Focal adhesion kinase (FAK) controls the assembly of focal adhesion sites and transduces signals from several membrane receptors. Controlled activation and localization of FAK functionally links cell adhesion, migration and survival. FAK is overexpressed in many cancer types, promoting tumor invasiveness and metastasis. The molecular mechanisms allowing FAK to fulfil numerous different functions and act as versatile ‘nanomachines’ are poorly understood. We have previously revealed that ligand-induced dimerization along with intramolecular interactions control FAK activation and localization where the C-terminal focal adhesion targeting (FAT) domain is strictly involved. In this study, we combine NMR with X-ray crystallography, as well as biophysical and computational methods to understand the molecular mechanisms that link the large-scale dynamics and intramolecular and intermolecular interactions of FAT into FAK’s capacity to integrate various stimuli into a site-specific function. Our results reveal FAT-mediated dynamical interplays between binding of known and newly discovered FAT ligands, and multimerization and autoactivation of FAK. Additionally, we investigate the impact of neuronal alternative splicing on FAT dynamics and interactions. Collectively, our results elucidate FAT’s role in allosterically controlling various FAK functions, and might inspire allosteric protein-protein interaction inhibitors against FAK-dependent cancer cell proliferation.

Adhesion complexes (ACs) are large macromolecular complexes of integrins and associated proteins that connect the actin cytoskeleton to the extracellular matrix. In migrating cells, ACs are highly dynamic -- forming and maturing at the cell front and disassembling at the cell rear. The turnover of ACs enables and localises the necessary traction forces required for cell migration. There is evidence for the spatiotemporal recruitment of specific proteins during AC maturation or disassembly; however, a holistic understanding of the compositional changes to ACs during these states is lacking. To this end, we sought to characterise the dynamic changes that occur at ACs during turnover using a mass spectrometry (MS)-based proteomics approach. A major challenge in studying AC turnover is the desynchronised nature of AC formation, maturation and disassembly within a population of cells. Therefore a nocodazole-washout assay was used to synchronise microtubule-induced AC maturation and disassembly. To study the dynamics of AC turnover by MS, an AC isolation method was optimised for use with the nocodazole-washout assay. Subsequently, the maturation of ACs by the loss of microtubules was studied by MS-based proteomics, and it was found that this resulted in the overall accumulation of adhesion proteins, and also the conversion of fibrillar adhesions to focal adhesions. Studying the dynamic process of AC disassembly requires a sensitive MS quantification method; as such, label-free quantitative methods were compared, and it was found that LC-MS peak ion intensity quantification performed better than spectral counting. Using optimised methodologies for isolation of ACs and MS quantification, the dynamics of AC disassembly was analysed over the course of the nocodazole-washout assay. It was found that in general, microtubules were enriched around ACs, whereas many structural AC proteins decreased over time. In summary, we have optimised methods for the study of ACs by MS-based proteomics, and applied these methods to the study of AC turnover.

京都大学 / 0048 / 新制・課程博士 / 博士(農学) / 甲第20587号 / 農博第2239号 / 新制||農||1052(附属図書館) / 学位論文||H29||N5076(農学部図書室) / 京都大学大学院農学研究科応用生命科学専攻 / (主査)教授 植田 和光, 教授 矢﨑 一史, 教授 宮川 恒 / 学位規則第4条第1項該当 / Doctor of Agricultural Science / Kyoto University / DFAM

Here I describe a new function of Focal Adhesion Kinase (FAK) in driving anti-tumour immune evasion. The kinase activity of FAK in squamous cancer cells drives the recruitment of regulatory T-cells (Tregs) by transcriptionally regulating chemokine/cytokine and ligand-receptor networks, including the transcription of CCL5 and TGFβ, which are required for enhanced Treg recruitment. In turn, these changes inhibit antigen-primed cytotoxic CD8+ T-cell activity in the tumour microenvironment, permitting survival and growth of FAK-expressing tumours. I show that immune evasion requires FAK’s catalytic activity, and a small molecule FAK kinase inhibitor, VS-4718, which is currently in clinical development, drives depletion of Tregs and permits CD8+ T-cell-mediated tumour clearance. It is therefore likely that FAK inhibitors may trigger immune-mediated tumour regression, providing previously unrecognized therapeutic benefit.

