Microbiome multi-omics analysis reveals novel biomarkers and mechanisms linked with CD etiopathology

Abstract

Background: The gut microbiome plays a key role in the development of inflammatory bowel disease (IBD), as imbalances in microbial composition are associated with immune dysfunction. However, the specific mechanisms by which certain microorganisms contribute to this process remain unclear.

Methods: Here, we employed a multi-omics approach on fecal samples to identify novel microbiome markers and elucidate mechanisms underlying IBD. Shotgun metagenomics was applied to 212 samples (850 in total with validation cohort), shotgun metatranscriptomics to 103 samples and metabolomics to 105 samples. Machine learning techniques were used to predict disease and the three omics data were integrated to propose a mechanistic role of the microbiota.

Results: Metagenomic analysis identified Crohn's disease (CD)-specific microbiome signatures, including a panel of 20 species that achieved a high diagnostic performance, with an area under the ROC curve (AUC) of 0.94 in an external validation set. Metatranscriptomic analysis revealed significant alterations in microbial fermentation pathways in CD, but not in ulcerative colitis (UC), highlighting disruptions that explain the depletion of butyrate-a key anti-inflammatory metabolite-observed in metabolomics analysis. Integrative multi-omics analyses further identified active virulence factor genes in CD, predominantly originating from the adherent-invasive Escherichia coli (AIEC). Notably, these findings unveiled novel mechanisms, including E. coli-mediated aspartate depletion and the utilization of propionate, which drives the expression of the ompA virulence gene, critical for bacterial adherence and invasion of the host's macrophages. Interestingly, these microbiome alterations were absent in UC, underscoring distinct mechanisms of disease development between the two IBD subtypes.

Conclusions: In conclusion, our study not only identifies promising novel biomarkers with strong diagnostic potential, which could be valuable in challenging clinical scenarios, but also offers an integrated multi-omics perspective on the microbial mechanisms underlying inflammation and virulence in Crohn's disease.

Keywords: Adherent-invasive Escherichia coli; Inflammatory bowel disease; Microbiome; Multi-omics; Virulence gene.

Fig. 1

Microbial signatures for CD. Heatmap of the log2-transformed normalized abundance of the top 10 enriched and depleted species in CD (n = 68) patients compared to healthy controls (n = 67) (A). Performance (ROCAUC) of the random forest classifier built using the 174 DA species found (B) and a panel comprising the top 10 enriched and depleted species in CD (C). Differentially abundant species were obtained using ANCOM-BC

Fig. 2

Pathway shifts in CD patients compared to HC. Differentially expressed pathways in the microbiome of CD patients (n = 27) compared to healthy controls (n = 49) using linear models (A). Pathways were grouped according to their level 1 MetaCyc classes (broader functionality) and colored by their level 2 MetaCyc classes (more specific). CD-depleted pathways are represented with negative log2FC, while CD-enriched pathways are positive. Differentially expressed fermentation between CD patients and healthy controls (B). In HC, fermentation pathways (blue) lead to the production of butyrate, while in CD (red), they lead to the production of propionate and 1,3-propanediol. Differentially expressed pathways were obtained using mixed-effects linear models

Fig. 3

Differentially abundant metabolites in feces of IBD patients and HC. Enriched and depleted metabolites in stool samples of the participants, categorized by disease condition (CD n = 28, HC n = 51, UC n = 26) (A). S-plot of the OPLS-DA model between CD patients (n = 28) and healthy controls (n = 51) (sevenfold CV, 100 permutations, R2Y(cum) = 0.39, Q2Y(cum) = 0.19, pR2Y = 0.01, pQ2Y = 0.01) using metabolite concentration (B). The line in the boxplot depicts the median, the hinges depict the first and third quartile, and the whiskers depict values equal to the first quartile + 1.5 × IQR and the third quartile—1.5 × IQR. All individual points and p-values are shown. The S-plot represents both the covariance and correlation between the metabolites and the modeled class of the OPLS-DA model

Fig. 4

Study of virulence factor genes. Comparison of virulence factor (VF) genes detected at DNA level (CD n = 68, HC n = 67, UC n = 77) (A) and actively transcribed at RNA level (CD n = 27, HC n = 49, UC n = 27) (B) depending on disease condition. Representation of CD-enriched actively transcribed virulence factor genes compared to healthy controls (C). VF genes were grouped by microorganisms of provenance and coloured by their function, according to the Virulence Factor Database (VFDB). The line in the boxplot depicts the median, the hinges depict the first and third quartile, and the whiskers depict values equal to the first quartile + 1.5 × IQR and the third quartile—1.5 × IQR. All individual points and p-values are shown

Fig. 5

Correlation between metabolites, pathways and virulence factor genes. Spearman’s correlation between ompA and propionate (A) in CD, between P108-PWY and propionate in CD (B), and between P108-PWY and ompA in both CD (C) and healthy controls (D) (CD n = 25, HC n = 23, UC n = 22). All individual points, p-values, and correlation coefficients are shown

Fig. 6

Bacterial dysbiosis drives profound changes in the gut environment of CD patients. These changes include the displacement of commensal bacteria, predominantly F. prausnitzii in healthy individuals, by dysbiotic bacterial communities dominated by F. umbilicata in CD patients. Additionally, microbial metabolic pathway activity is altered, leading to shifts in metabolite concentrations within the GI tract of affected individuals. Notably, integrative analysis of metagenomics, metatranscriptomics, and metabolomics revealed that E. coli utilizes aspartate to promote its growth and consumes propionate to express virulence factor genes such as ompA, which is involved in host adherence and invasion