Role of Gut Commensal Microflora in the Development of Experimental Autoimmune Encephalomyelitis¹

Oral tolerance has been associated with the control of experimental autoimmune diseases, including experimental autoimmune encephalomyelitis (EAE),³ the animal model of multiple sclerosis (MS) (1). Induction of oral tolerance as a protective pathway in human MS has been attempted, yet it has been met with little success. There is limited information available on the role of gut commensal bacteria. However, dietary concerns remain an important clinical issue for those diagnosed with MS.

The increasing numbers of studies on the role of commensal microorganisms as immune modulators show that microbial colonization of mucosal surfaces begins shortly after birth. This generates a highly diverse endogenous microflora population comprising over 10¹⁴ resident bacteria that create a relationship that confers benefits to both colonizers and host (2, 3, 4). Pathogens share the same mucosa and it is the role of the immune system to control concurrently the responses to commensal and pathogenic organisms (3). Absence of bacteria in germ-free mice (axenic) that are born and raised in sterile isolators demonstrates that the presence of commensal bacteria is essential for normal immune development (5). Alterations in the immune profile in these mice exhibit a default Th2 bias and a significant reduction in proinflammatory IL-17-producing CD4⁺ T cells compared with mice with an intact communal gut bacterial profile (6). Recent studies show that alterations of commensal populations can determine the regulatory T (T_reg)/Th17 cell balances in the GALT (7, 8), and the role of dendritic cells (DCs) in this process appears to be critical; CD103⁺ gut-derived DCs drive the conversion of naive CD4⁺CD25⁻ effector T cells into a regulatory FoxP3⁺ T_reg cell subpopulation and the acquisition of a gut-homing phenotype in the presence of retinoic acid and TGF-β has been described (9, 10).

Peripheral immunity/tolerance induced in the gut has been shown to be a powerful tool to control autoimmune disorders (1). Recent studies have demonstrated that tolerance induction through oral immunization with foreign Ags can control the development of autoimmune diseases. Oral immunization with a single dose of an attenuated strain of Salmonella typhimurium expressing the CFA/I fimbriae of enterotoxigenic Escherichia coli conferred prophylactic (11) and therapeutic (12, 13) protection against EAE in SJL mice. Salmonella-CFA/I elicited FoxP3⁺ T_reg cells (12, 13), protective against EAE. Since these T_reg cells induced by Salmonella vaccines were elicited after an oral immunization and protected against an inflammatory neurodegenerative disease, it is tempting to suggest a connection between the immune responses elicited in the gut and the immune consequences that may take place in the periphery and, in the case of EAE, within the CNS. Recently, it has been shown that NKT cells may be involved in protection against EAE in B6 mice following antibiotic treatment (14).

In this work, we demonstrate that alterations of bacterial populations of the gut by oral treatment with antibiotics can induce the accumulation of FoxP3⁺ T_reg cells in distal peripheral lymph nodes and reduce the severity of EAE in a T_reg cell-dependent manner. Moreover, adoptive transfer of FoxP3-enriched CD25⁺CD4⁺ T cells from cervical lymph nodes (CLN) of animals treated with oral antibiotics conferred protection against EAE, suggesting that gut-associated T_reg cells can be manipulated by alteration of commensal bacterial populations. Of interest is that this immune modulation in the gut by antibiotic treatment is not restricted to the GALT, as suppressive regulatory T cells accumulate in distal effector sites such as CLN in response to changes in the bacterial content of the gut. This would implicate the GALT as a potential reservoir of both the effector and regulatory cell populations that appear to be involved in the control of EAE and perhaps human multiple sclerosis.

Female 6-wk-old SJL mice were obtained from The Jackson Laboratory. All mice were maintained in accordance with institutional policies for animal health and well-being at Dartmouth College Animal Resources Center under pathogen-free conditions in individual ventilated cages under HEPA-filtered barrier conditions and were fed sterile food and water ad libitum. Mice were treated with the following antibiotics dissolved in drinking water or dissolved in water: ampicillin (1 g/ml), vancomycin (0.5 g/ml), neomycin sulfate (1 g/ml), and metronidazole (1 g/ml) (15). When required, dissolved antibiotics were administered by oral gavages or i.p. injections at daily single 200-μl doses of 1 g/ml. Mice treated i.p. or by oral gavages received normal drinking water during the length of the experiments. Serial dilutions of intestinal and fecal samples were cultured in general bacteriological agar plates (CDC blood agar; BD Biosciences) for 48 h at 37°C. Plates were cultured in aerobic and anaerobic conditions. Total bacteria per gram of sample was calculated based on the CFU counted in each serial dilution.

