Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Gut microorganisms act together to exacerbate inflammation in spinal cords

Abstract

Accumulating evidence indicates that gut microorganisms have a pathogenic role in autoimmune diseases, including in multiple sclerosis1. Studies of experimental autoimmune encephalomyelitis (an animal model of multiple sclerosis)2,3, as well as human studies4,5,6, have implicated gut microorganisms in the development or severity of multiple sclerosis. However, it remains unclear how gut microorganisms act on the inflammation of extra-intestinal tissues such as the spinal cord. Here we show that two distinct signals from gut microorganisms coordinately activate autoreactive T cells in the small intestine that respond specifically to myelin oligodendrocyte glycoprotein (MOG). After induction of experimental autoimmune encephalomyelitis in mice, MOG-specific CD4+ T cells are observed in the small intestine. Experiments using germ-free mice that were monocolonized with microorganisms from the small intestine demonstrated that a newly isolated strain in the family Erysipelotrichaceae acts similarly to an adjuvant to enhance the responses of T helper 17 cells. Shotgun sequencing of the contents of the small intestine revealed a strain of Lactobacillus reuteri that possesses peptides that potentially mimic MOG. Mice that were co-colonized with these two strains showed experimental autoimmune encephalomyelitis symptoms that were more severe than those of germ-free or monocolonized mice. These data suggest that the synergistic effects that result from the presence of these microorganisms should be considered in the pathogenicity of multiple sclerosis, and that further study of these microorganisms may lead to preventive strategies for this disease.

Main

Mice were given an antibiotic cocktail that contained ampicillin, vancomycin, neomycin and metronidazole in their drinking water one week before sensitization to MOG peptide fragment 35–55 (MOG35–55), and the antibiotic treatment was continued throughout the experiment (Extended Data Fig. 1a). Consistent with previous reports7, depletion of the gut microbiota by this antibiotic cocktail resulted in attenuated symptoms of experimental autoimmune encephalomyelitis (EAE) (Extended Data Fig. 1b). Notably, the treatment with ampicillin alone—but not with the other antibiotics in isolation—also limited the development of EAE (Fig. 1a). Mice that were orally treated with ampicillin were protected from demyelination of the spinal cord (Fig. 1b), and from infiltration of cells (including CD4+ T cells that produce IFNγ and IL-17A) into the spinal cord (Fig. 1c, d). To address whether treatment with ampicillin ameliorated the peripheral activation of MOG-specific T cells, splenocytes from each mouse were restimulated ex vivo with MOG35–55 for cytokine production. The cells from ampicillin-treated mice secreted significantly fewer cytokines in response to MOG35–55 than did those from mice treated with the other antibiotics (Extended Data Fig. 1d). Mice that were intraperitoneally injected with ampicillin displayed symptoms of EAE (measured as clinical scores) that were comparable to those of control mice injected with phosphate-buffered saline (PBS) (Extended Data Fig. 1c), which indicates that ampicillin acts in the intestine—but not systemically—to ameliorate EAE. We thus analysed MOG-specific immune responses in intestinal lamina propria and gut-associated lymphoid tissues. Ex vivo restimulation with MOG35–55 increased the production of IFNγ and IL-17 in lamina propria cells, especially in those from the small intestine of mice in which EAE had been induced (EAE mice) (Fig. 1e); by contrast, only small responses were observed in cells from mesenteric lymph nodes and Peyer’s patches. Small-intestinal CD4+, but not CD8+, T cells co-cultured with CD11c+ cells also produced elevated levels of these cytokines in response to restimulation with MOG35–55 in vitro (Extended Data Fig. 1e). CD4+ T cells, as well as a subset of these cells that expresses the proliferative marker Ki67, in the small intestine were increased in EAE mice compared to naive mice (Fig. 1f, g). Importantly, MOG-specific cytokine production and proliferation of CD4+ T cells in the small intestine were reduced by oral treatment with ampicillin (Fig. 1e, f). In addition, FOXP3+CD4+ regulatory T cells were not increased by treatment with ampicillin (Extended Data Fig. 1g–k). These observations support the hypothesis that MOG-specific T cells are activated by microorganisms in the small intestine.

Fig. 1: Suppression of EAE by manipulation of the gut microbiota.
figure1

a, EAE clinical scores of mice given normal drinking water (control), or drinking water containing an antibiotic. The combined results of two independent experiments are shown (n = 8). Asterisks indicate a significant difference between the control group and the ampicillin-treated group. Detailed statistics of the EAE clinical scores are summarized in Supplementary Table 3. Amp, ampicillin; Met, metronidazole; Neo, neomycin; Van, vancomycin. b, Left, representative images of spinal-cord sections stained with luxol fast blue, from control (n = 7) and ampicillin-treated (n = 7) groups. Scale bar, 500 μm. Right, the demyelinated area in the white matter was calculated. c, Representative fluorescence-activated cell sorting (FACS) plots of TH1 and TH17 cells in the spinal cord (gated on CD3+ and CD4+ cells). d, Absolute numbers of total cells, TH1 cells and TH17 cells in the spinal cord of control- (n = 6) and ampicillin-treated (n = 4) mice. e, The cells from small-intestinal lamina propria (SI LP), colonic lamina propria (LI LP), mesenteric lymph nodes (mLN) and Peyer’s patches (PP) of naive (naive), EAE (control) or ampicillin-treated EAE mice were cultured in the presence or absence of MOG35–55 for three days. The cytokine concentrations in the supernatants were measured by enzyme-linked immunosorbent assay (ELISA) (n = 6). f, Representative FACS plots (gated on CD3+ cells) and summary data of Ki67+CD4+ T cells in the small-intestinal lamina propria are shown (n = 6 mice). Con, control. g, Number of CD4 T cells in the small-intestinal lamina propria (n = 6 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05; two-way analysis of variance (ANOVA) with Tukey’s test (a, e), two-tailed unpaired t-test (b, d) or one-way ANOVA with Tukey’s test (f, g). Exact P values are in Source Data.

Source data

Although some antibiotics (including ampicillin) significantly decreased gut microorganisms in terms of 16S rRNA gene copy number (Extended Data Fig. 2a), the bacterial load itself seems insufficient to explain the effect of antibiotics on the severity of EAE. Each antibiotic modulated the composition of the small-intestinal microbiota in a distinct way, at the phylum level (Extended Data Fig. 2b); the principal coordinate analysis of unweighted UniFrac showed a clear separation for each group, and ampicillin-treated mice, especially, showed a unique microbiota structure in the small intestine (Extended Data Fig. 2c). Indicator species analysis showed that operational taxonomic unit (OTU)0002 (assigned to Allobaculum, for which an association with T helper 17 (TH17) cells has previously been suggested8) was the sole sequence that was almost completely depleted only from the small intestine of ampicillin-treated mice (Fig. 2a). We isolated a bacterial strain from the contents of the small intestine that possesses a 16S rRNA gene sequence identical to of OTU0002, and sequenced its genome. According to a phylogenetic analysis on the basis of 16S rRNA gene sequences, Allobaculum stercoricanis DSM 13633 was the closest type strain to OTU0002 (Extended Data Fig. 2d). The sequence identity of the two strains was 93.7%, which indicates that the OTU0002 strain is probably a newly isolated bacterium of the Erysipelotrichaceae family. Notably, we also detected bacteria of the Erysipelotrichaceae that are closely related to OTU0002 in other mammals, including humans (Extended Data Fig. 3e). OTU0002 was dominant both in luminal and mucosal microbiota in the small intestine, and the relative abundance of this strain in mucosa was comparable to that of segmented filamentous bacteria (SFB) (Fig. 2b), which colonize ileal mucosa by attaching to the epithelial cells9. Scanning electron microscopy (SEM) of intestinal mucosa from mice that were monocolonized with OTU0002 revealed that this strain attached to epithelial cells of the small intestine, but not to those of the caecum or colon (Fig. 2c, Extended Data Fig. 2f). Genome analysis also indicated that OTU0002 possessed several genes that are involved in colonization and biofilm formation (for example, capsular polysaccharide biosynthesis protein; OTU2_01244 in Supplementary Table 1).

