Introduction

The understanding of the importance of the gut microbiota and their metabolites in host physiology and pathology is advancing apace. Important, if not crucial, effects of gut commensal bacteria and their products have been revealed in both the intestinal and extraintestinal environments, including adipose tissues, the systemic immune system, the neuroendocrine system, and the central nervous system [1–3]. Qualitative and quantitative changes in a range of metabolites produced by gut microbiota appear to be involved in the development of a wide range of diseases including metabolic syndrome, obesity, inflammatory bowel disease, type 1 and type 2 diabetes mellitus, asthma autism, colon cancer, and major depression [1, 2, 4, 5].

One of the mechanisms by which abnormalities in the composition of the microbiota and their metabolites could provoke pathology is via adverse effects on the gut-brain axis. The gut-brain axis describes well-documented, bidirectional immune, hormonal, and neurally mediated routes of communication between the gut and the central nervous system aimed at maintaining or restoring the homeostatic performance of the gastrointestinal (GI) tract [6, 7]. There is now copious evidence demonstrating the importance of gut microbiota in influencing these interactions [8]. Accordingly, this and other evidence has led to the establishment of the microbiota-gut-brain axis in place of the well-established brain-gut-axis concept [9, 10].

The optimal gut-brain functioning of this axis is essential for maintaining GI and central nervous system homeostasis [7, 11]. Disorders in the microbiome-gut-brain axis are associated with the pathophysiology of a wide variety of diseases such as irritable bowel syndrome, depression, chronic fatigue syndrome, anxiety, and autism [9, 12, 13].

Dysbiosis refers to detrimental changes in the normal composition of the microbiota. Such a state may be caused by a range of dietary and environmental/genetic factors, increased intestinal permeability, or as a result of impaired communication between gut commensals and the antigen-presenting cells in the mucosal immune system. One common cause is increased physical or psychological stress, and in such circumstances, the resultant dysbiosis contributes to stressor-induced immune dysregulation in the gastrointestinal mucosa and systemic compartments of the immune system, leading to a systemic inflammation and other immune changes capable of affecting behavior and central nervous system function [9, 12, 14–17]. In addition to stress, high-fat and high sugar western-type diets, exceedingly common in many populations, are well-established promoters of dysbiosis and intestinal permeability [18, 19], while other aspects of unhealthy diets—such as emulsifiers in processed foods—could also potentially promote intestinal permeability and thus systemic inflammation.

One potential mechanism contributing to such inflammatory alterations is increased intestinal permeability and subsequent bacterial translocation or translocation of bacterial components such as the lipopolysaccharides (LPS) of the cell wall of Gram-negative bacteria into the systemic circulation [12, 20, 21]. However, another but certainly not mutually exclusive mechanism promoting illness involves increased neuroactive microbial metabolites such as tryptophan catabolites (TRYCATs); these are thought to play a role in the genesis of several neuroimmune diseases including major depression [12, 22, 23]. Other microbial metabolites are described as short chain fatty acids (SCFAs) and are also increasingly recognized as vital in human physiology and health. An insufficiency in SCFAs is a likely contributor to the development of several illnesses, including inflammatory bowel disease, obesity, metabolic syndrome, type 1 and 2 diabetes mellitus, autism, major depression, and colon cancer [1, 2, 4, 5]. Abnormalities in the concentrations in SCFAs have been described in these conditions [24–28]. This review aims to describe the mechanisms by which abnormalities in metabolite levels produced by gut microbiota as a result of dysbiosis may influence the genesis of disease, beginning with TRYCATs and a number of other neuroactive metabolites, before going on to focus on the specific effects of SCFAs.

Gut Dysbiosis and TRYCATs

Gut-Derived TRYCAT Formation and Immune Modulation

Commensals play crucial roles in metabolizing tryptophan and other amino acids such as tyrosine, with the subsequent production of tryptophan metabolites (e.g., indole-3-propionic acid). These are capable of modulating the immune system by various mechanisms, including shaping the differentiation of innate lymphoid cells or T cells, notably Th17 lymphocytes [29, 30]. Moreover, some species of Clostridia clusters show a high expression of indoleamine 2,3-dioxygenase (IDO), an enzyme that is induced by proinflammatory cytokines and induces the catabolism of tryptophan lowering plasma tryptophan and increasing TRYCAT levels [23]. TRYCATs also interact with the aryl hydrocarbon receptor, expressed on a number of immune or somatic cells, such as differentiating Th17 T cells [30]. The aryl hydrocarbon receptor is involved in the differentiation of Th17 and Tregs and the production of interleukin-22 (IL-22) by human Th22 cells. IL-22 regulates intestinal immune homeostasis and barrier function [29, 31]. This aryl hydrocarbon receptor (Arh) also regulates the activity of several classes of innate lymphoid cells, which play a major role in the formation and function of gut-associated lymphoid tissue (GALT) [31]. Hence, tryptophan catabolites and other aryl hydrocarbon receptor ligands can alter a number of innate and adaptive immune responses, including the outcomes of naive T cell differentiation [32].

IDO expression by dendritic cells causes accumulation of kynurenine, itself capable of upregulating aryl hydrocarbon receptors, resulting in a cascade of tolerogenic responses including the differentiation of Tregs [32, 33]. L-kynurenine, in particular, activates aryl hydrocarbon receptors in lymphoid tissue, resulting in IL-22 production by naive and activated T cells, leading to the development of Tregs [33–36]. Interestingly, agonism of the aryl hydrocarbon receptor is required to induce IDO expression [37], and once synthesized, kynurenine may activate the aryl hydrocarbon receptor for IDO induction in an autocrine manner, and form an aryl hydrocarbon receptor/kynurenine positive-feedback loop [38]. The aryl hydrocarbon receptor is also expressed by IL-22-producing innate lymphoid cells and intestinal intraepithelial lymphocytes [35, 39].

