Purification and macromolecular properties of a sialic acid-specific lectin from the slug Limax flavus.

A lectin (LFA) which is highly specific for sialic acid has been purified from the slug Limax flavus by a combination of ammonium sulfate fractionation and affinity chromatography on bovine submaxillary mucin coupled to Sepharose 4B. The affinity-purified lectin appeared homogeneous by electrophoresis in the presence of sodium dodecyl sulfate. Below 1 mg/ml at pH 7, LFA exists as a species of Mr = 44,000 which is composed of two equal sized subunits. Above 1 mg/ml, the protein solution was observed to behave as a rapidly associating-dissociating system. N-acetylneuraminic acid and N-glycolylneuraminic acid gave a 50% inhibition of agglutination of erythrocytes by LFA at 0.13 and 0.81 mM, respectively. Galactose, N-acetylgalactosamine, galactosamine, glucose, N-acetylglucosamine, glucosamine, mannose, arabinose, xylose, fucose, glucuronic acid, alpha-methyl-D-glucoside, alpha-methyl-D-mannoside, lactose, and sucrose were ineffective inhibitors at concentrations up to 10-25 mM. Bovine submaxillary mucin, a sialoprotein, was a potent inhibitor of hemagglutination by LFA. Upon treatment of the mucin with neuraminidase, loss of inhibitory activity was observed which was proportional to the loss of sialic acid from the mucin.

Lectins are a group of proteins that interact with glycoproteins and glycolipids by binding to specific carbohydrate residues (Lis and Sharon, 1973;Sharon and Lis, 1975;Pereira and Kabat, 1979;Goldstein and Hayes, 1978). Because of the high degree of specificity exhibited by individual lectins in their interaction with glycolipids and glycoproteins, they have been employed ever increasingly as highly discriminating agents in studies of membranes of normal and cancerous cells, in blood typing, in the purification of glycoproteins, in the subfractionation of cell populations, and in studies of the mitogenesis of lymphocytes.
In spite of their ubiquitous nature, relatively few lectins which are specific for N-acetylneuraminic acid have been identified. Because of their macromolecular properties and/or because their specificity is not limited to sialic acid, the lectins which have been purified and studied to date have not proven to be fully satisfactory for studies like those enumerated above. Among the prominent examples of sialic acid binding lectins are limulin and carcinoscorpin, which have been purified from the hemolymph of the American horseshoe crab * These studies were supported in part by a grant from South Carolina State Appropriations for Biomedical Research. 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.

Recipient of United States Public Health Service Grants
Limulus polyphemus (Marchalonis and Edelman, 1968) and the Indian horseshoe crab Carcinoscorpius rotunda cauda (Bishayee and Dorai, 1980), respectively. It has been reported, however, that limulin will also bind to N-acetylglucosamine (Marchalonis and Edelman, 1968) and glucuronic acid (Nowak and Barondes, 1975) while the carbohydrate binding properties of carcinoscorpin have not been extensively studied. Additionally, the large size of liiulin (Mr -350,000) and its tendency to dissociate (Marchalonis and Edelman, 1968) have made it somewhat more difficult to handle. An agglutinin (LAg-1) which binds AcNeu' has been isolated from the lobster, Homarus americanus (Hall and Rowlands, Jr., 1974). Hemagglutination by this lectin was inhibited by AcNeu, GlyNeu, and ManNAc. These studies also indicated that LAg-1 contains a second binding site of undefined specificity. The agglutination of human erythrocytes by wheat germ agglutinin is inhibited by AcNeu and AcNeua2 -+ 3-lactose; however, compounds such as GlcNacPl -+ 4GlcNAc and GalPl "+ 6GlcNAcP + C6H4NOJp) were much more potent inhibitors (Bhavanandan and Katlic, 1979), It was earlier observed (Pemberton, 1970) that extracts of the slug Liman flaws contain hemagglutinins for erythrocytes. Extending Pemberton's observation, we have recently reported (Miller, 1981) that hemagglutination of erythrocytes by L. flavus extracts is strongly inhibited by AcNeu and GlyNeu and to a lesser extent by GlcN, GlcNAc, GalN, and GalNAc. It is the purpose of this communication to describe the purification and physical characterization of a lectin (LFA) from L. f l a w s which exhibits a narrow specificity for sialic acid. Its narrow range of specificity together with the lectin's macromolecular properties suggest that the L. flauus lectin has excellent potential as a discriminator for sialic acid-containing glycoproteins.

