acase@tulane.edu / 1 / Jeremiah Hartner

This thesis explores the role of path integration on the firing of hippocampal place cells and medial entorhinal grid cells. Grid cells fire at equidistant locations in an environment, indicating that they keep track of the distance and direction an animal has moved in an environment. One class of model of path integration uses a continuous attractor network to update position information. The first part of this thesis showed that such a network can generate a "look-ahead" of neural activity that sweeps through the positions just visited and about to be visited, on the short time scale that is observed<italic>in vivo</italic>. Adding intrinsic currents to the neurons in the network model allowed this look-ahead to recur every theta cycle, and generate grid fields of a size comparable to data. Grid cells are a major input the hippocampus, and are hypothesized to be the source of the place specificity of place cells. When an animal explores an open environment, place cells are active in a particular location regardless of the direction in which the animal travels through it. While performing a specific task, such as visiting specific locations in the environment in sequence, however, most place cells are active only in one direction. The second part of this thesis studied the development of this directionality. It was determined that upon the initial appearance of place fields in a novel environment, place cells fired in all directions, supporting the hypothesis that the path integration is the primary determinant of place specificity. The directionality of place fields developed gradually, possibly as a result of learning. Ideas about how this directionality could develop are explored.

Spatial memory and navigation relies on estimation of location. This can be achieved through several strategies, including the use of landmarks and by path integration. The latter involves inferring location from direction and distance moved relative to a known start point. The neural mechanisms of path integration are not well understood and implementation of experiments that dissociate path integration from alternative strategies is challenging. The roles of specific cell types are also unknown. Although grid cells in layer 2 of the medial entorhinal cortex (MEC) are theorised to be involved given their periodic and repeating firing fields that form a grid-like map that tiles the environment. Two excitatory cell populations have been identified in layer 2 of the MEC. Clusters of pyramidal cells that project to the CA1 are surrounded by dentate gyrus (DG) projecting stellate cells. Both populations have been shown to exhibit grid-like activity. The extent to which these cell types contribute to path integration or other strategies for solving spatial tasks is unknown. To investigate these issues, I developed a spatial memory task for mice, which uses virtual reality to generate sensitive measures of an animal’s ability to path integrate. In this task mice are trained to locate a reward zone marked with a visual cue within a virtual linear track. Use of path integration strategies can be tested in trials in which the reward zone is unmarked. In this task mice can locate the reward zone using either a local beaconing cue or path integration strategies. To assess whether self-motion derived motor information or visual feedback is used for path integration, I manipulated the translation between physical and virtual movement, putting optic and motor feedback in conflict. These manipulations suggest that mice use motor information to locate the reward zone on path integration trials. To test roles of stellate cells in the task I injected adeno-associated virus expressing the light chain of tetanus toxin, conditionally on the presence of Cre, into the MEC of mice expressing Cre specifically in stellate cells. This abolishes synaptic output from stellate cells therefore preventing them from influencing downstream neurons. I find mice with dorsal expression of the tetanus toxin virus in layer 2 stellate cells are unable to locate the reward zone using a local beaconing cue or path integration strategies. In contrast, mice with expression of green fluorescent protein (GFP) were able to locate the reward zone using both strategies. Locating the reward zone using path integration strategies first requires animal’s to learn the reward zone location, as denoted in trials with a beacon cue. To distinguish the role of stellate cells in learning versus execution of the tasks, I temporally modified the activity of stellate cells after mice had learnt to locate the reward zone using both strategies. Temporal control was achieved by use of cre-dependent adeno-associated viruses expressing mutant human muscarinic 4 receptor (hM4). When activated by clozapine - N - oxide (CNO), this receptor opens G-protein inwardly rectifying potassium (GIRK) channels and attenuates neuronal firing. Using this method, the activity of stellate cells can be temporally controlled during task execution and potentially distinguish their involvement in learning and execution of spatial memory tasks. No effect on behavioural performance was seen under these conditions. This may indicate stellate cells are required for learning but not execution of spatial memory tasks that require the use of local beaconing cues or path integration.