While tryptophan metabolism appears to play a pivotal role in T cell homeostasis, the uptake of tryptophan and other dietary amino acids by intestinal epithelial cells plays an indispensable role in antimicrobial peptide production and is thus a crucial factor in maintaining the homeostasis of the microbiota [40]. Aryl hydrocarbon receptors are also involved in the development of a beneficial phenomenon known as endotoxin tolerance, whereby tryptophan 1,2-dioxygenase (TDO) 2 expression, induced by lipopolysaccharides in a Toll-like receptor (TLR)4-dependent manner, results in the production of L-kynurenine and the activation of aryl hydrocarbon receptor in cells of the innate immune system. Subsequent challenge by LPS does not provoke an immune-inflammatory response by those cells in the absence of IDO [33, 37]. However, this defense clearly has its limits in an environment of increased intestinal permeability and translocation of LPS into the circulation, which is invariably accompanied by chronic immune activation, and systemic inflammation and an oxidative and nitrosative stress (O&NS) response [20, 21, 41, 42].

Moreover, elevated LPS and LPS-induced inflammatory cytokines provoke the synthesis of kynurenine 3-monooxygenase (KMO) which drives further catabolism of kynurenine toward neurotoxic molecules such as quinolinic acid [43]. Chronic immune activation and Th17 T cell differentiation can also downregulate tryptophan catabolism [44]. Other tryptophan catabolites such as tryptamine and N-dimethyl tryptamine, produced by the competitive pathways of tryptophan metabolism, inhibit IDO activation [45], while 6-formylindolol [3,2-b] (FICZ) actively enhances Th17 T cell development [46, 47]. Clearly dysbiosis-induced changes in the levels or relative proportions of TRYCATs, notably kynurenine, could be a source of systemic neuropathology. However, while TRYCATs are readily found in the systemic circulation, the same is true of microbial derived tryptophan [47, 48]. The presence of these and other neuroactive metabolites in the systemic circulation goes some way to explain the major role played by the microbiota in the development of the central nervous system (CNS) disorders leading to neuropathology [9, 12, 22, 23, 49, 50].

The Role of Microbial Derived TRYCATs in CNS Function

The composition of the microbiota affect, or possibly even control, circulating levels of tryptophan, which is an essential neurotransmitter at all levels of the gut-brain axis [49, 51, 52]. Tryptophan absorption from the gut into the systemic circulation regulates central glutamate and serotonin systems [53]. The alternative degradation of tryptophan, along the TRYCAT pathway, via the microbial manipulation of IDO and TDO levels, is a well-known pathway enabling neuromodulation as previously discussed [23, 54, 55]. It is also worth re-emphasizing that these pathways are the subject of immune and glucocorticosteroid regulation, respectively, which means that the performance of these pathways may change in an inflammatory environment [49, 51, 52]. Some commensal species can affect the concentration of circulating tryptophan via other means. For example, Bacteroides fragilis utilizes tryptophan as an energy source via the production of a tryptophanase enzyme [56]. Some other species are capable of increasing gut tryptophan levels by producing a range of synthase enzymes [57], and yet others display a capacity ex vivo for using tryptophan as a substrate for the synthesis of serotonin [58]. However, the predominant ability of the microbiota to regulate tryptophan metabolism along the TRYCAT pathway has the simultaneous effect of reducing the activity of the pathway involved in serotonin production and increasing the synthesis of quinolinic acid and kynurenic acid, as well as a range of other neuroactive metabolites [55, 59]. This is highly significant as serotonin is a crucial neurotransmitter at each signaling terminus of the gut-brain axis [60]. There are a range of neural processes in the gastrointestinal tract, which may be influenced by localized changes in serotonin concentrations provoking qualitative and quantitative changes in signals propagated along the scaffolding of the gut-brain axis, which in turn provokes changes in central nervous system neurotransmission [60, 61]. This is another route by which changes in the composition of the microbiota could compromise the bidirectional signaling between the brain and the gut [60, 61].

Microbiota also influence the levels of γ-aminobutyric acid (GABA), brain-derived neurotrophic factor (BDNF), dopamine, and noradrenaline, which are key players in the communication between the GI tract and the brain [1, 62]. GABA is metabolized from glutamate by a range of Lactobacillus and Bifidobacterium species [63, 64]. Moreover, a relationship between changed levels of GABA in the gut and in the brain has now been established [65]. Several commensal species have the capacity to modulate levels of BDNF in the gut with consequent changes in the levels and performance of this factor in the brain where BDNF supports neuronal survival, differentiation, and growth [1, 66, 67]. There is also some evidence that changes in the composition of the microbiota affect levels of the N-methyl-D-aspartate receptor (NMDA) subunits NR2A and NR2B and other proteins involved in synaptic function, such as postsynaptic density protein-95 and synaptophysin, which play a vital role in memory formation and retrieval [1, 67, 68]. While TRYCAT abnormalities have an acknowledged role in the pathogenesis and persistence of diseases such as major depression, chronic fatigue syndrome, and somatoform disorder [12, 22, 23], abnormalities in the levels of neurotoxic and immune-modulatory TRYCAT are also seen in a range of neurodegenerative and neuroimmune diseases, including multiple sclerosis [69]. Hence, given considerable evidence indicating the existence of increased intestinal permeability in patients with these illnesses [70], there would seem to be an urgent need for studies directed at the metagenomic analysis of the commensal population in such illnesses, which are currently sparse or even nonexistent in humans. A diagrammatic representation of the effects of microbial metabolites on the production of neuroactive molecules and on the performance of the immune system is provided by Fig. 1. We next turn to a consideration of microbiota metabolites, which have an established role in the regulation of human physiology and immune responses and are increasingly implicated in the pathophysiology of several illnesses. These are described as SCFAs and are the fermentative end products of insoluble fiber breakdown by anaerobic bacteria largely concentrated in the colon [48].

