Integrating and storing visuo-spatial cues in the retrosplenial cortex

Where and how memory is encoded and stored in the brain is one of the 'holy grails' of neuroscience. Nearly 100 years ago Richard Semon hypothesised that external stimuli produce a "permanent record,... written or engraved on the irritable substance," i.e. the brain, and coined the term 'engram' for this memory trace. But only a handful of studies have so far been able to identify engrams for specific situations, mostly involving fear memory in brain areas such as the hippocampus and amygdala. In our pilot study we were the first to demonstrate that a specific pattern of neuronal activity in an area of the brain known as the retrosplenial cortex (RSC) is directly correlated with the performance of mice in a spatial memory task in a so-called radial arm maze. The RSC has emerged as a key area involved in episodic and topographical memory in humans as well as spatial memory in rodents. The dysgranular portion (Rdg) receives dense inputs from both visual areas and the hippocampal complex (parahippocampal region and subiculum), and it contains spatially-responsive cells, making it a prime candidate for integrating navigational information.
Our pilot study did not tell us what those RSC engrams actually represent: they could reflect a representation of the environment, a representation of rewarded, remembered locations in the environment, or a representation of the animal's movement and navigation within the environment. This question is linked to the relative importance of hippocampal vs. visual inputs for RSC engram formation. In order to tackle it we will pharmacologically inactivate either the hippocampus or the visual cortex and carry out a number of manipulations on the maze and the reward the mice receive for remembering the correct locations. We will then compare the performance of the animals with stability of the RSC engrams.
The second question we will tackle is whether visual input to the Rdg is necessary for spatial memory and if this role is time-limited such that other inputs, such as tactile or self-motion cues, can substitute for visual input. We will achieve this by either training mice in a normally lit environment and then successively remove visual cues, finally transferring them into a dark room during the memory testing phase, or by training and testing animals in darkness, thereby maximising the potential for cross-modal plasticity, and re-testing them after removal of tactile cues, leaving only self-motion (path integration) cues.
The third question concerns how visual and locomotion inputs are integrated in the Rdg. We will again use fluorescent activity indicators to test whether cells activated as constituents of a spatial memory engram (<10% of all Rdg cells) also respond to visual stimuli and/or during locomotion. We will train head-fixed mice on a virtual corridor task on a linear treadmill and analyse the correlation of individual neurons with task performance. By removing visual stimuli or pharmacologically silencing visual inputs to Rdg or by creating a mismatch between visual stimuli and path length, we will establish whether Rdg neurons involved in spatial memory engrams depend primarily on visual input or path integration.
Finally, we will investigate how the spatial memory network, involving the hippocampus, retrosplenial cortex and visual cortex, interacts over time, and to what extent engram formation and/or retrieval depends on them. We will pharmacologically inactivate either the hippocampus or the visual cortex and assess the effects on engram formation or expression. In addition, we will record electrical activity in retrosplenial cortex, hippocampus and visual cortex at the same time, and we will analyse the direction of information flow between the three brain areas at different stages of the acquisition and retrieval of spatial memory.

The retrosplenial cortex (RSC) is a key brain area supporting spatial memory and navigation. It receives strong inputs from the hippocampal complex (parahippocampal region, subiculum) as well as visual cortical areas. In a pilot study, we demonstrated for the first time spatial memory engrams in mouse dysgranular RSC (Rdg) in the form of specific patterns of activity. Their stability is correlated with the degree of spatial memory retention. We also found robust visual responses in Rdg which were strongly modulated by locomotion.
Our initial study did not tell us what those engrams represent: the environment, rewarded locations in the environment, or the animal's movement and navigation within the environment. It also did not tell us whether engrams depend on hippocampal or visual inputs for their formation, and how these inputs interact in Rdg over time during memory acquisition. We will address these questions using calcium imaging in mice that will either (1) be implanted with a miniscope and will be freely moving in a radial arm maze, or (2) will be trained in a virtual corridor and imaged head-fixed using a standard two-photon microscope.
We will manipulate intra- and extra-maze visual cues, spatial parameters of the maze as well as reward magnitudes. We will use DREADDs to inactivate specific populations of neurons in visual areas of the dorsal stream or the dorsal hippocampus, both of which form functional networks with Rdg. We will further test the role of Rdg engrams in cross-modal object memory by successively removing visual cues or placing mice in the dark at different points during memory acquisition or testing. Finally, we will investigate the direction of information flow between Rdg, hippocampus and the visual cortex, and how this develops over time during of the acquisition and retrieval of spatial memory. This will be achieved through recordings of local field potentials in the 3 areas at the same time, and Granger causality analysis.

Memory traces ('engrams') are one of the holy grails of neuroscience. In a study leading up to this project we have been able to demonstrate engrams of spatial memory in the retrosplenial cortex, a brain region that is important in humans as well as rodents for navigation and spatial memory. The proposed research will add to the growing body of knowledge about how connected brain regions interact to encode and store memories and how this varies according to the types of cue that is being used. While the most immediate beneficiaries will be other researchers with an interest in memory there are a number of likely beneficiaries outside this group. Foremost among these will be the large number of scientists in both academia and industry working on memory loss, dementia and neurodegenerative diseases such as Alzheimer's disease: the retrosplenial cortex is one of the first brain regions to show deficits at prodromic stages of such conditions and in mild cognitive impairment. We have already established a collaboration with Prof. Mark Good (School of Psychology, Cardiff University) and Prof. Patrick Kehoe (University of Bristol) who have experience with behavioural and electrophysiological assessments of mouse models of AD as well as therapeutic approaches. Two-photon imaging will provide us with a powerful tool for longitudinal assessment of neuronal activity that we will apply to physiologically relevant mouse models of AD which display age-dependent amyloid accumulation and cognitive impairments.
We are planning to hold a symposium on the retrosplenial cortex in Cardiff, in order to bring together the growing number of researchers in the field to learn about the latest advances, exchange ideas and forge new collaborations.
An understanding of how memory representations are stored in this brain region normally is essential to understand how these processes may be disrupted in animal models of neurodegenerative diseases. Cardiff is at the forefront of research in this field, with both the MRC Centre for Neuropsychiatric Genetics and Genomics and the Cardiff Dementia Research Institute.
We will achieve impact by communicating our findings to the scientific community by means of publications in high-profile journals, by presenting them at major conferences as well as more focused specialist meetings both in the UK and worldwide, and we will make our data available for other researchers to build on. We will explore new collaborations with researchers in the field of dementia and neurodegenerative diseases.
We will achieve wider impact by communicating our findings to members of the public in an accessible manner, from school children and sixth-form students to patient groups, charities and policy advisors. This will be achieved through our annual Brain Games, through public lectures and open afternoons showcasing our research.
A third way to achieve impact is through advocacy. This is most effectively done as part of an organisation. The PI is a Trustee of the Physiological Society, a charitable organisation whose purpose is to sustain the discipline of physiology (which includes neuroscience) through the advancement of science and education and thereby the advancement of human and animal health. The Trustees are legally responsible for ensuring that the charitable objectives are met. They engage with a number of stakeholders and policymakers, both within and outside of Government (e.g. Parliamentary Links Days, party conferences).