Limulus intracellular coagulation inhibitor type 3. Purification, characterization, cDNA cloning, and tissue localization.

We reported that limulus intracellular coagulation inhibitor type-1 (LICI-1) (Miura, Y., Kawabata, S., and Iwanaga, S. (1994) J. Biol. Chem. 269, 542-547) and LICI type-2 (LICI-2) (Miura, Y., Kawabata, S. , Wakamiya, Y., Nakamura, T., and Iwanaga, S. (1995) J. Biol. Chem. 270, 558-565) found in the hemocyte lysate belong to the serpin family. The LICI-1 specifically inhibits limulus lipopolysaccharide-sensitive serine protease, factor C (k1 = 2.5 x 10(6) M-1 s-1), whereas LICI-2 inhibits preferentially limulus clotting enzyme (k1 = 4.3 x 10(5) M-1 s-1). In our ongoing studies on limulus serpin, we found another inhibitor, named LICI type-3 (LICI-3), which strongly inhibits (1,3)-beta-D-glucan-sensitive serine protease, factor G (k1 = 3.9 x 10(5) M-1 s-1). Thus, the limulus hemolymph coagulation cascade is effectively regulated by at least the three endogenous serpins. LICI-3, newly identified in hemocytes, is a single chain glycoprotein with an apparent Mr = 53,000, the largest one among known limulus serpins. A cDNA sequence for LICI-3 coded a mature protein of 392 amino acids, of which 68 residues were confirmed by peptide sequencing. LICI-3 showed significant sequence similarity to LICI-1 (45.8% identity) and LICI-2 (33.7% identity). LICI-3 contained a putative reactive site, -Arg-Ser-, distinct from that of LICI-2 (-Lys-Ser-) but the same as that of LICI-1. Expression of LICI-3 mRNA was detected only in hemocytes, and not in heart, brain, stomach, intestine, coxal gland, and skeletal muscle. Immunoblotting of the hemocyte-derived large and small granules with antiserum against LICI-3 suggested that it is stored specifically in large granules, as in the case of LICI-1 and LICI-2, and is released in response to external stimuli.

The limulus hemolymph coagulation cascade induced by lipopolysaccharide is composed of three serine protease zymogens, factor C (1), factor B (2), and proclotting enzyme (3) and a clottable protein, coagulogen (4). This cascade is activated not only by lipopolysaccharide, but also by (1,3)-␤-D-glucan, the major cell wall component of fungi (5,6). The (1,3)-␤-D-glucansensitive system activates directly the proclotting enzyme through a glucan-sensitive serine protease zymogen factor G (6), resulting in the formation of coagulin gel. These responses are thought to be important for the host defense in engulfing invading microbes, in addition to preventing the leakage of hemolymph.
In our ongoing studies on the molecular mechanism of the limulus clotting pathway, we obtained evidence for the existence of serine protease inhibitors (serpins) in hemocytes of the Japanese horseshoe crab (Tachypleus tridentatus) (7,8). These serpins, named limulus intracellular coagulation inhibitors (LICI-1 and LICI-2), 1 specifically inhibit lipopolysaccharidesensitive serine proteases, factor C and clotting enzyme, and are functionally and structurally related to protease inhibitors of the mammalian plasma serpin superfamily. We have now identified a third serpin, designated LICI type-3 (LICI-3), with a unique inhibitory spectrum. Purification, characterization, cDNA cloning, and tissue localization of this serpin is described herein.
Assay for LICI-3-The LICI-3 activity was expressed as an inhibitory activity against the limulus-purified clotting enzyme. The amidase activity of clotting enzyme was routinely assayed, using the chromogenic substrate, Boc-Leu-Gly-Arg-pNA. Limulus clotting enzyme, 0.4 unit (3), was preincubated with a sample in 200 l of 0.1 M Tris-HCl buffer, pH 8.0, containing 0.05% bovine serum albumin and 0.1 M NaCl at 37°C for 30 min, and 50 l of 2 mM substrate were then added. The reaction mixture was incubated at 37°C for 5 min and terminated by the addition of 750 l of 0.6 M acetic acid. The resulting chromophore * This work was supported by a grant-in-aid for scientific research from Ministry of Education, Science, Sports and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  was measured at 405 nm. One unit of the inhibitor was defined as the amount of protein that inhibited 1 unit of clotting enzyme (3).