MATERIALS AND METHODS
The carbohydrates were obtained from Sigma. The N-acetylneuraminic acid was Type VI from Escherichia coli. Sepharose 4B was obtained from Pharmacia. Chemicals employed were reagent grade unless otherwise indicated. Charcoal-treated and doubly deionized water was used in all preparations. The slugs, L. flavus, were collected from the Charleston, SC area.
Coupling of Bovine Submaxillary Mucin to Sepharose 4B-Twenty ml of Sepharose 4B were washed with 200 ml of deionized water by filtration, and the excess water was removed. The Sepharose was subsequently washed with 20 ml of 2 M potassium phosphate (pH 12) by flitration and then placed in another 20 ml of the phosphate buffer. Two and one-half ml of dioxane containing 1.0 g of cyanogen bromide were added over a 6-min period while stirring the Sepharose Tris-saline buffer, 50 mM Tris-Cl,IOO mM NaCl, pH 7.5 at 5 "C; BSM, Sialic Acid-specific Lectin from Limax flavus 7575 in an ice bath. The Sepharose was stirred for an additional 10 rnin and washed with 400 ml of deionized water. Ten ml of BSM (10 mg/ m l ) in 1.0 M NaCl and 10 ml of 0.2 M NaHC03, pH 9.0, were added to the activated Sepharose and the mixture was stirred for 2 h at 22 "c. The BSM-Sepharose was washed with 100 ml of 1.0 M NaCl and 100 ml of Tris-saline buffer. The absorbance of the washes at 280 and 260 nm indicated that approximately 85% of the BSM had been coupled to the Sepharose. Purification of LFA-Approximately 150 g of slugs were washed with deionized water, eviscerated, and further washed with Tris-saline buffer to remove surface mucous material. One hundred g of washed tissue were placed in 400 ml of Tris-saline buffer containing 4.0 ml of phenylmethanesulfonyl fluoride (5 mg/ml in 2-propanol). The tissue was minced with scissors and then homogenized by use of a Polytron homogenizer at full speed for 3 min. The homogenate was centrifuged at 16,000 X g for 15 min and the supernatant fraction was saved (SF-16). Powdered ammonium sulfate was added to the SF-16 to 40% saturation and the precipitate was collected by centrifugation at 16,000 x g for 15 min. The supernatant fraction was decanted and adjusted to 80% saturation by additions of powdered ammonium sulfate. The precipitate (AS-40-80) was collected by centrifugation at 16,000 x g for 15 min. The AS-40-80 pellets were resuspended in 200 ml of Tris-saline buffer and dialyzed against three I-liter volumes of this buffer at 5 "C for 24 h.
A column (1.6 X 10 cm) of BSM-Sepharose was equilibrated with Tris-saline buffer. One hundred ml of the dialyzed AS-40-80 fraction at approximately 10 mg of protein/ml were pumped (50 ml/h) onto the column. The column was washed with Tris-saline buffer until the nonbound proteins were eluted. The lectin was then eluted with Trissaline buffer containing 10 mM AcNeu. The fractions containing protein which eluted with AcNeu were combined and dialyzed against Tris-saline buffer before storage at -70 O .
Removal of Bound AcNeu from Purified LFA-Extensive dialysis of affinity-purified LFA against Tris-saline did not completely remove AcNeu. Assays of dialyzed LFA by the thiobarbituric acid assay of Warren (1959) indicated that the LFA contained approximately 8 mol of noncovalently bound AcNeu/mol of LFA. Chromatography of LFA on a column (1 X 37 cm) of Dowex 501-X8(D) which had been equilibrated with 0.01 M ammonium acetate, pH 7.0, reduced its AcNeu content to less than 0.20 mol of AcNeu/mol of LFA with little or no change in the specific activity of the LFA preparation. Because the yield of protein from such a treatment was -70% and because the removal of AcNeu had no effect on the lectin's specific activity, samples of LFA which had been chromatographed'on the ion-exchange column were utilized for only selected experiments.
