Portal de Periódicos da CAPES

Human Intelectin Is a Novel Soluble Lectin That Recognizes Galactofuranose in Carbohydrate Chains of Bacterial Cell Wall

2001; Elsevier BV; Volume: 276; Issue: 26 Linguagem: Inglês

10.1074/jbc.m103162200

1083-351X

Shoutaro Tsuji, Junji Uehori, Misako Matsumoto, Yasuhiko Suzuki, Akio Matsuhisa, Kumao Toyoshima, Tsukasa Seya,

Glycosylation and Glycoproteins Research

Galactofuranosyl residues are present in various microorganisms but not in mammals. In this study, we identified a human lectin binding to galactofuranosyl residues and named this protein human intelectin (hIntL). The mature hIntL was a secretory glycoprotein consisting of 295 amino acids and N-linked oligosaccharides, and its basic structural unit was a 120-kDa homotrimer in which 40-kDa polypeptides were bridged by disulfide bonds. The hIntL gene was split into 8 exons on chromosome 1q21.3, and hIntL mRNA was expressed in the heart, small intestine, colon, and thymus. hIntL showed high levels of homology with mouse intelectin,Xenopus laevis cortical granule lectin/oocyte lectin, lamprey serum lectin, and ascidian galactose-specific lectin. These homologues commonly contained no carbohydrate recognition domain, which is a characteristic of C-type lectins, although some of them have been reported as Ca2+-dependent lectins. Recombinant hIntL revealed affinities to d-pentoses and ad-galactofuranosyl residue in the presence of Ca2+, and recognized the bacterial arabinogalactan ofNocardia containing d-galactofuranosyl residues. These results suggested that hIntL is a new type lectin recognizing galactofuranose, and that hIntL plays a role in the recognition of bacteria-specific components in the host. AB036706 Galactofuranosyl residues are present in various microorganisms but not in mammals. In this study, we identified a human lectin binding to galactofuranosyl residues and named this protein human intelectin (hIntL). The mature hIntL was a secretory glycoprotein consisting of 295 amino acids and N-linked oligosaccharides, and its basic structural unit was a 120-kDa homotrimer in which 40-kDa polypeptides were bridged by disulfide bonds. The hIntL gene was split into 8 exons on chromosome 1q21.3, and hIntL mRNA was expressed in the heart, small intestine, colon, and thymus. hIntL showed high levels of homology with mouse intelectin,Xenopus laevis cortical granule lectin/oocyte lectin, lamprey serum lectin, and ascidian galactose-specific lectin. These homologues commonly contained no carbohydrate recognition domain, which is a characteristic of C-type lectins, although some of them have been reported as Ca2+-dependent lectins. Recombinant hIntL revealed affinities to d-pentoses and ad-galactofuranosyl residue in the presence of Ca2+, and recognized the bacterial arabinogalactan ofNocardia containing d-galactofuranosyl residues. These results suggested that hIntL is a new type lectin recognizing galactofuranose, and that hIntL plays a role in the recognition of bacteria-specific components in the host. AB036706 mannose-binding lectin 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate human intelectin recombinant human intelectin Tris-buffered saline 2-acetamido-2-deoxy-4-O-β-d-galactofuranosyl-d-glucopyranose culture supernatant polyacrylamide gel electrophoresis polymerase chain reaction polyvinylidene difluoride In host defense, the recognition of bacterial components is important for induction of immune responses. The cell wall components of pathogens have various biological activities and contain the bacteria-specific carbohydrate chains that do not exist in mammals. The recognition of these carbohydrate chains is useful to induce the cellular responses and fluid-phase immune reactions for elimination of pathogens. In the innate immune response, the bacterial carbohydrate chains are recognized by the animal lectins that are present on cells as phagocytosis receptors or in plasma as opsonins or agglutinins. As a phagocytosis receptor, the mannose receptor binds materials containing terminal mannosyl residues such as zymosan and enhances their clearance by phagocytes (1Warr G.A. Biochem. Biophys. Res. Commun. 1980; 93: 737-745Crossref PubMed Scopus (107) Google Scholar, 2Ezekowitz R.A.B. Sastry K. Bailly P. Warner A. J. Exp. Med. 1990; 172: 1785-1794Crossref PubMed Scopus (420) Google Scholar). The collectins and the ficolins are soluble lectins, and these lectins function as opsonins or agglutinins for bacteria (3Weis W.I. Taylor M.E. Drickamer K. Immunol. Rev. 1998; 163: 19-34Crossref PubMed Scopus (902) Google Scholar, 4Kuhlman M. Joiner K. Ezekowitz R.A.B. J. Exp. Med. 1989; 169: 1733-1745Crossref PubMed Scopus (391) Google Scholar, 5Matsushita M. Endo Y. Taira S. Sato Y. Fujita T. Ichikawa N. Nakata M. Mizuochi T. J. Biol. Chem. 1996; 271: 2448-2454Abstract Full Text Full Text PDF PubMed Scopus (293) Google Scholar, 6Sugimoto R. Yae Y. Akaiwa M. Kitajima S. Shibata Y. Sato H. Hirata J. Okochi K. Izuhara K. Hamasaki N. J. Biol. Chem. 1998; 273: 20721-20727Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar). In addition, the mannose-binding lectin (MBL),1 a typical collectin, and ficolin/P32 form complexes with MBL-associated serine proteases in plasma. Binding of these complexes to targets activates the complement system, and complement activation induces opsonization of the targets by phagocytes and the target killing by formation of the membrane attack complex (7Matsushita M. Fujita T. J. Exp. Med. 1992; 176: 1497-1502Crossref PubMed Scopus (567) Google Scholar, 8Thiel S. Vorup-Jensen T. Stover C.M. Schwaeble W. Laursen S.B. Poulsen K. Willis A.C. Eggleton P. Hansen S. Holmskov U. Reid K.B.M. Jensenius J.C. Nature. 