Abstract

   How group 3 innate lymphoid cells (ILC3) regulate mucosal protection in
   the presence of T cells remains poorly understood. Here, we examined
   ILC3 function in intestinal immunity using ILC3-deficient mice that
   maintain endogenous T cells, Th17 cells, and secondary lymphoid organs.
   ILC3 were dispensable for generation of Th17 and Th22 cell responses to
   commensal and pathogenic bacteria, and absence of ILC3 did not affect
   IL-22-production by CD4 T cells before or during infection. However,
   despite presence of IL-22-producing T cells, ILC3 and ILC3-derived
   IL-22 were required for maintaining homeostatic functions of the
   intestinal epithelium. T cell-sufficient, ILC3-deficient mice were
   capable of pathogen clearance and survived infection with low dose
   Citrobacter rodentium. However, ILC3 increased pathogen tolerance at
   early timepoints of infection by activating tissue-protective immune
   pathways. Consequently, ILC3 were indispensable for survival of high
   dose infection. Our results demonstrate a crucial context-dependent
   role for ILC3 in immune-sufficient animals and provide a blueprint for
   uncoupling of ILC3 and Th17 cell functions.

One Sentence Summary:

   In T cell-sufficient mice, ILC3 are required for protection from
   high-dose, but dispensable for protection from low dose infection.

Introduction

   Group 3 innate lymphoid cells (ILC3) are an integral component of host
   immunity during mucosal infections, most notably through production of
   the effector cytokines IL-17 and IL-22. IL-17 and IL-22 are also
   produced by Th17 cells and both ILC3 and Th17 cells require the
   transcription factor RORγt for their development ([42]1, [43]2). Thus,
   ILC3 are considered the innate immune counterpart of Th17 cells
   ([44]3-[45]6). IL-22 is essential for protection against infection with
   Citrobacter rodentium (C. rod) and ILC3 are the main source of
   protective IL-22 in early stages of infection ([46]7-[47]10). Moreover,
   in T cell-deficient animals, ILC3 are required for protection and
   survival from C. rod infection ([48]11). However, CD4 T cells and CD4 T
   cell-derived IL-22 are also crucial for successful clearance of C. rod
   infection and have been proposed to play a dominant role in protection
   ([49]12-[50]14). Despite the established role of ILC3 in mucosal
   defense in absence of T cells, whether and how ILC3 contribute to
   infection in the presence of T cells, and in particular Th17 cells,
   continues to be a matter of debate.

   Evaluating the role of ILC3 in T cell sufficient animals has been
   impeded by the absence of genetic models in which ILC3 can be
   specifically depleted without compromising T cell or Th17 cell
   development or function ([51]15-[52]17). As opposed to antigen-specific
   T cell responses, which require weeks to develop, ILC3 are either
   tonically activated or can quickly respond to cytokine signals ([53]15,
   [54]18, [55]19). In the context of C. rod infection, whether ILC3 serve
   simply as placeholders for adaptive T cell responses or play additional
   roles in protection remains unresolved. In general, how ILC3 and Th17
   cells work together to provide protection, and the relevant
   contribution of these two subsets to type 3 immunity, is poorly
   understood ([56]20). Moreover, whether and how ILC3 affect primary and
   secondary microbial T cell responses remains unclear.

   To dissect non-redundant roles of ILC3 in protective immunity to C.
   rod, and other intestinal microbes, we generated ILC3-deficient mouse
   models that display intact T cell development and Th17 cell
   differentiation. We find that ILC3 increase pathogen tolerance and ILC3
   contribution to protection in Th17 cell-sufficient animals depends on
   pathogen load. ILC3-deficient, Th17 cell-sufficient mice successfully
   mount a protective T cell response and clear low dose infection but
   succumb to high dose infection. Our work underscores a
   context-dependent collaboration between ILC3 and Th17 cells in mucosal
   defense.

Results

ΔILC3 mice have normal T cell development and Th17 cell differentiation but
lack ILC3

   To examine the role of ILC3 in mucosal protection in the presence of
   Th17 cells, we generated mice with specific ablation of ILC3 that can
   produce Th17 cells. We took advantage of the fact that the
   transcription factor RORγt controls both ILC3 and T cell development
   ([57]1, [58]2, [59]21). We first generated RORγ-STOP-flox (RORγ^STOP or
   STOP) knock-in animals ([60]Figure S1A). In these animals the
   endogenous RORγ locus was modified by generating an inversion of the
   RORγ genomic sequence containing Exons 3–6, which inactivates the gene.
   The inversion was flanked by two sets of mutually exclusive LoxP sites
   ([61]Figure 1A and [62]S1A) in opposite orientation. In the presence of
   Cre recombinase, recombination by either pair of complementary LoxP
   sites leads to inversion of the intervening sequence. This recovers the
   RORγ locus into its wildtype (WT) configuration. It also places the
   second pair of LoxP sites in direct orientation, which allows for
   deletion of the intervening sequence and “locks in” the WT
   configuration of the locus ([63]Figure 1A and [64]S1A). There are two
   notable advantages of this strategy. First, RORγ expression is
   completely prevented in absence of Cre in STOP animals. Second, in
   Cre-expressing mice, the locus is recovered in WT configuration (except
   for two non-complementary LoxP sites in intronic regions) ([65]Figure
   S1A). Therefore, functional expression of RORγ is placed under
   physiological regulation, i.e., RORγ expression is not forced in Cre
   expressing cells and is expected to match WT levels. RORγ^STOP animals
   phenocopy RORγ-KO mice. This includes deficiencies in T cell
   development (loss of double-positive (DP) thymocytes, decrease in naïve
   peripheral CD4 T cells and development of thymomas), Th17 cell
   differentiation (lack of intestinal Th17 cells) and ILC3 development
   (lack of lymph nodes (LNs) / Peyer’s patches (PPs) and lack of adult
   ILC3) ([66]Figure 1 and [67]S1B). To allow for T cell development and
   Th17 cell differentiation we crossed RORγ^STOP mice to CD4-Cre animals
   to generate RORγ^STOP/CD4 (or STOP/CD4) mice, in which Cre-recombinase
   expression is restricted to αβ T cells ([68]22) ([69]Figure 1A).

Figure 1. RORγ^STOP/CD4 mice specifically lack ILC3 in the presence of T
cells, including Th17 cells.

   [70]Figure 1.
   [71]Open in a new tab

   (A) Schematic of generation of RORγ^STOP/CD4 (STOP/CD4) mice. (B) Total
   thymus cellularity in WT, STOP, and STOP/CD4 mice. (C) Thymoma-mediated
   mortality of WT, STOP, and STOP/CD4 mice. n = 27–33 mice per genotype.
   (D-F) Thymocyte development in WT, STOP, and STOP/CD4 mice, including
   (E) double-positive (DP) and (F) CD4 single-positive (CD4 SP)
   thymocytes. One out of two independent experiments with similar
   results, n = 4–5 mice per group. (G-I) Small intestinal RORγ^+Foxp3^–
   Th17 cells (G, H) and Foxp3^+RORγt^– Tregs (I) in WT, STOP, and
   STOP/CD4 mice. Plots gated on TCRβ^+CD4^+ cells (G, H) or total
   lymphocytes (I). Combined data from two out of four independent
   experiments, n = 5–8 mice per group. (J-L) Proportions (J, K) and
   number (L) of small intestinal ILC3 in WT, STOP and STOP/CD4 mice.
   Plots gated on Lin^– cells (J) or total lymphocytes (K). Combined data
   from several independent experiments, n= 7–12 mice per group.

   In the thymus, as expected, STOP mice had dramatically reduced thymic
   cellularity due to accelerated spontaneous apoptosis ([72]21). In
   contrast, thymic cellularity in STOP/CD4 mice was equivalent to WT
   ([73]Figure 1B). STOP/CD4 animals demonstrated considerable recovery of
   thymic T cell development compared to STOP littermates, exemplified by
   absence of mortality due to thymoma formation ([74]Figure 1C). Numbers
   of DP and CD4 single-positive (SP) thymocytes in STOP/CD4 mice were
   similar to WT controls ([75]Figure 1D-[76]F). However, frequency of CD8
   SP thymocytes was elevated ([77]Figure 1D and [78]S1C), likely because
   progenitor immature CD8 single-positive (ISP) cells in STOP/CD4 mice
   retained an immature phenotype ([79]Figure S1D). Peripheral mature CD4
   and CD8 T cell proportions, as well as the distribution of naïve and
   activated CD4 T cells in spleens of STOP/CD4 mice were similar to WT
   controls ([80]Figure S1E, [81]F). In the intestine, RORγt^+ and IL-17^+
   CD4 T (Th17) cells, were present in normal numbers in the small
   intestinal (SI) lamina propria (LP), colonic LP, and cecum LP of
   STOP/CD4 mice ([82]Figure 1G-[83]I and [84]S2). Therefore, STOP/CD4
   mice demonstrate significant recovery of T cell development and can
   generate peripheral Th17 cells.

