.
CEP135, a centriole assembly protein, is dysregulated in some breast cancers. We previously identified a short isoform of CEP135, CEP135mini that represses centriole duplication. CEP135mini represses centriole duplication by limiting localization of essential proteins such as SAS-6 and CPAP to centrioles. Interestingly, the levels of the two CEP135 isoforms, CEP135full relative to CEP135mini (the CEP135full:mini ratio) is higher in centrosome amplified breast cancer cell lines. Specifically, I demonstrate that inducing expression of CEP135full increases CA. In contrast, elevating CEP135mini reduces centrosome number in breast cancer cells.
I find that the CEP135mini isoform is generated in vivo by alternative polyadenylation. A directed genetic mutation near the CEP135mini alternative polyadenylation signal reduces the CEP135full:mini ratio and decreases CA, accordingly.
Thus, dysregulation of the CEP135 isoforms can promote CA in breast cancer cells
(Dahl et al., 2015; Ganapathi Sankaran et al., 2019).
127 Furthermore, in order to characterize the consequences of CA in microtubule organization and chromosome segregation, we designed a semi-automatic algorithm using machine learning to detect and analyze centrosomes and microtubule organization around centrosomes. Using this algorithm, I find that centrosome amplified breast cancer cells have microtubule organization defects such as increased microtubule density and EB3 density. Altered microtubule organization from CA affects mitotic progression. Dysregulation of CEP135 isoforms that promote CA also increases the frequency of multipolar spindles, anaphase-lagging chromosomes, and micronuclei leading to chromosome segregation defects (Ganapathi Sankaran et al., 2019). I conclude that dysregulation of the CEP135 isoforms can promote centriole overduplication and results in chromosome segregation errors in breast cancer cells.
CEP135 Isoform-Dependent Centriole Assembly and Mechanism Of Centriole
Number Control
Centriole duplication is tightly regulated such that only two new centrioles are formed in total per cell cycle. Several core centriolar proteins such as PLK4, SAS-6, and
STIL are tightly regulated along the cell cycle to facilitate the formation of only two new centrioles. The underlying mechanisms that control these core centriolar proteins along the cell cycle include auto-phosphorylation to proteasomal degradation. My work in this thesis describes a novel and unique regulatory arm of the core centriolar assembly protein, CEP135. We illustrate that the localization and the levels of the two antagonistic
CEP135 isoforms regulate centriole assembly to facilitate centriole assembly only at S phase (Dahl et al., 2015). However, the underlying mechanisms that regulate the
CEP135 isoform-dependent centriole assembly are not clear. Depletion of CEP135full in
128 certain unicellular organisms results in complete centriole loss while its loss in chordates including humans results in phenotypes varying from centriole assembly defects to complete loss of centrioles (Matsuura et al., 2004; Jerka-Dziadosz et al.,
2010; Bayless et al., 2012; Inanc et al., 2013; Lin et al., 2013; Dahl et al., 2015). These results suggest that CEP135full is important in centriole duplication but the exact phenotypes vary depending on the species and the exact mode of action requires further investigation.
Cell Cycle Regulation of CEP135 Isoforms
CEP135full acts as a physical link between SAS-6 and CPAP and promotes the formation of centrioles (Hiraki et al., 2007; Guichard et al., 2013; Lin et al., 2013).
CEP135full localizes to the pinhead hook of the centrioles. Unlike CEP135full, CEP135mini represses centriole duplication. We find that CEP135mini localizes primarily to the centriolar barrel during G1 phase and expands to the PCM during G2-M phase (Dahl et al., 2015). It is unclear how the redistribution of CEP135mini to PCM occurs. Several
PCM proteins exhibit similar expansion during the G2-M transition (Mennella et al.,
2012, 2014). Super-resolution imaging suggests that PCM transitions from an ordered state during interphase to a disordered state during mitosis (Woodruff et al., 2014,
2015). The change to the disordered state can be due to several reasons including forces experienced by centrosomes during mitosis and PLK1 driven centrosome maturation. It is plausible that similar reasons underlie the expansion and re-distribution of CEP135mini. However, whether such expansion is indeed driven by PLK1 or mitotic forces requires further investigation.
129 Investigation of how CEP135mini changes its localization from centrioles to PCM will not only reveal how CEP135mini organization changes but also provide insights into how PCM’s organization changes during mitosis. The changes in localization of
CEP135 isoforms along the cell cycle suggests that the levels of these two isoforms are also regulated through the cell cycle.
The levels of these two CEP135 isoforms are regulated such that centriole duplication is limited to S-phase. We find that CEP135full levels rise during the G1/S phase transition and protein accumulates at the daughter centrioles(Dahl et al., 2015).
How CEP135full rise at G1/S phase transition is not known. One plausible model is that
PLK4 phosphorylation controls this accumulation. PLK4 is a suicide kinase that activates itself and also initiates its own destruction (Rogers et al., 2009; Holland et al.,
2010, 2012a). How PLK4 controls its timely accumulation, phosphorylation of several other core centriolar proteins and its destruction is also not known and will be a target for future investigation.
In contrast to CEP135full accumulation during S phase, CEP135mini levels are at their lowest during the G1/S transition and increase through the rest of the cell cycle until metaphase. During the metaphase to anaphase transition, CEP135mini levels drop
(Dahl et al., 2015). Consistent with the protein levels, transcript levels of CEP135mini also drop during mitosis. It is not known how CEP135mini transcript and protein levels decrease during mitosis. Many proteins get degraded during the metaphase to anaphase transition through the anaphase-promoting complex dependent pathway
(Acquaviva, 2006). It is possible that CEP135mini gets degraded through the APC/C. In corroboration with the idea of proteasome and ubiquitination-dependent regulation,
130 depletion of UCHL5, a deubiquitinating enzyme, alters the accumulation of CEP135full
(Chadchankar et al., 2018). This suggests that the proteasome can be important in the regulation of cell cycle levels of CEP135 isoforms.
CEP135mini in Repression of Centriole Duplication
We find that CEP135mini functions opposite to the CEP135full isoform. Elevated expression of CEP135mini limits localization of SAS-6 and CPAP to centrioles and represses centriole duplication (Dahl et al., 2015). It is not known how CEP135mini limits the localization of these proteins to centrioles. The N-terminus of CEP135full forms a two-stranded coiled-coil structure with chains in parallel that interacts with the negatively charged outer surface of microtubules and the C-terminus of CPAP (Kraatz et al., 2016;
Lin et al., 2013). The C-terminus of CEP135full also has a second coiled-coil region that interacts with SAS-6. CEP135mini has an N-terminus identical to that of CEP135full.
It is known that coiled-coil proteins can readily exchange chains between each other and form heterodimers (Truebestein and Leonard, 2016). One plausible model as to how CEP135mini might limit SAS-6 and CPAP localization to centrioles is that it may form a heterodimer through its interactions with coiled-coil regions both at the N- and C- termini of CEP135full and interfere with its interactions with its binding partners SAS-6 and CPAP (Kraatz et al., 2016). Consistent with this idea, the ectopic expression of
CEP135full and CEP135mini suggests that CEP135full is capable of interaction with
CEP135mini (Dahl et al., 2015). Indeed, whether CEP135mini forms a heterodimer with
CEP135full requires future investigation. Moreover, how the divergent C-terminus of
CEP135mini contributes to this heterodimer formation is also not known and will be a
131 target of the future investigation. These questions can provide interesting insights into how CEP135mini represses duplication and regulates the centriole number control.
CEP135 Isoform Dysregulation in the Cause of Centrosome Amplification In
Cancer
Centrosomal aberrations often occur in several cancers. Centrosomal aberrations can be classified into structural and numerical aberrations in cancer(Godinho and Pellman, 2014). Structural aberrations such as centriole elongation have been characterized in cancer cells (Marteil et al., 2018). I investigated how PCM is structurally aberrated in cancer cells. I find that PCM proteins -tubulin and pericentrin are at a higher density at amplified centrosomes. Furthermore, the distribution of pericentrin also expands at amplified centrosomes. How this expansion occurs and impacts the cellular function is yet to be studied. Understanding the complete spectrum of structural defects of centrioles and centrosomes is limited by the optical resolution of microscopes. The diameter of centrioles is about 200nm which is below the optical resolution of standard fluorescent microscopes. With the technological advancement of newer microscopes, the characterization of centrosomal structural defects in cancer is evolving.
Centriole Overduplication in Centrosome Amplification of Breast Cancers
Numerical aberration of centrosomes is often observed in several types of cancers. The burden of CA is specifically much higher in solid tumors such as breast cancer, colorectal cancers, cervical and pancreatic cancers relative to hematological tumors (Chan, 2011). Several mechanisms can lead to CA in cancers. Some of them include cell cycle aberrations, centriole overduplication, centrosome fragmentation, and
132 de novo centriole assembly. Cell cycle aberrations include events such as cytokinesis failures and mitotic slippage (Fujiwara et al., 2005; Ganem et al., 2007). In the last decade, the underlying causes of CA have studied through mostly through overexpression studies of various core centriole components. The relative contribution of the above mechanisms to CA in breast cancer cells has been very unclear. I asked which amongst the above mechanisms may be candidates in causing CA in breast cancer cells. I investigated whether centriole numbers increase with a corresponding increase in the PCM. Indeed, I find that centriole numbers increase with a corresponding increase in the PCM (Ganapathi Sankaran et al., 2019). This suggests that centriole overduplication and cell cycle aberrations play an important role in CA in cancers. I investigated the contribution of centriole over-duplication to CA in breast cancer cells. I detected instances of multiple daughter centriole formations from a single mother centriole. These instances can represent the initiation of centriole over- duplication in breast cancer cells. Furthermore, the amplified centrioles were capable of assembling new daughter centrioles. In other words, these data suggest that centriole overduplication propagates existing duplication (Ganapathi Sankaran et al., 2019).
