PubMed was used as search engine for this review. Combinations of the following keywords were used: microbiota, microbiome, experimental autoimmune encephalomyelitis, multiple sclerosis, probiotic, prebiotic, synbiotic, fecal microbiota transplantation, and diet (see Box 1). In addition, preliminary reports acquired from conference abstracts have been used in the case of multiple sclerosis microbiota studies, to increase the amount of studies that our findings are based on. The studies of which only abstracts were available are noted in Table 3 as author, abstract. Since this article is not a systematic review, the 155 cited articles are a balanced selection out of approximately 200 papers. PubMed was last checked for new articles on July 10, 2017.

Roughly 2.5 million people worldwide are affected by multiple sclerosis (MS), an autoimmune neurological disease of the central nervous system (CNS). Frequently observed symptoms are fatigue, numbness, loss of coordination, vision loss, dizziness, pain, cognitive defects, depression, and bladder and bowel dysfunction (1). MS can lead to serious motoric disability, as approximately 50% of patients require permanent use of a wheelchair 25 years after diagnosis (2). Although the cause of MS is not known, several lines of evidence point to a crucial pathogenic role of the immune system. Genome-wide association studies, neuropathological analyses, and successful therapy trials support the concept that peripheral interactions of environmental risk factors with MS-predisposing genetic factors elicit an autoimmune attack on the CNS causing the formation of lesions. Classically, lesions are defined as usually sharply edged demyelinated areas within the white matter with a variable degree of inflammation, axonal damage, and gliosis. The presence of immune cells (T and B cells, macrophages) and immune molecules (antibodies, complement) supports the characterization of MS as an autoimmune-mediated inflammatory disease.

A growing body of evidence indicates that gut microbiota can modify the incidence and/or course of immune-mediated, extraintestinal diseases (3). Since the gut microbiota are greatly impacted by diet, studies have also started to assess the contribution of gut microbiota and diet in MS and animal models of autoimmune demyelinating diseases such as experimental autoimmune encephalomyelitis (EAE). The aim of this review is to determine whether MS can be treated via modification of gut microbiota and dietary components. We describe key players of the immune system that are involved in the pathogenesis of MS, and we discuss how the cross talk of gut microbiota with the immune system can affect the expression of MS/EAE. We then review recent studies on gut emicrobiota and dietary components in the animal model of EAE and MS patients. We also review how gut microbiota can be modulated and propose future research topics.

Multiple sclerosis presents itself in several forms of symptoms and disease courses. 85% of patients are diagnosed with relapsing remitting (RR) MS, which is characterized by alternating episodes of neurological symptoms (relapses) and recovery (remissions). During relapses, inflammatory active lesions can be detected in the CNS with imaging techniques, such as magnetic resonance imaging (MRI). As the disease progresses, recovery declines and 80% of RR MS patients develop secondary progressive (SP) MS within 10–20 years after initial diagnosis. SP MS does not come with periods of relapses and recovery, but increased disability gradually occurs along with axonal loss and decreased brain volume (atrophy). 10% of MS patients are diagnosed with primary progressive MS, which presents itself as SP MS directly from the start of the disease (2). The cause of MS is unknown, but various concepts have been proposed to explain disease etiology. An inside-out concept suggests that MS starts as a primary lesion in the CNS, for instance caused by an infection or by primary neurodegeneration inducing the release of self-antigens against which autoreactive T and B cells react. An outside-in concept, supported by animal research in EAE models, postulates that autoreactive T cells that have escaped thymic selection are activated by a peripheral infection. This activation can occur through antigens that closely resemble CNS antigens, bystander activation, novel autoantigen presentation, or recognition of sequestered CNS antigens. In both concepts, activated CD4⁺ T_h1 and Th17 cells infiltrate into the CNS where they can be reactivated by resident antigen-presenting cells (APC), microglia for example. CD4⁺ T_h1 and T_h17 produce IFNγ and IL-17A, and the inflammatory reaction that follows increases the permeability of the blood–brain barrier and recruits other immune cells such as B cells and monocytes to the CNS. The inflammatory milieu also activates microglia, which in turn produce pro-inflammatory mediators which elicit demyelination and axonal loss (4). Naïve CD4⁺ T cells infiltrating the CNS can broaden the pattern of autoimmune reactions by epitope spreading, adding to the inflammatory milieu (2). CD8⁺ T cells likely also contribute to MS pathogenesis. They are found in high frequency in demyelinating lesions and correlate with axonal damage. Myelin-specific CD8⁺ T cells can be activated by epitope spreading, and up to a quarter of CD8⁺ T cells in active lesions are thought to be mucosa-associated invariant T (MAIT) cells. MAIT cells are able to produce IL-17 and are associated with the gut and liver (5). MS patients have increased serum levels of IL-18 (6), which activates MAIT cells and is inversely correlated with MAIT cell blood frequency (7). When costimulated by IL-18 and T-cell receptor stimulation, MAIT cells upregulate integrin very late antigen-4, which is involved in cell migration into the CNS (7). In addition, MAIT cells are depleted after efficacious autologous hematopoietic stem cell transplantation, a treatment which can be beneficial in MS patients (8). Therefore, MAIT cells are likely involved in the immunopathogenesis of MS. The relative significance of the different T cell subsets in human MS is not yet completely understood, as EAE is usually induced via complete Freund’s adjuvant (CFA), and interspecies immunological differences exist. Additionally, a more predominant CD4⁺ driven disease course is seen in EAE, while a more CD8⁺ T cell-driven immune response is seen in MS (2).