   At the same time, STOP/CD4 mice lacked ILC3, including lymphoid tissue
   inducer-like (LTi) cells in all intestinal tissues, including small
   intestine, colon, and cecum ([85]Figure 1J-[86]L and [87]S3A, [88]B).
   The numbers and frequencies of other intestinal ILC subsets, e.g. ILC1
   and ILC2, were similar to WT littermates ([89]Figure S3C, [90]D). We
   conclude that STOP/CD4 mice represent an ILC3-deficient, but T
   cell-sufficient model, that preserves near normal T cell development
   and ability to generate Th17 cells. We, therefore, refer to these
   animals as ΔILC3 mice.

   ΔILC3 mice had normal numbers and subsets of intestinal intraepithelial
   lymphocytes (IELs) ([91]Figure S4A-[92]E) and Foxp3^+RORγt^– Treg cells
   in SI and colon LP ([93]Figure 1I and [94]S4F) but had significantly
   decreased frequencies and numbers of Foxp3^+RORγt^+ Treg cells in SI
   and colon LP ([95]Figure S4G, [96]H). Although γδ T cells developed
   normally, since the CD4-Cre construct expression should be restricted
   to αβ T cells, STOP/CD4 animals had significantly decreased RORγt^+ γδ
   T cells in SI LP ([97]Figure S4I). Due to the lack of LTi cells,
   STOP/CD4 mice lacked lymph nodes (LNs) and Peyer’s patches (PPs)
   ([98]Figure S1B). The loss of ILC3/LN/PP also resulted in a notable
   decrease in SI LP IgA^+ plasma cells ([99]Figure S4J, [100]K).

ILC3 preferentially control the transcriptional program of intestinal
epithelial cells at steady state

   To evaluate the relative roles of ILC3 and Th17 cells in steady state
   intestinal homeostasis we performed RNA-Seq on purified SI LP CD4 T
   cells and ileal intestinal epithelial cells (IEC) from WT, STOP and
   ΔILC3 mice. The transcriptional program of both CD4 T cells and IEC was
   significantly affected in STOP mice reflecting perturbed steady state
   intestinal homeostasis in the absence of type 3 immunity ([101]Figure
   2A, [102]B and [103]S5A, [104]B). In contrast, specific loss of ILC3 in
   ΔILC3 mice led to differential effects on the transcriptional programs
   of CD4 T cells and IEC. MDS ordination analysis revealed that ΔILC3 CD4
   T cells closely resembled WT CD4 T cells ([105]Figure 2A top). In
   contrast, the transcriptional program of IEC from ΔILC3 mice was
   significantly different from that of WT IEC and overlapped with that of
   IEC from STOP animals ([106]Figure 2A bottom). A large proportion (64%)
   of CD4 T cell genes that were decreased in STOP animals were not
   decreased, or marginally decreased, in ΔILC3 compared to WT mice
   ([107]Figure 2B). These included genes involved in Th17 cell
   differentiation (e.g., Il17, Il22, Ccr6, Rorc) ([108]Figure 2B,
   [109]C). Pathway analysis revealed enrichment in inflammatory and
   adaptive immune response pathways, as well as Th17 cell differentiation
   in WT and ΔILC3 CD4 T cells compared to STOP CD4 T cells ([110]Figure
   2D), confirming general recovery of intestinal CD4 T cell functionality
   in ΔILC3 mice. Further confirming normalized intestinal T cell
   development and functionality, intestinal CD4 T cells from ΔILC3 mice
   demonstrated similar T cell receptor repertoire gene usage to that of
   WT animals ([111]Figure 2B and [112]S5A, [113]B). A small set of genes
   decreased in ΔILC3 intestinal CD4 T cells compared to WT CD4 T cells.
   These included genes associated with residence in secondary and
   tertiary lymphoid organs, e.g., resting T cell genes (Sell, Tcf7,
   S1pr1). In contrast to CD4 T cells, an overwhelming proportion of genes
   (87%) decreased in STOP IEC were also similarly decreased in ΔILC3 IEC
   compared to WT controls ([114]Figure 2B). Expression of a small
   fraction of IEC genes, including Saa1, Nos2, and Gata4, were decreased
   in ΔILC3 mice, but not to the same level as in STOP mice, suggesting
   that they are partially controlled by ILC3 ([115]Figure 2B, [116]E, and
   [117]S5C). Only a handful (3%) of downregulated IEC genes, e.g.,
   Ceacam10, were similarly expressed between WT in ΔILC3 IEC ([118]Figure
   2B). Similar results were obtained with upregulated genes ([119]Figure
   S5A, [120]B). Altogether, these data suggest that ILC3, rather than
   Th17 cells, are crucial regulators of steady state IEC function.

Figure 2. ILC3 control steady state intestinal epithelial cell function.

   [121]Figure 2.
   [122]Open in a new tab

   (A) Multidimensional scaling (MDS) ordination of transcriptomes of SI
   LP CD4 T cells (top) and terminal ileum intestinal epithelial cells
   (IEC) (bottom) from WT, STOP and ΔILC3 mice. n = 2–4 mice per group.
   (B) Differentially expressed genes decreased in STOP compared to WT
   mice are plotted and compared between WT and ΔILC3 mice for either LP
   CD4 T cells (top) or IEC (bottom). “Like WT”, similar expression
   between WT and ΔILC3 mice. “Low in ΔILC3”, decreased in ΔILC3 compared
   to WT. T-cell receptor (TCR) genes, Th17 signature genes or IL-22
   controlled genes are highlighted below the heatmaps. (C) Th17 cell
   signature profile genes in RNA-Seq data from SI LP CD4 T cells from WT,
   STOP and ΔILC3 mice. (D) Pathways similarly enriched in LP CD4 T cells
   from WT and ΔILC3 mice compared to STOP mice; Top 5 enriched pathways
   of GO biological process and KEGG categories are shown. Pathways
   related to intestinal T cell functionality are shaded in red. (E)
   IL-22-controlled genes in RNA-Seq data from SI LP IEC from WT, STOP and
   ΔILC3 mice. (F) Gene set enrichment analysis (GSEA) of IL-22-regulated
   genes in transcriptomes of IEC from WT versus ΔILC3 mice. (G)
   Quantitative RT-PCR for Il22 mRNA from purified CD4 T cells from SI LP
   of WT, STOP and ΔILC3 mice. (H) IL-22 protein levels in supernatants of
   in vitro stimulated SI LP CD4 T cells. (I) Intracellular staining for
   IL-22 in SI LP TCRβ^+CD4^+ T cells from WT, STOP and ΔILC3 mice (see
   also [123]Figure S5D). Data from one of two independent experiments
   with similar results, n = 3–4 mice per group.

   IL-22 is the main functional ILC3 cytokine ([124]10, [125]11, [126]19,
   [127]23). However, IL-22 is also produced by CD4 T cells ([128]12,
   [129]24) and the relative contribution of T cell and ILC3-derived IL-22
   to immune phenotypes has been difficult to define. As expected, CD4 T
   cell-derived IL-22 was lost in STOP mice ([130]Figure 2C, [131]G and
   [132]S5D). In contrast, intestinal CD4 T cells from ΔILC3 mice produced
   WT amounts of IL-22 both at mRNA and protein level ([133]Figure 2C,
   [134]2G-[135]I and [136]S5D). ΔILC3 mice and WT controls contained
   similar levels of SI LP IL-22^+IL-17^neg cells and IL-22^+IL-17^+ Th17
   cells ([137]Figure 2I and [138]S2D). These results suggest that ILC3 do
   not control steady state production of IL-22 by intestinal CD4 T cells.
   ILC3 provided the bulk of IL-22 in the small and large intestine at
   steady state ([139]Figure S5E, [140]F). Therefore, despite similar
   production of IL-22 by CD4 T cells, the overall number of
   IL-22-producing cells in small intestine and colon was significantly
   decreased in ΔILC3 mice ([141]Figure S5E, [142]F). IEC from ΔILC3 mice
   showed significant defects in expression of IL-22-controlled genes
   ([143]Figure 2B, [144]E). The overwhelming majority of IL-22-dependent
   genes in ΔILC3 IEC, e.g., Reg3γ, Saa2, Saa3, Fut2, showed significantly
   decreased levels of expression, similar to that in STOP mice, and
   therefore were dependent on ILC3, but not T cell-derived IL-22
   ([145]Figure 2B, [146]E and [147]S5A, [148]C). Gene set enrichment
   analysis also confirmed that IL-22 signaling-dependent genes in IEC
   were underrepresented in IEC from ΔILC3 mice compared to WT IEC
   ([149]Figure 2F). The expression of only a few IL-22 dependent genes
   was partially recovered in ΔILC3 mice ([150]Figure 2B, [151]E and
   [152]S5C). Most notably Saa1 and Nos2 (iNOS) expression was controlled
   by both ILC3 and Th17 cells ([153]Figure 2E). The foregoing results
   suggest that ILC3-derived IL-22 is the main regulator of steady state
   intestinal IEC homeostasis. More generally, ΔILC3 mice can be used to
   establish the relative contribution of ILC3 and Th17 cells to mucosal
   immune phenotypes.

