Renal IL-23–Dependent Type 3 Innate Lymphoid Cells Link Crystal-induced Intrarenal Inflammasome Activation with Kidney Fibrosis

Crystal-induced inflammasome activation drives inflammation and fibrosis in various diseases and organs, including arteriosclerosis, gout, Alzheimer’s disease, liver fibrosis, and silica- or asbestos-induced lung fibrosis (1–5). The urine concentration and acidification function of the kidney favors intratubular crystal formation, when the solubility coefficient of metabolites is exceeded (6, 7). Crystals may cause kidney disease by obstructing renal tubules, but recent work uncovered that they may also promote inflammasome activation in renal mononuclear phagocytes (MNPs) (7–11). These include dendritic cells and macrophages, and play important roles in kidney homeostasis, defense against infections, and in sterile nephritis (12–15). They were also shown to phagocytose crystals, which evokes the activation of the NACHT, LRR, and PYD domain-containing protein 3 (NLRP3)-dependent inflammasome that produces IL-1β and IL-18. The important role of these cytokines in crystal-induced nephritis has been demonstrated in knockout (KO) mice lacking them, their receptors, or NLRP3 (8, 16), and in experiments where pharmacological inhibitors of these mediators were administered (17). Also, in models of primary hyperoxaluria and 2,8 dihydroxy-adenine urolithiasis, these maneuvers lowered local and systemic IL-1β and IL-18 levels and reduced kidney fibrosis (18).

In recent years, innate lymphoid cells (ILCs) have gained much attention in inflammatory diseases (19–22). ILCs are a family of leukocytes with lymphoid morphology that lack Ag-specific receptors and arise in the bone marrow from the common lymphoid progenitor (23, 24). The current classification includes NK cells and three subsets of ILCs, termed type 1 ILC (ILC1), type 2 ILC (ILC2), and type 3 ILC (ILC3) (25–27). ILC1s express the transcription factor T-bet, are activated by IL-12, and produce IFN-γ upon activation (28, 29). ILC2s express the transcription factor GATA-3, are induced by IL-4, IL-25, and IL-33, and release IL-4, IL-5, and IL-13 (30). ILC3s express the transcription factor RAR-related orphan receptor γt (Rorγt), are maintained by IL-23 (31–33), and produce IL-17, IL-22 and GM-CSF (34–36). Other cells capable of producing type 3 cytokines include Th17 cells, γδ T cells, and mucosal-associated invariant T cells (37, 38). These cytokines promoted glomerulonephritis (39), but their role in inflammasome-mediated nephritis is not yet fully understood (40, 41).

ILCs have been detected in several organs, especially in barrier organs, such as the lung, the intestine, and the skin, but also in the kidney (42–44). As tissue-resident cells, they are thought to act as sentinels against bacterial, viral, and parasitic infections (45, 46). However, they may also contribute to autoimmune or chronic inflammatory diseases (47), for example, in the liver, lung, and intestine (48). ILCs are activated by environmental stimuli and then produce cytokines locally (49). Stimuli include alarmins released in the context of tissue damage or from resident immune cells, such as macrophages.

Recently, different ILC populations were identified within the kidney, and their role in nephritis was investigated (50, 51). A rare population of ILC2s has been described that promoted IL-33–dependent repair in Adriamycin-induced murine glomerulosclerosis (43, 52, 53). ILC3s have very recently been described to aggravate adaptive immunity in a model of lupus nephritis (54) and to aggravate fibrosis resulting from unilateral ureteral obstruction (55), but a role in inflammasome-dependent kidney inflammation has not been described yet. In this study, we investigated the immune mechanism underlying crystal-induced, inflammasome-mediated kidney fibrosis (56). We detected increasing numbers of intrarenal ILC3s in response to IL-1β and demonstrate that they mechanistically link inflammasome activation with kidney fibrosis.

