Durant mon travail de thèse, j'ai étudié le rôle central de l'ADF/cofiline, une protéine qui se lie au cytosquelette d'actine, décore spécifiquement les parties ‘âgées' des filaments d'actine, diminue localement par un facteur 5 la rigidité du filament et provoque la fragmentation du filament à l'interface entre les sections nues et décorées. Dans ma première étude (Suarez et al., Current Biology, 2011), j'ai utilisé la microscopie à onde évanescente et une ADF/cofiline fluorescente pour démontrer que l'ADF/cofiline est un marqueur de l'état nucléotidique (ATP, ADP-Pi ou ADP) des sous-unités d'un filament d'actine en cours de polymérisation. De plus, l'ADF/cofiline, en accélérant la dissociation du phosphate inorganique (Pi), limite la taille du cap ATP/ADP-Pi du filament d'actine, sans toutefois le réduire à une taille zéro. Des analyses statistiques sur filaments isolés établissent une corrélation parfaite entre la densité de fixation de l'ADF/cofiline et son efficacité de fragmentation. Paradoxalement, l'efficacité de fragmentation est maximale pour une densité d'ADF/cofiline de 0.5. Ceci est confirmé par des analyses supplémentaires qui montrent que les sites de fragmentation du filament coïncident avec la position des frontières entre zones décorées et zones nues. Les conséquences de ce dernier résultat paradoxal sont l'objet de ma seconde étude (McCullough et al., 2011, Biophysical Journal). En combinant différentes sources d'ADF/cofilines (vertébré et levure) et d'actines (vertébré et levure), nous montrons, sur les quatre couples actine-ADF/cofiline possibles, qu'il existe une très forte corrélation entre (1) l'efficacité de fragmentation (qui dépend de la combinaison entre actine et ADF/cofiline) et (2) la déformation du filament, mesurée à la frontière entre zone décorée et zone nue. Au cours de ma troisième étude (Reymann et al., Molecular Biology of the Cell, 2011), nous montrons que le mécanisme de fragmentation ADF/cofiline-dépendant, établi à l'échelle d'un filament isolé, peut s'appliquer aussi à l'échelle d'une comète d'actine qui comporte un réseau complexe de filaments. Mon travail de thèse a montré que le mode d'action de l'ADF/cofiline se situe à l'intersection entre mécanismes microscopiques et macroscopiques, d'une part, et entre chimie et physique, d'autre part. Les caractéristiques microscopiques des interactions de cette protéine avec un filament d'actine isolé sont fondamentales pour expliquer des évènements macroscopiques, comme la fragmentation de filaments ou de structures complexes. D'autre part, nous avons montré comment les propriétés chimiques de l'ADF/cofiline modifient les propriétés physiques locales du filament et conduisent à la fragmentation. L'ADF/cofiline a un rôle central pour l'intégration de mécanismes physico-chimiques, à l'échelle microscopique, afin d'assurer un comportement cohérent à l'échelle de la cellule. / During my thesis, I have studied the pivotal role of ADF/cofilin, a protein that binds to the actin cytoskeleton, specifically decorates ‘old' actin filament parts, decreases by a factor of 5 the local filament rigidity and triggers filament fragmentation at boundaries between decorated and non-decorated filament sections. In my first study (Suarez et al., Current Biology, 2011), I have used evanescent wave microscopy and labeled ADF/cofilin to demonstrate that ADF/cofilin is a marker of the nucleotide state (i.e. ATP, ADP-Pi or ADP) associated with the actin sub-units in actively polymerizing filaments. In addition, because ADF/cofilin accelerates inorganic phosphate (Pi) release, the size of the ATP/ADP-Pi cap is diminished, although it cannot be reduced to zero. Fragmentation events frequency, determined from a thorough analysis of a population of single filaments decorated with labeled ADF/cofilin, is perfectly correlated with the binding density of ADF/cofilin on filaments. However, the maximal severing efficiency is obtained for half ADF/cofilin density. This paradoxical result is confirmed by analysis showing that severing sites are mainly associated with boundaries between decorated and bare actin filament sections. In consequence, in a second paper (McCullough et al., Biophysical Journal, 2011), I have took part in the study of actin filament deformation in relation with severing efficiency. Using different ADF/cofilin (vertebrate and yeast) and actin (vertebrate and yeast), we have shown that filament deformation at the boundary between bare and ADF/cofilin-decorated filament sections (which depends on the ADF/cofilin/actin combination) and severing are highly correlated. During my third study, (Reymann et al., Molecular Biology of the Cell, 2011), we established that stochastic dynamics, discovered at the molecular level for single filaments (or bundles of them), is also relevant to describe the macroscopic fragmentation of a comet tail consisting of hundreds of thousands filaments. I have shown that ADF/cofilin activity is at the crossroad between macroscopic and microscopic systems, on one hand, and physics and chemistry, on the other hand. The characteristics of microscopic interactions of ADF/cofilin with a single filament are fundamental to understand the macroscopic dynamics of a fragmenting comet. In addition, we have established how the binding of ADF/cofilin (chemistry) controls the mechanical properties of the filament (physics) before fragmentation. ADF/cofilin is essential in the integration of physical and chemical mechanisms at the microscopic level, to ensure consistent behavior at the cell scale.

