Cannabinoid Receptor II engagement promotes ILC2 expansion and enhances ILC2-dependent airway hyperreactivity

Abstract

Background:

Cannabinoids modulate the activation of immune cells and physiological processes in the lungs. Group-2 innate lymphoid cells (ILC2)s are central players in type-2 asthma, but how cannabinoids modulate ILC2 activation remains to be elucidated

Objective:

Our goal is to investigate the effects of cannabinoids on ILC2s and their role in asthma.

Methods:

A combination of Cannabinoid receptor (CB)2 KO mice, CB2 antagonist and agonist were used in the mouse models of IL-33, IL-25 and Alternaria alternata ILC2-dependent airway inflammation, and RNA sequencing was performed to assess transcriptomic changes in ILC2s. Humanized mice were used to assess the role of CB2 signaling in human ILC2s.

Results:

We here provide evidence that CB2 signaling in ILC2s is important for the development of ILC2-driven airway inflammation in both mice and human. We showed that both naïve and activated murine pulmonary ILC2s express CB2. CB2 signaling did not affect ILC2 homeostasis at steady state, but strikingly stimulated ILC2 proliferation and function upon activation. As a result, ILC2s lacking CB2 induced lower lung inflammation, as we made similar observations using a CB2 antagonist. Conversely, CB2 agonism remarkably exacerbated ILC2-driven airway hyperreactivity and lung inflammation. Mechanistically, transcriptomic and protein analysis revealed that CB2 signaling induced CREB phosphorylation in ILC2s. Human ILC2s expressed CB2, as CB2 antagonism and agonism showed opposing effects on ILC2 effector function and development of airway hyperreactivity in humanized mice.

Conclusion:

Collectively, our results define CB2 signaling in ILC2s as an important modulator of airway inflammation.

Clinical implication:

Our findings highlight the stimulatory capacity of cannabinoids on ILC2s and offer new therapeutic avenues, including the use of substances or pathways able to modulate CB2 capable of alleviating lung function in patients with lung inflammation.

Capsule summary:

This study is the first to highlight the role of CB2 signaling in promoting ILC2-dependent airway inflammation, opening avenues for the development of novel adapted therapeutic strategies.

INTRODUCTION

Group 2 innate lymphoid cells (ILC2)s are found in a variety of tissues including the lungs, where they play a crucial role in promoting airway inflammation (1, 2). They contribute to an ongoing type-2 immune response, but also remarkably can induce type-2 inflammation independently of adaptive immunity (3–6). ILC2s develop in the bone marrow (BM) from common lymphoid progenitors (CLPs), as they strategically localize in the lungs near the bronchioles for a rapid response to local stimuli (7, 8). In response to allergen exposure, epithelium-derived interleukin (IL)-25, IL-33 and thymic stromal lymphopoietin (TSLP) rapidly induce ILC2 expansion and production of large amounts of type-2 cytokines. Together, ILC2-derived IL-5 and IL-13 promote airway inflammation by inducing eosinophil recruitment and smooth muscle cell contraction, respectively (9). In recent years, we and others have shown that the severity of asthma symptoms can be affected by modulating ILC2 function (10). ILC2s notably respond to environmental soluble factors including cytokines (11, 12), lipid mediators (13, 14), hormones (15, 16) and neuronal factors (17, 18), but also communicate with surrounding cells via the expression of co-stimulatory (19, 20) and adhesion receptors (21, 22). Therefore, a better understanding of the mechanisms controlling ILC2 activation in the lungs, particularly their immune modulatory effects on asthma symptoms, will help to develop potential therapeutic targets, either alone or in combination with other therapies.

The endocannabinoid system (ECS) is an essential signaling pathway involved in many physiological processes ranging from affecting the central nervous system to modulating inflammation in peripheral tissues (23). It is composed of two endogenous signaling lipids – N-arachidonoyl-ethanolamine (AEA) (24) and 2-arachidnoyglycerol (2-AG) (25, 26) – and their two main G protein-coupled receptors (GPCRs) – cannabinoid receptor (CB)1 and CB2 (27, 28). While CB1 is mainly found in the brain and involved in the control of the nervous system, CB2 is found in peripheral tissues and primarily on immune cells (29, 30). The ECS can however be manipulated by external cannabinoids, such as compounds present in psychotropic preparations of Cannabis sativa, as it is now well appreciated that phytocannabinoids such as Δ⁹-tetrahydrocannabinol (THC) can bind to and activate cannabinoid receptors (31, 32). Furthermore, synthetic CB agonists and antagonists have been widely used to study CB1 and CB2 functions, opening up avenues for the therapeutic potential of ECS manipulation (33, 34). Alterations in the ECS are associated with diseases such as allergy, and a better understanding of the molecular mechanisms and mode of action driving relevant immune cell functions is becoming progressively necessary (30). In human asthmatics, allergen challenge leads to increased concentrations of AEA in the bronchoalveolar lavage (BAL) fluid, correlating with the levels of IL-5 (35, 36), while several studies suggest that cannabinoids exacerbate lung inflammation in mice (37, 38). Interestingly, observational studies further report that frequent combustible THC-containing marijuana smoking is associated with increased shortness of breath, wheeze and bronchitic symptoms compared to regular cigarette smoking (39–43), as a recent study shows similar results with marijuana vaping in young Southern California adults (44). While these findings show that cannabinoids are involved in the pathophysiology of asthma, how they enhance lung inflammation still remains largely unknown.

Here, we show that pulmonary ILC2s constitutively express CB2, and that CB2 engagement promotes ILC2 function. Mice treated with a CB2 agonist developed exacerbated lung inflammation in response to IL-33 challenge, associated with increased ILC2 expansion and cytokine secretion. As a proof of concept, CB2 KO mice developed less airway inflammation associated with decreased numbers of lung ILC2s following IL-25, IL-33 or Alternaria alternata intranasal challenge. Although CB2 KO and WT ILC2s showed comparable survival rates in the lungs, CB2 KO ILC2s proliferated less and produced less IL-5 and IL-13. As a result, CB2 KO ILC2s adoptively transferred to alymphoid mice failed to induce efficient airway inflammation in response to IL-33. Importantly, we observed similar effects using a selective CB2 antagonist. Of note, our transcriptomic analysis revealed that CB2 signaling induced cAMP response element binding protein (CREB) phosphorylation in ILC2s. Finally, we confirmed our findings in a human setting using a humanized mouse model. Human blood ILC2s constitutively expressed CB2, while CB2 antagonism inhibited human ILC2-driven airway inflammation and CB2 agonism remarkably exacerbated it. Our murine and human findings define CB2 signaling in ILC2s as an important modulator of airway inflammation and reveal a previously unappreciated mechanism driving ILC2 function.