In addition to T cells, autoreactive B cells may be activated in the periphery. B cells infiltrating the CNS locally produce autoantibodies, which bind myelin and cause damage to myelin via complement- and/or macrophage-mediated cytopathic reactions (CDC and ADCC). These B cells are also able to migrate out of the CNS and mature in the lymph nodes before migrating back to the CNS (9). Protective T_reg cells and their anti-inflammatory effects could also be defective. Protective cells include CD4⁺ Foxp3 expressing T_reg, IL-10 producing T_r1, and CD39⁺ T_reg cells. In MS patients, these cells are found in reduced frequency in the periphery and they have reduced immunosuppressive capacity compared to healthy individuals. Other regulatory cell types could also play a role, as they may increase after treatments. These include the CD8⁺ T_reg and IL-10 producing B_reg cells. In addition to defective regulatory cells, effector cells may escape their regulation when they are less sensitive to the suppression by T_reg cells (2). A clear imbalance of effector cells and regulatory cells is seen in early MS, which leads to a pro-inflammatory milieu in the CNS and promotes demyelination and axonal damage (see Figure 1) (2, 4, 10). In later stages of the disease immune cell migration from the periphery into the CNS subsides, but chronic CNS inflammation and neurodegeneration may continue to take place. This is associated with the formation of tertiary lymphoid-like structures within the CNS and associated meninges and with dysfunction of astrocytes and microglia. Microglial activation can promote dysfunction of astrocytes, after which astrocytes inhibit the maturation of myelin producing oligodendrocyte progenitors, resulting in reduced remyelination. Astrocytes can also produce CCL2 and GM-CSF, which further recruits and activates microglia and creates a self-sustaining feedback loop. Pro-inflammatory mediators such as reactive oxygen species (ROS) produced by astrocytes and microglia are neurotoxic, and the continuous CNS inflammation promotes gradual neurodegeneration (2) (see Figure 2).

The human gut serves as a host to many microbes (bacteria, archaea, viruses, and fungi). Newborn humans have a sterile gut, and colonization occurs through exposure to new flora depending on the mode of delivery, diet, and hygiene (11). Over the years, the gut microbiota increase in diversity and reach a maximum at adolescence, after which they remain fairly stable (11). The diversity of gut microbiota can be expressed as α-diversity, which shows the richness and distribution of taxa within one population. β-Diversity is used to measure differences between multiple populations, and it measures how many taxa are shared between populations (12). Gut microbiota display a low diversity at the phylum level, as bacteria from only 8 out of 55 phyla have been detected in the human gut (13). Yet, large interindividual variation exists in the relative abundance of microbiota members. This variation is not only greatly impacted by short-term dietary alteration but also influenced by long-term dietary habits, host genotype, and stochastic processes such as history of colonization and ecological processes, such as selection and evolution (14) (see Figure 3). Additionally, the use of antibiotics can reduce the number and diversity of gut microbiota, but these are largely restored to the pretreatment composition after a recovery period (15).

Studies investigating the human gut microbiota often use the sequencing of specific genes such as the 16s ribosomal RNA gene in stool samples to determine which species are present. This has to be interpreted with care, as the microbiota composition differs from the small intestine to the large intestine. The abundance of microbiota is increased along the gastrointestinal tract and while the jejunum is host to mainly aerobic species, the colon is dominated by anaerobic species. Sequencing of stool samples is therefore most informative of microbiota in the large intestine (16).

In addition, gut microbiota may influence distant host tissues such as the CNS. Accumulating evidence indicates that gut microbiota affect various behaviors such as social interaction (26), nociceptive responses (27), depression (28), stress responsiveness, and anxiety (29). Gut microbiota also influence hippocampal neurogenesis (30), blood–brain barrier integrity (31), and microglia maturation (32).

Gut microbiota affect the CNS through multiple mechanisms. The gut enteric nervous system controls the motility and homeostasis of the gut, which influences the gut microbiota composition (33). Conversely, gut microbiota influence the enteric nervous system, and the vagus nerve provides a direct communicatory link between the gut and the CNS. Gut microbiota can produce neurotransmitters or precursors such as tryptophan, and microbiota metabolites can also directly influence the brain. For instance, fermenting bacteria produce SCFA, which can translocate into the brain and inhibit deacetylases, resulting in epigenetic changes (32). In addition, bacterial RNA, DNA, and proteins are detected in the human brain (34). Fragments of bacterial cell walls, such as peptidoglycan (PGN) may translocate to the brain and possibly influence brain development and social interaction (35). Bacteria also produce microbial-associated molecular patterns, which can be recognized by the host immune system with pattern recognition receptors, such as Toll- or Nod-like receptors and influence gut physiology (32, 35). The GALT is located along the small and large intestines and functions as immune surveillance of the gut. Gut microbiota can elicit an immune response in the GALT (36), which can be pro-inflammatory or anti-inflammatory, depending on the involved microbiota (37). The pro-inflammatory response of segmented filamentous bacteria (SFB) is best characterized. Studies in mice show that SFB enhance antigen presentation by DC resulting in greatly increased numbers of pro-inflammatory intraepithelial lymphocytes (IEL), such as γδ T cells and CD8⁺ T cells. SFB also increase IgA⁺ B cells and Th17 cells (13, 36). Th17 cells produce the cytokines IL-17A, IL-17F, IL-21, and IL-22. IL-17A and IL-17F control bacterial and fungal infections through the recruitment of neutrophils and more Th17 cells and increased production of β-defensin by epithelial cells. IL-22 induces epithelial cell proliferation, survival, tissue repair, and increased expression of antimicrobial molecules (13). In addition, gut bacteria such as Helicobacter hepaticus elicit a pro-inflammatory response, marked by increased numbers of T_h17 and Th1 cells and pro-inflammatory cytokines excretion (38). Th1 cells are instrumental in fighting intracellular bacteria and viruses, via the production of IFNγ (13).

