Multiple Sclerosis

Multiple sclerosis (MS) is a chronic neuroinflammatory disease of the central nervous system (CNS) [1]. Clinical presentation varies among patients, which may include sensory, motor, and/or cognitive deficits, and is typically preceded by inflammatory and demyelinating lesions in the CNS white matter [1]. MS is thought to be an autoimmune disease caused by aberrant T-cell responses to myelin self-peptides [2]. Encephalitogenic T-cell responses in the CNS are followed by additional immune-cell infiltration, leading to inflammation, demyelination, and neurodegeneration [3]. The etiology of MS is complex, with both genetic and environmental factors playing major roles in disease pathogenesis. Genetic factors account for about 30% of disease risk (identical twins), with human leukocyte antigen (HLA) genes being the strongest genetic component [4, 5]. Although 70% of disease risk stems from nongenetic components, considerably less is known about environmental contributions to disease pathogenesis. Possible environmental contributing factors include smoking, exposure to the Epstein–Barr virus or other microbes, and low vitamin D levels due to insufficient exposure to sunlight/ultraviolet [3]. However, recent evidence suggests that the gut microbiota can make a major contribution to both susceptibility and protection from the disease.

Gut Microbiota

Throughout the twentieth century, the incidence of MS (and other autoimmune and allergic diseases) has increased in developed countries; a phenomenon that correlates inversely with the incidence of infectious diseases such as measles and polio [6]. This observation led to development of the “hygiene hypothesis” according to which individuals who live in cleaner environments (developed countries where there is a strong emphasis on personal hygiene) have little exposure to orofecal microbes at an early age and are thus more likely to develop allergic and autoimmune diseases [6, 7]. The corollary is that individuals who live in developing countries and are exposed to a high orofecal microbial load are more likely to develop infectious diseases while being less likely to develop allergic and autoimmune diseases. However, in recent years the hygiene hypothesis has been revisited because advances in the microbiome field have suggested that clean hygienic practices, such as the chlorination of water, overuse of antibiotics, loss of green space, small family size, and common delivery by cesarean section, have resulted in alterations of host–microbial flora collectively termed the microbiota [8, 9]. Indeed, it is now appreciated that the microbiota has a crucial role in keeping the host in a healthy state.

Microbiota refers to the trillions of bacteria, viruses, and fungi that live inside and on the human body, and their collective genetic pool is termed the microbiome. Bacterial species colonizing the small and large intestines are referred to as the gut microbiota, and their role in health and diseases are being investigated as a primary driver of host physiology [10,11,12]. The gut microbiota functions like a bioreactor that influences nutrient uptake, food metabolism, energy homeostasis, and mucosal and systemic immune responses. A healthy gut microbiota is characterized by its diversity and resilience [13], and it helps to keep the host healthy in multiple ways, including maintenance of an intact intestinal barrier, inhibition of colonization by pathogenic organisms, and regulation of host physiology and immune responses [11, 14,15,16].

Alterations of the gut microbiota, resulting in changes in its metabolic network, perturb this homeostasis, and can result in intestinal and systemic disorders [10, 17, 18]. These alterations are reflected in fecal samples which are utilized to profile the gut microbiota because they provide an easily accessible source of biospecimens .

Gut Microbiota and Autoimmunity

Advancements in culture-independent analysis, through improved DNA sequencing technology, have led to remarkable progress in our understanding of the roles of the gut microbiota in autoimmune diseases. Microbiome profiling, using 16S rRNA and metagenomic shotgun sequencing technologies, has been performed for several autoimmune/autoinflammatory diseases, including MS [19,20,21,22,23,24,25,26,27,28,29], type 1 diabetes (T1D) [30, 31], inflammatory bowel disease (IBD) [32, 33], celiac disease [34, 35], and rheumatoid arthritis [36,37,38]. The consensus of these studies is that patients with autoimmune/autoinflammatory diseases exhibit microbial dysbiosis, i.e., an alteration in the composition of the microbial community characterized by a decrease in beneficial bacteria and an increase in harmful bacteria (pathobionts). A healthy mixture of commensal bacteria helps in maintaining homeostasis at mucosal surfaces in the gut and perturbation of this community can result in colonization by pathobionts, potentially promoting a proinflammatory environment and predisposing the host to inflammatory diseases [39]. How beneficial bacteria maintain a healthy state and/or how pathobionts might predispose to or accelerate autoimmune disease is unknown. However, considerable efforts are being made to determine whether alterations in the gut microbiota are the cause or consequence of autoimmunity.

Studies performed in germ-free (GF) mice have established that the gut microbiota is crucial for the development of a healthy immune system; mice born and raised in GF facilities lack well-developed immune cells and secondary immune organs [40]. The initial interaction between the host immune system and the gut microbiota is a major determinant of the composition of the gut flora [41, 42]. Constant cross-talk between immune cells, intestinal epithelial cells (IECs), and the gut microbiota results in up- or downregulation of inflammatory mediators from the host (pattern recognition receptors, e.g., Toll-like receptors) and microbes [toxins, e.g., lipopolysaccharide (LPS)] [43]. Thus, the balance between host immune responses and gut microbiota is intricate, and the host adapts to discriminate between beneficial commensal bacteria and pathogenic bacteria. IECs play a significant role in this discrimination; their tight junctions join adjacent cells to form a barrier between immune cells of the host and bacteria in the gut [11, 16]. This cross-talk between IECs, immune cells and the microbiota help in maintaining homeostasis at the mucosal surface. Interestingly, in pathological states, this barrier is breached, resulting in increased gut permeability and systemic distribution of bacterial products and/or the bacteria itself. Several autoimmune diseases are associated with a condition called “leaky gut syndrome”, which is characterized by increased gut permeability [44, 45].

