The last decade has seen substantial advances in single-molecule tracking methods with nano-metre level precision. A powerful tool in single-molecule tracking is fluorescence imaging. One particular application, total internal reflection microscopy, can capture biological processes at high contrast video rate imaging at the single-particle level. This thesis presents methodologically novel methods in analysing single particle tracking data. Presented here is an application of a Bayesian statistical approach that can discriminate between the different diffusive modes that appear with the presence of membrane architecture. This algorithm is denoted BARD; a Bayesian Analysis to Ranking Diffusion. These algorithms are applied to a total internal fluorescence microscopy based experimental data of a novel membrane probe in Escherichia coli. This probe is a plasmid expressed, non-native membrane integrating trans-membrane helix and thus acts as an ideal protein based probe under no specific native control. Two experiments were performed using a combination of varying helix probe size and growth temperature experiments effectively altering the transition temperature of the membrane. These data are suggestive of a passive partitioning of the helix protein into mobile and immobile domains that emerge from the underlying phase behaviour of the membrane.

Artificial membranes composed of natural lipids are not stable when exposed to air/vacuum, surfactant, organic solvent, etc. Polymerizable lipids provide an opportunity to broaden the use of lipid membranes to study ligand-receptor pairs under harsh experimental conditions. This dissertation presents the utilization of polymerizable lipids in matrix assisted laser desorption and ionization-mass spectrometry (MALDI-TOF MS) for analysis of ligands bound to membrane receptors. This platform may be applied to rapid drug-screening for membrane receptors including transmembrane proteins. Bacterial toxins and their membrane receptors were used as model ligand-receptor pairs to demonstrate the feasibility of using polymerizable lipids to detect and identify ligands by MALDI-TOF MS. Cholera toxin B (CTB) was successfully detected bound to polymerized lipid membranes with incorporation of its membrane receptor, GM1, while no CTB was detected in non-polymerizable lipid membranes. This affinity capture platform based on poly(lipid) showed a high resistance to interferences. On-plate digestion of bound CTB was performed and 57% amino acid sequence coverage was achieved. Total internal reflection fluorescence microscopy (TIRF-M) was applied to compare CTB-GM1 binding affinity in polymerized and unpolymerized membranes. Under a static flow system, the binding between CTB and GM1 was found to be stronger in polymerized membranes than other membranes. However, the ligand concentration under a static flow system is not in excess and the apparent binding affinity is likely to be significantly different than the true value. The true binding affinity can be approached under a continuous flow system, however equilibration time was found to be too long to address experimentally. Membrane fluidity, which may be required to maintain the membrane receptor activity, is suppressed in poly(lipid) membranes compared to unpolymerized membranes. In order to maintain fluidity, a non-polymerizable lipid was mixed into a polymerized lipid. Fluorescence recovery after photobleaching (FRAP) data showed that fluidity of membrane composed of the mixed lipid was maintained compared to pure poly(lipid). Phase segregation of polymerized lipid and non-polymerizable lipid was detected by atomic force microscopy (AFM). CTB bound to GM1 in mixed lipid membranes was detected by MALDI-MS, indicating the mixed lipid membranes retain stability under MALDI-MS analysis conditions.

Artificial membranes composed of natural lipids are not stable when exposed to air/vacuum, surfactant, organic solvent, etc. Polymerizable lipids provide an opportunity to broaden the use of lipid membranes to study ligand-receptor pairs under harsh experimental conditions. This dissertation presents the utilization of polymerizable lipids in matrix assisted laser desorption and ionization-mass spectrometry (MALDI-TOF MS) for analysis of ligands bound to membrane receptors. This platform may be applied to rapid drug-screening for membrane receptors including transmembrane proteins. Bacterial toxins and their membrane receptors were used as model ligand-receptor pairs to demonstrate the feasibility of using polymerizable lipids to detect and identify ligands by MALDI-TOF MS. Cholera toxin B (CTB) was successfully detected bound to polymerized lipid membranes with incorporation of its membrane receptor, GM1, while no CTB was detected in non-polymerizable lipid membranes. This affinity capture platform based on poly(lipid) showed a high resistance to interferences. On-plate digestion of bound CTB was performed and 57% amino acid sequence coverage was achieved. Total internal reflection fluorescence microscopy (TIRF-M) was applied to compare CTB-GM1 binding affinity in polymerized and unpolymerized membranes. Under a static flow system, the binding between CTB and GM1 was found to be stronger in polymerized membranes than other membranes. However, the ligand concentration under a static flow system is not in excess and the apparent binding affinity is likely to be significantly different than the true value. The true binding affinity can be approached under a continuous flow system, however equilibration time was found to be too long to address experimentally. Membrane fluidity, which may be required to maintain the membrane receptor activity, is suppressed in poly(lipid) membranes compared to unpolymerized membranes. In order to maintain fluidity, a non-polymerizable lipid was mixed into a polymerized lipid. Fluorescence recovery after photobleaching (FRAP) data showed that fluidity of membrane composed of the mixed lipid was maintained compared to pure poly(lipid). Phase segregation of polymerized lipid and non-polymerizable lipid was detected by atomic force microscopy (AFM). CTB bound to GM1 in mixed lipid membranes was detected by MALDI-MS, indicating the mixed lipid membranes retain stability under MALDI-MS analysis conditions.

