ST2/T1 protein functionally binds to two secreted proteins from Balb/c 3T3 and human umbilical vein endothelial cells but does not bind interleukin 1.

The ST2/T1 receptor, a homologue of the interleukin 1 receptor (IL-1R), was expressed in COS and Drosophila S2 cells as a human IgG-Fc fusion protein. While a type I IL-1RFc fusion protein bound human IL-1 in vitro, the ST2Fc fusion protein did not. Furthermore, IL-1 stimulated a synthetic interleukin-8 promoter reporter gene that was cotransfected into Jurkat cells with a full-length IL-1R type I (IL-1RI) or a chimeric receptor composed of the IL-1RI extracellular domain and ST2 intracellular domain. In contrast, IL-1 did not stimulate the interleukin-8 promoter when cotransfected with a full-length ST2 or an ST2 extracellular/IL-1R intracellular domain fusion protein. Both IL-1RI and the IL-1R/ST2R chimeric receptor also activated a receptor-associated kinase and CSBP/p38 MAP kinase. Using ST2Fc receptor, we have identified, through receptor precipitation, receptor-dot blot and surface plasmon resonance, a putative ligand of ST2 secreted from Balb/c 3T3 and human umbilical vein endothelial cells. The putative ligand was also able to stimulate CSBP/p38 MAP kinase through the ST2 receptor. These results suggest that the ST2 is not an IL-1 receptor but rather has its own cognate ligand.

The ST2/T1 receptor, a homologue of the interleukin 1 receptor (IL-1R), was expressed in COS and Drosophila S2 cells as a human IgG-Fc fusion protein. While a type I IL-1RFc fusion protein bound human IL-1 in vitro, the ST2Fc fusion protein did not. Furthermore, IL-1 stimulated a synthetic interleukin-8 promoter reporter gene that was cotransfected into Jurkat cells with a fulllength IL-1R type I (IL-1R1) or a chimeric receptor composed of the IL-1RI extracellular domain and ST2 intracellular domain. In contrast, IL-1 did not stimulate the interleukin-8 promoter when cotransfected with a fulllength ST2 or an ST2 extracellular/IL-1R intracellular domain fusion protein. Both IL-1RI and the IL-1R/ST2R chimeric receptor also activated a receptor-associated kinase and CSBP/p38 MAP kinase. Using ST2Fc receptor, we have identified, through receptor precipitation, receptor-dot blot and surface plasmon resonance, a putative ligand of ST2 secreted from Balb/c 3T3 and human umbilical vein endothelial cells. The putative ligand was also able to stimulate CSBP/p38 MAP kinase through the ST2 receptor. These results suggest that the ST2 is not an IL-1 receptor but rather has its own cognate ligand.
ST2/T1 was identified as a late response gene induced by either serum or overexpression of v-mos or Ha-ras oncogenes in Balb/c 3T3 or NIH 3T3 cells (1,2). The ST2/T1 (designated ST2 hereafter) gene encodes a 38.5-kDa peptide that is secreted from 3T3 cells as a heavily glycosylated protein of 50 -60 kDa (3). Subsequently, an alternatively spliced form of murine ST2 and rat Fit-1 were cloned that encoded a single transmembrane-spanning protein retaining the extracellular domain found in the soluble ST2 receptor (4,5).
ST2 belongs to the immunoglobulin superfamily and bears significant amino acid identity (ϳ25%) to the extracellular portion of both type I and type II interleukin 1 receptors (IL-1R) 1 (2). Some of the intracellular residues that are required for signal transduction through the IL-1R and are conserved in the Drosophila Toll protein are also found in the membrane form of ST2 (6,7). Furthermore, the gene for ST2 was mapped to mouse chromosome 1 closely linked to the il-1r locus containing both type I and type II receptor genes in support of their common ancestry (8).
Both soluble and membrane bound ST2 receptors are predominantly expressed in hematopoietic tissues in vivo and in established hematopoietic, epithelial, and fibroblast cell lines in vitro (5,9). This expression pattern partially overlaps with that of the type I and type II IL-1Rs (10,11). Soluble IL-1Rs have also been identified from various sources (12)(13)(14) including vaccinia virus. The vaccinia virus coded protein, B15R, binds to IL-1␤ and has been shown to be involved in viral pathogenesis by attenuating host response elicited due to IL-1 production (15,16). Thus, soluble IL-1Rs may modulate IL-1mediated responses by sequestering it and inhibiting its proinflammatory responses (17). The ST2 receptor may play a similar role for its ligand.
We wished to determine if ST2 is a receptor for IL-1 or some other ligand in order to further understand its function. In the present work we have expressed a soluble and membrane form of ST2 and show that it is not a receptor for IL-1. Instead, we show for the first time that the ST2 receptor binds a previously uncharacterized ligand and signals in a manner similar but not identical to IL-1.

