INTRODUCTION

A major problem of comparative immunology is the identification and characterization of the immunoglobulin-independent defense systems that lyse foreign cells, such as microbial pathogens (1, 2). Lysis must be selective to avoid collateral damage to the organism's own tissues, necessitating the discrimination between host and foreign cells. In vertebrates, cytolysis is mediated by the complement system, which is comprised of an ensemble of serially activated proteases and effector proteins and a number of receptors and regulatory proteins (3). In the invertebrates, recognition and lysis must be independent of immunoglobulin-based antibodies because antibodies are restricted to the vertebrates (4). The possibility that invertebrates may possess isolated elements of the vertebrate complement system has been suggested (5, 6, 7, 8, 9), but as yet there has been no convincing demonstration that any invertebrate has anything resembling the complement activation cascade characteristic of vertebrates (10).

The lectins, proteins that bind selectively to limited subsets of complex carbohydrates, are a potentially important class of recognition proteins that are found in invertebrates (11). Lectins are one of the most widely distributed classes of recognition molecules, with representatives in bacteria, animals, and plants (12, 13). We document here that limulin (14, 15), one of the plasma lectins of Limulus, serves as the principal cytolytic agent in the plasma of that organism. Limulin has sialic acid- and 2-keto-3-deoxyoctonate-binding affinities and is a member of the pentraxin protein family (16). In common with other pentraxins, it also binds phosphorylethanolamine and phosphocholine. The best studied pentraxins in mammals, C-reactive protein and serum amyloid protein, are acute-phase proteins (17) whose unique function(s) are still uncertain because of considerable functional overlap with other constituents of the plasma. Although not directly cytolytic, human C-reactive protein can activate the classical pathway of the complement system (18, 19). In Limulus, limulin is necessary and sufficient for the expression of a hemolytic activity found in the plasma. Limulin-free plasma lacks the hemolytic activity of whole plasma and purified limulin is cytolytic at 3-5 nM. Hemolysis is dependent on the sialic acid-binding capabilities of limulin, because the process is inhibited by sialylated glycoconjugates and by desialylation of the target erythrocytes. The taxonomy of the pentraxins and sialic acid-binding lectins in Limulus has been predicated on the suppositions that limulin and Limulus C-reactive protein are identical and that limulin is the only sialic acid-binding lectin in the plasma (16). We show here that both suppositions are incorrect because we can separate limulin from the much more abundant pentraxin, Limulus C-reactive protein, and Limulus plasma contains other, nonpentraxin sialic acid-binding lectins.¹

MATERIALS AND METHODS

Blood was obtained from adult Limulus by cardiac puncture under sterile, lipopolysaccharide-free conditions from pre-chilled animals as described previously (20), and the blood cells were removed immediately by centrifugation. Animals were released into the ocean unharmed after bleeding. Most of the hemocyanin was removed from the plasma by ultracentrifugation (141,000 × g for 16 h; Ref. 21) or by incubation with 3% polyethylene glycol-8000 (PEG)² with centrifugation at 30,000 × g for 0.5 h. The supernatant was then made 10% in PEG and centrifuged as above, and the precipitate was redissolved in buffer A (0.15 M NaCl, 10 mM CaCl₂, 50 mM Tris, pH 8.0). This fraction (3-10% PEG cut) was then depleted of Sepharose-binding proteins by passage over a column of Sepharose 4B (Pharmacia Biotech Inc.) equilibrated with buffer A (0.2 volumes of resin/volume of 3-10% PEG cut). The unbound fraction then was incubated with phosphorylethanolamine-agarose (PE-agarose) (Sigma), which binds the pentraxins (0.1 volumes of resin/volume of plasma). The PE-agarose was washed with buffer A modified to contain 1 M NaCl and eluted with 0.1 M sodium citrate, pH 6.7, to recover the pentraxin fraction. Following dialysis into buffer A, the pentraxin fraction was further fractionated by passage over a column of fetuin-Sepharose equilibrated with buffer A. The breakthrough fraction from the fetuin-Sepharose column is Limulus C-reactive protein. The bound fraction, which was subsequently eluted with 0.1 M sodium citrate, pH 6.7, is the lectin limulin.

