Microbiota, regulatory T cell subsets, and allergic disorders


 Epidemiologic studies revealed a crucial role of the environment for the increased prevalence of allergic disorders. The microbiota as part of our immediate environment promotes immune diversity that facilitates a well-equilibrated balance between immunity and tolerance. Alterations of our symbiotic microbiota especially in early life is thought to play a fundamental role in defining susceptibility to the development of allergic diseases during adult life on the population level. Due to a high density of bacteria, viruses and fungi and a large contact surface area for host-microbiota interactions, the most relevant interaction between microbes and our immune system are thought to occur in the gut. The immune system co-evolved with the symbiotic microbiota and adopted a variety of mechanisms to allow a dynamic state of tolerance, including the induction of regulatory T cells (Tregs). Foxp3-expressing Tregs are well-described immune regulators in autoimmune and allergic disorders. However, recent years have shown that Tregs can come in different flavours with different regulatory potential and outcome for our immune system. This review summarizes novel findings from basic immunology research that may help to better understand the interaction between the microbiota, differentiation of Tregs and its consequences for the onset and regulation of allergic disorders.


Introduction e di erentiation of naive T cells to interleukin-4-(IL-4)-secreting T cells is one of the hallmarks of allergy. For a long time, it was completely unclear why and how harmless antigens/allergens provoke in certain individuals the di erentiation of T helper ( ) 2 cells. In contrast, it is well known that exposure of naive T cells to the cytokine IL-4 is a key event for triggering 2 di erentiation which will become self-sustaining as soon as the di erentiating 2 cells start to produce this cytokine themselves [1]. e cellular source of the initial IL-4 has remained enigmatic because neither naive CD4 + T cells nor dendritic cells are able to produce IL-4 themselves. However, a variety of innate or "innate-like" cells that carry the transcriptional capacity to rapid ly secrete IL-4 without prior di erentiation steps were identi ed in newly generated reporter mice; these include basophils, eosinophils, mast cells, γδ T cells, and natural killer T (NKT) cells [2]. Additionally, a newly identi ed cell type called innate lymphoid cell type 2 (ILC2) is thought to be critical for the induction and ampli cation of type 2 immune responses [3].
is resulted in the 1/ 2 concept of T helper cell di erentiation: IL-4 counteracts the e ect of IFN-γ and thus the di erentiation of 1 cells whereas IFN-γ blocks the di erentiation of 2 cells. e basis of this discovery is still used today for the treatment of patients that su er from 2-dominated allergic diseases: the repeated exposure to dened amounts of allergens is used to induce a 1-dominated immune response that helps the immune system to overcome a pro-allergic 2 immune response. A second possibility of the immune system for regulation of 2-dominated immune responses lies in the generation of allergen-speci c regulatory T cells (Tregs).
However, it is not clear which extrinsic factors determine whether allergen-speci c 2 cells or allergen-speci c Tregs are induced or whether certain Treg subsets are generated to suppress pro-allergic immune responses. e decision of the immune system between 2 and Treg induction upon recognition of harmless antigens may indeed re ect the immunological di erence between allergic and non-allergic individuals and a better understanding of this decision is therefore of high relevance. In recent years, it has also become clear that the induction of Tregs is in uenced by symbiotic microbes and may therefore provide one possible link between our environment and the susceptibility to allergic disorders.

Link between Tregs and microbiota
More than 25 years ago, Strachan was the rst to propose the so-called "hygiene hypothesis" [5]. He observed an increased frequency of atopic diseases such as hay fever in households of wealthier families. Strachan attributed this phenomenon to reduced infection rates during early childhood due to declining family sizes, improvements in household amenities, and higher standards of personal hygiene [5]. ese attributes may be seen as part of a more broad change in lifestyle during the last century that includes reduced exposure to livestock and helminth infections, higher rates of caesarean sections, and increased use of antibiotics. All of these factors have been proposed to contribute to the allergy epidemic particularly observed in "westernized" countries. Fundamental support of the hygiene hypo thesis comes from epidemiological data comparing rural to urban areas as well as children grown up with farm animal exposure vs. no such exposure. From such studies it has become clear that growing up in rural areas protects from allergic disorders possibly by a relative increase in bacterial or fungal diversity [6,7]. Nonetheless, the immunologic mechanism remains to be established. Very recently, it was shown that low-dose exposure to lipopolysaccharide (LPS), a cell-wall component of gram-negative bacteria, has protective e ects in a mouse model of house dust mite allergic lung in ammation [8]. is protective e ect required expression of a factor called A20 (encoded by Tnfaip3) by epithelial cells. Importantly, asthmatic patients showed reduced levels of A20 in epithelial cells which might render these patients more susceptible to allergic asthma due to failed LPS tolerance induction [8]. e proposed mechanism can be mimicked in murine models and might explain how early life exposure to farm dust can protect from the development of allergic disorders later in life.