However, the instances of initiation of centriole overduplication were much less frequent than events that propagate CA. Almost all new centrioles form in a once-and-only-once event where each existing centriole forms only one new centriole to maintain either normal or excess numbers of centrioles. We, therefore, suggest that most centriole overduplication events simply maintain existing CA in breast cancer cells (Ganapathi
Sankaran et al., 2019). This indicates that breast cancer cells, despite having lost some aspects of centrosome number control, retain regulatory mechanisms that limit centriole
133 duplication to a single daughter centriole for each mother centriole. This suggests that intrinsic mechanisms such as engagement of mother and daughter centriole that limit centriole overduplication in normal cells are still maintained in most instances in cancer cells (Tsou and Stearns, 2006b, 2006a). Furthermore, this is consistent with the idea that I do not observe a dramatic accumulation of CA between multiple passages of breast cancer cells culture.
It is possible that the initiation of CA occurs at a higher frequency through other mechanisms such as cell cycle aberration, but homeostatic mechanisms such as cell death due to genome instability eliminate the CA population. Furthermore, the determination of the first initiating event that causes CA in breast cancers requires a much higher temporal resolution as to when tumorigenesis initiation occurs in breast cancer patients. Ductal carcinoma in situ is a class of early pre-invasive breast tumors.
CA is observed as early as in DCIS patients. This can be an ideal tumor and time to understand the initiation of CA in breast cancers (Kronenwett et al., 2005). Aurora A kinase, the PCM protein is overexpressed in ductal carcinoma in situ patient samples.
Overexpression of Aurora A kinase promotes centrosome number aberration and cell division failures (Li et al., 2004). Long-term live cell imaging of such primary samples can aid in understanding the initiation and homeostasis of CA in breast cancers. Apart from cell division aberrations and centriole overduplication, the contribution of the deuterosome mediated CA or centrosome fragmentation in breast cancers is not well studied. It would be interesting to understand whether these mechanisms also contribute to CA in breast cancers.
The underlying mechanisms that support the initiation and propagation of
134 centriole overduplication in breast cancers are not known. Overexpression of core centriolar proteins promotes CA. PLK4 overexpression promotes the formation of centriole rosettes leading to duplication of multiple new centrioles. Centriole rosette-like structures with multiple daughter centrioles surrounding a single mother were reported in several malignancies, suggesting a high frequency of multiple daughter centriole overduplication from a single mother centriole (Cosenza et al., 2017). In my studies of cultured breast cancer cells, I did not observe rosettes. This suggests that PLK4 may not be continuously overexpressed in breast cancer cells. This is consistent with the idea that PLK4 is a master regulator, whose levels need to be very finely tuned.
Changes in PLK4 levels drive very high levels of CA. High levels of CA inhibit tumor evolution (Kleylein-Sohn et al., 2007; Holland et al., 2012b; Kulukian et al., 2015; Vitre et al., 2015). Hence, PLK4 doesn’t seem to be a promising candidate for the cause of misregulation of centriole number in breast cancer cells. However, this is a target for future investigation. Another candidate is the cartwheel protein CPAP, which is overexpressed in breast cancer cells. Its overexpression promotes centriole amplification through over-elongation of centrioles. Ectopic pro-centriole formation and centriole fragmentation are some models through which CPAP may promote centriole overduplication and CA (Marteil et al., 2018). Hence, CPAP overexpression can contribute to centriole overduplication in breast cancers.
Alternative Isoforms and their Regulation in Centrosome Amplification
The chromosomal locus of CEP135 is amplified in aggressive breast cancers
(Yu et al., 2009; Johansson et al., 2011). CEP135full and CEP135mini, the two isoforms of
CEP135 perform the opposite function in centriole duplication (Dahl et al., 2015). I
135 examined how these two isoforms are regulated in breast cancer cells. My results suggest that the CEP135 isoforms are differentially regulated in breast cancer cells. The
CEP135full is highly upregulated relative to CEP135mini isoform in aggressive breast cancer cells. This elevated CEP135full:mini ratio can promote CA in breast cancers
(Ganapathi Sankaran et al., 2019). Furthermore, CEP135 is a candidate oncogene that is frequently mutated and amplified in colorectal cancers. This suggests that CEP135 isoform dysregulation can be important in CA of other solid tumors (Tuupanen et al.,
2014). The underlying mechanism that leads to the differential expression of the
CEP135 isoforms in breast cancer cells is not known. I investigated whether CEP135mini is an alternatively spliced or alternatively polyadenylated isoform. My data support an alternatively polyadenylated model in which a proximal poly(A) site is utilized for transcription termination of CEP135mini(Ganapathi Sankaran et al., 2019). However, the underlying molecules/elements that control the usage of CEP135’s proximal or distal poly(A) sites are not known.
Alternative polyadenylation/splicing is a widespread means of regulating gene expression that is commonly disrupted in several cancers (David and Manley, 2010).
The 3’ cis-elements with the 3’ processing machinery control the alternative cleavage and polyadenylation(Nunes et al., 2010; Di Giammartino et al., 2011). I find that mutations near CEP135mini’s poly (A) signals in breast cancer cells are sufficient to increase the expression of CEP135mini. These mutations are insertions that contain AU rich elements that may disrupt the 3’ cis-elements for the cleavage and polyadenylation of CEP135mini. mRNA stability is often altered by the presence of AU rich elements (Jing et al., 2005). Hence it is plausible that these mutations increase the CEP135mini mRNA
136 stability. These results require further investigations to understand the mechanism by which the CEP135mini expression is upregulated through these mutations. Furthermore, the underlying 3’ mRNA processing factors that control the usage of CEP135mini’s poly
(A) signal are not known. HuR, an RNA binding protein, binds near CEP135mini’s poly
(A) signal (Mukherjee et al., 2011). Our preliminary data suggest that the loss of HuR reduces the expression of CEP135mini isoform in a cell cycle-dependent manner.
Furthermore, HuR knockdown regulates centrosome number (Filippova et al., 2012).
These results suggest that mRNA processing events can control the centrosome number in breast cancer cells.
Consistent with this idea, a genome-wide siRNA screen identified 38 regulators, which when knock-down caused centriole under-duplication or lower number of centrioles. Strikingly, 14/38 of the implicated genes required for centriole biogenesis play a role in mRNA processing, suggesting an unsuspected link between alternative mRNA processing and centriole biology (Balestra et al., 2013). Amongst these 14 RNA processing factors SNRNP200, SNW1 are a part of core spliceosome machinery, while others such as hnRNPA1, SFRS1, PNN, SF3A3, and SON are highly implicated in alternative processing events (Achsel et al., 1998; Matlin et al., 2005; Sato et al., 2014).
The targets of these mRNA processing factors that affect the centriole number have not been identified. Identification of these alternative isoforms that are controlled by the above mRNA processing factors and whether these events directly regulate centriole number are avenues for future investigation.
137 Consequences of Centrosome Amplification
In the early 1900s, Boveri postulated that CA causes cancer (Boveri, 1888). A century later now, the role of CA in tumorigenesis is still being investigated. CA and its consequences such as aneuploidy were considered bystanders of tumorigenesis, until the last decade where studies show that CA actively promotes tumor initiation and progression. However, the consequences of CA in tumorigenesis have been mainly investigated using PLK4 overexpression systems (Godinho et al., 2014; Vitre et al.,
2015; Levine et al., 2017). While PLK4 is a master regulator of centriole duplication, it is not clear whether PLK4 is dysregulated in cancers. Furthermore, this master regulator also affects numerous other processes other than CA (Sillibourne and Bornens, 2010).
In the future, it will be important to address/recapitulate the consequences of CA populations that better represent CA populations in cancers.
