Human Ecalectin, a Variant of Human Galectin-9, Is a Novel Eosinophil Chemoattractant Produced by T Lymphocytes*

A 1.6-kilobase pair cDNA was isolated from a human T-cell-derived expression library that encodes a novel eosinophil chemoattractant (designated ecalectin) expressed during allergic and parasitic responses. Based on its deduced amino acid sequence, ecalectin is a 36-kDa protein consisting of 323 amino acids. Although ecalectin lacks a hydrophobic signal peptide, it is secreted from mammalian cells. Ecalectin is not related to any known cytokine or chemokine but rather is a variant of human galectin-9, a member of the large family of animal lectins that have affinity for b -galactosides. Recombinant ecalectin, expressed in COS cells and insect cells, exhibited potent eosinophil chemoattractant activity and attracted eosinophils in vitro and in vivo in a dose-dependent manner but not neutrophils, lymphocytes, or monocytes. The finding that the ecalectin transcript is present in abundance in various lymphatic tissues To determine if there are multiple ecalectin-like genes in the human 6- m g samples of human blood-derived genomic DNA were After electro- a DNA probed under of high stringency with a - 32 P]dCTP-labeled probes (sequence the 3 9 -untranslated region) of the human ecalectin cDNA.

A 1.6-kilobase pair cDNA was isolated from a human T-cell-derived expression library that encodes a novel eosinophil chemoattractant (designated ecalectin) expressed during allergic and parasitic responses. Based on its deduced amino acid sequence, ecalectin is a 36-kDa protein consisting of 323 amino acids. Although ecalectin lacks a hydrophobic signal peptide, it is secreted from mammalian cells. Ecalectin is not related to any known cytokine or chemokine but rather is a variant of human galectin-9, a member of the large family of animal lectins that have affinity for ␤-galactosides.

Recombinant ecalectin, expressed in COS cells and insect cells, exhibited potent eosinophil chemoattractant activity and attracted eosinophils in vitro and in vivo in a dose-dependent manner but not neutrophils, lymphocytes, or monocytes. The finding that the ecalectin transcript is present in abundance in various lymphatic tissues and that its expression increases substantially in antigen-activated peripheral blood mononuclear cells suggests that ecalectin is an important T-cell-derived regulator of eosinophil recruitment in tissues during inflammatory reactions. We believe that this is the first report of the expression of an immunoregulatory galectin expressed by a T-cell line that is selective for eosinophils.
Eosinophil accumulation is a common feature of many inflammatory diseases such as bronchial asthma, allergic rhinitis, helminth infection, and atopic dermatitis (1). Although eosinophils are proinflammatory cells that normally reside in the circulation, they are induced to extravasate into tissues by a diverse array of chemoattractants and viability-enhancing factors. The eotaxins (2)(3)(4)(5)(6)(7)(8) are the most specific of the known peptide eosinophil chemoattractants (ECA), 1 but granulocyte-macrophage colony-stimulating factor, interleukin (IL)-2, IL-3, IL-5, IL-8, IL-16, RANTES, macrophage inflammatory protein 1␣, monocyte chemotactic peptide (MCP)-2, MCP-3, MCP-4, complement component 3a (C3a), and C5a also can regulate the chemotaxis, viability, growth, differentiation, and/or activation of certain populations of eosinophils (9 -18, 20, 22-25). Platelet-activating factor (26) and 5-lipoxygenase metabolites of arachidonic acid (27)(28)(29) are lipid mediators with potent chemotactic activity for eosinophils. Because these regulatory factors often work in concert, an eosinophil generally has to be cytokine-primed to respond to an ECA.
Increased numbers of eosinophils in tissues that contain antigen-activated T-cells have been observed by several groups (30 -34). However, it has been assumed that T-cells do not directly recruit eosinophils into tissues because they do not produce substantial amounts of an eotaxin. Most ECA are positively charged proteins that are small in size. For example, human eotaxin-1 consists of only 74 amino acids, yet 16 of them are either Lys or Arg (5,6). We previously noted that concanavalin A-activated or antigen-activated CD4 ϩ T-cells from patients infected with Schistosoma mansoni (35) produce an apparently novel 30 -40-kDa ECA with a relatively neutral isoelectric point. The virus-transformed human T-cell line STO-2 also appears to spontaneously produce this ECA (36). Using blocking antibodies, we concluded that the T-cellderived ECA was not IL-3, IL-5, or granulocyte-macrophage colony-stimulating factor. We now describe the purification, molecular cloning, and functional expression of ecalectin, a novel ECA produced by T-cells. Ecalectin is a variant of a protein, designated human galectin-9 (37), of unknown function recently found in the tumor of a patient with Hodgkin's lymphoma.