Remarkably, nearly all of the lifestyle factors that have been included in the hygiene hypothesis can be expected to directly or indirectly in uence the composition of our microbiota. Especially the intestinal microbiota is known to heavily impact on the local and systemic immune system due to a long co-evolution of the host with symbiotic bacteria, the presence of numerous immune cells, and a large intestinal contact area to the microbial world. e most relevant observations for immune tolerance are derived from mice raised under germfree conditions because such mice show drastically reduced frequencies of Tregs in the gut [9,10] but not in the skin [11]. Additional murine studies identi ed individual bacterial strains that have a very strong impact on the immune system. For instance, a product of Bacteroides fragilis called Polysaccharide A (PSA) has been shown to be su cient for the maturation of the adaptive immune system and correcting the 1/ 2 balance in germfree mice [12], most likely by promoting the induction of Tregs and IL-10 secreting T cells [13]. Furthermore, a consortium of clostridia species has been shown to enhance the induction of Tregs in the intestinal lamina propria [9].
Mechanistically, the fermentation of complex carbohydrate bres by the microbiota leads to the pro-duction of short-chain fatty acids (SCFA) that have been shown to boost the generation of Tregs [14,15]. SCFA and most notably butyrate alter the acetylation pattern at certain histone residues of the Foxp3 promoter or so-called conserved noncoding sequence 3 (CNS3) next to promoter of the forkhead box protein 3 (Foxp3) [14,15] and may thus serve as an epigenetic rheostat for Treg function and thus tolerance [16]. Besides their direct e ect on the immune system, SCFA can also serve as a fuel for colonocytes and thereby enforce the epithelial barrier to prevent overt immune activation due to a too close contact of the immune system and the symbiotic microbiota [17]. A complex microbiota leads directly to mutualistic Treg induction when re-introduced into germfree mice [10].
Besides the promotion of Tregs, the microbiota also induces primarily pro-in ammatory T helper cell subsets such as 1 or 17 cells. A certain member of the microbiota, the so-called Segmented Filamentous Bacteria (SFB), by default provokes a drastic increase of 17 cells in the lamina propria of the ileum [18]. Similarly, colonization of the skin with Staphylococcus epidermidis results in restoration of normal numbers of dermal 17 cells compared to germfree mice [11]. e physiologic role of a dominantly type-17 response in the intestine supposedly lies in the induction of innate mechanisms protecting the epithelium from bacterial or fungal invasion as well as to allow rapid expulsion of those pathogens that do breach the epithelial barrier [19,20].
17 cells, however, also play a prominent role in many autoimmune and autoin ammatory diseases such as experimental autoimmune encephalomyelitis (EAE) as well as in severe forms of allergic asthma [21,22]. Indeed, in a number of autoimmune disease models the presence of SFB as part of the microbiota resulted in more severe pathology [23,24,25]. Whether 17 cells di erentiated in the intestine do also contribute to asthma exacerbation in the lung is not known. However, a recent report shows that 17 cells from SFB-positive mice contain a T cell receptor repertoire that is essentially speci c to SFB-derived peptides [26]. is nding does not support the possibility that 17 cells primed in the gut in response to SFB colonization exacerbate immune reactions to allergens elsewhere.

Microbiota and type 2 immune responses
How does the microbiota impact on allergic disorders and notably on the generation of 2 cells?
Currently, there is no evidence for a direct in uence of the microbiota on the di erentiation of 2 cells. However, a number of studies addressed this issue indirectly by analyzing 2-dominated immune responses in mice that had been treated with broad-spectrum antibiotics or raised under germfree conditions. First, in the lung, the complete absence of microbes leads to a more severe form of ovalbumin-induced allergic airway in ammation [27]. Similarly, the treatment with antibiotics early in life resulted in a profound increase of susceptibility to allergic airway in ammation [28]. Of note, the frequency of Tregs, whereas reduced in the gut, was not a ected in the lung [28]. Furthermore, in a model for allergic airway in ammation to house dust mite (HDM) more severe allergic immune reactions were observed in the lung of very young mice compared to older mice and this e ect correlated with incomplete microbial colonization of the lung at this age [29]. In this study, an altered phenotype of antigen-presenting cells and reduced frequencies of Tregs have been observed in the lung which may link the developing microbiota to the induction of Tregs and the perinatal acquirement of tolerance in the lung [29].