MATERIAL AND METHODS

Mouse experiments

Experimental protocols were approved by the USC institutional Animal Care and Use Committee (IACUC) and conducted in accordance with the USC Department of Animal Resources’ guidelines. 5-10 weeks old age and sex-matched mice were used in the studies. C57BL/6J, BALB/cByJ, Rag2^−/− (C.B6(Cg)-Rag2^tm1.1Cgn/J), Rag2^−/− IL-2Rg^−/−(C;129S4-Rag2tm1.1Flv Il2rgtm1.1Flv/J) and CB2 KO (B6.129P2-Cnr2^tm1Dgen/J) were purchased from The Jackson Laboratory and bred in our animal facility at the Keck School of Medicine, University of Southern California (USC).

Murine ILC2, in vitro culture and cytokine measurements

Murine ILC2s were FACS-sorted to a purity of >95% on a FACSARIA III system. ILC2s were gated as live lineage (CD3ε, CD4, CD5, TCRβ, TCRγδ, CD45R/B220, CD335, CD11c, CD11b, Gr1, FcεRIα and Ter119) negative CD45⁺, ST2⁺, CD127⁺ cells. When indicated, ILC2s were sorted from the lungs of mice challenged for 3 consecutive days intranasally with either PBS (naïve, nILC2) or rmIL-33 (activated, aILC2). Isolated ILC2s were washed in PBS 1X and either a.) directly adoptively transferred to Rag2^−/−IL-2Rg^−/− mice as described above or b.) cultured ex vivo for the indicated readouts. For adoptive transfer and ex vivo culture experiments, ILC2 yields of 5-10 mice were pooled to achieve a sufficient amount of ILC2s to perform the experiments. ILC2s were cultured at 37°C (5x10⁴/mL) for 24 hours in 96-well U-bottom plates with rmIL-2 (10ng/mL, Biolegend) and rmIL-7 (10ng/mL, Biolegend) in complete RPMi (cRPMi) supplemented with 10% heat-inactivated FBS (Omega Scientific), 100 units/mL penicillin and 100mg/mL streptomycin (GenClone). When indicated, CB2 agonist (JWH133, Cayman Chemical, 10μg/mL), CB2 antagonist (AM630, Tocris Bioscience, 10μg/mL) or vehicle (0.0002% DMSO) were added to the culture. In some experiments, TSLP (eBioscience, 10ng/mL) or DTA-1 (Bioxcell, 5μg/mL) were added to cultures. Culture supernatants were collected, and cytokines measured using the LEGENDplex™ Mouse Th2 Panel (Biolegend), used according to the manufacturer’s instructions. When indicated, in vitro proliferation was measured using the CellTrace™ Violet Cell proliferation kit for flow cytometry (Thermofisher), used according to the manufacturer’s instructions.

Measurement of lung function and lung histology

Following intranasal challenges, lung function was measured using the FinePointe RC system (Buxco Research Systems) as described previously (20). Briefly, mice were surgically tracheotomized under deep anesthesia and placed on the mechanically ventilated system where increasing doses of methacholine (acetyl-β-methylcholine chloride (Sigma) are sequentially nebulized. Methacholine is a bronchoconstrictor inducing airway contraction and was nebulized at various doses in a total volume of 3.5μL, ranging from 0 mg/mL to 40mg/ml. For each dose, lung resistance and dynamic compliance were measured and computed over a period of 3 minutes. When stated, one lobe per lung was collected and stored in PFA 4% for histology. Briefly, lungs were embedded in paraffin and sections of 4mm prepared for hematoxylin and eosin (H&E) staining. Histology pictures were acquired on a Leica DME microscope and Leica ICC50HD camera (Leica) and analyzed with ImageJ.

Statistical analysis

Experiments were repeated at least two times (n=3-15) and data are shown as a representative experiment, except for the RNAseq performed in Figure 4. Non-parametric tests were used: Mann-Whitney U tests were used to compare the differences between two groups, except for multi-group comparisons where Kruskal Wallis tests were used. All tests were performed using Prism Software (GraphPad Software Inc.). The degree of significance was indicated as: *p<0.05, **p<0.01, ***p<0.001.

Fig. 4 |. CB2 signaling enhances ILC2 function via the CREB pathway.

WT and CB2 KO mice were challenged intranasally for 3 days with rmIL-33. On day 4, lung ILC2s were FACS-sorted and cultured with rmIL-2 and rmIL-7 for 24 hours. (A-F), RNA-seq was performed after 24 hours of culture. (A) Schematic overview of downstream CB2 regulatory signaling elements. (B) Heatmap illustrating gene set analysis (GSA) for RNA-seq expression, showing the differentially regulated genes in CB2 KO compared to WT ILC2s. Differentially expressed genes were selected based on p-value <0.05. (C) Ingenuity pathway analysis (IPA) results showing associated statistically significantly downregulated canonical pathways in CB2 KO ILC2s. Differentially regulated genes relevant to (D) ILC2 and CB2 signaling, (E) Cyclins and cell cycle regulation and (F) ILC2 proliferation signature. Differentially expressed genes were selected based on p-value <0.05. (G) ILC2 apoptosis and corresponding quantitation showing frequencies of early apoptotic (E.A., AnnexinV⁺ DAPI⁻) and late apoptotic/necrotic (L.A., AnnexinV⁺ DAPI⁺) ILC2s. Intranuclear expressions of (H) Bcl2, (I) Ki67, (J) pCREB, (K) pERK1/2, (L) p38, (M) pAKT and (N) GATA-3 and corresponding quantitation presented as MFI. Histograms show mean +/− SEM of a representative experiment of 2-3 with n=3-5. ns= non-significant, * p<0.05, ** p<0.01.

RESULTS

CB2 is constitutively expressed on murine ILC2s:

It has been known for many years that CB2 is expressed on a variety of both human and rodent immune cells, based on a combination of genetic and protein analysis (28, 46). Since innate lymphoid cells such as ILC2s were only recently discovered, whether they express receptors for cannabinoids – in particular in the lungs – remains unknown. To this end, WT and CB2 KO mice were challenged intranasally (i.n.) with recombinant mouse (rm)IL-33 or PBS on 3 consecutive days followed on day 4 with lung ILC2 isolation and staining for CB2. Murine lung ILC2s were gated as live CD45⁺ Lineage⁻ CD127⁺ ST2⁺ cells and expressed high levels of major transcription factor GATA-3 (Fig. 1A). Compared to CB2 KO ILC2s, both PBS and IL-33-treated ILC2s remarkably expressed CB2 (Fig. 1B). Of interest however, CB2 was not inducible by IL-33 as both PBS and IL-33 treated ILC2s expressed similar levels (Fig. 1C). Together, these results therefore suggest that both naïve and activated ILC2s express CB2 on the surface.

Fig. 1 |. CB2 is expressed on ILC2s and enhances ILC2-dependent AHR.

