The hippocampus plays an important role in spatial and non-spatial episodic memory. Hippocampal cell recording found not only a direct link between the external world and its representation in the brain (e.g. place cells), but also an effect of contextual properties as timing, history and reward on encoding of experiences and formation of memories⁽¹⁾. However, cellular recording is spatially limited and cannot offer information at the system level. MRI is a potent tool for investigating whole-brain structures in-vivo, and DTI in particular is a sensitive marker of quick plasticity processes⁽²⁾. Using DTI and the Morris water maze we investigated spatial preference for place and time as revealed by changes in diffusion indices induced by learning of a specific location in the maze and the overall training experience. We set to explore whether system-level mapping of spatial and temporal preference can be found in the hippocampus.

Rats completed 3 blocks in the maze (each comprised of 4 trials). First group was divided into 4 subgroups; each swam to a specific platform location (SE or SW or NSE or NW). Time interval between blocks was 45 minutes. (G45: SE45 n=13; SW45 n=12; NE45 n=11; NW45 n=16). A second group was also divided into subgroups based on platform location, but time interval between blocks was 2 hours (G2h: SE2h n=8; SW2h n=8; NE2h n=9; NW2h n=10). A third group included 2 more subgroups that completed maze training when time interval between blocks was 10 minutes or 3 hours, and platform location was in the SE quadrant (GrTime: SE10m n=8; SE3h n=10).

Scanning protocol: Rats were scanned with a DTI protocol 2-3 days before the task and 45 minutes following the last maze training (matrix size: 128X128, 32 directions, 2 b0, repeated 3 times).

Image processing and statistical analyses: DTI was calculated using Explore DTI³. Images of mean diffusivity (MD) were normalized to a rat template using SPM⁴. Voxel-based repeated measures ANOVA were calculated for each group using MATLAB (GrTime included SE45 and SE2h).

Behavioral results: Time to reach the hidden platform (latency) reduced as the learning progressed (paired t-test, p<0.05, Figure 1).

Imaging results: Reduction in MD was found in the hippocampus, striatum, thalamus and cortical areas, which is in good agreement with previous DTI short-term plasticity studies2. The reduction in MD in the hippocampus of each subgroup was used as an input for spatial preference analyses.

Preference analyses: Preference of change was computed in hippocampal voxels that showed significant reduction in MD as the change of each subgroup relative to the overall change in the group. Subgroups present distinct local areas with higher probability of change (Figure 2). A follow-up analysis compared all 10 subgroups together, and the percentage of change was calculated for the 10 subgroups in voxels significant for the main effect of training. Points of no preference were evaluated based on low variance of preference in each voxel. 21 points were found at the borders of the dentate gyrus in both hippocampi (Figure 3A). Hypothesizing that these points act as spatial/time pin-wheels, we created spheres at radius of 6 pixels around the no-preference points (Figure 3B-D), and analyzed the preference pattern on the sphere surface (spheres were computed around these pinpoints separately for each group). Registration to a randomly chosen sphere from group Gr45 was performed for all spheres in that group. The same transformation was then written on spheres of GR2h and GrTime. Following registration an averaged "ball" was computed for each group (Figure 4), revealing a similar pattern of spatial preference across most spheres, as well as preference for time, or experience duration.

In the current study we found a system-level mapping of space and time in the rat hippocampus. Using diffusion MRI, a marker of tissues microstructure, we located regions within the hippocampus that exhibit higher probability to undergo structural changes based on learning of a specific location in the maze and the length of the training episode. It seems that at list in the rat brain, the spatial and temporal characteristics of experience are interrelated and presented beyond the cellular level.