Gut microbiota are potent activators of innate lymphoid cells (ILC). ILC arise from common lymphoid precursors and respond rapidly to cytokines produced by the epithelium. ILC can be divided into three groups based on molecular markers: type 1 which expresses T box transcription factor (TBX21 or T-bet) in T cells, type 2 which expresses GATA binding protein and type 3 which expresses retinoid-related orphan receptor (ROR)γt. Type 3 ILC produce IL-17 and IL-22, and IL-22 production can be induced by bacterial metabolites acting on the aryl hydrocarbon receptor (AHR) (17). The pro-inflammatory reactions caused by gut microbiota may contribute to MS pathogenesis via the activation of autoreactive T-cells (36). Conversely, gut microbiota may also elicit an anti-inflammatory response, characterized by the induction of CD4⁺ T_reg cells. T_reg cells produce TGF-β and IL-10, which decrease pro-inflammatory cytokine production and cell proliferation. Clostridia species promote production of TGF-β, which helps maintain an anti-inflammatory milieu and is involved in the induction of T_reg cells (13, 36). Additionally, Bacteroides species increase the production of anti-inflammatory cytokines and suppress IL-17 production (36).

Gut microbiota also affect invariant natural killer T cells (iNKT cells); GF mice have increased iNKT cell numbers in their colon. These are a pro-inflammatory subset of T cells that express the invariant T-cell receptor α, with which they detect lipid antigens presented by CD1d molecules. They can promote T_h1 and T_h17 responses by producing IFNγ, IL-2, IL17A and tumor necrosis factor (TNF) (13).

The influence of phages on the immune system is likely mostly indirect, through modulation of bacteria by mechanisms mentioned earlier. Phages may also directly communicate with the immune system. Humans are constantly sensitized with phage antigens present in food. This induces a constant low level of phage-neutralizing antibodies in human serum. After systemic phage administration in animals, these neutralizing antibodies are produced at higher titers. In humans, administration of enterobacteria PhiX174 phages induced IgM antibodies, followed by IgG antibodies after a second exposure (39). The function of antibodies is clearance of phages as illustrated by the lower clearance in B cell-deficient mice. A reduction of macrophages, natural killer, or T cells does not influence phage clearance (40). Conversely, phages can influence T cell functions in vitro. Mycobacteriophages were found to suppress phytohemagglutinin-induced activation of T lymphocytes from guinea pigs in a dose-dependent matter (41). Moreover, purified T4 phages can inhibit human T cell proliferation via the CD3 T-cell receptor complex, while phage lysates of Staphylococcus aureus can induce T-cell proliferation (41). In addition, phages can affect host phagocytes. Phagocytosis and ROS production may limit bacterial densities in the gut. Excessive phagocyte activity may also lead to pathology. Preincubation of monocytes and neutrophils with T4 and F8 phages from Pseudomonas aeruginosa reduced in vivo phagocytosis of Escherichia coli. T4 phages also reduced the ROS production by phagocytes exposed to LPS and bacterial cells, although this effect may be phage strain-dependent, because staphylococcal A3/R-purified phage (lysates) did not affect ROS production (41). The effect of phages depends also on the type of preparation, as T4 lysates promote IL-6 production by monocytes but purified T4 phages do not (41). Most effects of phages are anti-inflammatory, as reflected by mitigated phagocytosis, ROS production, and lymphocyte proliferation. However, phage lysates of S. aureus are likely more pro-inflammatory. Overall, effects of phages on the host immune system are complex and depend on phage identity, phage tropism, and preparation.

Collectively, variable exposure to the environment and food greatly influences gut microbiota composition. Elements of the gut microbiota constantly interact with each other and with the host and are essential for the normal development of the immune system and CNS. Pro- and anti-inflammatory effects have been attributed to different microbial species and an imbalance may contribute to autoimmune disease may contribute to autoimmune disease (see Figure 4).

Due to the inaccessibility of the CNS of MS patients, animal models are used for translational research into the pathogenesis of MS and for therapy development. As in none of the available models the immunological and pathological complexity of MS is fully replicated, multiple models are being used. Depending on the research question models of MS-like demyelination are induced chemically (cuprizone, lysolecithin), with neurotropic viruses (Theiler’s murine encephalitis virus, Semliki Forest virus), through specific cytokine overexpression in the CNS, diphtheria toxin-based depletion of oligodendrocytes, and active immunization with myelin components in suitable adjuvants (42, 43). The active immunization model, known as EAE, is the most widely used preclinical MS model. EAE can be induced in a wide variety of laboratory animal species including mice, rats, rabbits, guinea pigs, and non-human primates by inoculation of myelin antigens formulated with a strong adjuvant, such as CFA. For synchronous EAE induction at high incidence in mice, the immunization is usually supplemented with injection of Bordetella pertussis toxin. Different myelin antigens and dosages are used to model the heterogeneity of MS. For example, SJL mice develop a RR type of EAE after sensitization against residues 79–87 of myelin basic protein (MBP_79–87) or residues 131–151 of proteolipid protein (PLP_131–151). C57Bl/6 mice can develop RR EAE after immunization with a low dose of myelin oligodendrocyte glycoprotein residues 35–55 (MOG_35–55) and develop chronic EAE without remission upon immunization with a high dose of MOG_35–55 (44). It is increasingly felt that the usage of strong adjuvants precludes a role of subtle regulatory mechanisms. For such studies, passive EAE models induced by the transfer of activated T cells from a donor with actively induced EAE or spontaneous EAE models established in mice expressing transgenic T and/or B cell receptors specific for myelin antigens may be more useful (43).

