IL-17 controls central nervous system autoimmunity through the intestinal microbiome

See allHide authors and affiliations

Science Immunology  05 Feb 2021:
Vol. 6, Issue 56, eaaz6563
DOI: 10.1126/sciimmunol.aaz6563

Gut reaction to IL-17

Whether IL-17A/F or TH17 cells are necessary for the immunopathogenesis of human multiple sclerosis (MS) has been a topic of active debate. Using experimental autoimmune encephalomyelitis (EAE), a mouse model of MS, in mice deficient for IL-17A and IL-17F, Regen et al. found that IL-17 ablation reduced susceptibility to EAE. However, adoptive transfer of TH17 cells from autoantigen-primed IL-17–deficient mice recapitulated disease in naïve control mice but not in IL-17–deficient mice. Cohousing IL-17–deficient mice with IL-17–expressing mice, fecal transplantation from IL-17–expressing mice into IL-17–deficient mice, and reexpression of IL-17A in the gut of IL-17–deficient mice restored susceptibility to EAE. Thus, IL-17 expression in the gut rather than its effector function in the CNS is directly linked to EAE susceptibility by modulating the gut microbiome.


Interleukin-17A– (IL-17A) and IL-17F–producing CD4+ T helper cells (TH17 cells) are implicated in the development of chronic inflammatory diseases, such as multiple sclerosis and its animal model, experimental autoimmune encephalomyelitis (EAE). TH17 cells also orchestrate leukocyte invasion of the central nervous system (CNS) and subsequent tissue damage. However, the role of IL-17A and IL-17F as effector cytokines is still confused with the encephalitogenic function of the cells that produce these cytokines, namely, TH17 cells, fueling a long-standing debate in the neuroimmunology field. Here, we demonstrated that mice deficient for IL-17A/F lose their susceptibility to EAE, which correlated with an altered composition of their gut microbiota. However, loss of IL-17A/F in TH cells did not diminish their encephalitogenic capacity. Reconstitution of a wild-type–like intestinal microbiota or reintroduction of IL-17A specifically into the gut epithelium of IL-17A/F–deficient mice reestablished their susceptibility to EAE. Thus, our data demonstrated that IL-17A and IL-17F are not encephalitogenic mediators but rather modulators of intestinal homeostasis that indirectly alter CNS-directed autoimmunity.


The interleukin-17 (IL-17) family of cytokines comprises six members, IL-17A to IL-17F. Of those, IL-17A and IL-17F signal through the same receptor complex (IL-17RA/IL-17RC) and share a high degree of homology (1). IL-17A was initially described in antibacterial immunity and is strongly linked to granulopoiesis (24). IL-17A is also implicated as a pathogenicity factor in a number of chronic inflammatory diseases, including multiple sclerosis (MS), arthritis, and psoriasis (5). Successful anti–IL-17A therapy has been demonstrated in patients with plaque psoriasis (6). In animal models, dysregulated IL-17A expression leads to a psoriasis-like skin disease, which can be reversed by IL-17A neutralization (7, 8). Dysregulated IL-17A also induces vascular inflammation associated with severe skin disease (9) and results in severe pancreatitis when overexpressed in intestinal epithelial cells (10), further supporting a role for IL-17A in the human pathology. However, despite massively increased IL-17A serum levels in these models, none of these studies reported possible effects on the brain vasculature or parenchyma.

In rodents, MS can be modeled by experimental autoimmune encephalomyelitis (EAE). For this purpose, genetically susceptible mice are immunized with components of the myelin sheath, the most widely used model being the peptide p35-55 of myelin oligodendrocyte glycoprotein (MOG35-55)–induced disease in mice of the C57BL/6 genetic background (11). The immunized mice develop acute central nervous system (CNS) inflammation with clinical manifestations of ascending paresis and partial paralysis, very similar to the pathology seen in the brain and spinal cord of patients with MS (11). In EAE, IL-17A overexpression by T cells does not alter disease development, despite the fact that these cells enter the CNS and deliver large amounts of the cytokine to the brain and spinal cord parenchyma (12). In contrast, systemic IL-17A deletion has varying effects on EAE development, ranging from essential to redundant, depending on the report (1214), which conflicts with the proposed importance of IL-17(A) in human MS (15).

Here, we found that mice deficient for IL-17A and IL-17F were resistant to EAE induction, although their T cells had full encephalitogenic potential. IL-17A and IL-17F were dispensable for EAE induction but were critical modulators of the intestinal microbiota. These alterations of the intestinal microbiome correlated with EAE resistance of mice deficient for IL-17A and IL-17F. Moreover, through experimental manipulation of the IL-17–sensitive microbiota, we were able to reestablish EAE susceptibility in IL-17–deficient mice. Our data challenge the role of IL-17 as encephalitogenic cytokine in CNS autoimmunity. Instead, our results establish IL-17 as a contributor to intestinal homeostasis, where disturbance of this homeostasis can affect (auto)immunity.


IL-17−/− mice were hyporesponsive to CNS autoimmunity

To investigate the role of IL-17 as encephalitogenic cytokine, we immunized mice lacking the expression of both IL-17A and IL-17F (hereafter referred to as IL-17−/−; figs. S1, A and B, and S2A) (16) with the MOG35-55 peptide and found these animals to be hyposusceptible to EAE induction (Fig. 1A), with less than 50% showing clinical disease symptoms. Those mice with clinical manifestations showed a significantly delayed onset and a relatively mild disease peak with only partial paresis of the lower body, from which they recovered quickly (Fig. 1, A and B; see details on disease scoring in Materials and Methods). Although EAE is a T helper (TH) cell–driven disease (17) and IL-17−/− mice developed few clinical symptoms, we detected no substantial alterations in their general composition of the immune compartment in secondary lymphoid tissues, including the various T cell subsets (Figs. 1, C and D, and fig. S2B), suggesting that IL-17 does not alter the composition of the disease-causing immune cells. We also found no differences in the frequency of peripheral FoxP3+ regulatory T (Treg) cells, arguing against Treg-mediated globally enhanced effector T cell suppression in these mice (figs. S1, A and C, and S2, C to E). Although MOG-responding T cells failed to enter the CNS of IL-17−/− mice in large numbers (figs. S1, A and D, and S2F), the frequency of these cells in the periphery and their production of the disease-relevant cytokines interferon γ (IFNγ) and granulocyte-macrophage colony-stimulating factor (GM-CSF) was undisturbed in IL-17−/− compared with wild-type C57BL/6 (WT) mice (figs. S1, A and E, and S2, G and H). Long-term treatment of IL-17+/− mice with IL-17A–neutralizing antibodies (18) led to delayed EAE (fig. S2I). To more precisely determine the roles of IL-17A/F as encephalitogenic effector cytokines of TH cells, we established adoptive transfer experiments with MOG-primed and TH17-skewed T cells, as previously described (19). TH17 cells isolated from IL-17−/− mice were fully capable of inducing disease when transferred to unimmunized WT recipients (Fig. 1E) but failed to do so when transferred into IL-17−/− host mice (fig. S2J). Together, these findings demonstrated that IL-17 produced by T cells is not needed to establish CNS autoimmune disease and that a mechanism other than the absence of encephalitogenic IL-17 inhibits EAE development in IL-17−/− mice.