Besides the microbiota, reduced infection rates with helminth parasites in westernized countries have been proposed to contribute to higher frequencies of immune-mediated diseases [30]. Helminth parasites can induce de novo Treg di erentiation [31] but also shape the microbiota in a way to induce higher production of SCFA and thus Treg induction [14,15] which results in protection from allergic airway in ammation [32,33]. In an attempt to analyze the impact of the human microbiota on allergic susceptibility, Finlay and colleagues recently identi ed a cluster of individual bacterial strains in the feces of human newborns prior to disease onset that negatively correlated with the later development of atopy and wheeze -both hallmarks of high asthma risk [34]. Re-colonization of germfree mice with human microbiota supplemented with these bacteria resulted in lower allergic airway in ammation pathology in the following generation indicating a protective e ect of these bacterial strains [34].
Not only the microbiota does impact immunity during an ongoing immune response, it also shapes the immune status at steady state. To that e ect, IgE levels in germfree mice are in general severely increased, which is in sharp contrast to all other immunoglobulin subtypes [35]. Neonatal recolonization of these mice with a standard ora reversed this increase of IgE [35]. In terms of IgE-dependent e ector cells, the microbiota also limits basophil differentiation from hematopoietic precursors in the bone marrow as systemic basophil numbers are enhanced in germfree as compared to microbiota-sufcient mice [36]. e highest microbial exposure to the host takes place at the gut epithelium and thus, its e ects on type 2 immune responses can be expected to be strongest at this site. In a murine model of orally in-Allergo J Int 2016; 25: 114-23 duced food allergy antibiotic-treated as well as germfree mice showed more e cient sensitization to peanut extract that resulted in more severe anaphylactic reactions upon systemic allergen challenge [37]. Besides its role for the development of the immune system, the microbiota also a ects the mucus layer and regulates the permeability of the epithelial barrier. Indeed, more peanut allergens were found in the serum of sensitized germfree mice compared to conventionally housed mice [37]. Besides allergic in ammation, ulcerative colitis is another non-infectious disease which is (at least in part) based on a type 2 immunopathology [38]. Again, the induction of an ulcerative colitis-like disease in mice leads to an exaggerated immunopathology in mice raised under germfree conditions compared to conventionally housed mice [39]. Certain NKT cell subsets within mucosal surfaces may at least in part account for this e ect, even though additional players may be involved [39]. Finally, colonization of mice with murine or human symbiotic Clostridia consortia -which boost Treg numbers in the intestinal tract -led to reduced secretion of IL-4 from restimulated splenocytes and ameliorated the pathology in an allergic diarrhea model [9,40].
ese observations from murine studies can be seen in light of the observation that T cells isolated from the cord blood of (almost sterile) newborns show a bias towards the secretion of type 2 cytokines [41]. It has therefore been proposed that humans are born with a "type 2 immune bias", the reason of which remains to be established [42]. is bias is thought to be counteracted through the induction of 1, 17 or Tregs a er colonization with a symbiotic microbiota. Altogether, there is accumulating evidence that colonization with a complex microbiota plays a key role in dampening type 2 immune responses at various mucosal sites.

Regulatory T cells and control of type 2 immune responses
Tregs are known key players in the regulation of adaptive immune responses. e identi cation of the transcription factor Foxp3 was the starting point to a precise identi cation of these immune regulatory T cells and allowed to study their function in a variety of diseases: the ablation of the Foxp3 gene in mice leads to an autoimmune disorder known as the "scurfy" phenotype characterized by hyper-activation of T cells, production of auto-antibodies, wasting disease, and early death [43]. Similarly, dysfunction of Foxp3 and thus Treg function in humans is known as the IPEX syndrome (immunodysregulation polyendocrinopathy X-linked syndrome), a rare genetic disease that results in autoimmune enteropathy, different forms of dermatitis, and autoimmune endocrinopathy [44,45].