Although EAE models have proven their relevance as a preclinical test system for new therapeutics they also have their limitations. The highly artificial way by which the autoimmune process is activated in CFA-based EAE models does not necessarily reflect the natural immune response toward self-antigens. As an illustration, the EAE model is dominated by CD4⁺ T cells, whereas CD8⁺ T cells are likely more immunodominant in MS patients (44). Another difference between EAE and MS is that mice with EAE mostly show lesions in the spinal cord, whereas MS patients mostly show lesions in the brain (43). A noticeable exception is formed by the atypical EAE models in marmosets, a small-bodied Neotropical primate, which more closely approximate MS with respect to clinical, pathological, and immunological presentation (45).

The first evidence that gut microbiota are involved in the pathogenesis of EAE stems from decades ago. Already in 1993, it was discovered that transgenic mice expressing T-cell receptors specific for MBP fail to develop EAE when they were housed in a sterile, specific pathogen-free (SPF) environment, while mice housed in a non-sterile environment did develop EAE (46). To determine the role of gut microbiota in EAE, GF models, antibiotic treatments, probiotic mixtures, bacterial products, and diet-based interventions have been used. These studies are summarized in Tables 1 and 2.

The notions that SCFA (<6 carbons) produced by gut microbiota modify immune functions and that SCFA are consumed through the diet underlie the hypothesis that the diet influences microbiota, the immune system, and ultimately EAE. Prophylactic oral treatment with propionic acid, a SCFA that is ingested with food, increases T_reg cells and reduces the EAE disease score. In contrast, long-chain fatty acids (LCFA; 13–21 carbons) increase EAE scores, which is associated with increased T_h1 and Th17 cells (64). Of note, SCFA can migrate into the CNS and serve as fuel for CNS neurons and glial cells. The microbiota composition of these mice also differs, as those fed with lauric acid, a saturated LCFA, have reduced Prevotellaceae and Bacteroidetes in their gut microbiota. This illustrates the impact that dietary components can have on the immune system and EAE expression, possibly through the modification of gut microbiota. Fatty acids can also have beneficial effects on EAE, regardless of their length, when they are unsaturated. These lipids have one or more double bonds between the carbon atoms of their hydrocarbon chain. Fatty acids with a double bond at the third carbon atom counted from the methyl (–CH3) tail of the chain are Ω-3 fatty acids and with a double bond at the sixth carbon atom are Ω-6 fatty acids. Oral treatment with the Ω-6 fatty acid γ-linolenic acid reduced the EAE clinical scores in MOG_92–106-immunized SJL mice. This was associated with increased TGF-β, prostaglandin E₂ production by spleen mononuclear cells (71). Oral treatment with oils containing γ-linolenic acid also reduced the EAE clinical scores in Lewis rats, where EAE was induced with guinea pig CNS matter homogenate (72). Ω-3 fatty acids can also reduce the EAE clinical scores. A diet rich in docosahexaenoic acid (DHA) starting 5 weeks before induction of EAE with MOG_35–55 reduced the EAE clinical scores in C57Bl/6 mice. This was found associated with reduced T_h1 and Th17 cell differentiation, and reduced amounts of these cells were found in the spleen and spinal cord of mice on a DHA-rich diet. In vitro, DHA reduced the expression of costimulatory molecules on DC and reduced their production of pro-inflammatory cytokines (73). A diet with the triglyceride form of DHA, starting before EAE induction also reduced the EAE clinical scores in the same mouse model. In vitro, pretreatment of microglia cells with the triglyceride form of DHA reduced microglial oxidative stress and production of nitric oxide and inflammatory cytokines (103). Eicosapentaenoic acid (EPA) supplementation starting 7 days after EAE induction reduced the EAE clinical scores in MOG_35–55-immunized C57Bl/6 mice. EPA is a ligand for PPAR α, β, and γ and increased their expression on CD4⁺ T cells in the spinal cord, while reducing IFNγ and IL-17 cytokine production. CD4⁺ T cells from the spleen of EPA-treated mice expressed increased mRNA levels of Foxp3, but also of IL-17 and RORγt (74). When an EPA rich diet was given at the time point of EAE induction, it delayed the time before EAE symptoms appeared in guinea pig spinal cord homogenate-immunized dark agouti rats. This was associated with increased myelination of axons in the spinal cord (75). A ketogenic diet, characterized by a high fat to protein and carbohydrate diet, also reduced the EAE clinical score when given before EAE induction. This was associated with reduced T_h1 and Th17 cells in the CNS and lymph nodes (the authors did not specify which lymph nodes). Additionally, mice on a ketogenic diet had reduced levels of pro-inflammatory cytokines in their lymph nodes and CNS (76). Another study confirmed the beneficial effect of the same ketogenic diet but did not investigate effects on the immune system (77). The proportion and types of lipids in the diet influence the gut microbiota composition. In healthy C57Bl/6 mice, mice were subjected to a high fat diet containing palm oil (high in saturated fatty acids), olive oil (high in monounsaturated fatty acids), safflower oil (high in Ω-6 polyunsaturated acids), or flaxseed/fish oil (high in Ω-3 polyunsaturated fatty acids). These diets were also compared to two low-fat diets, of which one contained a high percentage of calories from maize and one diet in which most calories came from sucrose. The cecum contents were tested for microbiota composition. In the palm oil group, the relative abundance of Bacteroidetes and Bacteroidaceae was reduced, while Lachnospiraceae was increased. Mice that received an olive oil-based diet had an increased proportion of Bacteroidaceae. In the flaxseed/fish oil group, the proportion of Bifidobacteriaceae and Bifidobacterium was increased. Within the low-fat diets, the relative abundance of Ruminococcaceae was increased and Erysipelotrichaceae was reduced in the high sucrose diet. Interestingly, cecal concentrations of SCFA were increased in the palm oil supplemented group (104). This indicates that dietary fats greatly influence the gut microbiota composition, which influences SCFA production or absorption. In this way, shifts in gut microbiota composition as a result from dietary fats may influence EAE/MS.