Place cells in the hippocampus fire in specific locations within an environment. The aim of this thesis is to investigate the different inputs to the hippocampus and what they contribute to place cell activity and performance of hippocampus-dependent tasks. Place cell activity can also be modulated by relevant features of a task such as a future destination or trajectory. Initial experiments investigated the origin and function of this trajectory-dependent activity and later experiments targeted the medial entorhinal cortex inputs to the hippocampal formation and investigated what they contributed to place cell activity and behaviour. The purpose of the first study was to determine whether trajectory dependent activity occurs in CA3 in a hippocampus-dependent serial-reversal task on the double-Y-maze and to compare it with that seen in CA1. Place cells in both CA3 and CA1 were recorded in rats trained on a serial-reversal task on a double-Y-maze. Rats were trained to run from a start box through two Y-junctions to one of four goal locations. After 10 trials the reward was moved to a new location, until all the boxes had been rewarded. Previous research has found that 44% of CA1 place cells with fields in the start areas of the maze show trajectory-dependent activity in rats trained on the task. This study found that a similar proportion of CA3 place cells also show trajectory-dependent activity in rats trained on this task and that this activity develops at the same time point as the task is learned. This result suggests that trajectory-dependent activity may be generated earlier in the circuit than CA1. Secondly, the contribution of the nucleus reuniens (N.Re) to spatial tasks was investigated. Previously, trajectory-dependent activity has been found to reach the hippocampus via N.Re, however this was shown in a hippocampus-independent task. To investigate the possible role that this input may play in behaviour, N.Re was lesioned and animals were tested on acquisition and performance of the double-Y-maze serial-reversal task described previously. Surprisingly, lesions had no effects on either learning or performance. Taken together with previous data from other studies, this suggests that trajectory dependent activity is not one unique phenomenon but is rather multiple similar phenomena which may originate in different brain regions and fulfil different roles in navigation depending on the demands of the task. In addition, animals were tested on tasks involving allocentric or egocentric navigation. Results suggest that N.Re may have a role in the selection or performance of allocentric navigation but not egocentric navigation. Thirdly, the role of inputs from the medial entorhinal cortex (MEC) to place cells was investigated. Consistent with previous research, MEC lesions resulted in larger, less precise place fields in CA1 place cells. By performing cue-rotation experiments using either distal or proximal cues it was observed that place fields in the MEC lesion animals were not anchored to distal cues but were either stable or anchored to other aspects of the environment. However, place cells in the MEC lesion group still followed proximal cues suggesting that the deficit is restricted to distal landmarks. This suggests that the MEC may process distal landmark information allowing the use of distal landmarks for orientation and self-location within an environment. This thesis contributes a better understanding of the role and origins of trajectory dependent activity as well as a novel finding that the MEC contributes information about distal landmarks to the hippocampus.

