Induced and Natural Regulatory T Cells in the Development of Inflammatory Bowel Disease

Abstract

The mucosal immune system mediates contact between the host, and the trillions of microbes that symbiotically colonize the gastrointestinal tract. Failure to tolerate the antigens within this “extended self” can result in inflammatory bowel disease (IBD). Within the adaptive immune system, the most significant cells modulating this interaction are Foxp3⁺ regulatory T (T_reg) cells. T_reg cells can be divided into two primary subsets: “natural” T_reg (nT_reg) cells, and “adaptive” or “induced” T_reg (iT_reg). Recent research suggests that these subsets serve to play both independent and synergistic roles in mucosal tolerance. Studies from both mouse models and human patients suggest defects in T_reg cells can play distinct causative roles in IBD. Numerous genetic, microbial, nutritional, and environmental factors that associate with IBD may also affect T_reg cells. In this review we summarize the development and function of T_reg cells, and how their regulatory mechanisms may fail, leading to a loss of mucosal tolerance. We discuss both animal models and studies of IBD patients suggesting T_reg cell involvement in IBD, and consider how T_reg cells may be used in future therapies.

Introduction

The gastrointestinal tract is the largest lymphoid organ in the body. This complex network consists of specialized epithelial cells, lymphatic vessels, dendritic cells, lymphocytes, and secondary lymphoid structures. The gut immune system mediates contact between the host and the microbes that symbiotically colonize this mucosal surface (termed the microbiota), while protecting the host against pathological invasion. Exposure to nutritional, microbial, and host antigens in the intestine creates an “extended self” that varies dynamically in this complex environment. Thus, the division between “self” and “non-self” is often indistinct, and multiple regulatory mechanisms have evolved to restrain misdirected immune responses. Failure of these regulatory pathways and/or imbalances in the composition of the microbiota can result in pathologic inflammatory processes such as inflammatory bowel disease (IBD), which primarily consists of ulcerative colitis (UC) and Crohn’s disease (CD).

Numerous tissues and specialized cells are involved in maintaining mucosal tolerance in the intestine. Paneth cells at the base of the intestinal crypts of the small intestine produce antimicrobial peptides and lectins, limiting direct contact between the epithelia and microbiota.¹ M cells of the follicle-associated epithelium translocate luminal antigens across the epithelium into the dome region of Peyer’s patches (PP) and isolated lymphoid follicles (ILF) for uptake and processing by dendritic cells (DCs).² Recent data suggests that a subset of goblet cells in the small intestine may also deliver antigens to DCs.³ The DCs within the gut lamina propria and secondary lymphoid tissues encounter and process the antigen. Alternatively, DC within the lamina propria can sample luminal antigens directly by extending dendrites between tight junctions of the epithelia, into the lumen. After encountering antigen, DCs migrate to T cell zones in the PP or mesenteric lymph node (mLN) to present antigen and activate effector T cells. Activated CD4 effector T cells can stimulate B cells, resulting in IgA-producing plasma cells. IgA plasma cells and activated effector T cells then populate the lamina propria, resulting in a low-grade inflammation that appears to be the normal state of the intestine.²

IBD appears to be a primarily T cell-mediated disease. Indeed, CD4 T cells are responsible for orchestrating the excessive inflammation involved in IBD.^{4, 5} However, T cells also appear to play the largest role in establishing and maintaining immunological tolerance by suppressing excessive or abnormal immune responses. Numerous subtypes of T cells appear to display immunosuppressive function in the gastrointestinal tract. Classical TCRαβ T cells such as CD8⁺ CTLs and CD4⁺ Tr1 cells exhibit immune suppression in the gut.^{6, 7} In addition, more atypical CD8αα and TCRγδ⁺ T cells have been shown to exhibit protection against mouse models of colitis.^{6, 7} Other innate-like T lymphocytes with invariant TCRs such as iNKT and mucosal-associated invariant T (MAIT) cells may also play immunoregulatory roles in the GI tract. Nonetheless, the most significant of the immunosuppressive T lymphocytes are TCRαβ CD4 ⁺ regulatory T (T_reg) cells that are characterized by the expression of the transcription factor Foxp3.⁸ This review will focus on these CD4⁺ Foxp3⁺ T_reg cells.

Global failure of T_reg cell development via Foxp3 deficiency results in the lethal multi-system autoimmune disease immunodysregulation polyendocrinopathy enteropathy X-linked syndrome (IPEX) that underscores the essential nature of T_reg cells in immune regulation. The most commonly affected organ in IPEX is the intestine, highlighting the essential role for T_reg cells in the gastrointestinal tract. In addition to IPEX, other genetic deficiencies suggest the importance of T_reg in IBD. Mutations in WASP, CD25, and IL-10 all lead to abnormal T_reg cell numbers and/or function, and also increase an individual’s risk for IBD.⁹ Although these genetic defects suggest a role for T_reg cells in human IBD, much of the mechanistic evidence has come from mouse models of disease. Mouse models of IBD can be separated into 4 relatively distinct categories: 1) chemically induced barrier disruption models, 2) spontaneous models due to genetic deficiency, 3) overexpression of inflammatory mediators, and 4) lymphopenic T cell transfer models (Table 1).

Table 1.

T_reg cell involvement in commonly used mouse models of IBD

Category of IBD Model	Model Name	Role for Regulatory T cells	References
1) Barrier disruption	DSS (chemical-induced)	- Absence of Treg or TGF-β1 signaling in Treg cells increases severity	20,217
	TNBS/DNBS/Oxazolone (hapten-induced)	- Ethanol in absence of TNBS leads to resistance	218
		- B. fragilis PSA increases iTreg cells leading to resistance	146
		- Treg cells transiently decrease then home to intestine during remission	219
2) Genetic deficiency	IL-10 deficiency	- Treg-produced IL-10 is necessary to prevent intestinal inflammation	75, 76, 77
	SAMP/YitFc (PPARG mutation)	- Treg cells from these mice are defective in suppressing colitis upon transfer	16, 220
	Mdr1 deficiency	- mutation leads to decreased in vivo and in vitro differentiation of iTreg	17
3) Overexpression of inflammatory mediators	TNFΔARE (TNF-α overexpression)	- increased differentiation of iTreg attenuates disease	221
	STAT4 overexpression	- unknown, Treg cells hypothesized to be normal	21
4) Lymphopenic T cell transfer	Naïve Tconv transfer into SCID/ RAG−/− mice	- co-transfer of nTreg cells prevents disease	19
		- nTreg + in vivo or in vitro iT reg cells treat disease	38

Commonly used chemically induced barrier disruption models include dextran sulfate sodium (DSS) and hapten-induced colitis via 2,4,6-trinitrobenzene sulfonic acid (TNBS) and oxazolone. These chemicals disrupt the epithelium, allowing microbial and immunogenic molecules to penetrate and initiate an inflammatory response.^{10, 11} Spontaneous genetic models include IL-10 deficient mice,¹² defective PPARG expression in SAMP1/YitFc mice,¹³ and Mdr1 deficiency.¹⁴ Numerous other genetic deficiencies also lead to intestinal inflammation, but IL10, PPARG, and MDR1 are particularly notable in that they play roles in T_reg cell number and/or function and are also human IBD susceptibility loci.^{11, 15–17} Other mouse models utilize gene overexpression to model intestinal inflammation. Such models include TNF-α overexpression in TNF^ΔARE mice and STAT4 overexpression.¹¹

Perhaps the most compelling data on the role of T_reg cells in IBD has come from the T cell transfer model of colitis. In this model, naïve CD4⁺ T cells, depleted of T_reg cells, (typically CD45RB^hi) are adoptively transferred into mice lacking B and T lymphocytes (SCID or RAG-deficiency). These effector T cells proliferate and become activated in response to bacterial antigens in the intestine, resulting in inflammation and colitis. Disease can be both prevented and treated via elimination of the microbiota or co-transfer of T_reg cells.^{18, 19} This model offers the distinct advantage that the pathogenic and regulatory T cell subsets can be genetically targeted independently. Thus, several important inflammatory and suppressive mechanisms have been identified through transfer of T_reg or effector T cells (T_eff ) from mutant mice.⁶

Numerous studies utilizing mouse models support a role for T_reg cells in IBD. In barrier models T_reg cells localize to the intestine and mLN in acute disease, and an absence of T_reg cells exacerbates disease.^{6, 20} A mild breech of the intestinal barrier via ethanol in the absence of TNBS does not result in colitis, suggesting regulatory responses predominate.⁶ Many genetic models of spontaneous IBD involve genes affecting T_reg function, and in treatment models T_reg cells home to the intestine to resolve inflammation.^{6, 21} Two subsets of T_reg cells have been described, “natural” T_reg (nT_reg) cells and “induced” or “adaptive” T_reg (iT_reg) cells. Both nT_reg and iT_reg subsets are characterized by the expression of Foxp3, and Foxp3 expression is necessary for their overall fate and function.^{8, 22} The nT_reg and iT_reg subsets are largely distinguished by their developmental origin and appear to play non-redundant roles enforcing gastrointestinal tolerance.

