Impact of Network Regulation on Migrating Cells

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation

vorgelegt von

Anne Pora, Ingénieur, Master

aus Rueil-Malmaison, Frankreich

Berichter: Univ.-Prof. Dr. Björn Kampa Univ.-Prof. Dr. med. Rudolf Leube

Tag der mündlichen Prüfung: 02.04.19

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek verfügbar. This work was performed at the Institute for Molecular and Cellular Anatomy at University Hospital RWTH Aachen by the mentorship of Prof. Dr. med. Rudolf E. Leube. It was exclusively performed by myself, unless otherwise stated in the text.

1. Reviewer: Univ.-Prof. Dr. Björn Kampa 2. Reviewer: Univ.-Prof. Dr. med. Rudolf E. Leube

Toulouse (FR), 30.11.18

2

Table of Contents

Table of Contents 3

Chapter 1: Introduction 6

1. Cell migration 6 A key process in physiological and pathological conditions 6 Different kinds of migration 6 Influence of the environment 7 2. The : a key player in cell migration 9 The cytoskeleton 9 and focal adhesions 9 12 Keratin intermediate filaments and 13 Cross-talk between keratin intermediate filaments and others 22 cytoskeletal components

Imaging cytoskeletal dynamics in migrating cells 24 3. Objectives 26

Chapter 2: Materials and Methods 27

1. Cell culture conditions 27 2. Keratin extraction and immunoblotting 28 3. Immunofluorescence 32 4. Plasmid constructs and DNA transfection into cultured cells 34 5. Drug treatments 35 6. Micropatterning 35 7. Preparation of elastic substrates 36 8. Imaging conditions 36

3

9. Image analysis 37 10. Statistical analysis 43

Chapter 3: Results 44

A. Influence of the mechanophysical environment on keratin 44 network dynamics in migrating cells

1. K5-YFP is a reliable reporter to measure keratin dynamics in migrating 44 normal human epidermal 2. The keratin flow pattern in migrating keratinocytes differs between the 47 cell front, center, and back with respect to speed and direction of keratin flow 3. High keratin flow correlates with high migration speed 50 4. The keratin flow pattern mirrors the trajectory of cell migration 52 5. Increased ECM coating density leads to decreased keratin flow 56 6. Decreased substrate stiffness increases keratin flow 59 7. Confinement of migrating normal human epidermal keratinocytes 62 reduces keratin flow 8. only mildly affect the mechanophysical-dependent regulation of 64 migration 9. Keratin flow lags behind actin flow in migrating keratinocytes 67

B. Hemidesmosomes and focal adhesions form treadmilling 72 arrays in migrating primary keratinocytes

1. Hemidesmosomal cluster in chevron-shaped arrays in migrating 72 primary human keratinocytes

2. Hemidesmosomal chevron arrays and focal adhesion sites are spatially 75 linked but segregated

3. Focal adhesion-decorated hemidesmosomal chevrons are formed at the 78 cell front and are removed in the cell rear in migrating keratinocytes

4

4. Hemidesmosomal chevron patterns form during cell adhesion and 80 spreading

5. Focal adhesions and hemidesmosomes affect each other's distribution 83 during chevron pattern formation

Chapter 4: Discussion 90

1. Keratin flow pattern during migration 90 2. Impact of the mechanophysical environment 92 3. Actin as an upstream regulator of keratin dynamics 93 4. Highly ordered chevron-like hemidesmosomal structures 95 5. Cross-talk between hemidesmosomes and focal adhesions 96 6. Cell-matrix adhesions as key players in actin-keratin cross-talks 98 7. Conclusion and future work 100 What we learn about the complexity of cytoskeletal cross-talk 100 during cell migration Future work 103

Summary 106 References 108 List of Figures 118 List of Tables 120 List of Movies 121 List of Abbreviations 123 Acknowledgements 125 Curriculum Vitae 127 Bibliography 129

5

Chapter 1: Introduction

1. Cell migration

A key process in physiological and pathological conditions

Cell migration is a highly complex process that is crucial in various processes, either physiological or pathological. Physiological processes include organ development during embryogenesis, maintenance of tissue homeostasis at every stage of life, as well as movement of immune cells in the body. Pathological processes include wound-healing but also migration of cancer cells away from the initial tumor to form metastases (Doyle et al., 2013).

Different kinds of migration

Among animal cells, migration behaviors are very diverse. Two main modes of migration are described in the literature for single cells: mesenchymal and amoeboid migration. Mesenchymal migration is characterized by a low speed, cells with irregular shapes, strong cell-matrix adhesions, and the development of flat protrusions at the leading-edge called lamellipodium and filipodium (Friedl and Wolf, 2010; Welch et al., 2015). Amoeboid migration is characterized by a higher speed, cells with round shapes, weak cell-matrix adhesions, and the formation of blebs (Gardel et al., 2010). Depending on the composition of the blebs (pseudopods or stable blebs), amoeboid migration can be subdivided into two sub-kinds of migration (Lämmermann and Sixt, 2009). Mesenchymal migration is typically adopted by fibroblasts and keratinocytes, while amoeboid migration is adopted by dendritic cells. Nevertheless, the cell behavior is very plastic, so that depending on the environment, cells can switch from one mode to another mode of migration. Typically, upon confinement fibroblasts can change in favor of amoeboid migration, in a process called mesenchymal-to-amoeboid migration (Friedl and Wolf, 2010).

6

This study focuses on epithelial cell migration. In the next parts, I will only focus on migration in the mesenchymal mode.

Influence of the environment

Migration is highly dependent on the cellular environment: cells are subjected to a high variety of biochemical signals, but also to different mechanical properties of their surroundings (Fig. 1.1) (Doyle et al., 2013). These two types of signal must be perceived and integrated for cellular responses. The ability of cells to sense the mechanical characteristics of their environment is called mechanosensation, while their ability to respond to it is called mechanotransduction (Bukoreshtliev et al., 2013). Relevant cell components and mechanisms for such processes are discussed in the next parts. Additionally, cells not only respond to their environment but also modify their environment; this bi-directional cross-talk is called dynamic reciprocity (Helvert et al., 2018).

The most relevant physical parameters to consider in the physiological environment for cells are the dimensionality and degree of confinement, the topography, and the extracellular matrix (ECM) composition, density and stiffness (Charras and Sahai, 2014). The way cells respond to these different stimuli is highly context dependent. Experimentally, when preparing a given matrix, changing one parameter might well affect several others. For example, the composition of 3D gels affects their elastics response: fibronectin fibers are covalently linked to each other so that they form a linear elastic gel, whereas fibers can move independently so that collagen gels do not have a linear elastic behavior (Pedersen et al., 2005).

In terms of dimensionality, 2D planar substrates are the most common geometry used in cell culture, even though they are poorly representative of physiological conditions encountered in the human body where cells are confronted with a 3D environment. Cells are mostly confined in physiological conditions, and 1D models (e.g., cells migrating on

7 fibers) can be representative of such situations. 2D geometries promote cell spreading while 1D and 3D models inhibit cell spreading due to their fibrillar or porous topography (Doyle et al., 2013). Confinement promotes blebbing when friction is enough to generate motion without the need of cell-matrix adhesions (Welch et al., 2015).

Figure 1.1: Influence of the environment on cell migration. Scheme depicting environmental factors affecting intracellular regulators and ultimately determining the migratory of cells. Adapted from Doyle et al. (2013).

ECM-ligand interactions also strongly affect cell migration. Different ECM proteins potentially interact with different ligands. The coating density as well as the substrates stiffness affect migration speed in a biphasic manner in 2D. At very low density, cell- matrix adhesions are hardly formed, while at very high density, cells fail at retracting in the rear (Palecek et al., 1997; Palecek et al., 1998). At fixed ECM concentration, migration speed is reduced on stiffer substrates, except for very soft substrates (i.e. in the sub-kPa range where migration speed is reduced) (Lo et al., 2000; Peyton and Putnam, 2005; Yeung et al., 2005; Zhong and Ji, 2013). Adjusting both parameters in parallel allows modulation of the migration speed.

The environment acts as an extracellular regulator of cell migration. At the intracellular level, these signals are integrated by a highly complex system called the cytoskeleton.

8

2. The cytoskeleton: a key player in cell migration

The cytoskeleton

In eukaryotic cells, a network of filaments extending throughout the cytoplasm is found and called the cytoskeleton. It has various roles including defining the cell shape, organizing the cytoplasm, maintenance of mechanical stability, and regulation of a wide range of signaling processes. With the help of associated motor proteins, the cytoskeleton is involved in cell division, differentiation, cell migration and organelle trafficking.

Contrary to what its name suggests, the cytoskeleton is a highly dynamic structure that undergoes continuous reorganization due to a very high number of regulators. It is composed of three main classes of filaments: actin filaments, the intermediate filaments, and microtubules. The structure and function of these networks, especially regarding cell migration, are detailed in the next parts.

Actin and focal adhesions

Actin filaments, also called , have a diameter of about 6 nm. They can be found in the cytoplasm of all eukaryotic cells. On distinguishes filamentous (F)-actin and monomeric (G)-actin. Actin filaments are polar, form a two-stranded helix, and grow by monomer addition preferentially at the plus end (Carlier and Pantaloni, 2007). They can also bundle to form thicker bundles formed of 10 to 30 actin filaments in which α- acts as a cross-linker (Tojkander et al., 2012).

The two main regulators of actin polymerization are Arp2/3 and formins. The Arp2/3 complex induces the branching of actin filaments with an angle of 70° from the existing filament. It is regulated by the small GTPases Rac (Weaver et al., 2003), which plays

9 therefore a key role in lamellipodia formation and membrane ruffling. Whereas formins induce the growth of actin filaments at the barbed end in a linear manner and are regulated by Cdc42 (Sagot et al., 2002; Watanabe and Higashida, 2004), which plays a key role in filipodia formation (Fig. 1.2). Still, the assembly and disassembly of actin filaments within the cytoplasm is regulated by a wide range of actin-binding proteins such as the capping (Zigmond, 2004) or the severing protein cofilin (Maciver and Hussey, 2002). By associating either with filaments or with monomers, they can shift the balance toward extension, branching, bundling, or severing of filaments, making the actin network a very highly dynamic system (Carlier and Pantaloni, 2007).

Actin filaments are associated with motor proteins to generate contractility in the cell. This process is regulated by members of the RhoGTPase family. The activation of RhoA and Cdc42 triggers the activation of ROCK, and consequently promotes cell contractility. The activation of Rac triggers the activation of PAK and consequently promotes cell spreading (Jaffe and Hall, 2005; Vicente-Manzanares et al., 2005).

Actin filaments associate with cell-matrix adhesions called focal adhesions (FA). Focal adhesions are composed of α3, α2, α5 and β1 , as well as a large number of signalling proteins including paxillin, , , α-actinin, kindlin, and focal adhesion kinase (Burridge, 2017) (Fig. 1.3). The existence of focal adhesions in vivo is still questioned even though they are very commonly observed in cells in culture (Underwood et al., 2009). Focal adhesions are very dynamic structures. After their formation, their composition changes upon maturation or they can be disassembled in a Rho GTPase- dependent process (Geiger and Yamada, 2011; Vicente-Manzanares et al., 2009). During maturation, their size increases upon phosphorylation due to higher tension being applied on them (Bershadsky et al., 1996; Riveline et al., 2001). Their highly complex composition allows them to be a hub for various signaling pathways, receiving signals both from the inside and the outside of the cell (Zaidel-Bar et al., 2007; Zaidel-Bar and Geiger, 2010). Considered to be the main points of force generation throughout the acto- myosin system (Geiger et al., 2001), focal adhesions are also key players in environment

10 sensing and, especially, with probing the mechanical characteristics of the cellular environment (Geiger and Yamada, 2011).

Figure 1.2: Elongation of actin filaments. Scheme depicting the 2 possible modes of elongation of actin filaments. Formin is a dimeric complex involved in the linear elongation of actin filaments at the plus end. It binds the most exposed subunit of the double helix to allow the assembly of a new subunit on the other helix. The Arp2/3 complex is involved in the branching of actin filaments at the barbed-end. Upon activation by their activating factor, Arp2 and Arp3 bind to form a complex resembling the plus end of actin filaments. This complex binds on the side of an existing actin filament and allows

11 the growth of a new branch of actin with an angle of 70°. Adapted from Alberts et al. (2012).

Figure 1.3: Focal adhesions composition. Scheme depicting the simplified composition of focal adhesions (cell-matrix adhesions that anchor actin filaments). Adapted from Deakin et al. (2008); Backert et al. (2013).

Microtubules

Microtubules have a diameter of 24-25 nm and form a polar hollow rod-like structure. Microtubules are composed of non-covalently bound dimers consisting of one α- tubulin and one β-tubulin (Weisenberg et al., 1968). The tubulin dimers form the walls of the hollow microtubules constituting 13 parallel protofilaments (Fig. 1.4) (Evans et al., 1985; Mandelkow et al., 1986). Addition of new dimers is faster at the plus end (side with a β-tubulin) than at the minus end (side with an α-tubulin). Such balance favors the

12 growth of microtubules from an organizing center such as the centrosome. The rapid switch from growth to microtubule disassembly in cells is known as dynamic instability (Akhmanova and Setinmetz, 2008; Gardner et al., 2008).

Figure 1.4: Microtubules composition and mode of assembly. The scheme depicts the structure and composition of microtubules. They are hollow tubes that are made up of 13 parallel protofilaments, which consist of α-/β-tubulin heterodimers. Dimers incorporate preferentially at the plus end to elongate microtubules. Adapted from Moch (2015); Conde and Caceres (2009); Litjens et al. (2006).

During cell division, microtubules play a crucial role since they form the mitotic spindle that allows segregation. Microtubules are involved in cell organization by positioning the nucleus, organelles like the Golgi apparatus and the microtubules organizing center (MTOC). Reorientation of the two latter toward the leading edge of migrating cells serves the maintenance of cell polarity. This is why microtubules are key players in persistent and directed migration. Microtubules are also tracks for motor proteins, with moving toward the plus end and moving towards the minus end (Etienne-Manneville, 2013).

Keratin intermediate filaments and hemidesmosomes

The name comes from their diameter of 8 nm to 12 nm, which is intermediate to that of actin filaments and microtubules. In mammals, around 70

13 different kinds of intermediate polypeptides exist; which are expressed in a tissue-specific and differentiation-specific manner (Table 1.1) (Hesse et al., 2001; Kim and Coulombe, 2007).

Despite this wide variety of intermediate filament polypeptides, they all share the same overall structure: each is composed of a long α-helical rod domain composed of 300-350 residues, flanked by highly variable head (amino-terminus) and tail (carboxy-terminus) domains (Steinert et al., 1983) (Fig. 1.5).

Table 1.1: Intermediate filaments nomenclature. Adapted from Leduc and Etienne- Manneville (2015).

Intermediate Filament Proteins Main Tissue Distribution Type I Keratins II Keratins Epithelium Mesenchyme Glial Fibrillary Acidic Protein Glia III (GFAP) Neurons Muscle Muscle IV Neurons α- Neurons V Ubiquitous (nuclear) Neurons, Astrocytes, Muscle VI Neural Stem Cells, Muscle, … Endothelium

14

During filament assembly (Fig. 1.5), two monomers assemble in a parallel manner to form a polar coil-coil dimer. Then, two such dimers assemble in an antiparallel manner to form an apolar tetramer. Eight tetramers can associate laterally to form a unit-length filament (ULF). Finally, the longitudinal association of ULFs allows the growth of a mature filament (Leube and Schwarz, 2016). Consequently, and contrary to actin filaments and microtubules, intermediate filaments are apolar filaments. Eventually, several intermediate filaments can associate laterally to form thicker bundles. Depending on the cell type, one filament can be composed of different types of intermediate filament polypeptides (e.g., in astrocytes vimentin, GFAP and nestin can be found along one filament (Leduc and Etienne-Manneville, 2017)).

Figure 1.5: Organization and mode of assembly of intermediate filaments. Scheme depicting the successive steps leading to the assembly of intermediate filaments. Each

15 monomer is composed of a rod domain flanked by a head domain and a tail domain. Parallel association of two monomers leads to the formation of a polar parallel dimer. In the case of keratin, association of one type I and one type II monomer is compulsory for the formation of a coiled-coil dimer. Antiparallel association of two dimers allows to form an apolar tetramer. Unit-length filaments are formed by lateral association of eight tetramers. Finally, filaments grow by longitudinal association of ULFs. Adapted from Leube and Schwarz (2016).

Many different post-translational modifications can be applied to intermediate filaments, such as phosphorylation, O-glycolysation, sumoylation, ubiquitination and acetylation. These modifications modulate the organization, properties and functions of intermediate filaments (Chung et al., 2013).

The diversity among intermediate filaments due to the variability in proteins, variants, and post-translational modifications, make them a complex system to study (Leduc and Etienne-Manneville, 2015).

This work deals with epithelial cell migration. Keratins are the main type of intermediate filaments expressed in the epithelium, so I will focus on them. They are expressed both in simple epithelium such as the intestine, or stratified epithelium such as the skin (Moll et al., 1982; Pekny and Lane, 2007). Nevertheless, like every other cell with a nucleus, epithelial cells express lamins. Additionally, in case of transformation into cancer cells, epithelial cells undergo epithelial-mesenchymal transition (EMT), which induces the expression of vimentin (Thiery et al., 2009; Nakamura and Tokura, 2011).

Keratins belong to two large multigene families, with about 50 coding for different keratin isoforms. Every keratin dimer contains one type I (acidic) and one type II (basic) monomer, so that a keratin filament always contains equal amounts of type I and type II keratins (Franke et al., 1983; Hatzfeld and Franke, 1985). In epithelial cells, keratin intermediate filaments are found in the entire cytoplasm, either in their filamentous (insoluble) or in their soluble form. Within the , the subset of keratin isoforms expressed in a cell depends on the cell layer, functional state, and the environment (Moll

16 et al., 2008). The mammalian skin is composed of several layers with increasing differentiation from bottom to top, i.e., the , the stratum spinosum, the stratum granulosum, and the stratum corneum (Fig. 1.6) (Radtke and Raj, 2003). In this configuration, the extracellular matrix is the : a thin layer that contains no cells and separates the epithelium for the connective tissue. The basal layer, which is mitotically active, is attached to the basement membrane. The basement membrane forms a network of various large molecules including fibronectin, collagen, , proteoglycans, , and hyaluronic acids (Hynes, 2009). Non- differentiated cells in the basal layer express mostly keratins 5 and 14; keratins 1 and 10 are expressed upon differentiation, and keratins 6, 16 and 17 are expressed in activated keratinocytes, i.e. typically during skin injury (Moll et al., 2008).

Figure 1.6: Main keratin isoforms expressed in the skin. Representation of the various layers in human skin and the main keratin isoforms expressed in these layers. The basal layer of the epidermis contains highly proliferative cells that express K5/K14 as the major keratin pair. In the suprabasal layers, cells undergo differentiation and switch their keratin expression in favour of the K1/K10 pair. Adapted from Radtke and Raj (2003).

17

Keratin intermediate filaments were shown to be highly dynamic in sessile cells, undergoing a spatially well-defined cycle of assembly and disassembly. Indeed, at the cell periphery, filaments are nucleated, then while moving toward the cell center, they elongate, integrate to the keratin network, bundle, and eventually either integrate into the nuclear cage or disassemble and diffuse back toward the cell periphery (Windoffer et al., 2011; Moch et al., 2013) (Fig. 1.7). The dynamics of keratin intermediate filaments in migrating cells remains unknown.

Figure 1.7: Keratin turnover cycle in a single sessile cell. The scheme depicts the spatially well-defined cycle of assembly and disassembly of keratin intermediate filaments in a sessile cell. At the cell periphery, filaments nucleate and elongate as they are transported toward the cell center. Then, they are integrated into the keratin network, and eventually bundle to form thicker filaments. Ultimately, either they are integrated into the highly stable nuclear cage, or they are disassembled and diffuse back toward the periphery of the cell. Adapted from Windoffer et al. (2011).

Keratins were initially shown to play a key role in the mechanical integrity of tissues and in stress protection. Keratin intermediate filaments are the main contributors to the mechanical resilience of cells (Ramms et al., 2013), thanks to their low bending stiffness, ability to be stretched as far as 300 % without rupturing, and strain-stiffening at large deformations (Kreplak et al., 2005; Sivaramakrishnan et al., 2008). Genetic in the genes coding for K5 or K14 are responsible for the disease simplex (EBS) (Alberts and Fuchs, 1993). At cellular level, the formation of keratin

18 aggregates has been observed. which are further increased upon mechanical stress (Chan et al., 1993). Macroscopically, patients suffer from skin blisters in response to minor physical traumas (Coulombe et al., 1991). Also, keratins carry numerous phosphorylation sites that allow the buffering of an excessive increase in kinase activity upon stress- response (Ku and Omary, 2006); this is why keratins are commonly called the “phosphate sponge”. Keratin cycling was shown to support rapid cell shape changes to adapt to changing environmental requirements (Leube et al., 2011; Windoffer et al., 2011).

However, the role of keratin intermediate filaments in cell migration remains unclear. For each type of intermediate filament, its role can be modulated by diverse factors like the level of expression, the associated proteins, the intracellular organization, and the post- translational modifications. Also, intermediate filaments may affect migration due to their impact on the mechanical properties of the cells, on the cytoplasmic organization, or on signaling pathways.

The effect of the expression of a given keratin isoform on migration depends on the cell type and the environment (Chung et al., 2013; Leduc and Etienne-Manneville, 2015). Typically, the knockdown of K8/K18 reduces the invasion capacities in squamous carcinoma cells (Alam et al., 2011), but it increases collective cell migration in epithelial cancer cells (Fortier et al., 2013). The expression of different isoforms of keratin alters the mechanical properties of cells and therefore their ability to migrate. Wong and Coulombe (2003) suggested that expression of keratins 6/16 in a context of wound healing corresponds to a compromise between cellular pliability (best achieved with keratins 5/14) and mechanical resilience (best achieved with keratins 1/10), in order to allow migration, even at a lower speed in the harshness of the wound sites. The depletion of the entire type II keratin cluster in mouse keratinocytes, making it impossible for the cells to assemble a keratin network, induces an increase in migration speed and invasive capacities, but alters the ability of the cells to sustain mechanical constraints and to migrate in persistent way (Seltmann et al., 2013; Ramms et al., 2013).