Partial Purification of Ecalectin and Generation of Ecalectin-specific
Antibodies-The STO-2 cell line was previously derived by transformation of normal human T-cells with human T-cell lymphoma-leukemia virus (35). Fluorescence-activated cell sorting analysis revealed that the cell line is of T-cell origin. It expresses CD2, CD3, CD4, CD5, and CD8 but lacks granulocyte/macrophage and B-cell markers such as CD16, CD19, and Leu7. Supernatants from STO-2 cells (2 ϫ 10 Ϫ5 /ml) that had been cultured for ϳ72 h in SF-02 serum-free medium (Sanko-Junnyaku Co., Tokyo, Japan) supplemented with 0.1% human serum albumin, 100 units/ml IL-2 (Tokushima Research Institute, Tokushima, Japan), 50 M 2-mercaptoethanol, 100 g/ml streptomycin, 100 units/ml penicillin, and 5 g/ml fungizone were adjusted to pH 5.0 with 50 mM sodium acetate buffer. The resulting conditioned medium (5 liters) was applied to replicate 2.6 ϫ 40-cm columns of CM-Sepharose (Amersham Pharmacia Biotech) that had been equilibrated with 50 mM sodium acetate, pH 5.0. Each ion exchange column was washed extensively with equilibration buffer to remove the major albumin contaminate in the conditioned medium. Bound ecalectin was eluted by exposing the * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AB005894 (ecalectin transcript and protein).
§ To whom correspondence and reprint requests should be addressed: Dept. of Bacterial Infection, Institute of Medical Science, University of Tokyo, 4-6-1 Shirokanedai, Minato-Ku, Tokyo 108, Japan. Tel.: 81-3-5449-5313; Fax: 81-3-3444-3197; E-mail: ryoji@hgc.ims.u-tokyo.ac.jp. 1 The abbreviations used are: ECA, eosinophil chemoattractant; C3a and C5a, complement components 3a and 5a, respectively; FCS, fetal calf serum; IL, interleukin; MCP, monocyte chemotactic peptide; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; RANTES, regulated on activation normal T-cell expressed and secreted. column to 20 mM sodium phosphate and 0.1 M NaCl, pH 7.0. After dialysis of the pooled eluate (ϳ300 ml) against 1% glycine, the material was subjected to preparative isoelectric focusing. Those 40-ml fractions having proteins with a pI of 7-8 were collected on a Rotofor system (Bio-Rad). The ecalectin-enriched material was then concentrated ϳ10fold in an ultrafiltration apparatus with a YM-5 membrane (Amicon Corp., Lexington, MA). After the material was dialyzed against phosphate-buffered saline (PBS), it was fractionated further on a 1 ϫ 30-cm column of Superose-12 (Amersham Pharmacia Biotech) equilibrated with the same buffer. Based on a calibration of the gel filtration column, most of the ECA activity was 30 -40 kDa. The ecalectin-enriched fractions were precipitated with 60% saturated ammonium sulfate, dissolved in 20 mM sodium phosphate (pH 7.0) containing 30% ammonium sulfate, and applied to a 4.6 mm ϫ 7.5-cm reverse phase column of TSKgel phenyl-5PW (Tosoh Co., Tokyo, Japan) equilibrated in the same buffer. The column was subjected to a linear gradient, decreasing the ammonium sulfate concentration in the buffer from 30 to 0%. SDSpolyacrylamide gel electrophoresis (SDS-PAGE) analysis of a portion of the final biologically active fraction failed to reveal a Coomassie Bluestainable protein of a defined molecular weight. Nevertheless, New Zealand White rabbits were immunized with the resulting partially purified ecalectin in Freund's complete adjuvant, and after booster injections in Freund's incomplete adjuvant, sera were collected.
Once the ecalectin cDNA described below was isolated and its amino acid sequence was deduced, a more specific antibody was generated in rabbits using an anti-peptide approach. A 13-mer peptide (AFSSSQAPYLSPA) corresponding to residues 2-14 of human ecalectin was synthesized by Peptide Inc. (Osaka, Japan). A Cys residue was then added to the C terminus of the ecalectin peptide to facilitate the coupling of the resulting 14-mer peptide through its thiol group to m-maleimidobenzoic acid N-hydroxysuccinimide esteractivated keyhole limpet hemocyanin (Sigma). A rabbit was immunized with the coupled peptide, sera were collected, and the relevant antibodies were purified using an affinity column consisting of the same peptide coupled to 2-fluoro-1-methylpyridinium toluene 4-sulfonate cellufine (Seikagaku Corp., Tokyo, Japan).