Natural regulatory T cells

“Natural” T_reg (nT_reg) cells arise as a discrete and largely stable lineage originating in the thymus. Foxp3⁺ cells are first detectable in a small fraction of CD4⁺ CD8⁺ double positive thymocytes, and are subsequently more frequent in CD4⁺ single positive thymocytes.⁸ The nT_reg subset exhibits a TCR repertoire that is distinct from Foxp3⁻ conventional T cells (T_conv) and from iT_reg cells. Data shows that the TCRs of nT_reg cells may have increased affinities for self-peptides.^23–26 Moreover, mutations in MHC and TCR signaling suggest that a strong TCR signal is required for nT_reg development. TCR signaling activates the NF-κB pathway, and several conditional mutations in NF-κB members show nT_reg defects.²⁷ NF-κB family member c-Rel binds directly to the conserved non-coding sequence 3 (CNS3) region of the Foxp3 promoter, providing a link between TCR signaling and Foxp3 expression.²⁸

TCR-ligand affinity alone does not determine Foxp3 expression.²⁸ nT_reg cells also require IL-2 for their development and maintenance. In fact, regulatory T cells were first identified by their elevated expression of the high-affinity IL-2 receptor CD25 (IL-2Rα). Mice lacking IL-2 signaling via antibody neutralization or genetic deficiency of IL-2 or IL-2 receptors show nT_reg deficiencies and spontaneous autoimmune disease including IBD.^29–32 CD25 deficiency in human patients also results in an IPEX-like disease, supporting the importance of IL-2 signaling in nT_reg development.³³ Downstream signaling from IL-2 receptors is mediated through STAT5, and STAT5 binds to the conserved non-coding sequence 2 (CNS2) enhancer region of the Foxp3 gene. This suggests that IL-2 signaling directly promotes the initial expression and/or maintenance of Foxp3 to support nT_reg development.²⁷

In addition to IL-2 signaling, normal T_reg development in the thymus requires autophagy. Transplant of autophagy-deficient thymi from Atg^−/− mice results in abnormal T cell selection and spontaneous colitis.³⁴ This suggests that a subset of intestinal auto-antigens may require autophagy for proper presentation and deletion in the thymus. In fact, autophagy genes are a commonly identified class of IBD susceptibility alleles. Thus, there may be a role for autophagy defects in clinical disease, though T_reg specificity and function in IBD patients with autophagy mutations has not been investigated.

Adaptive or induced regulatory T cells

Whereas nT_reg cells are defined by Foxp3 expression that originates during thymic T cell development, “adaptive” or “induced” T_reg (iT_reg) cells are generated during the course of an immune response from naive T_conv cells in the periphery. In healthy mice, the majority of peripheral Foxp3⁺ cells in the spleen and lymph nodes are thymically-established nT_reg cells. In contrast, the iT_reg subset of T_reg cells make up a large fraction of T cells in the lamina propria (LP) and gut-associated lymphoid tissue (GALT) of the intestine.^{34, 35} The formation of iT_reg cells involves the peripheral activation of T_conv cells in the presence of TGF-β1 (Figure 1). This conversion can also be performed in culture, resulting in Foxp3⁺ in vitro iT_reg cells.³⁶ These in vitro iT_reg cells may serve as cell-based therapies for autoimmune diseases such as IBD since they can be derived in large numbers from the patient’s own cells, and can potentially be enriched for relevant antigen specificities. Indeed, clinical trials utilizing in vitro-derived iT_reg cells as a new therapy are currently under way.³⁷

Figure 1.

Model of T_reg cell development. The nT_reg (green) and T_conv (grey) cell populations develop as separate lineages in the thymus, based in part on differences in the threshold for affinity-based selection. The result is a peripheral population of nT_reg and T_conv cells with distinct TCR (orange) repertoires. In the gut and at other environmental interfaces, TGF-β1 induces T_conv cells to become iT_reg cells (red) or Th17 cells (purple). The peripheral T_reg cell pool is therefore comprised of both iT_reg and nT_reg cells that share suppressive mechanisms such as IL-10 production. Increased levels of IL-6 in the local environment promote production of Th17 cells while blocking iT_reg cell formation. In general, iT_reg cells are unstable and may lose Foxp3 expression (ex-iT_reg cells), although certain factors may increase iT_reg stability. The ex-iT_reg cells (pink) are available to become Th1 (blue) or Th17 cells, or to reacquire Foxp3 expression and cycle back into the iT_reg pool. Importantly, nT_reg and iT_reg cells act synergistically and are non-redundant, a feature based largely on their different TCR repertoires.

In vivo, antigen exposure and/or lymphopenia can result in Foxp3 expression in up to 15% of antigen experienced T_conv cells.³⁸ These in vivo iT_reg cells appear to be particularly prevalent at mucosal surfaces where tolerance must be expanded to antigens within the colonizing microbiota.^{27, 39} In addition to the microbiota, ingested food contains a plethora of antigens. Immune responses to these dietary molecules can be actively suppressed by inducing oral tolerance. Antigen feeding experiments identified a TGF-β1-producing CD4⁺ T cell population in the LP and GALT coined “Th3” that was important in oral tolerance.⁴⁰ These studies were performed prior to the discovery of Foxp3, and subsequent studies have shown that most “Th3” cells express Foxp3 and are de facto iT_reg cells. However, a small fraction of CD4⁺ TGF-β1⁺ Foxp3⁻ Th3 cells may yet exist.⁴¹

Similarly to in vitro iT_reg, the peripheral induction of Foxp3 in vivo appears to be dependent upon TCR activation in the presence of TGF-β1.^{36, 42, 43} TGF-βR signaling leads to phosphorylation of SMAD molecules and the binding of SMADs to the CNS1 enhancer element in the Foxp3 locus.⁴⁴ This association together with the binding of other transcription factors leads to the TGF-β1-mediated activation of Foxp3 transcription, which drives differentiation of the iT_reg lineage.^{28, 45} The vitamin A metabolite retinoic acid (RA) further increases Foxp3 induction.⁴⁶

In vivo induction of Foxp3 seems to be particularly important in the gut. In fact, the GI tract is a TGF-β1 rich environment, LP and GALT dendritic cells (DCs) produce RA, and RA drives gut tropism in T_reg cells.^{6, 47} Additionally, APCs derived from the intestine, including CD103⁺ DCs and CD11b^hi CD11c⁻ macrophages preferentially promote iT_reg development in comparison to spleen or peripheral lymph node derived APCs.^{6, 27, 39} These data strongly support a role for induction of iT_reg and iT_reg-mediated suppression in the gastrointestinal environment.

Distinguishing nT_reg and iT_reg cells

There are few mechanisms for differentiating nT_reg and iT_reg cells. Most molecules showing increased expression on T_reg cells (including CD25, CTLA4, and GITR) appear to be targets of Foxp3 transcriptional activation, and are thus expressed similarly between iT_reg and nT_reg cells. Also, both nT_reg and iT_reg cells suppress immune responses using similar antigen-dependent and antigen-independent mechanisms.² Other cell types can also express ostensible T_reg markers, further confusing the discrimination between T_reg subsets.^{48, 49}

Helios is one molecule that has been suggested to differentiate nT_reg and iT_reg cells. Helios is a transcription factor that is expressed at elevated levels in nT_reg but not iT_reg cells.⁵⁰ However, genetic deficiency of Helios does not appear to affect T_reg differentiation and function.⁵¹ Subsequent studies have suggested that Helios is a marker of activation and proliferation that can be expressed on both T_reg and T_conv cells.⁵² Furthermore, in some studies only a small fraction of in vivo-derived iT_reg cells lacked Helios expression²⁴ and both in vitro and in vivo-derived iT_reg could express Helios in response to APC stimuli.⁵³

