Group 3 innate lymphocytes make a distinct contribution to type 17 immunity in bladder defence

Summary Bladder infection affects a hundred million people annually, but our understanding of bladder immunity is incomplete. We found type 17 immune response genes among the most up-regulated networks in mouse bladder following uropathogenic Escherichia coli (UPEC) challenge. Intravital imaging revealed submucosal Rorc+ cells responsive to UPEC challenge, and we found increased Il17 and IL22 transcripts in wild-type and Rag2−/− mice, implicating group 3 innate lymphoid cells (ILC3s) as a source of these cytokines. NCR-positive and negative ILC3 subsets were identified in murine and human bladders, with local proliferation increasing IL17-producing ILC3s post infection. ILC3s made a more limited contribution to bladder IL22, with prominent early induction of IL22 evident in Th17 cells. Single-cell RNA sequencing revealed bladder NCR-negative ILC3s as the source of IL17 and identified putative ILC3-myeloid cell interactions, including via lymphotoxin-β-LTBR. Altogether, our data provide important insights into the orchestration and execution of type 17 immunity in bladder defense.

Type 17 immune gene network is the major transcriptional response post bladder infection Group 3 innate lymphoid cells (ILC3s) identified in mouse and human bladder ILC depletion in Rag2 À/À mice resulted in increased bladder bacterial load IL17 production by ILC3 shape bladder macrophage activation and polarization

INTRODUCTION
Bacterial infection of the lower renal tract (cystitis) is common, affecting half of women at some point in their lifetime, with significant associated socioeconomic cost (Chertow et al., 2005;Foxman, 2014). The majority of uncomplicated infections are caused by uropathogenic Escherichia coli (UPEC) that can also ascend to the kidneys causing pyelonephritis and renal scarring (Foxman, 2014;Svensson et al., 2011). Tissue-resident immune cells are vital for initiating and propagating local immune responses, but our understanding of the nature and function of these cells in the bladder, particularly in humans, is incomplete.
In recent years, innate lymphoid cells (ILCs) have been identified in a number of environment-facing organs, including the gastrointestinal tract, skin, and lung, as the innate counterparts of T cells, providing an early source of cytokines that have the potential to shape innate and adaptive immune responses (Klose and Artis, 2020;Panda and Colonna, 2019). Group 3 ILCs are dependent on the transcription factor RORgt for their development (Eberl et al., 2004;Sanos et al., 2009) and secrete canonical Th17 cytokines (IL22, IL17, and GM-CSF) in response to stimulation with myeloid cell-derived cytokines such as IL1b, IL23, and TL1A (Melo-Gonzalez and Hepworth, 2017). In the intestine ILC3-derived IL22 contributes to the maintenance of epithelial stem cells (Aparicio-Domingo et al., 2015;Lindemans et al., 2015) and increases epithelial production of anti-microbial peptides (AMPs) to protect against enteric pathogens (Ahlfors et al., 2014;Cella et al., 2009;Guo et al., 2014). Human primary urothelial cells have been shown to express both the membrane-associated IL22 receptor subunit, IL22RA1, and the secreted soluble form IL22RA2 and to respond to ex vivo IL22 stimulation by producing S100A9 and lipocalin-2 (LCN2) (Le et al., 2014), two important AMPs. However, the cellular sources of IL22 in the bladder have not been delineated, and the role of IL22 in bladder defense in the context of bacterial infection is unknown. In the intestine and lung, ILC3s also produce IL17 and GM-CSF in the context of infection or inflammation (Buonocore et al., 2010;Castro-Dopico et al., 2020;Mortha et al., 2014;Pearson et al., 2016;Van Maele et al., 2014). In the bladder, IL17 is protective in murine models of UPEC-associated cystitis, with gdT cells suggested as a major source of this cytokine (Sivick et al., 2010), in keeping with a previous study showing increased susceptibility of gdT cell-deficient mice to acute cystitis (Jones-Carson et al., 1999). Interestingly, higher IL17 levels in infected bladders of
Here we sought to investigate type 17 immunity in bladder defense, because our unbiased transcriptomic analysis of UPEC-infected bladders identified this axis as one of the most up-regulated networks in acute bacterial infection. We validated an increase in bladder Il17 and IL22 transcripts in acute UPEC cystitis in both wild type (WT) and Rag2 À/À mice, implicating ILC3s in type 17 responses in the bladder. A network of Rorc+ cells were visualized in the submucosa using intravital, two-photon imaging, and were sessile in homeostasis, but showed increased motility post UPEC challenge. We confirmed that ILC3s were present in human and murine bladders, including both natural cytotoxicity receptor (NCR)+ and NCR negative (neg) subsets. Rorc-deficient mice had increased bacterial counts in the bladder, with reduced Il17 and Il22 transcripts, as well as a reduction in neutrophil recruitment and impaired monocyte differentiation into tissue macrophages. IL22R-deficient mice had no acute impairment of bacterial clearance but showed a reduction in bladder AMPs and in urothelial cell proliferation following UPEC challenge. ILC depletion in Rag2 À/À mice resulted in increased bladder bacterial load and a marked reduction in Il17 transcripts and an increase in an M2 macrophage transcriptional signature in the bladder, emphasizing the importance of bladder ILC3s in shaping tissue macrophage polarization. Single-cell RNA sequencing of murine bladder post UPEC challenge confirmed the relative contributions of Th17, ILC3, and gdT cells to IL17, IL22, and GMC-CSF production in the bladder and also identified IFNg and LTb secretion as potential mediators of ILC3-macrophage interactions. Altogether, our data show distinct populations of tissue type 17 immune cells that are primed for rapid responses to bacterial challenge, orchestrating epithelial and myeloid cell function for bladder defense.

