Novel strain-level resolution of Crohn's disease mucosa-associated microbiota via an ex vivo combination of microbe culture and metagenomic sequencing

Abstract

The mucosa-associated microbiota is widely recognized as a potential trigger for Crohn's disease pathophysiology but remains largely uncharacterised beyond its taxonomic composition. Unlike stool microbiota, the functional characterisation of these communities using current DNA/RNA sequencing approaches remains constrained by the relatively small microbial density on tissue, and the overwhelming amount of human DNA recovered during sample preparation. Here, we have used a novel ex vivo approach that combines microbe culture from anaerobically preserved tissue with metagenome sequencing (MC-MGS) to reveal patient-specific and strain-level differences among these communities in post-operative Crohn's disease patients. The 16 S rRNA gene amplicon profiles showed these cultures provide a representative and holistic representation of the mucosa-associated microbiota, and MC-MGS produced both high quality metagenome-assembled genomes of recovered novel bacterial lineages. The MC-MGS approach also produced a strain-level resolution of key Enterobacteriacea and their associated virulence factors and revealed that urease activity underpins a key and diverse metabolic guild in these communities, which was confirmed by culture-based studies with axenic cultures. Collectively, these findings using MC-MGS show that the Crohn's disease mucosa-associated microbiota possesses taxonomic and functional attributes that are highly individualistic, borne at least in part by novel bacterial lineages not readily isolated or characterised from stool samples using current sequencing approaches.

Trial registration: ClinicalTrials.gov NCT00989560.

Conflict of interest statement

The authors declare no competing interest.

© 2021. The Author(s).

Figures

Fig. 1. The mucosa-associated microbiota profiles produced from microbial cultures are very similar to those produced using culture-independent methods.

The 16 S rRNA gene amplicon profiles were generated using either DNA extracted directly from biopsy tissue (Biopsy), the same DNA subjected to subtractive enrichment of microbial DNA (Enriched), or microbial cultures produced from matched biopsy samples (MC) using either a mixture of diet-based (M2Diet) or host-based (M2Host) carbohydrates (see Methods for more details). Biopsies were collected from each patient at the anastomotic (ANA) or below the anastomotic site (BA), as well as the rectum (REC). The shaded boxes denote those bacterial taxa present in the datasets at >0.1% relative abundance.

Fig. 2. Principal coordinates analysis (PCoA) of mucosa-associated microbiota profiles based on OTU-level Bray–Curtis dissimilarity shows a patient-specific rather than DNA-based clustering of samples.

The lack of aggregation of the community profiles arising from the microbial cultures further supports that the patient-specific diversity of the mucosa-associated microbiota is retained in these samples.

Fig. 3. The MC-MGS approach produces a high-resolution footprint of the mucosa-associated microbiota compared to direct sequencing approaches.

The taxonomic profiles arising from MetaPhlAn2 analysis of the 3 Gbp MGS datasets produced using either total biopsy DNA (Biopsy), the same DNA subjected to subtractive enrichment of microbial DNA (Enriched), or the microbial cultures produced from matched biopsy samples (MC) using either a mixture of diet-based (M2Diet) or host-based (M2Host) carbohydrates (see Methods for more details). Denoted values are represented as log2-transformed relative abundances.

Fig. 4. Heatmap of KEGG BRITE Hierarchy levels represented as log2-transformed relative abundance counts per million (cpm), stratified according to DNA-type.

Although the microbial enrichment step (Enriched) did increase the amount of functional information from MGS as compared to the Biopsy group, the MC-MGS dataset provided a much more holistic assessment of the CD-MAM functional characteristics.

Fig. 5. Principle component analysis (PCA) plot generated through between-class analysis (BCA) of identified KEGG BRITE hierarchy functional pathways of the MC-MGS dataset (n = 30) and coloured according to patient.

The identified functional pathways were shown to retain the patient-specificity as observed in the MGS taxonomic data (Fig. 3).

Fig. 6. The genome-based phylogenetic tree of all 47 high-quality MAGs constructed based on 120 bacterial marker genes identified using Genome Taxonomy Database Toolkit (GTDB-Tk).

Three of the 47 MAGs (denoted by *) were determined to be novel, and were found to cluster separately from the other 44 MAGs.