Visualizing centrosome movement in mouse embryonic stem cells

Visualizing centrosome movement in mouse

embryonic stem cells

THESIS

submitted in partial fulfillment of the requirements for the degree of

BACHELOR OFSCIENCE

in PHYSICS

Author : Otto van Henten

Student ID : S1682644

Supervisor : Stefan Semrau

2ndcorrector : John van Noort

Daily Supervisor : Esm´ee Adegeest

Visualizing centrosome movement in

mouse embryonic stem cells

Otto van Henten

Huygens-Kamerlingh Onnes Laboratory, Leiden University P.O. Box 9500, 2300 RA Leiden, The Netherlands

June 20, 2019

Abstract

In order to more easily track single cells in 3D cellular structures over time, an mCherry-γ-tubulin construct was implemented into mouse E14 cells. With this construct the cells produce

γ-tubulin with a fluorophore attached. Since γ-tubulin aggregates around the centrosome, the

centrosome appears fluorescent, showing a small dot in the cell upon excitation, which allows for easier tracking of the cell during a time-lapse. However, phototoxic damage to the cells and photobleaching of the fluorophore limit the amount of time that we are able to image the mESCs. With minimal excitation light intensity and exposure time, the cells were able to survive a 16 hour time-lapse while still having clearly visible centrosomes. With the centrosome visu-alized during time-lapses, an interesting behaviour of two centrosomes just after cell division was observed. The centrosomes of both daughter cells appeared to be correlated in movement over time, even though the mother cell had already divided. To see whether more centrosomes were spatially-correlated, two spatial distribution functions, the nearest neighbor distance and Ripley’s K, were calculated. Improvements of the analysis should show whether nearby centro-somes are indeed spatially correlated over time or not.

Contents

2.4 Mouse E14 Cells 14

3 Methods 15

3.1 Cell culture 15

3.2 Transfection of E14 with the mCherry γ-tubulin construct 16

3.3 From a polyclonal to a monoclonal cell population through FACS 16

3.4 Microscope Setup 17

3.5 The Nearest Neighbor distance and Ripley’s K 18

3.6 Data Processing 20

4 Results 23

4.1 Selecting the best clone 23

4.2 Time-lapse 25 4.3 rnnand Ripley’s K 29 5 Discussion 33 5.1 Selecting Clones 33 5.2 Ripley’s K 33 5.3 Future perspectives 35 6 Supplementary Information 37 6.1 Supplementary Figures 37 6.2 Appendix 1 44 6.3 Appendix 2 53 6.4 Appendix 3 55 6.5 Appendix 4 55

Chapter

1

Introduction

In live cell microscopy, one often wants to observe cellular processes such as, intracellular traf-ficking [1], movement of cells within a 3D structure or cell division over time. Several methods exist to observe these processes over time, for example, Sangmi Jun et al.[2] combined live cell fluorescence microscopy with Cryo-electron tomography (cryoET) to directly visualize the spa-tial and temporal behavior of HIV-1 particles in living cells. With this combination of cryoET with fluorescence microscopy, they were able to directly observe HIV-1 particles interacting with living (HeLa) cells at different stages of infection.

In fluorescence microscopy, fluorophores are used to fluorescently label certain parts of the cell, such as, proteins, the DNA or the cell membrane, which enables the visualization of ob-jects within the cell. By exciting the fluorophores with light of a specific wavelength, light of a shorter wavelength is emitted, which is the signal used to visualize the aforementioned objects. Subsequently, the fluorescent signal can be imaged over time in live cells, to observe cellular processes.

Critically, the success of a fluorescence microscopy experiment hinges on the the effectiveness of the fluorophore. The ideal fluorophore is bright, can be flexibly implemented into living cells and absorbs at long wavelengths. The long wavelengths are important because of the inverse re-lationship between energy and wavelength. A shorter wavelength means higher energy which can be detrimental to the experiment, since the high energy will cause damage to the cells, in the form of phototoxicity, and bleaching of the fluorophores.. Phototoxicity is a general term for the damaging effects light can have on cells. Both the excitation light as well as the emitted light causes damage to the cell [3]. The extent of the damage is influenced by a variety of factors such as, excitation wavelength, intensity and exposure time. Therefore, when doing fluorescent mi-croscopy experiments, a trade-off between signal strength and sample health needs to be made since the validity of the found results can be called into question if the cells no longer behave like they normally would, because of the phototoxic damage. In fluorescence microscopy, it is not possible to completely eliminate phototoxicity, but it is possible to minimize the photo-toxic damage by keeping the intensity of the excitation light low and the exposure time short. There also exist different microscope designs that minimize the phototoxic effects. Furthermore, when the fluorophores are in the excited state they can react with oxygen and become degraded [4]. This is called bleaching and will cause a decrease of the received signal. Both the bleach-ing and phototoxic effects are exacerbated by the fact that the cells usually need to be imaged

8 Introduction

over extended periods of time, a time-lapse, which will increase the damage done to the cells. On top of this, when using conventional fluorescence microscopes the entire sample is excited when taking an image of the plane that is in focus. This exciting of the entire sample causes increased bleaching, phototoxic effects and the fluorophores that are not in focus will also be excited, which, will increase the background signal and, therefore, decrease the signal to noise ratio (SNR). Despite these limiting factors, live cell fluorescence microscopy is an extremely useful tool in studying biological processes such as cell division, intracellular trafficking and embryonic developmental processes.

When studying embryonic developmental processes, single cells within a 3D structure would ideally be tracked with the help of fluorescent labels. To this end one has several options in choosing a part of the cell to visualize. One could choose to visualize the nucleus of a cell but with the mouse embryonic stem cells (mESCs) we have in culture, mouse E14 cells, this raises a problem. Specifically, the nucleus of the cells is quite big and, therefore, the signal from one cell could overlap with the signal from another cell in a 3D structure, which makes tracking a single cell over time more difficult. In order to clearly visualize a single cell in a 3D structure generated by the mESCs a novel idea is presented in this thesis. This idea is based on visual-izing the centrosome of the cell, a component of the cell that, among others, facilitates in cell division. A cell is comprised of exactly one centrosome which, compared to the nucleus, is a relatively small organelle, around 0.5 µm. The small size of this organelle reduces the probabil-ity of overlapping signal, making it a suitable object to track over time in 3D structures. This idea was based on the research of Jaensch et al. [5], who fluorescently labelled the centrosome in C. elegans embryos.

Additionally, to research the extent of the phototoxic and bleaching effects, the intensity of the excitation light was decreased such that the centrosome was still visible while damaging the cells as little as possible, with the end goal that the cells would survive an overnight timelapse (16 hours) and signal of the centrosome would still be clearly visible. To remedy the phototoxic-ity, bleaching and decreased SNR, steps were taken to develop a Light Sheet Microscope (LSM). A LSM minimizes the illumination light to a µm-thin light sheet, which is exactly placed at the focal plane, such that the other planse of the sample are not illuminated. Thereby, the cells are exposed to less illumination light during a typical imaging experiment, enabling much longer imaging times because the phototoxic and bleaching effects are decreased.[6]. On top of that the SNR is also decreased since no out-of-focus light is collected.

In this thesis, we take steps towards the tracking of a single cell within a 3D embryonic body. In the Theory, we review the principles underlying the behaviour of the centrosome, fluorescence microscopy and the cells used in the varying experiments. This is followed by Methods, were we discuss, how the cells were kept in culture and how the centrosomes were made fluorescent. Furthermore, the microscope setup, data processing and clone selection are explained. Lastly, in Results we discuss how we selected the clones used for the experiments, the extent and limiting of phototoxic damage and bleaching and the spatial distribution of the centrosomes and nuclei.