Developing thymocytes are negatively selected in the thymus by their reactivity against self-MHC-II complexes: whereas strong a nities are deleted (negative selection), intermediate responses can induce Treg di erentiation [46]. erefore, it is believed that such thymic Tregs (tTregs) preferentially recognize self-antigens and are crucial to actively prevent autoimmunity [47]. In contrast, the majority of Tregs isolated from the colon have a T cell receptor (TCR) speci city that recognizes microbiota-derived antigens and may therefore not be thymus-derived [48] even though some thymic Tregs may contribute to microbiota recognition [49]. Naive T cells in the periphery naturally encounter tissue-restricted self-antigens that are not expressed in the thymus. is can lead to their di erentiation into Tregs in the periphery [50]. Interestingly, these peripherally-induced Tregs -that are speci c for self-antigens -do express Helios (encoded by Ikzf2) [50], a transcription factor that has been proposed to be mainly expressed by thymic-derived Tregs [51]. Naive T cells can also encounter a variety of non-self antigens including potential allergens in the periphery. e highest probabilities for such contacts are at mucosal surfaces and associated lymphoid tissues especially within the intestinal tract. At this site, both foodand bacterial-derived antigens (or even viral or fungal antigens) create an enormous reservoir for potential antigens during the lifetime of an individual. It can be expected that naive T cells encountering such antigens become anergic or even get deleted and only few T cells will di erentiate to Tregs because there is only a restricted "niche" for Tregs at each anatomical site (Fig. 1).
Is there a general link of Tregs to allergic disorders? ere is accumulating evidence that the presence and function of Foxp3 + Tregs is directly linked to IgE levels and thus to susceptibility to allergic disorders.
Firstly, for both mice and humans it was shown that a constitutive de ciency in the expression of Foxp3 -and therefore also in functional Tregsresults spontaneously in very high levels of serum IgE [44,52]. e absence of functional antigen-speci c Tregs has also been associated with a dysregulated 2 response in ovalbumin-induced allergic airway in ammation even though other features of lung in ammation remained unaltered [53].
Secondly, the speci c knockout of several pathways or costimulatory molecules within Foxp3 + Tregs in mice has been associated with spontaneously increased IgE levels: the knockout of interferon regulatory factor 4 (IRF4) in Tregs results in enhanced generation of 2 cells and plasma cells that generate high levels of IgE [54].
Similarly, the knockout of the cytotoxic T lymphocyte antigen 4 (CLTA-4) in Tregs results in fatal autoimmunity characterized among other defects by heightened IgE levels and increased numbers of IL-4 producing 2 cells [55].
irdly, the speci c knockout of a conserved non-coding DNA sequence element next to the Foxp3 promoter called CNS1 leads to a complete failure in the generation of induced Tregs. is deciency results in an age-dependent mucosal inammation characterized by higher frequencies of 2 cells in the intestinal lamina propria and increased IgE and IgA levels [56].
Fourthly, a recent human study with children of high risk for peanut allergy revealed a clear protective e ect for an early in life contact to peanuts: when children ate a prede ned weekly amount of peanuts in the rst ve years, the children were protected from the later development of peanut-speci c allergy [57]. is study changed the previous recommendation for allergic parents to avoid early contact of their children with potential allergens whenever possible. In contrast, this study indicates an active induction of peripheral tolerance in the gut and one can speculate that induction of antigen-speci c Tregs may be one reason for this active tolerance induction. In fact, all kinds of potential immunologically not-ignored allergens may induce Tregs in the periphery in healthy individuals that will suppress the potential di erentiation of allergen-speci c 2 responses and IgE antibodies. It is possible that this Treg induction is more likely to occur early in life than in adulthood for reasons that remain to be determined.