Other dietary components can also modify EAE disease scores. Potentially relevant for MS is the beneficial effect of vitamin D3 on EAE, as vitamin D3 deficiency is an established MS risk factor. Intraperitoneal administration of vitamin D prevents EAE development in SJL/J mice immunized with rat spinal cord homogenate when given prophylactically (65). Additionally, dietary vitamin D supplementation reduces EAE scores when given to MBP_79–87-immunized B10.PL mice (66) and MOG_35–55-immunized C57Bl/6 mice with established EAE, in which the effect was IL-10 dependent (67). DC isolated from MOG_35–55-immunized C57Bl/6 mice were also able to reduce EAE severity when incubated with 1,25-dihydroxyvitamin D3 prior to adoptive transfer into mice with EAE. Even though the proportions of T_h1 and Th17 cells in lymph nodes and the spleen were increased during this treatment, their proportions in the spinal cord were reduced. Therefore, DC that were exposed to 1,25-dihydroxyvitamin D3 may reduce migration of pathogenic T cells from the periphery into the CNS (105). In vitro culture with high concentrations of vitamin D inhibits CD4⁺ T cell proliferation, reduces IL-6, IL-17 producing T-cells while enhancing IL-10 producing and CD4⁺CD25⁺Foxp3⁺ T cells (106). In other models, vitamin D reduced demyelination and increased remyelination (107) and exerted additional anti-inflammatory effects, such as inhibition of the T_h1 and B-cell response and modulation of DC (108). Vitamin D may also inhibit the bacteria-induced pro-inflammatory NF-κB pathway and affect tight junction expression, contributing to proper intestinal barrier function (108).

Excess calorie intake and/or fasting may have a negative effect on MS/EAE. A prophylactic diet of 66% caloric restriction protected Lewis rats from developing EAE (69), while a prophylactic caloric restriction of 40% reduced the EAE score in several mouse models, which was associated with reduced spinal cord inflammation, demyelination, and axonal injury (68). Additionally, increased plasma levels of corticosterone, adiponectin and reduced plasma levels of IL-6 and leptin were found in animals with reduced food intake (68). These altered plasma concentrations are interesting, as corticosterone has broad inhibitory effects on the immune system, and adiponectin reduces IL-6 and TNFα production as well as induces production of IL-10R and IL-1R antagonists. Leptin induces T cell proliferation, T_h1 differentiation, and pro-inflammatory cytokine production (68). Intermittent feeding (ad libitum access to food on alternating days) also had a positive effect on the EAE clinical score, but its effect on the immune system was not investigated (70). The beneficial effect of fasting on EAE disease score was confirmed in another study, in which mice were fed in cycles of 4 days ad libitum, followed by 3 days of fasting (first day of fasting 50%, then 2 days of 10% of normal caloric intake). This intervention was able to reduce the clinical EAE score, even in mice with established EAE. The suppression of EAE was associated with reduced immune cell infiltration into the spinal cord, reduced splenic CD11⁺ DC, reduced pro-inflammatory cytokines, increased corticosterone in serum, increased T_reg, and reduced T_h1 and Th17 cells in the lymph nodes and spleen. In addition, fasting protects oligodendrocytes from apoptosis and stimulates maturation of oligodendrocyte precursors (77). The effect on oligodendrocytes is seen in EAE and also in a cuprizone model, which demonstrates that fasting affects oligodendrocytes also in an autoimmunity-independent way (77).