ILC3 and pre-existing Th17 cells participate in early protection from
Citrobacter rodentium infection

   ILC3 play crucial role in early protection from infection with C. rod
   in T cell-deficient animals ([154]10, [155]11, [156]23). However, how
   ILC3 and T cells work together is unclear ([157]11, [158]12, [159]20).
   To investigate how presence of endogenous T cell development and Th17
   cell differentiation benefit protection in the absence of ILC3, we
   infected WT, STOP, and ΔILC3 mice with 10^9 CFU/animal of C. rod and
   examined disease development. As expected, STOP mice had increased
   pathogen colonization early on and rapidly succumbed to infection
   ([160]Figure 3A, [161]B). ΔILC3 mice showed similar deficiency in
   controlling the infection and succumbed to it as rapidly as STOP mice
   ([162]Figure 3A, [163]B). Because ILC3-deficient animals died by Day 8,
   the lack of protection by T cells with high pathogen load could be due
   to the delayed kinetics of T cell activation. To address this
   possibility, we infected mice with a lower dose of C. rod (5 X 10^7
   CFU/mouse). This significantly increased the survival of ILC3-deficient
   mice and thus provided more time for activation of additional immune
   responses. At this lower dose of infection ΔILC3 mice were
   significantly better protected than STOP mice and survived on average 3
   days longer ([164]Figure 3C). This suggests that CD4 T cells
   participate in early protection from C. rod infection at lower
   infection doses. The degree of this protection could not be ascertained
   from these experiments, because WT controls have LNs and PPs and
   survived infection ([165]Figure S1B). To address this difference, we
   generated bone marrow (BM) chimeras by lethally irradiating RORγ-KO
   recipients and transferring BM from WT, STOP or ΔILC3 mice ([166]Figure
   3D). All resulting BM chimeras lack LNs and PPs but have different
   levels of reconstitution of their type 3 immune compartment. RORγ-KO
   mice reconstituted with WT BM possess Th17 cells and ILC3, RORγ-KO mice
   reconstituted with ΔILC3 BM lack ILC3 but possess intestinal Th17 cells
   and RORγ-KO mice reconstituted with STOP BM lack both ILC3 and Th17
   cells ([167]Figure 3D and [168]S6A). Following infection with low dose
   C. rod all three groups of BM chimeras succumbed to infection
   demonstrating the crucial role of secondary lymphoid organs for
   sterilizing immunity. However, survival differed significantly
   depending on the presence of ILC3 and Th17 cells. Consistent with loss
   of ILC3 function, both STOP and ΔILC3 BM chimeras showed increase in
   pathogen burden early on, at Day 5 ([169]Figure 3E). However, ΔILC3 BM
   chimeras were significantly more protected and survived longer than
   STOP BM chimeras, suggesting a role for Th17 cells in protection
   ([170]Figure 3F). At the same time, ΔILC3 mice were more susceptible to
   infection than WT chimeras suggesting that ILC3 provide an additional
   level of protection even in the presence of Th17 cells ([171]Figure
   3F).

Figure 3. ILC3 and bystander Th17 cells participate in early Citrobacter
protection.

   [172]Figure 3.
   [173]Open in a new tab

   (A, B) Survival (A) and pathogen CFUs in feces at day 4 post infection
   (B) of WT, STOP and ΔILC3 mice infected with high dose (10^9
   CFU/animal) Citrobacter rodentium (C. rod). One out of several
   independent experiments with similar results, n = 4–5 mice per group.
   (C) Survival of WT, STOP and ΔILC3 mice following infection with low
   dose C. rod (5×10^7 CFU/animal). Data from two out of six independent
   experiments with similar results, n = 5–6 mice per group. p value
   refers to ΔILC3 vs STOP mice. (D) Lethally irradiated
   ILC3/Th17-deficient and LN/PP-deficient RORγ-KO mice were reconstituted
   with BM from WT, STOP or ΔILC3 mice. (E, F) Pathogen CFUs at day 5 (E)
   and survival (F) of RORγ-KO BM chimeras (D) following infection with
   low dose C. rod. Data from two independent experiments with similar
   results, n = 7–9 mice per group. (G) STOP (left) and ΔILC3 (right) mice
   were pre-treated with anti-CD4 antibody to deplete CD4 T cells, or
   isotype antibody as control, and infected with low dose C. rod. One out
   of two independent experiments with similar results, n = 4–6 mice per
   group. (H) STOP (left) and ΔILC3 (right) mice were treated with
   neutralizing anti-IL-22 antibody (or isotype control) and infected with
   low dose C. rod. One out of two independent experiments with similar
   results, n = 2–5 mice per group. (I) Low dose C. rod infection in
   SFB-negative or SFB-positive STOP and ΔILC3 mice. One experiment, n =
   3–5 mice per group. (J) RORγ-KO BM chimeras (D) were infected with low
   dose C. rod and supplemented with exogenous IL-22-Fc or isotype control
   every other day starting at Day 0. One out of two independent
   experiments with similar results, n = 3–5 mice per group. Statistics
   survival curves, Mantel-Cox test.

   To confirm that Th17 cells are responsible for protection in ΔILC3
   compared to STOP mice ([174]Figure 3F), we depleted CD4 T cells in vivo
   using an anti-CD4 monoclonal antibody prior to infection with low dose
   C. rod ([175]Figure 3G and [176]S6B). Depletion of CD4 T cells in STOP
   mice did not affect the level of survival ([177]Figure 3G, left). In
   contrast, depletion of CD4 T cells in ΔILC3 mice resulted in loss of
   protection and CD4 T cell-depleted ΔILC3 mice were as susceptible to
   infection as STOP mice ([178]Figure 3G, right). To investigate whether
   protection is due to Th17 cell effector cytokines we used neutralizing
   antibodies to IL-17 and IL-22 ([179]Figure 3H and [180]S6C). Both
   treatments, and particularly neutralization of IL-22, led to loss of
   protection in ΔILC3 mice, but not in STOP controls ([181]Figure 3H and
   [182]S6C). Therefore, IL-22 producing CD4 T cells can provide increased
   protection to infection with low pathogen dose in the absence of ILC3.
   ΔILC3 mice lack LNs and PPs and therefore are not expected to mount a
   significant antigen-specific T cell response to C. rod, at least at
   early stages of infection. However, T cell protection can be mediated
   by IL-22 production from pre-existing non-antigen-specific Th17 cells.
   We previously showed that most steady state Th17 cells in mouse adult
   intestine in our SPF colony are generated in response to segmented
   filamentous bacteria (SFB) and produce IL-22 ([183]25, [184]26). To
   investigate whether pre-existing SFB-induced Th17 cells provide
   protection in ΔILC3 mice we infected SFB-negative and SFB-positive
   ΔILC3 mice with C. rod ([185]Figure 3I). Th17 cell-mediated protection
   of ΔILC3 mice was lost in the absence of SFB (and SFB Th17 cells)
   ([186]Figure 3I). We conclude that in the absence of ILC3 pre-existing
   commensal IL-22-producing Th17 cells can increase protection to
   infection with low dose C. rod in a bystander manner.

   ILC3-derived IL-22 early during infection is the main mechanism by
   which ILC3 mediate protection against C. rod infection ([187]7,
   [188]11). To examine whether ILC3 mediate protection through IL-22 we
   provided exogenous IL-22 starting at Day 0 to both STOP and ΔILC3 BM
   chimeras. Exogenous IL-22 decreased pathogen burden at Day 5 and
   increased protection in both STOP and ΔILC3 BM chimeras ([189]Figure 3J
   and [190]S6D). Notably, ΔILC3 BM chimeras treated with exogenous IL-22
   were as protected and survived as long as WT BM chimeras ([191]Figure
   3J). Combined, these results demonstrate that both ILC3 and
   pre-existing commensal Th17 cells can contribute to protection against
   low dose infection with C. rod.