The Min protein system regulates cell division in the bacterium Escherichia coli. The protein MinD undergoes a pole-to-pole oscillation, antagonizing formation of the division septum at the cell poles, thereby confining the septum formation to the mid-cell. The MinD oscillation period is 40 s at room temperature in healthy cells, but has been shown to be sensitive to stress on the cell. By fluorescently labeling MinD with green fluorescent protein (GFP), we are able to measure the MinD oscillation period as an in situ metric of cell viability using high resolution total internal reflection fluorescence (TIRF) microscopy. We have made several improvements to the method by which we measure and analyse the MinD oscillation period. A microscopy flow cell was designed and constructed and it provides temperature control and stability to a precision of 0.05 °C in addition to allowing controlled addition of bacterial cells and reagents of interest to the imaging region of the flow cell. This flow cell enabled us to make a precise measurement of the temperature dependence of the MinD oscillation period, for which we observed an Arrhenius dependence with an activation energy of 11.8 kcal/mol. We developed a centroid-tracking method, performed in a custom MATLAB program, to extract the values of the MinD oscillation periods from our time series of TIRF microscopy images. We measured the effect on the MinD oscillation period of exposure to the cationic antimicrobial peptide polymyxin B (PMB) and the related compound polymyxin B nonapeptide (PMBN), which does not have antimicrobial activity. Exposure to PMB resulted in a 60% increase in the average MinD oscillation period tau, whereas exposure to PMBN resulted in an 20% decrease in tau. After exposure to PMB and PMBN, we measured the Arrhenius temperature dependence of the MinD temperature dependence and calculated the associated activation energy Ea. We found that exposure to PMB resulted in a 40% increase in Ea, whereas exposure to PMBN did not significantly change the value of Ea. These results indicate that careful measurements of the MinD oscillation can yield information that can be helpful in evaluating the mechanism of action of antimicrobial compounds. / Natural Sciences and Engineering Research Council

The cyclical interaction between the force-generating protein myosin and actin is the mechanism responsible for muscle contraction among all muscle types. Cardiac muscle contraction is tightly controlled to ensure that blood pumps effectively and efficiently from the heart to peripheral organs. Mutations in various cardiac proteins can lead to cardiac dysfunction and a number of cardiomyopathies. The first part of this dissertation studies two disease-linked mutations in the regulatory light chain of the cardiac myosin molecule, D166V and K104E, and assesses the kinetic and mechanochemical effects of the mutations via the in vitro motility assay. The data show that D166V mutant myosin force generation is reduced compared to wild type, and exogenous phosphorylation of the mutant light chain rescues force generation. In contrast, the K104E mutation showed no deficit in force production but exhibited increased calcium sensitivity of activation. These results are consistent with contractile defects associated with cardiomyopathies caused by various mutation-induced changes to protein function and mechanism of interaction. The second part uncovers the actin-binding mechanism of one of the chief muscle regulatory proteins tropomyosin. In cardiac and skeletal muscle, tropomyosin and troponin modulate muscle contraction. Tropomyosin binds along the length of actin filaments and blocks myosin-binding sites. Following an excitatory stimulus, calcium binds troponin and causes tropomyosin to shift its position on actin, allowing myosin to bind. The precise mechanism of how tropomyosin monomers with low actin affinity bind to form a stably bound, high affinity chain is unknown. By directly observing fluorescently labeled tropomyosin binding to actin filaments, it was shown that tropomyosin molecules bind randomly along the actin filament. Subsequent monomer binding, and formation of tropomyosin end-to-end bonds, increases the probability of sustained chain growth by decreasing the probability of detachment prior to additional monomer binding. Tropomyosin molecules added to the growing chain at approximately 100 monomers/(μM*s). Different tropomyosin isoforms segregate to distinct functional and structural regions of cells. The last chapter presents data that show spatial segregation of two different tropomyosin isoforms on actin filaments. This suggests that tropomyosin sorting in cells is, at least partly, an intrinsic property of the binding mechanism.