Previous research investigating the neural circuitry underlying memory consolidation has primarily focused on single “nodes” in the circuit rather than the neural connections between brain regions, despite the likely importance of these connections in mediating different aspects or forms of memory. This focus has, in part, been due to technical limitations; however the advent of optogenetics has altered our capabilities in this regard, enabling optical control over neural pathways with temporal and spatial precision. The current set of experiments took advantage of optogenetics to control activity in specific pathways connecting brain regions in rats immediately after different kinds of learning. Chapter 2 first established the use of optogenetics to manipulate activity in the basolateral amygdala (BLA), which has been shown to modulate memory consolidation for a variety of types of learning likely through its connections to various downstream regions. Using a one-trial inhibitory avoidance task, a simple and robust fear learning paradigm, we found that both post-training stimulation and inhibition of BLA activity could enhance or impair later retention of the task, respectively. Enhancement was specific to stimulation using trains of 40, but not 20, Hz light pulses. Chapters 3 and 4 examined the projections from the BLA to the ventral hippocampus (VH) and medial entorhinal cortex (mEC) as the BLA’s ability to influence the consolidation for many types of memory is believed to be mediated through discrete projections to distinct brain regions. Indeed, the BLA innervates both structures, and prior studies suggest that the mEC and VH have distinct roles in memory processing related to contextual and nociceptive (footshock) learning, such as those involved in contextual fear conditioning (CFC). Optogenetic stimulation or inhibition of the BLA-VH or BLA-mEC pathway after training on a modified CFC task, in which the nociceptive or emotional stimulus (the footshock) and the context are separated, enabled experimental manipulations to selectively affect the consolidation for learning about one component and not the other. Optogenetic stimulation/inhibition was given to each candidate pathway immediately after the relevant training to determine its role in influencing consolidation for that component of the CFC learning. Chapter 3 results showed that stimulation of the BLA-VH pathway following footshock, but not context, training enhanced retention, an effect that was specific to trains of 40 Hz stimulation. Post-footshock photoinhibition of the same pathway impaired retention for the task. Similar investigations of the BLA-mEC pathway in Chapter 4 produced complementary findings. Post-context, but not footshock, stimulation of the pathway enhanced retention. In this particular case, only trains of 8 Hz stimulation were effective at enhancing retention. These results are the first, to our knowledge, to find that BLA inputs to different structures selectively modulate consolidation for different aspects of learning, thus enhancing our understanding of the neural connections underlying the consolidation of contextual fear conditioning and providing a critical foundation for future research.

Schicht 2 des mediale entorhinale Kortex (MEK) beinhaltet die größte Anzahl von Gitterzellen, welche durch ein hexagonales Aktivitätsmuster während räumlicher Exploration gekennzeichnet sind. In dieser Arbeit wurde gezeigt, dass spezielle Pyramidenzellen, die das Protein Calbindin exprimieren, in einem hexagonalen Gitter im Gehirn der Ratte angeordnet sind und cholinerg innerviert werden. Es ist bekannt, dass die cholinerge Innervation wichtig für die Aktivität von Gitterzellen ist. Weiterhin ergaben neuronale Ableitungen und Methoden zur Identifikaktion einzelner Neurone in frei verhaltenden Ratten, dass Calbindin-positive Pyramidenzellen (Calbindin+) eine große Anzahl von Gitterzellen beinhalten. Reelin-positive Sternzellen (Reelin+) im MEK, zeigten keine anatomische Periodizität und ihre Aktivität orientierte sich an den Begrenzungen der Umgebung. Eine weitere Studie untersucht die Architektur des MEK in verschiedenen Säugetieren, die von der Etrusker Spitzmaus, bis hin zum Menschen ~100 Millionen Jahre evolutionäre Vielfalt und ~20,000 fache Variation der Gehirngröße umfassen. Alle Arten zeigten jeweils eine periodische Anhäufung der Calbindin+ Zellen, was deren evolutive Bedeutung unterstreicht. Eine Studie zur Ontogenese der Calbindin Anhäufungen ergab, dass die periodische Struktur der Calbindin+ Zellen, sowie die verstreute Anordnung der Reelin+ Sternzellen schon zum Zeitpunkt