The hemolytic activity of the plasma was determined in duplicate or triplicate samples using sheep red blood cells (22, 23, 24). Unactivated sheep erythrocytes in Alsevers solution were obtained from Cappel (reference number 55875; West Chester, PA) and Becton Dickinson (reference number 12388; Cockeysville, MD). The buffer system was modified DGVB buffer (0.19 M NaCl, 0.18 mM CaCl₂, 0.5 mM MgCl₂, 2.5% glucose, 0.1% gelatin, 2.5 mM sodium barbital, pH 7.3). The reaction mixtures contained 3 × 10⁷ washed sheep red cells, the sample to be tested, and the modified DGVB buffer to a final volume of 800 µl. The reaction mixtures were incubated with shaking at 22-23 °C for 4 h, and the reaction was terminated by adding 2 ml of ice-cold phosphate-buffered saline containing 5 mM EDTA, followed by centrifugation to remove the red cells. The extent of hemolysis was determined by monitoring released hemoglobin in the supernatant by the optical absorbance at 412 nm and was compared with full hemolysis produced by hypotonic lysis of the red cells.

The hemagglutination assay was performed using sheep erythrocytes obtained from Becton Dickinson (reference number 12388; Cockeysville, MD). Sheep erythrocytes were washed and suspended at 2% (v/v) in buffer A. 25-µl volumes of 2-fold serially diluted samples dissolved in buffer A were mixed with 25 µl of the erythrocyte suspension in 200-µl round bottom microtiter wells (Sigma, catalog number M-4029), incubated for 45 min at room temperature, and scored for hemagglutination. Hemagglutinated erythrocytes formed a uniform mat covering the entire curved lower surface of the microtiter well (``umbrella formation''); nonagglutinated erythrocytes formed a compact pellet at the very bottom of the well (``button formation'') (25). The hemagglutination end point was the highest dilution of sample that produced visible agglutination.

Desialylation of the red cells was accomplished by treatment with neuraminidase (EC) from Clostridium perfringens (0.3 ml of packed erythrocytes, 15 min, 37 °C, 5 × 10³ units in 0.3 ml of 50 mM phosphate, pH 7.5) and Vibrio cholerae (3 × 10¹⁰ erythrocytes, 30 min, 37 °C, 0.1 units in 0.3 ml of 0.15 M NaCl, 0.05 M sodium acetate, pH 6.5) (26, 27). Sialic acid was measured by the thiobarbituric acid method (28).

SDS-polyacrylamide gel electrophoresis (PAGE) was carried out according to the method of Laemmli (29) using a 4% stacking gel and a 10% resolving gel. Samples were reduced with 5% 2-mercaptoethanol. Gels were silver stained to visualize protein bands (30). Tris-tricine SDS-PAGE was performed according to Schägger and von Jagow (31). Samples were boiled in reducing Tris-tricine sample buffer, loaded on a precast 16.5% acrylamide gel, and electrophoresed for 2 h at 100 V. Gels were stained with 0.2% Coomassie Brilliant Blue in 50% methanol, 10% acetic acid (v/v) and destained with 20% methanol, 10% acetic acid.

Purified limulin and C-reactive protein samples destined for peptide sequencing were electrophoresed on 15 × 15-cm SDS-polyacrylamide gels, and then the protein was transferred to Immobilon polyvinylidene difluoride membranes (Millipore) in 10 mM CAPS buffer, pH 11, for 4 h at 600 mA. The membranes were washed three times in CAPS buffer to remove residual glycine and were stained for 10 min with 0.2% Coomassie Brilliant Blue R-250 in 45% methanol, 10% acetic acid. The stained protein band representing limulin was carefully cut out, and an amino-terminal sequence was determined in an Applied Biosystems 470 gas phase sequinator. Gel filtration chromatography of limulin and Limulus C-reactive protein utilized a 1.6 × 100-cm column of Sephacryl 300HR (Pharmacia). A 1-ml sample was loaded onto the column and eluted with buffer A at 1 ml/min. For CNBr cleavage, 56-µg samples of pure limulin and pure C-reactive protein were dried using a SpeedVac lyophilizer and were then suspended in 60 µl of 70% formic acid (Fisher), and a small crystal of CNBr was added and dissolved as described (32). The samples were incubated in the dark for 16 h at 25 °C, after which distilled water was added to bring the acid content to 7% and to stop the reaction. The samples were dried using a SpeedVac lyophilizer and resuspended in Tris-tricine electrophoresis sample buffer plus 2% 2-mercaptoethanol.