Determination of Second-order Rate Constants-The second-order rate constants (k 1 ) for the inhibition of LICI-3 toward factor G, clotting enzyme, factor C, and ␣-chymotrypsin were determined using the method of Ehrlich et al. (16) as follows; each enzyme (final concentration, 5 nM) was incubated with LICI-3 (final concentration, 25 nM) in 1 ml of 0.02 M Tris-HCl buffer, pH 8.0, containing 0.05% bovine serum albumin and 0.1 M NaCl at 37°C. At appropriate times (15 s to 15 min), 50-l aliquots were removed, and the reaction was terminated by the addition of 950 l of the buffer containing each substrate (0.6 mM). The concentrations of residual enzymes were calculated from their specific activities. The second-order rate constant (k 1 ) was calculated using the standard equation for a second-order reaction (16).
Antimicrobial Activity-Escherichia coli B was used for determination of the antimicrobial activity, by the method of Saito et al. (11,17).
Isolation of LICI-3-derived Peptides, Sequencing, and Amino Acid Analysis-The purified inhibitor (100 g) was digested with lysyl endopeptidase (18). Peptides were separated by reversed-phase high performance liquid chromatography, using a Bondasphere C 8 300A column (Nihon Waters Ltd., Tokyo). Amino acid sequence analysis of purified peptides was performed using a gas-phase sequencer model 473A (Applied Biosystems) with the chemicals and program supplied by the manufacturer. For amino acid analysis, samples were hydrolyzed in 6 M HCl in evacuated and sealed tubes at 110°C for 24, 48, and 72 h. Cysteine was determined as cysteic acid after performic acid oxidation or as S-carboxymethylcysteine after iodoacetic acid treatment, with or without dithiothreitol. The hydrolyzates were analyzed using a Hitachi L-8500 amino acid analyzer with the chemicals and program supplied by the manufacturer.
LICI-3-specific DNA Probes and Screening of cDNA Library-The degenerate nucleotide sequences of the primer used for PCR were based on amino acid sequences of peptides K20 (-KGFWET-) and K13 (-DMGMK-), since the location and spatial relationship of these peptides could be deduced from their sequence similarities to the conserved regions of the serpin superfamily. Sense and antisense nucleotides were synthesized with an EcoRI site at the 5Ј end, using a DNA synthesizer model 380B and the chemicals and program supplied by the manufacturer (Applied Biosystems). Reactions for PCR contained the cDNA template (corresponding to 0.1 g of poly(A) ϩ RNA) and 200 pmol of each primer and were carried out in a Perkin-Elmer thermal cycler. The PCR products were treated with EcoRI and fractionated on low melting point agarose (Life Technologies, Inc.). Fragments of interest were then ligated into pBluescript II (Stratagene, La Jolla, CA) for sequence analysis, as described by Sambrook et al. (19). One clone with a 2.3kilobase pair insert contained the sequence of LICI-3-derived peptides and was used as a probe for screening the ZipLox cDNA library. A ZipLox cDNA library was constructed from poly(A) ϩ RNA, as reported previously (20). The PCR fragment, labeled with [␣-32 P]dCTP using a Ready-To-Go DNA-labeling kit (Pharmacia) served as a probe to screen the library. After tertiary screening, the plasmids containing the cDNA insert were prepared from the positive plaques, following the manual supplied by the manufacturer (Life Technologies, Inc.).
Computer-assisted Analysis of Sequence Data-The LICI-3 sequence was compared with all entries in the SWISS-PROT data base (release 30, January 1995) using the Gene Works software package (version 2.4 IntelliGenetics, Inc., Mountain View, CA).
Enzyme-linked Immunosorbent Assay (ELISA) of LICIs with Big Defensin-Microtiter plates were coated with big defensin (50 l/well) in Tris-HCl buffer, pH 7.5, containing 0.15 M NaCl (TBS) for 1 h at 37°C or overnight at 4°C. TBS containing 3% dried milk (200 l/well) was added to block any nonspecific binding. After standing for 1 h at room temperature, the wells were washed three times with TBS buffer. Then, LICIs in blocking buffer (50 l/well) were added, and the preparation was incubated at 37°C for 1 h, after which the wells were washed three times with TBS buffer. Antisera against LICIs (50 l/well) were added to each well, and the plates were incubated at room temperature for 1 h. Then, the wells were washed three times with TBS buffer and incubated with 50 l of horseradish peroxidase-conjugated goat anti-rabbit IgG (H ϩ L) for 1 h, washed again three times with TBS buffer and then three times with distilled water. Finally, each well was incubated with 150 l of 10 mM o-phenylenediamine dissolved in 50 mM sodium acetate buffer, pH 5.0, containing 0.1% H 2 O 2 . The reaction was stopped with 1 M H 2 SO 4 containing 1% sodium sulfate, and the absorbance of each well was measured at 490 nm using a microplate reader, model 3550 (Bio-Rad). As a control, antithrombin III and human neutrophil defensin were used.