der Geburt erkennbar war. Weitere Ergebnisse zeigen, dass Calbindin+ Zellen strukturell später ausreifen als Reelin+ Sternzellen - passend zu der Erkenntnis, dass Gitterzellen funktionell später reifen als Grenzzellen. Eine Untersuchung des Parasubiculums ergab, dass Verbindungen zum MEK präferiert in die Calbindin Anhäufungen in Schicht 2 projizieren. Zusammenfassend beschreibt diese Doktorarbeit eine Dichotomie von Struktur und Funktion in Schicht 2 des MEK, welche fundamental für das Verständnis von Gedächtnisbildung und deren zugrundeliegenden Mikroschaltkreisen ist. / The medial entorhinal cortex (MEC) is an important hub in the memory circuit in the brain. This thesis comprises of a group of studies which explores the architecture and microcircuits of the MEC. Layer 2 of MEC is home to grid cells, neurons which exhibit a hexagonal firing pattern during exploration of an open environment. The first study found that a group of pyramidal cells in layer 2 of the MEC, expressing the protein calbindin, were clustered in the rat brain. These patches were physically arranged in a hexagonal grid in the MEC and received preferential cholinergic-inputs which are known to be important for grid-cell activity. A combination of identified single-cell and extracellular recordings in freely behaving rats revealed that grid cells were mostly calbindin-positive pyramidal cells. Reelin-positive stellate cells in MEC were scattered throughout layer 2 and contributed mainly to the border cell population– neurons which fire at the borders of an environment. The next study explored the architecture of the MEC across evolution. Five mammalian species, spanning ~100 million years of evolutionary diversity and ~20,000 fold variation in brain size exhibited a conserved periodic layout of calbindin-patches in the MEC, underscoring their importance. An investigation of the ontogeny of the MEC in rats revealed that the periodic structure of the calbindin-patches and scattered layout of reelin-positive stellate cells was present around birth. Further, calbindin-positive pyramidal cells matured later in comparison to reelin-positive stellate cells mirroring the difference in functional maturation profiles of grid and border cells respectively. Inputs from the parasubiculum, selectively targeted calbindin-patches in the MEC indicating its role in shaping grid-cell function. In summary, the thesis uncovered a structure-function dichotomy of neurons in layer 2 of the MEC which is a fundamental aspect of understanding the microcircuits involved in memory formation.

Es ist noch weitgehend ungeklärt, mittels welcher Mechanismen die elektrische Aktivität von Nervenzellpopulationen des Gehirns Verhalten ermöglicht. Die Orientierung im Raum ist eine Fähigkeit des Gehirns, für die im Säugetier der mediale entorhinale Teil der Großhirnrinde als entscheidende Struktur identifiziert wurde. Hier wurden Nervenzellen gefunden, die die Umgebung des Individuums in einer gitterartigen Anordnung repräsentieren. Die neuronalen Schaltkreise, welche diese geordnete Nervenzellaktivität im medialen entorhinalen Kortex (MEK) ermöglichen, sind noch wenig verstanden. Die vorliegende Dissertation hat eine Klärung der zellulären Architektur und der neuronalen Schaltkreise in der zweiten Schicht des MEK der Ratte zum Ziel. Zunächst werden die Beiträge zur Entdeckung der hexagonal angeordneten zellulären Anhäufungen in Schicht 2 des MEK sowie zur Beschreibung der Dichotomie der Haupt-Nervenzelltypen dargestellt. Im zweiten Teil wird erstmalig eine konnektomische Analyse des MEK beschrieben. Die detaillierte Untersuchung der Architektur einzelner exzitatorischer Axone ergab das überraschende Ergebnis der präzisen Sortierung von Synapsen entlang axonaler Pfade. Die neuronalen Schaltkreise, in denen diese Neurone eingebettet sind, zeigten eine starke zeitliche Bevorzugung der hemmenden Neurone. Die hier erhobenen Daten tragen zu einem detaillierteren Verständnis der neuronalen Schaltkreise im MEK bei. Sie enthalten die erste Beschreibung überraschend präziser axonaler synaptischer Ordnung im zerebralen Kortex der Säugetiere. Diese Schaltkreisarchitektur lässt einen Effekt auf die Weiterleitung synchroner elektrischer Populationsaktivität im MEK vermuten. In zukünftigen Studien muss insbesondere geklärt werden, ob es sich bei den hier berichteten Ergebnissen um eine Besonderheit des MEK oder ein generelles Verschaltungsprinzip der Hirnrinde des Säugetiers handelt. / The mechanisms by which the electrical activity of ensembles of neurons in the brain give rise to an individual’s behavior are still largely unknown. Navigation in