A Newly Identified Horseshoe Crab Lectin with Binding Specificity to O-antigen of Bacterial Lipopolysaccharides*

We identified a novel horseshoe crab hemocyte-derived lectin, which we named tachylectin-4. It has more potent hemagglutinating activity against human A-type erythrocytes than a previously identified hemocyte lectin with an affinity toN-acetylglucosamine, tachylectin-2. The purified tachylectin-4 is an oligomeric glycoprotein of 470 kDa, composed of subunits of 30 and 31.5 kDa. Ca2+ at 10 mmenhanced the hemagglutinating activity 4-fold, and the activity was inhibited by EDTA and o-phenanthroline.l-Fucose and N-acetylneuraminic acid at 100 mm completely inhibited the activity of tachylectin-4. The activity was also inhibited more strongly by bacterial S-type lipopolysaccharides (LPS) but not by R-type LPS lacking O-antigen. The most effective S-type LPS was from Escherichia coliO111:B4, and the minimum concentration required for inhibiting agglutination against human A-type erythrocytes (0.1 μg/ml) was 160-fold lower than those of S-type LPS from Salmonella minnesota. Therefore, colitose (3-deoxy-l-fucose), a unique sugar present in the O-antigen of E. coli O111:B4 with structural similarity to l-fucose, is the most probable candidate for a specific ligand of tachylectin-4. A cDNA coding for tachylectin-4 was isolated from a hemocyte cDNA library. The open reading frame of the 1344-base pair cDNA coded for the mature protein with 232 amino acids. There is no significant sequence similarity to any other known LPS-binding lectins, whereas tachylectin-4 is homologous to the NH2-terminal domain with unknown functions of Xenopus laevis pentraxin 1.

Arthropods have developed a unique immune system without the acquired immunoglobulin-dependent immunity found in vertebrates. Therefore, innate immunity, the pre-existing and immediate ability to prevent and limit invading microbes and pathogens, is likely to be a major host defense system in arthropods. The hemolymph of horseshoe crabs contains three abundant proteins, hemocyanin, C-reactive protein, and ␣ 2macroglobulin, and one type of granular cell, accounting for 99% of the total hemocytes (1,2). The granular cells are extremely sensitive to bacterial endotoxin, i.e. lipopolysaccharides (LPS), 1 and the cells release granular components in re-sponse to LPS stimulation (3)(4)(5)(6). This response is thought to be important for host defense in engulfing and killing invading microbes, in addition to preventing the leakage of hemolymph. The hemocytes contain large and small granules that selectively store proteins and defense molecules, including serine protease zymogens, a clottable protein that participates in the coagulation cascade, protease inhibitors, antibacterial peptides, lectins, and others (6).
Many kinds of lectins play crucial roles in innate immunity and host defense not only in vertebrates but also in invertebrates, with involvement in processes such as non-self recognition, inflammation, opsonization, cell-cell or cell-extracellular matrix interactions, fertilization, development, and regeneration (7)(8)(9)(10)(11). To better understand the biological role of lectins in host defense of the horseshoe crab, we have studied various lectins found in hemolymph of this animal (6). Recently, we identified new horseshoe crab hemocyte-derived lectins, named tachylectin-1 (L6) (12) and tachylectin-2 (L10) (13). Tachylectin-1 displays LPS-binding potential with antibacterial activity toward Gram-negative bacteria, but it has no hemagglutinin activity for human, sheep, and rabbit erythrocytes. On the other hand, tachylectin-2 has hemagglutinating activity against human A-type erythrocytes with specificity for N-acetylglucosamine. Furthermore, tachylectin-2 has agglutination activity against Staphylococcus saprophyticus KD. Both tachylectin-1 and tachylectin-2 are composed of unique tandem sequence repeats with no significant sequence similarity with other known proteins, including various animal and plant lectins. While continuing these studies on horseshoe crab lectins, we have now identified a new hemocyte lectin, which we named tachylectin-4; it has more potent hemagglutinating activity than tachylectin-2, shows a unique binding specificity to Oantigen of LPS, and has sequence similarity to the NH 2 -terminal domain with unknown functions of Xenopus pentraxin 1.