The importance of the gut bacteria, their enzymes, and metabolites in human health and disease are actively being deciphered. Early research in the field points towards the gut microbiota playing an important role in host physiology by regulating numerous metabolic pathways. We discuss in the following sections 1) differences in microbial compositions of the gut between healthy controls and patients with MS; 2) how depletion or enrichment of particular gut microbes in patients with MS might predispose to autoimmunity; and 3) the possibility that reversal of such dysbiosis might protect against autoimmunity by regulating various metabolic pathways.

MS and the Gut Microbiota

Recently, several groups, including ours, profiled fecal microbiota in patients with MS in an attempt to determine the extent to which the gut microbiome influences MS risk (Table 1). These studies revealed that patients with MS exhibit microbial dysbiosis [19,20,21,22,23,24,25,26,27,28].

Table 1 Bacterial families (F), genera (G), and species (S) whose abundance is often higher or lower in patients with multiple sclerosis (MS) versus healthy controls

The majority of MS microbiome studies have reported a reduced abundance of Prevotella and Parabacteroides belonging to the  Bacteroidetes phylum in patients with MS versus healthy controls (Table 1) [19, 20, 24, 26, 28]. As highlighted in Table 1, several human microbiome studies in adult patients with MS (at least > 10 subjects) across different geographical regions (USA, Japan, and Italy) have shown either depletion of Prevotella versus healthy controls [19, 20, 26, 28] or enrichment of Prevotella after treatment with disease-modifying drugs [24]. A recent study analyzing duodenal biopsies from patients with MS reported that patients with active disease showed a lower abundance of Prevotella than healthy controls or patients in remission [28]. Among the 3 MS microbiome studies from the USA, 2 have shown a lower abundance of Parabacteroides in patients with MS versus healthy controls [19, 26]. Jangi et al. [24] demonstrated that the abundance of Butyricimonas, another member of the Bacteroidetes, is low. Finally, Miyake et al. [20] demonstrated that the abundance of Bacteroides is lower in patients with MS than in healthy controls. The demonstration of reduced abundance of bacteria from the Bacteroidetes phylum by several groups suggests that certain bacteria from this phylum might play an important role in protecting against the development of MS.

Certain bacterial genera belonging to the phylum Firmicutes have been found to be either enriched or depleted in patients with MS. However, some discrepancies regarding the association of Firmicutes with MS have been reported  by different laboratories. For example, where we found an increase in the abundance of Blautia and Dorea (of the Lachnospiraceae family) in patients with MS [19], others observed an increases in the abundance of Streptococcus [20, 28] and Ruminococcus [25]. Also, where we observed a decrease in the abundance of Lactobacillus, Coprobacillus, Erysipelotrichaceae, and Veillonellaceae [19], others observed decreases in the abundance of Lachnospiraceae, Ruminococcaceae [22], and in Clostridia species [20, 25]. The observed discrepancies in the abundance of Lachnospiraceae and Ruminococcaceae could be due to many factors (discussed below), including specificity of 16S rRNA primers utilized for analysis of the microbiome and/or differences in the patient population. Details about the various primers and sequencing technologies utilized and their significance are discussed in a recent review [46].

With regard to the phylum Actinobacteria, we and Jangi et al. [19, 24] observed a decrease in the abundance of Collinsella in patients with MS. In addition, we found a reduced abundance of Adlercreutzia in patients with MS [19]. Although Jangi et al. [24] did not report a difference in Adlercreutzia, they did report a reduction in Slackia, another member of the Actinobacteria phylum. Other groups observed an increased abundance of Bifidobacterium [20, 22] and Eggerthella [20] in patients with MS.

In the case of Proteobacteria, the majority of the studies revealed an increase in the abundance of certain bacteria belonging to this phylum in patients with MS (Table 1). We observed an enrichment of Pseuodomonas, Mycoplana, and Haemophilus in patients with MS [19], whereas others observed an increase in abundance of Bilophila [22], and Acinetobacter [26]. Two studies reported a modulation of Sutterella in patients with MS, with Miyake et al. [20] reporting a decreased abundance in patients with MS and Jangi et al. [24] reporting an increased abundance in patients with MS on disease-modifying therapies. Finally, 3 groups observed an enrichment of Akkermansia, of the Verrucomicrobia phylum, in patients with MS [24, 26, 27].

Caution must be made when interpreting the results in the aforementioned microbiome studies as considerable variability exists between each study of patients with MS. The results from the handful of studies that profiled the gut microbiome in healthy controls and patients with MS suggest that there is no “MS-associated gut microbial signature”. Rather, the literature suggests that patients with MS exhibit dysbiosis, defined as an alteration of microbial architecture from healthy controls that shifts the immune balance towards an inflammatory phenotype [46]. This is supported by 2 recent studies showing that fecal transfer from patients with MS, but not healthy controls, to mice increased either disease incidence [27] or severity [26] of experimental autoimmune encephalomyelitis (EAE), a mouse model of MS. Based on this, we hypothesize that patients with MS exhibit a general increase in proinflammatory bacteria, rather than exhibiting an increase or decrease in a specific set of bacterial genera. That being said, there are certain bacterial genera found to be depleted (Prevotella) or enriched (Akkermansia) in multiple cohorts of MS patients from different continents (Table 1). Therefore, the possibility for commonalities for certain gut bacteria in patients with MS is not completely inconceivable. Overall, caution must be taken when interpreting results as bacterial alterations in one set of patients with MS might not represent the gut microbial state of all patients.