Chapter

2

Theory

2.1

The Centrosome

The centrosome plays a key role in several processes within the cell. First of all, it serves as the main microtubule organizing center (MTOC), an anchoring point for microtubules. Micro-tubules are part of the cytoskeleton, a network of filamentous polymers and regulatroy proteins [7] that, among others, dictates the shape of the cell, organizes the contents of the cell, connects the cell to the environment and allows the cell to move. Where the cytoskeleton consists of three main building blocks, namely, actin filaments, intermediate filaments and microtubules, the microtubules provide the longest structural component of the cytostkeleton, as they con-sist of, long hollow tubes made from α- and β-tubulin. The anchoring of microtubules to the main MTOC, the centrosomes enables the cell to divide in two daughter cells, one step in the multistep cell cycle process, see figure 2.1 [8]. The general shape of a cell is governed by the cytoskeleton, a network of filamentous polymers and regulatory proteins[7]. Besides dictating the shape of the cell, the cytoskeleton has three more functions. First, it spatially organizes the contents of the cell. Second, it connects the cell to the environment, both physically and bio-chemically. Third, it allows the cell to move, by means of changing its shape. There are three main building blocks which, with the help of regulatory proteins, make up the cytoskeleton. Namely, actin filaments, intermediate filaments and microtubules. Microtubules are hollow tubes made from α- and β-tubulin and on top of being a main component of the cytoskeleton, they also play a main part in ensuring that during cell division both daughter cells have chro-mosomes. Cell division is one of the four steps in the cell cycle which the cell continuously goes through, see figure 2.1[8].

10 Theory

Figure 2.1:Different phases of the cell cycle. G1, a gap phase where the cell grows and prepares for the

next phase. S phase, the phase where the DNA and the centrosome is duplicated. G2, the cell grows again

and the mitotic spindle begins to form. M phase, the phase where the chromosomes get divided into the daughter cell, mitosis, and where the actual cell divides, cytokinesis.

Figure 2.2:The different steps of M-phase[9]. 1). Interphase consist of the phases G1, G2and S of the cell

cycle. 2). Prophase, the chromosomes condense and pair up to form sister chromatids. 3). Prometaphase, the mitotic spindle attaches to the sister chromatids. 4). Metaphase, the chromosomes are positioned at the equator of the mitotic spindle. 5). Anaphase, the mitotic spindle pulls apart the sister chromatids. 6). Telophase, the chromosomes are packaged into separate nuclei

After a successful division the first phase of the cell cycle is the gap phase, G1, where the cell

grows and prepares for the next phase. This is followed by S phase where both the DNA and the centrosome, a microtubule organizing center (MTOC), are duplicated. Next comes another gap phase, G2, where again the cell grows and prepares for the next phase, M-phase (cell division).

2.1 The Centrosome 11

During G2the duplicated centrosomes are moved towards opposite ends of the nucleus where

they continuously recruit proteins that facilitate the anchoring of microtubules to the centro-some which will form the mitotic spindle. The phases G1, S and G2constitute interphase and it

is followed by the actual division of the cell, M-phase.

M-phase again consists of several different phases which are grouped into two bigger phases, specifically: mitosis, the distribution of the chromosomes into the daughter nuclei, and cy-tokinesis, where the actual cell divides in two. Mitosis starts with prophase which entails the duplication of the centrosome and the condensing of the chromosomes to form compact rods, sister chromatids. Between prophase and prometaphase the centrosomes move towards oppo-site ends of the nucleus while continuously recruiting different protein such as γ-tubulin and SPD-2 from the cytoplasm [5]. This is followed by prometaphase, where, the nuclear envelope breaks down and microtubules grow outward from the centrosomes to form the mitotic spindle, which subsequently, attaches to the sister chromatids. The chromatids will then be positioned at the equator of the spindle during metaphase. While this is happening the centrosomes also attach to opposite sides of the cell cortex so that when metaphase is completed and anaphase starts the chromosome pairs can be pulled apart, such that each daughter cell has a copy of each chromosome. Next the spindle is disassembled and the chromosomes are packaged into separate nuclei during telophase. This concludes mitosis and cytokinesis then cleaves the cell into two daughter cells so that each daughter cell inherits a nuclei and a centrosome. The cen-trosome is consists of two structures, the centrioles.

Figure 2.3:The centrioles, which consist of nine microtubule triplets, are arranged at right angels in relation to each other and form the centrosome. The entire structure is surrounded by the pericentriolar material which contains several proteins of which γ-tubulin is the most abundant.

12 Theory

These two centrioles are positioned at right angles in relation to each other and are compised of a set of nine microtubule triplets arranged in a circular fashion[8]. The centrioles are sur-rounded by the pericentriolar material (PCM). The PCM consists of a variety of proteins, such as γ-tubulin, cop152 and cop128[10]. Among these protiens γ-tubulin was found to be the most abundant in the PCM . Therefore γ-tubulin would be ideal as a proxy for visualizing the centrosome since if it is possible to visualize γ-tubulin, the centrosome will automatically be vi-sualized. For example, Jaensch et al. [5] were able to develop an automated system for tracking and measuring fluorescently labeled centrosomes in living C. elegans embryos. Using fluores-cence is one of the main ways we are able to visualize processes and objects within the cell by means of fluorescence microscopy.

2.2

Principles of Fluorescence

In fluorescence microscopy, molecules called fluorophores can be used to visualize (certain parts of) the cell. These fluorophores can be excited when they are exposed to light of a specific wave-length. Upon excitation, fluorophores absorb light energy of the excitation wavelength, which excites electrons from the ground state, S0, to an excited state, S2, see figure 2.4. Eventually,

the electrons will relax to the lower excited state S1and subsequently they will fall back to the

ground state, where the electrons emit a photon in the visible light spectrum which we observe as fluorescence. Since the energy of the emitted photon is lower than the energy of the excitation light, emission light of a shorter wavelength is detected as the energy is inversely related to the wavelength, as follows from equation 2.1

E= hc

λ (2.1)

with E the energy, h the Planck constant, c the speed of light and λ the wavelength.

Both the excitation as well as the emission light cause damage to not only the cell, in the form of phototoxic effects, but also to the fluorophore, as bleaching. Exciting fluorophores with light will always produce reactive oxygen species (ROS). Under normal circumstances cells have multiple ways to deal with ROS[4] but if these defense mechanisms get overwhelmed, for ex-ample by repeated exposure , the cell will no longer be able to prevent damage from happening. This will lead to the ROS reacting with a variety of easily oxidizable components, such as pro-teins, nucleic acids, lipids and fluorophores. When the ROS react with the fluorophores they will bleach, when the ROS react with any other components of the cells it will cause phototoxic effects which will eventually lead to cell death. The extent of the damage caused by phototoxic effects are dependent on a variety of factors including, excitation light wavelength and inten-sity, exposure time, time between images and fluorophore location and concentration. When setting up fluorescence microscopy experiments it is paramount to limit the phototoxic effects by, for example, lowering excitation light intensity and exposure time. Yet this will decrease the amount of received signal and, therefore, decrease the SNR. Accordingly, the camera setting should be tuned to use as low of an intensity and exposure time as possible while still receiving enough signal that allows the objects to be detected

2.3 Fluorescence Microscopy 13

Figure 2.4:The energy levels of the electron in fluorophores. The excitation light excites an electron from the ground state, S0, to the second excited state S2and relaxes back to excitated state S1. Subsequently it

falls back to the ground state, S0, while it emits a photon of a shorter wavelength which is the fluorescent

signal we observe.

