Glioblastoma is the most malignant primary brain tumor with the average patients surviving only one year after diagnosis, even with aggressive therapy. The formation of numerous micro-tumors dispersed into the brain due to rapid invasion of tumor cells, presents the primary challenge to the surgical removal of tumors and limits the effectiveness of current treatments. This dissertation presents studies aimed at understanding the molecular mechanisms regulating invasion of human glioblastoma cells. Transplantation of human glioblastoma cells in the zebrafish brain showed that the knockdown of calpain 2, a calcium-activated protease, resulted in a three fold decrease in the tumor cell invasion. The result was further verified in the organotypic mouse brain slices where the knockdown cells demonstrated 2-fold decrease in the area of dispersal compared to control cells. Our data show that calpain 2 plays a role in the process of tumor cell angiogenesis. Glioblastoma cells were transplanted into the brain of zebrafish expressing GFP in the blood vessels and we observed that 23% of animals injected with control tumor cells demonstrated angiogenesis. In contrast, only 9% of fish that received calpain 2 knockdown cells showed the formation of new vessels. Consistent to the reports from human glioblastoma patients and rodent models, we did not observe metastasis of transplanted cells outside of the brain in the zebrafish, supporting for the use of zebrafish as an important model for glioblastoma cell invasion studies. These results provide evidence that calpain 2 protease activity is required for the dispersal of glioblastoma cells in the brain microenvironment. To determine the mechanism of calpain 2 regulation of tumor cell invasion, proteolysis of filamin by calpain 2 was studied. Filamin is an important actin cross-linking protein which develops orthogonal actin networks in the periphery of the cell. In this study, we show that the expression of filamin inhibits glioblastoma cell invasion. Hence, knocking down filamin expression by 80% resulted in 220% increase in the invasion of glioblastoma cells through Matrigel extracellular matrix. The regulated proteolysis of filamin is a potential mechanism to facilitate the cyclic turnover of actin orthogonal networks which is required for glioblastoma cell invasion. In this study, we identified a novel mechanism that the PI3 kinase activity regulates the cleavage of filamin by calpain 2 in glioblastoma cells. Binding of a membrane phospholipid phosphatidylinositol (3,4,5) triphosphate [PtdIns (3,4,5)-P₃] to filamin induces its proteolysis by calpain 2 after the amino acid lysine 268, removing the actin binding domain which in-turn abolishes the actin binding ability of filamin. / Graduation date: 2011 / Access restricted to the OSU Community at author's request from Jan. 31, 2011 - Jan. 31, 2012

Abstract Filamins are large dimeric proteins that cross-link actin into three-dimensional bundles or orthogonal networks. In addition to an actin-binding domain, each filamin monomer contains 24 immunoglobulin-like domains separated by flexible regions between domains 15–16 and 23–24. Dimerisation of filamin occurs through the Ig-like domain 24. Filamins bind to a variety of molecules. They provide a link between the plasma membrane and the cytoskeleton through interactions with transmembrane receptors, and at the same time, serve as a platform for signalling molecules. Filamins are involved in several human diseases affecting the central nervous system, vascular system and muscle. In this study the structure of the the carboxy-terminus of filamin was resolved and details of filamins interaction with a platelet surface protein important in haemostasis were analysed. An x-ray structure of the Ig-like domain 24 of human filamin C was solved at the resolution of 1.43 Å. The asymmetric unit of the crystal contains one monomer; a crystallographic dimer is formed by 2-fold axis symmetry. Point mutation studies confirmed that the dimer seen in the crystal is also present in solution. The structure showed that the dimerisation mode of human filamin is completely different from that in the Dictyostelium discoideum amoeba filamin analogue. Human filamin dimerises through β-strands C and D, and the Dictyostelium protein through β-strands B and G located on the opposite edge of the β-sandwich. Based on the sequence homology between vertebrate filamins it was proposed that the interface seen in human filamin is common for all vertebrate filamins. The structure of human filamin C Ig-like domains 23–24 was solved by combining the techniques of x-ray crystallography and small angle x-ray scattering (SAXS). This structure provides further insight into the organization of the domains in the carboxy-terminal part of filamin molecule. One of the first structural examples of the interaction of filamin with a ligand was provided by this study. The x-ray structure of filamin A domain 17 in complex with the alpha subunit of the GPIb-V-IX receptor was solved at a resolution of 2.3 Å. The interaction between filamin and the GPIbα-V-IX receptor is important for maintaining the integrity and shape of blood platelets, as well as for regulating the receptor adhesive function. This study also revealed that the Ig-like domain 17 represents a major binding site of filamin to GPIbα. The Kd of the interaction, determined by calorimetric studies, was 11 μM. The specificity of the filamin A 17 - GPIbα interaction is mainly determined by hydrophobic contacts.