Fig. 1 IL-17−/− mice are hyporesponsive to active EAE induction.

(A and B) Mice were immunized with MOG35-55, and the disease course (A) and the relevant disease parameters (B) were monitored. Data are pooled from six individual experiments (at least n = 4 per group per experiment). (C) t-SNE plots depicting the proportions of immune cells within the indicated mice and conditions. Data were obtained using high-dimensional flow cytometry, and cells were isolated from spleens (n = 5 per group). (D) Quantification of the indicated cell subsets based on the analysis from (C). (E) Mice of the indicated genotype (before arrow) were immunized with MOG35-55; and 10 days later, the spleen and lymph nodes were isolated, and single-cell suspensions were in vitro cultured in the presence of MOG antigen and IL-23. After 4 days of culture, 5 × 106 blasting cells were injected into naïve WT mice, and disease was monitored. Data are representative for two individual experiments with similar results (at least n = 5 per group per experiment). Data in (A) and (E) are represented as means ± SEM. Black bars in (B) and (D) represent the mean. P value in (A) was obtained applying two-way ANOVA. P values in (B) and (D) were obtained from applying unpaired, two-tailed Student’s t test. DPI, days post-immunization; AUC, area under the curve; NKT cells, natural killer T cells.

IL-17−/− and littermate IL-17+/− mice harbored distinct gut microbiota

For the experiments described thus far, we kept IL-17−/− mice separated from control animals. However, when IL-17−/− mice were cohoused with WT mice, they displayed significantly enhanced disease susceptibility (Fig. 2A). IL-17A and IL-17F are implicated in intestinal health (20), and IL-17A was originally described to enhance the antimicrobial activity of immune cells (2). Furthermore, alterations to the microbiome can modulate EAE susceptibility (2124). We thus hypothesized that IL-17 loss may disturb intestinal homeostasis, including the composition of the gut microbiota, and subsequently alter the immune system in a manner that leads to the altered EAE susceptibility. Analysis of 16S ribosomal RNA (rRNA) reads obtained from fecal material revealed that the overall microbiota composition significantly differed between IL-17−/− and WT C57BL/6J mice, the latter being externally purchased for this particular experiment (fig. S3A). However, the microbiota of externally bred mice significantly differs depending on the specific vendor (fig. S3B), and, generally, such mice are inappropriate for microbiota comparisons with separately bred mutant mice (2527). We therefore adapted our microbiota-centered experiments to use IL-17+/− littermate control mice, which were fully susceptible to EAE induction (fig. S3C). 16S rRNA analysis of fecal material from IL-17−/− and IL-17+/− littermate mice revealed an overall significantly different microbiota (Fig. 2B) with a number of significant changes at the family level (fig. S3D). We identified 13 bacterial operational taxonomical units (OTUs) that were specifically overrepresented in IL-17−/− mice (Fig. 2C; OTU-specific taxonomic information can be found in table S1). In addition, we detected a mildly increased a-diversity in IL-17−/− mice (fig. S3E). Notably, we found a significant increase in segmented filamentous bacteria (SFB) inhabiting the IL-17−/− small intestine (fig. S3F), indicating that the mere presence of these bacteria, previously shown to induce the generation of pathogenic TH17 cells (28, 29), is not sufficient to promote EAE development in IL-17−/− mice. A nontargeted analysis of serum metabolic features did not detect an overall significantly changed metabolome in IL-17−/− mice and yielded only very few significant changes, of which none could be identified as direct bacterial metabolites (fig. S3G).

Fig. 2 IL-17−/− and IL-17+/− littermate mice harbor distinct gut microbiota.

(A) EAE disease course of WT mice compared with groups of IL-17−/− mice that were housed among IL-17−/− littermates (single) or together with WT mice (cohoused). Data are pooled from three individual experiments (at least n = 3 per group per experiment). Data are represented as means ± SEM, and P values were obtained from applying two-way ANOVA with Tukey’s multiple comparisons test. (B) NMDS analysis of Bray-Curtis distances obtained between samples collected from IL-17−/− and IL-17+/− littermate mice. (C) Heatmap of the indicated single OTUs, whose abundance significantly differs between IL-17−/− and IL-17+/− littermate mice. For (B) and (C), data were obtained from 16S rRNA sequencing of fecal material from IL-17+/− (n = 15) and IL-17−/− (n = 24) mice. P value in (B) was obtained from applying PERMANOVA test. P values for each OTU in (C) were obtained from applying the Wilcoxon rank sum test and FDR. Original values and taxonomic information are shown in table S1.

Transfer of WT microbiota recovered EAE susceptibility in IL-17−/− mice

IL-17A has previously been identified as an important regulator of intestinal health, because its neutralization aggravates dextran sulfate sodium–induced colitis in mice (30) and biological agents that suppress IL-17 signaling worsen inflammatory bowel disease (IBD), which resulted in the termination of clinical trials (31). Patients with plaque psoriasis and ankylosing spondylitis treated with the IL-17A antibody secukinumab also developed IBD (32). These adverse effects of IL-17 blockade might be due to changes in the gut microbiome that render the intestine more permissive to inflammation. Preliminary histological analysis indicated structural alterations to the small intestine of IL-17−/− mice, which included a compromised mucus layer, a higher abundance of bacteria in close proximity to epithelial cells, and an increased rate of apoptotic cells in the epithelial barrier (fig. S4). Together, these data favor a fundamental involvement of IL-17 in shaping the gut microbiota and maintaining intestinal barrier integrity.

On the basis of the observation that cohousing with WT mice resulted in the regain of EAE susceptibility of IL-17−/− mice (Fig. 2A), we analyzed the microbiome of IL-17−/− cohoused with IL-17+/− mice and found a shift in the gut microbial composition of IL-17−/− animals that more closely resembled that of their IL-17+/− littermates (Fig. 3A). Moreover, eight of the above identified 13 OTUs were significantly decreased in the cohoused mice compared with their single-housed littermates (Fig. 3B and table S1). At the peak of disease, we detected no significant difference in EAE severity between cohoused IL-17−/− and WT mice (Fig. 3C). In a similar approach, we reconstituted the microbiota of germ-free (GF) IL-17−/− mice and detected, as expected, an increased EAE susceptibility of ex-GF mice when reconstituted with WT compared with IL-17−/− microbiota (Fig. 3D).