Neuropilin-1 (Nrp-1) is another marker that may differentiate nT_reg and iT_reg cells. Recent studies suggest that Nrp-1 is limitied to the nT_reg subset.^{54, 55} The iT_reg cell subset expresses only low levels of Nrp-1 after in vitro conversion via TGF-β1. In vivo iT_reg cells differentiated via oral or intravenous antigen delivery, exposure to a peripherally-expressed antigen, or lymphopenia-induced proliferation also lack Nrp-1 expression. Interestingly, Nrp-1 expression differences are largely maintained, even when cells are activated.^{54, 55} Conversely, iT_reg cells differentiated under inflammatory conditions in the lung or CNS appear to gain Nrp-1 expression.⁵⁴ Thus, it will be interesting to examine Nrp-1 expression in mouse models of IBD. Unfortunately, Nrp-1 does not appear to be expressed by human PBMCs, so the utility of this marker for clinical studies may be limited.⁵⁶

Differences in the methylation status of the Foxp3 genetic locus are a more universally accepted difference between the nT_reg and iT_reg subsets. Thymus-derived nT_reg cells have demethylated CpG motifs in the T_reg cell-specific demethylation region (TSDR) of conserved noncoding sequence 2 (CNS2) in Foxp3, whereas the TSDR of iT_reg cells is only partially demethylated.⁵⁷ Heritable and stable Foxp3 expression are linked to TSDR demethylation. The factors that influence the demethylation status of the TSDR in iT_reg cells are unknown and the level of demethylation varies depending on the nature of the iT_reg differentiation.^{24, 58} Unfortunately, detection of Foxp3 TSDR methylation must be performed on the genomic DNA of large pools of cells, and thus cannot differentiate nT_reg and iT_reg cells on a per cell basis as with surface markers.

Since markers that differentiate nT_reg and iT_reg are controversial, current studies often identify all peripheral Foxp3⁺ T cells as “nT_reg”. This practice complicates the precise determination of nT_reg and iT_reg cell identity and function, because the peripheral Foxp3⁺ T cell pool undoubtedly contains some fraction of T_reg cells that induced Foxp3 after leaving the thymus. Thus, the population of cells identified as “nT_reg cells” are likely contaminated with in vivo-derived iT_reg cells. The fraction of iT_reg in a typical peripheral T_reg pool has been estimated by tracking Foxp3⁻ T_conv cells that upregulate Foxp3 expression in vivo. In these approaches, congenically marked Foxp3⁻ T_conv cells are transferred into lymphopenic, lymphoreplete, or Foxp3-deficient hosts. Using these transfer approaches, the overall fraction of iT_reg cells in the peripheral T_reg pool appears to be 4–15%, depending on the system used. As noted above, higher frequencies of iT_reg cells are generally present in the mLN and GALT than the peripheral lymph nodes and spleen, supporting a role for iT_reg cells at mucosal surfaces.^{24, 38, 59}

The size of the iT_reg pool was also estimated in Carma-1 deficient mice, which lack thymically-derived nT_reg cells but develop iT_reg cells in the periphery.³⁴ The overall frequency of Foxp3⁺ cells in the periphery of Carma-1^−/− is 3–4% of that in wild-type mice. Since Carma-1^−/− mice only have iT_reg cells yet have a higher threshold for TCR activation, 3–4% likely represents the low end of the iT_reg contribution to the peripheral T_reg pool, in agreement with transfer studies. Additionally, compared to the spleen, Foxp3⁺ cells in these mice are twice as frequent in the mLN and ten times as frequent in the LP, further supporting an increased frequency and role for iT_reg development at these sites.³⁴ Thus, when considering most studies, it appears that the typical contribution of the iT_reg subset to the peripheral T_reg pool ranges from 3 to 15% and is highly context-dependent. Consistently, the contribution of iT_reg cells increases greatly in the intestine, favoring a significant role for iT_reg cells in the control of IBD.

Due to the lack of a specific surface marker, it is currently difficult to identify human in vivo-derived iT_reg cells, although it is reasonable to assume that the iT_reg cell subset exists. Human T cells share a capacity for the in vitro differentiation of iT_reg cells and have similarly increased numbers and frequency of T_reg cells in mucosal locations. One important distiction between mice and humans is that activated non-regulatory human T_conv cells transiently express FOXP3 at low levels. Accordingly, human FOXP3⁺ T cell populations are subdivided as a resting/naive T_reg cell subset that is CD25^hiCD45RA⁺FOXP3^hi, an effector T_reg cell subset that is CD25^hiCD45RA⁻FOXP3^hi, and a non-regulatory cytokine-secreting FOXP3⁺ subset that is CD25^hiCD45RA⁻FOXP3^lo.⁶⁰ Additional markers such as low expression of CD127, and increased expression of CD39, MHCII, GITR, and CTLA-4 further distinguish human T_reg cells.⁶¹

Mechanisms of T_reg suppression in IBD

T_reg cells utilize numerous effector mechanisms to suppress immune responses, targeting both the innate and adaptive arms of the immune system. Interestingly, no one effector mechanism or pathway has been shown to be both necessary and sufficient for all T_reg-mediated suppressive activities. There is a great deal of functional redundancy between T_reg suppressive pathways, which results in compensatory mechanisms in targeted T_reg cells and the need for a very strong inflammatory stimulus to differentiate phenotypes. Also, early deletion of genes involved in T_reg function often leads to systemic immune issues before intestinal involvement can be detected.⁶ Studies of IBD are particularly sensitive to slight disparities in the colonizing microbiota between mouse colonies and differing and mouse strains, leading to inconsistent results. Despite these difficulties, several effector mechanisms for T_reg-mediated suppression of IBD have been described (Table 2). In general they can be divided into two categories 1) immunoregulation through consumption or production of soluble mediators (primarily cytokines) and 2) immunoregulation via direct interaction with APCs or T_eff cells.

Table 2.

T_reg cell suppressive mechanisms in IBD

	Regulatory Mechanism	Utilized by		Role in IBD		References
		nTreg	iTreg	Mouse Models	Human Patients
Soluble Mediators	TGF-β1	Yes	Yes	Yes	Yes	42, 222
	IL-10	Yes	Yes	Yes	Yes	12, 73–75,117
	IL-35	Yes	No	Yes	Unknown	79, 223
	Pericellular adenosine (CD39/CD73)	Yes	Unknown	Yes	Unknown	82–85
	Sink for IL-2 (CD25hi)	Yes	Yes	Yes	Yes	22, 33, 36, 83, 87
Cell-Cell Interaction	Cytotoxicity (perforin/granzymes)	Yes	Yes	Unknown	Unknown	89, 90
	CTLA-4	Yes	Yes	Yes	Yes	95, 98, 224, 225
	LAG-3	Yes	Unknown	Yes	Unknown	99, 226, 227
	PD-1/PD-L1	Yes	Yes	Yes	Unknown	100
	Nrp-1	Yes	No	Unknown	Unknown	54, 55
	Galectins	Yes	Unknown	Yes	Yes	228–230
	GITR	Yes	Yes	Yes	Unknown	231, 241
	OX40	Yes	Yes	Yes	Unknown	232, 233

Among the cytokine-based approaches, TGF-β1 and IL-10 are the best studied. TGF-β1 is a pleiotropic cytokine produced by many immune and non-immune cells. TGF-β1 controls cell proliferation and differentiation in numerous tissues throughout the body. In particular, TGF-β1 plays an important role in the extra-thymic induction of Foxp3 in iT_reg cells. In addition, T_reg cells themselves produce large amounts of TGF-β1. Global deficiency of TGF-β1 results in a lethal multi-organ lymphoproliferative disease similar to Foxp3 deficiency. Since TGF-β1 deficient mice die at weaning, analysis of gut-specific effects of TGF-β1 difficult.

The intestine appears to be a particularly TGF-β1 -rich environment and apoptotic cells stimulate TGF-β1 release.⁶² The high cell turnover and/or acute epithelial damage in the intestine may contribute to the secretion of TGF-β1 to prevent excessive immune responses in the chronically immune-stimulated intestinal environment. However, much of the TGF-β1 in the intestine is present in an inactive form that must be activated by integrins. Mice deficient in α_v or β₈ TGF-β1-activating integrins in APC populations develop spontaneous colitis, suggesting DCs and perhaps other APCs must activate TGF-β1 in the GALT.^{6, 34} This activated TGF-β1 likely has direct effects on T lymphocytes since depletion of T cells prolongs the survival of TGF-β1 deficient mice.^63–65 Further, T cell-specific deletion of TGF-β-receptor II results in a similar autoimmune phenotype to global deficiency of TGF-β1.^{66, 67} In the T cell transfer model of colitis, a TGF-β1 blocking antibody eliminates the ability of a T_reg-containing CD45RB^low pool to prevent colitis.⁶⁸ In a similar transfer model, transferred T cells expressing an endogenous inhibitor of TGF-β1, Smad7, were unable to be suppressed by co-transferred T_reg.⁴² This suggests that T_reg-produced TGF-β1 has direct immunosuppressive effects on colitogenic T_eff cells, and/or that TGF-β1-driven differentiation of iT_reg is necessary for prevention of intestinal inflammation.

