1. CS particles in the lung activated ILC cells and altered the phenotype of ILCs.
In the silicotic lung, we observed increased ILC cells at both percentage and cell number levels (Fig. 1a,b, Fig. S1a), implying its causal relationship to fibrogenesis. Cytokines in the immuno-microenvironment affected ILC functions, we further checked cytokine receptor expressions on the ILCs. Apparently, ILCs in the silicotic lung expressed analogic levels of CD127 (IL-7R), but decreased CD25 (IL-2Rα) and ST-2 (IL-33R) compared to those from saline-treated lung, implying decreased responsiveness to cytokines IL-2 and IL-33 (Fig. 1c). Notably, we noticed a substantially increased IL-18Rα levels in the ILCs from fibrotic lung, suggesting the ILC cells increased sensitiveness to IL-18 (Fig. 1d), while IL-17RB (receptor to IL-25) was not induced on the ILCs from silicotic lung (Fig.S1b). We also discovered increased CD69 and CD103, as well as elevated CXCR6 expressions in the ILCs (Fig. 1e), manifesting enhanced immobilizations in the fibrotic lung. The chemokine receptor CCR9, expressed by ILC precursors, was not affected, implying the ILCs were not recruited from the circulation(Fig.S1c). To gain more insights, we further checked effector molecule levels on the ILCs. Notably, ILCs from fibrotic lung expressed elevated MHC-II, indicating a lifting role in antigen-presenting control of adaptive immunity (Fig. 1f). Besides, ILCs expressed increased levels of ICOS (Inducible T-cell costimulator) (Fig. 1g), that promoting the survival and cytokine productions (18). Moreover, NRP-1 (Neuropilin-1) was elevated in the fibrotic ILCs (Fig. 1h), supporting the idea that NRP-1 enhancing ILC activation mediates pulmonary fibrogenesis (19). The decreased KLRG1 (Killer cell lectin-like receptor subfamily G member 1) levels suggested the ILCs in fibrotic lung were newly expanded and activated (Fig. 1i). Some inhibitory molecules, including PD-1, PD-L1, and CD73, but not TIGIT, LAG3, and CD39, were augmented in the ILCs from fibrotic lung (Fig. 1j, Fig. S1d), suggesting the activated ILCs could negatively regulate the immune microenvironment. Remarkably, we detected the ILCs in the silicotic lung expressed decreased IL-13 but increased IFN-γ (Fig. 1k), which was contradictory to the notion that ILC2 was the dominant subset within the lung, indicating the ILC components in the lung were affected by invaded CS particles. Collectively, these data indicated that ILCs in the silicotic lung were activated, while the constitutions of ILCs in the fibrotic lung were affected by CS particles.
2. The phenotype of ILCs in the silicotic lung was shifted from ILC2 to ILC1.
We next sought to explore whether the invaded particles affected the constitution of pulmonary ILCs. Basically, ST-2+ ILC2 cells take a dominant part of ILCs in the lung at physiological state. Strikingly, we noticed the proportion of ILC2 decreased while the percentage of NKp46+ILC1 gradually increased along with the progression of silicosis (Fig. 2a, b, Fig. S2a). To consolidate the findings, we checked fate-determine transcriptional factors T-bet and Gata-3 expressions which echoed our previous results (Fig. 2c). The ratio of ILC1 to ILC2 in silicotic lung augmented along with fibrosis progression (Fig. 2d). Furthermore, we utilized Rag1−/− mice, who lacks T and B cells but not ILCs, establishing silicosis model. Intriguingly, the shifted ILCs from group 2 to group 1 were recurrence in the Rag1−/− mice (Fig. 2e), indicating the altered ratio of ILC2 and ILC1 in silicosis progression was independent of adaptive immunity. Additionally, we utilized another common pulmonary-fibrosis model that was induced by bleomycin (Fig. 2f), proving there was still a shift from ILC2 to ILC1 (Fig. 2g). suggesting that the phenotype changes in ILCs were related to the progression of pulmonary fibrosis but not specifically in silicosis.