Abstract FAP52 (focal adhesion protein, 52 kDa) is a focal adhesion-associated protein composed of a highly α-helical NH2-terminus containing a poorly characterized FCH (Fes/CIP4 homology) domain, unstructured linker region and the COOH-terminal SH3 domain. FAP52 is also known as PACSIN 2 or syndapin II. Together with other PACSINs and syndapins FAP52 shares a common domain architecture. The aim of this study was to characterize FAP52 in structural and functional terms. The function was pursued by identifying binding partners for FAP52, and by overexpressing the recombinant FAP52 in cultured cells. For the structural studies, various physico-chemical methods, such as chemical cross-linking, gel filtration chromatography, circular dichroism and X-ray crystallography were applied. In addition, the histological distribution of FAP52 in chicken tissues was explored. FAP52 binds filamin, a protein that regulates the dynamics of the cytoskeleton by crosslinking actin filaments. The binding site in FAP52 was mapped to the NH2-terminal 184 amino acids, of which the residues 146–184 form the core of the binding. In filamin, the binding site resides in the repeats 15–16 in the rod-like molecule encompassing 24 such repetitive domains. Overexpression of FAP52 or its filamin-binding domain in chicken embryo heart fibroblasts induced the formation of filopodial extensions on the cell surface and reduced the number of focal adhesions, suggesting a role in the organization of the cellular cytoskeleton and in cell adhesion machinery. Experiments utilizing surface plasmon resonance analysis, size exclusion chromatography and chemical cross-linking showed that FAP52 self-associates in vitro and in vivo. The region responsible for the self-association was mapped to the amino acids 146–280, which is predicted to fold into a coiled-coil arrangement. FAP52 was crystallized by using the hanging-drop vapor-diffusion method and ammonium sulfate grid screen. Native dataset was collected from two crystals, which diffracted to 2.8 Å and 2.1 Å resolution. For one form of crystals, phasing was performed using the native dataset and the datasets from two xenon-derivatized crystals. X-ray crystallography studies revealed a dimer in asymmetric unit. Histological and in vitro studies showed that, in liver, FAP52 is preferentially expressed in bile canaliculi. In other tissues, FAP52 showed a specific staining pattern in gut, kidney, brain and gizzard. Together, these data show that FAP52 self-associates in vivo and, probably via its interaction with its binding partner filamin, participates in the organization of the cytoskeletal architecture, especially of the cell surface protrusions, such as filopodia and microvilli of bile canaliculi.

