Host genetics shapes crypt niche colonisation by keystone gut bacteria, influencing metabolic health

A “polygenic-first” design for causal inference was adopted to study the caecal microbiome of HS rats, combining shallow-shotgun metagenomics with linear-mixed models to map host polygenic effects on multi-level microbiome traits. A reproducible, shallow shotgun-optimised pipeline yielded better results for detecting such effects than 16S, with heritable signals concentrating in Bacteroides and Prevotella, their shared guild, and β-diversity. Then, genome-resolved enrichments suggested potential adaptations to the crypt niche, leading to a hypothesis-driven selection of host genes. A cross-validated set of sentinel SNPs for these candidate genes explained genus-level polygenic effects and served as instruments in mediation and Mendelian randomisation analyses, contributing to a structural equation model that linked keystone bacteria, community structure, and host metabolism. The effects of these genera on fasting glucose levels had opposite signs and were balanced by indirect effects routed through community traits, suggesting that genotype-aware, niche-engineering interventions could better modulate glycaemia and other health-related traits.