IL-10 is a potent anti-inflammatory cytokine produced by both non-immune and immune cells including T_reg cells. IL-10 has particular significance in IBD because IL-10 and IL-10R2 deficient mice do not develop lethal systemic autoimmunity, as seen in TGF-β1 deficient mice, but instead develop colitis.^{12, 69} Established intestinal inflammation in mice can be ameliorated via treatment with recombinant IL-10 protein, IL-10 expresing transgenic T cells, or intestinal bacteria engineered to produce IL-10.^70–72 Significantly, the IL10 gene is a susceptibility locus for human ulcerative colitis, and patients with mutations in the IL-10 receptor develop IBD at an earlier age and with higher penetrance than IPEX patients.^{73, 74} T_reg cell-specific deletion of IL-10 results in a similar, though less severe intestinal inflammation than global IL-10 knockout mice.⁷⁵ In the T cell transfer model of colitis, T_reg-produced IL-10 is particularly necessary in the presence of triggering microorganisms such as Helicobacter hepaticus.^{76, 77} These results suggest that IL-10 production by both T_reg and non-T_reg cells is important in maintenance of tolerance in the GI tract, depending on the inflammatory context.

Tr1 cells are one other potential source for IL-10 in the intestine. Tr1 cells produce high levels of IL-10 and require IL-10 for their differentiation. Tr1 cells seem to share other suppressive pathways with T_reg cells, but Tr1 cells are characterized by their lack of Foxp3 expression.⁶ In the colon LP, Foxp3⁺ T_reg cells are the predominant IL-10 producing T cells. However, in the small intestine, the majority of IL-10-producing intraepithelial lymphocytes (IEL) do not express Foxp3, suggesting a potential role for Tr1 cells.⁷⁸ Since Tr1 cells lack a specific marker, little is known about this population. However, the prevalence of Foxp3⁻ IL-10-producing T cells in the intestine, and less severe disease in T_reg cell-specific IL-10 knockout mice suggests that Tr1 cells may also contribute to IL-10-mediated suppression.

Another immunosuppressive cytokine secreted by T_reg cells is IL-35. IL-35 is a newly discovered heterodimeric cytokine consisting of subunits IL12p35 and Ebi3. IL-35 is preferentially produced by T_reg cells and suppresses T cell proliferation in vitro. In the T cell transfer model of colitis T_reg cells deficient in either IL12p35 or Ebi3 showed decreased ability to ameliorate established disease.⁷⁹ In addition, mice deficient for Ebi3 were more susceptible to EAE, and both Ebi3^−/− and IL-12p35^−/− mice showed increased airway hyperresponsiveness, suggesting T_reg-produced IL-35 may play roles in numerous tissues.^{80, 81}

Pericellular adenosine is another soluble immunoregulatory molecule utilized by T_reg cells. Both mouse and human T_reg cells express surface bound enzymes such as CD39 and CD73 that convert proinflammatory extracellular ATP into the immunosuppressive nucleoside adenosine. Adenosine produced by T_reg cells binds adenosine receptors such as A2A on DCs and T_eff cells to decrease their activation and function.^{82, 83} Pericellular adenosine appears to play a critical role in intestinal immune tolerance. In the T cell transfer model of colitis, wild-type T_reg cells could not suppress disease induced by A2A-deficient colitogenic T cells.⁸⁴ Treatment with A2A agonists also ameliorated a spontaneous mouse model of ileitis.⁸⁵

In addition to producing cytokines, T_reg cells can also cause immunosuppression through cytokine deprivation. T_reg cells express elevated levels of numerous cytokine receptors including the high affinity IL-2 receptor CD25.^{8, 22, 86} This increased expression of cytokine receptors is hypothesized to allow T_reg cells to outcompete T_eff cells for survival signals, causing T_eff cell apoptosis. In the T cell transfer model of IBD, transferred T_reg cells increase apoptosis in colitogenic T_eff cells. In vitro provision of exogenous IL-2 or other common gamma chain cytokines decreases T_eff cell apoptosis in T_reg/ T_eff co-cultures.⁸⁷ The Bcl-2 family member Bim signals apoptosis in response to cytokine deprivation. T_eff cells deficient in Bim are resistant to T_reg-driven apoptosis both in vitro and in the T cell transfer model of colitis, suggesting T_eff cells are undergoing apoptosis due to an absence of cytokine-derived survival signals.⁸⁷ But, a role for cytokine consumption by T_reg cells in vivo is difficult to prove. Since T_reg cells could utilize other mechanisms to initiate termination of the immune response, survival cytokines would be expected to subsequently decrease as a result of the decreasing immune response.⁶

T_reg cells also directly interact with other immune cells to mediate immune suppression. The increased apoptosis of T_eff cells via T_reg cells may not only be due to cytokine deprivation, but may also be due to direct cytotoxic effects of T_reg cells on T_eff cells. The direct killing of numerous cell populations including CD4⁺ T_eff, CD8⁺ CTL, and APCs by T_reg cells via perforins and granzymes is well established in vitro.^88–90 T_reg-mediated killing of APCs is a particularly intriguing effector mechanism since this regulatory approach could yield antigen specificity through the TCR/MHCII interactions of these two cell types. The in vivo evidence for cytotoxicity of T_reg cells is more limited. In a mouse model of transplantation tolerance, granzyme B expression by T_reg cells and granzyme targeting in T_eff cells was important for maintenance of the graft.⁹¹ In addition, T_reg cells were shown to control CD8⁺ T cell responses to lung infection in a granzyme B-dependent manner.⁹² The extent to which cytotoxicity plays a role in T_reg cell suppressive function in IBD is currently unknown. Still, it is interesting to note that the increased apoptosis of effector populations through both T_reg -mediated cytokine deprivation and cytotoxicity could result in increased TGF-β1 production, further amplifying the regulatory response.

T_reg cells also express numerous surface receptors that confer suppressive activity. Cytotoxic T lymphocyte antigen 4 (CTLA-4) is foremost among these receptors. CTLA-4 is a high-affinity inhibitory receptor related to CD28. CTLA-4 has greater affinity for its B7 ligands than CD28. This increased affinity seems to increase the duration of time T_reg cells engage with APCs, and modulate the APC function and maturation. In fact, both in vitro and intravital imaging suggests that these stable T_reg-APC contacts limit access of T_conv cells to APCs, thus limiting T_eff cell activation.^{93, 94} In the T cell transfer model of colitis, CTLA-4 blockade limits protection by T_reg cells. Also, CTLA-4 deficient nT_reg cells can prevent colitis induced by wild-type naïve T cells, but not disease induced by CTLA-4-deficient T cells.⁹⁵ This suggests a role for CTLA-4 on nT_reg and also T_conv and/or in vivo iT_reg cells. T_reg cell-specific CTLA-4 deficiency leads to a lethal autoimmune disorder. However, this lethality occurs much later than in Foxp3 deficiency, suggesting other T_reg cell effector mechanisms can temporarily compensate for the absence of CTLA-4.⁹⁶

CTLA-4 also appears to play a particularly important immunoregulatory role in the human intestine. Anti-CTLA-4 (ipilimumab or tremelimumab) treatment increases the immune response to some cancers by presumably decreasing T_reg cell function. However, data from over 4000 patients shows that these treatments can result in potentially lethal colitis.^{34, 97} In addition, treatment with CTLA4-Ig (abatacept) has shown negative effects including new-onset UC in arthritis patients, and exacerbation of symptoms in CD. These results suggest that CTLA-4 may not only block B7 molecules to promote regulation, but that CTLA-4 signaling may be required for T_reg cell function in the gut.⁹⁸

T_reg cells also express several other receptors that may play a role in immunoregulation. Lymphocyte Activation Gene-3 (LAG-3) is a homologue of CD4 that binds with high affinity to MHCII on the surface of APCs. This binding inhibits DC maturation and leads to more tolerogenic DCs.⁹⁹ T_reg cells are one of few cell types that express both programmed death (PD) receptors and ligands. Expression of PD-L1 and PD-1 may promote the differentiation of iT_reg cells. PD-L1 and PD-1 expressed by T_reg cells may also promote stable contacts between T_reg cells and APCs, and modulate APC function, similar to CTLA-4 and LAG-3.¹⁰⁰ These T_reg-APC contacts may be maintained and enhanced via T_reg cell expression of Nrp-1. T_reg cells preferentially expresses additional receptors such as galectins, GITR, and OX40, but their contribution to T_reg effector function is relatively unknown.¹⁰¹

