Crucial roles of XCR1-expressing dendritic cells and the XCR1-XCL1 chemokine axis in intestinal immune homeostasis

Intestinal immune homeostasis requires dynamic crosstalk between innate and adaptive immune cells. Dendritic cells (DCs) exist as multiple phenotypically and functionally distinct sub-populations within tissues, where they initiate immune responses and promote homeostasis. In the gut, there exists a minor DC subset defined as CD103+CD11b− that also expresses the chemokine receptor XCR1. In other tissues, XCR1+ DCs cross-present antigen and contribute to immunity against viruses and cancer, however the roles of XCR1+ DCs and XCR1 in the intestine are unknown. We showed that mice lacking XCR1+ DCs are specifically deficient in intraepithelial and lamina propria (LP) T cell populations, with remaining T cells exhibiting an atypical phenotype and being prone to death, and are also more susceptible to chemically-induced colitis. Mice deficient in either XCR1 or its ligand, XCL1, similarly possess diminished intestinal T cell populations, and an accumulation of XCR1+ DCs in the gut. Combined with transcriptome and surface marker expression analysis, these observations lead us to hypothesise that T cell-derived XCL1 facilitates intestinal XCR1+ DC activation and migration, and that XCR1+ DCs in turn provide support for T cell survival and function. Thus XCR1+ DCs and the XCR1/XCL1 chemokine axis have previously-unappreciated roles in intestinal immune homeostasis.

Intestinal immune homeostasis is regulated by a variety of innate and adaptive immune cells 1 , including dendritic cells (DCs) 2,3 . The DC population is heterogeneous, consisting of several subsets with specific functions, distinguished by characteristic patterns of surface marker expression 4,5 . In mice, CD103 + CD11b − and CD103 − CD11b + DC subsets are present in the spleen and lymph nodes (LNs), while in the lamina propria (LP) of the intestine, an additional CD103 + CD11b + subset exists 4,5 . The roles of the various intestinal DC subsets are variously well-studied: CD103 + CD11b + DCs are the most abundant and are involved in generating Th17 and regulatory T (Treg) cells [6][7][8][9] , and anti-fungal immunity 10 ; CD103 − CD11b + DCs are related to macrophages 2 and play an immunoregulatory role via the secretion of IL-10 and TGF-β 1 11,12 ; while in contrast, little is known of the role of CD103 + CD11b − DCs in this site. CD103 + CD11b − DCs also express the T cell co-receptor CD8α and the Scientific RepoRts | 6:23505 | DOI: 10.1038/srep23505 chemokine receptor XCR1 (lymphotactin receptor/G-protein-coupled receptor 5), and represent approximately 5% of murine intestinal DCs [13][14][15][16] . This subset (hereafter "XCR1 + DCs") also exists in other tissues including spleen, LNs and skin, where it specializes in the uptake of dead cells and cross-presentation of antigen to CD8 + T cells 17 , which is important for protection against viruses, bacteria and parasites, and for anti-tumour immunity 14,16,[18][19][20][21] . XCR1 + DCs are also present in other mammalian species 22,23 ; in humans, XCR1 is expressed on a DC subset that is present in various tissues, including skin and blood, and possesses high cross-presenting activity 24 . How the XCR1 + DC subset functions in either the murine or human intestine is currently unknown.
The ligand for XCR1 is XCL1 in mice and XCL1 and XCL2 in humans 25 . In both species, XCL1 is produced mainly by natural killer (NK) and activated CD8 + T cells [15][16][17][18]22,23 . Mice lacking XCR1 or XCL1 show diminished CD8 + T cell responses against the antigens cross-presented by CD103 + CD11b − DCs 15 ; XCL1 is also involved in regulating medullary accumulation of thymic XCR1 + DCs, and thymic generation of naturally-occurring regulatory T (Treg) cells 26 . Thus, the XCR1-XCL1 axis has the potential to modulate both the localization and function of T cells and DCs, though the extent to which this is relevant outside of the thymic environment is not yet clear.
In order to clarify the functions of XCR1 + DCs, we generated and analysed mutant mice in which these cells are constitutively ablated. These mice possessed significantly fewer intestinal T cells than their wildtype counterparts, and the remaining T cells exhibited an atypical phenotype. Consistent with the regulatory roles of intestinal T cells, XCR1 + DC-deficient mice showed exaggerated manifestations during chemically-induced colitis. Alongside, in mice lacking either XCR1 or XCL1, a similar decrease in T cell populations was seen, accompanied by an accumulation of CD103 + CD11b − DCs in the intestine. Thus, we have identified novel regulatory roles of XCR1 + DC and the XCR1-XCL1 axis in maintaining intestinal immune homeostasis, and have proposed a hypothetical model on which to base future studies to define the underlying mechanisms.

Results
Intestinal T cell populations are specifically decreased in mice lacking XCR1 + DCs. We exploited the specific expression of XCR1 on this DC subset to generate a mouse model in which CD103 + CD11b − XCR1 + DCs are constitutively ablated as a result of diphtheria toxin A subunit (DTA) expression (XCR1-DTA mice) ( Supplementary Fig. 1A,B). We first confirmed the ablation of CD103 + CD11b − DCs in the LP, mesenteric lymph nodes (MLNs) and spleens of XCR1-DTA mice, and the retention of the other DC subsets across these tissues ( Supplementary Fig. 1C). In the spleen, CD103 + CD11b − DCs were more severely ablated than CD8α + CD11b − DCs, which is consistent with the finding that CD8α + CD11b − DC populations contain more XCR1 − cells than CD103 + CD11b − DC populations 13,14 (Supplementary Fig. 1D).
We next asked whether T cell populations were affected by the absence of XCR1 + DCs. In thymus, spleen and MLNs, the size and composition of the T cell populations were comparable between control and XCR1-DTA mice (Supplementary Fig. 2A,B). Normally the intestine contains several distinct lymphocyte populations: within the LP, conventional TCRα β + CD4 + and CD8 + T cells predominate, while the intraepithelial lymphocyte (IEL) population additionally includes TCRγ δ + T cells and an atypical TCRα β + subset that expresses CD8α α instead of CD8α β 27 . In the LP of XCR1-DTA mice there were approximately two-thirds fewer TCRα β + T cells than in control mice, with significant decreases across both CD4 + and CD8 + subpopulations (Fig. 1A,B); all IEL subsets were also significantly reduced compared to controls (Fig. 1C,D). Moreover, taking into consideration the lower T cell numbers, there was a further relative decrease in the more mature CD4 + CD8α α + IEL subset, accompanied by a relative accumulation of the less mature CD4 + CD8 − IELs in XCR1-DTA mice (Fig. 1C). The relative scarcity of TCRβ -expressing cells was similarly evident histologically (Fig. 1E). Thus, XCR1 + DCs are critical for keeping both intestinal LP T cells and IEL populations.
Intestinal T cells are more prone to death in XCR1-DTA mice. The low numbers of intestinal T cells in XCR1-DTA mice might have been due to decreased proliferation or increased cell death. The proliferation marker Ki67 was expressed at a comparable frequency in LP T cell and IEL sub-populations from XCR1-DTA and control mice ( Supplementary Fig. 3A,B). However, there were markedly more dead and/or dying intestinal TCRα β + T cells in mice lacking XCR1 + DCs, in particular in the IEL compartment, as assessed by frequency of Annexin V and 7-amino-actinomycin D (7-AAD) staining (Fig. 1F). The findings indicate that the T cells present in XCR1-DTA mice were not defective in proliferation, but were more prone to undergo cell death, likely contributing to the T cell deficiency observed in these mice. Thus, the absence of XCR1 + DCs compromises the survival of intestinal T cell populations.

LP T cells exhibit an abnormal phenotype in XCR1-DTA mice.
We next compared the phenotype of the lymphoid tissue T cells from control and XCR1-DTA mice. In control mice, LP TCRα β + T cells expressed lower levels of the cell adhesion molecule CD62L and higher levels of the integrin CD103, compared with splenic and MLN TCRα β + T cells ( Fig. 2A, shaded histograms). However, LP TCRα β + T cells from XCR1-DTA mice showed defects in this intestinal T cell phenotype ( Fig. 2A). Expression of the gut-homing receptors α 4β 7 integrin and CCR9 28,29 was comparable between control and XCR1-DTA mice ( Fig. 2A).
We then asked whether LP T cells from XCR1-DTA and control mice further differed at the transcriptional level. LP CD4 + T cells from mice lacking XCR1 + DCs contained significantly more transcripts for Il17a, Il22 and Il4 and significantly fewer transcripts for Ifng and Il10 than those from control mice (Fig. 2B,C), suggesting polarization towards a Th2/17 profile, rather than the Th1 profile evident in control animals. Furthermore, it is notable that Runx3 expression in CD4 + T cells from XCR1-DTA mice was decreased compared with that of CD4 + T cells from control mice. A transcription factor, RUNX3, is required for transition of CD4 + CD8 − to CD4 + CD8α α + T cells 30,31 . Therefore, decrease of Runx3 expression might explain the observed block in the generation of CD4 + CD8α α + from CD4 + CD8 − T cells in XCR1-DTA mice (Fig. 1C). Thus, gene expression profile analysis revealed that LP T cell surface phenotype and gene expression profile is profoundly altered in mice lacking XCR1 + DCs.
Given the known role of XCL1 in the thymic generation of natural Treg cells 26 we next investigated whether the intestinal Treg population was affected in XCR1-DTA mice. The proportion of Treg cells, defined as CD25 + Foxp3 + , among CD4 + T cells, and the expression of the natural Treg marker, Nrp1 32 , were comparable  Fig. 4C), but the remaining cells existed at the expected frequency and exhibited a normal phenotype. Therefore, in contrast to other LP T cell populations, the generation and maintenance of Tregs is not specifically affected in mice lacking XCR1 + DCs.

XCR1-DTA mice are more susceptible to DSS-induced colitis.
While mice lacking all CD11c + DCs exhibit aberrant myeloid cell proliferation and autoimmunity 33 , the absence of CD103 + CD11b − DCs alone does not induce any overt signs of dysregulated immunity under steady-state conditions 34 . However, during dextran sodium sulphate (DSS) -induced colitis, while mice lacking CD103 + CD11b − DCs through deficiency in the transcription factor BATF3 (Basic leucine zipper transcription factor ATF-like 3) (BATF3) are as susceptible as their wild-type counterparts 34 , induced ablation of CD103 + CD11b − DCs in Clec9A-diphtheria toxin receptor (DTR) mice resulted in significant exacerbation of colitis 35 . As XCR1-DTA mice also did not show clear signs of dysregulated intestinal immunity/intestinal inflammation in the steady state, we assessed their susceptibility to DSS-induced colitis. After administration of DSS, XCR1-DTA mice lost significantly more weight, achieved significantly higher disease activity scores and underwent significantly more colon shortening compared to control mice (Fig. 3). Thus, while the marked differences in the intestinal immune compartment in XCR1-DTA mice do not result in inflammatory changes under homeostatic conditions, these animals are rendered exquisitely sensitive to induced intestinal inflammation.

Mice lacking XCR1 or XCL1 exhibit similar defects in intestinal T cell populations as XCR1-DTA mice.
We then asked whether it was possible that the XCR1-XCL1 axis itself was contributing to the observed defects in mice lacking XCR1 + DCs. We first generated XCL1-deficient mice ( Supplementary Fig. 5A,B), and found that the size and composition of the T cell population in spleen and MLNs was comparable with control mice ( Supplementary Fig. 6A,B). However, similar to XCR1-DTA mice, XCL1-deficient mice possessed significantly fewer TCRα β + T cells in their LP and intraepithelial compartment ( Fig. 4A-D). All the LP and intraepithelial TCRα β + T cell subsets, but not TCRγ δ + T cells, were significantly decreased in XCL1-deficient mice compared to controls. Relative proportions of CD4 + CD8 − and CD4 + CD8αα + T cells were also increased and decreased, respectively within the IEL population (Fig. 4C), as in XCR1-DTA mice (Fig. 1C).
We then analysed XCR1-deficient mice 16 . These animals also possessed significantly fewer LP and intraepithelial TCRα β + T cells than controls (Fig. 4E-H), while splenic and MLN T cell populations were comparable between control and mutant mice ( Supplementary Fig. 6C,D). Furthermore, all LP and intraepithelial TCRα β + T cell subsets, but not TCRγ δ + T cells, were decreased in XCR1-deficient mice.
In summary, these data show that mice lacking either XCL1 or XCR1 exhibit similar specific defects in intestinal T cell populations as XCR1-DC-deficient mice. Therefore it is possible that the XCR1-XCL1 axis itself is directly involved in intestinal T cell homeostasis, via a mechanism that might either involve interaction with XCR1 + DCs or be independent of this pathway.

XCR1 deficiency alters the gene expression profile of intestinal DCs.
To further dissect the direct effects of the XCR1-XCL1 axis on DCs in the gut, we compared the gene expression profiles of LP DC subsets from control and XCR1-deficient mice. The mutant CD103 + CD11b − DCs contained significantly fewer transcripts for 62 of the 547 immune-related genes measured than did control CD103 + CD11b − DCs. The most highly differentially-expressed genes within this subset were Xcr1, Ccr7, Cd40, Il12b, Ccl22 and Il6 (Fig. 5A). Ccr7 encodes the chemokine receptor, CCR7, which is required for DCs to migrate from the peripheral tissues to the draining LNs 36 ; thus low expression of Ccr7 might impede emigration from the LP and thus result in abnormal patterns of DC distribution. To assess the biological significance of this finding, we measured the relative abundance of CD103 + CD11b − DCs in the LP and MLNs of mice lacking either XCL1 or XCR1, and in control animals. As would be expected if DCs expressed insufficient CCR7 for effective migration to the LN, in both XCL1-and XCR1-deficient mice, the CD103 + CD11b − DC population was significantly smaller in the MLN and significantly larger in the LP, compared to control mice ( Fig. 5B-E). Meanwhile, the composition of the splenic DC compartment was unaffected by the absence of XCL1 or XCR1 ( Supplementary Fig. 7A-D). Thus it appears that the XCR1-XCL1 axis is required for optimal gene expression to support normal migration of intestinal XCR1 + DCs to the MLN.
As XCL1 is known to be produced by NK and CD8 T cells in other tissues [15][16][17][18]22,23 , and given the observed importance of XCL1 for optimal intestinal DC function, we then asked which cells within the intestinal compartment express Xcl1 and might therefore act as a source of the chemokine in vivo. Compared with splenic T cells, LP and intraepithelial T cell subsets from wildtype mice showed high expression of Xcl1 (Fig. 6A). Interestingly, we saw that Xcl1 expression in LP T cells of XCR1-DTA mice was decreased to approximately 20% of that in control mice (Fig. 2B), which might indicate that the high expression of Xcl1 in intestinal T cells is supported by the presence of XCR1 + DC. Immunofluorescence imaging of intestinal sections confirmed that XCR1 + DCs and T cells are closely associated in vivo (Fig. 6B). Furthermore, in similar to XCL1-or XCR1-deficient mice, RAG2-deficient mice, which lack any mature lymphocytes, showed the increase and decrease of CD103 + CD11b − DCs in the LP and MLN, respectively (Fig. 6C). Taken together, these data suggest the possibility that intestinal T cells might provide XCR1 + DCs with XCL1 while closely associated in the gut.

Discussion
Here we have generated and analysed XCR1-DTA mice which lack the CD103 + CD11b − XCR1 + DC subset with the aim of understanding the role of this DC subset. While these mice did not exhibit any overt signs of pathology as a result of the absence of XCR1 + DCs, all LP T cell and IEL populations were specifically and significantly decreased, while the abundance of T lineage cells in thymus, spleen and MLNs was not affected. Thus the presence of XCR1 + DCs is necessary for normal numbers of T cells in the intestinal LP and IEL compartments.
In contrast to the absence of pathology in the steady-state, we found that the defects induced by the absence of XCR1 + DCs rendered XCR1-DTA mice significantly more susceptible to induced colitis compared with control mice. While the pathology of DSS colitis does not require the presence of T cells 37 , the condition is exaggerated in various mutant mice in which IELs, or in particular CD4 + CD8α α + T cells, are depleted 27,[38][39][40][41][42] . Exacerbation of colitis in XCR1-DTA mice, with their profoundly diminished intestinal T cell populations, is consistent with these reports. Thus, while the absence of XCR1 + DCs does not cause inflammation per se, the associated quantitative and/or qualitative changes in the intestinal T cell compartment may reduce the ability of XCR1-DTA mice to control the response to DSS-induced gut inflammation.
Similar to our findings, mice lacking CD103 + CD11b − DCs do not exhibit signs of spontaneous inflammation in the steady state 34,35 , but their susceptibility to DSS-induced colitis seems to vary depending on additional factors: models using deficiency in the transcription factor BATF3 to induce CD103 + CD11b − DC deficiency   are as susceptible as their wild-type counterparts 34 , while employing diphtheria-toxin-mediated ablation of CD103 + CD11b − DCs in Clec9A-DTR mice renders them significantly more sensitive to induced colitis 35 . Alongside the difference in DC ablation method, susceptibility to DSS-induced colitis can be affected by differences in the commensal gut microbiota, mouse genetic background and source and dose of the DSS used. For example, Edelson et al. used 129S6/SvEv mice with 5% DSS from TDB Consultancy 34 , whereas we used C57BL/6 mice with 1.5% DSS from MP Biomedicals. Furthermore, it cannot be formally excluded that CD103 + CD11b − DCs are compensatory generated during the course of DSS-induced colitis in BATF3-deficient mice 43 . Meanwhile, our results are consistent with enhanced manifestations of DSS-induced colitis in Clec9A-DTR mice, in which CD103 + CD11b − DCs are inducibly ablated 35 . Muzaki et al. used BALB/c Clec9A-DTR mice treated with 2% DSS from MP Biomedicals and showed that CD103 + CD11b − DCs were critical for anti-inflammatory responses, dependent on T cell-derived IFN-γ ; although intestinal T cell populations were not investigated in their study, we also observed decreased Ifng expression in the remaining LP CD4 + T cells from XCR1-DTA mice prior to the induction of colitis (Fig. 2C).
In addition to the depletion of T cell populations and an inability to restrain induced inflammation, we detected further abnormalities in the intestinal T cell compartment of mice lacking XCR1 + DCs. Intestinal T cells normally express lower levels of CD62L and higher levels of CD103 than T cells in other tissues, however LP T cells from XCR1-DTA mice showed defects in this typical intestinal phenotype. Decreased expression of CD62L is indicative of the constitutive activation of intestinal T cells, most likely by antigens from the intestinal lumen, such as those of the commensal bacteria. Indeed, germ-free or antibiotic-treated mice also show decreased populations of intestinal T cells including IELs 44,45 . Therefore, the failure to downregulate CD62L of LP T cells from XCR1-DTA mice indicates a deficiency somewhere along the pathway leading to activation of these T cells, or in their ability to respond to activating stimuli. It is unknown whether XCR1 + DCs directly incorporate and present antigens in the gut, though there is evidence that CX3CR1 + macrophages can capture and transfer antigens to CD103 + DCs 46 . It will be intriguing to investigate further how the absence of XCR1 + DCs is leading to the emergence of a diminished T cell population with atypical activation characteristics.
The upregulated expression of CD103 by intestinal T cells is driven by the local cytokine milieu, and particularly by TGF-β 47 . The absence of either CD103 or functional TGF-β signalling in T cells results in impaired development of intestinal T cell populations 48,49 , as seen in XCR1-DTA mice. TGF-β is normally produced by intestinal DCs, including XCR1 + DCs 11 , therefore we speculate that the lack of effective CD103 upregulation by intestinal T cells in XCR1-DTA mice might be caused by an absence of XCR1 + DC-derived TGF-β , which could also contribute to the decrease in T cell populations.
Intestinal T cells in XCR1-DTA mice had poor survival capacity and a tendency to express genes related to Th2 or Th17 differentiation, rather than Th1. It is unclear how the absence of XCR1 + DC is linked to this defect, however BATF3-dependent DCs are known to be involved in Th1 cell differentiation 20 , so it is possible that XCR1 + DCs might also drive T cell polarization. Meanwhile, Treg cells were numerically reduced in line with the overall decrease in T cell populations, but we did not detect any abnormalities in their phenotype in XCR1-DTA mice. Similarly, BATF3-deficient mice maintained a normal population of Treg cells in their LP and MLNs 34 .
Absence of XCR1 + DCs also led to further relative decrease in CD4 + CD8α α + IELs, accompanied by a relative accumulation of the CD4 + CD8 − T cells in XCR1-DTA mice. The generation requires certain MHC class II-expressing antigen presenting cells and Runx3 expression in CD4 + CD8 − T cells 30,31 . Because Runx3 expression in CD4 + CD8 − T cells was decreased in XCR1-DTA mice, we assume that XCR1 + DCs are responsible for the generation of CD4 + CD8α α + T cells by keeping Runx3 expression levels. Runx3 expression is driven by TGF-β and retinoic acids, which are generated by aldehyde dehydrogenase from vitamin A 30,31 . XCR1 + DCs express both Tgf-b1 and aldh1a2 11 . Although further studies are necessary to clarify whether the decrease of CD4 + CD8α α + T cells in XCR1-DTA mice is caused by developmental block and/or impaired expansion, it is possible that XCR1 + DCs are required for the generation of CD4 + CD8α α + T cells by providing TGF-β and/or retinoic acids.
While the removal of XCR1 + DCs clearly induces profound changes in the intestinal immune compartment, the question remains: to what extent is the XCR1-XCL1 axis involved, or are the observed effects more likely to be due to other, XCR1-XCL1-independent, effects? While this study was not designed to answer this question specifically, we found that mice lacking XCR1 or XCL1 recapitulated important features of the T cell insufficiencies and imbalances present in XCR1-DTA mice. Moreover, intestinal T cells from wild type mice abundantly expressed Xcl1 and were observed in close association with XCR1 + DCs in tissue sections, while LP T cells from XCR1-DTA mice expressed significantly lower levels of Xcl1 than those from control mice (Fig. 2B). Thus it is possible that intestinal T cells provide XCL1 to local XCR1 + DCs. We also noted that in XCR1-and XCL1-deficient mice, CD103 + CD11b − DCs accumulated in the LP, and were reduced in number in the MLN: in XCR1-deficient mice at least, this could be due to the lower than normal level of Ccr7 expression in CD103 + CD11b − DCs, as CCR7 is required for intestinal DCs to migrate to the MLN 50 .
Taken together, our findings lead us to propose a hypothetical model of XCR1-DC-T cell crosstalk in the intestine as a basis for further study and discussion (Fig. 7): activated intestinal T cells produce XCL1, which specifically attracts nearby XCR1 + DC; the ensuing DC-T cell crosstalk supports T cell survival, promotes the typical CD103 high /CD62L low intestinal phenotype, and leads to maintenance of intraepithelial and LP T cells; continuing XCL1 expression by the T cells in turn enables DC maturation, and in particular CCR7 expression necessary for migration from the LP to the MLN.
In conclusion, our findings highlight the potential significance of the XCL1-XCR1 axis, and in particular the interaction of XCR1 + DCs and the intestinal T cell compartment, in maintaining gut homeostasis and ensuring appropriate control of inflammation. This both advances our understanding of the fundamental workings of the intestinal immune system and identifies the XCR1-XCL1 axis as a novel potential therapeutic target for the treatment of human intestinal immune disorders. Further work to identify the key molecular mediators and underlying mechanisms will be necessary to fully understand this intriguing system. Generation of mutant mice. XCR1-DTA mice were generated as follows: first, the entire murine XCR1 coding sequence was replaced with a gene encoding the Cre recombinase to generate mice in which Cre recombinase is expressed in XCR1-expressing cells ( Supplementary Fig. 1A). The Cre recombinase gene was derived from pPGK-Cre-bpA (kindly provided by Dr. Klaus Rajewsky, Max-Delbrück-Center for Molecular Medicine, Berlin, Germany) and carried a polyadenylation signal derived from the bovine growth hormone gene (bGHpA). A neomycin-resistance gene driven by the MC1 promoter and flanked by yeast FRT sequences was used as a selection marker. A herpes simplex virus thymidine kinase gene (HSV-TK) was inserted for negative selection. The C57BL/6-derived embryonic stem (ES) cell line, Bruce4 (kindly provided by Drs. Colin L. Stewart, Institute of Medical Biology, Singapore, through Masaki Hikida, Kyoto University, Japan), was transfected with the linearized targeting vector by electroporation and selected with G418 (Nacalai Tesque) and ganciclovir (Mitsubishi Tanabe Pharma). Doubly-resistant clones were screened for homologous recombination by PCR and verified by Southern blot analysis. Germline-transmitting chimeras were generated by injection of targeted ES clones into blastocysts from BALB/c mice which were bred with C57BL/6J mice. Xcr1 +/cre mice were further backcrossed with C57BL/6J mice for 2-6 generations. Xcr1 +/cre mice were crossed with R26:lacZbpA flox DTA mice (kindly provided by Dr. Dieter Riethmacher, University of Southampton, UK, through Dr. Shigekazu Nagata, Kyoto University, Japan) carrying the DTA gene, which is designed to be expressed specifically in the cells or tissues expressing Cre recombinase under the control of the ubiquitously-active Rosa26 promoter 51 . This crossing generated the XCR1-DTA mice, in which XCR1 + cells are ablated by DTA expression. Xcr1 +/cre littermates were used as controls.
Xcr1 +/venus mice were generated by knocking a gene encoding a fluorescent protein, venus, into the Xcr1 locus 16 . Xcr1 venus/venus mice were therefore rendered XCR1-deficient.
To generate XCL1-deficient mice, the targeting vector was designed to replace the murine XCL1 coding sequences with a neomycin resistance gene driven by the MC1 promoter, flanked by the yeast FRT sequence (Supplementary Fig. 5A). Bacteriophage P1 loxP sequences and the DTA gene were inserted for conditional ablation and negative selection, respectively. The targeted ES cells were generated as described above, but were selected with G418 alone. The neomycin resistance gene was removed by crossing with CAG-cre transgenic mice (kindly provided by Jun-ichi Miyazaki, Osaka University, Osaka, Japan) 52 , and the resultant Xcl1 +/− mice were then backcrossed with C57BL/6J mice for 4 generations. The resulting Xcl1 −/− mice were thus XCL1 deficient. Cell preparations. Thymocyte and splenocyte suspensions were prepared by grinding the appropriate organs through mesh filters. MLNs were digested with 400 units/ml collagenase III (Worthington) and 100 μg/ ml deoxyribonuclease (DNase) I (Sigma-Aldrich) at 37 °C for 20 min. Intestinal immune cells were prepared from the small intestine as follows: fat and Peyer's patches were dissected away before the small intestine was opened longitudinally and stirred in RPMI containing 2% FCS, 2 mM EDTA for 20 min at 37 °C. IELs were isolated from the resulting supernatant using a 40-75% Percoll density gradient (GE Healthcare) and by collecting the cells that layered between the 40 and 75% fractions. Following supernatant collection the intestinal tissue was stirred for an additional 20 min at 37 °C in RPMI containing 2% FCS, before mincing and further stirring in 400 units/ml of collagenese D (Roche Diagnostics) and 10 μg/ml DNase I for 20 min at 37 °C. Floating cells were collected and the collagenase digestion was repeated twice more. The pooled cell suspensions were then centrifuged on a 40-70% Percoll density gradient, and the cells that layered between the 40-75% fractions were collected as LP cells. Histology. For Fig. 1E, small intestines were embedded in FSC22 frozen section compound (Leica Microsystems). Five μm cryosections were labelled with rat anti-human Ep-CAM (G8.8; BioLegend) Ab and biotinylated armenian hamster anti-TCRβ (H57-597; BioLegend) Ab, and then labelling was visualised with streptavidin-Alexa Fluor 594 (Invitrogen) and Alexa Fluor 647-conjugated anti-rat IgG (H+ L) Ab (Invitrogen). For Fig. 6B, small intestines from Xcr1 +/venus mice were first fixed with 4% paraformaldehyde and then embedded in FSC22 frozen section compound. Five μm cryosections were labelled with rat anti-mouse CD4 Ab (BioLegend) and labelling visualised with goat anti-Rat IgG Ab conjugated-Alexa Fluor 594 (Invitrogen). Sections were examined using the FV10i microscope (Olympus).

Cell death and proliferation analysis. For analysis of cell death, LP cells and IELs were washed
with annexin-binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl2, pH 7.4) and incubated with fluorochrome-conjugated annexin-V and 7-AAD Viablity Staining Solution (eBioscience) for 15 min at room temperature. For analysis of cell proliferation, cells were surface-labelled with the appropriate conjugated Abs, fixed, permeabilized using Foxp3 Staining Buffer Set (eBioscience) and subsequently incubated with anti-Ki67-FITC (B56; BD Biosciences) for 30 min. After washing, cells were analysed on a FACSAria II.
Gene expression analysis. Gene expression profiles of LP CD4 + T cells (Fig. 2B) and DC subsets (Fig. 5A) were analysed with the Mouse Immunology Kit containing 547 immune-related genes and the NanoString nCounter gene expression system (NanoString Technologies, Seattle, WA). Briefly, 10,000 cells of each T cell or DC subset were lysed in RLT buffer (Qiagen, Valencia, CA). The lysates were hybridized for 16 h with the Mouse Immunology Kit and loaded into the nCounter prep station followed by quantification using the nCounter Digital Analyzer (NanoString Technologies). The nCounter data were normalized in two steps: firstly using the positive spiked-in controls provided by the nCounter instrument (NanoString Technologies), and secondly, according to the expression of 14 control genes (Alas1, Eef1g, G6pdx, Gapdh, Gusb, Hprt, Oaz1, Polr1b, Polr2a, Ppia, Rpl19, Sdha, Tbp and Tubb5).
Colitis model. Mice were allowed free access to filtered drinking water containing 1.5% DSS (MW = 36,000-50,000; MP Biomedicals) for 1 week. Mice were weighed at indicated days. Severity of colitis was assessed using the disease activity index based on presence of diarrhea and bleeding 53 . At day 10 all mice were euthanized and colons were dissected for measurement of length.
Quantitative real-time PCR. Total RNAs were obtained using the Sepasol RNA-I Super G (Nacalai tesque), reverse transcribed and analysed by an ABI PRISM 7000 (Applied Biosystems). TaqMan probes (TaqMan Gene Expression Assay; Applied Biosystems) were used for 18S rRNA (internal control). The following SYBR Green primers were used: Xcl1 (Forward 5′ -TTTGTCACCAAACGAGGACTAAA-3′ , Reverse 5′ -CCAGTCAGGGTTATCGCTGTG-3′ ). Gene expression was normalized to that of 18S rRNA and is represented as the ratio relative to the indicated reference samples. All primers were validated for linear amplification.
Statistical analysis. The data were analysed using two-tailed unpaired Student's t test with Welch's correction in cases of unequal variance. P values < 0.05 were considered statistically significant. All statistical analysis was performed using GraphPad Prism software (GraphPad).