Calcineurin-mediated IL-2 production by CD11chighMHCII+ myeloid cells is crucial for intestinal immune homeostasis

The intestinal immune system can respond to invading pathogens yet maintain immune tolerance to self-antigens and microbiota. Myeloid cells are central to these processes, but the signaling pathways that underlie tolerance versus inflammation are unclear. Here we show that mice lacking Calcineurin B in CD11chighMHCII+ cells (Cnb1CD11c mice) spontaneously develop intestinal inflammation and are susceptible to induced colitis. In these mice, colitis is associated with expansion of T helper type 1 (Th1) and Th17 cell populations and a decrease in the number of FoxP3+ regulatory T (Treg) cells, and the pathology is linked to the inability of intestinal Cnb1-deficient CD11chighMHCII+ cells to express IL-2. Deleting IL-2 in CD11chighMHCII+ cells induces spontaneous colitis resembling human inflammatory bowel disease. Our findings identify that the calcineurin–NFAT–IL-2 pathway in myeloid cells is a critical regulator of intestinal homeostasis by influencing the balance of inflammatory and regulatory responses in the mouse intestine.

T he intestine is a unique and challenging immune microenvironment in which homeostasis requires active tolerance towards self-antigens, dietary-antigens, and commensal microorganisms, balanced with the requirement to detect and respond rapidly to pathogen invasion. Dysregulation of the intestinal immune compartment can result in inflammatory bowel disease (IBD), which affects~3.5 million people in the US and Europe 1 .
The mouse intestinal lamina propria (LP) contains distinct subsets of mononuclear phagocytes that maintain tolerance and immune reactivity towards pathogens 2,3 . These subsets include conventional dendritic cells (DC), monocytes, and tissue-resident macrophages. LP DCs express high levels of CD11c, MHCII, and FMS-like tyrosine kinase 3 (FLT3, also known as CD135) and FLT3 ligand (FLT3L), and are negative for the lineage markers CD3, CD19, B220, NK1.1, and CD64, whereas the monocytes and macrophages predominantly express CD64 and CD11b 3 . Intestinal DCs comprise three major subsets divided by CD103 and CD11b expression: CD103 + CD11b − , CD103 + CD11b + , and CD103 − CD11b + . Similar to LP populations, these DCs are present in the mesenteric lymph nodes (MLN), suggesting that they contribute to initiation of immune responses. Despite the importance of these myeloid cells in supporting intestinal immune homeostasis, we know little about the cellular signaling networks that modulate tolerance-versus-inflammation in the intestine.
The nuclear factor of activated T-cells (NFAT) family of transcription factors is an important mediator of DC function 4 Fig. 1 Calcineurin B and NFAT expression in mouse intestinal myeloid cells. a Relative expression levels of Cnb1, Nfat1, and Nfat2 mRNAs in intestinal CD11c high MHCII + cells (CD11b + and CD11b − ) and MLN CD3 + T cells, assessed by qRT-PCR. Data represent the means ± standard error of three experiments (n ≥ 7 mice/exp). b, c Percentage of NFAT-1 + cells (b) and NFAT-1 protein levels (c) in CD11c high MHCII + cells (CD11b + and CD11b − ) obtained from spleen, MLN, PP, and colonic LP (LP-colon), assessed by flow cytometry. Splenic CD3 + CD4 + T cells are included for comparison. Data represent the means ± standard error of three experiments (n = 4-5 mice/experiment). **P < 0.01 versus CD4 + T cells (ANOVA followed by Dunnett's multiple comparisons test). d Representative images of NFAT-1 labeling in sorted colonic CD11c high MHCII + cells (CD11b + and CD11b − ). Scale bar 5 μm. e NFAT-1 nuclear translocation in CD11c high MHCII + cells (CD11b + and CD11b − ) after 30 min stimulation with thapsigargin (Thap; 200 nM), as assessed by confocal analysis. Data represent the means ± standard error of two experiments (n = 6-10 mice/experiment). Scale bar 5 μm. f NFAT-dependent luciferase activity measured in a DC cell line in response to TLRs (Poly I:C, LPS), dectin-1 ligands (PGG and WGP), and the calcium mobilizer thapsigargin. g Dose-dependent inhibition of NFAT nuclear translocation by cyclosporin A (CsA) and tacrolimus (FK506) in WGP-stimulated D1 cells, as assessed by NFAT-luciferase activity. AU arbitrary units, Ctrl control, Luc luciferase, LP lamina propria, LPS Lipopolysaccharide, MLN mesenteric lymph node, PGG soluble β-(1,3)-glucan, PP Peyer's Patches, WGP whole glucan particles, Thap thapsigargin Initially, NFAT members were thought to be involved solely in the transcriptional activation of T cells 6 , but different NFAT proteins are now known to have a complex influence on host tolerance to self and foreign antigens [7][8][9] . Upon calcium influx induced by ligation of pattern recognition receptors, activated NFAT triggers calmodulin-mediated activation of the protein phosphatase calcineurin, which in turn induces NFAT nuclear translocation 4,10 .