Interestingly, many of these receptors have the potential to interact with both APC and T_conv cells to regulate immune responses. Several studies suggest T_reg cell suppression may be more important in the lymph nodes than the site of inflammation. In the T cell transfer model of colitis, deletions of gut-homing integrins such as CD103 (αE) shows that migration to the intestine is necessary only in the APC population for T_reg to mediate suppression. In addition, T_reg cells must express lymph node homing receptors to suppress colitis. This data argues that in the T cell transfer model of colitis T_reg cells mediate most of their suppressor function through interactions with gut-homing DCs in the lymph node.^102–105

Independent and synergistic roles of nT_reg and iT_reg cells in mucosal tolerance

It is becoming clear in mouse models of IBD that both nT_reg and iT_reg cells are necessary for protection and perhaps also treatment of disease.³⁸ However, how the nT_reg vs iT_reg cell responses are calibrated is currently unknown. TCR repertoire and T_reg differentiation studies suggest that the dual requirement for nT_reg and iT_reg cells may be due to differing TCR specificities between the two subsets of T_reg cells. In particular, it seems that nT_reg cells are necessary to confer tolerance to self antigens, whereas iT_reg cells are induced at mucosal surfaces to confer tolerance to antigens in food and to the microbiota that makeup the “extended self”. Interestingly, IBD patients have increased immunoreactivity to both self and microbial antigens suggesting important roles for both nT_reg and iT_reg subsets and their corresponding specificities in human patients as well.^{106, 107}

The iT_reg subset develops after peripheral activation of T_conv cells in the presence of TGF-β1. However, when levels of the pro-inflammatory cytokine IL-6 are also high, TCR and TGF-β1 signaling results in upregulation of RORγt and development of pro-inflammatory Th17 cells that express the cytokine IL-17 (Figure 1). Thus, iT_reg and Th17 cells both arise from T_conv cells in the presence of TGF-β1. In fact, iT_reg and Th17 cells may share a common Foxp3⁺/RORγt⁺ precursor.¹⁰⁸ Under normal conditions, Th17 cells recruit neutrophils to combat infection by fungi and extracellular bacteria. However, these cells are also associated with numerous autoimmune diseases, including IBD. In fact, increased levels of characteristic Th17 cytokines are found in biopsies from IBD patients.⁴⁸ The role of Th17 cells in IBD is unclear though, as several studies have shown them to be both pathogenic and protective.³⁹ Mouse studies have disagreed, in particular, on the role of IL-17. Disease can be exacerbated, ameliorated, or unaffected by the absence of IL-17 depending on the method of depletion and whether the IL-17A or IL-17F isoform was targeted.⁴⁸

In addition to IL-17, Th17 cells produce IL-21 and IL-22. IL-21 appears to be largely pro-inflammatory, and promotes further amplification of additional Th17 cells. IL-21 expression is upregulated in lesions of IBD patients.¹⁰⁹ In contrast to the other Th17 cytokines, IL-22 appears to be protective in IBD. IL-22 promotes intestinal barrier integrity and prevent microbes from breeching the epithelium. IL-22 increases cell proliferation and survival of the epithelium, upregulates production of antimicrobial peptides (AMP) such as defensins, and stimulates mucus production.⁴⁸

The Th17 cell phenotype is maintained through binding of IL-23 to Th17-expressed IL-23R. IL-23 appears to be necessary for the development of IBD in several mouse models, suggesting a pathogenic role for Th17 cells.^{77, 110–112} In contrast, one study suggests that IL-23 deficient mice have increased susceptibility to colitis induced by TNBS barrier disruption, which agrees with the role of Th17-produced IL-22 in barrier maintenence.¹¹³ Genome-wide association studies identified IL-23R as one of the genes most strongly associated with both UC and CD in human patients.⁵ One allele of IL-23R (R381Q) is protective for IBD. This mutation results in decreased IL-23R signaling and thus decreased numbers of Th17 cells.¹¹⁴ Since this allele confers about three-fold protection from CD, it appears that Th17 cells may play largely pathogenic roles in human IBD.⁵

Therefore, although iT_reg and Th17 cells may share receptor specificities, developmental precursors, and differentiation signals, they have differing impacts on immune responses. In fact, the increased methylation of the Foxp3 locus of iT_reg cells versus that of nT_reg cells suggests that the iT_reg subset may serve as a more flexible lineage, able to adapt to the antigenic or inflammatory environment. Determining whether iT_reg cells can downregulate Foxp3 expression, becoming so-called “ex-iT_reg”, and revert to alternate fates such as Th17 is an active area of research. It is likely that control of the iT_reg/Th17 balance and the stability of iT_reg are key immune regulatory checkpoints.^{115, 116} Thus, iT_reg/Th17 conversion may control regulatory versus inflammatory responses to extended self and foreign antigens, while nT_reg cells control responses to self-antigens in the intestine (Figure 1).

A dual requirement for both nT_reg and iT_reg cells in mucosal tolerance has been supported through several studies. In the T cell transfer model of colitis, disease is much more severe when induced with T_conv cells that have an inability to become in vivo-derived iT_reg cells.³⁸ Mice with established colitis can be successfully treated with the transfer of nT_reg cells, but only if the colitogenic T_conv population can induce Foxp3. In mice lacking in vivo iT_reg cells, co-transfer of nT_reg cells and in vitro iT_reg can cure disease, suggesting a requisite role for these cells in reversing IBD.³⁸ Further studies demonstrated differences in TCR repertoire between the subsets and suggested the primary role for iT_reg cells is to expand the antigen diversity of immunoregulatory responses.²⁴ Though nT_reg and iT_reg cells appear to utilize similar effector mechanisms, iT_reg cells may serve as a flexible lineage able to expand immune regulation to antigens not expressed during thymic nT_reg development.¹¹⁷ The ability of iT_reg cells to aid in peripheral tolerance appears to be particularly important at mucosal interfaces, where the majority of iT_reg cell specificities are to antigens derived from the microbiota.¹¹⁸

Effects of the microbiota on T_reg cells

The microbiota have co-evoved with the host to aid in host metabolism, and also play significant roles in host protection and immunity. The microbiota provide signals leading to the maintenance of the epithelium, and stimulate production of antimicrobial peptides and lectins, thus preventing breech of the initial immune barrier.^{27, 119} The microbiota also directly interact with the innate immune system and aid in adaptive immunity by promoting the development of Peyer’s patches and isolated lymphoid follicles in the gut, as evidenced by GF mice lacking these structures.¹²⁰ In turn, the host immune system limits microbial access to the epithelium via the production of antimicrobial products and the secretion of mucus and IgA.¹²¹

IBD patients exhibit abnormal shifts in the taxa of their microbiota. It is not known whether this abnormal bacterial growth is a direct cause of IBD, or a side effect of increased inflammation in the intestine. Nonetheless, probiotics, antibiotics, and fecal diversion have all shown some beneficial effects in IBD.^{39, 122, 123} Thus, it seems increasingly likely that in IBD there are abnormalities in both the microbiota and the immune response, and that these dual abnormalities may be related.

The T_reg cell compartment seems to be particularly sensitive to changes in the microbiota. TGF-β1 and IL-10 expression by T_reg cells is necessary for their suppressive capabilities, but only in the presence of an activating microbiota.³⁴ T_reg cells from GF mice are generally less suppressive, and express less Foxp3 than T_reg cells from mice under normal conditions.^124–126 Though, one study has demonstrated significant IBD suppression mediated by a T_reg-containing population from GF mice.¹²⁷ The presence of colonizing microbes may also affect the prevalence of T_reg cells in the GALT. T_reg cells numbers are decreased in the colons of GF mice, but are increased in the small intestine.²⁷ Further, T_reg cell numbers in the small intestine LP increase in response to decreasing bacterial load via vancomycin treatment.¹²⁸ Consistent with these findings, TLR9-deficient mice have increased numbers of T_reg cells in the small intestine, suggesting that inflammatory signals from bacterial DNA plays a role in inhibiting iT_reg cell differentiation or T_reg cell proliferation.¹²⁹

Mouse studies of IBD have identified Helicobacter hepaticus and segmented filamentous bacteria (SFB) as two particularly pro-inflammatory members of the intestinal microbiota. Helicobacter hepaticus is a normal member of the microbiota in many mouse facilities. Helicobacter hepaticus and a related species Helicobacter bilus normally do not cause pathology, but can trigger inflammation in susceptible hosts.^{6, 130} In the T cell transfer model of colitis, disease severity was greatly increased in the presence of Helicobacter hepaticus.¹³¹ Further, mice with a T_reg-specific deletion of IL-10 do not normally present any spontaneous phenotype. When infected with Helicobacter, these mice develop spontaneous colitis.⁷⁵ Though Helicobacter hepaticus appears to drive inflammation, T_reg cells from animals infected with Helicobacter hepaticus appear to be more suppressive than T_reg cells from uninfected animals.¹³² Given the numerous effects of Helicobacter spp. on T_reg cells and IBD, it seems likely that the variable presence of Helicobacter in mouse colonies may be one potential explanation for inconsistencies in IBD studies between institutions.