Hemagglutination Assay-Aliquots of LFA were adjusted to 0.5 ml by addition of Tris-saline buffer which contained 0.4 mg/ml gelatin. To this was added 0.5 ml of a solution containing human erythrocytes that had been washed several times with 0.9% NaCl and whose absorbance at 620 nm was -2. After standing at room temperature for 30 min, the cells were gently pelleted by centrifugation in a Dynac centrifuge. The cells were resuspended by shaking and allowed to stand for 5 min in order to permit aggregated cells to settle. The absorbance at 620 nm of the upper 0.5 ml of the erythrocyte suspension was measured and the data were plotted as AB^^ versus micrograms or microliters of lectin. One unit of activity is defined as the amount of LFA which gave a 50% agglutination (Miller, 1981). Per cent agglutination was calculated as follows: Inhibition of Hemagglutination-An amount of purified LFA or partially purified LFA (i.e. AS-40-80 fraction) which gave greater than 95% agglutination of erythrocytes in the hemagglutination assay was mixed with varying amounts of potential inhibitors dissolved in Tris-saline buffer, and the volume was adjusted to 0.5 ml by addition of Tris-saline buffer. A 0.5-ml aliquot of erythrocytes was added to the lectin plus inhibitor solution and the per cent agglutination determined as described above. The per cent inhibition of agglutination represents the difference between the per cent agglutination with lectin alone and that obtained with lectin plus inhibitor.
ChemicalAnalyses-Amino acid analyses of LFA were determined on a computerized Durmm 600 amino acid analyzer. Protein samples were hydrolyzed at 110 "C in constant boiling HCI for 24, 48, and 72 h. Corrections were made for loss of threonine and serine during hydrolysis. Cysteine was determined after performic acid oxidation of the protein. Tryptophan and tyrosine were estimated by the spectroscopic method of Edelhoch (1967) as modified by Bredderman (1974). AcNeu was measured after acid hydrolysis or neuraminidase treatment of glycoproteins by the thiobarbituric acid assay described by Warren (1959). The acid hydrolytic release of AcNeu from glycoproteins was accomplished by incubation of the glycoprotein in 0.1 N H2S04 at 80 "C for 1 h.
Physical Measurements-Sedimentation velocity and equilibrium measurements were performed in a Beckman Model E analytical ultracentrifuge equipped with schlieren/interference optics and a photoelectric scanner.
Sedimentation velocity measurements were made according to standard procedures; no corrections were made for the Johnston-Ogston effect. Meniscus depletion sedimentation equilibrium measurements employed the high speed (Yphantis, 1964) or the long column (Chervenka, 1970) methods. For all sedimentation equilibrium runs, attainment of equilibrium was checked by measuring the fringe displacements at several radial distances of two successive exposures taken 3 to 6 h apart. The partial specific volume used for LFA, as calculated from its amino acid composition, was 0.724 ml/g. For LFA at pH 2, a correction was included for the Donnan effect on the equilibrium sedimentation pattern in a three-component system (Johnson et al., 1954;Huston et al., 1972).
Empirical estimations of the hydrodynamic size of native LFA were attempted by gel chromatography on a 50-cm column of Bio-Gel P-200. Standard proteins of known hydrodynamic properties were utilized to calibrate elution positions from the column as a function of the equivalent hydrodynamic radius, Re, (Fish, 1975) or of the molecular weight. Empirical estimations of the size of the constituent polypeptide chains of LFA were by thin slab sodium dodecyl sulfatepolyacrylamide gel electrophoresis (Laemmli, 1970) and by gel chromatography in the presence of 6 M GdmeCI (Fish, 1975) on Sephacryl s-300.
