Intracellular hippocampal attractor dynamics

Memory research constitutes one of the most popular research areas in Neuroscience. The keystone was the examination of the patient H.M. in the 1950s, diagnosed with severely affected capacity for generating long-term memories after the bilateral removal of his temporal lobe regions. Today, the hippocampus is implicated in spatial memory formation and retrieval and harbours principal cells that show spatially modulated receptor fields (i.e. place cells), while cellular mechanisms underlying memory recall in the hippocampus remain enigmatic. To address this question, this project aims to use the spatial representation system of the hippocampal CA3 sub- region. This region constitutes the second processing stage in the hippocampal trisynaptic loop and forms excitatory interconnections within itself (i.e. recurrent collaterals). A separate input system represent the monosynaptic innervations from the entorhinal cortex suggested to provide the hippocampus with sensory information of a given environment. Specifically, the recurrent network connections were proposed to facilitate autoassociative network operations which are essential for error correction dynamics when receiving partial or noisy incoming information of a memory (i.e. pattern completion). This project aims to address this theory and investigate error connection dynamics on the single cellular level. To unravel the cellular mechanisms for pattern completion, whole-cell patch-clamp and local field potential recordings will be combined and place cell activity recorded in awake, behaving mice running on a linear treadmill. With this experimental design intrinsic properties and spike-firing modes of CA3 place cells will be first described and quantitatively characterised. However, in order to investigate error correction network dynamics, environmental cues will be manipulated to differentially record from single cells representing either attractor units of the network or non-coherent representations of the altered environment characterised by disappearing place fields. The gained information constitutes an important step toward understanding attractor dynamics on the cellular level and bridges theories proposed by in-vitro experiments and modelling approaches with in-vivo behavioural studies and will be therefore directly applicable for hippocampal autoassociative network models. On a wider scope it might facilitate our understanding of pathophysiologies of memory related brain disorder including epilepsy and schizophrenia.

One fundamental aim of memory research is to develop a basic understanding of the characteristics and dynamics of memory processes. A key element was the examination of the patient HM in the 1950 who lost the capability of generating new long-term memories after his hippocampus was removed to treat his epileptic seizures. Today we know that the hippocampus is involved in spatial and episodic memory encoding. One dynamic the brain uses to create information-rich memory traces is to synchronize brain regions with oscillatory brain waves or rhythmic patterns of neural activity. In the hippocampus, the low frequency theta oscillations appear when a rodent is engaged in active motor behavior such as exploratory movement or immobile attentive states and represents an online state for temporal coding and decoding. As soon as a memory is formed, co-active neurons get sequentially activated in compressed time windows of 100ms to consolidate such a memory trace, also termed memory engram, which appear as high frequency oscillations termed sharp-wave-ripples in electric-field recordings. This study aimed to shed light on the intracellular activity-pattern of single neurons in the memory recall center, the CA3-region, during these distinct oscillatory or behavioral states. We started with a simple question: what activity-pattern do neurons use during active behavior in the hippocampal CA3-region, and can we detect an activity gradient within the CA3-region considering its involvement in diverse memory processes? To answer these questions, we trained mice to run on a linear treadmill apparatus. The treadmill was enriched with textures and objects to create short spatial memory traces. With an advanced electrophysiological setup, we recorded brain oscillations/network activity and intracellular activity of single CA3 principal neurons in parallel. As shown in various brain regions, CA3-neurons use a ternary code to encode information comprised of silence, single spikes and bursts of five-to-ten spikes in less than 60ms time windows. Spiking activity of CA3-neurons is highly theta-modulated, both single spikes and bursts of spikes. Phase-preferences of these activity-patterns during theta-oscillations suggest that single spikes or bursts are preferentially forwarded by distinct input regions. However, inputs occur during the memory retrieval phase. Bursting activity of CA3-neurons appears very heterogeneous in duration and number of spikes. Within the CA3-subfield we detected four main clusters of spike bursts including short firing episodes of spike-doublets, short complex spike bursts and sustained complex spike bursts with topological gradient along the CA3-transverse axis. The gained information constitutes an important step toward understanding neural-activity patterns during active behavior. In future, these activity patterns can be applied to study plasticity mechanisms at neural communication sites. Moreover, analog or digital coding mechanisms involved in complex spike bursts can be investigated to unravel the basis of memory encoding in the future.