Component Sugar and Sialic Acid Analyses of LICI-1, LICI-2, and LICI-3-Component sugar analyses were performed on samples hydrolyzed in 4 M trifluoroacetic acid at 100°C for 3 h, using derivatizations with 2-aminopyridine, and quantification of sialic acid was performed as described by Nishimura et al. (21).
Reverse Transcription-PCR Analysis-Total RNA was extracted from various tissues of an adult male Japanese horseshoe crab (T. tridentatus). First-strand cDNA synthesis from 10 g of total RNA was performed using SuperScript II Rnase H Ϫ reverse transcriptase (Life Technologies, Inc.) and random primers. Oligonucleotide primers for PCR amplification were the same as those used for screening of a cDNA library. One-twentieth of the first-strand cDNA and 50 pmol of each primer were subjected to PCR (30 cycles) with denaturation at 94°C for 0.5 min, annealing at 45°C for 1 min, and extension at 65°C for 2 min. PCR products were analyzed on a 2% agarose gel and visualized following ethidium bromide staining.
Preparation of Antiserum against LICI-3-An antiserum against LICI-3 was raised in rabbits, as described previously (22). LICI-3 (50 g) was emulsified in synthetic adjuvant, Titer Max (Vaxal, Inc., Norcross, GA) and given intradermally. Every 4 weeks, a booster with 50 g of LICI-3 in the same adjuvant was given. The blood sample taken 1 week after the third injection was stored in serum form at Ϫ80°C.
SDS-PAGE and Immunoblotting-SDS-PAGE was performed in 10% slab gels according to the method of Laemmli (23). The gels were stained with Coomassie Brilliant Blue R-250. For immunoblotting, proteins were transferred to nitrocellulose membranes, using an electroblot apparatus (Bio-Rad) overnight at 20 V. The membranes were then treated with the LICI-3 antiserum and incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG, as described elsewhere (8).
Protein Concentrations-Concentration of human ␣-thrombin and ␣-chymotrypsin was determined by the active site titration with pnitrophenyl pЈ-guanidinobenzoate (24). Active site concentrations of limulus clotting enzyme and factor C were titrated with human antithrombin III by the method of Lawrence et al. (25). Concentrations for other proteins were calculated from their extinction coefficients of 1% solution at 280 nm as follows: 7.6 for the clotting enzyme (M r ϭ 58,000) (3), 21.3 for factor C (M r ϭ 123,000) (1), and 10 for factor G (M r ϭ 110,000) (10). The concentration of the purified LICI-3 was determined by titration with ␣-thrombin.

RESULTS
Purification of LICI-3-The lysate (1,200 ml) prepared from 55 g (wet weight) of hemocytes was first fractionated on a dextran sulfate-Sepharose CL-6B column (5 ϫ 23 cm) with increasing concentrations of sodium chloride from the range 0.15 to 2.0 M (7), and fractions were assayed for inhibitory activity toward clotting enzyme, as described under "Experimental Procedures" (Fig. 1A). The 0.5 M NaCl fractions containing LICI-2 and a new inhibitor (LICI-3) which are indicated by a bar were pooled and concentrated by ultrafiltration. LICI-3 was separated from LICI-2 by gel filtration on a Sephacryl S-200 column (4 ϫ 142 cm), equilibrated with 50 mM sodium acetate, pH 5.5, containing 0.05 M NaCl (Fig. 1B). The LICI-3 fractions indicated by the bar were pooled and dialyzed against 20 mM Tris-HCl, pH 7.5, containing 0.05 M NaCl and applied to a DEAE-Sepharose CL-6B (2 ϫ 16 cm) column, equilibrated with 20 mM Tris-HCl, pH 7.5, containing 0.05 M NaCl (Fig. 1C). The column was washed extensively with equilibration buffer, then proteins were eluted with a linear gradient of 0.05 to 0.35 M NaCl. Active fractions (bar) were pooled and dialyzed against 20 mM Tris-HCl, pH 7.5, containing 0.05 M NaCl. The dialyzed sample was again applied to a DEAE-Sepharose CL 6B column equilibrated with the same buffer. After applying the sample, the column was washed with 20 mM Tris-HCl, pH 8.5, containing 0.1 M NaCl and the protein was eluted with a linear gradient of 0.1-0.5 M NaCl in the same buffer (Fig. 1D). All steps of purification procedure were performed at 4°C. The purified LICI-3 gave a single protein band with M r ϭ 53,000 on SDS-PAGE, under reduced conditions (Fig. 1E). The purification scheme is summarized in Table I. The overall yield was 4.1% with substantial losses occurring especially at the second step of chromatography, but this step was required to remove LICI-2 contaminated in the first step of chromatography.
Inhibition of Limulus Coagulation Factors and Other Serine Proteases by LICI-3-To determine whether LICI-3 inhibits limulus clotting factors and other mammalian serine proteases, a 10 M excess of the inhibitor was incubated with each protease at 37°C for 30 min, and the remaining amidolytic activity was assayed, as described under "Experimental Procedure." Limulus factor G, factor C, and clotting enzyme activities, in addition to ␣-chymotrypsin, were inhibited by LICI-3 (Table II). Furthermore, LICI-3 inhibited the activities of bovine plasmin, human ␣-thrombin, urokinase, tissue type plasminogen activa-tor, rat salivary kallikrein, and trypsin. Among the clotting factors, the inactivation of factor G was the most rapid (k 1 ϭ 3.9 ϫ 10 5 M Ϫ1 s Ϫ1 ). Within the LICIs, LICI-1 did not inhibit the ␣-chymotrypsin activity, yet LICI-2 and LICI-3 did strongly inhibit such activity. The inactivation of ␣-chymotrypsin by LICI-3 (k 1 ϭ 1.2 ϫ 10 5 M Ϫ1 s Ϫ1 ) was three times more rapid than that of LICI-2 (k 1 ϭ 3.8 ϫ 10 4 M Ϫ1 S Ϫ1 ). A thiol protease, papain, was not inhibited by LICI-3 (data not shown). LICI-3 formed a stable one-to-one complex with factor G and ␣-chymotrypsin, both of which did not dissociate in the presence of 2% SDS. For example, the interaction of LICI-3 (53 kDa) with the factor G-derived protease domain (B-chain, 34 kDa) (26) yielded a 87-kDa complex, under nonreducing conditions (Fig.  2, lane 2). The complex of LICI-3 (53 kDa)-␣-chymotrypsin with a 65-kDa product was also detectable (lane 3). During the incubation of LICI-3 with ␣-chymotrypsin, a new product (M r ϭ 44,000) appeared with immunoblotting, in addition to the complex between LICI-3 and ␣-chymotrypsin. This product was thought to be LICI-3 proteolytically modified by the free enzyme, since ␣-chymotrypsin was not autocatalytically degraded under these conditions and the 44-kDa product was recognized by anti-LICI-3 antiserum.
Peptide Sequencing-Intact LICI-3 (100 pmol) was subjected to amino acid sequence analysis and the partial NH 2 -terminal sequence of DYLLDEILHLSDVDQQQLSA-was determined. The NH 2 -terminal Asp was also confirmed by amino acid sequence analysis of the NH 2 -terminal-derived peptide. The six peptides derived from LICI-3 were isolated and sequenced, the results of which revealed sequences of 68 amino acids, as shown in Fig. 3.
Isolation of a cDNA Clone and Nucleotide Sequence of LICI-3-The LICI-3-specific probe with a 0.36-kilobase pair insert was identified with oligonucleotides corresponding to peptides derived from LICI-3, using PCR and DNA sequence analyses. When the probe was used to screen a hemocyte cDNA library (500,000 recombinant phages), the one positive clone with a 2.3-kilobase pair insert was subjected to restriction mapping followed by sequence determination of both strands, by sequential exonuclease digestion. The nucleotide and deduced amino acid sequences are shown in Fig. 3. The cDNA included 1,429 nucleotides with an open reading frame of 1,242 nucleotides. A stop codon TAA (nucleotide position 50) was followed by an initiation Met beginning at position 84. The open reading frame for the LICI-3 cDNA encoded for a mature protein of 392 amino acid residues and a signal sequence of 22 residues with a typical hydrophobic core. Cleavage at the Gly Ϫ1 -Asn ϩ1 bond with a typical motif for the recognition of signal peptidase (29) was presumed to give a mature protein with an NH 2 -terminal sequence identical with that of the purified LICI-3, except for the NH 2 -terminal Asp of the purified protein. A DNA fragment of 274 bp coding the NH 2 -terminal region of the mature protein was then amplified by PCR, using limulus hemocyte cDNA as a template. The nucleotide sequence of the PCR product also coded the Asn at the corresponding position (data not shown).
The amino acid analysis of LICI-3 agreed well with the amino acid composition deduced from the cDNA (Table III). Like LICI-2 (no cysteine residues in LICI-1), LICI-3 contains two cysteine residues in the deduced amino acid sequence, and 1.9 mol of cysteic acids were detected after performic acid oxidation. These cysteine residues probably form a disulfide bridge, since carboxymethylcysteines were detected only in the hydrolysate of the sample S-alkylated after reduction (data not shown).
Carbohydrate Compositions of LICIs-There are three potential N-linked glycosylation sites at Asn 26 , Asn 208 , and Asn 245 . LICI-1 and LICI-2 are also glycoproteins and each contains one N-linked sugar chain at Asn 380 and Asn 152 , respectively (7,8). Sugar compositions of the three LICIs were analyzed by derivatization with 2-aminopyridine. Under the conditions used, 2.2 and 2.3 residues of N-acetylglucosamine/ protein were quantitated in LICI-1 and LICI-2, and the ratios of sugar components between the two inhibitors were similar (Table III). On the other hand, LICI-3 contained 5.6 mol of N-acetylglucosamine, and the value was about three times higher than those of LICI-1 and LICI-2, suggesting that the three potential sites are all glycosylated in LICI-3. The calculated molecular weight of sum of sugar components of LICI-3 (M r ϭ 5,697) compensates for the difference between the mo-lecular weight based on the amino acid sequence (M r ϭ 43,945) and that estimated by SDS-PAGE (M r ϭ 53,000). N-acetylgalactosamine and sialic acid were not detected in any LICIs.
Sequence Similarity to Other Serpins-Based on the sequence data, LICI-3 was more closely related to LICI-1 (45.8% identity) than to LICI-2 (33.7%), as shown in Fig. 4. It also shared sequence similarities to other members of the serpin superfamily, including mammalian and insect serpins, such as human monocyte/neutrophil elastase inhibitor (35% identity) (30), human plasminogen activator inhibitor type 2 (29% identity) (31), human antithrombin III (29%), antitrypsin (30%) from Bombyx mori (32), elastase inhibitor from Manduca sexta (29%) (33), and antichymotrypsin 1 from B. mori (19% identity) (34). Comparing LICI-3 with LICI-1 and LICI-2 (Fig. 4), there was a putative reactive site, -Arg 359 -Ser 360 -, in the COOHterminal region as in the case of LICI-1 and also in the serpin superfamily. Furthermore, two specific regions important for a functional inhibitor, that is, the hinge region EEGTV (positions from 343 to 347) and Ala 348 at the NH 2 -terminal side of the reactive center loop of serpins, referred to as the P12 site (which is known as a critical determinant of inhibitor status of serpins), was completely conserved in LICI-3 as well as LICI-1 and LICI-2. As in the cases of LICI-1 and LICI-2, LICI-3 also contained the consensus sequence FLF(F/L)I in the corresponding region, which is known as a clearance signal, for instance, FVFLM for ␣ 1 -antitrypsin, exposed after formation of the serpin-enzyme complex.
Expression of LICI-3 mRNA in Various Tissues-To investigate tissue specific expression of LICI-3, reverse transcription-PCR analysis was carried out using mRNA extracted from hemocytes, heart, hepatopancreas, stomach, intestine, coxal gland, brain, and skeletal muscle. Using the LICI-3 oligonucleotide primers that produce the PCR product of 360 bases, the expression of LICI-3 was detected only in hemocytes but not in other tissues examined (data not shown).
Subcellular Localization and Release of LICI-3 from Hemocytes-Antiserum raised against purified LICI-3 was used to identify localization of this coagulation inhibitor in hemocytes. Large and small granules isolated from hemocytes (35) were first treated with 1% SDS at 100°C for 2 min and subjected to SDS-PAGE, under reducing conditions for immunoblotting. The anti-LICI-3 antiserum recognized the 53-kDa LICI-3 in the extract of large granules (data not shown). However, immunoreactive materials were never observed in the extract of small granules, indicating that LICI-3 is located in the large granule.  To determine if LICI-3 could be secreted from hemocytes by external stimulation, the hemocytes were treated with calcium ionophore A23187, which induces exocytosis of limulus hemocytes (36,37). The three LICIs were released into the extracellular fluid, as detected by ELISA, using each specific antiserum (data not shown). Under the conditions used, hemocytes were not lysed, since lactate dehydrogenase activity, a cytosolic marker enzyme, was negligible.
Interaction of LICI-1 with an Antimicrobial Peptide, Big Defensin-Recently, Panyutich et al. (38) reported that human plasma serpins such as ␣ 1 -proteinase inhibitor, ␣ 1 -antichymotrypsin, and antithrombin III form complexes with human neutrophil defensin, an antimicrobial and cytotoxic peptide, and this complex formation inactivates biological activities of both serpins and the defensin, thereby suggesting that the interaction may have a role in regulating inflammatory processes. In horseshoe crab, a defensin-like peptide, named big defensin, has been recently identified in the hemocytes (11). To investigate the interaction of LICIs with big defensin, the three LICIs were incubated separately with big defensin-precoated on a microtiter plate, and LICIs bound were quantitated by ELISA (Fig. 5). LICI-1 bound to big defensin in a dose-dependent manner but LICI-2 and LICI-3 showed no interactions with big defensin. Moreover, when increasing concentrations of big defensin (50 to 200 molar excesses of LICI-1) were preincubated with LICI-1, it did not effect the inhibitory activity against the clotting enzyme. Increasing the concentrations of LICI-1 (10 to 50 molar excess of big defensin) and the complex between LICI-1 and its target protease factor did not inhibit the antimicrobial activity of big defensin, thereby differing from human plasma serpins. DISCUSSION In the present study, a third intracellular protease inhibitor type-3 (LICI-3) from limulus hemocytes was purified and characterized and the entire sequence of a cDNA coding for LICI-3 was determined. LICI-3 is a single-chain glycoprotein consisting of 392 amino acids with an apparent M r ϭ 53,000, under nonreducing conditions. LICI-1 (7) specifically inhibits factor C, whereas LICI-2 (8) inhibits the clotting enzyme in addition to factor C. LICI-3 found here, on the other hand, inhibits preferentially factor G (k 1 ϭ 3.9 ϫ 10 5 M Ϫ1 s Ϫ1 ) by forming a covalent 1:1 complex, probably through the putative reactive site, -Arg 359 -Ser 360 -. However, LICI-2 and -3 exhibit broad specificity for inhibition to the hemolymph coagulation factors, compared to strict specificity of LICI-1 (Table II), although structural basis of the specificity of LICIs cannot be explained.
Glycosaminoglycans have no apparent effect on LICI-3 activity as is the case with LICI-1 and LICI-2, a finding clearly different from mammalian serpins, such as antithrombin III, heparin cofactor II, protease nexin I, and plasminogen activator inhibitor-1. Evans et al. (39) reported that there is a unique conserved heparin-binding site rich in positively charged residues on the external surface of the heparin-sensitive serpins.
In the sequences of the three LICIs, however, amino acids with a positive charge at the heparin-binding site of antithrombin III are replaced by hydrophobic or neutral amino acids in the corresponding region (Fig. 4). On the other hand, LICIs contain amino acid residues indispensable for functional activity. Skriver et al. (40) reported that an Ala residue at the NH 2terminal side of the reactive center loop of serpins, referred to as the P12 region, is a critical determinant of the inhibitor status of serpins. In the sequence of LICI-3, Ala 348 is also FIG. 3. Nucleotide and deduced amino acid sequences of LICI-3. Nucleotides and amino acid residues are numbered on the right. Underlines represent sequences determined by amino acid sequence analysis of isolated peptides. A box represents a putative reactive site. Diamonds indicate potential sites for N-linked sugar chain. conserved in the P12 region of the putative reactive site (Fig.  4). Joslin et al. (41) have advanced the hypothesis that serpinenzyme complexes are more rapidly cleared than the corresponding native proteins through a cell-surface receptor, socalled the serpin-enzyme complex receptor, which recognizes a pentapeptide neodomain exposed on the serpins of the complexes. The sequence of the pentapeptide is highly conserved in the carboxyl-terminal region of serpins, for instance, -F-V-F-L-Mfor ␣ 1 -antitrypsin. As LICIs also contain the consensus sequence, -F-L/V/M-F-F/L-I-in the corresponding region (Fig. 5), a similar clearance mechanism may function in horseshoe crab.