Induction of type 17 immunity during bladder infection
To gain an unbiased insight into immunological pathways that potentially play a role in the local immune response to bladder infection, we challenged wild type (WT) C57BL/6 mice with UPEC and performed RNA sequencing (RNA-seq) on bladder tissues. We found that several thousands of genes were significantly upand down-regulated following infection ( Figure 1A). Analysis of the top 100 most up-regulated genes implicated two major interactions nodes; IL1b-associated Th17 immunity ( Figures 1B and 1C) and, related to this, Cxcl1-driven neutrophil recruitment ( Figure 1B). Gene set enrichment analysis confirmed an increase in genes associated with cellular responses to IL1 and in epithelial responses to IL17 and IL22 ( Figure 1D). In (E) qPCR of Th17 cytokines in C57BL/6J bladders day 1 (blue) and day 2 post UTI (red) (n = 4-6 per group) relative to uncatheterized bladders (gray).
(F) qPCR of Th17 cytokines in C57BL/6J (blue) and Rag À/À (red) bladders day 1 post UTI (n = 4-6 per group) relative to uncatheterized bladders (gray iScience Article an independent experiment and a previously published dataset (GEO:GSE68220), we confirmed a significant increase in Il17, Il22, and, to a lesser extent, Csf2 transcripts, canonical Th17 cytokines, in the bladder in WT mice at early time points (24-48 h) following challenge with UPEC ( Figures 1E and S1A), when adaptive immunity would not be expected to make a major contribution to the immune response. Il17 and Il22 were also increased in the bladders of Rag2 À/À mice that are deficient in gd T cells and Th17 cells ( Figure 1F), suggesting that ILC3s can act as a source of these cytokines during bacterial infection in the bladder.
IL17 promotes immune cell infiltration into tissues both via the induction of monocyte and neutrophil recruiting chemokines (Veldhoen, 2017) and also the secretion of matrix metalloproteinases (MMPs), including MMP1, 2, 3, 8, 9, and 13 (Koenders et al., 2005), which make the extracellular matrix more accessible to incoming cells. We observed an increase in Cxcl1, Cxcl5, and Ccl1 in bladder post UTI ( Figure 1B), as well as in Mmp3 and Mmp9 (Figure S1B), consistent with increased IL17 activity.
Intravital imaging using a GFP reporter mouse for the canonical type 17 immune cell transcriptional factor Rorc demonstrated the presence of many Rorc+ cells in the bladder, particularly within the submucosa ( Figure 1H), that were frequently located adjacent to blood vessels ( Figures 1H, 1I, and S1C and Video S1). These cells were largely sessile in homeostasis but showed increased movement following UPEC challenge ( Figures 1J and S1D and Video S2). Together, these data show that the bladder is prearmed and populated with a network of cells capable of mounting a rapid type 17 response following bacterial challenge.
NCR + and NCR À ILC3s present in murine and human bladder and increase during infection Th17 cell and gd T cells have previously been described in the bladder (Sivick et al., 2010), but given the increase in Il17 and Il22 transcripts at early time points post infection, even in T cell-deficient mice, we sought to characterize bladder ILC3s, as previous reports had described only CD4 + ILC3 (Zychlinsky Scharff et al., 2019). Flow cytometric analysis of tissue homogenates identified lineage negative (Lin neg ), CD127+-RORgt+ ILC3 in murine bladders, as well as GATA3+ ILC2 (Figures 2A, S2A and S2B). ILC3s can be further subdivided into lymphoid tissue-inducer (LTi)-like cells and natural cytotoxicity receptor (NCR)-positive subsets based on surface marker expression (Melo-Gonzalez and Hepworth, 2017). In human and murine bladder we identified both NCR + and NCR neg subsets ( Figures 2B, 2C, and S2B-S2D). Confocal imaging confirmed a network of RORgt-positive cells in mouse and human bladder, which included CD3 neg ILC3s  (Table S1) and corresponding image of bacterial growth on agar plates; 1:30 dilution (right panel). N = 6-7 mice per group.  The bladder comprises a number of layers; from the outer muscularis through to the mucosa, submucosal muscularis, lamina propria, and inner urothelium. We found that ILC3s were localized to the submucosal lamina propria ( Figure 2F). Following intravesical UPEC challenge, we observed an increase in ILC3 numbers ( Figures 2G and 2H), suggesting either tissue recruitment or local proliferation. Consistent with the latter, there was a significant increase in the proportion of Ki67+ ILC3s evident in the bladder post infection ( Figure 2I). There was also an increase in Th17 cell and gd T cells post UTI ( Figure S2G).

