Disease Severity and Immune Activity Relate to Distinct Interkingdom Gut Microbiome States in Ethnically Distinct Ulcerative Colitis Patients

Abstract

Significant gut microbiota heterogeneity exists among ulcerative colitis (UC) patients, though the clinical implications of this variance are unknown. We hypothesized that ethnically distinct UC patients exhibit discrete gut microbiotas with unique metabolic programming that differentially influence immune activity and clinical status. Using parallel 16S rRNA and internal transcribed spacer 2 sequencing of fecal samples (UC, 30; healthy, 13), we corroborated previous observations of UC-associated bacterial diversity depletion and demonstrated significant Saccharomycetales expansion as characteristic of UC gut dysbiosis. Furthermore, we identified four distinct microbial community states (MCSs) within our cohort, confirmed their existence in an independent UC cohort, and demonstrated their coassociation with both patient ethnicity and disease severity. Each MCS was uniquely enriched for specific amino acid, carbohydrate, and lipid metabolism pathways and exhibited significant luminal enrichment of the metabolic products of these pathways. Using a novel ex vivo human dendritic cell and T-cell coculture assay, we showed that exposure to fecal water from UC patients caused significant Th2 skewing in CD4(+) T-cell populations compared to that of healthy participants. In addition, fecal water from patients in whom their MCS was associated with the highest level of disease severity induced the most dramatic Th2 skewing. Combined with future investigations, these observations could lead to the identification of highly resolved UC subsets based on defined microbial gradients or discrete microbial features that may be exploited for the development of novel, more effective therapies.

Importance: Despite years of research, the etiology of UC remains enigmatic. Diagnosis is difficult and the patient population heterogeneous, which represents a significant barrier to the development of more effective, tailored therapy. In this study, we demonstrate the clinical utility of the gut microbiome in stratifying UC patients by identifying the existence of four distinct interkingdom pathogenic microbiotas within the UC patient population that are compositionally and metabolically distinct, covary with clinical markers of disease severity, and drive discrete CD4(+) T-cell expansions ex vivo These findings offer new insight into the potential value of the gut microbiome as a tool for subdividing UC patients, opening avenues to the development of more personalized treatment plans and targeted therapies.

FIG 1

Comparison of healthy (n = 13) and UC-associated (n = 30) fecal microbiotas. (a) Bacterial α diversity. Horizontal bars represent means ± standard deviations. P values were obtained by two-tailed Student t test. (b) Bacterial community composition represented by nonmetric multidimensional scaling (NMDS) of pairwise weighted UniFrac distances. Compositional bar plots of the mean relative abundance of bacterial families (c) and fungal genera (d) are shown. (e) Bacterial community composition of UC patients stratified by ethnicity (18 EU UC, 12 SA UC) represented by NMDS of pairwise weighted UniFrac distances. In panels b and e, each dashed ellipse represents the 95% confidence interval for the centroid of each stratification group as calculated by ordiellipse.

FIG 2

Comparison of fecal bacterial communities among all UC patients (n = 30). (a) Hierarchical cluster analysis using pairwise weighted UniFrac distances. Approximately unbiased P values (red) computed by multiscale bootstrap resampling. EU UC, squares; SA UC, circles. (b) Compositional plots of bacterial family relative abundance for each UC patient.

FIG 3

Clinical measurements of UC severity among UC MCSs (11 for MCS1, 8 for MCS2, 4 for MCS3, 3 for MCS4). (a) Simple clinical colitis activity. (b) Number of extracolonic symptoms. (c) Number of family members diagnosed with IBD. (d) Duration of disease. All pairwise comparisons were done with a two-tailed Dunn test. Only P values of <0.1 are indicated. EU UC, squares; SA UC, circles.

FIG 4

Heat map of bacterial OTUs significantly enriched across UC MCSs. The OTUs shown were identified by Kruskal-Wallis test comparing distributions among UC MCSs (P value, <0.05; q value, <0.08). Column order is consistent with Fig. 2. Rows are ordered on the basis of phylogenetic relatedness. For visualization, read counts were normalized [log₂(x + 1)] and scaled by row.

FIG 5

Fecal metabolites significantly differentially enriched among UC patients classified as MCS1, MCS2, MCS3, or MCS4 on the basis of pairwise Welch t tests (P value, <0.05).

FIG 6

In vitro human T-cell activity following coculture with autologous DCs coincubated with sterile fecal water. Panels: a, Th1-to-Th2 ratio; b, Th1 frequency; c, Th2 frequency; d, Th17 frequency; e, regulatory T-cell frequency (48 healthy, 116 UC). Comparisons of the Th1 frequencies (f), Th2 frequencies (g), and Th1-to-Th2 ratios (h) of healthy and UC MCSs are shown (48 for healthy, 48 for MCS1, 40 for MCS2, 16 for MCS3, and 8 for MCS4). Concentrations of IL-4 (i), IL-5 (j), and IL-13 (k) in cell supernatant following coculture of human T cells with autologous DCs challenged with sterilized fecal water from healthy participants and MCS1 and MCS2 patients are shown (48 for healthy participants, 48 for MCS1 patients, and 40 for MCS2 patients). Data were generated from four (a to h) or two (i to k) replicate experiments with DCs/T cells obtained from two anonymous PBMC donors. Horizontal bars (mean fitted values for each group) and P values were determined by linear mixed-effect modeling (see Materials and Methods). P values of <0.1 are indicated.