NFAT is a master transcription factor that regulates expression of IL-2, an immuno-regulatory cytokine that controls T-cell expansion and differentiation, Treg-cell maintenance, and NKcell activation 6 . DCs are a source of IL-2 11,12 and IL-2 can regulate immune responses in the mouse lung during fungal infection 5,13 , and promote adaptive immune responses to alum immunization 14 . Whether the calcineurin-NFAT-IL-2 axis in antigen-presenting cells (APC) can contribute to immune homeostasis under basal and inflammatory conditions in other organs, such as the intestine, is unknown.
Here we use mice lacking calcineurin B (encoded by Cnb1) or IL-2 (encoded by Il2) in CD11c high MHCII + cells, and show that calcineurin-NFAT-IL-2 signaling in intestinal CD11c high MHCII + APCs prevents spontaneous chronic intestinal inflammation in acute and chronic models of induced colitis. Calcineurin-NFAT signaling in intestinal DCs mediates IL-2 production, which supports local Treg cell expansion in the intestinal LP and restrains inflammatory T-cell expansion and effector function. Our data shed light on the processes underpinning tolerance versus inflammation in the mouse intestine, and identify the calcineurin-NFAT-IL-2 pathway in intestinal APCs as a potential therapeutic target for IBD.
Calcineurin B in myeloid cells maintains intestinal homeostasis. We next asked whether the calcineurin-NFAT pathway has a functional role in intestinal immune homeostasis. We generated Cnb1 CD11c mice in which calcineurin B expression is lost in cells expressing CD11c at high level, by crossing Cnb1 fl/fl mice with CD11c-specific Cre-mice 15 . Cnb1 mRNA was significantly diminished in LP-colon of CD11c high MHCII + CD11b − and CD11c high MHCII + CD11b + myeloid cells (mostly belonging to the DC pool) of Cnb1 CD11c mice, compared to CD11c low CD11b + CD64 + cells (mostly macrophages) (Fig. 2a). Cnb1 deletion prevented thapsigargin-driven NFAT-1 nuclear translocation in CD11b + and CD11b − CD11c high MHCII + cell populations in the MLN (Fig. 2b). However, calcineurin-NFAT-1 deficiency in CD11c high MHCII + cells did not affect the relative abundance of these myeloid populations in LP-colon of Cnb1 CD11c mice (Fig. 2c), or the expression of maturation markers of CD11c high MHCII + myeloid cells in LPcolon and LP-small intestine (LP-SI) ( Supplementary Fig. 1). Calcineurin B expression remained intact in B, NK, and mast cells, and naïve, memory CD4 + T cells and Treg cells from MLN of Cnb1 CD11c mice, which also released normal cytokine levels in vitro ( Supplementary Fig. 2a-c).
To confirm that Cnb1 deletion in CD11c high MHCII + cells did not impact T-cell function in vivo, we monitored the development of colitis following adoptive transfer of naïve CD4 + CD45RB high CD25 − T cells from Cnb1 CD11c and Cnb1 fl/fl mice into immune-deficient Rag2 KO mice 16 . A normal course of colitis development was observed in Rag2 KO mice receiving naïve CD4 + T cells from Cnb1 CD11c mice ( Supplementary Fig. 2d, e). These data indicate that CD11c high MHCII + cells are the predominant cell population targeted by the Cnb1 deletion and the abundance and activation status of myeloid cells and effector function of CD4 + T cells in the intestine of Cnb1 CD11c mice remain unaffected.
Having confirmed the specificity of our Cnb1 deletion to CD11c high MHCII + cells, we asked whether disruption of calcineurin-NFAT signaling in these cells affected intestinal homeostasis in vivo. Macroscopic examination of Cnb1 CD11c mice (aged 10-14 weeks) revealed moderate enlargement of the MLN compared to Cnb1 fl/fl control mice (Fig. 3a), and spontaneous inflammation of the SI and colon, characterized by an inflammatory infiltrate in the submucosa with marked erosion of the mucosal lining (Fig. 3b). Although Cnb1 CD11c mice did not develop evident symptoms of colitis (such as weight loss, diarrhea or rectal bleeding), they exhibited significantly higher intestinal permeability, titers of fecal IgA (Fig. 3c), and myeloperoxidase activity in homogenates of the terminal ileum ( Fig. 3d) than Cnb1 fl/fl controls. Increased levels of mucosal TNF and IFNγ were evident in the SI and colon of Cnb1 CD11c mice, compared to Cnb1 fl/fl controls, while IL-17 levels were comparable (Fig. 3d).