Fig. 1
figure 1

A diagrammatic representation of the effects of microbial metabolites on the production of neuroactive molecules and on the performance of the immune system

Full size image

Short Chain Fatty Acids

Production and Mode of Action of SCFAs

Commensal bacteria inhabiting the intestinal tract gain energy primarily through the fermentation of nondigested dietary components and of host secretions, mainly mucin, generating SCFAs, including acetate, propionate, and butyrate. However, they differ greatly in their genetic capacity to synthesize enzymes needed to produce SCFAs [71, 72]. For example, acetate (C2) is the most abundant SCFAs produced by enteric and acetogenic (the process whereby acetate is produced from CO2 and an electron source) commensals [73], while Bacteroidetes and some Firmicutes species preferentially produce propionate [74]. Firmicutes include several species (Anaerostipes spp., Eubacterium spp., Roseburia spp., and particularly Faecalibacterium prausnitzii) identified as the dominant producers of butyrate and specialist degraders of indigestible polysaccharides [75, 76]. Butyrate production may depend on cross-feeding mechanisms that require the participation of enzymatic pathways encoded by different bacteria. For example, Bacteroides thetaiotaomicron can generate acetate via its wide array of glycoside hydrolases that can be further utilized by Eubacterium halli to generate butyrate [77]. This reveals the importance of bacterial interactions for ensuring ecosystem stability and defining the functional activity of the microbiota.

SCFAs ligate and activate a range of G protein-coupled receptors, which are involved in intracellular signal transduction and, consequently, cellular responses. For example, G protein-coupled receptor-41 and G protein-coupled receptor-43 are activated by almost all SCFAs [78] and are expressed at physiological levels on enteroendocrine cells and normal enterocytes [79–82]. G protein-coupled receptor-41 is also present on the cell membranes of other cells such as adipocytes, pancreatic cells, and enteric neuronal cells [83, 84]. Extraintestinal expression of G protein-coupled receptor-43 appears on some myeloid cells and all granulocytes [85, 86]. G protein-coupled receptor-109a is expressed by gut epithelial cells, dendritic cells, macrophages, and extraintestinal adipocytes and appears to be a specific receptor for butyrate and niacin [87, 88].

The activation of G protein-coupled receptors 41, 43, and 109a, on intestinal epithelial cells mediates a significant proportion of the functions of SCFAs [87, 89–91]. SCFAs ligation of these G protein-coupled receptors activates a range of signaling processes such as protein kinase A and ERK1/2 and many transcription factors, e.g., activating transcription factor 2 [92, 93]. Activation of this latter pathway is an indispensable element in the expression of pivotal immune and inflammatory cytokines and chemokines such as tumor necrosis factor alpha (TNF-α), interleukin-1 (IL-1), IL-6, and chemokine (C-X-C motif) ligand (CXCL)-1 and CXCL-2. SCFAs also activate G protein-coupled receptors 41 and 43 on specialized secretory epithelial cells leading to the synthesis and secretion of glucagon-like peptide (GLP)-1 [94].

SCFAs also modulate cell differentiation and proliferation and the secretion of hormones, including leptin and peptide YY [95, 96]. These actions, and others, enable SCFAs to regulate a wide range of regulatory effects on metabolic pathways and the immune response to infection or inflammation. The multiple metabolic effects of SCFAs are depicted in Fig. 2. We next turn to a consideration of their actions on the immune system. While these effects are varied, and at times somewhat contradictory, they can be usefully categorized into effects on the recruitment of leucocytes from the peripheral circulation to areas of inflammatory activity, effects on naive T cell differentiation, modulation of cytokine expression by antigen-presenting cells, and disruption of immune homeostasis by effects on epithelial barrier function. We will discuss each in turn.

Fig. 2
figure 2

Regulation of fatty acid metabolism. SCFAs increase hepatic and striated muscle AMPK activity, via an increase in the AMP/ATP ratio and/or activation of the GPR43-leptin pathway, leading to increased expression of peroxisome proliferator-activated receptor gamma coactivator (PGC)-1α expression and subsequent elevations of peroxisome proliferator-activated receptor (PPAR)α and farnesoid X receptor (FXR). The sum of these effects is enhanced fatty acid (FA) oxidation in both tissues and the decrease of a de novo hepatic fatty acid production. SCFAs stimulate PGC-1α and uncoupling protein (UCP)-1 activity in brown adipose tissue (BAT), thereby increasing FA oxidation and thermogenesis and fatty acid oxidation. The inhibition of lipolysis in adipose tissue mediated by SCFA engagement of GP43 likely involves suppression of the hormone-sensitive lipase (HSL), the activation of the Gi/o protein and the subsequent decrease in the levels of cAMP and protein kinase A (PKA). SCFA stimulation of adipose-specific GPR43 inhibits insulin signaling via suppression of Akt phosphorylation, subsequently inhibiting adipose tissue fat storage and promoting lipid and glucose metabolism in other tissues

Full size image

Effects of SCFAs on Leucocyte Recruitment to Sites of Inflammation

Inflammatory damage to tissue leads to the recruitment of neutrophils and other leucocytes from the bloodstream and migration to the site of inflammation. This is a complex process, composed of several steps involving the coordinated activity of selectins and integrins, which enable neutrophil egress into tissue via human umbilical vein endothelial cells (HUVEC), and a range of chemokines, which summon and guide leucocytes to the sites of damage [74, 97]. SCFAs suppress the adherence of leucocytes to HUVEC and hence impede the translocation of neutrophils to sites of inflammation and hence exert an anti-inflammatory effect [74, 76] in a process involving nuclear factor (NF)-kB and peroxisome proliferator-activated receptor (PPAR)γ inhibition, and a subsequent downregulation in the expression of selectins and integrins [98]. SCFAs can also modulate the expression of chemokines and adhesion molecules on leucocytes and endothelial cells [76, 99]. SCFAs can stimulate or inhibit the recruitment of circulating lymphocytes to the sites of inflammation dependent on physiological conditions [3, 98, 100]. For example, butyrate inhibits the LPS-stimulated increase in CXCL-2 and macrophage chemoattractant protein-1 (MCP-1), leading to a decrease in leucocyte recruitment.