ILC3 control steady state epithelial cell homeostasis in the presence of
lymph nodes and Peyer’s Patches

   The above experiments demonstrate that both ILC3 and Th17 cells can
   provide early protection during low dose C. rod infection. However,
   despite presence of endogenous T cell development and intestinal Th17
   cells, ΔILC3 mice succumb to infection. We reasoned that this is due to
   the absence of LNs and PPs, which precludes generation of robust or
   timely pathogen-specific T cell response. Indeed, we could not detect
   antigen-specific increase in cytokine production from CD4 T cells in
   ΔILC3 animals ([192]Figure S7A). This is further supported by the fact
   that WT BM chimeras (WT BM into RORγ-KO recipients), which have both
   ILC3 and Th17 cells, but lack LNs and PPs, cannot clear the pathogen
   and succumb to infection ([193]Figure 3F). Therefore, to investigate
   whether T cells can substitute for ILC3 in the context of C. rod
   infection, we generated a mouse model that not only lacks ILC3 (while
   maintaining ability to generate endogenous T cells and Th17 cells), but
   also preserves development of secondary lymphoid organs ([194]Figure
   4A). We took advantage of the fact that LNs and PPs develop in
   embryonic life. We crossed RORγt^flox mice ([195]27) to
   tamoxifen-inducible CAGG-CRE-ER mice ([196]28) to generate a strain
   where we can conditionally ablate RORγt expression in all cells and
   tissues. RORγt^flox/CAGG-Cre-ER mice were treated with tamoxifen
   shortly after birth, which allows for development of LNs and PPs.
   Consistent with RORγ deletion, tamoxifen treatment led to decrease in
   intestinal ILC3 ([197]Figure S7B) and defect in thymic T cell
   development ([198]Figure S7C) a few weeks later. Tamoxifen-treated
   RORγt^flox/CAGG-Cre-ER mice were lethally irradiated and used as
   recipients of BM from WT, STOP, or ΔILC3 donor animals. The resulting
   chimeric animals are referred to as WT^LN/PP, STOP^LN/PP, and
   ΔILC3^LN/PP respectively ([199]Figure 4A). All of these models
   developed peripheral LNs ([200]Figure S7D) and PPs ([201]Figure S7E)
   with similar numbers and morphology. In support of recovered secondary
   lymphoid organ function, and in contrast to LN/PP-deficient ΔILC3 mice,
   WT^LN/PP and ΔILC3^LN/PP mice contained similar numbers of SI LP B
   cells and IgA^+ plasma cells ([202]Figure S7F, [203]G).

Figure 4. ILC3 control intestinal epithelial cell IL-22 program in the
presence of lymph nodes and Peyer’s patches.

   [204]Figure 4.
   [205]Open in a new tab

   (A) Schematic of generation of LN/PP-sufficient, Th17 cell-sufficient,
   ΔILC3 mice (ΔILC3^LN/PP) and experimental controls. (B, C) RORγt^+ Th17
   cells in SI LP of ΔILC3^LN/PP mice and controls. Plots gated on
   CD4^+TCRβ^+ cells. (D) RORγt^+Foxp3^+ Treg cells in SI LP of
   ΔILC3^LN/PP mice and controls. Plots gated on CD4^+TCRβ^+ cells. (E-G)
   ILC3 in SI LP of ΔILC3^LN/PP mice and controls. Plots gated on Lin^–
   lymphocytes. Combined data from several independent experiments, n =
   5–6 mice per group. (H-J) RORγt^+ (H, I) and RORγt^neg (H, J) IL-22
   producing Lin^– lymphocytes in SI LP of ΔILC3^LN/PP mice and controls.
   Combined data from two independent experiments, n = 5–6. (K) RNA-seq of
   terminal ileum from WT^LN/PP, STOP^LN/PP, and ΔILC3^LN/PP mice.
   Differentially expressed genes decreased in STOP^LN/PP compared to
   WT^LN/PP mice are plotted and compared between WT and ΔILC3^LN/PP mice.
   Only mice without residual ILC3. “Like WT^LN/PP”, similar expression
   between WT^LN/PP and ΔILC3^LN/PP mice. “Low in ΔILC3^LN/PP”, decreased
   in ΔILC3^LN/PP compared to WT^LN/PP. IL-22 induced genes marked in
   green. n = 4 mice per group. (L) Quantitative PCR for IL-22 controlled
   IEC genes in terminal ileum of WT^LN/PP, STOP^LN/PP, and ΔILC3^LN/PP
   mice. Only mice without residual ILC3. RE, relative expression. n = 5–7
   mice per group. (M) Top GO biological process pathways enriched in
   terminal ileum of WT^LN/PP compared to ΔILC3^LN/PP mice. Only mice
   without residual ILC3. Pathways related to anti-microbial defense are
   indicated in blue.

   Following loss of RORγ expression, STOP^LN/PP mice showed dysregulated
   T cell development and Th17 cell differentiation ([206]Figure 4B-[207]D
   and [208]S7H-[209]J). In contrast, WT^LN/PP and ΔILC3^LN/PP mice
   possessed similar levels of intestinal Th17 cells, including
   RORγ^+Foxp3^- and IL-17^+ CD4 T cells ([210]Figure 4B, [211]C and
   [212]S7H-[213]J). However, we found a significant decrease in
   Foxp3^+RORγt^+ Treg cells, in agreement with recent studies
   ([214]29-[215]31) ([216]Figure 4D). Compared to WT^LN/PP mice,
   STOP^LN/PP and ΔILC3^LN/PP mice had similarly severely decreased ILC3
   in SI LP as identified by RORγ expression ([217]Figure 4E-[218]G).
   Moreover, ΔILC3^LN/PP mice also lacked RORγt^– ILC3-like cells
   ([219]32), as evident by the lack of IL-22 expression in
   lineage-negative LP cells ([220]Figure 4H-[221]J). Conditional deletion
   of RORγ in adult animals has been reported to result in loss of ILC3
   and IL-22 production in intestinal LP, but appearance of Lin^–RORγt^–
   ILC3-like cells in mesenteric LNs (mLN) that can be a source of IL-22
   ([222]32). In contrast, very few RORγ-negative IL-22 producing
   ILC3-like cells were detected in mLN of ΔILC3^LN/PP mice and their
   levels were similar to those in STOP^LN/PP mice ([223]Figure
   S7K-[224]M). Therefore, STOP^LN/PP and ΔILC3^LN/PP mice have similar
   deficiency of ILC3 and ILC3 function. ILC1 and ILC2 levels, as well as
   the representation of various myeloid cell populations did not
   significantly differ between WT^LN/PP, STOP^LN/PP and ΔILC3^LN/PP mice
   ([225]Figure S8A-[226]E). Combined, these data show that ΔILC3^LN/PP
   mice possess endogenous T cell development and Th17 cell
   differentiation, have functional LNs and PPs, but lack ILC3 and
   ILC3-like cells.

   We next examined how specific absence of ILC3 affects steady state
   mucosal homeostasis in the lamina propria by performing bulk
   RNA-sequencing of terminal ileum. As expected, due to the presence of
   LNs and PPs, fewer genes were dysregulated in STOP^LN/PP animals than
   STOP animals ([227]Figure 4K, [228]S8F and [229]2B). However, similarly
   to LN/PP-deficient ΔILC3 mice, the expression of most of these genes
   was also dysregulated in ΔILC3^LN/PP animals compared to WT^LN/PP
   animals, indicating that these genes are regulated by ILC3 ([230]Figure
   4K and [231]S8F). This included a number of IEC genes and pathways and
   particularly IL-22-response genes, as well as pathways associated with
   acute inflammatory responses, epithelial defense, and epithelial
   barrier function ([232]Figure 4K-[233]M and [234]S8G). In WT^LN/PP
   animals at steady state most of the LP IL-22 was produced by ILC3 and
   there was no compensatory increase in CD4 T cell-derived IL-22 in
   ΔILC3^LN/PP animals ([235]Figure S8H). Interestingly a subset of
   ΔILC3^LN/PP animals, present at random throughout experiments,
   maintained a small fraction of ILC3 (5–10% of WT) ([236]Figure S9A,
   [237]B). Despite lacking 90% of intestinal IL-22-producing ILC3
   ([238]Figure S9A), these animals showed a significant recovery of the
   IL-22-controlled epithelial cell program compared to animals that
   completely lacked ILC3 ([239]Figure S9C, [240]D). This result further
   underscores the crucial role that ILC3 play in regulation of steady
   state epithelial homeostasis. We conclude that ILC3-derived IL-22,
   rather than T cell-derived IL-22, controls steady state epithelial gene
   expression.