Construction and Screening of a cDNA Library from a T-cell Line-A STO-2 cell cDNA expression library was constructed with the ZAP Express TM cDNA Gigapack cloning kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. In brief, ϳ5 g of poly(A) ϩ mRNA was isolated from 1 ϫ 10 8 STO-2 cells with the FastTrack RNA isolation kit from Invitrogen (San Diego, CA). The first DNA strand was synthesized with Moloney murine leukemia virus reverse transcriptase and a 50-mer oligo(dT) primer (5Ј-GAGAGAGAGAGAGAGAGAGAAC-TAGTCTCGAGTTTTTTTTTTTTTTTTTT-3Ј) containing an internal XhoI site. Synthesis was carried out in the presence of 5-methyl-CTP to methylate internal XhoI sites in the cDNAs, and then EcoRI adaptors were ligated onto the 5Ј-ends of the resulting cDNAs. After digestion with XhoI (Stratagene), the cDNAs were size-fractionated on a 1-ml Sephacryl S-500 HR (Life Technologies, Inc.) gel filtration column that had been equilibrated in 0.1 mM EDTA, 25 mM NaCl, and 10 mM Tris-HCl, pH 7.5. cDNAs of Ͼ500 base pairs were pooled, directionally cloned into the bacteriophage expression vector, and then packaged into phage particles with Gigapack III Gold extract. Escherichia coli (strain XL-1 Blue MRFЈ) was transfected with the resulting phage to obtain a cDNA library with a titer of ϳ1 ϫ 10 11 plaque-forming units/ml. The STO-2 cell cDNA library was then probed with the picoBlue TM immunoscreening kit from Stratagene and the newly generated antibodies (ϳ10 ng/ml) directed against human ecalectin.
Generation and Purification of Recombinant Ecalectin-The ExAssist/XLOLR system (Stratagene) was used for the cloning and expression of ecalectin in COS-1 cells because it allows for efficient excision of pBK-CMV phagemid from the ZAP express vector. With the use of plasmid miniprep isolation kits from Qiagen (Santa Clara, CA), plasmid DNA was purified from the clones selected by the immunoscreening procedure. To determine which immunoreactive clones actually encoded a functional ECA, in each instance ϳ5 g of purified cDNA were transfected into COS cells via the DEAE-dextran method (38) or the SuperFect method employing a polycationic reagent from Qiagen. Transfectants were cultured in RPMI 1640 medium supplemented with 10% (FCS) and 50 M 2-mercaptoethanol at 37°C for 3 days, and then the resulting supernatants were examined as described above for ECA activity. For a negative control, replicate COS cells were transfected in a similar fashion with a pCMV ␤-gal cDNA (Stratagene). The ecalectin cDNAs, isolated from the three immunoreactive clones that exhibited ECA activity in the in vitro assay described below, were sequenced in both directions on an Applied Biosystems automated DNA sequencer as described by Sanger and co-workers (39) using vector-specific primers and newly created oligonucleotides corresponding to internal nucleotide sequences in the ecalectin cDNA.
Recombinant human ecalectin was also expressed in Spondoptera frugiperda 9 (Sf9) insect cells using the approach we previously used to generate a recombinant tryptase zymogen (40). Briefly, the ecalectin cDNA was liberated from pBK-CMV by digesting the mammalian cell expression plasmid with EcoRI and NotI. The resulting cDNA was then inserted into the corresponding restriction sites of pVL1393 (PharMingen, San Diego, CA) downstream of the promoter of the polyhedrin gene. Plasmid DNA, purified by CsCl density gradient centrifugation, was mixed with linearized BaculoGold TM DNA (PharMingen) and calcium phosphate, and the resulting DNA solution was added to Sf9 insect cells. The recombinant baculovirus that contained the ecalectin cDNA was identified by a plaque assay, amplified, and used to infect fresh Sf9 cells. Sf9 cells were cultured at 27°C in TNM-FH medium (Life Technologies) supplemented with 10% heat-inactivated FCS. After 3 days of culture, the infected cells (ϳ7 ϫ 10 6 ) were collected by centrifugation and suspended in 1 ml of 20 mM Tris-HCl, pH 7.5, containing 150 mM NaCl (TBS) and varied protease inhibitors (100 g/ml phenylmethylsulfonyl fluoride (Sigma), 2 g/ml leupeptin (Sigma), 1 g/ml pepstatin A (Sigma), 2 g/ml aprotinin (Sigma), and 1 mM EDTA). The cell suspension was sonicated on ice and centrifuged to remove the debris, and the resulting supernatant was applied to a 1-ml lactose column (Seikagaku Corp.). After the affinity column was washed extensively with TBS to remove contaminating insect cell-derived proteins, recombinant ecalectin was eluted by exposing the column to 200 mM lactose in TBS. Aliquots of the eluate fractions were evaluated for their protein content. Buffer containing SDS and 2-mercaptoethanol was added to each, and the sample was boiled for 5 min and then subjected to SDS-PAGE. Gels were silver-stained to evaluate the number of proteins in each column fraction. Protein blots, prepared from replicate gels, were stained with the above anti-peptide antibody and the ECL TM development system (Amersham Pharmacia Biotech) to demonstrate the presence of the recombinant ecalectin.