Additional explanation for high variability among studies stems from the lack of standarized sequencing methodologies. Several 16S rRNA primers, specific for certain hypervariable regions within the 16S gene, exist and differential use of these primers may show bias for particular taxa [46]. Additionally, the gut microbiome has been shown to be influenced by genetics [47], diet [48], geographical location [49], and use of disease-modifying therapies [24, 27]. These are all potentially confounding factors and must be kept in mind when designing and interpreting gut microbiome experiments, not just for MS but for any disease/disorder of interest.

The above-described MS microbiome studies have shown that there are differences in the bacterial populations between patients with MS and healthy controls, with some consensus between different studies. The next step is to elucidate the functional significance of these differences, i.e., to determine the mechanism whereby modulation of the host immune response by gut bacteria might predispose to, or protect from, disease. A better understanding of these mechanisms will aid in developing therapies for treating or even curing MS based on gut commensal bacteria. Gut bacteria may affect host physiology through either their own components (e.g., cell wall, polysaccharide A, LPS) or microbial metabolites produced as a result of their metabolism of the host diet and/or other bacterial breakdown products (a process known as cross-feeding) [50, 51]. In the following sections, we discuss various metabolic pathways in which gut bacteria participate, their known effects on host immunity, as well neurons and/or glial cells, and how these metabolites might influence MS. The major metabolic pathways used by commensals that influence host immune responses are metabolism of short-chain fatty acids (SCFAs), bile acids, phytoestrogens, tryptophan, and choline (Table 2; Fig. 1).

Table 2 Bacterial families (F) or genera (G) that are depleted or enriched in patients with multiple sclerosis (MS), and their association with specific metabolic pathways
Fig. 1
figure 1

Schematic of the bacterial metabolic pathways that influence the host immune system. Metabolism of indigestible carbohydrates, primary bile acids, phytoestrogens, and tryptophan by commensal gut bacteria results in production of metabolites such as short-chain fatty acids (SCFAs), secondary bile acids, phytoestrogen metabolites, and indoles/indole derivatives, respectively. These bacterial metabolites exert anti-inflammatory effects on mucosal immune cells, and has been demonstrated in vitro and in vivo. Bacterial metabolites can 1) promote barrier function and epithelial integrity; 2) induce tolerogenic dendritic cells (DCs), which trigger the polarization of CD4+ T cells into interleukin (IL)-10 and/or transforming growth factor (TGF)-β-producing FoxP3+ regulatory T cells (Tregs); and 3) directly induce IL-10 and/or TGF-β producing FoxP3+ Tregs. This tolerogenic intestinal environment may influence the peripheral immune system, and thereby lead to protection against and/or amelioration of experimental autoimmune encephalomyelitis (EAE)/multiple sclerosis (MS). IEC = intestinal epithelial cell

SCFA Metabolism

Among the bacterial metabolic pathways, SCFAs are the most extensively studied in the context of host immunity [51,52,53]. SCFAs, abundant in high-fiber foods, are 1 to 6 carbons in length and produced through the fermentation of indigestible starches and complex sugars [52, 53]. The majority of SCFAs in the colon are acetic acid, propionic acid, and butyric acid [54]. SCFAs contribute to host immunity at both intestinal and extraintestinal locations, modulating cell processes either by interacting with certain G-protein coupled receptor and/or actively or passively transporting into the cytoplasm. Once inside the cytoplasm, SCFAs can regulate transcription factors and certain enzymes, such as histone deacetylases [50].

Within the colonic lumen, IECs represent the first line of defense as they are in direct contact with the gut microbiota and their metabolites. Besides serving as an energy source, SCFAs can modulate the immune-defensive functions of IECs by altering barrier integrity and cytokine production in response to inflammatory stimuli [52]. In vitro treatment of human IECs with SCFAs leads to a decrease in the induction of proinflammatory cytokines by inflammatory stimuli [55, 56] but induces production of interleukin (IL)-18, a proinflammatory cytokine involved in epithelial integrity and homeostasis [57]. Notably, treatment with acetate-producing Bifidobacterium improves barrier function against enterohemorrhagic Escherichia coli O157:H7 [58]. Similarly, direct treatment with SCFAs enhances barrier function by inducing mucin genes in goblet cells [50].

SCFAs can also modulate the phenotype and effector function of leukocytes in the gut. In general, SCFAs induce a tolerogenic phenotype in lymphocytes and antigen presenting cells [50, 54, 56]. Previous studies showed that SCFAs downregulate the expression of major histocompatibility complex class II, co-stimulatory molecules, and inflammatory cytokines by dendritic cells (DCs) and potentiate a regulatory phenotype in CD4+FoxP3+ T cells (Tregs) [50, 52, 59,60,61]. Several studies demonstrated that SCFAs induce the differentiation of both FoxP3+ Tregs and IL-10-producing T cells in vitro, as well as in vivo, thereby maintaining mucosal homeostasis and protecting against colonic inflammation [50, 52, 54, 62,63,64]. Taken together, these results suggest that SCFAs induce a tolerogenic and immunosuppressive phenotype in the gut mucosa. It should be noted, however, that SCFAs have also been shown to potentiate a proinflammatory phenotype in both IECs and immune cells [52]. Presumably, the precise phenotype elicited by SCFA treatment depends on one or several factors such as the SCFA used, the dose used, the duration of the treatment, and the model system utilized.

In addition to modulating the immune system, bacterial metabolites generated from the gut microbiota have the ability to directly modulate CNS function. For example, in both in vivo and in vitro models of Parkinson’s disease, butyrate prevents neuronal degeneration and death [65]. Moreover, this modulation of CNS function may be direct, given that SCFAs are found in systemic circulation [66].