Chromosome Mis-Segregation
The long-standing relation between CA and the formation of multipolar spindles during mitosis is known (Nigg, 2006; Godinho et al., 2009). Centrosomes in multipolar spindles cluster and undergo clustered bipolar mitosis. However, such clustered bipolar divisions promote merotelic kinetochore attachments during anaphase leading to the formation of lagging chromosomes. How merotely changes the shape and structure of kinetochores is not well studied. Studies suggest that merotelic attachments stretch the kinetochores. Moreover, whether such expansions promote additional erroneous attachments is an unanswered question (Cimini et al., 2001; Ganem et al., 2009;
Khodjakov and Rieder, 2009). Furthermore, merotelic attachments promote chromosome breakage through increased tension at kinetochores (Guerrero et al.,
138 2010). However, the anaphase lagging chromosomes resulting from merotelic attachments do not exhibit increased DNA damage response. Hence, the effect of merotely on chromosome breakage is unclear. These anaphase lagging chromosomes manifest as micronuclei during interphase. Micronuclei are sites of extensive genome rearrangements and promote chromothripsis in cancer cells (Crasta et al., 2012; Zhang et al., 2015). Chromosomes are highly unstable in micronuclei and how such instability occurs in micronuclei are questions under active and future investigation. I investigated the effect of CEP135 isoform dysregulation on chromosome segregation. I find that dysregulation of CEP135 isoforms is sufficient to promote the formation of multipolar spindles, anaphase lagging chromosomes, and micronuclei. This suggests that dysregulation of centriole assembly factors have far-reaching consequences in promoting genome instability, a hallmark of cancer. However, whether CEP135 isoform dysregulation occurs in very early pre-invasive tumors to drive several genetic assaults of breast cancer will require further investigation.
Microtubule Assembly and Organization
Alterations to microtubule organization and dynamics from amplified centrosomes will explain the merotelic kinetochore attachments during mitosis (Ertych et al., 2014; Cosenza et al., 2017). Amplified centrosomes in the mitotic spindle have higher microtubule density around them. However, whether and how the interphase microtubule network around the amplified centrosomes is affected is not known. We designed a semi-automatic machine learning algorithm to characterize microtubule organization defects. This algorithm can significantly speed up analysis and aid in removing bias during analysis since it removes the need for the user to manually
139 identify every centrosome. An interesting aspect of these corrections which we did not explore in current work is the ability to use these corrections to further train the underlying machine learning model. This is especially beneficial when collecting enough annotations for training the model is expensive and time-consuming.
I investigated the interphase microtubule organization in centrosome amplified breast cancer cells using the above algorithm. I find that centrosome amplified breast cancer cells have higher microtubule density around them. This suggests an increased nucleation and branching of the microtubule network from the interphase amplified centrosomes (Petry et al., 2013; Ishihara et al., 2014). Consistent with this idea, I find that amplified centrosomes have a higher density of microtubule growing ends.
Surprisingly, the length of microtubule growing ends is also higher in cells with amplified centrosomes. This suggests that microtubules from amplified centrosomes may have higher microtubule assembly rates (Maurer et al., 2012). The underlying mechanism through which microtubule assembly rates are increased in cancer cells are avenues for future investigation.
In summary, centrosome number homeostasis has important implications in controlling cellular functions. Alternative mRNA processing in this context of control of the centrosome number holds promise as an emerging paradigm. Alternative polyadenylation of CEP135 isoforms is one such example of this paradigm that controls centrosome number homeostasis in breast cancer cells.
140 REFERENCES
Achsel, T, Ahrens, K, Brahms, H, Teigelkamp, S, and Lührmann, R (1998). The human U5-220kD protein (hPrp8) forms a stable RNA-free complex with several U5- specific proteins, including an RNA unwindase, a homologue of ribosomal elongation factor EF-2, and a novel WD-40 protein. Mol Cell Biol 18, 6756–6766.
Agircan, FG, Schiebel, E, and Mardin, BR (2014). Separate to operate: control of centrosome positioning and separation. Philos Trans R Soc Lond B Biol Sci 369, 20130461-.
Allen, C, and Borisy, GG (1974). Structural polarity and directional growth of microtubules of Chlamydomonas flagella. J Mol Biol 90, 381–402.
Anderson, R, and Brenner, M (1971). The formation of basal bodies (centrioles) in the rhesus monkey oviduct. J Cell Biol 50, 10–34.
Applegate, K, Besson, S, Matov, A, Bagonis, M, Jaqaman, K, and Danuser, G (2011). plusTipTracker: quantitative image analysis software for the measurement of microtubule dynamics. J Struct Biol 176, 168–184.
Arquint, C, Gabryjonczyk, A-M, and Nigg, EA (2014). Centrosomes as signalling centres. Philos Trans R Soc Lond B Biol Sci 369, 20130464-.
Arquint, C, and Nigg, EA (2014). STIL microcephaly mutations interfere with APC/C- mediated degradation and cause centriole amplification. Curr Biol 24, 351–360.
Arquint, C, Sonnen, KF, Stierhof, Y-D, and Nigg, E a. (2012). Cell-cycle-regulated expression of STIL controls centriole number in human cells. J Cell Sci 125, 1342–1352.
Bahe, S, Stierhof, YD, Wilkinson, CJ, Leiss, F, and Nigg, EA (2005). Rootletin forms centriole-associated filaments and functions in centrosome cohesion. J Cell Biol 171, 27–33.
Balestra, F, Strnad, P, Flückiger, I, and Gönczy, P (2013). Discovering regulators of centriole biogenesis through siRNA-based functional genomics in human cells. Dev Cell 25, 555–571.
141 Barretina, J et al. (2012). The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603–607.
Bayless, BA, Giddings, TH, Winey, M, and Pearson, CG (2012). Bld10/Cep135 stabilizes basal bodies to resist cilia-generated forces. Mol Biol Cell 23, 4820– 4832.
Beaudoing, E (2000). Patterns of Variant Polyadenylation Signal Usage in Human Genes. Genome Res 10, 1001–1010.
Bergen, LG, and Borisy, GG (1980). Head-to-tail polymerization of microtubules in vitro. J Cell Biol 84, 141–150.
Bornens, M, and Gönczy, P (2014). Centrosomes back in the limelight. Philos Trans R Soc B Biol Sci 369.
Bornens, M, Gönczy, P, and B, PTRS (2014). Centrosomes back in the limelight. Philos Trans R Soc B Biol Sci 369.
Bouchet, BP, and Akhmanova, A (2017). Microtubules in 3D cell motility. J Cell Sci 130, 39–50.
Boveri, T (1888). Zellen-Studien 2: Die Befruchtung und Teilung des Eies von Ascaris megalocephala. Jenaische Zeitschr Med Naturw 22, 685– 882 22, 685– 882.
Bowne-Anderson, H, Zanic, M, Kauer, M, and Howard, J (2013). Microtubule dynamic instability: A new model with coupled GTP hydrolysis and multistep catastrophe. BioEssays 35, 452–461. van Breugel, M et al. (2011). Structures of SAS-6 suggest its organization in centrioles. Science 331, 1196–1199.
Brinkley, BR (1985). Microtubule Organizing Centers. Annu Rev Cell Biol 1, 145–172.
Carvalho-Santos, Z et al. (2012). BLD10/CEP135 Is a Microtubule-Associated Protein that Controls the Formation of the Flagellum Central Microtubule Pair. Dev Cell 23, 412–424.
142 Carvalho-Santos, Z, Azimzadeh, J, Pereira-Leal, JB, and Bettencourt-Dias, M (2011). Tracing the origins of centrioles, cilia, and flagella. J Cell Biol 194, 165–175.
Carvalho-Santos, Z, Machado, P, Branco, P, Tavares-Cadete, F, Rodrigues-Martins, A, Pereira-Leal, JB, and Bettencourt-Dias, M (2010). Stepwise evolution of the centriole-assembly pathway. J Cell Sci 123, 1414–1426.
Caviston, JP, and Holzbaur, ELF (2006). Microtubule motors at the intersection of trafficking and transport. Trends Cell Biol 16, 530–537.
Chan, JY (2011). A clinical overview of centrosome amplification in human cancers. Int J Biol Sci 7, 1122–1144.
Chang, Y-C, Nalbant, P, Birkenfeld, J, Chang, Z-F, and Bokoch, GM (2008). GEF-H1 Couples Nocodazole-induced Microtubule Disassembly to Cell Contractility via RhoA. Mol Biol Cell 19, 2147–2153.
Chavali, PL, Pütz, M, and Gergely, F (2014). Small organelle, big responsibility: the role of centrosomes in development and disease. Philos Trans R Soc Lond B Biol Sci 369, 20130468-.
Cimini, D (2008). Merotelic kinetochore orientation, aneuploidy, and cancer. Biochim Biophys Acta - Rev Cancer 1786, 32–40.
Cimini, D, Howell, B, Maddox, P, Khodjakov, A, Degrassi, F, and Salmon, ED (2001). Merotelic kinetochore orientation is a major mechanism of aneuploidy in mitotic mammalian tissue cells. J Cell Biol 152, 517–527.
Coelho, PA et al. (2015). Over-expression of Plk4 induces centrosome amplification, loss of primary cilia and associated tissue hyperplasia in the mouse. Open Biol 5.
Conduit, PT, Brunk, K, Dobbelaere, J, Dix, CI, Lucas, EP, and Raff, JW (2010). Centrioles regulate centrosome size by controlling the rate of Cnn incorporation into the PCM. Curr Biol 20, 2178–2186.
Cosenza, MR et al. (2017). Asymmetric Centriole Numbers at Spindle Poles Cause Chromosome Missegregation in Cancer. Cell Rep 20, 1906–1920.
143 Cottee, MA, Muschalik, N, Wong, YL, Johnson, CM, Johnson, S, Andreeva, A, Oegema, K, Lea, SM, Raff, JW, and Breugel, M Van (2013). Crystal structures of the CPAP / STIL complex reveal its role in centriole assembly and human microcephaly. 1–23.