In Vitro and in Vivo Chemotaxis-Eosinophil chemotaxis of native, COS cell-derived recombinant, and insect cell-derived recombinant ecalectin was evaluated in vitro as described (35,36). In brief, CD16 Ϫ eosinophils were isolated by subjecting peripheral blood human mononuclear cells from healthy volunteers to a discontinuous density gradient of Percoll (Amersham Pharmacia Biotech), followed by immunomagnetic treatment of the obtained cells with anti-CD16 immunoglobulin (DAKO A.S., Glostrup, Denmark). The purity and viability of the obtained eosinophils were Ͼ97 and 95%, respectively. Chemotactic activity was evaluated with a 48-well chamber (Neuro Probe Inc., Cabin John, MD) containing a polyvinylpyrolidone-free membrane with 5-m standard pore sizes. Human eosinophils (0.5-1 ϫ 10 6 /ml) and varied concentrations of a test chemoattractant were placed in the top and bottom chambers, respectively. Each assay was performed in triplicate. After a 1-2-h incubation at 37°C in a humidified atmosphere of 5% CO 2 , the membrane separating the two chambers was removed and placed in Diff-Quik ® stain (Baxter Healthcare Corp., McGaw Park, IL), and the stained eosinophils that had migrated through the membrane were counted under the microscope. Human C5a (Sigma), IL-5 (Genzyme, Cambridge, MA), RANTES (Genzyme), and eotaxin-1 (Seikagaku Corp.) were used as positive controls.
The ability of ecalectin to attract peripheral blood human neutrophils, monocytes, and lymphocytes in vitro was evaluated in a similar fashion. Neutrophils were obtained by centrifuging human peripheral blood cells at 400 ϫ g for 20 min on a discontinuous density gradient of Percoll. After the contaminating erythrocytes were lysed with ammonium chloride, the purity of the resulting neutrophils was ϳ99%. Monocytes and lymphocytes were separated from one another by briefly culturing the resulting monocyte/lymphocyte-enriched fraction in 75cm 2 plastic culture flasks. The nonadherent lymphocytes were subjected to a second plastic adherence step. As assessed immunohistochemically with fluorescence-activated cell sorting and anti-CD14 immunoglobulin (DAKO A.S.), the monocyte contamination in the resulting lymphocyte preparation was 1% or less. The flasks containing the adherent monocytes were washed several times with PBS. Ice-cold RPMI 1640 containing 5% FCS and 0.5% EDTA were added, and the flasks were shaken on ice for 30 min. The recovered cells were almost pure monocytes. Chemotaxis was carried out essentially as described above for eosinophils except that the membranes used in some of the assays had 3-and 8-m standard pore sizes. Human C5a and IL-8 (Genzyme) served as positive controls in the neutrophil chemotaxis assay. Human lymphotactin (Genzyme) and MCP-1 (Genzyme) served as positive controls in the lymphocyte and monocyte chemotaxis assays, respectively.
For evaluation of its ECA activity in vivo, purified insect cell-derived recombinant ecalectin was dialyzed extensively against PBS to remove the lactose. PBS containing 1% human serum albumin (1.0 ml) with or without purified ecalectin (ϳ10 Ϫ7 M) was then injected into the peritoneal cavities of replicate C57BL/6 mice. 14 -24 h later, each mouse was sacrificed, 6 ml of PBS containing 2% FCS was injected into the peritoneal cavity, and the obtained cells in the peritoneal exudate were centrifuged onto slides. The resulting slides were stained with Diff-Quick, and the number of eosinophils that had extravasated into the peritoneal cavity of the mouse was determined.
DNA and RNA Blot Analyses-To determine if there are multiple ecalectin-like genes in the human genome, 6-g samples of human blood-derived genomic DNA (Noven, Madison, WI) were incubated overnight at 37°C with 30 -60 units of BamHI, BglII, EcoRI, HindIII, KpnI, NotI, PstI, SalI, XbaI, or XhoI (New England Biolabs). After electrophoretic separation of the resulting digests in a 0.9% agarose gel (41), a DNA blot was prepared and probed under conditions of high stringency with [␣-32 P]dCTP-labeled probes that correspond to residues 1-460 or residues 1073-1579 (sequence corresponding to the 3Ј-untranslated region) of the human ecalectin cDNA.
Premade blots containing either total RNA (OriGene Technologies, Rockville, MD) or poly(A) ϩ RNA (CLONTECH, Palo Alto, CA) from various human tissues were analyzed with the radiolabeled ecalectin probes. RNA blots were washed twice for 15 min each at room temperature in 2ϫ SSC containing 0.1% SDS and then twice for 30 min each at 60°C in 0.1ϫ SSC containing 0.1% SDS. After autoradiography, the images were analyzed with a BAS-2000. 32 P imaging apparatus (Fuji Film Co., Tokyo, Japan).