Rorc-deficient mice have increased susceptibility to bladder infection
We next sought to address the functional importance of type 17 immunity in bladder defense using Rorcdeficient mice that lack ILC3, as well as gdT cells and Th17 cells, and therefore have a significantly diminished capacity to produce Th17 cytokines ( Figures 3A and S3A). At 48 h following intravesical UPEC challenge, we observed a significant increase in bladder colony forming units (CFUs) in Rorc-deficient mice relative to WT mice (Figures 3B and S3A and Table S1). Il17 and Il22 transcripts were reduced in Rorc À/À bladders following infection compared with controls, with similar levels of Csf2 ( Figure 3C). Intracellular cytokine staining in WT bladders following in vitro stimulation identified IL17 production in gd T cells in both homeostasis and infection ( Figures 3D and S3B), implying that they provide a basal level of defense. In contrast, there was little cytokine production evident in Th17 cells in homeostasis, but a marked induction of IL22 and IL17 ( Figures 3D and S3B), even at this early time point post infection before a primary adaptive response would have time to occur, suggestive of a recall response. There was also little basal cytokine production in ILC3s but a 20-fold increase in the proportion of IL17-producing ILC3s post UPEC challenge, with a more modest increase in IL22 production ( Figures 3D and S3B). Similarly, these cytokines were increased in whole bladder lysates post UPEC challenge by ELISA (S3C). This pattern of cytokine production is in sharp contrast to ILC3s in the gastrointestinal tract that make little contribution to tissue IL17 during inflammation, but rather act as an early source of IL22 (Lee et al., 2015).
In the gut IL22 stimulates AMP production by intestinal epithelial cells to promote barrier integrity (Ahlfors et al., 2014;Cash et al., 2006;Guo et al., 2014;Sano et al., 2015). Consistent with a similar paradigm in bladder, we observed a significant reduction in Lcn2 and Reg3g ( Figure 3E). In keeping with the known effects of IL17, there was a reduction in immune cell infiltrates into the bladder in RorcÀ/À mice, particularly neutrophils, and also F4/80+ macrophages ( Figure 3F). Infiltrating monocytes entering tissues progressively lose Ly6C expression and increase MHC-II expression as they become tissue macrophages, the so-called waterfall phenomenon, first described in the intestine (Tamoutounour et al., 2012) (Mora-Bau et al., 2015. In RorcÀ/À mice, Ly6C+ MHCII neg cell numbers were similar to WT counterparts but we observed a reduction in MHCII+ cells, consistent with an impaired maturation of infiltrating monocytes ( Figure 3G).
Together, these data show that type 17 immune cells are important for coordinated defense in the bladder early in the time course of infection, enabling myeloid cell infiltration and maturation, as well as epithelial production of AMPs.