Female SJL/J or C57BL/6 mice (four per group) were challenged s.c. with 200 μg of PLP_139–151 or 250 μg of MOG_35–55 (Peptides International), respectively, in 200 μl of CFA (Sigma-Aldrich). Mice received i.p. 200 ng (SJL/J) or 400 ng (C57BL/6) of Bordetella pertussis toxin (PT; List Biological Laboratories) (days 0 and 2) (12). Spinal cords were harvested 12 days after challenge and fixed with neutral buffered formalin (VWR International), embedded into paraffin, sectioned at 3 μm, and stained with H&E and Luxol fast blue for pathological changes and inflammatory cell infiltration (12).

Spleen and CLN cell suspensions were costimulated with PLP_139–151 (30 μg/ml) or with anti-CD3 mAb (coated well; 10 μg/ml; BD Pharingen) and soluble anti-CD28 mAb (5.0 μg/ml; BD Pharmingen) for 3 days (12). Luminex was employed to quantify triplicate sets of samples to measure cytokines (IFN-γ, TNF-α, MIP-1α, MIP-1β, MCP-1, IL-6, IL-17, IL-10, IL-4, and IL-13). Capture ELISA was used to quantify IFN-γ, IL-17, IL-10, and IL-13 from supernatants of cells cultured after FACS sort, as previously described (13). For PCR detection of IL-13 mRNA (r-GGTCCTGTAGATGGCATTGCA; l-GGAGCTGAGCAACATCACACA; Invitrogen) a total of 1.0 μg of RNeasy-purified mRNA (Qiagen) was reverse transcribed using MultiScribe RT (Amersham Biosciences). cDNA (200 ng) was amplified using the 2× SYBR Green mix (Applied Biosystems) on a Bio-Rad iCycler.

Single-cell preparations were prepared as described above (12). Cell subsets were analyzed using fluorochrome-conjugated mAbs (BD Pharmingen). For the analysis of T cell subpopulations, CD3, CD4, CD8, CD25, and CD45Rb (BD Pharmingen) were used. Intracellular staining for FoxP3 was done using flourochrome-labeled anti-FoxP3 mAb (clone FJK-16S; eBioscience). For NK cells, DX5, B220, and CD11b were used. For B cells, CD19 and B220 (BD Pharmingen) were used. For macrophages and DC subpopulations, CD11b, CD11c, CD103, B220, CD8, Gr-1, and F4/80 mAb were used (BD Pharmingen). For DC activation, MHC class II (I-Ad; clone AMS-32.1) and CD80 (BD Pharmingen) were used. Fluorescence was analyzed with a FACSCanto (BD Biosciences).

CD4⁺ T cells and CD8⁺ T cells were obtained with magnetic beads (Dynal Biotec). The enriched CD4⁺ T cells were cell-sorted for FITC-anti-CD4 and allophycocyanin-anti-CD25 mAbs (BD Pharmingen). CD11c⁺ cells were obtained with magnetic beads (StemCell Technologies). The enriched CD11c⁺ cells were cell-sorted (FACSVantage with Turbo-Sort; BD Biosciences) following staining with FITC-anti-CD103 into CD11c^highCD103⁺ cells.

To assess T_reg cell suppressor activity, 1.5 × 10⁵ responder CD25⁻CD4⁺ T cells were labeled with CFSE and subsequently cocultured in triplicate with CD25⁺CD4⁺ T cells at 1:1, 1:0.1, 1:0.01, and 1:0.001 CD25⁻/CD25⁺ T cell ratios. Feeder cells (T cell-depleted mitomycin C-treated) splenocytes prepared from naive mice (16) were added at 1.5 × 10⁵ cells per well. Cells were incubated at 37°C in 5% CO₂ for 72 h. CD4⁺ T cell proliferation was compared by FACS.

To asses cytokine production by CD25⁻CD4⁺ T cells and CD25⁺CD4⁺ T cells, sorted cells (2 × 10⁵) were stimulated in vitro with PLP_139–151 (30 μg/ml) or with anti-CD3 mAb-coated wells (10 μg/ml; BD Pharmingen) plus soluble anti-CD28 mAb (5 μg/ml; BD Pharmingen) (13). Capture ELISA was used to quantify IFN-γ, IL-17, IL-10, and IL-13 production.

To assess the role of CD11c^highCD103⁺ and CD11c^highCD103⁻ DCs in the in vitro conversion of splenic CD4⁺ T cells sorted from naive SJL/J mice into FoxP3⁺ T_reg cells, 1.5 × 10⁵ naive CD25⁻CD4⁺ T cells were cocultured in triplicate with CD11c^highCD103⁻ or CD11c^highCD103⁺ DCs. Anti-CD3 mAb (10 μg/ml; BD Pharmingen) and IL-2 (20 U/well) were added. Retinoic acid (4 nM) and TGF-β (1 ng/ml) were added to some cultures to enhance the T_reg cell conversion rates. Cells were incubated at 37°C in 5% CO₂ for 72 h. Conversion of naive CD25⁻CD4⁺ T cells into FoxP3⁺ T_reg cells was compared by FACS.