2.3

Fluorescence Microscopy

For the experiments in this thesis, an inverted (epi)fluorescence widefield microscope was used. In an inverted microscope one is looking at the sample from below, which is advantageous in cell biology since most samples contain cells that are grown on the bottom of the dish. There-fore, an inverted microscope will allow for the cells to be observed better since the working distance of objectives with higher magnification, for example 40x or 100x, is very small. The objective has to be close to the sample to be able to visualize it. There are several limitations with widefield microscopes though. First of all, the excitation and emission light will travel through the entire sample leading to more phototoxicity and bleaching.The illumination of the entire sample will also lead to exciting fluorophores which are out-of-focus, which will increase the noise, resulting in a decreased SNR. To block this out-of-focus light, the confocal microscope was designed, which uses a pinhole in front of the camera, to block all out of focus light. Mi-croscope techniques have been developed that reduce photoxocity and bleaching of the sample, such as the two photon microscope or a LSM.

In two-photon microscopy, two photons, instead of one, with half of the excitation energy are used to excite a fluorophore. To enable excitation with two photons of lower energy, a high local instantaneous intensity needs to be created by both a tight focus of the laser and high temporal concentration of the laser pulses. When this light is applied to a fluorophore, the probability of the fluorophore to absorb two photons becomes significant. There are two main advantages of using two photon microscopy. First of all it can penetrate up to 1 mm into tissue which allows for improved optional sectioning. Secondly, since two-photon microscopy only excites a single focal point at a time, the phototoxicity is kept at a minimum. [11].

14 Theory

into the sample and another detection arm which directs the emitted light to the camera[12]. With a laser and a scanning mirror a µm thin sheet of light is created which is directed into the sample. This sheet is positioned in the plane of focus and, therefore, only in-focus fluorophores will be excited. There are two main advantages to a LSM: First of all, since only a small part of the sample is excited at a time the phototoxic effects will be reduced, which enables better sam-ple health and, therefore, longer imaging times. Secondly, the SNR will be increased since the fluorophores that are out of focus are not excited which reduces the background noise. Depend-ing on the design of the LSM, 3D images can be made by either movDepend-ing the sample through the light sheet or moving the light sheet through the sample. Several designs of the LSM exist. The LSM discussed in this thesis is based on the inverted LSM designed by Strnad et al.[13]. With these microscopes a lot of different types of cells can be imaged.

Figure 2.5:The light path of the LSM worked on in this thesis. The laser light is generated by a lasers within a laser combiner with four different lasers, which allows for the excitation of a variety of fluorophores. The laser light passes through a fiber (F) and is, subsequently, scanned across the sample by a scanning mirror (SM), but before the light reaches the sample it is directed through a scanning lens (SL), a tube lens (TL), several kinematic mirrors (KM), a beam splitter (BS) and into the illumination objective (IL). The emmited photons, from the excited fluorophores, are collected by the detection objective (DE) and directed into another tube lens, through a motorized filter wheel and ultimately into the IRIS 15 CMOS camera.

2.4

Mouse E14 Cells

The cell line used in all the experiments of this thesis is the E14 mouse embryonic stem (ES) cell line. This ES cell line was first derived from strain 129/Ola mouse blastocyst by Dr Martin Hooper in Edinburgh[14]. These ES cells are often used in Semrau’s lab to model early em-bryonic events in vitro. Single cells in culture aggregate and form embryoid bodies. When treated as described by Van den Brink et al. [15] they form gastruloids, which show similari-ties with mouse gastrualting embryos. To further study the processes happening within these gastruloids, single cells would ideally be tracked over time. To this end, E14s were generated expressing fluorescent γ-tubulin, allowing for easier single cell tracking by following the cen-trosomes.

Chapter

3

Methods

3.1

Cell culture

The E14 mouse ES cells were cultured in two different media Serum with LIF and 2i. 2i was used to brings ES cells back to a naive ground state, where they exhibit greater pluripotent gene expression than ESC cultivated in Serum with LIF. 2i was named after it was found that adding two inhibitors (PO & CHIR) to the medium brought these ES cells back to their ground state[16]. The serum with LIF ESC medium (Knockout DMEM (Gibco, US), contains 10% ES certified FBS (Sigma, US), 1% NEAA, 0.144mM β-mercaptoethanol , 2mM L-glut , 1000 unit-s/ml mLIF, 1x pen/strep. The 2i ESC medium consists of DMEM (Sigma,US), 0.5% NEAA, 0.1001mM β-mercaptoethanol , 1x pen/strep, 1x N2 supplement, 1x B27 supplement, 0.5mM L-glut, 0.02mg/ml Human insulin, 1µM PD0325901, 1µM CHIR99021 and 1000 units/ml mLIF. The function of each ingredient in the two ESC media is summarized in Appendix 2. The cells were grown in an incubator at 37°C 5%, CO2and every 1 or 2 days when the dish was around

80% confluent they were passaged into a new dish under sterile conditions. Below the passage process is described, the amounts mentioned are the amounts needed for a 6cm petri dish. Us-ing a different size dish will require different amounts.

To prepare the new dish, 4 ml of gelatin (0.2%) was pipetted into the dish and the dish was swerved to ensure that the gelatine was spread everywhere. The dish was then put into the incubator for a minimum of 10 minutes. Next, the medium of the old dish was removed and the dish was washed with 4-5ml PBS (Sigma, US). Afterwards, 1 ml of trypsin (0.25% in PB-S/EDTA) was added if the ESC medium was Serum + LIF and in the case of 2i medium, 0.5ml of accutase, (EMD Milli crop, US) was used instead. The dish was then put into the incubator for 3 minutes and subsequently, 2ml of the correct ESC medium was added to deactivate the trypsin or the accutase. Next a single cell suspension was made by pipetting the cell up and down several times. When a single cell suspension was reached, it was pipetted into a tube and spun down in a centrifuge (3 minutes, 1200 rpm). Simultaneously, the gelatine of the new dish was removed and 5 ml of ESC medium was added. When the centrifuge was done, the supernatant was removed and the pellet of cells was resuspended with 2 ml of ESC medium. In general, the cells were passaged in a 1:6-1:8 ratio. Sometimes, though, when it was necessary to be more exact in the amount of cells that were passaged, the cells were counted using a TC 20

16 Methods

automated cell counter (Bior rad, US). To count the cells 10 µl of cell suspension was combined with 10 µl trypan blue, (Sigma, US) in an eppendorftube. Subsequently, this suspension was placed in a counting slide (Bio rad, US) to count the amount of live cells.

Cells were maintained in culture up to passage number 25. After this passage number, the cells are more likely to incorporate mistakes in their DNA. Therefore, it is no longer possible to do experiments with the cells since it will be unclear to what extent these mistakes will influence the experiments.