MATERIALS AND METHODS
Cell Lines, Culture Conditions, and Metabolic Labeling-COS-1 and human umbilical vein endothelial cells (HUVEC) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (Life Technologies Inc.). Drosophila Schneider 2 (S2) cells were grown in M3 medium supplemented with 10% heat-inactivated fetal bovine serum. Jurkat cells were maintained in RPMI 1640, 10% heat-inactivated fetal bovine serum. Balb/c 3T3 cells were maintained in DMEM, 10% heat-inactivated calf serum. For metabolic labeling, serum-starved (quiescent) cells were incubated for 4 h in methionine and cysteine-free medium containing 100 -150 Ci/ml of trans-35 S-label ([ 35 S]methionine/cysteine), specific activity 1000 Ci/ mmol (ICN Biomedicals, Costa Mesa, CA). For exponentially growing cells, this medium was supplemented with 5% dialyzed fetal bovine serum. Cells were made quiescent by serum starvation, and serum-free medium was added and collected after 48 h. This medium was concentrated using a 10 K cut-off Centriprep spin column (Amicon, Danvers, MA).
Cloning of Murine and Human ST2 cDNA-Human ST2 cDNA was generated by reverse transcriptase-mediated polymerase chain reaction (18) from RNA isolated from Balb/c 3T3 cells based on the published sequence (2). The forward primer, 5Ј-GAATTCGGTTACCGATATCTT-GCTCTTGATTGATAAAC-3Ј, corresponds to Ϫ23 to Ϫ4 (the ϩ1 base being the first base of initiator methionine codon) and contains EcoRI and BstEII restriction sites. The reverse primer, 5Ј-CGGCCGGTAC-CCCTTCCCTCGATGAAACACTCCTTACTTGGATTTTTCC-3Ј, corresponding to codons 321-328, contained an in frame recognition sequence for factor Xa (IEGR) and an in frame KpnI site. PCR products * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed. were first cloned into the PCRII vector (Invitrogen, San Diego, CA), to confirm the sequence, and insert was then excised with BstEII and KpnI and cloned into the MtalFc vector for stable expression in Drosophila cells (driven by the inducible metallothionine promoter) and into the COSFc vector for transient expression in COS cells (driven by the constitutive cytomegalovirus promoter). Both of these vectors contain the Fc portion of human IgG1, beginning with an in frame KpnI site at the start of the hinge segment (19). Mouse ST2Fc and human IL-1RFc constructs were generated by reverse transcriptase-mediated PCR in a similar manner using the following forward and reverse primers, respectively: mouse ST2Fc, 5Ј-GAATTCGGTTACCTGCAGC-CTCAGCCATCAATCACTA-3Ј (Ϫ24 to Ϫ5) and 5Ј-CGGCCGGGTAC-CCCTTCCCTCGATAGCAATGTGTGAGGGACACTCCTTAC-3Ј (codons 330 -337); and human IL-1RFc, 5Ј-GGAATTCGGTTACCAATATGA-AAGTGTTACTCAGACTT-3Ј (Ϫ7 to ϩ2) and 5Ј-ATGGTACCCCTTCC-CTCGATCTTCTGGAAATTAGTGACTGGATATA-3Ј (codons 328 -336). The final constructs of human ST2R, mouse ST2R, and human IL-1R Fc fusion proteins contained 566, 575, and 614 amino acids, respectively.
The sequence for the intracellular portion of MST2 was also amplified by reverse transcriptase-mediated PCR using the forward primer 5Ј-AAGTTCCAGCAATGACATGGATTG-3Ј corresponding to codons 280 -287 of membrane MST2 containing a 5Ј XcmI restriction site. The reverse primer, 5Ј-GTCTCTAGATCACAAGTCCTCTTCAGAAATGA-GCTTTTGCTCAAAGTGTTTCAGGTCTAAGCATGCCTTG-3Ј, corresponding to codons 559 -567, contained a sequence for the myc epitope (20) 9E10 (EQKLISEEDL) and a stop codon followed by an XbaI restriction site. After confirmation of sequence, the PCR product was cloned in place of the 3Ј end of the soluble ST2 and Fc sequence in the COSFc vector between the XcmI and XbaI sites, yielding the 567-amino acid full-length COSMST2R. The MST2/IL-1R chimera was constructed by amplifying the intracellular portion of IL-1R from amino acids 330 -567 with the following primers: 5Ј-CCAATTGATCACACTAATTTCCA-GAAGCACATGATT-3Ј (codons 330 -337 with in frame BclI restriction site) and 5Ј-CTTTCTAGATCACAAGTCCTCTTCAGAAATGAGCTTT-TGCTCCCCGAGAGGCACGTGAGCCTCTCTTTGCAGTTT-3Ј (codons 559 -567 followed by the myc epitope, stop codon, and a XbaI site). The PCR product was used to replace the intracellular portion of the ST2 receptor in COSMST2R from BclI to XbaI. The final product MST2/ IL-1R contained amino acids 1-331 of the ST2 receptor fused to amino acids 330 -568 of the IL-1R.
The full-length IL-1R construct and the truncated version, IL-1R⌬360, which lacks all but five amino acids after the transmembrane domain, was a kind gift from Dr. R. Einstein. For the IL-1R/MST2R chimera, the intracellular portion of ST2 receptor was amplified using the forward 5Ј-AAAGCTTCAGATGGCAAGCTCTACGATGCGTAC-3Ј (codons 378 -385 with a HindIII site) and the reverse primer 5Ј-CAG-GTGACCTCACAAGTCCTCTTCAGAAATGAGCTTTTGCTCAAAGTG-TTTCAGGTCTAAGCATGCCTTGCCACT-3Ј (codons 557-567 with myc epitope, stop codon, and a BstEII site).
