An integrative multi-omic analysis defines gut microbiota, mycobiota, and metabolic fingerprints in ulcerative colitis patients

Abstract

Background: Ulcerative colitis (UC) is a multifactorial chronic inflammatory bowel disease (IBD) that affects the large intestine with superficial mucosal inflammation. A dysbiotic gut microbial profile has been associated with UC. Our study aimed to characterize the UC gut bacterial, fungal, and metabolic fingerprints by omic approaches.

Methods: The 16S rRNA- and ITS2-based metataxonomics and gas chromatography-mass spectrometry/solid phase microextraction (GC-MS/SPME) metabolomic analysis were performed on stool samples of 53 UC patients and 37 healthy subjects (CTRL). Univariate and multivariate approaches were applied to separated and integrated omic data, to define microbiota, mycobiota, and metabolic signatures in UC. The interaction between gut bacteria and fungi was investigated by network analysis.

Results: In the UC cohort, we reported the increase of Streptococcus, Bifidobacterium, Enterobacteriaceae, TM7-3, Granulicatella, Peptostreptococcus, Lactobacillus, Veillonella, Enterococcus, Peptoniphilus, Gemellaceae, and phenylethyl alcohol; and we also reported the decrease of Akkermansia; Ruminococcaceae; Ruminococcus; Gemmiger; Methanobrevibacter; Oscillospira; Coprococus; Christensenellaceae; Clavispora; Vishniacozyma; Quambalaria; hexadecane; cyclopentadecane; 5-hepten-2-ol, 6 methyl; 3-carene; caryophyllene; p-Cresol; 2-butenal; indole, 3-methyl-; 6-methyl-3,5-heptadiene-2-one; 5-octadecene; and 5-hepten-2-one, 6 methyl. The integration of the multi-omic data confirmed the presence of a distinctive bacterial, fungal, and metabolic fingerprint in UC gut microbiota. Moreover, the network analysis highlighted bacterial and fungal synergistic and/or divergent interkingdom interactions.

Conclusion: In this study, we identified intestinal bacterial, fungal, and metabolic UC-associated biomarkers. Furthermore, evidence on the relationships between bacterial and fungal ecosystems provides a comprehensive perspective on intestinal dysbiosis and ecological interactions between microorganisms in the framework of UC.

Keywords: dysbiosis; gut metabolism; gut microbiota; inflammatory bowel disease; intestinal biomarkers; multi-omic integrated approaches; ulcerative colitis.

Figure 1

Compositional analysis at the genus level of UC and CTRL gut microbiota (left panel) and mycobiota (right panel). Unsupervised multivariate analysis [principal component analysis (PCA) plot] (A, E); supervised multivariate analysis plot [partial least squares-discriminant analysis (PLS-DA)] (B, F) and loading variables plot (filtered for VIP > 1 and for fungi, for loading coefficient > 0.1) (C, G). Bacterial PLS-DA is characterized by root mean square error (RMSE) = 0.336, R ² value = 0.544, and Q ² = 0.418. Fungal PLS-DA is characterized by RMSE = 0.226, R ² = 0.761, and Q ² value = 0.168. LDA plots on LEfSe univariate analysis (D, H). Bacterial taxa enriched in UC patients have negative LDA scores (orange), while bacterial and fungal taxa enriched in CTRL have positive scores (blue).

Figure 2

Bacterial and fungal interkingdom correlation network in UC (A) and CTRL (B). Each node represents bacteria (orange circles) and fungi (blue circles). Green and red edges indicate positive and negative correlation values, respectively. Only correlations statistically significant (p-value < 0.05) are reported.

Figure 3

Multivariate and univariate analyses on metabolic profiles of UC and CTRL. The biplot shows the first 24 loadings predicted by PCA analysis (A). The second biplot shows the sample clustering calculated with PLS-DA analysis [blue dots (CTRL) and orange triangles (UC)] (B). The barplot describes the value of loadings in each group, which are calculated by PLS-DA analysis and filtered for loading coefficient >0.1 and VIP value >1 (C). Root mean square error (RMSE) = 0.329, R ² = 0.526, and Q ² value = 0.382. Univariate plot based on log2 fold change values (D). The Mann–Whitney test confirms that phenylethyl alcohol is increased in the UC group.

Figure 4

Integrated multi-omic analyses confirm the presence of a typical shape and function of UC gut microbiota. Multiblock principal component analysis (MBPCA) plot (A), loadings plot (filtered for loading coefficient > 0.1) (B), and multiblock partial least squares-discriminant analysis (MBPLS-DA) plot (C). Root mean square error (RMSE) = 0.128, R ² = 0.935, and Q ² value = 0.567. VIP values are reported on the horizontal axis (D). ROC analysis of the MBPLS-DA model (E). The value of AUROC = 0.9947 indicates a high accuracy of the prediction model.

Figure 5

Bacterial, fungal, and metabolic markers in UC and CTRLs. Biplots show the result of ComDim analysis. Teal circles represent the UC patients and red circles represent CTRL subjects (A). Bacterial, fungal, and metabolic markers are labeled in red, green, and blue, respectively (B).