CD25⁺CD4⁺ T cells (4 × 10⁵) or CD25⁻CD4⁺ T cells (4 × 10⁵) were i.v. injected into naive recipients. In different experiments, 1 × 10⁶ CD4⁺ T cells and CD8⁺ T cells were adoptively transferred. One day after the adoptive transfer of T cells, mice were challenged with PLP_139–151 to induce EAE.

Mice were orally treated with antibiotics 7 days before EAE challenge with PLP_139–151 and PT. To inactivate CD25⁺CD4⁺ T cells, the same mice were given 0.3 mg of anti-CD25 mAb (American Type Culture Collection no. TIB-222, clone PC 61.5.3) on days 4 and 2 before EAE challenge (12). As a control group, treated and naive mice received 0.3 mg of purified rat IgG Ab on the same days before EAE challenge. CD25 depletion was confirmed by FACS analysis of peripheral blood samples obtained 2 days after the administration of the second dose of anti-CD25 or rat IgG Abs. A separate control group was immunized with PBS 7 days before EAE challenge.

Student’s t test was applied to show differences of combined experiments in clinical scores, body, spleen, and cecum weights, Luminex detection of cytokines, as well as in the flow cytometry of T_reg cell and DC experiments. ANOVA followed by post hoc Tukey’s test was applied to show differences in EAE clinical scores. Values of p <0.05 and <0.01 are indicated.

Studies in germ-free animals have demonstrated that microbial populations of the gut are essential for the complete development of a normal immune response. To assess whether modification of normal gut commensal microflora would influence the induction and development of an acute demyelinating process in the CNS, mice were treated with an antibiotic cocktail that has been shown to effectively reduce the gut intestinal bacterial burden (15). Drinking water was supplemented with ampicillin (1 g/L; Sigma-Aldrich), neomycin sulfate (1 g/L; Sigma- Aldrich), metronidazole (1 g/L; Sigma-Aldrich), and vancomycin (0.5 g/L; Sigma-Aldrich) for 7 days (15). EAE was induced with PLP_139–151 in SJL and MOG_35–55 in C57BL/6 mice previously treated with antibiotics (Fig. 1). Control mice were treated with PBS and i.p. with the same antibiotics. There have been different reports implicating a direct neurological effect by injections of minocycline, a second generation type of tetracycline. Minocycline provides partial protection against EAE when combined with glatiramer acetate and IFN-β (17, 18), provoking a down-regulation in the Ag presentation capability of blood monocyte-derived DC Ag presentation in mice and activation capability in MS patients (17).

Oral treatment with antibiotics previous to challenge with PLP_139–151 significantly reduced the severity of EAE when compared with PBS control and i.p. treated animals (Fig. 1^,A). Continuous oral antibiotic treatment in the drinking water resulted in body weight reduction of SJL/J mice. However, treatment of SJL/J mice with the same combination of antibiotics administered by once daily oral gavages (200-μl doses of 1 g/ml) and free access to untreated drinking water had no effect on body weight (Fig. 1 A). These mice were protected against EAE compared with control mice that had free access to normal drinking water and were not treated with antibiotics. Both PBS- and i.p. treated mice developed significant clinical disease compared with mice treated with oral antibiotics. Differences were observed in the onset of the disease and the cumulative scores of PBS vs i.p. compared with orally treated mice (supplemental Table SI).⁴

Similar studies were done with oral antibiotic treatment in C57BL/6 mice in which EAE was induced following immunization with MOG_35–55 (Fig. 1,B). The antibiotic-treated mice were protected against disease and there was no significant alteration in body weight (Fig. 1 B). The EAE cumulative scores of mice treated orally with antibiotics were significantly reduced (p < 0.01) when compared with PBS-treated (13.1 ± 1.2 vs 52.0 ± 2.1) and i.p. treated mice (13.1 ± 1.2 vs 49.7 ± 1.1) (supplemental Table SII). No significant differences in the EAE clinical scores were observed between C57BL/6 mice treated with PBS and mice treated i.p. with antibiotics. These results suggest that the protection observed was independent of any physiologic stress induced by continuous oral antibiotic treatment and more importantly the genetic background of the animals used in the studies.