The intracellular portion of IL-1R from the HindIII site (amino acid 378 onward) was then replaced with the PCR-amplified intracellular portion of MST2 (from amino acid 378 to 567). The expression vector for these receptors contain a cytomegalovirus promoter and a bovine growth hormone polyadenylation sequence. The authenticity of each construct was confirmed by transient expression of the corresponding proteins in COS cells as analyzed by immunoprecipitation from 35 S metabolically labeled cells.
The IL-8 promoter from Ϫ185 to ϩ44 (21,22) was made by PCRmediated gene synthesis containing a 5Ј HindIII and a 3Ј XbaI site. The PCR product was first cloned into PCRII to confirm the sequence. The insert was then removed by digestion with HindIII and XbaI and subcloned into corresponding sites in the PCATE vector (Promega, Madison, WI), which contains a bacterial chloramphenicol acetyl transferase (CAT) gene cassette, to generate the reporter plasmid IL-8P/CAT.

Expression of Fc Fusion Protein and Its
Purification-For transient expression, all Fc fusion constructs were transfected into COS cells by the DEAE-dextran method (18). 24 h post-transfection, serum-free medium was added to cells and collected after an additional 48 h. Stable Drosophila S2 cells were made by cotransfection of Fc fusion constructs with a plasmid containing the hygromycin resistance gene by the calcium phosphate coprecipitation method (23) and induced with 500 M copper sulfate for 7 days. Supernatants from COS or Drosophila cells were harvested, adjusted to pH 7.5, and passed over a protein A-Sepharose CL-4B (Pharmacia Biotech Inc.) column. The column was washed with 20 column volumes of 0.1 M Tris-HCl, pH 7.5, and the Fc fusion protein eluted with 0.1 M glycine buffer, pH 2.8. The peak eluate fractions were neutralized immediately, pooled, and dialyzed against phosphate-buffered saline (PBS) and stored frozen at Ϫ70°C. The yield was 2-5 g/ml. These recombinant proteins were detected by immunoblotting with anti-Fc, anti-ST2, or anti-IL-1R antibodies and horseradish peroxidase-conjugated secondary antibody followed by enhanced chemiluminescence detection (ECL, Amersham Corp.). Soluble ST2 was obtained by digesting ST2Fc fusion protein with factor Xa (New England Biolabs, Beverly, MA) according to the manufacturer's instructions and passing the digest over a protein A-Sepharose column. The identity of each protein was also confirmed by N-terminal sequencing.
IL-1 Binding Assays and Receptor Precipitation or Immunoprecipitation-For binding assays, Fc fusion proteins were first allowed to bind protein A-Sepharose for 15 min at room temperature. 5,000 -250,000 cpm (specific activity 25,000 -60,000 cpm/ng) of iodinated human IL-1␣ or IL-1␤ (Amersham Corp.) and 0.01-0.2 g of immobilized Fc fusion proteins were mixed in a final volume of 600 l of binding buffer (25 mM Hepes, pH 7.5, 0.1% bovine serum albumin, and 0.01% Tween 20) and incubated at room temperature for 3 h. For competition assays a 1000fold molar excess of unlabeled IL-1 and a 200-fold molar excess of soluble receptors were used. The reaction mixture was then centrifuged in a microfuge at high speed, and the pellets were washed 3 times with 500 l of ice-cold binding buffer. The pellets were counted in a ␥ counter (Beckman Instruments) and resuspended in SDS sample buffer for SDS-PAGE analysis. All experiments were done in triplicate, and results from one representative experiment are shown. Less than 10% of the total iodinated ligand bound at the highest concentration tested and the standard deviation did not exceed 5% in any case.
For receptor precipitation, 1-2.5 g of various purified Fc fusion proteins were mixed with 35 S-labeled 3T3 conditioned medium and 20 l of protein A-agarose (Life Technologies, Inc.) and incubated overnight at 4°C. Protein A-agarose pellets were collected by centrifugation and washed 3 times with PBSTDS buffer (PBS containing 1% Triton X-100, 0.1% SDS, and 0.01% sodium deoxycholate). Pellets were solubilized in sample buffer and resolved through SDS-PAGE, fixed, dried, and visualized by autoradiography. In some experiments the pH of 3T3 conditioned medium was lowered to 3.0 by 0.1 M HCl and then immediately neutralized before the receptor precipitation assay. For immunoprecipitation, polyclonal rabbit antiserum (preimmune or immune) generated against Drosophila-expressed ST2 was used instead of Fc fusions. Cytoplasmic extracts from 3T3 cells were made by washing the cells in ice-cold PBS and lysis in PBSTDS containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 5 units/ml aprotinin for 20 min on ice followed by centrifugation at 15,000 ϫ g for 20 min to remove nuclei and cell debris. For cross-linking experiments, the homobifunctional cross-linker disuccinimidyl suberate (Pierce) was added to preformed ligand-receptor complexes in binding buffer at a final concentration of 1 mM for 30 min, followed by the addition of Tris-HCl, pH 7.4 to 10 mM. Sample buffer was then added, and cross-linked product was resolved by SDS-PAGE. Endogenous soluble ST2 protein was partially purified by immunoprecipitation from concentrated 3T3 cell conditioned medium using anti-ST2 polyclonal antibody. ST2 was eluted from antibody-agarose beads using 10 mM sodium carbonate buffer, pH 11, neutralized to pH 7.4 and used for cross-linking studies.