19

Even though motor proteins cannot move in a defined direction on keratin filaments, and only very few cross-linkers of keratins with others cytoskeletal components are known, these results suggest that the role of keratin intermediate filaments in keratinocytes is not limited to sustaining the mechanical integrity of the epithelium, but that they also contribute to the regulation of cell migration.

Keratin intermediate filaments associate with cell-cell connections called in confluent layers (Delva et al., 2009), and with cell-matrix adhesions called hemidesmosomes (HD) (Walko et al., 2015).

Hemidesmosomes are composed of 3 transmembrane proteins: α6 and β4 integrins and BP-180, two intracellular proteins: BP-230 and , and a member of the tetraspanin family: CD151 (Borradori and Sonnenberg, 1996; Green and Jones, 1996) (Fig. 1.8).

Hemidesmosomes are also responsible for the mechanical integrity of tissues since they mediate the attachment of cells of the basal layer of the epidermis to the basement membrane throughout the ligand called laminin-332 (also previously referred to as laminin-5). They have major adhesive functions in the epithelium as shown by the effect of a knockout and by associated diseases showing a wrong adhesion of the epidermis (e.g., junctional epidermolysis bullosa associated to mutations is laminin-332, BP180 or α6/β4 ) (Fine et al., 2008). Hemidesmosomes are observed in vivo, but bona fide Hemidesmosomes are usually not observed in vitro. Instead, some hemidesmosomes- enriched protein complexes, commonly named stable anchoring complexes (SACs) are observed (Carter et al., 1990).

It was initially thought that hemidesmosomes were inhibiting migration by stabilizing the attachment of cells to the extracellular matrix and had to be disassembled to allow the cell to migrate (Carter et al., 1990; Hopkinson et al., 2014). However, growing evidences show that hemidesmosomes are also highly dynamic structures during cell migration (Tsuruta et al., 2003). Plectin-deficient keratinocytes as well as β4 integrin-deficient

20 keratinocytes display increased migratory properties (Andra et al., 1998; Raymond et al., 2005). BPAG1 knockout cells were shown to have normal hemidesmosomes density despite their absence of attachment to keratin intermediate filaments. However, they show a delay to cover wounds (Guo et al., 1995). Keratin knockout cells show disorganised hemidesmosomes, which correlates with faster adhesion to the substrate and increased cell migration (Seltmann et al., 2013). These results suggest that hemidesmosomes play a very complex role in cell migration.

Figure 1.8: Hemidesmosomes composition. Scheme depicting the composition of hemidesmosomes (cell-matrix adhesions that anchor keratin intermediate filaments). Adapted from Moch (2015); Borradori and Sonnenberg (1999).

21

Cross-talk between keratin intermediate filaments and others cytoskeletal components

The cytoskeleton is at the core of effective locomotion by enabling successive steps of protrusion at the cell front and contraction at the cell rear, in parallel with the regulation of cell-matrix adhesions (Ladoux et al., 2016). Mesenchymal migration is commonly described as a cyclic repetition of the following steps (Vicente-Manzanares et al., 2005). First, a front-rear polarity axis must be established. Then, the membrane protrudes at the leading edge while new adhesions are formed. These adhesions mature into focal adhesions that are connected to stress fibers which are responsible for the contraction of the cell body. Finally, upon disassembly of adhesions at the cell rear, the cell detaches and the rear-end retracts (Fig. 1.9). Such description puts the acto-myosin system and its associated focal adhesions at the center of cell migration. It is allowed by a well-defined distribution of actin flow: retrograde flow in the entire cell body except at the rear-end where upon rip-off of adhesion, the flow is anterograde. Focal adhesions distribution and dynamics of assembly and disassembly in migrating keratinocytes is known to determine actin flow and traction force organisation (Möhl et al., 2012). Actin polymerization at the cell-front drives the formation of protrusions (Parsons et al. 2010). Myosin is responsible for the pulling of actin towards the cell center (Chen et al., 1981; Mitchison and Cramer, 1996).

Nevertheless, this protrusive-contractile behavior of the acto-myosin system is regulated by a very wide range of modulators including the others cytoskeletal components namely keratin intermediate filaments and its associated hemidesmosomes as well as microtubules. Such cross-talk may be based on direct interactions, steric interactions, or indirect interactions typically involving cross-linkers and molecular motors (Chung et al., 2013).

22

Figure 1.9: Cycle of protrusion and retraction during cell migration. The scheme depicts how successive cycles of protrusion at the front and retraction at the back allow a cell to move. The actin cytoskeleton is the central for the model. At the leading edge, actin polymerisation is followed by formation of new cell-matrix adhesions to stabilize the new protrusion. At the rear-end, acto-myosin contractility leads to disassembly of adhesions, and subsequent retraction of the cell back. Adapted from Alberts et al. (2012).

Many cross-talk between vimentin intermediate filaments and microtubules, or between vimentin and focal adhesions have been described, involving various molecules, e.g., integrins, FAK, plectin 1f… (Leduc and Etienne-Manneville, 2015; Burgstaller et al., 2010; Havel et al., 2015). However, cross-talk involving keratin intermediate filaments and other cytoskeletal compounds are less well-known. In sessile cells, keratin dynamics were shown to be dependent on a functional actin and microtubule network (Wöll et al., 2005). Evidence has also been obtained showing that keratin intermediate filaments affect other cytoskeletal components and vice-versa in the context of cell migration (Chung et al., 2013; Leduc and Etienne-Manneville, 2015).

23

Plakins, which contribute to the organization of the cytoskeleton, anchor cytoskeletal filaments at diverse strategic locations in the cells. Thus, plectin anchors keratin intermediate filaments to hemidesmosomes and anchors keratin intermediate filaments to desmosomes (Leung et al., 2001; Delva et al., 2009). Notably, keratin-associated plectin 1c was shown to induce an increase in microtubule dynamics and to affect focal adhesion turnover in keratinocytes (Valencia et al., 2013).

Other evidence points to cross-talk between the keratin- scaffold and the actin-focal adhesions scaffold. knockout in hepatoma cells disturbs the actin organization in a Rho dependent manner and reduces the local stiffness in proximity to focal adhesions; this phenotype is associated with impaired cell migration in scratch wound assays (Bordeleau et al., 2010). Knockdown of the actin-binding protein actinin-4 induces a mislocalization of α6/β4 integrin and BP-230, inducing an alteration in the directionality of migration, in cell polarity, and in the dynamics of the lamellipodium (Hamill et al., 2013). Also, 14-3-3σ forms complexes with soluble , and actin contributing to invasivity of migrating breast tumor cells (Boudreau et al., 2013). Evidence of cross-talk between focal adhesions and hemidesmosomes regulating speed and persistence of cell migration have been obtained by live-cell imaging techniques (Geuijen and Sonnenberg, 2002; Tsuruta et al., 2003; Spinardi et al., 2004; Ozawa et al., 2010; Hiroyasu et al., 2016; Osmani et al., 2018).

Imaging cytoskeletal dynamics in migrating cells

Time-lapse imaging is a powerful tool to capture dynamic biological processes, and most particularly to decipher the spatiotemporal dynamics of cytoskeletal compounds in migrating cells (Dormann and Weijer, 2006). Fluorescence imaging has been extensively used since genetically encoded fluorescent labels have been available (Giepmans et al., 2006). Additionally, the development of confocal microscopes has allowed a substantial increase in spatial resolution as compared to widefield fluorescence microscopy thanks

24 to the removal of the out-of-focus light allowed by the addition of a pinhole (Peterson, 2010).

Nevertheless, these techniques remain challenging for several reasons. One must obtain sufficient spatiotemporal resolution. The imaging frequency must be adapted to the time scale of the observed phenomenon. The pixel size has to be adjusted considering the size and geometry of the observed components. When different components with different fluorophores are observed concomitantly, the dynamic range of the detector must be carefully set to capture the variations of intensities of all studied components. These parameters have an important cost in terms of imaging time and quantity of data. In addition, phototoxicity has to be taken into account, since excitation light is harmful to the cells, especially when using long exposure times and low wavelengths (Hoebe et al., 2007). During long repeated scanning sequences, the ability of fluorophores to emit light may be impaired, this process called photobleaching induces a decrease of the signal to noise ratio (Bernas et al., 2004). Also, fitting the sample into the imaging chamber of the microscope does not necessarily allow all possible geometries. The cell environment may have to be adapted to the constraints of the equipment. Due to the impact the environment has on cell migration (see Chapter 1, 1), one must be careful considering to what extend the imaging process may disturb the cell behavior.

Moreover, the analysis of images of migrating cells is particularly complex. Not only the cytoskeleton moves within the cell, but the entire cell moves inducing changes in the cell shape. Analysis requires the recording of cell shape changes, migration speed and direction, as well as subcellular processes such as filament dynamics and adhesions assembly and disassembly. Comparison between cells is difficult. It requires the calculation of the relative position of cytoskeletal components within the cell. Normalization of the cell shape to a circle was shown to be a promising technique to average results of actin flow and focal adhesions dynamics over numerous cells (Möhl et al., 2012).

25

3. Objectives

First, I aim at understanding how keratin dynamics is integrated in migrating cells. First, I will optimize an experimental system that allows capturing keratin dynamics in a migrating epithelial cell over a sufficient amount of time without inducing excessive phototoxicity or photobleaching, and without disturbing cell migration. Quantitative image analysis routines will be developed to allow mapping of keratin flow in a shape- normalized standard cell and increasing the significance of the results by averaging over numerous cells. The variations of keratin flow depending on migration features such as speed and persistence will be evaluated.

Second, I will alter the mechanophysical properties of the environment and determine how keratin dynamics are affected by it. This will allow to induce perturbations in cellular migration features without biochemical stimulation. I will determine whether keratins play a prominent role in the regulation of migration in different environments. I will obtain insights how these variations of keratin dynamics correlate with changes in the dynamics of other cytoskeletal components.

Third, I will determine how hemidesmosomes organize during migration in the conditions of our study. I will optimize systems so that I can capture in a given cell the dynamics of hemidesmosomes and focal adhesions during migration. I will determine the interdependence of hemidesmosomes and focal adhesions during migration. I will get insight into how hemidesmosomes might be key players in the interplay between the keratin and the actin during cell migration.

Above all, I will determine how keratin-hemidesmosome scaffold dynamics promote smooth migration in keratinocytes.

26

Chapter 2: Materials and Methods

1. Cell culture conditions

Normal human epidermal keratinocytes prepared from neonatal foreskin (nHEKs) (Cell Systems) were grown in Dermalife K Medium Complete without TGFα (Cell Systems) in

-1 the presence of penicillin-streptomycin (100 µg.mL , Invitrogen) at 37°C in a 5% CO2 humidified atmosphere. The cells were purchased as “Passage 0” and used for experiments as “Passage 3” (corresponding to approximatively 10 population doublings). They could be frozen at each passage in Cryo-SFM freezing medium (Promocell) and were thawed directly for experiments. They were passaged using Trypkit (Cell Systems) to trypsinize the cells. For experiments, cells were seeded on human fibronectin-coated (2.5 µg.cm-2 corresponding to a concentration of 17 mg.L-1, or 5 µg.cm-2 corresponding to a concentration of 33 mg.L-1 in a total volume of 1.5 mL of PBS for 30 minutes at 37°C, VWR) glass coverslips (24 mm diameter high precision coverslips #1.5 (Marienfeld) for 3D z- stacks acquisition, or 12 mm diameter coverslips #1.5 (ThermoScientific) for other immunostaining, or in 35 mm diameter glass bottom dishes (MatTek) for live-cell imaging) at a density of 5,000 cells per cm². When seeded on elastic substrates, the human fibronectin coating density was 5 µg.cm-2.

Murine epidermis-derived keratinocytes (either wild-types (WT) or type II keratin knock- out (KO)) established by Kröger et al. (2013) were cultivated in DMEM/Ham’s F12 medium with low calcium (PAA) complemented with the compounds listed in Table 2.1 at 32°C in a 5% CO2 humidified atmosphere. They were grown in flasks coated with bovine collagen I (3.75 µg.cm-2 corresponding to a concentration of 43 ng.mL-1, Corning) and were passaged twice a week at a ratio of 1:3 using trypsin (Biochrom). K5 rescue in keratin-free keratinocytes (RES) was obtained by stably transfecting the K5-YFP construct in the KO cells and selecting a single clone. The cells were kindly provided by Mugdha Sawant (Sawant et al., 2018).

27

Table 2.1: Supplements for DMEM/Ham’s F12 cell culture medium

Compound Final concentration Manufacturer Fetal Calf Serum Gold (Chelex- 10% PAA treated) Glutamate 2 mM Invitrogen Pyruvate 1 mM PAA Penicillin/Streptomycin 100 µg.mL-1 Invitrogen Adenine 0.18 mM Sigma-Aldrich Epidermal growth factor (EGF) 10 ng.mL-1 Invitrogen Insulin 25 µg.mL-1 Sigma-Aldrich Hydroxycortison 0.5 µg.mL-1 Sigma-Aldrich Choleratoxin 100 pM BIO-RAD

2. Keratin extraction and immunoblotting

Keratin enrichment was based on the ability of keratin intermediate filaments to precipitate in high salt buffer using the following protocol. Cells were cultivated in 6 diameter 100 mm dishes until they reached confluence. They were washed with cold PBS (Sigma), then scrapped with 750 µL low salt buffer (LSB, Table 2.2) per dish. Homogenisation of cells (3 periods of 5 seconds with 5 seconds interval) was performed with an Ultra TURRAX T8 mixer (IKA labortechnik), and followed by centrifugation at 5,000 g for 10 minutes at 4°C. The pellet was suspended in 1 mL high salt buffer (HSB, Table 2.2) and incubated on ice for 30 minutes. The homogenate was centrifuged at 1,500 g for 10 minutes at 4°C. The resulting supernatant contains the soluble keratin fraction. The pellet was suspended in 1 mL HSB and incubated on ice for 30 minutes. The homogenate was centrifuged at 1500 g for 10 minutes at 4°C. The resulting pellet was suspended in 500 µL LSB and centrifuged at 1,500 g for 10 minutes at 4°C. After one additional washing

28 of the pellet in distilled water, the final pellet containing the insoluble keratin fraction was suspended in 200 µL of 2 x SDS sample buffer (Table 2.3).

Table 2.2: Composition of the HSB and LSB buffers

Compound Concentration Manufacturer Compounds common to HSB and LSB buffers Tris pH 7.5 10 mM Biomol NaCl 140 mM Carl Roth Ethylenediaminetetraacetic acid 5 mM Sigma-Adrich (EDTA) Phenylmethylsulfonyl fluoride 2 mM Merck KGaA (PMSF) Protease inhibitor (Complete 2 tablets per litre Roche ULTRA tablets Mini EASY Pack) Additional compounds specific for HSB Triton X-100 1 % Sigma-ALdrich Dithiothreitol (DTT) 1 mM Merck KGaA KCl 1.5 mM SERVA Electrophoresis

Table 2.3: Composition of the 5 x SDS sample buffer (based on Laemmli buffer)

Compound Quantity Manufacturer 0.6 M Tris-HCl pH 6.8 + 0.4% 15 mL Life Technologies, (w/v) Sodium Dodecylsulfate SERVA Electrophoresis (SDS) Glycerol 12.5 mL Carl Roth Beta-Mercaptoethanol 2.5 mL Carl Roth SDS 2.5 g SERVA Electrophoresis Bromophenol blue 0.4% (w/v) SERVA Electrophoresis

29

For immunoblotting, sodium dodecyl sulfate - polyacrylamide gel electrophoresis (SDS- PAGE) was performed using SDS-PAGE gels that were prepared in-house with the following composition (Table 2.4).

Table 2.4: Composition of the resolving gel and the stacking gel

Compound Quantity Manufacturer Resolving gel

H2O 6 mL 40% Acrylamide (Acrylamide/Bisacrylamide 3 mL Carl Roth 30%, mixing ratio 29:1) Resolving gel buffer (1.65 M Tris and 6 N HCl pH 3 mL Life technologies 8.8) Tetramethylethylenediamine (TEMED) 12 µL Carl Roth 10% ammonium persulfate (APS) 120 µL Sigma-Aldrich Stacking gel

H2O 1.6 mL 40% Acrylamide 313 µL Carl Roth Stacking gel buffer (1.65 M Tris and 6 N HCl pH 625 µL Life technologies 6.8) TEMED 2.5 µL Carl Roth 10% APS 25 µL Sigma-Aldrich

For SDS-PAGE gel loading, the insoluble fractions obtained from keratin extraction were heated at 100°C for 10 minutes in 2x SDS sample buffer. Then, they were submitted to electrophoresis (100 V for 2 hours in running buffer (Table 2.5)). The ladder for protein molecular weight was ProSieve Quad Color Protein Marker 4.6 – 300 kDa (Biozym).

30

Table 2.5: Composition of the 10 x SDS running buffer

Compound Concentration Manufacturer Tris 250 nM Biomol Glycine 1.921 M Biomol SDS 1% (w/v) SERVA Electrophoresis

Proteins were transferred from the SDS-PAGE gel to a PVDF membrane (Immobilon-P transfer membrane, Merck Millipore) that had been activated in methanol (Carl Roth). Transfer was performed for 1 hour at 100 V in transfer buffer (Table 2.6).

Table 2.6: Composition of the immunoblot transfer buffer Compound Quantity Manufacturer

H2O 800 mL Methanol 200 mL Carl Roth Tris base 14.4 g Biomol Glycine 3.03 g Biomol

Afterwards, membrane blocking was performed for 1 hour at room temperature (RT) in 10% Rotiblock blocking reagent (Carl Roth). Incubation with primary antibodies (Table 2.7) diluted in 10% Rotiblock solution was done overnight at 4°C. Afterwards, the membranes were washed for 3 periods of 5 minutes in Tris-buffered saline-tween-20 solution (TBS-T, Table 2.8). Incubation with the secondary antibodies (Table 2.9) diluted in 10% Rotiblock solution was done for 1 hour at RT. Afterwards, the membranes were washed for 3 periods of 5 minutes in TBS-T.

For imaging, the membranes were treated with AceGlow kit (PEQLAB), mixing equal volumes of stabilized peroxide buffer and luminol enhancer solution, for 3 minutes at RT. Imaging was performed with a Fusion-Solo.WL.4M CCD-camera (Vilber Lourmat) using the Fusion Capt Advance Software (Vilber Lourmat).

31

Table 2.7: List of primary antibodies used for immunoblotting

Antigen Species Designation / Source of Supply / Reference Clone Keratin 5 Guinea Pig GP-CK5 Progen Guinea Pig CK 14.2 Langbein et al. (2001) Guinea Pig K 1.1 Langbein et al. (2005) Guinea Pig K 10.1 Langbein et al. (2006) Keratin 6 Guinea Pig K 6/2.1 Langbein et al. (2005) Guinea Pig K 16.1 Langbein et al. (2005) Keratin 17 Mouse Mo-Ab E3 Moll et al. (1993)

Table 2.8: Composition of TBS-T solution (pH adjusted to 7.6)

Compound Concentration Manufacturer NaCl 130 mM Carl Roth Tris base 50 mM Biomol Tween-20 0.1% Merck KGaA

Table 2.9: List of secondary antibodies used for immunoblotting

Antigen Conjugate Species Designation Source of Supply Guinea Pig HRP Rabbit P0141 Dako Mouse HRP Goat P0441 Dako

3. Immunofluroescence

Three different fixation methods were used for immunofluorescence. Fixation was performed without prior washing steps to minimize cell shape changes.

32

For methanol-acetone fixation, cells were fixed for 2 minutes in methanol (Carl Roth) at -20°C and were permeabilized immediately afterwards in acetone (Carl Roth) at -20°C for 20 s.

For paraformaldehyde (PFA)-acetone fixation, cells were fixed for 10 minutes at RT in 4% (w/v) PFA (Merck) in PBS and were subsequently permeabilized in acetone at -20°C for 30 s.

For PFA-Triton-X fixation, cells were fixed for 10 minutes at RT in 4% (w/v) PFA in PBS and were subsequently permeabilized for 10 minutes at RT in PBS containing 0.2% Triton-X- 100 (Sigma-Aldrich).

For the staining procedure, cells were saturated for 20 minutes at RT in a solution of 5% bovine serum albumin (BSA, SERVA Electrophoresis) in PBS. Incubation with the primary antibodies (Table 2.10) suspended in a solution of 1% BSA in PBS was performed for 1 hour 30 minutes at RT, and followed by three washing steps of 5 minutes in PBS at RT. Incubation with secondary antibodies (Table 2.11) suspended in a solution of 1% BSA in PBS was performed for 35 minutes at RT, and followed by two washing steps of 5 minutes in PBS and one washing step of 10 minutes in water at RT. When Alexa-488 phalloidin (Invitrogen) was used, it was mixed with the primary antibodies. When 4’,6-diamidino-2- phenylindole (DAPI, 2ng.µL-1, Roche) was used, it was mixed with the secondary antibodies. Mounting reagent was Mowiol (Carl Roth).

33

Table 2.10: List of primary antibodies used for immunohistofluorescence Antigen Species Clone / Source of Supply / Reference Designation β4 integrin CD104 Rat 439-9B / 55719 BD Pharmingen α6 integrin CD49f Rat GoH3 / IM0769 Beckman Coulter Pan Mouse PAN-CK / MA5- ThermoFisher 13203 Paxillin Mouse 349 / 610051 BD Biosciences Plectin Guinea Pig Schröder et al. (1999) (Harald Herrmann) BPAG1 (BP230) Mouse 279 / NU-01-BP1 CosmoBio

Table 2.11: List of secondary antibodies used for immunohistofluorescence

Antigen Species Conjugate Ordering Number Source of Supply Rat IgG Goat Alexa-555 A-21434 Invitrogen Rat IgG Donkey DyLight TM 712-476-153 Jackson-Dianova 405 Mouse IgG Goat Alexa-633 A-21053 Invitrogen Mouse IgG Goat Alexa-488 A-11029 Invitrogen Guinea Pig IgG Goat Alexa-488 A-11073 Invitrogen

4. Plasmid constructs and DNA transfection into cultured cells

nHEKs were transiently transfected one day after seeding with transit- transfection reagent (Mirus Bio LLC, MIR 2804). A total amount of 2.5 µg DNA was mixed with 3.75 µL reagent and 250 µL medium for each dish.