(A) Lung ILC2s were gated as single live CD45⁺, Lineage⁻, ST2⁺ and CD127⁺ cells, expressing high levels of GATA-3. (B-H) C57BL/6 (WT) or CB2 KO mice were challenged intranasally for 3 days with rmIL-33 or PBS. On day 4, lung ILC2s were isolated by FACS. (B) CB2 expression on ILC2s and (C) corresponding quantitation presented as CB2 MFI. On day 4, analysis of lung function, BAL cellular content and lung histology were performed. (D) Lung resistance and (E) dynamic compliance in response to methacholine. (F) Total BAL CD45⁺ cells and (G) BAL cellularity. (H) Lung H&E staining. (I-O) WT or CB2 KO mice were challenged intranasally for 5 days with A. alternata or PBS. On day 6, lung ILC2s were isolated by FACS. (I) CB2 expression on ILC2s and (J) corresponding quantitation presented as CB2 MFT On day 6, analysis of lung function, BAL cellular content and lung histology were performed. (K) Lung resistance and (L) dynamic compliance in response to methacholine. (M) Total BAL CD45⁺ cells and (N) BAL cellularity. (O) Lung H&E staining. Histograms show mean +/− SEM of a representative experiment of 3 with n=4-9. Scale bars 50μm. eosinophils (eos.), polymorphonuclear neutrophils (PMN). ns= non-significant, * p<0.05.

CB2 signaling is required for the development of efficient ILC2-dependent airway inflammation:

We next investigated the functional requirements of CB2 for the development of ILC2-dependent airway hyperreactivity (AHR) and lung inflammation. WT and CB2 KO mice were challenged i.n. with PBS or IL-33 on 3 consecutive days. Compared to PBS controls, both WT and CB2 KO mice challenged with IL-33 i.n. showed higher lung resistance, but that of CB2 KO mice was lower compared to WT mice (Fig. 1D). Furthermore, CB2 KO mice showed an improved dynamic compliance as compared to WT controls in response to IL-33 (Fig. 1E). Whereas both WT and CB2 KO mice challenged i.n. with PBS had similarly low numbers of BAL CD45⁺ cells, IL-33 i.n. challenge increased BAL CD45⁺ cells in both WT and CB2 KO mice (Fig. 1F). In line with our previous findings however, CB2 KO mice showed lower lung inflammation compared to WT controls in response to IL-33 i.n. In particular, the numbers of eosinophils were reduced in the BAL of CB2 KO mice (Fig. 1G). Furthermore, WT and CB2 KO mice showed similar numbers of polymorphonuclear neutrophils (PMN), SiglecF⁺ CD11c⁺ cells and - although minimal – a significant reduction in CD3+ lymphocytes, together indicating that IL-33 induced lung inflammation is impaired in the absence of CB2 signaling (Fig. 1H). We next sought to confirm our findings in a more physiologically relevant setting using Alternaria alternata (A. alternata), a common allergen known to induce ILC2-dependent AHR. We found that i.n. challenge with A. alternata did not induce CB2 expression in ILC2s, as both naïve and A. alternata-treated mice expressed similar levels of CB2 on the surface (Fig. 1I and J). Importantly however and similar to our previous finding, mice lacking CB2 developed less lung resistance as compared to WT controls in response to increasing doses of methacholine (Fig. 1K). Furthermore, CB2 KO mice showed an improved dynamic compliance as compared to WT controls in response to A. alternata (Fig. 1L). These observations were associated with lower lung inflammation, in particular as the numbers of eosinophils were reduced in the BAL of CB2 KO mice (Fig. 1M–O). Taken together, these findings suggest that CB2 is important for the development of ILC2-induced AHR and airway inflammation.

Lack of CB2 on ILC2s inhibits ILC2 tissue expansion:

To investigate the role of CB2 signaling in the development of ILC2s, we next analyzed the frequencies of ILC2 progenitors in the bone marrow (BM) of WT and CB2 KO mice at steady state. Progenitors were gated as CLPs (CD45⁺ Lin⁻ CD127⁺ α4β7⁻ Flt3⁺), α-LPs (CD45⁺ Lin⁻ CD127⁺ α4β7⁺ Flt3⁻), ChILPs (CD45⁺ Lin⁻ CD127⁺ α4β7⁺ Flt3⁻ CD25⁻), and ILC2Ps (CD45⁺ Lin⁻ CD127⁺ α4β7⁺ Flt3⁻ CD25⁺) as shown in Fig. 2A and recently described (7). We did not observe any significant effects of CB2 signaling on BM progenitor frequencies (Fig. 2B). Furthermore, the numbers of immature ST2⁺ ILC2s found in the BM were similarly unaffected by CB2 signaling (Fig. 2C). As a result, both WT and CB2 KO mice showed similar low numbers of ILC2s in the lungs at steady state (Fig. 2D,E, left panels). Of note, numbers of ILC2s were similarly unaffected in other peripheral tissues such as the adipose tissue (data not shown). Strikingly however, although IL-33 i.n. challenge expanded lung ILC2 numbers in both WT and CB2 KO mice, we found lower numbers of ILC2s in CB2 KO mice as compared to controls (Fig. 2D,E, right panels). Comparable trends were observed when representing Lineage⁻ ST2⁺ CD127⁺ ILC2s as a proportion of live CD45⁺ cells (Fig. E1A) (47). Furthermore, CB2 KO mice challenged with A. alternata also showed a decreased lung ILC2 expansion compared to WT controls (Fig. E3A and B). Taken together, these findings reveal that although CB2 is not involved in ILC2 development at steady state, it contributes to lung ILC2 accumulation during inflammation.

Fig. 2 |. CB2 signaling stimulates ILC2s but is not required for ILC2 development and homeostasis.

(A) BM ILC2 progenitors gated as CLPs (CD45⁺ Lin⁻ CD127⁺ α4β7⁻ Flt3⁺), αLPs (CD45⁺ Lin⁻ CD127⁺ α4β7⁺ Flt3⁻), ChILPs (CD45⁺ Lin CD127⁺ α4β7⁺ Flt3⁻ CD25⁻), ILC2Ps (CD45⁺ LiN⁻ CD127⁺ α4β7⁺ Flt3⁻ CD25⁺). (B) Frequencies within CD45⁺ BM cells of CLPs, αLPs, ChILPs and ILC2Ps in naïve WT and CB2 KO mice. (C) Numbers of ST2⁺ ILC2s in the BM (one femur). (D-L) WT or CB2 KO mice were challenged intranasally for 3 days with rmIL-33 or PBS and on day 4, lung ILC2 numbers, activation, apoptosis and proliferation were analyzed. (D) Live CD45⁺ Lin⁻ CD127⁺ ST2⁺ lung ILC2s and (E) corresponding quantitation showing the numbers of ILC2s per lung. (F) Early apoptotic (E.A., AnnexinV⁺ DAPT⁻) and late apoptotic/necrotic (L.A., AnnexinV⁺ DAPI⁺) ILC2s (black dots: WT mice, red dots: CB2 KO mice) and (G) corresponding quantitation showing frequencies of E.A. and L.A. (H) Lung ILC2 intranuclear Ki67 expression and (I) corresponding quantitation showing Ki67 MFI. (J) IL-13 (top panel) and IL-5 (bottom panel) positive ILC2s and corresponding quantitation of (K) IL-13- and (L) IL-5-positive lung ILC2s, showing frequencies of protein expression. Histograms show mean +/− SEM of a representative experiment of 3 with n=4-14. ns= non-significant, * p<0.05. FMO: Full-minus-one.