C57BL/6J, Rag2 KO, Rorc (RAR-related orphan receptor C gene) KO (57), IL-18R KO (58), IL-1R KO, IL-17A KO (59), IL-22 KO (60), and IL-23p19 KO (61) mice between 8 and 12 wk of age were bred in the animal facility of the University Clinic of Bonn. To induce crystal nephritis, we fed female or male littermate mice 14 or 21 d with 0.2% adenine-enriched diet, as described previously (18). Thy1.2⁺ cells were depleted with anti-murine Thy1.2 Ab (BioXCell). IL-1β was neutralized with anti-murine IL-1β Ab. IL-1β was neutralized by injecting 45 µg/mouse of anti-murine IL-1β Ab (BioXCell) i.p. on days 1, 4, 7, and 10 of adenine-enriched diet. All animal experiments were approved by German state government authorities (LANUV, Düsseldorf, Germany).

Kidneys of mice were excised, cut into pieces, and digested in RPMI medium enriched with 2% FCS, 1% penicillin-streptomycin (Pen/Strep), DNase (0.1 mg/ml), and Collagenase (1 mg/ml) as described previously (62). The cell suspension was filtered, washed, and incubated for 4 h at 37°C in RPMI medium enriched with 10% FCS, 1% Pen/Strep, 1% 2-ME, 20 ng/ml rmIL-7 (PeproTech), 1 µg/ml Ionomycin (Sigma-Aldrich), 0.5 μg/ml PMA (Sigma-Aldrich), and brefeldin A (GolgiPlug) (eBioscience) 1:1000 diluted. After restimulation, cells were washed and stained.

Kidneys were digested and homogenized, RBCs were lysed, and lineage⁻ cells were enriched as described earlier (Miltenyi). Thy1.1⁺ ILC precursor cells were isolated using a kit (Miltenyi), and 5 × 10⁴ were incubated for 10 d in a 96-well plate at 37°C and 5% CO₂. Cells were resuspended in 200 µl of X-Vivo medium supplemented with IL-2 (25 ng/ml; PeproTech), IL-7 (25 ng/ml; PeproTech), SCF (25 ng/ml; PeproTech), IL-23 (10 ng/ml; Miltenyi), IL-1β (10 ng/ml; PeproTech), IL-18 (10 ng/ml; R&D Systems), and TGF-b (2 ng/ml; PeproTech). Medium was changed every other day. ILC3s were enriched for MNP-coculture experiments by MACS-based depletion of ILC1s and ILC2s using NK1.1-bio (BD) or ST2-bio (BD) and streptavidin beads, respectively. ILC3 differentiation was verified by FACS.

Primary skin fibroblasts were purified from ears of C57BL/6 mice, in passage 3 seeded into 96-well plates and cocultured with purified CD11c⁺ renal MNPs isolated using a kit (Miltenyi). In brief, ears were digested in FCS-free RPMI medium containing DNAse I (0.1 mg/ml; Roche) and Collagenase P (0.2 mg/ml; Roche) for 2–3 h. Cells were collected, counted, and plated in a six-well plate at a concentration of 5 × 10⁵ cells/2 ml in RPMI media containing 10% FCS + 1% Pen/Strep per well. On the next day, unattached cells were removed and wells were washed with PBS. Medium was exchanged on day 2, and cells were split and passaged when confluent. Fibroblasts were then seeded into 96-well plates and used in the fibroblast coculture assays with renal MNPs.