Increased salt intake increased the EAE disease score in MOG_35–55-immunized C57Bl/6 mice (78–81). This effect was associated with increased activity of Th17 cells in the spinal cord and spleen. It is thought that the effect of high sodium is DC independent, as DC function is not altered by exposure to salt (78). Instead, sodium directly influences T_h17 differentiation. Two studies found that high sodium induces the expression of serum/glucocorticoid kinase 1 (SGK-1) in naïve T cells (80, 81). SGK-1 promotes IL-23R expression and induces T_h17 differentiation (80). The importance of SGK-1 signaling is shown by the observation that a high-salt diet increases the severity of EAE in mice, while SGK-1-deficient mice develop less severe EAE. This effect was associated with reduced Th17 cells in the CNS and mesenteric lymph nodes in SGK-1-deficient mice (80). Dietary tryptophan can be metabolized into a variety of AHR ligands. AHR ligands have shown different effects on EAE when given prophylactically. The tryptophan-derived AHR ligand 6-formylindolo[3-2b]carbazole (FICZ) accelerates EAE onset and increases pathology in MOG_35–55-immunized C57Bl/6 mice. This is associated with increased IL-17 and IL-22 producing CD4⁺ T cells in the spinal cords (82). Another study verified that FICZ increases EAE severity and shows that FICZ-treated mice have increased IL-17⁺CD4⁺ and IFNγ⁺CD4⁺ T cells in the spleen (83). In addition, they show that tryptophan-derived 2-(1′H-indole-3′carbonyl)-thiazole-4-carboxylic acid methyl ester (ITE) reduces EAE severity. Stimulation with 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), a less natural AHR ligand, induces T_reg cells, which are protective of EAE also after adoptive transfer (83). Even though in these studies the AHR ligands were injected rather than supplemented through diet, it shows the immunomodulatory effects of different AHR ligands which may also be taken through the diet. AHR signaling may also have protective effects through astrocytes. Mice fed a tryptophan-deficient diet develop increased EAE severity. This effect could be reversed by tryptophan supplementation in control mice, but not in astrocyte-specific AHR knockout mice. Gut microbiota mediate the conversion of tryptophan into AHR ligands. Depletion of these microbiota by ampicillin treatment increased EAE disease scores, which can be reduced by supplementation with multiple tryptophan metabolites and the bacterial enzyme tryptophanase. This shows a protective effect of the interplay between dietary tryptophan, gut microbiota, AHR ligands, and astrocytes (84).

Zinc can also influence EAE. In vitro, salt composed of zinc with the amino acid aspartate (zinc aspartate = ZnAsp) reduces the proliferation of stimulated human T-cells (85, 86) and also reduces their production of IL-2, IL-10, IL-17 (85), IFNγ, TNFα, GM-CSF, and IL-5 (86). The same effects of ZnAsp were seen in mouse splenocyte cultures (85, 86). In PLP_139–151-immunized SJL mice, intraperitoneal injection of 30 µg ZnAsp per day reduced the EAE clinical score prophylactically as well as therapeutically, while 120 µg increased the EAE severity (85). An oral dose of 6 or 12 µg ZnAsp per day reduced the clinical disease score in the same EAE model, while an oral dose of 30 µg increased the disease score (86). Zinc supplementation also affected EAE in the MOG_35–55-immunized C57Bl/6 model. Zinc supplementation in drinking water (87) and daily intraperitoneal injections of 6 and 30 µg ZnAsp both reduced the EAE clinical score when given before EAE induction (88). The mechanisms underlying the effect of ZnAsp injections include reduced systemic Th17 cells and increased Foxp3⁺ T-cells in the spinal cord (88). Lower plasma or serum concentrations of zinc have been found in MS patients compared to healthy individuals, while others have found no differences. Mice with EAE had reduced zinc plasma levels on day 21 after EAE induction compared to naïve mice (86). Given the beneficial effects of zinc supplementation in EAE, normalizing zinc levels in MS patients with zinc deficiency may be tested in clinical trials, but a high dosage or long-term supplementation may also have detrimental effects, thus patient serum concentrations in these trials must be tightly monitored (86). A zinc-deficient diet of 10 days did not alter the microbiota composition at phylum level and total bacterial abundance in healthy C57Bl/6 mice compared to those that were fed a normal diet. The relative abundance of several bacterial genera did differ between these diet groups, which included the genera Enterococcus, Enterobacteriaceae, Paenibacillus, Granulicatella, Clostridium, Akkermansia, and Burkholderia (109).

High dietary selenium (10× normal intake) increased EAE incidence and severity in guinea pigs, while those on normal and half amounts of selenium developed EAE in a normal fashion (89).

Intraperitoneal injection of diphenyl diselenide was toxic for whole MBP-immunized Wistar rats, while oral administration 1 week after immunization resulted in reduced EAE incidence and symptoms. This was associated with reduced macrophage numbers in the CNS and reduced MBP-specific T-cell proliferation. The EAE-reducing mechanism of diphenyl diselenide is thought to be based on reduced NF-κB signaling in macrophages and T-cells. In addition, diphenyl diselenide may increase ROS clearance as it mimics glutathione peroxidase activity, which protects against oxidative damage (90). The different effects of dietary selenium on EAE may be explained by differences in formulation.

Vitamin E has potent antioxidant properties. Multiple compounds are considered part of the vitamin E group, of which tocopherol is best studied. Intraperitoneal administration of tocopherol reduced the EAE clinical scores in MOG_35–55-immunized C57Bl/6 mice, which was associated with reduced MOG-specific splenocyte proliferation. In addition, splenocytes incubated with tocopherol produced less IFNγ (91). Vitamin E also increases remyelination and reduces demyelination in an animal model where demyelination is chemically induced (107). TFA-12, a synthetic tocopherol derivative, reduced the EAE clinical scores in MOG_35–55-immunized C57Bl/6 mice when injected intraperitoneally at the onset of EAE symptoms. This was associated with reduced inflammation of the CNS, astrogliosis, and demyelination. TFA-12 also accelerated remyelination in a chemically induced demyelination model. The mechanism behind this is believed to be due to the induction of oligodendrocyte maturation (92).

