PPARɣ drives IL-33-dependent ILC2 pro-tumoral functions

Group 2 innate lymphoid cells (ILC2s) play a critical role in protection against helminths and in diverse inflammatory diseases by responding to soluble factors such as the alarmin IL-33, that is often overexpressed in cancer. Nonetheless, regulatory factors that dictate ILC2 functions remain poorly studied. Here, we show that peroxisome proliferator-activated receptor gamma (PPARγ) is selectively expressed in ILC2s in humans and in mice, acting as a central functional regulator. Pharmacologic inhibition or genetic deletion of PPARγ in ILC2s significantly impair IL-33-induced Type-2 cytokine production and mitochondrial fitness. Further, PPARγ blockade in ILC2s disrupts their pro-tumoral effect induced by IL-33-secreting cancer cells. Lastly, genetic ablation of PPARγ in ILC2s significantly suppresses tumor growth in vivo. Our findings highlight a crucial role for PPARγ in supporting the IL-33 dependent pro-tumorigenic role of ILC2s and suggest that PPARγ can be considered as a druggable pathway in ILC2s to inhibit their effector functions. Hence, PPARγ targeting might be exploited in cancer immunotherapy and in other ILC2-driven mediated disorders, such as asthma and allergy.

Innate lymphoid cells (ILCs) are the most recently identified subset of innate lymphocytes. ILCs can be classified in three principal groups according to their transcription factor expression and cytokine secretion, mirroring to some extent CD4 T helper subsets1. Group 1 ILCs (ILC1s) express T-bet and secrete IFN-γ. Group 2 ILCs (ILC2s) express GATA3 and secrete IL-13 and IL-5. Tissue-resident group 3 ILCs (ILC3s) express RORγT and secrete IL-22 and IL-17 while, in the human peripheral circulation, ILC3s also comprise a population of progenitor cells (ILCPs) able to differentiate into all ILC subsets and natural killer (NK) cells2. In the last few years, different reports highlighted the important contribution of ILCs to different physiological and pathophysiologic processes3. In particular, ILC2s have been reported to play a critical role in several inflammatory diseases including asthma, chronic rhinosinusitis, and allergic rhinitis4,5,6. Moreover, we and others have shown that ILC2s mediate tumor immune responses, with either pro- or anti-tumor effects depending on the tumor type7,8,9,10,11. Nonetheless, external and cell-intrinsic regulatory factors that dictate ILC2 activation and function remain poorly studied, particularly in humans. Better knowledge on ILC2-drivers is warranted to enable pharmacological interference with ILC2 functions in disease settings. In that regard, alarmins (e.g., IL-33) were shown to directly activate ST2-expressing ILC2s and induce their Type-2 cytokine secretion12. Further, it was reported that certain types of lipids or lipid-derivatives act as key mediators involved in ILC2 activation. In particular, in the context of helminth infections, ILC2 function is dependent on fatty acid synthesis and oxidation, which is known to be regulated by peroxisome proliferator-activated receptors (PPARs)13,14. PPARs are a class of nuclear hormone receptors that act as transcription factors regulating gene expression15. They are divided in three major isoforms: α, β/δ, and γ that differ in their tissue expression and functions15. PPARγ regulates the transcription of genes associated with lipid metabolism and is expressed in different immune cells, including lymphocytes, monocytes, dendritic cells, and platelets, where it mainly exerts anti-inflammatory effects16. Although different synthetic PPARγ ligands are used in the clinics, 15-deoxy-Δ12-14-PGJ2 (15d-PGJ2) is one of the few mediators that is often referred as an endogenous PPARγ ligand17. 15d-PGJ2 is a prostaglandin D2 (PGD2)-derived product that has been reported to be essential for ILC2 activation18,19,20. In particular, it has recently been shown that PGD2 and its metabolites (including 15d-PGJ2) are endogenously synthetized after ILC2 stimulation with IL-33, IL-25, and TSLP19. Furthermore, elevated levels of PPARγ and other genes involved in prostaglandin (PG) synthesis and response were recently reported in a single cell mRNA-sequencing analysis of ILC2s sorted from human tonsil tissue21. In this work, we characterize the expression and functional role of PPARγ in human and mouse ILCs to assess whether PPARγ can be targeted in the context of an ILC-directed immunotherapy.