34

The Paxillin-DsRed2 construct was kindly provided by Rick Horwitz and is described in Webb et al. (2014). The GFP-β4 integrin construct was generously provided by Susan Hopkinson and is described in Tsuruta et al. (2003). The K5-EYFP construct was described in Moch et al. (2013). The LifeAct-RFP construct (Ibidi) was provided by Bernd Hoffmann and is described in Riedl et al. (2008). The pEYFP-N1 construct was obtained from Clontech (#6006-1).

5. Drug treatments

The ROCK inhibitor Y-27632 was purchased from StemCell Technologies and used at a concentration of 15 µM or 45 µM. Anti-β4 integrin CD104 (BD Pharmingen) and anti-α6 integrin CD49f (Beckman Coulter) were used as blocking antibodies. In these instances, 10,000 cells suspended in a volume of 100 µL of medium containing the blocking antibodies and were seeded on fibronectin-coated glass coverslips.

6. Micropatterning

Deep-UV micropatterning was performed by Claudia Schmitz according to the protocol described in Azioune et al. (2010). Briefly, 24 mm diameter high precision coverslips were spin-coated with TI PRIME adhesion promoter (MicroChemicals), then with polystyrene (PS, Sigma-Aldrich). Afterwards, they were exposed to deep-UV. They were incubated in Poly(L-lysine)-graft-poly(ethylene glycol) (PLL-g-PEG, Surface Solutions), and then exposed to deep-UV with a customer-designed mask manufactured by Compugraphics.

-2 They were then incubated in human fibronectin (0.5 µg.cm ) suspended in 100 mM

-2 NaHCO3 pH 8.5 (Carl Roth) before cells were seeded at a density of 6,600 cells.cm .

35

7. Preparation of elastic substrates

Elastic substrates were obtained by spin-coating silicon elastomer Polydimethylsiloxane (PDMS) Sylgard 184 (Dow-Corning) on thin glass coverslips according to the method described in Cesa et al. (2007). Optimal total thickness for imaging was 170 µm. Mixing different ratios of silicon oil and curing agent allowed the fabrication of substrates with defined stiffness (Table 2.12).

Table 2.12: Silicon elastomer composition and associated elastic modulus

Ratio mass of curing agent (m2) / Resulting substrates elastic

mass of silicon oil (m1) modulus 1:10 1.2 MPa 1:70 1.5 kPa

A mass m1 of silicon oil was measured. A mass m2 of curing agent equal to m2= α m1 with α = 1/10 or 1/70 was added in order to obtain the desired stiffness. The two compounds were manually mixed for 5 minutes. The mixture was degassed for 20 minutes. Then, it was spin-coated on a glass coverslip (22 mm x 22 mm, #0, Menzel-Gläser) at 1,800 rpm for 15 seconds. For use in live-cell imaging experiments, the coated coverslips were immediately coupled to 35 mm dishes with an 18 mm diameter hole (Cell E&G). Curing was performed for 16 hours at 60 °C.

8. Imaging conditions

Structured illumination fluorescence microscopy was performed with an ApoTome.2.microscope (Zeiss) equipped with an oil immersion objective 63 x (N.A. 1.4, DIC, Plan apochromat).

Live-cell imaging and 3D Z-stack recordings of immunostained samples were performed using a LSM 710 DUO confocal microscope (Zeiss) equipped with a DefiniteFocus device

36

(Zeiss) and an oil immersion objective 63 x (N.A. 1.4, DIC M27) or alternatively an objective 20 x (N.A. 0.8).

Live-cell imaging was performed in a humidified chamber with 5% CO2 set at a temperature of 37°C for nHEKs and 32°C for murine keratinocytes.

9. Image analysis

Manual cell tracking: For tracking of cells based on images acquired over time with the 20x objective and the bright field camera, Cell Tracker programme (Piccinini et al., 2016) developed in Matlab was used in the manual mode.

CMove programme (developed by Georg Dreissen (ICS-7 Jülich)): For quantitative analysis of cytoskeletal dynamics based on images acquired over time with the 63x objective in one or two fluorescent channels, CMove programme was developed using Matlab. In brief, the following procedure was carried out (for definitions of the calculated parameters see Fig. 2.1 and for the procedure see Fig. 2.2):

1. Extraction of a cell mask by thresholding the fluorescent signal 2. Determination of the cell’s trajectory based on the position of the mask’s centroid 3. Registration of the cell correcting both for translation and rotation 4. To calculate the flow on a regular grid within the entire mask, squared shaped templates at time point t were searched in time point t+1 via normalized cross- correlation. 5. Optional: In the case of analysis of two channels (Fig. 2.3), the procedure described for 1 to 4 is also applied for the chosen reference channel. For the second channel, the mask is calculated based on the thresholding of channel 2. But registration is performed as for channel 1. 6. Optional: Normalisation of the flow to a standard cell shape (principle described in Möhl et al. (2012)). When 2 channels are present, normalisation is performed

37

for channel 2 as for channel 1, so that the normalised flow calculated for channel 2 may give values that are outside the standard cell shape when the mask of channel 2 is larger than the mask of channel 1 (Fig. 2.4). 7. Quantitative analysis was done at the entire cell level or at the subcellular level. In the second case, the cells were split in five areas of interest as shown in Fig. 2.4. The mean values in each of these areas were taken into account.

The settings used are shown in Table 2.13.

Table 2.13: Parameters used for keratin and actin flow measurements with the CMove programme Parameter Value Maximum search width displacement 10 pixels Distance between 2 grid points 10 pixels Threshold for cross-correlation 0.6 Pixel size 0.088 µm

Standard image manipulation: General procedures including adjusting contrast, merging channels, performing z-projections, plotting kymographs etc. were performed using Fiji.

38

Figure 2.1: Definitions of speed, flow and directionality. The frame of the dish (Odish, Xdish,

Ydish) does not move and its origin is named Odish. A cell moves in the frame of the dish.

The time interval between two images of the cell is ti+1 - ti = dt. At each time point ti, the centroid of the cell named Ocell i has a different position in the frame of the dish due to cell migration. The position vector (vectors written in bold) of the cell is defined as Rcell i =

OdishOcell i. The migration speed is defined as the norm of the speed vector of the cell in the frame of the dish, namely vcell i = dRcell i / dt = (Rcell i - Rcell i-1) / dt. The mean migration speed is the average over time of the norm of the cell speed vector. At each time point ti, the frame of the cell is (Ocell i, xcell i, ycell i). It is defined so that ycell i is parallel to the short axis of the ellipse into which the mask has been fitted. A piece of keratin filament at

39 position Kker i is defined as Ocell i Kker i = u ker i. The flow of this piece is then defined as its speed vector in the frame of the cell: f K ker,i = d u ker i / dt = (u ker i - u ker i-1) / dt. The mean keratin flow for a cell is defined as the average over time of the norm of the flow for all pieces of filaments in the cell.

Figure 2.2: CMove programme routine for keratin flow measurements. The keratin fluorescent channel is used for analysis. For each frame, the signal outline is extracted automatically by thresholding to create a mask. The position of the centroid of the mask is extracted and allows the calculation of the migration speed and of the cell trajectory. Subsequently, the mask is registered by correcting for translation and rotation to obtain a stable frame of the cell. Keratin motion analysis is then performed using an algorithm based on cross-correlation. The keratin flow is averaged over time. It is finally normalized to a standard D-shaped mask.

40

Figure 2.3: CMove programme routine for parallel measurements of keratin and actin flow. The keratin and the actin fluorescence channels are used for analysis. For the keratin channel, the exact same procedure as described in Fig. 2.2 is applied. For the actin channel, a different mask is calculated that fits the contour of the actin signal. However, it is registered based on the transformation applied to the keratin channel for registration

(R1) (and not based on the movements of the centroid of the actin mask). The actin flow

41 analysis is then performed based on the same algorithm. It is normalized based on the transformation applied to the keratin channel (N1). As a result, the normalized actin flow has a different shape for each cell, but the border between the area with keratin and actin and the area with actin only retains the typical D shape (see also Fig. 2.4).

Figure 2.4: Areas of interest in normalized cells. On the left panel, two different zones are represented. Both keratin and actin can be detected in the green area (i.e., this zone contains the entire keratin signal). The green area is given a standard shape upon normalization of the flow. In pink is the area where some actin can be found but no keratin. It has a variable shape depending on cells, including after normalisation of the keratin flow to a standard shape. The green zone is further subdivided into three areas of interest: the cell front comprising the lamellipodium (excluding the very front of it where no keratin is detected), the cell center comprising the area behind the lamellipodium and below the nucleus and the back comprising the back of the cell with thick keratin bundles (excluding the very end of it where no keratin is detected). The red zone is subdivided into two areas: the leading edge comprising the entire front part of the cell where no keratin is detected, and the rear-end comprising the entire back of the cell where no keratin is detected.

42

10. Statistical analysis

All statistical analyses were performed with GraphPad Prism software. For every graph, mean ± standard deviation (SD) are plotted, except for Fig. 3.9 where the 5-95 % confidence intervals are plotted.

Distributions were considered Gaussian if they passed the d’Agostino & Pearson k2 test with a non-significant P value.

When two conditions with Gaussian distribution were compared, testing was performed with an unpaired Student t-test. When variances were significantly different, Welch’s correction was added. When at least one of the distributions was not Gaussian, then a Mann-Whitney test was used.

When at least three conditions were compared, one-way analysis of variance (ANOVA) followed by Tukey’s test were used if all distributions were Gaussian. Kruskal-Wallis test followed by Dunn’s test on all selected pairs of columns were used in the alternative case.

For correlation analyses, Pearson test was used for Gaussian populations, Spearman test in the alternative situation. In case of positive results, both were followed by linear regression.

* shows a P-value with P < 0.05, ** for P < 0.01, and *** for P < 0.001; n.s. stands for non- significant.

43

Chapter 3: Results

This section is separated into two parts, the first one dealing with keratin dynamics, and the second one dealing with hemidesmosome dynamics during migration. The same experimental model was used in both parts. Keratin intermediate filaments and hemidesmosomes are intrinsically linked, and therefore both parts highlight the dynamics of a single system at different levels: the filament level or the cell-matrix adhesion level.

Some passages have been quoted from manuscripts that are either in preparation or in revision (see Bibliography section).

A. Influence of the mechanophysical environment on keratin network dynamics in migrating cells

1. K5-YFP is a reliable reporter to measure keratin dynamics in migrating normal human epidermal keratinocytes

It has been suggested that the keratin cycle of assembly and disassembly supports rapid shape changes of epithelial cells (Windoffer et al., 2011). However, to date the keratin turnover cycle has not been examined during cell migration. To do this, spontaneously migrating normal human epidermal keratinocytes (nHEKs) from neonatal foreskin were used. Experiments were performed at a low enough passage to avoid cell differentiation (Schafer et al., 2010). When seeded on fibronectin-coated glass, single nHEKs migrated spontaneously adopting a characteristic D-shape with numerous lamellipodia and filopodia at the cell front (red arrow; Fig. 3.1 A) and long retraction fibers at the back (blue arrow; Fig. 3.1 A). Immunoblotting of high salt cytoskeletal extracts prepared from nHEKs showed that they produce all keratins that are typically found in foreskin (Fig. 3.1 B; (Moll et al., 2008)). Besides abundant K5 and K14, the hyperproliferation-associated K6, K16 and K17 were also detectable. These keratins are also a hallmark of the migratory phenotype (Chung et al., 2013). In addition, low amounts of the suprabasal K1 and K10 were also visible most likely produced by a small subpopulation of differentiating keratinocytes.

44

K5-YFP-encoding constructs were prepared to monitor keratin dynamics. Fluorescence microscopy of transiently transfected nHEKs (referred to as K5-YFP nHEKs) revealed a typical cytoplasmic keratin network and co-localization of the fluorescent transgene products with the endogenous keratin network throughout the cells (Fig. 3.1 C-C''). Immunostaining with pan-keratin antibodies further showed that the different keratins are contained within the same network. I therefore conclude that the K5-YFP fusion protein serves as a reliable reporter for the entire keratin network in transiently transfected nHEKs. Furthermore, overexpression of K5-YFP did not visibly affect the morphology of migrating nHEKS or their unique keratin network organization (Fig. 3.1 D- E). The keratin network was concentrated around the nucleus (white arrow) with lateral whisker-like extensions in the rear part of the cell (red arrows). Filament density and bundle thickness decreased towards the cell front.

Given the reported effects of keratins on cell migration (Chung et al., 2013), I assessed the consequences of K5 overexpression on nHEK migration. In comparison to nHEKs transfected with a construct coding for cytoplasmic YFP, K5-YFP nHEKs migrated slower (0.66 ± 0.26 µm.min-1 vs. 0.89 ± 0.33 µm.min-1, respectively; Fig. 1 F). On the other hand, increased persistence was noted in K5-YFP nHEKs as evidenced by an elevated directionality ratio (Fig. 3.1 G; for definitions see Fig. 2.1).

45

Figure 3.1: Migrating normal human epidermal keratinocytes can be used to measure keratin filament dynamics. nHEKs migrate spontaneously on fibronectin-coated glass (A) and express epidermal keratinocyte-specific keratins (B). Fluorescent K5-YFP reporter integrate into the endogenous keratin network of sessile and motile cells (C-E) reducing migration speed (F) and increasing persistence of migration (G). (A) Live-cell confocal phase contrast image (objective 63 x) of a migrating wild-type nHEK. The outlined cell contour reveals a polarized D-shape with multiple lamellipodia and filopodia at the curved

46 cell front (red arrows) and prominent retraction fibers at the straight cell rear (blue arrows). The nucleus (yellow asterisk) is shifted toward the back of the cell. (B) Immunoblot of a 10% SDS-polyacrylamide gel after electrophoretic separation of a high salt buffer extract prepared from wild-type nHEKs using antibodies against K5, K14, K6, K16, K17, K1, K10. The co-electrophoresed size markers are shown at left. (C) Confocal fluorescence microscopy (maximum intensity projections, objective 63 x) of a methanol- acetone fixed nHEK producing K5-YFP after immunostaining with murine pan-keratin antibody cocktail PAN-CK. Note the perfect overlap of the fluorescent reporter (shown in green in (C)) with the immunosignal ((C'); merged image in (C'')). (D, E) show a comparison of confocal fluorescence image of a vital migrating K5-YFP nHEK ((D); taken from Movie 1) with a structured illumination fluorescence image of a migrating methanol-acetone fixed nHEK immunostained with PAN-CK (E). Note the similarity in overall keratin distribution with prominent whiskers extending laterally from the nucleus (red arrows) and a cytoplasmic network thinning towards the cell front. The differences in relative fluorescence intensity between (D) and (E), especially around the nucleus (white arrow), can be explained by restricted epitope accessibility in immunostaining. All scale bars, 10 µm. (F, G) The column scatter plots depict mean migration speeds (F) and directionality ratios (G) ± SD of control nHEKs synthesizing YFP (n = 31) and nHEKs producing K5-YFP (n = 25). Values were extracted from live-cell confocal images (objective 63 x) recorded every 60 seconds for 30 minutes in migrating nHEKs. K5-YFP transfection induces a decrease in the migration speed and an increase in directionality as compared with YFP transfection. Unpaired Student t-test test was used in (F) (P = 0.0053); Mann-Whitney test was used in (G) (P = 0.0401).

2. The keratin flow pattern in migrating keratinocytes differs between the cell front, center and back with respect to speed and direction of keratin flow

To examine keratin flow in migrating nHEKs, confocal time-lapse fluorescence microscopy (30 minutes recordings, 1 image per minute, see Movie 1) was performed on single K5- YFP nHEKs grown on fibronectin-coated glass. Fluorescence was recorded in the focal plane of the ventral part of the cell, which contains most of the keratin network (Fig. 3.1 D; see also Moch et al., 2013). This focal plane is slightly on top of the cortical actin network.

47

Quantitative image analysis was carried out using the cross-correlation based program CMove developed in Matlab. This program allows measuring keratin flow defined as the speed of keratin filaments in the frame of the cell (Fig. 2.1-2). To compare keratin flow patterns in different migrating cells, the results were normalized to a standardized D- shape. The resulting heatmaps revealed specific flow patterns with high and inward- directed flow rates in the cell periphery (Fig. 3.2 A-B). The maximum speed was up to 0.9 µm.min-1, which is 3-4 times higher than the maximum values determined previously in other sessile cells even after EGF stimulation (Moch et al., 2013).

To further dissect the keratin flow pattern, the cell area was split into front, center and back (Fig. 2.4). Highest flow was found in the cell periphery, while lowest values were found in the cell center close to the nucleus (Fig. 3.2 A-A’). Fig. 3.2 A'' further demonstrates that the flow in the periphery was equally elevated in the front and back by a factor of ≈ 2 in comparison to the cell center. Analysis of the direction of the keratin flow showed that it is retrograde in the entire front part of the cell and anterograde only in the back of the cell (Fig. 3.2 B-B’’).

48

Figure 3.2: The keratin flow in migrating normal human epidermal keratinocytes has a well- defined spatial distribution. Data were extracted from live-cell confocal fluorescence images (objective 63 x) of 27 transiently transfected nHEK K5-YFP cells migrating on fibronectin-coated glass (30 minutes recording, 1 image.min-1). (A-A'') The mean speeds of the keratin flow are depicted as a heat map (A) and as column scatter plots of the cell front, center and back (A'). The ratios of the mean speeds of keratin flow determined in different parts of the cells are shown in (A''). The heat map was obtained after shape normalization. Highest flow is found at the periphery of the cell, while lowest flow was

49 found close to the nucleus. For (A’, A''), ANOVA was used for statistical analysis (P < 0.0001) followed by Tukey’s test between all pairs of columns. (B-B’’) Heat map and column scatter plots representing the direction of keratin flow. 90° is defined as the direction of migration. Kruskal-Wallis test was used for statistical analysis (P < 0.0001) followed by Dunn’s multiple comparison test between all pairs of columns. The flow is retrograde in the front and center, and in the direction of migration at the back end, as shown in (B’’) where the average distribution of the retrograde keratin flow is represented (corresponding to areas where the direction of the flow is between 180° and 360°).

3. High keratin flow correlates with high migration speed

For each cell, the mean keratin flow and the mean migration speed were calculated. This showed that higher keratin flow correlated with higher migration speed (Fig. 3.3 A). The migration speed varied linearly with the keratin flow. The theoretical value obtained for a sessile cell according to this plot is ≈ 0.12 µm.min-1, which is consistent with the values determined for sessile A431- and HaCaT-derived cells (Moch et al., 2013).

To further dissect the effect of migration speed on keratin flow at the subcellular level, cells were sorted into a slow group (n = 13) and a fast group (n = 14). The average migration speed differed significantly between the two groups (Fig. 3.3 B). For each group, the normalized mean keratin flow was calculated (Fig. 3.3 C). The values of the keratin flow in the front, center and back of the cells were then compared between the slow and fast group. This showed that the keratin flow was higher in the entire cell body for faster moving cells presenting significantly higher keratin flow in all three zones as compared to the corresponding zones in the slow group (Fig. 3.3 D). The highest increase was noted in the cell center (Fig. 3.3 E).

50

Figure 3.3: Higher migration speed correlates with increased keratin flow. Data were extracted from live-cell confocal fluorescence images (same as those used for Fig. 3.2) of transiently transfected K5-YFP nHEKs. (A) Graph of the mean migration speed in relation

51 to the mean keratin flow. Each dot represents one cell (n = 27). Higher keratin flow correlates with higher migration speed and the increase is linear. Statistical analysis was performed using Spearman correlation (P < 0.0001, R² = 0.7256). In (B-E), the cells were grouped as slow (n = 13) comprising cells with a migration speed < 0.65 µm.min-1 and fast (n = 14) comprising cells with a migration speed > 0.65 µm.min-1. (B) Graphical representation of the mean migration speed for the slow and the fast group. Mann- Whitney test was used for statistical analysis (P < 0.0001). (C) Heat maps of the mean normalized keratin flow in the slow (left) and fast (right) group after shape normalization. (D) Column scatter plots of the mean keratin flow in the cell front, center and the back of both groups. (E) Column scatter plots of the ratios between the mean keratin flow in different areas of fast and slow moving nHEKs. Note that the two groups have significantly different mean migration speeds. There is an overall increase in the keratin flow between both groups. The strongest increase is found in the cell center, the lowest in the back of the cell. Statistical analysis was performed using unpaired Student t-test (P < 0.0001 in (D) (with Welch correction in center area); P = 0.0037 in (E) (front/back); P = 0.0457 in (E) (front/center); P < 0.0001 in (E) (back/center)).

4. The keratin flow pattern mirrors the trajectory of cell migration

To examine the relationship between keratin flow and the trajectory of cell migration, the directionality ratios were calculated for each cell and compared to the keratin flow pattern. Higher keratin flow correlated with higher directionality ratios (Fig. 3.4 A). To examine the effect of an increase in persistence of cell migration on the keratin flow at the subcellular level, cells were sorted into a low directionality group (n = 13) and a high directionality group (n = 14). The average directionality ratios differed significantly between both groups (Fig. 3.4 B).

The mean normalized keratin flow was then calculated and used for the heatmaps in Fig. 3.4 C and the column scatter plots in Fig. 3.4 D-E. A significantly higher keratin flow was found in the front and the center area for the group with higher directionality. The ratio between the flows in the front and the back of the cells was increased for the high directionality group whereas the ratio between the front and the center was unchanged and the ratio between the back and the center was decreased in comparison to the group

52 with lower directionality showing that the increase in keratin flow associated with higher directionality ratio was strongest in the center and front.