1997; 386: 506-510Crossref PubMed Scopus (757) Google Scholar, 9Matsushita M. Endo Y. Fujita T. J. Immunol. 2000; 164: 2281-2284Crossref PubMed Scopus (274) Google Scholar). This lectin-dependent complement activation pathway is named the lectin pathway and plays important roles in innate immunity (10Garred P. Madsen H.O. Hofmann B. Svejgaard A. Lancet. 1995; 346: 941-943Abstract PubMed Scopus (0) Google Scholar, 11Super M. Thiel S. Lu J. Levinsky R.J. Turner M.W. Lancet. 1989; 2: 1236-1239Abstract PubMed Scopus (468) Google Scholar). These biological defense lectins commonly have affinity to mannose orN-acetylglucosamine, and binding is sustained by Ca2+ (1Warr G.A. Biochem. Biophys. Res. Commun. 1980; 93: 737-745Crossref PubMed Scopus (107) Google Scholar, 2Ezekowitz R.A.B. Sastry K. Bailly P. Warner A. J. Exp. Med. 1990; 172: 1785-1794Crossref PubMed Scopus (420) Google Scholar, 3Weis W.I. Taylor M.E. Drickamer K. Immunol. Rev. 1998; 163: 19-34Crossref PubMed Scopus (902) Google Scholar, 4Kuhlman M. Joiner K. Ezekowitz R.A.B. J. Exp. Med. 1989; 169: 1733-1745Crossref PubMed Scopus (391) Google Scholar, 5Matsushita M. Endo Y. Taira S. Sato Y. Fujita T. Ichikawa N. Nakata M. Mizuochi T. J. Biol. Chem. 1996; 271: 2448-2454Abstract Full Text Full Text PDF PubMed Scopus (293) Google Scholar, 6Sugimoto R. Yae Y. Akaiwa M. Kitajima S. Shibata Y. Sato H. Hirata J. Okochi K. Izuhara K. Hamasaki N. J. Biol. Chem. 1998; 273: 20721-20727Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar), although the opposite results have been reported with regard to the Ca2+ dependence of ficolins (5Matsushita M. Endo Y. Taira S. Sato Y. Fujita T. Ichikawa N. Nakata M. Mizuochi T. J. Biol. Chem. 1996; 271: 2448-2454Abstract Full Text Full Text PDF PubMed Scopus (293) Google Scholar,6Sugimoto R. Yae Y. Akaiwa M. Kitajima S. Shibata Y. Sato H. Hirata J. Okochi K. Izuhara K. Hamasaki N. J. Biol. Chem. 1998; 273: 20721-20727Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar, 12Ohashi T. Erickson H.P. J. Biol. Chem. 1997; 272: 14220-14226Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar). On the other hand, animal lectins also include a group of lectins that have affinity to galactose. These galactose-binding lectins generally participate in cell differentiation (13Wells V. Mallucci L. Cell. 1991; 64: 91-97Abstract Full Text PDF PubMed Scopus (230) Google Scholar), apoptosis (14Perillo N.L. Pace K.E. Seilhamer J.J. Baum L.G. Nature. 1995; 378: 736-739Crossref PubMed Scopus (951) Google Scholar, 15Perillo N.L. Uittenbogaart C.H. Nguyen J.T. Baum L.G. J. Exp. Med. 1997; 185: 1851-1858Crossref PubMed Scopus (268) Google Scholar), recognition of tumor antigens (16Suzuki N. Yamamoto K. Toyoshima S. Osawa T. Irimura T. J. Immunol. 1996; 156: 128-135PubMed Google Scholar), and the uptake of galactosylated glycoproteins such as aged proteins (17Ozaki K. Ii M. Itoh N. Kawasaki T. J. Biol. Chem. 1992; 267: 9229-9235Abstract Full Text PDF PubMed Google Scholar, 18Spiess M. Lodish H.F. Proc. Natl. Acad. Sci. U. S. A. 1985; 82: 6465-6469Crossref PubMed Scopus (141) Google Scholar). However, there has been no report of galactose-binding lectins of mammals binding to bacterial components, although it has been reported in insects that theSarcophaga lectin functioned as a host defense lectin (19Takahashi H. Komano H. Kawaguchi N. Kitamura N. Nakanishi S. Natori S. J. Biol. Chem. 1985; 260: 12228-12233Abstract Full Text PDF PubMed Google Scholar,20Komano H. Natori S. Dev. Comp. Immunol. 1985; 9: 31-40Crossref PubMed Scopus (70) Google Scholar). The cell wall skeleton of Mycobacteria orNocardia acts as an immune response activator and is ingested by phagocytes (21Tsuji S. Matsumoto M. Takeuchi O. Akira S. Azuma I. Hayashi A. Toyoshima K. Seya T. Infect. Immun. 2000; 68: 6883-6890Crossref PubMed Scopus (365) Google Scholar, 22Izumi S. Ueda H. Okuhara M. Aoki H. Yamamura Y. Cancer Res. 1986; 46: 1960-1965PubMed Google Scholar, 23Izumi S. Hirai O. Hayashi K. Konishi Y. Okuhara M. Kohsaka M. Aoki H. Yamamura Y. Cancer Res. 1987; 47: 1785-1792PubMed Google Scholar). They contain little mannose but possess unique galactans consisting of galactofuranosyl residues, which mammals lack (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar). Thus, the specific recognition of these galactans is thought to be useful for their uptake into phagocytes. Although the cell wall skeleton of Mycobacterium contains no mannose, it is ingested by phagocytes (21Tsuji S. Matsumoto M. Takeuchi O. Akira S. Azuma I. Hayashi A. Toyoshima K. Seya T. Infect. Immun. 