Anterior Thalamic Inputs Are Required for Subiculum Spatial Coding, with Associated Consequences for Hippocampal Spatial Memory

Abstract

Just as hippocampal lesions are principally responsible for "temporal lobe" amnesia, lesions affecting the anterior thalamic nuclei seem principally responsible for a similar loss of memory, "diencephalic" amnesia. Compared with the former, the causes of diencephalic amnesia have remained elusive. A potential clue comes from how the two sites are interconnected, as within the hippocampal formation, only the subiculum has direct, reciprocal connections with the anterior thalamic nuclei. We found that both permanent and reversible anterior thalamic nuclei lesions in male rats cause a cessation of subicular spatial signaling, reduce spatial memory performance to chance, but leave hippocampal CA1 place cells largely unaffected. We suggest that a core element of diencephalic amnesia stems from the information loss in hippocampal output regions following anterior thalamic pathology.SIGNIFICANCE STATEMENT At present, we know little about interactions between temporal lobe and diencephalic memory systems. Here, we focused on the subiculum, as the sole hippocampal formation region directly interconnected with the anterior thalamic nuclei. We combined reversible and permanent lesions of the anterior thalamic nuclei, electrophysiological recordings of the subiculum, and behavioral analyses. Our results were striking and clear: following permanent thalamic lesions, the diverse spatial signals normally found in the subiculum (including place cells, grid cells, and head-direction cells) all disappeared. Anterior thalamic lesions had no discernible impact on hippocampal CA1 place fields. Thus, spatial firing activity within the subiculum requires anterior thalamic function, as does successful spatial memory performance. Our findings provide a key missing part of the much bigger puzzle concerning why anterior thalamic damage is so catastrophic for spatial memory in rodents and episodic memory in humans.

Keywords: amnesia; diencephalon; hippocampus; memory; space; subiculum.

Figure 1.

NeuN-reacted coronal sections showing the status of the AV (anteroventral), AM (anteromedial) and AD (anterodorsal) thalamic nuclei in control (A) and lesion (B) animals. The nucleus reuniens (Re, arrowed), as shown using calbindin-reacted sections, was intact in both control (C) and lesion (D) animals, indicating that reuniens damage was not responsible for deficits seen in ATN-lesioned animals. E, The ATN cell count was significantly reduced in lesioned animals (ATNx) compared with controls (Control). Nissl-stained coronal sections helped to confirm electrode placement in dorsal subiculum (F), with the electrode path indicated. G, Schematic representing cannula placement (blue) and the two targets of the infusion needle (red). H, Cresyl violet-stained section indicating cannula placement, with DAB-reacted Flurogold infused to indicate spread of muscimol. Black line indicates canula placement. Red line indicates the track of the infusion needle. Dashed white line indicates the spread of the muscimol. ***p < 0.001 (Welch's two-sample t test). Scale bar, 800 mm. RSg, retrosplenial cortex.

Figure 2.

A, Schematic diagram of spatial alternation task. ATN-lesioned animals (ATNx) showed a significant deficit in spatial alternation compared with both control and sham animals (B). C, Schematic of novel object recognition task. There was no difference between control (Control) and lesion rats in cumulative D1 (D), D2 (E), or total exploration time (F). ***p < 0.001 (ANOVA with Tukey post hoc). Error bars indicate SEM.

Figure 3.

Representative single units recorded in the dorsal subiculum in control (Control; A) and ATN lesion (ATNx; B) animals. For individual units, the figures illustrate the following: heatmap of spike location adjusted for time spent in each location; path in arena (black) with spike location (blue); head direction; mean spike waveform. For the Control group, different classes of spatial cells are displayed, along with a nonspatial cell. No spatial cells were recorded in the ATNx cases. HD, Head direction.

Figure 4.

Properties of bursting and nonspatial subicular cells. A, B, Waveforms and autocorrelation histograms were used for cell classification. C, Diagram of waveform properties. Bursting cells in Control showed a higher burst duration (E), had more spikes per burst (F), and had a higher propensity to burst (H). Bursting cells in ATNx had a greater spike width (D) and higher IBI (I), than nonbursting cells. J, K, Nonspatial cells in ATN had higher spike width and spike height than nonspatial cells in Control animals. D-K (boxplots), Filled circles represent outliers. Unfilled diamonds represent the mean. *p < 0.05; ***p < 0.001; Welch's two-sample t test or Mann–Whitney U test.

Figure 5.

A, When Control rats are compared with rats with permanent NMDA lesions of the ATN (ATNx), more short-duration waveforms were recorded in Controls, but the samples obtained from the two groups of rats are not distinct. B, C, In the permanent lesion study, narrow waveform cells were more often spatial, but the classification of wide and narrow waveform cells was mixed. B, Combined data for the Control and ATNx rats. C, Data for only Control rats. D, Similarly, in rats where the ATN was inactivated with muscimol, narrow waveform cells were more often spatial, but the groups were mixed.

Figure 6.

A1-A3, Examples of spatial units before, during, and after muscimol infusion. Spatial properties of single units decreased when ATN was inactivated. A3, Relative inactivity after 100-120 min, and the cell was not recorded the next day. B, Spatial alternation dropped to chance levels when the ATN were temporarily inactivated with muscimol, compared with the same animals before infusion. When saline was infused in place of muscimol, no deficit was present. C1-C3, Further examples of spatial units before and after ATN inactivation. C1, Disruption of place field shortly after muscimol infusion with some head directionality remaining, which was later disrupted. C3, No firing was detected after muscimol infusion. **p < 0.01 (ANOVA with Tukey post hoc).

Figure 7.

Spike properties following temporary inactivation of ATN with muscimol (M). A, Inactivation of ATN caused no significant decrease in single-unit firing frequency. B, Animals showed decreasing levels of activity throughout the experiment, although there was no significant correlation between distance traveled and spike frequency (C). D, Firing frequency of each cell recorded at baseline (left), in 5 min bins throughout the experiment. First white line indicates ATN inactivation with muscimol, after 15-20 min of baseline recording before infusion. In most cases, electrophysiological recording was paused for T-maze testing between 80 and 100 min (second and third white lines), then continued. Recordings in which the animal was largely inactive or asleep were excluded. Final white line indicates data from the day after infusion. E, There were no significant changes in spike firing in spatial units as a result of ATN inactivation and burst properties remained consistent, including the number of bursts (F). A, B, E, F, First red vertical line (M) indicates the infusion of muscimol. Second red vertical line indicates recordings taken the next day (ND). Each bin represents 20 min of recording. Data are compared with baseline, immediately before inactivation, with error bars indicating SEM. Theta-entrained cells are removed from A, B, C because of high firing frequency compared with other cell classes. ***p < 0.001 (Welch's two-sample t test with Bonferroni correction).

Figure 8.

Representative electrode placement in CA1 (A) and ATN lesion (B), in NeuN-reacted sections. C, Animals with ATN lesions and electrodes implanted in CA1 showed a significant deficit in spatial alternation task compared with Control animals (control data repeated from Experiment 1). D1-D3, The same cohort of ATNx animals showed no deficit in object recognition on bow-tie maze. E, Representative place cells recorded from CA1 in 3 ATNx animals. ***p < 0.001 (Welch's two-sample t test).