Fig. 2: Increased severity of EAE caused by colonization with OTU0002.
figure2

a, The contents of the small intestine were collected on day 0 (as shown in Extended Data Fig. 1a), and the bacterial compositions of each mouse were analysed (n = 5). OTUs associated with each antibiotic treatment were identified by indicator analysis. The heat map shows the relative abundance and indicator values of indicator OTUs (nonadjusted P < 0.05 from the one-tailed permutation test, indicator value >0.7). OTU identifier (ID) and taxonomy of the indicators are shown on the right. Unc., unclassified. b, Luminal and mucosal bacteria were collected from the small intestine of naive specific-pathogen-free mice (n = 8), and the relative abundance of SFB and OTU0002 was analysed. c, The isolated strain corresponding to OTU0002 was inoculated into germ-free mice. Representative SEM images of the ileum from the OTU0002-monocolonized mice are shown (n = 3). Left, 300× magnification; right, 3,000× magnification. d, EAE clinical scores of germ-free (GF) mice inoculated with L. reuteri or OTU0002. The combined results of two independent experiments are shown (germ-free, n = 5; L. reuteri and OTU0002, n = 9). Asterisks indicate a significant difference between the germ-free mice and the OTU0002-monocolonized group. Detailed statistics of the EAE clinical scores are summarized in Supplementary Table 4. e, The small-intestinal lamina propria cells were cultured in the presence or absence of MOG35–55 for three days. The cytokine concentrations in the supernatants were measured by ELISA (n = 5 mice). f, Representative FACS plots (gated on CD3+ and CD4+), percentage and absolute numbers of TH17 cells in the small-intestinal lamina propria (n = 5 mice). g, Percentage and absolute numbers of FOXP3+CD4+ regulatory T cells in the small-intestinal lamina propria (n = 5 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05; two-way ANOVA with Tukey’s (d) or Bonferroni’s (e) test, two-tailed unpaired t-test (b, d), one-way ANOVA with Tukey’s test (f) or Kruskal–Wallis with Dunn’s test (g). Exact P value are in Source Data.

Source data

To examine the effect of OTU0002 colonization on the development of EAE, we induced EAE in mice that were monocolonized with OTU000. As a control, the bacterium corresponding to OTU0001—the most abundant OTU across all samples, and which belongs to Lactobacillus reuteri (100% match to the 16S rRNA gene of L. reuteri strains H4 and LMG 18238)—was also isolated from the content of the small intestines of specific-pathogen-free mice, and inoculated to germ-free mice. Colonization with OTU0002 resulted in an increased severity of EAE symptoms compared to those of germ-free mice, whereas colonization with L. reuteri did not noticeably affect EAE symptoms (Fig. 2d). Cytokine production—especially of IL-17A in cells of the lamina propria of the small intestine and splenocytes isolated from the EAE mice—in response to ex vivo restimulation with MOG35–55 was higher with OTU0002 monocolonization than with L. reuteri monocolonization, or in germ-free mice (Fig. 2e, Extended Data Fig. 3a). The frequency and number of TH17 cells, both in the small-intestinal lamina propria and spleen, were increased in mice monocolonized with OTU0002 (Fig. 2f, Extended Data Fig. 3b). The concomitant increase in small-intestinal regulatory T cells was also observed in mice monocolonized with OTU0002 (Fig. 2g). Serum amyloid A (SAA) produced by epithelial cells and IL-23 produced by dendritic cells have pivotal roles in the induction of TH17 cells by mucosa-associated gut bacteria, including SFB10,11. In addition, IL-23 promotes the pathogenicity of TH17 cells by inducing the transcription factor BLIMP1 and its target genes, including Il17a, Csf2 (which encodes GM-CSF) and Il23r12. The expression of genes Saa1, Saa2, Il23a (which encodes IL-23p19) and Il12b (which encodes IL-12 (also known as IL-23p40)) was significantly increased in the small-intestinal tissue of mice monocolonized with OTU0002 (Extended Data Fig. 3c). To evaluate the role of SAA in EAE, we next prepared splenocytes from MOG35–55-immunized specific-pathogen-free mice and restimulated the cells with MOG35–55 in the presence or absence of SAA. SAA increased the gene expression of Prdm1 (which encodes BLIMP1) and Rorc in CD4+ cells (Extended Data Fig. 3d). The target genes of these transcription factors—including Il17a, Csf2 and Il23r—were also upregulated in CD4+ cells in the presence of SAA (Extended Data Fig. 3d). The concentrations of IL-17A and GM-CSF in the culture supernatants were also increased by the treatment with SAA (Extended Data Fig. 3e). Consistent with increased expression of Saa1 and Saa2, the expression of the genes Csf2 and Il23r was elevated in the small-intestinal tissues of mice monocolonized with OTU0002 (Extended Data Fig. 3f). In naive mice, monocolonization of OTU0002 induced a significant increase in TH17 cells (Extended Data Fig. 3g). Thus, OTU0002 may promote the TH17-cell responses of MOG-specific TH17 cells in the small intestine by inducing SAA and IL-23—most probably from epithelial cells and dendritic cells, respectively—and act in a manner similar to an adjuvant to enhance the pathogenicity of EAE, especially during disease development.

Although colonization of OTU0002 facilitated EAE, mice monocolonized with OTU0002 exhibited EAE symptoms that were less severe (Fig. 2d) than those of specific-pathogen-free mice (Fig. 1a), which indicated that other bacterial members of the microbiota could also be participating in the pathogenesis of EAE. A previous study suggested that antigens from gut microbiota cross-react with, and activate, autoreactive T cells in autoimmune uveitis13. We therefore hypothesized that antigens from bacteria in the small intestine that cross-react with MOG-specific T cells may have an essential role in the development of EAE. Evidence in support this idea was obtained with MOG35–55-specific Vα3.2 and Vβ11 T-cell-receptor (TCR)-transgenic 2D2 mice. These mice showed increased numbers of CD4+ T cells in the small intestine compared to their wild-type littermates, with significantly more of the proliferative Ki67+CD4+ T cells (Extended Data Fig. 4a). These differences were not observed in the spleen (Extended Data Fig. 4b). Depletion of the microbiota using the antibiotic cocktail significantly decreased the number of CD4+ T cells and Ki67+CD4+ cells in 2D2 mice (Extended Data Fig. 4c), and—more specifically—of Ki67+ TCR 2D2 T cells (Extended Data Fig. 4d), which collectively indicates that MOG-specific T cells are activated by microorganisms in the small intestine. We next used shotgun sequencing data of the contents of the small intestines from non-treated mice (control in Fig. 1a) and sought in silico candidate mimicry peptides with sequences that matched the TCR-binding residues of MOG peptide fragment 40–48 (MOG40–48) (Pro2, Pro5, Pro7 and Pro8)14. BLAST analysis against the NCBI Microbial Proteins database revealed that more than half of the candidate mimicry peptides were inferred to be derived from species of Lactobacillus (Extended Data Fig. 5a). The shotgun reads obtained from contents of the small intestines of antibiotic-treated mice were mapped to their representative mimics. Among these mimics, three proteins—including the UvrABC system protein A (UvrA)—were found in Lreuteri, and aminopeptidase was found in the OTU0002 genome (Fig. 3a). Stimulation of cells from the mesenteric lymph nodes from 2D2 mice, but not from wild-type mice, with MOG peptide fragment 38–50 (MOG38–50) strongly induced the proliferation of CD4+ T cells with verified Ki67 expression (Fig. 3b). The 2D2 CD4+ T cells also expressed Ki67 in response to synthesized UvrA, but not aminopeptidase, peptides in an MHC-class II-dependent manner (Fig. 3b, Extended Data Fig. 5b). The antigen-experienced marker CD44 in 2D2 CD4+ T cells was also significantly upregulated by MOG38–50 and weakly upregulated by UvrA peptides (Extended Data Fig. 5c). Stimulation of cells from the spinal cord of EAE mice with UvrA peptides also weakly induced cytokine secretion (Extended Data Fig. 5d); however, mice immunized with the UvrA peptide did not exhibit overt clinical symptoms of EAE (Extended Data Fig. 5e). Although gene expression of pro-inflammatory cytokines in the spinal cord was slightly increased (albeit not significantly), demyelination in the spinal cord was not induced by immunization with UvrA (Extended Data Fig. 5f, g), which indicates that UvrA-specific T cells may not target myelin because of weaker antigen recognition. In the intestine, the RNA/DNA ratio of the uvrA gene was high in the small intestine of naive mice (Extended Data Fig. 5g), which suggests that uvrA is highly expressed in L. reuteri—especially in the small intestine. Taken together, UvrA expressed by L. reuteri could cross-react with and weakly activate MOG-specific T cells.