ILC3 are indispensable for survival of high dose mucosal infection

   To examine the specific role of ILC3 in protection from infection with
   C. rod we infected WT^LN/PP and ΔILC3^LN/PP mice with high dose C. rod.
   WT^LN/PP mice survived and cleared infection. In contrast, ΔILC3^LN/PP
   mice were unable to clear the pathogen and all animals succumbed to
   infection ([241]Figure 5A). Infection led to increased expression of
   IL-22 and IL-22-regulated genes, e.g., Reg3g, in colonic and SI LP of
   WT^LN/PP mice, which was not evident in ΔILC3^LN/PP mice ([242]Figure
   5B and [243]S10A). ΔILC3^LN/PP mice had severely decreased numbers of
   IL-22 producing ILC3 in colonic and SI LP at day 4 post infection
   without significant change in the number of IL-22-expressing CD4 T
   cells ([244]Figure 5C, [245]D and [246]S10B, [247]C). In contrast to
   controls, ILC3-deficient animals showed systemic dissemination of the
   pathogen to mLN ([248]Figure 5E), liver ([249]Figure 5F), and spleen
   ([250]Figure 5G) as early as day 4 post infection. These results
   suggest an inability of ILC3-deficient animals to mount protective
   immunity even in the presence of functional LNs.

Figure 5. ILC3 are essential for survival of high dose Citrobacter rodentium
infection.

   [251]Figure 5.
   [252]Open in a new tab

   (A) Survival of WT^LN/PP and ΔILC3^LN/PP mice infected with high dose
   C. rod. One out of two independent experiments with similar results, n
   = 3–5 mice per group (B) Quantitative PCR for Il22 and Reg3g
   transcripts in colon at day 4 post infection. RE, relative expression.
   (C, D) Number of IL-22^+ ILC3 and CD4 T cells in colon LP of infected
   WT^LN/PP and ΔILC3^LN/PP mice at Day 4. Plots in C gated on Lin^–
   lymphocytes. (E-G) Pathogen burden in mLN (E), liver (F), and spleen
   (G) on day 4 post infection. (H) Multidimensional scaling (MDS)
   ordination of transcriptomes of colon tissue from infected WT^LN/PP and
   ΔILC3^LN/PP mice at Day 4. (I) Differentially expressed genes increased
   upon infection in WT^LN/PP mice at Day 4. Genes controlled by IL-22 are
   marked in green and intestinal epithelial cell (IEC) genes are marked
   in red. n = 2 mice per group. (J) Enrichment of top infection-activated
   pathways in infected WT^LN/PP mice versus infected ΔILC3^LN/PP mice.
   Infection-activated pathways were defined as pathways increased in
   infected WT^LN/PP mice compared to uninfected WT^LN/PP mice. GSEA,
   gene-set enrichment analysis. MPS, module perturbation score. (K)
   Differential gene expression in colon of infected ΔILC3^LN/PP versus
   WT^LN/PP mice at Day 4. Red, transcripts increased in ΔILC3^LN/PP mice
   compared to WT^LN/PP mice. Blue, transcripts decreased in ΔILC3^LN/PP
   mice compared to WT^LN/PP mice. Green, IL-22-induced genes.

   To better characterize the role of ILC3 we performed RNA-Seq of total
   intestinal tissues at day 4 post infection. The transcriptional profile
   of infected ΔILC3^LN/PP mice was significantly different from that of
   infected WT^LN/PP mice in both colon ([253]Figure 5H) and small
   intestine ([254]Figure S10D). ΔILC3^LN/PP mice showed significant
   deficiency in expression of two thirds of genes induced by infection in
   colon of WT^LN/PP animals ([255]Figure 5I). This included genes
   belonging to multiple immune defense pathways ([256]Figure 5J). In
   particular, induction of IL-22-regulated genes and IEC defense genes
   required presence of ILC3 ([257]Figure 5I, [258]K and [259]Figure S10E,
   [260]F).

   These results demonstrate that, in the presence of adaptive immunity,
   ILC3-derived IL-22 is required for appropriate activation of epithelial
   immune defenses to ensure survival in the context of high dose
   infection with C. rod.

ILC3 are dispensable for survival of low dose infection, but reduce early
disease severity

   To examine whether the requirement for ILC3 depends on the magnitude of
   infection we infected WT^LN/PP and ΔILC3^LN/PP mice with low dose C.
   rod.

   In contrast to WT^LN/PP mice, ΔILC3^LN/PP and STOP^LN/PP mice
   demonstrated increased pathogen burden in the lumen, as well as
   increased pathogen penetration of colonic tissue and dissemination to
   systemic organs at day 5 post infection ([261]Figure 6A-[262]C). In
   accordance with functional ILC3 deficiency, both STOP^LN/PP and
   ΔILC3^LN/PP mice also lacked induction of intestinal Il22 and Reg3g
   early post infection on Day 5 and Day 8 ([263]Figure S11A-[264]D).
   However, despite similar deficiency in ILC3 and ILC3 function at early
   stages, STOP^LN/PP and ΔILC3^LN/PP mice had very different outcomes in
   terms of survival. STOP^LN/PP mice were incapable of clearing infection
   and succumbed to it by Day 12. In contrast, ΔILC3^LN/PP mice survived
   infection and cleared the pathogen with similar kinetics to WT^LN/PP
   mice ([265]Figure 6D, [266]E). We, therefore, conclude that in the
   presence of functional T cell responses, ILC3 are dispensable for
   clearance of low dose infection with C. rod and that Th17 cells are
   crucial, and sufficient, for effective protection.

Figure 6. ILC3 decrease disease severity during low dose Citrobacter
rodentium infection.

   [267]Figure 6.
   [268]Open in a new tab

   WT^LN/PP, STOP^LN/PP and ΔILC3^LN/PP mice were infected with low dose
   C. rod. (A-C) Pathogen levels in feces (A), spleen (B), and colonic
   tissue (C) at day 4 post infection. (D, E) Fecal C. rod levels (D) and
   survival (E). (F) Weight loss. Data from one of two independent
   experiments with similar results, n=3–5 mice per group. (G-J)
   Quantitative PCR for Ccl2 (G), Cxcl1 (H), Tnfa (I), and Lcn2 (J)
   transcripts in distal colon at day 8 post infection. Combined data from
   two independent experiments, n = 4–6 mice per group. (K, L) Intestinal
   pathology in colon of infected WT^LN/PP and ΔILC3^LN/PP mice in early
   (Day 8) and late (Day 14) timepoints. Images taken at 50x
   magnification. Scale bars, 500 μm. Combined data from two independent
   experiments, n = 5–6 mice per group.

   Although ILC3 were not required for the eventual generation of T cell
   responses and clearance of low dose infection, ΔILC3^LN/PP mice showed
   decreased tissue IL-22 and increased pathogen dissemination early in
   infection. Because IL-22 plays tissue protective functions, we asked
   whether ILC3 can provide protection at early stages of infection by
   decreasing tissue pathology and disease severity. Although both
   ΔILC3^LN/PP and WT^LN/PP animals cleared and recovered from infection,
   ΔILC3^LN/PP mice were more severely affected by it, especially at early
   stages. Low dose infection did not cause overt changes in animal
   appearance or significant changes in body weight at early stages in
   WT^LN/PP mice ([269]Figure 6F). In contrast, ΔILC3^LN/PP mice
   consistently showed increased macroscopic signs of disease, e.g.,
   transient loss of body weight ([270]Figure 6F), in agreement with
   systemic infection. They also showed increased markers of intestinal
   tissue inflammation at early timepoints. These included increased
   neutrophil infiltration ([271]Figure S11E, [272]F), as well as an
   increase in inflammatory tissue mediators, such as Ccl2, Cxcl1, Tnfa
   and Lipocalin-2 mRNA ([273]Figure 6G-[274]J) as well as an increase in
   fecal Lipocalin-2, an indicator of intestinal inflammation ([275]Figure
   S11G). Tissue pathology in WT^LN/PP mice increased gradually from early
   infection (Day 8) to peak infection (Day 14) ([276]Figure 6K, [277]L).
   In contrast, ΔILC3^LN/PP mice showed high histology scores even at
   early timepoints, and not so much at later timepoints, when both tissue
   pathology and fecal Lipocalin-2 levels were similar to WT^LN/PP mice
   ([278]Figure 6K, [279]L and [280]S11H), suggesting a specific loss of
   tissue protection at early stages of infection.

   Combined, our results demonstrate that although ILC3 are not required
   for pathogen clearance, they are needed for pathogen sequestration and
   tissue tolerance at early stages of infection.

