Development of brain function is instructed by both genetically-determined processes (nature) and environmental stimuli (nurture). The relative importance of nature and nurture is a major question in developmental neurobiology. In this dissertation, I investigated the role of visual experience in the development and plasticity of the visual pathway. Each neuron that receives visual input responds to a specific area of the visual field- their receptive field (RF). Developmental refinement reduces RF size and underlies visual acuity, which is important for survival. By rearing Syrian hamsters (Mesocricetus auratus) in constant darkness (dark rearing, DR) from birth, I investigated the role of visual experience in RF refinement and plasticity. Previous work in this lab has shown that developmental refinement of RFs occurs in the absence of visual experience in the superior colliculus (SC), but that RFs unrefine and thus enlarge in adulthood during chronic DR. Using an in vivo electrophysiological approach, I show that, contrary to a widely held view, visual experience is not necessary for refinement of RFs in primary visual cortex (V1). In both SC and V1, RFs refine by postnatal day (P) 60, but enlarge by P90 with chronic DR. One week of visual experience was sufficient to prevent RF enlargement in SC and V1. How normal sensory experience prevents plasticity in mature circuits is not well understood. Using an in vitro electrophysiological approach, I demonstrated that GABAergic inhibition is reduced in DR SC, which in turn affects short-term (but not long-term) synaptic plasticity. The level of GABABR-mediated short-term synaptic depression (STD) that occurs during high-frequency afferent stimulation, such as occurs during vision, is reduced by DR. Using a computational model of RF size, I propose that, in addition to the effect of reduced inhibition, reduced STD of excitation could contribute to enlarged RFs. This work provides insight into mechanisms of development and plasticity of the nervous system. How plasticity is restricted in mature circuits is of fundamental importance in neuroscience and could instruct therapies to prevent maladaptive plasticity in disease and to enhance recovery of function in adults.

As theoretical neuroscience has grown as a field, machine learning techniques have played an increasingly important role in the development and evaluation of theories of neural computation. Today, machine learning is used in a variety of neuroscientific contexts from statistical inference to neural network training to normative modeling. This dissertation introduces machine learning techniques for use across the various domains of theoretical neuroscience, and the application of these techniques to build theories of neural circuits. First, we introduce a variety of optimization techniques for normative modeling of neural activity, which were used to evaluate theories of primary motor cortex (M1) and supplementary motor area (SMA). Specifically, neural responses during a cycling task performed by monkeys displayed distinctive dynamical geometries, which motivated hypotheses of how these geometries conferred computational properties necessary for the robust production of cyclic movements. By using normative optimization techniques to predict neural responses encoding muscle activity while ascribing to an “untangled” geometry, we found that minimal tangling was an accurate model of M1. Analyses with trajectory constrained RNNs showed that such an organization of M1 neural activity confers noise robustness, and that minimally “divergent” trajectories in SMA enable the tracking of contextual factors. In the remainder of the dissertation, we focus on the introduction and application of deep generative modeling techniques for theoretical neuroscience. Specifically, both techniques employ recent advancements in approaches to deep generative modeling -- normalizing flows -- to capture complex parametric structure in neural models. The first technique, which is designed for statistical generative models, enables look-up inference in intractable exponential family models. The efficiency of this technique is demonstrated by inferring neural firing rates in a log-gaussian poisson model of spiking responses to drift gratings in primary visual cortex. The second technique is designed for statistical inference in mechanistic models, where the inferred parameter distribution is constrained to produce emergent properties of computation. Once fit, the deep generative model confers analytic tools for quantifying the parametric structure giving rise to emergent properties. This technique was used for novel scientific insight into the nature of neuron-type variability in primary visual cortex and of distinct connectivity regimes of rapid task switching in superior colliculus.