SCFAs exert a somewhat paradoxical effect on neutrophils. They modulate the synthesis and secretion of chemokines and adhesion molecule expression on the surface of neutrophils, in a similar manner to leucocytes, and thus inhibit the recruitment of these PMBCs to sites of inflammation [100]. However, SCFAs induce an increase in neutrophil chemotaxis, which is an effect dependent on G protein-coupled receptor-43 ligation [101, 102] and activation of ERK and p38 pathways [102]. There is, however, some evidence that the interaction of G protein-coupled receptor-43 and SCFAs may render neutrophils refractory to the chemoattractant effects of chemokines [3]. Butyrate, in particular, inhibits LPS-induced migration of macrophages via inhibition of histone deacetylase activity within these antigen-presenting cells [72]. Butyrate also modulates the synthesis and secretion of IL-8, MCP-1, and CXCL-1 by intestinal epithelial cells in response to peptidoglycan lipopolysaccharides and proinflammatory cytokines [103, 104]. Butyrate and propionate ameliorate the LPS-stimulated synthesis of CXCL-2 by neutrophils [98], while butyrate alone inhibits interferon (IFN)-γ-inducible protein-10 (IP-10). Both SCFAs inhibit the production of MCP-1 in the presence or absence of LPS [92].

Effects of SCFAs on T Cell Differentiation

Effects on Regulatory T Cell Production

Recent studies, utilizing metabolome analysis and other investigative techniques, have revealed that the number of colonic Tregs in steady-state conditions are heavily influenced or even determined by the luminal concentration of propionate, butyrate, and perhaps acetate, produced by commensal bacteria [105–107]. The precise mechanisms by which SCFAs promote the production of colonic Tregs are still a matter of debate and may well vary with the SCFAs involved. For example, colonic Tregs express G protein-coupled receptor-43, which has been implicated in ameliorating intestinal inflammation, and the function of this receptor is an essential element in colonic Treg differentiation mediated by propionate [108]. Butyrate, on the other hand, appears to mediate Treg differentiation by engagement with the butyrate-specific receptor G protein-coupled receptor-109a, which is extensively expressed on intestinal epithelial cells and antigen-presenting cells, but—crucially—not on lymphocytes [109]. Butyrate engagement with this receptor leads to the production of IL-10 and a range of metabolites, including retinoic acid, leading to the production of Tregs [109]. However, while the evidence that Treg production is mediated by butyrate appears robust, it must be remembered that T lymphocytes as a whole do not seem to express physiologically relevant levels of G protein-coupled receptors [110]. It is also worthy of note that the effect of propionate on Treg differentiation appears to be limited to colonic Tregs and not to the wider inducible Treg population [108]. Hence, either the effect of propionate on G protein-coupled receptor-43 is specific for colonic Tregs, or there is an alternative explanation for the observations. This alternative explanation may stem from the fact that short chain fatty acids, notably butyrate, also act as histone deacetylase inhibitors and promote the production or expansion of Tregs by epigenetic modification of forkhead box-P3 promoter and enhancer regions [105, 106]. Butyrate also potentiates Treg differentiation by intestinal dendritic cells by suppressing the transcription of genes involved in the response to LPS, leaving intestinal dendritic cells, TLR4 receptors, hyporesponsive to the presence of this commensal polysaccharide [106, 111]. Several uncertainties remain regarding the precise effects of SCFAs on the colonic Treg population, such as whether the provocative effects of SCFAs on Treg numbers are limited to increased naive CD4 T cell differentiation, or whether the effects include expansion of pre-existing resident Tregs [105, 106, 108]. It must be noted that, under certain physiological conditions, SCFAs can induce the production of effector Th1 and Th17 lymphocytes, and we now turn to a consideration of potential mechanisms enabling this phenomenon [110, 112].

SCFAs and Th1 and Th17 Differentiation

T cells can potentially be regulated by SCFAs via three different mechanisms. The first mechanism involves the activation of G protein-coupled receptors 41, 43, and 109a, which can modulate cell activation and differentiation. However, the weight of evidence indicates that T lymphocytes lack significant levels of these receptors, and hence, the current consensus is that SCFA receptors are unlikely to be an important mechanism enabling direct regulation of these lymphocytes by SCFAs. Another route is the regulation of cellular bioenergetic metabolism via the conversion of SCFAs to acetyl-CoA and integration into the tricarboxylic acid (Krebs cycle), providing an upsurge in energy production and an increase in adenosine triphosphate/adenosine diphosphate levels. This increase amplifies mTOR activation [113], which skews T cell differentiation into Th17 and Th1 T lymphocytes at the expense of regulatory FoxP3+ T cells [114]. Increased mTOR activity also provokes the differentiation of IL-10+ T cells [111]. Hence, the modulation of SCFAs of cellular energy metabolism and subsequent activation of mTOR accounts for the increased production of Th1, Th17 lymphocytes, and IL-10+ generating cells. Finally, the third route involved in T cell regulation involves histone deacetylase inhibitor activity, shared to various degrees by all SCFAs [112, 115, 116]. Such histone deacetylase inhibitor activity necessitates the internalization of SCFAs into lymphocytes and contact-based enzymatic inhibition, mainly of class I/II histone deacetylase, although some studies also suggest inhibition of Sirt-1 and other class III histone deacetylase [117]. SCFAs can directly induce the formation of IL-17, IFN-γ, and IL-10 secreting T lymphocytes cells via inhibition of histone deacetylase, leading to acetylation of p70 S6 kinase and subsequent phosphorylation of rS6 and increased mTOR activation [112]. This is of interest as this route to mTOR activation is quite distinct from the effects of SCFAs on energy metabolism discussed above.

In vitro evidence suggests that SCFAs can also affect T cell differentiation patterns indirectly via a broadly immunosuppressive or tolerogenic effect on antigen-presenting cells. For example, SCFAs inhibit the development of myeloid dendritic cells from their progenitors [118], as well their functional maturation [119, 120]. A key mechanism is the attenuation of histone deacetylase activity. Among the SCFAs, butyrate is one of the most potent, while acetate is one of the least potent inhibitors of histone deacetylase [121, 122]. This enzyme, together with the histone acetyltransferases (HAT), controls the degree of protein acetylation. By inhibiting the histone deacetylase activity, SCFAs increase the acetylation of histone and nonhistone proteins including different transcription factors, including NF-kB and p53 [123].