Surface molecules were stained with CD45-PerCpCy5.5 or CD45-BV421 (30F11), NKp46-BV510 (29A1.4), Thy1.2-PE, Thy1.2-BV510, or Thy1.2-PECy7 (53-2.1) from BioLegend. IL-18R-PE (P3TUNYA) was from eBioscience, and IL-1R-Alexa 647 (JAMA147) from Bio-Rad. For the lineage mixture, these biotinylated Abs from eBioscience were combined: Ly6G (RB6-8C5) and CD5 (53-7.3). The following Abs were from Invitrogen: CD11c (N418), γδ TCR (UC7-13D5), CD19 (MB19-1), CD11b (M1/70), and CD3ε (145-2C11). TCRβ (H57-597) was from BioLegend. Renal lymphocytes were gated as CD45⁺Thy1.2⁺ cells and identified as lineage⁻ using the earlier-described lineage mixture to define ILCs except in Supplemental Fig. 6, where we stained separately for γδ TCR to identify (1) Thy1.2⁺lineage⁺γδ TCR⁻αβTCR⁺ lymphocytes, (2) Thy1.2⁺lineage⁺γδ TCR⁺ lymphocytes, and (3) Thy1.2⁺lineage⁻γδ TCR⁻ innate ILCs. The total cell number was calculated by using absolute counting beads (ThermoFisher) with the formula: ([no. of events acquired × no. of beads added in the sample]/no. of beads acquired) × fraction of kidney volume used for staining. ILC1s were identified as CD45⁺Thy1.2⁺lineage⁻αβTCR⁻γδ TCR⁻T-bet⁺ cells, ILC2s as CD45⁺Thy1.2⁺lineage⁻αβTCR⁻γδ TCR⁻GATA3⁺, and ILC3s as CD45⁺Thy1.2⁺lineage⁻αβTCR⁻γδ TCR⁻Rorγt⁺. After staining with the mixture, Abs were labeled with fluorochrome-labeled streptavidin conjugates (BioLegend). For intracellular staining, surface molecules were stained first, cells were fixed and permeabilized with a kit (eBioscience), then stained with GATA3-Alexa 647 (L50-823; BD Biosciences), IFN-γ-FITC, or IFN-γ-allophycocyanin (XMG1.2), IL-17A-BV421 (TC11-18H10.1), T-bet-BV421 (4B10), Rorγ-Alexa 647 (Q31-378), or Ki67-PE/Cyanine7 (11F6) from BioLegend or Rorγ-Pacific Blue (Q31-378) or Eomes-Alexa 488 (X4-83) from BD Biosciences. Dead cells were excluded by using a fixable viability dye (Invitrogen). Data were acquired with a BD Canto II and analyzed with FlowJo (Tree Star).

The IFN-γ, IL-17A, and TNF-α production of sorted living, CD45⁺Lin⁻, Thy1.2⁺ ILCs were determined by Luminex assay according to the manufacturer’s instructions (Affymetrix). In short, sorted ILCs were plated in 40 µl/well into a 96-well plate and restimulated by PMA (500 ng/ml) and Ionomycin (1 µg/ml) overnight. Culture supernatants were measured by Bio-Rad BioPlex analyzer (Bio-Rad).

Lineage⁻, Thy1.2⁺, Rorγt⁺, or Rorγt⁻ ILCs were sorted using Rorc reporter mice using a BD Aria and plated into a 96-well plate and were restimulated by PMA (500 ng/ml) and Ionomycin (1 µg/ml) overnight. Culture supernatants were collected and measured by LEGENDplex Mouse Inflammation panel according to the manufacturer’s instructions and analyzed by software from BioLegend.

Tissues were fixed in methyl Carnoy’s solution, paraffin embedded, and cut into 1-µm sections. Periodic acid–Schiff staining was used to assess renal morphology. Collagen type III (Southern Biotechnology Associates) and α-smooth muscle actin (α-SMA; Dako) staining were performed as described previously (16, 63). For negative controls, isotype-matched IgG was used, and no unspecific staining was observed. Whole slides were digitalized using a whole-slide scanner NanoZoomer 2.0-HT and analyzed using NDP.view (Hamamatsu Photonics) and ImageJ (NIH) software. The percentage of stained area was analyzed in 16–22 consecutive fields omitting larger vessels at ×20 magnification, representing almost the entire kidney area. All analyses were performed in a blinded manner.