Several bacterial genera whose abundance is low in patients with MS are capable of participating in SCFA metabolism. These include Clostridium, Faecalibacterium, Eubacterium, Ruminococcus, Butyricimonas, Bacteroides, and Prevotella (Table 2) [51, 53, 67]. Consistent with their association with SCFA metabolism, these bacterial genera induce an anti-inflammatory phenotype in IECs and immune cells. Clostridium species (Firmicutes) are dominant gut commensals and are strongly associated with SCFA metabolism [68]. Clostridium clusters IV and XIVa (distinct cohorts of Clostridium groups based on phylogeny) are implicated in maintaining gut homeostasis by inducing FoxP3+ Tregs, and their representation is reduced in patients who have MS, as well as active IBD [20, 68,69,70,71]. Several groups have demonstrated that immunosuppression by Clostridium involves SCFA production, as discussed by Nagano et al. [72]. The Clostridia family includes Eubacterium, Ruminococcus, and Faecalibacterium which have anti-inflammatory properties [53]. Eubacterium levels are low in stool samples from patients with IBD, and treatment with Eubacterium attenuates experimental colitis, presumably owing to its SCFA-producing abilities [73, 74]. Within the colon, Ruminococcus plays a crucial role in degrading resistant starch to SCFAs, and it leads to decreased proinflammatory cytokine production in vitro [75, 76]. Faecalibacterium prausnitzii is another commensal bacterium of the Firmicutes phylum, and low levels are associated with a higher risk of IBD [77, 78]. Faecalibacterium prausnitzii supernatant induces anti-inflammatory effects in human peripheral blood mononuclear cells and a model of colitis based on induction by 2,4,6-trinitrobenzenesulfonic acid [79,80,81].

Bacteroides fragilis is a well-studied member of the Bacteroides genus and has been shown to induce IL-10-producing FoxP3+ Tregs by producing polysaccharide A [82]. Bacteroides fragilis has also been shown to participate in SCFA metabolism [51]. Bacteroides thetaiotaomicron produces acetate, which is associated with modulation of mucus production in colonic epithelial goblet cells [83]. Butyricimonas is a SCFA-producing bacterium predominant in the gut, and its abundance correlates negatively with the expression of proinflammatory genes in T cells and monocytes from patients with MS [24]. Prevotella is a commensal bacterium that plays a dominant role in metabolizing xylans to SCFAs [67]; its abundance has been reported to be low in patients with MS [19, 20, 24, 28] and patients with rheumatoid arthritis (RA) [36]. Additionally Prevotella histicola has been shown to suppress EAE and collagen-induced arthritis in humanized mice [84, 85]. Although abundance of Prevotella in MS has been consistent with studies showing either a lower abundance [19, 20, 28] or an increase after treatment in patients with MS [24], the role of Prevotella in RA remains uncertain, as 1 study has reported a higher abundance of Prevotella in patients with early-onset RA [38]. It is possible that the apparent contradictions regarding the Prevotella results are due to differences in the specific species and strains utilized in the studies, and/or to differences in the patient populations [86]. Notwithstanding the uncertainty on a few points covered above, the majority of the available data suggest that a reduction in SCFA-producing bacteria contributes to the proinflammatory state in patients with MS.

Using the EAE model, Haghikia et al. [87] demonstrated that SCFA treatment induces Tregs in the gut and ameliorates EAE. Chitrala et al. [88] showed that fecal transplant from CD44–/– donors to wild-type (WT) hosts ameliorated EAE, which they attributed to the increase in abundance of SCFAs in the stool. Mizuno et al. [89] also demonstrated that oral treatment with SCFAs or a high-fiber diet (results in high SCFA levels in the cecum) ameliorated EAE in C57Bl/6 mice, and this was accompanied by a greater frequency of Tregs in the draining lymph nodes [89]. The anti-inflammatory properties of SCFAs have been studied in other models of autoimmune diseases as well. For example, Marino et al. [90] demonstrated that nonobese diabetic mice receiving special diets designed to release particular SCFAs in the colon were less prone to develop spontaneous T1D [90]. In addition to their EAE findings, Mizuno et al. [89] demonstrated that oral administration of SCFAs ameliorates collagen-induced arthritis. It should be noted, however, that SCFA administration exacerbated K/BxN serum transfer arthritis; a model that, unlike collagen-induced arthritis, does not require adaptive immune responses and instead relies on innate immunity [89]. Thus, SCFAs may exert bimodal effects dependent on the cell type they are influencing. In a model of intestinal inflammation, dextran sulfate sodium-induced colitis, lack of the SCFA receptor GPR43, expressed on a variety of myeloid cell types, resulted in the development of more severe disease than in WT mice [91]. Moreover, in GF mice, which also exhibit exaggerated dextran sulfate sodium-induced colitis, treatment with SCFAs ameliorated intestinal inflammation [91]. Indeed, SCFAs have a demonstrated therapeutic potential in patients with IBD when given as an enema [92].

Collectively, the published findings on SCFA-producing bacteria suggest that they could influence immune responses involved in local or systemic pathophysiological states through their modulation of IECs and/or leukocytes (Fig. 1). The development of novel and more effective therapeutic interventions is likely to benefit from future assessment of the contribution of SCFA-producing bacteria to pathological states, such as MS.