Oral treatment with antibiotics significantly reduced the bacterial counts present in fecal and intestinal samples of SJL/J (Fig. 1,A) and C57BL/6 mice (Fig. 1 B). Aerobic and anaerobic conditions were examined and, in both cases, a significant reduction of bacterial counts was found after 1 wk of treatment. No bacterial CFU were detected in fecal samples of mice treated orally with antibiotics as opposed to the culture of fecal intestinal contents, suggesting that fresh pellets might be insufficient to compare total bacterial loads. When fresh fecal pellets and intestinal fractions of mice were combined and cultured, we observed that only oral treatment by drinking water or by oral gavages but not i.p. treatment with antibiotics reduced gut commensal microflora. However, antimicrobial treatment did not completely deplete bacterial presence, showing that certain bacterial populations remain viable despite antibiotic treatment. When animals were subsequently provided with normal drinking water, intestinal recolonization was observed 1 wk later (supplemental Fig. 2A). The treatment with antibiotics does not render the gut sterile but rather substantially reduces the bacterial load and perhaps alters the composition of the normal gut microflora. No significant differences in the splenic and cecum weights were observed in mice treated i.p. with antibiotics when compared with naive mice (supplemental Fig. 2B).

Our results show that daily administration of an antibiotic cocktail by single oral gavages or continuous administration in the drinking water rendered SJL/J mice resistant to EAE. Antibiotic treatment by oral gavages did not affect mouse weight that remained stable compared with baseline (no treatment). As weight loss did not appear to mediate resistance, SJL/J mice were treated with antibiotics by continuous administration in the drinking water in all further studies.

To study the cytokine pattern of mice treated with antibiotics, Peyer’s patches (PP), mesenteric lymph nodes (MLN), and splenic and cervical lymph nodes (CLN) lymphocytes were harvested from naive and mice treated orally with antibiotics and costimulated with αCD3/αCD28 Abs (Table I). Results show that the reduction of gut commensal microflora significantly diminished the production of MIP-1α, MIP-1β, and IL-6 in PP. MLN of animals treated with antibiotics produced lesser amounts of IFN-γ, MIP-1α, MIP-1β, and IL-6 and significantly increased levels of IL-13. Splenic and CLN cells derived from these mice produced reduced IFN-γ, MIP-1α, MIP-1β, MCP-1, IL-17, and IL-6 levels, whereas IL-13 and IL-10 (in CLN) were significantly enhanced when compared with untreated control mice.

Flow cytometry was used to compare the populations of T cells, B cells, DCs, macrophages, NK cells, and NKT cells (supplemental Fig. 1). A significant reduction in CD4⁺ T cells and an enhanced CD8⁺ T cell response was observed in mice treated orally with antibiotics when compared with naive and i.p. treated mice. Phenotypic analysis of the various immune compartments within the PP of animals treated orally with antibiotics showed a significant reduction in T cell, B cell, and CD11c⁺CD11b⁻ percentages. Conversely, there was a significant increase in CD11c⁺CD11b⁺ DCs when compared with either naive or mice treated i.p. with the same antibiotic cocktail. Percentages of CD11b⁺F4/80⁺ monocytes, NK cells, and NKT cells of treated mice failed to show any significant difference when compared with untreated control mice. The MLN of mice treated with oral antibiotics showed a significant reduction in total T cells, but no change in B cells, CD11b⁺F4/80⁺ monocytes, NK cells, NKT cells, or CD11c⁺CD11b⁺ or CD11b⁻ DC populations. The percentage of splenic T cells was significantly higher in orally treated than in naive and i.p. treated mice. No alterations were observed in CD11c⁺CD11b⁺, CD11c⁺CD11b⁻, and CD11c⁺Gr-1⁺ DCs or in CD11b⁺F4/80⁺ monocytes. Interestingly, a significant reduction in NK and NKT cell percentages in the spleen was observed in mice after oral treatment with antibiotics. Finally, analysis of CLN showed that percentages of T cells were reduced significantly in mice treated orally with antibiotics, with no modifications in the rest of cellular populations compared.

Oral treatment with antibiotics altered significantly CD4⁺ T cell subpopulations (Fig. 2 and supplemental Fig. 5, A and B). FACS analysis revealed that the frequency of CD4⁺CD25⁺ T cells was reduced in PP of mice orally treated with antibiotics, but significantly increased (p < 0.01) in MLN, spleens, and CLN when compared with naive and i.p. treated mice (Fig. 2,A). Lymph nodes of treated mice showed reciprocal reduction and enhancement of activated CD45Rb^lowCD4⁺ T cells in MLN and CLN of CD25⁺ T cell populations when compared with naive and mice treated i.p. with antibiotics (data not shown). Interestingly, FACS analysis showed that oral treatment with antibiotics provoked a significant reduction (p < 0.01) in the frequency of FoxP3⁺CD25⁺/total CD4⁺ T cells in spleens but was otherwise unchanged from control values (Fig. 2,B). When total numbers of FoxP3⁺ T_reg cells were compared, significant reductions (p < 0.01) were measured in PP and spleens of mice subjected to oral treatment with antibiotics. However, gut flora alterations enhanced FoxP3⁺T_reg cell numbers significantly (p < 0.01) in MLN and CLN when compared with naive and mice treated i.p. (Fig. 2 C).

