Multi-omic analyses identify mucosa bacteria and fecal metabolites associated with weight loss after fecal microbiota transplantation

Abstract

Fecal microbiota transplantation (FMT) has shown promising results in animal models of obesity, while results in human studies are inconsistent. We aimed to determine factors associated with weight loss after FMT in nine obese subjects using serial multi-omics analysis of the fecal and mucosal microbiome. The mucosal microbiome, fecal microbiome, and fecal metabolome showed individual clustering in each subject after FMT. The colonic microbiome in patients showed more marked variance after FMT compared with the duodenal microbiome, characterized by an increased relative abundance of Bacteroides. Subjects who lost weight after FMT sustained enrichment of Bifidobacterium bifidum and Alistipes onderdonkii in the duodenal, colonic mucosal, and fecal microbiome and increased levels of phosphopantothenate biosynthesis and fecal metabolite eicosapentaenoic acid (EPA), compared with those without weight loss. Fecal levels of amino acid metabolism-associated were positively correlated with the fecal abundance of B. bifidum, and fatty acid metabolism-associated metabolites showed positive correlations with A. onderdonkii. We report for the first time the individualized response of fecal and mucosa microbiome to FMT in obese subjects and highlight that FMT is less capable of shaping the small intestine microbiota. These findings contribute to personalized microbe-based therapies for obesity.

Keywords: Fecal microbiota transplantation; fecal metabolome; fecal microbiome; mucosal microbiome; obesity.

Graphical abstract

Figure 1

Study schematic and clinical outcome (A) Longitudinal timeline of clinical information and sample collection from 9 obese subjects treated with FMT (expressed in weeks). (B) Longitudinal body weight change of obese recipients after FMT. The y axis represents the percentage of weight loss at different time points compared with body weight at baseline. “Donor” denotes FMT donor. “FB” denotes obese subject (FMT recipient). “W” indicates the nth week since the start of FMT, where weeks 1–4 represent the 1-month daily FMT period and weeks 5–16 represent time points after FMT.

Figure 2

Mucosal microbiome shifts in obese recipients during the 1-month FMT period (A) Microbiome community shifts in the duodenal and colonic microbiome were viewed by non-metric multidimensional scaling (NMDS) plot based upon Bray-Curtis dissimilarities. (B) Spearman correlation between the genus Bacteroides with longitudinal colonic microbiome shift along MDS1. The MDS1 value was calculated based on Bray-Curtis dissimilarities. ”W” and “D” represent time points since the start of FMT: “W” indicates the nth week since the date of the first FMT; “”D” denotes the nth day within the indicated week. “W1D1” denotes the subject baseline, and the biopsy sample was collected before FMT.

Figure 3

Post-FMT alterations in the fecal microbiome beta and alpha diversity of obese recipients (A) Microbiome community alterations after FMT, viewed by NMDS plot based upon Bray-Curtis dissimilarities. (B and D) Fecal microbiome Chao1 richness in FMT recipients and their corresponding donors at baseline. (C and E) Fecal microbiome Shannon diversity in FMT recipients and their corresponding donors at baseline. Comparisons of the microbiome richness and diversity between donors, pre-FMT, and post-FMT last follow-up were statistically tested by paired Wilcoxon signed rank test, ∗p < 0.05. “OB” denotes obese subject (FMT recipient). “Donor” denotes FMT donor. “baseline” denotes subject baseline, and the samples were collected before antibiotics treatment. ”W” and “D” represent time points after FMT: “W” indicates the nth week since the start of FMT, where weeks 1–4 represent the 1-month daily FMT period and weeks 5–16 represent time points after FMT; “D” denotes the nth day within the indicated week.

Figure 4

Post-FMT alterations in the fecal microbiome composition of obese recipients (A) Alterations in the fecal bacteria composition at the species level in obese recipients after FMT at different time points. Only the most abundant 50 species across all of the subjects were plotted. (B and C) The relative abundance of Bacteriodes vulgatus and Alistipes onderdonkii in the fecal microbiome of recipients at different time points after FMT.

Figure 5

Fecal microbiome functionality alterations in recipients after FMT, in association with post-FMT body weight loss (A) Alterations in the functionality of fecal microbiome during the 1-month FMT period and after FMT at the DNA level. The abundance of metabolic functions (pathways) was profiled via HUMAnN2 based on the metagenomic dataset. (B) Alterations in the functional activity of fecal microbiome during the 1-month FMT period and after FMT at the metatranscriptional (RNA) level. The expression level of metabolic functions (pathways) was profiled via HUMAnDornorN2 based on the metatranscriptomic dataset. Metabolic pathways differentially associated with body weight change were identified via LASSO and plotted in the heatmap. Only those significant pathways included in the LASSO regression model were plotted. (C) Longitudinal expression profile of the metabolic pathway, phosphopantothenate biosynthesis I, in recipient microbiome after FMT at the metatranscriptional level. The expressional contribution to this pathway was stratified by constituent bacteria.

Figure 6

Alterations of the fecal metabolome in recipients after FMT, in association with post-FMT body weight loss (A) Alterations of fecal metabolome after FMT, viewed by NMDS plot based upon Bray-Curtis dissimilarities. (B) The abundance of eicosapentaenoic acid in the fecal microbiome of recipients at different time points after FMT. (C) Positive Spearman correlations between fecal metabolites with Bifidobacterium bifidum, Bacteriodes vulgatus, and Alistipes onderdonkii. (D) Negative Spearman correlations between fecal metabolites with Bifidobacterium bifidum, Bacteriodes vulgatus, and Alistipes onderdonkii. The dots indicate significant correlations. The size and shading of dots indicate the magnitude of the correlation, where darker shades showed higher correlations than lighter ones.