Interleukin‐33 (IL‐33): A nuclear cytokine from the IL‐1 family

Interleukin‐33 (IL‐33) is a tissue‐derived nuclear cytokine from the IL‐1 family abundantly expressed in endothelial cells, epithelial cells and fibroblast‐like cells, both during homeostasis and inflammation. It functions as an alarm signal (alarmin) released upon cell injury or tissue damage to alert immune cells expressing the ST2 receptor (IL‐1RL1). The major targets of IL‐33 in vivo are tissue‐resident immune cells such as mast cells, group 2 innate lymphoid cells (ILC2s) and regulatory T cells (Tregs). Other cellular targets include T helper 2 (Th2) cells, eosinophils, basophils, dendritic cells, Th1 cells, CD8+ T cells, NK cells, iNKT cells, B cells, neutrophils and macrophages. IL‐33 is thus emerging as a crucial immune modulator with pleiotropic activities in type‐2, type‐1 and regulatory immune responses, and important roles in allergic, fibrotic, infectious, and chronic inflammatory diseases. The critical function of IL‐33/ST2 signaling in allergic inflammation is illustrated by the fact that IL33 and IL1RL1 are among the most highly replicated susceptibility loci for asthma. In this review, we highlight 15 years of discoveries on IL‐33 protein, including its molecular characteristics, nuclear localization, bioactive forms, cellular sources, mechanisms of release and regulation by proteases. Importantly, we emphasize data that have been validated using IL‐33‐deficient cells.

Recently, it has been discovered that IL-33 plays important roles in regulatory immune responses. 2,3,22 Highly suppressive ST2 + FoxP3 + Tregs subsets regulated by IL-33 were identified in several tissues, including lymphoid organs, 23,24 gut, 25 lung, 26 and adipose tissues. [27][28][29] These recent observations with tissue Tregs, 26 together with previous studies on ILC2s, 30 suggest important roles of IL-33 in the maintenance of tissue homeostasis, resolution of inflammation, and repair of tissue damage. 2,3 There are comprehensive reviews about the role of IL-33 in various diseases and activation of immune cells, 2,3 to which the reader is referred. In this review, we rather focus on IL-33 biology and we summarize the important advances that have been made over the past 15 years.
The reader will found specific answers to important questions such as: • How IL-33 was discovered?
• What are its molecular characteristics and bioactive forms?
• What is its subcellular localization and what are its cellular sources in vivo?
• How is it released from producing cells and how its activity is regulated?
Specificity of reagents has turned out to be an important issue in the IL-33 field. Therefore, whenever possible, we emphasize data that have been validated using IL-33 knockout (mouse) or knockdown (human) cells.

| THE DISCOVERY OF IL-33
We visualized endogenous IL-33 protein and mRNA for the first time in human tissues in 2003. 31 We initially designated the protein 'nuclear factor from high endothelial venules (NF-HEV)', 31 because we discovered it as a nuclear protein abundantly expressed in HEVs, specialized blood vessels that mediate the entry of lymphocytes into lymph nodes and other lymphoid organs. 32, 33 We detected high levels of IL-33/NF-HEV mRNA in HEVs from diverse human lymphoid organs, including tonsils, Peyer's patches and lymph nodes by in situ hybridization. 31 We also reported the location of the human IL-33 gene on the short arm of chromosome 9 at 9p24.1 and the identification of the murine IL-33/NF-HEV orthologue. 31 Although the primary sequence of NF-HEV did not exhibit significant homology to known proteins, we found a strong similarity with a predicted canine protein encoded by an unknown mRNA (DVS27), identified by Onda et al 34  proposed the name of interleukin-33 (IL-33) for this novel member of the IL-1 family. Schmitz et al did not mention that the protein was not novel and that it was in fact 100% identical to the human NF-HEV protein sequence identified 2 years earlier. 31 One year later, we reported the identity between IL-33 and NF-HEV, and demonstrated that IL-33 is a chromatin-associated nuclear factor in vivo. 36