Circular dichroic spectra were measured on a Cary 60 spectropolarimeter equipped with a Model 6002 CD attachment. The CD was calibrated wih IO-d-camphorsulfonic acid (Eastman) (Adler et al., 1973). Cell path lengths from 0.5 to 10 mm were employed to maintain optimal signal-to-noise ratios. Data treatment was in the usual fashion (Adler et al., 1973). A value of 105 was used for the mean residue molecular weight. The differential refractometric method of Babul and Stellwagen (1969) was employed to estimate protein concentration.
Absorption spectra were obtained on a Cary 15 spectrophotometer whose absorbance accuracies were checked with dichromate (Haupt, 1952).

RESULTS
Purification of LFA-Homogenates of whole slugs were highly viscous and were only partially clarified by centrifugation at 30,000 X g for 30 min. Ammonium sulfate precipitates from such homogenates yielded highly viscous solutions when resuspended in buffer; the high viscosity of these solutions greatly hampered further purification of the LFA. It was subsequently found that the problems with highly viscous solutions could be eliminated by evisceration of the slugs and use of only the body tissues for the lectin preparation. About 90% of the total agglutinin activity was associated with the body tissues, about 5-7% was associated with the hemolymph, and less than 2% was associated with the internal organs. Furthermore, most of the viscous material (presumably mucins) was associated with the viscera and thus was eliminated by this procedure.
As indicated in Table I, most of the agglutinin activity present in the tissue homogenate was recovered in the 4040% ammonium sulfate fraction. Subsequent purification by affinity chromatography of LFA on BSM-Sepharose ( Fig. 1) resulted in a 68-fold overall purification of LFA with a 22% recovery of agglutinin activity (Table I). The use of BSM in conjunction with AcNeu as an eluting agent in the affinity purification of LFA should result in the recovery of lectin(s) which bind AcNeu. The low recovery (22%) and the apparently rather low degree of purification (68-fold) of LFA may be due to a separation, at the affinity chromatography step, from other hemagglutinins which are specific for carbohydrate Sialic Acid-specific Lectin from Limax flavus residues other than sialic acid. This suspicion is supported by the data discussed below. In spite of the low recovery of hemagglutinin activity, approximately 17 mg of purified LFA were obtained from 100 g of eviscerated slug tissue. Homogeneity of Purified LFA-Electrophoresis of the affinity-purified LFA on sodium dodecyl sulfate-polyacrylamide gels (12.0%), after reduction and denaturation of the protein by sodium dodecyl sulfate and /3-mercaptoethanol (Laemmli, 1970), yielded a single Coomassie blue-staining band (Fig. 2). The migration of LFA relative to the migration of myoglobin, chymotrypsinogen A, and ovalbumin suggests that reduced and denatured LFA consists of a single polypeptide species of M , -22,000. Some preparations of purified LFA exhibited a faster migrating minor component when a relatively large amount of LFA (>lo pg) was applied to the gel.
Binding Properties of LFA-As indicated in Table 11, hemagglutination by the affinity-purified LFA was inhibited by 50% at 0.13 and 0.81 mM concentrations of AcNeu and GlyNeu, respectively. No inhibition was observed with the other carbohydrates tested, even at concentrations as high as 10-25 mM. In contrast, inhibition of hemagglutination by the 40-80% ammonium sulfate fraction by GalNAc, GlcNAc, and GlcN in addition to AcNeu and GlyNeu (Table 11) suggests that this fraction contains hemagglutinins with binding specificities for carbohydrate residues other than sialic acid. It is unclear why so much agglutinin activity was lost during the affinity chromatography step. No additional agglutinin activ-

TABLE I
Purification of a sialic acid-specific slug lectin One hundred g of eviscerated slug tissue were used in the preparation. " The following notations are used SF-16, the supernatant fraction obtained from a 16,000 X g centrifugation of the tissue homogenate; AS-40-80. proteins which precipitated between 40 and 808 saturation of SF-16 with ammonium sulfate; purified LFA, protein eluted from the BSM-Sepharose column with AcNeu, i.e. affinity-purified LFA.
One unit of activity is defined as that amount of lectin which gives a 50% agglutination of erythrocytes in the hemagglutination assay (see "Materials and Methods").