Crasta, K, Ganem, NJ, Dagher, R, Lantermann, AB, Ivanova, E V., Pan, Y, Nezi, L, Protopopov, A, Chowdhury, D, and Pellman, D (2012). DNA breaks and chromosome pulverization from errors in mitosis. Nature 482, 53–58.
Cunha-Ferreira, I, Rodrigues-Martins, A, Bento, I, Riparbelli, M, Zhang, W, Laue, E, Callaini, G, Glover, DM, and Bettencourt-Dias, M (2009). The SCF/Slimb Ubiquitin Ligase Limits Centrosome Amplification through Degradation of SAK/PLK4. Curr Biol 19, 43–49.
Cuschieri, A, and Bannister, LH (1975). The development of the olfactory mucosa in the mouse : electron microscopy. J Anat, 471–498.
D’Assoro, AB, Barrett, SL, Folk, C, Negron, VC, Boeneman, K, Busby, R, Whitehead, C, Stivala, F, Lingle, WL, and Salisbury, JL (2002). Amplified centrosomes in breast cancer: A potential indicator of tumor aggressiveness. Breast Cancer Res Treat 75, 25–34.
Dahl, KD, Sankaran, DG, Bayless, BA, Pinter, ME, Galati, DF, Heasley, LR, Giddings, TH, and Pearson, CG (2015). A Short CEP135 Splice Isoform Controls Centriole Duplication. Curr Biol 25, 2591–2596.
David, CJ, and Manley, JL (2010). Alternative pre-mRNA splicing regulation in cancer: Pathways and programs unhinged. Genes Dev 24, 2343–2364.
Decker, M, Jaensch, S, Pozniakovsky, A, Zinke, A, O’Connell, KF, Zachariae, W, Myers, E, and Hyman, AA (2011). Limiting amounts of centrosome material set centrosome size in C. elegans embryos. Curr Biol 21, 1259–1267.
Dehring, DAK, Vladar, EK, Werner, ME, Mitchell, JW, Hwang, P, and Mitchell, BJ (2013). Deuterosome-Mediated Centriole Biogenesis. Dev Cell 27, 103–112.
Denu, RA, Zasadil, LM, Kanugh, C, Laffin, J, Weaver, BA, and Burkard, ME (2016). Centrosome amplification induces high grade features and is prognostic of worse outcomes in breast cancer. BMC Cancer 16, 1–13.
144 DeRisi, J et al., Penland, L, Brown, PO, Bittner, ML, Meltzer, PS, Ray, M, Chen, Y, Su, YA, and Trent, JM (1996). Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nat Genet 14, 457–460.
Desai, A, and Mitchison, TJ (1997). Microtubule Polymerization Dynamics. 83–117.
Dictenberg, JB, Zimmerman, W, Sparks, CA, Young, A, Carrington, W, Zheng, Y, Vidair, C, Fay, FS, and Doxsey, SJ (2002). Pericentrin and γ-Tubulin Form a Protein Complex and Are Organized into a Novel Lattice at the Centrosome. J Cell Biol 141, 163–174.
Duelli, DM, Hearn, S, Myers, MP, and Lazebnik, Y (2005). A primate virus generates transformed human cells by fusion. J Cell Biol 171, 493–503.
Duelli, DM, Padilla-Nash, HM, Berman, D, Murphy, KM, Ried, T, and Lazebnik, Y (2007). A Virus Causes Cancer by Inducing Massive Chromosomal Instability through Cell Fusion. Curr Biol 17, 431–437.
Duensing, A, Liu, Y, Perdreau, S a, Kleylein-Sohn, J, Nigg, E a, and Duensing, S (2007). Centriole overduplication through the concurrent formation of multiple daughter centrioles at single maternal templates. Oncogene 26, 6280–6288.
Duensing, S (2005). A tentative classification of centrosome abnormalities in cancer. Cell Biol Int 29, 352–359.
Duensing, S, and Mu, K (2002). The human papillomavirus type 16 E6 and E7 oncoproteins independently induce numerical and structural chromosome instability. Cancer Res 62, 7075–7082.
Elkon, R, Ugalde, AP, and Agami, R (2013). Alternative cleavage and polyadenylation: extent, regulation and function. Nat Rev Genet 14, 496–506.
Ertych, N, Stolz, A, Stenzinger, A, Weichert, W, Burfeind, P, Kaulfuß, S, Bastians, H, Aigner, A, Wordeman, L, and Bastians, H (2014). Increased microtubule assembly rates influence chromosomal instability in colorectal cancer cells. Nat Cell Biol 16, 779–791.
Farache, D, Emorine, L, Haren, L, and Merdes, A (2018). Assembly and regulation of γ- tubulin complexes. Open Biol 8.
145 Faragher, AJ, and Fry, AM (2003). Nek2A kinase stimulates centrosome disjunction and is required for formation of bipolar mitotic spindles. Mol Biol Cell 14, 2876–2889.
Filippova, N, Yang, X, King, P, and Nabors, LB (2012). Phosphoregulation of the RNA- binding protein Hu antigen R (HuR) by Cdk5 affects centrosome function. J Biol Chem 287, 32277–32287.
Finak, G et al. (2008). Stromal gene expression predicts clinical outcome in breast cancer. Nat Med 14, 518–527.
Fırat-karalar, EN, and Stearns, T (2014). The centriole duplication cycle. Philos Trans R Soc Lond B Biol Sci 369, 1–10.
Fong, K-W, Choi, Y-K, Rattner, JB, and Qi, RZ (2007). CDK5RAP2 Is a Pericentriolar Protein That Functions in Centrosomal Attachment of the Gamma-Tubulin Ring Complex. Mol Biol Cell 19, 308–317.
Fonte, VG, Searls, RL, and Robert Hilfer, S (1971). The relationship of cilia with cell division and differentiation. J Cell Biol 49, 226–229.
Friedl, P, and Wolf, K (2003). Tumour-cell invasion and migration: diversity and escape mechanisms. Nat Rev Cancer 3, 362–374.
Fry, AM, Sampson, J, Shak, C, and Shackleton, S (2017). Recent advances in pericentriolar material organization: ordered layers and scaffolding gels. F1000Research 6, 1622.
Fu, J, and Glover, DM (2012). Structured illumination of the interface between centriole and peri-centriolar material. Open Biol 2, 120104.
Fuentealba, LC, Eivers, E, Ikeda, A, Hurtado, C, Kuroda, H, Pera, EM, and De Robertis, EM (2007). Integrating Patterning Signals: Wnt/GSK3 Regulates the Duration of the BMP/Smad1 Signal. Cell 131, 980–993.
Fujiwara, T, Bandi, M, Nitta, M, Ivanova, E V., Bronson, RT, and Pellman, D (2005). Cytokinesis failure generating tetraploids promotes tumorigenesis in p53-null cells. Nature 437, 1043–1047.
146 Galati, DF, Sullivan, KD, Pham, AT, Espinosa, JM, and Pearson, CG (2018). Trisomy 21 Represses Cilia Formation and Function. Dev Cell 46, 641–650.e6.
Galeotti, G (1893). Beitrag zum Studium Carcinome., Chromatins in den Epithelzellen der. Beitr Pathol Anat Allg Pathol 14, 249 – 271.
Gall, J (2004). Early studies on centrioles and centrosomes. In Centrosomes in development and disease., Weinheim, Germany, Germany: Wiley Blackwell.
Ganapathi Sankaran, D, Stemm-Wolf, AJ, and Pearson, CG (2019). CEP135 isoform dysregulation promotes centrosome amplification in breast cancer cells. Mol Biol Cell.
Ganem, NJ, Godinho, SA, and Pellman, D (2009). A mechanism linking extra centrosomes to chromosomal instability. Nature 460, 278–282.
Ganem, NJ, Storchova, Z, and Pellman, D (2007). Tetraploidy, aneuploidy and cancer. Curr Opin Genet Dev 17, 157–162.
Ghadimi, BM, Sackett, DL, Difilippantonio, MJ, Schröck, E, Neumann, T, Jauho, A, Auer, G, and Ried, T (2000). Centrosome amplification and instability occurs exclusively in aneuploid, but not in diploid colorectal cancer cell lines, and correlates with numerical chromosomal aberrations. Genes Chromosom Cancer 27, 183–190.
Di Giammartino, DC, Nishida, K, and Manley, JL (2011). Mechanisms and Consequences of Alternative Polyadenylation. Mol Cell 43, 853–866.
Giehl, M, Fabarius, A, Frank, O, Erben, P, Zheng, C, Hafner, M, Hochhaus, A, Hehlmann, R, and Seifarth, W (2007). Expression of the p210BCR-ABL oncoprotein drives centrosomal hypertrophy and clonal evolution in human U937 cells. Leukemia 21, 1971–1976.
Giehl, M, Fabarius, A, Frank, O, Hochhaus, A, Hafner, M, Hehlmann, R, and Seifarth, W (2005). Centrosome aberrations in chronic myeloid leukemia correlate with stage of disease and chromosomal instability. Leukemia 19, 1192–1197.
