L’avènement de la microscopie STED (Stimulated Emission Depletion) a bouleversé le domaine desneurosciences du au fait que beaucoup de structures neuronale, tels que les épines dendritiques, lesaxones ou les processus astrocytaires, ne peuvent pas être correctement résolu en microscopiephotonique classique. La microscopie 2-photon est une technique d’imagerie photonique très largement utilisée dans le domaine des neurosciences car elle permet d’imager les événements dynamique en profondeur dans le tissu cérébral, offrant un excellent sectionnement optique et une meilleure profondeur de pénétration. Cependant, la résolution spatiale de cette approche est limitée autour de 0.5 μm, la rendant inappropriée pour étudier les détails morphologiques des neurones et synapses. Le but de mon travail de thèse était à A) développer un microscope qui permet d'améliorer l'imagerie 2-photon en la combinant avec la microscopie STED et B) démontrer son potentiel pour l'imagerie à l'échelle nanométrique de processus neuronaux dynamiques dans des tranches de cerveau aigus et in vivo. Le nouveau microscope permet d'obtenir une résolution spatiale latérale de ~ 50 nm à des profondeurs d'imagerie de ~ 50 μm dans du tissu cérébral vivant. Il fonctionne avec des fluorophores verts, y compris les protéines fluorescentes communes telles que la GFP et YFP, offrant le contraste de deux couleurs basé sur la détection spectrale et linéaire ‘unmixing’. S’agissant d’un microscope droit, utilisant un objectif à immersion ayant une grande distance de travail, nous avons pu incorporer des techniques électrophysiologiques comme patch-clamp et ajouter une plateforme pour l'imagerie in vivo. J’ai utilise ce nouveau microscope pour imager des processus neuronaux fins et leur dynamique à l’échelle nanométrique dans différent types de préparations et des régions différentes du cerveau. J’ai pu révéler des nouvelles caractéristiques morphologique des dendrites et épines. En outre, j'ai exploré différentes stratégies de marquage pour pouvoir utiliser la microscopie STED pour imager le trafic des protéines et de leur dynamique à l'échelle nanométrique dans des tranches de cerveau. / The advent of STED microscopy has created a lot of excitement in the field of neuroscience becausemany important neuronal structures, such as dendritic spines, axonal shafts or astroglial processes,cannot be properly resolved by regular light microscopy techniques. Two-photon fluorescence microscopy is a widely used imaging technique in neuroscience because it permits imaging dynamic events deep inside light-scattering brain tissue, providing high optical sectioning and depth penetration. However, the spatial resolution of this approach is limited to around half a micron, and hence is inadequate for revealing many morphological details of neurons and synapses. The aim of my PhD work was to A) develop a microscope that improves on two-photon imaging by combining it with STED microscopy and to B) demonstrate its potential for nanoscale imaging of dynamic neural processes in acute brain slices and in vivo. The new microscope achieves a lateral spatial resolution of ~50 nm at imaging depths of ~50 μm in living brain slices. It works with green fluorophores, including common fluorescent proteins like GFP and YFP, offering two-color contrast based on spectral detection and linear unmixing. Because of its upright design using a long working distance water-immersion objective, it was possible to incorporate electrophysiological techniques like patch-clamping or to add a stage for in vivo imaging. I have used the new microscope to image fine neural processes and their nanoscale dynamics in different experimental preparations and brain regions, revealing new and interesting morphological features of dendrites and spines. In addition, I have explored different labeling strategies to be able to use STED microscopy for visualizing protein trafficking and dynamics at the nanoscale in brain slices.