Glycosidase F Treatment-The samples were reduced and S-alkylated with iodoacetamide in 50 mM Tris-HCl, pH 7.5, containing 8 M urea (15). The S-alkylated protein (5 g) was incubated with 0.1 unit of N-glycosidase F (Boehringer Mannheim) in 50 mM Tris-HCl, pH 7.5, containing 2 M urea at 37°C for 18 h and subjected to SDS-PAGE under reducing conditions.
Isolation of Tachylectin-4-derived Peptides-Tachylectin-4 (100 g) was reduced and then S-alkylated with iodoacetamide (15). The Salkylated protein was digested with lysyl endopeptidase (enzyme/substrate ϭ 1/100, w/w) in 0.1 M NH 4 HCO 3 containing 2 M urea at 37°C for 24 h. The peptides were separated by reverse-phase HPLC using a Chemcosorb 5-ODS-H column (2.1 ϫ 150 mm, Chemco Scientific Co., Ltd., Osaka). Peptides were eluted from the column with a linear gradient of 0 -80% acetonitrile in 0.06% trifluoroacetic acid at a flow rate of 0.2 ml/min. Absorbance was monitored at 210 nm. Amino acid sequence analyses were performed using an Applied Biosystems 473A or 477A sequencer with the chemicals and programs supplied by the manufacturer (Perkin-Elmer Japan Co., Ltd., Tokyo, Japan).
Amino Acid Analysis-Samples were hydrolyzed in 6 M HCl containing 1% phenol at 110°C for 20 h in evacuated tubes. The hydrolysates were analyzed using a Pico-Tag system (Waters Japan Co., Ltd., Tokyo, Japan) (17).
Tachylectin-4-specific DNA Probes and Screening of cDNA Library-The degenerate nucleotide sequences of the primers used for PCR were based on the peptides derived from lysyl endopeptidase digestion (-Asn-Ala-Tyr-Val-Glu-Thr-and -Ile-Thr-Asp-Asp-Tyr-Val-) of tachylectin-4. Sense and antisense nucleotides were synthesized with an EcoRI site at the 5Ј end. Reactions for PCR contained the cDNA template (corresponding to 0.1 g of poly(A) ϩ RNA) and 100 pmol each of the primer were carried out in a Perkin-Elmer Cetus thermal cycler. The PCR products were treated with EcoRI and purified with agarose gel electrophoresis. Fragments of interest were then ligated into plasmid Bluescript II SK (Stratagene, La Jolla, CA) for sequence analysis, as described (18). One clone (0.4 kilobase), containing the sequence of tachylectin-4, was used as a probe to screen the ZipLox cDNA library. The PCR fragment, labeled with [␣-32 P]dCTP using a Ready-To-Go™ DNA-labeling kit (Pharmacia Biotech Inc.) served as a probe. After secondary screening, the plasmids containing the cDNA insert were prepared from the positive plaques, following the manual supplied by the manufacturer.
Homology Search-Amino acid sequence was compared with all entries in the SWISS-PROT protein data base by the FASTA homology search system of the European Bioinformatics Institute.

RESULTS
Purification of Tachylectin-4 -The lysate prepared from 100 g (wet weight) of hemocytes was first fractionated on a dextran sulfate-Sepharose CL-6B column (4.5 ϫ 20 cm) with increasing concentrations of NaCl from the range of 0.15-2.0 M (14), and a typical elution pattern and hemagglutinating activity against human A-type erythrocytes is shown in Fig. 1A. The activity in the flow-through fraction is derived from tachylectin-2 (13). The 0.15 M NaCl fractions containing a new lectin, named tachylectin-4, indicated by a bar were pooled and applied to a ConA-Sepharose column (3 ϫ 15 cm), equilibrated with 20 mM Tris-HCl, pH 7.5, containing 0.15 M NaCl. After washing with equilibration buffer, proteins were eluted with the same buffer containing 1 M ␣-methyl-D-glucoside (Fig. 1B). agarose column equilibrated with the same buffer. The column was washed extensively with the equilibration buffer, and then proteins were eluted with the same buffer containing 1 M Lfucose (Fig. 1C). All steps of the purification procedures were performed at 4°C. The purification scheme is summarized in Table I. The tachylectin-4 was purified about 600-fold with a yield of 56%, as calculated from the pooled fraction of dextransulfate-Sepharose CL-6B chromatography free from the hemagglutinating activity of tachylectin-2.