3.2

Transfection of E14 with the mCherry γ-tubulin construct

The plasmids that were used to create the fluorescent γ-tubulin necessary for the experiments of this thesis were purchased from Addgene but they needed to be altered slightly before they could be used. These alterations start with replacing the CMV promoter since this promoter is turned off in mouse E14 cells during differentiation which results in the protein no longer being produced. The promoter was replaced by the EF-1α promoter which was also purchased from Addgene. To change the promoter, the plasmid was cut open by restriction enzymes at the SnabI and BmtI sites, highlighted in supplementary figure 6.2 . The SnaBI restriction enzyme gener-ates a blunt end while the BmtI resctriction enzyme genergener-ates an end with an overhang. When a restriction enzyme leaves an overhang, the open nucleotides create a very specific bonding site making it easier for the new promoter to bond to if it has the right sequence. In comparison, a blunt cut will only leave the sides of the nucleotides open creating a non-specific bonding site which is more difficult to bond to, see supplementary figure 6.1. Before the EF-1α promoter was placed into the plasmid, by means of ligation, it was duplicated with a polymerase chain reaction (PCR) which is a technique to amplify short DNA sequences, see suplementary figure 6.3. Next, it was placed in bacteria, by transformation, so that the entire plasmid was multi-plied. The plasmids were retrieved by means of several purification steps. Subsequently, the plasmids were transfected into mouse E14 cells with the help of lipofectamine 3000, a reagent which leverages nanoparticle technology to improve transfection efficiency. Aside from having the fluorescent γ-tubulin DNA, the plasmid also contains DNA for two types of antibiotic re-sistance, one for bacterial- and one for mammalian cells. These resistances can be used to select cells which have incorporated the construct properly by treating the cells with antibiotics. Be-fore the plasmids were transfected into the E14 cells they were cut open by restriction enzymes since the cells incorporate linearized DNA more easily. Next, the cells were put on fibroblasts to facilitate growth. After antibiotics treatment, however, the selected cells were polyclonal, meaning that some but not all cells have incorporated the construct. To ensure that all the cells in the culture had implemented the construct, the cells needed to be sorted.

3.3

From a polyclonal to a monoclonal cell population through

FACS

To get a monoclonal cell culture, the cells were sorted at the Leiden University Medical Center (LUMC) by means of fluorescence activated cell sorting (FACS) [17]. In FACS, fluorescent cells are integrated into a liquid stream while they are illuminated by a laser beam one at a time. Downstream of the laser, the stream is broken up into uniform sized droplets which could be selected and plated depending on the strength of the fluorescent signal. The cells were sorted

3.4 Microscope Setup 17

to have an intermediate fluorescence signal for mCherry, since a too weak signal would not allow for useful experiments because the centrosomes would not be visible enough during a time-lapse. Moreover, a too strong signal would indicate that the construct was incorporated multiple times, which is detrimental for cell viability. Single cells of the selected population were put into a single well of a 96-well plate. This plate was put in the incubator for a while so that the single cells could grow into bigger colonies. Subsequently each well was imaged to see whether the colonies exhibited fluorescence and to judge how strong this fluorescence was. If there were multiple colonies within one well or if the colony exhibited no fluorescence at all, the clone was immediately excluded from selection for further experiments. For the wells with fluorescent colonies it was gauged by eye whether the entire colony was fluorescent and from these colonies, five were selected that looked the best. With these clones a monoclonal culture was created and several vials of each clone were frozen when the passage number was still relatively low. Therefore, cells were trypsinated and placed in ESC medium with 10% DMSO and put into cryovials. These cryovials were put into a -80°C freezer and after 2-3 days the cryovials were put into a -150°C freezer for indefinite storage. Adding DMSO is necessary because it functions as a cryoprotectant by preventing the formation of ice crystals during the freezing process which in turn prevents the cells from dying. However, DMSO itself is also harmful for the cells and therefore the process between adding DMSO and freezing was done quickly. When the passage number of the existing cells reached 25, the cell line was restarted by thawing a cryo-vial and quickly adding 9 ml of medium to to the cell suspension. This was then spun down (3-5min, 1000-1500 rpm), the medium was removed and 10 ml of new medium was added which was again spun down to make sure all the DMSO was removed. Afterwards the 10 ml medium was removed and the cells were resuspended with 2 ml medium and plated like described in section 3.1. The exact protocol for freezing and thawing the cells can be found in the supplementary information.

3.4

Microscope Setup

The microscope used in this thesis was a Nikon Eclipse Ti. Furthermore a Tokai Hit TiZW was used to keep the cells at 37 °C and with 5% CO2while doing a time-lapse, see figure 3.1. NIS

elements AR 45000 was the imaging software used in all experiments. For both the Hoechst ex-periments and the first two time-lapse exex-periments, the excitation light for the mCherry signal, 594 nm, had a 15% intensity with an exposure time of 300ms and for Hoechst, a 390 nm, light with 2% intensity and 50 ms of exposure time was used. When adjusting the settings in order to decrease the phototoxicity and bleaching, both the intensity and the exposure time were first lowered to 11% and 200 ms, respectively. When this was not enough of a decrease for the cells to survive, in subsequent experiments the light intensity was lowered to 3% and the exposure time to 100 ms, such that the signal of the centrosomes was still strong enough to be detected In all the experiments the complete colonies were imaged by means of a z-stack. This means that at different heights (z positions) an image was taken. The steps size and the amount of steps differ per experiment, but such that, the entire colony was imaged in around 20 steps. In the first two experiments, the 100x objective was used to clearly see individual cells even within colonies. Later, the experiments were done with a 40x objective because this increases the field of view while single cells could still be distinguished. There were also experiments where 2µg/mL Hoechst 34580, solved in DMSO, was added to the culture so that the nuclei would be stained. The cultures were in 6 cm Ibidi imaging dishes, one for each clone, and after adding the Hoechst the cells were placed back into the incubator for 15 minutes. Before imaging

18 Methods

the medium was replaced with fresh medium and then z-stacks of ten colonies per dish were taken. With the nuclei visualized, we could investigate the spatial distribution of both the cen-trosomes and the nuclei.

Figure 3.1:The Tokai Hit TiZW to keep the cells in an incubator, 37 °C and 5% CO2, while doing

time-lapses.

3.5

The Nearest Neighbor distance and Ripley’s K

After the second experiment was done, it looked like two centrosomes were connected even after cell division. To quantify whether this was happening Pearson correlation coefficient was calculated:

P= Cov(X, Y)

σxσy (3.1)

This returns a value between -1 and 1, with -1 meaning perfect anti-correlation, 0 meaning no correlation and 1 meaning perfect correlation. To look into this matter further the spatial distribution of centrosomes was look at with the question, whether the spatial distributions of centrosomes is completely random or if there is a preferred distance at which centrosomes are positioned in relation to each other. To this end, both the nearest neighbor distance (rnn) and

Ripley’s K were calculated. The rnnis the distance to the closest centrosome from a reference

centrosome. Making a histogram of all rnnand comparing it to the expected rnnwhen the points

are distributed randomly, calculated with equation 3.2, a preferred distance might emerge. Ripley’s K is a measure of how clustered a group of points is. It is calculated using the following

3.5 The Nearest Neighbor distance and Ripley’s K 19 formula[18] : K(r) = A N

∑

∑

Figure 3.2:The calculation of Ripley’s K. The blue point is the reference point and the green points are the points within a radius r of the reference point and where I = 1 in equation 3.2. The red points are the points where I = 0 in 3.2. Furthermore, the nearest neighbor distance, rnn, is also shown in red.

Here A is the area, N is the number of points, I is 1 if a point is within a circle of radius r around a reference point and zero otherwise, and w adjust for the edge effects. The edge effects arise from the fact that points outside the study area are not counted even when they are within the radius r. Ignoring these effects biases K(r) especially for larger r. There are several ways to correct for these edge effects and in this thesis the method proposed by Ripley will be used[19]. Ripley’s edge corrector is given by the proportion of the circle which is still inside the study area, see figure 3.3. This proportion can be calculated by knowing the distance to each bound-ary, combined with the fact that the circle will intersect the boundary at radius r. The angles necessary to calculate the proportions of the circle within the boundary can then be calculated with trigonometry.

