Purification and characterization of two lectins from Caragana arborescens seeds.

A glycoprotein fraction with hemagglutinating activity was purified by affinity chromatography from seeds of the pea tree, Caragana arborescens. Subsequent fractionation resolved two components, which could be separated on a preparative scale using different affinity matrices. The major component binds to N-acetylgalactosamine coupled to Sepharose 4B. It is a glycoprotein with high hemagglutinating activity. It is composed of two types of polypeptides, present in nonstoichiometric amounts, with apparent molecular weights near 30,000. In the native molecule, the subunits are cross-linked by disulfide bonds to form dimers, which in turn appear to be in rapid equilibrium with tetramers. The minor component binds to underivatized Sepharose 4B. It too, is a glycoprotein but has low hemagglutinating activity. It is composed of three types of polypeptides which, although they have apparent molecular weights near 30,000 are distinguished from the subunits of the major hemagglutinin by a number of physical and chemical properties. The native molecule is dimeric, with a mass of 60,000 daltons. The major component has high affinity (K = 0.1 mM) for the haptenic sugar, N-acetylgalactosamine, but will also bind D-galactose. Neither lectin has ABO blood group specificity, nor are they transformed mouse fibroblasts to the same extent.

properties. The native molecule is dimeric, with a mass of 60,000 daltons.
The major component has high affinity (K = 0.1 IIIM) for the haptenic sugar, N-acetylgalactosamine, but will also bind o-galactose.
Neither lectin has ABO blood group specificity, nor are they toxic to cultured mouse fibroblasts. Both agglutinate normal and transformed mouse fibroblasts to the same extent.
Although carbohydrates on the outer surfaces of cells have been intensively studied, their function has not yet been elucidated.
They may exist as structural components, possibly defining the asymmetry of the membrane, or they may play an active role in determining the responses of cells to environmental changes.
The presence of sugar-specific proteins on cell surfaces (l-5) has led some authors ( Whole blood was obtained from the local blood bank. L,,,,, Ehrlich ascites, 3T3, and Simian virus 40.transformed 3T3 (SVlOl-3T3) cells have been carried in this laboratory (9, 11). All cells were washed before use to remove plasma and buffy coat or ascites fluid. Medium and calf serum for tissue culture were purchased from Gibco-Biocult (Glasgow, Scotland). Affinity Chromatography-Sepharose 4B (Pharmacia) was activated with cyanogen bromide (12) for 30 to 45 min. Synthesis of the affinity matrix, Sepharose-2-acetamido-O-(p-aminophenyl)-2-deoxy-~n-galactopyranoside, was as reported (13). The coupled resin contained 0.4 to 0.7 rmol of P-acetamido-0(p-aminophenyl)-2-deoxy-B-Bgalactoside/ml of packed resin and was diluted with 4 volumes of unmodified Sepharose 4B before use, to a total volume of 125 ml. When an alternative method for activating Sepharose with cyanogen bromide was used (14), the ability of the substituted resin to retain agglutinating activity was greatly reduced. Assays-Agglutination of 3T3 and SVlOl-ST3 cells was as described (15) (25).
Puri,?cation-Meal from C. arborescens (10 g), obtained by grinding seed frozen in liquid nitrogen to a fine powder, was suspended in 100 ml of cold buffered saline (consisting of 8 g of NaCl, 0.  Table  I, as is the purification away from contaminating glycosidases. To inhibit binding of @-galactosidase to the affinity column, D-gahCkd, a competitive inhibitor of Escherichia coli @-galactosidase (26), was present in the column buffers.
Gala&al at 0.25 mM inhibits galactosidase activity by 50% at 22", but it has no effect on the agglutination reaction (see below). Its presence, therefore, allows removal of this enzyme from the column while lectin remains bound to the matrix (Fig. 1). Because the peak of agglutinating activity in Fig. 1 was asymmetric and did not coincide with the peak of optical density, we suspected that we had purif'ied more than one protein.
We, therefore, examined our purified fractions by electrophoresis on cellulose acetate strips. Fig. 2 shows that two proteins were resolved. We designate these two species I and II, I being the component migrating more slowly.