2000; 68: 6883-6890Crossref PubMed Scopus (365) Google Scholar). In the present study, we purified a novel human galactose-binding lectin and cloned its cDNA. We demonstrated that this lectin, human intelectin (hIntL), is a new type of Ca2+-dependent lectin that has affinity to galactofuranosyl residues and recognizes bacterial arabinogalactan. Galactose-Sepharose was produced by incubating epoxy-activated Sepharose 6B (Amersham Pharmacia Biotech) with galactose according to the manufacturer's instructions. Human placental tissue was obtained from Kobayashi Maternity Hospital, Osaka, Japan. The syncytial trophoblast layer was scraped from the fresh placenta and stored at −80 °C. The following purification procedures were carried out at 4 °C until column chromatography. Samples of ∼100 g of frozen tissue were homogenized with a Waring blender in 400 ml of 20 mm Tris-buffered saline (pH 7.5) (TBS) containing 10 mm EDTA, 1 mm phenylmethanesulfonyl fluoride, and 5 mm iodoacetamide. The homogenate was centrifuged at 30,000 × g for 15 min, and the pellet was then washed twice by homogenization and centrifugation with 200 ml of TBS. The washed pellet was resuspended in 400 ml of TBS containing 10 mm CHAPS, 1 mm phenylmethanesulfonyl fluoride, and 5 mm iodoacetamide, and then stirred for 18 h. The extract was isolated by centrifugation (30,000 × g, 1 h), CaCl2 was added to a final concentration of 10 mm, and then the extract applied to a galactose-Sepharose column (2.0 × 4.8 cm) at a flow rate of 1 ml/min at room temperature. The column was washed with TBS containing 10 mm CHAPS and 10 mm CaCl2, and the galactose-binding proteins were eluted with 10 mm Tris-HCl (pH 7.5) containing 10 mm CHAPS and 10 mm EDTA. The fractions showing absorbance at 280 nm were collected, concentrated with a vacuum concentrator, and dialyzed against 20 mmTris-HCl (pH 7.5) containing 5 mm CHAPS. The samples were resolved by SDS-PAGE under non-reducing conditions and transferred onto polyvinylidene difluoride (PVDF) membranes (Immobilon-P, Millipore). Several bands were cut out from the PVDF membranes stained with Coomassie Blue, and the strips were treated with 0.6 m HCl for 24 h at 25 °C (deblocking of formylation) and stored at −20 °C. These procedures were repeated five times and the strips were used for N-terminal amino acid sequence analysis. The purified hIntL was obtained by electro-elution from the 120-kDa band isolated by SDS-PAGE under non-reducing conditions. The N-terminal amino acid sequence of hIntL was used to search an expressed sequence tag data base, and the full-length cDNA sequence was deduced. PCR primers were designed on the basis of the predicted sequence (5′-GTG GAG GGA GGG AGT GAA GGA G-3′ and 5′-GAG TCA ATA TGA TTT ATT GTT TTC TCT TCT G-3′). cDNA was synthesized from human placental poly(A)+ RNA using oligo(dT) primer and SuperScript II RNase H reverse transcriptase (Life Technologies, Inc.). A single band was amplified by PCR using these primers and cDNA (30 cycles of 94 °C for 1 min, 56 °C for 1 min, and 72 °C for 1.5 min), and the PCR products were directly sequenced using several internal sequence primers. The primer regions were determined using a 5′ or 3′ rapid amplification of cDNA ends kit (Marathon cDNA amplification kit,CLONTECH). The MTN blot membranes containing 2 μg of poly(A)+ RNA from various human tissues (CLONTECH) were hybridized at 65 °C for 1 h in ExpressHyb hybridization solution (CLONTECH) with 32P-labeled cDNA of open reading frame sequences of hIntL. The blots were washed twice at 37 °C for 30 min in 0.1× SSC (1× SSC is 15 mm sodium citrate containing 150 mm NaCl) containing 0.1% SDS, and twice at 50 °C for 30 min in 3× SSC containing 0.1% SDS, followed by autoradiography. As a control for hybridization, the blots were rehybridized with a β-actin probe. The rabbit kidney cell line RK-13 were cultured in Eagle's minimum essential medium containing 10% fetal bovine serum (BioWhittaker, Walkersville, MD). cDNA encoding the open reading frame of hIntL was inserted into the mammalian cell expression vector pEF-BOS and transfected into RK-13 with LipofectAMINE 2000 (Life Technologies, Inc.). The transfected cells were incubated for 24 h at 37 °C, and the culture supernatants were collected. Approximately 1 mg of recombinant human intelectin (rhIntL) was purified from 700 ml of culture supernatant by galactose-Sepharose column chromatography. For preparation of antibodies against hIntL, hIntL-transfected RK-13 (5 × 107 cells) were injected with complete Freund's adjuvant (Difco Laboratories) into a rabbit every week (25Inoue N. Fukui A. Nomura M. Matsumoto M. Toyoshima K. Seya T. J. Immunol. 