Next, we sought to explore the reason for the altered ILCs. Significantly, the decreased ILC2 cells were not due to cell death (Fig. S2b), or egress into the hilar lymph nodes (Fig. S2c). Furthermore, with the help of in vivo labelling, we demonstrated the alteration was not related to circulatory cell replenishment (Fig. S2d). Taking the notion that elevated IL-18Rα expression in the ILCs, and its causal relation to ILC1 function, we next sought to explore its expressions on ILC subsets. We proved the NKp46+ILC1 expressed high IL-18Rα, which increased along with the pathogenesis (Fig. S2e). Interestingly, we discovered the emergence of IL-18Rα+ST-2+ cells in the silicotic lung, but not in physiological condition, and expanded with silicosis progression (Fig. 2h). The increased IL-18Rα+ ILC cells were not attributed to increased circulating ILC cells (Fig. S2f). Strikingly, the loss of ST2 on ILCs was strongly correlated with a parallel increase in the frequency of the IL-18Rα+ ILCs (Fig. 2i), as well as NKp46+ ILC1 (Fig. S2g). The characterization of the IL-18Rα+ST-2+ ILC cells manifested high expressions of tissue-resident markers CD69, CXCR6, and CD103, but few were labeled with circulating CD45 antibodies, suggesting that the IL-18Rα+ILC2 were from tissue local ILC2 pools (Fig. 2j). Additionally, the result from BrdU labeling ruled out the possibility of ILC1 proliferation in fibrotic lung (Fig S2h). Collectively, these data demonstrated there were major functional and phenotypic changes in lung-resident ILCs with silicosis progression, characterized by a shift from ILC2 to ILC1.
3. IL-18 mediated conversion of pulmonary ILC2 to ILC1 in silicotic lung.
The strong negative correlation between ST2-expressing and IL-18Rα+ ILCs within silicosis progression indicated the ILC1 cells might emerge from local ILC2. Taking the notion that ILC cells exerted plasticity in some circumstances, we next did cell transfer experiments aiming to explore whether these ILC1s were converted from the local ILC2s. The congenic mice (CD45.1/2) were intranasally treated with IL-33 expanding lung ILC2 cells that were sorted by the FACS method (Fig. S3a). The sorted ILC2s were transferred into Rag1−/− mice (CD45.2/2), who were in the initial stage of fibrogenesis (Fig. 3a). The transferred ILCs, readily identifiable by CD45.1, infiltrated into the lungs of recipient mice (Fig. S3b). Though the sorted ILC2s were ST-2+ and NKp46–, they were converted into NKp46+ ST-2– phenotype and expressed high levels of IL-18Rα (Fig. 3b), implying the critical role of IL-18 in mediating ILCs conversion. We further checked the transcriptional factor expressions in the transferred cells. Significantly, the ILC2 cells downregulated ST-2 expressions and correlated striking upregulation of T-bet, the ILC1-specific transcriptional factor. Besides, the converted ILCs expressed high IL-18Rα while downregulated Gata-3 (Fig. 3c). Collectively, these results indicated that ILC2s were converted into ILC1 cells in silicosis.
To consolidate the results of ILCs conversion, we generated ST-2-eGFP reporter mice by inserting the eGFP gene into the exon of Il1rl1 (encoding ST2, a direct target of Gata-3, commonly used to label ILC2s) (Fig. 3d), in which the eGFP+ ILC cells were bonafide ILC2s. We also identified the ST2 (eGFP)+ cells were almost exclusively labeled with ST2 fluorescent antibodies that decreased with the progression of silicosis (Fig. S3c). By checking IL-18Rα expression on ILCs of the ST2-eGFP mice, we found it emerged on eGFP+ cells (Fig. 3e). Notably, we noticed a significant leaning shift from the ST-2 (eGFP)+ to IL-18Rα+ ILCs (Fig. 3e), recommending a conversion of ILC2 to ILC1. Apparently, the IL-18Rα expressions on ST-2 (eGFP)+ IL-18Rα+ ILC cells were lower than those on ST2 (eGFP)– IL-18Rα+ ILCs (Fig. 3f). Additionally, we explored the receptor for IL-12 that were related to ILC conversion but found few ILCs expressed IL-12Rβ2 (Fig. 3g), suggesting a weak relation to ILC conversion in silicotic lungs. To directly prove our hypothesis that IL-18 mediates ILCs conversion in silicosis, we introduced IL-18 knockout mice and checked the ratio of ILCs in the fibrotic lung. Expectedly, mice deficient in IL-18 exerted attenuated ILC2 to ILC1 conversion (Fig. 3h). In synthesis, these data demonstrated that the ILC1 cells in the fibrotic lung were derived from the local ILC2 pool rather than the outgrowth of existing ILC1 cells, which were mediated by IL-18, but its cellular origin still needs to be determined.