Hereafter, the mixture of both proteins obtained by affinity chromatography will be referred to as (I + II). We demonstrated that components I and II were different proteins, and not merely charge isomers, by eluting them from cellulose acetate strips (27)   and (  9.5 x 10-4 9.9 x 10 4 7.9 x 10 I 8.3 x 10 =, a Incubations for the sample obtained after affinity chromatography on Sepharose derivatized with N-acetylgalactosamine were for 10% h, for other samples, 2 h, in buffered saline. FIG. 2. Separation of two species of Caragana lectins by electrophoresis on cellulose acetate. Protein fractions obtained from affinity chromatography were dialyzed against 10 mM sodium phosphate and 10 mM MgCl,, pH 6.5. This pH was chosen after preliminary experiments showed the lectins to have isoelectric points in the range of 5.9 to 6.5. Approximately 100 pg were applied to the cellulose acetate strip, and electrophoresis was for 40 min at 13 mA, 4". The electrophoresis buffer was 20 mM sodium phosphate and 10 mM MgCl,, pH 6.5. Staining was with amido black. lectins were obtained by virtue of their abilities to bind to different components of the affinity matrix. Chromatography of heat-treated extract on a column of underivatized Sepharose 4B, followed by elution with lactose, yielded a protein solution containing only lectin II (Fig. 4A). Passage of the extract through a column of Sepharose derivatized with N-acetylgalactosamine, but not diluted with unsubstituted resin (see "Experimental Procedures"), yielded lectin I with small amounts of contaminating protein upon elution with lactose (Fig. 4B). Lectin I could be separated from this contaminant by overnight dialysis at 4" against 1 mM sodium phosphate and 1 mM MgCl,, pH 6.5. This afforded a precipitate which could be dissolved in buffered saline and which contained only component I. containing, in order of decreasing molecular weight, bovine serum albumin (68,000), pyruvate kinase (57,006), ovalbumin (43,000), glyceraldehyde-3-P-dehydrogenase CWW, carbonic anhydrase (29,000), and lysozyme (14,000) which runs with the tracker dye. Samples were boiled for 3 min in dissociation buffer (25): d tog, in the presence of 1% 2-mercaptoethanol; and a to c, in the absence of 2-mercaptoethanol. The acrylamide concentration in the gel was 10%.
Characterization of Two Pectins-The amino acid compositions of the two proteins, reported in Table III, show significant differences, particularly in the basic amino acids, proline, leucine, and half-cystine. As expected from the amino acid composition and the presence of interchain disulfide bonds (Fig. 3), lectin I has no free cysteine groups ( < O.l/polypeptide chain) when assayed by the method of Ellman (28) in the presence of 6 M guanidine hydrochloride.
Both lectins I and II are glycoproteins, as found using anthrone reagent (21). Determination of their sugar compositions using gas-liquid chromatography (22) showed them to contain the following sugars in approximately equal amounts per polypeptide chain: mannose (3); xylose (1); glucosamine (1 to 2); and galactosamine (0). Both contained galactose and glucose, but we could not exclude the possibility that their presence was due to small amounts of lactose. The only difference we could detect in their sugar compositions was the virtual absence of fucose in lectin I, but not in II, which Two LectiJzs from Caragana arborescens Seeds contained approximately one fucose per polypeptide chain. As fucose is easily lost during the preparation of samples and is difficult to determine as a shoulder over large background signals, this difference may be insignificant. Estimates of the molecular weights of lectins I and II were performed using both the mixture (I + II) and the purified proteins. Sedimentation velocity analysis showed the pure proteins to have single sedimentation constants (So,,,) of 6.4 Volumes of extract applied were: A, 300 ml; B, 100 ml. and 4.6 s, respectively. Similar values (6.2 * 0.4 s and 4.3 f 0.5 s) were obtained with the mixture. Sedimentation equilibrium studies of lectin II gave a plot of log concentration uersus the square of the radial distance which was linear (Fig. 5). The average molecular weight obtained from such experiments between 0.25 and 0.5 mg of protein/ml is 64,000 f 3,000. Similar plots for lectin I were nonlinear (Fig. 5). However, a tangential line can be drawn through most of the points at protein concentrations greater than 0.2 mg/ml yielding an average molecular weight of 103,000 f-3,000. The curvature seen at lower protein concentrations may be due to dissociation of this form. This would be consistent with results obtained in gel filtration experiments using the mixture (I + II), in which Fractions I and II eluted with distribution coefficients (K,) corresponding to those of globular proteins with masses of 85,000 and 55,000 daltons, respectively (data not shown).
Interactions with Cells-The lectin mixture (I + II) does not show ABO blood group specificity in hemagglutination tests. The final concentration of lectin needed to agglutinate about 75% of the erythrocytes is approximately 1.5 wg/ml, but much higher concentrations are needed to agglutinate L,,,, (50 pg/ml) or Ehrlich ascites (230 pg/ml) cells to the same extent. After separation, the two proteins have widely different agglutinating activities (Fig. 6). Lectin I is by far the more active, and it accounts for >99% of the total hemagglutinating activity in the (I + II) preparation.