Analysis of the immune-cell composition and cytokine profile in the Cnb1 CD11c LP revealed a higher frequency of immune cells (CD45 + and CD4 + T cells), and a higher percentage of antigenexperienced CD44 + CD62L − CD4 + T cells in the LP-SI and LPcolon compared to Cnb1 fl/fl controls (Fig. 3e). Cnb1 deficiency in CD11c high MHCII + myeloid cells also significantly increased the frequency of IL-17-producing and IFNγ-producing CD4 + T cells infiltrating the LP-SI, and IFNγ-producing CD4 + T cells in the LP-colon (Fig. 3e). No difference in IL-4-producing CD4 + T cells or the CD8 + T-cell phenotype was observed in the LP-colon of Cnb1 CD11c mice compared to Cnb1 fl/fl controls ( Supplementary  Fig. 3a). Importantly, Cnb1 CD11c mice exhibited significantly reduced frequencies of FoxP3 + Treg cells in the LP-colon, LP-SI, and MLN, but not in the spleen or thymus (Fig. 3f). Spontaneous chronic inflammation occurred exclusively in the intestine and systemic inflammation, immune phenotype and cytokine release from splenic CD4 + T cells were comparable between Cnb1 CD11c and Cnb1 fl/fl mice ( Supplementary Fig. 3b, c).
These data support that the calcineurin-NFAT axis in CD11c high MHCII + cells is critical for CD4 + T-cell homeostasis in the intestine under steady-state conditions. Cnb1 CD11c mice have unrestrained induced gut inflammation. We next asked how dysregulated T-cell responses would affect the ability of Cnb1 CD11c mice to restrain pathologic mucosal inflammation during acute and chronic models of IBD. Intrarectal instillation of trinitrobenzene sulfonate (TNBS) rapidly induces intestinal inflammation in mice that recapitulates key features of Crohn's disease 17,18 . Compared to TNBS-treated Cnb1 fl/fl controls, TNBS-treated Cnb1 CD11c mice exhibited a significant loss in body weight by 6 days post treatment, worse stool consistency (Fig. 4a), and severe ulceration and thickening of the colon wall (Fig. 4b). Using a histopathological scoring approach, we found that Cnb1 CD11c mice exhibited severe leukocyte infiltration that penetrated all colon layers (Fig. 4c) and a significant increase in myeloperoxidase activity, TNF, and IL-17 in colon homogenates (Fig. 4d). Treatment of Cnb1 CD11c mice with naproxen, a non-steroidal anti-inflammatory drug that can exacerbate colitis in susceptible IBD patients, also induced notable colonic mucosal inflammation compared to untreated Cnb1 CD11c mice or naproxen-treated Cnb1 fl/fl control mice  Supplementary Fig. 4). These results show that calcineurin B expression in CD11c high MHCII + cells not only prevents spontaneous colitis under steady-state conditions, but also helps suppress severe intestinal pathology in response to acute and chronic inflammation.
Cnb1 deletion in non-DC phagocytes reduces colitis severity.
To examine the effect of Cnb1 depletion in monocytes, macrophages, and granulocytes on intestinal inflammation, we generated a Cnb1 LysM mouse model, in which the Cnb1 floxed allele was deleted by Cre-recombinase in cells expressing LysM 19 . The intestinal phenotype of LP mononuclear cells from Cnb1 LysM mice under steady-state conditions was indistinguishable from Cnb1 fl/fl controls with respect to: maintenance of body weight, colon histology, LP-colon immune-cell phenotypes, and levels of pro-inflammatory cytokines ( Supplementary Fig. 5a-c). Moreover, upon TNBS treatment, symptoms of acute colitis were significantly less severe in Cnb1 LysM mice than in Cnb1 fl/fl controls ( Supplementary Fig. 5d-f). These data highlight that calcineurin B expression in CD11c high MHCII + LysM − cells, which mainly represent the conventional DCs in the intestine, participates in intestinal homeostasis by restraining pathological inflammation during colitis while calcineurin B expressed in LysM + monocytes/ macrophages and granulocytes has little effect on intestinal homeostasis during steady-state conditions, but exacerbates intestinal immunopathology during TNBS-induced colitis.
Calcineurin B partially regulates DC-derived IL-2 production. To delineate the mechanistic basis for calcineurin-mediated protection from colitis elicited by CD11c high MHCII + DCs, we examined a role for calcineurin B in the production of DC cytokines regulating T-helper differentiation. Colonic CD11c high MHCII + CD11b − cells from Cnb1 CD11c mice produced significantly less IL-2 following thapsigargin exposure compared to cells from Cnb1 fl/fl mice, while the levels of IL-6, IL-12, IL-23, IL-10, and TGFβ were comparable (Fig. 5a). Similarly, we found that bone marrow (BM)derived DCs released IL-2 in response to thapsigargin, LPS, and zymosan, as assessed by ELISA and intracellular IL-2 labeling (Fig. 5b, c and Supplementary Fig. 6a), whereas IL-2 production from Cnb1 CD11c DCs, but not IL-12p40 or IL-6, was significantly reduced. Moreover, exposure to NFκB and p38 MAPK inhibitors fully abrogated IL-2 release from Cnb1 CD11c DCs ( Supplementary  Fig. 6b). These data indicate that IL-2 synthesis in DCs depends on calcineurin-NFAT, NFκB and AP1 signaling.