Certains troubles visuo-moteurs observés dès le stade précoce de la maladie de Parkinson (MP) pourraient être liés à une altération du fonctionnement d’une structure sous-corticale reliée aux ganglions de la base, le colliculus supérieur (CS). L’objectif de cette thèse a été d’explorer l’état fonctionnel du CS chez le patient parkinsonien nouvellement diagnostiqué (de novo) avant et après instauration du traitement dopaminergique, afin d’évaluer son potentiel de biomarqueur. Pour cela, un paradigme expérimental d’Imagerie par Résonance Magnétique fonctionnelle (IRMf) a été développé, permettant d’imager avec succès l’activité fonctionnelle du CS et également du corps genouillé latéral (CGL) et de l’aire visuelle primaire V1 et de moduler leur activité via l’emploi de stimulation visuelle jouant sur de très faibles niveaux de contraste (<10%). Un test de psychophysique a également été développé, permettant d’estimer la réponse perceptuelle au contraste. Nous avons dans un premier temps testé notre protocole expérimental auprès de sujets sains d’âge variable afin d’évaluer le fonctionnement de ces trois régions d’intérêt (ROIs) au cours du vieillissement normal et de différencier les effets liés à l’âge de ceux potentiellement liés à la pathologie (Etude 1). Une diminution statistiquement significative de la réponse BOLD au sein du CGL et de V1 avec l’âge a été observée, ces réponses corrélant de plus parfaitement avec les réponses perceptuelles estimées en psychophysique. Les voies magnocellulaire et parvocellulaire semblent jouer un rôle dans cette perte de sensibilité au contraste de luminance liée à l’âge. Nous avons dans un second temps testé notre protocole auprès de patients parkinsoniens de novo avant et après instauration du premier traitement dopaminergique afin d’évaluer les effets de la MP et du traitement sur le fonctionnement de nos ROIs (Etude 2). Une altération précoce du traitement du contraste a été observée au sein du CS et du CGL chez les patients parkinsoniens, non normalisée par l’instauration du traitement dopaminergique. Ces travaux de thèse ont ainsi mis en évidence un déficit fonctionnel du CS et du CGL survenant précocement durant l’évolution de la MP, confirmé par nos analyses de connectivité effective. Ces résultats pourraient favoriser l’identification de déficits liés à un dysfonctionnement sensoriel de ces structures tout comme le développement de tests paraclinique et clinique impliquant ce système pour un diagnostic plus précoce de la maladie. / Some visuo-motor impairments observed in the early stages of Parkinson’s disease (PD) might be related to a dysfunction of a subcortical structure connected to the basal ganglia, the superior colliculus (SC). The aim of this PhD thesis was to explore the functional state of the SC in newly diagnosed (de novo) PD patients before and after dopaminergic treatment intake, in order to evaluate the potential value of the SC functioning as a biomarker. To do this, we developed a functional Magnetic Resonance Imaging (fMRI) experimental protocol, which successfully imaged the SC and also the lateral geniculate nucleus (LGN) and primary visual area V1 functional activity and modulate their activity by using visual stimuli with low luminance contrast levels (<10%). Additionally, we estimated the perceptual response to contrast by using a psychophysical task. We tested in a first time this experimental protocol on healthy subjects with varying age in order to evaluate the effect of normal aging on the functioning of these regions of interest (ROIs) and to distinguish the effects related to age from those potentially related to the pathology (Study 1). A significant progressive decrease of the BOLD amplitude with age was observed in the LGN and V1. These data were consistent with the response functions obtained with the psychophysical task. These results indicate a significant luminance contrast sensitivity decline with age of both the magnocellular and parvocellular pathways. In a second time, we tested our protocol on de novo PD patients before and after the introduction of the first dopaminergic treatment in order to assess the effects of PD and treatment on the ROIs functioning (Study 2). Our results highlighted an early alteration of the contrast processing for the SC and LGN in PD patients, with no normalization after dopaminergic treatment introduction. These findings indicate a functional deficit of the SC and LGN that appears early in the disease course, in line with our effective connectivity analyses. These results could favor the identification of deficits linked to sensory dysfunction of these structures as well as the development of paraclinical and clinical tests involving this system for an early diagnosis of the disease.