Tumour angiogenesis is a hallmark of cancer. Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase involved in endothelial cells (ECs) survival, proliferation and migration. FAK has several tyrosine phosphorylation sites thought to be involved in FAK function but the requirement of phosphorylation of these residues in vivo is unknown. We have generated mice where endogenous FAK is deleted simultaneously with the expression of nonphosphorylatable FAK-Y397F or FAK-Y861F mutated or wild type forms of FAK in adult endothelium in order to test this. My data show that EC-FAK-Y397FKI mice present with decreased tumour angiogenesis (in sygeneic B16F0, CMT19T and LLC) but impaired B16F0 and CMT19T tumour growth only, with increased tumour hypoxia. FAK-Y397F tumour endothelium is not perfusion, leakage or vascular maturation defective. This mutation affects VEGF-, PlGF- and bFGF-driven angiogenesis in vivo and VEGF+Ang2 administration is able to partially rescue this phenotype ex vivo. In contrast, endothelial FAK-Y861F mutation leads to an initial delay in B16F0 tumour angiogenesis, that subsequently resolves, and does not affect B16F0 tumour growth. LLC and CMT19T tumour growth and angiogenesis are not affected by the endothelial FAK-Y861F mutation; neither are tumour blood vessel perfusion, leakage, vascular maturation or tumour hypoxia. VEGF-, PlGFand bFGF-driven angiogenesis in vivo and ex vivo was not affected by the endothelial FAK-Y861F mutation, whereas increased in vivo angiogenesis was triggered by Ang2 administration. Lastly, to understand whether cytokine profiles that might affect angiocrine signalling are affected differentially in FAK-Y397F vs FAK-Y861F endothelial cells, I show that CCL1 and CCL2 are increased in FAK-Y397F but IL- 13, IL-1F3, CCL4, IL-1F1, CCL2 and others are increased in FAK-Y861F endothelial cells. Overall my data indicates that endothelial-specific FAK mutations on two phosphorylation sites has different effects on tumour angiogenesis, tumour growth, growth factor stimulated angiogenesis in vivo and ex-vivo and cytokine production.

Integrin recycling has been generally assumed to be important for cell migration but the trafficking pathways and the molecules regulating integrin trafficking remain poorly characterized. Furthermore, little is known about the activation status of endocytosed integrins and how it affects the recycling of these receptors. It is likely that FA-engaged integrins will follow different trafficking pathways than bulk integrins and here I sought to study the endocytic fate of this particular integrin pool using the MT-induced FA disassembly assay. I found that integrins previously resident at FAs travel through different Rab compartments after FA disassembly and that their return to the plasma membrane is Rab11- and Src-dependent. Strikingly, I unveiled new functions for FAK and Src family kinases in this process by showing that these kinases are critical to keep integrins active during endocytic trafficking. This finding is unprecedented since it was not known whether endocytosed integrins were kept active during their trafficking. Interestingly, reassembly of FAs from endocytosed integrin occurred preferentially at the leading edge of migrating cells suggesting that integrins are trafficked in a polarized fashion. Furthermore, the recycling of integrins from the Rab11-positive compartment to the plasma membrane is a long-range transport implying the existence of a MT motor committed to this task. Consistently, I identified that a kinesin-II motor, Kif3AC, is engaged in this process. My work establishes a FAK- and Src family kinases-based mechanism for integrin "adhesion memory" during endocytic trafficking and identifies a direct link between FA disassembly and reassembly through an endocytic recycling pathway involving Rab5 and Rab11 and a kinesin-II family member.