| IL-33 gene
The gene encoding IL-33 ( Figure 2A) comprises eight exons both in humans and mice. 3 The promoter for the major human IL-33 mRNA form (NM_033439; 2718 nucleotides) is located upstream of an untranslated exon (exon 1 or 1a), more than 20 kb upstream of exon 2 that contains the AUG initiation codon. 3,37 It is active in normal human keratinocytes, endothelial cells and fibroblasts, and it contains an interferon stimulated response element (ISRE) and several gamma F I G U R E 1 The discovery of IL-33. The important dates and events are indicated above or below the timeline interferon activation sites (GAS). 37 An alternative exon 1 sequence (exon 1b), located 4.6 kb upstream of exon 2, corresponding to the use of an alternative promoter has been described both in humans and mice. 37,38 Transcription start sites are located downstream of consensus TATA box sequences in both promoters. 38 Interestingly, many of the single nucleotide polymorphisms (SNPs) in the human IL-33 gene associated with asthma are located in the promoter and intron 1. 11 They could thus influence the activity of the two promoters of the IL-33 gene. SNPs in IL-33 are generally associated with increased asthma susceptibility. 1,[11][12][13][14][15][16][17] However, a rare loss of function mutation in IL-33 that disrupts a canonical splice acceptor site before exon 8 and is predicted to result in the production of a truncated IL-33 protein, has been found to cause reduced numbers of eosinophils in blood and to protect against asthma. 39 Eosinophils counts are also reduced in IL-33-deficient mice. 39 Splice variants lacking exons 3, 4 and/or 5 have been described. [40][41][42] However, they have only been detected after amplification by qPCR and they appear to be very minor components in primary cells. At this point, it is therefore unclear whether alternative splicing is a relevant mechanism of IL-33 regulation.

| IL-33 protein
IL-33 protein is composed of two evolutionary conserved domains, the N-terminal nuclear domain and the C-terminal IL-1-like cytokine domain, separated by a divergent central part ( Figure 2B). Human IL-33 exhibits 58% and 52% identity over 270 residues ( Figure 2C) with its canine and murine orthologues, respectively. 31

| N-terminal nuclear domain
The N-terminal part of IL-33 contains a chromatin-binding motif (CBM, aa 40-58) 43 and a predicted bipartite nuclear localization sequence (NLS, aa 61-78). 31 The N-terminal domain (aa 1-65), that contains the IL-33 CBM, has been shown to be necessary and sufficient for nuclear targeting of epitope-tagged (GFP-fused) IL-33 in human cells, 36 but the possibility that the bipartite NLS may be required for nuclear targeting of untagged IL-33 has not been excluded. The Nterminal part of IL-33 containing the CBM and NLS (aa 1-112) has been shown to be necessary and sufficient for targeting of a dsRed reporter to the nuclei of producing cells in vivo in a genetically engineered mouse model. 44,45 Based on structural predictions, the Nterminal domain of IL-33 was initially believed to contain a putative homeodomain-like helix-turn-helix DNA-binding domain. 31 However, F I G U R E 2 The human IL-33 gene and protein. (A) Structure of the human IL-33 gene. The gene spans >42 kb, contains 8 exons, and is located on chromosome 9 at 9p24.1. Exon 1 (also designated exon 1a) is located 25.8 kb upstream of exon 2. An alternative exon 1b, located 4.6 kb upstream of exon 2 has also been described. Most single nucleotide polymorphisms (SNPs) associated with asthma are found in the promoter and intron 1 regions. (B) Primary structure of the human IL-33 protein. The different protein domains are indicated and the cleavage sites for proteases involved in IL-33 activation (inflammatory proteases) or inactivation (caspases) are shown above the primary structure. (C) Sequence alignment of human (h), canine (c) and murine (m) IL-33 proteins. The IL-33 CBM (bold line), putative NLS (dashed line), and cleavage sites for inflammatory proteases (black arrows) and caspases (red arrow) are indicated. Sequence alignment was performed with CLUSTALW. Black boxes: identical residues; shaded boxes: similar residues this was not confirmed in later studies which revealed that IL-33 associates with chromatin through CBM-mediated protein-protein interactions rather than through protein-DNA interactions. 43 Characterization of the IL-33 CBM uncovered a unique example of molecular mimicry between a chromatin-associated cytokine and a DNA tumor virus. 43 Indeed, the IL-33 CBM shares striking structural similarities with the CBM of Kaposi sarcoma herpes virus LANA (latency-associated nuclear antigen), which mediates attachment of viral genomes to chromosomes of infected host cells. 46 Similar to LANA CBM, IL-33 CBM adopts a tight hairpin structure stabilized by an array of intramolecular hydrogen bonds, and uses a conserved MXLRSG hexapeptide to bind to the acidic pocket formed by histones H2A and H2B at the surface of chromatin. 43 Strikingly, a single-point mutation of IL-33 Arg48 (or mutations of H2A acidic patch residues) has been shown to abrogate the interaction between IL-33 and the H2A-H2B heterodimer. 43