La neuropathie optique glaucomateuse est une pathologie du système visuel entrainant une cécité irréversible qui affectera 80 millions de personnes en 2020. Le principal facteur de risque du glaucome est l'élévation de la pression intraoculaire qui mène à la mort progressive des cellules ganglionnaires de la rétine (CGR) du nerf optique jusqu'aux voies visuelles dans le cerveau. Il a été montré que le glaucome partageait des mécanismes neuro-inflammatoires communs avec les pathologies neurodégénératives. Nous avons émis l'hypothèse que ces mécanismes contribueraient à la progression du glaucome. L'objectif de ma thèse a été d'analyser les processus neuro-inflammatoires observés de la rétine jusqu'aux colliculus supérieurs sur un modèle expérimental d'hypertension oculaire unilatérale chez le rat obtenu après la cautérisation des veines épisclérales. Par des approches de biologie cellulaire et moléculaire, nous avons montré que ce modèle animal se caractérise par 1) une atteinte neuronale des CGR de l'oeil cautérisé; 2) l'augmentation de marqueurs pro-inflammatoires dans la rétine de l'oeil cautérisé, dans la rétine de l'oeil controlatéral, dans le nerf optique et dans les colliculus supérieurs et 3) la transmission de la neuro-inflammation à l'oeil controlatéral se fait majoritairement par les fibres des CGR qui projettent dans les deux colliculus supérieurs. Toutes ces données mettent en évidence le rôle complexe joué par le colliculus supérieur chez le rat dans la propagation de la neuro-inflammation induite par l'hypertension oculaire unilatérale. / Glaucoma is a visual system disorder leading to irreversible blindness and affecting 80 millions people worldwide by 2020. The major risk factor is elevated intraocular pressure leading to progressive retinal ganglion cell (RGC) death from the optic nerve (ON) to visual pathways in the brain. Glaucoma has been reported to share neuroinflammatory mechanisms with neurodegenerative disorders. We therefore hypothesize that mechanisms in central visual pathways may contribute to the spread of glaucoma disease. The aim of the present study was to analyze the neuroinflammation processes that occur from the pathological retina to the superior colliculi (SCs) in a rat model of unilateral ocular hypertension induced by episcleral vein cauterization. By molecular and cell biology methods, we have shown that this animal model is characterized by 1) neuronal damage of CGR from the cauterized eye; 2) the increase in proinflammatory markers in the retina of the cauterized eye, in the retina of the contralateral eye, in the optic nerve and in the superior colliculus and 3) transmission of neuroinflammation in contralateral eye is done mainly by CGR fibers that project into the two superior colliculus. All these data evidence the complex role played by the SCs, in rat, in the propagation of neuroinflammatory events induced by unilateral ocular hypertension.

<p dir="ltr">The ability to selectively respond to a desirable stimulus while ignoring the surroundings is essential for efficient decision making, such as sensing food amongst rocks. This requires accurate discrimination of the input features and their associated values. Current understanding of value-modulated sensory processing is highly biased towards the visual modality, and little is known about how tactile signals evolve across cortical and subcortical regions in the same goal directed task. It is also unclear whether task engagement and action outcome modulates neural representation of the sensory and value space.</p><p dir="ltr">To answer these questions, we investigated the mouse primary somatosensory cortex (S1) and the superior colliculus (SC) while the animal was performing an active spatial discrimination task. The mice were trained to actively touch and associate a whisker-dependent stimulus with reward (GO whisker) while ignoring tactile input to the adjacent whisker (NoGo whisker). We hypothesize that sensory information in the S1 neocortex is encoded in the form of physical features such as location, which is then converted into a map of stimulus value in the SC. We found that both S1 and SC neurons accurately discriminated between adjacent whisker stimuli with the SC displaying a much stronger preference than S1 for the higher valued stimulus. This bias was unique to SC wherein spiking activities were facilitated for the rewarded stimulus and suppressed for the negative stimulus. Importantly, removing the opportunity for positive stimulus selection reduced but did not abolish positive bias in the SC, reinforcing its role in value-based sensory modulation. Moreover, the spontaneous activity in SC but not S1 predicted response latency and performance accuracy. Taken together, this study demonstrated a transformation of stimulus priority from a somatotopic map in S1 to a value map in SC.</p>

