Decreased path integration, measured in a virtual reality task, is associated with NFTs in the EC, as observed in human brain autopsies.
Maintenance of spatial memory across ages in humans. To investigate age-dependent changes in brain function, we studied path integration and spatial memory in volunteers ranging from 20 to 89 years old. All volunteers self-reported that they were free of dementia, and their spatial learning and spatial memory were assessed. Spatial memory was assessed in these human volunteers using a task similar to the Morris water maze test in rats (26); subjects were given five learning trials in which they explored a specific location in a virtual 3D space using virtual reality (VR) goggles (Fig. 1A), and spatial memory was then assessed with a probe trial. All age groups showed a similar traveling distance in the learning curve (Fig. 1B). In the probe trial, the error score, which is the total distance from the goal during the trial, tended to increase (Fig. 1C) with aging, and the time spent in the target quadrant decreased with increasing age (Fig. 1D). However, even the oldest age group (subjects in their 80s) spent at least 15 s in the target quadrant (Fig. 1D), suggesting that most subjects, including older subjects, maintained their spatial memory, even though it tended to decrease with aging.
Impaired spatial navigation with increasing age in humans. Next, we investigated path integration in the same participants of varying ages who demonstrated the maintenance of spatial learning and memory. Path integration was measured by having the participant go to a specific location A (yellow flag) in the virtual scene, then to a different location B (red flag), and then return to the starting point (Fig. 2A). The distance between the subject’s final location and the actual point (the error distance) was used as an index of path integration performance. Figure 2B shows the error distances in the eight age groups, from subjects in their 20s to those in their 80s. The average error distance in each age group remained relatively constant through subjects in their 40s but increased beginning from subjects in their 50s to those in their 80s. Compared to the average error distance for the 20s age group, a significant increase in error distance was observed for the 70s and 80s age groups. When the probability distributions were calculated for each age group (Fig. 2C), the error distances fell near the mean for subjects in their 20s to 40s, while the error distances for those in their 50s to 80s were much further from the mean. The variance in the probability distributions of error distances was similar in the 20s-40s age groups but increased significantly in the 50s-80s age groups (F test, P < 0.05). After the age of 50, individual error distances in the path integration test were substantially more variable, and the mean value increased, suggesting that the proportion of people with impaired EC-related navigation increases with age.
Positive correlation between impaired spatial navigation and NFT formation in the EC. Since the presence of NFTs was very low in people in their 20s, we used 5 virtual meters (vm), which is close to the maximum error distance of people in their 20s, as a threshold to calculate the percentage of people with increased error distances in each age group. We next sought to determine whether NFT formation in the EC could be related to the observed decrease in spatial navigation with aging. The proportion of people in each age range with NFTs was calculated based on the report by Braak and Braak (27) and compared to the proportion of people in that specific age range for which an error distance greater than 5 vm was measured. Both proportions increased with age (Fig. 2D). The proportion of subjects in each age range with NFTs in the EC and the proportion of people for which an error distance greater than 5 vm was measured were highly correlated (correlation coefficient = 0.96, Fig. 2D inset). When the threshold was set at 6 vm or higher, this high correlation with the frequency of NFTs was no longer obtained. Therefore, this analysis suggests that subjects for which an error distance greater than 5 vm was measured may have NFTs in the EC.
Accumulation of phosphorylated tau in the EC impairs path integration in a mouse model.
To investigate whether path integration performance reflects the formation of NFTs in the EC, we developed a behavioral paradigm to evaluate path integration in mice, the so-called L-maze. We investigated the effects of suppressed EC activity, overexpression of mutant tau (in PS19 mice that overexpress tau in a manner that results in NFT formation), and distribution of tau pathology on path integration in this behavioral paradigm.
Suppression of EC activity with DREADDs and impaired path integration. First, we tested whether L-maze performance reflected neuronal activity in the EC. To inhibit neural activity in the EC in wild-type (WT) mice, we used designer receptors exclusively activated by designer drugs (DREADDs): AAV-hSyn-hM4Di-mCherry was injected into the bilateral EC (Fig. 3A). Mice performed in the L-maze with and without an intraperitoneal administration of clozapine N-oxide (CNO). In the CNO group, neural activity in the EC should be suppressed, and path integration in the L-maze was impaired compared to that in the vehicle injection group (Fig. 3B), indicating that path integration in the L-maze depended on neural function in the EC.
Impaired path integration in tau-overexpressing mutant mice. Next, we assessed the performance of PS19 mice at 3 and 6 months of age in this behavioral paradigm, because at 6 months old, pathological changes in phosphorylated tau begin to be observed in the EC/ hippocampus regions. At 3 months of age, both WT and PS19 mice behaved similarly in the L-maze (Fig. 4A). However, at six months of age, PS19 mice showed a significant increase in the angular error toward the goal in the L-maze compared to that of WT mice of the same age (Fig. 4B), suggesting that path integration was decreased in the PS19 mice by 6 months of age.
Six-month-old PS19 mice were also examined in other behavioral paradigms, including spatial memory and novel object recognition. A Barnes maze, which depends on hippocampal activity, was used to evaluate spatial learning and memory in mice. Two-way analysis of variance (ANOVA) showed no significant difference in the total distance traveled to the goal between PS19 and WT mice (Fig. 4C) during learning, but a significant effect of day was detected. In the probe trial, there was no significant difference in the error score (Fig. 4D) or the percentage of time spent in the target quadrant between PS19 and WT mice (Fig. 4E), suggesting that six-month-old PS19 mice showed typical hippocampal-dependent learning and memory. In a novel object recognition test, no significant difference in discrimination score was observed between WT and PS19 mice at 6 months of age (Fig. 4F), and no significant difference in total search time was observed (Fig. 4G). Taken together, these results suggest that 6-month-old PS19 mice do not exhibit impairments in hippocampal-dependent spatial cognition and novel object recognition, which both use visual cues; however, they exhibit a deficit in EC-dependent path integration.
Tau pathology in the EC of mutant mice and impaired path integration. To understand the relationship between deficits in path integration and pathological changes in PS19 mice, we investigated tau pathology using an AT8 phosphorylated tau antibody (Fig. 5). At 4 months of age, no difference in AT8 staining was observed between the WT and PS19 mice under low-magnification microscopy (Fig. 5A). At higher magnification, PS19 mice showed sparse positives in the neurites, but staining in the cell body was rarely observed in the EC (Fig. 5B). In mice at 6 months of age, we observed a relatively strong positive signal in the EC and a faint signal in the hippocampus (Fig. 5C). At higher magnification, some somatodendritic cells and most neurites in the EC were strongly stained with AT8 (Fig. 5D). In the hippocampus, very few somatodendritic cells and neurites showed positive staining (Fig. 5E-G). To compare age-related changes in the number of phosphorylated tau-positive cell, we next counted the AT8-positive neurons in the EC and hippocampus. The number of AT8-positive neurons in the EC was significantly increased at 6 and 9 months of age compared to the number in mice at 3 months of age (Fig. 5H). The number of AT8-positive neurons in the dentate gyrus and CA1 and CA3 of the hippocampus showed a similar trend. Six-month-old PS19 mice did not show a significant increase in the number of AT8-positive neurons compared to that in 4-month-old mice. However, at 9 months of age, the number of AT8-positive neurons was significantly increased (Fig. 5I-K). These results suggest that the accumulation of phosphorylated tau in the EC, but not the hippocampus, of PS19 mice at the age of 6 months caused path integration deficits without affecting hippocampal-dependent spatial cognition or novel object recognition.