This thesis deals with the differential response of healthy and cancerous cells to surface topography at the nanoscale and the microscale. Using a statistical method we developed we studied the interactions of cells with grooves of nanoscale depth. We demonstrate that healthy cells have a greater ability to align with deeper grooves, whereas cancerous cells are more sensitive to shallow grooves. Analysis reveals that the nucleus follows the alignment of the cell body more closely in cancerous cells, and that the nucleus of cancerous cells is more sensitive to shallow grooves.On microscale pillars we demonstrate for the first time that osteosarcoma cells deform to adopt the surface topography and that the deformation extends to the interior of the cell and in particular to the nucleus. We show that healthy cells only deform during the initial stages of adhesion and that immortalized cells show intermediate deformation between the healthy and cancerous cells. When the spacing between the pillars is reduced, differences in the deformation of different cancerous cell lines are detected. Deformation was also found to be related to the malignancy in keratinocytes, and related to the expression of Cdx2 in adenocarcinoma. The mechanism of deformation is tentatively attributed to the cytoskeleton and attempts to identify the main actors of deformation were performed using confocal microscopy and cytoskeleton inhibitors. Live cell imaging experiments reveal that the deformed cells are very mobile on the surfaces, loss of deformation is necessary for mitosis to occur and deformation after mitosis is more rapid than initial deformation upon adhesion to surfaces.

<p>The Ras superfamily of GTPases embraces six major branches of proteins: the Ras, Rab, Ran, Arf, Rho and Miro subfamilies. The majority of GTPases function as binary switches that cycle between active GTP-bound and inactive GDP-bound states. This thesis will focus primarily on the biological functions of the Rho and Miro proteins. The Rho GTPases control the organization of the actin cytoskeleton and other associated activities, whereas the Miro GTPases are regulators of mitochondrial movement and morphology. </p><p>A diverse array of cellular phenomena, including cell movement and intracellular membrane trafficking events, are dependent on cytoskeletal rearrangements mediated by Rho GTPases. Although human Rho GTPases are encoded by 20 distinct genes, most studies involving Rho GTPases have focused on the three representatives RhoA, Rac1 and Cdc42, which each regulate specific actin-dependent cellular processes. In an effort to compare the effects of all Rho GTPase members in the same cell system, we transfected constitutively active Rho GTPases in porcine aortic endothelial (PAE) cells and examined their effects on the organization of the actin cytoskeleton. We identified a number of previously undetected roles of the different members of the Rho GTPases. Moreover, we demonstrated that the downstream effectors of Rho GTPases have a broader specificity than previously thought. </p><p>In a screen for novel Ras-like GTPases, we identified the Miro GTPases (Mitochondrial Rho). In our characterization of Miro, we established that these proteins influence mitochondrial morphology and serve functions in the transport of mitochondria along the microtubule system. Additionally, we provided evidence that Miro can be under control of calcium signaling pathways. Mitochondria are highly dynamic organelles that undergo continuous change in shape and distribution. Defects in mitochondrial dynamics are associated with several neurodegenerative diseases. In conclusion, our findings have contributed to a deeper understanding of the biological roles of Rho and Miro GTPases.</p>