To examine whether deficient IL-2 expression in CD11c high MHCII + DCs of Cnb1 CD11c mice was the key factor involved in dysregulation of intestinal immune responses, we administered IL-2:anti-IL-2 complexes (IL-2C), which increase IL-2 bioactivity in vivo, and assessed whether normal intestinal homeostasis could be restored in these mice. While the proportions of Foxp3 + Treg cells were increased to a similar extent in the spleens of both Cnb1 CD11c and Cnb1 fl/fl mice after IL-2C treatment, these cells expanded only in the intestinal mucosa of Cnb1 CD11c mice compared to Cnb1 fl/fl controls, suggesting that lack of Cnb1 in CD11c + cells is partly responsible for altered numbers of Treg cells in these mice ( Supplementary  Fig. 10a). To confirm that the impairment in DC-derived IL-2 signaling induces intestinal CD4 + T-cell dysregulation observed in Cnb1 CD11c mice, we adoptively transferred wild-type or Il2-null DCs into the Cnb1 CD11c mice. Wild-type DCs selectively restored the levels of activated CD4 + T cells producing IFNγ in the LPcolon of Cnb1 CD11c mice compared to untreated or Cnb1 CD11c mice receiving Il2-null DCs (Supplementary Fig. 10b). Moreover, wild-type DC transfer into Cnb1 CD11c mice normalized the percentage of Treg cells in MLN, whereas the injection of Il2-null DCs induced an additional reduction of colonic FoxP3 + Treg cells. (Supplementary Fig. 10c).
DC IL-2 production prevents severe spontaneous chronic colitis. A role for T-cell-derived IL-2 in regulating T-cell responses and maintaining Treg-cell-mediated self-tolerance is well-established, but a distinct contribution of myeloid IL-2 to intestinal homeostasis in vivo is unknown. To address this issue, we used an Il2 CD11c knock-out mouse model 5 to monitor the effects of myeloid Il2 deficiency. Il2 CD11c mice were initially indistinguishable from Il2 fl/fl mice, but from~5 weeks of age they exhibited significant weight loss, resulting in~40-45% lower body mass compared to Il2 fl/fl mice by 22 weeks (Fig. 6a). Enlargement of the MLN and macroscopic thickening and shortening of the colon without splenomegaly was observed in 7-10-week-old Il2 CD11c mice compared to Il2 fl/fl mice (Fig. 6b). Histopathological analysis of colons from 8-week-old Il2 CD11c mice revealed moderate leukocyte infiltration of the mucosa and sub-mucosa, which affected 50-100% of the colon surface (Fig. 6c); this pathology increased in severity with age, until systemic inflammation (high serum levels of CRP and IgG1) had developed by 14-16 weeks (Fig. 6d). Consistently, Il2 CD11c mice also exhibited substantially higher numbers of CD45 + leukocytes and total, antigen-experienced, and IFNγ-producing or IL-17-producing CD4 + T cells (Fig. 6e), and a markedly smaller population of FoxP3 + Treg cells in colons and the MLN (but not thymus and spleen), compared to Il2 fl/fl mice (Fig. 6f). MLN CD4 + T cells isolated from Il2 CD11c   mice robustly expanded with age ( Supplementary Fig. 11a) and produced more IFNγ and IL-17, while IL-2 release was normal when cells were re-stimulated ex vivo with antiCD3/CD28 antibodies ( Supplementary Fig. 11b).
To address the contribution of myeloid IL-2 in LP during differentiation of naïve CD4 + T cells in vivo, syngeneic naïve T helper cells from wild-type donor mice were adoptively transferred into Il2 CD11c and Il2 fl/fl , Rag-deficient mice (Rag2-KO Il2 CD11c and Rag2 KO Il2 fl/fl , respectively), alone or in combination with wild-type Treg cells 16 . Loss of myeloid IL-2 in the absence of lymphocytes did not affect the LP composition of colonic immune cells (Supplementary Fig. 12). By contrast, Rag2 KO Il2 CD11c mice receiving naïve CD4 + T cells showed an early and severe onset of wasting disease characterized by substantial weight loss (Fig. 7a), shortening of colon (Fig. 7b), mucosal inflammation (Fig. 7c), and an increase in activated CD4 + T cells (Fig. 7d) secreting higher levels of IL-17 and IFNγ (Fig. 7e) compared to Rag2 KO Il2 fl/fl control mice. Co-transfer of Treg cells prevented colitis by reducing the expansion of activated IFNγ-producing and IL-17-producing CD4 + T cells (Fig. 7a-e). After the induction of T-cell-mediated experimental colitis, IL-2 loss in CD11c high MHCII + cells was also associated with increased signs of skin and joint inflammation, which are the most common extra-intestinal manifestations of IBD in humans ( Supplementary Fig. 13) 21 . These results indicate that a deficiency in myeloid IL-2 is necessary and sufficient to trigger spontaneous autoimmune colitis characterized by spontaneous and chronic mucosal damage reminiscent of Crohn's disease 22,23 .