Novel intestinal Treg subpopulations
e di erentiation of T helper cells is regulated through cytokine signals that result in the expression of key lineage-determing transcription factors. Surprisingly, Tregs can also co-express some of these transcription factors [58]. Others and we recently found that non-self speci c Tregs in the gut do not express Helios but surprisingly express the pro-in ammatory transcription factor retinoic acid-related orphan receptor gamma t [ROR(γt)]* [59,60]. ROR(γt) has been identi ed as key transcription factor of so-called innate lymphoid cells type 3 (ILC3) including lymphoid tissue inducer cells [61]. erefore, ROR(γt)-de cient mice are completely devoid of secondary lymphoid organs [62]. Subsequently, ROR(γt) was found to be a necessary transcription factor for the di erentiation of 17 cells [63]. We found that ROR(γt)-de cient mice show an increased pathology a er induction of a dextran sulfate sodium (DSS)-mediated colitis possibly through the generation of supernumerary numbers of tertiary lymphoid tissues in the colon and thus bacteria-speci c, pathological IgG antibodies [64]. However, Tregs expressing ROR(γt) can be found in the gut at high frequencies and may also contribute to dampen in ammatory diseases in the gut [65,66]. Indeed, ROR(γt) + Tregs were at least as e cient as their ROR(γt) − Treg counterparts in in vitro suppression assays and are therefore true regulatory T cells [65]. For a long time, ROR(γt)/Foxp3-co-expressing cells have been seen as intermediates between 17 and Tregs that are not yet fully committed to one or the other lineage [66].
Others and we have now shown that these ROR(γt) + Tregs represent a stable population of Tregs that exert a critical regulatory function in the gut [59,60,67]: rstly, the highest frequency of ROR(γt) + Tregs can be found in the intestinal lamina propria of mice a er weaning and their frequency seems to correlate with the bacterial load (and thus possibly with antigen exposure) because their frequency increases from esophagus to small intestine to colon.
Phenotypically, these cells have a very active phenotype including low expression of CD62L, high expression of CD44, high levels of the regulatory molecules CTLA-4, ICOS, CD39, and CD73. ey *The human counterpart of ROR(γt) is RORC2. To improve readability, only the abbreviation "ROR(γt)" is used throughout this article. do not express Helios and only low levels of Neuropilin-1. is phenotype is in accordance with an induced nature of Tregs in response to foreign non-self antigens. Importantly, the analysis of ROR(γt)-de cient mice revealed that the Treg compartment in the intestinal lamina propria is completely devoid of Helios − Tregs [60] and thus iTregs with a speci city for non-self antigens [60]. Additionally, the transfer of naive T cells followed by the oral application of the antigen leads to a preferential induction of ROR(γt) + Tregs in the intestinal lamina propria (Fig. 1). In line with this observation, the constitutive ablation of dendritic cells or MHC-II molecules prevents the induction of normal amounts of ROR(γt) + Tregs but not ROR(γt) -Tregs [60]. Constitutive ablation of dendritic cells has been previously associated with failed central tolerance induction and thus autoimmunity but not with failed Treg induction [68]. Importantly, ROR(γt) + -Tregs show a stable phenotype in vitro and in vivo and also have a fully demethylated TSDR (Treg-speci c demethylated region) of the Foxp3 promoter and other key regulatory molecules [67]. erefore, one can expect that ROR(γt) + Tregs are rather stable and execute an important regulatory function in the gut at steady state.
To assess the direct impact of the microbiota on ROR(γt) + Tregs, broad-spectrum antibiotic treated mice or mice raised under germfree conditions have been analyzed. Indeed, the frequency of ROR(γt) + -Tregs was drastically reduced in both cases and the transfer of germfree mice to a microbial-containing environment rapidly re-induced normal ROR(γt) + Treg numbers [60]. e group of Benoist C. analyzed in detail which bacterial strains are the best inducers of ROR(γt) + Tregs and found indeed a selective capacity of individual strains to perform better than others [59]. However, there was no general pattern detectable among the inducers and the non-inducers until now. is nding may indicate that rather than bacterial family relativeness shared metabolic pathways play a dominant role in terms of ROR(γt) + Treg induction. It should be noted that colonization with a clostridia consortium -that has been previously associated to Treg induction -preferentially but not exclusively induces ROR(γt) + Tregs [59,60]. Similarly, we found a biased induction of ROR(γt) + Tregs a er treatment with the SCFA butyrate but Se k et al. did not. Notably, also SFB were able to induce some ROR(γt) + Tregs albeit to a lower extent than 17 cells [59,60].
Besides the microbiota, also diet can in uence the intestinal immune system. e vitamin A metabolite retinoic acid has been shown to enhance the induction of Tregs in vitro [65,69]. Importantly, the absence of retinoic acid can alter the balance between ILC2 and ILC3 and thereby cause a type 2 immune deviation within the mucosal immune system [70]. We found that both blocking of the receptor for vitamin A with an inhibitor or feeding mice a vitamin A-de cient diet results in reduced frequencies of ROR(γt) + Tregs [60]. Whether retinoic acid-mediated immune deviation from type 2 to type 3 immunity in both adaptive and innate immunity is functionally linked remains an open question.
