Microbiota and host immune responses: a love–hate relationship

Microbiota: intestinal effects

The establishment of a mature microbiota is a dynamic process during the first 2 years of life5 and coincides with the development of the immune system. Throughout the early developmental period innate immune components play key roles in protecting the infant from pathogens and shaping microbiota assembly. IgA is found in breast milk and can prevent immune activation in infants by binding microbial antigens. Similarly, secretory IgA produced along the intestinal tract continues to be important for maintaining mucosal homeostasis through adulthood.8 The development of the mature microbiota is regulated by host immune system components, which can also be influenced by members of the microbiota. Recent work with gnotobiotic mice suggests that proteobacteria, the dominant phylum in newborns, triggers a proteobacteria‐specific IgA response in mice that plays a key role in controlling proteobacteria levels in the adult microbiota.9 Faecal IgA levels (low versus high) are partly controlled by members of the microbiota; a phenotype that is vertically transmissible and independent of host genetic factors.10 16S rRNA sequencing revealed that Sutterella species are partly responsible for variable IgA levels, most probably by degrading IgA secretory component10 (Fig. 1). Noteably, expansion of proteobacteria/Enterobacteriaceae abundance is observed in patients with IBD and in pre‐clinical models.11 Whether this bloom of microorganisms is related to microbe‐mediated faecal IgA levels is unknown.

The microbiota continues to affect immune function well after development. Studies of Paneth cells using organoids generated from mice reveal that degranulation (release of antimicrobial products) is controlled by immune‐cell‐derived interferon‐γ, which may be induced in vivo during viral or bacterial challenge.12 Thymic and induced T regulatory (Treg) lymphocytes prevent autoimmunity and maintain tolerance to the microbiota, and a recent study suggests that most colonic Treg cells are thymic Treg cells that recognize bacterial antigens, including antigens from Clostridiales, Bacteroides and Lactobacillus. Importantly, antibiotic‐induced alterations in the microbiota, which decrease Clostridiales members among others, reduce intestinal Treg cells and alter colonic thymic Treg T‐cell receptor repertoire, suggesting that microbial composition influences the dynamic response of Treg cells.13

One of the most studied immunomodulatory microbes are segmented filamentous bacteria (SFB), which colonize the terminal ileum in mice, induce IgA production and increase effector T cells, particularly T helper type 17 (Th17) cells.1 Work from several groups suggests that SFB induction of Th17 cells occurs in the intestinal lamina propria rather than in Peyer's patches or the mesenteric lymph nodes.14, 15, 16 Further studies revealed that MHC class II (MHCII) ‐dependent antigen presentation by intestinal dendritic cells (DCs) is essential for SFB‐induced Th17 cells.15, 16 Additionally, Goto et al. provide evidence that MHCII presentation by innate lymphoid cells (ILCs) may constrain Th17 cell differentiation.15 The SFB also stimulate expansion of germinal centres and induce IgA‐secreting cells in Peyer's patches, isolated lymphoid follicles and tertiary lymphoid tissue.14 Recently, Schnupf et al. cultured SFB in vitro and provided evidence that SFB attachment in vivo is required to elicit ileal epithelial responses.17 Future studies examining the specific structural component(s) of SFB responsible for stimulating IgA production and inducing Th17 cells could be aided by the in vitro culture system. Although SFB have not yet been isolated from the human gastrointestinal tract, human cell lines support SFB growth17 and SFB‐specific 16S rRNA has been detected within human stool samples.18 Gram‐stained human ileal–caecal biopsies from a small set of patients with IBD and non‐inflamed patients suggest that SFB is present in patients with ulcerative colitis but absent in those with Crohn's disease.19 Hence, future studies may reveal a role for SFB or SFB‐related bacteria in human immune development and IBD.

In addition to interactions with the immune system, microbes interact with other microorganisms. Although many symbionts have beneficial immune properties, some bacteria facilitate host infection by viruses. For example, human norovirus probably binds to histo‐blood group antigen‐expressing bacteria such as Enterobacter cloacae, which promotes attachment and infection of B cells.20 Conversely, treating mice with the bacterial product, flagellin, prevents rotavirus infection by modulating host innate immune signalling.21 Hence, microbiota composition can promote or inhibit viral infection depending on the type of virus.

Microbiota effects on extra‐intestinal immunity

The effects of the microbiota on host immunity extend beyond the intestine. Germ‐free zebrafish have fewer and less active neutrophils compared with zebrafish colonized with a normal microbiota, as well as impaired neutrophil recruitment in a tail‐fin injury model; a phenomenon linked to microbial induction of serum amyloid A.22 Neonatal mice that are GF or born from antibiotic‐treated dams have fewer circulating and bone marrow neutrophils and are more susceptible to Escherichia coli K1 and Klebsiella pneumoniae sepsis, probably through microbiota induction of granulocytosis.23 Hence, the microbiota contributes to neutrophil development, homeostasis and function in both mice and zebrafish.22, 23 The GF mice have increased invariant natural killer T (iNKT) cells in the lung and colon due to enhanced CXCL16 expression, making them more susceptible to an ovalbumin‐driven model of allergic asthma.24 These studies also suggest that early exposure to microbes is important, as the iNKT cell levels returned to low levels when GF mice were exposed to specific pathogen‐free conditions upon birth but not as adults.24 Consequently, the intestinal microbiota has both local and systemic effects on innate and adaptive immunity (Fig. 1).

Although the majority of the microbiota resides within the intestine, the microbial communities located in extra‐intestinal regions also influence local host immunity.25 Studies comparing GF mice to conventionally raised mice suggest that the skin microbiota regulates expression of complement component C5a receptor, which regulates innate immune defence genes, thereby impacting microbiota diversity and composition.26 Certain skin microbiota community members, particularly Staphylococcus epidermidis interactions with CD103⁺ DCs, can induce CD8⁺ T‐cell migration to the epidermis, which enhances barrier function and limits epicutaneous Candida albicans infection through induction of interleukin‐17 (IL‐17).27 In a skin wound healing model, wound closure rate was restored in GF mice conventionalized with microbiota, a phenomenon associated with increased neutrophil accumulation and lower macrophage infiltration into the injured region.28 Microbial dysbiosis has also been implicated in extraintestinal diseases, such as increased Staphylococcus aureus, which is associated with inflammatory skin conditions.29 Hence, microbes occupying extra‐intestinal niches also influence host immunity, although the specific organisms and mechanisms responsible for these responses (Table 1) can differ between regions.

Structures detected by pattern recognition receptors

The innate immune system detects microbial components or products through several different families of pattern recognition receptors (PRRs), found on numerous cell types including macrophages, DCs and epithelial cells.3 Toll‐like receptors (TLRs) are a class of transmembrane PRRs located on either the cell surface or in endosomes.31 One of the most characterized bacterial immunomodulators is Bacteroides fragilis polysaccharide A (PSA), which is recognized by TLR2 and capable of influencing T‐cell development and homeostasis.32 Recent studies reveal that PSA activates TLR2 on plasmacytoid DCs rather than conventional DCs, leading to the induction of IL‐10 secretion by CD4⁺ T cells and mucosal protection during a 2,4,6‐trinitrobenzene sulphonic acid model of colitis.33 PSA–TLR2 activation of pDCs and Treg cell induction can also mediate protection in extra‐intestinal inflammatory diseases such as experimental autoimmune encephalomyelitis, a multiple sclerosis animal model.33, 34 Additional work characterizing PSA‐induced Treg cell activation via MHCII‐mediated antigen presentation suggests that the interaction depends on the zwitterionic (carries positive and negative charges) properties of PSA and induces a specific clonal expansion of Treg cells.42 Lactobacillus plantarum teichoic acid d‐alanylation (component of Gram‐positive bacterial envelope) also signals through TLR2 and promotes a pro‐inflammatory cytokine response in DCs, which modulates effector and regulatory T‐cell populations45 (Fig. 2).

Bacteria flagellin is recognized by TLR5 expressed on various cells including intestinal epithelial cells (IECs) and DCs. The IEC‐derived TLR5 signalling appears to influence microbiota composition and host response because TLR5^ΔIEC mice have an altered microbiota compared with co‐housed sibling wild‐type controls, develop low‐grade inflammation and metabolic syndrome, and have delayed clearance of adherent invasive Escherichia coli (AIEC).46 How TLR5 activation controls composition of the microbiota is unclear but could involve immune cell recruitment to clear pathogens in close proximity to the epithelium, stimulation of epithelial antimicrobial peptide production, or induction of flagellin‐specific IgA.46 The microbiota also impacts vaccine immunity through TLR5 signalling in B cells and macrophages, which is critical for mounting an antibody response to trivalent inactivated influenza vaccine and the inactivated polio vaccine.47

Most of the downstream signalling from TLRs occurs through either myeloid differentiation primary response protein 88 (MyD88) or Toll–interleukin receptor domain‐containing adaptor protein inducing interferon‐β (TRIF) adaptor proteins, resulting in activation of nuclear factor‐κB or interferon regulatory factors, respectively.48 Luminal bacteria promote mucus secretion and movement of monocytes closer to epithelial stem cells through an epithelial MyD88‐signalling pathway. Increased proximity of monocytes to epithelial stem cells results in increased crypt cell proliferation and intestinal stem cell numbers,49 which could be beneficial during intestinal injury response. Studies comparing GF mice to mice colonized with three strains of bacteria (E. coli K‐12, Staphylococcus xylosus and Enterococcus faecalis) reveal that GF mice have delayed microbial clearance, reduced inflammatory responses to intravenous E. coli K12 infection and a decreased myeloid cell pool size.50 Heat‐stable microbial antigens in the serum are able to restore bone marrow myeloid cell numbers through MyD88/Toll‐interleukin 1 receptor domain‐containing adaptor molecule (TICAM)‐dependent TLR signalling.50 MyD88‐dependent TLR signalling also plays a role in microbiota‐mediated tolerance to a non‐invasive strain of Salmonella enterica serovar Typhimurium by preventing CX₃CR1^hi mononuclear phagocyte‐mediated transport of luminal bacteria to the mesenteric lymph nodes.51 Furthermore, tumour necrosis factor receptor associated factor 6 (TRAF6), a component of TLR signal transduction, has MyD88‐independent effects on immune and microbiota homeostasis. Mice with TRAF6‐deficient DCs develop Th2‐driven small intestine inflammation and have decreased Treg cells, both of which are microbiota‐dependent.52

Nod‐like receptors (NLRs) are a class of cytosolic PRRs that act as intracellular sensors.3 Nod2 recognizes bacterial peptidoglycan through muramyl dipeptide. Studies using Nod2 ^−/− mice reveal reduced intraepithelial lymphocytes (IELS) and administering muramyl dipeptide to antibiotic‐treated mice restored IEL numbers through up‐regulation of IL‐15, suggesting that Nod2‐mediated recognition of the microbiota affects IEL homeostasis.53 Nod2 has also been implicated in preventing goblet cell dysfunction and restricting expansion of Bacteroides vulgatus in the small intestine, which prevents piroxicam‐induced intestinal inflammation in mice.54 The NLRs also have extra‐intestinal effects, as NLR ligands have been implicated in innate immunity in the lung, which is important for Klebsiella pneumoniae clearance.43

Inflammation or injury can cause members of the microbiota to become pathogenic and stimulate the immune system to induce inflammation. In the context of a dextran sulphate sodium (DSS) mouse model, Proteus mirabilis can induce IL‐1β production via NLR family, pyrin domain containing 3 (NLRP3) inflammasome activation in recruited inflammatory monocytes, promoting intestinal inflammation.44 After comparing different strains, the authors determined that Proteus mirabilis HpmA haemolysin induces K⁺ efflux, which is required for NLRP3‐induced inflammasome activation.30, 44 In vitro studies show that AIEC isolated from patients with IBD are also able to induce IL‐1β through NLRP3 activation in macrophages.55

Outer membrane vesicles

Outer membrane vesicles (OMVs) are produced by Gram‐negative bacteria and contain various bacterial components, many of which activate PRRs.39 The OMVs can promote immune homeostasis or enhance bacterial pathogenesis; effects that probably depend on the type of bacteria, OMV content and host environment.39 For example, Bacteroides thetaiotaomicron OMVs containing homologues of mammalian inositol phosphatase interact with IECs in vitro to promote intracellular calcium signalling.40 This signalling confers nutritional benefits and potentially anti‐carcinogenic properties, as dietary inositol hexaphosphate administration reduces tumorigenesis in carcinogen (1,2‐dimethylhydrazine or azoxymethane) ‐induced cancers in rats and mice.40 On the other hand, spontaneous colitis‐prone CD4‐dnTgfb2;IL10rb ^−/− mice exposed to B. thetaiotaomicron develop inflammation due to OMVs containing sulphatase activity, which degrades mucin glycans and allows B. thetaiotaomicron to interact with host macrophages.35 Bacterodies fragilis PSA is also released in OMVs, which can be detected by TLR2.36, 56 Some bacteria produce OMVs that have adverse effects on host immunity. For example, enterotoxigenic B. fragilis secrete B. fragilis toxin‐dependent particles that can induce host IECs to secrete sphingolipids (specifically, sphingosine‐1‐phosphate) in exosome‐like particles, which induce Th17 cells and enhance tumorigenesis in multiple colon cancer mouse models.56

Metabolites

Besides structural components, bacteria also generate a wide spectrum of metabolites that have the capacity to engage and trigger numerous host responses (Fig. 2). Multiple intestinal Bacteroides species are able to synthesize sphingolipids, which are structurally similar to host lipid agonists of iNKT cells. Bacteroides fragilis sphingolipids have been shown to modulate cellular homeostasis by both promoting iNKT cell activation37 and inhibiting activation and expansion of iNKT cells during mouse neonatal development, which protects against oxazolone‐induced colitis in adulthood.38

Short‐chain fatty acids (SCFAs) are bacterial metabolites generated as by‐products of dietary fibre fermentation. Butyrate, propionate and acetate are the most common intestinal SCFAs and are normally present in the millimolar range in the gut. The mechanisms by which SCFAs impact immunity include activation of G protein‐coupled receptors (GPRs), inhibition of histone deacetylases, and regulation of autophagy.57 Levels of SCFAs depend on two interdependent factors: dietary fibre and microbiota composition. The SCFAs may modulate protection against chemically induced DSS colitis through GPR43 and GPR109A receptor interactions, which are dependent on the NLRP3 inflammasome in non‐haematopoietic cells.58 Butyrate exerts anti‐inflammatory effects on bone marrow‐derived and colonic macrophages via histone deacetylase inhibition. However, gavaging mice with butyrate does not impact the outcome of DSS colitis, suggesting that butyrate promotes bacterial tolerance rather than tissue repair.59 Additionally, butyrate has been shown to increase barrier function by stimulating epithelial metabolism in the colon to promote oxygen depletion, stabilizing hypoxia‐inducible factor and inducing hypoxia‐inducible factor‐dependent target genes that promote barrier function.60 SCFAs from the microbiota can also have systemic effects. In a mouse model of allergic inflammation in the lung, high levels of propionate are protective, probably through GPR41 signalling, which results in DCs with high phagocytic capability and an impaired capacity to induce Th2 differentiation.41

In addition, microbial metabolites can influence host immune responses through an indirect route. For example, SCFAs augment 5‐hydroxytryptamine (serotonin) production from intestinal enterochromaffin cells through up‐regulation of the rate‐limiting biosynthetic enzyme tryptophan hydroxylase.61 The wide impact of serotonin on host biological response including immunity62 suggests that microbes could shape immune responses through complex mechanisms. SCFAs also directly impact innate immune cells in the brain and central nervous system. Germ‐free mice have defects in microglial (tissue macrophages of the brain) maturation, differentiation and function with a diminished response to lipopolysaccharides and viral challenges, while administering a mixture of propionate, butyrate and acetate to the drinking water restores microglial maturation.63 Indigenous bacteria metabolites may have key roles in inhibiting colonization of specific pathogens. For instance, Clostridium scindens inhibition of Clostridium difficile is associated with secondary bile acid synthesis.64

Tryptophan catabolites from the microbiota expand lactobacilli that produce an aryl hydrocarbon receptor (AhR) ligand, indole‐3‐aldehyde. Activation of AhR results in IL‐22 transcription that promotes antimicrobial peptide expression and mucosal homeostasis, providing colonization resistance against gastrointestinal or vaginal Candida albicans infection and DSS colitis models.65 Group 3 ILCs also rely on AhR signalling to inhibit Th17 cell expansion and regulate SFB levels, which could play a role in IBD because ~ 40% of mice that lack AhR signalling abilities in ILCs develop spontaneous colitis between 12 and 20 weeks of age and have exacerbated inflammation in a CD45RB^hi T‐cell transfer model of colitis.66 In summary, bacterial components and metabolites affect both innate and adaptive immunity in the intestine (Fig. 2).