Gilmore, EC, and Walsh, CA (2013). Genetic causes of microcephaly and lessons for neuronal development. 2, 461–478.
147 Glover, DM, Leibowitz, MH, Mclean, DA, and Parry, H (1995). Mutations in Aurora Prevent Centrosome Separation Leading to the Formation of Monopolar Spindles. Cell 81, 95–105.
Glück, S, Ross, JS, Royce, M, McKenna, EF, Perou, CM, Avisar, E, and Wu, L (2012). TP53 genomics predict higher clinical and pathologic tumor response in operable early-stage breast cancer treated with docetaxel-capecitabine ± trastuzumab. Breast Cancer Res Treat 132, 781–791.
Godinho, S, Kwon, M, and Pellman, D (2009). Centrosomes and cancer: How cancer cells divide with too many centrosomes. Cancer Metastasis Rev 28, 85–98.
Godinho, SA, and Pellman, D (2014). Causes and consequences of centrosome abnormalities in cancer. Phil Trans R Soc B 369.
Godinho, SA, Picone, R, Burute, M, Dagher, R, Su, Y, Leung, C, Polyak, K, Brugge, J, Théry, M, and Pellman, D (2014). Oncogene-like induction of cellular invasion from centrosome amplification. Nature 510, 167–171.
Gonczy, P (2015). Centrosomes and cancer: revisiting a long-standing relationship. Nat Rev Cancer 15, 639–652.
Gönczy, P (2012). Towards a molecular architecture of centriole assembly. Nat Rev Mol Cell Biol 13, 425–435.
Gopalakrishnan, J, Guichard, P, Smith, AH, Schwarz, H, Marco, S, Avidor-Reiss, T, and Agard, DA (2010). Self-assembling SAS-6 Multimer Is a Core Centriole Building Block. J Biol Chem 285, 8759–8770.
Gosti-testul, F, Marty, M, Berges, J, Maunoury, R, and Bornens, M (1986). Identification of centrosomal proteins in a human lymphoblastic cell line. EMBO J 5, 2545– 2550.
Grady, L (2006). Random walks for image segmentation. IEEE Trans Pattern Anal Mach Intell 28, 1768–1783.
148 Guerrero, AA, Gamero, MC, Trachana, V, Fütterer, A, Pacios-Bras, C, Nuria Panadero Díaz-Concha, JCC, Martínez-A., C, Wely, KHM van, and Garcia-Bellido, A (2010). Centromere-Localized Breaks Indicate the Generation of DNA Damage by the Mitotic Spindle. Proc Natl Acad Sci U S A 107, 4159–4164.
Guichard, P et al. (2013). Native architecture of the centriole proximal region reveals features underlying its 9-fold radial symmetry. Curr Biol 23, 1620–1628.
Guo, H, Gao, M, Ma, J, Xiao, T, Zhao, L, Gao, Y, and Pan, Q (2007). Analysis of the cellular centrosome in fine-needle aspirations of the breast. Breast Cancer Res 9, R48.
Habedanck, R, Stierhof, Y-D, Wilkinson, CJ, and Nigg, E a (2005). The Polo kinase Plk4 functions in centriole duplication. Nat Cell Biol 7, 1140–1146.
Hanahan, D, and Weinberg, R (2010). The Hallmarks of Cancer. Cell 18, 399.
Hansemann, D (1890). Ueber asymmetrische Zelltheilung in Epithelkrebsen und deren biologische Bedeutung. Arch Fu¨r Pathol Anat Und Physiol Und Fu¨r Klin Med 119, 299–326.
Haren, L, Stearns, T, and Lüders, J (2009). Plk1-dependent recruitment of γ-tubulin complexes to mitotic centrosomes involves multiple PCM components. PLoS One 4.
He, K, and Sun, J (2015). Convolutional neural networks at constrained time cost. Proc IEEE Comput Soc Conf Comput Vis Pattern Recognit 07–12–June, 5353–5360.
Heck, JN, Garcia-Mendoza, MG, Pehlke, CA, Ponik, SM, Inman, DR, Eliceiri, KW, and Keely, PJ (2012). Microtubules regulate GEF-H1 in response to extracellular matrix stiffness. Mol Biol Cell 23, 2583–2592.
Hinchcliffe, EH, Li, C, Thompson, EA, Maller, JL, and Sluder, G (1999). Requirement of Cdk2 – Cyclin E Activity for Repeated Centrosome Reproduction in Xenopus Egg Extracts. Science (80- ) 283, 851–855.
Hiraki, M, Nakazawa, Y, Kamiya, R, and Hirono, M (2007). Bld10p Constitutes the Cartwheel-Spoke Tip and Stabilizes the 9-Fold Symmetry of the Centriole. Curr Biol 17, 1778–1783.
149 Hirono, M (2014). Cartwheel assembly. Philos Trans R Soc Lond B Biol Sci 369.
Hoh, RA, Stowe, TR, Turk, E, and Stearns, T (2012). Transcriptional Program of Ciliated Epithelial Cells Reveals New Cilium and Centrosome Components and Links to Human Disease. 7, 1–13.
Holland, a. J, Fachinetti, D, Han, JS, and Cleveland, DW (2012a). Inducible, reversible system for the rapid and complete degradation of proteins in mammalian cells. Proc Natl Acad Sci 109, E3350–E3357.
Holland, AJ et al. (2012b). Polo-like kinase 4 controls centriole duplication but does not directly regulate cytokinesis. Mol Biol Cell 23, 1838–1845.
Holland, AJ, Lan, W, Niessen, S, Hoover, H, and Cleveland, DW (2010). Polo-like kinase 4 kinase activity limits centrosome overduplication by autoregulating its own stability. J Cell Biol 188, 191–198.
Hoque, M, Ji, Z, Zheng, D, Luo, W, Li, W, You, B, Park, JY, Yehia, G, and Tian, B (2013). Analysis of alternative cleavage and polyadenylation by 3′ region extraction and deep sequencing. Nat Methods 10, 133–139.
Van Horck, FPG, Ahmadian, MR, Haeusler, LC, Moolenaar, WH, and Kranenburg, O (2001). Characterization of p190RhoGEF, A RhoA-specific Guanine Nucleotide Exchange Factor That Interacts with Microtubules. J Biol Chem 276, 4948–4956.
Hurtado, L, Caballero, C, Gavilan, MP, Cardenas, J, Bornens, M, and Rios, RM (2011). Disconnecting the Golgi ribbon from the centrosome prevents directional cell migration and ciliogenesis. J Cell Biol 193, 917–933.
Hussain, MS et al. (2012). A Truncating Mutation of CEP135 Causes Primary Microcephaly and Disturbed Centrosomal Function. Am J Hum Genet 90, 871– 878.
Hyman, AA, and Karsenti, E (1996). Morphogenetic Properties of Microtubules and Mitotic Spindle Assembly. Cell 84, 401–410.
Inanc, B, Putz, M, Lalor, P, Dockery, P, Kuriyama, R, Gergely, F, and Morrison, CG (2013). Abnormal centrosomal structure and duplication in Cep135-deficient vertebrate cells. Mol Biol Cell 24, 2645–2654.
150 Inoue’, S, and Salmon, ED (1995). Force Generation by Microtubule Assembly/Disassembly in Mitosis and Related Movements Shinya. Mol Biol Cell 6, 1619–1640.
Ishihara, K, Nguyen, P a., Groen, AC, Field, CM, and Mitchison, TJ (2014). Microtubule nucleation remote from centrosomes may explain how asters span large cells. Proc Natl Acad Sci 111, 17715–17722.
Itoh, K, Jenny, A, Mlodzik, M, and Sokol, SY (2009). Centrosomal localization of Diversin and its relevance to Wnt signaling. J Cell Sci 122, 3791–3798.
Jana, SC, Marteil, G, and Bettencourt-Dias, M (2014). Mapping molecules to structure: Unveiling secrets of centriole and cilia assembly with near-atomic resolution. Curr Opin Cell Biol 26, 96–106.
Jerka-Dziadosz, M, Gogendeau, D, Klotz, C, Cohen, J, Beisson, J, and Koll, F (2010). Basal body duplication in paramecium: The key role of Bld10 in assembly and stability of the Cartwheel. Cytoskeleton 67, 161–171.
Jing, Q, Huang, S, Guth, S, Zarubin, T, Motoyama, A, Chen, J, Di Padova, F, Lin, SC, Gram, H, and Han, J (2005). Involvement of MicroRNA in AU-Rich Element- Mediated mRNA Instability. Cell 120, 623–634.
Johansson, I, Aaltonen, KE, Ebbesson, A, Grabau, D, Wigerup, C, Hedenfalk, I, and Rydén, L (2011). Increased gene copy number of KIT and VEGFR2 at 4q12 in primary breast cancer is related to an aggressive phenotype and impaired prognosis. Genes, Chromosom Cancer 51, 375–383.
Kalay, E et al. (2010). CEP152 is a genome maintenance protein disrupted in Seckel syndrome. Nat Genet 43, 23–26.
Kearns, WG, Yamaguchi, H, Young, NS, and Liu, JM (2004). Centrosome amplification and aneuploidy in bone marrow failure patients. Genes Chromosom Cancer 40, 329–333.