To determine if the steady-state level of the ecalectin transcript in a T-cell can be antigen-regulated, heparinized peripheral blood was obtained from two volunteers sensitive to Dermatophagoides farinae, and the peripheral blood mononuclear cells were isolated by Ficoll-Hypaque density gradient centrifugation. The obtained cells (5 ϫ 10 6 /ml) were suspended in RPMI 1640 medium supplemented with 5% FCS and cultured for 48 h in the absence or presence of 5 g/ml of D. farinae. Under these in vitro conditions, the T-cells in the peripheral blood mononuclear cell preparation should be the cells that are preferentially activated. Total RNA was isolated from control and D. farinae-stimulated cells with RNeasy Midi kits from Qiagen. For blot analysis of each sample, ϳ5 g of denatured total RNA were electrophoresed in a 1.4% agarose/formaldehyde gel, and the separated RNA was blotted onto Hybond™ nylon membranes (Amersham Pharmacia Biotech) (42). After UV-cross-linking of the RNA, the membranes were incubated at 68°C for 30 min in QuikHyb (Stratagene) containing 40 g/ml of denatured salmon sperm DNA (Stratagene) alone and then for 2 h in the same hybridization buffer containing a radiolabeled ecalectin probe. To ensure that approximately equal amounts of RNA were loaded in the individual lanes of the blot, the same blot generally was reprobed with the 548-base pair fragment (CLONTECH) that consists of residues 256 -804 of a human glyceraldehyde-3-phosphate dehydrogenase cDNA (43). Supernatants from antigen-activated T-cells from normal donors were also evaluated by SDS-PAGE/immunoblot analysis with the antipeptide antibody described above for determining whether or not immunoreactive ecalectin was exocytosed from these nontransformed cells.

Isolation and Characterization of Human Ecalectin
Transcript and Protein-32 immunoreactive clones were obtained when the expression cDNA library prepared from the human T-cell line was screened with anti-ecalectin immunoglobulin directed against the semipurified factor. When the cDNAs from these clones were individually transfected into COS cells and the supernatants of the transfectants were analyzed 2 days later, three of the clones exhibited potent ECA activity in the in vitro chemotaxis assay. Subsequent nucleotide sequence analysis revealed that these clones encoded the same protein. The consensus nucleotide and deduced amino acid sequences of the STO-2 cell-derived ecalectin cDNAs are depicted in Fig. 1. If the translation-initiation and translation-termination sites begin at residues 60 and 1028, respectively, the open reading frame of the transcript encodes a 36-kDa protein consisting of 323 amino acids. The ecalectin transcript has a typical polyadenylylation signal sequence at residues 1576 -1581. 59 and Ͼ560 nucleotides are present in its 5Ј-and 3Ј-untranslated regions, respectively. Repetitive sequences are often found in the 3Ј-untranslated regions of transcripts that have short halflives (44). Although no repetitive "AUUUA" motif resides in the 3Ј-untranslated region as it does in eotaxin-1 and most other chemokine and cytokine transcripts, four copies of a "CCCUCC" motif reside between residues 1341 and 1484 in the 3Ј-noncoding portion of the ecalectin transcript. Native ecalectin is secreted from the STO-2 cell line. Although the deduced amino acid sequence of the STO-2-derived transcript indicates that ecalectin contains large numbers of Phe, Leu, and Val residues, this ECA lacks a hydrophobic signal peptide at its amino terminus. Ecalectin is also predicted to have a pI of ϳ8.1. Ecalectin has four Cys residues and three potential Asnlinked glycosylation sites at positions 34, 79, and 137.
Expression of the Ecalectin Transcript in Cells and Tissues-Blot analysis of genomic DNA digested separately with 10 different restriction enzymes indicated multiple genes in the human genome encode ecalectin-like transcripts. Nevertheless, the 5Ј (Fig. 2, lanes 3, 5, 6, 8, and 10) and 3Ј (data not shown) ecalectin probes often hybridized strongly to one major DNA fragment in the digested genomic DNA. When varied human tissues were analyzed with these probes, the ϳ1.8-kb ecalectin transcript or its homolog was found in relative abundance in lymphoid tissues (Fig. 3). Because of the prominence of the ecalectin transcript in T-cell-enriched tissues, we investigated whether or not its level increases in peripheral blood mononuclear cells after antigen stimulation. As shown in Fig. 4 for one of the two allergic patients, these cells contained a small amount of ecalectin mRNA. However, the amount of ecalectin transcript increased ϳ50and ϳ90-fold in the two experiments after the cells from the patients were exposed to D. farinae for 48 h. Although the steady-state level of a transcript does not always correlate with the amount of protein that the cell produces, these RNA blot findings suggest that the T-cells in the preparation dramatically increased their expression of ecalectin when they were exposed to the relevant antigen. SDS-PAGE/immunoblot analysis with the peptide antibody directed against the N terminus of ecalectin revealed that immunoreactive ecalectin was present in the conditioned medium of activated, nontransformed human T-cells (Fig. 4). (Fig. 5). Nevertheless, because insect cells express measurable amounts of ecalectin by silver staining analysis of an SDS-PAGE gel (Fig. 5A), we attempted to affinity-purify the immunoreactive (Fig. 5B) recombinant protein from the lysates of these cells. The observation that the ecalectin cDNA appears to encode a protein with carbohydrate-binding domains suggested that the recombinant protein probably would bind to a lactose column. Thus, soluble lysates of insect cells