Microbiota in mesenteric adipose tissue from Crohn's disease promote colitis in mice

Abstract

Background: Mesenteric adipose tissue (mAT) hyperplasia, known as creeping fat is a pathologic characteristic of Crohn's disease (CD). The reserve of creeping fat in surgery is associated with poor prognosis of CD patients, but the mechanism remains unknown.

Methods: Mesenteric microbiome, metabolome, and host transcriptome were characterized using a cohort of 48 patients with CD and 16 non-CD controls. Multidimensional data including 16S ribosomal RNA gene sequencing (16S rRNA), host RNA sequencing, and metabolome were integrated to reveal network interaction. Mesenteric resident bacteria were isolated from mAT and functionally investigated both in the dextran sulfate sodium (DSS) model and in the Il10 gene-deficient (Il10-/-) mouse colitis model to validate their pro-inflammatory roles.

Results: Mesenteric microbiota contributed to aberrant metabolites production and transcripts in mATs from patients with CD. The presence of mAT resident microbiota was associated with the development of CD. Achromobacter pulmonis (A. pulmonis) isolated from CD mAT could translocate to mAT and exacerbate both DSS-induced and Il10 gene-deficient (Il10-/-) spontaneous colitis in mice. The levels of A. pulmonis in both mAT and mucous layer from CD patients were higher compared to those from the non-CD group.

Conclusions: This study suggests that the mesenteric microbiota from patients with CD sculpt a detrimental microenvironment and promote intestinal inflammation. Video abstract.

Keywords: Bacterial translocation; Crohn’s disease; Mesenteric adipose tissue; Microbiota.

Conflict of interest statement

The authors declare no competing interests.

© 2021. The Author(s).

Figures

Fig. 1

Multi-omics signatures of mesenteric adipose tissue in CD versus non-CD. a Separation of mesenteric microbiome between patients with 48 CD and 16 non-CD controls, revealed by principal component analysis (PCA) (Adonis p = 0.009). b LDA score computed from features differentially abundant between CD patients and non-CD controls. The criteria for feature selection is log LDA score > 3. c Separation of the mesenteric transcriptome between 46 CD patients and 15 non-CD controls, revealed by PCA (Adonis p = 0.001). d Heatmap of the immune-associated DEGs (FDR p < 0.05) from 46 patients with CD and 15 non-CD controls. Ordering by diagnosis, clustering within diagnosis. e Separation of mesenteric metabolome between 48 CD patients and 16 non-CD controls, revealed by PCA (Adonis p = 0.003). f Twenty-eight metabolites were significantly (Studentʼs t-test, p < 0.05) enriched in mAT from 48 CD, while 19 metabolites were significantly enriched in 16 non-CD controls. The violin plot indicated the difference of mean and 95% confidence interval for an individual metabolite. The statistical significance values are denoted as *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001

Fig. 2

Critical role of isolated bacterial biomarkers in CD inflammation. a The top 16 mAT bacterial biomarkers were identified by applying Random Forests regression of their relative abundance. Biomarker taxa are ranked in decreasing order of importance to the accuracy of the model. The model based on these biomarkers achieved an AUC of 0.987. b ROC analysis of these mAT bacterial biomarkers in the RF model achieved an AUC of 0.852 to classify patients with endoscopic recurrence. c Circularized plot showing the abundance of mesenteric resident bacteria isolated from different culture mediums. d Significant association of host-microbiome interactions: isolated species, immune transcripts, and differential metabolites. Network shows the significant correlations (Spearman’s correlation, p < 0.05) between two omics variables. Nodes are colored by different omics variables and sized by the number of connections. e Kaplan–Meier (KM) survival curves classify patients into high- and low-risk groups based on the major pathogens in the CD cohort