CB2 signaling stimulates activated but not naïve pulmonary ILC2s:

Since WT and CB2 KO mice showed different lung ILC2 numbers upon activation but not at steady state, we next measured the effects of CB2 signaling on tissue ILC2 homeostasis and activation by IL-33. WT and CB2 KO mice were challenged with PBS or IL-33 i.n. on 3 consecutive days and we measured lung ILC2 apoptosis and proliferation on day 4 in vivo. We found that CB2 signaling did not affect ILC2 survival at steady state nor following IL-33 i.n. challenge, as evidenced by the frequencies of both early apoptotic (AnnexinV⁺ DAPI⁻) and late apoptotic/necrotic (AnnexinV⁺ DAPI⁺) ILC2s (Fig. 2F–G). Although IL-33 administration efficiently expanded both WT and CB2 KO lung ILC2s compared to controls, we found that the proliferative capacity of activated ILC2s was lower in the absence of CB2, as evidenced by the lower expression of intranuclear proliferative marker Ki67 (Fig. 2H,I). Importantly, this finding was further confirmed with an ex vivo proliferation assay, as we found that IL-33-activated CB2 KO ILC2s cultured for 24 hours ex vivo showed a lower proliferation index compared to controls, associated with lower cell divisions (Fig. E2A). To further assess the ability of CB2 to stimulate activated – but not steady state – ILC2s, we next measured the capacity of ILC2s to produce effector cytokines after ex vivo stimulation with PMA and ionomycin. We found that WT and CB2 KO ILC2s at steady state produced similar basal levels of both IL-13 (Fig. 2J,K left panels) and IL-5 (Fig. 2J,L, left panels). Although IL-33 challenge enhanced cytokine production by lung ILC2s, both IL-13 (Fig. 2J,K right panels) and IL-5 (Fig. 2J,L, right panels) intracellular expressions were decreased in the absence of CB2 signaling. Furthermore, intranuclear GATA-3 expression was also decreased in CB2 KO ILC2s as compared to WT controls (Fig. E2B). Interestingly we found that compared to controls, CB2 KO ILC2s expressed lower levels of ST2, CD25 and KLRG1 but comparable levels of CD127 and CD90 (Fig. E2D–H). Importantly, we made similar findings on ILC2 survival, proliferation and activation following A. alternata (Fig. E2C and Fig. E3C–G) and IL-25 (Fig. E4) challenges. We next incubated FACS-sorted naïve and IL-33-activated lung ILC2s in the presence of IL-2 and IL-7, two cytokines sustaining ILC2 proliferation and function. Following 24 hours of culture, we found that naïve WT and CB2 KO ILC2s both produced low and similar levels of IL-5 and IL-13 (Fig. E1C). In line with our previous findings however, activated CB2 KO ILC2s produced less IL-5 and IL-13 compared to WT controls.

To assess whether the observed effects are inducible, we performed a series of ex vivo and in vivo experiments using AM630, a potent and selective CB2 antagonist (Fig. E5). Synthetic CB2 antagonists have been widely used to study the effects of cannabinoid receptors (33). We challenged Rag2^−/− mice – lacking T-cells and B-cell – with IL-33 i.n. with or without CB2 antagonist, and on day 4 measured lung resistance and ILC2 function (Fig. E5A). Similar to our findings using CB2 KO mice, mice challenged with CB2 antagonist showed lower lung resistance in response to methacholine (Fig. E5B). This phenotype was associated with lower BAL eosinophilia (Fig. E5C) and lung ILC2 numbers (Fig. E5D). Furthermore, ILC2 survival was unaffected by the antagonist (Fig. E5E) while ILC2 proliferation was significantly reduced following antagonist treatment, as evidenced by intranuclear Ki67 expression (Fig E5F). Similar to our findings shown in Fig. 2, both IL-5 (Fig. E5G) and IL-13 (Fig. E5H) were significantly reduced in the antagonist-treated mice, as well as intranuclear GATA-3 (Fig. E5I). Finally, activated ILC2s treated for 24h ex vivo with a CB2 antagonist produced less IL-5 and IL-13 compared to WT controls (Fig. E1D). Taken together, although not required for the maintenance of lung ILC2s at steady state, these findings suggest that CB2 signaling is important for the development of AHR and airway inflammation by promoting ILC2 tissue expansion and activation.

CB2 KO ILC2s exhibit impaired ability to induce airway hyperreactivity:

We next investigated whether the effects of CB2 signaling we observed on IL-33-driven lung inflammation were ILC2-specific. To this end, we performed a series of adoptive transfer experiments into host alymphoid Rag2^−/− IL-2Rg^−/− mice which lack T-cells, B-cells and ILC2s as previously described (11, 20). WT or CB2 KO activated lung ILC2s (aILC2s) were adoptively transferred into Rag2^−/− IL-2Rg^−/− mice which were then challenged on 3 consecutive days with PBS or IL-33 i.n (Fig. 3A). On day 4, mice transferred with both WT and CB2 KO ILC2s and challenged with IL-33 developed higher lung resistance compared to PBS controls (Fig. 3B). However, CB2 KO ILC2s induced a lower lung resistance as compared to WT controls (Fig. 3B). CB2 KO ILC2s also induced improved dynamic compliance compared to WT controls in response to i.n. IL-33 (Fig. 3C). In line with our previous findings, BAL eosinophilia induced in response to IL-33 was lower in mice transferred with CB2 KO ILC2s compared to controls (Fig. 3D). Importantly, we further found less CB2 KO ILC2s in the lungs at day 4 as compared to WT controls, although identical numbers of ILC2s were initially transferred into host mice (Fig. 3E). We therefore generated adoptive transfer experiments using the WT CD45.1 and CB2 KO CD45.2 mouse models, allowing us to track ILC2 proliferation and survival of different genetic backgrounds within the same host (Fig. 3F). A 1:1 mixture of CD45.1:CD45.2 sorted aILC2s (Fig. 3G) was adoptively transferred into Rag2^−/− IL-2Rg^−/− mice which were then challenged on 3 consecutive days with IL-33 i.n. Transferred aILC2s were detected in the lungs at day 2 post transfer, as this population expanded over time at day 4 (Fig. 3H,I). Although both WT and CB2 KO ILC2s migrated to the lungs as efficiently (Fig. E1E–F), we found that over the period of 48 hours ranging from day 2 to day 4, WT ILC2s expanded approximately 4-fold, whereas CB2 KO ILC2s only approximately 2-fold (Fig. 3J). In line with this observation, we found that CB2 KO ILC2s expressed less intranuclear proliferative marker Ki67 both at day 2 and day 4 after transfer as compared to WT controls (Fig. 3K). Finally, we did not see any differences in ILC2 survival between WT and CB2 KO ILC2s at day 2 nor at day 4, confirming our previous findings that CB2 signaling does not play a significant role in ILC2 survival (Fig 3L). Taken together, our findings suggest that CB2 KO ILC2s fail to efficiently induce IL-33-driven airway inflammation.