Movement is necessary for the survival of most animals. The spinal cord contains neuronal networks that are capable of motor coordination and of producing different movements. In particular, a very reduced neuronal network in the spinal cord can produce simple rhythmic outputs even in the absence of descending or sensory inputs. This basic circuit was discovered by Thomas Graham Brown (reported in 1911) and is termed central pattern generator. For over a century a large number of studies have been carried out in order to identify the neuronal components that are part of these networks. In project 1 we focused on Renshaw cells, which are a population of spinal interneurons expressing the alpha-2 subunit of the nicotinic acetylcholine receptors (Chrna2). Renshaw cells are the only identified central targets for motor neuron inputs, and in turn they mediate inhibition of the motor neurons. We analyzed the activity pattern of Renshaw cells on a cell-population level in neonates when the circuit is still developing. At segment 1 of the lumbar spinal cord, Renshaw cells show significantly greater activity response to functional sensory and motor inputs from rostral compared to the caudal segments. Contrarily, the suppression of the monosynaptic stretch reflex was more pronounced when caudal roots were stimulated. Our data underline the importance of sensory input during motor circuit development and help to understand the functional organization of Renshaw cell connectivity. Several neurons that play distinct roles in locomotor central pattern generation have been identified with the help of genetics. For instance, the V0 population of spinal interneurons are identified by the expression of transcription factor developing brain homeobox 1 (Dbx1). V0 neurons are necessary for producing an alternating rhythm at all locomotor speeds. In project 2 we have looked at a population of dorsally derived ventrally projecting interneurons that express the transcription factor doublesex and mab-3 related transcription factor 3 (Dmrt3). On a behavioral level Dmrt3 neurons are involved in regulating coordination across different locomotor speeds. On a microcircuit level, we have shown that individual Dmrt3 neurons show distinct frequencies of oscillations for a constant locomotor rhythm. In addition, removal of inhibitory neurotransmission from Dmrt3 neurons results in uncoupling of rhythm in motor neurons. In project 3 the activity patterns in populations of flexor related motor neurons are characterized during fictive locomotion in neonatal mice. An interesting and intriguing finding in project 3 is the presence of multiple rhythmicities in motor neurons. Multiple rhythmicities are seen even when the locomotor output shows a constant frequency.

In recent years, the confocal and two photon microscopes have become ubiquitous tools in life science laboratories. The reason for this is that both these systems can acquire three dimensional image data from biological specimens. Specifically, this is done by acquiring a series of two-dimensional images from a set of equally spaced planes within the specimen. The resulting image stack can be manipulated and displayed on a computer to reveal a wealth of information. These systems can also be used in time lapse studies to monitor the dynamical behaviour of specimens by recording a number of image stacks at a sequence of time points. The time resolution in this situation is, however, limited by the maximum speed at which each constituent image stack can be acquired. Various techniques have emerged to speed up image acquisition and in most practical implementations a single, in-focus, image can be acquired very quickly. However, the real bottleneck in three dimensional imaging is the process of refocusing the system to image different planes. This is commonly done by physically changing the distance between the specimen and imaging lens, which is a relatively slow process. It is clear with the ever-increasing need to image biologically relevant specimens quickly that the speed limitation imposed by the refocusing process must be overcome. This thesis concerns the acquisition of data from a range of specimen depths without requiring the specimen to be moved. A new technique is demonstrated for two photon microscopy that enables data from a whole range of specimen depths to be acquired simultaneously so that a single two dimensional scan records extended depth of field image data directly. This circumvents the need to acquire a full three dimensional image stack and hence leads to a significant improvement in the temporal resolution for acquiring such data by more than an order of magnitude. In the remainder of this thesis, a new microscope architecture is presented that enables scanning to be carried out in three dimensions at high speed without moving the objective lens or specimen. Aberrations introduced by the objective lens are compensated by the introduction of an equal and opposite aberration with a second lens within the system enabling diffraction limited performance over a large range of specimen depths. Focusing is achieved by moving a very small mirror, allowing axial scan rates of several kHz; an improvement of some two orders of magnitude. This approach is extremely general and can be applied to any form of optical microscope with the very great advantage that the specimen is not disturbed. This technique is developed theoretically and experimental results are shown that demonstrate its potential application to a broad range of sectioning methods in microscopy.