Filamin A is a well-characterised cytoskeletal protein which regulates cell shape and migration by cross-linking with actin. Filamin A mutations cause a number of human developmental disorders, many of which exhibit skeletal dysplasia. However, the molecular mechanisms by which Filamin A affects skeletal development are unknown. The transcription factor Runx2 is a master regulator of osteoblast and chondrocyte differentiation. Data presented in this thesis show that Filamin A forms a complex with Runx2 in osteoblastic cell lines. Moreover, it is demonstrated that Filamin A is present in the nucleus in several cell lines, including those of osteoblastic origin. The data presented show that the Filamin A/Runx2 complex suppresses the expression of the gene encoding the matrix-degrading enzyme, matrix metalloproteinase-13 (MMP-13), which is an important osteoblastic differentiation marker. ChIP assays were employed to demonstrate that endogenously expressed Filamin A associates with the promoter of the MMP-13 gene. In addition, Filamin A is not only located in the nucleus but also in the nucleolus, an important nuclear compartment involved in ribosomal RNA (rRNA) transcription. Ribosomal DNA promoter-driven reporter assays, Filamin A-knockdown experiments and exogenous Filamin A transfections demonstrated that Filamin A and Runx2 can repress ribosomal gene expression activity. Importantly, Filamin A is recruited to the human ribosomal DNA promoter, suggesting its direct involvement in the regulation of rRNA transcription. These findings reveal a novel role of Filamin A in the direct regulation of ribosomal gene expression. Finally, by using microarray technology, changes in gene expression were identified when Filamin A was downregulated. Some of the differentially expressed genes were known orchestrators of bone development. The data presented in this thesis strengthen the link between Filamin A and bone development and provide a molecular rationale for how Filamin A, acting as a regulator of gene expression, might influence osteoblastic differentiation.

Les hétérotopies nodulaires périventriculaires (HNP) correspondent aux malformations cérébrales les plus fréquemment découvertes à l'âge adulte. Survenant au cours de la migration, elles consistent en l'apparition de nodules de neurones ectopiques positionnés le long de la paroi des ventricules latéraux. Sur le plan clinique, les HNP associent une épilepsie et/ou un retard mental. Les mutations dans le gène FLNA (Xq28) représentent la cause majeure des HNP. Une forme récessive rare d'HNP liée à des mutations du gène ARFGEF2 (20q13) et des réarrangements chromosomiques identifiés chez des patients présentant une HNP ont également été rapportés. Alors que le lien entre les HNP associées à des mutations du gène FLNA et leurs manifestations cliniques a été clairement établi, les mécanismes physiopathologiques sous-jacents restent à ce jour inconnus. Deux lignées de souris knockout pour FlnA ont été développées mais aucune de ces deux lignées n'a développé d'HNP. Nous avons donc choisi de créer un nouveau modèle, chez le rat, par inactivation in utero du gène FlnA en utilisant la technique de l'ARN interférence (RNAi). Par cette approche, nous avons reproduit avec succès un phénotype d'HNP chez le rat comparable à celui observé chez les patients. Sur ce modèle, nous avons montré que l'HNP est associée à une désorganisation de la glie radiaire et à une incapacité des progéniteurs neuronaux de progresser dans le cycle cellulaire. En accord avec ces observations, une désorganisation de la glie radiaire a été également observée dans des cerveaux post-mortem de deux patientes présentant une HNP associée à une mutation de FLNA. / Periventricular nodular heterotopia (PNH) is a brain malformation caused by defective neuronal migration resulting in ectopic neuronal nodules lining the lateral ventricles. Most patients have epilepsy, with normal to borderline cognitive function. Mutations in FLNA are the main cause of PH. A rare recessive form caused by mutations in the ARFGEF2 gene (20q13) and chromosomal rearrangements identified in patients with PNH have been reported. The link between FLNA-trelated PH and clinical manifestattions has been wee established but the underlying pathological mechanism remains unknown. Though two FlnA knockout mice strains have been developed, progress has been hindered by the fact that none of them showed the presence of ectopic nodules. Therefore, to recapitulate the loss of FlnA function in the developing rat brain, we used an in utero RNA interference (RNAi)-mediated knockdown approach and successfully reproduced a PNH phenotype in rats comparable to that observed in patients. Using this FlnA knockdown rodent model, we demonstrated that PNH is associated with a disruption in radial glial scaffold integrity in the ventricular zone and also an inability for neuroprogenitor cells to progress adequately through the cell cycle.Consistent with the observations made in rodents, we found similar alterations of radial glia in postmortem brains of two PNH patients harboring distinct FLNA mutations. These data highlights the complexity of the pathogenesis of PNH, the likelihood that several mechanisms are coalescing to lead to disrupted neuronal migration.