Fig. 3

Mesenteric resident bacteria exacerbate colitis in mice. a SPF C57BL/6 mice were treated by an antibiotic cocktail for 4 days. One day post antibiotics, the mice were daily orally colonized with bacterial cocktail (5-Mix or E. fergusonii, 109 CFU/mouse/dose) or BHI until being euthanized. After 5 days bacterial colonization, the mice were exposed to 3% DSS for 7 days, followed by regular water for 3 days. b and c Changes of body weight (b) and disease activity index (DAI) after administration of 3% DSS (c). d–g Representative colons (d), colon length (e), representative colonic histological images (Scale bar = 40 μm and 10 μm) (f), colonic histological score (g) in mice treated with 5-mix bacterial cocktail, E. fergusonii and BHI. h Messenger RNA levels of key cytokines (TNF-α, IL-6, and IL-1β) in the terminal ileum (n=5 per group). i SPF Il10−/− mice were treated with antibiotics cocktail for 4 days. One day post antibiotics, the mice were daily orally colonized with bacterial cocktail (5-Mix or E. fergusonii, 109 CFU/mouse/dose) or BHI for 3 weeks. j–m Representative colons (j), colon length (k), representative colonic histological images (Scale bar= 40 μm and 20 μm) (l), colonic histological score from Il10−/− mice treated with 5-Mix, E. fergusonii and BHI (m). n Inflammatory cytokines from macrophage cell line RAW264.7 after co-culture with 5 different bacteria or 5-Mix. Results are shown as the mean ± SEM. Each dot indicates an individual mouse. The statistical significance values are denoted as: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. One-way ANOVA following Tukey’s multiple comparison test (e, g, h, k, m, and n); two-way ANOVA following Tukey’s multiple comparison test (b and c). A.p, A. pulmonis; O.a, O.anthropi; P.a, P.alcaliphila; A.d, A.deleyi; D.r, D.riboflavina

Fig. 4

A. pulmonis is sufficient to elicit a strong inflammatory response in mice. a SPF C57BL/6 mice were treated with an antibiotic cocktail for 4 days. One day post antibiotics, the mice were daily orally colonized with bacterial cocktail (A. pulmonis, 5-Mix or residual 4-Mix, 109 CFU/mouse/dose) or BHI until being euthanized. After 5 days bacterial colonization, the mice were exposed to 3% DSS for 7 days, followed by regular water for 3 days. b and c Changes of body weight (b) and disease activity index (DAI) (c) after 3% DSS administration. d–g Colon length (d), representative colons (e), representative colonic histological images (Scale bar = 40 μm and 10 μm) (f), colonic histological score (g) in mice treated with A. pulmonis, 5-mix bacterial cocktail, 4-mix bacterial cocktail and BHI. h SPF Il10−/− mice were treated with antibiotics cocktail for 4 days. One day post antibiotics, the mice were daily orally colonized with bacterial cocktail (A. pulmonis or E. fergusonii, 109 CFU/mouse/dose) or BHI for 3 weeks. i-l Colon length (i), representative colons (j), representative colonic histological images (Scale bar = 40 μm and 20 μm) (k), colonic histological score (l) from Il10−/− mice treated with A. pulmonis, E. fergusonii and BHI. Results are shown as the mean± SEM. Each dot indicates an individual mouse. The statistical significance values are denoted as: *p < 0.05, **p < 0.01, *** p < 0.001, ****p < 0.0001. One-way ANOVA following Tukey’s multiple comparison test (d, g, j, and l); two-way ANOVA following Tukey’s multiple comparison test (b and c)

Fig. 5

Bacterial translocation into mesenteric adipose tissue is associated with development of colitis. a Representative images of mucous depth in the colonic section stained by alcian blue (Scale bar = 40 μm and 10 μm) from mice colonized with bacterial cocktail (A. pulmonis, 5-Mix or residual 4-Mix, 109 CFU/mouse/dose) or BHI in DSS-colitis model. b Mucous depth from mice treated with A. pulmonis, 5-Mix, 4-Mix, or BHI in the DSS-colitis model. c The quantity of A. pulmonis in mesenteric tissue from mice was assessed by qPCR in the DSS-colitis model. Results are shown as the mean± SEM. d Detection of A. pulmonis in the human mesenteric tissue (mAT) from CD patients and non-CD controls by fluorescence in situ hybridization (FISH). Mesenteric tissue sections were stained with DAPI (blue), A. pulmonis-specific probes (green dot) and EUB338 (red dot). (Scale bar = 10 μm). e The quantity of A. pulmonis in human mAT from CD and non-CD controls was assessed by qPCR. f The quantity of A. pulmonis in human mAT from CD patients with and without endoscopic recurrence was assessed by qPCR. g and h The quantity of A. pulmonis in human mAT (g) or mucous layer (h) from a new validation cohort. Each dot indicates an individual mouse. The statistical significance values are denoted as *p < 0.05, **p < 0.01, ***p < 0.001. One-way ANOVA following Tukey’s multiple comparison test (b and c); two-tailed Student’s t-test (e–h)