53

Figure 3.4: Higher directionality ratio of migration correlates with increased keratin flow. Data were extracted from live-cell confocal images (same as those used for Fig. 3.2 and Fig. 3.3) of transiently transfected K5-YFP nHEKs. (A) Graph of the directionality ratio in relation to the mean keratin flow. Each dot represents one cell (n = 27). Higher keratin flow correlates with higher directionality ratio. Statistical analysis was performed using Spearman correlation test (P = 0.0001, R² = 0.3896). In (B-E), the cells were divided into a low directionality group (n = 13) comprising cells with a directionality ratio < 0.84 and a high directionality group (n = 14) comprising cells with a directionality ratio > 0.84. (B) Column scatter plot of the directionality ratios determined for the low and the high directionality groups. Student t-test with Welch’s correction was used for statistical analysis (P < 0.0001). (C) Heat maps of the mean normalized keratin flow in the low (left) and high directionality groups (right) after shape normalization. (D-E) Column scatter plots of the mean keratin flow in the cell front, center and the back of both groups, and their ratios. The two groups show significantly different directionality ratios. There is an overall increase in keratin flow in the high directionality versus the low directionality group. The strongest increase is found in the front and center of the cell. The following statistical tests were used: Mann-Whitney test (P = 0.0005 for front area in (D); P = 0.0053 for center area in (D)), unpaired Student t-test (P = 0.532 for back area in (D); P < 0.0001 for front/back in (E)), Student t-test with Welch’s correction (P = 0.9424 for front/center in (E); P = 0.0051 for back/center in (E)).

Next, migrating K5-YFP nHEKs were sorted into left turning cells (n = 13) and right turning cells (n = 18). In each case, a symmetry break was observed for the keratin flow pattern in the lateral back part of the cells with elevated keratin flow on the side opposite to the direction of migration (compare areas circled in red and in blue in Fig. 3.5 A).

To standardize the experimental conditions, K5-YFP nHEKs were seeded on micropatterned coverslips with fibronectin-coated sinusoidal lines (15 µm width, curvature 0.02 µm-1, see Movie 2). nHEKs adopted an elongated shape (Fig. 3.5 B). They migrated spontaneously with speeds ranging between 0.2 µm.min-1 and 1 µm.min-1. Keratin dynamics were recorded by confocal time-lapse fluorescence microscopy and analyzed with the help of CMove. The results were shape normalized and the average flow was calculated. The heat map in Fig. 3.5 C shows that higher keratin flow occurred on the convex side in comparison to the concave side. For comparison, cells were also grown on straight 15 µm-wide fibronectin-coated stripes (See Movie 3) and analyzed

54 similarly. Quantitative assessment revealed an asymmetric keratin flow with increased flow on the convex side of cells moving on sinusoidal lines in contrast to the symmetric keratin flow in cells moving on straight stripes (Fig. 3.5 D). Together, the analyses of nHEKs migrating on defined micropatterns fully confirmed the results obtained for freely migrating nHEKs. I therefore conclude that keratin dynamics are finely co-regulated with the speed and trajectory of cell migration.

55

Figure 3.5: Change in direction of migration induces symmetry break in keratin flow patterns. Keratin flow was determined in K5-YFP nHEKs in different situations. (A) Fluorescence recordings of nHEKs producing K5-YFP and migrating on fibronectin-coated glass slides were selected that turned either left or right during 30 minutes confocal time- lapse imaging (n = 11 with 60 s intervals and n = 20 with 120 s intervals) as judged from CMove analysis. The heat maps depict the shape-normalized mean keratin flow patterns of 13 cells turning left and 18 cells turning right. Note that areas with higher flow are located directly opposite to the new direction of migration, i.e. at the right back corner of cells turning left and vice versa (compare corresponding areas at the left and right back of the cells circled in red (higher flow) and blue (lower flow)). (B) The live-cell confocal fluorescence and corresponding phase contrast image (objective 63 x) of an K5-YFP nHEK migrating toward the right on a micropatterned fibronectin-coated sinusoidal line (width 15 µm, curvature 0.02 µm-1) is taken from corresponding Movie 2. The elongated cell adapts to the line width and the nucleus is shifted towards the back. Highly dynamic filopodia and lamellopodia are seen at the front and convex cell margins (red arrows) whereas the concave margin is straight (green arrows) and the cell rear extends long retraction fibers (blue arrow). (C) Heat map of the mean normalized keratin flow derived from fluorescence recordings of 21 K5-YFP nHEKs migrating on a sinusoidal line after shape normalization. Highest flow is found in the front and convex margin. (D) Column scatter plots depicting the ratio between the average keratin flow in the convex and concave part of cells migrating on sinusoidal lines (n = 21). For comparison K5-YFP nHEKs migrating on straight lines (width 15 µm; n = 26) were imaged. Statistical analysis was performed using Mann-Whitney test (P = 0.0007). Higher asymmetry in the keratin flow is seen for cells on waves than on stripes.

5. Increased ECM coating density leads to decreased keratin flow

To further examine the relationship between cell migration and keratin flow patterns, I studied the impact of ECM coating density, which is known to affect the speed of cell migration (Doyle et al., 2013; Palecek et al., 1997). To this end, K5-YFP nHEKs were seeded on glass coverslips coated with either 2.5 µg.cm-2 or 5 µg.cm-2 fibronectin. Keratin dynamics were then measured in migrating cells as described above. The mean cell area and cell shape were unchanged in both conditions (Fig. 3.6 A). But the migration speed and the directionality ratio were lower in cells on high density fibronectin compared to cells on low fibronectin (Fig. 3.6 B). The keratin flow was also lower when the coating density was high (Fig. 3.6 C). I further observed that the keratin flow for a given migration

56 speed was the same independent of the fibronectin concentration (Fig. 3.6 D). The heat maps in Fig. 3.6 E revealed that the keratin flow was reduced in the entire cell body for cells on substrates with higher coating density. Quantification showed that keratin flow was decreased in the cell front, center and back (Fig. 3.6 F) and that the decrease in keratin flow associated with higher coating density is the strongest in the center of the cells (Fig. 3.6 G).

57

Figure 3.6: Increased ECM coating density induces a decrease in migration speed and keratin flow. Data were extracted from live-cell confocal images (30 minutes recordings, 1 image.min-1; objective 63 x) of transiently transfected K5-YFP nHEKs migrating on glass coated either with 2.5 µg.cm-2 fibronectin (n = 14) or 5 µg.cm-2 fibronectin (n = 10). (A) Column scatter plots showing fibronectin density in relation to mean cell area (left) and

58 mean cell eccentricity (right). Statistical analyses were performed using an unpaired Student t-test for mean cell area (P = 0.8011) and a Student t-test with Welch’s correction for mean cell eccentricity (P = 0.3966; n.s.). Note that the cell morphology is not affected by a change in the coating density. (B) Graphical representation of the migration characteristics depending on the coating density. At left, mean migration speed is shown (Mann-Whitney test; P = 0.0434); at right, the directionality ratio is depicted (unpaired Student t-test; P = 0.0017). An increase in coating density correlates with a decrease in migration speed and directionality. (C) depicts the mean keratin flow depending on the coating density. The keratin flow is reduced at the higher fibronectin coating density ([Fn]). Statistical analysis was performed using an unpaired Student t-test (P = 0.0029). (D) Graph of the mean migration speed versus mean keratin flow. For a given migration speed, the corresponding keratin flow is similar irrespective of the coating density. Statistical analysis was performed using Pearson correlation (2.5 µg.cm-2: P = 0.0063, R² = 0.4765; 5 µg.cm-2: P = 0.0047, R² = 0.6530). (E) Heat maps showing the mean normalized keratin flow in shape-normalized nHEKs migrating on low and high density fibronectin (n = 14 and n = 10, respectively). (F) Quantification of the effect of fibronectin coating densities on keratin flow in the cell front, center and is seen in all cell regions; the strongest decrease, however, is detectable in the cell center. Statistical analysis was performed using the unpaired Student t-test (P = 0.0174; front), the Student t-test with Welch’s correction (P = 0.0044; center) and Mann-Whitney test (P = 0.0109; back). (G) The column scatter plots show the ratios of keratin flow in different cell regions at low and high fibronectin coating density. Unpaired Student t-test (P = 0.0684; front/back), Mann-Whitney test (P = 0.3641; front/center), and unpaired Student t-test (P = 0.0148; back/center).

6. Decreased substrate stiffness increases keratin flow

It is known that substrate stiffness affects cell migration (Doyle et al., 2013; Gupta et al., 2015; Lo et al., 2000; Peyton and Putnam, 2005; Yeung et al., 2005; Zhong and Ji, 2013). To study its impact on keratin flow, K5-YFP nHEKs were seeded on fibronectin-coated (5 µg.cm-2) elastic substrates with high stiffness (1.2 MPa) and low stiffness (1.5 kPa). The cells spread equally on both substrates (Fig. 3.7 A) but the cell shape was modified. A higher eccentricity was found for cells on soft substrates because of increased elongation of cells perpendicular to the direction of migration (Fig. 3.7 A). The migration speed and the directionality ratio of cells migrating on soft substrates were higher than for cells migrating on stiff substrates (Fig. 3.7 B). Keratin flow was higher on soft than on stiff

59 substrates (Fig. 3.7 C). I further noted that for a given migration speed the associated keratin flow was the same irrespective of the substrate stiffness (Fig. 3.7 D). Subcellular mapping further showed an overall increase of the keratin flow with the highest increase in the cell back (Fig. 3.7 E-F). The ratios between the keratin flows in the front, center and back of the cells were the same (Fig. 3.7 G).

60

Figure 3.7: Decreased substrate stiffness induces an increase in migration speed and keratin flow. Data were extracted from live-cell confocal images (30 minutes recordings, 1 image.min-1; objective 63 x) of transiently transfected K5-YFP nHEKs migrating on elastomeric substrates with high stiffness (1.2 MPa; n = 12) and low stiffness (1.5 kPa; n = 26). (A) Column scatter plots showing the relationship between substrate stiffness and

61 mean cell area (left) and mean cell eccentricity (right). Statistical analysis was performed using an unpaired Student t-test (P = 0.4327 for mean cell area; P = 0.0007 for mean cell eccentricity). The spreading area is unchanged whereas the cells elongate more perpendicular to the direction of migration on soft substrates. (B) Graphical representation of the migration characteristics depending on the elastic modulus of the substrate. Mean migration speed is shown at left (Mann-Whitney test; P = 0.0035), directionality ratio at right (unpaired Student t-test; P = 0.0524). On softer substrates, the migration speed is increased while the directionality ratio remains unchanged. (C) Column scatter plots of the mean keratin flow depending on the elastic modulus of the substrate. The keratin flow is higher in cells migrating on soft substrates (unpaired Student t-test; P = 0.0012). (D) Graph showing the relationship between mean migration speed and mean keratin flow in cells grown on substrates with different elastic moduli. Statistical analysis was performed using Pearson correlation (glass coated with 5 µg.cm-2 fibronectin (fn): P = 0.0047, R² = 0.6530 (n = 10); 1.2 MPa: P < 0.0001, R² = 0.8434; 1.5 kPa P = 0.0035, R² = 0.3034). For a given migration speed the associated keratin flow does not depend on substrate stiffness. (E) Heat maps of the mean keratin flow in normalized nHEKs migrating on PDMS substrates with an elastic modulus of 1.2 MPa (n =12) or 1.5 kPa (n = 26). (F) The column scatter plots show the effect of the elastic modulus of the substrate for keratin flow in the cell front, center and back. Statistical analysis was performed using unpaired Student t-test (P = 0.3434, front; P = 0.2165, center; P = 0.0105, back). (G) The column scatter plots show the ratios of keratin flow in different cell regions on substrates with high (1.2 MPa) and low stiffness (1.5 kPa). Unpaired Student t-test was applied in all instances (P = 0.1014, front/back; P = 0.3876, front/center, P = 0.9756, back/center). On softer substrates the increase in keratin flow is the strongest in the back of the cells.

7. Confinement of migrating normal human epidermal keratinocytes reduces keratin flow

Confinement has been shown to affect migration (Friedl and Wolf, 2010; Welch, 2015). I therefore wanted to find out, how this might modulate keratin flow patterns. To do this, I studied K5-YFP nHEKs on straight 15 µm-wide fibronectin-coated stripes (See Movie 3). Similar to cells moving on the sinusoidal line pattern, the cells drastically elongated with a leading edge characterized by lamellipodia and filopodia and a cell rear with typical retraction fibers (compare Fig. 3.8 A with Fig. 3.5 B). As a control, cells were grown under identical conditions on coverslips that had been treated in the same way as

62 micropatterned coverslips, but without a mask. The cell eccentricity was strongly increased in nHEKs moving on the stripes in comparison to those freely-moving on the evenly coated surface, but the cell area was reduced (Fig. 3.8 B). The migration speed was reduced on stripes as compared with the control (Fig. 3.8 C). In accordance, keratin flow was lower on stripes in comparison to the control (Fig. 3.8 D). Remarkably, I observed that for a given migration speed, the corresponding keratin flow was lower for cells migrating on stripes than for freely-moving cells (Fig. 3.8 E). The heat map in Fig. 3.8 F shows that the keratin flow was highest in the back of the cells moving on straight stripes. The slower flow at the cell front was still significantly higher than in the cell center (Fig. 3.8 F-H).

Figure 3.8: Confinement impedes migration speed and keratin flow. K5-YFP nHEK migration was restricted to micropatterned 15 µm-wide fibronectin-coated stripes. As control, K5-YFP-expressing nHEKs were seeded on fibronectin-coated coverslips that were prepared by deep UV illumination exactly like the micropatterned coverslips but

63 covering the entire glass surface of the coverslip (i.e. with no mask). Fluorescence images were recorded by confocal laser microscopy (objective 63 x, 1 image.min-1 for 30 minutes; n = 26 for cells on micropattern; n = 13 for the control). (A) Live-cell confocal fluorescence and corresponding brightfield image (objective 63 x) of a K5-YFP nHEK migrating on a fibronectin-coated stripe prepared by deep-UV micropatterning. The cell is elongated and the nucleus (green arrow) is shifted towards the back. The front of the cell (red arrow) is rich in lamellipodia and filopodia, while the back (blue arrow) contains multiple retraction fibers. (B) Column scatter blots depicting the effect of confinement for mean cell area and mean cell eccentricity. Statistical analysis was performed using an unpaired Student t-test (P = 0.018 for mean cell area; P < 0.0001 for mean cell eccentricity). The cell spreading area is reduced while the cell eccentricity is dramatically increased for cells on stripes. (C) Graphical representation of the stripe-induced pseudo-confinement on mean migration speed. Statistical analysis was performed using a Mann-Whitney test (P < 0.0001). The migration speed is significantly reduced in comparison to the control. (D) Graphical representation of the mean keratin flow depending on stripe-induced confinement. Statistical analysis was performed using an unpaired Student t-test (P < 0.0001). The keratin flow is reduced upon pseudo-confinement on a stripe. (E) Graph depicting the relationship between the mean migration speed and mean keratin flow with or without confinement. For a given migration speed, keratin flow is slower for cells forced to migrate on stripes than for free migrating cells. Statistical analysis was performed using Pearson correlation (Control: P < 0.0001, R² = 0.7859; Stripes: P = 0.009, R² = 0.2509). (F) Heat map representing the mean speed of the keratin flow after shape normalization in K5-YFP nHEKs migrating on a stripe. (G) Column scatter plots of the speed of the keratin flow in the front, center and back of cells migrating on stripes. (H) Column scatter plots of the ratios between the speed of the keratin flow in different areas of cells migrating on stripes. For (G) and (H) Kruskal-Wallis test was used for statistical analysis (P < 0.0001) then Dunn’s test between all pairs of columns. Highest flow is found at the back of the cell, while lowest flow is found close to the nucleus.

8. Keratins only mildly affect the mechanophysical-dependent regulation of migration

The results described in the preceding paragraphs demonstrated that both migration and keratin dynamics are dependent on the cellular environement. I therefore wanted to know whether the migration features observed in the different mechanophysical environments was dependent on keratin. To address this question, I used the recently described keratin-free (KtyII-/-) mouse keratinocytes (KO), in which the entire keratin type

64

II gene cluster was depleted (Kroger et al., 2013). As controls, I used the wild-type cells (WT; (Kroger et al., 2013)) and K5-YFP rescue cells (RES; (Sawant et al., 2018)). When seeded at low enough density, these cells migrate spontaneously.

First, cells were seeded on 15 µm stripes and on control coverslips. Phase contrast images were recorded the next day for 1 hour (1 image every 3 minutes). In control conditions, KtyII-/- cells were significantly faster than WT and RES in accordance with previous observations (Seltmann et al., 2013). All cell types were slower on stripes when compared to the control. But no significant differences could be observed for the migration speed of KtyII-/-, WT and RES cells on stripes (Fig. 3.9 A).

Next, I compared the behavior of KtyII-/- keratinocytes to control cells migrating on elastic substrates with low and high stiffness. No significant increase in the migration speed could be observed for WT on soft substrates compared to stiff substrates. However, the migration speed of KtyII-/- and RES cells was significantly decreased on soft substrates. (Fig. 3.9 B).

The results indicated that the presence of a functional keratin network has a complex effect on the ability of cells to migrate in different environments. However, the keratin network does not seem to play a key role in the regulation of the speed of migration in different environments. I therefore assume that keratin flow is regulated downstream of the actin flow. This assumption predicts that regulators of actin flow should be affected in the different paradigms. To test this idea, I examined focal adhesions known to be key players in environment sensing and regulators of actin dynamics (Geiger and Yamada, 2011; Möhl et al., 2012). I studied focal adhesion density in relation to fibronectin coating density and substrate stiffness (Fig. 3.9 C-D). As predicted, focal adhesion density was increased on high fibronectin density and increased substrate stiffness.

65

Figure 3.9: Mechanophysical-dependent regulation of migration is only mildly affected by keratins. (A-B) Mean migration speed was determined in wild-type murine keratinocytes (WT), KtyII-/- keratinocytes (KO) and K5-YFP rescue keratinocytes (RES) migrating on soft 1.5 kPa substrates, on stiff 1.2 MPa substrates, on 15 µm stripes and on deep UV- illuminated fibronectin-coated glass (control). Brightfield images were recorded by live- cell confocal laser microscopy (objective 20 x) for 60 minutes at 3 minutes intervals. Cells were tracked manually with the help of Cell Tracker software. The derived migration speeds are presented as whisker box plots (5-95%) with outliers shown as individual dots. (A) Stripes: WT, n = 163; KO, n = 133; RES, n = 241. Control: WT, n = 163; KO, n = 218; RES, n = 317. (B) 1.2 MPa: WT, n = 172; KO, n = 149; RES, n = 167. 1.5 kPa: WT, n = 113; KO, n = 102; RES, n = 201. Kruskal-Wallis test was used for statistical analysis in A and B (P < 0.0001) followed by Dunn’s multiple comparison tests on all selected pairs of columns. Only significant differences are shown. Note that migration speed is increased for KO keratinocytes on fibronectin-coated glass which is reversed in the rescue (control in (A)). Migration speed on stripes, however, is independent of the presence of keratins and

66 always reduced in comparison to the control. The plot in (B) reveals that, in contrast to the WT, the migration speed of KO keratinocytes is reduced on soft substrates. In this instance, the phenotype is not rescuable by K5-YFP. (C) Quantitation of focal adhesion immunostaining using paxillin antibodies in PFA-acetone fixed nHEKs grown on fibronectin-coated glass coverslips (2.5 µg.cm-2, n = 75; 5 µg.cm-2, n = 61). The area covered by focal adhesions was determined after manual thresholding of the fluorescence signal recorded by structured illumination microscopy. The total cell area was measured following manual extraction of the cell contour. The whisker box plot shows the relative area covered by focal adhesions. Mann-Whitney test was used for statistical analysis (P < 0.0001). (D) Whisker box plot showing the relative focal adhesion density in nHEKs grown on elastomeric substrates with different stiffness (1.2 MPa, n = 148; 1.5 kPa, n = 114) immunostained for paxillin (PFA-Triton X 100 fixation). Analysis was done as described in (C). Mann-Whitney test was used for statistical analysis (P < 0.0001). For cells seeded on softer substrates, the area covered by focal adhesions is lower.

9. Keratin flow lags behind actin flow in migrating keratinocytes

To investigate the relationship between actin and keratin dynamics in migrating nHEKs, cells were doubly transfected with the K5-YFP encoding construct and LifeAct-RFP for subsequent time-lapse recordings (30 minutes recording period, 2 images per minutes, see Movie 4). Actin and keratin dynamics were measured using CMove as specified in Fig. 2.4. The mean actin flow was in the expected range when compared to values reported in the literature for non-migrating cells using other methods (Caspi et al., 2001; Vallotton et al., 2004). When averaging the flows in the entire cell, actin flow was higher than keratin flow (Fig. 3.10 A). For a given migration speed, the associated actin flow was higher than the associated keratin flow (Fig. 3.10 B). To study actin and keratin flows at the subcellular level, each flow pattern was normalized separately (Fig. 2. 4). The resulting heat maps revealed that actin flow was particularly high in the most peripheral part of the cell front that is devoid of keratin (Fig. 3.10 C). Actin flow rates dramatically decreased at the border to the area with keratins (dashed line in Fig. 3.10 C). Within the overlap regions actin and keratin flow patterns were highly similar with comparable speed and direction. Yet, keratin flow was always slightly slower than actin flow indicating a delay between both. The difference was only minor in the cell front and back and more

67 pronounced in the cell center likely indicating a reduced degree of coupling (Fig. 3.10 D- E). The delay between keratin and actin seems to be reduced when the migration speed is higher (Fig. 3.10 B). The coupling between actin and keratin flow pattern is also evident in turning cells where an equivalent symmetry break in the actin flow is observed (Fig. 3.11 to compare to Fig. 3.5 A).