were subject to lactose affinity chromatography. After the column was exposed to 200 mM lactose, a single ϳ36-kDa silver-stained protein (Fig. 5C) was eluted, which was recognized by anti-ecalectin antibody (Fig.  5D). The culture supernatant of COS cells transfected with an ecalectin cDNA (data not shown) and purified, recombinant, insect cell-derived ecalectin (Fig. 6) exhibited dose-dependent ECA activity. In our in vitro system, the optimal chemotactic activity of IL-5, RANTES, eotaxin-1, and C5a for freshly isolated peripheral blood eosinophils was achieved at 10 Ϫ11 M, 10 Ϫ8 M, 10 Ϫ8 M, and 10 Ϫ8 M, respectively. Insect cell-derived ecalectin was active at 10 Ϫ10 M, and its activity at 10 Ϫ8 M was approximately one-half of that of eotaxin-1. Although IL-5 was active at a lower molar concentration, in comparative assays carried out on replicate eosinophils from the same patients, recombinant ecalectin induced ϳ4-fold more eosinophils to migrate through the membrane than the optimal dose of recombinant IL-5. COS cell-derived (data not shown) and insect cell-derived (Fig. 7) recombinant ecalectin did not induce chemotaxis of peripheral blood neutrophils, lymphocytes, or monocytes. 4 -24 h after C57BL/6 mice were given insect cell-derived ecalectin, 15.1 Ϯ 2.7% (n ϭ 3; mean ϮS.D.) of the cells in the peritoneal cavity exudates consisted of eosinophils. In contrast, the number of eosinophils was below detection (Ͻ1 eosinophil/500 cells) in the exudates of control mice given buffer lacking ecalectin. Thus, ecalectin has potent ECA activity in vitro and in vivo. DISCUSSION We previously reported that antigen-activated T-cells produce an ECA with unique biochemical properties (35, 36). We now describe the isolation, cloning, and expression of this ECA, which is discovered to be a member of the galectin family of lectins. Polyclonal antibodies prepared against the partially purified factor from T-cell conditioned medium were used to screen a cDNA expression library prepared from a T-cell line to obtain candidate cDNAs that encoded the novel ECA (Fig. 1). To confirm that the isolated cDNAs encoded a protein possessing ECA, the cDNAs were expressed in COS cells, and the resulting supernatants from the transfectants were evaluated for their chemotactic activity. Because native ecalectin was isolated from the conditioned medium of a T-cell line and because recombinant ecalectin was isolated from the conditioned medium of a COS cell transfectant, ecalectin can be secreted by two very different types of mammalian cells. Nevertheless, because we were unable to induce insect cells to exocytose recombinant ecalectin, we cannot rule out the possibility that some ecalectin is retained by activated T-cells in vivo. Recombinant ecalectin induced the specific chemotaxis of eosinophils (Figs. 6 and 7). On a weight or molar basis, ecalectin exhibited an ECA that is approximately one-half as potent as that of eotaxin-1. Native ecalectin was isolated from a human T-cell line, whereas recombinant ecalectin was isolated from both COS cells and insect cells. Only one silver-stained protein was detected after insect cell-derived ecalectin was purified by lactose affinity chromatography (Fig. 5). Although this finding suggests that ecalectin does not require an insect cell-derived cofactor to exert its chemotactic activity, we still cannot rule out the possibility that an unrecognizable low molecular weight peptide, oligosaccharide, or glycolipid stays tightly bound to the galectin during the affinity chromatography step or after it is dissolved in medium.

Recombinant ecalectin was spontaneously secreted from COS cells but not insect cells
Genomic blot analysis revealed that there are a number of closely related ecalectin-like genes in the human genome (Fig.  2). As assessed by RNA blot analysis, ecalectin (Figs. 3 and 4) is expressed by T-cell-rich lymphoid tissues or cells. However, a relatively high level of the ecalectin transcript was also found in the stomach RNA sample. Patients with chronic gastritis often have large numbers of activated T-cells and eosinophils in their stomach (45). Because the stomach RNA was obtained from a commercial source, it was not possible for us to ascertain whether some of the transcripts in the sample originated from activated T-cells. Nevertheless, because the ecalectin was found in the stomach, as well as in the lung and small intestine (Fig. 3A), it is likely that T-cells are not the only cell type that expresses ecalectin.
A homology search of the GenBank TM protein and nucleotide data bases revealed that STO-2 cell-derived ecalectin is not related to any known cytokine or chemokine. Rather, it is related to the galectin family of carbohydrate-binding proteins (46). Based on a comparison of its deduced amino acid sequence (Fig. 8), ecalectin most closely resembles the proteins designated human (37) and mouse (54) galectin-9. Mouse galectin-9 cDNA was isolated from an embryonic kidney cDNA library, and expression studies revealed that its transcript is particularly abundant in liver. Although the mouse galectin-9 transcript is present in the thymus, in situ hybridization and immunohistochemical analyses revealed that this galectin is expressed primarily by thymus epithelial cells rather than by T-cells (54). Micromolar concentrations of recombinant mouse galectin-9 induced cultured T-cells to undergo apoptosis (55). Because the ecalectin-expressing STO-2 cell line is a T-cell line that is not undergoing apoptosis and because the ecalectin transcript is not abundant in human liver (Fig. 3A), either ecalectin is not the human homolog of mouse galectin-9 or this  galectin evolved in the two species to carry out different functions.