space is one important capacity of the brain, for which the medial entorhinal cortex (MEC) is a pivotal structure in mammals. At the cellular level, neurons that represent the surrounding space in a grid-like fashion have been identified in MEC. These so-called grid cells are located predominantly in layer 2 (L2) of MEC. The detailed neuronal circuits underlying this unique activity pattern are still poorly understood. This thesis comprises studies contributing to a mechanistic description of the synaptic architecture in rat MEC L2. First, this thesis describes the discovery of hexagonally arranged cell clusters and anatomical data on the dichotomy of the two principle cell types in L2 of the MEC. Then, the first connectomic study of the MEC is reported. An analysis of the axonal architecture of excitatory neurons revealed synaptic positional sorting along axons, integrated into precise microcircuits. These microcircuits were found to involve interneurons with a surprising degree of axonal specialization for effective and fast inhibition. Together, these results contribute to a detailed understanding of the circuitry in MEC. They provide the first description of highly precise synaptic arrangements along axons in the cerebral cortex of mammals. The functional implications of these anatomical features were explored using numerical simulations, suggesting effects on the propagation of synchronous activity in L2 of the MEC. These findings motivate future investigations to clarify the contribution of precise synaptic architecture to computations underlying spatial navigation. Further studies are required to understand whether the reported synaptic specializations are specific for the MEC or represent a general wiring principle in the mammalian cortex.

Lebewesen sind jeden Tag Sequenzen von Ereignissen ausgesetzt, die sie sich merken wollen. Es ist jedoch ein allgemeines Problem, dass sich die Zeitskalen des Verhaltens und der Induzierung von neuronalem Lernen um mehrere Größenordnungen unterscheiden. Eine mögliche Lösung könnte "Phasenpräzession" sein - das graduelle Verschieben von Aktionspotential-Phasen relativ zur Theta-Oszillation im lokalen Feldpotential. Phasenpräzession ermöglicht es, Verhaltens-Sequenzen zeitlich zu komprimieren, herunter bis auf die Zeitskala von synaptischer Plastizität. In dieser Arbeit untersuche ich das Phasenpräzessions-Phänomen im medialen entorhinalen Kortex der Ratte. Ich entdecke, dass entorhinale Gitterzellen auf der für das Verhalten relevanten Einzellaufebene Phasenpräzession zeigen und dass die Phasenpräzession in Einzelläufen stärker ist als in zusammengefassten Daten vieler Läufe. Die Analyse von Einzelläufen zeigt zudem, dass Phasenpräzession (i) in Zellen aus allen Schichten des entorhinalen Kortex existiert und (ii) von den komplexen Bewegungsmustern der Ratten in zweidimensionalen Umgebungen abhängt. Zum Abschluss zeige ich, dass Phasenpräzession zelltyp-spezifisch ist: Sternzellen in Schicht II des medialen entorhinalen Kortex weisen klare Phasenpräzession auf, wohingegen Pyramidenzellen in der selben Schicht dies nicht tun. Diese Ergebnisse haben weitreichende Implikationen sowohl für das Lokalisieren des Ursprungs als auch für die m"oglichen Mechanismen von Phasenpräzession. / Every day, animals are exposed to sequences of events that are worth recalling. It is a common problem, however, that the time scale of behavior and the time scale for the induction of neuronal learning differ by multiple orders of magnitude. One possible solution could be a phenomenon called "phase precession" - the gradual shift of spike phases with respect to the theta oscillation in the local field potential. Phase precession allows for the temporal compression of behavioral sequences of events to the time scale of synaptic plasticity. In this thesis, I investigate the phase-precession phenomenon in the medial entorhinal cortex of the rat. I find that entorhinal grid cells show phase precession at the behaviorally relevant single-trial level and that phase precession is stronger in single trials than in pooled-trial data. Single-trial analysis further revealed that phase precession (i) exists in cells across all layers of medial entorhinal cortex and (ii) is altered by the complex movement patterns of rats in two-dimensional environments. Finally, I show that phase precession is cell-type specific: stellate cells in layer II of the medial entorhinal cortex exhibit clear phase precession whereas pyramidal cells in the same layer do not. These results have broad implications for pinpointing the origin and possible mechanisms of phase precession.