68

Figure 3.10: Actin and keratin interact dynamically during migration. Data were extracted from live-cell confocal images (objective 63 x) of nHEKs transiently co-transfected with K5-YFP and LifeAct-DsRed constructs (n = 21) migrating on fibronectin-coated glass (30 minutes recording, 2 images.min-1). (A) Column scatter plot of the mean keratin and actin flow. Actin flow is higher than keratin flow. Statistical analysis was performed using an

69 unpaired Student t-test (P < 0.0001). (B) Graph showing the mean actin and keratin flow in relation to the migration speed. For a given migration speed, actin flow is higher than keratin flow. The difference between actin flow and keratin flow is lower at higher migration speed. Statistical analysis was performed using Pearson correlation (for keratin P = 0.0022, R2 = 0.3974 and for actin P = 0.0152 and R2 = 0.2724). (C) Heat maps of the mean actin and keratin flows in shape-normalized migrating nHEKs. The signal shown for the actin flow outside the normalized shape corresponds to peripheral areas where actin can be found but no keratin. In these areas, the average flow was calculated only over the number of cells in which actin was detectable. (D) Left: Graphical representation of the speed of actin flow in 5 different areas of the cell. Statistical analysis was performed using ANOVA (P < 0.0001) followed by Tukey’s test on all pairs of columns. Middle: Graphical representation of the speed of keratin flow in 3 different areas of the cell. Statistical analysis was performed using ANOVA (P < 0.0001) followed by Tukey’s test on all pairs of columns. Right: Graphical representation of the ratio between actin and keratin flows in 3 different areas of the cells where both cytoskeletal components are detected. Kruskal-Wallis test was used for statistical analysis (P < 0.0001) followed by Dunn’s test. (E) Column scatter plots of the actin and the keratin flow in the front, center and back of cells where both cytoskeletal components are detected. For statistical analysis, the following tests were used: left: unpaired Student t-test (P = 0.0932), middle: unpaired Student t-test (P < 0.0001), right: t-test with Welch’s correction (P = 0.9412). Both actin and keratin flows are higher at the cell periphery than in the cell center. Actin flow shows particularly high values at the leading edge of the cell where keratin is absent, and is reduced as soon as it is in an area where keratin is detected. The difference between actin and keratin flows is highest in the cell center.

70

Figure 3.11: Asymmetry of actin flow in turning normal human epidermal keratinocytes. Representation of the mean normalized actin flow in a migrating nHEK, obtained by averaging the mean actin flow over time normalized to a standard shape for the cells sorted between two groups: the ones turning left (n = 7) and the ones turning right (n = 10). Data was extracted from live-cell confocal images (objective 63 x) of nHEKs transiently transfected with K5-YFP and LifeAct-DsRed constructs migrating on fibronectin-coated glass (30 minutes recording with an imaging frequency of 2 images.min-1). Higher flow is found on the opposite side of the direction change (i.e., in the areas demarcated by red lines compared to areas demarcated by blue lines).

71

B. Hemidesmosomes and focal adhesions form treadmilling arrays in migrating primary keratinocytes

1. Hemidesmosomal proteins cluster in chevron-shaped arrays in migrating primary human keratinocytes

Hemidesmosomes are cell-matrix adhesions anchoring keratin intermediate Filaments. I therefore wanted to investigate their organization and dynamics during migration to determine how they could contribute to the observations made on keratin in the previous part. Spontaneously migrating nHEKs were seeded at low density on fibronectin-coated glass coverslips. Immunostaining of these cells two days after seeding with antibodies against the hemidesmosome-specific β4 integrin revealed a highly ordered pattern consisting of chevron-shaped arrays (Fig. 3.12 A). The arrays radiate from the cell rear toward the entire semi-circular lamellipodial leading edge. The same pattern was observed with antibodies against α6 integrin, which associates with β4 integrin in hemidesmosomes (Fig. 3.12 B). To show that the α6/β4 integrin-positive sites are hemidesmosome-like structures, co-immunostaining was performed for β4 integrin with the hemidesmosomal -type plaque proteins BP230 and plectin. Extensive co- localization was detected with highest co-distribution for BP230 (Fig. 3.12 C). The anti- plectin staining pattern was more difficult to interpret because of the additional non- hemidesmosomal localization of plectin (Wiche et al., 2015).

72

Figure 3.12: Hemidesmosomal proteins localize to arrays of chevron-shaped structures in migrating normal human epidermal keratinocytes. In (A-D), the panel shows immunofluorescence images (epifluorescence microscopy) of single polarized nHEKs that had been seeded on glass coverslips coated with fibronectin and were fixed after two days with methanol-acetone. (A, B) Anti-β4 integrin and anti-α6 integrin antibodies stain similar arrays consisting of aligned chevron-shaped structures that radiate from the rear to the front and lateral sides. (C, D) Anti-β4 integrin antibodies co-localize with either anti- BP230 antibodies ((C), (C'), merge in (C'')) or anti-plectin antibodies ((D), (D'), merge in (D'')) in the same chevron-like structures. (E) Immunofluorescence images (epifluorescence) of confluent nHEKs seeded on fibronectin-coated glass coverslips and fixed with paraformaldehyde-acetone 7 days after seeding. Staining for β4 integrin and paxillin was performed with antibodies; labelling of actin was performed with Alexa-488 phalloidin. Note that hemidesmosomal chevron patterns cannot be detected in these conditions. Scale bars 10 µm.

73

Double immunofluorescence microscopy with antibodies against β4 integrin and keratin further showed that the β4 integrin-positive chevrons were connected to keratin intermediate filaments (Fig. 3.13 A-B). Taken together, I conclude that chevrons are a unique type of hemidesmosomal arrangement that is specific for migrating cells.

Figure 3.13: Hemidesmosomal chevron patterns anchor keratin intermediate filaments in migrating normal human epidermal keratinocytes. The panel shows the indirect immunofluorescence of anti-β4 integrin (A) and anti-keratin antibodies (A') (merge in (A'') and (B)) in the periphery of a migrating nHEK. (A) Maximum intensity projections recordings. (B) Images of single 0.1435 µm-thick focal planes. Note the association of keratin filaments with β4 integrin-positive structures. Scale bars: 10 µm.

74

2. Hemidesmosomal chevron arrays and focal adhesion sites are spatially linked but segregated

To examine the relationship between hemidesmosomal chevrons and actin-anchoring focal adhesions in migrating nHEKs, fluorescence microscopy analyses were performed (Fig. 3.14). The resulting images revealed that paxillin-positive and actin stress fiber- associated focal adhesions are intercalated between the hemidesmosomal chevrons. Despite their close neighbourhood hemidesmosomes and focal adhesions do not overlap, forming regularly patterned superstructures.

Figure 3.14: Focal adhesions intercalate between hemidesmosomal chevrons in migrating cells. (A-E) Quadruple epifluorescence images show anti-4 integrin immunofluorescence, anti-paxillin immunofluorescence, Alexa-488 phalloidin labelling and DAPI labelling either alone or in different combinations (false white colour for anti- paxillin fluorescence in (C)) in single nHEKs. nHEKs were seeded on fibronectin-coated glass coverslips and fixed with paraformaldehyde and acetone after two days prior to labelling. Note that paxillin-positive focal adhesions align next to but do not overlap with 4 integrin-positive hemidesmosomal chevrons. Furthermore, actin stress fibers associate selectively with paxillin-positive structures. Scale bars 10 µm.

To more precisely examine the spatial relationship between hemidesmosomes and focal adhesions, nHEKs were seeded on X-shaped micropatterned fibronectin islands.

75

Fluorescence staining showed that 4 integrin clustered in chevron-like structures, which were arranged symmetrically along the four arms of the pattern (Fig. 3.15 A1). Chevrons increased in size towards the cell center. Anti-paxillin staining further revealed that focal adhesions were precisely localized between the hemidesmosomal chevrons (Fig. 3.15 A2- A3). As expected, the focal adhesions were associated with prominent actin stress fibers, while chevrons were not (Fig. 3.15 A4-A6). When nHEKs were grown on fibronectin- coated D-shaped patterns, a different yet still non-overlapping distribution of hemidesmosomal and focal adhesion proteins was detected (Fig. 3.15 B3). β4 integrin- signal was detected in a broad circumferential region with multiple indentations on both sides harbouring focal adhesions. The outer focal adhesion-rich region consisted of small attachment sites and was directly next to the actin-rich cell cortex, while larger, actin stress fiber-associated focal adhesions localized to the gaps along the inner border of the hemidesmosome-rich region (Fig. 3.15 B4-B6).

76

Figure 3.15: Focal adhesions are sandwiched in between hemidesmosomes in normal human epidermal keratinocytes seeded on X- or D-micropatterns Quadruple epifluorescence of a single paraformaldehyde-acetone fixed nHEK that had been seeded on fibronectin-coated micropatterns ((A): X; (B): D) and were fixed after one day before labelling. Actin was detected with phalloidin-Alexa 488 nm, and staining for β4 integrin and paxillin was done with antibodies. (A) On X-patterns, β4 integrin signal (false colour in (A3)) shows a striking resemblance with chevrons observed in Fig. 3.13. Paxillin-positive focal adhesions align next to but do not overlap with 4 integrin-positive hemidesmosomal chevrons. Actin stress fibers associate selectively with paxillin-positive structures. (B) Note that the β4 integrin signal (false white colour in (B3)) is sandwiched

77 between focal adhesions at the outermost cell periphery next to the cortical actin and actin stress fiber-associated focal adhesions located in the more central part of the cell. Scale bars: 10 µm.

3. Focal adhesion-decorated hemidesmosomal chevrons are formed at the cell front and are removed in the cell rear in migrating keratinocytes

Time-lapse fluorescence microscopy was done to study the dynamics of hemidesmosomes and focal adhesions in migrating nHEKs. To this end, nHEKs were transiently co-transfected with a β4 integrin-eGFP and a paxillin-DsRed construct to label hemidesmosomes and focal adhesions, respectively. Assembly of focal adhesions and hemidesmosomes was exclusively detected at the cell front (Fig. 3.16 A). Paxillin-positive nascent focal adhesions appeared first (Fig. 3.16 A'; Movie 6). They subsequently enlarged into ellipsoid focal adhesions at the lamellipodium-lamellum interface. Pairs of ellipsoid focal adhesions usually faced each other. They were arranged at an oblique angle to the direction of migration. The space between the focal adhesions was subsequently filled with hemidesmosomal β4 integrin extending the pre-existing chevron-like structures (Movie 6). At the same time, focal adhesions enlarged slightly. Once the characteristic focal adhesion-hemidesmosome chevron pattern was established, it stayed in place with respect to the substratum without positional changes of its components (Fig. 3.16; Movie 5). Yet, the size and shape of individual focal adhesions was subject to variation. The continuous growth of the focal adhesion- hemidesmosome chevrons at the cell front was paralleled by translocation of the cell body in the direction of migration. As a result, old focal adhesion-hemidesmosome chevrons, which remained tightly attached to the substratum, became localized towards the cell rear. At the very back of the chevron arrays a slight sliding of focal adhesions occurred eventually leading to removal of paxillin-positive structures just prior to retraction fiber formation (Movie 5 and kymograms in Fig. 3.16 B). Typically, β4 integrin- positive regions persisted until retraction fibers were ripped off leaving behind substratum-attached 4 integrin patches (see also Fig. 3.14 A). The scheme in Fig. 3.16 C summarizes the observed morphogenesis and removal of the focal adhesion- hemidesmosome chevron pattern.

78

Figure 3.16: Spatial organization and dynamics of focal adhesions and hemidesmosomes are coordinated during cell migration. Time-lapse fluorescence confocal microscopy of a single migrating nHEK doubly transfected with plasmids encoding GFP-labelled β4 integrin (GFP-hb4) and DsRed2-labelled paxillin (paxillin-DsRed2). Cells were seeded on fibronectin-coated glass coverslips and transiently transfected the next day. Live-cell imaging (60 minutes, 60 seconds interval; Movies 5 and 6) was performed one day after transfection. (A) Focal adhesions are formed near the proceeding cell prior to the appearance of nearby β4 integrin-positive structures (higher magnification of boxed area

79 in (A'); nascent focal adhesions are marked by white arrows). Repetitive cycles of this process lead to the formation of new chevrons with intercalated focal adhesions. Note the remarkable positional stability of individual focal adhesions throughout the recording period (examples denoted by yellow and blue arrows). (B) The kymograms were prepared along the lines shown in (A1). They reveal that the composite and highly organized β4 integrin-positive hemidesmosomal chevrons fill the spaces between newly formed paxillin-positive focal adhesions and that both maintain their position on the substrate in the migrating cell during most of their life time. In the cell rear, paxillin-positive structures are removed prior to β4 integrin (right side of the kymograms). (C) Scheme representing the cycles of assembly at the front and disassembly at the back undergone by hemidesmosomal chevrons and focal adhesions during migration. Scale bars: 10 µm in (A1-A6); 5 µm in (A'1-A'6); 3 µm in (B).

4. Hemidesmosomal chevron patterns form during cell adhesion and spreading

The formation of hemidesmosomal chevrons in the leading front of migrating cells suggested that a similar process may occur in nHEKs during adhesion and spreading. To test this idea, nHEKs were placed on fibronectin-coated glass slides and fixed after defined time intervals for staining with anti-4 integrin and anti-paxillin antibodies as well as phalloidin labelling (Fig. 3.17 A). 30 minutes after seeding, cells were circular presenting multiple paxillin-positive focal adhesions in the cell periphery that were associated with phalloidin-positive actin filaments. At this time point, only a multipunctate and disordered β4 integrin-distribution was detectable in the cell center. By 75 minutes after seeding, cells had spread further and focal adhesions were not any more restricted to the outermost cell periphery. Conversely, β4 integrin-positive structures had formed nearby with no visible overlap. But alignment of both adhesion sites could be seen (arrows at 75 minutes after seeding in Fig. 3.17 B and high magnification in Fig. 3.17 B’-B’’’ from the area pointed by the yellow arrow). 165 minutes after seeding, cells started to migrate. By this time, aligned chevrons were discernible in the entire leading front of cells (see arrows at 165 minutes after seeding in Fig. 3.17 C and high magnification in Fig. 3.17 C’-C’’’ from the area pointed by the yellow arrow).

80

To find out, how the morphogenesis of the hemidesmosome chevrons relate to keratin network organization, spreading cells were co-stained for β4 integrin and keratin. Co- localization was obvious 45 minutes after seeding, i.e. at a time when hemidesmosome- like structures were just appearing. Each of the 4 integrin dots in the cell periphery was associated with radial keratin filaments (Fig. 3.17 D). In the outermost part of the spreading cells, many small filamentous keratin particles were also seen. They are known to correspond to nascent keratin filaments (Windoffer et al., 2011).

81

Figure 3.17: Chevron-like hemidesomosomes are formed next to focal adhesions and in association with keratin filaments during substrate adhesion. (A-C) Quadruple epifluorescence microscopy of single nHEKs seeded on fibronectin-coated glass coverslips and fixed with paraformaldehyde-acetone at the indicated time intervals after seeding. Cells were stained with phalloidin-Alexa 488 nm, DAPI and antibodies against β4 integrin and paxillin as indicated. 30 minutes after seeding, focal adhesions are localized exclusively in the cell periphery (A3) with associated radial actin stress fibers and

82 concentric actin filament bundles (A1), while β4 integrin localizes to disordered dot-like structures predominantly in the cell center. After 75 minutes, focal adhesions are also seen further towards the cell center (B3). Conversely, β4 integrin accumulates in more peripheral parts of the cell (B2). After 165 minutes, typical focal adhesion-chevron arrangements appear near the cell front in a fully polarized migrating cell (C2-C4; see arrows in (B4) and (C4)). (B, C) show enlargements of the areas demarcated by yellow arrows in (B4) and (C4). (D) nHEKs were fixed with methanol-acetone and immunostained for β4 integrin and keratin 45 minutes after seeding. 3D z-stacks were recorded by confocal laser microscopy. (D1) depicts a maximum intensity projection of an entire cell, (D2) a 0.1435 µm-thick single focal plane of the area delineated in (D1). At this stage, the keratin is mostly concentrated around the nucleus with prominent radial filament that are associated with β4 integrin-positive dots. Note also the presence of multiple keratin filament particles of variable size in the most peripheral part of the cell. Scale bars: 10 µm in (A-C) and (D1); 2.5 µm in (B’-B’’’), (C’-C’’’) and (D2).

5. Focal adhesions and hemidesmosomes affect each other's distribution during chevron pattern formation

Since the formation of focal adhesions slightly precedes the formation of hemidesmosomes, I wondered whether hemidesmosomes are able to cluster into chevron patterns when focal adhesion formation is impaired. I therefore treated nHEKs with the Rho-associated protein kinase (ROCK) inhibitor Y-27632 during substrate adhesion. Focal adhesions were still formed in the presence of 20 µM Y-27632. But they were smaller than in control cells and remained exclusively in the cell periphery as described (Humphries et al., 2007). β4 integrin formed patches around these residual focal adhesions but lacked the characteristic chevron shape (Fig. 3.18). I therefore concluded that the formation of bona fide focal adhesions is necessary for hemidesmosomes to cluster into chevron patterns.

83

Figure 3.18: Focal adhesions disruption affects hemidesmosomal chevrons formation during substrate adhesion. The panel presents immunofluorescence images (epifluorescence microscopy) of single nHEKs seeded on fibronectin-coated glass coverslips and fixed with PFA-acetone at different times after seeding as specified. Cells were stained for β4 integrin and paxillin. Cells were treated with the ROCK inhibitor Y- 27632 (15 µM) during seeding until fixation. Y-27632 prevents proper focal adhesion formation, i.e. focal adhesions remain in the cell periphery and are smaller than in the control. Note that hemidesmosomes fail to form chevron-like structures and concentrate in the cell periphery. Scale bars 10 µm.

84

I then investigated how, conversely, the distribution of focal adhesions is affected by impairment of hemidesmosome adhesion. To this end, nHEKs were treated with blocking antibodies against β4 integrin. Although β4 integrin-positive patches still remained after this treatment in the cell periphery, the hemidesmosome chevrons were completely absent (Fig. 3.19 A). An abnormally high number of newly forming focal adhesions was detectable in the cell periphery after 30 minutes of β4 integrin-antibody treatment. At later time points, only very few and more mature focal adhesions were seen towards the cell center with little change over time (75 minutes and 165 minutes in Fig. 3.19 A). Very similar results were obtained, when spreading nHEKs were treated with blocking antibodies against α6 integrin (Fig. 3.19 B).

Taken together, the observations suggest that a crosstalk between focal adhesions and hemidesmosomes is needed for the formation of hemidesmosome chevron patterns with intercalated focal adhesions.

85

Figure 3.19: Formation of hemidesmosomes and focal adhesions is impaired by treatment with blocking anti-β4 or anti-α6 integrin antibodies during substrate adhesion. The epifluorescence images were recorded in single nHEKs seeded on fibronectin-coated glass coverslips and fixed with paraformaldehyde-acetone at different times after seeding prior to staining with anti-β4 integrin and paxillin. Cells were treated with anti- β4 integrin (A) or with anti- α6 integrin (B) blocking antibodies directly during seeding until fixation. Both treatments prevent the formation of hemidesmosomal chevrons but still allow accumulation of β4 integrin in dot-like structures. In this scenario, an increased number of focal adhesions is detected after 30 minutes. The cells, however, fail to form mature focal adhesions in the cell center later on. Scale bars: 10 µm.

Finally, I investigated how hemidesmosomes rearrange upon partial disruption of focal adhesions during migration. Two days after seeding, cells were treated with Y-27632 for

86

20 minutes, fixed and stained with antibodies against paxillin and β4 integrin. Again, only small focal adhesions were formed and their distribution was restricted to the most peripheral part of the migrating cells. In contrast, the overall organisation of hemidesmosomes was not altered and hemidesmosome chevrons remained aligned throughout the 20 minutes treatment period (Fig. 3.20). But the individual chevrons were not as sharply delineated as in the control cells.

87

Figure 3.20: Hemidesmosomal chevrons persist upon focal adhesion disruption. The epifluorescence images were taken from single nHEKs seeded on fibronectin-coated glass

88 coverslips and fixed 2 days after seeding with paraformaldehyde-acetone. Cells were stained for β4 integrin and paxillin with antibodies. Cells were treated with the ROCK inhibitor Y-27632 (45 µM, 20 minutes) prior to fixation. Note that Y-27632 depletes focal adhesions, with only a few very small focal adhesions remaining in the cell periphery. Yet, hemidesmosomal chevrons conserve their overall organisation. Scale bars: 10 µm (except zooms: 5 µm).

89

Chapter 4: Discussion

Keratins have been shown to fulfil two a priori antagonistic functions: maintaining tissue integrity while being highly dynamic. Additionally, studies of keratin KO cells suggest that keratins inhibit cell migration (Seltmann et al., 2013). However, the dynamics of the keratin network in migrating cells and how it affects cell migration has not been investigated so far. The goal of the present work was therefore to determine how the dynamic behavior of the epithelial keratin intermediate filament cytoskeleton and its associated hemidesmosomes are integrated in migrating cells.

Primary human keratinocytes from the basal layer of the epidermis were used as an experimental model. This system presents pros and cons. A major advantage is that these primary cells are genetically and phenotypically closer to their tissue of origin than immortalized cell lines. nHEKs are spontaneously migratory with a reproducible half-circle shape. They contain a prominent and highly dynamic keratin network, which is compositionally very similar to reactive keratinocytes. Most importantly, they form highly ordered hemidesmosomes-like structures. On the other hand, primary cells are more sensitive at all steps of cultivation (thawing, passaging…), and genetic manipulations are more difficult to perform. Isolating stable genetically modified cell lines takes more than five passages. Therefore, all transfections performed in this work were transient transfections, most of which had very low efficiency.

1. Keratin flow pattern during migration

The measurement of keratin dynamics was facilitated by transient overexpression of fluorescently-tagged keratin 5. The exact level of overexpression could not be measured by standard methods such as immunoblotting since only a minority of cells was transfected in each dish. Controls were made to check that the method allowed to capture the dynamics of the entire keratin network and that the morphology of the keratin network was not affected by the induced keratin overexpression. Interestingly, the controls showed that all expressed keratin isoforms are contained within the same

90 keratin network together with the exogenous keratins. Additionally, the effect of overexpression induced only a mild decrease in migration speed and increase in directionality ratio. These effects are in accordance with the inverse effects of keratin KO on cell migration in murine keratinocytes (Seltmann et al., 2013).

