TgF344-AD Rat Model of Alzheimer’s Disease: Spatial Disorientation and Asymmetry in Hemispheric Neurodegeneration

INTRODUCTION

Alzheimer’s disease (AD) remains incurable despite advances in understanding this condition’s neuropathology and continuous efforts for new therapeutic approaches. One main obstacle to effective therapy is that the brain has already suffered extensive or unredeemable damage when patients are diagnosed. At the diagnostic stage, the AD brain is characterized by the formation of extracellular amyloid plaques (AP) from amyloid-beta (Aβ) peptide fibrils and of intracellular neurofibrillary tangles (NFT) from hyperphosphorylated tau (p-tau) protein [1]. Accumulation of AP and NFT leads to degeneration of neurons and synapses and eventually to the progressive cognitive decline typical of AD [2]. Accordingly, the AP and neuroinflammation correlate with the severity of cognitive impairment in AD patients, while drugs independently targeting both biomarkers fail at yielding significant clinical benefits [3]. AD starts decades before its clinical presentation and diagnosis [4]. Unfortunately, carrying out studies at such an early stage in prospective patients is difficult for practical and ethical reasons. Nevertheless, many challenges to answering pressing questions about the neuropathology of AD can be tackled with suitable transgenic rat models.

The “amyloid cascade hypothesis” states that Aβ-aggregation initiates a chain of pathophysiological reactions that ends in neuronal loss and the development of dementia [5]. The hypothesis is supported by genetic evidence that any mutation in the amyloid precursor protein (APP) or in presenilin 1 or 2 (PS1/PS2) eventually leads to Aβ accumulation and thus to the development of early-onset AD in humans [1]. This line of evidence led to the generation of Aβ-overproducing transgenic mice, which until recently served as the “gold standard” in AD animal modeling, but which did not develop robust tauopathy nor neuronal loss unless additional human transgenes were introduced [6–8]. Rats, on the other hand, are phylogenetically closer to humans with regards to key physiological aspects such as the number of tau isoforms [9, 10]. Therefore, the transgenic rat line TgF344-AD, which was generated from Fischer 344 rats, is one of the most suitable animal models for AD research, as it manifests in an age-dependent manner, similar to patients, the complete repertoire of AD’s pathological hallmarks: cerebral amyloidosis, tauopathy, oligomeric Aβ, gliosis, apoptotic loss of neurons, and behavioral impairment [9]. Specifically, the TgF344-AD rat line expresses two mutant human genes: the “Swedish” APP (APPsw) and the Δ exon 9 PS1 (PS1ΔE9). Numerous studies have validated the phenotype of TgF344-AD rats [11–14], which, since their first description, have been used to test efficacy of deep brain stimulation [15, 16], Aβ attenuation therapy [17], anti-inflammatory agents [18], and low-dose brain radiation [19]. Converging data suggest that the time resolution of the symptoms’ progression in this transgenic line not only allows for studying AD at both early and advanced stages but also at a prodromal stage[20–23].

In our previous study, we showed that spatial memory is impaired in TgF344-AD rats, yet the exact mechanism remained unknown [15]. We, therefore, set out to determine if spatial memory deficits in TgF344-AD rats result from the development of spatial disorientation, which, as in AD patients [2, 24–26], may result from the asymmetry in hemispheric neurodegeneration. We used positron emission tomography (PET) to detect changes in glucose metabolism and microglial activity as a means of assessing neurodegeneration in vivo [27]. Also, we subjected TgF344-AD rats to the delayed match-to-sample (DMS) task to assess their ability to differentiate between left and right sides as well as to simultaneously measure their working memory performance.

METHODS

RESULTS

PET imaging

PET imaging with the TSPO ligand [¹⁸F]DPA-714 showed confined clusters of increased microglial activity in transgenic rats compared to wild-type rats, with no significant global changes observed (Fig. 2A). Bilateral clusters were found in the dorsal subiculum and temporal association cortex, while unilateral clusters were found in the retrosplenial cortex (left), reticular formation (left), ventral thalamus (right), lateral posterior thalamus (left), and hippocampus (left), with the latter having the highest SUVR_bg (1.21±0.06 in transgenic rats vs 1.03±0.06 in wild-type rats).

Similarly, PET imaging with [¹⁸F]FDG showed confined clusters of altered glucose metabolism in transgenic rats compared to wild-type rats, with no significant global changes observed (Fig. 2B). Hypometabolic clusters were found in the dorsal subiculum (right), hippocampus (right), hypothalamus (right), ventral thalamus (left), subthalamic nucleus (left), and pituitary gland (bilateral). The most pronounced hypometabolic cluster was found in the reticular formation (bilateral), with an SUVR_Cer of 0.88±0.03 in transgenic rats vs. 0.99±0.05 in wild-type rats. The only hypermetabolic cluster was found in the retrosplenial cortex (bilateral), with an SUVR_Cer of 1.21±0.03 in transgenic rats vs. 1.13±0.03 in wild-type rats.

OPEN IN VIEWER

Fig. 1

DMS task. A) Five out of ten transgenic (TG) rats were unable to complete the last stage of the training part, that is, when the delay between the match and sample phases was still absent. B) For statistical analysis, their last day of training was used as the minimum length necessary to learn the DMS task (^## unpaired two-tailed t-test, t(18) = 2.84, p = 0.01). C, D) The difference in the learning rate between transgenic and wild-type (WT) rats was due to a strong preference that the non-learners had for one of the two levers (^# unpaired two-tailed t-test, t(18) = –2.45, p < 0.05). E) Only rats without a strong preference for either lever proceeded to the testing part where the delay between the match and sample phases was 1–16 seconds. The results of a two-way mixed ANOVA with GG correction showed a decline in performance as the delay increased (F(4, 64) = 33.49, GG corrected p < 0.0001). However, this decline was independent of genotype (genotype: F(1, 16) = 1.68, uncorrected p = 0.21; interaction: F(4, 64) = 1.06, uncorrected p = 0.38). Different symbols are used to show individual rats.