Ex vivo and in vitro expanded human ILC2s specifically express PPARγ

In order to characterize ILC2s in terms of function and activation, we mined a previously performed RNA-sequencing (RNA-seq) analysis of ILC1s, ILC2s, and ILCPs sorted from healthy donor (HD) PBMCs, and compared their transcriptional signature (ArrayExpress E-MTAB-8494). In particular, focusing on fatty acid metabolism-related genes, that have been associated to ILC2 response during helminthic infection, we found that PPARγ (PPARG) was more expressed in ILC2s, as compared to the other ILC subsets (Fig. 1a, left panel), while no differences in PPARα (PPARA) and PPARβ (PPARD) expression were observed among the subsets (Fig. 1a, right panel), as also confirmed by qPCR (Fig. 1b). Similar observations were made by the analysis of PPAR expression in in vitro expanded ILC subsets, both at mRNA and protein level (Fig. 1c, d). Given that ILCs are considered as the innate mirrors of CD4 T helper subsets, we analyzed the expression of PPARs in the adaptive counterparts of ILC1s, ILC2s, and ILCPs, i.e., Th1, Th2, and Th17 cells, respectively. PPARγ expression was preferentially observed in human Th2 and Th17 but not Th1 cells upon expansion (Supplementary Fig. 1a, d), in contrast to the mouse where PPARγ protein is expressed predominantly in Th2 cells22. Taken together, these findings show specific expression of PPARγ in human ILC2s, which is also maintained during in vitro culture.

PPARγ sustains Type-2 cytokine secretion in human ILC2s

To better elucidate the functional role of PPARγ in ILC2s, we analyzed the cytokine production of IL-33/IL-25 (referred as IL) stimulated ILC2s, after pre-incubation with non-toxic concentrations of T0070907, a selective and irreversible PPARγ antagonist23. As shown in Fig. 2a, b, ILC2s treated with T0070907 produced significantly less IL-13 as compared to control ILC2s. Moreover, a trend for reduction was also observed for other ILC2-specific cytokines, including IL-4 and amphiregulin (AREG) (Fig. 2b)24. These results were confirmed by the quantification of cytokines in cell culture supernatants of ILC2s, pre-treated or not with T0070907 (Fig. 2c). This additional readout allowed us to observe that T0070907 not only reduced the secretion of IL-13 in a concentration-dependent manner (Supplementary Fig. 2a), but also of IL-5 in pre-treated ILC2s (Fig. 2c). To test if the inhibitory effects of T0070907 were due to the selective inhibition of PPARγ, we transfected human in vitro expanded ILC2s with a small interfering RNA (siRNA) targeting PPARγ. The knockdown of PPARγ was confirmed by qPCR analysis (Supplementary Fig. 2b). Consistent with the impact of pharmacological inhibition, PPARγ silencing led to significantly reduced cytokine production and secretion (Supplementary Fig. 2c–e). This effect on Type-2 cytokine secretion might be at least in part a direct consequence of PPARγ binding to peroxisome proliferator response elements (PPREs), since we found motives partially matching the human PPARγ-retinoid X receptor alpha (RXRα) heterodimer binding motif in the promoter region of both IL-5 and IL-13 genes (Fig. 2d, Supplementary Data 1). To substantiate this finding, we performed Chromatin immunoprecipitation (ChIP) on human expanded ILC2s stimulated with the cytokine cocktail and treated or not with T0070907. The results indicated that PPARγ directly bound to the promoter of IL-13 in cytokine-activated ILC2s (Supplementary Fig. 2f). An increasing number of reports demonstrated a direct correlation between PG synthesis and PPARγ activation25,26,27. In particular, it has been recently reported that the inhibition of the cyclooxygenase (COX), which is the key rate-limiting enzyme in PG production, prevents IL-5 and IL-13 secretion by cytokine-stimulated ILC2s19. Based on these data, we speculated that COX and PPARγ could exert a synergistic effect on ILC2 cytokine secretion. Hence, we treated ILC2s with a combination of T0070907 and celecoxib (CXB), a selective inhibitor of COX-2, a gene upregulated in ILC2s after cytokine stimulation19. Although not statistically significant, we observed that this treatment induced a partial reduction in IL-13 and IL-5 secretion (Supplementary Fig. 2g).