Multiepitope-ligand cartography was performed as described previously (64). In brief, 4% PFA-fixed kidney sections were cut in 5-µm sections with a MH560 cryotome (ThermoFisher) on 3-aminopropyltriethoxysilane–coated cover slides (24 × 60 mm; Menzel-Gläser). Samples were rehydrated, blocked, and permeabilized with 0.2% Triton X-100 in PBS for 10 min at room temperature, and unspecific binding was blocked with 10% goat serum and 1% BSA in PBS for at least 20 min. A fluid chamber was created with “press-to-seal” silicone sheets (1.0 mm thickness; Life Technologies) with a circular cutout (10-mm diameter), which was attached to the coverslip surrounding the sample holding 100 μl of PBS. The tissue section was then incubated with PE, allophycocyanin, or FITC-coupled Abs (anti-B220, anti-CD5, anti-Gr1, anti-FcER1a, anti-CD45, anti-CD90.2, anti-CD127, anti-Rorγt, anti-CD11b, anti-F4/80, anti-CD3, anti-CD31, and DAPI). A fluorescence image was taken with three fluorescence channels acquired in parallel in one cycle, followed by repetitive washing steps. Afterward, a phase-contrast image was taken to adjust the autofocus and to ensure a stable image focus between imaging cycles. After image acquisition, the fluorescence was bleached by light exposure and a bleaching control image was taken. All steps were performed in an automated system, using a pipetting robot and a Leica microscope (Toponome Imaging Cycler). To analyze the multiplexed imaging data and computationally identify ILCs, T cells, and MNPs, we used an automated analysis pipeline in CellProfiler that segments single cells within images and measures mean fluorescence intensities (MFIs) of all markers included in the experiment for each cell (65). With this single-cell information, cell types can be identified based on thresholding of MFI for relevant markers. ILCs were defined as cells with MFI RORγt > 0.2, MFI CD3 < 0.04, MFI B220 < 0.05, MFI Gr1 < 0.02, and FcεR1a < 0.05. T cells were defined as cells with MFI CD45 > 0.1 and MFI CD3 > 0.1. We defined MNPs as cells with MFI CD11b > 0.07 and MFI F4/80 > 0.12. The area within a 10-µm radius (corresponding to the average diameter of a hematopoietic cell) around each ILC was defined as the ILC neighborhood or niche, and the distribution of MNP and T cells in this area was quantified. The same neighborhood analysis was performed for T cells with respect to MNP.

Results are expressed as mean ± SEM if not otherwise stated. The statistical analysis to compare three or more groups was performed with one-way ANOVA with Tukey posttest; Mann–Whitney was performed to compare two groups. Data were analyzed with GraphPad Prism version 10.

To study how inflammasome-mediated inflammation causes kidney fibrosis, we fed mice an adenine-enriched diet for 14 or 21 d (18, 66), which induces the formation of intratubular crystals. These activate the inflammasome in renal MNP, which causes progressive inflammation and eventually kidney fibrosis (8, 16, 17). When we investigated the immune cell infiltrate in the inflamed kidney by flow cytometry on day 14 of the diet (gating scheme in Fig. 1A), we observed an increase in CD45⁺Thy1.2⁺lineage⁻ γδ TCR⁻ immune cells (Fig. 1B), indicating an accumulation of ILCs. Staining for transcription factors revealed an increase of Tbet⁺ ILC1s and Rorγt⁺ ILC3s, but not of GATA3⁺ ILC2s (Fig. 1C–E), which was unexpected because usually ILC2s are associated with fibrogenic processes (43). EOMES⁺ NK cells were also unaltered (Supplemental Fig. 1). The accumulation of Rorγt⁺ ILC3 was dependent on proliferation, as evidenced by Ki67 expression (Supplemental Fig. 2). The Rorγt⁺ ILCs produced the prototypical type 3 immunity cytokine IL-17A and the Rorγt⁻ ILCs secreted the type 1 immunity signature cytokine IFN-γ, but only little of the type 2 cytokines IL-4, IL-5, and IL-13 (Fig. 1F–H, Supplemental Fig. 3). These findings identified an accumulation of ILC1 and ILC3 in inflammasome-mediated nephritis.

To investigate whether ILCs promote inflammasome-mediated kidney fibrosis, we fed the adenine-enriched diet to Rag2 KO mice that harbor ILCs, but no adaptive immune cells, and included an additional group of Rag2 KO mice treated with anti-Thy1.2 Ab to deplete also the ILCs. This strategy is frequently used to perform loss-of-function studies for ILCs (67–69), because no selective depletion method for ILCs exists. Histological analysis revealed that Collagen III and α-SMA deposition in the kidneys of adenine-treated Rag2 KO mice was reduced compared with wild-type (WT) mice (Fig. 2A, 2B), indicating a role for adaptive immune cells in inflammasome-mediated kidney fibrosis. Importantly, anti-Thy1.2 Ab–treated Rag2 KO mice (Rag2KO + anti-Thy1.2) showed an even greater reduction of Collagen III and α-SMA deposition than ILC-competent Rag2 KO (Rag2KO) and WT mice (Fig. 2A, 2B), and also a reduction of serum creatinine compared with WT mice (Fig. 2C) indicative of renal function. These findings confirmed a profibrotic role of ILCs in inflammasome-mediated nephritis.

