CCR7-dependent trafficking of RORγ+ ILCs creates a unique microenvironment within mucosal draining lymph nodes

Presentation of peptide:MHCII by RORγ-expressing group 3 innate lymphoid cells (ILC3s), which are enriched within gut tissue, is required for control of CD4 T-cell responses to commensal bacteria. It is not known whether ILC populations migrate from their mucosal and peripheral sites to local draining secondary lymphoid tissues. Here we demonstrate that ILC3s reside within the interfollicular areas of mucosal draining lymph nodes, forming a distinct microenvironment not observed in peripheral lymph nodes. By photoconverting intestinal cells in Kaede mice we reveal constitutive trafficking of ILCs from the intestine to the draining mesenteric lymph nodes, which specifically for the LTi-like ILC3s was CCR7-dependent. Thus, ILC populations traffic to draining lymph nodes using different mechanisms.

S econdary lymphoid tissues, such as lymph nodes (LNs), provide a highly organized microenvironment that promotes the chance encounter of antigen-specific lymphocytes with their cognate antigen. The system of lymphatic vessels that connects these structures allows immune cells and antigen to drain from sites of infection, resulting in an efficient adaptive immune response to immunological challenge. The formation of LNs is dependent upon the correspondingly named lymphoid tissue inducer (LTi) cell, a retinoic acid receptor-related orphan receptor gt (RORgt)-dependent population in the embryo that provides the critical lymphotoxin signals to developing stromal cells [1][2][3][4] . LTi cells are now described among the group 3 innate lymphoid cells (ILC3) of the ILC family 5 . Within adult mice, LTilike cells persist 6,7 , alongside NKp46 þ ILC3s 8 and colitogenic NKp46 À ILC3s (that appear distinct from LTi-like cells) 9 . All of these ILC3s are thought to be the key cytokine producers within the intestine, aiding epithelial barrier integrity through the production of IL-22, although they may also drive intestinal inflammation 9 . Within the gut, ILC3s are also required for development of isolated lymphoid follicles and T-cellindependent switching to IgA 10,11 . Consistent with a role in CD4 T-cell responses, only the LTi-like cells found in the adult express co-stimulatory molecules such as OX40L and CD30L, associated with CD4 T-cell survival 6,12 and indicating distinct functions in the developed immune system. Furthermore, the absence of RORg-expressing cells resulted in impaired memory CD4 T-cell survival 13 . Crucially, it was recently demonstrated that LTi-like ILC3s can present antigen in the context of major histocompatibility complex II (MHCII) 14 , and this was required for normal regulation of CD4 T-cell responses to commensal bacteria. The mechanisms by which ILC3s regulate CD4 T-cell responses and the site where this occurs are unknown. How, for example, do the ILC populations in secondary lymphoid tissue relate to those in the periphery? It is possible that specific ILC subsets migrate to secondary lymphoid tissue in order to present peptides, akin to dendritic cells (DCs).
To better understand this, here we investigate the ILC composition of a range of LNs, comparing those that drain peripheral tissues to those that drain mucosal sites. While ILCs can be detected in all LNs analysed, RORg þ ILC3s are enriched within mesenteric (m) and mediastinal (md) LNs, which drain mucosal tissues. Within the mLNs, ILC3s reside exclusively within the interfollicular spaces, where they sit within close proximity to GATA-3 þ ILC2s and form a microenvironment that was not detected within inguinal (i), brachial (b) or popliteal (p) LNs. Studies with Kaede transgenic mice demonstrate constitutive trafficking of ILCs from the gut to the mLN. Migration of LTi-like ILC3s, but not other ILC subsets, is dependent upon CCR7. Therefore, our data reveal that ILC populations utilize different mechanisms to traffic to secondary lymphoid tissue, with mucosal-draining LNs containing a distinct interfollicular microenvironment populated by LTi-like ILC3s.