The role of CD11c^highCD103⁺ DCs in the conversion of naive CD4⁺ T cells into Foxp3⁺ T_reg cells has been demonstrated (19), and the potential role for commensal bacteria in this conversion was hypothesized (19, 20). It has been suggested that CD103⁺ DCs might migrate from the intestine to the MLN, where they could generate T_reg cells (21). We investigated whether the enhancement of FoxP3⁺ T_reg cells observed in the lymph nodes of mice treated with antibiotics could be driven by this specific mucosal DC subpopulation. We observed a significant increase in the percentages and total numbers of CD11c^highCD103⁺ and CD11c^highCD103⁻ DCs in PP and MLN of mice orally treated with antibiotics when compared with naive and i.p. treated mice (Fig. 3 A; gating strategy and surface characterization of CD11c^highCD103⁺ and CD11c^highCD103⁻ DCs are shown in supplemental Fig. 4A).

We compared the effect of CD11c^highCD103⁺ and CD11c^highCD103⁻ DCs purified from naive and mice treated with antibiotics in the conversion of splenic naive CD4⁺ T cells into FoxP3⁺ T_reg cells (Fig. 3,B). The acquisition of FoxP3 by T cells was significantly augmented when cells were cocultured with CD11c^highCD103⁺ DCs of mice treated orally with antibiotics when compared with naive and i.p. treated mice (Fig. 3,B). The effect of CD11c^highCD103⁻ DCs in the conversion of naive CD4⁺ T cells into FoxP3⁺ T_reg cells was significantly lower, and no substantial differences in the conversion rates were observed between naive mice treated orally or mice treated i.p. with antibiotics (Fig. 3 B). No differences in cell proliferation were observed in cultures with CD11c^highCD103⁻ DCs or CD11c^highCD103⁺ DCs (data not shown). The role of TGF-β and retinoic acid (RA) in the enhancement of T_reg cell conversion by MLN CD103⁺ DCs has been described before (19). We compared the effect of TGF-β and RA in our experimental groups (supplemental Fig. 4B). As described before, rates of conversion by CD11c^highCD103⁺ DCs were increased for all groups tested. The conversion rates induced by CD11c^highCD103⁻ DCs were significantly reduced even in the culture with RA and TGF-β (supplemental Fig. 4B and Ref. 19).

The reduction of proinflammatory responses in the GALT and other peripheral lymph nodes after oral treatment with antibiotics was shown previously (Table I) (14). We tested whether the changes in gut commensal microflora would protect against EAE by alterations in the cytokine patterns (Fig. 4). Total lymphocytes were harvested from CLN 13 days after EAE induction and cytokines were quantified by capture ELISA after restimulation with PLP_139–151. Lymphocytes obtained from the oral treatment protected group of mice released to the media reduced levels of IL-17 and IFN-γ, whereas IL-10 and IL-13 were enhanced when compared with PBS- and i.p. treated mice lymphocytes (Fig. 4 A). IL-13 was enhanced in lymphocytes of i.p. treated compared with PBS-treated mice. However, the level of cytokine production was significantly lower when compared with mice treated orally with antibiotics.

To identify the CD4⁺ T cell population responsible for the switching of the cytokine profiles of mice protected against EAE, CLN lymphocytes were sorted in CD4⁺CD25⁺ and CD4⁺CD25⁻ T cells and pulsed with PLP_139–151. CD4⁺CD25⁻ T cells from orally treated mice produced significantly reduced levels of IFN-γ and IL-17 when compared with PBS- and i.p. treated mice (Fig. 4,B). However, IL-10 and IL-13 were significantly increased in these orally treated mice. When the cytokine profile of CD4⁺CD25⁺ T cells was compared, cells sorted from mice treated orally with antibiotics produced significantly higher levels of IL-10 when compared with PBS- and i.p. treated mice. No significant differences were observed in IFN-γ, IL-17, or IL-13 cytokines in the CD4⁺CD25⁺ T cell populations compared (Fig. 4 B). Our results show that oral treatment with antibiotics caused immune deviation in mice characterized by a reduction in proinflammatory responses and significant enhancement in Th2-type immunity.

PCR analysis showed enhanced levels of IL-13 expression in the brains of animals protected against EAE by oral treatment with antibiotics when compared with PBS-treated mice and animals treated i.p. with antibiotics (supplemental Fig. 3). No significant differences in IL-13 production were observed in brains of mice treated i.p. and control PBS-treated mice. Demyelination and nucleated cell infiltration levels were reduced in orally treated mice. No significant differences were observed between PBS- and i.p. treated mice (supplemental Table SIII and supplemental Fig. 3). Interestingly, when mice were treated with the antibiotics during the entire length of the experiment, mice were fully protected with no evidence of disease development as determined by clinical score (supplemental Fig. 3B). These data suggest that intestinal colonization with certain bacterial populations can evoke clinical disease consistent with EAE.