The relationship of ecalectin and the Hodgkin's lymphoma protein designated human galectin-9 (37) also remains to be determined. A BLASTN (56) computer search of the Gen-Bank TM /EMBL nonredundant nucleic acid sequence data base, focusing on the six regions that differ in the ecalectin and human galectin-9 transcripts, revealed clones that correspond to one or the other transcript. Thus, the differences in the nucleotide sequences between the ecalectin and galectin-9 transcripts are not cloning artifacts. Relevant to our study, one of the ecalectin-like expressed sequence tags in the data base originated from the human Jurkat T-cell line (accession number EST63150). The inability of Tü reci and co-workers (37) to find an allelic variant of galectin-9 in a preliminary reverse transcription-polymerase chain reaction analysis of the peripheral blood of normal patients coupled with the finding that four of the five amino acid differences in galectin-9 and ecalectin are substantial raises the possibility that ecalectin represents a new galectin. However, because ecalectin and galectin-9 differ only in a few amino acids, because the 5Ј-untranslated regions of their transcripts are identical, because the 3Ј-untranslated regions of their transcripts differ by only one nucleotide at residue 1547, and because galectin-9 is also expressed in a peripheral blood cell, the most likely interpretation of the data from the two studies is that ecalectin and galectin-9 are two allelic variants of the same human gene.
Galectins differ from Ca 2ϩ -dependent C-lectins. Galectins have conserved sequences that enable them to interact with ␤-galactoside residues. As noted in Fig. 7, those residues thought to be the most important for carbohydrate binding in galectin-2 (57) are present in ecalectin. Galectin-1 and galectin-2 are dimers, whereas the other galectins are monomers. Galectin-1, -2, -3, -5, and -7 have a single carbohydrate-binding domain, whereas galectin-4, -8, and -9 have two homologous carbohydrate-binding domains. In this regard, ecalectin resembles the latter family of galectins. Recombinant ecalectin readily binds to a lactose column (Fig. 5, C and D), as do other galectins. How ecalectin selectively induces eosinophil chemotaxis remains to be determined at the molecular level. However, the finding that ecalectin has two putative domains that both bind carbohydrate suggests that this galectin might function as a bridging protein interacting with two or more galactoside-containing molecules on the surface of the eosinophil. Because ecalectin has four Cys residues (Fig. 1), it could have two disulfide bonds. However, structural analysis of crystallized galectin-2 (57) suggests that these residues do not form disulfide bonds.
Because all galectins lack a hydrophobic signal peptide, they are often found in the cytosol. It has been proposed that galectins have intracellular functions. However, galectins can reside on the surface of a cell and often are spontaneously secreted by a mechanism that has not been deduced. For example, galectin-3 (49) is the Mac-2 antigen found abundantly expressed on the surface of inflammatory macrophages (58). When periph-

FIG. 5. Expression of human ecalectin in insect cells and affinity purification of the resulting recombinant protein.
Lysates of insect cells before (lane 1) and after (lane 2) they had been induced to express human ecalectin were subjected to SDS-PAGE gel analysis. The gel depicted in A was silver-stained to detect all of the proteins in these cells. The protein blot in B was prepared from the replicate gel; it was stained with the peptide antibody directed against the N terminus of human ecalectin. In C and D, the soluble lysate of ecalectin-expressing insect cells was applied to a lactose affinity column. After the column was washed with equilibration buffer, it was exposed to buffer containing 200 mM lactose. Samples of the resulting six fractions were subjected to SDS-PAGE gel analysis. The gel depicted in C was stained with silver, and the corresponding protein blot in D was stained with anti-ecalectin antibody. Molecular weight makers are shown on the left of each. The arrows point to recombinant ecalectin. eral blood mononuclear cells are exposed to antigen, ecalectin mRNA increases dramatically (Fig. 5), and the amount of ECA recovered in the conditioned medium increases accordingly (35). Exocytosis of ecalectin could occur via a pathway that does not involve post-translational modification of the protein. However, ecalectin has two sites in its amino terminus at residues 27 and 57 that might be susceptible to IL-1␤-converting enzyme (59). Thus, some ecalectin may be released into the conditioned medium by the same proteolytic mechanism that T-cells use to generate soluble IL-1␤.