<p dir="ltr">Localizing movements in the external space is crucial for animals to navigate safely, find food, avoid predators, and interact with their surroundings. Efficient localization during body movements requires the brain to distinguish between externally generated movements and self-generated ones. This involves integrating external stimulation with a continuous estimate of one's body position, to isolate external motion by suppressing sensations arising from self-motion.</p><p dir="ltr">To explore the neural mechanisms underlying object localization during active touch, we focused on the mouse superior colliculus (SC), which harbors multiple egocentric maps of sensorimotor space. Our studies revealed that SC neurons exhibit a rapidly adapting tactile response during externally generated touch. The response is significantly attenuated during self-generated touch, thus enhancing the ability to distinguish between external and self-induced tactile stimuli. Additionally, the direction of external motion is precisely encoded in the firing rates of these tactile-responsive neurons, indicating a specialized localization mechanism within the SC.</p><p dir="ltr">In scenarios devoid of external stimuli, SC neural activity accurately reflects the kinematics of self-motion, such as whisker position and locomotion speed, capturing past, present, and future body positions. Half of the neurons that encode self-motion also respond to external tactile stimuli. This dual functionality suggests that these neurons not only track self-motion but also engage in the processing of external tactile information. The magnitude of the external tactile response in these neurons is modulated by the state of self-motion upon touch. These results suggest that SC neurons integrate internal estimates of body movements with external tactile inputs to compute the egocentric distance of objects.</p>

Das Ziel dieser Arbeit war es, die Modulation der GABAergen synaptischen Transmission in den visuellen Schichten des Colliculus superior der Maus durch das Neurotrophin BDNF zu charakterisieren. Hierzu wurden bdnf+/+ und -/- Mäuse kurz vor und nach der Augenöffnung in einer die Morphologie erhaltenden Schnittpräparation elektrophysiologisch und molekularbiologisch untersucht. Das Fehlen von BDNF veränderte das Präparat hinsichtlich Neurondichte und -größe nicht. Ebenso blieben der Membranwiderstand und die Ganzzellkapazität unbeeinflusst von der chronischen Abwesenheit von BDNF. Im Gegensatz dazu zeigten sich deutliche funktionelle Defizite im Entladungsverhalten und in der GABAergen Hemmung. Durch Registrierung von Aktionspotentialen wurde demonstriert, dass BDNF für die Aufrechterhaltung der Netzwerkaktivität erforderlich ist. Durch Applikation eines GABAA-Rezeptor-Inhibitors konnte die Suppression der GABAergen Hemmung durch BDNF als zugrunde liegender Mechanismus aufgedeckt werden. Daraufhin durchgeführte Ganzzellableitungen bestätigten dies und legten einen postsynaptischen, TrkB-vermittelten Mechanismus der BDNF-Wirkung nahe. Es war möglich, den Einfluss der chronischen Abwesenheit von BDNF durch akute lokale Superfusion von BDNF vollständig aufzuheben. Die ausschließlich postsynaptische Blockade der PKC reichte aus, dies zu verhindern. Hierdurch wird unterstrichen, dass in diesem Präparat der Angriffspunkt von BDNF an der GABAergen Synapse auf der postsynaptischen Seite liegt. Um den genauen Wirkungsmechanismus von BDNF an der GABAergen Synapse zu beleuchten, wurde die mRNA-Expression der GABAA-Rezeptor-Untereinheiten alpha 1-3 untersucht. Diese ist in Anwesenheit von BDNF höher. Demzufolge sollte eine reduzierte Expression dieser Untereinheiten in bdnf-/- Tieren zu einer verringerten Rezeptoranzahl und somit zur Sättigung postsynaptischer Rezeptoren führen. Durch die Analyse von Amplitude und Kinetik GABAerger IPSC und die Applikation von Zolpidem wurde dies bestätigt. Demnach führt die Abwesenheit von BDNF zur Aufregulation der GABAergen Inhibition, obwohl die Rezeptorzahl in der Postsynapse wahrscheinlich niedriger ist. Als zentraler Mechanismus der akuten BDNF-Wirkung kommt deshalb am ehesten eine PKC-vermittelte Phosphorylierung und nachfolgende Veränderung