IL22 promotes epithelial proliferation and AMP production in the bladder post infection
To further explore the effect of IL22 on the bladder epithelium, we challenged IL22RA1-deficient mice with UPEC. There were no differences in bladder CFUs at 24 or 48 h between WT and IL22ra1 À/À mice (Figures 4A, S4A-S4B and Table S1), and similar immune cell infiltration was observed ( Figure 4B). However, there was a significant attenuation of AMP transcripts in IL22ra1 À/À bladders post infection ( Figure 4C). Gene set enrichment analysis (GSEA) of bladder bulk RNA-seq data indicated a significant enrichment in E2F target genes and G2M checkpoint pathway genes in WT compared with IL22ra1 À/À bladders (Figure 4D). E2F transcription factors are important regulators of genes involved in G1 to S-phase progression, many of which showed decreased expression in IL22ra1 À/À bladders ( Figures 4E and 4F). Consistent with an  (Table S1) and corresponding image of bacterial growth on agar plates; 1:30 dilution (right panel). N = 7-8 mice per group.
(C) Heatmap of AMPs from RNA sequencing of bladders infected with UTI89 in C57BL/6N (n = 4) or IL22ra À/À (n = 4) mice 24 h after challenge. Data represent four biological replicates per group (IL22ra À/À and C57BL/6N). iScience Article important role for IL22 in epithelial barrier repair in the bladder, we observed fewer Ki67-positive cells within the bladder epithelium in IL22ra1 À/À mice at 24 h following infection ( Figure 4G) as well as a reduction in Mki67 transcripts ( Figure 4H).
ILC depletion in Rag2 À/À mice leads to reduced Il17 and increased infection severity To investigate whether ILC3s play a significant role in bacterial defense in the bladder, and to further probe their contribution to epithelial-maintaining IL22 production versus IL17 production in the absence of T cells, we challenged Rag2 À/À mice with UPEC following ILC depletion with an anti-Thy1 antibody (Figures 5A and S5A-S5C). We observed a significant increase in CFUs in ILC-depleted mice post infection, consistent with an important role in bladder defense (Figures 5B, S5D and Table S1). Notably, ILC-deficient mice had a significant reduction in bladder Il17 transcripts, as observed in the Rorc À/À mice, but in contrast to Rorc À/À mice, we observed little difference in bladder Il22 in the absence of ILC3 ( Figure 5C), suggesting that the major contribution of ILC3s is in IL17 production rather than IL22. In keeping with this conclusion, there was a similar abundance of bladder AMP transcripts in anti-Thy1-treated and control mice ( Figure 5D). Neutrophil infiltration was also similar in WT and ILC-depleted mice, but there was a significant reduction in F4/80 hi macrophages in the latter ( Figure 5E). Ly6C hi monocytes were increased in ILC-depleted mice, with reduced MHCII hi cells, again suggestive of impaired progression of monocytes down the ''waterfall'' ( Figure 5F), as we had observed in Rorc À/À mice.
To further understand the role of ILCs in bladder defense, we performed RNA-seq on bladder samples taken from Rag2 À/À mice treated with isotype control or anti-Thy-1 antibody at 24 h following infection. Il17 transcripts were significantly reduced in ILC-depleted bladders, and there was also a more variable reduction in Il22 ( Figure 5G), and correspondingly little difference in AMP transcripts ( Figure 5H). GSEA showed a reduction in a number of inflammatory pathways in the absence of ILCs, including Interferon gamma response and Interferon alpha response pathway genes, while cell proliferation-associated pathways were increased ( Figure 5I). Cellular deconvolution indicated an increase in M2 macrophage-associated genes in ILC-depleted bladders, in keeping with a reduction in IFN-g stimulation ( Figure 5J) and conversely, enrichment for an IL17-stimulated macrophage gene signature in isotype-treated compared with anti-Thy1-treated bladders ( Figure 5K).
Altogether, these data indicate that, in contrast to intestinal ILC3s, IL22-mediated epithelial cross talk is a less important part of their function in the context of bacterial challenge in the bladder; rather, IL17-mediated myeloid cell stimulation plays a more prominent role in this context.

Bladder macrophages produce ILC3-stimulating cytokines during infection
Myeloid cell-ILC3 interactions are of critical importance in the gastrointestinal tract, with macrophages producing cytokines such as IL23, IL1b, and TL1A that stimulate ILC3 cytokine production (Longman et al., 2014). In uninfected mouse bladder, we observed a dense network of MNPs, poised to respond to infectious challenge ( Figure 6A), with ILC3s and RORgt+ T cells found in close proximity to bladder macrophages ( Figure 6B). Following UPEC challenge, there was an increase in Il1b, Il23, and Tnfsf15 transcripts in the bladder, with the potential to stimulate type 17 cytokines ( Figure 6C).
To determine if macrophages were a source of these transcripts, we flow sorted bladder macrophages and performed bulk RNA-seq. This confirmed a significant increase in Il1b at 24 h following infection, with minimal induction of Il23a evident ( Figure 6D), as well as higher levels of monocyte and T cell chemoattractants iScience Article Ccl4 and Ccl5 ( Figure 6E). In keeping with the conclusion that macrophage IL1b is an important activating stimulus for IL17-producing cells, we observed a strong positive correlation between bladder Il1b and Il17 transcripts ( Figure 6F). This correlation was less evident with Il23 ( Figure 6F). Furthermore, IL23R-deficient mice showed no increase in bladder bacterial load post UPEC challenge, although there was a variable reduction in Il17 transcripts compared with WT counterparts ( Figure S6 and Table S1).
Of note, bladder macrophages also expressed Csfr2a and IL17ra transcripts ( Figure 6G), suggesting a capacity to respond to Th17 cytokines and the potential for cross talk between bladder macrophages and Th17, gd T cells, and ILC3s. Cellular responses to IL-17A and IL-17F require the ubiquitously expressed IL-17RA paired with the inducible IL-17RC (Toy et al., 2006); Il17rc transcripts were also detectable at a lower level in bladder macrophages. To model the potential effects of IL17 on macrophage uptake of UPEC in the bladder, we quantified the phagocytosis of fluorescently labeled UPEC by bone marrow-derived macrophages in vitro and observed that the addition of exogenous IL17 augmented bacterial phagocytosis (Figure 6H), suggesting a synergistic relationship between bladder macrophages and ILC3s, and indeed other IL17-producing lymphocytes, whereby IL1b production by bladder macrophages promotes IL17 production by ILC3s and Th17/gd T cells, which in turn acts on macrophages to improve their defensive capabilities.

scRNAseq of infected bladders maps cytokine production and myeloid interactome of type 17 immune cells
To further explore the interaction between type 17 immune cells and MNPs in the bladder beyond Th17 cytokines we performed droplet-based RNA sequencing (scRNAseq) on infected and uninfected bladders using the10x Genomics platform. Fourteen clusters of T cells and innate lymphocytes were evident, which we annotated according to canonical marker gene expression, including Th17 cells, gdT cells, and ILC3s ( Figures 7A and S7A-S7D). Notably, cells the Th17 cluster showed a marker expression profile associated consistent with a tissue-resident memory T cell phenotype, being Cd44+, Ccr7-, and Sell-negative, with some Itage (CD103)+ cells ( Figure 7A). Following infection, Th17 cells showed the greatest expression of Il22 transcripts, with some contribution from ILC3s but little detectable Il22 in gdT cells ( Figure 7B). Induction of Il17a post infection was evident in ILC3s, Th17, and gdT cells, with the highest level of expression in ILC3s ( Figure 7B). Consistent with this, reactome pathway analysis demonstrated enrichment of ''interleukin signaling'' pathway genes in all three cell types, but particularly prominent in ILC3s, with Il17a and Il17f among the most enriched in this pathway ( Figure S7E). When considering the ILC3 cluster in isolation, two subsets of cells were evident, one that was Ncr1-negative and the other containing many Ncr1+ cells (B) Colony-forming units per bladder 24 h after infection with UTI89 in Rag2 À/À + isotype (gray) and Rag2 À/À + anti-Thy1 (red) mice (left panel) (Table S1) and corresponding image of bacterial growth on agar plates; 1:400 dilution (right panel). N = 7-8 mice per group.
In our dataset, we were also able to identify neutrophils, monocytes, and two broad macrophage clusters, which others have described using flow cytometric analysis (Lacerda Mariano et al., 2020), annotated as F4/ 80+ or CX3CR1+ ( Figure 7F). To investigate how these myeloid cell subsets might interact with type 17 lymphocytes in the bladder we used CellPhone DB, a platform that predicts cellular interactions based on receptor and ligand expression (Vento-Tormo et al., 2018). This identified several additional mechanisms by which these cell types may interact during bladder infection beyond IL17. Notably, IFNG-IFNGR2-mediated interactions between ILC3s and monocytes and macrophages were significantly increased in the context of UTI, and specific to ILC3s ( Figure 7G), consistent with the decrease in Interferon gamma response pathway genes we observed in ILC-depleted Rag2 mice ( Figure 5I). Ltb (encoding lymphotoxin [LT]-b)-and Tnfsf12 (encoding TWEAK)-mediated interactions were also increased in ILC3s during UTI (Figure 7G). Our analysis of bulk RNA-seq data confirmed a significant increase in Ltb in UPEC-infected bladders, with a decrease observed in ILC-depleted Rag2 À/À mice ( Figure 7H). Confocal imaging confirmed the expression of both Ifng and Ltb on RORgt-positive, CD3 neg ILC3s in mouse bladders ( Figures 7I and 7J). In gd and Th17 T cells, TNF-mediated interactions were increased in UTI, with F4/80+ macrophage secretion of CCL7 predicted to attract CCR5-expressing gd and Th17 T cells ( Figure 7G). F4/80+ cells also showed increased CXCL16-mediated interactions with CXCR6-expressing ILC3, gd, and Th17 T cells ( Figure 7G), identifying tissue macrophages as key orchestrators of type 17 immunity in the bladder.

DISCUSSION
Previous studies have largely approached the question of which pathways, cytokines, or immune cells are important in bladder infection in a hypothesis-driven manner, measuring or knocking out a favored cytokine (like IL17), or a particular cell type based on prior knowledge cytokine (Jones-Carson et al., 1999;Sivick et al., 2010;Zychlinsky Scharff et al., 2019). Here, we analyzed bulk bladder transcriptomic data in an unbiased way and asked which pathways were most up-regulated in the context of infection, identifying type 17-associated transcripts (including Il22, Il17, Rorc) as the major axis induced post UPEC challenge. Our analysis places this pathway at the heart of the host response to bacterial challenge in the bladder. Notably, IL22 and IL17 have also been detected in human urine in the context of candida UTI (Ahmadikia et al., 2018), supporting the conclusion that this axis may play a similarly important role in human bladder infection. An increase in bladder Il17 and IL22 transcripts was evident in both WT and T cell-deficient mice implicating ILC3s as contributors to type 17 responses, in addition to the previously described contribution of gd T cells to bladder IL17 (Sivick et al., 2010). There is a single previous description of murine bladder ILC3s, which identified CD4 + ILC3s (Zychlinsky Scharff et al., 2019), but here we show that NKp46+ ILC3s are also present and we directly profiled their cytokine production and function in the context of bladder infection. Importantly, we also identify ILC3s in human bladders in homeostasis, emphasizing the translational clinical relevance of our data.
Our functional studies using the Rag2 À/À ILC depletion model, together with the scRNAseq analysis following UPEC challenge, identified IL17 production as the major contribution of ILC3s in acute bladder infection, with a more limited contribution to IL22 production. This presents a marked contrast to the function of ILC3 in the gastrointestinal tract, where IL17 production is not a dominant feature (Lee et al., 2015) and has not been robustly described. However, in the lung, IL17+ ILC3s have been noted in the early stages of injury models (Muir et al., 2016), as we find here in the bladder. In addition to IL17, our analyses also implicated bladder ILC3s as a source of IFNg and LTb during acute bacterial cystitis. The importance of ILC-derived IFNg in shaping bladder defense was further supported by our bulk RNA-seq analysis of ILC-depleted Rag2 À/À bladders, which showed a reduction in Interferon gamma response pathway genes. IFNg is a (H) Efficiency of A647-labeled UPEC phagocytosis by murine bone marrow-derived macrophages with and without prior stimulation with Il17a for 24 h. Flow cytometry gating strategy for macrophages-Live/CD45 + /CD64 + /F4/80 + . Each circle represents a technical replicate (n = 4-6). The 4 C negative control is denoted in blue. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by two-way ANOVA with Sídá k's multiple comparisons test (C, H) and linear regression analysis (F). All bladders used were from female mice unless otherwise stated. iScience Article canonical Th1 cytokine and, as such, is classically associated with ILC1s. However, intestinal ILC3s exhibit functional plasticity, with signals such as IL-12 and IL-18 promoting up-regulation of T-bet and production of IFNg by NCR + ILC3 (Klose et al., 2013;Vonarbourg et al., 2010). We found Ifng+ Ncr1+ ILC3s in our scRNAseq data, and in bulk bladder transcriptomic data Il18 transcripts were up-regulated post UPEC challenge, providing a signal that may drive bladder ILC3s toward an IFNg-producing phenotype.
LT signaling is important for the generation of secondary lymphoid organs, including those in the gastrointestinal tract (Alimzhanov et al., 1997;De Togni et al., 1994;van de Pavert and Mebius, 2010), and may also drive tertiary lymphoid structure formation in inflamed tissue (GeurtsvanKessel et al., 2009). ILC3 production of LT is well recognized in NCR neg subsets that include lymphoid-tissue inducer (LTi)-like cells (Klose and Artis, 2016). Notably, ILC3 production of LT has been described in the intestine during Citrobacter rodentium infection, and LTbR signaling in intestinal epithelial cells was required for recruitment of neutrophils to the infection site via production of CXCL1 and CXCL2 (Wang et al., 2010). In addition, LTa1b2 and LTbR signaling has been shown to be essential for IL22 production, which subsequently promotes AMP production in epithelial cells, with Rorc-expressing cells a major source of LT (Ota et al., 2011;Tumanov et al., 2011). The role of LT signaling in bladder infection has not been previously investigated, although increased LTbR expression has been described in chronic cystitis and in bladder cancer in humans (Shen et al., 2015). Our data implicate ILC3s as an important source of LT during acute infection, with the potential to augment IL22 production and epithelial immune function.
While IL17 has been shown to be important in bladder defense in mouse models, and higher IL17 levels have been identified in infected bladders of female compared with male mice (Zychlinsky Scharff et al., 2019), much less is known about other contributions from type 17 immune cells beyond IL17 production, for example, via IL22 secretion. We identified Th17 cells as the major source of IL22 in the bladder, even in the acute phase of infection 24-48 h after UPEC challenge. This rapid response has the hallmarks of a recall response in memory T cells, and indeed Cd44 expression was evident in the Th17 cluster in the scRNAseq dataset. IL22R-deficient mice did not show any difference in bacterial numbers in the bladder (although we only examined very early time points, and this may be evident later in the course of infection), but we observed clear effects on AMP production and epithelial cell proliferation. In particular, Reg3g, Lcn2, and Ptx3 were significantly increased in UTI bladders, and this effect abrogated in IL22R-deficient animals. Previous studies have identified Reg3g as the most up-regulated AMP in the murine bladder post UPEC challenge and increased concentration of its human orthologue, hepatocarcinoma-intestinepancreas/Pancreatitis-associated protein (HIP/PAP) in human urine during UTIs (Spencer et al., 2015). However, Reg3g-deficient mice showed no impairment in bladder UPEC defense in terms of increased bacterial CFUs (Spencer et al., 2015). In contrast, Lcn2-deficient mice had increased bacterial counts in the bladder post UPEC challenge (Steigedal et al., 2014). Similarly, Ptx3-deficient mice also showed increased susceptibility to bacterial cystitis, with Ptx3 capable of opsonizing bacteria to enhance macrophage phagocytosis (Jaillon et al., 2014). This study suggested that Ptx3 secretion by urothelial cells was mediated via cell iScience Article intrinsic TLR4-MyD88 signaling (Jaillon et al., 2014). However, our data strongly implicate IL22 in the generation of this important AMP. Notably, these studies showing the effects of Lcn2 and Ptx3 deficiency on bladder CFUs examined time points later (day 5) than those used in our study, which may explain why we did not observe a statistically significant difference in bacterial numbers in our experiments with IL22R-deficient mice.
Several publications implicate macrophages in bladder defense, with macrophage depletion associated with an increased bacterial burden in primary infection (Carey et al., 2016;Schiwon et al., 2014) but a reduction in the subsequent development of adaptive immune responses (Mora-Bau et al., 2015). Macrophages have previously been shown to orchestrate neutrophil infiltration in bladder defense (Schiwon et al., 2014); here we find additional functions in stimulating type 17 immune cells via IL1b production. In addition, F4/ 80-high macrophages were a key source of chemokines that attract CXCR6-expressing ILC3s (CXCL16), as well as CCR5-expressing gd and Th17 cells (CCL7). Of note, CXCL16 plays a key role in recruiting CXCR6+ NKp46+ intestinal ILC3 in C. rodentium infection (Satoh-Takayama et al., 2014), identifying an additional facet of cross-tissue similarity in ILC3 biology between gut and bladder. We also found evidence for reciprocal cross talk between ILC3s and macrophage, with IL17 increasing macrophage phagocytosis and impaired maturation of infiltrating monocytes to a tissue macrophage phenotype also evident following ILC depletion, reminiscent of effects observed in gut macrophages in this context (Castro-Dopico et al., 2020).
In summary, our study emphasizes the importance, and provides an overview, of the role of type 17 immunity in bladder defense against bacterial infection, uncovering differing contributions from ILC3, gdT cells, and Th17 cells that inform our understanding of this important clinical condition.