Fig. 3: Coordinate effects of L. reuteri and OTU0002 on EAE development.
figure3

a, Reads from shotgun sequencing of the contents of the small intestine were mapped to candidate mimicry-peptide-containing proteins, as shown in Extended Data Fig. 5a. The sequences corresponding to MOG40–48 (TCR-binding residues are shown in bold) and annotations are shown. b, Mesenteric lymph nodes from wild-type (WT) or 2D2 mice were stimulated with the indicated peptides for four days. Representative FACS plots (gated on CD3+ cells) and summary data are shown (n = 6). c, The abundance of L. reuteri and OTU0002 in the contents of the small intestine (n = 5 mice). CFU, colony-forming units. d, e, EAE clinical scores (d) and incidence (e) of germ-free mice monocolonized with OTU0002 or co-colonized with L. reuteri and OTU0002. The combined results of two independent experiments are shown (germ-free, n = 9; OTU0002 and L. reuteri + OTU0002, n = 10 mice). Asterisks indicate a significant difference between the monocolonized group and the co-colonized group. f, Representative FACS plots (gated on CD3+ and CD4+ cells), percentage, and absolute numbers of TH17 cells in the small-intestinal lamina propria (n = 5 mice). g, Representative FACS plots (gated on CD3+ cells) and summary data of Ki67+CD4+ T cells in the small-intestinal lamina propria are shown (germ-free and OTU0002, n = 4; L. reuteri + OTU0002, n = 5 mice). h, i, EAE clinical scores (h) and incidence (i) of germ-free mice co-colonized with OTU0002 and a wild-type or uvrA-deficient (ΔuvrA) strain of L. reuteri. The combined results of two independent experiments are shown (n = 7). j, Ki67+CD4+ cells in the small-intestinal lamina propria (n = 4 mice). Data are mean ± s.d. Detailed statistics of the EAE clinical scores are summarized in Supplementary Table 5, 6. ***P < 0.001, **P < 0.01, *P < 0.05; Kruskal–Wallis with Dunn’s test (b, c), one-way (f, g) or two-way (d, h) ANOVA with Tukey’s (d, f, g) or Bonferroni’s (h) test, or two-tailed unpaired t-test (j). EAE incidence was analysed by two-tailed Mantel–Cox test (e, i). Exact P values are in Source Data.

Source data

As shown in Fig. 2d, L. reuteri monocolonization did not affect the severity of EAE, which indicates that stimulus by the L.-reuteri-derived mimicry peptides alone is not sufficient for the pathogenesis of EAE. We therefore considered that the distinct signals from L. reuteri and OTU0002 act together to activate MOG-specific T cells and worsen EAE. Mice co-colonized with L. reuteri and OTU0002 showed an abundance of OTU0002 in contents of the small intestine similar to that in mice monocolonized with OTU0002 (Fig. 3c). Nevertheless, the EAE clinical score and incidence were significantly higher in the co-colonized mice compared to the monocolonized mice (Fig. 3d, e). Demyelination and cell infiltration in the spinal cord were also accelerated by the co-colonization (Extended Data Fig. 6a, b). IL-17 production by splenocytes or small-intestinal lamina propria cells restimulated with MOG35–55 did not differ between the monocolonized and co-colonized mice (Extended Data Fig. 6c). The frequency and number of TH17 cells and regulatory T cells in the small intestine were also comparable between the monocolonized and co-colonized mice (Fig. 3f, Extended Data Fig. 6d). Accordingly, gene expression of Saa1, Saa2 and Il23a was unchanged between the monocolonized and co-colonized mice (Extended Data Fig. 6e). In the spleen, the number of TH17 cells was higher in the co-colonized mice than in the monocolonized mice, although this was not significant (Extended Data Fig. 6f). By contrast, Ki67+ proliferating CD4+ T cells were significantly increased in the small intestine of the co-colonized mice (Fig. 3g). Similar results were observed when mice were co-colonized with L. reuteri and TH17-cell-inducible SFB (Extended Data Fig. 7). Thus, L. reuteri cooperates with OTU0002 to trigger robust EAE symptoms not by amplifying antigen-nonspecific TH17 cell responses in the small intestine, but instead probably through antigen-specific TH17 cell activation among the total TH17 cell population by cross-reacting with MOG-specific T cells.

To further clarify the involvement of UvrA in the pathogenesis of EAE, we constructed a uvrA-deficient L. reuteri strain (L. reuteri ΔuvrA) using homologous recombination (Extended Data Fig. 8). Monocolonization of germ-free wild-type mice with L. reuteri wild type or L. reuteri ΔuvrA induced equivalent levels of Ki67+CD4+ T cells in the small intestine (Extended Data Fig. 9a). On the other hand, monocolonization with L. reuteri ΔuvrA led to significantly impaired activity to induce Ki67+CD4+ T cells (Extended Data Fig. 9a), especially Ki67+ 2D2 TCR T cells (Extended Data Fig. 9b), in the 2D2 mice. We next induced EAE in mice co-colonized with OTU0002 and L. reuteri wild type or L. reuteri ΔuvrA. Deletion of the uvrA gene did not affect the abundance of L. reuteri in the small intestine (Extended Data Fig. 9c). EAE incidence tended to be lower (although not significantly), and EAE symptoms were significantly less severe in the mice co-colonized with L. reuteri ΔuvrA compared to those with the L. reuteri wild type (Fig. 3h, i). TH17 cells in the small intestine were comparable between mice co-colonized with L. reuteri wild type or L. reuteri ΔuvrA strain (Extended Data Fig. 9d). However, Ki67+CD4+ cells were significantly decreased in mice colonized with L. reuteri ΔuvrA (Fig. 3j), which supports the hypothesis that UvrA expressed in L. reuteri cross-reacts with MOG-specific T cells, and that signals from OTU0002 increase the pathogenesis of those proliferated cells (Extended Data Fig. 10).

Our study emphasizes the necessity of considering the synergic effects of gut microorganisms on autoimmune diseases, and suggests that the expansion and education of autoreactive T cells by microorganisms in the small intestine may have a critical role during the development of EAE. Nevertheless, there is no direct evidence to support the notion that these small-intestinal MOG-reactive T cells themselves migrate into the central nervous system to exacerbate EAE, and further studies are required to explore this—as demonstrated in models of other autoimmune diseases15,16,17. As patients with multiple sclerosis show increased TH17 cells in the small intestine1, it may be worth investigating the microbiota of the small intestine—which can be captured only by analyses of endoscopic samples or biopsies, and not from faecal samples—for further studies in humans.

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized and investigators were not blinded to allocation during experiments and outcome assessment.

Mice

C57BL/6 mice were purchased from CLEA Japan, and acclimatized for one week in the specific-pathogen-free facility at the RIKEN Yokohama Campus before being used for experiments. Germ-free C57BL/6 mice were originally purchased from Sankyo Labo Service and CLEA Japan, and bred in the germ-free facility of the RIKEN Yokohama Campus. The 2D2 TCR transgenic mice (C57BL/6-Tg(Tcra2D2,Tcrb2D2)1Kuch/J) were obtained from The Jackson Laboratory and bred in the specific-pathogen-free facility. Both male and female 2D2 mice at 6–8 weeks of age were used for experiments. Mice were randomly allocated into each treatment group. No statistical estimations were performed to determine sample size. For antibiotic treatment, mice were given ampicillin (0.1 g/l), vancomycin (0.5 g/l), neomycin (1 g/l) or metronidazole (1 g/l) in their drinking water. In some experiments, mice were given a cocktail of these antibiotics (ampicillin 1 g/l, vancomycin 0.5 g/l, neomycin 1 g/l and metronidazole 1 g/l) in their drinking water to deplete intestinal microorganisms. All experimental procedures were approved by, and performed in accordance with, the Institutional Animal Care and Use Committee of the RIKEN Yokohama Campus.