Toll-like receptor (TLR)-4 is a pattern recognition receptor (PRR) that recognises the pathogen-associated molecular pattern (PAMP) lipopolysaccharide (LPS) produced by Gram-negative bacteria. LPS binds to Myeloid differentiation 2 (MD-2)/TLR-4 heterodimers, driving their dimerisation and inducing a conformational change of the intracellular TLR-4 toll/interleukin-1 receptor (TIR) domains. The adaptor protein Myeloid differentiation primary response gene 88 (MyD88)-adaptor-like (Mal)/TIR domain-containing adaptor protein (TIRAP) then binds to the TIR domains of TLR-4 and acts as a bridge for MyD88 which goes on to form the myddosome, a large protein complex of six to eight MyD88 molecules and four Interleukin-1 receptor- associated kinase (IRAK) 4 and four IRAK1/2 molecules. This triggers a signalling cascade which results in nuclear factor (NF)-κB transcription factor activation and production of pro-inflammatory effector molecules such as the cytokine Tumour Necrosis Factor (TNF)-α. Upon activation TLR-4 is also endocytosed where it interacts with a second set of adaptor proteins TIR-domain-containing adaptor- inducing interferon (IFN)-β (TRIF)-related adaptor molecule (TRAM) and TRIF to initiate the type I IFN response. How TLR-4 dimerisation results in the formation of the oligomeric myddosome is not fully understood, but it is possible that the stoichiometry of Mal/TIRAP may be important in the formation of this protein complex. The aim of my thesis was to determine the stoichiometry of Mal/TIRAP at the plasma membrane of immortalised bone marrow derived macrophages (iBMDMs) and whether this stoichiometry changes upon stimulation with different TLR-4 ligands. To investigate Mal/TIRAP stoichiometry I first developed a viral transduction experimental cell model to visualise fluorescently labelled Mal/TIRAP. Mal/TIRAP-/- iBMDMs were lentivirally transduced with a Mal/TIRAPHALO construct. The halotag was fluorescently labelled then the cells were stimulated with TLR-4 ligands, such as LPS, fixed at different time points, then imaged. Total internal reflection fluorescence (TIRF) microscopy was used to image the plasma membrane and photobleaching experiments performed to determine Mal/TIRAP stoichiometry. I developed a computer-based analysis pipeline to analyse the resulting photobleaching data. Under resting conditions, Mal/TIRAP is present at the plasma membrane in clusters of approximately ten Mal/TIRAP molecules per cluster. After five minutes of stimulation with 10 ng/ml LPS Mal/TIRAP redistributes into cluster sizes of approximately six, twelve and much larger. After ten and fifteen minutes stimulation with 10 ng/ml LPS the clusters return to the resting size of approximately ten Mal/TIRAP molecules per cluster with a few much larger clusters remaining present. This confirms the rapid time frame within which TLR-4 signalling occurs at the plasma membrane and is consistent with myddosome stoichiometry of six MyD88 molecules or proposed super myddosomes of twelve MyD88 molecules. The computer-based analysis pipeline developed can be used to analyse any protein of interest at the plasma membrane. Protein ligands have also been found to activate TLR-4; for example allergens, such as Fel d 1 and Der p 2, as well as endogenous damage associated molecular patterns (DAMPs), such as extracellular matrix (ECM) proteins, for example fragments of fibronectin and tenascin-C. The mechanism by which these proteins interact with TLR-4 and induce signalling is unclear. Proteins from the ECM (fragments FNIII1c, FNIII13-14, FNIII9-E and FNIII9-E-14 from fibronectin and the fibrinogen-like globe (FBG) domain of tenascin-C) were tested using a transient transfection assay in HEK293 cells and shown to activate TLR-4. In conclusion, I have developed new tools and methodology to investigate how TLR-4 signals in response to LPS and DAMPs in living cells. Whether DAMP- activated TLR-4 forms similar signalling complexes to those induced by LPS will form part of a future study.