The purified tachylectin-4 gave a doublet on SDS-PAGE under reducing (M r ϭ 30,000 and 31,500) and non-reducing (M r ϭ 23,500 and 26,000) conditions (Fig. 1D). On the other hand, an apparent molecular weight of tachylectin-4 in solution was determined by gel filtration to be 470,000, indicating an oligomeric protein. The two subunits of 31.5 and 30 kDa could be separated by reverse-phase HPLC on a YMC C4 column (4.6 ϫ 150 mm, Yamamura Chemical Laboratories Co., Ltd., Kyoto, Japan) (data not shown), and their NH 2 -terminal sequences proved to be identical up to 18 residues as follows: Trp-Arg-Met-Leu-Tyr-Leu-Pro-Val-Ile-Val-Lys-Tyr-Gly-X-Met-Lys-Leu-Asp-. Furthermore, amino acid compositions of 20-h hydrolysates and peptide mapping patterns of the two subunits were almost indistinguishable (data not shown). The purified tachylectin-4 was treated with N-glycosidase F and subjected to SDS-PAGE (Fig. 2). The N-glycosidase F treatment resulted in disappearance of the upper band, and the lower band was only observed in the gel with the same mobility as the un-treated lower band (lanes 2 and 3), indicating that the different molecular weights between the subunits is caused by partial modification of the subunit with N-linked sugar(s).
The extinction coefficient of tachylectin-4 at 280 nm for 1.0% solution in Tris-HCl buffer (pH 7.5) was calculated from data derived from amino acid analysis. The value of 14.0 was used for subsequent determinations of tachylectin-4 concentrations.
Hemagglutinating Activity of Tachylectin-4 -Tachylectin-4 agglutinated all types of human erythrocytes and A-type erythrocytes were most sensitive (Table II). In the presence of 10 mM CaCl 2 , tachylectin-4 agglutinated A-type erythrocytes at the minimum concentration of 0.014 g/ml, and 60-and 120-fold higher concentrations were required for O-type and B-type erythrocytes, respectively. No hemagglutination was observed for erythrocytes derived from horse, rabbit, and sheep (Table  II). The hemagglutinating activity for A-type erythrocytes was enhanced 4-fold by 10 mM Ca 2ϩ , but other divalent cations including Zn 2ϩ , Co 2ϩ , Cu 2ϩ , and Mn 2ϩ at 0.1 mM and 10 mM had no effects. The agglutinating activity was completely inhibited by 10 mM EDTA and 0.2 mM o-phenanthroline, suggesting the presence of metal ion(s) other than Ca 2ϩ in tachylectin-4.
Effects of Carbohydrates on Hemagglutination-Effects of various carbohydrates on the hemagglutinating activity of tachylectin-4 are shown in Table III. For mono-and disaccharides, L-fucose and N-acetylneuraminic acid at the minimum concentration of 100 mM inhibited the hemagglutination. On the other hand, a bacterial cell wall component, LPS, was a more potent inhibitor. An LPS derived from E. coli O111:B4 was the most potent inhibitor at the minimum concentration of 0.1 g/ml. Interestingly, LPS isolated from rough mutants of E. coli O111:B4 (smooth) such as E. coli J5 (Rc) and E. coli F583 (Rd) showed no inhibitory effect, suggesting that the O-antigen of the S-type E. coli is a determinant for carbohydrate recognition of tachylectin-4. This was the case for other Gramnegative bacteria. LPS derived from S-types of both S. minnesota and S. typhimurium inhibited the hemagglutinating activity at the minimum concentrations of 15.6 and 7.8 g/ml, respectively, but LPS isolated from several rough mutants had no effect. To confirm the binding activity to O-antigen, hemagglutinating activity was measured using sheep erythrocytes coated with LPS derived from a S-strain (E. coli O111:B4) and a rough mutant (S. minnesota R595 (Re). Tachylectin-4 signif-   icantly agglutinated sheep erythrocytes sensitized with LPS of E. coli O111:B4, but not those sensitized with LPS of S. minnesota R595 (Re). This agglutinating activity was inhibited by free LPS derived from E. coli O111:B4, but not from S. minnesota R595 (Re). Lipoteichoic acid, a cell wall component of Gram-positive bacteria, also inhibited the hemagglutination, but lipoteichoic acid from Bacillus subtilis was not so effective (Table III).