4. Mechanics-activated fibroblast secreting IL-18 modulated the conversion of ILC2 to ILC1.
We next sought to explore the cellular origin of IL-18 that mediated ILC conversions in fibrotic lungs. It was commonly believed that IL-18 was tightly linked to pyroptotic macrophages. To test the hypothesis, we applied clodronate liposomes (CL) depleting macrophages within silicosis progression (Fig. S4a, b) and checked the ratio of ILCs in the lung. Interestingly, macrophage depletion did not attenuate pulmonary ILCs conversion (Fig. 4a), implying other cellular sources mediating the effects. Various sources of IL-18 have been identified, including leukocytes and stromal cells (20). To further investigate whether leukocyte-derived IL-18 regulated ILCs conversion, we constructed bone marrow (BM) chimera using wildtype and IL-18–/– mice (Fig. S4c).
By using the chimera mice with silicosis model, we found hematopoietic cell–derived IL-18 did not boost ILC1s in the lung (Fig. S4d). Thus, IL-18 from non-hematopoietic cells (stromal cells) would be necessary for the ILCs conversion.
Taking the notion that fibroblasts actively proliferated in fibrosis progression and secreted cytokines modulating the immune microenvironment, we speculated that fibroblasts might control the ILC conversion. With immunofluorescence (IF) staining, we observed that ILC2 cells (Lin–Gata-3+) were clustered around or interacted with the activated fibroblasts (Fig. 4b), highly recommending that fibroblast modulated the ILCs conversion. Further, we prospectively detected whether activated fibroblast secreting IL-18 with an in vitro model that matrix stiffness-activated fibroblasts, which mimic fibroblast activation in the silicotic lung and avoid the effects of exogenous cytokines (Fig. S4e). Notably, the stiff matrix-cultured fibroblasts exerted a sharp increase of pro-IL-18 in the cells (Fig. 4c). Moreover, mature form IL-18 was also increased in the culture medium (Fig. 4d).
To directly certify that the activated fibroblast modulates the conversion of ILC2 to ILC1, we did some in vitro experiments. First, we cultured sorted pulmonary ILC2 with conditional medium (CM) from different stiffness-treated fibroblasts (Fig. 4e). A high portion of NK1.1+ ILCs was detected in the ST-2+ILC2s cultured with CM from high stiff-activated fibroblasts, while the ratio of ST-2+NK1.1– was decreased. Accordingly, the number of ST-2+NK1.1+ ILCs increased (Fig. f, g), implying a potential for the CM mediating ILCs conversion. To consolidate these findings, we further introduced a trans-well cocultured system of fibroblasts with ILC2 (Fig. 4h). After 36h co-culture, we observed a lift ratio of NKp46+ ILC1 and a decreased ST-2+ ILC2 (Fig. 4i, j), which was analog to the CM-cultured ILC2s. Strikingly, when did the co-cultured experiments for 60h, we discovered a boosted ratio of NKp46+ILC1s (Fig. 4k, l), indicating that activated fibroblasts modulated ILCs conversion in a time-dependent manner. Collectively, we demonstrated that IL-18 derived from the activated fibroblast would modulate the conversion of ILC2 to ILC1.