To respond to rapidly-changing stresses in their environment, bacterial cells must be able to sense a variety of chemical cues and respond to them by activating the relevant genes. The lysine riboswitch is a short RNA motif, located just upstream of a gene encoding a lysine biosynthesis protein, that suppresses the expression of that gene when sufficient lysine is present in the cell. It acts by binding a lysine monomer in a region called the aptamer, which in turn rearranges an adjacent domain called the expression platform, sequestering the ‘start' sequence of the gene and preventing it from being transcribed. In this thesis, the lysine riboswitch's ligand-binding transition is studied using single-molecule fluorescence microscopy, optical tweezers, and a hybrid optical force/fluorescence technique. Förster Resonance Energy Transfer (FRET) is used with a fluorescently-labeled aptamer to show that it has a previously-undescribed, partially-folded structural state with enhanced ligand affinity compared to the unfolded structure. The Mg²⁺ dependence of the transition between these states is shown to resolve existing debates in the literature about the sensitivity of the riboswitch. The kinetics of the folding transition are explored using FRET, optical force, and hybrid ‘Fleezers' to map the free energy landscape of ligand binding and show that the ligand itself promotes transitions into the aptamer's folded state, a so-called ‘induced fit' mechanism rare among riboswitches. Finally, high-resolution optical tweezers are used to explore the link between the aptamer's secondary structure (the sequence of paired nucleotides) and its tertiary structure (three-dimensional folding) to illuminate the role of ligand binding in gene regulation, which depends on the equilibrium between competing secondary structures. Hybrid biophysical techniques like optical force/fluorescence microscopy are shown to be indispensable for addressing all the states in the reaction pathways of complex biomolecules like riboswitches and for discriminating between multiple levels of structure formation and interaction with the environment. Not only do the results presented here shed light on the RNA folding problem, particularly the role of tertiary structure in determining the minimum-energy configuration of an RNA sequence, but they could have implications for biomedical research, as the lysine riboswitch has already been shown to be a potential target for next-generation antibiotics.

Eukaryotic flagella and cilia are long rod-like extensions of cells, which play a fundamental role in single cell movement, as well as in fluid transport. Flagella and cilia contain a highly evolutionary conserved mechanical structure called the axoneme. The motion of the flagellum is generated by dynein motor proteins, located all along the length of the axonemal structure. Fluorescent ATP analogs have been a useful tool to study ATPase activity of various motor proteins. \acrfull{mant} has been previously used to probe the activity of various ATPases, including dynein. It has been shown by various authors, that MANT-ATP supports dynein activity as well as the axonemal beat. However, direct observations of binding to the axonemal structure were not previously reported. Using highly sensitive fluorescent microscopy to monitor the binding of the fluorescent ATP analog, I probed dynein activity directly in the immobilized intact axoneme for the first time. To understand these kinetics a kinetic model was developed. By fitting this model to experimental data I was able to identify ATP-binding sites with distinct kinetic properties in the axoneme. I report a turnover rate of k = 0.02 s−1 at 1μM mant-ATP for dynein. Moreover, I discovered that there is binding of the ATP analog to the axoneme with a much higher rate of k = 11 s−1 at 1μM mant-ATP. By the application of this method to axonemes with reduced dyneins, it has been identified that the slow rate belongs to dynein.