Imaging conditions were optimized to minimize phototoxicity and bleaching. Since nHEKs are quite flat when seeded on fibronectin-coated substrates, it was sufficient to image single focal planes to capture keratin network dynamics missing only the upper part of the nuclear cage. Furthermore, given the comparatively slow kinetics of keratin cycling, an imaging frequency of one frame per minute was enough for tracking algorithms. However, measuring actin dynamics required higher imaging frequency.

The use of a powerful image analysis tool for flow analyses of cytoskeletal components in combination with cell shape normalisation allowed to collectively determine and compare flow patterns for multiple cells under different conditions. The cell shape normalisation step allowed to average results over many cells into a single map, reducing artefacts and singularities. Thus, topological maps of keratin dynamics were produced with subcellular resolution in migrating cells subjected to defined environments. The resulting precision superseded visual inspection as well as other currently available methods for measuring cytoskeletal dynamics such as fluorescence recovery after photobleaching (FRAP) or photoactivation. Indeed, these two techniques primarily give information on the time needed for a population of fluorescent molecules to be renewed in a defined area. To deduce the speed of displacement of molecules from these results, the bleached or photoactivated area cannot be the entire cell, and hypothesis have to be made on the direction of the flow. With the technique used here, no hypothesis was needed, and the flow was measured in the entire cell.

My results show that keratins are highly dynamic during cell migration, with retrograde flow in the entire cell except for the back of the cell where anterograde flow can be found. The two thick keratin whiskers on the side of the nuclear cage are found at the border

91 between these two zones. The flow is therefore inward directed in the entire cell. This spatial distribution resembles an adaptation of the flow distribution in circular sessile cells (Moch et al., 2013; Moch et al., 2016; Schwarz et al., 2016; Leube et al., 2017) to the geometry of migrating cells (D-shape). Highest keratin flow is found at the periphery of the cell, while lowest keratin flow is found in proximity to the nucleus. Additionally, keratin flow was shown to depend on the cell trajectory. Higher keratin flow correlates with higher directionality ratio and keratin flow is higher on the lateral back convex side when a cell turns. The asymmetry in keratin flow in turning cells coincides with asymmetric actin flow. This symmetry break of the subcellular keratin flow pattern was also observed for nHEKs migrating on sinusoidal line patterns.

These adaptations of the keratin flow pattern during migration reflect alterations in polarization of migrating cells both at the longitudinal (front/back) and lateral (convex/concave) level. Moreover, differences in keratin flow within a cell may contribute to variations of the local mechanophysical properties of the cell body. Indeed, keratin flow variations are associated to changes in the cell shape, while cell shape as well as keratin filaments have a dominant role in cell stiffness (Tee et al., 2010; Ramms et al., 2013).

2. Impact of the mechanophysical environment

In standard culture conditions (fibronectin-coated glass, 2.5 µg.cm-2), higher keratin flow was shown to be associated to higher migration speed. I aimed at understanding how keratin dynamics is affected by changes in the mechanophysical environment of the cells during migration. Mechanical cues were chosen to be physiologically relevant for epithelial cell migration. Additionally, altering the environment of cells allowed me to change migration features such as speed and directionality without disrupting any other cytoskeletal components and avoiding off-target effects that often hamper drug inhibition and genetic modification experiments.

92

When changing the mechanophysical environment of cells, the migration speed was altered as well as the keratin flow. Higher coating density and higher substrate stiffness were shown to induce lower migration speed as previously shown for other cell types (Doyle et al., 2013; Gupta et al., 2015; Palecek et al., 1997). In both cases, an increase in migration speed was associated with a higher keratin flow. In each situation, a strict correlation was seen between migration speed and speed of keratin flow, independent of the mechanophysical environment. This suggests that both processes are mostly under control of the same mechanosensing mechanisms. Still, some mild differences in the keratin flow distribution could be observed at the subcellular level that may probably be attributed to differences in perception of the environment (differences in substrate stiffness versus differences in coating density) by the cell-matrix adhesions. Indeed, the forces needed at the front versus at the back of the cell to protrude and assemble new adhesions versus to disassemble adhesions may be different in these two situations (Zhong and Ji, 2013), and therefore induce differences in the ratio of keratin flow at the front and at the back of the cell.

When nHEKs were restricted to 15 µm-wide stripes, both their migration speed and keratin flow decreased as expected. However, for a given migration speed, the associated keratin flow was lower than in the conditions described in the previous paragraph. This observation suggests that additional migration-independent restrictions act on keratin dynamics in pseudo-confinement, probably involving distinct pathways. Also, in this special situation, the cell shape is drastically modified, and most likely the cell stiffness as well (Tee et al., 2010).

Keratin-free murine keratinocytes (Kröger et al., 2013) were used to study the contribution of keratins to the regulation of migration speed. These cells are also spontaneously migratory, but they have more diverse morphologies than nHEKs. Some cells have the same D-shape morphology while others have a more fibroblast-like shape. They tend to grow large protrusions and to switch direction of migration more often.

93

Examining these keratin-free mouse keratinocytes showed that the presence of keratin had only a mild effect on cell migration. It limits migration of cells on fibronectin-coated glass (confirming the results of Seltmann et al., 2013) but has no obvious major regulatory function for pseudo-confinement-dependent or substrate stiffness-dependent migration speed. Interestingly, Zarkoob et al. (2018) used keratin type I KO keratinocytes to investigate if the effect of substrate stiffness on colony formation is dependent on the keratin network. The ability of individual cells to join to form multicellular groups was reduced on soft gels in KO cells while it was increased in WT cells. These results go in the same direction as mine, but differences of are much stronger. This may be due to an effect of neighbouring cells, as well as to the use of different cells (type I versus type II KO). Still, these results suggest that keratin may mildly modulate the effects of actin on migration.

3. Actin as an upstream regulator of keratin dynamics

Previous evidences (e.g., the coordination in the assembly of focal adhesion, hemidesmosomes, adherens junctions and desmosomes at the front of epithelial cell sheet (Hopkinson et al., 2014; Pastar et al., 2014)) were also pointing out a certain interdependence between the actin and the keratin cytoskeleton. I was therefore interested in the variations in actin flow that could be associated with the variations in keratin flow I had observed in different environments. The increase in focal adhesion density on stiff substrates as well as on densely-coated substrates indicated a decrease in actin flow and therefore in migration speed in these conditions. Indeed, focal adhesion distribution affects actin flow (Möhl et al., 2012) and subsequently migration speed (Maiuri et al., 2015). In this context, variations in keratin flow were associated to variations in actin flow, and the actin system was implicated as a major upstream regulator of cell migration speed during changes in the mechanophysical environment that could be modulated by keratin.

One could hypothesize that keratin dynamics is regulated downstream actin dynamics. To test this hypothesis, I investigated a possible correlation between actin and keratin

94 flow patterns at the subcellular level in migrating doubly transfected cells. The spatial distribution of actin and keratin flows showed remarkable similarities with highest flow found in the cell periphery for both cytoskeletal networks. Actin flow shows a strong decrease when it reaches the zone of detection of keratin. In the very front of the cell, actin flow is particularly high, however in the area where both keratin and actin are present, keratin flow is only slightly slower than actin flow. The delay between actin and keratin due to their mild speed difference seems to decrease at higher speed. These results strongly support the notion that actin is an upstream regulator of keratin dynamics during migration.

The keratin flow pattern reflects the cell migratory behaviour and is regulated downstream of actin dynamics. In all the paradigms considered in this study, increased keratin flow correlates with higher migration speed. Still, the small delay of keratin compared to actin suggests that keratin acts as a break affecting the pro-migratory cues of actin mediated by the propulsive acto-myosin contractility. This counterbalance effect on actin’s action is possible due to the special organisation of both networks in migrating cells. Thus, actin and keratin networks are localized in separate layers of the cytoplasm with limited connectivity (Quinlan et al., 2017). Also, they anchor to different cell-matrix adhesions: focal adhesions for the actin network, and hemidesmosomes for keratin network (Geiger and Yamada, 2011; Hopkinson et al., 2014; Walko et al., 2015). The mechanical coupling of these two biophysically very different cytoskeletal networks allows harmonization of actin flow and consequently an increase in migratory persistence. Another evidence of coupling between cytoskeletal components has been shown for vimentin and microtubules during migration of retinal pigment epithelial cells (Gan et al., 2016). Nevertheless, the mechanism described for vimentin and microtubules does not seem to apply for actin and keratin. Indeed, vimentin filaments template the older microtubules filaments so that the new ones can form following these tracks. This way, vimentin insures the persistence of cell polarity over a longer time frame than microtubules that have comparatively a shorter lifetime. However, the resemblance between vimentin and microtubules in the architecture of the networks is striking,

95 whereas actin and keratin networks show very different morphologies. Still, the interplay between actin and keratin dynamics is reminiscent of the observations made in sessile cells where keratin motility was shown to rely on a functional actin network (Wöll et al., 2005).

Taken together, these results suggest the existence of multiple biochemical and steric interactions between the keratin and acto-myosin cytoskeleton, possibly implicating their associated cell-matrix adhesions namely hemidesmosomes and focal adhesions.

4. Highly ordered chevron-like hemidesmosomal structures

Subsequently, I aimed at understanding hemidesmosomal organization and dynamics in relation to focal adhesions during migration of nHEKs.

I show that hemidesmosomes and focal adhesions arrange into a hitherto unknown highly ordered chevron-like structure. The hemidesmosomal chevron-pattern with intercalated focal adhesions were observed in three different paradigms involving polarized cells: during cell migration, during cell spreading, and for cells seeded on micropatterns. However, the specific chevron-type arrangement was absent in confluent non-polarized cells.

These chevron structures are very static during migration. Assembly takes place at the leading front of the cells, while disassembly occurs at the rear end. Between assembly and disassembly, they remain in place, although size changes might happen, and intrinsic polypeptide turnover may occur throughout.

These characteristics differ from previous reports both in terms of organization and dynamics. Cross talk between focal adhesions and hemidesmosomes has been investigated in migrating 804B rat urothelial cells and in immortalized human keratinocytes (Tsuruta et al., 2003; Spinardi et al., 2004; Ozawa et al., 2010). In these studies, sequential occupation of former focal adhesions sites by hemidesmosomes were

96 observed. Yet, no such phenomenon could be observed in my conditions. I could not detect any movement of focal adhesions or hemidesmosomes with respect to the substrate. Furthermore, focal adhesions and hemidesmosomes remained in place without detectable exchange between both. I can therefore distinguish two different models: the “substitution model” proposed by Tsuruta et al. (2011), and the “linkage model” based on my observations. In the “substitution model”, focal adhesions prepare a specialized membrane domain to be subsequently occupied by hemidesmosomes. α6/β4-integrins would interact with actin filaments before hemidesmosomes are formed (Tsuruta et al., 2011; Hiroyasu et al., 2016). In the “linkage model”, hemidesmosomes are formed in close proximity of focal adhesions. No interaction of β4-integrins with actin was found; but instead keratin intermediate filaments were shown to link immediately to nascent hemidesmosomes. Discrepancies between these two models may be explained by differences in the cell systems used. In transformed or immortalized epithelial cells, the formation of focal adhesions-decorated hemidesmosomal chevrons was not observed. The “cat-paw” pattern of hemidesmosomes detected in 804G cells (Tsuruta et al., 2003) hardly resembles the aligned chevrons described here. The fully mature focal adhesions-hemidesmosomes chevron structure might be needed to form stable cell-ECM connections, otherwise slippage may occur and no persistent migration is observed.

5. Cross-talks between hemidesmosomes and focal adhesions

The complex and highly ordered geometry of the arrangement of hemidesmosomes and focal adhesions suggested that they might affect each other during migration. A crucial outcome of my experiments is that the relationship between focal adhesions and hemidesmosomes is not purely hierarchical but mutual: the cross-talk is bidirectional. Focal adhesions are a prerequisite for the formation of chevron hemidesmosomal patterns, but also hemidesmosomes are necessary for focal adhesion maturation and localization. The cross-talk may involve different mechanisms: - Steric facilitation: focal adhesions may locally bring the adjacent plasma membrane close enough to the extracellular matrix to allow the clustering α6/β4 integrins and subsequently the formation of hemidesmosomes.

97

- Ligand competition: focal adhesions and hemidesmosomes can associate with common compounds (e.g., α3/β1 and α6/β4 integrins compete for the extracellular matrix-ligand laminin-332, the cytoskeletal cross-linker plectin has both a hemidesmosome-binding domain and an actin-binding domain that cannot be used simultaneously (Geerts et al., 1999; Litjens et al., 2006), and both focal adhesions and hemidesmosomes may interact with actin through alpha- 1 and 4 (Gonzalez et al., 2001; Hamill et al., 2013; Hamill et al., 2015)). These different possible combinations of association may induce specific local arrangements at the plasma membrane region where the individual components may be sorted locally according to their binding strength to each other associate component. - Mechanical linkage: Structural proteins such as plectin may bridge the border between focal adhesions and hemidesmosomes. - Signaling: Recent studies have shown that mechanical signals are involved in hemidesmosome formation (Osmani et al., 2018). Such mechanisms may involve focal adhesions as primary mechanosensors. Additionally, the focal adhesion proteins kindlin and syndecans are implicated in the regulation of hemidesmosome formation (Larjava et al., 2008; Wang et al., 2010; Hopkinson et al., 2014). In addition, the α6/β4 integrin-dependent activation of the Akt-mTOR pathway has been shown to affect the translation of α3 integrin (Kligys et al., 2007); so, signaling mechanisms may affect not only the relative localization but also the dynamics of focal adhesions and hemidesmosomes during migration.

An interesting consideration in the formation of chevron patterns is the different affinity to fibronectin and laminin-332 of focal adhesions and hemidesmosomes. The focal adhesion component α5/β1 integrin binds preferentially to fibronectin whereas the hemidesmosomal component α6/β4 integrin is not able to bind to fibronectin but only to laminin-332 (Jones et al., 1998). Since only a coating suitable for focal adhesions formation was provided in my experiments, nHEKs must have secreted laminin-332 to support the formation of hemidesmosomes. This fits with the idea, the matrix is

98 continuously remodelled during wound healing throughout degradation and deposition of ECM proteins (Goldfinger et al., 1999; Nguyen et al., 2000). Additionally, migrating cells leave membranous macroaggregates behind them that can be used as tracks for following cells (Geuijen and Sonnenberg, 2002; Rigort et al., 2004). Rigort et al. reported the intercalation and strict separation of type I macro-aggregates inherited from disassembled focal adhesions, and type II macro-aggregates inherited from disassembled hemidesmosomes. Consequently, once the cell has moved forward, the extracellular matrix retains traces of the intercalation of focal adhesions and hemidesmosomes that might be used by follower cells.

6. Cell-matrix adhesions as key players in actin-keratin cross-talk

What is the advantage of the ordered focal adhesion-hemidesmosome patterns? Hemidesmosomes stabilize focal adhesions and subsequently modulate actin flow (Möhl et al., 2012), resulting in slower but more persistent migration (Maiuri et al., 2015). Persistence of migration persistence is further reinforced by the polarized geometry of the chevron arrays. Moreover, the synchronised motion of focal adhesions and hemidesmosomes induces a coordination of their associated filaments during migration. Actin fibers are under tension between focal adhesions thereby tightly linking the movement of both through a tunable clutch mechanism (Giannone et al., 2009; Gardel et al., 2010). In contrast, the keratin network is not submitted to such a direct tension by hemidesmosomal anchorage and can therefore move more independently. This may explain why keratin flow can retain a mild delay in comparison to actin flow. The spatial arrangement of the hemidesmosomal chevrons with intercalated focal adhesions may thereby serve as a mean to co-ordinate actin and keratin responses without using direct cross-linkers between both. The coordination of focal adhesions and hemidesmosomes may thereby support coordinated cytoskeletal remodelling and dynamics.

The consequences of losing chevron patterns become evident during epithelial- mesenchymal transition when epithelial cells switch from keratin expression to vimentin expression (Mendez et al. 2010). Vimentin filaments are not known to anchor

99 hemidesmosomes but instead to interact with focal adhesions, (Burgstaller et al., 2010; Havel et al., 2015), even though vimentin can associate with α6/β4 integrin-containing fibrillar structures in endothelial cells (Homan et al., 1998). Therefore, the described system of stable caterpillar tracks allowing slow persistent migration would be impaired upon EMT. This conclusion is supported by observations in KO mouse keratinocytes which present mislocalized hemidesmosomes and compromised migration persistence (Seltmann et al., 2013). On the other hand, altered focal adhesion distribution and actin dynamics may contribute to the observed increase in invasiveness of keratin KO mouse keratinocytes, which may be enhanced by the higher deformability of these cells (Seltmann et al., 2013). It fits with the idea that upon EMT, the loss of epithelial cadherins would be accompanied by the disappearance of chevron patterns; subsequently, the cells are no more tightly associated within the epithelium and can invade the neighbouring tissues: intravasation can occur as a first step in the development of (Niit et al., 2015).

7. Conclusion and future work

What we learn about the complexity of cytoskeletal cross-talk during cell-migration

Using a simplistic model of epithelial migration, namely single primary human keratinocytes migrating in 2D, in combination with powerful tools of quantitative image analysis, I could unravel aspects of the dynamic interplays between the keratin and actin cytoskeleton. Fig. 4.1. schematically summarizes major conclusions of the present study.

Environment sensing is primarily affecting focal adhesion density and distribution. Consequently, actin flow is regulated (Möhl et al., 2012), directly influencing migration speed and persistence (Maiuri et al., 2015). The newly described ordered arrangement of focal adhesions and hemidesmosomes allows coordination of the keratin and actin flow. The mild delay of keratin slows down actin flow and thereby reduces overall migration speed but enhances migration persistence by harmonizing actin flow, ensuring that propulsive cues are continuously moderated and that contractile cues do not induce

100 excessive ripping off of adhesions to avoid cell detachment. Hemidesmosomes affect focal adhesions, typically by stabilizing their orientation, which in turn increases the persistence.

The above mechanisms, which are strengthened through feedback loops, strongly favour persistent migration but only mildly affect migration speed. This is physiologically relevant, since cell migration is only efficient when it is persistent over reasonable distances. This is essential during epithelial morphogenesis and wound healing.

101

Fig. 4.1: A unique chevron pattern involving hemidesmosomes and focal adhesions is at the heart of the cross-talk between the keratin and actin cytoskeleton during epithelial cell migration. The scheme depicts processes highlighted in this work, showing how chevron

102 patterns involving focal adhesions and hemidesmosomes play a central role in cell migration by influencing actin and keratin dynamics.

Future work

Keratin diversity: The expression of different keratin isoforms has different effects depending on the cell type and the environment (Leduc and Etienne-Manneville, 2015). For simplicity’s sake, I have considered the keratin network as a homogenous unit but am aware that the isoform-specific composition of cells within the monolayer, and even the specific composition of individual keratin filament bundles, are likely relevant for cell migration and the associated cell shape changes and mechanically driven processes (Woung and Coulombe, 2003; Chung et al., 2013). To what extend would different ratios of keratin isoforms alter the keratin dynamics, and through what mechanism would it impact cell migration?

Effect of pathologic keratin mutations on migration: The characterization of keratin KO cells has shown that keratins do not play a direct prominent role in migration and its adaptation to the environment. Yet, my observations suggest that keratins contribute in a more general way to environmentally-determined migration by improving persistence of migration. Further analyses are therefore wanted to investigate how altered keratin flow, which is expected to occur in cells with disease-causing keratin mutations, affects migration. In this situation, the keratin flow may lack the synchronization with the actin flow. Indeed, mutant cells with phosphomimetic or phosphodeficient threonine 150 of keratin 5 have an altered keratin network either composed of granules or of thicker filaments with reduced branching, and this correlates with an altered keratin flow in sessile cells (Sawant et al., 2018). One can wonder to what extend it does affect actin flow and the ability of such cells to migrate in a persistent way.

Impact of the pivotal role of lamins in mechanosensation on cytoskeletal dynamics and migration: Lamins are other types of intermediate filaments that play a role in

103 mechanosensation typically when the nucleus is squeezed in confined situations, and in cell migration (Houben et al., 2009). How lamins interfere in this cross-talk between actin and keratin in epithelial cell migration remains unknown. Nesprin-3 binds to intermediate filaments through plectin and to the nuclear membrane through Sun proteins, while nesprin-1 and -2 connect with the actin cytoskeleton (Wilhelmsen et al., 2005). One would test how keratin and actin dynamics are affected in cells with an altered functionality of nesprin-3 and to what extend it does affect cell migration, especially in a confined environment. This would be another contribution to the cross-talk between the actin and keratin networks mediated by components of the nuclear membrane.

Hemidesmosomes-induced forces on the matrix: While focal adhesions are considered to be the main points of force application by the cell on the matrix (Lauffenburger and Horwitz, 1996), no data are available on the forces that hemidesmosomes might apply to the matrix. Both practical aspects (low intensity of the fluorescent signal) and analytical aspects (ability to calculate the forces applied by distinct but so close structures) prevented successful traction force microscopy experiments. I would assume, however, that hemidesmosomes apply only very low forces in comparison to focal adhesions. But these forces may increase substantially, when focal adhesions are disrupted. Hemidesmosomes would act as an emergency system to maintain substrate adhesion in this scenario.

Hemidesmosomes dynamics at the molecular level: To gain insight into the intrinsic dynamics of hemidesmosomes, FRAP analysis need to be performed. The high background caused by the strong integrin signal in the endoplasmic reticulum of the transiently transfected cells was a major problem preventing the success of this experiment. In nHEKs, focal adhesions remain at the same position on the substrate while the cells migrate. However, focal adhesions are highly dynamic at the molecular level, with high rates of turnover and changes in their phosphorylation states upon maturation (Vicente-Manzanares et al., 2009). The composition of hemidesmosomes is way less

104 complex than the one of focal adhesions, but their molecular turnover rates may be as quick.