Limitations of the study
Although our study showed that IL22-dependent AMP transcripts were reduced in Il22ra knockout mice compared with WT, CFUs were comparable at the early time points examined (24 and 48 h). However, given previous studies demonstrating the functional importance of AMPs in bladder defense (Jaillon et al., 2014;Steigedal et al., 2014) (as discussed above), we cannot definitively conclude that IL22 plays no important role in bladder defense against bacterial infection. It would be important to examine the impact of Il22ra deficiency on bladder bacterial load at later time points following infection, which we were unable to complete owing to practical limitations relating to access to relevant mouse strains. Therefore, further studies will be required to evaluate the temporal effects of IL22 in UTI defense. CFU measurements were performed on single-cell suspensions rather than on cell-disrupted samples, offering the advantage of yielding paired CFU and flow cytometry data from a single experiment. The same protocol was employed across all experiments within this paper enabling a robust comparison between wild type and genetically modified mice.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:
d Microscopy data reported in this paper will be shared by the lead contact upon request.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.  Figure S2C.

Microbial strains
Uropathogenic Escherichia coli UTI89 was a kind gift from Scott Hultgren, Washington University, USA and was recovered from a patient with acute cystitis. Prior to bladder inoculation E.coli UTI89 was grown statically in Luria-Bertnai (LB) broth medium for 18 h at 37 C to ensure type 1-pilius expression as previously described. Bacterial culture was adjusted to an OD 600 nm of 0.4-0.5 for the mouse infection.

In vivo urinary tract infection model
Mice were anaesthetised, and residual urine expelled with gentle external pressure. Under aseptic conditions, a polyethelene sheath was inserted into the bladder and 100mL of UPEC or sterile PBS instilled using a 1 mg disposable insulin syringe. Mice were euthanised 24 or 48 hours post infection (as documented) and bladders digested and homogenised to a single cell suspension. Cell homogenates were plated on LB agar and incubated at 37 C overnight for quantification of colony forming units.

Murine tissue homogenisation
Prior to euthanasia mice were injected with 1:50 dilution of anti-CD45-A488 antibody (clone: 30-F11, Biolegend) in 200mL sterile PBS to label circulating leukocytes. Mice were left for three minutes before cervical dislocation. Following terminal procedure blood was obtained via cardiac puncture and transferred into an EDTA tube. Bladders were harvested from experimental mice and sliced into approximately 10 mm 3 pieces and digested for 30 min at 37 C with agitation, in a digestion solution containing 25 mg/mL Liberase TM (Roche) and 50 mg/mL DNase (Sigma) in 5 mL RPMI (Gibco) in gentleMACS C tubes (Miltenyi Biotec). Samples were then processed using a gentleMACS Dissociator (Miltenyi Biotec) on program spleen 4. The suspension was passed through a 100 mm cell strainer, washed with PBS and blocked with 50:50 mix of normal mouse and rat serum prior to staining. Cell counts per organ were calculated with the addition of 123count ll OPEN ACCESS iScience 25, 104660, July 15, 2022 In vivo ILC3 depletion Rag À/À mice were given either rat IgG2b isotype control or anti-Thy-1.2 at 0.25mg per mouse (1.25 mg/mL in sterile PBS) via intraperitoneal injection one day prior to and on the day of catheterisation.

OPEN ACCESS
using RMA and limma. Probes were reduced to one probe per gene by selecting the probe with the greatest variance across the samples using the gene filter package. Differential expression was varied out using limma with an appropriate design matrix. Public datasets used in this study are as follows: GEO:GSE68220 (Bladders from wild-type mice treated with PBS or UPEC for 24 h); GEO:GSE95404 (BMDMs treated with GM-CSF for 24 h) and GEO:GSE20087 (BMDMs treated with IL17a for 12 and 24 h).
Single cell RNA sequencing N = 5 bladders from C57BL/6J mice catheterised with UTI89 for 24 h and N = 5 bladders from naive mice were pooled before processing to a single cell suspension as described earlier (see 'Murine tissue homogenisation and intravascular labelling of circulating leukocytes'). Cell suspensions were counted using a hemocytometer and adjusted to 1 3 10 6 cells/mL. Two lanes per condition were loaded according to the standard protocol of Chromium single 3' (V2 chemistry) to capture 20,000 cells/channel. Libraries were prepared according to the manufacturer's protocol, followed by Bioanalyzer quality checks. Sequencing was performed on an Illumina Hiseq 4000.
Data was processed using the Cell Ranger 3.0.0 pipeline (103 Genomics). The FASTQ files were then aligned to the mouse genome reference sequence, mm10. The processed data was analysed using Seurat 3.2.0, doublets (Stuart et al., 2019) were detected with DoubletFinder 2.0.2 and removed, and multi-sample integration was performed with canonical correlation analysis. Cells with >250 and <2500 genes, > 1000 UMIs, and <10% mitochondrial genes were maintained. UMI counts, mitochondrial and ribosomal genes, and cell cycle phase scores were subtracted during data scaling. T cells or monocytes and macrophages compartments were isolated and re-integrated for further investigation in this study. GSEA was performed in the clusterProfiler package and was visualised with the GOChord function in the GOplot package. Briefly, genes were ranked according to their log2 expression levels (UTI vs PBS) in a descending order and were analysed with GSEA for Reactome database. The leading-edge genes from the significantly enriched immune pathways (p < 0.05, NES >0) were then subject to visualisation. Ligand-receptor analysis was performed with CellPhoneDB (Vento-Tormo et al., 2018) and was visualised with the plot_cpdb function in the ktplots package. The normalised counts and meta data extracted from Seurat objects were applied for the statistical analysis from CellPhoneDB in python 3.7.9. The resulting p values and means were then filtered and visualised with ktplots.

QUANTIFICATION AND STATISTICAL ANALYSIS
Statistical analysis was performed using GraphPad Prism software, R, IPA, or GSEA and have been described in the relevant methods sections and figure legends accordingly. For in vitro stimulation experiments, mean G standard error of mean (SEM) is shown. For RNAseq bioinformatics analyses, p values were calculated using the standard DE-Seq 2 method with multiple correction using BH procedure. For microarray experiments, p values were calculated using the limma package with multiple correction using BH procedure. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. Sample sizes (n) for all shown data can be found in figure legends. In vitro stimulations were performed in triplicate, unless stated, and sample sizes for in vivo experiments were determined based on initial experiments.