Two Novel IL-1 Family Members, IL-1δ and IL-1ε, Function as an Antagonist and Agonist of NF-κB Activation Through the Orphan IL-1 Receptor-Related Protein 2¹

Interleukin 1 family members are known to alter the host response to an inflammatory, infectious, or immunological challenge (1). The biological activity of IL-1 is tightly controlled under physiological conditions. The classical IL-1 family comprises several ligands (i.e., IL-1α, IL-1β, and IL-1 receptor antagonist (IL-1ra)³ (2, 3, 4), and surface and soluble IL-1 receptors (IL-1RI, IL-1RII, and IL-1R accessory protein (5, 6, 7), termed IL-1R1, IL-1R2, and IL-1R3 in this paper, respectively, in keeping with our previously proposed numbering system (8). IL-1 signaling is initiated by high-affinity binding of IL-1α and β to IL-1R1, which gets subsequently bound by IL-1R3 (5, 7). This results in an intracellular signaling cascade quite similar to stress-induced signal transduction (9), with the end effect being activation of NF-κB (10). IL-1ra and IL-1R2 antagonize the response to IL-1α and β at the ligand and (co)receptor levels (2, 3, 11, 12). Numerous studies have shown that perturbation of such control contributes to the pathogenesis of inflammatory and immunological diseases (i.e., leukemia, rheumatoid arthritis, and psoriasis; Refs. 13 and 14).

Recently, new members of the IL-1 family were identified based on both sequence homology and the presence of key structural patterns. For example, IL-18 (15, 16) is predicted to fold as a β-rich trefoil, typical for IL-1 ligands (17). Moreover, with respect to processing, receptor usage, and signaling, IL-18 can be classified as an IL-1 family member (i.e., IL-1γ; Refs. 18, 19, 20, 21, 22, 23). In addition, new IL-1 ligands have been identified that show strong structural similarities to IL-1ra (24, 25, 26, 27, 28, 29). To date, the expression and especially the receptor usage and function of these IL-1ra-like IL-1 ligands have only been characterized to a limited extent. At the IL-1 receptor level, there also exist additional IL-1R-like molecules. Many of these molecules are currently orphan receptors, such as T1/ST2 (termed IL-1R4; Refs. 30 and 31), IL-1R-related proteins 1 (IL-1R5; Ref. 32) and 2 (IL-1R6; Ref. 33), IL-1R accessory protein ligand (IL-1R7; Ref. 18), single Ig domain IL-1R-related protein (IL-1R8; Ref. 34), IL-1R accessory protein-like (IL-1R9) (35), and IL-1R10 (36), all harboring extracellular Ig-folds and an intracellular domain homologous to the cytosolic part of the Drosophila Toll protein. It is interesting to note that the majority of the IL-1 ligands (i.e., IL-1αβ/ra and some IL-1ra-like IL-1 ligands) and IL-1Rs (i.e., IL-1R1, IL-1R2, IL-1R4, IL-1R5, IL-1R6, and IL-1R7) are clustered and localized to chromosome 2 (18, 25, 28, 37, 38, 39, 40).

The presence of several orphan IL-1Rs suggests the existence of additional corresponding IL-1 ligands. In line with recent reports (24, 25, 26, 27, 28), we have independently identified two novel IL-1 ligands based on sequence homology with IL-1ra, which we termed IL-1δ and IL-1ε. IL-1δ corresponds to the reported sequences of IL1Hy1 (24), FIL1δ (25), murine IL1H3 (26), IL-1RP3 (27) and IL-1L (28), whereas IL-1ε corresponds to IL1H1 (26) and IL-1RP2 (27). These novel IL-1s are strongly expressed in embryonic tissue and epithelial cells, such as skin keratinocytes. The expression of IL-1ε, and to a lesser extent of IL-1δ, is significantly up-regulated in IL-1β/TNF-α-stimulated human keratinocytes. Human IL-1δ and IL-1ε proteins do not activate NF-κB through the classical IL-1Rs, i.e., the IL-1Rs used by IL-1α and β (IL-1R1 and IL-1R3) or IL-18 (IL-1R5/IL-1R7). Instead, IL-1ε activates this transcription factor via IL-1R6, and this response is potently and specifically antagonized by IL-1δ. Lesional psoriasis skin shows a substantially increased expression of both the IL-1 ligands as well as their IL-1R. IL-1δ and ε and IL-1R6 probably constitute an independent signaling system, present in epithelial barriers of our body, which may take part in local inflammatory responses.

