Spatial transcriptomics identifies disrupted circadian gene expression in a mouse model of Alzheimer’s disease

Abstract Background Disruptions in circadian rhythm are a common symptom of Alzheimer’s disease (AD), which occur early in disease progression and may contribute to the neurodegenerative process. However, the mechanisms that link alterations in rhythmic transcription and disease pathology are poorly understood. Here we used brain‐wide spatial transcriptomics to elucidate progressive disruptions in diurnal transcriptional rhythms in a mouse model of AD. Method Using Visium spatial transcriptomic technology, we investigated transcriptional changes with spatiotemporal resolution in APP23 transgenic mice, which overexpress human APP with the Swedish mutation. Sagittal brain slices from APP23 and littermate control mice were collected at 4 Zeitgeber times (ZT). We used community detection to cluster the spatial locations according to shared patterns of gene expression, which corresponded to major anatomical brain regions. Pseudobulk profiles were computed for each cluster in each sample, and these were used in subsequent differential expression analysis using DESeq2. Rhythmic gene expression was modeled as a sinusoidal function of ZT, while controlling for covariates (sex, genotype, and age) and their interaction with time. Results were further validated using non‐parametric tests for rhythmicity implemented in Metacycle. Result Our analysis revealed a large increase in the number of diurnally rhythmic genes in APP23 mice in several brain structures, including regions affected by neurodegeneration in AD like the hippocampus and the cortex. Additionally, many genes showed increased rhythmicity or altered circadian phase in APP23 mice in specific brain regions, when compared to controls. For example, the expression of Slc5a3, a gene involved in osmotic stress response and deregulated in human AD brains, presented strong circadian rhythms in both APP23 and controls in cortical clusters, but the amplitude of expression was significantly larger in the dentate gyrus only in APP23 mice. Importantly, this effect was observed before and after the onset of amyloid pathology. Conclusion These data map the changes in rhythmic gene expression over the course of AD progression in the spatiotemporal context of cortical and subcortical brain regions. Our findings elucidate molecular disruptions occurring early in the development of AD, which may drive disease progression.