Fig. 3 Transfer of WT microbiota recovers EAE susceptibility in IL-17−/− mice.

(A) NMDS analysis of Bray-Curtis distances obtained between fecal samples collected from IL-17+/− (n = 15), single-housed IL-17−/− (n = 24), and cohoused IL-17−/− (n = 14) mice. P values were obtained from applying PERMANOVA test. (B) OTUs whose abundance significantly differ between single-housed IL-17−/− and littermate IL-17+/− mice as well as between single-housed and cohoused IL-17−/− mice but that do not differ between cohoused IL-17−/− and littermate IL-17+/− mice. Data are shown as means ± SEM. P values were obtained from applying the Wilcoxon rank sum test. (C) Depicted is the clinical score at disease peak of the indicated MOG35-55–immunized mice as shown in Fig. 2A. Mice that did not develop disease were excluded. Data are pooled from four individual experiments (at least n = 3 per group per experiment were immunized). Black bars represent the mean. P values were obtained from applying one-way ANOVA followed by Bonferroni’s multiple comparisons test. (D) EAE disease course of MOG35-55–immunized SPF WT (n = 7), GF IL-17−/− cohoused with SPF WT (n = 4), and GF IL-17−/− cohoused with SPF IL-17−/− (n = 5) mice. Data are shown as means ± SEM. P values were obtained from applying two-way ANOVA with Tukey’s multiple comparisons test. uc, unclassified.

The IL-17−/− microbiota could not confer resistance to EAE in WT mice

In contrast to the partial establishment of the IL-17+/− microbiota in the IL-17−/− animals (Fig. 3A) or the WT microbiota in GF IL-17−/− mice and the subsequent effects to EAE susceptibility (Fig. 3, C and D), we did not observe a reciprocal relationship in terms of ameliorated EAE in WT mice cohoused with IL-17−/− mice (fig. S5A). Likewise, introducing IL-17−/− fecal material into previously antibiotic-treated WT animals did not ameliorate the EAE disease course (fig. S5B). This unchanged EAE phenotype was confirmed by microbiota analysis, where the cohousing of IL-17+/− with littermate IL-17−/− mice failed to establish the previously identified OTUs, representative of the IL-17−/− microbiota, in cohoused IL-17+/− animals (fig. S5C). In an attempt to fully establish the IL-17−/− microbiota in WT-like mice, we made use of GF WT C57BL/6 animals where the microbiota of IL-17−/− donor material does not encounter niche competition from a local microbiota. Cohousing these GF WT with specific pathogen–free (SPF)–housed IL-17−/− mice resulted in full susceptibility to EAE in these ex-GF WT mice (fig. S5D). This notable absence of EAE alteration led us to analyze the fecal microbiota of the ex-GF mice. When we calculated the degree of similarity between the microbiota of ex-GF mice with that of their microbiota-donating IL-17−/− cage mates, there was no significant difference to the comparison of ex-GF with SPF-housed WT animals (fig. S5E). Together, these results suggested that the intestinal flora of IL-17−/− mice was strictly associated with lost IL-17 activity, thus could not be transferred in its entirety to IL-17–sufficient animals.

IL-17A expression in the gut restored EAE susceptibility in IL-17–deficient animals

To conclusively demonstrate that IL-17 was a key factor determining the gut microbiota composition, we tested whether reintroduction of IL-17A in the intestinal compartment of otherwise IL-17−/− mice affected the microbiota composition and EAE susceptibility. We crossed IL-17−/− mice with mice allowing Cre-mediated transgenic expression of IL-17A controlled by the Villin promotor (10), resulting in IL-17A production from intestinal epithelial cells (IL-17−/−/IL-17A–Villtg/+ mice; Fig. 4A). Analysis of green fluorescent protein expression, which coincides with IL-17A expression in these mice, confirmed target gene expression confined to the gut epithelium (fig. S6A). We then investigated whether ectopic IL-17A expression in the gut epithelium could modulate the gut microbiota. The commensal bacterial communities of IL-17−/−/IL-17A–Vill+/+ mice, i.e., mice that do not express Cre, significantly differed from the microbiome of their IL-17−/−/IL-17A–Villtg/+ littermates, confirming the importance of IL-17A in controlling the gut microbiome composition (Fig. 4B). Furthermore, IL-17−/−/IL-17A–Vill+/+ mice cohoused with littermate IL-17−/−/IL-17A–Villtg/+ mice showed a shift in microbiota composition toward that of IL-17A–expressing mice (Fig. 4B and fig. S6B). In accordance with this change, we found several OTUs from different phyla present in single-housed IL-17−/−/IL-17A–Vill+/+ mice but less abundant or even absent in littermate IL-17−/−/IL-17A–Villtg/+ animals or IL-17−/−/IL-17A–Vill+/+ mice that were cohoused with them (Fig. 4C and table S1). Twelve of the previously identified 13 OTUs (see Fig. 2C) were also reduced or absent in IL-17−/−/IL-17A–Villtg/+ mice (Fig. 4D), highlighting the specific microbiota-modulating capacity of IL-17A.

Fig. 4 IL-17A expression in the gut restores microbiota and EAE susceptibility.

(A) Schematic representation of the targeting approach. CAG, chicken beta-actin-based promoter; NEO, neomycin resistance gene cassette; STOP, stop codon; IRES, internal ribosomal entry site; EGFP, enhanced green fluorescent protein. (B and C) NMDS analysis of Bray-Curtis distances obtained between samples (B) and heatmap of single OTUs, whose abundance significantly differed between IL-17−/−/IL-17A–Vill+/+ single-housed mice compared with the two other groups, but did not differ between IL-17−/−/IL-17A–Villtg/+ and cohoused IL-17−/−/IL-17A–Vill+/+ mice (C), of the results obtained from 16S rRNA sequencing of fecal material from IL-17−/−/IL-17A–Villtg/+ (n = 8), single-housed (n = 8), and cohoused (n = 14) IL-17−/−/IL-17A–Vill+/+ mice. P values in (B) were obtained from applying PERMANOVA test. P values for each OTU in (C) were obtained applying the Wilcoxon rank sum test. Original values and taxonomic information are shown in table S1. (D) Difference in the abundance of the 13 OTUs identified in Fig. 2C comparing IL-17−/−/IL-17A–Villtg/+ (n = 8) and single-housed IL-17−/−/IL-17A–Vill+/+ (n = 8) littermate mice. Data are shown as means ± SEM. P values were obtained applying the Wilcoxon rank sum test. (E and F) Indicated mice were immunized with MOG35-55, and the disease course (E) and the relevant disease parameters (F) were monitored. Data are representative for three individual experiments with similar results (at least n = 4 per group per experiment were immunized). Data in (E) are shown as means ± SEM, and P values were obtained from applying two-way ANOVA with Tukey’s multiple comparisons test. Black bars in (F) represent the mean, and P values in (F) were obtained applying one-way ANOVA followed by Bonferroni’s multiple comparisons test.