If a collection of points is randomly distributed in space (complete spatial randomness, CSR) the expected value for K is,

K=πt2 (3.3)

With this fact, we can calculate the L function: L=

r K

π (3.4)

20 Methods

(a) (b)

Figure 3.3:When taking the edge effects into account, the proportion of the circle, of radius r, that lies within the boundaries of the of the study area is calculated. When the distance to the boundary, d1−4, is

combined with the fact that the circle and the boundary intersect at length r the angles needed, to calculate the proportion of the circle that is outside the boundary, can be calculated with simple trigonometry. When the circle is near the corner two cases have to be taken into account, one where the corner is inside the circle (a) and one where the corner is outside the circle (b)

are randomly distributed. Furthermore the points are clustered when L−r > 0 and they are spatially regular when L−r<0.

3.6

Data Processing

To process the z-stacks, they were first loaded into Fiji and a maximum z-projection was made. This means that a single image is created with every pixel being the brightest pixel of the entire z-stack at that position. In case of time-lapse experiments this was done for every time-step. To enable improved calculation of rnnand Ripley’s K, for both the Hoechst as well as the mCherry

signal, the images were cropped to only contain a single colony. Furthermore, the background was subtracted, a Gaussian blur was added, σ = 2, and a bandpass filter was applied only to the mCherry signal, where structures between 0 and 10 µm were kept, and bigger structures where filtered. With these settings only singular dots of the centrosomes where left. Then, to analyze the data, four python scripts were written which can be found in the Appendix 1. The first script made was made to measure the degradation of the signal in single cell time-lapses and started with looking at the brightest pixel in the first time-step. It then measured the signal and position of this brightest pixel and looked in the next time-step for the brightest pixel within a radius of a 150 pixels to ensure that a single centrosomes was followed the entire time. These steps where consequently repeated for the entire time-lapse resulting in a graph of the degradation of the signal. The problem with this method was that when colonies were, imaged a centrosome could move farther in between time-steps than the distance between neighboring centrosomes. The first script was therefore not able to ensure that a single centrosome was followed the entire time. To still measure the degradation of the signal, centrosomes needed to be followed manually, by using the multipoint tool in Fiji and clicking on the centrosome every time-step. All the necessary data was then extracted and run through a the second python program which read out the data from Fiji and made the graphs of the signal over time. The third python script was made to calculate rnn and Ripley’s K. To this end, first a local

3.6 Data Processing 21

binary image was created by looking at where the original image had higher pixel values than the image with the threshold applied.

Next, objects in the images where identified with a function, see Appendix 1. With the objects found, rnn could be calculated with Pythagoras’ theorem and plotted in a histogram. In this

same graph the expected rnnfor spatially random points was plotted according to the formula:

P(r) =2πrσe−σπr2 (3.5)

The full derivation of this formula can be found in the supplementary information. Further-more, Ripley’s K was calculated with the use of a python function, see supplementary figures 3.2. With Ripley’s K calculated, first the L function and then L−r could be calculated and plotted. To compare the L−r graphs of the acquired data to L−r from a spatially random dis-tribution of points, the fourth script was made. This script follows the same steps to calculate Ripley’s K, but instead of using the processed images as data, it uses a simulated matrix with randomly distributed points with the same density as in the acquired data.

Figure 3.5:A simulation of the centrosomes in the form of a matrix with randomly distributed points. With this Ripley’s K and the L and L-r functions were calculated to compare with the results from the experiments.

22 Methods

Figure 3.4:The workflow used when processing data from the experiments. 1) The collection of the data with the experiments in the form of z-stacks. 2) Make a maximum z-projection of the z-stacks. 3) Subtract the background from the maximum projection. 4) Add a Gaussian blur to the image (σ=2). 5) A bandpass filter was applied where structures between 0 and 10 µm were kept, and bigger structures where filtered. 6) With the python program a binary image was created with the help of a local threshold and in this image a function was used to detect the objects within the image. (Here the yellow is where the original image has higher pixel values than the thresholded image, the purple is were the original image had lower pixel values than the thresholded image and the red dots are the found objects.) 7) From these detected objects the relevant data was extracted and graphs were made. For the processing of the degradation of the signal this, workflow was followed up till step 2. For the processing of the Hoechst images, step 5 was skipped.

Chapter

4

Results

4.1

Selecting the best clone

(a)

(b)

Figure 4.1:Colonies from the 96-well plated that were sorted by FACS with colonies that did exhibited fluorescence (a) and those that did not (b).

As can be seen in figure 4.1 some clones exhibited no fluorescence, some a lot and in some wells multiple colonies were formed. The wells with multiple colonies and the wells where colonies exhibited no fluorescence were immediately excluded from further selection. From the wells with fluorescent colonies five were selected. To select the ’best’ clone from these five, several tests were done to determine how it would meet the following criteria. First of all, the selected clone would need a reasonably high signal intensity, making centrosomes easily visible, with as little variance between the intensity of different centrosomes, as possible. In addition, the clone should have a high SNR making it easier to distinguish the centrosome signal from the background signal. To test which clone would meet these criteria best, the five different clones were plated in an 8-well ibidi imaging chamber which was placed into the incubator for a day so that they would grow colonies. Unfortunately, clone two died and therefore no measurements were done on this clone and it was automatically excluded from selection. The next day, the clones were imaged in 3D by taking a z-stack using the widefield microscope. For each clone, images were taken of 10 different colonies. The images were processed using a python script,

24 Results

see Appendix 1.

Figure 4.2:Selecting a significant amount of centrosomes and the background signal next to the centrosomes. With this data, the signal intensity, standard deviation and SNR was calculated.

From the maximum projections of the z-stacks, the average intensity of the cenrosome signal and the background signal was measure, see figure 4.2. By dividing the average centrosome signal with the average background signal, the average SNR was calculated for each clone, see table 4.1. Clone 1 showed the lowest average intensity signal for the centrosomes. Therefore, clone 1 was not selected for further tests

Clone Signal±Std Coefficient of variation Noise SNR N

1 4218±1302 0.31 2307 1.789 258

3 32787±12315 0.38 17887 1.833 242

4 10306±5166 0.50 5320 1.937 257

5 9412±2825 0.30 5478 1.718 255

Table 4.1:The table used in selecting the clones for further experiments. In this table the signal intensity with the standard deviation (std), coefficient of variation (std/mean), noise, SNR and the number of used points, N, are shown. With these values clone 1 was excluded from selection since the signal intensity was too low. Clone 3 was excluded since the signal intensity was too high and, therefore, clone 4 and 5 were selected since they had comparable values.

Furthermore, clone 3 was not selected, because both the intensity and the standard deviation were too high. Clone 4 and 5 showed relatively comparable values for the centrosome signal intensity, where for clone 4 a higher SNR was found and for clone 5 a lower standard deviation.

4.2 Time-lapse 25

Therefore, both clones were selected for further tests.

To determine whether each cell had a fluorescent centrosome, an experiment was done where the nuclei were colored by adding Hoechst to the medium. Subsequently, z-stacks were made for 10 positions for both clone 4 and 5. Obtaining a quantitative number on the amount of cells that had a fluorescent centrosome was not possible due to time constraints. To still get a gauge of how many cells had a fluorescent centrosome a composite image was made form the binary image of the Hoechst signal and the found objects from the centrosome signal, see figure 4.3 From this image it can be discerned that a lot of cells have a detected centrosome, yet cells

Figure 4.3:A composite image of the Hoechst signal, blue, with the brightfield, grey, with the detected centrosomes, white dots. At position 1 and 3 we see a lack of detected centrosomes. At 2 we see detected centrosomes on either side of a nucleus.

without a fluorescent centrosome are also visible.