Mammalian serpins such as ␣ 1 -proteinase inhibitor, ␣ 1 -antichymotrypsin, and antithrombin III form complexes with neutrophil defensin, an antimicrobial and cytotoxic peptide, and both biological activities are neutralized after complex formation (38). In limulus, we also found that a 8.5 kDa protein, now named big defensin (11), is co-purified during the purification of LICI-1, and the complex is separated at the last step of the purification (7). As expected, Fig. 5 clearly indicates the interaction between big defensin and LICI-1 but not LI-CIs-2 and -3. Loss of their biological activities did not occur after complex formation. LICIs, coagulation factors, and big defensin co-locate in large granules of hemocytes (11,35,42,43). Thus, this interaction may have physiological importance in neutralization or intracellular sorting of big defensin.
Component sugar analysis suggests that the content of sugars in LICI-3 is three times higher than in LICI-1 and LICI-2, therefore, the three potential sites in LICI-3 are probably all glycosylated by sugar chain similar to those for LICI-1 or LICI-2. No sialic acid is present in any LICI. Mammalian serpins, such as plasma serpins from horse and wallaby (44) FIG. 4. Alignment of the amino acid sequence of LICI-3 with those of LICI-1 (7) and LICI-2 (8). Manual alignment and position numbers are based on the sequence of LICI-3, with appropriate gaps. Residues identical to LICI-3 are boxed. The positions of LICIs that correspond to the high affinity heparin-binding site of antithrombin III are represented by asterisks. The reactive sites in these serpins are indicated by a shadowed box.
FIG. 5. Interaction of LICIs with limulus big defensin. Big defensin (20 g/ml) was coated on a microtiter plate and various concentrations of LICIs in TBS buffer were added and the preparation incubated at 37°C for 1 h. The anti-LICIs antisera and goat-anti-rabbit IgG were used as first and second antibodies. Experimental details are presented under "Experimental Procedures." Open circles, triangles, and closed circles represent LICI-1, LICI-2, and LICI-3, respectively. and human ␣ 1 -protease inhibitor (45), contain sialic acids in their sugar chains. The roles of carbohydrate chains in serpins are postulated to function in secretion, retention of reasonable half-life in the circulation, and recognition by receptors for the complex uptake (46). Recombinant ␣ 1 -antichymotrypsin without sugar chains has the same second-order rate constant as the serum-derived inhibitor for association toward ␣-chymotrypsin, thereby indicating no effect of the carbohydrate chain on inhibitory activity (47). Whether or not glycosylations function in the same in limulus remains to be analyzed. The expression of mRNA for LICI-3 was detected only in hemocytes, and the same results were also obtained for LICI-1 and LICI-2 (7,8). Although mammalian plasma serpins are mainly expressed in the liver, mRNAs for LICIs are not detected in hepatopancreas with functions analogous to the liver and pancreas of vertebrates. LICI-1, LICI-2, and LICI-3 all locate in large granules in hemocytes. Therefore, in limulus, all the serine protease zymogens for coagulation and the three serpins with different specificities co-localize in the same granules, which means an effective coagulation and regulation at local lesions. In mammalian blood coagulation system, factor Xa bound to platelets can be inactivated by antithrombin at much slower rate than the free enzyme, indicating that the effect of antithrombin as a clotting inhibitor is significantly decreased at local lesions (48,49). By analogy, in vivo, factor C and factor G, the initiators of hemolymph coagulation of horseshoe crab, are probably activated on the surfaces of Gramnegative bacteria and fungi, respectively, and the bound proteases could be protected from inactivation by LICIs. An important function of LICIs may be as scavengers of the proteases that have escaped into the hemolymph from the site of injury.