Fig. 4

Validation studies in primary hepatocytes and FMT mice. a Scheme of the experimental design for study 1. Mice were fed for 9 weeks diets containing low- (LI), low-normal- (LNI), high-normal- (HNI), moderately high- (MHI) and high- (HI) iron doses. b Heatmap displaying genus relative abundances for each mouse. c Principal coordinate analysis (PCoA) depicting dissimilarities between groups based on unifrac distance metrics. d Scheme of the experimental design for study 2. Mice were fed either a high fat diet (HFD) or a no-HFD diet containing four different iron doses (LI, LNI, HNI, MHI) for 10 weeks. e Variations in the Shannon diversity index, f Chao1 richness estimator and g observed species of mice fed either a HFD or a no-HFD with different iron doses (LI, LNI, HNI, MHI). h PCoA based on Canberra distance metric for the no-HFD-fed mice and i the HFD-fed mice with different iron doses. Differences in microbial composition between iron doses for each diet were assessed by PERMANOVA using 999 permutations. j, k Permutation tests for the O-PLS models between iron dose and bacterial families or genera in HFD-fed mice, respectively. l Significant families and m genera identified from O-PLS regression loadings to be associated with iron dose. n Scheme of the experimental design for study 3. Low-ferritin (n = 3) and high-ferritin (n = 3) microbiota human donors were selected and for each donor their faecal samples were transplanted n = 6–8 mice after antibiotic treatment. After 14 days following colonization gavage mice were sacrificed and iron and liver fat accumulation-related genes (n = 22) were measured by PCR. o Permutation test for the O-PLS-DA model between mice genes and the human donor group (low- or high- ferritin). p Significant mouse genes associated with donor group from O-PLS-DA regression loadings. q Ferroportin (Slc40a1) and r Tfrc expression according to the donor ferritin concentration