The gut microbiota in multiple sclerosis varies with disease activity

Abstract

Background: Multiple sclerosis is a chronic immune-mediated disease of the brain and spinal cord resulting in physical and cognitive impairment in young adults. It is hypothesized that a disrupted bacterial and viral gut microbiota is a part of the pathogenesis mediating disease impact through an altered gut microbiota-brain axis. The aim of this study is to explore the characteristics of gut microbiota in multiple sclerosis and to associate it with disease variables, as the etiology of the disease remains only partially known.

Methods: Here, in a case-control setting involving 148 Danish cases with multiple sclerosis and 148 matched healthy control subjects, we performed shotgun sequencing of fecal microbial DNA and associated bacterial and viral microbiota findings with plasma cytokines, blood cell gene expression profiles, and disease activity.

Results: We found 61 bacterial species that were differentially abundant when comparing all multiple sclerosis cases with healthy controls, among which 31 species were enriched in cases. A cluster of inflammation markers composed of blood leukocytes, CRP, and blood cell gene expression of IL17A and IL6 was positively associated with a cluster of multiple sclerosis-related species. Bacterial species that were more abundant in cases with disease-active treatment-naïve multiple sclerosis were positively linked to a group of plasma cytokines including IL-22, IL-17A, IFN-β, IL-33, and TNF-α. The bacterial species richness of treatment-naïve multiple sclerosis cases was associated with number of relapses over a follow-up period of 2 years. However, in non-disease-active cases, we identified two bacterial species, Faecalibacterium prausnitzii and Gordonibacter urolithinfaciens, whose absolute abundance was enriched. These bacteria are known to produce anti-inflammatory metabolites including butyrate and urolithin. In addition, cases with multiple sclerosis had a higher viral species diversity and a higher abundance of Caudovirales bacteriophages.

Conclusions: Considerable aberrations are present in the gut microbiota of patients with multiple sclerosis that are directly associated with blood biomarkers of inflammation, and in treatment-naïve cases bacterial richness is positively associated with disease activity. Yet, the finding of two symbiotic bacterial species in non-disease-active cases that produce favorable immune-modulating compounds provides a rationale for testing these bacteria as adjunct therapeutics in future clinical trials.

Keywords: Faecalibacterium prausnitzii; Gordonibacter urolithinfaciens; Gut microbiota; Multiple sclerosis; Shotgun sequencing.

Fig. 1

Contrasting bacterial species (metagenomics species (MGS)) and functional modules. A Barcode illustration of contrasted bacterial species after deconfounding for covariates (age, sex, BMI, smoking, and drug treatment). The 50 “tracer” genes are in rows, abundance is indicated by color gradient (white, not detected; red, most abundant); individuals, ordered by status (cases or HC) and by increasing species richness, are in columns. B Boxplots of contrasted bacterial modules (gut metabolic module (GMM) and gut-brain module (GBM)) after deconfounding for the same covariates. P-values associated with Wilcoxon test are displayed. MS = multiple sclerosis patients, HC = healthy controls

Fig. 2

Associations of contrasted bacterial species (metagenomics species (MGS)) with inflammatory markers. A Spearman’s correlations between contrasted bacterial species and fasting circulating inflammatory markers in the subgroup of treatment-naïve patients only. Only features with at least one p-value under 0.05 are displayed. Black dots denote correlations with FDR ≤ 0.1, while empty circles indicate correlation with P ≤ 0.05. The right side bars indicate the Cliff’s Delta (CD, effect size) of the feature in the cases/HC contrast (red: more abundant in cases; blue: more abundant in HC). B Relationships between abundance of Flavonifractor plautii and a group of fasting circulating inflammation markers. C Relationships between abundance of Clostridium leptum and expression of selected blood leukocyte genes. Spearman’s correlation coefficients along with the associated p-values are displayed. CD = Cliff’s Delta; MGS = metagenomics species. MS = multiple sclerosis patients, HC = healthy controls, EDSS = expanded disability status scale; MSSS = multiple sclerosis severity score (0: at baseline; 1: after 2-year follow-up)

Fig. 3

Species (metagenomics species (MGS)) richness and disease activity in treatment-naïve patients. A–C Relationships between A species richness or C gene richness adjusted for covariates (age, sex, BMI, smoking status, and fecal water content) and number of relapses during follow-up, in treatment-naïve patients only. Spearman’s correlation coefficients along with the associated p-values are displayed. B–D Distribution of B adjusted species richness or D adjusted gene richness, according to disease activity in treatment-naïve patients. P-values associated with Wilcoxon tests are displayed. CNA = clinically not active; CA = clinically active; MS = multiple sclerosis patients, MGS = metagenomics species

Fig. 4

Bacterial species (metagenomics species (MGS)) and bacteriome modules related to MS activity. A Bacterial species and D predicted bacteriome functional modules that are contrasted between CA and CNA treatment-naïve patients (after deconfounding for age, sex, BMI, smoking status, and fecal water content). Along is their effect size (Cliff’s Delta) in the contrasts (1) CA vs CNA, (2) CA vs HC, (3) CNA vs HC. B,C,E Distribution of bacterial species or bacteriome functional modules that are more abundant in CNA patients. P-values associated with Wilcoxon tests are displayed. F Correlation between contrasted bacterial species and fasting circulating inflammation markers in treatment-naïve patients. Only bacterial species with at least one p-value under 0.05 are displayed. Black dots denote correlations with FDR ≤ 0.1, while empty circles indicate correlation with P ≤ 0.05. The right side bars indicate the Cliff’s Delta (CD, effect size) of the feature in the CA/CNA contrast (green: more abundant in CNA; yellow: more abundant in CA). CD = Cliff’s Delta; CA = clinically active; CNA = clinically not active; HC = healthy controls; MS = multiple sclerosis patients; NS = non-significant, EDSS = expanded disability status scale; MSSS = multiple sclerosis severity score (0: at baseline; 1: after 2-year follow-up)

Fig. 5

Alteration of viral gut microbiota composition in cases and HC subjects. A Relative abundance of gut viral orders in cases and HC groups. B Relative abundance of Caudovirales in cases and HC individuals. C Shannon’s diversity for the viral gut microbiota between patients and HC at the virus species level. Statistical significance was determined by Wilcoxon’s rank sum test between two groups. D Principal coordinate analysis (PCoA) of the Canberra distance showing the stratification of patients from HC by viral gut microbiota at species level. Statistical significance for the Canberra distance was determined by PERMANOVA with permutations done 999 times. E Relative abundance of bacteriophage Enterococcus phage EFC-1 in all cases and HC. F Relative abundance of bacteriophage Enterococcus phage EFC-1 in treated or never treated cases compared with HC. G Gut viral species associate with blood cell expression of inflammation markers. Statistical significance was determined by Wilcoxon’s rank sum test between two groups. Kruskal-Wallis test, followed by Wilcoxon’s rank sum test with Benjamini-Hochberg correction was performed between the three groups. HC = healthy controls