Recombinant human IL-1α, IL-1β, IL-4, IFN-γ, and TNF-α were provided by R&D Systems (Minneapolis, MN). Recombinant human IL-18 and IL-1ra were produced at DNAX Research Institute of Molecular and Cellular Biology (Palo Alto, CA). The Q293 and 293-T cell lines were maintained in DMEM supplemented with 5% FBS, 0.3 mg/ml l-glutamine, 100 U/ml penicillin G, and 100 μg/ml streptomycin (Life Technologies, Paisley, U.K.). Human primary epidermal keratinocytes, dermal fibroblasts, dermal microvascular endothelial cells, and melanocytes (Clonetics, San Diego, CA) were cultured in specialized growth medium according to the suppliers recommendations. The Jurkat E6.1 cell line was maintained in RPMI 1640 medium supplemented with 10% FBS, glutamine, and antibiotics.

BLAST searches in the public mouse expressed sequence tag (EST) database with the common portion of murine IL-1ra revealed EST mb49b11.r1 (GenBank accession no. W08205). The insert contained the full-length sequence of a novel IL-1-like molecule, designated IL-1δ. With this mouse sequence as a query, a human EST (5120028H1), derived from RNA from bronchial smooth muscle cells, was found in our proprietary Incyte database (Palo Alto, CA) that contained the full-length open reading frame of the human ortholog of mouse IL-1δ. The same query sequence revealed an additional EST mi08c10.r1 in the public mouse database (GenBank accession no. AA030324), which contained partial sequence of a second novel IL-1-like molecule, designated IL-1ε. The full-length sequence of murine IL-1ε was obtained by extending the 5′ sequence by PCR on murine 17-day-old embryo Marathon-Ready library cDNA (Clontech, Palo Alto, CA). Separately, a Hidden Markov Model HMMer search (http://hmmer.wustl.edu/) with a PFAM alignment of IL-1α, IL-1β, and IL-1ra (http://pfam.wustl.edu/) revealed an EST (HAICR08) derived from RNA from epithelial cells in the Human Genome Sciences database (Rockville, MD) that contained the full-length open reading frame of human IL-1ε.

A multiple alignment of these novel IL-1 sequences and published IL-1 sequences was created using CLUSTALW (41), guided by tertiary structures and predicted secondary structures (with a consensus derived from several algorithms at http://circinus.ebi.ac.uk:8081/submit.html), and refined by eye. Conserved alignment patterns were drawn by Consensus (http://www.bork.embl-heidelberg.de/Alignment/consensus.html). Evolutionary tree analysis was performed with a neighbor-joining algorithm and viewed with TreeView 1.5 (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html).

Adenoviral vectors containing full-length human IL-1δ and IL-1ε sequences were constructed by PCR and used to transfect Q293 packaging cells. Viruses were subsequently purified, with all procedures according to the manufacturer’s protocols (Invitrogen, Carlsbad, CA). Q293 cells (5 × 10⁸) were infected (adenoviruses used at 10 multiplicity of infection) and incubated for 5 days in a cell factory in a total volume of 1 L of serum-free CMF-1 medium (Life Technologies). Culture medium was dialyzed (Spectra/Por membrane tubing; molecular mass cut-off, 6–8 kDa; Spectrum Laboratories, Rancho Dominguez, CA) against 50 mM Tris-HCl, pH 8.0, and 1 mM EDTA, and subsequently passed through hitrap Q Sepharose and heparin columns. The flow-through, containing the IL-1 proteins, was sterile-filtered and concentrated ∼70 times with an Amicon 8400 ultrafiltration cell with a 10-kDa molecular mass cut-off membrane (Millipore, Bedford, MA). The samples were dialyzed against PBS, and the protein content was quantified by PAGE and Coomassie blue staining with lysozyme as a standard. Protein identities were confirmed by N-terminal sequencing. Identically treated culture medium of Q293 cells infected with adenovirus encoding green fluorescent protein served as a negative control. Endotoxin levels were determined by using the Limulus amebocyte lysate assay (BioWhittaker, Walkersville, MD) and were <1.5 EU/100 μg protein. Protein samples were stored at 4°C.