Bile Acid Metabolism

Bile acids—steroid acids produced by the liver—aid in the digestion and absorption of dietary fats and fat-soluble vitamins [93]. Hepatocytes synthesize primary bile acids, which undergo a conjugation reaction with glycine or taurine before being secreted to the gallbladder [94]. After a meal, the gallbladder releases bile into the duodenum. In the distal ileum, conjugated bile acids are later reabsorbed by active transport and returned to the liver via the portal circulation to be utilized for subsequent digestions [94]. This process is termed enterohepatic circulation. While 95% of primary conjugated bile acids are efficiently returned to the portal circulation, the remaining 5% escape active transport and become substrates for bacterial metabolic reactions [93, 95]. The major microbial metabolic modifications, including deconjugation, the oxidation of hydroxy groups, and 7α/β dehydroxylation, result in the production of secondary bile acids [95]. Unlike the primary bile acids, which are conjugated, the secondary bile acids cannot participate in active reuptake. Some secondary bile acids are absorbed through the colon by passive diffusion, whereas others are excreted in feces [94]. Both primary and secondary bile acids can modulate host hormonal and immunological processes by stimulating nuclear receptor farnesoid X receptor (FXR) or cell-surface receptor G protein-coupled bile acid receptor 1 (GPBAR1 also known as TGR5) [54, 96].

Within the intestinal lumen, bile acids can influence barrier function and defense mechanisms via FXR+ and GPBAR+ IECs. A study using the biliary obstruction model of intestinal injury demonstrated that FXR plays a crucial role in protecting against bacterial overgrowth and the disruption of epithelial integrity [97]. In a separate study, feeding rats primary bile acids prior to bile-duct ligation led to a decrease in intestinal injury [98]. Similarly, GPBAR1–/– mice develop abnormal intestines characterized by altered tight junctions, irregular mucus-cell morphology, and increased susceptibility to colitis [99]. Furthermore, treatment of the human intestinal epithelial cell line Caco-2 with secondary bile acids resulted in decreased IL-8 production after stimulation with IL-1β [55].

Bile acids may also be involved in regulating leukocytes in the gut. FXR–/– mice exhibit a proinflammatory colonic phenotype and develop exacerbated trinitrobenzenesulfonic acid-induced colitis [100]. Additionally, macrophages from these animals exhibit enhanced proinflammatory cytokine response to stimulation with LPS [100]. This study also revealed that FXR agonists attenuated the expression of IL-1β and tumor necrosis factor (TNF)-α in LPS-stimulated murine lamina propria CD11b+ cells and human THP-1 cells [100]. Several other studies demonstrated that GPBAR1 agonists decrease nuclear factor kappa B (NF-κB) activity in macrophages and monocytes [101,102,103]. Recently, it was shown that GPBAR1–/– mice developed exacerbated colitis and enhanced recruitment of classically activated M1 macrophages [104]. In contrast, GPBAR1 agonist treatment alleviated colitis and shifted the polarization of lamina propria macrophages from M1 to alternatively activated M2 macrophages; a process reliant on IL-10 [104]. GPBAR1 agonists also inhibit the production of proinflammatory cytokines in primary human macrophages but not anti-inflammatory IL-10 [105]. The differentiation of monocytes into DCs in the presence of bile acids results in an IL-12 hypoproducing phenotype that is dependent on GPBAR1 [106]. However, once DCs are terminally differentiated, they do not respond to GPBAR1 stimulation as the expression of this receptor is downregulated over the course of differentiation [106]. Further, bile retinoids (which are distinct from bile acids) imprint CD103+ DCs within the lamina propria of the small intestine with retinol-metabolizing activities, enabling them to generate gut-tropic effector T cells and inducible Tregs [107]. Interestingly, both the retinoic acid receptor and FXR dimerize with retinoid X receptor, suggesting possible interplay between these 2 pathways [107]. Collectively, these studies indicate that bile acids have a central role in maintaining intestinal homeostasis.

The majority of these studies do not directly test the contribution of primary versus secondary bile acids. However, fecal samples from patients with IBD have increased and decreased abundance of primary and secondary bile acids, respectively. This suggests that a decrease in the amount of secondary bile acids may play a role in the proinflammatory state of patients with IBD [55]. Indeed, primary bile acids can stimulate Clostridium difficile growth and spore formation, whereas secondary bile acids inhibit outgrowth [93, 108]. Recent studies have shown that antibiotic treatment results in a higher primary to secondary bile acid ratio in fecal samples, which may promote C. difficile infection [93, 109]. Furthermore, primary bile acids generated from a high milk-fat diet promote the growth of Bilophila wadsworthia, leading to exacerbated colitis in genetically susceptible IL-10–/– mice [110]. These finding suggest that the ability of particular primary or secondary bile acids to promote or protect against disease is context dependent. Bile acids may also modulate CNS function directly. Primary and secondary bile acids are present in rat brain under homeostatic conditions, and may interact with GPBAR1+ microglia [111, 112].

Several bacteria capable of bile acid metabolism are depleted (Lactobacillus, Clostridium, Erysipelotrichaceae, and Parabacteroides) or enriched (Acinetobacter, Bifidobacterium, Pseudomonas, and Bilophila) in patients with MS (Table 2) [19, 20, 22, 26]. Lactobacillus, Clostridium, Erysipelotrichaceae, and Parabacteroides have been shown to exert anti-inflammatory effects. Lactobacillus species are a common ingredient in a number of probiotics, and exert anti-inflammatory effects in both in vitro and in vivo settings [113]. THP-1 cells exposed to Lactobacillus supernatant produced lower levels of TNF-α [114], possibly owing to its participation in bile acid metabolism [115, 116]. Certain Clostridium species have also demonstrated anti-inflammatory activities via bile acid metabolism. Clostridium scindens, which belongs to cluster XIVa, protects against C. difficile infection by their production of secondary bile acids [117, 118]. Patients with MS are depleted of Clostridia clusters XIV (as well as IV), further supporting a role for producers of secondary bile acids in maintaining immune homeostasis in the gut [20]. The Erysipelotrichaceae family plays a major role in bile-acid metabolism [119], and some members exert anti-inflammatory effects on the host immune system [19, 120]. Parabacteroides can also generate secondary bile acids [121], and this might be responsible for its anti-inflammatory effects [122].