Results
ILC3s are the main ILC group in mucosal-draining LNs. Many studies of ILCs have focused on their role at barrier sites of the body such as the gastrointestinal tract, the lung and the skin 9,[15][16][17][18][19][20] . Although ILC populations have been described within the secondary lymphoid tissue 9,12,14,17,20,21 , this has often been in mice lacking B and T cells 9,12,20 . Therefore, initially we sought to analyse the ILC populations present within LNs that drain distinct anatomical sites in wild-type (WT) mice. Given concerns that surface markers such as Thy1 are not definitive, we first identified IL-7Ra þ Lin (B220, CD3, CD5 and CD11c) À cells, and then used the expression of T-bet, GATA-3 and RORg to identify the ILC1, ILC2 and ILC3 groups, respectively (Fig. 1a). This strategy excludes conventional natural killer (NK) cells and focuses the analysis to the 'helper' ILC subsets recently shown to derive from an Id2 þ IL-7Ra þ progenitor 22 . Intracellular staining for CD3 was included to ensure exclusion of T cells. Per bLN or iLN, comparable numbers of ILC1s, ILC2s and ILC3s were detected (Fig. 1b). Notably, the number of ILC1s, ILC2s and ILC3s per individual mLN was significantly higher than in the peripheraldraining LNs (Fig. 1b), and within mLNs there were significantly more ILC3s than any other type of ILC, notable since these nodes drain the ILC3-enriched intestinal tissue 10,16 . Analysis of the ILC composition of each LN, shown as % of the IL-7Ra þ Lin À population, further highlighted the dominance of the ILC3 population within the mLN, although ILC3s were still present in all the peripheral LNs assessed (Fig. 1c,d). Surprisingly, the mdLN ILC populations were also dominated by ILC3s (Fig. 1c,d), despite the strong association of ILC2s with the lung and the lack of ILC3s in lung tissue 16,23,24 . In contrast, the proportion of ILC3s was lowest within the pLN, a small peripheral LN (Fig. 1c,d), revealing an ILC3 bias within mucosal tissue-draining LNs. In the gut, ILC3s can be subdivided based on cytokine expression and expression of the surface markers CCR6, CD4, MHCII and NKp46 (refs 9,18,25). We found that the majority of ILC3s in the mLN (Fig. 1e,f) were CCR6 þ CD4 þ / À MHCII þ NKp46 À , consistent with the previously described LTi-like population 26 . Although a minor population of NKp46 þ ILC3s were also detected, the CCR6 À CD4 À RORg þ T-bet þ population detected in the gut 25 was largely absent from WT mLNs. Within the ILC1 group, a population of ex-ILC3 cells has been described, where expression of RORg is lost and T-bet and NKp46 expression is turned on. Using RORg cre x ROSA26 eYFP fate mapping mice (Supplementary Fig. 1) we found that in all LNs analysed, a substantial population of T-bet þ eYFP þ RORg À cells were evident. The ability of the LN ILC3s to produce IL-22 was confirmed by in vitro culture with IL-23 (Fig. 1g), with ILC3s in the mLN significantly better at producing IL-22 than their counterparts in iLNs (Fig. 1h).
Since ILC3s regulate CD4 T-cell responses to commensal bacteria 14 , we sought to further assess whether ILC3s limit CD4 T-cell responses within LNs following immunization. Mice in which ILC3s lack MHCII (MHCII DILC3 ) 14 were immunized intraperitoneally (i.p.) with 2W1S peptide and the endogenous CD4 T-cell response assessed in a pool of ILC3-containing peripheral LNs. Within the peripheral LNs of MHCII DILC3 mice 6 days post immunization, the 2W1S response was substantially enhanced compared with controls and 2W1S-specific CD4 T cells showed increased frequencies of cells with an effector ARTICLE phenotype (CD44 þ CD62L lo ; Fig. 1i), consistent with a regulatory function for ILC3s in limiting CD4 T-cell responses through an MHCII-dependent mechanism.
ILC3s form a distinct microenvironment in mucosal LNs. Immunofluorescence staining was used to assess the location of ILC3s within LNs. Cells identified as ILC3s (RORg þ IL-7Ra þ CD3 À ) (Fig. 2a,b) were located almost exclusively within the interfollicular spaces of the mLN, with a scattering of cells further into the T zone at the interface with the follicle (Fig. 2a,b). Besides ILC3s, T cells expressing RORg (RORg þ IL-7Ra þ CD3 þ ) were identified within the interfollicular areas and T zone. To quantify ILC3s and RORg þ T cells within the interfollicular areas, regions between B-cell follicles were marked and the percentage of RORg þ IL-7Ra þ cells lacking CD3 calculated ( Fig. 2c and Supplementary Fig. 2). These data demonstrate that the interfollicular areas of the mLN are dominated by ILC3s, while in the iLN and pLN most cells expressing RORg are T cells, consistent with the flow cytometric analysis detecting fewer ILC3s within these tissues. Given their distinctive location, we also investigated whether ILCs are important for maintaining stromal cell populations within these tissues, a role that would be analogous to ILC3 function in the embryo. Different stromal cells were analysed in sections of LN from bone marrow (BM) chimeras where ILC3s were heavily depleted. No gross effects were evident in LN architecture with normal segregation of B and T cell areas. Lymphatic vessels and high endothelial venules appeared comparable between depleted and control chimeras ( Supplementary Fig. 3). The marginal reticular cell population, identified by expression of RANKL and located at the outer edge of follicles and thus in close proximity to ILCs also appeared normal ( Supplementary Fig. 3). Therefore no overt changes in LN structure or the composition of LN stromal populations were evident after ILC3 depletion.