Cellulose is the most abundant biopolymer on earth and the main load-bearing structure in plant cell walls. Cellulose microfibrils are laid down in a tight parallel array, surrounding plant cells like a corset. Orientation of microfibrils determines the direction of growth by directing turgor pressure to points of expansion (Somerville et al., 2004). Hence, cellulose deficient mutants usually show cell and organ swelling due to disturbed anisotropic cell expansion (reviewed in Endler and Persson, 2011). How do cellulose microfibrils gain their parallel orientation? First experiments in the 1960s suggested, that cortical microtubules aid the cellulose synthases on their way around the cell (Green, 1962; Ledbetter and Porter, 1963). This was proofed in 2006 through life cell imaging (Paredez et al., 2006). However, how this guidance was facilitated, remained unknown. Through a combinatory approach, including forward and reverse genetics together with advanced co-expression analysis, we identified pom2 as a cellulose deficient mutant. Map- based cloning revealed that the gene locus of POM2 corresponded to CELLULOSE SYNTHASE INTERACTING 1 (CSI1). Intriguingly, we previously found the CSI1 protein to interact with the putative cytosolic part of the primary cellulose synthases in a yeast-two-hybrid screen (Gu et al., 2010). Exhaustive cell biological analysis of the POM2/CSI1 protein allowed to determine its cellular function. Using spinning disc confocal microscopy, we could show that in the absence of POM2/CSI1, cellulose synthase complexes lose their microtubule-dependent trajectories in the plasma membrane. The loss of POM2/CSI1, however does not influence microtubule- dependent delivery of cellulose synthases (Bringmann et al., 2012). Consequently, POM2/CSI1 acts as a bridging protein between active cellulose synthases and cortical microtubules. This thesis summarizes three publications of the author, regarding the identification of proteins that connect cellulose synthases to the cytoskeleton. This involves the development of bioinformatics tools allowing candidate gene prediction through co-expression studies (Mutwil et al., 2009), identification of candidate genes through interaction studies (Gu et al., 2010), and determination of the cellular function of the candidate gene (Bringmann et al., 2012). / Zellulose ist das abundanteste Biopolymer der Erde und verleiht pflanzlichen Zellwänden ihre enorme Tragkraft. Mit der Reißfestigkeit von Stahl umwickeln Zellulosefibrillen pflanzliche Zellwände wie ein Korsett. Die Orientierung der Zellulosefibrillen bestimmt zugleich die Wachstumsrichtung, indem sie den Zellinnendruck (Turgor) in die entsprechende Ausdehnungsrichtung dirigiert (Somerville et al.,2004).Folglich zeigen Mutanten mit gestörter Zellulosesynthese oft geschwollene Organe und Zellen, die sich nicht mehr gerichtet ausdehnen können (zusammengefasst von Endler und Persson,2011). Wie aber erhalten die Zellulosefibrillen ihre parallele Orientierung? Erste Experimente aus den1960ern führten zur Vermutung, kortikale Mikrotubuli leiten die Zellulosesynthasen auf ringförmigen Bahnen um die Zellen herum (Green, 1962; Ledbetter and Porter, 1963). Diese Theorie wurde 2006 mit Hilfe moderner mikroskopischer Methoden bestätigt (Paredez et al., 2006). Wie jedoch dieser Leitmechanismus funktioniert, blieb bisher unentdeckt. Durch die Kombination verschiedener genetischer und bioinformatischer Methoden, konnten wir pom2 als Zellulose defiziente Mutante identifizieren. Die Ermittlung des Genlocus durch Map-based cloning zeigte, dass es sich bei POM2 um CELLULOSE SYNTHASE INTERACTING 1 (CSI1) handelt, ein Gen, dessen korrespondierendes Protein, wie vorher von uns gezeigt, mit dem zytosolischen Teil der primären Zellulosesynthasen interagiert (Gu et al., 2010). Durch ausführliche zellbiologische Charakterisierung von POM2/CSI1 konnten wir seine zelluläre Funktion entschlüsseln. Mit Hilfe konfokaler Spinning- Disc-Mikroskopie konnten wir zeigen, dass in Abwesenheit von POM2/CSI1, Zellulosesynthasen von den Mikrotubuli- Bahnen abweichen. Der ebenfalls von den Mikrotubuli abhängige Transport der Zellulosesynthasen zur Zellmembran hingegen, war nicht beeinflusst (Bringmann et al., 2012). Demzufolge ist POM2/CSI1 das gesuchte Bindeglied zwischen aktiven Zellulosesynthasen und Mikrotubuli. In dieser Dissertationsschrift werden drei Publikationen des Autors zusammengefasst, die wa ̈hrend der Arbeit an der Dissertiation entstanden sind. Sie beinhalten die Entwicklung bioinformatischer Methoden zur Ko- Expressionsanalyse, um Kandidatengene zu ermitteln (Mutwil et al., 2009), die Identifikaton des Kandidatengens POM2/CSI1 in einer Interaktionsstudie (Gu et al., 2010), sowie die Bestimmung der zellula ̈ren Funktion des korrespondieren- den Proteins POM2/CSI1 (Bringmann et al., 2012).