Dynamics of the keratin-hemidesmosome anchorage: Even though hemidesmosomes have been broadly defined as cell-matrix adhesions that anchor keratin intermediate filaments, very few images of this anchorage are available in the literature. My images suggest that keratin filaments do not end at the anchorage point. Such hypothesis is supported by the images made by super-resolution microscopy in fixed cells (Nahidiazar et al., 2015). Still, the dynamics of keratin intermediate filaments in proximity with hemidesmosomes remains to be investigated: are the anchorage points fixed or sliding, and how do they affect filament turnover and assembly dynamics?

105

Summary

Cell migration is a highly complex process whereby different physical and chemical signals act on many different cell components, most notably the force-generating cytoskeleton.

Intermediate filaments are one of the three major components of the cytoskeleton. Keratin intermediate filaments are the main type of cytoplasmic intermediate filaments of epithelial cells. They are anchored to hemidesmosomes, which are involved in the attachment of epithelial cells to the extracellular matrix of the underlying basement membrane. Although keratin intermediate filaments are involved in the mechanical resilience of tissues, they are highly dynamic structures. They are subject to continuous turnover in sessile keratinocytes. This turnover is part of a spatially-defined cycle of assembly and disassembly. Similarly, hemidesmosomes contribute to the mechanical integrity of the epithelium, but their role in epithelial cell migration remains unclear.

I aimed at understanding how the dynamic behavior of the epithelial keratin intermediate filament cytoskeleton and its associated hemidesmosomes are integrated in migrating cells. Most particularly, I investigated how the distribution and the kinetics of the keratin turnover cycle are affected by cell migration; and how this process is dependent on the mechanophysical environment of the cell. I determined how hemidesmosomes are organized and maintained during migration; and how this process depends on the adjacent actin-associated focal adhesions.

In migrating human primary keratinocytes, I show that keratin is highly dynamic presenting a well-defined spatial distribution with highest flow found at the cell periphery. Keratin dynamics are co-regulated with the speed of cell migration and the cell’s trajectory. Upon changes in the mechanophysical environment, migration features and keratin flow are altered. The changes in keratin flow are most likely downstream of focal adhesion-dependent mechanosensation and actin flow. As a result, keratin dynamics correlate with migration speed. Keratin flow and actin flow show conspicuous similarities in their spatial distribution. The observed dynamic interplay suggests the existence of multiple feedbacks between the actin and keratin cytoskeleton. My

106 observations further suggest that the associated cell-matrix adhesions play a key role in this feedback.

I then show that in migrating cells, hemidesmosomes cluster into highly ordered chevron- like patterns with intercalated focal adhesions in migrating primary human keratinocytes. This arrangement is maintained during migration by continuous assembly of adhesions at the cell front and disassembly of adhesions at the cell back. I further observed that the specialized hemidesmosome-focal adhesion distribution patterns emerge during substrate adhesion and cell spreading within three hours after seeding. During cell adhesion and migration, hemidesmosomes and focal adhesions affect each other’s distribution.

Bidirectional cross-talk between focal adhesions and hemidesmosomes is key for the coordination of the actin and keratin cytoskeleton during cell migration. Mechanical coupling at the cell-matrix adhesion level and at the cytoskeletal network level together induce an increase in migratory persistence. Persistence in cell migration is crucial for efficient migration in a complex 3D environment.

107

References

Akhmanova and Setinmetz, 2008 to chromatin: role of anoxia and photon Tracking the ends: a dynamic protein flux. Nat Biotech. 25: 249-253 network controls the fate of Bershadsky et al., 1996 Involvement of microtubules. Nat Rev Mol Cell Biol. 9: microtubules in the control of adhesion 309-322 dependent signal transduction. Curr Biol. Alam et al., 2011 Loss of keratin 8 and 18 6: 1279-1289 leads to alterations in alpha6beta4- Bordeleau et al., 2010 Keratin 8/18 integrin-mediated signaling and modulation of protein kinase C- decreased neoplastic progression in an mediated integrin dependent adhesion oral-tumor-derived cell line. J. Cell Sci. and migration of liver epithelial cells. 124: 2096-2106 Mol Bio Cell 21: 1698-1713 Albers and Fuchs, 1993 The molecular Borradori and Sonnenberg, 1996 biology of intermediate filament Hemidesmosomes: roles in adhesion, proteins. Int Rev Cytosol. 134: 243-279 signaling and human diseases. Curr Opin Alberts et al., 2012 Molecular Biology of Cell Biol. 8: 647-656 the Cell 5th Edition. Chapter 16 The Borradori and Sonnenberg, 1999 Cytoskeleton. Structure and function of Andra et al., 1998 Not just scaffolding: hemidesmosomes: more than simple plectin regulates actin dynamics in adhesion complexes. J. Invest Dermatol. cultured cells. Genes Dev. 12: 3442-3451 112: 411-418

Azioune et al., 2010 Protein Boudreau et al., 2013 114-3-3sigma micropatterns: a direct printing protocol stabilizes a complex of soluble actin and using deep UVs. Methods in Cell Biology intermediate filaments to enable breast Vol. 97 Chapter 8: 133-146 tumor invasion. Proc Natl Acad Sci. USA 110: 3937-3944 Backert et al., 2013 Transmigration route of Campylobacter Jejuni across polarized Bukoreshtliev et al., 2013 Mechanical intestinal epithelial cells: paracellular, cues in cellular signaling and transcellular, or both? Cell Comm and communication. Cell Tissue Res. 352: 77- Signaling 11: 72-87 94

Bernas et al., 2004 Minimizing Burgstaller et al., 2010 Keeping the photobleaching during confocal vimentin network under control: cell- microcopy of fluorescent probes bound matrix adhesion-associated plectin-1f

108 affects cell shape and polarity of Chung et al., 2013 Networking galore: fibroblasts. Mol Biol Cell 21: 3362-3375 intermediate filaments and cell migration. Curr Opin Cell Biol. 25: 1-13 Burridge, 2017 Focal adhesions: a personal perspective on a half century of Conde and Caceres, 2009 Microtubule progress. FEBS J. 284: 3355-3361 assembly, organization and dynamics in axons and dendrites. Nat Rev Carlier and Pantaloni, 2007 Control of Neuroscience 10: 319-332 actin assembly dynamics in cell motility. J. Biol Chem. 282: 23005-23009 Coulombe et al., 1991 A function for keratins and a common thread among Carter et al., 1990 Distinct functions for different types of epidermolysis bullosa integrins alpha 3 and beta 1 in focal simplex diseases. J. Cell Biol. 115: 1661- adhesions and alpha 6 beta 4/bullous 1674 phemgoid antigen in a new stable anchoring contact (SAC) of Delva et al., 2009 The . Cold keratinocytes: relation to Spring Harb Perspect Biol. 1: a002543 hemidesmosomes. J. Cell Biol. 1111: Deakin et al., 2008 Paxillin comes of age. 3141-3154 J Cell Sci. 121: 2435-2444 Caspi et al., 2001 A new dimension in Dormann and Weijer, 2006 Imaging of retrograde flow: centripetal movement cell migration. The EMBO Journal 25: of engulfed particles. Biophys J. 81: 3480-3493 1990-2000 Doyle et al., 2013 Dimensions in cell Cesa et al., 2007 Micropatterned silicone migration. Curr Opin Cell Biol. 25: 642- elastomer substrates for high resolution 649 analysis of cellular force patterns. Rev Sci Instrum. 78: 34301-34310 Etienne-Manneville, 2013 Microtubules in cell migration. Annu Rev Cell Dev Biol. Chan et al., 1993 The genetic basis of 29: 471-499 Weber-Cockayne epidermolysis bullosa simplex. Proc Natl Acad Sci. USA 90: Evans et al., 1985 Influence of the 7414-7418 centrosome on the structure of nucleated microtubules. J. Cell Biol. 100: Charras and Sahai, 2014 Physical 1185-1191 influences of the extracellular environement on cell migration. Nat Rev Fine et al., 2008 The classification of Mol Cell Biol. 15: 813-824 inherited epidermolysis bullosa (EB): report on the third international Chen et al., 1981 Mechanism of consensus meeting on diagnosis and retraction of the trailing edge during classification of EB. J. Am Acad Dermatol. fibroblast movement. J. Cell Biol. 90: 58: 931-950 187-200

109

Fortier et al., 2013 Keratin 8 and 18 loss Geiger et al., 2001 Transmembrane in epithelial cancer cells increases extracellular matrix-cytoskeleton collective cell migration and cisplatin crosstalk. Nat Rev Mol Cell Biol. 2: 793- sensitivity through claudin1 up- 805 regulation. J. Biol Chem. 288: 11555- Geiger et al., 2009 Environmental 11571 sensing through focal adhesions. Nat Rev Franke et al., 1983 Protein complexes of Mol Cell Biol. 10: 21-33 intermediate-sized filaments: melting of Geuijen and Sonnenberg, 2002 Dynamics cytokeratin complexes in urea reveals of alpha 6 beta 4 integrin in different polypeptide separation keratinocytes. Mol Biol Cell 13: 3845- characteristics. Proc Natl Acad Sci. USA 3858 80: 7113-7117 Giannone et al., 2009 Multilevel Friedl and Wolf, 2010 Plasticity of cell molecular clutches in motile cell migration: a multiscale tuning model. J. processes. Trends in Cell Biol. 19: 475- Cell Biol. 188: 11-19 486 Gan et al., 2016 Vimentin intermediate Giepmans et al., 2006 The fluorescence filaments template microtubule toolbox for assessing protein location networks to enhance persistence in cell and function. Science 312: 217-224 polarity and directed migration. Cell Syst. 3: 500-501 Goldfinger et al., 1999 The alpha 3 laminin subunit, alpha 6 beta 4 and alpha Gardel et al., 2010 Mechanical 3 beta 1 integrin coordinately regulate integration of actin and adhesion wound healing in cultured epithelial cells dynamics in cell migration. Annu Rev Cell and in the skin. J. Cell Sci. 112: 2615- Dev. 26: 315-333 2629 Gardner et al., 2008 Microtubule Gonzalez et al., 2001 Interactions of a assembly dynamics: new insights at the hemidesmosome component and nanoscale. Curr Opin Cell Biol. 20: 64-70 actinin family members. J. Cell Sci. 114: Geerts et al., 1999 Binding of integrin 4197-4206 alpha 6 beta 4 to plectin prevents plectin Green and Jones, 1996 Desmosomes and association with F-actin but does not Hemidesmosomes: structure and interfere with intermediate filament function of molecular components. binding. J. Cell Biol. 147: 417-434 FASEB J. 10: 871-881 Geiger and Yamada, 2011 Molecular Guo et al., 1995 Gene targeting of architecture and function of matrix BPAG1: abnormalities in mechanical adhesions. Cold Spring Harb Perspect strength and cell migration. Cell 81: 233- Biol 3: 1-21 243

110

Gupta et al., 2015 Adaptive rheology and traction forces in motile keratinocytes. ordering of cell cytoskeleton govern FASEB J. 30: 2298-2310 matrix rigidity sensing. Nat Comm. 6: Hoebe et al., 2007 Controlled light- 7525-7544 exposure microscopy reduces Hamill et al., 2013 Actinin-4 in photobleaching and phototoxicity in keratinocytes regulates cell motility via fluorescence live-cell imaging. J. an effect on lamellipodia stability and Microsc. 215: 281-296 matrix adhesions. FASEB J. 27: 546-556 Homan et al., 1998 Endothelial cells Hamill et al., 2015 Alpha actinin-1 assemble two distinct alpha 6 beta 4- regulates cell-matrix adhesion containing vimentin-associated organization in keratinocytes: structures: roles for ligand binding and consequences for skin cell motility. J. the beta 4 cytoplasmic tail. J. Cell Sci. Invest Dermatol. 135: 1043-1052 111: 2717-2728

Hatzfeld and Franke, 1985 Pair formation Hopkinson et al., 2014 Focal contact and and promiscuity of : hemidesmosomal proteins in formation in vitro of heterotypic keratinocyte migration and wound complexes and intermediate sized repair. Advances in wound care 3: 247- filaments by homologous and 263 heterologous recombination of purified Houben et al., 2009 Disturbed nuclear polypeptides. J. Cell Biol. 110: 1199-1210 orientationand cellular migration in A- Havel et al., 2015 Vimentin regulates type deficient cells. Biochemica cancer cell adhesion through a and Biophysica Acta 1793: 312-324 VAV2-Rac1 pathway to control focal Humphries et al., 2007 Vinculin controls adhesion kinase activity. Oncogene 34: focal adhesion formation by direct 1979-1990 interactions with talin and actin. J. Cell Helvert et al., 2018 Mechanoreciprocity Biol. 179: 1043-1057 in cell migration. Nat Cell Biol. 20: 8-20 Hynes, 2009 The extracellular matrix: not Hesse et al., 2001 Genes for intermediate just pretty fibrils. Science 326: 1216- filament proteins and the draft sequence 1219 of the : novel keratin Jaffe and Hall, 2005 RHO-GTPases: genes and a surprisingly high number of biochemistry and biology. Annu Rev. pseudogenes related to keratin gene 8 Cell. Dev Biol. 21: 247-269 and 18. J. Cell Sci. 114: 2569-2575 Jones et al., 1998 Structure and assembly Hiroyasu et al., 2016 A hemidesmosomal of hemidesmosomes. Bioessays 20: 488- protein regulates actin dynamics and 494

111

Kim and Coulombe, 2007 Intermediate Langbein et al., 2005 Characterization of filament scaffolds fulfil mechanical, a novel human type II epithelial keratin organisational, and signalling functions K1b, specifically expressed in eccrine in the cytoplasm. Genes Dev. 21: 1581- sweat glands. J. Invest Dermatol. 125: 1597 428-444

Kligys et al., 2007 The slingshot family of Langbein et al., 2006 K25 (K25irs1), K26 phosphatases mediates Rac1 regulation (K25irs2), K27 (K25irs3), and K28 of cofilin phosphorylation, laminin-332 (K25irs4) represent the type I inner root organization, and motility behavior of sheath keratins of the human hair keratinocytes. J. Biol Chem. 282: 32520- follicle. J. Invest Dermatol. 126: 2377- 32528. 2386

Kreplak et al., 2005 Exploring the Larjava et al., 2008 Kindlins: essential mechanical behavior of single regulators of integrin signalling and cell- intermediate filaments. J. Mol Biol. 354: matrix adhesion. EMBO Rep. 9: 1203- 569-577 1208

Kröger et al., 2013 Keratins control Lauffenburger and Horwitz, 1996 Cell intercellular adhesion involving PKC-α- migration: a physically integrated mediated desmoplakin phosphorylation. molecular process. Cell 84: 359-369 J. Cell Biol. 201: 681-692 Leduc and Etienne-Manneville, 2015 Ku and Omary, 2006 A disease and Intermediate filaments in cell migration phosphorylation-related non- and invasion: the unusual suspects. Curr mechanical function for keratin 8. J. Cell Opin Cell Biol. 32: 102-112 Biol. 174: 115-125 Leduc and Etienne-Manneville, 2017 Ladoux et al., 2016 Front-rear Regulation of microtubule-associated polarization by mechanical cues: from motors drives intermediate filament single cells to tissues. Trends Cell Biol. network polarization. J. Cell Biol. 216: 26: 420-433 1689-1703

Lämmermann and Sixt, 2009 Mechanical Lele et al., 2006 Mechanical forces alter modes of amoeboid cell migration. Curr zyxin unbinding kinetics within focal Opin Cell Biol. 21: 636-644 adhesions of living cells. J. Cell Physiol. 207: 187-194 Langbein et al., 2001 The catalog of human hair keratins II. Expression of the Leube et al., 2011 «Panta Rhei»: six type II members in the hair follicle perpetual cycling of the keratin and the combined catalog of human type cytoskeleton. Bioarchitecture 1: 39-44 I and II keratins J. Biol Chem. 276: 35123- 35132

112

Leube et al., 2017 Intracellular motility of Moch, 2015 Regulation of keratin intermediate filaments. Cold Spring Harb filament network dynamics. Perspect Biol. 9 Moch et al., 2013 Measuring the Leube and Schwarz, 2016 Intermediate regulation of keratin filament network Filaments. Encyclopedia of Cell Biology dynamics. Proc Natl Acad Sci. USA 110: (Bradshaw and Stahl), Vol. 2: 569-578 10664-10669

Leung et al., 2001 The plakin family. J. Moch et al., 2016 Effects of plectin Cell Sci. 114: 3409-3410 depletion on keratin network dynamics and organization. Plos One 11: e0149106 Litjens et al., 2006 Current insight into the formation and breakdown of Möhl et al., 2012 Quantitative mapping hemidesmosomes. Trends Cell Biol. 16: of averaged focal adhesion dynamics in 376-383 migrating cells by shape normalization. J. Cell Sci. 125: 155-165 Lo et al., 2000 Cell movement is guided by the rigidity of the substrate. Biophys Moll et al., 1982 Different keratin J. 79: 144-152 polypeptides in epidermis and other epithelia of human skin: a specific Maciver and Hussey, 2002 The cytokeratin of molecular wright 46,000 ADF/Cofilin family: actin remodeling in epithelia of pilosebaceous tract and proteins. Genome Biol. 3: 3007.1- basal epitheliomas. J. Cell Biol. 95: 285- 3007.12 295 Maiuri et al., 2015 Actin flow mediates a Moll et al., 1993 Special program of universal coupling between cell speed differentiation expressed in and cell persistence. Cell 161: 374-386 keratinocytes of human Haarscheiben: Mandelkow et al., 1986 On the surface an analysis of individual cytokeratin lattice of microtubules: helix starts, polypeptides. J. Invest Dermatol. 100: protofilament number, seam and 69-76 handedness. J. Cell Biol. 102: 1067-1073 Moll et al., 2008 The human keratins: Mendez et al., 2010 Vimentin induces biology and pathology. Histochem Cell changes in cell shape, motility, and Biol. 129: 705-733. adhesion during the epithelial to Nahidiazar et al., 2015 The molecular mesenchymal transition. FASEB J. 24: architecture of hemidesmosomes as 1838-1851 revealed with super-resolution Mitchison and Cramer, 1996 Actin-based microscopy. J. Cell Sci. 128: 3714-3719 cell motility and cell locomotion. Cell 84: Nakamura and Tokura, 2011 Epithelial- 371-379 mesenchymal transition in the skin. J. Dermatol. Sci. 61: 7-13

113

Nguyen et al., 2000 Deposition of laminin Pedersen et al., 2005 Mechanobiology in 5 by keratinocytes regulates integrin the third dimension. Ann Biomed Eng. adhesion and signaling. J. Biol Chem 275: 33: 1469-1490 31896-31907 Pekny and Lane, 2007 Intermediate Niit et al., 2015 Cell-cell and cell-matrix filaments and stress. Exp Cell Res. 313: adhesion in survival and metastasis: 2244-2254 Stat3 versus Akt. Biomol Concepts 6: Peterson, 2010 Encyclopedia of 383-399 movement disorders Osmani et al., 2018 An Arf6- and Peyton and Putnam, 2005 Extracellular caveolae-dependent pathway links matrix rigidity governs smooth muscle hemidesmosome remodeling and motility in a biphasic fashion. J. Cell mechanoresponse. Mol Biol Cell 29: 435- Physiol. 204: 198-209 351 Piccinini et al., 2016 Cell Tracker (not Ozawa et al., 2010 Dynamic relationship only) for dummies. Bioinformatics 32: of focal contacts and hemidesmosomes 955-957 protein complexes in live cells. J. Invest Dermatol. 130: 1624-1635 Quinlan et al., 2017 A rim-and-spoke hypothesis to explain the biomechanical Ozawa et al., 2014 The role of roles for cytoplasmic intermediate hemidesmosomes and focal contacts in filament networks. J. Cell Sci. 130: 3437- the skin visualized by dual-color live-cell 3445 imaging. Med Mol Morphol. 47: 185-188 Radtke and Raj, 2003 The role of Notch in Palecek et al., 1997 Integrin-ligand tumorigenesis: oncogene or tumor binding properties govern cell migration suppressor? Nat Rev Cancer 3: 756-767 speed through cell-substratum adhesiveness. Nat. 385: 537-540 Ramms et al., 2013 Keratins as the main component for the mechanical integrity Palecek et al., 1998 Physical and of keratinocytes. Proc Natl Acad Sci. USA biochemical regulation of integrin 110: 18513-18518 release during rear detachment of migrating cells. J. Cell Sci. 111: 929-940 Raymond et al., 2005 Keratinocytes display normal proliferation, survival and Parsons et al., 2010 Cell adhesion differentiation in conditional beta4- integrating cytoskeletal dynamics and integrin knock-out mice. J. Cell Sci. 118: cellular tension. Nat Rev Mol Cell Biol. 1045-1060 11: 633-643 Riedl et al., 2008 LifeAct: a versatile Pastar et al., 2014 Epithelialization in marker to visualize F-actin. Nat Methods wound healing: a comprehensive review. 5: 605-607 Adv. Wound Care 3: 445-454

114

Rigort et al., 2004 Release of integrin and impact invasive behavior. Proc Natl macroaggregates as a mechanism of rear Acad Sci. USA 110: 18507-18512 detachment during keratinocytes Seltmann et al., 2013 Keratins mediate migration. Eur J Cell Biol. 83: 725-733 localization of hemidesmosomes and Riveline et al., 2001 Focal contacts as repress cell motility. J. Invest Dermatol. mechanosensors: externally applied 133: 181-190 local mechanical force induces growth of Sivaramakrishnan et al., 2008 focal contacts by an mDia1-dependent Micromechanical properties of keratin and ROCK-independent mechanism. J. intermediate filaments networks. Proc. Cell Biol. 153: 1175-1186 Natl Acad Sci. USA 105: 889-884 Sagot et al., 2002 An actin nucleation Spinardi et al., 2004 A dynamic mechanism mediated by Bni1 and podosome-like structure of epithelial . Nat Cell Biol. 4: 626-631 cells. Exp Cell Res. 295: 360-374 Sawant et al., 2018 Threonine 150 Steinert et al., 1983 Complete amino- phosphorylation of keratin 5 is linked to acid sequence of a mouse epidermal epidermolysis bullosa simplex and keratin subunit and implication for the regulates filament assembly and cell structure of intermediate filaments. Nat. viability. J. Invest Dermatol. 138: 627- 302: 794-800 636 Tee et al., 2010 Cell shape and substrate Schäfer et al., 2010 The key feature for rigidity both regulate cell stiffness. early migratory processes. Dependence Biophys J. 100: 25-27 of adhesion, actin bundles force generation and transmission on Thiery et al., 2009 Epithelial- filopodia. Cell Adhesion & Migration 4: mesenchymal transition in development 215-255 and disease. Cell 139: 871-890