Fig. 3 |. CB2 KO ILC2s exhibit impaired ability to induce airway hyperreactivity.

(A-E) FACS-sorted WT and CB2 KO activated ILC2s (aILC2) were adoptively transferred to Rag2^−/−IL-2Rg^−/− mice (5x10⁴ ILC2s/mouse), and then challenged intranasally for 3 days with rmIL-33 or PBS. On day 4, lung function, BAL eosinophilia and lung ILC2 numbers were analyzed. (A) Experimental design. (B) Lung resistance and (C) dynamic compliance in response to methacholine. (D) BAL eosinophils and (E) lung ILC2s. (F-M) WT (CD45.1) and CB2 KO (CD45.2) mice were challenged intranasally for 3 days with rmIL-33. On day 4, lung ILC2s were FACS-sorted and 5x10⁴ of a 1:1 mix of CD45.1:CD45.2 ILC2s/mouse was adoptively transferred to Rag2^−/−IL-2Rg^−/− mice. Host mice were then challenged intranasally for 3 days with rmIL-33 and lung ILC2 numbers, survival and proliferation were measured. (F) Experimental design. (G) Ratio of transferred CD45.1:CD45.2 aILC2s. (H) Total lung ILC2s 24h and 72h after transfer and (I) corresponding quantitation showing ILC2 numbers per lung. (J) CD45.1 and CD45.2 ILC2 fold-increase over 48h (24h to 72h post transfer). (K) ILC2 Ki67 MFI and (L) frequencies of live CD45.1 and CD45.2 ILC2s 24h and 72h post transfer. Histograms show mean +/− SEM of a representative experiment of 2 with n=5. ns= non-significant, * p<0.05, *** p<0.001. NT: no transfer.

CB2 signaling enhances ILC2 function via the CREB pathway:

Although CB2 signaling is well established in the central nervous system (48) there is a deficit of knowledge regarding immune cells. A recent report however convincingly showed that CB2 signals via the downstream activation of CREB which then affects cellular functions such as proliferation and cytokine secretion (49). The activation of CREB involves several components including cyclic AMP (cAMP) signaling, phosphorylation of ERK1/2 and AKT, and to a lesser extent that of p38 (Fig. 4A). We therefore performed a transcriptomic analysis to establish CB2 signaling in a pure population of sorted murine WT and CB2 KO activated lung ILC2s. Gene specific analysis (GSA) for RNA-seq showed that a significant number of genes were statistically modulated by the lack of CB2 expression (Fig. 4B and Table E1). More specifically, we analyzed major pathways modulated by CB2 signaling in ILC2s by using the ingenuity pathway analysis (IPA) tool (Fig. 4C). We found that kinetochore metaphase signaling as well as cyclins and cell cycle regulation pathways, both involved in cell cycle and cell division, were strongly downregulated in CB2 KO ILC2s. In addition, ILC2 signaling elements such as ERK5, Jak/Stat and IL-2 pathways were downregulated in CB2 KO ILC2s, and we further found that CB2-related element cAMP signaling was decreased in the absence of CB2 (Fig. 4C). Importantly, the undetected expression of Cd3, Tbx21 and Rorc transcripts in WT and CB2 KO ILC2s confirmed that our analysis did not include closely related T-cell contamination (Fig. 4D). Il2r and Il7r were downregulated in CB2 KO ILC2s, as were signaling elements Grb2 and Jak and major transcription factors Gata3 and Nfkb1, together confirming that ILC2s are functionally impaired in the absence of CB2 signaling (Fig. 4D). Furthermore, while both WT and CB2 KO ILC2s failed to express Cnr1 transcripts, WT mice expressed Cnr2 transcripts that were expectedly at background levels in CB2 KO ILC2s. We found that CB2 major signaling element Mapk1 – encoding for ERK1/2 – was downregulated in the absence of CB2 (Fig. 4D). Mechanistically, the downregulation of Mki67 – encoding for proliferative marker Ki67 – confirmed our previous results on the positive effects of CB2 signaling on ILC2 expansion (Fig. 4D). In line with this, we found that the majority of detected Ccn and Cdk genes were downregulated, as well as transcription factor E2f8, all involved in cell cycle progression (Fig. 4E). These observations were further strengthened with a general downregulation of detected genes recently shown to correlate with ILC2 proliferation (18) (Fig. 4F). At the protein level, although CB2 signaling affected neither AnnexinV/DAPI (Fig. 4G) nor anti-apoptotic marker Bcl2 (Fig. 4H) expressions, it promoted that of Ki67 in ILC2s (Fig. 4I). Mechanistically, we found that the absence of CB2 on ILC2s led to statistically significant decreases in the phosphorylation of CREB (Fig. 4J), ERK1/2 (Fig. 4K), p38 (Fig. 4L) and AKT (Fig. 4M). As a result, we observed that GATA-3 – a key ILC2 transcription factor GATA-3 involved in ILC2 function – was decreased in CB2 KO mice (Fig. 4N). Our findings therefore confirm at both the transcriptomic and protein levels that CB2 signaling promotes ILC2 function, at least in part via the phosphorylation of CREB.

CB2 agonism exacerbates ILC2-driven AHR and lung inflammation:

Cannabinoid receptors can be activated by external cannabinoids such as synthetic cannabinoids or phytocannabinoids (32). Since we observed that CB2 signaling promotes ILC2 proliferation and function, we therefore next assessed the effects of JWH133, a commonly used potent CB2 agonist (50), on the development of ILC2-driven airway hyperreactivity and lung inflammation. We first incubated freshly FACS-sorted WT ILC2s isolated from the lungs of PBS and IL-33 i.n.-challenged mice with or without CB2 agonist for 24 hours and measured IL-5 and IL-13 levels in the culture supernatants. Although the agonist did not affect the low levels of IL-5 and IL-13 in PBS-challenged ILC2s, we found that CB2 agonism strikingly induced ILC2 secretion of IL-5 and IL-13 when ILC2s were isolated from mice challenged i.n. with IL-33 (Fig. 5A). We further found that ILC2 survival following IL-33 stimulation was unaffected by CB2 agonism as evidenced by AnnexinV and DAPI expression (Fig. 5B). However, activated ILC2s treated with CB2 agonist induced a modest but statistically significant increase in proliferation, as evidenced by the ILC2 intranuclear Ki67 expression (Fig 5C). In contrast, CB2 agonism did not increase ILC2 proliferation when incubated with naïve ILC2s (Fig. E1G). We next assessed the effects of CB2 agonism in vivo. Similar to our previous approach using a CB2 antagonist (Fig. E5), Rag2^−/− mice were challenged with IL-33 i.n. with or without CB2 agonist, and on day 4 we measured lung resistance and ILC2 function (Fig. 5D). We found that mice that received CB2 agonist remarkably showed increased lung resistance as compared to controls (Fig. 5E). The phenotype was associated with an increased lung inflammation as evidenced by the greater lung inflammation (Fig. 5F), number of BAL CD45⁺ cells (Fig. 5G) and in particular increased BAL eosinophilia (Fig. 5H). Strikingly, we found significantly more lung ILC2s in mice challenged with CB2 agonist (Fig. 5I–J). Comparable trends were observed when representing Lineage⁻ ST2⁺ CD127⁺ ILC2s as a proportion of live CD45⁺ cells (Fig. E1B). In confirmation of our previous findings, although CB2 agonism did not affect ILC2 survival (Fig. 5K), it appeared to induce Ki67 expression following IL-33 i.n. challenge (Fig. 5L). Finally, we measured the capacity of ILC2s to produce effector cytokines in response to 4 hours of ex vivo PMA and ionomycin stimulation. We found that CB2 agonism increased the frequency of IL-5-producing ILC2s (Fig. 5M and N) but not that of IL-13 (Fig. 5O and P) following 3 days of IL-33 i.n. challenge. These findings suggest that CB2 agonism directly induces ILC2 expansion and effector function following IL-33 i.n. challenge, together enhancing AHR and airway inflammation.

Fig. 5 |. CB2 agonist enhances ILC2-dependent AHR.

(A-C) BALB/c mice were challenged intranasally for 3 days with rmIL-33 or PBS. On day 4, lung ILC2s were FACS-sorted and cultured with rmIL-2 and rmIL-7 with or without 10μg/mL CB2 agonist for 24 hours. (A) Levels of IL-5, IL-13 in the culture supernatants. (B) ILC2 apoptosis and corresponding quantitation showing frequencies of early apoptotic (E.A., AnnexinV⁺ DAPI⁻) and late apoptotic/necrotic (L.A., AnnexinV⁺ DAPI⁺) ILC2s. (C) ILC2 Ki67 expression and corresponding quantitation showing Ki67 MFI. (D-O) Rag2^−/− mice were challenged with rmIL-33 for 3 days intranasally with CB2 agonist (JWH133, 1mg/kg) or vehicle and euthanized on day 4. (D) Experimental design. (E) Lung resistance in response to methacholine. (F) Lung H&E staining, scale bars 50μm. (G) Total BAL CD45⁺ cells and (H) BAL eosinophils. (I) Live lung ILC2s and (J) corresponding quantitation showing the numbers of ILC2s per lung. (K) E.A. and L.A. ILC2s and corresponding quantitation showing frequencies of E.A. and L.A. (L) Lung ILC2 Ki67 expression and corresponding quantitation showing Ki67 MFI. (M) IL-5 and (O) IL-13 positive ILC2s and corresponding quantitation (N, P), showing frequencies of protein expression. Histograms show mean +/− SEM of a representative experiment of 3 with n=4-6. ns= non-significant, * p<0.05, ** p<0.01.

CB2 is expressed on human ILC2s and promotes ILC2-dependent AHR in humanized mice:

We next investigated the effects of CB2 signaling on human blood ILC2s (hILC2s) that were isolated from peripheral blood mononuclear cells (PBMCs) of healthy volunteers. We first measured the levels of human CB2 expression on hILC2s incubated with or without IL-33 for 72 hours ex vivo (Fig. 6A). Human ILC2s were gated as CD45⁺, Lineage⁻, CD127⁺ CRTH2⁺ cells (Fig. 6B). In accordance with our previous murine models, hILC2s expressed and maintained the same level of CB2 with or without IL-33 stimulation (Fig. 6C), as the levels of CB2 on hILC2s were similar (Fig. 6D). We then measured the effects of CB2 antagonism on hILC2 function ex vivo (Fig. 6E). Following incubation with a CB2 antagonist, we found that inhibiting CB2 signaling reduced IL-5, IL-6 and IL-13 concentrations in the culture supernatants (Fig. 6F). Conversely, we next investigated whether CB2 agonism could enhance hILC2 function ex vivo, as observed in our murine model (Fig. 6G). Remarkably, incubation of hILC2s with a CB2 agonist increased IL-5, IL-6 and IL-13 secretion in the culture supernatants (Fig. 6H). We therefore next assessed whether CB2 signaling was involved in the development of hILC2-dependent AHR and lung inflammation using the previously described humanized mouse model (20). Sorted hILC2s were adoptively transferred into host Rag2⁻ IL-2Rg^−/− mice which we then challenged on 3 consecutive days with PBS or IL-33 i.n. On day 4, we measured lung resistance and airway inflammation in response to IL-33 (Fig. 6I). Compared to PBS-challenged mice, IL-33 induced higher lung resistance (Fig. 6J, black line), associated with higher BAL eosinophilia (Fig. 6K). However strikingly, mice that received a CB2 antagonist developed less lung resistance (Fig. 6J, red line), associated with a decrease in BAL eosinophilia compared to controls (Fig. 6K). Conversely, we found that mice co-challenged with IL-33 and CB2 agonist remarkably developed higher lung resistance (Fig. 6J, green line), associated with an increase in BAL eosinophilia compared to controls (Fig. 6K). Similar to our previous findings, the numbers of ILC2s were further affected by CB2 agonistic or antagonistic treatments (Fig. 6L). Taken together, our results reveal that similar to our murine studies, CB2 is expressed on hILC2s, and that CB2 signaling promotes hILC2 function and development of hILC2-driven AHR and lung inflammation.

Fig. 6 |. CB2 is expressed on human ILC2s and contributes to ILC2-dependent AHR in humanized mice.

(A) Human blood ILC2s (hILC2s) were FACS-sorted from PBMCs and cultured for 72h with rhIL-2 and rhIL-7 with or without rhIL-33. (B) Flow cytometry plots of isolated hILC2s. (C) CB2 expression on hILC2s cultured with or without rhIL-33 and (D) corresponding quantitation showing CB2 MFI. (E-H) Isolated hILC2s were cultured for 72h with rhIL-2, rhIL-7, rhIL-33 with or without (E) CB2 antagonist (AM630, 10μg/mL) or (G) CB2 agonist (JWH133, 10μg/mL). Levels of IL-5, IL-13 and IL-6 in the culture supernatants following (F) CB2 antagonist and (H) CB2 agonist treatment. Histograms show 4 independent healthy blood donors. ns= non-significant, * p<0.05, ** p<0.01. (I) Isolated hILC2s were cultured for 72h with rhIL-2, rhIL-7, and rhIL-33 prior to adoptive transfer of 50,000 hILC2s in Rag2^−/−IL-2Rg^−/− mice. Host mice were then challenged intranasally for 3 days with rhIL-33 and either CB2 antagonist or agonist (both 1mg/kg) or the vehicle (control). On day 4, lung function, BAL eosinophils and lung hILC2s were measured. (J) Lung resistance in response to methacholine. Number of (K) BAL eosinophils and (L) hILC2s per lung. Histograms show mean +/− SEM, n=4. ns= non-significant, * p<0.05, ** p<0.01. Data representative of 2 independent experiments.

DISCUSSION

Activation of the cannabinoid system can modulate inflammatory responses in many peripheral tissues such as the liver, colon, brain and the lungs (51–54). Similar to cigarette smoking, it is now well known that marijuana smoking has significant effects in the airways. Notably, chemical agents such as the nicotine found in cigarettes can either activate or inhibit relevant immune functions, including in ILC2s (17, 55). However, it is important to note that activation of the cannabinoid system in the lungs by THC-containing marijuana can increase lung injury compared to cigarette smoking (56), as this observation was confirmed in multiple recent observational studies (39–43). This public health issue has led the CDC in 2019 to report an outbreak of lung injury associated with the use of E-cigarettes, and national data particularly showed that THC-containing products played a major role in the outbreak. In the context of diseases such as asthma where immune events driving pathology are multiple and diverse, it is becoming crucial to characterize cannabinoid receptor expression, signaling and function in relevant immune cells (57). To the best of our knowledge, our results are the first to show that cannabinoid signaling in lung ILC2s promotes airway inflammation in a CB2-dependent manner.

Both endocannabinoids AEA and 2-AG are detected in mouse lungs (58), as inflammation can modulate these levels in multiple organs including the colon and the brain (52, 53). Quantitative analysis of AEA and 2-AG and related lipids from biological matrices is technically challenging as it nearly always utilizes chromatographic separation, largely due to the complex nature of tissue matrices even after lipid extraction and purification. However, based on previously published studies, allergen challenge directly increases AEA levels more than fourfold in the BAL of patients with allergic asthma (35). In particular, the level of AEA was positively correlated with the level of BAL IL-5 and eosinophilia. To elucidate the source of AEA and 2-AG in the lungs, we have analyzed a single cell RNAseq data set recently published in Nature Medicine comparing healthy versus asthmatic lungs (59) (Fig. E6). Since endocannabinoids are lipids, we focused on critical enzymes catalyzing the formation of AEA and 2-DG. The in vivo biosynthesis of AEA has been shown to occur through several pathways mediated by N-acylphosphatidylethanolamide-phospholipase D (NAPE-PLD), and 2-AG through the action of selective enzymes including diacylglycerol lipase (DAGL) (60). Interestingly, the results of the analysis suggest that multiple cell types express both NAPE-PLD and DAGL, with fibroblasts, smooth muscle cells, and the endothelium inducing the most in the inflamed lungs. Whether allergens directly induce the secretion of endocannabinoids or act via cytokines such as IL-33 or IL-25 is an interesting question that warrants further studies.

CB2 is expressed on various immune cells in the lungs including eosinophils, monocytes, mast cells, macrophages, dendritic cells, T-cell and B-cells (61–63). Of note, these cells were further shown to express different levels of CB1, but interestingly neither CB1 nor CB2 were detected in neutrophils at the protein level (63, 64). To illustrate ILC2 responsiveness to cannabinoids, we showed that CB2 is expressed on both naïve and activated ILC2s, an observation previously made at the transcriptomic level on a publicly available data set (18). Recently, Baban et al. showed that kidney ILCs express CB2, and in particular the frequencies of ILC2s increased following cannabinol treatment in a model of acute kidney injury (65). Our transcriptomic analysis showed that activated ILC2s respond to cannabinoids in a strictly CB2-dependent manner, and while it confirmed Cnr2 expression by activated ILC2s, it remarkably showed that they do not express Cnr1 transcripts. To elucidate the role of CB2 in airway inflammation, we challenged WT and CB2 KO mice with IL-33 intranasally. Compared to WT mice, mice lacking CB2 showed an improved lung function and associated lower airway inflammation, suggesting that cannabinoid signaling promotes airway inflammation in response to IL-33. A similar role for CB2 signaling was observed in a house dust mite (HDM)-driven model of airway inflammation (38). However, since the immune mechanisms driving inflammation in response to HDM are complex and mainly T cell driven, this does not suggest a direct role of CB2 signaling in ILC2s. IL-33 causes the rapid expansion and activation of local ILC2s, together driving airway inflammation. In line with this, our results show that CB2 signaling in ILC2s sufficiently promotes airway inflammation independently of adaptative immunity, as evidenced in particular by our adoptive transfer experiments.

The contribution of cannabinoids in airway inflammation is clear, but their effects can be bidirectional depending on the severity of asthma (66). While it can induce bronchodilation in chronic asthmatics (67), several reports reveal that some asthmatics develop severe bronchoconstriction in response to THC (68, 69). This dichotomy may be explained in part by the different immune mechanisms driving airway inflammation, likely causing heterogenous responsiveness to cannabinoids. Although a number of studies demonstrate that CB2 signaling may be inhibitory in particular in T cells and macrophages (70, 71), studies have suggested that it is not general to all immune cells, highlighting the complexity of the cannabinoid system. In particular in the context of B cells, CB2 agonism enhanced proliferation and chemotaxis (72, 73), as CB2 KO mice showed a defect in B cell formation/maintenance in the spleen (74). Furthermore, CB2 KO mice showed an impaired T-independent humoral response following NP-Ficoll immunization (75). Our results clearly demonstrate that CB2 signaling stimulates pulmonary ILC2 expansion and function. While it did not affect ILC2 development and homeostasis at steady state, CB2 signaling promoted ILC2 proliferation and effector cytokine production in response to IL-33, IL-25 and A. alternata stimulation. Importantly, these effects were observed in both CB2 KO mice and in mice challenged with a CB2 antagonist, suggesting that the contribution of cannabinoids in ILC2s is dynamic and reversible. In line with this, CB2 agonism remarkably promoted the proliferation and function of activated ILC2s, leading to the exacerbation of ILC2-driven lung inflammation. It is important to note that while CB2 KO mice constitutively lack CB2 signaling, we carefully established the optimal effective dose for both CB2 agonist and antagonist based on previously published studies, in particular for the use in chronic a model. Interestingly however, whereas ILC2-derived IL-5 production was increased by the agonist, we surprisingly did not observe an upregulation of IL-13 in ILC2s. Since IL-33 is a strong stimulus inducing a robust ILC2 activation, it is plausible to believe that the IL-13 expression in ILC2s has reached a plateau, and the effect of the agonist is therefore undetectable in vivo, at least in respect to IL-13 expression. Furthermore, the effects of IL-5 and IL-13 are distinct and may act independently from on another (19), with in particular the CREB pathway recently shown to stimulate IL-5 secretion in ILC2s (76). The transcription factor GATA-3 plays a pivotal role in the regulation of ILC2 effector functions (77). Notably, while GATA-3 intranuclear expression was inhibited in CB2 KO mice and upon antagonist treatment, it was upregulated in mice that received agonist, confirming that CB2 signaling regulates ILC2 effector functions. Although ILC2s are sufficient to induce lung inflammation independently of adaptive immunity, it is worth pointing out that cannabinoids were shown to stimulate regulatory T cell (Tregs) (78). Given the involvement of ILC2s in Treg expansion and function, it remains to be elucidated whether ILC2s could contribute to these effects (3).

It is important to point out that we observed a similar CB2 expression pattern in ILC2s following IL-33, A. alternata and IL-25 stimulation. While ILC2s all express CB2 upon inflammation, it was not induced as compared to PBS-treatment, suggesting that ILC2s constitutively express CB2 and do not induce it regardless of the inflammation stimuli. However, although both naïve and activated ILC2s expressed CB2, we only saw an effect on ILC2 proliferation and effector function in the context of inflammation, suggesting that CB2 signaling only stimulates activated ILC2s. Interestingly this observation has been previously described in ILC2s in the context of TNFRSF stimulation. While GITR is expressed at similar levels on both naïve and IL-33-activated ILC2s, GITR agonism only stimulates activated but not naïve ILC2s (19). This effect contrasts with DR3 agonism which activated naïve ILC2s and enhanced activated ILC2 function (79). Toki et al. elegantly showed that TSLP increases ST2 expression and therefore synergistically activates ILC2s with the IL-33-ST2 signaling (80). Interestingly, we found similar synergistic effects of CB2 agonism on both GITR agonist and TSLP-treated ILC2s (Fig. E7). Although further studies are required to fully understand the mechanisms behind this observation, it nevertheless confirms the capacity of CB2 agonism to enhance ILC2 effector functions, regardless of the stimulus. Interestingly, we did not see an effect of CB2 signaling on ILC2 development in the bone marrow, but our observations do not rule out an effect of exogenous cannabinoids on ILC2 precursors. Whether prolonged exposure to exogenous cannabinoids affects ILC2 development in the bone marrow is an exciting topic in need of future studies to clearly define the role of exogenous CB2 signaling on ILC2 development.

One known downstream mediator of CB2 signaling is the CREB pathway (81). Interestingly in human primary leukocytes, CB2 agonism remarkably induced IL-6 and IL-10 production (49). Monitoring the downstream events leading to CREB phosphorylation, Saroz et al. showed that 85% of the response was driven by ERK1/2 and AKT phosphorylation as well as cAMP signaling in a dose-dependent manner, whereas p38 phosphorylation accounted for the remaining response. ERK1/2 and CREB phosphorylation were also shown to promote ILC2 effector functions. Whereas Lei et al showed that ICAM-1 controlled ILC2 effector functions by modulating ERK1/2 phosphorylation (21), Nagashima et al in particular showed that CREB phosphorylation enhanced IL-5 production by ILC2s (76). In addition, CREB expression was previously associated with the stimulation of proliferation and inhibition of apoptosis (82). Consistent with these observations, we found that the signaling elements downstream of CB2 included CREB phosphorylation, as we observed an inhibition of CREB phosphorylation in lung CB2 KO ILC2s. In particular in our transcriptomic analysis, we found that cAMP signaling was decreased and Mapk1 – encoding for ERK1/2 – was downregulated in CB2 KO ILC2s. Furthermore at the protein level, ERK1/2, AKT and p38 phosphorylation were all inhibited in CB2 KO ILC2s – with pERK1/2 representing the biggest difference among the three – together suggesting that the CREB pathway may regulate CB2 signaling in ILC2s. Interestingly, CREB phosphorylation is associated with asthma severity in humans, as the level of phosphorylation was directly linked to different asthma phenotypes (83). Targeting of CREB phosphorylation using selective inhibitors reduced cytokine secretion in ILC2s (76), as several studies showed in humans that airway remodeling in asthma is associated with downregulation of CREB signaling (84, 85). Together, our findings reveal that CB2 signaling via CREB may offer a novel therapeutic approach for the development of ILC2-dependent asthma.

Several lines of evidence suggest that cannabinoids such as THC can promote airway inflammation in humans (39–43, 56, 68, 69). Similar to murine ILC2s, we found that human blood ILC2s expressed CB2 whether they were stimulated with IL-33 or not. Furthermore, treatment of hILC2s in the presence of IL-33 with CB2 antagonist inhibited cytokine secretion, while the use of a CB2 agonist remarkably increased ILC2 function. Notably, human and bovine serum are rich in endocannabinoids, in particular containing bioactive 2-AG (86, 87), together suggesting that human blood ILC2s are in contact with endocannabinoids both in the human body as well as in culture. We further validated our observations using a previously described humanized mouse model, allowing us to study the function of human ILC2-related pathways in the context of asthma (20). We showed that humanized mice challenged with IL-33 and a CB2 antagonist developed less airway resistance and lung inflammation compared to controls. Conversely however, IL-33 and CB2 agonistic treatment of humanized mice led to airway resistance exacerbation and induction of lung inflammation. Of note, the numbers of pulmonary ILC2s were similarly affected, suggesting that CB2 signaling is involved in human ILC2 accumulation in the lungs. These findings highlight the importance of CB2 signaling in the development of human ILC2-dependant airway hyperreactivity and lung inflammation.

In summary, our data clearly indicate that CB2 is expressed by pulmonary ILC2s, and engagement of CB2 by cannabinoids enhances ILC2-dependent airway hyperreactivity and lung inflammation. Importantly, we used a combination of tools including a CB2 antagonist as well as CB2 KO mice, allowing us to focus on both the inducible and constitutive effects of CB2 signaling. We were able to show that cannabinoids stimulate ILC2s by enhancing their expansion and effector functions, together contributing to lung inflammation and airway hyperreactivity. Remarkably, we further demonstrated that engagement of CB2 by a CB2 agonist was able to exacerbate lung inflammation. It is important to point out that the modulation of CB2 signaling in naïve ILC2s had no significant effects, and in particular CB2 agonism enhanced ILC2 expansion and function of activated but not naïve ILC2s. Our transcriptomic analysis further showed that the stimulatory effects of CB2 engagement in ILC2s were at least in part mediated via the CREB pathway. Finally, we observed that human ILC2s express CB2 and clearly showed using humanized mice that CB2 engagement in human ILC2s enhances ILC2-dependent airway hyperreactivity. Such findings provide novel therapeutic avenues for the treatment of asthma, in particular in patients showing increased airway inflammation caused by substances able to engage CB2.

Abbreviations

AHR

Airway hyperreactivity

BAL

Bronchoalveolar lavage

GSA

Gene specific analysis

IL-

Interleukin-

ILC2

Group-2 innate lymphoid cell

KO

Knock-out

RNAseq

RNA-sequencing

WT

Wild-type

CLP

Common lymphoid progenitor

TSLP

Thymic stromal lymphopoietin

ECS

Endocannabinoid system

AEA

N-arachidonoyl-ethanolamine

2-AG

2-arachidnoyglycerol

GPCRs

G protein-coupled receptors

CB

Cannabinoid receptor

THC

Δ⁹-tetrahydrocannabinol

rm

recombinant mouse

rh

recombinant human

cAMP

cyclic AMP

CREB

cAMP response element binding protein

H&E

hematoxylin and eosin

Treg:

Regulatory T-cell

Data and materials availability:

Data related to Fig. 4 RNAseq will be uploaded on GEO and accession number will be provided.