Disordered states of proteins include the biologically functional intrinsically disordered proteins and the unfolded states of folded proteins which are important for protein folding and stability. Just as solving the structures of folded proteins has been extremely valuable in understanding their functions and properties, obtaining a comprehensive understanding of the structural characteristics of disordered states at a molecular level is crucial. The focus of this thesis is on combining experimental data with computational methods in order to probe the structural characteristics of disordered states at a molecular level. I developed a new method that combines different chemical shifts into a single residue-specific secondary structure propensity (SSP) score which I used to compare fractional secondary structure in alpha- and gamma-synuclein. Significant differences between the two suggested that gamma-synuclein might be protected from fibrillation due to increased helical propensity. I also introduced a new method for calculating residual dipolar couplings (RDCs) from disordered state ensembles by calculating local alignment tensors for short protein fragments. Using this method, I was able to predict experimental RDCs from statistical coil models containing far fewer structures than when global alignment is used, demonstrating that RDCs in disordered proteins are primarily determined by local structure. Finally, I made major improvements to the ENSEMBLE program which is used for calculating structural models of disordered states. I utilized large amounts of experimental data in order to calculate ensemble models of the Drosophila drkN SH3 domain unfolded state. Although highly heterogeneous and having broad molecular size distributions, the calculated ensembles have very different properties than expected for random or statistical coils and possess significant non-native alpha-helical structure and both native-like and non-native tertiary structure. This has significant implications for our understanding of the structural properties of protein disordered states in general.

Disordered states of proteins include the biologically functional intrinsically disordered proteins and the unfolded states of folded proteins which are important for protein folding and stability. Just as solving the structures of folded proteins has been extremely valuable in understanding their functions and properties, obtaining a comprehensive understanding of the structural characteristics of disordered states at a molecular level is crucial. The focus of this thesis is on combining experimental data with computational methods in order to probe the structural characteristics of disordered states at a molecular level. I developed a new method that combines different chemical shifts into a single residue-specific secondary structure propensity (SSP) score which I used to compare fractional secondary structure in alpha- and gamma-synuclein. Significant differences between the two suggested that gamma-synuclein might be protected from fibrillation due to increased helical propensity. I also introduced a new method for calculating residual dipolar couplings (RDCs) from disordered state ensembles by calculating local alignment tensors for short protein fragments. Using this method, I was able to predict experimental RDCs from statistical coil models containing far fewer structures than when global alignment is used, demonstrating that RDCs in disordered proteins are primarily determined by local structure. Finally, I made major improvements to the ENSEMBLE program which is used for calculating structural models of disordered states. I utilized large amounts of experimental data in order to calculate ensemble models of the Drosophila drkN SH3 domain unfolded state. Although highly heterogeneous and having broad molecular size distributions, the calculated ensembles have very different properties than expected for random or statistical coils and possess significant non-native alpha-helical structure and both native-like and non-native tertiary structure. This has significant implications for our understanding of the structural properties of protein disordered states in general.

The regulation of protein synthesis and protein folding is crucial for normal cell function. The endoplasmic reticulum (ER) has crucial roles in safeguarding the correct folding and assembling of proteins through the use of ER molecular chaperones. Homeostasis disruption of the ER leads to activation of the Unfolded Protein Response. The UPR is a three-arm pathway that plays a role in regulating ER stress and ultimately leads to cell survival or cell death if the cell fails to recover. There are three major proteins for sensing Endoplasmic Reticulum stress: RNA dependent protein kinase RNA like endoplasmic reticulum kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring ER-to-nucleus signal kinase 1 (IRE1). PERK activation leads to the phosphorylation of the α-subunit of the translation initiation factor eIF2α on Serine 51 in activating its function. EIF2α phosphorylation leads to up-regulation of GADD34 and GADD34 bind protein phosphatase 1 (PP1) to dephosphorylate eIF2α and brings the cell back into homeostasis. CReP, similar to GADD34, binds to PP1, to dephosphorylate eIF2α. The RVxF motif, RARA sequence, and amino acids throughout the GADD34 sequence play a role in PP1 binding and are essential for dephosphorylating eIF2α in cells. The first 180 amino acids of GADD34 play a role in subcellular localization whereas the first 300 amino acids of CReP play a role for localization to the ER. Early on in the UPR the levels of binding immunoglobulin protein (BiP), CHOP, GADD34, and CReP increase; however, the mRNA levels of CReP drop during the 24-HR Thapsigargin treated stage. Two primary proteins that bind CReP were COPS5 and SNAPIN. Understanding the UPR is important because the inhibiting of GADD34 and CReP have been shown to improve many neurodegenerative diseases.