<p>Cellular processes like motility, chemotaxis, phagocytosis and morphogenesis are dependent on the dynamic regulation of the actin cytoskeleton. This cytoskeleton system is tightly controlled by a number of diverse actin-binding proteins (ABPs) by various mechanisms described as nucleation, polymerization, capping, severing, depolymerization and sequestration. The ABPs are grouped based on sequence identity as in the Wiskott-Aldrich Syndrome protein homology domain 2 (WH2), and the calponin homology domain (CH) containing proteins.</p><p>In this work, we elucidate the crystal structures of hybrids of gelsolin domain 1 with thymosin β4, ciboulot domain 2, and the second WH2 domain of N-WASP each bound to actin. We show that the single WH2 motif containing protein thymosin β4 in part sequesters actin by binding its pointed end via a C-terminal helix. This interaction prevents the addition of bound actin protomers to the barbed end of the filament. We propose that sequence variations in some WH2 motifs conferred F-actin binding ability to multiple repeat-containing proteins. These F-actin binding domains interact with the barbed end of a filament and the adjacent WH2 motifs are then freed to add their bound actin to the growing filament end. We demonstrate the binding of ciboulot domains 2 and 3 to both G- and F-actin and that full length ciboulot is capable of binding two actin monomers simultaneously. </p><p>We have also cloned, expressed, purified and crystallized rod domains 14-16 from the actin crosslinking protein a-filamin. Preliminary X-ray crystallography data gives us hope that we shall be able to solve the structure of this triple domain repeat.</p>

Cellular processes like motility, chemotaxis, phagocytosis and morphogenesis are dependent on the dynamic regulation of the actin cytoskeleton. This cytoskeleton system is tightly controlled by a number of diverse actin-binding proteins (ABPs) by various mechanisms described as nucleation, polymerization, capping, severing, depolymerization and sequestration. The ABPs are grouped based on sequence identity as in the Wiskott-Aldrich Syndrome protein homology domain 2 (WH2), and the calponin homology domain (CH) containing proteins. In this work, we elucidate the crystal structures of hybrids of gelsolin domain 1 with thymosin β4, ciboulot domain 2, and the second WH2 domain of N-WASP each bound to actin. We show that the single WH2 motif containing protein thymosin β4 in part sequesters actin by binding its pointed end via a C-terminal helix. This interaction prevents the addition of bound actin protomers to the barbed end of the filament. We propose that sequence variations in some WH2 motifs conferred F-actin binding ability to multiple repeat-containing proteins. These F-actin binding domains interact with the barbed end of a filament and the adjacent WH2 motifs are then freed to add their bound actin to the growing filament end. We demonstrate the binding of ciboulot domains 2 and 3 to both G- and F-actin and that full length ciboulot is capable of binding two actin monomers simultaneously. We have also cloned, expressed, purified and crystallized rod domains 14-16 from the actin crosslinking protein a-filamin. Preliminary X-ray crystallography data gives us hope that we shall be able to solve the structure of this triple domain repeat.