Treg cells are essential to intestinal homeostasis as they monitor effector T helper-cell activation 17 . As we observed a consistent reduction in Treg cells in both Cnb1 CD11c and Il2 CD11c mice in the MLN and LP, but not in the thymus or spleen (Figs. 3f and 6f), we examined whether myeloid IL-2 is important for the induction of antigen-specific Treg cells in the LP. Naïve CD4 + T cells from OTII mice were adoptively transferred into Rag2 KO Il2 CD11c and Rag2 KO Il2 fl/fl mice, and the mice were then fed with ovalbumin (OVA) for 5 days. Conversion of naïve OTII CD4 + T cells into FoxP3 + Treg cells was significantly impaired, while Th1 differentiation (and to a lesser extent Th17) was significantly enhanced in the MLN of Rag2 KO Il2 CD11c mice compared to controls (Fig. 8a); similar results were obtained in Cnb1 CD11c mice (Fig. 8b). These data indicate that myeloid IL-2 regulates Treg cell induction, which can directly suppress effector T helper cell expansion and activity.

DC and T cell IL-2 have distinct functions in gut homeostasis.
To dissect the contributions of myeloid cell-derived IL-2 and T helper cell-derived IL-2 to intestinal immune homeostasis, we compared the systemic and intestinal phenotypes of Il2 CD11c , Il2 CD4 (generated by crossing Il2 fl/fl and CD4-cre mice), and Il2 KO (carrying IL-2 deletion in myeloid and CD4 + T cells) mice. Il2 deletion in the CD4 + T-cell compartment resulted in characteristic IL-2 Deficiency Syndrome 24-26 originally described in Il2 KO mice and characterized by severe anemia (Fig. 9a), splenomegaly (Fig. 9b), uncontrolled expansion of splenic T cells associated with reduction of Treg cells and loss of B cells. None of these features were observed in Il2 CD11c mice (Fig. 9b), indicating that IL-2 Deficiency Syndrome is mostly driven by the loss of "adaptive" IL-2. The intestinal phenotypes of Il2 CD11c mice were even more divergent from Il2 CD4 mice: although both Il2 CD11c and Il2 CD4 mice developed colitis at 10-13 weeks, the intestinal pathology was exacerbated in Il2 CD11c mice compared to Il2 CD4 mice, and more similar to that of Il2 KO mice (Fig. 9c). A higher increase in total leukocytes, total and activated CD4 + T cells, and IL-17-producing CD4 + T cells was observed in the LP-colon of Il2 CD11c mice compared to Il2 CD4 mice, and was similar to that seen in Il2 KO mice, whereas accumulation of CD8 + T cells was specific to Il2 KO mice (Fig. 9d, e). Similar to the LP-colon of Il2 KO mice, frequencies of Treg cells were reduced in the LP-colon of Il2 CD11c and to a lesser extent in Il2 CD4 mice compared to Il2 fl/fl mice (Fig. 9f). These results indicate that innate (myeloid cells) and adaptive (CD4 + T cells) IL-2 exert both distinct and overlapping functions at the systemic level and in the intestine.

Discussion
This study shows that activation of the calcineurin-NFAT pathway in CD11c high MHCII + cells (representing the intestinal DC pool) prevents spontaneous intestinal inflammation in vivo by orchestrating a balance between inflammatory and regulatory responses in the colonic mucosa. By characterizing mice in which Cnb1 gene was deleted either in CD11c high MHCII + APCs (mostly representing the DC pool) but not CD64 + macrophages, or LysM-expressing cells (mainly macrophages, granulocytes, and monocytes), we identified two distinct roles for the calcineurin-NFAT pathway: calcineurin activation has an antiinflammatory function in intestinal CD11c + MHCII high DCs, but a pro-inflammatory role in macrophages, granulocytes, and monocytes. Accordingly, Cnb1 CD11c mice developed spontaneous intestinal inflammation, whereas Cnb1 LyzM mice were unaffected by targeted Cnb1 deletion, and were protected from induced colitis. We verified that spontaneous chronic colitis in Cnb1 CD11c mice was not due to anomalous clonal composition of the myeloid compartment, as the DC and macrophage populations were normally represented in the LP-colon and no overt differences in co-stimulatory markers expressed by these cells were found in our model. The CD4 + T-cell compartment of Cnb1 CD11c mice was also normal: T helper cells (both naïve and memory) and Treg cells expressed normal calcineurin B levels, were able to produce effector cytokines (including IL-2, IFNγ, and IL-10) at normal levels, and naïve T helper cells isolated from these mice were able to induce severe colitis when adoptively transferred into lymphopenic mice.