Plasmids encoding full-length human R1, mouse R3, mouse R4, human R5, and human R7 sequences were constructed by inserting PCR-generated cDNA fragments into pME18S (42). Human IL-1R6 cDNA, a generous gift of Dr. R.A. Maki (Neurocrine Biosciences, San Diego, CA), was subcloned directly into pME18S. The reporter gene plasmid pNF-κB-Luc (Stratagene, La Jolla, CA) contains five NF-κB sites and a basic promoter element to drive luciferase expression, and pRSV-βGal results in constitutive expression of β-galactosidase.

Mouse Northern blots containing ∼2 μg of poly(A)⁺RNA per lane, derived from either total embryo at different days postgestation (Clontech) or from various adult tissues (Origene Technologies, Rockville, MD), were hybridized to the mouse IL-1δ and IL-1ε cDNA probes containing the complete open reading frames. Probes were labeled with ³²P by using the Redivue labeling kit (Amersham Pharmacia Biotech, Uppsala, Sweden). Prehybridization, hybridization, stringency washes, and stripping were conducted according to the manufacturer‘s protocols. Membranes were exposed to a phosphorimager.

A panel of various human skin-derived cells, i.e., primary epidermal keratinocytes, dermal fibroblasts, dermal microvascular endothelial cells, and melanocytes, as well as skin biopsies and PBMC were used for TaqMan-PCR analyses. Skin-derived cells were left untreated or were treated with IL-4 (50 ng/ml), IFN-γ (20 ng/ml), or a combination of IL-1β (5 ng/ml) and TNF-α (10 ng/ml) for 18 h before RNA isolation. Biopsies from lesional psoriasis skin and normal healthy skin were kindly donated by Dr. T. Ruzicka (Department of Dermatology, University of Dusseldorf, Germany) and homogenized before RNA isolation. PBMC from a healthy donor, prepared by standard protocols, stimulated with and without PHA served as controls. RNA isolation, cDNA synthesis, and PCR were performed as described elsewhere (43). The amplicons for human IL-1δ (nt 17–90, with numbers starting at first methionine codon), human IL-1ε (nt 337–409), and human IL-1R6 (nt 1378–1448) were analyzed with 6-carboxy-fluorescein-labeled probes. 18S RNA quantities were measured with a VIC-labeled probe and served as internal controls to normalize for the total amount of cDNA. Values are expressed as fg/5 ng total cDNA.

Jurkat E6.1 cells (4 × 10⁶) were transiently transfected with pNF-κB-Luc reporter gene plasmid, pRSV-βGal plasmid, and IL-1R-encoding cDNA plasmid(s) as described previously (43). Twenty hours after transfection, cells were stimulated with 20 ng/ml of human IL-1α, IL-1β, IL-18, or IL-1ra, or 50 ng/ml human IL-1δ or IL-1ε for 6 h. Cells were lysed with reporter lysis buffer (Promega, Madison, WI), and luciferase and β-galactosidase activities were assessed with luciferase assay reagent (Promega) and Galacto-Light Kit (Tropix, Bedford, MA), respectively. Luciferase activities (in RLU) were normalized on the basis of β-galactosidase activities. For inhibition studies of IL-1R1-mediated activation of NF-κB, IL-1α was used at 50 pg/ml in the presence of IL-1ra and IL-1δ at concentrations ranging from 10 pg/ml to 10 μg/ml. Inhibition of the IL-1R6-mediated response was analyzed with IL-1ε used at 50 ng/ml and IL-1δ or IL-1ra at concentrations ranging from 64 pg/ml to 10 μg/ml.

Computational analyses led to the discovery of two novel IL-1 ligands: IL-1δ and IL-1ε. In short, the strategies used both homology-based and probabilistic-based (HMMer) searches (see Materials and Methods for details). The sequences, and a comparison with the previously known IL-1 family members, are given in Fig. 1,A. Fig. 1 B, the corresponding dendrogram, shows evolutionary relationships.

Northern blot analyses show that IL-1δ and IL-1ε are expressed in embryonic tissue and tissues containing epithelial cells (i.e., stomach and skin; Fig. 2). Quantitative PCR analyses on a large panel of mouse and human tissue cDNAs (including various lymphoid organs, kidney, heart, lung, brain, liver, organs of the digestive tract, reproductive organs, and skin) confirmed these findings. Messenger RNA expression in lung tissue appears to be unique to IL-1δ (Fig. 2). It should also be noted that IL-1δ mRNA analysis shows the presence of multiple variants. More in-depth studies, based on quantitative PCR, revealed that in skin, keratinocytes but not fibroblasts, endothelial cells, or melanocytes are the main producers of IL-1δ and IL-ε (Fig. 3). In vitro-cultured keratinocytes contained ∼10-fold more IL-1δ mRNA relative to IL-1ε mRNA. Stimulation with IL-4 or IFN-γ hardly affected the expression levels of IL-1δ and ε mRNA, whereas stimulation with a combination of IL-1β and TNF-α resulted in an enormous increase in the expression of IL-1ε mRNA and to a lesser extent of IL-1δ mRNA (Fig. 3).

The ability of the novel IL-1s to initiate IL-1R-mediated signaling was studied via an NF-κB-dependent reporter assay with ligand-stimulated Jurkat T cells transiently transfected with different pairs of IL-1Rs. The R1/R3 combination, conferring responsiveness to IL-1α and IL-1β (5, 44), did not generate a response to IL-1δ or IL-1ε. Also, the R5/R7 combination, required to mediate a response to IL-18 (18, 43), did not result in signaling on addition of the novel IL-1 ligands (Fig. 4,A). The next step was to test the orphan IL-1R-like molecules IL-1R4 and IL-1R6 (classified as potential ligand-binding receptors based on their homology to IL-1R1; Ref. 36) paired with various other IL-1R-like molecules, i.e., IL-1R3, IL-1R7, IL-1R9, and IL-1R10 (classified as potential signaling receptors based on their homology to IL-1R3; Ref. (36) for their capacity to confer responsiveness to IL-1δ and IL-1ε. Data consistently showed an IL-1R6-mediated activation of NF-κB upon stimulation with IL-1ε, but not IL-1δ or the mock control (Fig. 4 B).

In line with the striking similarity between IL-1δ and IL-1ra, we tested the possibility of IL-1δ being an antagonist of IL-1 responses rather than being an IL-1 agonist. With the same reporter assay with IL-1R-transfected Jurkat T cells, we showed that IL-1ra, but not IL-1δ, is able to antagonize the IL-1R1-mediated activation of NF-κB on stimulation with IL-1α (Fig. 5,A). Vice versa, IL-1δ, but not IL-1ra, is able to antagonize the IL-1R6-mediated activation of NF-κB on stimulation with IL-1ε (Fig. 5 B). Importantly, IL-1ra shows a 50% inhibition of the IL-1R1-mediated response to IL-1α at about a 1000-fold excess over IL-1α, whereas IL-1δ results in a similar inhibition of the IL-1R6-mediated response to IL-1ε at concentrations similar to or even less than IL-1ε.

In lesional psoriasis skin, characterized by chronic cutaneous inflammation, the expression of the novel IL-1 ligands, IL-1δ and IL-1ε, and IL-1R6 are all significantly increased relative to skin from a healthy individual (Fig. 6). The increase is most prominent for IL-1ε, in line with in vitro-cultured keratinocytes stimulated with the pro-inflammatory cytokines IL-1β/TNF-α (Fig. 3). Activation of PBMC also leads to increased levels of both IL-1 ligands and their receptor, albeit to a lesser extent than observed in lesional psoriasis skin.

This paper describes the discovery of two novel members of the IL-1 family, termed IL-1δ and IL-1ε. Several recent studies have reported on the cloning and molecular characterization of IL-1δ and IL-1ε (24, 25, 26, 27, 28). However, the present study is the first to report on the expression in human skin-derived cell types, receptor usage, and initial functional characterization of IL-1δ and IL-1ε.

The structurally aided alignment of both the novel and classical IL-1s in Fig. 1,A shows the conservation of the core 12 β-strands, making up the β-trefoil structure. The presented sequence of human IL-1δ protein is identical with the reported sequences of IL1Hy1 (24), FIL1δ (25), IL-1RP3 (27), and IL-1L (28). At the amino acid level, human IL-1δ and IL-1ra show a high degree of similarity, which is confirmed by the evolutionary tree analysis (Fig. 1 B). With respect to IL-1ε, the presented sequence is identical with the sequences of IL1H1 (26) and IL-1RP2 (27). It is interesting to note that several public mouse ESTs exist, mostly derived from tongue epithelium, with only slight variations relative to the IL-1ε sequence. In addition, FIL1ε (25) also shows a very high similarity to the human IL-1ε sequence presented in this paper (i.e., 51%). How these IL-1ε variants are generated and their biological significance remain unclear.

IL-1δ and IL-1ε are strongly expressed in embryonic development and in tissues such as stomach, lung, and skin (Fig. 2 and PCR analyses on a panel of various tissue cDNAs not shown). Lung tissue only showed expression of IL-1δ messenger RNA (at the Northern blot level), although at the PCR level, expression of both IL-1δ and IL-1ε can be detected in lung-derived cDNAs. Sizes of the predominant messages for IL-1δ are 1.4 and 2.7 kb for stomach and skin tissues, respectively, and 2.0 kb for lung tissue. Interestingly, IL-1δ messenger RNA in lung tissue is reported to lack the second exon relative to other tissues (25). In analogy to IL-1ra, the different IL-1δ messages might reflect different (tissue-specific) splice variants (2, 45, 46, 47). In fact, both IL-1δ and IL-1ra mRNA sequences diverge at the 5′ ends because of usage of alternative first exons (27). A detailed analysis of human skin was performed with quantitative PCR analysis on a panel of first-strand cDNAs derived from various skin-specific cell types. Keratinocytes, but not fibroblasts, endothelial cells, or melanocytes were identified as the major source for IL-1δ and IL-1ε, with levels of IL-1δ being ∼10-fold higher than those of IL-1ε (Fig. 3). In addition, Langerhans cells but not skin-homing T cells, freshly isolated from skin biopsies, showed some expression of IL-1δ and ε (data not shown). In vitro stimulation of keratinocytes with proinflammatory cytokines (i.e., IL-1β/TNF-α) but not with IL-4 or IFN-γ significantly up-regulated the expression of IL-1ε, and to a lesser extent of IL-1δ (Fig. 3). Our observations are in line with reports on the constitutive and induced expression of keratinocyte IL-1δ and IL-1ε mRNA (26, 27). Preliminary in situ hybridization data using mouse tissue sections confirmed that cells of epithelial origin, such as the parietal and chief cells in stomach, and basal keratinocytes in skin are the predominant cellular sources of these IL-1s (not shown). In addition, esophageal squamous epithelium is also reported to express IL-1ε (27). The presence of IL-1δε in epithelial barriers of our body (i.e., skin, digestive, and respiratory tracts), suggests that these novel IL-1s fulfill similar roles as their known family members (i.e., IL-1α and IL-1β) to promote a response to injury or infection (1, 48). In fact, Kumar and colleagues have shown that the epidermal expression of murine IL-1ε is up-regulated in vivo in response to contact hypersensitivity or a viral infection (26).

It is important to note that IL-1δ and IL-1ε neither possess a classical leader sequence (as does secreted IL-1ra; Ref. 2) nor do they possess a distinct pro-form (as do IL-1αβ and IL-18; Refs. 4, 15 , and 16). However, monitoring the presence of C-terminally tagged versions of IL-1δ and IL-1ε in the supernatants and lysates of transfected 293-T cells (human epithelial cells) revealed that these molecules are secreted as 20-kDa proteins (data not shown). This is in agreement with the finding that the human trophoblastic tumor cell line JEG-3 is able to secrete IL-1δ (28), and argues that an alternative mechanism exists to secrete these novel IL-1s. To functionally characterize the novel IL-1s, we expressed and purified adenovirally derived human IL-1δ and IL-1ε and tested these proteins for their capacity to initiate IL-1 signaling, with NF-κB activation as a read-out. The observation that IL-1R1/3 and IL-1R5/7 do not respond to these new protein preparations (Fig. 4,A) might be explained by the fact that receptor-ligand combinations within the IL-1 system are very specific (43). Therefore, we subsequently tested the orphan receptors IL-1R4 and IL- 1R6 paired with various other IL-1R-like molecules. These studies consistently showed that IL-1R6 responded to IL-1ε but not IL-1δ in activating NF-κB in Jurkat cells (Fig. 4 B). Even IL-1R6 single transfectants showed this response. The IL-1 system, as we know it today, typically requires two receptors, a ligand-binding subunit and a signaling subunit, to get an IL-1 response (5, 18). Because IL-1R6 is very homologous to IL-1R1 (33), a ligand-binding type of receptor, we believe that Jurkat cells endogenously express a second signaling type of receptor that can pair with IL-1R6 in the presence of IL-1ε. We know that the following IL-1R-like molecules are expressed by nontransfected Jurkat cells: IL-1R3, IL-1R4, IL-1R8, IL-1R9, and IL-1R10 (PCR data, not shown). Cotransfection of IL-1R6 with either IL-1R3, IL-1R9, or IL-1R10 does not potentiate the response to IL-1ε relative to IL-1R6 single transfectants (not shown). In addition, studies by others with IL-1R1 chimeras and IL-1α-mediated activation of NF-κB as a read-out do not support a combination of IL-1R6 and IL-1R8 to mediate an IL-1 response (34). The search for the additional IL-1ε receptor(s) is currently ongoing.

IL-1δ is most closely related to IL-1ra, and, like IL-1ra, lacks the loop between the fourth and fifth β-strands (see Fig. 1,A), which is typical for IL-1 agonists: IL-1α, IL-1β, IL-18 and IL-1ε. In fact, insertion of the loop amino acids QGEESN of IL-1β confers agonist activity to IL-1ra (49). Therefore, we hypothesized that IL-1δ acts as an antagonist. Indeed, IL-1δ is a very potent antagonist of the IL-1R6-mediated response to IL-1ε at a ratio of IL-1δ:IL-1ε <1 (Fig. 5). Note that the potency of IL-1ra to antagonize the IL-1R1-mediated response to IL-1α is ∼3 orders of magnitude less. The observation by others (25) that their FIL1δ and FIL1ε proteins do not bind to IL-1R6 is in our opinion not contradictory to our findings. Binding studies with partially purified IL-1δ and IL-1ε proteins from conditioned medium of transfected cells and an Fc fusion of IL-1R6 might not be sensitive enough to show binding to these new IL-1s. Moreover, the second receptor might actually be needed for affinity conversion and for binding to become detectable (5, 43).

IL-1δ is a highly specific antagonist of the IL-1R6-mediated response to IL-1ε. For instance, IL-1δ does not respond through IL-1R1, either as an agonist or antagonist (see Figs. 4 and 5), which confirms the reported lack of IL-1δ to induce the production of IL-6 or inhibit the IL-1αβ-induced production of IL-6 by cultured fibroblasts or endothelial cells (28). Moreover, IL-1δ does not respond through IL-1R5 because IL-1δ does not induce the production of IFN-γ or inhibit the IL-18-induced production of IFN-γ by KG-1 cells (28). In fact, a recently cloned IL-1ra homologue, termed IL-1H (with various isoforms: FIL1ζ (25), IL1H4 (26), and IL-1RP1 (27)) was shown to bind to IL-1R5 but not IL-1R1 (29), and may act as a specific IL-18 antagonist.

Expression of human IL-1R6 is restricted to lung epithelium and brain vasculature (33). In extension to these findings, we observed expression of IL-1R6 mRNA in monocytes and in skin-derived keratinocytes, fibroblasts and to a lesser extent endothelial cells. With respect to skin cells, IL-1R6 may in fact mediate proliferation and production of matrix metalloproteinases in response to IL-1ε (preliminary data, not shown). Activated monocytes also show an up-regulated expression of IL-1δ and IL-1ε mRNA that probably explains the presence of these IL-1s in activated PBMC (Fig. 6). The expression of IL-1δ and IL-1ε, as well as IL-1R6, mRNA are all, but most notably IL-1ε mRNA, highly increased in lesional psoriasis skin samples relative to normal control skin samples (Fig. 6). These data are momentarily followed up, but already confirm the involvement of these novel IL-1s in response to skin inflammation (26) and extend the notion that IL-1 ligands and receptors contribute to the pathogenesis of psoriasis (13).

Taken together, IL-1δ and ε and IL-1R6 may constitute an independent signaling system analogous to IL-1αβ/ra and IL-1R1. The IL-1R6 system, present in epithelial barriers of our body, as a result from the coexpression of IL-1δ and IL-1ε, may be in a default off-state. However, perturbation of homeostasis can shift this balance to an IL-1ε-mediated inflammatory or proliferative response, as seen in lesional psoriatic skin.

We thank Deborah Ligett for synthesizing oligonucleotides, Dan Gorman for help in generating and sequencing of expression DNA constructs, Alice Mui for helpful technical and scientific discussions, and Gerard Zurawski for critical reading of the manuscript.