<p>Bid is a member of the Bcl-2 family of proteins, which are known as the regula- tors of apoptosis. Bid recruits Bax, another Bcl-2 family protein, which forms large oligomers that permeabilize the mitochonrdial outer membrane during apoptosis. In this thesis, Bid complexes embedded in a mitochondria-like supported lipid bilayer were investigated using single molecule fluorescence techniques. The bilayer, con- taining a lipophilic tracer, was formed on a mica surface and ATTO647 labelled Bid was added to it. For experiments where the effect of Bax on Bid complexes was investigated, a wild type Bax or a HiLyte488 labelled Bax was added as well. The protein-bilayer sample was imaged using total internal reflection fluorescence (TIRF). The formation of a fluid bilayer was confirmed by the observation of the lateral diffusion of DiD. Single particle tracking of the lipid molecules was used to measure the diffusion coefficent of DiD which was determined to be 2.2μm2 /s. The TIRF images also revealed two populations of Bid complexes, immobile and mobile. The diffusion coefficient of the observed Bid complexes was determined to be about three times slower than that of DiD (0.8±0.5μm2 /s). This provides evidence that mobile Bid is embedded in the bilayer. Image analysis of immobile Bid complexes showed a step-wise decrease in the fluorescence intensity due to photobleaching. The oligomeric distribution of the immobile Bid complexes was determined from the num- ber of steps, which corresponds to the number of particles in each complex. From these distributions it was concluded that the imaged immobile Bid existed mainly as monomers. However dimer and trimer complexes of Bid were also observed. The detected oligomeric distribution was not affected by the presence of either wild type Bax or Hilyte488 Bax. However Bid was imaged for the first time participating in Bax complexes. The acquired results somewhat differ from what had been observed in confocal imaging of the same samples, where mostly larger Bid complexes (dimers and up) were detected. We attribute the difference to the superior sensitivity of the TIRF method presented here.</p> / Master of Science (MSc)

Cette thèse s'intéresse à deux problèmes rencontrés en traitement du signal et des images. Le premierconcerne la reconstruction 3D de structures biologiques à partir d'acquisitions multi-angles enmicroscopie par réflexion totale interne (MA-TIRF). Dans ce contexte, nous proposons de résoudre leproblème inverse avec une approche variationnelle et étudions l'effet de la régularisation. Une batteried'expériences, simples à mettre en oeuvre, sont ensuite proposées pour étalonner le système et valider lemodèle utilisé. La méthode proposée s'est montrée être en mesure de reconstruire avec précision unéchantillon phantom de géométrie connue sur une épaisseur de 400 nm, de co-localiser deux moléculesfluorescentes marquant les mêmes structures biologiques et d'observer des phénomènes biologiquesconnus, le tout avec une résolution axiale de l'ordre de 20 nm. La deuxième partie de cette thèseconsidère plus précisément la régularisation l0 et la minimisation du critère moindres carrés pénalisé (l2-l0) dans le contexte des relaxations continues exactes de cette fonctionnelle. Nous proposons dans unpremier temps la pénalité CEL0 (Continuous Exact l0) résultant en une relaxation de la fonctionnelle l2-l0 préservant ses minimiseurs globaux et pour laquelle de tout minimiseur local on peut définir unminimiseur local de l2-l0 par un simple seuillage. Par ailleurs, nous montrons que cette relaxation éliminedes minimiseurs locaux de la fonctionnelle initiale. La minimisation de cette fonctionnelle avec desalgorithmes d'optimisation non-convexe est ensuite utilisée pour différentes applications montrantl'intérêt de la minimisation de la relaxation par rapport à une minimisation directe du critère l2-l0. Enfin,une vue unifiée des pénalités continues de la littérature est proposée dans ce contexte de reformulationexacte du problème / This thesis is devoted to two problems encountered in signal and image processing. The first oneconcerns the 3D reconstruction of biological structures from multi-angle total interval reflectionfluorescence microscopy (MA-TIRF). Within this context, we propose to tackle the inverse problem byusing a variational approach and we analyze the effect of the regularization. A set of simple experimentsis then proposed to both calibrate the system and validate the used model. The proposed method hasbeen shown to be able to reconstruct precisely a phantom sample of known geometry on a 400 nmdepth layer, to co-localize two fluorescent molecules used to mark the same biological structures andalso to observe known biological phenomena, everything with an axial resolution of 20 nm. The secondpart of this thesis considers more precisely the l0 regularization and the minimization of the penalizedleast squares criteria (l2-l0) within the context of exact continuous relaxations of this functional. Firstly,we propose the Continuous Exact l0 (CEL0) penalty leading to a relaxation of the l2-l0 functional whichpreserves its global minimizers and for which from each local minimizer we can define a local minimizerof l2-l0 by a simple thresholding. Moreover, we show that this relaxed functional eliminates some localminimizers of the initial functional. The minimization of this functional with nonsmooth nonconvexalgorithms is then used on various applications showing the interest of minimizing the relaxation incontrast to a direct minimization of the l2-l0 criteria. Finally we propose a unified view of continuouspenalties of the literature within this exact problem reformulation framework