PPARγ inhibition results in mitochondrial dysfunction in human ILC2s

As a lipid sensor, PPARγ plays a key role in the regulation of metabolism by affecting, among others, mitochondrial biogenesis, glucose uptake, and lipid intake28. To further dissect the mechanisms underlying the PPARγ-mediated inhibition of ILC2s, we measured mitochondrial mass by MitoTracker Green uptake and MitoTracker Deep Red fluorescence in these cells. As shown in Fig. 3a–d, both MitoTracker Green and Deep Red dye uptake were decreased in ILC2s after treatment with the PPARγ inhibitor, suggesting reduced mitochondrial mass. We also used tetramethylrhodamine methyl ester (TMRM) that, by accumulating in intact and active mitochondria, allows to assess changes in their membrane potential. Consistent with the MitoTracker results, the TMRM assay showed a significant decrease of mitochondrial potential after incubation of ILC2s with T0070907 (Fig. 3e, f). To corroborate our data, we performed electron microscopy (EM) analysis that showed reduced percentage of mitochondrial volume per cell in T0070907-treated ILC2s as compared to control (Fig. 3g, h). These findings indicate that the inhibitory effects of T0070907 correlates with a decline in mitochondrial function, which in turn likely impairs ILC2 cellular fitness.

Tumor-derived IL-33 sensitizes ILC2s to produce IL-13

Over the last decade, alarmins have been identified as signaling mediators involved in cancer development and progression29,30. In particular, different studies reported that IL-33 and its receptor ST2 play a critical role in the pathogenesis of different types of cancers, including breast cancer, hepatocellular carcinoma, and colorectal cancer (CRC)31,32,33,34. We used CRC as a model to address a putative role of tumor-derived IL-33 on ILC2 functions in the context of tumor immunity. First, we confirmed expression of IL-33 in human CRC tissues using a Tissue Micro Array (Fig. 4a). GATA3+CD3− cells were identified in proximity to IL-33+ CRC epithelial cells, suggesting that IL-33 released from these malignant cells may stimulate ILC2s in human CRC lesions (Fig. 4b). In line with these findings, we observed elevated IL-33 levels in the serum of CRC patients as compared to HDs (Fig. 4c). Next, we evaluated ILC2 frequency and function, in terms of cytokine production, in both PBMCs and tumor-infiltrating lymphocytes (TILs) in CRC patients. As previously reported, circulating ILC2 frequency was comparable in HDs and CRC patients (Fig. 4d)35. Importantly, ILC2s are present among TILs, as assessed by flow cytometry analysis of dissociated CRC tissues (Fig. 4d). In terms of function, ILC2s from CRC patients’ PBMCs produced more IL-13 compared to ILC2s from HDs’ PBMCs (Fig. 4e, f). To gain further insight into the role of PPARγ in ILC2s in the CRC setting, we evaluated the expression of PPARγ, as well as of CPT1A, one of the well-characterized PPARγ-direct target genes36, in freshly-sorted ILC2s from both PBMCs and TILs of CRC patients. As shown in Supplementary Fig. 3a, b, a trend for higher expression of PPARγ and CPT1A was observed in CRC patients’ PBMCs compared to HDs, suggesting that PPARγ is not only increased but also more active in the CRC setting. In addition, exposure of expanded ILC2s from CRC patients to T0070907 resulted in the reduction of IL-13 and IL-5 secretion as well as in impairment of mitochondrial functions, thus recapitulating the effects observed in HD ILC2s (Supplementary Fig. 3c–e). Collectively, these results indicate that ILC2s from cancer patients are more prone to secrete cytokines, which may be related to their increased exposure to IL-33, both systemically and in the tumor microenvironment.