With regard to bile acid-metabolizing bacteria for which patients with MS are enriched, Bilophila and Pseuodomonas promote proinflammatory responses. As mentioned above, Bilophila utilizes primary bile acids that promote colonic inflammation in genetically susceptible mice [110]. Furthermore, bile-acid signaling via FXR promotes Pseudomonas biofilm formation, and this likely contributes to its proinflammatory effects [123, 124]. Similarly, Acinetobacter baumannii promotes biofilm formation, and is an important pathogen associated with nosocomial infections [125]. Other members of the same genus have also been shown to participate in bile-acid metabolism [126]. Finally, Bifidobacterium expresses enzymes that are necessary for bile-acid metabolism [127]. Although interactions between bacteria and bile acids can result in proinflammatory effects, the majority of studies indicate that bile-acid signaling pathways promote anti-inflammatory responses. In support of this, 2 separate studies showed that the activation of either FXR or GPBAR1 suppresses lymphocyte and myeloid cell activation, resulting in an attenuation of EAE [128, 129]. Collectively, these studies support the notion that bile-acid metabolism plays a major role in immune regulation both within and outside of the intestine (Fig. 1). Future work distinguishing the roles of primary versus secondary bile acids in modulating immune responses in EAE/MS will aid in our understanding of the contribution of this metabolic pathway to disease.

Phytoestrogen Metabolism

Phytoestrogens are plant-derived compounds that are present in diet and can be categorized into 4 major classes: isoflavones, prenylflavonoids, coumestans, and lignans [130, 131]. Gut bacteria metabolize isoflavones and lignans into equol and enterolactone, respectively [130, 131]. Isoflavones are most abundant in soybeans/soy products and lignans are most abundant in flaxseed [132]. Phytoestrogens are structurally similar to mammalian estrogen and can be either agonistic or antagonistic to estrogen receptors (ER), with a preference for ERβ [133, 134]. Gut bacteria play a critical role in the metabolism and bioavailability of lignans [135] and isoflavones [136] by providing the enzymes necessary for their conversion to enterolactone and equol, respectively. After gut bacteria hydrolyze and deglycosylate phytoestrogens, the resulting metabolites are more readily absorbed by the intestine [130, 136]. From there, they may be transported to the liver to have systemic effects on immune modulation.

Phytoestrogens may have an impact on various aspects of immunity, both in the intestine and systemically. With regard to the colonic lumen, studies on IECs (in vitro) have shown that they modulate host-cell behavior, as both primary intestinal cells and IEC cell lines express ERβ [137]. Several studies have demonstrated that isoflavones suppresses the production of proinflammatory cytokines and exert protective effects on barrier function in IEC cell lines [138, 139]. Isoflavones also inhibit LPS-induced nitric oxide production and inducible nitric oxide synthase expression in a dose-dependent manner [140], and downregulate NF-κB and extracellular signal–regulated kinase activation [141]. Lignans also have an immunosuppressive effect on IECs, reducing their expression of monocyte chemoattractant protein-1 in response to LPS treatment [142].

Phytoestrogens may also modulate the phenotype of immune cells in the gut. Isoflavones and lignans suppress macrophage/monocyte activity by inhibiting reactive oxygen species release, cyclooxygenase-2 and NF-κB activation, and proinflammatory cytokine production after LPS stimulation in vitro [139, 143,144,145,146]. DCs express ERs, and therefore their behavior may also be modulated by phytoestrogens [147]. Similar to their effects on macrophages/monocytes, phytoestrogens downregulate LPS-induced production of proinflammatory cytokines and antigen-presenting machinery in DCs [139]. Phytoestrogens may also influence Treg function, as treatment with them increases FoxP3, transforming growth factor-β, and cytotoxic T lymphocyte-associated protein 4 expression in splenocytes [148]. Genistein, an abundant isoflavone, is a known tyrosine kinase inhibitor, and, at high concentrations, may inhibit lymphocyte proliferation [130]. Phytoestrogen metabolites, which are found in circulation, may be able to access the CNS and protect against neurodegenerative insults as various regions in the CNS express ER [134, 149,150,151]. These observations suggest that bacterial metabolites not only modulate immune function, but also may influence the CNS directly.

Patients with MS have a reduced abundance of Prevotella, Parabacteroides, Adlercreutzia, Slackia, and Lactobacillus, that have the ability to metabolize phytoestrogen compounds (Table 2) [19, 24]. Prevotella species were shown to metabolize phytoestrogens into beneficial metabolites [152], and this may contribute to its anti-inflammatory effects [84, 153]. The ability of P. histicola to induce Tregs and suppress EAE in humanized mice [85] suggests that metabolism of phytoestrogen by Prevotella might be one of the major mechanism of disease suppression. Parabacteroides metabolize phytoestrogens and their presence is correlated with remission in patients with Crohn’s disease [19, 122, 154]. Adlercreutzia equolifaciens and Slackia are other equol producers in the gut [136, 155, 156]. Lactobacillus, a major milk fermenter, can confer anti-inflammatory effects on immune cells, possibly through its bioconversion of phytoestrogens [157, 158]. Given these findings, it is reasonable to hypothesize that phytoestrogens and their metabolites might confer protection in EAE and/or MS by inducing anti-inflammatory responses. Several studies have demonstrated that therapeutic administration of phytoestrogens (genistein and daidzein) ameliorates EAE, which is correlated with a decrease in the production of proinflammatory cytokines and an increase in the production of anti-inflammatory cytokines [159,160,161,162,163,164]. Phytoestrogens are also protective in the nonobese diabetic mouse model of T1D [165] and were associated with shifts in the gut microbiome [166]. A recent study by Berer et al. [27] demonstrated that mice transplanted with fecal samples from healthy human controls, which developed a lower frequency of spontaneous EAE than mice transplanted with MS fecal samples, harbored a higher abundance of Adlercreutzia then MS fecal-treated mice. This suggests that phytoestrogen metabolism may confer protection in EAE [27]. However, a direct role of equol and enterolactone (end metabolites of phytoestrogen) in EAE/MS is unknown. Studies are currently underway in our laboratory to decipher their role in modulation of immune responses and EAE.