Khodjakov, A, and Rieder, CL (2009). The nature of cell-cycle checkpoints: Facts and fallacies. J Biol 8.
151 Kilburn, CL, Pearson, CG, Romijn, EP, Meehl, JB, Culver, BP, and Yates III, JR (2007). New Tetrahymena basal body protein components identify basal body domain structure. J Cell Biol 178, 905.
Kim, T-S et al. (2013). Hierarchical recruitment of Plk4 and regulation of centriole biogenesis by two centrosomal scaffolds, Cep192 and Cep152. Proc Natl Acad Sci 110, E4849–E4857.
Kitagawa, D et al. (2011). Structural basis of the 9-fold symmetry of centrioles. Cell 144, 364–375.
Kleylein-Sohn, J, Westendorf, J, Le Clech, M, Habedanck, R, Stierhof, YD, and Nigg, E a. (2007). Plk4-Induced Centriole Biogenesis in Human Cells. Dev Cell 13, 190– 202.
Klingseisen, A, and Jackson, AP (2011). Mechanisms and pathways of growth failure in primordial dwarfism. 2011–2024.
Kollman, JM, Merdes, A, Mourey, L, and Agard, D a. (2011). Microtubule nucleation by γ-tubulin complexes. Nat Rev Mol Cell Biol 12, 709–721.
Kraatz, S, Guichard, P, Obbineni, JM, Olieric, N, Hatzopoulos, GN, Hilbert, M, Sen, I, Missimer, J, Gönczy, P, and Steinmetz, MO (2016). The Human Centriolar Protein CEP135 Contains a Two-Stranded Coiled-Coil Domain Critical for Microtubule Binding. Structure 24, 1358–1371.
Krämer, A, Mailand, N, Lukas, C, Syljuåsen, RG, Wilkinson, CJ, Nigg, EA, Bartek, J, and Lukas, J (2004). Centrosome-associated Chk1 prevents premature activation of cyclin-B-Cdk1 kinase. Nat Cell Biol 6, 884–891.
Krämer, A, Neben, K, and Ho, AD (2005). Centrosome aberrations in hematological malignancies. Cell Biol Int 29, 375–383.
Kronenwett, U, Huwendiek, S, Castro, J, Ried, T, and Auer, G (2005). Characterisation of breast fine-needle aspiration biopsies by centrosome aberrations and genomic instability. Br J Cancer 92, 389–395.
152 Kulukian, A, Holland, AJ, Vitre, B, Naik, S, Cleveland, DW, and Fuchs, E (2015). Epidermal development, growth control, and homeostasis in the face of centrosome amplification. Proc Natl Acad Sci 112, E6311–E6320.
Kwon, M, Godinho, SA, Chandhok, NS, Ganem, NJ, Azioune, A, Thery, M, and Pellman, D (2008). Mechanisms to suppress multipolar divisions in cancer cells with extra centrosomes. Genes Dev 22, 2189–2203.
Lambrus, BG, Uetake, Y, Clutario, KM, Daggubati, V, Snyder, M, Sluder, G, and Holland, a. J (2015). P53 Protects Against Genome Instability Following Centriole Duplication Failure. J Cell Biol 210, 63–77.
Leber, B (2010). Proteins required for centrosome clustering in cancer cells. Sci Transl Med 2, 33ra8.
LeCun, Y, Bottou, B, Bengio, Y, and Haffner, P (1998). Gradient-Based Learning Applied to Document Recognition. Proc IEEE 86, 2278–2324.
Leda, M, Holland, AJ, and Goryachev, AB (2018). Autoamplification and Competition Drive Symmetry Breaking: Initiation of Centriole Duplication by the PLK4-STIL Network. IScience 8, 222–235.
Lee, K, and Rhee, K (2011). PLK1 phosphorylation of pericentrin initiates centrosome maturation at the onset of mitosis. J Cell Biol 195, 1093–1101.
Leidel, S, and Gönczy, P (2003). SAS-4 is essential for centrosome duplication in C. elegans and is recruited to daughter centrioles once per cell cycle. Dev Cell 4, 431–439.
Levine, MS et al. (2017). Centrosome Amplification Is Sufficient to Promote Spontaneous Tumorigenesis in Mammals. Dev Cell 40, 313–322.e5.
Levitt, N, Briggs, D, Gil, A, and Proudfoot, NJ (1989). Definition of an efficient synthetic poly ( A ) site. 1019–1025.
Li, J, D’Angiolella, V, Seeley, ES, Kim, S, Kobayashi, T, Fu, W, Campos, EI, Pagano, M, and Dynlacht, BD (2013). USP33 regulates centrosome biogenesis via deubiquitination of the centriolar protein CP110. Nature 495, 255–259.
153 Li, JJ, Weroha, SJ, Lingle, WL, Papa, D, Salisbury, JL, and Li, SA (2004). Estrogen mediates Aurora-A overexpression, centrosome amplification, chromosomal instability, and breast cancer in female ACI rats. Proc Natl Acad Sci U S A 101, 18123–18128.
Li, S, Fernandez, JJ, Marshall, WF, and Agard, DA (2012). Three-dimensional structure of basal body triplet revealed by electron cryo-tomography. EMBO J 31, 552– 562.
Lin, Y-C, Chang, C-W, Hsu, W-B, Tang, C-JC, Lin, Y-N, Chou, E-J, Wu, C-T, and Tang, TK (2013). Human microcephaly protein CEP135 binds to hSAS-6 and CPAP, and is required for centriole assembly. EMBO J 32, 1141–1154.
Lingle, WL, Barrett, SL, Negron, VC, D’Assoro, AB, Boeneman, K, Liu, W, Whitehead, CM, Reynolds, C, and Salisbury, JL (2002). Centrosome amplification drives chromosomal instability in breast tumor development. Proc Natl Acad Sci U S A 99, 1978–1983.
Liu, T, Niu, Y, Yu, Y, Liu, Y, and Zhang, F (2009). Increased γ-tubulin expression and P16INK4Apromoter methylation occur together in preinvasive lesions and carcinomas of the breast. Ann Oncol 20, 441–448.
Liu, W, Anguelov, D, Erhan, D, Szegedy, C, Reed, S, Fu, CY, and Berg, AC (2016). SSD: Single shot multibox detector. Lect Notes Comput Sci (Including Subser Lect Notes Artif Intell Lect Notes Bioinformatics) 9905 LNCS, 21–37.
Lopes, CAM et al. (2018). Centrosome amplification arises before neoplasia and increases upon p53 loss in tumorigenesis. J Cell Biol 217, 2353–2363.
Ly, P, and Cleveland, DW (2017). Rebuilding Chromosomes After Catastrophe: Emerging Mechanisms of Chromothripsis. Trends Cell Biol 27, 917–930.
Mahjoub, MR, and Stearns, T (2014). Supernumerary centrosomes nucleate extra cilia and compromise primary cilium signaling. Curr Biol 154, 2262–2265.
Mahmood, S, Ahmad, W, and Hassan, MJ (2011). Autosomal recessive primary microcephaly ( MCPH ): clinical manifestations , genetic heterogeneity and mutation continuum. Orphanet J Rare Dis 2011, 1–15.
154 Marteil, G et al. (2018). Over-elongation of centrioles in cancer promotes centriole amplification and chromosome missegregation. Nat Commun 9.
Martinho, O et al. (2009). Expression, mutation and copy number analysis of platelet- derived growth factor receptor A (PDGFRA) and its ligand PDGFA in gliomas. Br J Cancer 101, 973–982.
Masood, S (2016). Breast cancer subtypes: Morphologic and biologic characterization. Women’s Heal 12, 103–119.
Matera, a. G, and Wang, Z (2014). A day in the life of the spliceosome. Nat Rev Mol Cell Biol 15, 108–121.
Matlin, AJ, Clark, F, and Smith, CWJ (2005). Understanding alternative splicing: towards a cellular code. Nat Rev Mol Cell Biol 6, 386–398.
Matov, A, Applegate, K, Kumar, P, Thoma, C, Krek, W, Danuser, G, and Wittmann, T (2011). Analysis of Microtubule Dynamic Instability Using a Plus End Growth Marker. 7, 761–768.
Matsuura, K, Lefebvre, PA, Kamiya, R, and Hirono, M (2004). Bld10p, a novel protein essential for basal body assembly in Chlamydomonas: Localization to the cartwheel, the first ninefold symmetrical structure appearing during assembly. J Cell Biol 165, 663–671.
Maurer, SP, Cade, NI, Bohner, G, Gustafsson, N, Boutant, E, and Surrey, T (2014). EB1 accelerates two conformational transitions important for microtubule maturation and dynamics. Curr Biol 24, 372–384.
Maurer, SP, Fourniol, FJ, Bohner, G, Moores, CA, and Surrey, T (2012). EBs Recognize a Nucleotide-Dependent Structural Cap at Growing Microtubule Ends. Cell 149, 371–382.
Mayr, C, and Bartel, DP (2009). Widespread Shortening of 3′UTRs by Alternative Cleavage and Polyadenylation Activates Oncogenes in Cancer Cells. Cell 138, 673–684.