IL-13 drives the crosstalk between ILC2s and cancer cells

In the last few years, IL-13 and its receptors have been identified as novel targets for cancer therapy, and inhibition of IL-13-producing cells as a strategy to reach this goal37. To better understand the crosstalk between PPARγ, ILC2s, and the CRC cells, we pre-treated ILC2s with T0070907 or solvent-control and co-cultured them with the SW1116 CRC cell line. As shown in Fig. 5a, b, the ILC2-SW1116 co-culture resulted in increased IL-13 production by ILC2s that was abolished by pre-treating ILC2s with T0070907. IL-13 production in ILC2s was possibly triggered by increased expression of IL-33 in co-cultured CRC cells (Fig. 5c). To address this hypothesis, we performed co-culture experiments using the helminth parasite-secreted protein (HpARI), known to suppress type 2 immune responses through interference with the IL-33 pathway38. As shown in Supplementary Fig. 4a, the addition of HpARI in the medium significantly reduced the production of IL-13 by ILC2s, after co-culture with SW1116 colorectal cancer cells. Since IL-13 is involved in CRC progression and metastasis development by affecting epithelial to mesenchymal transition (EMT)39, we hypothesized that ILC2-derived IL-13 could affect EMT in CRC cells. To address this point, we performed wound healing and clonogenic assays using conditioned medium (CM) of activated ILC2s, pre-treated or not with T0070907 (ILC2 CM and ILC2 + T0070907, respectively). No effect was observed on the CRC cell proliferation (Supplementary Fig. 4b). However, we found that the presence of ILC2 CM increased the migration of SW1116 cells, whereas addition of ILC2 + T0070907 CM lead to the opposite effect (Fig. 5d, e). Likewise, the colony formation assay confirmed that ILC2 CM increased the number of SW1116 colonies compared to control and to ILC2 + T0070907 CM (Fig. 5f). Moreover, the addition of an anti-IL-13 blocking antibody abrogated the development of SW1116 colonies, confirming the involvement of IL-13 in the ILC2-PPARγ dependent effect on CRC cells (Fig. 5f). In order to corroborate our findings on the pro-tumor effects of ILC2 CM for the migration and invasion of CRC, we evaluated in the SW116 line the expression of MMP9 and N-cadherin, two molecules that have been linked to the progression and invasion of tumor cells40,41,42. We found that the addition of ILC2 CM significantly increased the expression of these two EMT markers on CRC cells, whilst the addition of ILC2 + T0070907 CM or the anti-IL-13 blocking antibody reduced their expression towards basal levels (Fig. 5g). Taken together our data suggest a crosstalk between cancer cells and ILC2s, in which cancer cells trigger ILC2s to produce IL-13, which in turn sustains their migratory and invasive capacity, yet without affecting their proliferation rate. Inhibition of PPARγ in this circuit thwarts the pro-tumorigenic role of ILC2s.

PPARγ is expressed and functional in murine ILC2s

To translate our findings into an in vivo setting, we assessed the expression of PPARg in freshly-sorted ILC subsets from the lung of C57BL/6 mice. In agreement with our human data, PPARg was specifically expressed in ILC2s, while it was not detectable in ILC1s and ILC3s (Fig. 6a). Therefore, we exposed murine ILC2s to T0070907 in vitro and could thereby recapitulate the results obtained using human ILC2s, with a reduction in the production of both IL-13 and IL-5 (Fig. 6b, c). Next, to define the role of PPARγ in ILC2s beyond its pharmacological inhibition, we used a transgenic mouse model with an ILC-specific PPARγ inducible knockout (KO) by crossing Ppargfl/fl animals with mice expressing the Cre enzyme under the Id2 promoter, known to control ILC development43. This model allows tamoxifen-mediated induction of Cre activity in time and in an ILC-specific manner44 (Fig. 6d). We next sensitized Ppargfl/flId2CreERT2 positive or negative mice (also reported as Ppargfl/fl Id2Cre+and Ppargfl/fl Id2Cre, respectively) with IL-33 and IL-25 i.p. for 3 days, after 5 days of tamoxifen injection. As shown in Fig. 6e, ILC2s sorted from Ppargfl/fl Id2Cre+ mice secreted significantly less IL-13 and IL-5 compared to control Ppargfl/fl Id2Cre− mice. In addition, similarly to the results observed in humans, we found motives partially matching the mouse PPARγ-Rxrα heterodimer binding motif in the promoter region of both Il-5 and Il-13 genes (Fig. 6f, Supplementary Data 2). Taken together these results confirm the role of PPARγ as a key regulator of ILC2 function in vivo.