Vitamin A and its metabolites all-trans retinoic acid and 9-cis-retinoic acid can also reduce EAE severity. Prophylactic dietary 13-cis-retinoic acid and 4-HPR (a synthetic retinoid derivative) reduced EAE incidence in SJL mice which received MOG-incubated lymph node cells to induce EAE. Dietary 4-HPR can also reduce EAE severity in a therapeutic manner (93). In MOG_35–55-immunized C57Bl/6 mice, intraperitoneal injection with all-trans retinoic acid reduced the EAE clinical score when given prophylactically (94) as well as therapeutically (95). This was associated with reduced DC maturation and reduced pro-inflammatory monocytes in the subarachnoid space. Reduced T_h1 and Th17 cells were also found in the draining lymph nodes. In addition, bone mesenchymal DC pretreated with all-trans retinoic acid were able to reduce T_h1 and T_h17 differentiation and lymphocyte proliferation in vitro (94). All-trans retinoic acid also directly affected naïve CD4⁺ T cells, as it inhibited T_h17 differentiation by reducing expression of the IL-6 and IL-23 receptor (95). Even though all-trans retinoic acid can induce Foxp3⁺ T cells in vitro, the frequency of these cells was not increased in EAE-affected mice treated with the metabolite (95). In vitro, 9-cis-retinoic acid reduced pro-inflammatory cytokine production in LPS-stimulated microglia and reduced TNF-α and nitric oxide production in astrocytes (110). Mechanisms underlying gut microbiota and dietary interventions are summarized in Box 2.

How different food components may affect MS has been reviewed comprehensively by Schmitz et al. (141) of which the most important ones are summarized in Table 4. Multiple dietary intervention studies have been done in MS patients, but most were unsuccessful in reducing MS severity and have not examined the effects on microbiota. Vitamin D supplementation did show positive outcomes as it led to fewer relapses and reduced pro-inflammatory cytokines such as IFNγ and IL-4 in T-cells. Additionally, vitamin D supplementation increased peripheral IL-10⁺CD4⁺ T cells, suppressed T-cell proliferation, and reduced the number of gadolinium-enhancing regions per patient (13, 141). However, these results came from small studies and more randomized controlled trials investigating the effect of vitamin D for MS treatment are now taking place. The effect of vitamin D might in part be exerted via the gut microbiota, as in healthy individuals high vitamin D intake was associated with increased proportions of Prevotella and reduced proportions of Haemophilus and Veillonella (148). Vitamin D supplementation in vitamin D-deficient MS patients altered the relative proportions of different genera, but this study included very few subjects (123). In a study of 70 and a replication study of 59 RR MS patients, the influence of salt on MS was studied. As 80–90% of salt intake is excreted in urine, salt excretion rate in urine was measured as a proxy for salt intake. High salt excretion was associated with increased relapse rates, increased risk on developing a new lesion detected by MRI scans, and increased T2 lesion load. However, causality was not established. Patients with relapses are often treated with steroids, which influences salt excretion (137). Clinical trials with controlled salt intake are needed to assess the impact of increased salt intake on MS severity.

In a cohort of 2,087 MS patients, a general healthy diet (based on fruit/vegetable, fat, meat, and dairy consumption) was associated with better clinical scores (138). Therapeutically, two studies have shown effect of a general healthy diet (low intake of saturated fatty acid, sugar, coffee, alcohol, and high intake of vegetables, fruits, fish and whole grain products) on MS clinical scores when combined with supplementation of oils rich in Ω-6 or Ω-3 fatty acids (139). Of note, one of these studies also supplemented multiple vitamins including vitamin A, B, D, and E (140). Since these studies used multiple dietary components as intervention, it is hard to determine which part of the intervention exerted the beneficial effects. A difference in fat consumption may be responsible, since a third study found that reduced fat intake supplemented with Ω-3 fatty acids reduces MS clinical scores (127). It is noteworthy that the abovementioned studies related results to baseline measurements instead of a control group. Because dietary studies in MS can hardly be blinded and patients in the control groups have little incentive to complete a study, control groups have high dropout rates. Efforts are now being made to develop study designs that increase the participation of patients in control groups and to test if an observation period prior to a test diet can serve as a control group (149). Multiple studies are currently listed on www.clinicaltrials.gov as recruiting or ongoing, including studies testing diets such as ketogenic, low sodium, low saturated fat, caloric restriction, and a variation of Paleolithic diet (low grain, high vegetables). These studies should provide additional insight into which compounds may be responsible for the positive effects of a generally healthy diet. Preliminary results of oral treatment with propionic acid are promising. Healthy individuals and patients were treated with propionic acid, which increased T_reg frequency and reduced T_h17 frequency in both groups without any side effects; only an abstract of this study is available (136).

