Recent studies have suggested a transient glucose hypermetabolism in early phases of Alzheimer’s Disease (AD), which is followed by a characteristic glucose hypometabolism in dementia stages. This phenomenon desveres further investigation and it is suggested to be associated to glial/inflammatory or compensatory neuronal responses. Here, we aimed to longitudinally investigate brain glucose metabolism in an AD animal model and explore associated cellular and inflammatory changes. Longitudinal assessments, including cerebral glucose metabolism ( 18 F]FDG-PET), behavioral tasks and cerebrospinal fluid (CSF) sample collection were performed at 3, 6, 9, and 12 months of age (mo) in wild type (WT) and TgF344-AD rats (Tg). Glial and inflammatory markers were evaluated in the CSF via ELISA multiplex. A cross-sectional cohort was used to follow the spatial distribution of Aβ plaques (IHC), to assess the brain content of glial and neuronal proteins (western blot), and to analyze the cortical glutamate uptake (ex-vivo slices) at the same time points. At 9mo, three months after the initial appearance of Aβ plaque deposits, Tg animals exhibited cortical glucose hypermetabolism ( Figure 1A-C ). At the same age, astrocytic glutamate uptake was increased in the cortex and hippocampus ( Figure 2 ). Declines in performance on the Y-maze and Novel Object Recognition tasks were observed at 9 and 12mo ( Figure 1D ). CSF analysis revealed elevated GFAP levels at 6, 9 and 12mo, and reduced S100B at 9mo ( Figure 1E e F ). Tissue GFAP immunocontent increased in the temporoparietal cortex at 6 and 9mo, in the hippocampus at 9mo, and was reduced in the frontal and temporoparietal cortices at 12mo ( Figure 3 ). Our findings suggest the presence of an early, transient phase of brain glucose hypermetabolism in TgF344-AD rats, consistent with observations in recent animal and human studies. This phenomenon seems to be closely linked to the astrocyte response, reflected in variations of crucial astrocyte proteins such as GFAP and S100B, along with an increase in the glutamate uptake by these cells. In contrast, neuronal, microglial and inflammatory markers did not exhibit changes during this timeframe.

The Model Organism Development and Evaluation for Late-Onset Alzheimer’s Disease (MODEL-AD) Consortium has been established to develop the next generation of Alzheimer’s disease (AD) models based on human genomic findings. As new models are developed, phenotypic data are compared to established models, including the 5xFAD mouse model. We undertook measurements of diffusion metrics to characterize the temporal alterations in brain structure in male and female 5xFAD mice, including connectivity analysis between the hippocampus and infralimbic prefrontal cortex (PFC). 5xFAD mice were compared to age-matched littermates (C57BL/6J, WT) at 4, 8, and 12 months (mo). Mice underwent high resolution diffusion tensor imaging (DTI) magnetic resonance imaging (MRI) at 9.4T (5 B0, 30 directions b=3000 mm 2 /sec) to assess regional white and gray matter. Regional tissue features were extracted from fractional anisotropy (FA), radial (RD), axial (AxD) and mean diffusivity (MD) parametric maps. Tractography was performed based on the AMBMC mouse atlas. 18F-AV45 PET was performed in a separate cohort at 4, 6, and 12mo to quantitate amyloid β (Aβ) load. Elevated FA within hippocampal CA1 was first observed in 5xFAD males at 8mo, but at 12mo 5xFAD females had increased FA in bilateral CA1 compared to WT mice. Connectivity between CA1 and PFC found 12mo 5xFAD females had significantly increased RD along the tract with concomitant increases in MD and AxD. AV45-PET uptake was increased at 6mo (∼20%) and remained elevated at 12mo (∼25%) in the PFC in male and female 5xFAD compared to WT. In the hippocampus there was a ∼20% increase in AV45 binding at 6mo, that then declines by 12mo in 5xFAD mice. Phenotyping of mouse models using DTI identifies altered brain connectivity and regional tissue modifications that may be indicative of increasing Aβ deposition. These data support that increasing Aβ deposition results in altered DTI metrics and connectivity using clinically relevant imaging modalities.

The development and characterization of new mouse models late-onset Alzheimer’s Disease (LOAD) are critical for understanding the progression of the pathology and as a platform for theevaluation of new drugs. The UCI MODEL-AD project focuses on the development of LOAD mouse models based on human GWAS data. We evaluate each new modeldeveloped in our group by performing a comparative network analysis of hippocampal transcriptomes in order to relate changes in expression to other phenotypic changes. As part of our standard pipeline, we test each new GWAS variant model at the ages of 4 and 12 months crossed with 5XFAD mice to regular 5XFAD mice and GWAS variant-only mice. Differential gene expression analysis allows us to quantify the impact of the ABI3, ABCA7, PICALM and TREM2 variants on the pathology. Each model that we have analyzed have both shared as well as unique gene modules. Several variants showed less transcriptional differences compared to wild type than 5xFAD at 4 months but increase inflammation at 12 months. RNA transcriptome analysis combined with phenotypic characterization is essential to the characterization and identification of the most appropriate LOAD mouse model.

Mouse models of human diseases such as Alzheimer disease (AD) are one of the most important tools for studying pathogenic mechanisms and testing interventions and therapeutics. While much of the characterization of gene expression changes to date has been done with “bulk” RNA-seq in tissues such as hippocampus and cortex, the advent of single-cell and single nucleus functional genomics assays allow us to analyze those changes at the level of individual cell subtypes. However, we would like to compare the relatively sparse and noisy single-cell assays to bulk RNA-seq to deconvolve which cells are driving the changes observed in the previously collected bulk datasets. Here we develop a python package called PyWGCNA that is designed to identify and compare gene expression modules in bulk RNA-seq data as well as compare them to scRNA-seq data and snRNA-seq clusters. As a proof of concept, we apply this to bulk RNA-seq of the cortex and hippocampus from 5xFAD as well as matching control mice from MODEL-AD to identify gene modules associated with the genotype and phenotypes. Comparisons of these modules to single-nuclei reveal the specific cell-types driving specific modules in a region-specific manner. This deconvolution approach allows us to interrogate existing bulk datasets using single-cell data to look for the signature of cell-type specific changes in gene expression.

Transcriptomic studies of Alzheimer's disease (AD) have identified both tissue-level and cell-type specific gene expression changes. However, with both "bulk"-tissue and single-cell approaches, we lose pertinent spatial information, such as cell-to-cell proximity or proximity to pathological features. Recently several techniques have emerged, aiming to profile gene expression while preserving the spatial architecture, and we have leveraged one of these techniques—spatial transcriptomics—to interrogate AD gene expression changes in the 5XFAD, an amyloid mouse model, in a spatial and temporal manner. We generated spatial transcriptomic (10x Genomics Visium) data from 5XFAD and wildtype mice at the ages of 4, 6, 8, and 12 months (n = 80 total samples, sex-balanced). Prior to generating libraries, we stained our tissue sections with Amylo-glo and the conformation-specific antibody OC to analyze gene expression changes in spatial relativity to amyloid pathology. We profiled 18,000-20,000 total genes per sample with 1,800-2,700 genes per spatial spot and identified brain region-specific transcriptionally distinct clusters. We also examined the spatial distribution of AD risk genes, identified by GWAS, throughout the disease progression. Further, we identified gene expression changes spatially related to amyloid pathology localization. We have characterized the 5XFAD transcriptome, identifying spatiotemporal AD gene expression changes and providing further insight into the role of amyloid pathology in the modulation of gene expression in AD.