2001; 166: 424-431Crossref PubMed Scopus (26) Google Scholar). Three days after the fourth immunization, antiserum was collected, and the polyclonal antibodies were purified by precipitation with ammonium sulfate at 33% saturation. The specific anti-hIntL antibodies were isolated from the 33% ammonium sulfate fraction by affinity chromatography using Affi-Gel 10 (Bio-Rad) covalently bound to purified rhIntL. rhIntL (0.2 μg) or placental hIntL (0.1 μg) was dissolved in 20 μl of 200 mm Tris-HCl (pH 8.0) containing 0.1% SDS, 50 mm 2-mercaptoethanol, and 50 mmEDTA, and was denatured by boiling for 5 min. Each solution was divided into two aliquots to which 5 μl of 7.5% Nonidet P-40 and 13.8 μl of distilled water were added. 1.2 μl (0.3 unit) ofN-glycanase (Genzyme Co.) was added to each aliquot, and incubated for 18 h at 37 °C. These samples were resolved by SDS-PAGE under reducing conditions and transferred onto PVDF membranes. The hIntL treated with N-glycanase was detected by Western blotting. The blotted membranes were blocked with 5% nonfat milk and treated with rabbit anti-hIntL polyclonal antibodies (1.5 μg/ml). After washing, the blots were treated with horseradish peroxidase-conjugated goat anti-rabbit IgG and developed with ECL (Amersham Pharmacia Biotech). 2-Acetamido-2-deoxy-4-O-β-d-galactofuranosyl-d-glucopyranose (GalfG) was obtained from Toronto Research Chemicals Inc. (North York, Canada), and other saccharides were obtained from Pfanstiehl Laboratories Inc. (Waukegan, IL). Aliquots of 1 ml of culture supernatant of hIntL-transfected RK-13 were incubated at 4 °C for 2 h with 15 μl of galactose-Sepharose. The gels were washed five times with TBS containing 10 mm CaCl2, and incubated for 10 min with 20 μl of TBS, containing 10 mmEDTA or containing 10 mm CaCl2 and 100 mm various mono/disaccharides. The eluted samples were resolved by SDS-PAGE, and the gels were stained with Coomassie Blue. The stained gel images were imported into a computer with a scanner, and the densities of bands were measured using the image analysis software (NIH Image for Macintosh). The arabinogalactan of Nocardia rubra (a gift from Prof. I. Azuma, Hokkaido University, Sapporo, Japan) was covalently bound to universal binding 96-well plates (Costar) by UV cross-linking, and the plates were blocked with 2% bovine serum albumin. The plates were incubated for 3 h with 100 μl of the culture supernatant of hIntL-transfected RK-13, washed five times with TBS containing 0.05% Tween 20 and 2 mm CaCl2, and then incubated for 1 h with TBS containing 2 μg/ml specific anti-hIntL antibodies, 5% fetal bovine serum, and 2 mm CaCl2. After washing, the plates were incubated for 1 h with horseradish peroxidase-conjugated goat anti-rabbit IgG, developed witho-phenylenediamine and H2O2, and the absorbance at 492 nm was determined. To identify Ca2+-dependent galactose-binding protein, we employed an affinity chromatography method using galactose-Sepharose and protein analysis by SDS-PAGE, as described under "Experimental Procedures." Human placental tissue was used as a protein source, since it produces a variety of lectins and host defense proteins. The N-terminal amino acids of several proteins that showed affinity to galactose-Sepharose were sequenced and an unknown sequence was obtained from a 120-kDa protein under nonreducing conditions. As shown in Fig.1 A, this protein appeared as a 40-kDa single band under reducing conditions. Thus, this 120-kDa protein should be a homotrimer linked by disulfide bonds. We searched the expressed sequence tag data base for the N-terminal amino acid sequence of this 120-kDa protein (Fig. 2,solid underline), and the full-length sequence was deduced. Based on the deduced sequence, PCR primers were designed and used for reverse transcription-PCR using human placental mRNA. The PCR products were directly sequenced, and the primer regions were sequenced by 5′ or 3′ rapid amplification of cDNA ends. As shown in Fig. 2, the cDNA sequence of the 120-kDa protein (GenBank™ accession no. AB036706) contained a ATG translational initiation codon in accordance with Kozak's rule, a 939-base pair open reading frame encoding 313 amino acids, and a polyadenylation signal, AATAAA, followed by a poly(A) tail in the 3′-untranslated sequence. The transcript of this 120-kDa protein cDNA revealed a high degree of homology with the hypothetical protein mouse intelectin (26Komiya T. Tanigawa Y. Hirohashi S. Biochem. Biophys. Res. Commun. 1998; 251: 759-762Crossref PubMed Scopus (162) Google Scholar). Hence, we named this protein human intelectin (hIntL). The N-terminal amino acid of the purified placental hIntL was Thr-19, and analysis of the cDNA predicted the presence of a N-terminal hydrophobic signal sequence (amino acids 1–18). In addition, hIntL did not contain a significant transmembrane domain. These observations suggested that mature hIntL is a secretory protein. The predicted molecular size of the mature hIntL was calculated to be about 33 k, but the actual size of the placental hIntL was shown to be about 40 kDa on SDS-PAGE under reducing conditions (Fig.1 A). hIntL contained two potentialN-glycosylation sites at Asn-154 and Asn-163 (Fig. 2). Following N-glycanase treatment, the molecular sizes of both placental and recombinant hIntL were decreased to 34 kDa, which nearly matched the predicted molecular size (Fig. 1 B). Thus, the difference of molecular size appeared to be largely due to the presence of N-linked oligosaccharides. As shown in Fig. 3, the amino acid sequence of hIntL was 81.5% identical to mouse intelectin (26Komiya T. Tanigawa Y. Hirohashi S. Biochem. Biophys. Res. Commun. 