| C-terminal IL-1-like cytokine domain
The three-dimensional structure of the IL-1-like cytokine domain of human IL-33 has been solved both by nuclear magnetic resonance 47 and X-ray crystallography. 48 The domain adopts a β-trefoil fold similar to that described for IL-1α, IL-1β and IL-18. 47,48 The 12 β strands of the β-trefoil (β1-β12) are arrayed in three pseudo-repeats of four β strand units. 47 Crystal structure of the IL-33/ST2 complex ( Figure 3A) revealed that IL-33 interacts with all three Ig-like domains of the ST2 ectodomain through a large binding interface composed of two distinct binding sites. 48 The surface of IL-33 at site 1 exhibits strong negative charge, and the corresponding surface of ST2 is dominated by positive charge. 48 IL-33 acidic residues Glu144, Glu148, Asp149 and Asp244 make critical salt-bridge interactions with ST2 basic residues Arg38, Lys22, Arg198 and Arg35. At site 2, IL-33 acidic residue Glu165 has salt-bridge interaction with Arg313 of ST2, and a significant hydrophobic cluster is formed by Tyr163 and Leu182 of IL-33 and Leu246, Leu306 and Leu311 of ST2. IL-33 binds to ST2 with an affinity of 0.45-0.74 nM. 47,48 Single-point mutations of IL-33 acidic residues at sites 1 and 2 strongly decreased the ST2 binding affinity, indicating that charge complementarity plays a critical role in specific recognition of IL-33 by ST2. 48

| ACTIVATION OF ST2-EXPRESSING CELLS
Binding of IL-33 allows membrane ST2 (also designated ST2L) to interact with IL-1RAcP, a co-receptor shared with other IL-1 family members (IL1α, IL1β, IL-36). The IL-33/ST2/IL1RAcP complex then induces signaling through MyD88 adaptor, IRAK1 and IRAK4 kinases, and TRAF6, that culminates in the activation of MAP kinases and NFκB transcription factors. 3,35 The signaling pathways activated by IL-33 in target cells are very similar to those activated by IL-1 and IL-18. Differential expression of ST2, IL-1R, and IL-18R on target cells is thus likely to be critical to explain the unique biological effects of IL-33. 2,3 Tissue-resident immune cells that express ST2 constitutively are major targets of IL-33. 1-3 These include mast cells, 49-51 group 2 innate lymphoid cells (ILC2s) 4-7 and tissue regulatory T cells (Tregs) [23][24][25][26][27][28][29] ( Figure 3B). Many (but not all) mast cells, ILC2s and tissue Tregs express high levels of ST2, and convincing evidence that IL-33 is a critical regulator of these tissue-resident immune cells has been provided in many studies. [1][2][3] Modulation of ST2 expression on immune cells (constitutive vs inducible) in different conditions could explain why IL-33 has been reported to play crucial roles in both type-2 and type-1 immune responses and associated diseases. For instance, in a murine model of COPD, ST2 was shown to be downregulated on ILC2s, and induced on NK cells and macrophages that don't express it at steady state. 21 Induction of ST2 expression has also been observed on Th1 cells and CD8 + T cells after activation. 20,22 Although they are less well characterized, non-immune cells may also constitute important cellular targets of IL-33 in vivo. These  48 The two critical binding sites are indicated. (B) IL-33 activates various ST2 + immune cell types implicated in type-2, type-1 and regulatory immune responses. Tissue-resident mast cells, ILC2s and Tregs, that express high levels of ST2 constitutively, are among the major targets of IL-33 in vivo include endothelial cells, epithelial cells, fibroblasts, astrocytes, and neurons. 52-56

| Nuclear localization
We were surprised in 2005 when Schmitz et al 35 reported that the carboxy-terminal part of nuclear factor NF-HEV corresponds to a cytokine ligand for a membrane receptor. Therefore, we confirmed the strict nuclear localization of IL-33/NF-HEV in vivo 31 using three distinct antisera directed against the N-terminal nuclear and Cterminal IL-1-like cytokine domains. 36 We stained more than 50 distinct human tissues and found that endogenous IL-33 accumulates in the nuclei of endothelial cells in almost all tissues analyzed. 57 We also observed strict nuclear localization of IL-33 in other producing cells, such as epithelial cells and fibroblasts. 57 We made similar observations in mice. 58 For instance, we found strong nuclear expression of IL-33 in epithelial cells from barrier tissues and fibroblastic reticular cells (FRCs) from lymphoid organs. 58 Importantly, we validated the specificity of IL-33 staining using IL-33 −/− mice. 58 Nuclear localization of IL-33 in vivo has been confirmed by many other groups and it was validated in most studies using IL-33deficient cells or mice. 29,53,[59][60][61][62] In contrast, although cytoplasmic localization of IL-33 has been reported in some studies, it has not been validated using IL-33-deficient cells. Validation is essential because we have observed that even the best IL-33 antibodies available can give specific nuclear staining of IL-33 producing cells (staining lost in IL-33 −/− mice) and non-specific cytoplasmic staining of other cells (staining still observed in IL-33 −/− mice, despite absence of IL-33). 58 We believe that, in the absence of validation

| Role of IL-33 nuclear localization
IL-33 accumulates in the nucleus of producing cells, 31,36,57 binds to histones and chromatin. 36,43 In addition, evolutionary conservation of the N-terminal nuclear domain ( Figure 2C) supports a critical role of nuclear localization and chromatin-association of IL-33. Initial studies indicated that IL-33 exhibits some transcriptional repressor properties when overexpressed in mammalian cells. 36,65 We thus proposed that IL-33 could be a dual function protein, acting intracellularly as a nuclear factor regulating transcription, and extracellularly as a potent cytokine. 36 However, after more than 10 years of research efforts, convincing evidence that endogenous nuclear IL-33 regulates gene or protein expression is still lacking. Using a global proteomic approach, we recently showed that extracellular IL-33 cytokine, and not endogenous nuclear IL-33, regulates protein expression in primary human endothelial cells. 55 High-resolution mass spectrometry analyses revealed that knockdown of endogenous nuclear IL-33 using two independent RNA silencing strategies had no reproducible effect on the endothelial cell proteome. 55 A single protein was modulated on a global scale and it was IL-33 itself. Although it will be important to confirm these results in other cell types producing IL-33, particularly in mucosal epithelial cells, these proteome-wide analyses do not support the previous view of IL-33 as a dual function protein. 55 Nuclear localization and chromatin association of IL-33 may thus have been selected during evolution for purposes different from regulation of gene expression. As discussed below, an essential role of nuclear localization appears to be the regulation of IL-33 cytokine activity though nuclear sequestration. 44

| IL-33 full length precursor (IL-33 FL )
Schmitz et al 35 initially proposed that processing of IL-33 full length (IL-33 FL ) precursor was required for biological activity. During several years, IL-33 was thus believed to be activated by caspase 1 and inflammasomes, 35 similar to IL-1β and IL-18. 66 However, in 2009, we 67 and others 68,69 discovered that maturation of IL-33 is not essential for activation of ST2, and that IL-33 FL is a bona fide bioactive form of IL-33 that can induce ST2-dependent NFκB activity and cytokine production in target cells. Since shorter forms of IL-33 were observed during the bioassays used in some of these studies, 69 it was questioned whether IL-33 FL itself represented the bioactive species. 69 To address this issue, we performed additional experiments and demonstrated that recombinant IL-33 FL precursor produced in three different systems had biological activity and, importantly, that it was still in a full length form at the end of the bioassays. 70 These later data provided definitive evidence that IL-33 FL precursor is a bioactive form of the cytokine. Thus, unlike IL-1β and IL-18, and similar to IL-1α, 66 IL-33 has biological activity as a full length molecule.

| IL-33 mature forms
The observation that IL-33 FL precursor is a bioactive form of the protein did not exclude the possibility that shorter mature forms encompassing the IL-1-like cytokine domain may also be produced during inflammation. Indeed, using a combination of in vitro and in vivo approaches, we demonstrated that inflammatory serine proteases from neutrophils 70 and mast cells 71 can cleave IL-33 FL precursor to release shorter mature forms of 18-21 kDa. The sites of cleavage of IL-33 FL by inflammatory proteases were precisely identified by mass spectrometry, within the central part of IL-33, upstream of the IL-1-like cytokine domain. 70,71 Six distinct inflammatory serine proteases were shown to cleave human IL-33 FL precursor, including neutrophil cathepsin G, elastase and proteinase 3 (PR3), as well as mast cell chymase, tryptase and granzyme B. Human IL-33  and IL-33  are likely to be major mature forms of IL-33 in vivo because they can be generated by both activated neutrophils and IgE-activated mast cells. In mouse, mature forms of IL-33 of 19-20 kDa have been detected in BAL fluids after acute lung injury 70 or exposure to fungal aeroallergen Alternaria alternata, 64,72 and in the lungs in response to chitin or migratory helminths. 73 Interestingly, we observed that IL-33 mature forms generated by inflammatory proteases had significantly increased biological activity compared to IL-33 FL precursor. 70,71 Indeed, at least 10 to 30 fold higher concentrations of IL-33 FL were required to obtain similar levels of IL-6 secretion by MC/9 mast cells, and IL-5 or IL-13 secretion by ILC2s, respectively. 70,71 Processing of IL-33 FL into the highly active mature forms could play a crucial role in vivo, particularly when IL-33 FL is present at low concentrations and has no or very little activity. The observation that multiple proteases are able to cleave and activate the IL-33 FL precursor 71 suggests that proteolytic maturation is an important step in the regulation of IL-33 bioactivity. Neutrophil proteases could be critical for IL-33 activation during infection and type-1 inflammatory responses 70 and mast cell proteases during allergic type-2 inflammation. 71

| CELLULAR SOURCES OF IL-33
Analyses of the cellular sources of IL-33 in human and mouse tissues have revealed that IL-33 is a tissue-derived nuclear cytokine principally produced by endothelial cells (Figure 4A and B), epithelial cells ( Figure 4B-E), fibroblast-like cells and myofibroblasts, rather than CD45 + hematopoietic cells, both during steady state and inflammation. 36,57,58,74

| Endothelial cells
The endothelium is likely to constitute the major cellular source of IL-33 in the human body during steady state. 36,57,74 We observed constitutive and abundant expression of IL-33 along the vascular tree (>50 tissues analyzed), in the nuclei of endothelial cells from large and small blood vessels of healthy tissues. 57 Primary human endothelial cells in culture, particularly quiescent endothelial cells, were found to express nuclear IL-33, and specificity of the nuclear staining and identity of IL-33 FL band were validated using knockdown cells. 55,67,74,75 Notch signaling was shown to play a key role in the induction of nuclear IL-33 in quiescent endothelial cells. 75 Endothelial cells are also major sources of IL-33 during human disease. For instance, IL-33 was abundantly expressed in blood vessels from chronically inflamed tissues of patients with rheumatoid arthritis and Crohn's disease, 36  ( Figure 4A).
Although not constitutive, IL-33 expression in murine endothelial cells can be induced during inflammation. For instance, we observed strong upregulation of nuclear IL-33 in endothelial cells of the colon during colitis. 58,78 Others have shown that IL-33 is induced in cardiac endothelial cells after myocardial pressure overload, and that endothelium-derived IL-33 induces systemic inflammation, and plays a critical role in the heart's response to pressure overload. 59

| Epithelial cells
Although IL-33 expression was restricted to endothelial cells in many human tissues during homeostasis, we also observed high levels of nuclear IL-33 in the nuclei of epithelial cells in barrier tissues that are exposed to the environment. 57 Indeed, mucosal epithelial cells constitute major sources of IL-33 in the respiratory and digestive tracts ( Figure 4B-D), the skin ( Figure 4E) and the female reproductive system, both in humans and mice. 57,58,60,79,80 In mouse, we observed a very good correlation between expression of endogenous IL-33 protein in the nuclei of barrier epithelial cells and activity of the endogenous IL-33 promoter, visualized using an IL-33-LacZ reporter strain. 58 Nuclear IL-33 staining was specific because it was not observed in IL-33-deficient mice.
The nasal epithelium constitutes an important source of IL-33 in the respiratory tract, both in humans ( Figure 4D) and mice, with critical implications for allergic rhinitis and allergic airway inflammation. 60,81 In the lungs, the major sources of epithelial IL-33 may differ according to species. In humans, bronchial epithelial cells appear to be the major producers ( Figure 4B), particularly in patients with severe asthma 82,83 or COPD. 21,84 In COPD, IL-33 expression was traceable to a subset of airway basal cells with progenitor characteristics. 84 IL-33 is generally not expressed in murine airway epithelium but it can be induced after viral infection 84 or exposure to cigarette smoke. 21 In the murine lung, alveolar type II (ATII) pneumocytes expressing prosurfactant protein C constitute the major cellular source of IL-33 during homeostasis. 21 in humans 57 and mice. 58 In the skin, IL-33 expression is constitutive in murine keratinocytes but inducible in human keratinocytes, 80 particularly during atopic dermatitis 88 where it may be regulated by p53 family member p63 (specifically ΔNp63). 89 In our studies, we consistently observed nuclear IL-33 staining in human skin keratinocytes and epithelial cells of the stomach, but we noticed significant variability between different cells, different part of the tissues and different individuals, suggesting local regulation by environmental cues. 57 IFNγ appears to be a critical inducer of IL-33 in human skin keratinocytes. 80,90 Inducible and heterogeneous expression of IL-33 has also been seen in epithelial cells of the colon and the intestine, during colitis and intestinal polyposis, respectively. 52,78,91

| Fibroblast-like cells and myofibroblasts
We discovered that IL-33 is abundantly expressed in the nuclei of

| Other cell types
In addition to epithelial barrier tissues and lymphoid organs, we detected abundant expression of IL-33 in the brain and eye of wild-type mice during steady state. 58 Strong activity of the IL-33 promoter was observed in the corpus callosum, hippocampus (dentate gyrus), thalamus and the cerebellum (granular layer and white matter). 58 In the healthy brain, IL-33 is produced by glial cells and astrocytes, but not by microglia and neurons. 53,103,104 Post-mitotic OLIG2+ oligodendrocytes are the major sources of nuclear IL-33, and they rapidly release IL-33 following brain injury to promote recovery. 53 Strikingly, about one-third of isolated murine brain cells are IL-33 + . 53 It is not yet known whether IL-33 is expressed at such high levels in the human brain. In the eye, in addition to the optic nerve, we identified Sox2 + Muller glial cells from the retinal inner nuclear layer as major producers of IL-33. 58 Strong nuclear expression of IL-33 in Sox2 + Muller glial cells was also observed by others in the human eye, 45 and IL-33 was shown to mediate increased inflammatory responses following retinal injury in mouse. 45 In addition to Muller glial cells, nuclear IL-33 is also produced by epithelial cells of the ciliary body 58 and the conjunctiva 105 in the murine eye.
We have observed nuclear expression of IL-33 in visceral SMCs of the gastrointestinal and urogenital tracts in healthy human tissues but it was significantly lower than in endothelial cells or epithelial cells. 57 In addition, vascular SMCs did not express IL-33 in vivo during steady state, 57 despite the fact IL-33 mRNA was previously detected in vascular SMCs in culture. 35 Airways SMCs have been shown to express nuclear IL-33 in severe asthma, but again expression levels were low, compared to epithelial cells and endothelial cells in the same lung tissue. 83 Although IL-33 is not expressed in liver hepatocytes during homeostasis, high levels have been detected in the nuclei of murine hepatocytes during acute hepatitis. 76,106 CD45 + hematopoietic cells, such as macrophages and mast cells, have been claimed to represent major cellular sources of IL-33 in several reports. [107][108][109][110][111][112][113] However, convincing evidence to support these claims has not yet been provided. Bona fide cellular sources of IL-33 Although the mechanisms of IL-33 release in vivo remain incompletely defined, cellular damage and tissue injury appear to be critical. Indeed, in a murine model of acute lung injury associated with alveolar epithelium damage and neutrophil accumulation in the alveolar wall, we showed that IL-33 FL and cleaved IL-33 forms were released in BAL fluids 2 hours after the injury. 70 Viral infection can also induce lung epithelium damage, and infection with influenza virus, rhinovirus or respiratory syncytial virus has been shown to trigger IL-33 release from primary bronchial epithelial cells. 21 In subjects with asthma, rhinovirus infection resulted in increased levels of IL-33 in nasal fluids. 117 In mice, specific loss of IL-33 + epithelial cells in infected lung areas was observed following influenza virus-induced cell damage in vivo. 21 Disappearance of IL-33 producing cells (oligodendrocytes) from the lesion site have also been observed in the brain after spinal cord physical injury. 53 This was associated with rapid release of endogenous IL-33 in cerebrospinal fluid 1 hours postinjury 53 followed by activation of meningeal ILC2s. 118 IL-33 release was very transient since the cytokine was not detected at 3 hours or at 24 hours postinjury. 53 Transient release of endogenous IL-33 has also been reported after mechanical skin injury. 119 Serum IL-33 levels increased significantly 1 hour after injury (tape stripping) and returned to normal by 6 hours and at 24 hours. 119 Loss of nuclear IL-33 staining in keratinocytes was observed 6 hours after hapten-induced immediate-type contact hypersensitivity, 120 skin wounding (ear punch model), or intradermal injection of live Staphyloccus aureus. 80 Although the epidermis became overtly necrotic 24 hours after S. aureus injection, there were no signs of epidermal necrosis at 6 hours supporting the possibility of prenecrotic secretion of IL-33. 80 In primary cultures of human endothelial cells, we found that IL-33 FL was released extracellularly after mechanical stress induced by cell scraping, a process that mimics the transient sub-lethal membrane disruptions that are observed in cells subjected to mechanical forces in vivo. 67 Therefore, in certain circumstances, IL-33 could be released after a 'near cell death experience' followed by cell recovery or death at later time points.
In one study, IL-33 has been proposed to be secreted from living cells upon biomechanical stress in the absence of cell death. 63 However, this was essentially demonstrated using an NIH3T3 fibroblast cell line stably transfected with a lentiviral construct coding for an exogenous form of human IL-33 fused to a tetracysteine tag, and under the control of the strong EF1 promoter. 63 Transfection, exogenous overexpression in a cell line, and epitope-tagging may have modified the normal behavior of the protein, particularly the tetracysteine tag, since cysteine residues are critical regulators of IL-33 folding and activity. 64 Therefore, although mechanical stress is likely to be an important inducer of IL-33 release in vivo, 59,67,121 definitive evidence that this can occur in the absence of cell death or cell damage remains to be provided.
Other agents that cause cellular damage or necrotic cell death were found to induce rapid release of IL-33 in extracellular fluids.
For instance, bee venom phospholipase A2, a major allergen that is known to cause membrane damage, 122 and alum, a potent adjuvant that induces cellular necrosis, 123 were found to induce IL-33 release in peritoneal fluids 30 minutes after intraperitoneal injection. 122,123 In addition, IL-33 has been shown to be released in plasma after necroptosis, a specialized pathway of programmed necrosis. 124 Fungal allergen A. alternata has been proposed to induce IL-33 release in the absence of cell death, via cellular stress, through extracellular secretion of ATP, increases in intracellular calcium concentration, and generation of reactive oxygen species (ROS). 72,125,126 A. alternata is a potent inducer of IL-33 release in vivo. 64,125,127,128 Indeed, a single intranasal exposure of naïve mice to A. alternata extract resulted in a very rapid increase in IL-33 levels (peak level between 15 minutes and 1 hours after exposure) in BAL fluids, 64,125,127,128 that correlated with a significant reduction in IL-33 + ATII cells at 1 hours. 127  that can be cleaved extracellularly by proteases from inflammatory cells such as neutrophils or mast cells. IL-33 FL is biologically active, however, its cleavage results in the generation of highly active mature forms that exhibit 10-to 30-fold higher activity glutathione 72 has been shown to attenuate IL-33 release in BAL fluids after intranasal exposure to A. alternata. Protease allergens, such as cysteine protease papain, have also been found to induce IL-33 release in BAL fluids shortly (1-3 hours) after intranasal exposure. 130 Papain, which is known to cause lung tissue injury when used at high dose, 130 was proposed to act through release of endogenous uric acid into the airway lumen. 131 The capacity of A. alternata to generate oxidative stress 72,126 is a property shared with other allergens, particularly plant pollens. 132 In humans, IL-33 has been shown to be released very rapidly (10 minutes) in BAL fluids from asthma patients after segmental lung challenge with plant pollens and house dust mite. 133 In mice, nasal challenge with ragweed pollen has been shown to result in a prompt (1-4 hours) increase in IL-33 levels in nasal fluids. 60 IL-33 release was associated with a decrease in IL-33 + nasal epithelial cells in the nasal mucosa 1 hours after challenge. 60 In addition, nuclear IL-33 staining in the nasal epithelium was considerably lower in mice after continuous challenge with ragweed pollen 60 or house dust mite, 81 and in human patients with allergic rhinitis, 60 suggesting that exposure to allergens reduced IL-33 expression by causing epithelial cells to secrete IL-33.
It has been suggested in several studies that IL-33-producing primary cells release IL-33 extracellularly after exposure to extracellular ATP, 84,103,125,126,134 uric acid 131 or thapsigargin (an inducer of intracellular calcium). 125 However, the specificity of IL-33 detection was not validated with IL-33-deficient cells in most cases, and the possibility that IL-33 release may have occurred through cell death was not excluded in some of these studies. Indeed, it is difficult to exclude some level of cell death/necrosis in cell culture experiments.
Moreover, the capacity of ATP or thapsigargin to induce cytoplasmic translocation and extracellular release of IL-33 has not been confirmed in other studies. 80 Therefore, additional studies are needed to firmly establish the capacity of extracellular ATP and other agents to induce IL-33 release through cellular stress, in the absence of cell death. One possibility is that cellular and/or mechanical stress may induce a low level of IL-33 release, and that high levels of IL-33 release may only occur upon cell death and/or extensive cell or tissue damage.

| REGULATION OF IL-33 ACTIVITY
IL-33 is a very potent cytokine constitutively expressed in healthy tissues. Therefore, its activity needs to be tightly regulated at multiple levels. Several distinct mechanisms appear to be involved ( Figure 6): nuclear localization or sequestration, regulation by proteases, inactivation by oxidation of cysteine residues, or sequestration by soluble decoy receptor sST2.

| Nuclear sequestration
A major mechanism of regulation appears to be through nuclear lo- (DGVDG) (Figure 2). This cleavage site, which is located in a flexible loop between β4 and β5 strands in the middle of the IL-1 cytokine domain, is not found in other IL-1 family members. This suggests specific regulation of IL-33 during apoptosis, probably related to the high levels of constitutive expression of the cytokine in healthy tissues.
Recombinant caspase-3 and caspase-7, another caspase activated during apoptosis, processed IL-33 at the DGVDG site in vitro. 67,68 Recombinant caspase-1 was also able to induce the cleavage of IL-33 after Asp178 when added to cell extracts, 67 although the cleavage may be indirect and mediated by caspase-7. 137 Interestingly, caspase-7, which is a substrate for caspase-1, 138 translocates to the nucleus following activation by caspase-1. 139 It could thus play an important role in processing and inactivation of IL-33 in the nucleus. Activation of caspase-1 and caspase-7 has been proposed to be involved in resolution of type-2 allergic inflammation, through the inactivation of IL-33. 137,140 Inflammatory proteases are other critical regulators of IL-33 bioactivity. As discussed above, they process IL-33 FL in the central domain, upstream of the IL-1 cytokine domain, and generate mature forms with increased biological activity. 70,71 Calpain has also been reported to cleave IL-33 141 but since the processing site has not been mapped, it is not known whether this cleavage results in IL-33 activation or inactivation.
Inflammatory proteases, when present in large amounts, could also be involved in IL-33 inactivation, through protein degradation. Indeed, mast cell chymase has been shown to play a role in IL-33 degradation in mouse, 142 although this was not observed in human, 71 suggesting the existence of species differences. Proteinase 3, another protease of neutrophils, has a dual role in IL-33 activation and inactivation. 143 It has been shown in several models that IL-33 accumulates during a couple of hours after release and that it is generally no longer detectable after 6 hours. 53,60,64,70,72,119,122,125,130 IL-33 degradation by inflammatory or other proteases is thus likely to be one of the important mechanisms that limit IL-33 activity in vivo.

| Inactivation by oxidation
Biological activity of IL-33 in the extracellular microenvironment has been found to be rapidly terminated by the formation of two disulfide bridges. 64 It was shown that IL-33 is oxidized on critical cysteine residues shortly after extracellular release in a reduced form. 64 Cysteine oxidation then drives a conformational switch to a biologically inactive disulfide-bonded form of IL-33. Bioactive reduced IL-33 was no longer detectable in BAL fluids 2 hours after exposure to A. alternata. 64 The four critical cysteine residues were identified as Cys208, Cys227, Cys232, and Cys259, and mutation of these four cysteines resulted in prolonged activity of IL-33 in the lung microenvironment. 64 Oxidation of cysteines, shortly after extracellular release, is likely to be an important mechanism of IL-33 inactivation in vivo. It could be even more rapid than protein degradation. In any case, this means that the first few hours or even the first few minutes after IL-33 release are likely to be critical for activation of target cells.

| Sequestration by soluble ST2 receptor
A third mechanism that may limit IL-33 activity after extracellular release is its possible sequestration by soluble (decoy) receptor sST2. 144 The mRNA encoding sST2 is generated by alternative splicing and lacks the sequence encoding the transmembrane domain of full length ST2 (ST2L). sST2 is produced by various cell types, including mast cells and Th2 cells. 145,146 High amounts of sST2 are present in the serum of patients with various inflammatory diseases, and often correlate with disease severity. 144,145 Soluble sST2 can neutralize IL-33 in biological fluids 121,144,147 and high levels of sST2 may thus significantly inhibit IL-33 activity in vivo.

| CONCLUDING REMARKS
More than 2000 publications have been published on IL-33 since the first visualization of the protein in human tissues 15 years ago. 31  terized. In addition, very little is known about the bioactive forms of IL-33 in vivo, particularly in human tissues and diseases, and how they access target cells (local effects or diffusion). The relative importance of the different cellular sources of IL-33 will need to be defined, for instance, using tissue-or cell-specific knock-out approaches. It will be also crucial to determine how the observations made in mouse models are applicable to humans, because important species differences have already been identified. It is likely that new questions will arise as we get closer to fully understanding the modes of action, mechanisms of regulation and functions of IL-33 in health and disease.