The mechanical properties of the tissue vary widely around the body, from the soft brain to the rigid bone. Tissue cells are able to sense mechanical signals from their environment, which influence many aspects of cell behaviour such as migration, proliferation and differentiation. Focal adhesions (FAs) are large protein complexes that form the bridge between the extracellular matrix (ECM)-binding integrins and the contractile actin cytoskeleton. Here, they sense the rigidity of the local environment and translate this information into a cellular response, a process known as mechanotransduction. However, the FA proteins required for mechanotransduction, and the molecular mechanisms involved in this fundamental process, remain to be elucidated. Talin, vinculin, FAK and paxillin are four core FA-associated proteins that are thought to be involved in mechanotransduction. These proteins associate and dissociate from the complex in a constant state of flux. Using a live-cell imaging approach, I found that the rate of dynamic exchange of an FA protein correlates to its function. The FA appears to have a modular organisation; the slowest proteins have a structural role, such as talin and vinculin, responsible for directly linking integrin to actin and sensing the ECM stiffness. The signalling proteins are turned over more rapidly, including FAK and paxillin, and are responsible for directing the cellular response to force-generated signals from the ECM.The second results chapter focused on the force-dependent interactions between talin, vinculin and actin. The talin domains R2R3 were identified as the key mechanosensitive vinculin-binding sites, which are exposed upon the application of force across the talin rod. Vinculin binding to R2R3 led to actin associating with the central actin-binding site in the talin rod (ABS2), which is required for the transmission of actomyosin tension onto the underlying substrate as cellular traction force. Finally, the protein turnover data were incorporated into two mathematical models, describing talin and vinculin turnover, which were able to simulate the dynamic exchange of various talin and vinculin mutants in response to changing ECM stiffness. Using these models, the talin ABS2-actin and vinculin tail-actin interactions were found to be extremely important for sensing the stiffness of the ECM. These findings significantly increase our knowledge of the molecular mechanisms underpinning cellular mechanotransduction. Increased understanding of how mechanical signals are sensed and interpreted by the cell could lead to a number of novel therapies for a wide range of associated diseases, such as atherosclerosis, muscular dystrophy and cancer.

My thesis, written in an alternative format, consists of three manuscripts. The first one is published in Journal of Cell Science and is entitled "GAS2-like proteins mediate communication between microtubules and actin through interaction with end-binding (EB) proteins." This article describes the mechanisms of how members of the GAS2 family of proteins mediate the crosstalk between actin and microtubules (MTs). We show that in particular GAS2-like 1 (G2L1) and GAS2-like 2 (G2L2) coordinate this cross-communication, as their exogenous expression leads to the stabilisation of MTs and guidance along actin stress fibres. We found that the association of GAS2-like members with MTs is mediated through their binding to EB proteins. The second article is a follow up story of the first article, in which we further elucidate the role of GAS2-like proteins during cell division. We show that G2L1 localises to the mitotic spindle and cleavage furrow during cell division. G2L1 knockdown leads to reduced cell division rates, multinucleation and nuclear deformation. As for MT guidance along actin filaments, we demonstrate that the binding of G2L1 to EB proteins plays an important role in cell division. Although overexpression of G2L1 had no effect, the expression of a mutant that blocks the association with EB proteins phenocopies the knockdown effect of G2L1 on cell division. Actin and MTs undergo major reorganisation during cell division. This reorganisation involves the fast remodelling of focal adhesions (FAs) but the mechanisms of this remodelling were not clear. In the third paper we demonstrate that the majority of FAs disassemble shortly before cell division and reassemble in newly formed daughter cells during cytokinesis. Interestingly, our data suggest that the regulation of FA disassembly during cell division differs from the disassembly processes during cell migration. While in migrating cells FAs can be stabilised by the expression of constitutively active vinculin (vinT12, known to circumvent the requirement forces for FA stability), this was not case for FAs during cell division. Further experiments using inhibitors suggested that calpain-driven cleavage of FA components but not endocytosis play a key role in FA disassembly during cell division. Altogether, the three manuscripts provide insight into important molecular aspects involved in the regulation of cell cytoskeletal networks and cell adhesion during cell division.