Aside from their effects on autoimmune models, phytoestrogens are well known for their health benefits in treating menopausal symptoms and various diseases such as cardiovascular diseases, type 2 diabetes, and cancer [130, 167,168,169]. However, the results of clinical trials for the therapeutic use of phytoestrogens have been mixed [170, 171]. A possible explanation for the failure of phytoestrogens to provide health benefits in clinical trials is the loss of specific commensal bacteria in the gut required for the conversion of these phytoestrogens into beneficial metabolites such as equol and enterolactone. Based on our data and published reports, we propose that metabolism of phytoestrogen into equol and enterolactone by microbes of the gut plays a critical role in maintaining immune homeostasis because they induce anti-inflammatory responses (Fig. 1).

Tryptophan Metabolism

Tryptophan, an essential amino acid present in a variety of protein-based foods, is metabolized by certain gut bacteria into indole and indole derivatives and these can modulate the effectiveness of both the intestinal barrier and immune cells by activating the aryl hydrocarbon receptor (AhR) or pregnane X receptor [50, 172]. AhR plays a major role in regulating mucosal immune responses, as demonstrated by microbial perturbations in AhR–/– mice [173]. In addition, the generation of AhR ligands by commensal bacteria confers protection from inflammation and leads to modulation of the expression of both pro- and anti-inflammatory genes [172].

Tryptophan metabolites strengthen barrier function, increase mucin production, and attenuate TNF-α-mediated production of proinflammatory cytokines [174, 175]. Activation of AhR by tryptophan metabolites induces expression of the IL-10 receptor on IECs, suggesting an anti-inflammatory phenotype [176]. AhR activation also influences intestinal epithelial integrity indirectly through activation of group 3 innate lymphoid cells (ILC3), which require AhR for complete functionality and to produce IL-22, a cytokine essential for maintaining barrier integrity [177]. Tryptophan metabolites also activate pregnane X receptor, which helps maintain integrity of the gut barrier and enhances the healing of intestinal wounds in vivo [178, 179]. Tryptophan metabolites may also modulate immune cell function via AhR, as its expression in CD11c+ DCs is critical for immune homeostasis [180]. Further, AhR is required for the expression of indoleamine 2,3-dioxygenase, an immunosuppressive enzyme involved in tryptophan metabolism [181]. Given that indoleamine 2,3-dioxygenase-expressing DCs can limit effector responses and promote the expansion of Tregs [182], AhR activation modulates tolerogenic DC function, possibly by bacterial-derived metabolites. Indeed, tryptophan catabolites from gut bacteria confer mucosal homeostasis by engaging AhR [183]. In addition to modulating immune function, indole metabolites may modify neural function directly [184].

Lactobacillus, which patients with MS are depleted of, is well known for its involvement in tryptophan metabolism (Table 2) [19, 50]. Transfer of Lactobacillus to mice genetically susceptible to colitis (CARD9–/–) resulted in attenuation of inflammation in an AhR-dependent manner [185]. With regard to EAE, rats highly resistant to EAE harbor a diverse group of Lactobacillus species [186], and mice administered such a mixture exhibited milder EAE accompanied by IL-10-producing Tregs [187]. Furthermore, we observed that suppression of EAE by P. histicola in HLA transgenic mice was associated with an increase in the abundance of Lactobacillus [85]. In addition to Tregs, other gut-associated immune cells can exert regulatory function in EAE. Kadowaki et al. [188] demonstrated that transfer of CD4+ induced intraepithelial autoreactive T cells from myelin oligodendrocyte glycoprotein-specific T-cell receptor transgenic (2D2) mice suppressed EAE. Gut environmental stimuli that contribute to the development of these cells include gut microbiota and gut microbiota-derived dietary compounds such as Ahr ligands [188]. Furthermore, administration of microbial tryptophan metabolites to mice reduced CNS inflammation and EAE, and patients with MS have lower circulating levels of AhR agonists [189]. These studies highlight the potentially significant role of tryptophan metabolites in MS.

Choline Metabolism

Gut bacteria can modulate lipid levels through their effects on choline metabolism [172]. Choline is a water-soluble nutrient essential for cell signaling, the structural integrity of cell membranes, neurotransmission, and biosynthetic reactions [190, 191]. Certain gut bacteria metabolize dietary choline into an intermediate precursor, trymethylamine, which is further metabolized in the liver to generate trimethylamine N-oxide (TMAO) [172]. Most of the studies examining the relationship between the gut microbiota and choline metabolism have been performed in the contexts of obesity, atherosclerosis, and cardiovascular disease. However, obese individuals have an increased risk of MS development [192], and obesity has been shown to modulate host physiology by influencing the gut microbiota and systemic inflammation [193, 194].