PPARγ expression in ILC2s influences cancer progression in a CRC murine model

Next, to investigate the in vivo role of ILC2s in CRC development and progression, we used a heterotopic murine model of CRC in RORafl/sgIl7rCre mice (referred to as ILC2 KO mice)45. We injected subcutaneously (s.c.) MC-38 murine genetically modified CRC cells to constitutively produce IL-33 (reported as MC-38-IL33). Tumor cells were injected into the right flank of ILC2 KO and littermate control mice, and survival was monitored. As displayed in Fig. 7a, tumor progression was slower in ILC2 KO mice compared to littermate controls, resulting in better survival advantage in the KO strain, suggesting a pro-tumorigenic function of ILC2s in this model. This effect was reverted after the adoptive transfer of expanded ILC2s sorted from the lungs of C57BL6 mice (Fig. 7b). Next, to further assess the contribution of PPARγ to ILC2s during tumorigenesis, we injected MC-38-IL33 cells into Ppargfl/flId2Cre+ mice and Ppargfl/fl Id2Cre− mice (Fig. 7c). Strikingly, after 18 days, tumor volume was reduced in Ppargfl/fl Id2Cre+ mice compared to control (Fig. 7d). In line with these data, Ppargfl/flId2Cre+ mice showed a longer survival compared to control mice (Fig. 7e). To confirm our in vitro results on the involvement of ILC2s in the EMT phenomenon, we evaluated by qPCR analysis the expression of murine MMP-9 and N-cadherin in dissociated tumor tissues. As shown in Fig. 7f, higher expression of both Mmp9 and Ncad in Ppargfl/flId2Cre− compared to Ppargfl/flId2Cre+ tumors were observed. In addition, to evaluate the therapeutic potential of PPARγ inhibition in CRC development and progression, we treated tumor-bearing mice with T0070907. As shown in Supplementary Fig. 6a, treatment with T0070907 significantly reduced both tumor volume and weight, suggesting a possible therapeutic use of PPARγ inhibitors in CRC patients. Furthermore, we also assessed the pro-tumor effects of IL-13 in vivo by treating tumor-bearing mice with an anti-IL-13 neutralizing antibody. After 18 days of treatment, both tumor volume and weight were significantly reduced in treated mice compared to controls (Supplementary Fig. 6b). Taken together, these results underscore the relevance of ILC2s and PPARγ in fostering cancer cell progression and subsequent metastatic potential.

ILC2s are emerging as key regulators of Type-2 immune responses, by influencing downstream adaptive immunity in homeostasis and disease. However, both extrinsic and cell-intrinsic regulatory pathways dictating ILC2 functions remain elusive. Here, we demonstrate that PPARγ is selectively expressed in ILC2s, both in humans and mice. Its pharmacological inhibition impairs mitochondrial fitness and reduces Type-2 cytokine secretion in ILC2s upon IL-33 stimulation. Furthermore, in the environment of a human IL-33-enriched cancer type, such as CRC, ILC2s show PPARγ-dependent pro-tumoral functions. Lastly, by inducible and conditional deletion of PPARγ in ILCs using PPARgfl/flId2CreERT2mice and a heterotypic tumor model, we show significant reduction of cancer development and progression in vivo. Overall, our results point to a PPARγ-driven pro-tumorigenic role of ILC2s in IL-33-dependent tumors and suggest that PPARγ targeting in ILCs may become an attractive add-on therapy for ILC2-driven diseases, including cancer.