We also identified the importance of calcineurin-NFATmediated production of IL-2 by the intestinal DC pool in regulating intestinal homeostasis. The colonic CD103 + DC subset is the main producer of IL-2, whereas CD64 + F4/80 +/− macrophages do not produce this cytokine. Colonic DCs from Cnb1 CD11c mice still retain the ability to release IL-6, IL-12, IL-23. IL-2 deletion in CD11c high MHCII + DCs caused more severe, spontaneous intestinal inflammation in Il2 CD11c mice compared to Cnb1 CD11c mice, which was characterized by expansion of activated CD4 + effector T cells and substantial reduction of Treg cells. IL-2 complex treatment and DCs transfer results support that reduced levels of DC-derived IL-2 could be the primary cause triggering the dysregulation of intestinal CD4 + T cells observed in Cnb1 CD11c mice.
Spontaneous colitis in Cnb1 CD11c mice was milder compared to Il2 CD11c mice, suggesting that other signaling pathways that trigger DC-IL-2 production may exist. Using pharmacological inhibitors, we found that IL-2 release also depends on NFκB in response to LPS, zymosan, and thapsigargin stimulation, and on p38 MAPK in response to zymosan and thapsigargin stimulation. A recent study reported that deletion of TRAF6, a component of the TLR-NFκB pathway, is associated with~50% reduction of DC-derived IL-2 27 . Our study identifies the calcineurin-NFAT pathway as an additional axis that operates in parallel with the TRAF6-NFκB pathway in myeloid cells to produce IL-2, suggesting that both NFκB and NFAT cooperate to activate IL-2 transcription. Although these pathways contribute to the overall production of IL-2, the nature of the intestinal immunepathologies observed in the Traf6-DC and Cnb1-DC-deficient mice diverge: loss of Traf6 in DCs causes reduced inflammatory cytokines secretion, including IL-12, whereas Cnb1-deficient DCs release normal levels of inflammatory cytokines. Moreover, partial loss of IL-2 in Traf6-deficient DCs causes spontaneous Th2mediated enteritis characterized by eosinophilic infiltration and decreased Treg-cell numbers specifically in the small intestine 27 . This phenotype markedly differs from Cnb1 CD11c mice, as these mice develop a spontaneous form of enteritis in both the small and large intestine resembling Crohn's disease, and is associated with Th1 and Th17 colitogenic T-cell expansion, mucosal accumulation of mononuclear cells and progressive colon thickening 28 . Through a series of in vivo experiments, our data suggest that the calcineurin-NFAT-IL-2 pathway in myeloid cells is crucial for the induction of Treg cells, which contributes to maintaining immune tolerance in the gut by preventing autoimmunity. Indeed, Il2 CD11c and lymphopenic Rag2 KO Il2 CD11c mice receiving normal naïve T cells developed an exacerbated form of spontaneous colitis associated with extra-intestinal manifestations, including cutaneous and articular inflammatory reactions, which are the foremost comorbidities of IBD 21 .
To date, the regulatory and effector functions of the calcineurin-NFAT-IL-2 axis in orchestrating intestinal immune homeostasis has been largely studied in terms of T helper-cell and Treg-cell biology 6 . The current literature supports that IL-2 is pivotal for maintaining tolerance to self and microbial antigens 8,9,18,22 , which is achieved via various mechanisms. First, CD4 + naïve T cells constitutively produce IL-2 in the intestinal mucosa, which induces and maintains the intestinal Treg-cell pool, as these cells are uniquely dependent upon IL-2 29 . Second, IL-2 can directly suppress the differentiation of pathogenic Th17 cells 30 . Although T-cell-derived IL-2 likely has an important role in promoting intestinal tolerance, our data also highlight the requirement for DC-derived IL-2 in this process. Of note, in the absence of ex vivo restimulation, the detected IL-2 signals in intestinal DCs are relatively weak. However, it is important to bear in mind that levels of IL-2 and intensity of the signal cannot be compared to the IL-2 levels in T cells. Like DCs, unstimulated T cells do have weak IL-2 transcription/production levels, which is dramatically boosted upon stimulation. In this scenario, low level IL-2 production by DCs seems to have an important biological function and can be directly used by naïve CD4 + T cells to differentiate into Treg cells thereby suppressing the differentiation of pathogenic Th17 cells. It is also possible that some regulatory feedback mechanisms might be in place in DCs to limit and fine tune the level of IL-2 expression, as very high levels could disrupt optimal function of these cells, as well as affecting their interaction with other cells such as CD4 + T cells. Using various in vivo models, we have shown that DC-derived IL-2 in the gastrointestinal tract plays a crucial biological role in maintaining the immune homeostasis. Thus, we are confident that the immune phenotype observed in our DC-IL-2-deficient mouse model was mostly due to reduction of IL-2 production by DCs.