To determine the functional effects of IL-17A expression in the gut, we tested EAE susceptibility in IL-17−/−/IL-17A–Villtg/+ mice. Although, here, IL-17 expression was excluded from the immune compartment, including expression by TH17, IL-17 producing γδ T cells, and group 3 innate lymphoid cells, IL-17A expression in the gut epithelium fully restored EAE susceptibility (Fig. 4E). As expected, single-housed IL-17−/−/IL-17A–Vill+/+ mice were hyposusceptible to EAE induction, despite the fact that MOG-directed T cell priming was unaltered in both IL-17−/−/IL-17A–Villtg/+ and IL-17−/−/IL-17A–Vill+/+ littermate mice (Fig. 4, E and F; and figs. S1, A and E, and S6C). Moreover, when IL-17−/−/IL-17A–Vill+/+ mice were cohoused with IL-17−/−/IL-17A–Villtg/+ littermates, they regained partial EAE susceptibility (Fig. 4, E and F), resembling the phenotype of IL-17−/− mice that were cohoused with WT animals (Fig. 2A). Together, these data eliminate IL-17A as an encephalitogenic effector cytokine and instead demonstrate that intestinal IL-17A expression shapes the gut flora to restore EAE susceptibility in IL-17−/− mice.


In this study, we showed a critical role for IL-17A and IL-17F in the regulation of the gut microbiota and subsequently the susceptibility of IL-17–deficient mice to EAE. Specifically, cohousing of IL-17−/− mice with IL-17–producing littermates resulted in a shift of the intestinal microbial composition in IL-17–deficient mice, and these alterations were sufficient to reconstitute their ability to develop EAE. Moreover, reintroducing IL-17A to the gut of otherwise IL-17–deficient mice led to similar alterations of their gut microbiota and resulted in full susceptibility to EAE in these mice.

Although IL-17 is linked with chronic inflammatory diseases, the mechanism by which IL-17A/F influences tissue inflammation is a matter of some debate. The loss of IL-17 in mice leads to limited EAE susceptibility ranging from large effects (14) to minor disease resistance (12, 13), suggesting that differences between animal facilities may influence the EAE susceptibility of IL-17−/− mice. This, in turn, suggests that an environmental factor within the different animal facilities may be responsible for this phenomenon. Here, we found that mice lacking IL-17A and IL-17F exhibit a reduced susceptibility to MOG35-55–induced EAE, which was strongly associated with an altered intestinal microbiota. Shifting the microbial composition toward a WT-like commensal gut flora restored disease susceptibility in IL-17−/− mice. The specific reintroduction of IL-17A into the intestinal compartment was sufficient to shift the microbiota and to restore susceptibility to EAE. Together, these findings support an essential role for IL-17 in regulating intestinal homeostasis, which, when disturbed, has direct consequences for CNS autoimmunity.

Concomitantly, ectopic targeting of IL-17A to TH cells did not alter EAE susceptibility (12), in contrast to other cytokines such as GM-CSF, which, when targeted to TH cells, triggered spontaneous neuroinflammation (33). Global overexpression of IL-17A in mice, however, leads to marked inflammation of barrier tissues, more specifically epithelia, which is where the IL-17R complex is most prominently expressed (2, 34, 35). These preclinical findings match observations in human disease, where dysregulation of the IL-23/IL-17 axis is associated with the skin and gut inflammation and, moreover, where targeting of these cytokines has a markedly beneficial clinical effect (5). In MS, however, targeting IL-17A did not deliver a prominent clinical effect, and the drug secukinumab is not used for the treatment of MS.

One obvious limitation of our study is that all experiments were performed in only one specific mouse facility. Although this does not interfere with our general conclusions of the data, it is well possible that a different basic microbiota, as present in other mouse facilities, could result in somewhat different outcomes. We also would like to point out that we did not conclusively investigate the specific contribution of IL-17A versus IL-17F to the gut microbiota. Whereas this question is subject to follow-up experiments, our data on intestinal epithelium–restricted expression of specifically IL-17A point toward a decisive role for this particular cytokine. Future investigation will also amplify our preliminary observations on the influence of IL-17 on intestinal barrier integrity and structural homeostasis and decipher the cellular and molecular mechanisms that link altered intestinal homeostasis in IL-17−/− mice to their hyposusceptibility to EAE.

In conclusion, our data support the notion that IL-17 is primarily a barrier cytokine, vital for the maintenance of microbial homeostasis. Whereas these findings put previous discordant interpretations regarding the role of IL-17 in neuroinflammation into perspective, they foremost provide a conceptual advance in uncoupling IL-17 activity from encephalitogenicity. We believe that our data support further research efforts, which eventually will settle this long-standing debate in the neuroimmunology field. From a translational point of view, our data on the integral relationship between IL-17 and intestinal health already have important implications in long-term therapies targeting IL-17 signaling in the context of chronic inflammatory diseases.


Study design

This study investigated susceptibility to CNS autoimmunity of IL-17−/− and WT mice using the MOG35-55 peptide–induced EAE model. For selected experiments, these studies were complemented by the TH17 transfer EAE model (19), thereby omitting the need for adjuvant-based immunization but instead directly investigating the encephalitogenic potential of in vitro polarized T cells. Sample sizes vary per experiment and are indicated in the figure captions. For EAE experiments data collection, i.e., disease scoring was blinded. Because of the difference in genotype, experimental groups could not be randomized, except for those experiments where mice of the same genotype received different treatments. No data, including outliers, were excluded for analysis. Experiments were performed at least three times with some exceptions, as indicated in the figure captions.


C57BL/6J WT mice were purchased from Janvier Laboratories, Charles River Laboratories, or Envigo. IL-17−/− mice were provided by I. Prinz. IL-17Aind (12), and Villin-Cre (36) mice have been described elsewhere. All mice were bred and maintained in SPF conditions at the Translational Animal Research Center of the University Medical Center Mainz. GF C57BL/6 WT mice were maintained and provided by C. Reinhardt and T. Strowig. GF IL-17−/− mice were generated, maintained, and provided by C. Reinhardt.

Experimental control mice were littermates, termed IL-17+/−, or nonlittermate IL-17+/+ mice, termed WT throughout the manuscript and figures, unless stated differently. Littermate mice were cohoused (where indicated) from weaning on. For cohousing experiments involving nonlittermate mice, female microbiota donating and receiving mice were pooled in the same cage, whereas male receiving mice were provided with fresh supplemented with used embedding of microbiota donor males (1:1 mixture) every 3 days. Both procedures were maintained for at least 4 weeks before the induction of EAE or cellular analysis. For all experiments, 8-to 20-week old, sex-matched animals were used. All animal experimentation was approved by the local administration and was performed in accordance with federal and state policies.