Les chaperons moléculaires, une famille de protéines diverses en structure et taille, sont dédiés à accompagner, replier et protéger d’autres protéines afin qu’elles atteignent leur conformation finale et leur emplacement dans la cellule. Dans ce but, les chaperons moléculaires doivent être hautement spécialisés dans l’exécution de tâches spécifiques, telles que le repliement, le transport ou la désagrégation, et polyvalents dans leur motifs de reconnais- sance, afin de pouvoir interagir avec un grand nombre de protéines di érentes. Di érents chaperons moléculaires collaborent au sein de la cellule, formant ainsi un réseau complexe qui assure le contrôle de la qualité du protéome. Les interactions entre les di érents partenaires de ce réseau et entre les chap- erones et leurs substrats sont souvent dynamiques, ce qui rend leur obser- vation structurale particulièrement di cile pour les techniques de biologie structurale. Par conséquent, il y a à ce jour peu d’information sur les struc- tures et mécanismes d’interaction au sein des complexes chaperon-substrate. Dans cette thèse, je présente des études sur la structure, la dynamique et les interactions entre les substrats de deux chaperons moléculaires, en utilisant diverses méthodes biophysiques et in vivo.Dans la première partie, je montre que la chaperone TIM910, située dans l’espace inter-membranaire des mitochondries, lie ses substrats, des protéines membranaires destinées aux deux membranes mitochondriales, d’une manière très dynamique. Non seulement le complexe TIM910 est en échange constant entre les espèces monomèriques et hexameriques, mais aussi le substrat lié échange entre mulitples conformations à une échelle de millisecondes. Sur la base de la résonance magnétique nucléaire (RMN), de small-angle X-ray scat- tering (SAXS), de l’ultracentrifugation analytique (AUC) et des expériences mutationnelles in vivo et des tests fonctionnels d’import dans les mitochon- dries, je propose un modèle structurale de l’interaction entre le chaperon et la protéine membranaire. TIM910 lie ses substrats dans une poche hydrophobe à l’extérieur du chaperon. Cette interaction est modulaire et se fait avec un ou deux hexamères de TIM910, en fonction de la longueur du substrat.Dans la deuxième partie, nous avons étudié le comportement du récepteur N-terminal du unfoldase ClpC1 de M. tuberculosis en présence d’antibiotiques et de ligands di érents. Le domaine N-terminal de ClpC1 est le site de liai- son de divers antibiotiques nouveaux contre M. tuberculosis. L’antibiotique Cyclomarin A supprime complètement la dynamique induite par le ligand arginine-phosphate. Nous proposons que cette suppression de la dynamique soit le principe fondamental du mécanisme d’action de cet antibiotique.Dans les deux cas, les structures X-ray des chaperons dans leur état apo et la structure de ClpC-NTD liée à des antibiotiques étaient disponibles, mais ces structures statiques ne su sent pas pour expliquer le mécanisme d’action. La structure X-ray de TIM910 n’a pas fourni d’ indication sur l’endroit ou la façon dont les substrats sont liés. De même, les structures X-ray du domaine N-terminal de apo et de Cyclomarine A de ClpC1 ne présentent que des di érences de structure mineures. Les deux exemples montrent que les données structurelles statiques souvent ne permettent pas d’expliquer le fonctionnement d’un système moléculaire, donc la combinaison de di érentes techniques et le développement de nouvelles méthodes pour étudier les complexes chaperon-substrat sont primordiaux pour comprendre leur fonction. / The diverse group of molecular chaperones is dedicated to accompany, fold and protect other proteins until they reach their final conformation and loca- tion inside the cell. To this end, molecular chaperones need to be specialized in performing specific tasks, like folding, transport or disaggregation, and versatile in their recognition pattern to engage many di erent client pro- teins. Moreover, molecular chaperones need to be able to interact with each other and with other components of the protein quality control system in a complex network. Interactions between the di erent partners in this network and between the substrate and the chaperone are often dynamic processes, which are especially di cult to study using standard structural biology tech- niques. Consequently, structural data on chaperone/substrate complexes are sparse, and the mechanisms of chaperone action are poorly understood. In this thesis I present investigations of the structure, dynamics and substrate- interactions of two molecular chaperones, using various biophysical and in vivo methods.In the first part I show that the mitochondrial membrane protein chap- erone TIM910 binds its substrates in a highly dynamic manner. Not only is the TIM910 complex in constant exchange between monomeric and hex- americ species, but also the bound substrate samples multiple conformations on a millisecond timescale. Based on nuclear magnetic resonance (NMR), small-angle X-ray scattering (SAXS), analytical ultracentrifugation (AUC) and in vivo mutational experiments I propose a structural model of the chap- erone/membrane protein interaction. TIM910 binds its substrates in a hy- drophobic pocket on the exterior of the chaperone in a modular fashion, where the number of TIM910 complexes bound depends on the length of the substrate.In the second part I studied the behavior of the N-terminal receptor do- main of the ClpC1 unfoldase from M.tuberculosis in the presence of di erent antibiotics and ligands. The N-terminal domain of ClpC1 is the binding site for various new antibiotics against M.tuberculosis. The antibiotic cyclomarin completely abolishes dynamics induced by the ligand arginine-phosphate. We propose that this suppression of dynamics is the underlying principle for the mechanism of action of this antibiotic.In both cases X-ray structures of the apo or antibiotic bound form were available, but not su cient to explain the mechanism of action. The X- ray structure of TIM910 provided no evidence on where or how substrates are bound. Likewise, X-ray structures of the apo and cyclomarin-bound N-terminal domain of ClpC1 show only minor di erences in structure.Both examples show that static structural data is often not enough to explain how a molecular system works, and only the combination of di er- ent techniques, including newly developed methods enable the atomic-level understanding of chaperone/substrate complexes.