Isolation of cDNA Clone and Nucleotide Sequence of Tachylectin-4 -
The tachylectin-4-specific probe of 0.4 kilobase was identified with oligonucleotides corresponding to peptides derived from tachylectin-4, using PCR and DNA sequence analysis. When the probe was used to screen a hemocyte cDNA library (400,000 recombinant phages), one positive clone with a 1.3-kilobase pair insert was subjected to restriction mapping followed by sequence determination of both strands and by sequential exonuclease digestion. The nucleotide and deduced amino acid sequences are shown in Fig. 3. The cDNA included 1,344 nucleotides with an open reading frame of 699 nucleotides. The open reading frame for the cDNA encoded for a mature protein of 232 amino acid residues with the calculated molecular weight of 26,625, and a signal sequence of 23 residues with a typical hydrophobic core. The candidate for an initiation codon ATG was found at nucleotide position 21. The stop codon at position 786 was followed by a polyadenylation signal, AATAAA, starting at position 1322. Amino acid sequences of the isolated peptides derived from tachylectin-4 corresponded exactly to the protein sequence deduced from the cDNA sequence, clearly indicating that the isolated cDNA clone codes for tachylectin-4. The deduced protein sequence contains one potential N-linked glycosylation site at Asn-108. Tachylectin-4 was sensitive to N-glycosidase F treatment (Fig. 2) and contained glucosamine, based on amino acid analysis (data not shown), indicating the presence of an N-linked sugar chain at this site. The isoelectric point of tachylectin-4 was calculated from amino acid composition to be 6.05 (19), which is rather acidic, compared with findings in tachylectin-1 (9.69) and tachylectin-2 (9.63).
Sequence Similarity to Other Proteins-A search of SWISS-PROT showed the sequence similarity of tachylectin-4 to Xenopus laevis pentraxin 1 (20). Tachylectin-4 has a 24% sequence identity to the NH 2 -terminal domain of Xenopus pentraxin 1 and the sequence of Pro-120 to Lys-159 of tachylectin-4 is highly conserved in the corresponding region of Xenopus pentraxin 1 (50% identity), as shown in Fig. 4. DISCUSSION We previously identified two kinds of lectins with no significant sequence similarity with other known proteins. These were named tachylectin-1 with LPS binding activity (12) and tachylectin-2 with hemagglutinating activity against human A-type erythrocytes (13). In the present study, we identified a horseshoe crab hemocyte lectin, tachylectin-4, with hemagglutinating activity against human erythrocytes and with binding specificity to O-antigen of bacterial LPS. Tachylectin-4 has different characteristics from both tachylectin-1 and tachylectin-2. 1) The minimum agglutination concentration of tachylectin-4 required for human A-type erythrocytes (0.014 g/ml) is about 100-fold lower than that of tachylectin-2 (1.6 g/ml) (13).
2) The addition of Ca 2ϩ at 10 mM enhances the hemagglutinating activity of tachylectin-4 by 4-fold, whereas Ca 2ϩ is not required for the activity of tachylectin-2 and EDTA has no effect on the activity of tachylectin-2 (13). However, the hemagglutinating activity of tachylectin-4 was completely inhibited by EDTA or o-phenanthroline, indicating that tachylectin-4 contains metal ion(s) with an important role for the sugar binding.
3) The activity of tachylectin-2 is inhibited by 0.1 mM N-acetylglucosamine (13), which had no effect on the activity of tachylectin-4 at 100 mM. On the other hand, L-fucose (6-deoxy-L-galactose) specifically inhibits the activity of tachylectin-4, with no inhibition for tachylectin-2. L-Rhamnose (6-deoxy-Lmannose), a stereoisomer of L-fucose at C-2 and C-4, has no effect on the activity of tachylectin-4, suggesting an important role for the configuration of C-2 or C-4 in sugar recognition. 4) Tachylectin-2 (27 kDa) exists as a monomer in solution (13), whereas tachylectin-4 is present as a high molecular mass oligomer of 470 kDa. Based on an assumption of 30 kDa for one subunit, tachylectin-4 is composed of 15-16 subunits, under physiological conditions. 5) Tachylectin-1 agglutinates sheep erythrocytes coated with LPS derived from both a wild-type (smooth) and a Re mutant of S. minnesota (12). In contrast, the hemagglutinating activity of tachylectin-4 was inhibited by S-type LPS, not by R-type LPS lacking O-antigen.