EAE induction

Active EAE was induced as previously described (37). Briefly, mice were immunized with 50 μg of MOG35-55 peptide (GenScript) emulsified in complete Freund’s adjuvant (CFA; BD Biosciences) and supplemented with heat-inactivated Mycobacterium tuberculosis H37RA (10 mg/ml; BD Biosciences). The emulsion was injected subcutaneously at the tail base. Mice also received 200 ng of pertussis toxin (List Biological Laboratories) as intraperitoneal injection at the day of immunization and 2 days later.

For passive EAE induction, donor mice were immunized with MOG/CFA only. Ten days later, the spleen and inguinal lymph nodes of the immunized mice were removed, and single-cell suspensions were prepared and were in vitro cultured in the presence of MOG peptide (10 μg/ml), anti-IFNγ (10 μg/ml; BioXCell), and IL-23 (20 ng/ml; Miltenyi Biotec). After 4 days of culture, cells were harvested and examined for blasting lymphocytes, as based on forward scatter (FSC) and side scatter (SSC) properties in flow cytometry acquisition. Cell suspensions were adjusted to 25 × 106 blasting cells/ml, and 200 μl of this suspension was injected intravenously in the tail veins of recipient mice, accompanied by an intraperitoneal injection of pertussis toxin (200 ng) at the same day and 2 days later. Clinical assessment of EAE was performed daily according to the following criteria: 0, no disease; 1, decreased tail tone; 2, abnormal gait (ataxia) and/or impaired righting reflex (hind limb weakness or partial paralysis); 3, partial hind limb paralysis; 3.5, complete hind limb paralysis; 4, hind limb paralysis with partial fore limb paralysis; 5, complete fore limb paralysis or moribund; and 6, dead.

Tissue preparation

For the isolation of lymphocytes from the spleen and lymph nodes, the tissues were mechanically disintegrated and passed through a cell strainer (with a pore size of 40 μm) to obtain single-cell solutions. Spleen homogenates were further treated with ammonium-chloride-potassium (all from Merck) solution to lyse erythrocytes. Cells were kept on ice at all times in phosphate-buffered saline (PBS) (Merck) supplemented with 2% fetal calf serum (FCS) (Thermo Fisher Scientific).

For the isolation of lamina propria–resident cells, the small intestine was freed from fat and Peyer’s patches and washed vigorously in ice-cold PBS. Tissues were then predigested in Hanks’ balanced salt solution (Thermo Fisher Scientific) supplemented with Hepes, EDTA (both Merck), and dithiothreitol (Thermo Fisher Scientific) to disintegrate the epithelial cell layer, followed by mincing the tissue and enzymatic digestion with collagenase D, dispase II (both Roche), and deoxyribonuclease I (DNase I) (Merck) at 37°C for 30 min. Cell suspensions were then separated on a 40/80% Percoll gradient, and lymphocytes were harvested from the interphase to be used in downstream applications.

Flow cytometry

All biotinylated and fluorescence-conjugated antibodies were purchased from BD Biosciences, BioLegend, or Thermo Fisher Scientific. For surface staining, single-cell suspensions were incubated with anti-mouse CD16/CD32 (2.4g2; BioXCell) at 4°C for 10 min to prevent unspecific antibody capturing by Fc receptor–bearing cells. Dead cells were excluded by adding fixable viability dyes (Thermo Fisher Scientific) to the staining panel. Surface staining was performed in PBS containing 2% FCS at 4°C for 30 min, using the following antibodies: CD4 (GK1.5/RM4-5), CD11b (M1/70), CD19 (6D5), CD45.2 (104), CD127 (A7R34/SB/199), T cell receptor β (TCRβ) (H57-597), TCRγδ (GL3), and Thy1.2 (53-2.1). When biotinylated antibodies were used, surface staining was followed by incubation with fluorescence-coupled streptavidin (BD Biosciences or Thermo Fisher Scientific) at 4°C for 20 min. For intracellular staining, cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) or FoxP3/Transcription Factor Staining Buffer (Thermo Fisher Scientific), respectively, according to the manufacturer’s recommendation. Cells were stained in the respective buffers at 4°C for 45 to 90 min, using the following antibodies: CD154 (MR1), FoxP3 (FJK-16s), GM-CSF (MP1-22E9), IFNγ (XMG1.2), IL-17A (17B7/TC11-8H4), and retinoic acid receptor (RAR)–related orphan nuclear receptor gamma t (RORγt) (B2D/Q31-378).

Stained cells were acquired with a FACSCanto II cytometer (BD Biosciences) using FACS Diva software (BD Biosciences). Flow cytometry data were analyzed with FlowJo software version 7 or higher (TreeStar). For all analysis, doublets (FSC and SSC properties) and dead cells (dye inclusion) were excluded. Representative gating strategies for all flow cytometry analysis can be found in fig. S1.

High-dimensional flow cytometry

Multiparameter flow cytometry data were acquired and generated on a FACSSymphony (BD Biosciences) using the antibodies described in table S2. For t-distributed stochastic neighbor embedding (t-SNE)/FlowSOM plots, singlet/live/CD45+ cells were exported from FlowJo as .fcs files and uploaded into RStudio, where arcsinh cofactors were subsequently applied to every channel. The script was performed as previously described (38). t-SNE were generated mapping scholastically selected cells from all samples, and FlowSOM-guided metaclustering was displayed for all conditions.

MOG antigen recall assay

For T cell priming analysis, the spleens and inguinal lymph nodes of immunized mice were removed 10 days after immunization, and single-cell suspensions were prepared. For peak EAE analysis, mice were transcardially perfused with 0.9% NaCl solution, and the spinal cords were removed and digested with collagenase II (Thermo Fisher Scientific) and DNase I (Roche), before being separated on 30/37/70% Percoll gradient. Single-cell suspensions of all organs were in vitro cultured in the presence of MOG peptide (20 μg/ml). After 2 hours, cultures were supplemented with brefeldin A (1 μg/ml; Merck) to block the secretion of cytokines. After additional 4 hours of culture, cells were harvested and stained for flow cytometry analysis. T cell expression of CD154 was assessed as marker for recent activation, thereby serving as surrogate marker for MOG antigen specificity. CD154high T cells were then further analyzed for cytokine expression.