1998; 251: 759-762Crossref PubMed Scopus (162) Google Scholar) and 59.4% identical to Xenopus laevis cortical granule lectin (GenBank accession no. X82626), the sequence of which was the same as that of X. laevis oocyte lectin (27Lee J.K. Buckhaults P. Wilkes C. Teilhet M. King M.L. Moremen K.W. Pierce M. Glycobiology. 1997; 7: 367-372Crossref PubMed Scopus (57) Google Scholar) except for 5 amino acids. The highly homologous region (residues 37–313) of these proteins showed partial homology with the C-terminal part ofLampetra japonica lamprey serum lectin (GenBank™ accession no. AB055981) (residues 53–333, 58.1%) or Halocynthia roretzi ascidian galactose-specific lectin (28Abe Y. Tokuda M. Ishimoto R. Azumi K. Yokosawa H. Eur. J. Biochem. 1999; 261: 33-39Crossref PubMed Scopus (54) Google Scholar) (residues 69–348, 43.0%). Thus, hIntL is likely to be a human homologue of these molecules. Placental hIntL showed Ca2+-dependent saccharide binding activity as similarly to cortical granule lectin/oocyte lectin, but hIntL did not contain the carbohydrate recognition domain, which is a characteristic of C-type lectins. Moreover, hIntL did not contain any known functional domains, except the region from Pro-38 to Val-82 that was similar to part of the fibrinogen domain. A human genome sequence data base search indicated that the human intelectin gene (Itln) was contained in a chromosome 1 clone (GenBank™ accession no. AL354714). The exon-intron junctions were determined by comparing the genomic sequence with that of hIntL cDNA. As shown in Fig. 4, human Itln was split into 8 exons and the open reading frame was located within exons 2–8. The nucleotide sequences of all exon-intron junctions fulfilled the GT-AG rule. Analysis of some sequences (GenBank™ accession nos.AL162592, AL354714, AC068728, and AL121985) indicated thatItln was located between junction adhesion molecule and Ly-9, which was followed by CD48. Since these molecules were reported to be located on chromosome 1q21.3–22 (29Naik U.P. Naik M.U. Eckfeld K. Martin-DeLeon P. Spychala J. J. Cell Sci. 2001; 114: 539-547PubMed Google Scholar, 30Sandrin M.S. Henning M.M. Lo M.F. Baker E. Sutherland G.R. McKenzie I.F. Immunogenetics. 1996; 43: 13-19PubMed Google Scholar, 31Staunton D.E. Fisher R.C. LeBeau M.M. Lawrence J.B. Barton D.E. Francke U. Dustin M. Thorley-Lawson D.A. J. Exp. Med. 1989; 169: 1087-1099Crossref PubMed Scopus (57) Google Scholar), Itln would be located on chromosome 1q21.3–22. Northern blotting analysis of hIntL mRNA showed that the transcript size was 1.35 kilobase pairs (Fig. 5). It has been reported that mouse intelectin is specifically expressed in the small intestine (26Komiya T. Tanigawa Y. Hirohashi S. Biochem. Biophys. Res. Commun. 1998; 251: 759-762Crossref PubMed Scopus (162) Google Scholar), but hIntL mRNA was also expressed in the heart, small intestine, colon, and thymus, and to a lesser degree in the uterus and spleen. hIntL was purified and cloned from the placenta, but placental hIntL mRNA was barely detected on Northern blotting. Therefore, hIntL appeared not to be a major product in the placenta. For biochemical analysis of hIntL, the hIntL cDNA was cloned into the vector pEF-BOS and transfected into the rabbit kidney cell line RK-13. The recombinant hIntL (rhIntL) was affinity-purified from the cell lysates and the culture supernatants with galactose-Sepharose in the presence of Ca2+ and was eluted with EDTA. As shown in Fig.6, rhIntL was purified from both lysates and culture supernatants of hIntL-transfected cells, but not those of mock cells transfected with the empty vector. The rhIntL was the same molecular size as native placental hIntL under non-reducing or reducing conditions (Fig. 1 A). Moreover, the molecular sizes of rhIntL and native placental hIntL were similarly decreased byN-glycanase treatment (Fig. 1 B). These results indicated that rhIntL has Ca2+-dependent galactose binding activity, homotrimeric structure via disulfide bonds, and N-linked oligosaccharides, similarly to the native placental hIntL. The rhIntL was purified as a soluble protein from the culture supernatant of hIntL-transfected cells, but the purified and concentrated rhIntL tended to become insoluble and inactive in the presence of Ca2+ (data not shown). The rhIntL that was absorbed to galactose-Sepharose was eluted by various mono/disaccharides in the presence of Ca2+ (Fig.7 A). The absorbed rhIntL was completely eluted by 10 mm EDTA (data not shown). Approximately 50% of the absorbed rhIntL was eluted by buffers containing 100 mm galactose, 100 mm N-acetylgalactosamine, or 100 mm fructose. At the same concentration of monosaccharide, rhIntL was hardly eluted by other hexoses or derived saccharides (mannose, glucose,N-acetylmannosamine, N-acetylglucosamine, sorbose, d-fucose, l-fucose,l-rhamnose, and 2-deoxy-d-glucose). rhIntL was not eluted by cellobiose, maltose, trehalose, sucrose, or raffinose (data not shown). It has been reported that X. laeviscortical granule lectin/oocyte lectin has affinity to melibiose and lactose that contain an α- or β-galactopyranosyl residue, respectively (27Lee J.K. Buckhaults P. Wilkes C. Teilhet M. King M.L. Moremen K.W. Pierce M. Glycobiology. 