<p>Synaptic plasticity is the leading candidate for the cellular/molecular basis of learning and memory. One of the key molecules involved in synaptic plasticity is Calcium/calmodulin-dependent Kinase II (CaMKII). Synaptic plasticity can be expressed at a single dendritic spine independent of its neighboring dendritic spines. Here, we investigated how long the activity of CaMKII lasts during synaptic plasticity of single dendritic spines. We found that CaMKII activity lasted ~2 minutes during synaptic plasticity and was restricted to the dendritic spines undergoing synaptic plasticity while nearby dendritic spines did not show any change in the level of CaMKII activity. Our experimental data argue against the persistent activation of CaMKII in dendritic spines undergoing synaptic plasticity and suggest that the activity of CaMKII is a spine-specific biochemical signal necessary for synapse-specificity of synaptic plasticity. We provide a biophysical explanation of how spine-specific CaMKII activation can be achieved during synaptic plasticity. We also found that CaMKII is activated by highly localized calcium influx in the proximity of Voltage-dependent Calcium Channels (VDCCs) and a different set of VDCCs and their respective Ca2+ nanodomains are responsible for the differential activation of CaMKII between dendritic spines and dendritic shafts.</p> / Dissertation

In vivo real-time data of ultrasound and microbubble assisted blood-brain barrier disruption is centrally based on low-resolution magnetic resonance images. Additional information can be gained using online microscopic monitoring. This study presents the first ever in vivo two-photon microscopy, four-dimensional data sets of ultrasound and microbubble assisted blood-brain barrier disruption. It characterized the threshold pressures and mechanical index needed to disrupt the vasculature with 800 kHz ultrasound, and found three different leakage constants from the compromised vasculature. Furthermore, using numerical models, an ultrasound array was designed and optimized to perform specifically with our two-photon microscope. It was fabricated, fully characterized, and its performance met both the required pressure field profile and the pressure values needed for our in vivo two-photon microscopy experiments. This array is an important step in microscopically characterizing ultrasound and microbubble assisted blood-brain barrier disruption.

The current work investigated the stimulation induced spatial response of the cerebral microvascular network by reconstruction of the 3D microvascular morphology from in vivo two photon fluorescence microscopy (2PFM) volumes using an automated, model based tracking algorithm. In vivo 2PFM imaging of the vasculature in the forelimb representation of the primary somatosensory cortex of alpha-chloralose anesthetized rats was achieved via implantation of a closed cranial window, and intravascular injection of fluorescent dextran. The dilatory and constrictory responses of the cerebral microvascular network to functional stimulation were heterogeneous and depended on resting vascular radius and response latency. Capillaries experienced large relative dilations and constrictions, but the larger vessel absolute volume changes dominated the overall network cerebral blood volume change.

In vivo real-time data of ultrasound and microbubble assisted blood-brain barrier disruption is centrally based on low-resolution magnetic resonance images. Additional information can be gained using online microscopic monitoring. This study presents the first ever in vivo two-photon microscopy, four-dimensional data sets of ultrasound and microbubble assisted blood-brain barrier disruption. It characterized the threshold pressures and mechanical index needed to disrupt the vasculature with 800 kHz ultrasound, and found three different leakage constants from the compromised vasculature. Furthermore, using numerical models, an ultrasound array was designed and optimized to perform specifically with our two-photon microscope. It was fabricated, fully characterized, and its performance met both the required pressure field profile and the pressure values needed for our in vivo two-photon microscopy experiments. This array is an important step in microscopically characterizing ultrasound and microbubble assisted blood-brain barrier disruption.