Jurkat Cell Transfection, CAT, and Kinase Assays-For transfection studies, 1 ϫ 10 6 Jurkat cells were plated, and 24 h later 4 g of receptor and 2 g of IL-8P/CAT and 0.5 g of pCMV-␤-galactosidase reporter DNAs were transfected using Lipofectamine reagent (Life Technologies, Inc.) according to the manufacturer's recommendations. After recovery of cells in serum containing media for 24 h, cells were split into two flasks. One flask was left untreated, and the other flask was treated with 10 ng of IL-1␣ or IL-1␤ for 16 h. Cells were washed with PBS, and extracts were prepared by three freeze-thaw cycles. The extracts were normalized for ␤-galactosidase expression, which was used as an internal control for transfection efficiency. The CAT assay was performed in duplicate as described (24). The radioactivity was quantitated in a Betascope 603 blot analyzer (Betagen, Mountain View, CA).
For the CSBP/p38 kinase assay, 2 ϫ 10 7 cells were transfected with different receptor constructs, and 48 h later one-half of cells were stimulated with IL-1␣ or -␤ or concentrated 3T3 conditioned medium for 5-20 min and lysed, and immune complex kinase assays were carried out for 30 min at 30°C as in Raingeaud et al. (25) using anti-CSBP polyclonal antibodies (26). The kinase reaction was stopped by the addition of SDS-PAGE buffer, boiled for 2 min, and resolved by SDS-PAGE. The bottom half of the gel containing the myelin basic protein (MBP) was prepared for autoradiography, and the top half of the gel containing CSBP was prepared for immunoblotting. The radioactivity in each band was quantitated in a Betascope. For the receptorassociated kinase assay, cells were lysed and immunoprecipitated with a monoclonal anti-IL-1RI antibody (Genzyme, Cambridge, MA), and the kinase assay was performed as in Croston et al. (27) with the exception that MBP (5 g) was also added as an exogenous substrate in the kinase assay. These experiments were performed with cells stimulated with both IL-1␣ and -␤, and comparable results were obtained. Therefore, the data only for IL-1␤ are presented.
Size Exclusion Chromatography of 3T3 Conditioned Medium and Dot Blot Assay-5 ml of serum-free conditioned medium from 3T3 cells was concentrated 10-fold by ultrafiltration (Amicon YM-10 membrane) and applied to a 25-ml (10 ϫ 300-mm) Superdex-75 column (Pharmacia) equilibrated with 50 mM sodium acetate (pH 3.0) and 100 mM NaCl. After the void volume, 1-ml fractions were collected. Molecular weights for each fraction were calculated by linear regression based on standard proteins (Bio-Rad) for gel filtration. 100 l of each of the Superdex-75 fractions were blotted onto the nitrocellulose membrane placed in a BIO-DOT apparatus (Bio-Rad) for dot blot assay. The membrane was blocked with 1% gelatin in PBS for 1 h, washed with PBS containing 0.1% Tween 20 (PBST) and probed with ST2Fc at 2 g/ml in PBST for 1 h. After washing, the membrane was incubated with protein Ghorseradish peroxidase (1:1000, Bio-Rad) and visualized using ECL reagent (Amersham Corp.) and quantitated by densitometry (IS-1000, Alpha Innotech, CA).
BIAcore Analysis-BIAcore, a surface plasmon resonance based biosensor (Pharmacia Biosensors, Piscataway, NJ), was used to study the binding of various ligands to Fc fusion receptors (28,29). The carboxymethyl-dextran surface of the BIAcore chip was activated with Nhydroxysuccinimide and ethylcarbodiimide as described (19). 40 l of protein A (200 g/ml) in 10 mM sodium acetate, pH 4.5, or 40 l of different Fc fusion proteins (20 g/ml) at pH 5.0 were then immobilized through lysine ⑀-groups followed by the blocking of remaining active groups with 1 M ethanolamine. To measure the binding of ligands, Fc fusion proteins were then captured on the protein A surface by two injections of 40 l of solution of purified Fc fusion at 200 g/ml. Binding was analyzed with both directly and indirectly (through protein A) immobilized Fc fusion proteins. Typically 5000 -7000 response units (1000 response units ϭ 1 ng/mm 2 ) of Fc fusion protein were immobilized. After each injection the receptor surface was regenerated by a brief 2-min pulse of 0.1 M glycine, pH 2.8. 20 -40 l of solution containing various concentrations of IL-1 or serum free 3T3 or other cell conditioned medium were passed over the chip surface at a flow rate of 5 l/min, and binding in real time was observed for 4 -8 min, followed by a wash with running buffer (Hepes-buffered saline, pH 7.4, containing 3.4 mM EDTA and 0.005% surfactant P20). The difference between the base line before and after injection of ligands was indicative of the extent of binding.