Current evidence suggests that MS patients may have a different gut microbiota composition compared to healthy individuals and that gut microbiota may affect the relapse rate (see Table 3). If these results will be verified in larger studies, it would be interesting to test whether manipulation of the gut microbiota of MS patients induces immunosuppression. Multiple approaches can be used to manipulate the gut microbiota (see Box 3). Probiotics involve the treatment with live microorganisms that are absent or present in too low abundance in the host. In this way, bacteria that induce immunosuppression can be introduced in patients. Probiotics are already being used in intestinal diseases such as IBD or antibiotic-associated diarrhea (17). Prebiotics are food ingredients with which gut microbiota may be modified. Resistant starches, which can reach the colon before being metabolized, alter the gut microbiota and have been associated with increased butyrate production. Additionally, fibers can be fermented by gut microbiota that produce SCFA. However, the effect of prebiotics depends on microbial strains that are already present in the host. Therefore, high individual variation and a large number of non-responders are seen (150). Probiotics and prebiotics can also be used in conjunction, i.e., as synbiotic, to augment their individual effectiveness. The probiotics part introduces new strains, while the prebiotic part synergistically increases the efficacy of the introduced species due to increased survival, competition, or metabolic activity (150). Antibiotics can be used to remove or suppress unwanted, immunogenic strains from the gut microbiota. A downside of antibiotic treatment is that multiple strains are targeted and beneficial bacteria may also be depleted. This increases the risk of opportunistic infections. Moreover, in mice, vancomycin treatment disrupts complex carbohydrate fermentation and leads to reduced concentrations of SCFA, including butyrate and propionic acid (151). Phages may be employed for more specific targeting of immunogenic parts of the gut microbiota. In this way, beneficial bacteria remain untouched and immunosuppression may still occur. Phages may also be used to increase the competitiveness of strains that need to be introduced. However, phage therapy still requires more research before it can be implemented, as there is limited evidence of efficacy in human clinical trials (152). Fecal microbiota transplantation involves the administration of a liquid filtrate of a healthy person’s stool to a diseased recipient. In this way, the recipient gut microbiota can be modulated to resemble the gut microbiota of the donor. New strains may be introduced by the transplantation, and through competition and possibly cross talk between gut flora and the immune system, excessive strains may be reduced and dysbiosis could be normalized. Fecal microbiota transplantation has been effective in small randomized controlled trials for the treatment of recurrent Clostridium difficile infection. Based on these successes, clinical trials are now being done to investigate the efficacy of fecal microbiota transplantation in IBDs (153). Fecal microbiota transplantation may be a relatively safe option for restoration of a healthy microbiome in MS patients. Instead of manipulating microbiota, microbial products that induce immunosuppression may also be used for therapeutic purposes. Chemical alterations may have to be made to make sure that the compounds reach their site of action and act through their intended mechanism. SCFA are produced by bacteria in the colon, but when orally administered, a large proportion of SCFA are thought to be absorbed in the small intestine (100). Therefore, dosage and kinetic studies will have to point out how optimal use can be made of these products. The advantage of bacterial compounds over live bacterial species is that bacterial compounds can be dosed like any other drug, while live bacterial species are affected by colonization and may therefore lead to variable interindividual effects (37).

This review shows that many different microbial species and dietary interventions affect EAE disease expression and differences in gut microbiota between MS patients and healthy individuals may exist. Therefore, altering the gut microbiota of MS patients, possibly through dietary interventions is a potential therapeutic aim. Thus far, multiple small studies found that SCFA producing bacteria are reduced in patients with MS, and the PSA producing bacterium B. fragilis may also be reduced although B. fragilis is only found differently present in the gut by one study. Potentially pro-inflammatory Methanobrevibacter and Enterobacteriaceae may be increased in patients with MS. In addition, small studies have tried to determine if a causal relationship exists between gut microbiota and MS severity or immune responses. To determine if the results found in small studies are true and to determine if a causal relationship exists, controlled, prospective studies are eagerly awaited. Because it is relatively easy to frequently collect stool samples, it is feasible to analyze a few 100 samples of MS patients. Of these patients, clinical data need to be gathered, such as which treatments they are on, when they have relapses, as well as immunological data such as the frequency of T cell subsets and their cytokine production. In this way, the effects of vitamin D, glatiramer acetate, and other treatments on gut microbiota can be assessed, and it will be possible to determine whether a causal relationship exists between microbiota of MS patients and their immune response in relation to disease progression.

We also propose to collect mucosa biopsies to investigate the contribution of mucosa-associated bacterial species in MS. Thus far, all studies on gut microbiota in MS patients used fecal samples, which focus on luminal microbiota. Therefore, differences in mucosa-associated species may have been left undetected. Comparing intestinal biopsies of MS patients with healthy individuals may show differences, which could be additional targets for treatment. Geva-Zatorsky et al. have recently monocolonized gnotobiotic mice with one of 53 different bacterial species and assessed their immunomodulatory effects. Even though the effects of bacteria may be different in SPF or conventionally housed mice, this does provide additional immunosuppressive species which may be combined to rationally develop probiotic mixtures (37).

In the case that species that worsen or improve MS will be identified, the gut microbiota of MS patients may be modified for therapeutic benefit. This may be achieved by using probiotics, prebiotics, or a mixture of these two: synbiotics. The role of viruses in this process needs deeper investigation. As viruses can greatly influence the bacterial diversity, phage therapy may be used in conjunction with pre-, pro-, or symbiotic mixtures to reduce competitiveness of microbiota that are already present and increase colonization of introduced microbiota. These may also be used to reduce the abundance of unwanted microbial species. An already used and relatively safe option for altering the gut microbiota is fecal microbiota transplantation. The gut microbiota of MS patients may be restored with fecal microbiota transplantation to resemble the gut microbiota of healthy individuals.

An alternative for restoration of the gut microbiota of patients is treatment with bacterial products or dietary components that may dampen immune responses, such as purified PSA or SCFA. PSA is effective in EAE models through well-known mechanisms (99), and in vitro studies indicate that these results may translate well into humans (116). Therefore, it would be interesting to investigate if PSA would be equally effective in MS patients. Preliminary results of treatment with the dietary component propionic acid in MS patients are promising, as it shows immunological efficacy without safety concerns (136). This makes it a very promising therapeutic that needs to be tested in controlled clinical trials. The efficacy of vitamin D supplementation and reduced salt intake will also need to be tested in large randomized controlled trials.

In conclusion, gut microbiota and dietary interventions are promising treatments for MS, but there are still many questions that need to be investigated before we can conclude which bacterial species and which dietary components play a role in MS pathogenesis and before it will be possible to selectively target detrimental species as a treatment in MS patients.

All the authors developed tables, boxes, and figures. WH prepared and revised the manuscript. BH wrote part of the EAE/MS paragraphs. JL and BH critically reviewed the different versions of the manuscript and provided comments and additional research articles.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.