Generation of primary T cell responses to Citrobacter rodentium in the
absence of ILC3

   ΔILC3^LN/PP mice clear low dose C. rod infection in the absence of ILC3
   ([281]Figure 6D), which suggests generation of protective
   antigen-specific T cell responses. To assess in more detail how ILC3
   affect primary antigen-specific T cell responses, we compared the
   kinetics of CD4 T cell responses during C. rod infection in ΔILC3^LN/PP
   and WT^LN/PP mice. In contrast to STOP^LN/PP animals, both WT^LN/PP and
   ΔILC3^LN/PP mice generated antigen-specific Th1, Th17, and Th22
   responses. However, the kinetics of these responses differed between
   the two groups. In control WT^LN/PP mice IL-17-producing Th17 cells and
   IL-22-producing CD4 T cells were induced and could be detected in the
   colon as early as day 8 post infection ([282]Figure 7A, [283]B). At Day
   14, in addition to Th17 and IL-22^+ CD4 T cells, the response contained
   large numbers of IL-17^negIFNγ^+ Th1 and IL-17^+IFNγ^+ Th1/17 cells
   ([284]Figure 7C, [285]D). The T cell response in ΔILC3^LN/PP mice
   followed similar general sequence of induction of effector cytokines.
   However, at Day 8, ΔILC3^LN/PP mice contained much fewer IL-22^+IL-17^+
   and IL-22^+IL-17^neg CD4 T cells than WT^LN/PP mice ([286]Figure 7A,
   [287]B) consistent with overall deficiency in IL-22 production
   ([288]Figure S11A, [289]C). In contrast, LP T cell responses did not
   differ significantly between ΔILC3^LN/PP and WT^LN/PP animals at Day 14
   ([290]Figure 7A-[291]D) and the overall levels of mucosal Il22 and
   Reg3g transcripts were similar between the two groups at this stage
   ([292]Figure S11I, [293]J). Interestingly, despite the presence of LNs
   and PPs, CD4 T cells from STOP^LN/PP animals lacked any detectable
   induction of IL-17, IL-22, or IFNγ at Day 8 ([294]Figure 7A-[295]D)
   underscoring the severe defect in generation of antigen-specific
   pathogen responses in the absence of both ILC3 and Th17 cells, as well
   as the crucial role of Th17 and IL-22-producing T cells in protection.

Figure 7. Effects of ILC3 on primary T cell responses following Citrobacter
rodentium infection.

   [296]Figure 7.
   [297]Open in a new tab

   (A-D) ΔILC3^LN/PP mice and LN/PP-sufficient controls were infected with
   low dose C. rod and effector CD4 T cell responses in colonic LP
   examined at early (Day 8) and late (Day 14) stages of infection by flow
   cytometry. Plots gated on CD4^+TCRβ^+ cells. Data from two independent
   experiments, n = 3–7 mice per group.

   We conclude that absence of ILC3 does not affect the overall robustness
   and functionality of the primary T cell response at late stages of
   infection. However, lack of ILC3 results in diminished early T cell
   responses, which suggests that ILC3 assist timely generation of optimal
   primary CD4 T cell responses against C. rod.

ILC3 are not required for generation of SFB-specific Th17 cell responses

   Intestinal Th17 cells are also generated under homeostatic conditions
   in response to commensal microbes ([298]33-[299]36). Segmented
   filamentous bacteria (SFB) are members of the endogenous mouse
   microbiota that induce intestinal Th17 cells ([300]25, [301]37). SFB
   have also been shown to activate ILC3 ([302]38) and ILC3 activation has
   been suggested to participate in generation of SFB Th17 cells
   ([303]39). ILC3 and Th17 cells have both been shown to control SFB
   levels ([304]40, [305]41), though their relative contribution has not
   been compared. ΔILC3 mice had more than 100-fold higher SFB levels than
   WT littermate (LM) controls and only slightly lower SFB levels than
   STOP mice ([306]Figure 8A). This suggests that although Th17 cells do
   contribute, SFB levels are predominantly controlled by ILC3. To examine
   whether ILC3 are required for induction of SFB-specific Th17 cells, we
   colonized SFB-negative WT, STOP and ΔILC3 mice with SFB and examined
   Th17 cell induction. WT and ΔILC3 mice showed similar induction of
   endogenous intestinal Th17 cells, in contrast to STOP^LN/PP animals
   ([307]Figure 8B, [308]C). To examine effects on the antigen-specific
   Th17 cell response, we adoptively transferred SFB-specific naïve 7B8
   TCR Tg CD4 T cells ([309]42) into WT, STOP and ΔILC3 recipient mice
   ([310]Figure 8D) and evaluated SFB-specific Th17 cell differentiation
   in the small intestine. SFB-specific CD4 T cells expanded significantly
   more in SI LP of ΔILC3 mice compared to WT animals, though not as much
   as in STOP mice ([311]Figure 8E, [312]F). This supports a role for ILC3
   in regulating commensal-specific T cell responses as previously
   reported ([313]26, [314]43). Despite differences in proliferation,
   SFB-specific CD4 T cells differentiated into Th17 cells similarly in
   the presence or absence of ILC3 in WT, STOP and ΔILC3 mice ([315]Figure
   8G, [316]H). These results demonstrate that ILC3 are required for
   control of luminal SFB levels at steady state but are dispensable for
   the generation of SFB-specific Th17 cells.

Figure 8. ILC3 are not required for induction of SFB-specific Th17 cells.

   [317]Figure 8.
   [318]Open in a new tab

   (A) Fecal SFB levels in ΔILC3 mice and littermate controls. (B, C)
   Endogenous Th17 cells in SI LP of SFB-colonized WT, STOP and ΔILC3
   mice. Plots gated on TCRβ^+CD4^+ cells. (D) Schematic of adoptive
   transfer experiments. (E, F) Expansion of CD45.1^+ SFB-specific 7B8 TCR
   Tg CD4 T cells in SI LP of SFB-colonized WT, STOP and ΔILC3 recipient
   mice. Plots gated on TCRβ^+CD4^+ cells. (G, H) IL-17A (GFP) expression
   in transferred CD45.1^+Vβ14^+ SFB-specific 7B8.IL-17A-GFP TCR Tg CD4 T
   cells in SI LP of SFB-colonized WT, STOP and ΔILC3 recipient mice.
   Plots gated on CD45.1^+TCRβ^+CD4^+ cells. Data from two independent
   experiments. n = 3–5 mice per group.

Discussion

   Despite an established role in host immunity and physiology, the
   contribution of ILC3 to immune phenotypes in the presence of T cells
   has been debated ([319]15-[320]17). Here we examined the differential
   role of ILC3 in mucosal infection in the presence of T cells and, in
   particular, Th17 cells. As the two major components of type 3 immunity,
   ILC3 and Th17 cells share effector cytokines and immune functions.
   However, the relative contribution of these two subsets has been
   difficult to compare directly in the absence of ILC3-specific depletion
   mouse models. Using ΔILC3 and ΔILC3^LN/PP mice we find that ILC3, and
   ILC3-derived IL-22, are crucial for the establishment and maintenance
   of epithelial barrier function and the activation of innate immune
   defense pathways that increase tissue tolerance to infection as well as
   control pathogen dissemination. Absence of ILC3 is accompanied with
   inability to activate intestinal epithelial defenses and ILC3-deficient
   animals showed increased pathology and pathogen dissemination at early
   stages of infection regardless of pathogen dose. At high doses of
   infection this leads to mortality prior to development of
   antigen-specific T cell responses. To our knowledge this is the first
   direct demonstration that ILC3 are necessary for survival in the
   context of functional adaptive immune system.

   Because ΔILC3^LN/PP mice can develop protective antigen-specific
   adaptive immune responses they can clear low dose C. rod infection
   despite increased early pathology. Thus, antigen-specific T cell
   responses are required to provide long-term protection by clearing the
   pathogen. However, in the absence of ILC3, Citrobacter-specific Th17
   cell responses were delayed and decreased in magnitude at early stages
   of infection. Whether this is due to decrease in tissue tolerance and
   what ILC3 functions mediate these effects will be important to
   investigate in future studies. We also find that pre-existing commensal
   (SFB)-specific Th17 cells can provide transient bystander protection,
   which however is insufficient for pathogen clearance. SFB induce Th17
   cells in the small intestine, and to a lesser degree in the large
   intestine, and we find that the bystander protection depends on IL-22
   and IL-17 production. Therefore, ILC3 and Th17 cells have complementary
   sequential roles in immunity to C. rod with ILC3 providing protection
   from excessive tissue pathology and Th17 cells providing pathogen
   clearance. This underscores the protective role of ILC3 in early stages
   of infection prior to development of Th17 cell responses and supports
   an evolutionary role for ILC3 in providing such protection.

   Our transcriptional analysis showed that ΔILC3 mice recover steady
   state Th17 and Th22 cells. Overall, based on bulk RNA-Seq, we detected
   recovery of IL-17-controlled programs, but a deficiency in
   IL-22-controlled genes in ΔILC3 mice at steady state. Similar results
   were obtained in ΔILC3^LN/PP mice. Therefore, despite the ability of
   ILC3 and Th17 cells to produce both cytokines, at steady state,
   ILC3-derived IL-22 and Th17 cell-derived IL-17 control intestinal
   homeostasis. This was further underscored by the fact that presence of
   5–10% of intestinal ILC3 in some ΔILC3^LN/PP mice was sufficient to
   maintain near steady state IL-22-regulated epithelial programs. At the
   same time low ILC3 numbers were insufficient to support the high levels
   of IL-22 required during infection. This argues that the relatively
   large number of ILC3 observed in intestinal lamina propria are needed
   to provide protection against acute immunological challenges, such as
   infection, but not so much for maintenance of the steady state.