The current work investigated the stimulation induced spatial response of the cerebral microvascular network by reconstruction of the 3D microvascular morphology from in vivo two photon fluorescence microscopy (2PFM) volumes using an automated, model based tracking algorithm. In vivo 2PFM imaging of the vasculature in the forelimb representation of the primary somatosensory cortex of alpha-chloralose anesthetized rats was achieved via implantation of a closed cranial window, and intravascular injection of fluorescent dextran. The dilatory and constrictory responses of the cerebral microvascular network to functional stimulation were heterogeneous and depended on resting vascular radius and response latency. Capillaries experienced large relative dilations and constrictions, but the larger vessel absolute volume changes dominated the overall network cerebral blood volume change.

Gold nanoparticles, a class of plasmonic nanoparticle, have increasingly been explored as an imaging and therapeutic agent to treat cancer due to their characteristic surface plasmon resonance phenomenon and penchant for tumor accumulation. Photothermal therapy has been shown as a promising cancer treatment by delivering heat specifically to the tumor site via gold nanoparticles. In this study, we demonstrate that gold nanorod (GNR)-mediated photothermal therapy can be more effective through the understanding of cell death mechanisms. By targeting GNRs to various cellular localizations, we explored the association of GNR localization with cell death pathway response to photothermal therapy. Furthermore, we compared the 2D monolayer experiments with 3D in vitro tumor models, multicellular tumor spheroids (MCTS), to mimic the structure of in vivo tumors. With MCTS, we evaluated the cell death response with GNRs distributed only on the periphery, as seen in typical in vivo studies, and distributed evenly throughout the tumor. We demonstrated that GNR localization influences the cell death response to photothermal therapy by showing the power threshold necessary to induce significant apoptotic and necrotic increases was lower for internalized GNRs than membrane-bound GNRs. Furthermore, apoptosis was found to increase with increasing laser power until the necrotic threshold and decreased above it, as necrosis became the dominant cell death pathway response. A similar trend was revealed with the 3D MCTS; however, the overall cell death percentages were lower, most likely due to the upregulated cell repair response and varied GNR distributions due to the presence of cell-cell and cell-matrix interactions. Furthermore, the uniformly distributed GNRs induced more apoptosis and necrosis than GNRs located in the MCTS periphery. In conclusion, we quantitatively analyzed the cell death pathway response to GNR-mediated photothermal therapy to establish that it has some dependence on GNR localization and distribution to gain a more thorough understanding of this response for photothermal therapy optimization. / text