Lectin
II has 400-fold lower activity, and we cannot exclude at this time that hemagglutination by II is due to low levels of contamination with I. Lectin II appears to bind to the cells, however, as addition of large excesses of II slightly inhibits agglutination by I (Fig. 6). Treatment of erythrocytes with neuraminidase (29) or with trypsin (30) increases the agglutination by both lectins approximately 5-fold, but it does not render agglutination by either blood group-specific.
When assayed with fibroblasts, the lectins, either separately or together, agglutinate transformed and normal cells with the same efficiency (Fig. 7). While lectin I can agglutinate both qglutmin (uglml) FIG. 5 (left). Sedimentation equilibrium analysis of the two Carugana lectins. Lectins were purified as described in the text and the legend to Fig. 5. They were then concentrated by ultrafiltration and dialyzed against buffered saline to free them of lactose. Sedimentation was performed in this buffer at 20" using an AN-F rotor. The equilibrium distribution of protein in the cell was measured using the photoelectric scanning absorption system of the analytical ultracentrifuge. Logarithms of the optical density at 280 nm are plotted against the square of the radial distance, R, from the axis of rotation. 3T3 and SVlOl-3T3 cells completely, II can clump only about one-half of the cells. Mixtures of I and II give intermediate results. This suggests that the presence of II inhibits agglutination of fibroblasts as well as of erythrocytes.
The sugar specificity of lectin I has been investigated using inhibition of hemagglutination as a measure of interaction. The concentration of sugar which decreases the agglutination reaction from nearly complete agglutination (f + + +) to a point where approximately 75% (or ++) of the cells are clumped can be taken as a measure of the affinity of the lectin for the sugar (15). The results obtained with several monosaccharides and glycosides are reported in Table IV. Lectin I has highest affinity for N-acetylgalactosamine but is also inhibited by galactose. The lectin seems to prefer monosaccharide to disaccharide haptens, suggesting that the sugar binding site of lectin I accommodates only 1 monosaccharide unit. We found no consistent preferential interaction with (Y or p anomers.
Similar measurements with lectin II are not yet possible. As the hemagglutinating activity of this lectin may be due to contaminating amounts of lectin I, we are now beginning to study hapten binding using methods which do not rely on agglutination.
Because our ultimate aim is to use the lectins from Caragana arborescens with growing and differentiating cells, we tested their effects on the mouse fibroblast line, 3T3, in culture. The data in Table V show that 3T3 cells cultured in the presence of 100 or 200 pg/ml of the mixture (I + II) grow as well as controls. Hemagglutinating activity after incubations with cells was undiminished.
These lectins are, therefore, nontoxic to 3T3 cells under conditions in which they retain full activity. Similar results were found both with the purified lectins and with SVlOl-3T3 cells in log phase growth using the mixture. DISCUSSION We have described the preparation and properties of two lectins from C. arborescens seeds. Both are heterogeneous with The values given are the sugar concentrations (K-'$'") which inhibit hemagglutination by 25% (from + + + + to + +). All values are subject to a 50% S.D. inherent in the assay procedure.  Perhaps the most striking distinction between the two proteins is their different specificities for the two components of the affinity matrix on which they were purified.' This almost 'We do not know why lectin II fails to bind to Sepharose coupled with N-acetylgalactosamine, but we believe that the relatively long reaction we used to activate Sepharose with cyanogen bromide (see "Experimental Procedures") destroyed most of the galactose residues in the gel to which this lectin presumably binds. Preparation of coupled certainly means that the sugar hapten specificities of the two proteins are also different, but we have not yet been able to confirm this inference in quantitative experiments using lectin II.
The two Caragana lectins have a number of properties which distinguish them from other lectins with specificity for Nacetylgalactosamine (31)(32)(33)(34)(35)(36)(37)(38) and which also make them potentially useful for studying growing and differentiating cells. Foremost in this respect is their lack of toxicity to cultured mouse fibroblast cells. They also have rather high affinity for sugar haptens.
As cells in culture are often severely affected by sugars added to the culture medium,* this is a further advantage.
Finally, the fact that these lectins agglutinate normal and transformed mouse cells to the same extent offers a good opportunity to compare their properties to those of galactose and N-acetylgalactosamine-specific lectins which preferentially agglutinate transformed cells (31,37