   We also investigated whether ILC3 are required for a commensal-specific
   T cell response to SFB and control of the commensal. Both ILC3 and Th17
   cells can regulate luminal SFB levels ([321]40, [322]41) and
   ILC3-derived IL-22 can induce IEC SAA production to modulate SFB Th17
   cell differentiation ([323]25, [324]38). Our results show that despite
   some effect of IL-17 and Th17 cells, luminal SFB levels are mainly
   controlled by ILC3 and not so much by SFB-specific Th17 cells.
   SFB-specific Th17 cell differentiation was unperturbed (or slightly
   increased, presumably due to the higher SFB levels) in the absence of
   ILC3. Therefore, SFB-specific Th17 cell differentiation does not
   require ILC3 or ILC3-derived IL-22 and SFB-specific Th17 cells are not
   sufficient to control luminal SFB. Moreover, SFB-specific Th17 cells
   were generated in both STOP and ΔILC3 animals and therefore do not
   require mLN and PPs, in agreement with previous findings in
   LTα-deficient animals ([325]26, [326]44). Because mLN play a role in
   SFB-specific Th17 cell induction in WT mice ([327]38), commensal Th17
   cells are likely generated through several mechanisms. Indeed, we
   recently showed that SFB antigens acquired by enterocytes contribute to
   SFB-specific Th17 cell induction ([328]45). The relative contribution
   and biological significance of these pathways remain to be
   investigated. It will be interesting for example to examine how
   induction through different pathways affects the phenotype and function
   of commensal-specific Th17 cells.

   ILC3-deficient mice contained normal frequencies of Foxp3^+ Treg cells
   in SI and colon LP compared to WT littermates. However, we expect that
   composition and repertoire of Treg cells will differ between ΔILC3 and
   ΔILC3^LN/PP mice. The LN/PP-deficient and LN/PP-sufficient ΔILC3 mouse
   models we describe here can be instrumental in parsing out the role of
   LNs/PPs or ILC3 in intestinal Treg cell induction and heterogeneity in
   future studies. ILC3-deficient animals had a significant reduction in
   RORγt^+Foxp3^+ inducible Treg cells in both SI and colon LP. This
   agrees with recent studies reporting a role for ILC3, or other RORγt^+
   cells, in the induction of this Treg cell subset ([329]29-[330]31).
   However, despite significant decrease, RORγt^+Foxp3^+ Treg cells were
   present in the gut in the absence of ILC3, independent of LN/PP, and
   even after accounting for the presence of residual ILC3 in some
   animals. This warrants further study, considering the increased
   functional relevance of this population.

   The ILC3-deficient models we describe here have notable limitations.
   For example, generation of LN/PP-sufficient ΔILC3^LN/PP mice requires
   lethal irradiation and comes with all the drawbacks of bone marrow
   reconstitution, e.g., potential long-term changes in intestinal
   epithelial function and variable radio-sensitivity of intestinal immune
   cells. Some of these effects can be alleviated by increasing the period
   between bone marrow reconstitution and experimental analysis.
   Therefore, one of the most important limitations is the fact that this
   model does not allow investigation of early life events, such as role
   of ILC3 shortly after birth.

   Despite the above limitations, LN/PP-sufficient ΔILC3^LN/PP mice allow
   for studying of T cell responses in the absence of ILC3 in various
   contexts. Importantly this includes not only primary, but also
   secondary T cell responses, such as immunological memory, which will be
   exciting to investigate in future studies. Our work demonstrates that
   we can successfully uncouple ILC3 and Th17 cell function and increases
   the armament for studying relative contribution of type 3 immune cells
   in host physiology.

Materials and Methods

Study Design

   This study investigated non-redundant roles of ILC3 in mucosal
   protection by generating mouse models with ILC3-specific deletion in
   the context of existing T cell development and T cell differentiation.
   Gene-targeted mice, irradiation and bone marrow chimeras were used to
   generate ILC3-deficient models. Flow cytometry, bulk RNA-sequencing and
   in vivo experiments were used to identify ILC3-specific functions.
   Dose-dependent infection with Citrobacter rodentium was used to
   identify context-dependent functions of ILC3. All mice were bred and
   maintained under specific pathogen-free conditions at Columbia
   University Medical Center. All animals were SFB-positive unless
   specifically noted. Experiments were performed with littermate controls
   and genotypes were housed in the same cage to control for microbiota
   and cage effects. For experiments animals from different cages were
   randomized into groups based on genotype. Animals of both genders were
   used. Animals were generally 8–14 weeks of age, except for bone marrow
   chimera experiments, where animals were 18–25 weeks old. Experimental
   parameters, animal numbers, experimental replicates and statistical
   approaches are listed in the [331]Materials and Methods section or
   figure legends. Histology analysis was double-blinded, the rest of the
   study was not blinded.

Animals

   C57BL/6J, Ptprc (CD45.1), CD4-Cre, CAGG-Cre-ER™, 7B8 TCR Tg, RORγ-KO,
   RORγt^flox, IL-17-GFP mice were purchased from the Jackson
   Laboratories. 7B8 mice were bred to CD45.1 and IL-17-GFP mice at
   Columbia University to generate 7B8.CD45.1.IL-17-GFP animals. RORγ^STOP
   mice were generated by homologous recombination in C57BL/6 ES cells.
   The targeting vector generated an inversion of the Rorc genomic
   sequence containing Exons 3–6 surrounded by two pairs of LoxP and
   LoxP2272 sequences in opposite orientation in intron 2 and intron 6
   ([332]Figure S1A). The targeting vector and the targeting itself were
   confirmed by sequencing and Southern blot. To generate ILC3-deficient
   and CD4 T cells replete mice RORγ^STOP mice were crossed to CD4-Cre
   mice to obtain ΔILC3 mice.

Generation of LN/PP-sufficient ΔILC3 mice

   To generate ΔILC3 mice with LN/PP we bred RORγt^flox with CAGG-Cre-ER™
   mice, in which ubiquitous Cre expression is induced by tamoxifen. To
   delete RORγ-expressing ILC3 and T cells, 3-week-old
   RORγt^flox/CAGG-CreER™ mice were treated with 0.5 mg/animal of
   tamoxifen (Sigma-Aldrich #T5648) diluted in sunflower oil
   (Sigma-Aldrich) every other day by oral gavage for a total of 4
   administrations. At 8 weeks old, the mice were lethally irradiated (two
   times 550 rad four hours apart) and used as recipients of bone marrow
   from WT, STOP and ΔILC3 mice. Chimeric animals were analyzed 10–15
   weeks after BM reconstitution.

SFB quantification in feces

   SFB colonization levels were confirmed by quantitative polymerase chain
   reaction (PCR) of fecal bacterial DNA and normalized to levels of total
   bacteria (UNI) as previously described ([333]46).

Lamina propria cell isolation

   Lamina propria (LP) lymphocytes, intracellular cytokine staining, and
   transcription factor staining were performed as previously described
   ([334]25). Briefly, small or large intestines were dissected, washed in
   PBS to remove the fecal contents and cut into 1 cm pieces. Intestinal
   pieces were then incubated at 37º C in dissociation buffer (HBSS, 5 mM
   EDTA, 10 mM Hepes and 2.5% FCS) two times for a total of 20 min. For
   the isolation of lamina propria lymphocytes, tissues were cut to 1 mm^2
   using razor blades and incubated at 37ºC in digestion buffer (RPMI, 0.5
   mg/mL Collagenase D and DNAse, 50 U/mL Dispase and 5% FCS) for 20
   minutes. After vortexing, cells were filtered through 100 μm cell
   strainer and undigested tissue collected for two more digestions.
   Combined digestions were filtered through 40 μm cell strainer and
   resuspended in 40/80% Percoll gradient, centrifuged at 2500 rpm without
   brake for 20 min at room temperature, and LPL collected at the
   interphase. LP lymphocytes were washed in media and cells were
   resuspended in FACS buffer for staining and further analysis.

In vitro LP CD4 T cell IL-22 production

   SI LP CD4 T cells were purified by cell sorting
   (B220^negCD11c^negTCRβ^+CD4^+). Sorted CD4 T cells (5× 10^4/well) were
   cultivated with plate bound anti-CD3 (1 µg/mL) and soluble anti-CD28 (2
   µg/mL) in 96-well flat-bottom plates. T cell supernatants were
   collected after 72 hours for assessment of IL-22 production by ELISA.