The LPS from E. coli O111:B4 was most effective, and the minimum concentration required to inhibit hemagglutination for human A-type erythrocytes was 160-fold and 80-fold lower than those of LPS from S. minnesota (smooth) and S. typhimurium (smooth), respectively. The O-antigen of E. coli O111:B4 is built up by a unique repeating unit of a main chain containing D-galactose, D-glucose, and D-N-acetylgalactosamine, and a monosaccharide side chain of colitose (3,6dideoxy-L-galactose or 3-deoxy-L-fucose) (21). In these monosaccharides, colitose is the most probable candidate for a specific ligand of tachylectin-4, since it is a unique monosaccharide present in the O-antigen of E. coli O111:B4, its structure is similar to that of L-fucose, and three other monosaccharides had no effect on the hemagglutinating activity of tachylectin-4  ( Table III). On the other hand, the O-antigen of S. typhimurium is composed of a main chain containing of D-mannose, L-rhamnose, and D-galactose, and two monosaccharide side chains of D-glucose and abequose (3,6-dideoxy-D-galactose or 3-deoxy-D-fucose) (21). Abequose is also a candidate for another ligand, since the hexoses except abequose have no effect on the activity of tachylectin-4 (Table III). Abequose is the D-isomer of colitose, which may cause a reduced affinity to tachylectin-4. A sequence homology search showed no significant sequence similarity between tachylectin-4 and other known LPS-binding lectins, such as Limulus endotoxin-binding protein-protease inhibitor (22), mammalian plasma LPS-binding protein (23), and Periplaneta lectins from the American cockroach (11,24,25), whereas the search indicates the sequence similarity of tachylectin-4 to X. laevis pentraxin 1 (20) (Fig. 4). Pentraxins are decameric or dodecameric proteins composed of identical protomers, arranged in two pentameric or hexameric rings interacting face-to-face (26,27). They constitute a family of carbohydrate binding proteins with divalent cation dependence, and some pentraxins, such as human C-reactive protein, belong to acute phase proteins, rapidly increasing their concentrations in response to stress, injury or infection (7, 28 -30).
However, Xenopus pentraxin 1 gene, identified from a Xenopus cDNA library using pentraxin-specific oligonucleotide probes, unexpectedly encodes a novel fusion protein with an NH 2terminal domain of a unique sequence and the COOH-terminal domain with sequence similarity to C-reactive proteins. The sequence similarity of tachylectin-4 to the NH 2 -terminal domain suggests that Xenopus pentraxin 1 is a chimeric protein with two different lectin-like activities.
The binding activities of tachylectin-4 with LPS and lipoteichoic acids suggest that tachylectin-4 plays a important role in recognition of invading bacteria. There is no evidence that horseshoe crabs have immunoglobulins (31), the recognition molecules produced by gene rearrangements, which help to distinguish self from non-self. Horseshoe crabs, however, do have an innate constitutive immune system that includes recognition by cell surface receptors or diverse lectin-like substances, phagocytosis (32), and killing reactions by antimicrobial molecules. Therefore, horseshoe crabs may oppose invaders through a combinatorial method using preexisting components in hemolymph plasma and hemocytes. Co-localization of these defense molecules in granules and their release into the hemolymph, in response to the stimulation of LPS, FIG. 3. Nucleotide and deduced amino acid sequences of tachylectin-4. A, a restriction map and sequencing strategy. B, the nucleotide and deduced amino acid sequences. Nucleotides and amino acid residues are numbered on the right. A single underline represents sequences determined by amino acid sequence analysis of isolated peptides. Asn-108 circled is an attachment site for N-linked sugar chain. The NH 2 -terminal amino acid of the mature protein is shown by an arrowhead. A dotted line and an asterisk represent a poly(A) additional signal and the termination codon, respectively. suggest that they serve synergistically to accomplish an effective host defense system against invading microbes and foreign substances.