Bacteria

To isolate both the OTU0001 (L. reuteri) and OTU0002 bacteria, the contents of the small intestine from naive C57BL/6 mice were homogenized and diluted with PBS. The dilutions were plated on GAM agar plates (Nissui) and incubated anaerobically in an Anaero-Pack (Mitsubishi) at 37 °C for 3 days. For isolation of the OTU0002 bacterium, GAM agar plates were supplemented with colistin (8 μg/ml, Wako) and novobiocin (4 μg/ml, Wako). The 16S rRNA genes from individual colonies were amplified using the 27F and 1492R primers, and fragments, including the V4 region, were sequenced using an Applied Biosystems 3130xl Genetic Analyzer (Thermo Fisher) to determine the colonies corresponding to Lreuteri and OTU0002. The isolated bacteria were grown in GAM broth (Nissui) and stored in 20% glycerol at −80 °C.

Gnotobiotic mice

Female germ-free C57BL/6 mice at 5–7 weeks of age were gavaged with overnight cultures of L. reuteri or OTU0002 bacteria. To colonize SFB, germ-free mice were gavaged with faeces from SFB-monocolonized mice. All gnotobiotic mice were kept in isolators for 2–4 weeks before being used for experiments.

EAE induction

For induction of active EAE, female mice (9–12 weeks old) were immunized subcutaneously at 2 sites on the back with 200 μg of MOG35–55 emulsified in incomplete Freund’s adjuvant (Difco) supplemented with 5 mg/ml heat-killed Mycobacterium tuberculosis (Difco). Mice were also injected intraperitoneally with 400 ng of pertussis toxin (PTX) (List Biological Laboratories) on days 0 and 2 after immunization. EAE clinical scores were evaluated by a blinded observer as follows; 0, no sign of symptoms; 0.5, partial limp tail; 1, complete limp tail; 2, limp tail and weakness of hind limbs; 3, limp tail and complete paralysis of hind limbs; 4, limp tail, complete paralysis of hind limbs, and weakness of front limbs; 5, minimal movement owing to paralysis. The mice given a score of 5 were euthanized at this point, and a score of 5 was given for these mice for the rest of the experiments.

Luxol fast blue staining of spinal cords

Demyelination of spinal cords was assessed at day 20 after immunization. Mice were anaesthetized and perfused with PBS followed by 4% paraformaldehyde (PFA). Cervical spinal cords were post-fixed in 4% PFA at 4 °C overnight, embedded in paraffin and sectioned at 8-μm thickness. The sections were incubated with 0.1% luxol fast blue solution at 56 °C overnight and differentiated in 0.05% lithium carbonate solution for 5 min followed by rinsing in 70% ethanol for 30 s. When differentiation was completed, the sections were counterstained with 0.1% cresyl violet solution for 5 min, differentiated and dehydrated by 95% and 100% ethanol, and rinsed in xylene. The demyelinated regions in the white matter were quantified using ImageJ (v.1.8.0, NIH), and the percentages of demyelinated areas in the total white matter areas were calculated.

Preparation of lymphocytes

Single cell suspensions from spleen, mesenteric lymph nodes, Peyer’s patches and inguinal lymph nodes were prepared by mechanical disruption on a 70-μm cell strainer. Spleen suspensions were incubated in red blood cell lysis buffer (BioLegend). To isolate lymphocytes from the spinal cord, the tissues from PBS-perfused mice were minced and incubated in RPMI 1640 supplemented with 2% fetal bovine serum (FBS) (Biological Industries), 400 U/ml collagenase D (Roche) and 20 μg/ml DNase I (Wako) for 45 min at 37 °C with agitation. The digested tissues were suspended in 40% Percoll (GE Healthcare) and overlaid onto 70% Percoll followed by centrifugation at 800g for 20 min at 20 °C. Lymphocytes at the interface were collected. Isolation of lamina propria lymphocytes was performed as previously described18,19. In brief, intestines were incubated in Hank’s balanced salt solution supplemented with 2% FBS, 1 mM dithiothreitol and 20 mM EDTA for 30 min at 37 °C with agitation, and then the epithelial layer was removed by vigorous shaking in PBS. The remaining tissues were minced and incubated in RPMI 1640 supplemented with 2% FBS, 400 U/ml collagenase D, 0.25 U/ml dispase (BD Biosciences) and 0.1 mg/ml DNase I for 30 min at 37 °C with agitation. Lamina propria lymphocytes were collected from the 40%/80% Percoll interphase.

ELISA

Single-cell suspensions (1 or 2 × 106 cells per ml) were cultured in the presence or absence of MOG35–55 (50 μg/ml) for 3 days. In some experiments, SAA (10 μg/ml) was added in combination with MOG35–55. The concentrations of cytokines were measured by ELISA according to the manufacturer’s instructions.

Flow cytometry

For intracellular cytokine staining, the cells were stimulated with 25 ng/ml phorbol myristate acetate (PMA; Sigma) and 1 μg/ml ionomycin (Sigma) in the presence of brefeldin A (BioLegend) for 4 h at 37 °C. Dead cells were stained with Zombie Aqua dye (BioLegend). After Fc receptors were blocked with anti-CD16/32 (BD Biosciences), the cells were stained with anti-CD3 (APC–Cy7 or PE–Cy7), anti-CD4 (APC or PerCP–Cy5.5), anti-Vα3.2 (PE), anti-Vβ11 (Alexa 647), anti-CD44 (APC–Cy7) and anti-CD62L (Pacific Blue) antibodies. For intracellular staining, the cells were fixed and permeabilized with the FOXP3 staining buffer set (Thermo Fisher Scientific) and stained with anti-IL-17A (PE–Cy7), anti-IFN-γ (FITC), anti-FOXP3 (PE) and anti-Ki67 (FITC) antibodies. All data were collected on a FACS Canto II cytometer (BD Biosciences) and analysed using FlowJo (v.10.2, Tree Star).

Co-culturing

After Fc receptors were blocked with anti-CD16/32, small-intestinal lamina propria cells were stained with anti-TCRβ (FITC), anti-CD4 (APC) and anti-CD8α (PE) antibodies. CD4+ and CD8+ T cells were sorted on a FACS Aria III (BD Biosciences). Splenic CD11c+ cells were enriched with anti-CD11c beads (Miltenyi) according to manufacturer’s instructions. The sorted T cells (5 × 104 cells per well) and CD11c+ cells (5 × 104 cells per well) were co-cultured in round-bottom 96-well plates in the presence or absence of MOG35–55 (50 μg/ml) for 3 days. To block MHC class II, CD11c+ cells were pretreated with anti-MHC class II blocking antibody or its isotype control antibody (Biolegend) for 1 h at 37 °C before co-culturing.

qPCR analysis

The small intestinal tissues and spinal cords were homogenized in RLT buffer (Qiagen) using 3-mm steel beads in a Shake Master Neo (Biomedical Science) for 2–4 min at 1,500 rpm. CD4+ cells were enriched from MOG35–55 and SAA-stimulated splenocytes using CD4 Magnetic Particles (BD Biosciences) and suspended in RLT buffer. Total RNA was extracted using the RNeasy mini kit (Qiagen), and cDNA was obtained with RiverTra Ace (TOYOBO). qPCR was performed using SYBR Premix Ex Taq (Takara) on a Thermal Cycler Dice Real Time System (Takara). The primers used are listed in Supplementary Table 2.