4.2

Time-lapse

To investigate the extent of the phototoxic damage and the bleaching of the fluorophores, sev-eral time-lapse experiments were performed. In the first two experiments, the plates were pas-saged and plated in the morning and imaged in the afternoon. Therefore, the plates were still sparse allowing for clear visualization of single cells with a 100x objective. For each clone, the cells were imaged at 10 positions in the dish for a duration of 16 hours taking an image every 15 minutes with the excitation light intensity at 15% and an exposure time of 300 ms. Using the

26 Results

first python script, see Appendix 1, the signal intensity of the centrosome was measured and plotted over time, see figure 4.4.

Figure 4.4:The degradation of the signal intensity of the second time-lapse experiment. The settings used for this experiment were, 15% excitation light intensity and an exposure time of 300 ms. The first and last images taken from this time-lapse can be found in supplementary figures 6.4

From the time-lapses we could see that the cells were not able to survive the full 16 hours. The damage caused by the light was likely the main reason and therefore the light intensity and exposure time were lowered. Furthermore, the fact that the plates were sparse also was a contributing factor, since cells are more viable when they grow in colonies. To accommodate for this all further time-lapses were done a day after the cells were plated which resulted in more confluent plates and in turn this resulted in more viable cells. The setting for the next time-lapse were: 3% light intensity and an exposure time of 100 ms. On top of this, to image the centrosomes in a more 3D environment, the cells were placed in 2i medium such that they would from more round colonies comared to when they grow in Serum + LIF medium To pro-cess these time-lapse the maximum z-projection was made and the second python script, see Appendix 1, was used to measure the signal. With these settings the cells were able to survive the full 16 hours and the bleaching was minimal, see figure 4.5 and 4.6:

4.2 Time-lapse 27

Figure 4.5:The degradation of the signal in the sixth time-lapse experiment with clone 4. The settings used for this experiment were, 3% excitation light intensity and an exposure time of 100 ms. With these settings the cells were able to survive and the centrosomes were clearly visible at the end of the

experiment.The first and last images taken from this time-lapse can be found in supplementary figures 6.5

Figure 4.6:The degradation of the signal in the seventh time-lapse experiment with clone 5. The settings used for this experiment were, 3% excitation light intensity and an exposure time of 100 ms. With these settings the cells were able to survive and the centrosomes were clearly visible at the end of the

experiment.The first and last images taken from this time-lapse can be found in supplementary figures 6.6

In conclusion, when the time-lapses were done with the cells growing in colonies and the inten-sity and exposure time lowered significantly, the cells were able to survive and the centrosome signal remained clearly detectable. To improve cell viability and decrease photobleaching even

28 Results

further, work has begun for designing a LSM. To this end, several steps were taken for the cre-ation of a LabVIEW environment to operate both the camera and the lasers. As a consequence of doing time-lapses with the centrosomes visualized, interesting behaviour of two centrosomes after a cell division was observed, see figure 4.7. In the experiment with sparse plates, the

divi-(a) (b) (c)

(d) (e) (f)

Figure 4.7:Six frames from the dividing cell where it looks as if the centrosomes are still connected, even after the cell has divided. a) The two centrosomes have been moving away from each other to pull the chromosomes into the daughter cell. b) Cytokinesis cleaves the mother cell into two daughter cells each with one centrosome. c) After the two daughter cells have been separated from each other, the

centrosomes move back to the line of division. d-f)After having moved up and down on the line of division the centrosomes move with similar trajectories and speed away from the line of division. The brightfield images of this can be found in supplementary figures 6.8

sion of a cell was observed. Interestingly, after the cell had divided, and any connection between the two centrosomes should have been severed, it was observed that the centrosomes behave as if they were still connected. After the centrosomes moved apart to pull the chromosomes into the new daughter cells, both centrosomes moved with similar speed and mirrored trajec-tories back to the line of division, see figure 4.6 c. There, the centrosomes together moved up and down for a while and ultimately, again with mirrored trajectories and similar speed, they moved away, see figure 4.6 d-f. Note that when the centrosomes moved towards the opposite sites, the cell was presumably in the process of dying since this was at the end of the time-lapse with the high intensity and exposure time. To quantify the correlation of the movement of the two centrosomes, the Pearson correlation coefficient was calculated according to equation 3.1 which gave a value of 0.92 meaning that the movement is correlated. Furthermore the correla-tion coefficient of two centrosomes in neighboring cells was calculated as a comparison and this returned -0.54 meaning that the movement is not or slightly anti correlated. This observation raised the question of to what extent the movement of all centrosome pairs are correlated.

4.3 rnnand Ripley’s K 29

4.3

r

and Ripley’s K

To analyze if there is an interaction between the centrosomes which correlates the movement of centrosome pairs, both the nearest neighbor distance, rnn and Ripley’s K were calculated

from their trajectories. Calculation of rnnresulted in the following graph, where a histogram of

rnn is plotted with the expected rnn when the points were distributed randomly in space, see

figure 4.8 From this graph it can be discerned that the centrosomes are not distributed randomly

(a) (b)

Figure 4.8:The histogram of the nearest neighbor distance, blue, plotted, for two positions, with the expected nearest neighbor distance, red, if the centrosomes were distributed randomly through the image. From this graph it can be discerned that the centrosomes are not distributed randomly across the image since we see a smaller second peak right from the larger peak, indicating that there is a second distance centrosome pairs prefer to be at compared to a random distribution of points

across the image since we see a smaller second peak right from the larger peak, indicating that there is a second distance centrosome pairs prefer to be at compared to a random distribution of points. From these graphs, however, it is not possible to conclude whether this second peak is due to an interaction between centrosomes or due to the fact that the centrosomes are packed within a colony. Next, to ascertain how clustered the centrosomes might be, Ripley’s K and the corresponding L and L-r functions were calculated. This was first done for the simulation of randomly distributed points. Then, the Hoechst and centrosomes signals were used to calculate Ripley’s K:

30 Results

Figure 4.9a:Ripley’s K calculated for the centrosome signal, yellow, the hoechst signal, blue, and the matrix with randomly distributed points, green, with the 90% confidence interval. From this graph we can see that the nuclei of the cell are distributed randomly, while the centrosomes are increasingly clustered at longer radii.

Figure 4.9b:Ripley’s K calculated for the fifth position. In this graph the centrosome signal, yellow, the hoechst signal, blue, and the matrix with randomly distributed points, green, with the 90% confidence interval.From this graph we can see that the nuclei of the cell are distributed randomly, while the centrosomes are increasingly clustered at longer radii.

If there is a preferred distance between centrosomes of neighboring cells, it was expected that a small bump above zero was observed for short distances in the L-r graph, meaning that for short distances the centrosomes were clustered. At longer distances, the L-r graph was expected to be around zero or not exceed the clustering of the nuclei since we do not have information on long distances, since the movie only showed one dividing cell and, therefore, no long range clustering beyond the clustering of the cells themselves was expected. When looking at the

4.3 rnnand Ripley’s K 31

L-r graphs, however, we can see that neither expectations are expressed by the data. Rather, the centrosome signal shows a gradual increase meaning that the centrosomes are increasingly clustered at longer r values. The Hoechst signal hovers around the confidence interval of the simulated data suggesting that the cell nuclei are randomly distributed through the image. Note that the initial drop below zero, of both graphs, can be explained by the fact that there is only one centrosome and nucleus per cell and therefore a neighboring centrosome or nucleus has to be a minimal distance away. In order to see if the observed behaviour was dependent on the positions that were imaged, the L-r graphs of all ten positions were averaged.