FIG.
1. Affinity purification of LFA. One hundred ml of the AS-40-80 fraction were pumped (50 ml/h) onto a column (16 X 10 cm) of BSM-Sepharose that had been equilibrated with 50 mM Tris-CI, 100 mM NaCI, pH 7.5. The column was washed with the above buffered solution until the A m of the effluent approached zero. Elution of the column with 50 mM Tris-CI, 100 mM NaCI, 10 mM AcNeu, pH 7.5, was begun at fraction 32. A 1.0-pl aliquot of those fractions indicated was assayed for agglutinin activity. Fractions 34 through 37 were combined, dialyzed against two I-liter volumes of tris-saline buffer, and stored at -70 "C. ity was recovered by further elution of the BSM-Sepharose column with 50 mM AcNeu in Tris-saline or with 50 mM AcNeu in 50 mM Tris-C1, 1.0 M NaC1, pH 7.5. Also, little additional protein was recovered by washing the column with 0.01 N NaOH. Those fractions containing proteins which did not bind to the affinity matrix were without significant hemagglutinin activity (Fig. 1). Loss of agglutinin activity may be due to inactivation of lectins by denaturation or to loss of essential cations.
As mentioned above, the data of Table I1 suggest that the purified LFA binds free AcNeu in a highly specific manner. This specificity was further tested by use of the AcNeu-rich glycoprotein BSM. As shown in Table 111, as little as 9 pg/ml of BSM gave a 50% inhibition of hemagglutination in the standard assay containing 0.34 pg of LFA. Upon treatment of BSM with Vibrio cholerae neuraminidase, there was a timedependent release of AcNeu from the BSM with a concomitant loss in agglutination inhibition. The total content of AcNeu in BSM was 5% (w/w) as measured after its release by acid hydrolysis. Release of 12 pg of AcNeu/mg of BSM, i.e. 25% of its total AcNeu, virtually abolished its inhibitory activity. Thus, it appears that only about 25% of the AcNeu residues in BSM are the functional ligands in its interaction with LFA. This suggests that BSM may contain AcNeu resi- Inhibition of hemagglutinin activity of slug lectin by carbohydrates

Sialic Acid-specific Lectin from Limax f l a w s
The standard hemagglutination assay was used in these comparative studies. Twenty pg of protein from the 40-80% ammonium sulfate fraction (AS-40-80) or 0.34 pg of the purified LFA was utilized in each assay. at 37 "C. Aliquots were removed at times indicated above, diluted 10fold with 0.05 M Tris-C1, 0.1 M NaCl, pH 7.5, and assayed with 0.34 pg of slug lectin for inhibitory activity. Aliquots were assayed for released AcNeu by the thiobarbituric acid assay of Warren (1959).
The BSM was incubated as described above except the neuraminidase was omitted from the incubation mixture.
dues that are inaccessible to LFA or that the LFA has a specificity for a sialic acid which is preferentially cleaved (Corfleld et al., 1981) by the V. cholerae neuraminidase.
Physiochemical Properties of LFA-The amino acid composition of LFA is summarized in Table IV. The lectin contains 40 residues of Glx plus Asx and a total of 32 residues of Lys, Arg, and His/subunit. The high content of basic amino acids and the basic PI of the protein suggest that a relatively large proportion of the Glx and Asx are present in the protein in the amidated form.
Although LFA exhibited a single electrophoretic component under dissociating and denaturing conditions (Fig. 2), appreciable size heterogeneity was evident during sedimentation velocity measurements at protein concentrations greater than 1 rng/ml. As illustrated by Fig. 3, these solutions exhibited two boundaries in the analytical ultracentrifuge. The magnitude of the sedimentation coeffkient of the faster sedimenting species exhibited a marked positive dependence on the protein concentration, and its relative amount decreased with decreasing protein concentration to the point that at concentrations below 1 rng/ml only a single species of 3.4 S was observed. The 3.4 S species was present under all conditions and its sedimentation coefficient exhibited a slight positive concentration dependence. The relative ratios of the ultracentrifuge species could not be perturbed with AcNeu or chelating agents (Fig. 3). That these sedimentation velocity data are consistent with a rapidly associating-dissociating protein system (Gilbert, 1959;Gilbert, 1963) is supported by the behavior of the near  UV circular dichroism spectra of LFA at protein concentrations at or below which only the 3.4 S species could be observed in the analytical ultracentrifuge.
The amplitude of the ellipticity band at 282 nm increased with increasing protein concentration above -0.6 mg/ml; this increase produced a marked qualitative change in the near UV spectrum (cf curues I and 2 of Fig. 4A). Below -0.6 mg/ ml, the near UV spectra were qualitatively identical, but the amplitudes of the ellipticity bands fluctuated slightly among different preparations. In the absorbance range which could A, near UV spectra. Measurements were made in 1-cm cuvettes at the protein concentrations indicated. Curve S, the absorption spectrum at pH 7.5. The circular dichroic spectra in 0.1 M NaCl, 0.05 M Tris-C1, pH 7.5, are as follows: curve 1, protein concentration = 0.60 mg/ml. Concentrations less than this gave essentially identical spectra. Curve 2, protein concentration = 1.41 mg/ml. Curve 3, deionized LFA with 40 mol of AcNeu/mol of protein; protein concentration = 0.57 mg/ml. B, far UV circular dichroic spectra. Measurements were made in 0.5-to 1.0-cm cuvettes at the protein concentrations specified. Curve I , protein concentration 50.6 mg/ml for three different preparations at four different concentrations (below 0.6 mg/ ml). The range of ellipticity values was 13% of the curve shown; curve 2. Drotein Concentration ( 1.41 mdml; curve 3, the unreduced protein be measured at 280 nm (i.e. protein concentrations ~1 mg/ ml), LFA obeyed Beer's Law (absorption spectrum given in Fig. 4A). Thus, the near UV spectral data suggest that the increased amplitudes of the near UV circular dichroism bands reflect an induced or enhanced chiral environment of one or more aromatic side chains upon association of lectin molecules. Slight increases in the ellipticity values were observed at all wavelengths in the far UV at protein concentrations above -0.6 mg/ml (curues I and 2 of Fig. 4B), but the magnitudes of these increases, up to -13%, can be accounted for by the induced chirality of the aforementioned aromatic amino acid residues (Beychok, 1964;Goodman et al., 1968).
A cross-linked polyacrylamide gel chromatographic support medium was employed in an attempt to utilize it as a purification step and to estimate the equivalent hydrodynamic radius of native LFA. This type of support was employed rather than a gel of polysaccharide matrix to diminish the possibility of lectin-gel interactions. However, the elution position of LFA from the polyacrylamide gel was markedly dependent upon the ionic strength of the eluting buffer; at r/0.14, it eluted at a position equivalent to a protein of M, = 15,000, and at I'/1.04, it eluted at a position equivalent to a protein of M, = 37,000. This was apparently the result of ionic attractions between the basic protein and the support medium since a similar behavior was observed for chymotrypsinogen A, the only standard protein utilized with a pI (Kubacki et al., 1949) similar to that of LFA. Lines 1, 2, and 3 of Table V summarize the results of our sedimentation equilibrium measurements on native LFA. Linear pIots of In (fringe displacement) versus (radius)' were obtained from individual runs in dilute buffer at protein concentrations between 0.5 and 0.05 mg/ml. When examined collectively, however, the weight-average molecular weights  from these runs exhibited a tendency to increase slightly with increasing initial protein concentration. Neither the presence of a 30-fold molar excess of AcNeu (line 3) nor the complete absence of AcNeu (line 2) had an effect on the sedimentation equilibrium behavior of the protein. The native molecular weight estimated for LFA was 44,000 2,000. From this molecular weight, an of 3.4 S, and a 6 of 0.724 mg/ml, the estimated f/fmin of 1.35 suggests that LFA is a globular protein.
A limited number of sedimentation equilibrium measurements at or below pH 3 suggest that like a number of other lectins (McKenzie et al., 1972;Nagata and Burger, 1974;Fish et al., 1978) the quaternary structure of LFA is disrupted at low pH (Table V, line 4).