The Lon protease is a mitochondrial matrix quality-control protease belonging to the family of AAA+ proteins (ATPases associated with many cellular activities). We had previously found Lon to be upregulated in lung tumours with a non-angiogenic phenotype in a microarray study comparing these to conventional angiogenic tumours. In this project I set out to investigate whether Lon had any role in modulating the hypoxic response of tumour cells. Using a novel monoclonal antibody against Lon, I found that upregulation of Lon was present in breast and lung tumours and that higher levels of Lon are correlated with shorter overall survival in breast cancer patients. Targeting Lon with siRNA and shRNA in tumour cell lines reduced the normoxic and hypoxic stabilisation of HIF-α subunits. This is mediated through a mechanism independent of the activity of HIF-prolyl hydroxylases and independent of any changes in mitochondrial transcription. I found that the pre-imported form of Lon could bind and chaperone VHL in the cytoplasm potentially modulating VHL activity. In cell lines and human tumours, I observed that the proline-hydroxylated form of HIF-1α is induced by hypoxia and the hydroxylated form of HIF-1α is associated with shorter overall survival in breast cancer patients. This observation supports the notion that higher levels of Lon is associated with poor survival by downregulating VHL leading to higher levels of hydroxylated HIF. Finally I show that targeting Lon in cell lines is able to inhibit growth in a cell-line dependent fashion and partially reverses the Warburg effect, increasing oxygen consumption and reducing lactate production. In conclusion, I have demonstrated the broad therapeutic potential of targeting the Lon protease in tumours and highlighted a mechanism of post-hydroxylation HIF-regulation that has not been previously recognised in VHL competent tumours.

Près de 40% des protéines présentes dans les cellules sont prédites partiellement ou complètement désordonnées. Ces protéines dépourvues de structure tridimensionnelle à l'état natif sont impliquées dans de nombreux mécanismes biologiques, la flexibilité jouant un rôle moteur dans les mécanismes de reconnaissance moléculaire. La prise en considération de l'existence de flexibilité au sein des protéines et des interactions protéines-protéines a nécessité le renouvellement de nos connaissances, de notre appréhension des fonctions biologiques ainsi que des approches pour étudier et interpréter ces phénomènes. La méthode retenue pour étudier ces transitions conformationnelles est la spectroscopie par résonance magnétique nucléaire. Elle dispose d'une sensibilité unique, d'une résolution à l'échelle atomique et permet par diverses expériences d'accéder à l'ensemble des échelles de temps définissant les mouvements de ces protéines. Nous combinons ces mesures expérimentales à un modèle statistique représentant l'ensemble du paysage énergétique des protéines désordonnées : la description par ensemble explicite de structures. Ce modèle est une représentation discrète des différents états échantillonnés par ces protéines. Il permet, combinant les déplacements chimiques, les couplages dipolaires et la relaxation paramagnétique, de développer une description moléculaire de l'état déplié en caractérisant à la fois l'information locale et l'information à longue portée présente dans les protéines intrinsèquement désordonnées. / Around 40% of the human genome does not fold into stable three-dimensional structures but are either unfolded, or contain unfolded regions of significant length. The inherent flexibility of this class of proteins is essential for their function in a vast range of biomolecular process such as molecular recognition. In order to take into account the specificity of these interactions, it has been necessary to invent new approaches to study and interpret their behaviour. Nuclear magnetic resonance spectroscopy is a unique atomic resolution probe which is sensitive to a very large range of time scales. We combine experimental NMR data with a statistical model describing the energy landscape of unfolded state : the explicit ensemble description. This model is a discrete representation of the different states of theses proteins. Combining chemical shifts, residual dipolar couplings and paramagnetic relaxation enhancement, it is then possible to develop a molecular description of the unfolded state caracterising both the local and long-range information of intrinsically disordered proteins.