5. Mechanics-activated fibroblast secreting IL-18 was dependent on the Notch3 signaling pathway.
To gain insights into the mechanism that fibroblast secreting IL-18, we performed RNA-Seq to the primary lung fibroblasts cultured on matrix gels, identified 1281 upregulated and 936 downregulated transcriptions in stiff matrix cultured fibroblast (Fig. 5a). By Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis to upregulated genes, we determined multiple signaling pathways implicated, including HIF-1 and PI3K-Akt and rheumatoid arthritis-related signaling, cytokine-cytokine receptor interaction, etc. (Fig. b). Moreover, Gene ontology (GO) analysis showed significant enrichment in inflammatory response and wound healing, implying fibroblast’s critical role in immuno-regulation (Fig. c). Further, by protein-protein interaction analysis, we predicted that Notch signaling would be involved in process (Fig. 5d). Given that Notch3 in fibroblast recently emerged as a novel mediator in several disease pathogenesis, including fibrogenesis and rheumatoid arthritis (21, 22), also, the molecule reported to modulate fibroblast activation and survival (21), we then focused on Notch3 signaling in the fibroblasts. To gain insights about Notch3 in activated fibroblast, we first did immunoblot and proved a sharply increased Notch3 intracellular domain (Notch3ICD) in the activated fibroblast nuclei (Fig. 5e). Consistently, increased nuclear Notch3ICD localization were observed, further confirming that high mechanics activated Notch3 signaling in the fibroblasts (Fig. 5f). To directly explore the relationship between Notch3 and IL-18 expression in fibroblasts, we generated fibroblast-specific Notch3 knock-out mice, by crossing Notch3fl/fl mice to Col1a2-creERT mice (Fig. Sa, b), and sorted the pulmonary fibroblasts, repeated previous in vitro experiments. IF staining demonstrated Notch3 abrogating decreased pro-IL-18 expressions in stiff matrix-activated fibroblasts (Fig. 5g). Accordingly, the bioactive form IL-18 in the culture medium was not elevated in the Notch3 abrogating fibroblast though stimulated by high mechanics (Fig. 5h). In synthesis, these data proved that the Notch3 signaling was necessary for the mechanics-activated fibroblasts secreting IL-18.
6. Notch3 signaling in fibroblasts mediated the ILCs conversion that associated with attenuated fibrotic phenotype.
Now that we proved that Notch3 mediates IL-18 production in mechanics-activated fibroblasts in vitro, we next attempted to explore whether it modulated ILCs conversion in vivo. Consistent with the previous notion, we observed upregulated Notch3 expressions in the activated fibroblasts from silicotic lungs (Fig. 6a). Further, we used fibroblast-specific Notch3 knockout mice and treated with tamoxifen (TAM), which allowed us to disrupt Notch3 in fibroblasts at a specific time schedule. First, TAM was applied from Week 3 to 4 after CS treatment, when relays of acute inflammation and fibrogenesis (Fig. 6b). We prospectively analyzed the ratio of ILC1/ILC2 in the fibrotic lung at the times of Week 8 to 10 but did not get a significant altered ILC ratios (Fig. c-e). Unexpectedly, with fibrosis progression, we determined the ratio of ILC1 to ILC2 was apparently decreased in Notch3 knockout mice at Week 12 to 14 (Fig. 6c-e), demonstrating fibroblast would affect ILCs conversion dependent on Notch3. Further, taking the notion that fibroblasts proliferated actively in fibrogenesis, we tried to apply TAM from Week 8 to 9, expecting to gain more insights about its role at the fibrogenesis stage (Fig. 6f). By flow analysis to the ILCs, we noticed a restored ILC2 proportion, notwithstanding the ratio of ILC1 was not appreciably impacted by the Notch3 ablation at the time phase (Fig. 6g-i). Collectively, these results indicated that Notch3 signaling in fibroblast was essential for ILCs conversion in silicosis, but in a time-dependent manner.
We further aimed to explore whether the reduced ILCs conversion was linked to attenuated fibrotic phenotype in Notch3 cKO mice. Histological analysis with H&E and Masson staining demonstrated the Notch3fl/fl Colla2-Cre mice with reduced ILC1/ILC2 ratio exerted relatively attenuated cellular inflammation and collagen deposition (Fig. 6j, k). demonstrating Notch 3 signaling in fibroblast controlled ILCs conversion and fibrogenesis. The results were consolidated by qPCR analysis of the fibrotic-related gene expressions in the lung (Fig. 6l). Conversely, in the mice with unaffected ILC ratios, we did not observe apparently alleviated fibrogenesis (Fig. S6a). Furthermore, by monitoring the tidal volume (TV) and breathe frequency (f) of the mice, we extended our conclusion that the attenuated fibrosis phenotype with fast recovered pulmonary functions (Fig. S6b). Taken together, these data indicated fibroblast controlled ILC2 to ILC1 conversion in the fibrotic lung which was dependent on the Notch3 signal. The lack of Notch3 in fibroblast confers protection to the lung not only by a more preserved morphology with decreased collagen deposition but also by a fast recovery of lung function.