Effects of SCFAs on Cytokine Production by Antigen-Presenting Cells

SCFAs inhibit LPS or cytokine-stimulated production of nitric oxide (NO), IL-6, TNF-α and other inflammatory molecules in macrophages, monocytes, and PMBCs [110]. Butyrate and propionate suppress the synthesis of proinflammatory mediators, e.g., TNF-α and NO, in neutrophils via a mechanism involving the attenuation of NF-kB expression and/or activation [98]. The effect of propionate or butyrate on reactive oxygen species synthesis by neutrophils, on the other hand, remains controversial. Some authors have reported increased reactive oxygen species production by SCFAs [124, 125], while others have demonstrated inhibitory effects [126, 127]. Controversy also surrounds the possible impact of SCFAs on microglial production of reactive oxygen species, NO, prostaglandin-E2, etc. [128, 129]. However, the majority of studies demonstrate that these fatty acids ameliorate microglial activation via a mechanism once again involving histone deacetylase inhibitors [129, 130]. Accumulating evidence from animal models [130] also supports the view that SCFAs may be useful in reducing inflammatory responses in the central nervous system [131].

Effects of SCFAs on Gastrointestinal Epithelial Barrier Integrity

Butyrate-producing bacteria play a vital role in regulating the functions and integrity of the intestinal epithelial barrier [132–135], and hence, dysbiosis involving a reduction in butyrate producers could disrupt the immune homeostasis in the gastric mucosa with profound systemic effects [136]. SCFAs stimulate the secretion of glucagon-like peptides 1 and 2 via ligation of glucagon-like peptide-43 [137–139] leading to decreased epithelial permeability and increased mucosal antibacterial defenses [140]. Glucagon-like peptides 1 and 2 improve barrier function via a number of different mechanisms, including via the stimulation of L class enteroendocrine cells and the inhibition of apoptosis of intestinal epithelial cells [141]. These hormones also engage in a highly complex interaction with endocannabinoid receptors, whose actions are arguably the major players in regulating the permeability of the intestinal epithelium [142, 143]. When considered as a whole, the weight of evidence indicates that epithelial barrier integrity is governed by a complex cross talk between endocannabinoids, commensal LPS, SCFAs, and glucagon-like peptide hormones [142, 143]. It probably comes as no surprise to learn that changes in the commensal population, leading to a higher proportion of SCFA producers, increase the expression of glucagon-like peptides 1 and 2 and an increase in barrier integrity [144, 145]. On the other hand, dysbiosis leading to a loss of SCFA producers can result in adverse changes in the localization and distribution of two epithelial tight junction proteins, occludin and zonula occludens-1, leading to increased epithelial permeability in rodent models of type 2 diabetes [142, 145–147]. Increased epithelial permeability, linked to changes in the microbiota or other factors such as poor diet or stress, can lead to the translocation of LPS into the circulation and resulting systemic immune activation: this is a phenomenon often described as “metabolic endotoxemia” [145–148]. This phenomenon resulting from increased intestinal permeability is a triggering factor in the development of metabolic syndrome, insulin sensitivity, and type 2 diabetes (TD2) in humans [145, 146, 149]. Clearly, dysbiosis leading to a reduction in SCFAs can result in impaired barrier function leading to metabolic endotoxemia, loss of broadly anti-inflammatory effects, and metabolic dysregulation, notably in the domain of energy homeostasis. These tripartite effects have the potential to be a source of pathology in a host.

Effects of SCFAs on Energy Homeostasis

SCFAs manufactured by gut commensals, notably propionate and acetate, can be detected in hepatic, portal, and peripheral blood, suggesting that they have the ability to cross the gut-blood barrier [148, 150, 151]. This capacity for migration enables SCFAs to influence lipids, glucose, and cholesterol in a range of remote tissues [152–154] (Figs. 2 and 3). Positive benefits on satiety, cholesterol levels, and glucose metabolism have been recorded by increasing SCFA production in animals fed resistant starch [155, 156]. The positive effects of butyrate-producing bacteria on blood glucose and lipid metabolism are also evidenced by fecal transplantation studies [157, 158]. The positive effects on glucose are achieved largely by increases in glucagon-like peptide-1, which increases insulin secretion, lowering plasma glucose and preserving beta cell function [159, 160]. SCFAs also appear to affect epigenetic regulation of gene expression in type 2 diabetes via activation of G-coupled receptors and their capacity to act as histone deacetylation inhibitors [148, 161].

Fig. 3
figure 3

Regulation of glucose metabolism. Schematic representation of the likely pathways by which SCFAs regulate glucose metabolism. SCFAs can increase GLP-1 and PYY in the colon via engagement of GPR42 and GPR43. The latter molecule increases the uptake of glucose in muscle and adipose tissue, whereas GLP-1 decreases glucagon production and increases the production of insulin in the pancreas. In addition, SCFAs possess the capacity to decrease gluconeogenesis in the liver by increasing the phosphorylation and activity of AMPK

Full size image

SCFAs ligate and regulate the balance between fatty acid oxidation and lipolysis, and increase fatty acid oxidation and decrease lipolysis in a mechanism involving increased activation and/or upregulation of PGC-1a peroxisome proliferator-activated receptors and leptin [95, 162, 163]. SCFAs also exert profound effects on cholesterol metabolism by inhibiting the activity of hydroxymethylglutaryl coenzyme A synthase, reducing hepatic glucogenesis [164, 165].

Raised levels of GLP-1 following engagement of SCFAs with free fatty acid receptors lead to elevated production of insulin in the pancreas and decreased levels of glucagon, while raised levels of pancreatic peptide (PYY) increase glucose uptake by striated muscle and adipose tissue [163, 166]. Gut commensals can also exert a regulatory influence on energy homeostasis via SCFA-induced increase in the activity of the sympathetic nervous system, which in turn may cause immune-inflammatory responses [167, 168]. All in all, it is clear that a complex interplay between the microbiota and SCFA production has a profound effect on energy homeostasis. We now move on to examine the potential role of impaired SCFA levels in the pathogenesis of complex illness where there is evidence of immune and metabolic dysregulation, using type 2 diabetes and autism spectrum disorders as exemplars.