We further analyzed the distinct contributions of innate and adaptive IL-2 to intestinal immune homeostasis. Here, we compared the systemic and intestinal phenotype of Il2 CD11c mice (lacking Il2 in the myeloid compartment), Il2 CD4 mice (lacking Il2 in the CD4 + T cells), and Il2 KO mice (in which both myeloid and lymphoid Il2 is absent). Il2 CD4 mice exhibited IL-2 Deficiency Syndrome 24-26 characterized by severe anemia, splenomegaly, uncontrolled expansion of T cells associated with a reduction of Treg cells, and loss of B cells. The same phenotypes noted in the Il2 KO model. Mice deficient in myeloid IL-2 (Il2 CD11c mice) did not develop the same pathology, but instead showed a transmural infiltration of mononuclear cells, without affecting CD8 + T cells compartment involved in erosive mucosal lesion observed in Il2 KO mice, indicating that lymphoid IL-2 is crucial for immune homeostasis in the peripheral immune tissues. The difference between Il2 CD4 and Il2 CD11c is even more striking in the intestine, where loss of myeloid IL-2 resulted in a higher exacerbation of spontaneous colitis and a higher increase in the total number of leukocytes and activated CD4 + T effector cells producing IFNγ and IL-17 compared to mice lacking T-cell-produced IL-2. A significant reduction in the frequencies of Treg cells in the LPcolon of both Il2 CD11c and Il2 CD4 mice was observed, indicating that both myeloid and adaptive IL-2 are equally important for the induction and/or maintenance of Treg cells in the mouse intestine 20,31 . Of note, the aberrant Treg:T effector-cell ratio that accompanied pathologic inflammation in Cnb1 CD11c and Il2 CD11c mice seems to be restricted to the intestine of these mice, as we did not detect any changes in Treg-cell frequency in the lung or axillary and popliteal lymph nodes, suggesting a fundamental role for the gut microbiota in the activation of this pathway and thus IL-2 production by intestinal myeloid cells.
The results presented here pose interesting questions for future research. It would now be interesting to investigate how DCderived IL-2 mechanistically regulates the induction of intestinal Treg cells, as well as how it modulates effector T-cell functions in the mouse intestine. Moreover, the requirement of DC-derived IL-2 for the development and maintenance of innate cells, such as innate lymphoid cells, should be examined. Considering the differential effects of calcineurin signaling in various myeloid cell subsets, it will be important to investigate the effects of calcineurin inhibitors on different cell subsets in vivo. Calcineurin inhibitors are already in clinical use and elicit some benefits in patients with selected forms of IBD, including acute severe ulcerative colitis and steroid-refractory IBD. Data remain insufficient to recommend the use of calcineurin inhibitors (cyclosporin A or FK506) in Crohn's disease 32 .
In summary, our data support that the calcineurin-NFAT-IL-2 pathway in intestinal DCs has a central role in eliciting proper Tcell responses in the mouse intestine under both steady-state and inflammatory conditions, by supporting Treg-cell induction and by suppressing excessive expansion of Th1/Th17 cells (directly or indirectly). We propose that therapeutic strategies that selectively boost DC-derived IL-2 production may have a beneficial effect in patients with IBD.
All the experiments described in this study were allowed and carried out in accordance with the guidelines of the Institutional Animal Care and Use Committee of the Biological Resource Centre (A*STAR, Singapore). B6.TgIL-2-GFP mice 20 were maintained in the animal facility at the Institut Pasteur (Paris, France) in accordance with European animal welfare regulations. These animal studies were approved by the Institut Pasteur Safety Committee and by the Paris Ethics Committee 1.
Mouse models of colitis. Mice were exposed to 150 µl TNBS (Sigma-Aldrich) presensitization solution (1% TNBS dissolved in 4:1 acetone:olive oil solution) that was applied to a shaved area of abdominal skin. Control mice were subjected to the same protocol with pre-sensitization solution lacking TNBS. After 7 days, mice were anesthetized by intraperitoneal ketamine/xylazine injection (100 µl/10 g body weight of 100 mg/ml ketamine and 20 mg/ml xylazine in saline) and then administered 1 mg TNBS (dissolved in 50% ethanol) intra-rectally using a 3.5 French catheter equipped with a 1 ml syringe. To ensure even distribution of TNBS throughout the entire colon and cecum, mice were held in an inverted vertical position for 30 s after instillation. In some experiments to induce colitis, naproxen (40 mg/kg/day) was administrated for 4 weeks via drinking water. The mice were then monitored daily for weight loss and fecal score. The criteria and numerical scores for macroscopic damage were as follows: 0, no ulcer, no inflammation; 1, no ulcer, local hyperemia; 2, colon wall thickening/edema; 3, ulceration and inflammation at one site only; 4, two or more sites of ulceration and inflammation. Additional points were scored for the presence of indurations, intestinal adhesions, and each additional site of inflammation or ulceration >1 cm 2 in surface area.