The Ras superfamily of GTPases embraces six major branches of proteins: the Ras, Rab, Ran, Arf, Rho and Miro subfamilies. The majority of GTPases function as binary switches that cycle between active GTP-bound and inactive GDP-bound states. This thesis will focus primarily on the biological functions of the Rho and Miro proteins. The Rho GTPases control the organization of the actin cytoskeleton and other associated activities, whereas the Miro GTPases are regulators of mitochondrial movement and morphology. A diverse array of cellular phenomena, including cell movement and intracellular membrane trafficking events, are dependent on cytoskeletal rearrangements mediated by Rho GTPases. Although human Rho GTPases are encoded by 20 distinct genes, most studies involving Rho GTPases have focused on the three representatives RhoA, Rac1 and Cdc42, which each regulate specific actin-dependent cellular processes. In an effort to compare the effects of all Rho GTPase members in the same cell system, we transfected constitutively active Rho GTPases in porcine aortic endothelial (PAE) cells and examined their effects on the organization of the actin cytoskeleton. We identified a number of previously undetected roles of the different members of the Rho GTPases. Moreover, we demonstrated that the downstream effectors of Rho GTPases have a broader specificity than previously thought. In a screen for novel Ras-like GTPases, we identified the Miro GTPases (Mitochondrial Rho). In our characterization of Miro, we established that these proteins influence mitochondrial morphology and serve functions in the transport of mitochondria along the microtubule system. Additionally, we provided evidence that Miro can be under control of calcium signaling pathways. Mitochondria are highly dynamic organelles that undergo continuous change in shape and distribution. Defects in mitochondrial dynamics are associated with several neurodegenerative diseases. In conclusion, our findings have contributed to a deeper understanding of the biological roles of Rho and Miro GTPases.

All blood vessels in the microvasculature are embedded in loose connective tissue, which regulates the transport of fluid to and from tissues. The intersti-tial fluid pressure (IFP) is one of the forces that control this transport. A lowering of IFP in vivo results in an increased transport of fluid from the circulation into the underhydrated connective tissues, resulting in edema formation. During homeostasis, contractile connective tissue cells exert a tension on the connective tissue fibrous network by binding with β1 in-tegrins, thereby actively controlling IFP. During inflammation, the IFP is lowered but platelet-derived growth factor (PDGF)-BB induces an IFP nor-malization dependent on integrin αVβ3. We demonstrate that extracellular proteins from Streptococcus equi subspecies equi modulated cell-mediated and integrin αVβ3-directed collagen gel contraction in vitro. One of these proteins, the collagen- and fibronectin binding FNE, stimulated contraction by a process dependent on fibronectin synthesis. This study identified a pos-sible novel virulence mechanism for bacteria based on the ability of bacteria to modulate the edema response. Another protein, the collagen-binding pro-tein CNE, inhibited contraction and this led to the identification of sites in collagen monomers that potentially are involved in connecting αVβ3 to the collagen network. PDGF-BB and prostaglandin E1 (PGE1) stimulate and inhibit collagen gel contraction in vitro and normalize and lower IFP, respec-tively. We showed that these agents affected both similar and different sets of actin-binding proteins. PDGF-BB stimulated actin cytoskeleton dynamics whereas PGE1 inhibited processes dependent on cytoskeletal motor and adhesive functions, suggesting that these different activities may partly ex-plain the contrasting effects of PGE1 and PDGF-BB on contraction and IFP. Mutation of the phosphatidylinositol 3’-kinase (PI3K), but not phospholipase C (PLC)γ activation site, rendered cells unable to respond to PDGF-BB in contraction and in activation of the actin binding and severing protein cofilin. Ability to activate cofilin after PDGF-BB stimulation correlated with ability to respond to PDGF-BB in contraction, suggesting a role for cofilin in this process downstream of PDGF receptor-activated PI3K. Many proteins can modulate contraction either by affecting the extracellular matrix and cell adhesions or by altering cytoskeletal dynamics. Knowledge on how these proteins might influence IFP is likely to be of clinical importance for treat-ment of inflammatory conditions including anaphylaxis, septic shock and also carcinoma growth.

Physical forces in the cellular microenvironment play an important role in governing cell function. Forces transmitted through the cell cause distinct deformation of the nucleus, and possibly play a role in force-mediated gene expression. The work presented in this thesis drew upon innovative strategies employing simultaneous atomic force and laser-scanning confocal microscopy, as well as parallel optical stretching experiments, to gain unique insights into the response of eukaryotic cell nuclei to external force. Non-destructive approaches confirmed the existence of a clear anisotropy in nuclear mechanical properties, and showed that the nucleus' mechanical response to extracellular forces is differentially governed by both nuclear and cytoskeletal prestress: nuclear prestress regulates shape and anisotropic deformation, whereas cytoskeletal prestress modulates the magnitude and degree of deformation. Importantly, the anisotropic mechanical response was conserved among diverse differentiated cell types from multiple species, suggesting that nuclear mechanical anisotropy plays an important role in cell function.