155 Mennella, V, Agard, D a., Huang, B, and Pelletier, L (2014). Amorphous no more: Subdiffraction view of the pericentriolar material architecture. Trends Cell Biol 24, 188–197.
Mennella, V, Keszthelyi, B, McDonald, KL, Chhun, B, Kan, F, Rogers, GC, Huang, B, and Agard, D a. (2012). Subdiffraction-resolution fluorescence microscopy reveals a domain of the centrosome critical for pericentriolar material organization. Nat Cell Biol 14, 1159–1168.
Meraldi, P (2016). Centrosomes in spindle organization and chromosome segregation: a mechanistic view. Chromosom Res 24, 19–34.
Meraldi, P, Honda, R, and Nigg, E a. (2002). Aurora-A overexpression reveals tetraploidization as a major route to centrosome amplification in p53-/- cells. EMBO J 21, 483–492.
Mitchison, T, and Kirschner, M (1984). Dynamic instability of microtubule growth. Nature 312, 237–242.
Moreira, a, Wollerton, M, Monks, J, and Proudfoot, NJ (1995). Upstream sequence elements enhance poly(A) site efficiency of the C2 complement gene and are phylogenetically conserved. EMBO J 14, 3809–3819.
Moritz, M, Braunfeld, MB, Sedat, JW, Alberts, B, and Agard, DA (1995). Microtubule nucleation by γ-tubulin-containing rings in the centrosome. Nature 378, 638–640.
Mountain, V, Simerly, C, Howard, L, Ando, A, Schatten, G, and Compton, DA (1999). The Kinesin-related Protein, HSET, Opposes the Activity of Eg5 and Cross-links Microtubules in the Mammalian Mitotic Spindle. J Cell Biol 147, 351–365.
Mukherjee, N, Corcoran, DL, Nusbaum, JD, Reid, DW, Georgiev, S, Hafner, M, Ascano, M, Tuschl, T, Ohler, U, and Keene, JD (2011). Integrative Regulatory Mapping Indicates that the RNA-Binding Protein HuR Couples Pre-mRNA Processing and mRNA Stability. Mol Cell 43, 327–339.
Munnik, SA De, Hoefsloot, EH, Roukema, J, Schoots, J, Knoers, NVAM, Brunner, HG, Jackson, AP, and Bongers, EMHF (2015). Meier-Gorlin syndrome. Orphanet J Rare Dis, 1–7.
156 Nakamura, T, Saito, H, and Takekawa, M (2013). SAPK pathways and p53 cooperatively regulate PLK4 activity and centrosome integrity under stress. Nat Commun 4, 1713–1775.
Nam, DK, Lee, S, Zhou, G, Cao, X, Wang, C, Clark, T, Chen, J, Rowley, JD, and Wang, SM (2002). Oligo(dT) primer generates a high frequency of truncated cDNAs through internal poly(A) priming during reverse transcription. Proc Natl Acad Sci 99, 6152–6156.
Neben, K, Tews, B, Wrobel, G, Hahn, M, Kokocinski, F, Giesecke, C, Krause, U, Ho, AD, Krämer, A, and Lichter, P (2004). Gene expression patterns in acute myeloid leukemia correlate with centrosome aberrations and numerical chromosome changes. Oncogene 23, 2379–2384.
Neve, RM et al. (2006). A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell 10, 515–527.
Nigg, EA (2006). Origins and consequences of centrosome aberrations in human cancers. Int J Cancer 119, 2717–2723.
Nigg, EA, and Raff, JW (2009). Centrioles, Centrosomes, and Cilia in Health and Disease. Cell 139, 663–678.
Nigg, EA, and Stearns, T (2011). The centrosome cycle: Centriole biogenesis, duplication and inherent asymmetries. Nat Cell Biol 13, 1154–1160.
Nunes, NM, Li, W, Tian, B, and Furger, A (2010). A functional human Poly(A) site requires only a potent DSE and an A-rich upstream sequence. EMBO J 29, 1523–1536.
Oegema, K, Wiese, C, Martin, OC, Milligan, RA, Iwamatsu, A, Mitchison, TJ, and Zheng, Y (1999). Characterization of Two Related Drosophila Gamma-tubulin Complexes that Differ in Their Ability to Nucleate Microtubules. 144, 721–733.
Ohta, M, Ashikawa, T, Nozaki, Y, Kozuka-Hata, H, Goto, H, Inagaki, M, Oyama, M, and Kitagawa, D (2014). Direct interaction of Plk4 with STIL ensures formation of a single procentriole per parental centriole. Nat Commun 5, 5267.
157 Ohta, T, Essner, R, Ryu, JH, Palazzo, RE, Uetake, Y, and Kuriyama, R (2002). Characterization of Cep135, a novel coiled-coil centrosomal protein involved in microtubule organization in mammalian cells. J Cell Biol 156, 87–99.
Paintrand, M, Delacroix, H, Bornens, M, and Moudjou, M (1992). Centrosome Organization Their Sensitivity and Centriole Architecture : to Divalent Cations. 128, 107–128.
Pannu, V, Mittal, K, Cantuaria, G, Reid, MD, Li, X, Donthamsetty, S, Mcbride, M, and Klimov, S (2015). Rampant centrosome amplification underlies more aggressive disease course of triple negative breast cancers. Oncotarget 6, 10487–10497.
Parsons, JT, Horwitz, AR, and Schwartz, MA (2010). Cell adhesion: integrating cytoskeletal dynamics and cellular tension. Nat Rev Genet 11, 633–643.
Pazdur, R, Kudelka, AP, Kavanagh, JJ, Cohen, PR, and Raber, MN (1993). The taxoids: paclitaxel (Taxol®) and docetaxel (Taxotere®). Cancer Treat Rev 19, 351–386.
Petry, S, Groen, AC, Ishihara, K, Mitchison, TJ, and Vale, RD (2013). Branching microtubule nucleation in Xenopus egg extracts mediated by augmin and TPX2. 152, 768–777.
Piel, M, Meyer, P, Khodjakov, A, Rieder, CL, and Bornens, M (2000). The respective contributions of the mother and daughter centrioles to centrosome activity and behavior in vertebrate cells. J Cell Biol 149, 317–329.
Pihan, GA, Wallace, J, Zhou, Y, and Doxsey, SJ (2003a). Centrosome Abnormalities and Chromosome Instability Occur Together in Pre-invasice carcinomas. Cancer Res 63, 1398–1404.
Pihan, GA, Wallace, J, Zhou, Y, and Doxsey, SJ (2003b). Centrosome abnormalities and chromosome instability occur together in pre-invasive carcinomas. Cancer Res 63, 1398–1404.
Quintyne, NJ, Reing, JE, Hoffelder, DR, Gollin, SM, Saunders, WS, Quintyne, NJ, Reing, JE, Hoffelder, DR, Gollin, SM, and Saunders, WS (2005). Spindle Multipolarity Is Prevented by Centrosomal Clustering. Science (80- ) 307, 127– 129.
158 Rajagopalan, H, and Lengauer, C (2004). Aneuploidy and cancer. Nature 432, 338– 341.
Rauch, A et al. (2008). Mutations in the Pericentrin (PCNT) Gene Cause Primordial Dwarfism. Science (80- ) 319, 816–820.
Rhys, AD et al. (2018). Loss of E-cadherin provides tolerance to centrosome amplification in epithelial cancer cells. J Cell Biol 217, 195–209.
Rios, RM (2014). The centrosome - Golgi apparatus nexus. Philos Trans R Soc B Biol Sci 369.
Rodrigues-Martins, A, Riparbelli, M, Callaini, G, Glover, DM, and Bettencourt-Dias, M (2007). Revisiting the Role of the Mother Centriole in Centriole Biogenesis. Science (80- ) 98, 102502.
Rogers, GC, Rusan, NM, Roberts, DM, Peifer, M, and Rogers, SL (2009). The SCF Slimb ubiquitin ligase regulates Plk4/Sak levels to block centriole reduplication. J Cell Biol 184, 225–240.
Salaycik, KJ (2005). Quantification of microtubule nucleation, growth and dynamics in wound-edge cells. J Cell Sci 118, 4113–4122.
Salisbury, JL, D’Assoro, AB, and Lingle, WL (2004). Centrosome amplification and the origin of chromosomal instability in breast cancer. J Mammary Gland Biol Neoplasia 9, 275–283.
Sato, N et al. (2014). Inhibition of SNW1 association with spliceosomal proteins promotes apoptosis in breast cancer cells. Cancer Med.
Scheer, U (2014). Historical roots of centrosome research: discovery of Boveri’s microscope slides in Würzburg. Philos Trans R Soc Lond B Biol Sci 369, 20130469-.
Schiff, PB, and Horwitz, SB (2006). Taxol stabilizes microtubules in mouse fibroblast cells. Proc Natl Acad Sci 77, 1561–1565.
159 Schneeweiss, A, Sinn, H-P, Ehemann, V, Khbeis, T, Neben, K, Krause, U, Ho, AD, Bastert, G, and Krämer, A (2003). Centrosomal aberrations in primary invasive breast cancer are associated with nodal status and hormone receptor expression. Int J Cancer 107, 346–352.
