How did the immune system cope with this massive co-evolving community of diverse microbial species in close contact with the host tissues? The innate immune system uses inherited pattern recognition receptors (PRR) to recognise conserved microbial components. However, pathogens can evolve quickly and produce structures that are not recognized by existing PRRs. The innate immune system can try to adapt by duplicating and modifying PRR genes, but this process is slow and results in a cluttered genome. The adaptive immune system in vertebrates solves this problem by using somatic hypermutation to create a large variety of receptors with little increase in genetic complexity. This involves mutating the genes that encode B and T lymphocyte receptors to produce a diverse array of clones. While this reduces the genetic load, it also creates new problems such as the production of useless or autoreactive lymphocytes. However, since each clone expresses only one receptor, any cell line that recognises nothing or that recognises the host's own tissues can be eliminated. Most autoreactive cells are eliminated in the thymus which expresses self-antigens. However in order to select and retain lymphocyte clones that recognise a diversity of microorganisms, and that can manage and tolerate a diverse microbiota while eliminating pathogens, the adaptive immune system must obtain data from microbial inputs, acquired mostly from mother, family and the environment (31). Thus the immune repertoire of each new individual is matched to the microbial environment into which that individual was born.

An input of microbial antigens and epitopes is needed in early life to select the mutated lymphocyte clones that need to be retained as memory cells. All life forms are constructed at least in part from variants of the same building blocks that evolved long ago in early microorganisms (24, 25). This is important because it means that if individual humans or animals are exposed to a sufficiently diverse range of microbial epitopes, the lymphocyte clones that are retained and expanded will, by chance, contain some that recognise viruses or other pathogens to which the individual was never previously exposed such as HIV or COVID-19 (31–33). This point emphasises the need for exposure to microbial biodiversity (19). Interestingly, biodiversity of the gut microbiota has a strong correlation with health (64). Even declining health in old age is associated with a decrease in gut microbiota biodiversity (35, 65). However other possible functions of biodiversity are suggested in the next paragraph.

It may not be only the diversity itself that is important but also the increased likelihood of having essential species that drive functions such as immunoregulation (discussed later) (36,37), or other undiscovered necessary functions. For example, a single protein synthesised by a single organism is required for the expansion of the pancreatic β cells in the zebra fish (66). We do not know if there are similar hidden requirements in the human microbiota because we cannot recolonise germ-free humans with one species at a time in order to identify such dependencies.

Another possibility is that complex ecosystems are more stable. Species diversity can protect ecosystems from excessive damage caused by environmental change because having many species increases the chance that some can quickly adapt to the new conditions (67).

Biodiversity also prevents dangerous biofilm formation. An organism's physiology changes when it forms biofilm. Some organisms become more pathogenic, resistant to the immune system and even resistant to antimicrobials. Much of the pathology caused by

Candida albicans occurs when it switches from yeast to hyphal forms during biofilm formation (68). In patients with IBD, gut microbiota can penetrate the mucus barrier and form biofilm that adheres to the epithelial surface. Bacteria from the microbiota of healthy donors do not cross human intestinal epithelial cell monolayers in vitro but organisms from the biofilm can do so (69). High biodiversity of the gut microbiota may affect quorum sensing signals and prevent the switch to biofilm.

The roles of the major gut microbiota-derived metabolites in the regulation of immune function were reviewed recently (12) and will be outlined very briefly here and in Table 1. Short chain fatty acids (SCFA) such as acetate, propionate and butyrate, derived from the microbial fermentation of dietary fibre, can enhance production of IL-10, TGF-β and Treg (12, 41). Secondary bile acids generated by the microbiota, notably derivatives of lithocholic acid, can downregulate Th17 cells and enhance production of RORγ + Treg (12, 44). Several bacterial tryptophan catabolites including various indoles and indolepropionic acid promote Treg differentiation via the aryl hydrocarbon receptor (AhR) (12, 43). Branched-chain amino-acids such as valine, leucine and isoleucine are present in the diet, but are also generated by the microbiota, and maintain Treg (45).