The ultimate goal of antitumor vaccines is to develop memory CD8+ cytotoxic T lymphocytes (CTLs), which are critical mediators of antitumor immunity. Previous work in our lab demonstrated that the ovalbumin (OVA)-specific CD4+ T cell-based (OVA-TEXO) vaccine generated using OVA-pulsed dendritic cell (DCOVA)-released exosomes (EXOOVA) stimulates CTL responses via interleukin (IL)-2 and costimulatory CD80 signaling. To assess the potential involvement of other costimulatory pathways and to define the key constituent of costimulation for memory CTL development, we first immunized wild-type (WT) C57BL/6 and gene-knockout mice with WT CD4+ OVA-TEXO cells or OVA-TEXO cells with various molecular deficiencies. We then assessed OVA-specific primary and recall CTL responses using PE-H-2Kb/OVA257–264 tetramer and FITC-anti-CD8 antibody staining by flow cytometry. We also examined antitumor immunity against the OVA-expressing B16 melanoma cell line BL6-10OVA. We demonstrate that CD4+ OVA-TEXO cells form immunological synapses with cognate CD8+ T cells in vitro. By assessment of the pattern of ex vivo interactions between OTI CD8+ T cells and OVA-TEXO or (Kb-/-)TEXO cells lacking peptide/major histocompatibitity complex (pMHC)-I expression, we provide the first visible evidence on the critical role of exosomal pMHC-I in targeting OVA-TEXO to cognate CD8+ T cells using two-photon microscopy. By assessing primary and recall CTL responses in mice immunized with OVA-TEXO cells or with OVA-TEXO cells lacking the costimulatory molecules CD40L, 4-1BBL or OX40L, we demonstrated that these costimulatory signals are dispensable for CTL priming by OVA-TEXO cells. Interestingly, CD40L, but not 4-1BBL or OX40L, plays a crucial role in the development of functional memory CTLs against BL6-10OVA tumors. Overall, this work suggests that a novel CD4+ T cell-based vaccine that is capable of stimulating long-term functional CTL memory via CD40L signaling may represent a novel, efficient approach to antitumor vaccination. Breast cancer is the most common cancer among women in the western world. Approximately 20-30% of invasive breast carcinomas are proto-oncogene human epidermal growth factor receptor (HER)-2 positive and associated with increased metastatic potential and poor prognosis. The survival benefit of anti-HER2 driven therapies demonstrated in clinical trials indicates that HER2 is one of the most promising molecules for targeted therapy to date. Above results prompt us to assess whether CD4+ T-cell-based vaccine can stimulate efficient HER2-specific CD8+ CTL responses and antitumor immunity in transgenic mice with HER2-specific self-immune tolerance. We prepared HER2-specific HER2-TEXO using ConA-stimulated CD4+ T cells with uptake of exosomes released from HER2-expressing AdVHER2-transfected DCs. We found that HER2-TEXO vaccine is capable of inducing HER2-specific CTL responses and protective immunity against transgene HLA-A2/HER2-expressing B16 melanoma BL6-10HLA-A2/HER2 in 2/8 double transgenic HLA-A2/HER2 mice with HER2-specific self-immune tolerance. The remaining 6/8 mice had significantly prolonged survival. Therefore, the novel T cell-based HER2-TEXO vaccine may provide a new therapeutic alternative for women with HER2+ breast cancer. In contrast to CD4+CD25+ regulatory T cells (Tregs), mechanisms of CD8+CD25+ Treg-mediated immunosuppression are not well understood. In this study, we purified polyclonal CD8+CD25+ Tregs from C57BL/6 mouse splenocytes and expanded them in culture medium containing CD3/CD28 microbeads. By using these amplified CD8+CD25+ Tregs, we demonstrated that CD8+CD25+ Tregs inhibit naive CD4+ T-cell proliferation and induce naive T-cell anergy by up-regulating T-cell anergy-associated early growth response 2 (EGR2), and by decreasing T-cell proliferation and IL-2-secretion upon stimulation. They also impact the expression of perforin on effector CTLs and directly induce perforin-mediated CTL apoptosis. CD8+CD25+ Tregs, when pulsed with OVA323-339 peptide, exert an enhanced inhibition. Interestingly, CD8+CD25+ Tregs, when pulsed with myelin oligodendrocyte glycoprotein (MOG)35-55 peptide, become capable of inhibiting MOG35-55-induced experimental autoimmune encephalomyelitis (EAE). Two-photon microscopic observations suggest that OVA323-339-pulsed (armed) CD8+CD25+ Tregs reduce the interactions between DCs and cognate CD4+ T cells ex vivo by increasing velocities of T cells in mouse lymph nodes. Therefore, redirecting antigen-specificity to nonspecific CD8+CD25+ Tregs can be achieved for enhanced immunosuppression through their arming with the antigen-specific pMHC-II complexes. This approach may have great impact on improvement of endogenous polyclonal Treg-mediated immunotherapy for autoimmune diseases. Taken together, our studies demonstrate that nonspecific polyclonal CD4+ T cells and CD8+CD25+ Tregs, when armed with HER2 and MOG antigen-specific pMHC-I and -II complexes, become capable of stimulating enhanced HER2-specific CTL responses and antitumor immunity in double transgenic HLA-A2/HER2 mice and inducing enhanced MOG-specific immunosuppression in MOG-induced EAE mice, respectively. Therefore, redirecting antigen specificity to nonspecific CD4+ T and CD8+CD25+ Tregs by pMHC complex arming may have great impact in development of novel T cell-based vaccines for treatment of cancer and autoimmune diseases.