Stratification of human gut microbiomes by succinotype is associated with inflammatory bowel disease status

Abstract

Background: The human gut microbiome produces and consumes a variety of compounds that interact with the host and impact health. Succinate is of particular interest as it intersects with both host and microbiome metabolism. However, which gut bacteria are most responsible for the consumption of intestinal succinate is poorly understood.

Results: We build upon an enrichment-based whole fecal sample culturing approach and identify two main bacterial taxa that are responsible for succinate consumption in the human intestinal microbiome, Phascolarctobacterium and Dialister. These two taxa have the hallmark of a functional guild and are strongly mutual exclusive across 21,459 fecal samples in 94 cohorts and can thus be used to assign a robust "succinotype" to an individual. We show that they differ with respect to their rate of succinate consumption in vitro and that this is associated with higher concentrations of fecal succinate. Finally, individuals suffering from inflammatory bowel disease (IBD) are more likely to have the Dialister succinotype compared to healthy subjects.

Conclusions: We identified that only two bacterial genera are the key succinate consumers in human gut microbiome, despite the fact that many more intestinal bacteria encode for the succinate pathway. This highlights the importance of phenotypic assays in functionally profiling intestinal microbiota. A stratification based on "succinotype" is to our knowledge the first function-based classification of human intestinal microbiota. The association of succinotype with IBD thus builds a bridge between microbiome function and IBD pathophysiology related to succinate homeostasis. Video Abstract.

Fig. 1

Fecal microbiomes consume succinate at different rates. a We performed enrichment cultures of whole fecal microbiota from 13 different donors in a basal medium supplemented with 30 mM succinate as the primary carbon source. The triplicate cultures were sampled after 2 and 7 days, respectively. b The supplied succinate is consumed differently across fecal microbiota and replicates. We classify a culture as fast (red), intermediate (blue), or slow (orange) consumer, or non-consumer (green) of succinate. Each circle is a measurement of the succinate concentration in a replicate enrichment. The lines connect multiple time points from the same replicate. c The increase in optical density (OD) at day 2 was highest for the fast consuming cultures. Each thin line is a replicate enrichment from (b) and the thick lines are the average in each class. d The consumed succinate was converted to propionate at a molar ratio of 1:1. Each point corresponds to a point in (b). e Genus-level composition of the enrichment cultures after either 2 or 7 days. Only genera with at least 5% relative abundance are shown. f The “enrichment” is the difference in relative abundance between enrichments with supplemented succinate (SU+) and the corresponding enrichment without supplemented succinate (SU−) of the same donor. Only genera with positive enrichment values are shown. g Coefficients from a linear regression of the enrichment value on the rate category for each genus. Only genera with positive estimates are shown. Filled boxes are estimates with $p<0.05$

Fig. 2

Succinate-consuming gut bacteria differ in their succinate uptake rate. a To estimate the rate at which different representative isolates consume succinate, we performed replicate cultures of a panel of eleven isolates in media supplemented with 80 mM of succinate as the primary carbon source. Each replicate culture was destructively sampled at a different time point between 0 and 60 h. We then estimated the succinate uptake rate by fitting a mathematical model of succinate uptake and growth to the data. We accounted for the observed diauxic growth of Flavonifractor sp. by adding a second (unobserved) resource to the model (blue terms). b Experimental data and model fits for one representative isolate of each genus. For each isolate, two biological replicate “sets” were inoculated (circles and triangles). Initial bacterial concentrations, x₀, and succinate concentrations, A₀, were estimated separately for each replicate set and are shown as separate lines. Data for Flavonifractor isolates used the diauxic model. c Posterior mean estimates (bars) and 90% highest-probability density intervals (black lines) for the eleven strains. P1: P. faecium DSMZ 14760; P2: P. faecium PB-SDVAP; P3: P. faecium PB-SJWFW; P4: P. faecium PB-SPUPY; P5: P. succinatutens DSMZ 22533; D1: D. hominis DSMZ 109768; D2: D. invisus PB-SARUR; D2: D. succinatiphilus DSMZ 21274; F1: F. plautii DSMZ 24814; F2: F. plautii PB-SCBYV; F3: F. plautii PB-SSJQB

Fig. 3

Human fecal microbiota can be classified into “succinotypes” based on their dominant succinate consuming bacterium. a Relative abundances of the four identified succinate consuming genera in the fecal microbiota. b The three genera Phascolarctobacterium,Phascolarctobacterium_A, and Dialister are strongly mutually exclusive. The position in the ternary diagram shows to the relative abundance of the three genera in each of the fecal microbiota. Points in the corners indicate full mutual exclusivity. c Relative abundances of the succinate consuming taxa in the fecal microbiota resolved at the species level. The asterisk indicates the dominant species in a fecal microbiota. We assign fecal microbiota with a dominant Phascolarctobacterium,Phascolarctobacterium_A to a succinotype “P” and those with a dominant Dialister to a succinotype “D”

Fig. 4

Clear succinotypes are found broadly and are associated with disease. a The distribution of the relative abundance ratio of Dialister (D) compared to Phascolarctobacterium/Phascolarctobacterium_A (P) in fecal samples is strongly bimodal. The histogram encompasses 9 pooled cohorts of 16S amplicon data with a total of 8911 individuals. b Transition probabilities between succinotypes over time estimated individuals with repeat samples in the BIO-ML cohort. c Distribution of relative abundances of D (red) and P (blue) in the fecal samples of healthy individuals. There is no significant difference between D and P across cohorts. d Succinate concentrations in fecal samples from the BIO-ML cohort (Mann-Whitney U test, p = 0.001619). e Fraction of subjects assigned the D and P succinotype across cohorts and disease status. Dark colors are healthy subjects and light colors are IBD patients. f Succinate concentrations in fecal samples from the PRISM cohort. There are no significant differences in fecal succinate between healthy subjects and IBD patients. AGP: American Gut Project; UCC CAN: UCC Canadian Cohort; UCC IRE: UCC Irish Cohort; IBD Fam: IBD Families Cohort; BIO-ML: Broad Institute-OpenBiome Microbiome Library