Intermittent Fasting Confers Protection in CNS Autoimmunity by Altering the Gut Microbiota

Abstract

Multiple sclerosis (MS) is more common in western countries with diet being a potential contributing factor. Here we show that intermittent fasting (IF) ameliorated clinical course and pathology of the MS model, experimental autoimmune encephalomyelitis (EAE). IF led to increased gut bacteria richness, enrichment of the Lactobacillaceae, Bacteroidaceae, and Prevotellaceae families and enhanced antioxidative microbial metabolic pathways. IF altered T cells in the gut with a reduction of IL-17 producing T cells and an increase in regulatory T cells. Fecal microbiome transplantation from mice on IF ameliorated EAE in immunized recipient mice on a normal diet, suggesting that IF effects are at least partially mediated by the gut flora. In a pilot clinical trial in MS patients, intermittent energy restriction altered blood adipokines and the gut flora resembling protective changes observed in mice. In conclusion, IF has potent immunomodulatory effects that are at least partially mediated by the gut microbiome.

Trial registration: ClinicalTrials.gov NCT02411838.

Keywords: diet; experimental autoimmune encephalomyelitis; gut microbiota; intermittent fasting; multiple sclerosis.

Conflict of interest statement

COMPETING FINANCIAL INTERESTS

The authors declare no competing financial interests. AHC has served as a paid consultant for: AbbVie, Bayer, Biogen, EMD Serono, Genentech/Roche, Genzyme, Novartis and Teva.

Copyright © 2018 Elsevier Inc. All rights reserved.

Figures

Figure 1.. Intermittent fasting (IF) ameliorates EAE clinical course and pathology.

C57BL/6 mice underwent IF or were fed ad libitum (control group) for a month before immunization (n=10/group). (A) EAE clinical course of a representative experiment (dots represent the mean clinical scores for all 10 mice in each group, error bars are SEM.; P<0.0001 by two-way ANOVA). Four EAE experiments were performed with similar results. (B) Spinal cord pathology: the upper panel shows a histological staining (solochrome cyanine) for myelin (in blue) and the lower panel is an immuno-staining for SMI-32+ damaged axons (in red) and MBP (in green). (C) Quantification of inflammation, demyelination (evaluated by histology and MBP staining) and axonal damage (evaluated by SMI-32+ staining) in the spinal cord in the two groups (n=10/group) on day 26 post-immunization. Each dot represents a mouse and the bars are means ± SD. (D) Percentages of CD4+ T cells producing IL-17A, IFN-γ and GM-CSF measured by flow cytometry in lymph nodes draining the immunization site on day 6 post-immunization. Each dot represents a mouse and the bars are means ± SD. This is one of three different experiments performed with similar results (n=4–5/group in each experiment; Table S2 reports the results of all experiments performed). (E) Representative flow-cytometry plots for T cell production of IL-17A and IFN-γ in CD4+ T cells isolated from the draining lymph nodes at day 6 post-immunization in the two groups. * P<0.05; **P<0.005. All P values were calculated by Mann-Whitney test.

Figure 2.. IF was associated with decreased leptin and increased adiponectin, corticosterone and β-hydroxybutyrate.

Serum levels of (A) leptin (n=6/group), (B) adiponectin (n=5/group), (C) corticosterone (n=6/group) and (D) β-hydroxybutyrate (n=10/group) were measured by ELISA at different time points during the experiment: baseline before starting the diet (IF or ad libitum feeding-T1), after 4 weeks on the diet but before immunization (T2), during clinical EAE (day 18–20 post-immunization-T3). The box in the graphs extends from the 25th to 75th % percentiles, the bars are median, the whiskers indicate the smallest and largest values. Measurements for leptin, adiponectin and corticosterone were performed in 3 different experiments with similar results (Table S2 reports the results of all experiments performed); β−hydroxybutyrate was measured in 2 different experiments with similar results. In the ELISA assays each sample was run in duplicate. * P<0.05; **P<0.005, *** P<0.0005. AllP values were calculated by Mann-Whitney test.

Figure 3.. IF increases diversity and has a profound effect on gut microbiome composition.

Stool samples were collected from the IF and ad libitum groups at T1 (baseline, n=10 in the IF group, n=9 in the ad libitum group), T2 (after 4 weeks on the diet, prior to immunization, n=9 in IF, n=10 in ad libitum) and T3 (clinical EAE, n=10 in IF, n=8 in ad libitum). (A) Non-metric multidimensional scaling (NMDS) plots illustrate microbiome similarity in IF and ad libitum groups (each dot is one sample; X-axis and Y-axis are first and second dimension of microbiome data). At T1, samples from two groups intermingled, indicating similar microbiome (P>0.05, PERMANOVA test). At T2 and T3, samples from the two groups clustered separately, indicating two distinct microbiome communities (P<0.05, PERMANOVA test). (B) Richness is a measure of alpha diversity for microbial community. Bacterial richness increased significantly over the three time points in the IF group, but not in the ad libitum group (linear mixed regression #P<0.05; Y-axis is number of different bacteria families); (C) Blood leptin level was negatively correlated with bacterial richness (r=−0.51; P<0.005, Pearson correlation). (D) Bacterial families with significantly different relative abundance between the two groups at T2 and T3. Reported here only those bacterial families with the same direction of difference at T2 and T3 between the two groups. Y-axis is the relative abundance of the bacterial family (q<0.05, ANCOVA). * q<0.05; ** q<0.01. (E)Pearson or Spearman correlations of corticosterone or leptin and microbiome (data were log transformed). All samples from IF and ad libitum groups are included. (F) Different trajectories ofLactobacillus species in the IF and ad libitum groups were seen over the three time points. In the IF group, L. sp are significantly higher at the T2 and T3 time points (P<0.05, linear mixed regression). In b-d, IF is in grey and ad libitum in black.

Figure 4.. IF results in significantly different metabolic pathways in the gut microbiome.

Metagenomic whole genome shotgun was performed for stool samples collected at T3 in each group (n=5/group). A linear discriminate analysis (LDA) was conducted to identify differentially represented pathways in the two groups. Pathways with LDA score >2.5 (x-axis) and P<0.01 are shown. The right side of the figure represents pathways whose abundance was significantly higher in IF group. The left side of the figure represents pathways whose abundance was significantly higher in the ad libitum group. The absolute LDA value is the effect size between two groups for a particular pathway. The color of the bar represents main metabolic functions to which the different pathways belong to as indicated in the legend.

Figure 5.. Fecal microbiota transplantation (FMT) from mice on IF is protective in EAE.

Mice were pre-treated with an antibiotic cocktail for a week, and then subjected to FMT from donor mice that were on IF (for 4 weeks) or fed ad libitum. FMT was administered for a week before and a week after EAE immunization. (A) EAE clinical course in mice transferred with fecal matters from mice on IF or fed ad libitum. Shown is one representative experiment out of two performed with similar results (each dot represent the mean clinical scores for all 5 mice in each group; error bars are SEM.;P<0.0001 by two-way ANOVA; Table S2 reports clinical characteristics for the two experiments performed). (B) Spinal cord pathology: the upper panel shows histological staining (solochrome cyanine) for myelin (in blue) and the lower panel shows immuno-staining for SMI-32+ damaged axons (in red) and MBP (in green). (C) Quantification of inflammation, demyelination (evaluated by histology and MBP staining) and axonal damage (evaluated by SMI-32+ staining) in the spinal cord in the two groups (n=10/group). Each dot represents a mouse and the bars are means ± SD.

Figure 6.. Reduced proportion of IL-17- producing T cells and increased proportion of Tregs in the gut lamina propria after 4 weeks of IF.

(A) Intracellular cytokine production by CD4+ T cells in the small intestine lamina propria (SI LP) after 4 weeks on IF or normal ad libitum analyzed by flow cytometry. (B) Proportion of T regs in the SI LP in the 2 groups. Representative flow cytometry plots for one mouse/group are shown on the right panel. (C) Proportion of Tregs in mesenteric lymph nodes (MLN), peripheral lymph nodes (PLN) and spleen. Each dot represents a mouse, bars are mean ± SD. These results are from one experiment out of two (in A) or three (in B-C) performed with similar results (Table S2 reports the results of all experiments performed). * P<0.05; **P<0.005, ***P<0.0005. AllP values were calculated by Mann-Whitney test.

Figure 7.. Changes of blood metabolites and the gut microbiome after 15 days of IF in RRMS patients.

(A) Comparison of serum levels of leptin and adiponectin on day 15 in RRMS patients in the IF and ad libitum groups in the human trial. (B) At phylum level, the alteration of the gut microbiome after IF for 15 days in RRMS patients (n=5 patients/group, top panel) shows similar trends as that in mice (n=8 mice/group, bottom panel). Y-axis represents the percentage of change of the relative abundance of the gut microbiome. In human studies, the percentage change is calculated based on the microbial abundance at baseline and day 15. In mouse studies, the percent change is calculated based on the microbial abundance on T2 and T3. (c) Levels of serum adiponectin are strongly positively correlated with the relative abundance ofFaecalibacterium in the MS study participants (r=0.86,P=0.009 Pearson correlation).