1997; 7: 367-372Crossref PubMed Scopus (57) Google Scholar, 32Roberson M.M. Barondes S.H. J. Biol. Chem. 1982; 257: 7520-7524Abstract Full Text PDF PubMed Google Scholar, 33Nishihara T. Wyrick R.E. Working P.K. Chen Y.H. Hedrick J.L. Biochemistry. 1986; 25: 6013-6020Crossref PubMed Scopus (80) Google Scholar). hIntL is likely to be a human homologue of cortical granule lectin/oocyte lectin because hIntL showed a high degree of homology to X. laevis cortical granule lectin (Fig. 3). Thus, rhIntL was expected to have affinity to melibiose and lactose. However, rhIntL was not eluted from galactose-Sepharose by melibiose or lactose. Surprisingly, rhIntL was effectively eluted byd-pentoses. d-Xylose, d-ribose, and 2-deoxy-d-ribose completely eluted the absorbed rhIntL, andd-lyxose eluted 50% of the rhIntL.d-/l-Arabinose and l-xylose eluted <20% of the rhIntL. hIntL had affinity to galactose, but not melibiose or lactose, which contain a galactopyranosyl residue. These results suggested that hIntL recognized furanosides more effectively than pyranosides. Moreover, the high affinities of hIntL to pentoses supported this suggestion, since pentoses more effectively form furanoside rings than hexoses in the aqueous phase. As shown in Fig. 7 B, rhIntL absorbed to galactose-Sepharose was completely eluted by 100 mm GalfG containing a fixed galactofuranosyl residue, similarly to EDTA ord-ribose. On the other hand, <30% of the absorbed rhIntL was eluted by 100 mm2-acetamido-2-deoxy-4-O-β-d-galactopyranosyl-d-glucopyranose, the galactopyranoside isomer of GalfG, and rhIntL was hardly eluted by lactose or melibiose each containing a fixed galactopyranosyl residue. These results suggested that hIntL is a Ca2+-dependent lectin that has affinity to furanosides such as the galactofuranosyl residue. Bacterial carbohydrate chains such as arabinogalactan of cell wall ofMycobacterium or Nocardia contain galactofuranosyl residues (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar). We investigated whether hIntL bound to the arabinogalactan purified from N. rubra. As shown in Fig.8, rhIntL bound to the arabinogalactan ofN. rubra covalently fixed on the plate. The binding was completely inhibited by EDTA, d-ribose,d-galactose, and d-arabinose, but notd-glucose. Table I lists the saccharides tested as inhibitors of rhIntL binding to arabinogalactan and the concentration of each required to give 50% inhibition of binding. Pentoses (d-ribose, d-xylose,d-lyxose, and d-arabinose) andd-galactose inhibited the binding of rhIntL to arabinogalactan more effectively than hexoses (d-mannose and d-glucose). Moreover, GalfG containing a galactofuranosyl residue efficiently inhibited the binding of rhIntL, but not lactose or melibiose each containing a galactopyranosyl residue. These results suggested that the natural ligands of hIntL are bacterial carbohydrate chains such as arabinogalactan of N. rubra, and that hIntL recognizes the furanosyl residues contained in these oligosaccharides in a Ca2+-dependent manner.Table ISaccharide concentrations for 50% inhibition of binding of rhIntL to arabinogalactan of N. rubraSaccharideConcentration for 50% inhibitionmmd-Ribose 200Lactose>200Melibiose>20080% culture supernatants of hIntL-transfected cells containing various concentrations of saccharides were added to the plate binding arabinogalactan of N. rubra, and the amount of rhIntL was measured by enzyme-linked immunosorbent assay as described under "Experimental Procedures." Each saccharide was used at the concentrations of 5–200 mm. The inhibition curves were obtained from the mean of duplicate determinations, and the saccharide concentrations for 50% inhibition of rhIntL binding were calculated from these results. This experiment was repeated twice with similar results. Open table in a new tab 80% culture supernatants of hIntL-transfected cells containing various concentrations of saccharides were added to the plate binding arabinogalactan of N. rubra, and the amount of rhIntL was measured by enzyme-linked immunosorbent assay as described under "Experimental Procedures." Each saccharide was used at the concentrations of 5–200 mm. The inhibition curves were obtained from the mean of duplicate determinations, and the saccharide concentrations for 50% inhibition of rhIntL binding were calculated from these results. This experiment was repeated twice with similar results. In host defense, the recognition of specific structures of pathogens is important for induction of immune responses. Some animal lectins induce responses of the innate immune system by recognition of the carbohydrate chains that host cells lack. In this study, we demonstrated that hIntL is a new type of lectin that recognizes galactofuranosyl residues, which hardly exist in mammals. This is the first report of an animal lectin recognizing galactofuranose. Galactofuranose is contained in the carbohydrate chains of various microorganisms (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar, 34Abeygunawardana C. Bush C.A. Cisar J.O. Biochemistry. 