Komplexe Berechnungen im Gehirn werden durch das Zusammenspiel von exzitatorischen und hemmenden Neuronen in lokalen Netzwerken ermöglicht. In kortikalen Netzwerken, wird davon ausgegangen, dass hemmende Neurone, besonders Parvalbumin positive Korbzellen, ein „blanket of inhibition” generieren. Dieser Sichtpunkt wurde vor kurzem durch Befunde strukturierter Inhibition infrage gestellt, jedoch ist die Organisation solcher Konnektivität noch unklar. In dieser Dissertation, präsentiere ich die Ergebnisse unserer Studie Parvabumin positiver Korbzellen, in Schichten II / III des entorhinalen Kortexes und Präsubiculums der Ratte. Im entorhinalen Kortex haben wir dorsale und ventrale Korbzellen beschrieben und festgestellt, dass diese morphologisch und physiologisch ähnlich, jedoch in ihrer Konnektivität zu Prinzipalzellen dorsal stärker als ventral verbunden sind. Dieser Unterschied korreliert mit Veränderungen der Gitterzellenphysiologie. Ähnlich zeige ich im Präsubiculum, dass inhibitorische Konnektivität eine essenzielle Rolle im lokalen Netzwerk spielt. Hemmung im Präsubiculum ist deutlich spärlicher ist als im entorhinalen Kortex, was ein unterschiedliches Prinzip der Netzwerkorganisation suggeriert. Um diesen Unterschied zu studieren, haben wir Morphologie und Netzwerkeigenschaften Präsubiculärer Korbzellen analysiert. Prinzipalzellen werden über ein vorherrschendes reziprokes Motif gehemmt die durch die polarisierte Struktur der Korbzellaxone ermöglicht wird. Unsere Netzwerksimulationen zeigen, dass eine polarisierte Inhibition Kopfrichtungs-Tuning verbessert. Insgesamt zeigen diese Ergebnisse, dass inhibitorische Konnektivität, funktioneller Anforderungen der lokalen Netzwerke zur Folge, unterschiedlich strukturiert sein kann. Letztlich stelle ich die Hypothese auf, dass für lokale inhibitorische Konnektivität eine Abweichung von „blanket of inhibition― zur „maßgeschneiderten― Inhibition zur Lösung spezifischer computationeller Probleme vorteilhaft sein kann. / Local microcircuits in the brain mediate complex computations through the interplay of excitatory and inhibitory neurons. It is generally assumed that fast-spiking parvalbumin basket cells, mediate a non-selective -blanket of inhibition-. This view has been recently challenged by reports structured inhibitory connectivity, but it’s precise organization and relevance remain unresolved. In this thesis, I present the results of our studies examining the properties of fast-spiking parvalbumin basket cells in the superficial medial entorhinal cortex and presubiculum of the rat. Characterizing these interneurons in the dorsal and ventral medial entorhinal cortex, we found basket cells of the two subregions are more likely to be connected to principal cells in the dorsal compared to the ventral region. This difference is correlated with changes in grid physiology. Our findings further indicated that inhibitory connectivity is essential for local computation in the presubiculum. Interestingly though, we found that in this region, local inhibition is lower than in the medial entorhinal cortex, suggesting a different microcircuit organizational principle. To study this difference, we analyzed the properties of fast-spiking basket cells in the presubiculum and found a characteristic spatially organized connectivity principle, facilitated by the polarized axons of the presubicular fast-spiking basket cells. Our network simulations showed that such polarized inhibition can improve head direction tuning of principal cells. Overall, our results show that inhibitory connectivity is differently organized in the medial entorhinal cortex and the presubiculum, likely due to functional requirements of the local microcircuit. As a conclusion to the studies presented in this thesis, I hypothesize that a deviation from the blanket of inhibition, towards a region-specific, tailored inhibition can provide solutions to distinct computational problems.