The location of ILC2s within secondary lymphoid tissue has not yet been described. To identify ILC2s within mLN sections, expression of GATA-3 and ICOS or GATA-3 and KLRG-1 (ref. 27) were assessed. Cells lacking CD3 and expressing either GATA-3 and KLRG-1 (ref. 27) or GATA-3 and ICOS 17 were identified within the interfollicular spaces alongside ILC3s (Fig. 2d). Although fewer in number (Fig. 1b), this suggests that both ILC2s and ILC3s reside within close proximity within a LN.
Given that T cells expressing RORg reside in similar areas in the LN to ILC3s, we also sought to understand the contribution that ILC3s make to the total RORg þ population ( Supplementary  Fig. 4). Gating on RORg þ B220 À CD11b À CD11c À cells, we found that ILC3s constituted a very minor (o5%) proportion of RORg þ cells in peripheral LNs; however, a significantly larger part of the RORg þ population in mLNs (420%). Collectively, our data reveal that ILC3s are more numerous in LNs draining  mucosal tissue, and that they form a distinct microenvironment within the interfollicular spaces.
Altered ILC frequencies in lymphopenic mice. Studies of ILC3 function have often relied upon Rag À / À mice and anti-Thy1 antibody (Ab) treatment to deplete these cells in vivo 9,20 . To enable comparison of our data with previously published studies, we sought to assess the ILC composition of LNs in Rag À / À mice using the flow cytometric analysis described above (Fig. 3a). Surprisingly, within the mLN, this differed vastly from that of WT mice, with GATA-3 þ ILC2s replacing RORg þ ILCs as the dominant population ( Fig. 3a-c). Furthermore, this difference was mLN-specific, since within iLNs a comparable percentage of each ILC type was detected (Fig. 3d,e). The majority of the ILC3s in Rag À / À mLNs lacked CCR6 and CD4 expression (Fig. 3f,g), thus resembling a distinct population described within the intestine 25 , which is not detected in WT mLN (Fig. 1e,f). Given the disrupted architecture within Rag À / À LNs, mLNs from other T-celldeficient mice were also assessed with similar results (Fig. 3g-k). Therefore, the composition of ILCs within the gut-draining mLN is altered in T-cell-deficient mice highlighting the importance of using WT mice. In addition, the high frequency of ILC2 cells within T-cell-deficient mLNs enabled us to confirm our previous location studies. Within ZAP70 À / À mLN, as in WT, ILC2s were readily detected within the interfollicular spaces and at the B:T interface, interspersed with the ILC3s ( Supplementary Fig. 5).
ILCs traffic from the gut to the mLN. Since ILC3s are concentrated within the gut, we sought to test whether an ILC3 bias in the mLN reflected direct trafficking of ILC3s from the intestine to the mLN using transgenic Kaede mice 28 . Here cells express the photoconvertible Kaede protein that converts irreversibly from green to red fluorescence following violet light exposure. Following laparotomy via a small midline incision, the small intestines of Kaede mice were carefully exposed to violet light (labelled 'photoconverted', or PC), while the mLN and Peyer's patches (PPs) were shielded, thus specifically labelling intestinal cells. In all experiments with Kaede mice, expression of Kaede-red protein was based upon comparison with ambient light-treated surgery control mice (labelled 'non-converted' or NC). In PC mice immediately after surgery, ILCs (Fig. 4a,d) and CD4 T cells (Fig. 4b,e) in the small intestine expressed Kaede-red protein compared with NC controls. ILCs within the shielded mLN were not labelled by this process (Fig. 4c). Analysis of the mLN B24 h after exposure of the intestine to violet light revealed a significant number of Kaede-red þ ILCs within this site (Fig. 4f) demonstrating the direct trafficking of these cells from the intestine to the mLN. Gating specifically on the CCR6 þ ILC3s and CCR6 À Sca-1 þ putative ILC2s revealed that both populations had migrated from the intestine (Fig. 4f). While only a modest proportion of CD4 T cells in the mLN of PC mice at 24 h expressed Kaede-red protein (Fig. 4g), substantially more of the ILCs were labelled (Fig. 4h) and this was true of both the ILC3 and ILC2 subsets (Fig. 4i,j). Within the iLN at 24 h post surgery, very few ILCs or T cells expressing Kaede-red were detected (Fig. 4k,l). These data demonstrate that ILC populations in the gut are able to traffic to the draining mLN.