In mice, segmented filamentous bacteria (Candidatus arthromitus) also play an important role in intestinal immune responses. SFB are able to penetrate the mucus layer and directly contact the epithelium, making them particularly equipped to impact host immune responses.¹²¹ SFB are approximately 25 times more prevalent in C57BL/6 mice from the vendor Taconic Farms than in C57BL/6 mice from Jackson Labs, providing sources for relatively SFB sufficient and deficient mice.¹³³ In the T cell transfer model of colitis, the addition of SFB to a specific pathogen free bacterial cocktail led to severe colitis.¹³⁴ Further studies determined that SFB likely promote inflammation by driving Th17 cell differentiation in the small intestine LP.^{133, 135, 136} Monocolonization of GF animals with SFB also leads to IgA production and increased numbers of IEL in the gut.^{135, 137} SFB induction of Th17 cells appears to be dependent on SFB production of serum amyloid A and ATP, which act on LP DCs, that in turn drive Th17 differentiation.^{128, 138}

The impact of SFB on human IBD is unknown however, since this bacterium is rarely found in the human microbiota. This may be due to prevention of SFB colonization by specific human defensins. Indeed, transgenic expression of human defensin 5 in mice leads to a lack of SFB colonization and a subsequent elimination of Th17 cells in the intestinal LP.¹³⁶ One report suggests that in addition to the induction of Th17 cells, SFB may also play a role in differentiation of iT_reg cells.¹³⁵ However, others found SFB did not increase T_reg cells, suggesting the source of SFB may play a role.¹³³

Though T_reg cell numbers generally increase when overall bacterial load is decreased, several studies demonstrated that specific bacterial taxa may drive peripheral Foxp3 induction and/or T_reg cell proliferation. In fact, the colon T_reg TCR repertoire was shown to differ from the T_reg repertoire at other peripheral sites, and these TCR sequences were specific for members of the microbiota. Further, these TCRs were colitogenic when expressed in the absence of Foxp3. These results demonstrate a role for peripheral differentiation of iT_reg cells specific for the bacterial microbiota and provide exciting support for the hypothesized role of iT_reg cells in broadening tolerance to the “extended self” of commensal microorganisms.¹¹⁸

Specific bacterial species may play roles in iT_reg differentiation. Though GF mice lack T_reg cells in the colonic LP, the colonization of GF mice with a distinct combination of 46 strains of Clostridium spp. led to normal development of T_reg cells in the gut ¹³⁹. These Clostridium-colonized mice had greater levels of intestinal TGF-β1 and increased levels of CTLA4 and IL-10 expression in the intestinal T_reg pool. Faecalibacterium prausnitzii is one species included in the 46 Clostridium spp.¹³⁹ F. prausnitzii has been shown to induce IL-10 production and reduce severity of colitis in mouse models. Interestingly, this bacterium is decreased in the intestines of IBD patients and is associated with greater risk of disease recurrence in post-surgical Crohn’s disease patients.¹⁴⁰

Numerous lactic acid bacteria and Bifidobacteria are used as probiotics, and several strains have shown effects on T_reg cells. Lactobacillus acidophilus, Lactobacillus rhamnosus, and Lactobacillus reuteri have all been shown to modulate inflammation in mouse models of disease.³⁹ Lactobacillus reuteri were specifically shown to increase the frequency of T_reg cells in the mLN and spleen of host animals.^{141, 142} Bifidobacterium infantis increase the frequency and number of T_reg cells in the spleen and PP and decrease intestinal inflammation induced by Salmonella typhimurium infection.¹⁴³ Consuming B. infantis also increased the percent of Foxp3⁺ T_reg cells in the peripheral blood of human subjects.¹⁴⁴ VSL#3 is a probiotic mixture of numerous Lactobacillus and Bifidobacteria designed for the management of IBD. Treatment of mice with VSL#3 increases the production of IL-10 and number of TGF-β1-producing T cells, though it is unknown whether these cells are Foxp3⁺. VSL#3 treatment also decreases the severity of a chemically-induced model of colitis.¹⁴⁵

Generally, the presence or absence of bacterial taxa only correlates with effects on the T_reg cell pool. Administered microbes may exert their effects by modifying other members of the microbiota, and not through direct interaction with the immune system. However, Bacteroides fragilis exhibits a distinct mechanism for direct modification of host immune development. Monocolonization of GF mice with B. fragilis leads to increased differentiation of IL-10-producing iT_reg cells.¹⁴⁶ This induction is completely dependent upon expression of a single B. fragilis molecule, polysaccharide A (PSA). Further, this PSA-dependent iT_reg cell development requires TLR2 signaling, suggesting B. fragilis modifies the host APC to drive iT_reg cell differentiation. B. fragilis-produced PSA inhibits Th17 induction and promotes iT_reg development, thereby protecting the host from both chemically-induced and H. hepaticus-driven colitis.^{146, 147} Conversely, enterotoxigenic strains of B. fragilis, which naturally lack PSA expression, play a role in inducing Th17 cells and promote intestinal inflammation.¹⁴⁸

Numerous bacterial strains can affect the host T_reg cell compartment. Many of these bacteria are promising candidates for probiotic or antibiotic treatments for IBD. However, most of these taxa, including Lactobacillus, Bifidobacterium, Helicobacter, F. prausnitzii, and B. fragilis are of low or variable abundance in the human microbiota, and SFB are largely absent in humans.^{39, 121} Further, probiotic and antibiotic-driven changes in the microbiota often do not endure once treatment is discontinued. Therefore, additional studies will be necessary to identify the role indigenous bacteria play on the T_reg cell lineage, and how these interactions may promote or prevent IBD.

Failure of regulatory mechanisms in IBD

Since IBD involces a failure in immune regulation, it seems likley that IBD is caused in part by defects in T_reg cells. These defects could be T_reg cell-intrinsic, resulting from inherently defective T_reg cells in the patient. Alternatively, the causes of defective T_reg cells could be T_reg cell-extrinsic, and a function of the abnormal intestinal environment and cytokine milieu in IBD patients. These T_reg-intrinsic or extrinsic defects can lead to insufficient T_reg cell suppression of IBD in at least 4 ways: 1) deficient T_reg cell numbers, 2) defective T_reg cell function, 3) unstable T_reg cell phenotype, and 4) pathogenic T cell resistance.¹⁴⁹

The majority of what we know about the mechanisms of T_reg involvement in IBD comes from mouse models. Mouse models allow better identification and tracking of bona fide T_reg cells and easier analysis of the intestinal milieu and sites of inflammation. Complete T_reg cell deficiency, such as that in scurfy or other Foxp3 mutant mice, leads to fatal autoimmunity that includes intestinal inflammation, underscoring deficient T_reg number as a potential mechanism for IBD.^{8, 150} In T cell transfer models of colitis the colitogenic T cell population must be depleted of T_reg cells in order to cause disease.¹⁹ Further, injection of both T_reg cell subsets into these mice can prevent or cure IBD in a dose-dependent fashion. ³⁸ Mutations in other potentially T_reg cell-intrinsic genes such as CD25, IL-2, CD28, and TGF-β1 also lead to decreased numbers of T_reg cells, though IL-2 and TGF-β1 expression by other cell types likely contributes.^{2, 149} Other T_reg cell-extrinsic factors also play a role in mouse models. Cytokines such as IL-6, IL-21, IL-27, and IL-23 may suppress differentiation of iT_reg, suggesting a role for the intestinal milieu on T_reg cell number.⁶

T_reg cell function may also be affected by the same cytokines that lead to changes in T_reg cell number. In addition, other T_reg cell-extrinisic cytokines such as TNF, IL-4, IL-12, IL-7, and IL-15 may limit T_reg function.¹⁴⁹ Several molecules have been shown to play T_reg cell-intrinsic roles in T_reg suppressive function. Mutations in T_reg surface receptors CTLA-4, CD39, LAG-3, and Fas have been shown to limit T_reg cell suppression.¹⁴⁹ In addition, T_reg cells deficient in the soluble factors TGF-β1, IL-10, and IL-35 have decreased function (^{Table 2).149}