In vitro T cell culture

Spleens were harvested from the indicated mice, and naïve CD4+ T cells were magnetic-activated cell sorting (MACS) purified according to the manufacturer’s recommendations (Miltenyi Biotec). Purified cells (2 × 105) were then seeded in anti-CD3/anti-CD28–coated 96-well plates and cultured in the presence of IL-12 (25 ng/ml) and IL-18 (25 ng/ml) (both R&D Systems) for TH1 polarization and IL-6 (50 ng/ml), TGF-β1 (2 ng/ml; both R&D Systems), and anti-IFNγ antibodies (10 μg/ml; BioXCell) for TH17 polarization. After 4 days of culture, cells were harvested and cultured for additional 4 hours in the presence of phorbol 12-myristate 13-acetate (50 ng/ml; Merck), ionomycin (500 ng/ml; PromoCell), and brefeldin A ( Merck) before analyzed by flow cytometry.

Anti–IL-17A treatment

Naïve IL-17−/− mice were treated with 300 μg of IL-17A–neutralizing antibodies (18) diluted in PBS and intraperitoneally injected every 3 to 4 days starting directly after weaning (3 to 4 weeks of age) and throughout the course of the experiment. Control mice received intraperitoneal injections of PBS only. After 6 weeks of continuous treatment, active EAE was induced in all mice as described above.

Isolation of bacterial DNA

Feces samples were freshly drawn, immediately snap-frozen in liquid nitrogen, and then stored at −20°C until further processing. For the isolation of bacterial DNA, samples were homogenized with a FastPrep-24 instrument using Lysing Matrix E tubes (both MP Biomedicals). To support lysis of Gram-positive bacteria, samples were incubated with Gram-positive lysis buffer, containing lysozyme (20 mg/ml; Merck), 20 mM tris (pH 8.0), 2 mM EDTA, and 1.2% Triton X-100, for 30 min at 37°C. Samples were further processed using the QIAmp Fast DNA Stool Mini Kit (QIAGEN) according to the manufacturer’s recommendations. DNA was finally eluted into tris/EDTA (TE) buffer.

Samples from the experiment shown in fig. S4E were processed in a slightly different manner, according to an established protocol (39). Briefly, each sample was treated with 500 μl of extraction buffer [200 mM tris, 20 mM EDTA, and 200 mM NaCl (pH 8.0)], 200 μl of 20% SDS, 500 μl of phenol:chloroform:isoamyl alcohol (24:24:1), and 100 μl of zirconia/silica beads (0.1 mm in diameter). Samples were homogenized twice using a bead beater (BioSpec) for 2 min. After precipitation of DNA, crude DNA extracts were resuspended in TE buffer with ribonuclease I (1 mg/ml) and column purified to remove polymerase chain reaction (PCR) inhibitors.

16S rRNA gene amplification and sequencing

16S rRNA gene amplification and high-throughput sequencing was performed using the 454 platform (Roche) for the initial microbiota analysis (fig. S2A). Afterward, all samples (as presented in Figs. 2, B and C; 3, A and B; and 4, B and D; and figs. S2, B, D, and E; S4, C and E; and S5B) were sequenced on an Illumina MiSeq platform due to the discontinuation of the 454 platform. Data obtained from the different platforms were never compared with each other throughout the study.

Samples sequenced on the 454 platform were processed as we have previously described (40). For each sample, two replicates of 25 μl of PCRs were performed, each containing 20 ng of purified DNA, 0.25 mM deoxynucleotide triphosphates, 0.6 U of Taq DNA polymerase (Thermo Fisher Scientific), 2.5 μl of 10× PCR buffer, and 0.2 mM of the modified primer 8F and 534R, designed to amplify the V1 to V3 regions. The primer sequences are CCTATCCCCTGTGTGCCTTGGCAGTCTCAGAGAGTTTGATCCTGGCTCAG for 8F and CCATCTCATCCCTGCGTGTCTCCGACTCAGNNNNNNNATTACCGCGGCTGCTG for 534R, where sequences in bold correspond to the 454 Lib-L primer, whereas NNNNNNN codes for a unique barcode.

Cycling conditions were 94°C for 5 min, and 22 cycles of 94°C for 30 s, 56°C for 30 s, and 68°C for 30 s, with a final elongation cycle at 68°C for 5 min. Replicate PCRs were pooled, and amplicons were purified using the ExcelaPure 96-well UF PCR Purification Kit (EdgeBio). A library of the PCR products was prepared using the emPCR Kit Lib-L and sequenced using the GS FLX Titanium Sequencing Kit XL+ on a 454 GS FLX Titanium platform following the procedures recommended by 454 Roche.

Sequencing using the MiSeq platform (Illumina) was performed as described before (40). In this case, the V3 to V5 regions of the 16S rRNA gene were amplified (KAPA HiFi HotStart ReadyMix), indexed with the Nextera XT Index Kit (96 indexes, 384 samples), and sequenced as described in the manual for “16S Metagenomic Sequencing Library Preparation” of the MiSeq platform (Illumina) using the MiSeq Reagent Kit V3. Samples from the experiment shown in fig. S4E were processed in a different manner. Amplification of the V4 region (F515/R806) of the 16S rRNA gene was performed according to a previously described protocol (41). Briefly, 25 ng of DNA were used per 30 μl of the PCR reaction. The PCR amplification was performed using Q5 polymerase (New England Biolabs). The PCR conditions consisted of initial denaturation at 98°C for 30 s, followed by 25 cycles of 98°C for 10 s, 55°C for 20 s, and 72°C for 20 s. Each sample was amplified in triplicates and subsequently pooled. After normalization, PCR amplicons were sequenced on an Illumina MiSeq platform (PE250).

16S rRNA gene sequence analysis

Sequence data were compiled and processed using mothur v1.36 (42). We specify below the parameters used for the analysis of the sequences obtained using the different platforms.

454 sequencing data processing

Sequences shorter than 250 base pairs (bp), containing undetermined bases or homopolymer stretches longer than 8 bp, with no exact match to the forward primer or a barcode, or that did not align with the appropriate 16S rRNA variable region were not included in the analysis. Using the 454 base quality scores, sequences were trimmed using a sliding window technique, such that the minimum average quality score over a window of 50 bases never dropped below 25. Sequences were trimmed from the 3′-end until this criterion was met.

MiSeq sequencing data processing

For each sample, a forward and reverse pair-end sequence was retrieved. Quality assessment of sequences was performed using FASTP program (43). Sequences were trimmed using the sliding window technique, such that the minimum average quality score over a window of 10 bases never dropped below 30. Sequences were trimmed from the 3′-end until this criterion was met. Then, trimmed forward and reverse pair-end sequences were assembled using FLASH (44), applying default parameters. Assembled pair-end sequences larger than 400 bp were kept for the subsequent analysis. Sequences from fig. S4E (V4 region) were analyzed using the same parameters, except that all the assembled pair-end sequences larger than 250 bp were kept for subsequent analysis.