We next asked how inflammasome activation regulates ILCs during crystal-induced nephritis, focusing on the inflammasome-derived cytokines IL-1β and IL-18 (8). Mice treated with IL-1β blocking Abs or deficient for IL-18R were fed the adenine diet. After 2 wk, we found a significant reduction of intrarenal Rorγt⁺ ILC3s in anti–IL-1β–treated WT mice compared with WT mice, in contrast with IL-18R KO mice (Fig. 3A), whereas the decrease of IL-17A–producing Rorγt⁺ ILC3 was not significant (Fig. 3B). Numbers of Rorγt⁻ ILCs, including ILC1 and ILC2, were unaltered in the absence of either cytokine (Fig. 3C), nor were IFN-γ–producing cells (Fig. 3D). Thus, IL-1β, but not IL-18, selectively expanded ILC3, although we cannot formally exclude that IL-1β acted indirectly on ILC3. Histological analysis of kidney sections of adenine-fed mice showed less α-SMA, but not Collagen III, staining of kidney sections after IL-1β neutralization (Fig. 3E, 3F). To further validate the involvement of IL-1β signaling, we used IL-1R KO mice instead of the blocking Ab, and again noted significantly reduced numbers of ILC3s, but not of Rorγt⁻ ILCs (Supplemental Fig. 4A, 4B). In this study, deposition of Collagen III was reduced (Supplemental Fig. 2C), but not of α-SMA (Supplemental Fig. 4C). IL-18R KO deficient mice did not show significantly reduced kidney fibrosis parameters (Fig. 3E, 3F). The effects of IL-1β on intrarenal ILC3 numbers and the partial effect on kidney fibrosis endorsed a profibrotic role of ILC3 in our model.

The unchanged ILC3 numbers in IL-18R KO mice (Fig. 3A) did not exclude effects of this cytokine on the phenotype of ILC3. To investigate this possibility, we examined ILC3s isolated from kidneys of WT, IL-18R KO, and as a control, IL-1R KO mice on an adenine-enriched diet. We found that a proportion of renal ILC3s expressed the NK cell marker NKp46 (Supplemental Fig. 5A), which has recently been described to be expressed by ILC3s that responded to environmental stimuli in nonlymphatic tissues (44, 70, 71). NKp46 was increased on ILC3s from WT and on IL-1R KO mice under an adenine-enriched chow, but not on ILC3s from IL-18R KO mice (Supplemental Fig. 5B, 5C). Among the Rorγt⁻ ILCs, no significant differences in NKp46 expression were seen (Supplemental Fig. 5D, 5E). These findings showed that IL-18, but not IL-1β, promoted the emergence of NKp46⁺ ILC3, but not of other ILC subtypes expressing this marker during inflammasome-mediated nephritis.

To link inflammasome-mediated nephritis with type 3 immunity, we isolated MNPs from nephritic kidneys and measured their production of typical cytokines. MNPs of nephritic mice produced more IL-6, TGF-β, and IL-23 (Fig. 4A–C), all of which have been associated with differentiation of Rorγt⁺ cells (70, 72, 73). Moreover, whole kidney homogenates of adenine-fed mice contained the type 3 cytokine IL-17A, consistent with the earlier findings (Fig. 1), whereas IL-22 and GM-CSF were unchanged (Fig. 4D–F).