des Desensitisierungsverhaltens in Betracht. Außerdem muss an eine Reduktion der Öffnungswahrscheinlichkeit oder der Leitfähigkeit des Rezeptors gedacht werden. Unmittelbar vor der Augenöffnung hatte die BDNF-Defizienz keinerlei Einfluss auf die GABAerge Hemmung. Es ist also davon auszugehen, dass BDNF im CS erst nach der Augenöffnung eine wesentliche Rolle in der Modulation der GABAergen Synapsen und damit in der Kontrolle der Netzwerkaktivität spielt. Dies steht in Einklang mit der Vorstellung, dass sowohl Translation und Freisetzung als auch Transport von BDNF durch neuronale Aktivität reguliert werden. Der Ablauf der neurotrophinergen Regulation im Colliculus superior stellt sich wie folgt dar: Durch Aktivation colliculärer Afferenzen wird BDNF vermehrt freigesetzt. BDNF reduziert nun zunächst über die Modulation von Rezeptoreigenschaften die GABAerge Hemmung und disinhibiert die Netzwerkaktivität. Längerfristig kommt es über eine vermehrte Expression von GABAA-Rezeptor-Untereinheiten zum Anstieg der Rezeptorzahl, damit zur Wiederherstellung der GABAergen Hemmung und letztlich zu einer Reduktion der Netzwerkaktivität. BDNF ist also in der kritischen Zeitperiode der Augenöffnung, wenn das Mustersehen einsetzt, ein wichtiger Faktor in der Regulation neuronaler Aktivität. / The aim of the study was to characterise the influence of the neurotrophin BDNF on the GABAergic synaptic transmission in the visual layers of the mouse superior colliculus. Acute slices prepared from bdnf+/+ and -/- mice shortly before and after eye opening were employed in the experiments. The absence of BDNF altered neither the density or size of neurons nor their membrane resistance or whole cell capacity. However, registration of action potentials revealed a decreased firing rate in the absence of BDNF. Stronger disinhibition induced by application of a GABAA receptor blocker suggested an enhanced GABAergic inhibition as an underlying mechanism. This assumption was confirmed by performing whole cell experiments. Further analysis indicated a postsynaptic enhancement of GABAergic synaptic transmission in the absence of BDNF. In bdnf+/+ slices, blockade of BDNF signalling through the TrkB receptor strengthened GABAergic synaptic transmission. Contrariwise, superfusion of exogenous BDNF in bdnf-/- suppressed GABAergic synaptic transmission. An exclusively postsynaptic block of the PKC abolished the effect of BDNF application. Therefore, a BDNF induced, TrkB mediated, PKC dependent suppression of the GABAergic synaptic transmission at the postsynaptic site can be assumed. To further elucidate the mechanism of BDNF action the expression of the GABAA receptor subunits alpha 1-3 mRNA was studied. In the presence of BDNF an elevated expression was observed. A lower expression of these subunits in bdnf-/- slices could result in a reduced number of GABAA receptors. During synaptic transmission they may become saturated. Two observations support this idea: In bdnf-/- slices application of Zolpidem does not induce an increase of GABAergic IPSC amplitude as present in bdnf+/+ slices. Amplitude and decay kinetics of GABAergic IPSCs correlate in bdnf-/- but not +/+ slices. Therefore, absence of BDNF may strengthen GABAergic synaptic transmission although mediated by a reduced number of postsynaptic receptors. A BDNF induced PKC dependent receptor phosphorylation followed by a change in receptor desensitisation is most likely the underlying mechanism. Nevertheless, a reduction of the receptor opening probability or a reduced receptor conductance have to be considered as well. Shortly before eye opening BDNF deficiency had no impact on GABAergic inhibition. In accordance with the general idea that expression and release of BDNF is activity dependent, this finding suggests that BDNF controls network activity by modulating GABAergic synaptic transmission only after eye opening. Taken together, the following sequence of BDNF action in the superior colliculus during eye opening can be proposed: Activation of collicular inputs triggers an increased BDNF release. BDNF suppresses GABAergic synaptic transmission leading to a disinhibition of network activity. Later, the increased expression of GABAA receptor subunits by prolonged BDNF release results in an increased GABAA receptor number, the recovery of the GABAergic inhibition and finally a gently tuned network activity. In summary, BDNF constitutes a major regulator of neuronal activity in the critical period of eye opening marking the transition to pattern vision.