Mouse T_reg cells have also shown variable stability. IL-6 exposure drives a fraction of T_reg cells into the Th17 lineage in vitro.¹⁵¹ In the T cell transfer model of colitis, transfer of T_reg cells into an IL-10 deficient host led to a loss of Foxp3 expression and conversion of T_reg cells into Th1 pathogenic effector cells.¹⁵² Adoptive transfer of T_reg cells into T cell deficient hosts led to the loss of Foxp3 expression and conversion of T_reg cells into T follicular helper cells in the intestinal PP.¹⁵³ Other lineage-tracing studies showed that T_reg cells that lost Foxp3 expression converted to Th17 cells in the PP, but became more Th1-like in the spleen and mLN. Further, these studies showed that in a mouse model of diabetes T_reg cells that lost Foxp3 expression could be pathogenic upon re-transfer.¹⁵⁴ In the transfer model of colitis, the transfer of in vitro-derived iT_reg cells as treatment resulted in 85% of transferred cells losing Foxp3 expression, becoming ex-iT_reg cells with pathogenic potential.¹¹⁷ Conversely, another study utilizing an inducible cell-tracking system found T_reg to be rather stable.¹⁵⁵ The differences in these studies may reflect the differences in iT_reg cell vs nT_reg cell involvement in each system. In general, these studies have shown the iT_reg cell lineage to be more plastic and able to become pathogenic ex-iT_reg, whereas the nT_reg cell lineage demonstrates greater stability (Figure 1).

Mouse models have shown that even in the presence of functional T_reg cells, pathogenic effector cells can be resistant to T_reg-mediated suppression. In addition to affecting T_reg cell function, the cytokines IL-2, IL-4, IL-7, and IL-15 can drive proliferation of effector T cells in the presence of T_reg cells in vitro.¹⁵⁶ Also, agonistic stimulation of TNF receptor family members OX40 or 4-1BB on conventional T cells leads to effector T cells that are resistant to T_reg suppression in vitro.^{157, 158} For OX40, this action has been demonstrated in vivo in the T cell transfer model of colitis.¹⁵⁸

Human studies of IBD have a distinct advantage over other autoimmune diseases such as multiple sclerosis and lupus because tissue samples from the site of inflammation can be removed and studied. However, since human activated effector T cells also express Foxp3, histological analysis of T_reg cells in IBD patients can be difficult. Nonetheless, studies with human patients have been able to support regulatory mechanisms first discovered in mouse models. Research in human patients has similarly investigated T_reg cell number, T_reg cell function, T_reg cell stability, and effector cell resistance to T_reg as potential causes of IBD.

Numerous studies have investigated T_reg cell number in human IBD patients. In general, these reports observe greater numbers of Foxp3⁺ cells in the intestine of patients, particularly in active inflammatory lesions.^{77, 159–164} Numbers of T_reg cells in the peripheral blood are less consistent than the intestine. However, most studies identify a decreased frequency of T_reg cells in the peripheral blood of IBD patients. This decrease is intensified as T_reg cell numbers increase in the intestine during active disease, suggesting a potential redistribution of these cells in response to active inflammation.^{77, 162–165} Interestingly, the increase in T_reg cells in the intestine does not appear to be specific to autoimmune disease, as similar or greater numbers of T_reg cells are seen in other inflammatory conditions such as diverticulitis and enteritis. ^{77, 162}

T_reg cells at the site of inflammation fail to protect against IBD despite being present in increased numbers. This suggests that T_reg function may be compromised in these patients. To investigate this point, several groups removed T_reg cells from IBD patients and tested their suppressive function in vitro. These studies showed uniformly that T_reg cells from the peripheral blood, mLN, and LP of IBD patients were all equally suppressive to T_reg cells from normal controls.^{159, 161–163, 166–168} Thus, T_reg cells from IBD patients appear to be present in sufficient numbers and function normally ex vivo. These results suggest that there may not be an intrinsic defect of T_reg cells in IBD, but an extrinsic effect of the intestinal milieu on T_reg cells. In fact, numerous IBD susceptibility genes such as autophagy genes, NOD2, and IL23R, could play a role in creating an abnormal intestinal microenvironment that limits T_reg cell function in vivo. Previous studies in mouse models suggest that cytokines within an inflammatory microenvironment may also promote T_reg cell instability.¹⁶⁹ Similar results have been seen for human T_reg cells. Human T_reg cells from peripheral blood can convert to Th17 cells when stimulated in vitro in the presence of IL-2. This conversion is increased by inflammatory cytokines IL-1β IL-23, and IL-21 or APCs activated by microbial stimuli.¹⁷⁰

Effector T cell hyporesponsiveness to suppression is another T_reg cell-extrinsic factor that may play a role in human IBD. Two studies investigated T_eff responsiveness and found no difference between IBD patients and controls.^{159, 167} Other groups found T_eff cells from patients to be resistant to T_reg suppression. These studies suggested that T_eff cells from IBD patients overexpress SMAD7, thus making them resistant to TGF-β1 secreted by T_reg cells.^{42, 171} Further, T_reg cells may suppress disease by driving T_eff apoptosis, and T_eff cells in IBD patients are generally resistant to apoptosis.^{172, 173} Given the data suggesting a potential role for the intestinal milieu, it is important to note that these studies of T_eff responsiveness are performed on T cells outside of the inflammatory intestinal niche.

The potential for T_reg cells in IBD therapies

Since T_reg cells play a requisite role in preventing autoimmunity, these cells may be ideal targets for developing new treatments for IBD. There are currently over 50 products being tested for use in the treatment of IBD. These include small molecules, siRNA, peptides, vaccines, and numerous monoclonal antibodies and cell-based treatments.⁹⁸ Several of these treatments center on T_reg cells. However, additional research has suggested several other potential mechanisms for T_reg-based therapies. In order to find new therapies that rely on T_reg cells researchers must characterize T_reg cells during the disease process, and determine the mechanisms of impaired T_reg cell suppression in IBD. This knowledge will aid in developing treatments that will increase T_reg cell number, function, and stability.¹⁴⁹ Developing treatments involving T_reg cells is particularly challenging for IBD since the precise role for T_reg cells in IBD patients has not been fully elucidated.

Traditional treatments for IBD appear to have effects on the T_reg cell compartment. Standard treatment of UC patients with aminosalicylates or glucocorticoids led to increased peripheral blood frequencies of CD4⁺CD45RO⁺CD25⁺ T cells, a population enriched in T_reg cells.¹⁶⁴ In CD however, treatment with azathioprine or mercaptopurine led to decreased frequencies of T_reg cells in the peripheral blood.¹⁶³ Treatment with monoclonal antibodies to TNF-α (infliximab) was shown to increase T_reg cells in the peripheral blood and intestinal LP by two separate groups.^{174, 175} However, another group did not see any difference in T_reg cell frequency upon infliximab treatment.¹⁷⁶ These classically used IBD treatments have broad effects, and a disadvantage of these approaches is down-modulation of both innate and adaptive immunity and increased risk for opportunistic infection.⁴⁸ Therefore, there is a need to increase the specificity of IBD treatments, with T_reg cells as one particular area of interest.

IBD treatments focused on T_reg cells fall into two categories: cell-based treatments that involve the transfer of in vitro expanded or stimulated T_reg cells into patients, or pharmacological approaches that attempt to influence T_reg cells in vivo. The small number of T_reg cells that can be obtained from a patient’s blood limits cell-based therapies. Thus numerous approaches are being investigated to expand and activate a patient’s own peripheral blood T_reg cells for re-transfer. Such approaches have been shown to increase the number of T_reg cells up to 13000 fold in 3–4 weeks. Yet, a fraction of these cells lose Foxp3 expression, indicating instability of the T_reg cells or outgrowth of T_conv cells in this population.⁶¹ Another potential source for T_reg cell-based treatments is the in vitro differentiation of iT_reg cells from T_conv precursors via TCR stimulation and TGF-β1. This approach can yields an average of 240 × 10⁹ iT_reg cells in approximately 2 weeks in culture.¹⁷⁷

Both in vitro expansion of nT_reg and differentiation of iT_reg are limited by several factors. These T_reg cells may still have relatively few antigen specificities or the incorrect antigen specificities for proper suppression. Tetramer-based approaches could lead to expansion of relevant T_reg cell specificites, but rely on knowledge of the target antigens for disease.¹⁷⁸ Further, transferred T_reg cells may harbor potentially harmful specificities if they do not remain stably suppressive.^{117, 154} Addition of rapamycin to cultures leads to increased numbers of T_reg cells for both in vitro nT_reg expansion and iT_reg differentiation.^{177, 179} Further, rapamycin reduces the outgrowth of contaminating effector T cells, and increases T_reg stability.¹⁸⁰ Another limitation of cell-based approaches is that transferred T_reg cells may not localize to the inflamed target tissue. T_reg cells could be engineered with receptors such as tissue-specific integrins, driving a T_reg cell to home the intestine. However, for these approaches to work in human patients gene therapy would have to be utilized, which raises safety concerns.