Many galectins are developmentally regulated (46). An ecalectin-like expressed sequence tag has been identified in Jurkat T-cells. However, our finding that the steady-state level of ecalectin mRNA is low in human peripheral blood mononuclear cells (Fig. 4) until these cells are activated with antigen may explain why there has been no published in depth report of ecalectin expression in a T-cell line. Although members of the large galectin family are expressed in numerous cell types, some galectins exhibit immunoregulatory activities, and some galectins are expressed by certain hematopoietic cells. Galectin-1, expressed in abundance in muscle, can induce T-cells to undergo apoptosis (60). Galectin-3 has been localized in macrophages, basophils (61), eosinophils (62), and mast cells (61). Immunoelectron microscopic studies have indicated that much of the galectin-3 found in a mast cell is granule-associated. Because it readily binds to the oligosaccharides attached to both IgE and its receptor, galectin-3 can induce most, if not all, Fc ⑀ RI-bearing mast cells and basophils to degranulate (63). Galectin-3 prevents T-cells from undergoing apoptosis (64), induces neutrophils to increase their production of superoxide (65) and to adhere to laminin (66), and potentiates the production of IL-1 by human monocytes (67). Because both galectin-1 and galectin-3 bind to laminin, these two galectins could influence cell proliferation, cell-cell interactions, and cell-matrix interactions in vivo. Erythrocytes express galectin-5, which FIG. 6. ECA activity of recombinant ecalectin. In the in vitro assay used in these experiments, a porous membrane separates the top chamber that contains human eosinophils from the bottom chamber that contains medium with or without an ECA. After the incubation, the membrane that separates the two compartments in each well was removed and stained with Diff-Quik. Depicted in A is the stained side of the membrane that faces the bottom chamber of cells exposed to conditioned medium from a mock transfectant (left panel), conditioned medium from an ecalectin transfectant of COS cells (center panel), or culture medium supplemented with an optimal amount of eotaxin-1 (right panel). The colored cells are the eosinophils that had migrated through the black and white holes in the membrane. Depicted in B are the quantitative chemotaxis data from multiple experiments using insect cell-derived ecalectin that had been purified by lactose affinity chromatography. In these experiments, the ECA activity of ecalectin (q) was similar to that obtained with an optimal dose of RANTES (f) and IL-5 (Ⅺ) and was approximately one-half of that obtained with an optimal dose of eotaxin-1 (‚). The number of eosinophils that had migrated through the membrane is indicated on the y axis.

FIG. 7. Analysis of the chemotactic activity of recombinant human ecalectin for neutrophils, lymphocytes, and monocytes.
The ability of purified, insect cell-derived ecalectin (ϳ10 Ϫ8 M) to attract peripheral blood neutrophils (A), lymphocytes (B), and monocytes (C) in the in vitro assay was determined. PBS was used as a negative control in each assay. IL-8, lymphotactin, and MCP-1 were used as positive controls for the chemotaxis of neutrophils, lymphocytes, and monocytes, respectively. The numbers of neutrophils, lymphocytes, and monocytes that had migrated through the membranes are indicated on the y axis. The depicted results are the mean Ϯ S.D. of an experiment in which each assay was carried out in triplicate. exhibits weak agglutinin activity (51).
While our study was in progress, galectin-9 was identified during an analysis of the antigens that are selectively expressed by a patient with Hodgkin's lymphoma (37). Although the cellular origin of the galectin-9 transcript was not identified in the patient's tumor, galectin-9 mRNA or a related transcript also was found to be expressed in an undefined, peripheral blood leukocyte. Hodgkin's disease, which accounts for nearly 1% of all cancers in United States, is a neoplastic disorder primarily involving lymphoid tissue (68). Neoplastic Reed-Sternberg cells are found in affected tissue and are surrounded by a variable inflammatory infiltrate of B-cells, histiocytes, plasma cells, stromal cells, neutrophils, T-cells, and eosinophils. Although different types of Hodgkin's lymphoma have been identified, eosinophils regularly appear in those patients with the "nodular sclerosis" form of the disease. Interestingly, the galectin-9 transcript was isolated from a cDNA library prepared with mRNA isolated from a patient with the "nodular sclerosis" form of Hodgkin's lymphoma (37). While it has been proposed that the increased number of eosinophils in this subtype of Hodgkin's lymphoma is primarily a consequence of increased production of IL-5 and/or granulocyte-macrophage colony-stimulating factor by the Reed-Sternberg cells (19,21,68,69), galectin-9 and/or its variant ecalectin could work in concert with the two viability-enhancing cytokines to cause the prominent eosinophilia. If galectin-9 and ecalectin are allelic variants of the same gene, galectin-9 probably originates, in part, from the antigen-activated T-cells present in Hodgkin's lymphoma. Whatever its mechanism of action at the molecular level, ecalectin appears to be a newly recognized factor by which antigen-activated T-cells regulate eosinophil extravasation into tissues. Because the eotaxins are expressed primarily by epithelial and endothelial cells rather than by T-cells (3), ecalectin could be the dominant ECA expressed during certain T-cell-mediated inflammatory reactions in humans. FIG. 8. Comparison of the deduced amino acid sequence of human ecalectin with those of related galectins. Dashes and dots in the amino acid sequences indicate identical residues and insertions, respectively. The numbers indicate the residues in human ecalectin. The solid and dashed thick lines under the sequences indicate the putative residues that correspond to first and second carbohydrate-binding regions, respectively, in ecalectin. The conserved amino acids reported to be essential for lectin activity are shaded. Human ecalectin and human galectin-9 differ in their amino acid sequences at residues 5, 88, 135, 238, and 281. The depicted amino acid sequences of human galectin-1, human galectin-2, human galectin-3, rat galectin-4, rat galectin-5, human galectin-7, rat galectin-8, mouse galectin-9, and human galectin-9 have been reported by others (37,(47)(48)(49)(50)(51)(52)(53)(54). The first 100 amino acids of human galectin-3 are not shown in this alignment figure.