The molecular mechanisms underlying TMAO-induced pathogenicity are poorly understood because its receptor is unknown [172]. TMAO exacerbates atherosclerosis, and its levels are elevated in patients with cardiovascular disease and obesity [195, 196]. In the atherosclerotic ApoE–/– mouse model, supplementation of diet with high levels of choline promotes atherosclerosis and the formation of lipid-rich macrophages (foam cells) [197]. Additionally, in mice, atherosclerosis susceptibility can be transferred from donor to host by fecal transplantation [198]. Acinetobacter baumannii participates in choline metabolism, and the abundance of Acinetobacter is high in patients with MS (Table 2) [26, 199]. Thus, choline-metabolizing bacteria potentially play a role in MS pathogenesis by modulating lipid levels and/or obesity.

Mucin Degradation

The mucus layer of the digestive tract, secreted by goblet cells, consists of a heterogenous population of glycoproteins (mucins), salts, lipids, and other proteins such as immunoglobulins and growth factors [200, 201]. The mucus layer sits at the interface between the intestinal epithelium and the lumen, providing a selective barrier against damage, coming into direct contact with bacteria. The mucus layer is also an adhesive surface that is colonized and used as an energy source by commensal bacteria [200, 201]. Thus, the mucus layer is a crucial physical layer between bacteria and the host cells, and serves as a scaffold for the attachment of commensal bacteria. The colonic epithelium of GF mice has a significantly thicker mucus layer than that of its WT counterparts, owing to lack of mucin degradation by commensals and subsequent water retention [200]. Indeed, a certain amount of mucin degradation by commensal bacteria is an important aspect of homeostatic mucus turnover [200]. However, excessive mucin degradation may allow direct exposure of luminal antigens to the intestinal immune system, and/or provide by-products for other pathogenic bacteria to scavenge [202]. Whether mucin degradation is protective or pathogenic for the host depends on the bacterium utilizing the pathway and the context of the model system.

Goblet cells are an integral part of the mucosal immune system, as they not only provide mucus, which separates host cells and the external environment; they also provide luminal antigens to DCs of the lamina propria in the small intestines [203, 204]. Interestingly, only tolerogenic CD103+ DCs were found to be capable of participating in this interaction. Thus, it may be possible that the modulation of goblet cell function by mucin degradation can influence the anti-inflammatory aspects of the intestinal immune system.

Several bacterial species that are capable of degrading mucins are increased in patients with MS including Akkermansia muciniphilia Ruminococcus, Bifidobacterium, and Dorea [19, 22, 25,26,27]. Whether these bacteria exhibit proinflammatory activities by mucin degradation is controversial. Whereas the presence of A. muciniphilia exacerbates colonic pathology in Salmonella typhimurium-induced intestinal inflammation [205], it has also been shown to exert anti-inflammatory properties and correlate negatively with IBD, T1D, and obesity [32, 206, 207]. Ruminococcus gnavus, however, is present at high levels in patients with IBD and may promote dysbiosis [32]. Bifidobacterium can induce different levels of expression of pro- and anti-inflammatory cytokines, depending on the strain [208]. Dorea, which does not directly metabolize mucins but feeds on by-products released by mucin degradation, is associated with IBD [207, 209]. Further, ceratin species of Dorea might be pro-inflammatory as they can induce interferon-γ [210]. Collectively, these studies suggest that mucin-degrading bacteria may promote inflammation in patients with MS, depending on the composition of the bacterial community.

Other Mechanisms: Molecular Mimicry

Molecular mimicry is defined as significant structural homology between a given microbial antigen and self-antigen, resulting in aberrant immune responses to the latter peptide following exposure to the microbe [211]. Although molecular mimicry is proposed as a possible cause of autoreactive T-cell activation, conclusive supporting experimental evidence has been elusive. Certain components of the gut microbiota have been shown to share sequence similarities with encephalitogenic myelin peptides, suggesting that gut bacteria may contribute to MS through molecular mimicry [212]. Pseudomonas, which is present at high levels in patients with MS [19], has structural similarity to myelin basic protein (a dominant CNS antigen) and can activate myelin basic protein-specific T-cell clones [213]. Peptides from Escherichia coli, a common gut commensal, have been shown to provoke MS-like disease in a mouse model expressing a humanized T-cell receptor transgene [214]. However, the biological relevance of these findings to patients with MS is unclear.

Concluding Remarks

Recent studies showing that patients with MS are enriched or depleted of certain bacteria highlight the importance of the gut microbiota in the development of this disease. Gut microbiota may participate in MS pathogenesis by modulating host immunity through the regulation of multiple metabolic pathways (including those for SCFAs, bile acids, phytoestrogens, tryptophan, choline) and mucin degradation. The participation of certain bacteria in more than one metabolic pathway adds a layer of complexity, requiring an understanding of potential cross-talk between bacterial metabolites and host immune responses. For example, the addition of isoflavones to anaerobic bacteria from fecal samples results in the enrichment of SCFAs, as well as equol [215]. Lactobacillus, Prevotella, and Parabacteroides are examples of such bacteria because they can induce production of SCFAs in addition to metabolizing phytoestrogens. Similarly, Lactobacillus can participate in the metabolism of SCFAs, phytoestrogens, and tryptophans. Future studies examining these aspects in a single experiment might help to better understand cross-talk among bacteria and various metabolic pathways in the context of MS. However, the majority of microbiome studies are based on a single time point in patients with MS with established disease and thus it is possible that some of the reported changes in the microbiota are due to inflammation itself. Future studies should be based on longitudinal collection of samples to determine whether disease relapse is associated with presence of certain gut bacteria. Additionally, the colonization of GF mice with gut bacteria that are associated positively or negatively with MS can be tested for the ability to either induce disease or protect against it.