Mouse studies have utilized T_reg cells that become activated by a bystander Ag in the target tissue to increase T_reg localization.¹⁸¹

One recent phase 1/2a clinical study demonstrated both safety and efficacy of ovalbumin-specific T_reg cells in a small CD patient population.¹⁸² Autologous T_reg cells were expanded in vitro, cloned by limiting dilution, and selected for IL-10 production in response to ovalbumin. Since ovalbumin is a common food antigen, ovalbumin specificity was used as a bystander antigen in an attempt to drive gut homing and local immune regulation in the intestine. Five weeks post-treatment 8/20 patients showed significant clinical improvement, with 6/8 showing improvement at one dose. However, the T_reg subset being used in these studies is indeterminate since they express numerous molecules associated with Tr1 cells, yet are activated in vitro and can also express Foxp3. ¹⁸²

Cell-based treatments may also utilize populations other than T_reg cells. Transfer of tolerogenic DCs could drive iT_reg differentiation in vivo.¹⁸³ Mesenchymal stem cells can produce IL-10 and TGF-β1, which could lead to increased iT_reg differentiation or overall T_reg function.^{98, 184} Transfer of mesenchymal stem cells ameliorated a mouse model of chemically induced colitis and more than doubled the frequency of T_reg cells in the mLN of treated mice.¹⁸⁵

Recent studies underscore the need for both nT_reg and iT_reg cells for complete tolerance.²⁴ Thus, clinical studies may need to utilize transfer of both nT_reg and iT_reg cells. Alternatively, nT_reg cells could be transferred in addition to T_conv cells that cannot become pathogenic, or nT_reg cells in combination with another cell population to drive in vivo iT_reg differentiation. One can also imagine utilizing either nT_reg or iT_reg cells for different clinical outcomes.³⁵ For example, an inflammatory response against the extended self may benefit more from iT_reg cell specificities, whereas an autoimmune response may be better treated via nT_reg cells. However, as noted T_reg cell numbers are not generally decreased in human patients, and T_reg cells from these patients are generally suppressive ex vivo. Thus, T_reg cells in IBD patients may not be intrinsically flawed, but may fail to suppress disease due to local factors affecting T_reg cell function. Thus, other therapeutic approaches may be necessary to increase T_reg cell number or function in situ.

Probiotics are one approach to increase T_reg cell responses specifically in the intestine. Since IBD patients often display an abnormal microbiome, reshaping of the intestinal microbiota has been suggested as a potential treatment for disease. In fact, complete microbiota transplant, also known as fecal bacteriotherapy, has been used to treat Clostridium difficile-associated colitis, and has shown promise in some cases of IBD.^{123, 186, 187} More commonly, specific species or of bacteria are utilized as probiotics. Many of these treatments were found to increase T_reg cell numbers and function in both humans and mice, likely through local effects on gut-associated DCs.^{144, 188–190} Probiotic strains can also be genetically engineered to better modify the inflammatory intestinal milieu.¹⁹¹ Lactococcus lactis engineered to secrete IL-10 transiently colonized mice and decreased two models of colitis.⁷² A phase I clinical trial demonstrated safety of a similar approach in human patients, but efficacy has not been determined.¹⁹²

Vitamins A and D may also affect T_reg cells and serve as potential therapeutic agents for IBD. The vitamin A metabolite RA appears to increase T_reg cell induction and homing in the intestine, and RA treatment ameliorated mouse models of IBD.¹⁹³ Further, RA treatment increases Foxp3 expression in human intestinal biopsies, suggesting potential therapeutic applications for vitamin A metabolites in IBD.¹⁹³ Vitamin D is a sunlight-dependent precursor to the hormone calcitriol. Vitamin D metabolites and analogues have been variably shown to stimulate mouse and human T_reg cell and T_eff differentiation and function.^194–196 Interestingly, IBD risk increases significantly with increasing latitude, suggesting a role for vitamin D as an environmental risk factor in IBD.¹⁹⁷ Also, higher predicted serum levels of vitamin D reduce the risk for IBD, and the vitamin D receptor is a potential susceptibility allele for UC and CD.^{198, 199} Numerous mouse studies also support a protective role for vitamin D in IBD.²⁰⁰ However, few intervention studies treating IBD patients with vitamin D have been performed. A recent meta-analysis suggested a potential benefit to vitamin D supplementation of IBD patients, but underscored the need for randomized placebo-control studies.²⁰¹

In atopic disease allergen-specific hyposensitization is a common approach for inducing tolerance to an antigen. This involves repeated parenteral or mucosal vaccination with the antigen at increasing doses, and has been shown to be safe and decrease hypersensitivity. Significantly, this approach seems to rely on the differentiation or activation of antigen-specific T_reg cells.⁷ These approaches have been used for food allergy, suggesting this approach is effective for intestinal antigens.²⁰² Hyposensitization may be difficult to apply to IBD since the specific antigens for this hypersensitivity are largely unknown. Nevertheless, a recent study in a mouse model showed promising results. Oral administration of components of one specific species of the microbiota led to increased T_reg cell number and frequency in the mLN. Further, this treatment prevented a chemically induced model of colitis, and attenuated established disease.²⁰³

As mentioned, the classical treatments of IBD may obtain some of their beneficial results via effects on T_reg cells. However numerous other pharmacologic agents are also used to treat IBD, and often have notable effects on T_reg cells. In mouse models, histone deacetlylase inhibitors increased T_reg cell number and function, and decreased chemically induced colitis.^{204, 205} Other small molecules such as siRNA and nanoparticles are currently being used to target intestinal tissue, and may also have future applications to increase T_reg cell number and/or function in the gut.⁹⁸

Monoclonal antibodies are the largest class of pharmaceuticals in development for treatment of IBD. Anti-p40 (briakinumab, ustekinumab) limits Th17 and Th1 responses in IBD patients by blocking the cytokines IL-12 and IL-23.^206–208 Anti-p40 may also have beneficial effects on T_reg cells since IL23Rko mice have increased T_reg cell numbers, particularly in the GI tract.^{209, 210} Anti-CD3 has shown beneficial effects in mouse models of autoimmune disease, leading to deletion and anergy of pathogenic T cells, while promoting T_reg cell responses.²¹¹ However, one anti-CD3 monoclonal antibody (visilizumab) recently failed to show efficacy after parenteral injection, despite promising pilot studies.²¹² Clinical studies treating IBD patients with an oral anti-CD3 antibody (muromonab) are currently underway.²¹³

Other integrin-specific monoclonal antibodies are being used to limit pathogenic T cell access to intestine. Natalizumab, anti-α4 integrin (CD49d) has been approved for use in IBD in the U.S. However, unexpected complications arose with this drug due to α4 integrin’s role in lymphocyte homing to the CNS.²¹⁴ Thus, newer studies are utilizing vedolizumab to target the α4β7 integrin dimer, which appears to control homing only to the intestine.^{215, 216} One concern with monoclonal antibodies targeting integrins it that they may also limit T_reg cell access to the site of inflammation. However, any constrains on T_reg access to the intestine maybe counteracted by the blockade of pathogenic T cells and the ability of T_reg cells to suppress autoimmune responses in the mLN and other peripheral lymphoid tissues.

In general, the development of potential treatments for IBD is outpacing the ability to test these approaches in human patients. Clinical studies are often statistically limited by the ability of a small number of willing patients to differentiate relatively small effects on disease progression.⁹⁸ Future studies may benefit greatly from personalized medicine approaches to stratify treatment groups in a clinical study. Genetic screens and serum biomarker analysis may identify the patients most likely to respond to a T_reg-based therapy. Examination of a patient’s microbiota, and other non-genetic factors such as nutritional and vitamin status may also increase the success of treatments. Exciting new treatments from IBD will likely result from combining the increased ability to develop novel therapies with new technologies to identify the patients most likely to respond. A combination of genetics, microbiota, and other non-genetic/environmental influences are thought to contribute to IBD risk. Importantly, T_reg cells interact with, and can be affected by all three of these contributing factors. It therefore seems highly likely that future treatments for IBD will include therapies targeting and capitalizing on the potent immunomodulatory effects of T_reg cells.