16S rRNA gene analysis

Bacterial DNA was isolated according to a previously published method20 with minor modifications. The contents of the small intestine were suspended in PBS and filtered on a 100-μm cell strainer. After washing in PBS, bacterial cells were pelleted and resuspended in TE10 buffer (10 mM Tris-HCl, 10 mM EDTA, pH 8.0) containing lysozyme (15 mg/ml; Wako). The suspensions were incubated for 1 h at 37 °C with mixing at 1,000 rpm, followed by incubation with purified achromopeptidase (2,000 units per ml) (Wako) for 30 min at 37 °C. Subsequently, the samples were further incubated with 1% sodium dodecyl sulfate and proteinase K (1 mg/ml) (Merck) for 1 h at 55 °C. DNA was extracted with phenol:chloroform:isoamyl alcohol (25:24:1) and treated with RNase A (0.1 mg/ml) (Nippon Gene) for 30 min at 37 °C. High-molecular-weight DNA was precipitated by polyethylene glycol 8000 and washed with 70% ethanol. The isolation of DNA from luminal and mucosa-associated bacteria was performed as previously reported21. After collection of luminal contents, the tissues of the small intestine were washed with PBS three times. The tissues were shaken vigorously in 0.1% Tween 80 in PBS twice to release mucosa-associated bacteria, and the cell suspensions were centrifuged for 5 min at 10,000g. The bacterial DNA was extracted and purified. The total copy number of 16S rRNA genes was quantified by qPCR using the 340F/514R universal primers (Supplementary Table 2). Genomic DNA from Escherichia coli K-12 was used to generate the standard curve.

For 16S rRNA gene sequencing, the V4 variable region was amplified by PCR with dual barcoded primers as previously described22. The amplicons were purified using AMPure XP (Beckman Coulter) and quantified using the Quant-iT PicoGreen ds DNA Assay Kit (Thermo Fisher Scientific). The pooled amplicons were further qualified and quantified by a Bioanalyzer 2100 with the High Sensitivity DNA Kit (Agilent) and the KAPA Library Quantification Kit for Illumina (Kapa Biosystems). The denatured amplicons were mixed with 20% PhiX Control v.3 and sequenced on a MiSeq (Illumina, 2 × 250-bp paired-end reads). The 16S rRNA reads were processed with Mothur (v.1.36.1)23 following the mothur MiSeq standard operating procedure22. After trimming low-quality and chimeric reads, the aligned reads were taxonomically classified against the Ribosomal Database Project database (trainset9_032012.pds), and non-bacterial reads were removed. The remaining reads were clustered into 97% identity OTUs. Low-abundance OTUs (less than 0.01%) were filtered out, and the resulting OTU table was rarefied to 10,000 reads per sample. The OTU corresponding to SFB (Fig. 3b) was determined using the RDP SeqMatch tool (https://rdp.cme.msu.edu). Principal coordinate analysis of unweighted UniFrac distances and the Adonis test (1,000 permutations) were performed in QIIME (v.1.8.0)24 and the R package vegan (v.2.4.4). Indicator species analysis was performed using the R package Indicspecies (v.1.7.6)25 and the significance of indicator values (IndVal.g) was assessed with 1,000 permutations. Multiple sequence alignment of 16S rRNA gene sequences was performed by MAFFT (v.7.305b), and phylogenetic trees were constructed using FastTree (v.2.1.10)26.

SEM

After removal of intestinal contents, the tissues of the small intestine were washed with PBS and fixed in 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 2 h. The samples were post-fixed in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer for 1.5 h and dehydrated with increasing concentrations of ethanol. The dried samples were coated with a gold layer using MSP-1S (Shinku Device) and observed by SEM (VE-7800, KEYENCE).

Genome analysis

The L. reuteri and OTU0002 strains were cultured in GAM broth overnight, and 3–5 μg of genomic DNA was extracted and purified as described in ‘16S rRNA gene analysis’. The DNA was sonicated with a Covaris S220 (Covaris) to generate fragments (approximately 400 bp). Libraries with unique indexes were prepared using the TruSeq Nano DNA Library Preparation Kit (Illumina) according to the manufacturer’s instructions. The libraries were qualified and quantified by an Agilent 2200 TapeStation system with the High-Sensitivity D1000 ScreenTape (Agilent) and the NEBNext Library Quant Kit for Illumina (New England BioLabs). The pooled library was mixed with PhiX Control and sequenced on an Illumina MiSeq sequencer using MiSeq Reagent Kit v.2.

After trimming index, adaptor and PhiX sequences, the reads were quality-trimmed using Sickle (v.1.33)27. The resulting reads were assembled using SPAdes (v.3.9.0)28. Annotation of the assemblies was performed using Prokka (v.1.11)29. The virulence factors in the OTU0002 genome were retrieved from the core dataset in VFDB (http://www.mgc.ac.cn/VFs/) using BLASTP. The thresholds used were an E value of 1× 10−5 and a bit score of 100 (Supplementary Table 1).

Shotgun sequencing of the contents of the small intestine

Metagenomic libraries were constructed using the Illumina TruSeq kit according to the manufacturer’s protocol. Sequencing was performed using an Illumina MiSeq sequencer using MiSeq Reagent Kit v.3. The demultiplexed reads were interleaved using the interleave-reads.py script from khmer (v.1.4.1)30, and then low-quality reads were removed using the fastq_quality_filter script from the FASTX Toolkit (v.0.0.13). The filtered reads (quality score >30 in >50% of the sequence) were processed using extract-paired-reads.py in khmer to remove orphaned reads. For de novo assembly, the optimal k-mer length was determined using KmerGenie (v.1.7.038)31. The samples from the control group (Con in Extended Data Fig. 5a) were assembled using Velvet (v.1.2.10)32 and MetaVelvet (v.1.2.02)33. We used the following parameters; -ins_length 450 for velveth, -ins_length 450 -exp_cov auto -cov_cutoff auto -min_contig_lgth 200 for velvetg, and -ins_length 450 -exp_cov auto -cov_cutoff auto -min_contig_lgth 200 -scaffolding yes for the meta-velvetg script. The contigs were annotated using Prokka (v.1.12-beta), and bacterial gene contents among the samples were summarized using Roary (v.3.6.0)34. Proteins containing the peptide sequence XRXXFXRVX (in which X denotes any residue) were searched for using an in-house Perl script. The fitted proteins were blasted against the NCBI Microbial Proteins database at the NCBI website (http://www.ncbi.nlm.nih.gov/blast) to predict which bacteria contain these proteins. Gene sequences of the fitted proteins observed in all five control samples were used as references for mapping, and were indexed using bowtie2-build in Bowtie2 (v.2.3.1)35. The quality-filtered reads of each sample were mapped against the reference set using Bowtie2, allowing for one mismatch. The counts of the mapped reads were summarized using SAMtools (v.0.1.19)36 following the EDAMAME course materials (http://www.edamamecourse.org).

Generation of uvrA-deficient L. reuteri strain

A targeting vector with the erythromycin-resistance gene (ermAM) was constructed as previously described37. The 5′ and 3′ homology regions (2.2 kb and 3.6 kb, respectively) of the uvrA gene were amplified from L. reuteri genomic DNA by PCR with the 5-arm and 3-arm primers, respectively (Supplementary Table 2). A 1.1-kb ermAM cassette was amplified from plasmid pIL253 with the ermAM primers (Supplementary Table 2). The purified PCR products were assembled into a linearized pUC19 vector using the In-Fusion HD cloning kit (Takara). The constructs were transformed into DH5α competent cells (Toyobo) and plated on LB agar containing ampicillin (50 μg/ml). Ten individual colonies were screened by colony PCR using the cloning primers. The purified targeting vector was introduced into L. reuteri by electroporation at 2.5 kV, 25 μF and 400 Ω in a 0.2-cm cuvette using a Gene Pulser Xcell (BioRad). The transformants were recovered in MRS broth containing erythromycin (5 μg/ml) for 2 days and then plated on MRS agar containing erythromycin. Ten individual colonies were screened for homologous recombination events by colony PCR using uvrA- and ermAM-specific primers (Supplementary Table 2).

The absence of random integration of the targeting vector was confirmed by Southern blotting. Genomic DNA from wild-type and uvrA-deficient L. reuteri was digested with NdeI or SpeI (New England BioLabs), separated on a 0.8% agarose gel and then transferred to a nylon membrane Hybond N+ (GE Healthcare). The probes indicated in Extended Data Fig. 8a were amplified from the L. reuteri genome and the plasmid pIL253 by PCR using the primers listed in Supplementary Table 2. The probes were labelled with digoxigenin–dUTP, hybridized and detected using the DIG High Prime DNA Labelling and Detection Starter Kit II (Roche) according to the manufacturer’s instructions.