Which cell-intrinsic factors do regulate the induction of ROR(γt) + Tregs? e shared expression of ROR(γt) between 17 cells and iTregs in the gut indicates a shared transcriptional program for the di erentiation of both subtypes. Indeed, the knockout of IL-6 and its signal transducer STAT3 (both of which critical factors for the di erentiation of 17 cells) results in reduced frequencies of ROR(γt) + Tregs. For the e ect of IL-23, a cytokine involved in 17 cell di erentiation, the Benoist group and we obtained divergent results: in the absence of IL-23, we observed a reduced frequency of ROR(γt) + Tregs whereas the group of Benoist did not [59,60]. e role of IL-23 for 17 is still incompletely understood but it seems that IL-23 is only required for stabilization rather than initiation of the 17 program [21]. e contrasting results of IL-23 for the induction or maintenance of ROR(γt) + Tregs may therefore re ect the local microbiota composition and its relative stability in di erent animal facilities.
It should be noted that during the resolution of an in ammation, ROR(γt) + 17 can also trans-differentiate into Foxp3 − regulatory T cells expressing the anti-in ammatory cytokine IL-10 [71]. Only few ROR(γt) + Foxp3 + Tregs are directly generated from ex 17 cells [67,71]. is may re ect the circumstance that a massive expansion of 17 cells during in ammation occurs and only a limited niche for Tregs exists to enable tolerance to symbiotic microbes. Nevertheless, it seems that ROR(γt) + Tregs co-opt the transcriptional machinery of the 17 di erentiation program and the local micromilieu dictates the outcome between 17 or ROR(γt) + Tregs.
Given that ROR(γt) + Tregs are highly abundant in the intestinal lamina propria, the functional role of these cells for the intestinal immune homeostasis remained an open question. To this end, we rst analyzed ROR(γt)-de cient mice at steady state and found sometimes drastically enhanced frequencies of Gata3 + 2 cells and increased levels of serum IgE, one of the hallmarks of allergic predisposition [60].
In an attempt to speci cally knockout ROR(γt) in Tregs, the group of Benoist and we crossed mice expressing the Cre recombinase under control of the Foxp3 promoter with mice bearing oxed ROR(γt) alleles. We found that these Foxp3 Cre × ROR(γt) FL/FL mice showed slightly reduced frequencies of Helios − Foxp3 + iTregs in the small intestinal lamina propria. In parallel to the genetically induced reduction of ROR(γt) + Tregs, we found a relative increase of Gata3 + Tregs and a slight increase of Gata3 + 2 cells and serum IgE levels. ese results indicate that ROR(γt) + Tregs may regulate the induction of intestinal 2 cells and may therefore enhance the systemic susceptibility to allergic disorders even at steady state.
In order to boost a 2-driven in ammatory disorder, we treated mice with oxazolone to induce a type 2-dominated in ammation in the colon resembling human ulcerative colitis [38]. e selective knockout of ROR(γt) in Tregs resulted in a drastically enhanced pathology, enhanced secretion of type 2 cytokines of restimulated T cells, and sometimes death [60]. In contrast, the group of Benoist observed an exaggerated 1-and 17-driven immune pathology in a di erent colitis model using the same conditional knockout mice [59]. e combination of both studies may re ect the in uence of the individual microbiome on the type of immune deviation in the absence of ROR(γt) + Treg-mediated immune regulation.
Our results point towards a ROR(γt) + Treg-mediated control of 2-driven immune responses. In contrast, it is well known that helminth infections are cleared by the immune system through the induction of sometimes very strong type 2 immune responses. We reasoned therefore that in such a scenario the absence of ROR(γt) + Tregs could be benecial for the host. Indeed, we found a lower egg burden a er infection with the helminth parasite Heligmosomoides polygyrus in Foxp3 Cre × ROR(γt) FL/FL mice that was paralleled by increased secretion of type 2 cytokines [60].