Expression and Purification of ST2Fc
Protein-In order to study the role of ST2 protein, we chose to express it as an IgG-Fc fusion. Several fusion proteins including Fc fusions have been successfully used for receptor-ligand binding experiments (30,31). Both human and murine ST2 and human type I IL-1R Fc fusion proteins were expressed in COS and Drosophila cells and purified by affinity chromatography on a protein A-Sepharose CL-4B column. Fig. 1 shows the immunoblot of recombinant ST2Fc and IL-1RFc proteins with an anti-Fc antibody. Proteins expressed in COS cells (lanes 1-3) are Ͼ100 kDa, whereas proteins expressed in Drosophila cells (lanes 4 -6) are slightly smaller, ϳ100 kDa. Since the predicted molecular weight for these proteins is ϳ70 kDa, the apparent increase is likely due to glycosylation, in agreement with previous reports (3,32) with differences between Drosophila versus COS expressed proteins reflecting the differences in glycosylation complexity between mammals and insects. Unless otherwise indicated, all experiments were done with both Drosophila-and COS-expressed proteins, and results were comparable.
Binding of ST2 and IL-1R Fc Fusion Proteins to IL-1-We used soluble human ST2Fc and IL-1RFc proteins in a receptor precipitation assay to study the binding of human IL-1. The IL-1RFc showed saturable binding of 125 I-IL-1␣ (Fig. 2), whereas human ST2Fc failed to show any significant binding of 125 I-IL-1␣ over that of control IgG. We also performed a competition assay to determine if soluble ST2 competed in the binding of IL-1 to IL-1RFc. Human IL-1RFc bound an average of 7000 cpm (0.06%) of 125 I-IL-1␣; this binding was competed by excess cold IL-1␣, IL-1␤, IL-1 receptor antagonist, or soluble IL-1 receptor (sIL-1R) but not by soluble ST2 (sHST2R) (Fig. 3,  lanes 1-6). In contrast, HST2Fc fusion protein did not precipitate any significant amount of labeled IL-1␣ (Fig. 3, lanes  7-12). Purified human IgG, IL-5RFc, or protein A-agarose beads also did not precipitate IL-1␣ above background. sHST2R did not compete with the binding of IL-1␣ to IL-1RFc even when added in a 1000-fold molar excess. In contrast, more than 50% inhibition of binding was observed with a 200-fold molar excess of sIL-1R (lane 5). Also, no binding was observed when sIL-1R was included with HST2Fc receptor in the precipitation assay (Fig. 3, lane 11), suggesting that ST2 is not a second subunit of the IL-1R, which might have led to an increased binding. These data show that the HST2Fc fusion protein does not bind IL-1. Similar results were obtained when iodinated IL-1␤ or IL-1 receptor antagonist were used as ligands (data not shown). We were also unable to show any binding of IL-1 to ST2 by adding metal ions in this assay. Metal ions, especially zinc, have been shown to increase the binding affinity of growth hormone to prolactin receptor (33).
Cross-linking of IL-1 to IL-1 Receptor but Not to ST2-The binding of IL-1 to ST2 receptor may have a fast off rate, and the receptor-precipitation assay may not be suitable for such rapidly dissociating receptor-ligand interaction. Therefore, we evaluated the binding of IL-1 to ST2Fc by cross-linking with the homobifunctional agent disuccinimidyl suberate. As shown in Fig. 4, the ST2Fc fusion protein was not cross-linked to   9 and 11), whereas sIL-1R or IL-1RFc was cross-linked to both (lanes 3, 7, and 13). The cross-linked product of sIL-1R with IL-1␣ (lane 7) was slightly larger than that of sIL-1R with IL-1␤ (lane 3). The reason for this difference is not clear at this time. However, these crosslinked bands were specific since they could be competed by the appropriate excess cold IL-1 (lanes 4, 8, and 14). To exclude the possibility that recombinant ST2 protein expressed as Fc fusion may not behave as authentic ST2 protein, we used immunopurified ST2 protein from 3T3 cell conditioned medium, but it also failed to cross-link to either IL-1␣ or IL-1␤ (Fig. 4, lanes  1 and 2 and lanes 5 and 6).
BIAcore Assays-As a more sensitive means of detection we used a BIAcore biosensor instrument to determine if ST2 could bind IL-1. Human ST2Fc protein was captured on the biosensor chip surface via protein A immobilized by cross-linking it to the activated carboxymethyl dextran surface. As shown in Fig. 5, no binding was observed with any of the three IL-1 ligands (white brick bars). The results were negative at various concentrations of IL-1 (10 pM to 10 M) and over a wide concentration range of captured ST2Fc protein. In contrast, a polyclonal immune serum (IM-ST2-Ab) against ST2 protein but not the preimmune serum (PI-Ab) showed distinct binding (white brick bars) and the various IL-1s also bound to IL-1RFc captured in a similar way (black brick bars).