OPEN IN VIEWER

Fig. 2

PET imaging. Shown are averaged PET images of (A) microglial activation measured with [¹⁸F]DPA-714 (6 transgenic and 6 wild-type rats) and (B) glucose metabolism measured with [¹⁸F]FDG (5 transgenic and 6 wild-type rats). Respective t-maps show significant differences (corrected for multiple testing) between groups. Red voxels: higher tracer uptake in transgenic rats. Blue voxels: higher tracer uptake in wild-type rats. White dashed lines in A indicate regions obscured by spillover from the ventricular system and pituitary gland, where natural TSPO expression leads to high tracer binding. DS, dorsal subiculum; Hip, hippocampus; Hyp, hypothalamus; LP, lateral posterior thalamus; Pi, pituitary gland; Ra, raphe; RC, retrosplenial cortex; Rt, reticular formation; STh, subthalamic nucleus; TeA, temporal association cortex; VT, ventral thalamus. Scale bar: 5 mm.

DISCUSSION

Our behavioral data demonstrate that half of TgF344-AD rats developed a clear preference for one of the two levers (left vs. right) during the training phase of the DMS task (delay = 0 s), indicating the presence of spatial disorientation. This is consistent with the results of our previous study showing that TgF344-AD rats exhibit spatial memory deficits in a modified Barnes maze, which can be attenuated by deep brain stimulation [15]. Interestingly, our present study also shows that TgF344-AD rats without spatial disorientation perform equally well as their wild-type littermates in the testing phase of the DMS task (delay = 1–16 s), suggesting that their working memory is intact. This finding is consistent with the results of Muñoz-Moreno et al., who used the nonmatch version of the same task and showed that TgF344-AD rats exhibit a similar percentage of correct responses as their wild-type counterparts across all delays [35]. In a novel object recognition (NOR) test, TgF344-AD rats are reported to show no side bias [20], which somewhat contradicts our results of spatial disorientation in the DMS task. However, the reason for this difference is likely because in the NOR test, rats can face objects from any direction, while in the DMS task, rats can face levers only from one direction. This ensured that, in our DMS task, the left lever always remained on the left, while the right lever on the right.

In a transgenic mouse model of AD, whole-brain glucose hypermetabolism was shown to precede glucose hypometabolism [36, 37], which in turn is shown to positively corelate with microglial activity in early-onset AD patients [27]. Activation of microglia and astrocytes around AP is classically regarded to mark the onset of neuroinflammation in AD [38]. Our PET data on resting glucose metabolism and microglial activity suggest a mosaic of focal changes, which is in line with magnetic resonance imaging studies showing disruption of structural and functional networks in TgF344-AD rats as early as 5 months of age, confirming the notion of AD as a disconnection syndrome [35, 39].

Our PET results show an increase in microglial activity in the hippocampus, dorsal subiculum, thalamic nuclei, as well as in the temporal association and retrosplenial cortices of TgF344-AD rats. This increase is consistent with another PET study that used the same TSPO ligand (i.e., [¹⁸F]DPA-714), showing no change in TgF344-AD rats at 6 months of age, yet an increase in neuroinflammation at 12 months in the hippocampus and at 18 months in the thalamus and frontal cortex [20]. In the hippocampus of TgF344-AD rats, TSPO overexpression was previously reported at 12 months of age in astrocytes, and at 24 months of age in microglia, supporting the notion of TSPO as a neuroinflammatory marker [40]. Our PET findings are also consistent with existing magnetic resonance spectroscopy (MRS) data on increased neuronal dysfunction in the hippocampus of TgF344-AD rats. However, while the MRS data are based on a decrease in N-acetyl-aspartate levels occurring at 18 months of age [20], our study found neuronal dysfunction in the form of impaired glucose metabolism and increased microglial activity in TgF344-AD rats that were 14–16 months of age.

Of relevance in our study are hypometabolic changes in the hypothalamus and pituitary gland of TgF344-AD rats, as they support the recent hypothesis that dysregulation of the hypothalamic-pituitary-adrenal axis may help in the diagnosis of prodromal AD [41]. Also important in our study is the coincidence of spatial disorientation with increased microglial activity and altered glucose metabolism in the thalamus and retrosplenial cortex since dysfunction of these two interdependent structures represents one of the earliest signs of prodromal AD in patients [42]. Similarly, in a mouse model of amyloid pathology, dysfunction of the thalamus and retrosplenial cortex precedes overt AP formation [43].

Asymmetric neuroinflammation has previously been reported in Aβ mouse models (expressing mutant human APP alone or in combination with PS1/PS2) and was accompanied by fibrillary amyloidosis in the affected areas [44]. The asymmetry in our PET data is in line with clinical observations that brain atrophy in AD patients is bilateral but not always symmetrical [2, 24–26]. Therefore, some AD patients may develop atrophy predominantly in one of the hemispheres. Often such patients have contralateral spatial neglect, which is not compensated by the healthier hemisphere since it is also partially affected by neurodegeneration [24]. This is consistent with our PET findings, showing that unilateral microglial activation in TgF344-AD rats is accompanied by contralateral glucose hypometabolism in structures such as the ventral thalamus and hippocampus.

In conclusion, TgF344-AD rats display hemispherically asymmetrical neurodegeneration and spatial disorientation, with past clinical research suggesting a causal relationship between the two [2, 24–26]. When, however, TgF344-AD rats do not develop spatial disorientation, their working memory remains intact.