Komplexe kognitive Funktionen wie Gedächtnisbildung, Navigation und Entscheidungsprozesse hängen von der Kommunikation zwischen Hippocampus und Neokortex ab. An der Schnittstelle dieser beiden Gehirnregionen liegt der entorhinale Kortex - ein Areal, das Neurone mit bemerkenswerten räumlichen Repräsentationen enthält: Gitterzellen. Gitterzellen sind Neurone, die abhängig von der Position eines Tieres in seiner Umgebung feuern und deren Feuerfelder ein dreieckiges Muster bilden. Man vermutet, dass Gitterzellen Navigation und räumliches Gedächtnis unterstützen, aber die Mechanismen, die diese Muster erzeugen, sind noch immer unbekannt. In dieser Dissertation untersuche ich mathematische Modelle neuronaler Schaltkreise, um die Entstehung, Weitervererbung und Verstärkung von Gitterzellaktivität zu erklären. Zuerst konzentriere ich mich auf die Entstehung von Gittermustern. Ich folge der Idee, dass periodische Repräsentationen des Raumes durch Konkurrenz zwischen dauerhaft aktiven, räumlichen Inputs und der Tendenz eines Neurons, durchgängiges Feuern zu vermeiden, entstehen könnten. Aufbauend auf vorangegangenen theoretischen Arbeiten stelle ich ein Einzelzell-Modell vor, das gitterartige Aktivität allein durch räumlich-irreguläre Inputs, Feuerratenadaptation und Hebbsche synaptische Plastizität erzeugt. Im zweiten Teil der Dissertation untersuche ich den Einfluss von Netzwerkdynamik auf das Gitter-Tuning. Ich zeige, dass Gittermuster zwischen neuronalen Populationen weitervererbt werden können und dass sowohl vorwärts gerichtete als auch rekurrente Verbindungen die Regelmäßigkeit von räumlichen Feuermustern verbessern können. Schließlich zeige ich, dass eine entsprechende Konnektivität, die diese Funktionen unterstützt, auf unüberwachte Weise entstehen könnte. Insgesamt trägt diese Arbeit zu einem besseren Verständnis der Prinzipien der neuronalen Repräsentation des Raumes im medialen entorhinalen Kortex bei. / High-level cognitive abilities such as memory, navigation, and decision making rely on the communication between the hippocampal formation and the neocortex. At the interface between these two brain regions is the entorhinal cortex, a multimodal association area where neurons with remarkable representations of self-location have been discovered: the grid cells. Grid cells are neurons that fire according to the position of an animal in its environment and whose firing fields form a periodic triangular pattern. Grid cells are thought to support animal's navigation and spatial memory, but the cellular mechanisms that generate their tuning are still unknown. In this thesis, I study computational models of neural circuits to explain the emergence, inheritance, and amplification of grid-cell activity. In the first part of the thesis, I focus on the initial formation of grid-cell tuning. I embrace the idea that periodic representations of space could emerge via a competition between persistently-active spatial inputs and the reluctance of a neuron to fire for long stretches of time. Building upon previous theoretical work, I propose a single-cell model that generates grid-like activity solely form spatially-irregular inputs, spike-rate adaptation, and Hebbian synaptic plasticity. In the second part of the thesis, I study the inheritance and amplification of grid-cell activity. Motivated by the architecture of entorhinal microcircuits, I investigate how feed-forward and recurrent connections affect grid-cell tuning. I show that grids can be inherited across neuronal populations, and that both feed-forward and recurrent connections can improve the regularity of spatial firing. Finally, I show that a connectivity supporting these functions could self-organize in an unsupervised manner. Altogether, this thesis contributes to a better understanding of the principles governing the neuronal representation of space in the medial entorhinal cortex.