Examples of How Dysbiosis May Play a Role in Human Disease

Type 2 Diabetes

Evidence for Dysbiosis in the Pathophysiology of Type 2 Diabetes

The altered ratio of Firmicutes/Bacteroides seen in patients with type 2 diabetes is associated with greater energy harvesting and the development of insulin resistance [169, 170]. Recently, a potential relationship between gut microbiota and TD2 pathophysiology has been suggested by two independent studies comparing metagenomes from TD2 and healthy controls. They have provided particularly strong evidence for the role of gut dysbiosis in the pathophysiology of the disease and TD2 subjects [171, 172]. These authors demonstrated an increase in Clostridium clostridioforme numbers and a decrease in the levels of Roseburia 272 in TD2 patients in both European and Chinese populations [171, 172]. Moreover, increased levels of Roseburia following fecal transplants from lean donors to patients with metabolic syndrome seemingly resulted in improved insulin sensitivity [173]. These positive effects appeared to be related to the presence of SCFAs, propionic and butyric acids [174], which ameliorate low-grade inflammation and stimulate the release of hormones such as leptin and adipokine in adipose tissue as discussed above [95, 175, 176]. These authors reported reduced counts of Prevotella and Atopobium and Clostridium coccoides group, but significantly elevated numbers of Lactobacillus spp. in patients with TD2. Such a microbiota profile was also associated with reduced production of butyrate and other SCFAs. Furthermore, this study reported significant increases in the number of commensals in the bloodstream of their diabetic patients compared to controls, indicating increased intestinal permeability and commensal translocation from the gut lumen into the systemic circulation [174]. Moreover, bacterial DNA (mostly belonging to the phylum Proteobacteria) has also been detected in the blood of subjects before diabetes onset and was higher in those who had abdominal adiposity suggesting that translocation could predict disease manifestation [177]. We now move on to consider this phenomenon and disease-specific consequences in a little more detail.

Increased Intestinal Permeability in Patients with Type 2 Diabetes

Several teams have reported a relationship between impaired GI permeability and the translocation of live bacteria or LPS fragments into the bloodstream with the development of TD2 [178–180]. Translocation of bacterial fragments or whole bacteria into the systemic circulation leads to chronic immune activation and low-grade inflammation, which is the hallmark of metabolic endotoxemia as discussed previously. Such metabolic toxemia is related to the development of the metabolic syndrome, insulin sensitivity, and ultimately TD2 [181–183]. Interestingly, the severity of low-grade inflammation, as evidenced by biomarkers including C-reactive protein, TNF-α, and IL-6, is considered as a predictor of the progression from metabolic syndrome to TD2 at least in high-risk patients [184–186].

Translocated LPS exert pathophysiological effects adversely affecting the function of several organs and tissues. For example, elevated levels of TNF-α, IL-1β, IL-6, and NF-kB, produced by LPS-induced activation of TLR2 and TLR4, are upregulated in the liver, as well as in adipose tissue and striated muscles, and make a major contribution to the development of increased insulin resistance [184, 187, 188]. LPS activation of TLR4 and the subsequent increased production of proinflammatory cytokines in macrophages and beta cells conspire to decrease the function, and ultimately the viability, of the latter cells; this is a critical element in the development of insulin resistance and TD2 [189, 190]. The Th17/Treg balance is disturbed in patients with TD2, with elevated numbers of circulating Th17 T cells and reduced numbers and function of Tregs [191, 192]. The direct adverse effects on glucose metabolism of elevated LPS in the systemic circulation have been reported in an interesting study [193]. These authors reported that LPS activation of TLR4 in myotubes leads to the creation of an inflammatory environment and phosphorylation of AKT and impaired glucose transport. LPS-mediated activation of TLR2 also leads to the phosphorylation and activation of AKT [194]. This is highly significant, given data demonstrating that blunted AKT activity is required for the persistent immunosuppressive properties of Tregs [195, 196]. Metabolic endotoxemia may also have a direct inhibitory effect on Treg function leading to the unrestrained differentiation of Th17 T cells. Engagement of TLR2 receptors on naive T cells and on the surface of Tregs drives their differentiation toward the Th17 phenotype, with a loss of immunosuppressive properties [197].

It is clear that the abnormalities discussed above could, at least in part, stem from dysbiosis leading to impoverished levels of SCFA production and increased intestinal permeability. However, TD2 is a disease with a strong genetic component, with interactive polymorphisms in HLA and vitamin D receptor (VDR) being a critical element in increased disease susceptibility [198]. It is, however, of interest that these authors reported VDR polymorphisms to be correlated with HLA-DRB1*04 and with metabolic and immunologic parameters [198] as reduced SCFAs would likely compound this association.

TRYCATs and SCFAs in Type 2 Diabetes

As discussed above, dysbiosis leading to decreased SCFA production and resultant increase in the translocation of LPS into the systemic circulation likely underpins the state of metabolic endotoxemia seen in patients with metabolic syndrome, insulin resistance, and TD2. This state is characterized by increased levels of proinflammatory cytokines, which upregulate the activity of IDO and hence the activation of the TRYCAT pathway seen in these patients [199]. Metabolic endotoxemia also plays a major role in the development of insulin resistance, metabolic syndrome, and TD2 via the activation and dysregulation of the TRYCAT pathway in the blood, immune cells, pancreas, and CNS; this produces a range of adverse effects on the synthesis, release, and performance of insulin [200]. The activation of this pathway in patients with TD2, coupled with low levels of pyridoxal-5′-phosphate, leads to chronically upregulated levels of kynurenic acid (KYNA) and xanthurenic acid and a number of other TRYCATs, all of which have a range of adverse effects on insulin production and function [201–204]. Briefly, LPS translocated into the systemic circulation provokes the synthesis of proinflammatory cytokines from antigen-presenting cells, resulting in the activation of IDO.