Glucagon-like peptide-1 (GLP-1) is secreted from the enteroendocrine L cell following nutrient ingestion. Although GLP-1 regulates several aspects of nutrient homeostasis, one important function is to enhance glucose-dependent insulin secretion. In type 2 diabetes, post-prandial GLP-1 secretion is impaired. Insulin resistance, which is required for the pathogenesis of type 2 diabetes, is also associated with impaired GLP-1 secretion. I, therefore, hypothesized that insulin modulates GLP-1 secretion from the intestinal L cell and, furthermore, insulin resistance directly impairs the function of the endocrine L cell. In well-characterized L cell models, I established that insulin stimulates GLP-1 secretion through the MEK1/2-ERK1/2 pathway, and induction of insulin resistance in vitro attenuated insulin- and heterologous secretagogue-induced GLP-1 release. Furthermore, glucose-stimulated GLP-1 secretion was decreased in hyperinsulinemic-insulin resistant MKR mice, demonstrating that insulin resistance is associated with impaired L cell function. I next examined the role of the actin cytoskeleton in insulin-stimulated GLP-1 secretion. Insulin treatment transiently induced actin depolymerization, and depolymerization of the actin cytoskeleton potentiated insulin-stimulated GLP-1 release from the L cell, demonstrating that the cytoskeleton functions as a permissive barrier. Central to insulin’s effects on actin dynamics is the Rho GTPase, Cdc42, as siRNA-mediated knockdown and over-expression of a dominant-negative mutant, prevented insulin-stimulated actin remodeling and GLP-1 release. Insulin also promoted activation of PAK1, the downstream kinase of Cdc42, and over-expression of a kinase-dead PAK1 mutant attenuated insulin-stimulated GLP-1 release. In cells that expressed dominant-negative Cdc42 or kinase-dead PAK1, activation of ERK1/2 following insulin treatment was attenuated, demonstrating that the Cdc42-PAK1 axis regulates the activity of the canonical ERK1/2 pathway. In summary, this thesis demonstrates, for the first time, that insulin is a GLP-1 secretagogue, and this effect of insulin is mediated through the canonical ERK1/2 pathway and the Cdc42-PAK1 axis. Insulin resistance in the L cell impairs the responsiveness of the L cell to heterologous secretagogues. Collectively, these findings suggest that an alternative approach to treat type 2 diabetes and/or insulin resistance may be to directly improve the function of the L cell, thereby enhancing endogenous GLP-1 release.

High osmolarity causes amoebae of the cellular slime mould Polysphondylium pallidum to individually encyst, forming microcysts. During microcyst differentiation, actin is tyrosine phosphorylated. Tyrosine phosphorylation of actin is independent of encystment conditions and occurs during the final stages of microcyst formation. During microcyst germination, actin undergoes dephosphorylation prior to amoebal emergence. Renewed phosphorylation of actin in germinating microcysts can be triggered by increasing the osmolarity of the medium which inhibits emergence. Immunofluorescence reveals that actin is dispersed throughout the cytoplasm in dormant microcysts. Following the onset of germination, actin is observed around vesicles where it co-localizes with phosphotyrosine. Prior to emergence, actin localizes to patches near the cell surface. Increasing osmolarity disrupts this localization and causes actin to redistribute throughout the cytoplasm, a situation similar to that observed in dormant microcysts. Together, these results indicate an association between actin tyrosine phosphorylation, organization of the actin cytoskeleton, and microcyst dormancy.

High osmolarity causes amoebae of the cellular slime mould Polysphondylium pallidum to individually encyst, forming microcysts. During microcyst differentiation, actin is tyrosine phosphorylated. Tyrosine phosphorylation of actin is independent of encystment conditions and occurs during the final stages of microcyst formation. During microcyst germination, actin undergoes dephosphorylation prior to amoebal emergence. Renewed phosphorylation of actin in germinating microcysts can be triggered by increasing the osmolarity of the medium which inhibits emergence. Immunofluorescence reveals that actin is dispersed throughout the cytoplasm in dormant microcysts. Following the onset of germination, actin is observed around vesicles where it co-localizes with phosphotyrosine. Prior to emergence, actin localizes to patches near the cell surface. Increasing osmolarity disrupts this localization and causes actin to redistribute throughout the cytoplasm, a situation similar to that observed in dormant microcysts. Together, these results indicate an association between actin tyrosine phosphorylation, organization of the actin cytoskeleton, and microcyst dormancy.