Introduction

Alzheimer's disease (AD) is a progressive neurodegenerative disease characterized by a gradual decline in cognitive function and memory. It is distinguished by the presence of abnormal protein aggregates, including amyloid beta (Aβ) plaques and neurofibrillary tau tangles, which disrupt neuronal function and ultimately lead to cell death^1-3. AD is the leading cause of dementia worldwide, affecting nearly 50 million individuals⁴. Currently, the definitive diagnosis of AD requires post-mortem examination of brain tissue. In living patients, a probable AD diagnosis can be established through cognitive assessments, blood tests for plasma amyloid beta (Aβ), and PET imaging. While these tools combined form an effective diagnostic model, they are not capable of early AD detection⁵. Advancements in neuroimaging techniques capable of detecting deficits prior to significant pathological accumulation and cognitive decline would significantly aid in early detection, diagnosis, and potential therapeutic intervention. Resting state functional magnetic resonance imaging (rs-fMRI) has revealed deficits in functional connectivity in brain regions crucial for memory functions in both AD patients and mouse models. These deficits precede cognitive dysfunction, making rs-fMRI a promising clinical marker of AD^6,7. Resting state functional connectivity (rsFC) assesses spontaneous, coordinated brain activity in the absence of external stimulation. Using fMRI, blood-oxygen-level-dependent (BOLD) signal fluctuations, reflecting neural activity, can be measured in various brain regions. Analyzing correlations in BOLD signals uncovers functional connections between regions⁸. This study aims to characterize alterations in functional connectivity (FC) and investigate the relationship between FC and cognitive decline in AD using the Cg-Tg(APPswe,PSEN1dE9)85Dbo mouse model (an Aβ mouse model of AD). Importantly, awake, unanesthetized imaging was employed to avoid potential confounding effects of anesthesia. To elucidate the mechanisms underlying changes in FC, NanoString technology was used to conduct proteomic analyses of nine distinct brain regions involved in learning and memory processes. These investigations were conducted at three age points: pre-plaque formation (3 months), initiation of plaque formation (6 months), and after extensive plaque formation (10 months) to gain a comprehensive understanding of disease progression.

Methods

Morris Water Maze was used to test spatial memory performance. Mice underwent 2 blocks of 4 training trials per day for 4 days. Latency to escape was recorded. On Day 4, a probe trial assessed quadrant preference. Trajectories were tracked with EthoVision. To acclimate mice for awake imaging, mice underwent a five-day acclimation paradigm to the imaging holder and scanner environment. Acclimation times increased from 10 minutes on day 1 to 50 minutes on day 5. Imaging was performed using a 9.4T Bruker MRI. rs-fMRI assessed connectivity between 66 brain regions. Data was preprocessed in SPM v12, including realignment and warping to a template. CONN Toolbox v19 was used to quantify functional connectivity between all regions of interest. To perform spatial proteomics with Nanostring, animals were first cardiac perfused, brains were sectioned, frozen, and sent for analysis. GeoMx Mouse Protein Assays for NGS provided spatial proteomic data, utilizing oligotagged antibodies and morphology markers for ROI selection. Oligos were released upon UV exposure, then quantified using nCounter. (Figure 1)

Results and Discussion

APP/PS1 and wildtype mice were tested for spatial learning deficits in the Morris Water Maze at 3, 6, and 10 months. Memory deficits emerged at 6 months and persisted at 10 months (Figure 2). Following behavioral tasks, the same mice underwent rs-fMRI to assess connectivity changes, focusing on learning/memory regions and the Default Mode Network (DMN), a network of regions more active during times of rest. The 3-month cohort showed hyperconnectivity in areas including the hippocampus, temporal association areas, and retrosplenial cortex and hypoconnectivity in the ectorhinal cortex and anterior cingulate cortex. The 6-month cohort exhibited hyperconnectivity in the hypothalamus and perirhinal cortex, and hypoconnectivity in the hippocampus, anterior cingulate cortex, and retrosplenial cortex. The 10-month cohort displayed hyperconnectivity in the hippocampus, retrosplenial cortex, and ectorhinal cortex, and hypoconnectivity in the amygdala, insula, and temporal association areas. Overall, hyperconnectivity was observed before plaque accumulation, shifting towards hypoconnectivity as plaques formed. After significant plaque formation at 10 months, hyperconnectivity reemerged (Figure 3A). These connectivity patterns were consistent in the analysis of the DMN (Figure 3B). Proteomic analysis of the 6-month cohort in nine memory-related brain regions revealed distinct protein expression profiles in APP/PS1 mice compared to controls in both AD pathology and immune response panels (Figure 4). Future work includes using machine learning to determine directional relationships from these readouts. For the first time, the relationship between behavior, functional connectivity, and proteomic profiles are being investigated in a mouse model of AD. This may lead to breakthroughs in early detection and clinical diagnostics for patients with AD.