Recent studies have explored the spatial transcriptomics patterns of Alzheimer's disease (AD) brain by spatial sequencing in mouse models, enabling the identification of unique genome-wide transcriptomic features associated with different spatial regions and pathological status. However, the dynamics of gene interactions that occur during amyloid-β accumulation remain largely unknown. In this study, we performed analyses on ligand-receptor communication, transcription factor regulatory network, and spot-specific network to reveal the dependence and the dynamics of gene associations/interactions on spatial regions and pathological status with mouse and human brains. We first used a spatial transcriptomics dataset of the AppNL-G-F knock-in AD and wild-type mouse model. We revealed 17 ligand-receptor pairs with opposite tendencies throughout the amyloid-β accumulation process and showed the specific ligand-receptor interactions across the hippocampus layers at different extents of pathological changes. We then identified nerve function related transcription factors in the hippocampus and entorhinal cortex, as well as genes with different transcriptomic association degrees in AD versus wild-type mice. Finally, another independent spatial transcriptomics dataset from different AD mouse models and human single-nuclei RNA-seq data/AlzData database were used for validation. This is the first study to identify various gene associations throughout amyloid-β accumulation based on spatial transcriptomics, establishing the foundations to reveal advanced and in-depth AD etiology from a novel perspective based on the comprehensive analyses of gene interactions that are spatio-temporal dependent.