Various innate and adaptive lymphocytes can secrete IL-17A (74, 75) and express Rorγt. To identify the IL-17A–producing cell type(s), we examined by flow cytometry three lymphocyte populations known to produce this cytokine: (1) αβ T cells (Thy1.2⁺Rorγt⁺lineage⁺γδ TCR⁻, including both Th17 cells and IL-17–producing CTLs), (2) γδ T cells (Thy1.2⁺Rorγt⁺lineage⁺γδ TCR⁺ population), and (3) ILC3s (Thy1.2⁺Rorγt⁺lineage⁻γδ TCR⁻ population) (Supplemental Fig. 6A). Rorγt⁺IL-17⁺ cells increased in all three subsets in the kidneys of diseased mice (Supplemental Fig. 6B–D). When we measured IL-22 on protein and mRNA levels, no elevation was detected (Supplemental Fig. 7A, 7B). Furthermore, disease was not ameliorated in IL-22 KO mice, nor were renal ILC3s altered in comparison with WT mice (Supplemental Fig. 7C, 7D). Together with the data in Rag2KO mice treated with anti-Thy1.2 (Fig. 2), this indicated that IL-17⁺ ILCs played a profibrotic role in our model.

To formally demonstrate a role of type 3 immunity, we fed Rorc KO mice that lack Rorγt the adenine-enriched diet. These animals developed less severe kidney fibrosis compared with WT mice, as evident by significantly lower α-SMA and Collagen III deposition in immunohistochemistry (Fig. 4G). Furthermore, Rorc KO mice showed significantly diminished creatinine serum (Fig. 4H). This confirmed that type 3 immune cells such as ILC3 were involved in inflammasome-mediated nephritis.

To test for a role of type 1 immunity, we fed the adenine-enriched diet to mice deficient for IFN-γ (IFN-γ ΚΟ). These mice showed similar deposition of α-SMA and Collagen III and similar serum creatinine as did WT mice (Supplemental Fig. 8A–C), confirming that type 1 immunity was dispensable in our model.

To determine which type 3 cytokines were important, we fed the adenine-enriched diet to mice deficient for IL-17A. We found reduced deposition of Collagen III, but no significant changes of α-SMA compared with WT mice (Fig. 4I, 4J), indicating partial protection from fibrosis. This may be explained by other type 3 effector cytokines that IL-17A–deficient mice still might be able to use. We therefore focused on IL-23, known to promote IL-17 production of ILC3, but not of other ILCs (32, 70, 72, 73). The kidneys of IL-23p19–deficient mice under the adenine diet contained 70% less ILC3 than WT mice (Fig. 5A), whereas Rorγt⁻ ILC1 and ILC2 numbers were unchanged (Fig. 5B). Likewise, the IL-17A–producing ILC3 subset was reduced by even 85% (Fig. 5C), and other IL-17A–producing cells, determined as IL-17A⁺lineage⁺Thy1.2⁺ cells, were not significantly reduced when IL-23 was lacking (Fig. 5D), arguing against major IL-23 effects on other type 3 immune cells, such as Th17 cells and γδ T cells in our model. Importantly, kidneys of IL-23p19–deficient mice showed significantly less staining for both α-SMA and Collagen III (Fig. 5E–G). These findings supported a pathogenic role of IL-23 in inflammasome-mediated nephritis.

To understand why IL-23 was more important for renal fibrosis than IL-1β in our model (Figs. 3, 5), we examined on renal ILC3 the expression of their receptors and of the IL-18R as a control. The receptors for IL-23 and IL-18 were expressed throughout the disease, whereas IL-1R was not detectable any longer on ILC3 after 14 d of adenine-enriched diet (Fig. 5H–M), implying that these cells had lost the ability to respond to IL-1β after activation. By contrast, they retained the capability to respond to IL-23, implying that this cytokine could affect ILC3 longer than IL-1β, possibly explaining the different effects.

We next wished to visualize ILC3 in the nephritic kidney. To this end, we fed Rorc-reporter mice with the adenine-enriched diet and analyzed kidney sections by multiepitope-ligand cartography. Automatized image analysis revealed Rorγt⁺ cells that were lineage⁻ (B220⁻, Gr1⁻, FcεR1a⁻, CD11b⁻, CD3⁻) (Fig. 6A, white circles), classifying ILC3s. Notably, a certain proportion of these Rorγt⁺ ILC3s were located in the vicinity of CD11b⁺F4/80⁺ MNPs during inflammasome-mediated nephritis. Computational image analysis showed that 8 of 10 ILC3s colocalized with MNPs in the section (80%) (Fig. 6B), whereas only 20 of 113 (17.7%) CD3⁺ T cells, more precisely, 2 of 3 Th17 cells (66.7%), colocalized with MNPs (Fig. 6C), 4 of 10 ILC3s colocalized with CD3⁺ T cells (Fig. 7D), and 3 of 15 (20%) Rorγt⁻ ILC1s and ILC2s colocalized with MNPs (Supplemental Fig. 9). These findings suggested cross-talk between ILC3s and MNPs in inflammasome-mediated nephritis.