It has been reported that fatty acid metabolism-related genes (including hexokinase 2 (HK2), pyruvate dehydrogenase kinase 1 (PDK1), and FATP-6 play a key role in ILC2 functions during helminthic infections13. In other immune cell types, PPARγ was shown to act as a master regulator of lipid metabolism in alveolar macrophages46, as inhibitor of pro-inflammatory cytokine secretion47, and to promote type-2 responses48. In that regard, Nobs et al. found that IL-33 drives Th2 responses and the development of pulmonary allergic inflammation by induction of PPARγ in CD4+ T cells and DCs22. More recently, it has been reported that Th9 function is positively regulated by PPARγ, which promotes IL-9 secretion in acute allergic skin diseases49. Different ILC transcriptomic analysis previously reported on PPARγ mRNA expression in ILC2s, in both human and mouse tissues21,50. Moreover, it has been recently reported that PPARγ is an important regulator of ILC2 during allergic airway inflammation51,52. In addition, Batyrova et al., demonstrated that the expression of PD1 on ILC2s is controlled by PPARγ and affects the production of IL-5 and IL-13 in vivo53. However, the functional relevance of this nuclear receptor in ILC2s is still limited, particularly in humans. In contrast to the activation-induced PPARγ expression in CD4+ Th2 cells22, our results argue for a steady-state expression of PPARγ in ILC2s, that apparently acts as an homeostatic regulator of Type-2 cytokine secretion, likely in the presence of endogenous ligands such as PGD2-derived 15d-PGJ2, as it was previously reported19. It has been demonstrated that PPARγ stimulation, via the induction of PGC-1α, promotes mitochondrial biogenesis in different cell types54. Our findings are consistent with a role of PPARγ in the maintenance of energetic homeostasis also in ILC2s, possibly by boosting PGC-1α expression and promoting mitochondrial activity. Further, the expression of IL-13 and IL-5 could be directly regulated by the PPARγ-RXRα heterodimer in Type-2 cytokine epigenetically-poised ILC2s55, as suggested by the presence of potential PPARγRXRα binding elements in the promoter regions of these genes. Whether or not PPARγ expression is heterogeneous in distinct developmental stages or/and in sub-populations of ILC2s across tissues remains to be elucidated by high-definition screening methods, such as single-cell mRNA sequencing.

Accumulating evidence has indicated that IL-33 and its receptor ST2 (also known as IL1RL1) represent a key inflammatory pathway in tumor biology and immunology34,56. In particular, several studies reported that IL-33 and ST2 are implicated as potent modulators of the TME promoting immune cells recruitment and tumor malignancy in different types of cancer, including breast cancer, hepatocellular carcinoma, gastric cancer, and CRC31,57,58. In CRC, different reports showed that IL-33 and ST2 are highly expressed in tumor tissues and correlated with tumor development31 or advanced tumor stage and poor prognosis59,60,61. ILC2 tumor infiltration has been reported by us and others in different cancer types, being linked with both pro- and anti-tumor activities depending on the context10. Signaling through ST2 can directly stimulate Type-2 cytokine production by ILC2s62. In turn, ILC2-derived, IL-33-triggered IL-13 was shown to favor myeloid-derived suppressor cells (MDSCs) induction and to be linked to recurrence in bladder cancer7,57,63. Moreover, IL-13 has been reported to promote EMT via IL-13R/STAT6 signaling, but nothing was previously reported on a possible link between EMT and ILCs39. Our results show that ILC2s exert an essential role in the tumor-driven IL-33/ST2/IL-13 axis, by promoting the migration and invasion of CRC cancer cells, which are key factors promoting metastasis in vivo64. Our results indicate that PPARγ is key in this interaction between ILC2s and cancer cells, by supporting ILC2s in their pro-tumoral functions, that might be conserved across different IL-33 dependent tumors, such as breast and liver cancer. However, in other tumor types, ILC2s might exert anti-tumor functions, thus arguing that the benefit of targeting ILC2s is tumor type-dependent65. To date, conflicting data have been reported on PPARγ in cancer development and progression66. On the one hand, PPARγ expression has been associated with good prognosis in CRC67 and the use of the PPARγ agonists troglitazone and rosiglitazone inhibited tumorigenesis in CRC and bladder cancer models by a tumor-directed effect68,69. On the other hand, in Apcmin mice the same treatment increases the number of tumors, and the use of PPARγ inhibitors (including T0070907) suppressed breast cancer cell proliferation in vitro70 and restrained CRC cell migration and invasion both in vitro and in vivo71,72. In our setting of PPARγ targeting in ILC2s, the observed anti-tumoral phenotype obtained by the in vivo conditional and inducible genetic deletion of PPARγ in ILCs suggests that ILC-specific and temporal targeting of PPARγ might beneficial in cancer. Thus, we identified PPARγ as a druggable target to suppress the potent and rapid ILC2 effector functions. Its pharmacologic inhibition might be used as an add-on therapy for cancer treatment and/or in other ILC2-mediated diseases, as recently demonstrated in allergy and asthma22,48,73.