But how can ROR(γt) + Treg exert this speci c function? In general, it is not really known by which of their multiple regulatory mechanism Tregs exert their regulatory function in a de ned immune response. Nevertheless, we found that ROR(γt) + Tregs may alter the phenotype of intestinal dendritic cells because Foxp3 Cre x ROR(γt) FL/FL mice showed lower levels of the costimulatory receptors CD80 and CD86 [60] similar to what has been observed in CT-LA-4-de cient Tregs [55]. ROR(γt) + Tregs do express high levels of CTLA-4 and IRF4 at steady state which is in line with the 2 bias observed in mice with a selective defect of these molecules in Foxp3 + Tregs [54,55]. Altogether, these results point towards a fundamental role of ROR(γt)+ Tregs in the control of 2-dominated immune responses under certain circumstances (Fig. 2). As the di erentiation of 17 cells and ROR(γt) + Tregs seem to share at least some transcriptional programs, one can ask whether this program is generally used to counteract 2 immune responses. Indeed, one study recently highlighted a possible 2 and 17 counter regulation in the case of allergic asthma [72]. is does not exclude that mixed 2/ 17 phenotypes of allergic disorders exist because genetic factors may unleash this reciprocal regulation and result in even more severe forms of allergic asthma [73,74].
Besides ROR(γt) + Tregs, another population of Tregs expressing the transcription factor Gata3 has been observed in the intestinal tract [75,76]. Both ROR(γt) + and Gata3 + Treg subsets are mutually exclusive under healthy conditions [60,75]. Powrie and colleagues recently provided evidence that Gata3 expression in Tregs is induced upon stimulation with the epithelial-derived cytokine IL-33 [77]. Furthermore, Gata3 expression in Tregs is necessary for prevention of a spontaneous in ammatory disorder and e ective suppression of a transfer colitis [76,77]. However, ROR(γt) + Tregs have been shown to protect more e ciently than ROR(γt) − Tregs in this transfer colitis model [67] raising the question of whether both subsets use di erent mechanisms to suppress T cell proliferation in a lymphopenic environment. Two observations make it unlikely that Gata3 + Tregs are induced upon recognition of foreign antigens/allergens: rstly, Gata3 + Tregs do at least to a certain degree express Helios (and may thus be rather of thymic origin and/or recognize self antigens) and secondly, Gata3 + Tregs can still be found in germfree mice [60] (Fig. 1). In germfree mice, a di erent cytokine pattern of IL-33 and IL-6 among others may shape the distribution of Treg subsets in the gut [60,77]. In a model of food allergy, Gata3 + Tregs emerge as a consequence of enhanced IL-4 receptor signaling making such Gata3 + Tregs less e cient in protection from and possibly even a driver of food allergy due to high IL-4 expression [78]. Conversely, Gata3-de cient Tregs have a higher chance to adopt a 17-like cyto kine expression pro le [76]. ese observations may have important consequences for immunotherapy of allergic patients because the presence of 2like Tregs may prevent therapy or even aggravate immune pathology in established food allergy and possibly explains limited success of oral immunotherapy so far [79].

Conclusion
e hygiene hypothesis postulates a change of life style factors in westernized countries as one reason for an increase of allergic disorders. Of note, most of these life style factors can in uence directly or indirectly the composition of our microbiota. is "altered" microbiota may -for reasons that remain to be discovered -favor the di erentiation of intestinal 2 cells in the absence of Treg-mediated im-mune regulation (Fig. 2). Ine cient mucosal Treg induction or maintenance may be the result of a genetic predisposition or itself the result of an altered microbiota. According to this hypothesis, a combination of both factors results in an enhanced susceptibility to allergic disorders or other in ammatory diseases.
Several questions remain open: is Treg-mediated control antigen-speci c or rather the result of a general setting of the immune status? If the gut-derived immune regulation is antigen-speci c, are allergic disorders the result of cross-reactivity with microbiota-derived antigens? Or are most allergens normally present in the intestinal tract and induce antigen-speci c tolerance at steady state conditions? erefore, it remains to be determined whether the nature and speci cities of intestinal Treg subpopulations can explain the observed increases in allergic disorders in westernized countries. Most of the studies including our own do not directly address the role of Tregs for allergic onset or allergic suscep-tibility. It will be important to analyze how the absence or an imbalance between Treg subpopulation re ect an alteration of allergic susceptibilities and whether this knowledge can be used to design novel prevention strategies.  Possible mechanism for the relation between microbiota, ROR(γt) + Tregs, and prevention of pro-allergic Th2 immune responses. Unwanted Th2 di erentiation in response to harmless foreign antigens such as allergens is suppressed through the induction of ROR(γt) + Tregs under healthy conditions. Th2 di erentiation may be favored through an "altered" microbiota at mucosal sites.