Given that no tools exist to selectively remove ILC3 in vivo, we decided to directly demonstrate the profibrotic activity of ILC3 using a coculture system. Because ILC3s were very rare in the kidney, we instead purified renal CD11c⁺F4/80⁺ MNPs from mice under an adenine-enriched diet, which should have been influenced by ILC3, cocultured them with primary fibroblasts isolated from syngeneic mice, and measured their ability to stimulate collagen I release into the supernatant after 72 h (Fig. 7A). Indeed, renal MNPs from WT mice under an adenine-enriched diet stimulated fibroblasts to produce collagen I (Fig. 7B). Such production was seen also after adding the profibrotic cytokine TGF-β, which is produced by renal MNPs (76) (Fig. 7B), and it was reduced when a TGF-β–blocking Ab was added, but not when IL-1β was inhibited (Fig. 7B), arguing for TGF-β as a mediator of phagocyte-induced fibrosis. Importantly, neither phagocytes from RAG KO mice depleted with anti-Thy1.2 Ab from ILCs nor phagocytes from Rorc KO mice stimulated collagen production (Fig. 7C), consistent with the roles of ILCs and a type 3 immune response demonstrated in our in vivo studies (Figs. 2, 4). Furthermore, renal MNPs from mice lacking a functional IL-1β and IL-23 axis were unable to drive collagen production as much as phagocytes from WT mice (Fig. 7D), consistent with the fibrogenic role of these cytokines in vivo (Figs. 3, 5).

Finally, we aimed to directly demonstrate that ILC3s can render renal MNPs profibrogenic. To this end, we isolated ILCs precursors from mouse kidneys and differentiated them into ILC3s. Rorγt expression of the generated ILC3s was confirmed by flow cytometry (Supplemental Fig. 10). Such ILC3s were then cocultured with primary renal MNPs, followed by addition of fibroblasts as in the earlier experiments (Fig. 7E). After 72 h, ILC3-exposed renal MNPs stimulated collagen production by fibroblasts more than renal MNPs did, which were not exposed to ILC3s or to Rorγt⁻ ILCs (Fig. 7F), confirming the ability of ILC3s to promote the profibrotic properties of renal MNPs. In summary, these findings support the scenario that inflammasome-stimulated ILC3s interacted via IL-1β or IL-23 with renal MNPs, which then used TGF-β to promote renal fibrosis.

Fibrosis is a common consequence of inflammasome activation and chronic inflammation in several tissues (77), including the kidney (6, 9, 56, 78, 79). In this article, we propose a (to our knowledge) novel pathway how inflammasome activation can promote kidney fibrosis via ILC3s that respond to the inflammasome-derived cytokine, IL-1β, and to the type 3 immunity cytokine, IL-23, which emerged as an amplifier of NLRP3 inflammasome activation and resultant kidney fibrosis. These findings were unexpected because fibrosis is usually associated with type 2 immunity. Consistently, ILC2 promoted tissue repair in Adriamycin-induced murine glomerulosclerosis (43, 52). We found no upregulation of type 2 cytokines in our model, which argued against a role of ILC2 but did not formally exclude an additional role of these cells. ILC3s have very recently been implicated in lupus nephritis (54) and unilateral ureteral obstruction (55), but a role in inflammasome-dependent kidney inflammation has not been described yet.