Schvartzman, JM, Sotillo, R, and Benezra, R (2010). Mitotic chromosomal instability and cancer: Mouse modelling of the human disease. Nat Rev Cancer 10, 102– 115.
Serçin, Ö, Larsimont, JC, Karambelas, AE, Marthiens, V, Moers, V, Boeckx, B, Le Mercier, M, Lambrechts, D, Basto, R, and Blanpain, C (2016). Transient PLK4 overexpression accelerates tumorigenesis in p53-deficient epidermis. Nat Cell Biol 18, 100–110.
Siegrist, SE, and Doe, CQ (2007). Microtubule-induced cortical cell polarity. Genes Dev 21, 483–496.
Silkworth, WT, Nardi, IK, Scholl, LM, and Cimini, D (2009). Multipolar spindle pole coalescence is a major source of kinetochore mis-attachment and chromosome mis-segregation in cancer cells. PLoS One 4.
Sillibourne, JE, and Bornens, M (2010). Polo-like kinase 4: the odd one out of the family. Cell Div 5, 25.
Skyldberg, B, Fujioka, K, Hellström, A, Sylvén, L, Moberger, B, and Auer, G (2001). Human Papillomavirus Infection, Centrosome Aberration, and Genetic Stability in Cervical Lesions. Mod Pathol 14, 279–284.
Sonnen, KF, Gabryjonczyk, A-M, Anselm, E, Stierhof, Y-D, and Nigg, EA (2013). Human Cep192 and Cep152 cooperate in Plk4 recruitment and centriole duplication. J Cell Sci 126, 3223–3233.
Sorokin, S (1962). Centrioles and the formation of rudimentary cilia by fibroblasts and smooth muscle cells. J Cell Biol 15, 363–377.
Starita, LM, Machida, Y, Sankaran, S, Elias, JE, Griffin, K, Schlegel, BP, Gygi, SP, and Parvin, JD (2004). BRCA1-Dependent Ubiquitination of Gamma-Tubulin Regulates Centrosome Number. Mol Cell Biol 24, 8457–8466.
160 Stearns, T, and Kirschner, M (1994). In vitro reconstitution of centrosome assembly and function: The central role of γ-tubulin. Cell 76, 623–637.
Stehbens, S, and Wittmann, T (2012). Targeting and transport: How microtubules control focal adhesion dynamics. J Cell Biol 198, 481–489.
Strnad, P, Leidel, S, Vinogradova, T, Euteneuer, U, Khodjakov, A, and Gönczy, P (2007). Regulated HsSAS-6 Levels Ensure Formation of a Single Procentriole per Centriole during the Centrosome Duplication Cycle. Dev Cell 13, 203–213.
Tang, C-JC, Lin, S-Y, Hsu, W-B, Lin, Y-N, Wu, C-T, Lin, Y-C, Chang, C-W, Wu, K-S, and Tang, TK (2011). The human microcephaly protein STIL interacts with CPAP and is required for procentriole formation. EMBO J 30, 4790–4804.
Thein, KH, Kleylein-Sohn, J, Nigg, EA, and Gruneberg, U (2007). Astrin is required for the maintenance of sister chromatid cohesion and centrosome integrity. J Cell Biol 178, 345–354.
Thompson, SL, Bakhoum, SF, and Compton, D a. (2010). Mechanisms of Chromosomal Instability. Curr Biol 20, R285–R295.
Thompson, SL, and Compton, DA (2011). Chromosome missegregation in human cells arises through specific types of kinetochore-microtubule attachment errors. Proc Natl Acad Sci 108, 17974–17978.
Tian, B, Pan, Z, and Ju, YL (2007). Widespread mRNA polyadenylation events in introns indicate dynamic interplay between polyadenylation and splicing. Genome Res 17, 156–165.
Tong, T, Zhong, Y, Kong, J, Dong, L, Song, Y, Fu, M, Liu, Z, Wang, M, Guo, L, and Lu, S (2004). Overexpression of {Aurora-A} contributes to malignant development of human esophageal squamous cell carcinoma. Clin Cancer Res 10, 7304–7310.
Tran, PT, Walker, RA, and Salmon, ED (1997). A metastable intermediate state of microtubule dynamic instability that differs significantly between plus and minus ends. J Cell Biol 138, 105–117.
Truebestein, L, and Leonard, TA (2016). Coiled-coils: The long and short of it. BioEssays 38, 903–916.
161 Tsang, WY, and Dynlacht, BD (2013). CP110 and its network of partners coordinately regulate cilia assembly. Cilia 2, 1.
Tsou, MFB, and Stearns, T (2006a). Controlling centrosome number: Licenses and blocks. Curr Opin Cell Biol 18, 74–78.
Tsou, MFB, and Stearns, T (2006b). Mechanism limiting centrosome duplication to once per cell cycle. Nature 442, 947–951.
Tsuda, Y, Iimori, M, Nakashima, Y, Nakanishi, R, Ando, K, Ohgaki, K, Kitao, H, Saeki, H, Oki, E, and Maehara, Y (2017). Mitotic slippage and the subsequent cell fates after inhibition of Aurora B during tubulin-binding agent-induced mitotic arrest. Sci Rep 7, 1–12.
Tuupanen, S et al. (2014). Identification of 33 candidate oncogenes by screening for base-specific mutations. Br J Cancer, 1657–1662.
Viola, P, and Jones, M (2018). Robust Real-time Object Detection Paul. In: Second International Workshop on Statistical and Computational Theories of Vision– modeling, Learning, Computing and Sampling., 203–207.
Vitre, B et al. (2015). Chronic centrosome amplification without tumorigenesis. Proc Natl Acad Sci, 201519388.
Vladar, EK, and Stearns, T (2007). Molecular characterization of centriole assembly in ciliated epithelial cells. 178, 31–42.
Vogelstein, B, and Kinzler, K (1993). The multistep nature of cancer. Cancer 9, 138– 141.
Vorobjev, IA, and Chentsov, YS (1982). Centrioles in the cell cycle. I. Epithelial cells. J Cell Biol 93, 938–949.
Vulprecht, J et al. (2012). STIL is required for centriole duplication in human cells. J Cell Sci 125, 1353–1362.
162 Walker, RA, O’Brien, ET, Pryer, NK, Soboeiro, MF, Voter, WA, Erickson, HP, and Salmon, ED (1988). Dynamic Instability of Individual Microtubules. J Cell Biol 107, 1437–1448.
Waterman-Storer, CM, Worthylake, R a, Liu, BP, Burridge, K, and Salmon, ED (1999). Microtubule growth activates Rac1 to promote lamellipodial protrusion in fibroblasts. Nat Cell Biol 1, 45–50.
Watt, FM, and Sever, R (2008). Cell biology as the centuries change. J Cell Sci 121, 1– 2.
Winey, M, Toole, EO, and B, PTRS (2014). Centriole structure. Philos Trans R Soc B 369.
Wingate, H, Puskas, A, Duong, M, Bui, T, Richardson, D, and Keyomarsi, Y (2009). Low molecular weight cyclin E is specific in breast cancer and is associated with mechanisms of tumor progression. Cell Cycle 8, 1062–1068.
Wong, C, and Stearns, T (2003). Centrosome number is controlled by a centrosome- intrinsic block to reduplication. Nat Cell Biol 5, 539–544.
Woodruff, JB et al. (2015). Regulated assembly of a supramolecular centrosome scaffold in vitro. Science (80- ) 348, 808–812.
Woodruff, JB, Wueseke, O, Hyman, A a, and B, PTRS (2014). Pericentriolar material structure and dynamics.
Yamashita, YM (2009). The centrosome and asymmetric cell division. Prion 3, 84–88.
Yang, J, Adamian, M, and Li, T (2006). Rootletin Interacts with C-Nap1 and May Function as a Physical Linker between the Pair of Centrioles / Basal Bodies in Cells. Mol Biol Cell 17, 1033–1040.
Yang, Z, Lončarek, J, Khodjakov, A, and Rieder, and CL (2008). Extra centrosomes and/or chromosomes prolong mitosis in human cells: Control of mitotic duration. Nat Cell Biol 10, 748–751.
163 Yu, W, Kanaan, Y, Baed, YK, and Gabrielson, E (2009). Chromosomal changes in aggressive breast cancers with basal-like features. Cancer Genet Cytogenet 193, 29–37.
Zack, TI et al. (2013). Pan-cancer patterns of somatic copy number alteration. Nat Genet 45, 1134–1140.
Zhang, CZ, Spektor, A, Cornils, H, Francis, JM, Jackson, EK, Liu, S, Meyerson, M, and Pellman, D (2015). Chromothripsis from DNA damage in micronuclei. Nature 522, 179–184.
Zheng, D, Liu, X, and Tian, B (2016). 3′READS+, a sensitive and accurate method for 3′ end sequencing of polyadenylated RNA. Rna 22, 1631–1639.
Zheng, Y, Wong, ML, Alberts, B, and Mitchison, T (1995). Nucleation of microtubule assembly by a y-tubulin-containing ring complex. Nature 378, 578–583.
164