T-cell transfer model of chronic colitis. Splenocytes were collected from C57BL/6 mice aged 6-10 weeks, and naïve CD4 + CD45RB high T cells and CD4 + CD45RB low CD25 + Treg cells were sorted by fluorescence-activated cell sorting (FACS). Naïve T cells (3×10 5 /mouse) were injected intravenously alone or in combination with Treg cells (1×10 5 /mouse) into Rag2 KO Il2 CD11c and Rag2 KO Il2 fl/fl recipient mice. Mouse weight was recorded to evaluate irritable bowel disease. Mice were then killed by CO 2 asphyxiation at 110 or 140 days after adoptive cell transfer for macroscopic analysis, histopathological scoring and flow cytometric analyses of leukocytes.
Intestinal permeability assay. Mice were fasted for 6 h and then administered 500 mg/kg FITC-dextran (4 kDa; Sigma-Aldrich) by oral gavage. Blood samples were collected 1 and 3 h later, centrifuged at 12,000 x g at 4°C for 3 min, and the plasma was collected for analysis. The plasma was diluted in an equal volume of PBS and then analyzed for FITC-dextran concentration using a fluorescence spectrophotometer (485 nm excitation, 535 nm emission). A standard curve was obtained by diluting FITC-dextran in untreated plasma diluted with PBS (1:3 v/v).

Measurement of tissue inflammatory markers.
Frozen fragments of small intestine and colon were homogenized using Tissue Protein Extraction Reagent (Thermo Scientific) and the homogenates were used to assess protein concentrations and cytokine levels. Neutrophil infiltration into the colon was monitored by measuring MPO activity using a spectrophotometric assay with trimethylbenzidine (TMB) as the substrate and myeloperoxidase activity expressed as mU/mg of protein.
DC transfer in vivo. Bone marrow cells from C57BL/6 CD45.1 or Il2 KO mice were cultured in conditioned media containing recombinant mouse granulocytes macrophages colony-stimulating factor (GM-CSF) or recombinant mouse Flt3-L for 11 days. CD103 + and CD11b + DCs (CD11c + MHCII high ) were FACS-sorted and adoptively transferred intravenously into 12-week-old Cnb1 fl/fl or Cnb1 CD11c mice at a ratio of 1:6 CD103:CD11b (total number of transferred cells: 5×10 5 ). Two weeks post-transfer, mice were killed and the immune phenotype of mononuclear cells from spleen, SI, and colon was investigated by flow cytometry.
Isolation of LP leukocytes from the colon. Isolated colons were cut longitudinally and divided into segments of 0.5-1 cm in length. The tissue segments pieces were then added to RPMI medium supplemented with 2% FBS, and incubated for 20 min at 37 o C with constant stirring. After incubation, the resultant suspension was passed through a 70 µM sterile strainer and pieces of colon tissue were placed into a 50 ml tube containing 15 ml serum-free RPMI, shaken vigorously for 30 s, and then filtered twice through a 70 µM cell strainer. The remaining fragments of colon were washed in calcium-free and magnesium-free HBSS and treated with 1 mM EDTA in PBS (twice for 20 min) to remove the epithelium. The tissue was then digested with type-IV collagenase (0.8 mg/ml; Sigma-Aldrich) containing DNase and the leukocytes were enriched on a 40:75 Percoll gradient (Pharmacia) and the interface collected after centrifugation at 700 × g for 20 min.
Isolation of LP leukocytes from the small intestine. Isolated SI was flushed with HBSS, the PP was removed, and then were cut longitudinally and divided into pieces of 1 cm in length. Epithelial cells were removed by incubation with 2 mM EDTA in HBSS supplemented with 10% FCS for 40 min at 37°C, followed by vigorous shaking for 20 s. The samples were then collected and incubated with type-VIII collagenase (0.1 mg/ml, Sigma-Aldrich) and DNase I (50 µg/ml) in HBSS containing 10% FCS and 10 mM HEPES for 45 min at 37°C. After digestion, the samples were shaken vigorously for 10 s, the supernatants were collected by filtration through a nylon mesh, and the tissue underwent a second round of enzymatic digestion. Leukocytes were enriched on a 40:75 Percoll gradient and the interface was collected after centrifugation at 700 × g for 20 min.