Identification of OTUs and taxonomy assignment for both 454 and MiSeq data

Sequences were aligned to the 16S rRNA gene, using as a template the SILVA reference alignment (45) and the Needleman-Wunsch algorithm with the default scoring options. Potentially, chimeric sequences were removed using the ChimeraSlayer option for 454 data and UCHIME for MiSeq Illumina data in mothur. To minimize the effect of pyrosequencing errors in overestimating microbial diversity (46), rare abundance sequences that differ in 1% from a high-abundance sequence were merged to the high-abundance sequence using the pre.cluster option in mothur. Because different number of sequences per sample could lead to a different diversity, to compare the diversity of different fecal samples, we rarefied all samples to the number of sequences obtained in the sample with the lowest number of sequences: 1419 sequences per sample for the 454 data (fig. S2A), 26,234 sequences per sample for MiSeq Illumina data obtained from the V3 to V5 regions (Figs. 2, B and C; 3, A and B; 4, B to D; and figs. S2, B, D, and E; S4C; and S5B), and 4687 for MiSeq Illumina data obtained from the V4 region (fig. S4E). Subsequently, sequences were grouped into OTUs using cluster.split with the average neighbor algorithm method for 454 data, and the faster implementation of VSEARCH (47) using cluster with the abundance-based agc method for the MiSeq Illumina data. Sequences with distance-based similarity of 97% or greater were assigned to the same OTU.

Substitution type miscalls are a major source of errors for Illumina sequencing (48), which could lead to artificial OTUs and overestimating diversity. For this reason, OTUs that were detected only once in just one sample (singletons) were removed for analysis of the MiSeq Illumina data. Subsequently, the data without singletons were rarefied to obtain the same number of sequences per sample, resulting in 24,771 sequences (Figs. 2, B and C; 3, A and B; and 4, B to D; and figs. S2, B, D, and E; S4C; and S5B) and 4550 sequences per sample (fig. S4E), respectively.

Phylogenetic classification was performed for each sequence using the Bayesian classifier algorithm described by Wang and colleagues (49) with the bootstrap cutoff of 60%. Subsequently, the consensus taxonomy for each OTU was assigned using the classify.otu command in mothur with default parameters.

α- and β-diversity analysis

OTU-based microbial α-diversity was estimated by calculating the Shannon diversity index (50) using mothur.

β-diversity comparisons were performed using the Bray-Curtis distance among different pairs of samples, which was calculated using the OTU abundance matrix and mothur. In figs. S4E and S5B, we calculated and plotted the mean of the Bray-Curtis distances obtained between each mouse of a particular group against all mice from the reference group (indicated on the y axis of the figure). In addition, nonmetric multidimensional scaling (NMDS) was performed using the Bray-Curtis distance (Figs. 2B, 3A, and 4B and fig. S2B).

Statistical analysis

To determine the OTUs significantly different among control and IL-17−/− mice (Fig. 2C and fig. S2A) or among IL-17−/−/IL-17A–Villtg/+ and IL-17−/−/IL-17A–Vill+/+ mice (Fig. 4C), the Wilcoxon rank sum test was applied. Only prevalent OTUs, i.e., present in at least 50% of the mice from one of the two groups under comparison, were included in the analysis. Obtained P values were adjusted for multiple testing using the false discovery rate (FDR) method. The relative abundance of significant OTUs (P < 0.05; q < 0.2) is shown in heatmaps in Figs. 2C and 4C and fig. S2A. The exact P values and q values for each OTU are shown in table S1. Subsequently, we studied whether the relative abundance of the identified OTUs shown in Fig. 2C was significantly different between IL-17−/− single-housed versus IL-17−/− cohoused mice using the Wilcoxon test and the FDR method (Fig. 3B). The same method was used to evaluate whether the relative abundance of the identified OTUs shown to be significantly different between IL-17−/−/IL-17A–Villtg/+ mice and single-housed IL-17−/−/IL-17A–Vill+/+ mice did also significantly differ between IL-17−/−/IL-17A–Vill+/+ single-housed versus cohoused mice (Fig. 4C). To analyze community-level differences in the microbiome among groups of samples, the nonparametric permutational multivariate analysis of variance (PERMANOVA) test was applied using the adonis function from the R vegan package v2.4-2 ( Obtained P values are indicated in the respective figures. To compare whether the Bray-Curtis distance to the reference group of samples was significantly different between two groups of samples, the Wilcoxon rank sum test was applied.

Quantification of SFB

Bacterial DNA was extracted from freshly drawn feces samples as described above. The relative abundance of SFB-specific DNA (rRNA) was determined by real-time PCR using QuantiTect SYBR Green reagents (QIAGEN) and StepOnePlus instrumentation (Applied Biosystems). SFB-specific values were normalized to results obtained from the quantification of universal bacterial DNA using these previously described primers (51): UniF340, ACTCCTACGGGAGGCAGCAGT; UniR514, ATTACCGCGGCTGCTGGC; SFB736F, GACGCTGAGGCATGAGAGCAT; and SFB844R, GACGGCACGGATTGTTATTCA.

Mouse serum metabolomics

For metabolite extraction, 100 μl of mouse serum was mixed with 300 μl of cold acetonitrile and incubated on ice for 10 min, with vortexing every 2 to 3 min. After centrifugation at 13,000 rpm and 4°C for 15 min, the supernatant was transferred to a fresh reaction tube and evaporated to dryness in a centrifugal evaporator. Extracts were stored in dry form at −80°C until analysis.

Metabolite extracts were redissolved in 50 μl of MeOH, and quality control (QC) samples were generated by pooling aliquots from all samples. These QC samples have been analyzed every eight samples. Analysis of metabolites was performed using a Sciex Exion AD liquid chromatography coupled to a Sciex X500R quadrupole/time-of-flight (Q-ToF) mass spectrometer. Separation was achieved on a Phenomenex Kinetex F5 column (150 mm by 2.1 mm inner diameter; particle size, 2.6 μm) with a gradient from eluent A (100% H2O + 0.1% formic acid) to eluent B (100% acetonitrile + 0.1% formic acid). Gradient conditions were as following: 100/0 at 0 min, 100/0 at 2.1 min, 5/95 at 14 min, 5/95 at 16 min, 100/0 at 16.1 min, and 100/0 at 20 min. Flow rate was 200 μl/min, and the column temperature was 30°C. Mass spectrometric detection was carried out in positive and negative ionization mode using information-dependent acquisition of MS2 spectra. The MS was automatically recalibrated every five samples in MS and MS2 mode.

Data processing was performed in Genedata Expressionist for MS 13.5. Preprocessing included noise subtraction, chromatographic alignment, peak detection, isotope, and adduct grouping. Data were exported as Genedata .gda file for further statistical processing in the Analyst module. MS1 and MS2 data were exported as either .gda or .mgf file for further processing and metabolite identification in R.