1991; 30: 8568-8577Crossref PubMed Scopus (48) Google Scholar, 35de Lederkremer R.M. Colli W. Glycobiology. 1995; 5: 547-552Crossref PubMed Scopus (198) Google Scholar, 36Suzuki E. Toledo M.S. Takahashi H.K. Straus A.H. Glycobiology. 1997; 7: 463-468Crossref PubMed Scopus (50) Google Scholar, 37McConville M.J. Homans S.W. Thomas-Oates J.E. Dell A. Bacic A. J. Biol. Chem. 1990; 265: 7385-7394Abstract Full Text PDF PubMed Google Scholar). The cell wall skeleton ofMycobacteria or Nocardia has biological activities as an immune activator (21Tsuji S. Matsumoto M. Takeuchi O. Akira S. Azuma I. Hayashi A. Toyoshima K. Seya T. Infect. Immun. 2000; 68: 6883-6890Crossref PubMed Scopus (365) Google Scholar, 22Izumi S. Ueda H. Okuhara M. Aoki H. Yamamura Y. Cancer Res. 1986; 46: 1960-1965PubMed Google Scholar, 23Izumi S. Hirai O. Hayashi K. Konishi Y. Okuhara M. Kohsaka M. Aoki H. Yamamura Y. Cancer Res. 1987; 47: 1785-1792PubMed Google Scholar) and contains galactofuranose as a constituent of arabinogalactan (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar). In a previous study, we found that galactose effectively inhibited phagocytosis of the cell wall skeleton of Mycobacterium bovis by dendritic cells, but not mannose orN-acetylglucosamine. 2S. Tsuji, J. Uehori, M. Matsumoto, and T. Seya, unpublished data. This result allowed us to presume existence of a receptor or an opsonin that recognizes galactofuranosyl residues. hIntL is different from the lectins that induced phagocytosis of the cell wall skeleton of M. bovis because the results of saccharide affinity of hIntL were not similar to the results of phagocytosis inhibition. However, hIntL bound to the arabinogalactan that was purified as a pathogen constituent fromN. rubra. Thus, hIntL would bind to various pathogens containing galactofuranosyl residues such as Nocardia,Mycobacteria (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar), Streptococcus (34Abeygunawardana C. Bush C.A. Cisar J.O. Biochemistry. 1991; 30: 8568-8577Crossref PubMed Scopus (48) Google Scholar), andLeishmania and Trypanosoma (35de Lederkremer R.M. Colli W. Glycobiology. 1995; 5: 547-552Crossref PubMed Scopus (198) Google Scholar, 36Suzuki E. Toledo M.S. Takahashi H.K. Straus A.H. Glycobiology. 1997; 7: 463-468Crossref PubMed Scopus (50) Google Scholar, 37McConville M.J. Homans S.W. Thomas-Oates J.E. Dell A. Bacic A. J. Biol. Chem. 1990; 265: 7385-7394Abstract Full Text PDF PubMed Google Scholar). The soluble biological defense lectins, MBL and ficolin/P35, function as agglutinins. The agglutination activity of lectin is important to lock the targets into a point. hIntL also forms a polyvalent structure that is indispensable for agglutination. Thus, hIntL was suggested to be able to agglutinate the ligands. However, rhIntL did not exert agglutination activities toward some bacteria or erythrocytes (data not shown). This may have been due to requirement of non-covalent multimer structures larger than a trimer unit for agglutination. It has been reported that the native X. laevis cortical granule lectin/oocyte lectin, a homologue of hIntL, exists as multimers of approximately 500 kDa and agglutinated trypsinized and glutaraldehyde-fixed rabbit erythrocytes (32Roberson M.M. Barondes S.H. J. Biol. Chem. 1982; 257: 7520-7524Abstract Full Text PDF PubMed Google Scholar) or natural ligands (33Nishihara T. Wyrick R.E. Working P.K. Chen Y.H. Hedrick J.L. Biochemistry. 1986; 25: 6013-6020Crossref PubMed Scopus (80) Google Scholar). Thus, hIntL may exhibit agglutination activity with its natural ligands only when it forms the native multimer structure. MBL and ficolin/P35 also function as opsonins and complement activating factors (4Kuhlman M. Joiner K. Ezekowitz R.A.B. J. Exp. Med. 1989; 169: 1733-1745Crossref PubMed Scopus (391) Google Scholar, 5Matsushita M. Endo Y. Taira S. Sato Y. Fujita T. Ichikawa N. Nakata M. Mizuochi T. J. Biol. Chem. 1996; 271: 2448-2454Abstract Full Text Full Text PDF PubMed Scopus (293) Google Scholar, 6Sugimoto R. Yae Y. Akaiwa M. Kitajima S. Shibata Y. Sato H. Hirata J. Okochi K. Izuhara K. Hamasaki N. J. Biol. Chem. 1998; 273: 20721-20727Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar, 7Matsushita M. Fujita T. J. Exp. Med. 1992; 176: 1497-1502Crossref PubMed Scopus (567) Google Scholar, 8Thiel S. Vorup-Jensen T. Stover C.M. Schwaeble W. Laursen S.B. Poulsen K. Willis A.C. Eggleton P. Hansen S. Holmskov U. Reid K.B.M. Jensenius J.C. Nature. 1997; 386: 506-510Crossref PubMed Scopus (757) Google Scholar, 9Matsushita M. Endo Y. Fujita T. J. Immunol. 2000; 164: 2281-2284Crossref PubMed Scopus (274) Google Scholar). Although these activities are sustained by the collagen-like domains of these molecules (3Weis W.I. Taylor M.E. Drickamer K. Immunol. Rev. 1998; 163: 19-34Crossref PubMed Scopus (902) Google Scholar, 9Matsushita M. Endo Y. Fujita T. J. Immunol. 