Statistical analyses

Statistical analyses were performed using the Prism software suite (v.6.0e) and R (v.3.2.4 revised). To compare two groups of data, statistical differences were evaluated using the two-tailed unpaired Student’s t-test (for normally distributed variables) or the two-tailed unpaired Mann–Whitney test (non-normally distributed variables). Differences between more than three groups were tested using one-way ANOVA followed by Tukey’s multiple comparisons test (for normally distributed variables) or the Kruskal–Wallis test followed by Dunn’s multiple comparison test (for non-normally distributed variables). Two-way repeated ANOVA followed by Bonferroni’s multiple comparisons test was used if needed. For EAE incidence, the two-tailed log–rank (Mantel–Cox) test was performed. Grubbs test (α = 0.05) was performed, and significant outliers were removed in Extended Data Fig. 9a. P values < 0.05 were considered significant (*P < 0.05, **P < 0.01 and ***P < 0.001).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.

Data availability

The 16S rRNA sequence data, shotgun sequence data, and genome sequence data of L. reuteri and OTU0002 are available from the DDBJ Sequence Read Archive under DDBJ BioProject identifiers PRJDB6615, PRJDB6620, and PRJDB6618 and PRJDB6619, respectively. Source data are provided with this paper.

References

  1. 1.

    Cosorich, I. et al. High frequency of intestinal TH17 cells correlates with microbiota alterations and disease activity in multiple sclerosis. Sci. Adv. 3, e1700492 (2017).

    Article  ADS  Google Scholar 

  2. 2.

    Yokote, H. et al. NKT cell-dependent amelioration of a mouse model of multiple sclerosis by altering gut flora. Am. J. Pathol. 173, 1714–1723 (2008).

    CAS  Article  Google Scholar 

  3. 3.

    Berer, K. et al. Commensal microbiota and myelin autoantigen cooperate to trigger autoimmune demyelination. Nature 479, 538–541 (2011).

    CAS  Article  ADS  Google Scholar 

  4. 4.

    Miyake, S. et al. Dysbiosis in the gut microbiota of patients with multiple sclerosis, with a striking depletion of species belonging to Clostridia XIVa and IV clusters. PLoS ONE 10, e0137429 (2015).

    Article  Google Scholar 

  5. 5.

    Jangi, S. et al. Alterations of the human gut microbiome in multiple sclerosis. Nat. Commun. 7, 12015 (2016).

    CAS  Article  ADS  Google Scholar 

  6. 6.

    Chen, J. et al. Multiple sclerosis patients have a distinct gut microbiota compared to healthy controls. Sci. Rep. 6, 28484 (2016).

    CAS  Article  ADS  Google Scholar 

  7. 7.

    Ochoa-Repáraz, J. et al. Role of gut commensal microflora in the development of experimental autoimmune encephalomyelitis. J. Immunol. 183, 6041–6050 (2009).

    Article  Google Scholar 

  8. 8.

    Cox, L. M. et al. Altering the intestinal microbiota during a critical developmental window has lasting metabolic consequences. Cell 158, 705–721 (2014).

    CAS  Article  Google Scholar 

  9. 9.

    Ivanov, I. I. et al. Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell 139, 485–498 (2009).

    CAS  Article  Google Scholar 

  10. 10.

    Atarashi, K. et al. Th17 cell induction by adhesion of microbes to intestinal epithelial cells. Cell 163, 367–380 (2015).

    CAS  Article  Google Scholar 

  11. 11.

    Sano, T. et al. An IL-23R/IL-22 circuit regulates epithelial serum amyloid A to promote local effector Th17 responses. Cell 163, 381–393 (2015).

    CAS  Article  Google Scholar 

  12. 12.

    Jain, R. et al. Interleukin-23-induced transcription factor Blimp-1 promotes pathogenicity of T helper 17 cells. Immunity 44, 131–142 (2016).

    CAS  Article  Google Scholar 

  13. 13.

    Horai, R., Sen, H. N. & Caspi, R. R. Commensal microbiota as a potential trigger of autoimmune uveitis. Expert Rev. Clin. Immunol. 13, 291–293 (2017).

    CAS  Article  Google Scholar 

  14. 14.

    Petersen, T. R. et al. Characterization of MHC- and TCR-binding residues of the myelin oligodendrocyte glycoprotein 38–51 peptide. Eur. J. Immunol. 34, 165–173 (2004).

    CAS  Article  Google Scholar 

  15. 15.

    Krebs, C. F. et al. Autoimmune renal disease is exacerbated by S1P-receptor-1-dependent intestinal Th17 cell migration to the kidney. Immunity 45, 1078–1092 (2016).

    CAS  Article  Google Scholar 

  16. 16.

    Morton, A. M. et al. Endoscopic photoconversion reveals unexpectedly broad leukocyte trafficking to and from the gut. Proc. Natl Acad. Sci. USA 111, 6696–6701 (2014).

    CAS  Article  ADS  Google Scholar 

  17. 17.

    Benakis, C. et al. Commensal microbiota affects ischemic stroke outcome by regulating intestinal γδ T cells. Nat. Med. 22, 516–523 (2016).

    CAS  Article  Google Scholar 

  18. 18.

    Shimokawa, C. et al. Mast cells are crucial for induction of group 2 innate lymphoid cells and clearance of helminth infections. Immunity 46, 863–874 (2017).

    CAS  Article  Google Scholar 

  19. 19.

    Furusawa, Y. et al. Commensal microbe-derived butyrate induces the differentiation of colonic regulatory T cells. Nature 504, 446–450 (2013).

    CAS  Article  ADS  Google Scholar 

  20. 20.

    Okai, S. et al. High-affinity monoclonal IgA regulates gut microbiota and prevents colitis in mice. Nat. Microbiol. 1, 16103 (2016).

    CAS  Article  Google Scholar 

  21. 21.

    Gong, J. et al. 16S rRNA gene-based analysis of mucosa-associated bacterial community and phylogeny in the chicken gastrointestinal tracts: from crops to ceca. FEMS Microbiol. Ecol. 59, 147–157 (2007).

    CAS  Article  Google Scholar 

  22. 22.

    Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K. & Schloss, P. D. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl. Environ. Microbiol. 79, 5112–5120 (2013).

    CAS  Article  Google Scholar 

  23. 23.

    Schloss, P. D. et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl. Environ. Microbiol. 75, 7537–7541 (2009).

    CAS  Article  Google Scholar 

  24. 24.

    Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336 (2010).

    CAS  Article  Google Scholar 

  25. 25.

    De Cáceres, M. & Legendre, P. Associations between species and groups of sites: indices and statistical inference. Ecology 90, 3566–3574 (2009).

    Article  Google Scholar 

  26. 26.

    Price, M. N., Dehal, P. S. & Arkin, A. P. FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol. Biol. Evol. 26, 1641–1650 (2009).

    CAS  Article  Google Scholar 

  27. 27.

    Joshi, N. & Fass, J. Sickle: a sliding-window, adaptive, quality-based trimming tool for FastQ files (v.1.33), https://github.com/najoshi/sickle (2011).

  28. 28.

    Nurk, S. et al. Assembling single-cell genomes and mini-metagenomes from chimeric MDA products. J. Comput. Biol. 20, 714–737 (2013).

    MathSciNet  CAS  Article  Google Scholar 

  29. 29.

    Seemann, T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 30, 2068–2069 (2014).

    CAS  Article  Google Scholar 

  30. 30.

    Crusoe, M. R. et al. The khmer software package: enabling efficient nucleotide sequence analysis. F1000Res. 4, 900 (2015).

    Article  Google Scholar 

  31. 31.

    Chikhi, R. & Medvedev, P. Informed and automated k-mer size selection for genome assembly. Bioinformatics 30, 31–37 (2014).

    CAS  Article  Google Scholar 

  32. 32.

    Zerbino, D. R. & Birney, E. Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res. 18, 821–829 (2008).

    CAS  Article  Google Scholar 

  33. 33.

    Namiki, T., Hachiya, T., Tanaka, H. & Sakakibara, Y. MetaVelvet: an extension of Velvet assembler to de novo metagenome assembly from short sequence reads. Nucleic Acids Res. 40, e155 (2012).

    CAS  Article  Google Scholar 

  34. 34.

    Page, A. J. et al. Roary: rapid large-scale prokaryote pan genome analysis. Bioinformatics 31, 3691–3693 (2015).