The molecular weights of the constituent polypeptide chains of LFA were estimated by sedimentation equilibrium in 6 M Gdm . C1 with and without reduction of disulfide bonds.
In both cases (Table V, lines 7 and 8), the data were consistent with two polypeptide chains of the same molecular weight which are held together in the native molecule by noncovalent interactions. The circular dichroic spectrum of LFA in 6 M Gdm-Cl (Fig. 4B, curve 3 ) supports the assumption that all noncovalent interactions in the protein were broken and that the minimal subunit had been obtained. These results are also supported by empirical molecular weight estimation methods in denaturing solvents (Table V, lines 9 and 11). A comparison of the gel chromatographic behavior in 6 M Gdm. C1 between reduced and unreduced LFA suggests that there is only limited restraint imposed on the polypeptide by intrachain disulfide bonds (Table V, lines 9 and 10). This observation is consistent with the half-cystine content of the protein (Table IV) which suggests a maximum of only 1.5 disulfide bonds/100 amino acid residues (Tanford, 1968).
When LFA was exposed to a 40-fold molar excess of its receptor ligand, AcNeu, the only change observed in the circular dichroic spectrum of the protein was a diminution of about 10% in the amplitudes of the ellipticities between 240 and 285 nm (cfi curves Z and 3 of Fig. 4A). Although this is a rather subtle change, it is consistent with previous observations on other lectins where the carbohydrate(s) for which the lectin is specific elicit a change in the near UV circular dichroic spectra (Pflumm et al., 1971;Thomas et al., 1977;Shimazaki et al., 1975;Pere et al., 1975;Fish et al., 1978;Decastel et al., 1981). The far UV circular dichroic spectrum of LFA from which all free AcNeu had been removed exhibited no measurable change upon exposure of the lectin to a 40-fold molar excess of AcNeu.

DISCUSSION
This report describes the purification and characterization of a highly specific sialic acid binding lectin (LFA) from the slug L. pavus. This lectin is readily purified in milligram quantities (-17 mg/100 g of eviscerated slug tissue) by ammonium sulfate fractionation and affinity chromatography of slug tissue extracts. The purified LFA is demonstrated to be homogeneous by a variety of techniques. Sedimentation equilibrium studies indicate that native LFA has M, -44,000 and s&,&, of approximately 3.4 S. LFA consists of two equal sized subunits with M, -22,000 as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and sedimentation equilibrium in 6 M Gdm-Cl. The lectin dissociates into subunits at pH 2.0 and tends to associate into larger molecular species at concentrations above -1 mg/ml at pH 7.5, as evidenced by sedimentation velocity and near UV circular dichroism data.
LFA resembles the horseshoe crab lectins carcinoscorpin and limulin in its high content of glutamic acid and aspartic acid residues (Dorai et al., 1981;Roche and Monsigny, 1974).
The high PI, approximately 9-9.5, of LFA suggests that many of its acidic amino acid residues must be amidated. Although LFA contains six residues of cysteine, disulfide bonds are not involved in subunit interactions.
Hemagglutination by affinity-purified LFA was inhibited only by AcNeu and GlyNeu of the many carbohydrates tested. Hemagglutination by partially purified slug extracts was strongly inhibited by AcNeu and GlyNeu and to a lesser extent by GlcN and GlcNAc. These data suggest that the slug contains more than one lectin, some of which do not bind sialic acid. Hemagglutination by LFA was also inhibited by the sialoprotein bovine submaxillary mucin. Treatment of the mucin with neuraminidase resulted in a loss of inhibitory activity which was proportional to the release of AcNeu from the mucin. Thus, the slug lectin LFA appears to be highly specific for sialic acid residues. Although a number of lectins have been very useful in the affinity purification of analytical quantities of glycoproteins, none with a high specificity for sialic acid has been successfully employed in this manner. As a result of its macromolecular properties and high degree of specificity for sialic acid, LFA wiU likely be very useful in studies on the sialoproteins of cellular membranes and in the affinity purification of sialoproteins.