The ability to adhere to the extracellular matrix (ECM) is critical for numerous cell types and tissues including epithelia and muscle. Cell-ECM adhesion is primarily mediated by integrins which provide a direct link between the ECM and the actin cytoskeleton. Integrin adhesions are frequently associated with a core of 60 different proteins (integrin-associated proteins, IAPs). Integrins are required for muscle attachment and in Drosophila, loss of integrins and several IAPs results in embryonic lethality and muscle detachment. However, the IAPs FAK, RSU1, tensin, vinculin and zyxin are not required for viability or embryonic muscle attachment. Furthermore, FAK, RSU1, tensin and vinculin have been observed to localise to muscle attachment sites in Drosophila, indicating that they have some function in muscle attachment. Unlike FAK, RSU1, tensin and vinculin, it was not previously known whether zyxin is expressed in Drosophila muscles. To test this, I generated a genomic zyxin-GFP construct that should contain most of the endogenous zyxin promotor. The genomic zyxin-GFP construct was not observed at muscle attachment sites, suggesting that it is not normally expressed in muscle. I wished to know whether FAK, RSU1, tensin and vinculin are required for muscle function. Various behavioural assays were employed to test for muscle function in larvae and adult flies. The results suggest that larval muscle function was normal in flies lacking these IAPs, but that adult muscle function might be impaired, although it proved difficult to demonstrate a clear functional defect. I then tested whether the IAPs FAK, RSU1, tensin and vinculin are required for normal morphology of adult muscles, focusing on the adult indirect flight muscles (IFMs). The IFMs are fibrillar muscles which attach to the cuticle via specialised epithelial cells known as tendon cells. At the end of the myofibril, where the myofibril attaches to the tendon cell, is a dense region of actin and IAPs known as the modified terminal Z-band (MTZ). I have found that the MTZ is not a homogenous zone of proteins, but is instead organised into at least three distinct layers. Because of the similarity between the structure of the MTZ with that of a hand, I refer to the layers as ‘fingers’, ‘palm’ and ‘wrist’. I discovered that the IAPs FAK, RSU1, tensin and vinculin are each required for the proper structure of the MTZ in unique ways. The fingers were elongated in IFMs lacking FAK, RSU1, tensin or vinculin, while the palm was disrupted in IFMs lacking RSU1, tensin or vinculin. Finally, I was intrigued by the enrichment of the actin-binding protein filamin/Cheerio in the palm and wished to know if it is required for palm function. Deletion of the C-terminus of filamin/Cheerio resulted in a reduction in palm length. Filamin/Cheerio is a mechanosensitive protein which exists in a closed and open conformation. I found that filamin/Cheerio must be open in order to help form a normal palm. Furthermore, vinculin is required to convert filamin/Cheerio from and closed to an open filamin/Cheerio state so that it can perform its function in the palm.

In cell culture, entry into mitosis of many adherent mammalian cells is accompanied by substantial changes in cellular architecture. Flat, spread-out interphase cells detach from the extracellular matrix and become more spherical. These changes in cell shape are mediated by rearrangements in the actin cytoskeleton, a dynamic network of actin filaments that are organized by actin-binding proteins. Filamin A (FLNa) is a 280 kD actin-binding protein that crosslinks actin filaments into parallel bundles or three-dimensional orthogonal networks. We previously identified FLNa as an in vitro substrate of cyclin-dependent kinase 1 (Cdk1), a kinase that regulates entry into mitosis, and hypothesized that Cdk1 phosphorylation of FLNa regulates mitotic actin remodelling. Using mass spectrometry and a p-FLNa antibody, we show that FLNa is phosphorylated in vivo in HeLa cells on multiple Cdk1 sites, including serines 1084, 1459 and 1533. All three sites match the phosphorylation consensus sequence of Cdk1. We further show that p-FLNa is almost fully dephosphorylated by anaphase, consistent with it being a cell cycle-regulated substrate. Using a phospho-specific antibody, we find that p-FLNa has decreased cortical actin localization compared to total FLNa in mitotic cells. To investigate the functional role of mitotic FLNa phosphorylation, we mutated serines 1084, 1459 and 1533 to nonphosphorylatable alanine and expressed this FLNa mutant (FLNa-S1084A, S1459A, S1533A, referred to as “FLNa-AAA GFP”) in FLNa-deficient human M2 melanoma cells. FLNa-AAA GFP-expressing cells have enhanced FLNa-AAA GFP localization at sites of contact between daughter cells and this correlates with defects in cell division and impaired cell migration. Therefore, mitotic delocalization of cortical FLNa is critical for successful cell division and interphase cell behaviour.

No description available.