    CAS  Article  Google Scholar 

  35. 35.

    Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012).

    CAS  Article  Google Scholar 

  36. 36.

    Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).

    Article  Google Scholar 

  37. 37.

    Morita, H. et al. Comparative genome analysis of Lactobacillus reuteri and Lactobacillus fermentum reveal a genomic island for reuterin and cobalamin production. DNA Res. 15, 151–161 (2008).

    CAS  Article  ADS  Google Scholar 

  38. 38.

    Gevers, D. et al. The treatment-naive microbiome in new-onset Crohn’s disease. Cell Host Microbe 15, 382–392 (2014).

    CAS  Article  Google Scholar 

Download references

Acknowledgements

We thank T. Kanaya, T. Kato and T. Takeuchi for their technical support. E.M. is supported by the RIKEN Special Postdoctoral Researcher Program. This work was supported in part by the RIKEN Interdisciplinary Research Program ‘Integrated Symbiology’, the RIKEN Pioneering Project ‘Biology of Symbiosis’, Grants-in-Aid for Young Scientists (B) (26850090 to E.M.) and Scientific Research (A) (19H01030 to H.O.), AMED-CREST (19gm0710009h0006 to H.O.) and the Food Science Institute Foundation (to H.O.).

Author information

Affiliations

Authors

Contributions

E.M. and H.O. conceived the study. E.M. designed and performed the experiments and analyses, and cowrote the manuscript. S.-W.K. and T.D.T. contributed to analysis of the bacterial genome. M.K. and N.T.-A. contributed to 16S rRNA and bacterial genome sequencing. S.O. prepared gnotobiotic mice and helped with the experiments. W.S. and M.H. performed shotgun sequencing. H.M. contributed to generating the gene-deficient bacterial mutant. H.O. directed the research and cowrote the manuscript.

Corresponding author

Correspondence to Hiroshi Ohno.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature thanks Martin Blaser, Vijay Kuchroo and Harmut Wekerle for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Suppression of MOG-specific immune responses by oral treatment with antibiotics.

a, Schematic of treatment of EAE mice with antibiotics (ABX). PTX, pertussis toxin. b, EAE clinical scores for mice given normal (control, con) drinking water, or drinking water containing an antibiotic cocktail (ampicillin, vancomycin, neomycin and metronidazole (AVNM))) (n = 5). c, EAE clinical scores for mice given ampicillin in their drinking water (amp-oral) or by daily intraperitoneal (i.p.) injections (amp-i.p.) (n = 5). Asterisks indicate a significant difference between the orally treated group and the intraperitoneally treated group. d, Mice were killed on day 10, and splenocytes were restimulated with or without MOG35–55. After culturing for three days, the concentrations of IFNγ and IL-17 were measured by ELISA (d, n = 3). e, Small-intestinal CD4+ or CD8+ T cells from naive mice (naive), EAE mice (con) and ampicillin-treated EAE mice (amp) were co-cultured with splenic CD11c+ cells from naive mice in the presence or absence of MOG35–55 for three days. The cytokine concentrations in the supernatants were measured by ELISA (n = 4). ND, not detectable. f, Percentage and absolute numbers of TH1 and TH17 cells in the small-intestinal lamina propria (n = 5 mice). gk, Representative FACS plots (gated on CD3+CD4+ cells) and summary data of FOXP3+CD4+ regulatory T cells in the small-intestinal lamina propria (g; n = 4 mice) and other tissues (hj, n = 4 mice; k, n = 5 mice). Data are mean ± s.d. Detailed statistics of the EAE clinical scores are summarized in Supplementary Tables 7, 8. ***P < 0.001, **P < 0.01, *P < 0.05. Two-way ANOVA with Bonferroni’s (b) or Tukey’s (c) test, or one-way ANOVA with Tukey’s test (dk). Exact P values are in Source Data.

Source data

Extended Data Fig. 2 Bacterial loads and composition of the small-intestine microbiota, and phylogeny of OTU0002.

a, Bacterial loads in the contents of the small intestine, as determined by qPCR of 16S rRNA (n = 5 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. One-way ANOVA with Tukey’s test. Exact P values are in Source Data. b, Phylum-level relative abundance of small-intestine microbiota. c, Principal coordinate analysis plot of unweighted UniFrac distances. Each ellipse shows an 80% confidence interval (n = 5 mice). PC1, PC2, principle coordinates 1 and 2. d, Phylogenetic tree based on 16S rRNA gene sequences of OTU0002 and deposited isolates. e, Phylogenetic tree of OTU0002 and related uncultured bacteria. The bacterial reads from human faeces (human faeces 1 and 2) were retrieved from ref. 38 and ref. 6, respectively. f, Representative SEM images of the caecum and colon from OTU0002-monocolonized mice (n = 4).

Source data

Extended Data Fig. 3 Modulation of TH17-cell function by SAA.

a, Splenocytes from germ-free mice inoculated with L. reuteri or OTU0002 were cultured in the presence or absence of MOG35–55 for three days. The cytokine concentrations in the supernatants were measured by ELISA (n = 5 mice). b, Percentage and absolute numbers of TH17 cells in the spleen (n = 5 mice). c, f, mRNA expression of the indicated genes in tissues of the small intestine (n = 5 mice). d, e, Splenocytes from specific-pathogen-free EAE mice were restimulated with MOG35–55 in the presence or absence of SAA. d, After culturing for two days, CD4+ cells were enriched by magnetic beads and the mRNA expression of the indicated genes was quantified by qPCR (n = 4 mice). e, The cytokine concentrations in the supernatants were measured by ELISA (n = 4 mice). g, Representative FACS plots (gated on CD3+ and CD4+ cells), percentage and absolute numbers of TH17 cells in the small-intestinal lamina propria of naive mice (germ-free, n = 6 mice; L. reuteri and OTU0002, n = 5 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. Two-way ANOVA with Bonferroni’s test (a), one-way ANOVA with Tukey’s test (bg) or Kruskal–Wallis with Dunn’s test (b, number of IL-17A+CD4+ T cells; c, Saa1, Saa2 and Il12b; d, Rorc and Il17a; f, Csf2; g, number of IL-17A+CD4+ T cells). Exact P values are in Source Data.

Source data

Extended Data Fig. 4 Microbiota-dependent proliferation of MOG-specific 2D2 TCR T cells in the small intestine.

a, b, Representative FACS plots (gated on CD3+ cells) and percentage of Ki67+CD4+ T cells and absolute number of CD4+ T cells in the small-intestinal lamina propria (a, n = 6 mice) and spleen (b, n = 4 mice) of wild-type and 2D2 mice. c, Percentage of Ki67+CD4+ T cells and absolute numbers of CD4+ T cells in the small-intestinal lamina propria of non-treated or AVNM-treated (+AVNM) 2D2 mice (n = 4 mice). d, Percentage of Ki67+ in 2D2 TCR (Vα3.2+Vβ11+) cells (n = 4 mice). Data are mean ± s.d. P values were calculated using two-tailed unpaired t-test.

Source data

Extended Data Fig. 5 Induction of EAE by mimicry peptide.

a, Candidate mimicry proteins (matching TCR-binding residues of MOG40–48, shown in bold) were observed in the contents of the small intestines of control mice. Left, presence or absence of each candidate. The candidates observed in all five mice were used as references for the analysis shown in Fig. 3a. b, Splenic CD11c+ cells pretreated with indicated antibodies (Ab) were co-cultured with CD4+ T cells from mesenteric lymph nodes of 2D2 TCR mice in the presence of indicated peptides for four days. Representative FACS plots (gated on CD3+ cells) and summary data are shown (n = 4). c, Mesenteric lymph nodes from wild-type or 2D2 mice were stimulated with indicated peptides for four days. Representative FACS plots (gated on CD3+CD4+cells) and summary data are shown (n = 6). d, CD4+ T cells from the spinal cord of EAE wild-type mice were co-cultured with splenic CD11c+ cells from naive wild-type mice in the presence of indicated peptides for three days. The concentrations of IFNγ and IL-17 were measured by ELISA (n = 6). e, f, Wild-type mice were immunized with MOG38–50, UvrA or vehicle (PBS). EAE clinical scores are shown in e (n = 6). mRNA expression of the indicated genes in the spinal cord is shown in f (MOG, n = 5; PBS and UvrA, n = 6). g, Left, representative images of spinal cord sections stained with luxol fast blue. Right, demyelinated area in the white matter was calculated. Scale bars, 500 μm (g) (PBS, n = 3; MOG and UvrA, n = 4). h, RNA and DNA were extracted from the small intestine, caecum (Cec) and colonic contents (Col) of naive mice, and qPCR was performed using uvrA-specific primers. The RNA/DNA ratios of the uvrA gene are shown (n = 5). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. Two-way ANOVA with Bonferroni’s test (b), Kruskal–Wallis with Dunn’s test (c, d IFNγ, f, h) or one-way ANOVA with Tukey’s test (d IL-17, g). Exact P values are in Source Data.