Signal Transduction through the ST2 and IL-1R-It is possible that the soluble ST2 receptor has a very low binding affinity for IL-1 but that the membrane-anchored full-length ST2 receptor may respond to IL-1 binding. To test this hypothesis, we transiently coexpressed membrane ST2 (MST2R) or IL-1R in Jurkat cells together with a synthetic IL-8 promoter-CAT reporter gene. Jurkat cells lack IL-1Rs (34), but have previously been shown to be responsive to IL-1 once transfected with the type I IL-1R cDNA (6). It has also been shown that IL-1 induces IL-8 production in many cell types (35), and IL-8 promoter sequences responsible for this induction have been identified (36). We also created fusions of the extracellular domain of the ST2 receptor with the intracellular domain of IL-1R and vice versa (Fig. 6A, top panel). As shown in Fig. 6A, Jurkat cells cotransfected with the IL-1R and the IL-8P/CAT expression vectors showed a ϳ5-fold induction of CAT activity in response to IL-1␤. An IL-1R construct truncated at amino acid 360 (IL-1R⌬360), with all but five amino acids of the intracellular portion deleted, did not respond to IL-1, showing that the intracellular domain of IL-1R was required for signal transduction. Expression of a chimeric protein containing the extracellular portion of human IL-1R and the intracellular portion of mouse ST2 receptor (IL-1R/MST2R), also resulted in a ϳ7-fold induction of CAT activity in response to IL-1, suggesting that the intracellular domain of ST2 shared signaling determinants with IL-1R. In contrast, neither the full-length ST2 receptor (MST2R) nor a chimeric receptor containing the extracellular portion of MST2 and the intracellular portion of IL-1R (MST2R/IL-1R) responded to IL-1, suggesting that it does not bind IL-1 even when expressed on the cell surface. Similar data were obtained when a HIV1LTR/CAT, another  4 and lanes 7 and 8) or 100 ng of HST2Fc (lanes 9 -12) or 10 ng of human IL-1RFc (lanes 13 and 14) were incubated with 2 ng of iodinated IL-1s (specific activity, 60,000 cpm/ng) as indicated. After 3 h at room temperature, the homobifunctional cross-linker disuccinimidyl suberate was added to a 1 mM final concentration and incubated for an additional 30 min. Cross-linked products were analyzed by SDS-PAGE and autoradiography. Even numbered lanes show cross-linking in the presence of 1000-fold molar excess of unlabeled ligands. IL-1-responsive promoter, was used as a reporter gene. 2 The observation that the intracellular portion of the ST2 receptor can substitute for the intracellular portion of IL-1R suggests that the signal transduction pathways for the intracellular portion of the two receptors are similar. IL-1 is known to activate a recently discovered stress-activated MAP kinase known as CSBP/p38 (25,26,37,38). We next investigated whether the chimeric receptor also activated this MAP kinase. As shown in Fig. 6B, CSBP/p38 was activated in response to IL-1␤ in cells expressing the IL-1R and the IL-1R/MST2R chimera but not in cells expressing the truncated IL-1R⌬360 receptor. It has also been reported that an ϳ80-kDa IL-1R-associated protein kinase is required for IL-1-mediated activation of NF-B (27) and that another protein kinase that phosphorylates MBP co-immunoprecipitates with type I IL-1R in response to IL-1 in T cells (39). We therefore examined if either kinase was activated by the chimeric receptor. While we were unable to detect the ϳ80-kDa autophosphorylating kinase, we did detect an MBP-phosphorylating protein kinase activity that was induced within 5 min following IL-1␤ stimulation in cells transfected with the IL-1R and the IL-1R/MST2R chimera but not with the truncated IL-1R⌬360 receptor (Fig. 6c). These data suggest that at least part of the signal transduction pathway between the IL-1 and the ST2 receptors are common. We could not detect activation of these kinases by either MST2R or the MST2R/IL-1R chimera in response to IL-1 (data not shown).
Identification of a Putative Ligand for ST2-These data suggest that ST2 is a receptor for a ligand other than IL-1. In order 2 S. Kumar, unpublished observation. The extent of acetylation was also calculated and represented graphically. B, result of immune complex kinase assay with CSBP isolated from Jurkat cells transfected with IL-1R⌬360, IL-1R, or IL-1R/MST2R and treated with (white bar) or without (black bar) 10 ng/ml of IL-1␤ for 20 min. 32 P-phosphorylated MBP used as a substrate is indicated. The radioactivity in each band was quantitated and presented in a graphical format. The basal kinase activity in unstimulated cells in each case is considered as 1. From the kinase reaction, an immunoblot of CSBP is also presented. IgG is the heavy chain of anti-CSBP antibody used for immunoprecipitation. C, same as B except that the cells were treated with IL-1␤ for 5 min, and anti-IL-1R antibody was used for immunoprecipitation followed by immune complex kinase assay with MBP as substrate. The basal level of kinase activity in unstimulated cells is considered as 1.