To assess whether protection against EAE was associated with an increase in the T_reg cell populations, lymphocytes derived from PP, MLN, spleen, and CLN were analyzed for presence of FoxP3⁺ T_reg cells (Fig. 5 and supplemental Fig. 5, C and D). A significantly higher (p < 0.01) percentage of FoxP3⁺T_reg cells was observed in spleen and MLN in animals treated with oral doses of antibiotics and protected against EAE when compared with untreated mice and animals treated i.p. with antibiotics. A nonsignificant increase in the frequency of FoxP3⁺ T_reg cells in CLN of orally treated mice was observed; however, when total numbers were compared a significant increase of FoxP3⁺ T_reg cells in CLN, MLN, and spleens of orally treated mice was observed when compared with PBS- and i.p. treated mice (Fig. 5 B and supplemental Fig. 5, C and D).

Our results suggest that a combination of Th2-type immune responses and the induction of regulatory T cell subpopulations may provide an important framework that can offer protection against EAE when bacterial communities of the gut are challenged with antibiotics.

Adoptive transfer experiments were performed to determine whether the elicited T cell response following oral antibiotic treatment was responsible for the reduced susceptibility to EAE. In the first experiment, the protective role of CD4⁺ or CD8⁺ T cells was compared (Fig. 6, A and B). SJL/J mice were treated for 7 days with ampicillin, vancomycin, neomycin sulfate, and metronidazole dissolved in drinking water, or with normal drinking water (naive control group). After the treatment, CLN were harvested and CD4⁺ or CD8⁺ T cell populations were enriched by selection with magnetic microbeads. Adoptive transfer of 1 × 10⁶ cells per mouse (≥96% pure) was performed 1 day before EAE induction with PLP_139–151. CD4⁺ T cells isolated from CLN of mice treated with antibiotics significantly reduced the EAE clinical scores of SJL mice when compared with CD4⁺ T cells obtained from naive mice (Fig. 6,A). In contrast, no significant differences were observed in the clinical outcome of the disease after adoptive transfer of CD8⁺ T cell-enriched populations from CLN of mice treated with antibiotics when compared with PBS-treated mice or mice treated with naive CD8⁺ T cells (Fig. 6 B). These results suggest that CD8⁺ T cells of mice treated with antibiotics do not play a role in the protection against EAE observed previously.

We next evaluated whether CD25⁺CD4⁺ or CD25⁻CD4⁺ T cells obtained from CLN of mice treated with antibiotics would be suppressive in vitro and would confer protection against EAE after adoptive transfer. The suppressive capacity of antibiotic-treated, FoxP3-enriched CD25⁺CD4⁺ T cells was significantly enhanced at a 1:10 suppressor T-to-effector T cell ratio (supplemental Fig. 6). To analyze a potential protective role of these cell populations, naive recipient SJL mice were adoptively transferred with 4 × 10⁵ cells per mouse of CD25⁺CD4⁺ or CD25⁻CD4⁺ T cells obtained from CLN of naive mice or mice previously treated with antibiotics 1 day prior to EAE induction with PLP_139–151. When CD25⁺CD4⁺ T cells (>75% FoxP3⁺) purified from CLN of SJL mice treated with antibiotics were transferred, a significant reduction of the EAE clinical scores was observed (Fig. 6 C). No protection was observed after adoptive transfer of the control arms, including CD25⁻CD4⁺ T cells purified from mice treated with antibiotics and CD25⁺CD4⁺ and CD25⁻CD4⁺ T cells obtained from naive mice.

Analysis of the cytokine profile of adoptively transferred CD25⁺CD4⁺ and CD25⁻CD4⁺ T cells showed that protective CD4⁺CD25⁺ T cells (>75% FoxP3⁺) sorted from mice treated orally with antibiotics produced significantly enhanced levels of IL-10 (p < 0.01) and IL-13 (not significant) when compared with naive CD4⁺CD25⁺ T cells. When CD25⁻CD4⁺T cells were compared, those obtained from orally treated mice showed significant reductions in IFN-γ and IL-17 and not significant differences in IL-10 and IL-13 when compared with naive levels (Fig. 6 D).

To confirm the protective capacity of the T_reg cells from oral antibiotic-treated mice, in vivo neutralization of CD25-expressing cells was performed using a depleting anti-CD25 mAb (clone PC-61). Two doses of 300 μg/mouse on days 3 and 5 after the initiation of oral antibiotic treatment reduced the CD25⁺ in CD4⁺ T cells of naive mice as well as mice treated with either orally or i.p. with antibiotics when compared with control treatment with rat IgG isotype control (Fig. 7,A). Partial reversion of protection was observed by depletion of CD25⁺ T cells in mice treated with oral antibiotics (Fig. 7 B). The onset of clinical disease occurred earlier (p < 0.05) in all groups treated with anti-CD25 mAb when compared with rat IgG-treated mice (supplemental Table SIV). The cumulative scores and mortality of mice treated orally with antibiotics and subsequently with anti-CD25 mAb were significantly more severe (p < 0.05) when compared with mice treated orally with antibiotics and injected with rat IgG. Interestingly, EAE clinical scores were also significantly reduced in CD25-neutralized mice previously treated with antibiotics when compared with either naive (p < 0.05) or i.p. treated (p < 0.05) mice.

The findings we report herein demonstrate that gut commensal bacteria appear critical for the mediation of immune homeostasis in the periphery distal to the lumen of the gut and the GALT. Moreover, commensal bacteria within the gut can directly control the development of EAE, the experimental model of human multiple sclerosis. Our results implicate that the protection conferred by the alteration of gut microflora in response to oral antibiotic treatment might be mandated by a combination of regulatory and antiinflammatory cell populations.

To test whether gut commensal bacteria are involved in the control of inflammatory demyelinating disease, we utilized a broad spectrum of antibiotics (ampicillin, vancomycin, neomycin sulfate, and metronidazole) to reduce the gut commensal bacteria in EAE-susceptible SJL mice (15). The reduction of existing gut commensal microflora rather than the total absence provided by germ-free animals was preferred for our purposes to assess the impact of these changes in humans with MS and perhaps other autoimmune conditions. We have observed that oral antibiotic treatment does not render the gut sterile but rather substantially reduces the bacterial load and in all likelihood selectively alters the composition of the normal gut microflora (Fig. 1). Administration of antibiotics via an i.p. route does not have any quantitative effect on bacterial load in the gastrointestinal tract. Interestingly, treatment of EAE mice with minocycline, a second generation tetracycline, provided partial protection against disease when combined with IFN-β and glatiramer acetate (17, 18). Importantly, combination therapy with glatiramer acetate and minocycline in those individuals with MS provoked a down-regulation in the Ag presentation capability of blood monocyte-derived DC Ag presentation and activation capability (17). To test whether the selected cocktail of antibiotics had a direct neurological protective effect, mice were treated i.p. with the same antibiotic regimen used for oral administration. Our results demonstrate that i.p. antibiotic administration failed to confer protection against EAE. In contrast, oral treatment significantly reduced EAE clinical symptoms, and long-term treatment conferred full protection, suggesting that intestinal colonization with certain bacterial populations is necessary for the development and persistence of murine EAE.

Both continuous treatment with oral antibiotics in the drinking water and single daily oral gavages with antibiotics reduced clinical disease. As the reduction in disease severity in these two therapeutic approaches were similar, the weight loss observed after continuous antibiotic treatment does not appear to be important to the protection observed. Phenotypic and functional analysis of FoxP3⁺ T_reg cell populations were performed only after oral antibiotic treatment by drinking water, thus limiting our ability to determine whether weight loss influenced the FoxP3⁺ T_reg cell response and protective capacity. However, we observed that oral treatment with antibiotics in C57BL/6 mice reduced the severity of EAE without any significant impact on body weight. These results and unpublished preliminary data showing similar FoxP3⁺ T_reg cells in MLN and CLN of C57BL/6 treated mice suggest an active role of the alteration of the gut flora in the regulatory cell populations of mice.

The accumulation of FoxP3⁺ T_reg cells in the MLN and in the CLN might explain, at least in part, the protection observed in mice with reduced bacterial populations in the gut. The role of DCs in the generation of peripheral FoxP3⁺ T_reg cells has been extensively addressed (10, 19, 22) and it has been shown that gut CD11c^highCD103⁺ DCs enhance the conversion of naive CD4⁺ T cells into Foxp3⁺ T_reg cells (19). Moreover, the potential role for commensal bacteria in this conversion has been suggested (19, 20). CD103⁺ DC migration from the intestine to the MLN appears to involve the generation of T_reg cells (21). Our data suggest a role of MLN CD11c^highCD103⁺ DCs, but not CD11c^highCD103⁻ DCs, in the accumulation of FoxP3⁺ T_reg cells observed when the gut microflora is modified with antibiotics (Fig. 3). The role of TGF-β (23, 24) and RA (25, 26, 27, 28, 29, 30) in the generation of T_reg cells has been extensively studied. TGF-β and RA enhanced the T_reg cell conversion rates induced by mesenteric CD11c^highCD103⁻ dendritic cells (supplemental Fig. 4B) (19). We are currently investigating the role of RA and specific commensal components that could enhance TGF-β levels, or could be directly presented by DCs to naive T cells and induce the conversion and accumulation of FoxP3⁺ T_reg cells. The trafficking patterns from the gut to distal lymph nodes and possibly to the CNS is also under our investigation. A recently published work has reported that regulatory FoxP3⁺ T_reg cells isolated from germ-free animals have reduced suppressive effects on responder cells (31) when compared with naive specific pathogen-free mice. The differences with our model might be due to the fact that our system does not eliminate bacterial populations. Furthermore, the immune system in germ-free animals is not completely developed, as opposed to normal animals treated with antibiotics.

A recently published report showed a potential role for NKT cells in mediating protection against EAE in B6 mice in response to antibiotic treatment. The T_reg cell populations were reduced in the GALT similar to our observations in the PP (Fig. 2), but they were not studied in the distal peripheral lymph nodes (14). Our adoptive transfer experiments demonstrate that low numbers of CD4⁺CD25⁺ (FoxP3⁺ of >75%) T cells that accumulate in the CLN of mice treated orally with antibiotics confer protection against EAE, whereas the same numbers of naive CD4⁺CD25⁺ T cells fail in preventing the disease. When cytokine profiles were compared in the sorted cells used for the adoptive transfer experiments, a significantly enhanced production of IL-10 in CD4⁺CD25⁺ T cells sorted from orally treated mice was observed when compared with naive mice, which could explain the increased protective effect observed when low numbers of these cells were adoptively transferred. Our results confirm that transfer of CD4⁺CD25⁻ T cells with lower capacity to produce IFN-γ or IL-17 do not confer any protection against EAE, confirming the potential role of IL-10-producing T_reg cells in the reduction of the disease severity. When CD25⁺ cells were neutralized in vivo with anti-CD25 mAb, the protection conferred by antibiotic treatment was lost. Furthermore, CD25⁺ T cell depletion in these oral antibiotic-treated mice was associated with enhanced morbidity, as evidenced by increased cumulative clinical score and earlier clinical onset. These results confirmed the critical role of CD25⁺ T cells in disease prevention. However, despite CD25 depletion, clinical disease was significantly reduced in oral antibiotic-treated mice when compared with naive mice or mice treated i.p. with antibiotics and depleted of CD25 T cells. This may be due to other regulatory cell populations that might be important in the protection observed in Fig. 3 and might be related to the potential role for NKT cells in the protection against EAE in B6 mice in response to antibiotic treatment (14).

Our data suggest that the reduction of bacterial presence in the gut diminishes proinflammatory responses and also enhances the frequency of FoxP3⁺ T_reg cells in immune effector sites that are distal to the gut mucosa. It has been reported that IL-13 is protective against EAE (13, 32, 33). IL-13 levels were significantly enhanced in the brains of mice protected against EAE after oral treatment with antibiotics when compared with PBS- and i.p. treated mice. The role of IL-17 in the development of EAE has been of interest (34, 35). We found a significant reduction in IL-17 levels in spleens and CLN when animals were treated with oral antibiotics (Table I). Furthermore, treatment altered the percentages of FoxP3⁺ T_reg cells in the GALT, suggesting that oral antibiotic therapy can induce a shift in the T_reg/Th17 cell axis in normal mice. We postulate that the observed combination of enhanced T_reg cells in peripheral lymph nodes, a reduction in IL-17 and retinoic acid-related orphan receptor (ROR)γt, and a significant increase of IL-13 in the CNS are responsive to the modification in bacterial populations and accordingly confer the protection observed in SJL mice. Cytokine profiles of cells sorted from EAE-induced mice could confirm the immune switching toward an enhancement in regulatory and antiinflammatory Th2-type immune responses when mice are treated orally with antibiotics (Fig. 4).

Collectively, our data demonstrate that the reduction of commensal microflora prevents the development of EAE in SJL mice. When animals are recolonized with commensal bacteria, EAE clinical manifestations are reportable, whereas long-term control of bacterial populations with oral treatment with antibiotics confers complete protection against EAE. Our data do not reflect a direct neuroprotective effect of the cocktail of antibiotics. Protection might be caused by a nonspecific bystander effect of FoxP3⁺ T_reg cells and a reduction in the global levels of proinflammatory responses due to the reduced presence of bacterial populations and/or their products. We are currently comparing the microbial populations that selectively persist in the gastrointestinal tracts of mice treated with antibiotics that could be inducing the antiinflammatory reactions reported in this study.

The authors thank Dr. Azizul Haque, Dr. Jacqueline Y. Smith, John DeLong, Kathleen Smith, Alan J. Bergeron and Emily C. Colgate for technical support and critical review of the manuscript.

The authors have no financial conflicts of interest.