Source data

Extended Data Fig. 6 Effects of co-colonization with L. reuteri and OTU0002 on TH17 cells.

a, Left, representative images of spinal cord sections from EAE-induced germ-free mice or germ-free mice colonized with the indicated strains, stained with luxol fast blue (left). Scale bars, 500 μm. Right, the demyelinated area in the white matter was calculated. The combined results of two independent experiments are shown (germ-free and OTU0002, n = 8; L. reuteri + OTU0002, n = 6). b, Absolute numbers of CD4+ T cells in the spinal cord (germ-free and OTU0002, n = 8; L. reuteri + OTU0002, n = 6). c, The splenocytes and small-intestinal lamina propria cells were cultured in the presence or absence of MOG35–55 for three days. The cytokine concentrations in the supernatants were measured by ELISA (n = 5 mice). d, Percentage and absolute numbers of FOXP3+CD4+ regulatory T cells in the small-intestinal lamina propria (n = 5 mice). e, mRNA expression of the indicated genes in the small-intestine tissue (n = 5 mice). f, Percentage and absolute numbers of TH17 cells in the spleen (n = 5 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. Kruskal–Wallis with Dunn’s test (a, e), one-way ANOVA with Tukey’s test (b, d, f) or two-way ANOVA with Bonferroni’s test (c). Exact P values are in Source Data.

Source data

Extended Data Fig. 7 Coordinated effects of SFB and L. reuteri on EAE development.

a, The abundance of SFB and L. reuteri in the contents of the small intestine. qPCR with specific primers for SFB and Lactobacillus was performed (n = 5 mice). P value (two-tailed unpaired Mann–Whitney test) for the difference in SFB abundance is shown. b, c, EAE clinical scores (b) and incidence (c) of germ-free mice monocolonized with SFB and co-colonized with SFB and L. reuteri (n = 5). Asterisks indicate a significant difference between the monocolonized group and the co-colonized group. Detailed statistics of the EAE clinical scores are summarized in Supplementary Table 9. d, mRNA expression of the indicated genes in the small-intestine tissue (n = 5 mice). e, Representative FACS plots (gated on CD3+ and CD4+ cells), percentage and absolute numbers of TH17 cells in the small-intestinal lamina propria (n = 5 mice). e, Representative FACS plots (gated on CD3+ cells) and summary data of Ki67+CD4+ T cells in the small-intestinal lamina propria are shown (n = 5 mice). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. Two-way ANOVA with Tukey’s (b) or one-way ANOVA with Tukey’s test (df). EAE incidence was analysed by two-tailed log–rank (Mantel–Cox) test (c). Exact P values are in Source Data.

Source data

Extended Data Fig. 8 Generation of uvrA-deficient L. reuteri.

a, Schematic of the genomic structure of the uvrA gene of L. reuteri (L. reuteri wild type), the targeting vector and the resultant mutant (L. reuteri ΔuvrA) generated by homologous recombination containing the erythromycin-resistance gene (ermAM). The NdeI and SpeI restriction sites are indicated by NI and SI, respectively. b, PCR analysis of genomic DNA from L. reuteri and L. reuteri ΔuvrA using uvrA- and ermAM-specific primers. c, Southern blot analysis of NdeI- and SpeI-digested genomic DNA from L. reuteri and L. reuteri ΔuvrA. Probes A and B, depicted in a, were used to detect genomic fragments present in both strains and only in L. reuteri ΔuvrA, respectively. The experiments were independently repeated two times with similar results (b, c). For gel and blot source data, see Supplementary Fig. 1.

Extended Data Fig. 9 Effect of uvrA-deficient L. reuteri on MOG-specific 2D2 TCR T cells.

a, Germ-free wild-type or 2D2 mice were monocolonized with L. reuteri wild-type strain or a uvrA-deficient strain (ΔuvrA). The percentage and absolute numbers of Ki67+CD4+ T cells are shown (wild-type mice with ΔuvrA, n = 3; wild-type mice and wild-type strain, and 2D2 mice and ΔuvrA strain, n = 4; 2D2 mice and wild-type strain, n = 5). b, Representative FACS plots (gated on CD3+CD4+ cells) and summary data of Ki67+ 2D2 TCR (Vα3.2+Vβ11+) cells in the small-intestinal lamina propria are shown (n = 5 mice). c, d, Germ-free wild-type mice were co-colonized with OTU0002 and a wild-type or uvrA-deficient L. reuteri strain, and EAE was induced. The abundance of L. reuteri and OTU0002 in the contents of the small intestine was quantified by qPCR with specific primers for Lactobacillus and Allobaculum (d) (n = 4 mice). Representative FACS plots (gated on CD3+ and CD4+), percentage and absolute numbers of TH17 cells in the small-intestinal lamina propria are shown (e) (n = 4). Data are mean ± s.d. ***P < 0.001, **P < 0.01, *P < 0.05. One-way ANOVA with Tukey’s test (a) or two-tailed unpaired t-test (b). Exact P values are in Source Data.

Source data

Extended Data Fig. 10 Model of the role of microorganisms of the small intestine in EAE.

Two key microorganisms coordinately activate MOG-specific T cells in the small intestine. Mimicry peptides (UvrA) expressed in L. reuteri (OTU0001) trigger TCR signals, and OTU0002-induced pro-inflammatory factors (SAA and IL-23) increase the pathogenicity of MOG-specific TH17 cells. The activated cells may migrate into the spinal cord and induce demyelination.

Supplementary information

Supplementary Figure 1:

Supplementary Figures Raw gel and blot data from Extended Data Fig. 8b and 8c. Red boxes indicate the cropped area used in Extended Data Fig. 8b and 8c.

Reporting Summary

Supplementary Table 1:

Supplementary Table Predicted virulence factors in OTU0002 genome.

Supplementary Table 2:

Supplementary Table Sequences of primers and probes.

Supplementary Table 3:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Fig. 1a. Data represent the mean ± s.d. (a). P values were calculated with one-way (a) or two-way ANOVA with Tukey’s test (b).

Supplementary Table 4:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Fig. 2d. Data represent the mean ± s.d. (a). P values were calculated with one-way (a) or two-way ANOVA with Tukey’s test (b).

Supplementary Table 5:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Fig. 3d. Data represent the mean ± s.d. (a). P values were calculated with one-way (a) or two-way ANOVA with Tukey’s test (b).

Supplementary Table 6:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Fig. 3h. Data represent the mean ± s.d. (a). P values were calculated with two-tailed unpaired t-test (a) or two-way ANOVA with Bonferroni’s test (b).

Supplementary Table 7:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Extended Data Fig. 1b. Data represent the mean ± s.d. (a). P values were calculated with two-tailed unpaired t-test (a) or two-way ANOVA with Bonferroni’s test (b).

Supplementary Table 8:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Extended Data Fig. 1c. Data represent the mean ± s.d. (a). P values were calculated with one-way (a) or two-way ANOVA with Tukey’s test (b).

Supplementary Table 9:

Supplementary Table Detailed statistics of the EAE clinical scores shown in Extended Data Fig. 7b. Data represent the mean ± s.d. (a). P values were calculated with one-way (a) or two-way ANOVA with Tukey’s test (b).

Source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Miyauchi, E., Kim, SW., Suda, W. et al. Gut microorganisms act together to exacerbate inflammation in spinal cords. Nature 585, 102–106 (2020). https://doi.org/10.1038/s41586-020-2634-9

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing