The role of the dorsal and ventral hippocampal subregions and their impact on the medial prefrontal cortex in spatial working memory

Abstract

Spatial working memory (SWM), or the ability to remember a spatial cue over a brief period of time to guide a subsequent action, is a strategy employed by a wide variety of species to navigate their environment. Conceptually, this cognitive process can be divided in three different phases that consist of encoding a behaviorally relevant spatial stimulus, its maintenance over a delay period, and its retrieval to inform an upcoming choice. In the rodent brain, SWM engages the medial prefrontal cortex (mPFC) along with the hippocampus (HPC) and depends on the communication between these two regions. Previous studies have shown that both the dorsal and the ventral hippocampal subdomain interact with the mPFC during SWM performance, which manifests for example through increased synchronous activity in the two brain regions across various frequencies, and can also be mediated via direct axonal projections originating mainly in the ventral hippocampal subdomain. But while recent findings suggest that the mPFC is critically involved in all SWM phases, it is not clear which processing stages depend on the dorsal or on the ventral hippocampus (dHPC, vHPC), and which task-relevant information is transmitted from each subdomain to the mPFC. I addressed these questions by silencing neuronal activity in the dHPC or vHPC during the individual phases of SWM and simultaneously recording electrophysiological signals in the mPFC. To implement this approach, I first compared the efficacy of two different optogenetic methods to silence the activity of excitatory neurons in the HPC. In transgenic VGAT-ChR2 mice expressing channelrhodopsin in GABAergic cells throughout the brain, neuronal silencing can be achieved indirectly by activating inhibitory interneurons in the brain region of interest, which then target surrounding excitatory cells. On the other hand, optogenetic silencing can be realized in wildtype mice by inducing the expression of the inhibitory opsin ArchT in excitatory neurons through injection of a viral vector in the respective brain region. When I attempted to transiently inactivate neurons in the dHPC or in several cortical regions using VGAT-ChR2 mice, I found that the inhibitory effect extended far beyond the light source causing decreased firing rates in areas outside of the targeted brain region. This was even the case when using relatively low light intensities which came at the expense of silencing strength closer to the light source. In contrast, virus-mediated expression of ArchT enabled robust silencing of excitatory neurons that was restricted to the area of opsin expression. These results demonstrated so far undescribed constraints on using inhibitory interneuron activation to silence neuronal firing and additionally highlighted an effective alternative approach to directly inhibit excitatory neurons. This laid the groundwork for inactivating the dHPC or vHPC during different SWM phases to decipher their role in behavior and their influence on neuronal activity in the mPFC. In the second part of this study, I therefore expressed ArchT in excitatory neurons of the dHPC or vHPC and trained the mice to perform a SWM task in a T-maze, which was designed in a way that the phases encoding, maintenance and retrieval were separated in time and could be manipulated individually. Inhibition of these brain regions during the different SWM phases revealed that both the dorsal and the ventral hippocampal subdomain are necessary for SWM, but assume complementary functions. While inhibition of the dHPC during goal encoding or retrieval, but not during maintenance, impaired successful SWM performance, the vHPC was only required in the encoding phase and its inhibition during goal retrieval or maintenance did not impact task execution. However, both dHPC and vHPC inactivation led to a slightly reduced running speed of the animal, irrespective of the SWM phase. This finding contrasts the hippocampal role in SWM with its involvement in other behavioral parameters. Neuronal activity in the mPFC was modulated both by dHPC and by vHPC inactivation. However, the influence of the two hippocampal subdomains on task-related spatial firing patterns in the mPFC was more complex than previously expected. Prefrontal activity representing the different goal locations in the T-maze was selectively impaired by vHPC, but not dHPC, inhibition when a spatial goal had to be encoded, while mPFC goal representations without relevance for the imminent future were unaffected. Since the process of goal encoding is critical for successful performance of the SWM task, it is all the more surprising that the dHPC with its prominent spatial code is not involved in these goal representations in the mPFC. However, I found that spatial firing patterns relating to the distance between start and goal of the maze were altered both by dHPC and vHPC inhibition, but again only during the encoding phase and not during memory retrieval. Yet even in the encoding phase the mPFC did not lose its ability to code for space when the HPC was inhibited as instead new, but equally consistent, spatial firing patterns were formed. Nevertheless, specifically when information from the dHPC was missing during memory encoding, prefrontal neurons synchronized their spiking activity more strongly to ongoing theta oscillations in the dHPC, while vHPC inactivation had no effect on its theta-mediated synchrony with the mPFC. Taken together, these results point to a complex functional division between the dorsal and ventral hippocampal subdomains regarding their own role in behavior, but also in signaling spatial information to the mPFC, with both of these factors highly depending on the momentary requirements of the SWM task. This study complements the growing literature on the hippocampal-prefrontal network and its critical function in spatial learning and memory.