LTi-like ILC3s require CCR7 to traffic to LNs. Homing of lymphocytes and DCs to LNs requires CCR7 (ref. 29), and both embryonic LTi cells and adult splenic LTi-like cells express CXCR5 and CCR7 mRNA 26,30 ; therefore, we hypothesized that CCR7 was required for trafficking of ILC3s into LNs. In CCR7 À / À mice, the total numbers of LTi-like ILC3s were significantly reduced in both mLNs and iLNs compared with WT tissues ( Fig. 5a and Supplementary Fig. 6). Notably, total numbers of NKp46 þ ILC3s, as well as both ILC1 and ILC2 populations, were not significantly different in mLNs. Furthermore, no significant difference in the number of LTi-like ILC3s was detected in WT and CCR7 À / À spleens, indicating that the difference in LNs was not simply due to a systemic reduction in LTi-like ILC3s in CCR7 À / À mice (Fig. 5a). To investigate whether LN entry of LTilike ILC3s was dependent upon direct signalling through CCR7, mixed BM chimeras were generated using an B1:1 mix of either WT (CD45.1):WT (CD45.2) or WT (CD45.1):CCR7 À / À (CD45.2) BM (Fig. 5b,c and Supplementary Fig. 7). The use of CD45.1 þ CD45.2 þ host mice enabled any residual host cells within these chimeras to be identified and excluded from the analysis, thus removing any possible bias from the persistence of radio-resistant host cells. Consistent with a dependency on CCR7 for CD4 T-cell entry into LNs 29 , in WT:CCR7 À / À chimeras, B90% of the LN CD4 T cells were of WT BM origin. This bias was not seen in the WT:WT BM chimeras, validating the mixed BM chimera approach. The vast majority of LTi-like ILC3s in the LNs of WT:CCR7 À / À BM chimeras were derived from WT BM, while this was not the case for ILC2s in LNs (Fig. 5b). The homing of both CCR7 À / À CD4 T cells and LTi-like ILC3s to the spleen was much less impaired than in LNs, as CCR7 is not critical for entry to this tissue (Fig. 5c). To further test whether ILC3s were functionally sensitive to CCR7 ligands, a transmigration assay using WT mLN cells was performed. ILC3s, but not ILC2s, migrated towards CCL21, albeit less efficiently than CD4 T cells or DCs (Fig. 5d), consistent with functional CCR7 expression. In summary, LTi-like ILC3s, but not other ILC subsets, require CCR7 to enter LNs.

Accumulation of LTi-like ILC3s post infection requires CCR7.
Our studies using the Kaede mice demonstrated that ILCs, particularly the ILC3s, were able to traffic from the intestine to the mLN, but we sought to further investigate ILC trafficking using an alternative system. To this end, we used an infection model with the natural mouse parasite Heligmosomoides polygyrus. Following oral gavage with infective larvae, the helminths penetrate the submucosae of the small intestine, eliciting a strong Th2 response in the draining mLN. Control of this infection involves both innate and adaptive responses with accumulation of ILC2 cells previously reported 31,32 . It was also recently shown that H. polygyrus infection can induce IL-1b production 33 , dampening IL-25 and IL-33 responses and aiding pathogen chronicity, while within the lung IL-1b signals cause expansion of ILC3s 34 . Consistent with previous studies, in BALB/c mice, where the infection is cleared, we observed a significant increase in ILC2s in the mLN 7 days post infection (dpi; Fig. 6a) 31,32 . In contrast, within C57BL/6 mice, which fail to clear the infection, no significant accumulation of ILC2 was observed by 7 dpi, although by 14 dpi numbers appeared more comparable to infected BALB/ c mice, again consistent with previous observations 31 . Strikingly, in both strains of mice there was also a significant increase in the number of ILC3s within the mLN after infection, likely reflecting IL-1b production and/or intestinal damage caused by the parasite (Fig. 6a). To investigate whether ILC accumulation in the mLN reflected trafficking or in situ proliferation, expression of Ki-67 was assessed (Fig. 6b,c)    (d) Enumeration and (e) percentage of ILC1, ILC2 and ILC3 cells in Rag À / À iLN (n ¼ 7, 9,9). (f) Example of CCR6 and CD4 or T-bet staining on RORg þ ILCs in mLN of WT and Rag À / À mice. (g) Percentage of RORg þ ILC3s expressing CCR6, CD4, MHCII, NKp46 and T-bet in Rag À / À and ZAP70 À / À mLN (n ¼ 7, 6, 13, 10, 4, 6, 13, 10, 10 and 10), pooled from minimum of two independent experiments except MHCII, which is 1. Number of ILC3 (h) and ILC2 (i) cells per individual mLN from WT, Rag À / À , ZAP70 À / À and TCRa À / À mice (n ¼ 3, 12, 11 and 5). Percentage of ILC population with ILC3 (j) and ILC2 (k) phenotype in mLN from WT, Rag À / À ZAP70 À / À and TCRa À / À mice (n ¼ 3, 12, 11 and 5). Data pooled from minimum of two independent experiments. Bars represent the median values in all graphs.
significantly reduced in CCR7 À / À mLNs 7 dpi compared with C57BL/6 WT (Fig. 6d). Furthermore, this reduction was specific to the LTi-like ILC3 population, since there was no significant difference in the number of NKp46 þ ILC3s in WT versus CCR7 À / À mLN at 7 dpi (Fig. 6e). Surprisingly, analysis of Ki-67 expression revealed that in the absence of CCR7, proliferation of both the LTi-like and NKp46 þ ILC3 populations was substantially increased after infection (Fig. 6f,g), indicating dysregulation of the normal response to this parasite. In summary, these data show that accumulation specifically of LTi-like ILC3s in the draining mLN is dependent upon CCR7.

Discussion
Here we demonstrate that the interfollicular areas of mucosal draining LNs contain a microenvironment populated by  ¼ 4, 4). *Po0.05 (Mann-Whitney non-parametric, two-tailed test, except d,e where unpaired t-test was used). Data representative of three independent experiments, bars represent the median values in all graphs. SI, small intestine, NC, non-converted ambient light-exposed surgery controls, PC, photoconverted mice exposed to violet light. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6862 ARTICLE numerous LTi-like ILC3s. Using Kaede mice we show that ILC3s are able to traffic to the draining mLN. We report that in the absence of cell-intrinsic CCR7 signalling, the LTi-like ILC3 subset, which forms the majority of ILC3s in LNs, is significantly reduced. In vitro, ILC3s are able to migrate towards a ligand of CCR7, CCL21. Finally, using a helminth infection model we show that accumulation of LTi-like ILC3s in the draining LN requires CCR7 expression.
Although a rare population within secondary lymphoid tissue, concentrated clusters of ILC3s can be found within interfollicular areas of LNs. This region is a key site for interactions during the initial stages of adaptive responses, with both antigen-specific CD4 T and B cells specifically located here before migration into the follicle 35 . Given their expression of MHCII, but a lack of CD80 and CD86 (ref. 14), ILC3s are unlikely to drive T-cell proliferation under homeostatic conditions. Rather, these cells   CD4 þ T cells, CD11c þ DCs, Lin À IL-7Ra þ CCR6 þ ILC3, Lin À IL-7Ra þ CCR6 À ICOS þ ILC2 or B220 þ B cells prepared from WT or CCR7 À / À mLN that have migrated towards CCL21 in a transmigration assay (n ¼ 4, 2 and 4). *Po0.05, **Po0.01, ***Po0.001 and ****Po0.0001 (Mann-Whitney nonparametric, two-tailed test). Data are pooled from three (a) or two (b-d) independent experiments, bars represent the median values in all graphs.
likely interact with effector cells after priming, regulatory T cells or memory CD4 T cells recirculating through the tissue. While many more ILC3s were detected within the interfollicular areas of the mLN compared with peripheral LNs, antigen-specific CD4 T cells were increased in the peripheral LNs of MHCII DILC3 mice after peptide immunization, indicating that ILC3s function in a regulatory manner within these tissues as well as within the gut and mLN 14 . Integration of innate stimulatory signals induced by adjuvants or infectious agents may shape the outcome of ILC3:CD4 T-cell interactions 36 , perhaps through inducing expression of co-stimulatory molecules. While we know that ILC3s are concentrated in the intestine, and now the mLN, the relationship between the populations at these sites was unclear. Our observation that photoconverted ILC3s within the gut of Kaede mice can readily traffic from the intestine to the draining mLN leads us to question where does ILC3-mediated control of CD4 T-cell responses to commensal bacteria 14 actually occur in vivo? Is antigen transferred to ILC3 within the mLN from other migrating cells, or could ILC3 potentially acquire antigen in the gut and traffic directly to the mLN? Given both their ability to migrate and their location in LNs, it is possible to propose a model where these cells acquire commensal antigens in the gut and present it to responding CD4 T cells within interfollicular areas. Although deletion of MHCII on ILC3s did not affect their frequency in the mLN 14 , it is not clear whether antigen uptake is required for trafficking from the gut. Further understanding of the specific role of co-stimulatory molecules by ILCs, and whether or not these change following ARTICLE immunization, is required to really understand how these cells might regulate CD4 T cells. A surprising discovery from the phenotyping of LN ILCs was that a high proportion of ILCs in the mdLN were ILC3s, which, along with the mLN, appears to distinguish it from peripheral LNs, although ILC3s were still found here and also regulated CD4 T-cell numbers. In contrast to the gut, however, there is little evidence of many ILC3s within the lung 16 , arguing against trafficking from this site to the mdLN, perhaps pointing instead to a role in regulating responses to bacterial antigens from the lower respiratory tract. In all LNs analysed, however, it appears that the number of ILC3s is highly CCR7-dependent. This would suggest a common mechanism by which these cells enter LNs, likely through the afferent lymphatics, a CCR7dependent process based upon comparisons with T cells 37 . It remains possible that ILCs may enter LNs from the circulation through high endothelial venules in a CCR7-dependent manner, however, ILC3s are largely CD62L À 4 . Progenitor populations have been described in the BM and blood 22,27 ; however, it is not clear that 'mature' ILC populations exist in the circulation. Furthermore, we cannot rule out the possibility that the observed reduction in numbers of LTi-like ILC3 in CCR7 À / À LNs does not, to some extent, result from the additional absence of T cells in these tissues. The partial rescue of CCR7deficient ILC3 numbers in the mesenteric, but not inguinal, LNs of mixed BM chimeras is suggestive of some cell-extrinsic effects in this tissue, which may be especially relevant, given our observation of vastly reduced numbers of ILC3 specifically in the mLN of T-cell-deficient mice. Our data also show that different ILC populations utilize distinct homing receptors for migration. Even among the ILC3s, those expressing NKp46 showed no clear requirement for CCR7 in reaching the mLN, an observation that also applies to ILC2s. On the basis of expression of Ki-67, ILC2s appear more proliferative than ILC3s, and it seems likely that changes in the number of ILC2s reflect a combination of proliferation and trafficking, with data from the Kaede mice showing that they are capable of migrating from the intestine to the mLN. It is possible that activation through IL-33 or thymic stromal lymphopoietin (TSLP) 19,38 may be driving migration to the draining LN, with a reliance on alternative chemokine receptors to CCR7.
While the effects of ILC populations on B-cell responses in LNs have not been explored, ILC3s do support T-independent switching to IgA within isolated lymphoid follicles 11 , and it was recently shown that within the spleen, marginal zone B-cell responses are influenced by ILC3s 39 . It is noteworthy in this latter study that the phenotype of the human splenic ILC3 population described was distinct (CD40L þ NKp46 þ ) from the majority of those ILC3s detected in mouse secondary lymphoid tissue. It is also worth noting that DCs migrate through the interfollicular space after arrival in the afferent lymphatics 40 ; thus, indirect effects on the adaptive response via effects on DCs may also occur and ILC2s have recently been shown to influence DC migration from the lung to the LN 41 . Previous studies with lymphopenic mice have provided valuable insight into what ILCs can do; however, the challenge now is to understand their roles and contribution within a complete immune system. Since both ILC2 and ILC3 populations share a similar location within LNs and can process and present antigen 14,42,43 , one can speculate that a balance between different ILC populations may help to shape the outcome of adaptive immune responses. Evidence presented here and in several recent papers indicates that ILC3s play important roles in the control of adaptive immune responses. It will be important to understand exactly how ILCs can influence CD4 T cells and whether we can target this population to enhance or restrict responses.

Methods
Mice. Animals were used in accordance with the Home Office guidelines at the University of Birmingham and the University of Glasgow. The following mice were bred and maintained in the Biomedical Services Unit at the University of Birmingham: Balb/c, C57BL/6, CCR7 À / À , Rag À / À , TCRa À / À , ZAP70 À / À (all C57BL/6 background). Kaede mice were bred and maintained at the University of Glasgow. For all these experiments, a mixture of male and female mice was used at 6-12 weeks of age. H2-Ab1 fl/fl and MHCII DILC3 mice (all female, aged 8 weeks) were previously described 14 . Genetically modified mouse strains were compared with WT controls from the same facility, while WT mice used in Fig. 1 were obtained from Harlan Laboratories, all female 6-10 weeks of age. Mixed BM chimeras were generated through intravenous transfer of an approximate 1:1 mix of CD45.1 þ WT and either CD45.2 þ WT or CCR7 À / À BM into lethally irradiated CD45.1 þ CD45.2 þ WT hosts (2 Â 600 rad). RORg À / þ hosts were lethally irradiated (2 Â 450 rad) before intravenous transfer of WT or RORg À / À BM. After B4 weeks, RORg À / À into RORg À / þ chimeras were additionally treated with anti-Thy1 Abs and then assessed a further 4 weeks later. Control (WT into RORg À / þ ) chimeras were treated with control rat IgG (Sigma-Aldrich).
Mice were infected with B200 L3 larvae of H. polygyrus larvae by oral gavage. Mice were immunized i.p. with 100 mg 2W1S peptide on D0 and D2, and then taken 6 days later. A pool of peripheral (auxillary, brachial, cervical and inguinal) LNs were compared. WT or RORg À / À into RORg À / þ chimeras were immunized in the front paw pads with 2 Â 10 6 ActA-deficient L. monocytogenes-expressing 2W1S peptide (Lm-2W1S). Bacteria were grown overnight in a shaking 37°C incubator in Luria-Bertani broth supplemented with 20 mg ml À 1 chloramphenicol, and then subcultured and grown under the same conditions until the OD 600 ¼ 0.1. Bacteria were diluted in PBS for injection.
Surgical procedures. Under anaesthesia, the small intestine of Kaede transgenic mice was exposed following laparotomy with a small midline incision. Two-to three-centimetre sections of tissue were exposed to violet light (395 nm UV LED) for 2-3 min each, while PPs and mLN were protected from exposure using highdensity black card. Following replacement of the small intestine, the muscle layer was sutured and the skin closed with surgical clips. Zero-hour controls were taken immediately, while mice in 24-h time point were allowed to recover. The small intestine of control mice was exposed to ambient light. Before surgery, the mice received subcutaneous analgesics Carprofen (Rimadyl, Pfizer) and Buprenorphine (Vetergesic, Reckitt Benckiser Healthcare) at 0.1 ml and 0.15 ml per 100 g, respectively.
Cell culture. Cells were cultured overnight in RPMI/10% FCS with 10 ng recombinant mouse IL-23, and then Brefeldin A was added to a final concentration of 10 mg ml À 1 for B4 h before intracellular cytokine staining.
Transmigration assay. In all, 2 Â 10 6 cells (WT or CCR7 À / À mLN) in 100 ml RPMI 2% FBS were loaded into transwell inserts with a diameter of 6.5 mm and pore size of 5 mm (Corning Transwell Permeable Supports, 3421, Sigma-Aldrich) and placed in wells containing ± 20 nM CCL21 in RPMI with 2% FBS. Plates were incubated for 3 h at 37°C, and then 5.5 mM EDTA was added to bottom well and plate left on ice for 10 min, before migrated and input cells were analysed by flow cytometry. Percentages of migrated cells shown are as a proportion of the total number of cells retrieved from migrated and input wells. All samples were analysed in triplicate in each experiment, with the exception of CCR7 À / À controls, which were analysed in duplicate, and average values displayed.
Statistics. Data were analysed using the GraphPad Prism (versions 5.01 and 6.04). Sample sizes of atleast four data points were used where possible. Samples were only excluded from analysis where total cell yield isolated from tissue was greatly reduced because of technical error. Non-parametric Mann-Whitney test was used to determine significance, which was set at P ¼ o0.05. The median values were calculated and used in all analyses.