Elevated levels of these TRYCATs display pro-oxidative, neurotoxic, and apoptotic effects via upregulation of inducible NO synthase (iNOS), arachidonic acid, prostaglandins, phospholipase A2, and leukotriene cascades, each of which play a major role in the development of metabolic syndrome and insulin resistance (reviewed in [205]). Kynurenine may be alternatively metabolized to form KYNA and, ultimately, quinaldic acid (QA) in a reaction catalyzed by the action of kynurenine aminotransferase (KAT) or 3-hydroxy kynurenine (3-HK), catalyzed by KYN 3-monooxygenase [54]. 3-HK, in turn, may either be metabolized to produce quinolinic acid and, ultimately, nicotinamide, with the first step being catalyzed by kynureninase or metabolized to produce xanthurenic acid, and 8-hydroxyquinaldic acid (8-HQ), catalyzed by 3-HK-transaminase. IDO, KAT, and IMO activity is far higher in the periphery than in the brain, and hence, the activation of the TRYCAT pathway in the periphery has a major impact on the performance of this pathway in the CNS [206]. IDO activation in the periphery results in the production of kynurenine, 3-HK, and quinolinic acid, which can readily cross the blood brain barrier and be further metabolized in the CNS. KMO is relatively inactive in microglia and is readily saturated by chronically elevated levels of kynurenine entering the brain, which is the case in the presence of metabolic endotoxemia in the periphery. This inactivation results in increased production of KYNA by the relatively active KAT [206]. Levels of KYNA are elevated in patients with TD2 [206], which is pathologically significant as this TRYCAT can inhibit the conversion of proinsulin to insulin in the pancreas [201].

Many patients with TD2 also present with low levels of pyridoxal-5′-phosphate (P5P), an active form of vitamin B6 that acts as an essential cofactor for kynureninase [203, 204]. Hence, this enzyme is extremely sensitive to the depletion of P5P [200], and the upregulated TRYCAT pathway, in an environment of decreased P5P, leads to an increased availability of 3-HK for 3-HK-transaminase and subsequently increased production of xanthurenic acid and 8-HQ. This increases in the availability of kynurenine for KAT and thus increases the levels of KYNA and QA in the brain, blood, urine, and pancreatic islets [207–210]. Chronically elevated levels of these TRYCATs, notably xanthurenic acid, can have a number of deleterious effects on the production and function of insulin. Xanthurenic acid inhibits the release of insulin and induces apoptosis of pancreatic cells [211, 212]. This TRYCAT can also chelate insulin, which can reduce the activity of the hormone [203, 213].

The cause or causes of P5P depletion in patients with TD2 remain a matter of controversy, but it is fair to say that this phenomenon is not unique to people suffering from this illness as it also occurs in patients with inflammatory bowel disease, rheumatoid arthritis, cardiovascular disease, and other illnesses underpinned by the presence of chronic systemic inflammation [214]. Indeed, there is accumulating evidence that the depletion of this essential coenzyme is a consequence of chronic inflammation, which drives the catabolism of pyridoxol to pyridoxic acid and hydrolysis of P5P by alkaline phosphatase [215, 216].

Now we turn from a consideration of an illness where excessive levels of SCFA production might play a pathological role, to a group of illnesses where excessive SCFA production, notably propionate, may well underpin the symptoms of a disorder.

Autism Spectrum Disorder

Evidence for Dysbiosis in the Pathophysiology of Autism

Autism spectrum disorders are a heterogeneous set of neurobiological illnesses whose etiology cannot be explained by genetics alone [217, 218]. Idiopathic gastrointestinal abnormalities are a common comorbidity in children with autism spectrum disorder, suggesting that abnormalities in the microbiota and microbial metabolites may play a role in the pathophysiology of these illnesses [217, 219–221]. Indeed, numerous reports, utilizing both animal models and human epidemiological evidence, have linked disruptive changes in the composition in the microbiota and subsequent abnormalities in microbial metabolites leading to immune dyshomeostasis and changes in the gut-brain axis, with the development of autism spectrum disorder [217–219, 221–223].

Several authors have reported increased intestinal permeability and a range of gastrointestinal abnormalities, such as dysmotility, in some children with autism [224, 225]. A recent study by Emanuele and colleagues [226] reported the presence of LPS and elevated levels of proinflammatory cytokines in autism. In a similar vein, several research teams have reported the presence of gut dysbiosis, with an altered composition of the microbiota in individuals with autism spectrum disorder [223, 227–234]. Spore-bearing anaerobes and microaerophilic bacteria, mainly Clostridia, are elevated in many people with autism spectrum disorder [228, 235, 236]. Infectious pathobionts are also implicated in the etiology of autism spectrum disorders [236–238].

SCFAs and Autism

One animal model of autism spectrum disorders is based on administration of high levels of the SCFA propionate. We now focus on the consequences of such administration and highlight the mechanisms by which dysbiosis, leading to an overproduction of propionic acid and other SCFAs, could underpin many of the abnormalities seen in children with a diagnosis of autism spectrum disorder.

Propionate acid is a fermentation product of many bacteria, including Clostridia, Desulfovibrio, and Bacteroidetes, that are markedly overabundant in stool samples from patients with autism spectrum disorder [228, 238]. Importantly, this dysbiosis corresponds with excessively elevated propionic acid levels in the stool samples of autism spectrum disorder patients [221]. Other studies report alterations in Bacteroides and Prevotella species, but by far the most consistent finding is a major increase in the levels and diversity of Clostridia species, with one study reporting almost a 50-fold increase [232, 235, 238]. This gives further grounds for suspecting the presence of excessive SCFAs in at least some patients afforded this diagnosis. The study conducted by Kang and others cited above is of particular interest as they reported that the presence and severity of “autistic” symptoms was significantly correlated with lower levels of Coprococcus and Prevotella and several species of currently unclassified Veillonellaceae species [231]. These species are all propionate producers, and hence, the results of this study refute the suggestion that excessive propionate production is the cause of abnormalities seen in all children with an autism spectrum diagnosis. However, results of other studies do support the proposition that excessive SCFAs could partly explain the observed pathology in some of these illnesses. The possible mechanisms whereby propionic acid might cause these effects are not completely understood but we will now consider the mechanistic explanations as understood thus far.