The Treg populations in the skin are attracted and expanded by signals from skin microbiota (15). Lipases from components of this microbiota (Corynebacterium, Staphylococcus, and Micrococcus) act on lipids secreted by sebaceous glands to generate butyrate and other SCFA (70), which have immunoregulatory properties as outlined above. Moreover, some authors suggest that ammonia-oxidising bacteria (AOB) and archaea that colonised human skin in the past converted the high levels of ammonia and nitrate in human sweat into nitrite and nitric oxide which entered through the skin and exerted immunoregulatory functions (71), notably, downregulation of Th2 responses (72). Because these organisms are exquisitely sensitive to alkylbenzene sulfonate detergents this may no longer occur, though clinical trials applying AOB to the skin are being performed (73).

Exposures to microbial components that activate PRRs such as TLR4, TLR2, TLR9 or AhR or PI3K/Akt/mTORC1 signaling systems drive the establishment of immunoregulatory mechanisms. These microbial components transiently trigger inflammation, but repeated low dose exposures may prime anti-inflammatory mechanisms due to the release of IL-1β which can induce tolerance to itself and to endotoxin (LPS) (74). Therefore exposure to these microbial signals informs the immune system about the nature of the microbial environment and triggers immunoregulatory epigenetic adjustments. For example, LPS induced Treg via tolerogenic dendritic cells and TGF-β in an animal model (75). LPS in dust in the farming environment may protect against allergic responses by inducing A20 in lung epithelial cells (76). A20 is a potent inhibitor of the NF-κB signaling pathway and its expression is increased in Amish farmers using traditional farming methods (46, 47).

In mice, administering a TLR2 agonist resulted in a decrease in Th17 cells and an increase in type 1 regulatory T cells in the spleen. This led to a reduction in the severity of Experimental Autoimmune Encephalomyelitis (EAE) (77). The same researchers discovered that patients with Multiple Sclerosis (MS) had significantly lower levels of a TLR2 agonist derived from bacteria in their blood compared to healthy individuals (77).

TLR9 is an intracellular PRR that recognizes CpG motifs that are not methylated. These motifs, common in microbes, typically trigger an inflammatory response. However, some variants of CpG motifs and other microbial DNA sequences may have anti-inflammatory properties (78, 79). This appears to be the case for many species of Lactobacillus (80) which could explain why the probiotic effects of lactobacilli depend on the presence of TLR9 in the gut (81).

DNA is exchanged by horizontal gene transfer (HGT) both between different gut-resident microorganisms, and between gut microbiota and organisms in the environment (50–52). This exchange can occur between species that diverged in an evolutionary sense in the distant past, and constitutes a global network of gene exchange that can help strains to adapt rapidly to new diets and metabolic needs (51, 52). It is not clear whether HGT modulates the organisms that drive development of immunoregulatory mechanisms.

Host-derived microRNAs (miRNA) in membrane-bound extracellular vesicles (EV) can modify microbial gene expression. Mice that lack the miRNA-processing enzyme, Dicer, cannot form miRNAs. Such mice developed abnormal gut microbiota and exacerbated colitis, but administrating faecal miRNA corrected these abnormalities (82). So the host regulates gene expression in the gut microbiota. But this type of communication is 2-way. Recent work has shown that microbiota-derived EV have profound physiological effects on host metabolism, on the regulation of the immune system and on Treg numbers (53).

The microbial diversity of air is comparable to that of seawater, soil, and the human gut (83). Moreover, the lungs and airways constitute a sense organ with cellular sensors that can detect biogenic aerosols in inhaled air whether derived from microorganisms or from other sources such as plants. Plant polyphenols like quercetin, resveratrol and curcumin can reduce inflammation through the AhR (84). Microbial pigments such as phenazines and naphthoquinones can also regulate inflammation and anti-bacterial responses (48). All of these molecules and others from algae and higher plants can inhibit protein kinases of the PI3K/Akt/mTORC1 signaling system, which is believed to have an anti-inflammatory effect (49). This is likely to be relevant to the immunoregulatory benefits of exposure to the natural environment discussed later.

The seminal observation that hay fever was less prevalent in children with older siblings (85) led initially to speculation that modern domestic hygiene was reducing exposure to the common infections of childhood leading to imbalances within the immune system. However this hypothesis was unlikely because humans cannot be in a state of evolved dependence on these infections. They are mostly “crowd infections” that could not have persisted in isolated ancestral hunter-gatherer groups. Measles, for example, probably did not hit human populations until late in the Roman empire when appropriately large populations existed (86–88).

The debate was rapidly resolved when epidemiological studies revealed that these crowd infections of childhood do not protect from allergic disorders (89–91), and often trigger or exacerbate them (89, 92, 93).

We now understand that the protective effect of older siblings, while unquestionably correct, is likely to be due to increased transmission of the microbiota of mother and of the natural environment, as discussed later (94, 95). This does not mean that the common infections of childhood have no effect on the immune system but their effect, apart from inducing immunity to themselves, is non-specific activation of the innate immune system (“Trained Immunity”) rather than amplification of down-regulatory anti-inflammatory pathways, as briefly explained in the next section.

The transfer of the co-evolved human microbiota (29, 30) from mother (and siblings) to infant is critical for the development of the infant's microbiota, as well as the immune and metabolic systems (94). Certain lifestyle factors, such as Caesarean deliveries, lack of breastfeeding, poor diet (discussed in greater detail below and in Table 2), antibiotic use, and insufficient mother-infant intimacy, can reduce this transfer, and are associated with an increased risk of immunoregulatory disorders (94, 95, 101). Some organisms in the child's microbiota appear later and continue to accumulate until 5 years of age (131). These organisms are probably acquired from other family members and at day-care centres, as well as from the natural environment. Microbial strains can be transmitted person-to-person through normal social and mother-infant interactions both within and outside the home (132, 133). Such transfers may be diminished by modern lifestyles.

Can exposure to the microbiota of modern homes be considered essential? If we think in terms of evolution the response must be that it depends on the home. In the past, humans lived in natural shelters such as caves, or constructed homes from natural materials like stones, mud, branches, and leaves. Later more sophisticated homes were constructed by rearranging natural materials such as timber, stone, straw, soil, clay, animal dung, thatch or turf. The microbial makeup of such homes would have been similar to that of the surrounding natural environment. Even when damp and decaying, the organisms present would have been those with which humans co-evolved. However, modern homes made from synthetic materials, biocide-treated timber, plywood, and synthetic gypsum board harbour a microbiota that is different from that of the natural environment (134, 135). This difference is more pronounced in urban homes that are distant from nature (136). Furthermore, when a modern home is damp and deteriorating, as is often the case in households of low Socioeconomic Status (SES), the bacterial and fungal microbiota can produce secondary metabolites that are hazardous to human health, leading to varying degrees of “Sick Building Syndrome” (104–137) and increased likelihood of children being hospitalized for respiratory infections (107). Therefore, it is improbable that the unnatural microbiota of modern homes, especially when the home is of low SES, provides necessary or desirable microbial exposure for infants. However, when the microbiota of homes is similar to that of farms and the natural environment, it can be beneficial, particularly for disorders related to faulty immunoregulation like asthma, as discussed later (138–140).

In addition to the microbiota of mother and family, our evolving hunter-gatherer omnivore ancestors were inevitably exposed to the microbiota of the natural environment and of animals that were hunted, eaten or domesticated. During the late 19th century Blackley noted that farmers were less prone to hay fever than were people residing in urban areas (1). Since then, numerous studies have supported the notion that early-life exposure to a farming environment can reduce the prevalence of allergic disorders (46, 138). Some of the protective effect appears to come from early life exposure to farm animals (152) or dogs (139, 153) which cause measurable changes to the gut microbiota (154). Furthermore, living close to green spaces can also decrease the risk of allergic sensitization (155). Some of these studies have revealed immunological evidence that strongly suggests a cause-and-effect relationship, rather than a mere coincidental association (46, 47, 155). For example, Amish farmers who use traditional farming techniques have a very low incidence of allergic disorders and were found to have increased biomarkers of immunoregulation, compared to industrialized Hutterite farmers (46, 47). Similarly, exposing children to natural biodiversity in their school playgrounds in a controlled clinical trial resulted in an increase in peripheral blood biomarkers of immunoregulation (109). The evidence supporting the protective effect of exposure to the microbiota of the natural environment is less complete for other chronic inflammatory disorders, but there is suggestive evidence for IBD (156), autoimmune diseases (38) and psychiatric disorders (110) and metabolic and cardiovascular diseases (111).

Inevitably there are sometimes pathogens in respired air, but very low doses typically lead to protective immune responses rather than disease. When bacteria attach to the nasal mucosa, released exosomes may transfer inducible nitric oxide synthase to neighbouring epithelial cells and increase release of nitric oxide (167). Pathogens also trigger the release of cathelicidin, and other human antimicrobial peptides (AMP) of which there are more than 100 (39). When bacteria and cathelicidin enter the cell the NLRP3 inflammasome is activated and a cascade of events is initiated, including the activation of caspase 1, the death of infected cells, and the release of pro-inflammatory cytokines IL-1β and IL-18. These events enhance inflammation and recruit neutrophils which are induced to form networks of extracellular fibres consisting mainly of DNA called Neutrophil Extracellular Traps (NET). These NETs contribute to inactivation of microorganisms (39).

Ultimately, these mechanisms in the airways kill or disarm the respired organisms, which are then taken up by the lymphoid tissue of Waldeyer's ring or exposed to acid in the stomach before being sampled by the dendritic cells in the small bowel (40). In this way, inspired low doses of pathogens may provide useful data to the immune system and prime immunity to potential pathogens, but they do not appear to be crucial for setting up immunoregulation.

The order in which live or killed vaccines are given can determine the nature of their non-specific effects (59), indicating that the order in which vaccines are given determines the epigenetic changes that they induce. This may also apply to the sequence in which infections are experienced (168). For example it has been suggested that acute lymphocytic leukaemia (ALL) may result from delayed exposure to an agent that, during human evolution, would have been encountered in very early life (169). Interestingly ALL is more common when early life exposure to microorganisms is reduced, such as after Caesarean delivery, lack of breastfeeding and lack of older siblings (169). This is similar to risk factors for immunoregulatory disorders like allergies (94, 95).

A similar concept has been proposed to explain the increase in T1D that seems to have appeared in parallel with the development of the modern Western lifestyle. The prevalence of T1D fell after the introduction of rotavirus vaccines administered at 2 months (170).

It was reported 20 years ago that hay fever and asthma were less frequent in subjects seropositive for hepatitis A virus (HAV), Toxoplasma gondii, and herpes simplex virus 1 (171). It was not clear whether these infections were markers of plentiful food-borne and orofaecal microbial exposure to family and environment, or whether the infections were themselves enhancing immunoregulation. Some of these are persistent infections which deserve to be considered because they may drive immunoregulation in order to limit immunopathological damage to the host. A few particularly obvious examples of infections that exert non-specific effects on the immune system are considered briefly below.

Epidemiologists seeking links between low SES and allergic disorders should perhaps take note of the items in this list which can perhaps explain discordant results in this literature. In developing countries, low SES is often associated with subsistence agriculture in rural settings with abundant exposure to the natural environment. On the other hand, in wealthy urbanised countries low SES is associated with living in polluted urban slums with little access to green space while the wealthy have gardens and rural holiday homes. Similarly in developing countries access to Caesarean deliveries and antibiotics may be difficult for people of low SES, whereas in some rich countries, where these things are readily available, misuse of Caesarean deliveries and antibiotics can be more frequent amongst people of low SES, perhaps because they lack awareness of the disadvantages (193). These points can explain the enormous discrepancies between different epidemiological studies of allergic disorders which, for example, may or may not find links with low SES (194).

Treg biology provides another explanation for discordant epidemiology. There are several different subsets of Treg, which act at different stages in the development of an allergic response, including the initial expansion of T cell clones, polarisation towards Th2, regulation of IgE vs. IgG4 production and late effector pathways such as control of mast cell activation (195). Thus some patterns of Treg activation will block sensitisation, while other patterns of activation will fail to block sensitisation, but may still block clinical manifestations. This explains some instances of a lack of correlation between IgE or skin-prick test positivity and clinical manifestations of allergic disease (196). Different Treg subpopulations, located in different sites, might also explain the very variable correlation between skin, airway and gut manifestations of allergy (197, 198). We need more understanding of the role of different microbial exposures in expanding Treg cells that operate at each stage of the allergic response (195), and at different epithelial surfaces (198).