to identify cells that make putative ST2 ligand, we made use of the observation that soluble receptor secretion often accompanies ligand expression. We screened several cell lines for receptor expression by immunoprecipitation. To facilitate the detection of both soluble and membrane-anchored forms of the ST2 receptor, we generated polyclonal antibodies in rabbits using purified soluble ST2 protein expressed in Drosophila cells. These antibodies were used to immunoprecipitate both soluble and full-length ST2 proteins from metabolically labeled cells. As shown in Fig. 7A, a soluble ST2 protein of ϳ50 -60 kDa (arrow) is precipitated from exponentially growing 3T3 medium (lane 2). The anti-ST2 antibodies precipitated two proteins of ϳ40 -50 kDa (open arrowhead, lane 4) and ϳ70 -90 kDa (filled arrowhead, lane 4) from a 3T3 cell extract. The ϳ40 -50-kDa protein is probably the soluble receptor in the process of being secreted or the unglycosylated full-length receptor, whereas the 70 -90-kDa protein is likely to be the cell surface form. These proteins are reported to be highly glycosylated, which is consistent with their higher than predicted molecular weight (1,3,32). Immune serum preabsorbed with recombinant soluble ST2 protein or preimmune serum did not precipitate these proteins (Fig. 7A, lanes 1 and 3, and data not  shown). Similarly, immune serum but not preimmune serum also precipitated ϳ50 -60-kDa soluble ST2 protein (Fig. 7B, lane 2, filled arrowhead) from HUVEC.
We used mouse and human ST2Fc fusion proteins to identify ST2 binding proteins in metabolically labeled media from HU-VEC and 3T3 cells made quiescent by serum starvation. As shown in Fig. 7B, an ϳ18-kDa (arrow) and an ϳ32-kDa protein (open arrowhead) were precipitated from HUVEC medium by HST2Fc (lane 3) but not by control IgG (lane 4). There are additional proteins also precipitated by HST2Fc. However, only the ϳ18and the ϳ32-kDa proteins were precipitated by the mouse ST2Fc (MST2Fc) from quiescent 3T3 cell medium (Fig.  7C, lane 2, arrow and open arrowhead) but not by protein A-agarose beads alone (lane 1) or by control IgG (lane 3). Preincubation of labeled conditioned medium with soluble ST2 protein inhibited the precipitation of both the ϳ18and the ϳ32-kDa proteins by ST2Fc (data not shown). These two proteins were not precipitated by ST2Fc from either HUVEC or 3T3 cell lysates (data not shown).
The experiment was repeated with metabolically labeled conditioned media from exponentially growing 3T3 cells in the presence of serum. Both human and murine ST2Fc fusion proteins precipitated an ϳ18-kDa protein (Fig. 8, lanes 1 and 2,  arrow). However, the intensity of this band was very faint. Since exponentially growing cells secrete a large amount of soluble ST2 protein, whereas quiescent cells do not (32), 2 we suspected that most of the ligand may be bound to the secreted endogenous ST2. To release this potential pool of ligand, labeled conditioned medium from these cells was briefly treated with acid and neutralized before the addition of various Fc fusion proteins. As shown in Fig. 8, lanes 6 and 7, the intensity of the ϳ18-kDa band increased dramatically following this brief acid treatment. All control Fc fusion proteins were negative in this assay (Fig. 8, lanes 3-5 and lanes 8 -10). Acid treatment also led to an increase in the signal of other proteins in the high molecular weight range which was not reproducible and varied among different experiments. These high molecular weight proteins probably result from aggregation of ST2Fc fusion protein alone or with other labeled proteins in the conditioned medium, perhaps due to denaturation of serum proteins and/or ST2 following acid treatment. Alternatively some of these proteins may represent other accessory proteins coprecipitated with the ST2 ligand-receptor complex.  1-3). The conditioned media were made from exponentially growing cells for immunoprecipitation and serum-starved cells for receptor-precipitation, respectively.
To confirm the size of ST2 ligand, we passed the concentrated serum-free 3T3 conditioned medium over a Superdex 75 gel filtration column and assayed the resulting fractions by a dot blot assay using ST2Fc. As shown in Fig. 9, fractions corresponding to Ͻ47 and Ͼ15 kDa were positive in this assay, with maximum signal obtained with fraction corresponding to ϳ20 kDa. These data are consistent with our earlier results from receptor-precipitation studies.
As further evidence for the existence of the ST2 binding proteins, we used BIAcore analysis. A similar assay has been successfully used to identify the ligand for the ECK receptor protein-tyrosine kinase (28). As shown in Fig. 10, both unconcentrated (3T3 1ϫ) and a 10-fold concentrated (3T3 10ϫ) 3T3 cell conditioned medium showed significant binding (white brick bars) to ST2Fc protein captured through immobilized protein A. Soluble ST2 competed for this binding, thus showing its specificity (ϩMST2). Similarly, a 10-fold concentrated conditioned medium from HUVEC (HUVEC 10ϫ) also showed specific binding. The 20 -50-kDa fraction, obtained from concentrated 3T3 conditioned medium after passage through a Superdex 75 gel filtration column (see Fig. 9) was also positive in this assay (data not shown). No binding was observed with concentrated control media (DMEM 10ϫ) to ST2Fc or with various conditioned media to IL-1RFc (black brick bars). Conditioned media from either 3T3 or HUVEC did not show any binding to unactivated chip surface, protein A, or unrelated immobilized Fc fusions. We screened conditioned media from several other cell lines including Jurkat cells for this binding activity, but we were unable to find any other cell lines positive in this assay.
To look for signal transduction by the putative ST2 ligand, we examined CSBP/p38 activation and IL-8 promoter/CAT stimulation in Jurkat cells that have endogenous ST2 receptor. 2 The concentrated 3T3 conditioned medium was able to activate CSBP/p38 MAP kinase (Fig. 11) similar to IL-1␤. The activation of CSBP could be blocked Ͼ80% by preincubation of medium with MST2Fc protein, suggesting that the ST2 ligand is functional. However, the same 3T3-concentrated medium failed to induce transfected IL-1 promoter/CAT or HIV1LTR/ CAT reporter genes (data not shown). Since Jurkat cells have endogenous ST2 receptor we could not test the chimeric ST2/ IL-1 receptor. FIG. 11. Activation of CSBP MAP kinase by 3T3 conditioned medium. 2 ϫ 10 7 Jurkat cells were treated either with 1 ml of 10ϫ concentrated DMEM or 3T3 conditioned medium or with concentrated 3T3 conditioned medium preincubated with 10 g/ml MST2Fc. After 20 min cells were lysed, CSBP was immunoprecipitated, and a kinase assay was performed using MBP as a substrate (upper panel). CSBP was also immunoblotted from the kinase reaction (middle panel). IgG is the heavy chain of anti-CSBP antibody used for immunoprecipitation. The quantitated radioactivity is represented graphically (lower panel). The basal level of kinase activity in unstimulated cells is considered as 1.

DISCUSSION
To identify the ligand(s) of ST2, we have used an ST2Fc fusion protein to assess binding to purified IL-1s and crude cell lysates and media. Our data establish very clearly that none of the IL-1s are ligands for ST2. We did not detect binding via receptor precipitation, cross-linking, BIAcore, or signal transduction assays. This is in contrast to a recent report published while this paper was under review, which detected weak binding of rat ST2/Fit-1 to murine IL-1␤ (40). While we occasionally did detect weak, competable binding of human IL-1␣ and IL-1␤ to high concentrations of ligands and human ST2Fc in receptor precipitation assays, the binding was not saturable. Furthermore, we could not detect any binding of these proteins (including ST2 and IL-1␤ from mouse) by the more sensitive BIAcore, which can detect affinities in the M range, so we concluded that ST2 does not bind IL-1. We agree with these authors, however, that IL-1 does not signal through ST2. A second preliminary report is in agreement with our data (41).
In contrast to the negative data with IL-1, we were able to identify a ligand in 3T3 and HUVEC conditioned media using some of the same assays. In both cell media, two proteins of ϳ18 and 32 kDa were specifically precipitated by ST2Fc. Size exclusion data also indicated a ligand with a molecular mass of ϳ20 kDa, which, along with the variable appearance of the 32-kDa protein, suggests that the ligand binds as a monomer rather than a heterodimer. Although we do not know the relationship of the 32-and 18-kDa proteins, it is possible that the 32-kDa protein is a precursor of the 18-kDa protein, reminiscent of IL-1. However, we did not detect these proteins in cell extracts. The intensity of the 18-kDa protein also varied, depending upon the presence of endogenous ST2 as evidenced by an increase in signal after acid treatment of conditioned media made from exponentially growing cells. In quiescent cells, where no ST2 is made, there was no effect of acid treatment. These data suggest that ST2 ligand is continuously made, whereas the expression of soluble ST2 is modulated by serum and growth conditions. Although we have not been able to define a biological activity for the ST2 ligand(s) we have discovered, the ability of this ligand to activate the stress-activated MAP kinase CSBP/p38 in cells expressing the receptor argues that it is functional. ST2 shares this signal transduction pathway with IL-1RI. Indeed, a chimeric receptor consisting of the extracellular domain of IL-1RI and the intracellular domain of ST2 functions like IL-1RI, inducing a receptor-associated protein kinase, CSBP/p38, and IL-8 promoter-dependent transcription in response to IL-1 binding. This contrasts with the response of ST2 receptor to the ligand we have discovered, where only CSBP/p38 activation was observed.
Several key residues of the IL-1RI required for signal transduction have been defined. Three basic residues (Arg-431, Lys-515, and Arg-518) and three aromatic residues (Phe-513, Trp-514, and Tyr-519) that are conserved in human, murine, and chicken IL-1 receptors are required for IL-1 signal transduction (6,7). All six are conserved in murine ST2, and all but Tyr-519 are also conserved in the Drosophila toll protein (6,7). The region 435-484 of IL-1R is also similar in sequence to the box 1-and box 2-like elements present in gp130, the signal-transducing subunit of the IL-6 receptor family (42), and deletion of this region in IL-1RI abolishes its capacity to induce IL-8 gene expression (22). The experiments with the chimeric IL-1R/ST2 receptor suggest that these regions are functionally conserved in ST2, so that the differences observed between IL-1 and the putative ST2 ligand must be due to other components of ST2. One possibility is that association with a second subunit is required for signaling, as has been suggested for IL-1R (43).
Differences might then be due to the association of ST2 with a different second subunit after ligand binding, which does not trigger the IL-8 promoter.
To conclude, our experiments provide the first evidence for a unique ligand for ST2 that is distinct from IL-1. To further characterize the putative ST2 ligand, we are purifying it in sufficient quantities for sequencing and cloning. The availability of cloned material should then allow further evaluation of the biological role of ST2 and its potential intracellular signaling pathways.