For further statistical analysis, data were normalized by intensity drift normalization using regularly measured QC samples as reference. Only metabolites that were detected in all QC samples and had a relative SD lower than 30% were further analyzed. Principal component analysis was used to detect potential outlier samples. After removal of outliers, a Welch test was conducted to identify statistically different metabolites.

Data exported from Genedata Expressionist for MS 13.5 were imported into R using the genedataRutils package (, and an MS1 isotope pattern was reconstructed. MS1 and MS2 spectra from all samples were combined into consolidated spectra, which were exported as .ms files for analysis with Sirius and CSI:FingerID (52). Metabolite MS2 spectra were matched against an in-house spectral library using MS1, MS2, and retention time information from the same instrument and method (MSI level 1 identifications). Furthermore, matching against an in-house database from a different instrument (MSI level 2 identifications; and results from Sirius and CSI:FingerID were used (MSI level 2 to 3 identifications).

Antibiotic treatment and microbiota transfer

Mice were treated with ampicillin (0.5 mg/ml; Ratiopharm), gentamycin (0.5 mg/ml), metronidazole (0.5 mg/ml), neomycin (0.5 mg/ml; all Merck), and vancomycin (0.25 mg/ml; Hikma), all dissolved in autoclaved drinking water, for 4 weeks. The antibiotics cocktail was freshly prepared every 3 to 4 days. At the day of recolonization, the fecal content of small intestine, cecum, and large intestine of control and IL-17−/− mice, respectively, was collected in Thioglycollate Medium (BD Biosciences) and homogenized, and 200 μl of this suspension was administered to the previously antibiotic-treated recipient mice by oral gavage. The procedure of recolonization was repeated 2 weeks later. After an additional 2 weeks, EAE was induced in the recolonized mice.


Tissue samples from the small intestines of the indicated mice were embedded in tissue freezing medium, and 6- to 10-μm-thick sections were cut on a CM1900 cryostat (Leica). Sections in fig. S2 were fixed with 4% paraformaldehyde and stained with primary antibodies specific for cleaved caspase 3 (Asp175; Cell Signaling Technology), EpCAM (Abbiotec), and TCRγδ (clone UC7-13D5; Thermo Fisher Scientific). Primary antibodies were subsequently labeled with biotinylated species–specific secondary antibodies (Jackson ImmunoResearch) and stained using the tyramide signal amplification system (PerkinElmer), according to the manufacturer’s recommendations. Small intestinal M cells were stained with fluorescence-conjugated Ulex europaeus agglutinin I (Vector Laboratories). Small intestinal–resident bacteria were visualized by hybridization of bacterial rRNA to the fluorescence-conjugated probe 5′-Cy3-GCTGCCTCCCGTAGGAGT-3′, serving as universal detector of all eubacteria [fluorescence in situ hybridization (FISH)–Eubac; Metabion]. For all stainings, nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) as component of the mounting medium (Vector Laboratories). For the sections in fig. S5, the cell-intrinsic fluorescence was captured, whereas nuclei were counterstained with DAPI.

Statistical analysis

The experimental sample size was not predetermined statistically, experiments were not randomized, and investigators were not blinded to group allocation during experimentation and analysis of results. Experimental data were analyzed with unpaired, two-tailed Student’s t test or with one- or two-way ANOVA followed by Bonferroni’s or Tukey’s multiple comparison tests, as indicated. All statistical tests were performed with GraphPad Prism (v.5 or higher) software. P values of less than 0.05 were considered to be statistically significant. Original P values are stated only for statistically significant differences and can be found directly in the figures or in table S3.


Fig. S1. Representative gating strategies for flow cytometry analysis.

Fig. S2. The encephalitogenic T cell response is fully functional in IL-17−/− mice.

Fig. S3. IL-17−/− mice show distinct changes in their gut microbiota.

Fig. S4. IL-17−/− mice present with disturbances in gut homeostasis.

Fig. S5. The microbiota influence on EAE susceptibility is unidirectional.

Fig. S6. Ectopic expression of IL-17A in the intestine of IL-17−/− mice.

Table S1. List of OTUs from microbiota sequencing (separate Excel file).

Table S2. List of antibodies used for high-dimensional flow cytometry (separate Excel file).

Table S3. Raw data (separate Excel file).


Acknowledgments: We thank N. Puertas, C. El Gaz, E. Barleon, and A. Frömming for technical assistance. Funding: Work in the A.W. laboratory is supported by the National Multiple Sclerosis Society (RG-1707-28780), the DFG (SFB/TRR128 and SFB1292), the Sobek Foundation, the Hertie Foundation (P1150062), the Stiftung Rheinland-Pfalz für Innovation (961-386261/1141), and the Research Center for Immunotherapy (FZI) Mainz. Work in the C.B. laboratory is supported by the MINECO grant SAF2017-90083-R and a MINECO FPI predoctoral fellowship (to S.I.). Work in the B.B. laboratory is supported by the Swiss National Science Foundation (grants 310030_170320, 310030_188450, 316030_150768, and CRSII5_183478) and the European Union FP7 Project NeuroKine (to A.W. and B.B.). Work in the T.B. laboratory is supported by DFG grants SFB/TRR128 and SFB1292. Work in the I.M.P. laboratory is supported by the Sobek Foundation, the Ernst-Jung Foundation, the DFG (SFB992, SFB1160, Reinhart Koselleck Grant, and SFB/TRR167), the Ministry of Science, Research and Arts, Baden-Württemberg (Sonderlinie “Neuroinflammation”), and the BMBF-funded competence network of multiple sclerosis (KKNMS). N.Y. is supported by a grant from the Center for Molecular Medicine Cologne (CMMC B08). Author contributions: T.R. and A.W. conceived the project. T.R. performed most of the experiments and analyzed the data. J. Hauptmann, A.S., M.K., I.A.M., N.Y., J. Huppert, F.W., A.N., and M.B. helped with the experiments, data analysis, and discussion. S.I. and C.U. performed the microbiota sequencing and analyzed the data. A.A., N.G.N., and B.B. performed the high-dimensional flow cytometry and analyzed the data. A.G. and C.R. generated the GF IL-17−/− mice and provided the GF research models and expertise. E.J.C.G. and T.S. helped with the GF experiments and microbiota sequencing and provided expertise. R.S. and M.P. helped with histology. M.W. and P.S.-K. performed the mouse serum metabolite measurement and analyzed the data. I.P. and T.B. provided essential research models, scientific advice, and expertise. T.R., B.B., C.U., and A.W. wrote the manuscript with input from all authors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Microbiota sequencing data generated during this study are available in the NCBI database through accession number PRJNA682734. All other data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. Requests for research materials should be addressed to A.W.

Stay Connected to Science Immunology

Navigate This Article