The type 3 signature cytokine IL-17 has been shown to promote fibrosis in other contexts, like liver fibrosis (49) or unilateral ureteral obstruction (40, 55). However, it played only a partial profibrotic role in our model, which was reminiscent of the situation in multiple sclerosis, where Th17 cells are important for disease, but not because of their production of IL-17 (80), but of GM-CSF (81). However, in our model, GM-CSF was not increased. Instead, the type 3 immunity maintenance cytokine IL-23 played a major profibrotic role in our system, as did the signature inflammasome cytokine IL-1β, which expanded ILC3. This finding was consistent with previous work showing that this cytokine induced expansion of pulmonary ILC3 in a mouse model of obesity-associated airway hyperactivity (67). ILC3s express a broad range of cytokine receptors, including IL-1R, IL-18R, and Il-23R (67, 82), explaining their response to inflammasome products (83, 84). However, the IL-1R was downregulated during the course of disease, perhaps as a result of receptor endocytosis upon ligand binding (85–87), so that it could no longer stimulate ILC3s. By contrast, sustained IL-23R expression may explain the greater protection seen in mice lacking IL-23 than IL-1β. Another explanation may be an effect of IL-23, but not IL-1β, on Th17 and/or γδ T cells, although we did not detect significant alterations in their numbers. Adaptive type 3 immune cells also played a role in our model and depend on IL-23 (32, 59). Future studies are warranted to address their role in inflammasome-mediated fibrosis.

Sustained expression was also noted for the receptor of IL-18, the other main inflammasome-derived cytokine. However, genetic removal of this receptor failed to affect renal ILC3 numbers and cytokine production. In this respect, ILC3s seem to differ from NK cells and ILC1, which did proliferate in other models in response to IL-18 (88, 89). Consistent with the role of ILC3, inflammasome-mediated fibrosis did not depend on IL-18. This contrasts with the situation in acute crystal-induced kidney injury, where IL-18 did play a role (8). IL-18 affected NKp46 expression on ILCs, as it did in NK cells, but this did not noticeably impact fibrosis development in our model. We did not use fate-mapping mouse models to formally prove the conversion of NKp46⁻ ILCs in NKp46⁺ ILCs or to determine the relationship between ILC3 and NK cells (90, 91), but focused on changes of NKp46⁺ ILC3s in the presence or absence of IL-18 during inflammasome-mediated inflammation. To mechanistically unravel the role of the NKp46⁺ ILC3 subpopulation, tools need to be developed to generate mice exclusively lacking NKp46 in ILC3s.

The cross-talk between ILC3s and renal MNPs was supported by the analysis of these cells in their tissue context by multiplexed histology, an imaging technique coupled to a customized analysis pipeline that allowed us to computationally analyze and quantify single cells within tissues, while retaining their spatial information. This analysis showed colocalization between ILC3s and renal MNPs in inflamed kidneys. The molecular pathways identified during our in vivo experiments were further corroborated by in vitro studies, where we queried the ability of renal MNPs isolated from mice with adenine-induced nephritis to elicit collagen production by cocultured fibroblasts. Such production was lower when the MNPs were obtained from nephritic mice lacking ILC, Rorc, IL-1β or IL-23, providing further experimental support for the pathomechanism we identified in this study. This mechanism is consistent with the recent proposal that ILC3-derived cytokines may support tissue homeostasis by inducing immunosuppressive and/or repair functions of MNPs (47, 92) and with the recent identification of a type 3 immunity-related profibrotic macrophage subset in the liver and lung (93). An exaggerated repair function triggered by NLRP3 inflammasome activation may cause ILC3s to promote renal MNP-driven fibrosis. Our in vitro studies suggested that fibroblasts are instructed by TGF-β, a long-known critical mediator of kidney fibrosis (76). This finding differs from a recent study proposing that ILC3s use IL-17A to stimulate fibroblasts (55), which was not the case in our model. Further studies are needed to clarify how ILC3s cause MNPs to produce TGF-β.

In conclusion, ILC3, IL-23, and IL-1β are mediators in inflammasome-mediated fibrosis in the kidney and perhaps also in other organs and might represent therapeutic targets.

The authors have no financial conflicts of interest.

We acknowledge support by the Flow Cytometry Core Facility and the central animal facilities of the Medical Faculty of Bonn University. We especially thank Andreas Diefenbach (Charité) for help with in vitro expansion of ILC3. The visual abstract was created using BioRender.com.