Schröder et al., 1999 Immunogold EM Tojkander et al., 2012 Actin stress fibers reveals a close association of plectin and assembly, dynamics and biological roles. the desmin cytoskeleton in human J. Cell Sci. 125: 1855-1864 skeletal muscle. ECJB 78: 288-295 Tsuruta et al., 2003 Hemidesmosomes Schwarz et al., 2016 Multidimensional protein dynamics in live epithelial cells. monitoring of keratin intermediate Cell Motility and the Cytoskeleton 54: filaments in cultured cells and tissues. 122-134 Methods Enzymol. 568: 59-83 Tsuruta et al., 2011 Hemidesmosomes Seltmann et al., 2012 Keratins and focal contact proteins: functions and significantly contribute to cell stiffness cross-talks in keratinocytes, bullous

115 diseases and wound healing. J. Derm Sci. Weaver et al., 2003 Integration of the 62: 1-7 signals to the Arp2/3 complex. Curr Opin Cell Biol. 15: 23-30 Underwood et al. 2009 Ultrastructural localization of integrin subunits β4 and Webb et al., 2004 FAK-Src signalling α3 within the migrating epithelial tongue through paxillin, ERK and MLCK regulates of in vivo human wounds. J. Histochem adhesion disassembly. Nat Cell Biol. 6: Cytochem. 57: 123-142 154-161

Valencia et al., 2013 Intermediate Weisenberg et al., 1968 Tubulin- filament-associated cytolinker protein nucleotide interactions during the plectin 1c destabilizes microtubules in polymerization and depolymerization of keratinocytes. Mol Biol Cell 24: 768-784 microtubules. Biochemistry 15: 4248- 4254 Vallotton et al., 2004 Simultaneous mapping of filamentous actin flow and Welch et al., 2015 Cell migration, freshly turnover in migrating cells by squeezed. Cell 160: 581-582 quantitative fluorescent speckle Wiche et al., 2015 Networking and microscopy. Proc Natl Acad Sci. USA 101: anchoring through plectin: a key to IF 9660-9665 functionality and mechanotransduction. Vicente-Manzanares et al., 2005 Cell Curr Opin Cell Biol. 32:21-29. migration at a glance. J. Cell Sci. 118: Wilhelmsen et al., 2005 Nesprin-3, a 4917-4919 novel outer nuclear Vicente-Manzanares et al., 2009 Non- associates with the cytoskeletal linker muscle myosin-II takes centre stage in protein plectin cell adhesion and migration. Nat Rev Mol Windoffer et al., 2011 Cytoskeleton in Cell Biol. 10: 778-790 motion: the dynamics of keratin Wang et al., 2010 Interaction of intermediate filaments in epithelia. J. syndecan and alpha 6 beta 4 integrin Cell Biol. 194: 669-678 cytoplasmic domains: regulation of Wöll et al., 2005 Dissection of keratin ErbB2-mediated integrin activation. J. dynamics: different contributions of the Biol Chem. 285: 13569-13579 actin and microtubule systems. Euro. J. Walko et al., 2015 Molecular Cell. Biol. 84: 311-328 architecture and function of the Wong and Coulombe, 2003 Loss of hemidesmosome. Cell Tissue Res. 360: keratin 6 (K6) proteins reveals a function 429-544 of intermediate filaments during wound Watanabe and Higashida, 2004 Formins: repair. J. Cell Biol. 2003: 327-337 processive cappers of growing actin filaments. Exp Cell Res. 301: 16-22

116

Yeung et al., 2005 Effects of substrate Zaidel-Bar et al., 2007 Functional atlas of stiffness on cell morphology, integrin adhesome. Nat Cell Biol. 9: 858- cytoskeletal structure, and adhesion. 867 Cell Motility and the Cytoskeleton 60: Zarkoob et al., 2018 Mouse keratinocytes 24-34 without keratin intermediate filaments Zaidel-Bar and Geiger, 2010 The demonstrate substrate stiffness switchable integrin adhesome. J. Cell Sci. dependent behaviors. Cellular and 123: 1385-1388 Molecular Bioengineering 11: 163-174

Zaidel-Bar et al., 2007 A paxillin tyrosine Zhong and Ji, 2013 Impact of cell shape phosphorylation switch regulates the on cell migration behaviour on elastic assembly and form of cell-matrix substrates. Biofabrication 5: 1-10 adhesions. J. Cell Sci. 120: 137-148 Zigmond, 2004 Beginning and ending an actin filament: control at the barbed end. Curr Top Dev Biol. 63: 145-188

117

List of Figures

Fig. 1.1: Influence of the environment on cell migration 8 Fig. 1.2: Elongation of actin filaments 11 Fig. 1.3: Focal adhesions composition 12 Fig. 1.4: Microtubules composition and mode of assembly 13 Fig. 1.5: Organization and mode of assembly of intermediate filaments 15 Fig. 1.6: Main keratin isoforms expressed in the skin 17 Fig. 1.7: Keratin turnover cycle in a single sessile cell 18 Fig. 1.8: Hemidesmosomes composition 21 Fig. 1.9: Cycle of protrusion and retraction cycle during cell migration 23

Fig. 2.1: Definitions of speed, flow and directionality 39 Fig. 2.2: CMove programme routine for keratin flow measurements 40 Fig. 2.3: CMove programme routine for parallel measurements of keratin and 41 actin flow Fig. 2.4: Areas of interest in normalized cells 42

Fig. 3.1: Migrating normal human epidermal keratinocytes can be used to 46 measure keratin filament dynamics Fig. 3.2: The keratin flow in migrating normal human epidermal keratinocytes has 49 a well-defined spatial distribution Fig. 3.3: Higher migration speed correlates with increased keratin flow 51 Fig. 3.4: Higher directionality ratio of migration correlates with increased keratin 53 flow Fig. 3.5: Change in direction of migration induces symmetry break in keratin flow 55 patterns Fig. 3.6: Increased ECM coating density induces a decrease in migration speed 58 and keratin flow Fig. 3.7: Decreased substrate stiffness induces an increase in migration speed and 61 keratin flow Fig. 3.8: Confinement impedes migration speed and keratin flow 63 Fig. 3.9: Mechanophysical-dependent regulation of migration is only mildly 66 affected by keratins Fig. 3.10: Actin and keratin interact dynamically during migration 69 Fig. 3.11: Asymmetry of actin flow in turning normal human epidermal 71 keratinocytes Fig. 3.12: Hemidesmosomal proteins localize to arrays of chevron-shaped 73 structures in migrating normal human epidermal keratinocytes

118

Fig. 3.13: Hemidesmosomal chevron patterns anchor keratin intermediate 74 filaments in migrating normal human epidermal keratinocytes Fig. 3.14: Focal adhesions intercalate between hemidesmosomal chevrons in 75 migrating cells Fig. 3.15: Focal adhesions are sandwiched in between hemidesmosomes in 77 normal human epidermal keratinocytes seeded on X- or D-micropatterns Fig. 3.16: Spatial organization and dynamics of focal adhesions and 79 hemidesmosomes are coordinated during cell migration Fig. 3.17: Chevron-like hemidesomosomes are formed next to focal adhesions 82 and in association with keratin filaments during substrate adhesion Fig. 3.18: Focal adhesions disruption affects hemidesmosomal chevrons 84 formation during substrate adhesion Fig. 3.19: Formation of hemidesmosomes and focal adhesions is impaired by 86 treatment with blocking anti-β4 or anti-α6 integrin antibodies during substrate adhesion Fig. 3.20: Hemidesmosomal chevrons persist upon focal adhesion disruption 88

Fig. 4.1: A unique chevron pattern involving hemidesmosomes and focal 102 adhesions is at the heart of the cross-talk between the keratin and the actin cytoskeleton in epithelial cell migration

119

List of Tables

Table 1.1: Intermediate filaments nomenclature 14

Table 2.1: Supplements for DMEM/Ham’s F12 cell culture medium 28 Table 2.2: Composition of the HSB and LSB buffers 29 Table 2.3: Composition of the 5 x SDS sample buffer (based on Laemmli buffer) 29 Table 2.4: Composition of the resolving gel and the stacking gel 30 Table 2.5: Composition of the 10 x SDS running buffer 31 Table 2.6: Composition of the immunoblot transfer buffer 31 Table 2.7: List of primary antibodies used for immunobotting 32 Table 2.8: Composition of TBS-T solution (pH adjusted to 7.6) 32 Table 2.9: List of secondary antibodies used for immunobotting 32 Table 2.10: List of primary antibodies used for immunohistofluorescence 34 Table 2.11: List of secondary antibodies used for immunohistofluorescence 34 Table 2.12: Silicon elastomer composition and associated elastic modulus 36 Table 2.13: Parameters used for keratin and actin flow measurements with 38 Cmove programme

120

List of Movies

Movie 1: K5-YFP nHEK migrating on fibronectin-coated glass (corresponding Fig. 3.1 D) Laser confocal microscopy images (objective 63x, imaging frequency 1 image.min-1) of a K5-YFP nHEK migrating on fibronectin-coated glass (2.5 µg.cm-2) for 30 minutes. The brightfield and the green fluorescent channels are shown.

Movie 2: K5-YFP nHEK migrating on a fibronectin-coated sinusoidal line pattern (corresponding Fig. 3.5 B) Laser confocal microscopy images (objective 63x, imaging frequency 1 image.min-1) of a K5-YFP nHEK migrating on fibronectin-coated sinusoidal line pattern for 120 minutes. The brightfield and the green fluorescent channels are shown.

Movie 3: K5-YFP nHEK migrating on a fibronectin-coated stripe (corresponding Fig. 3.8 A) Laser confocal microscopy images (objective 63x, imaging frequency 1 image.min-1) of a K5-YFP nHEK migrating on a fibronectin-coated 15 µm wide stripe for 30 minutes. The brightfield and the green fluorescent channels are shown.

Movie 4: nHEK K5-YFP LifeAct-RFP migrating on fibronectin-coated glass (corresponding Fig. 3.10) Laser confocal microscopy images (objective 63x, imaging frequency 2 images.min-1) of a nHEK K5-YFP LifeAct-RFP migrating on fibronectin-coated glass (2.5 µg.cm-2) for 30 minutes. The two fluorescent channels (green and red) are shown.

Movie 5: Monitoring of 4 integrin and paxillin in a normal human keratinocyte migrating on fibronectin-coated glass (corresponding Fig. 3.16 A) The movie shows laser confocal time-lapse fluorescence microscopy (63x objective) of a migrating nHEK producing fluorescent 4 integrin (construct GFP-hb4; green) and paxillin (construct Paxillin-DsRed2; red). Images were recorded every 30 s for 50 minutes.

Movie 6: Monitoring of 4 integrin and paxillin in the leading edge of a normal human keratinocyte migrating on fibronectin-coated (corresponding Fig. 3.16 A’, extracted from Movie 6) The high magnification images correspond to a region taken from the leading edge of the cell shown in movie 1. The movie shows laser confocal time-lapse fluorescence microscopy (63x objective) of a migrating nHEK producing fluorescent 4 integrin (construct GFP-hb4; green) and paxillin (construct Paxillin-DsRed2; red). Images were

121 recorded every 30 s for 50 minutes. Note the growth of the chevron pattern toward the leading edge of the cell.

122

List of Abbreviations

iD i-dimensions (i = 1, 2, 3) Ki Keratin Isoform i Ab Antibody KIF Keratin Intermediate Filament ANOVA Analysis of Variance KO Knockout APS Ammonium Persulfate KtyII-/- Keratin type II Knockout Mouse ARP Actin-Related Proteins Keratinocyte BP- Bullous Phemgoid Antigen LSB Low Salt Buffer BSA Bovine Serum Albumine LSM Laser Scanning Microscope CCD Charged Coupled Device MIP Maximum Intensity Projection CDi Cluster of Differentiation Mo-Ab Monoclonal Antibody Cryo-SFM Cryopreservation Medium for MTOC Microtubules Organizing Center nHEKs N.A. Numerical Aperture DAPI 4',6-diamidino-2-phenylindole nHEK Normal Human Epidermal DIC Differential Interference Contrast Keratinocyte DMEM Dulbecco’s Modified Eagle’s PAGE Polyacrylamide Gel Medium Electrophoresis DNA Deoxyribonucleic Acid PAK p21-activated Kinases DsRed Discosoma Red Fluorescent PBS Phosphate-Buffered Saline Protein PDMS Polydimethylsiloxane DTT Dithiothreitol PFA Paraformaldehyde EBS Epidermolysis Bullosa Simplex PLL-g-PEG Poly(L-lysine) grafted and ECM Extracellular Matrix poly(ethylene glycol) EDTA Ethylenediaminetetraacetic Acid PMSF Phenylmethylsulfonyl Fluoride EGF Epidermal Growth Factor PVDF Polyvinylidene Fluoride EMT Epithelial-Mesenchymal Transition RES K5 Rescue (p)EYFP Enhanced Yellow Fluorescent RFP Red Fluorescent Protein Protein ROCK Rho-associated Protein Kinase FA Focal Adhesion RT Room Temperature [Fn] Fibronectin Concentration SAC Stable Anchoring Complex FRAP Fluorescent Recovery after SD Standard Deviation Photobleaching SDS Sodium Dodecylsulfate GFAP Glial Fibrillary Acidic Protein TBS-T Tris-Buffered Saline, 0.1% Tween GFP Green Fluorescent Protein 20 GTP Guanosine Triphosphate TGF Transforming Growth Factor HD Hemidesmosome TEMED Tetramethylethylene Diamine HRP Horseradish Peroxidase SIM Structured Illumination Microscopy HSB High Salt Buffer TGF Transforming Growth Factor

123

ULF Unit Length Filament UV Ultraviolet WT Wild Type w/v Ratio of Weight out of Volume YFP Yellow Fluorescent Protein Y-27632 (1R,4r)-4-((R)-1-aminoethyl)-N- (pyridin-4-yl)cyclohexanecarboxamide (ROCK inhibitor)

124

Acknowledgements

This project has been funded by the European Commission through a Marie Skłodowska Curie-Innovative Training Network of the Horizon 2020 programme (grant agreement No. 642866; Integrated Component Cycling in Epithelial Cell Motility (InCeM); incem.rwth- aachen.de).

I would like to express my sincere gratitude to Prof Rudolf Leube for his supervision during my doctoral thesis. Both your guidance and the freedom you let me conducting my research made this all a very nice and useful experience.

I would also like to thank Prof Björn Kampa for accepting to review my work and for the yearly guidance.

I would like to thank Prof Reinhard Windoffer for his expertise on microscopy techniques and for the related image analysis skills.

A warm thank you to all members of the keratin subgroup for their fruitful ideas when discussing my projects and for the technical help: Dr Nicole Schwarz, Dr Marcin Moch, Dr Mugdha Sawant, Nadieh Kuijpers and Sonja Lehmann. But also thanks to Prof. Irmgard Classen-Linke, Dr Claudia Krusche, Dr Sebastian Kant, Dr Florian Geisler, Dr Volker Buck, Richard Coch, and all the other medical students in MOCA for your help and advices. You were all responsible for the nice atmosphere in the lab during all this time!

I would also like to thank all the technical assistants in MOCA. Most particularly Claudia Schmitz for the micropatterning, Sabine Eisner for the electron microscopy, Sabina Hennes-Mades for the cell culture and Ursula Wilhelm for the maxi-preps. And also Barbara Bonn, Christine Eherer, Marina Lürkens-Weber. A big thank you to Elke Broekmeulen for all the administrative help, to Adam Breitscheidel for his graphic design skills, and to Felix Meuer, Mykhaylo Anokhin and Justus Schwarzott for solving all computer problems.

I deeply thank people in the ICS-7 in the Research Center Jülich, in particular Prof Rudolf Merkel and Dr Bernd Hoffmann for their helpful guidance and for giving me access to techniques not available in the MOCA Institute. In particular, thank you to Nico Hampe for the help with elastomeric substrates and traction force microscopy, to Nils Hersch for the RICM microscopy, and to Georg Dreissen for the incredible help for image analysis thanks to the CMove programme.

125

I cannot finish before I have thanked all members of the InCeM network. First, I would like to acknowledge the European Commission for the funding and most particularly for giving me the opportunity to do my PhD in these excellent conditions. Thank you so much to all supervisors for their extremely helpful comments and their fruitful discussions, especially to Prof. Perihan Nalbant, Prof. Alexander Bershadsky, Prof. Benny Geiger, Prof. Anotida Madzvamuse, and Prof. Stephanie Portet. And also to all students: Lea, Nikos, Kritika, Galia, Laura, Tina, Rutuja, Katerina, Roman, Victor, Shore, Dmytro and Davide. It has been such a great pleasure to work with you!

126

Curriculum Vitae

Anne PORA, Ingénieur, Master

Date of birth: 22.02.1993 French Citizenship

Professional Experience

Molecular and Cellular Anatomy Institute, University Hospital Aachen (Germany) | PhD Fellow | 3 years (2015-2018)

Impact of keratin network regulation on cell migration.

Sterol Metabolism and Therapeutic Innovations in Oncology Lab, Cancer Research Center Toulouse (France) | Research Intern | 6 months (2015)

Mechanism of action of a metabolite of cholesterol responsible for proliferation and resistance to hormonotherapy in breast cancer.

Biomedical Ultrasonics, Biotherapy and Biopharmaceuticals Lab, University of Oxford (UK) | Research Intern | 4 months (2014)

Development of a focused-ultrasound stimuli responsive gene delivery system using adenoviral vectors.

Guerbet Group, Research Pole for Imaging and Biology, Aulnay-Sous-Bois (France) | R&D trainee | 6 months (2013)

Optimization of tests for the preclinical phases of MRI contrast agents for brain tumors.

Award

ESPCI Alumni Fellowship 2014-2015

Awarded to 4 students for the excellence of their professional project and educational accomplishment.

127

Education

PhD fellow in Biology | RWTH Aachen University (Germany) | July 2015 – July 2018 Within the prestigious international and highly interdisciplinary Marie Sklodowska Curie- Initiative Training Network InCeM (Integrated Component Cycling in Epithelial Cell Motility)

MSc in Medical Biology, specialized in Oncology | Toulouse University (France) | (2014- 2015)

Master awarded the grade A+ by the National Agency in charge of the evaluation of higher education programs

Engineering Degree, Physical-chemistry major | ESPCI ParisTech (France) | (2011-2015)

Top 5 gradudate engineering school in France. “Ingénieur” title obtained in 2014, equivalent to a MSc in Physics and Chemistry. Advanced Master in Science and Technology obtained in 2015.

Classes Préparatoires | Lycée Pierre de Fermat, Toulouse (France) | (2009-2011)

Highly intensive higher education program in Maths, Physics and Chemistry, to prepare the competitive exams for top class graduate engineering schools.

Baccalauréat Science Major with First Class Honors | June 2009

Scientific major, specialty in mathematics. European Section awards (intensive English courses and one subject taught in English).

Languages

French: Mother tongue English: Fluent Spanish: Intermediate German: Basic daily conversation

128

Bibliography

Publications

Hemidesmosomes and focal adhesions form treadmilling arrays in migrating primary keratinocytes, Anne Pora, Reinhard Windoffer, Rudolf E. Leube [In revision, J. Invest Dermatol. 2018] Regulation of keratin network dynamics by the mechanophysical environment in migrating cells, Anne Pora, Georg Dreissen, Bernd Hoffmann, Rudolf Merkel, Reinhard Windoffer, Rudolf E. Leube [In preparation]

Conference Presentations

Impact of the mechanophysical environment on keratin network organization in migrating cells Oral communication during the Cell Migration Session of the 2nd Joint Meeting of the French Society for Cell Biology and the French Society for Developmental Biology (Lyon (France), 26th-29th April 2017)

Mechanophysical regulation of the keratin network in migrating keratinocytes Oral communication during the European Conference on Intermediate Filaments (St Malo (France), 14th-17th June 2017).

Mechanophysical regulation of the keratin network in migrating cells Oral communication during the International Symposium on Measuring and Modelling Cell migration (Vienna (Austria), 22nd-23rd Feb 2018)

Posters

Impact of the mechanophysical environment on keratin network organization in migrating cells, Anne Pora, Bernd Hoffmann, Rudolf Merkel, Reinhard Windoffer, Rudolf E. Leube presented during:

- The 623 WE-Heraus Seminar on Cellular Dynamics (Physikzentrum Bad Honnef (Germany), 4th-7th Sept 2016) - The Annual Meeting of the German Society for Biomaterials (Aachen (Germany), 28th Sept-1st Oct 2016)

129

Mechanophysical regulation of the keratin network in migrating keratinocytes, Anne Pora, Bernd Hoffmann, Rudolf Merkel, Reinhard Windoffer, Rudolf E. Leube presented during the European Conference on Intermediate Filaments (St Malo (France), 14th-17th June 2017).

Keratin dynamics in response to changes in the mechanophysical environment during cell migration, Anne Pora, Bernd Hoffmann, Rudolf Merkel, Reinhard Windoffer, Rudolf E. Leube presented during the International Symposium on Measuring and Modelling Cell Migration (Vienna (Austria), 22nd-23rd Feb 2018)

Hemidesmosomes in migrating cells: organisation, dynamics, and interdependence with focal adhesions, Anne Pora, Reinhard Windoffer, Rudolf E. Leube presented during the International Symposium on Measuring and Modelling Cell Migration (Vienna (Austria), 22nd-23rd Feb 2018)

130