RESULTS AND DISCUSSION

The pentraxins are high molecular mass proteins (~300 kDa) organized as double-stacked pentameric or hexameric assemblies of smaller (25-30 kDa) subunits (33) that bind phosphorylcholine and/or phosphorylethanolamine. In Limulus, where they are the second most abundant protein class in the plasma (34), the pentraxins are represented by at least two different proteins, limulin, which binds sialylated glycoconjugates, and Limulus C-reactive protein, which lacks sialic acid-binding lectin activity. It was previously reported that limulin and Limulus C-reactive protein are the same protein (16). Although both proteins are pentraxins, they do possess distinct ligand affinities. Tandem affinity chromatography of Limulus plasma on PE-agarose and fetuin-Sepharose reveals that limulin and Limulus C-reactive protein can be resolved as distinct proteins (Fig. 1). When hemocyanin-depleted plasma (Fig. 1, lane 2) is passaged over PE-agarose, the pentraxins are bound and can be eluted with 0.1 M sodium citrate (Fig. 1, lane 3b). The breakthrough fraction of the PE-agarose column (Fig. 1, lane 3a) contains the other plasma proteins including ₂-macroglobulin and residual hemocyanin. When the eluted pentraxin fraction is subsequently passaged over fetuin-Sepharose, the lectin limulin, which reacts with sialylated glycoconjugates such as fetuin, is bound and can be eluted with 0.1 M sodium citrate (Fig. 1, lane 3d). Limulus C-reactive protein, the major plasma pentraxin, which lacks sialic acid binding activity, is found in the breakthrough fraction from the fetuin-Sepharose column (Fig. 1, lane 3c). Limulin represents less than 1% of the total pentraxin protein.

When the order of affinity chromatography is reversed, Limulus C-reactive protein partitions into the fetuin-Sepharose breakthrough fraction (Fig. 1, lane 4a). The fraction that binds to and is eluted from fetuin-Sepharose (Fig. 1, lane 4b), which represents less than 0.5% of the plasma protein, can be further fractionated by passage over PE-agarose. Limulin, a pentraxin, will bind to and can be eluted from PE-agarose in purified form (Fig. 1, lane 4d). The other sialic acid-binding lectins will appear in the breakthrough fraction from the PE-agarose column (Fig. 1, lane 4c). These two different affinity fractionation procedures demonstrate that the plasma lectin limulin has the affinity for phosphorylethanolamine of a pentraxin but can be separated from the major pentraxin, Limulus C-reactive protein, by its ability to bind sialylated glycoconjugates.

Although limulin and Limulus C-reactive protein are different proteins, they clearly are closely related. Both proteins bind to phosphorylethanolamine-affinity resins, both have the same native molecular masses (~300 kDa, by size exclusion chromatography on Sephacryl S-300 HR resin), and we have determined that limulin has the same amino-terminal peptide sequence (Leu-Glu-Glu-Gly-Glu-Gly-Ile-Thr-Ser-Lys-Val) as does Limulus C-reactive protein (35, 36), and a polyclonal antiserum produced against purified Limulus C-reactive protein cross-reacted with limulin (data not shown). However, the two proteins show differences in the pattern of protein bands seen by SDS-PAGE (Fig. 2) and differences in the peptides generated by proteolytic fragmentation (data not shown). By SDS-PAGE, C-reactive protein appears to be a heteromultimer composed of two major subunits of apparent molecular masses of 29 and 31 kDa, whereas limulin consists of a single subunit of apparent molecular mass of 33 kDa. The pattern of fragmentation by CNBr is markedly different (Fig. 3). C-reactive protein fragments into two major peptides of apparent molecular masses of 22 and 5 kDa, and limulin is cleaved into four peptides of apparent molecular masses of 6, 7, 8, and 10 kDa. More significantly, the two proteins show important functional differences; limulin, but not Limulus C-reactive protein, binds to sialic acid (Fig. 1) and lyses target sheep erythrocytes (see below).

Limulus plasma contains a cytolytic activity capable of lysing sheep erythrocytes (5, 9, 37, 38). Limulin is necessary for hemolysis because plasma that had been depleted of the pentraxins by passage over PE-agarose (Fig. 1, lane 3a) or that had been depleted of sialic acid-binding lectins by passage over sialomucin- or fetuin-Sepharose (Fig. 1, lane 4a) was hemolytically inactive (Table I). Hemolytic activity was restored by the addition of limulin purified by sequential affinity binding to PE-agarose and fetuin-Sepharose (Table II). Limulin is sufficient for hemolysis because purified limulin was hemolytic in a Ca⁺²-dependent manner at 3-5 nM in the absence of other plasma components (Fig. 4). The hemolytic activity of purified limulin is dependent on its sialic acid-recognition capabilities, because limulin-mediated hemolysis was abolished by desialylation of the target erythrocytes with V. cholerae or C. perfringens neuraminidase and was reduced 50% by inclusion of 0.1 M N-acetylneuraminic acid in the incubation medium. Removal of sialic acid by neuraminidase treatment of erythrocytes was verified by the thiobarbituric acid method on ghosts of the enzyme-treated cells (27, 28). Consistent with the proposition that limulin is the principal cytolysin in plasma, the hemolytic activity of hemocyanin-depleted plasma also was abolished by desialylation of the erythrocytes or by inclusion in the hemolysis buffer of 9 µM fetuin or 2 µM transferrin, both sialic acid-containing glycoproteins. The hemolytic activity of plasma was reduced by 50% by 0.1 M N-acetylneuraminic acid or 1.3 µg/ml colominic acid, a polysialic acid (Sigma catalog number C-5762, lot number 34H0034), whereas galactose, mannose, and N-acetylglucosamine (the other sugars of the oligosaccharide chains of fetuin) failed to inhibit hemolysis by hemocyanin-depleted plasma at 0.19 M.

Table I. Purification of Limulin from Limulus plasma: recovery of hemolytic activity Purification step Gel lane (Fig. 1) Protein Hemolytic activity Specific activitya Yieldb mg units/mg % Sepharose 4B flow throughc 2 519 1.53 100 Purification scheme A: PE-agarose followed by fetuin-Sepharose PE-agarose flow through 3a 151 0.0d PE-agarose boundc 3b 278 1.55 54 Fetuin-Sepharose flow throughf 3c 275 0.0d Fetuin-Sepharose boundg 3d 1.3 323 41 Purification scheme B: fetuin-Sepharose followed by PE-agarose Fetuin-Sepharose flow through 4a 502 0.0d Fetuin-Sepharose boundh 4b 1.7 224 47 PE-agarose flow throughi 4c 0.2 0.0d PE-agarose boundg 4d 0.9 383 43 a One unit of activity is that required to produce 50% hemolysis of sheep erythrocytes under standard conditions. b Yield of hemolytic activity. c 3-10% PEG-8000 cut of 100 ml of plasma passed over Sepharose to remove the Sepharose-binding proteins. This was the starting material for fractionation by purification schemes A and B. d This sample produced less than 5% hemolysis at the maximum volume permissible in the assay (400 µl). e Total pentraxins. f Limulin-free pentraxins. g Purified limulin. h Total sialic acid-binding lectins. i Nonlimulin sialic acid-binding lectins.

Table II. Reconstitution of the hemolytic activity of depleted plasma by limulin Sample Amount A280a Addition Hemolysis ml % Hemocyanin-free plasmab 0.1 5.6 53.7 Hemocyanin-free plasma passed over PE-agarosec 0.3 2.8  0.3 Hemocyanin-free plasma passed over PE-agarose 0.3 2.8 20 nM limulin 66.1 Hemocyanin-free plasma passed over fetuin-Sepharosed 0.3 4.8 1.3 Hemocyanin-free plasma passed over fetuin-Sepharose 0.3 4.8 20 nM limulin 71.2 a Optical absorbance of the plasma sample at 280 nm. b Plasma centrifuged 141,000 × g for 16 h. c Plasma lacking pentraxins; equivalent to lane 3a of Fig. 1. d Plasma lacking the sialic acid-binding lectins; equivalent to lane 4a of Fig. 1.

As indicated above, limulin can be separated from the other sialic acid-binding lectins in the plasma of Limulus by its unique ability to bind to PE-agarose (Fig. 1, lanes 4a-4d). Although the other sialic acid-binding lectins of Limulus plasma agglutinated sheep erythrocytes, they showed no ability to lyse the cells when they were present at hemagglutination titers equivalent to or greater than those of active concentrations of limulin (Table III). Thus hemolysis is not produced by any and all sialic acid-binding lectins and is not a direct result of hemagglutination per se.

Table III. Hemolytic and hemagglutinating activity of Limulus plasma lectins Protein Amount Hemagglutination titera Hemolysis µg % hemolysis Limulin 28 10 2.6 41 15 40 46.5 17 77 Nonlimulin sialic acid lectins 20 16 0.0 40 32 0.3 60 48  0.1b a Hemagglutination titer is the reciprocal of the greatest dilution that produces detectable agglutination of sheep erythrocytes. b Below background.

This study clarifies several issues of the functional and biochemical characterization of the pentraxins in Limulus. Firstly, it identifies the major cytolytic protein in the plasma as the lectin, limulin. Presumably, the recognition of foreign cells for subsequent cytolytic destruction by this system depends on the presentation of sugars recognized by limulin on the surfaces of the foreign cells. Although a number of plasma lectins have been identified in invertebrates, the physiological function(s) of this class of proteins is for the most part not well characterized (39, 40). As far as we are aware, this is the first demonstration of a direct cytolytic activity for a plasma lectin. Secondly, we show that limulin is but one of several plasma lectins in Limulus with sialyl specificity, although only limulin has cytolytic activity in the hemolysis assay. Finally, we have clarified the relationship of limulin and C-reactive protein in Limulus by showing that limulin is not identical to C-reactive protein but is a low abundance member of the pentraxins with the specialized properties of sialic acid-binding and hemolytic activities.