Genotype-Associated Molecular Phenotypes Across Brain and Gut Compartments in Male 5xFAD Mice: Shared Latent Structure and the Limits of Individual-Level Inference

Whether gut and brain molecular states covary within individual animals, rather than merely co-occurring at the group level, has rarely been tested in Alzheimer’s disease models. We profiled the hippocampus, cortex, cerebellum, intestine and feces of 18 male 5xFAD (5 familial AD mutations) (B6SJL background) mice and wild-type littermates at 3, 6 and 9 months by label-free proteomics, and feces additionally by shotgun metagenomics. The hippocampal proteome reproduced the amyloid-associated signature (App, Apoe, Hexb, Clu, Gfap), and protein-set analysis showed elevated microglial, lysosomal, astrocytic and endoplasmic reticulum chaperone proteins in the hippocampus and cortex, with no detectable difference in synaptic or mitochondrial sets. Fecal Shannon diversity was reduced, and the bile-acid-modifying genera Romboutsia and Turicibacter were nearly absent in all 5xFAD animals from 3 months (cage nested within genotype). Multi-omics factor analysis recovered a genotype-associated factor shared across brain and gut views (exact permutation p = 0.00125) that explained at most 6.3% of variance in any view and was preprocessing-dependent. After removal of group structure, correlation burden did not exceed a within-cell permutation null, and neither did same-animal positive controls. The 5xFAD genotype thus alters every compartment examined, and a modest latent structure links them; because the design excludes only strong or dense individual-level coupling, the null result bounds design resolution rather than showing that gut and brain are uncoupled.