2000; 164: 2281-2284Crossref PubMed Scopus (274) Google Scholar), hIntL has no collagen-like domain. This suggested that hIntL does not function as an opsonin via C1q receptor and/or a complement activating factor forming complexes with MBL-associated serine proteases. However, it has been reported that ascidian galactose-specific plasma lectin, an ascidian homologue of hIntL, enhanced phagocytosis of its target, sheep erythrocytes (28Abe Y. Tokuda M. Ishimoto R. Azumi K. Yokosawa H. Eur. J. Biochem. 1999; 261: 33-39Crossref PubMed Scopus (54) Google Scholar). Thus, hIntL may be able to function as an opsonin via an unknown receptor. hIntL is a Ca2+-dependent galactose-binding lectin, but there is no carbohydrate recognition domain, which is a characteristic of C-type lectins. In mammals, the galactose-binding C-type lectins, a macrophage lectin or asialoglycoprotein receptors, bind to asialoglycoproteins (16Suzuki N. Yamamoto K. Toyoshima S. Osawa T. Irimura T. J. Immunol. 1996; 156: 128-135PubMed Google Scholar, 17Ozaki K. Ii M. Itoh N. Kawasaki T. J. Biol. Chem. 1992; 267: 9229-9235Abstract Full Text PDF PubMed Google Scholar). However, the binding affinity of hIntL to asialoglycoproteins was weak (data not shown). This was thought to be because the galactosyl residues of asialoglycoproteins are galactopyranosides although hIntL has affinity to the galactofuranosyl residue. As the X. laevis oocyte lectin, a homologue of hIntL, has been shown to participate in formation of the fertilization envelope that blocks sperm entry (38Quill T.A. Hedrick J.L. Arch. Biochem. Biophys. 1996; 333: 326-332Crossref PubMed Scopus (43) Google Scholar, 39Gerton G.L. Hedrick J.L. Dev. Biol. 1986; 116: 1-7Crossref PubMed Scopus (56) Google Scholar), hIntL may also participate in fertilization. However, hIntL was also expressed on various tissues other than oocytes, and it has been reported that the other homologues are also present in various tissues (26Komiya T. Tanigawa Y. Hirohashi S. Biochem. Biophys. Res. Commun. 1998; 251: 759-762Crossref PubMed Scopus (162) Google Scholar, 28Abe Y. Tokuda M. Ishimoto R. Azumi K. Yokosawa H. Eur. J. Biochem. 1999; 261: 33-39Crossref PubMed Scopus (54) Google Scholar, 40Roberson M.M. Barondes S.H. J. Cell Biol. 1983; 97: 1875-1881Crossref PubMed Scopus (36) Google Scholar). Thus, hIntL and its homologues may not only participate in formation of the fertilization envelope but also have other physiological functions. The result of Northern blotting analysis suggested that hIntL was plentifully expressed in the heart (Fig. 5). It is known that viridans streptococci invading into blood attack heart and cause subacute infectious endocarditis, and the cell surface polysaccharide ofStreptococcus oralis, a viridans streptococci, contains galactofuranosyl residues (34Abeygunawardana C. Bush C.A. Cisar J.O. Biochemistry. 1991; 30: 8568-8577Crossref PubMed Scopus (48) Google Scholar). Thus, hIntL may function in heart as a defense protein against these pathogens. It is also possible that hIntL may be an acceptor molecule of viridans streptococci in heart. These possibilities could be tested using galactofuranoside-modifiedStreptococcus. Mouse intelectin was reported as the first intelectin homologue in mammals, and the gene was specifically expressed in intestinal paneth cells (26Komiya T. Tanigawa Y. Hirohashi S. Biochem. Biophys. Res. Commun. 1998; 251: 759-762Crossref PubMed Scopus (162) Google Scholar). However, the biochemical and physiological functions were not analyzed in the study because the recombinant mouse intelectin produced in Escherichia coli had toxic activity in bacteria. Although the ascidian galactose-specific plasma lectin, an ascidian homologue of hIntL, was reported as a lectin that bound to galactose and enhanced the phagocytosis of sheep erythrocytes (28Abe Y. Tokuda M. Ishimoto R. Azumi K. Yokosawa H. Eur. J. Biochem. 1999; 261: 33-39Crossref PubMed Scopus (54) Google Scholar), it has not been clarified whether the lectin bound to constituents of pathogens. In the present study, we demonstrated that hIntL has affinity to galactofuranosyl residues and recognizes bacterial carbohydrate chains. The galactofuranosyl residue is a constituent of pathogens and a dominant immunogen (24Daffe M. McNeil M. Brennan P.J. Carbohydr. Res. 1993; 249: 383-398Crossref PubMed Scopus (84) Google Scholar, 41Notermans S. Veeneman G.H. van Zuylen C.W. Hoogerhout P. van Boom J.H. Mol. Immunol. 1988; 25: 975-979Crossref PubMed Scopus (91) Google Scholar). Thus, our results suggested that a biological function of hIntL is the specific recognition of pathogens and bacterial components containing galactofuranosyl residues. We are grateful to Drs. H. Koyama, M. Tatsuta and T. Masaoka (Osaka Medical Center) for support of this work. Thanks are also due to Dr. I. Azuma (Hakodate National College of Technology) and A. Hayashe (Osaka Medical Center) for valuable discussions.