In vivo depletion

   For in vivo depletion, anti-CD4 monoclonal antibody (Clone: GK1.5,
   BioXCell) was administered intraperitoneally (200 µg/mouse) every other
   day starting on day 0 and ending on day 12 of infection. Anti-IL-22
   monoclonal antibody (Clone: 8E11, Genentech) was administered
   intraperitoneally (200 µg/mouse) every other day starting on Day 0 and
   ending on Day 10. Anti-IL-17 monoclonal antibodies (IL-17A clone:17F3
   and IL-17F clone MM17F8F5, BioXCell) were administered
   intraperitoneally every other day (300 µg/mouse) starting on Day 0 and
   ending on Day 10.

In vivo IL-22 supplementation

   RORγ-KO chimera mice were injected intraperitoneally every three days
   during Citrobacter rodentium infection (Day 0, 3, 6, 9, 12 and 15) with
   50 µg/mouse of mouse IL-22-Fc (PRO312045) or control Fc conjugate for
   PRO312045 ([control] Ragweed:9652 10D9.W.STABLE mIgG2a) (Genentech).

Adoptive T cell transfers

   WT, STOP and ΔILC3 mice were colonized with SFB by oral gavage. Two
   weeks after SFB colonization, 1× 10^6 naïve 7B8 CD4 T cells from
   spleens and mLN of 7B8/CD45.1/IL-17AGFP reporter mice (7B8 CD45.1
   IL-17A^GFP) were purified by MACS (95–98% purity) and transferred
   intravenously into each recipient. Th17 induction in SI LP was
   evaluated 10 days after transfer.

Bone-marrow (BM) chimeras

   For BM chimera generation, 6-week-old RORγ-KO or RORγt^flox/CAGG-CreER™
   recipient mice were lethally irradiated with two doses of 550 rad given
   4 hours apart. The same day, recipients were injected intravenously
   with 10 × 10^6 BM cells isolated from femurs and tibias of
   corresponding donors (WT, STOP, or ΔILC3 mice). Animals were kept on
   antibiotics in the drinking water (1g/L Ampicillin, Neomycin 1g/L and
   0.5g/L Vancomycin) for 2 weeks post irradiation. After 2 weeks,
   antibiotics were removed, and a fecal transplant of SPF feces performed
   to recover endogenous microbiota. BM recipients were then allowed to
   rest for 12–16 weeks before use.

Citrobacter rodentium infection

   For Citrobacter rodentium infection, the DBS100 strain was a gift from
   Daniel Mucida. Bacteria were grown overnight in LB-broth and 5 × 10^7
   CFU (low dose) or 1 × 10^9 CFU (high dose) administered to recipient
   mice by oral gavage. To monitor infection levels, feces were collected
   at different time points post infection, weighed, homogenized with a
   bead beater in PBS and plated in serial dilutions on MacConkey agar
   plates for CFU calculation. For tissue CFUs, spleen, liver or distal
   colon tissue pieces were collected, washed, weighed, homogenized in PBS
   with a bead beater and plated in serial dilutions on MacConkey agar
   plates. Individual colonies were counted after incubation at 37º C
   overnight.

IL-22 and fecal Lipocalin ELISA

   For fecal Lipocalin, fecal pellets were collected and homogenized in
   PBS (1 mg of feces per 10 µL of buffer) containing protease inhibitor
   cocktail (Roche). Lipocalin-2 levels were measured using a Lipocalin-2
   specific ELISA kit (R&D) following manufacturer’s instructions. For
   IL-22 protein, ELISA was performed on supernatants from sorted SI LP
   CD4 T cells restimulated with anti-CD3/CD28 (1 µg/mL) for 72 hours with
   an IL-22 ELISA kit (R&D) using manufacturer’s protocol.

Quantitative real-time PCR (Q-PCR)

   Total mRNA was isolated from 0.5 cm pieces of terminal ileum or distal
   colon, sorted CD4 T cells, or IECs from ileal scrapings after
   homogenization in Trizol using standard protocols. For cDNA synthesis
   we used SuperScript III First Strand Synthesis System (Invitrogen).
   Q-PCR was performed on LightCycler 480 (Roche)/ Primers are listed in
   [335]Table S3. Relative expression was calculated by the ΔΔCt method
   with normalization to Gapdh. SFB levels were normalized to a positive
   WT sample for better visualization.

RNA-sequencing and reads preprocessing

   Extracted RNA was subjected to RNA-seq library preparation and
   sequencing by Novogene and JP Sulzberger Columbia Genome Center.
   Briefly, RNA-seq libraries were constructed using Illumina Stranded
   mRNA Prep Ligation kit [Illumina 20040532] following manufacturer’s
   instructions and resulting libraries were then sequenced on NovaSeq
   6000 using 2×150bp setting or on Element AVITI using 2×75bp setting to
   generate 39.1 ± 5.4 million paired-end reads per sample. Raw reads were
   then processed by Cutadapt v2.1 with following parameters
   “--minimum-length 30:30 -u 15 -u −5 -U 15 -U −5 -q 20 --max-n 0
   --pair-filter=any” to remove low-quality bases and TruSeq adapters and
   remaining reads passing quality filtering were then analyzed by
   Kallisto v0.46.1 ([336]47) with default setting to quantify gene
   expression based on GRCm38 M. musculus reference transcriptome
   sequences.

Expression profile ordination and differential expression analysis

   Gene expression profiles sample were projected for ordination analysis
   by principal component analysis (PCA) or multidimensional scaling (MDS)
   and outlier samples were then removed in downstream analysis. To
   identify genes significantly up- or down-regulated in three conditions
   (WT, STOP and ΔILC3), reads counts of pseudo-alignment generated by
   Kallisto were analyzed by DESeq2 ([337]47) with default setting and
   Benjamini-Hochberg procedure was applied to correct P-values.
   Significantly differentially expressed genes (DEGs) (adjusted P-value <
   0.01 and fold-change > 2) between condition WT and STOP were classified
   based on their expression pattern in ΔILC3 samples and DEGs were then
   subjected to manual inspection to correct misclassification.

Pathway enrichment analysis and gene set enrichment analysis (GSEA)

   To identify enriched pathways of down-regulated genes in ΔILC3 CD4 T
   cells, DEGs were analyzed by ToppGene Suite ([338]48) and enriched GO
   and KEGG pathways were visualized by R v4.1.2. To better characterize
   DEGs identified in IEC RNA-seq dataset, GSEA was performed by fgsea
   ([339]49) based on expression profile of WT and ΔILC3 IECs. IL-22
   controlled genes identified in IL-22RA1^−/− mice were from ([340]50).

Histological analysis

   Colons were excised from WT^LN/PP and ΔILC3^LN/PP infected with low
   dose C. rodentium at day 8 and day 14 post infection. A distal colon
   segment (1 cm) was removed, fixed with 10% formalin overnight and then
   transferred to 70% ethanol. The tissue was embedded in paraffin and
   sections (4 μm) were cut, mounted on slides, and stained with
   hematoxylin and eosin. The severity of colitis was scored using a
   modification of the scheme used in ([341]51). Scoring was performed in
   a double-blind fashion, and four features were assessed for a total
   possible score ranging between 0 and 9. The features scored were
   epithelial injury (0–3), inflammation (0–3), goblet cell depletion
   (0–2), and crypt hyperplasia (0–1).

Statistical analysis

   All data are represented as mean ± standard deviation. Unless otherwise
   noted in the figure legend, statistical differences between groups were
   carried out using unpaired t-test with or without Welch’s correction as
   appropriate. Statistical tests used are also listed in the raw data
   file in the [342]supplementary material. p˂0.05 was considered
   statistically significant. Statistical analysis was performed using
   GraphPad Prism version 10.0 for Windows (GraphPad Software). * p <
   0.05, ** p < 0.01, *** p < 0.005, **** p < 0.001, ns, not significant
   (p ≥ 0.05).

Supplementary Material

   Supplementary Material

   Figure S1. STOP/CD4 mice have normal T cell development

   Figure S2. STOP/CD4 mice have intestinal Th17 cells

   Figure S3. STOP/CD4 mice specifically lack ILC3

   Figure S4. Intestinal T and B cell subsets in STOP/CD4 mice

   Figure S5. ILC3 control steady state IEC function

   Figure S6. ILC3 participate in early protection from Citrobacter
   rodentium

   Figure S7. ΔILC3^LN/PP mice are a lymph node/Peyer’s patch sufficient
   model of ILC3 deficiency

   Figure S8. ILC3 control IL-22 program in IEC in the presence of LN/PP

   Figure S9. A small number of residual ILC3 can support steady state IEC
   function

   Figure S10. ILC3 are required for induction of appropriate epithelial
   responses during infection

   Figure S11. ILC3-derived IL-22 is crucial at early stages of
   Citrobacter rodentium infection

   Table S1: Reagents and resources

   Table S2: Mice

   Table S3: Quantitative PCR Primers
   [343]NIHMS2036404-supplement-Supplementary_Material.docx^ (11.5MB,
   docx)

Acknowledgments: