Purification and molecular characterization of a novel 16-kDa galectin from the nematode Caenorhabditis elegans.

In our previous study (Hirabayashi, J., Satoh, M., Ohyama, Y., and Kasai, K. (1992) J. Biochem. (Tokyo) 111, 553-555), two beta-galactoside-binding lectins (apparent subunit molecular masses, 16 and 32 kDa, respectively) were identified in the nematode Caenorhabditis elegans. The subsequent study revealed that the 32-kDa lectin is a member of the galectin family. Since the 32-kDa galectin was found to consist of two homologous domains (approximately 16 kDa), 16-kDa lectin was thought to be a degradation product of the 32-kDa galectin. To clarify this, the 16-kDa lectin was purified by an improved procedure employing extraction with a calcium-supplemented buffer. The purified 16-kDa lectin was found to exist as a dimer (approximately 30 kDa) and showed hemagglutinating activity toward trypsinized rabbit erythrocytes, which was inhibited by lactose. Almost the whole sequence of the 16-kDa polypeptide (approximately 95%, 135 amino acids) was determined after digestion with various proteases. Based on the obtained information, a full-length cDNA was cloned with the aid of RNA-polymerase chain reaction. The clone encoded 146 amino acids including initiator methionine (calculated molecular mass, 15,928 Da). Based on these results, it was concluded that the 16-kDa lectin is a novel member of the galectin family, but not a degradation product of the 32-kDa galectin as had previously thought. However, the 16-kDa galectin showed relatively low sequence similarities to both the N-terminal and the C-terminal domains of the 32-kDa galectin (28% and 27% identities, respectively) and to various vertebrate galectins (14-27%). Nonetheless, all of the critical amino acids involved in carbohydrate binding were conserved. These observations suggest that, in spite of phylogenic distance between nematodes and vertebrates, both the 16-kDa and 32-kDa nematode isolectins have conserved essentially the same function(s) as those of vertebrate galectins, probably through recognition of a key disaccharide moiety, "N-acetyllactosamine."

C. elegans, belonging taxonomically to the pseudocoelomates of Protostomia, has a simple body structure (2 mm in length) consisting of only about 1,000 cells. Nevertheless, it develops a primitive nerve-muscular system. The worm is experimentally useful, because it is transparent and hermaphroditic, and has a short generation time (3 days). Such features are apparently advantageous for studies on anatomy, development, fertilization, and, in particular, genetics (Coulson et al., 1988), and the nematode has become an important model animal since the initial proposal by Brenner in 1965 (reviewed by Wood (1988)).
From the viewpoint of protein architecture, galectins can be classified into three structural types (Hirabayashi et al., 1992b;Hirabayashi and Kasai, 1993); that is (i) proto type (subunit molecular mass, 14,000 -16,000 Da), (ii) chimera type (29,000 -35,000 Da), and (iii) tandem-repeat type (32,000 -36,000). The proto type, having a single CRD, includes a number of small galectins of both mammalian and non-mammalian species. Galectin-3, which has been found only in mammals, is the sole member of the chimera type that consists of an N-terminal proline-, glycine-, and tyrosine-rich repetitive domain and a C-terminal CRD that shows homology to other members of the galectins (reviewed by Anderson and Wang (1992) and Hughes (1994)). Tandem-repeat type was firstly identified in C. elegans (32-kDa galectin; Hirabayashi et al., 1992b), but more recently similar galectins were also found in mammals. Rat (Oda et al., 1993;Tardy et al., 1995a) and porcine (Chiu et al., 1994) galectin-4, rat galectin-6 (Gitt et al., 1995a), and rat galectin-8 (Hadari et al., 1995) are categorized into this type. However, the latter mammalian galectins are not complete homologues of the nematode 32-kDa galectin, since sequence homology between them is relatively low, and the mammalian tandemrepeat type galectins (both galectin-4 and -8) have an extra linker domain between the two repeated CRDs. The sequence of a cDNA for a probable homologue in Onchocerca volvulus to the C. elegans 32-kDa galectin has recently been deposited in the GenBank™ data base (accession number U04046). Among these galectin CRDs (ϳ14 kDa), several amino acid residues are strictly conserved. All of them are located in a central region, which has been shown to be encoded by a single exon (Ohyama and Kasai, 1988;Gitt and Barondes, 1991;Gitt et al., 1992;Gritzmacher et al., 1992). Many of these conserved residues, i.e. and Arg-73 (residue numbers are those of human galectin-1) were shown to be critical for the saccharide binding by our systematic mutagenesis studies (Hirabayashi andKasai, 1991, 1994).
In recent x-ray crystallographic studies, all of them were proved to be involved in the interaction between lactose and human galectin-2 (Lobsanov et al., 1993), and in that between N-acetyllactosamine and human  or bovine galectin-1 (Bourne et al., 1994).
In our previous study (Hirabayashi et al., 1992a), we found that the nematode contained two soluble, metal-independent ␤-galactoside-specific lectins (apparent subunit molecular masses estimated by SDS-PAGE, 16,000 Da and 32,000 Da, respectively (designated hereafter as nematode 16-kDa and 32-kDa lectins, respectively). They were purified by a procedure essentially similar to that applied to galectin-1 purification (Hirabayashi and Kasai, 1984;Hirabayashi et al., 1987); briefly, the nematode lectins were extracted with lactose-containing MEPBS (20 mM lactose, 4 mM ␤-mercaptoethanol, 2 mM EDTA, 20 mM sodium phosphate, pH 7.2, 150 mM NaCl) and were adsorbed on asialofetuin-agarose. Although cDNA cloning revealed that the 32-kDa protein is the first invertebrate member of the galectin family, it remains to be determined whether the 16-kDa lectin also belongs to the same protein family or not. Since preliminary amino acid analysis showed that the 16-kDa lectin is very similar to 32-kDa galectin, it seemed likely that the 16-kDa protein was derived from the 32-kDa species through limited proteolysis during preparation, as in the case of galectin-4 (Leffler et al., 1989;Oda et al., 1993;Tardy et al., 1995a). On the other hand, antiserum raised against nematode 32-kDa galectin reacted only poorly with 16-kDa lectin. A major difficulty in answering the above question, however, was that the 16-kDa lectin was a minor component; the yields of 16-kDa and 32-kDa lectins in the previous purification procedure were approximately 20 -30 and 200 g, respectively, from 10 g of the worm. After trial-and-error studies, we found that the use of calcium-supplemented buffer (Ca-TBS) for extraction instead of the previously used MEPBS greatly improved the yield of 16-kDa lectin. In this paper, we describe an improved purification of the 16-kDa lectin, its protein-chemical characterization, and cDNA cloning, as well as preliminary identification of endogenous glycoprotein receptors for the lectin. It was concluded that the 16-kDa lectin is a novel member of the galectin family in the nematode C. elegans, belonging to the proto type category.

EXPERIMENTAL PROCEDURES
Cultivation of C. elegans and Purification of 16-kDa Lectin-Although the nematode C. elegans was cultivated previously in a liquid medium (S-medium) using fresh Escherichia coli (OP50 strain) as feed (Wood, 1988), a more efficient and convenient procedure was developed: briefly, sterile worms were added to an appropriate volume of S-medium (e.g. 700 ml to each 3-liter flask) supplemented with 2 g of the Micrococcus lysodeikticus powder per liter, and proliferated with continuous vigorous shaking (e.g. at 200 rpm, at 20°C with a Taitek rotary shaker NR-20), until OD 600 was lowered below 1.0. Then another 2 g of M. lysodeikticus powder/liter of culture was added to boost the proliferation. The worms were harvested and purified as described previously (Wood, 1988). Thus prepared worms, 3-6 g from a liter of culture, showed no apparent abnormality compared with those obtained by the conventional procedure using fresh E. coli OP50.
Two ␤-galactoside-binding lectins (16 and 32 kDa) were purified essentially as described previously (Hirabayashi et al., 1992a). However, to improve the purification yield of the 16-kDa lectin, Ca-TBS (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, supplemented with 5 mM CaCl2) was used instead of MEPBS. After the worms were disrupted by sonication in 5 volumes of Ca-TBS, most of the soluble proteins were removed by centrifugation (15,000 rpm, 4°C, 25 min), and lectins were extracted with lactose/Ca-TBS (Ca-TBS containing 20 mM lactose) by shaking the suspension for 30 min at 4°C. The derived supernatant containing lectins (sup-2) was dialyzed extensively against Ca-TBS to remove lactose, and then applied to an asialofetuin-agarose column (10 ml volume, approximately 9 mg of fetuin was immobilized/ml of gel; prepared according to De Waard et al. (1976)). After extensive washing of the column with Ca-TBS, the bound lectins were eluted with lactose/ Ca-TBS. Protein was determined by the method of Bradford (1976).
Extraction of 16-kDa Lectin with Various Saccharides-Worms (10 g, wet weight) were disrupted by sonication with 25 ml of cold Ca-TBS, and subjected to centrifugation (15,000 rpm, 4°C, 20 min). After removal of the supernatant solution (sup-1), precipitated material (including both 16-kDa and 32-kDa lectins) was divided into 10 portions. Thus divided pellet fractions were extracted with various saccharide solutions (1 ml each of 0.1 M galactose, glucose, mannose, L-fucose, maltose, lactose, melibiose, or sucrose). Portions of these extracts (sup-2) containing 20 g of protein, as well as sup-1, were subjected to SDS-PAGE (14% gel), and lectin bands were detected by means of Western blotting analysis using specific antisera directed to either nematode 16-kDa lectin (prepared in this work, see below) or 32-kDa lectin (Hirabayashi et al., 1992b) as the first antibodies, and horseradish peroxidase-conjugated goat anti-rabbit IgG antiserum as the second antibody (Seikagaku Kogyo, Tokyo, Japan). For peroxidase detection, a colorimetric Wako POD Immunostain Set (Osaka, Japan) was used.
Antiserum Production in Rabbits-A rabbit was immunized with affinity-purified nematode 16-kDa lectin (purity, approximately 90%) by repeated injections (50 -100 g each for 10 months at 2-3-week intervals; in total, approximately 1 mg) with Freund's complete adjuvant. Almost monospecific antiserum was obtained, which can detect less than 10 ng of 16-kDa lectin in conventional Western blotting analysis as described above.
Molecular Weight Estimation-Subunit molecular size under denaturing conditions was estimated by SDS-PAGE according to the method of Laemmli, and protein was detected with a Wako Silver Stain Kit prepared according to Morrissey (1981). Oligomeric structure was estimated by high-performance gel filtration analysis with a Tosoh TSK GS-2000SW-XL column (exclusion size determined for dextran, 100,000 Da), as described previously (Hirabayashi and Kasai, 1991). PCR was performed on a Perkin-Elmer thermal cycler 4800 using the following three-step cycles (30 cycles); 92°C for 1 min (denaturation), 42°C for 1 min (annealing), and 72°C for 1 min (extension). After the final cycle DNA was further extended at 72°C for 10 min. For analysis of the derived DNA fragments, 2% Nusieve agarose gel (FMC Co. Ltd.) was used. An amplified fragment derived by using primers F1 and R3 (0.27 kbp) was sequenced by the dye-termination cycle sequencing method with an Applied Biosystems DNA sequencer 373S according to the manufacturer's instructions.
Screening a ZAP cDNA Library and Nucleotide Sequence Determination-A ZAP cDNA library of C. elegans (mixed stages) was a generous gift from Dr. Waterstone of Washington University (Brastead and Waterstone, 1989). The library was screened with a 400-bp PCR fragment, which was derived by using F1 and R1 primers and found to encode almost the entire region of the coding sequence for the 16-kDa lectin. The probe was labeled by using a random-octamer Megaprime system (Amersham) with [␣-32 P]dCTP (110 TBq/mmol). Plaques were formed on LB plates supplemented with tetracycline (12.5 g/ml) for 16 h at 37°C (105 plaque-forming units/15-cm plate) after infecting E. coli XLI-Blue cells. Plaque hybridization was performed essentially according to Maniatis et al. (Sambrook et al., 1989) except that Rapid Hybridization Buffer (Amersham) was used for both prehybridization and hybridization. After three rounds of screening, a single phage clone, named 7B1, was obtained, which showed a stable hybridization signal. An inserted fragment (0.55 kbp) was automatically subcloned into Bluescript(SK Ϫ ) as described in a manual provided by the manufacturer. The derived plasmid was proliferated in E. coli XLI-Blue in the presence of ampicillin (50 g/ml) and sequenced as described above by using conventional forward and reverse M13 primers.
Expression of Recombinant Lectin Fused with ␤-Galactosidase ␣ Peptide (recN16␣)-Since the above derived cDNA, 7B1, was cloned under the control of lac promoter together with a part of the structural gene of E. coli ␤-galactosidase (␣-peptide), recombinant 16-kDa lectin was directly expressed as a fusion protein with ␤-galactosidase ␣-peptide (designated here recN16␣1). E. coli XLI-Blue clone containing the lectin cDNA was proliferated in 1-3 liters of LB medium supplemented with antibiotics, and lectin production was induced by adding isopropyl-␤thiogalactoside to give a final concentration of 0.1 mM for 2 h, as described previously (Hirabayashi et al., 1992b). E. coli cells were disrupted by sonication in MEPBS, and, after centrifugation (12,000 rpm, 4°C, 25 min) of the extract, the recombinant lectin contained in the supernatant solution was adsorbed on an asialofetuin-agarose column (10 ml bed volume, 9 mg of fetuin/ml of gel). The adsorbed protein was eluted with MEPBS containing 20 mM lactose.
Isolation of Lectin-binding Glycoproteins on a RecN16␣-agarose Column-For the preparation of recN16␣-agarose, 24 mg of recN16␣ (derived from 8 liter of bacterial culture) in 20 ml of lactose/Ca-TBS was dialyzed twice against 0.5 ϫ coupling buffer (1 ϫ: 0.1 M NaHCO 3 , 0.5 M NaCl, 20 mM lactose), concentrated to 10 ml by lyophilization, and then mixed with 5 ml of prewashed CNBr-activated Sepharose 4B (Pharmacia Biotech Inc.). After the coupling reaction for 2 h at 23°C, an excess of glycine was added to block the remaining activated groups. The resin was washed extensively with 0.5 M NaCl on a glass filter. Almost complete immobilization was confirmed by protein determination (Bradford, 1976) of the filtrate. RecN16␣-agarose thus prepared was packed into a disposable polystyrene column (Pierce, product 29920). The column (volume, 1-2 ml) was equilibrated with Ca-TBS containing 0.1% Triton X-100.
To prepare a membrane fraction, which was assumed to contain lectin ligand glycoconjugates, frozen nematodes (10 g wet weight) were thawed, and soluble proteins including 16-kDa lectin were removed, as described under "Purification of 16-kDa Lectin." To the resultant pellet, Ca-TBS containing 1% Triton X-100 (Triton/Ca-TBS) was added in a ratio of 2.5 volumes to the original weight of worms. Membrane fraction, designated sup-3, was obtained after shaking (30 min at 4°C) and centrifugation (16,000 rpm at 4°C for 25 min).
The prepared membrane fraction was applied to a column (volume, 2 ml) packed with the recN16␣-agarose. After extensive washing of the column with Ca-TBS, the bound protein was eluted first with the same buffer containing 0.1 M sucrose, then with that containing 0.1 M lactose. The eluted fractions were subjected to protein determination (Bradford, 1976), SDS-PAGE, and lectin-probed Western blotting analysis described below.
Lectin-probed Western Blotting Analysis-Samples, which were expected to contain lectin-binding glycoproteins, were electrophoresed and electro-blotted onto a nitrocellulose membrane (Millipore) by the conventional Western blotting procedure. The blotted membrane was treated with TBS containing 1% Tween 20 (TBS-Tween), and allowed to react with recN16␣ (10 g/ml of TBS-Tween) at 4°C for 16 h, or 23°C for 2 h. All the following procedures were performed at ambient temperature (22-24°C). After washing of the membrane twice with TBS-Tween (2 min each) and then once with 0.15 M NaCl (5 min), the bound lectin was fixed by reaction with 0.1% (w/v) glutaraldehyde for 30 min. Excess reagent was removed by washing the membrane twice with TBS-Tween (5 min each), and the membrane was blocked with TBS-Tween containing 1% bovine serum albumin (Sigma, A2153, purity Ͼ96%) for 20 min. After brief washing with TBS-Tween, the membrane was treated with anti-nematode 16-kDa lectin antiserum (1,000-fold diluted in TBS-Tween) and horseradish peroxidase-conjugated goat anti-rabbit IgG antiserum (Seikagaku Kogyo, Tokyo; 1,000-fold-diluted in TBS-Tween), for 1 h each. After extensive washing of the membrane (e.g. 5 min, four times), bound second antibody was detected by using an Amersham enhanced chemiluminescence peroxidase kit (RPN 2209).

RESULTS
Purification of 16-kDa Nematode Lectin-In the previous procedure, 16-kDa lectin was recovered only as a minor component compared with 32-kDa lectin. Therefore, it was necessary to facilitate either worm preparation or lectin purification. After several trials, we found that commercially available lyophilized powder of M. lysodeikticus (e.g. from Sigma) can substitute for fresh E. coli as a food source. The yield of worms reached 3-6 g (wet weight)/liter of culture, almost comparable to that obtained with fresh E. coli.
We also found that the purification yield of 16-kDa nematode lectin was greatly improved by replacing the extraction buffer lactose/MEPBS with lactose/Ca-TBS. When the nematode was homogenized with lactose/Ca-TBS, the yield of 16-kDa lectin was approximately 200 g from 10 g of worms, while it was 20 -30 g when lactose/MEPBS was used. On the other hand, the yield of the 32-kDa lectin was greatly reduced (20 -30 g from 10 g of worms) compared with that by lactose/MEPBS extraction (200 g from 10 g of worms) (Fig. 1B). This may be largely attributed to calcium-activated proteolysis of 32-kDa lectin, because time-dependent decrease in this lectin was observed in Western blotting analysis of the extract prepared in the presence of calcium (data not shown). However, it was not associated with the increase in 16-kDa lectin band detected with anti-16-kDa lectin antiserum. This result suggests that the 16-kDa lectin is not a degradation product of the 32-kDa galectin, though it seemed likely that a region linking the two homologous CRDs (ϳ16 kDa) of 32-kDa galectin is susceptible to proteolysis, as has been demonstrated for galectin-4 (Oda et al., 1993;Tardy et al., 1995a). It is not clear why the yield of 16-kDa lectin was greatly increased when Ca-supplemented buffer was used. Although calcium was effective for the purification, this does not mean that the lectin requires calcium for FIG. 1. A, purification of C. elegans lectins (16 and 32 kDa) on asialofetuin-agarose by an improved procedure. In the present experiment, lectins (16 and 32 kDa) were extracted with 0.1 M lactose dissolved in Ca-TBS instead of in MEPBS. Subsequent steps were as described previously (Hirabayashi et al., 1992a(Hirabayashi et al., , 1992b. activity, because the bound 16-kDa lectin was not eluted from the asialofetuin-agarose column with EDTA (Fig. 1A).
The 16-kDa Lectin Is Extracted with Galactose-containing Saccharides-Since in previous studies only lactose had been used to dissociate both 16-kDa and 32-kDa lectins from the insoluble materials, other saccharides were examined for lectin extracting ability. Saccharides tested were galactose, glucose, mannose, L-fucose, lactose, melibiose, maltose, and sucrose. At the concentration of 0.1 M, 16-kDa lectin was extracted with only galactose-containing saccharides, i.e. lactose, melibiose, and galactose, but not with other simple saccharides (data not shown). The effect of concentration (1, 10, and 100 mM) of these saccharides was also examined (Fig. 2). Lactose was most effective; it dissociated 16-kDa lectin even at the lowest concentration (1 mM; lane 1). On the other hand, galactose and melibiose were much less effective; 10 mM galactose (lane 5) and melibiose (lane 8) could only partially dissociate the lectin. However, melibiose seemed to be slightly more effective than galactose (compare lane 5 with lane 8). As a control, sup-1 (lane 10), which was the first extract prepared in the absence of sugar, contained no detectable 16-kDa lectin. This result shows that the nematode 16-kDa lectin is not free in the native state like most galectins. As in the present case, roughly 100-fold higher affinity for lactose than galactose has been observed for vertebrate galectins (Leffler and Barondes, 1986;Sparrow et al., 1987;Oda et al., 1993).
The 16-kDa Lectin Behaves as a Non-covalent Dimer-In contrast to 32-kDa galectin, which has been shown to occur as a monomer (32 kDa) by a previous gel-filtration analysis (Hirabayashi et al., 1992a), the 16-kDa lectin was eluted from a high performance gel-filtration column (TSK GS-2000SW-XL) at a position approximately corresponding to that of recombinant human galectin-1 (29 kDa as a dimer; Hirabayashi et al., 1989) under nondenaturing conditions (data not shown). The result indicates that the lectin exists as a dimer under nondenaturing conditions. As described later, recombinant 16-kDa lectin (recN16␣) showed significant hemagglutinating activity. On the other hand, the 16-kDa lectin migrated as a monomer (16 kDa) in SDS-PAGE under both reducing (Fig. 1B) and non-reducing conditions (data not shown). Therefore, the native state of 16-kDa lectin is considered to be a non-covalent dimer.
Peptide Sequence Analysis-Affinity-purified 16-kDa lectin (ϳ90% purity) was further purified for structural analysis to remove a trace amount of 32-kDa galectin by means of re-versed-phase chromatography on a Tosoh trimethylsilyl column (data not shown). Thus purified 16-kDa lectin showed a similar amino acid composition to that of 32-kDa galectin: they showed similar contents of aspartic acid (12.2% for both 16 kDa and 32 kDa lectins, expressed in mol%), glutamic acid (10.3% and 11.4% for 16 kDa and 32-kDa lectins, respectively), proline (4.2% and 4.7%), isoleucine (7.1% and 7.2%) and arginine (5.4% and 5.3%), but had significantly different contents of tyrosine (0.9% and 3.2%) and lysine (1.2% and 7.2%). Based on the low content of lysine, the 16-kDa lectin was first digested with lysyl-endospecific Achromobacter protease I. This was particularly important, because direct N-terminal sequencing analysis was ineffective for intact 16-kDa lectin, as in the case of 32-kDa galectin. Derived peptides were purified on the trimethylsilyl column (data not shown). Two major fragments (designated Lys-4 and Lys-5, approximately 6 and 10 kDa, respectively, in SDS-PAGE) were obtained, of which one (Lys-4) was positive in Edman degradation, and was sequenced to the C-terminal end, while the other (Lys-5) was negative (assigned as N-terminal fragment) (see Table I). Three minor fragments were also obtained. These were attributed to nonspecific proteolysis. Two of them (Lys-1 and Lys-2) were successfully sequenced, and their positions in the whole sequence were confirmed, while the other (Lys-3) was negative in the sequence analysis.
To confirm the above result and to analyze undetermined regions, the N-terminal (Lys-5) and C-terminal (Lys-4) fragments were subjected to second digestions with trypsin, AspN, or Staphylococcus V8 protease. The derived peptides were purified on a TSK-ODS-80TM column, and sequenced (see results in Table I). Finally, 130 amino acids were determined with relevant overlapping sequences, and the result showed unambiguously that the nematode 16-kDa lectin is a novel member of the galectin family (described in more detail under "Discussion"). The total residue number comprised 95% of that finally obtained by cDNA cloning (described below), because we failed to obtain a small N-terminal peptide.
cDNA Cloning of the Nematode 16-kDa Galectin-To determine the sequence of the remaining N-terminal region, cDNA cloning was performed. For this purpose, RNA-PCR was carried out to obtain a relevant probe to screen a C. elegans cDNA library (Brastead and Waterstone, 1989) by using specific primers (for positions, see Fig. 3) and a template cDNA derived by oligo(dT) priming. Some of the amplified segments (0.27 kbp, obtained with primers F1 and R3, and 0.4 kbp obtained with F1 and R1) were sequenced or subjected to second PCR to confirm proper amplification. The latter 0.4-kbp fragment was used as a probe to screen the C. elegans cDNA library. We also confirmed that the library actually contained a target clone by performing "library PCR" using library-derived recombinant phage DNA as a template and various primers including F1 and R1 (data not shown). This seemed particularly important in the present case, because this library had been subjected to size selection to remove relatively small cDNAs (Ͻ0.5 kbp).
After three rounds of screening, a single clone, designated 7B1, was obtained. The clone had an insert of 0.55 kbp and was positive in Southern hybridization using the 0.4-kbp probe and in a PCR check using F1 and R1 primers (which amplified a 0.4-kbp fragment). The nucleotide sequence of the clone 7B1 was determined by the dideoxy termination method with a DNA sequencer (Applied Biosystems 373S) after automatic subcloning into pBluescript(SK Ϫ ).
The deduced amino acid sequence agreed completely with that directly determined by peptide analysis except for the N-terminal region, which was determined only by cDNA analysis (Fig. 3). The clone had a putative initiation codon ATG for initiator methionine (iMet). This methionine was followed by a   FIG. 2. Effects of saccharides (lactose, galactose, and melibiose) of various concentrations on extraction of 16-kDa lectin from C. elegans. The 16-kDa lectin was extracted with either lactose (lanes 1-3), galactose (lanes 4 -6), or melibiose (lanes 7-9) and was subjected to a conventional Western blotting analysis to see how much of the lectin had been extracted. Polyacrylamide gel (14%) and specific antiserum against the nematode 16-kDa lectin were used. Saccharide concentrations: lanes 1, 4, and 7 (1 mM); lanes 2, 5, and 8 (10 mM); lanes 3, 6, and 9 (100 mM). As a control, an extract in the absence of sugar (sup-1) was included for lectin extraction (lane 10). To each lane, 20 g of protein was applied.
decapeptide, Ile 2 -Gly-Gly-Gly-Ile-Gly-Ile-Ser-Phe 10 -Cys, and further by the previously determined tryptic peptide sequence (Lys-5-T-8), Asn 12 -Glu-Phe-Phe-Asn-Pro-Gln-Thr-Pro-Val 20 -Asn-Ile-Pro-Val-Gln-Gly-Phe-Ser-Asn-Gly 30 -Ala-Arg. This result was unexpected in two respects. (i) The N-terminal decapeptide sequence is unusual in that it is relatively rich in glycine and isoleucine, and has no charged amino acid; (ii) this decapeptide sequence ended with cysteine, but not with arginine or lysine. We could not obtain any corresponding fragment by reversed-phase chromatography after digestion of Lys-5 with various proteases. The unusual amino acid composition may partly explain the unexpected proteolysis and failure to recover the N-terminal peptide, because such a glycine-rich structure does not seem to form stable secondary structures and may be extremely susceptible to proteolysis. Including this missing N-terminal decapeptide, the calculated molecular weight and total residue number of the nematode 16-kDa galectin are 15,928 Da and 146 amino acid residues, respectively (initiator methionine included). The deduced amino acid sequence had no consensus sequence (Asn-X-Ser/Thr) for attachment of asparagine-linked oligosaccharide.
Northern and Southern Hybridizations-Northern hybridization was performed by using total RNA derived from mixed stages of C. elegans. The 0.4-kbp PCR fragment described above was 32 P-labeled and used as a probe. A single hybridization signal was detected, which corresponded to about 0.7 kbp (Fig. 4). This is reasonable considering the size of the cloned cDNA (0.55 kbp). The result also suggests that 16-kDa lectin is transcribed as a single molecular species.
Genomic Southern hybridization was performed after digestion of the nematode genomic DNA with BamHI, EcoRI, Hin-dIII, or PstI. After hybridization with the 0.4-kbp probe used above, each digest gave a single hybridization signal in autoradiography (Fig. 5). The result, though not definitive, suggests that the nematode 16-kDa galectin gene is unique.
Expression of a Recombinant Nematode 16-kDa Galectin in E. coli-A recombinant 16-kDa galectin was expressed in E. coli as a fusion protein with the ␣-peptide (5 kDa) of E. coli ␤-galactosidase (designated recN16␣) and was purified, as described previously (Hirabayashi et al. 1992b), by affinity chromatography on an asialofetuin-agarose column (data not shown). The purified recombinant galectin showed a main protein band (approximately 80%) at 21 kDa. Multiple minor bands between 18 and 16 kDa were also observed, which may represent degradation products originated from the 21-kDa protein, but retaining carbohydrate-binding activity. The yield of recN16␣ was typically 4 mg from a liter of culture.
Purified recN16␣ showed significant hemagglutination activity toward trypsinized rabbit erythrocytes prepared according to Kobiler and Barondes (1978). The activity was completely inhibited by the addition of 100 mM lactose. The minimum concentration of recN16␣ required for hemagglutination was about 10 g of the recombinant fusion protein/ml (equivalent to 7.6 g of intact protein/ml); in terms of specific activity, 140 titer Ϫ1 /mg/ml. This value is somewhat lower than those obtained for vertebrate galectins (i.e. approximately 1,000 titer Ϫ1 /mg/ml). Lactose showed the highest inhibitory effect on hemagglutination (concentration required for 50% inhibition, 6 mM), and melibiose and galactose were much less inhibitory (both 50 mM).
Identification of Glycoprotein Ligands for Nematode 16-kDa Galectin-It is important to identify endogenous receptor molecules for the nematode galectins, because they were shown to be homologous to vertebrate ones. The latter galectins have been shown to recognize a key disaccharide moiety, N-acetyllactosamine, in the particular context of poly-Nacetyllactosamine (Oda and Kasai, 1984;Cooper et al., 1991;Sato and Hughes, 1992;Zhou and Cummings, 1993;Ozeki   Negative al., 1995). Nevertheless, there is little information on glycoconjugates of the nematode (Bacic et al., 1990;Link et al., 1992;Borgonie, et al., 1994). Therefore, as the first step, we attempted to detect candidate glycoconjugate ligands for the nematode 16-kDa galectin by means of affinity chromatography using immobilized rec16␣. The binding ability of the adsorbent (4.8 mg of recN16␣ immobilized on 1 ml of agarose gel) was checked by applying bovine asialofetuin (1 mg) and mouse laminin (0.1 mg), to which vertebrate galectins have been shown to bind. Both glycoproteins were completely adsorbed on the column (1-ml volume) and were eluted specifically with 0.1 M lactose (data not shown). As the next step, C. elegans membrane fraction was prepared by using 1% (w/v) Triton X-100. Since all of the soluble proteins including galectins had previously been extracted with lactose, the fraction did not contain nematode 16-kDa galectin. The membrane fraction was applied to a column (2 ml) equilibrated with Ca-TBS containing 0.1% Triton X-100. After extensive washing of the column (10 volumes), the bound materials were FIG. 3. Nucleotide sequence of the clone 7B1 obtained by screening a ZAP cDNA library and deduced amino acid sequence. The clone was isolated by plaque hybridization (see "Experimental Procedures") with a 0.4-kbp probe obtained by RNA-PCR (F1 and R1 primers were used). Other PCR primers used for analyses are also indicated. Codons corresponding to amino acid residues determined directly by peptide analyses (see also  eluted first with 0.1 M sucrose, and then with 0.1 M lactose dissolved in the same buffer (Fig. 6A). Analysis of the eluted proteins by SDS-PAGE followed by silver staining revealed the presence of multiple protein bands only in the lactose eluate. The bands were mainly in the molecular mass range larger than 30 kDa (Fig. 6B, lane 1). A few of them (approximately 300 and 150 kDa, shown with closed triangles in lane 2) were dominant in the fraction. Since these proteins were not eluted with 0.1 M sucrose prior to the lactose elution, the binding is considered to be specific. Yields of these glycoprotein receptor candidates were 160 g from 2 ml of the detergent-solubilized membrane fraction (equivalent to 0.8 g of the nematode).
To confirm the above result, lectin-probed Western blotting analysis was performed by using free recN16␣ as described under "Experimental Procedures." When the lactose eluate was used, a few protein bands, mostly corresponding to the above observed ones (i.e. 300, 150, 100, and 70 kDa), were visualized (Fig. 6B, lane 2). In the analysis, lectin binding was greatly reduced in the presence of 0.1 M lactose (Fig. 6B, lane 3).

DISCUSSION
In our previous work, in which lactose/MEPBS was used for lectin extraction, 16-kDa lectin was purified only as a minor component relative to 32-kDa lectin. Since the latter consists of two homologous domains (i.e. CRDs) as a tandem-repeat type galectin, it seemed likely that the 16-kDa lectin was derived from the 32-kDa galectin by limited proteolysis. In fact, the 16-kDa species became a major component when MEPBS was substituted with Ca-TBS. This observation suggested the presence of a Ca-activated protease(s) responsible for the formation of 16-kDa lectin from the 32-kDa galectin. However, this was not the case, and the 16-kDa lectin was confirmed here to be a novel member of the galectin family. The 16-kDa galectin reacted only poorly with anti-32-kDa galectin antibody, and it showed significant but relatively low amino acid sequence similarity to the 32-kDa galectin. Since the C. elegans 32-kDa galectin lacks an apparent linker domain, a linking region between the two CRDs which exists in mammalian tandemrepeat type galectins, i.e. galectin-4 (Oda et al., 1993;Tardy et al., 1995a) and galectin-8 (Hadari et al., 1995), it may not be really cleaved. However, it remains unclear why calcium increases the yield of 16-kDa galectin but not that of 32-kDa galectin.
Although the sequence of nematode 16-kDa galectin was homologous to other galectin family members of both nematode and vertebrates, observed similarities were not very high: i.e. in terms of amino acid identities, the nematode 16-kDa galectin is homologous to the N-terminal (28%) and C-terminal domains (27%) of 32-kDa nematode galectin, human galectin-1 (19%), human galectin-2 (20%), human galectin-3 (23%), and the Nterminal (26%) and C-terminal domains (27%) of rat galectin-4 (see Fig. 7). Therefore, we cannot assign the nematode 16-kDa galectin as a possible ancestral molecule of small proto type galectins. Rather, as far as amino acid identities are concerned, nematode 16-kDa galectin is more similar to either the chimera (e.g. galectin-3) or the tandem-repeat type (e.g. galectin-4) than the proto type (i.e. galectins-1 and -2). In this context, Ahmed and Vasta (1994) have recently pointed out that galectin-1 is distinct from other galectin family members from the viewpoints of both saccharide-binding specificity and sequence similarity.
C. elegans 16-kDa galectin was shown to have a short redundant N-terminal region. The region is not as hydrophobic as usual signal sequences, nor does it resemble any known functional protein motif. Its function is not yet known, but may be unique to this lectin molecule. In this context, it is notable that many other galectins have an extra segment (non-CRD segment) in the N-terminal region, but not in the C-terminal region, as typically represented by the glycine-, proline-, and tyrosine-rich repetitive domain of galectin-3. A short, redundant N-terminal region is also observed in nematode 32-kDa galectin (tandem-repeat type), recently cloned galectin-8 (tandem-repeat type) (Hadari et al., 1995), and galectin-5 (monomeric proto type) (Gitt et al., 1995b). These facts imply that the N-terminal region can more readily be linked to other segments than the C-terminal region. Such N-terminal modifications may contribute to developing the "identities" of individual ga- FIG. 6. Isolation of recN16␣-binding glycoproteins on an immobilized-lectin column. A, a typical chromatogram obtained with 2 ml of membrane fraction and 2 ml of immobilized lectin gel prepared as described under "Experimental Procedures." Fraction volume was 2 ml. Triton X-100 (1%) extract (2 ml of the membrane fraction) was applied to the column (packed with 2 ml of recN16␣-immobilized gel) equilibrated with Ca-TBS containing 0.1% Triton X-100. After extensive washing of the column with the equilibration buffer, the bound glycoproteins were eluted with 0.1 M sucrose and then 0.1 M lactose. B, analysis of the eluted glycoproteins by SDS-PAGE followed by either silver staining (lane 1) or lectin-probed Western blotting (lanes 2 and 3; for details, see "Experimental Procedures"). In the latter analysis, recN16␣ binding was performed both in the absence (lane 2) and in the presence (lane 3) of 0.1 M lactose. Various lectin-probed bands are observed (shown with triangles), among which most significant ones are represented with closed triangles. lectin molecules.
According to the previous classification, which was made simply based on protein architecture (Hirabayashi et al., 1992b), the 16-kDa nematode galectin belongs to the proto type. Here, "proto" does not correspond to "primitive" in terms of molecular evolution. However, the fact that all animal species contain this type of galectin(s) (summarized in Table II) implies that the present galectins have actually evolved from such a proto type ancestral protein, although the present members should have diverged greatly, and thus, have become much more specific in their functions. So far identified prototype galectins have been shown to exist as non-covalent dimers with a few exceptions of recently cloned rat galectin-5 (Gitt et al., 1995b) and human galectin-7 (Madson et al., 1995;Magnald et al., 1995) that are monomeric. In the present study, C. elegans 16-kDa galectin was also found to exist as a dimer. Furthermore, recombinant 16-kDa galectin (recN16␣) caused hemagglutination of trypsinized rabbit erythrocytes, although its specific activity was somewhat lower than those of vertebrate galectins.
Why are there two galectins, 16-kDa and 32-kDa, in C. elegans? It seems meaningful that all animal species in which galectins have been found have more than one galectin, although this tendency seems to be more prominent in mammals (see Table II). Notably, all of them have conserved the fundamental ability to recognize a key disaccharide structure, Nacetyllactosamine. In this regard, Poirier and Robertson (1993) recently reported that deficiency of one of the mouse isolectins (galectins-1 and -3) was possibly compensated by the other. These authors generated null-mutant mice as regards galectin-1 gene by a procedure of homologous recombination, but these "knocked-out" mice were found to be completely vital and also proliferative. In this case, galectin-3 was suggested to substitute for galectin-1. Surprisingly, such unexpected results have been found increasingly for various developmental factors, such as tenascin (Saga et al., 1992), PrP (Bueler et al., 1992), and myoD (Rudnicki et al., 1992). These observations strongly suggest that many important proteins, including galectins, have evolutionally become more and more redundant (by making multiple gene copies), essentially because they are fundamental for the life of multicellular animals. In this context, the effect of destruction of either (and both) of the nematode galectin genes is of great interest. Although the 32-kDa galectin has been shown to be most abundantly expressed in the epidermal layer of adult worms, 2 the localization of the 16-kDa galectin is not known at present. When the 16-kDa galectin was extracted from the nematode, galactose and melibiose were also effective to some degree. However, this was not the case for 32-kDa galectin, because even 100 mM of these saccharides failed to extract the 32-kDa galectin (data not shown). Therefore, the 16-kDa galectin may have a looser specificity than 32-kDa galectin. Alternatively, these isolectins may have distinct kinetic properties in saccharide binding. In this regard, we recently found that the 32-kDa galectin has considerably slower on/off rates (i.e. in terms of k a and k d ) in binding to immobilized asialofetuin than the 16-kDa nematode galectin, through analysis with a Biacore biosensor. 3 Thus, co-occurrence of proto type and tandem-repeat type galectin in many animal species could be meaningful from various viewpoints, i.e. (i) saccharide binding specificity, (ii) kinetic properties, and (iii) cross linking features (homologous or heterologous binding to branched saccharides), as has recently been demonstrated (Gupta and Brewer, 1994;Bourne et al., 1994;Dessen et al., 1995).
Although C. elegans was chosen as a model animal to elucidate the basic features of complicated multicellular systems, basic biochemical studies on the worm are not well advanced. This is particularly true in the field of glycobiology. In fact, there has been almost no information on glycoconjugates and glycoenzymes of the nematode. In the present study, however, we identified several candidate glycoprotein ligands for C. elegans 16-kDa galectin, and showed that the binding was metalindependent and lactose-sensitive. Considering the conservative properties of galectins, the results imply the presence of N-acetyllactosamine-type glycoconjugates even in the nematode. In this context, comparative studies of the two C. elegans galectins from both biochemical 4 and genetic viewpoints 5 are in progress.