Figure 4.10:The averages of Ripley’s K of all ten positions. The random signal was made by calculating Ripley’s K for a simulated matrix of randomly distributed points, which was run 100 times and averaged.

Figure 4.11:The averages of Ripley’s K were the tenth position is excluded. The random signal was made by calculating Ripley’s K for a simulated matrix of randomly distributed points, which was run 100 times and averaged. Due to limitations of the method, at the tenth position the local threshold function was unable to separate a lot of the nuclei that were positioned close together. The effect of this is that the Average of the Hoechst signal is lower and the centrosome signal increases less quickly.

32 Results

In the averaged graphs we still see the gradual increase of the centrosome signal and Hoechst is, after a small increase, constantly clustered. The reason that two graphs of the average over all the positions are shown, is that in figure 4.10 b a measurement that might be considered an outlier was excluded, because the method is limited in the local threshold function used for object detection. This function was, for the Hoechst signal, sometimes unable to separate nuclei that were positioned close together. Every image was affected by this, but the tenth position to such a higher degree that exclusion might be warranted. Furthermore, other limitations of the method affected the centrosomes signal as well and will be elaborated on in the discussion.

Chapter

5

Discussion

5.1

Selecting Clones

In this thesis, mESCs were generated that express mCherry-γ-tubulin, so that their centrosomes can be visualized by means of fluorescence microscopy. This visualization enablus us to track the cells over time in 3D structures, which we aim to do in the future with our gastruloids model. When generating clones of mCherry-γ-tubulin expressing cells, the best clone needs to be selected before experiments can start. However, this raises the question: What is meant with the best clone? Depending on the factors one deems most important, an argument could be made for almost every clone to be selected. For example, clone 1 showed the lowest signal intensity but also the lowest standard deviation in intensity, indicating that that the

mCherry-γ-tubulin construct was implemented fewer times in the DNA, which would mean that a cell

more closely resembles the original cell. On the other hand the low signal would make detecting every centrosomes more difficult. For clone 3, the bleaching of the fluorophores would have less of an effect since the signal was relatively high, but an extremely high signal could indicate that the mCherry-γ-tubulin construct was implemented multiple times, which would mean that cell viability could be impaired.

5.2

Ripley’s K

As eluded too before, there were some issues with the method which was used to calculate Ripley’s K. First of all, the offset parameter of the local threshold function had to be tuned man-ually to detect the centrosome signal properly. When the offset was too low, objects would be detected were none should have been detected. Erroneous objects were detected, at the border of the colony, in empty space where there were no cells or in places where no centrosomes were visible in the processed image. To counteract the detection of non existent objects, the offset had to be increased, but to determine how far it should be increased had to be done by eye, which is a very inaccurate procedure. Furthermore, increasing the offset also led to the missing of, by eye clearly visible, centrosomes, while objects in completely dark spaces were still found.

34 Discussion

Therefore, we aimed to find a balance between missing clearly visible centrosomes and finding objects where none should exist.

(a) (b)

Figure 5.1:The processed image with the detected centrosomes, white dots, on top. A low offset (a) will result in the overestimation of detected centrosomes. Increasing the offset will remedy overestimation of centrosomes, but will result in the missing of, by eye, clearly visible centrosomes while still finding centrosomes where none should exist. The red arrows point out the errors caused by the local threshold.

The gradually increasing L-r graph could therefore be explained with an overestimation of the amount of centrosomes that were found in the image, which would result in a high I value in equation 3.2 and thus a high L-r value. In addition, for the Hoechst signal the local threshold was sometimes unable to separate nuclei that were very close together, which would lead to the underestimation and wrong distribution of nuclei. These effects were most apparent in the tenth position, see figure 5.2

5.3 Future perspectives 35

Figure 5.2:The binary image created by the local threshold on the Hoechst signal. The red dots are the found objects, the yellow is where the original image has higher pixel values than the thresholded image, the purple is were the original image had lower pixel values than the thresholded image. The yellow within the colony should be separate nuclei, but they form prolonged objects resulting in an

underestimation of the amount of nuclei.

A case could therefore be made to exclude the tenth position from the average since the thresh-old is so bad there. Most of the issues with the local threshthresh-old could be solved by, for example, using a mask in conjunction with fore- and background seeds, yet, due to time constraints, it was not possible to implement these solutions.

5.3

Future perspectives

Considering all the aforementioned limitations of the method, the reliability of the Ripley’s K analysis is called into question and further analysis with better object detection is necessary to make any conclusive remarks about a preferred distance between centrosome pairs or their (possibly) correlated movement over time. It should be noted, however, that seemingly con-nected centrosomes were observed on multiple, three, occasions, see supplementary figure 6.7 for the other occasions. It is therefore possible that there is indeed a connection between cen-trosomes, but that this connection only rarely forms, exists for a short period of time or is not visible in the data due to the aforementioned problems with the method. Furthermore, there are several factors which were not taken into consideration in the analysis which should or could be investigated. First, the effects of a sparse plate compared to a confluent plate, since behaviour of single cells could be different from behaviour of cells within a colony. Second, the effect that the two different ESC media have on the observed phenomena could be investigated, since using 2i medium will bring the cells back to a naive ground state where they exhibit greater pluripotent gene expression than when Serum with LIF is used.

Chapter

6

Supplementary Information

6.1

Supplementary Figures

(a) _(b)

Figure 6.1:A blunt cut (a) will only leave the sides of the nucleotides open, creating a nonspecific binding site. An overhang (b) will leave several nucleotides open from above or below creating a very specific bonding site.

38 Supplementary Information

Figure 6.2:The plasmid used to create the fluorescent γ-tubulin. The CMV promoter of this plasmid is replaced by cutting the plasmid open at the highlighted SnaBI and BmtI sites. Furhthermore, the bracket indicates where the DNA coding for γ-tubulin is located.

6.1 Supplementary Figures 39

Figure 6.3:[20]The three steps of PCR: 1. The double stranded DNA is heated so that they separate. 2. An oligonucleotide is put on the beginning of the desired DNA region on one strand and another is put on the end of the desired DNA region on the other strand. 3. A new DNA strand is synthesized by

polymerase, starting at the oligonucleotide. These three steps are repeated 20 to 30 times to create a lot of duplicate DNA.

40 Supplementary Information

(a)

(b)

(c)

(d)

Figure 6.4:The before (a,b) and after (c,d) images of the second timelapse (Serum with LIF, Intensity = 15 %, Exposure time = 300 ms), with both the centrosome signal (a,c) and the brightfield image (b,d). In image d the death of the highest cell is visible, since small bubbles within the cell can be seen.

6.1 Supplementary Figures 41

(a) (b)

(c)

(d)

Figure 6.5:The before (a,b) and after (c,d) images of the sixth timelapse (clone 4, 2i, Intensity = 3 %, Exposure time = 100 ms), with both the centrosome signal (a,c) and the brightfield image (b,d).

42 Supplementary Information

(a)

(b)

(c) (d)

Figure 6.6:The before (a,b) and after (c,d) images of the seventh timelapse (clone 5, 2i, Intensity = 3 %, Exposure time = 100 ms), with both the centrosome signal (a,c) and the brightfield image (b,d).

6.1 Supplementary Figures 43

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k) (l)

Figure 6.7:Two centrosomes that moved synchronous in the sixth time-lapse (clone 4, 2i, Intensity = 3%, Exposure time = 100 ms). a-f is the centrosome signal, g-i the brightfield image.

44 Supplementary Information

(a) (b) (c)

(d) (e) (f)

Figure 6.8:The brightfield images of figure 4.7. a) The two centrosomes have been moving away from each other to pull the chromosomes into the daughter cell. b) Cytokinesis cleaves the mother cell into two daughter cells each with one centrosome. c) After the two daughter cells have been separated from each other, the centrosomes move back to the line of division. d-f)After having moved up and down on the line of division the centrosomes move with similar trajectories and speed away from the line of division.

6.2

Appendix 1

t h r e s h o l d l o c a l ( image , b l o c k s i z e , o f f s e t = C)

Here, image is the input image, blocksize is the size of a pixel neighborhood that is used to cal-culate the threshold value of the pixel and offset is a constant that is added to the threshold value. Ripley’s K function

R i p l y e s K E s t i m a t o r ( a r e a = A, xmax = sx , ymax = sy , xmin = 1 , ymin = 1 ) Here, A is the size of the study area, xmax and ymax are the maximum x and y coordinates of the study area and xmin, ymin are the minimum x and y coordinates of the study area.

6.2 Appendix 1 45

Here, the input data is data, radii are the radii for which Ripley’s K needs to be calculated and mode is the method of estimating the edge effects.

Python program to measure degradation of the signal:

# T h i s c a l c u l a t e s t h e d i s t a n c e w i t h t h e p r e v i o u s p o i n t def C a l c D i s t a n c e ( x1 , y1 , x0 , y0 ) : dx = x1−x0 dy = y1−y0 d i s t a n c e = np . s q r t ( dx **2+dy * * 2 ) r e t u r n d i s t a n c e # C a l c u l a t e s t h e p o s i t i o n o f t h e h i g h e s t v a l u e def P o s i t i o n ( x ) : i f x < 1 0 : p i c t u r e = fname + pre1 + s t r ( x )+ t i f e l s e: p i c t u r e = fname + pre2 + s t r ( x )+ t i f img=mpimg . imread ( p i c t u r e ) f l a t = np . ndarray . f l a t t e n ( img ) s o r t i n g = np . s o r t ( f l a t ) b i g = np . amax ( img )

indexmax = np . argmax ( img ) x1 = indexmax%1024 y1 = indexmax //1024 r e t u r n ( x1 , y1 , big , s o r t i n g , f l a t , img ) # T h e s e two r e m o v e v a l u e s f r o m f i b e r s w h i ch a r e c l o s e enough t o t h e c e n t r o s o m e def t o o b i g ( b i g ) : i f b i g > 1 0 0 0 : b i g = 750 r e t u r n b i g def twobig ( v a l ) : i f v a l > 5 0 0 0 : v a l = 500 r e t u r n v a l #To c a l c u l a t e i n t e n s i t y , p o s i t i o n and d i s t a n c e p e r s t e p # I n i t i a l i z e a l l t h e v a r i a b l e s t i =64 fname = ’ i n s e r t f i l e n a m e ’ t i f = ’ . t i f ’ pre1 = ’ 000 ’ pre2 = ’ 00 ’ x0 = 0 y0 = 0 x = 0 indexmax = 0 i n t e r v a l = 15

46 Supplementary Information t o t t i m e = ( t i ) * i n t e r v a l t = np . arange ( t i ) maxi = [ ] x t o t = [ ] y t o t = [ ] d i s t a n c e t o t = [ ] time = np . arange ( 0 , t o t t i m e , i n t e r v a l ) t 2 = [ z / 60 f o r z in time ] # Loop t h r o u g h a l l i m a g e s f o r x in range ( 1 , t i ) : pos = P o s i t i o n ( x ) x1 = pos [ 0 ] y1 = pos [ 1 ] b i g g e s t = pos [ 2 ] s o r t = pos [ 3 ] l e n g t h = pos [ 4 ] img = pos [ 5 ] d i s t a n c e = C a l c D i s t a n c e ( x1 , y1 , x0 , y0 ) #To make t h e f i r s t h i g h e s t v a l u e t h e s t a r t i n g p o i n t i f x == 1 : maxi . append ( b i g g e s t ) d i s t a n c e = 0 d i s t a n c e t o t . append ( d i s t a n c e ) # I f t h e d i s t a n c e i s t o o b i g i t w i l l s e a r c h f o r t h e n e x t # most i n t e n s e p i x e l w h ic h i s c l o s e enough i f d i s t a n c e > 1 5 0 : f o r i in range ( len ( l e n g t h )−1 , 0 , −1): value = s o r t [ i ]

r e s u l t = np . where ( img == value ) x1 = r e s u l t [ 1 ] [ 0 ]

y1 = r e s u l t [ 0 ] [ 0 ]

d i s t a n c e = C a l c D i s t a n c e ( x1 , y1 , x0 , y0 ) i f d i s t a n c e < 1 5 0 :

# v a l u e = t w o b i g ( v a l u e ) maxi . append ( value ) x t o t . append ( x1 ) y t o t . append ( y1 ) d i s t a n c e t o t . append ( d i s t a n c e ) break e l s e: # b i g g e s t = t o o b i g ( b i g g e s t ) x t o t . append ( x1 )

6.2 Appendix 1 47 y t o t . append ( y1 ) maxi . append ( b i g g e s t ) d i s t a n c e t o t . append ( d i s t a n c e ) x0 = x1 y0 = y1 # p r i n t ( x ) # P l o t t i n g a l l t h e g r a p h s x l e n = np . s i z e ( img , 1 ) ylen = np . s i z e ( img , 0 ) f i g , ( ax1 ) = p l t . s u b p l o t s ( nrows =1 , n c o l s =1 , f i g s i z e = ( 1 0 , 6 ) ) ax1 . s c a t t e r ( t2 , maxi ) ax1 . s e t t i t l e ( t i t l e , f o n t s i z e =16)

ax1 . s e t x l a b e l ( ’ Time ( Hours ) ’ , f o n t s i z e = 1 7 )

ax1 . s e t y l a b e l ( ’ I n t e n s i t y ( Grey value ) ’ , f o n t s i z e = 1 7 ) p l t . t i c k p a r a m s ( l a b e l s i z e =15 p l t . show ( ) # I n i t i a t e v a r i a b l e s pos = 1 c e n t r o = 1 fname = ’ I n s e r t f i l e n a m e ’ save = t i t l e + ’ . png ’ csv = ’ . csv ’ a l l v a l = [ ]

my data = genfromtxt ( fname , d e l i m i t e r = ’ , ’ ) s i z e = np . s i z e ( my data , 0 ) # E x t r a c t i n t e n s i t y v a l u e s f o r m c s v f i l e f o r j in range ( 1 , s i z e ) : value = my data [ j ] [ 2 ] a l l v a l . append ( value ) # C r e a t e x−a x i s t o t t i m e = len ( a l l v a l ) * 1 5 time = np . arange ( 0 , t o t t i m e , 1 5 ) t 2 = [ z / 60 f o r z in time ] # P l o t g r a p h f i g , ax1 = p l t . s u b p l o t s ( nrows =1 , n c o l s =1 , f i g s i z e = ( 1 0 , 6 ) ) ax1 . s c a t t e r ( t2 , a l l v a l ) ax1 . s e t t i t l e ( t i t l e )

ax1 . s e t x l a b e l ( ’ Time ( Hours ) ’ , f o n t s i z e = 1 8 )

ax1 . s e t y l a b e l ( ’ I n t e n s i t y ( P i x e l Value ) ’ , f o n t s i z e = 1 8 ) p l t . t i c k p a r a m s ( l a b e l s i z e =15)

p l t . show ( )

The python program used to calculate the Nearest Neighbor distance and Ripley’s K # R i p l e y ’ s K