Potential Mechanistic Explanations for Propionic Acid-Induced Pathology in Autism Spectrum Disorder

Many environmental toxins are implicated in the development of autism spectrum disorder, including propionic acid and other metabolic products of commensal bacteria [239–241]. These SCFAs readily cross the gut-blood barrier and gain access to the CNS, at least partly via uptake by monocarboxylate transporters in the gut lumen[242]. Excessive levels of SCFAs in the systemic circulation can lead to adverse metabolic and neurobiological effects, leading to mitochondrial and CNS pathology proposed to play a causative role in the pathophysiology of autism spectrum disorder [243]. It is also of interest that propionic acid has a number of direct actions on GI physiology very similar to the pattern of GI pathology seen in patients with autism spectrum disorder (reviewed in [218]). Propionic acid inhibits mitochondrial functions, via the production of the cytotoxic propionyl CoA, and the subsequent sequestration of carnitine impaired membrane stability and levels of cardiolipin 51 [239, 244, 245]. It is of interest that intraventricular administration of propionic acid results in an abnormal acylcarnitine profile and relative carnitine deficiency seen in some children with autism spectrum disorder [246–248]. Carnitine plays a key role in β-oxidation of fatty acids and hence makes a major contribution to energy production; thus, depleted levels could be a source of relative mitochondrial dysfunction [249].

The capability of circulating propionic acid and other SCFAs to directly influence the brain via the assistance of monocarboxylate transporters in the blood brain barrier(BBB) is perhaps underappreciated [250, 251]. Propionic acid is internalized by microglia, astrocytes and, to a lesser degree, neurons [251, 252], which also express monocarboxylate transporting receptors [253, 254]. Once internalized, these SCFAs appear to be a major energy source involved in many aspects of cellular metabolism, particularly during postnatal brain development [251, 252, 255].

Perhaps most compellingly, animal models of autism spectrum disorder based on pulsed infusions of propionic acid and other SCFAs have demonstrated that such administration produces neurocognitive defects, abnormal motor movements, impaired social interactions, and other symptoms reminiscent of autism spectrum disorder [217, 239–241]. Moreover, the brain tissue of rats treated with propionic acid shows neurochemical changes reminiscent of autism, including glutathione depletion, neuroinflammation, oxidative stress, and altered phospholipid/acylcarnitine profiles [217, 239–241]. The presence of SCFAs in the CNS also affects lipid metabolism [256], the release and synthesis of neurotransmitters [257], oxidative stress-related mitochondrial functions [258], and intracellular pH.

Propionic acid can also modulate serotonin, dopamine, and glutamate systems in a manner similar to that observed in autism spectrum disorder [259, 260], largely by potentiating the release of intracellular calcium [261]. In general, SCFAs may increase dopamine synthesis via the induction of tyrosine hydroxylase in the sympathetic ganglia and the adrenals, leading to neuroendocrine abnormalities such as the cardiovascular instability and increased sympathetic tone seen in many children with a diagnosis of autism spectrum disorder [262, 263]. It must be stressed, however, that levels of propionate achieved by the mode of administration in such animal models may exceed those which can occur in children with autism spectrum disorder, and thus, extreme caution must be exercised in extrapolating such data. However, there is evidence that increased intestinal permeability or apoptosis of intestinal epithelial cells may allow for increased SCFA levels in the systemic circulation and consequently the CNS [264, 265]. In any event, SCFAs are unlikely to be the only metabolites ultimately derived from the microbiota involved in the pathogenesis of these illnesses, as abnormal TRYCAT levels are also implicated [266].

Summary and Conclusions

Tryptophan metabolites and IDO produced by microbial metabolites can shape the immune response by engaging aryl hydrocarbon receptors and promoting the differentiation of Th17, or regulatory T cells and IL-22 production by Th22 T cells, which in turn regulates epithelial barrier function. Elevated levels of IDO produced by a range of gut commensals increase the levels of kynurenine, which can also engage aryl hydrocarbon receptors. This induces the differentiation of Treg cells and increases the levels of IDO, leading to profound immunosuppression. Increased levels of IDO in the periphery can transverse the BBB leading to the production of toxic metabolites such as quinolinic acid. The composition of the microbiota is largely responsible for the levels of tryptophan in the systemic circulation and influences the levels of other neuroactive substances in the intestine, blood, and brain.

The levels of systemic tryptophan influence the performance of the serotoninergic and glutamatergic systems in the CNS and the production of neurotoxic molecules by the TRYCAT pathway in astrocytes and microglia. Microbiota also influence the levels of GABA, BDNF, dopamine, noradrenaline, and serotonin in the gut, the systemic circulation, and in the CNS. The presence of these and other neuroactive metabolites in the systemic circulation goes some way to explain the major role played by the microbiota in the functioning and development of the central nervous system and their potential to disrupt the brain functions leading to neuropathology.

Microbially generated SCFAs activate G protein-coupled receptors 41, 43, and 109a, on intestinal epithelial cells, which mediate many of their functions including the activation of intracellular signaling and the expression of pivotal immune and inflammatory cytokines and chemokines. SCFAs also activate G protein-coupled receptors 41 and 43 on specialized secretory epithelial cells, leading to the synthesis and secretion of GLP-1, which plays a major role in regulating epithelial barrier integrity. SCFAs also exert profound effects on chemokine production and neutrophil recruitment to sites of inflammation. Their capacity to promote the differentiation of Tregs, IL-17 producing Th17 T cells, and the pattern of cytokine production by antigen-presenting cells is mediated by their capacity to act as histone deacetylase inhibitors or activators of mTOR signaling.

SCFAs, notably butyrate, stimulate the secretion of glucagon-like peptides 1 and 2 via ligation of glucagon-like peptide-43, leading to decreased epithelial permeability and increased mucosal antibacterial defenses. On the other hand, dysbiosis leading to a loss of butyrate or propionate producers can result in adverse changes in the localization and distribution of two epithelial tight junction proteins, occludin and zonula occludens-1, leading to increased epithelial permeability and translocation of bacterial LPS into the systemic circulation.