Breast cancer most commonly metastasizes to the bone, where it perpetuates the vicious cycle leading to osteolytic lesions. This occurs when secreted factors from breast cancer cells disrupt bone homeostasis by deregulation of osteoblast bone formation, and enhance osteoclast bone degradation thereby releasing bone matrix bound growth factors leading to further tumor growth. Although the use of osteoclast targeting agents, such as bisphosphonates and RANK-L inhibitors, are common practice for the treatment of bone metastasis, they have not been shown to increase patient survival. We therefore sought to investigate the role of focal adhesion kinase (FAK), a potential therapeutic target, in the treatment of breast cancer mediated osteolysis. FAK is a non-receptor tyrosine kinase known to directly regulate tumor progression and metastasis; it is also expressed in all of the cell types involved in breast cancer mediated osteolysis. Thus, we hypothesized that the inhibition of FAK would restore normal bone homeostasis, as well as mediate direct anti-tumor activity. FAK depletion resulted in the decrease of expression of several osteolytic factors secreted by breast cancer cells. However, the use of FAK depleted breast cancer conditioned media did not prevent breast cancer mediated osteoclastogenesis in an osteoblast/osteoclast coculture. In monoculture however, using the FAK inhibitor PF-271, we have shown that FAK inhibition leads to increased apoptosis of mature osteoclasts, and their decreased ability to degrade mineralized bone matrix, perhaps in part due to reduced expression of lytic factors such as tartrate resistant acid phosphatase and cathepsin K. Further, FAK inhibition in osteoblast monoculture led to a decrease in their ability to express the maturation factor alkaline phosphatase, and also inhibited their ability to induce mineralization. This inhibition may be due in part to the specific effects of FAK inhibition using PF-271, which may result in decreased levels of p53 in treated osteoblasts. These results suggests that the pharmacological inhibition of FAK can effect all three cell types involved in the vicious cycle of bone metastasis, and as such could be a beneficial therapeutic for patients with bone metastasis resulting in prevention of bone degradation along with direct inhibition of tumor growth. However, it may require further evaluation in animal models to determine if observed effects on osteoblast activity in vitro also occurs in vivo with possible detrimental effects on restoration of damaged bone.

Protein kinase A (PKA) regulates a diverse array of cellular activities including metabolism, differentiation, actomyosin contractility, and migration. The multifunctionality of this ubiquitous enzyme is achieved, in part, through subcellular targeting mediated by the A Kinase Anchoring Proteins (AKAP) family of proteins. AKAPs serve as scaffolding proteins that localize PKA to various cellular compartments and bring together specific targets and modulators of PKA activity. The importance of spatially restricted PKA signaling is particularly apparent in the context of cell motility. It has been observed that both anchoring through AKAPs and the subsequent localized activation of PKA at the leading edge of migrating cells are required for directed migration in multiple cell types. Despite the significant body of evidence linking PKA to the regulation of cellular adhesion, contractility, and migration, the mechanisms governing the spatiotemporal control of PKA signaling during these activities is not fully understood. Focal adhesion complexes, which connect the actin cytoskeleton to the extracellular matrix and are thus intimately involved in the adhesive and contractile state of the cell, are attractive potential sites of PKA signaling. We have evidence indicating that PKA is active within these complexes, and that this activity impacts focal adhesion dynamics. To address the question of how PKA may be recruited to adhesive complexes, we have developed a targeted screen to identify PKA interacting proteins within adhesive and cytoskeletal structures. This method utilizes proximity-dependent biotin labeling in combination with a focal adhesion purification preparation and downstream proteomic analysis. The results of this screen will be used to identify candidate AKAPs and will serve as the foundation for future inquiry into the complex role of PKA in the regulation of cell migration.